Cryptic Polyketide Synthase Genes in Non-Pathogenic Clostridium SPP

Modular type I polyketide synthases (PKS) produce a vast array of bacterial metabolites with highly diverse biological functions. Notably, all known polyketides were isolated from aerobic bacteria, and yet no example has been reported for strict anaerobes. In this study we explored the diversity and distribution of PKS genes in the genus Clostridium. In addition to comparative genomic analyses combined with predictions of modular type I polyketide synthase (PKS) gene clusters in sequenced genomes of Clostridium spp., a representative selection of other species inhabiting a variety of ecological niches was investigated by PCR screening for PKS genes. Our data reveal that all studied pathogenic Clostridium spp. are devoid of putative PKS genes. In stark contrast, cryptic PKS genes are widespread in genomes of non-pathogenic Clostridium species. According to phylogenetic analyses, the Clostridium PKS genes have unusual and diverse origins. However, reverse transcription quantitative PCR demonstrates that these genes are silent under standard cultivation conditions, explaining why the related metabolites have been overlooked until now. This study presents clostridia as a putative source for novel bioactive polyketides.


Introduction
Bacterial modular type I polyketide synthases (PKS) are giant multifunctional enzymes that are responsible for the biosynthesis of complex polyketides, such as macrolides, polyenes, and polyethers.These secondary metabolites possess a wide range of pharmaceutical properties including antibiotic, antifungal, antitumor, and immunosuppressive activities [1].Until recently, only PKSs of a few bacterial genera, predominantly actinomycetes and myxobacteria, were the focus of investigation.However, the number of genera known to produce polyketides has recently expanded to include Bacillus, Pseudomonas and Burkholderia spp., for example [2,3].Genetic studies have yielded detailed information on the organization and function of the PKS genes involved in the biosynthesis of polyketides [1].Bacterial modular PKS have been found to have significant homology to each other.Structurally PKS are encoded by a variable number of modules that perform a stepwise biosynthesis of carbon skeletons from simple activated acyl and malonyl acid units.Ketoacyl synthase (KS) domains in PKS are required for the condensation of extender units on a growing polyketide chain [2,4].The highly conserved nature of these KS domains can be employed to screen underexplored or neglected bacterial genome sequences for PKS genes to investigate the actual metabolic potential of hitherto neglected microorganisms, such as clostridia [3,5].
Recently, we have discovered the first antibiotic, closthioamide, from the strictly anaerobic bacterium Clostridium cellulolyticum [6].This result encouraged us to explore the virtually unexplored secondary metabolome of obligate anaerobes, specifically Clostridium spp., and screen their genome sequences for PKS genes.
Members of the genus Clostridium are gram-positive, obligately anaerobic bacteria and include prominent human pathogens, like Clostridium botulinum, and species extensively used for biofuel production, cellulose degradation, and other biotechnological applications [7].Due to their relevance as pathogens or producer strains in biotechnology, 100 Clostridium genomes have already been sequenced and published (42 completed and 58 whole genome shotgun sequences; see http://www.ncbi.nlm.nih.gov/genomes/lproks.cgi).
We observed a widespread occurrence of putative PKS genes in non-pathogenic clostridia, especially in cellulolytic species, whereas no PKS genes were found in any pathogenic Clostridium species.By RT-qPCR we have shown that several PKS coding genes appear to be silent or only basally expressed under standard cultivation conditions, giving a possible explanation as to why the related polyketides have been overlooked so far.The presented data gives an overview of the occurrence and nature of potential PKS genes in non-pathogenic clostridia and provides a foundation for the discovery of novel polyketides from a group of anaerobic bacteria with previously underestimated potential.

