T6SS secretes an LPS-binding effector to recruit OMVs for exploitative competition and horizontal gene transfer

The Fur-regulated T6SS1 plays an important role in iron acquisition in C. necator

To explore the function of T6SS1 (Reut_A1713 to Reut_A1733) in C. necator (Fig. S1A), we analyzed the T6SS1 promoter and identified a Fur binding site (AGAAATA) upstream of gene reut_A1733. This Fur binding site was highly similar to the Fur-box reported in E. coli [38], with a probability score of 2.25 (out of a maximum score = 2.45) (Fig. S1B), which was calculated by applying the position weight matrix to a sequence [39]. Incubation of the T6SS1 promoter probe with purified Fur protein led to decreased mobility of the probe in the electrophoretic mobility shift assay, suggesting a direct interaction between Fur and the T6SS1 promoter (Fig. 1A). To further determine the function of Fur on the expression of T6SS1, a single-copy P_T6SS1::lacZ fusion reporter was introduced into the chromosomes of C. necator wild-type (WT), Δfur deletion mutant, and the Δfur(fur) complementary strain. Compared to WT, the P_T6SS1::lacZ promoter activity was significantly increased in the Δfur mutant (about 2.2-fold), and this increase could be restored by introducing the complementary plasmid pBBR1MCS-5-fur (Fig. 1B). Similar results were obtained by analyzing the expression of T6SS1 core component genes (hcp1, clpV1, vgrG1, and tssM1) with qRT-PCR (Fig. S1C). These results demonstrate that the expression of T6SS1 in C. necator is directly repressed by Fur, the master regulator of genes involved in iron homeostasis in many prokaryotes [40, 41].

To examine whether C. necator T6SS1 plays a role in iron homeostasis, we measured the intracellular iron contents of relevant strains in M9 medium using ICP-MS, and found no difference between the WT and T6SS1 mutant (ΔtssM1) (Fig. S2A). We speculated that the iron transport capacity of T6SS1 was masked by other major iron transport systems in C. necator, such as the cupriabactin siderophore iron transport system [34], and the FeoABC ferrous iron transport system [42]. As expected, deletion of tssM1 in ΔcubEΔfeoB (hereafter referred to as Δ2Fe), a double mutant defected in both cupriabactin and FeoABC iron transport systems, significantly reduced intracellular iron level. However, the defect of the Δ2FeΔtssM1 mutant in iron accumulation was fully restored by complementation of tssM1 (Fig. 1C). By contrast, the accumulation of other metal ions (zinc, sodium, magnesium) was not affected by the deletion of tssM1 in the Δ2Fe mutant (Fig. S2B). These results demonstrate that the C. necator T6SS1 is directly regulated by ferric uptake regulator Fur and is involved in iron acquisition.

T6SS1 effector TeoL contributes to acquisition of iron from OMVs

Lin et al. [29] reported that P. aeruginosa T6SS is involved in iron uptake by recruiting OMVs through TseF, a PQS-binding effector. Downstream of vgrG1 in the C. necator T6SS1 gene cluster, we also identified a putative T6SS effector (Reut_A1725, hereafter, TeoL). While significant amounts of TeoL could easily be detected in culture supernatant of WT, the secretion of TeoL was completely abolished in ΔtssM1 (Fig. 2A), and almost completely abolished in the ΔclpV1 and Δhcp1 mutants (Fig. S3A). Even the residual TeoL secretion was completely abolished in ΔclpV1ΔclpV2 and Δhcp1Δhcp2 double mutants defected in both T6SSs in C. necator (Fig. S3B). Moreover, the secretion defects of these T6SS mutants could be completely restored to WT levels by complementation of corresponding T6SS1 component genes (Figs. 2A and S3). These results demonstrate that TeoL is an effector protein mainly secreted by T6SS1, though limited substrate cross recognition among T6SS1 and T6SS2 existed.

