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

Learn More →

Medial temporal pathways for contextual learning: Network c-fos mapping in rats with or without perirhinal cortex lesions:

Medial temporal pathways for contextual learning: Network c-fos mapping in rats with or without... Background: In the rat brain, context information is thought to engage network interactions between the postrhinal cortex, medial entorhinal cortex, and the hippocampus. In contrast, object information is thought to be more reliant on perirhinal cortex and lateral entorhinal cortex interactions with the hippocampus. Method: The ‘context network’ was explored by mapping expression of the immediate-early gene, c-fos, after exposure to a new spatial environment. Results: Structural equation modelling of Fos counts produced networks of good fit that closely matched prior predictions based on anatomically grounded functional models. These same models did not, however, fit the Fos data from home-cage controls nor did they fit the corresponding data from a previous study exploring object recognition. These additional analyses highlight the specificity of the context network. The home-cage controls, meanwhile, showed raised levels of inter-area Fos correlations between the many sites examined, that is, their changes in Fos levels lacked anatomical specificity. A total of two additional groups of rats received perirhinal cortex lesions. While the loss of perirhinal cortex reduced lateral entorhinal c-fos expression, it did not affect mean levels of hippocampal c-fos expression. Similarly, overall c-fos expression in the prelimbic cortex, retrosplenial cortex, and nucleus reuniens of the thalamus appeared unaffected by the perirhinal cortex lesions. Conclusion: The perirhinal cortex lesions disrupted network interactions involving the medial entorhinal cortex and the hippocampus, highlighting ways in which perirhinal cortex might affect specific aspects of context learning. Keywords Entorhinal cortex, hippocampus, nucleus reuniens, prefrontal cortex, retrosplenial cortex, spatial memory, subiculum Received: 6 December 2016; accepted: 25 January 2017 Introduction Models of medial temporal lobe processing increasingly assume activate one of these processing pathways. This IEG, which pro- two distinct functional pathways (Figure 1), one for object-based vides an indirect marker of neural activity (Bisler et al., 2002; information and the other for spatial and contextual information Chaudhuri, 1997; Zangenehpour and Chaudhuri, 2002), is known (Bucci and Robinson, 2014; Burwell, 2000; Eacott and Gaffan, to show increased hippocampal activity following contextual 2005; Diana et al., 2007; Knierim et al., 2014; Ranganath and change (e.g. Albrechet-Souza et al., 2011; Jenkins et al., 2004; Ritchey, 2012; Ritchey et al., 2015). In the case of the rodent Vann et al., 2000). The importance of this c-fos expression is brain, perirhinal cortex is presumed to process object-based strikingly highlighted by studies showing how reactivating those information, in concert with the lateral entorhinal cortex (LEC; dorsal hippocampal neurons that had previously expressed c-fos Barker et al., 2007; Mumby and Pinel, 1994; Naber et al., 1997). in a distinctive context can reinstate representations of that same In contrast, postrhinal cortex, along with the medial entorhinal context (Liu et al., 2012, 2014; Ramirez et al., 2013). Structural cortex (MEC), is presumed to provide spatial and contextual equation modelling (SEM; McIntosh and Gonzalez-Lima, 1991; information for the hippocampus (Bucci et al., 2000; Bucci and Schumacker and Lomax, 2010) was then used to test anatomically Robinson, 2014; Burwell, 2000; Furtak et al., 2007; Norman and Eacott, 2005; Ranganath and Ritchey, 2012). These distinct, functional pathways are highlighted in a number of models, School of Psychology, Cardiff University, Cardiff, UK including the binding of item and context (BIC; Diana et al., The Institute of Medical Sciences, University of Aberdeen, Aberdeen, UK 2007) framework as well as in subsequent anatomical refine- ments of this basic model (Aggleton, 2012; Ranganath and Corresponding author: Ritchey, 2012; Ritchey et al., 2015). Lisa Kinnavane, School of Psychology, Cardiff University, 70 Park This study quantified the expression of the immediate-early Place, Cardiff CF10 3AT, UK. gene (IEG) c-fos after placing rats in a novel context in order to Email: kinnavanel@cardiff.ac.uk Creative Commons CC BY: This article is distributed under the terms of the Creative Commons Attribution 3.0 License (http://www.creativecommons.org/licenses/by/3.0/) which permits any use, reproduction and distribution of the work without further permission provided the original work is attributed as specified on the SAGE and Open Access pages (https://us.sagepub.com/en-us/nam/open-access-at-sage). 2 Brain and Neuroscience Advances Figure 1. Posterior–medial (PM) context system and anterior–temporal (AT) item system of the binding of item and context (BIC) framework. Parallel cortico-hippocampal pathways link the PM and AT systems with the entorhinal cortex, CA1, and subiculum. Source: Adapted from Ritchey et al. (2015). plausible models based on refinements of the BIC framework at the beginning of the c-fos imaging study. Of these rats, 18 had (Ritchey et al., 2015). received perirhinal cortex lesions, while 11 served as their surgi- As already noted, current medial temporal models typically cal controls. Rats from cohort B (n = 27) were approximately assume the presence of two parallel pathways that emanate, 7 months old at the beginning of the present experiment. Of these, respectively, from the perirhinal and parahippocampal (postrhi- 15 received perirhinal cortex lesions, while 12 received sham nal) cortices to reach the hippocampal formation (Diana et al., surgeries. Prior to the current experiment, both cohorts received 2007; Witter, 2002). There are, however, reciprocal connections object recognition memory tasks in the bow-tie maze (full details between the postrhinal and perirhinal cortices and between are described in Albasser et al., 2015). Additionally, both cohorts the MEC and LEC (Burwell and Amaral, 1998a, 1998b; Furtak received a single, spontaneous object recognition test in an open- et al., 2007). These interconnections question the degree of inde- field apparatus. Rats were not behaviourally tested for at least pendence between these two functional pathways (Kent and 2 weeks before the current experiment, with water and food Brown, 2012; Knierim et al., 2014; Liu and Bilkey, 1998b, available ad libitum throughout this intervening period. 2001). For this reason, this study also examined how perirhinal cortex lesions might affect parahippocampal–hippocampal c-fos Surgery activity following exposure to a novel context. The question was whether the context pathway still shows normal activity patterns The rats were approximately 3 months old at the time of surgery, in the absence of perirhinal cortex. Finally, other intact rats and when they weighed between 285 and 300 g. In total, 33 rats rats with perirhinal cortex lesions were examined for c-fos received bilateral perirhinal cortex lesions (‘Peri’), while 23 rats expression after receiving no context shift, so providing a com- served as surgical controls (‘Sham’). Anaesthesia was induced parison baseline condition. using a mixture of oxygen and isoflurane gas (5% for induction and 2% thereafter), before placing each rat in a stereotaxic frame (David Kopf Instruments, Tujunga, CA, USA), with the incisor Materials and methods bar set at +5.0 mm above the horizontal plane. After making a midline sagittal incision in the scalp, the skin was retracted to Animals expose the skull. Following a craniotomy, the perirhinal lesions Subjects were 56 male, Lister Hooded rats (Harlan, Bicester, were made by injecting a solution of N-methyl-d -aspartate UK), housed in pairs under diurnal conditions (12-h light/12-h (NMDA; Sigma, Poole, UK) diluted to 0.09 M in phosphate- dark). The home cages measured 42 cm × 25 cm × 21 cm with a buffered saline (PBS; 0.1 M, pH 7.4) using a 1-µm Hamilton water bottle and food hopper at the front. Each cage, which had syringe (Bonaduz, Switzerland; gauge 26, outside diameter opaque plastic floors and walls (13 cm high), was lined with 0.47 mm) held with a micro-injector (model 5000; Kopf sawdust and contained a cardboard tube and chew stick. The Instruments, Tujunga, CA). Bilateral injections of NMDA (each rats, which were fed 2014 Teklad global 14% protein rodent of 0.225 µL) were made at a rate of 0.10 µL/min, with a subse- maintenance diet (Harlan), were on an ad libitum schedule. All quent diffusion time of 4 min before the needle was removed. experiments were performed in accordance with the UK Animals Three injections were made in each hemisphere. Injection coor- (Scientific Procedures) Act, 1986, and associated guidelines and dinates relative to bregma (in mm) were (1) anterior–posterior approved by local ethical committees at Cardiff University. (AP): −1.8, medial–lateral (ML): ±5.9, and dorsal–ventral (DV): The rats came from two cohorts of animals, which received −9.3; (2) AP: −3.4, ML: ±6.1, and DV: −9.6; (3) AP: −5.0, ML: the same experimental protocols throughout the present experi- ±6.2, and DV: −9.0. Rats in the surgical control group received ment. Rats in cohort A (n = 29) were approximately 11 months old identical treatments, except that the dura was perforated three Kinnavane et al. 3 times per hemisphere with a 25-gauge Microlance 3 needle Lesion analysis (Becton Dickinson, Drogheda, Ireland) and no fluid was infused Only one hemisphere in each Peri brain was analysed for Fos into the brain. expression, while the other was eliminated. This procedure ensured that Fos counts were only taken from those hemispheres with either no evidence of surgically induced hippocampal cell Apparatus – activity boxes loss or with loss restricted to just one coronal section (typically in For the novel context condition, rats were placed individually in the hippocampal subfield, CA1). Brains that suffered damage to an activity cage in a novel room (272 cm × 135 cm × 240cm). both hippocampi were excluded from the study. In those hemi- A 3 × 6 bank of activity cages was located along one wall of the spheres analysed for study, the boundaries of the lesions were room. Each activity cage (Paul Fray, Cambridge, UK) measured drawn onto five coronal plates (bregma: −2.80 to −6.72 mm) from 56 cm × 39 cm × 19 cm and contained two photobeams placed Paxinos and Watson (2005). These images were then scanned, and 20 cm apart, positioned 18 cm from the short walls. The floor of the area of damage was calculated using cellSens Dimension each cage was made of wire; otherwise, the cage was empty. The Desktop, version 1.12 (Olympus, Southend-on-Sea, UK). top of each cage was also made of wire, and the room was illuminated. Immunohistochemistry Brain sections were initially stored at −20°C in cryoprotectant. Behavioural testing Free-floating sections were then immunohistochemically stained Both the Peri and Sham animals were divided between the two with sections from one rat from each of the four behavioural behavioural conditions, creating four groups. Rats with perirhinal groups placed in the same reaction vessel, that is, sections from lesions were assigned to either the novel context condition (Peri all four groups were processed concurrently. This arrangement Novel, n = 18; nine from cohort A, nine from cohort B) or the sought to decrease staining variation between groups. The sec- home-cage control condition (Peri Baseline, n = 15; 9 from cohort tions were first washed six times in PBS to remove the cryopro- A, 6 from cohort B). Similarly, the sham surgical controls were tectant, then washed in 0.2% Triton-X 100 in 0.1M PBS (PBST), divided between the novel context condition (Sham Novel, once in 1% H O in PBST (to block endogenous peroxidases), 2 2 n = 11; 4 from cohort A, 7 from cohort B) and the home-cage and then four further times in PBST. The sections were then control condition (Sham Baseline, n = 12; 7 from cohort A, 5 incubated in a blocking solution of 3% normal goat serum from cohort B). Behavioural testing (activity boxes) took place (NGS) in PBST for 1 h followed by the primary antibody solu- either 8 (cohort A) or 4 (cohort B) months after surgery. tion; rabbit-anti-c-fos (1:15,000) and 1% NGS diluted in PBST For the novel context condition, each rat was placed indi- (Cat# PC38; Calbiochem; now part of Merek Millipore, vidually in a dark holding room for 30 min (to which they had Nottingham, UK), for 48 h at 4°C. The sections were then received two 30 min familiarisation sessions on the preceding washed four times in PBST, before being incubated in the sec- days). They were then taken into the novel test room and placed ondary antibody solution; biotinylated goat-anti-rabbit (1:200; individually inside an activity test cage for 20 min. (Due to an Vector Laboratories, Peterborough, UK) diluted in 1.5% NGS in equipment malfunction, two activity scores (one Peri, one Sham) PBST for 2 h at room temperature. The sections were washed were not recorded.) These rats were then returned to the dark four times in PBST. They were then incubated in avidin- holding room. For the baseline condition, individual rats biotinylated horseradish peroxidase complex in PBST (Elite kit; remained throughout in their home cages without exposure to Vector Laboratories) for 1 h at room temperature. The sections the dark room prior to perfusion. were washed four times in PBST and then twice in 0.05 M Tris buffer (pH 7.4). All washes were 10 min unless otherwise stated. Finally, diaminobenzidine (DAB substrate kit; Vector Perfusion and tissue sectioning Laboratories) was used as the chromogen to visualise the loca- tion of immunostaining. The reaction was stopped in cold PBS. For the novel context condition, rats were perfused 90 min after The sections were mounted onto double gelatine-subbed glass being returned to the dark holding room. This interval is within slides and allowed to air dry for at least 48 h, dehydrated in the time period when the expression of Fos, the protein product increasing concentration of alcohol washes, cleared in xylene, of c-fos, peaks, that is, between 60 and 120 min after the induc- and coverslipped using DPX as the mounting media. ing event (Bisler et al., 2002; Zangenehpour and Chaudhuri, 2002). Animals in the baseline control condition were taken directly from their home cage immediately prior to perfusion. Image capture and analysis of c-fos activation All rats then received a lethal overdose of sodium pentobarbital (60 mg/kg, Euthatal, Marial Animal Health, Harlow, Essex, UK) Images from each region of interest (ROI) were captured from and were transcardially perfused with 0.1 M PBS followed by six consecutive sections (each 120 µm apart) from one hemi- 4% paraformaldehyde in 0.1 M PBS (PFA). Brains were removed sphere per animal. The equivalent hemisphere (left or right) was from the skull, postfixed in PFA for 4 h, and then incubated in also analysed in the corresponding ‘Sham’ control animal. Image 25% sucrose at room temperature overnight on a stirrer plate. capture used a 5× objective lens (numerical aperture of 0.12) on The brains were cut in the coronal plane into 40-µm sections a Leica DMRB microscope with an Olympus DP70 camera. The using a freezing microtome. A series of one in four sections field of view was 0.84 × 0.63 mm, so that cortical regions only were collected in PBS and then stained with cresyl violet required one image per section to include all lamina. For the hip- (a Nissl stain), while another one in four series was retained pocampus, multiple images were taken and combined (Microsoft for immunohistochemistry. Image Composite Editor (ICE); Microsoft, Redmond, WA, 4 Brain and Neuroscience Advances USA). Using ANALYSIS^D software (Soft-Imaging Systems, Olympus Corporation). Fos-positive cells were quantified by counting the number of immunopositive nuclei (mean Feret diameter of 4−20 µm) stained above a greyscale threshold set 60−70 units below the peak grey value measured by a pixel inten- sity histogram. ROIs The borders of the perirhinal and postrhinal cortices follow the description of Burwell and Amaral (1998a; see also Burwell, 2001), while those of the other brain areas correspond to Swanson (1992). The AP coordinates (mm from bregma) given in the descriptions below and in Figure 2 are from Paxinos and Watson (2005). The regional groupings are those subsequently used in the statistical analyses of Fos counts. Hippocampal subfields. Hippocampal subfields (dentate gyrus, CA1, and CA3) were subdivided into their septal (dorsal), inter- mediate (dorsal), and temporal (ventral) divisions(Bast, 2007; Strange et al., 2014). The septal hippocampus counts (dentate gyrus, CA3, and CA1) were obtained from sections from AP −2.52 to −3.24, while those for the intermediate dorsal hippo- campus (dentate gyrus, CA1, and CA3) came from sections Figure 2. Regions of interest for c-fos analyses. Sites examined: CA between AP −4.80 and −5.52. The border between the dorsal fields – intermediate (inter), septal (sept), and temporal (temp); intermediate and temporal hippocampus corresponds to −5.0 mm dentate gyrus (DG); dorsal subiculum (dorsal Sub); lateral entorhinal ventral from bregma (see Figure 2; Paxinos and Watson, 2005). cortex (LEC); medial entorhinal cortex (MEC); prelimbic cortex (PL); Within the temporal hippocampus, counts were made in the CA1 perirhinal cortex (PRH); postrhinal cortex (POR); nucleus reuniens of and CA3 fields at the same AP as the intermediate dorsal hippo- thalamus (Reuniens); retrosplenial cortex (RSP); and ventral subiculum campus (note that the dentate gyrus is not present at this level). (ventral Sub). The numbers below refer to the approximate distance in Additional Fos-positive cell counts were taken in both the dorsal millimetre from bregma. and ventral subiculum (from around AP −5.16). Source: Adapted from the