Results
Genome mining reveals PKS genes exclusively in nonpathogenic Clostridium spp.
First we aimed at analyzing PKS genes and investigating their relationship with Clostridium taxonomy in order to assess their ability to produce various type I PKS metabolites.In this context it should be noted that modular PKS frequently form hybrid assembly-lines with modules of nonribosomal peptide synthetases (NRPS) [8], and thus PKS/NRPS genes were also investigated in this study.To determine the PKS gene distribution in clostridia we analyzed all available genome sequences (both complete and incomplete, published sequencing projects) from the genus Clostridium for the occurrence of PKS genes.The module architectures encoded in each gene cluster was annotated using the SEARCHPKS software [9].Even though nearly 60% of the available genome sequences belong to pathogenic clostridia (six species), we could not detect any PKS genes in these genomes.In contrast, 23% of the remaining non-pathogenic genome sequences (7 out of 31 species) contained a wide variety of bacterial modular type I PKS gene clusters.The solventogenic species Clostridium acetobutylicum (CaceA) and Clostridium thermocellum (CtheA) feature relatively small putative cis-acyltransferase (cis-AT) PKS genes, whereas Clostridium cellulovorans (CcelvA), Clostridium beijerinkii (CbeiA), and Clostridium kluyveri (CkluA-C) possess putative hybrid trans-AT PKS/NRPS gene clusters (Table 1).These transacyltransferase PKS modules lack integrated AT domains [10].By far the most diverse gene cluster assemblies are found in the cellulolytic species Clostridium cellulolyticum and Clostridium papyrosolvens.C. cellulolyticum harbors three putative trans-AT hybrid PKS/NRPS gene clusters (CcelA, C, D) as well as one cryptic mixed cis/trans-AT hybrid PKS/NRPS gene cluster (CcelB).The C. papyrosolvens genome contains even more diverse gene clusters, with five putative trans-AT PKS gene clusters (CpapA, C, E, G, H), two trans-AT PKS/NRPS gene clusters (CpapB, I), and two mixed cis/trans-AT hybrid PKS/NRPS gene clusters (CpapD, F) (Table 1).Trans-AT pathways represent promising targets for natural product discovery as they are less well studied and possess unusual enzymatic features that often result in compounds with novel characteristics [11].The high frequency of these potential PKS genes in clostridia indicates that these organisms could be promising polyketide producers.

Clostridium PKS genes have unusual and diverse phylogenetic origins
For further phylogenetic analysis of the identified gene clusters we aligned all annotated KS domain sequences.In earlier studies it has been observed that cis-AT KS domains typically cluster tightly in natural product-specific clades, whereas KS domains from trans-AT systems cluster according to their substrate type [12].This results in the distribution of domains from individual pathways among different clades with many clades comprising KS domains from diverse bacterial phyla [12].Furthermore, KS domains from hybrid PKS/NRPS systems often form distinct clades in phylogenetic trees [2].It has been suggested that cis-AT PKS systems have evolved vertically from free-standing type II fatty acids, while trans-AT PKS systems seem to be patchworks acquired from diverse sources and assembled by extensive recombination events and multiple horizontal gene transfers [12].To investigate the phylogeny of the most highly conserved portions of the clostridial PKS systems, we aligned clostridial KS domains with KS domains from different phyla and PKS types (Figure 1).Surprisingly, the putative cis-AT PKS systems of the clostridia do not cluster with the most well-described cis-AT PKS systems (clade I), but fall into other clades (clades II to V) (Figure 1).The cis-AT PKS systems from C. thermocellum and C. acetobutylicum align with other unusual cis-AT PKS systems in clades II and III.The closest relatives to the KS domain of C. thermocellum (clade III) include those from putative cis-AT PKS systems within the cellulolytic anaerobe Acetivibrio cellulolyticus CD2 and from the gram-negative, strictly anaerobic Elusimicrobium minutum Pei191 [13].It appears that these cis-AT PKS from anaerobes form a novel group that evolved separately from the well studied systems within the actinomycetes.The putative cis-AT KS domain of C. acetobutylicum (clade II) shows homology with KS domains belonging to recently described systems with novel biosynthetic functions, including the cryptic cis-AT PKS of Saccharopolyspora erythraea NRRL 2338 [14].
The remaining non-canonical potential cis-AT hybrid PKS/NRPS systems in this clade II belong to a previously described group of biosynthetic systems conserved among diverse bacteria species for antifungal tetramic acid-containing macrolactams [15].The HSAF-type PKS/NRPS of Lysobacter enzymogenes is a rare example for an assembly line in which a single-module PKS assembles two separate polyketide chains that are then linked together via an amide bond of the same amino acid and further modified by tailoring redox enzymes [16].The putative trans-AT hybrid PKS/ NRPS gene clusters from the remaining non-pathogenic clostridial species cluster throughout clade V (Figure 1) with all kinds of trans-AT PKS and trans-AT hybrid PKS/NRPS systems, some with only poor bootstrap values pointing to uncharacterized related polyketides.While the pure trans-AT PKS and trans-AT hybrid PKS/NRPS systems tend to cluster closer with described trans-AT PKSs from Burkholderia and Bacillus spp.producing e.g.rhizoxin [17], bacillaene [18], or thailandamide [19] the KS domains belonging to putative mixed cis/trans-AT hybrid PKS/NRPS gene clusters (CcelB, CpapD, CpapF) are divided into the trans-AT KS domains, which cluster with described trans-AT PKSs, and the cis-AT KS domains, which cluster further apart with other clostridia PKS (Figure 1, Table 1).Taken together, the PKSs from Clostridium species appear to have unusual and diverse phylogenetic origins.

