Insights into the glycerol transport of Yarrowia lipolytica.

Cellular membranes separate cells from the environment and hence, from molecules essential for their survival. To overcome this hurdle, cells developed specialized transport proteins for the transfer of metabolites across these membranes. Crucial metabolites that need to cross the membrane of each living organism, are the carbon sources. While many organisms prefer glucose as a carbon source, the yeast Yarrowia lipolytica seems to favor glycerol over glucose. The fast growth of Y. lipolytica on glycerol and its flexible metabolism renders this yeast a fascinating organism to study the glycerol metabolism. Based on sequence similarities to the known fungal glycerol transporter ScStl1p and glycerol channel ScFps1p, ten proteins of Y. lipolytica were found that are potentially involved in glycerol uptake. To evaluate, which of these proteins is able to transport glycerol in vivo, a complementation assay with a glycerol transport-deficient strain of Saccharomyces cerevisiae was performed. Six of the ten putative transporters enabled the growth of S. cerevisiae stl1Δ on glycerol and thus, were confirmed as glycerol transporting proteins. Disruption of the transporters in Y. lipolytica abolished its growth on 25 g/L glycerol, but the individual expression of five of the identified glycerol transporters restored growth. Surprisingly, the transporter-disrupted Y. lipolytica strain retained its ability to grow on high glycerol concentrations. This study provides insight into the glycerol uptake of Y. lipolytica at low glycerol concentrations through the characterization of six glycerol transporters and indicates the existence of further mechanisms active at high glycerol concentrations.

INTRODUCTION

Plasma membranes build the interface between the cell and its environment, separating the intracellular metabolism from the extracellular space. To transfer nutrients and end products of the metabolism, to maintain the cellular homeostasis, and to enable intercellular communication, cells have developed specialized transport proteins that mediate the exchange of ions and molecules across this barrier. These plasma membrane transport proteins, together with transporters of intracellular membranes, account for approximately 10% of all cellular proteins (Saier et al., 2020). Currently, the Transporter Classification Database (tcdb.org) lists over 21,000 transport proteins and classifies them into more than 1500 families composed of proteins that share sequence, structural or functional attributes. Although the advent of metagenome sequencing and progress in bioinformatics aided the detection of novel transporters, only a fraction of these proteins has been functionally characterized so far. The transport proteins, that have already been described, feature a vast range of substrate specificities, such as the transport of diverse drugs (The International Transporter Consortium, 2010), carboxylic acids (Casal et al., 2008) or ions (Prakash et al., 2003), to name but a few. Even transport proteins that facilitate the transfer of water or urea through the plasma membrane are known, although these molecules are able to diffuse through lipid bilayers (Engel et al., 1994; You et al., 1993).

Essential molecules that need to cross the membrane of each living organism are the carbon sources. Diverse transport proteins for the import of carbon sources were discovered, whereupon especially transporters for glucose, the most abundant monosaccharide in nature, have been extensively studied in all kingdoms of life (Albers et al., 1999; Buhr & Erni, 1993; Huang & Czech, 2007; Özcan & Johnston, 1999; Sauer et al., 1990). As most organisms prefer glucose as a primary energy source, research mainly focused on the characterization of glucose and other hexose transporters. However, some organisms prefer other substrates over glucose and came to the fore because of their ability to efficiently utilize alternative carbon sources. One of these organisms is the yeast Yarrowia lipolytica.

Y. lipolytica is a dimorphic yeast that utilizes an array of carbon sources, including hydrophobic substrates as well as sugars, to accumulate lipids or to produce sugar alcohols and organic acids (Barth & Gaillardin, 1996). Due to its peculiar metabolism, Y. lipolytica became a yeast model for various processes, such as lipid production and protein secretion, or for studies on dimorphism (Nicaud, 2012). The flexible metabolism of Y. lipolytica, which allows it to adapt to various environmental conditions, and its preference for glycerol over glucose render Y. lipolytica a very interesting yeast for studies on the glycerol metabolism (Egermeier et al., 2017; Lubuta et al., 2019; Papanikolaou et al., 2002). Although many enzymes involved in the glycerol metabolism have already been identified, information on the proteins essential for glycerol uptake in Y. lipolytica is scarce (Rywińska et al., 2013).

The molecular mechanism of glycerol uptake in yeasts has been best studied in Saccharomyces cerevisiae, whereas details about the glycerol transport in other yeasts and fungi are limited. In S. cerevisiae, several proteins were proposed to be involved in the import of glycerol, including ScGup1p, ScGup2p (Holst et al., 2000), and the channel protein ScFps1p (Luyten et al., 1995; Sutherland et al., 1997). The proteins ScGup1p and ScGup2p were later identified as O‐acyltransferases that might influence the uptake of glycerol indirectly, instead of being glycerol uptake proteins themselves (Bosson et al., 2006). The initial assumption that ScFps1p mediates the import, as well as the export of glycerol by facilitated diffusion, was revised too. ScFps1p turned out to control the glycerol export under osmotic stress, rather than the import of glycerol (Oliveira et al., 2003; Tamás et al., 1999). Instead, the glycerol import of S. cerevisiae was shown to rely on the active transport by one single protein, the glycerol/H+‐symporter ScStl1p (Ferreira et al., 2005).

According to the Transporter Classification Database, ScStl1p is a member of the major facilitator superfamily (MFS), which is the largest family of secondary carriers (Pao et al., 1998). The MFS comprises transport proteins for a variety of molecules, including transporters for hexoses and carboxylic acids, some of which have already been characterized in Y. lipolytica (Erian et al., 2020; Guo et al., 2015; Lazar et al., 2017; Y.‐K. Park & Nicaud, 2020). Transporters of the MFS mediate the transport of solutes in response to chemiosmotic ion gradients (Pao et al., 1998), whereupon ScStl1p was shown to be an active symporter of glycerol and H+ (Ferreira et al., 2005). Based on sequence similarities to ScStl1p, glycerol transporters were identified and functionally characterized in several other yeasts, such as Candida albicans (Kayingo et al., 2009), Zygosaccharomyces rouxii (Dušková et al., 2015), Candida glycerinogenes (Ji et al., 2018), Wickerhamomyces anomalus (da Cunha et al., 2019) and Kluyveromyces marxianus (Zhang et al., 2020).

