Characterization and optimization of mnn11Δ-mediated enhancement in heterologous protein production in Kluyveromyces marxianus

Abstract

Background

N-glycosylation is a prevalent post-translational modification in eukaryotes, essential for regulating protein secretion. In Saccharomyces cerevisiae, glycosylation mutants have been shown to enhance the secretion of heterologous glycosylated proteins. However, whether these mutants can also increase the secretion of non-glycosylated proteins and whether the growth defects associated with glycosylation mutations can be mitigated remains unclear. This study aimed to characterize and optimize enhanced secretory expression in the promising yeast host Kluyveromyces marxianus by deleting MNN11, which encodes a subunit of the mannose polymerase II complex responsible for elongating α-1,6-linked mannose chains.

Results

Compared to wild-type cells, the mnn11Δ cells significantly increased the secretion activities of four glycosylated enzymes and three non-glycosylated enzymes in flasks, with increases ranging from 29 to 668%. Transcriptomic analysis of mnn11Δ mutant revealed upregulation of genes related to essential protein secretion processes, including vesicle coating and tethering, protein folding, translocation, and glycosylation. Additionally, genes involved in vacuolar amino acid transport and amino acid biosynthesis were upregulated, suggesting an amino acid shortage, which might contribute to the observed severe growth defect of the mnn11Δ mutant in a synthetic medium with inorganic nitrogen. Supplementation of the synthetic medium with amino acids or low concentrations of yeast extract alleviated this growth defect, reducing the specific growth rate difference between wild-type strain and mnn11Δ cells from 65% to as little as 2%. During high-density fermentation, the addition of 0.5% yeast extract substantially reduced the lag phase of mnn11Δ mutants and increased the secretory activities of α-galactosidase, endoxylanase, and β-glucanase, by 11%, 18%, and 36%, respectively, compared to mnn11Δ mutant grown without yeast extract.

Conclusion

In K. marxianus, deletion of MNN11 enhances the secretion of both glycosylated and non-glycosylated proteins by improving key protein secretion processes. The growth defect in the mnn11Δ mutant is closely tied to insufficient amino acid supply. Supplementing the synthetic medium with low concentrations of organic nitrogen sources effectively alleviates this growth defect and enhances secretory expression. This strategy could be applied to optimize the expression of other glycosylation mutants.

Background

N-glycosylation is a prevalent post-translational modification in eukaryotes [1]. In both yeasts and mammals, N-glycosylation begins in the endoplasmic reticulum (ER), where a pre-assembled oligosaccharide (Glc₃Man₉GlcNAc₂) is attached to specific asparagine residues on nascent peptides. Glucosidase I, glucosidase II, and an ER-resident α-1,2-mannosidase sequentially remove three glucose moieties and a terminal α-1,2-mannose [2]. The resulting Man₈GlcNAc₂-containing glycoprotein is transported to the Golgi apparatus, where glycosylation differs significantly between yeasts and mammals. In Saccharomyces cerevisiae, α-1,2-, α-1,3-, and α-1,6-mannosyltransferases, along with mannosylphosphate transferases, create N-glycan structures that are mannosylated and hypermannosylated to varying extents. This hyper-mannose modification is unique to yeast and absent in mammals [3]. About 50% of yeast proteins are N-glycosylated [4], and this modification is essential for protein folding, transport, and maturation [5].

Like native proteins, heterologous proteins expressed in yeast undergoes N-glycosylation [6]. Several studies have shown that deleting non-essential N-glycosylation genes in S. cerevisiae enhances the secretion of heterologous proteins. For example, deletion of genes responsible for oligosaccharide synthesis in the ER, such as ALG3, ALG5, ALG6, ALG8, ALG9, ALG10, or ALG12, promotes α-amylase secretion [7]. Deleting genes encoding glucosidase I (CWH41) and glucosidase II’s catalytic subunit (ROT2) enhances the secretion of cellulosomal enzymes [8]. Additionally, knocking out Golgi mannosyltransferase genes OCH1 and MNN9 improves the secretion of various cellulases [9], and deleting other mannosyltransferases genes MNN10 and MNN11 boosts amylase secretion [10]. Glycosylation mutants may enhance secretion through multiple mechanisms. First, glycosylation mutations might impair the function of cell wall-associated glycoproteins, increasing cell wall porosity and thereby facilitating protein secretion [9]. Second, these mutations reduce glycan moieties on heterologous proteins, potentially improving their permeability across membranes and the cell wall [11]. Third, glycosylation defects can activate stress response pathways such as the unfolded protein response (UPR), which may enhance heterologous protein secretion [9]. However, gaps remain in understanding N-glycosylation’s impact on heterologous protein production. First, studies have focused on glycosylated heterologous proteins; it is unclear if glycosylation mutants also boost the secretion of non-glycosylated proteins. Second, N-glycosylation mutants often exhibit growth defects [9, 12], reducing cell productivity. Whether these growth defects can be mitigated while maintaining improved heterologous protein production remains unknown. Third, there are limited studies on the impact of glycosylation defects in non-conventional yeasts.

Kluyveromyces marxianus and S. cerevisiae both belong to the Saccharomycetaceae family. K. marxianus, the most common yeast found in dairy products, has a long history of safe human consumption, which has led to its recognition as a food-safe species [13]. Its rapid growth, thermotolerance, and broad substrate range make it a promising cell factory for the production of bioethanol, chemicals, and heterologous proteins. Over 50 proteins [14], including industrial enzymes [15, 16], virus-like particles [17,18], and edible proteins [19], have been efficiently expressed in K. marxianus, with the highest yield reaching 16.8 g/L [20]. Various strategies have improved protein expression in K. marxianus, such as weakening autophagy [15], optimizing UPR and disulfide bond formation pathways [20, 21], and reducing cAMP cyclase activity [22]. However, studies on enhancing protein secretion through glycosylation gene mutations in K. marxianus are still lacking.

