ABSTRACT
Burkholderia pseudomallei, a tier 1 select agent and the etiological agent of melioidosis, transitions from soil and aquatic environments to infect a variety of vertebrate and invertebrate hosts. During the transition from an environmental saprophyte to a mammalian pathogen, B. pseudomallei encounters and responds to rapidly changing environmental conditions. Environmental sensing systems that control cellular levels of cyclic di-GMP promote pathogen survival in diverse environments. Cyclic di-GMP controls biofilm production, virulence factors, and motility in many bacteria. This study is an evaluation of cyclic di-GMP-associated genes that are predicted to metabolize and interact with cyclic di-GMP as identified from the annotated genome of B. pseudomallei 1026b. Mutants containing transposon disruptions in each of these genes were characterized for biofilm formation and motility at two temperatures that reflect conditions that the bacteria encounter in the environment and during the infection of a mammalian host. Mutants with transposon insertions in a known phosphodiesterase (cdpA) and a predicted hydrolase (Bp1026b_I2285) gene exhibited decreased motility regardless of temperature. In contrast, the phenotypes exhibited by mutants with transposon insertion mutations in a predicted diguanylate cyclase gene (Bp1026b_II2523) were strikingly influenced by temperature and were dependent on a conserved GG(D/E)EF motif. The transposon insertion mutant exhibited enhanced biofilm formation at 37°C but impaired biofilm formation at 30°C. These studies illustrate the importance of studying behaviors regulated by cyclic di-GMP under varied environmental conditions in order to better understand cyclic di-GMP signaling in bacterial pathogens.
IMPORTANCE This report evaluates predicted cyclic di-GMP binding and metabolic proteins from Burkholderia pseudomallei 1026b, a tier 1 select agent and the etiologic agent of melioidosis. Transposon insertion mutants with disruptions in each of the genes encoding these predicted proteins were characterized in order to identify key components of the B. pseudomallei cyclic di-GMP-signaling network. A predicted hydrolase and a phosphodiesterase that modulate swimming motility were identified, in addition to a diguanylate cyclase that modulates biofilm formation and motility in response to temperature. These studies warrant further evaluation of the contribution of cyclic di-GMP to melioidosis in the context of pathogen acquisition from environmental reservoirs and subsequent colonization, dissemination, and persistence within the host.
INTRODUCTION
Burkholderia pseudomallei is a Gram-negative bacterium and environmental saprophyte found in the soils and surface waters of regions of endemicity (1). In these environments, the ability of B. pseudomallei to form a biofilm or internalize in amoebic cysts may increase its persistence (2). Infection of susceptible humans and animals with B. pseudomallei results in melioidosis, an often fatal disease that is common in Southeast Asia and northern Australia (3, 4) and is also detectable in the Caribbean islands, Central America, South America, and eastern and western parts of Africa (4–7). A recent report predicts that B. pseudomallei and melioidosis are considerably more widespread than previously thought and that melioidosis rivals other, more prevalent tropical diseases in terms of the case fatality rate (8).
The transition of B. pseudomallei from an environmental reservoir to the establishment of an infection and subsequent dissemination within a human host requires the ability to sense and respond to changing environmental cues. Bacteria have evolved sophisticated sensory systems to detect and respond to a variety of signals from external stimuli (9). A key signaling system in bacteria that responds to environmental challenges relies on cyclic di-GMP (c-di-GMP), a nearly universal bacterial second messenger that can regulate a variety of bacterial behaviors, including virulence, motility, and biofilm formation (10, 11). These behaviors are inversely controlled through the modulation of intracellular cyclic di-GMP levels to control the transition between motile or planktonic states and sessile or biofilm modes of growth (12, 13). Bacterial species that routinely encounter changing environmental conditions generally have more sensory and response systems encoded in their genomes than bacteria that have adapted to a narrow niche or have become obligate pathogens of a specific host (10). Comprehensive characterizations have been conducted to describe these sensory systems at the genome scale (9, 10, 14). However, understanding the complex regulatory role of cyclic di-GMP signaling at the level of the whole organism remains a challenge.
Genes encoding the proteins involved in cyclic di-GMP metabolism can be found in almost all bacterial genomes and affect host-pathogen interactions (15). A complex metabolic network rapidly responds to environmental cues and alters the intracellular levels of cyclic di-GMP through the competing enzymatic activities of diguanylate cyclases (DGCs) and phosphodiesterases (PDEs). DGC activity is associated with the amino acid residues GG(D/E)EF, which are responsible for the catalysis of cyclic di-GMP from 2 molecules of GTP (16). PDE activity is associated with EAL or HD-GYP domains, which are responsible for cyclic di-GMP hydrolysis to pGpG (17) or 2 molecules of GMP (18), respectively. Enzymatic control of cyclic di-GMP levels facilitates a rapid cellular response to external signals through the sensory and regulatory domains often associated with EAL, HD-GYP, and GG(D/E)EF domains (19). PilZ domain-containing proteins also regulate diverse cellular activities in a cyclic di-GMP-dependent manner (10). Collectively, these protein domains exert control at the transcriptional and posttranslational levels to regulate the synthesis and activity of cell surface components (e.g., flagella, adhesins, pili, exopolysaccharides) and virulence factors (20–25).
Currently, there is a large gap in our knowledge of how B. pseudomallei senses and responds to its environment. To date, a single study has investigated the physiological role of cyclic di-GMP metabolism in B. pseudomallei and has provided evidence that high cyclic di-GMP levels in B. pseudomallei strain KHW resulted in autoaggregation, increased biofilm formation, and decreased motility, cell invasion, and cytotoxicity (26). Lee et al. used the genome of B. pseudomallei K96243 to identify a single candidate phosphodiesterase gene in order to study its homolog in the KHW strain of B. pseudomallei (26). However, the regulatory control of cyclic di-GMP metabolism is generally more complex in most bacteria, and 18 putative open reading frames (ORFs) predicted to encode DGCs or PDEs have been reported for B. pseudomallei K96243 (9, 14). Bioinformatics analysis of the cyclic di-GMP metabolic genes in B. pseudomallei strain KHW was not reported, so it is impossible to predict whether this analysis generally describes the role of cyclic di-GMP metabolism in B. pseudomallei without additional studies (26). It is also relevant to consider expanding analyses to include additional cyclic di-GMP-associated genes for this organism in order to better understand the overall complexity of the cyclic di-GMP signaling network.
