LcrQ Coordinates with the YopD-LcrH Complex To Repress lcrF Expression and Control Type III Secretion by Yersinia pseudotuberculosis

ABSTRACT Human-pathogenic Yersinia species employ a plasmid-encoded type III secretion system (T3SS) to negate immune cell function during infection. A critical element in this process is the coordinated regulation of T3SS gene expression, which involves both transcriptional and posttranscriptional mechanisms. LcrQ is one of the earliest identified negative regulators of Yersinia T3SS, but its regulatory mechanism is still unclear. In a previous study, we showed that LcrQ antagonizes the activation role played by the master transcriptional regulator LcrF. In this study, we confirm that LcrQ directly interacts with LcrH, the chaperone of YopD, to facilitate the negative regulatory role of the YopD-LcrH complex in repressing lcrF expression at the posttranscriptional level. Negative regulation is strictly dependent on the YopD-LcrH complex, more so than on LcrQ. The YopD-LcrH complex helps to retain cytoplasmic levels of LcrQ to facilitate the negative regulatory effect. Interestingly, RNase E and its associated protein RhlB participate in this negative regulatory loop through a direct interaction with LcrH and LcrQ. Hence, we present a negative regulatory loop that physically connects LcrQ to the posttranscriptional regulation of LcrF, and this mechanism incorporates RNase E involved in mRNA decay.

levels of yopD, yopE, and yopH genes in YpIII parental strain overexpressing lcrQ. Elevated cytoplasmic LcrQ abrogated mRNA levels of these genes under T3SS-induced conditions (Fig. 1A), which corroborated other reports (34,35). We next aimed to identify the regulatory element targeted by LcrQ. For this purpose, we used a transcriptional fusion assay. We constructed a series of chimeric clones composed of the promoter alone, the 59 UTR alone, or both promoter and 59 UTR of yopE and yopH genes in front of the promoterless lacZ reporter (Fig. 1B). Where the endogenous regulatory element was lacking, it was substituted by the equivalent element from the regulatory sequences of the lac operon (Fig. 1B). As shown in Fig. 1C, LcrQ did not repress the b-galactosidase activities of clones carrying the lac promoter fused with 59 UTR of yopE or yopH genes but significantly repressed the clones carrying promoters of yopE or yopH genes. Although we could not exclude the possibility that LcrQ may regulate expression of yopE and yopH through other regions (such as the coding region or 39 UTR), our data suggest that LcrQ can downregulate Yops expression by repressing the promoter activities of yop genes.
LcrQ represses the expression of the master transcriptional regulator LcrF. Since LcrF is the only transcriptional activator of the Ysc-Yop T3SS encoded on the pYV plasmid (16), we next asked if LcrQ could regulate the expression of lcrF. We first detected the mRNA levels of lcrF in LcrQ-overexpressed and DlcrQ strains. As expected, the mRNA level of lcrF was increased in a DlcrQ strain under T3SS-inducible conditions (Fig. S1A in the supplemental material), while it was repressed when LcrQ was overexpressed in the YpIII parental strain (Fig. 1D). To confirm this regulatory effect, we examined the LcrF protein levels by Western blotting in this strain overexpressing LcrQ. To facilitate LcrF detection, we inserted a Flag tag-encoding fragment at the 59 and 39 termini within the lcrF gene in cis in the YpIII genome. The transcription of lcrF mRNA was only slightly influenced by inserting this Flag tag at either end (Fig. S1C). However, the Flag::LcrF was barely detectable in Western blot assay using anti-Flag antibody (Fig. 1E), probably due to alterations in protein conformation or protein stability induced by the tag. Regardless, overexpression of LcrQ in these strains repressed the expression of recombinant LcrF in both Flag::LcrF and LcrF::Flag strains, which consequently abrogated T3SS production (Fig. 1E). These data taken all together confirmed that LcrQ downregulates the production of LcrF.
