Identification and application of keto acids transporters in Yarrowia lipolytica

Production of organic acids by microorganisms is of great importance for obtaining building-block chemicals from sustainable biomass. Extracellular accumulation of organic acids involved a series of transporters, which play important roles in the accumulation of specific organic acid while lack of systematic demonstration in eukaryotic microorganisms. To circumvent accumulation of by-product, efforts have being orchestrated to carboxylate transport mechanism for potential clue in Yarrowia lipolytica WSH-Z06. Six endogenous putative transporter genes, YALI0B19470g, YALI0C15488g, YALI0C21406g, YALI0D24607g, YALI0D20108g and YALI0E32901g, were identified. Transport characteristics and substrate specificities were further investigated using a carboxylate-transport-deficient Saccharomyces cerevisiae strain. These transporters were expressed in Y. lipolytica WSH-Z06 to assess their roles in regulating extracellular keto acids accumulation. In a Y. lipolytica T1 line over expressing YALI0B19470g, α-ketoglutarate accumulated to 46.7 g·L−1, whereas the concentration of pyruvate decreased to 12.3 g·L−1. Systematic identification of these keto acids transporters would provide clues to further improve the accumulation of specific organic acids with higher efficiency in eukaryotic microorganisms.

L ight have been shed on economic feasibility of production of organic acids by microorganisms, expecially considering the finiteness of fossil raw materials 1 . Although organic acids occupy a key position among the building-block chemicals, economic production by microorganisms is still a big chagllenge to replace petroleum-derived commodity chemicals 2 . In particular, bio-based production of organic acids is unharnessed by low titres, low yield and accumution of by-products [3][4][5] .
The metabolism and accumulation of a-ketoglutaric acid (a-KG) are subjected to a higher level of regulation than other organic acids involved in central carbon metabolism 6 , as a-KG occupies a key position in both carbon central metabolism 7 and the regulation of the carbon-nitrogen balance 8 . Metabolic strategies concerning the following have been investigated for a-KG production: regulation of key enzymes, including pyruvate dehydrogenase complex 9 , pyruvate carboxylase 10 , fumarase 3 , aconitase 11 , isocitrate lyase 12 , isocitrate dehydrogenase 13 and components of the a-ketoglutarate dehydrogenase complex 14 ; and co-factor engineering of acetyl-CoA biosynthesis and regeneration 15 .
Extracellular accumulation of non-target carboxylates is a common problem for the production of carboxylates 16,17 . Manipulation of transporters has been an efficient tool to improve the productivity for target carboxylates 5 . A successful metabolic engineering approach for the over-synthesis of organic acids also requires incorporation of an appropriate exporter to increase productivity 18 . A comparative study revealed that the highest malate yield was obtained once the malate transporter was recruited in the mutants 19 . Lactate production in S. cerevisiae could be significantly improved with the combined expression of lactate dehydrogenase and the lactate transporter 20 . Although many studies have described the mechanisms of action and regulation for carboxylate transport in yeast 20,21 , Y. lipolytica orthologous gene of the carboxylate transporter have not yet been identified.
In this study, a number of carboxylate transporters were identified from Y. lipolytica WSH-Z06, which exhibits flexibility in substrate specificity. The duplication of endogenous transporters might provide a powerful tool to ensure efficient carboxylate synthesis and maintain homeostasis of the intracellular environment. In this study, we observed that a competitive dual effect -a significant increase in a-KG production with a sharp decrease in pyruvate (PA) accumulation resulted from overexpression of YALI0B19470g. This result suggests a new and promising strategy for gene manipulation that can efficiently address a-KG accumulation. The identification of these transporters also have uncovered the mechanisms of extracellular accumulation of diverse organic acids in genome level, and provides new clues to orchestrate competition of extracellular accumulation between target and non-target carboxylates in yeast and other eukaryotic microorganisms.

