Equipping Saccharomyces cerevisiae with an Additional Redox Cofactor Allows F420-Dependent Bioconversions in Yeast

Industrial application of the natural deazaflavin cofactor F420 has high potential for the enzymatic synthesis of high value compounds. It can offer an additional range of chemistry to the use of well-explored redox cofactors such as FAD and their respective enzymes. Its limited access through organisms that are rather difficult to grow has urged research on the heterologous production of F420 using more industrially relevant microorganisms such as Escherichia coli. In this study, we demonstrate the possibility of producing this cofactor in a robust and widely used industrial organism, Saccharomyces cerevisiae, by the heterologous expression of the F420 pathway. Through careful selection of involved enzymes and some optimization, we achieved an F420 yield of ∼1.3 μmol/L, which is comparable to the yield of natural F420 producers. Furthermore, we showed the potential use of F420-producing S. cerevisiae for F420-dependent bioconversions by carrying out the whole-cell conversion of tetracycline. As the first demonstration of F420 synthesis and use for bioconversion in a eukaryotic organism, this study contributes to the development of versatile bioconversion platforms.


■ INTRODUCTION
F 420 is a naturally occurring deazaflavin cofactor synthesized only by certain bacteria and archaea, such as actinobacteria and methanogenic archaea. 1 While having a similar structure as the ubiquitous flavin cofactor FAD, the chemical properties of F 420 are more like nicotinamide cofactors, as it exclusively performs hydride transfer reactions due to the C5 of the 5-deazaisoalloxazine moiety.−4 Furthermore, the low redox potential of F 420 compared to the flavin cofactors FMN and FAD, and even to NAD(P)H, allows the reduction of recalcitrant substrates, expanding the scope of the currently available applications of enzymatic reductions. 2,5espite the potential use of F 420 for various industrial applications, the biosynthesis of this cofactor is limited to the use of natural producers such as Mycobacterium smegmatis, which hinders the cofactor availability and thus the related research.Therefore, its heterologous production in more versatile organisms such as Escherichia coli and yeast can be an attractive solution for easy access to this deazaflavin cofactor.−9 As a substitute to F 420 , a synthesis of structurally much simpler and yet functional non-natural deazaflavin analogue FOP has also been explored using both E. coli and Saccharomyces cerevisiae, offering an attractive alternative solution. 10ither naturally or non-naturally, F 420 has so far been synthesized only in prokaryotic organisms.In this study, we explored the biosynthesis of F 420 in S. cerevisiae to extend the F 420 -dependent biosynthesis platform even further to eukaryotic organisms.S. cerevisiae is a widely used organism across laboratories and industries due to its robust and harmless nature as well as its well-understood biophysical properties and well-developed molecular biological tools.Therefore, producing F 420 in this versatile organism can expand biotechnological means for related research and applications.
−14 Using the available information, we explored the use of these enzymes for the production of F 420 in S. cerevisiae by testing the expression and their in vivo functions.Initial studies were performed using plasmid-based expression of the involved enzymes, and we confirmed that expression of CofC (guanylyltransferase) and CofD (FO transferase) from M. jannaschii along with FbiB (glutamyl ligase) from M. smegmatis can produce F 420 in a good yield when the precursor FO is provided.Based on this result, we constructed a F 420 -producing S. cerevisiae strain by CRISPR-Cas mediated genomic integration of the three genes.The F 420 yield after some optimization was comparable to the yield from natural producer M. smegmatis.
We then further explored the F 420 -producing S. cerevisiae strain for use in F 420 -dependent bioconversions.To demonstrate that the attained intracellular levels of F 420 in S. cerevisiae can support new metabolic activities, we introduced several bacterial enzymes that catalyze the last steps of tetracycline synthesis.These involve the selective reduction of the C5a− C11a double bond of dehydrotetracycline, which requires reduced F 420 as electron donor. 15By expressing the last two enzymes of the tetracycline biosynthesis as well as an F 420reducing enzyme from Cryptosporangium arvum (FSDcryar, an F 420 -dependent sugar-6-phosphate dehydrogenase) 16 in the F 420 -producing S. cerevisiae strain, we could successfully convert anhydrotetracycline into tetracycline.This clearly shows the potential use of the strain for F 420 -dependent bioconversions.
