Cyclic Di-GMP Binding by an Assembly ATPase (PilB2) and Control of Type IV Pilin Polymerization in the Gram-Positive Pathogen Clostridium perfringens

ABSTRACT

The Gram-positive pathogen Clostridium perfringens possesses type IV pili (TFP), which are extracellular fibers that are polymerized from a pool of pilin monomers in the cytoplasmic membrane. Two proteins that are essential for pilus functions are an assembly ATPase (PilB) and an inner membrane core protein (PilC). Two homologues each of PilB and PilC are present in C. perfringens, called PilB1/PilB2 and PilC1/PilC2, respectively, along with four pilin proteins, PilA1 to PilA4. The gene encoding PilA2, which is considered the major pilin based on previous studies, is immediately downstream of the pilB2 and pilC2 genes. Purified PilB2 had ATPase activity, bound zinc, formed hexamers even in the absence of ATP, and bound the second messenger molecule cyclic di-GMP (c-di-GMP). Circular dichroism spectroscopy of purified PilC2 indicated that it retained its predicted degree of alpha-helical secondary structure. Even though no direct interactions between PilB2 and PilC2 could be detected in vivo or in vitro even in the presence of c-di-GMP, high levels of expression of a diguanylate cyclase from C. perfringens (CPE1788) stimulated polymerization of PilA2 in a PilB2- and PilC2-dependent manner. These results suggest that PilB2 activity is controlled by c-di-GMP levels in vivo but that PilB2-PilC2 interactions are either transitory or of low affinity, in contrast to results reported previously from in vivo studies of the PilB1/PilC1 pair in which PilC1 was needed for polar localization of PilB1. This is the first biochemical characterization of a c-di-GMP-dependent assembly ATPase from a Gram-positive bacterium.

IMPORTANCE Type IV pili (TFP) are protein fibers involved in important bacterial functions, including motility, adherence to surfaces and host cells, and natural transformation. All clostridia whose genomes have been sequenced show evidence of the presence of TFP. The genetically tractable species Clostridium perfringens was used to study proteins involved in polymerizing the pilin, PilA2, into a pilus. The assembly ATPase PilB2 and its cognate membrane protein partner, PilC2, were purified. PilB2 bound the intracellular signal molecule c-di-GMP. Increased levels of intracellular c-di-GMP led to increased polymerization of PilA2, indicating that Gram-positive bacteria use this molecule to regulate pilus synthesis. These findings provide valuable information for understanding how pathogenic clostridia regulate TFP to cause human diseases.

INTRODUCTION

Extension and retraction of extracellular type IV pili (TFP) drive bacterial motility, attachment to host cells, uptake of DNA, and many other functions (1). TFP were first discovered in Gram-negative bacteria, but more recent studies indicate that they are also found throughout the clostridia and a few other Gram-positive genera (2, 3). Analysis of TFP systems in Gram-negative bacteria have identified most, if not all, of the components involved in polymerization and retraction of the pilus. Using the nomenclature for the Pseudomonas aeruginosa TFP system, these include (i) major and minor pilins (PilA, PilV, etc.), (ii) a prepilin peptidase (PilD), (iii) an assembly ATPase (PilB), (iv) an inner membrane core protein (PilC), (v) inner membrane accessory proteins (PilM, PilN, and PilO), (vi) a retraction ATPase (PilT), and (vii) a secretin to allow passage of the pilus through the outer membrane (PilQ). Other proteins, such as PilP, which anchors the PilMNO complex to PilQ, play supportive roles in TFP functions (4, 5).

Reconstructions of the in vivo architecture of the TFP apparatus in the Gram-negative bacteria Thermus thermophilus and Myxococcus xanthus were recently proposed using images obtained from electron cryotomography and mutants lacking specific components of the TFP (6, 7). The architecture predicted for the T. thermophilus TFP assembly was focused more on the PilQ secretin than on the pilus base (6). In the M. xanthus model, the base of the pilus straddled the cytoplasmic membrane and was composed of the following: (i) the pilus fiber (comprised of pilin); (ii) a ring composed of PilN and PilO on the periplasmic face of the membrane; (iii) PilC embedded in the membrane at the base of the pilus fiber; (iv) a ring on the cytoplasmic face of the membrane, formed by PilM and the N terminus of PilN; and (v) PilB, located in close proximity to PilC and the PilM/N ring on the cytoplasmic side of the membrane (7). We expect that these structures are conserved in Gram-positive versions of TFP, which lack the outer membrane and associated TFP proteins, such as PilQ and PilP.

A recent review extensively described the similarities and differences in TFP systems found in Gram-negative and Gram-positive bacteria (2). In summary, there are high degrees of sequence similarity between all of the core components, with the exception of the secretin proteins, which are absent in Gram-positive bacteria (likely because these bacteria lack an outer membrane). In most classification systems, there are three types of pili along with their associated assembly systems, called type IVa, type IVb, and Tad pilins. Clostridial pilus systems have characteristics found in both type IVa and type IVb but not Tad pili (2).

Clostridium perfringens lacks flagella and any recognizable flagellum-related genes but does have TFP (8). It exhibits a type of gliding motility in which cells are attached in an end-to-end orientation to form curvilinear filaments that, while attached to the colony, can be observed extending away in a growth-dependent manner (2, 8). TFP also contribute to the formation of biofilms by C. perfringens (9). Precisely how TFP contribute to this motility is still not known, but a transposon-mediated screen for mutants that cannot glide revealed that several classes of proteins were involved, but with only a single TFP-related gene (pilT) identified (10).

C. perfringens has four identifiable pilin proteins, named PilA1 to PilA4, and two pairs of PilB/PilC proteins, PilB1/PilC1 and PilB2/PilC2 (2, 8). We consider PilA2 to be the major pilin involved in adherence to host cells, since PilA2 is present on the surface of C. perfringens (8) and since a strain of Neisseria gonorrhoeae expressing the pilA2 gene but lacking its endogenous pilin, PilE, changed its adherence profile from human genital-urinary tract epithelial cells to mouse myoblasts and fibroblasts (11). Given the location of pilA2 on the chromosome, immediately downstream of the pilB2 and pilC2 genes, it seems likely that these encode the core assembly proteins responsible for polymerizing PilA2 into a pilus.

PilB proteins are members of the GspE superfamily of secretion ATPases. The structures of bacterial and archaeal secretion ATPases have been determined by X-ray crystallography and electron cryomicroscopy (12–16), and a recent publication reported the structure of a TFP PilB protein (but lacking the N-terminal domain) from the thermophile T. thermophilus (17). Purified TFP and type II secretion system (T2SS) assembly ATPase homologs often (18), but not always (19), form hexamers in solution or in crystals with distinct N-terminal and C-terminal domains. The GspE family protein XpsE, from Xanthomonas campestris, oligomerizes to form hexamers only in the presence of ATP (20).

The inner membrane core proteins (IMCP), or PilC proteins, are also called platform or polytopic inner membrane proteins. They are members of the GspF protein secretion superfamily and are usually predicted to have three transmembrane domains along with single cytoplasmic N-terminal and C-terminal domains. Since they are often essential for TFP assembly and lie at the base of the pilus fiber (7), PilC proteins are likely to play an important role in TFP assembly.

