Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Assessment of MRI to estimate metastatic dissemination risk and prometastatic effects of chemotherapy

Assessment of MRI to estimate metastatic dissemination risk and prometastatic effects of... www.nature.com/npjbcancer ARTICLE OPEN Assessment of MRI to estimate metastatic dissemination risk and prometastatic effects of chemotherapy 1,2,3,4,15 5,15 6,7 8,9 8,9 George S. Karagiannis , Anthony Bianchi , Luis Rivera Sanchez , Kamal Ambadipudi , Min-Hui Cui , 10 11 3,4,6 3,4,6 2,7 3,11 12 Jesus M. Anampa , Saeed Asiry , Yarong Wang , Allison S. Harney , Jessica M. Pastoriza , Yu Lin , Xiaoming Chen , 6,11 2,3,4,11 13 8 8 10 Joan G. Jones , David Entenberg , Dana Haddad , Laura J. Hodges , Timothy Q. Duong , Joseph A. Sparano , 2,3,4,11 3,8,9,14 2,3,4,7,12 ✉ ✉ Maja H. Oktay , Craig A. Branch and John S. Condeelis Metastatic dissemination in breast cancer is regulated by specialized intravasation sites called “tumor microenvironment of metastasis” (TMEM) doorways, composed of a tumor cell expressing the actin-regulatory protein Mena, a perivascular macrophage, and an endothelial cell, all in stable physical contact. High TMEM doorway number is associated with an increased risk of distant metastasis in human breast cancer and mouse models of breast carcinoma. Here, we developed a novel magnetic resonance imaging (MRI) methodology, called TMEM Activity-MRI, to detect TMEM-associated vascular openings that serve as the portal of entry for cancer cell intravasation and metastatic dissemination. We demonstrate that TMEM Activity-MRI correlates with primary tumor TMEM doorway counts in both breast cancer patients and mouse models, including MMTV-PyMT and patient-derived xenograft models. In addition, TMEM Activity-MRI is reduced in mouse models upon treatment with rebastinib, a specific and potent TMEM doorway inhibitor. TMEM Activity-MRI is an assay that specifically measures TMEM-associated vascular opening (TAVO) events in the tumor microenvironment, and as such, can be utilized in mechanistic studies investigating molecular pathways of cancer cell dissemination and metastasis. Finally, we demonstrate that TMEM Activity-MRI increases upon treatment with paclitaxel in mouse models, consistent with prior observations that chemotherapy enhances TMEM doorway assembly and activity in human breast cancer. Our findings suggest that TMEM Activity-MRI is a promising precision medicine tool for localized breast cancer that could be used as a non-invasive test to determine metastatic risk and serve as an intermediate pharmacodynamic biomarker to monitor therapeutic response to agents that block TMEM doorway-mediated dissemination. npj Breast Cancer (2022) 8:101 ; https://doi.org/10.1038/s41523-022-00463-5 1,5,6 INTRODUCTION the lungs (and other secondary sites), as well as metastases . Consistent with these findings, TMEM doorway density in the Cancer cell dissemination occurs through specialized intravasation primary tumor has been validated as an independent prognostic portals on blood vessels called Tumor Microenvironment of biomarker for distant recurrence in human breast cancer in three Metastasis (TMEM) doorways . TMEM doorways consist of a 2,3,7 independent cohorts, including ~1150 patients , and thus may perivascular macrophage, a tumor cell overexpressing the actin- serve as a biomarker for distinguishing potentially lethal from non- regulatory protein Mammalian-enabled (MENA), and an endothe- 2–4 lethal cancers. lial cell, all in direct physical contact with each other . Cancer cell Consistent with these findings, TMEM doorway density in the intravasation at TMEM doorways occurs during tightly regulated primary tumor has been associated with distant recurrence in transient localized vascular opening events triggered by the TMEM human breast cancer in three independent cohorts, including doorway macrophage, which expresses the angiopoietin receptor 2,3,7 ~1150 patients , and thus may serve as a biomarker for TIE2. Upon stimulation, TIE2 macrophages at TMEM doorways distinguishing potentially lethal from non-lethal cancers. In the secrete vascular endothelial growth factor-A (VEGFA), leading to first proof-of-concept study involving 30 case-control pairs of the localized disruption of the underlying endothelial adherens patients with and without distant recurrence, TMEM doorway and tight junctions and, as a consequence, the opening of the density was significantly higher in patients with recurrence vessel wall (TMEM-associated vascular opening, or TAVO), localized vascular leakiness, and subsequent tumor cell transen- (P = 0.00006) . In a subsequent prospective validation in 259 dothelial migration and intravasation . As such, increased TMEM case-control pairs with and without distant recurrence from a doorway density and activity in the primary tumor microenviron- population-based cohort, TMEM doorway density was likewise associated with an increased risk of distant metastasis in the ment have been associated with an increased incidence of circulating tumor cells (CTCs), disseminated tumor cells (DTCs) in subset of those with hormone receptor-positive, HER2-negative 1 2 Department of Microbiology & Immunology, Albert Einstein College of Medicine, Bronx, NY, USA. Tumor Microenvironment and Metastasis Program, Albert Einstein Cancer 3 4 Center, Bronx, NY, USA. Integrated Imaging Program, Albert Einstein College of Medicine, Bronx, NY, USA. Gruss-Lipper Biophotonics Center, Albert Einstein College of 5 6 Medicine, Bronx, NY, USA. California State University, Bakersfield, CA, USA. Department of Anatomy and Structural Biology, Albert Einstein College of Medicine, Bronx, NY, USA. 7 8 Department of Surgery, Montefiore Medical Center, Albert Einstein College of Medicine, Bronx, NY, USA. Department of Radiology, Montefiore Medical Center, Albert Einstein 9 10 College of Medicine, Bronx, NY, USA. Gruss Magnetic Resonance Research Center, Albert Einstein College of Medicine, Bronx, NY, USA. Department of Medical Oncology, Montefiore Medical Center, Albert Einstein College of Medicine, Bronx, NY, USA. Department of Pathology, Montefiore Medical Center, Albert Einstein College of 12 13 Medicine, Bronx, NY, USA. Department of Cell Biology, Albert Einstein College of Medicine, Bronx, NY, USA. Maediclinic Middle East, Department of Breast Imaging, Dubai, 14 15 United Arab Emirates. Department of Physiology and Biophysics, Albert Einstein College of Medicine, Bronx, NY, USA. These authors contributed equally: George S. Karagiannis, Anthony Bianchi. email: craig.branch@einsteinmed.edu; john.condeelis@einsteinmed.edu Published in partnership with the Breast Cancer Research Foundation 1234567890():,; G.S. Karagiannis et al. breast cancer (P trend = 0.004), but not in triple-negative or HER2- typical approach for clinical assessment simply compares one or positive breast cancer. In the second prospective validation several post-GBCA images to a single pre-GBCA image, looking for cohort, including 600 patients from a clinical trial cohort treated features reflecting high contrast agent passage associated with with adjuvant chemotherapy, proportional hazards models tissue pathology . revealed a significant positive association between continuous One method that has been employed to estimate permeability TMEM doorway density score and early distant recurrence is k , which derives a bulk transfer constant (a rate constant) trans (P = 0.001) and locoregional plus distant recurrence between intravascular and extravascular spaces from a combina- (P = 0.00006) within 5 years of diagnosis in the subset of 297 tion of permeability and surface area . Its estimation requires patients with hormone receptor-positive, HER2-negative disease, measurements of both the arterial input and the tissue response but not in triple-negative or HER2-positive breast cancer. TMEM to the GBCA at a high temporal rate for several minutes. This doorway density score correlated poorly with the 21-gene method would allow measurement of TAVO by separating the Recurrence Score (r = 0.29) and was significantly prognostic for extravascular transfer of GBCA that does not reflect typical early locoregional and distant recurrence (p = 0.05) in multivariate permeability from the bolus efflux of plasma-borne GBCA into models including tumor size, grade, nodal metastasis, and the 21- the tissue space. When many TMEM doorways exist within a tissue gene Recurrence Score, with a trend toward first distant region, their opening is stochastic and temporally independent, recurrence (p = 0.10). Finally, TMEM doorway density did not which would appear in contrast-enhanced MRI as an overall significantly correlate with tumor size or nodal status, and increase in tissue permeability with individual TAVO events compared with hormone receptor-positive, HER2-negative dis- obscured. Thus, a dynamic measure of the tissue permeability, ease, was significantly higher in triple-negative or HER2-positive one which allows rates of efflux to be distinguished, may provide a breast cancer (P = 0.001 and P = 0.003, respectively), breast cancer means to detect TMEM doorways via this increased efflux rate. subtypes associated with higher recurrence rate. The totality of Further restriction to the initial time period post-GBCA adminis- the data, therefore, suggests a strong positive association tration may further enhance discrimination of TMEM doorway between TMEM density in the primary tumor and breast cancer activity. We, therefore, implemented a ‘limited first pass’ assess- recurrence. ment of the GBCA transfer rate between intravascular and Assessment of cancer cell dissemination and metastasis in extravascular spaces, calculated using only data from the initial preclinical models is currently performed using histological bolus of the GBCA uptake, leading to the ability to select a range endpoints, such as the number and localization of disseminated of transfer rates from the rate-histogram more likely to represent tumor cells at metastatic sites, as well as contextual changes in the 30 TMEM doorway activity, especially in highly permeable tissues . tumor microenvironment. Although such methods provide some In this study, we demonstrate that such a measure herein termed degree of morphological and molecular information, they are “TMEM Activity-MRI”, corresponds to the extent of TAVO events. limited by the need to euthanize the animal and thus lack We do not only demonstrate that TMEM Activity-MRI can be evidence regarding the dynamic nature of the phenomena they utilized in basic and translational settings to answer questions describe. Furthermore, only a small portion of the tumor is related to the molecular mechanism of cancer cell dissemination interrogated with the limited tissue sectioning typically used for but also in the clinical setting as a newly proposed and non- the assessment of histological endpoints. The ability to observe invasive companion diagnostic for breast cancer patients. cancer cell dissemination in situ, over short or long periods of time, and without the need for terminal procedures, has the potential to make a tremendous addition to our understanding of RESULTS spatiotemporal changes in the tumor microenvironment, asso- TMEM activity-MRI measurement correlates with TMEM ciated with cancer cell dissemination and metastasis. Indeed, such doorway-associated vascular opening (TAVO) events in vivo observations have been now made possible through the To measure TAVO events in tumors using magnetic resonance emergence of sophisticated imaging modalities in live animals, 8–26 imaging (MRI), we developed an algorithm (Materials and such as multiphoton intravital imaging . Although these Methods) to generate a dynamic contrast-enhanced first-pass techniques offer great potential in basic research, they cannot deconvolution MRI map for mouse primary breast carcinomas. As be used clinically because they require genomic incorporation of discussed above, prior evidence indicates that tumor cell artificially created fluorescent transgenes or direct injection of intravasation occurs exclusively in association with localized fluorescent reporters in the test subjects, and they also have poor vascular openings at TMEM doorways . To establish the TMEM depth of imaging in whole tissues. To circumvent this problem, we Activity-MRI method, we utilized a previously established trans- have investigated Magnetic Resonance Imaging (MRI) as an in vivo genic mouse model of spontaneous breast carcinoma, the Mouse versatile and non-invasive imaging tool to simultaneously Mammary Tumor Virus (MMTV) Polyoma Middle-T Antigen (PyMT) measure multiple tissue properties (e.g., structural, functional, model (Supplementary fig. 1), also known as MMTV-PyMT (or and metabolic) associated with TMEM activity and metastatic simply PyMT), which recapitulates human breast cancer develop- dissemination in breast cancer. 31–33 ment and progression in a clinically relevant fashion . In a clinical setting, contrast-enhanced MRI is commonly used MMTV-PyMT mice bearing spontaneous breast tumors were for the detection and characterization of breast cancer. During a subjected to a dynamic MRI protocol using a 9.4 T 31 cm Agilent contrast-enhanced MRI exam, a Gadolinium-based contrast agent Direct Drive imaging system. A baseline T1 map was acquired (GBCA) is typically injected, and the passage of GBCA into the using the variable flip angle (FA) approach and gradient recalled tumor parenchyma is visualized by comparing image intensity 34,35 echo (GRE) imaging (Fig. 1a), similar to previous publications , changes pre- and post-contrast injection, also known as with slight modifications as described in Materials and Methods. subtraction-based contrast . Alternatively, high temporal resolu- To delineate regions of interest (ROIs) that correspond to the tion MRI of the dynamic passage of GBCA through tissue beds tumor mass, the GRE image with a FA of 12 is used as it delineates using longitudinal (T1) relaxation-based imaging can be used to the tumor tissue versus the normal tissue with the highest measure the tissue’s permeability to the GBCA. Typically, GBCA contrast compared to the other images, and it facilitates the affects MRI signal intensity in direct relation to its concentration anatomical demarcation of the tumor margins (Supplementary within tissues, allowing for mathematical determination of its rate of exchange across the capillary. Endothelial permeability of GBCA Fig. 2a–d). A dynamic GRE sequence is initiated, and the can thus be determined directly from the observed arterial input gadolinium-based contrast agent (GBCA) is injected after 60s and tissue responses of the MRI dynamic data. However, the using a standard dose of 140 μL of 0.1 mmol/Kg gadopentetate npj Breast Cancer (2022) 101 Published in partnership with the Breast Cancer Research Foundation 1234567890():,; G.S. Karagiannis et al. Fig. 1 Development of TMEM activity-MRI assay. a T1 map of PyMT late carcinoma tumor (Left). A black arrow pointing at a potential blood/ necrotic region. Example voxel intensity over varying flip angle (Right). This graph is fitto Eq. 5 at every voxel to calculate the baseline T1 map. b Dynamic GRE images over time post-contrast injection (at 60 s). Yellow arrow, a small amount of visual change over time. c Gadolinium- based contrast agent (GBCA) concentration map over time, calculated using the baseline T1 map and dynamic GRE using Eq. 8. Yellow arrow, very slow leak area and most likely necrotic, showing up as small in the K image, indicating necrotic regions may have low K values. d GBCA fp fp concentration of the arterial source (left), where the first pass of the contrast agent is the first peak, and the second smaller peak suggests GBCA recirculation. An example voxel from within the tumor with fast uptake of contrast agent (right). Note the scale difference with the arterial source. e Final TMEM Activity-MRI map calculated using Eq. 4. Orange arrow, hyper-intense voxel, i.e., larger value in K is indicative of fp increased TAVO events. f Experimental design and mouse cohort composition. MMTV mouse mammary tumor virus, PyMT Polyoma Middle-T antigen, EC early carcinoma, LC late carcinoma. g TMEM identification by triple-stain immunohistochemistry (IHC) and representative images from early (EC) and late (LC) carcinoma MMTV-PyMT mice. Scale = 100um h Quantification of TMEM doorways (TMEM doorway score), assessed in 10 high-power fields (HPFs) in mice shown in (g). Mann–Whitney U-test. I Frequency histogram of the combined Early (EC) and Late (LC) Carcinoma mouse cohorts. The threshold is representative of creating the TMEM Activity-MRI Eq. 9. The optimal threshold was calculated to be ~0.001. j Representative TMEM Activity-MRI maps of mouse tumors by magnetic resonance imaging (MRI) from early (EC) and late (LC) carcinoma MMTV-PyMT mice. Hyper-intense voxels correspond to TMEM hotspots (i.e., increased number of TAVO events). Scale = 1 mm. k Quantification of TAVO events (a.k.a. TMEM activity), assessed via the TMEM Activity-MRI assay, in mice shown in (j). Mann–Whitney U-test. l Correlation between TMEM score (as quantified in h) and TMEM Activity-MRI (as quantified in k) in early and late carcinomas of the MMTV-PyMT mice shown in (j). Spearman’s rank correlation coefficient. Error bars: standard deviation (SD). Published in partnership with the Breast Cancer Research Foundation npj Breast Cancer (2022) 101 G.S. Karagiannis et al. dimeglumine at a rate of 20 µL/sec (Fig. 1b). Using the baseline T1 specifically assessing the tumor compartment that contains the map and the dynamic GRE sequence, a GBCA dynamic leakiest blood vessels in the entire tumor. In addition, TMEM concentration map is then calculated from the relaxivity map, as Activity-MRI avoids potential bias coming from tumor size previously reported , giving an estimate of GBCA concentration variations since it gives the percentage, and not the absolute for each individual voxel (Fig. 1c). From the dynamic concentration value, of the voxels in a given tumor that has the highest image series, the arterial source is identified (Fig. 1d; left panel), permeability. Upon quantification, it became evident that TMEM and an arterial input concentration curve is subsequently Activity-MRI was significantly (p < 0.05; Mann–Whitney U-test) determined. An example region of the tumor with fast uptake is higher in the LC compared to the EC cohort in MMTV-PyMT mice shown in the pipeline (Fig. 1d; right panel). Importantly, finding (Fig. 1j, k). In support of this, TMEM Activity-MRI significantly the arterial source also allows for the normalization among (p < 0.05; Spearman rho = 0.55; Spearman’s rank correlation) different subjects (Materials and Methods). As a next step, the time correlates with TMEM doorway score (Fig. 1l), suggesting a series of each voxel is fit to equation #4 (Materials and Methods) positive correlation between TMEM Activity-MRI and tumor to generate the corresponding TMEM Activity-MRI map (Fig. 1e). progression-dependent increases of TMEM doorway number. Importantly, high TMEM doorway activity is reflected by voxels Taken together, observations shown in Fig. 1 suggest that the with hyperintensities in the TMEM Activity-MRI map (Fig. 1e; newly developed TMEM Activity-MRI measurement can success- orange arrow) due to large endothelial openings. Voxels with low fully capture changes in TMEM doorway score, for instance, the intensity in the TMEM Activity-MRI map likely reflect inactive increase in TMEM doorway number during the progression from TMEM doorways or regions devoid of TMEM doorways. Necrotic or early- to late-stage carcinoma) in a (patho)physiologically relevant non-perfused regions appear as either hyper-intense or hypo- preclinical model of mammary carcinoma. intense voxels on T1W images, depending on whether fast- (open vasculature, Fig. 1a; black arrow) or slow-contrast-uptake (avas- Clodronate-mediated reduction of TMEM doorways decreases cular) regions (Fig. 1c, yellow arrow) are present respectively, thus TMEM Activity-MRI both yielding little or no permeability in the first-pass GBCA TMEM doorways are dynamic structures, and as such, TMEM uptake (i.e., TMEM Activity-MRI) maps (Fig. 1e). doorway activity oscillates in time . Multiphoton intravital imaging MMTV-PyMT mice, which spontaneously develop breast carci- studies in live animal tumors suggest that a higher density of noma, have been shown to form an increasing number of active TMEM doorways in the tumor microenvironment proportionally TMEM doorways throughout tumor progression . Therefore, a corresponds to a higher probability of TAVO occurrence . Based comparison between early (EC) and late (LC) carcinomas not only on this premise, along with data shown in Fig. 1j–l, we reasoned represents an excellent way to study TAVO events and cancer cell that targeted suppression of TMEM doorway formation should dissemination but also an excellent model to develop a sensitive overall reduce TMEM Activity-MRI, because fewer TMEM doorways MRI measurement corresponding to TMEM-dependent contrast would provide fewer TAVO events per voxel. To suppress TMEM agent leakage, as described in the pipeline above (Fig. 1a–e). To doorway formation, we adopted the treatment with clodronate this end, we first generated an MMTV-PyMT mouse cohort (Fig. 1f), liposomes, a pharmacologic macrophage depletion strategy that which included mice belonging to age groups corresponding to reduces TMEM doorways in mice . Clodronate liposomes or early carcinoma (EC; 6–8-week old, N = 9) and late carcinoma (LC; vehicle control (i.e., PBS liposomes) were administered for two 10–13-week old, N = 14), as described previously . The individual weeks in 7-week-old PyMT mice with palpable tumors, and, after mouse tumors were also examined histologically in a retrospective the completion of treatment, mice were subjected to TMEM manner (i.e., upon tumor resection following an MRI session) to Activity-MRI, TMEM score, and circulating tumor cells (CTC) confirm early- or late-stage carcinoma status by histopathology measurements (Fig. 2a). As a positive control, clodronate-treated (Supplementary Fig. 1). As expected and also reported pre- 1,36 mice presented with significantly (p < 0.01; Mann–Whitney U-test) viously , TMEM doorway assembly was significantly (p < 0.01; fewer macrophages overall, as assessed by IBA1 immunohisto- Mann–Whitney U-test) increased in LC compared to EC samples chemistry (Fig. 2b, c and Supplementary Fig. 3b). Because (Fig. 1g, h and Supplementary Fig. 3a). Because TMEM activity is macrophages are integral components of the TMEM doorway cell increased in late-stage compared to early-stage PyMT carcinomas, 4,38 triads , we also confirmed significantly (p < 0.05; Mann–Whitney we anticipated that such differences should be reflected in the U-test) fewer TMEM doorways upon clodronate treatment (Fig. 2d, TMEM Activity-MRI maps, as developed above (Fig. 1a–e). To look e), in line with the clodronate-mediated macrophage depletion into this possibility, all PyMT breast tumor images were manually (Fig. 2b, c). Despite the significant depletion of macrophages in segmented from the TMEM Activity-MRI images, and their clodronate-treated mice, we did not observe any significant corresponding histograms were pooled for each group together changes in histological features between vehicle- and clodronate- and graphed (Fig. 1i). In this analysis, k rates less than 0 and fp −3 treated animals (Supplementary Fig. 1). Importantly, however, above 40 × 10 were masked off as error voxels (refer to Materials TMEM Activity-MRI was significantly (p < 0.05; Mann–Whitney U- and Methods). This histogram analysis suggested that late test) reduced in the clodronate-treated mice (Fig. 2f, g), indicating carcinomas had a lower frequency of hypo-intense voxels that the reduction of TMEM doorways via elimination of tumor- (0.0–0.001) but a higher frequency of hyper-intense voxels associated macrophages may affect the number of TAVO events (0.001–0.04) compared to early carcinomas (Fig. 1i), which was per voxel. Clodronate treatment slightly affected tumor growth at consistent with our hypothesis of an expected pattern of the endpoint (Supplementary Fig. 4a–c), but this did not bias the increased TMEM doorway-dependent vascular opening during measurement of TMEM Activity-MRI, because the variable breast cancer progression. represents a ratio of hyper-intense voxels to the total number of To establish easily interpretable and biologically relevant tumor voxels and thus is unaffected by tumor size. Moreover, endpoints for statistical comparisons, we next calculated an MRI TMEM Activity-MRI correlated significantly (p < 0.05; Spearman feature by combining the histogram analysis (Fig. 1i) from the rho = 0.84; Spearman’s rank correlation) with TMEM score (Fig. TMEM Activity-MRI maps with permeability thresholding (Fig. 1i; 2h), suggesting a linear/proportional correlation between TMEM black line). The calculated MRI measurement, simply termed score and TMEM Activity-MRI, as we hypothesized. As mentioned “TMEM Activity-MRI,” is calculated from the corresponding TMEM above, an increased number of TMEM scores in the tumor Activity-MRI map and represents the ratio of the number of tumor voxels presenting with a permeability score above a certain microenvironment is expected to correlate with an increased threshold divided by the number of total voxels within the tumor probability of TAVO events and, as such, cancer cell dissemination. ROI. As such, TMEM Activity-MRI signifies a suitable metric for We thus finally measured circulating tumor cells (CTCs) and found npj Breast Cancer (2022) 101 Published in partnership with the Breast Cancer Research Foundation G.S. Karagiannis et al. Fig. 2 TMEM activity-MRI is suppressed, as a result of reduced TMEM doorway formation, using clodronate-mediated macrophage depletion. a Experimental strategy and mouse cohort composition for MMTV-PyMT mice subjected to clodronate-dependent macrophage depletion. PyMT Polyoma Middle-T antigen, CTC circulating tumor cell. b Identification of macrophages by IBA1 immunohistochemistry and representative images from MMTV-PyMT mice, treated with either control or clodronate liposomes. Scale = 100 um. c Quantification of IBA1 macrophages, as averaged in 10 high-power fields (HPFs) in MMTV-PyMT mice shown in (b). Mann–Whitney U-test. d TMEM identification by triple-stain immunohistochemistry (IHC) and representative images from MMTV-PyMT mice, treated with either control or clodronate liposomes. Scale = 100 um. e Quantification of TMEM doorways (TMEM score), as assessed in 10 high-power fields (HPFs), in MMTV-PyMT mice shown in (d). Mann–Whitney U-test. f Representative TMEM Activity-MRI maps of mammary carcinoma tumors by magnetic resonance imaging (MRI) from MMTV-PyMT mice treated with clodronate (second column) or vehicle control (first column). Scale = 1 mm. g Quantification of TMEM-mediated vascular opening events (TMEM doorway activity), assessed via TMEM Activity-MRI assay in MMTV- PyMT mice shown in (f). Mann–Whitney U-test. h Correlation between TMEM score (as quantified in e) and TMEM Activity-MRI (as quantified in g) in MMTV-PyMT mice treated with either control or clodronate liposomes. Spearman’s rank correlation coefficient. i Circulating tumor cell (CTC) counts in MMTV-PyMT mice treated with either control or clodronate liposomes. Mann–Whitney