Complete genome sequence and comparative genomics of the probiotic yeast Saccharomyces boulardii.

The probiotic yeast, Saccharomyces boulardii (Sb) is known to be effective against many gastrointestinal disorders and antibiotic-associated diarrhea. To understand molecular basis of probiotic-properties ascribed to Sb we determined the complete genomes of two strains of Sb i.e. Biocodex and unique28 and the draft genomes for three other Sb strains that are marketed as probiotics in India. We compared these genomes with 145 strains of S. cerevisiae (Sc) to understand genome-level similarities and differences between these yeasts. A distinctive feature of Sb from other Sc is absence of Ty elements Ty1, Ty3, Ty4 and associated LTR. However, we could identify complete Ty2 and Ty5 elements in Sb. The genes for hexose transporters HXT11 and HXT9, and asparagine-utilization are absent in all Sb strains. We find differences in repeat periods and copy numbers of repeats in flocculin genes that are likely related to the differential adhesion of Sb as compared to Sc. Core-proteome based taxonomy places Sb strains along with wine strains of Sc. We find the introgression of five genes from Z. bailii into the chromosome IV of Sb and wine strains of Sc. Intriguingly, genes involved in conferring known probiotic properties to Sb are conserved in most Sc strains.

Introduction

The probiotic yeast Saccharomyces boulardii (Sb) has unique physiological properties such as tolerance to variations in pH, temperature and local stresses like the presence of GI enzymes, bile salts, and organic acids[1]. Several clinical studies have been performed to present Sb as a unique organism that inhibits pathogens[2-5] and restores gut flora and improves digestion[6, 7]. The mechanism of elimination of the pathogenic bacteria is mainly attributed to the adhesion proteins of Sb that binds to bacteria and inhibits their adhesion to the mucous-intestinal membrane[8, 9]. Certain proteins in Sb have been reported previously for their key roles in providing protection against Escherichia coli, Clostridium difficile, Vibrio cholera, and Helicobacter pylori infections[10]. A 63 kDa protein phosphatase of Sb dephosphorylates the E. coli endotoxin[11], a 54 kDa serine protease provides protection against Clostridium difficile infections by cleaving toxins A and B[12, 13] and a 120 kDa heat and trypsin-labile non-proteolytic protein of Sb neutralizes the secretions induced by cholera toxin by possibly reducing cyclic Adenosine Monophosphate (cAMP) levels[14].

A French Scientist Henri Boulard initially isolated Sb from the fruits lychee and mangosteen in 1923 and the organism was characterized as ‘Saccharomyces boulardii’ a novel species of genus Saccharomyces possibly to differentiate its probiotic effects and application from other yeast species[15]. Characterization of Sb as a separate species was further supported by the lack of galactose utilization and sporulation as compared to S. cerevisiae (Sc)[16]. Molecular phylogenetic and typing techniques suggested that Sb forms a separate cluster but belong to species Sc [17]. Comparative genomic hybridization experiments also established that Sc and Sb are different strains of the same species but the loss of all intact Ty1/2 elements was reported only in Sb. The loss of Ty elements was hypothesized to be related to Sb’s non-sporulation and diploidy[18] as the transcription of these mobile elements is under diploid control[19, 20]. The numbers of Ty elements are maintained via transposition during sporulation and haploid mitotic growth, but the absence of these stages can lead to the loss of Ty elements[21, 22].

The unique formulations of Sb, isolated from different sources like lychee, mangosteen, pineapple, etc., explored for treating different disorders have been patented by different companies and laboratories[23-26]. The mode of action of the probiotic yeast is not completely known[10], however the beneficial effects of the yeast have been established through various clinical studies[27-33]. However, the probiotic yeast has also been found to be associated with fungemia in immune-compromised patients[34-37]. Our group reported the first genome of Sb-EDRL strain (Econorm from Dr. Reddy’s Laboratory) and tried to trace the genomic reasons for the probiotic behavior of this yeast[38]. There we explored that the proteins appeared to be specifically present in Sb were also present in the Sc strains[38]. However, a single draft genome may not be sufficient to study the probiotic properties associated with the organism. Whole genome sequence of more Sb strains including complete genomes will be required to understand the evolution and quantitative variations among Sb strains.

We have sequenced whole genomes from five Sb strains marketed by Laboratories Biocodex, Kirkman Labs, Unisankyo Ltd. (Now Sanzyme Ltd.) and Unique Biotech to find the reasons for probiotic properties of Sb. Two strains isolated from Sb sachets marketed by Laboratories Biocodex and Unique Biotech was assembled to completion for comparison and to address biologically relevant differences. We report the complete genomes of Sb-biocodex and Sb-unique28 in this paper along with draft genomes of Sb-kirkman and Sb-unisankyo and updated version of the Sb-EDRL genome. We have analyzed genomes of seven different strains of Sb to find if they are different and can account for different probiotic properties or are species specific. We have compared the Sb-biocodex with Sb-unique28 genome to find differences among these probiotics at the strain level. To address the differences between probiotic yeasts and other strains of Sc we compared all five Sb genomes with all available genomes of Sb and Sc strains. This genomic study for probiotic yeast relates to the previous clinical and molecular studies and reports variations at genome level among strains of Sb and Sc.