The negative regulatory role of LcrQ is dependent on a YopD-LcrH complex. Previous analyses have shown that LcrQ does not contain any DNA or RNA binding motif (21,23,31). Our recent study also indicated that LcrQ does not directly interact with LcrF (34). Therefore, we suppose that LcrQ may downregulate LcrF expression by interacting with other proteins. To test this hypothesis, we screened proteins interacting with LcrQ using a bacterial two-hybrid system configured to contain a library of about 60 ysc-yop T3SS functional genes derived from the pYV plasmid but excluding genes involved in plasmid replication. Interestingly, LcrQ interacted with itself ( Fig. 2A). Additionally, LcrQ interacted with SycH (pYV0020), SycE (pYV0024), LcrH (also known as SycD, pYV0056), and YscB (pYV0078) ( Fig. 2A and B), which are customized T3S chaperones specific to the secreted Yops.
To understand the relevance of LcrQ-T3S chaperone interactions, we first overexpressed LcrQ in mutants lacking these T3S chaperones or their cognate Yop substrate. As shown in Fig. 2C, only deletion mutations of yopD (designated DyopD) or lcrH genes (DlcrH) abolished the downregulation function by LcrQ. Moreover, overexpression of LcrQ in the absence of YopD or LcrH could not inhibit the accumulation of lcrF-, yscW-, and yopE-specific mRNA ( Fig. 3A and Fig. S1B). This suggests that the negative regulatory role of LcrQ depends upon the presence of functional YopD and LcrH.
During this analysis, it became evident that the intracellular level of LcrQ was much lower when overexpressed in the DyopD or DlcrH background than the wild-type (WT) background (Fig. 3B). Consistent with this, a large portion of LcrQ was secreted into the supernatants of these mutants (Fig. 3B). Hence, it appears that a YopD-LcrH complex may inhibit LcrQ secretion. To explore this relationship, we appended the GST tag to the N terminus of LcrQ, which had been observed to abolish the secretion of YscM (a LcrQ homologue) (31). Surprisingly, a portion of GST-LcrQ was observed in the clear supernatant fractions of the DyopD or DlcrH strain, although not by the parental strain that contained functional YopD and LcrH (Fig. 3C). This is likely to be active secretion to the culture supernatant rather than by contamination of bacterial cellular material because cytoplasmic-located RpoA was not detected in our supernatant samples (Fig. 3C). Critically, GST-LcrQ trapped in the cytoplasm of the DyopD or DlcrH strain had no repressive effect on YopE synthesis (Fig. 3C), although it does repress both expression and secretion of YopE when overexpressed in the YpIII parental strain (34). Together, these data suggest that intracellular LcrQ functions through the YopD-LcrH complex, and this complex retains LcrQ in the bacterial cytoplasm.
Since intracellular LcrQ requires the presence of the YopD-LcrH complex for its negative regulatory role, we next tested if the repressive effect of intracellular YopD/LcrH requires the presence of LcrQ. Noticeably, overexpression of YopD and LcrH only slightly repressed Yops secretion and synthesis in a DlcrQ strain, whereas it caused a dramatic repression in the YpIII parental background ( Fig. 3D and E). On the other hand, lcrF-specific mRNA was repressed in both the parental and the DlcrQ backgrounds upon YopD/LcrH overexpression (Fig. 3F). These data suggest that both LcrQdependent and independent pathways can promote the repressive effects of YopD-LcrH.
Mapping regulatory regions within LcrQ. In the absence of any predicted structural elements within LcrQ, we wanted to define regions that were important for its regulatory role. To facilitate this, we constructed an LcrQ-mCherry mutant library whereby 102 of 115 LcrQ residues were substituted for alanine. The remaining 12 preexisting alanine residues and the methionine initiation codon were left unchanged. Fusion to mCherry enabled convenient monitoring of the recombinant LcrQ mutant expression level. A biosensor assay based upon the lcrG promoter transcriptionally fused to promoterless lacZ was established as a screen for the repressive effect of LcrQ on T3SS expression. The repressive effect was determined by calculating the ratio of the fold repression relative to the respective LcrQ mutant expression level. As seen in Fig. 4A and Table S1, the relative repression fold of the three mutants, LcrQ F46A , LcrQ L68A , and LcrQ L102A , was considerably lower than observed for all other variants, including wild-type LcrQ. Hence, these three residues are important for the full repressive function of LcrQ. Interestingly, no single mutant totally abolished the repressive role of LcrQ (Fig. 4A). As a consequence, we constructed the F46A, L68A, and L102A mutations in double and triple combinations. This generated stable LcrQ variants with far greater regulatory defects, with the triple mutation combination, LcrQ F46A, L68A, L102A , being particularly defective (Fig. 4B). As expected, ectopic overexpression of this stable LcrQ F46A, L68A, L102A variant failed to repress the accumulation of lcrFand yopE-specific mRNA levels (Fig. S2) and the synthesis and secretion of Yops (Fig. 4C) under T3SS-permissive conditions. Hence, this scanning mutagenesis approach has identified crucial LcrQ residues that support its negative regulatory role.