Overall, the significance of the current study lies in the demonstration of the first eukaryotic production of F 420 and the F 420 -dependent bioconversions using yeast.In addition to the previously reported E. coli-based production of the cofactor, this will expand the tools for F 420 -related research.

■ RESULTS AND DISCUSSION
In Vivo FO Synthesis in S. cerevisiae is Hindered by Deficient Expression of FO Synthases.The catalytic core of F 420 , the 7,8-didemethyl-8-hydroxy-5-deazariboflavin (FO) moiety, is synthesized from tyrosine and 5-amino-6-(ribitylamino)-uracil.The reaction is performed by an FO synthase which mostly is (1) a bifunctional enzyme (FbiC) in actinobacteria, or (2) involves two separate enzymes (CofG and CofH) in Archaea. 1 For in vivo production of FO, we attempted the expression of several FO synthases from different organisms including FbiCs from M. smegmatis (MsFbiC), M. tuberculosis (MtFbiC), and a eukaryote Chlamydomonas reinhardtii (CrFbiC)�some eukaryotes use FO as a chromophore in the DNA repair process 17 �as well as CofG and CofH from M. jannaschii (mjCofG and mjCofH).The codon-optimized FO synthases were transformed in S. cerevisiae, and in vivo FO synthesis was analyzed.However, none of the FO synthases seem to express or function in S. cerevisiae, as no FO was detected in the cell extracts or in the culture media after the growth of the cells.Supplementing the growth media with tyrosine and methionine which are the precursors of FO and the cofactor SAM, respectively, did not change the result.SDS-PAGE analysis revealed no apparent protein bands for expressed FO synthases except for MjCofG (Figure S1).Previous studies showed that in E. coli very low expression of FO synthases was sufficient for the in vivo FO production. 6,10It is plausible that the expression of the SAMdependent Fe−S cluster-containing enzymes can be problematic due to the different Fe−S cluster assembly pathways in prokaryotes and eukaryotes, causing the disturbed expression or malfunction of the enzyme. 18,19As the functional expression of the different FO synthases failed, we have focused on building the F 420 pathway using the chemically synthesized FO.This would be analogous to the use of riboflavin as a precursor for flavin cofactors.
Coexpression of CofC and CofD Enables the In Vivo Production of Dehydro F 420 (DF 420 ) in S. cerevisiae.The precursor in F 420 biosynthesis was initially identified to be 2-PL but in recent studies, PEP was also shown to be a precursor in some organisms. 6,7When PEP is used as a precursor, dehydroF 420 -0 (DF 420 -0) instead of F 420 -0 is produced and subsequently reduced to F 420 -0 by an FMN-dependent catalysis (Grinter 2020). 11In some organisms, yet another precursor, 3-PG, was found to be used as the precursor, which results in the production of a F 420 analogue, 3-PGF 420 . 7The substrate specificities of the enzymes involved in attaching these moieties to FO, i.e., guanylyltransferase (CofC or FbiD) and FO transferase (CofD or FbiA), are therefore different between enzyme homologues. 12Since 2-PL is not known to be a common metabolite in S. cerevisiae, a guanylyltransferase and a FO transferase that are active on the more accessible Figure 1.Biosynthetic pathway of F 420 .The scheme is adapted from Bashiri et al. 6 The pathway represents the F 420 synthesis using 2phosphenolpyruvate (PEP) as a precursor.F 420 can also be synthesized using 2-phospho-L-lactate (2-PL) or 3-phospho-D-glycerate (3-PG), in which case the dehydro-F 420 -0 is not formed, therefore no reduction step to F 420 -0 is required. 6,7Generally, the Fbi-prefix is used for the enzymes from mycobacteria, and Cof-represents the homologues from archaea.
substrates PEP and EGGP, respectively, seem more suited for in vivo F 420 synthesis in S. cerevisiae.
