Filling the Gaps in the Kirromycin Biosynthesis: Deciphering the Role of Genes Involved in Ethylmalonyl-CoA Supply and Tailoring Reactions

Kirromycin is the main product of the soil-dwelling Streptomyces collinus Tü 365. The elucidation of the biosynthetic pathway revealed that the antibiotic is synthesised via a unique combination of trans-/cis-AT type I polyketide synthases and non-ribosomal peptide synthetases (PKS I/NRPS). This was the first example of an assembly line integrating the three biosynthetic principles in one pathway. However, information about other enzymes involved in kirromycin biosynthesis remained scarce. In this study, genes encoding tailoring enzymes KirM, KirHVI, KirOI, and KirOII, and the putative crotonyl-CoA reductase/carboxylase KirN were deleted, complemented, and the emerged products analysed by HPLC-HRMS and MS/MS. Derivatives were identified in mutants ΔkirM, ΔkirHVI, ΔkirOI, and ΔkirOII. The products of ΔkirOI, ΔkirOII, and kirHVI were subjected to 2D-NMR for structure elucidation. Our results enabled functional assignment of those enzymes, demonstrating their involvement in kirromycin tailoring. In the ΔkirN mutant, the production of kirromycin was significantly decreased. The obtained data enabled us to clarify the putative roles of the studied enzymes, ultimately allowing us to fill many of the missing gaps in the biosynthesis of the complex antibiotic. Furthermore, this collection of mutants can serve as a toolbox for generation of new kirromycins.

kirromycin indicated that the biosynthetic steps involve non-ribosomal peptide synthetases (NRPSs), acyltransferase (AT)-harbouring PKSs (cis-AT PKSs), AT-less PKSs (trans-AT PKSs), and several tailoring enzymes. This first description of a biosynthetic pathway involving three different types of multi-modular enzymes has made the assembly line a relevant model for studying complex cis-/trans-AT PKS/NRPS pathways.
Directly upstream of the PKS/NRPS genes in the BGC of kirromycin, the Sfp-type phosphopantetheinyl transferase (PPTase) KirP, responsible for the post-translational activation of PKS acyl carrier proteins (ACPs) and NRPS peptidyl carrier proteins (PCPs), was identified 7 . The core PKS/NRPS enzymes in kirromycin are encoded by kirAI -kirAVI and kirB 4 . Closer examinations of the PKS-related genes and their products revealed KirAVI to be the only cis-AT PKS in the assembly line. The upstream PKSs, KirAI -KirAV, lack the integrated ATs. For these modules, extender units are provided by the discrete ATs KirCI 8,9 and KirCII 8,10 , which provide malonyl-CoA and ethylmalonyl-CoA to the ACPs of the assembly line. The last core enzyme of the kirromycin PKS/NRPS assembly line is encoded by kirB. The protein KirB is responsible for incorporation of the precursor β-alanine 4 , which is biosynthesised by the aspartate-α-decarboxylase encoded by kirD 11 . Upon release from the assembly line, this precursor is cyclised to form the pyridone ring 4 . Interestingly, no genes encoding a typical thioesterase (TE) were found in the kirromycin biosynthetic gene cluster (BGC). Recently, Gui et al. 12 reported the discovery of a new group of Dieckmann cyclases involved in the biosynthesis of tetramic acid and pyridone scaffolds. A member of this new family of enzymes was also identified in the kirromycin BGC. KirHI, previously assigned with a hypothetical function, was found to catalyse pyridone ring closure and the co-occurring release of the molecule chain from the PKS I/NRPS assembly line 12 .
While the precursor loading of the PKS I/NRPS modules and the kirromycin polyketide chain elongation are well investigated, only limited knowledge about the provision of ethylmalonyl-CoA and the final tailoring reactions is available for this pathway. To date, the underlying reactions leading to the THF ring closure, the introduction of hydroxyl groups at C-16 and C-30, methylation of the oxygen at O-42, and the formation of the double bond between C-5 and C-6 in the pyridone ring have remained elusive. These enzymatic reactions cannot be explained from the enzymatic domains encoded within the PKS I/NRPS complex, and thus it was speculated that those product modifications are achieved through tailoring reactions.
