Identification and utilization of two important transporters: SgvT1 and SgvT2, for griseoviridin and viridogrisein biosynthesis in Streptomyces griseoviridis

Background Griseoviridin (GV) and viridogrisein (VG, also referred as etamycin), both biosynthesized by a distinct 105 kb biosynthetic gene cluster (BGC) in Streptomyces griseoviridis NRRL 2427, are a pair of synergistic streptogramin antibiotics and very important in treating infections of many multi-drug resistant microorganisms. Three transporter genes, sgvT1–T3 have been discovered within the 105 kb GV/VG BGC, but the function of these efflux transporters have not been identified. Results In the present study, we have identified the different roles of these three transporters, SgvT1, SgvT2 and SgvT3. SgvT1 is a major facilitator superfamily (MFS) transporter whereas SgvT2 appears to serve as the sole ATP-binding cassette (ABC) transporter within the GV/VG BGC. Both proteins are necessary for efficient GV/VG biosynthesis although SgvT1 plays an especially critical role by averting undesired intracellular GV/VG accumulation during biosynthesis. SgvT3 is an alternative MFS-based transporter that appears to serve as a compensatory transporter in GV/VG biosynthesis. We also have identified the γ-butyrolactone (GBL) signaling pathway as a central regulator of sgvT1–T3 expression. Above all, overexpression of sgvT1 and sgvT2 enhances transmembrane transport leading to steady production of GV/VG in titers ≈ 3-fold greater than seen for the wild-type producer and without any notable disturbances to GV/VG biosynthetic gene expression or antibiotic control. Conclusions Our results shows that SgvT1–T2 are essential and useful in GV/VG biosynthesis and our effort highlight a new and effective strategy by which to better exploit streptogramin-based natural products of which GV and VG are prime examples with clinical potential. Electronic supplementary material The online version of this article (doi:10.1186/s12934-017-0792-8) contains supplementary material, which is available to authorized users.


Background
The actinomycete Streptomyces griseoviridis NRRL 2427 generates two types of unrelated streptogramins; these include griseoviridin (GV) belonging to the A-type cyclic polyunsaturated macrolactone subclass, and viridogrisein (VG), a B-type cyclic depsipeptide counterpart to GV also referred to as etamycin ( Fig. 1) [1,2]. GV and VG bind to the A and P sites, respectively, of the 50S bacterial ribosomal subunit thereby averting peptide bond formation during the elongation phase of protein translation. Moreover, GV and VG have been noted to work synergistically leading to antibacterial activities superior to those of either compound alone; this effect suggests that the combination of GV/VG has great potential with respect to treating multidrug resistant microbial pathogens [3][4][5].
Stemming from the diversity of structures and bioactivities of their secondary metabolites, the streptograminproducing actinomycetes commonly employ several transmembrane transporters as drug efflux pumps to avoid intracellular metabolite accumulation. These transporters typically belong to major facilitator superfamily (MFS) and ATP-binding cassette (ABC) families and constitute an essential self-resistance mechanism to efficiently secrete antibiotics as they are constructed thereby protecting the producing microbe from the effects of its own secondary metabolism [5][6][7][8][9][10]. This is a central idea in microbial homeostasis [11][12][13][14]. MFS transporters commonly possess 12 or 14 transmembrane segments (TMS) and employ a transmembrane electrochemical gradient to secrete secondary metabolites [11][12][13][14][15]. Conversely, the ABC transporters share a conserved domain organization of separate transmembrane domain (TMD) subunits and nucleotide-binding domain (NBD, also called ATP-binding cassette) subunits enabling ATP hydrolysis and application of the resulting energy to effect metabolite removal from the cell [11][12][13][14]16]. Notably, the majority of prokaryotic ABC transporters consist of stand-alone TMD or NBD polypeptides which must dimerize in some fashion to generate fully functional proteins [12,17].
Not surprisingly, overexpression of actinomycete efflux transporters significantly enhances antibiotic release rates. This translates to enhanced antibiotic production by reducing feedback inhibition in vivo [13,18]. This strategy, termed "transporter engineering" is widely applied in large-scale industrial antibiotic production processes; some examples include production of pradimicin, avermectin, doxorubicin and neomycin in Streptomyces [19][20][21][22]. Enabled by the recently identified 105-kb GV/VG biosynthetic gene cluster (Fig. 2) [1,2], we report herein: (i) the identification of three transporter encoding genes housed within the GV/VG biosynthetic gene cluster: sgvT1-T3 (SgvT1 and SgvT3 are MFS transporters and SgvT2 is an ABC transporter) that are regulated by γ-butyrolactone (GBL)-type signaling, (ii) SgvT1 and SgvT2 are both necessary for efficient GV/ VG biosynthesis with SgvT1 playing an indispensable role in maintaining stable expression throughout sustainable GV/VG biosynthesis, and (iii) a roughly threefold increase in GV/VG titers resulting from sgvT1-T2 overexpression. These studies provide new insight into host self-resistance mechanisms in streptogramin biosynthesis and highlight a practical approach to improving streptogramin production.

