Enterococcal PrgU Provides Additional Regulation of Pheromone-Inducible Conjugative Plasmids

ABSTRACT Efficient horizontal gene transfer of the conjugative plasmid pCF10 from Enterococcus faecalis depends on the expression of its type 4 secretion system (T4SS) genes, controlled by the PQ promoter. Transcription from the PQ promoter is tightly regulated, partially to limit cell toxicity caused by overproduction of PrgB, a T4SS adhesin. PrgU plays an important role in regulating this toxicity by decreasing PrgB levels. PrgU has an RNA-binding fold, prompting us to test whether PrgU exerts its regulatory control through binding of prgQ transcripts. We used a combination of in vivo methods to quantify PrgU effects on prgQ transcripts at both single-cell and population levels. PrgU function requires a specific RNA sequence within an intergenic region (IGR) about 400 bp downstream of PQ. PrgU interaction with the IGR reduces levels of downstream transcripts. Single-cell expression analysis showed that cells expressing prgU decreased transcript levels more rapidly than isogenic prgU-minus cells. PrgU bound RNA in vitro without sequence specificity, suggesting that PrgU requires a specific RNA structure or one or more host factors for selective binding in vivo. PrgU binding to its IGR target might recruit RNase(s) for targeted degradation of downstream transcripts or reduce elongation of nascent transcripts beyond the IGR. IMPORTANCE Bacteria utilize type 4 secretion systems (T4SS) to efficiently transfer DNA between donor and recipient cells, thereby spreading genes encoding antibiotic resistance as well as various virulence factors. Regulation of expression of the T4SS proteins and surface adhesins in Gram-positive bacteria is crucial, as some of these are highly toxic to the cell. The significance of our research lies in identifying the novel mechanism by which PrgU performs its delicate fine-tuning of the expression levels. As prgU orthologs are present in various conjugative plasmids and transposons, our results are likely relevant to understanding of diverse clinically important transfer systems.

IMPORTANCE Bacteria utilize type 4 secretion systems (T4SS) to efficiently transfer DNA between donor and recipient cells, thereby spreading genes encoding antibiotic resistance as well as various virulence factors. Regulation of expression of the T4SS proteins and surface adhesins in Gram-positive bacteria is crucial, as some of these are highly toxic to the cell. The significance of our research lies in identifying the novel mechanism by which PrgU performs its delicate fine-tuning of the expression levels. As prgU orthologs are present in various conjugative plasmids and transposons, our results are likely relevant to understanding of diverse clinically important transfer systems.
KEYWORDS type 4 secretion systems, conjugation, regulation H ospital-acquired (nosocomial) infections and antibiotic resistance are among the largest threats to global health according to the World Health Organization (1). Pathogens often acquire their resistance genes via transferable plasmids and other mobile genetic elements (2). A common opportunistic pathogen in nosocomial infections is the Gram-positive bacterium Enterococcus faecalis. Many clinical isolates of E. faecalis harbor conjugative plasmids. These plasmids contain type 4 secretion systems (T4SS) that are responsible for their own horizontal gene transfer. They most often also contain genes encoding resistance to antibiotics and various virulence factors. One of the best-studied E. faecalis plasmids is pCF10, a member of the superfamily of pheromone-responsive plasmids. pCF10 codes for resistance to tetracycline from an acquired Tn925 transposon, and its prgQ operon of ;30 kb encodes cell surface adhesins including PrgB (also termed aggregation substance [AS]) as well as the entire T4SS (Fig. 1A) (3,4).
The regulation of this system has been the topic of several reviews (3,4), and an overview is shown in Fig. 1B. Several negative regulatory checkpoints are involved. Transcription of the prgQ operon from the P Q promoter is governed by the ratio of two hydrophobic peptide sex-pheromones: cCF10 (called peptide C, for clumping; sequence: LVTLVFV) and iCF10 (called peptide I, for inhibiting; sequence: AITLIFI). Peptides I and C counteractively modulate the DNA binding activity of PrgX and, in turn, initiation of transcription from P Q . The intergenic region (IGR) positioned between prgQ and the prgRST genes carries two potential transcription terminators, IRS1 and IRS2. These two terminator sites, especially IRS2, have previously been indicated to be . Note that prgU is situated within the cassette containing the surface adhesin, even though PrgU belongs to the regulatory proteins. (B) The inhibitory peptide I is transcribed from the pCF10 plasmid (#1), whereas the inducing C peptide is transcribed from the genome of all E. faecalis cells (#2). Both I and C peptides are secreted to the outside of the cell, where the C peptide from donor cells gets partially degraded by the PrgY protease (#2) (38,39). Both I and C peptides are subsequently bound by PrgZ and imported via a permease (23). The transcriptional regulator PrgX is a dimer in its apo state but tetramerizes upon binding either I or C peptide. Depending on which pheromone is bound, it either represses transcription of the prgQ operon (PrgX/I) or induces it (PrgX/C) (#3) (22,38,40). Induction of the prgQ operon produces 3 transcripts, Q S , Q L and Full (#4) (40). The prgX operon also produces two transcripts, one of which is the anti-Q RNA that aids the formation of a terminator structure of the prgQ operon at the inverted repeat sequence 1 (IRS1) (#5) (7). In uninduced cells, anti-Q levels are sufficient to interact with all nascent prgQ transcripts, favoring formation of the IRS1 structure and transcript termination. Pheromone induction leads to the production of many more prgQ transcripts, which then overwhelm the pool of anti-Q. In the unpaired prgQ transcripts, IRS1 does not form, and transcription extends through the rest of the operon. The findings that we present in this article show that PrgU interacts with a region in the IGR, in between IRS1 and IRS2, thereby decreasing the expression of proteins transcribed from downstream genes (#6). relatively weak (5). Early studies also identified potential secondary structures resembling 23S and 16S rRNA helices in the IGR, which could serve as binding targets for ribosomal proteins (6). Northern blot analysis of pheromone-induced donor cells shows 2 predominant RNA species, Q S and Q L , whose 39 termini correspond to the position of IRS1 and IRS2, respectively, as well as extended transcripts that include the entire operon (7). A countertranscript RNA termed anti-Q regulates formation of the IRS1 terminator structure (7), and it is likely that IRS1 and IRS2 could also stabilize transcripts subject to degradation via 39-to-59 RNase activity, contributing to the total pool of these RNAs. Recently, a small (13-kDa) protein encoded by prgU has been shown to play a critical role in controlling overexpression of the prgQ operon (8).
