Heterologous expression of hydroxylase coupled with pathway optimization enables highly efficient ʟ-DOPA production in Corynebacterium glutamicum Authors: Mengzhen Jia, Ran Duan, Yan Jiang, Xiaohe Chu, Xiaoyu Wu, Hui Yang, Bangce Ye, Bin Zhang Published: 2026 ABSTRACT ʟ-3,4-dihydroxyphenylalanine (ʟ-DOPA), the direct precursor of the neurotransmitter dopamine, is a cornerstone therapeutic agent for Parkinson's disease. In this study, Corynebacterium glutamicum was used as a microbial chassis to systematically evaluate its potential as a cell factory for ʟ-DOPA production. The strain was initially evaluated for ʟ-DOPA tolerance to confirm its suitability for production. Key hydroxylases, HpaB and HpaC, were introduced, and promoter and ribosome-binding site (RBS) optimization was applied to establish a de novo ʟ-DOPA biosynthetic pathway. To minimize flux toward competing branches, enzymes encoded by pheA and trpE2 were deleted, and fermentation conditions were optimized to balance cell growth with product formation. Promoter engineering was further employed to enhance the expression of Cgl0226, Cgl0218, and Cgl2101, thereby increasing the availability of the precursor ʟ-tyrosine for ʟ-DOPA biosynthesis. Additionally, deletion of competing shikimate pathway enzymes (encoded by qsuB and qsuD) combined with enhanced expression of aroA and aroC elevated carbon flux through the shikimate pathway. The final engineered strain, KD20, achieved an ʟ-DOPA titer of 710.9 ± 13.2 mg/L in shake-flask fermentation. Upon scale-up to a 5 L bioreactor under optimized conditions, the final titer reached 4605.58 ± 21.73 mg/L, representing the highest ʟ-DOPA yield reported in C. glutamicum. These results highlight the robust production capacity of C. glutamicum and underscore its strong potential for ʟ-DOPA biosynthesis. Introduction ʟ-3,4-dihydroxyphenylalanine (ʟ-DOPA), the immediate precursor of the neurotransmitter dopamine, serves as a cornerstone therapy for Parkinson's disease (PD) and finds broad applications in pharmaceutical and chemical industries [1,2]. PD is characterized by the progressive degeneration of dopaminergic neurons in the substantia nigra, resulting in reduced brain dopamine levels [3]. ʟ-DOPA can cross the blood–brain barrier and is decarboxylated in vivo to dopamine, effectively alleviating motor dysfunction [4]. Since high-dose clinical trials conducted by George Cotzias and colleagues in 1968 confirmed its efficacy, ʟ-DOPA was approved by the U.S. FDA in 1970 and subsequently commercialized worldwide by Roche in 1973 [5,6]. Beyond Parkinson's disease, ʟ-DOPA has shown potential therapeutic effects in amblyopia, hepatic encephalopathy, schizophrenia, heart failure, and bone fracture repair [7,8]. With the global population aging, PD continues to rise, driving a rapid increase in ʟ-DOPA demand. Current annual consumption approaches 250 metric tons, with a market value exceeding USD 100 billion, and projections indicate that China alone will have over 8 million patients by 2050, highlighting the urgent need for efficient and sustainable manufacturing strategies [[9], [10], [11]]. Consequently, green biomanufacturing of ʟ-DOPA has become a major research focus. Traditionally, ʟ-DOPA is produced via plant extraction, chemical synthesis, enzymatic catalysis, or microbial fermentation [12]. Plant extraction suffers from raw material dependence, geographical and seasonal limitations, low yields, complex purification, and high costs [13]. Chemical synthesis, although well-controlled, often exhibits poor stereoselectivity and requires expensive catalysts or harsh conditions, producing by-products that are difficult to handle and limiting alignment with green manufacturing principles [14,15]. In contrast, biological synthesis offers milder, environmentally friendly processes. Microbial fermentation can directly convert inexpensive carbon sources into ʟ-DOPA, providing advantages in scalability, safety, and environmental sustainability. With advances in synthetic biology and metabolic engineering, constructing high-performance recombinant strains has become the key strategy for industrial ʟ-DOPA production. Several microbial hosts have been explored for ʟ-DOPA fermentation, including Escherichia coli [[16], [17], [18], [19]], C. glutamicum [9], Bacillus licheniformis [20], and Aspergillus oryzae [21]. Among these, E. coli has been widely used due to its well-characterized genetic background, mature manipulation toolkit, and relatively low engineering cost. For instance, Fordjour et al. [16] systematically optimized metabolic flux in E. coli BL21, yielding the engineered strain LP-8, which achieved 691.24 mg/L in shake flasks and 25.53 g/L in a 5-L bioreactor. However, E. coli is prone to bacteriophage contamination and presents biosafety concerns at industrial scale, which may compromise production stability and increase cost. In comparison, C. glutamicum is a General Regarded as Safety (GRAS) organism that does not produce endotoxins, reducing downstream purification pressure and production risk [22,23]. The model strain ATCC 13032 was fully genome-sequenced, revealing a clear and highly plastic metabolic network suitable for targeted engineering [24]. C. glutamicum is robust under high substrate or product concentrations and produces minimal off-odors, making it an ideal microbial chassis for industrial-scale biomanufacturing [25]. Recent studies demonstrated feasibility by introducing the tyrosinase tyrRS from Ralstonia solanacearum, producing 0.13 g/L ʟ-DOPA and confirming its potential as a green production platform. In this study, C. glutamicum ATCC 13032 was employed as the microbial chassis to comprehensively assess its intrinsic capacity for ʟ-DOPA biosynthesis. Initially, the tolerance of C. glutamicum to ʟ-DOPA was systematically assessed to verify its suitability as a robust production host. The key hydroxylase genes hpaBC were subsequently introduced, followed by promoter and ribosome binding site (RBS) rational optimization to enhance the efficiency of the de novo biosynthetic pathway. To maximize carbon flux toward ʟ-DOPA, the competing biosynthetic routes for aromatic amino acids were disrupted, coupled with systematic adjustment of fermentation conditions to maintain a balance between growth and metabolite accumulation. Promoter engineering was applied to upregulate the expression of Cgl0218, Cgl0226, and Cgl2101, thereby reinforcing the intracellular ʟ-tyrosine supply. Finally, the deletion of qsuB and qsuD, coupled with the overexpression of aroA and aroC, strengthened precursor flow through the shikimate pathway, effectively boosting metabolic throughput toward ʟ-DOPA biosynthesis. Collectively, these engineering strategies establish a comprehensive technical framework