In this study, the genes kirM, kirHVI, kirOI, kirOII, kirHIV, kirHV, and kirN, which are encoded in the kirromycin BGC but have no experimentally determined function, were examined by genetic inactivation and gene complementation. The compounds produced by the gene inactivation mutants were analysed and characterised by high performance liquid chromatography (HPLC)-high-resolution mass spectrometry (HRMS), and MS/ MS. In addition, nuclear magnetic resonance (NMR) was employed for structural elucidation of the derivative produced by the ΔkirHVI, ΔkirOI, and ΔkirOII mutants.

Methods
Bacterial strains and general growth conditions. Plasmids and strains used in this study are listed in Supplementary Table S1. Wild type strain Streptomyces collinus Tü 365 was obtained from the "Tübinger Stammsammlung". For routine cultivations of mutants, complemented mutants, and wild type strain, Tryptic Soy Broth (TSB) supplemented with nalidixic acid (25 μg/mL) was used. Furthermore, complemented mutants were propagated in the presence of apramycin (50 μg/mL). Sporulation of wild type and mutant strains was performed on modified SFM agar (2% mannitol, 2% full fat soy flour, 20 mM MgCl 2 , 20 mM CaCl 2 , tap water). Escherichia coli DH5-α was used for standard cloning procedure of pGUSA21, pDrive, pA18mob, pJet1.2/blunt, pGM1190, and pRM4 constructs. Transformation was carried out according to the manufacturer's specifications and including a 1 h recovery step in Super Optimal broth with Catabolite repression (S.O.C.) medium. E. coli was grown at 37 °C (200 rpm) in Lysogeny broth (LB) medium supplemented with appropriate antibiotics. In order to introduce the construct into streptomycetes, plasmid DNA was introduced into E. coli ET12567 (pUZ8002) by calcium chloride transformation. ET12567 strains were grown in LB supplemented with apramycin (50 μg/mL), chloramphenicol (25 μg/mL), and kanamycin (25 μg/mL) for 16 hours, shaking at 200 rpm at 37 °C.
Construction of kirM, kirN, kirHIV, and kirHV gene inactivation plasmids. All primers used in this study are listed in Supplementary Table S2. A 2.1 kb cassette, containing the ermE* promoter flanked by two 1 kb-fragments up-and downstream of the genes kirM, kirHIV, kirHV, and kirN, were amplified from pCRISPR-Cas9_USER-kirM/kirHIV/kirHV/kirN, which were constructed based on a previously described method 13 (see Supplementary Table S1). For the amplifications, primers KP1 -KP8 and PCR programs listed in Supplementary Table S2 were used. Each of the kir cassettes were assembled in pGUSA21 14 by Gibson cloning carried out at 50 °C for 30 min. Correct clones of pHR1 and pHR3-5 were identified by control PCR and confirmed by Sanger sequencing with KP37/KP42.

Construction of kirHVI gene inactivation plasmid.
To generate the kirHVI gene inactivation plasmid pHR2, 1 kb fragments, flanking the target gene, were amplified from the cosmid 2K05 DNA 4 using primers KP9/ KP10 and KP13/KP14. The ermE* promoter was amplified from plasmid pCRISPR-Cas9 15 with primers KP11/ KP12. 1 M betaine was added to the PCRs to improve amplification of GC-rich stretches. Upon gel purification, the three fragments were assembled in pGUSA21 14 by Gibson cloning carried out at 50 °C for 30 min, yielding pHR2. Correct clones of pHR2 were identified by control PCR and Sanger sequencing with KP37/KP38.

Construction of kirOI and kirOII gene inactivation plasmids.
To construct pTL-kirOI, 1.2 kb of the flanking regions of the gene were amplified by PCR with primer pairs KP15/KP16 and KP17/KP18 using the PCR programs listed in Supplementary Table S2. For construction of pDW-kirOII, 2 kb of the flanking regions of kirOII were amplified by PCR with primers KP21/KP22 and KP23/KP24 using the PCR program listed in the Supplementary Table S2.
The left and right fragments, up-and downstream of kirOI and kirOII, were cloned into pA18mob resulting in the plasmids pTL-kirOI and pDW-kirOII, respectively. The thiostrepton resistance cassette (1.1 kb) was amplified from plasmid pSLE61 16 with primers KP19/KP20. The 1.1 kb fragment was first cloned into pDrive (Qiagen, Hilden, Germany) and then inserted via the XbaI sites into pTL-kirOI and pDW-kirOII, resulting in the plasmids pTL-kirOI-thio and pDW-kirOII-thio, respectively. The gene replacement mutants were confirmed by control PCRs targeting the apramycin and thiostrepton resistance cassettes with the primer pairs KP39/KP40 and KP19/ KP20, respectively.