Results
Bioinformatics analysis of potential carboxylate transporters in Y. lipolytica. To screen putative carboxylate transporters, 6611 proteins encoded by Y. lipolytica CLIB122 genome were obtained from the UniProt database. Of these, 1104 proteins (Supplementary S1) were predicted as transmembrane proteins by TMHMM v.2.0 (Fig. 1A). Subsequently, 117 proteins were excluded from this set, due to the presence of a possible signal peptide at the N terminus, as predicted by SignalP 4.1. A sequence search using a conserved carboxylate transporter signature of JEN family, NXXS/THXS/TQDXXXT, identified six putative proteins encoded by the YALI0B19470g, YALI0C15488g, YALI0C21406g, YALI0D24607g, YALI0D20108g, and YALI0E32901g genes, which exhibited high similarity to the signature sequence. A multiple sequence alignment of these sequences using homologous carboxylate transporter sequences from other fungi confirmed the presence of the conserved sequence (Fig. 1B).
In addition, a high level of sequence similarity among these putative transporters and the characterized carboxylate transporters was illustrated by two BLAST (http://blast.ncbi.nlm.nih.gov/) protocols, in which ScJen1p and KlJen2p were used as query sequences [22][23][24] . In these cases, the identity to ScJen1p and to KlJen2p were prospected, respectively (Fig. 2). In silico analysis of these protein sequences using the TMRPres2D tool revealed that the conserved sequence was located in the predicted transmembrane helices (Fig. 3). Hence, these transmembrane proteins were suspected to be potential carboxylate transporters.
Exogenous keto acid treatment. To uncover the response of carboxylate transporter candidates for uptake of exogenous keto acids, glucose-grown (repression condition) Y. lipolytica WSH-Z06 cells were transferred at the exponential phase to YPK or YPP medium (A large amount exogenous organic acids were accumulated in yeast cells, Supplementary S2). Initially, neither a-KG nor PA was detected in these cultures. During the first hour, the intracellular concentration of both carboxylates increased quickly to maximums of 3.90 mmol?(mg?DCW) 21 for a-KG and 1.67 mmol?(mg?DCW) 21 for PA in a-KGand PA-treated cells, respectively. After incubation in YPK or YPP medium for 1 h, the concentration of a-KG gradually decreased 1 h to 3 h, until the minimum of 0.85 mmol?(mg?DCW) 21 was reached. During this period, the content of PA also decreased to 0.23 mmol?(mg?DCW) 21 (Fig. 4A).
Assessment of putative genes using heterologous expression in S. cerevisiae. To assess their possible roles in carboxylate transport, null mutants of endogenous transporters were constructed by gene disruption in S. cerevisiae CEN.PK2-1D. The ScJEN1 deletion mutant displayed reduced growth on lactate, acetate, PA, malate, The results showed that 28 orthologs of the JEN family from Aspergillus oryzae, Aspergillus fumigatus, Aspergillus nidulans, Y. lipolytica, Kluyveromyces lactis, Candida albicans, Kluyveromyces thermotolerans, S. cerevisiae, Debaryomyces hansenii and Neurospora crassa had highly conserved sequences. The name of these orthologs was based on protein ID in NCBI database, as one potential JEN1 from N. crassa could not find in the database, it was named as Q9P732 based on protein ID from Uniprot.    and a-KG compared to the parental strain. To completely abolish the uptake by endogenous transporters, the Ady2p was also disrupted in W1 strain. In contrast, the W2 strain, in which both ScJEN1 and ScADY2 were disrupted, cannot grow on lactate, acetate, PA, malate, and a-KG containing media. Moreover, the parental strain as well as the W1 and W2 strains did not grow on citrate (Fig. 5). All putative transporter genes were introduced into the W2 strain by genetic transformation based on Dhis3 complementation. The resulting strains grew well on all sources including citrate, supporting the conclusion that these genes encode carboxylate transporters (Fig. 5).
Substrate specificity assay. To investigate the substrate specificity of these proteins, glucose-grown S. cerevisiae CEN.PK2-1D and mutants were incubated in YPA, YPL, YPP, YPM, YPK, and YPC medium for 2 h after exhaustion of the endogenous carbon source in saline water. The intracellular carboxylate content in S. cerevisiae CEN.PK2-1D cells was measured (   The transport of a-KG was restored in all heterogeneous transporter-containing strains, with the highest accumulation (1.41 mmol? (mg?DCW) 21 ) occurring in the W4 strain. Among the set, only the W4 could accumulate PA intracellularly (0.17 mmol?(mg?DCW) 21 ). Based on these observations, we conclude that YALI0B19470g and YALI0D24607g encode proteins that transport dicarboxylates and tricarboxylates. In addition to PA, W4 cells (containing YALI0C15488p) also accumulated acetate, malate, a-KG, and citrate, indicating the corresponding protein also transported these carboxylates. In a similar manner, we conclude the proteins encoded by the corresponding genes transport the following carboxylates: YALI0C21406p for lactate, malate, a-KG, and citrate; YALI0D20108p for lactate, a-KG and citrate, and YALI0E32901p for acetate, malate, and citrate.