Results

Complete genomes of Sb strains and their characteristics

The complete genomes were obtained for Sb-biocodex and Sb-unique28 (Fig. 1A) sequenced using PacBio P6C4 chemistry at ~200x coverage. The final assembly of Sb-biocodex (12 Mbp genome and N50 792,172 bp) comprises of 16 complete chromosomes and 14 unplaced contigs. Similarly, Sb-unique28 was finalized with 14 complete chromosomes, two chromosomes (Chromosome 5 and 9) comprising of two contigs each and nine unplaced contigs (12.1 Mbp genome and N50 929,172 bp) (Table 1; Supplementary File I). Detailed genomic features of Sb-biocodex and Sb-unique28 are discussed in Supplementary methods (Supplementary Figure 1). The complete chromosomes of Sb-biocodex and Sb-unique28 were identified with either the telomeres or the telomeric and sub-telomeric regions. The Chromosome I of Sb-biocodex had a shorter length as compared to Sb-unique28 and Sc S288C where we found that the sub-telomeric region of approximately 0.02 Mbp on the right arm with genes annotated as dubious or uncharacterized ORFs (YAR053W to YAR071W) in Sc S288C genome (http://www.yeastgenome.org) is absent.

Figure 1

(A) Circular representation of the Sb-biocodex and Sb-unique28 complete genome. Circles (from inside to outside): circle 1 (GC content), circle 2 (Sc S288C complete chromosomes with alternative blue and red color); circle 3 (GC skew); circle 4 (Sb-biocodex complete chromosomes); circle 5 (Sb-unique28 complete chromosomes). BRIG 0.95 was used to build the circular representation. Mapping studies were done using BLASTn with an E-value cut-off 1e-5. (B) Core orthologous proteins among all Sb proteome represented by orthology diagram depicting dispensable proteome and homologs among these strains.

Table 1

Genome Assembly statistics of all Sb strains.

Sb strains	Number of Contigs	N50 (bp)	Genome Size (Mbp)
Sb-biocodex	31	792,172	12.0
Sb-unique28	30	909,172	12.1
Sb-EDRL	77	306,308	11.5
Sb-kirkman	115	416,209	11.7
Sb-unisankyo	164	262,146	11.6

Sb-EDRL, sequenced using 454 sequencing data, was assembled in 107 contigs (Genome Size: 11.5 Mb and N50: 271,789 bp) and was further scaffolded into 77 gapless contigs (Genome Size: 11.5 Mb and N50: 819,652 bp) using Illumina HiSeq PE and MP shotgun data (Table 1). The shotgun reads of Sb-kirkman and Sb-unisankyo were assembled in 115 contigs (Genome Size: 11.7 Mb; N50: 621,720 bp) and 164 contigs (Genome Size: 11.6 Mb; N50: 262,146 bp), respectively (Table 1).

Complete 2-micron plasmid was retrieved from Sb-biocodex and Sb-unique28 matching to Sc YJM993 plasmid (Length: 6318 bp; Genbank identifier: CP004528.1). Similarly, the complete circular plasmid was also retrieved from Sb-EDRL, Sb-kirkman, and Sb-unisankyo by mapping their reads to the Sb-biocodex plasmid sequence. The complete circular plasmid obtained from all the Sb strains was found to be 100% identical. G5297 -> A and A5582 -> G polymorphisms in the rep2 gene were observed in Sb plasmids as compared to Sc YJM993 plasmid. The mutation A5582- > G was non-synonymous but G5297- > A corresponds to A296 -> V amino acid change in the Rep2 protein (Supplementary Figure 2).

Approximately 5500 CDS and 300 tRNAs were predicted in all the Sb strains (Supplementary File I). The Sc S288C genome was also re-annotated using a similar method. All these ORFs were characterized functionally based on the gene names and description provided in Saccharomyces Genome Database (SGD)[39]. The core proteome comprised of 5140 proteins for all Sb strains (Fig. 1B) and approximately 200 proteins are found to be unique in each Sb strain (Supplementary File II). The unique proteins in all the strains of Sb were extracted and subjected to BLASTp against the proteome of all other Sb to find if any homologs to those proteins are present in other strains of Sb. All these unique proteins have homologs in other strains of Sb. All the Sb genomes assembled and sequenced in our study were compared with two already available Sb genomes (Sb-17 and Sb-ATCC-MYA-796) and 145 Sc genomes (Supplementary File III).

