Bacterial strains and growth conditions

Bacterial strains and plasmids used in this study are listed in Table S1, while primers (Eurofins, MWG Operon, Germany) used are listed in Table S2. E. coli EPI300::pCC1FOS (Epicentre Biotechnologies, Madison, WI, USA) was grown in Luria-Bertani (LB) medium containing 12.5μg/ml chloramphenicol (Cm) and in 12.5μg/ml chloramphenicol plus 50μg/ml kanamycin (Kan) following EZ-Tn5 transposon mutagenesis reactions E. coli MKH13 [22] and Lactococcus lactis MG1363 [23] were grown in LB and M17 (+0.5% glucose; GM17 media) media respectively. Media was supplemented with 20µg/ml Cm for strains transformed with the plasmid pCI372 [24]. Media was supplemented with 1.5% (w/v) agar when required. Overnight cultures of E. coli were grown at 37°C with shaking, while L. lactis cultures were grown statically at 30°C.

Construction and screening of metagenomic library

A previously constructed fosmid clone library [25,26], created from metagenomic DNA isolated from a faecal sample from a healthy 26 year old Caucasian male was used to screen for salt-tolerant clones. The library was screened as outlined previously [21]. Briefly, a total of 23,040 clones from the library were screened on LB agar supplemented with 6.5% (w/v) NaCl and 12.5μg/ml chloramphenicol using a Genetix QPix 2 XT™ colony picking/gridding robotics platform. Plates were incubated at 37°C for 2-3 days and checked periodically for growth of likely salt-tolerant clones.

Transposon mutagenesis

Transposon mutagenesis was carried out in accordance with the manufacturer’s instructions, using the EZ-Tn5 <oriV/ KAN-2> in vitro transposition kit (Epicentre Biotechnologies). E. coli EPI300 cells were transformed with the transposon reaction mixture and selected on plates containing Cm and Kan (12.5 and 50µg/ml, respectively). The transposon insertion clones were subsequently replica plated onto LB with and without 6.5% added NaCl. Clones which grew on LB but not on LB + 6.5% NaCl suggested a likely insertion event in a gene involved in salt tolerance. Presumptive salt-tolerant knock-outs were grown overnight and a fosmid DNA extraction was performed. The extracted fosmid containing metagenomic DNA was subjected to sequencing from the ends of the transposon using the primers EZ-Tn FP-1 and EZ-Tn RP-1 (Table S2.). All sequencing was performed by GATC Biotech (Konstanz, Germany).

DNA manipulations

Induction of fosmids from low to high copy number for downstream applications such as transposon mutagenesis and sequencing was performed as per manufacturer’s instructions and as described previously [21]. The Qiagen QIAprep® Spin mini-prep kit was used to extract fosmids as per manufacturer’s instructions. PCR products were purified with a Qiagen PCR purification kit and digested with restriction enzymes XbaI and PstI (Roche Applied Science), followed by ligation using the Fast-Link DNA ligase kit (Epicentre Biotechnologies) to similarly digested plasmid pCI372. Electro-competent E. coli MKH13 and L. lactis MG1363 were transformed with the ligation mixture and plated on LB and GM17 agar respectively, containing 20µg/ml Cm for selection. Colony PCR was performed on resistant transformants using a primer on the stlA gene (stlA FP) and a primer on the plasmid (pCI372 RP) to confirm the presence and size of the insert.

Detection of the stlA gene in metagenomic DNA isolated from human stool samples was attempted using PCR. Twenty samples from the ELDERMET study [27]; which consisted of five community (healthy), five long stay (frail) old subjects and five healthy young and five young subjects with irritable bowel syndrome (IBS) were used as template DNA. Furthermore, five samples from healthy adults from a separate study [28] were also tested using the following primer pairs: stlA FP and stlA RP, stlA-J FP and stlA-J RP, stlA-OUT FP and stlA-OUT RP, stlA-IN FP and stlA-IN RP (see Table S2).

