Psychrotrophic Strain of Janthinobacterium lividum from a Cold Alaskan Soil Produces Prodigiosin

Abstract

We have explored the microbial community in a nonpermafrost, cold Alaskan soil using both culture-based and culture-independent approaches. In the present study, we cultured >1000 bacterial isolates from this soil and characterized the collection of isolates phylogenetically and functionally. A screen for antibiosis identified an atypical, red-pigmented strain of Janthinobacterium lividum (strain BR01) that produced prodigiosin when grown at cool temperatures as well as strains (e.g., strain BP01) that are more typical of J. lividium, which produce a purple pigment, violacein. Both purple- and red-pigmented strains exhibited high levels of resistance to β-lactam antibiotics. The prodigiosin pathway cloned from J. lividium BR01 was expressed in the heterologous host, Escherichia coli, and the responsible gene cluster differs from that of a well-studied prodigiosin producer, Serratia sp. J. lividum BR01 is the first example of a prodigiosin-producer among the β-Proteobacteria. The results show that characterization of cultured organisms from previously unexplored environments can expand the current portrait of the microbial world.

Introduction

Soil microbial communities are the most phylogenetically and functionally diverse communities on Earth. A single gram of soil may contain 10⁹ to 10¹¹ bacteria distributed among thousands of species, which play a significant role in biogeochemical cycling, bioremediation, and the production of antimicrobials and industrially useful small molecules and enzymes (Whitman et al., 1998; Janssen, 2006). Despite estimates that fewer than 1% of soil bacteria are readily culturable (Amann et al., 1990), culturing from soil continues to produce a new understanding of genetic and biochemical diversity, and more intensive efforts to culture from unexplored soils will accelerate the rate of discovery in microbiology. Combining traditional culturing with metagenomics and other culture-independent methods will be a powerful strategy to enhance our understanding of soil microbial communities.

As part of an effort to define the diversity of bacteria in undisturbed cold soils, we established the Alaskan Cold Microbial Observatory in the Bonanza Creek Long-Term Ecological Research site on an island in the Tanana River in interior Alaska where the soil temperature rarely exceeds 16°C (Viereck et al., 1993). The bacteria in this habitat represent a reservoir of unsampled biodiversity. Although there has been intense interest in sampling from extreme, cold environments (Shi et al., 1997; Zhou et al., 1997; Priscu et al., 1999; Christner et al., 2001; Schadt et al., 2003; Lavire et al., 2006; Alekhina et al., 2007; Hansen et al., 2007), less attention has focused on cold, nonpermafrost soils.

Previous study of this Alaskan soil microbial community using culture-independent approaches revealed high species richness based on the sequences of 1033 16S rRNA genes from a single 0.5-g sample of soil (Schloss and Handelsman, 2006). The sequences were distributed among 19 bacterial phyla, 4 of which had no cultured representatives. The Alaskan soil contains an estimated 5000 different species per gram, using ≥97% identity in 16S rRNA gene sequence to define members of the same species (Schloss and Handelsman, 2006). Moreover, functional metagenomic analyses indicate diverse genes involved in LuxR-based quorum sensing (Williamson et al., 2005a) and resistance to β-lactam antibiotics (Allen et al., 2009).

The previous culture-independent analyses are complemented here by a culture-dependent analysis of bacterial isolates collected from the Alaskan Cold Microbial Observatory. Two pigmented isolates, one red and one purple, were identified as Janthinobacterium lividum. The characterizations of these strains are presented here, including the genetic analysis of pigment production and antibiotic resistance. We report that the red-pigmented isolate produces prodigiosin and the purple-pigmented isolate produces violacein, both in a temperature-dependent manner.

Methods

Culture conditions

A soil core was obtained from a site, designated BP-1, on an island in the Tanana River in the Bonanza Creek Long Term Ecological Research site approximately 30 km southwest of Fairbanks, Alaska, United States (64°40.24′ N, 148°14.46′ W) (Williamson et al., 2005a; Schloss and Handelsman, 2006). The top 3 cm of the soil core was removed. Two soil horizons were visible: a silt layer that was deposited in a past flooding event topped by an organic layer resulting from years of biological activity after deposition of the silt. Each horizon was passed through a 5-mm sieve to remove large roots and organic debris. A 1-g soil sample from each horizon was suspended in 10 mL sterile deionized water and sonicated for 30 s in a bath sonicator (Model 2210 Branson, Danbury, CT). The sonicated soil suspension was serially diluted, and the dilutions were plated on 10%, 1%, and 0.1% tryptic soy agar (TSA) medium; nutrient broth gellan gum; VL55 gellan gum (Sait et al., 2002); and 10% tryptic soy gellan gum (TSG). All culture plates were incubated at 6°C, 12°C, and 28°C for up to 2 months. At various intervals, colonies were picked, inoculated into the appropriate broth media, and stored in 96-well plates in 15% glycerol at −80°C.

