A Gene Cluster for Biosynthesis of Mannosylerythritol Lipids Consisted of 4-O-β-D-Mannopyranosyl-(2R,3S)-Erythritol as the Sugar Moiety in a Basidiomycetous Yeast Pseudozyma tsukubaensis

Mannosylerythritol lipids (MELs) belong to the glycolipid biosurfactants and are produced by various fungi. The basidiomycetous yeast Pseudozyma tsukubaensis produces diastereomer type of MEL-B, which contains 4-O-β-D-mannopyranosyl-(2R,3S)-erythritol (R-form) as the sugar moiety. In this respect it differs from conventional type of MELs, which contain 4-O-β-D-mannopyranosyl-(2S,3R)-erythritol (S-form) as the sugar moiety. While the biosynthetic gene cluster for conventional type of MELs has been previously identified in Ustilago maydis and Pseudozyma antarctica, the genetic basis for MEL biosynthesis in P. tsukubaensis is unknown. Here, we identified a gene cluster involved in MEL biosynthesis in P. tsukubaensis. Among these genes, PtEMT1, which encodes erythritol/mannose transferase, had greater than 69% identity with homologs from strains in the genera Ustilago, Melanopsichium, Sporisorium and Pseudozyma. However, phylogenetic analysis placed PtEMT1p in a separate clade from the other proteins. To investigate the function of PtEMT1, we introduced the gene into a P. antarctica mutant strain, ΔPaEMT1, which lacks MEL biosynthesis ability owing to the deletion of PaEMT1. Using NMR spectroscopy, we identified the biosynthetic product as MEL-A with altered sugar conformation. These results indicate that PtEMT1p catalyzes the sugar conformation of MELs. This is the first report of a gene cluster for the biosynthesis of diastereomer type of MEL.

collection and analysis, decision to publish, or preparation of the manuscript. The specific roles of these authors are articulated in the 'author contributions' section.
Competing Interests: SY and TK are employees of Toyobo Co., Ltd. HK, TF, SY, TK and TM are inventors of submitted patent which related to this study (PCT/JP2016/061760). This does not alter the authors' adherence to PLOS ONE policies on sharing data and materials. The authors have declared that no other competing interests exist.
concentrations and temperatures [34] and shows higher hydration ability than conventional type of MELs [35]. Therefore, diastereomer type of MEL-B produced by P. tsukubaensis may facilitate the use of MELs in aqueous solutions. Over the past decade, a gene cluster involved in the biosynthesis of various glycolipids (such as MELs, cellobiose lipids and sophorolipids) has been identified [36][37][38][39][40]. In U. maydis, the MEL biosynthetic pathway consists of five proteins: an erythritol/mannose transferase (Emt1p), two acyl transferases (Mac1p and Mac2p), an acetyltransferase (Mat1p) [36,37] and a putative transporter (Mmf1p) (Fig 2). While the products of P. antarctica are similar to those of U. maydis, the sugar conformation of MEL produced by P. tsukubaensis differs from those of the other Ustilago and Pseudozyma species. Furthermore, the degree of acetylation of the mannose moiety in P. tsukubaensis differs from that of P. antarctica, because the main product of P. tsukubaensis is a monoacetylated MEL (MEL-B). Therefore, the reaction of MEL biosynthesis in P. tsukubaensis differs from the reactions in P. antarctica and U. maydis, particularly with respect to sugar conformation and acetylation. We therefore focused the current study on the gene cluster for MEL biosynthesis in P. tsukubaensis.
We identified the gene cluster responsible for biosynthesis of the diastereomer type of MEL-B in P. tsukubaensis NBRC1940 based on amino acid sequence analysis. The putative amino acid sequence encoded by the gene PtEMT1 exhibited high identity to that of P. antarctica, but had an independent position on the phylogenic tree. We altered the sugar conformation of MELs in P. antarctica from S-form to R-form by introducing PtEMT1 from P. tsukubaensis into a gene-disrupted mutant of P. antarctica lacking MEL biosynthesis.

Sequence analysis
The draft genome sequence of P. tsukubaensis NBRC1940 was performed (reported elsewhere). The BLAST program was used for sequence similarity searching in a database available on the NCBI website (http://blast.ncbi.nlm.nih.gov/Blast.cgi). Multiple sequence alignments were displayed using the ClustalW program. Phylogenetic analysis was performed using the neighborjoining method [41] with the program MEGA6 [42] and bootstrap analysis based on 1,000 replicates [43].

