Gleaning evolutionary insights from the genome sequence of a probiotic yeast Saccharomyces boulardii

To amplify the ITS regions, primers ITS1 (TCCGTAGGTGAACCTGCGG) and ITS4 (TCCTCCGCTTATTGATATG) were used. Amplification was done using a mixture containing 1× Standard Taq Reaction Buffer, dNTPs (200 μM), BSA (0.3 μg/μL), template DNA (500 ng), 1.25 units/50 μl PCR Taq DNA Polymerase (Taq DNA Polymerase with Standard Taq Buffer; New England BioLabs) and forward and reverse primers (0.2 μM each). The cycling parameters used for amplification were: initial denaturation for 5 minutes at 95°C, followed by 30 cycles of 30 seconds at 95°C, 30 seconds at 50°C and 90 seconds at 72°C, with a final extension for 10 minutes at 72°C and cooling to 4°C. The amplified products were separated on 1.2% agarose gels, visualized and photographed on an Alpha Image Analyzer (Alpha Innotech Corporation, CA). The DNA from the amplified bands was eluted with QIAquick Gel Extraction Kit (Qiagen N.V.). The eluted DNA was further amplified for sequencing using Terminator Ready Reaction Mix (1 μl), Sequencing Buffer (1 μl; 5×; 200 mM Tris-Cl, 5 mM MgCl₂, pH 9.0), PCR amplified DNA (35 ng), primer (3.2 pmol) and Milli-Q water to make up the volume to 10 μl. PCR cycling conditions were: initial denaturation for 1 minute at 96°C, followed by 24 cycles of 10 seconds at 96°C, 5 seconds at 50°C and 4 minutes at 60°C and cooling to 4°C. The final PCR product was sequenced on a Sanger sequencer and the resulting ITS sequence was compared against all available ITS sequences on the NCBI database.

Microsatellite fingerprinting based on the Post Meiotic Segregation (PMS) marker [30], PMS1, PMS2 and PMS3 was performed. Amplification was achieved using a mixture containing 1× Standard Taq Reaction Buffer, dNTPs (200 μM), BSA (0.3 μg/μL), template DNA (500 ng), 1.25 units/50 μl PCR Taq DNA Polymerase (Taq DNA Polymerase with Standard Taq Buffer; New England BioLabs) and 0.2 μM of forward and reverse primers (PMS1: GTGGTGGTGGTGGTG; PMS2: GACGACGACGACGAC and PMS3: GACAGACAGACAGACA). The following cycling parameters were used for microsatellite fingerprinting: initial denaturation for 5 minutes at 94°C, followed by 40 cycles of 1 minute at 94°C, 2 minutes at 45°C and 3 minutes at 72°C, with a final extension for 10 minutes at 72°C and cooling to 4°C. The amplified products were separated on 1.2% agarose gels, visualized and photographed on an Alpha Image Analyzer (Alpha Innotech Corporation, CA).

Sb EDRL (Econorm - Dr. Reddy’s Laboratories) was sequenced using the 454/Roche GS FLX Titanium system. Library preparation was carried out according to the GS FLX Titanium Rapid Library Preparation Kit (Roche Applied Sciences) at Centre for Cellular and Molecular Platforms (C-CAMP), Bangalore, India. The genomic DNA was first sheared into 400–1000 base-pair-long fragments and its quality was assessed using a BioAnalyzer DNA 7500 LabChip. Blunt ends were generated and adaptor ligation was followed by removal of small fragments. After library immobilization, the library was quantified using the RiboGreen method and the final yield was calculated. Half PicoTiter Plate of 454 shotgun sequencing was performed and resulted in a total of 733,390 reads for Sb EDRL with ~50× coverage.

The 454 shotgun reads were assembled de novo using Newbler v2 [31]. Multiple assemblies were obtained by varying the parameters for the minimum overlap length (ml) and minimum overlap identity (mi). The assembly that contained the lowest number of contigs and best N50 score (parameters: minlen 45; mi 96; ml 100) was chosen for further analysis. The quality of the assembly was further checked by mapping back the reads on the draft genome and visually checking for errors. Feature annotation was carried out by the MAKER pipeline [32] and tRNA was predicted by tRNAscan-SE 1.23 [33]. Features thus annotated were subjected to BLASTp [34] for functional characterization of the proteins with an E-value cutoff of 1e^-5. In addition to de novo assembly, the 454 shotgun reads were mapped onto 34 draft and complete genomes of S. cerevisiae available at NCBI (Additional file 1) using the mapping algorithm of CLCbio Genomics wb6 (http://​www.​clcbio.​com).