In comparison to the active transport mechanism of ScStl1p, ScFps1p is a glycerol channel of the family of major intrinsic proteins (MIPs) that transports glycerol passively. The MIP family diverts into three subfamilies, the aquaporins, the aquaglyceroporins, and the superaquaporins (J. H. Park & Saier, 1996). These channel proteins span the plasma membrane typically with six transmembrane helices and are selective for small uncharged molecules such as water (aquaporins) or glycerol (aquaglyceroporins). Additional to the export of glycerol, ScFps1p of S. cerevisiae facilitates the uptake of arsenite, antimonite, and the undissociated form of acetic acid (Mollapour & Piper, 2007; Wysocki et al., 2001). Several orthologues to ScFps1p were identified in other yeasts, including Fps1p of Z. rouxii (Dušková et al., 2015), K. marxianus, and Kluyveromyces lactis (Neves et al., 2004). While these ScFps1p orthologues have a function similar to ScFps1p concerning glycerol export and osmotic stress response, several proteins were found that have a similar sequence to ScFps1p but the reverse transport function. The proteins Fps2p of Pachysolen tannophilus and Fps1p of Cyberlindnera jadinii, for example, facilitate the glycerol uptake in S. cerevisiae stl1Δ and enhance the growth of wild‐type S. cerevisiae on glycerol (Klein et al., 2016).

The loss of a single protein (i.e.Stl1p) renders S. cerevisiae unable to grow on glycerol, but it requires the removal of at least seven transport proteins to stop its growth on glucose (Reifenberger et al., 1995; Wieczorke et al., 1999). Likewise, some yeasts that achieve higher growth rates on glycerol than S. cerevisiae possess more than only one glycerol transporter. For instance, two glycerol transporters were identified in C. glycerinogenes (Ji et al., 2018) and four potential glycerol transporters were found in the genome of Komagataella phaffii (Mattanovich et al., 2009). Remarkably, Y. lipolytica has even seven putative active glycerol transporters (Lazar et al., 2017) and two putative glycerol channels (Pettersson et al., 2005). A similarly high level of genetic redundancy is present in Y. lipolytica for the transport of hexoses. Six hexose transporters can be found in Y. lipolytica, three of which are broad‐range hexose transporters mediating the transport of glucose, galactose, mannose, and fructose (Lazar et al., 2017). The loss of two of these broad‐range transporters (YALI0C06424p and YALI0E23287p) drastically reduces growth on fructose, glucose, and mannose.

The expression of the putative glycerol transporters of Y. lipolytica has already been analyzed in several transcriptome and proteome studies. In particular, one orthologous protein to ScFps1p, that is, YALI0F00462p, was shown to be strongly induced by glycerol (Lubuta et al., 2019), whereas some of the ScStl1p‐like proteins were found to be upregulated on xylose (Ryu et al., 2016; Walker et al., 2021) or at low nitrogen concentrations (Hapeta et al., 2020). Although several expression studies are available, only one experimental study with one of the nine proteins has been conducted so far to support the expected glycerol transport function. In the study of Klein et al. (2016), one of the putative glycerol channels (YALI0E05665p) was expressed in a wild‐type strain of S. cerevisiae, which significantly increased its growth rate on glycerol and therefore suggests a role of the channel protein in glycerol import. However, the final proof that this protein is a transport protein is still lacking. Moreover, the role of the other eight putative transport proteins in glycerol uptake is still unclear.

In this study, we provide further evidence for the glycerol transport activity of YALI0E05665p and evaluated the transport function of nine other proteins of Y. lipolytica. To this end, we performed a complementation assay with glycerol‐transporter deficient strains of S. cerevisiae and Y. lipolytica, the latter of which was obtained during this study. In total, six proteins of Y. lipolytica, including YALI0E05665p, were experimentally verified as having glycerol import activity. Moreover, an indication was obtained that further transport proteins or transport mechanisms exist in Y. lipolytica which allow glycerol uptake at high glycerol concentrations.

MATERIALS AND METHODS

Strains and media

Y. lipolytica strain DSM 3286 was obtained from the Leibnitz Institute DSMZ‐German Collection of Microorganisms and Cell Cultures. Precultures were grown in a rich liquid medium containing per liter deionized water: 9 g yeast extract, 18 g soy peptone, and 10 g glycerol (YPG) or 10 g glucose (YPD), adjusted to pH 7.5. For solid media, 20 g/L agar–agar were added to the medium. Shake flask cultivations were performed in media (YNB) containing per liter deionized water: 1.7 g yeast nitrogen base without amino acids (BD Difco, Thermo Fisher Scientific), 6.25 g (NH4)2SO4, 1.99 g K2HPO4, 1.85 g KH2PO4, and either 25 g glycerol (YNBG) or glucose (YNBD). A defined medium was used for all bioreactor cultivations containing per liter deionized water: 100 g glycerol, 3.1 g (NH4)2SO4, 1.0 g KH2PO4, 1.3 g Na2HPO4 × 2H2O, 1.0 g MgSO4 × 7H2O, 0.2 g CaCl2 × 2H2O, 0.5 g citric acid, 21 mg FeCl3, 1 mg Thiamine‐HCl, 0.5 mg H3BO3, 0.06 mg CuSO4 × 5H2O, 0.1 mg KI, 0.45 mg MnSO4 × H2O, 0.71 mg ZnSO4 × 7H2O and 0.23 mg Na2MoO4 × 2H2O. For selection purposes and to maintain plasmids during cultivation, 400 mg/L nourseothricin were added. S. cerevisiae BY4741 stl1Δ was obtained from the EUROSCARF collection under accession number Y05831 and was maintained on YP agar with 20 g/L glucose and 300 mg/L hygromycin B if applicable. Precultures of S. cerevisiae were grown in YPD medium with 300 mg/L hygromycin B. Plates for spot test assays of S. cerevisiae contained per liter deionized water: 20 g agar, 1.7 g YNB, 50 g (NH4)2SO4, 10 g glycerol, 100 mg l‐leucine, 20 mg l‐methionine, and 20 mg l‐histidine.

