Conservation of the Chitin Utilization Pathway in the Vibrionaceae

Vibrionaceae are regarded as important marine chitin degraders,and attachment to chitin regulates important biological functions;yet, the degree of chitin pathway conservation in Vibrionaceaeis unknown. Here, a core chitin degradation pathway is proposedbased on comparison of 19 Vibrio and Photobacterium genomeswith a detailed metabolic map assembled for V. cholerae frompublished biochemical, genomic, and transcriptomic results.Further, to assess whether chitin degradation is a conservedproperty of Vibrionaceae, a set of 54 strains from 32 taxa weretested for the ability to grow on various forms of chitin. Allstrains grew on N-acetylglucosamine (GlcNAc), the monomer ofchitin. The majority of isolates grew on ${alpha}$ (crab shell) and β(squid pen) chitin and contained chitinase A (chiA) genes. chiAsequencing and phylogenetic analysis suggest that this geneis a good indicator of chitin metabolism but appears subjectto horizontal gene transfer and duplication. Overall, chitinmetabolism appears to be a core function of Vibrionaceae, butindividual pathway components exhibit dynamic evolutionary histories.

INTRODUCTION

Top

INTRODUCTION

Chitin is the second most abundant biopolymer after celluloseand, particularly in the marine environment, may comprise animportant source of organic carbon and nitrogen (2, 25). Chitinis composed of chains of N-acetylglucosamine (GlcNAc) residuesarranged in antiparallel ( ${alpha}$ ) or parallel (β) configurations.Both forms are found in the marine environment: β-chitinis produced by diatoms and is a major component of squid pens,while the more recalcitrant ${alpha}$ form makes up crustacean shells.While the ability to grow on the chitin monomer GlcNAc is thoughtto be widespread among bacteria (37), likely because it is acomponent of peptidoglycan, chitinoclastic ability is limitedto a number of bacterial groups within Proteobacteria, Bacteroidetes,Actinobacteria, and Firmicutes (8).

We focus here on bacteria of the family Vibrionaceae, whichincludes Vibrio and Photobacterium as its primary genera, sincethey have been studied extensively with respect to growth onchitin. Vibrios are ubiquitous and easily cultivatable membersof the coastal marine bacterioplankton community; as obligateheterotrophs, they can utilize a wide range of carbon sourcesfor energy (42). Moreover, there has been growing interest inthe effect of chitin on pathogenicity and regulation of geneexpression in the vibrios (27). Attachment of pathogenic V.cholerae to chitinous zooplankton may not only provide a nutrient-richhabitat (12) but could play a role in enhancing human diseasetransmission (7, 13). Chitin has been shown to change the physiologyof the vibrios by inducing competence (26), upregulating attachment/colonizationproteins involved in pathogenesis (20, 36), and increasing survivalduring temperature stress and exposure to stomach acid (3, 29).Thus, chitin has a strong influence on the growth and physiologyof vibrios. It is estimated that chitin can support up to 10%of marine bacterial production (19), and it has been speculatedthat the ubiquity of the vibrios can be explained by their abilityto degrade chitin (37).

Chitin degradation is achieved by a complex pathway includingmultiple chitinases (39); however, most studies of chitinasediversity in the aquatic environment focus on the distributionof the extracellular endochitinase chitinase A (chiA), sincethis gene is thought to be conserved in both Proteobacteriaand Firmicutes (8, 22). Additionally, for organisms with multiplechitinases, chiA appears to have the highest expression andactivity in response to crab shell chitin (31, 39), suggestingthat it may be most active in the environment and thus is apotentially useful indicator of chitinoclastic ability.

In this study, we propose a chitin degradation pathway for V.cholerae by incorporating bioinformatic predictions, biochemicalstudies, and expression data. We then ask how this pathway mapsonto sequenced Vibrio and Photobacterium genomes to determinewhether there is a conserved chitin degradation core. Second,we evaluate how widespread chitin metabolism is among Vibrionaceaeisolates, which cover the coexisting diversity in temperatecoastal waters (41, 43), by assaying growth on different formsof chitin ( ${alpha}$ and β). Third, we explore conservation of thechitin degradation pathway in strains by using the chiA gene,and we evaluate its evolutionary dynamics in Vibrionaceae.