atlas of Paxinos and Watson (2005). Parahippocampal cortices. Separate Fos-positive cell counts were taken from the LEC and MEC, as well as the postrhinal repeatedly broken, and ‘beam crossovers’, that is, the front and cortex. The LEC counts were taken from more caudal parts back beams broken sequentially. These data were compared with of the area to ensure that there was no encroachment from the a one between-subject factor (surgical condition) and one within- perirhinal lesion in the Peri groups. In the Sham cases only, Fos- subject factor (‘same beam’ or ‘beam crossovers’) analysis of positive cell counts were made in the caudal perirhinal cortex variance (ANOVA). The total number of beam breaks across the (areas 35 and 36; see Burwell, 2001). This caudal portion (from 20-min exposure to the novel context was then divided into four AP −4.80 to −5.52) was selected as previous studies indicate that bins of 5 min and compared with a one between-subject factor this region is particularly involved in processing novel visual (surgical condition) and one within-subject factor (bin). stimuli (Albasser et al., 2009, 2010; Kinnavane et al., 2014; Olarte-Sánchez et al., 2014). Fos-positive cells counts. To analyse group differences (Sham vs Peri lesion; baseline vs novel context) in the ROIs, two Other hippocampal-related areas. Fos-positive cell counts were between-subject factors (surgical condition and baseline/novel made within the prelimbic cortex (PL; AP +3.72 to +2.76), the gran- context) and one within-subject factor (ROI) ANOVA was calcu- ular retrosplenial cortex (RSP; AP −2.28 to −3.36), and nucleus lated. This analysis was carried out separately for three regional reuniens (AP −1.44 to −2.28). The granular retrosplenial cortex groupings: (1) hippocampal subfields, (2) parahippocampal cor- (area 29) was selected as it is both the principal recipient of the tices, and (3) other hippocampal-related areas. These regional projections from the hippocampal formation to this region and the groupings helped to reduce type 1 errors by limiting the number source of its projections to entorhinal cortex (Van Groen and Wyss, of comparisons. The Fos counts in perirhinal cortex (Sham 1990, 1992, 2003). There are also direct projections from the tem- groups only) were compared using a one between-subject (base- poral region of CA1 and the subiculum to prelimbic cortex, with line/novel context) by one within-subject factor (areas 35 and 36) return projections via nucleus reuniens of the thalamus (Conde ANOVA. When an interaction was significant (p ⩽ 0.017 cor- et al., 1995; Prasad and Chudasama, 2013; Vertes et al., 2007). rected for multiple tests), simple effects were examined. While the Fos counts from the novel groups were normally distributed, cell counts from both baseline control groups (‘Peri Statistical analysis Baseline’ and ‘Sham Baseline’) were not (Shapiro–Wilk test). Behavioural results. Activity scores from the ‘novel’ groups As the baseline Fos counts in all ROIs were positively skewed were separated between ‘same beam’, that is, a single beam being and their means were proportional to their variance, a square-root Kinnavane et al. 5 transformation was applied to the data (Howell, 2011) when the McIntosh, 2006; Schumacker and Lomax, 2010). Subsequently, analyses involved only these groups. Other analyses involving all each path can be independently unconstrained and the fit com- four groups used the raw Fos counts for comparability, mindful pared to the structural weights model, again using a χ difference that ANOVA is relatively robust to violations of the normality test to determine in which path the difference occurs (Protzner assumption when group sample sizes are equal (Howell, 2011). and McIntosh, 2006). When there were marked differences in the Pearson product-moment correlation coefficients were calcu- overall fit of the same model between two or more groups, the lated for the Fos-positive cell counts in the various sites, as well alpha level of the first χ difference test was slightly relaxed in as with the activity of animals in the novel context condition. order to explore the potential reasons why one group had poor fit. In both baseline control groups, the Pearson product-moment correlation coefficients were calculated based on the trans- Results formed scores as these data were subsequently used for SEM, where normality is assumed (Arbuckle, 2011). The histological and behavioural analyses only relate to those animals used for the c-fos analyses. As explained, the findings came from two cohorts of rats. The data from these two cohorts Structural equation modelling were repeatedly compared, though the outcome is only presented when there was a significant cohort difference (p ⩽ 0.05). Rats SEM uses multiple-equation regression models to quantify from both cohorts populated all conditions. potentially causal relationships between sets of variables in a theoretical structure, thereby testing models that can include the potential direction of effects (Schumacker and Lomax, 2010). Lesion analysis (In some cases, a direction of effect could not be inferred as the Based on the exclusion criteria (see section ‘Lesion analysis’), fit of the models did not change when the path direction was seven animals were removed from group Peri Novel and three reversed. This situation is indicated in the figures by a double- were excluded from Peri Baseline. Following these exclusions, headed arrow.) The SEM software package, SPSS AMOS version the group numbers were as follows: Peri Novel, n = 11 (3 from 20.0 (IBM Corp, Armonk, NY, USA) computed the analyses. cohort A, 8 from cohort B); Peri Baseline, n = 12 (6 from cohort Maximum likelihood estimation, which is recommended for use A, 6 from cohort B); Sham Novel, n = 11 (4 from cohort A, with smaller sample sizes (Arbuckle, 2011), allowed the pro- 7 from cohort B); Sham Baseline, n = 12 (7 from cohort A, gramme to estimate effects among variables. All models tested 5 from cohort B). Group Peri Novel contained six left and five were based on well-established anatomical connections (Furtak right hemispheres. For group Peri Baseline, three left and nine et al., 2007; Van Strien et al., 2009; Witter, 2002). right hemispheres were analysed. The corresponding hemi- An anatomically plausible model was specified and the covar- spheres were analysed in the matched Sham surgical controls. iance matrix of the regional Fos counts estimated the strength The lesions involved much of the AP extent of perirhinal cor- of the relationship (path) between regions as set out in this tex, with only small regions of tissue sparing (Figure 3). The model. The path coefficient of a connection between two regions extent of perirhinal tissue loss in those hemispheres analysed for (Arbuckle, 2011) estimates the ‘effective connectivity’ or the Fos expression in the Peri Novel context condition ranged from extent to which one region directly influences the other (Protzner 40.6% to 73.9% (cohort A) and 67.8% to 98.5% (cohort B). The and McIntosh, 2006). Models were assessed based on how well corresponding extent of tissue loss in the Peri Baseline condition the implied (estimated) variance–covariance matrix replicates ranged from 70.7% to 89.9% (cohort A) and 68.6% to 100% the sample (observed) variance–covariance matrices of the (cohort B). The extent of tissue loss did not differ between the observed data (Schumacker and Lomax, 2010). A model with 2 2 Peri Novel and Peri Baseline groups (t < 1). good fit has a non-significant χ , and the ratio of χ to the degrees The attempt to make near-complete perirhinal cortex lesions of freedom is <2 (Tabachnick and Fidell, 2001). The comparative led to some extra-perirhinal damage. This additional damage was fit index (CFI) and the root mean square error of approximation typically in the most ventral parts of area Te2 and the most dorsal (RMSEA) are additional measures of fit that are applicable for parts of the piriform cortex and LEC, that is, those cortical areas smaller sample sizes (Fan and Wang, 1998; Hu and Bentler, immediately adjacent to perirhinal cortex (Figure 3). 1998). A CFI of >0.9 is considered acceptable (Schumacker and Lomax, 2010), while a RMSEA of <0.08 is considered accepta- ble (Tabachnick and Fidell, 2001). Given the relatively small Behavioural testing group sizes, each model should contain twice as many cases as the number of variables to be estimated (Bollen and Long, 1992; Analyses of the beam breaks over the 20 min session (‘same Wothke, 1993). Finally, the squared multiple correlation (R or beam’ and ‘beam crossovers’) found no overall effect of coefficient of determination) is presented, indicating the amount perirhinal cortex lesions (F(1, 18) = 1.36, p = 0.26). Similarly, of variance in each brain region accounted for by the model there was no interaction between lesion and type of beam break (Arbuckle, 2011). (F(1, 18) = 3.21, p = 0.09). When the beam breaks were divided The various groups were compared on the same network into 5 min bins, the activity levels showed a highly significant model by a stacking procedure. In this procedure, the path coef- reduction across the 20 min session (F(3, 36) = 11.4, p < 0.001), ficients of all paths in the model are constrained so that they must with no effect of surgery (F < 1) and no interaction between have the same value for all groups, creating a ‘structural weights these factors (F < 1). Both the perirhinal lesion and sham control model’. If the model fit when the paths are constrained is signifi- rats showed a significant decrease in activity. (Note: these data cantly worse than when the paths are free to have different values were only available for 14 rats from cohort B.) This reduction for each group (as determined by a χ difference test), this indi- in activity is assumed to principally reflect habituation to a cates that the paths differ among the groups (Protzner and novel environment. 6 Brain and Neuroscience Advances Figure 3. Perirhinal lesion reconstructions. Diagrammatic reconstructions of the perirhinal cortex lesions showing the individual cases with the largest (grey) and smallest (black) lesions for rats from cohorts A and B in groups Peri Novel and Peri Baseline. The left panel illustrates regions involved for comparison. Sites highlighted: areas 35 and 36 of the perirhinal cortex, insular cortex, lateral entorhinal cortex (LEC), and piriform cortex. The numbers refer to the distance (in mm) from bregma (note that the hemispheres analysed came from both right and left hemispheres). Source: Adapted from Paxinos and Watson (2005). context seemingly differed among subfields (this novelty dif- Comparisons of Fos-positive cell counts ference was highly significant in all subfields, F(1, 42) > 27, Hippocampal subfields. Being placed in the novel context dra- p ⩽ 0.001; Figure 4, upper panel). While this increase seemed matically increased c-fos activity, although the perirhinal cortex most evident in CA1, scaling effects were present. Comparisons lesions had no apparent effect on the mean Fos counts in the of Fos-positive cell counts across the 10 hippocampal subfields hippocampal formation (Figure 4). A significant Mauchly test found no overall effect of perirhinal lesions (F(1, 42) = 1.43, (p ⩽ 0.001) indicated that the assumption of sphericity of the p = 0.24; Figure 4, upper panel). Similarly, there was no lesion by within-subject variable (ROI) was violated and so Greenhouse– context interaction (F < 1), lesion by subfield interaction (F(2.8, Geisser corrected degrees of freedom and p-values are presented 116) = 1.12, p = 0.34), or three-way interaction (F < 1). for the within-subject comparisons (Howell, 2011). Hippocampal Fos counts in the novel context rats were con- Parahippocampal cortices. Overall, novel context exploration sistently considerably higher than those of the rats in the baseline produced higher Fos counts in the MEC, LEC, and postrhinal (home-cage) controls (F(1, 42) = 166, p ⩽ 0.001), with individual cortex than remaining in the home cage (F(1, 40) = 113, p ⩽ 0.001; areas showing different levels of Fos expression (F(2.8, Figure 5(a)). Rats with perirhinal cortex lesions had lower 116) = 101.1, p ⩽ 0.001; Figure 4, upper panel). There was a Fos-positive cell counts across the parahippocampal cortices, significant context by subfield interaction (F(2.8, 116) = 48.2, with the LEC seemingly most affected (Figure 5(a)). As in the p ⩽ 0.001) as the increase in Fos counts from baseline to novel hippocampus, the assumption of sphericity was violated, and so Kinnavane et al. 7 Figure 4. Mean Fos-positive cell counts per group in the hippocampal formation. Top panel: Graph of results from all hippocampal sites analysed: CA fields – intermediate (inter), septal (sept), and temporal (temp); dentate gyrus (DG); and subiculum (Sub). Exposure to a novel context reliably increased Fos-related activity in all regions analysed (p < 0.001). Data are presented as means ± SEM. Middle panel: Representative photomicrographs from coronal sections that depict Fos-positive cells in intermediate and temporal levels of the hippocampus for all behavioural conditions. Scale bar: 1000 µm. Bottom panel: Higher magnification photomicrographs of regions corresponding to the dashed rectangle of the photomicrograph above. Scale bar: 50 µm. Greenhouse–Geisser corrected degrees of freedom and p-values numerically greater Fos increase in area 35 (F(1, 21) = 44.2, are presented (Howell, 2011). p ⩽ 0.001) than area 36 (F(1, 21) = 25.5, p ⩽ 0.001; Figure 5(b)). The context manipulation differentially affected the three Across the three parahippocampal regions analysed in all four cortical areas (F(1.2, 49) = 113.5, p ⩽ 0.001) although simple groups, Fos counts were numerically lower in the Peri rats than in effects revealed that MEC, LEC, and postrhinal cortex all had the Sham controls (Figure 5(a)). While this contrast did not reach higher Fos-positive cell counts when exposed to the novel con- the corrected levels of significance (F(1, 40) = 5.41, p = 0.025), text compared to baseline (all F > 46.8, p ⩽ 0.001; Figure 5(a)). there was a significant region by lesion interaction (F(1.2, Numerically, this increase appeared greatest in LEC, although 49) = 6.00, p = 0.013). This interaction indicated that the various this may have been due to scaling effects generated by the com- parahippocampal areas were differentially affected by the per- paratively higher Fos counts in LEC than in MEC or the pos- irhinal lesions. Simple effects revealed that this interaction trhinal cortex. For perirhinal cortex (Figure 5(b)), counts could reflected decreased Fos counts in the LEC of the rats with per- only be made in the two Sham groups. Once again, the novel irhinal lesions (F(1, 40) = 6.56, p = 0.014; Figure 5(a)), a lesion context condition raised Fos counts (F(1, 21) = 41.7, p ⩽ 0.001; effect that did not extend to the MEC or postrhinal cortex Figure 5(b)). Overall, the Fos counts in areas 35 and 36 did (F(1, 40) = 3.44, p = 0.071; F(1, 40) = 1.59, p = 0.21, respectively). not differ (F < 1), although the context manipulation affected Finally, the three-way interaction (area, lesion, and context) was the two areas differently (F(1, 21) = 9.84, p = 0.005) with a non-significant (F(2, 80) = 1.61, p = 0.21). 8 Brain and Neuroscience Advances Figure 5. Mean Fos-positive cell counts per group in the parahippocampal formation. (a) Sites analysed in all four groups: lateral entorhinal cortex (LEC), medial entorhinal cortex (MEC), and postrhinal cortex (POR). (b) Sites analysed in only the Sham controls: areas 35 and 36 of the perirhinal cortex. Exposure to a novel context reliably increased Fos-related activity in all regions analysed (p < 0.001). Data are presented as means ± SEM. Other hippocampal-related areas. While exposure to a novel context dramatically increased Fos expression (F(1, 42) = 74.7, p ⩽ 0.001) in all three areas (Figure 6), the perirhinal cortex lesions did not affect overall Fos activity in the prelimbic cortex, retrosplenial cortex, or nucleus reuniens of the thalamus (F < 1). While Fos counts differed between areas (F(2, 84) = 105.9, p ⩽ 0.001), there was no lesion by context interaction (F < 1) or region by lesion interaction (F < 1). Although the context by region interaction was significant (F(2, 84) = 49.2, p ⩽ 0.001), with the retrosplenial cortex showing a greater increase than prelimbic cortex, the relatively low counts in nucleus reuniens created a scaling effect. Activity behaviour and Fos-positive cell counts. For each of the two relevant groups (Sham Novel and Peri Novel), only one site had an initial significant correlation (p < 0.05). In both cases, the Fos-positive cell counts correlated positively with ‘same beam’ breaks (Sham Novel, intermediate CA3, r = 0.75, p = 0.012: Peri Novel, temporal CA3, r = 0.63, p = 0.038). However, in neither case did these effects survive correction for multiple comparisons, suggesting that the Fos counts were not a direct Figure 6. Mean Fos-positive cell counts per group in other hippocampal- product of the amount of locomotor activity. related areas. Sites analysed: prelimbic cortex (PL), retrosplenial cortex (RSP), and nucleus reuniens of the thalamus (Reuniens). Exposure to a novel context reliably increased Fos-related activity in all regions Structural equation modelling analysed (p < 0.001). Data are presented as mean ± SEM. Initial inspection of all of the inter-area correlations revealed an apparent difference between the novel context and baseline direction of these connections. A valid model of context learning control conditions. For both the Sham Baseline group (120/170) would be expected to have good fit for the novel context condi- and the Peri Baseline group (87/120), a large majority of the tion but not the baseline (home-cage) condition. inter-area Fos count correlations were significant at an uncor- rected level (~70% at p < 0.05). In contrast, for both the Sham 1. Is novel context exposure associated with specific network Novel group (41/170) and the Peri Novel group (32/120), the patterns of c-fos activity predicted by the BIC framework and is corresponding proportion was much lower (~25%, i.e. almost this affected by perirhinal cortex damage? The first model to be one-third of the number). tested used the parahippocampal (postrhinal)–medial entorhinal For SEM, all of the networks examined had to have anatomi- network described by Ritchey et al. (2015). In this refined cal plausibility with respect to their interconnections and the Kinnavane et al. 9 Figure 7. Testing the posterior–medial system of the BIC framework. (a) The posterior–medial system has good fit for group Sham Novel. (b) The same network model for group Peri Novel has poor fit. The same model also has poor fit for (c) group Sham Baseline and (d) group Peri Baseline. Model fit is noted at the bottom of each model (comparative fit index (CFI); root mean square error of approximation (RMSEA)). The strength of the causal influence of each path is denoted both by the thickness of the arrow and by the path coefficient next to that path. The number above the top right corner of each area box is the R value, denoting the variance accounted for by the inputs to that region. Sites depicted: medial entorhinal cortex (MEC); postrhinal cortex (POR); retrosplenial cortex (RSP); and hippocampal subfields CA1, CA3, and dorsal subiculum (sub). (Models with a grey background have poor fit.) *p < 0.05; **p < 0.01; ***p < 0.001. version of the BIC framework, interactions between the para- Fos activity data of the two groups did not differ between the hippocampal (postrhinal) cortices and retrosplenial cortex are regions set out in the model, this contrast was close to the level of included, creating what is referred to as the posterior–medial significance, and as the model had poor fit for group Sham (PM) system (Figure 1). For this initial analysis, the Fos counts Baseline, further examination took place. When the pathways along the longitudinal hippocampal axis were combined, that is, that compose the model were individually unconstrained, the the temporal, intermediate, and septal subregions of the dentate functional connection between postrhinal cortex and MEC was χ = 49 . 