Engineering Pichia pastoris as a Heterologous Platform for Non-Ribosomal Peptide Biosynthesis of the β‑Lactam Intermediate δ‑(l‑α-Aminoadipyl)‑l‑cysteinyl‑d‑valine (ACV) Authors: Shraddha Jogi, Rutuja Jagtap, Yalamanchili Venkata Subrahmanyam, Monika Sharma, Mahesh Kulkarni, Anu Raghunathan Published: 2026 ABSTRACT The production of nonribosomal peptides (NRPs) in non-native hosts remains challenging due to the size and complexity of the biosynthetic enzymes. Here, we report the first successful reconstruction of the δ-(L-α-aminoadipyl)-l-cysteinyl-d-valine (ACV) pathway in Pichia pastoris. By introducing codon-optimized versions of the pcbAB and npgA genes directly into the genome, we were able to rebuild the minimal pathway required for ACV synthesis, which we confirmed using LC–MS/MS. Although the starting strain produced only small amounts of ACV (∼5 ng/mL), yields increased 10-fold after adaptive evolution (∼50 ng/mL) and further improved with precursor supplementation (∼60 ng/mL). Transcriptomic analysis showed that the engineered strain reorganized its metabolism to meet the energy and precursor demands of NRPS activity. Overall, this study suggests that P. pastoris can serve as a practical, proof-of-concept host for complex NRP biosynthesis and lays the groundwork for future engineering of β-lactam pathways. Introduction The escalating global challenge of antibiotic resistance necessitates advanced metabolic engineering strategies to enhance the production of β-lactam antibiotics and their key precursors. NRPs encompass a diverse range of bioactive compounds. Synthesized by complex, multidomain nonribosomal peptide synthetases (NRPSs) incorporating non-proteinogenic amino acids, they undergo diverse post-translational modifications. Low native titers and intricate structural complexity, hinder effective metabolic engineering strategies in native producer organisms. NRPSs facilitate the incorporation of non-proteinogenic amino acids and various post-translational modifications, including methylation, formylation, and epimerization. , Due to the scarcity of nonribosomal peptide production in nature and their structural complexity, large-scale production through biological source purification or chemical synthesis is rarely feasible. These bottlenecks demand a paradigm shift toward robust heterologous platforms capable of enhanced NRP biosynthesis. The ACV tripeptide, a nonribosomal peptide (NRP) and a key precursor for β-lactam antibiotics such as penicillin and cephalosporins, , represents the initial intermediate and rate-limiting step in the biosynthetic pathways of these antibiotics. The δ-(l-α-aminoadipyl)-l-cysteinyl-d-valine (ACV) tripeptide is both the first committed intermediate and the rate-limiting step in β-lactam biosynthetic pathways. Naturally produced by filamentous fungi such as Penicillium chrysogenum and Acremonium chrysogenum, , ACV belongs to the class of nonribosomal peptides (NRPs), whose complex structures and low natural abundance make large-scale isolation or chemical synthesis impractical. In native producers, ACV is rapidly converted to downstream β-lactam intermediates, leaving little opportunity for direct recovery. This reflects a broader challenge in natural product biosynthesis: many bioactive molecules originate from organisms that are difficult to cultivate or genetically manipulate, limiting their availability for mechanistic studies or therapeutic development. The methylotrophic yeast Pichia pastoris offers an attractive alternative production host. It can be cultivated to high cell densities (>400 g/L wet weight) in inexpensive, chemically defined media, and its genetic tractability comparable to Saccharomyces cerevisiae enables stable genomic integration of heterologous biosynthetic genes without continuous selective pressure. ACV synthetase, an NRPS, catalyzes ACV tripeptide biosynthesis through the condensation of l-α-aminoadipic acid, l-cysteine, and l-valine. This process involves the activation and joining of these amino acids to form an ACV tripeptide, accompanied by the epimerization of the l-valine molecule to the d-configuration. Awan et al. (2017) successfully engineered the baker’s yeast Saccharomyces cerevisiae to produce and secrete the antibiotic penicillin, a β-lactam nonribosomal peptide. Pichia pastoris, classified as generally recognized as safe (GRAS), has emerged as a valuable and versatile host organism for producing foreign proteins. Its advantages include rapid growth on defined media, ease of genetic manipulation, quick expression time, and efficient protein processing. , Compared to Saccharomyces cerevisiae, P. pastoris offers higher cell density and increased secreted protein yields. Consequently, it has been used to express a wide range of heterologous proteins, including eukaryotic and integral membrane proteins, as well as genes such as npgA and atX to produce compounds like 6-methylsalicylic acid. This study establishes the methylotrophic yeast Pichia pastoris as a suitable heterologous platform for the de novo production of ACV tripeptide, a critical β-lactam precursor. To engineer this non-native biosynthetic pathway, a strategy integrating advanced synthetic biology and metabolic engineering techniques was designed. This included redesigning the complete gene