Matrix regulation: a plug-and-tune method for combinatorial regulation in Saccharomyces cerevisiae

The ability to fine-tune a pathway with multiple genes remains challenging in the field of life sciences and metabolic engineering^1,2. Underexpression of rate-limiting enzymes often restricts the metabolic flux of the whole pathway, whereas overexpression of certain pathway enzymes beyond what is needed would cause both accumulated intermediates and a waste of cellular resources. A typical approach is therefore overexpressing all genes in the target pathway, followed by multiple rounds of the design-build-test cycles until the optimal expression level of each enzyme is achieved. This overall process is laborious and time-consuming.

To dial in the correct amount of enzymes in a metabolic pathway, two main strategies based on library screening have been developed. One is directly generating genomic diversity within a set of functional elements, e.g., promoters³, ribosome binding sites⁴, 5′ untranslated regions (5′ UTR), and terminators⁵, which may cause untargeted mutations because of massive genome editing. The other approach is combinatorially assembling diverse regulatory modules, such as transcription factors⁶, Kozak variants⁷, and constitutive promoters^8,9, which may be restricted by current molecular engineering techniques, such as the amount of DNA that can be transformed into a given host strain or the number of genome integrations that can be multiplexed, etc. Thus, both combinatorial strategies are often accompanied by high-throughput assays, which are unfortunately unavailable for most target molecules. Therefore, a low-cost, facile, and less context-dependent fine-tuning technology is still needed to facilitate the optimization of cell factories and the basic research into the relationships between genotypes and phenotypes.

With the development of Clustered Regularly Interspaced Short Palindromic Repeats (CRISPR) mediated transcriptional reprogramming^10,11 and advanced gRNA assembly methods^12,13, CRISPR-based combinatorial regulation strategies have been reported^14,15. Yet, CRISPR-based activations rely on Cas proteins binding to the regions upstream of the 5′ UTRs that often composed of AT-rich sequences¹⁶, whereas their PAM sequences, such as NGG for the most commonly used SpCas9 (Cas9 from Streptococcus pyogenes)¹⁷, often restrict the available operating sites of a given promoter. Furthermore, the dynamic range of activation is limited by the inefficiency of activation domains (ADs) in yeast¹⁸. For example, the hybrid tripartite activation domain VP64-p65-Rta (VPR) in S. cerevisiae could only upregulate randomly picked genes by 2-fold^18,19, while it enables target gene activation by over 10,000-fold in mammalian cells²⁰.

Here, we develop Matrix Regulation (MR), a plug-and-tune technology that allows fine-tuning of pathways involving up to eight genes in S. cerevisiae. Briefly, to realize the combinatorial assembly of multiple groups of gRNAs and efficient processing of the gRNA arrays, we characterize multiple tRNAs and propose mixed tRNA arrays. To achieve a broadened targeting scope and ensure sufficient gRNA varieties within a given promoter, we assess the efficiency of dCas9 variants and identify dSpCas9-NG with efficient regulation capacities over all NG PAMs in S. cerevisiae. To address the limited regulation dynamic range caused by insufficient ADs in S. cerevisiae, we characterize 101 candidate activation domains, followed by further enhancing the best one by 3-fold through random mutagenesis and high-throughput screening. Finally, to prove the utility of our toolkit in biological research, we apply MR for optimizing the production of both squalene (without the high-throughput screening method reported) and heme (with the 96-wellplate screening method reported). Briefly, through a single-step assembly and transformation, we generate a combinatorial library with a size of 6⁸, in which all eight genes in the mevalonate (MVA) pathway are up-regulated at six activation levels, respectively. Through random picking of 50 colonies without the help of any phenotype aids, squalene production is enhanced by 37-fold. Moreover, we generate a two-dimensional combinatorial library of the heme biosynthesis pathway with the size of 6⁸, and through random picking of 500 colonies, heme production is enhanced by 17-fold.

Hybrid tRNA arrays enabled the construction of combinatorial gRNA array libraries

In this study, we aim to build a combinatorial regulation method with a library size of 6⁸, fine-tuning pathways with up to eight genes, with each gene regulated at six levels based on different gRNA positions. Yet, the current gRNA assembly method needs to be optimized for combinatorial purposes.

For genome engineering in yeast, recombinant plasmids generally have to be transformed into Escherichia coli (E. coli) for plasmid amplification and verification, which is both time-consuming and, more importantly, results in a considerable loss of combinatorial assembly possibilities. The Lightning GTR-CRISPR method developed for multiplexed genome knockouts which employs a gRNA-tRNA^Gly array can skip the E. coli cloning step and be directly transformed into yeast¹³. However, this system cannot yet facilitate E. coli-less yeast-based combinatorial regulations. In the Lightning GTR-CRISPR system, only tRNA^Gly is used to slice gRNAs of all gene targets, and the recognition site of the Golden Gate assembly enzymes could only be selected within the 20-base pair (bp) targeting sequences, which in turn imposes constraints on the combinatorial assembly of multiple groups of gRNAs. Moreover, using the same tRNA^Gly for all gRNAs may cause more possibilities of mis-constructed plasmids¹³. Briefly, for each gene to be regulated, the same 76-bp scaffold RNA has to be assembled, which together with the 71-bp tRNA^Gly would be 147 bp. Whereas in S. cerevisiae 100 bp homology arms would cause efficient gap repairs, especially in MR where all parts are directly transformed into yeast rather than preassembled in vitro. Thus, the gRNA-tRNA array needs to be optimized with varied tRNAs to facilitate the assembly of varied targeting sequences of one gene (Fig. 1a).

To facilitate subsequent assembly and enable a compact architecture, we selected seven short tRNAs with four distinct nucleotides at the 3′ end, including tRNA^Leu, tRNA^Gln, tRNA^Asp, tRNA^Arg, tRNA^Lys, tRNA^Thr, and tRNA^Ser. Sequences for tRNAs used in this study can be found in Supplementary Data 1. The five-gene gRNA-tRNA array format¹³ disrupting CAN1, LYP1, TRP2, FAA1, and ADE2 genes was selected for tRNA characterizations. Based on our previous knowledge, in this format, the transcription efficiency of the SNR52 promoter becomes insufficient when the number of gRNA transcripts increases to more than five¹³. Therefore, we replaced the last tRNA^Gly with candidate tRNAs (Fig. 1b). The disruption of ADE2 results in a red phenotype of the cells, allowing for rapid assessment of the splicing efficiency of the tRNA under investigation. The results indicated that all tested tRNAs exhibited comparable gene disruption efficiencies with tRNA^Gly, due to the similar RNA splicing efficiencies of these native tRNAs.

Identification of dSpCas9-NG with broadened PAM recognitions for combinatorial regulations

The widely used CRISPR-Cas9 system from S. pyogenes recognizes the NGG PAM sequence. Meanwhile, activation for the gene of interest is typically achieved by targeting upstream of 5’’UTR regions, which tend to be AT-rich^16,21, posing challenges in selecting alternative target sequences for combinatorial regulations. To expand targeting scopes and enhance the magnitude of regulation, SpCas9-NG and the xCas9 with NG PAMs reported in human cells^22,23 and demonstrated in S. cerevisiae with editing capacities²⁴ were tested (Fig. 2a).

