Sensing of intracellular Hcp levels controls T6SS expression in Vibrio cholerae

Significance

Assembly, firing, and recycling of the type 6 secretion system (T6SS) organelle is a costly process as each secretion event ejects hundreds of inner-tube Hcp molecules. In comparison to bacteria whose T6SSs are tightly regulated, bacterial strains that constitutively display T6SS activity would seem inefficient in their deployment of T6SS, especially if secretion is interrupted by environmental conditions that reduce available cellular energy. Here we have discovered a quantity-sensing feedback control that avoids wasteful accumulation of secreted proteins in several pathogens that express the T6SS. By sensing the efficiency of secretion of its major substrate (Hcp), this regulatory mechanism likely conserves energy and would therefore be helpful in responding to a variety of stress conditions that interrupt T6SS secretion.

Abstract

The type 6 secretion system (T6SS) is a bacterial weapon broadly distributed in gram-negative bacteria and used to kill competitors and predators. Featuring a long and double-tubular structure, this molecular machine is energetically costly to produce and thus is likely subject to diverse regulation strategies that are largely ill defined. In this study, we report a quantity-sensing control of the T6SS that down-regulates the expression of secreted components when they accumulate in the cytosol due to T6SS inactivation. Using Vibrio cholerae strains that constitutively express an active T6SS, we demonstrate that mRNA levels of secreted components, including the inner-tube protein component Hcp, were down-regulated in T6SS structural gene mutants while expression of the main structural genes remained unchanged. Deletion of both hcp gene copies restored expression from their promoters, while Hcp overexpression negatively impacted expression. We show that Hcp directly interacts with the RpoN-dependent T6SS regulator VasH, and deleting the N-terminal regulator domain of VasH abolishes this interaction as well as the expression difference of hcp operons between T6SS-active and inactive strains. We find that negative regulation of hcp also occurs in other V. cholerae strains and the pathogens Aeromonas dhakensis and Pseudomonas aeruginosa. This Hcp-dependent sensing control is likely an important energy-conserving mechanism that enables T6SS-encoding organisms to quickly adjust T6SS expression and prevent wasteful build-up of its major secreted components in the absence of their efficient export out of the bacterial cell.

The bacterium Vibrio cholerae is the causative agent of the diarrheal disease cholera with an estimated global disease burden of up to 4 million annual cases during the ongoing seventh cholera pandemic (1). V. cholerae deploys its type 6 secretion system (T6SS) in the environment and when colonizing host gastrointestinal systems to kill competing bacteria, improving colonization and virulence (2–4). The T6SS is a molecular weapon broadly distributed among gram-negative bacteria that acts analogously to a speargun, delivering long pointed tubular structures into nearby neighboring host or bacterial cells (5, 6). The T6SS spear is composed of hundreds to thousands of subunits of the Hcp protein, forming a tubular structure that is decorated with a spearhead consisting of a VgrG protein trimer sharpened to a point by a PAAR domain-containing protein (7–9). Toxic effector proteins can be attached to the spear through covalent extensions of these components or through noncovalent interactions (7, 10–14). Within the cytosol, a sheath forms around the Hcp tube that is made up of TssB and TssC protein subunits and this sheath in its extended state is anchored to the cell envelope through a membrane complex and baseplate (15, 16). The sheath is composed of hundreds of TssB and TssC subunits that are folded and assembled in a high-energy extended state within the cytosol (17). Upon an unknown signal, “firing” occurs and the sheath contracts to half its size in milliseconds, propelling the T6SS spear out of the cell with enough force to penetrate neighboring cells and thus deliver the toxic effectors loaded on its tip or present within the lumen of the Hcp tube component of the spear (12, 17–19). After contraction, the sheath is disassembled by the ClpV ATPase, which allows for recycling of sheath, baseplate, and membrane complex proteins for multiple rounds of sheath extension in a high-energy state (20–23), which is analogous to the stretched bands of a loaded speargun. After T6SS firing, the ejected components of the spear are lost to the extracellular milieu or to penetrated target cells and thus need to be replenished by de novo synthesis unless delivered by sister cells (18).

Regulation of the T6SS in V. cholerae is diverse, with some strains employing a constitutively active systems, including an older V52 clinical isolate of the O37 serogroup and environmental El Tor lineage strains, such as 2740-80 (6, 17, 24). More recent seventh pandemic V. cholerae strains, such as C6706 and A1552, have more tightly regulated T6SS that are not active under standard laboratory conditions but are activated in response to environmental signals including those within the infected host (4, 25–28). V. cholerae strains that regulate or constitutively express their T6SS all have a common genetic organization of their T6SS genes (Fig. 1A). The large virulence-associated secretion (VAS) cluster (under the control of the P_tssB promoter) primarily encodes T6SS components of the “speargun,” including the genes for the sheath, baseplate complex, membrane complex, and the bacterial enhancer binding protein (bEBP), VasH (6). There are also three auxiliary clusters that encode most secreted “spear” components. The hcp1 and hcp2 clusters have nearly identical P_hcp promoters and encode not only identical copies of the Hcp protein but also distinct VrgG and effector proteins (6, 29). The physical separation of the VAS cluster (encoding the speargun) and the auxiliary clusters, (encoding the spear) into distinct operons places them under different transcriptional regulation. To restore the T6SS apparatus following firing, cells reassemble de novo-synthesized spear components with recycled speargun components. At the same time, cells have to avoid undesirable and wasteful production of excess secreted substrates. How this continuous demand for secreted components is coordinated with their supply remains poorly understood.