PrgU has a pseudouridine synthase and archaeosine transglycosylase (PUA)-like fold as shown by its crystal structure (PDB: 2GMQ). PUA domains are conserved RNA binding motifs and can be found in all three kingdoms of life (9). They are often found in proteins involved in posttranscriptional modifications of tRNA and rRNA and in proteins involved in ribosome biogenesis and translation (10). Mutations in PUA motifs are reported to be involved in human diseases such as cancer and dyskeratosis congenita (11,12). The length of the PUA domains usually varies between 64 and 96 amino acids. In most known PUA-containing proteins, these RNA binding motifs are joined to a catalytic domain responsible for RNA modifications (13)(14)(15). In contrast, the PUA fold encompasses the entire PrgU sequence (119 amino acids), although PrgU has short insertions in a few loop regions not found in other PUA domains. Deletion of prgU from pCF10 leads to increased levels of the proteins encoded by the prgQ operon (Fig. 1B). This includes PrgB, which is toxic to the cells at high levels (8). Prolonged pheromone induction of pCF10DprgU mutants is therefore lethal, and the rare survivors contain suppressor mutations that display a noninducible phenotype (pCF10:: prgU Res ) (8). It has been suggested that other factors, such as the PrgR/PrgS proteins or bacterial host proteins, could coordinate with PrgU to regulate the expression of the prgQ operon (8).
Using a combination of in vivo and in vitro assays, we show that PrgU regulates expression of prgQ operon-carried genes. It does so by interacting with a specific sequence within the IGR, just downstream of the IRS1 sequence. Our cumulative data indicate that this interaction depends on another unidentified plasmid-or hostencoded factor(s). In bacteria, regulatory mechanisms controlling abundance of transcripts are well documented at the levels of transcription initiation and elongation/termination. Regulation at the levels of message stability and mRNA processing has also been well documented, although the details of these mechanisms are not as thoroughly understood. This report describes new information about the mechanism by which PrgU prevents toxic overexpression of a conjugation protein by degrading or inhibiting synthesis of transcripts extending past the IGR during induction of donor cells by a peptide sex pheromone.

RESULTS
PrgU suppresses expression of the prgQ operon. Because PrgU adopts a PUA fold implicated in binding rRNA or tRNA, we hypothesized that PrgU might exert its regulatory control by binding RNA structural folds within the IGR. To test this model, we first constructed reporter plasmids carrying the P Q promoter and a downstream lacZ reporter gene. One plasmid (pMC2) contains the IGR in between the promoter and lacZ, while another (pMC3) lacks the IGR. A third plasmid (pMC9) contains the IGR (like pMC2) but lacks the prgX gene ( Fig. 2A). We then monitored lacZ expression from these plasmids in induced OG1RF cells (2) or OG1RF cells harboring pCF10 or pCF10DprgU by a b-galactosidase activity assay (Fig. 2B). These strains also carried the vector pDL278p23, which has a constitutive P 23 promoter (designated P 23 ) or a variant of this plasmid, pMB11, that constitutively expresses prgU (designated P 23 ::U).
OG1RF cells with pMC2, which contains the IGR ( Fig. 2A), and the P 23 vector plasmid exhibited robust b-galactosidase activity. Isogenic strains that also contained P 23 ::U had only background levels of reporter activity (Fig. 2B, black bars). Introduction of pCF10 into OG1RF(pMC2, P 23 ) resulted in lower b-galactosidase activity, presumably due to PrgU production from native prgU. Reporter activity remained at background levels in cells carrying pCF10 and P 23 ::U due to abundant PrgU production. (Fig. 2B, middle, black bars). As expected, the mitigating effect of pCF10 on b-galactosidase activity was not observed upon introduction of pCF10DprgU into OG1RF(pMC2, P 23 ); in fact, this strain exhibited elevated b-galactosidase activity relative to OG1RF lacking pCF10 altogether (Fig. 2B, left and right, black bars).
In cells with pMC3, which lacks the IGR ( Fig. 2A), PrgU production from either pCF10 or P 23 ::U had no effect on b-galactosidase activities (Fig. 2B, compare dark gray bars, P 23 and P 23 ::U). Cells harboring pMC3 and pCF10 did exhibit slightly reduced b-galactosidase activities compared to cells with only pMC3. However, this effect was independent of PrgU, as pCF10DprgU caused a similar reduction of lacZ expression from pMC3 (Fig. 2B). We also observed that cells with pMC9, which lacks prgX, exhibited slightly higher b-galactosidase activity levels in the absence of PrgU overproduction, In pMC9 the prgX transcriptional repressor upstream from the P Q promoter is deleted. (B) b-Galactosidase activities originating from expression of the lacZ reporter gene on the pMC plasmids listed in panel A (coded in grayscale) in OG1RF cells without a pCF10 plasmid (2) or with wild-type pCF10 or pCF10DprgU. These cells also contain the pDL278p23 vector (P 23 ) or this vector constitutively expressing prgU (P 23 ::U). n = 3 independent biological replicates, and the error is the standard deviation. compared to pMC2, which carries prgX (compare P 23 black and light gray bars). This was expected because, in contrast to pMC2-carrying cells, the pMC9-carrying cells completely lack the transcriptional regulator PrgX. This lack of PrgX gives rise to a high constitutive transcription level of the P Q operon partly due to less anti-Q (16), in the OG1RF background (2), and lower PrgX levels in cells carrying coresident pCF10 or pCF10DprgU. Importantly, b-galactosidase activity from the pMC9 plasmid was still strongly reduced to near-background levels by PrgU overproduction (Fig. 2B, compare light gray bars, P 23 and P 23 ::U). Taken together, these results establish that PrgU is dependent on the presence of the IGR to inhibit expression of the downstream gene(s) and that PrgU acts independently of PrgX.