In this study, we screened K. marxianus mutants for enhanced heterologous protein expression, with a deletion mutant of the glycosylation gene MNN11 emerging as the top performer. The mnn11Δ mutant not only improved the secretion of glycosylated proteins but also non-glycosylated proteins. It showed upregulation in key protein secretion pathways. Additionally, the mnn11Δ mutant exhibited a potential amino acid shortage, which may contribute to the severe growth defect observed in synthetic medium with inorganic nitrogen. Supplementing the medium with individual amino acids or low concentrations of yeast extract effectively alleviated the growth defect. In high-density fermentation, the addition of 0.5% yeast extract further enhanced heterologous protein expression. This strategy may be applied to optimize the expression of other glycosylation mutants.

Materials and methods

Plasmids and strains

Plasmids used in this study are listed in Table S1. To construct plasmids for expressing heterologous proteins, the open reading frames (ORFs) of MEL1 (Uniprot ID: P04824, encoding α-galactosidase from S. cerevisiae) and SpChi1 (Uniprot ID: A0A8D4BAQ1, encoding chitosanase from Streptomyces pratensis) were inserted between the Sma I and Not I sites of pUKDN132. This resulted in the plasmids LHZ1626 and LHZ1627, respectively. Plasmids expressing Xyn-CDBFV (LHZ443/pZP46), RuCelA (LHZ442/pZP52), MAN330 (LHZ444/pZP42), AnFaeA (LHZ766) and BadGLA (LHZ1020), were described previously [16, 20, 21]. To construct CRISPR plasmids (LHZ1608-1625, 1645-1649), primers containing 20 bp target sequence were annealed in pairs and inserted into Sap I sites of LHZ531 [21]. Primers used in the construction are listed in Table S2.

All yeast strains used in the study are listed in Table S3. FIM-1ΔU was used as a wild-type strain in this study [16] and cultured in YPD medium (1% yeast extract, 2% hipolypepton, 2% glucose, 2% agar for plates). To delete MNN11 (gene ID: FIM-1_496), 500 bp sequence upstream and downstream of MNN11 ORF were amplified and ligated together as the donor sequence. CRISPR plasmid, LHZ1608 was co-transformed with the donor sequence into FIM-1ΔU by a lithium acetate method [23]. The transformants were selected on the synthetic complete minus uracil (SC-Ura) plate. A positive mnn11Δ clone was identified by PCR and named LHP1126. Using the same strategy, CWH41 (gene ID: FIM-1_900), FLC1 (gene ID: FIM-1_3021), NVJ3 (gene ID: FIM-1_3640), UBX2 (gene ID: FIM-1_3063), KEX1 (gene ID: FIM-1_1117), OPT2 (gene ID: FIM-1_3542), GDS1 (gene ID: FIM-1_5027), ASE1 (gene ID: FIM-1_1956), HUB1 (gene ID: FIM-1_4502), OST5 (gene ID: FIM-1_2393), PPM1 (gene ID: FIM-1_3059), CAP1 (gene ID: FIM-1_1767), TRM1 (gene ID: FIM-1_2607), ANK1 (gene ID: FIM-1_2414), GAG1 (gene ID: FIM-1_2904), YBL081W (gene ID: FIM-1_5156), MNN10 (gene ID: FIM-1_2211), ANP1 (gene ID: FIM-1_133) and HOC1 (gene ID: FIM-1_1731) were individually deleted in FIM-1ΔU to obtain LHP1126-1143, 1343 and 1344. MNN11 was deleted in LHP1143 to obtain LHP1157.To construct strains overexpressing MNN11, the ORF of MNN11 was inserted between the Spe I and Sac I sites of LHZ424. Both sites are located between the upstream and downstream sequences of the INU1 ORF (gene ID: FIM-1_326) [21]. The donor sequence was amplified from the resultant plasmid (LHZ1628) and co-transformed with a CRISPR plasmid targeting INU1 (LHZ759). The transformants were selected on the SC-Ura plates and the positive clone was named LHP1248. Strains overexpressing other genes were constructed using the same strategy. Primers used in the construction are listed in Table S2.

To express heterologous proteins, plasmids LHZ442-444, LHZ766, LHZ1020, LHZ1626 and LHZ1627 were transformed into FIM-1ΔU and mutants. The transformants were selected on SC-Ura plates and then grown at 30 °C in either YD medium (2% yeast extract, 4% glucose) or synthetic mineral (SM) medium, which includes synthetic mineral seed (SMS) medium and synthetic mineral fed-batch (SMF) medium [17]. Furthermore, 0.1% of individual amino acid and various amounts of yeast extract were added into the SMS and SMF medium as specified.

Enzymatic assays and SDS-PAGE

Transformants expressing heterologous proteins were cultivated in 50 mL of YD medium for 72 h. The cells were pelleted, and the supernatant was collected. A total of 16 µL of supernatant was mixed with 4 µL of 5×SDS-PAGE loading buffer, boiled, and then subjected to SDS-PAGE analysis. The activities of Mel1, AnFaeA, MAN330, RuCelA, BadGLA and Xyn-CDBFV in the supernatant were measured as described previously [16, 20, 24]. To measure the activity of chitosanase, 100 µL of supernatant was mixed with 100 µL of a 1% (w/v) chitosan solution in 50 mM NaAc buffer (pH 6.5). The sample was incubated at 55℃ for 15 min, after which 200 µL of DNS solution (D7800, Solarbio) was added. The mixture was then incubated at 100℃ for 10 min, and the released N-acetylglucosamine was determined by measuring the optical density at 540 nm (OD₅₄₀). One unit of chitosanase activity was defined as the amount of enzyme required to release 1 µmol of N-acetylglucosamine per minute.

