Abstract Background Rational metabolic pathway engineering is capable of boosting upstream flux towards downstream synthesis of target products, such as aromatic amino acid derivatives. However, coordinated synthesis of multiple downstream derivatives faces difficulty of combinatorial optimization of cellular metabolism. Results We developed a strategy combining metabolic engineering optimization with the global transcriptional regulation of transcription factors (TFs) Spt15p and Gcn4p to optimize the synthesis of aromatic amino acid derivatives in yeast. It is verified that the special mutants of these TFs can respectively improve the biosynthesis of betaxanthin, a tyrosine derived edible pigment. Comparative transcriptome analysis shows that significant transcriptional tuning occurs in glycolysis, pentose phosphate pathway, aromatic amino acid synthesis pathways, etc. In addition, global transcriptional engineering is proved to enhance the coordinated biosynthesis of both tyrosine derived pigment betaxanthin and tryptophan derived pigment violacein by more than 50%. Finally, we obtain an optimized production of 208 mg/L betaxanthin in yeast cells by flask fermentation. Conclusions Our strategy supplies an effective way to optimize the coordinated synthesis of two structurally divergent pigments downstream of the common aromatic amino acid pathway. Supplementary Information The online version contains supplementary material available at 10.1186/s12934-025-02799-6. Keywords: Rational pathway engineering, Global transcriptional engineering, Aromatic amino acid, SPT15, GCN4, Betaxanthin Background Recent advances in metabolic engineering have enabled the assembly of extended biosynthetic pathways (exceeding twenty enzymatic steps) for the production of structurally complex natural products, such as aromatic amino acid-derived compounds opioids, vinblastine and catharanthine [[34]1–[35]3]. To optimize precursor flux, researchers have engineered pathway-specific strategies, including (i) overexpression of rate-limiting upstream enzymes, (ii) knockout of competitive branch pathways, and (iii) deregulation of feedback repression with gene mutants [[36]4]. The overexpression of ARO1 (pentafunctional aromatic protein) and ARO2 (chorismate synthase), mutation of feedback-inhibited genes ARO4 (to ARO4^K229L) and ARO7 (to ARO7^G141S) to encode feedback-insensitive DAHP synthase and chorismite mutase, and knockout of ARO10 have been employed to alleviate feedback inhibition, enhance carbon flux into the aromatic amino acid pathway, and produce derived compounds such as flavonoids [[37]5–[38]7] tyrosol [[39]8] salidroside [[40]8] resveratrol [[41]9] 2-phenylethanol (2-PE) [[42]10] and caffeic acid [[43]11]. However, the combinatorial optimized synthesis of complex compounds faces a common challenge of coordinating the supply of multiple precursors especially localizing at competitive branch pathways. Besides, these precursors also participate in several other metabolic reactions to transfer into other compounds [[44]12–[45]14]. The effective push-and-pull of these precursors toward desirable products is difficult, requiring systematic modulation of cellular metabolism. Both the local gene transcription and global gene transcription shall be modulated to boost target product synthesis. A series of studies has demonstrated that the regulation of transcription factor expression can effectively modulate the synthesis of aromatic amino acid derivatives. In S. cerevisiae YS58, the transcription factors Gln3p and Gat1p regulate genes responsible for aromatic amino acid transport and utilization. Overexpression of their encoding genes GLN3 and GAT1 in strain MT21515 increased the production of the phenylalanine derivative 2-PE by 149%, achieving a final titer of 3.59 g/L [[46]15]. Improving the supply of cofactor NADPH and/or SAM could significantly enhance the production of aromatic organic acids [[47]11, [48]16]. Through coordinated engineering of upstream pathways and cofactor metabolism, the engineered strain produced 3.8 g/L ferulic acid, representing a 2.5-fold increase (from 1.5 g/L) compared with the reference strain [[49]11]. By introducing the global transcription factors SPT3 and SPT5, along with Kozak sequences for enhanced L-Pipecolic Acid (L-PA) production using engineered strain Saccharomyces cerevisiae BY4743. Approaches above led to an impressive 8.6-fold increase in L-PA yield, reaching 5.47 g/L in shake flask cultures [[50]17]. Compared with the methods requiring iterative evolution of strains and comprehensive genomic sequencing, editing the sequence of global transcriptional factors (TFs) would offer a better choice to explore genotype-phenotype correlations in a short time [[51]17–[52]19]. Here, we established a strategy combining metabolic engineering optimization and global TF transcriptional engineering to optimize the synthesis and coordination of aromatic amino acid derivatives. We choose to overexpress upstream pathway genes as the rational pathway engineering method. Besides, the sequences of SPT15 (expressing a TATA-binding TF involved in RNA polymerase-II transcription) and/or GCN4 (expressing a TF regulating amino acid biosynthesis and many other functions) are also modulated as global transcriptional engineering method to introduce large-scale tuning to yeast cells (Fig. [53]1). These two methods are utilized to optimize the synthesis of betaxanthin, a tyrosine derived edible pigment, in yeast libraries screened by a visual judgment of the colony’s yellow coloration. This color-judging strategy was also used in previous chromosomal SCRaMbLE-driven genotype-phenotype association studies in yeast [[54]20–[55]23]. In addition, the coordinated synthesis of betaxanthin and a tryptophane derived purple-pigment violacein in yeast libraries is screened by visual judgment of the colony’s