sequences of pcbAB and npgA genes that encode ACV synthetase and the 4′-phosphopantetheinyl transferases (PPTase). To enhance expression in P. pastoris, codon optimization of pcbAB and npgA resulted in high codon adaptation indices (CAI). GC content and problematic cis-acting sequence motifs (e.g., internal TATA-boxes, chi-sites, cryptic splice sites) were adjusted. These synthetic genes were integrated into the P. pastoris genome to construct stable production strains. Precursor amino acid supplementation and adaptive laboratory evolution were used to increase yields of ACV. Finally, we employed comprehensive transcriptomic analysis using RNA sequencing to dissect, at a molecular level, the complex metabolic reprogramming induced by ACV production, revealing significant differential expression in key pathways for further optimization. This work establishes Pichia pastoris as a viable heterologous chassis for complex nonribosomal peptide production. Methods Design of Synthetic Constructs The synthesis of genes pcbAB and npgA was outsourced to GeneArt (Thermo Fisher Scientific). Codon optimization was done using GeneOptimizer algorithm for host-specific gene design, which employs a multiparametric optimization technique instead of merely substituting codons. This algorithm considers host-specific codon usage biases, codon contexts, GC content, the avoidance of rare codon clusters, the minimization of stable mRNA secondary structures (especially in the 5′ region), and the removal of sequence motifs linked to transcription issues, mRNA instability, or premature termination (such as cryptic splice sites, internal polyadenylation signals, and transcriptional pause sites). Additionally, internal restriction sites that would interfere with downstream cloning were eliminated, while maintaining the original amino acid sequence. The resulting DNA sequences were synthesized de novo and cloned into pPink-LC_A333 vector for downstream integration. Strains and Media Conditions The strains and plasmids used in the present study are listed in . The base strain used in this study is Pichia pink (ade2-), derived from Pichia pastoris GS115 strain. Synthetic pPink_LC_pcbAB and pPink-LC_npgA were maintained in E. coli DH5α strain, cultivated in Luria–Bertani and Terrific broth media. P. pastoris was grown in YPD media (10 g/L yeast extract, 20 g/L peptone and 20 g/L glucose), or in BMGY or BMMY media [10 g/L yeast extract, 20 g/L peptone, 13.4 g/L yeast nitrogen base without amino acids, 0.4 mg/L biotin, 100 mM potassium phosphate buffer (final, pH 6.0) and 10 g/L glycerol (for BMGY) or 10 g/L methanol (for BMMY)]. Pichia adenine-dropout (PAD) agar medium (PichiaPink Media Kit) was used to select the Pichia pastoris_pcbAB strain. Ampicillin (100 μg/mL) and Zeocin (100 μg/mL) were used wherever necessary. Methanol was used for ACV tripeptide extraction from cell pellets. All the molecular grade chemicals and reagents were procured from HiMedia Laboratories, and LC-MS grade chemicals were obtained from Fisher Scientific. Pure ACV standard was procured from bioelsa company (https://bioelsa.com/) and 13C-labeled valine from Sigma-Aldrich was used as internal standard for LC-MS/MS studies. Genomic Integration of Synthetic pcbAB Gene in Pichia Pink Strain 1 The synthetic pPink-LC_pcbAB plasmid was linearized using the SacI restriction enzyme (NEB) for integration at the AOX1 locus of the Pichia pink genome. Transformation of the Pichia pink strain 1 with the linearized pPink-LC_pcbAB plasmid was performed using the electroporation method described in the Pichia pink expression manual (MAN0000717). The transformed cells were plated on PAD agar, and white colonies were selected as positive transformants based on pink-white screening. Genomic DNA from the Pichia pastoris_pcbAB strain was extracted using a Thermo Scientific Yeast DNA Extraction Kit (catalog number: 78870), and positive colonies were confirmed by PCR (primers are listed in Table S4). Design of pPink-LC_npgA_ble Construct The DNA fragments, vector backbone pPink-LC_npgA, and insert fragment PTEF1-PEM7-ble-CYCtt, were obtained from pPink-LC_npgA and pGAPz-alpha B plasmids, respectively. Primers were designed using the NEBuider tool (https://nebuilder.neb.com/) to amplify DNA fragments from the plasmids. Each DNA fragment was amplified by PCR using high-fidelity PrimeSTAR HS DNA Polymerase (Takara, Catalog number: R010A) according to the manufacturer’s protocol. Amplified DNA fragments were extracted using a QIAquick Gel Extraction Kit (Catalog number: 28704) and further purified using the QIAquick PCR Purification Kit (Catalog number: 28104) to remove the residual primers, nucleotides, and enzymes. The Gibson assembly reaction was performed using NEBuilder HiFi DNA Assembly Master Mix (Catalog Number: NEB #E2621) according to the manufacturer’s instructions. Primers are listed in . Glycerol stocks of E. coli_pPink-npgA-ble were stored at −80 °C. Pichia pastoris _ACV Strain Development The assembled plasmid construct, pPink-npgA-ble, was linearized using SpeI (NEB) restriction enzyme for integration at TRP2 locus in the Pichia pastoris_pcbAB genome. The linearized pPink-npgA-ble DNA fragment was extracted and purified using 3 M sodium acetate and 100% ethanol. Electrocompetent cells of Pichia pastoris_pcbAB