Locked Nucleic Acid and Flow Cytometry-Fluorescence In Situ Hybridization for the Detection of Bacterial Small Noncoding RNAs

Abstract

We describe the development and testing of a high-throughput method that enables the detection of small noncoding RNAs (ncRNAs) from single bacterial cells using locked nucleic acid probes (LNA) and flow cytometry-fluorescence in situ hybridization (flow-FISH). The LNA flow-FISH method and quantitative reverse transcription-PCR (qRT-PCR) were used to monitor the expression of three ncRNAs (6S, CsrB, and TPP-2) in Vibrio campbellii ATCC BAA-1116 cultures during lag phase, mid-log phase, and stationary phase. Both LNA flow-FISH and qRT-PCR revealed that CsrB and TPP-2 were highly expressed during lag phase but markedly reduced in mid-log phase and stationary phase, whereas 6S demonstrated no to little expression during lag phase but increased thereafter. Importantly, while LNA flow-FISH and qRT-PCR demonstrated similar overall expression trends, only LNA flow-FISH, which enabled the detection of ncRNAs in individual cells as opposed to the lysate-based ensemble measurements generated by qRT-PCR, was able to capture the cell-to-cell heterogeneity in ncRNA expression. As such, this study demonstrates a new method that simultaneously enables the in situ detection of ncRNAs and the determination of gene expression heterogeneity within an isogenic bacterial population.

INTRODUCTION

The ability to monitor and measure changes in mRNA transcripts from single bacterial cells becomes particularly important when trying to understand the role that stochastic cellular processes play in determining cell-to-cell variation in gene expression and how this variation can lead to physiological heterogeneity in microbial populations (6, 23, 27, 32). While it is clear that transcriptional analyses in single cells can greatly contribute to our understanding of individual and community behaviors, there continues to be a shortage of facile methods for making these measurements. This is especially the case where high-throughput measurements are required for the simultaneous analysis of thousands of cells to better assess reproducibility and variability for strong statistical analyses (21).

One of the few tools that microbiologists have at their disposal for the high-throughput analysis of single bacterial cells is flow cytometry (4). Combined with fluorescence in situ hybridization (FISH), flow cytometry (flow-FISH) has become a powerful tool that has been used to quantify rare bacterial cells from mixtures (3, 7), enrich for subpopulations from complex environmental matrices (15, 44), monitor the effectiveness of antimicrobial treatments (38), identify physiological and ecological diversity (16), measure the modulation of microbiota in response to nutrient shifts (5), and enable the detection of pathogenic microbes (12). Flow-FISH probes are typically designed to target rRNA due to the relative cellular abundance of rRNA molecules and their accepted use in species identification. However, flow-FISH has also been used to detect specific bacterial mRNA (14, 29), though far less frequently, due to the comparatively lower abundance and shorter half-life of mRNA, properties which ultimately strain signal detection limits. To counter this limitation, researchers have replaced the use of strict DNA probes in flow-FISH with high-affinity locked nucleic acid (LNA)-incorporated DNA probes (30, 31). In LNAs, the ribose sugar is constrained by a methylene bridge between 2′-oxygen and 4′-carbon, resulting in an N-type conformation (17, 26) that dramatically increases the melting temperature of the LNA-target hybridization product (2, 17). As a result, LNA probes have been shown to increase the hybridization efficiency, sensitivity, and specificity of FISH (18, 34, 36), and these properties suggest that LNA probes should be well suited for the detection of bacterial mRNAs and small noncoding RNAs (ncRNAs).

The recognized importance of ncRNAs as key regulatory elements of critical cellular processes (9, 25, 42) and the versatility of flow-FISH have led to the prediction that this method may also be adapted to examine the regulation and expression of ncRNAs (39). In this study, we describe the development and testing of an LNA flow-FISH method that enabled the detection of ncRNA expression in individual bacterial cells and the measurement of stochastic fluctuations in ncRNA expression within homogeneous bacterial populations.