Mating Locus and Sporulation in Sb

Mating type of yeast was determined by the two different alleles of Mating-type (MAT) Locus MATa and MATα [40]. Sb has been suggested to be diploid in previous studies[1, 41, 42] and should comprise both MATa and MATα sequences on the Chromosome III at a heterozygous locus. The characterized locus in Sc was used as a query to search the Sb-biocodex genome and both MATa, and MATα sequences were retrieved from Sb-biocodex at 99% identity in chromosome III and unplaced scaffold, respectively. MATa sequence in Sb-biocodex is 2438 bp in length which is 99.7% identical to Sc MATa locus (GI: V01313) with eight substitutions and seven insertions (Fig. 2). The region was divided into W, X, YA and Z1 regions based on the alignment to MATa locus of Sc (GI: V01313). Similarly, the MATα gene was retrieved by subjecting the MATα region of Sc S288C from SGD to BLASTn against Sb genomes. The MATα gene was 2507 bp with only one substitution T267 -> G and could be divided into W, X, Y, Z1 and Z2 regions based on the alignment to Sc S288C MATα gene. The ORFs coded by MATa and MATα were identified through Augustus, and very short ORFs were predicted using DNA to protein translation tool (http://insilico.ehu.es/translate/). In yeast the MATα locus codes for bud5, MATα1, and MATα2 genes and MATa locus codes for bud5, MATa1, and MATa2 genes. Homothallic switching endonuclease (HO) gene is required for gene conversion at MAT Locus in haploid cells[43]. In heterothallic strains T189A, G223S, L405S, H475L substitutions and deletion of 36 amino acids (524–559) in HO protein results in loss of endonuclease activity[44, 45] whereas homothallic cells express HO [45]. The HO gene in all strains of Sb did not have any of the above-mentioned substitutions or deletions, and hence, the probiotic yeast Sb is homothallic diploid wherein both the MAT loci are present in the genome.

Figure 2

MAT Locus of Sb biocodex depicting the annotated regions. (A) MATa locus with yellow arrows depicting the MAT region divided in W, X, YA and Z1 regions with coded MATa1 and MATa2 regulatory proteins. (B) MATα locus with yellow arrows depicting the MAT region divided in W, X, Y, Z1 and Z2 regions with coded MATα1 and MATα2 regulatory proteins.

As it is known that both the heterothallic and homothallic diploid strains sporulate under conditions of nutrient deficiency[46], we performed sporulation phenotype assay (detailed in Supplementary Methods), to investigate the sporulation in Sb. We found that the Sb did not sporulate on sporulation media even after one week of incubation concordant with previous studies stating non-sporulation behavior of the probiotic yeast[41]. To investigate the role of sporulation proteins in the non-sporulating behavior of Sb the presence and the absence of all the sporulation proteins mentioned in SGD were identified using BLASTp in Sb proteome. Further, to rule out the possibility of the lack of any meiotic and mitotic genes hindering the sporulation pathway, we searched for 110 meiotic genes and 56 mitotic genes in Sb genome and found all were present (Supplementary File IV). An earlier report suggesting the divergence in CDC16, DMC1, and MND2 sequences as a possible reason for the defective sporulation was also investigated[41]; but all these proteins of Sb were >99% identical to the respective proteins in Sc. Also, Sb was grown on non-fermentable carbon source (Glycerol) to find if the non-sporulation of Sb is governed by respiration-sensing pathway[47] (detailed in Supplementary Methods). The growth of Sb was found on the non-fermentable carbon source (Supplementary Figure 3). Thus, the non-sporulation phenotype of Sb is unlikely be due to a mutation in sporulation and respiration pathway genes. Since functional MAT loci are also critical for sporulation[48], we speculate that mutations at these loci, as seen in MATa sequence in Sb-biocodex with eight substitutions and seven insertions compared to Sc MATa (Fig. 2), might be responsible for the non-sporulation phenotype of Sb.

Gene copy number variations in Sb strains as compared to Sc

Genes absent

The mapping studies to Sc genes of yeastmine database[49] revealed that 50 genes had no reads mapped onto them. Also, we found that 94 genes apart from these 50 genes had read coverage <20 reads. The absence of these genes was confirmed by subjecting these genes as a query to BLASTn against the Sb-biocodex and Sb-unique28 PacBio assembly with complete chromosomes. Out of these 144 genes, 85 were dubious ORFs, 32 were uncharacterized genes, and 27 were functionally verified genes. The verified genes include two maltase utilization genes (MAL11 and MAL13), two hexose transporters (HXT9 and HXT11), four asparagine catabolism gene (ASP3-1, ASP3-2, ASP3-3 and ASP4-4), three palatinose utilization genes (IMA2, IMA3, IMA4), two putative membrane glycoprotein (VTH1 and VTH2) and ARN2, REE1, AYT1, AIF1, COS10, ENB1 and BDS1 (Table 2). Except ASP3 locus, all these absent genes belong to telomeric or subtelomeric regions of chromosome.

Table 2

Absent genes in Sb-biocodex and Sb-unique28.