Based on the available data from previous studies, we selected two guanylyltransferases from M. smegmatis and M. jannaschii (MsFbiD and MjCofC, respectively) as well as three FO transferases from M. smegmatis, M. jannaschii, and M. mazei (MsFbiA, MjCofD, and MmCofD, respectively).In order to select the best combination of two enzymes for the in vivo synthesis of DF 420 -0, we first expressed these enzymes individually in S. cerevisiae and performed reactions with mixtures of cell extracts.Among the possible six combinations of the cell extract mixtures of guanylyltransferase and FO transferase, the combination of MjCofC−MjCofD and MjCofC−MmCofD seemed to function, showing an additional peak when compared with the control reaction on HPLC analysis (Figure 2a).The F 420 spiked reaction product of MjCofC−MmCofD shows that the additional peak is from a potentially F 420 -like product.The products were purified and further analyzed using LC−MS, which indeed showed the mass corresponding to that of DF 420 -0 (Figure 2b).Previously it was shown that MsFbiA exclusively uses EPPG as a substrate and MsFbiD preferably accepts PEP over 2-PL when studied in vitro. 6,12However, in this study, reactions with MsFbiD and/ or MsFbiA did not show any products, which might be due to insufficient expression.SDS-PAGE analysis (Figure S2) showed no apparent overexpression of MsFbiA and visibly lower expression of MsFbiD compared to its homologous enzyme MjCofC.Based on this result we selected MjCofC and MjCofD for in vivo F 420 production.When coexpressed in S. cerevisiae, these two enzymes produced DF 420 -0 in vivo using the FO provided in the media, which was confirmed by HPLC and LC−MS analysis (Figure 2c,d).Although it was previously suggested that mainly 2-PL was used as a precursor in archaea, 12,20 the archaeal enzymes MjCofC and MjCofD expressed in S. cerevisiae seemed to exclusively use PEP, producing DF 420 -0 as the only detectable product.This result indicates that the in vivo PEP concentration in S. cerevisiae is sufficient for these enzymes to produce DF 420 -0 and that no or an insignificant amount of 2-PL is present in the yeast cells.

S. cerevisiae Strain Expressing FbiB from M. smegmatis along with MjCofC and MjCofD Produces
F 420 -n.The final step of F 420 biosynthesis is the elongation of F 420 -0 with glutamyl tails in varying length by CofE or FbiB.When PEP is used as a precursor, the reduction of the intermediate product dehydroF 420 -0 (DF 420 -0) to F 420 -0 is additionally required.In mycobacteria, the bifunctional glutamyl ligase FbiB also performs the FMN-dependent reduction of DF 420 -0 to F 420 -0 11 in addition to the glutamyl ligation reaction.However, in organisms that use monofunctional glutamyl ligase CofE, such as methanogenic archaea, the reduction is possibly performed by a "stand-alone nitroreductase" 21 or a yet-unknown enzyme.Considering the convenience of using the bifunctional enzyme for both the reduction of DF 420 -0 and the glutamyl ligation, we chose FbiB from M. smegmatis (MsFbiB) for the final step of F 420 biosynthesis.Three enzymes, MjCofC, MjCofD, and MsFbiB, were successfully expressed in S. cerevisiae on two separate plasmids (MjCofC and MjCofD on one and MsFbiB on the other).Gratifyingly, when the strain was grown in the media containing FO in vivo, the production of F 420 was detected (Figure S3a).The HPLC analysis showed several peaks of potential F 420 species, which were confirmed by the LC−MS analysis.The mass spectroscopy data showed that the produced F 420 species mostly contained five or six glutamyl moieties (Figure S3b).
