An Assay for Quantifying Protein-RNA Binding in Bacteria

Summary

In this method, we quantify the binding affinity of RNA binding proteins (RBPs) to cognate and non-cognate binding sites using a simple, live, reporter assay in bacterial cells. The assay is based on repression of a reporter gene.

Abstract

In the initiation step of protein translation, the ribosome binds to the initiation region of the mRNA. Translation initiation can be blocked by binding of an RNA binding protein (RBP) to the initiation region of the mRNA, which interferes with ribosome binding. In the presented method, we utilize this blocking phenomenon to quantify the binding affinity of RBPs to their cognate and non-cognate binding sites. To do this, we insert a test binding site in the initiation region of a reporter mRNA and induce the expression of the test RBP. In the case of RBP-RNA binding, we observed a sigmoidal repression of the reporter expression as a function of RBP concentration. In the case of no-affinity or very low affinity between binding site and RBP, no significant repression was observed. The method is carried out in live bacterial cells, and does not require expensive or sophisticated machinery. It is useful for quantifying and comparing between the binding affinities of different RBPs that are functional in bacteria to a set of designed binding sites. This method may be inappropriate for binding sites with high structural complexity. This is due to the possibility of repression of ribosomal initiation by complex mRNA structure in the absence of RBP, which would result in lower basal reporter gene expression, and thus less-observable reporter repression upon RBP binding.

Introduction

RNA binding protein (RBP)-based post-transcriptional regulation, specifically characterization of the interaction between RBPs and RNA, has been studied extensively in recent decades. There are multiple examples of translational down-regulation in bacteria originating from RBPs inhibiting, or directly competing with, ribosome binding^1,2,3. In the field of synthetic biology, RBP-RNA interactions are emerging as a significant tool for the design of transcription-based genetic circuits^4,5. Therefore, there is an increase in demand for characterization of such RBP-RNA interactions in a cellular context.

The most common methods for studying protein-RNA interactions are the electrophoretic mobility shift assay (EMSA)⁶, which is limited to in vitro settings, and various pull-down assays⁷, including the CLIP method^8,9. While such methods enable the discovery of de novo RNA binding sites, they suffer from drawbacks such as labor-intensive protocols and expensive deep sequencing reactions and may require a specific antibody for RBP pull-down. Due to the susceptible nature of RNA to its environment, many factors can affect RBP-RNA interactions, emphasizing the importance of interrogating RBP-RNA binding in the cellular context. For example, we and others have demonstrated significant differences between RNA structures in vivo and in vitro^10,11.

Based on the approach of a previous study¹², we recently demonstrated¹⁰ that when placing pre-designed binding sites for the capsid RBPs from the bacteriophages GA¹³, MS2¹⁴, PP7¹⁵, and Qβ¹⁶ in the translation initiation region of a reporter mRNA, reporter expression is strongly repressed. We present a relatively simple and quantitative method, based on this repression phenomenon, to measure the affinity between RBPs and their corresponding RNA binding sites in vivo.