2 gyrus, CA3, and CA1. This decision reflects the way in which a found to be stronger in the Sham baseline group ( , 1Diff coronal section across entorhinal cortex will include connections p = 0.027), while the functional connection between CA3 and χ = 66 . 7 along the full longitudinal axis of the hippocampus (Furtak et al., CA1 was stronger in group Sham Novel ( , p = 0.009; 1Diff χ < 19 . 2007; Van Strien et al., 2009). A maximum of six nodes could be all other paths: ). 1Diff included in each model given the sample size (Bollen and Long, The specificity of the PM system was tested in two further 1992; Wothke, 1993). ways. First, we tested the complementary item division of the This PM system, which is depicted in Figure 7, was found to updated BIC framework, that is, the anterior–temporal (AT) χ = 6.26 have good fit for group Sham Novel context ( , p = 0.51; system (Figure 1; see Ritchey et al., 2015). Fos counts from the CFI = 1.0; RMSEA = 0.0; Figure 7(a)). In contrast, the same perirhinal entorhinal cortex and LEC replaced those from the χ =10.3 model did not fit the Sham Baseline Fos data ( , p = 0.17; postrhinal entorhinal cortex and MEC, while the ventral sub- CFI = 0.96; RMSEA = 0.21; Figure 7(c)). When compared iculum replaced the dorsal subiculum. This AT model had only χ = 67 . 4 directly by stacking the data from these two groups on the same poor fit for group Sham Novel ( , p = 0.15; CFI = 0.71; model, the model in which the path coefficients were all free to RMSEA = 0.26). vary did not have significantly better fit than the model in which Second, data were taken from a previous c-fos experiment all path coefficients were constrained to be the same for both that matched this study in all respects, except for one critical fea- χ =13.87 groups ( , p = 0.085). While this indicates that the ture. Rats in that experiment were exposed to multiple novel 8Diff 10 Brain and Neuroscience Advances Figure 8. Optimal model for group Sham Novel. (a) This network, which is expanded to include prelimbic cortex, has optimal fit for data from group Sham Novel. (b) The same network for group Peri Novel has poor fit. The same model also has poor fit for group (c) Sham Baseline and (d) group Peri Baseline. Model fit is noted at the bottom of each model (comparative fit index (CFI); root mean square error of approximation (RMSEA)). The strength of the causal influence of each path is denoted both by the thickness of the arrow and by the path coefficient next to that path. The number above the top right corner of each area box is the R value, denoting the variance accounted for by the inputs to that region. Sites depicted: medial entorhinal cortex (MEC); prelimbic cortex (PL); retrosplenial cortex (RSP); and hippocampal subfields CA1, CA3, and dorsal subiculum (sub). (Models with a grey background have poor fit.) *p < 0.05; **p < 0.01; ***p < 0.001. object recognition problems in a familiar environment (Kinnavane early in the model. All of the paths in the model were significant, et al., 2016). Consequently, activity should be biased towards and all paths have directionality as the fit of the model was worse the AT item system, that is, the perirhinal cortex and LEC. when the direction of each path was reversed. It should be noted Consistent with that prediction, models based on the AT system that the hippocampal Fos data presented here are counts com- had good fit (Kinnavane et al., 2016). However, when the bined along the longitudinal hippocampal axis. If dorsal CA3 and postrhinal–medial entorhinal network was tested using the Fos CA1 counts are substituted for the combined counts, the model counts from that same object recognition study (Kinnavane et al., retains acceptable but inferior fit. Whereas if ventral (temporal) χ =15.2 CA3 and CA1 counts are substituted, this produces a poorly fit- 2016), the resulting model was of very poor fit ( , ting model (data not presented). p = 0.004; CFI = 0.74; RMSEA = 0.50). Surprisingly, when the This same optimal network model did not have acceptable lev- same postrhinal–medial entorhinal network was applied to the els of fit for any of the other three behavioural groups (Figure control condition from that study, which involved novel objects 8(b)–(d)). When the two intact groups were directly compared by but no familiarity discrimination, the model retained its fit χ = 32 . 3 stacking their Fos data on the same model (Figure 8(a)), the over- ( , p = 0.52; CFI = 1.0; RMSEA = 0.0). all group difference between model fit was close to being signifi- Finally, evidence that the perirhinal lesions disrupted the PM cant ( χ =11.5 system of the BIC framework (Figure 7) came from the finding , p = 0.075). As the model had poor fit for group 6Diff that the Fos data from the Peri Novel group had only poor fit Sham Baseline, additional analyses were conducted. When each χ = 25.75 ( , p = 0.001; CFI = 0.51; RMSEA = 0.52). Consistent of the component paths was allowed to vary individually, only χ = 44 . 1 with the above results, the updated BIC framework also failed freeing the paths from retrosplenial cortex to MEC ( , 1Diff 2 2 χ = 93 . 9 χ = 66 . 2 to fit the data from group Peri Baseline ( , p = 0.27; p = 0.036) and from CA3 to CA1 ( , p = 0.010) sig- 7 1Diff χ < 12 . CFI = 0.96; RMSEA = 0.18). nificantly improved fit (all other paths: ). This differ- 1Diff ence potentially reflects the strengthening of intrinsic hippocampal 2. What is the optimal model for the Sham Novel context connections with novel context exploration (Figure 8(a)). group? For group Sham Novel Context, the optimal model None of the network models with acceptable fit for group involved many of the regions implicated in the PM network of Sham Novel transferred over to the Sham Baseline group (e.g. χ = 76 . 6 the updated BIC framework ( , p = 0.57; CFI = 1.00; Figures 7 and 8). This failure again suggests that the context- RMSEA = 0.00; Figure 8(a)). Interestingly, the prelimbic and ret- driven models are specific and not simply driven by correlations rosplenial cortices had better predictive value when positioned associated with baseline Fos expression. Kinnavane et al. 11 Finally, evidence that perirhinal lesions disrupted network This specificity was tested in two further ways. First, com- activity in the medial temporal lobe came from the fact that it was parable models were examined using perirhinal cortex and not possible to generate a network model of acceptable fit with LEC, instead of the postrhinal entorhinal cortex and MEC. data from group Peri Novel. Additionally, when the data from the A decision was made not to divide the subiculum and CA1 Fos two novel context groups (Sham Novel and Peri Novel) were counts based on their distal and proximal locations, in order to stacked on the optimal model for Sham Novel (Figure 8), the ensure that all aspects of the models to be compared were held the activity-related Fos data differed significantly between the two same, aside from the introduction of the perirhinal entorhinal cor- χ =17.6 groups ( , p = 0.007). To investigate further, each of 6Diff tex and LEC. The resulting analyses, which tested the AT item the pathways was individually unconstrained revealing signifi- system of the BIC framework (Ritchey et al., 2015), failed to pro- χ = 83 . 0 cant differences in the steps from MEC to CA1 ( , 1Diff vide models of acceptable fit in the novel context groups. Second, χ = 40 . 0 p = 0.004) and CA1 to subiculum ( , p = 0.047; 1Diff data were taken from a previous experiment that examined medial Figure 8). Additionally, for group Sham Novel, the correlation temporal c-fos activity after a test of object recognition memory, between Fos counts in MEC and CA1 was strong and positive again in rats with perirhinal lesions and their surgical sham con- (r = 0.79, p = 0.004), whereas in group Peri Novel, this correlation trols (Kinnavane et al., 2016). The Fos counts from that surgical was negative and non-significant (r = −0.24, p = 0.48). Formal sham group (analysis not presented) failed to fit the PM system, comparison of these correlations using Fisher’s r-to-z transforma- but did fit the anterior–temporal system. Somewhat surprisingly, tion revealed that these correlations were significantly different the Fos data from their control group (Kinnavane et al., 2016), (z = 2.6, p = 0.009). Taken together, these results indicate that the which was exposed to novel objects but did not make recognition perirhinal cortex lesions altered coordinated activity between the discriminations, could fit the PM system. entorhinal cortex and CA1 when animals explored a novel context. Other evidence for the specificity of the context network models came from the baseline home-cage control groups. A striking feature in both the Sham Baseline and Peri Baseline Discussion groups was the high level of correlations between Fos levels in the different areas sampled (around 70% of all sites examined), This study sought to test networks of interlinked c-fos activity which contrasted with that found in the novel context groups associated with context learning in both intact rats and rats with (both around 25%). In the resting condition, the default state perirhinal cortex lesions. In one condition, rats were placed in a appears to involve widespread levels of inter-correlated activity. novel environment (unfamiliar cages in an unfamiliar room), and This pattern changes in the face of a particular learning challenge, in the other, they remained in their home cages. Although this for example, new contextual information. Now, more specific comparison brings additional changes in locomotor and arousal networks become engaged, so decreasing overall site-to-site levels between the two conditions, it has the benefit of creating interactions across multiple brain areas. robust, marked differences in c-fos expression, so more reliably This study also assessed the impact of perirhinal cortex testing any impact of perirhinal cortex loss. A further point is that lesions on medial temporal lobe c-fos activity. Perirhinal lesions the study did not include additional tests to confirm learning about did not disrupt the size of the hippocampal Fos increase when the novel context, aside from the evidence of habitation that came rats are moved to a novel context. Similarly, overall levels of from the locomotor scores. It should, however, be remembered c-fos expression in prelimbic cortex, retrosplenial cortex, and that context learning is regarded as spontaneous (Dix and nucleus reuniens of the thalamus all appeared unaffected by the Aggleton, 1999; Good et al., 2007) and that the context shift used perirhinal cortex lesions. Perirhinal lesions did, however, reduce in this study would be considered highly salient. Consequently, it c-fos expression in the parahippocampal region, an effect most cannot be excluded that changes in c-fos expression may have apparent in the LEC. This result closely matches the findings been driven by those differences in arousal, locomotor activity or from a related study that used object recognition to examine the anxiety, associated with experiencing a novel context. impact of perirhinal lesions on c-fos expression (Kinnavane The neural networks tested were based on recent refinements et al., 2016). The common finding of LEC hypoactivity under- of the BIC framework (Diana et al., 2007; Ranganath, 2010; lines the particularly close anatomical and functional links Ritchey et al., 2015) which emphasises relationships between between the perirhinal cortex and this entorhinal division (see parahippocampal (postrhinal), medial entorhinal, and hippocam- also Burwell and Amaral, 1998a; Wilson et al., 2013a, 2013b; pal areas for context learning. The resulting PM system (Ritchey Witter et al., 2000). Further evidence of perirhinal lesion effects et al., 2015) was tested using SEM. Networks closely based on came from the repeated failure to find medial temporal networks the PM system had good fit for the intact novel context group of acceptable fit for the Peri Novel group. (Figure 7). Furthermore, the optimal network model for this Kent and Brown (2012) suggest that the perirhinal cortex role novel context group incorporated much of the PM system, while in item processing extends to learning about complex features also adding further inputs from prelimbic cortex (Figure 8). This within contextual surroundings based on unitising stimulus rep- optimal novel context model retained its fit when the Fos counts resentations. Support comes from evidence that perirhinal lesions came from just the dorsal hippocampus but not the ventral can impair fear conditioning to complex auditory cues, as well as hippocampus. This result is consistent with the outcome of con- contextual conditioning (Bucci et al., 2000, 2002; Burwell et al., text reactivation studies based on c-fos expression in the dorsal 2004; Corodimas and LeDoux, 1995; Kholodar-Smith et al., hippocampus (Liu et al., 2012, 2014; Ramirez et al., 2013). The 2008a, 2008b; Lindquist et al., 2004; Sacchetti et al., 1999). In same spatial networks, that is, those based on the PM system contrast, perirhinal lesions spare fear conditioning to continuous (Ritchey et al., 2015), did not have acceptable fit for either of the tones (Bucci et al., 2000; Kholodar-Smith et al., 2008a; Lindquist baseline (home-cage) groups. These null results point to the specificity of the BIC framework for contextual learning. et al., 2004). Additionally, increased c-fos expression in the 12 Brain and Neuroscience Advances perirhinal cortex is associated with context shifts (VanElzakker from single unit recordings that the rat LEC also plays a role in et al., 2008; Vann et al., 2000), as well as with contextual fear spatial processing, often in relation to item location (Deshmukh conditioning, but not cued fear conditioning (Albrechet-Souza et al., 2012; Hunsaker and Kesner, 2013; Knierim et al., 2014; et al., 2011). Thus, the perirhinal cortex may be involved in dis- Neunuebel et al., 2013). The conclusion is that while there is a criminating and, hence, helping to bring together novel compo- major division of information processing pathways in the medial nents within a given context, even though this cortical area may temporal lobe and beyond, there remain important interactions be insensitive to their relative spatial disposition (Aggleton et al., between these same pathways at multiple levels, including those 2010; Deshmukh et al., 2012; Jenkins et al., 2004; Wan et al., between parahippocampal areas. 