Systematic	Genes	Gene Function
YOL165C	AAD15	Aryl-Alcohol Dehydrogenase
YNR074C	AIF1	Mitochondrial cell death effector
YHL047C	ARN2	Transporter
YLR155C	ASP3-1	Cell-wall L-asparaginase II involved in asparagine catabolism
YLR157C	ASP3-2	Cell-wall L-asparaginase II involved in asparagine catabolism
YLR158C	ASP3-3	Cell-wall L-asparaginase II involved in asparagine catabolism
YLR160C	ASP3-4	Cell-wall L-asparaginase II involved in asparagine catabolism
YLL063C	AYT1	Acetyltransferase
YOL164W	BDS1	Bacterially-derived sulfatase
YLR465C	BSC3	Bypass of Stop Codon
YNR075W	COS10	Protein of unknown function
YGR295C	COS6	Protein of unknown function
YOL158C	ENB1	Endosomal ferric enterobactin transporter
YOL156W	HXT11	Putative hexose transporter that is nearly identical to Hxt9p
YJL219W	HXT9	Putative hexose transporter that is nearly identical to Hxt11p
YOL157C	IMA2	Isomaltase
YIL172C	IMA3	Isomaltase
YJL221C	IMA4	Isomaltase
YGR289C	MAL11	High-affinity maltose transporter (alpha-glucoside transporter)
YGR288W	MAL13	MAL-activator protein
YIR041W	PAU15	seripauperin
YKL224C	PAU16	seripauperin
YJL217W	REE1	Cytoplasmic protein involved in the regulation of enolase (ENO1)
YAL064C-A	TDA8	Topoisomerase I Damage Affected
YOR068C	VAM10	Vacuolar Morphogenesis
YIL173W	VTH1	Putative membrane glycoprotein
YJL222W	VTH2	Putative membrane glycoprotein

Also, the yeastmine database genes were subjected to BLASTn against all strains of Sc included in this study to find genes unique to Sb (Supplementary File V). MAL11, MAL13, and ARN2 were present in more than 70% of the strains of different subgroups of Sc strains but were absent in all the probiotic strains. We found that the BDS1 gene was present in tree isolates, laboratory strains, and environmental samples but was absent in wine strains, beer strains, clinical, fruit derived, bakery and bioethanol producing strains. Similarly, the REE1 gene was present in tree isolates, laboratory strains and environmental samples and a few strains belonging to subgroup wine, beer and clinical but was absent in probiotic Sb strains.

The ASP3 locus was present in more than 80% of the laboratory or industrial strains and bioethanol producing strains and Sc strainYJM1383, a fruit derived strain; Sc strain CLIB324, a bakery strain, and a few clinical strains (YJM248, YJM339, YJM451, YJM693, YJM1078, and YJM1311). It was absent in the wine, distillery, and probiotic strains. The ORFs coding for this locus on chromosome XII is adjacent to the ribosomal DNA locus. The hexose transporter family is large and comprises of HXT1-17 genes[50] of which HXT11 and HXT9 were absent from all strains of Sb. HXT11 and/or HXT9 null mutants of Sc are resistant to cycloheximide, sulfomethuron methyl, and 4-NQO (4-nitroquinoline-N-oxide)[51] indicating Sb strains also could be resistant to these chemicals.

Multi-copy genes

The variation in the copy number of genes in a genome can have phenotypic and physiological differences[52]. Genes for PAU proteins, a member of the seripauperin multigene family, were found to be present in 18–20 copies in the genome, and gag-pol fusion genes were present in 15 copies in the whole genomes of Sb-biocodex and Sb-unique28 (Table 3). THI13 is present in five copies at the sub-telomeric regions and IMD3 and COS3 are present in four copies at the telomeric regions of Sb chromosomes. Imd3 catalyzes the rate-limiting step in the de novo synthesis of GTP[53] and Cos3 is involved in salt resistance[54] in Sc. The clusters of duplicated and triplicated genes mostly encode stress-related proteins, elongation factors, ribosomal proteins, kinases and transporters, fluoride export and altering replication stress tolerance. These duplicated genes could be helping in better adaptation of Sb to the harsh conditions of the mammalian host.

Table 3

Multicopy genes in Sb-biocodex, Sb-unique28 and Sc S288C.

	Sb biocodex	Sb unique28	Sc S288C
Total clusters	163	150	146
Clusters with two proteins	148	137	127
Clusters with more than two proteins
Genes in multiple copies	Sb biocodex	Sb unique28	Sc S288C
Seripauperin PAU	18	20	7
gag-pol fusion proteins	16	16	49
Thi13	5	6	1
IMP dehydrogenase IMD3	4	2	1
Cos3p	4	5	1
YIL169C-like protein	3	4	2
Aad4p	3	2	1
Fex1p	3	3	1
ribosomal 60S subunit protein L2B	3	4	1
Hsp32p	3	3	1
Y’ element ATP-dependent helicase protein 1 copy 1	2	5	5

Ty elements in Sb

Ty1, Ty3, and Ty4 elements were absent in all the Sb genomes whereas Ty2 elements were present in Sb-biocodex, Sb-kirkman and Sb-unique28 and one Ty5 element was present in all Sb strains except Sb-17 and Sb-MYA-796 (Supplementary Figure 4). Also, the presence of genes encoding gag-pol and gag-co-pol fusion proteins was confirmed by read mapping and coverage analysis. These were either contained in the Ty2 or Ty5 elements. The presence of Ty elements was compared in between complete genomes of Sb viz. Sb-biocodex and Sb-unique28; where the elements are present in the same chromosomes with some deviations in positions (Table 4; Supplementary File VI). Ty1–4 elements integrate near tRNA or RNA polymerase III genes[55], but analyzing the neighboring genes of these Ty elements, we found only one Ty element in each of Sb-unique28 and Sb-biocodex has Ribosomal 40S subunit protein upstream of it.

Table 4

Distribution of Ty elements in Sb strains.