An isolate of J. lividum that produced red colonies was isolated on 10% TSG at 6°C in the above collection, and this strain was designated BR01. This and other J. lividum isolates were cultured on 1/10th-strength TSA at temperatures no higher than 25°C, and were transferred to 4°C for several days to allow for pigment development. To improve pigment production, J. lividum isolates were also cultured on PGA medium (0.5% peptone, 1% glycerol, and 1.5% agar).

16S rRNA gene sequencing and analysis

Frozen stocks of 273 strains isolated on 1/10th-strength TSA were streaked for isolation and used to inoculate 1/10th-strength tryptic soy broth; broth cultures were grown until turbid. The cultures were lysed by freezing at −20°C followed by thawing. Genomic DNA from the lysed cells was used as the template for PCR with the primers 27f (5′-AGRGTTTGATYMTGGCTCAG) and 1492r (5′-GGYTACCTTGTTACGACTT) (Integrated DNA Technologies, Coralville, IA). Cycle conditions were an initial 5-min denaturation at 95°C, followed by 30 cycles of 45 s at 94°C, 45 s at 55°C, 90 s at 72°C, and a final extension of 10 min at 72°C. PCR products were purified using AmpPure (Agencourt Bioscience, Beverly, MA) and sequenced using the 27f primer and BigDye version 3.1 (Applied Biosystems, Foster City, CA). BLASTN (Altschul et al., 1990, 1997) was used to determine the phylum-level designation of each isolate.

Genomic DNA was purified from cultures of J. lividum according to the manufacturer's instructions with the GenElute Bacterial Genomic Kit (Sigma-Aldrich, St. Louis, MO), and the 16S rRNA gene was amplified, purified, and sequenced as described above. Full-length sequences were obtained for the J. lividum 16S rRNA genes using primers 27f, 1492r, 787r (5′-CTACCAGGGTATCTAAT), and 787f (5′-ATTAGATACCCYGGTAG) (Integrated DNA Technologies).

Screen for antibiotic production

To identify putative antibiotic-producing isolates, the Alaskan soil culture collection was replica printed from storage onto three types of media: 1/10th-strength TSA, VL55, and 1/10th-strength TSG. Duplicates were made of each plate to test the isolates for inhibition of Bacillus subtilis 168 and Pseudomonas putida F117. The plates were incubated at 6°C, 12°C, or 28°C until visible colonies formed. Colonies were fixed by spraying 0.5% water agar in a thin layer across the plate. A 100-μL overnight culture of each tester strain, B. subtilis 168, and P. putida F117, was mixed with 5 mL of soft Luria-Bertaini agar (LBA) (0.8% agar) and poured over each plate. After overnight incubation at 28°C, isolates that produced a zone of inhibition were then identified by 16S rRNA gene analysis as described above. J. lividum BR01 and BP01 were tested in the same manner for antibiosis against additional tester strains: Salmonella enterica, Streptomyces griseus, Bacillus cereus UW85, Bacillus mycoides, E. coli DH5α, Erwinia herbicola, and Pseudomonas aeruginosa.

In vivo transposon mutagenesis

The EZ-Tn5 <DHFR-1> transposome (Epicentre, Madison, WI) was introduced into electrocompetent J. lividum BR01 and BP01 according to the manufacturer's instructions. The mutant phenotype was confirmed and genomic DNA was isolated from each nonpigmented mutant (GenElute™ Bacterial Genomic DNA Kit; Sigma-Aldrich) and cloned into the PstI site of pGEM-3zf(+) (Promega, Madison, WI). Ligation reaction products were electroporated into E. coli DH5α and plated onto LBA supplemented with 15 μg/mL of trimethoprim (to select for the transposon) and 100 μg/mL ampicillin (to select for the vector). Thousands of colonies from each mutagenesis reaction were screened. Plasmids were isolated from the resulting clones according to manufacturer's instructions with the ExpressMatrix (Bio101; Morgan, Irvine, CA). The manufacturer-supplied transposon-based sequencing primers were used to determine the sequence of the insertion site.