PCR targeting of diverse PKS genes in not yet sequenced Clostridium species
Although many genomes of members of the genus Clostridium (in particular the human pathogens) have been sequenced, the genomes of many species isolated from a diverse range of habitats are as not yet sequenced.To determine whether some of these not sequenced species also contain putative PKS gene clusters, we designed degenerate oligonucleotide primers to amplify DNA from KS domains within these PKS genes.The primers were designed from the consensus sequences of the aligned putative KS domain sequences from sequenced Clostridium spp.Conserved regions with sufficient sequence homology to design Clostridium KS-domainspecific primers were identified.The upstream primer was directed against the active-site CSS (59-TKG AYA CWR YNT GYT CAT C-39) and the downstream primer was designed to encompass the second active-site histidine HGTGT (59-TCT CCY AAN GWW GTW CCB GTA CCR TG-39; product size ,470 bp).A second degenerate primer pair was designed using the primer encompassing the second active-site HGTGT as upstream primer (59-CAY GGT ACV GGW ACW WCN TTR GGA GA-39) and a downstream primer directed to the conserved region FGGG, approximately 115 amino acids downstream of the HGTGT histidine (59-ATT KGW RCC KCC SRM ACC RAA-39; product size ,370 bp).Primer binding sites were distinct from other degenerate primers typically used to screen for KS domains in bacterial genomes.However, due to the homology amongst active site regions within KS domains, correct amplicons were also obtained from Streptomyces as well as gram-negative Burkholderia spp.harboring type I PKS genes (data not shown).In order to show that these primer sets were able to amplify KS domains from the clostridia, PCR amplification of KS domains from the reference species C. cellulolyticum and C. kluyveri was performed using both primer sets, and the products were cloned and sequenced.The resulting sequences were identified as known KS domains from these bacteria.In contrast, PCR amplification of DNA from the sequenced C. butyricum, which does not possess any putative PKS genes and thus served as a negative control, gave no product (data not shown).The primer sets were then used to screen taxonomically diverse non-sequenced Clostridium strains from terrestrial as well as marine environments to get an overview on the occurrence of modular type I PKS genes in non-pathogenic clostridia (Table S1).All chosen Clostridium spp.strains were cultured in media recommended from the supplier (DSMZ) under strict anaerobic conditions, and their DNA was isolated.PCR products were cloned and sequenced and the resulting sequences were queried with BLASTX software (http://blast.ncbi.nlm.nih.gov/Blast.cgi)).In some cases different KS domains from one Clostridium strain were amplified.Out of fifteen cultivated nonsequenced, non-pathogenic Clostridium species three (C.hungatei, C. chartatabidum, and C. akagii) were found to possess putative PKS genes.The deduced sequences of the KS domains showed significant similarity ($75%) to published and annotated KS domain sequences from established polyketide producers (Table 2).
Adding these three Clostridium species to the ones that we have identified as potential sources for novel polyketides indicates that the frequency of PKS genes in non-pathogenic clostridia may be as high as 20%.
The deduced KS sequence from DNA fragment cloned and sequenced from C. hungatei (DSM 14427, isolated from soil under a pile of rotting wood chips; amplified with degenerate primer set two) shows high homology (similarity of 88%) to a cis-AT bketoacyl synthase from Acetivibrio cellulolyticus CD2 and a similarity of 83% to a putative Bacillus cereus subsp.cytotoxicus NVH 391-98 trans-AT beta-ketoacyl synthase (Table 2)  synthase from Acetivibrio cellulolyticus CD2 (Table 2).Taken together, these observations are consistent with the phylogeny of KS domains from sequenced Clostridium species, where the novel putative KS domains of non-sequenced species show only limited sequence similarity to well characterized PKS genes of other genera.This further highlights the distinctiveness of the putative polyketide synthases identified within the clostridia and reinforces their potential as producers of diverse polyketides.

