1. Comparative bioinformatic analysis of genomes of industrial Streptomyces rimosus strains M4018 and R6 with the parent strain ATCC 10970
As both M4018 and R6-500 strains had significantly greater oxytetracycline titres than the ancestral strain ATCC 10970, we first sought to determine the genomic changes that had occurred during the strain improvement programs carried out by Pfizer and PLIVA respectively. We generated complete genome sequences of M4018 and R6-500 (see Supplementary Information 1 for details of these assemblies and deduced rearrangements) and noticed that both strains had undergone extensive rearrangements in both left hand end (LHE) and right hand end (RHE) (Fig. 1A and 1B). Detailed comparison of the chromosomes is displayed in Supplementary Fig. 1. BGCs present in M4018 and R6-500 were identified by AntiMSASH 6.027 and compared (Supplementary Table 1) with BGCs identified previously in ATCC 1097021. In Streptomyces species, the chromosomal arms are abundant with biosynthetic gene clusters44(BGCs). The genomic alterations observed had resulted in extensive “shuffling” of the repertoire of BGCs and their locations in these industrial strains. For example, M4018 had lost 78,065bp from the LHE of the ATCC 10970 chromosome (genes SRIM_000005 to SRIM_000425) and lost 284,432bp (genes SRIM_038260 to SRIM_039170) from the RHE, adjacent to the OTC BGC, which was now located almost at the end of right chromosomal arm (Fig. 1D and Supplementary Fig. 1). The terminal 323,390p from the RHE of the ATCC 10970 chromosome (genes: SRIM_039175 to SRIM_040530), had fused to the LHE of the M4018 chromosome. In this way, BGCs 5 to 8 were deleted from RHE and BGCs 1 to 4 were relocated to the opposite chromosomal arm in an inverted orientation (marked yellow on Fig. 1D).
In R6-500, the terminal 869,667bp from the RHE of the ATCC 10970 chromosome (genes: SRIM_037035 to SRIM_040530) had duplicated and fused to the LHE of the chromosome (Supplementary Fig. 1). 555,093bp (genes: SRIM_000005 to SRIM_002375) were lost from the LHE of the ATCC 10970 chromosome during the origination of R6-500. Subsequently, a deletion had occurred between gene positions SRIM_036310 to SRIM_038540. Consequently, eight BGCs were deleted in R6-500: BGCs 11–13 and 40–46 (Supplementary Table 1, Supplementary Information 2). Furthermore, the R6-500 genome now contained two copies of the DNA fragment encompassing BGCs 1–7 (Fig. 1E, Supplementary Fig. 1) Although not duplicated, the OTC cluster is found at the LHE of the R6-500 chromosome as opposed to the RHE of ATCC 10970. The linear plasmid has undergone some minor deletions in M4018 whereas ~ 30% of the LHE has been deleted in R6-50021.
2. Identification of the chromosomal regions for targeted deletion experiments in S. rimosus ATCC 10970.
Our genome analysis showed that both M4018 and R6-500 had large-scale rearrangements at one end of their linear chromosomes, centred around the OTC BGC. We had already observed that relocation of the OTC BGC to a different chromosomal location had a positive effect on OTC production28. We therefore tested whether changing the location of the OTC BGC relative to the end of the chromosome had a positive effect on OTC production. This was done by introducing a large deletion in ATCC 10970, using the deletion observed in M4018 as a guide. We evaluated two large deletions with sizes of 145kb and 240kb (Fig. 2). The 145kb deletion removed two BGCs (6 and 7) from the ATCC 10970 genome (strain ATCC Δ145kb). Deletion of the 240kb region removed BGC 5 in addition to BGCs 6 and 7 (strain ATCC Δ240kb).