Plasmid construction

Expression plasmids were created by Golden Gate cloning in Escherichia coli DH10B as described in detail by Egermeier et al. (2019). Plasmid maps of standard vectors used in this study are available at www.addgene.com and can be accessed with their corresponding plasmid number. All coding sequences (CDS) were cured in silico from BsaI and BpiI restriction sites before cloning. CDS of YALI0D05665g (FPS1), YALI0F00462g (FPS2), YALI0D01111g (STL1), YALI0C04730g (STL2), YALI0C16522g (STL3), YALI0B17138g (STL6), YALI0F25553g (STL7), and YALI2C00079g (STL8) were ordered as plasmids or gBlocks from Integrated DNA Technologies (www.idtdna.com) with suitable fusion sites for Golden Gate cloning. The CDS of YALI0F06776g (STL4) and YALI0A08998g (STL5) were amplified by polymerase chain reaction (PCR) from genomic DNA of Y. lipolytica DSM 3286 simultaneously removing potential BsaI/BpiI restriction sites. The plasmids were directly used for cloning and CDS were assembled with the native TEF1 or GPD1 promoter and CYC1 transcription terminator of either Y. lipolytica or S. cerevisiae to expression cassettes. The PCR amplified fragments were first cloned into a plasmid (BB1_L_23_syn_BsaI; Plasmid #89915) without a promoter or terminator, were sequenced (Microsynth Austria GmbH), and subsequently assembled with the appropriate promoters and terminators (BB2_L_AB_syn_BbsI; Plasmid #89917). The expression cassettes were further cloned into expression vectors for Y. lipolytica or S. cerevisiae. The expression vector pMEG_BB3_YL68N_AC (Plasmid #117830) for Y. lipolytica contained an autonomous replication sequence (CEN/ARS68) and the resistance marker for nourseothricin (natMX). The vector BB3_arscen_URA3_Hygro (Plasmid #118077) for S. cerevisiae carried URA3, an hph expression cassettes and an autonomous replication sequence (CEN/ARS6). The final plasmids and empty vectors were used for yeast transformations according to the lithium‐acetate method established by Barth and Gaillardin (1996).

For the generation of sGFP‐fusion proteins, the CDS of FPS1 and sGFP were PCR amplified. Fusion sites for the assembly with a vector were added 5′ to the CDS of FPS1 and 3′ to the CDS of sGFP. At the 3′ end of FPS1, the first part of a glycine‐serine linker (amino acids GGG) and a unique fusion site were added, simultaneously removing the stop codon. The same fusion site and the second part of a glycine‐serine linker (amino acids GS) were added at the 5′ end of sGFP. Both PCR fragments were assembled in a vector BB1_L_23_syn_BsaI (Plasmid #89915) and verified by Sanger Sequencing (Microsynth Austria GmbH). Further cloning was performed as described above with promoter TEF1 and terminator CYC1 of Y. lipolytica.

Genome editing with CRISPR/Cas9

A CRISPR/Cas9 based system was used to disrupt genes by insertion/deletion mutations in the 5′ region of the CDS as previously described by Egermeier et al. (2019). Briefly, a plasmid was created by Golden Gate cloning which expressed a humanized Cas9 and a single guide RNA (gRNA) flanked by a Hammerhead type ribozyme and a hepatitis delta virus ribozyme. Y. lipolytica DSM 3286 was transformed with this plasmid and after an outgrowth phase of 4 days at 30°C, the cells were plated on selection plates. Cells carrying a frameshift mutation were detected by Sanger sequencing (Microsynth Austria GmbH) and were cured from the plasmid by restreaking them 2–3 times on YPG plates. The entire CDS of STL8 was knocked out using the same CRISPR/Cas9 based system with a single vector containing two gRNAs. Protospacers were selected that are directly up‐ and downstream of the CDS. The successful knockout was confirmed by colony PCR and Sanger sequencing, and the strains were cured as described before.

Bioinformatics tools

Standard in silico analysis, for example, Sanger sequencing analysis and DNA and protein sequence alignments, were performed with CLC Main Workbench (Version 8.1). The Y. lipolytica DSM 3286 genome assembly ASM1449061v1 was used as a reference genome. BLAST searches were done with the online protein–protein blast tool of NCBI using the predefined parameters, that is, search in the nonredundant protein sequences database with a BLOSUM62 scoring matrix, restricted to Y. lipolytica (tax id: 4952). Phylogenetic trees were generated with the Constraint‐based Multiple Alignment Tool (COBALT) available at NCBI. Prediction of protein localizations was done with the online tool ProtComp 9.0 (Softberry Inc.). Prediction of transmembrane helices was performed with the online tool TMHMM 2.0 (Krogh et al., 2001). Gene clusters of FPS1‐like genes were identified with the Sequence Similarity DataBase (SSDB, available on kegg.jp) using YALI0F00462g as query (Kanehisa et al., 2004).

Spot test assay

A total of 10 ml YPD was inoculated with a single colony of S. cerevisiae and cells were grown overnight at 30°C and 180 rpm. Cells were harvested, washed once with deionized water and a 10‐fold serial dilution from OD600 1 to 10−4 was prepared. Four microliters of each dilution were spotted on YNBD and YNBG agar and plates were incubated for 7 days at 30°C.

Shake flask cultivations

A total of 10 ml YPD was inoculated with a single colony and incubated overnight on a shaker at 180 rpm and 30°C. Cells were harvested and washed once with sterile deionized water. The cell density was determined with a photometer (Biochrom WPA CO8000 Cell Density Meter) at 600 nm and the cells were used for inoculation of 10 ml YNBG or YNBD medium (in 100 ml shake flasks) with an OD600 of 1. The shake flasks were incubated at 30°C on a rotary shaker at 180 rpm. All cultivations were performed in triplicates and an unpaired t‐test was performed for statistical analysis.