RNA-seq and data analysis

FIM-1ΔU and LHP1126 were grown in YPD liquid medium overnight and then transferred into 50 mL of fresh YD medium, starting at an OD₆₀₀ of 0.01. The cells were collected after growth periods of 24 or 48 h. At each time point, three biological replicate samples were collected. RNA extraction and RNA-seq were performed by Biozeron (Shanghai, China). All gene expression levels were listed in Table S4. To identify differentially expressed genes (DEGs), the expression level for each gene was calculated using the fragments per kilobase of exon per million mapped reads (FRKM) method. R statistical package edgeR (http://www.bioconductor.org/packages/release/bioc/html/edgeR.html/) was used for differential expression analysis. Genes with|log₂ Fold Change|≥1 and p-value < 0.05 are defined as significantly differentially expressed. GO functional enrichment and KEGG pathway analysis were carried out by Goatools (https://github.com/tanghaibao/Goatools) and KOBAS (http://kobas.cbi.pku.edu.cn/home.do). DEGs were significantly enriched in GO terms and metabolic pathways when their p-value was less than 0.05. Results of GO and KEGG enrichment analysis were listed in Table S5 and Table S6, respectively.

Cell growth profiling

FIM-1ΔU and LHP1126 were grown in YPD liquid medium overnight and then transferred into 50 mL of fresh YD, SMS, or SMS medium supplemented with different amino acids or various amounts of yeast extract. The OD₆₀₀ of the culture was adjusted to 0.01. Cells were cultured for 72 h, and the OD₆₀₀ was measured at intervals of 8 or 12 h. Each test was performed in biological triplicate.

Spot assay

FIM-1ΔU and LHP1126 were grown in YPD liquid medium overnight. Cells were collected and adjusted to an OD₆₀₀ of 1.0. Then, fivefold serial dilutions were performed and dilutions were spotted by a pinpad onto YPD and SMS plates containing hygromycin B (H8080, Solarbio). Plates were incubated at 30 °C for 24 h before imaging.

High cell-density fermentation

High cell-density fermentation was conducted in a 5-L fermenter (BXBIO, Shanghai, China) as previously described [17]. Cells expressing Mel1, Xyn-CDBFV or RuCelA were inoculated into 150 mL of SMS medium and grown for 16-18 h. The seed culture was then transferred into a fermenter containing 1.5 L of SMF medium, with or without 0.5% yeast extract. During fermentation, the dissolved oxygen level was maintained at approximately 10%, and the temperature was controlled at 30℃. The pH was adjusted to approximately 5.5 using ammonium hydroxide. At specified intervals, 50 mL of culture was taken to determine the OD₆₀₀ and enzyme activities.

Results and discussions

Screening genes to enhance secretory expression in K. marxianus

To enhance the secretion capacity of K. marxianus, we selected 17 candidate genes from the literature. In S. cerevisiae, deletion of homologs of these candidate genes has been shown to improve secretion expression [7,8,9,10, 25, 26]. Predicted locations of the proteins encoded by the candidate genes include the ER (Cwh41, Ost5, Flc1, Ubx2), Golgi apparatus (Mnn11, Kex1), mitochondria (Gds1), nucleus (Trm1, Ase1), peroxisome (Opt2), bud neck (Cap1, Hub1), and vacuole (Nvj3) (Fig. 1A). We evaluated the effects of deleting or overexpressing the candidate genes on the secretion activities of two glycosylated enzymes: α-galactosidase Mel1 from S. cerevisiae and feruloyl esterase AnFaeA from Aspergillus niger. Eight deletion mutants (mnn11Δ, cwh41Δ, flc1Δ, gds1Δ, trm1Δ, ybl081wΔ, gag1Δ, ank1Δ) displayed significantly increased Mel1 activity. None of the overexpression significantly increased Mel1 secretion (Fig. 1B). Five deletion mutants (mnn11Δ, cwh41Δ, ybl081wΔ, ank1Δ, ubx2Δ) exhibited improved AnFaeA activity. Overexpression of FLC1, TRM1, or OST5 increased AnFaeA activity (Fig. 1B). Interestingly, deletion, but not overexpression of FLC1 and TRM1 promoted Mel1 secretion. FLC1 encodes a flavin adenine dinucleotide transporter [27], and TRM1 encodes a tRNA methyltransferase [28]. The opposite effects of deletion and overexpression of FLC1 and TRM1 on the secretory activities of Mel1 and AnFaeA suggest these genes play pleiotropic roles in controlling the secretion of different proteins.

Fig. 1

Effects of deletion or overexpression of candidate genes on the secretory activities of Mel1 and AnFaeA. (A) Schematic view of the protein secretion pathway. Directions of protein secretion are indicated by arrows, and the subcellular locations of proteins encoded by candidate genes are shown, based on SGD annotation. (B) Effects of deletion or overexpression of candidate genes on the secretory activities of Mel1 and AnFaeA. Cells were grown in YD medium for 72 h at 30 °C, and enzyme activities in the supernatant were measured. Activities expressed by the wild-type (WT) cells were set as unit 1. Relative activities of cells harbouring deletions (left) or overexpression (right) of candidate genes are shown. Increases or decreases in activities are indicated in red and blue, respectively. Values are expressed as mean ± SD (n = 3) from biological replicates. * p < 0.05, ** p < 0.01. We failed to obtain strains overexpressing CWH41 or HUB1, probably because overexpressing either gene causes toxicity to the cells. Therefore, the effects of overexpressing CWH41 or HUB1 on the activities of Mel1 and AnFaeA were not measured. (C) Growth curves of wild-type strain and selected mutants during expression of Mel1 and AnFaeA. (D) Comparison between the maximum specific growth rate (µ_max) and improvement in the enzymatic activities in the selected mutants

A total of four deletion mutants (mnn11Δ, cwh41Δ, ybl081wΔ, ank1Δ) simultaneously enhanced the secretory activities of both Mel1 and AnFaeA. The growth curves of these strains were investigated during the expression of Mel1 or AnFaeA (Fig. 1C). Compared with the wild-type strain, deletion of CWH41 and YBL081W had little negative effect on growth. The ank1Δ mutant exhibited a prolonged lag phase, but its biomass and maximum specific growth rate (µ_max) were barely affected (Fig. 1C, D). The mnn11Δ mutant displayed substantially reduced biomass and µ_max (Fig. 1C, D). Meanwhile, compared with other mutants, the mnn11Δ mutant exhibited the highest average improvement in Mel1 and AnFaeA activities, suggesting a trade-off between cell growth and improved protein production in the mnn11Δ mutant (Fig. 1D).