BGC 6 (Fig. 2, which is alongside the OTC BGC24) encodes biosynthesis of rimocidin (a major metabolite in S. rimosus), a polyene macrolide with antifungal activity29. In our recent study we demonstrated that ATCC 10970 produces not only rimocidin, but also rimocidin congeners CE-108 and amid-CE-10821, which are biosynthesized by a PKS type I enzyme complex30 containing 14 modules, making it one of the larges BGCs present in the genome of S. rimosus. Based on AntiSMASH analysis27 BGC 7 encodes a NRPS gene cluster for a yet unknown secondary metabolite (Supplementary Fig. 2; genes presented in Supplementary Table 2). BGC 5, removed additionally in ATCC Δ240kb, encodes biosynthesis of a yet unknown PKS type I metabolite. Supplementary Table 2 lists genes deleted in ATCC Δ145kb (SRIM_038560-SRIM_038870), including genes of BGC 6 (rimocidin) and BGC 7 (unknown metabolite) and the additional genes deleted in ATCC Δ240kb (SRIM_038560-SRIM_039255), which includes those of BGC 5. The AntiSMASH analysis of BGCs 5, 7 and 22_2 is presented in Supplementary Fig. 2.
3. Deletion of chromosomal regions adjacent to the OTC BGC in S. rimosus ATCC 10970
Using pREP_P1_cas9 tsr as a foundation 26,31, two plasmid constructs (pRep_P1_cas9_Δ145kb and pRep_P1_cas9_Δ240kb) were used to introduce deletions into the ATCC 10970 chromosome (Supplementary information 3). They contained two successive gRNA cassettes, designed to target both ends of the deleted regions, along with adjacent homologous regions of approximately 2Kbp in size (Supplementary Table 3). Plasmids were introduced to S. rimosus ATCC 10970 via conjugation (see Methods section). The overall procedure for creating the deletions in S. rimosus using CRISPR-Cas9 had been optimized previously using the Cas9-SD-GusA tool26. Deletion of the 145kb region was carried out in ATCC 10970 and ATCC 10970 Δotc, already containing a deletion of the OTC cluster28. Verification PCR reactions on colony isolates confirmed that the anticipated deletion events had occurred (Supplementary Fig. 3, Supplementary information 4, Supplementary Table 4). We next performed detailed analysis of the best performing ATCC strain with the 145kb deletion, designated ATCC Δ145kb, but another independent clone with an introduced 145kb deletion (designated ATCC Δ145kb-b) was also included in OTC fermentations and transcriptome analysis (Supplementary Fig. 4).
4. Comparative analysis of OTC production in S. rimosus M4018, R6, ATCC 10970 and engineered strains with 145kb and 240kb deletions
OTC production of the best performing S. rimosus mutants with 145kb and 240kb deletions from a pre-screening experiment were compared with ATCC 10970, and the industrial OTC-producing strains M4018 and R6-500. Remarkably, ATCC Δ145kb, had a significant increase in OTC production of 2.9 g/L - a remarkable 7-fold increase compared to 0.452 g/L produced by its ATCC 10970 parent. This is comparable with the range of titers of the industrial production strains M4018 (2.3 g/L) and R6-500 (1.2 g/L) (Fig. 3) derived by intensive strain selection. ATCC Δ240kb produced 0.844 g/L OTC, still significantly higher than its parental strain ATCC 10970 (Fig. 3). These increases in OTC titers were notable as the growth rates (based on DNA concentration measurements throughout the fermentation process) of the engineered strains were comparable to parental ATCC 10970 strain (Supplementary Fig. 5). Results of OTC production for another independently isolated strain with 145kb deletion (ATCC Δ145kb-b) are presented in Supplementary Fig. 4A.
Alongside significant differences in OTC titers, ATCC Δ145kb but not ATCC Δ240kb also displayed an unusual morphology compared to the ATCC 10970 strain when cultivated on SM agar plates (Supplementary Fig. 6). Most notably, ATCC Δ145kb had aerial mycelium of the “peeling” phenotype, not observed with ATCC 10970 (Supplementary Fig. 6A-6B). On the other hand, we did not detect large differences in growth rates during evaluation of growth characteristics in OTC seed medium (GOTC-V) and production medium (GOTC-P) (Supplementary Fig. 5). However, culturing of ATCC Δ145kb decreased the pH more rapidly in both vegetative and production medium compared to other strains, indicating changes in primary metabolism in comparison to the Type Strain (Supplementary Fig. 5B).