Discovery of SgvT1-T3 as a two class-based transporter system
Streptomyces griseoviridis NRRL 2427 is a well-known producer of GV and VG; the highest yield of GV is 33.04 ± 0.70 μg/mL and that of VG is 31.56 ± 0.51 μg/mL both over the course of 108 h fermentation. Subsequent early analyses confirmed that ≈ 75% of GV/VG produced in such systems is found in the fermentation supernatant (Additional file 1: Table S6) suggesting that S. griseoviridis NRRL 2427 houses an effective efflux system in accordance with the annotated set of transporters genes sgvT1-T3 (Fig. 2).
The gene sgvT1 encodes a 531 aa MFS type transporter and is located proximal to the upstream boundary of the sgv biosynthetic gene cluster. Conversely, sgvT3 which codes for another MFS type transporter composed of 464 aa, is positioned downstream of sgvT1 and proximal to the 10 kb regulatory region within the sgv gene cluster (Fig. 2) [1]. The assignment of SgvT1 and SgvT3 as MFS type proteins is based on the alignment of conserved sequences and domains with previously characterized acinomycete MFS transporters, such as CmcT (Amycolatopsis lactamdurans and S. clavuligerus), EncT (S. maritimus) and MctT (S. lavendulae) [11,[23][24][25]. Both SgvT1 and SgvT3 possess TMS for functional exploitation of transmembrane electrochemical gradients [14]. SgvT3 contains 14 highly conserved TMS regions whereas SgvT1 contains 15 TMS fragments with two different regions and one more TMS region in the C terminal, even though the other 13 TMS are quite conserved (Additional file 1: Figure S1). By contrast, SgvT1 is slightly larger than many of its counterpart transporters by an average of ≈ 50-70 aa; this increased size may explain the additional fold seen with SgvT1 but not in related counterparts.