Fig. 1.

Expression of T6SS auxiliary clusters are down-regulated in T6SS-null strains. (A) Genetic organization of the main T6SS gene clusters in V. cholerae. Genes encoding cellular structural components of the T6SS are shaded light gray, while secreted components are shaded dark gray and cytoplasmic accessory genes are white. (B) RNA-seq of WT and T6SS mutant (ΔtssM) strains after coincubation with E. coli prey at a 10:1 ratio for 30 min. The ratio of average RPKM values for each gene in the clusters are plotted and Student’s t test P values for each gene were compared to a Bonferroni-corrected significance value. The bars for each gene cluster are grouped and presented left to right in gene order. (C) qRT-PCR measuring relative expression of T6SS genes in WT and a T6SS (ΔtssC) mutant in liquid culture. Student’s t tests were used to compare expression differences for each gene. (D) Luminescent images of strains (upper label) with chromosomal lux reporters (lower label) and expressing from plasmids (left label) induced with 0.01% arabinose. Images are representative of three independent replicates. (E) Relative luminescence values were quantified from P_hcp2-lux strains in D and normalized to WT pBAD. One-way ANOVA with Sidak’s multiple comparison tests were used to compare WT with ΔtssC when expressing from either empty plasmid or pTssC. (F) Relative luminescence values were quantified from P_tssB-lux strains in D and normalized to WT pBAD. Graphs in C, E, and F show the mean ± SD of three independent replicates with each replicate represented as a dot. **P < 0.01, ***P < 0.001; ns, not significant.

Interestingly, the alternative σ-factor RpoN controls expression of these auxiliary clusters but not that of the VAS cluster (30), reflecting the different cellular demand for recyclable structural components and single-use secreted components that need to be newly synthesized after secretion. RpoN-mediated transcription requires bEBP proteins, which bind to enhancer sequences upstream of the promoter and then oligomerize into hexamers to activate their ATPase activity, and allow for interactions with RpoN and initiation of transcription (31, 32). bEBPs encode diverse N-terminal domains that are used to sense a variety of signals and modulate the activity state of the bEBP (32). For the V. cholerae T6SS, VasH, encoded in the VAS operon, is the bEBP responsible for activating transcription from the RpoN-dependent P_hcp promoters of auxiliary operons and mutants lacking VasH activity have no T6SS function (6, 33). Since expression of the auxiliary operons requires this potentially signal-sensing VAS-encoded gene, VasH may represent a key regulatory site for modulating expression of secreted components of the V. cholerae T6SS.

In this study we describe a regulatory mechanism of the V. cholerae T6SS that allows the bacteria to repress expression of T6SS-secreted genes when the T6SS structure or its recycling is disrupted. We show that this regulation occurs through a negative feedback loop in which increased cytoplasmic levels of Hcp are sensed by the N-terminal domain of VasH resulting in reduced expression from auxiliary cluster P_hcp promoters that drive the expression of the secreted components, including Hcp. We show that this regulation is present in strains with constitutively active T6SS as well as seventh pandemic lineage strains. We also show that similar regulatory phenotypes for controlling expression of T6SS genes are present in other T6SS-encoding species of the genera Pseudomonas and Aeromonas. This study thus uncovers a general quantity-sensing feedback mechanism for regulating the T6SS system based on its functionality in secretion, and thus represents a new example of a common mechanism to conserve energy when deploying and assembling a complex organelle.

Results

Expression from Auxiliary T6SS Clusters Is Reduced in T6SS-Null Strains.

In previous work we used transcriptomics to analyze the response of Escherichia coli cells when attacked by V. cholerae strain V52, which constitutively expresses a functional T6SS apparatus (34). Here we reanalyzed this dataset to compare WT V52 and its T6SS null ΔtssM (ΔvasK) structural gene mutant in competition with E. coli prey. Of the 13 differentially regulated genes, 10 are encoded in the auxiliary clusters and down-regulated in the ΔtssM mutant relative to WT (Fig. 1B and Dataset S1). Except for vgrG3, a spear gene with its own RpoN-dependent promoter (30), no VAS-encoded genes were found to be differentially regulated, suggesting that T6SS structural components were not subject to down-regulation in the T6SS-null ΔtssM mutant. Using a different T6SS structural mutant ΔtssC, we verified this phenotype by qRT-PCR and observed similar hcp (auxiliary cluster) down-regulation in the T6SS-null mutant, while tssB (VAS cluster) expression was unaffected (Fig. 1C). These data show that in the absence of T6SS activity, expression of secreted T6SS components from the auxiliary T6SS clusters is reduced, while that of VAS-encoded cellular/recyclable components is unaffected.