	Sb-17	Sb-biocodex	Sb-EDRL	Sb-kirkman	Sb-MYA-796	Sb-unique28	Sb-unisankyo	Sc S288C
gag	1			1	1	1	1	1
gag co pol		3	5	3		3	1	45
gag pol		2				3		45
pseudo	2		2	2	2	2	2	2
Suppressor SPT7	1	1	1	1	1	1	1	1
Ty1 element								31
Ty1 LTR	147	191	173	190	151	200	169	483
Ty2 element		10		1		7		13
Ty2 LTR	5	4	12	9	8	14	11	32
Ty3 element								2
Ty3 LTR	26	46	40	45	29	45	33	42
Ty4 element								3
Ty4 LTR	15	16	17	17	15	17	15	35
Ty5 element		1	1	1		1	1	1
Ty5 LTR	5	3	3	4	4	3	3	8
Ty A								1
Ty B								1

Flocculation and adhesion

FLO1, FLO5, FLO8, FLO9, FLO10, FLO11, FIG2, and AGA1 encode flocculation proteins that belong to yeast adhesin families, and their sufficient expression leads to flocs, flor, biofilms or filaments formation by either binding to other yeast cell receptors or foreign surfaces[56]. The protein sequences of these flocculins obtained from SGD were subjected to BLASTp against Sb and Sc proteomes (Supplementary File VII). Flocculation genes are characterized by the presence of a large number of repetitive sequences with linear correlation to their size[57]. FLO1 and FLO5 are paralogs that arose from segmental duplication[58] of which we could trace the presence of FLO1 gene in our Sb genomes. FLO8, FLO10, FLO11, FIG2, and AGA1 were also present in all strains of Sb. FLO1, FLO5, FLO8, FLO9, FLO10, and FLO11 are telomeric genes and have repeats[59, 60]. These were searched in other Sc genomes, but we found that most of them encode truncated proteins possibly owing to limitations of sequencing technology. In Sb genomes, we found that these genes encode either truncated proteins or could not be traced in Sb-EDRL, Sb-kirkman, Sb-unisankyo, Sb-17, and Sb-MYA-796. The complete sequences of flocculins from Sc and Sb strains were retrieved and were aligned using MUSCLE[61] and were edited manually to trace variation in the number and period of the repeats in sequences (Supplementary Figure 5). Seven copies of 45 residues were identified in FLO1 protein of Sb and Sc YJM1385 (fruit borne) and Sc YJM1129 (brewery strain) whereas only one copy of this period was present in other Sc strains. The FLO8 protein of Sb was identical to that in most of the Sc strains except Sc strains S288C, BY4741, BY4742, FY1679, JK9-3d, SEY6210, W303, X2180-1A and YPH499 where a point mutation has resulted in the truncation of the protein[62]. FLO10 and FLO11 domains are present in all strains of Sb with repeats where the copy number and a period length of the repeats were similar to all the Sc strains.

Introgression of Zygosaccharomyces bailii proteins into Sb and Sc wine strains

Ten genes on chromosome IV of Sb-biocodex and Sb-unique28 were found to encode proteins >90% identical to Zygosaccharomyces bailii ISA1307 proteins (Fig. 3). On investigating this region, we found that five genes of Z. bailli had introgressed and further undergone duplication. Three of these genes were annotated as encoding uncharacterized proteins; one encodes a probable 5-oxoprolinase, and one is an allantoate transporter. These genes were searched in the sequenced Sc strains too, where a single copy of these genes was present in similar fashion in Sc UFMG A-905 strain with probiotic properties and wine strains YJM339, RM11-1a, L1528, and YS9. Three copies of these genes were also present in Sc wine strain BC187. Few of these introgressed genes were also present in bioethanol producing strain Sc JAY291; wine strains Sc EC1118, Vin13, VL3, AWRI796 in single copy and Sc wine strain LalvinQA23 has two copies of these Z. bailli genes.

Figure 3

Introgression of five genes in two copies from Zygosaccharomyces bailii to Chromosome IV of Sb genomes.

Taxonomic position of Sb

Sb and Sc shares more than 99% genomic relatedness as determined by Average Nucleotide identity (ANI) (Supplementary File VIII). Thus we retrieved the core proteome from all 145 Sc strains and 7 Sb strains and a outgroup species S. kudriavzevii. 182 proteins for which orthologs could be obtained in all 153 organisms were retreived and concatenated to find the taxonomic position of Sb with comparison to Sc. The ML based tree rooted the tree at S. kudriavzevii, the outgroup species, which further clustered all the Sc strains (Fig. 4). The Sc strains in the phylogenetic tree were grouped as per their isolation source. All the Sb strains were grouped in a clade, where Sc UFMG A-905 strain is closer to Sb-unique28. Sb-biocodex was present at the root of the Sb clade that groups Sb-17, Sb-EDRL, Sb-unisankyo, Sb-kirkman, and Sb-MYA-796. The Sb strains share the clade with wine strains Sc strains BC187, YJM1387, YJM1417, YJM1332 and R008, brewery strains Sc YJM1477 and Sc strain YJM1242 isolated from fruits. Separate clusters were observed in case of laboratory strains whereas clinical isolates were grouped into three distinct clusters in the tree. Sc strains isolated from fruits were scattered across tree but were closer to Sc wine strains or tree isolates. In terms of taxonomy, it is clear that the Sb strains are closely related to the Sc wine strains, and it would be of interest to explore the probiotic potential of these wine strains.