To avoid gRNA bias, we first constructed a platform strain, integrated into the genome XII-2 locus²⁵ with a 140 bp minimal CYC1 promoter, the mCherry gene, and the ADH1 terminator. We then fixed a 20 bp targeting sequence and replaced its CGG PAM with all other possible 15 NG PAMs, generating in total 16 platform strains for Cas9 variant characterizations. As shown in Fig. 2b, the results suggested that dxCas9-VPR did not exhibit significant regulatory capacities for the other 12 NG PAMs except for NGG PAMs in S. cerevisiae, nor did dCas9-VPR, while dSpCas9-NG-VPR exhibited efficient regulatory potencies.

These results also suggested that, under all conditions, the highly effective activator (VPR) in mammalian cells could only achieve a 2-fold activation in S. cerevisiae, indicating insufficient combinatorial activation potency that needs optimization, which is consistent with the results previously reported in S. cerevisiae^18,19.

Enhancement of the CRISPR-AD VPR through high-throughput screening

The use of a protein scaffold system that recruits multiple copies of ADs to amplify gene expression signals has been reported^18,26,27. Yet, this approach substantially increases the size of the dCas-AD protein that may impose a cellular burden and reduce the cell viability during multiplexing^18,28. Thus, we chose to increase the dynamic range of CRISPR-based activations by optimizing its activation domains.

We started by identifying potent endogenous activators from 101 candidates through the Saccharomyces Genome Database (SGD). Specifically, three criteria were taken to identify potential DNA sequences: (1) annotated as transcription factors, activators, or regulators, (2) with no reported functions related to repression or inhibition, (3) for those containing DNA-binding domains, the remaining parts were investigated separately. Details of these 101 candidate activators can be found in Supplementary Data 2.

The activation efficiencies were tested based on their abilities to activate the Zeocin resistance gene ZeoR through spot growth assays (Supplementary Fig. 1). Specially, potential candidates from Taf4, Pdr1, Snf12, and Med2 exhibited strong activation efficiencies. Taf4 forms part of the TFIID complex involved in RNA polymerase II transcription initiation²⁹. Pdr1 is a zinc cluster-containing transcription factor (TF) that regulates the pleiotropic drug response³⁰. Snf12 is a subunit of the SWI/SNF chromatin remodeling complex involved in the transcriptional regulation of 5% of the genes in yeast³¹. These three domains have not yet been reported as CRISPR ADs. Yet, results showed that all tested endogenous activators exhibited weaker ZeoR activation capabilities compared with VPR, which may be due to the complexity of in vivo regulatory mechanisms suppressing their effectiveness as CRISPR ADs³². In contrast to our finding, Med2, a subunit of the RNA polymerase II mediator, slightly outperformed VPR as a CRISPR AD in a recent report¹⁸.

Next, we sought to enhance the commonly used activation domain VPR through random mutagenesis and screening, activating both mCherry and ZeoR using the same gRNA targeting CYC1p. Briefly, the transformants were screened by the Fluorescence-Activated Cell Sorting (FACS), and cells exhibiting intense mCherry signals were sorted into 96-well plates with Zeocin. Then, the more efficient VPR variants were identified by spot growth assay and real-time qPCR (Fig. 3a).

Considering that the dCas9-VPR activated ZeoR expression could already endow cell growth on plates with 800 mg/L Zeocin (Supplementary Fig. 1), while Zeocin concentration higher than this level caused false positive growth, we first attenuated the activity of ZeoR protein. Briefly, the secondary structure of ZeoR protein was predicted using Phyre2, followed by alanine substitutions introduced at 13 conserved charged sites (Fig. 3b). Next, spot growth assays were performed and identified mut8 and mut13 with weak resistance to 50 mg/L of Zeocin (Fig. 3c), while upon VPR activation mut13 exhibited an improved resistance to 100 mg/L of Zeocin (Fig. 3d). Then, during FACS sorting, cells with the VPR activation group displayed increased mCherry signals and could grow at Zeocin concentrations up to 100 mg/L, while cells without VPR activation could proliferate only at concentrations up to 50 mg/L (Fig. 3e). This result also suggested that positive VPR mutants could be isolated using high-throughput screening.

A total of three consecutive rounds of FACS screening were carried out, followed by selections in 96-well plates and spot growth assays with increased Zeocin concentrations (Fig. 3f). Plates without Zeocin were used as growth control. For the initial round, cells with the top 5% mCherry signal were sorted to plates with 200 mg/L or 300 mg/L Zeocin, leading growth of 22 colonies (0–2 for each plate) after three days of cultivation. Spot growth on Zeocin of these colonies indicated that the VPR mutant 13 (named e13), was the most potent mutant after Round 1 (Supplementary Fig. 3). To avoid potential genome mutation effects, e13 together with the other seven variants were verified by reconstructing the mutated VPR plasmids and retransforming them into the starting strain with a clean background (Supplementary Fig. 5). Similarly, mutants e137 and e13711 were successively obtained from Round 2 and Round 3, respectively (Supplementary Fig. 4). In total, 1.5 × 10⁸ yeast cells (three rounds of FACS sorting were performed, each using 1 mL of cell culture at an OD₆₀₀ of 5, and each OD₆₀₀ was calculated as 10⁷ cells) were screened through FACS, and 4032 cells (3 rounds, twenty 96-well plates per round, with 70% screening efficiency calculated by the control plate) were selected on 96-well plates with Zeocin, and 63 mutants were validated through spot growth assays. Compared with the wild-type VPR that allows yeast tolerance up to 100 mg/L Zeocin, the final enhanced VPR mutant (e13711) could enhance yeast tolerance up to 600 mg/L Zeocin (Fig. 3f).

Finally, these mutated VPR domains were tested through real-time qPCR analysis (Fig. 3g). The result indicated that three mutants improved the expression of mCherry gene by 7.8-fold, 11.6-fold, and 15.5-fold, respectively. The final mutant, e13711, displayed an activation intensity almost three times that of the initial VPR, which—to the best of our knowledge—represents a record for the best CRISPR activation domains in S. cerevisiae. Sequences of VPR variants identified in this study can be found in Supplementary Data 3.

Fine-tuning of the MVA pathway

The MVA pathway is essential for eukaryotes for production of cholesterol and many terpenoids that are important for membrane integrity^33,34. It has now been intensively used for production of a wide range of terpenoids applied in fuels, perfumes, cosmetics, as well as nutraceuticals and pharmaceuticals^35,36. The MVA pathway in S. cerevisiae consists of eight genes, including ERG10, ERG13, HMG1, ERG12, ERG8, ERG19, IDI1, and ERG20, yet the ideal expression ratio of different genes has not been identified. Thus, we first applied MR to modulate the MVA pathway, using squalene production as a marker for pathway flux.

