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

Learn More →

Rapid generation of sequence-diverse terminator libraries and their parameterization using quantitative Term-Seq

Rapid generation of sequence-diverse terminator libraries and their parameterization using... Synthetic biology and the rational design and construction of biological devices require vast numbers of characterized biological parts, as well as reliable design tools to build increasingly complex, multigene architectures. Design principles for intrinsic terminators have been established; however, additional sequence-structure studies are needed to refine parame- ters for termination-based genetic devices. We report a rapid single-pot method to generate libraries of thousands of ran- domized bidirectional intrinsic terminators and a modified quantitative Term-Seq (qTerm-Seq) method to simultaneously identify terminator sequences and measure their termination efficiencies (TEs). Using qTerm-Seq, we characterize hundreds of additional strong terminators (TE > 90%) with some terminators reducing transcription read-through by up to 1000-fold in Escherichia coli. Our terminator library and qTerm-Seq pipeline constitute a flexible platform enabling identifica- tion of terminator parts that can achieve transcription termination not only over a desired range but also to investigate their sequence-structure features, including for specific genetic and application contexts beyond the common in vivo systems such as E. coli. Key words: transcriptional regulatory elements; genetic circuit engineering; biological techniques; gene expression regula- tion; synthetic biology. 1. Introduction facilitated predictive models to assist with synthetic cistron Forward engineering of genetic devices to carry out useful func- construction (5, 11, 12). tions is at the very core of synthetic biology. However, achieving Engineered genetic circuits consisting of one or a few genes predictable control of gene expression requires a detailed un- have been successfully applied to create a variety of genetic derstanding of relevant biological processes and the availability devices such as: oscillators (13), toggle switches (14) and logic of a sufficient number of characterized genetic parts for gene gates (15). However, de novo engineering of more complex, construction (1). Genetic parts libraries have been developed multigene synthetic devices are still limited by an incomplete mostly for bacteria (especially Escherichia coli) and they have pro- understanding of fundamental mechanisms of gene expression vided genetic engineers with repertoires to modulate gene ex- and/or accurate modeling of complex cellular environments pression at the transcriptional (2–5), post-transcriptional (6, 7) and gene interactions. Indeed, standardizing biological parts and translational levels (8–10). Experimental characterization of with predictable properties remains challenged by the finding single or collections of natural or synthetic parts has also that part performance may differ drastically in different DNA Submitted: 23 August 2019; Accepted: 3 October 2019 V The Author(s) 2019. Published by Oxford University Press. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. 1 Downloaded from https://academic.oup.com/synbio/article-abstract/4/1/ysz026/5609127 by DeepDyve user on 25 March 2020 2| Synthetic Biology, 2019, Vol. 4, No. 1 (or RNA transcript) sequence contexts (16). Development of design parameters using only two DNA oligonucleotides and high-throughput methods that characterize the performance of commonplace molecular biology enzymes and equipment. large collections of genetic parts under specific physiological We also apply a variation of the high-throughput terminator conditions and genetic contexts are, therefore, needed to refine RNA-sequencing methodology Term-Seq (21) to examine syn- genetic design principles and afford finer control over gene thetic terminator performance in vivo under a variety of condi- expression. tions. Applying these methods, we characterize hundreds of Transcription termination is a fundamental step of gene ex- additional strong intrinsic terminators in forward and reverse pression in all organisms and serves to prevent unintended orientations that can be used for engineering bacterial cistrons. transcription of flanking gene sequences, to define RNA tran- These methods provide a foundation for larger-scale studies to script 3 ends and to recycle RNA polymerase for subsequent dissect sequence-structure features influencing transcription rounds of transcription (17). In bacteria, transcriptional termi- termination and to identify synthetic riboregulators that nators are found near the end of operons where they modulate achieve a desired TE and/or respond to changing cellular condi- transcription of downstream genes and, in some cases, make tions and chassis. regulatory decisions in response to changing cellular conditions (e.g. small molecule recognition by terminator/anti-terminator 2. Materials and methods riboswitches) (18). Compared with other basic biological parts (e.g. transcription promoters and ribosome binding sites), 2.1 Terminator library construction transcription terminators have been less explored as tools to Terminator libraries were assembled in four steps consisting of: control gene expression; however, their utility as synthetic gene (i) oligo annealing, (ii) oligo extension, (iii) oligo ligation and expression regulators is exemplified by their repurposing as (iv) polymerase chain reaction (PCR; Supplementary Note S1). All adjustable high-pass and low-pass biological filters (19) and oligonucleotides for the study were synthesized by Integrated their integration into functional synthetic riboswitches (20). DNA Technologies and are detailed in Supplementary Table S1. Moreover, recent high-throughput RNAseq strategies such as Term-Seq (21) have facilitated the discovery of novel riboregula- i. Oligo annealing 0 0 tors which provide additional gene expression regulation tools One hundred picomoles of 5 half (oTerm13) and 5 -end available to synthetic biologists and provide additional mecha- monophosphorylated 3 half (oTerm14) oligonucleotides nistic insights into factor-dependent and intrinsic termination were mixed in 100 ml of ribonuclease-free ultrapure water mechanisms (21, 22). (MilliQ ) and heated in a thermocycler to 95 C for 2 min, The mechanism and sequence features affecting intrinsic followed by slowly cooling at a rate of 2 C per minute until termination efficiency (TE) have been extensively characterized the sample reached 21 C. Samples were then placed on ice. (17). In E. coli, intrinsic terminators minimally consist of a G-C- ii. Oligo extension rich hairpin (T ) that is immediately followed by a 7–9 nucleo- hp Annealed oligos were extended by T4 DNA polymerase for tide poly(U) tract (23). Intrinsic termination begins with stalling 10 min at 21 Cin 20 ml reactions containing 10 pmol of of RNA polymerase on the poly(U) tract due to the especially annealed oligos, 1 NEBuffer 2.1 (New England Biolabs, weak rU-dA hybrid and this creates a kinetic window for T for- hp NEB), 250 nmol each dNTP and 3 units of T4 DNA polymer- mation in the nascent RNA (24). Invasion of the T into the hp ase (NEB). Extended oligo complexes were then extracted RNA exit channel of RNAP then results in shearing of the rU-dA via phenol-chloroform at 21 C, ethanol precipitated and hybrid (25, 26) or hyper-translocation of RNAP (25, 27) with suspended in 15 ml of nuclease-free MilliQ ddH O. concomitant structural changes in RNAP that promote tran- iii. Oligo ligation scription elongation complex (TEC) dissociation (17). Stronger 0 0 Oligo 5 and 3 halves were ligated in 20 ml reactions contain- base pairing of the T and a perfect, extended poly(U) tract are hp ing 10 pmol annealed and extended oligos, 1 T4 DNA li- associated with increased TE (12) and the strength of the four gase buffer containing rATP (NEB) and 10 U T4 DNA ligase closing base pairs of the T and first three uridines of the hp (NEB). Ligations were incubated for 20 min at 16 C and then poly(U) tract are particularly conserved features of strong intrin- heat-inactivated at 65 C for 15 min. sic terminators (12, 23). Some intrinsic terminators also possess iv. Polymerase chain reaction a poly(A) tract upstream of the T and this feature is generally hp One microliter of ligated oligo reaction was used as tem- associated with increased TE and can enable terminators to act plate in 100 ml PCR reactions containing 1 ThermoPol in both forward and reverse directions—so-called bidirectional Buffer (Thermo Scientific), 200 mM dNTPs, 10 pmol each for- terminators (17). ward (oTerm16) and reverse (oTerm17) primers and 0.2 U Intrinsic terminator libraries are valuable tools for Phusion DNA polymerase (Thermo Scientific). PCR was car- examining RNA/DNA sequence features that affect TE and for ried out with an initial denaturation step at 98 C for 30 s, 25 identifying sequence-diverse intrinsic terminators with speci- cycles consisting of 98 C for 10 s, 55 C for 10 s and 72 C fied efficiencies for genetic circuit design (12); however, to for 20 s, and a final extension step at 72 C for 5 min. PCR the best of our knowledge, no current method can generate products were verified on 8% native polyacrylamide gels, bacterial transcriptional terminator libraries with randomized blunt-end cloned into the pJET1.2 (Thermo Scientific) using hairpin sequences without requiring many oligonucleotides manufacturer’s protocols and fifty clones were subjected to (i.e. one or two oligos per terminator) or relatively expensive Sanger sequencing to evaluate library quality and sequence massively parallel DNA synthesis strategies. Development of complexity. inexpensive and straightforward methods would facilitate high-throughput studies to further dissect RNA and DNA 2.2 Dual fluorescence reporter and library cloning sequences and/or structures affecting intrinsic termination and provide a greater collection of characterized terminators for The pBeRG reporter plasmid was created by cloning a synthetic genetic engineering. Here, we describe a novel workflow to gen- gBlock cassette (Integrated DNA Technologies) containing an erate thousands of bacterial intrinsic terminators with flexible E. coli codon-optimized, N-terminally FLAG-tagged monomeric Downloaded from https://academic.oup.com/synbio/article-abstract/4/1/ysz026/5609127 by DeepDyve user on 25 March 2020 A.J. Hudson and H.