Characterizing transcriptional interference between converging genes

Subscriber access provided by UB der LMU Muenchen

Letter

Characterizing transcriptional interference between converging genes in bacteria Stefan Hoffmann, Nan Hao, Keith E Shearwin, and Katja M Arndt ACS Synth. Biol., Just Accepted Manuscript • DOI: 10.1021/acssynbio.8b00477 • Publication Date (Web): 05 Feb 2019 Downloaded from http://pubs.acs.org on February 5, 2019

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 13 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Synthetic Biology

Characterizing transcriptional interference between converging genes in bacteria Stefan A. Hoffmann1, Nan Hao2,3, Keith E. Shearwin2,* and Katja M. Arndt1,* Molecular Biotechnology, Institute for Biochemistry and Biology, University of Potsdam, KarlLiebknecht-Str. 24-25, 14476 Potsdam-Golm, Germany 1

Discipline of Biochemistry, Department of Molecular and Biomedical Science, School of Biological Sciences, The University of Adelaide, Adelaide, South Australia 5005, Australia 2

3

CSIRO Synthetic Biology Future Science Platform

*To whom correspondence should be addressed. Tel: +61 8 83135361; Email: [email protected]. Tel: +49 331 977 5261; Email: [email protected]

Abstract Antisense transcription is common in naturally occurring genomes and is increasingly being used in synthetic genetic circuitry as a tool for gene expression control. Mutual influence on the expression of convergent genes can be mediated by antisense RNA effects and by transcriptional interference (TI). We aimed to quantitatively characterize long-range TI between convergent genes with untranslated intergenic spacers of increasing length. Controlling for antisense RNA-mediated effects, which contributed about half of the observed total expression inhibition, the TI effect was modelled. To achieve model convergence, RNA polymerase processivity and collision resistance were assumed to be modulated by ribosome trailing. The spontaneous transcription termination rate in regions of untranslated DNA was experimentally ACS Paragon Plus Environment

ACS Synthetic Biology 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

determined. Our modeling suggests that an elongating RNA polymerase with a trailing ribosome is about 13 times more likely to resume transcription than an opposing RNA polymerase without a trailing ribosome, upon head-on collision of the two.

Keywords gene regulation, antisense transcription, transcriptional interference, mathematical modelling, Escherichia coli Antisense transcription is found in all three domains of life.1 Whole transcriptome studies have revealed that in some bacterial species, antisense transcripts are found for the majority of annotated genes.1 In E. coli, more than 1000 antisense transcripts have been identified.2 While a proportion of antisense transcripts may just be transcriptional noise arising from cryptic promoters, antisense transcription also provides a recognized means of gene regulation.3 Notable examples have been described for both eukaryotes, like IME4 in S. cerevisiae4 and the Airn/Igf2r locus in mouse.5 as well as for bacteria, e.g. the prgX/prgQ operon in E. faecalis.6 There are two general mechanisms of mutual influence of two convergent, overlapping transcriptional units: antisense RNA (asRNA), which functions both in cis and in trans, and cis-acting transcriptional interference (TI). In the prgX/prgQ example, both mechanisms apparently play a role. Antisense RNA can modulate the expression of the sense gene in different ways, most notably (i) by promoting decay of the sense mRNA, (ii) by inhibition of its translation due to blocking of the ribosome binding site and (iii) by attenuation of sense transcription via direct binding of asRNA to the nascent sense RNA to terminate mRNA transcription.7,8 In contrast, transcriptional interference is the direct, typically mutually inhibitory interaction of RNA polymerase (RNAP) complexes. In the case of convergent transcription, the head-on collision of two elongating RNAPs, the dislodgement of an initiating RNAP complex and the prevention of loading of an initiation complex to the promoter by an elongating RNAP can occur. Those three mechanisms were dubbed collisional, sitting duck and occlusion transcriptional interference, respectively.9 Apart from naturally occurring examples, antisense transcription is increasingly being used in artificial genetic circuitry. Recently, it has been suggested as a way to achieve tunable gene expression10,11. In addition to having a further option for tuning expression strength, employing antisense transcription for this purpose might also reduce variability of expression: genes regulated by antisense transcription in yeast showed lower-than-expected expression noise.12 We recently demonstrated that substantial mutuallyinhibitory effects on gene expression can be obtained in bacteria when two full-sized genes face each other.13 This effect allows an inverse coupling of the expression of two genes without additional regulatory elements, making it an appealing approach for synthetic genetic circuitry. We leveraged antisense transcription to create a dual selection system solely relying on positive selection markers.13 Examples of further potential uses are sensitive in vivo DNA binding assays, toggle switches and oscillators. In order to reduce trial-and-error-based circuit tuning, having a predictive model of the expected inhibitory effect will help to guide the design of respective artificial circuitry. Further, being able to gauge potential inhibitory effects from antisense transcription will also provide a better understanding of naturally occurring convergent gene arrangements. Here, we aimed to combine quantitative reporter protein expression data with mathematical modelling to further understand the key determinants of long-range TI between two convergent genes. ACS Paragon Plus Environment