Briefly, we have introduced the mixed gRNA-tRNA array for rapid library construction, the dSpCas9-NG with expanded PAM recognition to target AT-rich promoter regions, and the mutated VPR activation domain (e13711) with enhanced potency for combinatorial regulations. For the eight genes in the MVA pathway, we designed six gRNAs for each gene. The gRNAs of each gene were spaced by approximately 50 bp within the region between 100 bp to 500 bp upstream of the transcription start site (TSS) to avoid potential hindrance of RNA polymerase (Fig. 4a). These eight groups of gRNAs were mixed and combinatorially assembled into the p2μ-gRNA vector to create the expression library with a potential library size of 6⁸. Then, fifty colonies were randomly selected from the plate to examine the capability of MR in fine-tuning a given pathway without clear phenotype selection aids. Compared with the control strain, half of the colonies exhibited elevated squalene production ranging from 4- to 37-fold (Fig. 4b). The other half that produced less may be due to the aberrant up-regulation of gene expression ratios in the pathway, causing enzyme waste generation, intermediate accumulation, and cell stress. This result demonstrated the significant variety of our combinatorial library based on MR, as well as the importance of the method for fine-tuning unknown pathways.

The gRNAs from high producers (colony numbers 7, 14, 24, 26, and 28) were identified by sequencing and restored into the experimental strain, resulting in strains 7 R, 14 R, 24 R, 26 R, and 28 R, which maintained comparable high levels of squalene accumulation (Fig. 4c). Interestingly, the HMG1 gRNA in these five strains was consistently gRNA 4, while the gRNAs for the other genes exhibited diversity (Fig. 4d). It has been reported that the HMG-CoA reductase, coded by HMG1, is one of the fluxes controlling enzymes in the MVA pathway^37,38,39,40. Therefore, examining the strength of all HMG1 gRNAs may help to explain the convergence of gRNA 4 in these high-producing strains. Indeed, as shown in Fig. 4e, gRNA 4 was the most effective among all designed HMG1 gRNAs, indicating the fundamental role of key gene overexpression in increasing the pathway flux, as well as the importance of the expression ratios of other genes on the final production (ranging from 12-fold to 37-fold).

Transcript changes of the five best producers, 7R, 14R, 24R, 26R, and 28R, were examined using real-time qPCR to determine whether differences in gRNA profiles altered the relative expression levels of the eight genes (Fig. 4f). The results demonstrated that despite using the same HMG1 gRNA for the five strains, the expression of HMG1 varied between strains, indicating an influence on HMG1 expressions from other genes in the pathway. Similar results have been reported that overexpression of ERG2 and ERG3 increased the transcription of ERG1⁴¹. As a rate-limiting gene in the MVA pathway, HMG1 plays a crucial role in maintaining membrane structure, electron transport, protein formylation, and the biosynthesis of glycoproteins. Its expression stability is regulated by multiple transcription factors^42,43, for instance, the CGGNNNTA motif located 400 bp upstream of the TSS serves as a DNA recognition site for the zinc finger transcription factor Hap1. Nevertheless, the potential regulation of the pathway genes would not affect the capability of MR to fine-tune the overall pathway flux, since the restored MR plasmids could generate comparable production levels as the original selected strains (Fig. 4b and c).

Moreover, although 24R and 26R had comparable HMG1 expressions, they exhibited significant differences in squalene production, with 24R showing a 16-fold increase and 26R showing a 37-fold increase, possibly because the eight genes in 26R exhibited high expression levels, including the other generally believed key gene of the pathway, ERG20. Notably, the strain 28R, with both HMG1 and ERG20 moderately expressed, exhibited comparable squalene production to 26R, and both exhibited improved cell growth compared with the control strain (Supplementary Fig. 6), highlighting the complexity of the pathway, as well as the importance of coordinated gene expression rather than the arbitrary overexpression of the key genes.

We have then measured the concentrations of the MVA pathway metabolites in 26R and 28R to determine whether there is a potential inherent limit to the squalene production. As shown in Table 1, although the eight pathway genes were differentially regulated (Fig. 4f), the concentrations of pathway metabolites were comparable, except for MVA and MVA-PP (Table 1). Although the expression level of ERG19 varied significantly among the three strains, all of which exhibited excessive accumulation of MVA-PP, indicating that enzymatic activity may severely limit this step. Therefore, enzyme engineering of Erg19 is expected to be an effective strategy for further optimization of the MVA pathway, thereby enhancing the performance of microbial cell factories for terpenoid production.

Two-dimensional MR for fine-tuning the heme production pathway

Heme availability within the cell is crucial for the proper folding and function of enzymes using heme cofactors^44,45. Due to their biological significance, heme and heme-containing proteins are key subjects in molecular cell biology, advancing fundamental research alongside applications in medicine and technology^{46,47,48,49,50}. To demonstrate that the production improvement is not subject to overall up-regulation but rather a balanced pathway expression, we up-graded MR that could endow both up- and down-regulation. Specifically, for each of the eight genes in the heme biosynthesis pathway, six gRNAs were designed with three targeting the upstream region of the core promoter for upregulation, and the other three targeting the downstream region of the core promoter for down-regulation (Fig. 5a). Furthermore, we integrated MR with 96-well plate measurements and tested 500 colonies randomly picked from plates. With the two-dimensional combinatorial regulation, the heme production exhibited a diversity range of 152-fold (FI/OD: 5000–760,000 AU), as shown in Fig. 5b. Compared with the control strain, seven strains were identified with > 10-fold increases in heme production, with a maximum increase of 17.4-fold.

To identify the potential regulatory mechanisms of the heme biosynthesis pathway, the relative expression levels of heme pathway genes were analyzed by real-time qPCR in all seven strains with >10-fold production and another three strains with <0.5-fold production. As shown in Fig. 5c, HEM2 exhibited completely opposite expression patterns between the high-producing and low-producing strains, with HEM2 upregulation favoring the heme production. Specifically, according to the mRNA profile of strain 4C10, once HEM2 was suppressed overexpressing other genes failed to improve the heme production. This result indicated HEM2 being the potential rate-limiting gene in the pathway. Interestingly, in the best heme producing strain 4B7, all genes were moderately regulated, with each gene having strongly regulated cases in other selected strains. These results again demonstrate that balanced pathway expression outcompetes simple overexpression of all the pathway genes.

Additionally, to compare the capability of dSpCas9-NG and dSpCas9-NG on the combinatorial regulation, we built another strain library based on MR equipped with dSpCas9 and screened 500 strains as well. As expected, although the dSpCas9 system also contributed to a significant diversity of heme production, with a diversity range of 87.5-fold (FI/OD: 4000–350,000 AU) and three strains exhibiting >10-fold increases in heme production (Supplementary Fig. 7), the dSpCas9-NG system still demonstrated its strength (Fig. 5b).