To examine the role of TeoL in iron acquisition, we produced a Δ2FeΔteoL mutant that consisted of a teoL deletion in the Δ2Fe background. While the Δ2FeΔteoL mutant grew equally in M9 medium as the Δ2Fe mutant, its growth was severely impaired compared to the Δ2Fe mutant in the iron-depleted M9 medium that contained 4.0 µM of the iron chelator ethylenediamine-N,N′-bis(2-hydroxyphenylacetic acid) (EDDHA) (Fig. S4A). However, the growth defect of the Δ2FeΔteoL mutant was completely rescued by plasmid-borne expression of teoL, or by adding excessive Fe³⁺ (0.5 μM) to the iron-depleted medium (Fig. S4A). Moreover, the Δ2FeΔteoL mutant exhibited significantly reduced intracellular iron levels compared to the Δ2Fe mutant and the Δ2FeΔteoL(teoL) complemented strain (Fig. 2B), though the accumulation of other metal ions was not affected (Fig. S2C). These results suggest that TeoL is involved in iron acquisition. However, we were unable to detect interactions between TeoL and Fe³⁺ (Fig. S5), suggesting that TeoL may not directly sequester iron as in the case of metal-binding T6SS effectors for metal ions transportation [43, 44].

To examine whether TeoL is involved in iron utilization from OMVs, we determined the effects of OMVs on the growth of Δ2FeΔteoL in iron-depleted M9 medium containing 5.5 μM EDDHA. As shown in Fig. S4B, both Δ2Fe and Δ2FeΔteoL(teoL) strains, but not Δ2FeΔteoL, exhibited increased growth with adding OMVs purified from distantly related Gram-negative bacteria, P. aeruginosa PAO1 and Yersinia pseudotuberculosis YPIII. Unexpectedly, the growth of Δ2FeΔteoL also increased following adding OMVs purified from the C. necator WT and ΔteoL(teoL) complemented strain (Fig. 2C). However, the adding of OMVs purified from the C. necator ΔteoL mutant had no effect on increasing Δ2FeΔteoL growth (Fig. 2C). These results demonstrate that TeoL plays crucial roles in acquiring iron derived from OMVs.

TeoL is required for OMV recruitment in C. necator

The involvement of TeoL in acquiring iron derived from OMVs prompted us to further explore the role of TeoL in OMV recruitment. Thus, we incubated mCherry-labeled C. necator WT and ΔteoL mutant cells with FITC-labeled OMVs derived from the ΔteoL mutant. After 4 h of incubation, cells were washed and imaged with confocal microscopy and the percentages of cells exhibiting both mCherry and FITC fluorescence were quantified to measure the direct association between OMVs and bacterial cells. Although 33.3% of WT cells exhibited both mCherry and FITC fluorescence after incubation with fluorescent OMVs derived from ΔteoL mutant, the percentage of co-localized ΔteoL mutant cells decreased to 6.1% following incubation with OMVs derived from the ΔteoL mutant (Fig. 3), indicating that TeoL is involved in OMV recruitment.

To gain further insight into the role of TeoL in OMV recruitment, the interaction between TeoL and OMVs was examined using an assay based on glutathione-S-transferase (GST) pull-down. First, we introduced a plasmid expressing the OMV marker OmpW [45] tagged with the VSVG epitope into the ΔteoL mutant of C. necator. OmpW-VSVG containing OMVs purified from this strain were incubated with glutathione beads coated with GST-TeoL or GST, respectively, and OMVs captured on the glutathione beads were detected by immunoblot after SDS-PAGE using an anti-VSVG antibody for detecting the OmpW marker. As shown in Fig. S6A, capture of OmpW-VSVG containing OMVs was observed for the GST-TeoL fusion protein but not for the GST protein or beads-only control. This indicated that GST-TeoL directly interacts with OMVs prepared from C. necator. Interestingly, OMVs prepared from P. aeruginosa and Y. pseudotuberculosis showed the same binding results with GST-TeoL (Fig. S6A), suggesting that the interaction between TeoL and OMVs is not species-specific.

The interaction between TeoL and OMVs prompted us to further predict that secreted TeoL may associate with OMVs. Indeed, TeoL-VSVG was detected in OMVs purified from ΔteoL mutant expressing the teoL-vsvg fusion protein. Similarly, the OMV marker OmpW tagged with VSVG was also present in OMVs purified from the ΔteoL mutant expressing this fusion protein. By contrast, the VgrG1-VSVG protein, a core component of T6SS1, was not detectably associated with OMVs as predicted (Fig. S6B). These results suggest that TeoL directly associates with OMVs after secretion. We therefore concluded that TeoL contributes to OMV recruitment via direct interaction.

TeoL recruits OMVs through binding LPS

Above results suggest that TeoL targets OMVs for recruitment to the bacterial cell, yet the OMV component that determines TeoL targeting is unknown. Because LPS is the main component of OMVs, we investigated whether LPS was necessary and sufficient to link TeoL with OMVs. As shown in Fig. 4A, LPS immobilized on Sepharose beads efficiently precipitated the GST-TeoL protein but not GST, indicating direct binding between TeoL and LPS. The disassociation constant (K_d) between TeoL and LPS was 0.58 μM (Fig. S7A) as measured using isothermal titration calorimetry (ITC), comparable to that of CD4, a well-known LPS-binding protein [46]. The negative control GST did not bind LPS, as detected under the same binding conditions (Fig. S7A).

