
==== Front
iScience
iScience
iScience
2589-0042
Elsevier

S2589-0042(24)01999-0
10.1016/j.isci.2024.110774
110774
Article
A library-based approach allows systematic and rapid evaluation of seed region length and reveals design rules for synthetic bacterial small RNAs
Brück Michel 12
Köbel Tania S. 1
Dittmar Sophie 2
Ramírez Rojas Adán A. 1
Georg Jens 3
Berghoff Bork A. bork.berghoff@uni-ulm.de
25∗
Schindler Daniel daniel.schindler@mpi-marburg.mpg.de
146∗∗
1 Max-Planck-Institute for Terrestrial Microbiology, Karl-von-Frisch-Straße 10, 35043 Marburg, Germany
2 Institute for Microbiology and Molecular Biology, Justus Liebig University Giessen, Heinrich-Buff-Ring 26-32, 35392 Giessen, Germany
3 Institut für Biologie III, Albert-Ludwigs-Universität Freiburg, Schänzlestraße 1, 79104 Freiburg, Germany
4 Center for Synthetic Microbiology, Philipps-University Marburg, Karl-von-Frisch-Straße 14, 35032 Marburg, Germany
∗ Corresponding author bork.berghoff@uni-ulm.de
∗∗ Corresponding author daniel.schindler@mpi-marburg.mpg.de
5 Present address: Institute of Molecular Biology and Biotechnology of Prokaryotes, University of Ulm, Albert-Einstein-Allee 11, 89081 Ulm, Germany

6 Lead contact

20 8 2024
20 9 2024
20 8 2024
27 9 11077425 4 2024
14 6 2024
15 8 2024
© 2024 The Author(s)
2024
https://creativecommons.org/licenses/by/4.0/ This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
Summary

All organisms must respond to environmental changes. In bacteria, small RNAs (sRNAs) are an important aspect of the regulation network underlying the adaptation to such changes. sRNAs base-pair with their target mRNAs, allowing rapid modulation of the proteome. This post-transcriptional regulation is usually facilitated by RNA chaperones, such as Hfq. sRNAs have a potential as synthetic regulators that can be modulated by rational design. In this study, we use a library-based approach and oxacillin susceptibility assays to investigate the importance of the seed region length for synthetic sRNAs based on RybB and SgrS scaffolds in Escherichia coli. In the presence of Hfq we show that 12 nucleotides are sufficient for regulation. Furthermore, we observe a scaffold-specific Hfq-dependency and processing by RNase E. Our results provide information for design considerations of synthetic sRNAs in basic and applied research.

Graphical abstract

Highlights

• A library-based approach for synthetic sRNAs with varying seed region length (SRL)

• Synthetic RybB sRNAs are processed by RNase E at AU-rich sequence motifs

• Accessibility of SgrS seed regions is an important determinant for functionality

• Importance of Hfq and SRL for regulatory efficiency depends on the sRNA scaffold

Molecular mechanism of gene regulation; Molecular microbiology; Genetic engineering

Subject areas

Molecular mechanism of gene regulation
Molecular microbiology
Genetic engineering
Published: August 20, 2024
==== Body
pmcIntroduction

Bacterial small RNAs (sRNAs) are short regulatory RNA sequences that allow cells to rapidly adapt to environmental changes. sRNAs typically alter the translation of their mRNA targets through imperfect antisense binding, which is mediated by the sRNA seed region. In most cases, sRNAs inhibit translation and decrease the stability of target mRNAs by base-pairing within or in close proximity to the ribosome binding site (RBS).1,2,3 However, activation of translation by sRNAs is occasionally observed.4 sRNA-mediated regulation is commonly facilitated by RNA chaperones, with Hfq being one of the best characterized to date. Hfq assembles into a hexameric complex with three RNA interaction sites: the proximal face, the distal face, and the lateral rim.5,6,7 sRNAs and mRNAs use distinct RNA sequence motifs to differentially occupy these interaction sites, and Hfq serves as a platform for base-pairing between the RNAs.8,9,10 With the help of Hfq, the number of nucleotides (nt) that are needed for sRNA-mRNA interactions is often less than ∼10 nt.9 In addition, Hfq is important for the stability of sRNAs; Hfq-dependent sRNAs are typically less abundant in the absence of Hfq.5 The majority of mRNA decay in bacteria is initiated by endoribonucleases (predominantly RNase E), and subsequent decay is performed by 5′ and 3′ exoribonucleases that degrade the remnants of RNase E-cleaved mRNAs.11,12 The ternary complex of Hfq, sRNA, and mRNA may facilitate RNase E-mediated decay of both the mRNA and the sRNA, depending on the 5′ phosphorylation state of the sRNA.13,14 Hence, RNase E-mediated cleavage considerably contributes to many aspects of sRNA-based regulation.15,16 Because of their regulatory properties, sRNAs are gaining interest as tools to control gene expression and fine-tune phenotypes of bacterial cells.

Synthetic sRNAs have tremendous potential to expand the toolbox of genetic engineering techniques, allowing for strain improvement to increase the yield of the target product.17 Synthetic sRNAs have the advantage of allowing rapid plasmid-based assays18 to identify efficient target regulation prior to performing labor-intensive genome engineering procedures. While synthetic sRNAs are promising tools for genetic engineering, antisense peptide nucleic acids (PNAs) have the potential to serve as on-demand programmable antibiotics by targeting essential or resistance genes of specific target organisms.19,20,21 Unlike an antibiotic, such PNAs can be rapidly modified by changing the antisense sequence, which has enormous potential to address the antibiotic resistance crisis. Furthermore, species-specific targeting might be achievable with such an antisense strategy, leaving the majority of the microbiome intact.22 For these reasons, computational tools have been developed to predict the optimal seed region for antisense binding of synthetic sRNAs or antisense oligomers.23,24 Such prediction tools allow systematic testing but do not guarantee sRNA functionality.25 The binding probability can be estimated by calculating the binding energy, an approach that depends on careful consideration of structural RNA features.26 It is generalized that the lower the binding energy the stronger the binding between two RNA species. However, in addition to the binding energy, other factors such as the accessibility and location of the antisense-mediated binding region and the secondary structures of the synthetic sRNA or mRNA may play important roles. There is a lack of knowledge to predict and apply perfect synthetic sRNAs on demand. Systematic studies are needed to improve prediction tools and to generate training data for artificial intelligence-based prediction in the future for time- and cost-effective application of synthetic sRNAs in basic and applied research.

Here, we use a library-based Golden Gate cloning approach and laboratory automation workflow to obtain a library of synthetic sRNAs with increasing seed region lengths (SRL) and to characterize the functionality of these synthetic sRNAs in Escherichia coli. The experimental design is based on our previously characterized synthetic sRNAs and an established oxacillin susceptibility assay.18 We use the designed library to identify the minimum SRL for functional synthetic sRNAs. We investigate the steady-state levels of synthetic sRNAs, the potential for Hfq-independent post-transcriptional regulation with increasing SRL and RNase-dependent processing of selected candidates. We further extend the library-based approach from the RybB sRNA containing a 5′ seed region to the structurally complex SgrS sRNA containing a 5′ open reading frame (sgrT) and an internal seed region. Our results show that the SRL differentially affects RybB and SgrS. RybB functionality seems to be independent of the SRL, albeit RNase E-mediated processing occurs with increasing SRL. In case of SgrS, such processing is not observed. Instead, SgrS functionality is sensitive to SRL-mediated changes in local secondary structures, and an increasing SRL helps to alleviate the Hfq-dependency of SgrS. These observations underscore that library-based approaches provide efficient means to optimize synthetic sRNA design and to study sRNA biology.

Results

Concept of Golden Gate-based sRNA library construction and validation

sRNAs have a modular structure and can be as simple as a 5′ seed region, capable of binding complementary mRNAs, and a short single-stranded region followed by a stem-loop terminator structure (Figure 1A). Their small size and the modular structure allow the standardized, high-throughput construction of synthetic sRNAs to be tested and used as post-transcriptional regulators in synthetic biology and biotechnology applications. We have previously described an easy-to-use plasmid toolset for efficient generation and benchmarking of synthetic small RNAs in bacteria.18 The system is based on the Golden Gate DNA assembly of synthetic sRNA expression constructs. This has been the basis for the development of a Modular Cloning toolbox utilizing SapI, an enzyme generating 3-bp overhangs and thereby reducing the size of the fusion sites compared to the 4-bp overhangs that are generated by other Type IIS restriction enzymes (e.g., BbsI, BsmBI, and BsaI).27,28,29 Briefly, this system allows the generation of level 0 constructs which can be combined to transcription units (TU) in designated acceptor plasmids with minimal fusion sites reducing potential scar sequences (Figure 1B). For this study we created a pBAD-derived acceptor plasmid for cloning of distinct sRNA TUs via SapI.25 The plasmid has the same properties and features as in our previous study, including the mCherry-ccdB counter-selection cassette, only the Type IIS enzyme was swapped from BbsI to SapI.18 In our previous study, we tested a set of 16-nt seed regions in combination with the RybB scaffold to repress acrA mRNA which allows us to perform oxacillin susceptibility assays to validate sRNA functionality (acrA repression leads to an increase in oxacillin susceptibility). We have shown that targeting different positions in the translation initiation region (TIR) reduces AcrA abundance with high probability and causes oxacillin susceptibility. We picked 16-nt seed regions because it resembles the length of the natural RybB seed region.30 However, for synthetic biology applications we were interested in the effects of SRL variation and its correlation to post-transcriptional regulation. We created a library based on the coordinates of the previously described 16-nt seed region s8 (s8 = seed sequence 8 of the initial screen; cf. Köbel et al.18), with seed regions ranging from 2 to 82 nt in length and 2-nt increments toward the transcription start site (TSS) of acrA mRNA (Figures 1C and S1). The corresponding sRNAs were denoted SRL2 to SRL82, respectively. The 5′ untranslated region (UTR) of acrA was previously determined to have a length of 79 nt.31 Notably, we were expecting that based on the cloning fidelity the smaller seed regions will not be generated. The reason for lower efficiencies for short seed sequences in the library may be that the double-stranded DNA can denature under the Golden Gate DNA assembly reaction conditions of 37°C. The different lengths of the seed sequence library can introduce a bias into the library, which should be considered especially as the library complexity increases. The SRL-library was ordered as an oligonucleotide pool, in which each individual SRL-oligonucleotide contained overhangs for PCR amplification and SapI recognition sites for Golden Gate cloning (Figure 1C). To create the plasmid library, the level 0 parts for the promoter, the scaffold, the destination plasmid, and the PCR-amplified library are combined in one Golden Gate reaction. The reaction was directly transformed into E. coli MG1655 wild-type cells and selection was performed on solid media to obtain individual colonies.Figure 1 Workflow of synthetic sRNA cloning, library generation and characterization

(A) Synthetic sRNAs for post-transcriptional control can have a short length and a simple structure consisting of the 5′ seed region and the sRNA scaffold. The seed region can be modified for specific targeting of mRNAs.