U-test. j Correlation of TMEM activity-MRI score with circulating tumor cells (CTCs) in MMTV-PyMT mice, treated with either control or clodronate liposomes. Spearman’srankcorrelation coefficient. Error bars: standard deviation (SD). that clodronate-mediated macrophage depletion resulted in a detect rebastinib-mediated suppression of TMEM activity. We used two independent mouse models of breast carcinoma, both significant (p < 0.05; Mann–Whitney U-test) reduction of CTCs of which received either rebastinib or vehicle control, to alter (Fig. 2i). In line with this, TMEM Activity-MRI also correlated TMEM doorway activity. The first model was developed via significantly (p < 0.05; Spearman rho = 0.82; Spearman’s rank syngeneic transplantation of PyMT tumors from late-stage PyMT correlation) with the number of CTCs (Fig. 2j). The data presented donors into wild-type FVB hosts (Fig. 3a). The second model was in this section collectively suggest that TMEM Activity-MRI developed via xenogeneic transplantation of patient-derived HT17 assesses TMEM-dependent metastatic dissemination. In this tumor chunks [previously established from an estrogen receptor- section, clodronate treatment was not performed in experimental negative (ER-) breast cancer patient ] into immunocompromised mice for purposes of proposing clinical intervention but only for SCID hosts (Fig. 3a′). Both animal models have been detailed purposes of indirectly suppressing TMEM doorway assembly and previously in studies involving TMEM-dependent cancer cell activity and confirming that such suppression could be detected 1,5,36,39 dissemination and metastasis . Following the development in the newly established TMEM Activity-MRI assay. of palpable tumors in both models (4–6 weeks after transplanta- tion), mice received a 3-week treatment with rebastinib or vehicle Rebastinib inhibition of TMEM doorway function decreases [administration protocol detailed in ref. ], followed by an MRI TMEM Activity-MRI session and subsequent measurement of TMEM Activity-MRI and Prior studies have demonstrated that TMEM doorways induce other metastatic endpoints (Fig. 3a, a′). Histological assessment of localized and transient vascular opening associated with tumor the resected PyMT and HT17 tumors, as expected, revealed cell intravasation, which are both tightly regulated by the TIE2 features of late-stage carcinomas (Supplementary Fig. 1), consis- macrophage at TMEM doorways . Indeed, TMEM doorway activity tent with prior observations . Importantly, TMEM Activity-MRI was can be suppressed by targeting the TIE2 signaling pathway in significantly (p < 0.05; Mann–Whitney U-test) reduced in both the 5,6 perivascular macrophages . The pharmacological suppression of PyMT (Fig. 3b, c) and the HT17 (Fig. 3b, c) rebastinib-treated mice the TIE2 signaling pathway in macrophages at TMEM doorways compared to vehicle-treated mice. It should be underscored that can be achieved by the specific TIE2 inhibitor, rebastinib, and at rebastinib, as opposed to the clodronate treatment experiments doses which demonstrate minimal off-target effects based on in described above (Fig. 2), specifically affects the function, but not vivo-relevant cellular assays measuring inhibition of kinase the assembly of TMEM doorways in the breast tumor microenvir- 5,6 5,6 activity . To examine if TMEM Activity-MRI is indeed indicative onment . Indeed, the TMEM doorway score remained unaltered of TAVO events, we tested the TMEM Activity-MRI’s ability to between vehicle- and rebastinib-treated animals for both the Published in partnership with the Breast Cancer Research Foundation npj Breast Cancer (2022) 101 G.S. Karagiannis et al. PyMT and the HT17 models (Supplementary Fig. 5a, b). Interest- reduction of TMEM doorway activity in rebastinib-treated mice. ingly, in the absence of an extrinsic factor (e.g., rebastinib) capable This observation further indicates that TMEM Activity-MRI better of modulating TMEM doorway activity (as, for example, in the mirrors the activity, and not as much the assembly, of TMEM transition from early to late breast carcinoma where TMEM doorways in the tumor microenvironment. doorway activity increases with TMEM doorway score), TMEM activity-MRI correlates well with TMEM doorway score (Fig. 1l). TMEM Activity-MRI correlates with established endpoints of However, upon treatment of breast carcinomas with rebastinib, TMEM doorway-associated vascular opening and metastatic we did not observe any correlation between TMEM score and dissemination TMEM activity-MRI in either of the two models tested (Supple- Previously, we have developed a multichannel immunofluores- mentary Fig. 5c). This observation was completely expected since cence assay to specifically visualize and quantify localized TAVO rebastinib is known to inhibit TMEM doorway activity without events in breast cancer . This assay, here termed “TMEM Activity- affecting the assembly of new, or the breakdown of existing, TMEM doorways . As a consequence, these observations collec- Dextran,” is based on the intravenous injection of high-molecular- tively show that TMEM Activity-MRI detects the significant weight (155 kDa) dextran conjugated to tetramethylrhodamine npj Breast Cancer (2022) 101 Published in partnership with the Breast Cancer Research Foundation G.S. Karagiannis et al. Fig. 3 TMEM activity-MRI corresponds to metastatic dissemination endpoints in a mouse model of spontaneous breast carcinoma (MMTV-PyMT; a–k) and a breast cancer patient-derived xenograft (HT17; a′–h′), and is suppressed by the specific TMEM doorway inhibitor, rebastinib. a–a′ Experimental strategy and mouse cohort composition for syngeneic (A) and xenogeneic (A′) mouse models of breast carcinoma. PyMT polyoma middle-T antigen, CTC circulating tumor cell, SCID severe combined immunodeficiency, FVB friend virus B. b–b′ Representative TMEM Activity-MRI maps of mammary carcinoma tumors by magnetic resonance imaging (MRI) from PyMT (B) and HT17 (B′) mice treated with rebastinib (second column) or vehicle control (first column). Scale = 1 mm. c–c′ Quantification of TMEM-mediated vascular opening events (TMEM doorway activity), assessed via TMEM Activity-MRI assay, in PyMT (c) and HT17 (c′) mice. Mann–Whitney U-test. d–d′ Binarized images of extravascular dextran in mammary carcinoma tumors from PyMT (d) and HT17 (d′) mice treated with rebastinib (second column) or vehicle control (first column). Scale = 20 um. e-e′ Quantification of TAVO events (a.k.a. TMEM activity), assessed via the TMEM activity-dextran assay, in PyMT (e) and HT17 (e′) mice. Mann–Whitney U-test. f–f′ Correlation of TMEM doorway activity scores, as quantified with the TMEM activity-dextran and TMEM activity-MRI assays, in PyMT (f) and HT17 (f′) mice, treated with either rebastinib or vehicle control. Spearman’s rank correlation coefficient. g–g′ Circulating tumor cell (CTC) counts in PyMT (g) and HT17 (g′) mice treated with either rebastinib or vehicle control. Mann–Whitney U-test. h–h′ Correlation of TMEM activity-MRI score with circulating tumor cells (CTCs) in PyMT (h) and HT17 (h′) mice, treated with either rebastinib or vehicle control. Spearman’s rank correlation coefficient. i Multichannel immunofluorescence of disseminated tumor cells (DTCs) in MMTV-PyMT mice treated with either rebastinib (right panel) or vehicle control (left panel), as assessed by co-staining of the PyMT antigen, Pancytokeratin (PanCK) and DAPI. Magnified inserts show individual fluorescent channels for the cells outlined with squared boxes in the main images. Scale = 40 um. j Quantification of disseminated tumor cells (DTCs) in MMTV-PyMT mice, treated with either rebastinib or vehicle control. Mann– Whitney U-test. k Correlation of TMEM activity-MRI score with disseminated tumor cells (DTCs) in PyMT mice, treated with either rebastinib or vehicle control. Spearman’s rank correlation coefficient. Error bars: standard deviation (SD). (TMR) in experimental mice a few minutes before the termination factors, such as the dwell time of CTCs in the circulation resulting of the experiment. In intact blood vessels, the fluorescent probe is from prolonged survival and evasion of CTCs from innate/adaptive 41,42 restrained in the vascular lumens, because its molecular weight immunity . prevents it from passing between the endothelial cells . However, Finally, we assessed the correlation of the newly described TME under circumstances where endothelial cell tight junctions are Activity-MRI measurement with the most direct metastatic dissolved as a result of TMEM doorway activity , the fluorescent dissemination outcome, the presence of disseminated tumor cells probe can leak into the tumor tissue and be visualized and (DTCs) in secondary sites, in particular the lungs. DTCs were quantified as a high TMR signal accumulation in the perivascular detected as single cells in the lung parenchyma, co-expressing space . Here, we adapted the TMEM Activity-Dextran assay by pancytokeratin (PanCK ), which is a generic epithelial marker, and (following the termination of the MRI session) injecting Dextran- the PyMT antigen (PyMT ), which is specific to cancer cells in the TMR directly into the tail vein and sacrificing the mice 15’ later MMTV-PyMT model (Fig. 3i). As expected, rebastinib treatment (Fig. 3a-a′ and Supplementary Fig. 6a). It should be noted that the results in significantly (p < 0.01; Mann–Whitney U-test) fewer DTCs appearance of vascular profiles, as well as the baseline values of in the lung parenchyma of PyMT mice (Fig. 3j). In support to this the TMEM Activity-Dextran assay in PyMT mice following an MRI finding, TMEM Activity-MRI significantly correlated (p < 0.01; session (adapted protocol with the addition of GBCA), are similar Spearman rho = 0.9; Spearman’s rank correlation) with DTCs in to those of PyMT mice not subjected to an MRI session [published this model (Fig. 3k). Taken together, the data presented in this protocol without the addition of GBCA as in ref. ] (Supplemen- section show that TMEM Activity-MRI is an accurate measure of tary fig. 6a, b). Moreover, no extravascular dextran staining is TMEM doorway function during dissemination and indicative of detected in tissue sections of mice receiving GBCA without the the prometastatic potential of breast carcinomas. dextran injection (Supplementary Figs. 6c), indicating the pre- sence of GBCA in the peripheral circulation of the experimental TMEM activity-MRI reveals that mammalian-enabled (MENA) is mice does not interfere with the endpoint measurement of the not essential for TMEM doorway-associated vascular opening TMEM Activity-Dextran assay. Importantly, we noticed that TMEM but is necessary for cancer cell dissemination Activity-Dextran was significantly (p < 0.05; Mann–Whitney U-test) So far, we demonstrated that TMEM Activity-MRI is a potentially suppressed in rebastinib-treated compared to vehicle-treated useful MRI-based measurement of TMEM doorway activity by PyMT (Fig. 3d, e) and HT17 (Fig. 3d′,e′) mice. In support of this using diverse mouse models of breast carcinoma with perturba- observation, TMEM Activity-MRI significantly (p < 0.05; Spearman’s tions in either the TMEM doorway number (Fig. 2) or TMEM rank correlation) correlated with TMEM Activity-Dextran in both doorway function/activity (Fig. 3). Next, we sought to investigate PyMT (Spearman rho = 0.69) and HT17 (Spearman rho = 0.74) whether this novel TMEM Activity-MRI assay could be used in models (Fig. 3f, f′), further strengthening the notion that TMEM mechanistic studies related to the molecular basis of cancer cell Activity-MRI captures TMEM doorway activity within the tumor dissemination. microenvironment. Amid the two main prerequisites of cancer cell dissemination, TMEM doorway function results not only in the localized and i.e., the presence of TMEM doorway and the presence of a transient vascular opening but also in the intravasation of highly proinvasive/promigratory cancer cell subpopulation, prior data invasive, highly migratory tumor cells into the peripheral have shown that the invasive isoform of the actin-regulatory circulation . Consistent with these observations, the administra- INV 4,38 protein MENA, MENA , is critical for the latter prerequisite . tion of rebastinib in both PyMT (Fig. 3g) and HT17 (Fig. 3g′) mice Indeed, PyMT mice lacking the MENA gene (MENA-KO) fail to significantly (p < 0.01; Mann–Whitney U-test) decreased the 43 INV establish metastatic disease . Moreover, MENA is necessary for number of CTCs in the peripheral circulation. Importantly, TMEM the transendothelial migration of tumor cells since it regulates the Activity-MRI significantly (p < 0.05; Spearman’s rank correlation) development and maturation of invadopodia, which are essential correlated with CTCs in both the PyMT (Spearman rho = 0.54) and 44–47 the HT17 (Spearman rho = 0.46) models (Fig. 3h, h′). The cytoplasmic protrusions for the migratory/invasive process . correlation of TMEM Activity-Dextran with TMEM Activity-MRI is Nevertheless, it is not clear whether MENA simply defines the stronger than that of CTCs with TMEM Activity-MRI (compare Fig. invasive properties of the prometastatic cancer cell subpopulation 3f, f′ with Fig. 3h, h′). This is in consistent with the fact that the or whether it is also necessary for TMEM doorway activity (Fig. 4a). number of CTCs in the peripheral circulation does not only The newly developed TMEM Activity-MRI assay is ideal to answer depend on TAVO events but may also be determined by other this question, as TMEM Activity-MRI is specific for measuring TAVO Published in partnership with the Breast Cancer Research Foundation npj Breast Cancer (2022) 101 G.S. Karagiannis et al. Fig. 4 TMEM activity-MRI demonstrates that MENA is not necessary for TMEM-associated vascular opening events but is essential for cancer cell dissemination. a Experimental hypothesis under investigation using the TMEM Activity-MRI assay. The illustration displays two prerequisites of metastatic dissemination: TMEM doorway [(a cell triad composed of a macrophage (green), a tumor cell (blue), and an endothelial cell (red)], and the invasive/migratory tumor cell subset migrating alongside macrophages towards underlying TMEM doorways. Although it is known from prior literature (see text for details) that MENA is necessary for inducing invasive and migratory behavior to the prometastatic tumor cells, it is not known if MENA is also necessary for TMEM-associated vascular opening (TAVO), as depicted with the +/+ −/− question mark. b Experimental strategy and mouse cohort composition for MMTV-PyMT (MENA and MENA ) mice. PyMT polyoma INV +/+ middle-T antigen, CTC circulating tumor cell. c Representative images of MENA immunofluorescence from MENA (MENA-WT) and −/− +/+ −/− MENA (MENA-KO) MMTV-PyMT mice. Scale = 50 um. d Circulating tumor cell (CTC) counts in MENA (MENA-WT) and MENA (MENA- KO) MMTV-PyMT mice. Mann–Whitney U-test. e Representative TMEM Activity-MRI maps of mammary carcinoma tumors by magnetic +/+ −/− resonance imaging (MRI) from MENA (MENA-WT) (first column) and MENA (MENA-KO) (second column) MMTV-PyMT mice. Scale = 1 mm. f Quantification of TMEM-mediated vascular opening events (TMEM doorway activity), as assessed via the TMEM Activity-MRI +/+ −/− assay, in MENA (MENA-WT) and MENA (MENA-KO) MMTV-PyMT mice. Mann–Whitney U-test. g Binarized (thresholded) images of +/+ −/− extravascular dextran in mammary carcinoma tumors from MENA (MENA-WT) and MENA (MENA-KO) MMTV-PyMT mice. +/+ Scale = 200 um. h Quantification of TAVO events (a.k.a. TMEM activity), as assessed via the TMEM Activity-dextran assay, in MENA −/− (MENA-WT) and MENA (MENA-KO) MMTV-PyMT mice. Mann–Whitney U-test. Error bars: standard deviation (SD). events and TMEM doorway activity. In this regard, we crossed Activity-MRI could be generally used to investigate the molecular MMTV-PyMT mice with MENA heterozygotes to develop MMTV- mechanisms behind cancer cell intravasation and endothelial −/− PyMT MENA (herewith referred to as MENA-KO mice), while the permeability during metastasis. +/+ MMTV-PyMT MENA (MENA-WT) litter served as the wild-type control (Fig. 4b and Supplementary Fig. 1). We subjected 7–9- Translational relevance of TMEM activity-MRI: Potential utility week-old MENA-KO and MENA-WT mice bearing an average as a companion diagnostic diameter of PyMT tumors of ~0.5 cm, to TMEM Activity-MRI assay Finally, we examined if TMEM activity-MRI could be extended into and experimental endpoints of metastatic dissemination (Fig. 4b). the preclinical setting to provide any insights into its potential Retrospectively, we confirmed that the resected MENA-KO tumors clinical importance. Our group has previously demonstrated that INV did not express the prometastatic MENA isoform using treatment with neoadjuvant paclitaxel or doxorubicin/cyclopho- immunofluorescence (Fig. 4c), and, as expected from our prior 43 sphamide can significantly increase TMEM doorway assembly and work , displayed significant (p < 0.05; Mann–Whitney U-test) activity as a result of the infiltration of prometastatic macrophages suppression of CTCs, compared to MENA-WT mice (Fig. 4d). in both mouse and human breast cancer . Such modifications in Surprisingly, we did not observe any difference (p >0.05; the tumor microenvironment are capable of delaying tumor Mann–Whitney U-test) in TMEM Activity-MRI between MENA-WT growth in the short term but otherwise obfuscate the long-term and MENA-KO mice (Fig. 4e, f), implying that MENA is not directly clinical benefits of chemotherapy treatment. These may also involved in the regulation of TMEM doorway activity, but may contribute to the observed distant relapse following treatment exclusively contribute to the establishment of the proinvasive/ 50–52 with chemotherapy in some patients . Because not all breast promigratory cancer cell subpopulation that disseminates via TMEM 46–49 cancer patients respond with the development of the aforemen- doorways .Toconfirm that the genetic elimination of the MENA tioned prometastatic macrophage infiltration upon treatment with gene in MMTV-PyMT mice did not interfere with the MRI assay giving chemotherapy, we have previously indicated the importance of false-negative results, we also evaluated these observations by using developing non-invasive approaches for monitoring the tumor TMEM Activity-Dextran in the same animals that were imaged with microenvironment while patients undergo pre-operative che- MRI (Fig. 4g, h). Similarly, there was no significant (p >0.05; 50–52 motherapy . We reasoned that TMEM activity-MRI is a Mann–Whitney U-test) difference in TMEM Activity-Dextran between promising tool for this purpose because it correlates with TMEM MENA-WT and MENA-KO mice (Fig. 4g, h), suggesting that TMEM doorway function in all preclinical models tested thus far (Figs. Activity-MRI is indeed specific for measuring TAVO events. Overall, these data served a dual purpose. Foremost, they indicate that MENA 1–3), and importantly, TMEM Activity-MRI can be acquired in a is notnecessary forTMEM doorway-dependent vascular opening, non-invasive manner. To examine if TMEM Activity-MRI can despite it being necessary for metastatic dissemination (Fig. 4a). capture chemotherapy-mediated changes in TMEM doorway Second, these data provide an accurate proof of principle that TMEM function in breast tumors, we again utilized the HT17 npj Breast Cancer (2022) 101 Published in partnership with the Breast Cancer Research Foundation G.S. Karagiannis et al. Fig. 5 Translational significance of the newly developed TMEM activity-MRI assay. a Experimental strategy and mouse cohort composition for HT17 breast cancer patient-derived xenograft. SCID severe combined immunodeficiency. b Representative TMEM activity-MRI maps of mammary carcinoma tumors by magnetic resonance imaging (MRI) in HT17 mice treated with vehicle control (Ctrl; left panel), paclitaxel chemotherapy (Ptx; middle panel), or paclitaxel plus rebastinib (Reb + Ptx; right panel). Scale = 1 mm. c Quantification of TAVO events (a.k.a. TMEM activity), assessed via the TMEM activity-MRI assay, in HT17 mice shown in (b). Kruskal–Wallis analysis of variance with Mann–Whitney U- test for post hoc analysis. d Circulating tumor cells (CTCs) in HT17 Mice treated with chemotherapy alone or chemotherapy in combination with rebastinib. Kruskal–Wallis analysis of variance with Mann–Whitney U-test for post hoc analysis. e Experimental strategy and mouse cohort composition for HT17 breast cancer patient-derived xenograft. SCID, severe combined immunodeficiency. Black arrowheads indicate time- points, in which TMEM activity-MRI was assessed. f Representative TMEM activity-MRI maps of mammary carcinoma tumors by magnetic resonance imaging (MRI) in HT17 mice treated with vehicle control (first row), paclitaxel chemotherapy (second row), or paclitaxel plus rebastinib (third row) across multiple time-points (first column, day 0; second column, day 5; third column, day 10; fourth column, day 15). Scale = 1 mm. g Quantification of TMEM-mediated vascular leakiness (TMEM activity), assessed via the TMEM activity-MRI assay across multiple time-points in the HT17 mice shown in (f). h Breast cancer patient demographic and histopathologic data. i, j Representative images from TMEM triple-stain immunofluorescence (i) and TMEM Activity-MRI maps of the corresponding tumor ROIs (j) in the pilot patient cohort of breast carcinoma, here presented as “low” TMEM doorway score patient (first row), and “high” TMEM doorway score patient (second row). Scale = 100 um (i) and 1 cm (j). k Correlation of TMEM activity-MRI score with TMEM doorway score in the pilot patient cohort. Spearman’s rank correlation coefficient. Error bars: standard deviation (SD). patient-derived xenograft (PDX) model, which demonstrates variance; post hoc analysis: Mann–Whitney U-test) reduction of excellent preclinical utility (Supplementary Fig. 1), especially in TMEM Activity-MRI (Fig. 5b) and CTCs (Fig. 5d). the context of chemotherapy-induced metastasis . It should be Based on these preliminary data, we reasoned that TMEM noted that the HT17 mice treated with paclitaxel, with or without Activity-MRI could be utilized as a companion diagnostic to rebastinib, were generated from the same cohort as the HT17 monitor the effects of chemotherapy in breast cancer patients. To mice shown in Fig. 2; as a result, the same untreated animal group evaluate such potential in a preclinical setting, we designed could be re-graphed as a reference group to demonstrate baseline another mouse carcinoma study, again using the HT17 patient- TMEM Activity-MRI values (Fig. 5a). Importantly, our results derived xenograft (Supplementary Fig. 1). In a longitudinal fashion, indicate that TMEM activity-MRI is significantly increased HT17 mice either received paclitaxel chemotherapy alone or (p < 0.05; Kruskal–Wallis analysis of variance; post hoc analysis: paclitaxel along with rebastinib, while TMEM Activity-MRI was Mann–Whitney U-test) in HT17 mice receiving neoadjuvant measured in frequent intervals to mimic the clinical scenario of paclitaxel when compared to the vehicle controls (Fig. 5b, c). As monitoring breast cancer patients in the course of neoadjuvant expected, the addition of rebastinib in paclitaxel-treated HT17 treatment (Fig. 5e). As expected , we found that paclitaxel mice results in a significant (p < 0.05; Kruskal–Wallis analysis of chemotherapy significantly (p < 0.05; repeated-measures ANOVA) Published in partnership with the Breast Cancer Research Foundation npj Breast Cancer (2022) 101 G.S. Karagiannis et al. increased TMEM Activity-MRI during the course of treatment, procedures, requiring the collection of dynamic (high temporal compared to vehicle alone (Fig. 5f, g). Importantly, however, the and spatial) resolution images starting at contrast infusion and co-administration of rebastinib with paclitaxel prevented the lasting for 90s of contrast circulation. Traditionally, standard-of- expected chemotherapy-induced increase of TMEM Activity-MRI, care images are acquired after this time period. In total, this thus bringing the raw TMEM Activity-MRI values down to the same acquisition and the accompanying pre-contrast T1-relaxation levels as the mice treated with vehicle controls (Fig. 5f, g). These quantification adds roughly 3 to-4 min to the entire clinical exam, observations suggest that the newly established TMEM Activity- does not alter the clinical effectiveness, and therefore can be MRI measurement represents a potential surrogate of TMEM routinely added to the stand-of-care exam. MRI sequences doorway activity in preclinical animal models of breast carcinoma, necessary for this rapid acquisition of dynamic data are routinely thus demonstrating the significant potential for clinical utility. available on most clinical high-field (1.5 and 3 Tesla) MRI systems, TMEM doorway activity assays are not currently available in so these methods have been easily incorporated into clinical human clinical practice because it is not possible to inject practice in the clinical setting. In conclusion, a single combined fluorescent dextran in human patients to evaluate TMEM Activity- MRI exam, which includes both the standard-of-care clinical Dextran in a clinical setting. Thus, we evaluated the possibility that assessment of the patient and our proposed TMEM Activity-MRI the new TMEM Activity-MRI measurement could serve as a non- assay, will not delay the patient’s clinical assessment or induce invasive surrogate for TMEM doorway activity and, therefore, additional discomfort to the patient and could be readily available cancer cell dissemination in humans. Thus, we assembled a “pilot” for use in the clinic as a companion diagnostic. patient cohort that accrued 11 patients without distant metastasis Several studies have previously reported on the use of dynamic who had a wide distribution in tumor size (T1 = 3, T2 = 5, T1 = 3) contrast-enhanced (DCE) MRI to estimate the metastatic state and 54–56 and axillary lymph node involvement (node-positive = 5). This outcome of breast tumors . Typically, these studies have used cohort included patients with a wide range of TMEM doorway the volume transfer constant between the blood plasma and the scores determined by using the standardized TMEM scoring extravascular extracellular space (K ) as a marker, which trans 57,58 method on tissue sections (Fig. 5h), expected to present disparate evaluates the full wash-in of the contrast agent . In contrast, 2,3,7 metastatic risk according to prior clinical investigations . Tumor TMEM Activity-MRI only uses the extravasation of GBCA during the ROI acquisition and TMEM Activity-MRI calculations were per- first pass through the vasculature, exclusively corresponding to formed in an analogous fashion to the mouse protocol, albeit with the tightly controlled vascular opening of the TMEM doorway. As minor modifications, as described in Materials and Methods. As such, TMEM doorway-independent mechanisms of vascular representative examples, TMEM doorway immunostaining images leakage (i.e., necrosis) are efficiently isolated in our established (with either low or high TMEM scores) from biopsies obtained MRI measurement, as shown in Fig. 1c (yellow arrow). Therefore, 54–56 from the tumor site prior to the MRI session are presented along compared to prior methods , TMEM activity-MRI is more with corresponding MRI analyses (Fig. 5i, j). In this pilot patient