MATERIALS AND METHODS

Top

MATERIALS AND METHODS

In silico analysis of the chitin pathway.
The annotated protein and DNA sequences from Vibrionaceae genomesequences (complete and unfinished) were obtained from the NationalCenter for Biotechnology Information (NCBI) website on 10 May2007; Shewanella oneidensis MR-1 was included as an outgroup.A list of genomes and their accession numbers is contained inTable S1 in the supplemental material. OrthoMCL (23) was usedto identify orthologous groups (families) in the sequenced genomes.This program takes all-against-all BLASTp as input and definesputative pairs of orthologs or recent paralogs based on thereciprocal best BLAST hit. Recent paralogs are identified asgenes within the same genome that are more similar to each otherthan any sequence from another genome. OrthoMCL then convertsthe reciprocal BLASTp values to a normalized similarity matrixthat is analyzed by a Markov Cluster algorithm (MCL). In return,the MCL yields a set of clusters, with each cluster containinga set of orthologs and/or recent paralogs. OrthoMCL was runwith a BLAST E-value cutoff of 1e–6 and an inflation parameterof 1.5. Families related to chitin metabolism were obtainedfrom the chitin pathway defined for Vibrio cholerae (Fig. 1)and by using a keyword search for "chitin" in the annotatedgenomes. For each of the chitin-related families, the orthologousgenes were identified for all Vibrionaceae genomes with OrthoMCL,and a presence/absence profile was constructed. A complete listof locus tags and gene locations is provided in Table S2 inthe supplemental material.

View larger version (37K):
[in this window]
[in a new window]

FIG. 1. Schematic of the chitin catabolic cascade in V. cholerae. Enzymes and transporters are given gene identifiers from V. cholerae N16961 when possible. The boxes around gene identifiers denote how functions were predicted: gray shading, biochemical evidence in the vibrios; thick outline, microarray expression data (27); thin lines, bioinformatic prediction only; and dashed lines, predicted functions based on experimental evidence. (Expanded from reference 34 with permission of the publisher.)

Vibrionaceae genome phylogeny.
A "whole-genome phylogeny" was generated for the annotated genomesby taking 100 randomly selected, single-copy genes present inall genomes. These were aligned in MUSCLE (9), and poorly alignedregions were removed; this concatenated alignment was used toestimate a maximum-likelihood phylogenetic tree by using thePhyML program with 100 bootstrap replicates (10), with options"0 i 1 100 GTR e e 4 e BIONJ y y."

Growth assays.
Vibrionaceae strains were tested for growth on GlcNAc, ${alpha}$ , andβ chitin as the nitrogen and carbon nutrient sources (Table1). Cultures were grown overnight in 0.25x 2216 medium (Difco)and diluted 1:100 in minimal medium containing chitin substrates.The minimal medium was derived from that used by Meibom et al.(26), i.e., 234 mmol/liter (brackish) or 428 mmol/liter (marine)NaCl, 27.5 mmol/liter MgSO₄, 4.95 mmol/liter CaCl₂, 5.15 mmol/literKCl, 0.07 mmol/liter Na₂B₄O₇, 0.187 mmol/liter K₂HPO₄, 1x "K"trace metals (14), and 50 mmol/liter HEPES, pH 7.4, and supplementedwith a filter-sterilized vitamin mixture (30). β-Chitinwas isolated from squid pen (Loligo pealei) by treatment with1 mol/liter NaOH for 5 h to remove protein, followed by extensivewashing to remove residual base (6). Tubes containing media(15 ml) were supplemented with 25 mmol/liter GlcNAc or 0.05g of either crab shell ${alpha}$ -chitin (Sigma) or β-chitin. Strainswere grown at room temperature ( $~$ 22°C) with shaking at 150rpm, and growth was assessed every 2 days. A starting opticaldensity at 600 nm of less than or equal to 0.01 that increasedto a value of at least 0.1 by day 30 was scored as positive.