(B) Based on the simple structure, sRNAs can be modularized and constructed by Golden Gate cloning. The transcription unit of the constructed synthetic sRNAs consists of the promoter (in this case PLlacO-1), a seed region and the sRNA scaffold (in this case RybB). The three parts are combined into the appropriate acceptor plasmid for subsequent experiments.

(C) Concept of oligonucleotide pool-based libraries. The pooled oligonucleotide sequences contain 5′ and 3′ universal sequences providing the required Type IIS recognition and cut sites (the Type IIS recognition site is visualized by the gray box with arrows for its direction and the cut site is highlighted by the indicated nucleotide sequence of 3 nt in length). The universal sequences serve as primer binding sites (indicated by the outermost arrows) to convert the single-stranded oligonucleotide pool into a double-stranded DNA library that can subsequently be used for cloning approaches. Different colors of oligo pool sequences indicate different seed region lengths in the library.

(D) Schematic workflow for construction and selection of synthetic sRNAs. The workflow can be completed within <10 days depending on the chosen sequence validation method to obtain characterized synthetic sRNAs for their application.

For subsequent characterization, we have established a workflow supported by automated equipment to increase throughput (Figure 1D). This workflow is designed to isolate individual candidates from the library and perform an initial screening. Working with a pooled library is not possible in this case because functional candidates have reduced viability under our chosen selection pressure, the β-lactam antibiotic oxacillin. We also tested whether a rapid workflow for colony PCR-based sequencing could be established using Nanopore sequencing to identify individual candidates.32,33 To achieve this in a cost-effective and sustainable manner, we used our previously described highly multiplexed barcoding approach and downscaled reaction volumes using a combination of a bulk dispenser, an acoustic dispenser, and a screening robot to minimize reagent use and plastic waste. Using this workflow, 66% of the candidates could be assigned an SRL. For the remaining 34%, either coverage was insufficient, the result was ambiguous (possibly not a single isolate), or no amplicon was obtained. Using this protocol, four experiments were sequenced simultaneously on one Flongle cell, resulting in 1,536 tested candidates, resulting in a cost of less than 0.10 € per barcoded amplicon, including colony PCR, sequencing library generation and Nanopore sequencing.32

Functional characterization of the synthetic SRL-library using the RybB scaffold

To create a RybB library with varying SRL, two level 0 plasmids containing either the PLlacO-1 promoter (hereafter, PL) (pSL135)16,34 or the RybB scaffold sequence (pSL123), the destination plasmid (pSL137), and the amplified SRL-library were assembled in one Golden Gate reaction. The reaction was transformed into E. coli wild type MG1655 (hereafter referred to as “wild type”), and dilutions were plated to obtain plates with optimal density for automated colony picking. For the purpose of workflow establishing, 380 candidates were picked into a 384-well format (Figure 2A). The last four positions were kept empty to accommodate controls. The processing of the samples was performed according to the described workflow (Figure 1D). In a first step, the 380 candidates and four controls were screened on solid media containing varying concentrations of oxacillin (0–200 μg/mL) (Figure 2A, not all data shown). The controls are the wild type with an empty plasmid, the ΔacrA strain with an empty plasmid, the previously characterized pPL-RybB-s8 strain (s8 = 16-nt seed region from Köbel et al.18), and a media control. Strains were cultivated in the 384-well format overnight and spotted onto solid media for the initial screen (Figure 2A). Under the reference condition, without oxacillin, all patches show growth. Under the testing condition with 75 μg/mL oxacillin, however, growth is reduced for several candidates. Importantly, the previously characterized pPL-RybB-s8 control is showing no strong phenotype, indicating that the SRL-library contains candidate sRNAs with a superior post-transcriptional regulation. Based on these results, candidates which show increased oxacillin susceptibility could be individually analyzed by Sanger sequencing. However, we were interested in the composition of the whole library. Therefore, we decided to sequence all candidates in house on the Nanopore sequencing platform with a custom dual barcoded amplicon sequencing procedure (cf. Methods).32,33 Based on the results, we successfully generated an SRL-collection from 2 to 82 nt incrementing in 2-nt steps. In addition, the library contained a control with no seed region at all (denoted “SRL0”). Three constructs with an SRL of 4, 8, and 12 nt were not obtained within the library, presumably due to low cloning fidelity of short fragments.Figure 2 Initial screening and detailed characterization of the RybB SRL-library targeting acrA mRNA

(A) Screening of RybB SRL-library. 380 candidates and four respective controls were spotted from fresh overnight cultures into 3 × 3 grids on LB-agar plates with and without oxacillin. Comparing the reference and the testing condition indicates multiple candidates render cells susceptible toward oxacillin. Comparing the established pPL-RybB-s8 construct18 with the library indicates sRNA constructs with a superior regulation. Violet box indicates the position of indicated controls.

(B) Liquid growth analysis of the SRL-library indicates regulation for the majority of synthetic sRNAs by comparing the area under the curve (AUC) in the presence and absence of oxacillin. According to the assay, at least 14 nt are needed for regulation of acrA mRNA. Further, the SRL does not correlate with its functionality despite possessing lower binding energies. Binding energies were calculated by IntaRNA (for display reasons the values are multiplied by −1). Wild type (wt) and ΔacrA with an empty plasmid serve as positive and negative control, respectively. Oxacillin susceptibility assay was performed in quadruplicate. The Wilcoxon-Mann-Whitney test was used to assess the significance of the difference between growth with and without oxacillin. p-value adjustment was performed using the Benjamini-Hochberg procedure; ∗ = p < 0.05 (non-parametric, unpaired).

(C) Oxacillin susceptibility test on solid medium reveals that 12 nt seem to be the minimal SRL for functionality of synthetic sRNAs. A representative replicate is shown.

(D) Oxacillin susceptibility assay in liquid culture shows similar results. However, in liquid media an SRL of 11 nt already shows regulation, presumably based on the increased forces onto the bacterial cell wall compared with growth on solid substrate. Binding energies were calculated by IntaRNA (for display reasons the values are multiplied by −1). Oxacillin susceptibility assay was performed in quadruplicate. The Wilcoxon-Mann-Whitney test was used to assess the significance of the difference between growth with and without oxacillin. p-value adjustment was performed using the Benjamini-Hochberg procedure; ∗ = p < 0.05 (non-parametric, unpaired). pPL-RybB-s8 is abbreviated as s8.

We used the defined SRL-library to repeat the screening on solid agar by spotting dilution series onto medium with increasing oxacillin concentration for a qualitative susceptibility assessment (data not shown). For quantitative evaluation we performed plate reader growth kinetics in five replicates to determine the area under the curve (AUC) for susceptibility assessment (Figure 2B). It is not surprising that with an increased SRL the binding energy decreases, as assessed by IntaRNA predictions.26 However, we did not observe a linear increase of post-transcriptional regulation for the synthetic sRNAs with increasing SRL (Figure 2B). Furthermore, regulation strength was not correlated with the expression level of sRNAs, as indicated by northern blot analysis (Figure S2).

Based on our obtained results, we identified an SRL of 10 nt to not be functional in our assay, while an SRL of 14 nt was functional (Figures 2C and 2D; note: s8 = 16-nt seed region as control from Köbel et al; 18). To dissect the threshold for our assay and determine if there are quantitative differences, we individually generated sRNA constructs with an SRL of 11, 12 and 13 nt, and performed their characterization on solid and in liquid media in comparison to the respective controls. An SRL of 11 nt showed a slight post-transcriptional regulation which significantly improved with a length of 12 nt (Figures 2C and 2D). It is, however, possible that these observations are specific for the targeted mRNA sequence and may differ in regard to other targets. The results indicate a stronger regulation in liquid culture which is most likely caused by increased forces onto the bacterial cell wall making oxacillin under this condition more potent.

We selected the synthetic RybB sRNAs with an SRL of 12, 16, 30, 40, 52, and 64 nt for further analyses. The sRNAs were selected because they represent (1) the minimal length needed for regulation (SRL12), (2) the reference length from our previous study (SRL16), (3) the best performing (SRL30 and 40), or (4) well-performing sRNAs with an SRL of >50 nt (SRL52 and 64).