J. lividum fosmid library construction and analysis

Genomic DNA was isolated from J. lividum BR01 and BP01 (Table 1) using the GenElute Bacterial Genomic Kit (Sigma-Aldrich). A Hamilton syringe was used to shear the genomic DNA to fragments of 40 kb, which served as inserts for fosmid libraries using the CopyControl Fosmid Library Production Kit (Epicentre) (Table 1). Among 1000 colonies, clones harboring the pigmentation genes (pBRred and pBPpur) (Table 1) were identified by growing the two libraries for 40 h at room temperature and identifying red or purple colonies. Clones harboring genes conferring resistance to β-lactam antibiotics (pBRbeta and pBPbeta) (Table 1) were identified by growing the two libraries overnight at 37°C on LBA containing ampicillin (100 μg/mL).

Table 1.

Strains and Plasmids Used in This Study

Strain or plasmid	Relevant characteristicsa	Reference
Strains
Janthinobacterium lividum  strain BR01	Isolated from Alaskan soil; produces red pigment (prodigiosin) when grown at low temperatures	This study
J. lividum strain BP01	Isolated from Alaskan soil; produces purple pigment (violacein) when grown at low temperatures	This study
Escherichia coli  strain EPI300	DH10B derivative with inducible promoter in front of trfA; induction increases copy number of pCC1FOS from oriV	Epicentre, Madison, WI
E. coli strain DH5α	General purpose lab strain	Hanahan (1983)
Plasmids
pCC1FOS	ChlR; fosmid cloning vector with inducible copy number when propagated in E. coli EPI300	Epicentre
pGEM-3zf(+)	AmpR; standard cloning vector	Promega, Madison, WI
pBRred	ChlR; subclone of J. lividum BR01 in pCC1FOS that turns E. coli EPI300 pink at cool temperatures	This study
pBPpur	ChlR; subclone of J. lividum BP01 in pCC1FOS that turns E. coli EPI300 purple at cool temperatures	This study
pBRbeta	ChlR; ampR; subclone of J. lividum BR01 in pCC1FOS that confers resistance to β-lactam antibiotics on E. coli EPI300	This study
pBPbeta	ChlR; ampR; subclone of J. lividum BP01 in pCC1FOS that confers resistance to β-lactam antibiotics on E. coli EPI300	This study

^aAntibiotic resistance phenotypes are abbreviated as follows: Chl^R, chloramphenicol resistant; Amp^R, ampicillin resistant.

Fosmid DNA was isolated from induced overnight cultures of pBRred, pBPpur, pBRbeta, and pBPbeta by alkaline lysis with buffers P1, P2, and N3 (Qiagen, Valencia, CA) followed by ethanol precipitation (Sambrook and Russell, 2001). To identify the genes responsible for the respective phenotypes, fosmid DNA was mutagenized with a transposon via the GPS-1 Genome Priming System (New England Biolabs, Ipswich, MA). Each mutagenesis reaction was transformed into E. coli EPI300 and plated on LBA supplemented with antibiotics to select for the fosmid (20 μg/mL chloramphenicol) and transposon (20 μg/mL kanamycin). DNA was isolated from mutants that had lost the respective phenotype, and the transposon-based sequencing primers were used to sequence the genes of interest using BigDye version 3.1 (Applied Biosystems). Additional random insertion mutants were prepared to sequence the entire fosmid insert to at least threefold coverage. The region of pBRred containing pigJK was sequenced to ∼15-fold coverage, and both the pigBC and pigJK junctions were PCR amplified and sequenced to confirm the annotation. Sequences were assembled using the STADEN software package (v. 1.4.0; http://staden.sourceforge.net/) and annotated using open reading frame (ORF) Finder (www.ncbi.nih.gov/gorf/gorf.html). ORF maps were drawn in SeqBuilder (Lasergene; DNASTAR, Madison, WI).