efficient in eliminating background signals from various sources cohort, TMEM Activity-MRI correlated positively and significantly not associated with the active process of cancer cell intravasation with TMEM score (p < 0.05; Spearman rho = 0.81; Spearman’s Rank and dissemination. Correlation), irrespective of tumor size, age, race, ethnicity, or TMEM doorways are composed of three individual cells in direct lymph node status (Fig. 5k). Along with the animal preclinical data and stable physical contact: a perivascular macrophage, an shown in this study, this human pilot study additionally indicates endothelial cell, and a tumor cell highly expressing the actin- 2,4,53 that TMEM Activity-MRI could serve as a companion diagnostic in regulatory protein Mena . As such, a single voxel in a TMEM the clinical management of breast cancer patients. Activity-MRI map can contain multiple TMEM doorways which are approximately 40–60 um in diameter (the approximate average diameter of a single TMEM doorway), and the observed signal DISCUSSION intensity is likely the result of cumulative signals from multiple It has long been known that TMEM doorway activity correlates active TMEM doorways in the tumor microenvironment. Despite with increased metastatic potential in preclinical mouse models of the overall lower resolution of MRI, the hyper-intense voxels 4,38,53 breast carcinoma . Clinical investigations have since demon- within a TMEM Activity-MRI map likely correspond to large strated that an increased number of TMEM doorways correlates densities (i.e., hotspots) of active TMEM doorways, given that 2,3,7 with increased metastatic risk in breast cancer patients . background signal from other sources is very low, as explained However, protocols for measuring TMEM doorway activity, and above. However, super-resolution kinetic analysis of TMEM door- therefore metastatic risk, in a non-invasive manner (i.e., without way activity via multiphoton intravital imaging has previously the surgical extraction of the primary tumor or a core biopsy) do demonstrated that TMEM doorways remain open for only ~20 min not exist in the clinical setting. These observations have together before the endothelium is spontaneously re-sealed . It is thus clear inspired the pursuit of a novel, non-invasive tool/assay for the that the hyper-intense voxels within the TMEM Activity-MRI map successful measurement of TMEM doorway activity in breast can capture regions where metastatic dissemination is currently cancer patients, which could theoretically be embedded in active. This measurement, therefore, offers exciting possibilities for standard-of-care clinical practice. In this regard, this study focused physician-based monitoring of the metastatic potential in the era on the algorithmic development and validation of the TMEM of personalized medicine. Activity-MRI assay, a dynamic contrast-enhanced first-pass decon- Metastasis is the primary cause of death in breast cancer, yet no volution MRI approach that measures TAVO events, known to clinically validated imaging modalities are available that reflect the biologically correlate with cancer cell intravasation and dissemina- ability of primary breast cancer to metastasize. Although screening tion in the peripheral circulation. Furthermore, this study provides mammography has contributed to greater detection rates and critical insights for the subsequent transfer of the TMEM Activity- reduced breast cancer mortality, it also results in over-diagnosis or MRI assay to the clinical setting, with the vision of facilitating detection of cancers that pose no threat to life. There is, therefore, treatment decision-making for breast cancer patients undergoing an unmet need to develop screening modalities, used as a primary neoadjuvant treatment. screening test, or a reflex test after initial screening mammo- While the translational potential of the TMEM Activity-MRI assay graphy, that may distinguish non-lethal versus potentially lethal is highlighted in the current research-oriented pilot study, the cancers. Multiparameter gene expression assays, including Onco- 59 ®TM clinical potential of the assay requires its incorporation into the type DX Recurrence Score , MammaPrint , Prosigna™, and standard-of-care MRI exam. Indeed, the proposed TMEM Activity- Breast Cancer Index℠) provide similar prognostic information MRI assay is highly compatible with standard-of-care MRI that is driven largely by proliferation and estrogen-dependent npj Breast Cancer (2022) 101 Published in partnership with the Breast Cancer Research Foundation G.S. Karagiannis et al. genes and not by the intrinsic propensity of tumor cells to presents with any significant translational application in meta- 61–63 metastasize or interact with their microenvironment . Inter- static cancers, as well. estingly, TMEM doorway Score correlates poorly with the While our study has primarily focused on breast cancer, it is now Oncotype DX Recurrence Score, captures different biologic known that other types of cancer, such as pancreatic neuroendo- information, and provides complementary prognostic informa- crine tumors, also utilize TMEM doorways as the primary cancer cell dissemination machinery for the initial steps of the metastatic tion . Thus, the TMEM doorway score, and its highly correlated cascade . Previously, pancreatic tissue and tumors have been TMEM-MRI score, offer the potential to more accurately determine successfully visualized using low or high-resolution imaging prognosis regarding recurrence beyond what is possible for lower modalities, including magnetic resonance imaging and multi- ranges of Oncotype DX and other multiparameter gene expres- 84–88 photon intravital microscopy . It would therefore be extremely sion scores. This could lead to treatment decisions that differ from interesting to extend our studies into other types of cancer and those made based on Multiparameter gene expression investigate whether TMEM activity-MRI or an equivalent MRI- assays alone. based measurement could be universally used as an assessment Another potential clinical application of TMEM Activity-MRI is as tool for metastatic potential. a companion diagnostic to monitor the pharmacodynamic effects In conclusion, this article describes the development and of standard cytotoxic therapy and also therapeutic interventions validation of a novel MRI measurement that correlates with designed to induce a blockade of TMEM doorways. It is known metastatic dissemination in preclinical models of breast carcinoma that cytotoxic chemotherapy has profound effects on the tumor 50,52,64 and breast cancer patients. The technological and conceptual microenvironment , including promoting an influx of proan- 64–69 innovation of this newly proposed measurement, herewith known giogenic M2 macrophages and the formation of TMEM 5,70 as TMEM Activity-MRI, is the quantification of only the first pass of doorways . Chemotherapy may also increase the density of 5,71–73 a gadolinium-based contrast agent (GBCA) into the tumor tissue. cancer cells with high dissemination potential . Therefore, This procedure notably eliminates background signals from other quantifying the effects of chemotherapy on the dissemination sources and highly correlates with the biological activity of cancer potential of the tumor microenvironment could be used as a cell dissemination doorways, known as TMEM doorways, whose pharmacodynamic biomarker to stratify distant metastatic risk in activity is the rate-limiting step of metastatic dissemination. The patients with residual disease after neoadjuvant chemotherapy or TMEM Activity-MRI assay is sensitive to various anti-cancer and adding agents such as the Tie2 kinase inhibitor rebastinib that anti-metastatic therapies (i.e., chemotherapy, TMEM doorway blocks TMEM doorway-mediated cancer cell intravasation. Addi- inhibitors, etc.) and has been successfully utilized here to classify tional studies are required to further evaluate the clinical validity patients into high and low-risk individuals for developing of TMEM Activity-MRI for distant recurrence risk and its potential metastasis as a non-invasive TMEM score surrogate. We propose clinical utility in breast cancer management. that TMEM Activity-MRI could be utilized as a promising All the validation experiments performed in this study have companion diagnostic to facilitate physicians with decision- consistently confirmed that TMEM Activity-MRI significantly making, especially during treatment, as well as with the clinical correlates with endpoints of metastatic dissemination, including management of breast cancer patients. TMEM doorway activity, circulating tumor cells (CTCs), and disseminated tumor cells (DTCs), suggesting that TMEM Activity- MRI can serve as a specific tool to study the biology of cancer cell METHODS dissemination, which is often seen as the rate-limiting step of the Animal subjects 4,38 metastatic cascade . Here, as a proof-of-concept, we documen- Ethics statement. All studies involving mice were carried out in ted that MENA expression is not critical for the function of TMEM accordance with the National Institutes of Health (NIH) regulation doorways, even though it is necessary for the generation of a concerning the care and use of experimental animals and with the proinvasive/promigratory tumor cell subset (Fig. 4). These approval of the Institutional Animal Care and Use Committee (IACUC) of observations foster a subsequent question pertaining to the Molecular Imaging, Inc. (Ann Arbor, MI), a facility accredited by the precise MENA-independent mechanism, via which cytoskeletal Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC), or with the approval of the Albert Einstein College of Medicine dynamics at TMEM doorways may regulate endothelial cell Animal Care and Use Committee. opening to facilitate transendothelial migration. Indeed, other members of the Ena/VASP family could play a key role in this 74,75 MMTV-PyMT (spontaneous model). Transgenic mice expressing the Poly- process and should be explored in the future to fully oma Virus Middle-T (PyMT) antigen under the control of mammary tumor appreciate the complex regulation of TAVO events. Interestingly, virus long terminal repeat (MMTV-LTR) were bred in-house at the Albert TMEM doorways are microenvironmental niches that may Einstein College of Medicine (Condeelis lab), maintained on the FVB cultivate induction and maintenance of stemness and localized background, and the resulting tumors could be palpated at ~6 weeks of 76,77 immunosuppression , indicating they may confer suitable age. Depending on the experimental question, MMTV-PyMT mice were niches for the development of immune-privileged, metastatic used in different age groups. stem cells . The particularly high correlation observed between TMEM Activity-MRI and DTCs (Fig. 3i–k) is supportive of this notion MMTV-PyMT (transplantation model). Syngeneic transplantation models and further indicates that TMEM Activity-MRI may harbor were generated through orthotopic transplantation of 1 mm tumor additional potential in studying the immune tumor chunks from 12–16-week-old MMTV-PyMT donor mice bearing late-stage carcinomas of ~1 cm in diameter into 5–6-week-old FVB recipients. These microenvironment. tumors were not passaged in culture or dissociated but propagated as In this study, TMEM Activity-MRI was only measured in primary 39,43 tumor chunks in vivo . The tumor chunk was implanted on the fourth tumors of mice and humans. However, more recent observations mammary pad on the right side of the recipient mouse. suggest that TMEM doorways do not only assemble in the primary tumor microenvironment but also in local (i.e., lymph nodes) or −/− MMTV-PyMT/ Mena (spontaneous model). Generation of the MMTV- 79–81 distant (i.e., lungs) metastatic sites . These observations raise −/− PyMT Mena mice was achieved by crossing MMTV-PyMT mice with the intriguing possibility that metastatic breast tumors may 43 MENA heterozygotes as described . The forward and reverse primer potentially utilize hematogenous routes, based on the assembly sequences used to identify the transmission of the disrupted MENA allele −/− of TMEM doorways, to efficiently re-disseminate to tertiary sites in MENA mice are the following: LACZ-F: CGATCGTAATCACCCGAGTGT; 80,82,83 after achieving metastatic colonization . As such, future LACZ-R: CCGTGGCCTGACTCATTCC; Enah-3-F: ACCGCAGTCTCCCTTACAT AACTTA; Enah-3-R: GCACTGCACTTTTAATCAGGTGTCT. efforts should establish whether the TMEM Activity-MRI assay Published in partnership with the Breast Cancer Research Foundation npj Breast Cancer (2022) 101 G.S. Karagiannis et al. Patient-derived xenograft (PDX) model. The generation of the estrogen Development of the mouse TMEM activity-MRI assay receptor-negative (ER ) HT17 patient-derived xenograft has been devel- The MRI images were obtained on a 9.4 Tesla Agilent Direct Drive MRI/MRS oped in-house, as described . Similar to the PyMT-transplantation model system. Images were acquired with a 24 mm diameter surface receive-only described above, the HT17 tumors were also never passaged in culture or coil within a 12 cm diameter pin-switch driven volume transmit coil. High- dissociated but only propagated as tumor chunks in vivo (passage resolution T2-weighted images were used to determine the placement of number <5). Each tumor chunk was implanted on the fourth mammary the two slices in the permeability estimation, one for the arterial source pad on the right side of severe combined immunodeficiency (SCID) and one for the tumor slice. The GRE images were collected with a field of recipient mice. The resulting tumors can be palpated 4–6 weeks following view of 23 mm , matrix size of 96 × 128, thickness of 1 mm, image the transplantation. repetition time (TR) of 23 ms, echo time (TE) of 3 ms, flip angle (FA) was 28 degrees, acquisition bandwidth of 100 kHz, for an image temporal resolution of 2 s (signal averages= 1) or 4 s (SA= 2). The samples in the Chemicals and reagents administered to mice dynamic study were sampled every 4 s and for 10 min, with the contrast (gadolinium) being injected at 1 min. Prior to dynamic image acquisition, a Rebastinib reconstitution and administration. Rebastinib was reconstituted T1 measurement was completed by varying the FA between 2 and 60 ms, at a concentration of 10 mg/mL in 0.4% hydroxypropyl methylcellulose to which T1 was fit on a pixel-by-pixel basis, and the low FA data (FA 2, 4, 6, (HPMC). Each mouse in the experimental group received p.o. doses of and 8 degrees) were used along with the dynamic data to fit dynamic T1 to 10 mg/kg rebastinib (100 μL total volume) twice per week, for 4 weeks. The the dynamic curve pixel-by-pixel for determination of CA concentration. control group received p.o. 100 μL of HPMC. During a contrast-enhanced MRI exam, a gadolinium-based contrast agent (GBCA) is typically injected and CA passage into the tumor tissue is Chemotherapy reconstitution and administration. Mice were treated with visualized by comparing image intensity changes pre- and post-contrast the taxane-based chemotherapeutic paclitaxel (Sigma-Aldrich). Paclitaxel (subtraction-based contrast) . Alternatively, high temporal resolution MRI was reconstituted at a concentration of 10 mg/mL in 1:1 EtOH:Cremophor- of the dynamic passage of the GBCA through tissue beds using EL (Millipore). Each mouse in the experimental group received an i.v. dose longitudinal (T1) relaxation-based agents can be used to measure tissue of 10 mg/Kg paclitaxel (200 μL total volume) every 5 days, for a total of permeability to the GBCA. Typically, these contrast agents are T1-based three doses. The control group received an i.v. injection of 200 μL 1:1 agents, for which the change in tissue relaxivity associated with the EtOH:Cremophor-EL. contrast agent’s presence can be used to approximate the agent’s blood and tissue time-dependent concentration. Mathematical deconvolution Clodronate and PBS liposome reconstitution and administration. The methods can be used to extract the tissue response function from the clodronate liposomes were administered in experimental mice with an tissue signal response function using the measured or inferred arterial i.v. injection at a dose of 5 mL/Kg (200 μL total volume) every other day for input concentration function of the contrast agent, the form of which is a total of seven doses. The control group received an equivalent i.v. derived from the differential equation: injection of PBS liposomes. dC ðtÞ (1) V ¼ PSρ C ðÞ t  C ðtÞ ; e p e dt High-molecular weight (155 kDa) dextran reconstitution and administra- where C (t) and C (t) are the GBCA concentration in the blood plasma and tion. 155-kDa Tetramethylrhodamine-Dextran (TMR-Dextran) solution p e tissue, respectively, P is the GBCA endothelial permeability, S is the was reconstituted at a concentration of 20 mg/mL in sterile phosphate- capillary wall surface area, and ρ is the tissue density. buffered saline (PBS). Each mouse in control or experimental groups The product PS is often called the permeability-surface area product, received 100 μL total volume of TMR-Dextran via right-sided retro-orbital which assumes that delivery of the GBCA and perfusion are sufficient to injection, 1-h before sacrifice. ensure that the permeability is the dominant determinant of GBCA exflux into the tissue. Endothelial permeability of the GBCA is usually represented Gadolinium reconstitution and administration. The gadolinium-based as the permeability transfer constant, or k ; trans contrast agent (Magnevist Bayer HealthCare, Pittsburgh, PA) was used in experimental mice at a dose of 0.1 mm/kg, diluted in gadopentetate K ¼ PSρ (2) trans dimeglumine (140 ul total volume), and was administered at 20 ul/sec via If we use the convention that the concentration of GBCA in the imaging tail vein catheter injection with a 50 µl dead space in the line filled of voxel, Ct(t) is given by the volume weighted concentrations of the plasma saline. V and tissue V ; p e C ðÞ t ¼ V C ðÞ t þ V C ðtÞ (3) t p p e e Human subjects Solution of the differential leads to a convolution integral, which can be TMEM-MRI activity was measured in breast cancer patients. This study was simplified by assuming that the uptake of GBCA into the extravascular/ designed by Montefiore-Einstein cancer center investigators, approved by extracellular space is minimal during the measurement. Because the entire the Albert Einstein Institutional Review Board, and conducted in measurement process outlined above is limited to the ‘first pass’ of the accordance with the ethical principles derived from international guide- GBCA, this assumption is valid and simplifies the convolution integral to lines, including the International Council for Harmonisation Good Clinical the following: Practice guidelines, the Declaration of Helsinki, and local regulations on the conduct of clinical research. All the participants provided written informed 0 0 (4) C ðÞ t ¼ k C ðÞ t dt t fp p consent before enrollment. The inclusion criteria were the following: breast mass >1 cm with biopsy-proven histology of invasive breast carcinoma which shows that the Gd concentration in the tissue is modeled as a (any histologic type and ER, PR, HER2 status), age ≥18 years, ECOG convolution of the tissue’s permeability-surface product (k ) and the fp performance status 0–1, ability to undergo MRI with gadolinium integrated GBCA delivery. enhancement, no known or suspected renal impairment, normal organ This estimation is valid during the first pass of the contrast bolas, before and marrow function, weight less than or equal to the MRI table limit, the recirculation of the agent. The use of this first-pass leakage profile has ability to understand and willingness to sign a written informed consent. been shown to give a more accurate estimation of the TMEM-associated The exclusion criteria were the following: prior chemotherapy of radiation vascular opening (TAVO, also shown as k ) compared to the Multi- fp therapy to the ipsilateral breast, breast prosthetic implants (silicone or compartmental Model . saline), use of any investigational agent within 30 days of starting the A baseline T1 map needs to be estimated to allow for the calculation of a study, uncontrolled intercurrent illness including, but not limited to, concentration map in real-time. Multiple GRE images with varying flip ongoing or active infection, symptomatic congestive heart failure, unstable angles (FA) are acquired and Eq. 5, based on the Ernst formula, is fit to give angina pectoris, cardiac arrhythmia, or psychiatric illness/social situations a T1 estimate: that would limit compliance with study, pregnancy, and lactation. The TR=T10 1  e clinical and pathological characteristics of the tumors are summarized in (5) s ¼ mðÞ sinFA TR=T1 1 ðÞ cosFA e Fig. 5h. npj Breast Cancer (2022) 101 Published in partnership with the Breast Cancer Research Foundation G.S. Karagiannis et al. To fit this model for T1 ,FA= (2, 4, 6, 8, 10, 12, 16, 20, 25, 30, 40) degrees panMena antibody (510693; BD Biosciences) to detect Mena-expressing were acquired, with TR = 15 ms and TE = 3.2 ms. Levenburg–Marquart cancer cells. To visualize macrophages, we used anti-IBA1 antibody (019- nonlinear least squares algorithm is used to fit Eq. 5. This gives a baseline 19741; Wako) for mouse and anti-CD68 (MO876; Dako) for human tumors. T1 estimate. To visualize endothelial cells, we used anti-endomucin (SC-65495; Santa After gaining the baseline T1 map, the GBCA is injected and the Cruz) for mouse and anti-CD31 (MO823; Dako) for human tumors. dynamic GRE series (S(t)) of images is collected, with FA= 25 degrees Appropriate areas containing invasive cancer tissue suitable for TMEM sampled every 4 s. A dynamic T1 series (T1(t)) is estimated directly using doorway analysis were identified by low-power scanning using the the following: following criteria: high density of tumor, adequacy of a tumor, lack of necrosis or inflammation, and lack of artifacts such as retraction or folds. 1 1 1  D 2,36 TMEM doorway scoring was performed as previously described . The ¼  ln (6) T1ðÞ t TR 1 ðÞ cos FA D assessment of TMEM doorway scores was performed with Adobe Photoshop on ten high-power (400X) digital images of the most TR=T10 StðÞ Sð0Þ 1  e representative areas of the tumor. The total number of TMEM doorways (7) D ¼ þ TR=T1 0 for each image were tabulated, and the scores from all ten images were m sinFA 1 ðÞ cosFA e summed to give a final TMEM doorway density for each patient sample, The Gd concentration map (C(t)) is then calculated with the following: expressed as the number of TMEM doorways per total area (ten high- 1 1 power [400X] fields). A representative high-power magnification image T1ðÞ t T1 (8) CtðÞ ¼ showing the tripartite TMEM doorway is indicated in Supplementary fig. R1 3a. IBA1 single immunohistochemistry was performed with the IBA1 −1 −1 where the relaxivity of the Gd is R1 = 3.2 s mM . primary antibody used in the TMEM immunohistochemistry, and a To narrow our analysis on vascular leakage elicited by TAVO events, an representative high-power magnification image is indicated in Supple- arterial source must first be found. So, we acquire two slices in our scans, mentary fig. 3b. one through the tumor (C ) and one through the artery (C ). Using Eq. 4 t a with C = C , weighted least squares is used to estimate k with higher p a fp Multichannel immunofluorescence weights given to the points at the top of the C curve. For all multichannel immunofluorescence experiments, slides were first To graphically compare the TMEM-Associated Vascular Opening (TAVO) deparaffinized by melting at 60 C in an oven equipped with a fan for events of mouse breast carcinomas among different conditions, we 20 min, followed by 2X xylene treatment for 20 min. Slides were then calculated the ratio of values above a permeability threshold in the entire rehydrated, and antigen retrieval was performed in 1 mM EDTA (pH 8.0) at histogram (Eq. 9). The optimal threshold for class separation (early versus 97 C for 20 min in a conventional steamer, followed by incubation in a late carcinomas in PyMT mice) was found to be ~0.001. In the following blocking buffer solution (10% FBS, 1% BSA, 0.0025% fish skin gelatin in equation (Eq. 9), H represents the histogram of k values found within the i fp 0.05% PBST) for 60 min at room temperature. Specific considerations were tumor. then considered depending on the assay performed, as described below. :04 :04 X X TMEM Activity  MRI ¼ H = H (9) i i INV MENA immunofluorescence. After standard slide preparation as i¼th i¼0 INV described above, slides were incubated with chicken anti-MENA (0.25 μg/mL, generated in the lab of Dr. John S. Condeelis) in a blocking buffer for 60 min at room temperature. Samples were washed three times in 0.5% PBST and incubated with an HRP-conjugated IgG anti-chicken Development of the human TMEM activity-MRI assay secondary antibody for 60 min at room temperature. After washing, slides TMEM activity was measured in a small preliminary study of breast cancer were incubated with biotinylated tyramide (Perkin Elmer; Opal 4-color patients after consent in accord with an approved IRB protocol. TMEM Fluorescent IHC kit) diluted at 1:50 in amplification buffer for 10 min. After activity was measured on a Philips 3 T Ingenia Elition. Dynamic contrast- washing, slides were incubated with spectral DAPI for 5 min and mounted enhanced (DCE) imaging was conducted following intravenous adminis- with ProLong Gold antifade reagent (Life Technologies). The slides were tration of Gadoterate (0.1 mm/kg) administered as a bolus a power imaged on the Pannoramic 250 Flash II digital whole slide scanner, using a injector. The dynamic contrast-enhanced MRI (DCE-MRI) protocol used for 20 × 0.75NA objective lens. Tissue suitable for scanning was automatically clinical assessment was modified to permit the collection of a T1 detected using intensity thresholding. relaxometry data set prior to CA injection and a dynamic data set collected during the CA injection. The DCE-MRI data set began 1 min prior TMEM activity-dextran assay. Assessment of TMEM-mediated vascular to injection and continued for at least 90 s following injection using a 4D- opening (TAVO) was performed using multichannel-IF in an FFPE section TRAK XD imaging protocol which employed compressed sense (factor 8) with a sequential TMEM triple-IHC section already stained. Each slide was and both keyhole (20%) and half scan (factors 0.625 and 0.878) parallel stained with a primary antibody mixture cocktail against rat anti- imaging providing rapid (3 to 5 s) volume acquisitions. Image volumes 3 endomucin (1:500; sc-65495; Santa Cruz) and rabbit anti-TMR (1:1,000; A- were acquired in rapid succession, with a high-resolution matrix (1 mm 6397; Life Technologies). Slides were then washed three times in 0.05% isotropic or nearly isotropic voxels), and (typically) a TR of 3.9 ms, TE PBST, and incubated with a secondary antibody mixture cocktail, including 1.95 ms, temporal 3D volume resolution of 3.2 to 4.6 s, flip angle 28 donkey anti-rabbit Alexa-488 and goat anti-rat Alexa-568, both at 1:200 degrees, and between 60 and 120 sequential 3D volumes. Imaging dilution for 60 min at room temperature. After washing (0.05% PBST 3X), covered both breasts, as well as the heart, from which an arterial signature slides were incubated with spectral DAPI for 5 min