The ideal expression of each gene endows the optimal pathway flux and reduces cell stress. Yet, this knowledge is still missing for most pathways and has to be achieved through iterative engineering. Various strategies have been used to achieve multiplexed regulations, attempting to bypass the conventional DBTL cycle. For example, CRISPR-AID¹⁴ combines transcriptional activation, transcriptional interference, and gene deletion targeting three loci within the whole genome, and is a powerful tool identifying potential targets towards the engineering purposes. ScrABBLE¹⁵ employed three sets of crRNA targeting three open reading frames, enabling combinatorial suppression of three genes in competing pathways. BETTER⁴, GEMbLeR⁵, and COMPASS⁶ have also been proven to be effective combinatorial regulation strategies to fine-tune a given pathway, yet they often require major pre-integration of ribosome binding sites⁴, LoxPsym sites together with upstream promoter elements and terminators⁵, and transcription factor binding sites to the targeted genes⁶. Thus, an easy-to-use method with the ability to fine-tune long pathways is still needed.

In this study, we developed MR, the plug-and-tune technology that allows for fine-tuning pathways with up to eight genes in S. cerevisiae. This method has three key features: (1) an efficient gRNA matrix assembly system for library construction using a mixed gRNA-tRNA array; (2) an expanded PAM range, which increases targeting scope in AT-rich regions; and (3) the most potent activation domain reported in S. cerevisiae that provides greater regulatory amplitude. We have demonstrated the utility and efficiency of MR using both the MVA pathway and the heme biosynthesis pathway, neither of which requires high-throughput screening approaches. This is because target molecules lack biosensors or related phenotypes suitable for high-throughput screening in most scenarios, and the development of efficient combinatorial regulation methods to optimize target pathways without high-throughput screening is critical. Nevertheless, integration of MR with high-throughput screening approaches and machine learning techniques will endow optimal gene expression equations of the targeted pathways, facilitating the rapid and quantitative scope of metabolic engineering.

Future optimization of MR may include: expansion of the number of target genes beyond eight by improving plasmid assembly efficiency, improvement of activators for regulatory research. Moreover, the off-target possibilities of MR could also be determined, especially in more complex hosts. Cas9 variants have been reported with enhanced specificity⁵¹ and targeting ability⁵². By leveraging these advanced Cas9 proteins, MR is expected to be useful in a broad number of host species.

Altogether, MR provides a more convenient and efficient option for fine-tuning pathways with multiple genes. This toolkit not only benefits fundamental research on genotype-phenotype relationships but also accelerates metabolic engineering in yeast and sheds light on the key steps in the establishment of efficient cell factories.

Strains and culture conditions

The GTR-CRISPR system was used for all genome modifications in this work¹³. The S. cerevisiae strain CEN.PK113-5D (MATa, MAL2-8c, SUC2, ura3-52)⁵³ was used for tRNA testing and also served as the base strain for experiments in this study. To investigate the gene activation efficiency of different dCas9 variants on NGN PAMs without the gRNA bias, we constructed a strain platform containing 16 strains, each with X-2::gRNA-P_CYC1-140bp-mCherry-T_ADH1 and different NGN PAM sequences. CEN.PK113-5D X-3::P_CYC1-140bp-ZeoR-T_ADH1 was used for endogenous activator screening. CEN.PK113-5D Gal80∆::KanMX X-3::P_CYC1-140bp-ZeoR*-T_ADH1 XII-2::P_CYC1-140bp-mCherry-T_ADH1 was used for VPR enhancement. For the combinatorial regulation in the MVA pathway and the heme biosynthesis pathway, CEN.PK113-5D leu2∆ his3∆ X-2::P_TEF1-dSpCas9-NG-e13711-T_ADH1 was created. For a full list of strains used in this study, see Supplementary Data 4.

For yeast transformation, a single yeast colony from a fresh plate was inoculated into YPD medium and grown overnight. The culture was then diluted into 50 mL of YPD medium to an OD₆₀₀ of 0.3. After ~5 h of growth at 30 °C, 220 rpm, when the culture reached an OD₆₀₀ of 1.6, the cells were harvested by centrifugation (3 min, 3000 rpm, 4 °C). The cell pellet was washed once with 20 mL of ice-cold 1 M sorbitol. Then the washed yeast cells were resuspended in a solution containing 16 mL of 1 M sorbitol, 2 mL of 10× TE buffer (100 mM Tris-HCl, 10 mM EDTA, pH 7.5), and 2 mL of 1 M lithium acetate. The cell suspension was incubated in a shaker for 30 min at 30 °C. Subsequently, 200 µL of 1 M dithiothreitol (DTT) was added, and incubation continued for an additional 15 min. The cells were then washed twice with 20 mL of ice-cold 1 M sorbitol. After the final wash, all supernatant was carefully removed. The resulting competent cells were resuspended in 400 µL of 1 M sorbitol. 100 µL competent cells were then dispensed into a sterile microcentrifuge tube for transformation. Less than 10 µL of the target DNAs was added to the competent cells. The mixture was subjected to electroporation at 1.5 kV using a 2 mm electroporation cuvette. After pulsing, 1 mL of 1 M sorbitol was rapidly added to the cuvette, and the cells were transferred to a 50 mL culture tube. The cuvette was washed once with an additional 1 mL of sorbitol, which was also transferred to the culture tube. Finally, 2 mL of YPD medium was added to the tube. The transformed cells were recovered by incubating the culture tube at 30 °C by shaking (250 rpm) for 5 h. Transformants were then cultured on SC agar media without the nutrient supplemented by vectors. SC-URA media supplemented with Zeocin were employed for activator screening and VPR enhancement. SC-URA-LEU-HIS media was employed to screen strains harboring plasmids for MVA pathway regulation by MR. SC-URA-LEU media were adopted for screening strains with plasmids for heme biosynthetic pathway regulation.

Plasmid construction

A full list of plasmids used in this study can be found in Supplementary Data 4. Recombinant plasmids were constructed using Gibson Assembly or Golden Gate Assembly. The GTR-CRISPR system was used for gRNA-tRNA fragment assembly in this work¹³. pdCas9-VPR, pdxCas9-VPR, and pdSpCas9-NG-VPR for dCas9 variant characterization were constructed based on the pCas vector in the GTR-CRISPR system. The iCas9 gene in the pCas was replaced by dCas9-VPR, dxCas9-VPR, and dSpCas9-NG-VPR, respectively. pCen-L-Kan vector was built by replacing the 2-micron sequence in pCas with CEN/ARS, iCas9 with dCas9, and lacZα with KanMX. pKan-ori-VPR vector, used as a template for the initial error-prone PCR, was created by assembling the ori, KanR, and VPR sequences. Subsequently, pKan-ori-e13 and pKan-ori-e137 were constructed by replacing VPR gene with e13 and e13 with e137, respectively. The p2μ-gRNA vector was also built on pCas by removing the iCas9 sequence. The psgtRNA-Arg, psgtRNA-Thr, psgtRNA-Ser, and psgtRNA-Asp used for gRNA amplification were formed by replacing the tRNA^Gly gene sequence in psgtRNA in the GTR-CRISPR system with the relevant tRNAs. The pHis3-tRNA-Lys, pKlura3-tRNA-Leu, and pLeu2-tRNA-Ser vectors used for gRNA and marker amplification were generated by replacing Ura3 and tRNA^Gly sequences in pScURA3 in the GTR-CRISPR system with the relevant markers and tRNAs.

