Physiological and Metabolic Challenges of Flocculating Saccharomyces cerevisiae in D-Lactic Acid Fermentation Under High-Glucose and Inhibitory Conditions

Abstract

Lactic acid is an important biobased chemical widely used in the production of biodegradable plastics, food, and pharmaceuticals. However, the application of flocculant Saccharomyces cerevisiae remains limited in addressing stresses such as high-glucose and inhibitor-rich conditions derived from biomass, particularly in D-lactic acid (D-LA) production. This study investigates two genetically engineered S. cerevisiae F118 strains, ΔCYB2::LpDLDH and ΔPDC1::LpDLDH, for D-LA production under high-glucose and inhibitor-stress conditions that mimic lignocellulosic hydrolysates in shake-flask fermentation. At 150 g/L glucose, ΔCYB2::LpDLDH produced 41 ± 0.73 g/L D-LA, whereas ΔPDC1::LpDLDH yielded 80 ± 1.78 g/L, corresponding to 27% and 53% of the theoretical yield, respectively. Calcium carbonate (CaCO₃) supplementation enhanced glucose consumption and strengthened flocculation in ΔPDC1::LpDLDH. The addition of 5% inhibitory chemical compounds (ICCs) consisting of furfural, HMF, and weak acids redirected carbon flux in ΔCYB2::LpDLDH toward D-LA formation and reduced ethanol byproduct accumulation. Transcriptomic analysis revealed the upregulation of stress-response genes (HOG1, TPS1) and cell-wall remodeling genes (CRH1, SCW10) in response to high-glucose stress. The strongly flocculent F118ΔCYB2::LpDLDH strain exhibited greater tolerance to weak acids and furfural than the weakly flocculent F118ΔPDC1::LpDLDH strain. Metabolomic profiling indicated that under inhibitor stress, carbon flux was diverted from the TCA cycle toward lactate synthesis to maintain redox balance. These findings highlight the multifaceted benefits of flocculation in enhancing strain robustness and D-LA productivity under harsh fermentation environments, providing insights for developing resilient yeast platforms for lignocellulosic bioprocessing.

1. Introduction

The budding yeast Saccharomyces cerevisiae is one of the most well-established microbial hosts in biotechnology. Its natural ability to thrive in high-sugar and high-ethanol environments, combined with its Generally Recognized as Safe (GRAS) status and ease of genetic manipulation, has made it a central platform for producing bioethanol and various value-added chemicals [1]. Despite its robustness, S. cerevisiae encounters multiple challenges during industrial fermentations, including high osmolarity, ethanol accumulation, low pH, and exposure to toxic compounds [2], all of which constrain productivity and yield. Therefore, developing yeast strains that withstand such stresses is essential for achieving cost-efficient bioprocessing [3].

Lignocellulosic biomass serves as a sustainable feedstock for biorefineries; however, its pretreatment generates inhibitory chemical compounds (ICCs) that hinder microbial growth and metabolism. These inhibitors include furan aldehydes (furfural and 5-hydroxymethylfurfural), phenolic derivatives (syringaldehyde and vanillin), and weak organic acids (acetic and formic acids) [4]. Such compounds impair S. cerevisiae growth, ethanol production, and cellular integrity, and collectively disrupt fermentation performance [5,6,7]. When combined with high sugar concentrations and ethanol toxicity, these ICCs create complex multi-stress environments that decrease overall fermentation robustness.

Yeast flocculation—the reversible aggregation of cells into multicellular clusters—is mediated by cell-wall glycoproteins encoded by the flocculin (FLO) gene family. This phenotype is common among wild and industrial S. cerevisiae isolates, although its intensity varies with genotype and environmental conditions [8]. Flocculation has been associated with improved tolerance to environmental stresses. For instance, deletion of FLO1 reduced acetic acid tolerance in the industrial strain SPSC01 [9], whereas strong flocculation increased survival under furfural stress [10]. These protective effects are attributed to the formation of microenvironments within flocs, where outer cells absorb or detoxify inhibitors, thereby shielding inner cells and enhancing collective survival [11]. Moreover, flocculation facilitates cell sedimentation and biomass reuse, offering operational advantages in repeated-batch and continuous fermentations [12]. Consequently, flocculation serves both as an evolutionary survival mechanism and as a valuable industrial trait.

Lactic acid is a key platform chemical with applications in food, pharmaceuticals, and biodegradable polymers, such as poly(lactic acid) (PLA) [13]. While lactic acid bacteria remain the primary producers, yeast offers advantages, including higher acid tolerance and immunity to phage contamination. By heterologously expressing lactate dehydrogenase (Ldh), S. cerevisiae can redirect pyruvate flux from ethanol toward D-lactic acid (D-LA). Targeted genomic integration at loci such as PDC1 or CYB2 enables metabolic rerouting, although ethanol formation often remains significant under glucose-rich conditions [14,15,16].

The wild-type S. cerevisiae F118 strain has previously demonstrated robustness against lignocellulosic inhibitors [11], making it a suitable host for lactic acid production. In this study, we used an existing engineered strain [17] to express a D-lactate dehydrogenase (LpDLDH) gene at either the CYB2 or PDC1 locus, generating two D-LA-producing strains with distinct flocculation behaviors—one strongly flocculant and the other weakly flocculant. Despite limited genomic modification, both retained the capacity to form ethanol under glucose-rich conditions. We evaluated their performance under stresses relevant to lignocellulosic fermentations, including osmotic stress (10–15% glucose), weak acids, furfural, and low pH, with minimal calcium carbonate neutralization to mimic cost-effective industrial operation.