LPS is composed of three distinct domains: the lipid A moiety, the core oligosaccharides, and the O-antigen [47]. To determine which part of LPS is required for TeoL binding, the interactions between TeoL and different rough (R) forms of LPS with varying polysaccharide chains (Ra, Rc, and Rd) were investigated. As shown in Fig. S7B, no binding was detected between TeoL and lipid A in ITC analyses, and compared to LPS and lipid A, rough LPS showed an intermediate binding affinity. Moreover, the longest R form tested (Ra) showed the strongest binding affinity (K_d = 14.3 μM) and the shortest R form tested (Rd) showed the weakest binding affinity (K_d = 105.6 μM). These results suggest that the O-antigen region in LPS may directly interact with TeoL.

To further verify the roles of lipid A and O-antigen in OMV recruitment, we produced the P. aeruginosa PAO1 lipid A biosynthesis mutant (ΔmsbB) [48] and O-antigen biosynthesis mutant (Δwzy) [49, 50]. OMVs prepared from PAO1 and ΔmsbB showed efficient TeoL binding while OMVs prepared from the Δwzy mutant failed to interact with TeoL (Fig. 4B). Moreover, the growth of Δ2Fe mutant under iron-depleted medium was efficiently increased by addition of OMVs prepared from WT and ΔmsbB mutant, while adding OMVs prepared from Δwzy mutant showed a very weak effect (Fig. 4C).

Since both OMVs and bacterial outer membranes contain LPSs, our next challenge was to uncover how TeoL distinguishes between LPSs on OMVs and LPSs on bacterial outer membranes. We speculated that TeoL might exhibit higher binding affinity to OMV-associated LPSs, enabling OMV-specific binding. Consistent with our hypothesis, ITC analysis revealed that TeoL exhibited a 4.7-fold higher affinity to LPSs purified from OMVs compared to those purified from bacterial cells (Fig. S8). Taken together, these results indicate that TeoL recognizes LPS, particularly LPS derived from OMVs, through binding to its O-antigen component.

TeoL guides OMV recruitment by binding to outer membrane receptors CubA and CstR

Despite the finding that TeoL recognizes OMVs through LPS, the mechanism of OMV recruitment by the bacterial cell is still unclear. We hypothesized that TeoL may direct OMVs to the bacterial cell surface by interacting with specific outer membrane receptors. To identify possible binding receptors, we performed affinity chromatography with GST-TeoL-coated beads against total cell lysates of C. necator WT. After washing with TEN buffer, proteins retained by GST-TeoL were visualized with silver staining after SDS-PAGE (Fig. 5A). Two specific bands around 80 kDa were identified by mass spectrometric analysis. These were identified as the cupriabactin siderophore receptor CubA (Reut_B3686) [34], and the catecholate siderophore receptor Reut_B4659 (hereafter refer to as CstR). Both CubA and CstR are siderophore-related TonB-dependent outer membrane receptor proteins. The specific interactions between TeoL and CubA or CstR were validated by in vitro binding assays with purified proteins (Fig. 5B). To determine the role of these receptors in iron acquisition, we constructed Δ2FeΔcubAΔcstR (hereafter referred to as Δ2FeΔ2R) mutant in which cubA and cstR were deleted in the background of strain Δ2Fe. While the Δ2FeΔ2R mutant showed severely reduced intracellular iron accumulation in M9 medium, this reduction was rescued by complementation with either cubA or cstR alone, thus confirming their roles in iron acquisition (Fig. 5C).

To further determine the roles of these receptors in OMV recruitment, we compared the growth of the Δ2FeΔ2R mutant with Δ2Fe in an iron-depleted medium supplemented with OMVs prepared from the ΔteoL mutant. The growth of the Δ2FeΔ2R mutant was significantly lower compared to the Δ2Fe mutant, which was completely restored by introducing a plasmid expressing either cubA or cstR (Fig. 5D). Similar results were obtained by adding OMVs purified from P. aeruginosa PAO1 to the iron-depleted medium (Fig. S9A). The role of CubA and CstR receptors in OMV recruitment was further confirmed by observing the direct association between mCherry-labeled bacterial cells and FITC-labeled OMVs purified from the ΔteoL mutant (Fig. S10). These results suggest that CubA and CstR are involved in TeoL-mediated OMV recruitment and iron acquisition.

