Intein-mediated temperature control for complete biosynthesis of sanguinarine and its halogenated derivatives in yeast

While sanguinarine has gained recognition for antimicrobial and antineoplastic activities, its complex conjugated structure and low abundance in plants impede broad applications. Here, we demonstrate the complete biosynthesis of sanguinarine and halogenated derivatives using highly engineered yeast strains. To overcome sanguinarine cytotoxicity, we establish a splicing intein-mediated temperature-responsive gene expression system (SIMTeGES), a simple strategy that decouples cell growth from product synthesis without sacrificing protein activity. To debottleneck sanguinarine biosynthesis, we identify two reticuline oxidases and facilitated functional expression of flavoproteins and cytochrome P450 enzymes via protein molecular engineering. After comprehensive metabolic engineering, we report the production of sanguinarine at a titer of 448.64 mg L−1. Additionally, our engineered strain enables the biosynthesis of fluorinated sanguinarine, showcasing the biotransformation of halogenated derivatives through more than 15 biocatalytic steps. This work serves as a blueprint for utilizing yeast as a scalable platform for biomanufacturing diverse benzylisoquinoline alkaloids and derivatives.

3) In Figures 3e and 3f, even at non-permissive temperatures, spliced 35 kDa bands are observed along with unspliced 85 kDa bands.And the ratio of spliced to unspliced forms does not appear to match the fluorescence in Figure 3d.Therefore, in Figures 3e and 3f, the quantification of the ratio of spliced to unspliced forms are necessary, and then the correlation between the ratio and the value of fluorescence should be discussed.4) In Figures 6c and 7c, the images are too small to be easily seen, so it is better to include an enlarged image of the cells as insets.5) In figures, the numbers a-d should be written in upper or lower case.(They are capitalised in the text).6) In Fig. 1, an explanation of the abbreviations NOR, RET, SCO, CHE, PRO and SAN is given in the diagram, but also in the legend of Fig. 4. If this is to be adapted, the explanation of the abbreviations should also be given in the legend of Figs. 5 and 6. 7) In Fig. 2D, SED, SEDG, SEG, SEGly and SEGlyG need to be explained.Reviewer #3: Remarks to the Author: Gou et al show a novel Splicing Intein-Mediated Temperature-responsive Gene Expression System, which they call SIMTeGES that enables regulation of expression of heterologous genes.Their system uses temperature as a regulatory signal to separate cell growth from product biosynthesis processes (e.g. for a two-stage process).At 30 °C, optimal for yeast growth, SIMTeGES keeps the GAL4 transcription factor inactive, which allows cell growth.Conversely, at 25 °C, GAL4 is activated by splicing and this promotes the expression heterologous genes under control of the GAL4 responsive promoter.The approach was used to separate cell growth and sanguinarine biosynthesis (a toxic compound for yeast).Additionally the authors show that its generally applicable and facilitates the temperature-controlled expression of additional target proteins (e.g., mCherry, GAL80).The case study of sanguinarine biosynthesis is impressive as it involves heterologous expression of many enzymes and optimization of several steps.The authors report the highest titer achieved for sanguinarine.The topic of two-stage process and dynamic control of microbial systems is very relevant and the manuscript is well written.I have few points that should be considered: 1) The manuscript combines two important developments: the SIMTeGES, and the sanguinarine production.The results are structured in this way, but Figure 1 starts with the sanguinarine pathway.I recommend to present the SIMTeGES first.In Figure 1, the expression system should be explained: which genes are under which promoter, where are they integrated into the genome.This would help to follow the metabolic engineering strategy that is described in the second part of the result section.
2) The authors should describe how the temperature sensitive (TS) intein VMAL212P was derived.Did they screen for the TS phenotype, did they find additional TS alleles?
3) It seems that the GAL 4M9 and GAL4 tsINT perform similar for lycopene and mCherry, and the largest benefit is visible in the sanguinarine case.The authors should better compare the pros and cons of these system also in terms of repression efficiency at 30°C.The Western Blot indicates significant splicing at 30°C for GAL4 tsINT.4) In Figure 6a the authors show a very nice example of metabolic engineering (or bottleneck identification) based on metabolite accumulation.However, the presentation is very basic.The authors should provide quantitative data (e.g.fold-changes) for the metabolites together with abundances of the respective enzymes.This would help to understand the systematic approach for changes in the pathway.5) The authors use complex medium for their bioprocess for sanguinarine biosynthesis.At the time point of sanguinarine synthesis the main carbon source glycerol and galactose are depleted.Therefore, the authors should show which carbon source goes into sanguinarine biosynthesis.It is not clear if tyrosine fed to the cultures in Figure 6 or if tyrosine was derived from the complex medium.

