Combinatorial biosynthesis and the basis for substrate promiscuity in class I diterpene synthases

Abstract

Terpene synthases are capable of mediating complex reactions, but fundamentally simply catalyze lysis of allylic diphosphate esters with subsequent deprotonation. Even with the initially generated tertiary carbocation this offers a variety of product outcomes, and deprotonation further can be preceded by the addition of water. This is particularly evident with labdane-related diterpenes (LRDs) where such lysis follows bicyclization catalyzed by class II diterpene cyclases (DTCs) that generates preceding structural variation. Previous investigation revealed that two diterpene synthases (DTSs), one bacterial and the other plant-derived, exhibit extreme substrate promiscuity, but yet still typically produce exo-ene or tertiary alcohol LRD derivatives, respectively (i.e., demonstrating high catalytic specificity), enabling rational combinatorial biosynthesis. Here two DTSs that produce either cis or trans endo-ene LRD derivatives, also plant and bacterial (respectively), were examined for their potential analogous utility. Only the bacterial trans-endo-ene forming DTS was found to exhibit significant substrate promiscuity (with moderate catalytic specificity). This further led to investigation of the basis for substrate promiscuity, which was found to be more closely correlated with phylogenetic origin than reaction complexity. Specifically, bacterial DTSs exhibited significantly more substrate promiscuity than those from plants, presumably reflecting their distinct evolutionary context. In particular, plants typically have heavily elaborated LRD metabolism, in contrast to the rarity of such natural products in bacteria, and the lack of potential substrates presumably alleviates selective pressure against such promiscuity. Regardless of such speculation, this work provides novel biosynthetic access to almost 19 LRDs, demonstrating the power of the combinatorial approach taken here.

1. Introduction

While often associated with highly complex cyclization and rearrangement reactions, as suggested by their nomenclature terpene synthases (TPSs) do not necessarily catalyze such reactions. Essentially, their basic catalytic function is simply lysis of the allylic diphosphate ester, accomplished with assistance from a trio of divalent magnesium (Mg²⁺) co-factors bound by conserved DDxxD and (N/D)Dxx(S/T)xxxE motifs, as well as conserved basic residues, followed by deprotonation (Christianson, 2008). Indeed, a significant portion of the complexity associated with TPS-mediated reactions can be attributed to the inherent reactivity of their isoprenoid substrates (Tantillo, 2017). The simplest TPS reaction is direct deprotonation of the initially generated tertiary carbocation to generate an olefin, although preceding addition of water can lead to formation of a tertiary alcohol as well. There are 3 potential olefin products, the exo-ene generated by deprotonation of the neighboring methyl, along with the cis and trans variants of the endo-ene generated by deprotonation of the neighboring methylene, which depends on the orientation of the allylic diphosphate isoprenyl unit relative to the rest of the precursor (Fig. 1).

Fig. 1.

Simple lysis (of the allylic diphosphate ester) reactions catalyzed by TPSs with variable length isoprenyl precursors (R = one or more isopentenyl units). Direct deprotonation of the initially formed tertiary carbocation intermediate leads to the depicted range of products, with configuration of the newly formed internal endo C=C presumably dependent on initial substrate conformation (as shown).

Diterpenes, composed of four isoprenyl units, are generally derived from (E,E,E)-geranylgeranyl diphosphate (GGPP, 1), although it has been shown that the cisoid analog (Z,Z,Z)-nerylneryl diphosphate (NNPP, 2) also can serve as a precursor (Zi et al., 2014b). Such metabolism is particularly prevalent in plants, where GGPP is required to form photosynthetic pigments (i.e., the phytol side-chain of chlorophyll as well as carotenoids). Moreover, GGPP must be cyclized to produce the gibberellin A (GA) phytohormone required for normal growth and development in all vascular plants. The relevant DTC and DTS, which produce 8(17)-ene(exo) copalyl diphosphate (CPP) and kaurene, are termed CPP synthases (CPSs) and kaurene synthases (KSs), respectively. These have given rise to diversified families of DTCs and DTSs (these latter sometimes referred to as KS-like or KSL) in many plant species, with the DTSs further serving as the ancestors of the plant TPS family more generally (Zi et al., 2014a). Accordingly, GA serves as the ancestral LRD and has given rise to extensive such metabolism in the plant kingdom. By contrast, LRD biosynthesis is only intermittently found, even rare, in microbes, such as fungi and bacteria (Zi et al., 2014a).