were transformed using purified linear pPink-npgA-ble cassette according to the protocol (Pichia pink expression system manual (MAN0000717)). The transformed cells were plated on YPD agar selection plates containing Zeocin (100 μg/mL) and the plates were incubated at 30 °C. Glycerol stocks of Pichia pastoris_ACV strains were stored at −80 °C. Induction of pcbAB and npgA Genes in Pichia pastoris _ACV Strain Five ml BMGY medium was inoculated with glycerol stock of the Pichia pastoris_ACV, grown overnight at 30 °C in a shaking incubator at 250 rpm. 100 mL BMGY medium was inoculated in 250 mL baffled shake flask with an overnight culture (OD600 of 0.1). The cells were grown until OD600 = 10–12, harvested and the supernatant discarded. The cell pellet was washed twice with sterile water, resuspended in 10 mL BMMY medium (0.5% methanol as inducer), and grown for 24 h at 30 °C in a shaking incubator set to 250 rpm. Extraction of ACV Tripeptide ACV tripeptide was extracted from cellular extracts based on previously validated methanol-lysis protocol. After 24 h of induction, the cells were harvested and the cell pellet resuspended in 10 mL of 100% methanol. The mixture was vortexed for 30 s with 5 mL acid-washed glass beads and kept on ice 30 s, followed by ten ice and vortex cycle for 30 s each. The mixture was centrifuged for 1h at 4 °C at 7800 rpm. The extract was filtered and collected in fresh 15 mL falcon tubes. LC-MS Analysis The analysis was performed on a Ultraperformance liquid chromatograph (UPLC), a SCIEX ExionLC AD system coupled to a SCIEX 5500 tandem mass spectrometer (MS/MS) with an electrospray ionization source in positive mode with following optimized mass interface parameters: curtain gas, 35 psi; ion source gas 1 (nebulizer), 35 psi; ion source gas 2 (turbo ion spray), 35 psi; ion spray voltage, 4500 V; temperature, 500 °C. An Accucore PFP column (4.6 × 150 mm, 2.6 μ) was used with the mobile phase flow rate 0.4 mL/min and the column oven temperature of 40 °C. 10 μL was injected for a total run time 10 min. Mobile phase A (0.1% formic acid in water), and mobile phase B (0.1% formic acid in acetonitrile) were used with elution gradient (Table S2). Data acquisition and analysis was performed using Analyst and Skyline software version 24.1. During the extraction process, 10 ppb of 13C-valine was used as internal standard in the extraction solvent (100% methanol). The internal standard was used for normalization and the internal standard response across all samples to ensure good extraction efficiencies. A standard calibration curve was generated to calculate the yield of ACV in ng mL–1. A six-point calibration curve for ACV was prepared using standard solutions ranging from 0 to 25 ng mL–1. The curve was constructed by plotting the integrated area versus the concentration of the standards. The calibration curve exhibited a strong linear relationship over the investigated concentration range, with a regression equation of (y = 0.1154x – 0.919) and a coefficient of determination R2 of 0.9921 (Supporting Information, Figure S12). The limit of detection (LOD) and limit of quantification (LOQ) for ACV were determined to be 1 and 2 ppb, respectively, based on LC-MS/MS signal-to-noise ratios (S/N = 3 for LOD; S/N = 10 for LOQ). Adaptive Laboratory Evolution of P. pastoris_ACV Adaptive laboratory evolution was performed by serial transfer ofP. pastoris_ACV strain in 5 mL of BMG (Buffered Minimal Glycerol) medium supplemented with increasing concentration of glycerol, ranging from 5 g/L to 20 g/L. The culture was transferred after every 24 h for total duration of 40 days (∼320 generations), and grown at 30 °C, 250 rpm. Populations were harvested every 24 h and stored in glycerol at – 80 °C. Whole Genome Sequencing of P. pastoris_ACV Evolved Strain Whole genome sequencing of the P. pastoris_ACV evolved strain was performed to identify genomic changes postadaptive laboratory evolution. The genome sequencing was outsourced to Strand Life Sciences. The DNA was extracted using Qiagen DNAesy Mini Kit, and the concentration of DNA was determined using a Qubit fluorometer and the quality was assessed using Tapestation. The xGen DNA Library Prep EZ Kit from IDT was used to perform library preparation. Adapter-ligated libraries were purified and PCR-enriched to generate final sequencing-ready libraries. Library quality and fragment size distribution were assessed using Tapestation, and library concentration was quantified using Qubit prior to sequencing. The libraries were normalized, pooled, and sequenced using NovaSeq X plus. Comparative Mutation Analysis of P. pastoris_ACV Evolved Strain in Reference to the Unevolved Strain Paired-end Illumina sequencing reads (evolved strain FASTQ files) and a reference genome FASTA file (unevolved strain) were provided as input. The raw sequencing reads were evaluated for quality by using FastQC. Low-quality bases and adapter sequences were removed by using Trimmomatic. Trimmed reads were again assessed using FastQC to confirm an improvement in read quality after trimming. The reference genome was indexed to enable efficient alignment and variant calling, which included indexing with BWA, creation of a FASTA index using SAMtools, and generation of a sequence dictionary using GATK. High-quality trimmed