An acetyltransferase controls the metabolic flux in rubromycin polyketide biosynthesis by direct modulation of redox tailoring enzymes

The often complex control of bacterial natural product biosynthesis typically involves global and pathway-specific transcriptional regulators of gene expression, which often limits the yield of bioactive compounds under laboratory conditions. However, little is known about regulation mechanisms on the enzymatic level. Here, we report a novel regulatory principle for natural products involving a dedicated acetyltransferase, which modifies a redox-tailoring enzyme and thereby enables pathway furcation and alternating pharmacophore assembly in rubromycin polyketide biosynthesis. The rubromycins such as griseorhodin (grh) A are complex bioactive aromatic polyketides from Actinobacteria with a hallmark bisbenzannulated [5,6]-spiroketal pharmacophore that is mainly installed by two flavoprotein monooxygenases. First, GrhO5 converts the advanced precursor collinone into the [6,6]-spiroketal containing dihydrolenticulone, before GrhO6 effectuates a ring contraction to afford the [5,6]-spiroketal. Our results show that pharmacophore assembly in addition involves the acetyl-CoA-dependent acetyltransferase GrhJ that activates GrhO6 to allow the rapid generation and release of its labile product, which is subsequently sequestered by ketoreductase GrhO10 and converted into a stable intermediate. Consequently, the biosynthesis is directed to the generation of canonical rubromycins, while the alternative spontaneous [5,6]-spiroketal hydrolysis to a ring-opened pathway product is thwarted. Presumably, this allows the bacteria to rapidly adjust the biosynthesis of functionally distinct secondary metabolites depending on nutrient and precursor (i.e. acetyl-CoA) availability. Our study thus illustrates how natural product biosynthesis can be enzymatically regulated and provides new perspectives for the improvement of in vitro enzyme activities and natural product titers via biotechnological approaches.


Cloning and recombinant production of GrhO10
To allow for the recombinant production of GrhO10, the respective gene was purchased from BioCat, codon optimized for Escherichia. coli and subcloned into the expression vector pET16b. Upon arrival, the plasmid DNA was dissolved in ddH2O to a final concentration of 100 ng µL -1 and transformed into E. coli BL21 (DE3) cells. For protein production, TB-medium supplemented with 100 µg mL -1 ampicillin was inoculated with a pre-culture grown in LB-medium containing an equal amount of antibiotic. Cultures were incubated at 37 °C and 130 rpm until and optical density at 600 nm (OD600) of ca. 0.5 was reached. Then, the temperature in the incubator was set to 18 °C and at an OD600 of 0.7-0.8 protein production was induced with 0.1 mM IPTG. To maximize the protein yield, cultures were incubated at 18 °C and 130 rpm overnight. The following day, cells were harvested by centrifugation (4 000 g for 15 min) and used for protein purification immediately.

Cloning and recombinant production HyalO10
To allow for the recombinant production of HyalO10, the respective gene was obtained from BioCat, codon optimized for Escherichia coli, flanked with NcoI (3') and NotI (5') restriction sites. After restriction digestion, the gene was cloned into the pETM11-His-TEV vector, in frame with the coding sequence for an N-terminal hexahistidine tag. Having confirmed the proper insertion of the gene into the vector by automated sequencing, the recombinant plasmid was transformed into E. coli BL21 (DE)cells for subsequent gene expression. For protein production, TB-medium supplemented with 50 µg mL -1 kanamycin was inoculated with a pre-culture grown in LB-medium containing an equal amount of antibiotic. Cultures were incubated at 37 °C and 130 rpm until and optical density at 600 nm (OD600) of ca. 0.5 was reached. Then, the temperature in the incubator was set to 18 °C and at an OD600 of 0.7-0.8 protein production was induced with 0.1 mM IPTG. To maximize the protein yield, cultures were incubated at 18 °C and 130 rpm overnight. The following day, cells were harvested by centrifugation (4 000 g for 15 min) and used for protein purification immediately.

Cloning and recombinant production MBP-GrhJ
To allow for the recombinant production of GrhJ as fusion protein with MBP (as his-tagged protein the enzyme is not soluble at all), the respective gene was amplified from isolated cosmid DNA using the following primers -fwd primer: 5ˈ -CCGGGCCATGGTGAGCCTCGAACTG -3; rev. primer: 5ˈ -CCGGTCCCTGCAGTCACGCCTTGACGG -3ˈ. After restriction digestion with NcoI (5') and PstI (3'), the gene was ligated into a pMAL-vector linearized with the same enzymes and proper insertion was confirmed by automated sequencing. For protein production, TB-medium containing ampicillin (100 µg mL -1 ) was inoculated with a pre-culture grown in LB-medium supplemented with the same amount of antibiotic. Cultures were incubated at 37 °C and 130 rpm until an OD600 of 0.6-0.7 was reached. Then, the temperature in the incubator was set to 18 °C and protein production was induced with 0.25 mM IPTG. To optimize the protein yield, cultures were incubated at 18 °C overnight, before harvesting the cells by centrifugation (4000 g for 15 min).