Reviewer #1 (Comments for the Author):
In this study, Gou and colleagues aimed to improve the production of sanguinarine in baker's yeast, building on their previous research.To uncouple cell growth from product biosynthesis, the researchers devised a splicing intein-mediated temperature-responsive gene expression system (SIMTeGES) to regulate gene expression.An intein sequence was inserted into the coding sequence of GAL4 to create a temperature-sensitive transcription factor.The regularity of this engineered GAL4-tsINT was assessed through lycopene biosynthesis and fluorescent protein expression, and its mode of action was analyzed using in silico methods.Next, two BBE-like proteins, CjBBE and McBBE, were identified through database mining and engineered by replacing their Nterminal peptide with maltose-binding protein.This led to an improvement in the conversion of reticuline to scoulerine compared to the wild-type PsBBE.Furthermore, the team analyzed the protopine 6-hydroxylases and engineered its transmembrane domains by creating a chimeric enzyme EcCFS1-83-McP6Hs 84-522.This resulted in enhanced protein expression and improved localization.Finally, the team optimized the precursor supply and made other routine improvements.All of these efforts led to the achievement of the highest reported level of sanguinarine production.With this foundation, the authors were able to produce fluorinated sanguinarines by introducing fluorinated precursors into the cell chassis.Overall, this work demonstrates how to improve the production of plant natural products by comprehensively utilizing pathway engineering and metabolic engineering.The workload for this project is substantial, and the outcomes are good.Publication in Nature Communications is recommended, but the following concerns need to be addressed.
Thanks for the reviewer's positive comments on our manuscript.
1.The current manuscript is lengthy, and it would be better to reduce the text by 10-20% and eliminate redundancy.
Thanks for the reviewer's suggestion.We agree with the reviewer that our manuscript is lengthy, as we tried our best to include as much info as possible.Accordingly, we have taken steps to address this issue, such as revising the Introduction and Discussion sections, moving some materials and methods into the supporting information, and trimming redundant background information at the beginning of each result section.
2. It appears that downstream enzymes cannot tolerate the halogen substitution, and only fluorine can be transferred to the end product.The authors should briefly discuss it.In addition, the characterization of F-sanguinarin is based solely on MS analysis.It is necessary to verify the chemical structure using NMR.
Thanks for the reviewer's comments.We agree with the reviewer that it is better to briefly discuss the characteristics of halogenated derivatives entering the sanguinarine pathway.Enzyme promiscuity confers adaptability to the yeast cell factory, allowing it to accommodate substrates like halogenated tyrosine, aiding exploration of new chemical spaces for drug discovery.However, halogenated derivatives, not natural substrates of the pathway enzymes, exhibit lower conversion yield, especially in longer pathways.Our current results leave some ambiguity regarding whether downstream enzymes, such as reticuline oxidase (BBE), cannot tolerate chlorine substitution or there's an insufficient supply of the precursor Cl-reticuline, resulting in undetectable Cl-scoulerine.Fluorination, owing to its reduced steric hindrance, appears to be more readily integrated into the pathway compared to chlorination and iodination.Similar Regarding the structural characterization of F-sanguinarine, as suggested by the reviewer, we agree with the reviewer to recognize NMR as the most reliable approach.
However, the low yield of the new-to-nature compound F-sanguinarine, currently only detectable, poses challenges in isolating and purifying sufficient quantities for NMR analysis.Similar studies relied solely on mass spectrometry for accurate molecular weight analysis as well, such as Bradley, S. A. et al. Nat Chem Biol 19, 1551-1560(2023), Han, J., Li, S. Commun Chem 6, 27 (2023), and Li, Y. et al Proc Natl Acad Sci U S A 115, 17 (2018).