and provide theoretical support for the development of high-performance C. glutamicum strains tailored for industrial-scale ʟ-DOPA production. Materials and methods ʟ-DOPA tolerance assay The reference strain C. glutamicum ATCC 13032 preserved in our laboratory was used to evaluate cellular tolerance toward ʟ-DOPA. Cryopreserved cells stored at −80 °C were streaked onto LB agar plates and sequentially activated twice for a total of 36 h. A single colony was inoculated into 10 mL of LB medium and cultivated at 32 °C with shaking at 220 rpm for 12 h to obtain the seed culture. To precisely control ʟ-DOPA concentrations, a filter-sterilized ʟ-DOPA stock solution was prepared and added to the fermentation medium prior to inoculation. The seed culture was transferred at an inoculation volume of 0.5 mL into media supplemented with different concentrations of ʟ-DOPA. Samples were collected at 2 h intervals from 0 to 12 h, followed by additional sampling at 16 and 24 h. After appropriate dilution, cell growth was determined by measuring the optical density at 600 nm (OD600) using a microplate reader. Growth curves were constructed to quantitatively assess the tolerance of C. glutamicum to increasing ʟ-DOPA concentrations. Strains and plasmids All bacterial strains and plasmids employed in this study are summarized in Table 1. E. coli DH5α was used as the cloning host for plasmid construction, while C. glutamicum ATCC 13032 served as the parental strain for metabolic engineering. Standard molecular biology techniques, including PCR amplification, DNA purification, restriction enzyme digestion, homologous recombination, transformation, and plasmid isolation, were performed according to established protocols.Table 1Strains and plasmids used in this work.Table 1Strain/PlasmidCharacteristicsSourceStrainE. coli DH5ɑClone host strainTransGenKC.glutamicum ATCC 13032KTC. glutamicum ATCC 13032 harboring pXMJ19-hpaBCLab stockKT1C. glutamicum ATCC 13032 harboring pXMJ19-hpaBC RBS1Lab stockKT2C. glutamicum ATCC 13032 harboring pXMJ19-hpaBC RBS2Lab stockKT3C. glutamicum ATCC 13032 harboring pXMJ19-hpaBC RBS3Lab stockKT4C. glutamicum ATCC 13032 harboring pXMJ19-hpaBC RBS4Lab stockKT5C. glutamicum ATCC 13032 harboring pXMJ19-Psod -hpaBC RBS1Lab stockKT6C. glutamicum ATCC 13032 harboring pXMJ19-Psod -hpaBC RBS2Lab stockKT7C. glutamicum ATCC 13032 harboring pXMJ19-Psod -hpaBC RBS3Lab stockKT8C. glutamicum ATCC 13032 harboring pXMJ19-Peftu -hpaBC RBS1Lab stockKT9C. glutamicum ATCC 13032 harboring pXMJ19-Peftu -hpaBC RBS2Lab stockKT10C. glutamicum ATCC 13032 harboring pXMJ19-Peftu -hpaBC RBS3Lab stockKT11C. glutamicum ATCC 13032 harboring pXMJ19-Peftu -hpaBC RBS4Lab stockKT12C. glutamicum ATCC 13032 harboring pXMJ19-PNcgl0824 -hpaBC RBS1Lab stockKT13C. glutamicum ATCC 13032 harboring pXMJ19-PNcgl0824 -hpaBC RBS2Lab stockKT14C. glutamicum ATCC 13032 harboring pXMJ19-PNcgl0824 -hpaBC RBS3Lab stockKT15C. glutamicum ATCC 13032 harboring pXMJ19-PNcgl0824 -hpaBC RBS4Lab stockKD0KT with pheA deletionThis studyKD1KD0 with trpE2 deletionThis studyKD2KD0 with Psod promoter inserted in front of cgl0218This studyKD3KD0 with Psod promoter inserted in front of cgl0226This studyKD4KD0 with Psod promoter inserted in front of cgl2101This studyKD5KD3 with Psod promoter inserted in front of cgl0218This studyKD6KD3 carrying expression vector PSB19This studyKD7KD3 carrying expression vector PSB19-TyrA7This studyKD8KD3 carrying expression vector PSB19-TyrA20This studyKD9KD3 with Psod promoter inserted in front of cgl2101This studyKD10KD5 with Psod promoter inserted in front of cgl2101This studyKD11KD9 with qsuB deletionThis studyKD12KD9 with qsuD deletionThis studyKD13KD11 with qsuD deletionThis studyKD14KD9 with PNcgl0824 promoter inserted in front of aroAThis studyKD15KD9 with Psod promoter inserted in front of aroAThis studyKD16KD9 with PNcgl0824 promoter inserted in front of aroCThis studyKD17KD9 with Psod promoter inserted in front of aroCThis studyKD18KD13 with Psod promoter inserted in front of aroAThis studyKD19KD13with PNcgl0824 promoter inserted in front of aroCThis studyKD20KD19 with Psod promoter inserted in front of aroAThis studyPlasmidThis studypXMJ19A shuttle expression vector, ChlRLab stockpK18mobsacBMobilizable vector, allows for genome edit in C. glutamicum, KmR, sacBLab stockpSR18Mobilizable vector, allows for genome edit in C. glutamicum, SpeR, sacBLab stockpSB19A shuttle expression vector derived from pXMJ19, SpeRLab stockpSR18-ΔpheAA derivative of pSR18, harboring △pheA fragmentLab stockpSR18-ΔtrpE2A derivative of pSR18, harboring ΔtrpE2 fragmentLab stockpSR18-Psod-cgl0218A derivative of pSR18, harboring Psod-cgl0218 fragmentLab stockpSR18-Psod-cgl0226A derivative of pSR18, harboring Psod-cgl0226 fragmentLab stockpSR18-Psod-cgl2101A derivative of pSR18, harboring Psod-cgl2101 fragmentLab stockpSB19-TyrA7A derivative of pSB19, harboring TyrA7 geneLab stockpSB19-TyrA20A derivative of pSB19, harboring TyrA20geneLab stockpK18-ΔqsuBA derivative of pK18mobsacB, harboring ΔqsuB fragmentLab stockpK18-ΔqsuDA derivative of pK18mobsacB, harboring ΔqsuD fragmentLab stockpSR18-PNcgl0824-aroAA derivative of pSR18, harboring PNcgl0824-aroA fragmentLab stockpSR18-Psod-aroAA derivative of pSR18, harboring Psod -aroA fragmentLab stockpSR18-PNcgl0824-aroCA derivative of pSR18, harboring PNcgl0824-aroC fragmentLab stockpSR18-Psod-aroCA derivative of pSR18, harboring Psod -aroC fragmentLab stock Table 1Strains and plasmids used in this work.Table 1Strain/PlasmidCharacteristicsSourceStrainE. coli DH5ɑClone host strainTransGenKC.glutamicum ATCC 13032KTC. glutamicum ATCC 13032 harboring pXMJ19-hpaBCLab stockKT1C. glutamicum ATCC 13032 harboring pXMJ19-hpaBC RBS1Lab stockKT2C. glutamicum ATCC 13032 harboring pXMJ19-hpaBC RBS2Lab stockKT3C. glutamicum ATCC 13032 harboring pXMJ19-hpaBC RBS3Lab stockKT4C. glutamicum ATCC 13032 harboring pXMJ19-hpaBC RBS4Lab stockKT5C. glutamicum ATCC 13032 harboring pXMJ19-Psod -hpaBC RBS1Lab stockKT6C. glutamicum ATCC 13032 harboring pXMJ19-Psod -hpaBC RBS2Lab stockKT7C. glutamicum ATCC 13032 harboring pXMJ19-Psod -hpaBC RBS3Lab stockKT8C. glutamicum ATCC 13032 harboring pXMJ19-Peftu -hpaBC RBS1Lab stockKT9C. glutamicum ATCC 13032 harboring pXMJ19-Peftu -hpaBC RBS2Lab stockKT10C. glutamicum ATCC 13032 harboring pXMJ19-Peftu -hpaBC RBS3Lab stockKT11C. glutamicum ATCC 13032 harboring pXMJ19-Peftu -hpaBC RBS4Lab stockKT12C. glutamicum ATCC 13032 harboring pXMJ19-PNcgl0824 -hpaBC RBS1Lab stockKT13C. glutamicum ATCC 13032 harboring pXMJ19-PNcgl0824 -hpaBC RBS2Lab stockKT14C. glutamicum ATCC 13032 harboring pXMJ19-PNcgl0824 -hpaBC RBS3Lab stockKT15C. glutamicum ATCC 13032 harboring pXMJ19-PNcgl0824 -hpaBC RBS4Lab stockKD0KT with pheA deletionThis studyKD1KD0 with trpE2 deletionThis studyKD2KD0 with Psod