(This work was presented in part at the 111th General Meeting of the American Society for Microbiology, New Orleans, LA, 2011.)

MATERIALS AND METHODS

Bacterial growth and fixation.

Vibrio campbellii ATCC BAA-1116 cultures were grown in Luria marine medium (20 g NaCl, 10 g Bacto tryptone, 5 g yeast extract per liter, pH 7.8) for 16 h at 30°C in a shaking incubator (200 rpm). The overnight cultures were used to seed triplicate 50-ml cultures with 2 × 10⁵ cells/ml, and aliquots from each flask were taken immediately (time zero) for growth curve analyses using the Bioscreen C MBR (Growth Curves USA, Piscataway, NJ). At the times of interest, the cultures were sampled and the cells were harvested via centrifugation and rinsed in 1× phosphate-buffered saline (PBS). Half of each cell pellet was processed using the MirVana RNA extraction kit according to the manufacturer's recommendations for total RNA extraction (Applied Biosystems/Ambion, Austin, TX). The other half of each cell pellet was incubated for 10 min in a fixative solution (4% paraformaldehyde, 5% acetic acid in 1× PBS) at room temperature, washed twice in 1× PBS, and stored at 4°C. Samples from three biological replicates were collected and analyzed for every time point.

LNA probes.

Four biotinylated LNA-modified DNA oligonucleotide probes were designed and purchased for the detection of the ncRNA (Exiqon, Inc., Woburn, MA). The negative-control Spot-42 ncRNA-specific probe had the following sequence: 5′-biotinTEG-aCatCttAccTctGtaCccTacG-3′ (melting temperature [T_m] =76°C; LNA monomers in uppercase letters). Three additional LNA probes were designed targeting the CsrB ncRNA (5′-biotin-TEG-gTccAttTccCgtCctTagCagC-3′, T_m =80°C), 6S ncRNA (5′-biotin-TEG-tAggTatTgcTtaTcgGctCagG-3′, T_m =77°C), and TPP-2 ncRNA (5′-biotin-TEG-cAagTggGttTgcTccCcgAtgA, T_m =81°C). The T_m of each probe was estimated using the Exiqon T_m prediction program (http://www.exiqon.com/ls/Pages/ExiqonTMPredictionTool.aspx) (37).

Flow-FISH.

The fixed bacterial aliquots were incubated in freshly prepared 0.1% (vol/vol) diethylpyrocarbonate in 1× PBS (Sigma-Aldrich, St. Louis, MO) for 12 min at room temperature and rinsed twice with 1× PBS. The cells were then permeabilized with lysozyme (1 mg/ml) (Sigma-Aldrich) in Tris-EDTA buffer for 30 min at room temperature, followed immediately by treatment with proteinase K (3 μg/ml) (Applied Biosystems/Ambion) in Tris-EDTA for 15 min at room temperature. The cells were then rinsed twice in 1× PBS and split into aliquots for the different LNA probes and controls. Probe-ncRNA hybridization was performed in 100 μl of hybridization solution (20 pmol specific LNA, 50% formamide [Applied Biosystems/Ambion], 10% dextran sulfate [Sigma-Aldrich], 1× Denhardt's solution [Sigma-Aldrich], 50 mM sodium phosphate buffer [pH 7.0], 2× SSC [1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate], 20 μg sheared salmon sperm DNA [Applied Biosystems/Ambion], and 20 μg yeast tRNA [Invitrogen]) for 1 h at 60°C. Following hybridization, 1 ml of 0.1× sodium citrate with 0.1% Tween 20 buffer (SSCT) was added to each sample, and the cells were pelleted and the supernatant discarded. The cell pellets were then resuspended and incubated in 200 μl of 50% formamide in 2× SSCT buffer at 65°C for 30 min. Again, 1 ml of 0.1× SSCT buffer was added to the mixture, the cells pelleted, and the supernatant removed. Each cell pellet was then resuspended in 500 μl 0.1× SSCT and incubated for 40 min at 65°C. The negative controls were subjected to the same conditions as described above and consisted of either the Spot-42 LNA or no LNA. Following the rinsing steps, the cells were incubated in 1× blocking buffer (Vector Laboratories, Burlingame, CA) for 30 min at room temperature and stained with 2 μg/ml of DyLight 488-conjugated streptavidin (Invitrogen) in 1× blocking buffer for 12 min at room temperature with constant shaking. Afterward, the cells were washed once with 0.1× SSCT buffer and then once with PBST (1× PBS plus 0.1% Tween 20 buffer). The signal was then amplified using a biotinylated antistreptavidin antibody (Vector Laboratories) (1 μg/ml in 1× PBS) for 45 min at room temperature. The cells were washed twice in PBST and then stained again with 2 μg/ml DyLight 488-conjugated streptavidin for signal enhancement. To measure cellular autofluorescence, an aliquot of cells were subjected to the LNA flow-FISH protocol, including hybridization buffer containing the Spot-42 LNA probe, but were not stained. The cells were rinsed as previously described, counterstained with propidium iodide (PI), and subjected to centrifugation at 1,200 rpm to remove any cell aggregates.

