ABSTRACT
Biological hydrolysis of cellulose at high temperatures relies on microorganisms that either secrete free enzymes or assemble cellulosomes. While these enzymatic systems appear to be opposites of one another, they may share an underlying mechanism of attachment. Extreme thermophile Caldicellulosiruptor bescii is highly cellulolytic, due in part to its freely secreted modular, multi-functional carbohydrate acting enzymes. Additionally, C. bescii also employs non-catalytic carbohydrate binding proteins, which likely evolved as a mechanism to compete against other heterotrophs in the carbon limited biotopes that these bacteria inhabit. Prior analysis of the Caldicellulosiruptor pangenome identified that a type IV pilus (T4P) locus is conserved among all Caldicellulosiruptor species. Interestingly, T4P loci are evolutionarily divergent between the highly and weakly cellulolytic members of the genus Caldicellulosiruptor. In this study, we sought to determine if C. bescii T4P plays a role in attachment to plant polysaccharides. Based on pilin-like protein domains, transcriptomics and protein expression data, we identified the major pilin (pilA) encoded for by the C. bescii genome. Using immunodot blots, we identified xylan as the main inducer of PilA production, in comparison to other representative plant polysaccharides. The extracellular location of PilA was further confirmed by immunofluorescence microscopy. Furthermore, recombinant PilA specifically disrupted C. bescii cell adhesion to xylan and crystalline cellulose in competitive cell binding assays. Based on these observations, we propose that PilA, the major C. bescii pilin, and by extension the T4P, plays a role in Caldicellulosiruptor cell attachment to plant biomass.
IMPORTANCE Most microorganisms are capable of attaching to surfaces either to persist or take advantage of an energy source. Here, we describe that the thermophilic, plant degrading bacterium, Caldicellulosiruptor bescii, uses type IV pili to attach to polysaccharides found in plant biomass. This ability is likely key to survival in environments where carbon sources are limiting, allowing C. bescii to compete against other plant degrading microorganisms. Interestingly, the polysaccharide that induced the highest expression of pilin protein was xylan, a hemicellulose that is not the majority polysaccharide in plant biomass. We also demonstrate that attachment to polysaccharides can be disrupted by the addition of recombinant pilin. This mechanism mirrors those recently described in pathogenic Gram-positive bacteria, and may indicate the ancient origins of type IV pilus systems.
INTRODUCTION
Thermophilic microorganisms capable of hydrolyzing all, or part of lignocellulosic plant biomass have been under considerable interest for biotechnological applications of their enzymes. Of note are cellulolytic microorganisms which produce the enzymes necessary to hydrolyze recalcitrant plant biomass. The Gram-positive, anaerobic, extremely thermophilic genus Caldicellulosiruptor employ an array of mechanisms for deconstruction of plant biomass (reviewed in 1). One hallmark of this genus is modular, multifunctional enzymes comprised of both catalytic and binding domains. Caldicellulosiruptor bescii is a highly cellulolytic member of the genus (2, 3) capable of attaching to and degrading plant biomass at temperatures as high as 90°C (4). Notably, C. bescii is able to degrade insoluble cellulose along with various other plant polysaccharides like xylan (5) and pectin (6), and can grow efficiently on untreated plant biomass with high lignin content (7-9). Modular enzymes with multiple catalytic domains, found primarily in the glucan degradation locus (GDL) (10) diversifies the substrates that these enzymes can hydrolyze (11-15).
In multiple studies, Caldicellulosiruptor cells have been observed adhering to plant biomass during growth (4, 9, 16-19), presumably as an adaptation to efficiently degrade lignocellulosic biomass. Given that the genus Caldicellulosiruptor does not produce a cellulosome, other proteins have been implicated in mediating this attachment. All members of genus Caldicellulosiruptor produce one or more S-layer bound proteins and enzymes (9, 10). Two S-layer located proteins from Caldicellulosiruptor saccharolyticus were demonstrated to adhere to cellulose (17). Additionally, S-layer associated enzymes from Caldicellulosiruptor kronotskyensis facilitated attachment to xylan when heterologously expressed in C. bescii (9). Aside from S-layer located proteins, other potential plant polysaccharide-interacting proteins are also produced by the genus Caldicellulosiruptor, including substrate-binding proteins, flagella, a type IV pilus (T4P) and uncharacterized hypothetical proteins which were enriched in a cellulose-bound fraction through a proteomics screen (10). Additional C. bescii substrate binding proteins implicated in attachment to plant biomass have also been identified through extracellular protein identification (20), including an expanded proteomics dataset comparing extracellular proteins produced during growth on plant biomass versus crystalline cellulose (21). Both these proteomics studies corroborate the significance of non-catalytic proteins in the process of lignocellulosic plant biomass deconstruction by the genus Caldicellulosiruptor. Structurally unique proteins called tāpirins are another mechanism by which strongly to weakly cellulolytic Caldicellulosiruptor species attach to cellulose from plant biomass (19, 22). Genes encoding for tāpirins are located directly downstream of the T4P locus in all cellulolytic Caldicellulosiruptor (10, 19, 22), however it remains to be determined if they interact with the T4P.