MNN11, OST5 and CWH41, are the three glycosylation genes included in this screen. Ost5 is located in the ER and functions as a zeta subunit of the oligosaccharyltransferase (OST) complex, which transfers 14-sugar branched oligosaccharides from dolichyl pyrophosphate to asparagine residues on nascent peptides [29]. Cwh41 possesses α-glucosidase I activity and is an integral membrane protein of the ER, responsible for removing the terminal glucose from core oligosaccharides immediately after they are transferred to proteins [30]. Mnn11 is a subunit of the Mannose polymerase II complex (M-Pol II) and contributes to the extension of α-1,6-linked mannose polymers in the Golgi [31]. Contrary to the improved activities observed in the mnn11Δ and cwh41Δ mutants, the ost5Δ mutant displayed significantly reduced secretory activities of Mel1 and AnFaeA. Overexpression of OST5 even increased AnFaeA activity. Several key proteins involved in protein secretion, such as Ero1 and Pdi1, contain conserved glycosylation sites (Asn-Xaa-Ser/Thr). The deletion of OST5 affects oligosaccharide addition, leading to significant changes in glycosylation [29, 32], which might impair the function of key proteins involved in secretion and reduce secretory activity. In contrast, the deletion of CWH41 affects the removal of terminal glucose but does not disrupt the addition of the oligosaccharide backbone [33]. Deletion of MNN11 reduces the length of the outside chain [34]. Neither deletion caused dramatic changes to the oligosaccharide backbone. Therefore, deletions of genes involved in late-stage glycosylation are more likely to maintain the activity of glycosylated proteins essential for secretion while promoting the secretion of heterologous proteins through other pathways.

Deletion of MNN11 enhances the secretory expression of both glycosylated and non-glycosylated proteins

To determine the universality of the mnn11Δ mutant in promoting secretory expression, we tested the secretion of five different enzymes. These enzymes include glucoamylase BadGLA, endo-1,4-β-endoxylanase Xyn-CDBFV, endo-1,4-β-mannanase MAN330, chitosanase SpChi1, and endo-1,4-β-glucanase RuCelA. BadGLA and Xyn-CDBFV are hypermannosylated [16, 20], while the other three proteins are not glycosylated. As shown in Fig. 2A-E, compared to wild-type strain, the mnn11Δ mutant increased the secretion activities of BadGLA, Xyn-CDBFV, MAN330, SpChi1, and RuCelA by 176%, 43%, 130%, 668%, and 62%, respectively, indicating that the mnn11Δ mutant promotes the secretion of both glycosylated and non-glycosylated proteins.

Fig. 2

Effect of MNN11 and MNN10 deletion on the secretory expression of glycosylated and non-glycosylated proteins. (A-E) Secretory activities of BadGLA (A), Xyn-CDBFV (B), MAN330 (C), SpChi1 (D), and RuCelA (E) in mnn11Δ, mnn10Δ and mnn10Δ mnn11Δ mutants. Cells were grown in 50 mL YD medium for 72 h at 30 °C, and enzyme activities in the supernatant were measured. Activities expressed by a wild-type cell (WT) were set as unit 1. Relative activities of mnn11Δ, mnn10Δ and mnn10Δ, mnn11Δ mutants are shown. Values are expressed as mean ± SD (n = 3) from biological replicates. * p < 0.05, ** p < 0.01, ns no significant. (F-J) SDS-PAGE of the supernatant. Arrows indicate the theoretical positions of the heterologous proteins. A smear that contained glycosylated proteins is indicated by the bracket. (K) Growth curves of mutants of M-Pol II complex subunits. Cells were inoculated into 50 mL YD medium at an initial OD₆₀₀ of 0.01 and grown at 30 °C for 72 h. Values were calculated as mean ± SD (n = 3) from biological replicates. (L) Secretory activities of heterologous enzymes in the mnn11Δ and hoc1Δ mutants. Activities were measured as described in (A)

SDS-PAGE showed that the glycosylated forms of BadGLA and Xyn-CDBFV expressed by wild-type cells appeared as smears above their theoretical molecular weights (Fig. 2F, G), consistent with previous studies [16, 20]. In the mnn11Δ mutant, the smears migrated faster (Fig. 2F, G). The differences in the glycosylation smears were more obvious in the SDS-PAGE using more samples (Fig S1). The results indicate that the deletion of MNN11 affects M-Pol II activity, leading to defects in outside chain extension and reduced molecular weight. Compared to the wild-type cells, the intensity of the smear increased significantly in the mnn11Δ cells, suggesting an increase in the yield of BadGLA and Xyn-CDBFV. Similarly, MNN11 deletion significantly increased the band intensity of MAN330, SpChi1, and RuCelA (Fig. 2H-J), indicating that the improved secretory activities in the mnn11Δ mutant are due to increased amounts of secreted proteins.