The objective of this study is to elucidate the impact of genomic integration location on D-LDH and D-LA production, as well as stress tolerance, in strains with CYB2 and PDC1 integrations. It evaluates fermentation performance under high-glucose and inhibitory conditions to find the most robust and efficient production host. The results provide theoretical insights into how genomic context influences metabolic regulation and offer practical guidance for engineering yeast strains optimized for organic acid production. Comparing two engineered S. cerevisiae strains with different flocculation behaviors shows that strong flocculation enhances tolerance, decreases neutralization needs, and supports biomass recycling for efficient, sustainable lignocellulosic biorefineries. These findings deliver both fundamental insights into yeast physiology and practical strategies for strain development.

2. Materials and Methods

2.1. Strains and Medium

The wild-type strain used for fermentation in this study was S. cerevisiae F118 (NBRC268), a flocculant strain obtained from the National Biological Resource Center (NBRC) culture collection in Tokyo, Japan. The primary strains, F118∆CYB2::LpDLDH and F118∆PDC1::LpDLDH, are derived from S. cerevisiae F118 and were developed in our previous study [17]. Yeast extract peptone dextrose medium (YPD), containing 100 g/L glucose, 10 g/L yeast extract, and 20 g/L peptone, was used to cultivate the yeast. Solid media for S. cerevisiae included 15 g/L agar added to the liquid medium. The medium was treated with the antibiotic geneticin/G418 (Cat. No. ant-gn-1, InvivoGen, San Diego, CA, USA).

2.2. Batch Fermentation

Fermentation was carried out in a small-scale 100 mL Erlenmeyer flask containing 12 mL of YPD100 medium (100 g/L glucose), inoculated with a single colony of the yeast strain grown on YPD agar that had been incubated for three to four days. The experimental method was adapted from Ishida et al. [14] and Long et al. [18], with several modifications implemented to suit the specific objectives of this study. To prepare the seed culture, cultures were incubated for one to two days at 30 °C with shaking at 150 rpm. The initial density for seed cultures was adjusted to an OD_600nm of 1 or approximately 0.4 g/L of dry cell weight. Depending on the experiment, the seed was inoculated with 12 mL of YPD, containing 50, 100, 150, or 300 g/L of glucose, to initiate the primary culture of batch fermentation. The final D-LA titers and pH values obtained for each condition are summarized in Supplementary Tables S1 and S2. The effect of an inhibitory chemical compound (ICC) was examined by adding 5% of five chemicals: 15 mM Acetic acid, 10 mM Formic acid, 15 mM Furfural, 1.25 mM Levulinic acid, and 2.5 mM Hydroxymethylfurfural (HMF). One normal sodium hydroxide solution was used as a neutralizer at the start of cultivation until the media pH reached 5–6. The cultures were then grown at 30 °C with shaking at 90 rpm until the glucose in the medium was depleted or no further decrease in glucose concentration was observed. Glucose concentration was monitored daily by sampling and analyzed using high-performance liquid chromatography (HPLC). Statistical analysis was performed in R (v4.4.2) using two-way ANOVA to evaluate the effects of strain and glucose concentration on D-LA production. Tukey’s HSD test was applied for multiple comparisons at a significance level of p < 0.05. Data are presented as mean ± SD from three biological replicates.

2.3. Batch Fermentation with Calcium Carbonate Supplementation

The fermentation was carried out in a small-scale 100 mL Erlenmeyer flask containing 12 mL of YPD50 (50 g/L glucose) to prepare a seed culture, inoculated with a single colony of the yeast strain grown on YPD agar. The colony grows for three to four days at 30 °C. The seed was inoculated with 12 mL of YPD containing 150 g/L with and without calcium carbonate. The concentration of CaCO₃ was 0.4% (w/v). The experimental method was adapted from Ishida et al. [14] and Baek et al. [15] with several modifications implemented to suit the specific objectives of this study. The cultures were then grown at 30 °C with shaking at 90 rpm until the glucose in the medium was depleted. Glucose concentration was monitored daily by sampling and analyzed using high-performance liquid chromatography (HPLC). The final D-LA titers and pH values obtained for each condition are summarized in Supplementary Tables S1 and S2. Experiments on this cultivation were conducted with three replicates per strain. The mean and standard deviation (SD) were calculated for each dataset. Data were analyzed using RStudio (v4.4.2). Two-way analysis of variance (ANOVA) was applied to determine the effects of strain and CaCO₃ supplementation (with or without) on final glucose and D-LA concentrations. Interaction effects between strain and CaCO₃ were also evaluated.

2.4. Measurements of Fermentation Products

Periodically, fermentation samples were collected and centrifuged for 5 min at 4 °C and 14,000× g. After collection, the supernatants were filtered into HPLC vials using a 0.45 µm polytetrafluoroethylene (PTFE) filter (Merck Millipore, Carrigtwohill, Ireland). Fermentation products were analyzed by high-performance liquid chromatography (HPLC) with a refractive index detector (RID-10A, Shimadzu, Kyoto, Japan), which measured various fermentation-related substances, including glucose, D-LA, glycerol, acetic acid, and ethanol. A 5 mM sulfuric acid solution in ultra-pure water served as the eluent in a Coregel-87H column (7.8 mm ID × 300 mm, Transgenomic Inc., New Haven, CT, USA) operated at 80 °C for 40 min at a flow rate of 0.6 mL/min.

2.5. Measurement of Cell Growth

The optical density could not be used to evaluate the flocculent F118 strains. Therefore, dry cell weight was used to measure yeast growth. Following batch fermentation, each sample’s culture was harvested by centrifugation at 14,000× g for 5 min at 4 °C. The cell pellets were then washed twice with sterile water. After that, the pellets were freeze-dried in a Labconco vacuum-drying machine Labconco Freezone 2.5 (Kansas City, MO, USA) overnight at −80 °C. The dry cell weight was then measured.

2.6. Cell Morphological Observation

A BIOREVO BZ-9000 digital microscope (Keyence, Osaka, Japan) was used to observe the morphological changes in the cells caused by glucose presence, calcium carbonate (CaCO₃) supplementation, and an inhibitory chemical substance in the F118 strain. The objective lens was set at ×100 magnification.