promoter inserted in front of cgl0218This studyKD3KD0 with Psod promoter inserted in front of cgl0226This studyKD4KD0 with Psod promoter inserted in front of cgl2101This studyKD5KD3 with Psod promoter inserted in front of cgl0218This studyKD6KD3 carrying expression vector PSB19This studyKD7KD3 carrying expression vector PSB19-TyrA7This studyKD8KD3 carrying expression vector PSB19-TyrA20This studyKD9KD3 with Psod promoter inserted in front of cgl2101This studyKD10KD5 with Psod promoter inserted in front of cgl2101This studyKD11KD9 with qsuB deletionThis studyKD12KD9 with qsuD deletionThis studyKD13KD11 with qsuD deletionThis studyKD14KD9 with PNcgl0824 promoter inserted in front of aroAThis studyKD15KD9 with Psod promoter inserted in front of aroAThis studyKD16KD9 with PNcgl0824 promoter inserted in front of aroCThis studyKD17KD9 with Psod promoter inserted in front of aroCThis studyKD18KD13 with Psod promoter inserted in front of aroAThis studyKD19KD13with PNcgl0824 promoter inserted in front of aroCThis studyKD20KD19 with Psod promoter inserted in front of aroAThis studyPlasmidThis studypXMJ19A shuttle expression vector, ChlRLab stockpK18mobsacBMobilizable vector, allows for genome edit in C. glutamicum, KmR, sacBLab stockpSR18Mobilizable vector, allows for genome edit in C. glutamicum, SpeR, sacBLab stockpSB19A shuttle expression vector derived from pXMJ19, SpeRLab stockpSR18-ΔpheAA derivative of pSR18, harboring △pheA fragmentLab stockpSR18-ΔtrpE2A derivative of pSR18, harboring ΔtrpE2 fragmentLab stockpSR18-Psod-cgl0218A derivative of pSR18, harboring Psod-cgl0218 fragmentLab stockpSR18-Psod-cgl0226A derivative of pSR18, harboring Psod-cgl0226 fragmentLab stockpSR18-Psod-cgl2101A derivative of pSR18, harboring Psod-cgl2101 fragmentLab stockpSB19-TyrA7A derivative of pSB19, harboring TyrA7 geneLab stockpSB19-TyrA20A derivative of pSB19, harboring TyrA20geneLab stockpK18-ΔqsuBA derivative of pK18mobsacB, harboring ΔqsuB fragmentLab stockpK18-ΔqsuDA derivative of pK18mobsacB, harboring ΔqsuD fragmentLab stockpSR18-PNcgl0824-aroAA derivative of pSR18, harboring PNcgl0824-aroA fragmentLab stockpSR18-Psod-aroAA derivative of pSR18, harboring Psod -aroA fragmentLab stockpSR18-PNcgl0824-aroCA derivative of pSR18, harboring PNcgl0824-aroC fragmentLab stockpSR18-Psod-aroCA derivative of pSR18, harboring Psod -aroC fragmentLab stock Strains and plasmids used in this work. DNA manipulation and strain construction For plasmid construction, genomic DNA was isolated from E. coli using a commercial extraction kit (Tiangen, Beijing, China) and used as the template for PCR amplification of hpaBC and its ribosome-binding-site-optimized variants (hpaBC(RBS1–4)). All primers are listed in Table S1. Strong constitutive promoters, including Peftu and Psod, were amplified from the genome of C. glutamicum ATCC 13032. Promoter fragments were fused with hpaBC by overlap extension PCR to generate complete expression cassettes, which were subsequently inserted into the pXMJ19 vector, yielding a series of recombinant plasmids. Using a similar strategy, tyrA7 and tyrA20 were amplified from the genome of E. coli BL21(DE3), fused by overlap extension PCR, and cloned into pSB19 to construct pSB19-TyrA7 and pSB19-TyrA20, respectively. Recombinant plasmids and the empty vector control were introduced into C. glutamicum strains by electroporation, and verified by colony PCR. Chromosomal gene deletions and promoter replacements were performed via a double-crossover homologous recombination strategy using the suicide plasmid pSR18, which carries the sacB counter-selection marker. Homologous arms flanking pheA and trpE2 were amplified, fused by overlap extension PCR, and cloned into pSR18 to generate deletion plasmids. For promoter replacement, the strong promoter Psod and homologous arms flanking Cgl0226, Cgl0218, and Cgl2101 were amplified and assembled into promoter-replacement fragments, which were cloned into pSR18. The resulting plasmids were introduced into C. glutamicum by electroporation, and correct recombinants were confirmed after two rounds of homologous recombination. Shake-flask production of ʟ-DOPA LB medium was used for routine cultivation of E. coli and C. glutamicum. For ʟ-DOPA production, recombinant strains were activated on LB agar plates for two consecutive cycles (36 h total). A loopful of cells was inoculated into 20 mL of seed medium in a 250 mL baffled flask and incubated at 32 °C and 220 rpm for 12 h. The seed medium (per liter) contained 25 g glucose, 10 g corn steep liquor powder, 5 g (NH4)2SO4, 1 g MgSO4·7H2O, 1 g KH2PO4, and 2 g urea, with the pH adjusted to 6.8. Seed cultures were transferred at 10% (v/v) into fermentation medium containing (per liter) 100 g glucose, 15 g corn steep liquor, 25 g (NH4)2SO4, 0.5 g MgSO4·7H2O, 1 g KH2PO4, 2 g sodium citrate, and 10 g CaCO3, with the pH adjusted to 7.0. After 6 h of cultivation, IPTG was added to a final concentration of 1 mM, and fermentation was continued for 48 h. Samples were collected to determine ʟ-DOPA concentration and cell growth. When required, chloramphenicol was added at 50 mg/L for E. coli and 25 mg/L for C. glutamicum. Fed-batch fermentation for ʟ-DOPA production in a 5 L bioreactor Fed-batch fermentation was conducted in a 5 L bioreactor. Engineered strains were activated on LB agar plates and transferred into 200 mL of seed medium in a 500 mL baffled flask. The seed medium (per liter) consisted of 30 g glucose, 10 mL corn steep liquor, 10 g yeast extract, 15 g (NH4)2SO4, 2.5 g MgSO4·7H2O, 1 g KH2PO4, 0.5 g K2HPO4, and 0.5 g Na2HPO4 (pH 6.8). After 12 h of cultivation, the seed culture was used to inoculate the bioreactor. The fermentation medium contained (per liter) 20 g peptone, 20 g yeast extract, 30 g (NH4)2SO4, 3 g KH2PO4, 2.5 g MgSO4·7H2O, 1.5 g K2HPO4, 1.5 g Na2HPO4, 0.09 g MnSO4, and 0.09 g FeSO4. IPTG was added to 1 mM after 6 h. Samples were collected every 12 h for analysis of ʟ-DOPA concentration, biomass, and residual glucose. Analytical methods Cell growth was monitored by measuring OD600. Prior to measurement, fermentation broth was treated with 0.125 mol/L HCl to dissolve CaCO3. Glucose concentration was determined using an SBA-40C biosensor after centrifugation and filtration. ʟ-DOPA and tyrosine concentrations were quantified by HPLC. Clarified supernatants were analyzed using a Waters C18 column at 30 °C with methanol/0.1 N acetic acid (10:1, v/v) as the mobile phase at 1.2 mL/min. Detection was performed at 280 nm with an injection volume of 20 μL. Standard calibration curves were generated using serially diluted authentic standards. Results and discussions Evaluation of ʟ-DOPA tolerance in C. glutamicum The tolerance of a microbial chassis to the target compound is a pivotal factor for industrial-scale