Antibiotic susceptibility

Minimum inhibitory concentration (MIC) assays were performed according to Clinical Laboratory Standards Institute guidelines (NCCLS, 2004). Briefly, serial dilutions of antibiotics were made in a 96-well plate from 512 to 0.5 μg/mL, with no antibiotics in the last well. The final volume in each well was 90 μL in Mueller–Hinton broth (Becton, Dickinson and Company, Sparks, MD). Ten microliters containing ∼1 × 10⁵ cfu of J. lividum BR01, J. lividum BP01, E. coli pBPpur, E. coli pBRbeta, and E. coli pBPbeta were added to the appropriate wells. MICs were tested with the following β-lactam antibiotics: ampicillin (Research Products International Corp., Mt. Prospect, IL), carbenicillin (Fisher Scientific, Fair Lawn, NJ), amoxicillin, penicillin G, piperacillin, cephalexin, cefamandole, and ceftazidime (Sigma, St. Louis, MO). Assays were performed in duplicate in three separate experiments using independent cultures with E. coli EPI300 carrying the pCC1FOS vector as the reference strain (Epicentre). Growth was evaluated within 24 h.

Purification of prodigiosin from J. lividum BR01

A 500 μL overnight culture of J. lividum BR01 grown in LB medium at 30°C was used to inoculate 500 mL of nutrient broth containing 1% glycerol in a 4 L Erlenmeyer flask, which was then incubated at 10°C with shaking at 250 rpm for 11 days, by which time the culture had become red-orange. Cells were removed by centrifugation, and the cell pellet, which retained most of the red color, was extracted three times by resuspending the cells in 50 mL of methanol, incubating on a rocker overnight, and centrifuging at 10,000 rpm for 15 min at 4°C. The final methanol extract was dried, and 486 mg of material was recovered. This extract was purified by column chromatography with a silica gel using a step gradient of dichloromethane and methanol. The active fractions were further purified by high performance liquid chromatography (HPLC) on a C-18 column using a water–methanol mixture acidified with 0.05% trifluoroacetic acid. The HPLC elution started with 4 min at 75% methanol and then increased to 100% methanol over 10 min followed by 5 min at 100% methanol (prodigiosin: tR = 11.3 min; heptylprodigiosin: tR = 15.7 min).

Purification of violacein from J. lividum BP01

Eleven 500 mL Erlenmeyer flasks each containing 50 mL of 0.5% (w/v) Bacto peptone and 1% (v/v) glycerol were inoculated with 50 μL of an overnight culture of J. lividum BP01. Cultures were incubated at room temperature on a platform shaker at 200 rpm. After 13 days the cultures were dark purple. The cultures were combined, frozen, and lyophilized. Approximately 150 mL of methanol was added to the lyophilized culture and this mixture was incubated at 4°C for approximately 1 h on a rocker. The mixture was centrifuged and the methanol extract was decanted. This extraction process was repeated twice except that the incubation at 4°C was extended to 24 h. The methanol extracts were combined and dried to yield approximately 7.5 g of oil, which was purified according to the protocol outlined in Brady et al. (2001). Briefly, the oil was dissolved in 800 mL of 80% methanol and extracted three times with 800 mL of hexanes. The aqueous portion was diluted to 50% methanol and extracted twice with 1000 mL of chloroform followed by two more extractions with 500 mL of chloroform. The chloroform extract yielded 143 mg of purple oil. This extract was subjected to normal-phase column chromatography on a silica gel. Fractions were eluted from the column using a hexanes:ethyl acetate gradient acidified with 0.01% trifluoroacetic acid (TFA). The purple fractions from the silica column were combined and further purified using reverse-phase HPLC on a C-18 column (isocratic: 40% acetonitrile: 60% water acidified with 0.05% TFA; violacein: tR = 9.2 min).

GenBank accession numbers

pBRred and pBPpur sequences have been deposited in GenBank under accession numbers EF063590 and EF063591, and pBRbeta and pBPbeta sequences have been deposited under accession numbers EF063589 and EF063592, respectively. The 16S rRNA gene sequences for J. lividum BR01 and BP01 have been deposited in GenBank under accession numbers EU330448 and EU330449, respectively. GenBank accession numbers of each unique 16S rRNA gene sequence of the bacteria that were isolated on 10% TSA have been deposited under numbers HM113566-HM113684. These sequences are associated with the Alaskan soil metagenome project (NCBI GenomeProject database ID 28853).

