ABSTRACT
Attachment is essential for microorganisms to establish interactions with both biotic and abiotic surfaces. Stable attachment of Caulobacter crescentus to surfaces requires an adhesive polysaccharide holdfast, but the exact composition of the holdfast is unknown. The holdfast is anchored to the cell envelope by outer membrane proteins HfaA, HfaB, and HfaD. Holdfast anchor gene mutations result in holdfast shedding and reduced cell adherence. Translocation of HfaA and HfaD to the cell surface requires HfaB. The Wzx homolog HfsF is predicted to be a bacterial polysaccharide flippase. An hfsF deletion significantly reduced the amount of holdfast produced per cell and slightly reduced adherence. A ΔhfsF ΔhfaD double mutant was completely deficient in adherence. A suppressor screen that restored adhesion in the ΔhfsF ΔhfaD mutant identified mutations in three genes: wbqV, rfbB, and rmlA. Both WbqV and RfbB belong to a family of nucleoside-diphosphate epimerases, and RmlA has similarity to nucleotidyltransferases. The loss of wbqV or rfbB in the ΔhfsF ΔhfaD mutant reduced holdfast shedding but did not restore holdfast synthesis to parental levels. Loss of wbqV or rfbB did not restore adherence to a ΔhfsF mutant but did restore adherence and holdfast anchoring to a ΔhfaD mutant, confirming that suppression occurs through restoration of holdfast anchoring. The adherence and holdfast anchoring of a ΔhfaA ΔhfaD mutant could be restored by wbqV or rfbB mutation, but such mutations could not suppress these phenotypes in the ΔhfaB mutant. We hypothesize that HfaB plays an additional role in holdfast anchoring or helps to translocate an unknown factor that is important for holdfast anchoring.
IMPORTANCE Biofilm formation results in increased resistance to both environmental stresses and antibiotics. Caulobacter crescentus requires an adhesive holdfast for permanent attachment and biofilm formation, but the exact mechanism of polysaccharide anchoring to the cell and the holdfast composition are unknown. Here we identify novel polysaccharide genes that affect holdfast anchoring to the cell. We identify a new role for the holdfast anchor protein HfaB. This work increases our specific knowledge of the polysaccharide adhesin involved in Caulobacter attachment and the general knowledge regarding production and anchoring of polysaccharide adhesins by bacteria. This work also explores the interactions between different polysaccharide biosynthesis and secretion systems in bacteria.
INTRODUCTION
Biofilm-forming bacteria produce different forms of cell surface-associated exopolysaccharides (EPS) to facilitate attachment to surfaces. For example, Sinorhizobium meliloti produces a symbiotic exopolysaccharide that mediates the invasion of root nodules of Medicago sativa (alfalfa) (1). Polysaccharide adhesins, such as the polysaccharide intercellular adhesin (PIA), are synthesized by both Gram-positive and Gram-negative bacteria, such as Staphylococcus spp. and Escherichia coli, for functions suited to their individual lifestyles (2–5). The Gram-negative aquatic bacterium Caulobacter crescentus synthesizes a polarly localized adhesin, known as the holdfast, that is made of an N-acetylglucosamine (GlcNAc)-containing polysaccharide and other unknown components (6). Although other polar organelles, such as the pili and flagellum, are involved in the initial attachment of C. crescentus cells to surfaces (7, 8), the holdfast plays a key role in strong permanent attachment of C. crescentus cells to a substrate or to each other (9). The holdfast has properties of an elastic gel, and its force of adhesion is extremely strong (10, 11).
Holdfast biogenesis is a complex process that is predicted to begin in the cytoplasm with the biosynthesis of lipid-linked precursors by glycosyltransferases and other enzymes. Following biosynthesis, translocation and assembly of the precursors generate the full-length holdfast, which is anchored to the cell surface via a protein complex (Fig. 1). The following three distinct classes of genes are major contributors to the elaboration of the holdfast: holdfast biosynthesis genes (hfs [holdfast synthesis]), holdfast anchor genes (hfa [holdfast anchor]), and certain pleiotropic developmental genes (for example, podJ and pleC) (12–19). The major hfs gene cluster comprises eight genes, namely, hfsDABC and hfsEFGH, and three unlinked genes, hfsI, hfsJ, and hfsK, also contribute to holdfast synthesis (16, 20–22). Biosynthesis and polymerization of the holdfast are performed by HfsEFGH, HfsJ, HfsK, and then HfsC and HfsI (Fig. 1). HfsE is a predicted integral membrane sugar transferase with similarity to the family of polyisoprenylphosphate hexose-1-phosphate transferase (PHPT) sugar transferases (23, 24). HfsJ is a member of the WecG/TagA enzyme family that is involved in transfer of an activated sugar to a previously glycosylated undecaprenyl phosphate (Und-PP-GlcNAc) (21). hfsF encodes a protein with similarity to polysaccharide Wzx flippases. hfsG encodes a cytoplasmic protein with similarity to family 2 glycosyltransferases. hfsH encodes a predicted carbohydrate esterase and hfsK an N-acetyltransferase that are required for holdfast adhesiveness and cohesiveness (22). HfsC and HfsI are paralogs with similarity to Wzy O-antigen polymerases (20). Deletions of hfsE (and its unlinked paralogs pssY and pssZ), hfsF, hfsG, hfsH, and hfsC (and its paralog hfsI) have various impacts on holdfast polysaccharide synthesis. ΔhfsE, ΔhfsF, ΔhfsC, and ΔhfsI mutants are still able to adhere to surfaces and synthesize holdfasts, but the ΔhfsG and ΔhfsJ mutants cannot (20, 21). The ΔhfsH and ΔhfsK mutants synthesize holdfast but shed holdfast due to reduced adhesiveness and cohesiveness (20, 22). For hfsC and hfsI, both polymerases must be deleted to abolish holdfast synthesis. This is also true for hfsE and its paralogs pssY and pssZ; all three genes must be deleted to abolish holdfast synthesis (20).
Model of holdfast biosynthesis, secretion, and anchoring. The model shows the predicted functions and membrane localization of proteins involved in the biosynthesis, secretion, and anchoring of the holdfast polysaccharide. The inner membrane (IM), outer membrane (OM), lipopolysaccharide (LPS), cell wall (CW), undecaprenyl phosphate (Udp-P), and N-acetylglucosamine (GlcNAc) are indicated. HfsE and the paralogs PssY and PssZ are the predicted initiating glycosyltransferases, which are predicted to attach a GlcNAc moiety to Udp-P. HfsJ and HfsG are glycosyltransferases that add additional sugars to the polysaccharide unit. HfsH is a predicted deacetylase that partially deacetylates the sugars in the polysaccharide subunit. HfsK is a predicted N-acetyltransferase that affects acetylation of the sugars in conjunction with HfsH. HfsF is a predicted polysaccharide flippase that transfers the polysaccharide subunit across the IM. HfsC and HfsI are predicted polysaccharide polymerases that assemble the polysaccharide units into the growing holdfast polysaccharide chain. The holdfast is secreted across the outer membrane by HfsD, HfsA, and HfsB, which have similarity to Wza, Wzb, and Wzc in the E. coli type I capsule secretion system. Finally, holdfast anchoring requires HfaA, HfaB, and HfaD. HfaA has properties of amyloid proteins, and HfaD is similar to other bacterial adhesins. HfaB is similar to the curlin secretion protein CsgG and is required for the proper localization of HfaA and HfaD to the OM. The mechanism by which the holdfast is anchored to the cells is unknown.
The hfsDAB locus, which encodes proteins involved in secretion of holdfast, is adjacent to but divergent from hfsEFGH (Fig. 1). HfsD has similarity to Wza, an outer membrane lipoprotein involved in polysaccharide translocation (16). HfsA has similarity to Wzc, a member of the membrane periplasmic auxiliary family (MPA-1) of polysaccharide export proteins (16). The complex of Wzc and Wza works to polymerize and translocate E. coli type I capsule to the cell surface (25). In Gram-positive bacteria, Wzc is composed of two separate proteins, which is also the case in C. crescentus. HfsB has sequence similarity to Wzc ATPase but has lost the ATP binding site (26). Mutations in any of the hfsDAB genes result in the complete loss of holdfast production (16).
The hfa locus comprises three genes, hfaA, hfaB, and hfaD, all of which are required for anchoring of the holdfast to the tip of the stalk (Fig. 1) (12–15, 17, 18). HfaA has some similarity to fimbrial proteins and to the curlin monomer CsgA (17, 18). HfaA contains several amyloid-like domains and forms heat- and SDS-resistant multimers whose assembly is dependent on HfaD (18). The lipoprotein HfaB is similar to CsgG, the outer membrane porin for CsgA and CsgB (27), and is necessary for targeting of HfaA and HfaD to the outer membrane (18). HfaD is an outer membrane protein with limited similarity to the paracrystalline surface S-layer and to several adhesins (12). HfaD also generates multimers that are dependent on both HfaA and HfaB but not on holdfast polysaccharide (18).