The LRDs are defined by the initiating bicyclization reaction mediated by DTCs (Peters, 2010). These catalyze protonation of the terminal carbon-carbon double bond (C=C), with anti addition of the two internal C=C leading to formation of the eponymous labda-13E-en-8-yl⁺ diphosphate intermediate. This intermediate can be formed in four different stereochemical configurations, depending on the initial conformation of the substrate, which can be distinguished by the configuration of carbons 9 and 10 (C9 and C10; see Fig. 2 for numbering). These are traditionally designated as normal (9S,10S), ent- (9R,10R), syn- (9R,10S) or ent-syn- (9S,10R) - e.g., the gibberellins are derived from the ent stereoisomer. Note that the decalin bridgehead configuration (i.e., relative orientation of the C5-hydrogen and C10-methyl) is always trans in this intermediate. Immediate deprotonation leads to formation of the corresponding stereoisomer of CPP. However, this intermediate also can undergo 1,2-shifts of the hydride and methyl substituents of the tertiary and quaternary carbons (respectively) in the decalin bicycle, creating a series of tertiary carbocations. The first methyl shift creates the halimane backbone (halima-13E-en-10-yl⁺ diphosphate), and the second generates the clerodane backbone, also referred to as kolavane (kolava-13E-en-4-yl⁺ diphosphate), which is used here to distinguish the resulting product (KPP) from CPP (Fig. 2). Note that this final 1,2-shift can occur with either methyl substituent of C4, so the final decalin bridgehead (C5,10) configuration can be either cis or trans, adding further stereochemical variation. Each carbocation can be deprotonated at alternative positions, such that each of these three basic backbones (labdane, halimane and clerodane) can be produced as 2–3 distinct olefins (i.e., C=C positioning), yielding isomers of CPP, HPP or KPP (respectively). In addition, each tertiary carbocation intermediate can undergo addition of water prior to deprotonation, yielding the corresponding hydroxylated derivative. Moreover, it has recently been demonstrated that DTCs can mediate ring rearrangement of the decalin bicycle as well (Xu et al., 2018). Accordingly, there are a wide variety of potential DTC products, although enzymes for only 19 are currently known (Tables 1 and S1). These DTC products serve as intermediates that undergo further biosynthetic elaboration, generally initiated by DTSs.

Fig. 2.

Basic DTC products from bicyclization and subsequent rearrangement (PP=diphosphate). Also shown are known stereoisomers for the initial bicycle, with derived products for which DTCs are known indicated by superscript (ⁿnormal, ^eent, ^ssyn, no ent-syn have yet been identified).

Table 1.

Potential substrates and the relevant synthases used in this study.

No.a	Nameb	DTC/IDSc	Origin	Reference
1	GGPP	GGPPS	Abies grandis	(Burke and Croteau, 2002)
2	NNPP	NNPPS	Solanum lycopersicum	(Zi et al., 2014b)
3	FPP	FPPS	Saccharomyces cerevisiae	(Martin et al., 2003)
4	GFPP	GFPPS	Arabidopsis thaliana	(Nagel et al., 2015)
5	copalyl diphosphate (CPP)	AgAS:D621A	Abies grandis	(Peters et al., 2001)
6	ent-CPP	An2/ZmCPS2	Zea mays	(Harris et al., 2005)
7	syn-CPP	OsCPS4	Oryza sativa	(Xu et al., 2004)
8	7-endo-CPP	SmCPS/KSL1:D501A/ D505A	Selaginella moellendorffii	(Mafu et al., 2011)
9	ent-7-endo -CPP	SdCPS2:C359F/W360S	Salvia divinorum	(Pelot et al., 2017)
10	8α-hydroxy-CPP	NgCLS	Nicotiana glutinosa	(Criswell et al., 2012)
11	8β-hydroxy-ent-CPP	AtCPS:H263A	Arabidopsis thaliana	(Potter et al., 2014)
12	peregrinol diphosphate (9α-hydroxy-CPP)	MvCPS1	Marrubium vulgare	(Zerbe et al., 2014)
13	kolavenyl diphosphate (KPP)	Haur_2145	Herpetosiphon aurantiacus	(Nakano et al., 2015)
14	ent-KPP	AtCPS:H263Y	Arabidopsis thaliana	(Potter et al., 2016b)
15	terpentedienyl diphosphate (syn-KPP)	KgTPS	Kitasatospora griseola	(Nakano et al., 2010)
16	tuberculosinyl diphosphate (halimadienyl diphosphate, HPP)	MtHPS	Mycobacterium tuberculosis	(Mann et al., 2009)
17	syn-HPP	OsCPS4:H561D	Oryza sativa	(Potter et al., 2016a)
18	syn-halima-5 (10),13E-dienyl diphosphate	MvCPS1:W323F/F505Y	Marrubium vulgare	(Mafu et al., 2016)
19	mutildienyl diphosphate	CpPS:D649L	Clitopilus passeckerianus	(Xu et al., 2018)

^aNumbering as defined in the text.

^bPreviously assigned common names, with semi-systematic names (as defined in the text) also given where necessary.

^cFull names of these synthases can be found in the abbreviation list.