Copy number analysis of putative transporter genes in recombinant strains. In order to estimate copy number of each expression cassette that integrated in genome of recombinant strains, quantitative-PCR (qPCR) analysis was carried out. ACT1 was utilized as an endogenous control 26 . With comparison to ACT1, no obvious relative changes of these transporters were observed (Fig. 6). This indicated that these transporter genes were presented in a single copy in parental strain, respectively. After individual transportation of each endogenous transporters gene into cells of parental strain, relative changes of 2.54 6 0.22-fold for YALI0B19470g, 2.33 6 0.27-fold for YALI0C15488g, 2.62 6 0.33fold for YALI0C21406g, 2.28 6 0.19-fold for YALI0D24607g, 2.39 6 0.26-fold for YALI0D20108g and 2.55 6 0.32-fold for YALI0E32901g were observed, which indicated one more copy of each transporter gene existed per genome in corresponding recombinant strain (Fig. 6).
The effects of transporter genes on carboxylate accumulation in Y. lipolytica. As previously reported carboxylate transporter possess bifunctions for carboxylate influx and efflux 25 . To clarify the roles of these transporters on carboxylate extracellular accumulation, they were overexpressed in Y. lipolytica WSH-Z06. The hp4d promoter was used to increase the transcription of these genes, and YALI0B19470g, YALI0C15488g, YALI0C21406g, YALI0D24607g, YALI0D20108g and YALI0E32901g expression was improved by 23.34 6 2.67-fold, 8.53 6 0.90-fold, 9.32 6 0.82-fold, 11.79 6 1.32-fold, 3.37 6 0.49-fold and 10.50 6 0.97-fold, respectively (Fig. 7). Due to the flexible substrate specificity of these transporters, the accumulation of PA and a-KG increased in T5 cells, and the ratio of extracellular a-KG/PA decreased from 2.06 to 1.87 compared to the wild-type strain. This ratio also decreased in YALI0C15488g, YALI0C21406g, YALI0D24607g, and YALI0E32901g overexpressing strains, as only extracellular PA increased for T2, T3, T4, and T6. A competitive dual effect was observed for strain T1: the transport of a-KG increased dramatically, whereas the concentration of PA dropped by 30.6%, resulting in an increase in the ratio of extracellular a-KG/PA from 2.06 to 3.79 (Fig. 8).
The intracellular carboxylate content, C in (mmol?(mg?DCW) 21 ), was also determined ( Table 2). The intracellular accumulation of a-KG decreased from 0.026 6 0.005 mmol?(mg?DCW) 21   S. cerevisiae, CEN.PK2-1D Djen1Dady2  208.56 for PA were observed for the T1 strain. Based on these observations, overexpression of YALI0A9470g was considered the best strategy to enhance a-KG transport and reduce PA accumulation.

Discussion
The object of the current work was to screen and identify carboxylate transporters, then determine whether the identified proteins regulate accumulation of non-target carboxylates. The duplication of transporters and the flexible substrate specificity demonstrated by the identified carboxylate transporters facilitated extracellular accumulation of carboxylates. Moreover, knowledge of the examined carboxylate transport mechanism is a prerequisite for improving carboxylate synthesis through metabolic engineering. The results provide new insights for regulating extracellular carboxylate accumulation in similar eukaryotic microorganisms.
To circumvent the issue of PA accumulation, previous studies have focused mainly on the regulation of intrinsic forces that redistribute carbon flux from other intermediates to a-KG production. Expression of PA carboxylase 10 , malate dehydrogenase, and fumarase 3 dramatically decreased the accumulation of PA. A strategy to regulate co-factor regeneration resulted in remarkable reduction of extracellular PA 15 . However, as PA has a pivotal role in the regulation of carbon metabolism 20 , these modifications could not entirely overcome PA accumulation. As accumulation of carboxylate is believed to be a yeast defense response to severe environmental conditions 27 ,   regulation of carboxylate transportation process might be another potent route for enhancement of a-KG production.
