Targeted mutagenesis of intergenic regions in the Neisseria gonorrhoeae gonococcal genetic island reveals multiple regulatory mechanisms controlling type IV secretion

Summary

Gonococci secrete chromosomal DNA into the extracellular environment using a type IV secretion system (T4SS). The secreted DNA acts in natural transformation and initiates biofilm development. Although the DNA and its effects are detectable, structural components of the T4SS are present at very low levels, suggestive of uncharacterized regulatory control. We sought to better characterize the expression and regulation of T4SS genes and found that the four operons containing T4SS genes are transcribed at very different levels. Increasing transcription of two of the operons through targeted promoter mutagenesis did not increase DNA secretion. The stability and steady-state levels of two T4SS structural proteins were affected by a homolog of tail-specific protease. An RNA switch was also identified that regulates translation of a third T4SS operon. The switch mechanism relies on two putative stem-loop structures contained within the 5’ untranslated region of the transcript, one of which occludes the ribosome binding site and start codon. Mutational analysis of these stem-loops supports a model in which induction of an alternative structure relieves repression. Taken together, these results identify multiple layers of regulation, including transcriptional, translational, and post-translational mechanisms controlling T4SS gene expression and DNA secretion.

Introduction

The majority of Neisseria gonorrhoeae strains carry the 57 kb gonococcal genetic island (GGI) (Dillard & Seifert, 2001, Hamilton et al., 2005, Ramsey et al., 2011). The GGI contains genes for a type IV secretion system (T4SS), and sequence characteristics suggest that it was horizontally acquired (Dillard & Seifert, 2001, Hamilton et al., 2005). The gonococcal T4SS secretes single-stranded chromosomal DNA into the extracellular environment, and this DNA is effective for the natural transformation of other members of the population (Dillard & Seifert, 2001, Salgado-Pabón et al., 2007). The DNA secreted by the T4SS also plays a role in the initial stages of biofilm formation (Zweig et al., 2013).

Twenty-one genes in the GGI are required for DNA secretion. These genes encode proteins responsible for DNA processing and recruitment as well as structural components of the secretion apparatus (Figure 1) (Hamilton et al., 2005, Pachulec et al., 2014, Ramsey et al., 2011). The T4SS genes in the GGI are organized into four transcripts, with many of the genes ordered in a similar manner to those in F-plasmid (Hamilton et al., 2005, Pachulec et al., 2014, Ramsey et al., 2011). There are several notable differences in the gene organization and content between the GGI and F-plasmid, however, including the fact that the GGI does not encode homologs of the known F-plasmid regulators (for example, TraJ, TraY, and TraM) (Frost et al., 1994, Salgado-Pabón et al., 2010, Wong et al., 2012). Another difference is the genetic context of the genes encoding the relaxase and putative coupling protein (traI and traD), which in N. gonorrhoeae are divergently transcribed from the rest of the T4SS genes (Figure 1) (Frost et al., 1994, Hamilton et al., 2005, Salgado-Pabón et al., 2010).

Figure 1.

Diagram of the GGI. tra genes are indicated with an uppercase letter, and trb genes are indicated with a lowercase letter. The four promoters predicted to transcribe the genes required for type IV secretion are indicated by stem arrows. The ellipsis indicates a 29 kb region containing 33 genes, mostly of unknown function, that are not required for type IV secretion. Black arrows indicate genes that are required for type IV secretion. The relaxase (TraI) and coupling protein (TraD), are encoded on a 4-gene operon downstream of P_yaf. Many structural components of the secretion apparatus are encoded on a 19-gene operon downstream of P_ltgX. The 4-gene operon downstream of P_traH encodes TraH and TraG, predicted components of the secretion apparatus, as well as AtlA, a peptidoglycanase. Two homologs of partitioning proteins thought to be involved in recruiting DNA to the secretion apparatus (ParA and ParB) are encoded in an operon downstream of P_parA.

Previously, we observed that TraK and TraB, outer membrane proteins of the secretion apparatus, are expressed at very low levels (Ramsey et al., 2014). We hypothesized that expression of the transcript containing traK and traB was regulated to prevent high-level expression of non-variable surface-exposed proteins during infection (Ramsey et al., 2014). However, very little is known about the regulation of the T4SS genes in N. gonorrhoeae. Previous work showed that piliated gonococci secrete more DNA than nonpiliated gonococci and that expression levels of traD and traI are partially responsible for this difference (Salgado-Pabón et al., 2010). Given the divergent orientation of traD and traI in the GGI, it was hypothesized that DNA secretion might be regulated in gonococci via regulation of traD and traI (Salgado-Pabón et al., 2010). A short ORF (yaa) located in the same operon as traD and traI has also been shown to affect transcript levels of T4SS genes (Pachulec et al., 2014). In both cases, the precise regulatory mechanisms are unknown, and we hypothesized that additional unidentified factors likely contribute to T4SS gene expression.

In the current study, we investigate the expression and regulation of the T4SS genes and examine how changes in gene expression affect DNA secretion. We found that expression levels of the four transcripts containing T4SS genes varied, with genes on two of the transcripts (traG and parA) expressed at levels 25-fold higher than traK. By performing targeted mutagenesis on three of the four predicted T4SS promoters, we generated strains that overexpress subsets of T4SS genes. Although DNA secretion was variable in the overexpression strains, it was not significantly different from that of the WT strain, and we found that a periplasmic protease similar to E. coli tail-specific protease (Tsp) was involved in the degradation of T4SS proteins, suggesting that T4SS protein levels are closely monitored in the cell. We also discovered that the T4SS transcript containing traG is regulated by an RNA switch mechanism. Two stem-loops in the 5’ untranslated region (UTR) repress translation, but formation of an alternative stem-loop structure relieves the repression. Altogether, these results demonstrate that T4SS gene expression in gonococci is controlled by a complex regulatory program involving transcriptional, translational, and post-translational mechanisms.

Results

Targeted mutagenesis of the P_ltgXand P_yafpromoters in the yaf-ltgX intergenic region

The majority of the genes required for DNA secretion in gonococci are contained within two divergent operons beginning with the yaf and ltgX genes (Figure 1) (Pachulec et al., 2014, Remmele et al., 2014). The operon beginning with yaf contains 4 genes, including those encoding the relaxase and coupling protein. The operon beginning with ltgX contains 19 genes, many of which encode structural components of the secretion apparatus (Figure 1) (Pachulec et al., 2014). The 630-bp intergenic region between yaf and ltgX is predicted to contain the promoters for both operons, as well as the putative origin of transfer (Salgado-Pabón et al., 2007), suggesting that the yaf-ltgX intergenic region could play an important role in regulating DNA secretion (Figure 1 and Figure 2A).

Figure 2.

Mutagenesis of predicted promoters in the yaf-ltgX intergenic region. (A) Sequence of the yaf-ltgX intergenic region. The predicted ATG start sites for yaf and ltgX are underlined. The predicted −35 and −10 hexamers for the predicted P_yaf and P_ltgX promoters are boxed. The sequences of the modified P_yaf^cs and P_ltgX^cs promoters are indicated in the gray boxes, with the introduced mutations indicated in bold lower-case letters. The single base-pair insertion in strain AY503 is indicated. The primary transcription start site for the yaf transcript determined by Remmele et al. (2014) is indicated with an asterisk, while a secondary transcription start site is indicated by double asterisks. Possible transcription start sites identified by 5’ RACE for the ltgX operon are indicated with circles. A 22-bp sequence on the reverse strand that differs by only one base-pair compared to sequence A in the traH 5’ UTR is underlined with an arrow. (B) Transcript levels of traK, traI, traH, traG, atlA, and parA in the WT strain MS11 n.v.. Transcript levels are normalized to the housekeeping gene rpoB. The average ± standard deviation for at least 3 biological replicates is shown. **, significantly different (P < 0.001) by Student’s two-tailed t-test; *, significantly different (P < 0.05) by Student’s two-tailed t-test; #, significantly different (P < 0.01) from transcript levels of both traK and traI by Student’s two-tailed t-test. (C) Fold change in transcript levels of traK and traI in the promoter mutants MR589 (P_ltgX^cs), MR590 (P_yaf^cs), and MR591 (P_ltgX^cs P_yaf^cs) compared to the WT MS11 n.v. strain. The average ± standard deviation for at least 3 biological replicates is shown. §, significantly different from traK levels in MS11 n.v. (P < 0.05) by Student’s two-tailed t-test. *, significantly different from traI levels in MS11 n.v. (P < 0.05) by Student’s two-tailed t-test. (D) Western blots for TraI, TraK, and TraN in MS11 n.v. (WT), MR589, MR590, and MR591. Solid black triangles indicate full-length protein, while gray triangles indicate predicted degradation products.

A recent transcriptomic analysis identified two transcriptional start sites upstream of yaf: a primary start site located 504-bp upstream of the predicted yaf start codon (single asterisk in Figure 2A), and a secondary start site located 145-bp upstream (double asterisks in Figure 2A) (Remmele et al., 2014). The Remmele et al. study did not identify a transcriptional start site for the ltgX operon, perhaps due to the low expression levels of this operon. Using 5’RACE, we identified several transcription start sites for the ltgX operon, including bases at −7, −56, −79, and −110 relative to the ltgX coding sequence (circles in Figure 2A). By searching the intergenic region for sequences with similarity to sigma-70 consensus motifs, we identified two possible promoter elements (Figure 2A). These putative promoters are consistent with the transcription start sites at −145 for yaf and at −110 for ltgX.