To distinguish between wild type and single/double crossover mutants, the primer pairs KP41/KP42 and KP43/KP44 were used for amplification of an internal region of (wild type) kirOI/kirOII or the thiostrepton resistance marker (single-or double crossover mutant).
In addition to PCR, Southern Blot analysis was used to confirm correct mutants of kirOI and kirOII. Here, the DIG-labelled probe for Southern Blot experiments was amplified from cosmid 1C24 DNA 4 with the primer pairs KP41/KP42 (kirOI) or KP43/KP44 (kirOII) using a 10 × DIG DNA labelling mix (Roche, Mannheim, Germany). Genomic DNA preparation from streptomycetes was performed with the NucleoSpin ® Tissue Kit from Macherey-Nagel, Düren, Germany.
Intergeneric conjugation in S. collinus Tü 365. To introduce the gene inactivation plasmids into S. collinus Tü 365, a standard protocol for intergeneric conjugation was used 17 . Single crossover mutants were selected based on blue-white screening 14 in which spore dilution plates were overlaid with 20 mM X-Gluc (5-bromo-4-chloro-3-indolyl-β-D-glucuronic acid). Control PCR was used to verify the single crossovers using primers KP72/KP73. For induction of the double crossover event, two-day-old cultures of single crossover mutants, cultivated in TSB medium, were stressed for 24 h at 37 °C, shaking at 160 rpm. Spore dilution plates were overlaid with 20 mM X-Gluc and white clones were picked for control PCRs with primers KP45 -KP58 (see Supplementary  Fig. S6).
Construction of complementation plasmids for kirromycin kir mutants. kir genes were PCR amplified using the cosmid 2K05 (kirM, kirHVI, kirOI, kirOII, and kirN) 4 as template and the primers KP27 -KP36. 1 M betaine was added to the PCRs to improve amplification of the GC-rich stretches. The DNA encoding the kir genes was purified from gel and cloned into the blunt end pJet1.2 vector as instructed by the manufacturer. Correct clones were verified by control PCR and Sanger sequencing with primers KP59/KP60. pJet1.2 clones of kirM, kirHVI, and kirN were subjected to restriction digestions with NdeI and HindIII together with DNA of integrative plasmid pRM4 18 . The pJet1.2 clone of kirOII was subjected to restriction digestion with NheI and HindIII together with pRM4. Finally, pJet1.2-kirOI was subjected to restriction digestion with HindIII and EcoRI together with replicative plasmid pGM1190. The genes purified from gel and linearised Kirromycin production assay, HPLC-HRMS, and MS/MS analyses. For the fermentations, 5 mL of two days-old precultures (grown in TSB) were used for inoculation of 95 mL kirromycin production medium composed of 1% full-fat soy flour, 1% D-mannitol, and 0.5% CaCO 3 dissolved in tap water and pH adjusted to 7.4 prior to autoclaving. Fermentations were carried out for six days at 30 °C in a rotary shaker at 160 rpm.
Cultures were extracted with 1:1 ethyl acetate for 90 minutes and evaporated in a Büchi ® Rotavapor ® RII evaporator equipped with jack and water bath. Dried extracts were redissolved in 500 μL methanol. HPLC-HRMS analysis was carried out with an Orbitrap Fusion connected to a Dionex Ultimate 3000 UHPLC pumping system (ThermoFisher Scientific, Waltham, MA, USA). UV-Vis detection was done using a DAD-3000 set to the range 200-600 nm. Samples were kept at 10.0 °C in the autosampler during the analysis. 2 μL of each sample was injected into a C18 Acquity UPLC F5-3 HPLC column (2.1 × 100 mm, 1.8 μm,) at a flow rate of 0.4 mL/min, 30.0 °C. Mobile phases A and B were 0.1% formic acid in water and acetonitrile, respectively. Elution was done with a 30 min multistep system. After 5% B for 1 min, a linear gradient started from 5% B to 100% B in 21 min, which was held for another 5 min and followed by re-equilibration to 5% B until 30 min. Data was collected in both positive and negative ion modes with a scan range of (m/z) = 200-2000. MS/MS fragmentation was carried out using Ion Trap for collision-induced dissociation (CID) with the collision energy (30%), RF Lens 60%, AGC target 5.0e4, and scan range (m/z) 230-800. Data analyses were performed with the software Xcalibur 3.0.63 (Thermo Fisher Scientific Inc.).