biomanufacturing. Excessive accumulation of the product can exert toxic or inhibitory effects on the host cell, compromising growth and metabolism. These consequences ultimately limit both productivity and process stability, underscoring the importance of engineering robust host strains for sustainable, high-yield biosynthesis [26,27]. Thus, evaluating product tolerance is a necessary step for rational metabolic engineering, as strains capable of maintaining growth and metabolic activity under high product concentrations enable greater pathway flux, longer fermentation performance, and more robust industrial applicability. In this study, the tolerance of C. glutamicum ATCC 13032 toward ʟ-DOPA was systematically assessed by monitoring its growth under increasingly elevated ʟ-DOPA levels. In the absence of ʟ-DOPA, the strain rapidly proliferated to an OD600 of 7.79 within 24 h, confirming its robust baseline growth physiology (Fig. 1). At concentrations of 1–5 g/L, the growth behavior was nearly identical to the control group, indicating that low to mid-level ʟ-DOPA accumulation imposes negligible physiological stress (Fig. 1). Upon exposure to higher ʟ-DOPA concentrations, transient growth inhibition emerged during the early fermentation phase (0-8 h). However, the cells quickly transitioned into exponential growth, suggesting the involvement of effective adaptive mechanisms. Even when challenged with 15 g/L ʟ-DOPA, the strain achieved a final biomass only 19.7% lower than the untreated culture, demonstrating strong intrinsic ʟ-DOPA tolerance and a remarkable ability to recover from biosynthetic stress (Fig. 1). These results collectively position C. glutamicum ATCC 13032 as a robust chassis for microbial ʟ-DOPA production. Its ability to remain metabolically active under high ʟ-DOPA accumulation provides a favorable physiological landscape for further metabolic pathway optimization.Fig. 1Cell growth dynamics following ʟ-DOPA supplementation in C. glutamicum. Reported data reflect the mean ± SD obtained from three independent replicates. Cell growth dynamics following ʟ-DOPA supplementation in C. glutamicum. Reported data reflect the mean ± SD obtained from three independent replicates. Assembly of an ʟ-DOPA biosynthetic pathway in C. glutamicum via heterologous expression of hpaBC ʟ-DOPA is biosynthesized from ʟ-tyrosine, with the conversion mediated by the flavin-dependent monooxygenase 4-hydroxyphenylacetate 3-hydroxylase (4HPA3H). In this study, hpaBC operon, which encode the two-component 4HPA3H enzyme system, from E. coli were heterologous introduced into C. glutamicum ATCC13032 to reconstruct a functional ʟ-DOPA biosynthetic pathway (Fig. 2a). The resulting engineered strain KT produced 77.20 ± 3.66 mg/L ʟ-DOPA in shake-flask fermentation, verifying effective hpaBC expression and demonstrating de novo ʟ-DOPA biosynthesis in C. glutamicum (Fig. 2b and Table 2). This result is consistent with previous reports demonstrating de novo ʟ-DOPA synthesis in E. coli through the heterologous expression of 4HPA3H [17,18]. Nevertheless, the ʟ-DOPA production titer remained limited. We hypothesized that ʟ-DOPA biosynthesis may be influenced by two key factors: the availability of the precursor ʟ-tyrosine and the expression level of hpaBC. To assess the impact of precursor utilization, 1 g/L ʟ-tyrosine was supplemented into the fermentation medium, which increased ʟ-DOPA production in the KT strain to 289.2 ± 9.79 mg/L—a 2.74-fold improvement—indicating that precursor availability constitutes a major bottleneck (Fig. 2b and Table 2). However, the substantial accumulation of residual ʟ-tyrosine in the fermentation broth (886.4 ± 52.25 mg/L) further suggests that insufficient hpaBC expression constitutes a limiting factor for ʟ-DOPA production (Fig. 2c and Table 2). To finely tune hpaBC expression, a combinatorial expression strategy was employed in which three strong promoters were randomly paired with four ribosome binding sites with predicted translation start strength of 5341.55, 25210.78, 57705.62, and 580091.25 au (Table S2) [23,28]. This systematic promoter–RBS assembly generated a panel of engineered strains, designated KT1 to KT15, enabling evaluation of expression-level effects on ʟ-DOPA biosynthesis. Unexpectedly, all 15 engineered strains showed reduced ʟ-DOPA titers and impaired growth compared with the control strain KT (Fig. 2b and c), indicating that overexpression of hpaBC likely caused metabolic stress by depleting cellular resources or disrupting overall pathway balance. These results emphasize that finely tuned, rather than maximal, gene expression is essential for achieving optimal pathway efficiency. To promote ʟ-DOPA biosynthesis without raising fermentation costs, increasing the endogenous supply of ʟ-tyrosine represents a preferred strategy.Fig. 2Construction and fermentation performance evaluation of recombinant C. glutamicum for de novo ʟ-DOPA biosynthesis. (a) Schematic illustration of the construction of recombinant C. glutamicum strains KT01–KT15. (b) Comparison of ʟ-DOPA titers among recombinant C. glutamicum KT01–KT15. (c) Biomass accumulation and ʟ-tyrosine synthesis in engineered C. glutamicum strains KT01–KT15. Reported data reflect the mean ± SD obtained from three independent replicates.Table 2Engineering the ʟ-DOPA production in recombinant C. glutamicum.Table 2Strains (C. glutamicum)ʟ-DOPA accumulation (mg/L)Cell Biomass (OD600)ʟ-tyrosine accumulation (mg/L)ʟ-DOPA/OD600KT(a)82.56 ± 4.8026.02 ± 1.11478.23 ± 15.173.17KT289.18 ± 9.8018.55 ± 0.58886.40 ± 52.2515.59KT1203.42 ± 3.2817.56 ± 0.66838.55 ± 29.2411.58KT296.94 ± 2.6814.98 ± 1.051001.84 ± 43.836.47KT3168.47 ± 4.2113.18 ± 1.13922.45 ± 29.3912.78KT4224.18 ± 7.1117.04 ± 1.04856.16 ± 39.0613.16KT5241.13 ± 8.9816.96 ± 1.25869.51 ± 19.1214.22KT6142.42 ± 3.8517.85 ± 1.12976.85 ± 39.577.98KT7119.14 ± 8.3419.77 ± 1.24976.33 ± 10.726.03KT852.85 ± 5.0416.23 ± 1.131026.52 ± 25.713.26KT9105.24 ± 8.7717.20 ± 0.99904.43 ± 5.866.12KT1093.13 ± 4.6416.33 ± 1.151011.36 ± 35.015.70KT11137.76 ± 9.8918.09 ± 1.63981.36 ± 46.437.62KT12111.14 ± 6.2518.34 ± 0.441055.91 ± 45.066.06KT13108.18 ± 5.2317.51 ± 1.051065.43 ± 45.246.18KT1497.30 ± 3.5318.00 ± 0.751073.55 ± 14.435.41KT1565.40 ± 2.7615.98 ± 1.131036.52 ± 35.614.10aRepresents the group in the absence of ʟ-tyrosine. Construction and fermentation performance evaluation of recombinant C. glutamicum for de novo ʟ-DOPA biosynthesis. (a) Schematic illustration of the construction of recombinant C. glutamicum strains KT01–KT15. (b) Comparison of ʟ-DOPA titers among recombinant C. glutamicum KT01–KT15. (c) Biomass accumulation and ʟ-tyrosine synthesis in engineered C. glutamicum strains KT01–KT15. Reported data reflect the mean ± SD obtained from three independent replicates. Table 