Bioreactor cultivations

All cultivations were carried out in a DASGIP Parallel Bioreactor System (Eppendorf AG) with four parallel bioreactors and a maximum working volume of 1.2 L. pH was monitored with a pH probe (Mettler‐Toledo) and adjusted to pH 5.5 by the automated addition of 5 M NaOH or 1 M H2PO4. The dissolved oxygen concentration was monitored by a VisiFerm DO 120 probe (Hamilton Company) and was controlled at 50% throughout the cultivation by adjusting the gassing rate of pressurized air and the stirrer speed. To reduce foam formation, 5% (wt/vol) Struktol (SB 2121; Schiller+Seilacher GmbH) were added dropwise whenever necessary. All cultivations were performed at 30°C.

For preculture, 100 ml YPD was inoculated with a single colony and incubated overnight on a shaker at 180 rpm and 30°C. Cells were harvested and washed once with sterile deionized water. The cell density was determined with a photometer (Biochrom WPA CO8000 Cell Density Meter) at 600 nm and the cells were used for inoculation of 500 ml cultivation medium with an OD600 of 1. All cultivations were performed in triplicates.

Metabolite and biomass determination

Glycerol concentrations were determined by high‐performance liquid chromatography analysis (Shimadzu) with an Aminex HPX‐87H column (300 × 7.8 mm; Bio‐Rad Laboratories). The column was operated at 60°C with 8 mM H2SO4 as mobile phase and a flow rate of 0.6 ml/min for 25 min. Peaks were detected and quantified with a refraction index detector (RID‐10A; Shimadzu) or with a photodiode array detector at 254 nm (SPD‐M20A; Shimadzu). To determine cell dry mass (CDM) of Y. lipolytica, 2 ml culture broth were centrifuged for 5 min at 10,000 g, the pellet was washed once with deionized water and dried for 48 h at 100°C. CDM was determined in duplicates.

Gene expression analysis

A comprehensive data set of previous bioreactor cultivations of Y. lipolytica DSM 3286 was used for differential gene expression analysis. In these experiments, a 72‐h batch cultivation was performed in bioreactors with 100 g/L glycerol or glucose either at pH 5.5 or pH 3.5. Samples for RNA‐Seq were taken in the early exponential growth phase at pH 5.5 (after 8 h, approx. 2 g/L CDM and 95 g/L residual glycerol) and in the nitrogen‐limited stationary phase at both pH conditions (after 28 h, approx. 14.5 g/L CDM and 18 g/L residual glycerol). A total of 18 samples for strain DSM 3286 was analyzed by Illumina sequencing (Eurofins Austria GmbH). Expression data are provided as log2‐FPKM values for biological triplicates. The full data set, including details on sample treatment and data processing, is available in the GEO database with entry number GSE151659.

Protein localization analysis

Single colonies of strains expressing the sGFP‐fusion proteins were inoculated in 10 ml YPG and were incubated overnight at 30°C and 180 rpm. Cells were harvested, washed once with deionized water and used for inoculation of fresh medium at OD600 of 2. After incubation at 30°C and 180 rpm for approx. 3 h, the cells were analyzed with an Axio Observer Z1 (Carl Zeiss Microscopy GmbH) using an HXP 120 V lightning unit (Carl Zeiss Jena GmbH), filter set 38 HE (Carl Zeiss Microscopy GmbH), and Software ZEN 2.3 pro. Pictures of representative cells were taken in the brightfield, and after excitation/emission at 488/509 nm.

RESULTS AND DISCUSSION

Characterization of glycerol transporters in a loss‐of‐function mutant of S. cerevisiae

In S. cerevisiae, the import of glycerol relies on the glycerol/H+‐symporter Stl1p. A BLAST search for proteins in Y. lipolytica similar to ScStl1p yielded a list of several proteins with e‐values ranging from 10−131 to 10−23. All of these proteins can be classified as transporters of the Sugar Porter Family and have been phylogenetically analyzed by Lazar et al. (2017). The phylogenetic analysis showed that eight of these proteins form a cluster with ScStl1p (Figure 1), which suggests a potential role in glycerol transport (Table 1). Therefore, these eight proteins were selected for analysis of their function. Two further proteins were considered as putative glycerol transporters that have amino acid sequences similar to the glycerol channel Fps2p of P. tannophilus (Table 1). The expression of one of these proteins, termed YlFps1p, enhanced the glycerol uptake of S. cerevisiae in a previous study (Klein et al., 2016). The other gene, termed FPS2, is located directly upstream of the glycerol kinase GUT1, forming a cluster. The same cluster of (experimentally characterized) glycerol facilitators with glycerol kinases can also be found for instance in the genomes of K. marxianus, K. lactis, and Z. rouxii.

Figure 1

Unrooted phylogenetic tree of glycerol transporters. (a) Putative active glycerol transporters of Yarrowia lipolytica are compared to ScStl1p (YDR536W) and known hexose transporters of Y. lipolytica (Yht1p (YALI0C06424p), Yht2p (YALI0C08943p)), Yht3p (YALI0F19184p), Yht4p (YALI0E23287p). (b) Putative glycerol channels of Y. lipolytica are compared to the known glycerol channels CjFps1p (ODV72620.1), KlFps1p (AAQ01788.1), KmFps1p (AAS47031.1), ScFps1p (CAA38096.1), PtFps1p (AFN43530.1), PtFps2p (AFN43531.1), and ZrFps1p (AAQ16650.1)

Table 1

Selected proteins of Yarrowialipolytica from a BLAST search with PtFps2p and ScStl1p as reference proteins​​​​​​

Locus tag			BLAST analysis			Gene expression analysis
DSM 3286	CLIB122	Name	Query	Query cover	E‐value	log2‐FPKM Growth	log2‐FPKM N‐Limit
YALI2F00006	YALI0F00462	YlFps2p	PtFps2p	87%	4E−107	6.91 ± 0.02	3.23 ± 0.24
YALI2E00127	YALI0E05665	YlFps1p	PtFps2p	86%	5E−105	6.99 ± 0.05	7.13 ± 0.06
YALI2D01102	YALI0D01111	YlStl1p	ScStl1p	95%	4E−131	5.90 ± 0.17	0.43 ± 0.07
YALI2C00080	YALI0C04730	YlStl2p	ScStl1p	86%	2E−130	0.04 ± 0.06	0.81 ± 0.19
YALI2C00079	YALI0C04686	YlStl8p	ScStl1p	95%	2E−129	n.a.	n.a.
YALI2B00179	YALI0C16522	YlStl3p	ScStl1p	93%	2E−129	0.09 ± 0.08	4.89 ± 0.01
YALI2E00997	YALI0F25553	YlStl7p	ScStl1p	95%	2E−126	1.44 ± 0.21	3.09 ± 0.40
YALI2C00742	YALI0B17138	YlStl6p	ScStl1p	88%	6E−120	0.24 ± 0.05	2.5 ± 0.19
YALI2A00136	YALI0A08998	YlStl5p	ScStl1p	90%	2E−120	0.19 ± 0.07	3.57 ± 0.07
YALI2F00135	YALI0F06776	YlStl4p	ScStl1p	82%	5E−78	5.29 ± 0.02	7.34 ± 0.06