Incongruence of PKS gene occurrence with taxonomy
Surprisingly, we observed that potential PKS assembly lines encoded in the genomes of bacteria belonging to the genus Clostridium is restricted to non-pathogens.Thus, we aimed at further correlating the distribution and types of PKS genes in clostridia with the phylogeny of the genus.A multiple alignment of 16S ribosomal DNA (16S rDNA) sequences of Clostridium spp.strains was used to construct a phylogenetic tree representing clostridial taxonomy (Figure 2).We made a multiple alignment by ClustalW and a phylogenic tree using MEGA version 3.1 with the NJ algorithm [20].Alignment of the 16S rDNA sequences shows that several separate clades are formed and that there is no distinct separation of pathogens and non-pathogens.C. difficile strains, for example, have a larger phylogenetic distance to other pathogenic Clostridium species than to non-pathogens.Furthermore, there seems to be little correlation between the evolution of the bacteria and the presence of putative PKS systems.This is shown for the occurrence of putative PKS genes in fully sequenced strains as well as the investigated non-sequenced strains.Particularly noteworthy is the high number of PKS genes in the cellulolytic organisms C. cellulolyticum and C. papyrosolvens.These results strongly suggest that Bootstrap values above 50% from 1000 re-samplings (neighborhood-joining (NJ) algorithm) are indicated at the nodes.N marks sequences from Clostridium species.(List of GenBank accession numbers for KS domains: see table S2).doi:10.1371/journal.pone.0029609.g001the clostridia have acquired these PKS gene clusters via horizontal gene transfer independent of strain evolution.This is further supported by the observation that in the proximity of nine out of the twenty identified putative PKS gene clusters in clostridia insertion sequences or other genetic mobility sequences are located.Several potential transposase and/or integrase genes are flanking these gene clusters.This observation correlates well with the hypothesis that trans-AT PKS systems are patchworks acquired from diverse sources and assembled by extensive recombination events and multiple horizontal gene transfers [12].

Cryptic PKS gene clusters are silent under standard cultivation conditions
Although the presence of putative PKS and hybrid PKS/NRPS gene clusters within members of the clostridia is indicative of the potential to produce novel compounds, expression of these genes is of course a prerequisite for secondary metabolite production.Thus, we investigated the expression of representative PKS genes (coding for KS domains) from C. cellulolyticum, C. papyrosolvens, and C. acetobutylicum.Each Clostridium spp.strain was cultivated under standard growth conditions and total RNA was isolated from cells harvested in the mid-logarithmic growth phase.PKS gene expression was measured using reverse transcription-quantitative PCR (RT-qPCR) with normalization against 16S rDNA using the expression values of the related non-template controls for calibration [21].This analysis indicated that in the three organisms tested, PKS gene clusters appear to be silent or only basally expressed under standard laboratory conditions (Figure 3, Figure S1).

