Rho-dependent Termination of ssrS (6S RNA) Transcription in Escherichia coli

It is well known that 6S RNA, a global regulatory noncoding RNA that modulates gene expression in response to the cellular stresses in Escherichia coli, is generated by processing from primary ssrS (6S RNA) transcripts derived from two different promoters. The 5′ processing of 6S RNA from primary transcripts has been well studied; however, it remains unclear how the 3′-end of this RNA is generated although previous studies have suggested that exoribonucleolytic trimming is necessary for 3′ processing. Here, we describe several Rho-dependent termination sites located ∼90 bases downstream of the mature 3′-end of 6S RNA. Our data suggest that the 3′-end of 6S RNA is generated via exoribonucleolytic trimming, rather than endoribonucleolytic cleavage, following the transcription termination events. The termination sites identified in this study are within the open reading frame of the downstream ygfA (putative 5-formyl-tetrahydrofolate cyclo-ligase) gene, a part of the highly conserved bacterial operon ssrS-ygfA, which is up-regulated during the biofilm formation. Our findings reveal that ygfA expression, which also aids the formation of multidrug-tolerant persister cells, could be regulated by Rho-dependent termination activity in the cell.

binds to RNAP 70 -holoenzyme (E 70 ) and represses transcription from 70 -dependent promoters and activates S -dependent promoters (7)(8)(9). The 6S RNA is highly induced during stationary growth, during which time it plays an important role in modifying the utilization of E 70 to E S (7,8,10). The characteristic secondary structure of 6S RNA consists of a largely double-stranded helix and a central bulge (11,12). This highly conserved structure resembles DNA templates in terms of its open promoter complex and might therefore mimic DNA promoters to encourage the binding of E 70 (7,8). Interestingly, 6S RNA can act as a template for the transcription of pRNA, a very short RNA molecule (10).
6S RNA is cotranscribed with ygfA, a gene expressed as a second gene from the ssrS promoters, that encodes a putative 5-formyl-tetrahydrofolate cyclo-ligase (7,13). This dual-functional operon structure is highly conserved in ␣and ␥-proteobacteria, as well as in certain members of the class ␤-proteobacteria (14,15). Recent reports have suggested that ygfA expression aids the formation of persister cells (13) and is upregulated during the biofilm formation (16). The expression of ygfA is of particular interest, considering that the antibiotic recalcitrance of biofilm infections is largely caused by persister cells. However, the mechanisms of ygfA regulation remain unclear.
The transcription of primary 6S RNA transcripts and processing of their 5Ј-ends has been well characterized in E. coli (17). Two alternative promoters (i.e. a proximal canonical 70 -dependent ssrS P1 and a distal 70 -and S -dependent P2) are involved in the transcription of 6S RNA. These promoters produce two transcripts: a short P1 transcript that begins at nucleotide position Ϫ9 and a long P2 transcript that begins at position Ϫ224 relative to the mature 5Ј-end of 6S RNA (ϩ1). Interestingly, transcription from these promoters changes in response to cellular stress: P1 transcription is the predominant type of transcription during exponential phase, while P2 transcription increases upon entry into stationary phase (17). Furthermore, the 5Ј-ends of the transcripts are removed by two functionally similar enzymes with different specificities: RNase E acts on the long transcript, while RNase E and RNase G act on the short transcript (17). On the other hand, the mechanism responsible for the formation of the 3Ј-end of 6S RNA is not clear, although it has been shown that exoribonucleases are involved in a plausible 3Ј trimming mechanism that forms heterogeneous 3Ј-ends from ϩ184 to ϩ191 (17,18).
Considering that 6S RNA is transcribed as part of a dicistronic ssrS-ygfA message, the 3Ј processing of 6S RNA likely occurs via endoribonuclease cleavage followed by exoribonucleolytic trimming. However, this model is not consistent with past observations that formation of the 3Ј-end of 6S RNA is not altered in the endoribonuclease RNase E-, RNase G-, or RNase III-deficient cells (17).