gray coloration. Our strategy aims to optimize the coordinated synthesis of multiple derivatives downstream of common aromatic amino acid pathways. Fig. 1. Fig. 1 [56]Open in a new tab Scheme of the strategy of the global transcriptional engineering for the coordinated synthesis of aromatic amino acid derivatives. (a) Schematic diagram of the process of global transcription engineering to regulate amino acid derivatives. (b) Simplified metabolic pathways for the betaxanthin and violacein screening strains. The heterologous pathways to betaxanthin and violacein illustrated are colored red and purple, respectively. (Abbreviations of the metabolites: G-6-P: glucose-6-phosphate; F-6-P: fructose-6-phosphate; G3P, glyceraldehyde 3-phosphate; PEP: phosphoenolpyruvate; Ru5P: ribulose 5-phosphate; R5P: ribose-5-phosphate; X5P, xylulose-5-phosphate; S7P: sedoheptulose 7-phosphate; E4P: erythrose-4-phosphate; DAHP: 3-deoxy-D-arabino-2-heptulosonic acid 7-phosphate; SHIK: Shikimic acid; CHA: chorismic acid; PPA: prephenate; HPP: p-hydroxyphenylpyruvate; ANA: anthranilic acid; CYP76AD*: the mutant CYP76AD; DOD: dopa-4,5-dioxygenase. Genes highlighted in red are overexpressed genes.) Results Modulation of betaxanthin synthesis via global transcriptional engineering In order to study the potential influences of yeast global transcriptional modulation on the synthesis of aromatic amino acid derivatives, the mutant libraries of SPT15 and GCN4 were generated in vitro using error-prone PCR and integrated into the chassis producing betaxanthin (CB) chromosome using homologous recombination to build colony libraries and named CBS and CBG. The betaxanthin pathway was constructed according to DeLoache’s work [[57]24] (details shown in Methods and Additional File [58]1: Text S1). The colony libraries producing different concentrations of betaxanthin, a tyrosine derivative, were screened according to the visual judgment of the different depths of the colony’s yellow coloration. The results showed that more than half of the visually screened strains owning one of the two TF mutants could synthesize higher titer of betaxanthin than the control strain owning only natural TFs (Fig. [59]2a, b and Figure [60]S1). The No. 19 strain with SPT15 mutants (CBS-19) presented the highest betaxanthin titer of 51.2 mg/L, which was 217% of the titer in the control strain (CB). Sequencing of the SPT15 mutant in this strain showed that a R238K amino acid mutation existed in the Spt15p, lying in the C terminus of this protein (Fig. [61]2c). The No. 9 strain with GCN4 mutants (CBG-9) presented highest betaxanthin titer of 44.6 mg/L, which was 189% of the titer in the control strain (CB). Sequencing of the GCN4 mutant in this strain showed that three amino acid mutations, namely S22Y, T51N and L71N, existed in the Gcn4p (Fig. [62]2d). A direct comparison of the culture media also indicated the difference in yellow depth positively correlated with the titer of betaxanthin in each strain (Fig. [63]2e). In order to investigate the effect of both SPT15 mutant and GCN4 mutant to synergistically enhance betaxanthin titer, we co-transfected the two mutants into strain CB and named it CBSG. The results demonstrated that the engineered strain CBSG achieved betaxanthin titer of 46.94 mg/L, which was 12.5% lower than CBS-19 (51.2 mg/L), but 5.2% higher than CBG-9 (44.6 mg/L). All three strains exhibited comparable biomass accumulation profiles, as evidenced by their similar OD[600] values throughout fermentation (Fig. [64]2f). These results suggest that the variant Spt15p^R238K has a better performance for betaxanthin production than the other variants under the test conditions. Fig. 2. [65]Fig. 2 [66]Open in a new tab Modulation of betaxanthin synthesis via global transcriptional engineering driven by mutants of Spt15p and Gcn4p. (a) The mutation libraries of SPT15 were introduced in the yeast chassis producing betaxanthin, then the constructed strain libraries were screened and the betaxanthin production was measured. (b) Similar to figure a), the yeast libraries containing GCN4 mutants were screened and the betaxanthin production was measured. (c) The genotype of the SPT15 mutant in the selected strain CBS-19. (d) The genotype of the GCN4 mutant in the selected strain CBG-9. (e) Comparison of the 50 mL fermentation liquids of the strains CB, CBG-9 and CBS-19. (f) Mutants of SPT15 in CBS-19 and GCN4 in CBS-9 were introduced into the betaxanthin-producing yeast CB, and the betaxanthin production was measured. (g) Detection of the impacts of CBS19-SPT15 mutant on the production of betaxanthin and violacein in the strain containing both synthesis pathways. The asterisk (*) represents the statistical significance (*p < 0.05, ** p < 0.01, *** p < 0.001) To evaluate the effects of transcriptional mutations in the CBS-19-SPT15 mutants-identified through single-pathway screening using be betaxanthin-on other aromatic amino acid derivatives, we introduced another plasmid containing a VioABEDC gene combination [[67]25] for expressing violacein-synthesizing enzymes to construct the control strain CBV and the globally transcribed engineered strain CBV-19S. It was shown that only the synthesis of betaxanthin was enhanced compared to the control strain CBV, but violacein synthesis was not improved (Fig. [68]2g). It was worth noting that the flux towards betaxanthin was not effectively distributed to violacein synthesis. This indicated that the coordinated synthesis of multiple aromatic amino acid derivatives faced difficulty of combinatorial metabolic optimization. The similar phenomenon also occurred in previous report where a SCRaMbLE method was used to screen the optimized tryptophan derived PDV-synthesizing strains showing mere synthesis improvement of tyrosine derived p-coumaric acid [[69]22]. Metabolic optimization synergizes with global transcriptional