The draft genome annotated by the MAKER [32] pipeline and the reference-mapped genomes were searched for the presence of proteins in the molecular weight range of 54 kDa and 120 kDa using an in-house Practical Extraction and Report Language (PERL) script. The 63 kDa protein was retrieved by BLASTp using as query a 38-amino acid sequence stretch of this protein previously reported in literature [17], and the molecular weight was further confirmed using our PERL script. In order to narrow down on the 54 kDa protease, proteins in the molecular weight range 50–60 kDa, retrieved as mentioned above, were subject to BLASTp against the MEROPS database (http://​merops.​sanger.​ac.​uk/​) and an independent hmmscan [35] run against the protein family (PFAM) database [36] with an E-value cutoff of 1e^-5. These proteins were further used as queries in the Fold and Function Assignment System (FFAS) [37] program to retrieve annotated homologs and subjected to BLASTp against the Gene Ontology (GO) database [38] with an E-value cutoff of 1e^-5. Likewise, the sequences of 119–121 kDa proteins obtained using our PERL script was subjected to BLASTp against the GO database with an E-value cutoff of 1e^-5 and Search Tool for the Retrieval of Interacting Genes/Proteins (STRING) [39] analysis was performed to find interacting partners. Galactose metabolizing enzymes of the Leloir pathway [40] were found in the annotated genome by initiating stand-alone BLASTp searches using sequences of these enzymes from other S. cerevisiae.

The genomic DNA was purified from a commercially available lyophilized Sb (Econorm sachet; Dr. Reddy’s Laboratories) and was further confirmed by ITS sequencing. The ITS sequence was >99% identical to that of the S. boulardii culture collection (KC2540481.1) strain. PMS1, PMS2 and PMS3 markers were used to perform microsatellite fingerprinting in order to verify the similarity between Sb EDRL to other commercially available Sb strains (Sb Uni-Sankyo Ltd. [Now known as Sanzyme Ltd.] and Sb Kirkman) (Figure 1).

Next-generation sequencing of Sb EDRL on the Roche 454 GS-FLX Titanium platform resulted in a total of 733,390 shotgun reads of length 40–1773 bp. High-quality reads with ~50× coverage were assembled using Newbler v2.8 to obtain a draft genome of 11.4 Mbp in 194 contigs (N50: 251,807 bp). The GC content was 38% and 285 tRNA were present in Sb EDRL. Feature annotation was performed using the MAKER pipeline with Augustus [41] as the gene predictor. Of the 5803 coding sequences (CDS) regions predicted, 4635 (79%) could find hits with S. cerevisiae proteins when subject to a BLAST analysis against the non-redundant (nr) database. Reference-based mapping of Sb data onto other S. cerevisiae genomes revealed that the maximum number of reads map to S. cerevisiae strains Lalvin QA23 followed by EC1118, RM11-1a and S288c suggesting high similarity to these genomes (Additional file 2).

Sb is being used in treatment of C. difficile-induced diarrhea and colitis [14]. Sb can inhibit the toxins A and B of C. difficile by producing a 54 kDa serine protease that cleaves these toxins [14, 15]. Approximately 600 proteins of the Sb genome in the molecular weight range of 50–60 kDa were subjected to BLASTp against the MEROPS database and of these, 221 hits were obtained. These proteins were further subjected to BLASTp runs against the GO database [38]. Four proteins were found to be putative serine proteases based on their GO annotation and presence of conserved serine protease signature motifs. These four proteins belong to the carboxypeptidase and subtilisin-like sub-classes of serine protease (Additional file 2 and Additional file 3). Independently, all the annotated proteins from the MAKER pipeline were subjected to hmmscan against the PFAM database. Twenty two serine proteases spanning 10 sub-classes of the serine protease family were obtained. Of these 22 proteins, 4 were in the molecular weight range of 50–60 kDa and were the same proteins identified by our previous search against the MEROPS database as putative serine proteases. It is therefore tempting to speculate that one or more of these four proteins could possibly play a role in cleaving the C. difficile toxins A and B.

A 120 kDa protein has been suggested to neutralize the effects of cholera toxin by reducing cAMP levels in mouse [17]. Fifteen proteins in the molecular weight range of 119–121 kDa were retrieved (Additional file 3). These proteins were subjected to BLASTp against the GO database [38] with an E-value cutoff of 1e^-5 and FFAS search was performed. These 15 proteins mostly belong to the family of kinases and transporters (Additional file 2). The interacting partners of these proteins were fetched out by STRING analysis (Additional file 2). It is possible that any of these proteins may be involved in the cAMP pathway to neutralize the effects of the cholera toxin.