In the present study, we found that in vitro transcription from the ssrS promoter continues beyond ygfA and can extend into sibC, a gene that encodes an anti-toxin small RNA (19,20). However, in vivo studies suggested that transcription from the ssrS promoters results in very low levels of ygfA expression. This low expression could reflect either internal cleavage of the polycistronic RNA or termination of transcription somewhere beyond the 3Ј-end of 6S RNA in the cell. We identified several Rho-dependent termination sites located ϳ90 bases downstream of the 3Ј-end of 6S RNA. No accumulation of Rho-terminated ssrS transcripts in cells lacking the four endoribocucleases, RNase E, RNase G, RNase III, and RNase P suggest that the 3Ј-end of 6S RNA is generated via exoribonucleolytic trimming after termination. The location of Rho-dependent termination sites within the N-terminal coding region of ygfA implies that ygfA expression depends on ssrS transcription continuing through the termination sites.
Preparation of Total Cellular RNA-E. coli-Overnight cultures grown at 37°C were diluted 1:100 in LB broth containing ampicillin (50 g/ml) or tetracycline (10 g/ml) and grown to an A 600 of 0.5 at the same temperature. If necessary, IPTG was added to the cell culture at 1 mM of final concentration and the culture was incubated further for 30 min. In the case of rne-1 or rnpA49 cells, the cells were grown at 30°C  GAA TGT GGC GCT CCG CGG TTG GTG AGT CAG  AAG AAC TCG TCA AG   Construction of MG1655 ssrS::kan   6SKO_dn  GCA GTT TTA AGG CTT CTC GGA CGG ACC GAG CAT GTA  TGG ACA GCA AGC GAA CCG  ssrS-92_F  CGG GGA TCC TTA CTT GAA CAA GGT  and then shifted to 44°C for 1 h. rng::cat cells also treated as above to compare with rne-1 effect even though the strain itself is not temperature-sensitive. Total cellular RNA was isolated by hot phenol extraction as described previously (25). Northern Blot Analysis-Total cellular RNA (15 g) was fractionated on a 5% polyacrylamide gel/7 M urea and electrotransferred onto a Hybond-XL membrane (GE Healthcare). 5Ј-End-radiolabeled oligonucleotides a6Sϩ185 and atRNA_ arg (Table 1) used as probes for 6S RNA and Brevibacterium albidum tRNA Arg , respectively. Hybridization was performed according to the manufacturer's instructions. The Northern blots were visualized and quantified using Image Analyzer FLA7000 (Fuji).
In Vitro Transcription by E. coli RNA Polymerase-The linear DNA templates used for in vitro transcription reactions were obtained with chromosomal or plasmid DNA via PCR with corresponding primer pairs (Table 1). Some primers had extra linker sequences at the 5Ј-end. In vitro transcription reaction was conducted using E. coli E 70 (Epicenter) according to the manufacturers' instructions with minor modifications. Briefly, the DNA templates (6 nM) were incubated at 37°C for 5 min in reaction buffer (40 mM Tris-HCl, pH 7.5, 150 mM KCl, 10 mM MgCl 2 , 0.05% Triton X-100, 10 mM DTT) containing E 70 (1 unit) and 2 mM ATP. The reaction was then initiated by adding mixtures containing rNTPs (0.5 mM rGTP, 0.5 mM rUTP, and 0.025 mM rCTP including 10 Ci of [␣-32 P]CTP) and rRNasin (4 units, Promega). After 25 min, the reactions were terminated by phenol-chloroform extraction and ethanol precipitation. The pellet was resuspended with gel-loading buffer II (Ambion) and loaded on a 5% polyacrylamide/8 M urea sequencing gel. The gel was visualized and analyzed by FLA7000 (Fuji).
Analysis of Transcription from ssrS-P1 Promoter in Vivo-ssrS-P1 promoter-containing DNA fragments (i.e. ranging from nucleotide position Ϫ92 to various downstream bases) were obtained via PCR amplification of genomic DNA with corresponding primer pairs ( Table 1). The resulting PCR products were digested with BamHI/HindIII and ligated into pKK232-8 (GE Healthcare) to generate pCAT plasmids. Cells containing fusion plasmids were exponentially grown in LB broth (1:300 dilution) of different chloramphenicol concentrations. The IC 50 was measured as previously described (26).