Synthetic RybB sRNAs rely on Hfq irrespective of the seed region length

It was previously reported that RybB is associated with Hfq.35,36 Having the comprehensive SRL-library on hand, we hypothesized that increasing the SRL will reduce Hfq-dependency of synthetic RybB sRNAs. The defined SRL-library was transformed into an Δhfq strain and subsequently analyzed on solid agar (data not shown) and in liquid culture (Figure S3). We observed that the Δhfq strain shows generally a higher susceptibility toward oxacillin which was confirmed by testing the minimum inhibitory concentration (MIC); the MIC of the Δhfq strain was reduced approximately 4-fold in contrast to the wild type (Figure S4). The oxacillin concentration was, therefore, adjusted for experiments with the Δhfq strain. For our detailed analysis we focused on the subset of functional RybB sRNAs chosen in the previous section (SRL of 12, 16, 30, 40, 52, and 64 nt). The comparative growth on solid and in liquid media showed the expected oxacillin-sensitive phenotype in the wild type but not in the Δhfq strain (Figures 3A and S5). To validate the results on the level of post-transcriptional regulation, we tested the selected sRNAs using an established fluorescent reporter construct; a translational fusion of the first nine amino acids of acrA with syfp2.18 As expected, all selected sRNAs caused an at least 10-fold reduction of sYFP2 fluorescence in the wild type. In the Δhfq strain, the fluorescence was only reduced by approximately 2-fold, irrespective of the SRL (Figure 3B). We conclude that (1) the SRL does not influence regulation strength when using the RybB scaffold and (2) that all tested synthetic RybB sRNAs are Hfq-dependent. To further explore the Hfq-dependency, expression levels of the selected sRNAs were analyzed on northern blots, which revealed the expected sizes of the sRNA primary transcripts in both the wild type and the Δhfq strain (Figure 3C). However, steady-state levels in the Δhfq strain were decreased, indicating a reduced stability of the synthetic RybB sRNAs, which supports the role of Hfq as a major determinant for sRNA stability.5 Interestingly, northern blot experiments revealed distinct processing patterns, which were especially apparent for longer seed regions (Figure 3C). Notably, the processing pattern seems not to show a positive or negative influence toward sRNA functionality.Figure 3 Hfq is required for functionality of synthetic RybB sRNAs

(A) Liquid growth oxacillin susceptibility assay for selected RybB candidates in Δhfq. No regulation of the acrA target can be observed, which is consistent with the solid media susceptibility assay (Figure S5). The analysis indicates a reduced viability of Δhfq in contrast to the wild type (wt) and ΔacrA, already in the absence of oxacillin. Oxacillin susceptibility assay was performed in quadruplicate. The Wilcoxon-Mann-Whitney test was used to assess the significance of the difference between growth with and without oxacillin. p-value adjustment was performed using the Benjamini-Hochberg procedure; ∗ = p < 0.13 (non-parametric, unpaired).

(B) The functionality of selected RybB candidates is validated by a previously established sYFP2-reporter assay using the chromosomal acrA-9′-syfp2 construct.18 The sRNAs show functionality in the presence of Hfq. In the absence of Hfq a fluorescence reduction of 50% can be observed for all sRNAs irrespective of the SRL. The control (ctrl) refers to empty plasmid pSL009. A wild type (wt) and Δhfq strain without reporter construct were used as background controls, respectively. Bars represent the mean of three biological replicates and error bars indicate the standard deviation.

(C) Comparison of abundance and processing of selected RybB sRNAs in wild type (wt) and Δhfq by northern blot analysis. The control (ctrl) refers to empty plasmid pSL009. 5S rRNA serves as a loading control. Abundance of RybB sRNAs is reduced on average to 37% +/− 16% (RI = relative intensity in Δhfq in comparison to wild type).

RybB sRNAs from the SRL-library are predominantly processed by RNase E

Two major factors for sRNA processing are the endoribonucleases RNase III and E. RNase III cleaves double-stranded RNAs and is required for the processing of non-coding RNAs.37,38,39 We, therefore, tested the selected synthetic RybB sRNAs in an RNase III deletion background and its isogenic wild-type strain. On northern blots, the processing pattern of the synthetic sRNAs was similar in both strains (Figure 4A). However, we observed a slight increase of sRNA abundance (both primary transcript and processing products) and the accumulation of smaller sized fragments in the RNase III deletion background, suggesting a reduced sRNA decay in this strain, which is in line with the general function of RNase III.37,40 RNase E is an essential RNase which cleaves single-stranded RNAs and plays a prominent role in sRNA processing and decay.41,42,43 We investigated the processing pattern of the selected sRNAs in a temperature-sensitive RNase E background under permissive (30°C) and non-permissive (44°C) conditions. Our results indicate that RNase E is at least one of the RNases responsible for the observed processing (Figure 4B). The processing pattern is visible under permissive but much less pronounced under non-permissive conditions. When inspecting the seed sequences, we observed potential RNase E cleavage sites as AU-rich motifs.41 The potential processing sites correspond with our experimental data (Figures 4C and S2). In summary, we conclude that the synthetic RybB sRNAs are predominantly processed by RNase E, probably due to AU-rich motifs in the seed regions. This finding may have important implications for synthetic sRNA design, in particular if target mRNAs have a high AU-content in the target region.Figure 4 Investigation of RNase-mediated processing of synthetic RybB sRNAs

(A) Northern blot analysis of synthetic RybB sRNAs in an RNase III deletion strain (rnc−) and an isogenic wild type (rnc+). The control (ctrl) refers to empty plasmid pSL009. 5S rRNA serves as a loading control.

(B) Northern blot analysis of synthetic RybB sRNAs in a temperature-sensitive RNase E background (rnets) and an isogenic wild type (rne+). The control (ctrl) refers to empty plasmid pSL009. 5S rRNA serves as a loading control.

(C) Distinct processing patterns with stable intermediates can be observed which matches potential RNase E cleavage sites.41 SRL26, SRL52, SRL64, and SRL72 are visualized because of the first appearance of the processing band in the northern blot of the whole SRL library (see Figure S2). Theoretical and detected processing pattern exemplary visualized for RybB SRL82. For the full northern blot panel of RybB SRL library see Figure S2.

Applying the SRL-library approach to the more complex SgrS sRNA

RybB has a rather simple structure. However, there are sRNAs which have more complex structures, as for example SgrS which has a length of 227 nt. Based on structure predictions it contains a 5′ stem-loop (containing the sgrT open reading frame), a 3′ stem-loop and an internal seed region (Figure 5A). To make use of the reusable part library, we generated two new level 0 parts, the first containing the PL promoter fused to the 5′ SgrS sequence (pSL133) and the second containing the 3′ SgrS sequence (pSL132). This strategy allowed the reuse of the SRL-library, which has been created for RybB (Figure 5B). The library of SgrS was created in parallel to the RybB library and treated in the same manner. In total, 380 candidates were screened for oxacillin susceptibility (Figure 5C) and subsequently sequenced in the same Nanopore sequencing run. Based on the results a defined SRL-library was obtained lacking the SRL of 0, 2, 26, 50 and 80 nt. The library was assessed on solid (data not shown) and in liquid media for oxacillin susceptibility. In contrast to the RybB library, there were only two lengths (36 and 42 nt) that showed a superior regulation at 50 μg/mL oxacillin (Figure 5D). In particular, SRL42 seems to have an increased repression of acrA, even in comparison to the tested RybB candidates. Notably, in the initial library we observed a sequence duplication in the downstream region of SgrS. However, it was experimentally proven that the sequence amplification did not have an influence (data not shown). Nevertheless, all subsequent experiments were performed with the corrected sequence for individually created synthetic SgrS sRNAs. We first tested the minimum SRL needed for functionality of synthetic SgrS sRNAs. As observed for RybB, we determined a minimal SRL of 12 nt for regulation of acrA using the oxacillin susceptibility assay on solid media (100 μg/mL oxacillin) (Figure 5E), which was also confirmed in liquid media at 75 μg/mL oxacillin (Figure S6).Figure 5 Adaptation of the SRL-library approach toward the SgrS sRNA for acrA targeting

(A) Predicted structure of SgrS according to RNAcentral44 visualizes the complex 5′ secondary structure and the 3′ terminator stem-loop. For simplification a schematic is introduced for the subsequent figures. The seed region is indicated in purple and was adapted from Figure S2 of Na et al. 2013.17

(B) Schematic of the Golden Gate assembly of the SRL-library using SgrS as scaffold. In contrast to RybB, the promoter sequence and the 5′ sequence of SgrS are fused allowing reuse of the PCR-amplified SRL-library. All other steps are identical to the cloning of RybB (Figure 1B).

(C) Screening of the SgrS candidate library. 380 candidates and four respective controls were spotted from fresh overnight cultures into 3 × 3 grids on LB-agar plates with and without oxacillin. In contrast to the RybB candidate library, fewer candidates show reduced growth under the screening condition. Violet box indicates the position of indicated controls.

(D) Liquid growth analysis indicates regulation of acrA for some of the selected candidates by accessing the area under the curve (AUC) in the presence and absence of oxacillin. According to the assay at least 14 nt are needed for regulation of acrA. Synthetic SgrS sRNAs with an SRL of 36 and 42 nt are the best performing. The SRL does not correlate with its functionality despite possessing lower binding energies. Binding energies are calculated by IntaRNA (for display reasons the values are multiplied by −1). Wild type (wt) and ΔacrA with an empty plasmid are serving as respective positive and negative controls. Oxacillin susceptibility assay was performed in quadruplicate. The Wilcoxon-Mann-Whitney test was used to assess the significance of the difference between growth with and without oxacillin after p-value adjustment using the Benjamini-Hochberg procedure; ∗ = p < 0.05 (non-parametric, unpaired).

(E) The minimum SRL for functionality of synthetic SgrS sRNAs was determined by oxacillin susceptibility assays on solid media. Similar to RybB, an SRL of 12 nt seems to be sufficient for regulation. A representative replicate is shown.