To test whether the observed transcriptional control is exerted through the promoters of the auxiliary clusters, we integrated a luciferase reporter gene under the control the P_hcp2 auxiliary or the P_tssB VAS cluster promoter into the genome of either the WT or the ∆tssC mutant strains. Reporter and their isogenic parental strains displayed equal levels of T6SS-dependent killing of E. coli, indicating that T6SS activity is unaffected by the chromosomally integrated reporter constructs (SI Appendix, Fig. S1A). Consistent with our transcriptome and qRT-PCR data, P_hcp2, but not P_tssB-driven reporter gene expression, was significantly reduced in the ∆tssC mutant compared to WT (Fig. 1 D–F). Exogenous complementation of TssC restored P_hcp2 reporter expression (Fig. 1 D and E), as well as T6SS-dependent killing activity, to WT levels (SI Appendix, Fig. S1B). Taken together, these data reveal the existence of a transcriptional control mechanism acting on the auxiliary cluster promoters that connects T6SS activity with the expression of T6SS auxiliary, but not the VAS-encoded genes.

Accumulation of Intracellular Hcp Signals Reduced Expression of Auxiliary Clusters.

To test our hypothesis that T6SS activity is linked to auxiliary cluster gene expression, we used our lux reporters in a panel of structural gene mutants with disruptions of the membrane complex, baseplate, sheath, and needle tip (ΔvgrG2) (18, 35), as well as the clpV gene required for recycling of the T6SS sheath. P_tssB-driven reporter expression was unaffected in all strains except the ΔtssA (encoding a chaperone) mutant (SI Appendix, Fig. S2A) (36). P_hcp2-driven reporter expression was significantly reduced in all mutant strains, with one notable exception: the ∆hcp1/2 mutant. Deletion of both hcp1 and hcp2 encoded on auxiliary clusters 1 and 2, respectively, fully restored gene expression from the P_hcp2 promoter as measured both by reporter gene expression and qRT-PCR (Fig. 2 A and B). Additionally, in this Δhcp1/2 background, deletion of tssC no longer reduced P_hcp2-lux or auxiliary cluster expression (Fig. 2 A and B).

Fig. 2.

Hcp accumulation in the cytoplasm results in reduced expression from auxiliary clusters. (A) Luminescent images of the indicated strains carrying chromosomal P_hcp2-lux reporters. (B) qRT-PCR measuring relative T6SS gene expression. One-way ANOVA with Sidak’s multiple comparisons tests were used to compare strains for each gene target. (C) Western blot analysis of Hcp in cell pellets (P), supernatant (SN), and total culture (T). α-RpoB acts as a cellular loading control. (D) Luminescent images of the indicated strains (upper label) carrying chromosomal P_hcp2-lux reporter and expression plasmids (left label). Grown on LB + 0.2% arabinose plates. (E) Relative luminescence values were calculated for the strains in D and normalized to WT pBAD. One-way ANOVA with Sidak’s multiple comparison tests were used to compare Hcp expressing strains with their empty plasmid controls. Images in A, C, and D are representative of three independent replicates. Graphs in B and E show the mean ± SD of three independent replicates with each replicate represented as a dot. *P < 0.05, **P < 0.01, ***P < 0.001; ns, not significant.

Based on these observations, we hypothesized that in the absence of T6SS activity, Hcp accumulation in the cytosol could serve as a signal to repress the expression of auxiliary T6SS clusters. To exert this function, Hcp should accumulate at detectable levels in the cytosol of a T6SS-null strain despite being transcriptionally repressed. Western blot analysis of Hcp revealed that this protein is indeed detectable in T6SS-null ΔtssC cells both in the cellular pellet and total (pellet plus supernatant) but not in the supernatant fraction (Fig. 2C). Active T6SS-dependent secretion of Hcp in the WT strain is evidenced by the band detected in the supernatant harvested from these cells and explains why this protein is detected at lower levels in the cellular pellet fraction compared to the ∆tssC mutant strain (Fig. 2C). These data suggest that the level of the Hcp protein in the cytosol may be a key signal mediating down-regulation of the T6SS auxiliary gene clusters.