1999). It is presumably this latter aspect, along with the relative preservation of inter-hippocampal activity, as seen in this study Declaration of conflicting interests (see also Kinnavane et al., 2016), which helps to explain why The author(s) declared no potential conflicts of interest with respect to perirhinal cortex lesions often spare those tests of allocentric spa- the research, authorship, and/or publication of this article. tial memory that are highly sensitive to hippocampal damage (Glenn and Mumby, 1998; Machin et al., 2002; Ramos, 2013; Funding Winters et al., 2004). Many of these same tests make additional This work was supported by the Wellcome Trust (WT087955 and demands on navigation, an ability closely linked with medial WT09520). entorhinal–hippocampal function, rather than perirhinal cortex (Buzsáki and Moser, 2013). References It would be wrong, however, to infer that perirhinal cortex lesions are without effect on hippocampal spatial processing. In Aggleton JP (2012) Multiple anatomical systems embedded within the this study, perirhinal lesions altered entorhinal cortex activity, so primate medial temporal lobe: Implications for hippocampal func- tion. Neuroscience and Biobehavioral Reviews 36(7): 1579–1596. disrupting parahippocampal–hippocampal interactions, for Aggleton JP, Albasser MM, Aggleton DJ, et al. (2010) Lesions of the example, those with CA1. With this result in mind, it may be rat perirhinal cortex spare the acquisition of a complex configural relevant that electrophysiological studies have shown that while visual discrimination yet impair object recognition. Behavioral perirhinal lesions do not appear to affect the initial formation of Neuroscience 124(1): 55–68. CA1 place fields, their stability is reduced (Muir and Bilkey, Aggleton JP, Kyd R and Bilkey D (2004) When is the perirhinal cortex 2001). Other reported hippocampal changes following perirhinal necessary for the performance of spatial memory tasks? Neurosci- lesions are a reduction in the proportion of theta cells, a phase ence and Biobehavioral Reviews 28(6): 611–624. shift in CA1 place cells, and an altered modulation by movement Albasser MM, Davies M, Futter JE, et al. (2009) Magnitude of the object velocity on place cells (Muir and Bilkey, 2003). At the same time, recognition deficit associated with perirhinal cortex damage in rats: subtle deficits after perirhinal cortex lesions have been reported Effects of varying the lesion extent and the duration of the sample period. Behavioral Neuroscience 123(1): 115–124. by some when using spatial tasks sensitive to hippocampal dam- Albasser MM, Olarte-Sánchez CM, Amin E, et al. (2015) Perirhinal cor- age (Aggleton et al., 2004; Liu and Bilkey, 1998a, 1998b, 2001; tex lesions in rats: Novelty detection and sensitivity to interference. Wiig and Bilkey, 1994a, 1994b). The current data suggest that Behavioral Neuroscience 129(3): 227–243. these milder deficits stem from parahippocampal dysfunctions, Albasser MM, Poirier GL and Aggleton JP (2010) Qualitatively differ- such as those that were evident in LEC. ent modes of perirhinal-hippocampal engagement when rats explore This study set out to test medial temporal models that novel vs. familiar objects as revealed by c-fos imaging. European describe the processing of context information. Expression of Journal of Neuroscience 31(1): 134–147. c-fos revealed inter-related activity networks that closely match Albrechet-Souza L, Borelli KG, Almada RC, et al. (2011) Midazolam those predicted by refinements of the BIC framework. Arguably, reduces the selective activation of the rhinal cortex by contextual fear one unexpected finding was the direction of influence in the opti- stimuli. Behavioural Brain Research 216(2): 631–638. Arbuckle JL (2011) IBM SPSS AMOS 20 User’s Guide. Chicago, IL: mal model for the intact novel context group (Figure 8). While all AMOS Development Corporation. connections were significantly connected, the best fit was Barker GRI, Bird F, Alexander V, et al. (2007) Recognition memory for obtained when the model progressed from prelimbic cortex to objects, places, and temporal order: A disconnection analysis of the retrosplenial cortex and then to the medial temporal lobe. In prac- role of the medial prefrontal cortex and perirhinal cortex. Journal of tice, the PM contextual system (Ritchey et al., 2015; Figure 1) Neuroscience 27(11): 2948–2957. can accommodate these findings as it includes reciprocal connec- Bast T (2007) Toward an integrated perspective on hippocampal func- tions between the retrosplenial cortex and medial temporal sites. tion: From a rapid encoding of experience to adaptive behaviour. A final point concerns the impression that there are distinct Reviews in the Neurosciences 18(3–4): 253–281. parahippocampal pathways for ‘what’ and ‘where’ information. It Bisler S, Schleicher A, Gass P, et al. (2002) Expression of c-Fos, ICER, has been argued that a more nuanced division might be more Krox-24 and JunB in the whisker-to-barrel pathway of rats: Time course of induction upon whisker stimulation by tactile exploration appropriate, such as that between local and global reference of an enriched environment. Journal of Chemical Neuroanatomy frames (Knierim et al., 2006, 2014). As part of this process, a 23(3): 187–198. more complete model will need to include crosstalk between the Bollen KA and Long JS (1992) Tests for structural equation models: two putative pathways, reflecting their interconnectivity (Burwell Introduction. Sociological Methods & Research 21(2): 123–131. and Amaral, 1998b). Indeed, evidence of lateral entorhinal– Bucci DJ and Robinson R (2014) Toward a conceptualization of retro- medial entorhinal interactions was found in this study. For exam- hippocampal contributions to learning and memory. Neurobiology of ple, when the Sham Novel context model (Figure 7) is modified Learning and Memory 116: 197–207. so that LEC replaces the postrhinal cortex, the model still retains Bucci DJ, Phillips RG and Burwell RD (2000) Contributions of post- good fit (not presented). It was also the case that the perirhinal rhinal and perirhinal cortex to contextual information processing. lesions disrupted the PM system. Similarly, there is evidence Behavioral Neuroscience 114(5): 882–894. Kinnavane et al. 13 Bucci DJ, Saddoris MP and Burwell RD (2002) Contextual fear discrimi- formation but not the parahippocampal cortices; a c-fos expression nation is impaired by damage to the postrhinal or perirhinal cortex. study. Neuroscience 124(1): 43–52. Behavioral Neuroscience 116(3): 479–488. Kent BA and Brown TH (2012) Dual functions of perirhinal cortex in Burwell RD (2000) The parahippocampal region: Corticocortical fear conditioning. Hippocampus 22(10): 2068–2079. connectivity. Annals of the New York Academy of Sciences 911: Kholodar-Smith DB, Allen TA and Brown TH (2008a) Fear conditioning 25–42. to discontinuous auditory cues requires perirhinal cortical function. Burwell RD (2001) Borders and cytoarchitecture of the perirhinal and Behavioral Neuroscience 122(5): 1178–1185. postrhinal cortices in the rat. Journal of Comparative Neurology Kholodar-Smith DB, Boguszewski P and Brown TH (2008b) Auditory 437(1): 17–41. trace fear conditioning requires perirhinal cortex. Neurobiology of Burwell RD and Amaral DG (1998a) Cortical afferents of the perirhinal, Learning and Memory 90(3): 537–543. postrhinal and entorhinal cortices of the rat. Journal of Comparative Kinnavane L, Amin E, Horne M, et al. (2014) Mapping parahippocam- Neurology 398(2): 179–205. pal systems for recognition and recency memory in the absence of Burwell RD and Amaral DG (1998b) Perirhinal and postrhinal cortices of the rat hippocampus. European Journal of Neuroscience 40(12): the rat: Interconnectivity and connections with the entorhinal cortex. 3720–3734. Journal of Comparative Neurology 391(3): 293–321. Kinnavane L, Amin E, Olarte-Sánchez CM, et al. (2016) Mapping Burwell RD, Bucci DJ, Sanborn MR, et al. (2004) Perirhinal and responses to novel objects: The impact of perirhinal cortex discon- postrhinal contributions to remote memory for context. Journal of nection on hippocampal activity patterns. Hippocampus 26(11): Neuroscience 24(49): 11023–11028. 1393–1413. Buzsáki G and Moser EI (2013) Memory, navigation and theta rhythm Knierim JJ, Lee I and Hargreaves EL (2006) Hippocampal place cells: in the hippocampal-entorhinal system. Nature Neuroscience 16(2): Parallel input streams, subregional processing and implications for 130–138. episodic memory. Hippocampus 16(9): 755–764. Chaudhuri A (1997) Neural activity mapping with inducible transcription Knierim JJ, Neunuebel JP and Deshmukh SS (2014) Functional correlates factors. NeuroReport 8(13): 3–7. of the lateral and medial entorhinal cortex: Objects, path integration Conde F, Marie-Lepoivre E, Audinat E, et al. (1995) Afferent connec- and local-global reference frames. Philosophical Transactions of the tions of the medial frontal cortex of the rat II: Cortical and subcorti- Royal Society B: Biological Sciences 369(1635): 20130369. cal afferents. Journal of Comparative Neurology 352(4): 567–593. Lindquist DH, Jarrard LE and Brown TH (2004) Perirhinal cortex sup- Corodimas KP and LeDoux JE (1995) Disruptive effects of posttraining ports delay fear conditioning to rat ultrasonic social signals. Journal perirhinal cortex lesions on conditioned fear: Contributions of con- of Neuroscience 24(14): 3610–3617. textual cues. Behavioral Neuroscience 109(4): 613–619. Liu P and Bilkey DK (1998a) Excitotoxic lesions centered on perirhinal Deshmukh SS, Johnson JL and Knierim JJ (2012) Perirhinal cortex rep- cortex produce delay-dependent deficits in a test of spatial memory. resents nonspatial, but not spatial, information in rats foraging in Behavioral Neuroscience 112(3): 512–524. the presence of objects: Comparison with lateral entorhinal cortex. Liu P and Bilkey DK (1998b) Lesions of perirhinal cortex produce Hippocampus 22(10): 2045–2058. spatial memory deficits in the radial maze. Hippocampus 8(2): Diana RA, Yonelinas AP and Ranganath C (2007) Imaging recollec- 114–121. tion and familiarity in the medial temporal lobe: A three-component Liu P and Bilkey DK (2001) The effect of excitotoxic lesions centred on model. Trends in Cognitive Sciences 11(9): 379–386. the hippocampus or perirhinal cortex in object recognition and spa- Dix SL and Aggleton JP (1999) Extending the spontaneous preference tial memory tasks. Behavioral Neuroscience 115(1): 94–111. test of recognition: Evidence of object-location and object-context Liu X, Ramirez S and Tonegawa S (2014) Inception of a false memory by recognition. Behavioural Brain Research 99(2): 191–200. optogenetic manipulation of a hippocampal memory engram. Philo- Eacott MJ and Gaffan EA (2005) The roles of perirhinal cortex, post- sophical Transactions of the Royal Society B: Biological Sciences rhinal cortex, and the fornix in memory for objects, contexts, and 369(1633): 20130142. events in the rat. Quarterly Journal of Experimental Psychology B: Liu X, Ramirez S, Pang P, et al. (2012) Optogenetic stimulation of Comparative and Physiological Psychology 58(3–4): 202–217. a hippocampal engram activates fear memory recall. Nature Fan X and Wang L (1998) Effects of potential confounding factors 484(7394): 381–385. on fit indices and parameter estimates for true and misspecified Machin P, Vann SD, Muir JL, et al. (2002) Neurotoxic lesions of the SEM models. Educational and Psychological Measurement 58(5): rat perirhinal cortex fail to disrupt the acquisition or performance of 701–735. tests of allocentric spatial memory. Behavioral Neuroscience 116(2): Furtak SC, Wei S-M, Agster KL, et al. (2007) Functional neuroanatomy 232–240. of the parahippocampal region in the rat: The perirhinal and postrhi- McIntosh AR and Gonzalez-Lima F (1991) Structural modeling of func- nal cortices. Hippocampus 17(9): 709–722. tional neural pathways mapped with 2-deoxyglucose: Effects of Glenn MJ and Mumby GD (1998) Place memory is intact in rats with peri- acoustic startle habituation on the auditory system. Brain Research rhinal cortex lesions. Behavioral Neuroscience 112(6): 1353–1365. 547(2): 295–302. Good MA, Barnes P, Staal V, et al. (2007) Context-but not familiar- Muir GM and Bilkey DK (2001) Instability in the place field location ity-dependent forms of object recognition are impaired following of hippocampal place cells after lesions centered on the perirhinal excitotoxic hippocampal lesions in rats. Behavioral Neuroscience cortex. Journal of Neuroscience 21(11): 4016–4025. 121(1): 218–223. Muir GM and Bilkey DK (2003) Theta- and movement velocity-related Howell DC (2011) Statistical Methods for Psychology (8th edn). firing of hippocampal neurons is disrupted by lesions centered on the Wadsworth, OH: Cengage Learning. perirhinal cortex. Hippocampus 13(1): 93–108. Hu L and Bentler PM (1998) Fit indices in covariance structure mod- Mumby DG and Pinel JP (1994) Rhinal cortex lesions and object recog- eling: Sensitivity to underparameterized model misspecification. nition in rats. Behavioral Neuroscience 108(1): 11–18. Psychology Methods 3(4): 424–453. Naber PA, Caballero-Bleda M, Jorritsma-Byham B, et al. (1997) Parallel Hunsaker MR and Kesner RP (2013) The operation of pattern separation input to the hippocampal memory system through peri- and postrhi- and pattern completion processes associated with different attributes nal cortices. NeuroReport 8(11): 2617–2621. or domains of memory. Neuroscience and Biobehavioral Reviews Neunuebel JP, Yoganarasimha D, Rao G, et al. (2013) Conflicts between 37(1): 36–58. local and global spatial frameworks dissociate neural representations Jenkins TA, Amin E, Pearce JM, et al. (2004) Novel spatial arrange- of the lateral and medial entorhinal cortex. Journal of Neuroscience ments of familiar stimuli promote activity in the rat hippocampal 33(22): 9246–9258. 14 Brain and Neuroscience Advances Norman G and Eacott MJ (2005) Dissociable effects of lesions Van Groen T and Wyss JM (2003) Connections of the retrosplenial gran- to the perirhinal cortex and the postrhinal cortex on memory ular b cortex in the rat. Journal of Comparative Neurology 463(3): for context and objects in rats. Behavioral Neuroscience 119(2): 249–263. 557–566. Van Strien NM, Cappaert NLM and Witter MP (2009) The anatomy Olarte-Sánchez CM, Kinnavane L, Amin A, et al. (2014) Contrasting of memory: An interactive overview of the parahippocampal– networks for recognition memory and recency memory revealed by hippocampal network. Nature Reviews Neuroscience 10(4): immediate-early gene imaging in the rat. Behavioral Neuroscience 272–282. 128(4): 504–522. VanElzakker M, Fevurly RD, Breindel T, et al. (2008) Environmental Paxinos G and Watson C (2005) The Rat Brain in Stereotaxic Coordi- novelty is associated with a selective increase in Fos expression in nates (5th edn). San Diego, CA: Elsevier Academic Press. the output elements of the hippocampal formation and the perirhinal Prasad JA and Chudasama Y (2013) Viral tracing identifies parallel cortex. Learning & Memory 15(12): 899–908. disynaptic pathways to the hippocampus. Journal of Neuroscience Vann SD, Brown MW, Erichsen JT, et al. (2000) Fos imaging reveals 33(19): 8494–8503. differential patterns of hippocampal and parahippocampal subfield Protzner AB and McIntosh AR (2006) Testing effective connectivity activity in response to different spatial memory tasks. Journal of changes with structural equation modeling: What does a bad model Neuroscience 20(7): 2711–2718. tell us? Human Brain Mapping 27(12): 935–947. Vertes RP, Hoover WB, Szigeti-Buck K, et al. (2007) Nucleus reuniens Ramirez S, Liu X, Lin P-A, et al. (2013) Creating a false memory in the of the midline thalamus: Link between the medial prefrontal cortex hippocampus. Science 341(6144): 387–391. and the hippocampus. Brain Research Bulletin 71(6): 601–609. Ramos JMJ (2013) Differential contribution of hippocampus, peri- Wan H, Aggleton JP and Brown MW (1999) Different contributions rhinal cortex and postrhinal cortex to allocentric spatial memory of the hippocampus and perirhinal cortex to recognition memory. in the radial maze. Behavioural Brain Research 247: 59–69. Journal of Neuroscience 19(3): 1142–1148. Ranganath C (2010) Binding items contexts: The cognitive neuroscience Wiig KA and Bilkey DK (1994a) Perirhinal cortex lesions in rats disrupt of episodic memory. Current Directions in Psychological Science performance in a spatial DNMS task. NeuroReport 5(11): 1405–1408. 19(3): 131–137. Wiig KA and Bilkey DK (1994b) The effects of perirhinal cortical lesions Ranganath C and Ritchey M (2012) Two cortical systems for on spatial reference memory in the rat. Behavioural Brain Research memory guided behaviour. Nature Reviews Neuroscience 13(10): 63(1): 101–109. 713–726. Wilson DIG, Langston RF, Schlesiger MI, et al. (2013a) Lateral ento- Ritchey M, Libby LA and Ranganath C (2015) Cortico-hippocampal rhinal cortex is critical for novel object-context recognition. systems involved in memory and cognition: The PMAT framework. Hippocampus 23(5): 352–366. Progress in Brain Research 219: 45–64. Wilson DIG, Watanabe S, Milner H, et al. (2013b) Lateral entorhinal Sacchetti B, Lorenzini CA, Baldi E, et al. (1999) Auditory thalamus, dor- cortex is necessary for associative but not nonassociative recognition sal hippocampus, basolateral amygdala, and perirhinal cortex role in memory. Hippocampus 23(12): 1280–1290. the consolidation of conditioned freezing to context and to acous- Winters BD, Forwood SE, Cowell RA, et al. (2004) Double dissocia- tic conditioned stimulus in the rat. Journal of Neuroscience 19(21): tions between the effects of peri-postrhinal cortex and hippocampal 9570–9578. lesions on tests of object recognition and spatial memory: Heteroge- Schumacker RE and Lomax RG (2010) A Beginners Guide to Structural neity of function within the temporal lobe. Journal of Neuroscience Equation Modelling (3rd edn). New York: Routledge. 24(26): 5901–5908. Strange BA, Witter MP, Lein E, et al. (2014) Functional organization Witter MP (2002) The parahippocampal region: Past, present, and future. of the hippocampal longitudinal axis. Nature Reviews Neuroscience In: Witter M and Wouterlood F (eds) The Parahippocampal Region: 15(10): 655–669. Organization and Role in Cognitive Function. Oxford: Oxford Swanson LW (1992) Brain Maps: Structure of the Rat Brain. University Press, pp. 1–20. Amsterdam: Elsevier. Witter MP, Naber PA, van Haeften T, et al. (2000) Cortico-hippocampal Tabachnick BG and Fidell LS (2001) Using Multivariate Statistics communication by way of parallel parahippocampal-subicular path- (4th edn). Needham, MA: Allyn & Bacon. ways. Hippocampus 10(4): 389–410. Van Groen T and Wyss JM (1990) Connections of the retrosplenial gran- Wothke W (1993) Nonpositive definite matrices in structural equation ular a cortex in the rat. Journal of Comparative Neurology 300(4): modeling. In: Bollen KA and Long JS (eds) Testing Structural Equa- 593–606. tion Models. London: SAGE, pp. 256–293. Van Groen T and Wyss JM (1992) Connections of the retrosplenial dys- Zangenehpour S and Chaudhuri A (2002) Differential induction and granular cortex in the rat. Journal of Comparative Neurology 315(2): decay curves of c-fos and zif268 revealed through dual activity 200–216. maps. Molecular Brain Research 109(1–2): 221–225. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Brain and Neuroscience Advances SAGE