After establishing a set of functional F 420 pathway enzymes in yeast, we developed an S. cerevisiae strain that has a F 420 synthetic pathway built in using CRISPR-Cas9-mediated genomic integration.One copy of each MjCofC, MjCofD, and MsFbiB genes was cloned at the HO locus as described in the Materials and Methods.The resulting strain is herein referred to as Sc-F 420 .In cell extracts of the Sc-F 420 strain grown in synthetic defined (SD) media supplemented with FO, peaks corresponding to F 420 were detected by HPLC-FLD, which were expectedly absent in the control sample of the wild-type S. cerevisiae strain (Figure 3).With the HPLC method used for verifying the in vivo F 420 production, four apparent peaks for F 420 were visible in both the standard samples purified from M. smegmatis and Sc-F 420 samples.In order to identify the peaks, we purified the in vivo reaction products in two purification steps using anion exchange chromatography and reverse phase HPLC.The F 420 species purified from the cell extracts using anion exchange chromatography were further separated by HPLC-FLD with an optimized elution method and collected manually.The purified F 420 products were analyzed using LC−MS.The mass analysis confirmed that all of the peaks corresponded to F 420 species ranging from F 420 -6 to F 420 -2 in the descending order of the retention time (Figure S4).Interestingly, the mass of one of the peaks corresponded to the unreduced (dehydro)F 420 species with a single glutamyl tail, DF 420 -1.It shows that the glutamate ligation reaction of FbiB can occur before the reduction reaction.Confirming that all of the peaks that were detected in HPLC-FLD correspond to F 420 species, the combined peak area was used for further quantification of F 420 production.
In order to optimize the F 420 yield, we tested the effect of the growth media (Table 1).80 mg/L glutamate was added in synthetic defined media (SD and Verduyn media) to support the glutamyl ligation reaction.Among the media tested, the highest yield per culture volume was shown when a rich medium (YPD) was used.When normalized by the amount of the biomass produced, Verduyn medium (VD) with the glutamate supplement was as effective as YPD yielding around 300 nmol/g dry biomass.SD medium was least favorable for F 420 production, yielding about 100 nmol/g dry biomass.Furthermore, when SD media was used, the strain only reached an OD 600 of ∼5.7, while OD 600 values of 12.8 and 23.6 were reached when VD medium or YPD medium was used, respectively.The lower cell density also contributed to the low yield per volume of culture, which improved slightly when glutamate was supplemented.The estimated in vivo concentration of F 420 in cells grown in YPD and VD media with  The experiments were performed in biological duplicates.b The correlation between the cell dry weight (DW) and the OD 600 was determined (1 OD 600 unit ≈ 0.37 mg/mL) to estimate the dry weight of the culture.c The in vivo F 420 was estimated based on the reported cell volume per biomass of S. cerevisiae. 26lutamate was similarly high, reaching over 100 μM.Therefore, while YPD would be the best choice of media for the F 420 production purpose, VD media would be the better choice for in vivo F 420 -dependent conversion as it offers more flexible options of using selective markers for expressing additional enzymes if required.−25 Throughout the study, 200 μM FO was added in the media for F 420 production.As the concentration of FO in the media can affect the physiological state of the cells as well as the F 420 yield, we evaluated the effect of different FO concentrations on the cell growth and F 420 production.The addition of up to 400 μM FO did not influence the growth of the Sc-F 420 strain as the final OD 600 upon harvest at 48 h incubation was similar, between 12 and 13, regardless of the FO concentrations added.The FO concentration showed positive correlation to the F 420 yield as expected and this may be an indication of the low FO import efficiency into the cell (Figure 4).Even though we tested only up to 400 μM FO due to its poor solubility, it is possible that higher concentrations (if solubilized) of FO could further increase the F 420 yield.For the purpose of F 420 production, the highest FO concentration possible should be used, while it seems that a lower concentration (ex.200 μM) is enough to produce sufficient in vivo F 420 concentration for a F 420 -dependent bioconversion using cells.