Effects of PrgU on Q transcripts in vivo. To evaluate the effect of PrgU on prgQ transcript levels, we used single-cell in situ hybridization chain reaction ( Fig. 3A and B). This technique both facilitates quantitative comparison of the levels of different transcripts in the same cell and determines the range of variation of expression dynamics within a population that is exposed to the same stimulus (17). The mRNA levels of fluorescently labeled prgB, an early gene in the operon behind the P Q promoter, and pcfG, a late gene in the same operon (11.5 kbp downstream of prgB), were examined in single cells. OG1RF cells carrying the following plasmids were analyzed: pCF10, pCF10DprgU, pCF10DprgU Res , and pCF10DprgABUC. pCF10DprgU Res has an additional mutation (prgRD271) that prevents C peptide-dependent induction of P Q transcription (8), and it served as a negative control together with uninduced cells (2, pCF10 in Fig. 3A and B). pCF10DprgABUC was used as an additional background control for prgB, as this variant lacks the prgB gene, as well as prgA, prgU, and prgC.
Cells with a mean fluorescence intensity higher than the threshold for either prgB or pcfG (Fig. 3A   were considered to be induced. As expected, in the negative-control strains there was no fluorescence observed for the prgB probe (with the exception of two of the 500 randomly chosen pCF10DprgABUC cells at 60 min), and very little for the pcfG probe (only two of the 500 randomly chosen pCF10DprgU Res cells above the threshold at 30 min). Also as expected, many pCF10DprgABUC cells displayed high fluorescence intensities for pcfG. These controls established that there is no overlap between the two hybridization probes and/or fluorescence channels, consistent with previous findings (18,19).
To study the effect of PrgU on prgB and pcfG transcript levels, we compared the levels of fluorescence in cells with pCF10DprgU to those in cells with wild-type pCF10. The cells from the pCF10DprgU strain had a tendency to aggregate and showed an increase in cell lysis upon induction, likely due to increased levels of PrgB as previously reported (8,20). This complicated the measurements in this strain and probably led to an underestimation of the number of induced cells (Table 1). However, the pCF10DprgU cells that were induced, and not aggregated, contained a larger amount of pcfG transcripts and kept those high levels for a longer time than wild type (compare pCF10 and pCF10DprgU at 60 min, Fig. 3B). The same trend, but not as prominent, can also be observed for prgB transcripts (Fig. 3A).
We observed that all pCF10-carrying cells with high levels of pcfG transcripts also had high prgB transcript levels, except for the pCF10DprgABUC-containing cells that lack prgB, whereas not all cells with high prgB levels had high pcfG levels (example shown in Fig. S1 in the supplemental material). This was noted previously and could be due to the fact that it takes the RNA polymerase roughly 7 min to traverse from the prgB gene to the pcfG gene (18).
Additionally, PrgU-mediated reduction of prgB transcripts in induced cells was also supported by transcriptome sequencing (RNA-seq) analysis of total mRNA from donor cells subjected to a 60-min induction by cCF10 pheromone (Fig. 3C and Data Sets S1 and S2). At this time point, wild-type cells have already progressed into the shutdown phase of the pheromone response as shown in the hybridization chain reaction (HCR) analysis ( Fig. 3A and B) and by previous findings (8,21). However, this RNA-seq data suggested that cells lacking PrgU retained about twice as much prgB mRNA as was observed for wild-type cells. We note that analysis of prgA transcript levels in these experiments is complicated by the fact that this gene and the adjacent prgT gene are subject to a significant basal level of transcription from a constitutive promoter located at the 59 end of prgT.
We further investigated the effect of PrgU on prgQ operon transcript levels with Northern blot assays. As shown in Fig. 1, transcription from P Q results in the formation of three transcripts, Q S and Q L, whose 39 termini correspond to the IRS1 and IRS2 structures in the IGR, respectively, and the full-length transcript encoding the surface adhesins and T4SS. We sought to compare Q S , Q L , and full-length transcript levels, but because the full-length transcript from pCF10 could not be resolved due to its length, we instead analyzed transcripts generated in strains harboring pMC2. This enabled detection of the full-length lacZ transcript as a proxy for the transcript of the full prgQ operon ( Fig. 2A). In total RNA extracts of OG1RF(pMC2) cells lacking prgU, the Q S and full-length [Full (pMC2)] transcripts were detected, but not Q L (Fig. 4, first lane). However, in extracts from cells expressing PrgU, either from the P 23 promoter or from the pCF10 plasmid, the Q L transcript was detected. Quantification of the bands showed that the amount of the Q S transcript in P 23 ::prgU, pCF10, and pCF10DprgU increased 3fold compared to OG1RF 1 pMC2. Full-length pMC2 transcripts were no longer detected in the strain that overproduced PrgU from the P 23 promoter, and the level of this transcript was reduced by ;40% in the strain with wild-type pCF10. Taken together, these data suggest that PrgU blocks the formation or enhances the breakdown of RNA transcripts downstream of IRS1/2. Since PrgU has a PUA fold that is predicted to bind RNA with rRNA-like structures, we hypothesized that PrgU binds Q L , which contains the whole IGR with the IRS1 and IRS2 site and possibly also to Q S , which contains over half of the IGR including the IRS1 site. To test this hypothesis in vivo, we conducted a pulldown experiment with total cell extracts of OG1RF(pCF10) strains engineered to produce FLAG-tagged PrgU or nontagged PrgU as a control. FLAGtagged PrgU was shown to be functional as it complements OG1RF(pCF10::DprgU). RNA recovered from the affinity pulldowns was subjected to Northern blot analysis with a probe designed to detect Q S and Q L . As seen in Fig. 5A, both Q S and Q L transcripts were identified in the pulled-down material in the presence of FLAG-tagged PrgU but, importantly, not with the nontagged PrgU control. By densitometry, we determined that the amounts of Q L transcripts were the same in the input and affinityenriched sample, but the level of Q S was reduced 5-fold in the affinity-enriched sample compared to the input. This suggested that FLAG-PrgU bound and pulled down more Q L (or extended Q transcripts processed back to Q L ) than Q S .