Medial temporal pathways for contextual learning: Network c-fos mapping in rats with or without perirhinal cortex lesions:

Loading next page...
 
/lp/sage/medial-temporal-pathways-for-contextual-learning-network-c-fos-mapping-mz4ygm0UC7

References (180)

Publisher
SAGE
Copyright
Copyright © 2022 by SAGE Publications Ltd and British Neuroscience Association, unless otherwise noted. Manuscript content on this site is licensed under Creative Commons Licenses
ISSN
2398-2128
eISSN
2398-2128
DOI
10.1177/2398212817694167
Publisher site
See Article on Publisher Site

Abstract

Background: In the rat brain, context information is thought to engage network interactions between the postrhinal cortex, medial entorhinal cortex, and the hippocampus. In contrast, object information is thought to be more reliant on perirhinal cortex and lateral entorhinal cortex interactions with the hippocampus. Method: The ‘context network’ was explored by mapping expression of the immediate-early gene, c-fos, after exposure to a new spatial environment. Results: Structural equation modelling of Fos counts produced networks of good fit that closely matched prior predictions based on anatomically grounded functional models. These same models did not, however, fit the Fos data from home-cage controls nor did they fit the corresponding data from a previous study exploring object recognition. These additional analyses highlight the specificity of the context network. The home-cage controls, meanwhile, showed raised levels of inter-area Fos correlations between the many sites examined, that is, their changes in Fos levels lacked anatomical specificity. A total of two additional groups of rats received perirhinal cortex lesions. While the loss of perirhinal cortex reduced lateral entorhinal c-fos expression, it did not affect mean levels of hippocampal c-fos expression. Similarly, overall c-fos expression in the prelimbic cortex, retrosplenial cortex, and nucleus reuniens of the thalamus appeared unaffected by the perirhinal cortex lesions. Conclusion: The perirhinal cortex lesions disrupted network interactions involving the medial entorhinal cortex and the hippocampus, highlighting ways in which perirhinal cortex might affect specific aspects of context learning. Keywords Entorhinal cortex, hippocampus, nucleus reuniens, prefrontal cortex, retrosplenial cortex, spatial memory, subiculum Received: 6 December 2016; accepted: 25 January 2017 Introduction Models of medial temporal lobe processing increasingly assume activate one of these processing pathways. This IEG, which pro- two distinct functional pathways (Figure 1), one for object-based vides an indirect marker of neural activity (Bisler et al., 2002; information and the other for spatial and contextual information Chaudhuri, 1997; Zangenehpour and Chaudhuri, 2002), is known (Bucci and Robinson, 2014; Burwell, 2000; Eacott and Gaffan, to show increased hippocampal activity following contextual 2005; Diana et al., 2007; Knierim et al., 2014; Ranganath and change (e.g. Albrechet-Souza et al., 2011; Jenkins et al., 2004; Ritchey, 2012; Ritchey et al., 2015). In the case of the rodent Vann et al., 2000). The importance of this c-fos expression is brain, perirhinal cortex is presumed to process object-based strikingly highlighted by studies showing how reactivating those information, in concert with the lateral entorhinal cortex (LEC; dorsal hippocampal neurons that had previously expressed c-fos Barker et al., 2007; Mumby and Pinel, 1994; Naber et al., 1997). in a distinctive context can reinstate representations of that same In contrast, postrhinal cortex, along with the medial entorhinal context (Liu et al., 2012, 2014; Ramirez et al., 2013). Structural cortex (MEC), is presumed to provide spatial and contextual equation modelling (SEM; McIntosh and Gonzalez-Lima, 1991; information for the hippocampus (Bucci et al., 2000; Bucci and Schumacker and Lomax, 2010) was then used to test anatomically Robinson, 2014; Burwell, 2000; Furtak et al., 2007; Norman and Eacott, 2005; Ranganath and Ritchey, 2012). These distinct, functional pathways are highlighted in a number of models, School of Psychology, Cardiff University, Cardiff, UK including the binding of item and context (BIC; Diana et al., The Institute of Medical Sciences, University of Aberdeen, Aberdeen, UK 2007) framework as well as in subsequent anatomical refine- ments of this basic model (Aggleton, 2012; Ranganath and Corresponding author: Ritchey, 2012; Ritchey et al., 2015). Lisa Kinnavane, School of Psychology, Cardiff University, 70 Park This study quantified the expression of the immediate-early Place, Cardiff CF10 3AT, UK. gene (IEG) c-fos after placing rats in a novel context in order to Email: kinnavanel@cardiff.ac.uk Creative Commons CC BY: This article is distributed under the terms of the Creative Commons Attribution 3.0 License (http://www.creativecommons.org/licenses/by/3.0/) which permits any use, reproduction and distribution of the work without further permission provided the original work is attributed as specified on the SAGE and Open Access pages (https://us.sagepub.com/en-us/nam/open-access-at-sage). 2 Brain and Neuroscience Advances Figure 1. Posterior–medial (PM) context system and anterior–temporal (AT) item system of the binding of item and context (BIC) framework. Parallel cortico-hippocampal pathways link the PM and AT systems with the entorhinal cortex, CA1, and subiculum. Source: Adapted from Ritchey et al. (2015). plausible models based on refinements of the BIC framework at the beginning of the c-fos imaging study. Of these rats, 18 had (Ritchey et al., 2015). received perirhinal cortex lesions, while 11 served as their surgi- As already noted, current medial temporal models typically cal controls. Rats from cohort B (n = 27) were approximately assume the presence of two parallel pathways that emanate, 7 months old at the beginning of the present experiment. Of these, respectively, from the perirhinal and parahippocampal (postrhi- 15 received perirhinal cortex lesions, while 12 received sham nal) cortices to reach the hippocampal formation (Diana et al., surgeries. Prior to the current experiment, both cohorts received 2007; Witter, 2002). There are, however, reciprocal connections object recognition memory tasks in the bow-tie maze (full details between the postrhinal and perirhinal cortices and between are described in Albasser et al., 2015). Additionally, both cohorts the MEC and LEC (Burwell and Amaral, 1998a, 1998b; Furtak received a single, spontaneous object recognition test in an open- et al., 2007). These interconnections question the degree of inde- field apparatus. Rats were not behaviourally tested for at least pendence between these two functional pathways (Kent and 2 weeks before the current experiment, with water and food Brown, 2012; Knierim et al., 2014; Liu and Bilkey, 1998b, available ad libitum throughout this intervening period. 2001). For this reason, this study also examined how perirhinal cortex lesions might affect parahippocampal–hippocampal c-fos Surgery activity following exposure to a novel context. The question was whether the context pathway still shows normal activity patterns The rats were approximately 3 months old at the time of surgery, in the absence of perirhinal cortex. Finally, other intact rats and when they weighed between 285 and 300 g. In total, 33 rats rats with perirhinal cortex lesions were examined for c-fos received bilateral perirhinal cortex lesions (‘Peri’), while 23 rats expression after receiving no context shift, so providing a com- served as surgical controls (‘Sham’). Anaesthesia was induced parison baseline condition. using a mixture of oxygen and isoflurane gas (5% for induction and 2% thereafter), before placing each rat in a stereotaxic frame (David Kopf Instruments, Tujunga, CA, USA), with the incisor Materials and methods bar set at +5.0 mm above the horizontal plane. After making a midline sagittal incision in the scalp, the skin was retracted to Animals expose the skull. Following a craniotomy, the perirhinal lesions Subjects were 56 male, Lister Hooded rats (Harlan, Bicester, were made by injecting a solution of N-methyl-d -aspartate UK), housed in pairs under diurnal conditions (12-h light/12-h (NMDA; Sigma, Poole, UK) diluted to 0.09 M in phosphate- dark). The home cages measured 42 cm × 25 cm × 21 cm with a buffered saline (PBS; 0.1 M, pH 7.4) using a 1-µm Hamilton water bottle and food hopper at the front. Each cage, which had syringe (Bonaduz, Switzerland; gauge 26, outside diameter opaque plastic floors and walls (13 cm high), was lined with 0.47 mm) held with a micro-injector (model 5000; Kopf sawdust and contained a cardboard tube and chew stick. The Instruments, Tujunga, CA). Bilateral injections of NMDA (each rats, which were fed 2014 Teklad global 14% protein rodent of 0.225 µL) were made at a rate of 0.10 µL/min, with a subse- maintenance diet (Harlan), were on an ad libitum schedule. All quent diffusion time of 4 min before the needle was removed. experiments were performed in accordance with the UK Animals Three injections were made in each hemisphere. Injection coor- (Scientific Procedures) Act, 1986, and associated guidelines and dinates relative to bregma (in mm) were (1) anterior–posterior approved by local ethical committees at Cardiff University. (AP): −1.8, medial–lateral (ML): ±5.9, and dorsal–ventral (DV): The rats came from two cohorts of animals, which received −9.3; (2) AP: −3.4, ML: ±6.1, and DV: −9.6; (3) AP: −5.0, ML: the same experimental protocols throughout the present experi- ±6.2, and DV: −9.0. Rats in the surgical control group received ment. Rats in cohort A (n = 29) were approximately 11 months old identical treatments, except that the dura was perforated three Kinnavane et al. 3 times per hemisphere with a 25-gauge Microlance 3 needle Lesion analysis (Becton Dickinson, Drogheda, Ireland) and no fluid was infused Only one hemisphere in each Peri brain was analysed for Fos into the brain. expression, while the other was eliminated. This procedure ensured that Fos counts were only taken from those hemispheres with either no evidence of surgically induced hippocampal cell Apparatus – activity boxes loss or with loss restricted to just one coronal section (typically in For the novel context condition, rats were placed individually in the hippocampal subfield, CA1). Brains that suffered damage to an activity cage in a novel room (272 cm × 135 cm × 240cm). both hippocampi were excluded from the study. In those hemi- A 3 × 6 bank of activity cages was located along one wall of the spheres analysed for study, the boundaries of the lesions were room. Each activity cage (Paul Fray, Cambridge, UK) measured drawn onto five coronal plates (bregma: −2.80 to −6.72 mm) from 56 cm × 39 cm × 19 cm and contained two photobeams placed Paxinos and Watson (2005). These images were then scanned, and 20 cm apart, positioned 18 cm from the short walls. The floor of the area of damage was calculated using cellSens Dimension each cage was made of wire; otherwise, the cage was empty. The Desktop, version 1.12 (Olympus, Southend-on-Sea, UK). top of each cage was also made of wire, and the room was illuminated. Immunohistochemistry Brain sections were initially stored at −20°C in cryoprotectant. Behavioural testing Free-floating sections were then immunohistochemically stained Both the Peri and Sham animals were divided between the two with sections from one rat from each of the four behavioural behavioural conditions, creating four groups. Rats with perirhinal groups placed in the same reaction vessel, that is, sections from lesions were assigned to either the novel context condition (Peri all four groups were processed concurrently. This arrangement Novel, n = 18; nine from cohort A, nine from cohort B) or the sought to decrease staining variation between groups. The sec- home-cage control condition (Peri Baseline, n = 15; 9 from cohort tions were first washed six times in PBS to remove the cryopro- A, 6 from cohort B). Similarly, the sham surgical controls were tectant, then washed in 0.2% Triton-X 100 in 0.1M PBS (PBST), divided between the novel context condition (Sham Novel, once in 1% H O in PBST (to block endogenous peroxidases), 2 2 n = 11; 4 from cohort A, 7 from cohort B) and the home-cage and then four further times in PBST. The sections were then control condition (Sham Baseline, n = 12; 7 from cohort A, 5 incubated in a blocking solution of 3% normal goat serum from cohort B). Behavioural testing (activity boxes) took place (NGS) in PBST for 1 h followed by the primary antibody solu- either 8 (cohort A) or 4 (cohort B) months after surgery. tion; rabbit-anti-c-fos (1:15,000) and 1% NGS diluted in PBST For the novel context condition, each rat was placed indi- (Cat# PC38; Calbiochem; now part of Merek Millipore, vidually in a dark holding room for 30 min (to which they had Nottingham, UK), for 48 h at 4°C. The sections were then received two 30 min familiarisation sessions on the preceding washed four times in PBST, before being incubated in the sec- days). They were then taken into the novel test room and placed ondary antibody solution; biotinylated goat-anti-rabbit (1:200; individually inside an activity test cage for 20 min. (Due to an Vector Laboratories, Peterborough, UK) diluted in 1.5% NGS in equipment malfunction, two activity scores (one Peri, one Sham) PBST for 2 h at room temperature. The sections were washed were not recorded.) These rats were then returned to the dark four times in PBST. They were then incubated in avidin- holding room. For the baseline condition, individual rats biotinylated horseradish peroxidase complex in PBST (Elite kit; remained throughout in their home cages without exposure to Vector Laboratories) for 1 h at room temperature. The sections the dark room prior to perfusion. were washed four times in PBST and then twice in 0.05 M Tris buffer (pH 7.4). All washes were 10 min unless otherwise stated. Finally, diaminobenzidine (DAB substrate kit; Vector Perfusion and tissue sectioning Laboratories) was used as the chromogen to visualise the loca- tion of immunostaining. The reaction was stopped in cold PBS. For the novel context condition, rats were perfused 90 min after The sections were mounted onto double gelatine-subbed glass being returned to the dark holding room. This interval is within slides and allowed to air dry for at least 48 h, dehydrated in the time period when the expression of Fos, the protein product increasing concentration of alcohol washes, cleared in xylene, of c-fos, peaks, that is, between 60 and 120 min after the induc- and coverslipped using DPX as the mounting media. ing event (Bisler et al., 2002; Zangenehpour and Chaudhuri, 2002). Animals in the baseline control condition were taken directly from their home cage immediately prior to perfusion. Image capture and analysis of c-fos activation All rats then received a lethal overdose of sodium pentobarbital (60 mg/kg, Euthatal, Marial Animal Health, Harlow, Essex, UK) Images from each region of interest (ROI) were captured from and were transcardially perfused with 0.1 M PBS followed by six consecutive sections (each 120 µm apart) from one hemi- 4% paraformaldehyde in 0.1 M PBS (PFA). Brains were removed sphere per animal. The equivalent hemisphere (left or right) was from the skull, postfixed in PFA for 4 h, and then incubated in also analysed in the corresponding ‘Sham’ control animal. Image 25% sucrose at room temperature overnight on a stirrer plate. capture used a 5× objective lens (numerical aperture of 0.12) on The brains were cut in the coronal plane into 40-µm sections a Leica DMRB microscope with an Olympus DP70 camera. The using a freezing microtome. A series of one in four sections field of view was 0.84 × 0.63 mm, so that cortical regions only were collected in PBS and then stained with cresyl violet required one image per section to include all lamina. For the hip- (a Nissl stain), while another one in four series was retained pocampus, multiple images were taken and combined (Microsoft for immunohistochemistry. Image Composite Editor (ICE); Microsoft, Redmond, WA, 4 Brain and Neuroscience Advances USA). Using ANALYSIS^D software (Soft-Imaging Systems, Olympus Corporation). Fos-positive cells were quantified by counting the number of immunopositive nuclei (mean Feret diameter of 4−20 µm) stained above a greyscale threshold set 60−70 units below the peak grey value measured by a pixel inten- sity histogram. ROIs The borders of the perirhinal and postrhinal cortices follow the description of Burwell and Amaral (1998a; see also Burwell, 2001), while those of the other brain areas correspond to Swanson (1992). The AP coordinates (mm from bregma) given in the descriptions below and in Figure 2 are from Paxinos and Watson (2005). The regional groupings are those subsequently used in the statistical analyses of Fos counts. Hippocampal subfields. Hippocampal subfields (dentate gyrus, CA1, and CA3) were subdivided into their septal (dorsal), inter- mediate (dorsal), and temporal (ventral) divisions(Bast, 2007; Strange et al., 2014). The septal hippocampus counts (dentate gyrus, CA3, and CA1) were obtained from sections from AP −2.52 to −3.24, while those for the intermediate dorsal hippo- campus (dentate gyrus, CA1, and CA3) came from sections Figure 2. Regions of interest for c-fos analyses. Sites examined: CA between AP −4.80 and −5.52. The border between the dorsal fields – intermediate (inter), septal (sept), and temporal (temp); intermediate and temporal hippocampus corresponds to −5.0 mm dentate gyrus (DG); dorsal subiculum (dorsal Sub); lateral entorhinal ventral from bregma (see Figure 2; Paxinos and Watson, 2005). cortex (LEC); medial entorhinal cortex (MEC); prelimbic cortex (PL); Within the temporal hippocampus, counts were made in the CA1 perirhinal cortex (PRH); postrhinal cortex (POR); nucleus reuniens of and CA3 fields at the same AP as the intermediate dorsal hippo- thalamus (Reuniens); retrosplenial cortex (RSP); and ventral subiculum campus (note that the dentate gyrus is not present at this level). (ventral Sub). The numbers below refer to the approximate