Nevertheless, we have devoted every effort to systematically characterize Fsanguinarine.Initially, we compared the MS 2 spectra of sanguinarine and Fsanguinarine using triple quadrupole (QQQ) mass spectrometry (Fig. 7c), confirming the same molecular weight differences in precursor ([F] -=18, sanguinarine [M] + =332, F-sanguinarine [M] + =350) and characteristic fragments (sanguinarine: 245.9, 273.9;Fsanguinarine: 263.9, 291.9).Building on this, we further analyzed F-sanguinarine's exact molecular weight (accurate to 4 decimal places, Fig. 7e and Supplementary Fig. S41) and isotopic distribution (Supplementary Table S3) using high resolution qTOF mass spectrometry.By comparing with the sanguinarine standard, analyzing the exact molecular weight, and considering isotopic distributions, we believe our conclusion is well-supported.Accordingly, we have revised the manuscript to provide a more systematic description of characterizing F-sanguinarine in Lines 493-495 of the revised manuscript and added the isotopic information of precursor and characteristic fragments for F-sanguinarine (Supplementary Table S3).Qualitative results analyzed using MassHunter software (Agilent Technologies) are summarized in the source data.S3 The isotopic distribution of F-sanguinarine Reviewer #2 (Comments for the Author):

Supplementary Table
The authors report the strategy of the engineered yeast to produce sanguinarine, a potential antimicrobial and anti-cancer compound, achieving a high yield of 715.12 mg/L.A temperature-responsive gene expression system was developed to mitigate cytotoxic effects and enhance production.The work paves the way for scalable yeast-based production of various benzylisoquinoline alkaloids and their derivatives.
It is supported by detailed research and the text is well theorised and described.The following points could be improved.
Thanks for the reviewer's positive comments on our manuscript.
1.In fed-batch fermentation, sanguinarine production increased 7-fold to 700 mg/L.Is this just an increase in biomass compared to flask fermentation?If not, you should discuss this and show data comparing the biomass in flask fermentation to prove that it is not just an increase in biomass.
Also, is it possible to show the validity of SIMTeGES by determining the production of sanguinarine under normal culture conditions in fed-batch fermentation?
Thanks for the reviewer's comments.As depicted in Supplementary Fig. S26, the OD600 of the SAN231 strain in shake flasks was ~10, while in the fermenter, the average OD600 reached ~130, representing a substantial increase in biomass by ~13-fold during fedbatch fermentation.This significant rise in biomass indeed contributes to the overall improvement in sanguinarine production (~7-fold).Nevertheless, we would like to revise the fed-batch fermentation results.Despite repeated attempts post-submission, only in a few cases could we achieve the results with the sanguinarine titer of ~700 mg L -1 and OD600 of ~130.Most attempts resulted in the sanguinarine titer of ~450 mg L -1 and OD600 of ~80.Therefore, we have updated Figures 6b and 6c with these more representative results.We acknowledge that this discrepancy could potentially stem from additional limitations in sanguinarine production, possibly due to cytotoxicity effects.Anyway, in the case of new results, biomass was increased ~8-fold, while sanguinarine titer was increased by ~4.4-fold.In other words, sanguinarine yield per cell during fed-batch fermentation is lower compared with that in shake flasks, indicating that sanguinarine production was still limited by its cytotoxicity and fermentation conditions should be further optimized.To address these challenges, we are incorporating relevant strategies such as transporter engineering and tolerance engineering into our ongoing research efforts.
To further demonstrate the validity of SIMTeGES by determining sanguinarine production under normal culture conditions, we replaced GAL4-tsINT in strain SAN231 with the wild-type GAL4.To ensure the stability of the strain due to leakage expression of the GAL promoter under low glucose concentrations, we also complemented the transcriptional repressor GAL80, resulting in the construction of strain SAN334.Subsequently, we assessed the sanguinarine production capability of strain SAN334 in shake flasks at both 25 °C and 30 °C, which was compared with strain SAN231.The results indicated minimal variation in sanguinarine production for SAN334 strain at different temperatures, with titers averaging around 26 mg L -1 , representing a decrease of approximately 4-fold when compared with strain SAN231 (Supplementary Fig. S26).This observation underscores the significance of SIMTeGES in enhancing sanguinarine biosynthesis, as highlighted in Lines 446-449 of the revised manuscript.Given the significantly lower sanguinarine production using strain SAN334, we think it unnecessary to perform fed-batch fermentation under normal culture conditions.Thanks for the reviewer's comments.We agree with the reviewer that the effectiveness of SIMTeGES should be further demonstrated in producing compounds without toxic effects.Although our study has already showcased its significant efficacy and adaptability in facilitating non-toxic lycopene biosynthesis, with GAL4-tsINT exhibiting a 15.5-fold increase in lycopene production over GAL4M9 and a 3.7-fold increase over wild-type GAL4 (Supplementary Fig. S7).