We hypothesized that the SgrS sRNAs with an SRL of 36 and 42 nt were performing superior in contrast to all other tested SgrS sRNAs because of their secondary structure. We used RNAfold45 to predict the sRNA structures and observed that in most cases the seed region is sequestered (data not shown). In contrast, the seed regions of SRL36 and SRL42 are less likely engaged in intramolecular base-pairing. We further predicted the structures for SgrS sRNAs that deviate by 1 nt from SRL36 and SRL42. According to these predictions, the seed regions of SRL35 and SRL43 are likely accessible, while the seed regions of SRL37 and SRL41 are potentially occluded via intramolecular base-pairing (Figure S7). We hypothesized that SRL37 and SRL41 show reduced functionality due to the lack of seed region accessibility. To test our hypothesis we constructed the respective synthetic SgrS sRNAs and tested their regulatory effect. Indeed, the two candidates, which were predicted to contain a more accessible seed region, performed similarly to SRL36 and SRL42. In contrast, the regulatory potential of SRL37 and SRL41 was abolished or at least significantly impaired on solid media (Figure 6A). Interestingly, the differences in liquid growth conditions are less drastic, but are in line with the functionality of SRL37 on solid media at higher oxacillin concentrations (Figure 6B). Our results show that the accessibility of the sRNA seed region is an important design feature for the construction of functional sRNAs and that a difference of 1 nt may decide whether or not the synthetic sRNA is functional.Figure 6 Synthetic SgrS sRNA functionality is affected by 1-nt changes

(A) Solid growth oxacillin susceptibility assay for SgrS SRL36 (upper panels) and SRL42 (lower panels) in comparison to synthetic SgrS sRNAs targeting acrA mRNA that deviate by 1 nt in their SRL. A representative replicate is shown.

(B) Susceptibility test for the same strains as in (A) under liquid growth conditions. Oxacillin susceptibility assay was performed in quadruplicate. The Wilcoxon-Mann-Whitney test was used to assess the significance of the difference between growth with and without oxacillin. p-value adjustment was performed using the Benjamini-Hochberg procedure; ∗ = p < 0.05 (non-parametric, unpaired).

An increasing seed region length helps to alleviate the Hfq-dependency of SgrS

The structurally complex sRNA SgrS was reported to depend on Hfq for riboregulation.46 When performing oxacillin susceptibility assays with the SgrS SRL-library in an hfq deletion background, we did not observe oxacillin-sensitive phenotypes on solid media (data not shown), suggesting that the synthetic SgrS sRNAs depend on Hfq for regulation of acrA. In liquid media however, increased sensitivity to oxacillin was observed for some SgrS variants in the Δhfq strain, especially for those with an SRL above ∼50 nt (Figure S8). To further elucidate the impact of an increased SRL on the Hfq-dependency, we applied our fluorescent reporter assay (translational fusion of acrA with syfp2) for selected synthetic SgrS sRNAs. In the wild-type background, repression of the acrA reporter correlated well with the observed oxacillin sensitivity (cf. Figure 5D): variant SRL4 did not repress the reporter fusion, and variants SRL36 and SRL42 showed superior repression in comparison to SRL12, SRL66 and SRL82 (Figure 7A). In the Δhfq background, the pattern changed, and enhanced repression was observed with an increasing SRL. Intriguingly, repression by SRL82 was comparable in both strain backgrounds (Figure 7A). In contrast to RybB, steady-state levels of the synthetic SgrS sRNAs were generally not reduced in the absence of Hfq, and some variants even showed increased levels (Figure 7B). Interestingly, the distinct processing pattern, which was observed for the RybB SRL-library, did not occur for SgrS. Our results indicate that an increasing SRL helps to alleviate the Hfq-dependency of SgrS. Furthermore, the position of the seed region (5′ end versus internal) seems to impact the SRL-dependent processing by RNase E. The higher sensitivity of 5′ seed regions to RNase E-mediated processing might be either caused by an increased proximity of the RNase E cleavage sites to the sRNA 5′ end or by structural changes that make the cleavage sites more accessible. Indeed, when truncating the 5′ stem-loop structure of SgrS to locate the seed region immediately at the 5′ end, some of the tested SRL-variants had a slightly more pronounced processing pattern on northern blots (data not shown).Figure 7 Regulation of acrA by selected synthetic SgrS sRNAs in wild type and Δhfq

(A) Fluorescence of reporter strain acrA-9′-syfp2 is normalized to the control strain, containing empty plasmid pSL009. In the wild type, reduction of fluorescence indicates sRNA functionality and correlates with phenotypes (cf.Figure 5D). In the absence of Hfq, the fluorescence intensity is generally increased. However, the Hfq dependence decreases with increasing SRL, and regulation by SgrS SRL82 is almost comparable between wild type and Δhfq. The control (ctrl) refers to empty plasmid pSL009. A wild type (wt) and Δhfq strain without reporter construct were used as background controls, respectively. Bars represent the mean of three biological replicates and error bars indicate the standard deviation.

(B) The control (ctrl) refers to empty plasmid pSL009. 5S rRNA serves as a loading control. Abundance of SgrS sRNAs is only slightly reduced on average to 83% +/− 44% (RI = relative intensity in Δhfq in comparison to wild type).

Discussion

We established a workflow to benchmark and characterize synthetic sRNAs in high-throughput supported by laboratory automation equipment. We applied our previously developed assay for targeting of acrA mRNA, rendering E. coli cells susceptible toward the β-lactam antibiotic oxacillin,18 in a library-based approach, and focused on the impact of the SRL on sRNA functionality. We combined the phenotypic assay with our recently described rapid and low-cost in-house long-read sequencing-based analysis workflow to connect phenotype and genotype.32 Our presented library approach can rapidly identify synthetic sRNAs with improved functionality. Construction, testing, and identification of optimal regulating sRNAs can be achieved within four to five days using the presented workflow (excluding sequence determination). To understand the mechanistic functions of synthetic sRNAs, this study went beyond the identification of improved synthetic sRNAs and characterized the library to improve our knowledge for future prediction, generation and application of synthetic sRNAs. A logical future iteration of the system could also allow the cloning of a UTR-library in front of the acrA reporter. This would facilitate the design and testing of synthetic sRNA regulators for virtually any gene of interest and sequence context. Furthermore, in combination with efficient barcode-based Nanopore sequencing, large seed- and UTR-libraries could be screened to generate sufficient training data for artificial intelligence design tools. Such a phenotype-based approach would complement fluorescence reporter assays and directly demonstrate sRNA functionality in a biological context.

Based on our results with RybB and SgrS, we conclude that a minimal SRL of 12 nt should be calculated for efficient regulation of target mRNAs. It may be important to avoid AU-rich sequences in short seed sequences to avoid RNase E-mediated processing. A previous study has shown that an sRNA-mRNA interaction length of ≥15 nt was preferable when MicF was used as scaffold.47 However, efficient target repression was also achieved with some of the shorter MicF constructs (interaction length of <15 nt). More importantly, the authors were able to show that the regulation strength saturated at an interaction length of ∼15 nt, and that synthetic sRNAs with an interaction length of ∼60 nt did not perform superior when compared to shorter sRNAs.47 This is in line with our findings that, in a wild-type background, increasing the SRL beyond the minimum of 12 nt does not necessarily improve regulatory performance of RybB. Since RybB and MicF are Hfq-dependent sRNAs, we conclude that synthetic seed regions as short as 12 nt are sufficient for efficient regulation as long as Hfq is present. Interestingly, it was observed for antisense PNAs conjugated to cell-penetrating peptides that 9-mer to 10-mer PNAs are sufficient for repression of target mRNAs.20,48 The slightly shorter minimal length that is needed for PNA-mediated repression might be founded in their higher intracellular stability and enhanced RNA-binding affinity.49 In the case of PNA-peptide conjugates, the PNA length needs to be minimized because it represents a constraint for the uptake into bacterial cells.48 In contrast, the SRL is not a limiting factor for endogenously produced synthetic sRNAs, suspending the need to design the shortest possible seed region.

We recently established a computational tool, SEEDling, for the selection of optimal seed regions based on structural considerations and potential off-target effects,24 but we were lacking information about the optimal SRL. In a first attempt, we applied the tool to RybB sRNAs with 16-nt long seed regions, which is the native SRL of RybB.25 The results of our current study confirm that an SRL of 16 nt is well suited for RybB and probably other Hfq-dependent sRNAs with a 5′ seed region, such as MicA and MicF.18,47 However, for other sRNAs, longer seed regions might be preferable for efficient target regulation. When transferring our SRL-library concept to the more complex sRNA SgrS,46 only two sRNAs (SRL36 and SRL42) showed strong repression of acrA. Hence, an SRL of above 35 nt was needed to achieve a performance similar to synthetic RybB sRNAs. In E. coli, native SgrS regulates the ptsG mRNA by imperfect base-pairing that involves 23 nt of the 31-nt long SgrS seed region.50,51 Even though the SgrS SRL may be minimized to 14 nt without losing ptsG regulation,51 we conclude that SgrS relies on longer seed regions for efficient target regulation than other Hfq-dependent sRNAs.9

The structure of sRNAs is an important aspect to consider in the design process,18,24 which was particularly evident for synthetic SgrS sRNAs. One-nt changes may strongly impede functionality, as observed for SgrS SRL37 and SRL41 in comparison to SRL36 and SRL42, respectively. Based on secondary structure predictions we assume that synthetic SgrS sRNAs are sensitive to sequence changes that increase the local base-pairing probability within the seed region. Hence, in the case of SgrS, the overall structure may be even more important than the actual seed region. We suggest that SgrS seed regions should be designed with the aid of the SEEDling tool to control for structural changes that might affect seed region accessibility and, hence, impede regulation.24

Our data indicate that synthetic sRNAs with a 5′ seed region, such as RybB, are more likely prone to processing by the endoribonuclease RNase E, and that the synthetic seed region sequence may introduce AU-rich motifs serving as cleavage sites. RNase E is known to target single-stranded RNAs at AU-rich sites in different bacteria.41,52,53 Our SRL-library fortuitously provides an increasing number of AU-rich motifs with increasing SRL, generating up to four stable processing products for synthetic RybB sRNAs with an SRL ≥72 nt. However, our data do not indicate a decrease of acrA regulation with the appearance of additional RybB processing products. Processing of the synthetic RybB sRNAs may occur only after binding to the target mRNA due to coupled degradation by RNase E.42 We, therefore, investigated potential alterations of the processing pattern in an acrA deletion strain. However, we did not observe any differences in comparison to the wild type and conclude that processing is at least independent of acrA (data not shown). It remains an open question whether the 5′ triphosphate of the RybB primary transcript needs to be converted into a 5′ monophosphate before processing can occur.13,14 The absence of distinct SgrS processing patterns further suggests that the synthetic SgrS seed regions are less likely to be accessible to RNase E as single-stranded regions, as also supported by our secondary structure predictions. Interestingly, the best performing SgrS sRNAs (SRL36 and SRL42) show at least one stable processing product, while the primary transcript is less abundant in the absence of Hfq. We conclude that these two sRNAs have a more single-stranded seed region, which allows efficient target regulation, but at the same time decreases stability.