Page 2 of 13

Page 3 of 13 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


To separate antisense RNA mediated effects from cis-acting TI, the antisense RNA was produced in trans on a second locus on the same plasmid. The remaining inhibitory effect, attributable to direct TI, was modelled by extending a recently published stochastic model.14,15 Initial fits revealed the necessity to consider the cooperation between translation and transcription in bacteria in the model. In bacteria, RNA polymerases elongating in translated regions are being trailed by a translating ribosome. This cotranscriptional translation protects the RNA polymerase from spontaneous NusG/Rho dependent transcription termination prevalent in untranslated regions,16,17 and from backtracking, 18evident by increased read-through upon encountering transcriptional roadblocks.18 Spontaneous termination in untranslated regions was experimentally determined. A free parameter was added to allow uneven read-through probabilities in the case of a head-on collision of two elongating RNA polymerases, one of which has a trailing ribosome. Fitting this parameter suggested that in a collision event, the RNA polymerase trailed by a ribosome is about 13 times more likely to resume transcription than the untrailed RNA polymerase.

Results and Discussion Measuring transcriptional interference For quantitative assessment of the distance dependence of the total inhibitory effects arising from converging transcription, a convergent arrangement of a BFP and an mCherry CDS was used as the basic construct (Figure 1). To achieve context independence of transcriptional activity, functional and scrambled promoters used in this study were flanked by transcriptional insulators,19 optimized for lack of cryptic promoter activity (Table S1, Figure S1). BFP was either preceded by Pstrong or a scrambled sequence, whereas mCherry was likewise either driven by Pweak or a scrambled promoter, with Pstrong being about 11.3 times stronger than Pweak (Figure S2). The resulting three combinations with either one or two functional promoters in the construct were cloned with three different distances between the BFP and mCherry CDS; with either a 21 bp landing pad for spacer insertion (0 bp spacer), an additional 500 bp spacer, or two consecutive copies of the 500 bp spacers. Spacer sequences were the first 500 bp (excluding the start codon) of the bacterial antibiotic resistance gene cat to ensure transcriptional processivity by bacterial RNAP. The sequence was tested to exhibit only minimal endogenous transcriptional activity (Figure S3). For both BFP and mCherry, expression without an opposing promoter was reduced with increasing spacer length, presumably due to reduced mRNA stability of the longer transcript. But more importantly, expression of mCherry driven by Pweak was also markedly reduced when it was facing Pstrong driving BFP. The strength of this effect tapered off with increasing spacer length (Figure 1). We next aimed to distinguish between the contribution of direct transcriptional interference due to inhibitory interactions of opposing RNA polymerase complexes, and the contribution of antisense RNA effects. In previous studies on convergent TI, separation of asRNA mediated effects was achieved by producing the asRNA in trans on a separate plasmid.10,11 We took a similar approach, but sought to achieve a more direct control over the number of produced asRNA transcripts. To avoid the confounding effects of plasmid copy numbers, we constructed single plasmid designs in which asRNA effects without TI could be observed. Therefore, the reporter setup of convergent BFP and mCherry was cloned twice into the ACS Paragon Plus Environment