Having identified LcrH as a novel regulatory target of LcrQ, we next examined if the single, double, and triple mutant combinations of LcrQ influenced the interaction with LcrH. Initially using the bacterial two-hybrid system, we found that LcrQ L68A maintained an ability  to engage with LcrH to a level observed for wild-type LcrQ (Fig. 4D). On the other hand, the single (LcrQ F46A and LcrQ L102A ) and double (LcrQ F46A, L68A , and LcrQ L68A, L102A ) mutant variants decreased the LcrQ-LcrH interaction as judged by a 2-to 3-fold reduction in reporter output (Fig. 4D). Furthermore, the double (LcrQ F46A, L102A ) and triple (LcrQ F46A, L68A, 102A ) mutation variants abrogated much of the interaction with LcrH (Fig. 4D). Critically, this was not due to protein instability because the fluorescence intensity of LcrQ F46A, L102A and LcrQ F46A, L68A, 102A in fusion with mCherry was comparable to wild-type LcrQ (Table S2). To further confirm these findings, we established a pulldown assay using strains producing His-tagged LcrQ variants together with either GST alone or a GST-LcrH fusion. GST-LcrH could be successfully coeluted with wild-type His-LcrQ but not with the His-LcrQ F46A, L68A, L102A variant (Fig. 4E). Crucially, GST alone did not coelute with either His-LcrQ variant (Fig. 4E). Moreover, neither GST-LcrH nor GST alone could bind to Ni-nitrilotriacetic acid (Ni-NTA) in the absence of His-LcrQ (Fig. S3). Taken all together, these data suggest that the residues at positions 46 and 102 are critical for interacting with LcrH, and this interaction permits LcrQ to exert a negative regulatory role. Intriguingly, we also identified position 68 to influence this LcrQ regulatory capacity, but this may occur independently of the LcrQ-LcrH pathway.
RNase E contributes to negative regulation of LcrF through LcrQ and YopD-LcrH interactions. Since LcrQ cooperates with YopD-LcrH complex and the YopD-LcrH complex regulates T3SS posttranscriptionally (21,23), we next tested if LcrQ also participates in posttranscriptional regulation. Consistent with our hypothesis, deletion of lcrQ increased the stability of lcrFand yopE-specific mRNA, but not mRNA of the control fragment pYV0023 encoding a likely transposase remnant (Fig. 5A). To examine whether RNA decay factors are also involved in this negative regulatory circuit, we Mechanism of LcrQ Negative Regulation to T3S ® overexpressed LcrQ and YopD with LcrH in four different RNase mutant strains, Drne, Dpnp, Drnr, and Drnb (36). As seen in Fig. 5B and C, the repressive impact on Yops secretion normally caused by accumulation of either LcrQ or the YopD-LcrH was diminished specifically in the Drne strain lacking RNase E production. This correlated with the observation that lcrF-specific mRNA was higher in this mutant than the WT strain (Fig. 5D). Crucially, overexpression of LcrQ in the Drne strain was less effective at repressing lcrF mRNA levels (4-fold reduction) than in the WT strain (9-fold) (Fig. 5D). Moreover, YopD/LcrH overexpression in the Drne strain had no repressive impact on lcrF-specific mRNA levels compared to the WT strain (Fig. 5D). Hence, the RNase E mRNA decay factor influences the negative role of LcrQ and YopD-LcrH complex.