5. Confirmation of targeted deletion of the selected DNA fragment by entire-genome sequencing
Mapping of Illumina reads to the S. rimosus ATCC 10970 reference genome21 confirmed that the anticipated 145kb region had been deleted in the ATCC Δ145kb strain (also in the second strain-ATCC Δ145kb-b; Supplementary Information 5). There was a complete absence of sequencing reads internal to the region of ATCC 10970 that had been deleted. To ensure reliable identification of possible non-specific mutations, two control samples of the ATCC 10970 paternal strain were also re-sequenced. The in-depth WGS analysis of the ATCC Δ145kb strain revealed only a few non-specific mutations per genome (Supplementary Table 5), suggesting that despite invasive perturbation of the genome to create the 145kb deletion, there was little CRISPR-Cas9 ‘off target’ activity in the engineered strain.
6. Targeted inactivation of BGC encoding rimocidin biosynthesis
The entire rimocidin BGC (BGC 6, see Fig. 2) was removed in both ATCC Δ145kb and ATCC Δ240kb strains. It was plausible that the increased OTC titres observed with these deletion strains might be attributed to increased availability of precursors (both OTC and rimocidin require malonyl-CoA14) or some cross-regulation issue between the OTC and rimocidin BGCs. Seco et al. (2004)30 proposed that inactivation of rimocidin biosynthesis in Streptomyces diastaticus var.108 had a significant effect on OTC biosynthesis. Therefore, to clarify the potential effect of deletion of the rimocidin BGC on OTC production by the Δ145kb strain, we introduced a small in-frame deletion inside the rimocidin BGC in ATCC 10970 that specifically inactivated the loading module, RimA. Similarly, although in a different Streptomyces species, Seco et al. (2004)30 also inactivated rimA via insertional inactivation. We constructed a CRISPR plasmid pRep_P1_cas9_ΔrimA (see Methods section), specifically targeting the ACP (acyl carrier protein) and KS (encoding ketosynthase) domains of the loading module RimA (Supplementary Fig. 3C). Deletion of the target DNA fragment was confirmed by PCR (Supplementary Fig. 3C) and absence of all rimocidin-related metabolites (RIM) was confirmed by full-scan LC-MS (see MS analysis results). Inactivation of rimA and therefore interruption of rimocidin biosynthesis had no effect on OTC titer. This was confirmed by HPLC analysis of fermentation broths from 4 independent rimA mutants (mean ATCC 10970 = 0.483 g/L, mean all rimA mutants = 0.402 g/L). For direct comparison, we included ATCC Δ145kb in this fermentation experiment, which on average produced 3.2 g/L (Fig. 4).
7. Comparative transcriptional analysis of S. rimosus ATCC10970 and engineered strains containing deletions of the 145kb and 240kb regions
To elucidate the observed changes in OTC production and morphology with genetic modifications, we have performed differential gene expression analysis of the wild type and the two engineered strains in two time points, 24 and 50 hrs, corresponding to the onset of production and late exponential stage (Supplementary Fig. 7). Cultivation procedure and isolation of mRNA are described in Supplementary Information 6. In line with OTC titers (Fig. 3) and strain morphology (Supplementary Fig. 6), we observed pronounced differences in the transcriptome of ATCC Δ145kb compared to ATCC Δ240kb (Figs. 5, 6, Supplementary Fig. 8). This is somewhat surprising considering that the Δ240kb strain had lost a larger number of genes (see Supplementary Table 2). Both engineered strains showed greater differences in gene expression when compared with ATCC 10970 at the 24-hour sampling time point (see Fig. 5). According to comparison of DNA content in OTC-production broths (Supplementary Fig. 9, Supplementary Information 8), there was no consistent differences in growth rate between tested strains. Principal Component Analysis (PCA) plots displaying the biological variability of samples are presented in Supplementary Fig. 8. To assess the normalization process and evaluate the overall consistency of the data, we compared expression (TPM) of house-keeping genes rpoB in gyrB between all samples32,33 (Supplementary Table 6). In addition, to verify the uniform growth phase of analysed strains at both sampling time points, we compared TPM counts of ftsZ and bldD (Supplementary Table 6), involved respectively in cell division and initiation of antibiotic production34,35.