Note: Log2‐FPKM expression values of a transcriptome analysis from previous cultivations of Y. lipolytica DSM 3286 in the growth phase at pH 5.5 and the nitrogen‐limited production phase at pH 3.5 with glycerol as substrate are given as mean values and standard deviations of biological triplicates. STL4 and FPS1 are the only genes showing high expression profiles in both cultivation phases. STL8 is not represented in the analyzed transcriptome data set and therefore, no gene expression data is available. All query cover and e‐values refer to the proteins of Y. lipolytica DSM 3286. The corresponding locus tags in Y. lipolytica CLIB122 are additionally indicated. Additional information on differential gene expression on glycerol and glucose can be found in Table S1.

Abbreviation: n.a., not available.

A high gene expression of important glycerol transporters was expected to occur in Y. lipolytica when glycerol is present in the medium. Therefore, we compared the transcriptome of Y. lipolytica DSM 3286 cultivated in either glycerol‐ or glucose‐containing medium. However, contrary to the expected upregulation of transporters on glycerol, the expression levels were similar at either condition or even higher on glucose than glycerol. The only gene being upregulated on glycerol compared to glucose is FPS1 during the growth phase of the culture (Table S1).

To evaluate if the putative transporters facilitate import of glycerol, a complementation assay was performed with a knock‐out strain of S. cerevisiae that lacks its native glycerol transporter Stl1p. S. cerevisiae BY4741 stl1Δ was transformed with plasmids carrying an expression cassette of 1 of the 10 putative transporters of Y. lipolytica under control of the endogenous TEF1 promoter and CYC1 terminator or with an empty vector (VC) as control. All strains were spotted on agar plates with 10 g/L glycerol and growth phenotypes were determined.

In comparison to the control strain S. cerevisiae BY4741 stl1Δ, which grew poorly (Figure 2), growth was enhanced by the expression of YALI0C04730g (STL2), YALI0C16522g (STL3), YALI0B17138g (STL6), YALI2C00079g (STL8), YALI0E05665g (FPS1), or YALI0F00462g (FPS2). The restored growth of these strains confirmed a glycerol transport function of YlStl2p, YlStl3p, YlStl6p, YlStl8p, YlFps1p, and YlFps2p. The expression of YlStl3p and YlStl6p improved growth of S. cerevisiae stl1Δ even substantially in comparison to the wild‐type strain expressing an empty vector (Figure 2).

Figure 2

Characterization of putative glycerol transporters of Yarrowia lipolytica in Saccharomyces cerevisiae. Growth of S. cerevisiae BY4741 carrying an empty vector (VC) and S. cerevisiae BY4741 stl1Δ expressing putative glycerol transporters of Y. lipolytica or an empty vector (VC) was determined on plates containing 1% (wt/vol) glycerol. To verify the growth of all strains, a control with 1% (wt/vol) glucose was performed and can be found in Figure S2

The ability to import glycerol via YlFps1p and YlFps2p contrasts with the function of ScFps1p which facilitates predominantly the export of glycerol from cells (Tamás et al., 1999). However, similar to YlFps1p and YlFps2p, proteins Fps1p of C. jadinii (Klein et al., 2016) and Fps2p of P. tannophilus (Liu et al., 2013) were demonstrated to enable glycerol import in S. cerevisiae stl1Δ. In a phylogenetic analysis, CjFps1, PtFps1, PtFps2, YlFps1, and YlFps2 grouped in a branch separate to the nonglycerol importers KlFps1, KmFps1, ZrFps1, and ScFps1 (Figure 1). Comparison of the amino acid sequences of all aforementioned Fps1p‐homologs shows significant differences in the length of the proteins. All known glycerol facilitators that enable import of glycerol in S. cerevisiae stl1Δ (i.e., CjFps1p, PtFps2p, YlFps1p, YlFps2p) consist of 322–385 amino acids, whereas proteins complementing a loss of ScFps1p in S. cerevisiae (i.e., KlFps1p, KmFps1p, ZrFps1p) have 564–658 amino acids. The core of all these proteins is rather conserved, but the glycerol importers lack the long N‐terminal domain present in ScFps1p, which is the target for MAP kinase ScHog1p involved in the closure of the channel in S. cerevisiae (Lee et al., 2013).

More than 70% of the amino acid sequence of YlStl2p, YlStl3p, YlStl6p, and YlStl8p are identical, differing mainly in the N‐ and C‐terminal regions (Figure S2). All four proteins are predicted by the stand‐alone software TMHMM 2.0 to form 12 transmembrane helices, which is typical for the Sugar Porter Family of transporters (Saier, 2000). Although YlStl1p and YlStl5p are very similar to YlStl2p, YlStl3p, and YlStl6p (∼70% identities; Table S2), they did not complement the loss‐of‐function in S. cerevisiae stl1Δ at the tested conditions (Figure 2).

To further characterize the six identified glycerol transporters, their impact on glycerol uptake in Y. lipolytica was determined. Additional to these proteins, we also considered YALI0F06776p (YlStl4p) as an interesting transport protein of Y. lipolytica although it did not complement the loss‐of‐function in S. cerevisiae stl1Δ (Figure 2). The preceding transcription analysis of Y. lipolytica DSM 3286 showed that STL4 is highly transcribed during the growth as well as the nitrogen‐limited production phase (Table 1). In contrast, all other ScStl1p‐like proteins were either weakly expressed (STL7) or downregulated during growth (STL2, STL3, STL5, STL6) or production (STL1). Due to the interesting expression profile of STL4, which is similar to the expression pattern observed for the previously reported glycerol transporter FPS1 by Klein et al. (2016), the encoded protein YlStl4p was also considered as putative glycerol transporter of Y. lipolytica, although the complementation assay in S. cerevisiae did not show such a result for the assumed phenotype. It may be noted at this point that the expression data for STL8 is not depicted in Table 1 because the respective gene is not represented in the available data set used for expression analysis as it is a pseudogene in the reference genome of Y. lipolytica CLIB122.