In addition to Mnn11, the M-Pol II subunits include Anp1, Mnn9, Hoc1, and Mnn10 [31]. Mnn10 is distantly related to Mnn11 (15.9% identity). Deletion of MNN10 also affects M-Pol II activity [10]. Studies in S. cerevisiae suggest a positive genetic interaction between MNN11 and MNN10 [35]. Therefore, we investigated the effect of deletion of MNN10 on secretion. Compared to the wild-type cells, the mnn10Δ cells did not significantly increase BadGLA activity (Fig. 2A). Although the mnn10Δ mutant improved the secretion activities of Xyn-CDBFV, MAN330, SpChi1, and RuCelA, the activities of three enzymes were significantly lower than those in the mnn11Δ mutant (Fig. 2A, C and D). Overall, the positive effects of MNN10 deletion on secretion were weaker than those of MNN11 deletion. The positions of the glycosylation smears of BadGLA and XynCDBFV in the mnn10Δ mutant differed from those in the mnn11Δ mutant (Fig. 2F, G). In S. cerevisiae, mnn10Δ and mnn11Δ mutants also displayed different migration speeds of glycosylated invertase [10]. These results suggest that the deletions of MNN10 and MNN11 affect M-Pol II activity to different extents, leading to varying degrees of defects in outside chain extension, which may explain the differential effects of MNN10 and MNN11 deletions on protein secretion.

To further investigate the relationship between MNN10 and MNN11 in promoting secretion, we constructed the mnn10Δ mnn11Δ double-deletion mutant. There was no positive synergistic effect of the double deletion on secretory activities (Fig. 2A-E). In contrast, the secretory activities of Xyn-CDBFV, MAN330, SpChi1, and RuCelA in the mnn10Δ mnn11Δ mutant were even lower than those in the mnn10Δ or mnn11Δ single mutant (Fig. 2B-E). These results suggest that a moderate defect (as seen in the mnn10Δ or mnn11Δ mutants), rather than a dramatic defect in M-Pol II complex (as seen in the mnn10Δ mnn11Δ mutant), can achieve better improvement in protein secretion.

We also constructed mutants for two other subunits of the M-Pol II complex: Anp1 and Hoc1. However, we encountered difficulty deleting the third subunit, Mnn9. Compared to the transformation plate for deleting HOC1, far fewer colonies formed on the plate for deleting MNN9, and no positive mnn9Δ clones were identified afterward (Fig. S2). The results suggest that deleting MNN9 causes a severe growth defect or even lethality in K. marxianus, reflecting the importance of the M-Pol II complex for cell growth. Consistent with this idea, mutants of other M-Pol II subunits (mnn11Δ, mnn10Δ, anp1Δ, and hoc1Δ) all exhibited growth defects to varying extents. The growth of the anp1Δ mutant was poor in overnight YD culture (12 h) (Fig. 2K). When transformed with a plasmid expressing a heterologous protein, the anp1Δ mutant obtained far fewer and slower-growing transformants compared to the wild-type and hoc1Δ cells (Fig. S3). Therefore, we focused on the effects of the hoc1Δ mutant. Results showed that the hoc1Δ mutant displayed improved expression of SpChi1 to a comparable level as the mnn11Δ mutant. However, deleting HOC1 had no effect on the activities of BadGLA, Xyn-CDBFV, and RuCelA, and even significantly reduced the activity of MAN330 (Fig. 2L). Consequently, mnn11Δ mutant stood out as the best performer in promoting protein expression among the M-Pol II subunit mutants.

mnn11Δ mutant upregulates key processes in protein secretion and exhibits a high demand for amino acids

To investigate the mechanism by which MNN11 deletion enhances heterologous protein expression, we compared the transcriptomes of wild-type and mnn11Δ cells after 24 and 48 h of growth. At 24 h, there were 1,428 differentially expressed genes (DEGs) between wild-type and mnn11Δ cells, accounting for 27% of the total gene count, with 1,169 genes significantly upregulated and 259 genes significantly downregulated. At 48 h, there were 1,744 DEGs, accounting for 33% of the total, with 1,492 genes significantly upregulated and 252 genes significantly downregulated. Notably, 826 genes were consistently upregulated at both 24 h and 48 h, while 141 genes were consistently downregulated at both time points (Fig S4). The deletion of MNN11 caused significant changes in the transcriptome.

We performed gene ontology (GO) term and KEGG enrichment analyses of the DEGs. Among the top 10 enriched GO terms present at both 24 h and 48 h were mitochondrial translation and carbohydrate metabolism, closely related to energy supply, as well as processes related to protein synthesis, including organic nitrogen component biosynthesis, peptide metabolism, and ribosome biogenesis. In the KEGG analysis, processes related to energy supply, such as the TCA cycle and oxidative phosphorylation, and those related to protein secretion, such as protein export and SNARE interaction, were enriched at 24 h and 48 h (Fig. 3B, C). These enriched processes are closely related to protein secretion, demonstrating the reliability of RNA-seq and DEGs analysis. Notably, the pyrimidine metabolism appeared in both the enriched GO and KEGG terms. Genes involved in pyrimidine metabolism generally upregulated in mnn11Δ mutant, which may be related to the enhanced gene transcription.

Fig. 3

mnn11Δ mutant upregulates key processes in protein secretion. (A) GO term enrichment analysis of DEGs. Wild-type and mnn11Δ cells were grown in YD liquid medium for 24 or 48 h. Cells were collected and subjected to transcriptomic analysis. Genes with a p-value < 0.05 and|log₂ fold change| ≥ 1 were defined as DEGs. GO enrichment was performed based on biological processes (BP). The top 10 BP terms at both time points with a p-value < 0.05 are shown. (B-C) KEGG enrichment analysis of DEGs. Terms with p-value < 0.05 are listed in (B) (24 h) and (C) (48 h). (D) Comparison of DEG regulation among mnn11Δ, P7, and K5 strains. A total of 238 DEGs at 48 h were extracted from enriched terms, and the DEGs were categorized based on their functional annotation. The regulation status of these DEGs in P7 and K5 at 48 h was extracted from a previous study [19]. Circles filled with pink indicate upregulation, and those filled with blue indicate downregulation

We selected 238 DEGs from enriched terms closely related to protein secretion and analyzed their upregulation and downregulation. In a previous study, we introduced heterologous or native disulfide bond formation modules into K. marxianus using artificial chromosomes, resulting in the P7 and K5 strains. Similar to the mnn11Δ mutant, the P7 and K5 strains exhibited broad-spectrum promotion of heterologous protein expression [20]. Therefore, we compared the regulation status of 238 DEGs with that in the P7 and K5 strains to identify shared regulatory characteristics in high-yield strains.