Cloning and recombinant production GB1-HyalJ
To allow for the recombinant production of GB1-tagged HyalJ, hyalJ was obtained from BioCat, codon optimized for E. coli and flanked with NcoI (5') and NotI (3') restriction sites. After restriction digestion, the gene was inserted into a pETM11-His-GB1-TEV vector, in frame with the coding sequences for an N-terminal hexahistidine tag and the solubility enhancer protein GB1 (B domain of protein G). Proper insertion of hyalJ was confirmed by automated sequencing and recombinant plasmids were transformed into E. coli BL21 (DE3) RP cells. For protein production, TB-medium supplemented with 50 µg mL -1 kanamycin and 20 µg mL -1 chloramphenicol was inoculated with a pre-culture grown in LB-medium containing an equal amount of antibiotic. Cultures were incubated at 37 °C and 130 rpm until and optical density at 600 nm (OD600) of ca. 0.5 was reached. Then, the temperature in the incubator was set to 18 °C and at an OD600 of 0.7-0.8 protein production was induced with 0.1 mM IPTG. To maximize the protein yield, cultures were incubated at 18 °C and 130 rpm overnight. The following day, cells were harvested by centrifugation (4 000 g for 15 min) and used for protein purification immediately.

Purification MBP-GrhJ
For purification of MBP-tagged GrhJ, cells were resuspended in a final volume of 50 mL 20 mM Tris, 400 mM NaCl, 1 mM EDTA, pH 8 (buffer A) + 1 mg mL -1 of lysozyme and 1 tablet of protease inhibitor.
The suspension was stirred on ice of 30 min, before lysing the cells by ultrasonication (3 cycles; 4 s pulse, 16 s pause, 1 min pulse time). Then, the lysate was cleared by centrifugation (18 000 g for 30 min) and loaded onto one 5 mL MBP-trap column pre-equilibrated with buffer A. Unspecifically bound proteins were removed by washing with 20 CV of buffer A and MBP-GrhJ was eluted with a gradient of buffer A and buffer B (buffer A + 10 mM maltose; 0% B to 100% B within 5 CV). MBP-GrhJ containing fractions identified by SDS-PAGE analysis were pooled, concentrated and supplemented with 10% glycerol (v/v). Finally, aliquots were prepared, flash frozen in lq. N2 and stored at -80 °C until further use.
Finally, aliquots were prepared, flash-frozen in liquid nitrogen and stored at -80 °C until further use.

Production and purification GrhO6
Production and purification of GrhO6 was carried out as described previously 1 .

Site-directed mutagenesiscloning of GrhO6 variants
In order to find out whether HyalJ acetylates one of GrhO6's Lys residues, three GrhO6 variants were generated by PCR-based mutagenesis. Mutations were introduced into the pETDuet-GrhO6 wild type construct with forward and reverse primers carrying the desired nucleotide replacements.

Isolation of 7 from Actinoplanes ianthinogenes
To obtain pure 7 as standard compound for comparison in all activity assays performed in this study, A.
ianthinogenes was cultured and 7 was isolated as described previously (see

Cultivation of S. albus pMP31 and analysis of the accumulating metabolites
Cultivation of the S. albus pMP31 strain was carried out as described previously 2 and metabolite analysis was carried out as described for the KR42 mutant strain below.

Cultivation of S. albus KR42 (grhJ) and analysis of accumulating compounds
To analyze the effect of knocking out the acetyltransferase GrhJ on griseorhodin A production in vivo, the corresponding S. albus strain (KR42) was cultivated in LB-medium supplemented with apramycin (50 µg µL) at 30 °C and 200 rpm for 2-3 days. Then, cells were harvested by centrifugation (4 000 g for 15 min) and the resulting pellet as well as the clear supernatant were acidified with HCl to pH 3 and extracted with EtOAc. Solvents were removed under reduced pressure and residues were redissolved in CH3CN or CH3CN:DMSO (1:1) and analyzed by semipreparative HPLC (NUCLEODUR 100-5 C18ec column; 250 x 10 mm ID, 5 µM, Macherey Nagel).