-J. Wieden | 3 blue-exciting red fluorescent protein (mBeRFP) (28) coding se- values from 0% to 50% (weak terminators), 50–87% (intermedi- quence complete with a strong 5 Shine-Dalgarno sequence and ate-weak), 87–95% (intermediate-strong) and >95% (strong; V R 3 flanking BioBrick prefix and suffix sequences at the EcoRI Supplementary Figure S3A). Sorted cell fractions were used to and NdeI restriction sites upstream of the enhanced green fluo- inoculate 5 ml of fresh LB medium, grown to mid-log phase at rescent protein (eGFP) gene of the host plasmid pBbE6a (29) (see 37 C with shaking and plated on LB plates. Fractioned Amp100 Supplementary Figure S1 for annotated plasmid sequence). clone libraries were then cultivated, aliquoted and stored as de- mBeRFP and eGFP strongly excite at 488 nm but have widely scribed above. separated emission peaks (610 and 510 nm, respectively) that minimize fluorescence resonance energy transfer and emission 2.4 Terminator RNAseq channel bleed-through. pBeRG was transformed into NEB5a E. Total RNA was isolated from terminator library E. coli cultures coli chemically competent cells (NEB) and subsequently verified as described by Bernstein et al. (30) with modifications. by Sanger sequencing. Terminator library PCR products were Terminator library FACS-sorted library clones and control ter- cloned nondirectionally into pBeRG at the two NotI sites of the minator control cultures were cultivated in LB media BioBrick prefix and suffix sequences and 100 ng of ligated plas- Amp100 overnight, sub-cultured into fresh LB media, grown until mid was transformed into NEB5a high-efficiency chemically Amp100 competent E. coli cells. Several transformation products were mid-log phase (OD ¼ 0.4) with shaking at 37 C and induced with 1 mM IPTG. Temperature-dependent termination experi- pooled and plated on LB agar with 100 mg/ml ampicillin (LB ). Transformation efficiency was estimated by plating ments were performed as described above except library clones Amp100 1% of transformation on separate LB plates. The final ter- were incubated at 14 C, 21 C, 30 Cor37 C for 15 min prior to in- Amp100 minator library (10 000 clones) was then suspended in 5 ml of duction with continued incubation at the specified temperature LB medium containing 20% glycerol, flash frozen in liquid Amp100 for 1 h (Supplementary Figure S3B). For both FACS-sorted and nitrogen and stored at 80 C. temperature-dependent experiments, 5 ml aliquots of induced cultures were added to tubes containing 1 ml of ice-cold 95% ethanol containing and 5% saturated phenol (pH 6.8; to inhibit 2.3 Flow cytometry and fluorescence-activated cell RNA degradation), rapidly mixed by vortexing, placed on ice and sorting experiments subsequently flash frozen in liquid nitrogen and stored at 80 C Flow cytometry analysis of library clones (n ¼ 180) as well as ter- until required (less than 1 week). Total RNA was extracted via minator control plasmid clones was performed by inoculating the hot phenolic extraction method followed by ethanol precipi- 5ml LB liquid medium with individual E. coli (NEB5a) colo- Amp100 R tation and elution in 50 ml of RNase-free water (MilliQ ). nies and growing at 37 C with shaking at 200 RPM for 16 h. Contaminating DNA was removed by DNase I digestion in 20 ml Cultures were then used to inoculate fresh 5 ml LB media, Amp100 reactions containing 10 mg of total RNA, 2 U DNase I (Thermo grown to mid-log phase (OD ¼ 0.4) with shaking at 37 C and Scientific) and 1 DNase I reaction buffer and incubating at induced by addition of 1 mM IPTG. Fluorescent proteins were 37 C for 1 h followed by phenol extraction and ethanol precipi- expressed overnight (16 h) to allow for maximal fluorescent pro- tation and elution in 20 ml of RNase-free water (MilliQ ). RNA tein maturation, prior to flow cytometry (see Supplementary sample quality was assayed on 1% agarose gels and quantitated Note S2 for detailed explanation). Ten to 100 ml aliquots of in- TM using a BioDrop spectrophotometer. duced overnight cultures were diluted in 1 ml of phosphate- RNA samples were prepared for Illumina MiSeq Next buffered saline and analyzed on a Becton-Dickenson V R TM Generation Sequencing using the NEBNext Multiplex Small FACSAria Fusion flow cytometer with excitation at 488 nm RNA Library Prep Set for Illumina (New England Biolabs) with and FITC and Texas Red channel filters for detection of eGFP modifications (Supplementary Figure S3C). Because the 3 end of and mBeRFP fluorescence, respectively. Forward and side scat- reporter read-through transcripts would lie several hundred tering gates were configured to select single-cell events and nucleotides downstream of the terminator cloning site, a new photomultiplier tube voltages were adjusted to 500–600 V to in- proximal 3 end was generated for read-through transcripts by crease sensitivity for weak eGFP emission and to maximize the annealing an antisense DNA oligo that binds approximately dynamic range of detection. Apparent TE was calculated from 30 nt downstream from the terminator site and digesting with measured fluorescent protein emissions using the formula: RNase H. Digestions were carried out in 20 ml reactions contain- 1 ing 10 mg of total RNA, 100 pmol antisense oligo (oTerm-Seq-R), hmBi0 hmBiTerm TE ¼ 1   100% 1 RNase H reaction buffer and 5 U RNase H (NEB) and incu- heGi0 heGiTerm bated for 1 h at 37 C followed by phenol extraction and ethanol precipitation as described above. RNase H-treated samples were Where, ‘mB’ and ‘eG’ are the mBeRFP and eGFP fluorescence V R then prepared using the NEBNext Multiplex Small RNA Library intensity, respectively. Clones containing pBeRG plasmids with Prep Set for Illumina (New England Biolabs) and barcoded as per terminator parts (‘Term’) were normalized to clones containing the manufacturer’s instructions, except the 5 linker ligation pBeRG plasmids that lack a terminator part (‘0’) to account for step was omitted. Library first-strand cDNAs were amplified us- cell-to-cell plasmid copy number variation. Flow cytometry ing the P7 reverse primer and a custom gene-specific forward experiments were performed in duplicate with TE mean and primer that binds 50 bp upstream of the terminator cloning standard deviation calculated from all cell events from each ex- site and bares the P5 oligo sequences at its 5 end (oTerm-Seq- periment. Plasmids from 50 selected clones from flow cytome- F5-P5). Barcoded terminator libraries were pooled and se- try experiments were subjected to Sanger sequencing quenced on an Illumina MiSeq Sequencer (GeneWiz) which (Supplementary Table S2). yielded approximately 40 million, 2 150 bp paired-end reads Fluorescence-activated cell sorting (FACS) experiments were performed as per flow cytometry, however, using entire E. coli with 73% of reads  Q30. Illumina adapters were removed using terminator library samples. E. coli terminator library clones were Cutadapt (31) with a threshold of 90% identity to adapter sorted into four bins with gates corresponding to apparent TE sequences. Only reads that passed quality control (Q30) and Downloaded from https://academic.oup.com/synbio/article-abstract/4/1/ysz026/5609127 by DeepDyve user on 25 March 2020 4| Synthetic Biology, 2019, Vol. 4, No. 1 bared a complete T sequence were considered for analysis even relatively small combinatorial libraries (100 sequences) hp (Supplementary Tables S3 and S4). both expensive and time-consuming. We, therefore, sought to Unique terminator sequence identifiers were assigned with develop methods to create libraries consisting of thousands of the designation ‘T’ and numbered in descending order of their unique intrinsic terminator variants while maintaining control sequence read abundance. Terminator counts were normalized over features such as the length and composition of the T , hp for each sample replicate (in transcripts per million, TPM) and poly(U) and poly(A) tracts and adjacent sequences (e.g. for TM custom Python scripts (see Section 2.6) were used to tabulate cloning or part assembly purposes). Although we attempted transcript sequencing read lengths which served to determine several strategies for terminator library synthesis (many other whether transcripts were terminated or read-through. strategies are likely possible), we found a strategy that uses two Transcripts were deemed as ‘terminated’ if their length corre- semi-randomized DNA oligonucleotides that provides a great sponded to þ1to þ8 nt downstream of the last nucleotide of the compromise in terms of economy (could be performed rapidly T while transcripts with length greater than this (beyond the hp and inexpensively, using readily available lab materials) and poly(U) tract) were counted as ‘read-through.’ The top 2000 ter- 0 flexibility over terminator library design. In this method, two 5 minators with the greatest number of total reads were then 0 and 3 half oligos are designed using a computer to contain assessed for TE by counting the number of terminated tran- desired features (see Supplementary Note S1 for design consid- scripts divided by total reads for each terminator variant. Only erations) and generated by conventional DNA oligo synthesis. terminator variants that had 200 or more reads in either FACS Oligos are then assembled into terminator libraries in four steps binning or differential temperature experiments are displayed (see Section 2.1). (see Supplementary Tables S3 and S4 for all terminator sequen- Because strong intrinsic terminators (>90% TE) with diverse ces). T minimal free-energy predications and predicted sec- hp sequences are particularly desirable for insulating synthetic cis- ondary structures were calculated for each terminator variant trons, we applied our terminator library methodology to create using the RNAfold program from the Vienna RNA Package (32) very strong intrinsic terminators: a 14 bp, perfectly paired hair- with folding parameters set to 37 C and -p -d2 –noLP modifiers. pin with a basal 8 bp randomized portion that is closed by two Terminators identified in FACS-sorted terminator RNAseq strong (G-C or C-G) pairs, a GAAA tetraloop (or UUUC in reverse), (Term-Seq) libraries were normalized to TPM values, based on a downstream 8 nt poly(U) tract and an upstream 8 nt poly(A) the total number of reads in each barcoded library. Terminators tract (12)(Figure 1A). We placed BioBrick RFC 10 standard prefix were examined for enrichment in FACS bins as per the bin with and suffix sequences in the upstream and downstream con- the greatest mean TPM values (most abundant bin), as well as stant regions to allow the library parts to be easily assembled by use of the two-tailed Student’s t-tests, which compared TPM into genetic circuits via BioBrick assembly (http://parts.igem. counts from each FACS bin from triplicate biological replicates org/Assembly: Standard_assembly). Our library design has a (df ¼ 1; Supplementary Table S3). If t-test P-values were <0.05 theoretical maximum diversity of 16 384 sequences and once for any one bin relative to all other bins, the terminator was designed, terminator library assembly could be performed in a considered significantly enriched in that bin. For differential single 8-h day. temperature experiments, TE values for terminators were calcu- lated for each temperature treatment from triplicate experi- 3.2 Synthetic terminator library characterization via ments. Individual terminator TEs were then compared for each temperature treatment using two-tailed Student’s t-tests for flow cytometry significant differences in terminator read-through efficiency We cloned our terminator library nondirectionally into a cus- (P< 0.05, df ¼ 1). V R tom bi-cistronic terminator testing device (pBeRG) at a BioBrick RFC 10 multicloning site (Figure 1B) and examined TEs of termi- 2.6 Data availability nators from our library (n ¼10 000 clones) using a fluorescence-based interference strategy and flow cytometry. All custom scripts used for Term-Seq have been made available Flow cytometry revealed a bimodal distribution for all termina- at GitHub (https://github.com/andyhudson42/TermSeq.git). Raw tor library clones and two distinct populations with median sequence data for Term-Seq were deposited and available from fluorescence interference TEs (designated TE ) of 13.6% and the NCBI Sequence Read Archive (SRA) under BioProject acces- IF 96.5%, relative to no terminator (Figure 1C and Supplementary sion: PRJNA503821. All other relevant data are available upon re- Figure S2). For comparison, we benchmarked terminators from quest from the authors. our library against two previously characterized intrinsic termi- nators: BBa_B0012 (medium-efficiency terminator) and 2.7 Materials availability BBa_B0013 (high-efficiency terminator) from the Registry of All DNA constructs will be provided upon request following the Standardized Biological Parts (http://parts.igem.org). In pBeRG, completion of a Materials Transfer Agreement and any other flow cytometry experiments predicted 90.9% and 96.5% TE for IF documentation that may be required. the BBa_B0012 and BBa_B0013 terminators, respectively (Figure 1C and Supplementary Table S2). TEs for 180 library clones were then measured individually using flow cytometry. 