2.7. Gene Expression Analysis by Quantitative PCR

Since the cells show their highest metabolic activity during the mid-exponential phase of aerobic fermentation, yeast cells were harvested at this stage. This was achieved by centrifuging 0.5–1 mL of cell suspension at 8000× g for 10 min at 4 °C. Before RNA extraction, the pellets were stored at −80 °C after a single wash with cold, sterile water. As recommended by the manufacturer, the NucleoSpin RNA kit (Macherey-Nagel, Duren, Germany) was used for extraction, and rDNase treatment during column digestion minimized DNA contamination. The ReverTra Ace qPCR RT Master Mix with gDNA remover (Toyobo, Osaka, Japan) was used for reverse transcription, and the resulting cDNA served for quantitative PCR (qPCR). Using 10 ng/μL of diluted cDNA and 0.4 μM of each forward and reverse primer, the KOD SYBR qPCR Mix (Toyobo) was employed to measure gene expression levels. A Mx3005P system (Agilent Technologies Ltd., Tokyo, Japan) was used for qPCR. Before the amplification cycles—comprising an initial denaturation at 95 °C for 10 min, followed by 60 cycles of 30 s at 95 °C, 1 min at 55 °C, and 1 min at 72 °C for elongation—the PCR program began with an initial denaturation at 95 °C for 10 min. The final phase involved one cycle of 1 min at 95 °C, followed by 30 s at 55 °C, and then 30 s at 95 °C.

Because of its similar C_t values and consistent expression across all samples, the ACT1 gene was used as an internal reference gene. Using calculations with the MxPro QPCR v4.10 (Agilent Technologies Japan, Ltd.), the expression levels of target genes were determined as fold changes with the formula 2^−∆∆Ct, where ∆∆C_t = ∆C_t sample − ∆C_t control and ∆C_t sample = C_t sample − C_{t ACT1}. According to Supplementary Table S3, the primer sequences used in the analysis were designed using the NCBI Primer-BLAST tool (version 2.5.0) (https://www.ncbi.nlm.nih.gov/tools/primer-blast/, accessed 11 November 2025) with a target product size of approximately 100 bp.

2.8. Extraction of Internal Metabolites

The cells were harvested during each strain’s logarithmic phase, which varies in timing from their fermentation profile. All fermentation cultures were collected by centrifugation at 11,000× g for 5 min to ensure consistency across experiments, as flocculating cells were difficult to sample. The cells were immediately frozen in liquid nitrogen to stop metabolic activity and then stored at −80 °C until extraction. Metabolites were extracted using 3 mL of preheated 75% ethanol, containing 10 µg/L of ribitol as an internal standard for GC/MS analysis and 20 µg/L (+)-10 camphorsulfonic acid for LC-MS analysis. After thorough vortexing, the mixture was heated to 95 °C for 3 min, then quickly cooled on ice. Cells were centrifuged at 14,000× g for 3 min to remove cell debris. Subsequently, 500 µL of the supernatant was transferred to a new sterile tube. The lysate was evaporated overnight in a Labconco vacuum-drying machine and stored at −80 °C until ready for further analysis.

2.9. GC/MS Analysis for Untargeted Metabolites

Untargeted metabolites were analyzed using gas chromatography–mass spectrometry (GC/MS). Derivatization, involving oximation and trimethylsilylation, was performed before GC/MS analysis. The first derivatization involved adding methoxiamine hydrochloride (100 µL of 20 mg/mL in pyridine) and incubating the freeze-dried lysate from the previous step at 30 °C for 90 min. For the second derivatization, MSTFA (N-methyl-N-(trimethylsilyl) trifluoroacetamide) was added at a volume of 50 µL. The mixture was incubated at 37 °C for 30 min before injection into the GC/MS [19]. The Shimadzu GC/MS QP2020, paired with an AOC-20i autosampler (Kyoto, Japan), features an inert Cap 5 MS/NP column (0.25 mm × 30 m × 0.25 μm) from GL Science Inc. (Tokyo, Japan). A one µL sample was analyzed in split mode at an injection temperature of 230 °C. Helium served as the carrier gas at a flow rate of 1.12 mL/min and a linear velocity of 39 cm/s. The column temperature was initially set to 80 °C and maintained for 2 min, then increased at 15 °C/min to 330 °C, where it was held for 6 min. The ion source was maintained at 200 °C, and the transfer line at 250 °C. Electron ionization was set at 0.93 kV. Spectra were recorded at 10,000 scans per second over a mass-to-charge (m/z) range of 85–500. Retention index calibration used a standard mixture of terminal alkanes (C₈–C₄₀) to facilitate tentative compound identification. Raw GC/MS data were converted to .abf files with an Abf converter (Reifycs Inc., Tokyo, Japan) and analyzed using MSDIAL (Riken, Yokohama, Japan). Detected peaks were identified and labeled by matching their retention times and mass spectrometry data against the RIKEN GL-Science database, a publicly available metabolite library.

2.10. LC/MS Analysis for Carbon Central Metabolites

The freeze-dried lysate was dissolved in 50 µL ultrapure water, then centrifuged at 10,000× g for 20 min to remove debris. The lysate was then transferred into vials for analysis of carbon central metabolites using liquid chromatography–tandem triple quadrupole mass spectrometry (LC/QqQ/MS). An Agilent 1260 high-performance liquid chromatography instrument, coupled with a 6460 Triple Quad MS (Santa Clara, CA, USA) and the Mastro2 column (3 μm particle size, 150 mm × 2.1 mm) from Shimadzu were used. For the mobile phase, 10 mM tributylamine with 15 mM acetate as mobile phase A and methanol as mobile phase B were used. The mobile-phase flow rate was set at 0.2 mL/min, and the temperature was 37 °C. The analysis was performed using a gradient elution program for mobile phase B. The gradient started at 0% for 1 min, then increased at 10%/min to 15%, 7%/min to 50%, and 5%/min to 100%. The system was held at 100% for 1 min before being re-equilibrated by a rapid decrease to 0% in 30 s. A 3 μL sample was injected. Instrument parameters were set to a DL temperature of 250 °C, a heat block temperature of 40 °C, a drying gas flow of 15 L/min, and a nebulizer ga 2 L/min [19]. Intracellular metabolites were detected in negative ion mode. Raw LC/QqQ/MS data were converted to .abf files using an Abf converter and analyzed with MRMPROBS (Riken, Yokohama Japan) [20].