reads were aligned to the reference genome by using BWA-MEM. The SAM file was converted to a BAM file using SAMtools. Genetic variants (SNPs and indels) were detected using GATK HaplotypeCaller with the ploidy set to 1, appropriate for Pichia pastoris haploid genome. The identified variants were filtered based on read depth (greater than 50) and quality metrics to retain high-confidence variants. Gene predictions were generated from the reference genome by using AUGUSTUS to produce a GFF annotation file. Using the reference genome FASTA file and the generated GFF annotation, a custom variant annotation database was built using SnpEff. The filtered VCF file was annotated using SnpEff to determine the functional effects of variants, including synonymous, missense, frameshift, and upstream/downstream variants. The annotated variants were mapped to the reference genome annotation to identify the affected GeneID, ProteinID, and corresponding protein product, enabling the interpretation of the potential biological impact of each variant. RNA Sequencing Sample Preparation Cultures of Pichia pink and P. pastoris_ACV (2 Biological Replicates each) were grown and sampled from the 24 h induction phase and centrifuged at 1500g for 5 min. The supernatant was discarded and cell pellet was snap-frozen in liquid nitrogen. The samples were then shipped to Eurofins Genomics India Pvt. Ltd. (Bangalore, India) for RNA isolation and transcriptome sequencing. Total RNA was extracted from the yeast cells using Quick RNA MiniPrep Plus Kit (Zymo Research) as per manufacturer’s instruction. Quality and quantity of the extracted RNA samples were checked on NanoDrop followed by Agilent Tape Station using High Sensitivity RNA ScreenTape. The paired end sequencing libraries were prepared from the QC passed RNA samples using NEBNext Ultra II Directional RNA Library Prep Kit for Illumina (NEB) as per manufacturer’s instructions. The PCR enriched libraries were purified using AMPureXP beads and analyzed on 4200 Tape Station system (Agilent Technologies) using high sensitivity D1000 Screen tape as per manufacturer’s instructions. RNA quality was confirmed using RIN (RNA Integrity Number) values (Table S7). TPM normalization (Transcripts Per Million) was used for transcript quantification and differential expression analysis accounting for both gene length and sequencing depth. Each sample was sequenced to a depth of approximately 20–25 million paired-end reads, sufficient for capturing genome-wide expression profiles in Pichia pastoris. The sequenced raw data of the samples were processed to obtain high quality clean reads using Trimmomatic v0.39 to remove adapter sequences, ambiguous reads and low-quality reads. The high-quality (quality threshold >25) paired-end reads were used for reference based read mapping which was done using STAR (v.2.7.10a) with default parameters. FeatureCounts (version 2.0.3) was used to quantify the read numbers mapped on each gene. Differentially expressed genes were identified using DESeq2, with standard thresholds (adjusted p < 0.05, log2 fold-change ≥ ± 1). RNA Sequencing Data Analysis RNA sequencing data analysis was performed using the iDEP 2.0 web-based tool (https://ge-lab.org/idep/) [accessed on 29 March, 2025]. iDEP (integrated Differential Expression and Pathway analysis) is a web-based tool that integrates 63 R/Bioconductor packages and comprehensive annotation and pathway databases. Differential gene expression analysis, functional enrichment, and pathway analysis were performed to identify genes with 2 or greater fold change in expression values. Hierarchical clustering based on Pearson distance of log2-fold-change values was used for gene ordering in the heatmap. The gene expression values were log-transformed and row-scaled using Z-score normalization. Gene set enrichment analysis (GSEA) is a rank-based algorithm for pathway enrichment. The gene set database used for annotation was the KEGG pathway database for fungi. GSEA considered all genes and is independent of arbitrary cutoffs, enhancing the robustness. Results and Discussion Gene Sequence Design and Optimization Given the complexity in size and structure of the ACV synthetase gene pcbAB (∼11 kb) from P. chrysogenum and the phosphopantetheinyl transferase npgA gene from Aspergillus nidulans, a rational codon-optimization approach was followed to maximize translation efficiency, avoid rare codons, and reduce mRNA secondary structures known to hinder expression in Pichia. CAI optimization was performed using GeneOptimizeralgorithm of the GeneArt technology platform. This integrates codon usage, mRNA secondary structure, GC content, and sequence motifs to enhance gene expression in the selected host. The codon-adaptive index (CAI) of the native pcbAB gene from P. chrysogenum was 0.80 and that of native npgA from A. nidulans was 0.77 (Figure C, Native). This translated to a CAI value of 0.66 and 0.65 (Figure C, recombinant) for expression of the pcbAB and npgA genes, respectively, in P. pastoris. Codon optimization was done to match the codon usage bias in P. pastoris to increase translational efficiency and resulted in higher CAI values of 0.83 and 0.87 (Figure C, synthetic) for optimized pcbAB and npgA genes indicative of effective adaptation for an optimal host. Design