To test if Hcp accumulation acts as a signal for the down-regulation of auxiliary clusters, we overexpressed Hcp from a plasmid in WT, T6SS-null ∆tssC, and Δhcp1/2 cells with the luminescent reporters. Overexpression of Hcp caused a significant reduction of P_hcp2-lux but not P_tssB-lux reporter gene expression (Fig. 2 D and E and SI Appendix, Fig. S2B). This reduction is most notable in the Δhcp1/2 ΔtssC strain where reintroducing Hcp resulted in the largest drop in expression to a level similar to the ΔtssC strain containing an empty plasmid (Fig. 2 D and E). Hcp overexpression also caused a drop in expression of P_hcp2-lux in WT and ΔtssC cells, while the Δhcp1/2 mutant displayed only a minor decrease (Fig. 2E). This suggests that the level of auxiliary cluster expression is inversely related to intracellular Hcp concentration. Notably, these observations can be extended to the T6SS-dependent E. coli killing activity observed in these cells, where overexpression of Hcp in WT cells causes a detectable decrease in killing capacity (SI Appendix, Fig. S2C). Collectively, these data suggest that Hcp is part of a negative feedback loop where it acts as a signal to decrease its own expression and this signal is amplified when Hcp secretion by T6SS is disrupted through mutations that affect organelle assembly (e.g., ΔtssM).

Intracellular Levels of Hcp Are Sensed by the N-Terminal Domain of VasH.

VasH, the bEBP responsible for σ⁵⁴ activation and expression of the auxiliary clusters, is composed of three conserved domains: the N-terminal regulatory GAF domain, the σ⁵⁴-activator domain (AAA+), and the C-terminal DNA binding domain (HTH) (Fig. 3C) (33). In bEBPs, the regulatory function of the GAF domain is usually modulated by direct sensing of a small-molecule ligand that alleviates steric inhibition of the AAA+ domain, which in turn activates the cognate σ-factor (32). Whether the N-terminal domain modulates VasH activity in V. cholerae and what signals it may respond to remains to be fully understood. Here, we tested the possibility that elevated levels of intracellular Hcp are sensed by the VasH N-terminal domain to repress RpoN activation and thus transcription from the P_hcp auxiliary cluster promoters. We reasoned that if this is the case, exogenous supply of ligand-free full-length VasH to the system may be sufficient to overcome P_hcp down-regulation in the T6SS ∆tssC mutant that displays elevated levels of intracellular Hcp. Our data show that this is indeed the case since overexpression of VasH, but not an unrelated protein control, fully restores P_hcp-lux reporter gene expression to WT levels in the ∆tssC mutant background (Fig. 3 A and B).

Fig. 3.

The N terminus of VasH responds to accumulating cytoplasmic Hcp. (A) Colonies of strains (left label) carrying the chromosomal hcp₂-lux reporter and indicated plasmid (upper label). Grown on LB + 0.1%arabinose. (B) Quantification of luminescence in A normalized to WT pGFP. One-way ANOVA with Sidak’s multiple comparison tests were used to compare WT with ΔtssC when expressing either GFP or VasH. (C) Domain organization of VasH and mutant VasH^ΔN mutant. VasH contains an N-terminal GAF domain, a central σ54 interacting domain and a C-terminal helix-turn-helix domain. VasH spans amino acids 1 to 530 while VasH^ΔN spans amino acids 190 to 530. (D) Colonies of indicated strains carrying chromosomal hcp₂-lux reporters. (E) Confirmation of vasH^ΔN mutant phenotypes seen with lux reporters by qRT-PCR. Student’s t tests were used to compare the strains for each gene target. (F) Colonies of strains (upper label) carrying chromosomal hcp₂-lux reporters with the indicated plasmid (left label). Grown on LB + 0.2% arabinose. (G) Relative quantification of luminescence in E normalized to vasH^ΔN pBAD. One-way ANOVA with Sidak’s multiple comparison tests were used to compare Hcp expressing strains with their empty plasmid controls. (H) Western blot showing lysate and elution fraction of His-pulldown. Hcp-His and VasH-FLAG or VasH^ΔN -FLAG were coexpressed from pBAD plasmids in V. cholerae. (A, D, F, and H) Images are representative of three independent replicates. (B, E, and G) Graphs show the mean ± SD of three independent replicates with each replicate represented as a dot. ***P < 0.001; ns, not significant.

Next, to directly measure the influence of the VasH N-terminal domain on auxiliary cluster gene expression, we generated vasH^ΔN mutants in both WT and T6SS-null ∆tssC backgrounds containing the luminescent reporters (Fig. 3C). We found that while luminescence intensity was overall lower in the vasH^ΔN strains, absence of the VasH N-terminal domain completely abolished the Hcp-mediated P_hcp-lux reporter gene and hcp repression observed in the T6SS-null ∆tssC mutant with full-length VasH (Fig. 3 D and E). We observed no changes in the VAS cluster nor P_tssB reporter gene expression in vasH^∆N strains, as measured by qRT-PCR (Fig. 3E and SI Appendix, Fig. S3A). The intermediate P_hcp2-lux reporter luminescence intensity of vasH^ΔN strains, which correlates with their reduced E. coli killing activity, suggests that the N-terminal domain is required for proper VasH-mediated activation of auxiliary cluster expression, as has been previously observed (Fig. 3D and SI Appendix, Fig. S3B) (33).