Figure 4

Maximum Evolution tree for 182 orthologous proteins of all strains of Sb, Sc and S. kudriavzevii as outgroup species. The taxa are shaded based on the isolation source of strains as boulardii: gold; Fruits: lightsalmon; Wine: hotpink; Tree isolate: skyblue; Brewery: lawngreen; Lab strains: palegoldenrod; Bakery: darkorange; Bioethanol: peru; Clinical: mediumseagreen; Environmental samples: cornflowerblue.

Discussion

The complete genomes of Sb determined in this study are the best assemblies of the yeast known, as the long PacBio reads used in the study assisted in the identification of complete chromosomes, telomeres and complete structures of Ty elements which could not be identified using Illumina HiSeq short read data. The comparative genomic hybridization analysis for Sb revealed at the first place that the Ty1-2 elements were absent from Sb [18, 41] which could not be identified in Sb-EDRL and Sb-unisankyo draft assemblies but in complete genomes of Sb-biocodex and Sb-unique-28, we were able to identify 15 complete Ty elements. Chromosome IX has been mentioned to exhibit trisomy[1], but such event could not be traced instead we found that the chromosome XII had double the read coverage (400x) as compared to the other chromosomes (200–270x) which could be owing to the aneuploidy of the chromosome XII. The rep2 gene of 2-micron circle plasmid had G5297 -> A synonymous mutation resulting in A296 -> V amino acid change as reported earlier[63]. Sb strains share ~3100 orthologous proteins with one or more Sc strains suggesting large conserved protein repertoire between Sc and Sb. The Sb proteome had high conservation level among the strains of Sb owing to the presence of the protein homologs in one or the other strains of Sb. Sb strains are likely to be homothallic and diploid due to the presence of both MAT alleles and HO gene. Being diploid the yeast should sporulate, but the sporulation was absent in Sb as revealed by sporulation assay which has already been established in a previous study[1]. The respiration efficiency of Sb strains was determined by its growth on non-fermentable carbon sources[47] and hence the non-sporulation behavior of the organism is not owing to the respiration deficiency. Even the absence or divergence of any sporulation gene was not observed in Sb proteome in comparison to that of Sc. Possibly the deficiency of the mating type genes[48] in Sb could be leading to the non-sporulation behavior.

Sb can prevent antibiotic-associated diarrhea, recurrent Clostridium difficile-associated diarrhea and colitis, Traveller’s diarrhea, acute bacterial and viral diarrhea, anti-inflammatory bowel diseases by various mechanisms[10]. Sb, with antimicrobial properties, secretes 54 kDa[12, 13], 63 kDa[11], and 120 kDa[64] proteins that exhibit the protection of gut microflora against pathogenic bacteria either through cleaving the toxin or by reducing the cAMP level. These proteins were searched in Sb and Sc, and it was established that these proteins are not unique to Sb but were present in Sc too[38] (Supplementary Methods). Sb has been shown to hinder the biofilm formation by pathogenic strains because of steric hindrance caused by its larger size as compared to bacteria[65]. Also, Sb exerts its antimicrobial effect by adhering to intestinal mucus membrane and eliminating pathogens by flow preventing their adhesion to the intestine[8]. Adhesion to other foreign surfaces has been reported to be a critical step for pathogenic as well as a probiotic organism[57]. Yeast also chooses its lifestyle according to its environment and can form different colonies such as non-adhesive colonies, self-adhesive non-dissolvable colonies, biofilms during foreign adhesion or flocs or flor[56]. All flocculin genes harbor a large number of repeats that tend to increase or decrease in copy numbers affecting the degree of flocculation and sensitivity to stress conditions[56]. We could identify complete flocculin genes in the whole genomes of Sb where the repeats and their copies were varying even within Sb genomes but were consistent in between Sb-biocodex and Sb-unique28. Sb harbors all the flocculin genes required for protection against environmental stress as ethanol and fungicides (FLO1)[66], floc formation (FLO10 and FLO11)[67], and biofilm formation (FLO10, FLO11, and FIG2)[67, 68]. FLO8 gene required for expression of FLO1 and FLO11 was present in Sb, but was truncated in laboratory strains, impairing their flocculation and adhesion to the foreign surfaces[62]. These genes are located at telomeres and are highly repetitive, and the maximum number of repeats identified in the strains of Sb could be conferring higher adhesive properties to the organism.

During evolution, the Saccharomyces yeast can undergo a process of gene duplication, polyploidy, chromosomal rearrangements, interspecific hybridization, and introgression[69]. The process of eukaryote-to-eukaryote gene transfer events and introgression in Sc strains have been validated through genetic experiments and certain regions among Sc strains have been reported to have similarity with Z. bailii, S. arboricola, S. bayanus and other yeasts[70, 71]. A particular region in wine yeast Sc EC1118 has been proposed earlier to be transferred from Z. bailii type strain CBS680[70]. Similarly, a specific region of 10 proteins in chromosome IV of Sb was syntenic to the five proteins in Z. bailii ISA1307 which has got duplicated in Sb and few wine strains of Sc.