Protocol

1. System Preparation

1. Design of binding-site plasmids
   1. Design the binding site cassette as depicted in Figure 1. Each minigene contains the following parts (5' to 3'): Eagl restriction site, ∼40 bases of the 5' end of the kanamycin (Kan) resistance gene, pLac-Ara promoter, ribosome binding site (RBS), AUG of the mCherry gene, a spacer (δ), an RBP binding site, 80 bases of the 5' end of the mCherry gene, and an ApaLI restriction site.
      NOTE: To increase the success rate of the assay, design three binding-site cassettes for each binding site, with spacers consisting of at least one, two, and three bases. See Representative Results section for further guidelines.
2. Cloning of binding site plasmids
   1. Order the binding-site cassettes as double-stranded DNA (dsDNA) minigenes. Each minigene is ∼500 bp long and contains an Eagl restriction site and an ApaLI restriction site at the 5' and 3' ends, respectively (see step 1.1.1).
      NOTE: In this experiment, mini-genes with half of the kanamycin gene were ordered to facilitate screening for positive colonies. However, Gibson assembly¹⁷ is also suitable here, in which case the binding site can be ordered as two shorter complementary single-stranded DNA oligos.
   2. Double-digest both the mini-genes and the target vector with Eagl-HF and ApaLI by the restriction protocol¹⁸, and column purify¹⁹.
   3. Ligate the digested minigenes to the binding-site backbone containing the rest of the mCherry reporter gene, terminator, and a kanamycin resistance gene²⁰.
   4. Transform the ligation solution into Escherichia coli TOP10 cells²¹.
   5. Identify positive transformants via Sanger sequencing.
      1. Design a primer 100 bases upstream to the region of interest (see Table 1 for primer sequences).
      2. Miniprep a few bacterial colonies²².
      3. Prepare 5 µL of a 5 mM solution of the primer and 10 µL of the DNA at 80 ng/µL concentration.
      4. Send the two solution to a convenient facility for Sanger sequencing²³.
   6. Store purified plasmids at -20 °C, and bacterial strains as glycerol stocks²⁴, both in the 96-well format. DNA will then be used for transformation into E. coli TOP10 cells containing one of four fusion-RBP plasmids (see step 1.3.5).
3. Design and construction of the RBP plasmid
   NOTE: Amino acid and nucleotide sequences of the coat proteins used in this study are listed in Table 2.
   1. Order the required RBP sequence lacking a stop codon as a custom-ordered dsDNA minigene lacking a stop codon with restriction sites at the ends (Figure 1).
   2. Clone the tested RBP lacking a stop codon immediately downstream of an inducible promoter and upstream of a fluorescent protein lacking a start codon (Figure 1), similar to steps 1.2.2-1.2.4. Make sure that the RBP plasmid contains a different antibiotic resistance gene than the binding-site plasmid.
   3. Identify positive transformants via Sanger sequencing, similar to step 1.2.5 (see Table 1 for primer sequences).
   4. Choose one positive transformant and make it chemically-competent²⁵. Store as glycerol purified plasmids at -20 °C and glycerol stocks of bacterial strains²⁴ at -80 °C in 96-well plates.
   5. Transform the binding-site plasmids (from step 1.2.6) stored in 96-well plates into chemically-competent bacterial cells already containing an RBP-mCerulean plasmid²¹. To save time, instead of plating the cells on Petri dishes, plate them using an 8-channel pipettor on 8-lane plates containing Luria-Bertani (LB)²⁶ agar with relevant antibiotics (Kan and Amp). Colonies should appear in 16 h.
   6. Select a single colony for each double transformant and grow overnight in LB medium with the relevant antibiotics (Kan and Amp) and store as glycerol stocks²⁴ at -80 °C in 96-well plates.

2. Experiment Setup

NOTE: The protocol presented here was performed using a liquid-handling robotic system in combination with an incubator and a plate reader. Each measurement was carried out for 24 inducer concentrations, with two duplicates for each strain + inducer combination. Using this robotic system, data for 16 strains per day with 24 inducer concentrations was collected. However, if such a device is unavailable, or if fewer experiments are necessary, these can easily be done by hand using an 8-channel multi-pipette and adapting the protocol accordingly. For example, preliminary results for four strains per day with 12 inducer concentrations and four time-points were acquired in this manner.