Since Hcp overexpression resulted in reduced expression of the P_hcp2-lux reporter in strains with full-length VasH (Fig. 2 D and E), we tested if deletion of the N-terminal domain of VasH would abolish this response. Interestingly, Hcp overexpression in vasH^ΔN strains did not result in changes in P_hcp2-lux or P_tssB-lux reporter gene expression in either the WT or ΔtssC backgrounds (Fig. 3 F and G and SI Appendix, Fig. S3C). This suggests that the N-terminal domain of VasH is important for sensing the intracellular accumulation of Hcp that leads to transcriptional repression of the auxiliary clusters. A simple mechanism that would explain this phenotype would be the existence of a direct protein–protein interaction between Hcp and the VasH N terminus. Indeed, when performing a pull-down assay Hcp was able to interact with full-length VasH but not VasH^ΔN (Fig. 3H). Collectively, our data suggest a model in which T6SS activity is logically linked by a negative transcriptional feedback to the synthesis of its secreted components; when T6SS secretion is disrupted, accumulation of the secretion substrate Hcp acts as a signal to repress its own synthesis

Negative Feedback Regulation Is Conserved in the Seventh Pandemic Lineage V. cholerae Strains and in Other Bacteria.

To determine if the negative feedback regulation of T6SS auxiliary clusters is conserved in the medically relevant seventh pandemic lineages of V. cholerae, we used the environmental O1 El Tor 2740-80 strain (17) and the O1 El Tor strain C6706 (37). Much like in V52, differential expression analysis of RNA-sequencing (RNA-seq) data obtained from 2740-80 WT and a ∆tssB mutant showed that only the hcp1 and hcp2 clusters are down-regulated in the absence of an active T6SS while expression of the VAS cluster genes is unchanged (Fig. 4A and Dataset S2). Consistent with these data, P_hcp2-lux, but not P_tssB-lux, -driven reporter gene expression in 2740-80 ∆tssC was down-regulated compared to WT (Fig. 4B). Conversely, upon deletion of the N-terminal domain of 2740-80 VasH, auxiliary cluster repression in the T6SS-null ∆tssC mutant was completely abolished. Notably, and in contrast with what we observed in V52, VasH^∆N presented no defect in its ability to activate transcription of the auxiliary cluster promoters and retained WT-level E. coli killing activity (Fig. 4B and SI Appendix, Fig. S4A). The 2740-80 colonies carrying the P_tssB-lux reporter are much brighter than V52 colonies, suggesting higher levels of vasH expression, which could explain the WT levels of hcp expression seen in vasH^ΔN 2740-80 strains (SI Appendix, Fig. S4B). These findings underscore our model that in the absence of its N-terminal regulatory domain, VasH fails to sense elevated intracellular Hcp levels remaining competent to activate σ⁵⁴ for continuous expression of T6SS substrates from the auxiliary clusters.

Fig. 4.

T6SS activity sensing regulation is conserved in V. cholerae strains and may be present in other species. (A) Heatmap depicting mean centered log₂ RPKM expression values of the T6SS auxiliary and main clusters of V. cholerae 2740-80 WT, ΔtssB, and complemented ∆tssB (ΔtssB ptssB). The differential expression log₂ ratio is the base 2 log of the ratio of the expression values between the WT and ΔtssB (DiffE log₂). (B) V. cholerae 2740-80 strains (upper label) carrying indicated luminescent reporter (left label). (C) V. cholerae C6706 strains (upper label) carrying indicated luminescent reporter (left label). (D) A. dhakensis SSU strains (upper label) carrying indicated luminescent reporter (left label). (E) qPCR measuring relative expression of T6SS genes under inducing conditions for P. aeruginosa. Student’s t tests were used to compare strains for each gene target. (B and C) Images are representative of three independent replicates. (E) Graph shows the mean ± SD of three independent replicates with each replicate represented as a dot. **P < 0.01, ***P < 0.001.

Because C6706 displays no T6SS activity under laboratory conditions, we used a tsrA::kan C6706 transposon mutant strain with a nonsense mutation in luxO shown to constitutively secrete Hcp (25). We confirmed this strain can kill E. coli (SI Appendix, Fig. S4C) and measured hcp and tssB expression using the luminescent reporter. Our results show reduced auxiliary reporter expression in ΔtssC relative to the parental strain, while the vasH^ΔN background strains had an intermediate phenotype similar to what was observed in V52 (Fig. 4C).

The Aeromonas dhakensis SSU T6SS has a similar genetic organization to that of V. cholerae in which the secreted components are encoded on auxiliary clusters under the control of an orthologous VasH bEBP encoded in the major cluster (38). Using luminescent reporters, we observed that T6SS-null ΔtssM cells had reduced auxiliary cluster expression and unchanged structural cluster expression relative to WT, suggesting that A. dhakensis shares a similar negative feedback mechanism with V. cholerae (Fig. 4D).