To garner additional insight into CubA and CstR functions in TeoL-mediated OMV recruitment, we incubated GST-tagged receptors with C. necator ΔteoL OMVs labeled with OmpW-VSVG in the presence or absence of the TeoL protein, respectively. After precipitation with glutathione beads, receptor-OMV complexes retained on the glutathione beads were resolved by SDS-PAGE and detected by immunoblot with an anti-VSVG antibody for detecting the OmpW marker. Although both CubA and CstR specifically bound to OMVs, the binding was strictly dependent on the presence of TeoL (Fig. 5E). Similar results were obtained with OprF-VSVG marked OMVs [51] purified from P. aeruginosa (Fig. S9B). These results suggest that during OMV recruitment, the role of TeoL is to tether iron-containing OMVs to specific receptors on the cell surface.

This conclusion was further supported by directly measuring the binding of GFP-TeoL proteins (preincubated with or without ΔteoL OMVs) to C. necator WT, the ΔcubAΔcstR double mutant, and the ΔcubAΔcstR(cubA) and ΔcubAΔcstR(cstR) complemented strains using a fluorescence spectrometer (Fig. 5F). While the GFP-TeoL protein alone exhibited weak binding affinities to all strains even C. necator WT, preincubation of the GFP-TeoL protein with ΔteoL OMVs greatly improved its affinity to C. necator WT and the ΔcubAΔcstR(cubA) and ΔcubAΔcstR(cstR) complemented strains. However, preincubation with ΔteoL OMVs did not improve the affinity of GFP-TeoL to the ΔcubAΔcstR double mutant. The finding that preincubation with OMVs enhanced the binding affinities of TeoL to bacterial cells further corroborated its role in tethering OMVs to the bacterial cell surface through recognition of the outer membrane CubA/CstR receptors.

We then speculated that secreted TeoL may exhibit a binding preference for OMVs over bacterial cells. To validate this hypothesis, we incubated GFP-TeoL proteins with ΔteoL cells or OMVs containing equal amounts of LPS (30 µg ml⁻¹ of phospholipids), respectively, and the amounts of GFP-TeoL associated with OMVs or bacterial cells were quantified using a fluorescence spectrometer after removing unbound GFP-TeoL proteins in the supernatant with ultracentrifugation. As predicted, GFP-TeoL showed stronger associations with ΔteoL OMVs than ΔteoL cells (Fig. S11), consistent with our finding that TeoL exhibited higher affinities to LPSs purified from OMVs than to those purified from bacterial cells (Fig. S8). Taken together, these results suggest that once secreted, the TeoL effector protein selectively binds to OMVs first, then brings the iron-containing OMVs to the bacterial cell surface by interacting with the CubA/CstR outer membrane receptors.

TeoL-mediated OMV recruitment is crucial for exploitation competition, oxidative stress resistance, and horizontal gene transfer

T6SSs enhance bacterial survival by delivering “anti-bacterial” toxins [52, 53] or by enhancing its ability to acquire essential micronutrients such as manganese and zinc during exploitative competition (such as consuming nutrients from the milieu) [43, 44, 54, 55]. The finding that TeoL/T6SS1 is required for iron acquisition from OMVs suggests that they play a role in mediating exploitation competition. To test this hypothesis, we performed intraspecies growth competition assays between C. necator strains with differed capabilities in TeoL secretion and OMVs recruitment, in M9 medium containing ΔteoL OMVs (20 µg ml⁻¹ of phospholipids). As shown in Fig. 6A, the Δ2Fe strain showed increased growth compared to the Δ2FeΔ2R strain, because although both strains can secrete TeoL, only the Δ2Fe strain can recruit OMVs with CubA/CstR receptors. Δ2FeΔ2R did not show competition advantage over Δ2FeΔteoL and Δ2FeΔtssM1, which cannot secrete TeoL but can recruit OMVs with CubA/CstR receptors. These results suggest that bacteria that possess functional OMV receptors can use TeoL-associated OMVs produced by other bacteria, regardless of their ability to secrete TeoL. Consistent with this conclusion, the Δ2FeΔ2R strain displayed a severe growth disadvantage when competing with not only Δ2Fe, but also with the Δ2FeΔteoL and Δ2FeΔtssM1 strains. One possible explanation is that the Δ2FeΔ2R strain, which cannot recruit OMVs, can still produce TeoL-associated OMVs to support the growth of the Δ2FeΔteoL and Δ2FeΔtssM1 strains. As expected, the Δ2FeΔteoL and Δ2FeΔtssM1 strains displayed no growth advantage over the Δ2FeΔ2RΔteoL (hereafter referred to as Δ5) strain, which cannot produce TeoL-associated OMVs. The role of TeoL/T6SS1 in mediating exploitative competition was further confirmed by interspecies growth competition assays between C. necator strains and Y. pseudotuberculosis. As shown in Fig. 6B, while the C. necator WT showed increased growth compared to Y. pseudotuberculosis in the absence of ΔteoL OMVs (1.8-fold), it was highly competitive against the Y. pseudotuberculosis competitor in the presence of ΔteoL OMVs (2.8-fold). However, the competitive advantage of C. necator WT was largely abolished in ΔteoL and ΔtssM1 mutants, and such deficits could be rescued by complementation with corresponding genes.