How are the activities of the assembly ATPases regulated at the cellular level? Recently, a direct connection to cyclic di-GMP (c-di-GMP)-dependent regulation was reported in which a family of PilB assembly ATPase proteins bound c-di-GMP directly (21). c-di-GMP is synthesized from 2 GTP molecules by diguanylate cyclase (DGC) enzymes (22). In the case of Vibrio cholerae, this leads to pilus biogenesis and changes in surface-based motility (23). c-di-GMP is believed to regulate, among other functions, the switch from flagellum-mediated swimming motility to surface colonization and biofilm formation (22), so having a direct effect on PilB activity fits within the confines of that model. In the Gram-positive bacterium Clostridium (Peptoclostridium) difficile, there is a c-di-GMP type II riboswitch directly upstream of the main TFP-encoding locus (24, 25). Expression of a DGC in C. difficile led to increased c-di-GMP and pilus biogenesis (26–29), but this riboswitch is absent in C. perfringens (data not shown).

We reported previously that a yellow fluorescent protein (YFP)-PilB1 fusion protein lost its polar localization in a pilC1 mutant of C. perfringens (30). To analyze potential interactions between PilB2 and PilC2 from C. perfringens, we purified these proteins from Escherichia coli and P. aeruginosa, respectively. We demonstrated that PilB2 binds to both ATP and the second messenger molecule c-di-GMP. PilB2 is stimulated to polymerize the PilA2 pilin by expression of a native DGC, thereby extending the role of c-di-GMP in regulating PilB activity to Gram-positive bacteria.

RESULTS

Localization of a YFP-PilB2 fusion is not altered in a pilC2 in-frame deletion strain grown on plates.

C. perfringens strain 13 contains two copies of PilB and PilC gene homologs, each in separate putative operons (Fig. 1A). We reported previously that a mutation in the pilC1 gene resulted in loss of polar localization of a YFP-PilB1 fusion protein (30), suggesting that PilC1 and PilB1 interacted directly or indirectly in the cell. To determine if this was also the case with PilC2 and PilB2, we constructed an in-frame deletion of the pilC2 gene, resulting in strain WH2. We placed a yfp-pilB2 gene fusion downstream of a lactose-inducible promoter on a plasmid in strain WH2 and measured the localization of the YFP-PilB2 protein fusion. In contrast to what we observed with the PilB1/PilC1 pair, there was no difference in the level of polar localization between the mutant and wild-type (HN13) strains (Fig. 1B). In addition, YFP-PilB2 localization was not changed in a strain with a pilC1 in-frame deletion, nor was YFP-PilB1 localization altered in the strain with a pilC2 deletion (Fig. 1C), suggesting that there is some specificity in the binding of the PilB/PilC pairs.

FIG 1.

(A) TFP-associated genes in C. perfringens strain 13. The two sets of PilB/PilC-encoding genes are colored gray. (B) Phase-contrast and fluorescence imaging of YFP-PilB2 in strains HN13 (wild type; left) and WH2 (pilC2 mutant; right). YFP-PilB2 appears as white spots in the cells. Note that some of the fluorescent foci are not at the poles. Bars = 5 μm. (C) Numbers of polar spots per cell for different strains of C. perfringens. The values shown are means and standard errors of the means (SEM). The fluorescent protein fusion used for each strain is shown in parentheses (pAH10 was used to express YFP-PilB2, and pAH12 was used for YFP-PilB1).

PilB2 binds zinc and forms hexamers in the absence of ATP.

PilB2-hemagglutinin (HA)-His₆ was purified from E. coli extracts by use of metal affinity chromatography, anion-exchange chromatography, and size exclusion chromatography (Fig. 2A). The pure protein had a faint but distinct pink color. To determine if any metals were complexed with PilB2-HA-His₆, inductively coupled plasma-atomic emission spectroscopy (ICP-AES) was carried out on a sample of the purified protein, which was analyzed for the following elements: S, Ni, Ca, Cd, Mn, Mg, Co, Cu, Zn, and Fe. Zinc and sulfur were detected in the sample at a ratio of approximately 1 mol Zn to 22 mol S. One PilB2 monomer contains 18 sulfur-containing amino acid residues, so it appears that monomers of PilB2 have 0.82 zinc bound, which suggests a level of <1 mol of zinc bound per monomer. PilB2 contains a tetracysteine motif near the C terminus (2) that has been shown in other PilB-like proteins to coordinate zinc ions (18). Other ions (Ca, Fe, and Ni) were found, but at much lower ratios (0.025, 0.012, and 0.002 mol per mol of S, respectively). The elements Mn, Mg, Co, Cd, and Cu were all below their respective detection limits. To determine the multimeric state of purified PilB2-HA-His₆, the protein was loaded onto a gel filtration column along with selected molecular size marker proteins. Based on its elution profile, the predicted mass of the protein complex was ∼489 to 537 kDa (Fig. 2B), which is 7.3 to 8 times that of a 67-kDa monomer. Given the consistent findings that assembly and retraction ATPases are hexamers (14, 16, 17, 20), it is likely that PilB2-HA is in a hexameric state as well. When the same experiment was repeated with 50 μM ATP in the buffer, there was no change in the elution profile of PilB2 (data not shown).

FIG 2.

(A) Purification of PilB2-HA-His₆. MW, molecular weight markers; Ni, elution fraction from nickel-charged immobilized-metal affinity chromatography column; Q, elution fraction from Q-Sepharose anion-exchange column; GF, elution fraction from Superose 6 column. (B) Gel filtration was performed using a Superose 6 10/300 column equilibrated with 0.5× Tris-buffered saline. The calibration sample contained thyroglobulin (669 kDa), pyruvate kinase (237 kDa), lactate dehydrogenase (140 kDa), and lysozyme (14.7 kDa) (shown as black diamonds in the graph). The PilB2-HA-His₆ size estimation was 489 to 537 kDa (white square).

Use of a coupled assay to measure ATPase activity of PilB2-HA-His₆.

The standard method for measuring the activity of TFP-T2SS-associated ATPases involves measuring the amount of phosphate released after a specified length of time (i.e., an endpoint assay). While it is sensitive, this assay has a low signal-to-noise ratio due to hydrolysis of ATP by the detection reagents themselves and is hindered by low levels of contamination by P_i. Therefore, we utilized a linked assay which includes excess amounts of the enzymes pyruvate kinase (PK) and lactate dehydrogenase (LDH) along with the substrates phosphoenol pyruvate and NADH. The ATP generated by PK is hydrolyzed to ADP and P_i by the ATPase. The pyruvate is reduced to lactate by the LDH with the oxidation of NADH to NAD⁺, which is continuously measured in a spectrophotometer. Using this assay, PilB2-HA-His₆ showed Michaelis-Menten kinetics, with a K_m of 53.2 ± 14.5 μM and a V_max of 0.27 ± 0.03 μM/min (Fig. 3A). The specific activity of the enzyme was 9.9 nmol/min/mg (Fig. 3B), which is in the same range as those seen with other assembly ATPases (13, 18, 21, 31). PilB2-HA-His₆ bound the fluorescent ATP analog trinitrophenyl-ATP (TNP-ATP) in a noncooperative manner, with a dissociation constant (K_d) of 18.7 ± 0.45 μM (Fig. 3C).