While DTS can catalyze elaborate reactions, as exemplified by that mediated by KSs, these require cyclization and, thus, rely on precise positioning of the internal C=C or hydroxyl group relative to the initially generated (tertiary) carbocation (Christianson, 2008). By contrast, the simplest (non-cyclizing) TPS reactions described above only require steric isolation of the allylic diphosphate isoprenyl unit. Accordingly, such DTSs might exhibit less stringent substrate binding - here termed substrate promiscuity, with fidelity in product outcome referred to as catalytic specificity, as previously defined (Hult and Berglund, 2007). Particularly if these DTSs exhibit significant substrate promiscuity yet are still catalytically specific, they can predictably build on the structural complexity mediated by the preceding DTCs, offering the possibility of rational combinatorial biosynthesis to generate substantial additional variety. This has been partially realized in previous work that identified two such DTSs that orthogonally catalyze two of the possible ‘simple’ product outcomes. In particular, the terpentetriene synthase from the bacterium Kitasatospora griseola (Dairi et al., 2001; Hamano et al., 2002), termed here KgTS, and sclareol synthase from the plant Salvia sclarea (Caniard et al., 2012; Schalk et al., 2012), termed here SsSS, which were found to react with a wide range of DTC products (i.e., all 12 known at that time), generally yielding the exo-ene or tertiary alcohol LRD derivatives (i.e., 13(16)-ene or 13-ol), respectively (Jia et al., 2016). This work was enabled by a previously developed modular metabolic engineering system (Cyr et al., 2007), which enables facile co-expression in Escherichia coli of combinations of the relevant isoprenyl diphosphate synthase (IDS) and/or DTC and DTS, with additional engineering to increase metabolic flux to the upstream isoprenoid precursors (Morrone et al., 2010), such that sufficient quantities of the resulting product can be readily isolated for de novo structural characterization. Here this approach was used to further investigate the utility of additional DTSs for such rational combinatorial biosynthesis by examining potential general production of either cis- or trans-endo-ene LRD derivatives, as well as to investigate the basis for promiscuity via similar analysis of a variety of other DTSs, testing not only the acyclic 20-carbon (20C) precursors 1 and 2, but also the transoid 15C precursor (E,E)-farnesyl diphosphate (FPP, 3) and 25C precursor (E,E,E,E)-geranylfarnesyl diphosphate (GFPP, 4), along with an expanded arsenal of 15 DTCs with distinct products (5 – 19; Table 1).

2. Results and Discussion

2.1. Further examination of KgTS and SsSS promiscuity

It was previously reported that KgTS will react with not only the 20C transoid isoprenyl diphosphate precursor 1, but also the transoid 15C acyclic precursor FPP (3), at least in vitro (Hamano et al., 2002). In the E. coli metabolic engineering system endogenous phosphatases dephosphorylate isoprenyl diphosphate precursors, yielding the corresponding primary alcohol derivative (designated here by prime notation of the corresponding compound number - i.e., 1’ - 19’), which are extractable and observable by the gas-chromatograph with mass spectral (GC-MS) detection analytical method utilized here. To further examine the promiscuity of KgTS, as well as SsSS, in the context of this system it seemed worth investigating their ability to react with transoid acyclic precursors that differ in length by a single isoprenyl unit - i.e., both the shorter 3 and longer 4. This was carried out via co-expression with IDSs that produce either 3 or 4 (i.e., FPP synthase, FPPS, or GFPP synthase, GFPPS). As the native (plant) genes contain N-terminal plastidial targeting peptides that are removed after import in planta, the recombinant genes used here for GFPPS and SsSS are truncated to remove the corresponding sequence and, thus, encode pseudo-mature enzymes. Note that analogous pseudo-mature constructs are used for the plant-derived IDSs that produce either 1 or 2 (i.e., GGPP synthase, GGPPS, or NNPP synthase, NNPPS), as well as all plant-derived DTSs and DTCs.

In this metabolic engineering system both KgTS and SsSS were found to react with the shorter 3, albeit somewhat inefficiently (i.e., at least relative to the E. coli phosphatases, as less than half of the observed sesquiterpenoids result from DTS activity, with the remainder representing the primary alcohol derivative produced by the phosphatases), but only KgTS reacts with 4 and does so reasonably efficiently. Although it was previously reported that only SsSS reacted with the cisoid C20 acyclic precursor 2 (Jia et al., 2016), the greater promiscuity observed for KgTS with 4 prompted reexamination of its reactivity with the cisoid 2 - i.e., via co-expression with NNPPS (Zi et al., 2014b). Indeed, extending the data collection time to cover earlier eluting compounds revealed that KgTS does react with 2, actually significantly more efficiently than does SsSS (Figs 3 and S1).

Fig. 3.

Extended promiscuity of SsSS and, particularly, KgTS. GC–MS chromatograms (either total or indicated extracted ion count, TIC or EIC, respectively) of extracts from E. coli engineered for production of a potential substrate by introduction of either the combination of a relevant DTC and GGPPS, or an alternative IDS only (Table 1), along with KgTS or SsSS, as indicated (peaks are labeled by compound numbers, as described in the text, with those corresponding to the dephosphorylated substrate derivatives indicated by prime’ notation).