Alternatively, a B. albidum tRNA Arg sequence (27) was amplified by PCR with a corresponding primer pair (Table 1) and inserted immediately upstream of the CAT gene in the pCAT plasmids to measure the exogenous amount of tRNA Arg (i.e. as an indicator of transcription activity). Total RNAs isolated from cells containing the exogenous tRNA gene were subjected to Northern blot analysis as described above.
Analysis of Rho-dependent Termination-His-tagged Rho and IcdA (isocitrate dehydrogenase) proteins were purified from cells containing ASKA-rho and ASKA-icdA plasmids, respectively, as previously reported (28). The linear DNA templates were prepared and subjected to in vitro transcription analysis as described above except that Rho, or IcdA protein as a control, was added at 40 nM.
Gel Mobility Shift Assay-DNA templates for in vitro transcription of 6S RNA and precursor 6S RNA carrying the downstream sequence to ϩ330 were obtained via PCR using pCAT330 DNA with primer pairs of ERIT76S/a6Sϩ185 and ERIT76S/cat_R (Table 1), respectively. In vitro transcription was carried out using T7 RNA polymerase (Promega). Gelpurified 6S RNA and precursor 6S RNA were 5Ј-end labeled with [␥-32 P]ATP and re-purified by gel elution. Labeled RNAs (2 nM) were incubated with the purified Rho protein in 10 l of binding buffer (10 mM Tris-HCl, pH 8.0, 1 mM MgCl, 1 mM DTT, 100 mM NaCl, 0.06% Triton X-100, 0.25% glycerol, 10 g of ytRNA) for 15 min at 25°C. The reaction mixtures were then analyzed on 5% polyacrylamide gels, as described previously (20).
3Ј-RACE Assay-3Ј RACE analysis was performed on the in vitro transcribed RNAs in the presence of Rho protein with the ssrS-containing DNA ranging from Ϫ92 to ϩ448 as previously described (29), with following modifications. The Adaptor-ligated RNA was reverse transcribed and PCR amplified using a One-Step RT-PCR PreMix kit (Intron) according to the manufacturer's instructions. The PCR products were separated on a 2% agarose gel, purified, and analyzed by DNA sequencing after cloning into a pGEM-Teasy vector (Promega).
Analysis of Rho-knockdown Effects on ygfA Expression-Plasmid pAKA (20), a derivative of pACYC184, was used to construct an anti-rho RNA expression plasmid. The anti-rho RNA sequence linked to the rnpB terminator was amplified with a primer pair of ARrho1_F/pnMTXbaI_R by PCR and cloned into the AatII/EcoRI sites of pAKA so that anti-rho RNA could be induced by IPTG. Total RNAs were isolated from cells containing the anti-rho RNA expression plasmid pARRho after the IPTG induction for 30 min as described above, and treated with Turbo DNase (Ambion) to remove contaminating DNA. DNase was heat-inactivated and RNA samples were subjected to reverse transcription with MMLV RT (Enzynomics) with primers Rho216_R for rho mRNA, 6Sϩ733_R for ssrS-ygfA dicistronic RNA, and 5S_R for 5S rRNA. The resulting cDNAs were amplified using Taq polymerase premix (Solgent) with primer pairs Rho20_F/ Rho196_R for rho mRNA, 6Sϩ104_F/6Sϩ316_R for ssrS-ygfA dicistronic transcripts, and 5S_F/5S_R for 5S rRNA. The primers used were listed in Table 1. The PCR products were electrophoresed on 2% agarose gels, stained by SYBR Safe (Invitrogen), and photographed by GelDoc 1000 (Bio-Rad). Quantitative real-time RT-PCR was done by ABI 7500 Realtime PCR (Applied Biosystems) with QuantiTect SYBR Green PCR premix kit (Qiagen) in triplicate experiments. The abundance of each RNA was normalized to the amount of 5S RNA and represented as a fold change. Data were analyzed using ABI 7500 SDS software (Applied Biosystems, ver. 1.3). The cycle threshold (C T ) values obtained were an average of the triplicates.