2Engineering the ʟ-DOPA production in recombinant C. glutamicum.Table 2Strains (C. glutamicum)ʟ-DOPA accumulation (mg/L)Cell Biomass (OD600)ʟ-tyrosine accumulation (mg/L)ʟ-DOPA/OD600KT(a)82.56 ± 4.8026.02 ± 1.11478.23 ± 15.173.17KT289.18 ± 9.8018.55 ± 0.58886.40 ± 52.2515.59KT1203.42 ± 3.2817.56 ± 0.66838.55 ± 29.2411.58KT296.94 ± 2.6814.98 ± 1.051001.84 ± 43.836.47KT3168.47 ± 4.2113.18 ± 1.13922.45 ± 29.3912.78KT4224.18 ± 7.1117.04 ± 1.04856.16 ± 39.0613.16KT5241.13 ± 8.9816.96 ± 1.25869.51 ± 19.1214.22KT6142.42 ± 3.8517.85 ± 1.12976.85 ± 39.577.98KT7119.14 ± 8.3419.77 ± 1.24976.33 ± 10.726.03KT852.85 ± 5.0416.23 ± 1.131026.52 ± 25.713.26KT9105.24 ± 8.7717.20 ± 0.99904.43 ± 5.866.12KT1093.13 ± 4.6416.33 ± 1.151011.36 ± 35.015.70KT11137.76 ± 9.8918.09 ± 1.63981.36 ± 46.437.62KT12111.14 ± 6.2518.34 ± 0.441055.91 ± 45.066.06KT13108.18 ± 5.2317.51 ± 1.051065.43 ± 45.246.18KT1497.30 ± 3.5318.00 ± 0.751073.55 ± 14.435.41KT1565.40 ± 2.7615.98 ± 1.131036.52 ± 35.614.10aRepresents the group in the absence of ʟ-tyrosine. Engineering the ʟ-DOPA production in recombinant C. glutamicum. Represents the group in the absence of ʟ-tyrosine. Blocking ʟ-phenylalanine and ʟ-tryptophan biosynthesis pathway and fermentation optimization to enhance ʟ-DOPA accumulation In the aromatic amino acid biosynthetic network of C. glutamicum, branch-chain keto acids function as central metabolic intermediates and serve as shared precursors for the biosynthesis of ʟ-tyrosine, ʟ-phenylalanine, and ʟ-tryptophan. Within this network, pheA encodes prephenate dehydratase, catalyzing the first committed step in the ʟ-phenylalanine biosynthetic branch, while trpE2 encodes anthranilate synthase, responsible for the initial committed step of the ʟ-tryptophan pathway. The coexistence of these competing pathways not only redirects carbon flux away from branch-chain keto acid formation but also leads to the accumulation of metabolic intermediates that impose feedback regulation on the expression or catalytic activity of key enzymes required for ʟ-DOPA biosynthesis. To mitigate these metabolic constraints, pheA was deleted in the engineered strain KT to generate strain KD0, and subsequent removal of trpE2 from KD0 yielded strain KD1 (Fig. 3a). Shake-flask fermentation of those strains demonstrated that KD0 produced 267.89 ± 2.67 mg/L ʟ-DOPA, representing a 2.24-fold increase compared to strain KT, confirming that inhibition of the phenylalanine biosynthetic pathway effectively improves ʟ-DOPA production (Fig. 3b and Table 3). Along with the increase in ʟ-DOPA production, the residual ʟ-tyrosine levels and OD600 also decreased markedly (Fig. 3b and Table 3). Notably, this observation aligns with previous reports in which deletion of pheA to generate an ʟ-phenylalanine auxotrophic strain (LP-4) led to an ʟ-DOPA titer of 411.72 mg/L, representing a 1.29-fold enhancement over the parental strain [29]. In contrast, KD1 exhibited substantial decreases in both ʟ-tyrosine and ʟ-DOPA titers—by 37% and 25.4%, respectively—along with a 51.5% reduction in OD600 compared with strain KT (Fig. 3b and Table 3). These results indicate that simultaneous disruption of both the ʟ-phenylalanine and ʟ-tryptophan pathways results in a severe deficit of aromatic amino acids, thereby impairing cellular growth and limiting biosynthetic capacity. This outcome highlights the necessity of balancing pathway rerouting with the essential metabolic demands of the host cell. Consequently, KD0, which maintained a more favorable trade-off between growth and ʟ-DOPA production, was selected for subsequent fermentation optimization.Fig. 3Effects of deleting competing metabolic pathways on ʟ-DOPA production in C. glutamicum. (a) Schematic diagram of the gene engineering approach. EPSP, 5-enolpyruvylshikimate-3-phosphate; CHA, Chorismate; PRE, Prephenate; HPPH, 4-hydroxyphenylpyruvate; trpE2, encoding anthranilate synthase; pheA, encoding chorismate mutase/prephenate dehydratase. (b) Comparison of fermentation parameters between recombinant strains KD0 and KD1. (c) Influence of inoculation time on ʟ-DOPA fermentation performance. (d) ʟ-DOPA synthesis under different pH conditions. (e) Influence of riboflavin dosage on ʟ-DOPA fermentation performance. (f) ʟ-DOPA fermentation profiles over time. Reported data reflect the mean ± SD obtained from three independent replicates.Table 3Engineering the ʟ-DOPA production in recombinant C. glutamicum.Table 3Strains (C.glutamicum)ʟ-DOPA accumulation (mg/L)Cell Biomass (OD600)ʟ-tyrosine accumulation (mg/L)ʟ-DOPA/OD600KD0267.89 ± 2.6718.91 ± 0.66296.26 ± 21.5714.17KD1199.61 ± 0.2312.62 ± 0.42180.46 ± 1.2115.82KD0a420.07 ± 0.3517.75 ± 0.9176.36 ± 0.2923.67KD263.16 ± 1.8018.27 ± 1.09285.82 ± 12.263.46KD3192.85 ± 0.2017.30 ± 0.94239.11 ± 1.8311.15KD4158.51 ± 0.1718.13 ± 0.95255.50 ± 0.218.74KD580.01 ± 0.2219.02 ± 0.84312.17 ± 9.014.21KD6120.73 ± 9.9618.08 ± 0.64249.14 ± 13.226.68KD762.20 ± 3.1120.05 ± 0.6895.57 ± 4.323.10KD895.55 ± 2.0719.07 ± 0.66237.57 ± 2.105.01KD9443.32 ± 9.8920.58 ± 0.7259.78 ± 3.7421.26KD10352.80 ± 6.6818.84 ± 0.74122.37 ± 3.1418.73KD9a507.16 ± 2.2832.35 ± 0.07104.56 ± 0.6715.68KD11561.15 ± 1.3830.83 ± 1.18161.12 ± 5.2818.20KD12569.55 ± 0.1729.12 ± 0.85129.68 ± 1.1519.56KD13617.70 ± 0.9330.10 ± 0.16164.19 ± 2.3120.52KD14543.31 ± 6.3433.56 ± 1.41207.31 ± 8.0416.19KD15556.74 ± 2.4234.33 ± 0.02210.37 ± 1.6516.22KD16562.43 ± 5.1733.17 ± 0.53215.32 ± 9.9216.96KD17558.26 ± 3.3232.73 ± 0.87204.93 ± 2.5517.06KD18624.34 ± 47.6415.00 ± 0.34134.92 ± 6.2141.62KD19697.93 ± 41.1515.32 ± 0.25159.44 ± 17.5845.56KD20710.98 ± 38.6115.27 ± 0.3799.76 ± 7.9646.56aRepresents the optimal fermentation condition. Effects of deleting competing metabolic pathways on ʟ-DOPA production in C. glutamicum. (a) Schematic diagram of the gene engineering approach. EPSP, 5-enolpyruvylshikimate-3-phosphate; CHA, Chorismate; PRE, Prephenate; HPPH, 4-hydroxyphenylpyruvate; trpE2, encoding anthranilate synthase; pheA, encoding chorismate mutase/prephenate dehydratase. (b) Comparison of fermentation parameters between recombinant strains KD0 and KD1. (c) Influence of inoculation time on ʟ-DOPA fermentation performance. (d) ʟ-DOPA synthesis under different pH conditions. (e) Influence of riboflavin dosage on ʟ-DOPA fermentation performance. (f) ʟ-DOPA fermentation profiles over time. Reported data reflect the mean ± SD obtained from three independent replicates. Table 3Engineering the ʟ-DOPA production in recombinant C. glutamicum.Table 3Strains (C.glutamicum)ʟ-DOPA accumulation (mg/L)Cell Biomass (OD600)ʟ-tyrosine accumulation (mg/L)ʟ-DOPA/OD600KD0267.89 ± 2.6718.91 ± 0.66296.26 ± 21.5714.17KD1199.61 ± 0.2312.62 ± 0.42180.46 ± 1.2115.82KD0a420.07 ± 0.3517.75 ± 0.9176.36 ± 0.2923.67KD263.16 ± 1.8018.27 ± 1.09285.82 ± 12.263.46KD3192.85 ± 0.2017.30 ± 0.94239.11 ± 1.8311.15KD4158.51 ± 0.1718.13 ± 0.95255.50 ± 0.218.74KD580.01 ± 0.2219.02 ± 0.84312.17 ± 9.014.21KD6120.73 ± 9.9618.08 ± 0.64249.14 ± 13.226.68KD762.20 ± 3.1120.05 ± 0.6895.57 ± 4.323.10KD895.55 ± 2.0719.07 ± 0.66237.57 ± 2.105.01KD9443.32 ± 9.8920.58 ± 0.7259.78 ± 3.7421.26KD10352.80 ± 6.6818.84 ± 0.74122.37 ± 3.1418.73KD9a507.16 ± 2.2832.35 ± 0.07104.56 ± 0.6715.68KD11561.15 ± 1.3830.83 ± 1.18161.12 ± 5.2818.20KD12569.55 ± 0.1729.12 ± 0.85129.68 ± 1.1519.56KD13617.70 ± 0.9330.10 ± 0.16164.19 ± 2.3120.52KD14543.31 ± 6.3433.56 ± 1.41207.31 ± 8.0416.19KD15556.74 ± 2.4234.33 ± 0.02210.37 ± 1.6516.22KD16562.43 ± 5.1733.17 ± 0.53215.32 ± 9.9216.96KD17558.26 ± 3.3232.73 ± 0.87204.93 ± 2.5517.06KD18624.34 ± 47.6415.00 ± 0.34134.92 ± 6.2141.62KD19697.93 ± 41.1515.32 ± 0.25159.44 ± 17.5845.56KD20710.98 ± 38.6115.27 ± 0.3799.76 ± 7.9646.56aRepresents the optimal fermentation condition. Engineering the ʟ-DOPA production in recombinant C. glutamicum. Represents the optimal fermentation condition. To further enhance ʟ-DOPA production, key fermentation parameters in strain KD0—including seed culture duration, initial medium pH, riboflavin supplementation, and total cultivation time—were systematically optimized. Extending the seed culture period to 16 h increased the ʟ-DOPA titer to 311.6 ± 10.6 mg/L, corresponding to a 20.2% improvement relative to the control (Fig. 3c), while optimization of the initial medium pH further revealed that the highest ʟ-DOPA concentration (331.49 ± 5.41 mg/L) was achieved at pH 7.0 (Fig. 3d). In contrast, pH values exceeding 7.5 led to visible precipitation, likely decreasing the effective availability of key medium components and consequently compromising fermentation performance. Furthermore, since 4HPA3H is an FAD-dependent hydroxylase and riboflavin serves as its metabolic precursor, exogenous riboflavin supplementation was evaluated to determine whether cofactor availability limited ʟ-DOPA production. Surprisingly, riboflavin addition did not lead to any appreciable increase in ʟ-DOPA titer (Fig. 3e), indicating that the endogenous riboflavin biosynthetic capacity was adequate to sustain FAD-dependent enzyme activity under the tested fermentation conditions. During the fermentation process, the appearance of black pigments upon standing suggested that ʟ-DOPA underwent oxidative degradation. Optimization of the fermentation duration showed that ʟ-DOPA titers peaked at 48 h (407.13 ± 1.51 mg/L) but declined to 343.91 ± 11.26 mg/L by 72 h, corresponding to a 22.8% reduction. (Fig. 3f). This decline is likely attributable to nutrient depletion and a metabolic shift toward cellular maintenance rather than product synthesis, as well as possible detoxification mechanisms that convert ʟ-DOPA into less harmful derivatives. Collectively, these results established that the optimal fermentation parameters for KD0 consisted of a 16 h seed culture, an initial medium pH of 7.0, and a fermentation duration of 48 h, under which ʟ-DOPA biosynthesis was markedly enhanced, thereby providing a solid basis for subsequent process development efforts. Promoter engineering enhances ʟ-tyrosine supplement for ʟ-DOPA production As demonstrated in Section 3.2, the addition of 1 g/L tyrosine increased ʟ-DOPA production by 2.74-fold, confirming that tyrosine supply is a major bottleneck in pathway efficiency. To achieve stable endogenous precursor supplement, promoter engineering was first applied to strengthen the expression of key upstream genes in the ʟ-tyrosine biosynthetic pathway of C. glutamicum. In this metabolic network (Fig. 4a), prephenate is converted to 4-hydroxyphenylpyruvate by prephenate dehydrogenase encoded by Cgl0226, followed by transamination catalyzed by tyrosine aminotransferase isoenzymes encoded by Cgl0218 and Cgl2101, yielding ʟ-tyrosine, which is subsequently hydroxylated by 4HPA3H to form ʟ-DOPA. Based on this pathway structure, the native promoters of Cgl0226, Cgl0218, and Cgl2101 were replaced with the strong promoter Psod to construct strains KD2, KD3, and KD4. However, fermentation results showed that the production titers of ʟ-DOPA decreased by 85%, 55%, and 62.3%, respectively, along with a multiplicative increase in ʟ-tyrosine synthesis (Fig. 4b and c). We hypothesize that this phenomenon may be due to reinforcing a single metabolic node disturbed pathway homeostasis rather than improving global metabolic flux.Fig. 4Effects of ʟ-tyrosine pathway amplification on ʟ-DOPA biosynthesis. (a) Schematic diagram of the gene engineering approach. tyrA and cgl0226, encoding chorismate mutase/prephenate dehydrogenase; cgl2101, encoding 4-hydroxyphenylpyruvate dioxygenase, cgl0218, encoding transaminase. (b) Comparison of ʟ-DOPA titers among recombinant C. glutamicum KD02–KD10. (c) Biomass accumulation and ʟ-tyrosine synthesis in engineered C. glutamicum strains KD02–KD10. (d) Influence of inducer concentration on ʟ-DOPA fermentation performance. (e) Influence of inducer supplementation at various time points on ʟ-DOPA fermentation. Reported data reflect the mean ± SD obtained from three independent replicates. Effects of ʟ-tyrosine pathway amplification on ʟ-DOPA biosynthesis. (a) Schematic diagram of the gene engineering approach. tyrA and cgl0226, encoding chorismate mutase/prephenate dehydrogenase; cgl2101, encoding 4-hydroxyphenylpyruvate dioxygenase, cgl0218, encoding transaminase. (b) Comparison of ʟ-DOPA titers among recombinant C. glutamicum KD02–KD10. (c) Biomass accumulation and ʟ-tyrosine synthesis in engineered C. glutamicum strains KD02–KD10. (d) Influence of inducer concentration on ʟ-DOPA fermentation performance. (e) Influence of inducer supplementation at various time points on ʟ-DOPA fermentation. Reported data reflect the mean ± SD obtained from three independent replicates. To further optimize pathway balance, multi-gene co-expression strategies were employed. Remarkably, co-expression of Cgl0226 and Cgl0218 (strain KD5) caused an 81% decrease in ʟ-DOPA titer while increasing tyrosine accumulation to 3.08-fold that of KD0 (Fig. 4b and c). This inverse relationship suggests that excessive flux toward tyrosine accumulation may impose feedback constraints or disrupt downstream hydroxylation efficiency, thereby impairing ʟ-DOPA biosynthesis. Combining Cgl0226 with Cgl2101 (strain KD9) resulted in a modest 5.6% improvement in ʟ-DOPA production (443.3 ± 9.89 mg/L), whereas simultaneous overexpression of all three genes (KD10) did not provide additive benefits and instead yielded an ʟ-DOPA titer 16.01% lower than that of KD0 (Fig. 4b and Table 3). However, no concomitant increase in ʟ -DOPA accumulation was observed. Notably, individual or combinatorial overexpression of Cgl0218 resulted in varying degrees of elevation in ʟ-tyrosine titers