Given the paucity of information about cyclic di-GMP signaling in the tier 1 select agent B. pseudomallei, we identified B. pseudomallei 1026b mutants with transposon insertions in genes predicted to be involved in cyclic di-GMP metabolism and binding. Our goal was to identify key components of cyclic di-GMP signaling, which exhibit differential responses to environmental changes that contribute to the survival of the pathogen in the environment, colonization of the host, dissemination during infection, and long-term persistence in the host. We utilized a bioinformatics analysis of publicly available databases to identify a set of cyclic di-GMP-associated genes and selected transposon insertional mutants with alterations in those genes from a B. pseudomallei 1026b sequence-defined transposon mutant library. We characterized these transposon insertion mutants under various environmental conditions in order to better understand how these cyclic di-GMP-associated genes affect biofilm formation, motility, and virulence, since these behaviors are crucial for pathogen survival and dissemination in the environment and in the host.
RESULTS
Bioinformatics analysis of cyclic di-GMP-associated genes.In these studies, bioinformatic analyses of the B. pseudomallei 1026b genome identified 23 predicted proteins that contain domains with potential cyclic di-GMP metabolic or binding activities (Fig. 1). Our bioinformatic analyses expand upon the proteins described for B. pseudomallei K96243 in the large-scale analysis published by Chou and Galperin (see the companion table entitled “Distribution of GGDEF, EAL, HD-GYP, and PilZ domains in bacterial genomes” in reference 14).
Predicted proteins that contain GGEEF, EAL, GGDEF plus EAL, HD superfamily, or PilZ domains in B. pseudomallei 1026b. Locus numbers are provided for multiple genome annotations of B. pseudomallei 1026b and K96243. The original NCBI Bp1026b gene locus number is given first, the reannotated NCBI Bp1026b gene locus next, and the K96243 gene last. The domains represented are as follows: EAL, putative phosphodiesterase; GG(D/E)EF, putative diguanylate cyclase; PAS, Per, ARNT, Sim; HDc, metal-dependent phosphohydrolase with an HD motif; REC, cheY-homologous receiver domain; TM, transmembrane domain; PAC, PAS accessory domain; HDOD, metal-dependent hydrolase; GAF, cGMP-specific phosphodiesterase, adenylyl cyclase, FhlA; HD5, metal-dependent phosphohydrolase; YcgR; PilZ; and glycosyl transferase family 2. Schematics are approximate representations but are not drawn to scale. Protein domains were identified using SMART and Pfam analysis.
GG(D/E)EF and EAL domain proteins.Five predicted proteins were identified that contain the canonical GG(D/E)EF catalytic domain and also retain the inhibitory (I) site, RXXD, with an exclusive preference for GGEEF amino acid residues as opposed to GGDEF (Fig. 2). These amino acid residues are indispensable for catalysis and metal ion coordination (10). Four of the five GG(D/E)EF proteins possess predicted transmembrane domains (ranging in number from two to eight), suggesting that these proteins are localized to the membrane. One notable exception is Bp1026b_II2523, which lacks a transmembrane domain but does have a PAS domain (Fig. 1). Proteins with PAS domains have been reported to exert their signaling functions by promoting protein-protein interactions, transferring signals, and binding small molecules that can detect signals associated with oxygen levels, redox potential, or visible light (for a review, see reference 27).
Sequence alignments of GGEEF, EAL, GGDEF plus EAL, HD superfamily, and PilZ domains from B. pseudomallei 1026b. Conserved residues, as determined by Clustal Omega, are boxed. Boldface indicates 100% conservation of residues.
All six of the predicted EAL-type PDEs from B. pseudomallei 1026b do not possess additional partner domains or any other identifiable domains, such as PAS domains (Fig. 1). None of the EAL proteins possess all seven residues (ENEEDKE) that are associated with functionally characterized cyclic di-GMP phosphodiesterases (10, 28); however, both Bp1026b_I2260 and Bp1026b_I3148 possess six of the seven residues and possibly retain functional PDE activity (Fig. 2). Functional metal coordination activity has been described for certain residues in EAL domain-containing proteins, including the ENEED amino acid residues and the putative catalytic K residue, and additional metal coordination is potentially facilitated by the E residue (28). Bp1026b_I2659 and Bp1026b_I0571 are the most degenerate EAL proteins and are not predicted to be active PDEs but may serve as accessory proteins (Fig. 2). Precedents for the roles of inactive PDEs in the regulation of motility and biofilm formation have been established (29). Interestingly, these two genes are conserved in the four Burkholderia species that were compared in this study (Table 1).
BLASTN comparison of B. pseudomallei 1026b cyclic di-GMP genes with the corresponding genes of B. mallei 23344, B. thailandensis E264, B. cenocepacia J2315, and B. glumae BGR1
GGDEF–EAL composite proteins.Five composite proteins that are predicted to constitute a single polypeptide with both EAL and GGDEF domains were identified in the B. pseudomallei 1026b genome. It is commonly reported that only one of these domains is catalytically active, while the other, inactive domain may serve as a scaffold; however, in a few cases, both domains have been shown to be active (for a review, see reference 10). The GGDEF motif was conserved in four of the five composite proteins in B. pseudomallei 1026b. Bp1026b_I2284 is a notable exception, in which ASDKF replaces the canonical GGDEF motif, suggesting that Bp1026b_I2284 does not possess diguanylate cyclase activity (Fig. 2). Lee et al. have previously characterized the Bp1026b_I2284 homolog in B. pseudomallei KHW and have demonstrated that this composite EAL and GGDEF domain-containing protein, designated CdpA, is a functional phosphodiesterase (26). Among the remaining dual-domain-containing proteins, which have a conserved GGDEF motif, Bp1026b_II0885 is the only protein that retains the canonical I site (RXXD) (Fig. 2); however, only half of all known DGCs that retain the inhibitory I site are inhibited by cyclic di-GMP (10). All five composite proteins (Fig. 2) retain the seven ENEEDKE residues associated with functionally characterized cyclic di-GMP phosphodiesterases (10, 28). All of these proteins, except for Bp1026b_I2284, are predicted to contain transmembrane domains (Fig. 1). Bp1026b_I2284, Bp1026b_I2928, and Bp1026b_II2498 also contain PAS domains, suggesting that these proteins may be responsive to environmental stimuli (Fig. 1).