1. Prepare, in advance, 1 L of bioassay buffer (BA) by mixing 0.5 g of tryptone, 0.3 mL of glycerol, 5.8 g of NaCl, 50 mL of 1 M MgSO₄, 1 mL of 10x phosphate-buffered saline (PBS) buffer pH 7.4, and 950 mL of double distilled water (DDW). Autoclave or sterile filter the BA buffer.
2. Grow the double-transformant strains at 37 °C and 250 rpm shaking in 1.5 mL LB with appropriate antibiotics (kanamycin at a final concentration of 25 μg/mL and ampicillin at a final concentration of 100 μg/mL), in 48-well plates, over a period of 18 h (overnight).
3. In the morning, make the following preparations.
   1. Inducer plate. In a clean 96-well plate, prepare wells with semi-poor medium (SPM) consisting of 95% BA and 5% LB²⁶ in the incubator at 37 °C. The number of wells corresponds to the desired number of inducer concentrations. Add C4-HSL to the wells in the inducer plate that will contain the highest inducer concentration (218 nM).
   2. Program the robot to serially dilute medium from each of the highest-concentration wells into 23 lower concentrations ranging from 0 to 218 nM. The volume of each inducer dilution should be sufficient for all strains (including duplicates).
   3. While the inducer dilutions are being prepared, warm 180 μL of SPM in the incubator at 37 °C, in 96-well plates.
   4. Dilute the overnight strains from step 2.2 by a factor of 100 by serial dilutions: first dilute by a factor of 10 by mixing 100 μL of bacteria with 900 μL of SPM in 48-well plates, and then dilute again by a factor of 10 by taking 20 μL from the diluted solution into 180 μL of pre-warmed SPM, in 96-well plates suitable for fluorescent measurements.
   5. Add the diluted inducer from the inducer plate to the 96-well plates with the diluted strains according to the final concentrations.
4. Shake the 96-well plates at 37 °C for 6 h, while taking measurements of optical density at 595 nm (OD₅₉₅), mCherry (560 nm/612 nm) and mCerulean (460 nm/510 nm) fluorescence via a plate reader every 30 min. For normalization purposes, measure growth of SMP with no cells added.

3. Preliminary Results Analysis

1. For each day of experiment, choose a time interval of logarithmic growth according to the measured growth curves, between the linear growth phase and the stationary (T­₀, T_final). Take approximately 6−8 time points, while discarding the first and last measurements to avoid error derived from inaccuracy of exponential growth detection (see Figure 2A, top panel).
   NOTE: Discard strains that show abnormal growth curves or strains where logarithmic growth phase could not be detected and repeat the experiment.
2. Calculate the average normalized fluorescence of mCerulean and rate of production of mCherry, from the raw data of both mCerulean and mCherry fluorescence for each inducer concentration (Figure 2A).
   1. Calculate normalized mCerulean as follows:
      
      ​where blank(mCerulean) is the mCerulean level [a.u.] for medium only, blank(OD) is the optical density for medium only, and mCerulean and OD are the mCerulean fluorescence and optical density values, respectively.
   2. Average mCerulean over the different time points (Figure 2B, top two panels) as follows:
      
      ​where #Time points is the number of data timepoints taken into account, T₀ is the time at which the exponential growth phase begins, and T_final is the time at which the exponential growth phase ends.
   3. Calculate mCherry rate of production (Figure 2B, bottom two panels) as follows:
      
      where mCherry(t) is the mCherry level [a.u.] at time t, OD is the optical density value, T₀ is the time at which the exponential growth phase begins, and T_final is the time at which the exponential growth phase ends.
3. Finally, plot the mCherry rate of production as a function of mCerulean, creating dose response curves as a function of RBP-mCerulean fusion fluorescence (Figure 2C). Such plots represent production of the reporter gene as a function of RBP presence in the cell.

4. Dose Response Function Fitting Routine and K_RBP Extraction

1. Under the assumption that the ribosome rate of translation with the RBP bound is constant, model the mCherry production rate as follows (see Figure 2D, green line):
   
   where [x] is the normalized average mCerulean fluorescence calculated according to Eq. 2, mCherry production rate is the value calculated according to Eq. 3, K_RBP is the relative binding affinity [a.u.], K_unbound is the ribosome rate of translation with the RBP unbound, n is the cooperativity factor, and C is the base fluorescence [a.u.]. C, n, K_unbound, and K_RBP are found by fitting the mCherry production rate data to the model (Eq. 4).
2. Using data analysis software, conduct a fitting procedure on plots depicting mCherry production rate as a function of averaged mCerulean (step 3.3), and extract the fit parameters according to the formula in Eq. 4.
   NOTE: Only fitting results with R² > 0.6 are taken into account. For those fits, K_RBP error is mostly in the range of 0.5% to 20% of K_RBP values, for a 0.67 confidence interval, while those with higher K_RBP error can be also verified by eye.
3. Normalize K_RBP values by the respective maximal value of averaged mCerulean for each dose-response function.
   