We wondered if this association was through a direct interaction between these proteins. Using a bacterial two-hybrid system assay, we found that YopD did not show any direct interaction with RNase E, but LcrQ and LcrH can both interact with RNase E and its associated protein RhlB (Fig. 5E). Importantly, the regulatory-deficient LcrQ mutants F46A, L68A and L102A, in either single, double, or triple combination, could all still interact with RNase E or RhlB (Fig. S4). Hence, RNase E or RhlB do not compete with LcrH for the same binding sites on LcrQ. Taken altogether, these data indicate that RNase E is an important contributor to Ysc-Yop T3SS downregulation by LcrQ and YopD-LcrH control in pathogenic Yersinia. Further, it is likely that RNase E works through interactions with LcrQ and LcrH.

DISCUSSION
A number of studies have highlighted the important regulatory role played by LcrQ in the control of Ysc-Yop T3SS by Yersinia (21,31,34). However, detailed knowledge of the molecular mechanism is lacking. In this study, we demonstrated that LcrQ inhibits expression of yop genes by downregulating the expression of lcrF encoding the master transcriptional regulator LcrF. This regulatory process depends on the presence of a posttranscriptional regulatory complex composed of YopD and LcrH. Furthermore, we demonstrated that coupling between LcrQ and this complex is achieved through a direct interaction of LcrQ with LcrH. Finally, these two proteins can both interact with RNase E, suggesting LcrQ, YopD/LcrH, and RNase E may combine to regulate T3SS in Yersinia.
Previous studies had indicated that the negative regulatory role of LcrQ may require the presence of the YopD-LcrH complex (22,35), but no direct mechanism underlying this possible relationship had been demonstrated experimentally. Moreover, additional studies using an in vitro translation system demonstrated that YopQ translation repression by the YopD-LcrH complex required the LcrQ homologue, YscM1 (13,27). Herein, we bridge all these studies by identifying that LcrQ interacts with LcrH to facilitate the negative regulatory role of the YopD-LcrH complex. Critically, stable LcrQ variants unable to physically interact with LcrH could no longer exert a repressive role on the T3SS. These findings are supported by the observation that YscM interacts with LcrH in Y. enterocolitica (37,38). We speculate that the purpose of this interaction might be to influence mRNA stability. The basis for this idea stems from observing that both LcrH and LcrQ interact with RNase E and its associated protein RhlB. We propose a model that suggests this interaction facilitates lcrF mRNA degradation (Fig. 6). Our future experiments will strive to confirm this coupling. Interestingly, previous studies with YopD have indicated a role in mRNA stability (24)(25)(26). In fact, the recent work of Kusmierek and colleagues indicates that this process involves an intricate array of RNA binding proteins and degradation factors (26). Our work corroborates and extends these findings by suggesting that the mRNA stability function attributed to YopD may actually depend upon LcrQ-LcrH, which acts as a molecular scaffold to recruit RNase E in the vicinity of YopD (Fig. 6).
RNase E, which recognizes a specific AU-rich RNA motif (39,40), is an established regulator of T3SSs in different bacteria. However, the effects can be either repression, such as in Yersinia (26, 41) and enterohemorrhagic Escherichia coli (EHEC) (42,43), or activation such as with Pseudomonas aeruginosa (44). There remains a lack of detail surrounding the action of RNase E in these different modes of regulation; to fill these knowledge gaps is worthy of further studies. Our data indicate that RNase E is an important contributor to Ysc-Yop T3SS downregulation by LcrQ and YopD-LcrH control in pathogenic Yersinia. However, we also observed that the repressive effects of LcrQ and YopD/LcrH were not completely abolished in our Drne strain (Fig. 5). This is not so surprising given the multifactorial nature of RNase E function. For example, the basis of our Drne strain is an incomplete deletion caused by a 39 truncation of the rne gene (36). It is evident that the nature of the rne mutation, coupled to the expression of other RNases in the organism, can affect the phenotypes displayed by rne mutants with respect to RNA degradosome assembly, mRNA turnover, maturation of rRNA and tRNA precursors, processing and degradation of regulatory RNAs, as well as rRNA quality control (45). Any of these situations may be at play in our Yersinia Drne background. Moreover, unidentified factors, such as additional RNA binding proteins, may also be involved in the regulatory roles of LcrQ and YopD/LcrH. Hence, further studies of our Drne mutant will likely identify additional players in the posttranscriptional regulation of lcrF expression and its impact on T3SS control by pathogenic Yersinia.