engineering to coordinate betaxanthin and violacein biosynthesis After exploration of the global transcriptional influences, to explore the effects of global transcription engineering combined with metabolic optimization on betaxanthin production. We chose to directly use the engineered yeast strains of LYTY1 and LYTY5 from Liu’s work as chassis [[70]8]. The LYTY1 owned overexpression of ARO4^K229L andARO7^G141S, and the LYTY5 owned an extra overexpression of TKL1, RKI1, ARO2 based on the LYTY1 (Fig. [71]2a). The SPT15 mutant in CBS-19 which enhances the highest titer of betaxanthin integrated into the chassis producing betaxanthin (LYTY-B) chromosome using homologous recombination and named LYTY-BS. LYTY1-B and LYTY1-BS betaxanthin titer were 36.45 and 39.19 mg/L (Fig. [72]2b). Similarly, the same results were obtained in the LYTY5 series of strains, LYTY5-B and LYTY5-BS betaxanthin titer were 34.44 and 36.21 mg/L (Fig. [73]2c). The detection result showed that the overexpression of upstream pathway genes could improve betaxanthin production by 54.2% in LYTY1 and by 45.8% in LYTY5 (Fig. [74]3a, b). It was worth noting that the betaxanthin production in LYTY5 was lower than that in LYTY1, although LYTY5 owned more upstream pathway optimization. The introduction of the CBS19-SPT15 mutant significantly increased betaxanthin production in the engineered strain, while also reducing biomass accumulation (OD[600]) compared to the parental strain LYTY-B. This suggests that the mutant may cause metabolic burden. Fig. 3. [75]Fig. 3 [76]Open in a new tab Modulation of betaxanthin synthesis via combining local and global transcriptional engineering. (a) Schematic illustration of the design. Red genes without borders denote genes expressed in LYTY1, while red genes with borders represent genes additionally expressed in LYTY5 compared to LYTY1. (ARO4^K229L: feedback-insensitive DAHP synthases; ARO7^G141S: feedback-insensitive chorismate mutase; RKI1: ribose-5-phosphateketol-isomerase 1; TKL1: transketolase 1) (b) The SPT15 mutant in CBS-19 integrated into the chassis producing betaxanthin (LYTY1-B) chromosome and named LYTY1-BS to test their impacts on betaxanthin production. (c) Similar to figure b), the only difference was that the LYTY5 was used instead of the LYTY1. (d) The CBS19-SPT15 mutant was introduced into the LYTY1-BV (containing both betaxanthin and violacein pathways) and named LYTY1-BVS to test their impacts on betaxanthin and violacein production. (e) Similar to figure d), the only difference was that the LYTY5 was used instead of the LYTY1. (f) Detection of betaxanthin and violacein production in the screened strains containing both product synthesis pathways and the mutation libraries of SPT15. All data represent the mean of n = 3 independent biological samples and error bars show standard deviation. Statistical analysis was carried out by using Student’s t-test (*p < 0.05, ** p < 0.01, *** p < 0.001) To further explore the effects of CBS19-SPT15 mutants on betaxanthin and violacein synthesis under the synergistic effect of global transcriptional regulation and metabolic engineering optimization. Both product pathways and CBS19-SPT15 were transformed into the strain LYTY1, presenting 30.4% increment of betaxanthin yield and rare improvement of violacein synthesis (Fig. [77]3c). The same genetic manipulation made the betaxanthin titer increase by merely 4.60% and the violacein titer increase by merely 5.11% in LYTY5 (Fig. [78]3d). It indicated that this combination of CBS19-SPT15 mutant and local pathway optimization was not beneficial to the metabolic flux towards the tryptophan branch pathway. The above experimental results show that transcription factor mutations screened by a single-pathway tyrosine derivative of betaxanthin could not be sufficient to efficiently coordinate the synthesis of the tyrosine derivative betaxanthin and the tryptophan derivative violacein. Based on the above results, a primarily screened TF mutant based on one pathway optimization would not benefit the coordinated synthesis of both products in branch pathways. We selected the SPT15 transcription factor—a key regulator shown to significantly enhance betaxanthin production—to construct an in vitro mutant library. The library was then integrated into the chromosomes of betaxanthin- and violacein-producing strains (designated as CBV strains) via homologous recombination. The resulting engineered strains were named CBVS-A to CBVS-X. It was shown that this library screening was available based on the visual judgment of different depths of gray color mixed with yellow and purple colors. After primary eye-judged screening and further fermentation measurement, we found half of the optimally screened colonies indeed got improved production of both betaxanthin and violacein, 1/3 of all got weakened synthesis of both products, and 1/6 of all got trade-off synthesis of both products (Fig. [79]3e). The optimal strain was named CBVS-1. CBVS-1 produced 31.63 mg/L of betaxanthin, which is 59.9% higher than that of the strain CBV (19.78 mg/L). Additionally, CBVS-1 achieved a violacein titer of 282.27 µg/L, which corresponded to a 66.7% improvement compared to the control strain CBV (169.34 µg/L) (Fig. [80]3f and Figure [81]S3). Sequencing of the SPT15 mutant showed that an isoleucine at position 211 was replaced into asparagine in Spt15p. It was concluded that the combinatorial screening of two pigment synthesis based on global TF engineering were effective for improving both product synthesis. Exploration of the regulated transcripts correlated with the enhanced pigment synthesis To explore the effects of TFs mutation on global transcription, the transcriptome analysis was done for three pairs of strains, namely CB vs. CBS-19, CB vs. CBG-9, and CBV vs. CBVS-1. Analysis of transcriptome results to explore genes with more than 