HD superfamily proteins.In addition to canonical EAL-containing PDEs, phosphodiesterase proteins containing HD-GYP domains facilitate the hydrolysis of cyclic di-GMP, although less is known about this class of enzymes. Hydrolysis of cyclic di-GMP mediated by a HD-GYP domain-containing protein was first described for RpfG in Xanthomonas campestris (18). RpfG has also been shown to interact with several DGCs from X. axonopodis (30). Bioinformatic analyses of the B. pseudomallei 1026b genome failed to identify proteins containing the canonical residues that make up the HD-GYP motifs HDXGK and HHEXXDGXGYP (31). However, this analysis identified four degenerate HD proteins of unknown function that lack a GYP motif (Fig. 1 and 2). As a result, we refer to these proteins as belonging to the HD superfamily of proteins rather than designating them as HD-GYP proteins. Bp1026b_I2285 and Bp1026b_II0700 are not included as HD-GYP proteins in the curated table entitled “Distribution of GGDEF, EAL, HD-GYP, and PilZ domains in bacterial genomes” (14). However, bioinformatics analyses indicate that these proteins contain HDOD and HDc domains, respectively (Fig. 1). Interestingly, Bp1026b_I2285 also contains a GAF domain (Fig. 1) that is known to bind cyclic nucleotides to modulate protein activity (32). GAF domains have also been reported to interact with a variety of small molecules, which include cyclic GMP (cGMP), cAMP, amino acids, tetrapyrrole, mononuclear nonheme iron, and nitric oxide (33, 34).
PilZ domain proteins.Bioinformatics analyses identified three proteins with predicted PilZ domains, and two of the genes had transposon insertions that were further characterized (Fig. 1). There are currently three different types of PilZ-containing proteins: type I, which retain the RXXXR and (D/N)XSXXG motifs; type II, which lack the RXXXR motif and cannot bind cyclic di-GMP; and type III, which are truncated and are unable to bind cyclic di-GMP (for a review, see reference 14). On the basis of its retention of the two conserved motifs, Bp1026b_II1683 (annotated as bcsA) is considered most likely a type I PilZ domain-containing protein and thus is predicted to bind cyclic di-GMP (Fig. 2). Bp1026b_II1683 is also a predicted glycosyl transferase, potentially involved in polysaccharide synthesis (35). Bp1026b_I3233 and Bp1026b_II0807 possess the (D/N)XSXXG motif but do not retain the RXXXR motif (Fig. 2) and are predicted to belong to the class II or III family of PilZ domain-containing proteins, which do not bind cyclic di-GMP. Bp1026b_II0807 is annotated as a 1,3-β-glucan synthase catalytic subunit, suggesting that it may play a role in polysaccharide (e.g., cellulose) biosynthesis. Cellulose production and regulation have not been characterized in B. pseudomallei. Bp1026b_I3233 is predicted to encode a homolog of the flagellar brake protein, YcgR, which is situated within one of the flagellum biosynthesis gene clusters.
Conservation of cyclic di-GMP-associated genes in Burkholderia species.In order to gain insight into the evolution of the cyclic di-GMP signaling pathway in B. pseudomallei, we used BLASTN (36) to conduct a comparative analysis of the cyclic di-GMP genes in different species of Burkholderia that occupy various niches. Comparison between the free-living opportunistic pathogen B. pseudomallei 1026b and the obligate pathogen Burkholderia mallei ATCC 23344, which evolved from B. pseudomallei (37), revealed that all but five genes are highly conserved; the exceptions are Bp1026b_I2284 (composite), Bp1026b_I2285 (HD superfamily), Bp1026b_I1579 (EAL), Bp1026b_I3148 (EAL), and Bp1026b_II0153 (GGEEF) (Table 1). A different set of five genes was found not to be conserved when B. pseudomallei 1026b was compared to the closely related soil saprophyte Burkholderia thailandensis E264. These genes include two EAL domains (Bp1026b_I1579 and Bp1026b_II0879), one composite domain (Bp1026b_II0885), one HD superfamily domain (Bp1026b_II0700), and one PilZ domain (Bp1026b_II0807) (Table 1). Interestingly, Bp1026b_I1579 (EAL) was not found in B. thailandensis E264, Burkholderia cenocepacia J2315, B. mallei ATCC 23344, or Burkholderia glumae BGR1; thus, it may represent a B. pseudomallei gene that has been horizontally acquired or is necessary for host adaptation during pathogenesis (Table 1). Bp1026b_II0879 (EAL), Bp1026b_II0700 (HD superfamily), and Bp1026b_II0807 (PilZ) are found only in B. pseudomallei 1026b and B. mallei ATCC 23344; they may represent genes that originated from a common ancestor prior to divergence (Table 1). Four genes, two of which encode predicted EAL domains (Bp1026b_I0571 and Bp1026b_I2659), while one encodes a composite protein (Bp1026b_I2456) and one a PilZ domain (Bp1026b_I3233), are conserved in all four Burkholderia species (Table 1). One of the more surprising aspects of this conservation is that two EAL genes, Bp1026b_I0571 and Bp1026b_I2659, are highly degenerate in comparison to the other B. pseudomallei 1026b EAL genes (Fig. 2). The proteins encoded by these genes may not be functional PDEs but may play roles as accessory proteins (29). The significance of the retention of Bp1026b_I2456 is not readily apparent. However, the conservation of Bp1026b_I3233, a homolog of ycgR that encodes a flagellar brake protein that regulates flagellum-based motility in a cyclic di-GMP-dependent manner (23, 38), indicates the potential importance of this protein in the cyclic di-GMP regulatory network.
Impact of cyclic di-GMP-associated genes on swimming motility.Intracellular cyclic di-GMP levels have been shown to modulate swimming motility in many bacteria through a variety of regulatory mechanisms (10). At 30°C, Bp1026b_I2284::T24 (cdpA) (composite) and Bp1026b_I2285::T24 (HD superfamily) mutants exhibited significant decreases in motility (Fig. 3A). A smaller decrease in swim zone diameter was observed with the Bp1026b_I0571::T24 (EAL) mutant (Fig. 3A). Two DGC (Bp1026b_I2235::T24 and Bp1026b_II2523::T24), one HD superfamily (Bp1026b_II1761::T24), and one PilZ domain-containing (Bp1026b_II0807::T24) mutant exhibited increased motility. The greatest increase in motility was observed in the Bp1026b_II2523::T24 mutant at 30°C, which could be rescued by conditional expression of full-length Bp1026b_II2523, as observed by a significant decrease in swim zone diameter (Fig. 3A; see also Fig. S1A in the supplemental material).