With this study, we have extended the standardized Golden Gate basic-part toolbox for cloning of synthetic sRNAs. Golden Gate cloning has the advantage of rapid and reliable DNA assembly of the basic-part collection and is automation compatible.27,54,55 We have standardized our system using SapI as the Type IIS restriction enzyme, which creates 3-nt fusion sites, reducing the size of scars in the process of generating synthetic sRNAs, which is downstream fully compatible with the Modular Cloning standard of the group of Sylvestre Marillonnet.28,29,56 The development of a large standardized parts library for the generation of synthetic sRNAs can be a useful resource for the community in the future, allowing the rapid construction and characterization of synthetic sRNAs for basic and applied research. Larger datasets and the application of various sRNA scaffolds could generate an increased number of data points which may be used for machine learning approaches to predict the optimal synthetic sRNAs for their application in synthetic and engineering biology.

Limitations of the study

Synthetic sRNAs are an important regulatory tool, for example, to control the metabolism of bacteria in biotechnological applications.17 Our library-based study provides new insights into the design of efficient synthetic sRNAs. However, our study focuses solely on E. coli but others have shown it can be adapted to other bacteria and used as orthogonal regulators.57

Although our study addresses RNA processing, it is limited in this aspect. We do not fully understand whether the different northern blot bands that we observe for synthetic RybB sRNAs are solely generated by RNA processing (based on our data presumably catalyzed by RNase E), or whether they are generated by alternative TSS. However, it seems unlikely that up to four additional TSS are provided by the synthetic seed regions with sizes ≥72 nt.

We have focused on two well-characterized scaffolds and made derivatives of them with synthetic seed regions. Therefore, the results may not reflect the global potential of synthetic sRNAs. For example, in several sRNAs, seed sequences are located within the loop regions of stem-loop structures, and the SRL might influence these sRNAs in a different manner than observed here for RybB and SgrS. Furthermore, the seed regions from the SRL-library target the TIR of acrA (i.e., the Shine-Dalgarno sequence and start codon) and are extended toward the acrA 5′ UTR. sRNA-based regulation is indeed most effective when the TIR is targeted,18,47,58,59 but targeting the coding regions (CDR) of mRNAs can provide efficient repression as well.16 An SRL-library that is designed for targeting of the CDR might produce different results than obtained here and will probably complement our conclusions regarding synthetic sRNA design.

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, D.S. (Daniel.Schindler@mpi-marburg.mpg.de).

Materials availability

This study did not generate new unique reagents. Plasmids and strains generated in this study are available from the lead contact on request.

Data and code availability

• All data are available within the study and the supplemental information.

• This study did not generate new unique code.

• Any additional information required to reanalyze the data reported in this study is available from the lead contact upon request.

Acknowledgments

This work was funded by the Max Planck Society in the framework of the MaxGENESYS project (D.S.), the European Union (NextGenerationEU) via the European Regional Development Fund (ERDF) by the state Hesse within the project “biotechnological production of reactive peptides from waste streams as lead structures for drug development” (D.S.), and an Exploration Grant from the 10.13039/501100008454 Boehringer Ingelheim Foundation (B.A.B). J.G. was supported by the 10.13039/501100001659 DFG (GE 3159/1-1 ). We are grateful to all laboratory members for extensive discussions on synthetic sRNAs. M.B. acknowledges that he was awarded the iFZ Masters 2023 Prize of the 10.13039/100009091 Justus Liebig University Giessen for this study. Open access funded by 10.13039/501100004189 Max Planck Society .

Author contributions

B.A.B. and D.S. conceived the study. M.B., T.S.K., S.D., A.A.R.R., B.A.B., and D.S. conducted the experiments. M.B., J.G., B.A.B., and D.S. evaluated and analyzed the data. B.A.B. and D.S. wrote the manuscript with support of M.B. and J.G. All authors read and approved the final manuscript.

Declaration of interests

The authors declare no competing financial interest.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER	
Bacterial and virus strains	
	
E. coli DB3.1	Invitrogen	Cat# 11782-018	
E. coli Top10	Invitrogen	Cat# C4040-10	
	
Chemicals, peptides, and recombinant proteins	
	
1 kb Plus DNA Ladder	New England Biolabs (NEB)	Cat# N3200L	
Deoxynucleotide (dNTP) Solution Set	NEB	Cat# N0446S	
NucleoMag NGS beads	Macherey Nagel	Cat# 744970.50	
Oxacillin sodium salt	Merck	Cat# 28221-5G	
Q5 DNA Polymerase	NEB	Cat# M0491L	
Rubidiumchlorid	Fisher	Cat# 11984191	
SapI	NEB	Cat# R0569L	
Sera-Mag SpeedBeads	Cytiva Healthcare	Cat# 45152105050250	
T4 DNA Ligase	NEB	Cat# M0202L	
T4 Polynucleotide Kinase	NEB	Cat# M0201	
[gamma-P32]ATP	Hartmann Analytic	Cat# SRP-301	
	
Experimental models: organisms/strain	
	
See Table S1 for Escherichia coli MG1655 K-12 strains used in this study	N/A	N/A	
	
Oligonucleotides	
	
Oligonucleotides and oligo pools are listed in Tables S2 and S3.	IDT	N/A	
	
Recombinant DNA	
	
Plasmids are listed in Table S4 and sequences are provided as GenBank files in Data S1.	N/A	N/A	
	
Software and algorithms	
	
DuBa.flow	Ramírez Rojas et al.32	https://github.com/RGSchindler/DuBA.flow	
Growthcurver	Sprouffske et al.60	https://github.com/cran/growthcurver	
Guppy (version 6.0.1)	Oxford Nanopore Technologies	https://nanoporetech.com/	
IGV viewer (version 2.8.10)	Robinson et al.61	https://software.broadinstitute.org/software/igv/	
Quantity One 1-D Analysis Software (version 4.6.6)	Bio-Rad	https://www.bio-rad.com/	
R
RStudio	R Core Team
Posit, PBC	https://www.r-project.org/
https://www.posit.co/	
RNAfold
ViennaRNA Package 2.0	Lorenz et al.45	http://rna.tbi.univie.ac.at//cgi-bin/RNAWebSuite/RNAfold.cgi	
Snapgene	GSL Biotech LLC	http://www.snapgene.com/	

Experimental model and study participant details

A full list of E. coli K-12 wild type MG1655 and its derivatives used may be found in Table S1. MG1655-derived strains were used for physiological experiments. If not stated otherwise, strains were cultivated in LB medium (1% (w/v) bacto-tryptone, 1% (w/v) NaCl, 0.5% (w/v) yeast extract) in Erlenmeyer flasks at 37°C under continuous shaking at 180 rpm. If appropriate, antibiotics were present in the following concentrations: 100 μg/mL ampicillin (Amp), 50 μg/mL kanamycin (Kan), 25 μg/mL chloramphenicol (Cm), 120 μg/mL spectinomycin (Spec), and 10 μg/mL tetracycline (Tet).

Method details

Oligodeoxynucleotides

All oligodeoxynucleotides were ordered from Integrated DNA Technologies (IDT, Coralville, USA) or Microsynth Seqlab (Göttingen, Germany) with standard desalting purification (Table S2). The oligonucleotide pool for the SRL library was ordered from Integrated DNA Technologies (IDT, Coralville, USA) in the form of an oPools oligo pool containing 42 distinct oligonucleotides (SLop2.1 to SLop2.42) at 50 pmol scale (Table S3). The primer used for the barcoding procedure for Nanopore sequencing were ordered in 96-well plates from Integrated DNA Technologies (IDT, Coralville, USA) with standard desalting purification.32

PCR purification and plasmid extraction

If not indicated otherwise PCR amplicons and plasmid extractions were performed according to the open source magnetic bead procedure published by Oberacker et al. 2019 using SeraMag Speed Beads (Cytiva, Marlborough, USA).62 For plasmid extractions cultures were grown in SMM medium (16 g/L tryptone, 10 g/L yeast extract, 5 g/L glycerol, and 1x M9 salts) with the appropriate antibiotic in 2–3 mL volume in tubes or 1 mL in 96-well deep-well plates.18 Deep-well plates were inoculated overnight in a Multitron HT shaker (Infors AG, Bottmingen, Switzerland) at 800 rpm and 80% humidity at 37°C with a gas permeable seal (AB-0718, Thermo Scientific Nunc, Waltham, USA).