To gain deeper insight into biological processes altered in the ATCC Δ145kb and ATCC Δ240kb strains, we first performed Gene Ontology (GO) analysis (Supplementary information 7; Supplementary data 2). Most affected molecular function in the engineered strains was transmembrane transport (GO:0055085), with the largest proportion of both overexpressed and down-regulated genes. The ATCC Δ145kb strain also had changes in the regulation of DNA-templated transcription initiation (GO:0006352) and proteolysis (GO:0006508), which were not observed in ATCC Δ240kb strain. Interestingly, ATCC Δ145kb strain also had a large number of affected genes related to metabolism of nitrogen compounds (GO:0006807). As Streptomyces genomes are not well annotated with GO terms, we suspected that GO analysis was not sufficient to cover all the differences between strains. Therefore, to gain a general perspective of observed changes in engineered strains and to emphasize the largest differences between engineered strains and native strain ATCC 10970, we performed a basic clustering analysis of RNA-sequencing data (see Methods section). Clustering was performed on a dataset of genes with log2 fold change of >-2.5 or < 2.5 in at least one engineered strain at least one time point (24h or 50h). The results of clustering analysis are presented as a hierarchically clustered heatmap (Fig. 5).
Based on expression dynamics between analysed strains, the heatmap was divided into 7 clusters (A-G on Fig. 5). Clusters F and G represent mostly down-regulated genes, while clusters A and B represent genes that were generally overexpressed in both engineered strains. On the contrary clusters C, D and E were downregulated in the Δ145kb strain, but remain unaffected or are even up-regulated in the Δ240kb deletion strain (see Fig. 5). These genes are probably responsible for observed differences between engineered strains (list of genes with annotations and expression levels - see Supplementary data 3). The most outstanding is cluster C (see Fig. 5), which is composed of only one operon with an unknown function (SRIM_017880 - SRIM_017875) putatively encoding a Beta-lactamase family protein, UTRA domain-containing protein, ATP-binding protein, NAD(P)/FAD-dependent oxidoreductase and one hypothetical protein.
Overall, genes involved in biosynthesis of secondary metabolites were the most affected group in both ATCC Δ145kb and ATCC Δ240kb strains. Almost exclusively, the genes in cluster G (Fig. 5) belong to secondary metabolism, namely BGC 41 and BGC 46, which are strongly downregulated in both engineered strains. On the other hand, clusters A and B (C9 on Fig. 5) consist largely of genes from BGC 22 (biosynthesis of longicatenamycin B/C and yet unknown metabolite) and BGC 9 – OTC, that were significantly overexpressed in both ATCC Δ145kb and ATCC Δ240kb strains. Together with genes involved in secondary metabolism, several proteases and ABC transporter proteins were also strongly upregulated as part of clusters A and B (see Supplementary Data 3). Overexpression of genes in clusters A and B is much more significant in the ATCC Δ145kb strain than the ATCC Δ240kb strain.
Interestingly, cluster D contains 6 putative genes involved in the synthesis of the rodlet layer, a typical surface layer of both aerial hyphae and spores in Streptomyces. Two rodlin genes rdlA36 and four different chaplin37 genes are significantly down-regulated only in the Δ145kb strain. In Streptomyces species, the chaplin genes are to some extent redundant, but rdlA is indispensable for the establishment of this rodlet structure36. Also, 4 other chaplin and rodlin genes were identified as downregulated in the Δ145kb strain as part of cluster E (Fig. 5). This suggests that changes in the rodlet layer of ATCC Δ145kb might explain the morphological changes in this strain (Supplementary Fig. 6). The large cluster E consists of several poorly annotated or hypothetical genes (presented in Supplementary data 3). We also performed a separate analysis focusing on the number of differentially expressed genes in 100kb intervals across genome of engineered strains (Supplementary Fig. 10). Interestingly, genes from the OTC BGC and BGC 22 are not only the most highly over-expressed genes (Fig. 6) but are also the two largest intervals of differentially expressed genes. Importantly, the greatest transcriptome changes were observed in identical regions for all engineered strains (Supplementary Fig. 10) but are most pronounced in the ATCC Δ145kb strain.
7.1 Comparative bioinformatics analysis of BGC expression in ATCC 10970 and engineered strains
RNA-seq data demonstrated that increases in OTC biosynthesis of the engineered strains are correlated with a significant increase in expression of otc biosynthetic genes. Significantly, otc biosynthetic genes were the most overexpressed group of in both ATCC Δ145kb and ATCC Δ240kb strains (Fig. 6), with average log2 fold ratios of 4.7 and 1.7, respectively (Table 1, Supplementary Data 1).
