Burkholderia multivorans requires species-specific GltJK for entry of a contact-dependent growth inhibition system protein.

Interbacterial antagonism and communication are driving forces behind microbial community development. In many Gram-negative bacteria, contact-dependent growth inhibition (CDI) systems contribute to these microbial interactions. CDI systems deliver the toxic C-terminus of a large surface exposed protein to the cytoplasm of neighboring bacteria upon cell-contact. Termed the BcpA-CT, import of this toxic effector domain is mediated by specific, yet largely unknown receptors on the recipient cell outer and inner membranes. In this study, we demonstrated that cytoplasmic membrane proteins GltJK, components of a predicted ABC-type transporter, are required for entry of CDI system protein BcpA-2 into Burkholderia multivorans recipient cells. Consistent with current CDI models, gltJK were also required for recipient cell susceptibility to a distinct BcpA-CT that shared sequences within the predicted "translocation domain" of BcpA-2. Strikingly, this translocation domain showed low sequence identity to the analogous region of an Escherichia coli GltJK-utilizing CDI system protein. Our results demonstrated that recipient bacteria expressing E. coli gltJK were resistant to BcpA-2-mediated interbacterial antagonism, suggesting that BcpA-2 specifically recognizes Burkholderia GltJK. Using a series of chimeric proteins, the specificity determinant was mapped to Burkholderia-specific sequences at the GltK C-terminus, providing insight into BcpA transport across the recipient cell cytoplasmic membrane.

INTRODUCTION

Contact‐dependent growth inhibition (CDI) systems are Two‐Partner secretion systems that mediate competition and communication among closely related Gram‐negative bacteria. CDI systems are composed of a large stalk‐like exoprotein, termed BcpA in Burkholderia species, that is secreted to the cell surface by an outer membrane β‐barrel protein, BcpB. BcpA proteins have an N‐terminal conserved region approximately 2,700 amino acids (aa) in length that contains multiple filamentous hemagglutinin (FHA) repeats and a C‐terminal variable region (~300 aa) termed the BcpA‐CT (Anderson et al., 2012). This variable region contains an antibacterial toxic effector domain that frequently has nuclease activity (Nikolakakis et al., 2012). Upon cell−contact, the BcpA‐CT is delivered to the cytoplasm of neighboring bacteria where it inhibits growth unless the recipient cells produce a specific immunity protein, BcpI. BcpI proteins are also variable, only binding to and inhibiting the toxicity of their cognate BcpA‐CT proteins.

CDI systems were first described in Escherichia coli (where they are termed CdiBAI), and the pathway for effector entry into recipient cells has been delineated for several CdiA variants in E. coli and related species (Aoki et al., 2005; Ruhe et al., 2017; Willett et al., 2015). The current model suggests that a variable “receptor binding domain” of CdiA, mapped to ~300 aa in the conserved N‐terminal region between the FHA‐1 and FHA‐2 repeats and located distally from the donor cell surface, interacts with a specific recipient cell outer membrane protein (Ruhe et al., 2017, 2018). Upon binding, the CdiA‐CT domain is translocated across the recipient cell outer membrane. Recipient cell inner membrane proteins are required for subsequent CdiA‐CT translocation to the cytoplasm and sequences at the N‐terminal region of the CdiA‐CT (termed the “translocation domain”) dictate which inner membrane protein is required (Willett et al., 2015). Among others, outer membrane proteins BamA, OmpC, and Tsx and inner membrane proteins PtsG, MetI, RbsC, FtsH, and DppBC have been identified as receptors for CdiA variants (Allen et al., 2020; Ruhe et al., 2013, 2017; Willett et al., 2015).

The process of CdiA‐CT entry into recipient cells restricts effector exchange to bacterial species that produce the required receptors. Further specificity has also been observed for the translocation of several CdiA‐CT polypeptides, due to sequence variability in the outer membrane proteins that serve as CdiA receptors. For example, CdiA proteins that utilize BamA as an outer membrane receptor are constrained by strain‐specific differences in the variable extracellular loops of BamA (Ruhe et al., 2013). Thus, although BamA is generally conserved among enterobacteria, these CdiA proteins are only delivered to bacteria that produce the “correct” BamA variant. It is not known whether similar sequence specificity may exist for CdiA inner membrane receptors.

Burkholderia CDI systems are functionally distinct from the CdiBAI proteins found in E. coli and related species. Differences include gene order (cdiBAI for “E. coli‐type,” bcpAIOB for “Burkholderia‐type”), the motif that separates the conserved CdiA/BcpA N‐terminal region from the variable C‐terminal toxin (VENN in E. coli, N(E/Q) × LYN in Burkholderia), the modularity of CdiA/BcpA proteins, and the presence of one or more open reading frames between bcpI and bcpB, termed bcpO genes (Anderson et al., 2012; Nikolakakis et al., 2012). Evidence suggests that the BcpA‐CT recipient cell entry pathway is similar to pathways described for CdiA‐CT proteins, but only one BcpA receptor, an inner membrane protein, has been conclusively identified (Koskiniemi et al., 2015; Willett et al., 2015).

We recently described the two CDI systems of Burkholderia cepacia complex (Bcc) strain Burkholderia multivorans CGD2M (Myers‐Morales et al., 2019). Like other Burkholderia species, Bcc organisms are commonly found in soil and other natural environments (Vial et al., 2010). They are also opportunistic pathogens that can cause chronic pulmonary infections in patients with cystic fibrosis, chronic granulomatous disease, or who are immunocompromised. Here we used a transposon mutagenesis approach to define factors in B. multivorans recipient cells that are required for intoxication by BcpA and we identified GltJ/GltK as the likely inner membrane receptors for BcpA‐2. Strikingly, although B. multivorans and E. coli GltJK share >60% sequence identity, only the Burkholderia homologs conferred susceptibility to BcpA‐2. Using a series of chimeric proteins, this specificity was mapped to the C‐terminus of GltK. Containing predicted periplasmic, transmembrane, and cytoplasmic sequences, our results support a model in which this GltK region is critical for translocation of BcpA‐2‐CT across the recipient cell cytoplasmic membrane.

RESULTS

Transposon insertions in gltJK confer resistance to BcpA‐2

B. multivorans CGD2M encodes two distinct functional CDI systems, but only CDI system‐2 mediates interbacterial competition in laboratory medium under conditions of native gene expression (Myers‐Morales et al., 2019). To determine factors needed for translocation and toxicity of BcpA‐2 in B. multivorans recipient cells, we used a sequential selection approach similar to approaches used to analyze E. coli CdiBAI (Aoki et al., 2008; Willett et al., 2015). The CDI‐sensitive ΔbcpAIB‐2 mutant was mutagenized with miniTn5 and the resulting mutant library competed against wild‐type B. multivorans in an interbacterial competition assay (Figure 1a and Table S1). Transposon mutants surviving the competition were pooled and re‐competed against wild‐type bacteria to enrich for insertions that conferred resistance to BcpA‐2. This process was repeated until, after four sequential rounds of competition, CDI susceptibility of the enriched transposon mutant pool was not significantly different from that of recipient cells expressing cognate bcpI‐2. CDI resistance was validated for 16 randomly chosen clones by testing them in individual competition assays against wild‐type bacteria (Figure 1b). Using nested arbitrary primed PCR, transposon insertion sites were identified for 12 of these mutants. All 12 mutants had insertions in gltJ (3 unique insertions) or gltK (5 unique insertions; Table S2). Predicted to encode inner membrane components of an ABC‐type transporter for glutamate and aspartate, gltJK are found in a locus that also includes gltL, encoding a predicted ATPase, and gltI, encoding a predicted periplasmic binding protein (Figure 1c).

FIGURE 1

Transposon mutagenesis and selection for contact‐dependent growth inhibition (CDI)‐resistant Burkholderia multivorans mutants. (a) B. multivorans CGD2M Δbcp‐2 recipient bacteria were mutagenized with miniTn5, pooled, and subjected to sequential rounds of co‐culture with wild‐type bacteria on agar. Competitive index (CI) values are shown for competition mixtures collected at 24 hr for CmR wild‐type donor bacteria co‐cultured with KanR Δbcp‐2 recipient bacteria (none; open symbols), Δbcp‐2 recipients complemented with cognate bcpI‐2 (black symbols), and the Δbcp‐2 transposon (Tn) mutant pool (gray symbols). Output CFUs for the Δbcp‐2 transposon mutant pool were collected after each competitive round, pooled, and used as the input for the next round. The CI was calculated as (output donor CFU/recipient CFU) divided by (input donor CFU/recipient CFU). Symbols represent CI values from one biological replicate and bars show the mean (n = 3). Dashed line shows CI = 1 (1:1 ratio of donor:recipient). CI values for the Tn mutant pool were compared to those of immune recipient bacteria (+bcpI‐2). Statistical differences were determined with two‐tailed Student’s t‐test. **p < .01. ns, not significant. (b) Competitive indices for wild‐type donor bacteria co‐cultured with Δbcp‐2 recipient bacteria (none; open symbols), Δbcp‐2 recipients complemented with cognate bcpI‐2, and individual Δbcp‐2 transposon mutants. Individual mutants were tested in two batches (left and right panels). Letters (a‒h) indicate the location of the transposon insertion for each mutant in panel C locus diagram. (c) Diagram showing the genetic organization B. multivorans CGD2M gltIJKL locus. Arrows indicate the approximate locations of unique transposon insertions conferring resistance to BcpA‐2‐mediated CDI (letters correspond to Tn mutants in b)