The taxonomic position of Sb as a separate species has been controversial[10, 41, 72–75]. It was initially considered as a separate species of genus Saccharomyces, but CGH analysis characterized it as a strain of Sc. The core proteome based phylogeny, obtained from the Sb, Sc, and S. kudriavzevii could resolve the clades. All Sb strains got clustered together in a clade along with Sc UFMG A-905 strain, a Sc strain with probiotic properties where Sb-unique28 shared the sister clade with Sc UFMG A-905. All the phylogenies drawn revealed concordantly that the wine strains of Sc are closer to the Sb.

There is no doubt that the Sb belongs to Sc species and is a strain of Sc, but the probiotic features of Sb make it a yeast with beneficial effects in gastrointestinal disorders[3, 5, 7, 12, 14, 24, 26, 41, 76–79]. The genomic perspective of the organism with relevance to its probiotic features was examined in the study where we couldn’t identify any specific and unique proteins in Sb, since the Sb proteins are homologous with one or the other proteins of Sc strains. The genomic perspective in this study revealed that the Sb probiotic strains are closer to wine strains of Sc than industrial or baking strains; as revealed by the absence of ASP3 locus, the introgression of Z. bailli proteins and the core proteome based taxonomic placement of probiotic Sb and wine Sc strains. Sc BC187 a wine strain shows maximum similarity with the Sb strains and might be explored for its probiotic properties similar to Sc strain UFMG A-905[80, 81].

Methods

Isolation and purification of Sb genomic DNA

The lyophilized yeast Sb available in the market as probiotics in sachets were used for isolation of source DNA for Sb-biocodex (Florastor), Sb-EDRL (Dr. Reddy’s Laboratories) and Sb-unique28 (Unique Biotech). Two cultures of Sb from Unisankyo Ltd. (Now Sanzyme Ltd.) and Kirkman Labs that were maintained at MTCC (IMTECH, India) since 2003 were obtained for DNA isolation and sequencing. DNA isolation was performed using the ZR Fungal/Bacterial DNA miniprep kit (Zymogen), and purity index was checked as the ratio of OD at 260/280 nm was >1.8 as observed by NanoDropND-1000 spectrophotometer. Detailed isolation and purification methods have been provided in supplementary material.

Genome Sequencing

Sb-unique28 and Sb-biocodex were sequenced using PacBio P6C4 chemistry using eight and nine SMRT cells, respectively. 101-bp paired-end (PE) shotgun data from Illumina HiSeq-1000 high-throughput sequencing technology was also obtained for Sb-biocodex. The sequencing depth for PacBio sequencing was approximately 200x for both Sb-biocodex and Sb-unique28. Sb-EDRL was sequenced earlier using Roche 454 sequencing technology which now has also been sequenced on Illumina HiSeq 1000 platform to obtain 101-bp PE shotgun data twice and one 2 K and one 8 K mate-pair (MP) library shotgun data. Sb-kirkman was sequenced using Illumina HiSeq1000 to get 101-bp PE shotgun data along with 2 K and 8 K MP reads. For Sb-unisankyo, only 101 bp shotgun data was obtained from Illumina HiSeq 1000 sequencing platforms. The Illumina HiSeq-1000 sequencing was performed at C-CAMP, Bangalore, India and PacBio P6C4 chemistry SMRT cells sequencing data was obtained from Genome Quebec Centre, McGill University, Canada. Detailed Sequencing protocols have been provided in supplementary material.

Genome Assembly and Annotation

Sb-biocodex and Sb-unique28 SMRT cells were assembled using Hierarchical Genome Assembly Process (HGAP) v2.0[82] pipeline of the SMRT Portal. The Sb-biocodex and Sb-unique28 contigs were aligned to Sc reference strain S288C using Mauve aligner[83] and the contigs completely mapping to the chromosomes were submitted to NCBI as complete chromosomes. Some unplaced contigs were obtained which were subjected to BLASTn[84] against the finalized chromosomes of respective strains and the NT database to find if any contig belongs to a plasmid or mitochondrial genomes.

Sb-EDRL, assembled earlier with Roche 454 shotgun data and submitted to NCBI (ATCS01000000), was further scaffolded and gapfilled using SSPACE v3.1[85] and GapFiller v1.10[86] with Illumina PE and MP data. Sb-kirkman and Sb-unisankyo were assembled de novo using SPAdes v3.1[87]. Further, these assemblies were scaffolded using SSPACE v3.0 and gapfilled using GapFiller v1.10. Also, the Illumina Next-Generation Sequencing (NGS) data for four Sb: Sb-biocodex, Sb-EDRL, Sb-kirkman and Sb-unisankyo were mapped on to the 16 chromosomes of Sb-biocodex to find the gaps that were not covered by Illumina reads. Final assemblies were submitted to GenBank under accession numbers Sb-biocodex: LIIL01, Sb-unique28: LIOO01, Sb-kirkman: LOMX01, Sb-unisankyo: LNQF01 and Sb-EDRL ATCS02.