Results

Biodiversity of Alaskan soil includes a red-pigmented J. lividum strain

To culture diverse bacteria, multiple media types, temperatures, and incubation lengths were employed. Dilute nutrients, such as those found in VL55 media, and long incubation times have been shown to increase the diversity of culturable isolates from soil (Janssen et al., 2002; Sait et al., 2002). Multiple cool temperatures were chosen for bacterial cultivation to capture a range of temperatures experienced in the Alaska soil habitat.

To assess the biodiversity of the Alaskan soil culture collection, the 16S rRNA genes of a subset of isolates were sequenced. Partial 16S rRNA sequences were obtained from 273 bacteria that were isolated on 10% TSA; the sequences represented 26 different genera distributed among the Proteobacteria (37.7%), Bacteroidetes (48%), Actinobacteria (12.8%), and Firmicutes (1.5%) (Fig. 1). The 16S rRNA gene sequences of a different subset of 138 randomly selected isolates, representing the various media used in the initial isolation, were sequenced and revealed greater abundance of Proteobacteria (52.2%), Actinobacteria (20.3%), and Firmicutes (2.8%), and fewer Bacteriodetes (24.6%) (data not shown) than were detected among the 16S rRNA gene sequences from 10% TSA isolates alone. Overall, far fewer phyla were isolated by culturing than by a culture-independent analysis of the same soil (Fig. 1).

FIG. 1.

The relative abundance of each phylum in the Alaskan soil culture collection isolated on 1/10th-tryptic soy agar (n = 273 16S rRNA gene sequences) and in a culture-independent, PCR-derived clone library (n = 1033 16S rRNA gene sequences) (Schloss and Handelsman, 2006). White bars represent sequences identified by culture-dependent methods, and black bars represent sequences identified by culture-independent methods. Candidate phyla OP10, WCHB1, BD group, and ACE have no sequenced representatives.

A screen of the Alaskan culture collection for antimicrobial activity identified six putative antibiotic producers. All inhibited the growth of B. subtilis 168 but not of P. putida F117. One isolate that produced a red pigment was further characterized.

The sequence of the 16S rRNA gene of the red antibiotic producer revealed that it was 100% identical to J. lividum (GenBank accession DQ640007). An additional 24 J. lividum isolates in the culture collection were identified with similar 16S rRNA gene sequences. Although the definition of J. lividum specifies that it produces a purple pigment (Janthinobacterium = purple bacterium, lividum = bluish), 9 of 24 J. lividum Alaskan soil isolates produced a red pigment (e.g., strain BR01) and the remaining 15 isolates produced the purple pigment typical of J. lividum (e.g., strain BP01) (Table 1). The full-length 16S rRNA gene sequences of a representative purple and red strain, J. lividum BP01 and J. lividum BR01, differed by two nucleotides over 1455 base pairs. Neither the red nor the purple pigments were visible in cultures grown above 20°C, although growth rate was not affected by temperature.

Given the difference in pigment production despite nearly identical 16S rRNA sequences, we examined other phenotypes of J. lividum BR01 and BP01. Except for the unique red pigmentation of nine J. lividum isolates and the temperature dependency of pigment production, the phenotypes of the Alaskan soil J. lividum isolates were similar to those of previously reported J. lividum isolates (De Ley et al., 1978). Phenotypes tested include motility, proteolysis, lactose fermentation (data not shown), antibiosis, and antibiotic resistance. Pigmented J. lividum BR01 and BP01 inhibited the growth of S. enterica, S. griseus, B. subtilis 168, B. cereus UW85, and B. mycoides, but not E. coli DH5α, E. herbicola, or P. aeruginosa. No growth inhibition was observed for any tester strain when the J. lividum strains were grown at temperatures that prevented the appearance of pigmentation. Both strains of J. lividum (BR01 and BP01) exhibited high levels of resistance to seven of eight β-lactam antibiotics tested (Table 2).

Table 2.

Minimum Inhibitory Concentration (μg/mL) of Eight β-Lactam Antibiotics on Janthinobacterium lividum and Subclones After 24 h of Growth

β-Lactam antibiotic	Minimum inhibitory concentration of strain or constructa
	J. lividum BR01	J. lividum BP01	E. coli pBRbeta	E. coli pBPbeta	E. coli pCC1FOS (empty vector)
Amoxicillin	256	128	>512	>512	4
Ampicillin	256	512	512	>512	8
Carbenicillin	128	256	128	512	4
Penicillin G	128	64	>512	>512	32
Piperacillin	>512	>512	>512	>512	4
Cephalexin	256	256	256	256	8
Cefamandole	256	256	256	512	0.5
Ceftazidime	8	16	8	4	0.5

^aNames are defined in Table 1.