In separate pulldown experiments, we precipitated the affinity-enriched RNA for analysis by RNA-seq ( Fig. 5B and Data Set S2). The data showed significant RNAs from the IRS1 to IRS2 regions in all samples. In particular, a very prominent RNA species with a deduced 39 end was detected, located 15 bp downstream from the 39 end of the IRS1 sequence (note that the IRS1 structure is not generated in nonterminated prgQ transcripts). An RNA containing the same 39 end was also seen in the nontagged sample, at slightly lower levels, but not significantly enriched in the pulldown samples. At present, the biological significance of this RNA is not clear, but it could represent a stable processing product of a PrgU-dependent degradation pathway; this RNA would not FIG 4 PrgU expression reduces the formation of full-length [Full (pMC2)] transcript and leads to formation of Q L and, at high concentrations, an increase in Q S transcripts. Northern blot analyses of total RNA extracts probed with an oligonucleotide specific for the 59 end of the IGR or 5S RNA as a loading control. All strains analyzed were OG1RF cells containing the pMC2 plasmid, induced by addition of exogenous C peptide. In one strain, no additional plasmid was present (2). The other strains contained either the pMB11 vector that constitutively expresses prgU (P 23 ::prgU), wild-type pCF10, or pCF10DprgU. Note that Q S and Q L are also formed from the pCF10 plasmids, while the fulllength transcript produced from pCF10 is much larger and not visible here.
likely be detected in the Northern analysis because its 59 region does not extend to the sequence complementary to the Northern blot probe. Whether generation of this RNA requires PrgU will be an important question for future mechanistic studies. We note that RNA concentrations from PrgU-FLAG samples were ;10-fold greater than that of the untagged negative controls. This suggests enrichment of PrgU-bound RNAs relative to background material by FLAG-independent attachment to the beads used for pulldowns.
In vitro characterization of PrgU. The results of our studies described above suggested that PrgU interacts with the Q transcripts. To test this directly, we overproduced and purified PrgU to characterize its biochemical properties in isolation. PrgU, with an N-or C-terminal His tag (15.9 kDa), was produced in Escherichia coli and purified to homogeneity using a 2-step purification protocol. Both His-PrgU and PrgU-His eluted at the same volume on the size exclusion chromatograph (SEC). We then assessed the oligomeric state of purified PrgU-His by GEMMA (gas-phase electrophoretic molecular mobility analysis) and SEC-MALS (size exclusion chromatography-multiangle light scattering). The GEMMA indicates that PrgU-His exists as both a dimer and a monomer (Fig. 6A). The SEC-MALS experiment was done with a much higher protein concentration (;1 mg/ml versus 0.01 mg/ml for the GEMMA) and shows almost exclusively a dimer form of PrgU-His (Fig. 6B). Taken together, these data indicate that PrgU is in a concentration-dependent monomer-dimer equilibrium.
We assayed whether purified PrgU binds double-stranded DNA (dsDNA) by electrophoretic mobility shift assays (EMSAs). The results showed that PrgU binds a dsDNA substrate corresponding to the IGR with low-micromolar affinity but that it has a similar affinity for the control dsDNA substrate (Fig. 7A and Fig. S2). Next, we investigated the binding of PrgU to single-stranded DNA (ssDNA), dsDNA, and RNA forms of the IGR, as well as to shorter sequences corresponding only to the IRS1 and IRS2, by microscale thermophoresis (MST) (Table S1). However, under the tested experimental conditions, we were unable to detect binding to IRS1 or IRS2 (for neither RNA, ssDNA, nor dsDNA). We did observe binding to the longer IGR RNA, but PrgU bound equally well to control RNA of similar length (Fig. 7B). Our data therefore suggest that PrgU has low-micromolar affinity to both dsDNA and RNA but that binding is not sequence specific under our experimental conditions.

DISCUSSION
Regulation of the conjugative plasmid pCF10 by pheromone sensing has been extensively studied (3,4,22,23). The results have revealed a multitude of mechanisms to regulate the expression from the P Q promoter (3,4,21,22,24,25). More recently, it was discovered that PrgU, a previously uncharacterized protein, is also involved in this regulation. Bhatty et al. (8,20) showed that induction with the C peptide was toxic to cells carrying a pCF10 plasmid lacking prgU (pCF10DprgU). This toxicity was relieved by extragenic suppressor mutations or by PrgU overproduction, both of which blocked the production of Prg adhesins. Levels of PrgA, PrgB, and PrgC were increased in cells with pCF10DprgU, and conversely, levels were strongly reduced upon prgU overexpression. These findings led to the proposal that PrgU integrates with one or  more of the P Q repression systems to control the synthesis of the Prg surface adhesins upon pheromone induction (8). Importantly, prgU genes are widely spread on both plasmids and chromosomes in enterococci and are most often found together with a prgB gene (8). The information about PrgU presented here is thus not solely linked to the pCF10 plasmid, but the PrgU function likely plays an important role in various other plasmids and strains.
Here, we showed that PrgU reduces expression from genes downstream of the IGR (Fig. 2). In our reporter fusion studies and Northern blot analyses, we utilized lacZ reporter plasmids that lacked all prg/pcf genes downstream of the IGR. The observed effects of PrgU thus required only the P Q promoter, prgQ, and the IGR and no other gene products from the prgQ operon, including PrgR and PrgS, which were previously postulated to coordinate with PrgU to control P Q transcription ( Fig. 2 and 4) (8). Since none of the reporter plasmids contained a prgU gene (Fig. 2), these experiments also confirmed that expression in trans allows for full PrgU function. We also showed that PrgU regulation was unaltered by the presence or absence of prgX (Fig. 2), indicating that PrgU does not exert its negative control through effects on formation, stability, and/or function of the PrgX/C or PrgX/I complexes. From our experiments, we cannot rule out whether PrgU enhances P X transcription or leads to increased levels of the anti-Q transcript, but in view of our data we deem these options unlikely. Based on our findings, we therefore propose that PrgU functions independently of other known regulators of the pCF10 pheromone response system.
PrgU's control of expression of the prgQ operon depends on the presence of the IGR positioned between the P Q promoter and the regulated genes (Fig. 2). PrgU was observed to have a PUA-like fold (8), which is often involved in RNA binding and found in proteins that play a role in posttranscriptional modifications of tRNA and rRNA. This was of special interest, given previous evidence for control of prgB expression at both transcriptional and translational levels (6). The data from these previous experiments suggested that ribosomes, sequences in the IGR upstream from IRS1, and a 59 untranslated region between prgA and prgB are all involved in a novel translational regulatory mechanism unique to prgB. Conceivably, this poorly understood mechanism for regulation of PrgB synthesis could be related to the specific effect of PrgU on reduction of prgB mRNA and PrgB protein levels.