respective plasmids; one reporter had a functional promoter (Pstrong) driving BFP, but no promoter for mCherry, while the other had a promoter only for mCherry (Pweak), but none for BFP. The respective constructs for each of the three inter-promoter distances were made. Both possible relative orientations of the two reporter layouts were considered (Figure 2). The two orientations did not produce significantly different results, and their averaged BFP/mCherry expressions were used to gauge the antisense RNA effect by relating it to the construct with the respective uninhibited expression (dashed lines, Figure 2). Note that it is likely that the antisense RNA effects we observed here represent the upper limit of the asRNA effects. In the case of converging transcription, RNA polymerase collision events will produce a larger proportion of aborted transcripts, which may have a smaller inhibitory effect than full length transcripts. At all three distances, substantial antisense RNA (asRNA) mediated silencing of mCherry was observed (Figure 2). We did not attempt to model effects mediated by antisense RNA, as they are presumably highly sequence dependent - secondary structure and in vivo half-life of RNAs varies drastically with sequence.20 Very different effects of asRNA in constructs with convergent transcription are evident from reported studies: They range from a contribution similar to that of direct TI,11 as was observed in the present study, to no noticeable effect of the antisense RNA in between convergent promoters.10 The remaining difference in expression between the antisense RNA control constructs and the respective constructs with convergent transcription was assumed to be due to direct transcriptional interference (Figure 2). The level of TI decreased with distance, from 3.2 fold TI for the no spacer construct down to just 1.1 fold with the 1000 bp spacer. Flow cytometry demonstrated a high correlation between mCherry and BFP fluorescence for constructs expressing both (Figure S4). Thus, the ratio between the two fluorescent proteins is highly similar among single cells with the same construct.

ACS Paragon Plus Environment

Page 4 of 13

Page 5 of 13 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 1. Total inhibitory effect of antisense transcription. Reporter constructs with either convergent promoters (Pstrong and Pweak) or a single promoter were generated to measure total inhibitory effect due to the combined effects of antisense RNA and direct transcriptional interference. Three sets of constructs were made with no added spacer between two reporter genes, a 500 bp spacer or a 1000 bp spacer. Data are mean ± SD, n = 4.



Figure 2. Separation of effects of TI and antisense RNA. To discern contributions from cis-acting TI and trans-acting antisense RNA effects to the total inhibitory effect, complementary BFP and mCherry expression modules were placed on separate loci of the same plasmid DNA. The reduced expression of BFP and mCherry in those constructs relative to uninhibited expression in constructs producing only one transcript was due to antisense RNA effects. The remaining reduction was attributed to cis-acting TI. The fold-changes in gene expression due to TI are indicated in the bar graph. The dotted lines bordering the shaded areas indicate observed BFP and mCherry expression levels without inhibition by antisense transcription or trans-produced antisense RNA (Figure 1). Data are mean ± SD, n = 4.

Spontaneous termination in untranslated regions TI results were simulated using stochastic simulations.14 Initial simulations of the observed transcriptional interference suggested the need to account for reduced bacterial RNAP processivity in untranslated regions. To quantify spontaneous termination in untranslated DNA stretches, the reporter genes BFP and mCherry were placed in a tandem arrangement, such that BFP with Pstrong is upstream of mCherry, which does not bear a functional promoter, but does have a ribosome binding site (Figure 3A). The distance between BFP and mCherry was varied by the insertion of untranslated spacers in between. With increasing distance, the expression of BFP did not vary but mCherry expression was significantly reduced, as expected (Figure 3A). Fitting the decay of mCherry fluorescence with increasing spacer length (Figure 3B), the rate of termination kT within the untranslated regions was estimated to be 0.085 s-1 (Figure S5A), assuming a constant RNA polymerase speed of 50 bp/s. Spontaneous termination in untranslated regions is likely to be Rho dependent and mediated by NusG binding to the nascent mRNA chain in the absence of a trailing


Page 6 of 13

Page 7 of 13 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


ribosome.16,17 This is consistent with the observation of increased read-through of the untranslated spacers when partially inhibiting Rho with bicyclomycin (Figure S6).

Figure 3. The processivity of RNAP is reduced in untranslated regions. (A) Tandem constructs consisting of Pstrong driving BFP and mCherry reporters, separated by untranslated spacing sequences. The distance between the two reporter genes was varied by inserting spacers of 100 bp, 200 bp, 500 bp and 1000 bp, and the fluorescent intensities for both BFP and mCherry were measured. (B) Data (circles) and simulation (lines) of the FI of mCherry relative to the 0 bp spacer construct. Data are mean ± SD, n = 4. The simulation lines shows the range of fits from 100 stochastic simulations performed at 5 bp increments from 0 bp to 1000 bp spacer, with kT = 0.085 s-1.