Identification of SgvT1/T2 as two necessary transporters in regular GV/VG biosynthesis
Using established λ-RED-mediated PCR-targeting mutagenesis methods we constructed gene insertion mutants ΔsgvT1-T3 for subsequent function analyses. Following validation of the desired mutations, all three mutant strains were cultured in liquid or solid medium and no obvious changes could be noted relative to the wild-type strain based on growth curves and morphological analyses (Fig. 3a, b). The constructed mutant strains displayed significantly different efficiencies in GV/VG production (Fig. 3c). GV/VG production by the ΔsgvT1 mutant strain was significantly impaired; metabolite generation (optimal) by this strain was only 14-16% of that realized with WT producer [yields: (5.28 ± 0.19 μg/mL for GV and 4.65 ± 0.13 μg/mL for VG after 60 h fermentation)]. The ΔsgvT2 strain also suffered sharply reduced GV/VG production efficiencies with yields only on par with 30-35% of that observed with WT producer; metabolite titers for the ΔsgvT2 were 10.05 ± 0.09 μg/mL GV and 11.21 ± 0.35 μg/mL VG after fermentation for 108 h (see: Fig. 4a, b and Additional file 1: Table S6). Importantly, the highest intracellular GV/VG concentrations in the ΔsgvT1 and ΔsgvT2 mutant strains were found to be ≈ 1.5 μg/mL. Intracellular GV/VG concentrations for the WT producer proved substantially higher at 8.12 ± 0.62 μg/mL GV and 7.12 ± 0.52 μg/mL VG using comparable fermentation conditions (Additional file 1: Table S6). The dramatically different intracellular concentrations for GV/VG between mutant and the unmodified WT strains suggest impaired biosynthesis correlating to sgvT1 and sgvT2 inactivation indicating that the encoded transporters are integral to efficient GV/VG biosynthesis. Then, we performed the trans complementation of ΔsgvT1 and ΔsgvT2 mutant strains and compared the fermentation with WT strains. As expected, we found that both two trans complemented stains' GV/VG biosynthesis could be readily restored to near WT levels (Additional file 1: Figure S11). These complementation assay also verified the indispensable roles of SgvT1 and SgvT2 in GV/VG biosynthesis, in turn.
By contrast, little to no impact on GV/VG titers were apparent following metabolite analysis of ΔsgvT3 strain fermentations (Fig. 3c). This finding indicates that the SgvT3 maybe a compensatory transporter responsible for transport of intermediates in GV/VG biosynthesis. Moreover, it is interesting to note that, whereas inactivation of SgvT1/T2 had a profound influence on GV/VG production, inactivation of SgvT3 had no influence upon metabolite titers. Thus, this finding also suggests the alternative role of SgvT3 in the transporter of GV/VG, which can likely be carried out by the intact SgvT1/T2 system.

Regulation of sgvT1-T3 expression by the GBL signaling pathway
Global on/off regulation via γ-butyrolactone (GBL)-type "quorum sensing" signal molecules of pathway-specific activation of GV/VG biosynthesis has been previously shown [1]. Inactivation of the essential GBL signal molecule biosynthetic gene sgvA will effectively shut down expression of GV/VG biosynthetic pathway-specific activators coding genes sgvR2 and sgvR3. Consequently, inactivation of either sgvR2 or sgvR3 leads to abolished expression of GV and VG biosynthetic genes such as sgvQ and sgvD1, respectively. This cascade of gene inactivations ultimately precludes GV/VG biosynthesis [1]. Using a ΔsgvA mutant devoid of GBL signaling capacity we found that sgvT1-T3 all failed to be expressed, which could be restored in strain ΔsgvA::sgvA (sgvA insertion mutant ΔsgvA complemented with sgvA) (Fig. 3d). However, mutant strains ΔsgvR2 and ΔsgvR3 in which regulatory genes had been inactivated, as well as strains in which key GV/VG backbone biosynthesis genes were inactivated (ΔsgvQ and ΔsgvD1) were all found to express sgvT1-T3 at rates similar to that seen with the WT producer (Fig. 3d). On the basis of these data it is clear that sgvT1-T3 expression is tightly regulated by global GBL signaling and is not significantly or directly influenced by GV/VG biosynthetic machinery (Fig. 5).