2.11. Multivariate Analysis

Multivariate analysis was performed using MetaboAnalyst 6.0 (https://www.metaboanalyst.ca) (accessed 20 August 2025) to conduct Principal Component Analysis (PCA) without data transformation. This method aimed to emphasize both differences and similarities within the dataset. The relative metabolite intensities were normalized with internal standards, ribitol for GC/MS and (+)-10 camphorsulfonic acid for LC/QqQ/MS, which served as explanatory variables. To minimize the masking effect caused by variability among metabolites, the data were automatically rescaled.

In addition to PCA, an ANOVA test was performed for each metabolite to evaluate significant differences between the two strains under ICC conditions and CaCO₃ supplementation from the F118ΔPDC1::LpDLDH strain. Metabolites with p < 0.05 were considered significantly different.

Heatmap analysis was performed to visualize the overall differences in metabolite profiles among samples. The normalized metabolite intensity data were used to construct a clustered heatmap using the R software environment (version v4.4.2). The heatmap was generated using the pheatmap package in R (version 1.10.13), and color gradients represent the relative abundance levels of each metabolite. Significant metabolites identified through ANOVA followed by False Discovery Rate (FDR) correction (adjusted p < 0.05) were highlighted in the heatmap to emphasize strain- and condition-specific metabolic changes.

3. Results

3.1. D-LA Production at Different Glucose Levels

We initially assessed the potential of two genetically modified yeast strains, S. cerevisiae F118ΔCYB2::LpDLDH and F118ΔPDC1::LpDLDH, for D-LA production under high glucose levels (150 g/L and 300 g/L). As shown in Figure 1, at 150 g/L glucose, D-LA production reached 41 ± 0.73 g/L for F118ΔCYB2::LpDLDH and 80 ± 1.78 g/L for F118ΔPDC1::LpDLDH, corresponding to 27% and 53% of the theoretical yield, respectively. This production was significantly different, as measured by the Student’s t-test (p-value < 0.0001). Conversely, at 300 g/L glucose, both strains exhibited poor substrate utilization, with F118ΔCYB2::LpDLDH consuming only about half of the available glucose and F118ΔPDC1::LpDLDH consuming even less. These results suggest that very high glucose levels induce significant osmotic and metabolic stress, which limits glucose uptake and fermentation efficiency. Even though both strains were incapable of completely consuming glucose, their capabilities were significantly different, as measured by the Student’s t-test (p-value < 0.001).

The PDC1-disrupted strain exhibited an increased D-LA yield, likely due to decreased ethanol production and a greater carbon flux into pyruvate breakdown. In contrast, the CYB2-disrupted strain produced less D-LA and more ethanol, but was more efficient at consuming glucose, suggesting a greater tolerance to osmotic stress. These results highlight a trade-off between product yield and physiological resilience, which is strongly influenced by the location of the LpDLDH expression cassette. To explore this further, both strains were tested across a broader range of glucose levels (50–150 g/L) to assess their tolerance to increasing osmotic stress [21].

At moderate glucose levels, the F118ΔCYB2::LpDLDH strain showed a significant increase in D-LA productivity (Figure 2a). The D-LA yield nearly doubled as glucose increased from 50 g/L to 150 g/L, demonstrating that this strain can tolerate osmotic stress up to 150 g/L without major growth inhibition. Statistical analysis revealed no significant difference in overall D-LA production between the two strains (p-value = 0.489). There is a strong effect with increasing glucose concentration (p-value < 0.0001), but when comparing the strains, the effect depended on the strain (p-value < 0.001). D-LA production from the F118ΔPDC1::LpDLDH strain, which had a lower efficiency, producing only 31.41 ± 0.46 g/L D-LA with a 39% yield under 150 g/L glucose conditions (Figure 2b), which is much lower than the 80 g/L observed in the previous batch at the same sugar level. Therefore, subsequent gene expression studies focused on the more resilient F118ΔCYB2::LpDLDH strain.

To identify genes responding to high-glucose stress, qPCR was performed during fermentation, with a focus on pathways involved in D-LA production, ethanol metabolism, and stress adaptation. The genes examined included those involved in D-LA formation (LpDLDH), ethanol synthesis (PDC1), cell-wall remodeling (CRH1, SCW10, SCW11, DSE2), cell-wall integrity and resistance (SED1, HSP150, YGP1, TOS6), flocculation (FLO5), osmotic-stress regulation (HOG1, SSK1), stress protection (TPS1), and iron homeostasis (AFT1). As shown in Figure 3, the transcriptional profiles demonstrated significant upregulation of stress-related genes at higher glucose concentrations. The TPS1 gene, which controls trehalose synthesis and overall stress resistance, was activated under both 100 g/L and 150 g/L glucose conditions. At the same time, HOG1, a key kinase in the high-osmolarity glycerol (HOG) pathway, was upregulated only at 150 g/L. Similarly, cell-wall remodeling genes CRH1 and SCW10 increased at 150 g/L, indicating reinforcement of cell-wall strength under osmotic stress. These responses correlated with decreased culture turbidity at high-glucose levels (Figure 2a), suggesting reduced flocculation and cell aggregation.