The experiments are performed in duplicate, and the error bas represent the standard deviation.
In the above-described end-point measurements, a rather long incubation time of 48 h was used in order to guarantee a sufficient time for the full growth and the maximum F 420 yield.In order to improve the efficiency of F 420 production of the Sc-F 420 strain, we tested the F 420 yield at different incubation times (Table 2).Between 24 and 48 h, the cells still seem to be growing as interpreted by the increasing OD 600 .The F 420 yield per volume culture increased between 24 and 36 h incubation time, which coincide with the increasing cell biomass.However, despite the further increased OD 600 at 48 h, the F 420 yield decreased slightly compared to that at 36 h incubation time.The productivity (nmol/g dry biomass), on the other hand, was the highest at 24 h showing almost 500 nmol/g DW.As a result, it also showed the highest estimated in vivo F 420 concentration of ∼180 μM.Therefore, for further experiments, we incubated the cells only for 24 h.The F 420 yield achieved with Sc-F 420 is comparable to the one with M. smegmatis but in a much shorter incubation time (Table 3).
F 420 Producing S. cerevisiae Strain Sc-F 420 Can be Used for F 420 -Dependent Bioconversion.Upon confirmation that in vivo F 420 production in yeast reached high F 420 levels, we explored the possibility to build a Sc-F 420 -based bioconversion system.The final steps of the tetracycline synthesis involve a specific F 420 -dependent enzyme-catalyzed reaction.OxyR and CtcM produce tetracycline by catalyzing the reduction at 5a(11a) of dehydrotetracycline, which is essential for the potency of the bacterial antibiotic. 29,30In the natural biosynthesis in Streptomyces species, OxyR and CtcM are known to be involved in the synthesis of oxytetracycline and chlorotetracycline, respectively. 15However, in vitro studies showed that both OxyR and CtcM can reduce 5a(11a)dehydrooxytetracycline as well as 5a(11a)-dehydrotetracycline, producing oxytetracycine and tetracycline, respectively (Figure 5). 15,31OxyS, a flavin-dependent enzyme in the pathway, performs a single or double hydroxylation on anhydrotetracycline producing 5a(11a)-dehydrotetracycline or 5a(11a)dehydrooxytetracycline, respectively, which can be subsequently reduced by the above-described F 420 -dependent enzymes.As a proof of concept of a Sc-F 420 -based bioconversion, we set out to demonstrate the F 420 -dependent last step of tetracycline conversion using OxyR and CtcM.As anhydrotetracycline, and not the hydroxylated product, is commercially available, we also employed OxyS for the hydroxylation reaction.
In order to produce tetracycline from anhydrotetracycline using in vivo-produced F 420 , we coexpressed OxyS together with either OxyR or CtcM on a plasmid in the Sc-F 420 strain.For an F 420 -dependent reduction reaction, the cofactor needs to be reduced and we employed the F 420 -dependent glucose-6phosphate dehydrogenase from C. arvum, expressed on a separate plasmid.This bacterial enzyme can conveniently use the available glucose-6-phosphate to generate reduced F 420 (F 420 H 2 ).After 24 h of cultivation in FO containing media and subsequent incubation with anhydrotetracycline, both Oxy-  The experiments were performed in triplicates.The values represent the average and the errors are the standard deviation.The yield per biomass and the in vivo F 420 concentration were calculated as described in Table 1.S_OxyR and OxyS_CtcM expressing Sc-F 420 strains seemed to be able to produce tetracycline which was analyzed by HPLC and LC−MS methods (Figure 6).In the control reactions, wild-type CEN.PK yeast strain expressing OxyS_CtcM and FSDcryar as well as FSDcyar-absent Sc-F 420 strain with or without OxyS_CtcM, no tetracycline related products were detected, indicating that the tetracycline was indeed produced by OxyS_OxyR or OxyS_CtcM using the in vivo produced and regenerated F 420 H 2 .