   where K_RBP in [a.u.] is the value extracted from the fitting procedure in Eq. 4, and max (averaged mCerulean) is the maximal averaged mCerulean signal [a.u] observed for the current strain.
   NOTE: The normalization facilitates correct comparison of the regulatory effect across strains by eliminating the dependence on the particular maximal RBP expression levels.

Representative Results

The presented method utilizes the competition between an RBP and the ribosome for binding to the mRNA molecule (Figure 1). This competition is reflected by decreasing mCherry levels as a function of increased production of RBP-mCerulean, due to increasing concentrations of inducer. In the case of increasing mCerulean fluorescence, with no significant changes in mCherry, a lack of RBP binding is deduced. Representative results for both a positive and a negative strain are depicted in Figure 2. In Figure 2A, the OD, mCherry, and mCerulean channels are presented as a function of time and inducer over a range of four hours, with T­₀ = 1 h and T_final = 3.5 h. In Figure 2B, averaged mCerulean fluorescence (top) and mCherry rate of production (bottom) are presented as a function of inducer concentration, for the two example strains. As can be seen, the results for a positive strain display a clear down-regulatory effect in the mCherry rate of production (Figure 2B,C), which translates into a significant non-zero value of K_RBP (Figure 2D). For the positive strain, the fitting procedure yielded the following values: K_RBP = 394.6 a.u., K_unbound = 275.6, n = 2.1, C = 11.2 a.u., and R² = 0.93. After normalization by the maximal mCerulean fluorescence, the K_RBP value was 0.24. For the negative strain, a lack of distinct response was observed (Figure 2C), and no K_RBP value was extracted (Figure 2D).

In Figure 3, we present the results of this assay for two phage coat RBPs, PP7 and MS2, on several mutated binding sites, at different locations within the initiation region of the mCherry mRNA. The results are roughly classified into three kinds of responses (Figure 3A): strains exhibiting a down-regulatory effect at a low mCerulean level, reflecting a low K_RBP value (high binding affinity); strains exhibiting down-regulatory effect at either intermediate or high mCerulean levels, reflecting a high K_RBP value (intermediate or low affinity); and strains exhibiting no distinct response to rising levels of mCerulean, reflecting a higher K_RBP value than the maximum RBP concentration in the cell (no detectible binding affinity). Figure 3B presents the minimal K_RBP value computed for every RBP−binding-site combination based on all combinations of the two RBPs and ten binding-sites at different positions. The binding sites include a negative control (no binding site), non-matching binding sites, and a positive control ― the native binding site for each RBP (PP7-wt for PP7 coat protein [PCP], and MS2-wt for MS2 coat protein [MCP]). The results match the predictions, as both RBPs present a high affinity for their positive controls, and a non-detectible binding affinity for the negative controls. Additionally, previous studies using these two RBPs^27,28 have observed that they are orthogonal, which is clearly conveyed in the heatmap presented: both MCP and PCP do not bind the native site of the other RBP. Furthermore, the mutated binding sites present varying results, where some binding sites displayed a similar level of affinity as that of the native site, such as PP7-mut-1, PP7-mut-2, and MS2-mut-3, while others displayed a significantly lower affinity, such as PP7-mut-3 and MS2-mut-2. Thus, the assay presented a quantitative in vivo measurement of the binding affinity of RBPs, yielding results that are comparable to those of past experiments with these RBPs.