and mounted with of CA uptake was acquired. Pre-contrast and post-contrast conventional 3D ProLong Gold antifade reagent (Life Technologies). The slides were imaged images were acquired after the dynamic images, allowing delineation of on the Pannoramic 250 Flash II digital whole slide scanner, using a the tumor volume. Fitting of the convolution equation followed that 20 × 0.75NA objective lens. Tissue suitable for scanning was automatically described for the animal data, with a similar calculation of the TMEM detected using intensity thresholding. Whole tissue images were uploaded Activity-MRI. in Pannoramic Viewer version 1.15.4 (3DHISTECH). To investigate whether highly permeable blood vessels were associated with TMEM doorways, Histology (H&E) and immunohistochemistry for IBA1 and multiple 40X fields were captured to obtain ~20–25 vascular profiles for TMEM doorways each case. A “vascular profile” was defined as an endomucin blood vessel with clear margins, either longitudinally or in cross-section. Vessels in the After mice were sacrificed, all mammary tumors were extracted and vicinity or continuing away from the field of view were excluded from this immersed in 10% formalin in a volume ratio of tumor to formalin of 1:7. analysis because it was not possible to access the entire perivascular area Tissues were fixed for 24 to 48 h and embedded in paraffin, then processed associated with them. In each image, the endomucin and dextran-TMR for histological examination. One 5 µm section from each tumor was channels were each thresholded just above the background based upon stained for hematoxylin and eosin (H&E) and one for TMEM. The TMEM intensity by using contrast adjustment. For each vascular profile, the doorway assay is a triple-stain IHC for predicting metastatic risk, in which endomucin channel was then used as an exclusion mask to the dextran three antibodies are applied sequentially and developed separately with channel to directly designate an ROI that belonged exclusively to the different chromogens on a Dako Autostainer. TMEM doorway stain was 2 extravascular portion of the dextran. Because IF staining may result in a performed as previously described , except that in this study we used anti- Published in partnership with the Breast Cancer Research Foundation npj Breast Cancer (2022) 101 G.S. Karagiannis et al. certain number of nonspecific “speckles” with a positive TMR signal, we As such, slight deviations can be observed in the total number of mice in used an arbitrary threshold of >20 pixels around a vascular profile to each individual figure panel, accounting for the exclusion criteria applied consider it a “leaky vascular profile”. The sequential TMEM IHC sections in each case. were then used to assess whether these leaky profiles had an associated Transgenic MMTV-PyMT animals, as well as animals transplanted with TMEM structure. To directly compare TAVO-dependent dextran leakage patient-derived tumors (e.g., HT17 xenografts), were housed in cages of among different groups of mice, the extravascular dextran ROI was five animals per the regulations of the Albert Einstein College of Medicine expressed in each image as an area fraction, and an average among all (AECOM) Animal Care and Use Committee. Once the mice reached the ROIs for each mouse was reported. criteria for inclusion into the experimental pipeline (tumors with a diameter of ~2–3 mm), they were randomly allocated to rebastinib-, chemotherapy-, or vehicle-treated groups. Quantification of disseminated tumor cells (DTCs) in mouse lungs. The The two pathologists (M.H.O. and J.G.J.) involved in TMEM scoring were assessment of DTCs in the lungs was performed in the MMTV-PyMT mouse model, based on the availability of commercial anti-PyMT antibodies for blinded to the specific group allocations, as were all the scientists performing CTC scoring, all IF/IHC analyses, and all MRI feature specific detection of PyMT-expressing tumor cells, indicative of this mouse quantifications. Importantly, the TMEM Activity-MRI maps, upon which model of breast carcinoma. Each lung section was stained with primary TMEM-MRI-Activity is determined, was conducted prior to and without antibody mixture cocktail against mouse anti-pancytokeratin (PanCK; knowledge of the TMEM pathology score (and scored by different 1:1,000; C2562; Sigma) and rat anti-PyMT (PyMT; 1:400; NB100-2749; Nobus individuals) and therefore was blinded. Biologicals). Slides were then washed three times in 0.05% PBST and incubated with a secondary antibody mixture cocktail, including goat anti- mouse Alexa-488 and donkey anti-rat Alexa-568, both at 1:200 dilution for 60 min at room temperature. After washing (0.05% PBST 3X), slides were DATA AVAILABILITY incubated with spectral DAPI for 5 min and mounted with ProLong Gold The original contributions presented in the study are included in the article and antifade reagent (Life Technologies). The slides were imaged on the supplementary material. Further inquiries can be directed to the corresponding Pannoramic 250 Flash II digital whole slide scanner, using a 20 × 0.75NA authors. No datasets were generated or analyzed during the current study. objective lens. Tissue suitable for scanning was automatically detected using intensity thresholding. Whole tissue images were uploaded in Received: 5 October 2021; Accepted: 11 July 2022; Pannoramic Viewer version 1.15.4 (3DHISTECH). DTCs were identified as + + + single PanCK PyMT DAPI cells and expressed as mean (i.e., DTCs per high-power field) for each mouse. Intravasation (circulating tumor cell) assay REFERENCES The in vivo intravasation assay was performed as previously 1. Harney, A. S. et al. Real-time imaging reveals local, transient vascular perme- 5,39,90,91 described . Mice were anesthetized with isoflurane, and blood was ability, and tumor cell intravasation stimulated by TIE2hi macrophage-derived taken from the right ventricle by heart puncture, using 25 G needles VEGFA. Cancer Discov. 5, 932–943 (2015). coated with heparin. Erythrocytes were lysed using 10 ml of 1X RBC lysis 2. Robinson, B. D. et al. Tumor microenvironment of metastasis in human breast buffer (multi-species, eBioscience). The samples were centrifuged at carcinoma: a potential prognostic marker linked to hematogenous dissemination. 200 × g for 5 min, then cell pellets were reconstituted in 10 ml of Clin. Cancer Res. 15, 2433–2441 (2009). Dulbecco’s modified Eagle medium (DMEM/F-12) supplemented with 3. Rohan, T. E. et al. Tumor microenvironment of metastasis and risk of distant 20% fetal bovine serum (FBS) and plated in 10-cm Petri dishes. After cells metastasis of breast cancer. J. Natl Cancer Inst. https://doi.org/10.1093/jnci/ were attached, single tumor cells were counted. The total number of cells dju136 (2014). counted was divided by the volume of blood taken. Tumor cells counted 4. Karagiannis, G. S., Goswami, S., Jones, J. G., Oktay, M. H. & Condeelis, J. S. Sig- were either CFP- or Dendra2-positive, thus confirming their identity as natures of breast cancer metastasis at a glance. J. Cell Sci. 129, 1751–1758 (2016). tumor cells. As a negative control, blood from a non-tumor-bearing mouse 5. Karagiannis, G. S. et al. Neoadjuvant chemotherapy induces breast cancer was analyzed, and the complete absence of epithelial cells was confirmed. metastasis through a TMEM-mediated mechanism. Sci. Transl. Med. https:// doi.org/10.1126/scitranslmed.aan0026 (2017). 6. Harney, A. S. et al. The selective Tie2 inhibitor rebastinib blocks recruitment and Statistical analysis function of Tie2(Hi) macrophages in breast cancer and pancreatic neuroendo- All statistical analyses, including graph/plot generation and statistical crine tumors. Mol. Cancer Ther. 16, 2486–2501 (2017). hypothesis testing, were performed in GraphPad Prism 8 software. The 7. Sparano, J. A. et al. A metastasis biomarker (MetaSite Breast Score) is associated variables in our study (TMEM score, TMEM Activity-MRI, TMEM Activity- with distant recurrence in hormone receptor-positive, HER2-negative early-stage Dextran, circulating tumor cells per mL blood, disseminated tumor cells) breast cancer. Nat. PJ Breast Cancer 3, 42 (2017). were all continuous and, as such, presented as dot-plots with means and 8. Kienast, Y. et al. Real-time imaging reveals the single steps of brain metastasis their corresponding Standard Deviation (SD). Statistical comparisons formation. Nat. Med. 16, 116–122 (2010). between two independent groups were performed using a two-tailed 9. Tanaka, K. et al. In vivo real-time imaging of chemotherapy response on the liver Mann–Whitney U-test, and those between more than two independent metastatic tumor microenvironment using multiphoton microscopy. Oncol. Rep. groups were performed using Kruskal–Wallis one-way analysis of variance. 28, 1822–1830 (2012). For longitudinal analyses, repeated-measures ANOVA was used. Statistical 10. Provenzano, P. P., Eliceiri, K. W. & Keely, P. J. Multiphoton microscopy and correlations were performed by Spearman’s correlation, with the data fluorescence lifetime imaging microscopy (FLIM) to monitor metastasis and the presented as scatterplots with fit lines and their corresponding P values. tumor microenvironment. Clin. Exp. Metastasis 26, 357–370 (2009). Mouse studies were conducted in at least three independent animal 11. Bonapace, L. et al. If you don’t look, you won’t see: intravital multiphoton imaging cohorts, which were pooled together for statistical analysis. of primary and metastatic breast cancer. J. Mammary Gland Biol. Neoplasia 17, Sample size calculations for all the validation studies were performed 125–129 (2012). using the following parameters: significance level (adjusted for sidedness) 12. Yan, J. et al. Use of multiphoton microscopy to diagnose liver cancer and lung of 0.025, probability of type II error of 0.2 (i.e., statistical power 0.8), and metastasis in an orthotopic rat model. Scanning 34, 271–277 (2012). expected difference in means equal to 1.5 SD units, based on the 13. Entenberg, D. et al. Time-lapsed, large-volume, high-resolution intravital imaging assumption that the SD of the response variable was 1 unit. Based on the for tissue-wide analysis of single cell dynamics. Methods 128,65–77 (2017). above, a total of 18 animals was calculated to enter the study. The 14. Entenberg, D. et al. In vivo subcellular resolution optical imaging in the lung experiments were initially designed to include >18 animals in each study, reveals early metastatic proliferation and motility. Intravital 4,1–11 (2015). to account for mice that died during an MRI session, mice that fitted one or 15. Entenberg, D. et al. A permanent window for the murine lung enables high- more exclusion criteria, and finally for statistical outliers. The following resolution imaging of cancer metastasis. Nat. Methods 15,73–80 (2018). animals were removed from the study based on exclusion criteria: Mice 16. Gligorijevic, B. et al. Intravital imaging and photoswitching in tumor invasion and intravasation microenvironments. Micros Today 18,34–37 (2010). whose tumors were overly necrotic or cystic; mice with inadequate tumor 17. Patsialou, A. et al. Intravital multiphoton imaging reveals multicellular streaming tissue to perform IHC or IF; mice with less than 0.5 ml of blood collected as a crucial component of in vivo cell migration in human breast tumors. Intra- through heart puncture for the CTC assay; and mice that were either obese vital 2, e25294 (2013). (>35 g), or emaciated (<17 g), upon reaching the experimental endpoints. npj Breast Cancer (2022) 101 Published in partnership with the Breast Cancer Research Foundation G.S. Karagiannis et al. 18. Alexander, S., Koehl, G. E., Hirschberg, M., Geissler, E. K. & Friedl, P. Dynamic 46. Pignatelli, J. et al. Invasive breast carcinoma cells from patients exhibit MenaINV- imaging of cancer growth and invasion: a modified skin-fold chamber model. and macrophage-dependent transendothelial migration. Sci. Signal 7, ra112 Histochem. Cell Biol. 130, 1147–1154 (2008). (2014). 19. Beerling, E., Ritsma, L., Vrisekoop, N., Derksen, P. W. & van Rheenen, J. Intravital 47. Pignatelli, J. et al. Macrophage-dependent tumor cell transendothelial migration microscopy: new insights into metastasis of tumors. J. Cell Sci. 124, 299–310 (2011). is mediated by Notch1/MenaINV-initiated invadopodium formation. Sci. Rep. 6, 20. Karreman, M. A. et al. Correlating intravital multi-photon microscopy to 3D 37874 (2016). electron microscopy of invading tumor cells using anatomical reference points. 48. Roussos, E. T. et al. Mena invasive (MenaINV) promotes multicellular streaming PLoS ONE 9, e114448 (2014). motility and transendothelial migration in a mouse model of breast cancer. J. Cell 21. Sahai, E. et al. Simultaneous imaging of GFP, CFP and collagen in tumors in vivo Sci. 124, 2120–2131 (2011). using multiphoton microscopy. BMC Biotechnol. 5, 14 (2005). 49. Oudin, M. J. et al. MenaINV mediates synergistic cross-talk between signaling 22. Sidani, M., Wyckoff, J., Xue, C., Segall, J. E. & Condeelis, J. Probing the micro- pathways driving chemotaxis and haptotaxis. Mol. Biol. Cell 27, 3085–3094 (2016). environment of mammary tumors using multiphoton microscopy. J. Mammary 50. Karagiannis, G. S., Condeelis, J. S. & Oktay, M. H. Chemotherapy-induced metas- Gland Biol. Neoplasia 11, 151–163 (2006). tasis: mechanisms and translational opportunities. Clin. Exp. Metastasis 35, 23. Soon, L., Braet, F. & Condeelis, J. Moving in the right direction-nanoimaging in 269–284 (2018). cancer cell motility and metastasis. Microsc Res. Tech. 70, 252–257 (2007). 51. Karagiannis, G. S., Condeelis, J. S. & Oktay, M. H. Chemotherapy-induced metas- 24. Wyckoff, J. et al. A paracrine loop between tumor cells and macrophages is tasis in breast cancer. Oncotarget 8, 110733–110734 (2017). required for tumor cell migration in mammary tumors. Cancer Res. 64, 7022–7029 52. Karagiannis, G. S., Condeelis, J. S. & Oktay, M. H. Chemotherapy-induced metas- (2004). tasis: molecular mechanisms, clinical manifestations, therapeutic interventions. 25. Wyckoff, J. B. et al. Direct visualization of macrophage-assisted tumor cell intra- Cancer Res. 79, 4567–4576 (2019). vasation in mammary tumors. Cancer Res. 67, 2649–2656 (2007). 53. Oktay, M. H. & Jones, J. G. TMEM: a novel breast cancer dissemination marker for 26. Harney, A. S., Wang, Y., Condeelis, J. S. & Entenberg, D. Extended time-lapse the assessment of metastatic risk. Biomark. Med. 9,81–84 (2015). intravital imaging of real-time multicellular dynamics in the tumor micro- 54. Ioannidis, G. S., Maris, T. G., Nikiforaki, K., Karantanas, A. & Marias, K. Investigating environment. J. Vis. Exp. https://doi.org/10.3791/54042 (2016). the correlation of Ktrans with semi-quantitative MRI parameters towards more 27. Lohrke, J. et al. 25 years of contrast-enhanced MRI: developments, current robust and reproducible perfusion imaging biomarkers in three cancer types. IEEE challenges and future perspectives. Adv. Ther. 33,1–28 (2016). J. Biomed. Health Inf. 23, 1855–1862 (2019). 28. Bar, A., Skorka, T., Jasinski, K. & Chlopicki, S. MRI-based assessment of endothelial 55. Thakran, S. et al. Characterization of breast lesion using T1-perfusion magnetic function in mice in vivo. Pharm. Rep. 67, 765–770 (2015). resonance imaging: qualitative vs. quantitative analysis. Diagn. Inter. Imaging 99, 29. Li, C. H. et al. Flow versus permeability weighting in estimating the forward 633–642 (2018). volumetric transfer constant (K(trans)) obtained by DCE-MRI with contrast agents 56. Cheng, Z. et al. Discrimination between benign and malignant breast lesions of differing molecular sizes. Magn. Reson. Imaging 36, 105–111 (2017). using volumetric quantitative dynamic contrast-enhanced MR imaging. Eur. 30. Li, K. L., Zhu, X. P., Waterton, J. & Jackson, A. Improved 3D quantitative mapping Radio. 28, 982–991 (2018). of blood volume and endothelial permeability in brain tumors. J. Magn. Reson. 57. Patlak, C. S., Blasberg, R. G. & Fenstermacher, J. D. Graphical evaluation of blood- Imaging 12, 347–357 (2000). to-brain transfer constants from multiple-time uptake data. J. Cereb. Blood Flow. 31. Guy, C. T., Cardiff, R. D. & Muller, W. J. Induction of mammary tumors by Metab. 3,1–7 (1983). expression of polyomavirus middle T oncogene: a transgenic mouse model for 58. Tofts, P. S. et al. Estimating kinetic parameters from dynamic contrast-enhanced metastatic disease. Mol. Cell Biol. 12, 954–961 (1992). T(1)-weighted MRI of a diffusable tracer: standardized quantities and symbols. J. 32. Lin, E. Y. et al. Progression to malignancy in the polyoma middle T oncoprotein Magn. Reson. Imaging 10, 223–232 (1999). mouse breast cancer model provides a reliable model for human diseases. Am. J. 59. Sparano, J. A. et al. Prospective validation of a 21-gene expression assay in breast Pathol. 163, 2113–2126 (2003). cancer. N. Engl. J. Med. 373, 2005–2014 (2015). 33. Fantozzi, A. & Christofori, G. Mouse models of breast cancer metastasis. Breast 60. Sparano, J. A. & Solin, L. J. Defining the clinical utility of gene expression assays in Cancer Res. 8, 212 (2006). breast cancer: the intersection of science and art in clinical decision making. J. 34. Enden, T. et al. Visualization of deep veins and detection of deep vein thrombosis Clin. Oncol. 28, 1625–1627 (2010). (DVT) with balanced turbo field echo (b-TFE) and contrast-enhanced T1 fast field 61. Paik, S. Is gene array testing to be considered routine now? Breast 20, S87–S91 echo (CE-FFE) using a blood pool agent (BPA). J. Magn. Reson. Imaging 31, (2011). Suppl 3. 416–424 (2010). 62. Sparano, J. A., Fazzari, M. & Kenny, P. A. Clinical application of gene expression 35. Lim, H. K. et al. MR diagnosis of facial neuritis: diagnostic performance of profiling in breast cancer. Surgical Oncol. Clin. North Am. 19, 581–606 (2010). contrast-enhanced 3D-FLAIR technique compared with contrast-enhanced 3D- 63. Wirapati, P. et al. Meta-analysis of gene expression profiles in breast cancer: T1-fast-field echo with fat suppression. Am. J. Neuroradiol. 33, 779–783 (2012). toward a unified understanding of breast cancer subtyping and prognosis sig- 36. Roussos, E. T. et al. Mena invasive (Mena(INV)) and Mena11a isoforms play distinct natures. Breast Cancer Res. 10, R65 (2008). roles in breast cancer cell cohesion and association with TMEM. Clin. Exp. 64. De Palma, M. & Lewis, C. E. Macrophage regulation of tumor responses to Metastasis 28, 515–527 (2011). anticancer therapies. Cancer Cell 23, 277–286 (2013). 37. Arwert, E. N. et al. A unidirectional transition from migratory to perivascular 65. Hughes, R. et al. Perivascular M2 macrophages stimulate tumor relapse after macrophage is required for tumor cell intravasation. Cell Rep. 23, 1239–1248 chemotherapy. Cancer Res. 75, 3479–3491 (2015). (2018). 66. De Palma, M., Murdoch, C., Venneri, M. A., Naldini, L. & Lewis, C. E. Tie2-expressing 38. Borriello, L. et al. The role of the tumor microenvironment in tumor cell intra- monocytes: regulation of tumor angiogenesis and therapeutic implications. vasation and dissemination. Eur. J. Cell Biol. 99, 151098 (2020). Trends Immunol. 28, 519–524 (2007). 39. Patsialou, A. et al. Selective gene-expression profiling of migratory tumor cells 67. Lewis, C. E., De Palma, M. & Naldini, L. Tie2-expressing monocytes and tumor in vivo predicts clinical outcome in breast cancer patients. Breast Cancer Res. 14, angiogenesis: regulation by hypoxia and angiopoietin-2. Cancer Res. 67, R139 (2012). 8429–8432 (2007). 40. Karagiannis, G. S. et al. Assessing tumor microenvironment of metastasis 68. Roodhart, J. M. et al. Late release of circulating endothelial cells and endothelial doorway-mediated vascular permeability associated with cancer cell dissemina- progenitor cells after chemotherapy predicts response and survival in cancer tion using intravital imaging and fixed tissue analysis. J. Vis. Exp. https://doi.org/ patients. Neoplasia 12,87–94 (2010). 10.3791/59633 (2019). 69. Squadrito, M. L. & De Palma, M. Macrophage regulation of tumor angiogenesis: 41. Dasgupta, A., Lim, A. R. & Ghajar, C. M. Circulating and disseminated tumor cells: implications for cancer therapy. Mol. Asp. Med. 32, 123–145 (2011). harbingers or initiators of metastasis? Mol. Oncol. 11,40–61 (2017). 70. Chang, Y. S., Jalgaonkar, S. P., Middleton, J. D. & Hai, T. Stress-inducible gene Atf3 42. Bidard, F. C., Proudhon, C. & Pierga, J. Y. Circulating tumor cells in breast cancer. in the noncancer host cells contributes to chemotherapy-exacerbated breast Mol. Oncol. 10, 418–430 (2016). cancer metastasis. Proc. Natl Acad. Sci. USA 114, E7159–E7168 (2017). 43. Roussos, E. T. et al. Mena deficiency delays tumor progression and decreases 71. Janni, W. J. et al. Pooled analysis of the prognostic relevance of circulating tumor metastasis in polyoma middle-T transgenic mouse mammary tumors. Breast cells in primary breast cancer. Clin. Cancer Res. 22, 2583–2593 (2016). Cancer Res. 12, R101 (2010). 72. Riethdorf, S. et al. Prognostic impact of circulating tumor cells for breast cancer 44. Weidmann, M. D. et al. MenaINV dysregulates cortactin phosphorylation to patients treated in the neoadjuvant “Geparquattro” trial. Clin. Cancer Res. 23, promote invadopodium maturation. Sci. Rep. 6, 36142 (2016). 5384–5393 (2017). 45. Eddy, R. J., Weidmann, M. D., Sharma, V. P. & Condeelis, J. S. Tumor cell inva- 73. Bidard, F. C. et al. Circulating tumor cells in breast cancer patients treated by dopodia: invasive protrusions that orchestrate metastasis. Trends Cell Biol. 27, neoadjuvant chemotherapy: a meta-analysis. J. Natl Cancer Inst. 110, 560–567 595–607 (2017). (2018). Published in partnership with the Breast Cancer Research Foundation npj Breast Cancer (2022) 101 G.S. Karagiannis et al. 74. Furman, C. et al. Ena/VASP is required for endothelial barrier function in vivo. J. Myles P. Dempsey. Views, opinions of, and endorsements by the authors do not Cell Biol. 179, 761–775 (2007). reflect those of the US Army or the Department of Defense. All experiments involving 75. Kris, A. S., Kamm, R. D. & Sieminski, A. L. VASP involvement in force-mediated rebastinib administration were conducted in part using rebastinib that was kindly adherens junction strengthening. Biochem Biophys. Res. Commun. 375, 134–138 provided by Deciphera Pharmaceuticals. The authors would finally like to thank the (2008). personnel of the Histopathology Facility and Analytical Imaging Facility at the Albert 76. Sanchez, L. R. et al. The emerging roles of macrophages in cancer metastasis and Einstein College of Medicine for constant support and assistance with the use of the response to chemotherapy. J. Leukoc. Biol. 106, 259–274 (2019). facilities. 77. Asiry, S. et al. The cancer cell dissemination machinery as an immunosuppressive niche: a new obstacle towards the era of cancer immunotherapy. Front. Immunol. 12, 654877 (2021). AUTHOR CONTRIBUTIONS 78. Oskarsson, T., Batlle, E. & Massague, J. Metastatic stem cells: sources, niches, and Developed the project concept and designed experiments: G.S.K., A.B., C.A.B., and vital pathways. Cell Stem Cell 14, 306–321 (2014). J.S.C. Contributed equally as first authors: G.S.K. and A.B. Executed animal 79. Entenberg, D. et al. A protocol for the implantation of a permanent window for experiments and corresponding data analyses: G.S.K., A.B., L.R.S., K.A., M.-H.C., Y.W., high-resolution imaging of the murine lung. Protocol. Exch. https://doi.org/ A.S.H., J.M.P., Y.L., X.C., J.G.J., D.E., M.H.O., C.A.B., and J.S.C. Performed human studies 10.1038/protex.2017.134 (2017). and corresponding data analyses: G.S.K., A.B., J.M.A., S.A., D.H., L.J.H., T.Q.D., J.A.S., 80. Coste, A. et al. Hematogenous dissemination of breast cancer cells from lymph M.H.O., C.A.B., and J.S.C. Wrote manuscript: G.S.K., A.B., M.H.O., C.A.B., and J.S.C. nodes is mediated by tumor microenvironment of metastasis (TMEM) doorways. Revised and approved the final manuscript: All authors. Front. Oncol. (in the press). 81. Ginter, P. S. et al. Tumor microenvironment of metastasis (TMEM) doorways are restricted to the blood vessel endothelium in both primary breast cancers and COMPETING INTERESTS their lymph node metastases. Cancers https://doi.org/10.3390/cancers11101507 The authors declare no competing interests. (2019). 82. Brown, M. et al. Lymph node blood vessels provide exit routes for metastatic tumor cell dissemination in mice. Science 359, 1408–1411 (2018). ADDITIONAL INFORMATION 83. Pereira, E. R. et al. Lymph node metastases can invade local blood vessels, exit the node, and colonize distant organs in mice. Science 359, 1403–1407 (2018). Supplementary information The online version contains supplementary material 84. Holmberg, D. & Ahlgren, U. Imaging the pancreas: from ex vivo to non-invasive available at https://doi.org/10.1038/s41523-022-00463-5. technology. Diabetologia 51, 2148–2154 (2008). 85. Cao, L., Kobayakawa, S., Yoshiki, A. & Abe, K. High resolution intravital imaging of Correspondence and requests for materials should be addressed to Craig A. Branch subcellular structures of mouse abdominal organs using a microstage device. or John S. Condeelis. PLoS ONE 7, e33876 (2012). 86. Alieva, M., Ritsma, L., Giedt, R. J., Weissleder, R. & van Rheenen, J. Imaging win- Reprints and permission information is available at http://www.nature.com/ dows for long-term intravital imaging: General overview and technical insights. reprints Intravital 3, e29917 (2014). 87. Coste, A., Oktay, M. H., Condeelis, J. S. & Entenberg, D. Intravital imaging tech- Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims niques for biomedical and clinical research. Cytometry A https://doi.org/10.1002/ in published maps and institutional affiliations. cyto.a.23963 (2019). 88. Morse, B. & Klapman, J. Imaging of pancreatic tumors. Monogr. Clin. Cytol. 26, 21–33 (2020). 89. Mann, R. M., Kuhl, C. K. & Moy, L. Contrast-enhanced MRI for breast cancer Open Access This article is licensed under a Creative Commons screening. J. Magn. Reson. Imaging 50, 377–390 (2019). Attribution 4.0 International License, which permits use, sharing, 90. Wyckoff, J. B., Jones, J. G., Condeelis, J. S. & Segall, J. E. A critical step in metastasis: adaptation, distribution and reproduction in any medium or format, as long as you give in vivo analysis of intravasation at the primary tumor. Cancer Res. 60, 2504–2511 appropriate credit to the original author(s) and the source, provide a link to the Creative (2000). Commons license, and indicate if changes were made. The images or other third party 91. Boimel, P. J. et al. Contribution of CXCL12 secretion to invasion of breast cancer material in this article are included in the article’s Creative Commons license, unless cells. Breast Cancer Res. 14, R23 (2012). indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http:// ACKNOWLEDGEMENTS creativecommons.org/licenses/by/4.0/. This research was supported by NIH K99 CA237851; T32 CA200561; Department of Defense (W81XWH-13-1-0010); CA216248; S10 OD019961 for the use of the Perkin Elmer 250 slide scanner; the Gruss-Lipper Biophotonics Center and its Integrated © The Author(s) 2022 Imaging Program; the Evelyn Gruss-Lipper Charitable Foundation, and Jane A. and npj Breast Cancer (2022) 101 Published in partnership with the Breast Cancer Research Foundation http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png npj Breast Cancer Springer Journals

Loading next page...