Genetic characterization of β-lactam resistance in J. lividum BR01 and BP01

The genetic basis for β-lactam antibiotic resistance in J. lividum BR01 and BP01 was explored by selecting a fosmid clone library of each strain on ampicillin, followed by transposon mutagenesis and sequence analysis of the clones that conferred resistance on E. coli. We identified five clones from the J. lividum BR01 library and three from the BP01 library, and a restriction digest revealed overlap between ampicillin-resistant clones from the same library (data not shown). Therefore, a representative clone from each library was characterized and sequenced (clones designated pBRbeta and pBPbeta). Both pBRbeta and pBPbeta conferred high levels of β-lactam antibiotic resistance on E. coli, and in some cases (when tested on amoxicillin and penicillin G, e.g.) exhibited greater resistance than did the J. lividum strain from which the genes were isolated (Table 2). Sequencing revealed much congruency between pBRbeta and pBPbeta; the former has a 33.1-kb insert containing 32 ORFs, the latter has a 38.9-kb insert containing 33 ORFs, and they contain 22 ORFs in common (Fig. 2). Within the overlapping region, pBPbeta and pBRbeta are syntenous except for one additional ORF in the pBPbeta sequence that is not found in pBRbeta (Fig. 2). The 22 overlapping ORFs are, on average, 95.2% (SD = 3.9) identical between the two fosmids. Both encode a putative β-lactamase repressor protein, a putative carboxypeptidase, and a putative metallo-β-lactamase, and the amino acid sequences of the pairs of homologs have 99.2%, 95.4%, and 93.3% identity, respectively. The predicted protein sequence encoded by the two metallo-β-lactamase genes share 81% identity with a previously published metallo-β-lactamase (THIN-B) from J. lividum (Rossolini et al., 2001). The genes encoding the metallo-β-lactamases from J. lividum BR01 and BP01 are designated blaLRA-20 and blaLRA-21, respectively (bla, β-lactamase; LRA, β-lactam resistance from Alaskan soil).

FIG. 2.

Maps of open reading frames (ORFs) from pBRbeta (33,133 bp) and pBPbeta (38,875 bp), aligned across homologous regions. The ORFs with dotted diagonal lines in each map encode putative β-lactamase repressor proteins (A), carboxypeptidases (B), and metallo-β-lactamases (LRA-20 and LRA-21). The asterisk below the map of pBPbeta designates the ORF that is missing in pBRbeta. ORFs filled with black, gray, and black dots have homology to hypothetical ORFs, regulatory proteins, and transport proteins, respectively. Those ORFs lacking a color have homology to a methyltransferase (1), hemerythrin-like protein (2), oxidoreductase (3), branched-chain amino acid aminotransferase (4), ribonuclease I (5), AMP-dependent synthetase and ligase (6), electron chain protein (7), adventurous gliding motility protein (8, 11), TPR repeat protein (9), alcohol dehydrogenase (10), ferrichrome-iron receptor (12), and isochorismate hydrolase (13).

Isolation and characterization of J. lividum BR01 pigment-producing gene cluster

The red compound produced by J. lividum BR01 was purified, and a nuclear magnetic resonance (NMR) and liquid chromatography coupled with tandem mass spectrometry (LC-MS/MS) analysis indicated that it was a mixture of prodigiosin and heptylprodigiosin (Fig. 3A, B). Transposon mutagenesis identified two genes, smpB and pigE, that were required for pigmentation and antibiosis. SmpB forms a complex with SsrA RNA that targets aberrant proteins and mRNAs for degradation (Keiler, 2008). This complex is ubiquitous among bacteria and is required for viability, response to stresses, and pathogenesis in many bacteria (Keiler, 2008). The other gene implicated in pigmentation, pigE, is part of the prodigiosin pathway in Serratia sp. (Williamson et al., 2005b).

FIG. 3.

Structures of compounds isolated in this study. Prodigiosin (A) and heptylprodigiosin (B) were purified from a culture of Janthinobacterium lividum BR01, and violacein (C) was purified from a culture of J. lividum BP01.