A 63-kDa phosphatase of Sb (PHO8P) containing a peptide of 38-amino acids (HFIGSSRTRSSDSLVTDSAAGATAFACALKSYNGAIGV), has been suggested to inhibit the toxicity of E. coli surface endotoxins by dephosphorylation [18]. The sequence has an activation site (VTDSAAGAT) that is present in S. cerevisiae and S. pombe. The presence of this protein was examined in most of the S. cerevisiae genomes available at the Saccharomyces genome database (SGD) [42] and in our Sb genome. S. cerevisiae strains AWRI1631, AWRI796, BY4741, BY4742, CBS7960, CEN.PK113-7D, CLIB215, CLIB324, CLIB382, EC1118, EC9-8, FL100, FostersB, FostersO, JAY291, Kyokai no.7, Lalvin QA23, M22, PW5, RM11-1a, Sigma1278b, S288c, T73, T7, UC5, Vin13, VL3, W303, Y10, YJM269, YJM789, YPS163 and ZTW1 were considered, out of which the phosphatase PHO8P was present in 28 strains and could not be found in strains CLIB324, T73, Y10, CLIB382 and M22. The pho8 gene is present in the Sb genome reported here. Fourteen genes (prp3, jip4, YDR476C, snf1, snm1, pex29, dig2, cwc21, kre2, vps52, vps72, vps60, rib3, pac11) that are in the genomic neighborhood of the phosphatase gene were analyzed. All these genes are completely conserved in Sb, S288c, AWRI796, BY4741, BY4742, CEN.PK113-7D, EC1118, FostersB, FostersO, JAY291, Kyokai no. 7, Lalvin QA23, Sigma1278b, T7, Vin13, VL3, W303, YJM789 and ZTW1 strains of S. cerevisiae (Figure 2A). The S. cerevisiae RM11-1a (Figure 2K) has likewise all genes conserved but they are present on the opposite strand. Other S. cerevisiae strains have one or more of these 14 genes missing (Figure 2B-J). The protein was conserved among all the S. cerevisiae strains with >95% identity suggesting the possibility of phosphatase activity in all the S. cerevisiae strains.

One of the distinguishing features of Sb is that unlike S. cerevisiae, it is incapable of utilizing galactose as a source of carbon [23, 28, 29, 40, 43]. However there are reports of some Sb strains, which can utilize galactose [29]. The enzymes galactose-mutarotase, galactokinase, galactose-1-phosphate uridyltransferase, UDP-galactose-4-epimerase and phosphoglucomutase are constituents of the Leloir pathway which helps to catalyze the conversion of galactose to glucose-6-phosphate [29]. Glucose-6-phosphate can then further be utilized via the glycolysis pathway. Galactokinase, galactose-1-phosphate uridyltransferase and UDP-galactose-4-epimerase are in synteny in most of the S. cerevisiae strains such as RM11-1a, S288c, AWRI796, BY4741, BY4742, Fosters B, FostersO, JAY291, LalvinQA23, Sigma1278b, T7, UC5, Vin13, VL3, YJM789, AWRI1631, CEN.PK113-7D, EC1118, Kyokai no. 7, ZTW1 and a similar synteny was observed for our Sb (Figure 3). S. cerevisiae strains CLIB382 and Y10 were not included in this analysis as their genomes are not annotated completely on the SGD. Surprisingly, our Sb genome has all the genes responsible for galactose uptake and fermentation, although it has been suggested that Sb is not able to utilize galactose as a carbon source [28, 44]. We have experimentally confirmed that Sb EDRL whose genome is reported here can assimilate galactose but cannot ferment it (unpublished results). Given that all genes responsible for galactose assimilation and fermentation are present in our genome, an effect similar to that observed by van den Brink et. al., 2009 [45] for S. cerevisiae CEN.PK113-7D wherein the yeast is unable to suddenly switch from glucose to galactose in the absence of oxygen (fermentation), due to energetic requirements, may help explain the inability of our Sb to ferment galactose.

The genome of Sb offers an important starting point to glean insights into the probiotic effects of this yeast and help differentiate it from other strains of S. cerevisiae. It is possible that the various probiotic effects that have been attributed to Sb do not necessarily correspond to the same strain of this organism. Therefore it will be of interest to obtain genomes of different commercially available strains of Sb in order to get a complete genome-level understanding of this important probiotic yeast.

This Whole Genome Shotgun project has been deposited at DDBJ/EMBL/GenBank under the accession ATCS00000000 for Saccharomyces cerevisiae strain boulardii EDRL. The version described in this paper is the first version ATCS01000000.