3. Results More than half of the measured library terminators displayed 3.1 Rapid synthesis of intrinsic terminator libraries apparent TE of 90% or greater and many terminators showed IF even higher efficiency than the strong BBa_B0013 terminator Terminator libraries have been valuable tools for examining (Figure 1C). Fifty clones were Sanger sequenced, which identi- RNA/DNA sequence–function relationships and for establishing fied only unique terminator variants and confirmed successful design principles for synthetic terminators with desirable TEs (5, 12). However, prior terminator library construction has re- terminator library cloning with sufficient diversity to warrant quired assembly of individually synthesized terminator parts, further examination using higher throughput methods making the ability to probe the functional sequence space of (Supplementary Table S2). Downloaded from https://academic.oup.com/synbio/article-abstract/4/1/ysz026/5609127 by DeepDyve user on 25 March 2020 A.J. Hudson and H.-J. Wieden | 5 Figure 1. Intrinsic terminator library synthesis and measurement by flow cytometry. (A) Terminator library design used in this study. The corresponding RNA se- V R quence for the terminator library is indicated with boxes representing BioBrick RFC 10 prefix and suffix sequences. The randomized portion of the T is in red text, hp where ‘S’ indicates a ‘C’ or ‘G’ nucleotide and ‘N’ is any nucleotide. The forward T loop sequence is shown with the reverse sequence shown above. (B). Schematic of hp the bi-cistronic pBeRG terminator testing device is shown using SBOL visual format and indicates the terminator cloning site (BioBrick multicloning site, MCS). See Supplementary Figure S1 for pBeRG vector sequence. (C) Flow cytometry measurements for 180 E. coli NEB5a terminator library clones were performed in duplicate and TE values were calculated (see Methods section). Standard deviations are shown for clones as red lines and the pBeRG (without terminator, first bar on left), BBa_B0012 terminator and BBa_B0013 terminator clones are indicated as blue, orange and gold bars. See Supplementary Table S2 for terminator sequences. identify terminator sequences and evaluate transcription TE 3.3 High-throughput measurement of terminators using of terminators from our synthetic library (see Section 2 and qTerm-Seq Supplementary Figure S3). Flow cytometry-based methods have been routinely applied to To determine whether TEs from flow cytometry-based (TE ) IF examine TEs of intrinsic terminators in vivo (5, 12); however, were comparable to those obtained by qTerm-Seq (TE ), we TS they are less amenable to high-throughput examination of used FACS to separate terminator library clones into four frac- terminators (e.g. hundreds of terminators) due to the need to in- tions prior to qTerm-Seq with FACS gates corresponding to: 0% dividually clone and measure terminator parts. Transcriptome- to 50% TE (Weak terminators), 50–87% TE (Intermediate-Weak), based methods are highly scalable and an attractive alternative 87–95% TE (Intermediate-Strong) and >95% TE (Strong). to flow cytometry for higher throughput studies examining qTerm-Seq identified over 10 000 unique library sequences; transcription termination, due to the ability to examine thou- however, some of these are expected to represent cloning or sands of transcripts in a single experiment. Term-Seq is a re- sequencing artifacts (e.g. only single-read representation). cently described RNAseq method that maps RNA transcript 3 We examined the distribution of 3 ends for all qTerm-seq reads ends to identify transcription termination regulators in E. coli relative to the pBeRG terminator reporter construct to identify (22) and other bacteria (21). However, while Term-Seq identifies putative termination sites, it has not been applied to quantitate putative transcription termination sites (Supplementary Figure S4A). Transcript 3 end positions showed similar distributions TE. Therefore, we developed a quantitative Term-Seq method and bioinformatic strategy (qTerm-Seq) that can simultaneously for all four FACS bins, with exception of weak terminators, Downloaded from https://academic.oup.com/synbio/article-abstract/4/1/ysz026/5609127 by DeepDyve user on 25 March 2020 6| Synthetic Biology, 2019, Vol. 4, No. 1 which suggests that terminators separated into the remaining However, stronger terminators generally showed less variation bins had similar transcriptional and degradation profiles in vivo between their forward and reverse orientations than weaker (Supplementary Figure S4A). A large proportion of the qTerm- terminators and several terminators maintained greater 0 0 0 Seq reads were 3 truncated within the 5 and 3 arms of the T than 99% TE in both directions (Figure 2D and Supplementary hp TS portion of terminator library sequences (Supplementary Figure Table S3). S4A). Although this collection may indeed represent bona fide transcription termination sites or stable reporter mRNA degra- 3.4 Expression temperature does not strongly affect TE TS dation products, a complete T sequence could not be deter- hp of library terminators mined for these reads and they were removed from further Terminator hairpin folding dynamics during transcription can analysis. A peak þ1to þ3 nt downstream of the T was also ev- hp be affected by expression temperature and plausibly this may ident and due to their close proximity to the expected termina- tion site for intrinsic terminators (þ6to þ8 nt downstream of confer changes in terminator efficiency. Therefore, we also expressed our terminator library at 14 C, 21 C, 30 C and 37 Cin T )(23) were deemed to have arisen from transcription termi- hp nation events (Supplementary Figure S4A). qTerm-Seq reads E. coli prior to qTerm-Seq to examine possible temperature- whose 3 ends extended beyond þ8 nt downstream of the T dependent termination effects in our library (Supplementary hp were classified as read-through events in which RNA polymer- Figure S3B). After removing low-abundance terminators (<200 ase did not terminate at the candidate terminator. reads), we identified 495 terminator variants, which did not After removing truncated reads with an incomplete T and hp overlap significantly with those from the FACS-enriched termi- terminator sequences with fewer than 200 reads, 622 unique nator pool (Supplementary Table S4). Although measured termi- full-length terminator variants remained for analysis nator TE spanned two orders of magnitude (0.62–99.96%), TS (Supplementary Table S3). The overall distribution of terminator most of the terminators did not display significant differences library TE from the 622 FACS-enriched terminators mimics TS in TE at expression temperatures from 14 Cto 37 C and very the distribution obtained from individual flow cytometry strong terminators showed the least temperature-dependent experiments (cf. Figures 1C and 2A). Terminator library TE TS variance in TE (Figure 2E and Supplementary Figure S6). TS predicted by qTerm-Seq also corresponded closely with enrich- Patterns of qTerm-Seq 3 end positions did not show obvious ment in the four FACS bins (Figure 2B), indicating that qTerm- deviations for different expression temperatures, except at an Seq and flow cytometry measurements generate comparable R upstream site within the BioBrick prefix sequence which oc- apparent TEs. curred more frequently at 14 C and 21 C(Supplementary Figure TE measurement for the BBa_B0012 and BBa_B0013 termi- TS S4B). Several terminators were found that had significantly dif- nators was 92.8% (SD ¼ 1.9%) and 99.1% (SD ¼ 0.21%), respec- ferent TE from 14 Cto 37 C(Figure 2E and Supplementary TS tively and 2–3% higher than TE values predicted by flow IF Table S4) and most of these exhibited a higher TE at lower TS cytometry experiments for the two terminators. Unfortunately, temperature with the largest difference being a 78% decrease at none of the 50 sequences obtained from Sanger sequencing 14 C as compared with 37 C (T528). were present in the 622 terminators measured in qTerm-Seq and these could not be directly compared using the two 3.5 Terminator features correlating to TE methods. From the collection of 622 FACS-enriched terminators, 423 Using the collection of library terminators measured by qTerm- terminators had a TE greater than 90% and 180 terminators TS Seq, we looked for terminator features that correlated to TE . TS were greater than 99% (Figure 2A and Supplementary Figure S5 T predicted free-energies somewhat correlated with TE and hp TS and Table S3). The strongest 26 terminators had TE greater TS there was an apparent division in T free-energy with termina- hp than 99.9%, corresponding to more than a 1000-fold difference tors with a DG of less than 20 kJ/mol having TEs greater than in read-through, relative to the lowest measured terminators 90% and terminators with greater DG typically having lower TE from the library. As expected for our library design, most strong (Figure 3A and B and Supplementary Tables S3 and S4). terminators featured a perfectly base-paired T . Weak termi- hp Terminator hairpin G/C content did not show any obvious rela- nators (< 50% TE ) were found to be mostly terminator assem- TS tionship to TE , although surprisingly, the strongest termina- TS bly or cloning artifacts that could not form a strong hairpin tors in the library generally had less G/C-rich hairpins structure, and this is consistent with the importance of T base hp (Figure 3A). It is possible that PCR amplification of the termina- pairing integrity for efficient intrinsic termination (24). tor library and our qTerm-Seq library preparation methods fa- Consistent with prior studies (12, 33), we observed that nucleo- vored more A/T-rich terminators and that even stronger (more tide substitutions that disrupt base pairing within the middle of G/C rich) terminators may exist in our library but escaped detec- the T reduced TE to a lesser extent than the two closing T hp TS hp tion using qTerm-Seq. base pairs and nucleotide insertions between the T and the hp Because our terminator library only varies the proximal 8 bp first uridine of the poly(U) tract drastically reduced TE . As ob- TS of the T , we were able to specifically examine the effect of hp served in other studies (5, 12), TE variance was inversely pro- TS basal hairpin sequence composition on TE in greater depth TS portional to terminator strength, with the strongest terminators (Figure 1A). Terminators were divided into four pools with 10- consistently having lower variance in TE between experimen- TS fold increment TE cut offs (<90%, >90%, >99%, >99.9%) to ex- TS tal replicates (Figure 2C). amine possible conserved sequence features of each pool. The Fifty-two terminators were measured in both forward and strongest measured terminators (TE >99%) showed a high fre- reverse orientations and differed only in their hairpin loop se- 0 0 quency of 5 -GG.. .CC-3 as the closing base pairs of the T (128 quence (either a ‘GAAA’ or ‘TTTC’; Supplementary Table S3). hp out of 187 terminators), while less strong terminators (< 99%) Neither loop sequence was consistently linked with increased or decreased TE nor differences in forward and reverse TE did not conserve this motif (115 out of 435 terminators; TS, TS 0 0 values were typically smaller than standard error between ex- Figure 3C). Conservation of a 5 -G...C-3 closing base pair for perimental replicates (Figure 2D and Supplementary Table S3). very strong terminators was also noted by Cambray and Downloaded from https://academic.oup.com/synbio/article-abstract/4/1/ysz026/5609127 by DeepDyve user on 25 March 2020 A.J. Hudson and H.