GltJK are required for recipient cell sensitivity to BcpA‐2

To test the role of gltJK in B. multivorans CDI, we constructed a ΔgltJK deletion mutation in a Δbcp‐1Δbcp‐2 double mutant that lacks both of its CDI systems. This quadruple mutant was resistant to interbacterial competition with wild‐type bacteria and its sensitivity was restored by complementation of gltJK to a chromosomal site (Figure 2a). During culture in rich medium, the second B. multivorans CDI system, BcpAIOB‐1, mediates a lower level of interbacterial competition than BcpAIB‐2 and overexpression of bcpAIOB‐1 is necessary to observe its effects in vitro (Myers‐Morales et al., 2019). To determine whether gltJK are also required for BcpA‐1 susceptibility, the quadruple ΔgltJK Δbcp‐1Δbcp‐2 mutant was competed against a Δbcp‐2 mutant that overexpressed bcpAIOB‐1. No significant difference in competitive index was seen between recipient cells with and without gltJK (Figure 2c), indicating that GltJK are not required for BcpA‐1 sensitivity and are specifically required for B. multivorans BcpA‐2. To determine whether loss of gltJK in donor bacteria alters CDI, we constructed a ΔgltJK mutant in wild‐type B. multivorans. Loss of gltJK had no effect on the ability of this mutant to inhibit the growth of recipient cells (Figure 2d), indicating that gltJK are not required in donor bacteria for efficient CDI.

FIGURE 2

Roles of gltJ, gltK and gltI in recipient cell susceptibility to Burkholderia multivorans contact‐dependent growth inhibition. Competitive indices (CI) for bacteria sampled from the edge of a colony biofilm after 24 hr co‐culture on agar are shown. CI was calculated as (output donor CFU/recipient CFU) divided by (input donor CFU/recipient CFU). Symbols represent CI from one biological replicate, and bars show the mean (n = 6 from two independent experiments). Dashed line shows CI = 1 (no competition). Statistical differences were determined with two‐tailed Student’s t‐test unless otherwise indicated. *p < .05; **p < .01; compared to corresponding recipient cells with no bcpI. (a) Wild‐type donor bacteria were competed against Δbcp‐1 Δbcp‐2 recipient cells (open circles), Δbcp‐1 Δbcp‐2 recipients carrying bcpI‐2 at an attTn7 site (black circles), and Δbcp‐1 Δbcp‐2 ΔgltJK recipient cells with (gray squares) or without (white squares) gltJK at an attTn7 site. Statistical differences were determined with the Mann−Whitney test. (b) Wild‐type donor bacteria were competed against mutants in a) and bcp‐1 Δbcp‐2 ΔgltI recipient cells with (gray triangles) or without (white triangles) gltI at an attTn7 site. Statistical differences were determined with the Mann−Whitney test. (c) B. multivorans Δbcp‐2 donor bacteria overexpressing bcpAIOB‐1 (from strong, constitutive promoter PS12) were competed against Δbcp‐1 Δbcp‐2 recipient cells (open circles), Δbcp‐1 Δbcp‐2 recipients carrying bcpI‐1 at an attTn7 site (black circles), and Δbcp‐1 Δbcp‐2 ΔgltJK recipient cells with (gray squares) or without (white squares) gltJK at an attTn7 site. (d) Wild‐type (left panel) or ΔgltJK (right panel) donor bacteria were competed against Δbcp‐1 Δbcp‐2 recipient cells with (black circles) or without (white circles) cognate bcpI‐2 at an attTn7 site

ABC‐type transporters like GltJK have been previously found to mediate translocation of CdiA‐CT effectors into recipient cells (Willett et al., 2015). The ATPase components of these transporters, including GltL, were not required for CdiA translocation, but the periplasmic components of the transporters have not been investigated in the context of CDI (Willett et al., 2015). To determine whether the predicted periplasmic binding protein GltI influences BcpA‐2 sensitivity, a ΔgltI mutant was constructed in a Δbcp‐1Δbcp‐2 double mutant. This mutant showed no difference in interbacterial competition against wild‐type B. multivorans (Figure 2b), indicating that gltI does not contribute to CDI susceptibility. Although the ATPase component (GltL) was not examined here, these data are consistent with the model that recipient cell sensitivity to BcpA‐2 requires the membrane‐localized components of the GltIJKL ABC‐type transporter.

GltJK are not required for BcpA‐2 intracellular toxicity

Predicted to be located in the inner membrane, GltJK are hypothesized to impact BcpA‐2 translocation to the recipient cell cytoplasm. To test this hypothesis, we examined the intracellular toxicity of BcpA‐2‐CT independently of effector translocation. A plasmid containing bcpA‐2‐CT under the control of a rhamnose‐inducible promoter was constructed and introduced into Δbcp‐1Δbcp‐2 and Δbcp‐1Δbcp‐2 ΔgltJK recipient cells by conjugation. Upon selection on medium containing glucose (allowing only basal bcpA‐2‐CT expression), B. multivorans transconjugants could not be obtained for either gltJK+ or gltJK− recipient bacteria, unless bacteria were complemented with bcpI‐2 (Figure 3). Thus, when BcpA‐2 entry into recipient cells was bypassed by producing BcpA‐2‐CT intracellularly, cells with and without GltJK were similarly susceptible to toxicity. These results indicate that GltJK are not required for BcpA‐2 intracellular toxicity and instead suggest that GltJK are needed for BcpA‐2 translocation into recipient cells, likely across the cytoplasmic membrane.

FIGURE 3

Role of gltJK in BcpA‐2‐CT intracellular toxicity. Transconjugants obtained after introduction of pSCrhaB3 empty vector (top) or plasmid carrying bcpA‐2‐CT (bottom) into Δbcp‐1 Δbcp‐2 mutant bacteria (left two columns) or Δbcp‐1 Δbcp‐2 ΔgltJK mutant bacteria (right two columns) that constitutively express cognate bcpI‐2 from an attTn7 site (first and third columns) or lack immunity (second and fourth columns)

E. coli gltJK cannot complement B. multivorans ΔgltJK

Previous research has already identified E. coli GltJK as a probable cytoplasmic membrane translocation factor for CdiA‐CT from Photorhabdus luminescens strain TTO1 (Willett et al., 2015). CdiA‐CT proteins that require the same translocation factor have been shown to share a high degree of sequence identity within their N‐terminal regions, termed the “translocation domain” (Willett et al., 2015). Unexpectedly, comparison of the GltJK‐utilizing BcpA‐2‐CTCGD2M and CdiA‐CTTTO1 showed only 24% sequence identity within the N‐terminal halves of the polypeptides (Figure S1). By contrast, GltJ and GltK are more similar between B. multivorans and E. coli, sharing 63% and 71% amino acid identity, respectively (78% and 84% similarity, respectively). To explore the impact of these differences on BcpA‐2 uptake, E. coli DH5α gltJK (gltJK) were expressed in the Δbcp‐1Δbcp‐2 ΔgltJK quadruple mutant. In interbacterial competition assays, E. coli gltJK failed to restore CDI sensitivity of this mutant (Figure 4a). In the event that the E. coli GltJK proteins require the cognate E. coli GltI and GltL proteins for optimal function or stability, we also attempted to complement the mutant with the complete gltIJKL locus from E. coli (gltIJKL). Expression of these genes was also unable to restore BcpA‐2 sensitivity to recipient cells lacking gltJK (Figure 4a). Burkholderia thailandensis E264 GltJK are very similar to their B. multivorans CGD2M counterparts (98% and 96% identical, respectively) and, as expected, the genes encoding these proteins (gltJK) also restored susceptibility of the B. multivorans recipient cells to BcpA‐2 toxicity (Figure 4b). These results suggest that CGD2M BcpA‐2 requires Burkholderia GltJK for translocation and led us to hypothesize that specific differences in the E. coli proteins block entry of BcpA‐2 into the recipient cell.