Supplementary
Regarding reference 31 highlighted by the reviewer, where the highest titer of reticuline (4.6 g L -1 ) was reported using pulsed fed-batch fermentation, direct comparison of reticuline production becomes complex due to differences in experimental parameters such as promoters (constitutive, e.g., PTDH3, PTEF1, PPGK1), culture media (2×SC), and carbon sources (sucrose).Additionally, the specialized pulsed fed-batch fermentation method employed by the authors resulted in a 10.29-fold increase in titer from 447 mg L -1 to 4.6 g L -1 .Nevertheless, to address the reviewer's concern, we evaluated reticuline accumulation under different carbon sources based on the SAN220, SAN220-M9, and SAN220-tsINT strains with the PsBBE gene knocked out.As illustrated in the figure below, when both glycerol and galactose were present (SEGlyG), GAL4-tsINT exhibited the highest reticuline titer of 270.15 mg L -1 , surpassing GAL4 by 1.89-fold and GAL4M9 by 2.00-fold, respectively (Fig. R1).Given our focus on optimizing sanguinarine production and the occasional need for controlled accumulation levels of reticuline as an intermediate metabolite, achieving equivalent or superior reticuline yields poses challenges without a similar fermentation system (pulsed fed-batch fermentation).However, considering the performance across multiple systems, we speculate that SIMTeGES holds promise for enhancing the biosynthesis of diverse target products by decoupling cell growth from product synthesis, thereby alleviating cellular metabolic burdens and fostering efficient target product biosynthesis.
3. In Figures 3e and 3f, even at non-permissive temperatures, spliced 35 kDa bands are observed along with unspliced 85 kDa bands.And the ratio of spliced to unspliced forms does not appear to match the fluorescence in Figure 3d.Therefore, in Figures 3e   and 3f, the quantification of the ratio of spliced to unspliced forms are necessary, and then the correlation between the ratio and the value of fluorescence should be discussed.
Thanks for the reviewer's comments.We agree with the reviewer that quantifying the ratio of spliced to unspliced forms of the protein in Figures 3e and 3f is important, although our primary aim was to employ western-blot to validate the direct correlation between protein activity and intein splicing.We conducted grayscale analysis of the bands corresponding to mCherry-tsINT at 35 kDa (spliced form) and 85 kDa (unspliced form) under permissive or non-permissive conditions using Image J software.The ratio of unspliced to spliced protein under permissive and non-permissive conditions is 1:5.32 and 4.59:1, respectively (Fig. R2).As noted by the reviewer, there is indeed a higher abundance of spliced protein under non-permissive conditions compared to the fluorescence results.It is noteworthy that the reduction in the 85 kDa protein under permissive conditions is much lower than the increase in the 35 kDa protein.We speculate two main reasons for the observed discrepancy: firstly, intein splicing is a cofactor or energy-independent intramolecular process mainly through bond rearrangement.Although we expedited the sample preparation process as quickly as possible, it still takes longer than fluorescence detection (flow cytometry and fluorescence microplate reader), and protein samples under non-permissive conditions are likely to undergo partial splicing during sample preparation (room temperature is 25 °C).Secondly, the transmembrane efficiency of large molecular weight proteins is relatively low, and there are differences in grayscale values corresponding to proteins of different molecular weights.Therefore, western-blot results can only serve as semiquantitative references and are more suitable for proving the conclusion that temperature-dependent activity relies on intein splicing.Accordingly, to prevent potential misunderstanding, we choose not to put the grayscale analysis results in the manuscript.Expression System, which they call SIMTeGES that enables regulation of expression of heterologous genes.Their system uses temperature as a regulatory signal to separate cell growth from product biosynthesis processes (e.g. for a two-stage process).At 30 °C, optimal for yeast growth, SIMTeGES keeps the GAL4 transcription factor inactive, which allows cell growth.Conversely, at 25 °C, GAL4 is activated by splicing and this promotes the expression heterologous genes under control of the GAL4 responsive promoter.The approach was used to separate cell growth and sanguinarine biosynthesis (a toxic compound for yeast).Additionally the authors show that its generally applicable and facilitates the temperature-controlled expression of additional target proteins (e.g., mCherry, GAL80).
The case study of sanguinarine biosynthesis is impressive as it involves heterologous expression of many enzymes and optimization of several steps.The authors report the highest titer achieved for sanguinarine.The topic of two-stage process and dynamic control of microbial systems is very relevant and the manuscript is well written.I have few points that should be considered: Thanks for the reviewer's positive comments on our manuscript.
1.The manuscript combines two important developments: the SIMTeGES, and the sanguinarine production.The results are structured in this way, but Figure 1 starts with the sanguinarine pathway.I recommend to present the SIMTeGES first.In Figure 1, the expression system should be explained: which genes are under which promoter, where are they integrated into the genome.This would help to follow the metabolic engineering strategy that is described in the second part of the result section.