Alongside BGC 9 (OTC BGC), a cryptic cluster annotated as BGC 22 caught our attention due to over-expression (> 32-fold change, Table 1, Fig. 6) in the ATCC Δ145kb strain. Upon more detailed analysis, we propose that the BGC 22 region consists of two separate BGCs each with its own NRPS and accessory genes, annotated here as 22_1 (SRIM_030975 - SRIM_031065) and 22_2 (SRIM_030895 - SRIM_030970). The two clusters are adjacent, possibly share accessory genes, and are overexpressed with identical intensity (see Supplementary Fig. 2.1). The first part (BGC 22_1) encodes a cyclic hexapeptide longicatenamycin B, recently discovered in S. rimosus by Li et al. (2022)38, and the second part (BGC 22_2) encodes a yet unknown NRPS metabolite (AntiSMASH analysis - Supplementary Fig. 2). The entire BGC 22 was moderately overexpressed in the ATCC Δ240kb strain (1.86-fold change) (Table 1, Fig. 6F).
Despite the observed overexpression of BGC 22 in ATCC Δ145kb, the absolute expression strength of cluster BGC 22 was only around 10% in comparison to the OTC BGC (see TPM scale at Fig. 6A-B). Therefore, the magnitude of BGC_22 overexpression is so high, considering it originates from its almost cryptic level of expression in ATCC 10970 (Fig. 6C). Importantly, BGC 22_1 encoding longicatenamycin congeners has been considered cryptic in S. rimosus, and until now its activation could only be detected when using high-throughput elicitor screening (HiTES)38. Similarly, longicatenamides A-D were detected in Streptomyces cultures only when using a co-cultivation approach39. Our results therefore indicate that targeted genome reductions can be used to induce expression of cryptic clusters. Also, it is important to emphasize that the expression of other BGCs, of which many are un-characterised, was also significantly affected in the ATCC Δ145kb and ATCC Δ240kb strains (see Table 1). A Volcano Plot depicting transcriptome changes in a second independent strain with 145kb deletion (ATCC Δ145kb-b) is presented in Supplementary Fig. 4B. Importantly, OTC BGC and BGC 22 remain the most overexpressed groups of genes in the ATCC Δ145kb-b strain, confirming the effect of the 145kb deletion on observed BGC overexpression.
Table 1
Significantly affected* BGCs after genome reductions in engineered S. rimosus strains. Average log2 ratio - fold change of expression at 24h time point. * BGCs with an average gene log2 fold change of > 1 or <-1 in at least one mutant strain was considered significantly affected.
Significantly affected BGC
|
Metabolite
|
Fold change-log2 ratio (x vs ATCC 10970)
|
ATCC Δ145kb
|
ATCC Δ240kb
|
BGC 22_1
|
longicatenamycin B
|
5,36
|
0,88
|
BGC 22_2
|
unknown
|
5,15
|
0,91
|
BGC 9 - OTC
|
oxytetracycline
|
4,7
|
1,7
|
BGC 45
|
momomycin
|
1,73
|
0,79
|
BGC 23
|
tyrobetaine-2
|
1,74
|
0,85
|
BGC 46
|
unknown
|
-1,97
|
-1,02
|
BGC 5
|
unknown
|
-2,33
|
/
|
BGC 41
|
unknown
|
-4,09
|
-2,33
|
8. Comparative analysis of metabolite profiles of S. rimosus ATCC 10970 and engineered strains by applying MS analysis
Full-scan MS analysis of the engineered strains was done to assess the impact of the two deletions on metabolite production profiles. We immediately observed peaks belonging to oxytetracycline (OTC; 461,7 [M + H]+obs / 460,4340 [M]) (OTC on Fig. 7) and rimocidins in S. rimosus ATCC 10970 (RIM on Fig. 7A). On full-scan LC-MS chromatograms (Supplementary Fig. 11) we could distinguish between production of 5 rimocidin congeners in the ATCC 10970 strain background. Additionally, in the next step, using high-resolution mass spectrometry (HR-MS) we confirmed production of rimocidin, CE-108, rimocidin (27-etyl), CE-108B and rimocidin B in the ATCC 10970 background (HR-MS data - Supplementary Table 7, RIM structures - Supplementary Fig. 13). The ATCC 10970 ΔOTC strain was included in the metabolite profile analysis to investigate if deletion of the OTC BGC affects production of other metabolites. Despite the removal of OTC as the most abundant metabolite, the