FIGURE 4

Ability of heterologous gltJK alleles to sensitize Burkholderia multivorans recipient cells to contact‐dependent growth inhibition mediated by BcpA‐2 and a related BcpA protein. Competitive indices (CI) for bacteria sampled from the edge of a colony biofilm after 24 hr co‐culture on agar are shown. CI was calculated as (output donor CFU/recipient CFU) divided by (input donor CFU/recipient CFU). Symbols represent CI from one biological replicate and bars show the mean (n = 6 from two independent experiments). Dashed line shows CI = 1 (no competition). Statistical differences were determined with Mann−Whitney test unless otherwise indicated. **p < .01; compared to corresponding recipient cells with no bcpI. (a) Wild‐type donor bacteria were competed against Δbcp‐1 Δbcp‐2 recipient cells (open circles), Δbcp‐1 Δbcp‐2 recipients carrying bcpI‐2 at an attTn7 site (black circles), and Δbcp‐1 Δbcp‐2 ΔgltJK recipient cells complemented at an attTn7 site with empty vector (none, white squares), or PS12‐driven B. multivorans gltJK (gray squares), Escherichia coli gltJK (dark gray squares) or E. coli gltIJKL (dark gray squares). (b) Wild‐type donor bacteria were competed against Δbcp‐1 Δbcp‐2 recipient cells (open circles), Δbcp‐1 Δbcp‐2 recipients carrying bcpI‐2 at an attTn7 site (black circles), and Δbcp‐1 Δbcp‐2 ΔgltJK recipient cells complemented at an attTn7 site with empty vector (none, white squares) or PS12‐driven B. multivorans gltJK (gray squares) or Burkholderia thailandensis gltJK (dark gray squares). Statistical differences were determined with two‐tailed Student’s t‐test. (c) Graphical representations of the BcpA‐CT domains and BcpI proteins from B. multivorans CGD2M (top), Burkholderia cenocepacia Bp8968 (middle) and chimeric strain constructed (CGD2M‐30, bottom). Arrow indicates the location of the VENN motif and theoretical N‐terminal end of the BcpA‐CT region. Approximate locations of the putative translocation domain (TD) and effector domain (ED) of the BcpA‐CT are shown. (d) Chimeric B. multivorans donor bacteria (CGD2M‐30) producing chimeric BcpA‐2 containing the BcpA‐CT from B. cenocepacia Bp8968 (along with its cognate BcpI and an additional hypothetical ORF) was competed against Δbcp‐1 Δbcp‐2 recipient cells (open circles), Δbcp‐1 Δbcp‐2 recipients carrying B. cenocepacia Bp8968 bcpI at an attTn7 site (gray circles), and Δbcp‐1 Δbcp‐2 ΔgltJK recipient cells complemented at an attTn7 site with empty vector (none, white squares), or PS12‐driven B. multivorans gltJK (gray squares) or E. coli gltJK (dark gray squares)

Related BcpA‐CT shows specificity for Burkholderia GltJK

Bioinformatic analyses of other Burkholderia CDI systems and comparison of the predicted BcpA‐CT translocation domains led to the identification of a putative BcpA from Burkholderia cenocepacia Bp8968. BcpA‐CTBp8968 and BcpA‐2‐CTCGD2M share 91% identity across their N‐terminal regions (~150 residues), but only 14% identity over the C‐terminal halves of BcpA‐CT (Figure S1). The large, conserved N‐termini (~3,000 residues) of the two BcpA proteins similarly share high sequence identity (93%). To test the hypothesis that BcpA‐CTBp8968 would also utilize GltJK for entry into recipient cells, we constructed a chimeric mutant by replacing the chromosomal B. multivorans bcpA‐2‐CT and bcpI with B. cenocepacia Bp8968 bcpA‐CT and bcpI (Figure 4c). An additional predicted B. cenocepacia Bp8968 ORF (DF018_16545) located between bcpI and bcpB was also included in the chimeric mutant. This mutant (named CGD2M‐30), expected to produce chimeric BcpA along with cognate BcpI, was used as a donor in competition assays. CGD2M‐30 outcompeted B. multivorans Δbcp‐1Δbcp‐2 recipient cells (Figure 4d) and Δbcp‐2 recipient cells (Figure S2a) by ~1.5 logs unless they produced cognate B. cenocepacia bcpI, indicating that the chimeric BcpA was functional. As expected, this mutant was unable to inhibit the growth of recipient cells lacking gltJK or expressing E. coli gltJK (Figures 4d and S2a), supporting the results obtained with native BcpA‐2. Although we cannot rule out contributions of the conserved BcpA N‐terminus (~3,000 aa), these results are consistent with previous data suggesting that the BcpA‐CT N‐terminal region functions in toxin translocation. These findings are also consistent with our hypothesis that the translocation domain found in BcpA‐2‐CTCGD2M and BcpA‐CTBp8968 is specific for Burkholderia GltJK.

BcpACGD2M‐CdiATTO1 chimera is not functional

Unlike E. coli CdiA proteins, previous data suggest that the N‐terminal “conserved” region of Burkholderia BcpA proteins is only capable of delivering certain, closely related BcpA‐CT toxin cargo (from BcpA proteins of the same subclass) (Anderson et al., 2014; Nikolakakis et al., 2012). Chimeric proteins consisting of “mismatched” BcpA N‐ and C‐terminal regions are nonfunctional for CDI (Anderson et al., 2014). Nonetheless, to further test our hypothesis, we attempted to switch the GltJK specificity of B. multivorans BcpA‐2 by generating a chimeric BcpA protein composed of the BcpA‐2 N‐terminus (~3,000 aa) and the P. luminescens TTO1 CdiA‐CT, joined at the VENN motif (Figure S2b). A Burkholderia mutant producing this chimeric protein and its cognate immunity protein, CdiITTO1, did not inhibit the growth of B. multivorans recipient cells expressing either B. multivorans or E. coli gltJK (Figure S2). These results suggest that E. coli GltJK are insufficient for CdiA‐CTTTO1 translocation/toxicity in B. multivorans or, more likely, that this chimeric BcpACGD2M‐CdiATTO1 protein is not functional. While we cannot rule out the possibility here that E. coli GltJK are improperly assembled in B. multivorans and not competent for CdiA‐CTTTO1 translocation, previous evidence indicates that Burkholderia BcpA proteins are not as modular as their CdiA counterparts, which can complicate generation of functional chimeras (Anderson et al., 2014).

Mixed GltJK complexes suggest GltK drives translocation specificity

To determine how each inner membrane protein contributes to BcpA‐2 translocation, we generated plasmids to deliver individual gltJ and gltK genes to each of the B. multivorans two attTn7 sites. Neither gltJ nor gltK expressed alone was sufficient to restore CDI susceptibility to recipient cells lacking gltJK (Figure 5a), suggesting that a membrane complex composed of both GltJ and GltK are needed for BcpA‐CT translocation and/or complex stability. Growth inhibition also did not occur when recipient cells expressed E. coli gltK paired with gltJ from either E. coli or B. multivorans (Figure 5b). Surprisingly, E. coli gltJ expressed with B. multivorans gltK conferred partial susceptibility to BcpA‐2, although the level of growth inhibition was significantly less than when both gltJ and gltK were from B. multivorans (Figure 5b). Overexpression of gltJK in recipient cells that expressed native gltIJKL did not enhance CDI susceptibility (Figures 5c and S4a). Together, these data indicate that both GtlJ and GltK are required for BcpA‐2 import into recipient cells, but suggest that GltK confers most of the species specificity to translocation.

FIGURE 5

Individual contributions of GltJ and GltK to sensitivity to BcpA‐2‐mediated contact‐dependent growth inhibition. (a) Wild‐type donor bacteria were competed against Δbcp‐1 Δbcp‐2 recipient cells (open circles), Δbcp‐1 Δbcp‐2 recipients carrying bcpI‐2 at an attTn7 site (black circles), and Δbcp‐1 Δbcp‐2 ΔgltJK recipient cells complemented at an attTn7 site with empty vector (none, white squares) or PS12‐driven Burkholderia multivorans gltJ or gltK (gray squares). Competitive indices (CI) for bacteria sampled from the edge of a colony biofilm after 24 hr co‐culture on agar are shown. CI was calculated as (output donor CFU/recipient CFU) divided by (input donor CFU/recipient CFU). Symbols represent CI from one biological replicate and bars show the mean (n = 6 from two independent experiments). Dashed line shows CI = 1 (no competition). Statistical differences were determined with two‐tailed Student’s t‐test. **p < .01; compared to corresponding recipient cells with no bcpI unless otherwise indicated. (b) Competition assays as described in a. Wild‐type donor bacteria were also competed against Δbcp‐1 Δbcp‐2 ΔgltJK recipient cells complemented at each of two attTn7 sites with combinations of PS12‐driven gltJ or gltK from Escherichia coli or B. multivorans. (c) Competition assays as described in a. Wild‐type donor bacteria were also competed against Δbcp‐1 Δbcp‐2 recipient cells complemented at each of two attTn7 sites with B. multivorans gltJ and gltK (gray circles). (d) Competition assays as described in a. Wild‐type donor bacteria were also competed against Δbcp‐1 Δbcp‐2 ΔgltJK recipient cells complemented at each of two attTn7 sites with empty vectors (none, white squares) or B. multivorans gltJ and untagged gltK or FLAG‐tagged gltK (gray squares). (e) Western blots of subcellular fractions of Δbcp‐1 Δbcp‐2 ΔgltJK bacteria complemented at each of two attTn7 sites with combinations of gltJ and FLAG‐tagged gltK from B. multivorans and E. coli (or untagged gltK; first lanes of each fraction). Equal amounts of each fraction (cytoplasmic, outer membrane [Sarkosyl‐soluble], and inner membrane [Sarkosyl‐insoluble]) were resolved on 12% sodium dodecyl sulfate‐polyacrylamide gel electrophoresis gels and blots probed with anti‐ FLAG peptide (bottom), anti‐E. coli RNA Polymerase β subunit (RpoB; top), and anti‐Salmonella typhii Outer Membrane protein A (OmpA, middle) antibodies. Expected masses for GltK, OmpA and RpoB are ~26, ~35, and 150 kDa, respectively

It is possible that E. coli GltJK produced heterologously in B. multivorans results in unstable, improperly folded, or incorrectly localized proteins. To determine whether the observed differences in CDI susceptibility were due to improper localization of E. coli GltK when produced in B. multivorans, we generated epitope‐tagged versions of each GltK that carried an in‐frame FLAG tag at its C‐terminus. The FLAG epitope tag did not appear to disrupt the function of B. multivorans GltK, as there was no difference in CDI susceptibility between recipient cells producing tagged or untagged B. multivorans GltK (Figure 5d).