Flow cytometry.

Flow cytometry analyses were performed using an Accuri C6 flow cytometer equipped with a 488-nm laser and standard emission filters for fluorescein isothiocyanate (FITC) and PI. For each sample, 2 × 10⁴ events were collected in a gate corresponding to the cell population.

qRT-PCR.

A DNA standard was produced for each ncRNA using primers that generated amplicons that were 50 to 79 bp longer than the internal primers used for quantitative reverse transcription-PCR (qRT-PCR) (see Table S1 in the supplemental material). The PCR products were purified using a QIAquick PCR purification kit (Qiagen, Valencia, CA) and quantified using a Qubit 2.0 fluorometer and Qubit double-stranded DNA (dsDNA) BR assay kit (Invitrogen, Carlsbad, CA). Quantitative reverse transcription-PCR assays were conducted on an iCycler (Bio-Rad Laboratories, Hercules, CA) using the iScript One-Step RT-PCR kit with SYBR green (Bio-Rad) according to the manufacturer's instructions. Real-time PCR mixtures consisted of 1× SYBR green PCR master mix (Applied Biosystems, Foster City, CA), 20 ng of total RNA from each extracted sample, 200 nM primers and were subjected to the following cycling conditions: one cycle at 50°C for 30 min and 95°C for 15 min, followed by 40 cycles of 94°C for 30 s, 54°C for 45 s, and 72°C for 45 s.

RESULTS

Method development.

LNA flow-FISH has been successfully used for the detection of high-abundance eukaryotic mRNA and viral RNA (30, 31). However, a number of variables (e.g., RNase inactivation, permeabilization, and signal amplification) had to be tested in order to adapt this method for the detection of relatively low-abundance and transient bacterial ncRNAs. A primary concern was our ability to prevent the degradation of ncRNAs by intracellular RNases. This concern was addressed early in the developed LNA flow-FISH protocol by treating the fixed bacterial cells with diethyl pyrocarbonate (DEPC), which irreversibly inactivates RNases. Compared to cells that were not treated with DEPC, DEPC-treated cells generated stronger and more reproducible flow-FISH signals (data not shown).

Another variable investigated was cell permeabilization, as it was required to enhance the LNA probe penetration into fixed cells and had a marked effect on the strength of the FISH signal. While too little permeabilization resulted in weak signals, overdigestion resulted in poor cell morphology and the loss of cells and target molecules. This was evidenced by the facts that higher lysozyme concentrations resulted in cell clumping and higher proteinase K concentrations resulted in cell loss (data not shown). Treatments with lysozyme (1 mg/ml) followed by a mild digestion with proteinase K (3 μg/ml) provided the best permeabilization conditions tested for V. campbellii.