As previously reported from in vitro assays (Hamano et al., 2002), with 3 KgTS seems to produce a mixture of the three C=C isomers of farnesene (20 - 22), while SsSS more specifically produces the tertiary alcohol (E)-nerolidol (23), as verified by comparison via GC-MS of retention time (RT) and mass spectra (MS) to an authentic standard (Fig. S2). With 4 KgTS produces a mixture of three sesterterpenes assumed to be the C=C isomers of geranylfarnesene (24 - 26), although the predominant 25, based on relative RT and MS, appears to be the trans-endo-ene, commonly referred to as the (E)-α isomer (Fig. S2). With 2 KgTS yields two products, the major of which, upon isolation and structural analysis by NMR (Figs. S3–S5 and Table S2), was found to be the expected exo-ene containing derivative of 2, β-nerylmyrcene (27). However, while another diterpene was observed as a minor product, surprisingly, upon isolation and structural analysis by NMR (Figs S6–S8 and Table S3), this was found to be β-springene (28). It is assumed that this is derived from production of (E,E,Z)-geranylneryl diphosphate from the endogenous FPP (3, from the E. coli host) by the NNPPS (with the putative dephosphorylated derivative of this indicated by P in Fig. 3). Regardless, the KgTS activity found here provides novel biosynthetic access to 27 and 25 (albeit alongside small amounts of C=C isomer co-products in each case).

As previously reported (Jia et al., 2016), although KgTS and SsSS react more efficiently with 1 than the endogenous phosphatases from E. coli, they cannot compete with DTCs. This enables facile examination of their ability to react with DTC produced bicyclic isoprenyl diphosphate precursors (i.e., by simple co-expression of the relevant DTC, as well as GGPPS). Such an approach was taken here to examine the specificity of KgTS and SsSS with three additional DTC products. These are an ent- version of 7-endo-CPP (ent-7-endo-CPP, 9), a novel example of the decalin ring with bridgehead C=C, syn-halima-5(10),13E-dienyl diphosphate (18), and, of particular interest, a 5–6 bicycle resulting from ring rearrangement (of the halima-13E-en-5-yl⁺ diphosphate intermediate) such that the product no longer has the prototypical decalin core, mutildienyl diphosphate (19). Both KgTS and SsSS react more efficiently than the endogenous E. coli phosphatases with 9 and 18, but only KgTS does so with 19, indicating that the bacterial KgTS exhibits more substrate promiscuity than the plant-derived SsSS (Figs 3 and S1).

Perhaps not surprisingly, the products of KgTS and SsSS with 9 appear to be enantiomers of their products with the normal stereoisomer (i.e., 7-endo-CPP, 8), and, on the basis of their identical RT and MS in (non-chiral) GC-MS analysis, were assigned as the expected exo-ene derivative ent-labda-7,13(16),14-triene (29) and tertiary alcohol derivative ent-labda-7,14-dien- 13-ol (30), respectively (Fig. S9). The KgTS and SsSS products with 18 were unknown and it was necessary to isolate these for de novo structural analysis by NMR, which determined that these are the expected exo-ene derivative (Figs. S10–S12 and Table S4), syn-halima-5(10),13(16),14-triene (31), and tertiary alcohol derivative (Figs. S13–S15 and Table S5), syn-halima-5(10),14-dien-13-ol (32), respectively. Although SsSS produces only 32, KgTS produces both, with actually slightly more 32 than 31. The KgTS product with 19 also was unknown, but upon isolation and de novo structural analysis by NMR was found to be the expected exo-ene derivative (Figs. S16–S18 and Table S6), termed here mutil-4(18),13(16),14-triene (33).

While 32 has been previously isolated from a plant extract (Nagashima et al., 2001), the relevant DTS is unknown. In addition, searches of the SciFinder database indicate that 29 - 31 and 33 do not appear to have been previously reported. Accordingly, the KgTS and SsSS activities reported here not only generally confirm their utility for rational combinatorial biosynthesis, but also provide novel biosynthetic access to 29 - 33 (albeit 31 is only produced in a mixture with 32).

2.2. Examining the promiscuity of endo-ene producing DTSs

As described above, there are four potential product outcomes with the simplest TPS reaction. Accordingly, in addition to the exo-ene and tertiary alcohol produced by KgTS and SsSS (respectively), it should also be possible generate endo-ene derivatives, in either the cis or trans configuration. Indeed, there is a DTS known to catalyze each of these outcomes. In particular, the cis-abienol synthase from the plant Abies balsamea (Zerbe et al., 2012), termed here AbCAS, which produces the cis-endo-ene derivative of its native substrate 8α-hydroxy-CPP (10), and a labdatriene synthase from the bacterium Streptomyces cyslabdanicus K04–0144 (Yamada et al., 2016), termed here ScLS, which produces the trans-endo-ene derivative of its native substrate CPP (5). Note that AbCAS is bifunctional, with both DTC and DTS activity, but a variant with a D405A substitution that negates DTC activity - i.e., this AbCAS construct only exhibits DTS activity - has been reported (Zerbe et al., 2012), and was used here.