Localization of GltK‐FLAG in these recipient cells was determined by subcellular fractionation using Sarkosyl solubility methods (Filip et al., 1973; Myers‐Morales et al., 2007). As controls, fractions were probed with anti‐E. coli RpoB antibodies, which recognize a cytoplasmic B. multivorans protein that is a similar size to the E. coli 153 kDa RpoB (Figure S3). Fractions were also probed with anti‐Salmonella OmpA antibodies, which recognize an outer membrane B. multivorans protein that is a similar size to the 35 kDa E. coli OmpA (Figure S3). Bands reacting with anti‐RpoB were observed primarily in the cytoplasmic fraction, with very faint bands observed in the membrane fractions (Figure 5e), suggesting little cytoplasmic contamination of the membrane fractions. Similarly, anti‐OmpA‐reactive bands were localized to the outer membrane (Sarkosyl‐insoluble) fraction, with a small amount of putative OmpA protein observed in the inner membrane (Sarkosyl‐soluble) fractions.

For all mutants, GltK ‐FLAG and GltK ‐FLAG were detected in the inner membrane‐enriched fractions, with fainter bands detectable in the cytoplasmic fractions, suggesting that all GltK proteins, regardless of their species or GltJ partner, were localized to the B. multivorans inner membrane (Figures 5e and S5). Even in the absence of GltJ, GltK ‐FLAG was detectable in the inner membrane fraction (Figure S4b,c). Both GltK ‐FLAG and GltK ‐FLAG ran at a slightly smaller molecular weight than predicted. Slight differences in GltK‐FLAG levels were detectable, with GltK ‐FLAG appearing to be the most abundant when produced with its cognate GltJ protein. Inadequate levels of GltK could affect BcpA‐2‐CT translocation in a non‐specific manner. However, GltK‐FLAG abundance in the inner membrane (Figure 5e) did not appear to correlate with BcpA‐2 susceptibility (Figure 5b). These results indicate that both GltK and GltK localize to the B. multivorans inner membrane fraction and suggest that these proteins differ in their ability to mediate BcpA‐2‐CT translocation.

Multiple regions of GltJ are required for BcpA‐2 translocation specificity

We hypothesized that the functional difference between GltJK and GltJK for B. multivorans CDI could be used as a tool to define the region(s) of the transporters required for BcpA import. We predicted that interaction of incoming BcpA‐2‐CT polypeptides would occur with the periplasmic region(s) of the GltJK channel. Prediction of transmembrane helices and protein topology indicated that several regions containing sequence divergence between GltK and GltK and, to a lesser extent between the GltJ proteins, occurred within predicted periplasmic regions (Figures 6a and 7a). To test this hypothesis and map the region(s) required for translocation specificity, we constructed a series of chimeric GltJ and GltK proteins that contained heterologous sequence within one or several predicted periplasmic regions (Figures 6b and 7b).

FIGURE 6

Contribution of GltJ to the specificity of BcpA‐2 translocation. (a) Amino acid alignment of GltJ from Burkholderia multivorans CGD2M and Escherichia coli MG1655. Yellow lines indicate the predicted periplasmic regions and black boxes show the regions changed in the constructed chimeras. (b) Graphic representation of the GltJ proteins from B. multivorans and E. coli and the chimeras constructed with indicated residue numbers. Percent identity of each protein/chimera to B. multivorans CGD2M GltJ is shown to the right. (c) Wild‐type donor bacteria were competed against Δbcp‐1 Δbcp‐2 recipient cells (open circles), Δbcp‐1 Δbcp‐2 recipients carrying bcpI‐2 at an attTn7 site (black circles), and Δbcp‐1 Δbcp‐2 ΔgltJK recipient cells complemented at each of two attTn7 sites with empty vectors (none, white squares) or PS12‐driven B. multivorans gltK and the indicated gltJ allele: B. multivorans gltJ (gray squares), E. coli gltJ (dark gray squares), or genes to produce chimeric GltJ proteins gltJ (red squares) or gltJ (blue squares). Competitive indices (CI) for bacteria sampled from the edge of a colony biofilm after 24 hr co‐culture on agar are shown. CI was calculated as (output donor CFU/recipient CFU) divided by (input donor CFU/recipient CFU). Symbols represent CI from one biological replicate and bars show the mean (n = 6 from two independent experiments). Dashed line shows CI = 1 (no competition). Statistical differences were determined with two‐tailed Student’s t‐test. *p < .05; **p < .01; compared as shown

FIGURE 7

Contribution of GltK to the specificity of the BcpA2 translocation. (a) Amino acid alignment of GltK from Burkholderia multivorans CGD2M and Escherichia coli MG1655. Yellow lines indicate the predicted periplasmic regions and black boxes show the regions changed in the constructed chimeras. (b) Graphic representation of the GltK proteins from B. multivorans and E. coli and the chimeras constructed with indicated residue numbers. Percent identity of each protein/chimera to B. multivorans CGD2M GltK is shown to the right. (c) Wild‐type donor bacteria were competed against Δbcp‐1 Δbcp‐2 recipient cells (open circles), Δbcp‐1 Δbcp‐2 recipients carrying bcpI‐2 at an attTn7 site (black circles), and Δbcp‐1 Δbcp‐2 ΔgltJK recipient cells complemented at each of two attTn7 sites with empty vectors (none, white squares) or PS12‐driven B. multivorans gltJ and the indicated gltK allele: B. multivorans gltK (gray squares), E. coli gltK (dark gray squares), or genes to produce chimeric GltK proteins gltK (red squares), gltK (blue squares) gltK (green squares), gltK (orange squares), gltK (purple squares), or gltK (yellow squares). Competitive indices (CI) for bacteria sampled from the edge of a colony biofilm after 24 hr co‐culture on agar are shown. CI was calculated as (output donor CFU/recipient CFU) divided by (input donor CFU/recipient CFU). Symbols represent CI from one biological replicate and bars show the mean (n = 6 for two independent experiments). Dashed line shows CI = 1 (no competition). Statistical differences were determined with two‐tailed Student’s t‐test. **p < .01; compared to corresponding recipient cells with no bcpI. (d) Western blots of subcellular fractions of Δbcp‐1 Δbcp‐2 ΔgltJK bacteria complemented at each of two attTn7 sites with PS12‐driven B. multivorans gltJ and the indicated gltK allele: B. multivorans gltK (untagged), or genes to produce chimeric GltK‐FLAG proteins gltK‐FLAG and gltK‐FLAG. Equal amounts of each fraction (cytoplasmic, outer membrane [Sarkosyl‐soluble], and inner membrane [Sarkosyl‐insoluble]) were resolved on 12% sodium dodecyl sulfate‐polyacrylamide gel electrophoresis (SDS‐PAGE) gels and blots probed with anti‐ FLAG peptide (bottom), anti‐E. coli RNA Polymerase β subunit (RpoB; top), and anti‐Salmonella typhii Outer Membrane protein A (OmpA, middle) antibodies. Expected masses for GltK, OmpA and RpoB are ~26, ~35 and 150, respectively. (e) Equal amounts of the inner membrane fraction samples shown in d) (derived from Δbcp‐1 Δbcp‐2 ΔgltJK mutant bacteria carrying the indicated gltJ and gltK alleles) were resolved on 12% SDS‐PAGE and visualized by silver staining. Boxed region indicates region shown in in FLAG Western Blot in d)

Chimeric gltJ genes were used to complement Δbcp‐1Δbcp‐2 ΔgltJK mutant recipient bacteria that expressed B. multivorans gltK. Replacement of two predicted periplasmic regions (residues 1–19 and 95–116) of GltJ with E. coli sequences decreased the ability of the resulting chimeric protein, GltJChim1, to confer CDI susceptibility to recipient cells (Figure 6c), suggesting that these regions are important for the ability of GltJ to mediate BcpA‐2 translocation. However, the chimeric GltJChim1 retained significantly more function than GltJ , indicating that other sequences within B. multivorans GltJ are also required for BcpA‐2 transport. Similarly, replacement of the predicted periplasmic N‐terminus of GltJ (residues 1–24) with B. multivorans sequence resulted in a chimeric protein, GltJChim2, that increased CDI susceptibility as compared to recipient cells producing GltJ , although not to the same level as the complete B. multivorans GltJ. Together, these data suggest that several B. multivorans‐specific putative periplasmic regions of GltJ (Figure 8a) contribute to entry of BcpA‐2 into recipient cells, but other regions of the protein are also necessary for this function.