Since the assay is based on repression of the mCherry gene, a viable mCherry signal is required. Therefore, when designing the binding site cassette, there are two design rules to keep in mind. First, the open reading frame (ORF) of the mCherry should be kept. Since the binding-site length can vary, inserting it into the gene can cause a shift of one or two bases from the original mCherry ORF. Therefore, if needed (Figure 4A), insert one or two bases immediately downstream to the binding site. For example, a binding site that is 20-base long, with a δ of two bases, will yield an addition of 22 bases to the mCherry gene. To keep the ORF, we need to add two bases, for a total of 24 bases. The second design rule is to avoid insertions of stop codons into the mCherry ORF. Some binding sites, as the MS2-mut-2 (Figure 4B, inset), contain stop codons when positioned in one or more of the three possible ORFs. Such an example is illustrated in Figure 4A, where the binding site contained a stop codon that is in-frame with the mCherry ORF only when no bases are added. As can be seen in the dose-response curve for that position (Figure 4B), mCherry production rate was undetectable, thus the binding affinity could not be measured.

A closer look at Figure 4B demonstrates the effect of the spacing δ on mCherry production. For instance, for δ = 4, basal production rate was a factor of six more than those for δ = 5, ensuring a higher fold-repression effect. For δ = 14, however, the basal production levels were too low to observe a down-regulatory effect.

Figure 1: Overview of system design and cloning steps. Illustration of the cassette design for the binding site plasmid (left) and RBP-mCerulean plasmid (right). The next step is consecutive transformations of both plasmids into competent E. coli cells, with RBP plasmids first. Double-transformants are then tested for their mCherry expression levels in increasing inducer concentrations; if the RBP binds to the binding site, mCherry levels decline as a function of mCerulean (gray bubble). Please click here to view a larger version of this figure.

Figure 2: Analysis scheme. (A) Three-dimensional (3D) plots depicting raw OD levels (top), mCerulean fluorescence (middle), and mCherry fluorescence (bottom) as a function of time and inducer concentration, for a positive strain. (B) Top: mCerulean steady-state expression levels for each inducer concentration is computed by dividing each fluorescence level by the respective OD and averaging over all values in the 2−3 h exponential growth time window for both the positive (left) and negative (right) strains. Bottom: mCherry production rate computed according to Eq. 3 for time-points 2−3 h after induction. (C) mCherry production rate plotted as a function of mean mCerulean fluorescence averaged over two biological duplicates for two strains. Error bars are standard deviation of both mCherry production rate and averaged mCerulean fluorescence acquired from at least two replicates. (D) Fit for K_RBP using the fitting formula in Eq. 4 shown for the positive strain (left), exhibiting a specific binding response. For the negative strain (right), no K_RBP value was extracted. Data is shown in duplicate. This figure has been adapted with permission from Katz et al.¹⁰. Copyright 2018 American Chemical Society. Please click here to view a larger version of this figure.

Figure 3: Representative final results. (A) Normalized dose-response curves for thirty different strains based on two RBPs and ten binding sites at different locations. Three types of responses are observed: high affinity, low affinity, and no affinity. (B) Quantitative K_RBP results for two RBPs (MCP and PCP) with five different binding site cassettes (listed). All RBP−binding-site strains were measured in duplicate. This figure has been adapted with permission from Katz et al.¹⁰. Copyright 2018 American Chemical Society. Please click here to view a larger version of this figure.

Figure 4: Example design and results for MCP with a mutant binding site. (A) Design illustration of the binding site cassettes in four different locations. Cassette including the ribosome binding site, start codon for the mCherry, δ spacer bases, the binding site tested, one or two bases to maintain the ORF, and the rest of the mCherry gene. Red stars indicate a stop codon. (B) Dose-response curves for MCP with a mutant binding site at four different locations. Inset: the sequence of the tested mutated binding site. Results presented are for duplicates of each strain. Please click here to view a larger version of this figure.