-J. Wieden | 7 Figure 2. High-throughput measurement of library terminators via qTerm-Seq. (A) TEs for 622 unique library terminators (>200 reads) from FACS binning experiments were determined using qTerm-Seq and are compared with those determined for the BBa_B0012 (orange bar) and BBa_B0013 (gold bar) test terminators. Red lines indi- cate TE standard deviations from three biological replicates. See Supplementary Table S3 for terminator sequences and other data and Supplementary Figure S5 for ad- ditional details. (B) Boxplot comparing qTerm-Seq-calculated TE values for terminators in each FACS bin. Box boundaries represent the upper and lower quartiles (median indicated as a horizontal line within) and vertical lines indicate extreme upper and lower TE values. (C) Relationship of mean TE with standard deviation nor- malized to mean TE. (D) Comparison of terminator strength for terminators measured using qTerm-Seq in forward and reverse orientations. (E) qTerm-Seq measure- ment of 495 library terminators at various expression temperatures. Terminator variants are ordered from low to high mean TE based on the 14 C sample data with corresponding mean TE measured for the 21 C, 30 C and 37 C overlaid. Gray bars indicate standard deviations from three biological replicates for the 14 C sample data. See Supplementary Table S4 for terminator sequences and other data and Supplementary Figure S6 for additional details. colleagues (Arkin and Endy Laboratories, BIOFAB) (5) in a library bottleneck by introducing novel methods for creating sequence- of natural and synthetic intrinsic terminators. diverse libraries of bidirectional intrinsic terminators and developed quantitative Term-Seq to evaluate the performance of hundreds to thousands of unique intrinsic terminator parts 4. Discussion in a single experiment. In our proof-of-concept, we employed Synthetic biologists will require even greater numbers of qTerm-Seq to characterize a synthetic terminator library with characterized biological parts than are currently available and strong terminator design features in E. coli at various growth tem- more accurate design parameters to create elaborate multigene peratures. We find that most strong terminators from out syn- devices and synthetic genomes. Here, we assist with this thetic library produce stable TEs from expression temperatures Downloaded from https://academic.oup.com/synbio/article-abstract/4/1/ysz026/5609127 by DeepDyve user on 25 March 2020 8| Synthetic Biology, 2019, Vol. 4, No. 1 Figure 3. Sequence and structural features affecting intrinsic terminator strength. (A, B) Terminator hairpin G-C content and predicted free-energy of folding are exam- ined with respect to TE. Colored dots indicate the corresponding FACS bin that terminators were enriched (see Supplementary Figure S3A for gating strategy). Terminator hairpin free-energies were calculated using RNAFold from the Vienna RNA Package (32). (C) Sequence logo representations (WebLogo (34)) of terminator hairpins sequence conservation for terminators with dependent on TE. Nucleotide letter heights indicate their frequency at each position of the terminator hairpin. from 14 Cto 37 C and our collection of characterized synthetic challenges in synthetic biology, focusing on improving the reli- ability of design by using libraries of well-characterized parts terminators adds hundreds of characterized BioBrick -compatible transcription terminators that reduce read-through by up to 1000- and their development. fold. We did not explore the TE of our library terminators in other genetic and cellular contexts; however, we expect that our library Supplementary data design and qTerm-Seq methods can be utilized to identify library Supplementary Data are available at SYNBIO Online. terminators with required functional properties in vivo for a wide variety of genetic contexts and cellular conditions. Reuse of common biological parts in large genetic assem- Acknowledgements blies contributes to undesirable homologous recombination at We would like to thank members of the Wieden lab for their repeat regions and limits design complexity. Consequently, helpful suggestions and C.R. Laing for assistance with large numbers of sequence-diverse terminators are needed to V R Term-Seq Python scripts and critical review in the prepar- reduce the occurrence of gene rearrangements, insertions and ing of this manuscript. deletions in synthetic genetic circuits (12). Due to flexibility of our terminator library method, it is now possible to create dis- parate terminator libraries that have low sequence similarity, Funding enabling selection of terminators with desired properties while This work was supported by an Alberta Innovates reducing the possibility of homologous recombination. The abil- Technology Futures (AITF) Postdoctoral Fellowship (A.J.H.) ity to manipulate the length and sequence of the T stem and hp and an AITF Strategic Chair (SC60-T2) in RNA Bioengineering loop sequence as well as the other terminator features will facil- (H.J.W.). itate more detailed studies that examine specific terminator DNA/RNA features that influence transcription termination. Conflict of interest statement. None declared. This is in particular interesting as it will allow the simple design of intrinsic terminator libraries for a variety of RNA polymer- References ases, either from different host organism or in a purified and reconstituted system, creating the framework for optimal gene 1. Wang,Y.-H., Wei,K.Y. and Smolke,C.D. (2013) Synthetic biol- circuit design for uses in vitro or in specific host organisms. Our ogy: advancing the design of diverse genetic systems. Annu. construction pipeline addresses, therefore, one of the founding Rev. Chem. Biomol. Eng., 4, 69–102. Downloaded from https://academic.oup.com/synbio/article-abstract/4/1/ysz026/5609127 by DeepDyve user on 25 March 2020 A.J. Hudson and H.-J. Wieden | 9 19. Lin,M.T., Wang,C.Y., Xie,H.J., Cheung,C.H.Y., Hsieh,C.H., 2. Davis,J.H., Rubin,A.J. and Sauer,R.T. (2011) Design, construc- tion and characterization of a set of insulated bacterial Juan,H.F., Chen,B.S. and Lin,C. (2016) Novel utilization of promoters. Nucleic Acids Res., 39, 1131–1141. terminators in the design of biologically adjustable synthetic filters. ACS Synth. Biol., 5, 365–374. 3. Siegl,T., Tokovenko,B., Myronovskyi,M. and Luzhetskyy,A. (2013) Design, construction and characterisation of a syn- 20. Wachsmuth,M., Domin,G., Lorenz,R., Serfling,R., Findeiß,S., Stadler,P.F. and Mo ¨ rl,M. (2015) Design criteria for synthetic thetic promoter library for fine-tuned gene expression in acti- riboswitches acting on transcription. RNA Biol., 12, 221–231. nomycetes. Metab. Eng., 19, 98–106. 21. Dar,D., Shamir,M., Mellin,J.R., Koutero,M., Stern-Ginossar,N., 4. Lucks,J.B., Qi,L., Mutalik,V.K., Wang,D. and Arkin,A.P. (2011) Cossart,P. and Sorek,R. (2016) Term-seq reveals abundant Versatile RNA-sensing transcriptional regulators for engi- ribo-regulation of antibiotics resistance in bacteria. Science, neering genetic networks. Proc. Natl. Acad. Sci. USA, 108, 352, aad9822. 8617–8622. 22. and Sorek,R. (2018) High-resolution RNA 3-ends map- 5. Cambray,G., Guimaraes,J.C., , Lam,C., Mai,Q.A., ping of bacterial Rho-dependent transcripts. Nucleic Acids Thimmaiah,T., Carothers,J.M., Arkin,A.P. and Endy,D. (2013) Res., 46, 6797–6805. Measurement and modeling of intrinsic transcription termi- 23. Peters,J.M., Vangeloff,A.D. and Landick,R. (2011) Bacterial nators. Nucleic Acids Res., 41, 5139–5148. transcription terminators: the RNA 3 -end chronicles. J. Mol. 6. Brophy,J.A. and Voigt,C.A. (2016) Antisense transcription as a Biol., 412, 793–813. tool to tune gene expression. Mol. Syst. Biol., 12, 854–854. 24. Gusarov,I. and Nudler,E. (1999) The mechanism of intrinsic 7. Carrier,T.A. and Keasling,J.D. (1999) Library of synthetic 5’ transcription termination. Mol. Cell, 3, 495–504. secondary structures to manipulate mRNA stability in 25. Larson,M.H., Greenleaf,W.J., Landick,R. and Block,S.M. (2008) Escherichia coli. Biotechnol. Prog, 15, 58–64. Applied force reveals mechanistic and energetic details of 8. Guiziou,S., Sauveplane,V., Chang,H.J., Clerte ´ ,C., Declerck,N., transcription termination. Cell, 132, 971–982. Jules,M. and Bonnet,J. (2016) A part toolbox to tune genetic ex- 26. Komissarova,N., Becker,J., Solter,S., Kireeva,M. and pression in Bacillus subtilis. Nucleic Acids Res., 44, 7495–7508. Kashlev,M. (2002) Shortening of RNA: DNA hybrid in the elon- 9. Mutalik,V.K., Qi,L., Guimaraes,J.C., Lucks,J.B. and Arkin,A.P. gation complex of RNA polymerase is a prerequisite for tran- (2012) Rationally designed families of orthogonal RNA regula- scription termination. Mol. Cell, 10, 1151–1162. tors of translation. Nat. Chem. Biol., 8, 447–454. 27. Santangelo,T.J. and Roberts,J.W. (2004) Forward translocation 10. Green,A.A., Silver,P.A., Collins,J.J. and Yin,P. (2014) Toehold is the natural pathway of RNA release at an intrinsic termina- switches: de-novo-designed regulators of gene expression. tor. Mol. Cell, 14, 117–126. Cell, 159, 925–939. 28. Yang,J., Wang,L., Yang,F., Luo,H., Xu,L., Lu,J., Zeng,S. and 11. Salis,H.M. (2011) The ribosome binding site calculator. Zhang,Z. (2013) mBeRFP, an improved large stokes shift red Methods Enzymol., 498, 19–42. fluorescent protein. PLoS One, 8, 6–11. 12. Chen,Y.J., Liu,P., Nielsen,A.A.K., Brophy,J.A.N., Clancy,K., 29. Lee,T.S., Krupa,R.A., Zhang,F., Hajimorad,M., Holtz,W.J., Peterson,T. and Voigt,C.A. (2013) Characterization of 582 nat- Prasad,N., Lee,S.K. and Keasling,J.D. (2011) BglBrick vectors ural and synthetic terminators and quantification of their de- and datasheets: a synthetic biology platform for gene expres- sign constraints. Nat. Methods, 10, 659–664. sion. J. Biol. Eng., 5, 12. 13. Stricker,J., Cookson,S., Bennett,M.R., Mather,W.H., 30. Bernstein,J.A., Khodursky,A.B., Lin,P.-H., Lin-Chao,S. and Tsimring,L.S. and Hasty,J. (2008) A fast, robust and tunable Cohen,S.N. (2002) Global analysis of mRNA decay and abun- synthetic gene oscillator. Nature, 456, 516–519. dance in Escherichia coli at single-gene resolution using 14. Gardner,T., Cantor,C. and Collins,J. (2000) Construction of a two-color fluorescent DNA microarrays. Proc. Natl. Acad. Sci. genetic toggle switch in Escherichia coli. Nature 403, 339–342. USA, 99, 9697–9702. 15. Bonnet,J., Yin,P., Ortiz,M.E., Subsoontorn,P. and Endy,D. 31. Marcel,M. (2011) Cutadapt removes adapter sequences from (2013) Amplifying genetic logic gates. Science, 340, 599–603. high-throughput sequencing reads. EMBnet J., 17, 10–12. 16. Vecchio,D. and Del Vecchio,D. (2015) Modularity, 32. Lorenz,R., Bernhart,S.H., Siederdissen,C.H.Z., Tafer,H., context-dependence, and insulation in engineered biological Flamm,C., Stadler,P.F. and Hofacker,I.L. (2011) ViennaRNA circuits. Trends Biotechnol., 33, 111–119. Package 2.0. Algorithms Mol. Biol., 6, 26. 17. Ray-Soni,A., Bellecourt,M.J. and Landick,R. (2016) 33. Schwartz,A., Rahmouni,A.R. and Boudvillain,M. (2003) The Mechanisms of bacterial transcription termination: all good functional anatomy of an intrinsic transcription terminator. things must end. Annu. Rev. Biochem., 85, 319–347. EMBO J, 22, 3385–3394. 18. Santangelo,T. and Artsimovitch,I. (2011) Termination and 34. Crooks,G., Hon,G., Chandonia,J. and Brenner,S. (2004) antitermination: RNA polymerase runs a stop sign. Nat. Rev. WebLogo: a sequence logo generator. Genome Res., 14, Microbiol., 9, 319–329. 1188–1190. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Synthetic Biology Oxford University Press

Rapid generation of sequence-diverse terminator libraries and their parameterization using quantitative Term-Seq

Loading next page...