relative to the parental strain. This finding suggests functional divergence among ʟ-tyrosine aminotransferase isozymes. We therefore propose that overexpression of Cgl0218 perturbs intracellular flux distribution, leading to pathway imbalance and activation of competing byproduct biosynthsis routes. Consequently, the precursor ʟ-tyrosine is likely redirected into alternative metabolic branches, thereby diminishing its effective availability for ʟ-DOPA biosynthesis and constraining product accumulation. Although multi-gene regulation did not markedly enhance ʟ-DOPA accumulation, tyrosine production in KD2, KD3, KD4, and KD5 increased to 3.74-, 3.13-, 3.35-, and 4.09-fold that of KD0 (Fig. 4c), confirming that reinforced expression of these rate-limiting enzymes substantially boosted tyrosine biosynthesis and shifted the primary bottleneck to insufficient downstream conversion. To mitigate feedback inhibition, two documented feedback-resistant TyrA variants (TyrA7 and TyrA20) [30] were introduced into KD3 to generate strains KD7 and KD8. The empty plasmid pSB19 was also introduced into KD3 and generate strain KD6. However, all three strains displayed markedly reduced titers of both tyrosine and ʟ-DOPA, which is presumably attributable to homologous recombination events among plasmids with shared backbones, compromised expression stability, or the increased metabolic burden imposed by sustaining multiple plasmids simultaneously. Since strengthening the biosynthetic pathway alone did not resolve the issue of “high tyrosine accumulation but limited conversion to ʟ-DOPA”, we next sought to optimize induction conditions of hpaBC expression as an alternative strategy. Fermentation data revealed that ʟ-DOPA production followed a characteristic “increase-then-decrease” pattern in response to both IPTG concentration and induction timing (Fig. 4d and e). Induction with 1 mM IPTG at 6 h post-inoculation resulted in the highest ʟ-DOPA titer of 538.1 ± 11.45 mg/L, corresponding to a 21.38% increase relative to the control (Fig. 4d and e). These findings suggest that moderate induction effectively enhances heterologous gene expression and metabolic flux, whereas strong or premature induction can trigger cytotoxic stress, impair cellular growth, or promote protein misfolding, ultimately diminishing product formation. Therefore, initiating induction enables cells to maintain metabolic stability, maximizing protein expression and ʟ-DOPA biosynthesis, thereby achieving a more favorable balance between expression and cellular metabolic burden. Strengthening the shikimate pathway to augment metabolic flux toward ʟ-DOPA biosynthesis In C. glutamicum, undesired branch reactions compromise the efficiency of the shikimate pathway. Specifically, the qsuD-encoded quinate/shikimate dehydrogenase and qsuB-encoded 3-dehydroshikimate dehydratase divert metabolic flux toward quinate (QA) and protocatechuate (PCA), respectively, thereby reducing precursor availability for ʟ-DOPA synthesis and generating by-products that complicate downstream purification (Fig. 5a). To address this issue, qsuB and qsuD were deleted individually and in combination, generating strains KD11, KD12, and KD13. Fermentation results showed that ʟ-DOPA titers for KD11, KD12, and KD13 reached 561.15 ± 1.38 mg/L, 569.55 ± 0.17 mg/L, and 617.70 ± 0.93 mg/L, representing increases of 10.6%, 12.3%, and 21.8%, respectively, relative to the control strain KD9 (507.16 ± 2.28 mg/L) (Fig. 5b and Table 3). In addition, promoter engineering was employed to enhance the expression of key enzymes within the shikimate pathway. Substituting the native aroA promoter with the strong promoters PNcgl0824 and Psod generated strains KD14 and KD15, resulting in ʟ-DOPA titers increased by 6.9% and 9.6%, respectively (Fig. 5c). Similarly, enhancing aroC expression through promoter replacement generated strains KD16 and KD17, which achieved 10.7% and 9.9% increases in ʟ-DOPA production, respectively (Fig. 5c). These results are consistent with a previous study reporting that overexpression of aroL, aroC, and aroA in E. coli achieved an ʟ-DOPA titer of 12.5 g/L in a bioreactor, collectively demonstrating that restricting metabolic branching while reinforcing bottleneck enzymes within the pathway can synergistically enhance precursor availability and product accumulation.Fig. 5Promotion of ʟ-DOPA biosynthesis through strengthening the shikimate primary metabolic pathway. (a) Schematic diagram of the gene engineering approach. PEP, Phosphoenolpyruvate; E4P, Erythrose-4-phosphate; DAHP, 3-Deoxy-d-arabino-heptulosonate-7-phosphate; DHQ, 3-Dehydroquinate; QA, Quinate; DHS, 3-Dehydroshikimate; S3P, Shikimate-3-phosphate; qsuB, encoding 3-dehydroquinate dehydratase; qsuD, encoding quinate/shikimate dehydrogenase; aroA, encoding 5-enolpyruvylshikimate-3-phosphate synthase; aroC, encoding chorismate synthase. (b) Comparison of ʟ-DOPA titers among recombinant C. glutamicum KD09, KD11–KD13. (c) Comparison of ʟ-DOPA titers among recombinant C. glutamicum KD09, KD14–KD17. (d) Comparison of ʟ-DOPA titers among recombinant C. glutamicum KD09, KD18–KD20. Reported data reflect the mean ± SD obtained from three independent replicates. Promotion of ʟ-DOPA biosynthesis through strengthening the shikimate primary metabolic pathway. (a) Schematic diagram of the gene engineering approach. PEP, Phosphoenolpyruvate; E4P, Erythrose-4-phosphate; DAHP, 3-Deoxy-d-arabino-heptulosonate-7-phosphate; DHQ, 3-Dehydroquinate; QA, Quinate; DHS, 3-Dehydroshikimate; S3P, Shikimate-3-phosphate; qsuB, encoding 3-dehydroquinate dehydratase; qsuD, encoding quinate/shikimate dehydrogenase; aroA, encoding 5-enolpyruvylshikimate-3-phosphate synthase; aroC, encoding chorismate synthase. (b) Comparison of ʟ-DOPA titers among recombinant C. glutamicum KD09, KD11–KD13. (c) Comparison of ʟ-DOPA titers among recombinant C. glutamicum KD09, KD14–KD17. (d) Comparison of ʟ-DOPA titers among recombinant C. glutamicum KD09, KD18–KD20. Reported data reflect the mean ± SD obtained from three independent replicates. To further enhance ʟ-DOPA production, both strategies were combined to synergistically maximize metabolic flux through the pathway. In strain KD13, the aroA promoter was replaced with Psod to generate KD18, the aroC promoter was substituted with PNcgl0824 to produce KD19, and both promoter replacements were simultaneously implemented to construct KD20. Fermentation analysis showed that ʟ-DOPA productionin strain KD18, KD19, and KD20 increased by 4.9%, 17.2%, and 19.3% relative to KD13, with KD20 achieving the highest titer of 710.98 ± 13.21 mg/L (Fig. 5d and Table 3). This value represents the highest ʟ-DOPA production reported for C. glutamicum using fermentation, highlighting the effectiveness of integrating