Swimming motilities of the wild type and transposon mutants. The swimming motilities of wild-type 1026b and the indicated transposon mutants were assessed on plates containing 0.3% agar. The plates were incubated at 30°C (A) or 37°C (B) for 24 h, and diameters were measured. An asterisk indicates that a significant difference was obtained with Student's t test utilizing the Bonferroni correction (P < 0.002) to account for multiple comparisons (n = 23).
Increasing the temperature to 37°C recapitulated the motility results observed during growth at 30°C. A positive correlation between increasing temperature and swim zone diameter was observed during the growth of the wild type at 30°C, 37°C, and 40°C, with a small decrease in motility at 42°C from that at 37°C (see Fig. S2A in the supplemental material). Both I2284::T24 (cdpA) (composite) and I2285::T24 (HD superfamily) mutants showed significantly decreased swimming at 37°C (Fig. 3B). A single mutant, II2523::T24 (GGEEF), exhibited a statistically significant increase in the swim zone diameter relative to the wild type at 37°C (Fig. 3B). To confirm this observation, a second allele of II2523::T24 was tested, and it also exhibited an increase in swim zone diameter (see Fig. S3A in the supplemental material). This phenotype was rescued during conditional expression of full-length Bp1026b_II2523 at a neutral intergenic Tn7 insertion site in the II2523::T24 mutant background, and a statistically significant reduction in the swim zone diameter was observed (Fig. S1B). Site-directed mutagenesis of the predicted GGEEF active site in Bp1026b_II2523 was conducted to change the DGC active site to GGAAF on the complementation construct, which caused the swim zone diameters at 30°C and 37°C to revert to the respective levels for the II2523::T24 EV (empty vector) mutant (Fig. S1A and B). It should also be noted that none of the mutants with transposon insertions in these genes exhibited growth defects under the conditions tested (see Fig. S4A and B in the supplemental material). Mutants with transposon insertions in fliC (I3555::T24) were evaluated as controls for validation of the swimming motility assay (see Fig. S5A in the supplemental material).
The decreases in motility observed for the I2284::T24 (cdpA) (composite) and I2285::T24 (HD superfamily) mutants suggest that these proteins play a role in motility independently of the temperatures tested (Fig. 3). Interestingly, there is a binding site for sigma 28 (FliA) (data not shown), a known regulator of flagellar biosynthesis (39), in the Bp1026b_I2284 (cdpA) promoter region. The presence of this binding site strongly supports the hypothesis that CdpA production is associated with flagellar biosynthesis in a cyclic di-GMP-dependent manner.
FliC production is correlated with swimming motility.In order to determine the effects of the cyclic di-GMP gene mutations on flagellum biosynthesis, protein extracts from planktonic cells were evaluated for FliC production by use of a B. pseudomallei-specific antibody for FliC (Fig. S5B and C in the supplemental material). FliC production was not detected in two transposon mutants (I3555::T24) with insertions in fliC (Fig. 4; see also Fig. S5C). Mutants I2284::T24 (cdpA) (composite) and I2285::T24 (HD superfamily) exhibited the greatest decreases in motility independently of temperature (Fig. 3). We conducted immunoblotting using a B. pseudomallei-specific FliC antibody to determine if changes in swimming were correlated with reduced FliC production for all cyclic di-GMP transposon mutants (see Fig. S6 in the supplemental material). Not surprisingly, there was a nearly 50% decrease in FliC production by the I2284::T24 (cdpA) mutant when it was grown planktonically (Fig. 4). We also noted a decrease in FliC production in the I2285::T24 mutant, which correlates with a decrease in motility (Fig. 3 and 4). The increase in the swim diameter zone of the II2523::T24 mutant was positively correlated with an increase in FliC production (Fig. 4).
Analysis of the production of flagellar protein FliC by B. pseudomallei 1026b and the indicated transposon mutants. Extracts were prepared from planktonic cells grown in biological triplicate at 37°C. Proteins were separated by SDS-PAGE analysis and were immunoblotted with monoclonal anti-FliC and anti-GroEL antibodies. FliC protein levels were normalized to GroEL protein levels. Asterisks indicate significant differences (P < 0.05) obtained by Student's t test.
Impact of cyclic di-GMP-associated genes on biofilm formation.Cyclic di-GMP has been shown to play an integral role in biofilm formation in numerous bacterial species (10). Recently, the role of temperature in the regulation of biofilm formation has been linked to cyclic di-GMP signaling in Vibrio cholerae; however, the mechanism has yet to be elucidated (40). To determine if any of the predicted cyclic di-GMP metabolic/binding proteins were involved in biofilm formation, we assessed biofilm formation in LB medium at two temperatures (30°C and 37°C) using a static microtiter biofilm assay (Fig. 5). The effects of temperature on biofilm formation have been reported previously for B. pseudomallei (41) and numerous other bacteria (40). Under our growth conditions over a range of temperatures from 30°C to 42°C, the level of biofilm formation by B. pseudomallei 1026b increased 2-fold at 37°C over that at 30°C but declined sharply at 40°C (Fig. S2B). A ∼6-fold decrease in biofilm formation was observed at 40°C and 42°C from that at 37°C. These data illustrate the role of temperature in the regulation of biofilm formation. Temperature appears to have a greater impact on biofilm formation than on the control of motility (Fig. S2A and B).
Biofilm formation by the wild type and transposon mutants. B. pseudomallei 1026b and the indicated transposon mutants were grown in 96-well polystyrene plates at 30°C (A) and 37°C (B) for 24 h. Attached biofilm bacteria were quantified with crystal violet. Asterisks indicate significant differences obtained by Student's t test utilizing Bonferroni's correction (P < 0.002) to account for multiple comparisons (n = 23).
A single mutant, II2498::T24 (composite), exhibited a statistically significant increase in biofilm formation at 30°C (Fig. 5A). In addition, I1579::T24 (EAL) and I2928::T24 (composite) also showed increased biofilm formation at 30°C. Two EAL mutants (I0571::T24 and I3148::T24), three of the four HD superfamily mutants (II1761::T24, I2285::T24, and I2818::T24), both of the PilZ mutants (II0807::T24 and II1683::T24 [bscA]), and one GGEEF mutant (II2523::T24) exhibited decreases in biofilm formation at 30°C (Fig. 5A). Both of the PilZ mutants may be involved in polysaccharide biosynthesis, based on the presence of a glycosyl transferase domain (Fig. 1). These data suggest that perhaps cellulose or an uncharacterized polysaccharide participates in B. pseudomallei biofilm formation at 30°C.