and analysis of synthetic gene sequences optimized for Pichia pastoris. (A) Synthetic genes pcbAB and npgA for ACV biosynthesis. (B) Reaction scheme for ACV synthesis. (C) CAI of pcbAB and npgA genes. (D) GC content of pcbAB and npgA genes. Beyond codon adaptation, the guanine-cytosine (GC) content of both genes was adjusted to a target of 48% in the Pichia host from native organism values of >50% (Figure D) to ensure it resided within a range conducive to efficient expression. Similarly, the npgA gene optimization process prioritized the removal of regions characterized by excessively high (>80%) or low (<30%) GC content, maintaining an average of 48%. Pairwise sequence analysis revealed percent identity between native sequence and codon-optimized sequence of pcbAB gene and npgA genes as 74.67% and 74.49% at nucleotide level, respectively, and 100% at protein level (MSA Supplement data). The final gene sequence design mitigated the influence of cis-acting sequence motifs, including elimination of internal TATA-boxes, chi-sites, ribosomal entry sites, AT-rich or GC-rich stretches, RNA instability motifs, repeat sequences, and potentially disruptive RNA secondary structures. Construction of Genome Engineered Pichia pastoris Strain for ACV Biosynthesis by Expression of ACVS and PPTase To engineer Pichia pastoris for ACV production, the optimized gene constructs pcbAB gene and the npgA gene were sequentially integrated into the Pichia pink genome, an ade2 auxotrophic strain of P. pastoris (Figure ). Both the pcbAB and npgA genes were placed under the control of the methanol-inducible AOX1 promoter, which is widely used in Pichia pastoris for tightly regulated, high-level expression of heterologous genes. Development of Pichia pink strain for production of ACV tripeptide. The strain development strategy is represented (created in https://BioRender.com). Specifically, the pcbAB gene was inserted into the AOX1 locus, creating the P. pastoris pcbAB strain. This integration strategy resulted in the generation of a MutS phenotype (methanol utilization slow) due to disruption of the AOX1 gene. Methanol-based induction of AOX1 promoter-driven genes was supported by another alcohol oxidase AOX2 in P. pastoris. The npgA gene was inserted into the TRP2 locus of the P. pastoris_pcbAB strain, generating the final P. pastoris_ACV strain. Successful integration of both genes was confirmed by PCR amplification using gene-specific primer pairs (), and the targeted insertions of pcbAB and npgA at the AOX1 and TRP2 loci were verified (). The integration loci were further confirmed through whole-genome sequencing using a hybrid platform of Illumina and nanopore sequencing technologies. The resulting sequences were analyzed using BLAST, and the alignment of pcbAB and npgA sequences against the assembled sequence of the P. pastoris_ACV strain confirmed their presence in the genome (MSA Supplemental Data). Hereafter, this genome-engineered strain of P. pastoris will be referred to as P. pastoris_ACV. To confirm gene expression, we performed RT-qPCR analysis of heterologous genes pcbAB and npgA. In the engineered P. pastoris_ACV strain, Ct values for pcbAB (22.81) and npgA (23.65) were markedly lower than those in the parental strain (AOX1 locus: 34.89; TRP2 locus: 33.19), indicating notable transcriptional activity (Table S8, Figure S10). Lower Ct values correspond to higher transcript abundance, consistent with strong expression driven by the methanol-inducible AOX1 promoter. These results indicated successful pathway reconstitution for ACV synthesis in P. pastoris. Heterologous Production of ACV Tripeptide Was Confirmed in P. pastoris _ACV Comparative growth rate analysis in YPD medium demonstrated a statistically significant difference (p value = 0.01) between P. pink (0.237 ± 0.005 h–1) and P. pastoris_ACV (0.191 ± 0.017 h–1) (Table S5) indicating a potential metabolic burden and differential physiology of the heterologous expression of ACVS and PPTase in the engineered strain. This is further substantiated by the fact that there is no decrease in final biomass yields (Figure S9) on external addition of ACV at concentrations detected by LC-MS (Figure A). ACV begins to exert inhibitory effects on cell growth at ≥ 10 ppm with the strongest decrease in yield observed between 100–1000 ppm (Figure S9). The concentrations were calculated using a 6-point calibration curve with a regression equation of (y = 0.1154x - 0.919) and a coefficient of determination R2 of 0.9921 (Figure S12). The optimal cellular growth phase for methanol induction and the addition of non-proteinogenic l-α-aminoadipic acid (AAA) for optimal yields of ACV were determined. ACV production levels were probed by harvesting cells at different phases of logarithmic growth (Figure A). Maximal ACV titers were observed when induction was initiated at an OD600 of 10 (DCW = 1.91g/L), with cells in early mid logarithmic phase (Figure B). ACV biosynthesis may be potentially due to higher intracellular concentrations of cysteine and valine (Figure D). ACV tripeptide production in different cellular growth phases. (A) The growth of P. pastoris_ACV indicating early log, mid log and late-log phase (Logistic Monod Model). (B) The ACV titers using cells for induction from different growth phases (n = 9). One-way