In order to perform a screen for suppressor mutations in holdfast synthesis and/or anchoring pathways, we used the nonadherent ΔhfsF ΔhfaD double mutant as a sensitized background. Mutations in the wbqV gene (CC_3210/CCNA_03316), rfbB (CC_3629/CCNA_03744), or rmlA (CC_1141/CCNA_01199) restored holdfast-dependent adherence to the ΔhfsF ΔhfaD mutant. WbqV and RfbB have sequence similarity to nucleoside-diphosphate dehydratases/epimerases, and RmlA has sequence similarity to nucleotidyltransferases. We hypothesize that loss of WbqV, RfbB, or RmlA results in the use of an alternate polysaccharide synthesis and translocation pathway, resulting in partial bypass of the holdfast anchor deficiency. Additionally, the loss of wbqV or rfbB can restore adhesion to ΔhfaA and ΔhfaD mutants but not to a ΔhfaB mutant, suggesting that HfaB may have a previously unidentified role in holdfast anchoring or may secrete an unknown factor (in addition to HfaA and HfaD) important for anchoring of the holdfast to the cell.
RESULTS
Synthetic adhesion phenotype of the ΔhfsF mutant and holdfast anchor gene mutations.In order to gain insight into the mechanisms of holdfast anchoring, we sought to isolate suppressor mutations that restore adherence to a holdfast anchor mutant. However, even though holdfast anchor mutants have reduced adherence, their relative adherence is still ∼10 to 30% of the wild-type level (18), making it difficult to perform a suppressor screen for restored adherence. We therefore sought to create a sensitized background in which holdfast anchoring suppressors could easily be enriched. Some mutations in holdfast synthesis genes, including combined mutations in redundant holdfast synthesis genes, completely abolished holdfast synthesis. The two exceptions to this were the predicted polysaccharide deacetylase gene hfsH, whose mutation decreased holdfast adhesiveness and cohesiveness, and the polysaccharide flippase gene hfsF, whose deletion resulted in a moderate loss of adherence. By staining cells adherent to polystyrene with crystal violet dye followed by elution and quantification of the bound dye, we found that deletion of hfsF reduced adherence to ∼50% of the wild-type level (Fig. 2A), confirming previous results (20).
The ΔhfsF deletion results in reduced attachment and holdfast production. (A) Short-term binding assay of a ΔhfsF mutant and controls. CB15 is the parent strain. CB15 ΔhfsDAB is a holdfast-negative mutant. The relative percent binding is normalized to the binding of CB15. The assay was performed in triplicate for at least three independent cultures. The mean ± standard error (SE) is shown for each strain. (B) Holdfast quantification by dot blot assay to compare the ΔhfsF mutant to the CB15 parent and the holdfast-negative strain CB15 ΔhfsDAB. PYE was used as a medium control. T is the total culture, C is cells only, and S is the culture supernatant. Four 2-fold serial dilutions were done for each sample. This is a representative sample. (C) Holdfast shedding assay and lectin binding assay to examine holdfast shedding and the amount of holdfasts. Top panels are phase-contrast images of cells, and bottom panels are fluorescence images of holdfasts stained with lectin. All images show representative areas of a glass coverslip that was submerged in a culture of each strain, washed, stained, and imaged. Each image was acquired and processed in the same manner.
Our previous analysis of the hfsF mutant holdfasts by transmission electron microscopy suggested that the hfsF mutant holdfasts were smaller than those of wild-type cells (20). This result was confirmed using a holdfast shedding assay and microscopy. Cells producing a holdfast bind to a glass coverslip and are visualized by fluorescent-lectin staining. Adherent cells elaborate holdfasts that remain attached to the glass slide, and an anchor defect is indicated by the inability of cells to remain associated with that holdfast. Holdfast staining indicated that ΔhfsF holdfasts had 67% of the fluorescence intensity of wild-type holdfasts (see Fig. S1A in the supplemental material). Wheat germ agglutinin (WGA) is specific to N-acetylglucosamine (GlcNAc) and sialic acid sugars, so there was a possibility that the reduction in fluorescence staining in the ΔhfsF holdfast was due to an alteration in the holdfast polysaccharide or the specificity of binding by the lectin to the holdfast (28, 29). To confirm that the size of the holdfasts was reduced in the ΔhfsF mutant, atomic force microscopy (AFM) was performed. AFM analysis showed that shed holdfasts of the ΔhfsF mutant had an average height of 40 nm, compared to 63 nm for shed wild-type holdfasts (Fig. S1B). Furthermore, the height distribution of ∼2,000 shed holdfasts exhibited a wide distribution for both mutants (Fig. S1C). The height heterogeneity in a mixed population of wild-type shed holdfasts is in agreement with the results of a previous AFM analysis (10), and we show here that while the height distribution in a ΔhfsF ΔhfaB population is even greater than that in a ΔhfaB population, the majority of ΔhfsF holdfasts are significantly smaller. To further examine the hfsF deletion phenotype, we determined the amount of holdfast polysaccharide by using a dot blot assay, which indicated that the ΔhfsF mutant had severely reduced holdfast production relative to that of the wild type (Fig. 2B). Figure 2C shows that the ΔhfsF mutant synthesized small holdfasts that remained attached to cells (Fig. 2C).
The above results indicate that a ΔhfsF mutant has decreased holdfast production, which may cause the reduction in adherence. We hypothesized that the defects of a ΔhfsF mutant and a holdfast anchor mutation would be additive. To test this, we constructed in-frame deletion double mutants of hfsF and hfaA, hfaB, or hfaD. The ability of the three double mutants to adhere to a polystyrene surface was assessed in a short-term binding assay which quantifies cells bound to polystyrene by using crystal violet staining. Each mutant was severely deficient for surface binding (Fig. 3A). Introduction of hfsF at the xylX locus in the ΔhfsF ΔhfaA and ΔhfsF ΔhfaD mutants restored adherence to the level of the Δhfa mutant, but its introduction into the ΔhfsF ΔhfaB mutant did not (Fig. 3B). A reconstructed ΔhfaB ΔhfsF double mutant also could not be complemented, ruling out the possibility that a second site mutation in the original strain prevented complementation (Fig. 3B). The pUJhfsEFGH plasmid, containing the full hfsEFGH region under the control of a xylose-inducible promoter, did not complement the ΔhfsF ΔhfaB mutant, indicating that the lack of complementation was not due to a polar effect of the ΔhfsF mutation (Fig. S2). Introduction of a plasmid harboring each hfa gene into the respective ΔhfsF Δhfa double mutants restored adherence to the level of the ΔhfsF mutant in all three double mutants (Fig. 3C). These data suggest that the loss of hfaB has an unknown effect on HfsF activity.
Short-term adherence assay of ΔhfsF, ΔhfaA, ΔhfaB, and ΔhfaD deletion combinations and complementing plasmids. The relative percent binding is normalized to the binding of CB15. The assays were performed in triplicate for at least three independent cultures. The mean and SE are shown for each strain. Empty vectors (ev) were used as controls. (A) ΔhfaA ΔhfsF, ΔhfaB ΔhfsF, and ΔhfaD ΔhfsF double mutants. (B) Complementation of the ΔhfsF deletion in double mutants by use of pKaS112. pXCHYN-2 was the empty vector (ev) control. Complementation was done with the addition of 0.1% xylose for hfsF induction. (C) Complementation of double mutants with pMR20hfaA, pMR20hfaB, or pMR20hfaD. pMR20 was the empty vector control. NA1000, which has a point mutation in hfsA, was used as a holdfast-negative control.
Holdfast size and shedding of the double mutants were assessed with the coverslip lectin binding assay. For each mutant, significant areas of the coverslip were covered with holdfasts but not cells, indicating a holdfast shedding defect (Fig. 4). In addition, the holdfasts were small, similar to those of the ΔhfsF mutant (Fig. 4). Thus, the combination of the reduced holdfast polysaccharide produced by the ΔhfsF mutant and the holdfast shedding phenotype of the holdfast anchor mutants resulted in severely reduced adherence.
Holdfast shedding assay and holdfast quantification for the ΔhfsF mutant combined with different anchor gene mutations. Cell panels are phase-contrast images of cells, and holdfast panels are fluorescence images of WGA-Alexa Fluor 488 lectin bound to the holdfasts. All images show representative areas of a glass coverslip that was submerged in a culture of each strain, washed, stained, and imaged. Each image was acquired and processed in the same manner.