Results
Sequence analysis of genes and encoded products presumably involved in kirromycin biosynthesis in S. collinus Tü 365. The genes kirM, kirHVI, kirOI, kirOII, kirHIV, kirHV, and kirN are located in the kirromycin gene cluster and thus a potential function of their gene products was postulated based on sequence analysis using the sequence aligner DIAMOND 19 against the MIBiG database (Table 1).
For the tailoring enzymes, the closest homologue of KirM was the O-methyltransferase RapM (75% aa identity and 87% similarity), involved in rapamycin biosynthesis in Streptomyces hygroscopicus NRRL 5491 20 . With 31% aa identity and 55% similarity, the hydroxylase Fum3p involved in C-5 hydroxylation in fumonisin biosynthesis in Gibberella fujikuroi 21 was the best hit for KirHVI. When the protein sequence of KirHVI was aligned against the Uniprot/Swissprot database, most of the hits were members of the phytanoyl-CoA dioxygenases enzyme family. KirOI and KirOII were putatively assigned to cytochrome P450 hydroxylases. KirOI displayed 49% aa identity and 64% similarity to the previously characterised cytochrome P450 hydroxylase TiaP2, responsible for hydroxylation of the C-20 residue in tiacumicin B biosynthesis 22 . For KirOII, the closest characterised homologue (40% aa identity and 59% similarity) was found to be the 6-deoxyerythronolide B hydroxylase, encoded by eryF, which is involved in biosynthesis of erythromycin in Saccharopolyspora erythraea 23 . Additional proteins downstream of the PKS/NRPS biosynthetic core have remained assigned with hypothetical function as it is the case for KirHIV and KirHV. The sequence analyses of KirHIV and KirHV, using DIAMOND 19 alignment against the MIBiG database, did not yield any significant hits. Furthermore, based on the alignment against the UniProtKB/Swiss-Prot database, KirHIV and KirHV displayed similarity only to proteins with uncharacterised functions. Here, the best match for KirHIV (76% aa identity and 88% similarity) was an uncharacterised protein found in Frankia sp. BMG5.23. KirHV displayed the highest aa identity and similarity (71 and 80%) to an uncharacterised protein from Streptomyces sp. CB00455. Hence, in respect of the sequence analyses the functions of KirHIV and KirHV remained elusive.
The gene kirN is located downstream of the PKS I/NRPS-encoding region in the BGC and shows 76% aa identity and 86% similarity to the crotonyl-CoA reductase/carboxylase (CCR) SfaR from the sanglifehrin BGC in Streptomyces flaveolus DSM 9954 24 (Table 1). In addition to this, the sequence analysis revealed that kirN displays between 70 to 76% aa identity with several other CCRs found in other Streptomyces species.
Gene inactivations of kirM, kirHVI, kirOI, and kirOII result in production of kirromycin derivatives. To determine the role of the six genes presumed to be involved in tailoring reactions, two different gene replacement strategies were carried out. For the mutants in kirM, kirHVI, kirHIV, and kirHV, the respective gene was replaced by the ermE* promoter, whereas for kirOI and kirOII, a thiostrepton resistance cassette together with the ermE* promoter were used to replace the genes. Mutants were verified by control PCRs (see Supplementary  Fig. S6), which resulted in either the expected bands or no product when the primers targeted internal regions of the deleted gene. The confirmed mutant clones were subjected to fermentation experiments carried out in parallel batches with wild type S. collinus Tü 365 as reference strain. The cultures were extracted with ethyl acetate, concentrated, and the extracts, including kirromycin and its derivatives, were dissolved in methanol. The samples were analysed by HPLC-HRMS.
In the 30 min chromatographic method used for separation of kirromycin (1) and its derivatives, 1 appeared at 12.2 min with the main ion m/z = 795.4110 [M-H] − (Fig. 2). Furthermore, MS/MS fragmentation of the derivatives was carried out to gain insights into differences in the chemical formulas (see Supplementary Fig. S2). Kirromycin derivatives were detected in the mutants of ΔkirM, ΔkirHVI, ΔkirOI, and ΔkirOII (see Fig. 2 and Supplementary Table S3). Gene inactivations of the hypothetical proteins encoded by kirHIV and kirHV did not affect the production of kirromycin and resulted in HPLC-HRMS data similar to the wild type (see Supplementary Fig. S1-I and S1-II). Thus, it was concluded that these proteins are not directly involved in kirromycin biosynthesis.