Both AbCAS and ScLS were tested with all the potential substrates already examined with KgTS and SsSS (i.e., 1 - 19). Strikingly, the plant-derived AbCAS was found to be quite specific, reacting efficiently only with its native substrate 10, with the only other accepted precursor being the structurally closely related 5, and even this reacts relatively poorly (Fig. 4 and Table S7). By contrast, the bacterial ScLS is highly promiscuous, reacting with all the potential substrates except the longer (25C) 4 (Fig. 5 and Table S7). While ScLS reacts relatively poorly with the acyclic substrates (i.e., 1 - 3) and the 5–6 bicycle 19, it efficiently out-competes the E. coli phosphatases with all the decalin containing substrates (i.e., 5 - 18).

Fig. 4.

Limited promiscuity of AbCAS. GC–MS chromatograms of extracts from E. coli engineered for production of the indicated potential substrate by co-expression of GGPPS and a relevant DTC (Table 1), along with AbCAS, as indicated (peaks are labeled by compound numbers, as described in the text, with those corresponding to the dephosphorylated substrate derivatives indicated by prime’ notation). Only the two examples exhibiting conversion to product(s) are shown, no turnover was observed with all other potential substrates.

Fig. 5.

Extreme promiscuity of ScLS. GC–MS chromatograms of extracts from E. coli engineered for production of a potential substrate by introduction of either the combination of a relevant DTC and GGPPS, or an IDS only (Table 1), along with ScLS, as indicated (peaks are labeled by compound numbers, as described in the text, with those corresponding to the dephosphorylated substrate derivatives indicated by prime’ notation).

The observed MS for the products of AbCAS and ScLS with their native substrates are consistent with the previously reported activity - i.e., the production of cis-abienol (labda-12Z,14-dien-8α-ol, 34) from 10 by AbCAS (Zerbe et al., 2012), and the production of labda-8(17),12E,14-triene (35) from 5 by ScLS (Yamada et al., 2016). Similarly, as previously reported (Yamada et al., 2016), ScLS efficiently reacts with 8 to specifically produce the trans-endo-ene derivative labda-7,12E,14-triene (36). The mass spectra for 34 – 36 are shown in the Supporting Information (Fig. S19).

In a number of cases the observed products seemed likely to correspond to the known products of KgTS, SsSS or other known DTSs, or be enantiomers of these, which was investigated by GC-MS based comparison of RT and MS. For example, with its acyclic substrates ScLS was found to specifically produce the expected trans-endo-ene derivative from 1, (E)-α-springene (37), as identified from the mix produced by KgTS (Nakano et al., 2010), but surprisingly specifically produce the same exo-ene derivative from 2 as KgTS (i.e., 27), and a similar mixture of the C=C isomers of farnesene (20 - 22) as KgTS, as well as the SsSS product (E)-nerolidol (23), from 3. Perhaps more interestingly, AbCAS yields two products from 5, with the less abundant one found to be the same exo-ene derivative sclarene (38) produced by KgTS with 5, while the major product was the expected cis-endo-ene derivative, (Z)-biformene (labda-8(17),12Z,14-triene, 39), as verified by comparison to the enantiomer known to be produced from 6 by KgTS (Fig. S20). Similarly, ScLS yields two products from 6, the enantiomer of its native substrate 5, with the major product being the exo-ene derivative ent-sclarene (40), as identified by comparison to the enantiomer produced by KgTS with 5 (Fig. S21), while the minor product is the known ent-kaurene (41). Interestingly, KgTS does not produce 40 from 6, but rather the cis-endo-ene derivative instead, such that ScLS and KgTS both exhibit unexpected yet orthologous activity with this substrate. With 7 ScLS also yields two products, which were found to be the same exo-ene derivative griseolaene (syn-labda-8(17),13(16),14-triene, 42) produced by KgTS, and the same tertiary alcohol derivative vitexifolin A (syn-labda-8(17),14-dien-13-ol, 43) produced by SsSS (Fig. S22). ScLS further yields two products from 9, and the major product was found to be the same exo-ene derivative 29 produced by KgTS, while the minor product was the expected trans-endo-ene derivative, ent-labda-7,12E,14-triene (44), as identified by comparison to the enantiomer produced by ScLS with 8 (Fig. S22). With 11 ScLS efficiently and selectively yields a single product, identified as the heterocyclic derivative ent-13-epi-manoyl oxide (45)(Fig. S22), as also produced by KSs from this hydroxylated derivative of their native substrate (Mafu et al., 2015). With 13 ScLS selectively produces the same tertiary alcohol derivative cleroda-3,14-dien-8-ol (46) as SsSS (Fig. S22). Finally, with 19 ScLS produces small amounts of the same exo-ene derivative 33 as KgTS, along with trace amounts of an unidentified diterpene (Fig. S22).

In the case of the remaining substrates, at least one of the observed ScLS products could not be identified by comparison to readily available diterpenes, requiring isolation and de novo structural analysis by NMR. For example, with 10 ScLS yields three products. By comparison to previously identified DTS products (Mafu et al., 2015), the two more abundant products were identified (Fig. S23) as manoyl oxide (47) and its C13 epimer, 13-epi-manoyl oxide (48), while the third required de novo structural analysis (Figs. S24–S26 and Table S8), and was identified as the expected trans-endo-ene derivative trans-abienol (labda-12E,14-dien-8α-ol, 49). With 12 ScLS yields two products and, upon de novo structural analysis (Figs. S27–S32 and Tables S9–S10), the major product was identified as syn-labda-9,13S-epoxy-14-ene (50), while the minor product was identified as the C13 epimer syn-labda-9,13R-epoxy-14-ene (51). With 14 ScLS yields a single product that, upon de novo structural analysis (Figs. S33–S35 and Table S11), was identified as the expected trans-endo-ene derivative ent-cleroda-3,12E,14-triene (52). With 15 ScLS yields two products, the predominant of which, upon de novo structural analysis (Figs. S36–S38 and Table S12), was identified as the expected trans-endo-ene derivative syn-cleroda-3,12E,14-triene (53), while the minor product was identified as the same exo-ene derivative syn-cleroda-3,13(16),14-triene (54) produced by KgTS (Fig. S23). With 16 ScLS yields a single product that, upon de novo structural analysis (Figs. S39–S41 and Table S13), was identified as the expected trans-endo-ene derivative halima-5,12E,14-triene (55). With 17 ScLS yields two products, the major of which was identified as the same exo-ene derivative syn-halima-5,13(16),14-triene (56) produced by KgTS (Fig. S23), while the minor product, upon de novo structural analysis (Figs. S42–S44 and Table S14), was identified as the expected trans-endo-ene derivative syn-halima-5,12E,14-triene (57). With 18 ScLS yields three products, the major of which, upon de novo structural analysis (Figs. S45–S47 and Table S15), was identified as the expected trans-endo-ene derivative syn-halima-5(10),12E,14-triene (58), while the minor products were found to be the same exo-ene derivative 31 produced by KgTS and tertiary alcohol derivative 32 produced by SsSS.

While 39 has been previously isolated (Noma et al., 1982), the relevant DTS is unknown. Moreover, although production of 50 and 51 was recently reported (Johnson et al., 2018), that for 40, 44, 52, 53, 55, 57 and 58 does not appear to have been previously reported. Accordingly, the AbCAS and, particularly, ScLS activities reported here provides novel access to 39, 40, 44, 52, 53, 55, 57 and 58 (albeit 39 is only produced inefficiently, and 39, 40, 44, 53, 57 and 58 are produced non-selectively).

It is evident that ScLS exhibits extreme substrate promiscuity and at least moderately specific catalytic activity, enabling (semi)-rational combinatorial biosynthesis as shown above. However, despite their catalysis of analogously simple TPS reactions, the substrate promiscuity of ScLS contrasts with the much greater specificity exhibited by AbCAS. Together with the greater promiscuity observed with KgTS versus SsSS this suggests the hypothesis that that bacterial DTSs might generally prove to be more promiscuous than those from plants.

2.3. Examining the phylogenetic basis for DTS promiscuity

To examine the hypothesis that substrate promiscuity depends more on phylogenetic origin than reaction mechanism, a variety of additional DTSs were chosen for investigation of their substrate specificity. Notably, this has already been examined to some extent with a number of plant DTSs. Although much of that work focused on selectivity for the various stereoisomers of (exo-)CPP (i.e., 5 - 7), more extensive studies also have been reported (Andersen-Ranberg et al., 2016; Zerbe et al., 2013). Of particular relevance here, it has been shown that the DTS activity of the enzymes most closely related to AbCAS (i.e., those from other gymnosperms) exhibit (exo-)CPP stereospecificity and will only react with their native substrate 5, but not 6 or 7. This includes not only the more closely related bifunctional abietadiene synthases (Peters et al., 2000), which catalyze more complex cyclization and rearrangement reactions, but also pimaradiene synthases that catalyze more straightforward simple cyclization reactions (Hall et al., 2013). Similarly, all the investigated KSs from plant gibberellin biosynthesis, which catalyze a highly complex (bi)cyclization with subsequent ring rearrangement reaction and yet are ancestral to all plant DTSs (Zi et al., 2014a), also exhibit CPP stereospecificity, reacting only with their native substrate, the enantiomer 6 (Cui et al., 2015; Heskes et al., 2018; Irmisch et al., 2015; Jackson et al., 2014; Kumar et al., 2016; Pelot et al., 2018; Shimane et al., 2014; Wu et al., 2012; Xu et al., 2007; Zerbe et al., 2014; Zerbe et al., 2013; Zhou et al., 2012). It has been reported that at least some plant pimaradiene synthases will react with two stereoisomers of CPP (Morrone et al., 2011; Pelot et al., 2018; Zhou et al., 2012), and a number of DTSs that catalyze cyclization reactions have been shown to readily react with the hydroxylated variant of their native substrate to carry out heterocyclization, forming various isomers of manoyl oxide (Andersen-Ranberg et al., 2016; Brückner et al., 2014; Ignea et al., 2015; Mafu et al., 2015; Pateraki et al., 2014). This latter group of DTSs includes the miltiradiene synthase from Salvia miltiorrhiza (Gao et al., 2009), termed here SmMS, which is closely related to SsSS (>60% amino acid, aa, sequence identity), but carries out both cyclization and rearrangement of the initially formed pimarane backbone (i.e., produces an abietane), and was chosen for broader analysis of substrate specificity. In addition, two plant KSs required for gibberellin biosynthesis, that from Arabidopsis thaliana (Yamaguchi et al., 1998), termed here AtKS, and rice (Sakamoto et al., 2004), Oryza sativa so termed here OsKS, as representative of dicots and monocots respectively, also were selected to reflect their ancestral role in evolution of the plant DTS family. While functionally analogous AtKS and OsKS share only ~42% aa sequence identity. There are relatively few bacterial DTSs and these have almost invariably just been coupled to the products of the DTC from the relevant biosynthetic operon, although a report on somewhat broader studies with KgTS in vitro prompted its use for combinatorial biosynthesis (Nakano et al., 2010). Hence, to examine a ‘simple’ cyclase, the pimaradiene synthase from the bacterium Salinispora arenicola (Xu et al., 2014), termed here SaPS, was selected, along with two bacterial KSs, both also involved in gibberellin biosynthesis, one from Bradyrhizobium japonicum (Morrone et al., 2009), termed here BjKS, and the other from Erwinia tracheiphila (Nagel and Peters, 2017), termed here EtKS, to examine the effect of increased reaction complexity and match the phylogenetic divergence of the pair of plant KSs (albeit these are more closely related, sharing ~51% aa sequence identity).

The substrate specificity of these six DTSs was investigated with all potential substrates 1 - 19 (Figs. 6–7; note that the product profiles for BjKS and EtKS are essentially identical, so only those for BjKS are shown here). Strikingly, it was immediately evident that the bacterial DTSs exhibit much greater promiscuity than those from plants, particularly when including all those investigated here, as, with the exception of SsSS, the plant derived DTSs react with only a couple of substrates, while those from bacteria react with almost all and do so reasonably efficiently with at least half of the decalin bicycle containing substrates (Tables 2 and S7). Indeed, in contrast to the plant KSs, which only efficiently react with their native substrate 6 and the hydroxylated variant 11, the bacterial KSs, if anything, are actually more promiscuous than SaPS, which catalyzes a simpler reaction. Nevertheless, the similarly limited substrate specificity exhibited by the plant-derived AtKS and OsKS, as well as promiscuity exhibited by the bacterial BjKS and EtKS, while distinct, is consistent with the analogous physiological function of each pair in the relevant biological kingdom. Intriguingly, while production of ent-kaurene requires complex (bi)cyclization and ring rearrangement, the promiscuity of the bacterial KSs indicates that this reaction need not be tightly chaperoned, which is consistent with the production of ent-kaurene by ScLS as well. Indeed, the ring rearrangement has been predicted to be concerted (albeit asynchronously) with secondary (tetra)cyclization (Hong and Tantillo, 2010). Nevertheless, this further emphasizes both the inherent reactivity of these isoprenyl reactants (Tantillo, 2017), and importance of catalytic base positioning to terminate the carbocationic (cascade) reactions initiated by TPSs (Pemberton and Christianson, 2016), as the lack of deprotonation of earlier intermediates presumably at least partially underlies the production of this complex diterpene by these promiscuous bacterial DTSs.

Fig. 6.

Limited promiscuity of plant DTSs. GC–MS chromatograms of extracts from E. coli engineered for production of the indicated potential substrate by co-expression of GGPPS and a relevant DTC (Table 1), along with a plant DTS, as indicated (peaks are labeled by compound numbers, as described in the text, with those corresponding to the dephosphorylated substrate derivatives indicated by prime’ notation). Only combinations exhibiting conversion to product(s) are shown, no turnover was otherwise observed.

Fig. 7.

Extreme promiscuity of bacterial DTSs. GC–MS chromatograms of extracts from E. coli engineered for production of the indicated potential substrate by co-expression of GGPPS and a relevant DTC (or GGPPS or FPPS alone), along with a bacterial DTS, as indicated (peaks are labeled by compound numbers, as described in the text, with those corresponding to the dephosphorylated substrate derivatives indicated by prime’ notation). Only combinations exhibiting conversion to product(s) are shown, no turnover was otherwise observed.

Table 2.