Identification of suppressors that restore holdfast anchoring, but not holdfast synthesis, to a ΔhfsF ΔhfaD mutant.The severe phenotype of the ΔhfsF ΔhfaD double mutant provided a sensitized background for screening for suppressors of the holdfast anchoring defect caused by the hfa mutations and the holdfast synthesis defect of the ΔhfsF mutant. Spontaneous suppressors were obtained by enrichment for adherent cells as described in Materials and Methods. We isolated 14 independent ΔhfsF ΔhfaD spontaneous suppressors, and suppressor S4 was chosen for initial characterization. Suppressor S4 had ∼31% adherence, compared to 5% for the ΔhfsFΔ hfaD mutant and 100% for the wild type (Fig. 5A; Table S3).
Short-term adherence assay of CB15 ΔhfaD ΔhfsF suppressors S4, S9, and S7. The relative percent binding is normalized to the binding of CB15. The assay was performed in triplicate for at least three independent cultures. The mean and SE are shown for each strain. (A) Data for the S4 suppressor, CB15 (wild type), NA1000 (holdfast negative), CB15 ΔhfaD, CB15 ΔhfsF, CB15 ΔwbqV, CB15 ΔhfaD ΔhfsF, and CB15 ΔhfaD ΔhfsF S4. (B) Adherence of the CB15 ΔhfaD ΔhfsF S9 suppressor, CB15 (wild type), and CB15 ΔhfsDAB (holdfast negative). (C) The ΔhfaD ΔhfsF rmlAC751T CC_1108::pLW93 mutant has a kanamycin resistance plasmid linked to rmlA. The ΔhfaD ΔhfsF rmlAWT mutant has the rmlAC751T suppressor mutation repaired to the parental genotype. CB15 rmlAC751T CC_1108::pLW93 has the rmlAC751T mutation in the CB15 background. The ΔhfaD ΔhfsF rmlAC751T* strain has the rmlAC751T suppressor mutation transduced into the clean background of the ΔhfaD ΔhfsF mutant.
We used Illumina sequencing to map the S4 suppressor mutation and identified a single nucleotide polymorphism (SNP) at reference coordinate 3466384 as the only difference between the ΔhfsF ΔhfaD S4 suppressor strain and the ΔhfsF ΔhfaD parent strain (see the supplemental material). This nonsynonymous SNP is located in the CC_3210/CCNA_03316 coding region and results in a single amino acid change, creating the WbqVE428G mutation (Fig. 6A). CC_3210 encodes a 627-amino-acid protein, WbqV, with similarity to proteins that belong to an enzyme family of UDP N-acetylglucosamine C6 dehydratases/C4 reductases. WbqV has sequence similarity to both FlaA1 and WbpM, both of which belong to the same family of dehydratases that catalyze the conversion of UDP-GlcNAc to Qui2NAc, a frequent component of lipopolysaccharide (LPS) (30, 31). WbqV was previously identified as being involved in the production of C. crescentus LPS (32). FlaA1 and WbpM define subfamilies within the dehydratases. WbqV is comparable to WbpM, which has a similar size, contains four N-terminal transmembrane domains, and has an MXXXK catalytic site (30) (Fig. S3 and 6A). In addition, WbqV contains Rossmann-fold NAD(P)H/NAD(P)(+) binding (NADB) and nucleoside-diphosphate sugar epimerase domains (Fig. S3 and 6A). Since many of the other independently derived spontaneous ΔhfsF ΔhfaD binding suppressors had phenotypes similar to that of S4, we sequenced their wbqV loci and found that 8 of the 13 remaining suppressors contained polymorphisms in wbqV (Fig. 6A). Five of them possessed deletion or insertion polymorphisms (DIPs) that resulted in frameshift mutations, while four of them contained nonsynonymous SNPs, suggesting that a loss of function in WbqV suppresses the adherence defect of the ΔhfsF ΔhfaD mutant. WbqV is similar to FlmA (CC_0233), which is predicted to be involved in flagellar glycosylation in C. crescentus but is smaller, like the FlaA1 subfamily of dehydratases (33). While the loss of flmA results in degradation of flagellin in C. crescentus (33, 34), the loss of wbqV did not affect steady-state levels of flagellins (Fig. S4), suggesting that WbqV is not involved in flagellin glycosylation.
Suppressors with wbqV, rfbB, and rmlA mutations. The suppressor number is indicated at the left side of each construct. The number above each construct indicates the codon where the frameshift or SNP occurs. TM, transmembrane. (A) Cartoon of WbqV (CC_3210/CCNA_03316), showing functional domains and indicating wbqV frameshift and SNP mutations and their locations within WbqV. (B) Cartoon of RfbB (CC_3629/CCNA_03744), showing the functional domain and indicating the SNP mutation of rfbB and its location within RfbB. (C) Cartoon of RmlA (CC_1141/CCNA_01199), showing the functional domains and indicating the SNP of rmlA.
We used Illumina sequencing to map the suppressor mutation in one of the remaining ΔhfsF ΔhfaD suppressors, S9, which had 29% adherence versus the 5% adherence of the ΔhfsF ΔhfaD mutant (Fig. 5B; Table S3). We identified a SNP in the rfbB (CC_3629) coding region that caused a nonsynonymous single amino acid change from isoleucine to asparagine at position 328, creating RfbBI328N (Fig. 6B). We sequenced the rfbB genes of the remaining four suppressors and found that none of these strains had a mutation in rfbB. RfbB (CC_3629) is an RmlB homologue that is a predicted dTDP-glucose 4,6-dehydratase that converts dTDP-d-glucose to dTDP-6-deoxy-d-xylo-hexos-4-ulose, a precursor for l-rhamnose (Fig. 6B) (35). RfbB also contains a conserved Rossmann-fold NADB domain and a conserved active site (YXXXK) (Fig. S5). RfbB is a 356-amino-acid protein with no predicted transmembrane domains, unlike WbqV, suggesting that it is probably cytoplasmic.
Finally, we sequenced one of the remaining ΔhfsF ΔhfaD suppressors, S7. Suppressor S7 had ∼40% adherence and 38% lectin binding, compared to 5% and 16% levels for the ΔhfsF ΔhfaD mutant and 100% and 58% levels for the wild type (Fig. 5C; Table S3). We identified a nonsynonymous SNP in rmlA (CC_1141) that resulted in a single amino acid change, at position 251, from proline to serine (RmlAP251S) (Fig. 6C). The region containing the SNP in rmlA was amplified from the remaining three suppressors. Suppressor S14 contained the same mutation as suppressor S7, but suppressors S12 and S2 did not contain mutations within rmlA and remain unidentified (Fig. 6C). RmlA is a predicted short-form glucose-1-phosphate thymidylyltransferase that converts d-glucose 1-phosphate to dTDP-d-glucose. RmlA has a conserved XDP-sugar pyrophosphorylase activator binding site (GXGTRXXPXT) and a conserved sugar binding domain (XEXP) (Fig. S6) (36). The RmlA enzyme is often involved in the first step of dTDP-rhamnose synthesis (37). The mutated and wild-type rmlA alleles were linked to a kanamycin resistance marker (CMS13) for transduction (38). When an rmlA mutant gene encoding a C-to-T change at position 751 (rmlAC751T) was backcrossed into the ΔhfsF ΔhfaD mutant, adherence was restored and was comparable to that of the original suppressor strain (Fig. 5C). When the wild-type rmlA gene was transduced into the S7 suppressor, adherence was strongly deficient and was similar to that of the original ΔhfsF ΔhfaD mutant (Fig. 5C). These results indicate that the rmlAC751T mutation suppressed the adherence defect of the ΔhfsF ΔhfaD double mutant. The rmlAC751T mutation transduced to the wild-type strain caused a moderate decrease in adherence relative to that of the wild type (Fig. 5C), similar to what was seen for the individual ΔrfbB or ΔwbqV null mutant (see below). The rmlA suppressor was not studied further.
Deletion of either wbqV or rfbB partially suppresses the ΔhfsF ΔhfaD phenotype in a holdfast-dependent manner.The above results suggest that null alleles of wbqV suppress the ΔhfsF ΔhfaD adherence defect. Since rfbB is also predicted to be involved in polysaccharide precursor synthesis, we sought to confirm that a loss-of-function mutation in either gene suppresses the ΔhfsF ΔhfaD phenotype by using in-frame deletions. Adherence and lectin binding of ΔwbqV and ΔrfbB mutants were similar to or moderately reduced compared to those of the wild-type strain CB15 (Fig. 7A and B and 8A and B), suggesting that loss of wbqV or rfbB alone does not have a significant impact on surface adhesion or holdfast production. Both the ΔhfsF ΔhfaD ΔwbqV and ΔhfsF ΔhfaD ΔrfbB mutants had increased levels of surface adherence, similar to the original suppressors (Fig. 7A and 8A; Table S3). In addition, both ΔwbqV and ΔrfbB in-frame deletions suppressed the lectin staining defect of the ΔhfsF ΔhfaD mutant, as reflected by the fact that a larger proportion of cells were stained by lectin and therefore harbored a holdfast (Fig. 7B and 8B). Finally, the combination of ΔwbqV and ΔrfbB deletions did not have an additive effect on adherence and lectin binding in either the wild-type or ΔhfsF ΔhfaD background (Fig. S7 and Table S3).