FIGURE 8

Schematic of predicted GltJ and GltK membrane topologies and contribution to BcpA‐CT‐2 translocation specificity. (a) Predicted GltJ membrane topology, showing the two putative periplasmic region(s) altered in the chimeric GltJ proteins (red). (b) Predicted GltK membrane topology, highlighting the three region(s) altered in chimeric GltK proteins. Blue shading denotes regions that did not contribute significantly to BcpA‐CT‐2 translocation specificity and orange shading shows the region that required Burkholderia sequence for BcpA‐CT‐2 translocation. Residue numbers for predicted transmembrane helices are indicated

C‐terminus of B. multivorans GltK is sufficient for BcpA‐2 translocation specificity

To determine the contribution of GltK regions to the species‐specificity of BcpA‐2 translocation, chimeric GltK proteins (Figure 7b) were produced in Δbcp‐1Δbcp‐2 ΔgltJK mutant recipient bacteria that expressed B. multivorans gltJ (Figure 7c). Unlike E. coli GltJ, which was partially functional for BcpA‐2 translocation, E. coli GltK did not confer any CDI susceptibility to recipient cells (Figures 5b and 7c). Similarly, replacement of three regions of B. multivorans GltK that contain predicted periplasmic domains (residues 1–29, 81–98, and 186–225) with E. coli sequences rendered the resulting chimeric protein, GltKChim1, nonfunctional for BcpA‐2 translocation (Figure 7c). By contrast, swapping these same regions of E. coli GltK with the corresponding B. multivorans sequences, resulting in the chimeric protein GltKChim2, restored the ability of this protein to mediate CDI in recipient cells.

Based on this “repaired” GltKChim2 protein, we constructed subsequent chimeras in the E. coli GltK backbone that contained two of the three B. multivorans sequence regions to test the contribution of each region to BcpA‐2 translocation. Of these, GltKChim3, containing the N‐terminal and center B. multivorans regions (Figure 8b), was nonfunctional in recipient cells. GltKChim4, containing the center and C‐terminal regions, and GltKChim5, containing the N‐ and C‐terminal regions, both enabled CDI in recipient cells to levels that did not differ from the CDI susceptibility of bacteria producing full‐length B. multivorans GltK. These results indicate that chimeric GltK proteins containing B. multivorans sequence at their C‐termini (GltKChim2, GltKChim4, and GltKChim5) were able to mediate BcpA‐2 translocation, while proteins containing E. coli sequence within this region (GltKChim1 and GltKChim3) could not.

To test the hypothesis that the GltK C‐terminus mediates the specificity of BcpA‐2 entry, we expressed gltK in recipient cells, producing a protein composed of the E. coli N‐terminus (residues 1–185) and B. multivorans C‐terminus (residues 186–225; Figures 7b and 8b). Recipient cells producing GltKChim6 were strongly outcompeted by B. multivorans donor cells (Figure 7c), indicating that the chimeric protein was functional for BcpA‐2 translocation and suggesting that sequences in the GltK C‐terminus are responsible for the specificity of BcpA‐2 for Burkholderia GltK.

To determine whether the functional differences among the chimeric GltK proteins were due to defects in abundance or localization, we generated FLAG‐tagged versions of one functional (GltKChim6) and one nonfunctional (GltKChim3) chimera. Similar to the full‐length B. multivorans and E. coli GltK‐FLAG proteins, GltKChim3‐FLAG and GltKChim6‐FLAG were detected predominantly in the inner membrane fractions (Figure 7d,e). Although we cannot rule out the possibility that chimeric GltK are improperly folded in the B. multivorans membrane, these data suggest that GltK abundance or localization do not account for the differences in CDI susceptibility mediated by GltKChim3 and GltKChim6.

The B. multivorans and E. coli GltK proteins share 56% identity within the C‐terminal region included in GltKChim6. Alignment of additional GltK proteins shows that this C‐terminal region is highly conserved among Burkholderia species, diverging slightly in the related species Ralstonia solanacearum and more significantly in E. coli and other γ‐proteobacteria (Figure S6). Together, these results support the model that Burkholderia‐specific sequences within the GltK C‐terminus (Figure 8b) are needed for translocation of BcpA‐2‐CT to the recipient cell cytoplasm.

DISCUSSION

In this study, we demonstrated that inner membrane proteins GltJK are required for entry of CDI system protein BcpA‐2 into B. multivorans recipient cells. Constituting the permease component of a predicted ABC‐type transporter, both GltJ and GltK were required for BcpA‐2 import. Although E. coli GltJ and GltK share significant sequence identity with their B. multivorans homologs, expression of E. coli gltJK did not confer BcpA‐2 susceptibility to recipient cells. These results suggest that BcpA‐2 requires Burkholderia‐specific sequences within GltJK for translocation and provided a useful tool to begin probing the mechanism of BcpA cytoplasmic entry. Using chimeric proteins, the specificity determinant was mapped to the C‐terminus of GltK. The presence of B. multivorans sequences at the GltK C‐terminus was both necessary and sufficient to render recipient cells sensitive to BcpA‐2 toxicity during interbacterial competition, suggesting that BcpA‐2‐CT may interact with this region.

Studies examining the structure and function of ABC‐type transporters offer clues toward possible roles for the GltK C‐terminal region in BcpA‐2 import. Topology models (UniProt) predict that this region contains a periplasmic region (residues 176–196), a transmembrane helix (residues 197–217), and a cytoplasmic region (residues 218–224; UniProt Consortium, 2019). Consistent with these models, studies using E. coli GltK‐PhoA and ‐GFP fusions demonstrated that the GltK C‐terminus is located in the cytoplasm (Daley et al., 2005). This region is also distinct from sequences shown in other ABC transporters to mediate interactions with the ATPase subunits (Rees et al., 2009).

We predicted that periplasmic regions of GltK, available to interact with periplasmic BcpA‐2‐CT that is received from neighboring bacteria, would be responsible for the specificity of BcpA‐2 for GltK (Figures 8b and S6). However, only two amino acids within the C‐terminal predicted periplasmic region vary between the B. multivorans and E. coli proteins, N186 and T193 (T186 and Q193 in GltK ). Of these, it is unlikely that N186 contributes specificity because P. luminescens TTO1 GltK, which is expected to be the native receptor for CdiA‐CTTTO1, shares the B. multivorans sequence (N186). A greater number of amino acid differences are found within the region’s predicted transmembrane helix (7 substitutions) and the cytoplasmic C‐terminus (4 substitutions). It is not known how BcpA‐CT or CdiA‐CT polypeptides translocate across the recipient cell cytoplasmic membrane nor the precise role that the identified inner membrane receptors, like GltJK, play in this process. Perhaps interactions of BcpA‐CT with membrane‐embedded or channel‐exposed portions of the GltK transmembrane helix or cytoplasmic region are necessary for translocation. Consistent with this hypothesis, a transmembrane helix of SecY was recently found to contribute to cytoplasmic transport of an E. coli CdiA‐CT (Jones et al., 2021).

Both GltJ and GltK were required for BcpA‐2 intoxication, although it is unknown whether both proteins participate directly in BcpA‐CT translocation. Our results also indicate that BcpA‐2 is only somewhat specific for Burkholderia GltJ, as recipient cells expressing E. coli gltJ were partially susceptible to CDI. Inner membrane levels of GltK did not depend on the presence of GltJ, suggesting that GltJ is not necessary for GltK localization or abundance. Chimeric GltJ proteins indicated that several periplasmic regions have minor contributions to the limited specificity of BcpA‐CT‐2 for GltJ . GltJ protein levels were also not examined here, so it is possible that defects in the localization or stability of heterologous GltJ contributed to our results. The precise function of GltJ in BcpA‐CT translocation remains unclear, but these results suggest that it plays a supporting role. Future examination of GltJ, particularly focusing on non‐periplasmic regions, may clarify its contribution to BcpA translocation.

In an attempt to swap GltJK specificity, our results demonstrating the inability of a chimeric BcpACGD2M‐CdiATTO1 protein to mediate CDI were somewhat expected. Unlike E. coli‐type CdiA proteins, for which functional chimeric proteins can be constructed by grafting the CdiA‐CT from other bacterial species (including CdiA‐CTTTO1) onto an E. coli CdiA N‐terminus (Aoki et al., 2010; Willett et al., 2015), Burkholderia BcpA proteins do not appear as modular. BcpA proteins fall into several subclasses and functional chimeras can be constructed by swapping BcpA‐CT domains among subclass‐1 or among subclass‐2 proteins (Anderson et al., 2014; Nikolakakis et al., 2012), but not between the subclasses (Anderson et al., 2014). Thus, CT domains of even closely‐related Burkholderia BcpA proteins cannot always be swapped to generate functional chimeras, due to yet unknown mechanisms. While it is possible that chimeric BcpACGD2M‐CdiATTO1 did not mediate CDI due to improper conformation of E. coli GltJK when produced in B. multivorans or the lack of a possible cytoplasmic “activator” protein (Diner et al., 2012; Koskiniemi et al., 2014), these previous data support the hypothesis that chimeric BcpACGD2M‐CdiATTO1, which fused a distantly‐related CT domain onto a BcpA N‐terminus, was nonfunctional. B. multivorans BcpA‐2 is an atypical BcpA, appearing to have a CdiA‐like “VENN” motif rather than the “Nx(E/Q)LYN” that precedes many BcpA‐CT domains (Myers‐Morales et al., 2019). Despite this commonality with CdiA, the results here suggest that BcpA‐2 is functionally similar to typical BcpA proteins in its lack of modularity. These results highlight gaps in the understanding of the molecular requirements for proper Burkholderia BcpA secretion, processing, and translocation into recipient cells.