Name	Binidng site location, A in AUG = 1	Binding site sequence (RBS for controls)	Site: ATG to second mCherry codon GTG Controls: RBS to second mCherry codon GTG	Source
MS2_wt_d5	5	acatgaggattacccatgt	atgcacatgaggattacccatgtcgtg	Gen9 Inc.
MS2_wt_d6	6	acatgaggattacccatgt	atggcacatgaggattacccatgtgtg	Gen9 Inc.
MS2_wt_d8	8	acatgaggattacccatgt	atggcgcacatgaggattacccatgt cgtg	Gen9 Inc.
MS2_wt_d9	9	acatgaggattacccatgt	atggcgccacatgaggattacccatg tgtg	Gen9 Inc.
MS2_U(-5)C_d8	8	acatgaggatcacccatgt	atgcacatgaggatcacccatgtgg tg	Gen9 Inc.
MS2_U(-5)C_d9	9	acatgaggatcacccatgt	atggcacatgaggatcacccatgtg tg	Gen9 Inc.
MS2_U(-5)C_d8	8	acatgaggatgacccatgt	atgcacatgaggatgacccatgtgg tg	Gen9 Inc.
MS2_U(-5)G_d9	9	acatgaggatgacccatgt	atggcacatgaggatgacccatgtg tg	Gen9 Inc.
MS2_struct_d9	9	cacaagaggttcacttatg	atggccacaagaggttcacttatgg tg	Gen9 Inc.
MS2_struct_d8	8	cacaagaggttcacttatg	atgccacaagaggttcacttatggg tg	Gen9 Inc.
PP7wt_d5'	5	taaggagtttatatggaaaccctta	atgctaaggagtttatatggaaacc cttacgtg	Gen9 Inc.
PP7wt_d6'	6	taaggagtttatatggaaaccctta	atgaataaggagtttatatggaaac ccttagtg	Twist Bioscience
PP7wt_d8'	8	taaggagtttatatggaaaccctta	atgaacataaggagtttatatggaa acccttacgtg	Twist Bioscience
PP7wt_d9'	9	taaggagtttatatggaaaccctta	atgaacaataaggagtttatatgga aacccttagtg	Twist Bioscience
PP7_USLSBm_d6	6	taaccgctttatatggaaagggtta	atggctaaccgctttatatggaaag ggttagtg	Gen9 Inc.
PP7_USLSBm_d15	15	taaccgctttatatggaaagggtta	atgggcgccggcgctaaccgcttta tatggaaagggttagtg	Gen9 Inc.
PP7_nB_d5	5	taagggtttatatggaaaccctta	atgctaagggtttatatggaaaccc ttagcgtg	Gen9 Inc.
PP7_nB_d6	6	taagggtttatatggaaaccctta	atggctaagggtttatatggaaacc cttatgtg	Gen9 Inc.
PP7_USs_d5	5	taaggagttatatggaaccctta	atgctaaggagttatatggaaccct tagtg	Gen9 Inc.
PP7_USs_d6	6	taaggagttatatggaaccctta	atggctaaggagttatatggaaccc ttagcgtg	Gen9 Inc.
No_BS_d1	-	-	ttaaagaggagaaaggtacccatgg tg	Gen9 Inc.
No_BS_d4	-	-	ttaaagaggagaaaggtacccatgg gcgtg	Gen9 Inc.
No_BS_d10	-	-	ttaaagaggagaaaggtacccatgg gcgccggcgtg	Gen9 Inc.
Sequencing primer for binding site cassettes			gcatttttatccataagattagcgg	IDT
Sequencing primer for RBP cassettes			gcggcgctgggtctcatctaataa	IDT

Table 1: Binding sites and sequencing primers. Sequences for the binding sites and binding site cassettes used in this study, as well as the primers for the sequencing reactions detailed in the protocol (steps 1.2.5.1 and 1.3.3).