Spontaneous hydrolysis of 7,8-dideoxy-6-oxo-griseorhodin C (5a/b) under assay conditions (in the presence of DTT to generate reducing conditions)
To determine the stability of 5a/b under assay conditions, 50 µM 5b and NADPH (0.7 mM) were added to 50 mM Tris, pH 8 (+ 1 mM DTT in a second reaction) and incubated to 30 °C and 750 rpm. Samples were withdrawn after 0, 1, 2, 5, 10 and 30 min and quenched and extracted with EtOAc + 10% FA and the organic layers were analyzed by HPLC-DAD after 30 s of centrifugation at 13 300 g.

Time course of 7,8-dideoxy-6-oxo-griseorhodin C (5a/b) formation in the presence of HyalJ
To study 5(a)b formation in the absence and presence of HyalJ, a reaction cascade involving GrhO5 (20 µM), GrhO1 (10 µM), GrhO6 (10 µM), ca. 10 HyalJ µM + AcCoA (300 µM) and 3b (200 µM) as substrate was set up (control without HyalJ). Reactions were started by the addition of NADPH (3 mM) and incubated at 30 °C and 750 rpm. Samples were withdrawn from the assay mixture after 0, 0.5, 1, 2, 4, 7, 10, and 15 min and quenched with 2 eq (v/v) of EtOAc + 10% FA, each. The 0 min sample was centrifuged at 13 000 g for 30 s and the organic layer was analyzed by HPLC-DAD immediately. All other samples were flash-frozen in liquid nitrogen right after quenching and thawed and centrifuged only shortly prior to HPLC-DAD analysis to avoid the formation of undesired oxidation products.
Reactions were started by the addition NADPH (3 mM) and incubated at 30 °C and 750 rpm for 4 min.
After quenching with 2 eq (v/v) EtOAc + 10% FA, samples were centrifuged for 30 s and the organic layers were analyzed by HPLC-DAD immediately.

Isolation of AcCoA from HyalJ
To find out, whether HyalJ freshly purified from E. coli or stored at -80 °C for several weeks binds AcCoA, 150 µL of a 510/580 µM protein solution containing about 100 µM HyalJ were mixed with 100 µL EtOAc + 10% FA. The organic layer was removed and the aqueous phase was centrifuged at 18 000 g for 10 min. Finally, the clear water phase was analyzed by UPLC-HRMS.

Incubation of HyalJ with 7
Turnover test of 7 with HyalJ and AcCoA. For that, purified 7 was mixed with HyalJ (10 µM) and AcCoA (300 µM) in 50 mM Tris, pH 8 and incubated at 30 °C and 750 rpm for 4-6 min, followed by analytical HPLC as described further below. Under these conditions, no conversion of 7 was observed.

Assay O5/O1/O10 to produce 8 for UPLC-HRMS analysis
To produce compound 8, 20 µM GrhO5, 10 µM GrhO1, 45 µM HyalO10 and 200 µM 3b were mixed and incubated at 30 °C for 2 min. Then, NADPH (3 mM) was added, and reactions were incubated at 30 °C and 750 rpm for 10 min. Reactions were quenched and extracted by the addition of 2 eq (v/v) EtOAc + 10% FA and the organic layers were subjected to HPLC-DAD after 30 s of centrifugation at 13 300 g. The peak corresponding to 8 was collected manually and 8 was again extracted from the solvent mixture using EtOAc. Subsequently, the organic phase was transferred to fresh reaction tubes and concentrated in the speed-vac to complete dryness. The residue was redissolved an CH3CN and 8 was subjected to UPLC-HRMS analysis.

Inhibition assay with inhibitor 8
To test the effect of 8 on the activity of GrhO6, 8 and 4 were produced in two separate enzymatic assays.
Then, 50 µL of the 4a(b) mixture were combined with varying amounts of the 8 product solution (+50 mM Tris pH 8) and GrhO6 was added to a final concentration of 10 µM. Reactions were incubated at 30 °C and 750 rpm for 10 min, before quenching and extracting them with 200 µL of EtOAc + 10% FA. Samples were flash-frozen in liquid nitrogen and thawed and centrifuged (13 000 g for 30 s) shortly before analyzing the organic layers by HPLC-DAD.