Protein expression studies using ruminal cellulolytic bacteria identified pilins as cellulose-binding proteins through comparison of binding-deficient mutants versus wild type for Fibrobacter succinogenes (23) and Ruminococcus flavefaciens (24). Celluose affinity pull-down assays using extracellular proteins from R. albus 8 (25) and extracellular proteome analysis of non-cellulosomal binding deficient R. albus 20 mutant also identified pilin-like proteins, further implicating pili in cellular attachment to cellulose (26). Transcriptomic analysis, however, determined that pilin genes encoded by R. albus 7 were not differentially expressed on cellulose in comparison to cellobiose, and may indicate that other polysaccharides act as the inducer for pilin expression (27). Taken together, these studies indicate that T4 pilins from other Gram-positive, cellulolytic bacteria, like the genus Caldicellulosiruptor, may facilitate the attachment of cells to cellulose.
Genes required for assembly of a Gram-negative like T4P are fairly widespread throughout the genus Clostridium (28, 29), including the pathogens Clostridium perfringens (28), and C. difficile (30). Major pilins from both C. perfringens (31) and C. difficile (32) have been demonstrated to play a role in adhesion, and heterologous expression of the C. perfringens major pilin gene, (pilA2), in T4P-deficient Neisseria gonorrhoea mutants resulted in attachment to muscle cells (31). By gaining a novel cell-adherence specificity in the complemented N. gonorrhea strain, but not restoration of motility, illustrates a limitation to the conserved structure and function of pilins. Recently, the ATPases required for twitching motility in C. perfringens were found to be upregulated in response to colonization and necrosis of murine muscle tissue (33) supporting the role of C. perfringens T4P adherence to muscle cells. Furthermore, C. difficile ΔpilA1 mutants, lacking T4P, were significantly reduced in their ability to attach to epithelial cells (32).
Gram-negative like T4P genetic loci are also present in the genomes of plant biomass degrading genus, Caldicellulosiruptor (10). Among the strongly cellulolytic Caldicellulosiruptor, this locus is located upstream of the tāpirins and modular, multi-functional cellulases (10). Available transcriptomics (4, 8, 34) and proteomics data (4, 10, 21, 35) indicate that this T4P locus is strongly upregulated, and that pilin peptides are also present during growth on plant biomass or plant-derived polysaccharides. Considering this data, along with compelling evidence from ruminal and pathogenic Firmicutes that indicate the involvement of T4P in adherence (23-25, 31, 32) we propose that the C. bescii T4P plays a role in attachment to plant polysaccharides during plant biomass deconstruction. Through analysis of publicly available transcriptomics and proteomics data, we have identified the major pilin in C. bescii. Our findings also reveal that xylan is the key inducer of the C. bescii major pilin, and the pilin protein is located to the extracellular surface of C. bescii during growth on complex polysaccharides. Recombinant C. bescii pilin also specifically inhibited adherence of C. bescii cells to plant polysaccharides, supporting the role of C. bescii T4P in attachment to plant biomass.
RESULTS
A type IV pilus is encoded for by the genome from Caldicellulosiruptor bescii
Based on genome data available for C. bescii (4) we confirmed that all essential genes required for assembly of a Gram negative-like type IV pilus (T4P) are present (Fig. 1a). Essential genes for T4P assembly in a Gram-positive bacterium include a pilin subunit, a pre-pilin peptidase, an assembly ATPase, and a membrane protein that recruits the ATPase (29). Moreover, like other Gram-positive bacteria (36) these genes appear to be arranged in an operon, as there are no gaps larger than 71bp between the T4P genes, and only a single hairpin sequence is predicted within the pre-pilin peptidase coding sequence in the T4P locus (Table S1).