Identification of the pivotal role of SgvT1 in sustainable GV/VG biosynthesis
Analysis of GV/VG production efficiencies (Fig. 4) reveals that ΔsgvT1 and ΔsgvT2 mutant strains both produce GV and VG with comparable efficiencies up to approx. 48 h. However, differences in biosynthetic efficiencies between the two strains become noticeable starting at about 48 h and continuing to 120 h where the difference is most pronounced (Fig. 4a, b). In efforts to correlate GV/VG biosynthesis to specific changes in gene expression in both ΔsgvT1 and ΔsgvT2 mutants we employed qPCR to monitor changes in mRNA levels for specific GV/VG-related genes. These included: sgvA (putative GBL signaling gene), sgvR1 (repressor gene), sgvR2, R3 (activator genes), GV/VG backbones biosynthetic genes sgvE1/sgvD1 (GV/VG backbone biosynthetic genes), and sgvT1-T3 [1]. We found that transcripts for ΔsgvT2 correlated very closely with those found for WT strain at during the early (up to 60 h) phases of GV/VG biosynthesis. This was the case also for the intermediate (120 h) stage of biosynthesis presumably because intact SgvT1 and SgvT3 could ensure GV/VG transmembrane transport needed to prevent intracellular accumulation of GV and/or VG (Fig. 4c, d) and subsequent feedback inhibition. Similarly, transcript analyses for the ΔsgvT1 mutant revealed normal levels of gene expression in the  Proposed transfer mechanism via SgvT1-T3 in GV/VG biosynthesis in S. griseoviridis NRRL 2427. GBL signal pathway not only activates the GV/VG biosynthesis but also prompts the synchronized sgvT1-T3 expression to avoid disorder GV/VG accumulation early stage of GV/VG biosynthesis on par with those observed for ΔsgvT2 and WT strains (Fig. 4c).
In considering GV/VG biosynthesis over the course of 120 h we noted that all GV/VG biosynthetic genes, except for sgvA, sgvR1 and sgvT3, were down-regulated in the ΔsgvT1 mutant (Fig. 4d). These changes indicated that only intact SgvT2 and SgvT3 could not potentially sustain GV/VG biosynthesis throughout all fermentation stages. In concert with the realization (Fig. 4a, b) that, over this time frame, GV/VG titers are most dramatically different between the ΔsgvT1 and ΔsgvT2 mutants, it becomes clear that SgvT1 plays an apparently critical role in modulating GV/VG biosynthesis and that its effects correlate to changes in sgvR2, sgvR3, sgvD1, and sgvE1. That no comparable decreases in transcripts were noted for the sgvT2 mutant (Fig. 4c, d) suggests SgvT2 is important as one element of a SgvT2/T3 efflux system but that its role in this manifold is secondary to that of the SgvT1/T2 system. These data firmly suggest that effective GV/VG biosynthesis in S. griseoviridis NRRL 2427 calls for an intact SgvT1/ T2 efflux system and that the effectiveness of this transport mechanism far exceeds that of a putative SgvT2/T3 system.

Generation of a GV/VG high yielding recombinant strain by overexpression of sgvT1/T2 in S. griseoviridis NRRL 2427
Given the importance of the SgvT1/T2 efflux system we sought to generate a high yielding GV/VG-producing strain via transport engineering. Notably, the trans complementation in ΔsgvT1 and ΔsgvT2 mutant strains could readily restore the GV/VG to near WT levels (Additional file 1: Figure S11). Predicated on this finding, and the clear importance of SgvT1/T2 efflux we envisioned that sgvT1/ T2 over-expression in the WT strain might readily afford a high yielding GV/VG producer. Accordingly sgvT1/T2 were incorporated into the WT producer to afford S. griseoviridis NRRL 2427::sgvT1-T2 (WT::sgvT1-T2). Placement under the control of the ermE promoter afforded approximately a 20-fold over-expression of sgvT1 and a 19-fold over-expression of sgvT2. As expected, the highyielding WT::sgvT1-T2 strain was found to generate GV/ VG in substantially greater yield than does the native S. griseoviridis. Commensurate with expectation, GV and VG yields relative to the WT producer increased by ≈ 3-fold [(106.85 ± 1.81 μg/mL GV/106.61 ± 1.12 μg/mL VG) see Fig. 6a-c, Additional file 1: Table S6]. Especially noteworthy is that GV/VG yields steadily increased while the expression of previously examined biosynthetic genes (Fig. 4c, d) remained constant; the latter is indicative of stable intracellular homeostasis (Fig. 6d). Additionally, we found that the higher yields of GV and VG obtained using WT::sgvT1-T2 were achieved even in the absence of antibiotic generally used to maintain selective pressure (Additional file 1: Figure S10).