 
/lp/oxford-university-press/rapid-generation-of-sequence-diverse-terminator-libraries-and-their-KZWc4RaeLI

References (71)

Publisher
Oxford University Press
Copyright
© The Author(s) 2019. Published by Oxford University Press.
eISSN
2397-7000
DOI
10.1093/synbio/ysz026
Publisher site
See Article on Publisher Site

Abstract

Synthetic biology and the rational design and construction of biological devices require vast numbers of characterized biological parts, as well as reliable design tools to build increasingly complex, multigene architectures. Design principles for intrinsic terminators have been established; however, additional sequence-structure studies are needed to refine parame- ters for termination-based genetic devices. We report a rapid single-pot method to generate libraries of thousands of ran- domized bidirectional intrinsic terminators and a modified quantitative Term-Seq (qTerm-Seq) method to simultaneously identify terminator sequences and measure their termination efficiencies (TEs). Using qTerm-Seq, we characterize hundreds of additional strong terminators (TE > 90%) with some terminators reducing transcription read-through by up to 1000-fold in Escherichia coli. Our terminator library and qTerm-Seq pipeline constitute a flexible platform enabling identifica- tion of terminator parts that can achieve transcription termination not only over a desired range but also to investigate their sequence-structure features, including for specific genetic and application contexts beyond the common in vivo systems such as E. coli. Key words: transcriptional regulatory elements; genetic circuit engineering; biological techniques; gene expression regula- tion; synthetic biology. 1. Introduction facilitated predictive models to assist with synthetic cistron Forward engineering of genetic devices to carry out useful func- construction (5, 11, 12). tions is at the very core of synthetic biology. However, achieving Engineered genetic circuits consisting of one or a few genes predictable control of gene expression requires a detailed un- have been successfully applied to create a variety of genetic derstanding of relevant biological processes and the availability devices such as: oscillators (13), toggle switches (14) and logic of a sufficient number of characterized genetic parts for gene gates (15). However, de novo engineering of more complex, construction (1). Genetic parts libraries have been developed multigene synthetic devices are still limited by an incomplete mostly for bacteria (especially Escherichia coli) and they have pro- understanding of fundamental mechanisms of gene expression vided genetic engineers with repertoires to modulate gene ex- and/or accurate modeling of complex cellular environments pression at the transcriptional (2–5), post-transcriptional (6, 7) and gene interactions. Indeed, standardizing biological parts and translational levels (8–10). Experimental characterization of with predictable properties remains challenged by the finding single or collections of natural or synthetic parts has also that part performance may differ drastically in different DNA Submitted: 23 August 2019; Accepted: 3 October 2019 V The Author(s) 2019. Published by Oxford University Press. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. 1 Downloaded from https://academic.oup.com/synbio/article-abstract/4/1/ysz026/5609127 by DeepDyve user on 25 March 2020 2| Synthetic Biology, 2019, Vol. 4, No. 1 (or RNA transcript) sequence contexts (16). Development of design parameters using only two DNA oligonucleotides and high-throughput methods that characterize the performance of commonplace molecular biology enzymes and equipment. large collections of genetic parts under specific physiological We also apply a variation of the high-throughput terminator conditions and genetic contexts are, therefore, needed to refine RNA-sequencing methodology Term-Seq (21) to examine syn- genetic design principles and afford finer control over gene thetic terminator performance in vivo under a variety of condi- expression. tions. Applying these methods, we characterize hundreds of Transcription termination is a fundamental step of gene ex- additional strong intrinsic terminators in forward and reverse pression in all organisms and serves to prevent unintended orientations that can be used for engineering bacterial cistrons. transcription of flanking gene sequences, to define RNA tran- These methods provide a foundation for larger-scale studies to script 3 ends and to recycle RNA polymerase for subsequent dissect sequence-structure features influencing transcription rounds of transcription (17). In bacteria, transcriptional termi- termination and to identify synthetic riboregulators that nators are found near the end of operons where they modulate achieve a desired TE and/or respond to changing cellular condi- transcription of downstream genes and, in some cases, make tions and chassis. regulatory decisions in response to changing cellular conditions (e.g. small molecule recognition by terminator/anti-terminator 2. Materials and methods riboswitches) (18). Compared with other basic biological parts (e.g. transcription promoters and ribosome binding sites), 2.1 Terminator library construction transcription terminators have been less explored as tools to Terminator libraries were assembled in four steps consisting of: control gene expression; however, their utility as synthetic gene (i) oligo annealing, (ii) oligo extension, (iii) oligo ligation and expression regulators is exemplified by their repurposing as (iv) polymerase chain reaction (PCR; Supplementary Note S1). All adjustable high-pass and low-pass biological filters (19) and oligonucleotides for the study were synthesized by Integrated their integration into functional synthetic riboswitches (20). DNA Technologies and are detailed in Supplementary Table S1. Moreover, recent high-throughput RNAseq strategies such as Term-Seq (21) have facilitated the discovery of novel riboregula- i. Oligo annealing 0 0 tors which provide additional gene expression regulation tools One hundred picomoles of 5 half (oTerm13) and 5 -end available to synthetic biologists and provide additional mecha- monophosphorylated 3 half (oTerm14) oligonucleotides nistic insights into factor-dependent and intrinsic termination were mixed in 100 ml of ribonuclease-free ultrapure water mechanisms (21, 22). (MilliQ ) and heated in a thermocycler to 95 C for 2 min, The mechanism and sequence features affecting intrinsic followed by slowly cooling at a rate of 2 C per minute until termination efficiency (TE) have been extensively characterized the sample reached 21 C. Samples were then placed on ice. (17). In E. coli, intrinsic terminators minimally consist of a G-C- ii. Oligo extension rich hairpin (T ) that is immediately followed by a 7–9 nucleo- hp Annealed oligos were extended by T4 DNA polymerase for tide poly(U) tract (23). Intrinsic termination begins with stalling 10 min at 21 Cin 20 ml reactions containing 10 pmol of of RNA polymerase on the poly(U) tract due to the especially annealed oligos, 1 NEBuffer 2.1 (New England Biolabs, weak rU-dA hybrid and this creates a kinetic window for T for- hp NEB), 250 nmol each dNTP and 3 units of T4 DNA polymer- mation in the nascent RNA (24). Invasion of the T into the hp ase (NEB). Extended oligo complexes were then extracted RNA exit channel of RNAP then results in shearing of the rU-dA via phenol-chloroform at 21 C, ethanol precipitated and hybrid (25, 26) or hyper-translocation of RNAP (25, 27) with suspended in 15 ml of nuclease-free MilliQ ddH O. concomitant structural changes in RNAP that promote tran- iii. Oligo ligation scription elongation complex (TEC) dissociation (17). Stronger 0 0 Oligo 5 and 3 halves were ligated in 20 ml reactions contain- base pairing of the T and a perfect, extended poly(U) tract are hp ing 10 pmol annealed and extended oligos, 1 T4 DNA li- associated with increased TE (12) and the strength of the four gase buffer containing rATP (NEB) and 10 U T4 DNA ligase closing base pairs of the T and first three uridines of the hp (NEB). Ligations were incubated for 20 min at 16 C and then poly(U) tract are particularly conserved features of strong intrin- heat-inactivated at 65 C for 15 min. sic terminators (12, 23). Some intrinsic terminators also possess iv. Polymerase chain reaction a poly(A) tract upstream of the T and this feature is generally hp One microliter of ligated oligo reaction was used as tem- associated with increased TE and can enable terminators to act plate in 100 ml PCR reactions containing 1 ThermoPol in both forward and reverse directions—so-called bidirectional Buffer (Thermo Scientific), 200 mM dNTPs, 10 pmol each for- terminators (17). ward (oTerm16) and reverse (oTerm17) primers and 0.2 U Intrinsic terminator libraries are valuable tools for Phusion DNA polymerase (Thermo Scientific). PCR was car- examining RNA/DNA sequence features that affect TE and for ried out with an initial denaturation step at 98 C for 30 s, 25 identifying sequence-diverse intrinsic terminators with speci- cycles consisting of 98 C for 10 s, 55 C for 10 s and 72 C fied efficiencies for genetic circuit design (12); however, to for 20 s, and a final extension step at 72 C for 5 min. PCR the best of our knowledge, no current method can generate products were verified on 8% native polyacrylamide gels, bacterial transcriptional terminator libraries with randomized blunt-end cloned into the pJET1.2 (Thermo Scientific) using hairpin sequences without requiring many oligonucleotides manufacturer’s protocols and fifty clones were subjected to (i.e. one or two oligos per terminator) or relatively expensive Sanger sequencing to evaluate library quality and sequence massively parallel DNA synthesis strategies. Development of complexity. inexpensive and straightforward methods would facilitate high-throughput studies to further dissect RNA and DNA 2.2 Dual fluorescence reporter and library cloning sequences and/or structures affecting intrinsic termination and provide a greater collection of characterized terminators for The pBeRG reporter plasmid was created by cloning a synthetic genetic engineering. Here, we describe a novel workflow to gen- gBlock cassette (Integrated DNA Technologies) containing an erate thousands of bacterial intrinsic terminators with flexible E. coli codon-optimized, N-terminally FLAG-tagged monomeric Downloaded from https://academic.oup.com/synbio/article-abstract/4/1/ysz026/5609127 by DeepDyve user on 25 March 2020 A.J. Hudson and H.