When comparing the organization of the T4P locus of C. bescii to 12 other sequenced Caldicellulosiruptor species, we observed that their T4P locus organization was highly conserved. Despite this, orthologous pilins shared between 41.8 to 100% percent amino acid sequence identity (Table S2), indicating that there may be some evolutionary adaptations among pilins between the strongly cellulolytic versus weakly cellulolytic species. A phylogenetic tree built from alignment of concatenated T4P loci from 12 sequenced Caldicellulosiruptor species was constructed to assess if there were correlations between cellulolytic ability and the T4P (Fig. 1b). As expected, genes from the T4P operons clustered by cellulolytic ability of the corresponding Caldicellulosiruptor species. This lends credence to our primary hypothesis that the T4P, on account of its proximity to modular, multi-domain glycoside hydrolases, plays a role in plant biomass attachment preceding enzymatic deconstruction.
Five putative pilins are encoded in the C. bescii genome
In order to identify putative pilins in the T4P locus, we screened genes for an N-terminal pre-pilin cleavage site that is post translationally modified by the prepilin peptidase (PilD) prior to pilus assembly (37). We also screened for typical Gram-positive sortase dependent amino acid motifs (LPxTG, 38) however none were identified in the T4P locus. Based on the presence of a pre-pilin cleavage site (N-terminal amino acid motif: GFxxxE), we identified five genes (Athe_1872, Athe_1876, Athe_1877, Athe_1880 and Athe_1881) as encoding for putative pilins (Table 1). Predicted protein lengths for these range in size from 130aa (Athe_1880) to 277aa (Athe_1872), furthermore, when we analyzed the leader peptide for all five predicted pilins, they all were of variable length, ranging from 5 up to 21 amino acids long (Table 1, Fig. S1). Based on the total predicted amino acid length of Athe_1880 and Athe_1881, these proteins are typical of T4a pilins (39), however, Athe_1880 has a leader peptide 15 amino acids in length which is not typical for T4a pilins.
Predicted secondary structures of these putative pilins indicate that these pilins share some common regions with the Gram negative T4 pilins as shown in (Fig. 2) (39, 40). These common regions include an N-terminal helix: α1 and a globular C-terminal domain. The N-terminal half of α1 is referred to as α1-N (peach) and the C terminal half as α1-C (cyan) regions, the αβ loop (red) separating the α helix from the antiparallel β sheets (yellow) of the C terminus (Fig. 2). Gram negative T4 pilins typically have cysteine residues that define and stabilize the D region. Such a D region cannot be defined for C. bescii putative pilins as they lack the cysteine residues. A predicted transmembrane domain (TMD) is located within the α1-N terminal region of each putative pilin (Fig. 2).
Predicted secondary structure and the amino acid length of mature Athe_1880 suggest that it resembles a T4a pilin, however the major pilin from C. difficile was originally predicted to be a T4a pilin on the basis of structural prediction (29), but its solved structure (PDB accessions, 4TSM, 4OGM, and 4PE2) is more closely aligned with T4b pilins (41). Given the prior inaccuracy of predicted protein models for Gram positive pilins, we first used C. bescii putative pilins as queries to search for amino acid similarity from proteins with crystal structures available on the Protein Data Bank website. Athe_1880 (37%) and Athe_1881 (59%) shared limited homology with poor E-values in the α1-N region with representative T4a pilin, Dichelobacter nodosus FimA (PDB accession 3SOK), so we did not attempt to use these structures to predict the tertiary structure of any C. bescii pilins. Surprisingly, none of the putative pilins shared significant homology with Clostridioides difficile (formerly ‘Clostridium difficile’) pilin PilA1 (PDB accessions, 4TSM, 4OGM, and 4PE2) (41) or Streptococcus sanguinis PilE1 (PDB accession 6I20) (42). Firmicutes that encode for Gram negative-like T4P may use different pilins as their major pilin, therefore, we widened our BLASTp analysis to determine if any C. bescii putative pilins matched other predicted pilin amino acid sequences from C. difficile, C. perfringens or S. sanguinis. Of the five putative pilins, only Athe_1880 and Athe_1881 shared low homology (28–29% identity over >70% query coverage) with C. difficile pilins, none of which were homologs of the major C. difficile pilin, PilA1 (30). While we could not find acceptable templates to build a predicted structure of C. bescii pilins with, these findings do suggest that in cases were a genome encodes for multiple pilins, orthologous pilins may not necessarily serve as the major pilin subunit across functionally diverse species.