metabolome profile of the ATCC 10970 ΔOTC strain did not change, with exception of a minor increase in base peak intensity belonging to the rimocidins (see Fig. 7A, Y axis on Supplementary Fig. 11). Interestingly, the precise in frame deletion of rimA, encoding the polyketide synthase loading module of rimocidin BGC did not have any effect on OTC production (see Fig. 9C). As expected, no rimocidin congeners could be detected in the Δ145kb and Δ240kb deletion mutants (see Fig. 7B, C, E, F and Supplementary Fig. 11). Comparable to the HPLC results (Fig. 3), differences in OTC titers between ATCC 10970 and ATCC Δ145kb and ATCC Δ240kb strains can clearly be observed also on full-scan-LC-MS spectra (see Fig. 7A, B, C). Absence of the OTC BGC accompanied with deletion of 145kb region in ATCC 10970 ΔOTC Δ145kb strain resulted in a clean background (Fig. 8); the metabolic profile contained virtually no major metabolites.
In addition to OTC and rimocidin congeners, we also attempted to identify metabolites related to cryptic BGCs, which were noticed through transcriptomics analysis (Table 1, Fig. 6A). Accordingly, full-scan LC-MS analysis of the ATCC Δ145kb sample revealed three newly emerged peaks, that became even more evident when metabolites were extracted using 4 volumes of acetonitrile (ACN) ([M + H]+obs 387,7; 764,9 and 778,9 (peaks 1, 2 and 3 on Fig. 9; see also Table 2). Peak no. 1 was predicted to be tyrobetaine-2, a member of the tyrobetaines, a class of nonribosomal peptides with an unusual trimethylammonium tyrosine residue41. Tyrobetaine-2 was also detected in the control ATCC 10970 extract, but with lower abundance (Y axis on Supplementary Fig. 12A). Interestingly, the tyrobetaine BGC (BGC 23) was significantly overexpressed in the Δ145kb strain (Fig. 6A; Table 1), which is consistent with the LC-MS results (Fig. 9B).
The other two new peaks ([M + H]+obs 764,9 and 778,9) were considered to be potential cryptic metabolites produced by one of the overexpressed BGC in the ATCC Δ145kb strain (Table 2). Indeed, peak no. 2 was identified as longycatenamycin B38,42 in the next step using HR-MS ([M + H]+obs 763.3546, [M + H]+cal 763.3546). BGC 22_1, encoding longycatenamycin B, and BGC 22_2 are the most overexpressed BGCs in the engineered strains, which is consistent with the appearance of this metabolite.
We speculated that peak No. 3 (Fig. 9B) represents a new metabolite produced by uncharacterised NRPS complex BGC 22_2 (presented in Supplementary Figs. 2 and 2.1), considering its strong overexpression in the ATCC Δ145kb strain. To our surprise, the HR-MS results (Supplementary Table 7) indicated that peak No.3 is in fact another member of the longicatenamycin (S-520 antibiotic) family, synthesized by the same NRPS as longicatenamycin B (BGC No. 22_1), with an additional methyl group (D-Val ->Ile) ([M + H]+obs 777.3699, [M + H]+cal 777.3702). This longycatenamycin congener was detected before by von Nussbaum et al. (2008)42 and possibly by Shoji and Sakazaki (1970)43, but only as a component of a complex mixture of congeners and without a specific name. We therefore designated this congener as longicatenamycin C (Supplementary Table 7).
No longicatenamycin could be detected by LC-MS in the native ATCC 10970 strain. We also attempted to evaluate if the presence of longicatenamycins in the ATCC 10970 strain could be masked by RIM congeners, which elute at same retention time and have similar [M + H]+obs (see Supplementary Fig. 7). We therefore analysed chromatograms obtained by full-scan MS of the rimA mutant strain, not producing RIM and found traces of both longicatenamycin B and C. In accordance with the work by Li et al. (2022)38 where longicatenamycin B was first characterised, these authors also used a RIM non-producing strain of S. rimosus. To conclude, we show here that longycatenamycins B and C are present in trace amounts in the ATCC 10970 parent strain, masked to LC-MS detection by larger amounts of RIM congeners.