distance in Additional Fos-positive cell counts were taken in both the dorsal millimetre from bregma. and ventral subiculum (from around AP −5.16). Source: Adapted from the atlas of Paxinos and Watson (2005). Parahippocampal cortices. Separate Fos-positive cell counts were taken from the LEC and MEC, as well as the postrhinal repeatedly broken, and ‘beam crossovers’, that is, the front and cortex. The LEC counts were taken from more caudal parts back beams broken sequentially. These data were compared with of the area to ensure that there was no encroachment from the a one between-subject factor (surgical condition) and one within- perirhinal lesion in the Peri groups. In the Sham cases only, Fos- subject factor (‘same beam’ or ‘beam crossovers’) analysis of positive cell counts were made in the caudal perirhinal cortex variance (ANOVA). The total number of beam breaks across the (areas 35 and 36; see Burwell, 2001). This caudal portion (from 20-min exposure to the novel context was then divided into four AP −4.80 to −5.52) was selected as previous studies indicate that bins of 5 min and compared with a one between-subject factor this region is particularly involved in processing novel visual (surgical condition) and one within-subject factor (bin). stimuli (Albasser et al., 2009, 2010; Kinnavane et al., 2014; Olarte-Sánchez et al., 2014). Fos-positive cells counts. To analyse group differences (Sham vs Peri lesion; baseline vs novel context) in the ROIs, two Other hippocampal-related areas. Fos-positive cell counts were between-subject factors (surgical condition and baseline/novel made within the prelimbic cortex (PL; AP +3.72 to +2.76), the gran- context) and one within-subject factor (ROI) ANOVA was calcu- ular retrosplenial cortex (RSP; AP −2.28 to −3.36), and nucleus lated. This analysis was carried out separately for three regional reuniens (AP −1.44 to −2.28). The granular retrosplenial cortex groupings: (1) hippocampal subfields, (2) parahippocampal cor- (area 29) was selected as it is both the principal recipient of the tices, and (3) other hippocampal-related areas. These regional projections from the hippocampal formation to this region and the groupings helped to reduce type 1 errors by limiting the number source of its projections to entorhinal cortex (Van Groen and Wyss, of comparisons. The Fos counts in perirhinal cortex (Sham 1990, 1992, 2003). There are also direct projections from the tem- groups only) were compared using a one between-subject (base- poral region of CA1 and the subiculum to prelimbic cortex, with line/novel context) by one within-subject factor (areas 35 and 36) return projections via nucleus reuniens of the thalamus (Conde ANOVA. When an interaction was significant (p ⩽ 0.017 cor- et al., 1995; Prasad and Chudasama, 2013; Vertes et al., 2007). rected for multiple tests), simple effects were examined. While the Fos counts from the novel groups were normally distributed, cell counts from both baseline control groups (‘Peri Statistical analysis Baseline’ and ‘Sham Baseline’) were not (Shapiro–Wilk test). Behavioural results. Activity scores from the ‘novel’ groups As the baseline Fos counts in all ROIs were positively skewed were separated between ‘same beam’, that is, a single beam being and their means were proportional to their variance, a square-root Kinnavane et al. 5 transformation was applied to the data (Howell, 2011) when the McIntosh, 2006; Schumacker and Lomax, 2010). Subsequently, analyses involved only these groups. Other analyses involving all each path can be independently unconstrained and the fit com- four groups used the raw Fos counts for comparability, mindful pared to the structural weights model, again using a χ difference that ANOVA is relatively robust to violations of the normality test to determine in which path the difference occurs (Protzner assumption when group sample sizes are equal (Howell, 2011). and McIntosh, 2006). When there were marked differences in the Pearson product-moment correlation coefficients were calcu- overall fit of the same model between two or more groups, the lated for the Fos-positive cell counts in the various sites, as well alpha level of the first χ difference test was slightly relaxed in as with the activity of animals in the novel context condition. order to explore the potential reasons why one group had poor fit. In both baseline control groups, the Pearson product-moment correlation coefficients were calculated based on the trans- Results formed scores as these data were subsequently used for SEM, where normality is assumed (Arbuckle, 2011). The histological and behavioural analyses only relate to those animals used for the c-fos analyses. As explained, the findings came from two cohorts of rats. The data from these two cohorts Structural equation modelling were repeatedly compared, though the outcome is only presented when there was a significant cohort difference (p ⩽ 0.05). Rats SEM uses multiple-equation regression models to quantify from both cohorts populated all conditions. potentially causal relationships between sets of variables in a theoretical structure, thereby testing models that can include the potential direction of effects (Schumacker and Lomax, 2010). Lesion analysis (In some cases, a direction of effect could not be inferred as the Based on the exclusion criteria (see section ‘Lesion analysis’), fit of the models did not change when the path direction was seven animals were removed from group Peri Novel and three reversed. This situation is indicated in the figures by a double- were excluded from Peri Baseline. Following these exclusions, headed arrow.) The SEM software package, SPSS AMOS version the group numbers were as follows: Peri Novel, n = 11 (3 from 20.0 (IBM Corp, Armonk, NY, USA) computed the analyses. cohort A, 8 from cohort B); Peri Baseline, n = 12 (6 from cohort Maximum likelihood estimation, which is recommended for use A, 6 from cohort B); Sham Novel, n = 11 (4 from cohort A, with smaller sample sizes (Arbuckle, 2011), allowed the pro- 7 from cohort B); Sham Baseline, n = 12 (7 from cohort A, gramme to estimate effects among variables. All models tested 5 from cohort B). Group Peri Novel contained six left and five were based on well-established anatomical connections (Furtak right hemispheres. For group Peri Baseline, three left and nine et al., 2007; Van Strien et al., 2009; Witter, 2002). right hemispheres were analysed. The corresponding hemi- An anatomically plausible model was specified and the covar- spheres were analysed in the matched Sham surgical controls. iance matrix of the regional Fos counts estimated the strength The lesions involved much of the AP extent of perirhinal cor- of the relationship (path) between regions as set out in this tex, with only small regions of tissue sparing (Figure 3). The model. The path coefficient of a connection between two regions extent of perirhinal tissue loss in those hemispheres analysed for (Arbuckle, 2011) estimates the ‘effective connectivity’ or the Fos expression in the Peri Novel context condition ranged from extent to which one region directly influences the other (Protzner 40.6% to 73.9% (cohort A) and 67.8% to 98.5% (cohort B). The and McIntosh, 2006). Models were assessed based on how well corresponding extent of tissue loss in the Peri Baseline condition the implied (estimated) variance–covariance matrix replicates ranged from 70.7% to 89.9% (cohort A) and 68.6% to 100% the sample (observed) variance–covariance matrices of the (cohort B). The extent of tissue loss did not differ between the observed data (Schumacker and Lomax, 2010). A model with 2 2 Peri Novel and Peri Baseline groups (t < 1). good fit has a non-significant χ , and the ratio of χ to the degrees The attempt to make near-complete perirhinal cortex lesions of freedom is <2 (Tabachnick and Fidell, 2001). The comparative led to some extra-perirhinal damage. This additional damage was fit index (CFI) and the root mean square error of approximation typically in the most ventral parts of area Te2 and the most dorsal (RMSEA) are additional measures of fit that are applicable for parts of the piriform cortex and LEC, that is, those cortical areas smaller sample sizes (Fan and Wang, 1998; Hu and Bentler, immediately adjacent to perirhinal cortex (Figure 3). 1998). A CFI of >0.9 is considered acceptable (Schumacker and Lomax, 2010), while a RMSEA of <0.08 is considered accepta- ble (Tabachnick and Fidell, 2001). Given the relatively small Behavioural testing group sizes, each model should contain twice as many cases as the number of variables to be estimated (Bollen and Long, 1992; Analyses of the beam breaks over the 20 min session (‘same Wothke, 1993). Finally, the squared multiple correlation (R or beam’ and ‘beam crossovers’) found no overall effect of coefficient of determination) is presented, indicating the amount perirhinal cortex lesions (F(1, 18) = 1.36, p = 0.26). Similarly, of variance in each brain region accounted for by the model there was no interaction between lesion and type of beam break (Arbuckle, 2011). (F(1, 18) = 3.21, p = 0.09). When the beam breaks were divided The various groups were compared on the same network into 5 min bins, the activity levels showed a highly significant model by a stacking procedure. In this procedure, the path coef- reduction across the 20 min session (F(3, 36) = 11.4, p < 0.001), ficients of all paths in the model are constrained so that they must with no effect of surgery (F < 1) and no interaction between have the same value for all groups, creating a ‘structural weights these factors (F < 1). Both the perirhinal lesion and sham control model’. If the model fit when the paths are constrained is signifi- rats showed a significant decrease in activity. (Note: these data cantly worse than when the paths are free to have different values were only available for 14 rats from cohort B.) This reduction for each group (as determined by a χ difference test), this indi- in activity is assumed to principally reflect habituation to a cates that the paths differ among the groups (Protzner and novel environment. 6 Brain and Neuroscience Advances Figure 3. Perirhinal lesion reconstructions. Diagrammatic reconstructions of the perirhinal cortex lesions showing the individual cases with the largest (grey) and smallest (black) lesions for rats from cohorts A and B in groups Peri Novel and Peri Baseline. The left panel illustrates regions involved for comparison. Sites highlighted: areas 35 and 36 of the perirhinal cortex, insular cortex, lateral entorhinal cortex (LEC), and piriform cortex. The numbers refer to the distance (in mm) from bregma (note that the hemispheres analysed came from both right and left hemispheres). Source: Adapted from Paxinos and Watson (2005). context seemingly differed among subfields (this novelty dif- Comparisons of Fos-positive cell counts ference was highly significant in all subfields, F(1, 42) > 27, Hippocampal subfields. Being placed in the novel context dra- p ⩽ 0.001; Figure 4, upper panel). While this increase seemed matically increased c-fos activity, although the perirhinal cortex most evident in CA1, scaling effects were present. Comparisons lesions had no apparent effect on the mean Fos counts in the of Fos-positive cell counts across the 10 hippocampal subfields hippocampal formation (Figure 4). A significant Mauchly test found no overall effect of perirhinal lesions (F(1, 42) = 1.43, (p ⩽ 0.001) indicated that the assumption of sphericity of the p = 0.24; Figure 4, upper panel). Similarly, there was no lesion by within-subject variable (ROI) was violated and so Greenhouse– context interaction (F < 1), lesion by subfield interaction (F(2.8, Geisser corrected degrees of freedom and p-values are presented 116) = 1.12, p = 0.34), or three-way interaction (F < 1). for the within-subject comparisons (Howell, 2011). Hippocampal Fos counts in the novel context rats were con- Parahippocampal cortices. Overall, novel context exploration sistently considerably higher than those of the rats in the baseline produced higher Fos counts in the MEC, LEC, and postrhinal (home-cage) controls (F(1, 42) = 166, p ⩽ 0.001), with individual cortex than remaining in the home cage (F(1, 40) = 113, p ⩽ 0.001; areas showing different levels of Fos expression (F(2.8, Figure 5(a)). Rats with perirhinal cortex lesions had lower 116) = 101.1, p ⩽ 0.001; Figure 4, upper panel). There was a Fos-positive cell counts across the parahippocampal cortices, significant context by subfield interaction (F(2.8, 116) = 48.2, with the LEC seemingly most affected (Figure 5(a)). As in the p ⩽ 0.001) as the increase in Fos counts from baseline to novel hippocampus, the assumption of sphericity was violated, and so Kinnavane et al. 7 Figure 4. Mean Fos-positive cell counts per group in the hippocampal formation. Top panel: Graph of results from all hippocampal sites analysed: CA fields – intermediate (inter), septal (sept), and temporal (temp); dentate gyrus (DG); and subiculum (Sub). Exposure to a novel context reliably increased Fos-related activity in all regions analysed (p < 0.001). Data are presented as means ± SEM. Middle panel: Representative photomicrographs from coronal sections that depict Fos-positive cells in intermediate and temporal levels of the hippocampus for all behavioural conditions. Scale bar: 1000 µm. Bottom panel: Higher magnification photomicrographs of regions corresponding to the dashed rectangle of the photomicrograph above. Scale bar: 50 µm. Greenhouse–Geisser corrected degrees of freedom and p-values numerically greater Fos increase in area 35 (F(1, 21) = 44.2, are presented (Howell, 2011). p ⩽ 0.001) than area 36 (F(1, 21) = 25.5, p ⩽ 0.001; Figure 5(b)). The context manipulation differentially affected the three Across the three parahippocampal regions analysed in all four cortical areas (F(1.2, 49) = 113.5, p ⩽ 0.001) although simple groups, Fos counts were numerically lower in the Peri rats than in effects revealed that MEC, LEC, and postrhinal cortex all had the Sham controls (Figure 5(a)). While this contrast did not reach higher Fos-positive cell counts when exposed to the novel con- the corrected levels of significance (F(1, 40) = 5.41, p = 0.025), text compared to baseline (all F > 46.8, p ⩽ 0.001; Figure 5(a)). there was a significant region by lesion interaction (F(1.2, Numerically, this increase appeared greatest in LEC, although 49) = 6.00, p = 0.013). This interaction indicated that the various this may have been due to scaling effects generated by the com- parahippocampal areas were differentially affected by the per- paratively higher Fos counts in LEC than in MEC or the pos- irhinal lesions. Simple effects revealed that this interaction trhinal cortex. For perirhinal cortex (Figure 5(b)), counts could reflected decreased Fos counts in the LEC of the rats with per- only be made in the two Sham groups. Once again, the novel irhinal lesions (F(1, 40) = 6.56, p = 0.014; Figure 5(a)), a lesion context condition raised Fos counts (F(1, 21) = 41.7, p ⩽ 0.001; effect that did not extend to the MEC or postrhinal cortex Figure 5(b)). Overall, the Fos counts in areas 35 and 36 did (F(1, 40) = 3.44, p = 0.071; F(1, 40) = 1.59, p = 0.21, respectively). not differ (F < 1), although the context manipulation affected Finally, the three-way interaction (area, lesion, and context) was the two areas differently (F(1, 21) = 9.84, p = 0.005) with a non-significant (F(2, 80) = 1.61, p = 0.21). 8 Brain and Neuroscience Advances Figure 5. Mean Fos-positive cell counts per group in the parahippocampal formation. (a) Sites analysed in all four groups: lateral entorhinal cortex (LEC), medial entorhinal cortex (MEC), and postrhinal cortex (POR). (b) Sites analysed in only the Sham controls: areas 35 and 36 of the perirhinal cortex. Exposure to a novel context reliably increased Fos-related activity in all regions analysed (p < 0.001). Data are presented as means ± SEM. Other hippocampal-related areas. While exposure to a novel context dramatically increased Fos expression (F(1, 42) = 74.7, p ⩽ 0.001) in all three areas (Figure 6), the perirhinal cortex lesions did not affect overall Fos activity in the prelimbic cortex, retrosplenial cortex, or nucleus reuniens of the thalamus (F < 1). While Fos counts differed between areas (F(2, 84) = 105.9, p ⩽ 0.001), there was no lesion by context interaction (F < 1) or region by lesion interaction (F < 1). Although the context by region interaction was significant (F(2, 84) = 49.2, p ⩽ 0.001), with the retrosplenial cortex showing a greater increase than prelimbic cortex, the relatively low counts in nucleus reuniens created a scaling effect. Activity behaviour and Fos-positive cell counts. For each of the two relevant groups (Sham Novel and Peri Novel), only one site had an initial significant correlation (p < 0.05). In both cases, the Fos-positive cell counts correlated positively with ‘same beam’ breaks (Sham Novel, intermediate CA3, r = 0.75, p = 0.012: Peri Novel, temporal CA3, r = 0.63, p = 0.038). However, in neither case did these effects survive correction for multiple comparisons, suggesting that the Fos counts were not a direct Figure 6. Mean Fos-positive cell counts per group in other hippocampal- product of the amount of locomotor activity. related areas. Sites analysed: prelimbic cortex (PL), retrosplenial cortex (RSP), and nucleus reuniens of the thalamus (Reuniens). Exposure to a novel context reliably increased Fos-related activity in all regions Structural equation modelling analysed (p < 0.001). Data are presented as mean ± SEM. Initial inspection of all of the inter-area correlations revealed an apparent difference between the novel context and baseline direction of these connections. A valid model of context learning control conditions. For both the Sham Baseline group (120/170) would be expected to have good fit for the novel context condi- and the Peri Baseline group (87/120), a large majority of the tion but not the baseline (home-cage) condition. inter-area Fos count correlations were significant at an uncor- rected level (~70% at p < 0.05). In contrast, for both the Sham 1. Is novel context exposure associated with specific network Novel group (41/170) and the Peri Novel group (32/120), the patterns of c-fos activity predicted by the BIC framework and is corresponding proportion was much lower (~25%, i.e. almost this affected by perirhinal cortex damage? The first model to be one-third of the number). tested used the parahippocampal (postrhinal)–medial entorhinal For SEM, all of the networks examined had to have anatomi- network described by Ritchey et al. (2015). In this refined cal plausibility with respect to their interconnections and the Kinnavane et al. 9 Figure 7. Testing the posterior–medial system of the BIC framework. (a) The posterior–medial system has good fit for group Sham Novel. (b) The same network model for group Peri Novel has poor fit. The same model also has poor fit for (c) group Sham Baseline and (d) group Peri Baseline. Model fit is noted at the bottom of each model (comparative fit index (CFI); root mean square error of approximation (RMSEA)). The strength of the causal influence of each path is denoted both by the thickness of the arrow and by the path coefficient next to that path. The number above the top right corner of each area box is the R value, denoting the variance accounted for by the inputs to that region. Sites depicted: medial entorhinal cortex (MEC); postrhinal cortex (POR); retrosplenial cortex (RSP); and hippocampal subfields CA1, CA3, and dorsal subiculum (sub). (Models with a grey background have poor fit.) *p < 0.05; **p < 0.01; ***p < 0.001. version of the BIC framework, interactions between the para- Fos activity data of the two groups did not differ between the hippocampal (postrhinal) cortices and retrosplenial cortex are regions set out in the model, this contrast was close to the level of included, creating what is referred to as the posterior–medial significance, and as the model had poor fit for group Sham (PM) system (Figure 1). For this initial analysis, the Fos counts Baseline, further examination took place. When the pathways along the longitudinal hippocampal axis were combined, that is, that compose the model were individually unconstrained, the the temporal, intermediate, and septal subregions of the dentate functional connection between postrhinal cortex and MEC was χ = 49 . 2 gyrus, CA3, and CA1. This decision reflects the way in which a found to be stronger in the Sham baseline group ( , 1Diff coronal section across entorhinal cortex will include connections p = 0.027), while the functional connection between CA3 and χ = 66 . 7 along the full longitudinal axis of the hippocampus (Furtak et al., CA1 was stronger in group Sham Novel ( , p = 0.009; 1Diff χ < 19 . 2007; Van Strien et al., 2009). A maximum of six nodes could be all other paths: ). 1Diff included in each model given the sample size (Bollen and Long, The specificity of the PM system was tested in two further 1992; Wothke, 1993). ways. First, we tested the complementary item division of the This PM system, which is depicted in Figure 7, was found to updated BIC framework, that is, the anterior–temporal (AT) χ = 6.26 have good fit for group Sham Novel context ( , p = 0.51; system (Figure 1; see Ritchey et al., 2015). Fos counts from the CFI = 1.0; RMSEA = 0.0; Figure 7(a)). In contrast, the same perirhinal entorhinal cortex and LEC replaced those from the χ =10.3 model did not fit the Sham Baseline Fos data ( , p = 0.17; postrhinal entorhinal cortex and MEC, while the ventral sub- CFI = 0.96; RMSEA = 0.21; Figure 7(c)). When compared iculum replaced the dorsal subiculum. This AT model had only χ = 67 . 4 directly by stacking the data from these two groups on the same poor fit for group Sham Novel ( , p = 0.15; CFI = 0.71; model, the model in which the path coefficients were all free to RMSEA = 0.26). vary did not have significantly better fit than the model in which Second, data were taken from a previous c-fos experiment all path coefficients were constrained to be the same for both that matched this study in all respects, except for one critical fea- χ =13.87 groups ( , p = 0.085). While this indicates that the ture. Rats in that experiment were exposed to multiple novel 8Diff 10 Brain and Neuroscience Advances Figure 8. Optimal model for group Sham Novel. (a) This network, which is expanded to include prelimbic cortex, has optimal fit for data from group Sham Novel. (b) The same network for group Peri Novel has poor fit. The same model also has poor fit for group (c) Sham Baseline and (d) group Peri Baseline. Model fit is noted at the bottom of each model (comparative fit index (CFI); root mean square error of approximation (RMSEA)). The strength of the causal influence of each path is denoted both by the thickness of the arrow and by the path coefficient next to that path. The number above the top right corner of each area box is the R value, denoting the variance accounted for by the inputs to that region. Sites depicted: medial entorhinal cortex (MEC); prelimbic cortex (PL); retrosplenial cortex (RSP); and hippocampal subfields CA1, CA3, and dorsal subiculum (sub). (Models with a grey background have poor fit.) *p < 0.05; **p < 0.01; ***p < 0.001. object recognition problems in a familiar environment (Kinnavane early in the model. All of the paths in the model were significant, et al., 2016). Consequently, activity should be biased towards and all paths have directionality as the fit of the model was worse the AT item system, that is, the perirhinal cortex and LEC. when the direction of each path was reversed. It should be noted Consistent with that prediction, models based on the AT system that the hippocampal Fos data presented here are counts com- had good fit (Kinnavane et al., 2016). However, when the bined along the longitudinal hippocampal axis. If dorsal CA3 and postrhinal–medial entorhinal network was tested using the Fos CA1 counts are substituted for the combined counts, the model counts from that same object recognition study (Kinnavane et al., retains acceptable but inferior fit. Whereas if ventral (temporal) χ =15.2 CA3 and CA1 counts are substituted, this produces a poorly fit- 2016), the resulting model was of very poor fit ( , ting model (data not presented). p = 0.004; CFI = 0.74; RMSEA = 0.50). Surprisingly, when the This same optimal network model did not have acceptable lev- same postrhinal–medial entorhinal network was applied to the els of fit for any of the other three behavioural groups (Figure control condition from that study, which involved novel objects 8(b)–(d)). When the two intact groups were directly compared by but no familiarity discrimination, the model retained its fit χ = 32 . 3 stacking their Fos data on the same model (Figure 8(a)), the over- ( , p = 0.52; CFI = 1.0; RMSEA = 0.0). all group difference between model fit was close to being signifi- Finally, evidence that the perirhinal lesions disrupted the PM cant ( χ =11.5 system of the BIC framework (Figure 7) came from the finding , p = 0.075). As the model had poor fit for group 6Diff that the Fos data from the Peri Novel group had only poor fit Sham Baseline, additional analyses were conducted. When each χ = 25.75 ( , p = 0.001; CFI = 0.51; RMSEA = 0.52). Consistent of the component paths was allowed to vary individually, only χ = 44 . 1 with the above results, the updated BIC framework also failed freeing the paths from retrosplenial cortex to MEC ( , 1Diff 2 2 χ = 93 . 9 χ = 66 . 2 to fit the data from group Peri Baseline ( , p = 0.27; p = 0.036) and from CA3 to CA1 ( , p = 0.010) sig- 7 1Diff χ < 12 . CFI = 0.96; RMSEA = 0.18). nificantly improved fit (all other paths: ). This differ- 1Diff ence potentially reflects the strengthening of intrinsic hippocampal 2. What is the optimal model for the Sham Novel context connections with novel context exploration (Figure 8(a)). group? For group Sham Novel Context, the optimal model None of the network models with acceptable fit for group involved many of the regions implicated in the PM network of Sham Novel transferred over to the Sham Baseline group (e.g. χ = 76 . 6 the updated BIC framework ( , p = 0.57; CFI = 1.00; Figures 7 and 8). This failure again suggests that the context- RMSEA = 0.00; Figure 8(a)). Interestingly, the prelimbic and ret- driven models are specific and not simply driven by correlations rosplenial cortices had better predictive value when positioned associated with baseline Fos expression. Kinnavane et al. 11 Finally, evidence that perirhinal lesions disrupted network This specificity was tested in two further ways. First, com- activity in the medial temporal lobe came from the fact that it was parable models were examined using perirhinal cortex and not possible to generate a network model of acceptable fit with LEC, instead of the postrhinal entorhinal cortex and MEC. data from group Peri Novel. Additionally, when the data from the A decision was made not to divide the subiculum and CA1 Fos two novel context groups (Sham Novel and Peri Novel) were counts based on their distal and proximal locations, in order to stacked on the optimal model for Sham Novel (Figure 8), the ensure that all aspects of the models to be compared were held the activity-related Fos data differed significantly between the two same, aside from the introduction of the perirhinal entorhinal cor- χ =17.6 groups ( , p = 0.007). To investigate further, each of 6Diff tex and LEC. The resulting analyses, which tested the AT item the pathways was individually unconstrained revealing signifi- system of the BIC framework (Ritchey et al., 2015), failed to pro- χ = 83 . 0 cant differences in the steps from MEC to CA1 ( , 1Diff vide models of acceptable fit in the novel context groups. Second, χ = 40 . 0 p = 0.004) and CA1 to subiculum ( , p = 0.047; 1Diff data were taken from a previous experiment that examined medial Figure 8). Additionally, for group Sham Novel, the correlation temporal c-fos activity after a test of object recognition memory, between Fos counts in MEC and CA1 was strong and positive again in rats with perirhinal lesions and their surgical sham con- (r = 0.79, p = 0.004), whereas in group Peri Novel, this correlation trols (Kinnavane et al., 2016). The Fos counts from that surgical was negative and non-significant (r = −0.24, p = 0.48). Formal sham group (analysis not presented) failed to fit the PM system, comparison of these correlations using Fisher’s r-to-z transforma- but did fit the anterior–temporal system. Somewhat surprisingly, tion revealed that these correlations were significantly different the Fos data from their control group (Kinnavane et al., 2016), (z = 2.6, p = 0.009). Taken together, these results indicate that the which was exposed to novel objects but did not make recognition perirhinal cortex lesions altered coordinated activity between the discriminations, could fit the PM system. entorhinal cortex and CA1 when animals explored a novel context. Other evidence for the specificity of the context network models came from the baseline home-cage control groups. A striking feature in both the Sham Baseline and Peri Baseline Discussion groups was the high level of correlations between Fos levels in the different areas sampled (around 70% of all sites examined), This study sought to test networks of interlinked c-fos activity which contrasted with that found in the novel context groups associated with context learning in both intact rats and rats with (both around 25%). In the resting condition, the default state perirhinal cortex lesions. In one condition, rats were placed in a appears to involve widespread levels of inter-correlated activity. novel environment (unfamiliar cages in an unfamiliar room), and This pattern changes in the face of a particular learning challenge, in the other, they remained in their home cages. Although this for example, new contextual information. Now, more specific comparison brings additional changes in locomotor and arousal networks become engaged, so decreasing overall site-to-site levels between the two conditions, it has the benefit of creating interactions across multiple brain areas. robust, marked differences in c-fos expression, so more reliably This study also assessed the impact of perirhinal cortex testing any impact of perirhinal cortex loss. A further point is that lesions on medial temporal lobe c-fos activity. Perirhinal lesions the study did not include additional tests to confirm learning about did not disrupt the size of the hippocampal Fos increase when the novel context, aside from the evidence of habitation that came rats are moved to a novel context. Similarly, overall levels of from the locomotor scores. It should, however, be remembered c-fos expression in prelimbic cortex, retrosplenial cortex, and that context learning is regarded as spontaneous (Dix and nucleus reuniens of the thalamus all appeared unaffected by the Aggleton, 1999; Good et al., 2007) and that the context shift used perirhinal cortex lesions. Perirhinal lesions did, however, reduce in this study would be considered highly salient. Consequently, it c-fos expression in the parahippocampal region, an effect most cannot be excluded that changes in c-fos expression may have apparent in the LEC. This result closely matches the findings been driven by those differences in arousal, locomotor activity or from a related study that used object recognition to examine the anxiety, associated with experiencing a novel context. impact of perirhinal lesions on c-fos expression (Kinnavane The neural networks tested were based on recent refinements et al., 2016). The common finding of LEC hypoactivity under- of the BIC framework (Diana et al., 2007; Ranganath, 2010; lines the particularly close anatomical and functional links Ritchey et al., 2015) which emphasises relationships between between the perirhinal cortex and this entorhinal division (see parahippocampal (postrhinal), medial entorhinal, and hippocam- also Burwell and Amaral, 1998a; Wilson et al., 2013a, 2013b; pal areas for context learning. The resulting PM system (Ritchey Witter et al., 2000). Further evidence of perirhinal lesion effects et al., 2015) was tested using SEM. Networks closely based on came from the repeated failure to find medial temporal networks the PM system had good fit for the intact novel context group of acceptable fit for the Peri Novel group. (Figure 7). Furthermore, the optimal network model for this Kent and Brown (2012) suggest that the perirhinal cortex role novel context group incorporated much of the PM system, while in item processing extends to learning about complex features also adding further inputs from prelimbic cortex (Figure 8). This within contextual surroundings based on unitising stimulus rep- optimal novel context model retained its fit when the Fos counts resentations. Support comes from evidence that perirhinal lesions came from just the dorsal hippocampus but not the ventral can impair fear conditioning to complex auditory cues, as well as hippocampus. This result is consistent with the outcome of con- contextual conditioning (Bucci et al., 2000, 2002; Burwell et al., text reactivation studies based on c-fos expression in the dorsal 2004; Corodimas and LeDoux, 1995; Kholodar-Smith et al., hippocampus (Liu et al., 2012, 2014; Ramirez et al., 2013). The 2008a, 2008b; Lindquist et al., 2004; Sacchetti et al., 1999). In same spatial networks, that is, those based on the PM system contrast, perirhinal lesions spare fear conditioning to continuous (Ritchey et al., 2015), did not have acceptable fit for either of the tones (Bucci et al., 2000; Kholodar-Smith et al., 2008a; Lindquist baseline (home-cage) groups. These null results point to the specificity of the BIC framework for contextual learning. et al., 2004). Additionally, increased c-fos expression in the 12 Brain and Neuroscience Advances perirhinal cortex is associated with context shifts (VanElzakker from single unit recordings that the rat LEC also plays a role in et al., 2008; Vann et al., 2000), as well as with contextual fear spatial processing, often in relation to item location (Deshmukh conditioning, but not cued fear conditioning (Albrechet-Souza et al., 2012; Hunsaker and Kesner, 2013; Knierim et al., 2014; et al., 2011). Thus, the perirhinal cortex may be involved in dis- Neunuebel et al., 2013). The conclusion is that while there is a criminating and, hence, helping to bring together novel compo- major division of information processing pathways in the medial nents within a given context, even though this cortical area may temporal lobe and beyond, there remain important interactions be insensitive to their relative spatial disposition (Aggleton et al., between these same pathways at multiple levels, including those 2010; Deshmukh et al., 2012; Jenkins et al., 2004; Wan et al., between parahippocampal areas. 