Increasing the temperature to 37°C resulted in a significant increase in biofilm formation for two mutants, I2928::T24 (composite) and II2523::T24 (GGEEF) (Fig. 5B). Statistically significant decreases in biofilm formation were observed for one composite mutant, I2284::T24 (cdpA), and a PilZ mutant, II0807::T24 (Fig. 5B). Interestingly, our biofilm results with I2284::T24 (cdpA) do not replicate those in a previous report by Lee et al. (26), who demonstrated that interruption of cdpA resulted in hyper-biofilm formation. To determine if the discrepancy was due to culture conditions, we set up microtiter biofilms in AB minimal medium containing 0.2% glucose and 0.5% Casamino Acids, as reported previously (26). Despite the change in the medium conditions, the cdpA (I2284::T24) transposon mutant in B. pseudomallei strain 1026b failed to exhibit an increase in biofilm production over that by the wild type (see Fig. S7 in the supplemental material). Therefore, we suggest that the discrepancy may be due to differences in B. pseudomallei strains or in the genetic elements used for gene disruption.
A diguanylate cyclase contributes to the thermoregulation of biofilm formation.The most intriguing finding of the biofilm assays was the >2-fold increase in biofilm formation by the II2523::T24 (GGEEF) mutant at 37°C and the reciprocal >2-fold decrease at 30°C (Fig. 6). To confirm that inactivation of Bp1026b_II2523 results in hyper-biofilm formation at 37°C, we identified a second transposon allele and confirmed that it also exhibited increased biofilm formation at 37°C (Fig. S3B). Complementation of the II2523::T24 mutant with full-length Bp1026b_II2523 under the control of an isopropyl-β-d-thiogalactopyranoside (IPTG)-inducible promoter at a neutral site on the chromosome reduced and restored biofilm formation to wild-type empty-vector (EV) levels at 37°C; however, complementation of the II2523::T24 mutant was abrogated when the GGEEF motif was mutated to GGAAF and conditionally expressed (Fig. 6B). The hyper-biofilm formation phenotype of the II2523::T24 mutant at 37°C is different from what might be expected when a diguanylate cyclase is functionally inactivated. The current model of cyclic di-GMP metabolism predicts that loss of a DGC should result in a decrease in biofilm formation, as observed for the II2523::T24 mutant grown at 30°C (Fig. 6A). This phenotype was also complemented with full-length Bp1026b_II2523 and was dependent on an intact GGEEF motif (Fig. 6A). Biofilm assays with a second transposon allele of Bp1026b_II2523 along with complementation of the first transposon mutant indicate that the hyper-biofilm formation phenotype observed for the II2523::T24 EV mutant at 37°C is due to the functional loss of Bp1026b_II2523 (Fig. 6B; see also Fig. S3B). The increased biofilm formation observed at 37°C for the II2523::T24 EV mutant and the decrease at 30°C (Fig. 6) indicate that the putative DGC encoded by Bp1026b_II2523 participates in the thermoregulation of biofilm formation.
Complementation of the II2523::T24 mutant biofilm phenotypes at 30°C and 37°C. The II2523 transposon mutant was complemented with full-length Bp1026b_II2523 or Bp1026b_II2523GGAAF expressed from the tac promoter on a chromosomally integrated mini-Tn7 element. Biofilm formation was assessed at 30°C (A) and 37°C (B) in the presence of IPTG. Asterisks indicate significance differences (P < 0.05) obtained with Student's t test.
Delayed disease progression in an acute model of B. pseudomallei infection.Based on the in vitro data, we chose to evaluate the virulence of three mutants, I2284::T24, I2285::T24, and II2523::T24, in a murine model of acute melioidosis. These transposon insertion mutants expressed either a significant decrease in motility or an increase in biofilm formation at 37°C, which would be predicted to contribute to acute and rapidly disseminating infection. In this acute-infection model, all three of the transposon insertion mutants showed slightly delayed disease progression relative to that with the wild type, indicating that the activities of these individual genes do not contribute appreciably to disease progression under the conditions tested (see Fig. S8 in the supplemental material).
DISCUSSION
B. pseudomallei is highly resistant to numerous antibiotics and is a tier 1 select agent for which no vaccines are currently available. Our understanding of cyclic di-GMP signaling pathways in this bacterium and related species is limited. To better understand the role of cyclic di-GMP in B. pseudomallei, we utilized bioinformatics and a sequence-defined transposon library to identify transposon insertions in genes predicted to participate in cyclic di-GMP signaling, which have important roles in the regulatory control of factors that contribute to pathogenicity, motility, and biofilm formation. To date, there has been only one cyclic di-GMP metabolic gene, a phosphodiesterase gene designated cdpA, characterized from B. pseudomallei (26). The goal of this study was to characterize a collection of B. pseudomallei mutants with mutations in predicted cyclic di-GMP genes that could be used to better understand the role(s) of c-di-GMP regulation in Burkholderia spp.
Current models of cyclic di-GMP signaling correlate low levels of signal with planktonic growth, increased virulence factor production, and motility, whereas high intracellular cyclic di-GMP levels promote antibiotic tolerance and persistence during infection, associated with a physiology that is linked to a sessile or biofilm state of growth. This simple model likely underestimates the complexity of cyclic di-GMP signaling networks in general (42) and does not account for key metabolites produced during cyclic di-GMP degradation. This point is further illustrated by the large number of proteins encoded for the production and degradation of cyclic di-GMP in the genomes of pathogens that survive in the environment. Although many bacteria have homologous genes, the components of the sensory machinery differ from organism to organism, and this signaling pathway has not been comprehensively characterized in B. pseudomallei. Libraries of transposon insertions and in-frame deletion mutants of DGCs, PDEs, and DGC-PDE composites in Pseudomonas aeruginosa, Escherichia coli, and V. cholerae have been instrumental in identifying genes that are involved in biofilm formation, motility, and cytotoxicity (29, 40, 42, 43). In P. aeruginosa, one of the model organisms for cyclic di-GMP studies, 41 proteins have been characterized and shown to modulate phenotypes dependent on cellular levels of cyclic di-GMP (19, 43). Kulasekara et al. used a library of transposon mutants to evaluate the contribution of each individual DGC and PDE to P. aeruginosa biofilm formation and pathogenesis in order to identify DGCs and PDEs that contribute to various phenotypes (43). The data in this published study are comparable to what we have found, where DGCs and PDEs contribute to different phenotypes when challenged under various environmental conditions. Although we observed statistically significant differences for some transposon insertion mutants, we also observed phenotypes that made observable contributions that were not statistically significant under the stringent statistical criteria that our analysis utilized. The effects of these mutations may become more apparent when they are combined in the same genetic background or analyzed under a wider range of environmental conditions.