Thanks for the reviewer's comments.We agree with the reviewer that presenting the SIMTeGES first is beneficial.Consequently, we have reorganized the manuscript to feature SIMTeGES first, with the sanguinarine pathway now described in Fig. 3 of the revised manuscript.We believe this adjustment will offer a clearer delineation of our two main developments.
Due to space limitations, we don't think it possible to include so much detailed info (integration loci, promoters, and genes) in Fig. 3 of the revised manuscript (Fig. 1 of the original manuscript).Instead, we have included the construction processes of the SIMTeGES and sanguinarine bioproduction-related strains in Supplementary Fig. S1 and S14 to further explain the expression system.These figures detail which genes are regulated by which promoter and where they are integrated into the yeast genome.
Additionally, in conjunction with the yeast strains and cell information in Table 1 and plasmid information in Supplementary Table S6, we have also provided detailed information about the integration sites in Supplementary Table S7.These additions hopefully enhance the understanding of the metabolic engineering strategy described in the results section.
Supplementary Fig. S1 Construction procedures for SIMTeGES-related strains.Detailed depiction of the integrated gene cassette with the utilized promoters and their corresponding integration sites.Detailed information regarding the integration sites can be found in Supplementary Table S7.
2. The authors should describe how the temperature sensitive (TS) intein VMAL212P was derived.Did they screen for the TS phenotype, did they find additional TS alleles?
doi: 10.1534/genetics.109.104794) and found that only the L212P mutant exhibited a distinct temperature-sensitive phenotype under a modest temperature difference (5 °C , Supplementary Fig. S4).While other mutants have been reported to show different temperature-sensitive characteristics, they did not meet the specific requirements for our application.Such discrepancy may result from the use of different expression system: plasmid expression in previous studies and genome integrated expression in the present study.Consequently, we focused our investigation primarily on the VMA L212P variant (a.k.a.tsINT), applying it to different target proteins (mCherry, GAL80, and GAL4) and in various host systems (S.cerevisiae, P. pastoris, and mammalian cells).Additionally, we utilized molecular dynamics (MD) simulations to gain insights into the temperature dependence mechanism.To address the reviewer's concern, we have incorporated additional details regarding the intein mutant screening process in Lines 157-159 of the revised manuscript.
5. The authors use complex medium for their bioprocess for sanguinarine biosynthesis.
At the time point of sanguinarine synthesis the main carbon source glycerol and galactose are depleted.Therefore, the authors should show which carbon source goes into sanguinarine biosynthesis.It is not clear if tyrosine fed to the cultures in Figure 6 or if tyrosine was derived from the complex medium.
Thanks for the reviewer's comments.We agree with the reviewer that providing additional clarity regarding the flux of carbon sources into sanguinarine biosynthesis is necessary.In our study, we primarily utilized 20 g L -1 of galactose as the main carbon source in synthetic medium (SE), which undergoes assimilation to produce glucose-6phosphate (G6P).Subsequently, phosphoenolpyruvate (PEP) and erythrose-4phosphate (E4P) are generated via the pentose phosphate pathway (PPP) and glycolytic pathway, respectively, contributing to sanguinarine biosynthesis through the shikimate pathway, as illustrated in Supplementary Fig. S17b.Additionally, glycerol served as a secondary carbon source, converted to dihydroxyacetone phosphate (DHAP), thereby bolstering the supply of PEP.Given the comparatively lower assimilation rates in S.
cerevisiae, we used 10 g L -1 glycerol as non-inhibitory and non-inducing carbon source for seed culture.As shown in Fig. 6b, sanguinarine biosynthesis was initiated until all glycerol was consumed during fed-batch fermentation.In other words, glycerol was mainly used for cell growth, while galactose was used for both cell growth and sanguinarine biosynthesis.Therefore, we tentatively conclude that galactose serves as the main carbon source for sanguinarine biosynthesis.
Supplementary Fig. S17b Schematic representation of DEGs within central carbon metabolic pathways when glycerol and galactose are carbon sources, with upregulated genes in red and downregulated genes in green.
8) Page 18, line 385: Fig. 5C should be corrected to Fig. 5B.Page 18, line 391: Fig. 5D should be corrected to Fig. 5C.9) Page 23, line 487: Is Fig. 7B a C? Please check the citations in Fig. 7A-E as the text and figures do not match.

Fig. 7a -
Fig. 7a-e not aligning with the figures.We have corrected this by updating the citation in Page 23, Line 488 from "Fig.7b" to "Fig.7c and 7e" and have meticulously reviewed the citation issues concerning Fig. 7a-e in the revised manuscript.