RESULTS
In Vitro Transcription of ssrS Extends to sibC-Although it has been reported that ssrS and ygfA are co-transcribed, it remains unknown where this transcription is terminated. To examine the presence of intrinsic termination sites in vitro first, we prepared DNA templates that extended downstream of the SibC sRNA gene and used these templates for in vitro transcription (Fig. 1). The DNA constructs were designed to initiate transcription at the ssrS P1 promoter. Because all the DNA constructs had a 9-bp linker sequence at the ends, runoff transcripts were expected to have extra 9 nucleotides. The Constructs 1 and 2 produced transcripts (marked with a and b), with the same sizes as the P1 run-off transcripts of expected 256 and 866 nucleotides, respectively. Construct 4 contained the sibC transcription unit where transcription is initiated at position ϩ869 and terminates at ϩ1009 by an in-trinsic terminator (T sibC ) (19,20). This construct generated a transcript (marked with c) with the estimated size of 141 nucleotides. We did not observe apparent run-off transcripts of expected 181 nucleotides in this construct, suggesting that the sibC terminator is very effective. Construct 5 generated two transcripts (marked with d and c) of estimated 1018 and 141 nucleotides, which would initiate from the ssrS P1 and sibC promoter, respectively, but both of which would terminated at the T sibC . In Construct 6, which lacks the T sibC , we observed a run-off transcript (marked with e) of estimated 949 FIGURE 1. In vitro transcription from the ssrS P1 promoter. A, schematic arrangement of the ssrS transcription unit and DNA templates used for in vitro transcription. Two ssrS promoters, P1 and P2, and the downstream ygfA and sibC genes are indicated. Six DNA templates (i.e. 1-6) were prepared by PCR and used as DNA templates for in vitro transcription with 70 RNA polymerase. Numbers are given relative to the 5Ј nucleotide within the mature 6S RNA, which is considered ϩ1. The ssrS P1 promoter starts transcription at Ϫ9, while the sibC gene has a promoter (P sibC ) and an intrinsic terminator (T sibC ), which allow for transcription initiation and termination at ϩ869 and ϩ1009, respectively. B, in vitro transcription products were analyzed on a 5% polyacrylamide gel containing 7 M urea. The template numbers used for in vitro transcription are indicated above each lane. Major transcripts were marked with a to e. C, E. coli ssrS transcription unit. The 6S RNA and SibC RNA sequences encoded by ssrS and sibC, respectively, are shown in bold. Rho-dependent termination sites are indicated by vertical arrows and the sizes of arrows indicate the approximate relative magnitude of termination at each site, as determined by 3Ј-RACE. Boxed C residues indicate repeated C residues with 12 Ϯ 1 spacing required for Rho-dependent termination. S/D indicates a putative ribosome binding site of ygfA mRNA.
nucleotides that would begin at the ssrS P1 promoter. These in vitro transcriptional data suggest that transcription from the ssrS P1 promoter continues through to sibC and terminates at the T sibC in vitro.