In contrast, the F118ΔPDC1::LpDLDH strain, which has reduced flocculation due to the downregulation of FLO [22], showed no noticeable change in aggregation as glucose concentrations increased. This supports a mechanistic link between flocculation behavior and osmotic-stress tolerance. Measurements of dry-cell weight across the three glucose levels showed no significant differences (Supplementary Figure S1), indicating that the variations in D-LA production are mainly metabolic rather than growth-related. Although the study examined a limited set of genes, these results establish an initial connection between metabolic output and stress-response regulation under high-glucose conditions. The upregulation of HOG1, TPS1, and cell-wall genes suggests that osmotic stress triggers adaptive responses that may improve flocculation stability and D-LA productivity in the robust F118ΔCYB2::LpDLDH strain. Future research should expand these findings by conducting transcriptome-wide analyses, targeted overexpression of protective genes, and adaptive evolution experiments to further enhance osmotic tolerance and D-LA production. Combining these strategies with rational genetic design will be key to optimizing yeast cell factories for lignocellulosic bioprocesses [23].

3.2. Calcium Carbonate Supplementation Effect on D-LA Production

To reduce the growth inhibition observed in S. cerevisiae F118∆PDC1::LpDLDH, caused by LpDLDH disruption that limits ethanol production and cell growth [17], CaCO₃ was added at the start of fermentation as a neutralizer. In industrial D-LA production, adding excess CaCO₃ is a common method to prevent the pH from dropping too much during lactic acid buildup [24]. During fermentation, CaCO₃ reacts with lactic acid to produce calcium lactate and carbonic acid, which subsequently decompose into water and carbon dioxide [25]. This buffering reaction maintains the fermentation broth at a pH of around 5.0, providing a more stable environment for yeast metabolism [26]. CaCO₃ addition maintained a slightly higher final pH in both strains. With CaCO₃ supplementation, the final pH of the CYB2-integrating strain increased to 3.0, while that of the PDC1-integrating strain was approximately 2.16, compared with about 2.0 in their respective control fermentations.

The results revealed that both strain type and CaCO₃ had a significant effect on the residual glucose at the end of fermentation (p < 0.001). There was also a significant interaction, indicating that the effect of CaCO₃ depends on the strain (p < 0.001). As shown in Figure 4a, the fermentation performance of F118∆CYB2::LpDLDH was not much different with CaCO₃ supplementation. Conversely, for F118∆PDC1::LpDLDH, the D-LA yield did not increase, but complete glucose consumption was achieved after about 48 h of fermentation (Figure 4b). This suggests that CaCO₃ improved metabolic stability and glucose utilization, although it did not directly enhance D-LA productivity. Interestingly, in the F118∆PDC1::LpDLDH strain, where disrupting the pyruvate decarboxylase pathway limits ethanol production and cell growth, the presence of CaCO₃ significantly reduced growth inhibition and increased overall lactic acid accumulation. This indicates that CaCO₃ not only prevents medium acidification but also indirectly helps maintain intracellular pH balance. At the transcriptional level, adding CaCO₃ significantly elevated LpDLDH expression in F118∆CYB2::LpDLDH by 983-fold, while most other genes remained unchanged (Figure 5). Among the few upregulated genes were YGP1, FLO5, and LpDLDH, which are associated with cell-wall integrity and flocculation. Under these conditions, increased cell aggregation and turbidity were clearly visible, demonstrating that CaCO₃ supplementation enhanced flocculation behavior not only in the naturally strong-flocculant F118∆CYB2::LpDLDH but also in the initially weak-flocculant F118∆PDC1::LpDLDH, as shown in Figure 4.

The relationship between CaCO₃ addition, cell-wall stability, and flocculation aligns with previous findings in S. cerevisiae, where FLO1, FLO5, and FLO11 govern cell–cell adhesion and aggregation [27,28]. Specifically, FLO5 has been recognized as a key determinant of flocculation [8] and of cell-wall assembly and organization [28], particularly in wine yeast strains.

Overall, while CaCO₃ supplementation did not increase D-LA yield, it promoted complete substrate utilization and enhanced cell flocculation, likely through the combined effects of pH stabilization, upregulation of FLO5 and YGP1, and improved cell-wall integrity. These findings suggest that mild buffering can influence cell physiology, support strong aggregation, and sustain metabolic activity under acidic fermentation conditions.

3.3. Effect of Inhibitory Chemical Compounds on D-LA Production

To evaluate the effect of lignocellulosic-derived inhibitors on D-LA production, 5% of an inhibitory chemical compound mixture (ICCs) was added to the fermentation medium. Furfural, the most toxic component among ICCs, is known to impair S. cerevisiae growth, metabolism, ethanol production, and cell morphology by inducing oxidative and membrane damage. To mitigate medium acidification and enhance strain performance, sodium hydroxide was added at the start of fermentation to maintain the initial pH between 5.0 and 6.0. In previous studies, the parental F118 recombinant strain producing L-LA was found to tolerate up to 20% ICCs without pH control, although LA productivity declined markedly [29].

A two-way ANOVA revealed a significant interaction between strain and ICC on D-lactic acid production (p < 0.001), indicating that the effect of ICC depended on the strain. ICC significantly increased D-LA in F118ΔCYB2::LpDLDH (20.42 ± 5.44 g/L→ 32.82 ± 1.14 g/L, p < 0.001) (Figure 6a) and significantly decreased D-LA in F118ΔPDC1::LpDLDH (35.3 ± 1.33 g/L→ 18.4 ± 0.2 g/L, p < 0.001) (Figure 6b). Consequently, the strain producing higher D-LA reversed under ICC treatment: F118ΔPDC1::LpDLDH > F118ΔCYB2::LpDLDH at control conditions, and F118ΔCYB2::LpDLDH > F118ΔPDC1::LpDLDH in the presence of ICC.