We also tested the H2-T6SS of Pseudomonas aeruginosa PAO1 since, like V. cholerae, it has a large cluster that encodes a bEBP and auxiliary clusters encoding secreted components (39). Unlike V. cholerae however, this bEBP represses expression from the major cluster through RpoN (39). Interestingly, when grown under conditions that induce expression of the H2-T6SS (39), a T6SS-null strain ΔtssB strain showed reduced expression of both auxiliary (hcp2) and major (tssC2) clusters (Fig. 4E). This suggests that P. aeruginosa also employs a T6SS secretion activity-linked feedback regulation mechanism.

Discussion

How cells sense and control the quantity of certain proteins and metabolites to avoid wasteful build-up and maintain homeostasis during growth is an intriguing question across all biological systems, from single-cell bacteria to multicellular organisms. Restricting the assembly of energetically expensive and complex organelles in low-energy environments or when they are defective in function is critical to avoid wasting energy needed for other biological processes. For the V. cholerae T6SS apparatus, structural proteins are recycled while secreted proteins need to be regenerated for each round of firing. The organization of structural genes and secreted genes into distinct operons suggests the possibility that they can be differentially regulated in response to an inactive system to save energy. Here we show that when T6SS secretion of V. cholerae V52 is disrupted by a structural mutation or recycling defect, expression of the secreted components but not the structural components is repressed by a mechanism that senses elevated intracellular Hcp levels. We further show that the N-terminal domain of VasH is essential to establish this negative feedback and likely senses Hcp levels by directly interacting with this protein. At this time we do not know whether VasH senses hexamers or monomers of Hcp, although the former is the major form of the Hcp protein when expressed in heterologous cells (8, 40, 41).

We propose a model whereby T6SS is logically linked to the synthesis of its components by a negative transcriptional feedback loop (Fig. 5). In fast-growing WT cells with an active T6SS, intracellular Hcp levels are maintained at steady state by continuous de novo synthesis, with most Hcp recruited to assembling T6SS tubes with some remaining free in the cytoplasm. Upon disruption of T6SS activity, Hcp accumulates in the cytosol, allowing for increased interaction with VasH and inhibition of its own synthesis. Additionally, reduced T6SS activity, perhaps in a low-energy environment, could also result in increased intracellular levels of Hcp, which could result in varying levels of Hcp repression along a Hcp concentration gradient. The repression of Hcp expression is dependent on the N-terminal domain of VasH, so its disruption makes cells unresponsive to varying intracellular concentrations of Hcp. Similar regulatory mechanisms have been previously defined for both the type 3 secretion systems that include the bacterial flagellum, as well as specialized systems involved in secretion of virulence-related effectors. In both instances, a substrate of the system accumulates in the cytosol while the system is inactive and exerts a repressive effect on expression of other secreted genes (42–46). Upon activation of secretion, cytoplasmic levels of the secreted repressor drop and expression of secreted genes is activated (42–46). Additionally, a link between Hcp and gene regulation has been recently described in Lysobacter enzymogenes (47). When grown in nutrient-poor conditions, Hcp is expressed but not secreted, allowing it to interact with the transcription factor Clp (cAMP-receptor like protein) in turn activating expression of heat-stable antifungal factor (47).

Fig. 5.

Model of Hcp⁻ feedback regulation in V. cholerae. (A) In T6SS active cells, Hcp is produced and secreted normally. A small amount is retained in the cytoplasm where it can interact with the N-terminal domain of VasH and slightly limit VasH’s ability to activate hcp expression. (B) In cells with reduced or abolished T6SS activity, Hcp accumulates in the cytoplasm. This accumulation of Hcp allows more to interact with VasH through its N-terminal domain allowing for a strong repression of hcp expression. (C) In cells where the N-terminal domain of VasH is deleted, no matter how much Hcp accumulates in the cytoplasm, it does not interfere with the ability of VasH^ΔN to mediate hcp expression. Green lines indicate an activating effect while red lines indicate an inhibitory effect. Additionally, the larger solid lines indicate a strong effect while light dashed lines indicate a weak effect.

The function of the VasH N-terminal domain has been elusive due to strain-to-strain variations in phenotype, despite encoding virtually identical proteins. In V. cholerae V52, deletion of the N-terminal domain reduces Hcp expression (33) (Fig. 3D), while in the strain A1552, overexpression of a N-terminally truncated VasH increases Hcp expression relative to the overexpression of the full-length protein (48). Our pull-down analysis suggests that the function of this regulatory domain is to sense intracellular Hcp accumulation (Fig. 3H). Considering that VasH-like bEBPs are also found in a large number of gamma-proteobacteria (48) and that T6SS inactivation results in down-regulation of Hcp in multiple V. cholerae strains as well as A. dhakensis and P. aeruginosa (Fig. 4 A–E), such Hcp quantity-sensing regulation is likely not limited to V. cholerae but conserved among diverse species that encode the T6SS.