Short-term adherence and lectin binding assays of ΔwbqV deletion combinations. For short-term binding assays, the relative percent binding is normalized to the binding of CB15, and the assay was performed in triplicate for at least three independent cultures. The mean ± SE is shown for each strain. For lectin binding, the assay was performed with at least two independent samples. The percent lectin binding was calculated as the percentage of predivisional cells that had a holdfast. The mean ± SE is shown for each sample. (A) Short-term binding of ΔwbqV triple and quadruple mutant combinations. (B) Lectin binding of ΔwbqV triple and quadruple mutant combinations. (C) Short-term binding of ΔwbqV double mutant and ΔhfaA ΔhfaD ΔwbqV triple mutant combinations. (D) Lectin binding of ΔwbqV double mutant and ΔhfaA ΔhfaD ΔwbqV triple mutant combinations.
Short-term adherence and lectin binding assays of ΔrfbB deletion combinations. For short-term binding assays, the relative percent binding is normalized to the binding of CB15, and the assay was performed in triplicate for at least three independent cultures. The mean ± SE is shown for each strain. For lectin binding, the assay was performed with at least two independent samples. The percent lectin binding was calculated as the percentage of predivisional cells that had a holdfast. The mean ± SE is shown for each strain. (A) Short-term binding of ΔrfbB triple and quadruple mutant combinations. (B) Lectin binding of ΔrfbB triple and quadruple mutant combinations. (C) Short-term binding of ΔrfbB double and ΔhfaA ΔhfaD ΔrfbB triple mutant combinations. (D) Lectin binding of ΔrfbB double and ΔhfaA ΔhfaD ΔrfbB triple mutant combinations.
Complementation analysis was performed to confirm that the suppression phenotypes of the ΔwbqV and ΔrfbB mutants were not due to polar effects. Plasmids that contained either wbqV or rfbB expressed from a xylose-inducible promoter were introduced into the ΔhfsF ΔhfaD ΔwbqV and ΔhfsF ΔhfaD ΔrfbB mutants. In both cases, the complemented triple mutants phenocopied the ΔhfsF ΔhfaD double mutant for adherence and lectin binding (Fig. S8A and B), indicating that the ΔwbqV and ΔrfbB deletions did not have polar effects on downstream genes. Taken together, the above results suggest that the ΔwbqV and ΔrfbB mutations suppress the holdfast shedding but not the reduced amount of holdfast produced by the ΔhfsF ΔhfaD mutant.
To determine if the loss of wbqV or rfbB affected the amount of holdfast produced, holdfast anchoring, or both, we examined the ΔhfsF ΔhfaD ΔwbqV and ΔhfsF ΔhfaD ΔrfbB mutants in a holdfast shedding assay (Fig. 9). In each case, the area of the coverslip with holdfasts and cells was significantly increased, indicating that the loss of wbqV or rfbB decreased the amount of holdfast shedding of the ΔhfsF ΔhfaD mutant. However, the fluorescence intensity of holdfasts produced by the triple mutants was weak and similar to that of the ΔhfsF single mutant, indicating that the ΔwbqV and ΔrfbB mutations did not suppress the holdfast synthesis defect of the ΔhfsF mutant (Fig. 2C; Fig. S1A).
Holdfast shedding assay and holdfast quantitation for wbqV or rfbB deletion combined with ΔhfaD and ΔhfsF mutations. Cell panels are phase-contrast images of cells, and holdfast panels are fluorescence images of WGA-Alexa Fluor 488 lectin bound to the holdfasts. All images show representative areas of a glass coverslip that was submerged in a growing culture of each strain, washed, stained, and imaged. Each image was acquired and processed in the same manner.
To determine if suppression of the ΔhfsF ΔhfaD mutant was dependent on holdfast synthesis, we deleted either hfsA, which is predicted to function in holdfast secretion (16), or hfsG, which encodes a predicted glycosyltransferase involved in holdfast biosynthesis (20). The addition of either the ΔhfsA or ΔhfsG deletion to the ΔhfsF ΔhfaD ΔwbqV and ΔhfsF ΔhfaD ΔrfbB triple mutants resulted in an almost complete loss of adherence and a complete loss of lectin binding, similar to the phenotype of a holdfast null mutant, indicating that suppression by the ΔwbqV and ΔrfbB deletions is holdfast dependent (Fig. 7A and B and 8A and B).
ΔwbqV and ΔrfbB deletions restore holdfast function to ΔhfaA and ΔhfaD mutants but not to the ΔhfsF mutant.The above results suggested that suppression of the ΔhfsF ΔhfaD mutant by deletion of wbqV or rfbB occurred by suppression of the holdfast anchoring defect resulting from the ΔhfaD mutation. To test this, we generated a series of double mutants by combining the mutation in the ΔwbqV or ΔrfbB mutant with individual in-frame deletions of the holdfast anchor genes, hfaA and hfaD, or the polysaccharide flippase gene, hfsF. Deletion of wbqV or rfbB combined with either the ΔhfaA or ΔhfaD mutation increased adherence and lectin binding, but adherence and lectin binding were only minimally affected when the deletion of wbqV or rfbB was combined with the ΔhfsF mutation (Fig. 7C and D and 8C and D). These data indicate that suppression of the ΔhfsF ΔhfaD double mutant phenotype by loss of wbqV or rfbB results from restoration of holdfast anchoring but not holdfast biosynthesis.
The above results were confirmed when we examined combinations of wbqV or rfbB deletion and deletion of either hfaA, hfaD, or hfsF in coverslip assays and analyzed the amount of holdfast produced and the level of holdfast shedding (Fig. 10). As with short-term binding and lectin binding, deletion of wbqV or rfbB in combination with the ΔhfaA or ΔhfaD mutation increased the number of cells bound to the coverslips and reduced areas with holdfasts but no cells, indicating a decrease in holdfast shedding (Fig. 10). The combination of the holdfast flippase ΔhfsF mutant with either the ΔwbqV or ΔrfbB mutation did not alter the number of cells binding to the coverslip, nor did it change the amount of holdfast produced (Fig. 10). These data indicate that the increase in adherence of the ΔhfsF ΔhfaD mutant by the addition of the ΔwbqV or ΔrfbB mutation is mediated through restoration of holdfast anchoring, not by an increase in holdfast production.
Holdfast shedding assay and fluorescent-lectin binding assay to examine holdfast shedding and holdfast amounts in ΔwbqV or ΔrfbB double mutants. Cell panels are phase-contrast images of cells, and holdfast panels are fluorescence images of WGA-Alexa Fluor 488 lectin bound to the holdfasts. All images show representative areas of a glass coverslip that was incubated with a growing culture of each strain, washed, stained, and imaged. Each image was acquired and processed in the same manner.
ΔwbqV and ΔrfbB cannot suppress the holdfast shedding defect of an hfaB mutant.HfaB is a lipoprotein required for the secretion and localization of both HfaA and HfaD. In a ΔhfaB mutant, HfaA and HfaD are not targeted to their normal outer membrane and polar locations but instead are localized to the periplasm, where they are degraded (18). We tested whether HfaB function was required for suppression of holdfast shedding defects. When either a ΔwbqV or ΔrfbB mutation was combined with the ΔhfaB mutation, there was no increase in adherence or lectin binding (Fig. 7C and D and 8C and D) and no rescue of the holdfast shedding phenotype (Fig. 10), indicating that HfaB is required for suppression of holdfast shedding by ΔwbqV and ΔrfbB mutations.