View this table:
[in this window]
[in a new window]

TABLE 1. Growth of vibrio isolates on different forms of chitin and GlcNAc

PCR amplification and phylogeny of chiA.
PCR primers designed to target all known proteobacterial chiAgenes were used to amplify and sequence this gene in vibrioisolates: chiAf (GGN GGN TGG CAN YTN WSN GAY CCN TT) (8) andchiAr (ATR TCN CCR TTR TCN GCR TC) (22). DNA was obtained usinga DNA extraction kit (Gentra) or Lyse'N Go (Pierce). The PCRmixture contained 1 µmol/liter final concentration ofchiAf and chiAr, 0.75 U Jumpstart Taq (Sigma), 200 µmol/literdeoxynucleoside triphosphates, and 1x buffer. The PCRs werethermocycled as follows: 3 min at 94°C, followed by 35 cyclesof 1 min at 94°C, 1 min at 50°C, and 2 min at 72°C,with a final 6-min extension at 72°C. Alternate primerstargeting Photobacterium profundum chiA family sequences weredesigned based on the sequences of strains SS9 and 3TCK andcontain all codon degeneracies. Primers Pprof_chiAf (AAR CAYTTY CCN GAR ATG GCN GC) and Pprof_chiAr (TCR TTR TCN ACD ATRTAY TGN GC) were amplified as described above.

An alignment, including chiA gene sequences from diverse isolates,previously analyzed taxa (22), and whole genomes, was preparedusing Clustal and refined manually. Ambiguously aligned regionswere excluded, yielding an alignment of 603 nucleotide positions.The maximum-likelihood tree was constructed using PHYML underthe GTR model with estimation of all parameters and generationof 100 bootstraps (10).

Additional gene sequencing.
The partial 16S rRNA gene was amplified as described previously(41) and identified based on similarity to database sequences(1). For a limited subset of isolates, adenylate kinase (adk)and malate dehydrogenase (mdh) sequences were amplified as describedpreviously (38).

Nucleotide sequence accession numbers.
Sequences were submitted to GenBank with accession numbers EU177043to EU177094.

RESULTS AND DISCUSSION

Top

RESULTS AND DISCUSSION

Chitinolytic pathway in V. cholerae.
The chitinolytic system in the vibrios channels chitin monomersinto the central metabolism as fructose-6-phosphate (fructose-6-P),acetate, and ammonium (18). We refine a previous representationof the chitinolytic pathway (34) by incorporating literaturedata on biochemical experiments, microarray expression data,and bioinformatic predictions to fill gaps in the pathway relatedto chitobiose metabolism and identify a core set of genes whichare responsible for chitin degradation in vibrios.

Figure 1 depicts the proposed chitin catabolic cascade in V.cholerae, beginning with the breakdown of chitin polymer intooligomers by extracellular chitinases (Fig. 1, label 1). Thesegenes are assumed to have differential activity or regulationand act collectively to degrade chitin into (GlcNAc)_n>2 oligosaccharides(31, 39), which are transported into the periplasmic space viaa specific porin (Fig. 1, label 2) (15). The monomer GlcNAcand dimer N,N'-diacetylchitobiose are thought to enter the periplasmby nonspecific porins. Once in the periplasm, chitin oligosaccharidesare degraded by periplamic chitinodextrinases (Fig. 1, label3) (17) and β-N-acetylglucosaminidases (Fig. 1, label 4)(16) to (GlcNAc)_1,2. (GlcNAc)₂ is transported across the innermembrane by an ABC-type transporter (Fig. 1, label 5) (24),whereas GlcNAc can be transported into the cytosol and phosphorylatedvia a PTS transporter (Fig. 1, label 8) (5). In the cytosol,(GlcNAc)₂ is converted into 2(GlcNAc-6-P) by an N,N'-diacetylchitobiosephosphorylase (Fig. 1, label 6) (32), a GlcNAc-1P-mutase (Fig.1, label 7) (24), and a predicted GlcNAc-specific ATP-dependentkinase (gene not identified) (4). The GlcNAc-6-P generated eitherduring uptake by the PTS or by the N,N'-deacetylchitobiose phosphorylasepathway is converted into fructose-6-P via the action of anN-acetylglucosamine-6-phosphate deacetylase (Fig. 1, label 9)and a glucosamine-6-phosphate deaminase (Fig. 1, label 10) (11).