Supplementation with 4 g/L CaCO₃ substantially improved this strain’s performance, enabling full glucose consumption within 48 h and increasing the D-LA titer from 18.42 ± 0.2 g/L to 42.36 ± 0.96 g/L (Figure 6c). This enhancement demonstrates the buffering and detoxification effects of CaCO₃ under weak-acid and aldehyde stress. Microscopic images (Figure 6, bottom panels) depict cellular morphology and flocculation behavior after 72 h of fermentation. The ΔCYB2 strain exhibited strong cell aggregation and compact floc structures, whereas the ΔPDC1 strain showed mostly dispersed cells with reduced adhesion. CaCO₃ buffering enhanced cell integrity and aggregation, indicating a stabilizing effect against ICC-induced acid stress and supporting sustained D-LA production.

Carbon flux analysis revealed that F118ΔCYB2::LpDLDH redirected pyruvate toward D-LA rather than biomass or ethanol production (Figure 7a), as reflected by reduced biomass yield and ethanol formation. This metabolic shift, coupled with higher ICC tolerance, suggests that F118ΔCYB2::LpDLDH can efficiently reallocate carbon resources under stress, while F118ΔPDC1::LpDLDH relies on external neutralization to sustain productivity. Despite its growth limitation, CaCO₃-supplemented F118ΔPDC1::LpDLDH achieved the highest D-LA titer among all tested conditions (Figure 7b).

To further elucidate transcriptional responses to ICCs, genes associated with carbon metabolism, stress adaptation, and energy regulation were analyzed by qPCR in cells cultured in YPD100 medium, with or without 5% ICCs. The target genes included those involved in glycolysis and ethanol metabolism (HXK1, PFK1, ENO1, ADH1, LpDLDH), lactic acid formation (LpDLDH), lactate transport (JEN1), stress and cell-wall remodeling (SLT2, RLM1, SCW11, HSP150, YGP1), and mitochondrial energy metabolism (NDI1). As illustrated in Figure 8, several genes were significantly upregulated in the presence of ICCs, including PFK1 and ENO1 (glycolysis), ADH1 (ethanol metabolism), LpDLDH (lactate production), and JEN1 (lactate export). Stress-related genes (SLT2, RLM1, YGP1, SCW11) also showed increased expression, indicating activation of the cell-wall integrity (CWI) and general stress response pathways. Notably, NDI1 expression was markedly induced in F118ΔCYB2::LpDLDH, suggesting enhanced mitochondrial NADH oxidation and energy turnover, which may contribute to increased stress tolerance. In F118ΔPDC1::LpDLDH, strong LpDLDH induction under CaCO₃ supplementation indicated partial recovery of lactate flux, compensating for the loss of pyruvate decarboxylase activity and growth inhibition.

To further elucidate the molecular responses of engineered S. cerevisiae strains to ICCs, we examined transcriptional and metabolomic alterations during fermentation. Genes associated with glycolysis, ethanol metabolism, D-LA production, stress adaptation, and mitochondrial energy regulation were analyzed via qPCR in cells cultivated in YPD100 medium, with or without 5% ICCs. The genes selected for analysis included HXK1, PFK1, ENO1, ADH1, and PDC1 (carbon metabolism); LpDLDH (D-LA production); JEN1 (lactate transport); SLT2, RLM1, SCW11, HSP150, and YGP1 (cell-wall integrity and stress response); and NDI1 (mitochondrial energy metabolism).

As depicted in Figure 8, several glycolytic genes (PFK1, ENO1) and ethanol-related genes (ADH1) were significantly upregulated under ICC exposure, indicating an enhanced flux through upper glycolysis and redox rebalancing. The strong induction of LpDLDH and JEN1 suggested that the lactate synthesis and export system remained active even under inhibitor stress, supporting continued D-LA accumulation. Stress-responsive genes (SLT2, RLM1, YGP1, SCW11) were also markedly upregulated, reflecting the activation of the cell-wall integrity (CWI) and general stress response pathways.

Notably, NDI1 expression was strongly induced in F118ΔCYB2::LpDLDH, suggesting reinforcement of mitochondrial NADH oxidation to sustain ATP generation and mitigate redox imbalance. Conversely, F118ΔPDC1::LpDLDH exhibited enhanced LpDLDH expression, particularly under CaCO₃ supplementation, indicating a partial recovery of lactate flux that compensated for the loss of pyruvate decarboxylase activity and associated growth inhibition.

Given the tight coupling between transcriptional and metabolic networks, metabolomic profiling was performed to assess differential metabolite accumulation in response to ICC exposure. As shown in Figure 9, F118ΔCYB2::LpDLDH exhibited a larger number of significantly altered metabolites, while F118ΔPDC1::LpDLDH displayed fewer but more specific changes, consistent with its lower metabolic turnover rate.

In F118ΔCYB2::LpDLDH, several amino acids (aspartic acid, histidine, serine, threonine, glycine) were significantly elevated, suggesting enhanced nitrogen assimilation and stress-driven amino acid turnover. Increased levels of adenine, nicotinamide, and tricarboxylic acid (TCA) cycle intermediates (citric, isocitric, and fumaric acids) indicate activation of nucleotide biosynthesis and reinforcement of oxidative metabolism to maintain redox balance. In contrast, decreases in glucose, mannose, ribose, and trehalose indicated glycolytic suppression and a diversion of carbon toward maintenance metabolism and osmolyte production—strategies commonly employed to withstand weak-acid and aldehyde stress.

In F118ΔPDC1::LpDLDH, the metabolomic profile showed accumulation of succinic acid and its anhydride derivative, pointing to a redirection of pyruvate metabolism through mitochondrial pyruvate carrier (MPC) as a gate transporter toward reductive branches of the TCA cycle. Simultaneous increases in citric and isocitric acids, coupled with declines in sugar alcohols (meso-erythritol, galactitol), suggest a shift from carbohydrate storage toward energy-generating intermediates. These results imply that in the absence of PDC1, the strain compensates for the blocked ethanol pathway by enhancing TCA cycle flux and maintaining NAD⁺/NADH homeostasis under ICC-induced oxidative stress.