Gene inactivation of kirM resulted in the complete loss of production of 1, and instead the derivative 20-O-demethyl-kirromycin (kirromycin-ΔkirM, 2), which was only present in low amounts in the wild type sample, was the main product of the mutant. With its chemical formula C 42 H 58 N 2 O 12 , 2 appeared in the chromatogram at 11. Similar to the ∆kirM mutant, the extracts of ∆kirHVI, ∆kirOI, and ∆kirOII were analysed using HPLC-HRMS and HRMS/MS. The chromatograms of these mutants revealed new prominent peaks, which were only detected with weak intensities in the extract of the wild type S. collinus Tü 365. Gene inactivation of kirHVI  Supplementary Fig. S2-I and S2-III). This mass corresponds to a fragmentation at the peptide bond at C-27 25 . As this pyridone-containing fragment is observed in both 1 and 3 this indicated that the hydroxyl group is missing on the sugar-containing fragment, presumably at position C-30, of 3 in the ΔkirHVI mutant. 1 H and 13 C NMR and Homonuclear Correlation Spectroscopy (COSY), Heteronuclear Single Quantum Correlation (HSQC), and Heteronuclear Multiple Bond Correlation (HMBC) corroborated the putative proposed structure of 3 as the analogue 30-deoxy-kirromycin (see Supplementary Table S4, Supplementary Fig. S4-I and Fig. S5-I).
In the case of KirOII, the difference of two oxygen atoms underlined the hypothesis of the enzyme being involved in oxidation at C-16 and presumably preparing the ring closure to the THF ring. However, the MS/MS fragmentation of 5, yielding the main fragment ion of m/z = 467.2548 [M-H] − , did not allow for prediction of the exact chemical structure of the derivative (see Supplementary Fig. S2-V). Therefore, NMR analyses, including 1 H NMR, 13 C NMR, COSY, HSQC, and HMBC were undertaken to gain more insight into the structure of 5 (see Supplementary Table S4, Supplementary Fig. S4-III, and Fig. S5-III). To elucidate the structure of 5, a 2 L-fermentation of the ΔkirOII mutant strain was carried out. Pure compound 5 (2.2 mg) was obtained by separation on a semi-preparative RP-18 column by HPLC. The 13 C NMR spectrum showed addition signals for double bonds and the loss of signals for three oxygen-bearing carbons indicating opening of the furan ring and corroborated the chemical nomenclature 30-hydroxy-5,6-dehydro-1-N-demethyl-16-deoxy-kirrothricin. 2D-NMR experiments revealed the tetraene fragment for 5, which is not present in 1 (see Supplementary Table S4 and Supplementary Fig. S4-III and S5-III).
To rule out polar effects caused by disruption of the kirromycin BGC, gene complementations for each of the mutants, ΔkirM, ΔkirHVI, and ΔkirOII, were carried out using conjugation and the integrative plasmid pRM4 18 , and for ΔkirOI using the replicative plasmid pGM1190 (see Supplementary Fig. S3-I-S3-IV). The complemented mutants were fermented in kirromycin production medium, extracted, and the extracts analysed by HPLC-HRMS. The complementation of the ΔkirM mutant resulted in similar kirromycin production levels compared to the wild type (100%). For the complementation of the ΔkirHVI mutant, approximately 30% of the wild type levels of 1 could be restored, which was confirmed by both MS and UV-Vis quantification. The complementation of the ΔkirOI and ΔkirOII mutants resulted in only 10% and 20% restored production of 1 compared to the wild type levels. Although there was no full complementation for these two mutants, the retention times and molecular masses, as determined by HPLC-HRMS, clearly indicated that the observed effects were due to the inactivation of the individual genes and not due to polar effects on the biosynthetic assembly line.