Using qRT-PCR, we investigated transcript levels of traI (part of the yaf transcript) and traK (part of the ltgX transcript) in piliated WT cells. Levels of the traK transcript are six-fold lower than those of traI (Figure 2B), consistent with the fact that TraI is detectable in Western blots of WT cells, while TraK is not (Figure 2D) (Salgado-Pabón et al., 2007, Ramsey et al., 2014). We wondered whether we could increase expression of TraK and other structural T4SS proteins by mutagenizing the sequences of the predicted P_yaf and P_ltgX promoters. In addition to facilitating laboratory studies of the apparatus proteins, these strains would also allow us to determine how changes in T4SS protein stoichiometry affect DNA secretion. We introduced mutations into both P_yaf and P_ltgX on the gonococcal chromosome that changed the predicted −35 and −10 hexamers to sigma-70 consensus motifs (cs) and optimized the spacing between the hexamers from 16 to 17 bp (Figure 2A).

Introduction of four mutations in P_ltgX (P_ltgX^cs, strain MR589, Figure 2A) resulted in a 7.7 (±0.67)-fold increase in traK levels compared to WT (Figure 2C). Interestingly, traI levels were slightly but significantly decreased in the P_ltgX^cs strain compared to WT (0.78 (±0.06), Figure 2C). Consistent with the changes at the RNA level, the amount of TraI decreased slightly while TraK increased to a detectable level (Figure 2D). To confirm that the P_ltgX^cs mutations also affected expression of other genes in the operon, we investigated levels of TraN, encoded by the seventeenth of 19 genes in the operon (Pachulec et al., 2014). TraN, like TraK, is not detectable in WT cells and increased to a detectable level in the presence of the P_ltgX^cs mutations (Figure 2D). Depending on the particular Western blot, we observed two or three bands for TraN, the largest of which is predicted to be the full-length protein (black triangle, Figure 2D).

Introduction of four mutations in P_yaf (P_yaf^cs, strain MR590, Figure 2A) resulted in a 14 (±1.2)-fold increase in traI levels compared to WT and no significant difference in traK levels (Figure 2C). TraI levels also increased in the P_yaf^cs strain (Figure 2D). P_yaf contains an A₈ repeat in the spacer between the −35 and −10 elements (Figure 2A), suggesting that slipped-strand mispairing could regulate transcription from this putative promoter element. A survey of gonococcal genomes reveals sequences with 6, 7, or 8 A’s at this location, suggesting that this region does vary. To test how changes in spacing would affect downstream gene expression, we changed the spacing between the predicted −35 and −10 hexamers from 16 bp to 17 bp (strain AY503, Figure 2A). We performed qRT-PCR for the downstream gene traD, which is co-transcribed with traI (Salgado-Pabón et al., 2010), and observed a slight change in traD levels (0.66 ± 0.24 compared to WT). This difference was not statistically significant, however, suggesting that increasing the spacing between the −10 and −35 hexamers from 16 to 17 bp has minimal effects on transcription.

We also constructed a strain combining the P_yaf^cs and P_ltgX^cs mutations (P_yaf^cs P_ltgX^cs, strain MR591). Transcript levels of traK and traI increased by 7.9 (± 0.93)-fold and 9.0 (± 0.82)-fold, respectively, in the P_yaf^cs P_ltgX^cs strain compared to WT (Figure 2C). TraI, TraK, and TraN levels also increased in this strain compared to WT (Figure 2D).

The P_yaf^cs P_ltgX^cs strain does not exhibit a significant increase in DNA secretion

The P_yaf^cs P_ltgX^cs strain produces increased levels of T4SS proteins transcribed from the yaf and ltgX operons, and we hypothesized that it might therefore secrete increased levels of DNA. We performed a fluorometric DNA secretion assay to quantify the amount of DNA present in culture supernatants from a P_yaf^cs P_ltgX^cs strain (MR588). The amount of DNA in culture supernatants from the P_yaf^cs P_ltgX^cs strain was elevated relative to WT, but was highly variable and was not statistically significant (Figure 3). Both the WT and P_yaf^cs P_ltgX^cs strain secreted increased levels of DNA relative to the ΔGGI negative control (Figure 3). These results suggest that increasing the expression of the transcripts downstream of P_yaf and P_ltgX is not sufficient to significantly increase DNA secretion (Figure 3).

Figure 3.

Fluorometric detection of secreted DNA. Gonococcal culture supernatants were treated with the DNA-binding dye PicoGreen to determine the concentration of secreted DNA from strains ND500 (ΔGGI), MS11 (WT), MR588 (P_yaf^cs P_ltgX^cs), and MR674 (P_yaf^cs P_ltgX^cs UTR^m16). The average ± standard error of the mean of 6 independent experiments is shown. The average background fluorescence (as determined using the ΔGGI strain ND500) was subtracted from all samples. §, significantly different from ND500 (P < 0.05) by Student’s two-tailed t-test. N.S., not significant from MS11 by Student’s two-tailed t-test.

Stabilization of TraK by the addition of a C-terminal FLAG3 tag

Previously, we showed that addition of a triple FLAG epitope to the C-terminus of TraK made it possible to detect TraK-FLAG3 expressed from the native locus, presumably due to signal amplification from the triple epitope (Figure 4A) (Ramsey et al., 2014, Wayne et al., 2010). We also expressed traK-FLAG3 from the native locus in the P_yaf^cs P_ltgX^cs background (MR598). Interestingly, TraK-FLAG3 levels were higher than those we had observed for TraK in the P_yaf^cs P_ltgX^cs background (Figure 4A, compare MR598 with MR591 in the top panel). This difference is not attributable to signal amplification given that we used an antibody against TraK and not against the triple epitope tag (Figure 4A). Using qRT-PCR, we found that transcript levels of traK and traK-FLAG3 were not significantly different (Figure 4B). We hypothesized that the addition of the FLAG3 epitope to the C-terminus of TraK might be blocking proteolytic degradation of TraK in the P_yaf^cs P_ltgX^cs background.

Figure 4.

A C-terminal FLAG3 tag stabilizes TraK in the P_yaf^cs P_ltgX^cs background. (A) Western blots using the 2E5 α-TraK and M2 α-FLAG monoclonal antibodies (MAb) to investigate TraK and TraK-FLAG3 levels in several gonococcal strains. The strain backgrounds and the traK alleles are indicated. MS11 (WT), MR602 (WT, traK-FLAG3 at the native locus), MR591 (P_yaf^cs P_ltgX^cs), and MR598 (P_yaf^cs P_ltgX^cstraK-FLAG3 at the native locus). The gray triangle indicates a predicted TraK-FLAG3 degradation product. The Bradford assay was used to normalize the amount of protein analyzed from each sample. (B) Transcript levels of traK and traK-FLAG3 in the same strains as in part (A) normalized to the housekeeping gene rpoB. N.S., not significantly different. Transcript levels were determined using primers located in traK.

Tail-specific protease (Tsp) is involved in degradation of T4SS proteins

TraK and TraK-FLAG3 both localize to the outer membrane in gonococci (Ramsey et al., 2014), so we hypothesized that a periplasmic protease might be degrading TraK. One candidate was tail-specific protease (Tsp), which has a preference for unfolded proteins with small, non-polar C-terminal residues in E. coli (Silber et al., 1992, Keiler et al., 1995, Keiler & Sauer, 1996). The C-terminal sequence of TraK is relatively non-polar (FIISRPGG). The FLAG epitope (DYKDDDDK) is both charged and polar, which would make TraK-FLAG3 a poor putative substrate for Tsp. A search of the N. gonorrhoeae MS11 genome sequence identified one open reading frame (NGFG_01224) with similarity to E. coli tsp. The protein encoded by NGFG_01224 is annotated as a C-terminal processing peptidase and is 26% identical to Tsp from E. coli (determined by alignment with Clustal Omega). Like E. coli Tsp, NGFG_01224 has a predicted signal sequence and is therefore predicted to localize to the periplasm. The active site residues identified in E. coli Tsp are conserved in NGFG_01224 (Keiler & Sauer, 1995), and we will therefore refer to NGFG_01224 as tsp.

We generated a complete, unmarked, deletion of tsp (Δtsp) in both the WT and P_yaf^cs P_ltgX^cs backgrounds. We then complemented the deletion by restoring tsp at a distant locus under the control of an IPTG-inducible promoter. Deletion of tsp in the P_yaf^cs P_ltgX^cs background resulted in an increase in TraK levels (Figure 5A). Complementation of tsp in this strain resulted in a reproducible decrease in TraK levels, although it did not completely reduce TraK levels to those of the parental P_yaf^cs P_ltgX^cs strain (Figure 5A). TraK remained undetectable when tsp was deleted in the WT background, suggesting either that Tsp does not affect TraK levels in WT cells or that the increase in TraK is not sufficient to allow detection by the TraK antibody (Figure 5A). In contrast, TraK-FLAG3 levels did not vary when tsp was deleted, either in the WT or the P_yaf^cs P_ltgX^cs background, suggesting that the addition of a C-terminal FLAG tag to TraK stabilizes the protein and blocks degradation (Figure 5B). We did observe a ~15 kDa TraK-FLAG3 degradation product in the Δtsp strain that was not present in the tsp complemented strain (Figure 5B). We hypothesize that another periplasmic protease may process TraK-FLAG3, and this smaller fragment can serve as a substrate for Tsp.

Figure 5.