In addition to data from the transcriptional reporter studies, several other experimental findings support our working model that PrgU exerts its negative control through interaction with RNAs produced from the IGR. We showed that PrgU reduces full-length transcript levels by three independent methods by measuring (i) the transcript levels from an early gene and a late gene from the prgQ operon by HCR analysis (Fig. 3A and B) and (ii) the mRNA levels of prgB by RNA-seq data (Fig. 3C) and by (iii) Northern blot analysis of the RNA from a reporter plasmid (Fig. 4). The Northern blot analysis also showed that PrgU overproduction leads to an accumulation of Q S and Q L transcripts, presumably at the expense of the full-length transcript (Fig. 4). Results of the affinity pulldown assays further showed that both of these transcripts (especially Q L ) copurified with FLAG-tagged PrgU (Fig. 5A). Furthermore, RNA-seq analysis of the affinity-enriched RNA suggests that a predicted linear region just downstream of the IRS1 sequence may be the actual binding target for PrgU (Fig. 5B). We speculate that this RNA was generated by PrgU binding to this region just downstream of IRS1 on longer transcripts (note that the IRS1 terminator structure is unlikely to have been formed in these transcripts). Bound PrgU would then block 39-to-59 degradation by host RNases at this binding site. Alternatively, PrgU binding to this site on nascent transcripts could also inhibit further elongation via effects on IRS2 RNA polymerase processivity, which would lead to a similar outcome. We also observe the terminated transcript in the control pulldown with nontagged PrgU (Fig. 5B). It is important to note that untagged PrgU is also overexpressed in the nontagged cells and thus produces the same terminated transcript as in the cells with FLAG-tagged PrgU. However, as mentioned earlier, the pulldown with the FLAG-PrgU contained ;10 times more RNA than the control. Thus, we believe that much of the signal from the untagged samples presented in Fig. 5 likely resulted from nonspecific binding to the resin.
Both the reporter constructs (pMC2 and pMC9 shown in Fig. 2) that were negatively regulated by PrgU contain the predicted PrgU binding site. Interestingly, the essential target region of PrgU is upstream from the prgU gene. Such an arrangement could facilitate a timing delay, where PrgU-mediated negative regulation can take place only after a period of protein expression. This would first enable the production of adequate levels of the conjugation machinery for efficient plasmid transfer, followed by a reduction in transcript levels to prevent protein levels rising further and becoming toxic for the donor cells. This is supported by our HCR analysis (Table 1 and Fig. 3A and B).
Despite several lines of evidence suggesting that PrgU interacts with a specific RNA target sequence in the IGR, we were unable to confirm that purified PrgU specifically bound the IGR mRNA in vitro. EMSAs with dsDNA and MST data with RNA of the IGR and control sequences indicated that PrgU nonspecifically binds to both dsDNA and RNA with an affinity in the low-micromolar range (Fig. 7A and B). The affinity and specificity were the same with both N-and C-terminally tagged PrgU. We also know that FLAG-tagged PrgU can complement a DprgU mutant in vivo. This indicates that the purification tag did not affect binding (compare Fig. 7A and Fig. S2 in the supplemental material). The observed low-affinity and sequence-nonspecific binding to dsDNA could indicate that PrgU is concentrated around DNA, perhaps associated with active RNA transcription elongation complexes. This would allow it to quickly bind to the specific IGR binding site once it is produced to block further RNA synthesis or to recruit nucleases involved in RNA turnover. We suspect that specificity for RNA of the IGR could be imparted by another PrgU binding partner, which is missing in our in vitro experiments. The candidates include one or more host-encoded regulators or even RNA polymerase itself. Another possibility, raised by Bhatty et al. (8), is that PrgU stabilizes the interaction of anti-Q RNA or another regulatory RNA with Q transcripts (41), which we did not test for in our in vitro experiments. Alternatively, it is possible that PrgU stabilizes the IRS1 and IRS2 terminator structures and in that way inhibits the production of the fulllength transcript. We also determined that PrgU exists in a concentration-dependent monomer/dimer equilibrium in solution, which might be related to its biological function. PrgU dimerization might enable binding of two RNA target sequences as a prerequisite for effective regulatory control of downstream transcription or might be necessary for recruitment of coregulatory protein(s) or antisense RNA.
In summary, we have supplied several lines of evidence supporting our model that PrgU interacts with IGR mRNA and in so doing strongly reduces the accumulation of full-length transcripts. Although various mechanisms can be posited to account for PrgU's inhibition of full-length transcription, three of the more likely ones are (i) transcriptional blockade, (ii) strengthening the IRS1 and IRS2 terminator structures, and (iii) recruitment of RNase(s) for selective degradation. Experiments to determine the exact mechanism of PrgU, as well as the presumptive protein or RNA cofactor, remain exciting avenues for future research.

MATERIALS AND METHODS
E. faecalis strains, growth, and pheromone induction. The bacterial strains, plasmids, and oligonucleotides used in the study are listed in Tables S1 and S2 in the supplemental material. For plasmid construction, E. coli DH5a served as a host. E. coli strains were grown in Lysogeny broth (LB) (Difco Laboratories) at 37°C with shaking. E. coli strains were grown with the following antibiotics as needed: chloramphenicol (20 mg ml 21 ), erythromycin (100 mg ml 21 ), and spectinomycin (50 mg ml 21 ). E. faecalis strains were grown in brain heart infusion broth (BHI; Difco Laboratories) or M9 minimal medium (26) at 37°C without shaking. E. faecalis strains were grown with the following antibiotics as needed: erythromycin (100 mg ml 21 final concentration for plasmid markers, 10 mg ml 21 for chromosomal markers), fusidic acid (25 mg ml 21 ), rifampin (200 mg ml 21 ), spectinomycin (1,000 mg ml 21 for plasmid markers, 250 mg ml 21 for chromosomal markers), streptomycin (1,000 mg ml 21 ), tetracycline (10 mg ml 21 ), and chloramphenicol (10 mg ml 21 ). Antibiotics were obtained from Sigma-Aldrich.