Complete degradation of chitin must also take into account theassimilation of deacetylated residues (GlcN), which can compriseup to a sixth of the residues in natural forms of chitin (28).Here, we propose a mechanism by which GlcN could be incorporatedinto the chitin catabolic cascade. Recently, a set of genesannotated as a cellobiose PTS transporter (Fig. 1, label 12)(VC1281 to VC1286) was demonstrated to transport (GlcN)₂ intothe cytosol (27). An adjacent gene (VC1280) was also upregulatedupon addition of (GlcN)₂ and has a predicted deacetylase function(Fig. 1, label 11), suggesting it converts GlcN-GlcNAc to (GlcN)₂.Once in the cytoplasm, the β1-4 linkage between the glucosamineresidues could be broken by an enzyme (Fig. 1, label 13) currentlycharacterized as a cellobiase (33). We reannotate this geneas a chitobiase, as V. cholerae does not grow on cellobioseand both substrates consist of β1-4-linked glucose. Further,this gene is upregulated by growth on chitin (27) and is adjacentto components of the chitin metabolic pathway. The cytoplasmicGlcN can be phosphorylated by an ATP-dependent glucosamine kinase(Fig. 1, label 14) (34) and converted to fructose-6-P (Fig.1, label 10). The proposed chitin utilization scheme describedabove identifies a predicted chitin degradation core; the questionis how well conserved this pathway is in the vibrios.

Distribution of chitin pathway genes in Vibrionaceae genomes.
The conservation of the chitin degradation pathway in the sequencedVibrionaceae genomes suggests that chitin metabolism is an ancestralfeature of the vibrios (Fig. 2). In Fig. 2, the left panel depictsthe phylogenetic relationships of the sequenced genomes, whichdemonstrate that gene presence/absence in the right panel hasa phylogenetic context (e.g., the second copy of chitinodextrinaseis shared among all V. cholerae genomes but not other isolates).The chitin degradation genes identified in V. cholerae (Fig.1) appear to be almost universally conserved, with homologsidentified for 91% of core gene matrix positions in sequencedgenomes (Fig. 2). The genes in V. cholerae which are not wellconserved in other genomes include the second copy of a GlcNAc-6-Pdeacetylase, an alternative chitinase (VC1952), and the (GlcN)₂PTS transporter. We note that the V. angustum S14 whole-genomephylogeny and 16S rRNA gene sequence place this strain withinthe genus Photobacterium, and it is included in this group forsubsequent analyses. Gene families annotated with chitin-relatedfunctions and present in at least two genomes are also shownin Fig. 2; these genes have a spotty distribution in Vibrionaceaegenomes, suggesting that outside of the chitin degradation core,there is tremendous gene content flexibility.

View larger version (56K):
[in this window]
[in a new window]

FIG. 2. Distribution of predicted chitin pathway genes among Vibrionaceae genomes. The phylogenetic relationship is based on maximum-likelihood analysis of a concatenation of 100 shared genes. Numbers at nodes represent values based on 100 bootstrap replicates. Each of the columns corresponds to a chitin metabolism-related gene family, with the family name indicating the predicted function and the number indicating the reaction or transport mechanism identifier from Fig. 1, with a representative gene designation in parentheses. The number within the box indicates the number of copies of that gene family in the corresponding genome, which is further indicated by light-gray shading for one gene copy or dark-gray shading for two or more gene copies. An asterisk indicates a complete genome sequence.