The regulatory mechanisms controlling expression of B. multivorans gltIJKL have not been delineated. In Salmonella, expression of gltIJKL and other amino acid transporter‐encoding genes is regulated by nutrient availability and small regulatory RNAs (Miyakoshi et al., 2015; Sharma et al., 2007). Whether modulation of cellular concentrations of GltJK (or other BcpA/CdiA receptors) by environmental conditions could impact CDI susceptibility remains an interesting outstanding question.

Mechanisms of CDI systems found in Burkholderia species are distinct and the receptors that mediate BcpA entry into recipient cells are largely unknown. This study identified proteins likely composing the inner membrane receptor for one Burkholderia BcpA‐CT variant. A distinct inner membrane protein has also been identified as a probable receptor for a Burkholderia pseudomallei BcpA‐CT (Koskiniemi et al., 2015; Willett et al., 2015). Evidence from B. thailandensis also indicates that bacteria producing altered LPS are resistant to a specific BcpA variant, although it is unclear whether LPS functions as the direct outer membrane receptor or whether altered LPS indirectly affects BcpA entry (Koskiniemi et al., 2015).

Our findings support the model that BcpA and CdiA proteins are exchanged between closely related species. This exchange is likely limited predominantly by interactions between BcpA/CdiA and their cognate outer membrane receptors, although it is possible that inner membrane receptors like GltJK also contribute. It is unclear whether the specificity determinant identified here, the GltK C‐terminus, is also important for translocation of E. coli GltJK‐utilizing CdiA‐CT toxins. Similar specificity may exist for other CDI system effectors, but the extent to which inner membrane receptor variability restricts natural exchange of CdiA/BcpA proteins remains to be determined. The recent finding that some CdiA proteins can be exchanged cross‐species among Enterobacteriaceae spp. implies that any inner membrane receptor variation among these species does not substantially limit CdiA‐CT delivery (Virtanen et al., 2019). We expect that during natural CDI system activity, BcpA‐2‐CTCGD2M mainly encounters Burkholderia GltJK, rather than γ‐proteobacterial GltJK. Thus, it is not unexpected that the translocation domain of BcpA‐2‐CTCGD2M evolved to recognize the GltJK homolog found in its native recipient cells. Overall, the results here contribute to the model of CDI system protein exchange and provide insight into the mechanism of BcpA translocation across the cytoplasmic membrane.

EXPERIMENTAL PROCEDURES

Bacterial strains and culture conditions

Bacterial strains used in this study are listed in Table 1. B. multivorans CGD2M was isolated from a patient with chronic granulomatous disease. B. thailandensis E264 is an environmental isolate (Brett et al., 1998). Plasmids were maintained in E. coli DH5α and mated into B. multivorans using the donor E. coli strain RHO3 (López et al., 2009). B. multivorans was cultured in low salt Luria‐Bertani medium (LSLB, 0.5% NaCl), supplemented, as appropriate, with 20 µg/ml chloramphenicol, 250 µg/ml kanamycin, 25–50 µg/ml tetracycline, or 0.2% glucose. Overnight cultures were aerated for ∼18 hr at 37℃ to OD600 ∼2, centrifuged at 16,000×g for 2 min, washed, and resuspended in fresh LSLB broth or sterile phosphate buffered saline (PBS) for subsequent use. E. coli strains were cultured in LSLB supplemented, as appropriate, with 100 µg/ml ampicillin, 50 µg/ml kanamycin, 10 µg/ml tetracycline, 200 µg/ml 2,6‐diaminopimelic acid (for RHO3 strains), or 0.2% glucose.

TABLE 1

Bacterial strains and mutants used in this study

Strain	Description	References
E. coli RHO3	Conjugation donor strain, Δasd	López et al. (2009)
E. coli DH5α	E. coli cloning strain
CGD2M	Wild‐type Burkholderia multivorans isolated from patient with chronic granulomatous disease
Δbcp‐2	In‐frame deletion of bcpAIB‐2 in CGD2M	Myers‐Morales et al. (2019)
Δbcp‐1 Δbcp‐2	In‐frame deletion of bcpAIOB‐1 and bcpAIB‐2 in CGD2M	Myers‐Morales et al. (2019)
ΔgltJK	In‐frame deletion of gltJ and gltK in CGD2M	This study
Δbcp‐1 Δbcp‐2 ΔgltJK	In‐frame deletion of gltJ and gltK in CGD2M Δbcp‐1 Δbcp‐2 mutant	This study
Δbcp‐2 ΔgltJK	In‐frame deletion of gltJ and gltK in CGD2M Δbcp‐2 mutant	This study
Δbcp‐1 Δbcp‐2 ΔgltI	In‐frame deletion of gltI in CGD2M Δbcp‐1 Δbcp‐2 mutant	This study
CGD2M‐30	CGD2M with the native bcpA‐2‐CT and bcpI‐2 replaced with Burholderia cenocepacia Bp8968 bcpA‐CT, bcpI and DF018_16545, starting at the VENN of bcpA‐2	This study
CGD2M‐32	CGD2M with the native bcpA‐2‐CT and bcpI‐2 replaced with Photorhabdus luminecens TTO1 cdiA‐CT and cdiI, starting at the VENN of bcpA‐2	This study

Genetic manipulations

Plasmids used in this study are listed in Table 2. All plasmids were confirmed by DNA sequencing (Eurofins Genomics or ACGT, Inc.) and bacterial mutant strains verified by PCR.

TABLE 2

Plasmids used in this study

Strain or plasmid	Description	References
pEXKm5	sacB‐based allelic exchange vector, KanR	López et al. (2009)
pUC18T‐miniTn7‐kan	Plasmid for attTn7 site delivery, AmpR, KanR	Choi et al. (2005)
pTNS3	Helper plasmid for attTn7 site delivery, AmpR	Choi et al. (2008)
pUCS12	PS12 promoter in pUC18T‐miniTn7‐kan, AmpR, KanR	Anderson et al. (2012)
pUCCm	CmR plasmid for attTn7 site delivery, AmpR, CmR	Anderson et al. (2012)
pUCTet	TetR plasmid for attTn7 site delivery, AmpR, TetR	Anderson et al. (2012)
pECG69	bcpI‐1 driven by PS12 in pUCTet, AmpR, TetR	Myers‐Morales et al. (2019)
pECG70	bcpI‐2 driven by PS12 in pUCTet, AmpR, TetR	Myers‐Morales et al. (2019)
pECG74	First ~500 bp of bcpA‐1 cloned 3′ to PS12 in pUCS12, for constitutive bcpAIOB‐1, KanR	Myers‐Morales et al. (2019)
pUT‐miniTn5‐Kn	Used for transposon mutagenesis, KanR	de Lorenzo et al. (1990)
pScrhaB3	Used for expression of genes under rhamnose induction, KanR	Cardona and Valvano (2005)
pTMM009	Deletion allele for gltJ and gltK in pEXKm5, KanR	This study
pTMM012	gltJKB. mult . operon (driven by PS12) in pUC18T‐miniTn7‐kan, AmpR, KanR	This study
pTMM013	gltJKE. coli operon (driven by PS12) in pUC18T‐miniTn7‐kan, AmpR, KanR	This study
pTMM016	gltJKB. thai operon (driven by PS12) in pUC18T‐miniTn7‐kan, AmpR, KanR	This study
pTMM018	gltJE. coli (driven by PS12) in pUC18T‐miniTn7‐kan, AmpR, KanR	This study
pTMM019	gltKE. coli (driven by PS12) in pUCTet, AmpR, TetR	This study
pTMM020	gltJB. mult (driven by PS12) in pUC18T‐miniTn7‐kan, AmpR, KanR	This study
pTMM021	gltKB. mult . (driven by PS12) in pUCTet, AmpR, TetR	This study
pJChim1	GeneArt plasmid containing gltJ Chim1 in pMK‐RQ backbone, KanR	This study
pKChim1	GeneArt plasmid containin gltK Chim1 in pMK‐RQ backbone, KanR	This study
pKChim2	GeneArt plasmid containin gltK Chim2 in pMA‐RQ backbone, AmpR	This study
pTiT01	GeneArt plasmid for bcpA‐2 chimera, includes ~500 bp upstream of CGD2M bcpA‐2‐CT in‐frame with cdiA‐CT and cdiI from Photorhabdus luminescens TTO1, and ~500 bp downstream of CGD2M bcpI‐2 in pMK‐RQ backbone, KanR	This study
pBp8968	GeneArt plasmid for bcpA‐2 chimera, includes ~500bp upstream of CGD2M bcpA‐2‐CT in‐frame with bcpA‐CT, bcpI and DF018_16545 from Burkholderia cenocepacia Bp8968, and ~500 bp downstream of CGD2M bcpI‐2 in pMK‐RQ backbone, KanR	This study
pTMM022	gltJ Chim1 (driven by PS12) in pUC18T‐miniTn7‐kan, AmpR, KanR	This study
ipTMM023	gltK Chim1 (driven by PS12) in pUCTet, AmpR, TetR	This study
pTMM024	gltIJKLE. coli complete locus (driven by PS12) in pUC18T‐miniTn7‐kan, AmpR, KanR	This study
pTMM025	CGD2M bcpA‐2‐CT in pSCrhaB3, KanR	This study
pTMM027	gltK Chim2 (driven by PS12) in pUCTet, AmpR, TetR	This study
pTMM030	Allelic exchange plasmid for bcpA‐2 chimera with bcpA‐CT, bcpI and DF018_16545 from Burkholderia cenocepacia Bp8968 and ~500 bp flanking CGD2M sequence in pEXKm5, KanR	This study
pTMM031	B. cenocepacia Bp8968 bcpI (driven by PS12) in pUC18T‐miniTn7‐kan, AmpR, KanR	This study
pTMM032	Allelic exchange plasmid for bcpA‐2 chimera with cdiA‐CT and cdiI from Photorhabdus luminescens TTO1 and ~500 bp flanking CGD2M sequence in pEXKm5, KanR	This study
pTMM034	Deletion allele for gltI in pEXKm5, KanR	This study
pTMM035	gltI from CGD2M (driven by PS12) in pUC18T‐miniTn7‐kan, AmpR, KanR	This study
pTMM036	gltJ Chim2 (driven by PS12) in pUC18T‐miniTn7‐kan, AmpR, KanR	This study
pTMM037	gltK Chim3 (driven by PS12) in pUCTet, AmpR, TetR	This study
pTMM038	gltK Chim4 (driven by PS12) in pUCTet, AmpR, TetR	This study
pTMM039	gltK Chim5 (driven by PS12) in pUCTet, AmpR, TetR	This study
pTMM040	gltK Chim6 (driven by PS12) in pUCTet, AmpR, TetR	This study
pTMM041	gltKB. mult ‐FLAG (driven by PS12) in pUC18T‐miniTn7‐kan, AmpR, TetR	This study
pTMM042	gltKE. coli ‐FLAG (driven by PS12) in pUCTet, AmpR, TetR	This study
pTMM046	gltK Chim6‐FLAG (driven by PS12) in pUCTet, AmpR, TetR	This study
pTMM047	gltK Chim3‐FLAG (driven by PS12) in pUCTet, AmpR, TetR	This study