Both the Sc-F 420 strains expressing OxyS_OxyR or OxyS_CtcM produced tetracycline but different analogues.While OxyR produced a significant portion of oxytetracycline along with the standard tetracycline, CtcM exclusively produced tetracycline and no detectable amount of oxytetracycline.This result is in line with a previous study where CtcM showed higher specificity toward dehydrotetracycline (single hydroxylation product of OxyS) thus producing tetracycline primarily. 15On the contrary, a previous study showed that in vitro reactions using FO and crude extract of yeast cells expressing OxyS, OxyR, and a F 420 -reducing enzyme FNO did not yield any tetracycline products. 31It is likely due to the nonactive OxyR when FO is used instead of F 420 .Even though it is possible to substitute F 420 with FO or a non-natural analogue FOP for some F 420 -dependent reactions, 32 the application is limited to the cofactor specificity of each enzyme.Therefore, conversion using the F 420 -producing system as described in this study is at an advantage for exploiting F 420 -dependent enzymes.
The production of tetracyclines in the Sc-F 420 strain demonstrates that the strain can produce enough F 420 for F 420 -dependent reduction and shows the potential for further exploration of the strain for F 420 -dependent bioconversions.Even though the F 420 yield is lower than the previously constructed and engineered F 420 -producing E. coli strains, 8,9 the development of the Sc-F 420 strain as a first F 420 producing eukaryotic organism expands the tools for F 420 related research.

■ CONCLUSIONS
F 420 is a unique cofactor that structurally resembles the canonical flavin cofactor FAD while having a similar chemical property as the nicotinamide cofactors NAD and NADP.Despite the potential value of F 420 for biotechnological applications, research on this cofactor and the respective F 420 -dependent enzymes has been limited by the constrained production of the cofactor using unconventional laboratory organisms such as M. smegmatis.In this study, we tackled this problem by producing the cofactor in a robust industrial microorganism, S. cerevisiae, through heterologous expression of part of the F 420 pathway.By optimizing the enzyme combination and the growth medium, we achieved comparable F 420 yields to that of the natural production by M. smegmatis but in a much shorter production time.Furthermore, we demonstrated the use of the F 420 -producing S. cerevisiae strain for F 420 -dependent bioconversion by showing the successful bioproduction of tetracycline from anhydrotetracycline. Together with the previously developed F 420 -producing E. coli strain by others, the first F 420 producing eukaryotic strain developed in this study extends the toolbox for further development in the field of F 420 -related research.

Strains and Plasmids.
All plasmids used in this study are E. coli�yeast shuttle vectors assembled using a modular cloning kit, Moclo-YTK from Addgene and the assembly was performed as described previously 34 with some modification in the Golden Gate assembly methods.E. coli NEB10-beta strain was used for cloning purposes.CEN.PK2-1C strain was purchased from Euroscarf and used for constructing the F 420producing yeast strain.
Growth Media.For the growth of E. coli, Lysogeny broth medium containing an appropriate type of antibiotic (100 μg/ mL ampicillin, 50 μg/mL kanamycin, or 50 μg/mL chloramphenicol) was used.Media for S. cerevisiae used in this study are SD medium, YPD medium (formedium), and VD.The SD medium is composed of 6.9 g/L yeast nitrogen base without amino acids (formedium), 0.77 g/L complete supplement mixture (formedium) that is appropriate for the auxotroph markers, and 2% (w/v) glucose.VD containing 2% glucose was made according to Verduyn et al. 35 For F 420 production in S. cerevisiae, 72 mg/L FO was added to respective media prior to autoclave sterilization.FO was chemically synthesized as described in Drenth et al. 32 For optimization of F 420 production, 80 mg/L glutamate was added to the respective medium.