pathway reinforcement and metabolic burden elimination to enhance microbial biosynthesis. Fed-batch fermentation in a 5 L bioreactor enhances ʟ-DOPA production The engineered strain developed through systematic metabolic engineering in the preceding stages ultimately required validation under scaled fermentation conditions. When the optimized strain KD20 was cultivated in a 5 L bioreactor using a fed-batch fermentation strategy, a previously unobserved metabolic bottleneck emerged compared with shake-flask cultivation. After approximately 36 h of fermentation, the fermentation broth rapidly darkened, ʟ-DOPA titers declined sharply, and the culture exhibited a lower OD600 value. We hypothesized that inappropriate nitrogen source or insufficient supplementation, leading to nutrient limitation during the later stages of fermentation. To identify an appropriate nitrogen source for ʟ-DOPA biosynthesis, corn steep liquor powder (CSL) and yeast extract (YE) were systematically evaluated. CSL is enriched in small peptides, free amino acids, and essential metal ions (e.g., Fe and Zn), which are readily assimilated and can enhance tricarboxylic acid (TCA) cycle activity while alleviating acetate accumulation. In addition, its relatively low tyrosine content (1.5–2.5%) supports precursor supply without inducing pronounced feedback inhibition. In contrast, YE provides abundant B vitamins and nucleotides, promotes robust cell growth, and offers high batch-to-batch consistency, making it advantageous for process standardization. Shake-flask studies demonstrated a positive correlation between CSL concentration and ʟ-DOPA production, with titers peaking at 712.5 ± 16.4 mg/L at 15 g/L CSL, after which further increases in CSL failed to enhance ʟ-DOPA accumulation (Fig. 6a). In YE-supplemented cultures, the maximum ʟ-DOPA titer of 541.93 ± 6.00 mg/L was obtained at 5 g/L, whereas further increases in YE concentration resulted in diminished product accumulation (Fig. 6b). Although CSL outperformed YE at the shake-flask level, its limited solubility, tendency to obstruct spray balls in large-scale fermenters, and higher associated costs constrain its suitability for industrial application. Conversely, YE is more practical for scale-up and contains precursors for glutathione and superoxide dismutase, which can mitigate oxidative stress and stabilize tyrosine hydroxylase activity, thereby supporting sustained ʟ-DOPA synthesis. Further optimization revealed that 5 g/L YE enabled better balance between biomass growth and product formation, allowing ʟ-DOPA to accumulate continuously until the end of fermentation and reaching a maximum titer of 946 ± 17.6 mg/L (Fig. 6c).Fig. 6Scale-up and fermenter optimization of ʟ-DOPA production in C. glutamicum. (a) Effect of different corn steep liquor powder concentrations on ʟ-DOPA fermentation. (b) Effect of different yeast extract concentrations on ʟ-DOPA fermentation. (c) Time-dependent ʟ-DOPA production under different yeast extract additions. (d) Fed-batch fermentation curves of KD20 at 5 L bioreactor scale. Reported data reflect the mean ± SD obtained from three independent replicates. Scale-up and fermenter optimization of ʟ-DOPA production in C. glutamicum. (a) Effect of different corn steep liquor powder concentrations on ʟ-DOPA fermentation. (b) Effect of different yeast extract concentrations on ʟ-DOPA fermentation. (c) Time-dependent ʟ-DOPA production under different yeast extract additions. (d) Fed-batch fermentation curves of KD20 at 5 L bioreactor scale. Reported data reflect the mean ± SD obtained from three independent replicates. Based on these findings, the optimized nitrogen-feeding strategy was implemented in 5 L fed-batch fermentation to comprehensively evaluate the scale-up potential of strain KD20. During the initial fermentation phase (0–11 h), rapid glucose consumption supported vigorous cell growth, while ʟ-DOPA accumulation remained limited, suggesting preferential allocation of nutrients toward biomass formation. Upon initiation of feeding at 11 h and maintenance of residual glucose below 25 g/L, cell growth slowed, and ʟ-DOPA biosynthesis entered a rapid accumulation phase that persisted until the end of fermentation. Notably, without exogenous ʟ-tyrosine supplementation, ʟ-DOPA production titer reached 4605.58 ± 21.73 mg/L after 72 h, representing an approximately 3.87-fold increase compared with shake-flask cultivation (Fig. 6d). Collectively, these results demonstrate the robust production capacity of strain KD20 under scaled fed-batch conditions and represent the highest ʟ-DOPA titer reported to date in C. glutamicum. Drawing on the breakthrough achievements in the microbial production of aromatic compounds such as protocatechuic acid [31] and shikimic acid [32,33], C. glutamicum holds strong promise as a powerful chassis for the fermentative production of ʟ-DOPA. Conclusion This study demonstrates that high-level ʟ-DOPA production in C. glutamicum can be achieved through mechanism-driven metabolic flux redistribution. The native tolerance of C. glutamicum to ʟ-DOPA provided a robust physiological basis for stable implementation of the heterologous hpaBC-mediated biosynthetic pathway. Promoter engineering of key l-tyrosine biosynthetic genes, together with disruption of pheA and trpE, expanded precursor pools and redirected metabolic flux toward l-DOPA synthesis. Notably, deletion of pheA yielded the largest fold increase in l-DOPA titer, highlighting the importance of relieving l-phenylalanine–mediated feedback inhibition. Optimization of induction conditions further balanced pathway expression with cellular burden, improving overall production efficiency. Deletion of competing shikimate pathway branches (qsuB and qsuD) and overexpression of aroA and aroC redirected carbon flux toward chorismate and tyrosine formation, effectively relieving upstream bottlenecks. These coordinated interventions synergistically strengthened pathway continuity, resulting in strain KD20 achieving 710.9 mg/L in shake flasks and 4.61 g/L in fed-batch fermentation without ʟ-tyrosine supplementation—the highest titer reported for C. glutamicum. This work elucidates key flux-control mechanisms underlying efficient ʟ-DOPA biosynthesis and provides a scalable framework for producing aromatic amino acid–derived compounds. CRediT authorship contribution statement Mengzhen Jia: Writing – original draft, Conceptualization. Ran Duan: Methodology, Investigation, Formal analysis, Data curation. Yan Jiang: Investigation. Xiaohe Chu: Conceptualization. Xiaoyu Wu: Supervision. Hui Yang: Conceptualization. Bangce Ye: Resources. Bin Zhang: Supervision, Resources, Project administration, Methodology, Funding acquisition. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.