Growth experiments

Cultures were grown overnight in appropriate media. Cells were subsequently harvested, washed in one quarter strength sterile Ringer’s solution and re-suspended in fresh broth. A 2% inoculum was sub-cultured in fresh broth containing the appropriate stress (i.e. sodium chloride (NaCl), potassium chloride (KCl), sucrose, glycerol, low pH or bile as required) and 200µl was transferred to individual wells of a sterile 96-well micro-titre plate (Sarstedt Inc. Newton, USA). Plates were incubated at 37°C (or 30° for L. lactis strains) for 24-48 hours in an automated spectrophotometer (Tecan Genios) which recorded the optical density at 595 nanometres (OD_595nm) every hour. For experiments using bile, uninoculated media containing bile were dispensed as blanks in the 96-well plate and their OD_595nm values were subtracted from the corresponding inoculated wells to give the OD_595nm for the microbial fraction. The data was subsequently retrieved and analysed using the Magellan 3 software program. Representative graphs were created using the Sigma Plot 10.0 software programme (Systat Software Inc, London, UK). Results are presented as the average of triplicate experiments, with error bars being representative of the standard error of the mean (SEM).

BIOLOG Phenotype Microarray (PM) Assay

The phenotype microarray (PM) osmolytes microplate (PM9) was used to compare the cellular phenotypes [29] of E. coli MKH13::pCI372 and MKH13::pCI372-stlA under 96 different conditions. The BIOLOG PM protocol for E. coli and other Gram-negative bacteria was followed for preparation of the different inoculating fluid (IF) solutions (IF-0 and IF-10; supplied by BIOLOG) and inoculation of the PM plates. Briefly, isolated colonies were added to IF-0 fluid until a cell suspension of 42% T (transmittance) was achieved. This was subsequently diluted in IF-0 + dye mix A to achieve 85% T. Finally, this was diluted in IF-10 + dye mix A and 100ul was inoculated to each well of the PM 9 microplates. Plates were incubated at 37°C and readings were taken over a 24 hour period using an automated plate reader (BIOTEK Synergy 2) which measured the absorbance at 590nm.

Sequencing and bioinformatic analysis

The fosmid insert from clone SMG 25 was fully sequenced and assembled by GATC Biotech (Konstanz, Germany) using the GS FLX (Roche) pyrosequencing platform on a titanium mini-run. Putative open reading frames were predicted using Softberry FGENESB bacterial operon and gene prediction software (available at www.softberry.com). Retrieved nucleotide and translated amino acid sequences were functionally annotated by homology searches using the Basic Local Alignment and Search Tool (BLAST) using a maximum e-value cut-off of 1e^-03, to identify homologous sequences from the National Centre for Biotechnology Information (NCBI) website: http://www.ncbi.nlm.nih.gov/blast/Blast.cgi. A list of proteins encoded on SMG 25 fosmid insert is presented in Table 1.

The following databases and tools were used to gain additional information on the StlA protein: Expasy ProtParam server, Conserved Domain Database (CDD), PROSITE motif search, SignalP 4.0, HMMER, TMHMM, HHPred, Softberry BProm promoter search (www.softberry.com), SOPMA, SWISS MODEL, iTasser and QUARK. Relevant information and results can be found in Table 2 [30-41].

The Fold and Functional Assignment System (FFAS03) is a profile-profile and fold recognition algorithm that can detect remote homology between proteins [42]. FFAS03 searches numerous databases including non-redundant NCBI protein sequence database (NCBI nr), Global Ocean Sampling (GOS) from JCVI (J. Craig Venter Institute), PDB (Protein Data Bank), SCOP (Structural Classification of Proteins), and COG (Clusters of Orthologous Groups), as well as numerous metagenome datasets (microbial metagenome samples from the Joint Genome Institute, human gut metagenome samples from the Hattori lab, human oral microbiome database from the Forsyth institute and GOS data from JCVI and CAMERA). Furthermore, FFAS03 searches against the MetaHit (Metagenomics of the Human Intestinal Tract) dataset [1], which contains over 3 million unique gene sequences from the human gut microbiome. The StlA protein sequence was submitted to the server to identify proteins with distant homology based on FFAS profiling or homologues by BLAST and PSI-BLAST (Position-Specific Iterated BLAST) against the databases and metagenome datasets. The FFAS03 server can be found at: http://ffas.burnham.org/ffas-cgi/cgi/document.pl.