As shown in Fig. 3D, six glucose transporter genes, ILV2 (involved in amino acid synthesis), PDB1, PYC2, and CIT3 (involved in TCA cycles), and TLG1 (involved in vesicle transport) were downregulated in the mnn11Δ mutant. All other DEGs in the mnn11Δ mutant were upregulated. The proportions of synchronously upregulated genes in mnn11Δ, P7, and K5 were 88%, 69%, 60%, and 60% for vesicle coating, protein translocation, protein glycosylation substrates, and vesicle tethering, respectively. This suggests that the upregulation of these processes is a common feature in high-secretion strains. Notably, even though P7 and K5 do not have glycosylation gene mutations, both strains, like mnn11Δ mutant, upregulated glycosylation genes and glycosylation substrate genes (43% and 60% synchronously upregulated, respectively). This suggests that the upregulation of glycosylation-related genes is not a compensatory response unique to glycosylation mutants but may be a general strategy used by K. marxianus to enhance secretory expression. All the processes containing high proportions of synchronously upregulated genes are essential for the protein secretion pathway. Enhancement of these processes in mnn11Δ mutant is expected to promote the secretory expression of various types of proteins, including both glycosylated and non-glycosylated proteins.

In addition to the shared characteristics with P7 and K5, mnn11Δ mutant exhibited unique gene expression features. The yeast vacuole is the primary intracellular degradation organelle and plays a key role in balancing amino acid concentrations [36]. In mnn11Δ mutant, amino acid transporters and V-ATPase on the vacuolar membrane were significantly upregulated, a trend not observed in P7 or K5 (Fig. 3D). The upregulation of V-ATPase may increase H⁺ pumping, enhancing proteolytic activity to break down short peptides into amino acids. Upregulated transporters may release more amino acids to maintain intracellular levels, suggesting that mnn11Δ mutant has a higher amino acid demand. Supporting this, genes involved in amino acid biosynthesis and amino acid transportation across the plasma membrane were also significantly upregulated in mnn11Δ mutant (Fig. 3D).

We conducted a detailed analysis of amino acid biosynthesis gene expression (Fig. 4). For the synthesis of 16 amino acids, at least one gene in one step was upregulated at 24 or 48 h. In contrast, only one gene involved in the fourth step of leucine synthesis was significantly downregulated in mnn11Δ mutant at 24 h. The general upregulation of amino acid synthesis in mnn11Δ mutant suggests an insufficient intracellular amino acid supply. Mutations in M-Pol II complex subunits affect cell wall function [9, 10], such as altering porosity. Cell wall defects can impair nutrient uptake, including amino acids [37], leading to intracellular amino acid shortages, which may stimulate the upregulation of amino acid synthesis and enhance amino acid flux between the vacuole and cytoplasm.

Fig. 4

mnn11Δ mutant upregulates amino acids biosynthesis. The biosynthesis pathways of 20 amino acids were depicted according to SGD YeastPathways (https://pathway.yeastgenome.org/). Boxes next to each amino acid represent steps in its biosynthesis. A solid or dashed line around each box indicates steps at 24 h and 48 h, respectively. One step can involve more than one gene. The expression levels of genes in mnn11Δ cells were compared to wild-type cells. If any gene in a step was significantly upregulated or downregulated, the step is labelled accordingly. There is no contradictory regulation within each step. Red indicates upregulation, and blue indicates downregulation. Relative gene levels in each step are shown in Table S7. Abbreviation: Glcnlac-6P, 6-phosphogluconolactone; Glcn-6P, 6-phosphogluconate; Rl-5P, ribulose 5-phosphate; R-5P, ribose 5-phosphate; Xul-5P, xylulose 5-phosphate; S-7P, sedoheptulose 7-phosphate; GAP, glyceraldehyde-3-phosphate; Ery-4P, erythrose 4-phosphate; Fru-6P, fructose-6-phosphate; Glc-6P, glucose-6-phosphate; Fru-1,6-BP, fructose-1,6-bisphosphate; DHAP, dihydroxyacetone phosphate; 1,3-BPG,1,3-bisphosphoglycerate; 3-PG, 3-phosphoglycerate; 2-PG, 2-phosphoglycerate; PEP, phosphoenolpyruvate; Pyr, pyruvate; OAA, oxaloacetate; Cit, citrate; Isocit, isocitrate; α-KG, α-ketoglutarate; Suc-CoA, succinyl-CoA; Suc, succinate; Fum, fumarate; Mal, malate

Supplementation with organic nitrogen sources alleviated the growth defect of mnn11Δ mutant

The capacity of mnn11Δ mutant to promote the secretion of various heterologous proteins suggests that it could serve as a promising host for industrial protein production. In industrial production, growth rate and media composition are key factors for reducing costs. YD is a common fermentation medium for K. marxianus, containing 2% yeast extract and 4% glucose. Synthetic mineral (SM) medium is a most commonly used medium for K. marxianus in high-density fermentation [38], and it includes seed medium (SMS) and feed-batch medium (SMF). The main components of SMS and SMF are (NH₄)₂SO₄, glucose, KH₂PO₄, MgSO₄, and trace amounts of vitamins and minerals. SMS and SMF use inorganic nitrogen sources, reducing their costs to 13% and 40% of YD, respectively, making them more suitable for the industrial production of bulk protein products. SM medium has already been successfully used for high-density fermentation of various heterologous proteins, including industrial enzymes, virus-like particles, and edible proteins [16,17,18,19, 21, 39]. Therefore, we compared the growth of wild-type and mnn11Δ cells in both YD and SMS. In YD, the maximum specific growth rate (µ_max) of the mnn11Δ cells decreased by 17%, and the OD₆₀₀ decreased by 26%, compared to wild-type cell (Fig. 5A). In SMS, the growth defect of the mnn11Δ cells was more pronounced, with µ_max and OD₆₀₀ both decreasing by 65% compared to wild-type cells (Fig. 5A).