 
/lp/springer-journals/assessment-of-mri-to-estimate-metastatic-dissemination-risk-and-NV9xzOb0aq

References (105)

Publisher
Springer Journals
Copyright
Copyright © The Author(s) 2022
eISSN
2374-4677
DOI
10.1038/s41523-022-00463-5
Publisher site
See Article on Publisher Site

Abstract

www.nature.com/npjbcancer ARTICLE OPEN Assessment of MRI to estimate metastatic dissemination risk and prometastatic effects of chemotherapy 1,2,3,4,15 5,15 6,7 8,9 8,9 George S. Karagiannis , Anthony Bianchi , Luis Rivera Sanchez , Kamal Ambadipudi , Min-Hui Cui , 10 11 3,4,6 3,4,6 2,7 3,11 12 Jesus M. Anampa , Saeed Asiry , Yarong Wang , Allison S. Harney , Jessica M. Pastoriza , Yu Lin , Xiaoming Chen , 6,11 2,3,4,11 13 8 8 10 Joan G. Jones , David Entenberg , Dana Haddad , Laura J. Hodges , Timothy Q. Duong , Joseph A. Sparano , 2,3,4,11 3,8,9,14 2,3,4,7,12 ✉ ✉ Maja H. Oktay , Craig A. Branch and John S. Condeelis Metastatic dissemination in breast cancer is regulated by specialized intravasation sites called “tumor microenvironment of metastasis” (TMEM) doorways, composed of a tumor cell expressing the actin-regulatory protein Mena, a perivascular macrophage, and an endothelial cell, all in stable physical contact. High TMEM doorway number is associated with an increased risk of distant metastasis in human breast cancer and mouse models of breast carcinoma. Here, we developed a novel magnetic resonance imaging (MRI) methodology, called TMEM Activity-MRI, to detect TMEM-associated vascular openings that serve as the portal of entry for cancer cell intravasation and metastatic dissemination. We demonstrate that TMEM Activity-MRI correlates with primary tumor TMEM doorway counts in both breast cancer patients and mouse models, including MMTV-PyMT and patient-derived xenograft models. In addition, TMEM Activity-MRI is reduced in mouse models upon treatment with rebastinib, a specific and potent TMEM doorway inhibitor. TMEM Activity-MRI is an assay that specifically measures TMEM-associated vascular opening (TAVO) events in the tumor microenvironment, and as such, can be utilized in mechanistic studies investigating molecular pathways of cancer cell dissemination and metastasis. Finally, we demonstrate that TMEM Activity-MRI increases upon treatment with paclitaxel in mouse models, consistent with prior observations that chemotherapy enhances TMEM doorway assembly and activity in human breast cancer. Our findings suggest that TMEM Activity-MRI is a promising precision medicine tool for localized breast cancer that could be used as a non-invasive test to determine metastatic risk and serve as an intermediate pharmacodynamic biomarker to monitor therapeutic response to agents that block TMEM doorway-mediated dissemination. npj Breast Cancer (2022) 8:101 ; https://doi.org/10.1038/s41523-022-00463-5 1,5,6 INTRODUCTION the lungs (and other secondary sites), as well as metastases . Consistent with these findings, TMEM doorway density in the Cancer cell dissemination occurs through specialized intravasation primary tumor has been validated as an independent prognostic portals on blood vessels called Tumor Microenvironment of biomarker for distant recurrence in human breast cancer in three Metastasis (TMEM) doorways . TMEM doorways consist of a 2,3,7 independent cohorts, including ~1150 patients , and thus may perivascular macrophage, a tumor cell overexpressing the actin- serve as a biomarker for distinguishing potentially lethal from non- regulatory protein Mammalian-enabled (MENA), and an endothe- 2–4 lethal cancers. lial cell, all in direct physical contact with each other . Cancer cell Consistent with these findings, TMEM doorway density in the intravasation at TMEM doorways occurs during tightly regulated primary tumor has been associated with distant recurrence in transient localized vascular opening events triggered by the TMEM human breast cancer in three independent cohorts, including doorway macrophage, which expresses the angiopoietin receptor 2,3,7 ~1150 patients , and thus may serve as a biomarker for TIE2. Upon stimulation, TIE2 macrophages at TMEM doorways distinguishing potentially lethal from non-lethal cancers. In the secrete vascular endothelial growth factor-A (VEGFA), leading to first proof-of-concept study involving 30 case-control pairs of the localized disruption of the underlying endothelial adherens patients with and without distant recurrence, TMEM doorway and tight junctions and, as a consequence, the opening of the density was significantly higher in patients with recurrence vessel wall (TMEM-associated vascular opening, or TAVO), localized vascular leakiness, and subsequent tumor cell transen- (P = 0.00006) . In a subsequent prospective validation in 259 dothelial migration and intravasation . As such, increased TMEM case-control pairs with and without distant recurrence from a doorway density and activity in the primary tumor microenviron- population-based cohort, TMEM doorway density was likewise associated with an increased risk of distant metastasis in the ment have been associated with an increased incidence of circulating tumor cells (CTCs), disseminated tumor cells (DTCs) in subset of those with hormone receptor-positive, HER2-negative 1 2 Department of Microbiology & Immunology, Albert Einstein College of Medicine, Bronx, NY, USA. Tumor Microenvironment and Metastasis Program, Albert Einstein Cancer 3 4 Center, Bronx, NY, USA. Integrated Imaging Program, Albert Einstein College of Medicine, Bronx, NY, USA. Gruss-Lipper Biophotonics Center, Albert Einstein College of 5 6 Medicine, Bronx, NY, USA. California State University, Bakersfield, CA, USA. Department of Anatomy and Structural Biology, Albert Einstein College of Medicine, Bronx, NY, USA. 7 8 Department of Surgery, Montefiore Medical Center, Albert Einstein College of Medicine, Bronx, NY, USA. Department of Radiology, Montefiore Medical Center, Albert Einstein 9 10 College of Medicine, Bronx, NY, USA. Gruss Magnetic Resonance Research Center, Albert Einstein College of Medicine, Bronx, NY, USA. Department of Medical Oncology, Montefiore Medical Center, Albert Einstein College of Medicine, Bronx, NY, USA. Department of Pathology, Montefiore Medical Center, Albert Einstein College of 12 13 Medicine, Bronx, NY, USA. Department of Cell Biology, Albert Einstein College of Medicine, Bronx, NY, USA. Maediclinic Middle East, Department of Breast Imaging, Dubai, 14 15 United Arab Emirates. Department of Physiology and Biophysics, Albert Einstein College of Medicine, Bronx, NY, USA. These authors contributed equally: George S. Karagiannis, Anthony Bianchi. email: craig.branch@einsteinmed.edu; john.condeelis@einsteinmed.edu Published in partnership with the Breast Cancer Research Foundation 1234567890():,; G.S. Karagiannis et al. breast cancer (P trend = 0.004), but not in triple-negative or HER2- typical approach for clinical assessment simply compares one or positive breast cancer. In the second prospective validation several post-GBCA images to a single pre-GBCA image, looking for cohort, including 600 patients from a clinical trial cohort treated features reflecting high contrast agent passage associated with with adjuvant chemotherapy, proportional hazards models tissue pathology . revealed a significant positive association between continuous One method that has been employed to estimate permeability TMEM doorway density score and early distant recurrence is k , which derives a bulk transfer constant (a rate constant) trans (P = 0.001) and locoregional plus distant recurrence between intravascular and extravascular spaces from a combina- (P = 0.00006) within 5 years of diagnosis in the subset of 297 tion of permeability and surface area . Its estimation requires patients with hormone receptor-positive, HER2-negative disease, measurements of both the arterial input and the tissue response but not in triple-negative or HER2-positive breast cancer. TMEM to the GBCA at a high temporal rate for several minutes. This doorway density score correlated poorly with the 21-gene method would allow measurement of TAVO by separating the Recurrence Score (r = 0.29) and was significantly prognostic for extravascular transfer of GBCA that does not reflect typical early locoregional and distant recurrence (p = 0.05) in multivariate permeability from the bolus efflux of plasma-borne GBCA into models including tumor size, grade, nodal metastasis, and the 21- the tissue space. When many TMEM doorways exist within a tissue gene Recurrence Score, with a trend toward first distant region, their opening is stochastic and temporally independent, recurrence (p = 0.10). Finally, TMEM doorway density did not which would appear in contrast-enhanced MRI as an overall significantly correlate with tumor size or nodal status, and increase in tissue permeability with individual TAVO events compared with hormone receptor-positive, HER2-negative dis- obscured. Thus, a dynamic measure of the tissue permeability, ease, was significantly higher in triple-negative or HER2-positive one which allows rates of efflux to be distinguished, may provide a breast cancer (P = 0.001 and P = 0.003, respectively), breast cancer means to detect TMEM doorways via this increased efflux rate. subtypes associated with higher recurrence rate. The totality of Further restriction to the initial time period post-GBCA adminis- the data, therefore, suggests a strong positive association tration may further enhance discrimination of TMEM doorway between TMEM density in the primary tumor and breast cancer activity. We, therefore, implemented a ‘limited first pass’ assess- recurrence. ment of the GBCA transfer rate between intravascular and Assessment of cancer cell dissemination and metastasis in extravascular spaces, calculated using only data from the initial preclinical models is currently performed using histological bolus of the GBCA uptake, leading to the ability to select a range endpoints, such as the number and localization of disseminated of transfer rates from the rate-histogram more likely to represent tumor cells at metastatic sites, as well as contextual changes in the 30 TMEM doorway activity, especially in highly permeable tissues . tumor microenvironment. Although such methods provide some In this study, we demonstrate that such a measure herein termed degree of morphological and molecular information, they are “TMEM Activity-MRI”, corresponds to the extent of TAVO events. limited by the need to euthanize the animal and thus lack We do not only demonstrate that TMEM Activity-MRI can be evidence regarding the dynamic nature of the phenomena they utilized in basic and translational settings to answer questions describe. Furthermore, only a small portion of the tumor is related to the molecular mechanism of cancer cell dissemination interrogated with the limited tissue sectioning typically used for but also in the clinical setting as a newly proposed and non- the assessment of histological endpoints. The ability to observe invasive companion diagnostic for breast cancer patients. cancer cell dissemination in situ, over short or long periods of time, and without the need for terminal procedures, has the potential to make a tremendous addition to our understanding of RESULTS spatiotemporal changes in the tumor microenvironment, asso- TMEM activity-MRI measurement correlates with TMEM ciated with cancer cell dissemination and metastasis. Indeed, such doorway-associated vascular opening (TAVO) events in vivo observations have been now made possible through the To measure TAVO events in tumors using magnetic resonance emergence of sophisticated imaging modalities in live animals, 8–26 imaging (MRI), we developed an algorithm (Materials and such as multiphoton intravital imaging . Although these Methods) to generate a dynamic contrast-enhanced first-pass techniques offer great potential in basic research, they cannot deconvolution MRI map for mouse primary breast carcinomas. As be used clinically because they require genomic incorporation of discussed above, prior evidence indicates that tumor cell artificially created fluorescent transgenes or direct injection of intravasation occurs exclusively in association with localized fluorescent reporters in the test subjects, and they also have poor vascular openings at TMEM doorways . To establish the TMEM depth of imaging in whole tissues. To circumvent this problem, we Activity-MRI method, we utilized a previously established trans- have investigated Magnetic Resonance Imaging (MRI) as an in vivo genic mouse model of spontaneous breast carcinoma, the Mouse versatile and non-invasive imaging tool to simultaneously Mammary Tumor Virus (MMTV) Polyoma Middle-T Antigen (PyMT) measure multiple tissue properties (e.g., structural, functional, model (Supplementary fig. 1), also known as MMTV-PyMT (or and metabolic) associated with TMEM activity and metastatic simply PyMT), which recapitulates human breast cancer develop- dissemination in breast cancer. 31–33 ment and progression in a clinically relevant fashion . In a clinical setting, contrast-enhanced MRI is commonly used MMTV-PyMT mice bearing spontaneous breast tumors were for the detection and characterization of breast cancer. During a subjected to a dynamic MRI protocol using a 9.4 T 31 cm Agilent contrast-enhanced MRI exam, a Gadolinium-based contrast agent Direct Drive imaging system. A baseline T1 map was acquired (GBCA) is typically injected, and the passage of GBCA into the using the variable flip angle (FA) approach and gradient recalled tumor parenchyma is visualized by comparing image intensity 34,35 echo (GRE) imaging (Fig. 1a), similar to previous publications , changes pre- and post-contrast injection, also known as with slight modifications as described in Materials and Methods. subtraction-based contrast . Alternatively, high temporal resolu- To delineate regions of interest (ROIs) that correspond to the tion MRI of the dynamic passage of GBCA through tissue beds tumor mass, the GRE image with a FA of 12 is used as it delineates using longitudinal (T1) relaxation-based imaging can be used to the tumor tissue versus the normal tissue with the highest measure the tissue’s permeability to the GBCA. Typically, GBCA contrast compared to the other images, and it facilitates the affects MRI signal intensity in direct relation to its concentration anatomical demarcation of the tumor margins (Supplementary within tissues, allowing for mathematical determination of its rate of exchange across the capillary. Endothelial permeability of GBCA Fig. 2a–d). A dynamic GRE sequence is initiated, and the can thus be determined directly from the observed arterial input gadolinium-based contrast agent (GBCA) is injected after 60s and tissue responses of the MRI dynamic data. However, the using a standard dose of 140 μL of 0.1 mmol/Kg gadopentetate npj Breast Cancer (2022) 101 Published in partnership with the Breast Cancer Research Foundation 1234567890():,; G.S. Karagiannis et al. Fig. 1 Development of TMEM activity-MRI assay. a T1 map of PyMT late carcinoma tumor (Left). A black arrow pointing at a potential blood/ necrotic region. Example voxel intensity over varying flip angle (Right). This graph is fitto Eq. 5 at every voxel to calculate the baseline T1 map. b Dynamic GRE images over time post-contrast injection (at 60 s). Yellow arrow, a small amount of visual change over time. c Gadolinium- based contrast agent (GBCA) concentration map over time, calculated using the baseline T1 map and dynamic GRE using Eq. 8. Yellow arrow, very slow leak area and most likely necrotic, showing up as small in the K image, indicating necrotic regions may have low K values. d GBCA fp fp concentration of the arterial source (left), where the first pass of the contrast agent is the first peak, and the second smaller peak suggests GBCA recirculation. An example voxel from within the tumor with fast uptake of contrast agent (right). Note the scale difference with the arterial source. e Final TMEM Activity-MRI map calculated using Eq. 4. Orange arrow, hyper-intense voxel, i.e., larger value in K is indicative of fp increased TAVO events. f Experimental design and mouse cohort composition. MMTV mouse mammary tumor virus, PyMT Polyoma Middle-T antigen, EC early carcinoma, LC late carcinoma. g TMEM identification by triple-stain immunohistochemistry (IHC) and representative images from early (EC) and late (LC) carcinoma MMTV-PyMT mice. Scale = 100um h Quantification of TMEM doorways (TMEM doorway score), assessed in 10 high-power fields (HPFs) in mice shown in (g). Mann–Whitney U-test. I Frequency histogram of the combined Early (EC) and Late (LC) Carcinoma mouse cohorts. The threshold is representative of creating the TMEM Activity-MRI Eq. 9. The optimal threshold was calculated to be ~0.001. j Representative TMEM Activity-MRI maps of mouse tumors by magnetic resonance imaging (MRI) from early (EC) and late (LC) carcinoma MMTV-PyMT mice. Hyper-intense voxels correspond to TMEM hotspots (i.e., increased number of TAVO events). Scale = 1 mm. k Quantification of TAVO events (a.k.a. TMEM activity), assessed via the TMEM Activity-MRI assay, in mice shown in (j). Mann–Whitney U-test. l Correlation between TMEM score (as quantified in h) and TMEM Activity-MRI (as quantified in k) in early and late carcinomas of the MMTV-PyMT mice shown in (j). Spearman’s rank correlation coefficient. Error bars: standard deviation (SD). Published in partnership with the Breast Cancer Research Foundation npj Breast Cancer (2022) 101 G.S. Karagiannis et al. dimeglumine at a rate of 20 µL/sec (Fig. 1b). Using the baseline T1 specifically assessing the tumor compartment that contains the map and the dynamic GRE sequence, a GBCA dynamic leakiest blood vessels in the entire tumor. In addition, TMEM concentration map is then calculated from the relaxivity map, as Activity-MRI avoids potential bias coming from tumor size previously reported , giving an estimate of GBCA concentration variations since it gives the percentage, and not the absolute for each individual voxel (Fig. 1c). From the dynamic concentration value, of the voxels in a given tumor that has the highest image series, the arterial source is identified (Fig. 1d; left panel), permeability. Upon quantification, it became evident that TMEM and an arterial input concentration curve is subsequently Activity-MRI was significantly (p < 0.05; Mann–Whitney U-test) determined. An example region of the tumor with fast uptake is higher in the LC compared to the EC cohort in MMTV-PyMT mice shown in the pipeline (Fig. 1d; right panel). Importantly, finding (Fig. 1j, k). In support of this, TMEM Activity-MRI significantly the arterial source also allows for the normalization among (p < 0.05; Spearman rho = 0.55; Spearman’s rank correlation) different subjects (Materials and Methods). As a next step, the time correlates with TMEM doorway score (Fig. 1l), suggesting a series of each voxel is fit to equation #4 (Materials and Methods) positive correlation between TMEM Activity-MRI and tumor to generate the corresponding TMEM Activity-MRI map (Fig. 1e). progression-dependent increases of TMEM doorway number. Importantly, high TMEM doorway activity is reflected by voxels Taken together, observations shown in Fig. 1 suggest that the with hyperintensities in the TMEM Activity-MRI map (Fig. 1e; newly developed TMEM Activity-MRI measurement can success- orange arrow) due to large endothelial openings. Voxels with low fully capture changes in TMEM doorway score, for instance, the intensity in the TMEM Activity-MRI map likely reflect inactive increase in TMEM doorway number during the progression from TMEM doorways or regions devoid of TMEM doorways. Necrotic or early- to late-stage carcinoma) in a (patho)physiologically relevant non-perfused regions appear as either hyper-intense or hypo- preclinical model of mammary carcinoma. intense voxels on T1W images, depending on whether fast- (open vasculature, Fig. 1a; black arrow) or slow-contrast-uptake (avas- Clodronate-mediated reduction of TMEM doorways decreases cular) regions (Fig. 1c, yellow arrow) are present respectively, thus TMEM Activity-MRI both yielding little or no permeability in the first-pass GBCA TMEM doorways are dynamic structures, and as such, TMEM uptake (i.e., TMEM Activity-MRI) maps (Fig. 1e). doorway activity oscillates in time . Multiphoton intravital imaging MMTV-PyMT mice, which spontaneously develop breast carci- studies in live animal tumors suggest that a higher density of noma, have been shown to form an increasing number of active TMEM doorways in the tumor microenvironment proportionally TMEM doorways throughout tumor progression . Therefore, a corresponds to a higher probability of TAVO occurrence . Based comparison between early (EC) and late (LC) carcinomas not only on this premise, along with data shown in Fig. 1j–l, we reasoned represents an excellent way to study TAVO events and cancer cell that targeted suppression of TMEM doorway formation should dissemination but also an excellent model to develop a sensitive overall reduce TMEM Activity-MRI, because fewer TMEM doorways MRI measurement corresponding to TMEM-dependent contrast would provide fewer TAVO events per voxel. To suppress TMEM agent leakage, as described in the pipeline above (Fig. 1a–e). To doorway formation, we adopted the treatment with clodronate this end, we first generated an MMTV-PyMT mouse cohort (Fig. 1f), liposomes, a pharmacologic macrophage depletion strategy that which included mice belonging to age groups corresponding to reduces TMEM doorways in mice . Clodronate liposomes or early carcinoma (EC; 6–8-week old, N = 9) and late carcinoma (LC; vehicle control (i.e., PBS liposomes) were administered for two 10–13-week old, N = 14), as described previously . The individual weeks in 7-week-old PyMT mice with palpable tumors, and, after mouse tumors were also examined histologically in a retrospective the completion of treatment, mice were subjected to TMEM manner (i.e., upon tumor resection following an MRI session) to Activity-MRI, TMEM score, and circulating tumor cells (CTC) confirm early- or late-stage carcinoma status by histopathology measurements (Fig. 2a). As a positive control, clodronate-treated (Supplementary Fig. 1). As expected and also reported pre- 1,36 mice presented with significantly (p < 0.01; Mann–Whitney U-test) viously , TMEM doorway assembly was significantly (p < 0.01; fewer macrophages overall, as assessed by IBA1 immunohisto- Mann–Whitney U-test) increased in LC compared to EC samples chemistry (Fig. 2b, c and Supplementary Fig. 3b). Because (Fig. 1g, h and Supplementary Fig. 3a). Because TMEM activity is macrophages are integral components of the TMEM doorway cell increased in late-stage compared to early-stage PyMT carcinomas, 4,38 triads , we also confirmed significantly (p < 0.05; Mann–Whitney we anticipated that such differences should be reflected in the U-test) fewer TMEM doorways upon clodronate treatment (Fig. 2d, TMEM Activity-MRI maps, as developed above (Fig. 1a–e). To look e), in line with the clodronate-mediated macrophage depletion into this possibility, all PyMT breast tumor images were manually (Fig. 2b, c). Despite the significant depletion of macrophages in segmented from the TMEM Activity-MRI images, and their clodronate-treated mice, we did not observe any significant corresponding histograms were pooled for each group together changes in histological features between vehicle- and clodronate- and graphed (Fig. 1i). In this analysis, k rates less than 0 and fp −3 treated animals (Supplementary Fig. 1). Importantly, however, above 40 × 10 were masked off as error voxels (refer to Materials TMEM Activity-MRI was significantly (p < 0.05; Mann–Whitney U- and Methods). This histogram analysis suggested that late test) reduced in the clodronate-treated mice (Fig. 2f, g), indicating carcinomas had a lower frequency of hypo-intense voxels that the reduction of TMEM doorways via elimination of tumor- (0.0–0.001) but a higher frequency of hyper-intense voxels associated macrophages may affect the number of TAVO events (0.001–0.04) compared to early carcinomas (Fig. 1i), which was per voxel. Clodronate treatment slightly affected tumor growth at consistent with our hypothesis of an expected pattern of the endpoint (Supplementary Fig. 4a–c), but this did not bias the increased TMEM doorway-dependent vascular opening during measurement of TMEM Activity-MRI, because the variable breast cancer progression. represents a ratio of hyper-intense voxels to the total number of To establish easily interpretable and biologically relevant tumor voxels and thus is unaffected by tumor size. Moreover, endpoints for statistical comparisons, we next calculated an MRI TMEM Activity-MRI correlated significantly (p < 0.05; Spearman feature by combining the histogram analysis (Fig. 1i) from the rho = 0.84; Spearman’s rank correlation) with TMEM score (Fig. TMEM Activity-MRI maps with permeability thresholding (Fig. 1i; 2h), suggesting a linear/proportional correlation between TMEM black line). The calculated MRI measurement, simply termed score and TMEM Activity-MRI, as we hypothesized. As mentioned “TMEM Activity-MRI,” is calculated from the corresponding TMEM above, an increased number of TMEM scores in the tumor Activity-MRI map and represents the ratio of the number of tumor voxels presenting with a permeability score above a certain microenvironment is expected to correlate with an increased threshold divided by the number of total voxels within the tumor probability of TAVO events and, as such, cancer cell dissemination. ROI. As such, TMEM Activity-MRI signifies a suitable metric for We thus finally measured circulating tumor cells (CTCs) and found npj Breast Cancer (2022) 101 Published in partnership with the Breast Cancer Research Foundation G.S. Karagiannis et al. Fig. 2 TMEM activity-MRI is suppressed, as a result of reduced TMEM doorway formation, using clodronate-mediated macrophage depletion. a Experimental strategy and mouse cohort composition for MMTV-PyMT mice subjected to clodronate-dependent macrophage depletion. PyMT Polyoma Middle-T antigen, CTC circulating tumor cell. b Identification of macrophages by IBA1 immunohistochemistry and representative images from MMTV-PyMT mice, treated with either control or clodronate liposomes. Scale = 100 um. c Quantification of IBA1 macrophages, as averaged in 10 high-power fields (HPFs) in MMTV-PyMT mice shown in (b). Mann–Whitney U-test. d TMEM identification by triple-stain immunohistochemistry (IHC) and representative images from MMTV-PyMT mice, treated with either control or clodronate liposomes. Scale = 100 um. e Quantification of TMEM doorways (TMEM score), as assessed in 10 high-power fields (HPFs), in MMTV-PyMT mice shown in (d). Mann–Whitney U-test. f Representative TMEM Activity-MRI maps of mammary carcinoma tumors by magnetic resonance imaging (MRI) from MMTV-PyMT mice treated with clodronate (second column) or vehicle control (first column). Scale = 1 mm. g Quantification of TMEM-mediated vascular opening events (TMEM doorway activity), assessed via TMEM Activity-MRI assay in MMTV- PyMT mice shown in (f). Mann–Whitney U-test. h Correlation between TMEM score (as quantified in e) and TMEM Activity-MRI (as quantified in g) in MMTV-PyMT mice treated with either control or clodronate liposomes. Spearman’s rank correlation coefficient. i Circulating tumor cell (CTC) counts in MMTV-PyMT mice treated with either