As shown in Figure 10, the observed increase in sedoheptulose-7-phosphate (S7P) under +ICC conditions supports the activation of the pentose phosphate pathway (PPP). This metabolic shift indicates enhanced NADPH regeneration, which is essential for the reductive detoxification of furfural and other aldehyde inhibitors [23]. Reinforcement of both the PPP and the TCA cycle contributes to maintaining a balanced supply of reducing equivalents (NADH/NADPH), thereby mitigating oxidative stress and sustaining metabolic activity during inhibitory fermentation.

The integrated transcriptional and metabolomic datasets revealed a complex adaptive response to ICC-induced stress. Upregulation of glycolytic and stress-protection genes, coupled with activation of PPP- and TCA-associated redox pathways, indicates a coordinated reprogramming of central metabolism (Figure 10). To statistically assess these metabolic differences, a two-way analysis of variance (ANOVA) was performed, with strain (F118ΔCYB2::LpDLDH and F118ΔPDC1::LpDLDH) and treatment (−ICC vs. +ICC) as independent factors. This approach evaluated the main effects and their interaction on intracellular metabolite levels determined by GC-MS and LC-MS (Supplementary Table S4). Post hoc comparisons were conducted using Tukey’s honestly significant difference (HSD) test, with significance defined at p < 0.05 (Figure 11).

In total, 25 metabolites were significantly influenced by ICC exposure, highlighting two predominant response patterns. First, samples under ICC stress formed distinct clusters from those in YPD, reflecting broad metabolic reprogramming driven by inhibitory compounds. Most glycolytic intermediates (F16P, PEP, G6P) and adenylate nucleotides (ATP, ADP, AMP) decreased, indicating suppressed energy generation and carbon flux. Second, genotype-specific responses were evident: F118ΔCYB2::LpDLDH maintained a more balanced metabolite distribution, whereas F118ΔPDC1::LpDLDH exhibited pronounced depletion of amino acids and glycolytic intermediates—consistent with redox and energy imbalance resulting from PDC1 disruption. In contrast, several TCA-cycle metabolites (succinate, aconitate, oxaloacetate) remained elevated, suggesting compensatory activation of mitochondrial metabolism to preserve redox homeostasis. Collectively, these findings demonstrate that inhibitory stress exerts a dominant metabolic influence across genotypes, yet F118ΔCYB2::LpDLDH sustains greater intracellular stability, underscoring its robustness as a production strain for D-LA fermentation under inhibitory conditions.

To evaluate how pH buffering by CaCO₃ influences metabolic flux during D-LA fermentation, intracellular metabolites in S. cerevisiae F118ΔPDC1::LpDLDH were quantified under cultures with (+CaCO₃) and without (−CaCO₃) supplementation (Figure 12). There were metabolites that are not shown in Figure 12; other metabolites (F16P, MEP, DXP, and PGA) changed significantly between the two conditions (p-value < 0.01). In addition, there are five metabolites were also significantly changed (p-values < 0.05): lactate, OXA, succinate, ADP, and PEP. The metabolomic map revealed that CaCO₃ markedly modulated carbon flow through glycolysis, the pentose phosphate pathway (PPP), and the tricarboxylic acid (TCA) cycle.

CaCO₃ buffering increased the intracellular levels of key glycolytic intermediates, including glucose-6-phosphate (G6P), fructose-1,6-bisphosphate (FBP), and phosphoenolpyruvate (PEP), suggesting enhanced glycolytic throughput under stabilized pH conditions. The elevated levels of G6P and 6-phosphogluconate (6-PGL) in the PPP indicate activation of NADPH-producing reactions that support antioxidant defense and biosynthetic reduction potential. This effect is consistent with the neutralizing role of CaCO₃ in mitigating intracellular acidification, thereby relieving feedback inhibition on glycolytic enzymes and allowing continued ATP production.

Under conditions of +CaCO₃, a consistent increase in iso-citrate and succinate was observed, indicating stimulation of the TCA cycle. The rise in oxaloacetate and malate further suggests that buffering maintains metabolic continuity between glycolysis and the TCA cycle through the pyruvate-to-acetyl-CoA node. These increases suggest that CaCO₃ supplementation facilitates smoother energy generation and NADH reoxidation, thereby counteracting the acid stress that would otherwise restrict oxidative metabolism during unbuffered fermentation. In the absence of buffering (−CaCO₃), the accumulation of acidic end products, such as D-LA, lowers intracellular pH and limits enzyme activity in both glycolysis and the TCA cycle. The observed metabolite restoration under +CaCO₃ thus reflects improved redox homeostasis, as the buffering system prevents proton overload and promotes continuous NADH oxidation through the TCA cycle. This redox balancing effect aligns with the previously observed transcriptional upregulation of NDI1 and LpDLDH under buffered conditions, highlighting a synergistic link between pH control and metabolic adaptation. Collectively, these results demonstrate that CaCO₃ supplementation stabilizes intracellular pH, sustains glycolytic flux, enhances PPP-derived NADPH generation, and strengthens TCA-cycle activity. This integrated response maintains energy production and redox balance, thereby supporting robust D-LA fermentation even under high-acid conditions. The metabolic stabilization provided by CaCO₃ complements the genetic modifications in PDC1-deficient strains, enabling improved tolerance and productivity in lignocellulosic hydrolysate-based bioprocesses.