There is also evidence for several gene duplications. For example,the chitodextrinases (Fig. 1, label 3) contain two orthologouscopies in all V. cholerae genomes and in P. profundum SS9, althoughone of the copies (VC1073) is not upregulated in the presenceof chitin (27), suggesting that this gene may no longer be activein chitin degradation. Additionally, the chiA gene family hastwo copies in Photobacterium sp. strain SKA34 and V. angustum,one of which clusters with the vibrios while the second is moreclosely related to other proteobacteria (Fig. 3). However, multiplecopies of PTS genes (Fig. 1, labels 8 and 12) may reflect similaritiesbetween transporters for different substrates rather than multiplecopies of the same gene.

View larger version (39K):
[in this window]
[in a new window]

FIG. 3. Phylogenetic relationships of partial chiA gene sequences from Vibrionaceae and related organisms based on maximum-likelihood analysis. Numbers shown at nodes represent values based on 100 bootstrap replicates; only nodes with values of >80 are shown. Branch length to the outgroup is truncated, as indicated by the arrow. GenBank accession numbers are given for previously sequenced genes. Gray boxes indicate potential LGT events. Round circles indicate the two copies of chiA family genes in Photobacteria.

Growth of Vibrionaceae environmental isolates on chitin substrates.
Although the genome analysis suggests that chitin utilizationis a universal characteristic among Vibrionaceae, a previousstudy indicated that growth on chitin was distributed spottilyamong Vibrionaceae isolates (35). Therefore, a more diverseset of isolates was tested for growth on ${alpha}$ and β chitinas well as GlcNAc, the monomer of chitin. Growth on GlcNAc iscommon among marine bacteria, even among those not capable ofmetabolizing chitin (44). Indeed, all 54 Vibrionaceae strainsassayed grew on GlcNAc as the sole nitrogen and carbon nutrient,including the few strains which did not grow on chitin and appearedto lack chiA genes (e.g., V. halioticoli and V. hispanicus)(Table 1). This suggests that growth on GlcNAc is not a goodindicator of chitin metabolism and is consistent with the previouslysuggested GlcNAc uptake by the PTS system, which is independentof chitin degradation (Fig. 1).

The majority of isolates also grew on both ${alpha}$ and β chitin,although 10 strains grew only on the more enzymatically accessibleβ form. Overall, the broad distribution of chitin metabolismsuggests that chitin degradation is indeed an ancestral capabilityof the vibrios. However, several isolates were incapable ofchitin degradation (Table 1), corroborating that it is not auniversally conserved characteristic within the vibrios andthat strains within a family may have alternate lifestyles.However, the fraction of isolates which displayed growth onchitin was much higher than that reported in the previous study(35), and we attribute this to more complete media containingtrace metals and vitamins. Several isolates, including bothV. ordalii strains, V. ichthyoenteri, and V. calviensis, produceda yellow pigment when attached to chitin but not when grownon rich media, glucose, or GlcNAc, indicating that chitin orperhaps biofilm growth regulates pigment production.

Diversity of chitinase A among Vibrionaceae.
The chiA gene fragments amplified from strains listed in Table1 were sequenced and found to be highly divergent, with a maximumnucleotide divergence of 55% within the genus Vibrio, comparedto 22% for recA (40) and $~$ 10% for the 16S rRNA gene within Vibrionaceae(21). The photobacterial chiA sequences are even more diverse,with the second copy of the strain S14 and SKA34 chiA genesgrouping with nonvibrio proteobacteria (Fig. 3), while the P.profundum chiA family genes share only $~$ 30% amino acid identitywith other vibrio sequences. The majority of Vibrionaceae strainsform a large clade albeit without strong bootstrap support (Fig.3), although the Enterovibrio sp. and V. fischeri sequencesare distinct from this large cluster.