Currently, the only known function of HfaB is to target HfaA and HfaD to the outer membrane. HfaB is still expressed and localized normally in the outer membrane in a ΔhfaA ΔhfaD double mutant (18). We next tested whether the requirement of HfaB for suppression of holdfast shedding by ΔwbqV and ΔrfbB mutations was due to its function in HfaA and HfaD targeting by assessing the effect of wbqV or rfbB deletion in a ΔhfaA ΔhfaD double mutant. In each case, adherence and lectin binding were increased relative to those of the ΔhfaA ΔhfaD mutant (Fig. 7C and D and 8C and D). This was also confirmed by the coverslip assay. The triple mutants demonstrated an increase in the number of cells present on the coverslips and a decrease in areas with holdfasts but no cells, indicating a reduction in holdfast shedding (Fig. 11). These results indicate that suppression of holdfast shedding by ΔwbqV and ΔrfbB mutations requires an unknown function of HfaB that is independent of HfaA and HfaD localization. We hypothesize that HfaB either plays a more fundamental role in anchoring of the holdfast polysaccharide or secretes an as yet unknown protein important for the anchoring process. In addition, the complexities associated with complementation of the ΔhfsF mutation in the ΔhfaB background may contribute to the inability of wbqV or rfbB mutation to suppress this double mutant.
Holdfast shedding assays to test suppression of hfaA hfaD double mutant phenotypes by wbqV and rfbB mutations. Cell panels are phase-contrast images of cells, and holdfast panels are fluorescence images of WGA-Alexa Fluor 488 lectin bound to the holdfasts. All images show representative areas of a glass coverslip that was incubated with a growing culture of each strain, washed, stained, and imaged. Each image was acquired and processed in the same manner.
Loss of wbqV and rfbB affects anchoring of the surface array protein.The surface array protein (RsaA) of C. crescentus is anchored to the bacterial cell via smooth LPS (SLPS) (32). Mutations within the LPS structure result in shedding of RsaA into the culture medium (32). Because rfbB is predicted to be involved in polysaccharide biosynthesis, one possibility is that it is involved in the biosynthesis of LPS. Previous work by Awram and Smit described wbqV as being involved in LPS core sugar biosynthesis and showed that a wbqV mutant affects SLPS levels (32). To examine the possibility that loss of wbqV and rfbB affects LPS biosynthesis, we performed Western blots with an anti-RsaA antibody to determine if wbqV or rfbB deletion resulted in shedding of the surface array protein into the culture medium (Fig. 12). Both cells and culture supernatant were examined for the presence of RsaA. The loss of rfbB alone or in combination with deletion of hfsF and hfaD resulted in shedding of RsaA. There was no shedding of RsaA detected in the ΔwbqV mutant alone, slight RsaA shedding in the ΔhfsF ΔhfaD double mutant, and significant RsaA shedding in the ΔwbqV ΔhfsF ΔhfaD triple mutant. The shedding of the surface array protein into the supernatant suggests that deletion of wbqV or rfbB affects the structure of Caulobacter LPS.
Effects of ΔrfbB and ΔwbqV mutations on surface array shedding. Western blots were performed with whole cells (C) and culture supernatants (S). Samples were normalized by use of the OD600, loaded onto a 10% SDS-PAGE gel, transferred to nitrocellulose, and incubated with anti-RsaA antibody. (A) CB15 and the ΔhfaD, ΔhfsF, and ΔwbqV mutants. (B) ΔrfbB, ΔhfaD ΔhfsF, ΔhfaD ΔhfsF ΔwbqV, and ΔhfaD ΔhfsF ΔrfbB mutants.
LPS is important for the adherence of some bacteria to both abiotic and biotic surfaces (39–41). Because LPS was exposed by loss of the S-layer in several of the mutants, it was important to determine if the changes in adherence, holdfast shedding, and lectin binding were holdfast mediated or facilitated by the exposure of LPS. To examine this possibility, the ΔhfaA and ΔhfaD anchor mutants were combined with ΔrsaA mutation. The loss of rsaA increased the overall adherence and numbers of cells bound in the coverslip assay for both ΔhfaA and ΔhfaD mutants (Fig. 13 and 14A). However, lectin binding of the ΔhfaA ΔrsaA and ΔhfaD ΔrsaA mutants was not increased relative to that of the ΔhfaA and ΔhfaD single mutants, despite the fact that lectin binding was increased in the ΔrsaA mutant relative to that in CB15 (Fig. 14B). These data suggest that while the loss of rsaA in conjunction with either the ΔhfaA or ΔhfaD mutation resulted in an overall increase in bacterial cell attachment, this increase was not holdfast mediated, because there was no corresponding increase in cell-associated holdfasts in the ΔrsaA double mutants. In contrast, the addition of a ΔrfbB or ΔwbqV mutation to a ΔhfaA, ΔhfaD, or ΔhfaA ΔhfaD mutant resulted in an increase in lectin binding and adherence and a decrease in holdfast shedding, and these changes were holdfast dependent (see the previous section) (Fig. 7C and D, 8C and D, and 11). It is unclear why the ΔrsaA mutant had an increased number of cells with holdfasts as indicated by lectin binding. Adherence was lost in a ΔrsaA ΔhfsA mutant (Fig. 14C), indicating that the increased adherence seen with loss of the surface array protein cannot overcome the loss of a holdfast. Therefore, loss of the surface array protein is clearly not responsible for the increased holdfast anchoring associated with the ΔrfbB and ΔwbqV mutants.
Effects of ΔrsaA mutation on holdfast shedding. The holdfast shedding assay was performed on ΔrsaA mutants. Cell panels are phase-contrast images of cells, and holdfast panels are fluorescence images of WGA-Alexa Fluor 488 lectin bound to the holdfasts. All images show representative areas of a glass coverslip or slide well that was incubated with a growing culture of each strain, washed, stained, and imaged. Each image was acquired and processed in the same manner.
Effects of ΔrsaA mutation on adherence and lectin binding. (A and C) For short-term binding assay of ΔrsaA mutants, the relative percent binding is normalized to the binding of CB15, and data are for assays performed in triplicate for at least three independent cultures. The mean and SE are shown for each strain. (B) For lectin binding assay of ΔrsaA mutants, the assay was performed with at least three independent cultures. The percent lectin binding was calculated as the percentage of predivisional cells that had a holdfast. The mean and SE are shown for each strain.
LPS may be altered in wbqV, rfbB, and rmlA mutants.The shedding of the S-layer in several of the combination mutants indicated that the LPS may be altered in these strains. LPS was isolated using a modified version of the method of Hitchcock and Brown (42) and analyzed in a Tris-Tricine gel by silver staining as described in Materials and Methods (Fig. 15). There was a slight decrease in the apparent molecular weights of the LPS bands for the ΔrfbB single mutant and the ΔhfaD ΔhfsF ΔwbqV, ΔhfaD ΔhfsF ΔrfbB, and ΔhfaD ΔhfsF rmlAC751T triple mutants, suggesting that there was an alteration in the core LPS structure. The surface array protein is associated with SLPS, which is only 10% of the total LPS (43). However, an alteration in the core sugars of the LPS would affect the SLPS, which is anchored via the core LPS structure, resulting in an alteration of surface array anchoring. The alteration of LPS in these three mutants also corresponds to shedding of the surface array protein from the cell (Fig. 12).
LPS of mutant combinations in a silver-stained Tris-Tricine gel. LPS was isolated by the method of Hitchcock and Brown (42), and the equivalent of 2.5 μg of LPS was run in a 16% Tris-Tricine gel. The gel was stained with silver stain. Five microliters of Novex Sharp prestained standard (M) was run as a reference.
DISCUSSION
In this work, we used the synthetic effect of combining an hfaD (holdfast anchor) deletion with an hfsF (holdfast polysaccharide flippase) deletion to generate a sensitized background for a suppressor screen. Our data indicate that the loss of holdfast anchoring conferred by the hfaD mutation combined with the reduced holdfast production conferred by the hfsF mutation resulted in the complete loss of adherence of the ΔhfsF ΔhfaD double mutant. Similarly, the combination of hfsF deletion with hfaA and hfaB holdfast anchor mutations resulted in abolished adherence. Interestingly, all 10 suppressor mutants restoring adherence to the ΔhfsF ΔhfaD double mutant that were analyzed had restored holdfast anchoring but still exhibited the low holdfast synthesis efficiency of the hfsF mutant. wbqV and rfbB mutations were also able to suppress the holdfast anchoring defect of a ΔhfaA mutant. The HfaA anchor protein has sequence similarity to and biochemical properties of functional amyloid proteins. HfaD is required for the stability of HfaA multimers, and vice versa (18). However, restoration of holdfast anchoring in the ΔhfsF ΔhfaD double mutant does not occur through the restoration of HfaA multimerization, because we found that the loss of wbqV or rfbB did not restore the stability of the HfaA multimers in the ΔhfaD background (data not shown).