2-fold change in transcript levels. It was found that all three pairs got more than 1000 regulated genes when introducing corresponding TF mutants (Fig. [82]4a). Both introductions of Spt15p^R238K and Gcn4p^S22Y, T51N, L71V mutants in the strain containing betaxanthin pathway caused up-regulation of around 1300 genes and down-regulation of similar number of genes. By contrast, the introduction of Spt15p^I212N in the strain containing both betaxanthin and violacein pathways caused respective up-regulation and down-regulation of around 1000 genes. The Venn analysis showed that the introduction of Spt15p^R238K or Gcn4p^S22Y, T51N, L71V led to a large number of commonly regulated genes as 2260, covering 80% and 85% of respectively total regulated genes (Fig. [83]4b). By contrast, although the pair of CBV vs. CBVS-1 got a sum of 800 regulated genes shared by the other two pairs, it also owned as many as 788 unique regulated targets, covering 39% of the corresponding total regulated genes. Fig. 4. [84]Fig. 4 [85]Open in a new tab Transcriptome analysis of the strains with optimized synthesis of aromatic pigments. (a) The volcano analysis of transcriptome comparison pairs of CB vs. CBS-19, CB vs. CBG-9 and CBV vs. CBVS-1. The dot in red indicates metabolite is up-expressed, and the dot in green indicates metabolite is down-expressed. The numbers under genes are log[2](fold change of transcriptional read count). (b) The Venn analysis of three transcriptome comparison pairs. (c) The pathway enrichment analysis of the regulated transcripts in CBVS-1 compared with CBV. (d) The drastically up-regulated transcripts shared by the CBS-19 and CBG-9 compared with CB. Rich factor indicates ratio of differentially expressed genes to total genes in a pathway; Diff indicates the number of changes in the level of transcription of a gene; bubble color indicates the q-value (red: q < 0.01, orange: 0.01 ≤ q < 0.05), reflecting statistical confidence after false discovery rate (FDR) correction. (e) Seven typical genes were chosen from the up-regulated pathways in CBS-19 and CBG-9 (relative to CB) and transformed as one more cassette into the same strain, and measuring betaxanthin production. All error bars indicate ± standard deviation, n = 3. Statistical analysis was carried out by using student’s t-test (*p < 0.05, ** p < 0.01, *** p < 0.001) The KEGG pathway enrichment was analyzed for every pair of strains. The pairs of CB vs. CBS-19 and CB vs. CBG-9 got highly similar enrichment maps, although the different TFs were engineered (Additional File [86]1: Figure S5-S7). The pathways with relatively higher rich factors were involved in the synthesis of ribosome, glycolysis and general carbon metabolism, indicating a large range of overall regulation of cell proliferation and metabolism. By contrast, the pair of CBV vs. CBVS-1 gained very detailed enriched regulation of several particular pathways, including biosynthesis of amino acids, glycolysis, 2-oxocarboxylic acid metabolism, pentose phosphate pathway, several other amino acid metabolism and organic acid metabolism (Fig. [87]4c). Most of these enriched pathways were metabolically related to the synthesis of aromatic amino acids. The engineered strains exhibited coordinated down-regulation of energy-intensive pathways, particularly through the strategy of inhibiting β-alanine metabolism (CBS-19), fatty acid degradation/TCA cycle (CBG-9), and BCAA synthesis (CBVS-1). This suppression likely redirects carbon flux away from ATP generation and lipid catabolism, leading to a detailed enrichment of specific metabolic pathways. This shift prioritizes precursor availability for biosynthesis, resulting in a significant increase in the production of aromatic natural pigments betaxanthin and violacein (Additional File [88]1: Figure S5-S7). It was concluded that the Spt15p^I212N mutant caused highly oriented and focused regulation of transcriptome for multiple pigment synthesis downstream of aromatic amino acid pathways. To identify potential key regulatory nodes, we analyzed each pairwise comparison for transcripts exhibiting marked differential expression, applying a stringent threshold of > 30-fold upregulation. The pair of CBV vs. CBVS-1 did not gain interesting regulated genes. However, the pairs of CB vs. CBS-19 and CB vs. CBG-9 got exactly the same list of up-regulated targets with more than 30-fold enhancement (Fig. [89]4d). To determine these obviously up-regulated pathways in CBS-19 contributed to the improved betaxanthin production, we chose the above seven typical up-regulated genes and transformed their individual expression cassette on plasmids into CBS-19, which are named CBS-19-1 ~ CBS-19-7. The results showed that most of the transformed strains with recombinant plasmids (from CBS-19-3 to CBS-19-7) gained higher betaxanthin production than the initial strain CBS-9 with blank plasmid (CBS-19-0). These genes except PDC5 and PDC1 did increase the flux of L-TYR synthesis. All the cells’ OD[600] values were slightly reduced when cultivated in the multiple defective media perhaps owing to the influence of introducing a new plasmid (Fig. [90]5e). The top two least significantly up-regulated targets were PDC5 and PDC1 encoding different isozymes of pyruvate decarboxylase to commit the end product of glycolysis, pyruvate, to ethanol production. PDC5 and PDC1 could also decarboxylate other 2-oxo acids such as indolepyruvate and 2-keto-3-methyl-valerate, and this activity contributed to the catabolism of the amino acids such as isoleucine, phenylalanine, tryptophan, and valine [[91]12]. In several reports, PDC5 was knocked out to enhance the flux towards tyrosine derivatives [[92]5, [93]11, [94]26, [95]27]. This is consistent with the experimental results. HXT1 and HXT3 encoded glucose transporters which were beneficial to glucose uptake [[96]28]. Ncw2p and Fit2p