-J. Wieden | 3 blue-exciting red fluorescent protein (mBeRFP) (28) coding se- values from 0% to 50% (weak terminators), 50–87% (intermedi- quence complete with a strong 5 Shine-Dalgarno sequence and ate-weak), 87–95% (intermediate-strong) and >95% (strong; V R 3 flanking BioBrick prefix and suffix sequences at the EcoRI Supplementary Figure S3A). Sorted cell fractions were used to and NdeI restriction sites upstream of the enhanced green fluo- inoculate 5 ml of fresh LB medium, grown to mid-log phase at rescent protein (eGFP) gene of the host plasmid pBbE6a (29) (see 37 C with shaking and plated on LB plates. Fractioned Amp100 Supplementary Figure S1 for annotated plasmid sequence). clone libraries were then cultivated, aliquoted and stored as de- mBeRFP and eGFP strongly excite at 488 nm but have widely scribed above. separated emission peaks (610 and 510 nm, respectively) that minimize fluorescence resonance energy transfer and emission 2.4 Terminator RNAseq channel bleed-through. pBeRG was transformed into NEB5a E. Total RNA was isolated from terminator library E. coli cultures coli chemically competent cells (NEB) and subsequently verified as described by Bernstein et al. (30) with modifications. by Sanger sequencing. Terminator library PCR products were Terminator library FACS-sorted library clones and control ter- cloned nondirectionally into pBeRG at the two NotI sites of the minator control cultures were cultivated in LB media BioBrick prefix and suffix sequences and 100 ng of ligated plas- Amp100 overnight, sub-cultured into fresh LB media, grown until mid was transformed into NEB5a high-efficiency chemically Amp100 competent E. coli cells. Several transformation products were mid-log phase (OD ¼ 0.4) with shaking at 37 C and induced with 1 mM IPTG. Temperature-dependent termination experi- pooled and plated on LB agar with 100 mg/ml ampicillin (LB ). Transformation efficiency was estimated by plating ments were performed as described above except library clones Amp100 1% of transformation on separate LB plates. The final ter- were incubated at 14 C, 21 C, 30 Cor37 C for 15 min prior to in- Amp100 minator library (10 000 clones) was then suspended in 5 ml of duction with continued incubation at the specified temperature LB medium containing 20% glycerol, flash frozen in liquid Amp100 for 1 h (Supplementary Figure S3B). For both FACS-sorted and nitrogen and stored at 80 C. temperature-dependent experiments, 5 ml aliquots of induced cultures were added to tubes containing 1 ml of ice-cold 95% ethanol containing and 5% saturated phenol (pH 6.8; to inhibit 2.3 Flow cytometry and fluorescence-activated cell RNA degradation), rapidly mixed by vortexing, placed on ice and sorting experiments subsequently flash frozen in liquid nitrogen and stored at 80 C Flow cytometry analysis of library clones (n ¼ 180) as well as ter- until required (less than 1 week). Total RNA was extracted via minator control plasmid clones was performed by inoculating the hot phenolic extraction method followed by ethanol precipi- 5ml LB liquid medium with individual E. coli (NEB5a) colo- Amp100 R tation and elution in 50 ml of RNase-free water (MilliQ ). nies and growing at 37 C with shaking at 200 RPM for 16 h. Contaminating DNA was removed by DNase I digestion in 20 ml Cultures were then used to inoculate fresh 5 ml LB media, Amp100 reactions containing 10 mg of total RNA, 2 U DNase I (Thermo grown to mid-log phase (OD ¼ 0.4) with shaking at 37 C and Scientific) and 1 DNase I reaction buffer and incubating at induced by addition of 1 mM IPTG. Fluorescent proteins were 37 C for 1 h followed by phenol extraction and ethanol precipi- expressed overnight (16 h) to allow for maximal fluorescent pro- tation and elution in 20 ml of RNase-free water (MilliQ ). RNA tein maturation, prior to flow cytometry (see Supplementary sample quality was assayed on 1% agarose gels and quantitated Note S2 for detailed explanation). Ten to 100 ml aliquots of in- TM using a BioDrop spectrophotometer. duced overnight cultures were diluted in 1 ml of phosphate- RNA samples were prepared for Illumina MiSeq Next buffered saline and analyzed on a Becton-Dickenson V R TM Generation Sequencing using the NEBNext Multiplex Small FACSAria Fusion flow cytometer with excitation at 488 nm RNA Library Prep Set for Illumina (New England Biolabs) with and FITC and Texas Red channel filters for detection of eGFP modifications (Supplementary Figure S3C). Because the 3 end of and mBeRFP fluorescence, respectively. Forward and side scat- reporter read-through transcripts would lie several hundred tering gates were configured to select single-cell events and nucleotides downstream of the terminator cloning site, a new photomultiplier tube voltages were adjusted to 500–600 V to in- proximal 3 end was generated for read-through transcripts by crease sensitivity for weak eGFP emission and to maximize the annealing an antisense DNA oligo that binds approximately dynamic range of detection. Apparent TE was calculated from 30 nt downstream from the terminator site and digesting with measured fluorescent protein emissions using the formula: RNase H. Digestions were carried out in 20 ml reactions contain- 1 ing 10 mg of total RNA, 100 pmol antisense oligo (oTerm-Seq-R), hmBi0 hmBiTerm TE ¼ 1   100% 1 RNase H reaction buffer and 5 U RNase H (NEB) and incu- heGi0 heGiTerm bated for 1 h at 37 C followed by phenol extraction and ethanol precipitation as described above. RNase H-treated samples were Where, ‘mB’ and ‘eG’ are the mBeRFP and eGFP fluorescence V R then prepared using the NEBNext Multiplex Small RNA Library intensity, respectively. Clones containing pBeRG plasmids with Prep Set for Illumina (New England Biolabs) and barcoded as per terminator parts (‘Term’) were normalized to clones containing the manufacturer’s instructions, except the 5 linker ligation pBeRG plasmids that lack a terminator part (‘0’) to account for step was omitted. Library first-strand cDNAs were amplified us- cell-to-cell plasmid copy number variation. Flow cytometry ing the P7 reverse primer and a custom gene-specific forward experiments were performed in duplicate with TE mean and primer that binds 50 bp upstream of the terminator cloning standard deviation calculated from all cell events from each ex- site and bares the P5 oligo sequences at its 5 end (oTerm-Seq- periment. Plasmids from 50 selected clones from flow cytome- F5-P5). Barcoded terminator libraries were pooled and se- try experiments were subjected to Sanger sequencing quenced on an Illumina MiSeq Sequencer (GeneWiz) which (Supplementary Table S2). yielded approximately 40 million, 2 150 bp paired-end reads Fluorescence-activated cell sorting (FACS) experiments were performed as per flow cytometry, however, using entire E. coli with 73% of reads  Q30. Illumina adapters were removed using terminator library samples. E. coli terminator library clones were Cutadapt (31) with a threshold of 90% identity to adapter sorted into four bins with gates corresponding to apparent TE sequences. Only reads that passed quality control (Q30) and Downloaded from https://academic.oup.com/synbio/article-abstract/4/1/ysz026/5609127 by DeepDyve user on 25 March 2020 4| Synthetic Biology, 2019, Vol. 4, No. 1 bared a complete T sequence were considered for analysis even relatively small combinatorial libraries (100 sequences) hp (Supplementary Tables S3 and S4). both expensive and time-consuming. We, therefore, sought to Unique terminator sequence identifiers were assigned with develop methods to create libraries consisting of thousands of the designation ‘T’ and numbered in descending order of their unique intrinsic terminator variants while maintaining control sequence read abundance. Terminator counts were normalized over features such as the length and composition of the T , hp for each sample replicate (in transcripts per million, TPM) and poly(U) and poly(A) tracts and adjacent sequences (e.g. for TM custom Python scripts (see Section 2.6) were used to tabulate cloning or part assembly purposes). Although we attempted transcript sequencing read lengths which served to determine several strategies for terminator library synthesis (many other whether transcripts were terminated or read-through. strategies are likely possible), we found a strategy that uses two Transcripts were deemed as ‘terminated’ if their length corre- semi-randomized DNA oligonucleotides that provides a great sponded to þ1to þ8 nt downstream of the last nucleotide of the compromise in terms of economy (could be performed rapidly T while transcripts with length greater than this (beyond the hp and inexpensively, using readily available lab materials) and poly(U) tract) were counted as ‘read-through.’ The top 2000 ter- 0 flexibility over terminator library design. In this method, two 5 minators with the greatest number of total reads were then 0 and 3 half oligos are designed using a computer to contain assessed for TE by counting the number of terminated tran- desired features (see Supplementary Note S1 for design consid- scripts divided by total reads for each terminator variant. Only erations) and generated by conventional DNA oligo synthesis. terminator variants that had 200 or more reads in either FACS Oligos are then assembled into terminator libraries in four steps binning or differential temperature experiments are displayed (see Section 2.1). (see Supplementary Tables S3 and S4 for all terminator sequen- Because strong intrinsic terminators (>90% TE) with diverse ces). T minimal free-energy predications and predicted sec- hp sequences are particularly desirable for insulating synthetic cis- ondary structures were calculated for each terminator variant trons, we applied our terminator library methodology to create using the RNAfold program from the Vienna RNA Package (32) very strong intrinsic terminators: a 14 bp, perfectly paired hair- with folding parameters set to 37 C and -p -d2 –noLP modifiers. pin with a basal 8 bp randomized portion that is closed by two Terminators identified in FACS-sorted terminator RNAseq strong (G-C or C-G) pairs, a GAAA tetraloop (or UUUC in reverse), (Term-Seq) libraries were normalized to TPM values, based on a downstream 8 nt poly(U) tract and an upstream 8 nt poly(A) the total number of reads in each barcoded library. Terminators tract (12)(Figure 1A). We placed BioBrick RFC 10 standard prefix were examined for enrichment in FACS bins as per the bin with and suffix sequences in the upstream and downstream con- the greatest mean TPM values (most abundant bin), as well as stant regions to allow the library parts to be easily assembled by use of the two-tailed Student’s t-tests, which compared TPM into genetic circuits via BioBrick assembly (http://parts.igem. counts from each FACS bin from triplicate biological replicates org/Assembly: Standard_assembly). Our library design has a (df ¼ 1; Supplementary Table S3). If t-test P-values were <0.05 theoretical maximum diversity of 16 384 sequences and once for any one bin relative to all other bins, the terminator was designed, terminator library assembly could be performed in a considered significantly enriched in that bin. For differential single 8-h day. temperature experiments, TE values for terminators were calcu- lated for each temperature treatment from triplicate experi- 3.2 Synthetic terminator library characterization via ments. Individual terminator TEs were then compared for each temperature treatment using two-tailed Student’s t-tests for flow cytometry significant differences in terminator read-through efficiency We cloned our terminator library nondirectionally into a cus- (P< 0.05, df ¼ 1). V R tom bi-cistronic terminator testing device (pBeRG) at a BioBrick RFC 10 multicloning site (Figure 1B) and examined TEs of termi- 2.6 Data availability nators from our library (n ¼10 000 clones) using a fluorescence-based interference strategy and flow cytometry. All custom scripts used for Term-Seq have been made available Flow cytometry revealed a bimodal distribution for all termina- at GitHub (https://github.com/andyhudson42/TermSeq.git). Raw tor library clones and two distinct populations with median sequence data for Term-Seq were deposited and available from fluorescence interference TEs (designated TE ) of 13.6% and the NCBI Sequence Read Archive (SRA) under BioProject acces- IF 96.5%, relative to no terminator (Figure 1C and Supplementary sion: PRJNA503821. All other relevant data are available upon re- Figure S2). For comparison, we benchmarked terminators from quest from the authors. our library against two previously characterized intrinsic termi- nators: BBa_B0012 (medium-efficiency terminator) and 2.7 Materials availability BBa_B0013 (high-efficiency terminator) from the Registry of All DNA constructs will be provided upon request following the Standardized Biological Parts (http://parts.igem.org). In pBeRG, completion of a Materials Transfer Agreement and any other flow cytometry experiments predicted 90.9% and 96.5% TE for IF documentation that may be required. the BBa_B0012 and BBa_B0013 terminators, respectively (Figure 1C and Supplementary Table S2). TEs for 180 library clones were then measured individually using flow cytometry. 