Transformation
The plasmid pUXV1_neo::PtEMT1 and pUXV1_neo::PaEMT1 were introduced into ΔPaEMT1 by electroporation according to Morita et al. [19] with suitable modification. ΔPaEMT1 was grown in 3 mL MEL production medium (1 g L -1 of yeast extract, 3 g L -1 of NaNO 3 , 0.3 g L -1 of KH 2 PO 4 and 0.3 g L -1 of MgSO 4 Á7H 2 O) containing 10% (w/v) glycerol at 25°C for 3 days as a seed culture. One milliliter of seed culture was inoculated into 50 mL of MEL production medium containing 10% (w/v) glycerol and cultivated at 25°C for 15 h with 250 stroke min -1 . The cells were harvested by centrifugation at 5,000 rpm for 5 min and washed twice with chilled 1 M sorbitol. 0.1 mL of cell suspension containing about 3 μg of plasmid was pulsed by electroporation using the Bio-Rad Gene Pulser II with Pulse Controller Plus (Bio-Rad, Tokyo, Japan). The cells were pulsed twice with a square-wave electroporation pulse of 1000 V and a pulse length of 1.0 ms at a pulse interval of 5 s. The electroporated cells were immediately diluted in 0.9 mL of chilled 1 M sorbitol and 3 mL of YM medium and incubated at 25°C for 1 h. After incubation, an aliquot (0.2 mL) was spread on the YM medium plate containing 500 μg mL -1 G418, and the resulting colonies were grown at 25°C for 4 days.

MEL production
ΔPaEMT1 harboring pUVX1_neo::PtEMT1 or pUXV1_neo::PaEMT1 were cultivated in 2 mL YM medium (3 g L -1 of yeast extract, 3 g L -1 of malt extract, 5 g L -1 peptone and 10 g L -1 glucose) containing 200 μg mL -1 G418 at 25°C for 3 days as a seed culture. In a 300-mL flask, the seed culture was inoculated with 30 mL MEL production medium containing 10% (w/v) glucose as a carbon source and incubated at 25°C for 7 days with 200 stroke min -1 . 200 μg mL -1 of G418 was supplied in the culture medium to maintain the plasmid. The produced MELs were detected using TLC [19]. The MELs were extracted from the cell culture with an equal volume of ethyl acetate, and 50 μL of ethyl acetate extracts were analyzed by TLC using chloroform, methanol and NH 4 OH in a 65:15:2 (v:v:v) ratio as an eluent. The MELs were detected on the TLC plate by spraying with 2% anthrone-sulfate reagent and heating at 90°C for 5 min. A mixture of purified MEL-A, MEL-B and MEL-C was used as a reference.

MEL purification
The ethyl acetate fractions containing MELs were evaporated. The concentrated MELs were dissolved in chloroform and purified using silica gel (Wako-gel C-200) column chromatography with a gradient elution of chloroform/acetone (10:0 to 0:10, v/v) mixtures as solvent systems [23]. The purified MEL-A was used in the following experiments.

Structural analysis
The structure of the purified MEL-A was characterized by 1 H and 13 C nuclear magnetic resonance spectroscopy (NMR) with a Bruker AVANCE 400 (400 MHz) at 30°C in a CDCl 3 solution. Tetramethylsilane [(CH 3 ) 4 Si] was used as an internal chemical shift standard and the purified MEL-A was used as a reference for NMR analysis.