In Vivo Transcription from ssrS P1 Promoter-Our in vitro transcription data revealed that transcription of ssrS extends to sibC (Fig. 1). To determine if this transcription extension occurs in vivo, we constructed reporter fusion plasmids by inserting ssrS P1 promoter-containing DNA fragments with unique downstream sequences into plasmid pKK232-8, which contained a promoterless CAT gene. We measured IC 50 in cells containing the ssrS-CAT fusion plasmids (Fig. 2). Interestingly, the IC 50 decreased significantly when the ssrS-CAT fusions contained the downstream sequences of 6S RNA beyond position ϩ238 up to ϩ260. The further decrease of IC 50 was observed when the downstream sequences were extended to or past position ϩ330. These data can be explained in two ways. First, it is possible that the in vivo transcription termination sites in the intergenic sequence between ϩ239 and ϩ330 resulted in ssrS transcription termination. Alternatively, ssrS transcription might extend into the downstream CAT gene, but the intergenic region would be susceptible to endoribonucleolytic cleavage and the cleaved 3Ј RNA fragment would be rapidly degraded. To discriminate between these two possibilities, we inserted a B. albidum tRNA Arg sequence between the ssrS sequences and the CAT coding sequence in the ssrS-CAT fusion constructs, and examined exogenous tRNA expression as well as 6S RNA in a 6S RNA knock-out strain (Fig. 3). We used heterologously expressed B. albidum tRNA Arg , which was previously shown to be metabolically stable in E. coli (30) and could be detected by Northern blot analysis without cross-hybridization with E. coli tRNAs. If our second hypothesis, of the susceptibility of the intergenic region to endoribonucleolytic cleavage and the rapid degradation of the cleaved 3Ј RNA fragment, were correct, we would expect to observe tRNA Arg expression in all the constructs even if the endoribonucleolytic cleavage could lead to RNA degradation. However, the amount of tRNA decreased in the presence of the sequences downstream of ϩ239 to ϩ260, and the decrease was prominent with the downstream sequences extended to or past ϩ330 (Fig. 3B), consistent with the IC 50 data obtained from the ssrS-CAT fusion constructs (Fig. 2). These findings suggested that the intergenic sequence between 6S RNA and CAT genes has in vivo termination sites rather than undergoing endoribonucleolytic cleavage. Therefore, it is likely that the in vivo termination sites lie within the sequence ranging from ϩ239 to ϩ330  (minor sites in the region of ϩ239 to ϩ260 and major sites in ϩ261 to ϩ330).
Rho Factor Is Involved in Termination of 6S RNA-It is well known that many in vivo transcription terminations are mediated by termination factors. As Rho factor is the major termination factor in E. coli (31), we examined whether Rho factor is involved in transcription termination in the region ranging from positions ϩ239 and ϩ330. We performed in vitro transcription assays in the presence of Rho factor (Fig. 4), using DNA templates spanning from Ϫ92 to ϩ238, Ϫ92 to ϩ330, and Ϫ92 to ϩ448, which were marked with 238, 330, and 448 above each lane of the figure, respec-tively. The Ϫ92 to ϩ330 and Ϫ92 to ϩ448 templates generated Rho-dependent termination products (marked with f), while the Ϫ92 to ϩ238 template did not. This is consistent with our in vivo results suggesting that the transcription termination is located beyond ϩ238 (Figs. 2 and 3).
We then examined whether this Rho-dependent termination occurs through binding of Rho protein to the 3Ј downstream sequences. Mature 6S RNA and precursor 6S RNA carrying the 3Ј downstream sequences to ϩ330 were synthesized in vitro and used for a gel shift assay with Rho protein.
Rho protein bound to the precursor 6S RNA, not mature 6S RNA (Fig. 4C), suggesting that Rho protein terminates the ssrS operon transcription through its binding to RNA. To identify the precise termination sites, we analyzed the in vitro transcripts produced in the presence of Rho protein by 3Ј RACE. The RACE data revealed the 3Ј heterogeneity of the in vitro transcripts, with the 3Ј-end of ϩ279 being predominant ( Table 2). These data suggest that transcription is terminated by Rho at multiple sites and that the major termination occurs at ϩ279 (Fig. 1C). We also found in vitro transcripts with 3Ј-end of ϩ253, ϩ255, ϩ260, which could explain the observation that the sequence of ϩ239 to ϩ260 provided a partial termination signal in the ssrS-CAT fusion assay (Figs. 2 and 3).
To determine whether the Rho-dependent termination takes place in vivo, we constructed a plasmid expressing antirho RNA by IPTG induction. This anti-rho RNA was designed to bind to the translation initiation region rho mRNA (the region of Ϫ22 to ϩ6 with the first nucleotide of the start codon as ϩ1) of rho mRNA so that the cellular level or translation of rho mRNA would be reduced. Cells containing the plasmid were treated with 1 mM of IPTG and total RNAs were subjected to RT-PCR to analyze the cellular levels of rho mRNA and ssrS-ygfA dicistronic RNA. The level of ssrS-ygfA dicistronic RNA was estimated by determining the amount of transcripts containing the region of ϩ104 to ϩ316 in the ssrS-ygfA transcription unit (Fig. 1C). The rho mRNA level was reduced to about 50% by induction of anti-rho RNA, while ssrS-ygfA dicistronic RNA as run-through transcripts at the Rho termination sites increased by more than 1.5-fold (Fig. 5), suggesting that ygfA expression could be regulated by Rho-dependent termination activity.