3. Results More than half of the measured library terminators displayed 3.1 Rapid synthesis of intrinsic terminator libraries apparent TE of 90% or greater and many terminators showed IF even higher efficiency than the strong BBa_B0013 terminator Terminator libraries have been valuable tools for examining (Figure 1C). Fifty clones were Sanger sequenced, which identi- RNA/DNA sequence–function relationships and for establishing fied only unique terminator variants and confirmed successful design principles for synthetic terminators with desirable TEs (5, 12). However, prior terminator library construction has re- terminator library cloning with sufficient diversity to warrant quired assembly of individually synthesized terminator parts, further examination using higher throughput methods making the ability to probe the functional sequence space of (Supplementary Table S2). Downloaded from https://academic.oup.com/synbio/article-abstract/4/1/ysz026/5609127 by DeepDyve user on 25 March 2020 A.J. Hudson and H.-J. Wieden | 5 Figure 1. Intrinsic terminator library synthesis and measurement by flow cytometry. (A) Terminator library design used in this study. The corresponding RNA se- V R quence for the terminator library is indicated with boxes representing BioBrick RFC 10 prefix and suffix sequences. The randomized portion of the T is in red text, hp where ‘S’ indicates a ‘C’ or ‘G’ nucleotide and ‘N’ is any nucleotide. The forward T loop sequence is shown with the reverse sequence shown above. (B). Schematic of hp the bi-cistronic pBeRG terminator testing device is shown using SBOL visual format and indicates the terminator cloning site (BioBrick multicloning site, MCS). See Supplementary Figure S1 for pBeRG vector sequence. (C) Flow cytometry measurements for 180 E. coli NEB5a terminator library clones were performed in duplicate and TE values were calculated (see Methods section). Standard deviations are shown for clones as red lines and the pBeRG (without terminator, first bar on left), BBa_B0012 terminator and BBa_B0013 terminator clones are indicated as blue, orange and gold bars. See Supplementary Table S2 for terminator sequences. identify terminator sequences and evaluate transcription TE 3.3 High-throughput measurement of terminators using of terminators from our synthetic library (see Section 2 and qTerm-Seq Supplementary Figure S3). Flow cytometry-based methods have been routinely applied to To determine whether TEs from flow cytometry-based (TE ) IF examine TEs of intrinsic terminators in vivo (5, 12); however, were comparable to those obtained by qTerm-Seq (TE ), we TS they are less amenable to high-throughput examination of used FACS to separate terminator library clones into four frac- terminators (e.g. hundreds of terminators) due to the need to in- tions prior to qTerm-Seq with FACS gates corresponding to: 0% dividually clone and measure terminator parts. Transcriptome- to 50% TE (Weak terminators), 50–87% TE (Intermediate-Weak), based methods are highly scalable and an attractive alternative 87–95% TE (Intermediate-Strong) and >95% TE (Strong). to flow cytometry for higher throughput studies examining qTerm-Seq identified over 10 000 unique library sequences; transcription termination, due to the ability to examine thou- however, some of these are expected to represent cloning or sands of transcripts in a single experiment. Term-Seq is a re- sequencing artifacts (e.g. only single-read representation). cently described RNAseq method that maps RNA transcript 3 We examined the distribution of 3 ends for all qTerm-seq reads ends to identify transcription termination regulators in E. coli relative to the pBeRG terminator reporter construct to identify (22) and other bacteria (21). However, while Term-Seq identifies putative termination sites, it has not been applied to quantitate putative transcription termination sites (Supplementary Figure S4A). Transcript 3 end positions showed similar distributions TE. Therefore, we developed a quantitative Term-Seq method and bioinformatic strategy (qTerm-Seq) that can simultaneously for all four FACS bins, with exception of weak terminators, Downloaded from https://academic.oup.com/synbio/article-abstract/4/1/ysz026/5609127 by DeepDyve user on 25 March 2020 6| Synthetic Biology, 2019, Vol. 4, No. 1 which suggests that terminators separated into the remaining However, stronger terminators generally showed less variation bins had similar transcriptional and degradation profiles in vivo between their forward and reverse orientations than weaker (Supplementary Figure S4A). A large proportion of the qTerm- terminators and several terminators maintained greater 0 0 0 Seq reads were 3 truncated within the 5 and 3 arms of the T than 99% TE in both directions (Figure 2D and Supplementary hp TS portion of terminator library sequences (Supplementary Figure Table S3). S4A). Although this collection may indeed represent bona fide transcription termination sites or stable reporter mRNA degra- 3.4 Expression temperature does not strongly affect TE TS dation products, a complete T sequence could not be deter- hp of library terminators mined for these reads and they were removed from further Terminator hairpin folding dynamics during transcription can analysis. A peak þ1to þ3 nt downstream of the T was also ev- hp be affected by expression temperature and plausibly this may ident and due to their close proximity to the expected termina- tion site for intrinsic terminators (þ6to þ8 nt downstream of confer changes in terminator efficiency. Therefore, we also expressed our terminator library at 14 C, 21 C, 30 C and 37 Cin T )(23) were deemed to have arisen from transcription termi- hp nation events (Supplementary Figure S4A). qTerm-Seq reads E. coli prior to qTerm-Seq to examine possible temperature- whose 3 ends extended beyond þ8 nt downstream of the T dependent termination effects in our library (Supplementary hp were classified as read-through events in which RNA polymer- Figure S3B). After removing low-abundance terminators (<200 ase did not terminate at the candidate terminator. reads), we identified 495 terminator variants, which did not After removing truncated reads with an incomplete T and hp overlap significantly with those from the FACS-enriched termi- terminator sequences with fewer than 200 reads, 622 unique nator pool (Supplementary Table S4). Although measured termi- full-length terminator variants remained for analysis nator TE spanned two orders of magnitude (0.62–99.96%), TS (Supplementary Table S3). The overall distribution of terminator most of the terminators did not display significant differences library TE from the 622 FACS-enriched terminators mimics TS in TE at expression temperatures from 14 Cto 37 C and very the distribution obtained from individual flow cytometry strong terminators showed the least temperature-dependent experiments (cf. Figures 1C and 2A). Terminator library TE TS variance in TE (Figure 2E and Supplementary Figure S6). TS predicted by qTerm-Seq also corresponded closely with enrich- Patterns of qTerm-Seq 3 end positions did not show obvious ment in the four FACS bins (Figure 2B), indicating that qTerm- deviations for different expression temperatures, except at an Seq and flow cytometry measurements generate comparable R upstream site within the BioBrick prefix sequence which oc- apparent TEs. curred more frequently at 14 C and 21 C(Supplementary Figure TE measurement for the BBa_B0012 and BBa_B0013 termi- TS S4B). Several terminators were found that had significantly dif- nators was 92.8% (SD ¼ 1.9%) and 99.1% (SD ¼ 0.21%), respec- ferent TE from 14 Cto 37 C(Figure 2E and Supplementary TS tively and 2–3% higher than TE values predicted by flow IF Table S4) and most of these exhibited a higher TE at lower TS cytometry experiments for the two terminators. Unfortunately, temperature with the largest difference being a 78% decrease at none of the 50 sequences obtained from Sanger sequencing 14 C as compared with 37 C (T528). were present in the 622 terminators measured in qTerm-Seq and these could not be directly compared using the two 3.5 Terminator features correlating to TE methods. From the collection of 622 FACS-enriched terminators, 423 Using the collection of library terminators measured by qTerm- terminators had a TE greater than 90% and 180 terminators TS Seq, we looked for terminator features that correlated to TE . TS were greater than 99% (Figure 2A and Supplementary Figure S5 T predicted free-energies somewhat correlated with TE and hp TS and Table S3). The strongest 26 terminators had TE greater TS there was an apparent division in T free-energy with termina- hp than 99.9%, corresponding to more than a 1000-fold difference tors with a DG of less than 20 kJ/mol having TEs greater than in read-through, relative to the lowest measured terminators 90% and terminators with greater DG typically having lower TE from the library. As expected for our library design, most strong (Figure 3A and B and Supplementary Tables S3 and S4). terminators featured a perfectly base-paired T . Weak termi- hp Terminator hairpin G/C content did not show any obvious rela- nators (< 50% TE ) were found to be mostly terminator assem- TS tionship to TE , although surprisingly, the strongest termina- TS bly or cloning artifacts that could not form a strong hairpin tors in the library generally had less G/C-rich hairpins structure, and this is consistent with the importance of T base hp (Figure 3A). It is possible that PCR amplification of the termina- pairing integrity for efficient intrinsic termination (24). tor library and our qTerm-Seq library preparation methods fa- Consistent with prior studies (12, 33), we observed that nucleo- vored more A/T-rich terminators and that even stronger (more tide substitutions that disrupt base pairing within the middle of G/C rich) terminators may exist in our library but escaped detec- the T reduced TE to a lesser extent than the two closing T hp TS hp tion using qTerm-Seq. base pairs and nucleotide insertions between the T and the hp Because our terminator library only varies the proximal 8 bp first uridine of the poly(U) tract drastically reduced TE . As ob- TS of the T , we were able to specifically examine the effect of hp served in other studies (5, 12), TE variance was inversely pro- TS basal hairpin sequence composition on TE in greater depth TS portional to terminator strength, with the strongest terminators (Figure 1A). Terminators were divided into four pools with 10- consistently having lower variance in TE between experimen- TS fold increment TE cut offs (<90%, >90%, >99%, >99.9%) to ex- TS tal replicates (Figure 2C). amine possible conserved sequence features of each pool. The Fifty-two terminators were measured in both forward and strongest measured terminators (TE >99%) showed a high fre- reverse orientations and differed only in their hairpin loop se- 0 0 quency of 5 -GG.. .CC-3 as the closing base pairs of the T (128 quence (either a ‘GAAA’ or ‘TTTC’; Supplementary Table S3). hp out of 187 terminators), while less strong terminators (< 99%) Neither loop sequence was consistently linked with increased or decreased TE nor differences in forward and reverse TE did not conserve this motif (115 out of 435 terminators; TS, TS 0 0 values were typically smaller than standard error between ex- Figure 3C). Conservation of a 5 -G...C-3 closing base pair for perimental replicates (Figure 2D and Supplementary Table S3). very strong terminators was also noted by Cambray and Downloaded from https://academic.oup.com/synbio/article-abstract/4/1/ysz026/5609127 by DeepDyve user on 25 March 2020 A.J. Hudson and H.