Discussion
Transporter-encoding genes within secondary metabolite biosynthesis gene clusters are central to cell viability and homeostasis; both rely heavily on a cell's ability to pump antibiotics and other toxic metabolites out of the cell [11][12][13][14]. Advances in genomics and gene inactivation assays enabled us to identify three transporter encoding genes: sgvT1-T3 distributed within the GV/VG biosynthetic gene cluster. SgvT3 is a conserved MFS transporter with 14 TMS. Interestingly, inactivation of sgvT3 failed to significantly impact GV/VG titers suggesting that SgvT3 may serve as a form of alternative transporter. Precedence for this logic has been reported for landomycin C and cephalosporin; inactivations of lndW (in the landomycin cluster) and cefT (in the cephalosporin cluster) also failed to impact natural product titers [30,31]. The failure of inactivations of putative transporters to significantly alter metabolite titers suggests that losses in their activity can be compensated for using substitute or alternative transport mechanisms. In the case of GV/VG biosynthesis, we found that loss of SgvT3 activity is likely compensated for by its counterparts SgvT1 and SgvT2.
SgvT1 and SgvT2 are both necessary transporters involved in GV/VG biosynthesis; inactivation of either protein significantly decreased GV/VG fermentation titers. SgvT1 is a rare 15 TMS-containing MFS transporter. Alternatively, SgvT2 is composed of two NBD domains but is devoid of any TMS. The absence of TMS elements in SgvT2 suggests a reliance upon other TBD domain-containing transporters in order for SgvT2 to be involved in any kind of effective GV/VG efflux system. That two distinct chemical species (GV and VG) are generated by the S. griseoviridis producer also supports the notion that effective efflux systems may require two dramatically different types of transporter proteins to work synergistically to maintain homeostasis within the producer.
Griseoviridin and VG are both excellent examples of streptogramins and previous work has placed a great deal of importance on understanding mechanisms of streptogramin cellular export during biosynthesis [4][5][6][7][8][9][10]. We focus herein much more intensely on SgvT1 and SgvT2. Careful analyses of both metabolite production profiles and mRNA levels throughout long periods up to 240 h of fermentation unveiled a new appreciation for the importance of SgvT1 in GV/VG biosynthesis. In particular, the careful tracking of both GV/VG titers and transcript levels for selected key genes using ΔsgvT1 and ΔsgvT2 mutant strains shed tremendous insight into the roles of these transporter genes. These analyses make clear that SgvT1 is pivotal to metabolite export and that its inactivation is only minimally compensated for by either SgvT2/T3 or perhaps other transporters within the producer genome. This is in stark contrast to the case of sgvT3 inactivation, which, based on our results, is readily compensated for by an intact SgvT1/T2 transport couple.
Inactivations of sgvT1-T3 do not appear to directly affect the expression of GV/VG biosynthetic or activating genes. Only the ΔsgvT1 mutant appeared to have any of these gene's expression indirectly inhibited, presumably in response to intracellular environmental adaptations. Moreover, we found that sgvT1-T3 are regulated by the GBL signaling pathway, which appears to synchronize with GV/VG biosynthesis without directly influencing GV/VG titers. On the basis of these findings we envision that, when the S. griseoviridis (WT or mutants) producer receives the GBL signal to activate GV/VG biosynthesis, sgvT1-T3 expression ensues to coordinate prompt GV/ VG secretion thereby averting intracellular GV/VG accumulation. In this way, we posit that GBL signaling plays a key role in GV/VG biosynthesis by maintaining cellular homeostasis and coordinating GV/VG biosynthesis and transmembrane export (Fig. 5).
An overarching goal of transport engineering is to increase natural product titers via overexpression of transport genes. Accordingly, and motivated by our findings with sgvT1-T3, we overexpressed sgvT1 and sgvT2 to enhance transmembrane secretion of GV/VG thereby improving natural product titers. Exploitation of ermE promoter technology enabled us to overexpress both sgvT1/T2 by ~ 20-fold which translated to a threefold increase in GV and VG titers. Notably, this approach does not disturb biosynthetic gene expression profiles and affords a producer no longer dependent on environmental antibiotic exposure to generate the antimicrobials of interest. Over-expression of sgvT1/T2 in the mutant GV/VG producer not only validates the importance of these transporters but also provide a new approach for future industrialized exploitation and use of the streptogramins GV and VG.