control or clodronate liposomes. Mann–Whitney U-test. j Correlation of TMEM activity-MRI score with circulating tumor cells (CTCs) in MMTV-PyMT mice, treated with either control or clodronate liposomes. Spearman’srankcorrelation coefficient. Error bars: standard deviation (SD). that clodronate-mediated macrophage depletion resulted in a detect rebastinib-mediated suppression of TMEM activity. We used two independent mouse models of breast carcinoma, both significant (p < 0.05; Mann–Whitney U-test) reduction of CTCs of which received either rebastinib or vehicle control, to alter (Fig. 2i). In line with this, TMEM Activity-MRI also correlated TMEM doorway activity. The first model was developed via significantly (p < 0.05; Spearman rho = 0.82; Spearman’s rank syngeneic transplantation of PyMT tumors from late-stage PyMT correlation) with the number of CTCs (Fig. 2j). The data presented donors into wild-type FVB hosts (Fig. 3a). The second model was in this section collectively suggest that TMEM Activity-MRI developed via xenogeneic transplantation of patient-derived HT17 assesses TMEM-dependent metastatic dissemination. In this tumor chunks [previously established from an estrogen receptor- section, clodronate treatment was not performed in experimental negative (ER-) breast cancer patient ] into immunocompromised mice for purposes of proposing clinical intervention but only for SCID hosts (Fig. 3a′). Both animal models have been detailed purposes of indirectly suppressing TMEM doorway assembly and previously in studies involving TMEM-dependent cancer cell activity and confirming that such suppression could be detected 1,5,36,39 dissemination and metastasis . Following the development in the newly established TMEM Activity-MRI assay. of palpable tumors in both models (4–6 weeks after transplanta- tion), mice received a 3-week treatment with rebastinib or vehicle Rebastinib inhibition of TMEM doorway function decreases [administration protocol detailed in ref. ], followed by an MRI TMEM Activity-MRI session and subsequent measurement of TMEM Activity-MRI and Prior studies have demonstrated that TMEM doorways induce other metastatic endpoints (Fig. 3a, a′). Histological assessment of localized and transient vascular opening associated with tumor the resected PyMT and HT17 tumors, as expected, revealed cell intravasation, which are both tightly regulated by the TIE2 features of late-stage carcinomas (Supplementary Fig. 1), consis- macrophage at TMEM doorways . Indeed, TMEM doorway activity tent with prior observations . Importantly, TMEM Activity-MRI was can be suppressed by targeting the TIE2 signaling pathway in significantly (p < 0.05; Mann–Whitney U-test) reduced in both the 5,6 perivascular macrophages . The pharmacological suppression of PyMT (Fig. 3b, c) and the HT17 (Fig. 3b, c) rebastinib-treated mice the TIE2 signaling pathway in macrophages at TMEM doorways compared to vehicle-treated mice. It should be underscored that can be achieved by the specific TIE2 inhibitor, rebastinib, and at rebastinib, as opposed to the clodronate treatment experiments doses which demonstrate minimal off-target effects based on in described above (Fig. 2), specifically affects the function, but not vivo-relevant cellular assays measuring inhibition of kinase the assembly of TMEM doorways in the breast tumor microenvir- 5,6 5,6 activity . To examine if TMEM Activity-MRI is indeed indicative onment . Indeed, the TMEM doorway score remained unaltered of TAVO events, we tested the TMEM Activity-MRI’s ability to between vehicle- and rebastinib-treated animals for both the Published in partnership with the Breast Cancer Research Foundation npj Breast Cancer (2022) 101 G.S. Karagiannis et al. PyMT and the HT17 models (Supplementary Fig. 5a, b). Interest- reduction of TMEM doorway activity in rebastinib-treated mice. ingly, in the absence of an extrinsic factor (e.g., rebastinib) capable This observation further indicates that TMEM Activity-MRI better of modulating TMEM doorway activity (as, for example, in the mirrors the activity, and not as much the assembly, of TMEM transition from early to late breast carcinoma where TMEM doorways in the tumor microenvironment. doorway activity increases with TMEM doorway score), TMEM activity-MRI correlates well with TMEM doorway score (Fig. 1l). TMEM Activity-MRI correlates with established endpoints of However, upon treatment of breast carcinomas with rebastinib, TMEM doorway-associated vascular opening and metastatic we did not observe any correlation between TMEM score and dissemination TMEM activity-MRI in either of the two models tested (Supple- Previously, we have developed a multichannel immunofluores- mentary Fig. 5c). This observation was completely expected since cence assay to specifically visualize and quantify localized TAVO rebastinib is known to inhibit TMEM doorway activity without events in breast cancer . This assay, here termed “TMEM Activity- affecting the assembly of new, or the breakdown of existing, TMEM doorways . As a consequence, these observations collec- Dextran,” is based on the intravenous injection of high-molecular- tively show that TMEM Activity-MRI detects the significant weight (155 kDa) dextran conjugated to tetramethylrhodamine npj Breast Cancer (2022) 101 Published in partnership with the Breast Cancer Research Foundation G.S. Karagiannis et al. Fig. 3 TMEM activity-MRI corresponds to metastatic dissemination endpoints in a mouse model of spontaneous breast carcinoma (MMTV-PyMT; a–k) and a breast cancer patient-derived xenograft (HT17; a′–h′), and is suppressed by the specific TMEM doorway inhibitor, rebastinib. a–a′ Experimental strategy and mouse cohort composition for syngeneic (A) and xenogeneic (A′) mouse models of breast carcinoma. PyMT polyoma middle-T antigen, CTC circulating tumor cell, SCID severe combined immunodeficiency, FVB friend virus B. b–b′ Representative TMEM Activity-MRI maps of mammary carcinoma tumors by magnetic resonance imaging (MRI) from PyMT (B) and HT17 (B′) mice treated with rebastinib (second column) or vehicle control (first column). Scale = 1 mm. c–c′ Quantification of TMEM-mediated vascular opening events (TMEM doorway activity), assessed via TMEM Activity-MRI assay, in PyMT (c) and HT17 (c′) mice. Mann–Whitney U-test. d–d′ Binarized images of extravascular dextran in mammary carcinoma tumors from PyMT (d) and HT17 (d′) mice treated with rebastinib (second column) or vehicle control (first column). Scale = 20 um. e-e′ Quantification of TAVO events (a.k.a. TMEM activity), assessed via the TMEM activity-dextran assay, in PyMT (e) and HT17 (e′) mice. Mann–Whitney U-test. f–f′ Correlation of TMEM doorway activity scores, as quantified with the TMEM activity-dextran and TMEM activity-MRI assays, in PyMT (f) and HT17 (f′) mice, treated with either rebastinib or vehicle control. Spearman’s rank correlation coefficient. g–g′ Circulating tumor cell (CTC) counts in PyMT (g) and HT17 (g′) mice treated with either rebastinib or vehicle control. Mann–Whitney U-test. h–h′ Correlation of TMEM activity-MRI score with circulating tumor cells (CTCs) in PyMT (h) and HT17 (h′) mice, treated with either rebastinib or vehicle control. Spearman’s rank correlation coefficient. i Multichannel immunofluorescence of disseminated tumor cells (DTCs) in MMTV-PyMT mice treated with either rebastinib (right panel) or vehicle control (left panel), as assessed by co-staining of the PyMT antigen, Pancytokeratin (PanCK) and DAPI. Magnified inserts show individual fluorescent channels for the cells outlined with squared boxes in the main images. Scale = 40 um. j Quantification of disseminated tumor cells (DTCs) in MMTV-PyMT mice, treated with either rebastinib or vehicle control. Mann– Whitney U-test. k Correlation of TMEM activity-MRI score with disseminated tumor cells (DTCs) in PyMT mice, treated with either rebastinib or vehicle control. Spearman’s rank correlation coefficient. Error bars: standard deviation (SD). (TMR) in experimental mice a few minutes before the termination factors, such as the dwell time of CTCs in the circulation resulting of the experiment. In intact blood vessels, the fluorescent probe is from prolonged survival and evasion of CTCs from innate/adaptive 41,42 restrained in the vascular lumens, because its molecular weight immunity . prevents it from passing between the endothelial cells . However, Finally, we assessed the correlation of the newly described TME under circumstances where endothelial cell tight junctions are Activity-MRI measurement with the most direct metastatic dissolved as a result of TMEM doorway activity , the fluorescent dissemination outcome, the presence of disseminated tumor cells probe can leak into the tumor tissue and be visualized and (DTCs) in secondary sites, in particular the lungs. DTCs were quantified as a high TMR signal accumulation in the perivascular detected as single cells in the lung parenchyma, co-expressing space . Here, we adapted the TMEM Activity-Dextran assay by pancytokeratin (PanCK ), which is a generic epithelial marker, and (following the termination of the MRI session) injecting Dextran- the PyMT antigen (PyMT ), which is specific to cancer cells in the TMR directly into the tail vein and sacrificing the mice 15’ later MMTV-PyMT model (Fig. 3i). As expected, rebastinib treatment (Fig. 3a-a′ and Supplementary Fig. 6a). It should be noted that the results in significantly (p < 0.01; Mann–Whitney U-test) fewer DTCs appearance of vascular profiles, as well as the baseline values of in the lung parenchyma of PyMT mice (Fig. 3j). In support to this the TMEM Activity-Dextran assay in PyMT mice following an MRI finding, TMEM Activity-MRI significantly correlated (p < 0.01; session (adapted protocol with the addition of GBCA), are similar Spearman rho = 0.9; Spearman’s rank correlation) with DTCs in to those of PyMT mice not subjected to an MRI session [published this model (Fig. 3k). Taken together, the data presented in this protocol without the addition of GBCA as in ref. ] (Supplemen- section show that TMEM Activity-MRI is an accurate measure of tary fig. 6a, b). Moreover, no extravascular dextran staining is TMEM doorway function during dissemination and indicative of detected in tissue sections of mice receiving GBCA without the the prometastatic potential of breast carcinomas. dextran injection (Supplementary Figs. 6c), indicating the pre- sence of GBCA in the peripheral circulation of the experimental TMEM activity-MRI reveals that mammalian-enabled (MENA) is mice does not interfere with the endpoint measurement of the not essential for TMEM doorway-associated vascular opening TMEM Activity-Dextran assay. Importantly, we noticed that TMEM but is necessary for cancer cell dissemination Activity-Dextran was significantly (p < 0.05; Mann–Whitney U-test) So far, we demonstrated that TMEM Activity-MRI is a potentially suppressed in rebastinib-treated compared to vehicle-treated useful MRI-based measurement of TMEM doorway activity by PyMT (Fig. 3d, e) and HT17 (Fig. 3d′,e′) mice. In support of this using diverse mouse models of breast carcinoma with perturba- observation, TMEM Activity-MRI significantly (p < 0.05; Spearman’s tions in either the TMEM doorway number (Fig. 2) or TMEM rank correlation) correlated with TMEM Activity-Dextran in both doorway function/activity (Fig. 3). Next, we sought to investigate PyMT (Spearman rho = 0.69) and HT17 (Spearman rho = 0.74) whether this novel TMEM Activity-MRI assay could be used in models (Fig. 3f, f′), further strengthening the notion that TMEM mechanistic studies related to the molecular basis of cancer cell Activity-MRI captures TMEM doorway activity within the tumor dissemination. microenvironment. Amid the two main prerequisites of cancer cell dissemination, TMEM doorway function results not only in the localized and i.e., the presence of TMEM doorway and the presence of a transient vascular opening but also in the intravasation of highly proinvasive/promigratory cancer cell subpopulation, prior data invasive, highly migratory tumor cells into the peripheral have shown that the invasive isoform of the actin-regulatory circulation . Consistent with these observations, the administra- INV 4,38 protein MENA, MENA , is critical for the latter prerequisite . tion of rebastinib in both PyMT (Fig. 3g) and HT17 (Fig. 3g′) mice Indeed, PyMT mice lacking the MENA gene (MENA-KO) fail to significantly (p < 0.01; Mann–Whitney U-test) decreased the 43 INV establish metastatic disease . Moreover, MENA is necessary for number of CTCs in the peripheral circulation. Importantly, TMEM the transendothelial migration of tumor cells since it regulates the Activity-MRI significantly (p < 0.05; Spearman’s rank correlation) development and maturation of invadopodia, which are essential correlated with CTCs in both the PyMT (Spearman rho = 0.54) and 44–47 the HT17 (Spearman rho = 0.46) models (Fig. 3h, h′). The cytoplasmic protrusions for the migratory/invasive process . correlation of TMEM Activity-Dextran with TMEM Activity-MRI is Nevertheless, it is not clear whether MENA simply defines the stronger than that of CTCs with TMEM Activity-MRI (compare Fig. invasive properties of the prometastatic cancer cell subpopulation 3f, f′ with Fig. 3h, h′). This is in consistent with the fact that the or whether it is also necessary for TMEM doorway activity (Fig. 4a). number of CTCs in the peripheral circulation does not only The newly developed TMEM Activity-MRI assay is ideal to answer depend on TAVO events but may also be determined by other this question, as TMEM Activity-MRI is specific for measuring TAVO Published in partnership with the Breast Cancer Research Foundation npj Breast Cancer (2022) 101 G.S. Karagiannis et al. Fig. 4 TMEM activity-MRI demonstrates that MENA is not necessary for TMEM-associated vascular opening events but is essential for cancer cell dissemination. a Experimental hypothesis under investigation using the TMEM Activity-MRI assay. The illustration displays two prerequisites of metastatic dissemination: TMEM doorway [(a cell triad composed of a macrophage (green), a tumor cell (blue), and an endothelial cell (red)], and the invasive/migratory tumor cell subset migrating alongside macrophages towards underlying TMEM doorways. Although it is known from prior literature (see text for details) that MENA is necessary for inducing invasive and migratory behavior to the prometastatic tumor cells, it is not known if MENA is also necessary for TMEM-associated vascular opening (TAVO), as depicted with the +/+ −/− question mark. b Experimental strategy and mouse cohort composition for MMTV-PyMT (MENA and MENA ) mice. PyMT polyoma INV +/+ middle-T antigen, CTC circulating tumor cell. c Representative images of MENA immunofluorescence from MENA (MENA-WT) and −/− +/+ −/− MENA (MENA-KO) MMTV-PyMT mice. Scale = 50 um. d Circulating tumor cell (CTC) counts in MENA (MENA-WT) and MENA (MENA- KO) MMTV-PyMT mice. Mann–Whitney U-test. e Representative TMEM Activity-MRI maps of mammary carcinoma tumors by magnetic +/+ −/− resonance imaging (MRI) from MENA (MENA-WT) (first column) and MENA (MENA-KO) (second column) MMTV-PyMT mice. Scale = 1 mm. f Quantification of TMEM-mediated vascular opening events (TMEM doorway activity), as assessed via the TMEM Activity-MRI +/+ −/− assay, in MENA (MENA-WT) and MENA (MENA-KO) MMTV-PyMT mice. Mann–Whitney U-test. g Binarized (thresholded) images of +/+ −/− extravascular dextran in mammary carcinoma tumors from MENA (MENA-WT) and MENA (MENA-KO) MMTV-PyMT mice. +/+ Scale = 200 um. h Quantification of TAVO events (a.k.a. TMEM activity), as assessed via the TMEM Activity-dextran assay, in MENA −/− (MENA-WT) and MENA (MENA-KO) MMTV-PyMT mice. Mann–Whitney U-test. Error bars: standard deviation (SD). events and TMEM doorway activity. In this regard, we crossed Activity-MRI could be generally used to investigate the molecular MMTV-PyMT mice with MENA heterozygotes to develop MMTV- mechanisms behind cancer cell intravasation and endothelial −/− PyMT MENA (herewith referred to as MENA-KO mice), while the permeability during metastasis. +/+ MMTV-PyMT MENA (MENA-WT) litter served as the wild-type control (Fig. 4b and Supplementary Fig. 1). We subjected 7–9- Translational relevance of TMEM activity-MRI: Potential utility week-old MENA-KO and MENA-WT mice bearing an average as a companion diagnostic diameter of PyMT tumors of ~0.5 cm, to TMEM Activity-MRI assay Finally, we examined if TMEM activity-MRI could be extended into and experimental endpoints of metastatic dissemination (Fig. 4b). the preclinical setting to provide any insights into its potential Retrospectively, we confirmed that the resected MENA-KO tumors clinical importance. Our group has previously demonstrated that INV did not express the prometastatic MENA isoform using treatment with neoadjuvant paclitaxel or doxorubicin/cyclopho- immunofluorescence (Fig. 4c), and, as expected from our prior 43 sphamide can significantly increase TMEM doorway assembly and work , displayed significant (p < 0.05; Mann–Whitney U-test) activity as a result of the infiltration of prometastatic macrophages suppression of CTCs, compared to MENA-WT mice (Fig. 4d). in both mouse and human breast cancer . Such modifications in Surprisingly, we did not observe any difference (p >0.05; the tumor microenvironment are capable of delaying tumor Mann–Whitney U-test) in TMEM Activity-MRI between MENA-WT growth in the short term but otherwise obfuscate the long-term and MENA-KO mice (Fig. 4e, f), implying that MENA is not directly clinical benefits of chemotherapy treatment. These may also involved in the regulation of TMEM doorway activity, but may contribute to the observed distant relapse following treatment exclusively contribute to the establishment of the proinvasive/ 50–52 with chemotherapy in some patients . Because not all breast promigratory cancer cell subpopulation that disseminates via TMEM 46–49 cancer patients respond with the development of the aforemen- doorways .Toconfirm that the genetic elimination of the MENA tioned prometastatic macrophage infiltration upon treatment with gene in MMTV-PyMT mice did not interfere with the MRI assay giving chemotherapy, we have previously indicated the importance of false-negative results, we also evaluated these observations by using developing non-invasive approaches for monitoring the tumor TMEM Activity-Dextran in the same animals that were imaged with microenvironment while patients undergo pre-operative che- MRI (Fig. 4g, h). Similarly, there was no significant (p >0.05; 50–52 motherapy . We reasoned that TMEM activity-MRI is a Mann–Whitney U-test) difference in TMEM Activity-Dextran between promising tool for this purpose because it correlates with TMEM MENA-WT and MENA-KO mice (Fig. 4g, h), suggesting that TMEM doorway function in all preclinical models tested thus far (Figs. Activity-MRI is indeed specific for measuring TAVO events. Overall, these data served a dual purpose. Foremost, they indicate that MENA 1–3), and importantly, TMEM Activity-MRI can be acquired in a is notnecessary forTMEM doorway-dependent vascular opening, non-invasive manner. To examine if TMEM Activity-MRI can despite it being necessary for metastatic dissemination (Fig. 4a). capture chemotherapy-mediated changes in TMEM doorway Second, these data provide an accurate proof of principle that TMEM function in breast tumors, we again utilized the HT17 npj Breast Cancer (2022) 101 Published in partnership with the Breast Cancer Research Foundation G.S. Karagiannis et al. Fig. 5 Translational significance of the newly developed TMEM activity-MRI assay. a Experimental strategy and mouse cohort composition for HT17 breast cancer patient-derived xenograft. SCID severe combined immunodeficiency. b Representative TMEM activity-MRI maps of mammary carcinoma tumors by magnetic resonance imaging (MRI) in HT17 mice treated with vehicle control (Ctrl; left panel), paclitaxel chemotherapy (Ptx; middle panel), or paclitaxel plus rebastinib (Reb + Ptx; right panel). Scale = 1 mm. c Quantification of TAVO events (a.k.a. TMEM activity), assessed via the TMEM activity-MRI assay, in HT17 mice shown in (b). Kruskal–Wallis analysis of variance with Mann–Whitney U- test for post hoc analysis. d Circulating tumor cells (CTCs) in HT17 Mice treated with chemotherapy alone or chemotherapy in combination with rebastinib. Kruskal–Wallis analysis of variance with Mann–Whitney U-test for post hoc analysis. e Experimental strategy and mouse cohort composition for HT17 breast cancer patient-derived xenograft. SCID, severe combined immunodeficiency. Black arrowheads indicate time- points, in which TMEM activity-MRI was assessed. f Representative TMEM activity-MRI maps of mammary carcinoma tumors by magnetic resonance imaging (MRI) in HT17 mice treated with vehicle control (first row), paclitaxel chemotherapy (second row), or paclitaxel plus rebastinib (third row) across multiple time-points (first column, day 0; second column, day 5; third column, day 10; fourth column, day 15). Scale = 1 mm. g Quantification of TMEM-mediated vascular leakiness (TMEM activity), assessed via the TMEM activity-MRI assay across multiple time-points in the HT17 mice shown in (f). h Breast cancer patient demographic and histopathologic data. i, j Representative images from TMEM triple-stain immunofluorescence (i) and TMEM Activity-MRI maps of the corresponding tumor ROIs (j) in the pilot patient cohort of breast carcinoma, here presented as “low” TMEM doorway score patient (first row), and “high” TMEM doorway score patient (second row). Scale = 100 um (i) and 1 cm (j). k Correlation of TMEM activity-MRI score with TMEM doorway score in the pilot patient cohort. Spearman’s rank correlation coefficient. Error bars: standard deviation (SD). patient-derived xenograft (PDX) model, which demonstrates variance; post hoc analysis: Mann–Whitney U-test) reduction of excellent preclinical utility (Supplementary Fig. 1), especially in TMEM Activity-MRI (Fig. 5b) and CTCs (Fig. 5d). the context of chemotherapy-induced metastasis . It should be Based on these preliminary data, we reasoned that TMEM noted that the HT17 mice treated with paclitaxel, with or without Activity-MRI could be utilized as a companion diagnostic to rebastinib, were generated from the same cohort as the HT17 monitor the effects of chemotherapy in breast cancer patients. To mice shown in Fig. 2; as a result, the same untreated animal group evaluate such potential in a preclinical setting, we designed could be re-graphed as a reference group to demonstrate baseline another mouse carcinoma study, again using the HT17 patient- TMEM Activity-MRI values (Fig. 5a). Importantly, our results derived xenograft (Supplementary Fig. 1). In a longitudinal fashion, indicate that TMEM activity-MRI is significantly increased HT17 mice either received paclitaxel chemotherapy alone or (p < 0.05; Kruskal–Wallis analysis of variance; post hoc analysis: paclitaxel along with rebastinib, while TMEM Activity-MRI was Mann–Whitney U-test) in HT17 mice receiving neoadjuvant measured in frequent intervals to mimic the clinical scenario of paclitaxel when compared to the vehicle controls (Fig. 5b, c). As monitoring breast cancer patients in the course of neoadjuvant expected, the addition of rebastinib in paclitaxel-treated HT17 treatment (Fig. 5e). As expected , we found that paclitaxel mice results in a significant (p < 0.05; Kruskal–Wallis analysis of chemotherapy significantly (p < 0.05; repeated-measures ANOVA) Published in partnership with the Breast Cancer Research Foundation npj Breast Cancer (2022) 101 G.S. Karagiannis et al. increased TMEM Activity-MRI during the course of treatment, procedures, requiring the collection of dynamic (high temporal compared to vehicle alone (Fig. 5f, g). Importantly, however, the and spatial) resolution images starting at contrast infusion and co-administration of rebastinib with paclitaxel prevented the lasting for 90s of contrast circulation. Traditionally, standard-of- expected chemotherapy-induced increase of TMEM Activity-MRI, care images are acquired after this time period. In total, this thus bringing the raw TMEM Activity-MRI values down to the same acquisition and the accompanying pre-contrast T1-relaxation levels as the mice treated with vehicle controls (Fig. 5f, g). These quantification adds roughly 3 to-4 min to the entire clinical exam, observations suggest that the newly established TMEM Activity- does not alter the clinical effectiveness, and therefore can be MRI measurement represents a potential surrogate of TMEM routinely added to the stand-of-care exam. MRI sequences doorway activity in preclinical animal models of breast carcinoma, necessary for this rapid acquisition of dynamic data are routinely thus demonstrating the significant potential for clinical utility. available on most clinical high-field (1.5 and 3 Tesla) MRI systems, TMEM doorway activity assays are not currently available in so these methods have been easily incorporated into clinical human clinical practice because it is not possible to inject practice in the clinical setting. In conclusion, a single combined fluorescent dextran in human patients to evaluate TMEM Activity- MRI exam, which includes both the standard-of-care clinical Dextran in a clinical setting. Thus, we evaluated the possibility that assessment of the patient and our proposed TMEM Activity-MRI the new TMEM Activity-MRI measurement could serve as a non- assay, will not delay the patient’s clinical assessment or induce invasive surrogate for TMEM doorway activity and, therefore, additional discomfort to the patient and could be readily available cancer cell dissemination in humans. Thus, we assembled a “pilot” for use in the clinic as a companion diagnostic. patient cohort that accrued 11 patients without distant metastasis Several studies have previously reported on the use of dynamic who had a wide distribution in tumor size (T1 = 3, T2 = 5, T1 = 3) contrast-enhanced (DCE) MRI to estimate the metastatic state and 54–56 and axillary lymph node involvement (node-positive = 5). This outcome of breast tumors . Typically, these studies have used cohort included patients with a wide range of TMEM doorway the volume transfer constant between the blood plasma and the scores determined by using the standardized TMEM scoring extravascular extracellular space (K ) as a marker, which trans 57,58 method on tissue sections (Fig. 5h), expected to present disparate evaluates the full wash-in of the contrast agent . In contrast, 2,3,7 metastatic risk according to prior clinical investigations . Tumor TMEM Activity-MRI only uses the extravasation of GBCA during the ROI acquisition and TMEM Activity-MRI calculations were per- first pass through the vasculature, exclusively corresponding to formed in an analogous fashion to the mouse protocol, albeit with the tightly controlled vascular opening of the TMEM doorway. As minor modifications, as described in Materials and Methods. As such, TMEM doorway-independent mechanisms of vascular representative examples, TMEM doorway immunostaining images leakage (i.e., necrosis) are efficiently isolated in our established (with either low or high TMEM scores) from biopsies obtained MRI measurement, as shown in Fig. 1c (yellow arrow). Therefore, 54–56 from the tumor site prior to the MRI session are presented along compared to prior methods , TMEM activity-MRI is more with corresponding MRI analyses (Fig. 