FIG 3.

(A) Using the oxidation of NADH as a marker for ATP hydrolysis, rates of ATP hydrolysis were determined for a 1 μM sample of PilB2-HA-His₆ in the presence of increasing amounts of ATP. Each point represents one replicate, and the line indicates the line of best fit, determined using least-squares regression to fit the data to the Michaelis-Menten equation. (B) Specific activity of PilB2-HA-His₆ was determined by adding increasing amounts of PilB2-HA-His₆ to assay mixtures containing 100 μM ATP. The mean and SEM for three samples were plotted for each point. (C) Increasing amounts of PilB2-HA-His₆ were incubated with 0.5 μM TNP-ATP. For each triplicate sample, the theoretical fluorescence curve, represented by the equation F = F₀ + PilB2-HA-His₆ × F_max(K_d + PilB2-HA-His₆), was fit to the data by using least-squares regression. The average of these curves was plotted. Error bars indicate SEM.

PilB2 binds the second messenger molecule c-di-GMP.

Recent reports showed that some assembly ATPases from TFP found in Gram-negative bacteria are able to bind the second messenger molecule cyclic di-GMP (c-di-GMP) and to regulate TFP polymerization (21, 23, 32). The binding site for c-di-GMP has been shown to lie in an extended N-terminal domain, such as that found in the MshE protein of Vibrio cholerae (21, 32). This domain is separate from the ATP binding and hydrolysis domains of the enzymes (21, 32). Conserved motifs present in the N-terminal domains of verified c-di-GMP-binding ATPases are conserved in the N-terminal domain of PilB2, making it a candidate for c-di-GMP binding (32). The crystal structure of the MshE N-terminal domain bound to c-di-GMP revealed that conserved tandem 24-residue motifs [RLGXX(L)(V/I)XXG(I/F)(L/V)XXXXLXXXLXXQ] linked by a 5-residue spacer form a 53-residue-long c-di-GMP-binding domain (32). The PilB2 N-terminal domain contains all but three of the conserved residues (data not shown).

A differential radial capillary action of ligand assay (DRaCALA) (33) was used to measure binding of c-di-GMP. PilB2-HA-His₆ bound to c-di-GMP (Fig. 4A); binding was cooperative, and nonlinear regression analysis using a one-site binding model gave a K_d of 1.34 ± 0.10 μM (Fig. 4A). This was similar to the K_d (1.9 ± 0.4 μM) for c-di-GMP binding to a His-MBP-MshE fusion protein derived from V. cholerae (21). The specificity of binding to c-di-GMP was shown by the addition of unlabeled competitors to the reaction mixture; only c-di-GMP exhibited inhibition (Fig. 4B). The DRaCALA method was also used to analyze the binding of ATP to PilB2-HA-His₆. The results from DRaCALA gave a K_d of 2.74 ± 0.26 μM for ATP binding to PilB2-HA-His₆ (Fig. 4C). This was specific, since only unlabeled ATP could inhibit the binding (Fig. 4D). The K_d for ATP binding was 6.8 times lower than that for TNP-ATP, suggesting that the different dissociation constants reflect differences in binding between the two nucleotides. However, despite binding of c-di-GMP with a high affinity, the addition of c-di-GMP did not increase the ATPase activity of PilB2 (Fig. 4E).

FIG 4.

The dissociation constants (K_d) and specificities of PilB2 binding to [³²P]c-di-GMP (A and B) and [³²P]ATP (C and D) were measured using DRaCALA as described in Materials and Methods. (A and C) K_d was calculated by using the equation y = B_max × x/(K_d + x). (B and D) The indicated unlabeled nucleotide competitor (100 μM) was used. (E) Increasing amounts of c-di-GMP were added to the PilB2 ATPase reaction mixture to measure its effects on activity. The means and SEM for triplicate samples are shown.

Purification of PilC2 synthesized in P. aeruginosa.

We attempted to purify the PilC2 protein from E. coli, but within a few minutes after induction of the gene by the addition of IPTG (isopropyl-β-d-thiogalactopyranoside) to the culture, the cells stopped growing (data not presented). Since different growth conditions did not relieve this problem, we used an alternative expression system, that of the TFP-containing Gram-negative bacterium P. aeruginosa. After induction of the pilC2 gene, P. aeruginosa continued to grow and produce the PilC2 protein as determined by Western blotting (data not presented). However, the majority of PilC2 in the cells was present in the insoluble fraction of lysed cells. Therefore, the purified inclusion bodies were denatured using Sarkosyl and refolded (see Materials and Methods). Fractions collected after elution with an imidazole gradient showed a single band of 43 kDa (Fig. 5A), which was confirmed to be 6×His-FLAG (HisFlg)-tagged PilC2 by Western blotting (Fig. 5B). To confirm that the protein had regained its secondary structure, circular dichroism (CD) spectroscopy was performed on the purified protein (Fig. 5C). An alpha-helix content of 70% and a beta sheet content of 3.9% were calculated from the CD spectrum. These closely match the predictions of 74% and 4.7%, respectively, made by the JPred4 secondary structure prediction server (http://www.compbio.dundee.ac.uk/jpred/), which uses Jnet, version 2.3.1 (34). These results suggest that the secondary structure was regained after refolding.

FIG 5.

(A) SDS-PAGE of elution fractions following on-column refolding of HisFlg-PilC2. Lanes: M, molecular size marker; 1, denatured sample applied to column; 2, flowthrough; 3 to 6, elution fractions of refolded PilC2. (B) Western blot of elution fractions (lanes 4 and 5 in panel A) following on-column refolding of PilB2-HA-His₆. The blot was probed with rabbit anti-PilC2 serum. (C) Circular dichroism of purified HisFlg-PilC2. The circular dichroism spectrum of PilC2 was measured with a Jasco J815 spectrometer. Secondary structure estimates were obtained from the K2D3 server (http://k2d3.ogic.ca//index.html).

Purified PilB2 does not bind to and its ATPase activity is not stimulated by PilC2.

We assayed for direct binding between PilB2 and PilC2 by using a staphylococcal protein A column that was preloaded with anti-PilB2 antibodies. PilB2-HA-His₆ and HisFlg-PilC2 were incubated alone or together for 1 h before loading onto the anti-PilB2 antibody column. The antibodies and bound proteins were eluted and analyzed by nonreducing SDS-PAGE to prevent the IgG from dissociating (Fig. 6A). Intact IgG was visible in all elution fractions, at 150 kDa. Column A was loaded with PilB2-HA-His₆ alone. PilB2-HA-His₆ (67 kDa) was visible in the eluate (Fig. 6A, lane 2) but not in the flowthrough (Fig. 6A, lane 1), indicating that it bound to the column. Column B was loaded with HisFlg-PilC2 alone. HisFlg-PilC2 (43 kDa) was visible in the flowthrough (Fig. 6A, lane 3) but not in the eluate (lane 4). Column C was loaded with a mixture of PilB2-HA-His₆ and HisFlg-PilC2. HisFlg-PilC2 was seen in the flowthrough only (Fig. 6A, lane 5) and PilB2-HA-His₆ in the elution fraction only (lane 6), suggesting that PilB2-HA-His₆ could bind to the column but that HisFlg-PilC2 could not. This experiment was repeated with 100 μM c-di-GMP added to all of the buffers, but there was no significant difference in the elution profile (data not presented), suggesting that c-di-GMP does not enhance the binding of PilC2 to PilB2. Analysis of the ATPase activity of PilB2 in the presence of purified PilC2 indicated that there was not a significant increase in activity (Fig. 6B). The lack of in vitro PilB2-PilC2 interactions may have been due to the presence of the hemagglutinin (HA)-His₆ tag on PilB2 and the HisFlg tag on PilC2, although we did not observe changes in the in vivo polar localization of YFP-PilB2 in the PilC2 mutant (Fig. 1C).