As the abundance of ncRNA was a concern, signal amplification was also investigated to increase ncRNA detection sensitivity. Following staining with streptavidin-DyLight 488, the protein-dye conjugate itself was detected with a biotinylated antistreptavidin antibody, followed by a secondary staining with streptavidin-DyLight 488. Because the antibody can recognize streptavidin through its antigen binding site and be recognized by streptavidin due to its conjugated biotin moieties, the localized hybridization signal was amplified many-fold, thereby enabling the detection of small changes in ncRNA expression over the background. A detailed listing of the variables tested and the resulting outcomes are presented in Table S2 in the supplemental material.

Detecting ncRNA expression via LNA flow-FISH.

Four previously identified V. campbellii ncRNAs were targeted in this study: Spot-42, CsrB, 6S, and TPP-2 (35). The Spot-42 ncRNA (116 nucleotides [nt]), which acts as an antisense RNA to differentially regulate the galactose operon in Escherichia coli (24), was chosen as a negative-control target as it is not expressed in V. campbellii under the growth conditions utilized in this study. In contrast, the mid-log-phase expression of the other three targeted ncRNAs—CsrB (315 nt), which inhibits the degradation of mRNA involved in glycogen metabolism by binding directly to the carbon storage regulator protein CsrA (22), 6S (185 nt), which inhibits the expression of genes during stationary phase by binding directly to RNA polymerase (41), and TPP-2 (150 nt), a thiamine pyrophosphate-dependent riboswitch that resides in the 5′ untranslated region of the V. campbellii tbpA mRNA which participates in thiamine metabolism (43)—had previously been confirmed (35). Applying the developed LNA flow-FISH protocol to cells harvested during lag-phase growth (4.5 h) validated the method by demonstrating the detection of 6S, CsrB, and TPP-2 ncRNA expression (Fig. 1). The histograms presented (Fig. 1A) demonstrate (i) detection of the three ncRNAs over the background (Fig. 1A, black trace, no DyLight 488, and gray trace, no LNA) and (ii) a clear peak shift denoting greater fluorescence intensity when using the LNA probes specific for the CsrB and TPP-2 ncRNAs. The corresponding dot plots (Fig. 1B) are gated in red on the Spot-42 population. Again, the CsrB and TPP-2 ncRNA plots show clear shifts to higher fluorescence (indicative of greater transcript abundance), while the 6S ncRNA plot shows a small shift (indicative of low levels of 6S expression) that is more easily visualized in the gated dot plot than in the cognate histogram.

Fig 1.

Detection of ncRNA expression using LNA flow-FISH. (A) Histograms demonstrating 6S, CsrB, and TPP-2 expression (blue) compared to that of the various controls: Spot-42 (red), no LNA (gray), and no DyLight 488 (black). (B) Dot plots gated on the Spot-42 LNA signal (red) reveal the change in fluorescence when using LNA probes specific for each ncRNA. All data are from lag-phase (4.5 h) V. campbellii cultures.

Analyses of ncRNA expression over time.

Based on their purported functions, we expected to see the greatest differences in expression of each of the targeted ncRNAs during lag phase and stationary phase. To determine if we could monitor changes in ncRNA expression over time, V. campbellii cells were harvested from lag-phase (4.5 h), mid-log-phase (10 h), and stationary-phase (30 h) cultures (Fig. 2A) and tested using the LNA flow-FISH method and a more established and accepted method, qRT-PCR. The LNA flow-FISH method revealed that the CsrB and TPP-2 ncRNAs were most abundant during lag-phase growth and demonstrated a reduction in expression during mid-log phase and stationary phase (Fig. 2B). In contrast, the 6S ncRNA demonstrated little to no expression during lag phase but increased during mid-log phase and stationary phase. Importantly, the targeted negative-control Spot-42 ncRNA was not expressed at any of the time points tested. Compared to the qRT-PCR analyses using the same cultures (Fig. 2C to E), both methods appeared to demonstrate similar overall expression trends, with the only exception being the discrepancy seen in the mid-log-phase (10 h) expression of the 6S ncRNA (Fig. 2B versus C). As was the case in the LNA flow-FISH analyses, the Spot-42 ncRNA was also not detected in the qRT-PCR analyses (data not shown).