Transcription of the T6SS large and auxiliary clusters is known to be controlled by multiple environmental cues including temperature, salinity, and quorum sensing (25, 27, 37, 49). These multiple signals may act in concert or independently to control T6SS expression, thus making its regulation highly complex and prone to strain variations. We observed strain variation in the phenotypes displayed when the VasH N-terminal domain was deleted. In V52 and C6706, deletion of the N-terminal GAF domain resulted in reduced expression of the P_hcp2-lux reporter (Figs. 3D and 4C) while in 2740-80 the deletion had no effect (Fig. 4B). Strain 2740-80 appears to have higher levels of T6SS expression overall and could potentially be producing excess VasH, which could be enough to mask this phenotype (SI Appendix, Fig. S4B). In addition, expression of RpoN was shown to be undetectable below OD₆₀₀ 2.0 in V. cholerae strain A1552 (48). This growth-phase–dependence may be also strain-dependent because RpoN is a known master regulator that controls a number of exponential-phase–active functions including motility and nitrogen utilization in the V. cholerae O1 classic strain 0395 (50, 51).

What is the evolutionary advantage of coupling T6SS activity with supply and demand for its own components? It is possible that in a low-energy environment, where ATP-dependent ClpV recycling of the T6SS sheath is limited, this mechanism allows for the cell to shift the amount of Hcp expressed to match the lower requirements of infrequent rounds of T6SS assembly and firing. Alternatively, this mechanism may increase fitness when T6SS-expressing strains encounter inhibitors of T6SS function. Considering the recent discovery of immunity-gene–independent protection mechanisms against the T6SS (52–54), and the coexistence of T6SS⁺ and T6SS⁻ species in diverse communities, production of a small molecule capable of disrupting T6SS function might have evolved as a defense mechanism. Several studies that have identified T6SS-inhibitory small molecules support the possibility of such a hypothesis (34, 55, 56), although no natural inhibitor has been described.

Finally, V. cholerae strains show diverse expression patterns of the T6SS, with some displaying constitutive activity and others tight regulation (6, 24, 25, 28). We show that 2740-80 (an environmental isolate) and C6706 (a clinical isolate) that are both related to the seventh pandemic O1 El Tor lineage, and that display constitutively active and tightly controlled T6SS, respectively, also display the described negative feedback mechanism (Fig. 4 A and C). This conserved T6SS regulatory mechanism will help us better understand the strategies employed by environmental and clinical V. cholerae isolates that contribute to their success in survival and virulence in diverse environmental and host conditions.

Materials and Methods

Bacterial Strains and Plasmids.

The strains and plasmids in this study are listed in SI Appendix, Tables S1 and S2. Strains were grown shaking at 37 °C in LB media (0.5% NaCl) or on LB agar unless otherwise indicated. Antibiotics were added for plasmid maintenance or strain selection at the following concentrations: 100 µg mL⁻¹ streptomycin, 100 µg mL⁻¹ carbenicillin, 50 µg mL⁻¹ kanamycin, 20 µg mL⁻¹ gentamicin, 2.5 µg mL⁻¹ chloramphenicol. The plasmids, reporters and suicide vectors to make chromosomal mutants were generated using standard molecular cloning techniques and verified by Sanger sequencing.

RNA-Seq.

Bacterial growth, RNA purification and preparation, and data analysis were performed as previously described (34). Briefly, 1 × 10⁹ V. cholerae V52 cells were mixed with 1 × 10⁹ E. coli cells, filtered onto a nitrocellulose membrane and incubated on an LB agar plate at 37 °C for 30 min. RNA was extracted by mixing cells in boiling SDS lysis buffer before mixing with acidic phenol. Samples were incubated at 65 °C for 5 min, on ice for 10 min, then centrifuged at 20,000 × g for 10 min. RNA was purified from the aqueous phase using a Direct-zol RNA MiniPrep Kit (Zymo Research). The MICROBExpress Bacterial mRNA Enrichment Kit (Life Technologies) was used to degrade ribosomal RNA for library preparation.

Sequencing libraries were prepared using a NEBNext mRNA Library Prep Kit (New England Biolabs) and a NEBNext Multiplex Oligos for Illumina Kit (New England Biolabs), as per the manufacturer’s instructions. Sequencing was performed using an Illumina HiSeq 2000 platform in the Biopolymer core facility at Harvard Medical School. Sequencing reads (50 nt in length) were mapped to the V. cholerae N16961 reference genome using the software CLC Genomics Workbench (Qiagen), as previously described (41). Genes without mapped reads in any of the biological replicate samples were filtered and excluded from downstream analysis. Gene-expression levels represented as the reads per kilobase per million (RPKM) mapped read values were used to calculate fold differences in expression. The Student’s t test was used for estimating statistical significance with computed P values compared to a Bonferroni-corrected P value of 1.47 × 10⁻⁵ (0.05 ÷ 3,409 genes).