Modelling of transcriptional interference The proportion of inhibition attributed to direct transcriptional interference in the convergent arrangement of BFP and mCherry was modelled on the basis of a previously developed stochastic TI model.14,15,21 It is a discrete time step model, in which each time step is the time taken for an elongating RNAP to advance one base pair, while the occurrence of other possible events is stochastically sampled. If an event is possible, it occurs in the next time step, if a randomly generated number between 0 and 1 is less than 1– e–k, where k is the relevant rate of the particular event. Firing rates of the two promoters were deduced from their relative strengths, accounting for promoter self-occlusion (Supplementary Method). Promoter firing was simulated as a two-step process, consisting of loading of RNAP holoenzyme to form an open complex, and transition of the open complex to an elongating complex. When an elongating RNAP collides with an open complex at the promoter, the open complex is removed, and the elongating RNAP continues (sitting duck TI). In collisions of two elongating RNAPs, one polymerase is terminated, whereas the other continues transcription (collisional TI). To achieve model convergence with the experimental data, two additional parameters had to be introduced, both pertaining to a lower RNA polymerase processivity of untranslated DNA stretches (Figure 4A). Without any of the two parameters, the predicted TI of Pstrong exerted on Pweak is much stronger than what was experimentally observed (Figure S7). The first parameter kT describes the spontaneous termination of RNA polymerases in untranslated regions and was estimated to be 0.085 s-1 (Figure 3B). The second parameter PTC_trans allows for an asymmetrical termination probability in head-on collisions of two elongating RNA polymerases, where one RNA polymerase is transcribing a coding sequence, and thus is likely to be coupled to a trailing ribosome. We assumed that upon head-on collision of two RNA polymerases, only one of the two continues transcription, and the other one is immediately removed from the system. If the survival of each polymerase is equally probable, i.e. in untranslated regions, then this probability becomes PTC_untrans = 0.5. Fitting the transcriptional interference data with the expanded model with PTC_trans as a free parameter yielded a value of 0.07 (Figure S5B), indicating a strong bias for termination of the RNA polymerase without trailing ribosome: Accordingly, the trailed ribosome is about 13 times more likely to continue transcription than the untrailed one. In a previous study on TI in bacteria using a single gene facing a convergent downstream promoter, the same effect was encountered: RNA polymerases transcribing in the sense direction were found to be more resistant to collision-induced termination than polymerases in an antisense direction.11 In the model used in that study, a termination probability upon collision was introduced for each direction as a free parameter. The sum of both termination probabilities was markedly below 1, suggesting the occurrence of bypassing of head-on colliding polymerases. To investigate the possibility of RNA polymerase bypass, we also included a bypass probability as a free parameter into our TI model. However, this parameter did not converge with a global fit of the data in our model. Thus, including the possibility of both RNA polymerases surviving head-on collisions did not result in a better explanation of the data. Such a collision bypass has only been experimentally shown for single-unit bacteriophage RNA polymerases,22 whereas in vitro studies with prokaryotic23 and eukaryotic24 RNA polymerases suggested that their multi-unit counterparts were not able to bypass each other. The refined model reproduced transcriptional interference of Pstrong on Pweak driving mCherry expression very well (Figure 4B). However, interference of Pweak on Pstrong was more substantial than the model ACS Paragon Plus Environment