Rho-terminaterd ssrS Transcripts Are Not Accumulated in Endoribonuclease-deficient Cells-Our previous study showed that 3Ј processing of ssrS transcript was not affected in RNase E-, RNase G-, or RNase III-deficient cells (17), while Deutscher and co-workers (18) reported that multiple exoribonucleases can participate in the 3Ј trimming reaction of 6S RNA and that any one of five exoribonucleases, RNase II, D, BN, T, and PH, can carry out the trimming reaction although either RNase T or RNase PH appear to be the most effective a 3Ј RACE products from in vitro transcripts obtained from the Ϫ92 to ϩ448 template in Fig. 4 were cloned and analyzed by DNA sequencing. b The numbers in parentheses indicate the frequency of occurrence. exoribonucleases. They also showed that the mutant strains lacking four of the five enzymes accumulated 6S RNA with extra sequences of up to 6 nt at the 3Ј-end. Because mutant cells lacking all the five exoribonucleases are not viable and the enzymes functionally overlap in vivo (32), we thought that it would be impossible to observe the accumulation of Rhoterminated ssrS transcripts in any of exoribonuclease-deficient cells. Therefore, we re-examined whether Rho-terminated ssrS precursors would be accumulated in endoribonuclease-deficient cells. In this experiment, we included cells lacking the endoribonuclease RNase P, which had not been included in our previous work (17). First, the mutant strains lacking RNase E, RNase G, RNase III, and RNase P were converted to the corresponding 6S RNA knock-out strains by transduction with ssrS::kan from strain MG1655 ssrS::kan. In these 6S RNA knock-out mutant cells, we analyzed 6S RNA transcripts generated from ssrS-CAT fusion plasmids (Fig. 6). The fusion plasmids generate only P1 transcripts without P2 transcripts, making it easier to analyze 3Јends of 6S RNA transcripts. All the mutant strains did not show any accumulation of Rho-terminated transcripts of about 280 nucleotides from plasmid pCAT848 carrying the Rho-dependent termination site. On the other hand, plasmid pCAT238, which did not carry the termination site, generated a large precursor (marked with h) in rne-1 cells at 44°C, suggesting that this precursor needs the action of RNase E before the 3Ј trimming by exoribonucleases. All together, these data suggest that the four endoribonucleases are not involved in 3Ј processing of 6S RNA.

DISCUSSION
In this study, we found that ssrS transcription is terminated by Rho factor. This finding completes the current model of 6S RNA biogenesis in the cell (Fig. 7). Transcription of 6S RNA from the P1 and P2 promoters is typically terminated by Rho factor at position ϩ279, generating P1 transcripts of 288 nucleotides and P2 transcripts of 503 nucleotides. The mature 6S RNA then forms from these transcripts. Processing and maturation of the 5Ј-end occurs via different utilization of RNase E or RNase G on P1 and P2 transcripts (17), while processing at the 3Ј-end seems to precede this 5Ј processing because 5Ј precursors with the already matured 3Ј-end were observed as major ssrS transcripts in vivo (17). Both transcripts are subjected to 3Ј processing to remove ϳ90 extra nucleotides. Processing at the 3Ј-end appears to be carried out by exoribonucleases because premature 3Ј-ends of 6S RNA were observed in exoribonuclease-deficient cells (18), not in endoribonuclease-deficient cells such as RNase E Ϫ , RNase G Ϫ , RNase III Ϫ , and RNase P Ϫ strains (Ref. 17 and Fig. 6). However, we do not exclude the possibility that unidentified endoribonucleases are involved in 3Ј processing of 6S RNA before the final trimming although RNase I, known as a nonspecific endoribonuclease, appears to be irrelevant because 3Ј processing of 6S RNA was not affected in RNase I Ϫ cells (33).