Discussion
Clostridium is one of the largest bacteria genera including over 150 metabolically diverse species that occur ubiquitously in soil and in the intestine of higher organisms.The aim of this study was to gain insight into the potential of anaerobic Clostridium spp. to produce polyketide metabolites.We analyzed all available Clostridium genome sequences and found that the presence of putative PKS genes is clearly restricted to non-pathogenic organisms.In total, 23% of the non-pathogenic genome sequences (7 out of 31 species) harbored diverse modular type I PKS gene clusters.This frequency is supported by PCR screening of fifteen cultivated non-sequenced, non-pathogenic Clostridium species, out of which three possess type I PKS genes.Interestingly, a phylogenetic analysis revealed that Clostridium PKS genes have unusual and diverse phylogenetic origins.We not only detected putative cisand trans-AT PKS genes, but also genes for unusual types of modular PKS, which fall into novel clades in the phylogenetic tree.Despite the exclusive distribution of PKS genes in non-pathogenic clostridia, the occurrence of PKS genes is not congruent with taxonomic relationships.Thus, it is plausible that the PKS gene clusters have evolved independent of strain evolution, and their diversity points to diverse sources involving extensive recombination events and multiple horizontal gene transfers.The finding that the Clostridium PKS systems apparently have unusual and diverse phylogenetic origins also suggests that the pathways encoded lead to diverse polyketide metabolites.However, we found that all hitherto investigated cryptic PKS gene clusters are silent under standard cultivation conditions.As has been shown for C. cellulolyticum, particular signaling compounds, nutrients or other unknown environmental triggers are be needed to activate the apparently silent closthioamide biosynthesis genes [6].This phenomenon is well known for other metabolic pathways in bacteria and fungi [22] and is plausible in an anaerobic world where energy is a rare commodity.The dormant state of these putative PKS gene clusters might also explain why the related metabolites have been overlooked until now.Whereas the high frequency of PKS genes in non-pathogenic clostridia indicates that these organisms could serve as promising polyketide sources, the on-going challenge is to understand the regulatory mechanisms and to find ways to activate gene expression.This may either involve genetic engineering including heterologous expression or applying particular stimuli for inducing gene cluster expression in the wild type.It stands to reason that non-pathogenic clostridia may use more than just the ability to live in anaerobic niches to be as successful as they are in their habitats and that the chemical weapons they might have developed could be novel bioactive molecules.The recent discovery of the first antibiotic from the strictly anaerobic world, closthioamide, encourages future research in this direction [6].Furthermore, we introduced degenerate primer pairs that can be used to screen further non-pathogenic Clostridium species for more potential PKS genes.Taken together, this study provides an insight into the distribution and diversity of PKS genes in non-pathogenic clostridia and facilitates the discovery of novel polyketides from a group of strict anaerobic bacteria with previously unexplored potential.

Bacterial strains and culture conditions
Table S1 contains a list of each of the strains utilized in this study.Strains were obtained from DSMZ GmbH (Braunschweig) as stock cultures and grown anaerobically in serum glass bottles sealed with butyl rubber stoppers and aluminium crimps in media derived from recommended media of the DSMZ GmbH at the appropriate temperature without shaking.Clostridium cellulolyticum  Strains marked with N possess PKS genes.NA: not appointed; + indicates new sequenced KS domain sequences; -indicates no positive PCR results.