Why is it harder to isolate suppressors of the holdfast synthesis defect of an hfsF mutant than those of the anchoring defect of the hfaD mutation? HfsF is a predicted Wzx flippase involved in translocation of the holdfast polysaccharide repeat units across the inner membrane (20). In Gram-negative bacteria, polysaccharides can be translocated across the inner membrane by either a Wzx flippase or an ABC transporter. Previous studies demonstrated that both Wzx flippases and ABC transport systems can translocate polysaccharides from other biosynthesis systems, albeit less efficiently (44). Increasing holdfast synthesis in the hfsF mutant would likely require a gain of function in a compensating flippase or ABC transporter, which would be much less frequent than the loss-of-function mutations in sugar precursor pathways that suppress the anchoring defect conferred by the hfaD mutation. Furthermore, holdfast precursors would likely still compete with the natural substrate of the compensating flippase or ABC transporter, making it even more difficult to increase holdfast precursor flipping.
In contrast, the nature of the suppressor genes and the resulting suppression of the anchoring defect of hfaA and hfaD mutants suggest why it is easier to isolate such suppressors through loss of function. The sequences of RmlA, RfbB, and WbqV suggest that they are part of the lipopolysaccharide or glycosylation biosynthesis pathway. There are numerous polysaccharide biosynthesis and utilization pathways, including those for lipopolysaccharide, peptidoglycan, capsule, and other extracellular polysaccharides. The pools of polysaccharide precursors used by these pathways often overlap. Therefore, the alteration of any pathway will likely have effects on the other pathways through modulation of the precursor pool levels. Loss of wbqV could result in an increase in nucleotide sugars, such as UDP-GlcNAc or UDP-glucose (UDP-Glc). Similarly, loss of rfbB would result in increased pools of UDP-Glc, and loss of rmlA would result in an increase of glucose-1-phosphate. Changes in nucleotide sugar pools may alter the composition or structure of the holdfast and increase its interaction with some other component of the cell envelope. Indeed, since ΔwbqV and rfbB mutations did not significantly affect holdfast synthesis or adherence in a wild-type strain background, it is unlikely that they play a direct role in the production of holdfasts. Since many different types of mutations can shift the pools of relevant nucleotide sugars, obtaining such mutations should be relatively frequent, as exemplified by the three suppressors we identified.
A link between holdfast anchoring and LPS might be suggested by the fact that combinations of mutants containing either a wbqV or rfbB mutation result in shedding of RsaA, which is anchored to the cell surface via the SLPS in C. crescentus (32). Both rmlA and rfbB are predicted to be involved in biosynthesis of rhamnose, which is a component of C. crescentus LPS (45). In addition, a wbqV mutant was previously shown to alter levels of SLPS in C. crescentus (32). However, while loss of the surface layer, which exposes the LPS, increases attachment in a wild-type background, the increased adherence of the suppressors is holdfast dependent, indicating that suppression does not occur solely by exposure of LPS through shedding of the surface layer. Indeed, as demonstrated by lectin binding levels, the ΔhfaA ΔrsaA and ΔhfaD ΔrsaA mutants did not exhibit the increase in cell-associated holdfasts seen for the suppressors. An alternative possibility is that shedding of the surface layer in the suppressors is a consequence of an altered structure of the LPS, which may facilitate modified anchoring of the holdfast at the cell pole. Examination of the lipopolysaccharide profiles of the ΔrfbB and ΔwbqV mutant combinations by SDS-PAGE and silver staining indicated a possible structural change to the LPS, as seen by faster migration of the LPS for these samples.
One of the most interesting observations from this work is the finding that HfaB may have an additional function in holdfast anchoring beyond the export of HfaA and HfaD. This is supported by the fact that neither the loss of wbqV nor that of rfbB could suppress a ΔhfaB mutant but could suppress a ΔhfaA ΔhfaD double mutant. Indeed, previous work showed that an hfaB mutant has more severe holdfast shedding and adherence phenotypes than an hfaA hfaD double mutant (18). HfaB may interact with the holdfast directly and facilitate anchoring. Alternatively, or in addition, HfaB may be required for secretion of an additional protein involved in holdfast anchoring. Examination of the chromosomal region surrounding hfaB did not reveal any obvious candidates for a protein secreted via HfaB. In addition, the inability to complement the hfsF deletion in the ΔhfsF ΔhfaB mutant suggests that the loss of HfaB has an effect on HfsF. HfaB and HfsF may interact directly or indirectly, as putative unknown proteins secreted via HfaB may interact with HfsF. Future studies will seek to identify other proteins secreted via HfaB and to elucidate the relationship between HfaB and HfsF, which may be important for communication between the holdfast biosynthesis, secretion, and anchoring systems in Caulobacter.
MATERIALS AND METHODS
Bacterial strains and growth conditions.Bacterial strains and plasmids used in this study are listed in Table S1 in the supplemental material. C. crescentus strains were grown in peptone-yeast extract (PYE) medium (46) at 30°C, with antibiotic and carbon supplements at the following concentrations, where necessary: kanamycin, 20 μg/ml (plate) and 5 μg/ml (broth); chloramphenicol, 1 μg/ml (plate and broth); nalidixic acid, 20 μg/ml (plate); tetracycline, 2 μg/ml (plate) and 1 μg/ml (broth); glucose, 0.3% (liquid) and 0.1% (plate); and xylose, 0.1 or 0.03% (liquid only). Escherichia coli strains were cultured at 37°C in Luria-Bertani (LB) medium. LB medium was supplemented with kanamycin (50 μg/ml or 25 μg/ml [plate] and 30 μg/ml [broth]), chloramphenicol (30 μg/ml [plate and broth]), and tetracycline (12.5 μg/ml [plate and broth]) where necessary.
DNA manipulations and sequencing.Restriction enzymes used for standard molecular cloning in this study were purchased from New England BioLabs Inc. (Ipswich, MA). All primers used in this study are listed in Table S2 and were purchased from Invitrogen, Grand Island, NY. Plasmid DNA was isolated using a Zyppy plasmid miniprep kit (Zymo Research Corp., Irvine, CA), and PCR products were purified using QIAquick spin columns (Qiagen, Valencia, CA) following procedures recommended by the manufacturer. Chromosomal DNA was isolated using the Promega Magic MiniPrep DNA purification system (Promega, Madison, WI) according to the manufacturer's instructions. Sequencing reactions were performed at the Indiana Institute for Molecular Biology at Indiana University, using an Applied Biosystems 3730 automated fluorescence sequencing system and an ABI Prism BigDye Terminator v. 3.1.1 cycle sequencing kit (Applied Biosystems, Foster City, CA). Sequence data were analyzed using Sequencher 4.8 software (Gene Codes Corporation, Ann Arbor, MI). Illumina genome sequencing was performed on an Illumina HiSeq2000 sequencer at the Tufts University Core Facility, using the massively parallel sequencing protocol (47).
Short-term and long-term surface binding and holdfast shedding assays.Short-term polystyrene binding assays were performed by incubating cells in 12-well polystyrene plates for 45 min, washing the cells, and staining them with 0.1% crystal violet as previously described (20). Long-term binding assays were performed by incubating cells for 24 h at 30°C in 12-well polystyrene plates, with a plastic coverslip inserted vertically into each well. Plastic coverslips were washed and stained with 0.1% crystal violet as previously described (18).
Holdfast shedding assays were performed by growing cells overnight in PYE at 30°C and diluting them to an optical density at 600 nm (OD600) of 0.15. Cells were incubated with glass coverslips or on 12-well slides for 4 to 5 h at 30°C, washed, and stained with fluorescent lectin to visualize holdfasts. Cells were imaged as described below.
Construction of single and combination deletion mutants.All deletions were generated in two steps by homologous recombination, using upstream and downstream fragments of a gene cloned into nonreplicating plasmid pNPTS138 or -9, carrying a kanamycin resistance gene cassette (nptI) along with the sacB cassette, which confers sucrose sensitivity, as previously described (48). The deletion mutants were confirmed by colony PCR using the primers used to clone the upstream and downstream fragments and were verified by sequencing.
For the wbqV (CC_3210/CCNA_03316) deletion, the Fup3210Eco, Rup3210Xba, Fdw3210Xba, and Rdw3210Hind primers were used to PCR amplify 500-bp DNA fragments directly upstream and downstream of the CC_3210 gene, leaving four codons into the annotated start of wbqV and seven codons from the end of the gene. For the rfbB (CC_3629/CCNA_03744) deletion, the rfbBupF, rfbBupR, rfbBdwnF, and rfbBdwnR primers were used to PCR amplify 500-bp DNA fragments directly upstream and downstream of the CC_3629 gene, removing the start codon and leaving the last 27 codons, which overlap the predicted start of CC_3628.