FIG 6.

(A) Antibody pulldown assay with protein A columns loaded with anti-PilB2 antibodies. Ten micrograms of each protein was applied to the columns. Each fraction was analyzed by nonreducing SDS-PAGE, and proteins were stained with Coomassie blue. Intact IgG is visible at 150 kDa in all elution fractions. Column A was loaded with PilB2-HA-His₆. PilB2-HA-His₆ (67 kDa) is visible in the elution fraction (lane 1) but not in the flowthrough (lane 2). Column B was loaded with HisFlg-PilC2. PilC2 (43 kDa) is visible in the flowthrough (lane 3) but in the elution fraction (lane 4). Column C was loaded with PilB2-HA-His₆ and HisFlg-PilC2. HisFlg-PilC2 is visible in the flowthrough only (lane 5), and PilB2-HA-His₆ is visible in the elution fraction only (lane 6). The vertical line visible in lane 1 is due to an imaging equipment defect and does not represent splicing of the image. (B) Effects of added HisFlg-PilC2 on PilB2 ATPase specific activity.

Expression of a DGC leads to increased polymerization of PilA2 in a PilB2- and PilC2-dependent manner.

Since we had determined that PilB2 could bind c-di-GMP but did not observe an increase in ATPase activity in vitro with c-di-GMP present, we increased the c-di-GMP level in vivo by expressing the gene encoding CPE1788, an endogenous protein with a GGDEF domain. As a measurement of PilB2 activity, we used an enzyme-linked immunosorbent assay (ELISA)-based method to quantify the amount of PilA2 on the surfaces of the bacteria. Expression of CPE1788 led to a significant increase in surface-exposed PilA2 (Fig. 7). When the CPE1788 expression vector was placed in a ΔpilB2 or ΔpilC2 strain, the increase was lost. Complementation of the ΔpilB2 strain with the YFP-PilB2 fusion protein restored the phenotype, suggesting that the YFP-PilB2 fusion was fully functional for polymerization of PilA2. Complementation of the ΔpilC2 strain with the PilC2-His6 protein also fully restored the PilA2 phenotype, indicating that the His tag did not interfere with PilC2 functions as they relate to PilA2 polymerization.

FIG 7.

ELISA using anti-PilA2 antibodies to detect surface-exposed PilA2. pKRAH, expression vector; pSRM18, pKRAH with cpe1788 (encoding a DGC); pAH10, pKRAH expressing a YFP-PilB2 fusion; pAH11, pKRAH expressing a YFP-PilB1 fusion; pSRM20, cpe1788 in pXEH, a xylose-inducible promoter-containing vector (45).

DISCUSSION

We purified and characterized PilB2 and PilC2 from C. perfringens; this is the first time that members of either of these protein families have been purified and characterized from a Gram-positive bacterium. We demonstrated that polymerization of PilA2, as evidenced by increased surface exposure, was dependent on PilB2 and PilC2. In C. difficile, polymerization of the major pilin PilA1 was dependent on the adjacent PilB1 assembly ATPase (27). Interestingly, the C. difficile PilB1 protein is also predicted to be a c-di-GMP-binding protein (32). In C. difficile, transcription of the operon containing the pilA1 and pilB1 genes is regulated by a c-di-GMP riboswitch, with increased levels of c-di-GMP leading to increased transcription (28, 29, 35). This pattern, where a single regulatory factor controls a process at multiple steps, has been referred to as sustained sensing (36). C. perfringens lacks the c-di-GMP riboswitch found upstream of the large TFP operon that is present in C. difficile, so at least for regulating pilus biosynthesis, it lacks sustained sensing. The reasons for this difference are unknown, but it may be related to the fact that C. perfringens, unlike C. difficile, lacks flagella and any flagellum-associated functions, which are another major aspect of c-di-GMP-regulated functions in bacteria (22). In this scenario, a single environmental input would be sufficient to activate PilB2, since the level of commitment in energy and biosynthetic capacity is much less for C. perfringens than for C. difficile.

PilB2 bound both ATP and c-di-GMP at separate sites, since only unlabeled ATP and c-di-GMP could interfere with the binding of their respective labeled forms (Fig. 4A to D). However, the addition of c-di-GMP to the reaction mixture did not increase the ATPase activity of PilB2 (Fig. 4E). In addition, c-di-GMP increased the ATPase activity of the V. cholerae MshE assembly ATPase by only 10% (21). These results suggest that, at least for these two proteins, the regulation of their activity does not seem to be at the level of enzyme catalysis. This is in contrast to other proteins that bind c-di-GMP, such as the cellulose synthase of Komagataeibacter xylinus (previously known as Acetobacter xylinus), which increases its activity in response to c-di-GMP (37). Further analysis of the role of c-di-GMP binding in regulating assembly ATPase activity may reveal the underlying mechanism(s) involved.

Besides binding c-di-GMP, PilB2 has other interesting characteristics. It was purified as a hexamer in the absence of ATP, a feature it shares with the PilF DNA transformation ATPase from T. thermophilus and the secretion ATPase GspE from the archaeal organism Archaeoglobus fulgidus (14, 16). Other assembly ATPases, such as PilB from M. xanthus (19) and GspE from V. cholerae (13), are monomers or, in the case of XpsE from Xanthomonas campestris, oligomerize to form hexamers in the presence of ATP (20). This is a small sample size, but c-di-GMP binding does not seem to correlate with forming a hexamer in vitro, since PilB2 from C. perfringens, PilB from M. xanthus, and XpsE from X. campestris bind or are predicted to bind c-di-GMP (32) but exist as hexamers (PilB2) or monomers (PilB and XpsE) in vitro. A. fulgidus, as an archaeal species, does not produce c-di-GMP.

Purified PilB2 also bound zinc at a ratio of just under one molecule of zinc per monomer of PilB2. Other assembly ATPases have been shown to bind zinc (18, 38, 39). Coordination of zinc occurs via a conserved tetracysteine motif that is found in assembly ATPases but not retraction ATPases, such as PilT (18). PilB2 and PilB1 in C. perfringens have the conserved tetracysteine motif (data not shown) and are likely to coordinate the zinc in each monomer. The role of zinc coordination in assembly ATPases was recently investigated with the type II secretion ATPase EpsE of V. cholerae (39). Mutation of the conserved cysteine residues resulted in reduction of ATPase activity in vitro and loss of function (i.e., secretion) in vivo (39). It seems likely that zinc coordination is important for the function of PilB2 in C. perfringens as well.