participated in cell wall remodeling and protection. Dak2p catalyzed the second step in the metabolism of glycerol to glycerone phosphate, converting dihydroxyacetone, a toxic substrate. It was concluded that the different Spt15p^R238K or Gcn4p^S22Y, T51N, L71V introduction caused the common up-regulation of these special functions to improve cell fermentation and protection for boosting betaxanthin production. Fig. 5. [97]Fig. 5 [98]Open in a new tab Comparison of the transcriptional regulation of direct upstream carbon metabolism caused by different TF mutations. The main genes encoding the enzymes participating in the carbon metabolism upstream of the aromatic amino acid synthesis were listed. The gene names were labelled in a gray box. The change folds of up-regulated gene transcription were labelled in red number and the change folds of down-regulated gene transcription were labelled in blue number. The yellow, green, and purple boxes represented the pairs of CB vs. CBG-9, CB vs. CBS-19 and CBV vs. CBVS-1, respectively Evaluation of the tuning of carbohydrate metabolism caused by global transcriptional modulation The impacts of different TF mutants on the carbohydrate metabolism from the substrate glucose to the end aromatic amino acids were analyzed (Fig. [99]5). In general, all three compared pairs presented their own different features. Most of the glycolysis genes were up-regulated when Spt15p^R238K was introduced in the strain producing both betaxanthin and violacein except for HXK2 encoding glucose phosphorylase. Especially, the two reaction steps from glyceraldehyde-3P to glycerate-3P were largely enhanced where the main genes of TDH1, TDH3 and PGK1 were highly up-regulated by dozens of folds. The glycolysis pathway was also overall up-regulated in CBS-19 compared with CB. As for the pentose phosphate pathway, the three pairs showed quite different features. For example, the ZWF1 was only not down-regulated in the pair of CB vs. CBG-9. Both genes of GND1 and GND2 showed opposite regulation modes in the pairs of CB vs. CBS-19 and CB vs. CBG-9 compared with the pair of CBV vs. CBVS-1, but the activity balance of this reaction from gluconate-6P to ribulose-5-P was guaranteed. The similar phenomenon also occurred in the reactions of transferring ribulose-5-P to the other C5 sugars of ribose-5-P and xylulose-5-P. In the last aromatic amino acid synthesis pathway, the key enzyme of Aro4p was obviously up-regulated in the pairs of CB vs. CBS-19 and CB vs. CBG-9, and another enzyme Aro2p charging multi-step reactions was only up-regulated in the pair of CBV vs. CBVS-1. The regulation of other genes showed that CBS-19 relatively performed the best. In all, the introduction of different TFs more or less enhanced the ability of carbon source to end product metabolism in yeast, especially the introduction of Spt15p^R238K. Optimization of fermentation conditions to improve the betaxanthin production To assess the potential of fermentation-based betaxanthin production, we used YNB medium and evaluated the optimal fermentation conditions where glucose, glycerol, and sucrose were used as the sole carbon sources. Glucose as a carbon source contributed to a higher titer of 82.4 mg/L betaxanthin in the 50 mL flask fermentation of the strain CBS-19 (Fig. [100]6a, Additional File [101]1: Figure S8-S9). Subsequently, glucose was used as the sole carbon source and was supplemented every 12–24 h (Fig. [102]6a). Supplementing glucose every 12 h resulted in the optimal fermentation titer of 113.1 mg/L. Therefore, the YNB medium supplemented glucose every 12 h was used as the nutrient of yeast fermentation for betaxanthin production. Fig. 6. [103]Fig. 6 [104]Open in a new tab Optimization of yeast fermentation conditions for betaxanthin production. a) The fermentation conditions using different types of sugars, pH or sugar supplementation frequencies were tested on the strain CBS-19. b) c) d) An amplification fermentation of CBS-19, CBG-9 and CBVS-1 was respectively done in 1 L flask under the optimized condition. All fermentations were carried out at 30 ℃, 220 rpm, pH 6. All error bars indicate ± standard deviation, n = 3. Statistical analysis was carried out by using Student’s t-test (*p < 0.05, ** p < 0.01, *** p < 0.001) Finally, Fed-Batch fermentation conditions were established to maximize the betaxanthin production. To test the amplified fermentation performance, the strains of CBS-19, CBG-9, and CBVS-1 were cultivated in the above optimized nutrient, respectively, in 1 L flasks (Figs. [105]6b-d, Additional File [106]1: Figure S10). Within the first 24 h, yeast strains grew rapidly, along with the rapid increment of betaxanthin production and the consumption of sugar (decreasing from 20 g/L to 1.1 g/L). At this time point, glucose was detected and supplemented every 12 h to restore the glucose concentration to 20 g/L. Afterwards, the production of betaxanthin in flask continued to increase, reaching its maximum within 84 h. The peak production of CBS-19 was 208.45 mg/L, and that of CBG-9 and CBVS-1 were respectively 164.98 mg/L and 178.60 mg/L. Afterwards, fermentation entered a period of decline, and continuous sugar supplementation alone could not increase the titer of betaxanthin. Additional process optimization might be needed to further improve production. Discussion In many scenarios, rational engineering of pathways is not enough to meet the requirement of optimal synthesis of multiple downstream derivatives, especially when these metabolites are further involved in many other metabolic functions. Therefore, global transcriptional engineering is needed to promote the potentials of various metabolic pathways and cellular functions to enhance target product synthesis. In this work, we develop a strategy to optimize the synthesis of different aromatic amino acid derivatives in yeast by combining