The Integrated Microbial Genomes and Metagenomes (IMG/M) [43] is a data management system for the comparative analysis of metagenome sequence data. IMG/M-HMP [44] specifically contains metagenome data from the Human Microbiome Project (HMP) [45]. It contains 748 metagenome datasets generated from sequencing samples from 17 different body sites (number of samples from each site are in brackets); anterior nares (94), keratinised gingiva (6), buccal mucosa (122), hard palate (1), retroauricular crease (20), left (2) and right retroauricular crease (5), palatine tonsils (6), saliva (5), throat (7), tongue dorsum (133), sub- (8) and supra-gingival plaques (127), mid-vagina (2), posterior fornix (60), vaginal introitus (3) and stool (147). It also provides tools for comparative analysis between hosted sequences and user supplied sequences. The StlA protein sequence was searched (maximum e-value cut-off of 1e^-50) against all the available metagenomes (748) from the HMP as well against all bacterial (9049), archaeal (323), viral/phage (2809) and eukaryotic genome (183) sequences (both assembled and draft), as well as all sequenced plasmids (1193) and all other non-human metagenome datasets (representing >1,300 non-human samples) from diverse environments, including terrestrial, aquatic and host-associated plants and animals (File S1) stored in the database at the time of writing (search date 19/09/13). The IMG/M-HMP server can be found at: http://img.jgi.doe.gov/cgi-bin/imgm_hmp/main.cgi. PhiSpy [46] was used to identify putative prophage genes and the boundaries of the putative prophage region on SMG 25.

Taxonomic assignment of scaffolds

The scaffolds on which an stlA homologue was identified were subjected to BLASTX analysis (maximum e-value cut-off of 1e^-50). The BLASTX results were downloaded and imported to MEGAN 4 (Metagenome Analyser 4) software program [47] for taxonomic assignment.

Screening the metagenomic library

Screening approximately 23,000 clones from a human gut metagenomic library led to the identification of 53 clones which were designated as conferring salt-tolerance (i.e. facilitating growth on LB supplemented with 6.5% NaCl, a concentration which inhibits the growth of the cloning host carrying an empty fosmid vector). Six clones (annotated Salt MetaGenome; SMG 1-6) grew within 24 hours and the remaining 47 grew in the following 24-48 hours. SMG 25 represents one of the “late-bloomers” and was chosen at random for analysis. End sequencing of the fosmid insert revealed it shared highest genetic identity to species from the genus Akkermansia, namely Akkermansia muciniphila ATCC BAA-835, Akkermansia muciniphila CAG:154 and Akkermansia sp. CAG:344. A muciniphila ATCC BAA-835, the type strain [48,49], is a mucin degrading member of the Phylum Verrucomicrobia which is commonly found in the human gut microbiome [50]. Growth was monitored spectrophotometrically, by measuring the optical density at 595_nm (OD_595nm). SMG 25 was shown to have a significant (unpaired student t-test, P <0.0001) growth advantage in the presence of NaCl compared to the EPI300 host strain carrying an empty fosmid vector (pCC1FOS) (Figure 1 A).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Growth experiments in NaCl and KCl.

(A) Growth of E. coli EPI300::pCC1FOS and clone SMG 25 in LB broth supplemented with 6.5% sodium chloride (NaCl) (P <0.0001). (B) Growth in LB broth and LB broth supplemented with 3% NaCl (P =0.0470), (C) 4% NaCl (P <0.0001) and (D) 4% KCl (P <0.0001). P-values were determined using the student t-test (unpaired). Results are presented as the average of triplicate experiments, with error bars being representative of the standard error of the mean (SEM).

https://doi.org/10.1371/journal.pone.0082985.g001

Fosmid sequencing and analysis

The fosmid insert from SMG 25 was fully sequenced and assembled by GATC Biotech (Konstanz, Germany) and was predicted to contain 45 putative open reading frames that encode proteins (see Table 1). Translated nucleotide sequences were subjected to BLASTP analysis to identify homologous sequences in the database. Twenty-six of the genes encoded proteins corresponding to different species of Akkermansia (ranging from 34-98% amino acid identity), but a sizeable proportion of the encoded proteins, approximately 27% (12/45) (highlighted in bold in Table 1) had no significant similarity to sequences in the database, indicating the presence of novel sequences. Overall, only 13 proteins could be assigned a putative function based on BLASTP searches, whilst the remaining genes encoded hypothetical or uncharacterised proteins. The full fosmid insert sequence of SMG 25 can be found in GenBank (accession number=JQ269600.1; gi=375342965). The G+C (guanine and cytosine) skew of the entire fosmid insert as well a picture of the G+C content of each individual gene are presented in Figure 2 B and C respectively.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Bioinformatic analysis of SMG 25 fosmid insert.