Fig. 5

Supplementation with organic nitrogen sources alleviated the growth defect of mnn11Δ mutant. (A-E) Growth curves of wild-type and mnn11Δ cells with different nitrogen sources. Cells were inoculated into 50 mL medium at an initial OD₆₀₀ of 0.01 and grown at 30 °C for 72 h. Growth curves were obtained by monitoring OD₆₀₀ in YD and SMS medium (A), SMS medium supplemented with 0.1% valine or tyrosine (B), 0.1% lysine or arginine (C), 0.1% methionine or glutamate (D), and different concentrations of yeast extract (YE) (E). Growth curves of wild-type cells in (B, C, D) were from the same sample. Values were calculated as mean ± SD (n = 3) from biological replicates. The maximum specific growth rate (µ_max) for each curve is shown in the inset. (F) Spot assay of wild-type and mnn11Δ cells. Overnight cultures were diluted 5-fold and spotted onto YD, SMS, and SMS plates supplemented with 0.5% YE. To test cell wall integrity, the medium was supplemented with hygromycin B. Plates were incubated at 30 °C for 24 h

Transcriptomic analysis of the mnn11Δ mutant suggested an insufficient intracellular amino acid supply (Figs. 3D and 4). Amino acids are essential for cell growth, and their shortage can cause growth defects [40]. Since SMS uses ammonium sulfate as its nitrogen source, all amino acids must be synthesized de novo, exacerbating the deficiency in mnn11Δ mutant, and leading to its severe growth defect. If this hypothesis is correct, supplementing SMS with amino acids should alleviate this defect. Therefore, we tested the effect of adding 0.1% of individual amino acid to SMS on the growth of wild-type and mnn11Δ cells. In wild-type cells, no amino acid significantly increased OD₆₀₀ at 72 h (Fig. 5B-D). However, adding glutamate, arginine, and lysine increased µ_max of wild-type cells by 48%, 44%, and 44%, respectively (Fig. 5C, D). Valine, tyrosine, and methionine had minimal impact on µ_max (Fig. 5B, D), likely due to yeast’s preference for amino acid assimilation. Glutamate and arginine are preferred nitrogen sources, valine is intermediate, and tyrosine and methionine are non-preferred nitrogen sources [41]. As a result, glutamate and arginine are more efficiently assimilated, improving µ_max. While lysine cannot be utilized by S. cerevisiae or other yeast species in the post-whole-genome-duplication (WGD) clade [42], K. marxianus, a pre-WGD species, harbours potential lysine assimilation pathways [43]. Our results suggest lysine can serve as a preferred nitrogen source in K. marxianus.

On the other hand, adding any amino acid substantially improved the growth of mnn11Δ cells. The addition of arginine, lysine, glutamate, valine, methionine, and tyrosine reduced the gap in µ_max between wild-type and mnn11Δ cells from 67% to 41%, 39%, 29%, 22%, 14%, and 2%, respectively. Similarly, the addition of these amino acids reduced the OD₆₀₀ gap between wild-type and mnn11Δ cells at 72 h from 63% to 35%, 38%, 18%, 19%, 27%, and 29%, respectively. These results suggest that the growth defect of mnn11Δ mutant in SMS is closely related to insufficient amino acid supply, and supplementation with amino acids effectively alleviates this defect. Since supplementing individual amino acids is costly for industrial production, we investigated whether low concentrations of yeast extract (YE) could achieve similar effects. The results showed that 0.1%, 0.5%, and 1% YE reduced the µ_max gap between wild-type and mnn11Δ cells from 56% to 19%, 15%, and 8%, respectively (Fig. 5E), and reduced the OD₆₀₀ gap at 72 h from 56% to 22%, 9%, and 7%. These results suggest that YE, as a complex organic nitrogen source, can alleviate the growth defect of mnn11Δ mutant in a dose-dependent manner.

As suggested in the previous section, the cell wall defect caused by MNN11 deletion may lead to insufficient amino acid supply. As shown in Fig. 5B-E, supplementing SMS with YE effectively alleviated the growth defect likely caused by an amino acid shortage. To confirm the connection between the cell wall defect and amino acid deficiency, we investigated the impact of YE supplementation on the cell wall defect phenotype. The spot assay revealed that mnn11Δ cells grew slightly worse than wild-type cells in YD (Fig. 5F). As the hygromycin B concentration increased, the growth difference between wild-type and mnn11Δ cells became more pronounced, indicating greater sensitivity to hygromycin B and a cell wall defect in mnn11Δ cells, consistent with findings in S. cerevisiae [10]. On SMS plates, mnn11Δ cells grew substantially worse than wild-type cells, consistent with the growth curve results in Fig. 5A. When hygromycin B was added to SMS, mnn11Δ cells exhibited a pronounced synthetic growth defect, suggesting a link between the amino acid shortage and the cell wall defect. Supplementing SMS + hygromycin B with 0.5% YE rescued the severe growth defect of mnn11Δ cells, restoring the growth difference between wild-type and mnn11Δ cells to the level observed in YD (Fig. 5F). This result indicates that supplementing SMS with low concentrations of organic nitrogen not only alleviates the growth defect induced by amino acid shortage but also mitigates the cell wall defect.