The periplasmic protease Tsp is involved in degradation of T4SS proteins. (A) Western blot using the α-TraK antibody to investigate TraK levels in MS11 (WT), MR621 (MS11 Δtsp), MR641 (MS11 Δtsp + tsp⁺), MR591 (P_yaf^cs P_ltgX^cs), MR623 (MR591 Δtsp), and MR634 (MR591 Δtsp + tsp⁺). (B) Immunoblot using the α-FLAG antibody to investigate TraK-FLAG3 levels in MR602 (traK-FLAG3 at the native locus), MR622 (MR602 Δtsp), MR642 (MR602 Δtsp + tsp⁺), MR598 (P_yaf^cs P_ltgX^cstraK-FLAG3 at the native locus), MR624 (MR598 Δtsp), and MR635 (MR598 Δtsp + tsp⁺). (C) Western blot using the α-TraN monoclonal antibody to investigate TraN levels in the same strains as in part (A) as well as in MR564 (ΔtraN) and HH578 (traN⁺). The black triangle indicates full-length TraN. Gray triangles indicate predicted degradation products. The asterisk indicates a faint band corresponding to full-length TraN in the Δtsp strain.

We attempted to test whether insertion of the FLAG3 epitope at the N-terminus of TraK (directly after the predicted Sec-dependent signal sequence; SS-FLAG3-TraK) would affect Tsp-dependent proteolysis. In Western blots of whole cell lysates, SS-FLAG3-TraK migrated with an apparent mass that was slightly larger than TraK-FLAG3 (Supplemental Figure 1). Since both proteins are predicted to be the same size following signal sequence cleavage, we hypothesize that the negatively charged N-terminal FLAG3 interfered with signal sequence insertion and/or processing, blocking export of SS-FLAG3-TraK to the periplasm and making it impossible to determine whether Tsp affects stability of this protein.

Given the multiple bands we observe in Western blots of TraN (Figure 2D), we also investigated whether TraN expression or stability was affected by Tsp. In the P_yaf^cs P_ltgX^cs strain, we detected two dominant TraN-specific bands (~50 kDa and ~30 kDa), as well as a fainter ~55 kDa band corresponding to the predicted size of full-length TraN (Figure 5C). Additional degradation products become apparent when TraN is overexpressed (Figure 5C). Deletion of tsp in the P_yaf^cs P_ltgX^cs background resulted in an increase in full-length TraN as well as the appearance of a new ~40 kDa degradation product (Figure 5C). Complementation of tsp restored the banding pattern observed in the P_yaf^cs P_ltgX^cs strain, and levels of full-length TraN decreased (Figure 5C). Interestingly, when tsp was deleted in the WT background, we consistently detected a faint band corresponding to full-length TraN (asterisk, Figure 5C), suggesting that Tsp affects TraN levels in WT cells as well as in the P_yaf^cs P_ltgX^cs background. The C-terminal amino acids of TraN are not particularly small or non-polar (VKNYYEYQ), suggesting that TraN may not be a direct target for Tsp.

To test whether TraK and TraN are direct substrates for Tsp, we asked whether the C-terminal sequences of TraK and TraN would affect degradation of alkaline phosphatase (PhoA). We added the C-terminal 8 amino acids of TraK or TraN to a truncated form of PhoA (PhoA’). Deletion and complementation of tsp had no effect on PhoA levels in the presence of the TraK or TraN tails (Supplemental Figure 2). Similar results were obtained in experiments where the TraK or TraN C-terminal sequences were added to beta-lactamase (data not shown). These results suggest that Tsp is not acting directly on TraK or TraN and suggest that another protease could be involved. HtrA (high temperature requirement A) family proteases are multi-copy suppressors of tsp in E. coli (Bass et al., 1996, Volokhina et al., 2011), so we tested whether N. gonorrhoeae DegQ affected T4SS protein levels. Deletion of degQ had no effect on TraK or TraN levels in either the WT or P_yaf^cs P_ltgX^cs backgrounds (data not shown).

Collectively, these results suggest that T4SS proteins are degraded by periplasmic proteases in the P_yaf^cs P_ltgX^cs background and that Tsp is involved, although possibly not directly (Figure 5). Since the P_yaf^cs P_ltgX^cs strain is predicted to overexpress most, but not all, of the genes required for DNA secretion (16 out of 21), we hypothesized that assembly of additional secretion apparatuses was limited in this strain and that buildup of the overexpressed proteins in the cell induced a stress response. We therefore decided to investigate expression of the remaining T4SS genes, which are organized in two operons downstream of the P_traH and P_parA promoters (Figure 1). These proteins include TraH, a predicted outer membrane protein specific to F-like T4SSs (Arutyunov et al., 2010), the inner membrane protein TraG (Kohler et al., 2013), the peptidoglycanase AtlA (Kohler et al., 2007), and two partitioning protein homologs ParA and ParB (Hamilton et al., 2005).

The parA transcript is present at high levels in WT cells

The parA and parB genes are transcribed in an operon located over 30 kb away from the rest of the T4SS genes (Figure 1) (Hamilton et al., 2005, Pachulec et al., 2014). Both parA and parB are required for DNA secretion and are thought to be involved in recruiting DNA to the secretion apparatus (Hamilton et al., 2005, Pachulec et al., 2014). We performed qRT-PCR to investigate parA transcript levels and found, in agreement with an earlier study (Pachulec et al., 2014), that parA was present at high levels in WT cells (0.51 ± 0.07 when normalized to the housekeeping gene rpoB) (Figure 2B). This is ~4-fold higher than traI levels, and ~25-fold higher than traK levels (Figure 2B). A single product was generated in the qRT-PCR reaction, and this product was not present in a GGI deletion strain, confirming that we were detecting parA contained within the GGI and not genomic parA (data not shown). Addition of a consensus sigma-70 promoter did not significantly increase transcript levels (data not shown). Given the expression levels of parA in WT cells, we hypothesized that inadequate transcription from the P_parA promoter is not a limiting factor for DNA secretion by the P_yaf^cs P_ltgX^cs strain. Furthermore, ParA or ParB would not be predicted to be a limiting factor for apparatus assembly since they are not predicted to be structural components of the secretion apparatus (Ramsey et al., 2011). We therefore investigated expression of the genes downstream of the fourth predicted promoter of type IV secretion genes: P_traH.

The traF-traH intergenic region encodes two predicted stem-loops

The predicted P_traH promoter is located in a 173 bp intergenic region between traF (the last gene in the ltgX operon) and the downstream gene traH (Figure 1). The −35 and −10 hexamers of P_traH (TTGACA and TAAAAT, respectively) are close to sigma-70 consensus motifs and are separated by 16 bp. The predicted P_traH promoter is located four base-pairs upstream of a transcription start site mapped by Remmele et al. (2014). We performed qRT-PCR for traG, the second gene in the operon downstream of P_traH, and found that traG transcript levels in WT cells were higher than those of traK and traI and similar to those of parA (0.47 ± 0.03 when normalized to rpoB) (Figure 2B). Despite the high levels of traG transcript, a previous study detected low levels of TraG in concentrated membrane fractions from WT cells (Kohler et al., 2013). This observation suggested that TraG protein levels are low despite apparently high levels of transcription from P_traH, so we investigated the intergenic region upstream of traH for potential regulatory elements.

Analysis of the sequence downstream of P_traH revealed two putative RNA stem-loops (SL) in the traH 5’ untranslated region (UTR). SL1 is formed from complementary sequences A and B, and SL2 is formed from complementary sequences C and D (Figures 6A and 6C). SL2 is predicted to occlude the ribosome binding site and TraH start codon, and we hypothesized that SL2 might inhibit translation of the downstream genes (Figure 6A). Interestingly, we also identified an alternative structure, SL3, formed from sequences B and C (Figures 6B and 6C). We predicted that SL3 would induce an open structure around the ribosome binding site and TraH start codon, resulting in increased translation of the downstream genes (Figure 6B).

Figure 6.

Identification of predicted stem-loops in the traH 5’ UTR. (A) Sequence and structure of two predicted stem-loops (SL1 and SL2) upstream of traH beginning with the transcriptional start point mapped by Remmele et al. (2014). The predicted ribosome binding site (RBS) and TraH start codon are indicated. The four stems are highlighted in red, blue, green, or purple to correspond with parts (B) and (C). The substitutions introduced to generate the A*, C*, and D* mutations are indicated. An unintended G-A mutation in the RBS of the CD* construct is shown. (B) Sequence and structure of an alternative stem-loop (SL3) upstream of traH. The predicted RBS and TraH start codon are indicated. (C) Sequence of the WT traH 5’ UTR beginning at the predicted transcriptional start point. The complementary sequences of three predicted stem-loops in this region (SL1, SL2, and SL3) are labeled A-D and are denoted by colored boxes. The m16 sequence shows the 16 mutations (lower case) that were introduced into the traH 5’ UTR on the gonococcal chromosome to generate the UTR^m16 strain. A 22-bp sequence that differs by only one base-pair compared to a sequence overlapping the ltgX start site (See Figure 2A) is underlined with an arrow. (D) Western blots for TraG (top panel) and TraK (bottom panel) in gonococcal strains MS11 n.v. (WT), MR647 (UTR^m16), MR591 (P_yaf^cs P_ltgX^cs), and MR648 (UTR^m16, P_yaf^cs P_ltgX^cs). (E) Fold change in transcript levels of traH, traG, and atlA in MR647 (UTR^m16) compared to MS11 n.v. The average ± standard deviation for at least 3 biological replicates is shown. *, significantly different from corresponding gene expression in the WT strain (P < 0.01) by Student’s two-tailed t-test.