ANOVA (p-value = 0.002) followed by post hoc Dunnett’s test was performed for statistical analysis. Liquid chromatography-tandem mass spectrometry (LC-MS/MS) with multiple reaction monitoring (MRM) was used to analyze the cell extracts of P. pastoris_ACV. LC-MS/MS analysis of the ACV standard identified a distinct peak with an elution time of 7.0 min with product ions at m/z 144 and 158, corresponding to its MRMs. A peak with a similar m/z of product ions and retention time as the ACV standard was observed in the spectra from the induced cultures. This signal was absent in extracts from uninduced cultures, validating the functional reconstitution of the ACV biosynthetic pathway. RT-q-PCR analysis verified the successful reconstitution of the pathway (). The engineered P. pastoris_ACV strain yielded a maximum of 4.95 ng/mL (0.01 μM) ACV tripeptide (paired, two-tailed t test, p-value = 0.04) (Figure A). Production of ACV tripeptide. Bar plot (A) represents production of ACV tripeptide in induced (0.5% methanol) and uninduced condition (paired, two-tailed t test, p-value = 0.04; n = 3). (B) Optimization of methanol inducer concentration for improved ACV production (one-way ANOVA, p-value = 0.026, followed by post hoc Dunnett’s test; n = 3). (C) The impact of addition of l-α-aminoadipic acid at distinct growth phases on ACV biosynthesis, C1 – Growth phase (BMGY+AAA), C2 – Induction phase (BMMY+AAA), C3 – Both growth and induction phase (BMGY+AAA and BMMY+AAA) (one-way ANOVA, p-value = 0.0001, followed by post hoc Dunnett’s test; n = 8). (D) Effect of precursor supplementation on ACV tripeptide production (unpaired, two-tailed t test, p-value = 0.0004; n = 3) (E) ACV titers of P. pastoris_ACV strain after adaptive laboratory evolution (paired, two-tailed t test, p-value = 0.0148; n = 3). Optimization of Heterologous Production of ACV Tripeptide in Pichia pastoris Optimization of methanol concentration for induction of optimal ACV in the engineered P. pastoris showed dose dependent relationship between inducer concentration and product titer. The highest ACV titer (4.95 ± 1.8 ng/mL) was observed at inducer concentration of 0.5% methanol (Figure B). One-way ANOVA test (p-value = 0.026) was used for statistical analysis, followed by post hoc Dunnett’s test. The groups. 0.5% vs 0.125% (p-value = 0.4144) methanol concentration does not show significant difference, however other groups, 0.5% vs 1% (p-value = 0.0278), 0.5% vs 1.5% (p-value = 0.0032) and 0.5% vs 2% (p-value = 0.0003) showed significant difference. The methanol range used (0.125–2%) was selected based on legacy data, and methanol is reported to be toxic to the cells beyond 3% concentration. − To investigate effect of l-α-aminoadipic acid (AAA) addition in different phases of growth and induction on ACV biosynthesis, P. pastoris_ACV cultures were cultivated under three experimental conditions: C1 (AAA supplemented in growth phase, BMGY medium), C2 (AAA supplemented in induction phase, BMMY medium), and C3 (AAA supplemented sequentially in both growth and induction phases, BMGY & BMMY media). Comparative analysis of ACV titers revealed 2.5-fold increase in ACV concentration (∼4.47 ng/mL) in C3 relative to the baseline production observed in C1 (∼1.73 ng/mL) (Figure C, one-way ANOVA, p-value = 0.0001, followed by post hoc Dunnett’s test). This suggests a synergistic effect of AAA addition in both growth and induction phases on ACV biosynthetic pathway efficiency. Precursor amino acids (AAA, l-cysteine, and l-valine) supplementation in the growth media showed a statistically significant increase (t test, p value = 0.00035) in ACV titer from 4.95 ± 1.8 ng/mL (only AAA) to 60.08 ± 5.1 ng/mL (0.15 μM) (Figure D, Unpaired, two-tailed t test, p-value = 0.0004). On adaptive laboratory evolution under the selection pressure of the substrate glycerol, this engineered P. pastoris_ACV strain yielded a maximum of 50.18 ng/mL (0.12 μM) ACV (Figure E, paired, two-tailed t test, p-value = 0.0148). ACV biosynthesis is energetically demanding and dependent on adequate pool of l-cysteine, l-valine, and l-α-aminoadipate, long-term adaptation on glycerol was expected to enrich for strains capable of sustaining higher intracellular precursor pools and improved ATP generation. Hence, glycerol was selected as the evolutionary pressure as Pichia can efficiently assimilate carbon during respiratory growth and potentially favor the emergence of metabolic variants with enhanced flux distribution and precursor regeneration. Following ∼312 generations of adaptive laboratory evolution, the evolved P. pastoris_ACV strain displayed higher ACV titers even though pcbAB and npgA transcripts were reduced. This also indicates the stability of integration of these heterologous genes in Pichia pastoris for ACV and NRP production. Although the evolved strain showed higher Ct values (Figure S10) for pcbAB and npgA, indicating lower transcript levels, it nevertheless produced more ACV (Figure E). The observed increase in ACV production despite reduced transcript levels of pcbAB and npgA suggests that adaptive laboratory evolution improved global regulatory rewiring, pathway efficiency, and precursor supply. Because transcript abundance frequently exhibits only a weak correlation with protein abundance and cannot accurately