Supplementation with yeast extract improves the yield of mnn11Δ mutant in high-density fermentation

To verify whether supplementing organic nitrogen sources in SM could be applied in industrial production, we conducted high-density fermentation in a 5-L fermenter. At the flask scale, the µ_max of mnn11Δ mutant in SMS supplemented with 0.5% YE was close to that in SMS with 1% YE (Fig. 5E). To reduce costs, we used SMS and SMF containing 0.5% YE for high-density fermentation, which cost 32% and 59% of YD containing 2% YE, respectively. In the initial stage of fermentation, YE supplementation increased the µ_0−12 h of wild-type cells expressing Mel1, Xyn-CDBFV, and RuCelA by 68%, 102%, and 69%, respectively. In contrast, the mnn11Δ cells exhibited a long lag phase in SMS without YE, leading to an extremely low µ_0−12 h. However, with YE supplementation, µ_0−12 h of mnn11Δ cells expressing Mel1, Xyn-CDBFV, and RuCelA increased by 1318%, 1143%, and 1037%, respectively (Fig. 6A-C). Thus, YE supplementation effectively shortened the lag phase of the mnn11Δ mutant, allowing it to enter the exponential phase more rapidly. This improvement is valuable for industrial production, as a shorter lag phase reduces contamination risk.

Fig. 6

Supplementation of low concentrations of yeast extract improves the yield of mnn11Δ mutant in high-density fermentation. (A-C) Growth curves and secretory activities during high-density fermentation of wild-type and mnn11Δ cells. Cells expressing Mel1 (A), Xyn-CDBFV (B), or RuCelA (C) were grown in SMS and SMF with or without 0.5% YE supplementation. Specific growth rates during the early stages (µ_0−12 h) are shown in the inset. (D-F) SDS-PAGE of the supernatant at 72 h. Arrows indicate the theoretical positions of heterologous proteins. A smear that contained glycosylated proteins is indicated by a bracket

During Mel1 expression, the OD₆₀₀ of the mnn11Δ cells with YE supplementation remained consistently higher than that of the mnn11Δ cells without YE (Fig. 6A). However, during Xyn-CDBFV and RuCelA expression, the OD₆₀₀ of the YE-supplemented mnn11Δ mutant fell below that of the mutant without supplementation after 48 h, likely due to the negative impact of overexpressed proteins on cell growth. Secretory activities in the YE-supplemented mnn11Δ mutant surpassed those in the mutant without supplementation after 12 h, and by 72 h, the secretory activities of Mel1, Xyn-CDBFV, and RuCelA were 11%, 18%, and 36% higher, respectively. Notably, YE did not affect the wild-type cells’ secretory activity, suggesting that YE specifically enhances the secretion advantage of the mnn11Δ cells. In terms of proportion, without YE supplementation, the activities of Mel1, Xyn-CDBFV, and RuCelA expressed by the mnn11Δ cells were 186%, 47%, and 245% of those expressed by wild-type cells, respectively. With YE supplementation, these proportions increased to 316%, 173%, and 470%, respectively. SDS-PAGE analysis of the supernatant was consistent with the secretory activities, as the amounts of heterologous proteins expressed by the YE-supplemented mnn11Δ mutant were significantly higher than those of the mutant without supplementation (Fig. 6D-F). Therefore, supplementing low concentrations of YE in SMS and SMF during high-density fermentation can shorten the lag phase of the mnn11Δ mutant and enhance its ability to express both glycosylated and non-glycosylated proteins.

Yet, it is worth noting that the addition of YE increases costs and causes issues in downstream protein purification. K. marxianus is recognized as a food-safe yeast and has been approved as a feed additive in China [44]. Therefore, in the large-scale expression of feed enzymes, such as Mel1, Xyn-CDBFV, and RuCelA, crude culture might be dried and used as feed additives without purification. When expressing high-value recombinant proteins, improved yields in mnn11Δ mutants might compensate for the cost of adding extra YE and further purification. In some cases, YE can be replaced by single amino acids, such as glutamate, in the fermentation of mnn11Δ strains, since single amino acids can be easily separated from the secreted proteins. A similar strategy has been applied in the fermentation of Escherichia coli [45], Bacillus subtilis [46], S. cerevisiae [47], Pichia pastoris [48], and CHO cells [49].

The effect of different amino acids on high-yield strains varies. For example, adding arginine or lysine, rather than isoleucine, to the medium of K5 can further increase BadGLA expression [20]. In a synthetic medium, adding tyrosine increased the yield of Plasmodium falciparum merozoite surface protein 3 in P. pastoris by 3.5-fold [50]. Among the six amino acids tested in this experiment, glutamate, arginine, and lysine provided better support for the growth of the mnn11Δ cells. Glutamic acid, arginine, and lysine are among the top six most abundant amino acids in YE [51], which explains the effectiveness of YE in alleviating the growth defect of mnn11Δ mutant and boosting recombinant protein expression. Soy peptone and tryptone, also commonly used organic nitrogen sources, have different proportions of preferred and non-preferred nitrogen compared to YE [52]. For different high-yield strains, selecting a nitrogen source that matches specific amino acid requirements may result in better expression enhancement at a lower cost.

Conclusions

This study demonstrates that glycosylation mutation offers a novel strategy to improve the secretory expression of heterologous proteins in K. marxianus. The mnn11Δ mutant enhances the secretion of both glycosylated and non-glycosylated proteins by upregulating key secretion processes. The mnn11Δ mutant exhibits a significant growth defect in the synthetic medium due to amino acid shortage. Supplementing the medium with low concentrations of organic nitrogen alleviates this defect and improves heterologous protein expression. The amino acid shortage in mnn11Δ mutant is linked to a cell wall defect, and most glycosylation mutants exhibit similar defects. Thus, this strategy of supplementing organic nitrogen sources might be applied to optimize the expression of other glycosylation mutants.