As an initial experiment to test the significance of the stem-loops on the expression of the downstream genes, we introduced 16 mutations to the 5’ UTR of traH (UTR^m16, the specific mutations are indicated by lower case letters in Figure 6C). These mutations were predicted to disrupt formation of all three stem-loops and to not result in any additional structures (Zuker, 2003). We introduced the UTR^m16 mutations in both the WT and the P_yaf^cs P_ltgX^cs backgrounds and performed a Western blot for TraG. We were unable to detect TraG in the WT or P_yaf^cs P_ltgX^cs strains, despite the presence of the near-consensus P_traH promoter and Shine-Dalgarno sequence (GGAG) (Figure 6D). In contrast, we detected high levels of TraG in the presence of the UTR^m16 mutations, suggesting that the predicted secondary structures repress expression of the downstream genes (Figure 6D). Transcript levels of traG also increased in the UTR^m16 strain compared to WT (16 ± 0.95-fold), as did expression of the neighboring genes traH and atlA (by 16 ± 0.80-fold and 11 ± 0.67-fold, respectively), suggesting that the UTR^m16 mutations affect gene expression at both the RNA and protein level (Figure 6E).

We investigated whether increased expression of the traH-traG-atlA transcript had a global effect on the levels of other T4SS proteins, perhaps by relieving the limiting factor(s) required for apparatus assembly. We investigated TraK levels in the presence of the UTR^m16 mutations, but found no change in protein levels in either the WT or P_yaf^cs P_ltgX^cs backgrounds (Figure 6D), suggesting that while the RNA hairpins may regulate expression of the traH-traG-atlA transcript, they do not play a global role in regulating other T4SS genes.

DNA secretion in the P_yaf^cs P_ltgX^cs UTR^m16 strain is highly variable

We tested whether DNA secretion was increased in a P_yaf^cs P_ltgX^cs UTR^m16 strain (MR674) compared to WT. The amount of DNA present in MR674 culture supernatants was significantly higher than the ΔGGI strain ND500, but was highly variable and was not significantly different from the WT strain MS11 or from the P_yaf^cs P_ltgX^cs strain MR588 (Figure 3). Thus, overexpression of three transcripts in the GGI was not sufficient to increase DNA secretion.

Analysis of the stem-loops in the traH 5’ UTR

To more precisely characterize the significance and individual contributions of SL1, SL2, and SL3 to expression of the downstream genes, we made a series of traH-lacZ translational fusions (Figure 7A). Different regions of the traH 5’ UTR were fused to lacZ such that none of the traH open reading frame was present. All of the fusions contained the constitutive opaB promoter (P_opaB) instead of P_traH in order to study the regulatory effects of the putative hairpins in isolation (see Supplemental Figure 3 for details). We expressed the traH-lacZ fusions in single-copy from the iga-trpB locus on the gonococcal chromosome. The WT traH-lacZ fusion has the capacity to form all three stem-loops, but we hypothesized that it would form SL1 and SL2 (Figure 7A). A gonococcal strain containing the WT traH-lacZ fusion exhibited low levels of β-galactosidase activity (2.1 ± 0.6 β-galactosidase units, Figure 7B), consistent with the low levels of TraG we observed in the WT strain (Figure 6D). The m16 traH-lacZ fusion contains 16 mutations that differ by one base-pair from those introduced in the UTR^m16 mutant (Figure 7A, see legend for details). A gonococcal strain containing the m16 traH-lacZ fusion exhibited a >8×10⁴-fold increase in β-galactosidase activity relative to the WT construct (Figure 7B), consistent with the increased levels of TraG we observed in the UTR^m16 gonococcal strains (Figure 6D).

Figure 7.

Regulatory roles of the stem-loops in the traH 5’ UTR. (A) Schematic representation of the different regions of the traH 5’ UTR included in the eight traH-lacZ constructs. All of the traH-lacZ fusions contain the P_opaBconstitutive promoter instead of P_traH. The mutations introduced in the m16 traH-lacZ construct differ by 1 bp from those introduced in the UTR^m16 gonococcal mutant (see Figure 6C; the CTCGAG XhoI site introduced in the UTR^m16 mutant was changed to the sequence CTCGCG). The remaining constructs include various portions of the traH 5’ UTR separated from P_opaB by a randomized sequence to maintain constant spacing with the first predicted stem-loop in each construct. Coloring is consistent with that used in Figure 6. (B) β-galactosidase activity of the traH-lacZ constructs expressed in single copy from the iga-trpB site on the gonococcal chromosome in strains MR661 (WT), MR662 (m16), MR663 (CD), MR665 (C*D), MR666 (CD*), MR667 (C*D*), MR664 (BCD), KH641 (A*BCD). The RNA sequences included in each construct are indicated, as well as the predicted stem-loop structures (SL). The average ± standard deviation of at least three independent biological replicates is shown. §, significantly different from β-galactosidase activity levels in MR661 (WT, P < 0.05) by Student’s two-tailed t-test. *, significantly different from β-galactosidase activity levels in the CD construct by Student’s two-tailed t-test. N.S., not significant. (C) Levels of lacZ transcript in the same strains as in (B) normalized to MR661 (WT). §, significantly different from lacZ levels in MR661 (WT, P < 0.05) by Student’s two-tailed t-test.

The CD traH-lacZ fusion only contains RNA sequences C and D and is predicted to form only SL2 (Figure 7A). A gonococcal strain containing the CD construct exhibited low β-galactosidase activity (4.3 ± 1.2 units), indicating that SL2 alone is sufficient to repress β-galactosidase activity, although perhaps not as efficiently as the WT sequence (Figure 7B). To determine whether repression was due to the predicted stem-loop structure or was dependent on the sequence, we made a set of three complementary mutations to each side of SL2 (the C*D or CD* constructs, Figures 6A and 7A). We predicted that the presence of the C* or D* mutations in isolation would disrupt formation of SL2, while the combination of the complementary C* and D* mutations would allow the stem-loop to reform. It should be noted that sequence analysis of candidate CD* constructs revealed multiple frame-shift mutations or deletions in the 5’ region of lacZ. As a result, the CD* construct used in this study contains an additional unintended mutation (G to A) in the predicted RBS (Figure 6A). We hypothesize that the D* mutations created a stronger ribosome binding site (GGAG to GGAGG) in addition to disrupting the formation of SL2, and these combined effects resulted in β-galactosidase levels that were toxic to the cell (Figure 6A). Gonococcal strains expressing the C*D and CD* constructs exhibited a >1×10⁴-fold increase in β-galactosidase activity compared to the CD strain (Figure 7B). In contrast, a gonococcal strain expressing the C*D* construct exhibited β-galactosidase activity similar to that of the CD strain (4.1 ± 0.7 units), suggesting that the complementary C* and D* mutations allow the SL2 structure to re-form (SL2*, Figure 7B). These results support the existence of the SL2 structure and provide strong support for its role in controlling translation.

To investigate the significance of SL3, we made a traH-lacZ construct lacking RNA sequence A (construct BCD, Figure 7A). A gonococcal strain containing the BCD construct exhibited a >1×10⁴-fold increase in β-galactosidase activity compared to the CD strain (Figure 7B). These results suggest that the presence of RNA sequence B is capable of converting SL2 into a non-repressive structure, and we hypothesize that it does so by forming the alternative SL3 structure (Figure 6B). We also generated a traH-lacZ construct containing two point mutations (C19A and C24A) in sequence A (construct A*BCD, Figure 7A). The two point mutations would be predicted to disrupt base pairing between sequence A and B, allowing formation of SL3. A gonococcal strain containing the A*BCD construct exhibited a >8×10⁴-fold increase in β-galactosidase activity compared to the WT traH-lacZ strain (Figure 7B), suggesting that disruption of SL1 favors formation of SL3. The signal that mediates the switch between the repressive SL1-SL2 structure and the permissive SL3 structure in vivo is currently unknown.

Since introduction of the m16 mutations at the native locus caused an increase in traG transcript levels, we used qRT-PCR to investigate whether lacZ transcript levels varied among gonococcal strains expressing the various traH-lacZ fusions. In a gonococcal strain containing the WT traH-lacZ fusion, lacZ transcript levels were high (>5-fold higher than traG transcript levels in the WT strain), confirming that the P_opaB promoter is stronger than the native P_traH promoter. In contrast to the increase in traG transcript that we observed in the UTR^m16 strain, lacZ transcript levels did not vary by more than 3-fold in gonococcal strains containing any of the different traH-lacZ fusions, and differences in lacZ transcript levels were not correlated with differences in β-galactosidase activity (Figure 7C). These results suggest that the stem-loops primarily affect translation and not transcription, likely because SL2 occludes the ribosome binding site and TraH translation start site (Figure 6A).

These results suggest a model for the function of the RNA switch in controlling translation of TraH, TraG, and AtlA (Figure 8D), described in more detail below. In this model, the more energetically-favored “OFF” structure comprised of SL1 and SL2 forms and represses expression of the T4SS genes. However, under certain conditions, a factor is produced that sequesters sequence A, leading to formation of the “ON” SL3 structure and production of the T4SS proteins.

Figure 8.