For the double deletion of hfsF and hfsG, the NewFuphfsF, NewRuphfsF, DwnHfsGHindF, and DwnHfsGBamR primers were used to PCR amplify 500-bp DNA fragments directly upstream of hfsF and downstream of hfsG, leaving 64 codons after the annotated start of hfsF to preserve a predicted promoter for hfsG and hfsH and 11 codons from the end of hfsG. The upstream product was digested with EcoRI and BamHI, and the downstream PCR product was digested with BamHI and HindIII. Both digested DNA fragments were cloned into pNPTS139 digested with EcoRI and HindIII. The ΔhfsF ΔhfsG strain was created by introducing pNPTS139ΔhfsFhfsG into CB15. A ΔhfsF ΔhfaD double mutant was created by introducing pNPTS138ΔhfsF (20) into CB15 ΔhfaD (18) by conjugation, giving CB15 ΔhfsF ΔhfaD. The ΔhfsF ΔhfaA and ΔhfsF ΔhfaB mutants were created essentially in the same way, by introducing pNPTS138ΔhfsF into CB15 ΔhfaA and CB15 ΔhfaB. The ΔhfaB ΔwbqV, ΔhfaD ΔwbqV, and ΔhfsF ΔwbqV mutants were created by introduction of pNPTS139ΔwbqV into CB15 ΔhfaB, CB15 ΔhfaD, and CB15 ΔhfsF. The ΔhfaA ΔwbqV deletion mutant was generated by introduction of pNPTS138ΔhfaA into CB15 ΔwbqV. The ΔhfaA ΔrfbB, ΔhfaB ΔrfbB, ΔhfaD ΔrfbB, and ΔhfsF ΔrfbB mutants were generated by introduction of pNPTS139ΔrfbB into CB15 ΔhfaA, CB15 ΔhfaB, CB15 ΔhfaD, and CB15 ΔhfsF. The ΔwbqV ΔrfbB mutant was created by introduction of pNPTS139ΔrfbB into CB15 ΔwbqV. The ΔhfaA ΔrsaA and ΔhfaD ΔrsaA double mutants were created by introduction of pNPTS138ΔhfaA or pNPTS138ΔhfaD into CB15 ΔrsaA. The CB15 ΔrsaA ΔhfsA double mutant was created by introduction of pNPTS138ΔhfsA into CB15 ΔrsaA.
The ΔhfsF ΔhfaD ΔwbqV and ΔhfsF ΔhfaD ΔrfbB triple mutants were created by introducing pNPTS139ΔwbqV or pNPTS139ΔrfbB into CB15 ΔhfsF ΔhfaD. The ΔhfaA ΔhfaD ΔwbqV and ΔhfaA ΔhfaD ΔrfbB triple mutants were created by introducing pNPTS139ΔwbqV or pNPTS139ΔrfbB into CB15 ΔhfaA ΔhfaD. The ΔhfaA ΔrfbB ΔhfsA triple mutant was created by introduction of pNPTS138ΔhfsA into CB15 ΔhfaA ΔrfbB. The ΔhfaA ΔhfaD ΔrsaA triple mutant was generated by introduction of pNPTS138ΔhfaD into CB15 ΔhfaA ΔrsaA.
The ΔhfsF ΔhfsG ΔhfaD ΔwbqV and ΔhfsF ΔhfsG ΔhfaD ΔrfbB quadruple mutants were created first by introduction of pNPTS139ΔhfsFΔhfsG into CB15 ΔhfaD followed by the introduction of either pNPTS139ΔwbqV or pNPTS139ΔrfbB into CB15 ΔhfsF ΔhfsG ΔhfaD. The ΔhfsF ΔhfaD ΔwbqV ΔhfsA and ΔhfsF ΔhfaD ΔrfbB ΔhfsA quadruple mutants were generated by the introduction of pNPTS138ΔhfsA into either CB15 ΔhfsF ΔhfaD ΔwbqV or CB15 ΔhfsF ΔhfaD ΔrfbB. The CB15 ΔhfsF ΔhfaD ΔwbqV ΔrfbB quadruple mutant was generated by introduction of pNPTS138ΔrfbB into CB15 ΔhfsF ΔhfaD ΔwbqV.
Isolation of suppressors of an hfsF hfaD mutant.We obtained suppressors of the ΔhfsF ΔhfaD mutant that restored binding by serial enrichment for increased adherence. Fifteen single colonies of the ΔhfsF ΔhfaD mutant were inoculated into separate culture tubes and cultured overnight at 30°C. The next morning, the liquid cultures were poured out and the culture tubes were rinsed by vortexing twice with fresh PYE medium before fresh medium was added back to the culture tubes, which were then incubated overnight. The cultures were processed in this fashion for 10 days, and cells were screened visually under the microscope for signs of rosette formation, indicating cells bound together in a group by their holdfasts. Subsequently, aliquots of any cultures with positive rosette-forming cells were plated on PYE plates. We then scored single colonies by fluorescent-lectin holdfast staining.
DNA preparation and adapter ligation for Illumina sequencing.Suppressor mutant genomic DNA (30 μg) was added to shearing buffer (40% glycerol, 10 mM Tris, pH 8, and 1 mM EDTA) in a total final volume of 800 μl and nebulized at room temperature for 4 min at 40 lb/in2 with argon. Three microliters of 30 μg/μl dextran was added, along with 1/10 the total volume of potassium acetate solution (3 M potassium acetate, 6 M glacial acetic acid), and the solution was vortexed. One volume of isopropanol was then added, and solution was vortexed and placed in a dry ice-ethanol bath for 5 min. Sheared genomic DNA was then concentrated to 333 ng/μl. Five micrograms or 15 μl of sheared genomic DNA was then blunted and 3′-A additions made using an NEB quick blunting kit and Klenow fragment (exo-), respectively. The adapter ligation was performed using an NEB quick ligation kit, with a reaction mixture containing template DNA, 50 μM IL-1/IL-2 adapter mix (OLJ131/OLJ132), 2× quick ligase buffer, and quick ligase. Ligated samples were then run in a 2% agarose gel, size selection was performed, and DNA was extracted using a Qiagen QIAquick gel extraction kit. The sequencing DNA library was then prepared following 15 cycles of PCR using oligonucleotides IL-3 and IL-4 and a Stratagene Easy A PCR kit. A NanoDrop spectrophotometer (Thermo Scientific, Wilmington, DE) was then used to quantitate the library concentration before Illumina sequencing.
Complementation and overexpression of wbqV and rfbB.A high-copy-number, xylose-inducible plasmid, pUJ142 (49), harboring either wbqV or rfbB, was constructed to complement ΔwbqV or ΔrfbB mutants. The wbqV gene was first PCR amplified from wild-type CB15 chromosomal DNA by using the primers F3210start and R3210end, digested with EcoRI and KpnI, and cloned into pUJ142. The plasmid carrying wbqV under the control of the xylX inducible promoter was introduced to the ΔwbqV mutant. The rfbB gene was first PCR amplified from wild-type CB15 chromosomal DNA by using the primers rfbBATGEco and rfbBstopHind and then cloned into pUJ142 by use of EcoRI and HindIII. pUJrfbB under the control of the xylX inducible promoter was introduced to the ΔrfbB mutant. The resultant complemented strains were assayed for surface adhesion and fluorescent-lectin holdfast staining. The xylX promoter is leaky and did not require the addition of exogenous xylose to the medium during growth to give sufficient expression of either wbqV or rfbB for complementation.
Transduction and repair of the SNP in rmlA.Transduction was performed as previously described (38). A CMS13 marker plasmid (pLW93) was mated into the ΔhfaD ΔhfsF rmlAC751T mutant. A ΦCR30 lysate of the ΔhfaD ΔhfsF rmlAC751T CC_1108::pLW93 strain was used to transduce kanamycin resistance into the ΔhfaD ΔhfsF mutant and CB15. Transductants were grown on PYE with kanamycin. Chromosomal DNA was isolated from transductants, and a PCR product around the rmlAC751T point mutation was generated by using primers CC1141-225F and CC1141-endR and sequenced to confirm that the mutation was transferred both to CB15 and again into CB15 ΔhfsF ΔhfaD.
To repair the CC_1141 (CCNA_01199) mutation, CB15 ΔhfaD ΔhfsF rmlAC751T was transduced for kanamycin resistance with a ΦCR30 lysate of CB15 that was previously transduced with CC_1108::pLW93. Transductants were selected on PYE with kanamycin. Five transductants were transferred to a fresh plate. Chromosomal DNA was isolated from each transductant, and PCR was used to confirm that the mutation had been repaired.
Holdfast quantitation by dot blotting.Cells were grown overnight in PYE, diluted to an OD600 of 0.15, and allowed to grow to mid-exponential phase (OD600 = 0.3 to 0.5). The OD600 was determined, and for each strain, two 500-μl aliquots were removed. One aliquot was used as a total culture (T). The second sample was centrifuged at 16,600 × g for 5 min. The top 400 μl of medium was saved and used as a supernatant sample (S). The remaining 100 μl of medium was removed from the cells and discarded. The cell pellet was suspended in 500 μl of fresh PYE and used as a cell sample (C). Approximately 50 μl of each sample was diluted in 250 μl of 50 mM Tris, pH 7.6. Samples were diluted to equivalent OD600 values. Twofold dilutions of each sample were made and applied to nitrocellulose by use of a Bio-Rad dot blot apparatus (Bio-Rad, Hercules, CA). The nitrocellulose was washed in 50 mM Tris, pH 7.6, and then in 20 mM Tris, 137 mM NaCl, and 0.05% Tween (TBST). The dot blot was blocked with 3% bovine serum albumin (BSA) in TBST for 1 h and incubated with WGA conjugated to horseradish peroxidase (HRP) at 1:150,000 for 1 h. The blot was washed three times for 5 min each, developed with Dura substrate (Thermo Fisher Scientific, Rockford, IL) for 5 min, and exposed on a Kodak image station (Carestream Health, Rochester, NY).
Fluorescent-lectin holdfast staining and microscopy.Fluorescent-lectin holdfast staining assays were performed as previously described (17), except that Alexa Fluor 488 (green)- or 594 (red)-conjugated WGA was used to label the holdfast (Invitrogen Molecular Probes). Briefly, lectin was used at a working concentration of 0.5 μg/ml. Cells at an OD600 of 0.3 to 0.5 were incubated for 20 min in PYE with lectin and imaged by microscopy on 1% agarose pads made with 1× M2 salts. Lectin binding was calculated by determining the percentage of predivisional cells with lectin binding indicating a holdfast was present. Either a Nikon Eclipse E800 or 90i light microscope equipped with a Nikon FITC-HyQ filter cube (Chroma Technologies) and a 100× Plan Apo oil objective was used for phase-contrast microscopy and epifluorescence microscopy. Images were captured using a Princeton Instruments cooled charge-coupled device (CCD) camera (model 1317) and MetaMorph imaging software v. 7.1.1 (Molecular Devices, Sunnyvale, CA) or Elements imaging software v 4.0 (Nikon).
Fluorescence intensity was determined using ImageJ (50).
Sample preparation, SDS-PAGE, and Western blotting.Cells were grown in PYE overnight at 30°C, and the OD600 was determined. For whole-cell samples, 1 ml of cell culture was centrifuged at 16,600 × g for 5 min. The supernatant was removed, and the cells were suspended in 50 μl of 10 mM Tris, pH 8, and 50 μl of 2× Laemmli sample buffer. For culture supernatant samples, 50 μl of 2× Laemmli sample buffer was added to 50 μl of culture supernatant. Samples were normalized based on equivalent OD600 values. Samples were boiled for 5 min before loading, except for surface array (RsaA) samples, which were incubated at 37°C for 5 min.
Samples were run in SDS-PAGE gels at 150 V and transferred to nitrocellulose for 1 h at 100 V. Blots were blocked with 5% nonfat milk in Tris-buffered saline with 0.05% Tween 20 (TBST) for 1 h at room temperature. Antiflagellin antibody was used at 1:1,000, and anti-RsaA antibody was used at 1:100,000. Blots were washed with TBST three times for 5 min each and incubated with HRP-conjugated goat anti-rabbit antibody (Bio-Rad, Hercules, CA), preadsorbed with C. crescentus NA1000 acetone powder, at 1:20,000 in blocking buffer for 1 h at room temperature. Blots were washed three times for 5 min each and developed with Supersignal West Pico chemiluminescence substrate (Thermo Fisher Scientific) for 5 min. Imaging was performed using either a Kodak 4000MM Pro image station (Carestream Health) or a Bio-Rad Chemidoc MP imager (Bio-Rad).
Tricine–SDS-PAGE gels and silver staining.A 14- by 16-cm Tricine step gradient gel was generated. First, a 16.5% T/3% C separating gel was poured for the bottom of the gel, where T is total acrylamide monomer and C is the percentage of cross-linker given as a ratio of cross-linker to total acrylamide/bisacrylamide concentration. This was followed by a 10% spacer gel and a 4% stacking gel, following the method of Schagger and von Jagow (51). Gels were run at 30 V until the loading dye reached the end of the stacking gel, and then gels were run at 90 V for a total of 8 to 16 h. Gels were silver stained using the method of Tsai and Frasch (52), with a few modifications (53). The gel was fixed twice with 40% ethanol and 5% acetic acid for 30 min each time. The oxidation step was increased to 15 min. Gels were imaged using a Bio-Rad Chemidoc MP imager (Bio-Rad).
LPS purification.LPS was isolated using a modified version of the method of Hitchcock and Brown (42) based on the work of Walker et al. (53). Briefly, the cell pellet from 5 ml of cell culture was washed with 10 mM HEPES, pH 7.2, and suspended in 250 μl of 10 mM Tris, 1 mM EDTA, pH 8. Cells were frozen at −20°C and thawed. One microliter of DNase (0.5 mg/ml), 1 μl of lysozyme (10 mg/ml), and 3 μl of 1 M MgCl2 were added, and the cells were incubated at room temperature for 15 min. Two hundred fifty microliters of SDS-PAGE sample buffer was added, the sample was boiled for 10 min and then cooled to room temperature, and proteinase K was added to a final concentration of 0.5 mg/ml. The cell suspension was incubated at 60°C for 1 h. A sample of LPS was quantitated by using a Purpald assay prior to the addition of SDS-PAGE sample buffer, with 2-keto-3-deoxyoctulosonic acid (KDO) as a standard (54). Samples were boiled for 5 min in LPS sample buffer (0.03 M Tris, pH 6.8, 0.5 mM EDTA, 1% SDS, 7.2% β-mercaptoethanol) and then loaded onto a Tricine gel with 5 μl Novex Sharp prestained protein standard and a lane with only sample buffer containing bromophenol blue.
Holdfast height measurement by atomic force microscopy.Holdfast samples were prepared as described previously (55), with some modifications. C. crescentus ΔhfaB and ΔhfaB ΔhfsF mutants were grown in PYE to an OD600 of 0.3 to 0.4, diluted to an OD600 of 0.1 in fresh PYE, and spotted onto a 10- by 10-mm piece of freshly cleaved mica. Samples were incubated at room temperature in a humid chamber for 4 h. The mica surface was then thoroughly rinsed with sterile distilled water (dH2O) to remove all cells and debris, and 100 μl sterile dH2O was placed on the surface for AFM imaging. AFM topographic images of holdfasts (10 × 10 μm) were obtained using the alternative contact mode on a Cypher AFM (Asylum Research) at 20°C, using Mikromasch cantilevers (HQ:CSC38/Cr-Au, 0.09 N/m, 20 kHz). At least 30 images of three independent replicates were analyzed, and holdfast height was determined using the built-in image analysis function of the AFM software (version 131395; Igor Pro/Asylum Research).
ACKNOWLEDGMENTS
We thank Bogdan Dragnea, Department of Chemistry, Indiana University, for use of his AFM and facilities for analysis of the ΔhfsF holdfasts. We thank John Smit, University of British Columbia, Vancouver, BC, Canada, for kindly providing the antibody to RsaA. We thank Urs Jenal, University of Basel, Basel, Switzerland, for providing the pKaS112 plasmid. We thank former lab member Rey Allen, currently at Arizona State University, Biodesign Program, Tempe, AZ, for creation of the pUJhfsEFGH construct. We thank the members of the Brun lab for helpful scientific discussions and critical readings of the manuscript.
This work was supported by grants GM102841 and R35GM122556 from the National Institutes of Health to Y.V.B. Some of this work was supported by the Indiana METACyt Initiative of Indiana University, funded in part through a major grant from the Lilly Endowment, Inc.
FOOTNOTES
- Received 5 October 2017.
- Accepted 10 November 2017.
- Accepted manuscript posted online 20 November 2017.
- Address correspondence to Yves V. Brun, ybrun{at}indiana.edu.
↵* Present address: Evelyn Toh, Indiana University School of Medicine, Department of Microbiology and Immunology, Indianapolis, Indiana, USA.
Citation Hardy GG, Toh E, Berne C, Brun YV. 2018. Mutations in sugar-nucleotide synthesis genes restore holdfast polysaccharide anchoring to Caulobacter crescentus holdfast anchor mutants. J Bacteriol 200:e00597-17. https://doi.org/10.1128/JB.00597-17.
Supplemental material for this article may be found at https://doi.org/10.1128/JB.00597-17.
REFERENCES
- Copyright © 2018 American Society for Microbiology.