We used an in vitro antibody binding column assay to test for interactions between PilB2 and PilC2 but were unable to show that the two proteins bind each other (Fig. 6A). It is possible that the purified PilC2 protein, despite having the two-dimensional structure predicted for the native protein (Fig. 5C), was not in its native conformation and was thus unable to bind PilB2. However, the in vivo localization data, obtained using a YFP-PilB2 fusion, also failed to show any difference in the localization pattern when the pilC2 gene was deleted (Fig. 1C). In contrast, YFP-PilB1 localization was affected by a deletion in the pilC1 gene (Fig. 1C), showing that the method does work for that PilB/PilC pair in C. perfringens. One possible explanation for these results is that PilB2-PilC2 interactions are weaker or more transient than PilB1-PilC1 protein-protein interactions, but determining if this is the case will have to await purification and characterization of PilB1 and PilC1.

Despite the lack of evidence for specific interactions between PilB2 and PilC2, we did find evidence that PilB2 responds to increased c-di-GMP in vivo to polymerize PilA2, as seen with a whole-cell ELISA using anti-PilA2 antibodies (Fig. 7). C. perfringens has six proteins that contain a GGDEF domain, which have the potential to function as DGCs, along with three EAL domain- and one HD-GYP domain-containing protein which can function as c-di-GMP phosphodiesterases (PDEs) (Fig. 8). CPE0245 has both GGDEF and EAL domains, while CPE1560 has both GGDEF and HD-GYP domains. We chose to use CPE1788 to increase the intracellular levels of c-di-GMP and to stimulate PilB2 activity because it is small, lacks other recognizable domains, and is not membrane bound (Fig. 8). Expression of CPE1788 did lead to a significant increase in surface-localized PilA2 (Fig. 7), but this does not prove that CPE1788 actually controls PilB2 activity under normal physiological conditions. Deletions of the pilB2 and pilC2 genes led to a complete loss of the increase in PilA2 polymerization (Fig. 7) in the presence of CPE1788, and complementation with genes encoding YFP-PilB2 and His-tagged PilC2 restored activity. We deliberately used the YFP-PilB2 fusion to test whether it could function as a tracer of PilB2 localization and still complement a pilB2 mutation. The PilC2-His₆ fusion that we used for complementation, although not identical to the purified HisFlg-PilC2 protein, was used for similar reasons. Successful complementation with each protein does support our hypothesis that PilB2 and PilC2 interact, albeit indirectly, perhaps through another protein known to be located at the base of the type IV pilus, such as PilM (7).

FIG 8.

Functional domains of each protein in C. perfringens strain 13 that possesses putative DGC (GGDEF) or PDE (EAL and HD-GYP) activity. PAS, signaling domain; PBP2, periplasmic binding protein type 2. CPE1788 (asterisk) was used to stimulate PilB2 activity in vivo.

If c-di-GMP binding by PilB2 does not increase the ATPase activity but does stimulate PilA2 polymerization, then the question arises as to what mechanism underlies these findings. It is possible that c-di-GMP binding in vivo does lead to increased ATP hydrolysis, although this is difficult to measure directly. Since the c-di-GMP-binding and ATP-binding sites are in different parts of the protein, another option is that c-di-GMP binding changes the conformation of the N-terminal domain such that it becomes more effective at polymerizing PilA2, even if the rate of ATP hydrolysis remains unchanged.

MATERIALS AND METHODS

Bacterial strains and culture conditions.

All bacterial strains used in this study are listed in Table 1. E. coli was grown in Luria-Bertani (LB) medium, with antibiotics added as necessary (100 μg/ml kanamycin, 20 μg/ml chloramphenicol, or 100 μg/ml ampicillin), along with 100 μg/ml X-Gal (5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside) for blue-white screening and 1.5% agar for plates. C. perfringens was grown in a Coy anaerobic chamber in brain heart infusion (BHI) medium with 20 μg/ml chloramphenicol and/or 1.5% agar when necessary. TY medium (30 g/liter tryptone, 20 g/liter yeast extract, 1 g/liter sodium thioglycolate), supplemented with 3% galactose and/or 1.5% agar as needed, was used for mutagenesis work with C. perfringens. Fastidious anaerobe broth (Lab M) with 20 g/liter glucose (FABG) was used for pilin surface expression experiments. P. aeruginosa was grown in LB medium with 150 or 30 μg/ml carbenicillin and with 1.5% agar for plates, as appropriate.

TABLE 1.

Strains, plasmids, and primers used in this study

Strain, plasmid, or primer	Relevant characteristics or sequence (5′ to 3′)	Source or reference
Strains
E. coli strains
DH10B	F− mcrA Δmrr-hsdRMS-mcrBC ϕ80dlacZΔM15 lacX74 deoR recA1 araD139 Δara,leu7697 galU galK ΔrpsL endA1 nupG
BL21-CodonPlus(DE3)-RIL	F− ompT hsdS(rB− mB−) dcm+ Tetr gal λ(DE3) endA Hte [argU ileY leuW Camr]
C. perfringens strains
HN13	galKT deletion mutant of parent strain 13; used for construction of in-frame deletions	43
WH2	In-frame deletion of the pilC2 gene in strain HN13	This study
SRM1	In-frame deletion of the pilC1 gene in strain HN13	This study
AH12	In-frame deletion of the pilB2 gene in strain HN13	This study
Plasmids
pGEM-T Easy	PCR cloning vector	Promega
pCR-Blunt II-TOPO	PCR cloning vector	Invitrogen
pET-24a	Protein expression vector	Novagen
pKRAH1	Lactose-inducible expression vector	30
pCM-GALK	Contains a Clostridium beijerinckii galK gene under the control of a ferredoxin promoter from C. perfringens	43
pAH9	pGEM-T Easy with yfp-pilB2	This study
pAH10	yfp-pilB2 in pKRAH1	This study
pAH11	pilC2-His in pKRAH1	This study
pAH12	yfp-pilB1 in pKRAH1	This study
pWH3	PilB2-HA-His6 construct in pET24a	This study
pWH5	Vector based on pCM-GALK; used to make a pilC2 in-frame deletion	This study
pMMB-HisFlg-PilC2	Vector for expressing a gene encoding HisFlg-PilC2 in P. aeruginosa	This study
pSRM2	Vector based on pCM-GALK; used to make a pilC1 in-frame deletion	This study
pSRM16	pCR-Blunt II-TOPO with cpe1788	This study
pSRM18	cpe1788 in pKRAH1	This study
pAH20	Vector based on pCM-GALK; used to make a pilB2 in-frame deletion	This study
pXEH	Vector with a xylose-inducible promoter for expression in C. perfringens	45
pSRM20	cpe1788 in pXEH	This study
Primers
OAH14	CCGCGGTAAATAACAAAAAGGAGAACGCATAATGTCAAAAGGAG
OAH36	CACACGGAATGGATGAATTATATAAGATGAAATATACCATTAAAGATATAGATATGAAGC
OAH37	CTTCATATCTATATCTTTAATGGTATATTTCATCTTATATAATTCATCCATTCCGTGTGTAATTCC
OAH38	GAAGCGAGCTCTTTAAACCCAAACATCTAA
OAH39	CACACGGAATGGATGAATTATATAAGTTGATTAGTTATCA GAAAAAGCGTTTAGGAG
OAH40	CTAAACGCTTTTTCTGATAACTAATCAACTTATATAATTC ATCCATTCCGTGTGTAATTCC
OAH41	CTTTCGAGCTCTTACATATCATAAGTTATATTTAAC
OWH17	GGATCCCCTATGATAACAGGAGAAAAGATAG
OWH18	CTTTTAATTCTTATTAATTATGCTTTTAACCTATATTTGCCATAATTCCTCTCC
OWH19	GGAGAGGAATTATGGCAAATATAGGTTAAAAGCATAATTAATAAGAATTAAAAG
OWH20	GTCGACTCTCACTTCCAATATCAATAGAAATAG
OWH32	GCTAGCATGATTAGTTATCAGAAAAAGCGTTTAGGAGATATACTAATTG
OWH34	GTCGACAGCGTAATCTGGAACATCGTATGGGTACATATCATAAGTTATATTTAACATTTCTTCAACTGTGGTATTCCCC
OSRM21	GGGCATAAAGTATATTCAACAATACACGCTAACAGTGG
OSRM2	GGGTAAATCCTTTTCTTTTACTATACATATTTCCCAAACATCTAACCCCTCACATTC
OSRM3	GAATGTGAGGGGTTAGATGTTTGGGAAATATGTATAGTAAAAGAAAAGGATTTACCC
OSRM23	CCATATCTATATTTCCTCATAAATCACTTAACATTAATTTTCCATATGCC
OSRM48	GTTCTAGGATCCGGAGGAAAGAATATTGGAAGTTTTACTAG
OSRM49	CTAAAACTGCAGATAATACTAAAATACAGTTTATTTCTTCC
OAH159	CCATGGCATATGGACGTCGACGAGTAATTCATTTTTGGATGATTTGGGAG
OAH160	CATTTCTTCAACTGTGGTATTCCCCTCCTGATAACTAATC AAAGAAAATACCTCC
OAH117	CTGCAGTAATTAGGTGAAAGAAAAGGAGAGGAATTATGGC
OAH118	GTCGACTTAATGATGATGATGATGATGACCTATACTGTTATACATTTTAAACATAGGTG
OAH161	GGAGGTATTTTCTTTGATTAGTTATCAGGAGGGGAATACC ACAGTTGAAGAAATG
OAH162	GATCTAGACTCGAGCTCAGACATTCCACTTGATGATACTATTCCC
5′-1 HisFlgPilC2	CGATTACAAAGACGATGACGATAAACTGCTGGTTGCAAATTTTAAATATAAAGCTAT
5′-2 HisFlgPilC2	GAGGAGGATATTCATGATGGTTCATCACCATCACCATCACGATTACAAAGACGATG
3′PilC2	AGAAAGCTGGGTTTCAACCTATACTGTTATACATTTTA