Strain construction

Chromosomal manipulations were performed by λ red recombineering as described previously.18 Briefly, for gene deletions, the chloramphenicol acetyltransferase (cat) gene was amplified to replace the gene(s) of interest. Chromosomal constructs were verified by diagnostic PCR. All relevant primer pairs for λ red recombineering are provided in Table S2. Constructs were transduced to other strain backgrounds using P1 phages according to standard procedures.63 Transductants were selected with chloramphenicol and verified by diagnostic PCR as before. If needed, resistance cassettes were removed by FRT-mediated flipping using plasmid 709-FLPe (Gene Bridges GmbH, Heidelberg, Germany) according to the manufacturer’s instructions.

Generation of plasmids

Plasmids were mostly generated by Golden Gate cloning (all sequences are provided as GenBank files in Data S1). The generation of basic Golden Gate parts and the library construction is described in the following sections. Golden Gate assembly reactions were prepared with equimolar amounts for each DNA part and acceptor vector (20–40 fmol). Reactions were carried out in a thermocycler in 10–20 μL total volume with 1 μL of SapI (NEB, Ipswich, USA) and 1 μL of T4 DNA ligase (NEB, Ipswich, USA) in 1x T4 DNA ligase buffer (NEB, Ipswich, USA) and incubated with the following program: 50 cycles of 4 min at 16°C, 3 min at 37°C followed by 10 min at 50°C and 10 min at 80°C, and storage at 4°C. 10 μL of Golden Gate reactions were transformed into RbCl competent cells according to the standard procedure18,64 and candidates were selected by overnight cultivation on appropriate LB agar plates. A complete list of plasmids is provided in Table S4.

Level 0 part generation

Level 0 parts were constructed by blunt end ligation as described before29 and all resulting plasmid files are provided as GenBank files in Data S1. Briefly, the primer for the PCR amplicons contained the corresponding overhangs and either the oligonucleotides or the obtained amplicon was phosphorylated using the T4 polynucleotide kinase (PNK) (NEB, Ipswich, USA). The level 0 acceptor plasmid was amplified with the primer pair SLo0765/SLo0766 using Q5 DNA Polymerase (NEB, Ipswich, USA) followed by a DpnI (NEB, Ipswich, USA) digest prior ligation. The phosphorylated amplicon and the PCR amplified vector were ligated using T4 ligase for 2 h at room temperature in a 4:1 ratio. Transformation was performed into in house generated RbCl competent E. coli Top10 cells.18,64 Selection was performed overnight on LB spectinomycin plates at 37°C. Level 0 parts were identified by colony PCR, followed by plasmid extraction and subsequent validation by external Sanger sequencing services (Microsynth, Göttingen, Germany).

Library generation

Double stranded DNA was generated from the oligonucleotide pool by PCR using the primer SLo1505/1506 and Q5 DNA Polymerase (NEB, Ipswich, USA) with the following settings: Initial denaturation 0:20 min 98°C, followed by 35 PCR cycles [0:20 min 98°C, 0:20 min 62°C, 0:10 min 72°C] and final storage at 8°C. One fmol of the oligonucleotide pool was used in the reaction as a template. The RybB and SgrS plasmid libraries were generated by standard Golden Gate assembly combining the purified amplicon-library with pSL135 and pSL123 (for RybB) or pSL133 and pSL132 (for SgrS) into pSL137. The Golden Gate reaction was transformed into in house generated RbCl competent E. coli K-12 MG1655 cells.18,64 The cell suspension was plated in dilution steps onto multiple 120 × 120 mm square petri dishes and incubated until single colonies appeared. 380 single colonies were picked onto 384-well microtiter plates with LB containing kanamycin utilizing the PIXL colony picking robot (Singer Instruments, Somerset, UK). Microtiter plates were incubated overnight on a Multitron HT (Infors AG, Bottmingen, Switzerland) incubator at 37°C with 500 rpm with 3 mm shaking and 80% relative humidity, only covered with the lid. The plates were used to inoculate liquid cultures in 384 well format for cryo-preservation and initial screening after overnight incubation with a Rotor HDA+ screening robot (Singer Instruments, Somerset, UK).

Phenotypic screening and analysis of generated libraries

The initial screening of the 380 candidate sized library and 4 controls was done using overnight cultures grown on a Multitron HT (Infors AG, Bottmingen, Switzerland) at 500 rpm with 3 mm shaking, 80% humidity at 37°C. A Rotor HDA+ screening robot was used to spot the library into a 3 × 3 grid using 384-pin pads onto LB kanamycin media with varying oxacillin concentrations (0–200 μg/mL). Data was documented with the PhenoBooth using default settings (Singer Instruments, Somerset, UK). The determined length library for RybB and SgrS of constitutively expressed synthetic sRNA constructs was performed based on manual dilution series of indicated overnight cultures generated in 96-well or 384-well flat bottom microtiter plates using multichannel pipettes. A Rotor HDA+ screening robot (Singer Instruments, Somerset, UK) was afterward used to transfer droplets onto solid agar plates containing LB kanamycin, and the indicated oxacillin concentrations. Programs either transfer 49 droplets for each dilution into a 7 × 7 grid (96-well source plates) or 9 droplets for each dilution into a 3 × 3 grid (384-well source plates), using 96-pin pads and 384-pin pads, respectively. After each droplet transfer the source plate was revisited to ensure sufficient material transfer. Plates were incubated at 37°C overnight, and images were taken with the PhenoBooth plate documentation system using default settings (Singer Instruments, Somerset, UK).

Phenotypic screening in liquid media was performed in CLARIOstar plate readers (BMG Labtech, Ortenberg, Germany) equipped with plate holders conceptualized for kinetic measurements with constant shaking. Briefly, E. coli precultures were inoculated in microtiter plates (polystyrene, flat bottom; Greiner Bio-One, Kremsmünster, Austria) overnight at 37°C, 1000 rpm with 75% relative humidity in 100 μL of LB medium containing kanamycin on an Multitron HT incubator (Infors AG, Bottmingen, Switzerland). Assay cultures were inoculated from the overnight cultures with the Rotor HDA+ screening robot (Singer Instruments, Somerset, UK) into 96-well microtiter plates containing 200 μL of the respective LB medium supplemented with kanamycin and indicated oxacillin concentrations (0–100 μg/mL). Plates were sealed with optical clear sealing film with a PlateLoc plate sealer (Agilent, Santa Clara, USA). Measurements were taken with the following program: 120 s orbital shaking (1 mm amplitude) and 120 s linear shaking (1 mm amplitude), followed by wavelength measurement at 600 nm for 24 h. The program was compared to our previously described method using TopSeal-A PLUS (PerkinElmer, Waltham, USA) and Infinite M Nano+ plate reader (Tecan, Männedorf, Switzerland),18 no differences were observed. Data was processed with the Growthcurver R package60 and visualized with custom R scripts using default settings. AUCs were extracted from the Growthcurver data output based on data points from t = 0 to t = 8 h for maximum dynamic range to compare growth.

Nanopore sequencing and analysis of generated libraries

The procedure was performed according to Ramírez-Rojas et al.32 Briefly, amplicons were directly generated from the obtained colonies by a two-step PCR procedure (for details see Ramírez-Rojas et al.32,33). The first step attached universal sequences which are used in the second step as binding sequence for the barcoding primers. The amplified, dual barcoded amplicons of the RybB and SgrS candidates were pooled and purified with two other, similar libraries resulting in 1536 barcoded samples. One microgram of the purified library was used as input for the ligation sequencing kit SQK-LSK109 (Oxford Nanopore Technologies). The procedure was carried out according to the manufacturer guidelines and the library was sequenced on two Flongle flow cells FLO-FLG001 (R9.4.1) generating a total output of 0.9 Gb. Basecalling of the raw data was performed using ONT Guppy basecalling software (version 6.0.1) in GPU mode with standard settings. The amplicons were demultiplexed using the standard settings of MiniBar,65 subsequently mapped against the references (RybB or SgrS with SRL82) with minimap2 using standard settings.66 Resulting sam files were processed into indexed bam files using standard settings in Samtools67 and visually inspected using Integrative Genomics Viewer.61 The procedure was later developed in an automated analysis pipeline described in Ramírez Rojas et al.32,33

Northern blot analysis

For northern blot analysis of synthetic sRNAs, expression plasmids were transformed to the indicated strain backgrounds. Total RNA was extracted from exponential-phase cultures (OD600 of approx. 0.4) using the hot acid-phenol method as described.68 Northern blot analysis was performed according to a standard protocol.69 In short, 10% polyacrylamide gels containing 1x TBE and 7 M urea were used for analysis of sRNAs. After gel separation, RNA was transferred to nylon membranes by semi-dry electroblotting. Oligodeoxynucleotides (Table S2) for detection of specific RNA species were end-labeled with [γ-32P]-ATP using T4 polynucleotide kinase (PNK) (NEB, Ipswich, USA) according to manufacturer guidelines. Prehybridization and hybridization were performed in Church buffer [0.5 M phosphate buffer (pH 7.2), 1% (w/v) bovine serum albumin, 1 mM EDTA, and 7% (w/v) SDS] at 42°C. After washing of membranes, phosphorimaging was applied for visualization using a Molecular Imager FX (Bio-Rad, Hercules, CA, USA). Signal intensities were quantified using the Quantity One 1-D Analysis Software (Version 4.6.6) (Bio-Rad, Hercules, CA, USA). sRNA signal intensities were normalized to the corresponding 5S rRNA signals.

sRNA binding affinity prediction

The interaction energy for the acrA mRNA and the sRNA variants was calculated with IntaRNA26 version 3.2.0 using the Vienna RNA package 2.4.14 and the “Turner 2004” energy model. IntaRNA was used in the heuristic mode (H) with the seed-extension strategy (model X) with lonely base pairs and GUs at helix ends allowed.

sRNA structure prediction using RNAfold

RNA structures were predicted using the RNAfold WebServer (http://rna.tbi.univie.ac.at//cgi-bin/RNAWebSuite/RNAfold.cgi).45 Predictions were performed with standard settings using the minimum free energy (MFE) and partition function, isolated base pairs were avoided and the “Turner 2004” Energy Parameters were applied with standard values.