To obtain the entire complement of genes required for pigment production, the J. lividum BR01 fosmid library was screened in E. coli for red-pigmented clones. One pink clone, pBRred, was identified and its insert (32.6 kb) sequenced, revealing a cluster of 11 genes that transposon mutagenesis indicated are required for prodigiosin biosynthesis (20.8 kb) (Fig. 4). The mean G+C content of the pig genes (61.7%) is consistent with the mean G+C content of the remaining ORFs on pBRred (62.8%) and the ORFs on the pBRbeta clone (63.4%). All of the deduced Pig proteins from J. lividum BR01 exhibit higher sequence identity to the proteins from the marine γ-Proteobacterium Hahella chejuensis (40–76%) (Kim et al., 2006) than to the proteins from either Serratia sp. ATCC 39006 (32%–60%) or S. marsescens ATCC 274 (34%–60%) (Supplemental Table 1, available online at www.liebertonline.com).

FIG. 4.

Maps of ORFs from pBRred (32,611 bp) (A) and pBPpur (32,163 bp) (B). The ORFs with diagonal lines correspond to homologues of the pig pathway, which encodes prodigiosin biosynthetic genes, and the ORFs with grid lines correspond to homologues of the vio pathway, which encodes violacein biosynthetic genes. ORFs filled with black, gray, and black dots have homology to hypothetical ORFs, regulatory proteins, and transport proteins, respectively. Those ORFs lacking fill have homology to a D-aspartate O-methyltransferase (1), rhodanase (2), glyoxalase (3), osmotically inducible protein OsmC (4), L-sorbosone dehydrogenase (5), NAD(+) kinase (6), RecN (7), class C β-lactamase (8), α-amylase (9), transposase (10, 11), DNA polymerase III ɛ subunit (12), ribonuclease H (13), and methyltransferase (14).

Growth of the clone at 28°C or 37°C resulted in reduced pigment production, and optimal pigment production was observed when clones were grown at room temperature. It is possible that the reduced pigmentation of pBRred (pink) compared with J. lividum BR01 (red) was due to limitations of heterologous expression of the pathway.

Isolation and characterization of J. lividum BP01 pigment-producing gene cluster

The proton nuclear magnetic resonance (¹H-NMR) spectrum and LC-MS/MS analysis of the purple compound produced by J. lividum BP01 matched that obtained with purified violacein (Fig. 3C). Random mutagenesis indicated that interruption of the smpB gene resulted in the loss of pigmentation, but no biosynthetic genes were identified by mutagenesis. When the fosmid library of J. lividum BP01 genome was then screened in E. coli, one clone (pBPpur) was identified that was dark purple when grown at room temperature but not when grown at 28 or 37°C. The complete sequence of the pBPpur fosmid clone (32.2 kb; Fig. 4) revealed the prototypical violacein biosynthetic operon (7.3 kb), which consists of the five genes vioABCDE (Sanchez et al., 2006) and was originally identified in Chromobacterium violaceum. In vitro transposon mutagenesis of pBPpur resulted in nonpigmented mutants, and these insertions mapped to each of the genes (vioABCDE) in the violacein pathway. PCR amplification of the violacein cluster in J. lividum BR01 and of the prodigiosin cluster in J. lividum BP01 was negative (data not shown). Additionally, annotation of the DNA flanking the violacein gene cluster revealed a putative β-lactamase (Fig. 4). pBPpur was therefore tested for resistance to eight β-lactam antibiotics in an MIC assay, but the clone did not confer resistance on E. coli (data not shown).