metabolic engineering optimization and global engineering of TF Spt15p and Gcn4p. In detail, the edible pigment betaxanthin was synthesized as a tyrosine derivative, and violacein is synthesized as a tryptophan derivative. The result showed that our strategy is effective to boost not only the synthesis of betaxanthin but also the coordinated synthesis of both competitive pathway products. The synthesis of betaxanthin is improved by 217% when a Spt15p^R238K mutant is introduced into the strain and by 189% when a Gcn4p^S22Y, T51N, L71V mutant is introduced. Ultimately, the coordinated biosynthesis of both betaxanthin and violacein was improved by more than 50% when a Spt15p^I212N replaces the yeast natural Spt15p. The introduction of particular TF mutants is not only beneficial to target end-product synthesis but also beneficial to several related metabolisms. The replacement of natural Spt15 with a Spt15p^I212N mutant contributes to the enriched regulation of central carbon metabolic pathways and amino acid metabolic pathways, including glycolysis, pentose phosphate pathway, biosynthesis of amino acids, 2-oxocarboxylic acid metabolism, etc. By contrast, although the respective introduction of Spt15p^R238K or Gcn4p^S22Y, T51N, L71V does not lead to enriched pathway regulation, there occurs greatly up-regulation of special targets including pyruvate decarboxylase Pdc5p and Pdc1p, glucose transporters Hxt1p and Hxt3p, and cell wall protecting protein Ncw2p and Fit2p. Using these seven genes as overexpression targets, it was demonstrated that most of the targets contributed to the improved production of betaxanthin. In all, different TF mutations contribute to different tuning modes of cellular overall transcription in a generalized way or a focused way. Most of the genes participating in the carbohydrate metabolism from glucose to the end aromatic amino acids are more or less up-regulated especially when Spt15p^R238K replaced its corresponding natural TF. Finally, this strain owning Spt15p^R238K mutant is able to use intermittent glucose carbon sources and produce 208 mg/L of betaxanthin in 1 L flask fermentation. In previous studies, the respective 17 mg/L, 30.8 mg/L and 134.1 mg/L betanin titers were achieved by engineering yeast [[107]29–[108]31]. To the best of our knowledge, our study obtained highest reported betaxanthin titer in yeast to date, although it was slightly lower than the highest reported 287.69 mg/L titer in E. coli [[109]32]. In addition, the precursors synthesizing tryptophan had the potential to be converted into tyrosine, according to a recently evolved enzyme TrpB working in vitro [[110]33]. Therefore, our coordinated pigment synthesis mode may promote the optimization of tyrosine derivative synthesis in the future. While this study focuses on the heterologous biosynthesis of betaxanthins, we emphasize that the complexity inherent in their biosynthetic pathway renders them an ideal model for exploring global engineering strategies of more complex natural products. Firstly, the upstream pathway intermediates such as CHA serve as common precursors for the biosynthesis of major classes of aromatic natural products, such as caffeic acid, 5-hydroxytryptophan, naringenin and derived flavanoids [[111]34]. Secondly, regarding spatial compartmentalization complexity, key reactions are distributed across distinct cellular compartments: L-DOPA is synthesized in the cytosol (via tyrosine hydroxylase), betalamic acid is generated in the endoplasmic reticulum (through CYP76AD1-mediated oxidation), and several derived products are stored in the vacuole [[112]32, [113]35, [114]36]. This necessitates engineering of intercompartmental transporters and coordination of redox balance, with a complexity isomorphic to the compartment synthetic pathway of various terpenoids, alkaloids [[115]37, [116]38]. This spatiotemporal dynamic complexity may be beneficial to the synthesis of complex natural products at the end by modifying the global transcription of cells. There are lots of aspects that can be further improved for the strategy. Firstly, only the direct upstream pathway optimization is used in this work. In recent reports, several other targets of not quite directly related were rationally engineered to enhance the production of aromatic derivatives [[117]7, [118]11, [119]39, [120]40]. Besides, recent work also proved that rational engineering was beneficial to the synthesis of red pigment betanin [[121]14]. These gene manipulations can be further combined with our global engineering strategy. Secondly, our selected TF mutants only contain replacement 1 to 3 amino acids based on one round mutation. Maybe iterative mutation will accumulate more amino acid replacements and contribute to further enhanced production of target products. Thirdly, the error-prone PCR method is not a high-efficiency way to build mutation library, especially not supporting automatic high-throughput library construction and screening. A recent automatic evolving strategy has utilized an orthogonal misreading DNA polymerase to amplify the specified gene sequence, which is able to build large libraries along with different cell proliferation generations [[122]41]. This or other new method shall be further beneficial to TF gene library construction and chassis overall engineering. Conclusions In this work, we engineer both the local pathways and global TFs of Spt15p and Gcn4p to boost the synthesis of aromatic amino acid derived pigments in yeast. The results demonstrate the availability of global TF engineering for enhancing the synthesis of a tyrosine derived pigment, betaxanthin, and a tryptophan derived pigment, violacein. The coordinated synthesis of pathway products is also realized in combination with the tuning of upstream carbon metabolism and several other related organic acid metabolic pathways. The strategy is