(A) FFAS03 analysis of the StlA protein and the encoded proteins of flanking genes was performed to identify putative distant structural homologues. A score of -9.50 or lower is considered significant. (B) Representation of the G+C skew of the entire fosmid insert of SMG 25. (C) Representation of the gene arrangement on SMG 25. Gene lengths are approximately to scale and colour coding represents G+C content of each individual gene which can be determined from the G+C content gradient bar. The presence of a phage-associated gene and clear separation in G+C content over the length of the fosmid insert indicates much of this region may have been acquired via lateral gene transfer (LGT). Phage-associated gene is marked “P”, while the stlA gene is indicated with an asterisk (*) symbol. Genes are numbered as indicated in Table 1 and as mentioned in the text. Numbering of some shorter genes has been excluded for clarity. Selected nucleotide positions (in base pairs) are displayed in bold italic font above genes. (D) A detailed view is presented of the nucleotide and amino acid sequence of the stlA gene and StlA protein respectively. The putative start codon is in green, while a 250 base-pair region upstream of this is shown to include putative -35 and -10 promoter regions (underlined) and a predicted rpoD transcription factor binding site (in bold). Amino acids surrounded by grey box indicate the predicted signal sequence of StlA and those highlighted in blue represent four transmembrane regions. The location of the EZTn5 transposon insertion is indicated with a red triangle.

https://doi.org/10.1371/journal.pone.0082985.g002

Transposon (EZ-Tn5) mutagenesis and cloning of the stlA gene

Transposon mutagenesis was performed on clone SMG 25 and a transposon insertion in gene 6 was identified which eliminated the growth advantage under osmotic stress; this locus (designated stlA) is predicted to encode a protein of 257 amino acids which, at the time of writing, has no homologues in the database. The transposon insertion was found to be between amino acid position 136 (alanine) and 137 (glutamine) of the protein. The stlA gene was cloned, along with some flanking DNA that was predicted to contain the native promoter region (predicted with BProm program; see Table 2 for details), into the shuttle plasmid pCI372 and transformed into E. coli MKH13 and L. lactis MG1363.

Growth experiments and BIOLOG phenotypic microarray

E. coli MKH13 cells transformed with a plasmid bearing a copy of the stlA gene were grown in LB broth containing various concentrations of NaCl (from 0-5% w/v added NaCl). It was observed that a statistically significant (unpaired student t-test) growth advantage was conferred upon the stlA⁺ cells compared to wild-type MKH13 carrying an empty plasmid in LB broth supplemented with both 3% (P =0.0019) and 4% NaCl (P <0.0001) (Figure 1 B and C, respectively), while growth was similar in LB alone (data not shown).

Due to the uncharacterised and non-homologous nature of the stlA gene and its encoded protein, growth of wild-type MKH13 and stlA⁺ strains was compared under 96 different conditions using BIOLOGs phenotypic microarray (PM) technology [29] to identify possible further phenotypic changes. Strains were tested on BIOLOG plate PM9, which contains different osmotic stress conditions and osmolytes for analysis. In addition to NaCl, the results indicated stlA⁺ had an increased growth phenotype in the presence of potassium chloride (KCl). Confirmatory growth curves were performed in LB broth supplemented with a concentration of 4% KCl. A statistically significant (unpaired student t-test) growth advantage was observed for stlA⁺ in LB supplemented with 4% KCl (P <0.0001) compared to wild-type MKH13 (Figure 1 D).

Growth of both strains was also assessed under conditions of non-ionic osmotic stress (in the form of glycerol and sucrose), low pH and in both porcine and human bile, as all three stress conditions are commonly encountered in the GI tract. Growth of both strains was inhibited at pH 2.5 and pH 3.5, while no significant difference in growth was observed at pH 4.5 or pH 5.5, in the presence of sucrose or glycerol, or in the presence of either porcine or human bile (Figure S1).

The stlA genes’ ability to increase salt tolerance was also tested in a Gram positive host; L. lactis MG1363. There was no observable increase in salt tolerance in L. lactis MG1363 carrying a plasmid encoded copy of stlA compared to L. lactis carrying an empty copy of the plasmid, while a similar growth rate and final OD value was observed for both strains in GM17 broth alone (Figure S2 A and B)

Bioinformatic analysis of StlA

The databases and tools used to identify features of StlA are presented in Table 2 below, along with the results of the analyses. An illustration of the stlA gene and its associated features is presented in Figure 2 D.