Conclusions
In this study we have identified the roles of SgvT1-T3 in GV/VG biosynthesis: SgvT1 and SgvT2 are both necessary for efficient GV/VG biosynthesis with SgvT1 playing an indispensable role in maintaining stable biosyntheic gene expression throughout sustainable GV/VG biosynthesis; but SgvT3 is an alternative MFS-based transporter that appears to serve as a compensatory protein in GV/VG biosynthesis. All these three transporters' coding genes are regulated by γ-butyrolactone (GBL)type signaling, which appears to synchronize with GV/ VG biosynthesis and avert excessive intracellular GV/VG accumulation. Enabled by the discovery and application of two critical transporters SgvT1/T2, we constructed a new steady GV/VG high-producing strain in titers about threefold greater than seen for the wild-type producer and without any notable disturbances to GV/VG biosynthetic gene expression or environmental antibiotic control. Consequently, our effort highlighted a new strategy to better exploit streptogramin-based natural products and demonstrate that enhancing self-resistance mechanisms in antimicrobial producing organisms is a valuable approach to safely and efficiently improving metabolite production processes.

Bacterial strains, plasmids, media and general experimental procedures
The general methods and materials used, as well as, all bacterial strains, plasmids, and culture conditions have all been previously described [1,2]. Fermentations and GV/VG production and HPLC analyses of S. griseoviridis NRRL 2427 and related mutant strains were executed as previous described [1,2].
The trans complementants of ΔsgvT1/T2 mutants were constructed using modified vector pSET152AKE, the manipulations of pSET152AKE have been previously described [1,2]. The complementant of S. griseoviridis NRRL 2427 WT strain (WT::sgvT1-T2) was constructed by two steps. The first step entailed introduction of sgvT1 by conjugation with pPWW50Apr-sgvT1 and screening via apramycin resistance as previous described [33]. The sgvT2 gene was then introduced by conjugation to pSET152AKE-sgvT2 and screening for desired fusions was carried out by kanamycin resistance screening as previous described [1,2]. The primers used are shown in Additional file 1: Table S3.

Quantitative analysis of bacterial growth and GV/VG production using S. griseoviridis NRRL 2427 WT and mutant strains
Four strains, the S. griseoviridis NRRL 2427 WT strains, two mutants ΔsgvT1/T2 and the WT::sgvT1-T2, were fermented and subsequent metabolite analyses carried out as previously described [1,2]. Ten flasks per strain were withdrawn every 12 h to measure cell dry weights needed to construct growth curves and to determine GV/VG yields. The cell cultures (200 μL/per flask) were collected on filter papers and dried at 55 °C before weight measurements were taken.
To measure GV/VG production averages included in the confidence interval, two standard curves were generated using analytically pure sample and HPLC analysis (Additional file 1: Figure S10). To establish the GV/VG fermentation curve, ten flasks/per strain of cell cultures (250 μL/per flask) were extracted with butanone and concentrated in vacuo affording oil residues. All residues were then dissolved in MeOH and subjected to HPLC analysis. To analyze extra/intracellular GV/VG production efficiencies, each 50 mL batch of fermentation broth was centrifuged (4000g, 10 min) to separate precipitates and cells from the supernatant [1]. Broths were then extracted with 100 mL EtOAc and the resulting organic extracts concentrated under reduced pressure to afford residues that were then dissolved into 2 mL MeOH. Samples were then centrifuged at 12,000g for 15 min; both the supernatants and pellet contents were then subjected to HPLC analysis.

RT-PCR and qPCR analysis of S. griseoviridis NRRL 2427 WT and mutant strains
Mycelia harvested at specific time points were powderized using in N 2 (l) and total RNA was extracted from the frozen pellet using the SV total RNA purification Kit (Promega). RNA samples were subjected to Dnase I (Promega) digestion according to the manufacturer instructions. First-strand cDNA synthesis was accomplished using Invitrogen's SuperScript ™ Kit and second step PCR was carried out under previously indicated conditions