The Biosynthesis of Spinosyn in Saccharopolyspora spinosa: Synthesis of Permethylated Rhamnose and Characterization of the Functions of SpnH, SpnI, and SpnK

Abstract

Spinosyn A is a polyketide-derived macrolide produced by Saccharopolyspora spinosa and is an active ingredient in several commercial insecticides. It is glycosylated by a tri-O-methylated rhamnose at C-9 and a forosamine at C-17. Previous studies indicated that the rhamnose methyltransferases are encoded by the spnH, spnI and spnK genes. To verify the functions of these methyltransferases and to study how they are coordinated to achieve the desired level of methylation of rhamnose, we studied the catalytic properties of the spnH, spnI and spnK gene products and validated their roles in the permethylation process of spinosyn A. Our data reported herein firmly established that SpnH, SpnI, and SpnK are the respective rhamnose 4′-, 2′-, and 3′-O-methyltransferase. Investigation of the order of the methylation events revealed that only one route catalyzed by SpnI, SpnK and SpnH in sequence is productive for the permethylation of the rhamnose moiety. Moreover, the completion of rhamnose permethylation is likely achieved by the proper control of the expression levels of the methyltransferase genes involved. These results set the stage for future exploitation of spinosyn biosynthetic pathway to produce targeted spinosyn derivatives and, perhaps, new analogues.

Spinosyn A (SPA, 1) is a polyketide-derived macrolide produced by Saccharopolyspora spinosa, that is an active ingredient in several commercial insecticides.¹ The structures of SPA and its many analogues have a characteristic perhydro-as-indacene core, which is glycosylated by a rhamnose (see 2) at C-9 and a forosamine at C-17. Both the aglycone (AGL, 3) and the sugar appendages contribute to the observed activity of the spinosyns, among which SPA is most potent.² Alteration of the tetracyclic nucleus or removal of either deoxy sugars significantly diminishes the pesticidal activity. Even subtle structural variations, such as the methylation pattern of the rhamnose moiety in spinosyns, change the LD₅₀ by as much as >200-fold against Heliothis virescens.^2a In view of these results and the global market value of spinosyns as a general purpose insecticide, a detailed investigation of the biosynthesis of spinosyns is warranted. Since the rhamnose moiety is methylated to a different extent in various spinosyns,³ the study of how the corresponding methyltransferases are coordinated to achieve the desired level of methylation of rhamnose has been a focus of this research. We report herein the function and substrate specificity of the three methyltransferases involved in the methylation reactions, the preferred reaction sequence of their catalyzed reactions, and the likely regulation of permethylation of rhamnose in 1.

The spinosyn biosynthetic gene cluster had been cloned from S. spinosa.⁴ Sequence analysis and gene disruption experiments revealed that the spnH, spnI and spnK genes,⁴ all of which show high sequence identity to those encoding S-adenosyl-L-methionine (SAM) dependent methyltransferases (MTs), are involved in the O-methylation of rhamnose in 1.^4,5 As illustrated in Scheme 1, methylations may take place before (2 → 4 → 6, route A) or after (3 → 5 → 6, route B) the attachment of rhamnose to the aglycone (AGL, 3). It is also possible that methylations are the final tailoring steps after both sugars have been coupled to the aglycone (3 → 5 → 7 → 1, route C). These methylation reactions are also not necessarily consecutive and may occur between the glycosylation steps catalyzed by SpnG and SpnP. To biochemically verify the functions of the MTs involved and to distinguish the possible methylation pathways, we studied the catalytic properties of the spnH, spnI and spnK gene products and validated their roles in the permethylation of 1.

Scheme 1.

The genes were amplified from S. spinosa genomic DNA and individually cloned into pET28b(+) expression vectors.⁶ Each was subsequently heterologously expressed in Escherichia coli BL21 star (DE3) or Rosetta II (DE3) host cells and was purified as an N-terminal His₆-tagged protein by affinity chromatography using Ni-NTA resin.⁶ Interestingly, the purified SpnH, SpnI or SpnK (20 μM each), upon incubation with 1 mM of 2 and 5 mM of SAM in 50 mM Tris•HCl buffer (pH 7.5) at 30 °C, failed to methylate TDP-L-rhamnose (TDP-R, 2), which was prepared from TDP-D-glucose as previously described via the actions of RfbB (a 4,6-dehydratase), RfbC (a 3,5-epimerase), and RfbD (a 4-reductase) from Salmonella enterica LT2.⁷ This result clearly ruled out route A as a possible biosynthetic pathway (Scheme 1).

To check whether route B is operative, the presumed substrate, R-AGL (5), was chemoenzymatically prepared using SpnG, which catalyzes the transfer of rhamnose moiety from 2 to the aglycone (AGL, 3).⁸ The competence of R-AGL (5) as substrate for SpnH, SpnI and SpnK was evaluated under the aforemen-tioned assay conditions (vide supra).⁶ In contrast to the assays with TDP-R (2) alone, HPLC analysis showed that all three MTs, in the presence of SAM, could convert 5 to new products (Figure 1). These new products (I, II, III) were isolated from preparative-scale incubation mixtures, and characterized by 2-D NMR spectroscopy and high-resolution mass spectrometry.⁶ As depicted in Figure 1, the regiospecificity of methylation reaction can be readily determined by comparing the chemical shifts of the rhamnose signals in 5 versus those in the products, and the heteronuclear long-range correlations between the newly introduced methyl carbon and the nearest O-methine proton. These data firmly establish that 5 is a common substrate for SpnH, SpnI, and SpnK, which are the respective rhamnose 4′-, 2′-, and 3′-O-methyltransferase. ^4,5 Since several MTs involved in O-methylation of deoxysugars (e.g., TylE/F and MycE)^9c,d are known to require divalent metals for optimal activities, the effects of Mg²⁺ ion on the activity of SpnH, SpnI, and SpnK were also examined. Monitoring the time course of product formation in each enzyme reaction clearly showed that the inclusion of Mg²⁺ in the incubation increases the rate of the reaction (~2-7-fold) in all cases.⁶

Figure 1.

HPLC analysis showing formation of compounds I, II and III upon incubation of (1) R-AGL (5) with (2) SpnH, (3) SpnI, and (4) SpnK. The pertinent chemical shifts of I, II and III are shown with arrows denoting the heteronuclear long-range correlation between the methyl carbon (in red) and the nearest O-methine proton (in blue). See Supporting Information for details.⁶

Having determined that SpnH, SpnI, and SpnK recognize and methylate R-AGL (5), we proceeded to investigate whether these enzymes can also accept the monomethylated compounds (I, II, III) as substrates. As shown in Figures 2a, 2b, 2c, SpnH is able to modify II and III to produce the corresponding 2′,4′- and 3′,4′-di-O-methylated products (IV and VI) under the above assay conditions. In contrast, SpnI can methylate I to give IV, but it is unable to accept III as a substrate. Similarly, SpnK methylates II to produce the 2′,3′-di-O-methylated product (V), but fails to modify I. Addition of Mg²⁺ ion in the incubation had no effect on the outcomes. These observations show that formation of three doubly methylated products (IV, V and VI) is possible, and also reveal that SpnI and SpnK do not tolerate the presence of a methoxy moiety at O-3′and O-4′positions of the rhamnose unit, respectively. To explore whether compounds IV, V and VI can be converted to the permethylated product PSA (6) by the appropriate MT, further experiments were carried out using these di-O-methylated intermediates in the incubation reactions. If the inability of SpnI and SpnK to process rhamnose carrying a methoxy group at C-3′ and/or C-4′ is an intrinsic property of these catalysts, the synthesis of PSA (6) can only be achieved by SpnH through intermediate V. As predicted (see Figures 2d, 2e, 2f), SpnI and SpnK were unable to process VI and IV, whereas SpnH successfully methylated the O-4′position of V resulting in the formation of PSA (6).

Figure 2.

HPLC analysis of methyl transfer reactions using monomethylated (I, II and III) and dimethylated (IV, V, and VI) substrates. a. Incubation of I (1) I standard, with (2) SpnI, and (3) SpnK; b. incubation of II (1) II standard, with (2) SpnH, and (3) SpnK; c. incubation of III (1) III standard, with (2) SpnH, and (3) SpnI; d. incubation of IV (1) IV standard, with (2) SpnK; e. incubation of V (1) V standard, with (2) SpnH; e. incubation of VI (1) VI standard, with (2) SpnI.

The six possible biosynthetic permethylation sequences of rhamnose in 5 are summarized in Scheme 2. Because di-O-methylated intermediates IV and VI are not processed by SpnK and SpnI, respectively, and the mono-O-methylated intermediate III is not a substrate for SpnI, permethylation of rhamnose must begin with SpnI, followed by SpnK, and end with SpnH (5 → II → V → 6). Since the great majority (> 95%) of spinosyns (spinosyn A and D) produced by S. spinosa bear permethylated rhamnose at C-9, reactions resulting in the accumulation of products with mono- and di-O-methylated rhamnose³ (such as I, II, III, IV, VI) are likely aborted paths. In view of the occurrence of various routes leading to dead-end intermediates (such as IV and VI) in vitro, the action of the three rhamnose MTs are likely rigorously orchestrated during spinosyn production in vivo to prevent the build up of dead-end intermediates.

Scheme 2.

The location of the encoding genes, spnH, spnI and spnK, of these MTs on different operons (see Scheme 1) suggests that differential expression of these MT genes may be a mechanism to regulate rhamnose methylations during spinosyns in vivo production, and minimize the formation of dead-end intermediates. This hypothesis could be tested in vitro by varying the concentration of each MT in the assays.⁶ As shown in Figure 3-(1), when equal concentration (20 μM) of three MTs are used in the incubation mixture, PSA (6) is the major product (75%). However, a significant amount of IV (25%) is also produced. The absence of the other dead-end intermediate, VI, in the reaction mixture is interesting. Since SpnK cannot process I, the only route leading to VI (5 → III → VI) requires SpnK and SpnH working in sequence. The fact that neither III was accumulated suggests that this path might be prevented by the lower catalytic efficiency of SpnK toward R-AGL (5) than with either or both SpnH and SpnI. Formation of IV can instead be accomplished via two routes involving either I or II as the intermediate. In both cases, SpnH plays a key role. In fact, the competition between SpnH and SpnK for II could determine the flux of the reaction to either the accumulation of IV or the production of V then PSA (6). On the basis of the results shown in Figure 3-(1) where IV is a significant byproduct, it was envisioned that reduction of the SpnH concentration in the incubation mixture might facilitate the formation of more PSA (6) than IV by decreasing the flux of 5 → I → IV and that of II → IV. Indeed, PSA (6) became almost the exclusive product when the concentration of SpnH was lowered to 2 μM and those of SpnI and SpnK were raised to 40 μM (Figure 3-(2) and (3)).

Figure 3.

Product distribution change upon the variation of enzyme concentration. Experiments were performed in duplicate (error bars represent standard deviations). In all cases, the substrate, R-AGL (5) was completely consumed and only negligible amounts (< 0.5%) of monomethylated products (I, II, III) were detected.

Alternatively, the product composition can be altered so that the formation of dimethylated products is favored (5 → I → IV or 5 → III → VI) by suppressing the SpnI-catalyzed conversion of 5 → II. Interestingly, little effect was discernible when the SpnI concentration was reduced 4-fold as compared to those of SpnH and SpnK (Figure 3-(4)). It appears that the preference of SpnI toward R-AGL (5) is much higher than those of the other two MTs. To gain support for this idea, the SpnI concentration was reduced to 20-fold lower than SpnH and SpnK. As anticipated, suppression of PSA synthesis (64%) is now evident and the production of both IV (16%) and VI (20%) are also observed (Figure 3-(5)). From these results, one can conclude that the majority of IV produced in the presence of sufficient SpnI (Figure 3-(1) and (4)) is derived via the route involving II (5 → II → IV). Therefore, it would also be possible to suppress the formation of IV by increasing the concentration of SpnK to enhance the flux from II to V. In accord with this prediction, when we increased the SpnK concentration and kept SpnH and SpnI concentrations low, the exclusive formation of PSA (6) was observed (Figure 3-(6)).

Previously, the permethylation of rhamnose in the elloramycin pathway had also been studied. The three MTs (ElmMI, ElmMII, and ElmMIII) identified were shown to be regiospecific at O-2′, O-3′, and O-4′of rhamnose, respectively.^9e The order of the methylation events (ElmMI → ElmMII → ElmMIII) was also proposed. However, these results, obtained by in vitro cell-freee-extracts assays and in vivo deletion/complementation assays, remained to be biochemically validated. The experiments described herein unambiguously establish that SpnH, SpnI, and SpnK are the corresponding 4′-, 2′-, and 3′-O-methyltransferase responsible for the permethylation of rhamnose during spinosyn biosynthesis.^4,5

Investigation of the sequence of the methylation events in the spinosyn case revealed that only one route is productive (5 → II → V → 6) for the permethylation of the rhamnose moiety. Moreover, the completion of rhamnose permethylation is likely achieved by the proper control of the expression levels of the methyltransferase genes involved. Specifically, it was demonstrated that the production of PSA (6) could be enhanced by reducing the cellular concentration of SpnH and/or by increasing the SpnK concentration. Our results also showed that the methylation reactions do not take place prior to the SpnG reaction (route A, Scheme 1), but instead occur after rhamnose has been coupled to the aglycone (route B). Recently, the heterologously expressed SpnH was shown to be capable of converting 4′-O-desmethyl-SPA (spinosyn K, 8) to SPA (1), indicating that SpnH is the 4′-O-methyltransferase and route C is also feasible.¹⁰

The methylation of deoxysugars found in various macrolides is a common modification and known to enhance the metabolic stability and often the biological activity of the parent compounds. Several MTs responsible for such processes have been studied.⁹ The permethylation reported herein represents one of the most complex post-glycosylation modifications of a deoxysugar in secondary metabolites. The methylation sequence is established by reconstituting the pathway with purified enzymes, and such information is otherwise inaccessible by genetic or bioinformatic experiments. Thus, this work not only provides valuable insight into the rhamnose permethylation events during spinosyn biosynthesis, but also highlights the significance of the biochemical approach in deciphering natural product biosynthetic pathways. Our results set the stage for future exploitation of spinosyn biosynthetic pathway to produce targeted spinosyn derivatives and, perhaps, new analogues.