Interestingly, others implicate one other RNA stability factor, PNPase, in the control of T3SS in Yersinia (46,47). In particular, secretion of YopE and YopD were inhibited in the absence of PNPase, but only upon a short exposure of bacteria to T3SS-inducing conditions (47). Intriguingly, prolonged exposure did not result in any defect, and this is consistent with our data (Fig. 5). Subsequently, however, PNPase was found to posttranscriptionally regulate lcrF expression through YopD (26). Yet, in our hands, an overexpressed YopD-LcrH complex still strongly repressed ysc-yop T3SS in our Dpnp mutant. These discrepancies probably reflect subtle genetic differences between the specific strains used in the various studies, which are impacted by the relative genes via a pathway that is either independent of LcrQ (1) or dependent on LcrQ (2). The LcrQdependent pathway also involves RNase E and its associated protein RhlB and possibly some other uncharacterized RNases. This involvement occurs via direct protein-protein interactions involving LcrQ with LcrH as well as LcrQ/LcrH with RNase E and RhlB. Importantly, the interaction between YopD-LcrH with LcrQ inhibits the secretion of LcrQ (x). LcrQ trapped in the cytoplasm subsequently promotes the repressive effect of the YopD-LcrH-LcrQ complex in a feedback pathway. expression levels of the various RNases comprising the RNA degradosome. It also suggests that the role of PNPase in this regulatory process may not be a dominant feature in all Yersinia strains.
Another aspect of this study was the observation that YopD-LcrH complex can retain cytoplasmic pools of LcrQ. This is probably a consequence of the direct interaction between LcrQ and LcrH. This corroborates specific secretion of LcrQ occurring from regulatory-deficient mutants of yopD and lcrH when grown in the nonpermissive secretion conditions of plus Ca 21 (25,48,49). Interestingly, reciprocal experiments showed that YopD was specifically secreted in a DlcrQ strain grown in the same nonpermissive conditions (21,29). This suggests that LcrQ may also retain critical cytoplasmic levels of YopD. The accumulation of cytoplasmic levels of both LcrQ and the YopD-LcrH complex would facilitate the repression of T3SS under noninducible conditions (Fig. 6). As LcrQ secretion is an obvious checkpoint in orchestrated control of Yop synthesis and secretion, an analysis of the LcrQ secretor domain is warranted. Precedent for the value of this type of study comes from an analysis of the equivalent YopD secretor domain that revealed features setting it aside from a classical T3SS substrate signal, including possible yopD translation control mechanisms (50).
Interestingly, we show that the negative regulatory function of YopD/LcrH was not completely abolished in the absence of LcrQ ( Fig. 3D and E). However, the negative regulatory function of LcrQ was completely abolished in the absence of YopD-LcrH (Fig. 3B). This suggests that the regulatory role of LcrQ is strictly dependent on the presence of the YopD-LcrH complex, but the YopD-LcrH complex can function through both LcrQ-dependent and independent mechanisms. Our model of posttranscriptional regulation of lcrF expression reflects the involvement of these two pathways (Fig. 6). At this point, the reason for these two pathways and the relative contribution of each to regulatory control is not known. The LcrQ-independent nature of YopD function is thought to manifest itself in the form of translation inhibition of Yop synthesis by direct binding to yop mRNA (24), association with the 30S ribosomal subunit (27), and hijacking of global RNA regulators (26). However, these findings could be reinvestigated in light of LcrQ dependency.
Finally, we identified the LcrQ L68A variant that had decreased ability to repress Yops synthesis and secretion despite maintaining an interaction with LcrH, RNase E, and RhlB. Although our interaction assay does not measure productive binding, we suggest that the phenotype associated with the LcrQ L68A variant implies that LcrQ-dependent regulation must incorporate additional regulatory targets. In this context, we and others showed that LcrQ and/or YscM1/YscM2 can also directly interact with several other T3S chaperones, including SycH, SycE, SycO, and SycB (37, 51; this study). Furthermore, we demonstrated herein that LcrQ has potential to bind to itself. Despite the established importance of the LcrQ-SycH interaction to efficient LcrQ secretion (31,33), roles for the other interactions in T3SS biogenesis, function, and regulation are not well established. However, all these interactions have potential to function in this regulatory process. Having access to the regulatory-deficient LcrQ L68A -producing mutant may provide an important genetic tool to revisit the biological consequences of these binding phenomena.