Page 8 of 13

Page 9 of 13 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


predicted for the smallest inter-promoter distance. We investigated whether this stronger-than-expected TI on BFP expression might be explained by the pulsatile nature of mRNA expression in bacteria.25 Due to transiently increased promoter firing rates, the weaker promoter might exert a higher TI on the strong promoter. Thus the model was extended to incorporate bursty promotor behavior (Figure S8) in a similar way as previously reported.26 However, the fit with the best overall score of the extended bursty model only marginally differed from the fit of the original model and did not explain the extent of TI on the strong promoter (Figure 4B). Having a mathematical model also makes it possible to track the fates of each RNAP originating from Pstrong and Pweak, which in turn provides mechanistic insights on the contributions of different TI mechanisms to the overall TI. Our simulation suggests that without any additional spacer between BFP and mCherry, an RNAP initiated from Pstrong almost never terminates until it has finished transcribing the BFP gene (Figure 4C). This is likely due to the large difference in strength between Pstrong and Pweak, meaning that only few RNAPs initiated from mCherry would reach the BFP coding region. In addition, the RNAPs originated from Pstrong are also protected by trailing ribosomes when transcribing the BFP gene. Once the BFP transcribing RNAP has moved past the BFP coding region however, it is no longer protected by the trailing ribosome, and is thus subjected to both spontaneous termination and collision-induced termination. Nevertheless, ~10% of the RNAPs originated from Pstrong are able to progress through the ~1 kb mCherry region to reach the mCherry promoter and cause sitting duck TI at Pweak. The sitting duck TI itself contributes to ~40% TI seen at Pweak, but is gradually reduced to ~20% for the 500 bp spacer construct and less than 10% for the 1000 bp spacer construct. Accordingly, the proportion of BFP RNAPs capable of progressing to the mCherry promoter is also significantly reduced in the longer constructs. In addition, increasing the intergenic distance will also increase the probability of RNAP collisions occurring within the intergenic region, when both RNAPs have fully transcribed their respective FP gene. The presented stochastic TI model demonstrates the importance of co-transcriptional translation in bacteria for TI outcomes - a trailing ribosome protects the elongating RNA polymerase from both spontaneous and collision-induced transcription termination. Both effects lead to suppression of antisense transcription. In the natural context, this can serve metabolic efficiency in two ways. First, transcription from antisense cryptic promoters is suppressed. As about 10% of random 100 bp sequences display promoter activity in E. coli,27 accidental promoters are likely to be quite frequent in bacterial genomes. Second, transcriptional interference between convergent genes is limited even without functional terminators, especially when they are separated by untranslated DNA stretches. There are also ramifications for synthetic genetic circuit design, as the described translation effects weaken achievable TI between convergent genes. Thus, to achieve a notable inhibitory effect of one gene onto the other, either a substantial difference in promoter strength is required, or the coding regions have to be placed outside of the inter-promoter space. Consequently, this study constitutes a quantitative general model for TI in convergent gene layouts in bacteria, explicitly taking into account the involvement of co-transcriptional translation. Firstly, this allows the assessment of TI achieved in naturally occurring convergent gene arrangements, providing a deeper understanding of gene regulation in the genomic context. Secondly, the model can guide the design of artificial genetic circuitry based on the inhibitory effect of antisense transcription. As convergent gene layouts can achieve inversely correlated gene expression of two genes, this presents a promising design ACS Paragon Plus Environment


strategy in synthetic biology. However, in order to assess the entire inhibition by antisense transcription, general rules to quantify the antisense RNA mediated effect would complement our model, warranting future studies to elucidate them.

Figure 4. Modelling of transcriptional interference. (A) Schematic of the described TI model indicating assumed RNAP footprints, elongation speed as well as collision-induced termination probabilities for left- and rightward transcription in the coding regions and the untranslated gap region. Green ellipses symbolize ribosomes trailing RNAPs in translated regions. (B) Data (symbols) and simulations (lines) of TI at Pweak (top) and Pstrong (bottom). Data are mean ± SD, n = 4. The shaded area shows the range of fits from 100 stochastic simulations, with kT = 0.085 s-1, PTC_untrans = 0.5, and PTC_trans = 0.07. Simulations were performed at 5 bp increments from 0 bp to 1000 bp insertions. The bursty promoter fits were shown as short green lines. (C) Cumulative distribution of RNAPs initiated from Pstrong (blue) and Pweak (red) for each of the 3 intergenic distances. Figures are drawn to scale.