Fig 2.

Lag-, mid-log-, and stationary-phase ncRNA expression. (A) V. campbellii growth curves. Arrows indicate the sampling time points (4.5, 10, and 30 h) for LNA flow-FISH and qRT-PCR analyses. OD, optical density. The symbols (circles diamonds, and triangles) represent each of the three biological replicates. (B) Changes in ncRNA expression over time as determined by LNA flow-FISH. The Spot-42 background has been subtracted from the data presented. (C to E) Matched-culture qRT-PCR results for 6S (C), CsrB (D), and TPP-2 (E). Data shown represent means ± standard deviations of the results of three independent experiments.

Monitoring ncRNA expression during the transition from lag-phase to mid-log-phase growth.

As the most dramatic changes in ncRNA expression observed were between lag phase and mid-log phase, we chose to monitor this transition with greater resolution by sampling cultures hourly. The LNA flow-FISH data confirmed the expression trends previously demonstrated during this transition (Fig. 2B): CsrB and TPP-2 ncRNA expression decreased over time, whereas 6S ncRNA expression remained constant and began to increase in mid-log phase (Fig. 3A). In addition, the data generated from triplicate cultures also revealed greater intersample variability at the earliest time points (3.5 h and 4.5 h) for the 6S and TPP-2 ncRNAs and this variability decreased over time (Fig. 3A). In contrast, the CsrB ncRNA demonstrated high intersample variability for every time point tested. We also examined the dot plots for each time point during this transition period to understand the intrasample variability (i.e., variation in ncRNA expression within the population as opposed to between replicate cultures) (Fig. 3B). Perhaps the most striking feature was the overall cohesiveness of the population with respect to TPP-2 ncRNA expression at the individual time points and over time. The limited noise in this population suggested that TPP-2 expression is tightly controlled during this transition. By comparison, the expression of the 6S ncRNA was first seen to have little variation in expression but became more variable over time. Of the three ncRNAs, the expression of CsrB was the most variable at each time point tested. Moreover, certain time points (5.5 h and 9.5 h) revealed the presence of two distinct cell populations that may be the result of transcriptional bursts that lead to a bimodal response in CsrB expression (Fig. 3B).

Fig 3.

Expression of ncRNAs during the transition from lag-phase to mid-log phase growth. (A) LNA flow-FISH analyses of cells collected hourly reveal changes in signal during the transition from lag-phase to mid-log-phase growth. Data shown represent means ± standard deviations of the results of three independent experiments, and the error is indicative of intersample variation. (B) Representative LNA flow-FISH dot plots simultaneously depicting the cell-to-cell heterogeneity and temporal change in ncRNA expression. Each ncRNA was individually gated on the 3.5-h time point (red, 6S; purple, CsrB; blue, TPP-2) and held constant for each plot. The values in the lower right corners of the plots indicate the percentage of cells that are included in the corresponding gate.

DISCUSSION

Flow-FISH is a versatile tool that has been used successfully in a wide range of microbiological applications. In this study, we developed a method that adapted flow-FISH for the sensitive detection of ncRNA expression in individual V. campbellii cells and demonstrated that this method enabled the observation of changes and noise in ncRNA expression over time.

FISH and flow-FISH are tools that are generally well suited to monitor mRNA/ncRNA transcript levels in individual cells but have not been widely adopted for this purpose. This is due in part to the lower abundance of mRNA and ncRNA molecules (compared to the abundance of rRNA), which often test signal detection limits. One strategy that has been employed to help overcome this limitation and improve detection sensitivity has been the use of multiple probes per target RNA (13, 23, 40). In one example, sufficient signal to count individual mRNA molecules in single bacterial cells was generated by hybridizing six multifluorophore-labeled and unique 31- to 36-mer DNA probes to the same mRNA molecule (23). Although successful, the use of multiple probes or long probes with multiple fluorophores is difficult to employ for the detection of small RNA transcripts, such as ncRNA, due to their size (typically ∼50 to 300 nt). A second strategy that has helped to improve FISH detection sensitivity has been the development of methods that enable an increase in the signal output per hybridization event via the localized accumulation of reporter molecules (11, 29). We used this strategy to generate fluorophore-protein and antibody complexes to localize and amplify the LNA-ncRNA hybridization signal. Signal amplification in this manner enabled the detection of the CsrB ncRNA, which is known to be present at ∼100 to 500 molecules per cell based on the phase of growth (10). This level of expression is comparable to the known number of Spot-42 ncRNA molecules per cell (100 to 200) when cells are grown under conditions that prompt the expression of Spot-42 (33).