MATERIALS AND METHODS
Plasmids, bacterial strains, and growth conditions. The Y. pseudotuberculosis YpIII and its derivate strains used in this study were cultured in YLB medium (1% tryptone, 0.5% NaCl, and 0.5% yeast extract) at 26°C. E. coli strains were grown in LB medium and incubated at 37°C for amplifying plasmids or at 20°C for protein expression. Ampicillin (100mg/ml), kanamycin (50 mg/ml), and chloramphenicol (30 mg/ml) were supplemented to the medium when needed. All bacterial strains and plasmids used in this study are listed in Table S3 in the supplemental material.
Plasmid construction. All oligonucleotides used in this study are listed in Table S4. To construct the LcrQ overexpression plasmid, the lcrQ gene was cloned into the pOVR plasmid (34) between the PstI and KpnI sites to obtain the plasmid designated pOVR-LcrQ. A gst-encoding region was amplified and inserted upstream of the lcrQ gene in pOVR-LcrQ. To overexpress the YopD-LcrH complex, the yopD and lcrH genes were both amplified and overlapped into one fragment using a ribosomal binding region as an internal linker. This overlapped fragment was then cloned into the pOVR plasmid. Clones composed of various promoter-lacZ transcriptional fusions were constructed based on the pZT plasmid as described earlier (23). The promoter and 59 UTR of yopH or yopE genes (35) were cloned upstream of promoterless lacZ using a ClonExpress II one step cloning kit (Vazyme). For the bacterial two-hybrid assay (52), genes were cloned into pKT25 or pUT18 using the ClonExpress II one step cloning kit (Vazyme).
Yops extraction and Western blotting assay. The Yops produced by various YpIII strains were extracted as previously described (34,53). Briefly, overnight cultures of YpIII strains in YLB were diluted (1:20) into Ca 21 -depleted medium (20 mM MgCl 2 and 5 mM EGTA) and cultured at 26°C for another 2 h. After that, cultures were transferred to 37°C and incubated for 4 h. Bacterial cell pellets were harvested by centrifugation. For each strain, an 8.1-ml supernatant fraction was carefully removed and then filtrated by a 0.22-mm filter to avoid bacterial contamination. Trichloroacetic acid (TCA) and acetone were used for protein precipitation from supernatant samples. The weights of bacterial cell pellets were determined for normalizing protein levels in bacterial pellets and supernatants. Proteins were dissolved in SDS-loading buffer and resolved by SDS-PAGE. For Western blotting, proteins resolved in SDS-PAGE were transferred into a polyvinylidene difluoride (PVDF) membrane (Millipore) by a semidry method. The membrane was then blocked with 5% nonfat milk. Protein-specific antiserum previously recovered from immunized rabbits (53) was diluted 1,000-fold and used to detect the protein levels of Yops. Mouse anti-Flag monoclonal antibody (1:2,000; Sigma) was used to detect the LcrF levels when it was fused with Flag tag. As appropriate, horseradish peroxidase (HRP)-labeled goat anti-rabbit or anti-mice IgG (1:10,000; Beyotime) was used as the secondary antibody. Enhanced chemiluminescence reagent (Bio-Rad) was used for signal generation. Image detection and collection used a ChemiDoc imaging system, and analysis was performed by the Image Lab software.
Protein purification and GST pulldown assay. E. coli strain BL21(DE3) was used for protein purification. The pET21a-LcrQ, pET21a-LcrQ3m, pGEX-KG, and pGEX-KG-LcrH plasmids were transformed into BL21(DE3) and the strains grown at 37°C in LB and incubated to an optical density of 0.4 at a wavelength of 600 nm. IPTG (isopropyl-b-D-thiogalactopyranoside) at a final concentration of 0.3 mM was used for protein production. Ni-NTA was used for His-LcrQ and His-LcrQ F46A, L68A, L102A (His-LcrQ3m) purification, and glutathione Sepharose was used for GST and GST-LcrH purification. For the pulldown assay, His-LcrQ, GST-LcrH, His-LcrQ3m, and GST-LcrH were incubated at 37°C for 1 h. Ni-NTA was used to trap the complex via the His tag. The combinations of His-LcrQ and GST alone, as well as His-LcrQ3m and GST alone, were used as negative controls.