-J. Wieden | 7 Figure 2. High-throughput measurement of library terminators via qTerm-Seq. (A) TEs for 622 unique library terminators (>200 reads) from FACS binning experiments were determined using qTerm-Seq and are compared with those determined for the BBa_B0012 (orange bar) and BBa_B0013 (gold bar) test terminators. Red lines indi- cate TE standard deviations from three biological replicates. See Supplementary Table S3 for terminator sequences and other data and Supplementary Figure S5 for ad- ditional details. (B) Boxplot comparing qTerm-Seq-calculated TE values for terminators in each FACS bin. Box boundaries represent the upper and lower quartiles (median indicated as a horizontal line within) and vertical lines indicate extreme upper and lower TE values. (C) Relationship of mean TE with standard deviation nor- malized to mean TE. (D) Comparison of terminator strength for terminators measured using qTerm-Seq in forward and reverse orientations. (E) qTerm-Seq measure- ment of 495 library terminators at various expression temperatures. Terminator variants are ordered from low to high mean TE based on the 14 C sample data with corresponding mean TE measured for the 21 C, 30 C and 37 C overlaid. Gray bars indicate standard deviations from three biological replicates for the 14 C sample data. See Supplementary Table S4 for terminator sequences and other data and Supplementary Figure S6 for additional details. colleagues (Arkin and Endy Laboratories, BIOFAB) (5) in a library bottleneck by introducing novel methods for creating sequence- of natural and synthetic intrinsic terminators. diverse libraries of bidirectional intrinsic terminators and developed quantitative Term-Seq to evaluate the performance of hundreds to thousands of unique intrinsic terminator parts 4. Discussion in a single experiment. In our proof-of-concept, we employed Synthetic biologists will require even greater numbers of qTerm-Seq to characterize a synthetic terminator library with characterized biological parts than are currently available and strong terminator design features in E. coli at various growth tem- more accurate design parameters to create elaborate multigene peratures. We find that most strong terminators from out syn- devices and synthetic genomes. Here, we assist with this thetic library produce stable TEs from expression temperatures Downloaded from https://academic.oup.com/synbio/article-abstract/4/1/ysz026/5609127 by DeepDyve user on 25 March 2020 8| Synthetic Biology, 2019, Vol. 4, No. 1 Figure 3. Sequence and structural features affecting intrinsic terminator strength. (A, B) Terminator hairpin G-C content and predicted free-energy of folding are exam- ined with respect to TE. Colored dots indicate the corresponding FACS bin that terminators were enriched (see Supplementary Figure S3A for gating strategy). Terminator hairpin free-energies were calculated using RNAFold from the Vienna RNA Package (32). (C) Sequence logo representations (WebLogo (34)) of terminator hairpins sequence conservation for terminators with dependent on TE. Nucleotide letter heights indicate their frequency at each position of the terminator hairpin. from 14 Cto 37 C and our collection of characterized synthetic challenges in synthetic biology, focusing on improving the reli- ability of design by using libraries of well-characterized parts terminators adds hundreds of characterized BioBrick -compatible transcription terminators that reduce read-through by up to 1000- and their development. fold. We did not explore the TE of our library terminators in other genetic and cellular contexts; however, we expect that our library Supplementary data design and qTerm-Seq methods can be utilized to identify library Supplementary Data are available at SYNBIO Online. terminators with required functional properties in vivo for a wide variety of genetic contexts and cellular conditions. Reuse of common biological parts in large genetic assem- Acknowledgements blies contributes to undesirable homologous recombination at We would like to thank members of the Wieden lab for their repeat regions and limits design complexity. Consequently, helpful suggestions and C.R. Laing for assistance with large numbers of sequence-diverse terminators are needed to V R Term-Seq Python scripts and critical review in the prepar- reduce the occurrence of gene rearrangements, insertions and ing of this manuscript. deletions in synthetic genetic circuits (12). Due to flexibility of our terminator library method, it is now possible to create dis- parate terminator libraries that have low sequence similarity, Funding enabling selection of terminators with desired properties while This work was supported by an Alberta Innovates reducing the possibility of homologous recombination. The abil- Technology Futures (AITF) Postdoctoral Fellowship (A.J.H.) ity to manipulate the length and sequence of the T stem and hp and an AITF Strategic Chair (SC60-T2) in RNA Bioengineering loop sequence as well as the other terminator features will facil- (H.J.W.). itate more detailed studies that examine specific terminator DNA/RNA features that influence transcription termination. Conflict of interest statement. None declared. This is in particular interesting as it will allow the simple design of intrinsic terminator libraries for a variety of RNA polymer- References ases, either from different host organism or in a purified and reconstituted system, creating the framework for optimal gene 1. Wang,Y.-H., Wei,K.Y. and Smolke,C.D. (2013) Synthetic biol- circuit design for uses in vitro or in specific host organisms. Our ogy: advancing the design of diverse genetic systems. Annu. construction pipeline addresses, therefore, one of the founding Rev. Chem. Biomol. Eng., 4, 69–102. Downloaded from https://academic.oup.com/synbio/article-abstract/4/1/ysz026/5609127 by DeepDyve user on 25 March 2020 A.J. Hudson and H.-J. Wieden | 9 19. Lin,M.T., Wang,C.Y., Xie,H.J., Cheung,C.H.Y., Hsieh,C.H., 2. Davis,J.H., Rubin,A.J. and Sauer,R.T. (2011) Design, construc- tion and characterization of a set of insulated bacterial Juan,H.F., Chen,B.S. and Lin,C. (2016) Novel utilization of promoters. Nucleic Acids Res., 39, 1131–1141. terminators in the design of biologically adjustable synthetic filters. ACS Synth. Biol., 5, 365–374. 3. Siegl,T., Tokovenko,B., Myronovskyi,M. and Luzhetskyy,A. (2013) Design, construction and characterisation of a syn- 20. Wachsmuth,M., Domin,G., Lorenz,R., Serfling,R., Findeiß,S., Stadler,P.F. and Mo ¨ rl,M. (2015) Design criteria for synthetic thetic promoter library for fine-tuned gene expression in acti- riboswitches acting on transcription. RNA Biol., 12, 221–231. nomycetes. Metab. Eng., 19, 98–106. 21. Dar,D., Shamir,M., Mellin,J.R., Koutero,M., Stern-Ginossar,N., 4. Lucks,J.B., Qi,L., Mutalik,V.K., Wang,D. and Arkin,A.P. (2011) Cossart,P. and Sorek,R. (2016) Term-seq reveals abundant Versatile RNA-sensing transcriptional regulators for engi- ribo-regulation of antibiotics resistance in bacteria. Science, neering genetic networks. Proc. Natl. Acad. Sci. USA, 108, 352, aad9822. 8617–8622. 22. and Sorek,R. (2018) High-resolution RNA 3-ends map- 5. Cambray,G., Guimaraes,J.C., , Lam,C., Mai,Q.A., ping of bacterial Rho-dependent transcripts. Nucleic Acids Thimmaiah,T., Carothers,J.M., Arkin,A.P. and Endy,D. (2013) Res., 46, 6797–6805. Measurement and modeling of intrinsic transcription termi- 23. Peters,J.M., Vangeloff,A.D. and Landick,R. (2011) Bacterial nators. Nucleic Acids Res., 41, 5139–5148. transcription terminators: the RNA 3 -end chronicles. J. Mol. 6. Brophy,J.A. and Voigt,C.A. (2016) Antisense transcription as a Biol., 412, 793–813. tool to tune gene expression. Mol. Syst. Biol., 12, 854–854. 24. Gusarov,I. and Nudler,E. (1999) The mechanism of intrinsic 7. Carrier,T.A. and Keasling,J.D. (1999) Library of synthetic 5’ transcription termination. Mol. Cell, 3, 495–504. secondary structures to manipulate mRNA stability in 25. Larson,M.H., Greenleaf,W.J., Landick,R. and Block,S.M. (2008) Escherichia coli. Biotechnol. Prog, 15, 58–64. Applied force reveals mechanistic and energetic details of 8. Guiziou,S., Sauveplane,V., Chang,H.J., Clerte ´ ,C., Declerck,N., transcription termination. Cell, 132, 971–982. Jules,M. and Bonnet,J. (2016) A part toolbox to tune genetic ex- 26. Komissarova,N., Becker,J., Solter,S., Kireeva,M. and pression in Bacillus subtilis. Nucleic Acids Res., 44, 7495–7508. Kashlev,M. (2002) Shortening of RNA: DNA hybrid in the elon- 9. Mutalik,V.K., Qi,L., Guimaraes,J.C., Lucks,J.B. and Arkin,A.P. gation complex of RNA polymerase is a prerequisite for tran- (2012) Rationally designed families of orthogonal RNA regula- scription termination. Mol. Cell, 10, 1151–1162. tors of translation. Nat. Chem. Biol., 8, 447–454. 27. Santangelo,T.J. and Roberts,J.W. (2004) Forward translocation 10. Green,A.A., Silver,P.A., Collins,J.J. and Yin,P. (2014) Toehold is the natural pathway of RNA release at an intrinsic termina- switches: de-novo-designed regulators of gene expression. tor. Mol. Cell, 14, 117–126. Cell, 159, 925–939. 28. Yang,J., Wang,L., Yang,F., Luo,H., Xu,L., Lu,J., Zeng,S. and 11. Salis,H.M. (2011) The ribosome binding site calculator. Zhang,Z. (2013) mBeRFP, an improved large stokes shift red Methods Enzymol., 498, 19–42. fluorescent protein. PLoS One, 8, 6–11. 12. Chen,Y.J., Liu,P., Nielsen,A.A.K., Brophy,J.A.N., Clancy,K., 29. Lee,T.S., Krupa,R.A., Zhang,F., Hajimorad,M., Holtz,W.J., Peterson,T. and Voigt,C.A. (2013) Characterization of 582 nat- Prasad,N., Lee,S.K. and Keasling,J.D. (2011) BglBrick vectors ural and synthetic terminators and quantification of their de- and datasheets: a synthetic biology platform for gene expres- sign constraints. Nat. Methods, 10, 659–664. sion. J. Biol. Eng., 5, 12. 13. Stricker,J., Cookson,S., Bennett,M.R., Mather,W.H., 30. Bernstein,J.A., Khodursky,A.B., Lin,P.-H., Lin-Chao,S. and Tsimring,L.S. and Hasty,J. (2008) A fast, robust and tunable Cohen,S.N. (2002) Global analysis of mRNA decay and abun- synthetic gene oscillator. Nature, 456, 516–519. dance in Escherichia coli at single-gene resolution using 14. Gardner,T., Cantor,C. and Collins,J. (2000) Construction of a two-color fluorescent DNA microarrays. Proc. Natl. Acad. Sci. genetic toggle switch in Escherichia coli. Nature 403, 339–342. USA, 99, 9697–9702. 15. Bonnet,J., Yin,P., Ortiz,M.E., Subsoontorn,P. and Endy,D. 31. Marcel,M. (2011) Cutadapt removes adapter sequences from (2013) Amplifying genetic logic gates. Science, 340, 599–603. high-throughput sequencing reads. EMBnet J., 17, 10–12. 16. Vecchio,D. and Del Vecchio,D. (2015) Modularity, 32. Lorenz,R., Bernhart,S.H., Siederdissen,C.H.Z., Tafer,H., context-dependence, and insulation in engineered biological Flamm,C., Stadler,P.F. and Hofacker,I.L. (2011) ViennaRNA circuits. Trends Biotechnol., 33, 111–119. Package 2.0. Algorithms Mol. Biol., 6, 26. 17. Ray-Soni,A., Bellecourt,M.J. and Landick,R. (2016) 33. Schwartz,A., Rahmouni,A.R. and Boudvillain,M. (2003) The Mechanisms of bacterial transcription termination: all good functional anatomy of an intrinsic transcription terminator. things must end. Annu. Rev. Biochem., 85, 319–347. EMBO J, 22, 3385–3394. 18. Santangelo,T. and Artsimovitch,I. (2011) Termination and 34. Crooks,G., Hon,G., Chandonia,J. and Brenner,S. (2004) antitermination: RNA polymerase runs a stop sign. Nat. Rev. WebLogo: a sequence logo generator. Genome Res., 14, Microbiol., 9, 319–329. 1188–1190.

Journal

Synthetic BiologyOxford University Press

Published: Jan 1, 2019

There are no references for this article.