Abbreviations: Amp, ampicillin; Cm, chloramphenicol; Kan, kanamycin; Tet, tetracycline.

In‐frame deletion mutations of gltJK and gltI were constructed by allelic exchange. DNA fragments corresponding to ~500 bp 5′ to gltJ (BURMUCGD2M_3194, including the first three codons of gltJ) and ~500 bp 3′ to gltK (BURMUCGD2M_3193, including the last ten codons of gltK), were joined by overlap PCR and cloned into EcoRI‐digested allelic exchange vector pEXKm5, resulting in plasmid pTMM009. A fragment consisting of ~500 bp 5′ to gltI (BURMUCGD2M_3196, including the first ten codons of gltI) and ~500 bp 3′ to gltI (including the last 9 codons of gltI) was generated and cloned into pEXKm5, resulting in plasmid pTMM034. Both plasmids were used for allelic exchange in CGD2M as described (López et al., 2009).

DNA fragments containing bcpA‐2 chimeras with Photorhabdus luminescens TTO1 (TTO1) cdiA‐CT or B. cenocepacia Bp8968 (Bp8968) bcpA‐CT were ordered from Thermo Fisher Scientific (GeneArts Strings Gene Synthesis) flanked by EcoRI sites on each end. The regions ordered contained 530bp upstream of the CGD2M bcpA‐2 (BURMUCGD2M_2430) VENN ending in nucleotide 9,297 from bcpA2, followed by either the TTO1 cdiA‐CT (plu0548) starting at 12,727bp of CdiA (TTO1 VENN) and ending at the stop codon of TTO1 cdiI (plu0547) or the Bp8968 bcpA‐CT (DF018_16555) starting at 9,297bp of B. cenocepacia Bp8968 BcpA (Bp8968 VENN) and ending at the stop codon of DF018_16545 (extra open reading frame after B. cenocepacia Bp8969 bcpI). Either region was followed by 569bp downstream of the CGD2M bcpI‐2 (BURMUCGD2M_2429) stop codon. Both chimeric fragments were received in plasmids we designated pTTO1 and pBp8968, respectively. After digest of plasmids with EcoRI, the region containing the chimeric inserts was band purified and ligated into EcoRI digested pEXKm5, resulting in plasmids pTMM032 (TTO1 chimera) and pTMM030 (Bp8968 chimera) which were used for allelic exchange in CGD2M as described (López et al., 2009) resulting in strains with chimeric bcpA‐2 alleles designated CGD2M‐32 and CGD2M‐30 respectively. For competitions, a gene conferring chloramphenicol resistance was added to CGD2M‐32 with plasmid pUCCm (previously described (Anderson et al., 2012)) and a gene encoding kanamycin resistance was added to CGD2M‐30 with plasmid pUC18T‐miniTn7T‐Kan by delivery to one of the attTn7 sites (Choi et al., 2008).

Constitutive expression of the bcp‐1 locus in CGD2M, achieved by integration of the PS12 promoter (promoter of B. thailandensis rpsL, encoding the ribosomal S12 subunit) immediately 5′ to bcpA‐1 with plasmid pECG74, has been described previously (Myers‐Morales et al., 2019).

Deletion mutants were complemented by introducing chromosomal copies of the indicated genes to attTn7 sites using miniTn7 transposition as described previously (Choi et al., 2008). gltJK was PCR amplified from CGD2M, Bukholderia thailandensis E264, or E. coli DH5α (DH5α) and cloned 3′ to PS12 in pUCS12 (Anderson et al., 2012), resulting in pTMM012, pTMM016 and pTMM013, respectively. The complete E. coli gltIJKL locus was PCR‐amplified from DH5α and cloned 3′ to PS12 in pUCS12 resulting in pTMM024. gltJ was PCR‐amplified from CGD2M or DH5α and cloned 3′ to PS12 in pUCS12 resulting in pTMM020 and pTMM018, respectively. gltK was PCR‐amplified from CGD2M or DH5α and cloned 3′ to PS12 in pUCTet (Anderson et al., 2012), resulting in pTMM021 and pTMM019, respectively. gltI was PCR amplified from CGD2M and cloned 3′ to PS12 in pUCS12 resulting in pTMM035. B. cenocepacia Bp8968 bcpI was amplified from pBp8968 and cloned 3′ to PS12 in pUCS12, resulting in plasmid pTMM031. Plasmids pECG69 and pECG70, encoding a PS12‐driven copy of CGD2M bcpI‐1 and bcpI‐2, respectively, have been described previously (Myers‐Morales et al., 2019). For use in competition assays, the mutants were tagged with a tetracycline resistance cassette at a second attTn7 site using pUCTet or a kanamycin resistance cassette using pUC18T‐miniTn7T‐Kan (Myers‐Morales et al., 2019). The mutant expressing gltJ alone (KanR) was additionally marked with a tetracycline resistance cassette at a second attTn7 site using pUCTet. Similarly, the mutant expressing bcpI‐2 (TetR) was additionally tagged with a kanamycin resistance cassette using pUC18T‐miniTn7T‐Kan.

DNA fragments containing chimeric gltJ and gltK genes composed of CGD2M and E. coli sequences were ordered from Thermo Fisher Scientific (GeneArts Strings Gene Synthesis). Chimeric gene gltJ consists of CGD2M gltJ. coding sequence with nucleotides 7–57 and 281–345 replaced with nucleotides 7–54 and 281–345 from E. coli MG1655 gltJ (b0654), respectively. Flanked by 5′ HindIII and 3′ StuI sites, restriction digestion was used to clone gltJ into pUCS12 downstream of PS12, resulting in plasmid pTMM022. Chimeric gene gltK Chim1 consists of CGD2M gltK coding sequence with nucleotides 4–87, 241–294, and 553–678 replaced with nucleotides 4–87, 241–294, and 553–675 from E. coli MG1655 gltK (b0653), respectively. Chimeric gene gltK Chim2 is the opposite of gltK Chim1, consisting of E. coli MG1655 gltK coding sequence with nucleotides 4–87, 241–294, and 553–678 replaced with nucleotides 4–87, 241–294, and 553–678 from CGD2M gltK. Flanked by 5′ HindIII and 3′ KpnI restriction sites, restriction digestion was used to clone gltK Chim1 and gltK Chim2 genes 3′ to PS12 in pUCTet, resulting in pTMM023 and pTMM027, respectively.

Chimeric gene gltJ Chim2 consists of E. coli MG1655 gltJ coding sequence with nucleotides 7–75 replaced with nucleotides 7–78 from CGD2M gltJ. gltJ Chim2 was made by amplifying PS12 and the 5′ end of gltJ. (nucleotides 1–78) from pTMM020 and the 3′ end of gltJ (nucleotides 76–741) from pTMM018. The two regions were joined by overlap PCR and cloned into pUC18T‐miniTn7T‐Kan, resulting in plasmid pTMM036.