Gene inactivation of kirN gives rise to lowered kirromycin production. The ΔkirN mutant was constructed based on the previously described GusA system for double crossover 14 . An ermE* promoter was used to replace the gene. The integrative plasmid pRM4 was used to construct the complemented mutant. The gene kirN encodes a putative CCR, which is postulated to provide the ethylmalonyl-CoA extender unit for the trans-AT KirCII, which then loads this substrate onto KirAII-ACP5. Unexpectedly, a deletion of kirN (ΔkirN) resulted in loss of kirromycin production. Instead, derivative 3, which was the main compound observed in the ΔkirHVI mutant, was detected. Genetic complementation of the ΔkirN mutant failed to restore production of 1. However, upon gene complementation of the ΔkirN mutant with pRM4 harbouring kirHVI, levels of 1 were partially restored (~30% of wild type levels). Since the replacement of kirN with the ermE* promoter was expected to drive expression through the downstream gene kirHVI, we examined the genetic organisation. This revealed overlapping open reading frames (ORFs) of kirN and kirHVI and it was concluded that the genetic inactivation of kirN led to the disruption of downstream gene kirHVI. To test if production of 1 could be fully restored, pRM4 harbouring a copy of both kirN and kirHVI was constructed and used for complementation of ΔkirN. This resulted in a restoration of 1 in the ΔkirN mutant, reaching between 40-100% of wild type levels (see Supplementary Fig.  S3-V and S3-VI). Taken together, this data confirmed that the downstream gene kirHVI was affected by the gene inactivation of kirN and a full recovery of the production of 1 was only possible when kirN and its neighbouring gene kirHVI were encoded on the complementation construct.

Discussion
In this study, the involvement of the seven genes, kirM, kirHVI, kirOI, kirOII, kirHIV, kirHV, and kirN in the biosynthesis of kirromycin in S. collinus Tü 365 was investigated experimentally based on gene inactivations, complementations, and product analyses by HPLC-HRMS, MS/MS, and NMR. Our studies provide experimental evidence that KirM acts as an O-methyltransferase, giving rise to the methyl group on O-42 in kirromycin. This finding is further strengthened by comparison of the role of the close homologue RapM, which is involved in rapamycin biosynthesis in Streptomyces hygroscopicus NRRL 5491. Gene complementation with rapM of the deletion mutant MG2-10 (ΔrapKIJMNOQL) resulted in the restored O-methylation at the C-16 position in this molecule 20 .
Based on the MS/MS fragmentation (see Supplementary Fig. S2-III) and NMR analysis (see Supplementary  Table S4, Supplementary Fig. S4-I and Fig. S5-I), the main derivative 3, produced by the ΔkirHVI mutant, lacks a hydroxyl group at position C-30 in the kirromycin analogue. From the sequence-based alignment against the MIBiG database, KirHVI displays similarity to the hydroxylase Fum3p, which is responsible for attachment of a hydroxyl group at the C-5 position in fumonisin B produced by Gibberella fujikuroi 21 . Heterologous expression of Fum3p in Saccharomyces cerevisiae could clarify the enzymatic function of the protein and based on in vitro and in vivo findings it was possible to determine Fum3p to be a 2-ketoglutarate-dependent dioxygenase 26 . Since most protein hits from the UniProtKB/Swiss-Prot sequence alignment of KirHVI belonged to the family of phytanoyl-CoA dioxygenases it can be postulated that the hydroxylation catalysed by KirHVI in kirromycin biosynthesis occurs in a similar fashion (see Supplementary Fig. S7-I).
Furthermore, the function of KirOI and KirOII has been analysed in this study. According to the data from the MS/MS fragmentation and NMR, the ΔkirOI mutant produced the derivative kirromycin-ΔkirOI, which was undoubtedly missing the double bond between C-5 and C-6 in the pyridone ring (see Supplementary Table S4, Supplementary Fig. S2-IV, Fig. S4-II, and Fig. S5-II). Here, the predicted cytochrome P450 hydroxylase KirOI most likely forms a hydroxylated intermediate, which then undergoes dehydrogenation giving rise to the fully unsaturated pyridone ring (see Supplementary Fig. S7-II).
The HRMS and MS/MS analyses of the derivative 5, produced by the ΔkirOII mutant, did not provide sufficient data for unambiguous structure prediction. Therefore, 5 was subjected to NMR studies, which enabled us to uncover a kirromycin analogue missing the THF ring structure and both hydroxyl groups at positions C-15 and C-16 (see Supplementary Table S4, Supplementary Fig. S4-III, and Fig. S4-III) in the kirromycin analogue. Based on our current hypothesis, the biosynthesis of kirromycin from 5 might be explained by a mechanism similar to that of the cytochrome P450 monooxygenase AurH, which is involved in the conversion of deoxyaureothin to aureothin through the formation of two C-O bonds, ultimately giving rise to a THF ring 27,28 . Similar to AurH, KirOII could introduce two hydroxyl groups at C-14 and C-16. Allylic substitution of the hydroxyl system could then lead to the formation of the THF ring system at the expense of H 2 O. The proposed enzymatic pathway is presented in Supplementary Fig. S7-III. In the future, enzymatic in vitro assays are necessary to fully establish the enzymatic reaction catalysed by KirOII.