We also found that mature 6S RNA was generated from two different CAT fusion transcripts containing ssrS sequences up to ϩ185 or ϩ238 (i.e. transcripts from pCAT185 and pCAT238), suggesting that correct 3Ј processing can occur even in the absence of Rho-dependent termination. Because a large 6S RNA precursor was observed in cells carrying pCAT238 in rne-1 cells at the nonpermissive temperature, RNase E might assist 3Ј processing to promote the generation of mature 6S RNA. In addition, the formation of mature 6S RNA from pCAT185 and pCAT238 implies that 3Ј processing is guided by the structural sequence of 6S RNA rather than by extra 3Ј sequences.
A recent genome wide study found a class of Rho-terminated loci leading to transcription termination of noncoding RNAs including tRNAs, but not 6S RNA, at the ends of their genes (34). Therefore, the Rho-terminated ssrS locus is likely to belong to this class. Rho factor prefers to bind to naked, untranslated, and C-rich RNA, reiterating C residues with about 12 Ϯ 1 nucleotide-spacing (35)(36)(37). The Rho-dependent termination complex consists of a hexamer of Rho proteins that can contact at least six C residues; therefore, Rho-dependent termination requires a repeated C region of ϳ60 nucleotides in nascent RNA transcribed by the elongation complex. The 3Ј downstream region of the 6S RNA sequence contains C residues form positions FIGURE 5. Effect of rho-knockdown on ygfA expression. A, total cellular RNAs were isolated from JM109 cells containing the anti-rho RNA expression plasmid pARRho grown in the presence of 1 mM IPTG, and subjected to RT-PCR for analysis of the cellular levels of rho mRNA, ssrS-ygfA dicistronic RNA, and control 5 S RNA, as described under "Experimental Procedures." RT-PCR products were electrophoresed on a 2% agarose gel and stained with SYBR Safe (Invitrogen). The relative band intensities of the IPTG-induced samples to the non-induced ones are indicated below each lane. B, quantitative real-time RT-PCR analysis. The RNA samples of A were also analyzed by quantitative real-time RT-PCR. The abundance of rho mRNA or ssrS-ygfA dicistronic RNA was normalized to the amount of 5S RNA and depicted as fold-changes by the IPTG induction for cells containing each plasmid. ϩ193 to ϩ313 with 12 Ϯ 1 nucleotide spacing (Fig. 1C). Therefore, it is likely that the C-residues ranging from ϩ193 to ϩ265 interact with Rho factor to terminate the transcription. Rho-dependent termination sites are located within the ygfA open reading frame. The definitive gene function of ygfA is not yet known, but encodes a protein with a high degree of sequence identity to mammalian 5-formyl-tetrahydrofolate cyclo-ligase (7, 13). As ygfA does not have its own promoter, its expression should depend on transcription initiation from the ssrS promoters (14,17), as well as antitermination at Rho-dependent termination sites. Consequently, ygfA expression can be regulated by the initiation rate from the ssrS promoters and the activity of Rho factor in the cell. When a rho-knockdown experiment was performed by expressing anti-rho RNA, we observed the increase of ssrS-ygfA dicistronic transcripts formed by read-through transcription at the Rho-dependent termination sites. Recently, Nudler's group also reported a significant increase in ygfA mRNA when RNAs from cells treated with bicyclomycin, which inhibits Rho factor function, were subjected to microarray analysis (38). It is likely, therefore, Rho factor participates not only in transcription termination for generating 3Ј-ends of 6S RNA from the termination sites but also in regulating ygfA expression. It is noteworthy that ygfA expression is implicated in the formation of persister cells, which contribute to the antibiotic resistance of biofilm infections (13). Moreover, several microarray analyses showed that ygfA gene is up-regulated during the biofilm formation (16,39,40). Thus, antitermination of ssrS transcripts at Rho-dependent termination sites under some Rho-factor function suppressing conditions could play an essential role in persister cell and biofilm formations.