Athe_1880 is the major pilin (PilA) based on transcriptomics and proteomics evidence
We expect that the major C. bescii pilin would be highly expressed and would constitute the majority of the T4P structure, therefore we examined publicly available transcriptomics and proteomics data available for C. bescii. The major C. bescii pilin should either be upregulated as determined by transcriptomics data, or enriched as peptide fragments in proteomics data. Datasets from three independent comparative transcriptomic studies of C. bescii grown on switchgrass versus glucose (8), filter paper versus glucose (4) and microcrystalline cellulose versus switchgrass (34) respectively indicated that Athe_1880 and Athe_1881 are the most highly upregulated genes among the candidate pilins (Table 2). Proteomics data for protein abundance on microcrystalline cellulose confirmed that Athe_1880 is the most abundant of all of the candidate pilins across three independent sets of proteomics data (4, 10, 35). These data were further corroborated by a recent study on the extracellular proteome of C. bescii where Athe_1880 was found to have a fold change greater than 2x on complex substrates like xylan, switchgrass and Avicel compared to simple substrates like xylose, glucose and cellobiose (21). Given both the transcriptomic and proteomic evidence, we confirm that Athe_1880 is the major pilin represented in the T4P of C. bescii, that we will refer to as PilA. We then sought to produce soluble, recombinant PilA using rational design, informed by predicted secondary protein structures (see Fig. 2). Soluble, recombinant PilA protein was produced by truncating Athe_1880 to remove the α-1N region of the alpha-helix (Fig. 3).
Xylan is the key inducer of PilA in C. bescii
Previously published transcriptomics and proteomics data confirmed that pilA gene expression and protein production were upregulated when C. bescii was grown on polysaccharides or plant biomass versus mono- or disaccharides (4, 8, 10, 21, 34, 35). Based on this evidence it is natural to assume that cellulose would be the main polysaccharide regulating the T4P operon. Since the other representative plant polysaccharides had yet to be tested, we sought to examine if hemicellulose polysaccharides played a role in the regulation of the T4P by monitoring T4P pilus biogenesis through immuno-dot blots. C. bescii cultures were cultured on 5 different sugars: cellulose, pectin, glucomannan, xylan, and xylose and sampled during early, mid-, and late exponential phase. After normalization to cell density, xylan induced a 10-fold higher amount of pilin production (7.7 pg ml-1 cell-1) compared to glucomannan (0.75 pg ml-1 cell-1) or xylose (0.51 pg ml-1 cell-1) at late exponential growth (Fig. 4). Furthermore, the amount of pilin protein quantified also increased in a growth phase dependent manner for all sugars tested, with no detectable PilA present during early exponential phase for growth on glucomannan and xylose. The most dramatic increase was noted during growth on xylan, with over a 3.5-fold increase in PilA from early to late exponential growth (Fig. 4). Interestingly, while cellulose is the major component of plant biomass, it was the least effective polysaccharide for inducing T4P production, and extracellular PilA below measurable limits, as was also the case for pectin (data not shown). Given that xylan induced the highest levels of PilA protein, we concluded that it is in fact, xylan, rather than cellulose that is the main inducer of PilA. This is not completely unsurprising, considering that after cellulose, xylan is the second most common polysaccharide present in secondary plant cell walls (43).
PilA is present on the extracellular surface of C. bescii cells
Although immunoblots confirmed the presence of PilA protein, we wanted to confirm that this protein is located to the external surface of the cell, and not secreted into the media. To visualize PilA, we labelled C. bescii cells grown on xylan with anti-PilA, using a fluorescently-tagged secondary antibody for visualization. Immunofluorescence microscopy images confirm the extracellular localization of PilA, as a green signal is present around DAPI-stained cells (Fig. 5B). Furthermore, the same samples processed with only the secondary antibody lack any fluorescent signal, establishing that non-specific binding was not the cause (Fig. 5A, C). Furthermore, we also attempted to detect PilA on C. bescii cells grown on xylose which is a monomer of xylan (Fig. 5D) to determine if PilA production would be induced by the monosaccharide component of xylan. Based on our immunofluorescence micrographs, PilA was not observed when C. bescii was grown on xylose. As a control, abiotically treated xylan was processed similar to biotic cultures, and the lack of fluorescence confirms that xylan was not interacting with either of the antibodies in a non-specific manner (Fig. 5E). However, when abiotically treated xylan was incubated with recombinant PilA, we observed a weak signal on xylan (Fig. 5F). This was surprising, as we could not detect any interaction of recombinant PilA with xylan or crystalline cellulose using pull down assays. Incubating rPilA with xylan, microcrystalline cellulose and filter paper (cellulose fiber) did not detect interactions as observed as eluted protein in an SDS-PAGE gel (Fig. 6). The overall amount of rPilA bound to xylan as observed in Fig. 5F is likely to be very low, as it is below the level of detection for stained SDS-PAGE gels (Fig. 6).