proved a promising tool for optimizing multiple competitive pathway end products, especially when these products were both important to cells or value-added in food applications. Materials and methods Strains and media The S. cerevisiae strain CEN.PK2-1 C (MATa; leu2-3,112; ura3-52; trp1-289; his3-Δ1 MAL2-8c SUC2) was used as the chassis for exogenous DNA construct transformation and for the synthesis of aromatic amino acid derived pigments. Escherichia coli strain DH5α was used for shuttle plasmid construction. YPD medium containing yeast extract (10 g/L), peptone (20 g/L) and D-glucose (20 g/L) was used to cultivate yeast strains ready for exogenous DNA transformation. Synthetic complete (SC) medium (0.67% yeast nitrogen base without amino acids, 2% glucose, and appropriate amino acid drop-out mix), lacking certain amino acid, was used to select and cultivate yeast transformants. Some other types of sugar were used to replace glucose or extra glucose was supplemented in some particular experiments (details shown in Results). For formation in flask, the yeast colonies on agar plates were picked and cultivated overnight in appropriate liquid YPD or SC medium to get seed culture, and then were inoculated into 50 mL/250 mL/1 L SC-marker (amino acid) medium with an initial OD[600] of 0.1 and cultivated at 30 °C and 250 rpm for 96 h under aerobic fermentation. All the fermentation experiments were performed in triplicate.The constructed yeast strain is shown in Table [123]1. Table 1. List of yeast strains constructed in this study Strain name Genotype CB CEN.PK2-1 C, pRS425K-CYP76AD1^W13L F309L-DOD CBS-19 CEN.PK2-1 C, SPT15::SPT15^R238K URA3, pRS425K-CYP76AD1^W13L F309L-DOD CBG-9 CEN.PK2-1 C, GCN4::GCN^S22Y, T51N, L71V HIS3, pRS425K-CYP76AD1^W13L F309L-DOD CBV CEN.PK2-1 C, pRS425K-CYP76AD1^W13L F309L-DOD, pRS413-VioABEDC CBV-19 S CEN.PK2-1 C, SPT15::SPT15^R238K URA3, pRS425K-CYP76AD1^W13L F309L-DOD, pRS413-VioABEDC CBVS-1 CEN.PK2-1 C, Spt15::Spt15p^I212N URA3, pRS425K-CYP76AD1^W13L F309L-DOD, pRS413-VioABEDC LYTY1-B LYTY1, pRS425K-CYP76AD1^W13L F309L-DOD LYTY1-BS LYTY1, SPT15::SPT15^R238K URA3, pRS425K-CYP76AD1^W13L F309L-DOD LYTY1-BV LYTY1, pRS425K-CYP76AD1^W13L F309L-DOD, pRS413-VioABEDC LYTY1-BVS LYTY1, SPT15^R238K URA3, pRS425K-CYP76AD1^W13L F309L-DOD, pRS413-VioABEDC LYTY5-B LYTY5, pRS425K-CYP76AD1^W13L F309L-DOD LYTY5-BS LYTY5, SPT15::SPT15^R238K URA3, pRS425K-CYP76AD1^W13L F309L-DOD LYTY5-BV LYTY5, pRS425K-CYP76AD1^W13L F309L-DOD, pRS413-VioABEDC LYTY5-BVS LYTY5, SPT15^R238K URA3, pRS425K-CYP76AD1^W13L F309L-DOD, pRS413-VioABEDC [124]Open in a new tab Construction of CYP76AD1 and DOD transcriptional units The CYP76AD1 and DOD transcriptional units (TUs) were designed based on the codon bias optimization of the original sequences (Additional File [125]1: Text S1). Then, the oligos were designed and synthesized (by GENEWIZ) for the assembly of the whole double-TUs sequence. An approach of “touch-down PCR” and “forward PCR” were used to amplify the four parts of the whole using PrimeSTAR Max DNA polymerase (Takara, China). The first part was assembled based on oligo (CYP76AD1-DOD)-1 to (CYP76AD1-DOD)-18; the second part was assembled based on oligo (CYP76AD1-DOD)-19 to (CYP76AD1-DOD)-36; the third part was assembled based on oligo (CYP76AD1-DOD)-37 to (CYP76AD1-DOD)-54; the fourth part was assembled based on oligo (CYP76AD1-DOD)-55 to (CYP76AD1-DOD)-72. All the oligos used were listed in Additional File [126]1: Table S2. The touch-down PCR was manipulated as follows. The synthesized oligos was diluted to a common concentration of 10 mM with ddH[2]O and 6 µL of each oligo was mixed with ddH[2]O to get a total volume of 200 µL. The above mixture was used as DNA template in ratio of 1:10 to build the total PCR system. A 50 µL system contained 5 µL DNA template, 25 µL PCR mix (PrimeSTAR, Takara) and 20 µL H[2]O. The PCR program was: 95 °C, 5 min, one cycle; 95 °C, 30 s, each cycle decreased by 0.5 °C from 65 °C, 30 s, 72 °C, 30 s, and the above three steps were carried out for 30 cycles; 72 °C, 10 min, one cycle; 4 °C for storage. After that, a forward PCR was operated. The above PCR product solution was directly used as DNA template in ratio of 1:10 to build the forward PCR system. A 50 µL system contained 5 µL DNA template, 2.5 µL forward primer (synthesized by GENEWIZ) and 2.5 µL reverse primer (synthesized by GENEWIZ), 25 µL PCR mix (PrimeSTAR, Takara) and 15 µL H[2]O. The PCR program was: 95 °C, 5 min, one cycle; 95 °C, 30 s, 55 °C, 30 s, 72 °C, 30 s, and the above three steps were carried out for 30 cycles; 72 °C, 10 min, one cycle; 4 °C for storage. Then, each PCR part was purified after gel electrophoresis. Finally, the whole sequence of CYP76AD1 and DOD TUs was assembled and ligated onto a shuttle vector pRS425K through the MultiF Seamless Assembly Mix (ABclonal, China) (Additional File [127]1: Table S1). Then, the reaction liquid was transformed into E. coli cells and selected on the agar plates adding kanamycin (100 µg/mL). Construction of TF mutation libraries The mutation was done by error-prone PCR. The PCR fragments were amplified via PCR using 2 × Rapid Taq Master Mix (Vazyme, China). The 100 µL PCR system contained 10 µL 10× FastTaq Buffer, 10 µL EP dNTP mixture (containing 4 mM dTTP, 4 mM dCTP, 3 mM dGTP and 3 mM dATG), 2 µL FastTaq enzyme, 0.15 mM MnCl[2], 1 µL template DNA, 0.05 mM forward primer, 0.05 mM reverse primer, and ddH[2]O for making up the remaining volume. The reaction was operated as follows: initial template denaturation at 94 °C for 3 min, followed by 33 cycles of denaturation at 94 °C for 30 s, primer annealing at 50 °C for 30 s and elongation at 72 °C for 30 s, following that, a last amplification at 72 °C for 10 min was run. Then, the mutation library sequences of SPT15 or GCN4 were assembled with other homologous arms and selection markers to get the