Similar to T6SSs reported in Y. pseudotuberculosis [44] and Burkholderia thailandensis [43], the C. necator T6SS1 also contributed to defense against oxidative stress (Fig. S12A). Indeed, deleting teoL alone was sufficient to decrease resistance to H₂O₂ in C. necator (Fig. S12A), suggesting that OMVs recruited by TeoL is important for resistance to oxidative stress. To determine the functions of OMVs in resisting oxidative stress, we used the Δ5 mutant, which has deficits in iron acquisition, OMV recruitment, and TeoL production. The survival rates of Δ5 and its corresponding single gene complemented strains were determined following exposure to H₂O₂ for 25 min, in the absence or presence of OMVs purified from WT, ΔteoL, and ΔteoL(teoL) strains, respectively. While adding all three types of OMVs significantly increased the survival rates of the WT strain, adding these OMVs had no effect on the Δ5 mutant, indicating that the capability to obtain OMVs is crucial for resisting oxidative stress (Fig. S12B). Moreover, adding OMVs purified from WT and ΔteoL(teoL) complementary strains, but not the ΔteoL mutant strain, substantially improved the survival rates of Δ5 complemented with OMV receptor genes cubA or cstR, but not teoL. These results suggest that the presence of TeoL (no matter provided by the bacteria cells themselves or by added OMVs) and one of the receptors allowed the bacteria to obtain OMVs for resisting oxidative stress.

OMVs are also known to be involved in HGT [11]. To determine whether the TeoL/T6SS1-mediated OMV recruitment pathway contributes to HGT, we evaluated plasmid DNA transfer mediated by OMVs. The C. necator ΔteoL mutant harboring pBBR1MCS-2 (Km^R) was grown in NB medium until the stationary phase, and plasmid-containing OMVs were extracted from the supernatant. About 1.87 × 10⁵ copies of pBBR1MCS-2 plasmid were detected to be associated with OMVs in 1 ml of the culture supernatant (7.30 × 10⁵ and 5.43 × 10⁵ copies ml⁻¹ in the supernatant before and after removing OMVs through ultracentrifugation, respectively). After degrading the external DNA surrounding OMVs by DNase I treatment, about 1.22 × 10⁵ copies of pBBR1MCS-2 plasmid in the OMVs from 1 ml culture supernatant (equivalent to 1.48 × 10⁷ copies mg⁻¹ OMV phospholipids) were detected. When relevant C. necator cells (~2.5 × 10³ cells ml⁻¹) were incubated with an excessive amount of OMVs (30 μg ml⁻¹ phospholipids), more than 19.1% of Δ2Fe transformants were obtained on selective plates containing kanamycin after incubation with ΔteoL OMVs for 4 h at 30 °C, suggesting that the plasmid contained in the OMVs was transferred to bacterial cells. However, the Δ2FeΔteoL and Δ2FeΔ2R mutants preincubated with ΔteoL OMVs showed a remarkable decrease in transformation efficiency (3.6% and 2.4%, respectively), and the decreased transformation efficiency could be substantially restored by complementation (Fig. 6C). Notably, natural transformation did not occur in C. necator when naked plasmid DNA (10 ng ml⁻¹) extracted from ΔteoL(pBBR1MCS-2) was directly added to bacterial cell suspension.

Together, these results demonstrate that the TeoL/T6SS1-mediated OMV recruitment pathway is crucial for obtaining cargos loaded in OMVs, thus performing pleiotropic physiological functions.

Source: Ecology - nature.com