predict protein levels or enzymatic activity due to layers of post-transcriptional and translational regulation, this apparent disconnect is expected. , Adaptive evolution potentially fine-tuned enzyme activity, precursor balance, or metabolic routing to channel flux more effectively toward ACV synthesis. This apparent decoupling between mRNA levels and product titer suggests that adaptive evolution helped achieve higher productivity through network-level optimization (also seen in the whole transcriptome analysis, Figure ) instead of transcriptional amplification alone. The observed increase in ACV production therefore reflects a systems-level enhancement in metabolic efficiency driven by glycerol-based selection. Mutational Analysis of P. pastoris _ACV Day 40 Strain (Evolved) Strain Comparative genomic analysis identified 81 mutations in the evolved strain relative to the unevolved strain, including 27 coding sequence mutations and 54 intergenic variants (Table ). Notably, multiple independent mutations were concentrated within the tRNA splicing endonuclease complex, with additional mutations observed in genes associated with the translational machinery, including a 60S ribosomal protein (RPL2). This clustering suggests strong selective pressure on host translational capacity during adaptive evolution. The subunit of the tRNA splicing endonuclease is involved in splicing of tRNA, which is a critical step in the tRNA maturation for intron-containing tRNAs. These mutations probably may have increased the translational efficiency and cellular tRNA pools. Such translation modulation may be a possible reason for increased ACV titer in the evolved strain. 1Mutations in P. pastoris_ACV Day 40 Evolved StrainTypeNumberChromosomeGene IDProductConservative in-frame deletion128198656Subunit of the tRNA splicing endonucleaseDisruptive in-frame insertion128198655Cell-wall l-asparaginase IIMissense mutation62 and 35 mutations in 8198656; 1 mutation in 8199621Subunit of the tRNA splicing endonuclease; Hypothetical proteinFrameshift mutation1028198656Subunit of the tRNA splicing endonucleaseStop-lost & Splice region mutation228198656Subunit of the tRNA splicing endonucleaseSynonymous mutation71, 2 and 38196722; 8198656; 8200145; 819984360S ribosomal protein L2; Subunit of the tRNA splicing endonuclease; Tubulin folding factor D; Putative ion transporter Mutations in P. pastoris_ACV Day 40 Evolved Strain Expression of ACV synthetase (pcbAB), a large nonribosomal peptide synthetase, likely imposes a substantial translational burden on the host cell. Mutations affecting tRNA maturation and ribosomal function may therefore enhance the availability of mature tRNAs and improve translation efficiency under this burden. A mutation was also identified in l-asparaginase II, which may influence intracellular nitrogen metabolism and amino-acid pools. Because α-aminoadipate, one of the three precursors required for ACV biosynthesis, is derived from amino-acid metabolism, such changes may contribute to improved precursor availability (Figure ). Rewiring of translation and amino-acid metabolism post adaptive laboratory evolution enhances ACV tripeptide production. Transcriptomic Profiling during ACV Production To delineate the system-level impact of pcbAB and npgA overexpression on the transcriptome, we performed strand-specific RNA-seq. The high-resolution quantification of gene expression or transcriptome (5078 genes) is presented as a clustered heatmap (Figure A), demonstrating a clear partitioning of samples based on the genotype. The P. pastoris_ACV (left column) exhibited a distinct expression profile compared to the control, reflecting a notable transcriptional adaptation to the introduced genetic modifications. The heatmap confirmed the successful overexpression of pcbAB and npgA by their pronounced enrichment (log2FC = [pcbAB = 10.32, npgA = 11.63], FDR-adjusted p = [pcbAB = 0.0008, npgA = 6.297e-09]) in the ACV strain. Differential gene expression analysis (Figure B) identified genes with significant and biologically relevant changes. Transcriptome analysis. (A) Differential expression of 5078 genes for both Pichia ACV (left column) and Pichia pink (right column) (B) Volcano plot for 5078 differentially expressed genes. Transcriptomic analysis of the ACV-producing P. pastoris strain revealed a significant upregulation of ADE2 ( PAS_CHR3_0085 ), encoding phosphoribosyl aminoimidazole carboxylase, a key enzyme in de novo purine biosynthesis. This response reflects the adenine auxotrophy complementation in the P. pastoris_ACV strain. Of the 5078 quantified transcripts, differential expression analysis revealed a discrete partitioning of regulatory response: 530 genes were uniquely upregulated (log2FC ≥ 2, FDR-adjusted p < 0.05), while 588 were downregulated. Transcriptome profiling focusing on genes exhibiting ≥5-fold differential expression, revealed strategic metabolic reprogramming. Hierarchical clustering of significantly enriched pathways, visualized as a dendrogram (Figure A), revealed a distinct topology, with a high-confidence cluster of upregulated pathways, including “fatty acid beta-oxidation” (p < 0.001), “fatty acid catabolic process” (p < 0.005), “IMP metabolic process” (p < 0.004), and “IMP biosynthetic process” (p < 0.004), indicating a coordinated