4. Discussion

The overall performance of the engineered S. cerevisiae strains under ICC stress conditions demonstrates multifaceted adaptation, including cell wall remodeling, redox regulation, and the dynamic rerouting of central carbon metabolism. The integration of physiological, transcriptional, and metabolomic data offers a comprehensive understanding of how these processes collectively sustain D-LA productivity in inhibitory and acidic environments. At the cellular level, the pronounced flocculation phenotype observed in F118ΔCYB2::LpDLDH significantly enhances tolerance against ICC-derived inhibitors. Flocculated cell aggregates limit the diffusion of weak acids and aldehydes across the plasma membrane, thereby maintaining intracellular homeostasis and mitigating oxidative stress [9,10]. The upregulation of FLO5, YGP1, and SCW11 in response to ICC exposure supports activation of the cell-wall integrity (CWI) pathway, leading to β-glucan and mannoprotein cross-linking that reinforces the outer cell wall. This structural adaptation minimizes osmotic and oxidative damage while fostering microenvironmental buffering and cooperative nutrient exchange among aggregated cells. These findings are consistent with recent studies in other yeast systems. Sae-Tang et al. [30] demonstrated that introducing ScFLO1 into Pichia pastoris significantly enhanced tolerance to lactic acid stress, resulting in a 1.6-fold increase in specific growth rate and a 2.6-fold increase in D-LA titer compared to non-flocculent controls. Transcriptomic analysis further revealed upregulation of lactate transporter and iron homeostasis genes, indicating that flocculation not only forms a physical barrier but also triggers adaptive transcriptional programs that facilitate metabolite export and redox balance. Similarly, Stewart et al. [31] highlighted that flocculation is not merely a sedimentation trait, but a dynamic, stress-responsive process influenced by calcium-dependent lectin–mannose interactions and cell-surface charge regulation, which can enhance survival under acid, ethanol, or osmotic stress. These parallels support the notion that flocculation represents an evolutionarily conserved protective mechanism that can be rationally exploited to improve organic acid fermentation performance. At the metabolic level, ICC exposure induced extensive reprogramming of carbon flux. In F118ΔCYB2::LpDLDH, the depletion of hexose sugars coupled with the accumulation of tricarboxylic-acid (TCA) intermediates—citric, isocitric, and fumaric acids—indicates partial TCA-cycle inhibition and redirection of pyruvate toward D-LA formation. This metabolic shift facilitates cytosolic NADH reoxidation while reducing mitochondrial ROS generation, consistent with the observed induction of NDI1, which encodes the alternative NADH dehydrogenase. Concurrent activation of amino acid biosynthesis and the pentose phosphate pathway (PPP) provides NADPH for antioxidant defense and macromolecular repair, thereby representing a dual mechanism for maintaining redox and biosynthetic balance.

In contrast, the deletion of LpDLDH in F118ΔPDC1::LpDLDH eliminated the primary decarboxylation pathway from pyruvate to acetaldehyde, thereby redirecting carbon flux through the reductive branch of the TCA cycle. The significant accumulation of succinic and isocitric acids suggests compensatory NADH oxidation via this reductive pathway. This metabolic reconfiguration, coupled with robust LpDLDH expression, ensures continuous NAD⁺ regeneration and supports D-LA synthesis, even under elevated inhibitor stress. However, the lack of intrinsic pH control rendered this strain more vulnerable to acidification, a limitation that was mitigated by CaCO₃ supplementation. CaCO₃ not only neutralized excess protons but also sequestered furfural and related aldehydes through carbonate adduct formation, thereby restoring both metabolic activity and D-LA yield.

These findings reveal an integrative tolerance mechanism wherein (i) flocculation establishes a protective physical barrier and localized buffering microenvironment, (ii) CaCO₃ supplementation maintains extracellular pH stability and chemically detoxifies inhibitory aldehydes, and (iii) metabolic rerouting through D-LA synthesis and PPP activation regenerates NADH/NADPH pools essential for redox homeostasis and antioxidant defense. The convergence of these processes enables S. cerevisiae F118 derivatives to sustain ATP generation, preserve intracellular redox balance, and protect macromolecular integrity during the fermentation of lignocellulosic hydrolysates. This coordinated stress-response network mirrors principles observed in P. pastoris [30] and native brewing yeasts [31], highlighting the universal value of flocculation-based tolerance strategies. Comparable physiological outcomes were also reported by [32], who achieved neutralizer-free lactic acid production in S. cerevisiae through adaptive evolution and multiplexed redox-pathway optimization, underscoring that efficient proton management and NADH recycling are central to acid tolerance.

Collectively, these insights suggest that future strain-engineering strategies should integrate flocculation control, exporter enhancement, metal-ion and redox regulation, and self-buffering capacity to minimize reliance on external neutralizers. Such a holistic design approach represents a forward-looking framework for constructing next-generation, self-neutralizing yeast biocatalysts capable of high-titer, high-tolerance D-LA production from lignocellulosic biomass under industrially relevant inhibitory conditions.

5. Conclusions

This study compared the influence of different genomic integration sites of LpDLDH (CYB2 vs. PDC1 loci) on D-LA production and stress tolerance in engineered S. cerevisiae under industrially relevant conditions. The results revealed a clear trade-off between productivity and robustness: F118ΔCYB2::LpDLDH exhibited superior tolerance to osmotic and inhibitory stress, maintaining growth and glucose consumption even without pH neutralization, whereas F118ΔPDC1::LpDLDH showed reduced tolerance under the same conditions. However, supplementation with a mild neutralizer (0.4% CaCO₃) restored its performance and enabled the highest D-LA titer under inhibitory stress. These findings reflect the metabolic consequences of PDC1 disruption, which redirects pyruvate flux from ethanol toward D-LA, enhancing productivity but increasing redox and energy imbalance. In practice, the CYB2-integrating strain provides a robust foundation for high-glucose fermentation, while PDC1 disruption becomes advantageous under buffered or controlled conditions. Future improvements using the CRISPR-Cas9-based genome editing technique in F118ΔCYB2::LpDLDH to simultaneously eliminate ethanol-related pathways could further optimize carbon flux, minimize byproduct formation, and advance the development of high-tolerance, self-neutralizing yeast biocatalysts for sustainable D-LA production.