Disruption of the identified glycerol transporters in Y. lipolytica abolishes growth

Apart from detecting the ability to transport glycerol, it was of great interest to investigate which of the glycerol transporters are essential for the import of glycerol into Y. lipolytica cells. Hence, we disrupted the putative glycerol transporters in Y. lipolytica DSM 3286 and determined the glycerol uptake of the generated strains.

While ScFps1p is a channel protein that enables active diffusion, ScStl1p is actively transporting H+ and glycerol into the cells. Thus, both proteins have fundamentally different transport properties, and we chose to examine both types of proteins separately in Y. lipolytica. To that end, two Y. lipolytica strains were generated that had a disruption of either FPS1 and FPS2, yielding strain YldFPS, or a disruption of STL2, STL3, STL4, STL6, and STL8, yielding strain YldSTL. Both strains and a wild‐type strain were cultivated in shake flasks with 25 g/L glycerol and their growth was determined. Additionally, all strains were cultivated with 25 g/L glucose to exclude a growth defect that is not related to glycerol uptake. Growth and substrate uptake of both strains were comparable to the wild‐type strain in glycerol (Figure 3) and glucose media (data not shown). Thus, loss of these transporters resulted in no phenotypic change of Y. lipolytica under the tested conditions and no vital glycerol import function could be attributed to them.

Figure 3

Shake flask cultivation of Yarrowia lipolytica DSM 3286 and mutant strains with a disruption of FPS1 and FPS2 (YldFPS), or a disruption of STL2, STL3, STL4, STL6, and STL8 (YldSTL) or all of the aforementioned seven genes (YldSF). The strains were cultivated in shake flasks at a starting pH of 6.5 with 25 g/L glycerol and the time‐course of (a) glycerol consumption and (b) growth was monitored. Mean values of triplicate cultivations are shown with standard deviations as error bars

The deletion of transporters may not result in any detectable change of growth as long as at least one other transporter is present that can compensate for the loss. Therefore, we generated a strain with a disruption of all seven genes (FPS1, FPS2, STL2, STL3, STL4, STL6, and STL8), yielding strain YldSF, and determined growth in shake flasks with 25 g/L glycerol or glucose, respectively. Growth of YldSF in the glycerol‐containing medium was not only slower than the wild‐type but was stopped and no glycerol was taken up within 96 h (Figure 3). Growth of YldSF in the glucose‐containing medium was unaffected (data not shown), thus, the growth defect of YldSF on glycerol can be assumed to be the consequence of lacking essential glycerol transporters.

The disruption of either ScStl1p‐like proteins (strain YldSTL) or ScFps1p‐like proteins (strain YldFPS) had no influence on the growth of Y. lipolytica DSM 3286, whereas the disruption of ScStl1p‐like and ScFps1p‐like proteins (strain YldSF) diminished growth. Although both types of proteins rely on different mechanisms for the transport of glycerol across the plasma membrane, Y. lipolytica seemed to be able to compensate the loss of active diffusion with active, H+‐coupled transport and vice versa at the tested conditions. Hence, an effect on growth and glycerol uptake was only detectable upon the simultaneous disruption of both transporter types.

Individual expression of each glycerol transporter restores the growth of YldSF

Disruption of STL2, STL3, STL4, STL6, STL8, FPS1, and FPS2 in Y. lipolytica DSM 3286 resulted in a strain that showed a growth deficiency when using glycerol as the sole carbon source. Hence, this strain was an excellent host to investigate the transport ability of each individual transporter. To this end, fourteen expression vectors were generated, each carrying one transporter‐coding gene under the control of either the endogenous GPD1 or TEF1 promoter of Y. lipolytica, whereof TEF1 has a higher relative expression strength (Blazeck et al., 2011). These constitutive promoters were used for the expression of all genes to ensure comparable results by avoiding unanticipated transcriptional regulations. All expression vectors were used for the transformation of YldSF. Additionally, YldSF and wild‐type Y. lipolytica DSM 3286 were transformed with a vector control, generating strains YldSF_VC and 3286_VC, respectively. The obtained strains were cultivated in shake flasks with 25 g/L glycerol and growth and substrate uptake were determined after 48 h.

The wild‐type derivative strain 3286_VC grew to an OD600 of 38 and consumed the entire glycerol within 48 h, whereas the deletion strain derivative YldSF_VC neither grew nor consumed any glycerol (Figure 4). Strains expressing STL2, STL3, STL4, STL6, or STL8 from the weak GPD1 promoter did not significantly grow either. However, growth and glycerol uptake were restored by the expression of FPS1 or FPS2. Expressing the protein YlFps1p even facilitated the growth of YldSF to OD600 values similar to the wild‐type strain and enabled the complete uptake of glycerol. When the transporters were expressed with a stronger promoter (TEF1), all of them reconstituted the growth of YldSF except for YlStl4p and YlStl8p (Figure 4). YlStl2p, YlStl3p, YlFps1p, and YlFps2p allowed wild‐type‐like growth, whereas the final OD600 of YlStl6p was significantly lower (p < 0.01). To exclude a detrimental effect of the constitutive gene expression on the growth of the recombinant strains, all strains were additionally cultivated in a medium with 25 g/L glucose. The strains grew synchronously and reached an OD600 of ∼39–43, showing no growth defect.