Model for regulatory mechanisms controlling T4SS gene expression. (A) WT gonococci secrete DNA into the extracellular environment. The four transcripts containing T4SS genes are expressed at very different levels (numbers indicate approximate relative abundance of each transcript compared to transcript levels of the ltgX operon). (B) In the P_yaf^cs P_ltgX^cs mutant, mutations in the P_yaf and P_ltgX promoters result in increased transcription and translation of the downstream genes. The P_yaf^cs P_ltgX^cs strain secretes similar levels of DNA compared to WT, but overexpression of T4SS proteins triggers a stress response and leads to degradation by periplasmic proteases. (C) The P_yaf^cs P_ltgX^cs UTR^m16 mutant contains 16 mutations in the 5’ UTR of traH in addition to the mutations in the P_yaf and P_ltgX promoters, resulting in increased transcription and translation of 19 of the 21 genes required for DNA secretion. The P_yaf^cs P_ltgX^cs UTR^m16 strain secretes similar levels of DNA compared to WT. In both (B) and (C), the numbers indicate transcript abundance relative to that of the ltgX operon in WT cells. (D) Model of the regulatory roles of the stem-loops (SL) in the 5’ UTR of the traH transcript. Under our current in vitro growth conditions, we predict that SL1 and SL2 form and that SL2 represses translation of the downstream genes by sequestering the ribosome binding site. The binding of an unknown factor, which could be protein or RNA or both, to sequence A allows formation of an alternative SL3, which creates a more open structure around the ribosome binding site and allows translation. A sequence with only 1 base-pair mismatch compared to sequence A is also found on the antisense strand overlapping the ltgX start codon, suggesting that the unknown factor may also bind to or affect expression levels of other T4SS transcripts.

Discussion

These studies enhance our understanding of the expression and regulation of the GGI-encoded T4SS genes and suggest that gene expression is integrated into a complex regulatory network that includes transcriptional, translational, and post-translational mechanisms of control, summarized in Figure 8. We showed that the four T4SS operons are expressed at very different levels, with the operon encoding most of the structural apparatus proteins expressed at much lower levels compared to the other transcripts (Figures 2B and 8A). We performed targeted mutagenesis of the putative promoters P_yaf and P_ltgX with the intention of generating a strain that would constitutively overexpress the T4SS apparatus. The P_yaf^cs P_ltgX^cs strain is predicted to overexpress 16 of 21 required genes, including most of the structural components of the secretion apparatus, but this strain does not secrete significantly higher levels of DNA compared to the WT strain (Figures 3 and 8B). A strain combining the P_yaf^cs and P_ltgX^cs mutations with the UTR^m16 mutations (and thus overexpressing 19 of 21 required genes) also exhibited variable, but not significantly increased, levels of DNA secretion (Figures 3 and 8C).

One of the initial goals of the current study was to generate a strain that overexpressed the T4SS and secreted consistently high levels of DNA. Such a strain would be useful for future studies of this unusual DNA secretion system. Intriguingly, our results indicate that increasing DNA secretion in gonococci is not as simple as merely overexpressing the genes in the GGI and suggest that the requirements for DNA secretion are more complicated than previously appreciated. It is possible that there are additional unidentified proteins required for DNA secretion in gonococci beyond those that have been identified in the GGI. It also may be that the DNA secretion substrate is a limiting factor for type IV secretion (Lang et al., 2014). A single oriT sequence is located on the chromosome in the yaf-ltgX intergenic region (Salgado-Pabón et al., 2007), and since gonococci are naturally diploid (Tobiason & Seifert, 2006), exactly two DNA substrates would be expected to be produced per cell. It is also possible that the relative stoichiometry of T4SS proteins, which is perturbed in our promoter mutants, is important for T4SS apparatus assembly. Consistent with this hypothesis, we found that a homolog of E. coli Tail-specific protease (Tsp) is involved degradation of the T4SS proteins TraK and TraN in the P_yaf^cs P_ltgX^cs strain, although likely not directly (Figure 8B, Supplemental Figure 2). These results suggest that a buildup of T4SS proteins in the periplasm of the P_yaf^cs P_ltgX^cs strain triggered a stress response. Little is known regarding the mechanistic details of periplasmic quality control in Neisseria, but in E. coli, Tsp acts in concert with several other proteases and chaperones to degrade or refold damaged, aggregated, or misfolded proteins (Merdanovic et al., 2011). The ability to detect TraN in the tsp deletion strain suggests that Tsp affects T4SS proteins not just in our overexpression strains but also in strains with WT levels of the T4SS proteins (Figure 5C). Our results suggest that levels of T4SS proteins are closely monitored in the cell and regulated at a post-translational level.

We also identified an RNA secondary structure consisting of two stem-loops (SL1 and SL2) in the traH 5’ UTR that regulates a >10,000-fold change in downstream gene expression (Figure 6). Using traH-lacZ fusions, we showed that SL2 alone is capable of repressing lacZ expression. Introduction of complementary mutations on either side of the stem-loop supports the hypothesis that this sequence does indeed form a secondary structure and has an important role in controlling translation of the downstream genes (Figure 7B). We also identified an alternative stem-loop (SL3) that we predicted would generate an open structure around the ribosome binding site (Figure 6B). Consistent with this hypothesis, traH-lacZ constructs predicted to form SL3 (the BCD and A*BCD constructs, Figure 7B) exhibited high levels of β-galactosidase activity.

Transcript levels of lacZ did not vary more than 3-fold amongst any of the traH-lacZ constructs despite several log-fold differences in β-galactosidase units, suggesting that the stem-loops primarily affect translation (Figure 7C). In contrast, we did observe an increase in traG transcript levels when the m16 mutations were introduced at the native locus (UTR^m16) (Figure 6E). It is unclear whether the increase in traG transcript represents an increase in transcription or an increase in RNA stability. Given that transcription and translation are coupled in bacteria, increased numbers of ribosomes on the traH-traG-atlA transcript could increase its stability. The most notable difference between the UTR^m16 strain and the m16 traH-lacZ fusion is the promoter: the traH-lacZ fusions all contain the strong constitutive P_opaB promoter instead of the native P_traH promoter. The fact that the m16 mutations affect transcript levels only at the native locus could suggest that an additional level of regulation affects transcription from P_traH. Alternatively, the fact that P_opaB is a stronger promoter than P_traH could mask potential differences in RNA stability between the traH-lacZ fusions.

Based on these results, we propose that an RNA switch mechanism regulates expression of the traH-traG-atlA transcript (Figure 8D). We predict that the SL1-SL2 structure represses translation of the downstream genes by sequestering the ribosome binding site. Since SL1 is more stable than SL3 (ΔG = −18.8 kcal/mol vs. −14.4 kcal/mol, calculated using the Mfold server) (Zuker, 2003), and since transcription of the sequences contributing to SL1 will occur first, we hypothesize that SL3 is unlikely to form unless sequence A is occluded by a regulatory factor (Figure 8D). Alternatively, it is possible that a regulatory factor stabilizes the SL1-SL2 structure by binding to SL1, preventing SL3 formation. If sequence A acts as a binding site for a regulatory factor, we wondered whether this sequence was located anywhere else in the genome. We performed a search for sequence A in the available gonococcal genome sequences and found one additional match in every genome that contained the GGI. This 22-bp sequence overlaps the ltgX start codon and Shine-Dalgarno sequence and is on the anti-sense strand. It contains only one mismatch compared to nucleotides 4–24 of the traH 5’ UTR (CCCTTAATCTCAtGTCTTTACT, underlined in Figures 2A and 6C). How the 22-bp sequence could affect expression of ltgX and the downstream T4SS genes is unclear, but we predict that the unknown regulatory factor could bind to both the ltgX and traH transcripts, thereby coordinating regulation and expression of T4SS genes in response to environmental stimuli.

The identities of any regulatory factors and the nature of the signal that mediates the switch between the repressive SL1-SL2 structure and the permissive SL3 structure are currently unknown. The regulatory factor could be a protein similar to the trp RNA binding attenuation protein (TRAP) in Bacillus subtilis, which regulates expression of the trpE gene in tryptophan biogenesis through a translational repression mechanism (Babitzke et al., 2009). In the presence of tryptophan, TRAP binds to the trpE 5’ UTR, which favors the formation of a repressive stem-loop, occluding the ribosome binding site and blocking trpE translation (Babitzke et al., 2009).

Alternatively, the regulatory factor could be a small RNA (sRNA) or sRNA-protein complex. N. gonorrhoeae encodes an Hfq homolog, and although a recent study did not detect any change in transcript levels of traH, traG, or atlA in an hfq mutant, this study does not preclude a role for Hfq and/or sRNAs in post-transcriptional regulation of the traH operon (Dietrich et al., 2009). Interestingly, an sRNA-protein complex (FinP/FinO) is involved in regulation of F-plasmid conjugation via repression of the primary activator TraJ. The FinP antisense RNA is transcribed from a promoter in the 5’ region of traJ and is complementary to the traJ 5’ UTR (Frost et al., 1994). The RNA chaperone FinO then facilitates the interaction between FinP and the 5’ UTR of traJ, thereby repressing TraJ translation (Wong et al., 2012, Arthur et al., 2003). It is unlikely that a similar anti-sense RNA is responsible for repressing traH in gonococci, however, since none of the traH open reading frame was present in any of the traH-lacZ constructs, and the WT traH-lacZ construct exhibited low levels of β-galactosidase activity.

The RNA switch regulated translation of our traH-lacZ fusion over 10,000-fold. The purpose of such strong regulation of the traH-traG-atlA transcript is currently unknown. TraG is an inner membrane protein (Kohler et al., 2013), while TraH is a predicted outer membrane protein that is a central component in an interaction group thought to be involved in F-pilus assembly in E. coli (Arutyunov et al., 2010, Harris & Silverman, 2004). AtlA is a peptidoglycanase that is required for gonococcal type IV secretion and is thought to make localized breaks in the peptidoglycan layer to allow apparatus assembly (Kohler et al., 2007). It is possible that these three proteins form a nucleation complex that initiates complex assembly and that independent regulation of this transcript could be important for regulating complex assembly as a whole.

Experimental Procedures

Bacteria and growth conditions

The bacterial strains used in this study are described in Table 1. Escherichia coli strains were grown on Luria-Bertani (LB) agar plates or in LB broth at 37°C (Sambrook et al., 1989). Gonococcal strains were grown on gonococcal base (GCB) agar plates (Difco) containing Kellogg’s supplements (Kellogg et al., 1963) or in GCBL liquid medium containing Kellogg’s supplements and 0.042% NaHCO₃ (cGCBL) (Morse & Bartenstein, 1974). Erythromycin was used at 500 µg ml⁻¹ for E. coli and 10 µg ml⁻¹ for gonococci. Chloramphenicol was used at 25 µg ml⁻¹ for E. coli and 10 µg ml⁻¹ for gonococci. Ampicillin was used at 100 µg ml⁻¹ for E. coli. Gene expression in gonococci was induced with 1 mM isopropyl β-D-thiogalactopyranoside (IPTG).

Table 1.

Bacterial strains and plasmids

Plasmid or strain	Properties	Reference or source
Plasmid
pAKK64	PCR of pMR47 with phoA_SacI-F and phoAtr_SpeI-R into pMR68 SacI- SpeI (truncated phoA gene lacking the last 12 codons)	This work
pAKK65	PCR of pMR47 with phoA_SacI-F and phoAtr-Ktail-R into pMR68 SacI- SpeI (truncated phoA gene fused to the last eight codons of TraK)	This work
pAKK66	PCR of pMR47 with phoA_SacI-F and phoAtr-Ntail-R into pMR68 SacI- SpeI (truncated phoA gene fused to the last eight codons of TraN)	This work
pAKK67	PCR of pUC19 with bla_SacI-F and bla_SpeI-R into pMR68 SacI-SpeI	This work
pAKK68	PCR of pUC19 with bla_SacI-F and bla_Ktail-R into pMR68 SacI-SpeI	This work
pAKK69	PCR of pUC19 with bla_SacI-F and bla_Ntail-R into pMR68 SacI-SpeI	This work
pAY3	PCR of pMR97 with yaf_MscI-F and yaf-profix-R, digestion with MscI, and self-ligation (for generating 17 bp spacing between the predicted −10 and −35 elements farther upstream from yaf)	This work
pIDN3	Cloning vector	(Hamilton et al., 2001)
pKH37	lctP-aspC complementation construct	(Kohler et al., 2007)
pKH176	SOE product pMR115+1 plasmid DNA (traH UTR SOE1 with traH UTR SOE2 and traH UTR SOE3 with traH UTR SOE4) into pMR115+1 AflII-StuI	This work
pKLD116	Cloning vector for constructing maltose-binding protein fusions	(Rocco et al., 2008)
pMR24	iga-trpB complementation construct, CmR	(Ramsey et al., 2012)
pMR33	iga-trpB complementation construct, EmR	(Ramsey et al., 2012)
pMR45	PCR of MS11 chromosomal DNA with TraNF-W128-StuI and TraNR-R243- NcoI into pKLD116 StuI and NcoI (encoding MBP-TraNW128-R243)	This work
pMR47	traWSS::’phoA (alkaline phosphatase signal sequence substituted with signal sequence from TraW)	This work
pMR68	iga-trpB complementation construct, contains the tetracyline-inducible promoter	(Ramsey et al, 2012)
pMR75	PCR of MS11 chromosomal DNA with delTraNF-SpeI and TraNstartR-XmaI into pIDN3 SpeI-XmaI (873 bp 5’ traN flanking DNA, includes 92 bp of traN)	This work
pMR76	PCR of MS11 chromosomal DNA with TraNendF-XmaI and delTraNR-ClaI into pIDN3 XmaI-ClaI (517 bp 3’ traN flanking DNA, includes 60 bp of traN)	This work
pMR77	pMR76 XmaI-ClaI into pMR75 XmaI-ClaI sites (traN deletion construct)	This work
pMR95	PCR of MS11 chromosome with 111F and 140R, digested with EcoRV and, PvuI blunted, into pIDN3 EcoRV site (yaf-ltgX IGR)	This work
pMR96	PCR of pMR95 with ProfixF and ProfixR, digestion with AfeI, and self- ligation (contains PltgXcs mutations)	This work
pMR97	PCR of MS11 chromosome with 111F and pNT01-Screen-1, digested with NaeI and DraIII, blunted, into pIDN3 EcoRV site (partial yaf-ltgX IGR)	This work
pMR98	PCR of pMR97 with yafprofix-F and yafprofix-R, digestion with MscI, and self-ligation (contains Pyafcs mutations)	This work
pMR104	Plasmid containing traK-FLAG3 with downstream flanking DNA	(Ramsey et al., 2014)
pMR108	PCR of MS11 chromosomal DNA with delTspF-SpeI & delTspstartR-XmaI into pIDN3 SpeI-XmaI sites (564 bp 5’ tsp flanking DNA)	This work
pMR109	PCR of MS11 chromosomal DNA with delTspendF-XmaI & delTspR-ClaI into pIDN3 XmaI-ClaI sites (677 bp 3’ tsp flanking DNA)	This work
pMR110	pMR109 with XmaI-ClaI into pMR108 XmaI-ClaI sites (tsp deletion construct)	This work
pMR111	PCR of MS11 chromosomal DNA with TspF-SpeI & TspR-XhoI into pKH37 SpeI-XhoI sites (tsp complementation construct)	This work
pMR112	Ligation of two PCRs from MS11 chromosomal DNA (using deldegQF-SpeI with degQstartR-XmaI and degQendF-XmaI with deldegQR-ClaI) into pIDN3 SpeI-ClaI sites (degQ deletion construct)	This work
pMR113	Stitching by overlap extension PCR using two PCRs from MS11 chromosomal DNA (traF-F-XmaI with breakhairpinsR and breakhairpinsF with traH-R-ClaI) into pIDN3 XmaI-ClaI sites (contains traH UTRm16 mutations)	This work
pMR114	PCR of traK to add FLAG3 after signal sequence in pIDN3 (SS-FLAG3-TraK)	This work
pMR115	‘lacZ from pKH39 (blunted MfeI and ClaI) into pMR24 blunted BseRI and ClaI (iga-trpB contstruct containing 3’ portion of lacZ with no promoter)	This work
pMR115+1	traH UTR sequence #1-WT into pMR115+7 XhoI and StuI (WT traH-lacZ)	This work
pMR115+2	traH UTR sequence #2-NoStructure into pMR115 XhoI and ClaI (mut16 traH-lacZ)	This work
pMR115+3	traH UTR sequence #3-DelStem1 into pMR115+7 XhoI and StuI (CD traH-lacZ)	This work
pMR115+4	traH UTR sequence #4-DelStem1A into pMR115+7 XhoI and StuI (BCD traH-lacZ)	This work
pMR115+5	traH UTR sequence #5-DelStem1(2A) into pMR115 XhoI and ClaI (C*D traH-lacZ)	This work
pMR115+6	traH UTR sequence #6-DelStem1A(2B) into pMR115 XhoI and ClaI (CD* traH-lacZ)	This work
pMR115+7	traH UTR sequence #7-DelStem1A(2A2B) into pMR115 XhoI and ClaI (C*D* traH-lacZ)	This work
N. gonorrhoeae strain
AKK640	MS11 transformed with pAKK64 (phoA’)	This work
AKK641	MR621 transformed with pAKK64 (phoA’, Δtsp)	This work
AKK642	MR641 transformed with pAKK64 (phoA’, Δtsp, IPTG-inducible tsp at lctP-aspC)	This work
AKK643	MS11 transformed with pAKK65 (phoA’-Ktail)	This work
AKK644	MR621transformed with pAKK65 (phoA’-Ktail, Δtsp)	This work
AKK645	MR641 transformed with pAKK65 (phoA’-Ktail, Δtsp, IPTG-inducible tsp at lctP-aspC)	This work
AKK646	MS11 transformed with pAKK66 (phoA’-Ntail)	This work
AKK647	MR621 transformed with pAKK66 (phoA’-Ntail, Δtsp)	This work
AKK648	MR641 transformed with pAKK66 (phoA’-Ntail, Δtsp, IPTG-inducible tsp at lctP-aspC)	This work
AKK651	MS11 transformed with pAKK67 (bla+)	This work
AKK652	MR621 transformed with pAKK67 (bla+, Δtsp)	This work
AKK653	MR641 transformed with pAKK67 (bla+, Δtsp, Δtsp, IPTG-inducible tsp at	This work
	lctP-aspC)
AKK654	MS11 transformed with pAKK68 (bla-Ktail)	This work
AKK655	MR621 transformed with pAKK68 (bla-Ktail, Δtsp)	This work
AKK656	MR641 transformed with pAKK68 (bla-Ktail, Δtsp, IPTG-inducible tsp at	This work
	lctP-aspC)
AKK657	MS11 transformed with pAKK69 (bla-Ntail)	This work
AKK658	MR621 transformed with pAKK69 (bla-Ntail, Δtsp)	This work
AKK659	MR641 transformed with pAKK69 (bla-Ntail, Δtsp, IPTG-inducible tsp at lctP-aspC)	This work
AY503	MS11 transformed with pAY3	This work
HH578	ΔtraN, traN+complement	(Hamilton et al., 2005)
KH641	MR658 transformed with pKH176 (locked pilE, A*BCD traH-lacZ at iga-trpB)	This work
MR564	MS11 transformed with pMR77 (ΔtraN)	This work
MR582	MS11 transformed with pMR96 (PltgXcs)	This work
MR588	MR582 transformed with pMR98 (Pyafcs, PltgXcs)	This work
MR589	MS11 n.v. transformed with pMR96 (locked pilE, PltgXcs)	This work
MR590	MS11 n.v. transformed with pMR98 (locked pilE, Pyafcs)	This work
MR591	MR589 transformed with pMR98 (locked pilE, Pyafcs, PltgXcs)	This work
MR598.1	MR591 transformed with pMR104 (locked pilE, Pyafcs, PltgXcstraK-FLAG3) (referred to as MR598)	This work
MR602.1	MS11 transformed with pMR104 (traK-FLAG3) (referred to as MR602)	(Ramsey et al., 2014)
MR621	MS11 transformed with pMR110 (Δtsp)	This work
MR622	MR602 transformed with pMR110 (traK-FLAG3, Δtsp)	This work
MR623	MR591 transformed with pMR110 (locked pilE, Pyafcs,PltgXcs, Δtsp)	This work
MR624	MR598 transformed with pMR110 (locked pilE, Pyafcs, PltgXcs, traK-FLAG3, Δtsp)	This work
MR627	MS11 transformed with pMR112 (ΔdegQ)	This work
MR628	MR602 transformed with pMR112 (traK-FLAG3, ΔdegQ)	This work
MR629	MR591 transformed with pMR112 (locked pilE, Pyafcs, PltgXcs, ΔdegQ)	This work
MR630	MR598 transformed with pMR112 (locked pilE, Pyafcs, PltgXcs,traK-FLAG3, ΔdegQ)	This work
MR634	MR623 transformed with pMR111 (locked pilE, Pyafcs, PltgXcs, Δtsp, IPTG-inducible tsp at lctP-aspC)	This work
MR635	MR624 transformed with pMR111 (locked pilE, Pyafcs, PltgXcs,traK-FLAG3, Δtsp, IPTG-inducible tsp at lctP-aspC)	This work
MR641	MR621 transformed with pMR111 (Δtsp, IPTG-inducible tsp at lctP-aspC)	This work
MR642	MR622 transformed with pMR111 (traK-FLAG3, Δtsp, IPTG-inducible tsp) at lctP-aspC	This work
MR643	MS11 n.v. transformed with pMR114 (locked pilE, SS-FLAG3-traK)	This work
MR644	MR591 transformed with pMR114 (locked pilE, Pyafcs, PltgXcs,SS-FLAG3-traK)	This work
MR647	MS11 n.v. transformed with pMR113 (locked pilE, UTRm16)	This work
MR648	MR591 transformed with pMR113 (locked pilE, Pyafcs, PltgXcs, UTRm16)	This work
MR658	MS11 n.v. transformed with pMR115 (locked pilE, ‘lacZ at iga-trpB)	This work
MR661	MR665 transformed with pMR115+1 (locked pilE, WT traH-lacZ at iga-trpB)	This work
MR662	MR658 transformed with pMR115+2 (locked pilE, m16 traH-lacZ at iga-trpB)	This work
MR663	MR665 transformed with pMR115+3 (locked pilE, CD traH-lacZ at iga-trpB)	This work
MR664	MR658 transformed with pMR115+4 (locked pilE, BCD traH-lacZ at iga-trpB)	This work
MR665	MR658 transformed with pMR115+5 (locked pilE, C*D traH-lacZ at iga-trpB)	This work
MR666	MR658 transformed with pMR115+6 (locked pilE, CD* traH-lacZ at iga-trpB)	This work
MR667	MR658 transformed with pMR115+7 (locked pilE, C*D* traH-lacZ at iga-trpB)	This work
MR674	MR588 transformed with pMR113 (Pyafcs, PltgXcs, UTRm16)	This work
MS11	WT N. gonorrhoeae	(Swanson et al., 1971)
MS11 n.v.	WT N. gonorrhoeae with non-variable pilin expression (locked pilE due to a point mutation in the G-quartet sequence upstream of pilE)	A. Criss

Given the effect of piliation on expression of the T4SS (Salgado-Pabón et al., 2010), we sought to prevent pilin variation and introduced all of the promoter mutations into the MS11 n.v. strain, a variant of the WT MS11 strain in which the pilE gene is expressed and non-variable due to the avd-1 point mutation in the G-quartet that controls pilE recombination (Cahoon & Seifert, 2009). Gonococci that express pili form clumps in liquid culture, and we observed a particularly strong clumping phenotype in the MS11 n.v. strain, perhaps due to the particular pilE sequence being expressed. These clumps interfered with optical density measurements necessary for the DNA secretion assay. Therefore, when testing the effects of the promoter mutations on DNA secretion, we introduced the mutations into the WT MS11 background instead of the MS11 n.v. background. Piliated colonies of MR588, which formed clumps that could be easily disrupted by gently vortexing, were used for analysis in the DNA secretion assay.

DNA techniques

The plasmids used in this study are described in Table 1, and primer sequences are provided in Table 2. Plasmid DNA was isolated using the QiaPrep miniprep kit (Qiagen), and digested DNA was gel-purified using the QiaQuick gel extraction kit (Qiagen). T4 DNA polymerase (New England Biolabs) was used to generate blunt DNA ends, and ligation was performed using T4 DNA ligase (New England Biolabs). DNA was transformed into chemically competent RapidTrans TAM1 E. coli cells (Active Motif). Potential transformants were screened by analysis of whole-cell lysates using the lysis solution of Kado and Liu (Kado & Liu, 1981), and additional screening was performed by restriction enzyme digest and PCR.

Table 2.

Primer Sequences

Primer Name	Sequence (5'→3') with restriction sites underlined
111F	CCGCTGACAAGCATTAGTGA
140R	GCATAGGGAGCCATTTCTTC
atlA-RT-F	CAAGTGGAAGACATGTGAGAGG
atlA-RT-R	GTTTTACACCTTCTGATGCCG
breakhairpinsR	GCTCCATGAAGACTAGTTAGCAGGGATTGGCTTCTCGAGACCAAG ATAAGTTATAGTAAAGACTTG
breakhairpinsF	TTGGTCTCGAGAAGCCAATCCCTGCTAACTAGTCTTCATGGAGCG GTTATGAAATTTCACATCTTTC
degPendF-XmaI	GGTGGTCCCGGGTAACGCACATAATTTAACC
degPstartR-XmaI	GGTGGTCCCGGGCACGTTTTGTCCTTTGTCGGATG
deldegPF-SpeI	GGTGGTACTAGTTCCGCCAAGAAATCTTCGAG
deldegPR-ClaI	GGTGGTATCGATCTTGAAACGGGCATCAGCAC
delTraNF-SpeI	GGTGGTACTAGTCGGCATGTCCTACGGAAACC
delTraNR-ClaI	GGTGGTATCGATAGGCGTGCAGGCTAAATATC
delTspendF-XmaI	GGTGGTCCCGGGTAGGTTTTGACGATGCCGTCT
delTspF-SpeI	GGTGGTACTAGTGCCGAAGACAGGAACATCCAAG
delTspR-ClaI	GGTGGTATCGATGATGAACAAATCGCGGCTGGG
delTspstartR-XmaI	GGTGGTCCCGGGCATTCTTCTTTAACTTTCTCT
lacZ-RT-F	GCCGAAATCCCGAATCTCTATC
lacZ-RT-R	AGCAGCAGCAGACCATTT
pNT01-Screen-1	AAGGGTAAAGGCTCTTAT
phoA_SacI-F	CTGAGAGCTCATCCATTCAGGGCAAATTATTGGTAAGGC
phoAtr SpeI-R	GACTACTAGTTAATCGGTCTGGTCGGTCAGTC
phoAtr-Ktail-R	GACTACTAGTCAACCTCCCGGACGGCTGATAATAAAATCGGTCTGGTC GGTCAGTC
phoAtr-Ntail-R	GTCTACTAGTTATTGATATTCATAATAGTTCTTCACATCGGTCTGGTC GGTCAGTC
bla_SacI-F	CTGAGAGCTCAGGAAGAGTATGAGTATTCAACATTTCC
bla_SpeI-R	GACTACTAGTCAGTTACCAATGCTTAATCAGTGAGG
bla_Ktail-R	GACTACTAGTCAACCTCCCGGACGGCTGATAATAAACCAATGCTTAAT CAGTGAGG
bla_Ntail-R	GTCTACTAGTTATTGATATTCATAATAGTTCTTCACCCAATGCTTAAT CAGTGAGG
ProfixF	GGTGGTAGCGCTTATATTTTGTCAAATTACCG
ProfixR	GGTGGTAGCGCTAACTTATAATTATACCATAATC
traF-F-XmaI	GGTGGTCCCGGGATCGGATTGCCCTTACTG
traH-R-ClaI	GGTGGTATCGATATCAAGGGCCTTAAGTGC
traH UTR SOE 1	CGCGTTACTGTACCGTAATG
traH UTR SOE 2	GATAAGTTATATTAAATACTTGAGATTAAGGG
traH UTR SOE 3	CCCTTAATCTCAAGTATTTAATATAACTTATC
traH UTR SOE 4	TGGGAAGGCCTATCGGTGCG
traG-RT-F	CTTACCCTGGAACTCTTTCGG
traG-RT-R	TGGTAGGTCAAATTCGGATGC
traH-RT-F	GCAATGGGAAAACTGGGTTC
traH-RT-R	TTATCGGCTTCATGGACAAGG
TraNendF-XmaI	GGTGGTCCCGGGAGGGAGGCATCCGTTAATGG
TraNF-W128-StuI	GGTGGTAGGCCTTTGGGATGAGTCGTCATGTCAG
TraNR-R243-NcoI	GGTGGTCCATGGTCACCGATCCTGACAACCGGATACC
TraNstartR-XmaI	GGTGGTCCCGGGCACGCTGCACTTTCCCTGAG
TspF-SpeI	GGTGGTACTAGTTTTGCCGTTTGCCGTACCTG
TspR-XhoI	GGTGGTCTCGAGCGCTCATAACCTGTCCTTTC
yaf_AgeI-F	CTGTACCGGTTGTCAAACCATTC
yaf_AgeI-R	GTCTACCGGTGTCTTTTTAATCAAAATG
yaf_MscI-F	GCACTGGCCAACTTTTTTTTATTATATGAATGG
yafprofix-F	GGTGGTTGGCCAACTTTTTTTTATATAATGAATGG
yafprofix-R	GGTGGTTGGCCATGTCAAAAATCAAAATAAAAAATG

Transformation of gonococci

Spot transformations were used to transform gonococci with plasmid DNA (Dillard, 2011). Plasmids smaller than 8000 bp were linearized before transformation. Between 1 – 10 µg plasmid DNA was diluted in water, spotted onto a GCB plate, and allowed to dry. Several piliated colonies of the appropriate gonococcal strain were streaked over the spot, and the plate was incubated overnight at 37°C with 5% CO₂. In cases where there was a selectable marker, a Dacron swab was used to transfer colonies from the spots onto plates containing the appropriate antibiotic to select for transformants. Alternatively, for unmarked mutations or deletions, colonies were swabbed from the spot into GCBL, serially diluted, and spread on GCB agar plates. Single colonies were restreaked and screened by PCR and restriction digest.

Protein purification

A portion of TraN (W128-R243) was purified as a C-terminal fusion to maltose binding protein (MBP) using pKLD116 (Rocco et al., 2008). Plasmid pMR45 (encoding the MBP-TraN_W128-R243 fusion) was transformed into BL21 E. coli for expression and purification. The following protocol was repeated twice to purify enough MBP-TraN_W128-R243 for monoclonal antibody production. Overnight cultures of BL21 E. coli containing pMR45 were used to inoculate 4 L of LB broth containing 100 µg ml⁻¹ ampicillin. Cultures were grown at 25°C with shaking for ~8 hours, protein production was induced with 0.3 mM IPTG, the temperature was dropped to 15°C, and the cultures were grown overnight with shaking. The cells were harvested by centrifugation at 3000 × g for 10 minutes and resuspended in 20 ml amylose column buffer (20 mM Tris-HCl pH 7.4, 200 mM NaCl, 10% glycerol) containing protease inhibitors (Roche complete protease cocktail). The cells were lysed by three passages through a French press cell at 1200 psi, and the lysate was centrifuged at 20,200 × g for 30 minutes to remove unlysed cells. The supernatant was incubated with 2 ml of amylose resin that had been washed in amylose column buffer (New England Biolabs) at 4°C with gentle shaking for 2 hours and was then loaded onto a gravity column. After washing the column with amylose column buffer containing protease inhibitors, MBP-TraN_W128-R243 was eluted in 3-ml fractions of amylose column buffer containing 0.001, 0.01, 0.1, 1, and 10 mM maltose. The maltose fractions were pooled, dialyzed overnight into nickel column buffer (20 mM Tris-HCl pH 8.0, 300 mM NaCl, 10% glycerol), and incubated for 2.5 hours with gentle mixing at 4°C with a 2.5-ml volume of nickel resin pre- washed in nickel column buffer (Sigma). The mixture was loaded on a gravity column, washed with nickel column buffer containing 10 mM imidazole, and eluted in 2-ml fractions of nickel column buffer containing 30, 60, 100, and 250 mM imidazole. Purified MBP-TraN_W128-R243 was dialyzed twice into storage buffer (50 mM Tris-HCl pH 8.0, 100 mM NaCl, 10% glycerol) and used for monoclonal antibody production. MBP was purified from the empty pKLD116 vector following a similar protocol.

TraN monoclonal antibody production

Purified MBP-TraN_W128-R243 and MBP were sent to Genscript for monoclonal antibody production. The resulting hybridoma lines were screened by ELISA to identify antibodies that reacted with MBP-TraN but not with the MBP tag. Two hybridoma lines were identified, and the 7E12 line was selected for Protein A/G affinity purification.

Western blots

Western blots were performed essentially as described before (Ramsey et al., 2012, Ramsey et al., 2014). The rabbit polyclonal TraI antibody was used at a 1:30 dilution (Salgado-Pabón et al., 2007), the mouse monoclonal TraK antibody was used at a 1:800 dilution (Ramsey et al., 2014), the mouse monoclonal TraN antibody was used at a 1:1300 dilution, the mouse monoclonal FLAG antibody was used at a 1:10,000 dilution, and the rat polyclonal TraG antibody was used at a 1:650 dilution (Kohler et al., 2013).

5’ RACE

RNA isolated from 3-hour cultures of N. gonorrhoeae was subjected to 5’ RACE according to the manufacturer’s instructions (Life Technologies). Prior to some reactions RNA was treated with Terminator 5’ phosphate-dependent exonuclease (TEX, Epicentre) to enrich for primary transcripts.

DNA secretion assays

DNA secretion assays were performed essentially as described previously (Jain et al., 2012, Ramsey et al., 2014).

RNA isolation

To isolate RNA from gonococci, cultures were inoculated in cGCBL at an OD₅₄₀ = 0.25 and grown for 3 hours into log phase. The cultures were vortexed, and a 2-ml volume of the culture was mixed with a 2-ml volume of −20°C methanol. The cells were centrifuged at 17,200 × g for 7 minutes at 4°C. The supernatant was removed, and the pellet was frozen at −80°C for approximately 20 minutes. The pellet was resuspended in a 1-ml volume of Trizol followed by the addition of 200 µl chloroform. The tubes were shaken vigorously for 15 seconds to mix, allowed to incubate for 3 minutes at room temperature, and then centrifuged for 15 minutes at 12,000 × g at 4°C. The top aqueous phase was removed to a new tube, 500 µl of isopropanol was added, and the tube was inverted to mix. RNA was precipitated at room temperature for 10 minutes followed by centrifugation for 15 minutes at 12,000 × g at 4°C to pellet the RNA. The pellet was washed once in 75% ethanol and allowed to dry before it was resuspended in DEPC-treated water. The Turbo DNA-free kit was used to treat 4 µg of RNA (Applied Biosystems). The iScript cDNA Synthesis Kit (Bio-Rad) was used to reverse transcribe the RNA, and this cDNA was used for quantitative real-time PCR.

qRT-PCR

Oligonucleotide primers for real-time PCR were designed using the SciTools feature of the IDT website. Sequences of the primers used for analysis of lacZ, traH, traG, and atlA are provided in Table 2. Analysis of traI, traD, traK, and rpoB was performed using primers as previously described (Salgado-Pabón et al., 2010, Ramsey et al., 2014). RNA was isolated from log-phase gonococci and reverse-transcribed into cDNA as described above. Control reactions were carried out in the absence of reverse transcriptase. Reaction mixtures for quantitative real-time PCR (qRT-PCR) were prepared using the iQ SYBR green supermix (BioRad) in a 25 µl volume and run in triplicate. Negative controls using the “no reverse transcriptase” reactions as a template were included in duplicate. Standard curves were generated using chromosomal DNA from gonococcal strain MS11 or MR661, and the relative standard curve method was used to determine transcript levels. The rpoB gene was used as a housekeeping control. Data were gathered using an Applied Biosystems StepOnePlus real-time PCR system. In all cases, at least three biological replicates were performed.

Beta-galactosidase assays

N. gonorrhoeae strains were grown in cGCBL for 3 hours from an OD₅₄₀=0.25. Because gonococci form aggregates during growth in liquid culture, it is difficult to obtain accurate measurements of optical density. A Bradford assay was therefore used to quantify protein amounts (mg protein), and this measurement was used in place of OD₆₀₀ in the Miller equation. A 2-ml volume of each culture was harvested by centrifugation following 20 minutes on ice, and the pellet was resuspended in 0.4 ml Z buffer containing 0.002% SDS (Miller, 1992). Three 100-µl aliquots of each sample were removed to a 96-well plate, and a 30-µl volume of ortho-nitrophenyl-β-galactoside (ONPG, 4 mg ml⁻¹ in Z buffer) was added. The OD₄₂₀ and OD₅₅₀ were measured at timed intervals (2 hours for strains expressing traH-lacZ fusions with high β-galactosidase activity and 8 hours for strains expressing traH-lacZ fusions with low β-galactosidase activity). Beta-galactosidase activity was calculated using the equation described by Miller except that OD₆₀₀ was replaced by mg of protein (Miller, 1992).