Page 10 of 13

Page 11 of 13 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Methods Assays Reporter constructs with BFP and mCherry were cloned into pSB4S5 (annotated sequences available on request), derived from pSB4K5 by swapping the nptII gene for an aadA gene. Assay plasmids were transformed into E. coli XL1-blue. Transformed cells were plated onto lysogeny broth agar (10 g/l casein tryptone, 5 g/l yeast extract, 5 g/l NaCl, 15 g/l agar agar) with 100 µg/ml spectinomycin. 96-well U-bottom plates with 200 µl double yeast tryptone (16 g/l casein tryptone, 10 g/l yeast extract, 5 g/l NaCl) with 100 µg/ml spectinomycin per well were inoculated from transformation plates with eight replicates per construct. Those assay cultures then were incubated at 37°C under orbital shaking for 24 hours. Cells were pelleted and resuspended in 200 µl phosphate buffered saline (PBS) per well after removal of the medium. For full fluorophore maturation, assay plates with cell suspensions were incubated for further 24 hours at 4°C. Again, cells were pelleted and resuspended in fresh PBS. Cell suspensions then were transferred to flat bottom plates for higher accuracy of density readings. In a Tecan M-1000 multimode plate reader optical density at 600 nm (OD600) and red (ex: 587 nm, em: 610 nm) and blue (ex: 398 nm, em: 451 nm) fluorescence were recorded for each well. For normalization with optical density, PBS blank values were subtracted from every reading and fluorescence values were divided by the respective optical density.

Stochastic simulations Stochastic simulations were performed as previously described14,15 with added functionalities to account for i) spontaneous RNAP termination at untranslated DNA and ii) biased termination towards ribosome free RNAPs following head-on RNAP collision. The simulation programs were written in FORTRAN (available on request). In a typical run, 5x107 time steps were simulated for each condition. A detailed model description can be found in the supplements.

Supporting Information Table S1. Insulators and insulated promoters with respective symbols used in schematics and annotated sequence. Figure S1. Removing cryptic promoter activity of transcriptional insulators. Figure S2. Strength mapping of insulated promoters. Figure S3. Test for cryptic promoter activity within the spacer sequence. Figure S4. Flow cytometry analysis of assay constructs with 0 bp spacer. Figure S5. Funnel plot analysis of parameter fits. Figure S6. Spontaneous termination in untranslated regions without and with Rho inhibition. Figure S7. Modelling outcomes omitting parameters for reduced RNAP processivity in untranslated regions. Figure S8. Bursty promoter model. ACS Paragon Plus Environment


Acknowledgement We thank Dr. Ian Dodd for discussion and critical reading of the manuscript and Jeannette Wenzel for technical assistance. The work was funded by an ARC Discovery Early Career Researcher Award to N.H. [DE150100091] and an ARC Discovery grant to KES [DP150103009]. N.H. was also part funded through the CSIRO Synthetic Biology Future Science Platform.

References (1)

Georg, J.; Hess, W. R. (2011) cis-antisense RNA, another level of gene regulation in bacteria. Microbiol. Mol. Biol. Rev. 75, 286–300.

(2)

Dornenburg, J. E.; DeVita, A. M.; Palumbo, M. J.; Wade, J. T. (2010) Widespread antisense transcription in Escherichia coli. MBio 1, 1–4.

(3)

Hao, N.; Palmer, A. C.; Dodd, I. B.; Shearwin, K. E. (2017) Directing traffic on DNA—How transcription factors relieve or induce transcriptional interference. Transcription 8, 120–125.

(4)

Hongay, C. F.; Grisafi, P. L.; Galitski, T.; Fink, G. R. (2006) Antisense transcription controls cell fate in Saccharomyces cerevisiae. Cell 127, 735–745.

(5)

Latos, P. A.; Pauler, F. M.; Koerner, M. V; Ba, H.; Hudson, Q. J.; Stocsits, R. R.; Allhoff, W.; Stricker, S. H.; Klement, R. M.; Warczok, K. E.; et al. (2012) Airn transcriptional overlap but not its lncRNA products induces Igf2R silencing. Science 338, 1469–1473.

(6)

Chatterjee, A.; Johnson, C. M.; Shu, C.-C.; Kaznessis, Y. N.; Ramkrishna, D.; Dunny, G. M.; Hu, W.-S. (2011) Convergent transcription confers a bistable switch in Enterococcus faecalis conjugation. Proc. Natl. Acad. Sci. U. S. A. 108, 9721–9726.

(7)

Brantl, S. (2007) Regulatory mechanisms employed by cis-encoded antisense RNAs. Curr. Opin. Microbiol. 10, 102–109.

(8)

Pelechano, V.; Steinmetz, L. M. (2013) Gene regulation by antisense transcription. Nat. Rev. Genet. 14, 880–893.

(9)

Shearwin, K. E.; Callen, B. P.; Egan, J. B. (2005) Transcriptional interference--a crash course. Trends Genet. 21, 339–345.