5i, j). In this pilot patient efficient in eliminating background signals from various sources cohort, TMEM Activity-MRI correlated positively and significantly not associated with the active process of cancer cell intravasation with TMEM score (p < 0.05; Spearman rho = 0.81; Spearman’s Rank and dissemination. Correlation), irrespective of tumor size, age, race, ethnicity, or TMEM doorways are composed of three individual cells in direct lymph node status (Fig. 5k). Along with the animal preclinical data and stable physical contact: a perivascular macrophage, an shown in this study, this human pilot study additionally indicates endothelial cell, and a tumor cell highly expressing the actin- 2,4,53 that TMEM Activity-MRI could serve as a companion diagnostic in regulatory protein Mena . As such, a single voxel in a TMEM the clinical management of breast cancer patients. Activity-MRI map can contain multiple TMEM doorways which are approximately 40–60 um in diameter (the approximate average diameter of a single TMEM doorway), and the observed signal DISCUSSION intensity is likely the result of cumulative signals from multiple It has long been known that TMEM doorway activity correlates active TMEM doorways in the tumor microenvironment. Despite with increased metastatic potential in preclinical mouse models of the overall lower resolution of MRI, the hyper-intense voxels 4,38,53 breast carcinoma . Clinical investigations have since demon- within a TMEM Activity-MRI map likely correspond to large strated that an increased number of TMEM doorways correlates densities (i.e., hotspots) of active TMEM doorways, given that 2,3,7 with increased metastatic risk in breast cancer patients . background signal from other sources is very low, as explained However, protocols for measuring TMEM doorway activity, and above. However, super-resolution kinetic analysis of TMEM door- therefore metastatic risk, in a non-invasive manner (i.e., without way activity via multiphoton intravital imaging has previously the surgical extraction of the primary tumor or a core biopsy) do demonstrated that TMEM doorways remain open for only ~20 min not exist in the clinical setting. These observations have together before the endothelium is spontaneously re-sealed . It is thus clear inspired the pursuit of a novel, non-invasive tool/assay for the that the hyper-intense voxels within the TMEM Activity-MRI map successful measurement of TMEM doorway activity in breast can capture regions where metastatic dissemination is currently cancer patients, which could theoretically be embedded in active. This measurement, therefore, offers exciting possibilities for standard-of-care clinical practice. In this regard, this study focused physician-based monitoring of the metastatic potential in the era on the algorithmic development and validation of the TMEM of personalized medicine. Activity-MRI assay, a dynamic contrast-enhanced first-pass decon- Metastasis is the primary cause of death in breast cancer, yet no volution MRI approach that measures TAVO events, known to clinically validated imaging modalities are available that reflect the biologically correlate with cancer cell intravasation and dissemina- ability of primary breast cancer to metastasize. Although screening tion in the peripheral circulation. Furthermore, this study provides mammography has contributed to greater detection rates and critical insights for the subsequent transfer of the TMEM Activity- reduced breast cancer mortality, it also results in over-diagnosis or MRI assay to the clinical setting, with the vision of facilitating detection of cancers that pose no threat to life. There is, therefore, treatment decision-making for breast cancer patients undergoing an unmet need to develop screening modalities, used as a primary neoadjuvant treatment. screening test, or a reflex test after initial screening mammo- While the translational potential of the TMEM Activity-MRI assay graphy, that may distinguish non-lethal versus potentially lethal is highlighted in the current research-oriented pilot study, the cancers. Multiparameter gene expression assays, including Onco- 59 ®TM clinical potential of the assay requires its incorporation into the type DX Recurrence Score , MammaPrint , Prosigna™, and standard-of-care MRI exam. Indeed, the proposed TMEM Activity- Breast Cancer Index℠) provide similar prognostic information MRI assay is highly compatible with standard-of-care MRI that is driven largely by proliferation and estrogen-dependent npj Breast Cancer (2022) 101 Published in partnership with the Breast Cancer Research Foundation G.S. Karagiannis et al. genes and not by the intrinsic propensity of tumor cells to presents with any significant translational application in meta- 61–63 metastasize or interact with their microenvironment . Inter- static cancers, as well. estingly, TMEM doorway Score correlates poorly with the While our study has primarily focused on breast cancer, it is now Oncotype DX Recurrence Score, captures different biologic known that other types of cancer, such as pancreatic neuroendo- information, and provides complementary prognostic informa- crine tumors, also utilize TMEM doorways as the primary cancer cell dissemination machinery for the initial steps of the metastatic tion . Thus, the TMEM doorway score, and its highly correlated cascade . Previously, pancreatic tissue and tumors have been TMEM-MRI score, offer the potential to more accurately determine successfully visualized using low or high-resolution imaging prognosis regarding recurrence beyond what is possible for lower modalities, including magnetic resonance imaging and multi- ranges of Oncotype DX and other multiparameter gene expres- 84–88 photon intravital microscopy . It would therefore be extremely sion scores. This could lead to treatment decisions that differ from interesting to extend our studies into other types of cancer and those made based on Multiparameter gene expression investigate whether TMEM activity-MRI or an equivalent MRI- assays alone. based measurement could be universally used as an assessment Another potential clinical application of TMEM Activity-MRI is as tool for metastatic potential. a companion diagnostic to monitor the pharmacodynamic effects In conclusion, this article describes the development and of standard cytotoxic therapy and also therapeutic interventions validation of a novel MRI measurement that correlates with designed to induce a blockade of TMEM doorways. It is known metastatic dissemination in preclinical models of breast carcinoma that cytotoxic chemotherapy has profound effects on the tumor 50,52,64 and breast cancer patients. The technological and conceptual microenvironment , including promoting an influx of proan- 64–69 innovation of this newly proposed measurement, herewith known giogenic M2 macrophages and the formation of TMEM 5,70 as TMEM Activity-MRI, is the quantification of only the first pass of doorways . Chemotherapy may also increase the density of 5,71–73 a gadolinium-based contrast agent (GBCA) into the tumor tissue. cancer cells with high dissemination potential . Therefore, This procedure notably eliminates background signals from other quantifying the effects of chemotherapy on the dissemination sources and highly correlates with the biological activity of cancer potential of the tumor microenvironment could be used as a cell dissemination doorways, known as TMEM doorways, whose pharmacodynamic biomarker to stratify distant metastatic risk in activity is the rate-limiting step of metastatic dissemination. The patients with residual disease after neoadjuvant chemotherapy or TMEM Activity-MRI assay is sensitive to various anti-cancer and adding agents such as the Tie2 kinase inhibitor rebastinib that anti-metastatic therapies (i.e., chemotherapy, TMEM doorway blocks TMEM doorway-mediated cancer cell intravasation. Addi- inhibitors, etc.) and has been successfully utilized here to classify tional studies are required to further evaluate the clinical validity patients into high and low-risk individuals for developing of TMEM Activity-MRI for distant recurrence risk and its potential metastasis as a non-invasive TMEM score surrogate. We propose clinical utility in breast cancer management. that TMEM Activity-MRI could be utilized as a promising All the validation experiments performed in this study have companion diagnostic to facilitate physicians with decision- consistently confirmed that TMEM Activity-MRI significantly making, especially during treatment, as well as with the clinical correlates with endpoints of metastatic dissemination, including management of breast cancer patients. TMEM doorway activity, circulating tumor cells (CTCs), and disseminated tumor cells (DTCs), suggesting that TMEM Activity- MRI can serve as a specific tool to study the biology of cancer cell METHODS dissemination, which is often seen as the rate-limiting step of the Animal subjects 4,38 metastatic cascade . Here, as a proof-of-concept, we documen- Ethics statement. All studies involving mice were carried out in ted that MENA expression is not critical for the function of TMEM accordance with the National Institutes of Health (NIH) regulation doorways, even though it is necessary for the generation of a concerning the care and use of experimental animals and with the proinvasive/promigratory tumor cell subset (Fig. 4). These approval of the Institutional Animal Care and Use Committee (IACUC) of observations foster a subsequent question pertaining to the Molecular Imaging, Inc. (Ann Arbor, MI), a facility accredited by the precise MENA-independent mechanism, via which cytoskeletal Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC), or with the approval of the Albert Einstein College of Medicine dynamics at TMEM doorways may regulate endothelial cell Animal Care and Use Committee. opening to facilitate transendothelial migration. Indeed, other members of the Ena/VASP family could play a key role in this 74,75 MMTV-PyMT (spontaneous model). Transgenic mice expressing the Poly- process and should be explored in the future to fully oma Virus Middle-T (PyMT) antigen under the control of mammary tumor appreciate the complex regulation of TAVO events. Interestingly, virus long terminal repeat (MMTV-LTR) were bred in-house at the Albert TMEM doorways are microenvironmental niches that may Einstein College of Medicine (Condeelis lab), maintained on the FVB cultivate induction and maintenance of stemness and localized background, and the resulting tumors could be palpated at ~6 weeks of 76,77 immunosuppression , indicating they may confer suitable age. Depending on the experimental question, MMTV-PyMT mice were niches for the development of immune-privileged, metastatic used in different age groups. stem cells . The particularly high correlation observed between TMEM Activity-MRI and DTCs (Fig. 3i–k) is supportive of this notion MMTV-PyMT (transplantation model). Syngeneic transplantation models and further indicates that TMEM Activity-MRI may harbor were generated through orthotopic transplantation of 1 mm tumor additional potential in studying the immune tumor chunks from 12–16-week-old MMTV-PyMT donor mice bearing late-stage carcinomas of ~1 cm in diameter into 5–6-week-old FVB recipients. These microenvironment. tumors were not passaged in culture or dissociated but propagated as In this study, TMEM Activity-MRI was only measured in primary 39,43 tumor chunks in vivo . The tumor chunk was implanted on the fourth tumors of mice and humans. However, more recent observations mammary pad on the right side of the recipient mouse. suggest that TMEM doorways do not only assemble in the primary tumor microenvironment but also in local (i.e., lymph nodes) or −/− MMTV-PyMT/ Mena (spontaneous model). Generation of the MMTV- 79–81 distant (i.e., lungs) metastatic sites . These observations raise −/− PyMT Mena mice was achieved by crossing MMTV-PyMT mice with the intriguing possibility that metastatic breast tumors may 43 MENA heterozygotes as described . The forward and reverse primer potentially utilize hematogenous routes, based on the assembly sequences used to identify the transmission of the disrupted MENA allele −/− of TMEM doorways, to efficiently re-disseminate to tertiary sites in MENA mice are the following: LACZ-F: CGATCGTAATCACCCGAGTGT; 80,82,83 after achieving metastatic colonization . As such, future LACZ-R: CCGTGGCCTGACTCATTCC; Enah-3-F: ACCGCAGTCTCCCTTACAT AACTTA; Enah-3-R: GCACTGCACTTTTAATCAGGTGTCT. efforts should establish whether the TMEM Activity-MRI assay Published in partnership with the Breast Cancer Research Foundation npj Breast Cancer (2022) 101 G.S. Karagiannis et al. Patient-derived xenograft (PDX) model. The generation of the estrogen Development of the mouse TMEM activity-MRI assay receptor-negative (ER ) HT17 patient-derived xenograft has been devel- The MRI images were obtained on a 9.4 Tesla Agilent Direct Drive MRI/MRS oped in-house, as described . Similar to the PyMT-transplantation model system. Images were acquired with a 24 mm diameter surface receive-only described above, the HT17 tumors were also never passaged in culture or coil within a 12 cm diameter pin-switch driven volume transmit coil. High- dissociated but only propagated as tumor chunks in vivo (passage resolution T2-weighted images were used to determine the placement of number <5). Each tumor chunk was implanted on the fourth mammary the two slices in the permeability estimation, one for the arterial source pad on the right side of severe combined immunodeficiency (SCID) and one for the tumor slice. The GRE images were collected with a field of recipient mice. The resulting tumors can be palpated 4–6 weeks following view of 23 mm , matrix size of 96 × 128, thickness of 1 mm, image the transplantation. repetition time (TR) of 23 ms, echo time (TE) of 3 ms, flip angle (FA) was 28 degrees, acquisition bandwidth of 100 kHz, for an image temporal resolution of 2 s (signal averages= 1) or 4 s (SA= 2). The samples in the Chemicals and reagents administered to mice dynamic study were sampled every 4 s and for 10 min, with the contrast (gadolinium) being injected at 1 min. Prior to dynamic image acquisition, a Rebastinib reconstitution and administration. Rebastinib was reconstituted T1 measurement was completed by varying the FA between 2 and 60 ms, at a concentration of 10 mg/mL in 0.4% hydroxypropyl methylcellulose to which T1 was fit on a pixel-by-pixel basis, and the low FA data (FA 2, 4, 6, (HPMC). Each mouse in the experimental group received p.o. doses of and 8 degrees) were used along with the dynamic data to fit dynamic T1 to 10 mg/kg rebastinib (100 μL total volume) twice per week, for 4 weeks. The the dynamic curve pixel-by-pixel for determination of CA concentration. control group received p.o. 100 μL of HPMC. During a contrast-enhanced MRI exam, a gadolinium-based contrast agent (GBCA) is typically injected and CA passage into the tumor tissue is Chemotherapy reconstitution and administration. Mice were treated with visualized by comparing image intensity changes pre- and post-contrast the taxane-based chemotherapeutic paclitaxel (Sigma-Aldrich). Paclitaxel (subtraction-based contrast) . Alternatively, high temporal resolution MRI was reconstituted at a concentration of 10 mg/mL in 1:1 EtOH:Cremophor- of the dynamic passage of the GBCA through tissue beds using EL (Millipore). Each mouse in the experimental group received an i.v. dose longitudinal (T1) relaxation-based agents can be used to measure tissue of 10 mg/Kg paclitaxel (200 μL total volume) every 5 days, for a total of permeability to the GBCA. Typically, these contrast agents are T1-based three doses. The control group received an i.v. injection of 200 μL 1:1 agents, for which the change in tissue relaxivity associated with the EtOH:Cremophor-EL. contrast agent’s presence can be used to approximate the agent’s blood and tissue time-dependent concentration. Mathematical deconvolution Clodronate and PBS liposome reconstitution and administration. The methods can be used to extract the tissue response function from the clodronate liposomes were administered in experimental mice with an tissue signal response function using the measured or inferred arterial i.v. injection at a dose of 5 mL/Kg (200 μL total volume) every other day for input concentration function of the contrast agent, the form of which is a total of seven doses. The control group received an equivalent i.v. derived from the differential equation: injection of PBS liposomes. dC ðtÞ (1) V ¼ PSρ C ðÞ t  C ðtÞ ; e p e dt High-molecular weight (155 kDa) dextran reconstitution and administra- where C (t) and C (t) are the GBCA concentration in the blood plasma and tion. 155-kDa Tetramethylrhodamine-Dextran (TMR-Dextran) solution p e tissue, respectively, P is the GBCA endothelial permeability, S is the was reconstituted at a concentration of 20 mg/mL in sterile phosphate- capillary wall surface area, and ρ is the tissue density. buffered saline (PBS). Each mouse in control or experimental groups The product PS is often called the permeability-surface area product, received 100 μL total volume of TMR-Dextran via right-sided retro-orbital which assumes that delivery of the GBCA and perfusion are sufficient to injection, 1-h before sacrifice. ensure that the permeability is the dominant determinant of GBCA exflux into the tissue. Endothelial permeability of the GBCA is usually represented Gadolinium reconstitution and administration. The gadolinium-based as the permeability transfer constant, or k ; trans contrast agent (Magnevist Bayer HealthCare, Pittsburgh, PA) was used in experimental mice at a dose of 0.1 mm/kg, diluted in gadopentetate K ¼ PSρ (2) trans dimeglumine (140 ul total volume), and was administered at 20 ul/sec via If we use the convention that the concentration of GBCA in the imaging tail vein catheter injection with a 50 µl dead space in the line filled of voxel, Ct(t) is given by the volume weighted concentrations of the plasma saline. V and tissue V ; p e C ðÞ t ¼ V C ðÞ t þ V C ðtÞ (3) t p p e e Human subjects Solution of the differential leads to a convolution integral, which can be TMEM-MRI activity was measured in breast cancer patients. This study was simplified by assuming that the uptake of GBCA into the extravascular/ designed by Montefiore-Einstein cancer center investigators, approved by extracellular space is minimal during the measurement. Because the entire the Albert Einstein Institutional Review Board, and conducted in measurement process outlined above is limited to the ‘first pass’ of the accordance with the ethical principles derived from international guide- GBCA, this assumption is valid and simplifies the convolution integral to lines, including the International Council for Harmonisation Good Clinical the following: Practice guidelines, the Declaration of Helsinki, and local regulations on the conduct of clinical research. All the participants provided written informed 0 0 (4) C ðÞ t ¼ k C ðÞ t dt t fp p consent before enrollment. The inclusion criteria were the following: breast mass >1 cm with biopsy-proven histology of invasive breast carcinoma which shows that the Gd concentration in the tissue is modeled as a (any histologic type and ER, PR, HER2 status), age ≥18 years, ECOG convolution of the tissue’s permeability-surface product (k ) and the fp performance status 0–1, ability to undergo MRI with gadolinium integrated GBCA delivery. enhancement, no known or suspected renal impairment, normal organ This estimation is valid during the first pass of the contrast bolas, before and marrow function, weight less than or equal to the MRI table limit, the recirculation of the agent. The use of this first-pass leakage profile has ability to understand and willingness to sign a written informed consent. been shown to give a more accurate estimation of the TMEM-associated The exclusion criteria were the following: prior chemotherapy of radiation vascular opening (TAVO, also shown as k ) compared to the Multi- fp therapy to the ipsilateral breast, breast prosthetic implants (silicone or compartmental Model . saline), use of any investigational agent within 30 days of starting the A baseline T1 map needs to be estimated to allow for the calculation of a study, uncontrolled intercurrent illness including, but not limited to, concentration map in real-time. Multiple GRE images with varying flip ongoing or active infection, symptomatic congestive heart failure, unstable angles (FA) are acquired and Eq. 5, based on the Ernst formula, is fit to give angina pectoris, cardiac arrhythmia, or psychiatric illness/social situations a T1 estimate: that would limit compliance with study, pregnancy, and lactation. The TR=T10 1  e clinical and pathological characteristics of the tumors are summarized in (5) s ¼ mðÞ sinFA TR=T1 1 ðÞ cosFA e Fig. 5h. npj Breast Cancer (2022) 101 Published in partnership with the Breast Cancer Research Foundation G.S. Karagiannis et al. To fit this model for T1 ,FA= (2, 4, 6, 8, 10, 12, 16, 20, 25, 30, 40) degrees panMena antibody (510693; BD Biosciences) to detect Mena-expressing were acquired, with TR = 15 ms and TE = 3.2 ms. Levenburg–Marquart cancer cells. To visualize macrophages, we used anti-IBA1 antibody (019- nonlinear least squares algorithm is used to fit Eq. 5. This gives a baseline 19741; Wako) for mouse and anti-CD68 (MO876; Dako) for human tumors. T1 estimate. To visualize endothelial cells, we used anti-endomucin (SC-65495; Santa After gaining the baseline T1 map, the GBCA is injected and the Cruz) for mouse and anti-CD31 (MO823; Dako) for human tumors. dynamic GRE series (S(t)) of images is collected, with FA= 25 degrees Appropriate areas containing invasive cancer tissue suitable for TMEM sampled every 4 s. A dynamic T1 series (T1(t)) is estimated directly using doorway analysis were identified by low-power scanning using the the following: following criteria: high density of tumor, adequacy of a tumor, lack of necrosis or inflammation, and lack of artifacts such as retraction or folds. 1 1 1  D 2,36 TMEM doorway scoring was performed as previously described . The ¼  ln (6) T1ðÞ t TR 1 ðÞ cos FA D assessment of TMEM doorway scores was performed with Adobe Photoshop on ten high-power (400X) digital images of the most TR=T10 StðÞ Sð0Þ 1  e representative areas of the tumor. The total number of TMEM doorways (7) D ¼ þ TR=T1 0 for each image were tabulated, and the scores from all ten images were m sinFA 1 ðÞ cosFA e summed to give a final TMEM doorway density for each patient sample, The Gd concentration map (C(t)) is then calculated with the following: expressed as the number of TMEM doorways per total area (ten high- 1 1 power [400X] fields). A representative high-power magnification image T1ðÞ t T1 (8) CtðÞ ¼ showing the tripartite TMEM doorway is indicated in Supplementary fig. R1 3a. IBA1 single immunohistochemistry was performed with the IBA1 −1 −1 where the relaxivity of the Gd is R1 = 3.2 s mM . primary antibody used in the TMEM immunohistochemistry, and a To narrow our analysis on vascular leakage elicited by TAVO events, an representative high-power magnification image is indicated in Supple- arterial source must first be found. So, we acquire two slices in our scans, mentary fig. 3b. one through the tumor (C ) and one through the artery (C ). Using Eq. 4 t a with C = C , weighted least squares is used to estimate k with higher p a fp Multichannel immunofluorescence weights given to the points at the top of the C curve. For all multichannel immunofluorescence experiments, slides were first To graphically compare the TMEM-Associated Vascular Opening (TAVO) deparaffinized by melting at 60 C in an oven equipped with a fan for events of mouse breast carcinomas among different conditions, we 20 min, followed by 2X xylene treatment for 20 min. Slides were then calculated the ratio of values above a permeability threshold in the entire rehydrated, and antigen retrieval was performed in 1 mM EDTA (pH 8.0) at histogram (Eq. 9). The optimal threshold for class separation (early versus 97 C for 20 min in a conventional steamer, followed by incubation in a late carcinomas in PyMT mice) was found to be ~0.001. In the following blocking buffer solution (10% FBS, 1% BSA, 0.0025% fish skin gelatin in equation (Eq. 9), H represents the histogram of k values found within the i fp 0.05% PBST) for 60 min at room temperature. Specific considerations were tumor. then considered depending on the assay performed, as described below. :04 :04 X X TMEM Activity  MRI ¼ H = H (9) i i INV MENA immunofluorescence. After standard slide preparation as i¼th i¼0 INV described above, slides were incubated with chicken anti-MENA (0.25 μg/mL, generated in the lab of Dr. John S. Condeelis) in a blocking buffer for 60 min at room temperature. Samples were washed three times in 0.5% PBST and incubated with an HRP-conjugated IgG anti-chicken Development of the human TMEM activity-MRI assay secondary antibody for 60 min at room temperature. After washing, slides TMEM activity was measured in a small preliminary study of breast cancer were incubated with biotinylated tyramide (Perkin Elmer; Opal 4-color patients after consent in accord with an approved IRB protocol. TMEM Fluorescent IHC kit) diluted at 1:50 in amplification buffer for 10 min. After activity was measured on a Philips 3 T Ingenia Elition. Dynamic contrast- washing, slides were incubated with spectral DAPI for 5 min and mounted enhanced (DCE) imaging was conducted following intravenous adminis- with ProLong Gold antifade reagent (Life Technologies). The slides were tration of Gadoterate (0.1 mm/kg) administered as a bolus a power imaged on the Pannoramic 250 Flash II digital whole slide scanner, using a injector. The dynamic contrast-enhanced MRI (DCE-MRI) protocol used for 20 × 0.75NA objective lens. Tissue suitable for scanning was automatically clinical assessment was modified to permit the collection of a T1 detected using intensity thresholding. relaxometry data set prior to CA injection and a dynamic data set collected during the CA injection. The DCE-MRI data set began 1 min prior TMEM activity-dextran assay. Assessment of TMEM-mediated vascular to injection and continued for at least 90 s following injection using a 4D- opening (TAVO) was performed using multichannel-IF in an FFPE section TRAK XD imaging protocol which employed compressed sense (factor 8) with a sequential TMEM triple-IHC section already stained. Each slide was and both keyhole (20%) and half scan (factors 0.625 and 0.878) parallel stained with a primary antibody mixture cocktail against rat anti- imaging providing rapid (3 to 5 s) volume acquisitions. Image volumes 3 endomucin (1:500; sc-65495; Santa Cruz) and rabbit anti-TMR (1:1,000; A- were acquired in rapid succession, with a high-resolution matrix (1 mm 6397; Life Technologies). Slides were then washed three times in 0.05% isotropic or nearly isotropic voxels), and (typically) a TR of 3.9 ms, TE PBST, and incubated with a secondary antibody mixture cocktail, including 1.95 ms, temporal 3D volume resolution of 3.2 to 4.6 s, flip angle 28 donkey anti-rabbit Alexa-488 and goat anti-rat Alexa-568, both at 1:200 degrees, and between 60 and 120 sequential 3D volumes. Imaging dilution for 60 min at room temperature. After washing (0.05% PBST 3X), covered both breasts, as well as the heart, from which an arterial signature slides were incubated with spectral