The carboxylate transport process of yeast is an intensively investigated field 20 . The S. cerevisiae ScJEN1 and ScADY2 genes were identified as key carboxylate transporters 28,29 . The duplication of transporters has been strongly implicated in the utilization of organic acids as a carbon source 24 . In Kluyveromyces lactis, the presence of two carboxylate transporters, Jen1p and Jen2p, guaranteed efficient uptake of lactic acid as a substrate from a lactic-acid-producing habitat 30 . As the cells of Y. lipolytica harbored a powerful potential to use a wide range of substrates as a sources of carbon and energy 31 , these results indicated that powerful carboxylate transporters that maintain intracellular environment homeostasis. A previous study reported that reduction of by-product resulted in enhanced synthesis of target carboxylate by de-repression of the feedback inhibition 3 . It was speculated that the enhanced synthesis of a-KG could be achieved through derepression of the feedback inhibition by efflux of intermediates.
In Y. lipolytica, efficient carboxylate transport was achieved by the duplication of iso-functional transporters. Evolution analysis and motif identification confirmed that a precursor form of Jen1p, preJen1p, arose from the duplication of an ancestral Jen2p 32 . In S. cerevisiae, the transport capacity and substrate affinity of Jen1p were determined by the conserved NXXS/THXS/TQDXXXT sequence 33 . Presence of this signature sequence also determined the flexibility of substrate specificity for these transporters. Previously, Jen1p was induced by lactate, PA, and propionate, whereas Ady2p and Jen2p were induced by acetate 29,30 . Expression of transporters from Y. lipolytica displayed different responses to exogenous carbon source, and single carboxylate induced multiple transporters.
The roles of these carboxylate transporters were assayed via double their copy numbers in the genomic DNA of a-KG producer. One more copy of each endogenous carboxylate transporter was observed (Fig. 6). Our observation that overexpression of carboxylate transporters resulted in enhanced accumulation of extracellular carboxylates (Fig. 7). Previously, uncovered the mechanism for malate efflux was mediated by monoanionic malate concentration gradient, in which the proton symport was major force 25 . Similar to the observations that production of carboxylate was benefited from overexpression of carboxylate transporter 13,34 , the efflux of carboxylate accompanied by symport of proton was speculated 25 . Based on measurement of total content of mixture of monoanionic, dianionic and undissociated form of the carboxylates, the intracellular content of carboxylates was not correlated with extracellular carboxylate content. While, the reported carboxylate transporter was specific for monoanionic form of carboxylate for efflux, the efflux specificity of this series of transporter would be the key to the contradictory observation in future studies.
The 6611 putative proteins encoded by Y. lipolytica CLIB122 genomic DNA were obtained from UniProt (http://www.uniprot.org/). A genome-wide analysis using a transmembrane-helix sequence was performed following a method described previously 38,39 . For each protein, the transmembrane protein topology was predicted using TMHMM (http://www.cbs.dtu.dk/services/TMHMM/). The predicted result was visualized using the TMRPres2D tool 40 . The following values were used to discriminate helical proteins from other proteins 41 : (i) the number of predicted transmembrane helices; (ii) the expected number of residues in the transmembrane helices; and (iii) the expected number of transmembrane helices. To avoid false prediction, we also analyzed all proteins with putative transmembrane helices at the N terminus with SignalP (http://www.cbs.dtu.dk/services/SignalP/) to predict whether the sequence encoded a signal peptide 42 . All screened transmembrane proteins were used for sequence-similarity searches using BLAST (http://blast.ncbi.nlm.nih.gov/) to identify orthologs that have been characterized as carboxylate transporters.
Keto acid treatment. The cells of Y. lipolytica WSH-Z06 were streaked onto YPD slants from glycerol stocks. Cells grown in glucose were harvested at the exponential phase. After a 2 h treatment in saline water, cells were transferred to 100 mL YPK or 100 mL YPP and incubated at 28uC. PA-or a-KG-treated cells were collected at regular time intervals for quantitative real-time PCR (qRT-PCR) analysis and determination of intracellular carboxylate content. YPA, YPL, YPP, YPM, YPK, and YPC media, supplemented as necessary with 50 mg?mL 21 uracil, leucine, tryptophan, and histidine, were used to assay carboxylate transportation of S. cerevisiae CEN.PK2-1D cells and derivatives.
Yeast cells were harvested by centrifugation at 10,000 3 g for 10 min, and washed twice with cold distilled water. The dry cell weight was determined according to protocol described previously 37 . The intracellular carboxylate content was determined according to the method previously, with little modification 43 . Harvested cells were stored in liquid nitrogen until extraction of intracellular organic acid which was followed by cell disruption of according to a protocol described previously 44 . Supernatants were used for determination of intracellular carboxylate concentrations by high-performance liquid chromatography (HPLC). The intracellular concentration of carboxylate was expressed in mmol?(mg?DCW) 21 .
Quantitative real-time PCR analysis. The cells of Y. lipolytica WSH-Z06 were harvested, centrifuged at 10,000 3 g for 10 min, and immediately frozen in liquid nitrogen until RNA extraction. Total RNA was extracted using Trizol reagent (Life Technologies, Carlsbad, CA), according to the manufacturer's instructions. cDNA was synthesized from 5 mg total RNA using the PrimeScript RT Reagent Kit Perfect Real Time (Takara, Dalian, China). qRT-PCR was performed with the synthesized cDNA and primers listed in Table 4 using the SYBR Premix Ex Taq TM Kit (Taraka, Dalian, China) and a LightCycler 480 II Real-time PCR instrument (Roche Applied Science, Mannheim, Germany). All experiments were performed in triplicate and mean values were used for further calculations. Fold changes were determined by the 2 -DDC T method and normalized to the ACT1 gene 45 .
Copy number analysis. In order to determine the copy number of the integrative expression cassettes in the recombinant strains, a qPCR analysis was performed on the genomic DNA template using ACT1 as the internal control 26 . Genomic DNA from parental strain (WSH-Z06) and six recombinant strains (T1, T2, T3, T4, T5 and T6) were isolated after disruption of yeast cells with glass beads (Sigma-Aldrich, St.Louis, MI) by FastPrep 24 (MP Biomedicals, Santa Ana, CA). qPCR was performed with the 5 ng genomic DNA and primers listed in Table 4 using the SYBR Premix Ex Taq TM Kit (Taraka, Dalian, China) and a LightCycler 480 II Real-time PCR instrument (Roche Applied Science, Mannheim, Germany). All experiments were performed in triplicate and mean values were used for further calculations. Fold changes were determined by the 2 -DDC T method and normalized to the ACT1 gene 45 .
Shake flask culture. Shake flask culture was performed in 500 mL flasks containing 50 mL fermentation medium following the protocol stated previously 35 . A yeast seed culture was inoculated from an agar slant and incubated in a 500 mL flask containing 50 mL medium for 18 h on a rotary shaker at 28uC. The culture was used to inoculate 500 mL flasks containing 50 mL fermentation medium. An inoculum volume of 10% (v/v) was used for a-KG accumulation assay. Flask cultures were incubated in a shaker at 200 r?min 21 for 144 h at 28uC.
HPLC analysis. Samples taken from shake flask culture were centrifuged at 10,000 3 g for 10 min. The supernatant was diluted 50 times and filtered through a membrane (pore size 5 0.22 mm). a-KG, pyruvic acid, acetate, lactate, malate, and citrate present in the supernatant were simultaneously determined by HPLC (Agilent 1200 series, Santa Clara, CA, USA) with an Aminex HPX-87H column (300 mm 3 7.8 mm; Bio-Rad Laboratories Inc., Hercules, CA, USA). The mobile phase was 5 mmol L 21 sulfuric acid in distilled, de-ionized water filtered to 0.22 mm. The mobile phase flow rate was 0.6 mL min 21 . The column temperature was maintained at 35uC, and the injection volume was 10 mL. a-KG, pyruvic acid, acetate, lactate, malate, and citrate were detected with a UV detector (wavelength at 210 nm) 37 . To determine the intracellular carboxylates, cells taken from shake flask culture were disrupted and lysates were centrifuged at 10,000 3 g for 10 min. The carboxylate content in supernatant was determined by HPLC followed the protocol above.