Figure 4

Shake flask cultivation of YldSF expressing putative glycerol transporters from different promoters. Yarrowia lipolytica DSM 3286 with a disruption of STL2, STL3, STL4, STL6, STL8, FPS1, and FPS2 (YldSF) expressing transporter genes from either GPD1 (Panels A and C) or TEF1 promoter (Panels B and D) were cultivated with 25 g/L glycerol for 48 h. Y. lipolytica DSM 3286 and YldSF carrying empty vectors (3286_VC, YldSF_VC) were cultivated as control strains. OD600 values (Panels A and B) and consumed glycerol concentrations (Panels C and D) are displayed as mean values of triplicate cultivations with standard deviations as error bars. Asterisks indicate a significant difference (p < 0.01) to YldSF_VC

Concluding, even a weak expression of either YlFps1p or YlFps2p compensated the lack of five other glycerol transporters, whereas ScStl1‐like proteins required a stronger expression to restore the growth of YldSF. YlStl4p, which was selected due to its interesting expression profile, was not verified as a glycerol transporter and may have another, not yet determined function. A recent proteome study emphasizes a role of YlStl4p in the transport of xylose (Walker et al., 2021). Surprisingly, YlStl8p did not enable the growth of YldSF on glycerol although it restored growth of S. cerevisiae stl1Δ in the preceding complementation assay (Figure 2). This observation might hint to a posttranscriptional regulation or posttranslational modification, which obstructs the glycerol uptake function in Y. lipolytica.

YlFps1p is localized in the plasma membrane

Proteins that import substrates into the cell are components of the plasma membrane, linking the cytoplasm to the extracellular environment. Hence, the identified glycerol transporters were expected to be localized in the plasma membrane, which was supported by bioinformatical predictions with very high scores of 9.9–10 using the stand‐alone software ProtComp.

Especially the transporter YlFps1p appeared in the shake flask experiments as a potent glycerol facilitator which enabled wild‐type‐like growth and glycerol uptake already upon weak expression. Thus, YlFps1p was chosen for a more detailed characterization and the experimental verification of its subcellular localization. For this purpose, YlFps1p was fused with a C‐terminal superfolder GFP (sGFP) tag and was expressed in Y. lipolytica DSM 3286 from the constitutive TEF1‐promoter. As a control, Y. lipolytica DSM 3286 was transformed with a plasmid carrying an untargeted sGFP. Both strains were cultivated in glycerol‐containing medium and were analyzed under a fluorescence microscope.

Fluorescence microscopy of Y. lipolytica DSM 3286 expressing the fusion protein showed that YlFps1p is localized in the plasma membrane of Y. lipolytica (Figure 5b), whereas sGFP was not visibly targeted to the plasma membrane (Figure 5a). A faint fluorescence signal inside the cells expressing the fusion protein was detected as well, which might be an artifact of strong expression of YlFps1p‐sGFP or due to a distorted structure of the fusion protein. However, the strong fluorescence of the plasma membrane clearly supports the bioinformatic prediction that YlFps1p is integrated into the plasma membrane of Y. lipolytica. We verified that the fusion protein is functional, as it complements the deletion phenotype of YldSF (not shown). The absence or presence of glycerol had no impact on localization (not shown).

Figure 5

Subcellular localization of YlFps1p. Exponentially growing cells of (a) Yarrowia lipolytica DSM 3286 expressing untargeted sGFP and (b) Y. lipolytica expressing YlFps1p with a C‐terminal sGFP‐tag were analyzed with a fluorescence microscope. Pictures of representative cells were taken from the brightfield (left) and after excitation/emission at 488/509 nm (right)

Transport mechanism of glycerol at high concentrations remains unknown

Although YlStl2p, YlStl3p, YlStl6p, YlFps1p, and YlFps2p were confirmed to transport glycerol, an influence on the uptake of glycerol had not been observed before the gene disruption of all five proteins (plus YlStl4p and YlStl8p). So far, all experiments were carried out with relatively low glycerol concentrations of 25 g/L. Therefore, it was of interest to analyze if the strains YldSTL and YldFPS were still able to compensate the loss of two or three transporters, respectively, at higher glycerol concentrations. Both strains, as well as YldSF and wild‐type Y. lipolytica DSM 3286, were cultivated in medium containing 100 g/L glycerol. The cultivations were performed under controlled conditions in bioreactors to ensure sufficient oxygen supply, which is essential at high glycerol concentrations.

In accordance with the results obtained from shake flask cultivations with 25 g/L glycerol (Figure 3), the glycerol uptake and growth of strain YldSTL was wild‐type‐like (Figure 6). Although growth of YldFPS was also wild‐type‐like, the uptake of glycerol differed. Glycerol concentrations, and hence, the glycerol uptake, of both strains remained similar for 26 h but started to diverge afterward (Figure 6). While glycerol was already depleted after 48 h by the wild‐type, 10.0 g/L glycerol were still present in the cultivations of YldFPS at the same time point. The cells required another 20 h to consume the residual glycerol. The difference in glycerol uptake of YldFPS between the cultivation in shake flasks and bioreactors might be due to the inherently different cultivation conditions. While the oxygen supply in shake flasks is limited and the pH fluctuates, both parameters were controlled in the bioreactor to ensure the best growth conditions. Hence, the regulation of gene expression might have changed due to the different conditions and altered glycerol uptake of YldFPS. Moreover, the better oxygenation in bioreactors allows high glycerol conversion rates, which might only be sufficiently supported by the glycerol uptake rates of ScFps1p‐like transporters but not ScStl1p‐like transporters at low concentrations. The slower uptake of strain YldFPS but not YldSTL re‐emphasizes the good transport capability of YlFps1p and YlFps2p, which became evident even at a weak expression in YldSF in the preceding shake flask experiment (Figure 4).