DAPI for 5 min and mounted with of CA uptake was acquired. Pre-contrast and post-contrast conventional 3D ProLong Gold antifade reagent (Life Technologies). The slides were imaged images were acquired after the dynamic images, allowing delineation of on the Pannoramic 250 Flash II digital whole slide scanner, using a the tumor volume. Fitting of the convolution equation followed that 20 × 0.75NA objective lens. Tissue suitable for scanning was automatically described for the animal data, with a similar calculation of the TMEM detected using intensity thresholding. Whole tissue images were uploaded Activity-MRI. in Pannoramic Viewer version 1.15.4 (3DHISTECH). To investigate whether highly permeable blood vessels were associated with TMEM doorways, Histology (H&E) and immunohistochemistry for IBA1 and multiple 40X fields were captured to obtain ~20–25 vascular profiles for TMEM doorways each case. A “vascular profile” was defined as an endomucin blood vessel with clear margins, either longitudinally or in cross-section. Vessels in the After mice were sacrificed, all mammary tumors were extracted and vicinity or continuing away from the field of view were excluded from this immersed in 10% formalin in a volume ratio of tumor to formalin of 1:7. analysis because it was not possible to access the entire perivascular area Tissues were fixed for 24 to 48 h and embedded in paraffin, then processed associated with them. In each image, the endomucin and dextran-TMR for histological examination. One 5 µm section from each tumor was channels were each thresholded just above the background based upon stained for hematoxylin and eosin (H&E) and one for TMEM. The TMEM intensity by using contrast adjustment. For each vascular profile, the doorway assay is a triple-stain IHC for predicting metastatic risk, in which endomucin channel was then used as an exclusion mask to the dextran three antibodies are applied sequentially and developed separately with channel to directly designate an ROI that belonged exclusively to the different chromogens on a Dako Autostainer. TMEM doorway stain was 2 extravascular portion of the dextran. Because IF staining may result in a performed as previously described , except that in this study we used anti- Published in partnership with the Breast Cancer Research Foundation npj Breast Cancer (2022) 101 G.S. Karagiannis et al. certain number of nonspecific “speckles” with a positive TMR signal, we As such, slight deviations can be observed in the total number of mice in used an arbitrary threshold of >20 pixels around a vascular profile to each individual figure panel, accounting for the exclusion criteria applied consider it a “leaky vascular profile”. The sequential TMEM IHC sections in each case. were then used to assess whether these leaky profiles had an associated Transgenic MMTV-PyMT animals, as well as animals transplanted with TMEM structure. To directly compare TAVO-dependent dextran leakage patient-derived tumors (e.g., HT17 xenografts), were housed in cages of among different groups of mice, the extravascular dextran ROI was five animals per the regulations of the Albert Einstein College of Medicine expressed in each image as an area fraction, and an average among all (AECOM) Animal Care and Use Committee. Once the mice reached the ROIs for each mouse was reported. criteria for inclusion into the experimental pipeline (tumors with a diameter of ~2–3 mm), they were randomly allocated to rebastinib-, chemotherapy-, or vehicle-treated groups. Quantification of disseminated tumor cells (DTCs) in mouse lungs. The The two pathologists (M.H.O. and J.G.J.) involved in TMEM scoring were assessment of DTCs in the lungs was performed in the MMTV-PyMT mouse model, based on the availability of commercial anti-PyMT antibodies for blinded to the specific group allocations, as were all the scientists performing CTC scoring, all IF/IHC analyses, and all MRI feature specific detection of PyMT-expressing tumor cells, indicative of this mouse quantifications. Importantly, the TMEM Activity-MRI maps, upon which model of breast carcinoma. Each lung section was stained with primary TMEM-MRI-Activity is determined, was conducted prior to and without antibody mixture cocktail against mouse anti-pancytokeratin (PanCK; knowledge of the TMEM pathology score (and scored by different 1:1,000; C2562; Sigma) and rat anti-PyMT (PyMT; 1:400; NB100-2749; Nobus individuals) and therefore was blinded. Biologicals). Slides were then washed three times in 0.05% PBST and incubated with a secondary antibody mixture cocktail, including goat anti- mouse Alexa-488 and donkey anti-rat Alexa-568, both at 1:200 dilution for 60 min at room temperature. After washing (0.05% PBST 3X), slides were DATA AVAILABILITY incubated with spectral DAPI for 5 min and mounted with ProLong Gold The original contributions presented in the study are included in the article and antifade reagent (Life Technologies). The slides were imaged on the supplementary material. Further inquiries can be directed to the corresponding Pannoramic 250 Flash II digital whole slide scanner, using a 20 × 0.75NA authors. No datasets were generated or analyzed during the current study. objective lens. Tissue suitable for scanning was automatically detected using intensity thresholding. Whole tissue images were uploaded in Received: 5 October 2021; Accepted: 11 July 2022; Pannoramic Viewer version 1.15.4 (3DHISTECH). DTCs were identified as + + + single PanCK PyMT DAPI cells and expressed as mean (i.e., DTCs per high-power field) for each mouse. Intravasation (circulating tumor cell) assay REFERENCES The in vivo intravasation assay was performed as previously 1. Harney, A. S. et al. Real-time imaging reveals local, transient vascular perme- 5,39,90,91 described . Mice were anesthetized with isoflurane, and blood was ability, and tumor cell intravasation stimulated by TIE2hi macrophage-derived taken from the right ventricle by heart puncture, using 25 G needles VEGFA. Cancer Discov. 5, 932–943 (2015). coated with heparin. Erythrocytes were lysed using 10 ml of 1X RBC lysis 2. Robinson, B. D. et al. Tumor microenvironment of metastasis in human breast buffer (multi-species, eBioscience). The samples were centrifuged at carcinoma: a potential prognostic marker linked to hematogenous dissemination. 200 × g for 5 min, then cell pellets were reconstituted in 10 ml of Clin. Cancer Res. 15, 2433–2441 (2009). Dulbecco’s modified Eagle medium (DMEM/F-12) supplemented with 3. Rohan, T. E. et al. Tumor microenvironment of metastasis and risk of distant 20% fetal bovine serum (FBS) and plated in 10-cm Petri dishes. After cells metastasis of breast cancer. J. Natl Cancer Inst. https://doi.org/10.1093/jnci/ were attached, single tumor cells were counted. The total number of cells dju136 (2014). counted was divided by the volume of blood taken. Tumor cells counted 4. Karagiannis, G. S., Goswami, S., Jones, J. G., Oktay, M. H. & Condeelis, J. S. Sig- were either CFP- or Dendra2-positive, thus confirming their identity as natures of breast cancer metastasis at a glance. J. Cell Sci. 129, 1751–1758 (2016). tumor cells. As a negative control, blood from a non-tumor-bearing mouse 5. Karagiannis, G. S. et al. Neoadjuvant chemotherapy induces breast cancer was analyzed, and the complete absence of epithelial cells was confirmed. metastasis through a TMEM-mediated mechanism. Sci. Transl. Med. https:// doi.org/10.1126/scitranslmed.aan0026 (2017). 6. Harney, A. S. et al. The selective Tie2 inhibitor rebastinib blocks recruitment and Statistical analysis function of Tie2(Hi) macrophages in breast cancer and pancreatic neuroendo- All statistical analyses, including graph/plot generation and statistical crine tumors. Mol. Cancer Ther. 16, 2486–2501 (2017). hypothesis testing, were performed in GraphPad Prism 8 software. The 7. Sparano, J. A. et al. A metastasis biomarker (MetaSite Breast Score) is associated variables in our study (TMEM score, TMEM Activity-MRI, TMEM Activity- with distant recurrence in hormone receptor-positive, HER2-negative early-stage Dextran, circulating tumor cells per mL blood, disseminated tumor cells) breast cancer. Nat. PJ Breast Cancer 3, 42 (2017). were all continuous and, as such, presented as dot-plots with means and 8. Kienast, Y. et al. Real-time imaging reveals the single steps of brain metastasis their corresponding Standard Deviation (SD). Statistical comparisons formation. Nat. Med. 16, 116–122 (2010). between two independent groups were performed using a two-tailed 9. Tanaka, K. et al. In vivo real-time imaging of chemotherapy response on the liver Mann–Whitney U-test, and those between more than two independent metastatic tumor microenvironment using multiphoton microscopy. Oncol. Rep. groups were performed using Kruskal–Wallis one-way analysis of variance. 28, 1822–1830 (2012). For longitudinal analyses, repeated-measures ANOVA was used. Statistical 10. Provenzano, P. P., Eliceiri, K. W. & Keely, P. J. Multiphoton microscopy and correlations were performed by Spearman’s correlation, with the data fluorescence lifetime imaging microscopy (FLIM) to monitor metastasis and the presented as scatterplots with fit lines and their corresponding P values. tumor microenvironment. Clin. Exp. Metastasis 26, 357–370 (2009). Mouse studies were conducted in at least three independent animal 11. Bonapace, L. et al. If you don’t look, you won’t see: intravital multiphoton imaging cohorts, which were pooled together for statistical analysis. of primary and metastatic breast cancer. J. Mammary Gland Biol. Neoplasia 17, Sample size calculations for all the validation studies were performed 125–129 (2012). using the following parameters: significance level (adjusted for sidedness) 12. Yan, J. et al. Use of multiphoton microscopy to diagnose liver cancer and lung of 0.025, probability of type II error of 0.2 (i.e., statistical power 0.8), and metastasis in an orthotopic rat model. Scanning 34, 271–277 (2012). expected difference in means equal to 1.5 SD units, based on the 13. Entenberg, D. et al. Time-lapsed, large-volume, high-resolution intravital imaging assumption that the SD of the response variable was 1 unit. Based on the for tissue-wide analysis of single cell dynamics. Methods 128,65–77 (2017). above, a total of 18 animals was calculated to enter the study. The 14. Entenberg, D. et al. In vivo subcellular resolution optical imaging in the lung experiments were initially designed to include >18 animals in each study, reveals early metastatic proliferation and motility. Intravital 4,1–11 (2015). to account for mice that died during an MRI session, mice that fitted one or 15. Entenberg, D. et al. A permanent window for the murine lung enables high- more exclusion criteria, and finally for statistical outliers. The following resolution imaging of cancer metastasis. Nat. Methods 15,73–80 (2018). animals were removed from the study based on exclusion criteria: Mice 16. Gligorijevic, B. et al. Intravital imaging and photoswitching in tumor invasion and intravasation microenvironments. Micros Today 18,34–37 (2010). whose tumors were overly necrotic or cystic; mice with inadequate tumor 17. Patsialou, A. et al. Intravital multiphoton imaging reveals multicellular streaming tissue to perform IHC or IF; mice with less than 0.5 ml of blood collected as a crucial component of in vivo cell migration in human breast tumors. Intra- through heart puncture for the CTC assay; and mice that were either obese vital 2, e25294 (2013). (>35 g), or emaciated (<17 g), upon reaching the experimental endpoints. npj Breast Cancer (2022) 101 Published in partnership with the Breast Cancer Research Foundation G.S. Karagiannis et al. 18. Alexander, S., Koehl, G. E., Hirschberg, M., Geissler, E. K. & Friedl, P. Dynamic 46. Pignatelli, J. et al. Invasive breast carcinoma cells from patients exhibit MenaINV- imaging of cancer growth and invasion: a modified skin-fold chamber model. and macrophage-dependent transendothelial migration. Sci. Signal 7, ra112 Histochem. Cell Biol. 130, 1147–1154 (2008). (2014). 19. Beerling, E., Ritsma, L., Vrisekoop, N., Derksen, P. W. & van Rheenen, J. Intravital 47. Pignatelli, J. et al. Macrophage-dependent tumor cell transendothelial migration microscopy: new insights into metastasis of tumors. J. Cell Sci. 124, 299–310 (2011). is mediated by Notch1/MenaINV-initiated invadopodium formation. Sci. Rep. 6, 20. Karreman, M. A. et al. Correlating intravital multi-photon microscopy to 3D 37874 (2016). electron microscopy of invading tumor cells using anatomical reference points. 48. Roussos, E. T. et al. Mena invasive (MenaINV) promotes multicellular streaming PLoS ONE 9, e114448 (2014). motility and transendothelial migration in a mouse model of breast cancer. J. Cell 21. Sahai, E. et al. Simultaneous imaging of GFP, CFP and collagen in tumors in vivo Sci. 124, 2120–2131 (2011). using multiphoton microscopy. BMC Biotechnol. 5, 14 (2005). 49. Oudin, M. J. et al. MenaINV mediates synergistic cross-talk between signaling 22. Sidani, M., Wyckoff, J., Xue, C., Segall, J. E. & Condeelis, J. Probing the micro- pathways driving chemotaxis and haptotaxis. Mol. Biol. Cell 27, 3085–3094 (2016). environment of mammary tumors using multiphoton microscopy. J. Mammary 50. Karagiannis, G. S., Condeelis, J. S. & Oktay, M. H. Chemotherapy-induced metas- Gland Biol. Neoplasia 11, 151–163 (2006). tasis: mechanisms and translational opportunities. Clin. Exp. Metastasis 35, 23. Soon, L., Braet, F. & Condeelis, J. Moving in the right direction-nanoimaging in 269–284 (2018). cancer cell motility and metastasis. Microsc Res. Tech. 70, 252–257 (2007). 51. Karagiannis, G. S., Condeelis, J. S. & Oktay, M. H. Chemotherapy-induced metas- 24. Wyckoff, J. et al. A paracrine loop between tumor cells and macrophages is tasis in breast cancer. Oncotarget 8, 110733–110734 (2017). required for tumor cell migration in mammary tumors. Cancer Res. 64, 7022–7029 52. Karagiannis, G. S., Condeelis, J. S. & Oktay, M. H. Chemotherapy-induced metas- (2004). tasis: molecular mechanisms, clinical manifestations, therapeutic interventions. 25. Wyckoff, J. B. et al. Direct visualization of macrophage-assisted tumor cell intra- Cancer Res. 79, 4567–4576 (2019). vasation in mammary tumors. Cancer Res. 67, 2649–2656 (2007). 53. Oktay, M. H. & Jones, J. G. TMEM: a novel breast cancer dissemination marker for 26. Harney, A. S., Wang, Y., Condeelis, J. S. & Entenberg, D. Extended time-lapse the assessment of metastatic risk. Biomark. Med. 9,81–84 (2015). intravital imaging of real-time multicellular dynamics in the tumor micro- 54. Ioannidis, G. S., Maris, T. G., Nikiforaki, K., Karantanas, A. & Marias, K. Investigating environment. J. Vis. Exp. https://doi.org/10.3791/54042 (2016). the correlation of Ktrans with semi-quantitative MRI parameters towards more 27. Lohrke, J. et al. 25 years of contrast-enhanced MRI: developments, current robust and reproducible perfusion imaging biomarkers in three cancer types. IEEE challenges and future perspectives. Adv. Ther. 33,1–28 (2016). J. Biomed. Health Inf. 23, 1855–1862 (2019). 28. Bar, A., Skorka, T., Jasinski, K. & Chlopicki, S. MRI-based assessment of endothelial 55. Thakran, S. et al. Characterization of breast lesion using T1-perfusion magnetic function in mice in vivo. Pharm. Rep. 67, 765–770 (2015). resonance imaging: qualitative vs. quantitative analysis. Diagn. Inter. Imaging 99, 29. Li, C. H. et al. Flow versus permeability weighting in estimating the forward 633–642 (2018). volumetric transfer constant (K(trans)) obtained by DCE-MRI with contrast agents 56. Cheng, Z. et al. Discrimination between benign and malignant breast lesions of differing molecular sizes. Magn. Reson. Imaging 36, 105–111 (2017). using volumetric quantitative dynamic contrast-enhanced MR imaging. Eur. 30. Li, K. L., Zhu, X. P., Waterton, J. & Jackson, A. Improved 3D quantitative mapping Radio. 28, 982–991 (2018). of blood volume and endothelial permeability in brain tumors. J. Magn. Reson. 57. Patlak, C. S., Blasberg, R. G. & Fenstermacher, J. D. Graphical evaluation of blood- Imaging 12, 347–357 (2000). to-brain transfer constants from multiple-time uptake data. J. Cereb. Blood Flow. 31. Guy, C. T., Cardiff, R. D. & Muller, W. J. Induction of mammary tumors by Metab. 3,1–7 (1983). expression of polyomavirus middle T oncogene: a transgenic mouse model for 58. Tofts, P. S. et al. Estimating kinetic parameters from dynamic contrast-enhanced metastatic disease. Mol. Cell Biol. 12, 954–961 (1992). T(1)-weighted MRI of a diffusable tracer: standardized quantities and symbols. J. 32. Lin, E. Y. et al. Progression to malignancy in the polyoma middle T oncoprotein Magn. Reson. Imaging 10, 223–232 (1999). mouse breast cancer model provides a reliable model for human diseases. Am. J. 59. Sparano, J. A. et al. Prospective validation of a 21-gene expression assay in breast Pathol. 163, 2113–2126 (2003). cancer. N. Engl. J. Med. 373, 2005–2014 (2015). 33. Fantozzi, A. & Christofori, G. Mouse models of breast cancer metastasis. Breast 60. Sparano, J. A. & Solin, L. J. Defining the clinical utility of gene expression assays in Cancer Res. 8, 212 (2006). breast cancer: the intersection of science and art in clinical decision making. J. 34. Enden, T. et al. Visualization of deep veins and detection of deep vein thrombosis Clin. Oncol. 28, 1625–1627 (2010). (DVT) with balanced turbo field echo (b-TFE) and contrast-enhanced T1 fast field 61. Paik, S. Is gene array testing to be considered routine now? Breast 20, S87–S91 echo (CE-FFE) using a blood pool agent (BPA). J. Magn. Reson. Imaging 31, (2011). Suppl 3. 416–424 (2010). 62. Sparano, J. A., Fazzari, M. & Kenny, P. A. Clinical application of gene expression 35. Lim, H. K. et al. MR diagnosis of facial neuritis: diagnostic performance of profiling in breast cancer. Surgical Oncol. Clin. North Am. 19, 581–606 (2010). contrast-enhanced 3D-FLAIR technique compared with contrast-enhanced 3D- 63. Wirapati, P. et al. Meta-analysis of gene expression profiles in breast cancer: T1-fast-field echo with fat suppression. Am. J. Neuroradiol. 33, 779–783 (2012). toward a unified understanding of breast cancer subtyping and prognosis sig- 36. Roussos, E. T. et al. Mena invasive (Mena(INV)) and Mena11a isoforms play distinct natures. Breast Cancer Res. 10, R65 (2008). roles in breast cancer cell cohesion and association with TMEM. Clin. Exp. 64. De Palma, M. & Lewis, C. E. Macrophage regulation of tumor responses to Metastasis 28, 515–527 (2011). anticancer therapies. Cancer Cell 23, 277–286 (2013). 37. Arwert, E. N. et al. A unidirectional transition from migratory to perivascular 65. Hughes, R. et al. Perivascular M2 macrophages stimulate tumor relapse after macrophage is required for tumor cell intravasation. Cell Rep. 23, 1239–1248 chemotherapy. Cancer Res. 75, 3479–3491 (2015). (2018). 66. De Palma, M., Murdoch, C., Venneri, M. A., Naldini, L. & Lewis, C. E. Tie2-expressing 38. Borriello, L. et al. The role of the tumor microenvironment in tumor cell intra- monocytes: regulation of tumor angiogenesis and therapeutic implications. vasation and dissemination. Eur. J. Cell Biol. 99, 151098 (2020). Trends Immunol. 28, 519–524 (2007). 39. Patsialou, A. et al. Selective gene-expression profiling of migratory tumor cells 67. Lewis, C. E., De Palma, M. & Naldini, L. Tie2-expressing monocytes and tumor in vivo predicts clinical outcome in breast cancer patients. Breast Cancer Res. 14, angiogenesis: regulation by hypoxia and angiopoietin-2. Cancer Res. 67, R139 (2012). 8429–8432 (2007). 40. Karagiannis, G. S. et al. Assessing tumor microenvironment of metastasis 68. Roodhart, J. M. et al. Late release of circulating endothelial cells and endothelial doorway-mediated vascular permeability associated with cancer cell dissemina- progenitor cells after chemotherapy predicts response and survival in cancer tion using intravital imaging and fixed tissue analysis. J. Vis. Exp. https://doi.org/ patients. Neoplasia 12,87–94 (2010). 10.3791/59633 (2019). 69. Squadrito, M. L. & De Palma, M. Macrophage regulation of tumor angiogenesis: 41. Dasgupta, A., Lim, A. R. & Ghajar, C. M. Circulating and disseminated tumor cells: implications for cancer therapy. Mol. Asp. Med. 32, 123–145 (2011). harbingers or initiators of metastasis? Mol. Oncol. 11,40–61 (2017). 70. Chang, Y. S., Jalgaonkar, S. P., Middleton, J. D. & Hai, T. Stress-inducible gene Atf3 42. Bidard, F. C., Proudhon, C. & Pierga, J. Y. Circulating tumor cells in breast cancer. in the noncancer host cells contributes to chemotherapy-exacerbated breast Mol. Oncol. 10, 418–430 (2016). cancer metastasis. Proc. Natl Acad. Sci. USA 114, E7159–E7168 (2017). 43. Roussos, E. T. et al. Mena deficiency delays tumor progression and decreases 71. Janni, W. J. et al. Pooled analysis of the prognostic relevance of circulating tumor metastasis in polyoma middle-T transgenic mouse mammary tumors. Breast cells in primary breast cancer. Clin. Cancer Res. 22, 2583–2593 (2016). Cancer Res. 12, R101 (2010). 72. Riethdorf, S. et al. Prognostic impact of circulating tumor cells for breast cancer 44. Weidmann, M. D. et al. MenaINV dysregulates cortactin phosphorylation to patients treated in the neoadjuvant “Geparquattro” trial. Clin. Cancer Res. 23, promote invadopodium maturation. Sci. Rep. 6, 36142 (2016). 5384–5393 (2017). 45. Eddy, R. J., Weidmann, M. D., Sharma, V. P. & Condeelis, J. S. Tumor cell inva- 73. Bidard, F. C. et al. Circulating tumor cells in breast cancer patients treated by dopodia: invasive protrusions that orchestrate metastasis. Trends Cell Biol. 27, neoadjuvant chemotherapy: a meta-analysis. J. Natl Cancer Inst. 110, 560–567 595–607 (2017). (2018). Published in partnership with the Breast Cancer Research Foundation npj Breast Cancer (2022) 101 G.S. Karagiannis et al. 74. Furman, C. et al. Ena/VASP is required for endothelial barrier function in vivo. J. Myles P. Dempsey. Views, opinions of, and endorsements by the authors do not Cell Biol. 179, 761–775 (2007). reflect those of the US Army or the Department of Defense. All experiments involving 75. Kris, A. S., Kamm, R. D. & Sieminski, A. L. VASP involvement in force-mediated rebastinib administration were conducted in part using rebastinib that was kindly adherens junction strengthening. Biochem Biophys. Res. Commun. 375, 134–138 provided by Deciphera Pharmaceuticals. The authors would finally like to thank the (2008). personnel of the Histopathology Facility and Analytical Imaging Facility at the Albert 76. Sanchez, L. R. et al. The emerging roles of macrophages in cancer metastasis and Einstein College of Medicine for constant support and assistance with the use of the response to chemotherapy. J. Leukoc. Biol. 106, 259–274 (2019). facilities. 77. Asiry, S. et al. The cancer cell dissemination machinery as an immunosuppressive niche: a new obstacle towards the era of cancer immunotherapy. Front. Immunol. 12, 654877 (2021). AUTHOR CONTRIBUTIONS 78. Oskarsson, T., Batlle, E. & Massague, J. Metastatic stem cells: sources, niches, and Developed the project concept and designed experiments: G.S.K., A.B., C.A.B., and vital pathways. Cell Stem Cell 14, 306–321 (2014). J.S.C. Contributed equally as first authors: G.S.K. and A.B. Executed animal 79. Entenberg, D. et al. A protocol for the implantation of a permanent window for experiments and corresponding data analyses: G.S.K., A.B., L.R.S., K.A., M.-H.C., Y.W., high-resolution imaging of the murine lung. Protocol. Exch. https://doi.org/ A.S.H., J.M.P., Y.L., X.C., J.G.J., D.E., M.H.O., C.A.B., and J.S.C. Performed human studies 10.1038/protex.2017.134 (2017). and corresponding data analyses: G.S.K., A.B., J.M.A., S.A., D.H., L.J.H., T.Q.D., J.A.S., 80. Coste, A. et al. Hematogenous dissemination of breast cancer cells from lymph M.H.O., C.A.B., and J.S.C. Wrote manuscript: G.S.K., A.B., M.H.O., C.A.B., and J.S.C. nodes is mediated by tumor microenvironment of metastasis (TMEM) doorways. Revised and approved the final manuscript: All authors. Front. Oncol. (in the press). 81. Ginter, P. S. et al. Tumor microenvironment of metastasis (TMEM) doorways are restricted to the blood vessel endothelium in both primary breast cancers and COMPETING INTERESTS their lymph node metastases. Cancers https://doi.org/10.3390/cancers11101507 The authors declare no competing interests. (2019). 82. Brown, M. et al. Lymph node blood vessels provide exit routes for metastatic tumor cell dissemination in mice. Science 359, 1408–1411 (2018). ADDITIONAL INFORMATION 83. Pereira, E. R. et al. Lymph node metastases can invade local blood vessels, exit the node, and colonize distant organs in mice. Science 359, 1403–1407 (2018). Supplementary information The online version contains supplementary material 84. Holmberg, D. & Ahlgren, U. Imaging the pancreas: from ex vivo to non-invasive available at https://doi.org/10.1038/s41523-022-00463-5. technology. Diabetologia 51, 2148–2154 (2008). 85. Cao, L., Kobayakawa, S., Yoshiki, A. & Abe, K. High resolution intravital imaging of Correspondence and requests for materials should be addressed to Craig A. Branch subcellular structures of mouse abdominal organs using a microstage device. or John S. Condeelis. PLoS ONE 7, e33876 (2012). 86. Alieva, M., Ritsma, L., Giedt, R. J., Weissleder, R. & van Rheenen, J. Imaging win- Reprints and permission information is available at http://www.nature.com/ dows for long-term intravital imaging: General overview and technical insights. reprints Intravital 3, e29917 (2014). 87. Coste, A., Oktay, M. H., Condeelis, J. S. & Entenberg, D. Intravital imaging tech- Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims niques for biomedical and clinical research. Cytometry A https://doi.org/10.1002/ in published maps and institutional affiliations. cyto.a.23963 (2019). 88. Morse, B. & Klapman, J. Imaging of pancreatic tumors. Monogr. Clin. Cytol. 26, 21–33 (2020). 89. Mann, R. M., Kuhl, C. K. & Moy, L. Contrast-enhanced MRI for breast cancer Open Access This article is licensed under a Creative Commons screening. J. Magn. Reson. Imaging 50, 377–390 (2019). Attribution 4.0 International License, which permits use, sharing, 90. Wyckoff, J. B., Jones, J. G., Condeelis, J. S. & Segall, J. E. A critical step in metastasis: adaptation, distribution and reproduction in any medium or format, as long as you give in vivo analysis of intravasation at the primary tumor. Cancer Res. 60, 2504–2511 appropriate credit to the original author(s) and the source, provide a link to the Creative (2000). Commons license, and indicate if changes were made. The images or other third party 91. Boimel, P. J. et al. Contribution of CXCL12 secretion to invasion of breast cancer material in this article are included in the article’s Creative Commons license, unless cells. Breast Cancer Res. 14, R23 (2012). indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http:// ACKNOWLEDGEMENTS creativecommons.org/licenses/by/4.0/. This research was supported by NIH K99 CA237851; T32 CA200561; Department of Defense (W81XWH-13-1-0010); CA216248; S10 OD019961 for the use of the Perkin Elmer 250 slide scanner; the Gruss-Lipper Biophotonics Center and its Integrated © The Author(s) 2022 Imaging Program; the Evelyn Gruss-Lipper Charitable Foundation, and Jane A. and npj Breast Cancer (2022) 101 Published in partnership with the Breast Cancer Research Foundation

Journal

npj Breast CancerSpringer Journals

Published: Sep 2, 2022

There are no references for this article.