The remaining chimeric gltK genes were constructed by restriction digestion of plasmids containing gltK, gltK Chim1 and gltK Chim2, taking advantage of naturally occurring HpaI and BglI restriction sites within the gltK sequence. Plasmids pTMM019 (E. coli gltK) and pTMM027 (gltK Chim2) were digested with SacI and StuI, and the resulting inserts (which include PS12 and the gltK gene) were further digested with: (1) HpaI to produce 438bp 5′ (H Up) and 519bp 3′ (H down) fragments or (2) BglI to produce 653bp 5′ (B up) and 309bp 3′ (B down) fragments. Combinations of these fragments were cloned into pUCTet: pTMM027 B Up with pTMM019 B Down resulting in pTMM037 (gltK Chim3), pTMM019 H Up with pTMM027 H Down resulting in pTMM038 (gltK Chim4) and pTMM019 B Up with pTMM027 B Down resulting in pTMM040 (gltK Chim6). Plasmid pTMM040 was similarly digested with SacI and StuI and the resulting gltK Chim6 insert was subsequently digested with HpaI to produce two fragments, H up (438bp) and H Down (519bp). Fragment H Down from pTMM040 was ligated with pTMM027 H Up into pUCTet to produce pTMM039 (gltK Chim5).

Nucleotides encoding a 3x alanine linker and 1x FLAG tag were added to the 3′ end of the indicated gltK genes by PCR amplification of gltK from CGD2M or DH5α or chimeric gltK genes from pTMM040 and pTMM037. The resulting gltK genes were cloned 3′ of PS12 into pUCTet, resulting in plasmids that produce C‐terminal FLAG‐tagged GltK: pTMM041 (GltK ‐FLAG), pTMM042 (GltK ‐FLAG), pTMM046 (GltKChim6‐FLAG), and pTMM047 (GltKChim3‐FLAG).

Nucleotides encoding the C‐terminus of CGD2M BcpA‐2 (nucleotides 9,298–10,272) were amplified with a 5′ primer that added an in‐frame start codon. This fragment was cloned 3′ of Prham in pScrhaB3 (Cardona & Valvano, 2005), resulting in pTMM025.

Transposon mutagenesis selection and arbitrary PCR

Random transposon mutagenesis of CGD2M ΔbcpAIB‐2 was achieved by delivering pUT‐MiniTn5‐Kn (de Lorenzo et al., 1990) by conjugation as described (López et al., 2009). The mating mixture was serially diluted and plated on LSLB with kanamycin. Colonies were enumerated, collected, and pooled in LSLB with 15% glycerol for storage of the transposon mutant library.

For selection of CDI resistant (CDIR) mutants, sequential interbacterial competition assays were used (Anderson et al., 2012; Aoki et al., 2008). Briefly, the transposon pool was cultured in LSLB with kanamycin and mixed at 1:1 ratio with CmR wild‐type CGD2M. The mixture was serially diluted and plated on antibiotic plates to determine the initial competition ratio. The mixture (20 µl) was spotted on LSLB agar in triplicate, air dried, and incubated at 25℃ for 24 hr. Colony spots were collected with a sterile loop, serially diluted, and plated on antibiotic plates to determine final competition ratios. KanR transposon mutant colonies were collected from output plates and pooled from all replicates in LSLB with 15% glycerol for storage. This process was repeated three times, using the output pool of transposon mutants as the inoculum for the next round of competition selection.

Transposon insertion sites of CDIR mutants were determined by arbitrary PCR, as described (Intile et al., 2015) with modifications. Genomic DNA was extracted from CDIR mutants and a wild‐type CGD2M control using the Wizard® Genomic DNA Purification system (Promega) according to the manufacturer’s protocol. Nested arbitrary‐primed PCR was performed using this genomic DNA as template and a first round primer annealing to the 3′ end of the transposon (Tn3out, 5′‐CACGCAGATGGGCCGGC) and an arbitrary primer (Arb1, 5′‐GGCCACGCGTCGACTAGTACNNNNNNNNNNACGCC). Products from the first reaction were treated with ExoSAP‐IT™ PCR Product Cleanup Reagent (Applied Biosystems) and used as template in a second, nested PCR reaction using primers Arb2 (5′‐GGCCACGCGTCGACTAGTAC) and Tn3in (5′‐CAAGCGCGAGATGTTCACCGACCC). Second‐round PCR products were analyzed by agarose gel electrophoresis, treated with ExoSAP‐IT™, and the transposon‐chromosome junctions sequenced with primer Tn3seq (5′‐CATCACACGAACAAAGATGG).

Interbacterial competition assay

Interbacterial competition assays were performed as previously described (Anderson et al., 2012), with modifications. Bacteria carrying antibiotic resistance markers were cultured overnight, washed with PBS, and diluted in PBS to OD600 = 2. Strains were mixed at a 1∶1 ratio, and 20 µl of the mixture was plated on LSLB agar without antibiotic selection. The culture inoculum was plated on LSLB with antibiotic selection to determine the ratio of each strain at 0 hr. Agar plates were incubated at 25℃ for 24–26 hr. Bacteria were picked from the edge of the resulting colony spots/biofilms with a sterile pipette tip, diluted in PBS, and plated on LSLB with antibiotic selection to quantify each strain at 24 hr. The competitive index (CI) was calculated as the ratio of the inhibitor strain to the target strain at time 24 hr divided by the ratio at time 0 hr. Two to three independent experiments were performed in triplicate.

Intracellular toxicity assay

Plasmids pScrhaB3 and pTMM025 (Rham‐bcpA‐2‐CT) were introduced by conjugation into CGD2M ΔbcpAIOB‐1 ΔbcpAIB‐2, CGD2M ΔbcpAIOB‐1 ΔbcpAIB‐2 ΔgltJK and each strain carrying bcpI‐2 at an attTn7 site. Conjugation mixtures were incubated on LSLB agar supplemented with DAP at 37℃ for 6 hr. Approximately half of each mating mixture was collected with a sterile polyester swab, streaked for isolation on plates supplemented with kanamycin and 0.2% glucose, incubated ~24 hr at 37℃, and imaged.

Subcellular fractionation and western blotting

Subcellular fractionation of CGD2M was performed by the Sarkosyl extraction method (Filip et al., 1973; Myers‐Morales et al., 2007) with modifications. Briefly, bacterial strains were cultured overnight at 37℃ with shaking in LSLB. Strains were resuspended to an OD600 of 4 in HEPES Resuspension Buffer (10 mM HEPES, pH 7.5 supplemented with Roche Complete Mini EDTA‐free Protease Inhibitor Cocktail and Pierce Universal Nuclease for Cell Lysis). Cells were broken by two passages through a chilled French pressure cell (20,000 lb/in2), and unbroken cells and large debris were removed by centrifugation at 10,000×g at 4℃ for 15 min, followed by an additional centrifugation for 10 min. Total membranes were separated by ultracentrifugation for 15 min at 100,000×g at 4℃ and supernatants collected for analysis of the cytoplasmic fraction. Total membranes were washed with HEPES Resuspension Buffer, and the pellet was resuspended in 0.3% (wt/vol) sodium laurylsarcosinate (Sarkosyl) in 10 mM HEPES, pH 7.5 and incubated with rotation at RT for 30 min. Samples were separated by ultracentrifugation at 100,000×g at 20℃ and supernatants collected for analysis of Sarkosyl‐soluble (inner membrane‐enriched) proteins. The pellet was washed with 0.3% Sarkosyl Buffer, and the Sarkosyl‐insoluble (outer membrane‐enriched) fraction was resuspended in 10 mM Tris, pH 8 with 1% sodium dodecyl sulfate (SDS) (Tris/SDS buffer). The cytoplasmic and inner membrane fractions were concentrated by methanol−chloroform precipitation, and the resulting pellets were resuspended in Tris/SDS buffer. Protein concentration for all fractions was determined by microplate BCA assay (Pierce).

Equal protein amounts of each fraction were analyzed by sodium dodecyl sulfate‐polyacrylamide gel electrophoresis (SDS‐PAGE) and transferred to nitrocellulose membrane. Immunoblots were probed with rabbit polyclonal anti‐Salmonella typhii Outer Membrane protein A (Biomatik) or mouse monoclonals anti‐FLAG M2 (Sigma) and anti‐E. coli RNA Polymerase β (Biolegend) and secondary antibodies coupled to IRDye 800CW (Licor). Silver staining of the gels containing the inner membrane (Sarkosyl‐soluble) fractions was used for visualization of protein loading. Immunoblots and gels were imaged on a BioRad ChemiDoc MP Imaging System.

Bioinformatics and statistics

BLASTp analysis using B. multivorans BcpA‐2‐CT as a query was used to identify BcpA from B. cenocepacia strain Bp8968 (encoded by DF018_16555). Protein sequences were aligned using the ClustalW alignment function in Geneious (v.6.1.8). Transmembrane helices and membrane topology of GltJ and GltK were predicted using Uniprot (UniProt Consortium, 2019) and the TMPred program available via ExPASY‐Swiss Institute of Bioinformatics. The Shapiro‐Wilk normality test was performed on all datasets to determine distribution. Normally distributed data were analyzed by two‐tailed Student’s t‐tests. Data failing the normality test were analyzed by the nonparametric Mann−Whitney test. All statistics were performed with GraphPad Prism (v.8).

CONFLICT OF INTEREST

The authors have no conflict of interest to declare.

Supplementary Material

Click here for additional data file.

Table S1

Click here for additional data file.