Figure 6

Bioreactor cultivation of Yarrowia lipolytica DSM 3286 wild‐type and mutant strains with 100 g/L glycerol. Y. lipolytica DSM 3286 with a disruption of FPS1 and FPS2 (YldFPS), STL2, STL3, STL4, STL6, and STL8 (YldSTL) or STL2, STL3, STL4, STL6, STL8, FPS1, and FPS2 (YldSF) were cultivated in minimal medium with 100 g/L glycerol at pH 5.5. The time course of (a) glycerol uptake and (b) biomass production are shown of cultivations in triplicate cultivations with standard deviations as error bars. The carbon balances close at 101 ± 3% and can be found in detail in Table S3. CDM, cell dry mass

Remarkably, strain YldSF was able to grow in the bioreactor cultivation (Figure 6), which is in stark contrast to the previous shake flask cultivations (Figure 3). After a long lag phase, the cells started to grow and reached the same maximum biomass concentration as the wild‐type. However, YldSF was not able to completely metabolize glycerol within 97 h, which is double the time the wild‐type required to take up the entire amount. The glycerol uptake rates of the wild‐type peaked between glycerol concentrations of 40–70 g/L with an average rate of 0.32 g/gCDM/h. In comparison, the uptake rate of YldSF in the same glycerol range was only 0.13 g/gCDM/h and decreased to less than 0.03 g/gCDM/h below 36 g/L glycerol. This very low uptake rate at low glycerol concentrations corresponds to the results of the shake flask experiments, where no uptake was detected at a concentration of 25 g/L and an OD600 1 (equal to approximately 0.25 gCDM/L).

During the bioreactor cultivation, wild‐type Y. lipolytica DSM 3286 secreted citrate, ketoglutarate, mannitol, arabitol, and erythritol. The same metabolites were also produced by YldSF, but the accumulation of erythritol was reduced significantly compared to the wild‐type (Figures 7 and S3). Erythritol concentrations of the wild‐type, YldFPS, and YldSTL peaked between 3.6 and 5.4 g/L but did not exceed 0.28 g/L during the cultivation of YldSF (Figure 7). Erythritol is typically accumulated by Y. lipolytica DSM 3286 until glycerol is depleted and is consumed afterward. In strain YldSF erythritol accumulated only slightly until the glycerol concentration was between 70 and 80 g/L and was subsequently taken up. An early reconsumption of erythritol was also detected in strain YldFPS, which started to consume erythritol after 26 h despite the availability of 40.8 g/L glycerol (Figures 6a and 7). To evaluate, if the identified glycerol transporters of Y. lipolytica are also involved in the transport of erythritol, Y. lipolytica DSM 3286 and YldSF were cultivated in shake flasks with glucose‐ or erythritol‐containing medium. When grown on 100 g/L glucose, both strains produced similar amounts of erythritol (approx. 1.25 g/L) within 96 h. When grown on 12.5 g/L erythritol, both strains grew to an OD600 of 25 and consumed the entire erythritol within 24 h. Concluding, erythritol appears to be better accessible than glycerol at high glycerol concentrations upon the loss of important glycerol importers, and none of the transporters seems to be crucial for erythritol import or export.

Figure 7

Erythritol production of Yarrowia lipolytica DSM 3286 and mutant strains. Y. lipolytica DSM 3286 with a disruption of FPS1 and FPS2 (YldFPS), STL2, STL3, STL4, STL6, and STL8 (YldSTL) or STL2, STL3, STL4, STL6, STL8, FPS1, and FPS2 (YldSF) were cultivated in bioreactors with 100 g/L glycerol at pH 5.5. The time course of erythritol production is shown of cultivations in triplicate cultivations with standard deviations as error bars

The ability of YldSF to import glycerol indicates the presence of other proteins that can transport glycerol into the cells additional to the five identified transporters. These additional transporters might facilitate glycerol import only at high concentrations due to a low affinity for glycerol or due to transcriptional repression or posttranslational regulation at low concentrations. The uptake of substrates is an intricately regulated process and is best described in yeasts for the hexose uptake of S. cerevisiae (Bisson et al., 2016; Özcan & Johnston, 1999). Several sensor proteins and transcription factors coordinate more than 20 hexose transporters to regulate hexose uptake of S. cerevisiae at different substrate concentrations. For instance, one of the regulated hexose transporters, ScHxt1p, is a low‐affinity glucose transporter whose expression increases linearly with glucose concentration and reaches maximal expression not before a glucose concentration of 40 g/L (Özcan & Johnston, 1995). Similar regulatory mechanisms might be present in Y. lipolytica that influence yet unidentified transporters and enable the glycerol uptake at high concentrations. The proteins YlStl1p, YlStl5p, and YlStl7p might also be reconsidered as glycerol transporters although they did not complement the loss of function in S. cerevisiae stl1Δ (Figure 2). Eventually, glycerol transport did not occur at the tested conditions in the heterologous host but might occur in the native host Y. lipolytica.

Moreover, passive diffusion of glycerol into the cells might play a role at high glycerol concentrations. Glycerol is generally recognized as a molecule that can pass the lipid bilayer of the plasma membrane by passive diffusion along a concentration gradient (Gancedo et al., 1968; Heredia et al., 1968; Romano, 1986). This concept of passive diffusion of glycerol into yeast cells was however questioned by Oliveira et al. (2003), who hypothesized that any detectable passive diffusion of glycerol into S. cerevisiae is in fact facilitated by ScFps1p, eventually together with other less specific proteins. Nevertheless, the same authors referred to the dynamic nature of plasma membranes and stated that passive diffusion of glycerol into cells cannot be completely excluded. The composition of plasma membranes is known to be highly dynamic and flexible and is influenced by environmental stresses (Beney & Gervais, 2001). It has been shown that the deletion of membrane proteins, such as glycerol channels GlpF of E. coli or Fps1p of S. cerevisiae, influences the lipid composition of the plasma membrane and thereby alters its permeability (Toh et al., 2001; Truniger & Boos, 1993). Hence, passive diffusion of glycerol into YldSF might occur at high glycerol concentrations and might be accompanied by changes in the lipid bilayer composition of the plasma membrane, reflected in the long lag phase of YldSF.

In conclusion, out of 10 putative glycerol transporters, five were experimentally confirmed to have glycerol transport activity in Y. lipolytica (YALI0B17138p, YALI0C04730p, YALI0C16522p, YALI0E05665p, YALI0F00462p). Each of these five proteins and YALI2C00079p complemented a loss of the sole glycerol importer Stl1p in S. cerevisiae and reconstituted the yeast's ability to take up glycerol. Although the disruption of the identified transporters in Y. lipolytica abolished growth in medium containing 25 g/L glycerol, it did not impair growth at higher concentrations. These results provide an excellent starting point for further studies on the intriguing glycerol transport mechanism and glycerol metabolism of efficiently growing yeast strains.

CONFLICTS OF INTEREST

The authors declare that there are no conflicts of interest.