PilA inhibits binding of C. bescii cells to xylan and cellulose
Given the proximity of the T4P locus to major cellulases used by C. bescii (10), we originally hypothesized that the major pilin was functioning as an adhesin, by binding to plant polysaccharides. Based on the apparent weak affinity of PilA for xylan (Fig. 5F), we examined its putative role in adherence by assessing if it would interfere with C. bescii attachment to insoluble polysaccharides. We measured the ability of C. bescii planktonic cells to attach to insoluble polysaccharides using 2×2 factorial design (Fig. S2). Factorial design allowed us to test whether the presence of PilA influenced the attachment of planktonic cells to insoluble substrates, by testing for a statistical interaction between the “Protein” (PilA presence) and “Substrate” (xylan or cellulose presence) treatment variables. Planktonic cell densities (PCDs) were measured after incubation with an insoluble substrate and/ or recombinant PilA, with a reduction in PCD after treatment is indicative of cell attachment. Based on the increased production of PilA protein in response to xylan, we tested C. bescii cells grown on xylan for their ability to attach to insoluble xylan or cellulose (Fig. 7 A, C). As a comparison, we also tested if cells grown on cellulose behaved similarly (Fig. 7B).
In all cases, the PCD after exposure to insoluble substrate (Fig. 7, dotted lines) were lower than the PCD of cells in buffer alone (Fig. 7, solid line), indicating that C. bescii cells were attaching to xylan and cellulose, as expected. Interestingly, while the attachment of C. bescii cells to xylan after growth on xylan (26% attachment, Fig. 7A) is expected, a higher proportion of cells were attaching to cellulose after growth on xylan (69% attachment, Fig. 7C). Moreover, after addition of rPilA (Fig. 7 A-C, PilA), PCDs increase indicating that the presence of rPilA during the attachment process is inhibiting the ability of cells to interact with the substrate. We observed complete inhibition of cell attachment to xylan (Fig. 7A) compared to only partial inhibition of cell attachment to cellulose (Fig. 7 B, C) with rPilA. Two-way ANOVA was then used to test for interaction effects between our independent variables, substrate and protein (Table S3). The interaction between treatments (substrate and PilA) were statistically significant with p-values below 0.05 (Table S3). This interaction is also specific, as a control protein, BSA, did not interfere with attachment based on a lack of statistical interaction in this experiment (Fig. 7D, Table S3). By extension, our cell binding data support that C. bescii T4P play an integral role in cellular attachment to xylan, but additional mechanisms are used for cellular attachment to crystalline cellulose.
DISCUSSION
In this study, we present an initial functional characterization of the major pilin from a Gram-negative T4P locus encoded for by C. bescii. Similar to other members of the class Clostridia (24, 29), the C. bescii genome encodes for the genes required to produce a Gram negative-like T4P in a single locus (Fig. 1A), that is likely arranged as an operon (Table S1). Additionally, based on nucleotide sequence homology, the concatenated C. bescii T4P locus clusters with other T4P loci from strongly cellulolytic Caldicellulosiruptor species (Fig. 1B). Initial bioinformatics analyses identified five putative pilin genes located in the C. bescii T4P locus (Table 1). Pathogenic Firmicutes, such as Streptococcus sanguinis (44, 45) and C. perfringens (28), also encode for multiple pilins in their respective T4P loci, and a hypervirulent C. difficile strain encodes for as many as nine pilins distributed across five genomic loci (30). Unexpectedly, none of the C. bescii putative pilins share similarity with the known crystal structures of major pilins from T4P producing Gram-positive bacteria, C. difficile and S. sanguinis. However, Athe_1880 and Athe_1881 align with other annotated pilin proteins from C. difficile. These analyses indicate that regardless of amino acid sequence homology between Firmicute pilins, the major pilin appears to be unique to each species.