Library construction

In this study, all gRNA-tRNA fragments were obtained through PCR, using vectors containing different tRNAs as templates. All primers used can be found in Supplementary Data 5, and gRNA sequences in Supplementary Data 6. After gel DNA extraction, gRNAs of the same group were mixed in one tube. In 30 μL total Golden Gate mixes, 10 μL NEBridege^® Ligase Master Mix buffer (M1100, New England Biolabs) and 1 μL BsaI-HFv2 (E1601, New England Biolabs) were added to the reaction mix. The quantity of vectors and fragments added was calculated using tools available at GoldenGate.neb.com. The mixtures were made according to the recommended procedure by NEBridege^® Ligase Master Mix for 60 cycles. To increase the number of successful assemblies, six identical Golden Gate reactions were performed simultaneously. Subsequently, every three reactions were purified using one Cycle-pure PCR Purification Kit (Omega, USA), and finally, both columns were eluted with 35 μL of water. 10 μL of the elution was transformed into yeast strains. After a 5 h recovery, each  1 mL of the culture was plated onto the corresponding SC selective medium plates for screening.

Extraction and quantification of squalene production

Colonies were picked from the plates and inoculated into seed cultures. After 24 h of cultivation, the cultures were inoculated to 20 mL of the selective SC medium in 100 mL shake flasks, with an initial OD₆₀₀ of 0.1. After 6 h, terbinafine was added to a final 30 mg/L concentration. Yeast cells were cultured by 72 h for squalene production determinations using a HPLC-based method at 210 nm. For squalene extraction⁵⁴, 0.2 mL of the culture was sampled and centrifuged at 3000 rpm for 5 min. Then, the cell pellet was mixed with 0.2 g acid-washed beads (0.5 mm in diameter), resuspended in 1 mL acetone, and vortexed for 20 min until residues were colorless. The acetone extracts were then centrifuged at 13,000 rpm for 10 min, filtered with a 0.45 μm nylon filter membrane before analyzing. An HPLC system (LC-20 A/SPD-20AV; SHIMADZU; LabSolutions v5.93) equipped with a C18 column (4.6 mm × 250 mm × 5 μm; SUPELCOSILTMLC-18) was used to detect the 210 nm signal at a flow rate of 1 ml/min at 40 °C. Samples were eluted with acetonitrile–methanol–isopropanol (5:3:2, v/v/v).

Quantification of metabolites in the MVA pathway

The 26R and 28R strains carrying gRNA served as experimental samples, with the gRNA-free strain used as the negative control (NC). Cell cultures were inoculated from fresh colonies (n = 3) and incubated in tubes overnight and then transferred to 100 mL shake flasks with an initial OD₆₀₀ of 0.1. At mid-log phase (36 h), 1.5mL samples were collected and centrifuged at 14,000 × g for 1 min, after which the cell pellet was vortexed with 250 µL of ice-cold methanol (the processed samples could be stored at −20 °C for up to one month). Prior to analysis, the samples were added with 250 µL of deionized water, vortexed, and then centrifuged at 4 °C and 14,000 × g for 5 min. The supernatant was then transferred to a 3000 Da MW/CO centrifuge filter tube (Amicon Ultra from MilliporeSigma) and centrifuged at −2 °C and 13,000 × g for 90 min. 400 µL of the filtrate was mixed with 500 µL of deionized water, flash-frozen in liquid nitrogen, and lyophilized for 24 h (protected from light). The pellet was then re-dissolved using 200 µL reconstitution solution (acetonitrile–methanol–water, 6:1:3, v/v/v), and measured by LC-MS.

Chromatographic separation was performed on a Waters ACQUITY UHPLC system equipped with a Waters BEH Amide column (1.7 μm, 3.0 × 100 mm) maintained at 40 °C. The injection volume was 5 μL with a constant flow rate of 400 μL/min. For positive ion mode analysis, the mobile phase consisted of (A) 10 mM ammonium formate + 0.1% ammonia in water and (B) 90% acetonitrile + 10 mM ammonium formate + 0.1% ammonia. An isocratic elution profile was employed: 30% A/70% B held from 0.00 to 5.00 min. Mass spectrometric detection was conducted using an AB Sciex 4500 triple quadrupole mass spectrometer operating in MRM mode with positive electrospray ionization. ESI source parameters were optimized as follows: ion spray voltage 4500 V, source temperature 500 °C, curtain gas 25 psi, collision gas 10 psi. Both primary and secondary mass spectrum were acquired. All data were analyzed using MultiQuant 3.0.3 software. Sample concentrations were calculated based on calibration curves constructed from the responses of reference standards versus their corresponding concentrations using the internal standard method⁵⁵.

Reagents used in this experiment: acetyl-CoA (S23156, OriLeaf), acetoacetyl-CoA (A341482, Aladdin), 3-hydroxy-3-methyl-glutaryl-CoA (H6132, Sigma), 3,5-dihydroxy-3-methylpentanoic acid (R921336, Macklin), (4S)-4-hydroxy-4-methyloxan-2-one (D869733, Macklin), mevalonic acid-5-phosphate (79849, Sigma), mevalonic acid-5-pyrophosphate (M990271, Macklin), isopentenyl diphosphate (I0503, Sigma), dimethylallyl pyrophosphate (D4287, Sigma), geranyl diphosphate (G6772, Sigma), farnesyl pyrophosphate (F6892, Sigma), DL-methionine sulfone (D863216, Macklin), deuterated d4-succinic acid (S874692, Macklin), deuterated d4-citric acid (C994619, Aladdin).

Determination of heme concentration

For the liquid medium used in the heme experiments, 24.5 mg of ferric citrate (to achieve a final concentration of 100 μM) was first dissolved in deionized water in a Schott-Duran bottle and heated using a magnetic stirrer at 60 °C. The bottle containing the ferric citrate solution was wrapped in aluminum foil to prevent light exposure. After complete dissolution, the components of SC-URA-LEU-HIS drop-out medium and 7.507 g of glycine (to reach a final concentration of 100 mM) were added. The pH was then adjusted to 6.0, deionized water was added to a final volume of 1 L, and the medium was filter-sterilized.