(10)

Bordoy, A. E.; Varanasi, U. S.; Courtney, C. M.; Chatterjee, A. (2016) Transcriptional interference in convergent promoters as a means for tunable gene expression. ACS Synth. Biol. 5, 1331–1341.

(11)

Brophy, J. A.; Voigt, C. A. (2016) Antisense transcription as a tool to tune gene expression. Mol. Syst. Biol. 12, 854–854.

(12)

Huber, F.; Bunina, D.; Gupta, I.; Khmelinskii, A.; Meurer, M.; Theer, P.; Steinmetz, L. M.; Knop, M. (2016) Protein abundance control by non-coding antisense transcription. Cell Rep. 15, 2625–2636.

(13)

Hoffmann, S. A.; Kruse, S. M.; Arndt, K. M. (2016) Long-range transcriptional interference in E. coli used to construct a dual positive selection system for genetic switches. Nucleic Acids Res. 44. Epub Feb 29, 2016. DOI: 10.1093/nar/gkw125.

(14)

Hao, N.; Palmer, A. C.; Ahlgren-berg, A.; Shearwin, K. E.; Dodd, B. (2016) The role of repressor ACS Paragon Plus Environment

Page 12 of 13

Page 13 of 13 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


kinetics in relief of transcriptional interference between convergent promoters. Nucleic Acids Res. 44, 6625–6638. (15)

Sneppen, K.; Dodd, I. B.; Shearwin, K. E.; Palmer, A. C.; Schubert, R. A.; Callen, B. P.; Egan, J. B. A (2005) Mathematical model for transcriptional interference by RNA polymerase traffic in Escherichia coli. J. Mol. Biol. 346, 399–409.

(16)

Peters, J. M.; Mooney, R. A.; Grass, J. A.; Jessen, E. D.; Tran, F.; Landick, R. (2012) Rho and NusG suppress pervasive antisense transcription in Escherichia coli. Genes Dev. 26, 2621–2633.

(17)

Richardson, J. P. (1991) Preventing the synthesis of unused transcripts by Rho factor. Cell 64, 1047– 1049.

(18)

Proshkin, S.; Rahmouni, A. R.; Mironov, A.; Nudler, E. (2010) Cooperation between translating ribosomes and RNA polymerase in transcription elongation. Science 328, 504–508.

(19)

Davis, J. H.; Rubin, A. J.; Sauer, R. T. (2011) Design, construction and characterization of a set of insulated bacterial promoters. Nucleic Acids Res. 39, 1131–1141.

(20)

Jain, C. (2002) Degradation of mRNA in Escherichia coli. IUBMB Life 54, 315–321.

(21)

Hao, N.; Krishna, S.; Ahlgren-Berg, A.; Cutts, E. E.; Shearwin, K. E.; Dodd, I. B. (2014) Road rules for traffic on DNA - systematic analysis of transcriptional roadblocking in vivo. Nucleic Acids Res. 42, 8861–8872.

(22)

Ma, N.; McAllister, W. T. (2009) In a head-on collision, two RNA polymerases approaching one another on the same DNA may pass by one another. J. Mol. Biol. 391, 808–812.

(23)

Crampton, N.; Bonass, W. A.; Kirkham, J.; Rivetti, C.; Thomson, N. H. (2006) Collision events between RNA polymerases in convergent transcription studied by atomic force microscopy. Nucleic Acids Res. 34, 5416–5425.

(24)

Hobson, D. J.; Wei, W.; Steinmetz, L. M.; Svejstrup, J. Q. (2012) RNA polymerase II collision interrupts convergent transcription. Mol. Cell 48, 365–374.

(25)

Golding, I.; Paulsson, J.; Zawilski, S. M.; Cox, E. C. (2005) Real-time kinetics of gene activity in individual bacteria. Cell 123, 1025–1036.

(26)

So, L. H.; Ghosh, A.; Zong, C.; Sepúlveda, L. A.; Segev, R.; Golding, I. (2011) General properties of transcriptional time series in Escherichia coli. Nat. Genet. 43, 554–560.

(27)

Yona, A. H.; Alm, E. J.; Gore, J. (2018) Random sequences rapidly evolve into de novo promoters. Nat. Commun. 9, 1–10.


Characterizing transcriptional interference between converging genes

Recommend Documents