Activity GrhO6 variants
Activity of the GrhO6-variants was tested both in the absence and presence of HyalJ. Therefore, 200 µM 3b, 20 µM GrhO5, 10 µM GrhO1 (+ 10 µM HyalJ) and 10 µM of a GrhO6-variant were mixed and incubated at 30 °C and 750 rpm for 2 min. Then, NADPH (3 mM) was added and reactions were incubated at 30 °C and 750 rpm for 4 min. Reactions were quenched and extracted with 2 eq (v/v) of EtOAc + 10% FA and flash-frozen in liquid nitrogen. Finally, samples were thawed and centrifuged (13,300 g for 30 s) shortly prior to analysis of the organic layers by HPLC-DAD.

Homology modeling
Homology models of GrhO10 and HyalJ were generated using the SWISS-MODEL server 3, 4 and the Phyre 2 server 5 , respectively. For GrhO10, the crystal structure of the C17/C19-ketoreductase ARX21 (PDB-ID, 5thq) was used as a template, as this protein exhibited the highest sequence identity and coverage (60% identity on amino acid sequence level and 100% coverage) of all crystallized proteins with GrhO10. For HyalJ, the crystal structure of a GCN5-related acetyltransferase from clavulanic acid biosynthesis (PDB-ID, 2wpw) sharing 23% sequence identity (95% coverage) with HyalJ was selected as template. Despite the not so high sequence similarity, a confidence of 100% was predicted for the model, which can be explained by the fact that GCN5-related acetyltransferases generally share high structural but low sequence similarity.

Multiple sequence alignment
Structure based multiple sequence alignments of ARX21 and several GrhO10-homologs were generated using the online server T-coffee 6 . The alignment results were exported in PHYLIP format and visualized in SeaView 7 . Final color editing was carried out in PowerPoint.

Analytical gel filtration general
To determine the biologically active oligomeric state of the proteins used in this study, analytical gelfiltration using a Superdex 200 GL 10/300 column pre-equilibrated with 50 mM Tris, 300 mM NaCl, 10% glycerol, pH 7.4 was used. Molecular weights of the proteins eluting in the main peak was calculated based on a previously generated calibration curve.
Samples were analyzed in MS positive mode with a capillary voltage of 3.0 kV, 100 °C source temperature, 300 °C desolvation gas temperature and 600 L h −1 N2 desolvation gas flow and a capillary voltage of 1.5 kV, 120 °C source temperature, 300 °C desolvation gas temperature and 500 L h −1 N2 desolvation gas flow for MS negative mode.  Analytical size-exclusion was performed for both proteins using a Superdex 200 GL 10/300 column pre-equilibrated with 50 mM Tris, 300 mM NaCl, 10% glycerol, pH 7.4. GrhO10 eluted as a clean single peak at around 15 mL, indicating that it forms dimers in solution. For HyalO10 also a peak corresponding to dimeric enzyme was detected at 15 mL, however, also a major aggregate peak around 8-8.5 min was observed. To test the activity of the KR, a reaction cascade with 3b, NADPH, GrhO5, GrhO1, GrhO6 and the KR (control without KR) was used. Interestingly, in either case, 6(a)b (6b standard, black line; highlighted in grey) formation could not be observed. Both in the reaction with as well as without the KR, 7 (highlighted in light red) was the main accumulating metabolite. The peak at 12 min in the blue trace is inhibitor 8. Figure S4: Time-dependent spontaneous hydrolysis of 5b under assay/reducing conditions. 5b was dissolved in DMSO and added to 50 mM Tris, 300 mM NaCl, 10% glycerol (50 µM) additionally containing 0.7 mM NADPH and 1 mM DTT to generate reducing conditions. Reactions were incubated at 30 °C and 750 rpm and samples were withdrawn after 0, 1, 2, 5, 10 and 30 min and analyzed by HPLC-DAD. As indicated by the time course graph, 5b (dark red) is slowly converted into 7 (light red).