Cloning.All S. cerevisiae codon-optimized genes tested for F 420 biosynthesis and F 420 -dependent bioconversion were purchased from Twist Bioscience.The gene fragments were first cloned into an entry vector and subsequently cloned into a preassembled E. coli-yeast shuttle vector by Goldengate assembly method according to the Moclo-YTK cloning protocols. 34All cloning products were initially transformed in E. coli NEB10b strain using the heat shock method, isolated, and analyzed by sequencing at Eurofins.The correct cloning products were then transformed to S. cerevisiae CEN.PK2-1C using the lithium acetate/single-stranded carrier DNA/PEG method that is optimized by Gietz et al. 36 Figure 5. Scheme of (oxy)tetracycline synthesis from anhydrotetracycline using OxyS, OxyR, or CtcM.
Construction of the F 420 -Producing Sc-F 420 Strain.The genes encoding MjCofC, MjCofD, and MsFbiB were integrated in the HO locus using a Crispr-Cas9-mediated method.A vector containing an sgRNA sequence and Cas9 expression cassette was assembled using a Moclo YTK kit.The 20mer target sequence of sgRNA is 5′-GCTCCAGCATTA-TAGCATGC-3′.The vector contained CEN6/ARS4 origin, and pPGK1 promoter was used for Cas9 expression.The repair fragment was constructed via assembling the F 420 pathway genes, HO locus homology fragments as well as a leu3 into a multigene plasmid and subsequently linearizing by NotI digestion.The resulting fragment contained 5′-HO homology sequence, cof D, cof C and f biB, leu3, as well as 3′-HO homology sequence in this order.The transformation of CEN.PK21-C strain with Cas9_sgRNA plasmid and the repair fragment was performed following the protocol of Gietz et al. 36 Approximately 500 ng of Cas9_gRNA plasmid and 5 μg of repair fragment was used for the transformation.The PCR verification of the correct integration was performed on the genomic DNA extracted from selected colonies.Cas9-gRNA plasmid was removed from the transformants by growing the cells in nonselective YPD media and confirming the loss of uracil selectivity that the plasmid was carrying.
In Vitro and In Vivo DF 420 Conversion.In order to find a functional combination of PEP guanylyltransferase (FbiB/ CofC) and FO transferase (FbiA/CofD) for in vivo DF 420 production, reactions using crude extracts mixture of S. cerevisiae expressing each of the enzyme types were tested.Cells expressing FbiB from M. smegmatis, CofC from M. jannaschii, FbiA from M. smegmatis, CofD from M. jannaschii, or CofD from M. mazei were grown in 20 mL SD media with 2% glucose for 24 h at 30 °C.Cells were harvested and washed with ddH 2 0 and resuspended in 1 mL of 50 mM KPi, pH 7.0 containing 1 mg/mL Zymolyase (Amsbio).The cell resuspension was then incubated for 20 min at 30 °C and subsequently vortexed in the presence of equal volume of glass beads (five repeats of 5 s vortexing and 10 s resting on ice).Total protein concentration in the crude extracts was measured by Bradford assay.One ml reaction mixtures in 50 mM KPi, pH 7.0 containing 1 mM GTP, 1 mM phosphoenolpyruvate, 5 mM MgCl 2 , 200 μM FO and mix of CofC (FbiD)-and CofD (FbiA)-containing crude extracts (normalized to 1 mg of total ).Peaks 1 and 3 also show the mass that corresponds to tetracycline with slightly different m/z compared with peaks 2 and 4.This peak is also appearing in the standard tetracycline sample (b) as a minor peak.It is possible that tetracycline is epimerized in the acidic LC−MS analysis condition and eluted separately. 33OxyS_OxyR expressing strain also produces oxytetracycline, which is shown in peak 1, representing the [M + H] + of 461.438.In contrast, OxyS_CtcM expressing strain did not produce any detectable amount of oxytetracycyline (peaks 3 and 4).
protein each) were incubated at 30 °C for 4 h.The reactions were stopped by heating at 95 °C for 10 min and subsequently centrifuged and filtered.The reaction samples were analyzed by HPLC method for DF 420 production.