constructs of “upstream promoter sequence-SPT15 mutation CDS-URA3 marker-downstream terminator sequence” and “upstream sequence-GCN4 mutation CDS-HIS3 marker-downstream terminator sequence” (Additional File [128]1: Text S1). Yeast transformation A classical LiAc/SS carrier DNA/PEG method was used for DNA transformation into S. cerevisiae cells. The yeast colony on agar plates was picked and cultivated overnight in appropriate liquid YPD or SC medium, and then was inoculated a fresh liquid medium by the volume ratio of 1:10. After 4 h, the yeast cells were collected by centrifugation and washed using ddH[2]O by twice. Then, 0.1 M LiAc was added to deal with the cells for 5 min to make the cells get into competent stage. Subsequently, the whole 360 µL transformation system was prepared following this order: the 74 µL target DNA mixture, 10 µL ssDNA, 240 µL 50% PEG3350 and 36 µL 1 M LiAc. The transformed yeast colonies were cultivated on selection agar plates at 30 ℃ for 3–5 days. Selection of yeast colony libraries The strain libraries on agar plates were screened, selected and sequenced. The genome of selected yeast cells was extracted using conventional methods. The GCN4/SPT15 mutants in all five sequential cassettes were cloned from genome by a normal PCR method. Also, the integration site was verified using the upstream primer pair of “GCN-F”,“GCN4up500-F”, “SPT15up500-F”and“SPT15-F”, and downstream primer pair of “GCN-R”, “GCN4Down500-R”,“SPT15Down500)-R”and“SPT15-R”. The amplificated DNA products were purified after gel and sequenced. Betaxanthin extraction and measurement Fermentation of the colonies directly selected from agar plates were performed in 50-mL tubes containing 20 mL media. The culture was performed in a shaking incubator (HNY-203T, Honor, Tianjin, China) at 30 °C and 250 rpm for 72 h. After fermentation, the betaxanthin content was determined, 1 ml of cultivation broth was transferred to a 2 ml ep tube containing 0.5 ml glass beads. The cells were lysed with vigorous shaking. After cell disruption, the debris was spun down (11,000 × g, 10 min) in a centrifuge and the betaxanthin content in the supernatant analyzed. The absorbance of the sample at 480 nm as well as the strain OD[600] were measured using a microplate reader (PerkinElmer, Instrument Co., Ltd., USA). Fermentation of 50 mL flask and 1 L flask for further optimization of fermentation conditions (details shown in Results). Transcriptome analysis Five strains including three mutants CBS-19, CBG-9 and CBVS-1as well as the control strain CB and CBV were subjected to transcriptome analysis using RNA sequencing. The nucleic acid was extracted by CLB + Eidley RN40 kit (Manufacturer: Eidley, Model: RN40)/TRIzol reagent. The concentration of the extracted whole-cell RNA was measured with Qubit RNA Assay Kit in Qubit 3.0 Fluorometer (Life Technologies, CA, USA). Then an amount of 3 µg RNA per sample was used as input. Sequencing libraries were generated using Illumina NovaSeq X plus platform (San Diego). Finally, the PCR products were purified using AMPure XP system and the library quality was assessed by Agient2100, LabChip GX system. Then the libraries of PCR products were sequenced on an Illumina Hiseq 4000 platform and 150 bp paired-end reads were generated. The data analysis was all done in Tsingke Biotechnology Company. Especially, to take insight to the change of phenotype, the enrichment of differential expression genes in KEGG pathways ([129]http://www.genome/jp/kegg/) and GO ([130]http://www.geneontology.org/) was tested by Phyper software. Transcriptomics data uploaded to the NCBI SRA database (SUB14835906). Electronic supplementary material Below is the link to the electronic supplementary material. [131]Supplementary Material 1^ (2.3MB, docx) Abbreviations TF: transcriptional factor TU transcriptional unit CB the yeast CEN.PK2-1 C strain containing betaxanthin synthesis pathway CBS the yeast CEN.PK2-1 C strains containing betaxanthin synthesis pathway and SPT15 libraries CBG the yeast CEN.PK2-1 C strains containing betaxanthin synthesis pathway and GCN4 libraries CBV the yeast CEN.PK2-1 C strains containing both betaxanthin and violacein synthesis pathways CBVS the yeast CEN.PK2-1 C strains containing both betaxanthin and violacein synthesis pathways and SPT libraries LYTY-B the yeast LYTY strain containing betaxanthin synthesis pathway LYTY-BS the yeast LYTY strains containing betaxanthin synthesis pathway and CBS-19-SPT15 mutant LYTY-BV the yeast LYTY strains containing both betaxanthin and violacein synthesis pathways LYTY-BVS the yeast LYTY strains containing both betaxanthin and violacein synthesis pathways and CBS-19-SPT15 mutant Author contributions H. X., D. L. and Y. C. designed and conducted the experiments, analyzed the data. H. X. and D. L. drafted the manuscript. M.L. assisted most experiments and analyzed the data., D. T., D. L. and H. W. supervised the whole work, revised the manuscript. D. L. and H. W. offered financial support. Funding This work is funded by the National Key Research and Development Program of China (2022YFC2106100); National Natural Science Foundation of China for Excellent Young Scholars (32122047); Tianjin Natural Science Foundation for Distinguished Young Scholars (23JCJQJC00210); Beijing-Tianjin-Hebei Basic Research Cooperation Project of Beijing Natural Science Foundation (23JCZXJC00370); Key Program of Tianjin Natural Science Foundation (22JCZDJC00230). Data availability No datasets were generated or analysed during the current study. Declarations Ethics approval and consent to participate Not applicable. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Huimin Xue, Mingshan Li and Yuhui Cui contributed equally to this work. Contributor Information Dongkui Tian, Email: 80079052@qq.com. Duo Liu, Email: duo_liu0@tju.edu.cn. Hanjie Wang, Email: wanghj@tju.edu.cn. References