transcriptional activation to enhance intracellular acetyl-CoA and purine nucleotide pools. The clustering, characterized by short branch lengths, indicates a high degree of gene overlap and functional relatedness among the precursor-generating pathways. (A) Dendrogram representing top 15 significant pathways. (B) Enrichment analysis of top 20 significant pathways [FDR < 0.05, Log2Fold ≥ 2]. (C) Pathway enrichment of top 15 significant pathways [FDR < 0.05, Log2Fold ≥ 5]. Amino acid precursors of ACV were metabolically regulated as anticipated. For l-α-aminoadipate (AAA), PAS_chr1–4_0153 (l-aminoadipate-semialdehyde dehydrogenase; log2FC = 0.79; p = 0.003) was upregulated, while PAS_chr4_0795 (homoaconitate hydratase; log2FC = −0.98; p = 0.001) was downregulatedsuggesting AAA accumulation through suppression of lysine branchpoints. Sulfite reductase genes PAS_chr4_0369 (log2FC = 1.45; p = 4.1 × 10–8) and PAS_chr3_1084 (log2FC = 1.41; p = 3.5 × 10–8) were strongly induced, supporting enhanced cysteine biosynthesis and sulfur assimilation, whereas PAS_chr1–4_0253 (sulfate adenylyltransferase; Log2FC = −0.69; p = 0.02) was repressed, suggesting nonlimiting sulfate supply. For l-valine, PAS_chr4_0248 (isoleucine transaminase; Log2FC = 0.92; p = 0.001) was upregulated, whereas PAS_chr4_0228 (2-acetolactate pyruvate-lyase; log2FC = −0.61; p = 0.015) was downregulated, potentially conserving pyruvate for other precursor-generating processes. Network-based enrichment analysis illustrated strong interconnectivity of “Fatty acid beta-oxidation” and “IMP biosynthesis” (green nodes, p < 0.02 and p < 0.004, respectively) While the fatty-acid beta oxidation and IMP biosynthesis are not directly involved in the ACV tripeptide biosynthesis, they indirectly support the metabolic rewiring. Upregulation of fatty-acid beta oxidation is commonly associated with increased generation of reducing equivalents and ATP generation, indirectly supporting ACV tripeptide biosynthesis. Furthermore, the observed downregulation of the vitamin B6 metabolic process (blue node, p < 0.05) presents an intriguing aspect of the engineered metabolic reconfiguration. The implied coordinated transcriptional responses driven by genes within metabolic pathways (Figure ) potentially impacted by the experimental conditions, suggesting hypotheses about shifts in metabolic flux, substrate utilization, and biosynthetic capabilities. Collectively, the transcriptional changes indicate coordinated reallocation of metabolic resourcesenhancing precursor supply, reducing competing fluxes, and prioritizing the energetically demanding ACV biosynthetic process. This study establishes Pichia pastoris as a viable chassis for complex nonribosomal peptide synthesis and successfully presents reconstructing of a complex ACV nonribosomal peptide synthesis pathway using optimized synthetic genes. The use of P. pastoris was motivated by its high-density growth, efficient post-translational processing, and the flexibility of its AOX1-based inducible system. This tight control minimizes plasmid and metabolic burden during biomass accumulation, ensuring reproducible induction once methanol is added. While AOX1 regulation provides excellent control, it can also introduce methanol-related stress (methanol toxicity above an optimal 0.5% concentration); future iterations could incorporate mixed-promoter or constitutive/inducible hybrids to balance growth and expression efficiency. Key limitations identified include suboptimal precursor pools of AAA, l-cysteine, l-valine (12-fold increase on supplementation). Adaptive laboratory evolution with glycerol selection pressure increased ACV yields 10-fold. Rewiring of translation and amino acid metabolism postadaptive laboratory evolution could be a possible reason for increased ACV titers in the evolved strain. Transcriptomics revealed significant host metabolic readjustments, including upregulation of energy and nucleotide supply pathways, likely to support the demands of the heterologous pathway. These findings highlight P. pastoris as a viable chassis but underscore the need for further targeted metabolic engineering, particularly addressing precursor (AAA) supply, optimizing sulfur and NADPH metabolism, increasing valine production, reducing resource competition from growth programs and fine-tuning fatty acid and nucleotide pools, to enhance ACV tripeptide production. In summary, this study establishes Pichia pastoris as a proof-of-concept chassis for heterologous production of complex nonribosomal peptides. While current ACV titers are modest, the integration of synthetic gene design, precursor supplementation, and transcriptomic analysis provides a foundation for subsequent optimization cycles. Future work incorporating the study of effect of precursor amino acids in combination, l-2-aminoadipic acid + l-Cysteine and or l-valine, on ACV tripeptide production, metabolic flux modeling, adaptive evolution sequencing, and targeted metabolomics will aim to enhance yield and dissect regulatory mechanisms. Engineering the modules of ACV synthetase using “XUT” approach could be used to generate novel, tailored peptides and various metabolic engineering strategies could be used in future to improve the ACV tripeptide yields in Pichia pastoris. Supplementary Material