Identification of the gene cluster of MEL biosynthesis in P. tsukubaensis NBRC1940
Based on BLASTP analyses, the genes involved in MEL biosynthesis (emt1, mac1, mac2, mmf1 and mat1 for U. maydis; PaEMT1, PaMAC1, PaMAC2, PaMMF1 and PaMAT1 for P. antarctica) were found in the draft genome of P. tsukubaensis NBRC1940 (PtEMT1, PtMAC1, PtMAC2, PtMMF1 and PtMAT1, respectively) (Fig 4). The gene arrangement in the MEL biosynthesis cluster of P. tsukubaensis NBRC1940 is more similar to U. maydis than P. antarctica strains JCM10317 and T-34 (Fig 4). The PtEMT1 and PtMAC2 of P. tsukubaensis NBRC1940 are rearranged, as well as U. maydis, compared with P. antarctica JCM10317 and T-34.
The amino acid sequence homology of five proteins involved in the biosynthesis of MELs (PtEMT1p, PtMAC1p, PtMAC2p, PtMMF1p and PtMAT1p) was compared to those from U. maydis and P. antarctica strains JCM10317 and T-34 (Table 1). According to BLASTP analysis, PtEMT1p, PtMAC1p, PtMAC2p and PtMMF1p have high (over 50%) homology to the corresponding proteins in U. maydis and P. antarctica.
Amino acid sequence analysis of PtEMT1 from P. tsukubaensis NBRC1940 The amino acid sequence of PtEMT1p is similar to those of the corresponding proteins in U. maydis and P. antarctica. However, the sugar conformation of MEL produced by P. tsukubaensis differed from those of other MEL producers. We thus performed further amino acid sequence analysis (Table 2). PtEMT1p consists of 612 amino acids, and the corresponding proteins from U. maydis UM521, P. antarctica JCM10317 and P. antarctica T-34 consist of 615, 617 and 617 amino acids, respectively. Surprisingly, only nine proteins, from strains in the genera Ustilago, Melanopsichium, Sporisorium and Pseudozyma, showed high sequence identities with PtEMT1p (69-72%) ( Table 2). All of these strains, except for those in the genus Pseudozyma, have been reported as plant pathogens [9,45,46]. Table 3 shows Emt1p sequence identities among P. tsukubaensis NBRC1940 and nine other strains. While PtEMT1p has approximately 70% identity with homologous proteins, other pairs of species (e.g. P. aphidis and P. antarctica or S. scitamineum and S. reilianum) share more than 90% sequence identity. We aligned nine proteins that exhibited high identity to PtEMT1p using ClustalW. While sequence identity was greater than 69%, we found three regions with relatively low identity (Fig 5). The three regions in P. tsukubaensis NBRC1940 each consisted of about 30 amino acids. The amino acid position of regions I, II and III were from 272 to 300, 372 to 397, and 569 to 597, respectively (Fig 6). Putative sugar binding domain was not found. Phylogenetic analysis based on the amino acid sequences of Emt1p indicated that PtEMT1p diverged significantly from the other nine proteins (Fig 7). Thus, amino acid sequence analysis suggests that PtEMT1p is a novel structure that differs substantially from homologous proteins.
MEL production by P. antarctica ΔPaEMT1 harboring pUXV1_neo:: PtEMT1 from P. tsukubaensis NBRC1940 Because ΔPaEMT1 derived from P. antarctica T-34 had its PaEMT1 function disrupted, it provides a useful host to investigate the function of PtEMT1 in MEL production. The plasmid harboring PtEMT1 from P. tsukubaensis NBRC1940 was introduced into ΔPaEMT1. The empty vector pUXV1_neo was used as a negative control, and pUXV1_neo::PaEMT1 which harboring PaEMT1 from P. antarctica T-34 was used as a positive control.
The ΔPaEMT1 harboring pUXV1_neo::PtEMT1 was cultivated in MEL production medium containing 10% (w/v) glucose for 7 days at 25°C. The produced MELs were extracted by ethyl acetate and detected by thin-layer chromatography (TLC) analysis (Fig 8). ΔPaEMT1 harboring pUXV1_neo::PtEMT1 and pUXV1_neo::PaEMT1 produced MELs from glucose, whereas ΔPaEMT1 harboring empty vector PUVX1_neo failed to produce MELs, as expected. Thus, the gene PtEMT1 from P. tsukubaensis NBRC1940 restored MEL production in ΔPaEMT1.  To determine the structure of MEL-A produced by ΔPaEMT1 expressing PtEMT1, we subjected purified MEL-A to NMR analysis and compared the signal pattern to that of conventional type of MEL-A produced by ΔPaEMT1 expressing PaEMT1 (Fig 9A and 9B). Each signal was assigned as previously described [47][48][49]. The diastereomer and conventional type of MELs showed very similar 1 H NMR spectra (Fig 9A). However, two resonances arising from H-4a and H-4b in the erythritol moiety were significantly different. In conventional type of MEL the H-4 signals were widely separated (H-4b: 3.98-4.01 ppm; H-4a: 3.80-3.83 ppm), while the signals observed from diastereomer type of MEL overlapped with one another (H-4a and H4-b: 3.87-3.97 ppm). Moreover, the H-1' signal from the mannose anomeric hydrogen was shifted to a lower field in MEL-A produced by recombinant strain ΔPaEMT1 harboring pUXV1_neo:: PtEMT1 (from 4.73 ppm to 4.74 ppm). In 13 C NMR analyses (Fig 9B), we observed a characteristic chemical shift in the C-2 and C-3 signals of the erythritol moiety (from 72.0 ppm to 71.8 ppm and 71.3 ppm to 71.5 ppm, respectively), which corresponded to the C-2 and C-3 signals in diastereomer type of MEL-A [47][48][49]. Based on these observations, we concluded that ΔPaEMT1 harboring pUXV1_neo::PtEMT1 produced the diastereomer type of MEL-A.