F 420 Production Using S. cerevisiae.For in vivo production of F 420 in S. cerevisiae, the respective cells were first grown in 5 mL of SD media with 2% glucose overnight at 30 °C.The precultures were then diluted to OD 600 ∼ 0.2 in an appropriate FO-containing media and grown at 30 °C.Unless otherwise stated, the cells were harvested after 24 h and washed with ddH 2 O.In order to extract the in vivo produced F 420 , the cells were incubated with approximately 4× cell volume of 70% boiling ethanol at 95 °C for 5 min.The supernatants were collected after centrifugation at 8000g for 10 min, and the process was repeated once.The collected solutions were subjected to vacuum centrifugation at 60 °C for 1 h in order to remove ethanol and concentrate.The products were resuspended in ddH 2 O, centrifuged, and filtered prior to the analysis.
Tetracycline Conversion.F 420 -producing yeast strain Sc-F 420 expressing OxyR or CtcM as well as OxyS and FSDcryar were grown in 20 mL VD media containing 80 mg/L glutamate and 72 mg/L FO for 24 h in order to accumulate the in vivo F 420 .Cells were collected by centrifugation at 5000g for 10 min and resuspended in 1 mL of reaction solution containing 1 mM anhydrotetracycline, 5 mM glucose, and 100 mM Tris•HCl, pH 7.5.The reactions were incubated for 24 h at 30 °C and subsequently extracted three times with 2 mL of EtOAc.The reactions were dried by vacuum centrifugation for 1 h, redissolved in 1 mL of ddH 2 0, and centrifuged for 10 min at 11.000g to remove any debris prior to analysis.

Figure 2 .
Figure 2. DehydroF 420 (DF 420 ) production in S. cerevisiae.(a) HPLC analysis of in vitro reaction using S. cerevisiae cell extracts expressing various CofC (FbiD) and CofD (FbiA).The control reaction was performed with wild-type S. cerevisiae cells containing an empty plasmid.The orange chromatogram shows the reaction product of MjCofC and MmCofD, which was spiked with purified F 420 .(b) LC−MS analysis shows DF 420 -0 as the reaction product of MjCofC−MjCofD and its fragmented ion with m/z of 424.399 [M − H] − .(c) The HPLC result and d.LC−MS identification of in vivo DF 420 -0 production was performed using S. cerevisiae expressing MjCofC and MjCofD.The calculated mass of DF 420 -0 is 513.350.The ionized molecule with a m/z of 424.396 [M − H] − is expected to be a fragment of DF 420 -0 as shown in panel b.

Figure 3 .
Figure3.F 420 production in S. cerevisiae.F 420 purified from M. smegmatis is used as the standard sample.The wild-type S. cerevisiae grown in media containing FO is used as the control.Only the samples from strain Sc-F 420 shows the F 420 -corresponding peaks.

Figure 4 .
Figure 4. Effect of the FO concentration on the F 420 yield.

Figure 6 .
Figure 6.Biosynthesis of tetracycline using Sc-F 420.HPLC (a) and LC−MS (b) analysis of production tetracycline."SM' and 'SR" refer to the expression of OxyS_OxyM and OxyS_OxyR, respectively.(c) Mass spectrum of tetracycline products.The m/z of peaks 2 and 4 corresponds to tetracycline (exact mass: 444.435).Peaks 1 and 3 also show the mass that corresponds to tetracycline with slightly different m/z compared with peaks 2 and 4.This peak is also appearing in the standard tetracycline sample (b) as a minor peak.It is possible that tetracycline is epimerized in the acidic LC−MS analysis condition and eluted separately.33OxyS_OxyR expressing strain also produces oxytetracycline, which is shown in peak 1, representing the [M + H] + of 461.438.In contrast, OxyS_CtcM expressing strain did not produce any detectable amount of oxytetracycyline (peaks 3 and 4).

Table 1 .
Effect of Media on the F 420 Yield a a

Table 2 .
Incubation Time-Dependent F 420 Yield a a