Purification of PilB2-HA-His₆.

The gene encoding PilB2 was amplified from C. perfringens strain 13 chromosomal DNA by use of primers OWH32 and OWH34 and Phusion polymerase (New England BioLabs). Taq polymerase was used to generate 3′-A overhangs, and the product was ligated into pGEM-T Easy (Promega). The insert encoding PilB2-HA was excised by digestion with SalI and NheI and ligated into pET-24a (also digested with SalI and NheI) to yield pWH3, encoding PilB2-HA-His₆. BL21-CodonPlus(DE3)-RIL cells were transformed with pWH3 by electroporation and grown at 37°C in LB medium supplemented with 20 μg/ml chloramphenicol and 100 μg/ml kanamycin with shaking to an optical density at 600 nm (OD₆₀₀) of 0.7. IPTG was then added to a final concentration of 1 mM, and the cells were grown overnight. Cells were harvested by centrifugation at 10,000 × g, resuspended in 0.5× Tris-buffered saline (TBS; 12.5 mM Tris, pH 7.4, 75 mM NaCl) containing one tablet of Complete Mini protease inhibitor cocktail (Roche) and 40 mM imidazole, and lysed with a model 500 Sonic Dismembrator instrument (Fisher). The lysate was cleared by centrifugation at 20,000 × g, followed by filtration through a 0.22-μm syringe filter. The clarified lysate was applied to a 5-ml HisTrap FF column (GE Healthcare Life Sciences) and eluted with a linear gradient of 0.5× TBS with 1 M imidazole. Fractions containing PilB2-HA-His₆ (as evaluated by SDS-PAGE) were pooled, applied to a 5-ml Q-Sepharose anion-exchange column (GE Healthcare Life Sciences), and eluted with a stepwise gradient of 0.5× TBS plus 1 M NaCl. The fractions containing PilB2-HA-His₆ were pooled, concentrated, and applied to a Superose 6 10/300 gel filtration column (GE Healthcare Life Sciences) equilibrated with 0.5× TBS. Unless otherwise indicated, these and all subsequent chromatography procedures were performed using an ÄKTApurifier 10 fast-performance liquid chromatography (FPLC) system (GE Healthcare Life Sciences) at 4°C. Fractions were evaluated for purity by SDS-PAGE, and those that contained >90% PilB2-HA-His₆ were pooled, concentrated, filter dialyzed into 0.5× TBS plus 25% glycerol by use of Amicon Ultra 2-ml centrifugal filters (100-kDa cutoff; Millipore), and stored at 4°C. The protein concentration was determined with a Micro BCA protein assay kit (Thermo).

Purification of HisFlg-PilC2.

To construct a PilC2 N-terminal 6×His-FLAG tandem fusion, the gene encoding PilC2 was amplified from C. perfringens strain 13 chromosomal DNA by use of primers 5′-1 HisFlgPilC2 and 3′ PilC2. 5′-1 HisFlgPilC2 added a portion of the 6×His-FLAG epitope. The resulting product was reamplified with primers 5′-2 HisFlgPilC2 and 3′ PilC2 to add the remainder of the 6×His-FLAG epitope (MVHHHHHHDYKDDDDKLLV-PilC2). Primers 5′-2 HisFlgPilC2 and 3′ PilC2 were tailed for Gateway cloning (Thermo Fisher Scientific) as previously described (40). The resulting amplicon was cloned by Gateway cloning technology into the P. aeruginosa expression vector pMMBGW (40) to yield plasmid pMMB-HisFlg-PilC2. The expression plasmid was mobilized into P. aeruginosa strain PAK by triparental mating (41). P. aeruginosa strain K cells containing pMMB-HisFlg-PilC2 were grown overnight in LB medium with 150 μg/ml carbenicillin at 37°C with shaking in a baffled flask. The following morning, the culture was subcultured into 2 liters of fresh, prewarmed LB medium containing 30 μg/ml carbenicillin and incubated at 37°C (also with shaking in a baffled flask) until it reached an OD₆₀₀ of ∼1.0. IPTG was added to a final concentration of 200 μM, and the cells were grown overnight. The following morning, the cells were harvested by centrifugation at 25,000 × g. The cells were then suspended in 1× TBS and lysed with a model 500 Sonic Dismembrator instrument. Urea and Sarkosyl were added to final concentrations of 8 M and 2%, respectively, and the lysate was incubated overnight at 4°C with gentle agitation (42). The following morning, the solubilized lysate was diluted with an equal volume of ultrapure water. Insoluble debris was removed by centrifugation at 40,000 × g, and the supernatant was passed through a 0.45-μm filter. The solubilized PilC2 protein was applied at a rate of 0.1 ml/min to a 5-ml HisTrap FF column equilibrated with 0.5× TBS plus 0.1% Sarkosyl. The column was washed extensively with 0.5× TBS plus 0.1% Sarkosyl, and this buffer was replaced to refold the protein with 0.5× TBS plus 0.1% beta-octyl glucoside (BOG) over the course of 10 h. The refolded HisFlg-PilC2 protein was then eluted with a linear gradient of 0.5× TBS plus 0.1% BOG plus 500 mM imidazole. Fractions were evaluated for purity by SDS-PAGE, and those that contained >90% HisFlg-PilC2 were pooled, concentrated, filter dialyzed into 0.5× TBS plus 0.1% BOG plus 25% glycerol by use of Amicon Ultra 2-ml centrifugal filters (10-kDa cutoff; Millipore), and stored at 4°C. The protein concentration was determined with a Micro BCA protein assay kit (Thermo).