IMG/M-HMP analysis

The StlA protein sequence was screened against all available metagenomes from the human microbiome project (HMP) using BLASTP on the IMG/M-HMP website (http://img.jgi.doe.gov/cgi-bin/imgm_hmp/main.cgi) [45], which were sampled from 17 body sites giving a total of 748 samples. In addition, all available finished, permanent draft and draft genome sequences for Bacteria, Archaea, Eukarya and viruses/phages, as well as all available sequenced plasmids were searched using BLASTP for homologous sequences to StlA. There was no significant similarity for the StlA protein to any of the bacterial, archaeal, eukaryotic, viral or plasmid genomes/sequences, nor to any non-human associated metagenomes (over 1,300 samples from more than 200 metagenomes, see File S1). The only similarity to StlA among the sampled microbiomes was to the stool microbiome samples, where 10 similar proteins from 8 different subjects (out of 100) (Table 3) were identified on different scaffolds. The date of the last search was on 19/09/13. The taxonomic assignment of the scaffolds can be seen in Figure S3.

The gene neighbourhoods around the genes homologous to stlA on each scaffold were investigated in an attempt to gain information on possible functions and conserved gene arrangements (Figure S4). The genes most commonly flanking the stlA homologues were on the same strand of DNA and encoded an ankyrin repeat protein (COG0666), a DnaJ class molecular chaperone with C-terminal zinc-finger domain (COG0484) and a predicted membrane protein (COG2314; Pfam05154 –TM2 domain). There are also a number of hypothetical proteins, for which no additional information is currently known. On two of the larger scaffolds, genes for a restriction modification system are present, as well as an integrase/site specific recombinase protein (COG4974; Pfam00589), indicating some of this region may have been acquired by lateral gene transfer (LGT) and may represent prophage DNA. A phage-associated protein is predicted to be encoded by gene 20 (designated “P” in Figure 2 C) indicating the presence of a prophage on SMG 25 also. The fosmid insert of SMG 25 was analysed with PhiSpy [46] to identify possible prophage genes and the boundaries of the prophage region. PhiSpy predicted the prophage region to run from the start of gene 3 (nucleotide position 2024) to the end of gene 42 (nucleotide position 39972).

FFAS03 analysis

The FFAS03 server [42] was used to detect distant homology and fold recognition to StlA. FFAS analysis was also carried out on the translated protein sequences of the neighbouring genes to stlA on SMG 25, which also lacked any homologues in the databases (i.e. gene 3, 4, 5 and 7; gene 6 is stlA). The results of FFAS03 analysis are summarised in Figure 2 A). .

In addition to a profile-profile and a fold and functional assignment, the FFAS03 server also carries out a BLAST and PSI-BLAST search of the user sequence against numerous databases and metagenome datasets. The StlA protein was found to share significant similarity to a protein from two individuals from the MetaHit dataset [1]. These sequences corresponded to samples MH0011 (a healthy Danish female) and V1.CD-14 (a Spanish female with Crohn’s disease) which shared 60% identity (over 210 amino acids) and 82% identity (over 224 amino acids) respectively to StlA.

Detection of stlA in metagenomic DNA from human stool samples using PCR

The primer pair (stlA FP and stlA RP) initially used to amplify the stlA gene for cloning was unable to amplify PCR products in any of the metagenomic DNA samples (isolated from human stool microbiota), so a set of primers (stlA-J FP and stlA-J RP) were designed to amplify an internal fragment of the gene. This set of primers amplified numerous products of the correct size but these were found to be false positives following sequencing. An alignment was generated for StlA and homologous sequences from the stool microbiome from the HMP and MetaHit datasets to identify the most highly conserved regions (Figure S5) and different primer pairs were designed (stlA-OUT FP and RP; stlA-IN FP and RP). Two of the 25 metagenomic DNA samples tested (isolated from human faecal samples from ELDERMET [27] and another study [28]) generated PCR products of the correct size, which were confirmed to be stlA homologues following sequencing by using the stlA-IN FP and RP primer pair. One positive PCR product shared 72% nucleotide identity over approximately 300 base pairs (BLASTN versus stlA gene) and 64% identity (over 100 amino acids) using BLASTX, while one ELDERMET [27] sample was positive (community care/ healthy old) and confirmed by sequencing (87% identity over 339 nucleotides and 85% identity over 112 amino acids).