Single colonies were randomly picked and inoculated into 96-well plates, with each well containing 800 μL SC-URA-LEU-HIS liquid medium supplemented with 100 mM glycine and 100 μM ferric citrate. After 12 h of cultivation, strains were subcultured into a fresh 96-well plate with started OD₆₀₀ at 0.3 and grown for an additional 24 h. The cell growth was measured. Then, 200 μL of each culture was transferred to PCR plates and centrifuged at 3000 rpm for 3 min. The pellets were resuspended in 120 μL of 20 mM oxalic acid and stored at 4 °C for 8 h. Subsequently, 120 μL of 2 M oxalic acid was added, and 120 μL of the suspension was transferred to a new PCR plate. One plate was heated at 98 °C for 30 min, while the other was incubated at room temperature for 30 min. Then, both plates were centrifuged at 3000 rpm for 3 min. 100 μL of supernatant from each plate was transferred to a black microplate for fluorescence detection using a Molecular Devices iD3 spectrophotometer (Ex: 400 nm, Em: 600 nm). The heme concentrations were calculated by subtracting the fluorescence value of the room temperature group from that of the heat-treated group.

RNA level determinations

An exponential phase culture was used to test the RNA level using real-time qPCR with PowerUp^TM SYBR^TM Green Master Mix (A25742, ThermoFisher Scientific), and the total RNAs were extracted using TRIzol^TM Reagent (15596026, ThermoFisher Scientific). 500 ng of the extracted RNAs were reverse transcribed into cDNA using the All-in-one^TM First-Strand cDNA Synthesis Kit (QP056, GeneCopoeia). A QuantStudio 3 (ThermoFisher Scientific) instrument equipped with QuantStudio Design&Analysis Desktop Software v1.4.1 was used to collect the data and Prism 10.1.2 software was used to analyze the data. Primers for real-time qPCR in this study can be found in Supplementary Data 7.

Assay of fluorescence intensity

Yeast colonies were inoculated to 96 deep-well plates with 800 μL of SC-URA-LEU and incubated at 30 °C, shaking at 1000 rpm for three days. Then, 50 μL of the saturated culture was diluted into 750 μL of fresh medium and cultured for one day. Fluorescence intensity was measured by a Molecular Devices iD3 plate reader spectrophotometer equipped with SoftMax Pro 7.1 software. 200 μL of the culture was transferred to a black 96-well plate for fluorescence intensity measurement, with an excitation wavelength of 578 nm and an emission wavelength of 618 nm for mCherry. Simultaneously, 10 μL of the culture was transferred to a clear-bottom microplate, and each well was diluted with 190 μL of water to measure OD₆₀₀.

Library construction and screening of VPR mutant

The GeneMorph II Random Mutagenesis Kit (200550, Agilent, USA) was utilized to amplify the sequence for the VPR via error-prone PCR. Thirty PCR cycles and 500 ng template plasmid were used to obtain 3-8 mutations per gene. Purification was performed using the Cycle-pure PCR Purification Kit (Omega, USA). This fragment contained a 150 bp C-terminal sequence of dCas9 encoding sequence, an SV40 NLS, the VPR gene library, and a 198 bp ADH1 terminator. The backbone pCen-L-Kan vector was digested with the enzyme BsaI at 37 °C, followed by purification. The VPR mutagenesis library and digested backbone dCas9 fragments were directly electroporated into yeast cells, and the pCen-L-VPR* plasmid was constructed with homology-directed repair (HDR). After a 5 h recovery, each 1 mL of yeast cells was plated onto a SC-URA plate and incubated at 30 °C for three days. The cells were then scraped off using sterile water and cultured overnight. The culture was transferred to a fresh medium and diluted to an OD₆₀₀ of 0.1, then cultured for three days in preparation for FACS. Single cells were sorted with a BD Influx (equipped with BD FACS Software sorter software) using a 561 nm laser for excitation and a 593/40 nm (RFP) filter for detection. The top 1% most fluorescent cells were collected into 96-well cell culture plates with SC-URA medium supplemented with Zeocin (R25001, Invitrogen, USA) and cultured for three days at 30 °C (Supplementary Fig. 8). The data were then analyzed using FlowJo v10.8.1 software.

Spot growth assay

Colonies were picked from fresh SC-URA plates and inoculated into a liquid medium for overnight culture. The seed culture was then transferred to a fresh medium with an initial OD₆₀₀ of 0.1. When the culture reached an OD₆₀₀ of 0.8–1.2, the cells were diluted to OD₆₀₀ values of 0.01, 0.001, and 0.0001. Then, 5 μL of each dilution was dropped onto SC-URA plates containing Zeocin and incubated at 30 °C for 3-5 days to observe growth.

Use of large language model

During the preparation of this work, the GPT-4 was used to ensure grammatical correction of manuscript.

Statistics and reproducibility

The experiments were repeated at least twice with similar results to ensure reproducibility. Colonies were selected randomly from agar plates when being prepared for experimental pre-cultures; a single colony represents one biological replicate. Sample sizes were chosen based on common practices in the field and previous studies with similar experimental designs. No statistical method was used to predetermine sample size. For tRNA characterization, the experiment was run with six biological replicates. For FI of mCherry, real-time qPCR, and growth assay, each experiment was run with three or four biological replicates. For the verification of the re-transformed strains, the experiment was run with two biological replicates. No data were excluded from the analyses. Error bars indicate mean ± SD of all replicates Statistical analysis was conducted using two-tailed unpaired t-tests. For all statistical analyses, p  <  0.05 was considered significant.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

1. Nielsen, J. & Keasling, J. D. Engineering cellular metabolism. Cell 164, 1185–1197 (2016).

   CAS  PubMed  Google Scholar 

2. Qin, C. et al. Precise programming of multigene expression stoichiometry in mammalian cells by a modular and programmable transcriptional system. Nat. Commun. 14, 1500 (2023).

   ADS  CAS  PubMed  PubMed Central  Google Scholar 

3. Liu, J. et al. CRISPR-assisted rational flux-tuning and arrayed CRISPRi screening of an L-proline exporter for L-proline hyperproduction. Nat. Commun. 13, 891 (2022).

   ADS  CAS  PubMed  PubMed Central  Google Scholar 

4. Wang, Y. et al. In-situ generation of large numbers of genetic combinations for metabolic reprogramming via CRISPR-guided base editing. Nat. Commun. 12, 678 (2021).

   ADS  CAS  PubMed  PubMed Central  Google Scholar 

5. Cautereels, C. et al. Combinatorial optimization of gene expression through recombinase-mediated promoter and terminator shuffling in yeast. Nat. Commun. 15, 1112 (2024).

   ADS  CAS  PubMed  PubMed Central  Google Scholar 

6. Naseri, G., Behrend, J., Rieper, L. & Mueller-Roeber, B. COMPASS for rapid combinatorial optimization of biochemical pathways based on artificial transcription factors. Nat. Commun. 10, 2615 (2019).

   ADS  PubMed  PubMed Central  Google Scholar 

7. Xu, L., Liu, P., Dai, Z., Fan, F. & Zhang, X. Fine-tuning the expression of pathway gene in yeast using a regulatory library formed by fusing a synthetic minimal promoter with different Kozak variants. Micro Cell Fact. 20, 148 (2021).