Discussion
In this study, we obtained for the first time the gene cluster involved in the diastereomer type of MEL-B biosynthesis in P. tsukubaensis. The gene PtEMT1, which plays a crucial role in determining the sugar conformation of ME, was introduced into ΔPaEMT1. PtEMT1 restored MEL production in ΔPaEMT1 and the product was determined to be diastereomer type of MEL-A. MEL biosynthesis clusters have been previously reported in conventional type of MEL producers such as U. maydis, P. antarctica, P. aphidis and P. hubeiensis [14-17, 36, 37]. Earlier studies assessed the function of Emt1 using gene disruption methods and found that Emt1 is essential for MEL biosynthesis [21,36]. Therefore, we strongly expected Emt1 to contribute to the chirality of MEL-B in P. tsukubaensis. To investigate this hypothesis, we introduced the plasmid pUXV1_neo::PtEMT1 into ΔPaEMT1, which lacks MEL production. NMR analyses revealed that the MEL produced was diastereomer type of MEL-A (Fig 9). We therefore concluded that PtEMT1p plays a critical role in the formation of the sugar moiety in MELs.
Morita et al. [50] reported that P. tsukubaensis JCM16987 synthesizes mannosyl-L-arabitol lipid-B from L-arabitol as a substrate but does not utilize D-arabitol. In contrast, P. parantarctica, a conventional type of MEL-A producer, biosynthesized only mannosyl-D-mannitol lipid from D-arabitol. This suggests that substrate specificity in P. tsukubaensis differs from that of conventional type of MEL producers [50]. Based on Emt1p amino acid sequence alignment (Fig 5), we observed three low-identity regions in the C-terminal half of P. tsukubaensis NBRC1940. We predict that these regions may be related to the substrate specificity of sugar alcohols in P. tsukubaensis. Important avenues for future research, include chimeric enzyme construction, crystal structure analysis, and determination of the active site of PtEMT1p, will provide us with greater insight into the function of PtEMT1p.
To our knowledge, P. tsukubaensis and P. crassa are the only species that biosynthesize diastereomer type of MEL [48,49]. P. crassa biosynthesizes a mixture of diastereomer type of MELs containing MEL-A, MEL-B and MEL-C; however, the genomics of this species has not yet been studied. Therefore, genetic and structural analysis of Emt1p in P. crassa will help to elucidate the catalytic mechanism of sugar conformation.
Various homologs of conventional type of MELs have been reported to date, such as MEL-A, MEL-B, MEL-C and MEL-D. Other MEL homologs, containing D-arabitol, D-mannitol and ribitol instead of erythritol, have also been biosynthesized [51,52]. In previous studies, diastereomer type of MEL-B and MEL-D exhibited higher hydrophilicity and water-holding properties than conventional type of MELs [35]. Based on these findings, it was determined that the sugar conformation of ME affects the function of MELs. Hitherto, MEL-B is the only diastereomer type of MEL capable of commercial-scale production. Diastereomer type of MEL-A and MEL-C are biosynthesized by P. crassa; however, this species produces only small amounts of MELs. While diastereomer type of MEL-D can be obtained in vitro by enzymatic reaction, there is no previously reported means of producing this glycolipid through microbial biosynthesis. In the present study, we demonstrated that diastereomer type of MEL-A can be produced by a conventional type of MEL-A producer modified to express PtEMT1. Further investigation of the PtEMT1p enzyme will facilitate the expansion of structural and functional varieties of MELs through gene engineering methods.
In conclusion, we identified for the first time the gene cluster involved in diastereomer type of MEL-B biosynthesis and demonstrated that PtEMT1p plays a crucial role in sugar conformation. Further elucidation of the MEL biosynthesis gene cluster will provide opportunities for metabolic engineering using this biosynthetic pathway.