The most striking finding reported in this study is the differential regulation of biofilm formation observed at different temperatures in the II2523::T24 mutant. The effects of mutations in DGCs on biofilm formation in response to changing temperatures have been reported previously (40); however, the identification of a predicted DGC, Bp1026b_II2523, that contributes to biofilm formation at 30°C and restricts biofilm formation at 37°C (Fig. 5) has not been reported previously. These data indicate that in the wild-type strain, this predicted DGC has activity that most likely increases cyclic di-GMP concentrations, which would correspond to a decrease in swimming. The molecular mechanism for temperature sensing, the corresponding thermoregulation of biofilm formation, and the participation of this DGC is unclear; however, the alterations in biofilm formation at 30°C and 37°C are fully complemented by conditional expression of Bp1026b_II2523 that is dependent on the GGEEF motif (Fig. 6).
The increase in biofilm formation by II2523::T24 at 37°C could indicate a potential role in modulating the function of another DGC. One possible explanation for this activity at 37°C is heterodimer formation with another DGC to decrease DGC activity, as hypothesized by Lindenberg et al. (44). Another possible explanation is that Bp1026b_II2523 interacts with another protein to disrupt DGC function. A precedent exists for the role of catalytically inactive DGCs and PDEs in the regulation of adherence and motility in uropathogenic E. coli (29). Mutations in DGCs that increase biofilm formation have also been reported previously in similar studies conducted on P. aeruginosa transposon and clean deletion mutant libraries (42, 43), indicating that the increased biofilm formation resulting from a transposon insertion in Bp1026b_II2523 is not a phenomenon unique to B. pseudomallei. Future studies will identify potential proteins that interact with Bp1026b_II2523 at various temperatures.
We identified two transposon insertion mutants, Bp1026b_I2284 and I2285, with decreased swimming motility, which is characteristic of increased cellular levels of cyclic di-GMP, corresponding to decreased inactivation of cyclic di-GMP. A previously described phosphodiesterase from B. pseudomallei KHW, cdpA, showed increases in cyclic di-GMP production, biofilm formation, and Congo red binding and decreases in motility, adhesion, and cytotoxicity (26). In this study, a T24 transposon insertion in Bp1026b_I2284 exhibited decreased motility, as expected; however, we did not observe increased biofilm formation or increased Congo red binding (data not shown). The decrease in biofilm formation by our cdpA mutant most likely results from the reduced production of flagella, which are known to be important for the initial phase of biofilm formation on abiotic surfaces (45). In addition, the reduction of motility in this study appears to differ from the findings in the initial report of cdpA. The motility of a 1026b cdpA transposon mutant is reduced approximately 50%, whereas interruption with a tetracycline resistance cassette in the KHW strain results in a nearly complete loss of motility (26).
These observed phenotypic differences between the KHW strain and 1026b could result from differences in the genetic backgrounds of the strains, assay conditions, or the different genetic methods employed in these two independent studies: gene inactivation with a transposon versus gene inactivation by the insertion of a tetracycline resistance gene. It has also been reported previously that phenotypic differences between the B. pseudomallei 1026b and B. pseudomallei KHW strains have been observed in mutational analyses with respect to antibiotic susceptibility and virulence factor expression (46).
The observation that Bp1026b_I2285 contributes to motility is also a novel finding. Bioinformatics analyses predict that this gene encodes a hydrolase with HDOD and GAF domains. The gene product does not fit the current criteria for a canonical HD-GYP phosphodiesterase (Fig. 2), which converts cyclic di-GMP to pGpG and ultimately GMP (18). Instead, it falls into a class of HD proteins of unknown function that lack the GYP motif. The identification of a GAF domain and an HD domain would potentially indicate that this protein is involved in binding nucleotides and hydrolyzing bonds. Hydrolysis of cyclic di-GMP mediated by HD-GYP domain-containing proteins was originally reported for Xanthomonas campestris (18), and further research has demonstrated that HD-GYP domain-containing proteins interact with DGCs in Xanthomonas axonopodis (30). The GYP motif is critical for complex formation between proteins containing HD-GYP and GGDEF domains but is not necessary for the PDE activity to completely degrade cyclic di-GMP to GMP (47). Our bioinformatics analyses of B. pseudomallei 1026b failed to identify HD-GYP proteins containing the GYP motif, indicating that additional proteins that metabolize cyclic di-GMP to GMP remain to be characterized. EAL-containing phosphodiesterases only degrade cyclic di-GMP to pGpG. The association of Bp1026b_I2285 and adjacent cdpA (Bp1026b_I2284), encoding a known EAL-containing cyclic di-GMP phosphodiesterase, suggests that these genes may work together in some unknown context. This finding potentially represents the discovery of a variation in the cyclic di-GMP metabolic pathway that has yet to be identified and may be critical for nucleotide turnover. The recent discovery of a gene (orn) for oligoribonuclease, which hydrolyzes pGpG to 2 molecules of GMP, in P. aeruginosa has demonstrated that additional proteins that contribute to cyclic di-GMP metabolism and turnover can be identified (48, 49). Future efforts will evaluate the biochemical role of Bp1026b_I2284 and identify the protein(s) that converts pGpG to GMP to complete the cyclic di-GMP metabolic pathway in B. pseudomallei.
In order to better understand the role of a subset of the cyclic di-GMP mutants in pathogenesis, we selected three mutants that we hypothesized would be attenuated in a murine model of melioidosis based on the motility and biofilm phenotypes observed in this study. Based on a previous report on the role of cdpA in B. pseudomallei KHW cellular invasion and cytotoxicity in cell culture assays (26), it was expected that I2284::T24 (cdpA) would exhibit reduced virulence in an acute model of melioidosis. I2284::T24 (cdpA), I2285::T24, and II2523::T24 all exhibited a marginal delay in the time from intranasal inoculation to severe disease requiring euthanasia. Although this difference contributed to a ∼6-h delay in clinically significant symptoms, these findings do not indicate that these genes are essential for acute pathogenesis. Surprisingly, these data also indicate that CdpA may not make a significant contribution to acute stages of melioidosis in B. pseudomallei 1026b. Additional studies will be conducted to fully evaluate dissemination and bacterial loads in both acute and chronic murine models of melioidosis as controlled by cyclic di-GMP. These data indicate that in order to rule out the possibility of functional redundancy, it may also be necessary to study pathogenicity and the corresponding role of cyclic di-GMP signaling for multiple components of the system.