   CAS  Google Scholar 

8. Mukherjee, M., Blair, R. H. & Wang, Z. Q. Machine-learning guided elucidation of contribution of individual steps in the mevalonate pathway and construction of a yeast platform strain for terpenoid production. Metab. Eng. 74, 139–149 (2022).

   CAS  PubMed  Google Scholar 

9. Gao, S., Zhou, H., Zhou, J. & Chen, J. Promoter-library-based pathway optimization for efficient (2S)-naringenin production from p-coumaric acid in Saccharomyces cerevisiae. J. Agric Food Chem. 68, 6884–6891 (2020).

   CAS  PubMed  Google Scholar 

10. Gilbert, L. A. et al. Genome-scale CRISPR-mediated control of gene repression and activation. Cell 159, 647–661 (2014).

    CAS  PubMed  PubMed Central  Google Scholar 

11. Zalatan, J. G. et al. Engineering complex synthetic transcriptional programs with CRISPR RNA scaffolds. Cell 160, 339–350 (2015).

    CAS  PubMed  Google Scholar 

12. Shaw, W. M. et al. Inducible expression of large gRNA arrays for multiplexed CRISPRai applications. Nat. Commun. 13, 4984 (2022).

    ADS  CAS  PubMed  PubMed Central  Google Scholar 

13. Zhang, Y. et al. A gRNA-tRNA array for CRISPR-Cas9 based rapid multiplexed genome editing in Saccharomyces cerevisiae. Nat. Commun. 10, 1053 (2019).

    ADS  PubMed  PubMed Central  Google Scholar 

14. Lian, J., HamediRad, M., Hu, S. & Zhao, H. Combinatorial metabolic engineering using an orthogonal tri-functional CRISPR system. Nat. Commun. 8, 1688 (2017).

    ADS  PubMed  PubMed Central  Google Scholar 

15. Wu, Y. et al. CRISPR-dCas12a-mediated genetic circuit cascades for multiplexed pathway optimization. Nat. Chem. Biol. 19, 367–377 (2023).

    CAS  PubMed  Google Scholar 

16. Sugihara, F., Kasahara, K. & Kokubo, T. Highly redundant function of multiple AT-rich sequences as core promoter elements in the TATA-less RPS5 promoter of Saccharomyces cerevisiae. Nucleic Acids Res. 39, 59–75 (2011).

    CAS  PubMed  Google Scholar 

17. Cong, L. et al. Multiplex genome engineering using CRISPR/Cas systems. Science 339, 819–823 (2013).

    ADS  CAS  PubMed  PubMed Central  Google Scholar 

18. Zhai, H. T., Cui, L., Xiong, Z., Qi, Q. S. & Hou, J. CRISPR-mediated protein-tagging signal amplification systems for efficient transcriptional activation and repression in. Nucleic Acids Res. 50, 5988–6000 (2022).

    CAS  PubMed  PubMed Central  Google Scholar 

19. Dong, C. et al. A single Cas9-VPR nuclease for simultaneous gene activation, repression, and editing in Saccharomyces cerevisiae. ACS Synth. Biol. 9, 2252–2257 (2020).

    CAS  PubMed  Google Scholar 

20. Chavez, A. et al. Highly efficient Cas9-mediated transcriptional programming. Nat. Methods 12, 326–328 (2015).

    CAS  PubMed  PubMed Central  Google Scholar 

21. Deaner, M., Mejia, J. & Alper, H. S. Enabling graded and large-scale multiplex of desired genes using a dual-mode dCas9 activator in Saccharomyces cerevisiae. ACS Synth. Biol. 6, 1931–1943 (2017).

    PubMed  Google Scholar 

22. Nishimasu, H. et al. Engineered CRISPR-Cas9 nuclease with expanded targeting space. Science 361, 1259–1262 (2018).

    ADS  CAS  PubMed  PubMed Central  Google Scholar 

23. Hu, J. H. et al. Evolved Cas9 variants with broad PAM compatibility and high DNA specificity. Nature 556, 57–63 (2018).

    ADS  CAS  PubMed  PubMed Central  Google Scholar 

24. Gong, G. et al. GTR 2.0: gRNA-tRNA array and Cas9-NG based genome disruption and single-nucleotide conversion in Saccharomyces cerevisiae. ACS Synth. Biol. 10, 1328–1337 (2021).

    CAS  PubMed  Google Scholar 

25. Jensen, N. B. et al. EasyClone: method for iterative chromosomal integration of multiple genes in Saccharomyces cerevisiae. FEMS Yeast Res. 14, 238–248 (2014).

    CAS  PubMed  Google Scholar 

26. Tanenbaum, M. E., Gilbert, L. A., Qi, L. S., Weissman, J. S. & Vale, R. D. A protein-tagging system for signal amplification in gene expression and fluorescence imaging. Cell 159, 635–646 (2014).

    CAS  PubMed  PubMed Central  Google Scholar 

27. Qu, J. L. et al. Synthetic multienzyme complexes, catalytic nanomachineries for cascade biosynthesis. Acs Nano 13, 9895–9906 (2019).

    CAS  PubMed  Google Scholar 

28. Barajas, C., Huang, H. H., Gibson, J., Sandoval, L. & Del Vecchio, D. Feedforward growth rate control mitigates gene activation burden. Nat. Commun. 13, 7054 (2022).

    ADS  CAS  PubMed  PubMed Central  Google Scholar 

29. Sanders, S. L. & Weil, P. A. Identification of two novel TAF subunits of the yeast Saccharomyces cerevisiae TFIID complex. J. Biol. Chem. 275, 13895–13900 (2000).

    CAS  PubMed  Google Scholar 

30. Nishikawa, J. L. et al. Inhibiting fungal multidrug resistance by disrupting an activator-Mediator interaction. Nature 530, 485–489 (2016).

    ADS  CAS  PubMed  PubMed Central  Google Scholar 

31. Sudarsanam, P., Iyer, V. R., Brown, P. O. & Winston, F. Whole-genome expression analysis of snf/swi mutants of Saccharomyces cerevisiae. Proc. Natl. Acad. Sci. USA 97, 3364–3369 (2000).

    ADS  CAS  PubMed  PubMed Central  Google Scholar 

32. Yim, S., Yu, H., Jang, D. & Lee, D. Annotating activation/inhibition relationships to protein-protein interactions using gene ontology relations. BMC Syst. Biol. 12, 9 (2018).

    PubMed  PubMed Central  Google Scholar 

33. Xia, Y. et al. The mevalonate pathway is a druggable target for vaccine adjuvant discovery. Cell 175, 1059–1073.e1021 (2018).

    CAS  PubMed  Google Scholar 

34. Caspeta, L. et al. Altered sterol composition renders yeast thermotolerant. Science 346, 75–78 (2014).

    ADS  CAS  PubMed  Google Scholar 

35. Zhang Y., Nielsen J., Liu Z. Engineering yeast metabolism for production of terpenoids for use as perfume ingredients, pharmaceuticals and biofuels. FEMS Yeast Res. 17, (2017).