Isolation of total RNA and gene expression analysis
Total RNA was isolated from 50 mL of culture of C. cellulolyticum DSM 5812, C. papyrosolvens DSM 2782 and C. acetobulylicum DSM 792, respectively.RNAprotect Bacteria Reagent (Qiagen GmbH, Hilden) was quickly added to the samples and cell pellets were harvested following the manufacturer's protocol, and stored at 220uC.For RNA isolation, cells were resuspended in TE buffer (30 mM Tris-HCl, 1 mM EDTA, pH 8.0) containing 15 mg mL 21 lysozyme (1 volume), vortexed for 10 s and incubated at room temperature for 1 h.RNA was purified using RNeasy Mini Kit (Qiagen GmbH, Hilden, Germany) following the manufacturer's protocol and afterwards stored at 280uC.DNA was removed by digestion utilizing the TURBO DNA-free Kit (Applied Biosystems, Ambion Inc., Austin, USA) following the manufacturer's protocol with incubation for 40-90 min at 37uC.RT-qPCR experiments were performed when RNA samples were confirmed to be DNA free, as indicated by a lack of product formation in a PCR amplification using DNA polymerase and the RT-PCR primers on an RNA template.RNA quantity was measured in a spectrophotometer (NanoDrop, ND-1000; PeQLab Biotechnologie GmbH, Erlangen).One-step reverse transcription-quantitative PCRs were performed using a One Step SYBR PrimeScript RT-PCR Kit (TAKARA Bio Inc., Otsu Shiga, Japan).100 ng of RNA were applied to each RT-qPCR reaction mixture.16S rRNA gene of each species was used as an internal standard for calculation of expression levels.For negative controls, experiments were performed using water instead of RNA, and for non-template controls without RT enzyme mix, respectively.The reactions were executed in an Eppendorf realplex 2 mastercycler (epgradientS, Eppendorf AG, Hamburg) in triplicate for each sample.Cycling parameters included an initial DNA denaturation step at 95uC for 2 min, followed by 40 cycles with DNA denaturation at 95uC for 5 s, primer annealing at 60uC for 15 s, and elongation at 72uC for 10 s.Relative expression levels for each mRNA sample were obtained using the DDCq method via normalization to 16S rRNA applying the formula 2 2(DCq,q-DCq,cb) for all samples.With DC q,q = C q,PKS target 2C q,16SrRNA and DC q,cb = C q,calibrator 2C q,16SrRNA .The expression values of the related non template controls of each sample were used for calibration (maximum value: C q,calibrator = 40).The appendent value of the calibrator is set as 1 (2 2(DCq,cb-DCq,cb) = 1) [21].

Isolation of genomic DNA
1.5 mL of cultured bacterial cells were centrifuged and the pelleted cells were resuspended in TE buffer and used for isolation of genomic DNA with MasterPure TM Gram Positive DNA Purification Kit (Epicentre Biotechnologies, Madison, Wisconsin, USA).The precipitated DNA was finally resuspended in 20 mL TE buffer and directly used for PCR analysis.
. The following closest homologues in the BLASTX search are all KS domains from Clostridium species (C.kluyveri, C. papyrosolvens and C. cellulolyticum) with similarities .78% to the PCR fragment.A sequence amplified from the genomic DNA of C. akagii (DSM 12554, isolated from beech litter) with primer pair one shows significant similarity 86% to a putative KS sequence of Bacillus thuringiensis serovar huazhongensis BGSC 4BD1 and 79% similarity with a KS domain of a b-ketoacyl synthase from C. cellulolyticum (CcelDKS2).An amplicon from C. akagii generated with primer set two shows 82% similarity to a section of a gene from C. papyrosolvens annotated as AMP-dependent synthetase and ligase and the next closest homologue (78% similarity) is a potential beta-ketoacyl synthase encoded in the C. cellulolyticum genome (CcelAKS13).Furthermore, the deduced KS sequence of the PCR fragment from C. akagii is to 78% similar to a KS domain from a putative PKS module encoded in the Paenibacillus polymyxa E681 genome (Table2).A PCR product of C. chartatabidum (DSM 5482, isolated from ovine rumen) generated with primer pair two codes for a KS fragment with highest similarity (91%) to the deduced KS from C. kluyveri DSM 555 and NBRC 12016 (CkluAKS1), and has significant (83%) similarity to a putative cis-AT beta-ketoacyl

Figure 1 .
Figure 1.Phylogenetic tree constructed with MEGA 3.1 analysis of KS domain protein sequences from bacterial PKS genes.

Figure 3 .
Figure 3. Relative quantity of the mRNA level determined by RT-qPCR of Clostridium species PKS genes (KS domains).Relative quantity is given as the log 2 of 2DDCq.Constitiutive expression of the 16S rRNA of each species was used as an internal control and the value of the related non template control of each sample was set to 1 for calibration.doi:10.1371/journal.pone.0029609.g003

Figure 2 .
Figure 2. Phylogeny of 16SrDNA sequences of Clostridium species.Bootstrap values above 70% from 1000 re-samplings (NJ algorithm) are indicated at the nodes.Saccharopolyspora erythraea NRRL 2338 was used as an outgroup for this analysis.Strains shaded in gray are pathogens.

Table 2 .
BLASTX results of sequences amplified from not yet sequenced Clostridium species genomic DNA using KS domain specific degenerated primers. doi:10.1371/journal.pone.0029609.t002