Higher-order deletion (double, triple, etc.) mutants will be generated in order to identify the additional contributions of the DGCs, PDEs, DGC-PDE composites, HD superfamily proteins, and PilZ proteins and will allow for a more detailed dissection of cyclic di-GMP signaling pathways in B. pseudomallei. Despite the potential drawbacks of using transposon mutants, such as polarity, residual activity from a promoter within the transposon, and partial loss of function, the mutants described in these studies, when combined appropriately with complementation studies, offer a rapid means of initially characterizing genes of interest.
These findings illustrate the significance of screening panels of cyclic di-GMP mutants under various environmental conditions. Our results indicate that the role of cyclic di-GMP metabolism with regard to the regulatory control of motility, virulence, and biofilm formation in B. pseudomallei is not as well understood as previously thought. The identification of cyclic di-GMP binding and metabolic proteins that have differential roles in biofilm formation at different temperatures will provide insight on how environmental pathogens sense and respond to the transition from survival in the environment to the establishment of infection in a homeothermic host.
MATERIALS AND METHODS
Bacterial strains and growth conditions.Experiments were performed in select-agent approved biosafety level 3 (BSL-3) and animal biosafety level 3 (ABSL-3) facilities in the Rocky Mountain Regional Biocontainment Laboratory at Colorado State University with appropriate approvals from the CDC and the Colorado State University Institutional Biosafety committee. B. pseudomallei strain 1026b, a clinically derived strain isolated from the blood of a patient with melioidosis (50), was used in these studies (Table 2). Transposon (T24) mutant derivatives generated during the production of a comprehensive two-allele sequence-defined transposon mutant library (unpublished data) were used exclusively in these studies (Table 2). T24 is a Tn5-derived transposon containing a select-agent approved kanamycin resistance selection marker that was constructed in Colin Manoil's laboratory at the University of Washington. Bacterial strains were grown in lysogeny broth (LB) medium with shaking at 250 rpm and 37°C unless otherwise stated. The LB medium used in these studies was the Lennox formulation and consisted of 10 g/liter tryptone (Fisher Scientific, Pittsburgh, PA), 5 g/liter yeast extract (Becton, Dickinson and Company, Sparks, MD), and 5 g/liter NaCl (Fisher Scientific, Pittsburgh, PA). For animal studies, mid-log-phase cultures were grown in LB broth; their titers were determined; and they were stored in growth medium supplemented with 20% (vol/vol) sterile glycerol at −80°C prior to use for inoculation. For complementation studies, cultures were induced with 1 mM IPTG (Gold Biotechnology, Saint Louis, MO). E. coli RHO3 conjugation strains were maintained on a medium with 400 μg/ml 2,6-diaminopimelic acid (DAP; Chem-Impex International, Wood Dale, IL). B. pseudomallei 1026b T24 mutants were maintained on media containing 300 μg/ml kanamycin (Gold Biotechnology, Saint Louis, MO).
Strains and plasmids used in this study
Bioinformatic analyses.Potential cyclic di-GMP metabolic domains were identified through bioinformatic analyses in order to find predicted proteins encoded in the B. pseudomallei 1026b genome. Transposon mutant insertions in genes identified by this analysis were identified from a sequence-defined B. pseudomallei 1026b transposon mutant library (unpublished data). The location of the transposon insertion was reconfirmed by sequencing. Predicted protein sequences from the B. pseudomallei 1026b genome (51) were obtained from the NCBI GenBank database. Sequences were searched against the Pfam database for the identification of protein domain families. B. pseudomallei 1026b chromosomes I and II (GenBank accession numbers NC_017831.1 and NC_017832.1 , respectively) were searched against the Pfam database for “GGDEF,” “EAL,” “HD” (for HD-GYP), and PilZ domains (52). Pfam entry identifiers for the following domains are given in parentheses after the domain names: GGDEF (PF00990), EAL (PF00563), PilZ (PF07238), and HD superfamily members HD (PF01966), HDOD (PF08668), HD_2 (PF12917), HD_3 (PF13023), and HD_5 (PF13487). A cutoff E value of 1.0 was used. Additional domain analysis for the identification of cyclic di-GMP metabolic proteins was conducted with SMART (Simple Modular Architecture Research Tool) (http://smart.embl-heidelberg.de/ ). Rational analysis of putative proteins containing HD domains was used to eliminate putative proteins showing strong homology to well-defined proteins with known functions that are not currently associated with cyclic di-GMP metabolism. Comparison of cyclic di-GMP genes with those of other Burkholderia species—B. mallei 23344 (taxonomic identifier [taxid] 243160), B. thailandensis E264 (taxid 271848), B. cenocepacia J2315 (taxid 216591), and B. glumae BGR1 (taxid 626418)—was conducted using NCBI's BLASTN default settings. Protein sequence alignments were performed using Clustal Omega (53). Predicted proteins and potential transmembrane domains were evaluated with TMHMM (http://www.cbs.dtu.dk/services/TMHMM/ ) (54).