ICP-AES of PilB2-HA-His₆.

PilB2-HA-His₆ was suspended in 0.5× TBS and was analyzed by ICP-AES for the analytes S, Ni, Ca, Cd, Mn, Mg, Co, Cu, Zn, and Fe by use of a Spectro Arcos spectrometer (Spectro Analytical Instruments, Inc.).

Binding of TNP-ATP to PilB2-HA-His₆.

Solutions containing 0.5 μM trinitrophenyl-ATP (TNP-ATP; a fluorescent ATP analog) and increasing amounts of PilB2-HA-His₆ were prepared in 0.5× TBS and placed in a black, flat-bottomed 96-well plate. Fluorescence was measured using a Tecan Infinite M200 instrument, using an excitation wavelength of 409 nm and an emission wavelength of 541 nm.

Kinetics and specific activity of PilB2-HA-His₆.

The rate of ATP hydrolysis of PilB2-HA-His₆ was determined using a linked continuous assay. PilB2-HA-His₆ was added to a solution containing 25 mM HEPES, pH 7.5, 10 mM dithiothreitol (DTT), 20 μg/ml bovine serum albumin (BSA), 200 μM phosphoenolpyruvate, 300 μM NADH, 25 mM MgCl₂, and 2.5 μl of a pyruvate kinase-lactate dehydrogenase solution (1.5 to 2.5 U pyruvate kinase plus 2.25 to 3.5 U lactate dehydrogenase) (Sigma). The reaction was initiated by the addition of ATP. Conversion of ATP to ADP results in pyruvate kinase transferring the phosphoryl group from phosphoenolpyruvate to ADP, yielding ATP and pyruvate. Lactate dehydrogenase converts pyruvate and NADH to lactate and NAD⁺. The disappearance of NADH, as a signal of ATP hydrolysis, was monitored by the reduction in absorbance at 340 nm. The data were plotted in Excel, and the rate of the reaction from 15 min to 30 min was determined by using the line of best fit.

Analytical gel filtration chromatography of PilB2-HA-His₆.

A Superose 6 10/300 column was calibrated using thyroglobulin, pyruvate kinase, lactate dehydrogenase, and lysozyme as molecular size standards. The approximate size of PilB2-HA-His₆ was determined by comparing the retention volume of purified protein (with and without 50 μM ATP) to the retention volumes of the standards.

Circular dichroism spectroscopy of HisFlg-PilC2.

A sample containing 1 μM HisFlg-PilC2 in 0.5× TBS plus 0.1% BOG was analyzed using a Jasco J815 spectrometer. Far-UV spectra (190 to 260 nm) were recorded at 25°C in a 2-mm-path-length cuvette, using a 1-nm bandwidth, 1-nm data pitch, 2-s data integration time, and 100-nm/min scanning speed. The spectra were recorded three times and averaged. The data were analyzed using the K2D3 server (5; http://k2d3.ogic.ca//index.html).

Antibody production and Western blotting.

Rabbit polyclonal antibodies (GenScript, Inc.) were produced against synthetic peptides from PilA2, PilB2, and PilC2. The peptide used for anti-PilA2 production was composed of 14 amino acids, starting at residue 110, with the sequence N-VFAVEVSGKEDSPV-C. For PilB2 antibodies, a peptide of 15 amino acid residues was used, starting from position 155 and having the sequence N-LTDKASDEESNELC-C. The peptide used for anti-PilC2 production was composed of 15 amino acid residues, starting from position 9 of PilC2, and had the sequence N-INSEGQRIEGSQSAC-C. All antibodies were affinity purified against the respective antigen peptides. Western blotting was performed using an iBlot 2.0 transfer system (Life Technologies) and a Snap i.d. 2.0 blot development system (Millipore). Antibodies were diluted in TBS with Tween 20 (TBS-T; Santa Cruz Biotechnology), and 0.5% gelatin (Sigma) in 1× TBS-T was used as a blocking agent. The secondary antibody used, unless otherwise noted, was horseradish peroxidase (HRP)-conjugated goat anti-rabbit IgG (Thermo). Reactive bands were visualized by incubation with SuperSignal West Dura extended-duration substrate (Thermo) and imaging with a Typhoon Trio variable mode imager (GE).

Coimmunoaffinity chromatography of PilB2-HA-His₆ and HisFlg-PilC2.

Rabbit anti-PilB2 antibodies were cross-linked to protein A HP SpinTrap (GE Healthcare) columns according to the manufacturer's instructions. Samples of PilB2-HA-His₆, HisFlg-PilC2, and a mixture of PilB2-HA-His₆ and HisFlg-PilC2 were preincubated in 0.5× TBS plus 0.1% BOG for 60 min at 37°C and then separated with the prepared anti-PilB2 columns according to the manufacturer's instructions. The flowthrough, wash, and elution fractions from the columns were collected and analyzed by SDS-PAGE and Western blotting with anti-PilB2 and anti-PilC2 antibodies.

Construction and complementation of pilC1, pilC2, and pilB2 in-frame deletion mutants.

A mutant strain of C. perfringens with an in-frame deletion in pilC2 was constructed according to the method of Nariya et al. (43). To create plasmid pWH5, the region upstream of the pilC2 gene, including the first three codons, was amplified from C. perfringens strain 13 chromosomal DNA by use of primers OWH17 and OWH18. The region downstream of the pilC2 gene, including the codons coding for amino acids 400 and 401 and a stop codon, was amplified from C. perfringens strain 13 chromosomal DNA by use of primers OWH19 and OWH20. The two partially overlapping products were then used as the template, allowing for amplification of pilC2Δ_4–399 by use of primers OWH17 and OWH20. Taq polymerase was used to give the product 3′-A overhangs to allow for ligation into pGEM-T Easy. The insert was then digested out of this plasmid by use of the enzymes BamHI and SalI and ligated into pCM-GalK digested with the same enzymes to form pWH5. This plasmid was introduced into strain HN13 by electroporation. The first crossover event was selected for by the addition of chloramphenicol to the growth medium. An overnight culture of HN13(pWH5) was grown at 37°C in BHI with 20 μg/ml chloramphenicol (BHI Cm²⁰), subcultured into TY medium plus 3% galactose, and grown for 24 h at room temperature. This culture was then serially diluted and plated onto TY-3% galactose plates. Resultant colonies were patched onto TY plates containing 3% galactose and BHI Cm²⁰. Clones that underwent a second recombination event, and thus were sensitive to chloramphenicol, were screened by PCR to confirm the deletion in pilC2. An identical strategy was used to delete the pilC1 and pilB2 genes of C. perfringens strain HN13. Primer pairs OSRM21-OSRM2 and OSRM3-OSRM23 were used to amplify the flanking DNA fragments used to construct the pilC1 deletion strain (SRM1) by using the pCM-galK-based mutagenesis plasmid pSRM2. Primer pairs OAH159-OAH160 and OAH161-OAH162 were used to construct the pilB2 deletion strain (AH12) mutagenesis plasmid, pAH20. Both mutants retained, in frame, the first three and final three codons of the pilC1 and pilB2 genes.