Based on our phylogenetic analysis, combined with past proteomics data (10), we hypothesized that the C. bescii T4P is likely used to facilitate cell attachment to polysaccharides found in plant biomass. To test this hypothesis, we first used transcriptomics and proteomics data (Table 2) to identify the C. bescii major pilin, (PilA) for molecular cloning and analysis. Direct pilin-mediated attachment to cellulose was previously observed in cellulosomal and non-cellulosomal members of the genus Ruminococcus (26, 46), and we sought to determine if members of the noncellulosomal genus Caldicellulosiruptor used T4P in a similar manner to facilitate attachment to plant polysaccharides. Pilins from C. bescii are unlikely to mediate direct interaction with plant biomass, we did not detect any adsorption of rPilA to crystalline cellulose or xylan, in qualitative binding assays (Fig. 6) in contrast to the tāpirins, which bind with moderate affinity to crystalline cellulose (19).
Regardless, previous proteomics studies have identified PilA peptides produced by C. bescii when grown on cellulose (4, 10, 21, 35) implying that the T4P overall play a potential role in attachment to plant polysaccharides. We used rPilA to generate polyclonal antibodies against PilA to monitor the presence of C. bescii pilins during growth on various plant polysaccharides. In this study, we cultured C. bescii on a chemically defined medium ensuring that the carbohydrate provided is the only available carbon source. Overall, the highest PilA production during growth was observed on xylan in comparison to glucomannan, xylose (Fig. 4), cellulose, or pectin (data not shown), implying that xylan is the main inducer of T4P locus expression in C. bescii. Immunofluorescence microscopy further confirmed that PilA is extracellularly localized during growth on xylan (Fig. 5), similar to observations on the subcellular localization of C. perfringens T4P (28). It was surprising that we could not detect any extracellular PilA from C. bescii cells grown on cellulose using immunoblots, and this discrepancy with prior data may be explained by their use of complex media in previous proteomics studies (4, 10, 35).
Response of a T4P locus to xylan has yet to be described, as R. flavefaciens and F. succinogenes pilins were identified in cultures grown on cellulose (23, 24) and/or glucose (23). However, response to soluble xylooligosaccharides in carbon-limited biotopes would be a likely adaptation to sense plant biomass in terrestrial hot springs. For example, Ruminiclostridium thermocellum (formerly ‘Clostridium thermocellum’) uses extracytoplasmic factor anti-sigma factors to upregulate genes encoding for cellulosomes-associated enzymes in response to soluble xylooligosaccharides in its environment (47, 48).
While the major C. bescii pilin does not appreciably adhere to xylan, planktonic C. bescii cells cultured on xylan attached to a greater extent to cellulose than to xylan. This adherence was, in part, mediated by T4P in a specific manner (Fig. 7), as supplemented rPilA interfered with the attachment of planktonic C. bescii cells to crystalline cellulose. Since rPilA did not adhere to xylan or cellulose in polysaccharide pulldown assays (Fig. 6), it is not likely that this observation is result of rPilA blocking binding sites for C. bescii cells. One explanation is that PilA may undergo a conformational change after pilus assembly, exposing a binding domain, similar to epitopes exposed by Neisseria T4P subjected to force (49, 50). However, this explanation does not account for the role of rPilA, as we do not expect N-terminus truncations of PilA to self assemble in a manner similar to the native T4P. Another possible explanation is that the C. bescii T4 pili are not directly interacting with insoluble substrates, but are attaching to cellulose through association with other proteins, such as the tāpirins, or to xylan using an as-of-yet unidentified xylan adhesin. This explanation is most plausible as it would explain the apparent lack of PilA affinity for polysaccharides, but the significant role of rPilA in interfering with C. bescii cell attachment. A similar mechanism has been recently described for a T4b pilus from enterotoxicogenic Escherichia coli (ETEC), where a secreted adhesin (CofJ) protein functions as a type of “scout” that would first associate with receptors on the host cell, and then interact with minor pilins present at the tip of the T4b pilus (51). Based on previously published data (19) and our findings, we propose a model (Fig. 8) for the putative role of the T4P in attachment to cellulose or xylan. Overall, our data supports the notion that non-cellulosomal species, like those from the genus Caldicellulosiruptor, have evolved alternate mechanisms to adhere to insoluble substrates.