Plasmid constructions. pMC1 is the shuttle vector plasmid pCI372 carrying prgX to prgA from pCF10 followed by lacZ with its own Shine-Dalgarno sequence. It was constructed by PCR amplification of the region of the IGR beginning at a natural XbaI site through the end of prgA using pCF10 as a template and amplification of lacZ from pCJK205. The two PCR products were digested with XbaI-BamHI and BamHI-SphI, respectively, and the resulting products were introduced into XbaI/SphI-digested plasmid pCIE.
pMC2 is pCI372 carrying prgX through the end of the IGR followed by lacZ. It was constructed by inverse PCR using pMC1 as a template and the primers listed in Table S2. pMC3 is pCI372 carrying prgX through the end of prgQ followed by lacZ. It was constructed by inverse PCR using pMC1 as a template and the primers listed in Table S2. pMC9 is pMC2 deleted of prgX. It was constructed by inverse PCR using pMC2 as a template and the primers listed in Table S2. pMC10 is plasmid pDL278p23 expressing prgU-FLAG from the P 23 promoter; it was constructed by inverse PCR using pMB11 as a template (8) and the primers listed in Table S2. pMC11 is plasmid pDL278p23 carrying the intergenic region (IGR) of pCF10; it was constructed by amplification of the IGR from pMC2, digestion with BamHI and SphI, and insertion into similarly digested pDL278p23.
To produce the His-tagged PrgU overexpression vectors for protein purification, p10-mini (8) was used as a template for PCR amplification. The prgU-containing PCR products were cloned into the initial cloning vector pINIT using SapI. The constructs were then transferred into p7XC3H or p7XNH3 via the FX cloning system (27).
b-Galactosidase assays. b-Galactosidase activity was assayed as previously described (28) with the following modifications: cells were cultured overnight in M9-yeast extract (YE) with selective antibiotics. The overnight culture was diluted 1:5 in fresh medium with 50 ng/ml cCF10 and grown for 90 min to mid-log phase. The cells were pelleted and resuspended in assay buffer (Z buffer: 60 mM Na 2 HPO 4 Á7H 2 O, 40 mM NaH 2 PO 4 ÁH 2 O, 10 mM KCl, 1 mM MgSO 4 , 50 mM b-mercaptoethanol, pH set to 7.0) to eliminate error due to the effects of different carbon sources in the growth medium on the b-galactosidase enzyme activity. Error bars indicate the standard deviation of three independent measurements.
Paraformaldehyde (PFA) with a final concentration of 4% was used to fix the cells followed by incubation overnight at 4°C before resuspending in phosphate-buffered saline (PBS) buffer supplemented with 2.7 mM KCl containing RNaseOUT (Invitrogen). HCR labeling was done as described before (19). In short, the cells were permeabilized and HCR labeling was done on cell suspensions. The hybridization probes were designed against an early gene (prgB) and a late gene (pcfG). The probes and hairpin amplifiers were obtained from Molecular Instruments (www.molecularinstruments.com). The B2H1 and B2H2 and the B5H1 and B5H2 amplifiers conjugated to Alexa Fluor 647 and 488 were used to detect prgB and pcfG transcripts. Cells were counterstained with Hoechst 33342.
Cells were then mounted as previously described (19). Images were acquired using a Zeiss Axio Observer.Z1 confocal microscope with an LSM 800-based Airyscan superresolution detector system (Zeiss). Confocal images were acquired through a 63Â 1.40-numerical-aperture (NA) objective (Zeiss) as z stacks at approximately 0.79-mm intervals using Zen software (version 2.1; Zeiss). Images were deconvolved and flattened using a maximum intensity projection (MIP) with Ortho Display, followed by image analysis by MATLAB (R2019b; MathWorks). For HCR analysis, the blue Hoechst fluorescence reference channel was used to determine pixel locations of individual cells and colocalized fluorescent overlap from HCR-labeled transcripts was quantified. The processed images (deconvolved, flattened by MIP, and subjected to background subtraction) were imported into MATLAB (R2019b; MathWorks) in a 16-bit TIFF format and analyzed as described previously (19). The reference channel images were binarized using Otsu's method (28) for thresholding. Objects between 3 and 50 pixels in size were analyzed as cells in the following analysis. HCR intensity value corresponding to each cell was calculated by taking the mean intensity of the pixels corresponding to each cell. To normalize the samples, 500 random cells from each image were used. The induction thresholds were decided based on the expression levels in the uninduced pCF10 population for pcfG, while for the threshold determination of prgB, the pCF10DABUC mutant was used. Graphs were created using GraphPad Prism 8.0.
Total RNA extraction from E. faecalis. E. faecalis strains were grown overnight in 5 ml M9 medium without antibiotics at 37°C. The culture was then diluted to an OD 600 of ;0.03 in 50 ml M9 medium and grown without shaking to an OD 600 of 0.1. Pheromone (cCF10) was added at a final concentration of 20 ng/ml, and cultures were grown to an OD 600 of 0.3. Cells were pelleted by centrifugation at 16,000 Â g at 4°C for 15 min. The pellet was resuspended in 1 ml of RNA-pro solution from a Fast RNA Blue kit (MP Biomedicals), and the mixture was transferred to a Fast Prep tube containing lysing matrix B. Cells were disrupted by bead beating five times at a setting of 400 rpm for 40 s with 1-to 2-min intervals on ice. Insoluble material was removed by centrifugation at 16,000 Â g for 5 min at 4°C, and the supernatant (;750 ml) was transferred to a fresh microcentrifuge tube and stood for 5 min at room temperature. After chloroform extraction, precipitation with absolute ethanol, and a final wash with cold 75% ethanol (made with diethyl pyrocarbonate [DEPC]-treated H 2 O), the RNA pellet was air dried and resuspended in 100 ml of DEPC-H 2 O. The RNA concentration was determined using a NanoDrop spectrophotometer, and RNA was stored at 280°C.
Northern blot analysis. RNA samples (2 mg) were loaded onto 10% Criterion Tris-buffered EDTA (TBE)-urea precast gels (Bio-Rad) and electrophoresed at 70 V until the dye front reached the bottom of the gel. RNA samples were transferred to a Zeta-Probe GT membrane (Bio-Rad) using a Trans-Blot SD semidry transfer apparatus (Bio-Rad) according to the manufacturer's guidelines. Transferred RNA was UV cross-linked and hybridized overnight with 100 ng/ml of 59-biotinylated probes (Table S2) in ULTRAhyb hybridization buffer (Ambion) at 42°C as described previously (29,30). Blots were developed using the BrightStar BioDetect kit protocol (Ambion) and imaged with a ChemiDoc MP imager (Bio-Rad). Quantifications of the bands were performed by densitometry tracing using Image J software.
PrgU-FLAG affinity pulldowns. E. faecalis strains were grown overnight in BHI (5 ml) and then diluted to an OD 600 of ;0.03 in 200 ml BHI and grown without shaking to an OD 600 of 0.1. cCF10 was added at a final concentration of 20 ng/ml, and cultures were grown to an OD 600 of 0.3. Cells were pelleted by centrifugation at 16,000 Â g at 4°C for 5 min, washed once with 10 ml Tris-buffered saline (TBS) buffer (50 mM Tris-HCl, pH 7.4; 150 mM NaCl), resuspended in 1 ml TBS, and transferred to a 1.5-ml Eppendorf tube. Cells were pelleted by centrifugation at 16,000 Â g at 4°C for 5 min, the supernatant was removed, and the cell pellet was flash frozen in a dry ice-ethanol slurry and stored at 280°C.
Cell lysates were prepared by suspending cells in 500 ml of TBS buffer with 5 ml of Halt protease inhibitor cocktail and 2 ml of SUPERase. Cells were added to a prechilled tube containing 500 ml of glass beads (0.1 mm) and disrupted by repeated cycles of maceration for 40 s at 400 rpm with intervals on ice. TBS buffer (500 ml) was added, and the slurry was vortexed for 30 s, after which cell fragments and other material were pelleted by centrifugation at 12,000 Â g for 30 min at 4°C. The supernatant was transferred to a prechilled tube, the volume was increased to 1 ml with cold TBS buffer, and 5 ml of SUPERase inhibitor was added.
An aliquot representing the input fraction was removed, and the remaining material was subjected to FLAG pulldowns by addition of 75 ml of anti-FLAG magnetic bead slurry to a 2.0-ml Dolphin tube (Costar 3213). The mixture was then placed on a nutator (rocker) for 2.0 h at 4°C, after which time the resin was pelleted by centrifugation at 4,000 Â g for 30 s at 4°C. The supernatant was removed, and the resin was washed 3 times with 1.5 ml of TBS buffer, with incubations on a nutator for 1 min, and then subjected to magnetic bead pulldowns at 4°C. Material bound to the FLAG beads was eluted by addition of 15 ml of 5-mg/ml 3ÂFLAG peptide solution and 250 ml of 3ÂFLAG elution buffer. Samples were incubated on a nutator at 4°C for 30 min and then centrifuged for 30 s at 4,000 Â g. The supernatant containing the eluted material was transferred to a new tube.
TBS (250 ml) and 3 M sodium acetate, pH 5.2 (50 ml), were added to the input and the eluted fractions (made as described above), and the RNA was isolated by acid phenol-chloroform extraction. The aqueous phase was transferred to a new tube containing 500 ml of neutral phenol-chloroform-isoamyl alcohol, followed by centrifugation at 14,000 rpm for 15 min. The supernatant was then mixed with an equal volume of isopropanol and placed at 280°C overnight. RNA was pelleted by centrifugation at 14,000 rpm for 15 min, washed once with 75% ethanol (EtOH), and air dried at room temperature. RNA was suspended in 20 ml DEPC-treated water and stored at 280°C until analysis via Northern blotting (see above).
RNA-seq analyses. For whole-cell RNA-seq, overnight bacterial cultures of OG1RF(pCF10) and OG1RF(pCF10DprgU) were diluted 1:20 in BHI medium and grown for 1 h at 37°C. Cells were then induced with 10 ng/ml cCF10 followed by incubation for 1 h at 37°C (31,32). rRNA was depleted using a MICROBExpress kit following manufacturer's instructions.
For RNA-seq on pulldowns, overnight cultures (OG1RF pCF10 carrying either pCIE::prgU or pCIE:: prgU-FLAG) were subcultured to an OD 600 of 0.03 in BHI. At an OD 600 of 0.1, cCF10 (20 ng/ml) was added. Cultures were grown to an OD 600 of 0.3 and pelleted (16,000 Â g at 4°C for 10 min). Cells were washed once with 10 ml 1Â TBS and once with 1 ml 1Â TBS, frozen on dry ice, and stored at 280°C. Cell pellets were resuspended in 500 ml TBS with 5 ml HALT protease inhibitor cocktail (ThermoFisher) and 2 ml SUPERase and then added to a prechilled tube containing 500 ml glass beads (0.1 mm). Tubes were placed in a BeadBug microtube homogenizer (Benchmark Scientific) for 5 cycles (40 s, 400 rpm), incubating on ice between cycles. Five hundred microliters TBS was added to each tube, and tubes were pelleted at 12,000 Â g for 30 min at 4°C. The supernatant was transferred to a prechilled microcentrifuge tube containing 5 ml SUPERase. Fifty microliters was removed as a protein sample, and 50 ml was removed for an "input" RNA sample. The remaining volume was incubated with 75 ml prewashed anti-FLAG slurry in a 2.0-ml tube for 2 h at 4°C. The resin was pelleted (4,000 Â g for 30 s at 4°C) and washed with 1.5 ml 1Â TBS three times. FLAG-tagged bacterial proteins were eluted with 3ÂFLAG peptide as follows. Two hundred fifty microliters elution buffer (15 ml of 5-mg/ml 3ÂFLAG peptide in 500 ml TBS) was added to the resin and incubated with rotation at 4°C for 30 min. After centrifuging the resin (4,000 Â g for 30 s at 4°C), the supernatant (containing eluted protein) was transferred to a new tube to which 50 ml 3 M sodium acetate (pH 5.2) was added. RNA was extracted with 1 volume acid phenol-chloroform. The aqueous phase was transferred to a new tube containing 500 ml neutral phenol-chloroform-isoamyl alcohol (25:24:1), extracted with 1 volume isopropanol, precipitated at 280°C overnight, and washed with 1 volume 75% ethanol. The pellet was dried and resuspended in 20 ml DEPC-treated water.
All RNA samples were treated with Turbo DNase (ThermoFisher; rigorous method described by the manufacturer) and submitted to the University of Minnesota Genomics Center for Illumina library prep. Whole-cell RNA samples were prepared using a nonstranded TruSeq RNAv2 kit, and libraries were size selected to generate inserts of 200 bp. Pulldown samples were processed using the Illumina TruSeq stranded mRNA preparation kit. To retain small RNAs, size selection was not performed between library preparation and sequencing for pulldown samples. All samples were sequenced as paired-end reads on an Illumina HiSeq 2500 in high output mode (125-bp read length for whole-cell samples and 50-bp read length for pulldown samples). Sequencing files were trimmed to remove contaminating adapter sequences and low-quality bases using Trimmomatic. Reads were mapped to the pCF10 reference sequence (NC_006827) using Rockhopper (33,34) and visualized using IGV (35). Rockhopper generates normalized expression values similar to reads per kilobase per million (RPKM) for all transcripts. The RNA-seq data, including RPKM and expression values, can be found (as Rockhopper output files) in Data Sets S1 and S2 in the supplemental material.
Protein production and purification. Both variants of PrgU, with either N-or C-terminal 10ÂHis tag and a 3C protease site, were produced in E. coli BL21(DE3) and grown in terrific broth (TB) medium in a LEX bioreactor (Epiphyte). The cells were grown at 37°C to an OD 600 of 1.0 to 1.5, and the temperature was then lowered to 18°C before induction with 0.5 mM IPTG (isopropyl-b-D-thiogalactopyranoside). The cells were incubated for 18 h at 18°C before being harvested via centrifugation and lysed in lysis buffer (50 mM KPi [pH 7.8], 150 mM NaCl, 15 mM imidazole [pH 7.8], and 10% glycerol) using a Constant cell disruptor (Constant Systems) with a pressure of 25 kilopounds/in 2 . Cell lysates were run over an immobilized metal-ion affinity chromatograph (IMAC) with Ni-nitrilotriacetic acid (NTA) beads using gravity flow and washed with 20 column volumes (CV) of wash buffer (50 mM KPi [pH 7.8], 500 mM NaCl, 50 mM imidazole [pH 7.8]) followed by 5 CV of wash buffer containing 2 M LiCl followed again by 20 CV of wash buffer. The proteins were then eluted with elution buffer (50 mM KPi [pH 7.0], 500 mM NaCl, and 500 mM imidazole [pH 7.0]). Eluted material was run over a size exclusion chromatograph (SEC) in 20 mM KPi (pH 7.0), 150 mM NaCl using either an S200 10/300 GL Increase (GE Healthcare) or an S75 10/ 300 GL (GE Healthcare) column.
GEMMA. Gas-phase electrophoretic mobility macromolecule analysis (GEMMA) on PrgU-His was performed as previously described (36). Briefly, the main peak of PrgU-His after SEC was collected and diluted to a concentration of 0.01 mg/ml in 20 mM ammonium acetate, pH 7.8. The sample was scanned 3 times to increase the signal-to-noise ratio.
Size-exclusion chromatography coupled to multiangle light scattering (SEC-MALS). Two hundred microliters of purified PrgU-His at 1 mg/ml was run over a Superdex 200 10/300 GL Increase column on an ÄKTA Pure system in line with Wyatt Treos II (light scattering) and Wyatt Optilab T-Rex (refractive index) in 20 mM potassium phosphate (pH 7.0), 150 mM NaCl. Astra software (Wyatt Technology; version 7.2.2) was used to collect and analyze the SEC-MALS data.
DNA and RNA oligonucleotides used for in vitro binding studies. The template used for producing the DNA of the IGR and the control was pCF10 or pRS01, respectively (Table S2). A touchdown PCR was done with Phusion polymerase and a starting annealing temperature of 72°C with a decrease of 1°C/cycle. The PCR products were cloned into pRAV23 (Addgene) using EcoRI and HindIII restriction sites and transformed into Top10 cells. The cells were grown followed by miniprep (Qiagen) for plasmid isolation. The plasmids were digested with EcoRI (Thermo Fisher) and HindIII (Thermo Fisher) using the Fast Digest buffer (Thermo Fisher) and used as a template for T7 in vitro transcription. The T7 reaction mixture consisted of 9 mM MgCl 2 , 4 mM ribonucleotide triphosphates (rNTPs), 1Â transcription buffer (Thermo Fisher), 2.5 to 5 mg DNA, and 0.05 mg/ml T7 RNA polymerase (Thermo Fisher) and in a 50-ml reaction volume, with incubation at 37°C for 1 h. The RNA was treated with proteinase K (Sigma) and DNase I (Sigma) according to manufacturer's instructions. Afterward, the RNA was isolated by successive rounds of EtOH precipitation, and the dried pellet was resuspended in double-distilled water (ddH 2 O) at the desired concentrations.
EMSA. Electrophoretic mobility shift assay (EMSA) was performed as described previously (37). DNA was from ThermoFisher or produced by PCR amplification with pCF10 Mini or pRS01 as the template (IGR and control DNA). DNA (100 nM) was mixed with increasing concentrations of PrgU with an N- (Fig. S2) or C-terminal (Fig. 7) His tag. Binding was allowed to occur for 20 min at room temperature (;25°C), and the complexes were then run at 4°C on 1% agarose gels at 100 V for 40 min. The visualization was done by poststaining with 3Â GelRed and recording on a ChemiDoc (Bio-Rad).
MST. The microscale thermophoresis (MST) was performed as follows. PrgU-His was labeled with the Red second-generation His tag labeling kit (NanoTemper Technologies) according to the manufacturer's instructions with a final concentration in the assay of 50 nM (molar dye/protein ratio of 4:1). A 16:1 dilution series was performed with RNA of IGR and a control sequence of the same length, with 40 mM as the highest concentration in the assay. The samples were mixed in a buffer containing 20 mM KPi (pH 7.0), 150 mM NaCl, 0.05% Tween 20, and 0.1 U/ml Protector RNase inhibitor (Sigma-Aldrich) and incubated for 10 to 20 min at room temperature. After incubation, the samples were quickly spun down and loaded into Monolith NT.115 standard capillaries (NanoTemper Technologies) at room temperature (;25°C). MST measurements were carried out with a Monolith (Monolith NT.115) using the instrument parameters of 60 to 100% LED power and medium MST power, the Red filter, and an MST on time of 20 s. Data were collected and analyzed using MO.Monolith Control software V1.6.1 and MO.Monolith Affinity Analysis V2.3. Three biological replicates were compared for each sample. Figures were made in GraphPad Prism 8.0.
Data availability. The data that support the findings of this study are available in the supplemental material of this article. RNA-seq data have been deposited with NCBI GEO under accession numbers GSE163794 and GSE168958.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.