RBP name in this work	source organism name, protein	source organism gene	source organism refseq	wt aa seq	changes from wt (and references)	aa seq used in this work	nt seq used in this work
MCP	Escherichia virus MS2	cp	NC_001417.2	MASNFTQFVLV DNGGTGDVTV APSNFANGVA EWISSNSRSQ AYKVTCSVRQ SSAQNRKYTI KVEVPKVATQT VGGVELPVA AWRSYLNMEL TIPIFATNSD CELIVKAMQG LLKDGNPIPS AIAANSGIY	delF-G [1] V29I [1] taken from addgene plasmid 27121	MASNFTQFVLV DNGGTGDVTV APSNFANGIA EWISSNSRSQ AYKVTCSVRQ SSAQNRKYTI KVEVPKG AWRSYLNMEL TIPIFATNSD CELIVKAMQG LLKDGNPIPS AIAANSGIY	ATGGCTTCTA ACTTTACTCA GTTCGTTCTC GTCGACAATG GCGGAACTGG CGACGTGACT GTCGCCCCAA GCAACTTCGC TAACGGGATC GCTGAATGGA TCAGCTCTAA CTCGCGTTCA CAGGCTTACA AAGTAACCTG TAGCGTTCGT CAGAGCTCTG CGCAGAATCG CAAATACACC ATCAAAGTCG AGGTGCCTAA AGGCGCCTGG CGTTCGTACT TAAATATGGA ACTAACCATT CCAATTTTCG CCACGAATTC CGACTGCGAG CTTATTGTTA AGGCAATGCA AGGTCTCCTA AAAGATGGAA ACCCGATTCC CTCAGCAATC GCAGCAAACT CCGGCATCTAC
PCP	Pseudomonas phage PP7	cp	NC_001628.1	MSKTIVLSVGEA TRTLTEIQST ADRQIFEEKV GPLVGRLRLT ASLRQNGAKT AYRVNLKLDQ ADVVDCSTSVC GELPKVRYTQ VWSHDVTIVA NSTEASRKSL YDLTKSLVAT SQVEDLVVNL VPLGR	delF-G [2] taken from addgene plasmid 40650	MLASKTIVLSVG EATRTLTEIQ STADRQIFEE KVGPLVGRLR LTASLRQNGA KTAYRVNLKL DQADVVDSG LPKVRYTQVW SHDVTIVANS TEASRKSLYD LTKSLVATSQ VEDLVVNLVP LGR	ATGCTAGCCTC CAAAACCATC GTTCTTTCGG TCGGCGAGGC TACTCGCACT CTGACTGAGA TCCAGTCCAC CGCAGACCGT CAGATCTTCG AAGAGAAGGT CGGGCCTCTG GTGGGTCGGC TGCGCCTCAC GGCTTCGCTC CGTCAAAACG GAGCCAAGAC CGCGTATCGC GTCAACCTAA AACTGGATCA GGCGGACGTC GTTGATTCCG GACTTCCGAA AGTGCGCTAC ACTCAGGTAT GGTCGCACGA CGTGACAATC GTTGCGAATA GCACCGAGGC CTCGCGCAAA TCGTTGTACG ATTTGACCAA GTCCCTCGTC GCGACCTCGC AGGTCGAAGA TCTTGTCGTC AACCTTGTGC CGCTGGGCCGT
References:
1.Peabody, D.S., Ely, K.R. Control of translational repression by protein-protein interactions. Nucleic Acids Research. 20 (7), 1649–1655 (1992).
2. Chao, J.A., Patskovsky, Y., Almo, S.C., Singer, R.H. Structural basis for the coevolution of a viral RNA–protein complex. Nature Structural & Molecular Biology. 15 (1), 103–105, doi: 10.1038/nsmb1327 (2008)

Table 2: RBP sequences. Amino acid and nucleotide sequences of the coat proteins used in this study.

Discussion

The method described in this article facilitates quantitative in vivo measurement of RBP-RNA binding affinity in E. coli cells. The protocol is relatively easy and can be conducted without the use of sophisticated machinery, and data analysis is straightforward. Moreover, the results are produced immediately, without the relatively long wait-time associated with next generation sequencing (NGS) results.