MATERIALS AND METHODS
Identification of Pilin Genes from C. bescii and bioinformatics analysis
Putative pilin genes in C. bescii were identified using the Joint Genome Institute Integrated Microbial Genomes (IMG) database (52). Functional annotation of the predicted amino acid sequences for each pilin used the Pfam database (53) available within IMG. Signal peptides were predicted by the SignalP 4.1 Server (54) and transmembrane domains were predicted using TMHMM, version 2.0 (55). Hairpin sequences that may serve as transcriptional terminators were identified using TransTermHP (56). BLASTp, hosted within the IMG database, was used to identify Caldicellulosiruptor orthologs to all genes in the C. bescii T4P operon (Athe_1872 to Athe_1886) (52, 57). Nucleotide sequences of orthologs were aligned using the MUSCLE program with partial gaps at 95%, and a phylogenetic tree was built using the Maximum Likelihood method and Hasegawa-Kishino-Yano model, and the results were confirmed using bootstrapping values of 500 replicates using MEGA version 6 (58). Secondary structures of putative C. bescii pilins were predicted using PSIPRED program, version 3.3 (59).
Media, growth conditions and estimation of cell densities
Minimal, low osmolality defined (LOD) medium (60) was used for culturing C. bescii DSM 6725 strains on various carbon sources. C. bescii DSM 6725 was provided by Robert M. Kelly (North Carolina State University, Raleigh, NC). C. bescii cultures were grown at 75°C under anaerobic conditions with one of the following substrates, all at 1 g/l: xylose, Sigmacell (20 µm, Sigma), xylan (beechwood, Sigma), glucomannan (konjac root, NOW Foods). Epifluorescence microscopy (Nikon Eclipse E400) was used to enumerate glutaraldehyde fixed cells stained with acridine orange as previously described (61).
Cloning, production and purification of recombinant PilA
Two truncated versions of PilA (GenBank accession number WP_015908263.1) were cloned into the pET-28b protein expression vector (Novagen). Construct Athe_1880C was truncated after glycine similar to the way mature C. bescii pilins are processed in the cells, whereas PilA was truncated after the hydrophobic transmembrane domain as predicted by TMHMM server described above. Oligonucleotide primer sequences involved in the cloning process are listed in Table 3. Both the constructs were cloned with an N-terminal histidine tag for immobilized nickel affinity purification using 1ml HisTrap columns (GE Healthcare) per the manufacturer’s protocol. Both recombinant proteins were produced using auto-induction medium (62) using 50 μg/ml kanamycin and 34 μg/ml chloramphenicol for selection. PilA was eluted using a linear gradient, and eluted fractions were confirmed for purity by SDS-PAGE. PilA was then dialyzed against 50mM sodium phosphate (pH 7.2) using SnakeSkin dialysis tubing (Thermo-Scientific) per the manufacturer’s protocol. Final concentrations of PilA was determined using the bicinchoninic acid assay (BCA, Thermo Scientific) per the manufacturer’s protocol.
Immunoblots
Actively growing cultures of C. bescii on plant polysaccharides xylan, glucomannan, cellulose and the sugar xylose were sampled throughout growth until cultures reached stationary phase. Growth rates for C. bescii on the substrates mentioned above were calculated from this data. Cultures were sampled at early, mid and late exponential growth and diluted 1:1 with sterile glycerol prior to storage at ×20° C until further use. Purified recombinant PilA was applied on each membrane as a standard curve for quantitative immunoblot analysis. PVDF membranes used for protein immobilization (Amersham Hybond 0.2 PVDF, GE) were pre-wet with 100% methanol per the manufacturer’s instructions and then equilibrated in 1X PBS buffer. Samples were applied to the pre-wetted membrane blocked in PBS-T blocking buffer (PBS with 0.1% Tween-20 (v/v), 5% milk (w/v)) for one hour on an orbital shaker at room temperature. Afterwards, the membrane was rinsed with PBS-T washing buffer. Primary antibody (chicken anti-PilA, GeneTel) was diluted 1:1,000 in PBST and incubated at room temperature. Afterwards, the membrane was washed in PBS-T. Secondary antibody (rabbit anti-chicken HPR conjugate, Immunoreagents) was diluted 1:1000 PBS-T and again incubated at room temperature. The membrane was then washed in PBS-T as previously described. Chemiluminescent detection of secondary antibody used ECL western blotting detection reagents (GE), per the manufacturer’s protocol. Membranes were imaged (ChemiDoc Touch Imaging System, Bio-Rad) using standard chemiluminescent settings and analyzed using Image Lab 5.2.1 (Bio-Rad). Three independent replicates were blotted for each time point. Cell densities determined while plotting growth curves were used to calculate the number of cells applied to the membrane at each time point. Protein concentrations were determined using a standard curve (R-squared = 0.98). Protein concentrations were then normalized to the number of cells and compared using a two-sample t-test in R studio statistical software (63) (v.3.3.3).