Measurement of sYFP2 fluorescence

Chromosomal syfp2 reporter strains, containing sRNA expression plasmids, were used for sYFP2 measurements as described previously.18,25 Briefly, stationary-phase cultures were diluted 100-fold into LB medium, and 150 μL was loaded into wells of a transparent 96-well plate (polystyrene, flat bottom; Greiner Bio-One, Kremsmünster, Austria). Cells were cultivated in an Infinite M Nano+ microplate reader (Tecan, Männedorf, Switzerland) at 37°C and orbital shaking with an amplitude of 3.5 mm. The optical density was measured at 600 nm (OD600), and the sYFP2 fluorescence was monitored using excitation and emission wavelengths of 510 and 540 nm, respectively. The gain was set to 100. Measurements from wells containing pure LB medium were used for background correction of OD600 and sYFP2 values. Finally, sYFP2 fluorescence values were normalized to the corresponding OD600. Three biological replicates were obtained for each strain.

Quantification and statistical analysis

Obtained data was processed and visualized using RStudio and the ggplot package70 using standard functions. Statistical tests to evaluate growth differences (AUC values) under defined treatment conditions (presence or absence of screening antibiotic) were performed using the non-parametric Wilcoxon-Mann-Whitney test, as normal distributions could not be identified. Wilcoxon-Mann-Whitney-Test was performed using the R package “stats“ with the function wilcox.test.71 p-values were adjusted using the Benjamini-Hochberg procedure using the function p.adjust of the R package “stats“. A p-value equal to or less than 0.05 (or as indicated) was used to determine significance.

Supplemental information

Document S1. Figures S1–S8 and Tables S1–S4

Data S1. GenBank sequences for all generated plasmids, related to STAR Methods

Supplemental information can be found online at https://doi.org/10.1016/j.isci.2024.110774.
==== Refs
References