1999). It is presumably this latter aspect, along with the relative preservation of inter-hippocampal activity, as seen in this study Declaration of conflicting interests (see also Kinnavane et al., 2016), which helps to explain why The author(s) declared no potential conflicts of interest with respect to perirhinal cortex lesions often spare those tests of allocentric spa- the research, authorship, and/or publication of this article. tial memory that are highly sensitive to hippocampal damage (Glenn and Mumby, 1998; Machin et al., 2002; Ramos, 2013; Funding Winters et al., 2004). Many of these same tests make additional This work was supported by the Wellcome Trust (WT087955 and demands on navigation, an ability closely linked with medial WT09520). entorhinal–hippocampal function, rather than perirhinal cortex (Buzsáki and Moser, 2013). References It would be wrong, however, to infer that perirhinal cortex lesions are without effect on hippocampal spatial processing. In Aggleton JP (2012) Multiple anatomical systems embedded within the this study, perirhinal lesions altered entorhinal cortex activity, so primate medial temporal lobe: Implications for hippocampal func- tion. Neuroscience and Biobehavioral Reviews 36(7): 1579–1596. disrupting parahippocampal–hippocampal interactions, for Aggleton JP, Albasser MM, Aggleton DJ, et al. (2010) Lesions of the example, those with CA1. With this result in mind, it may be rat perirhinal cortex spare the acquisition of a complex configural relevant that electrophysiological studies have shown that while visual discrimination yet impair object recognition. Behavioral perirhinal lesions do not appear to affect the initial formation of Neuroscience 124(1): 55–68. CA1 place fields, their stability is reduced (Muir and Bilkey, Aggleton JP, Kyd R and Bilkey D (2004) When is the perirhinal cortex 2001). Other reported hippocampal changes following perirhinal necessary for the performance of spatial memory tasks? Neurosci- lesions are a reduction in the proportion of theta cells, a phase ence and Biobehavioral Reviews 28(6): 611–624. shift in CA1 place cells, and an altered modulation by movement Albasser MM, Davies M, Futter JE, et al. (2009) Magnitude of the object velocity on place cells (Muir and Bilkey, 2003). At the same time, recognition deficit associated with perirhinal cortex damage in rats: subtle deficits after perirhinal cortex lesions have been reported Effects of varying the lesion extent and the duration of the sample period. Behavioral Neuroscience 123(1): 115–124. by some when using spatial tasks sensitive to hippocampal dam- Albasser MM, Olarte-Sánchez CM, Amin E, et al. (2015) Perirhinal cor- age (Aggleton et al., 2004; Liu and Bilkey, 1998a, 1998b, 2001; tex lesions in rats: Novelty detection and sensitivity to interference. Wiig and Bilkey, 1994a, 1994b). The current data suggest that Behavioral Neuroscience 129(3): 227–243. these milder deficits stem from parahippocampal dysfunctions, Albasser MM, Poirier GL and Aggleton JP (2010) Qualitatively differ- such as those that were evident in LEC. ent modes of perirhinal-hippocampal engagement when rats explore This study set out to test medial temporal models that novel vs. familiar objects as revealed by c-fos imaging. European describe the processing of context information. Expression of Journal of Neuroscience 31(1): 134–147. c-fos revealed inter-related activity networks that closely match Albrechet-Souza L, Borelli KG, Almada RC, et al. (2011) Midazolam those predicted by refinements of the BIC framework. Arguably, reduces the selective activation of the rhinal cortex by contextual fear one unexpected finding was the direction of influence in the opti- stimuli. Behavioural Brain Research 216(2): 631–638. Arbuckle JL (2011) IBM SPSS AMOS 20 User’s Guide. Chicago, IL: mal model for the intact novel context group (Figure 8). While all AMOS Development Corporation. connections were significantly connected, the best fit was Barker GRI, Bird F, Alexander V, et al. (2007) Recognition memory for obtained when the model progressed from prelimbic cortex to objects, places, and temporal order: A disconnection analysis of the retrosplenial cortex and then to the medial temporal lobe. In prac- role of the medial prefrontal cortex and perirhinal cortex. Journal of tice, the PM contextual system (Ritchey et al., 2015; Figure 1) Neuroscience 27(11): 2948–2957. can accommodate these findings as it includes reciprocal connec- Bast T (2007) Toward an integrated perspective on hippocampal func- tions between the retrosplenial cortex and medial temporal sites. tion: From a rapid encoding of experience to adaptive behaviour. A final point concerns the impression that there are distinct Reviews in the Neurosciences 18(3–4): 253–281. parahippocampal pathways for ‘what’ and ‘where’ information. It Bisler S, Schleicher A, Gass P, et al. (2002) Expression of c-Fos, ICER, has been argued that a more nuanced division might be more Krox-24 and JunB in the whisker-to-barrel pathway of rats: Time course of induction upon whisker stimulation by tactile exploration appropriate, such as that between local and global reference of an enriched environment. Journal of Chemical Neuroanatomy frames (Knierim et al., 2006, 2014). As part of this process, a 23(3): 187–198. more complete model will need to include crosstalk between the Bollen KA and Long JS (1992) Tests for structural equation models: two putative pathways, reflecting their interconnectivity (Burwell Introduction. Sociological Methods & Research 21(2): 123–131. and Amaral, 1998b). Indeed, evidence of lateral entorhinal– Bucci DJ and Robinson R (2014) Toward a conceptualization of retro- medial entorhinal interactions was found in this study. For exam- hippocampal contributions to learning and memory. Neurobiology of ple, when the Sham Novel context model (Figure 7) is modified Learning and Memory 116: 197–207. so that LEC replaces the postrhinal cortex, the model still retains Bucci DJ, Phillips RG and Burwell RD (2000) Contributions of post- good fit (not presented). It was also the case that the perirhinal rhinal and perirhinal cortex to contextual information processing. lesions disrupted the PM system. Similarly, there is evidence Behavioral Neuroscience 114(5): 882–894. Kinnavane et al. 13 Bucci DJ, Saddoris MP and Burwell RD (2002) Contextual fear discrimi- formation but not the parahippocampal cortices; a c-fos expression nation is impaired by damage to the postrhinal or perirhinal cortex. study. Neuroscience 124(1): 43–52. Behavioral Neuroscience 116(3): 479–488. Kent BA and Brown TH (2012) Dual functions of perirhinal cortex in Burwell RD (2000) The parahippocampal region: Corticocortical fear conditioning. Hippocampus 22(10): 2068–2079. connectivity. Annals of the New York Academy of Sciences 911: Kholodar-Smith DB, Allen TA and Brown TH (2008a) Fear conditioning 25–42. to discontinuous auditory cues requires perirhinal cortical function. Burwell RD (2001) Borders and cytoarchitecture of the perirhinal and Behavioral Neuroscience 122(5): 1178–1185. postrhinal cortices in the rat. Journal of Comparative Neurology Kholodar-Smith DB, Boguszewski P and Brown TH (2008b) Auditory 437(1): 17–41. trace fear conditioning requires perirhinal cortex. Neurobiology of Burwell RD and Amaral DG (1998a) Cortical afferents of the perirhinal, Learning and Memory 90(3): 537–543. postrhinal and entorhinal cortices of the rat. Journal of Comparative Kinnavane L, Amin E, Horne M, et al. (2014) Mapping parahippocam- Neurology 398(2): 179–205. pal systems for recognition and recency memory in the absence of Burwell RD and Amaral DG (1998b) Perirhinal and postrhinal cortices of the rat hippocampus. European Journal of Neuroscience 40(12): the rat: Interconnectivity and connections with the entorhinal cortex. 3720–3734. Journal of Comparative Neurology 391(3): 293–321. Kinnavane L, Amin E, Olarte-Sánchez CM, et al. (2016) Mapping Burwell RD, Bucci DJ, Sanborn MR, et al. (2004) Perirhinal and responses to novel objects: The impact of perirhinal cortex discon- postrhinal contributions to remote memory for context. Journal of nection on hippocampal activity patterns. Hippocampus 26(11): Neuroscience 24(49): 11023–11028. 1393–1413. Buzsáki G and Moser EI (2013) Memory, navigation and theta rhythm Knierim JJ, Lee I and Hargreaves EL (2006) Hippocampal place cells: in the hippocampal-entorhinal system. Nature Neuroscience 16(2): Parallel input streams, subregional processing and implications for 130–138. episodic memory. Hippocampus 16(9): 755–764. Chaudhuri A (1997) Neural activity mapping with inducible transcription Knierim JJ, Neunuebel JP and Deshmukh SS (2014) Functional correlates factors. NeuroReport 8(13): 3–7. of the lateral and medial entorhinal cortex: Objects, path integration Conde F, Marie-Lepoivre E, Audinat E, et al. (1995) Afferent connec- and local-global reference frames. Philosophical Transactions of the tions of the medial frontal cortex of the rat II: Cortical and subcorti- Royal Society B: Biological Sciences 369(1635): 20130369. cal afferents. Journal of Comparative Neurology 352(4): 567–593. Lindquist DH, Jarrard LE and Brown TH (2004) Perirhinal cortex sup- Corodimas KP and LeDoux JE (1995) Disruptive effects of posttraining ports delay fear conditioning to rat ultrasonic social signals. Journal perirhinal cortex lesions on conditioned fear: Contributions of con- of Neuroscience 24(14): 3610–3617. textual cues. Behavioral Neuroscience 109(4): 613–619. Liu P and Bilkey DK (1998a) Excitotoxic lesions centered on perirhinal Deshmukh SS, Johnson JL and Knierim JJ (2012) Perirhinal cortex rep- cortex produce delay-dependent deficits in a test of spatial memory. resents nonspatial, but not spatial, information in rats foraging in Behavioral Neuroscience 112(3): 512–524. the presence of objects: Comparison with lateral entorhinal cortex. Liu P and Bilkey DK (1998b) Lesions of perirhinal cortex produce Hippocampus 22(10): 2045–2058. spatial memory deficits in the radial maze. Hippocampus 8(2): Diana RA, Yonelinas AP and Ranganath C (2007) Imaging recollec- 114–121. tion and familiarity in the medial temporal lobe: A three-component Liu P and Bilkey DK (2001) The effect of excitotoxic lesions centred on model. Trends in Cognitive Sciences 11(9): 379–386. the hippocampus or perirhinal cortex in object recognition and spa- Dix SL and Aggleton JP (1999) Extending the spontaneous preference tial memory tasks. Behavioral Neuroscience 115(1): 94–111. test of recognition: Evidence of object-location and object-context Liu X, Ramirez S and Tonegawa S (2014) Inception of a false memory by recognition. Behavioural Brain Research 99(2): 191–200. optogenetic manipulation of a hippocampal memory engram. Philo- Eacott MJ and Gaffan EA (2005) The roles of perirhinal cortex, post- sophical Transactions of the Royal Society B: Biological Sciences rhinal cortex, and the fornix in memory for objects, contexts, and 369(1633): 20130142. events in the rat. Quarterly Journal of Experimental Psychology B: Liu X, Ramirez S, Pang P, et al. (2012) Optogenetic stimulation of Comparative and Physiological Psychology 58(3–4): 202–217. a hippocampal engram activates fear memory recall. Nature Fan X and Wang L (1998) Effects of potential confounding factors 484(7394): 381–385. on fit indices and parameter estimates for true and misspecified Machin P, Vann SD, Muir JL, et al. (2002) Neurotoxic lesions of the SEM models. Educational and Psychological Measurement 58(5): rat perirhinal cortex fail to disrupt the acquisition or performance of 701–735. tests of allocentric spatial memory. Behavioral Neuroscience 116(2): Furtak SC, Wei S-M, Agster KL, et al. (2007) Functional neuroanatomy 232–240. of the parahippocampal region in the rat: The perirhinal and postrhi- McIntosh AR and Gonzalez-Lima F (1991) Structural modeling of func- nal cortices. Hippocampus 17(9): 709–722. tional neural pathways mapped with 2-deoxyglucose: Effects of Glenn MJ and Mumby GD (1998) Place memory is intact in rats with peri- acoustic startle habituation on the auditory system. Brain Research rhinal cortex lesions. Behavioral Neuroscience 112(6): 1353–1365. 547(2): 295–302. Good MA, Barnes P, Staal V, et al. (2007) Context-but not familiar- Muir GM and Bilkey DK (2001) Instability in the place field location ity-dependent forms of object recognition are impaired following of hippocampal place cells after lesions centered on the perirhinal excitotoxic hippocampal lesions in rats. Behavioral Neuroscience cortex. Journal of Neuroscience 21(11): 4016–4025. 121(1): 218–223. Muir GM and Bilkey DK (2003) Theta- and movement velocity-related Howell DC (2011) Statistical Methods for Psychology (8th edn). firing of hippocampal neurons is disrupted by lesions centered on the Wadsworth, OH: Cengage Learning. perirhinal cortex. Hippocampus 13(1): 93–108. Hu L and Bentler PM (1998) Fit indices in covariance structure mod- Mumby DG and Pinel JP (1994) Rhinal cortex lesions and object recog- eling: Sensitivity to underparameterized model misspecification. nition in rats. Behavioral Neuroscience 108(1): 11–18. Psychology Methods 3(4): 424–453. Naber PA, Caballero-Bleda M, Jorritsma-Byham B, et al. (1997) Parallel Hunsaker MR and Kesner RP (2013) The operation of pattern separation input to the hippocampal memory system through peri- and postrhi- and pattern completion processes associated with different attributes nal cortices. NeuroReport 8(11): 2617–2621. or domains of memory. Neuroscience and Biobehavioral Reviews Neunuebel JP, Yoganarasimha D, Rao G, et al. (2013) Conflicts between 37(1): 36–58. local and global spatial frameworks dissociate neural representations Jenkins TA, Amin E, Pearce JM, et al. (2004) Novel spatial arrange- of the lateral and medial entorhinal cortex. Journal of Neuroscience ments of familiar stimuli promote activity in the rat hippocampal 33(22): 9246–9258. 14 Brain and Neuroscience Advances Norman G and Eacott MJ (2005) Dissociable effects of lesions Van Groen T and Wyss JM (2003) Connections of the retrosplenial gran- to the perirhinal cortex and the postrhinal cortex on memory ular b cortex in the rat. Journal of Comparative Neurology 463(3): for context and objects in rats. Behavioral Neuroscience 119(2): 249–263. 557–566. Van Strien NM, Cappaert NLM and Witter MP (2009) The anatomy Olarte-Sánchez CM, Kinnavane L, Amin A, et al. (2014) Contrasting of memory: An interactive overview of the parahippocampal– networks for recognition memory and recency memory revealed by hippocampal network. Nature Reviews Neuroscience 10(4): immediate-early gene imaging in the rat. Behavioral Neuroscience 272–282. 128(4): 504–522. VanElzakker M, Fevurly RD, Breindel T, et al. (2008) Environmental Paxinos G and Watson C (2005) The Rat Brain in Stereotaxic Coordi- novelty is associated with a selective increase in Fos expression in nates (5th edn). San Diego, CA: Elsevier Academic Press. the output elements of the hippocampal formation and the perirhinal Prasad JA and Chudasama Y (2013) Viral tracing identifies parallel cortex. Learning & Memory 15(12): 899–908. disynaptic pathways to the hippocampus. Journal of Neuroscience Vann SD, Brown MW, Erichsen JT, et al. (2000) Fos imaging reveals 33(19): 8494–8503. differential patterns of hippocampal and parahippocampal subfield Protzner AB and McIntosh AR (2006) Testing effective connectivity activity in response to different spatial memory tasks. Journal of changes with structural equation modeling: What does a bad model Neuroscience 20(7): 2711–2718. tell us? Human Brain Mapping 27(12): 935–947. Vertes RP, Hoover WB, Szigeti-Buck K, et al. (2007) Nucleus reuniens Ramirez S, Liu X, Lin P-A, et al. (2013) Creating a false memory in the of the midline thalamus: Link between the medial prefrontal cortex hippocampus. Science 341(6144): 387–391. and the hippocampus. Brain Research Bulletin 71(6): 601–609. Ramos JMJ (2013) Differential contribution of hippocampus, peri- Wan H, Aggleton JP and Brown MW (1999) Different contributions rhinal cortex and postrhinal cortex to allocentric spatial memory of the hippocampus and perirhinal cortex to recognition memory. in the radial maze. Behavioural Brain Research 247: 59–69. Journal of Neuroscience 19(3): 1142–1148. Ranganath C (2010) Binding items contexts: The cognitive neuroscience Wiig KA and Bilkey DK (1994a) Perirhinal cortex lesions in rats disrupt of episodic memory. Current Directions in Psychological Science performance in a spatial DNMS task. NeuroReport 5(11): 1405–1408. 19(3): 131–137. Wiig KA and Bilkey DK (1994b) The effects of perirhinal cortical lesions Ranganath C and Ritchey M (2012) Two cortical systems for on spatial reference memory in the rat. Behavioural Brain Research memory guided behaviour. Nature Reviews Neuroscience 13(10): 63(1): 101–109. 713–726. Wilson DIG, Langston RF, Schlesiger MI, et al. (2013a) Lateral ento- Ritchey M, Libby LA and Ranganath C (2015) Cortico-hippocampal rhinal cortex is critical for novel object-context recognition. systems involved in memory and cognition: The PMAT framework. Hippocampus 23(5): 352–366. Progress in Brain Research 219: 45–64. Wilson DIG, Watanabe S, Milner H, et al. (2013b) Lateral entorhinal Sacchetti B, Lorenzini CA, Baldi E, et al. (1999) Auditory thalamus, dor- cortex is necessary for associative but not nonassociative recognition sal hippocampus, basolateral amygdala, and perirhinal cortex role in memory. Hippocampus 23(12): 1280–1290. the consolidation of conditioned freezing to context and to acous- Winters BD, Forwood SE, Cowell RA, et al. (2004) Double dissocia- tic conditioned stimulus in the rat. Journal of Neuroscience 19(21): tions between the effects of peri-postrhinal cortex and hippocampal 9570–9578. lesions on tests of object recognition and spatial memory: Heteroge- Schumacker RE and Lomax RG (2010) A Beginners Guide to Structural neity of function within the temporal lobe. Journal of Neuroscience Equation Modelling (3rd edn). New York: Routledge. 24(26): 5901–5908. Strange BA, Witter MP, Lein E, et al. (2014) Functional organization Witter MP (2002) The parahippocampal region: Past, present, and future. of the hippocampal longitudinal axis. Nature Reviews Neuroscience In: Witter M and Wouterlood F (eds) The Parahippocampal Region: 15(10): 655–669. Organization and Role in Cognitive Function. Oxford: Oxford Swanson LW (1992) Brain Maps: Structure of the Rat Brain. University Press, pp. 1–20. Amsterdam: Elsevier. Witter MP, Naber PA, van Haeften T, et al. (2000) Cortico-hippocampal Tabachnick BG and Fidell LS (2001) Using Multivariate Statistics communication by way of parallel parahippocampal-subicular path- (4th edn). Needham, MA: Allyn & Bacon. ways. Hippocampus 10(4): 389–410. Van Groen T and Wyss JM (1990) Connections of the retrosplenial gran- Wothke W (1993) Nonpositive definite matrices in structural equation ular a cortex in the rat. Journal of Comparative Neurology 300(4): modeling. In: Bollen KA and Long JS (eds) Testing Structural Equa- 593–606. tion Models. London: SAGE, pp. 256–293. Van Groen T and Wyss JM (1992) Connections of the retrosplenial dys- Zangenehpour S and Chaudhuri A (2002) Differential induction and granular cortex in the rat. Journal of Comparative Neurology 315(2): decay curves of c-fos and zif268 revealed through dual activity 200–216. maps. Molecular Brain Research 109(1–2): 221–225.

Journal

Brain and Neuroscience AdvancesSAGE

Published: Mar 14, 2017

Keywords: Entorhinal cortex; hippocampus; nucleus reuniens; prefrontal cortex; retrosplenial cortex; spatial memory; subiculum

There are no references for this article.