Complementation of Bp1026b_I2284::T24 (cdpA), Bp1026b_I2285::T24, and Bp1026b_II2523::T24 mutants.Complementation was achieved by transposase-mediated integration of full-length Bp1026b_I2284, Bp1026b_I2285, Bp1026b_II2523, or Bp1026b_II2523GGAAF into the chromosome and conditional expression of the gene of interest from the tac promoter with IPTG induction, using methods for select-agent-compliant genetic manipulation of B. pseudomallei (55, 56). Full-length Bp1026b_I2284, Bp1026b_I2285, and Bp1026b_II2523 were amplified by PCR using primers 5′-NNNCCCGGGAGGAGGATATTCATGGAAGCCATCAGGAACAA-3′ and 5′-NNNAAGCTTTCATGCGGTGGCGTGCAGAT-3′ for Bp1026b_I2284, 5′-NNNCCCGGGAGGAGGATATTCATGCCTATCACCGCACAACTGCC-3′ and 5′-NNNAAGCTTCTACCGATCGTCGCCCTGCGCG-3′ for Bp1026b_I2285, and 5′-NNNCCCGGGAGGAGGATATTCATGAACCTGCTTTCCTCGCTG-3′ and 5′-NNNAAGCTTTCAACCGAACGCAACCGAGC-3′ for Bp1026b_II2523. All forward primers included an XmaI site and a ribosome binding site, while all reverse primers contained an HindIII site for cloning purposes. PCR products were digested with XmaI and HindIII and were then ligated into pUC18T-mini-Tn7T-Km-LAC, where gene expression is driven from the E. coli tac promoter (57). Site-directed mutagenesis of GGEEF to GGAAF in Bp1026b_II2523 was carried out using the QuikChange Lightning kit according to the manufacturer's recommendations (Agilent Genomics) with primers 5′-AGCAGCGCGAATGCCGCGCCGCCGAAGC-3′ and 5′-GCTTCGGCGGCGCGGCATTCGCGCTGCT-3′. All resulting plasmids were confirmed via sequencing. Prior to complementation, the kanamycin resistance cassette in the transposon mutants was removed via conjugation with RHO3/pFLPe2 (55). The removal of kanamycin and zeocin resistance markers was confirmed by PCR. Triparental matings of RHO3/pTNS3, RHO3/pUC18T-mini-Tn7T-Km-LAC expressing the full-length clone of interest, and the transposon mutant of interest (kanamycin-sensitive clone) were conducted by plating the strains on LB medium with 400 μg/ml DAP. After 24 h, transconjugants were selected on a medium containing 1,000 μg/ml kanamycin. Kanamycin-resistant colonies were further analyzed for the site of insertion, which was verified by PCR. Strains with a mini-Tn7 insertion adjacent to glmS2 were chosen for further analysis. Expression from the tac promoter was induced with 1 mM IPTG.
Swimming motility assays.Swimming motility plates were made using LB medium supplemented with 3 g/liter Bacto agar (Becton, Dickinson and Company, Franklin Lakes, NJ). The medium was poured into petri plates approximately 4 h prior to inoculation. The medium was stab inoculated with 2 μl from an overnight culture grown in LB medium with 300 μg/ml kanamycin as needed. Plates were incubated at 30°C or 37°C, and swim diameters were measured at 24 h. All mutants were tested at least twice in triplicate or quadruplicate.
Biofilm assays.Biofilm assays were performed as described previously (20) with the following modifications: overnight cultures were grown in LB medium and were diluted to an optical density at 600 nm (OD600) of 0.1 in LB medium. One hundred microliters of the diluted culture was added to each well (replicates of six) of a 96-well polystyrene plate (Nunc Microwell 96-well microplates; catalog no. 243656; Thermo Scientific, Grand Island, NY). Plates were incubated in a sealed Ziploc bag at 30°C or 37°C for 24 h. The supernatant was removed, and the wells were washed once with 1× phosphate-buffered saline (PBS) to remove planktonic bacteria and were stained with 0.05% crystal violet (Sigma-Aldrich, St. Louis, MO) for 15 min. The wells were washed again with 1× PBS, and the crystal violet was solubilized in 95% ethanol for 30 min. The solubilized dye was transferred to a new 96-well polystyrene plate, and the OD600 of each well was quantified on a Synergy HT plate reader (BioTek Instruments, Winooski, VT).
Immunoblotting.Lysates from planktonic cells were isolated from 1 ml of an overnight culture, which was centrifuged and resuspended in 200 μl 1× PBS to which 200 μl 2× Laemmli buffer containing β-mercaptoethanol (Sigma-Aldrich) was added. Samples were then boiled for 10 min. The 660nm Protein Assay kit (Pierce) was used to normalize sample loading (4 μg for FliC) onto 4-to-20% Criterion gradient gels (Bio-Rad, Hercules, CA). Gels were transferred to a 0.2-μm polyvinylidene difluoride (PVDF) membrane (Bio-Rad) and were blocked in 1× Tris-buffered saline–Tween 20 (TBST) containing 5% skim milk. Immunoblots were probed at a ratio of 1:1,000 with a primary B. pseudomallei-specific FliC (F3F8 IgG2b) or GroEL (8E4) antibody and were detected with a horseradish peroxidase (HRP)-conjugated goat anti-mouse polyclonal antibody (1:50,000) (Pierce). The immunoblots were visualized using a Thermo SuperSignal West Pico detection kit. The signal was quantified with Image Lab analysis software on a Bio-Rad ChemiDoc XRS+ system. FliC and GroEL values were separately normalized to the values for wild-type Bp1026b, and the normalized FliC value for each lane was divided by the respective normalized GroEL value where indicated.
Statistical analysis.All statistical analyses were performed using GraphPad Prism (GraphPad Software, Inc., La Jolla, CA). Mutant strain results were normalized to those for the wild type and were analyzed using a pairwise Student t test. Significance was defined as a P value lower than 0.05. Alternatively, when the sample size was increased to 23, significance was defined as a P value lower than 0.002. This was determined using the Bonferroni correction to account for multiple comparisons. Error bars in figures indicate standard errors.
Ethics statement.Animal studies were carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health (58). The animal use protocol was approved by the Institutional Animal Care and Use Committee at Colorado State University (Animal Welfare Assurance number A3572-01).
ACKNOWLEDGMENTS
This work was supported by grant AI065357 from the National Institute of Allergy and Infectious Diseases, National Institutes of Health, to the Rocky Mountain Regional Center of Excellence for Biodefense and Emerging Infectious Diseases Research, Colorado State University. G.I.B. and B.R.B. also received research support from the Boettcher Foundation's Webb-Waring Biomedical Research Program.
We acknowledge and thank Colin Manoil and Larry Gallagher at the University of Washington for providing preliminary transposon sequence insertion information. We thank Linnell Randall, Nicole Podnecky, Katherine Rhodes, and Nawarat Somprasong for helpful discussions and technical assistance. We are also grateful to S. Brook Peterson, Yasuhiko Irie, Mihnea Mangalea, Jason Cummings, and Lily Filipowska for critical review and comments on the manuscript prior to publication.
FOOTNOTES
- Received 4 November 2016.
- Accepted 7 December 2016.
- Accepted manuscript posted online 12 December 2016.
Supplemental material for this article may be found at https://doi.org/10.1128/JB.00780-16 .
- Copyright © 2017 American Society for Microbiology.