Unlike CsrB and Spot-42, 6S is not a low-abundance ncRNA. The 6S ncRNA is expressed during all phases of growth and is most abundant in stationary phase, when it reaches ∼10,000 molecules per cell (41). In addition, microarray-based expression-profiling experiments using mid-log-phase cultures of V. campbellii have previously revealed 6S to be one of the most highly expressed transcripts in the entire genome (35). Considering its relative abundance, it was surprising to note that the median fluorescence intensities were consistently comparable to or lower than the far less abundant CsrB ncRNA. One explanation that may account for this observation is that the secondary structure of the 6S ncRNA reduced the hybridization efficiency with the 6S LNA probe. We consider this explanation to be unlikely as one of the inherent benefits of LNAs as hybridization probes is the ability to use elevated hybridization temperatures (as the T_m can be fine tuned depending on the number and placement of LNA nucleotides) to facilitate the resolution of secondary structures in targeted molecules and discourage hybridization complex formation with native targets. Another possible explanation is that the LNA probe was poorly designed and lowered the LNA-ncRNA duplex formation hybridization efficiency. However, this too does not appear to be the case, as the testing of a second unique 6S LNA probe using the same LNA flow-FISH method yielded nearly the same signal intensity and overall result (data not shown). A third explanation for the relative lack of signal strength is that the necessary cell fixation step in the LNA flow-FISH protocol may cross-link the 6S ncRNA to its native σ⁷⁰-RNA polymerase target, thus rendering it inaccessible for hybridization to the LNA probe. During mid-log and stationary phase, >90% of the 6S molecules are bound to the σ⁷⁰-RNA polymerase (41). If the vast majority of 6S molecules cannot be detected due to cross-linking, it then becomes plausible that the fluorescent signal generated represents the detection of expressed and noncomplexed 6S ncRNA only, which would explain the unexpectedly low signal intensity. In this instance, comparing the measured expression abundance and trends of the LNA flow-FISH results (chemically fixed cells) (Fig. 2B) and the matched qRT-PCR results (no chemical fixation) (Fig. 2C) should have been informative but was further convoluted by the discrepancy between the mid-log-phase (10 h) results. Finally, as the 6S ncRNA is by far the most highly expressed of the ncRNA targeted in this study, it is also possible that the expected signal intensity is not observed not because of complications due to hybridization efficiency but rather because the number and proximity of fluorophores that may be loaded in each cell may result in fluorophore quenching. While some explanations are more likely than others, at the current time, we cannot fully explain the mediocre signal intensity or the observed discrepancy between the mid-log-phase LNA flow-FISH and qRT-PCR results for the 6S ncRNA.

In contrast to the 6S ncRNA expression trends, the LNA flow-FISH and qRT-PCR data corroborated that CsrB ncRNA expression is highest in lag phase and decreases over time. As CsrB participates in carbon storage regulation by sequestering the global carbon storage regulator protein CsrA (1, 22), there are a few parallels that can be drawn with the 6S ncRNA–σ⁷⁰-RNA polymerase complex. In this case, comparing the measured expression abundance and trends between the LNA flow-FISH results (chemically fixed cells) and matched qRT-PCR results (no chemical fixation) clearly revealed that cell fixation did not inhibit LNA probe binding to the target ncRNA. Thus, the lower signals detected in mid-log phase and stationary phase are not an artifact of the LNA flow-FISH protocol. It is interesting to note that CsrB and csrA expression in E. coli are known to increase as cultures approach stationary phase (10), whereas our data demonstrate that CsrB in V. campbellii decreases over the same period. As V. campbellii BAA-1116 has multiple CsrB homologs (35), it is possible that this observed difference in regulation is due to the targeted CsrB participating in regulatory circuits other than carbon metabolism (1) or that the CsrB expression profile in V. campbellii differs from that in E. coli.