36. Partow, S., Siewers, V., Daviet, L., Schalk, M. & Nielsen, J. Reconstruction and evaluation of the synthetic bacterial MEP pathway in Saccharomyces cerevisiae. PLoS One 7, e52498 (2012).

    ADS  CAS  PubMed  PubMed Central  Google Scholar 

37. Han, J. Y., Seo, S. H., Song, J. M., Lee, H. & Choi, E. S. High-level recombinant production of squalene using selected Saccharomyces cerevisiae strains. J. Ind. Microbiol. Biotechnol. 45, 239–251 (2018).

    CAS  PubMed  Google Scholar 

38. Zhao, J. et al. Dynamic control of ERG20 expression combined with minimized endogenous downstream metabolism contributes to the improvement of geraniol production in Saccharomyces cerevisiae. Micro Cell Fact. 16, 17 (2017).

    Google Scholar 

39. Jiang, G. Z. et al. Manipulation of GES and ERG20 for geraniol overproduction in Saccharomyces cerevisiae. Metab. Eng. 41, 57–66 (2017).

    CAS  PubMed  Google Scholar 

40. Xie, W., Lv, X., Ye, L., Zhou, P. & Yu, H. Construction of lycopene-overproducing Saccharomyces cerevisiae by combining directed evolution and metabolic engineering. Metab. Eng. 30, 69–78 (2015).

    CAS  PubMed  Google Scholar 

41. Wei, W. et al. Reengineering of 7-dehydrocholesterol biosynthesis in Saccharomyces cerevisiae using combined pathway and organelle strategies. Front. Microbiol. 13, 978074 (2022).

    PubMed  PubMed Central  Google Scholar 

42. MacIsaac, K. D. et al. An improved map of conserved regulatory sites for Saccharomyces cerevisiae. BMC Bioinforma. 7, 113 (2006).

    Google Scholar 

43. Hu, Z., Killion, P. J. & Iyer, V. R. Genetic reconstruction of a functional transcriptional regulatory network. Nat. Genet. 39, 683–687 (2007).

    CAS  PubMed  Google Scholar 

44. Bernhardt, R. Cytochromes P450 as versatile biocatalysts. J. Biotechnol. 124, 128–145 (2006).

    CAS  PubMed  Google Scholar 

45. He, L. et al. Antioxidants maintain cellular redox homeostasis by elimination of reactive oxygen species. Cell Physiol. Biochem. 44, 532–553 (2017).

    PubMed  Google Scholar 

46. Moradi, S., Jahanian-Najafabadi, A. & Roudkenar, M. H. Artificial blood substitutes: first steps on the long route to clinical utility. Clin. Med. Insights Blood Disord. 9, 33–41 (2016).

    PubMed  PubMed Central  Google Scholar 

47. Fraser, R. Z., Shitut, M., Agrawal, P., Mendes, O. & Klapholz, S. Safety evaluation of soy leghemoglobin protein preparation derived from pichia pastoris, intended for use as a flavor catalyst in plant-based meat. Int. J. Toxicol. 37, 241–262 (2018).

    CAS  PubMed  PubMed Central  Google Scholar 

48. Ryu, W. H. et al. Heme biomolecule as redox mediator and oxygen shuttle for efficient charging of lithium-oxygen batteries. Nat. Commun. 7, 12925 (2016).

    ADS  CAS  PubMed  PubMed Central  Google Scholar 

49. Mirts, E. N., Petrik, I. D., Hosseinzadeh, P., Nilges, M. J. & Lu, Y. A designed heme-[4Fe-4S] metalloenzyme catalyzes sulfite reduction like the native enzyme. Science 361, 1098–1101 (2018).

    ADS  CAS  PubMed  PubMed Central  Google Scholar 

50. Ishchuk, O. P. et al. Genome-scale modeling drives 70-fold improvement of intracellular heme production in Saccharomyces cerevisiae. Proc. Natl. Acad. Sci. USA 119, e2108245119 (2022).

    CAS  PubMed  PubMed Central  Google Scholar 

51. Casini, A. et al. A highly specific SpCas9 variant is identified by in vivo screening in yeast. Nat. Biotechnol. 36, 265–271 (2018).

    CAS  PubMed  PubMed Central  Google Scholar 

52. Walton, R. T., Christie, K. A., Whittaker, M. N. & Kleinstiver, B. P. Unconstrained genome targeting with near-PAMless engineered CRISPR-Cas9 variants. Science 368, 290–296 (2020).

    ADS  CAS  PubMed  PubMed Central  Google Scholar 

53. Entian K.-D., Kötter P. 25 Yeast genetic strain and plasmid collections. In: Methods in Microbiology (eds Stansfield I, Stark MJR). (Academic Press, 2007).

54. Zhao, Y., Zhang, Y., Nielsen, J. & Liu, Z. Production of β-carotene in Saccharomyces cerevisiae through altering yeast lipid metabolism. Biotechnol. Bioeng. 118, 2043–2052 (2021).

    CAS  PubMed  Google Scholar 

55. Baidoo E. E. K., Wang G., Joshua C. J., Benites V. T., Keasling J. D. Liquid chromatography and mass spectrometry analysis of isoprenoid intermediates in Escherichia coli. In: Microbial Metabolomics: Methods and Protocols (ed Baidoo E. E. K.) (Springer New York, 2019).

Download references

Author notes

1. These authors contributed equally: Xiaolong Teng, Zibai Wang, Yueping Zhang.

Authors and Affiliations

1. College of Life Science and Technology, State Key Laboratory of Green Biomanufacturing, Beijing Advanced Innovation Center for Soft Matter Science and Engineering, Beijing University of Chemical Technology, Beijing, China

   Xiaolong Teng, Zibai Wang, Binhao Wang, Guiping Gong, Jinmiao Hu, Yifan Zhu, Baoyi Peng, Junyang Wang, James Chen, Shuobo Shi, Jens Nielsen & Zihe Liu

2. College of Veterinary Medicine, China Agricultural University, Beijing, China

   Yueping Zhang

3. Department of Biology and Biological Engineering, Chalmers University of Technology, Gothenburg, Sweden

   Jens Nielsen

4. BioInnovation Institute, Copenhagen N, Denmark

   Jens Nielsen

Contributions

X.L.T., Z.B.W., Y.P.Z., J.N., and Z.H.L. designed the research; X.L.T., Z.B.W., B.H.W., G.P.G., J.M.H., Y.F.Z., and B.Y.P. carried out the experiments; X.L.T., Z.B.W., Y.P.Z., J.Y.W., S.B.S., and Z.H.L. analysed the data; X.L.T., J.C., J.N., and Z.H.L. wrote the paper; J.N. and Z.H.L. supervised the research. X.L.T., Z.B.W., and Y.P.Z. contributed equally.

Corresponding author

Correspondence to Zihe Liu.

Competing interests

The authors declare no competing interests.

Peer review information

Nature Communications thanks Riaan Den Haan and the other anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions