Growth and Autolysis of Kluyveromyces Marxianus Stimulated with Lactate Provides Insights into Symbiosis


 Kluyveromyces marxianus is a yeast that was identified from kefir and can use a broad range of substrates, such as glucose and lactate, as carbon sources. The lactate produced in kefir culture can be a substrate for K. marxianus. However, the complexity of the kefir microbiota makes the traits of K. marxianus difficult to study. In this research, we focused on K. marxianus cultured with lactate as the sole carbon source. The optimal growth and released protein in lactate culture were determined under different pH conditions, and the LC-MS/MS-identified proteins were associated with the tricarboxylic acid cycle, glycolysis pathway, and cellular stress responses in cells, indicating that autolysis of K. marxianus had occurred under the culture conditions. The abundant glyceraldehyde-3-phosphate dehydrogenase (GAP1) was crystallized in the cell-free fraction, and the low transcription level of the GAP1 gene indicated that the protein abundance under autolysis conditions was dependent on protein stability. These results suggest that lactate produced by lactic acid bacteria stimulates the growth and autolysis of K. marxianus, releasing proteins and peptides. These findings can be fundamental for K. marxianus probiotic and kefir studies in the future.


Introduction
Kluyveromyces marxianus is frequently isolated from dairy products, such as cheese and ke r. It is consistently detected in certain ke r cultures 1 and can be acquired from unpasteurized milk 2 . The long history of safe use of these products makes K. marxianus a generally recognized as safe (GRAS) yeast strain. In a recent review, K. marxianus was considered as a potential probiotic yeast 3 . K. marxianus can utilize a broad spectrum of carbon sources, including carbohydrates and organic acids, such as glucose, sucrose, lactose, fructose, galactose, xylose, lactic acid and malic acid 4 , making it suitable for use in the food industry. In ke r cultures, lactate can be a nonfermentable carbon source for K. marxianus. However, the complexity of the symbiotic bacteria and yeasts in ke r culture makes the interactivity of the microbiota di cult to study. To simplify the interaction between K. marxianus and lactate, we focused on the traits of the K. marxianus lactate culture.
Although several nonfermentable carbon sources can be used to grow K. marxianus, protein expression under these culture conditions is rarely discussed. Another issue with using nonfermentable carbon sources as substrates is autolysis. The known autolysis in Saccharomyces cerevisiae occurs after fermentable carbon sources are exhausted at stationary phase in the culture 5 . Substitution of a fermentable carbon source with a nonfermentable carbon source can lead to the formation of a stressful environment for yeast.
Lactate is a nonfermentable carbon source for yeast. It is known that some S. cerevisiae strains can express pectinase in the presence of lactate 6 . The induction of secreted pectinase expression implies that lactate can upregulate the expression of certain proteins in S. cerevisiae. Furthermore, both S. cerevisiae and K. marxianus can reduce lactate feedback inhibition of lactic acid bacteria 7 . With lactate as the sole carbon source, K. marxianus exhibits potential for biotechnological applications. The lactate uptake mechanism of K. marxianus involves uptake of lactate anions with a monocarboxylate uniport 8 . Therefore, the lactate uptake e ciency depends on the pH of the culture medium for the dissociation of lactate. Culture conditions with nonfermentable carbon sources and low pH cause autolysis in S. cerevisiae during winemaking 5 . Autolysis of yeast leads to the release of low-molecular-weight compounds and peptides that impart additional avor or sweetness to the wine 5,9 . In this study, the proteins that were released from K. marxianus autolysis and the transcription level of the genes that encoded the high-abundance proteins were investigated. The total released proteins were analyzed by SDS-PAGE and liquid chromatography-tandem mass spectrometry (LC-MS/MS). The results showed that maximum protein release occurred in culture medium at pH 5.2, and 11 out of the 15 most abundant proteins were possible moonlighting proteins.

Results
The initial pH value of the lactate culture medium affects the growth of K. marxianus The K. marxianus Bot3 + 7 strain was cultured in YPD medium for two days as a seed culture. Then, synthetic yeast culture medium with lactate as the sole carbon source, YNL, with different initial pH values was used for batch cultivation of the K. marxianus strain for ve days. The growth curves and the released protein concentrations are shown in Fig. 1a, b. Optimal growth was observed at pH 4.18 (optical density at 600 nm (OD 600 ) = 19) ( Table 1). Upon cultivation with YNL medium at pH values higher than 4.8, the cell concentration of the K. marxianus strain decreased (Table 1). This might be because the dissociation of lactate anions decreased with increasing pH in the medium. c The data are expressed as the mean ± standard deviation; n = 3 independent biological samples.
Protein released to the culture medium Table 1 shows that the concentration of protein released to the culture medium was associated with the initial pH of the culture medium. Maximum protein release was observed at pH 5.2 (178 µg mL − 1 ) (

Culture medium ltrate and protein concentration
The released protein level in the culture medium was too low for direct analysis by SDS-PAGE or LC-MS/MS. The culture media were rst ltered with membranes with a molecular weight cutoff of 3 k as described in the methods. The proteins were concentrated 20-fold for SDS-PAGE analysis ( Supplementary  Fig. S1). However, the amount of protein was not suitable for total protein LC-MS/MS analysis. Therefore, 300 mL of culture medium (at pH 5.44) in another batch experiment of K. marxianus Bot3 + 7 was collected to obtain more protein. Approximately 0.1 mg mL − 1 protein was obtained on the 12th day of incubation (data not shown). It took more than twice as much time to reach the optimum concentration of released protein than it did with the 3-mL culture (Fig. 1b). The medium was ltered through a 0.2-µm pore size membrane and stored at 4 °C for further concentration and desalting. Noticeable small crystals appeared in the storage medium after 2 days (Fig. 2a). The crystals formed in the culture medium did not dissolve in ddH 2 O and methanol but could be dissolved in Bradford reagent and low-pH solutions, such as 10% acetic acid. The crystals dissolved in Bradford reagent were analyzed by SDS-PAGE, and there were two major bands at approximately 37 kDa and 27 kDa (Fig. 2b). To determine the identity of the proteins, the bands were excised and sent for LC-MS/MS analysis. Peptide ngerprinting of the 37-kDa and 27-kDa proteins identi ed both proteins as glyceraldehyde 3-phosphate dehydrogenase 1 (GAP1) in K. marxianus (Fig. 3). Some protein regions were unmatched, which might have been due to insu cient in-gel trypsin digestion. Nonetheless, a notable ngerprint peptide (VAVFQEK) was absent just before the second acid cleavage site in the peptide (aspartic acid-proline, 79D-80P) among the 27-kDa peptide ngerprints compared with the 37-kDa ngerprints ( Fig. 3; Supplementary Table S1). The molecular mass of the N-terminal peptides (residues 1-79) after acid cleavage of G3P1 was approximately 9 kDa. These results suggest that the 27-kDa protein was derived from the acid cleavage of GAP1 when the protein was dissolved in acidic solutions.
General proteomics of released protein from K. marxianus in lactate culture medium Analysis of total released protein with LC-MS/MS Nonetheless, protein crystals were formed, and the proteins in YNL medium at pH 5.44 were further desalted and concentrated to 10 µg/µL for LC-MS/MS analysis as described in the methods. Note that since the samples were released by autolysis, the proteins would be cleaved at unexpected cleavage sites. To maintain reliability of the LC-MS/MS analysis, only trypsin cleavage-derived peptides were used as ngerprints to identify proteins. Therefore, the numbers of matched peptides were lower than expected. However, the identi ed proteins could be con rmed in the autolysis medium. A total of 513 proteins were matched in the NCBI K. marxianus protein database, and 473 out of the 513 identi ed proteins could be mapped to UniProtKB for further Gene Ontology (GO) classi cation.

Metabolism
Proteins associated with the tricarboxylic acid cycle, gluconeogenesis, and pentose phosphate pathway were identi ed (Supplementary Table S2). Among other carbohydrate metabolism processes, proteins related to galactose metabolism and exoinulinase were identi ed (Supplementary Table S2). Two proteins were identi ed as being related to lactate metabolism (D-lactate dehydrogenase [cytochrome] 2 and cytochrome b2) (Supplementary Table S2).

Stress response
The identi ed proteins included 29 proteins that were related to cellular stress responses, such as responses to starvation (5 proteins), osmotic stress (8 proteins), and oxidative stress (17). Notably, a 12 kDa heat shock protein (HSP12) was related to both osmotic and oxidative stress; therefore, it could be counted as only 1 protein, making the total 29. A total of 13 proteins (including heat shock proteins and chaperones) that help protein folding and refolding were identi ed in the sample (Supplementary Table S2).
Since the samples collected were of proteins released by autolysis, the observed proteins were relatively stable under the autolysis conditions. The most abundant protein, HSP12, had a notable abundance ratio compared to all the other matched proteins (Table 2). However, GAP1 crystallized during storage before LC-MS/MS analysis, so the abundance ratio of GAP1 in the total released proteins was underestimated. Potential strong promoters under lactate culture conditions High abundance of proteins in culture medium may be a result of pH-induced expression, protein stability, or both. To further investigate the transcription levels of the matched protein-encoding genes, the expression of the HSP12, GAP1, and ENO genes was measured by using quantitative real-time polymerase chain reaction (qPCR) (Fig. 4). In addition, the potential promoters of INU1 and PGU1, which encode inulinase and polygalacturonase, respectively, were also analyzed. The promoter of INU1 was applied for heterologous protein expression 13,14 , and the promoter of PGU1 has potential biotechnological applications in K. marxianus 15 . The transcription levels of all the genes were determined by qPCR and compared with that of the actin (ACT) gene to calculate the relative transcription levels. HSP12 presented relatively high transcription levels at pH 3.9 and pH 4.8, and the transcription level decreased at pH 6.23 (Fig. 4). The GAP1 transcription level was lower than that of ACT under all the tested conditions (Fig. 4). ENO transcription was high at pH 3.9 but decreased at pH 4.8 and pH 6.23 (Fig. 4). The notably high transcription levels of PGU1 at every tested pH indicated the potential for application in heterologous enzyme expression. On the other hand, the transcription levels of INU1 were lower than those of PGU1 (Fig. 4).

Discussion
K. marxianus produces different secreted enzymes when different carbon sources are used. For example, pectinase is produced when glucose or pectin is used as a carbon source 15 , and inulinase is produced when inulin or sucrose is used as a carbon source 16 . Therefore, it would be interesting to identify the proteins produced by K. marxianus in ke r culture. Lactate is the primary acid that is produced in ke r culture, and its level can range from 0.6 to 1% (w/v) 17,18 . On the other hand, application of the lactate assimilation trait of K. marxianus was reported to reduce the L-lactate feedback inhibition of Lactococcus lactis for nisin production 7 . This indicates that K. marxianus could play the same role during ke r production and the potential of lactate usage for biotechnological applications. In this study, to study microbiota interactions in ke r, we focused on using lactate as the sole carbon source to culture K. marxianus. The proteins released from K. marxianus under lactate culture conditions were identi ed.
DL-lactate was used as the sole carbon substrate for batch cultivation of K. marxianus at different pH values. Even under the optimal growth conditions at pH 4.18, as shown in the results (Table 1), the released protein concentrations and the SDS-PAGE analysis of released protein indicated that some of the yeast cells were still autolyzed ( Supplementary Fig. S1). The mechanism of autolysis in K. marxianus is based on the biosynthesis of lytic enzymes, which is similar to the mechanism in S. cerevisiae 19,20 , and optimum autolysis was observed at pH 4.5 in culture 20 . In this study, the optimum culture pH for autolysis was observed to be pH 5.2 (Table 1). However, the culture conditions were different from those in this study 20 , since this work lacked fermentable carbon sources in the culture medium. In the presence of 0.2 M MgCl 2 , the nal cell concentrations and the released protein concentrations in the culture medium increased and decreased, respectively (Fig. 1c, d). and HSP12 proteins found in the culture ltrate were also related to the maintenance of cell membrane integrity and stability 22,23 . These results suggest that K. marxianus may bear the stress that could decompose membranes. Nonetheless, the inhibition of autolysis suggested that the presence of 0.2 M MgCl 2 helped K. marxianus resist the stress occurring under the lactate culture conditions (Fig. 1c, d).
GAP1 is a moonlighting (multifunctional) protein that is involved in glycolysis and constitutively expressed when fermentable or nonfermentable carbon sources are used 24 . Some glyceraldehyde 3phosphate dehydrogenases are related to aggregation and apoptosis 25,26 . However, GAP1 was reported to be related to occulation and contributes to cell wall integrity in K. marxianus 22,27 . The expression of GAP1 from K. marxianus in S. cerevisiae induces occulation 27 . In this study, even though the yeast was separated from the proteins, stable GAP1 crystallized in the liquid (Fig. 2a). These results indicate that GAP1 plays an important role in K. marxianus occulation.
LC-MS/MS analysis of the total protein content revealed the identities of the proteins released after autolysis, other than the crystallized GAP1, and it also provided a glimpse of the metabolic pathways in K. marxianus. Enzymes in central metabolic pathways, such as the tricarboxylic acid cycle, glycolysis/gluconeogenesis, and pentose phosphate pathway, were identi ed (Supplementary Table S2). Although DL-lactate was the only carbon source that was supplied to the culture medium, proteins related to galactose and inulin metabolism were also expressed (Supplementary Table S2). These results suggest that K. marxianus has the potential to assimilate galactose and inulin under these culture conditions. Two lactate metabolism-related proteins, namely, D-lactate dehydrogenase [cytochrome] 2 and cytochrome b2, were identi ed (Supplementary Table S2). D-lactate dehydrogenase [cytochrome] 2 catalyzes the conversion of D-lactate to pyruvate with the reduction of two molecules of cytochrome. The association between these identi ed lactate assimilation proteins and the expression of the central metabolism enzymes indicates the lactate assimilation pathway in K. marxianus. It was reported that the PGU1 promoter is activated in the presence of lactate in S. cerevisiae 6 . In the lactate culture of K. marxianus, the transcription levels of PGU1 were higher than those of GAP1 and INU1 (Fig. 4). However, none of the analyzed peptides matched the PGU1 protein in our results. Among the 15 most abundant proteins, the average emPAI% of HSP12 was 18.7%, and that of the second abundant protein, GAP1, was 3.51% (Table 2). There was a 15.19% difference between these two proteins, and the third most abundant protein differed from GAP1 by only 1.21% (Table 2). These results suggest that most of the proteins were digested to unidenti ed peptides under the autolysis conditions. Therefore, PGU1 might be digested to unidenti ed peptides. However, none of the peptides matched PGU1, even though no enzyme digest ngerprints were considered (data not shown), which needs to be further investigated.
GAP1 was crystallized in the cell-free fraction and was found to be the second most abundant protein in the results (Table 2). However, the transcription level of GAP1 was lower than that of ACT (Fig. 4). These results suggest that the protein abundance in the autolysis condition was dependent on the stability of the protein, not on the gene transcription level. The most abundant protein, HSP12, was reported to increase membrane stability under different stress conditions in S. cerevisiae, such as heat shock and oxidative and osmotic stresses 23 . The expression of HSP12 in S. cerevisiae may also contribute to lifespan extension by protecting the membrane from desiccation 28,29 . The relatively high expression of HSP12 (Fig. 4) under the autolysis conditions suggests that HSP12 might play a similar role in protecting K. marxianus cells. Additionally, Marchal, et al. 30 demonstrated the contribution of peptides from HSP12 of S. cerevisiae to sweet perception after yeast autolysis in wine. HSP12 from K. marxianus autolysis might also contribute additional avor in ke r culture.
Protein abundance was related to protein stability and to K. marxianus abundance under the autolysis conditions. However, one cannot exclude the possibility that proteins might also be secreted into the medium rather than released from dead yeast cells. For example, GAP1, enolase and pyruvate decarboxylase are secreted into the extracellular space in Candida albicans without N-terminal signal peptides 31 . These nonconventional protein secretion pathways were also found in S. cerevisiae 32 and are related to nonclassical export (NCE) genes 33 such as NCE102, which is also encoded in the K. marxianus genome 34 . Although 513 proteins were identi ed, only 15 proteins had an abundance ratio higher than 0.5% (Table 2). Additionally, the 11 out of 15 most abundant proteins (Table 2) were reported as moonlighting proteins that do not have N-terminal signal peptides and might be secreted by nonconventional protein secretion pathways [35][36][37][38][39][40][41][42] . The alternative functions of these moonlighting proteins were reported to bind plasminogen, extracellular matrix protein ( bronectin, vitronectin and laminin) and kininogen and inhibit transcription [35][36][37][38][39]41,42 . Most of these studies focused on the extracellular matrix protein binding activities of these moonlighting proteins on the cell surface of the opportunistic pathogenic yeasts C. albicans, Candida parapsilosis and Candida tropicalis. Considering that the moonlighting proteins in this study were identi ed from the cell-free fraction, the alternative functions of these proteins might be different from those in previous reports. However, this could imply that the moonlighting proteins released from K. marxianus possibly competitively bind to extracellular matrix proteins and reduce the ability of those pathogenic yeasts to adhere to the cell surface. On the other hand, the secreted proteins might also be digested into peptides as the rest of the proteins are released by autolysis and become substrates for other symbiotic-related proteins in ke r. The well-known symbiotic interaction between Lactobacillus bulgaricus and Streptococcus thermophilus in yogurt is that S. thermophilus produces formate to stimulate L. bulgaricus, and L. bulgaricus liberates free amino acids and peptides from milk proteins to stimulate S. thermophilus 43 . In this study, lactate was used as a substrate for K. marxianus to grow and induce autolysis of K. marxianus to release proteins and peptides into the cell-free fraction. This suggests that the lactate produced by lactic acid bacteria would stimulate K. marxianus to release proteins and peptides. However, the functions of those proteins and peptides remain to be discovered. The proposed interaction model of lactic acid bacteria and K. marxianus is presented in Fig. 5. Nonetheless, the potential of K. marxianus lactate culture for preventing pathogenic yeast invasion and the roles of the released proteins and peptides in ke r culture need further investigation in the future.
K. marxianus has received recognition for its potential industrial applications in recent years due to its thermotolerance, rapid growth rates, and broad substrate spectrum 4 . Although knowledge of its biochemistry and genetics is limited compared to that for S. cerevisiae, many studies based on different biochemical principles, such as nonhomologous end joining 44 , homologous recombination 45,46 , and the CRISPR/Cas9 mechanism 47 , have been applied to develop genetic engineering tools. These studies used fermentable substrates for culture, such as glucose and galactose. Therefore, knowledge on culturing nonfermentable substrates is limited. In this study, the optimal lactate culture conditions, the released proteins and the transcription levels of several genes were determined in K. marxianus. Autolysis occurred in all of the experimental lactate culture conditions, and the LC-MS/MS results that present various heat shock proteins and moonlighting proteins can be the foundation of K. marxianus probiotic and ke r research.

Materials And Methods
Yeast strain and culture conditions The K. marxianus Bot3 + 7 strain was rst precultured for two days at 30 °C with rotation at 120 rpm in YPD medium containing 10 g of yeast extract, 20 g of peptone, and 20 g of glucose per liter of distilled water. To measure growth and protein release, precultured cells from a two-day culture were diluted 1:100 by adding 30 µL of the suspension to 3 mL of YNL medium (a synthetic medium composed of 6.7 g of yeast nitrogen-base without amino acids (Difco, Detroit, MI, USA) and 20 g of lactic acid (J.T. Baker, Philipsburg, NJ, USA) per liter of distilled water). The pH of the YNL medium was adjusted to 3.9, 4.18, 4.8, 5.2, 5.44 or 6.23, followed by the addition of NaOH (Showa, Tokyo, Japan). All yeast cultures were incubated at 30 °C in a rotating incubator (120 rpm). The medium was supplemented with MgCl 2 as indicated in the manuscript. For LC-MS/MS analysis of the total protein content, the precultured cells were diluted by adding 3 mL of the suspension to 300 mL of YNL medium (pH 5.44) and incubated at 30 °C in a shaking incubator (120 rpm) for LC-MS/MS protein analysis.

Growth curves and protein concentration determination
The growth of the K. marxianus Bot3 + 7 strain was determined by measuring the OD 600 and OD 595 with a GeneQuant 1300 system (GE Healthcare, Little Chalfont, Buckinghamshire, UK) and a Tecan Sunrise Basic microplate reader (Tecan Austria GmbH, Grödig, Austria), respectively. At each time point during the 5-day cultivation period, 20 µL of yeast culture was sampled and aliquoted into a 96-well microplate in triplicate and diluted 1:10 by adding 180 µL of distilled water to measure the growth. Additionally, protein secretion into the external medium was determined by measuring the OD 595 with a Tecan Sunrise Basic microplate reader using the Bradford method with bovine serum albumin as the standard by following the manufacturer's instructions (Bio-Rad Protein Assay Dye Reagent, Bio-Rad, Hercules, CA, USA). For this, 50 µL of yeast culture at each time point was centrifuged at 14,000 rpm for 3 min. After centrifugation, 40 µL of supernatant was sampled into a 96-well microplate in triplicate, and 160 µL of Bradford reagent (1.25fold) was added to be diluted 5-fold with the protein.

SDS-PAGE
The crystals were collected into 1. Fluka) was added, and the gel pieces were agitated at room temperature for 30-60 min and transferred to a new centrifuge tube. Second, 15 ~ 50 µL of 5% FA was added, and the tube was incubated at room temperature for 10 min, followed by the addition of 15 ~ 50 µL of 100% ACN and agitation for 30-60 min.
Third, the two extract solutions were combined and dried in a Speed Vac. The samples were desalted using C18 ZipTips (Millipore) with the following steps: the dried peptide mixtures were desalted in 10 µL of ddH 2 O containing 0.1% FA, a ZipTip was activated by rinsing 3 times with 10 µL of 100% ACN, the tip was equilibrated 3 times with 10 µL of 0.1% FA, peptides were bound to the tip by aspirating and dispensing the peptide solutions 10 times, the tip was washed 6 times with 10 µL of 0.1% FA, and peptides were eluted with 10 µL of 0.1% FA (dissolved in 50% ACN) by aspirating and dispensing 10 times. These steps were repeated to obtain a combined solution, and the combined solution was dried with a SpeedVac vacuum concentrator.

Proteins in autolysis medium
The proteins in the supernatant of the 12-day K. marxianus culture were concentrated to 10 µg/µL using Amicon Ultra- 15

Mascot search parameters
The type of search and enzyme were set as MS/MS Ion Search and trypsin, respectively. The variable modi cations included carbamidomethyl (C), deamidated (NQ), oxidation (HW), and oxidation (M).
Monoisotopic was used as the mass parameter, and the mass tolerance values for peptides and fragments were ± 0.05 Da and ± 0.03 Da, respectively. The max missed cleavages by enzyme was set as 2. The database used for the search was the NCBI K. marxianus database. The results are provided as a Supplementary Dataset.
Quantitative real time polymerase chain reaction procedure Total RNA extraction was performed using TRIzol Reagent (Invitrogen). The yeast cells were collected by centrifugation, and then, the total RNA was extracted according to the manufacturer's instructions. The RNA yield was determined by a Nano-100 Micro-Spectrophotometer (Medclub Scienti c Co., Ltd., Taiwan). A total of 0.1 µg of RNA was converted to cDNA using the GoScript™ Reverse Transcription System, and qPCR analysis was performed using LightCycler 480 SYBR Green Master Mix (Roche) on a Rotor-Gene 3000 Real-Time DNA Detection System (Corbett Research). The transcription levels were calculated for reverse transcriptase-treated and nontreated samples and normalized to those of actin (ACT). The changes in relative transcription levels were calculated as fold changes using the ΔΔCt method. All the primers used for analysis in this study are listed in Supplementary Table S3.

Data Availability
All data generated or analysed during this study are included in this published article (and its Supplementary Information les).   Schematic diagram of the peptides that matched to glyceraldehyde-3-phosphate dehydrogenase (GAP1) from the 37-kDa and 27-kDa protein bands in SDS-PAGE analysis.