Immunofluorescence microscopy
Immunolabelling of C. bescii cells followed methods from Conway et al. (9) with modifications. C. bescii cells were grown on 50 mL LOD medium in 125 ml serum bottles with either xylan or xylose as the carbon source and harvested at late exponential phase. Cells were pelleted using centrifugation 5000 x g for 10 minutes at room temperature (same conditions for all of the centrifugation steps). Anti-PilA polyclonal antibodies were raised in chicken (chicken anti-PilA, GeneTel), and used as the primary antibody. The secondary antibody is a goat, anti-chicken IgY polyclonal antibody conjugated to DyLight 488 (Novus Biologicals). After the final washing step, cells were stained with 1µg/ ml DAPI in 1X PBS for 5 minutes at 4°C. Cells were then vacuum-filtered onto 0.2µm polycarbonate track etched membrane filters (GVS Life Sciences) and mounted in Vectashield mounting medium for imaging (Vector laboratories). Epifluorescence imaging used a Nikon eclipse E400 microscope and images were captured using an Infinity 3 Lumenera camera. Control images were captured from secondary antibody labeled samples which were not incubated with the primary antibody. Biomass controls (uninoculated xylan in LOD) were processed similar to samples. Recombinant PilA labeled, xylan was incubated with 30 µM PilA at 300 rpm for an hour at room temperature. PilA was discarded and xylan was processed in the same manner described above. All assays had biological triplicates.
Qualitative and quantitative protein binding assays
For PilA binding assays, the substrates (cellulose or xylan) used were washed with distilled water five times followed by 16h air drying and then at 70°C for two hours. For pilin pull down assays, 10 mg of the substrate was incubated with 100 μl of PilA (30 µM) in the binding buffer, 50 mM sodium phosphate, pH 7.2 (same binding buffer used for all other assays) in a thermomixer (Eppendorf) at 700 rpm for four hours at room temperature. Insoluble substrates were pelleted by centrifugation for one minute (15000 x g). The supernatant (∼ 70µl) represented unbound protein. Pelleted substrate was washed with 1 ml of the binding buffer five times. After the final wash, the pellet was resuspended in 70µl of binding buffer representing the bound protein. Both the bound and unbound fractions were analyzed using SDS-PAGE.
Cell binding assays
C. bescii cells cultures were grown to early stationary phase on either xylan or cellulose (1 g l-1) and harvested at 5000 x g for 10 minutes at room temperature. Cells were resuspended in 50 mM sodium phosphate, pH 7.2 to a density of 109 cells ml-1. Each experimental condition consisted of a total volume of 1.2 ml comprised of 1 ml C. bescii cells, and 0.2 ml of either 30 µM rPilA, BSA or buffer. Washed xylan or cellulose (10 mg) was added to measure the amount of cells bound, and no binding substrate were added to the negative controls. All binding assays were incubated at room temperature with gentle rotary shaking at 100 rpm for one hour. After incubation, planktonic cells were enumerated using epifluorescence microscopy as described above. Each of the four treatments in all cell binding assays had at least three biological replicates. Results from each experiment were analyzed with a two-way ANOVA, using the functions “lm” and “Anova” in Program R (64). Each model tested for main and interactive effects of binding substrate (presence/absence of xylan or cellulose) and protein (presence/absence of rPilA or BSA) on the planktonic cell density. In this modeling framework, the significance of the interaction term (Substrate×Protein) indicates whether the presence of the protein (rPilA or BSA) influenced the binding affinity of cells to the substrate (xylan or cellulose).
ACKNOWLEDGEMENTS
This study was supported by start-up funds to S.E.B.-S. (Oakland University). A.K. was supported in part through funds from the Center for Biomedical Research (OU).