1 Hör J. Matera G. Vogel J. Gottesman S. Storz G. Trans-acting small RNAs and their effects on gene expression in Escherichia coli and Salmonella enterica EcoSal Plus 9 2020 10 1128 10.1128/ecosalplus.ESP-0030-2019
2 Holmqvist E. Wagner E.G.H. Impact of bacterial sRNAs in stress responses Biochem. Soc. Trans. 45 2017 1203 1212 10.1042/BST20160363 29101308
3 Wagner E.G.H. Romby P. Small RNAs in bacteria and archaea: Who they are, what they do, and how they do it Adv. Genet. 90 2015 133 208 10.1016/bs.adgen.2015.05.001 26296935
4 Frohlich K.S. Vogel J. Activation of gene expression by small RNA Curr. Opin. Microbiol. 12 2009 674 682 10.1016/j.mib.2009.09.009 19880344
5 Vogel J. Luisi B.F. Hfq and its constellation of RNA Nat. Rev. Microbiol. 9 2011 578 589 10.1038/nrmicro2615 21760622
6 Schu D.J. Zhang A. Gottesman S. Storz G. Alternative Hfq-sRNA interaction modes dictate alternative mRNA recognition EMBO J. 34 2015 2557 2573 10.15252/embj.201591569 26373314
7 Sauer E. Schmidt S. Weichenrieder O. Small RNA binding to the lateral surface of Hfq hexamers and structural rearrangements upon mRNA target recognition Proc Natl Acad Sci USA 109 2012 9396 9401 10.1073/pnas.1202521109 22645344
8 Updegrove T.B. Zhang A. Storz G. Hfq: The flexible RNA matchmaker Curr. Opin. Microbiol. 30 2016 133 138 10.1016/j.mib.2016.02.003 26907610
9 Santiago-Frangos A. Woodson S.A. Hfq chaperone brings speed dating to bacterial sRNA Wiley Interdiscip. Rev. RNA 9 2018 e1475 10.1002/wrna.1475
10 Wagner E.G.H. Cycling of RNAs on Hfq RNA Biol. 10 2013 619 626 10.4161/rna.24044 23466677
11 Hui M.P. Foley P.L. Belasco J.G. Messenger RNA degradation in bacterial cells Annu. Rev. Genet. 48 2014 537 559 10.1146/annurev-genet-120213-092340 25292357
12 Bandyra K.J. Luisi B.F. RNase E and the high-fidelity orchestration of RNA metabolism Microbiol. Spectr. 6 2018 10 1128 10.1128/microbiolspec.RWR-0008-2017
13 Schilder A. Gorke B. Role of the 5' end phosphorylation state for small RNA stability and target RNA regulation in bacteria Nucleic Acids Res. 51 2023 5125 5143 10.1093/nar/gkad226 36987877
14 Bandyra K.J. Said N. Pfeiffer V. Górna M.W. Vogel J. Luisi B.F. The seed region of a small RNA drives the controlled destruction of the target mRNA by the endoribonuclease RNase E Mol. Cell 47 2012 943 953 10.1016/j.molcel.2012.07.015 22902561
15 Prevost K. Desnoyers G. Jacques J.F. Lavoie F. Masse E. Small RNA-induced mRNA degradation achieved through both translation block and activated cleavage Genes Dev. 25 2011 385 396 10.1101/gad.2001711 21289064
16 Pfeiffer V. Papenfort K. Lucchini S. Hinton J.C.D. Vogel J. Coding sequence targeting by MicC RNA reveals bacterial mRNA silencing downstream of translational initiation Nat. Struct. Mol. Biol. 16 2009 840 846 10.1038/nsmb.1631 19620966
17 Na D. Yoo S.M. Chung H. Park H. Park J.H. Lee S.Y. Metabolic engineering of Escherichia coli using synthetic small regulatory RNAs Nat. Biotechnol. 31 2013 170 174 10.1038/nbt.2461 23334451
18 Köbel T.S. Melo Palhares R. Fromm C. Szymanski W. Angelidou G. Glatter T. Georg J. Berghoff B.A. Schindler D. An easy-to-use plasmid toolset for efficient generation and benchmarking of synthetic small RNAs in bacteria ACS Synth. Biol. 11 2022 2989 3003 10.1021/acssynbio.2c00164 36044590
19 Popella L. Jung J. Popova K. Ðurica-Mitić S. Barquist L. Vogel J. Global RNA profiles show target selectivity and physiological effects of peptide-delivered antisense antibiotics Nucleic Acids Res. 49 2021 4705 4724 10.1093/nar/gkab242 33849070
20 Popella L. Jung J. Do P.T. Hayward R.J. Barquist L. Vogel J. Comprehensive analysis of PNA-based antisense antibiotics targeting various essential genes in uropathogenic Escherichia coli Nucleic Acids Res. 50 2022 6435 6452 10.1093/nar/gkac362 35687096
21 Oh E. Zhang Q. Jeon B. Target optimization for peptide nucleic acid (PNA)-mediated antisense inhibition of the CmeABC multidrug efflux pump in Campylobacter jejuni J. Antimicrob. Chemother. 69 2014 375 380 10.1093/jac/dkt381 24084637
22 Vogel J. An RNA biology perspective on species-specific programmable RNA antibiotics Mol. Microbiol. 113 2020 550 559 10.1111/mmi.14476 32185839
23 Jung J. Popella L. Do P.T. Pfau P. Vogel J. Barquist L. Design and off-target prediction for antisense oligomers targeting bacterial mRNAs with the MASON web server RNA 29 2023 570 583 10.1261/rna.079263.122 36750372
24 Philipp N. Brinkmann C.K. Georg J. Schindler D. Berghoff B.A. DIGGER-Bac: Prediction of seed regions for high-fidelity construction of synthetic small RNAs in bacteria Bioinformatics 39 2023 btad285 10.1093/bioinformatics/btad285
25 Brück M. Berghoff B.A. Schindler D. In silico design, in vitro construction, and in vivo application of synthetic small regulatory RNAs in bacteria Methods Mol. Biol. 2760 2024 479 507 10.1007/978-1-0716-3658-9_27 38468105
26 Mann M. Wright P.R. Backofen R. IntaRNA 2.0: Enhanced and customizable prediction of RNA-RNA interactions Nucleic Acids Res. 45 2017 W435 W439 10.1093/nar/gkx279 28472523
27 Weber E. Engler C. Gruetzner R. Werner S. Marillonnet S. A modular cloning system for standardized assembly of multigene constructs PLoS One 6 2011 e16765 10.1371/journal.pone.0016765
28 de Vries S.T. Köbel T.S. Sanal A. Schindler D. In- & Out-Cloning: Plasmid toolboxes for scarless transcription unit and modular Golden Gate acceptor plasmid assembly Preprint at bioRxiv 2024 10.1101/2024.06.22.600171
29 de Vries S.T. Kley L. Schindler D. Use of a Golden Gate plasmid set enabling scarless MoClo-compatible transcription unit assembly Preprint at arXiv 2024 10.48550/arXiv.2402.03410
30 Papenfort K. Bouvier M. Mika F. Sharma C.M. Vogel J. Evidence for an autonomous 5' target recognition domain in an Hfq-associated small RNA Proc. Natl. Acad. Sci. USA 107 2010 20435 20440 10.1073/pnas.1009784107 21059903
31 Eguchi Y. Oshima T. Mori H. Aono R. Yamamoto K. Ishihama A. Utsumi R. Transcriptional regulation of drug efflux genes by EvgAS, a two-component system in Escherichia coli Microbiology (Read.) 149 2003 2819 2828 10.1099/mic.0.26460-0
32 Ramírez Rojas A.A. Brinkmann C.K. Köbel T.S. Schindler D. DuBA.flow - A low-cost, long-read amplicon sequencing workflow for the validation of synthetic DNA constructs ACS Synth. Biol. 13 2024 457 465 10.1021/acssynbio.3c00522 38295293
33 Ramírez Rojas A.A. Brinkmann C.K. Schindler D. Validation of Golden Gate assemblies using highly multiplexed Nanopore amplicon sequencing Preprint at arXiv 2024 10.48550/arXiv.2401.14191
34 Vogel J. Argaman L. Wagner E.G.H. Altuvia S. The small RNA IstR inhibits synthesis of an SOS-induced toxic peptide Curr. Biol. 14 2004 2271 2276 10.1016/j.cub.2004.12.003 15620655
35 Sauer E. Weichenrieder O. Structural basis for RNA 3'-end recognition by Hfq Proc. Natl. Acad. Sci. USA 108 2011 13065 13070 10.1073/pnas.1103420108 21737752
36 Thompson K.M. Rhodius V.A. Gottesman S. SigmaE regulates and is regulated by a small RNA in Escherichia coli J. Bacteriol. 189 2007 4243 4256 10.1128/JB.00020-07 17416652
37 Bechhofer D.H. Deutscher M.P. Bacterial ribonucleases and their roles in RNA metabolism Crit. Rev. Biochem. Mol. Biol. 54 2019 242 300 10.1080/10409238.2019.1651816 31464530
38 Robertson H.D. Dunn J.J. Ribonucleic acid processing activity of Escherichia coli ribonuclease III J. Biol. Chem. 250 1975 3050 3056 1091644
39 Court D.L. Gan J. Liang Y.H. Shaw G.X. Tropea J.E. Costantino N. Waugh D.S. Ji X. RNase III: Genetics and function; structure and mechanism Annu. Rev. Genet. 47 2013 405 431 10.1146/annurev-genet-110711-155618 24274754
40 Viegas S.C. Pfeiffer V. Sittka A. Silva I.J. Vogel J. Arraiano C.M. Characterization of the role of ribonucleases in Salmonella small RNA decay Nucleic Acids Res. 35 2007 7651 7664 10.1093/nar/gkm916 17982174
41 Chao Y. Li L. Girodat D. Förstner K.U. Said N. Corcoran C. Śmiga M. Papenfort K. Reinhardt R. Wieden H.J. In vivo cleavage map illuminates the central role of RNase E in coding and non-coding RNA pathways Mol. Cell 65 2017 39 51 10.1016/j.molcel.2016.11.002 28061332
42 Massé E. Escorcia F.E. Gottesman S. Coupled degradation of a small regulatory RNA and its mRNA targets in Escherichia coli Genes Dev. 17 2003 2374 2383 10.1101/gad.1127103 12975324
43 Quendera A.P. Seixas A.F. Dos Santos R.F. Santos I. Silva J.P.N. Arraiano C.M. Andrade J.M. RNA-binding proteins driving the regulatory activity of small non-coding RNAs in bacteria Front. Mol. Biosci. 7 2020 78 10.3389/fmolb.2020.00078 32478092
44 The RNAcentral Consortium RNAcentral: A hub of information for non-coding RNA sequences Nucleic Acids Res. 47 2019 D221 D229 10.1093/nar/gky1034 30395267
45 Lorenz R. Bernhart S.H. Honer Zu Siederdissen C. Tafer H. Flamm C. Stadler P.F. Hofacker I.L. ViennaRNA package 2.0 Algorithm Mol. Biol. 6 2011 26 10.1186/1748-7188-6-26
46 Wadler C.S. Vanderpool C.K. A dual function for a bacterial small RNA: SgrS performs base pairing-dependent regulation and encodes a functional polypeptide Proc. Natl. Acad. Sci. USA 104 2007 20454 20459 10.1073/pnas.0708102104 18042713
47 Hoynes-O'Connor A. Moon T.S. Development of design rules for reliable antisense RNA behavior in E. coli ACS Synth. Biol. 5 2016 1441 1454 10.1021/acssynbio.6b00036 27434774
48 Goltermann L. Yavari N. Zhang M. Ghosal A. Nielsen P.E. PNA length restriction of antibacterial activity of peptide-PNA conjugates in Escherichia coli through effects of the inner membrane Front. Microbiol. 10 2019 1032 10.3389/fmicb.2019.01032 31178830
49 Pellestor F. Paulasova P. The peptide nucleic acids (PNAs), powerful tools for molecular genetics and cytogenetics Eur. J. Hum. Genet. 12 2004 694 700 10.1038/sj.ejhg.5201226 15213706
50 Vanderpool C.K. Gottesman S. Involvement of a novel transcriptional activator and small RNA in post-transcriptional regulation of the glucose phosphoenolpyruvate phosphotransferase system Mol. Microbiol. 54 2004 1076 1089 10.1111/j.1365-2958.2004.04348.x 15522088
51 Maki K. Morita T. Otaka H. Aiba H. A minimal base-pairing region of a bacterial small RNA SgrS required for translational repression of ptsG mRNA Mol. Microbiol. 76 2010 782 792 10.1111/j.1365-2958.2010.07141.x 20345651
52 McDowall K.J. Lin-Chao S. Cohen S.N. A+U content rather than a particular nucleotide order determines the specificity of RNase E cleavage J. Biol. Chem. 269 1994 10790 10796 7511606
53 Förstner K.U. Reuscher C.M. Haberzettl K. Weber L. Klug G. RNase E cleavage shapes the transcriptome of Rhodobacter sphaeroides and strongly impacts phototrophic growth Life Sci. Alliance 1 2018 e201800080 10.26508/lsa.201800080
54 Engler C. Kandzia R. Marillonnet S. A one pot, one step, precision cloning method with high throughput capability PLoS One 3 2008 e3647 10.1371/journal.pone.0003647
55 Köbel T. Schindler D. Automation and miniaturization of Golden Gate DNA assembly reactions using acoustic dispensers Preprint at arXiv 2023 10.48550/arXiv.2310.00325
56 Taylor G.M. Mordaka P.M. Heap J.T. Start-Stop Assembly: A functionally scarless DNA assembly system optimized for metabolic engineering Nucleic Acids Res. 47 2019 e17 10.1093/nar/gky1182
57 Cho J.S. Yang D. Prabowo C.P.S. Ghiffary M.R. Han T. Choi K.R. Moon C.W. Zhou H. Ryu J.Y. Kim H.U. Lee S.Y. Targeted and high-throughput gene knockdown in diverse bacteria using synthetic sRNAs Nat. Commun. 14 2023 2359 10.1038/s41467-023-38119-y 37095132
58 Yoo S.M. Na D. Lee S.Y. Design and use of synthetic regulatory small RNAs to control gene expression in Escherichia coli Nat. Protoc. 8 2013 1694 1707 10.1038/nprot.2013.105 23928502
59 Bouvier M. Sharma C.M. Mika F. Nierhaus K.H. Vogel J. Small RNA binding to 5' mRNA coding region inhibits translational initiation Mol. Cell 32 2008 827 837 10.1016/j.molcel.2008.10.027 19111662
60 Sprouffske K. Wagner A. Growthcurver: An R package for obtaining interpretable metrics from microbial growth curves BMC Bioinf. 17 2016 172 10.1186/s12859-016-1016-7
61 Robinson J.T. Thorvaldsdóttir H. Winckler W. Guttman M. Lander E.S. Getz G. Mesirov J.P. Integrative Genomics Viewer Nat. Biotechnol. 29 2011 24 26 10.1038/nbt.1754 21221095
62 Oberacker P. Stepper P. Bond D.M. Höhn S. Focken J. Meyer V. Schelle L. Sugrue V.J. Jeunen G.J. Moser T. Bio-On-Magnetic-Beads (BOMB): Open platform for high-throughput nucleic acid extraction and manipulation PLoS Biol. 17 2019 e3000107 10.1371/journal.pbio.3000107
63 Lennox E.S. Transduction of linked genetic characters of the host by bacteriophage P1 Virology 1 1955 190 206 10.1016/0042-6822(55)90016-7 13267987
64 Hanahan D. Studies on transformation of Escherichia coli with plasmids J. Mol. Biol. 166 1983 557 580 10.1016/s0022-2836(83)80284-8 6345791
65 Krehenwinkel H. Pomerantz A. Henderson J.B. Kennedy S.R. Lim J.Y. Swamy V. Shoobridge J.D. Graham N. Patel N.H. Gillespie R.G. Prost S. Nanopore sequencing of long ribosomal DNA amplicons enables portable and simple biodiversity assessments with high phylogenetic resolution across broad taxonomic scale GigaScience 8 2019 giz006 10.1093/gigascience/giz006
66 Li H. Minimap2: Pairwise alignment for nucleotide sequences Bioinformatics 34 2018 3094 3100 10.1093/bioinformatics/bty191 29750242
67 Danecek P. Bonfield J.K. Liddle J. Marshall J. Ohan V. Pollard M.O. Whitwham A. Keane T. McCarthy S.A. Davies R.M. Li H. Twelve years of SAMtools and BCFtools GigaScience 10 2021 giab008 10.1093/gigascience/giab008
68 Berghoff B.A. Karlsson T. Källman T. Wagner E.G.H. Grabherr M.G. RNA-sequence data normalization through in silico prediction of reference genes: The bacterial response to DNA damage as case study BioData Min. 10 2017 30 10.1186/s13040-017-0150-8 28878825
69 Edelmann D. Berghoff B.A. Type I toxin-dependent generation of superoxide affects the persister life cycle of Escherichia coli Sci. Rep. 9 2019 14256 10.1038/s41598-019-50668-1
70 Wickham H. ggplot2: Elegant Graphics for Data Analysis 2016 Springer-Verlag
71 R Core Team R: A Language and Environment for Statistical Computing 2023 Foundation for Statistical Computing