All the Sb strains were annotated using Augustus[88] as gene predictor with species model ‘Sc S288C’ and tRNA was predicted by tRNAscan-SE 1.23[89]. Features thus annotated were subjected to BLASTp against Saccharomyces Genome Database (SGD)[39] protein dataset and non-redundant (NR) protein sequence database for functional characterization of the proteins with an E-value cutoff of 1e−5.

Gene copy number variations

The complete set of Sc genes present in yeastmine database[49] were mapped by the Illumina shotgun reads of Sb strains Sb-biocodex, Sb-kirkman, Sb-unisankyo, and Sb-EDRL. Genes with no read coverage were checked in the complete PacBio genome assembly of Sb-biocodex for their absence. The genes duplicated within Sb-biocodex and Sb-unique28 genomes with >90% identity, and >90% query coverage were identified.

Presence-Absence Variations (PAVs)

The proteins or genes involved in adhesion, flocculins, sporulation, meiotic, mitotic, galactose utilization and palatinose utilization were downloaded from SGD and were subjected to BLASTp or BLASTx against the proteome of all Sb and Sc and the hits thus obtained were filtered at >50% query coverage and >30% identity. The duplicate hits were removed, and the PAVs were plotted as matrix across all 7 Sb and 145 Sc genomes.

Genome datasets used for comparison

The Sb genomes were compared to 145 strains of Sc reported at SGD and NCBI. The annotations for all strains of Sc were obtained from SGD and NCBI. NCBI reports 168 genomes as on 01-May-2015 of which 50 were reported in SGD. Five strains (Sc FL100, Sc RM11-1a, Sc Sigma1278b, Sc W303 and Sc YPS163) were excluded as their updated versions were considered for the analysis. Two strains (Sc CLIB382 and Sc M22) were excluded from annotations as the number of scaffolds was more than 6000. Sc strain T73 and Y10 were also eliminated as the number of annotated features was less than 3000. Fourteen strains with no annotations available at NCBI and SGD were not used for the comparative analysis. Unannotated Sc strain UFMG A-905 available from NCBI was utilized in the comparative study as the strain has been mentioned to exhibit probiotic effects. Two Sb-17 and Sb-ATCC-MYA-796 were also used for the comparison. The features prediction was made using Augustus[88] for these strains. All these strains were divided into nine subgroups based on the origin of the strains as Fruit-derived, Wine, Tree-isolates, Beer, Laboratory, Bioethanol, Bakery, Clinical and Environmental.

Identification of Ty elements

Retrotransposons sequences downloaded from SGD database were subjected to BLASTn against the Sb genomes. The BLAST results obtained were further filtered with query coverage of 90% and best hits were retrieved. Further, the matched regions were screened manually.

Core-proteome Analysis

The orthologous pairs of proteins across all Sb and Sc proteomes were extracted using Proteinortho v2.3[90] PERL script and was considered as the core proteome of all Sb strains. The homologs of proteins of one Sb strain were identified by subjecting the proteins to BLASTp against proteome of other Sb strains.

Taxonomy

The core proteins retrieved from proteome of Sb and Sc and outgroup species S. kudriavzevii IFO 1802 were concatenated and aligned using MAFFT[91]. The alignment was fed to MEGA v6.0[92] for generation of Minimum Evolution (ME) tree at 100 bootstrap values. The evolutionary distances were calculated using Dayhoff matrix based method.

Data access

This Whole Genome Shotgun project has been deposited at DDBJ/EMBL/GenBank along with the plasmid and ITS sequences retrieved from all Sb. The version described in this paper is the first version for Sb-biocodex- LIIL01000000, Sb-unique28- LIOO01000000, Sb-kirkman- LOMX01000000, Sb-unisankyo- LNQF01000000 and second version of Sb-EDRL- ATCS02000000.

Conclusions

Two complete genomes and three draft genomes of Sb were sequenced and assembled. The complete genomes revealed the presence of Ty2 elements and gag-co-pol genes in Sb unlike the complete absence of Ty1/2 elements in Sb suggested previously. Homozygous diploid probiotic yeast Sb had non-sporulation phenotype for which the absence/divergence of sporulation genes or respiration efficiency is not responsible, but the deficiency in mating genes may be playing a role.

The physiological and molecular differences making Sb different from Sc were explored through genome analysis. We found that the HXT11 and HXT9 hexose transporter genes were absent in Sb only but were present in all Sc strains. The asparagine utilization (ASP3) locus was absent in Sb and Sc wine and distillery strains and were only present in Sc laboratory and bioethanol and some clinical strains. All flocculins except FLO5 protein and adhesins were present across all Sb, and we found that these have a larger number of repeats comparable to most of the Sc strains probably enabling adhesive properties to the organism.

The introgression of five genes, related to transporters, from Z. bailii ISA1307 to Sb was found to be present in similar fashion in Sc wine strains too. The taxonomic position of Sb was derived using 182 core proteins as the high genomic relatedness between Sb and Sc did not allow a few molecular markers to resolve the phylogeny. Taxonomically the probiotic yeast shares the clade with Sc UFMG A-905 and wine strains. In light of the genomic and taxonomic evidence we found that the probiotic yeast is closer to the wine strains as compared to industrial strains.