Discussion

This is the first report of a red-pigmented J. lividum strain, revealed by the construction and analysis of a collection of bacteria cultured from nonpermafrost Alaskan soil. This study extends the known diversity of J. lividum, which is a member of the Burkholderiales in the β-Proteobacteria and has been found primarily in aquatic environments (Saeger and Hale, 1993) and soils (Gleave et al., 1995; Rossolini et al., 2001; Du et al., 2007; Kawakami et al., 2007). J. lividum has been reported to control amphibian fungal pathogens (Harris et al., 2009), and can be toxic to other bacteria, viruses, and protozoa (Lichstein and Van De Sand, 1946; Andrighetti-Fröhner et al., 2003; Matz et al., 2004). Our results demonstrate the importance of culturing from novel environments and of challenging common assumptions; J. lividum BR01 and BP01 were nearly ignored because of the familiarity of the red and purple pigments. Our original predictions were that the purple isolate was C. violaceum and that the red isolate was S. marcescens. However, neither prediction was correct as 16S rRNA gene sequencing indicated that both isolates are J. lividum. The 16S rRNA gene sequences of BR01 and BP01 are virtually identical, both strains contain metallo-β-lactamases, and the genes surrounding the β-lactamase genes are highly similar, yet the strains diverged in the acquisition of pathways for the production of different small molecules (presumably after the evolution of the β-lactamase genes). This highlights the point that neither colony morphology nor 16S rRNA gene sequence alone indicates the taxa present in a community or their characteristics (Jaspers and Overmann, 2004).

Pigment production in J. lividum BR01 and BP01 requires growth at temperatures below 20°C. Serratia sp. produce prodigiosin at temperatures below 30°C but not at 37°C (Williams et al., 1971). J. lividum BP01 is the first example of cold-dependent pigment production among violacein producers. The temperature-dependent pigment and antibiosis by J. lividum BR01 and BP01 raise interesting questions about the ecological role of the pigment in these strains. The Alaskan soil rarely reaches temperatures above 16°C (Viereck et al., 1993), so this habitat may have selected for J. lividum strains that produce the pigments at cool temperatures. Further studies are required to investigate the role of cold-activated pigment production and antibiosis in J. lividum survival in situ. The cloned pathways maintain the cold regulation in E. coli that they had in the native organisms, and the clones may therefore provide the basis for studying the mechanism of cold-specific gene regulation.

J. lividum BR01 is the first reported β-Proteobacterium that produces prodigiosin. Prodigiosins are of clinical interest because they are attractive immunosuppressive and anticancer therapeutics (Williamson et al., 2006). Other prodigiosin-producing bacteria include Streptomyces coelicolor, Hahella chejuensis, Pseudomonas magnesiorubra, Vibrio psychroerythreus, and Serratia sp., which represent the Actinobacteria and γ-Proteobacteria (Williamson et al., 2006). Prodigiosin is produced by the condensation of 2-methyl-3-amyl-pyrrole (MAP) and 4-methoxy-2,2′-bipyrrole-5-carbaldehyde (MBC) (Williamson et al., 2005b). The gene clusters for prodigiosin production in Serratia sp. (Harris et al., 2004; Williamson et al., 2005b) and H. chejuensis (Kim et al., 2006) differ from the J. lividum BR01 cluster in several ways. First, the discrete pigB and pigC genes of Serratia sp., whose products sequentially produce MAP and execute the MAP-MBC condensation reaction (Williamson et al., 2005b), exist as a fused gene in pBRred. Further, the pigJ and pigK genes, which are also separate genes in Serratia sp., are a fused gene in pBRred. Both PigJ and PigK are involved in the biosynthesis of MBC, although the role of PigK is uncertain. The third difference is that pigN is absent from the J. lividum BR01 prodigiosin cluster. In Serratia sp., pigN and pigF are thought to work together to direct the final step in the synthesis of MBC (a methylation reaction), and pigN mutants produce norprodigiosin as well as prodigiosin (Harris et al., 2004; Williamson et al., 2005b). A BLASTN analysis for pigN in the pBRred cluster yielded no regions of identity, so it remains unclear how the final step in synthesis of MBC is accomplished in strain BR01. Further studies are required to determine the effects of the genetic differences on the biochemical functions of the prodigiosin pathway in J. lividum BR01. Comparison of expression of the prodigiosin genes in J. lividum BR01 and Serratia sp. may aid in understanding how the gene cluster contributes to the ecophysiology of these organisms. The discovery of a prodigiosin-producing J. lividum strain expands the diversity of pathways for the biosynthesis of prodigiosin.

The goal of this study was to explore the possibility that previously unexplored Alaskan soil harbors novel biodiversity. The red-pigmented J. lividum strain indicates potential for discovery in the Alaskan soil culture collection. In this study of a cold soil, we found known organisms and natural products in new configurations and combinations. When examined in new ways, the culturable soil community may continue to be a fruitful source of biochemical and genetic diversity.

Disclosure Statement

No competing financial interests exist.