Similar to the CsrB ncRNA expression trends, the LNA flow-FISH and qRT-PCR data also corroborated that TPP-2 ncRNA expression is highest in lag phase and decreases over time. The TPP-2 ncRNA riboswitch (thi-box) resides in the 5′ untranslated region of the V. campbellii tbpA mRNA, which encodes the thiamine ABC transporter substrate binding subunit. Although the genetic regulation of tbpA expression in V. campbellii is not known, one interpretation of the expression data may suggest that TPP-2 expression is elevated early in lag phase as the cells adapt from a thiamine-depleted environment (overnight cultures where intracellular thiamine pyrophosphate > extracellular thiamine pyrophosphate) to a thiamine-replete environment (fresh medium where the level of intracellular thiamine pyrophosphate is less than that of extracellular thiamine pyrophosphate). Once exogenous thiamine pyrophosphate is transported into the cell (via TbpA) and present in a sufficient concentration, it can repress the further expression of tbpA via feedback inhibition, resulting in the observed decline of detectable TPP-2 ncRNA from lag to stationary phase.

In addition to elucidating expression trends over time, LNA flow-FISH also enables the detection of cell-to-cell variation in gene expression within the context of the population, as each cell is treated as an independent observation. This is in contrast to traditional ensemble measurements (such as RT-PCR or microarray analyses) that reveal the average properties of a population of cells but do not reveal the properties of individual cells or the variation within the population (20). Recently, the term “noise” has been used to describe these levels of variation in gene expression from isogenic microbial populations (28, 32). This noise is generated by stochastic processes in the cell and/or differences in external microenvironment stimuli and can result in substantial and selectable physiological differences. One salient example of this is the demonstration that random intrinsic noise is responsible for the variations in Bacillus subtilis comK gene expression that select cells for competence (23). In this study, the use of the LNA flow-FISH method to monitor the transition from lag phase to mid-log phase not only revealed the transcriptional heterogeneity of the population with respect to each ncRNA but also demonstrated that the level of transcriptional noise varied greatly between time points and ncRNA species. This is best observed in the CsrB expression plots, which demonstrate an overall decrease in expression over time but with much greater variation than the TPP-2 profile, which shows a steady and less variable decrease in expression, suggesting tighter regulation of this ncRNA. Furthermore, the CsrB dot plots reveal two distinct cell populations that are most noticeable at 5.5 h and less so at 9.5 h (Fig. 3B) and may indicate transcriptional bursting (8, 19), which results in a bimodal response in CsrB expression. These populations were not observed at the same time points when using LNA probes specific for the 6S and TPP-2 ncRNAs, indicating that the changes are only in CsrB expression and are not due to global cellular changes. These are informative differences that are undetectable in widely used ensemble measurements, such as qRT-PCR, that only reveal the averaged expression.

The flow-FISH method developed in this study uses LNA probes and posthybridization signal amplification to permit the specific and sensitive in situ detection of small ncRNA transcripts. This method is adaptable for the detection of different species of ncRNA or mRNA molecules in many cell types and is amenable to multiplexed measurements. Furthermore, not only does this method allow for the high-throughput analysis of single bacterial cells but it also provides the potential to sort differing subpopulations for further experimentation and offers this capability without the need to develop custom software or experimental setups. These attributes suggest that LNA flow-FISH may be a tool that can be used to develop a better understanding of the heterogeneity of transcriptional processes in natural biological systems and may also have value in assessing the transcriptional response and function of synthetic biological constructs.