YpIII mutant construction. YpIII mutants or strains with integration of Flag tag at the 59 end or 39 end of the lcrF gene were constructed using the suicide plasmid pDM4 (54) as previously described (55). Briefly, an ;500-bp fragment upstream and downstream of the region to be deleted was amplified, joined together by the two-step overlap PCR procedure, and then cloned into pDM4 plasmid. The pDM4 derivative was then transformed into E. coli S17-1lpir by chemical transformation and then conjugated into YpIII by conjugal mating. Allelic exchange by homologous recombination was screened as previously described (55).
RNA isolation and qRT-PCR. The culture conditions of strains were the same as used for Yops extraction. The TRIzol reagent (Ambion) was used for RNA isolation. The reverse transcription-quantitative PCR (qRT-PCR) assay was performed as described (56). Briefly, 2 mg DNase I (Promega)-treated RNA was used in reverse transcription assay with Moloney murine leukemia virus (M-MLV) reverse transcriptase (Promega). SYBR green supermix and CFX Connect fluorescence quantitative PCR detection system (Bio-Rad) were used in quantification assay. The copy number of 16S rRNA was used for normalization. For each gene expression analysis, at least three biological repetitions were performed, and each repetition contains two technical replicates.
RNA stability assay. The overnight cultures of YpIII strains in YLB were diluted (1:20) into fresh YLB with 20 mM MgCl 2 and cultured at 26°C for 2 h, after which they were transferred to 37°C and incubated for a further 2 h. Rifampin was then added to a final concentration of 500 mg/ml. After determined time points (0 min, 2 min, 4 min, 6 min, and 8 min), samples were collected in the presence of 0.2 volumes of stop buffer (5% water-saturated phenol, 95% ethanol) and snap frozen in liquid nitrogen. RNA was isolated as described above, and the mRNA stability was detected by gene-specific qRT-PCR, also as described above.
Bacteria two-hybrid assay. The adenylate cyclase-based bacterial two-hybrid system was used to detect protein-protein interactions (52). E. coli BTH101 was cotransformed with various pKT25 and pUT18 derivatives. Three colonies from each transformation were used for testing the b-galactosidase activity using ONPG (o-nitrophenyl-b-D-galactopyranoside) (Songon) as the substrate. The empty plasmid pair of pKT25 and pUT18 was used as the negative control, and the pKT25-Zip and pUT18-Zip plasmid pair was used as the positive control. The b-galactosidase activity was examined according to previous descriptions (57).
LcrQ mutant library screening. For LcrQ point mutation library construction, the lcrQ gene was first translationally fused at the C terminus with mCherry and cloned into the pBAD22 plasmid (58). The sitedirected point mutations of LcrQ were performed by following the protocol provided by QuikChange site-directed mutagenesis kit (Stratagene). All the altered amino acids were mutated to alanine (Ala). This mutant library was cotransformed with the pZT-lcrGp plasmid (34) into the DlcrQ mutant to test the repressive effect of the LcrQ protein. The b-galactosidase activity was monitored to indicate the lcrG promoter activity. The fluorescence intensity of mCherry (excitation and emission wavelengths are 587 nm and 610 nm, respectively) was measured by a microplate reader (Biotek) to indicate the expression level of the LcrQ variants. Three colonies were tested for each strain harboring a unique LcrQ Mechanism of LcrQ Negative Regulation to T3S ® variant. The relative fold repression of lcrGp by the LcrQ variants was calculated on the basis of lcrG promoter activity against the LcrQ expression level.
Statistical analysis. All data for the b-galactosidase activity assays were shown as mean 6 standard deviation (SD) of the results of multiple independent experiments. Statistical analyses were performed using the unpaired Student's t test (two-tailed) between each of two groups.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.