Figure S6: Overlay of the crystal structure of the C17/C19 ketoreductase ARX21 (grey) with a homology model generated for GrhO10 (blue).
Comparison of the homology model of GrhO10 (blue) with the crystal structure of ARX21 (grey; PDB-ID, 5thq) reveals a virtually identical overall fold as well as conserved active site residues. Both the catalytic triad proposed for ARX21 consisting of Y144, Y157, and K161, as well as the binding residues (R154 and Y210) are found in GrhO10 as well (Y142, Y155 and K159 and R154 and Y208, respectively). All proposed key residues are shown as sticks (ARX21, grey; GrhO10, blue). The NADP + observed in the crystal structure of ARX21 is displayed in orange.         To test the effect of acetyl-CoA on HyalJ activity, turnover assays with 3b, NADPH, GrhO5, GrhO1, GrhO6 and HyalJ were performed in the absence and presence of free AcCoA. In line with its proposed function, the addition of acetyl-CoA strongly increased the enzymatic activity of HyalJ resulting in elevated 5a/b-levels (highlighted in grey) in this sample. -HyalJ / + AcCoA n.d. Figure S16: Effect of acetyl-CoA (AcCoA) on 5a/b formation. To test the effect of AcCoA on HyalJ activity, turnover assays with 3b, NADPH, GrhO5, GrhO1, GrhO6 and HyalJ were performed in the absence and presence of free AcCoA and analyzed by HPLC-DAD. While hardly any influence of free acetyl-CoA on 5b formation could be observed with freshly purified HyalJ (A), this effect became rather pronounced when then protein had already gone through several freezing-thawing cycles ("stored HyalJ"; B).
Figure S17: HyalJ is produced and purified as holo-enzyme with acetyl-CoA bound in the active site. To find out, whether HyalJ is co-purified with acetyl-CoA, freshly purified HyalJ was sacrificed with EtOAc + 10% FA. After removal of the organic phase, the water phase was centrifuged to remove protein particles and the aqueous layer was analyzed by UPLC-HRMS. In line with the enzymatic assays, acetyl-CoA could be detected in the water phase, confirming that HyalJ is produced and purified as holo-enzyme. A, Extracted ion chromatograms (positive ion mode) corresponding to acetyl-CoA ([M+H] + , 810.135 m/z) detected in the sample (top) as well as in a pure standard (bottom). B, Mass spectra corresponding to the peaks shown in panel A. As indicated, besides the mother ion ([M+H] + , 810.135 m/z), Na + -adducts were also detected. Figure S18: Acetyl-CoA content in fresh and stored HyalJ. To confirm that HyalJ "inactivation" upon longterm storage/after several freezing-thawing cycles results from acetyl-CoA degradation, "poorly active" HyalJ was sacrificed with EtOAc + 10% FA. After removal of the organic phase, the water phase was centrifuged to remove protein particles and the aqueous layer was analyzed by UPLC-HRMS. In line with the enzymatic assays, in contrast to the freshly purified enzyme (A) no acetyl-CoA could be detected in the water phase of the stored enzyme (B), confirming that HyalJ "inactivation" is the result of acetyl-CoA degradation. This is also in line with the fact that addition of free acetyl-CoA to "inactivated" HyalJ can restore its enzymatic activity. Figure S19: CoA derivatives identified to be co-purified with HyalJ. To find out, whether HyalJ is co-purified with acetyl-CoA or any other CoA-derivative, freshly purified HyalJ was sacrificed with EtOAc + 10% FA. After removal of the organic phase, the water phase was centrifuged to remove protein particles and the aqueous layer was analyzed by UPLC-HRMS. Interestingly, not only acetyl-CoA but also propionyl-CoA, malonyl-CoA and methylmalonyl-CoA could be detected. While malonyl-CoA was only present in small amounts, the signal intensity for propionyl-CoA and methylmalonyl-CoA was in the same range as for acetyl- CoA               Red line, To test the requirement of HyalJ for 8 formation, a reaction mixture with 3, NADPH, GrhO5, GrhO1, GrhO6 and HyalO10 was prepared and incubated at 30 °C and 750 rpm for 10 min. The reaction was quenched and extracted with EtOAc + 10% FA and the organic layer was analyzed by HPLC-DAD. Since even higher amounts of 8 were produced under these reaction conditions, compared to the original set up, HyalJ is not involved in 8 formation. Blue line, To further analyze the need of GrhO6 for 8 production, the same assay was repeated, however without the addition of GrhO6. In this case, 8 became the major reaction product, indicating that GrhO6 and HyalO10 act on the same substrate -8 is therefore generated from 4a(b). Figure S34: Conversion of 8 by GrhO6. To find out, whether GrhO6 can convert 8 (highlighted in yellow), GrhO6 was mixed with in situ produced 8 and the reactions were incubated at 30 °C and 750 rpm for 10 min. After quenching and extracting with EtOAc + 10% FA, the organic layers were analyzed by HPLC-DAD. As reflected by the chromatogram highlighted in blue, GrhO6 is not capable of converting 8 (the slightly different shape of the peak(s) corresponding to 8 probably is the result of tautomerism as the ratio of the two peaks both corresponding to 8 was different each time the assay was repeated).