The pilB2 deletion strain was complemented with a yfp-pilB2 fusion, located in plasmid pAH10. To create pAH10, the yfp gene was amplified from the pSW4-YFP plasmid (44) by use of primers OAH14 and OAH40. The pilB2 gene was amplified from C. perfringens strain 13 chromosomal DNA by use of primers OAH39 and OAH41. The two partially overlapping products were then used as the template, allowing for amplification of yfp-pilB2 by use of primers OAH14 and OAH41. Taq polymerase was used to give the product 3′-A overhangs to allow for ligation into pGEM-T Easy to create plasmid pAH9. The insert was then removed using the enzymes SacII and SacI and ligated into pKRAH1 digested with the same enzymes to form pAH10.

The pilC2 deletion strain was complemented with a PilC2-His₆-encoding gene in vector pAH11. pAH11 was made by amplifying the pilC2 gene from strain 13 chromosomal DNA by use of primers OAH117 and OAH118; the latter encoded the His residues. The PCR product was first cloned into pGEM-T Easy and then into pKRAH after digestion with PstI and SalI.

Since both pAH10 and pAH11 carried chloramphenicol resistance-encoding genes, in the complemented strains the CPE1788 gene was placed in the pXEH vector, a xylose-inducible promoter-containing plasmid that encodes erythromycin resistance (45), to make pSRM20. pSRM20 was made by digesting pSRM16 with EcoRI and cloning the CPE1788 gene into EcoRI-digested pXEH.

In vivo imaging of fluorescent protein fusions in C. perfringens.

pAH10, which contains a yfp-pilB2 fusion, was transformed by electroporation into C. perfringens strains HN13 and WH2. pAH12, which contains a yfp-pilB1 fusion, was made by using the same methods as those used for pAH10, except that the PCR primer pairs were OAH14-OAH37 and OAH38-OAH39 and the plasmid was transformed into C. perfringens by electroporation. All strains were incubated at 37°C overnight in liquid BHI plus 20 μg/ml chloramphenicol. The following morning, the cultures were inoculated onto BHI agar plus 20 μg/ml chloramphenicol plus 0.5 mM lactose. The plates were incubated at 37°C for 2 h, and the cells were scraped from the surface of the plate into 1× Dulbecco's phosphate-buffered saline (DPBS; 25 mM sodium phosphate, 150 mM NaCl, pH 7). The cell suspensions were placed onto polylysine-coated slides (Thermo) and allowed to stand for 10 min before a coverslip was added. The cells were then imaged at 37°C on an Olympus IX71 microscope with differential interference contrast (DIC) and YFP filters, using the DeltaVision deconvolution SoftWorx program (Applied Precision). The objective used was an Olympus 100× UPLS Apo phase objective with a numerical aperture of 1.40. The number of cells examined for each sample ranged from 120 to 612. The position of YFP-PilB2 in cells was analyzed using the MicrobeTracker suite for MATLAB and the SpotFinderZ function (46).

PilB2 DRaCALA methods.

The use of DRaCALA (differential radial capillary action of ligand assay) to identify protein-ligand interactions was described previously (33). To assay binding, purified PilB2 protein in binding buffer (10 mM Tris, pH 8.0, 100 mM KCl, 5 mM MgCl₂) was mixed with radiolabeled ligand (4 pM [³²P]c-di-GMP or [³²P]ATP), incubated for 1 min at room temperature, spotted onto dry nitrocellulose with a 2-μl pin tool, air dried, exposed to phosphorimager film, and imaged using a Fujifilm FLA-7000 phosphorimager (GE Healthcare). The fraction bounds were quantified as previously described, using Fujifilm Multi Gauge software v3.0 (33). To measure the dissociation constant (K_d), 2-fold serial dilutions of purified protein were made in binding buffer and mixed with radiolabeled ligand, and the fraction bound was determined by DRaCALA. The K_d was calculated by nonlinear one-site specific binding according to the equation y = B_max × x/(K_d + x) in GraphPad Prism. For competition assays, excess unlabeled nucleotide (100 μM) was premixed with radiolabeled ligand and added to purified protein in binding buffer, and the fraction bound was determined by DRaCALA.

Expression of a diguanylate cyclase enzyme in C. perfringens.

The gene encoding the DGC CPE1788 was amplified by PCR from C. perfringens chromosomal DNA by use of primers OSRM48 and OSRM49. The PCR product, which contained the entire coding region and the putative ribosomal binding site for the gene, was ligated to the PCR cloning vector pTOPO to create plasmid pSRM16, which was digested with the restriction enzymes PstI and BamHI and ligated to PstI-BamHI-digested plasmid pKRAH1 to produce plasmid pSRM18.

Measurements of surface-exposed PilA2.

HN13 and ΔpilA2, ΔpilB2, and ΔpilC2 in-frame deletion mutants were transformed with pSRM18 by use of previously described standard electroporation techniques. Cell suspensions of the WT and each mutant were spotted onto FABG plates containing 20 mg/ml chloramphenicol and 5 mM lactose and grown anaerobically for 24 h at 37°C. Cells were scraped from the edges of colonies and fixed anaerobically in 2% glutaraldehyde in 1× DPBS for 30 min. Cells were pelleted by centrifugation and washed once in 1× DPBS before normalization of the OD₆₀₀ to 0.5. Wells of a 96-well untreated opaque white microplate (Nalge Nunc Inc.) were coated in triplicate with 100 μl cell suspension before incubation for 40 min at 37°C. Wells were washed four times with 1× DPBS before being blocked with 200 μl blocking solution containing 4% milk and 2% bovine serum albumin in 1× DPBS. Blocking occurred at 37°C for 90 min. Wells were washed three times with 1× DPBS, and 100 μl affinity-purified rabbit antibody against PilA2 diluted to 0.5 μg/ml was incubated in wells at 37°C for 30 min. Wells were washed again, and 100 μl peroxidase-conjugated goat anti-rabbit secondary antibody (Thermo Scientific) was applied to wells at 0.1 μg/ml at 37°C for 30 min. Detection was performed using Quanta Red HRP substrate (Thermo Scientific) per the manufacturer's instructions. Peroxidase activity was stopped after 5 min of incubation at room temperature, and colorimetric values for each well were read in a clear 96-well plate with a SpectraMax M5 plate reader at 576 nm.