A positive chiA PCR assay was a good predictor of chitin metabolism;however, several Photobacterium and Vibrio strains gave negativePCR results but still grew on chitin. Indeed, the P. profundumgenomes harbor highly divergent sequences (see Table S2 in thesupplemental material), which are distinct from the other vibriochiA sequences but contain the conserved catalytic site motif,suggesting chitinase activity (22). Because the "universal"proteobacterial chiA primers do not match these P. profundumchiA sequences, new primers were designed for the divergentchiA genes (see Table S2 in the supplemental material). However,these new primers did not capture additional chiA sequencesin strains positive for growth on chitin, suggesting that chiAis either not necessary for chitin degradation or more diversethan previously anticipated. The second possibility is supportedby phylogenetic analysis using additional genes (hsp60, mdh,and adk) for five Vibrionaceae isolates, which grew on chitinbut had negative PCR results for chiA. Four of the strains,with 16S rRNA gene sequences most similar to those of P. damselaeand P. phosphoreum, formed two deep clades within the photobacteriadistinct from the sequenced genomes (see Fig. S1 in the supplementalmaterial). Given that the sequenced photobacterial genomes containdivergent chiA sequences, these additional clades (see Fig.S1 in the supplemental material) may harbor highly differentiatedchiA genes. This is an indication that even core chitin degradationgenes are apparently subject to duplication and transfer.

The use of chiA to identify chitin degraders (8, 22, 35) isproblematic, as even within a single bacterial family, the chiAgene family is too divergent to capture with PCR primers. Additionally,there is evidence for lateral gene transfer (LGT) or duplicationof this gene, which will make developing relationships withthe organismal phylogeny difficult (8). The phylogeny of thechiA gene suggests several other instances of LGT (Fig. 3);the most obvious is the placement of alphaproteobacterial sequencesin a node within the Enterovibrio group that has a well-supportedbootstrap value. While some Alphaproteobacteria contain thepathways to assimilate GlcNAc, chitinase-like sequences havenot been observed thus far in sequenced genomes (44). Moreover,Cottrell et al. (8) found that the chiA-containing alphaproteobacterialisolates did not grow on chitin, suggesting a nonfunctionalchitinoclastic pathway, potentially a hallmark of LGT into astrain without a complete metabolic pathway. Perhaps the chitinasegene in these strains has taken on another role, such as servingas a chitin attachment protein. Less well-supported evidenceof gene transfer includes the presence of a second chiA familygene in Photobacterium SKA34 and V. angustum S14 more closelyrelated to nonvibrio proteobacteria, as well as several gammaproteobacterialsequences that cluster within the vibrios (Fig. 2 and 3). AlthoughchiA appears subject to lateral transfer and/or duplication,there is no other gene that serves as a good indicator of growthon chitin; alternate exochitinases are either not present inall sequenced genomes (Fig. 2) or not upregulated in the presenceof chitin (Fig. 1). Additional genome sequencing of photobacteriamay reveal alternate genes/pathways of chitin metabolism.

ACKNOWLEDGMENTS

Squid pens were provided by the Marine Resources Center of theMarine Biological Laboratory, Woods Hole, MA. Thanks to S. P.Preheim, who sequenced additional housekeeping genes, and J.R. Thompson, who initiated this project.

This research was funded by the NSF-NIEHS-cosponsored WoodsHole Center for Ocean and Human Health, NSF-OCE, the DOE Genomesto Life program, and the Moore Foundation. D.E.H. also acknowledgesgraduate fellowships awarded by NSF and the Whitaker Foundation.D.G. is indebted to the Fund for Scientific Research—Flanders(Belgium) for a position as a postdoctoral fellow and researchfunding.

FOOTNOTES

* Corresponding author. Mailing address: Massachusetts Institute of Technology, 48-417, 77 Massachusetts Avenue, Cambridge, MA 02139. Phone: (617) 253-7128. Fax: (617) 258-8850. E-mail: mpolz{at}mit.edu

Published ahead of print on 12 October 2007.

Supplemental material for this article may be found at http://aem.asm.org/.

REFERENCES

Top