Finally, we have studied the putative CCR encoded by kirN, which is expected to be involved in biosynthesis of the ethylmalonyl-CoA extender unit incorporated in kirromycin at the C-28 position 4 . Studies on the involvement of CCRs in both primary and secondary metabolism have led to a revision of their function and today these enzymes are known for their involvement in acetyl-CoA assimilation through an alternative pathway distinct from the glyoxylate pathway 29,30 . Furthermore, sequence analysis of actinomycetes has revealed a tendency of co-clustering of CCRs with PKSs. This suggests an important role of the CCRs in polyketide derivatization as they provide extender units alternative to butyrate, which are incorporated into the polyketide structure 30,31 . The putative role of KirN is further strengthened by the 13 C NMR studies on the N-methylated kirromycin analogue aurodox, in which Liu et al. 32,33 confirmed the incorporation of an intact butyrate unit at the C-28 position in the molecule. We suspect that the presence of an additional ccr gene in the genome of S. collinus Tü 365 (locus B446_29770 in NCBI GenBank entry CP006259.1), likely involved in the ethylmalonyl pathway to assimilate C2-units in primary metabolism 29 , may give rise to an inherent complementation of ΔkirN. Therefore, complete abolishment of kirromycin production was not expected. Surprisingly, the ΔkirN mutant generated in this study gave rise to production of 3 instead of the expected lowered production of 1. Examination of the genetic organisation of the kirromycin BGC revealed the ORF of kirN to overlap with that of kirHVI, hence leading to disruption of the latter gene and production of 3. Partial and full restoration of kirromycin production was achieved by genetic complementation of ΔkirN with kirHVI separately and with kirN and kirHVI together, respectively (see Supplementary Fig. S3-V and S3-VI). The observed phenomenon should be kept in mind for future BGC engineering efforts to account for polar effects in these often tightly regulated systems.
In our study, we demonstrated that the enzymes KirM, KirHVI, KirOI, and KirOII play important roles in late stages of the biosynthesis of kirromycin. Derivatives of 1 were detected in all mutants, except for ΔkirN, ΔkirHIV, and ΔkirHV. Although the genes kirHIV and kirHV are located in the kirromycin BGC, their inactivation had no effect on kirromycin biosynthesis and resulted in similar production profiles in both the wild type and mutant clones. Consequently, the data suggest that these hypothetical proteins are not directly involved in kirromycin biosynthesis. For the ΔkirN mutant, we observed a reduced production of 1. While the ΔkirM mutant produces a derivative missing a methyl group at the O-42 position, the main derivative produced by the ΔkirHVI mutant lacks the hydroxyl group at the C-30 position in the sugar-like moiety. Finally, gene inactivations of kirOI and kirOII, encoding the two cytochrome P450 hydroxylases, result in a missing double bond in the pyridone ring and no THF ring closure, respectively. These results and the new knowledge have allowed us to close some of the gaps in the biosynthetic pathway of kirromycin (Fig. 3). Our studies does not allow for determining a defined sequence of the tailoring reactions. However, based on our findings, it could be argued that the enzymes act in an independent manner.
The better understanding of the complex kirromycin assembly line, which can be regarded as model for hybrid cis-/trans-AT PKS I/NRPSs, provides new incentive to the development of novel strategies for the production of polyketide derivatives. In particular the ΔkirN mutant, generated and analysed in this study, will support the applicability of a previously developed system 25 . In a previous study, we demonstrated that the promiscuity of the trans-AT KirCII, which also accepts non-natural malonate-derived extender units, can be used as a tool for polyketide diversification. By expressing a tailored malonyl-CoA synthetase and feeding of allyl-or propargylmalonic acid externally, the production of allyl-and propargyl-kirromycin was achieved 25 . However, the reported yields of propargyl-kirromycin were low compared to the production of the wild type kirromycin (22% propargyl-kirromycin compared to 78% kirromycin), likely due to the higher activity of KirCII towards the native substrate ethylmalonyl-CoA 34 . Hence, the use of the ΔkirN mutant with its impaired ethylmalonyl-CoA biosynthetic pathway is expected to result in improved yields of the novel derivatives.