One limitation to this method is that it works only in bacterial cells. However, a previous study¹² has demonstrated a repression effect using a similar approach for the L7AE RBP in mammalian cells. An additional limitation of the method is that the insertion of the binding site in the mCherry initiation region may repress basal mCherry levels. Structural complexity or high stability of the binding site can interfere with ribosomal initiation even in the absence of RBP, resulting in decreased mCherry basal levels. If basal levels are too low, the additional repression brought on by increasing concentrations of RBP will not be observable. In such a case, it is best to design the binding site cassette with the binding site still in the initiation region, but on the verge of the transition from initiation region to elongation region (δ in the range of 12−15 bp^10,29). We have shown that for such δ values a repression effect can still be observed. To increase the chances that the assay will work, regardless of structural complexity, we advise performing the assay on at least three different positions for a given binding site.

The main disadvantage of the method in comparison to in vitro methods, such as EMSA, is that the RBP-RNA binding affinity is not measured in absolute units of RBP concentration, but rather in terms of fusion-RBP fluorescence. This disadvantage is a direct result of the in vivo setting, which limits our ability to read out the actual concentrations of RBP. This disadvantage is offset by the benefits of measuring in the in vivo setting. For example, we have found differences in binding affinities when comparing results from our in vivo assay to previous in vitro and in situ assays. These differences may stem from discrepancies in the structure of the mRNA molecules in vivo that emerge from their presence inside cells^10,11,30,31. Such structural differences may lead to changes in the stability of the folded states in vivo which, in turn, either stabilize or de-stabilize RBP binding.

Since the method is relatively simple and inexpensive, we advise running multiple controls alongside the actual experiment. Running a negative control, i.e., a sequence that has no affinity to the RBP yet has similar structural features, can help avoid false positives stemming from non-specific interactions with the mRNA. In the representative results shown, the two negative controls were the mCherry gene alone (no binding site), and the native binding site of the other RBP (i.e., PP7-wt for MCP and MS2-wt for PCP). Moreover, we propose incorporating a positive control (such as an RBP and its native binding site). Such a control will help in quantifying the binding affinity by presenting a reference point, and in avoiding false-negatives stemming from low fold-repression.

Finally, for those who wish to obtain a structural perspective of RBP-RNA binding, we propose carrying out a selective 2′-hydroxyl acylation analyzed by primer extension sequencing (SHAPE-Seq)^11,32,33 experiment. SHAPE-Seq is an NGS approach combined with chemical probing of RNA, which can be used to estimate secondary structure of RNA as well as RNA interactions with other molecules, such as proteins. In our previous work we conducted a SHAPE-Seq experiment on a representative strain in both in vivo conditions³⁴ and in vitro with purified recombinant protein^10,35. In our case, the results revealed that RBP-binding affected a much wider segment of RNA than previously reported for these RBPs in vitro³⁶.

Materials

Name	Company	Catalog Number	Comments
Ampicillin sodium salt	SIGMA	A9518
Magnesium sulfate (MgSO4)	ALFA AESAR	33337
48 plates	Axygen	P-5ML-48-C-S
8-lane plates	Axygen	RESMW8I
96-well plates	Axygen	P-DW-20-C
96-well plates for plate reader	Perkin Elmer	6005029
ApaLI	NEB	R0507
Binding site sequences	Gen9 Inc. and Twist Bioscience		see Table 1
E. coli TOP10 cells	Invitrogen	C404006
Eagl-HF	NEB	R3505
Glycerol	BIO LAB	071205
Incubator	TECAN	liconic incubator
Kanamycin solfate	SIGMA	K4000
KpnI- HF	NEB	R0142
Ligase	NEB	B0202S
Liquid-handling robotic system	TECAN	EVO 100, MCA 96-channel
Matlab analysis software	Mathworks
Multi- pipette 8 lanes	Axygen	BR703710
N-butanoyl-L-homoserine lactone (C4-HSL)	cayman	K40982552 019
PBS buffer	Biological Industries	020235A
Platereader	TECAN	Infinite F200 PRO
Q5 HotStart Polymerase	NEB	M0493
RBP seqeunces	Addgene	27121 & 40650	see Table 2
Sodium Chloride (NaCL)	BIO LAB	190305
SV Gel and PCR Clean-Up System	Promega	A9281
Tryptone	BD	211705