Table 2
Major peaks/metabolites detected on full-scan LC-MS analysis in engineered strains and ATCC 10970 parent strain. +++ base peak intensity > 5,0 E5, ++base peak intensity > 1,0 E5, + base peak intensity > 5,0E3, - not detected.
BGC no.
|
Metabolite
|
Base peak intensity in strain
|
[M + H]+obs / [M]
|
Δ145kb
|
Δ240kb
|
ATCC 10970
|
|
9
|
oxytetracycline
|
+++
|
++
|
+
|
461,7 / 460.4
|
6
|
rimocidins
|
-
|
-
|
+++*
|
739,4; 740, 3;
754, 3; 768,34 / *
|
22_1
|
longicatenamycin B
|
++
|
+
|
-
|
764,8 / 763.3 *
|
22_2
|
longicatenamycin C
|
++
|
+
|
-
|
778,9* / 777.4 *
|
23
|
tyrobetaine-2
|
++
|
+
|
+
|
387,7 / 387.4
|
*see also HR-MS data (Supplementary Table 7) |
9. Repositioning of the entire otc BGC in S. rimosus ATCC Δ145kb and ATCC Δ240kb strains
To explore the effect on OTC titer by the location of the otc BGC and gene dosage, we designed an OTC cluster relocation experiment, similar to that described by Pikl et al. (2021)28. We introduced the identical 145kb deletion into the genome of the S. rimosus ATCC 10970 ∆OTC strain, which has almost the entire otc BGC deleted and therefore does not produce oxytetracycline (see Supplementary information 3 and 8). Three independent S. rimosus ATCC 10970 ∆OTC ∆145kb colonies were then complemented with a pYAC-ΦC31-Ts-OTC construct containing the entire OTC BGC integrated at the ΦC31 attB site. This results in the re-location of the otc BGC from the left terminal arm to the attB site - located in the central region of S. rimosus genome28 (Fig. 10C). Similarly, duplication of the OTC BGC was achieved by introducing plasmid pYAC-ΦC31-Ts-OTC plasmid into the attB site of the ATCC ∆145kb strain (see Fig. 10B).
9.1 Evaluation of the OTC titer and morphological properties of the engineered strains S. rimosus ATCC10970∆OTC∆145kb::otc and S. rimosus ATCC10970∆145kb::otc
All three isolates of ATCC 10970 ∆OTC ∆145kb::otc had increased OTC production with an average OTC titer of 3 g/L (Fig. 11A). Pikl et al. (2021)28 had already demonstrated that relocation of the OTC cluster to a more central region of the chromosome of the ATCC 10970 parent strain had a positive effect on OTC production, resulting in an OTC titer increase from around 200mg/L to around 1g/L. Interestingly, when relocation of the otc BGC was carried out in the ATCC 10970 ∆OTC_∆145kb strain background, the OTC titer was over 7 times higher, compared to the ATCC 10970 parental strain. Hence, this experiment confirmed our observation that the 145kb deletion has a strong positive effect on OTC biosynthesis and that positive effect is not exclusive to the native chromosomal location of the OTC BGC.
To evaluate the effect of copy number of the otc BGC on OTC titer, we also introduced the pYAC-ΦC31-Ts-OTC plasmid28 containing the entire otc BGC into the ATCC Δ145kb strain, thus generating a strain containing a second copy of the otc BGC (see Fig. 10B). Remarkably the transformant (ATCC Δ145kb + pYAC-ΦC31-Ts-OTC) had a titer of 5.2 g/L of OTC (Fig. 11), which is almost 10x higher compared to the parent ATCC 10970 strain and almost double the amount (2.8g/L, Fig. 2) produced by the ATCC Δ145kb strain. Considering the titers of industrial strains M4018 and R6-500, we have demonstrated that industry-relevant OTC titers can be achieved by introduction of a second copy of the OTC BGC to the genome of native parent strain ATCC 10970 containing the 145kb deletion (Fig. 11 compared to Fig. 2).