For 2740-80, two biological replicates of each strain were grown to OD_600nm of 0.5 and harvested by centrifugation, resuspended in 1 mL of TRIzol (Thermo Fisher Scientific), and incubated for 5 min at room temperature before the addition of 200 μL of chloroform. Samples were centrifuged at a relative centrifugal force of 13,000 for 10 min. The aqueous phase was mixed with 200 μL of 100% ethanol, and the mixture was transferred to Purelink RNA Mini columns (Thermo Fisher Scientific) for purification following the manufacturer’s instructions. RNA was treated with Turbo DNase (Thermo Fisher Scientific) according to the manufacturer’s instructions and stored at –80 °C. RNA-seq libraries were prepared following the Ovation Complete Prokaryotic RNA-Seq (NuGEN) kit instructions. Reads were aligned using the Geneious Prime (57), v2019.0.4, software package’s default settings to the V. cholerae O1 biovar El Tor strain N16961 (NC_002505.1, NC_002506.1) reference genome. Differential expression analysis was performed using the DESeq2 (58) R package within the Geneious interface. The 2740-80 RNA-seq datasets obtained in this study have been deposited in the National Center for Biotechnology Information (NCBI) Sequence Read Archive (SRA) under accession number PRJNA713318.

qRT-PCR.

V. cholerae was subcultured to OD_600nm of 0.025 and grown for 3 h 10 min (to OD_600nm of ∼1.1 to 1.2). A. dhakensis was subcultured 1/100 and grown for 3 h (to OD_600nm of ∼1 to 1.1). P. aeruginosa was subcultured 1/50 and grown for 5 h (to OD_600nm of ∼2.5). RNA was extracted by lysing cultures with hot SDS (1% SDS, 2 mM EDTA), then mixing with hot acidic phenol (65 °C, pH 4.5). Phenol mixtures were incubated at 65 °C for 5 min with occasional mixing, incubated on ice for 5 min before separating the phases by centrifuging at 16,000 × g for 2 min. RNA was purified from the aqueous phase using the Illustra RNAspin Kit (Cytiva). SuperScript IV Reverse Transcriptase (Invitrogen) was used for reverse transcription and qRT-PCR was performed in technical triplicate using Fast SYBR Green Master Mix (Applied Biosystems). Relative quantification was done using the standard curve method. In V. cholerae and A. dhakensis, the 16s gene rrsA was used as an internal control while proC was used in P. aeruginosa. All primers used are listed in SI Appendix, Table S3.

Luciferase Reporter Assays.

Luminescent strains were constructed using a modified pVIK112 plasmid (59) with the lacZ gene replaced with a lux operon. The P_hcp2-lux reporter contains 866-bp upstream of the predicted translation start site to the 99th codon of hcp2. The P_tssB-lux reporter contains 551-bp upstream of predicted translation start site to the 15th codon on tssB. The 16-h cultures grown shaking in a 96-well plate were diluted 1:5. Five microliters were spotted onto LB agar plates with required antibiotics and the indicated inducer and incubated for 6 h at 37 °C. Luminescence was measured using a Bio-Rad ChemiDoc-MP with images collected every 10 s for 3 min with no illumination or filters. Quantification of luminescence intensity was performed using Fiji.

Cell Fractionation.

Cells were subcultured to OD_600nm ∼1.0 then normalized to OD_600nm 1. Samples were taken and mixed with SDS sample buffer to make the “total culture” fraction. Next, 500 µL of culture was centrifuged at 10,000 × g for 5 min before using the supernatant to perform a secretion assay as described previously to make the “supernatant” fraction (60). The pelleted cells were resuspended in 500 µL fresh LB and used to make the “cell pellet” fraction.

Protein Pull-Down Assays.

Assays were performed as described previously (10). Briefly, V. cholerea cultures containing plasmids encoding Hcp-His, VasH-FLAG, or VasH^ΔN-FLAG were grown to OD_600nm 0.6 and induced with 0.1% arabinose for 3 h at 37 °C. Cultures were concentrated and lysed by sonication before incubation with cobalt-NTA beads. The beads were washed with 40 mM imidazole and protein eluted in 500 mM imidazole, then analyzed by Western blotting.

T6SS Competition Assay.

The assessment of T6SS activity was conducted as described previously (53). Briefly, killer strains were subcultured for 3 h (to OD_600nm ∼1.0 to 1.5) with chloramphenicol (2.5 μg/mL) to maintain plasmids as necessary. Prey (from overnight cultures) and killer strains were mixed at a 10:1 ratio, spotted on LB plates and incubated at 37 °C. Agar plugs were removed after 3 h using wide-bore pipette tips, followed by resuspension and serial dilution plating for colony forming units. For plasmid expression, arabinose was added for the last hour of subcultures at 0.2% or 0.01% (final concentration) and to the LB plate at 0.1% or 0.01% (respectively), as indicated in the figure legends.

Data Availability

RNA-seq data have been deposited in the NCBI BioProject (accession no. PRJNA713318). All other study data are included in the article and supporting information. Previously published data were used for this work (see ref. 34).