Relative reactivity of the ten DTSs and 19 potential substrates examined here.

substrate╲DTSa	SsSSb	AbCAS	SmMS	OsKS	AtKS	KgTSb	ScLS	SaPS	BjKS	EtKS
1	+ + +	-	-	-	-	+ + +	+ +	+ +	+	+ +
2	+ +	-	-	-	-	+ + +	+	-	-	-
3	+	-	-	-	-	+	+	-	+	+
4	-	-	-	-	-	+ +	-	-	-	-
5	+ + +	+	+++	-	-	+ + +	+++	+++	+ +	+ + +
6	+ + +	-	-	+++	+++	+ + +	+ + +	+ +	+++	+++
7	+ + +	-	+ + +	-	-	+ + +	+ + +	+ + +	+ + +	+
8	+ + +	-	-	-	-	+ + +	+ + +	+	+ + +	+ + +
9	+ + +	-	-	-	+	+ + +	+ + +	+ +	+ + +	+ + +
10	+++	+++	+ + +	-	-	+ + +	+ + +	+ + +	+ + +	+ + +
11	+ + +	-	-	+ + +	+ + +	+ + +	+ + +	+ + +	+ + +	+ + +
12	+ + +	-	-	-	-	+ + +	+ + +	+ +	+ + +	+ + +
13	+ + +	-	-	-	-	+ + +	+ + +	-	-	-
14	+ + +	-	-	-	-	+ + +	+ + +	-	+ + +	+ + +
15	+ + +	-	-	-	-	+++	+ + +	-	-	-
16	+ + +	-	-	-	-	+ + +	+ + +	-	-	-
17	+ + +	-	-	-	-	+ + +	+ + +	+	+	+ + +
18	+ + +	-	-	-	-	+ + +	+ + +	+	+	+ +
19	-	-	-	-	-	+ + +	+	-	-	-

“−” indicates substrate conversion percentage (P) < 10%; “+” 10% ≤ P < 40%; “++” 40% ≤ P < 70%; “+++” 70% ≤ P ≤ 100%; specific substrate conversion percentage values can be found in Fig. S1 and Table S7. Native substrate indicated by bold red text.

^aPlant DTSs indicated by green text and bacterial by blue text.

^bThe substrate conversion percentage values for SsSS and KgTS with substrates 5–8 and 10–17 were obtained from a previous report (Jia et al., 2016).

The observed mass spectra for the products of these DTSs with their native substrates are consistent with the previously reported native activity - i.e., the production of ent-kaurene (41) from 6 by all four KSs (Morrone et al., 2009; Nagel and Peters, 2017; Xu et al., 2007; Yamaguchi et al., 1998), production of miltiradiene (abieta-8,12-diene, 59) from 5 by SmMS (Gao et al., 2009), and production of isopimara-8,15-diene (60) from 5 by SaPS (Xu et al., 2014). In addition, as previously reported (Mafu et al., 2015), SmMS reacts with 10, the hydroxylated variant of its native substrate, and predominantly produces manoyl oxide (47), along with small amounts of 13-epi-manoyl oxide (48), while AtKS and OsKS react with 11, the hydroxylated variant of their native substrate, and specifically produce ent-13-epi-manoyl oxide (45). The mass spectra for 59 - 60 can be found in the Supporting Information (Fig. S48).

In a number of cases at least some of the observed products seemed likely to correspond to the known products of KgTS or SsSS, or be enantiomers of these, which was investigated by GC-MS based comparison of retention time and mass spectra. For example, AtKS reacts (albeit somewhat inefficiently) with 9, the 7-endo isomer of its native substrate, and the product was found to be the same ent-labda-7,13(16),14-triene (29) as identified here from KgTS. The bacterial DTS examined here (i.e., SaPS, BjKS and EtKS) also reacted with many substrates, generally producing similarly simple derivatives. For example, with 1 SaPS selectively produces the exo-ene derivative β-springene (28), while BjKS and EtKS produce both this and the tertiary alcohol derivative geranyllinalool (61). With 3 BjKS and EtKS selectively produce the tertiary alcohol derivative 23. With 5 BjKS and EtKS yield two products, one of which was identified as the tertiary alcohol derivative manool (62), and the other product was identified as pimara-7,15-diene (63) by comparison to its enantiomer, which is a known DTS product (Jia et al., 2017) (Fig. S49). With 6 SaPS yields two products, with the major one identified as the tertiary alcohol derivative ent-manool (64), while the minor product was identified as ent-pimara-8(14),15-diene (65), a known DTS product (Xu et al., 2007). With 8 SaPS, BjKS and EtKS all selectively produce the tertiary alcohol derivative labda-7,14-dien-13-ol (66). Analogous results were obtained with 9, as all three bacterial DTSs again selectively produce the tertiary alcohol derivative ent-labda-7,14-dien-13-ol (30). With 12 SaPS selectively produces the tertiary alcohol derivative viteagnusin D (syn-labda-14-en-9,13-diol, 67), while BjKS and EtKS also produce small amount of 67, they largely produce the exo-ene derivative syn-labda-13(16),14-dien-9-ol (68). With 14 BjKS and EtKS yield two products, one is the tertiary alcohol derivative ent-cleroda-3,14-dien-13-ol (69), and the other product is the exo-ene derivative ent-cleroda-3,13(16),14-triene (70). With 17 SaPS, BjKS and EtKS all produce the tertiary alcohol derivative syn-halima-5,14-dien-13-ol (71). Analogous results were obtained with 18, as all three bacterial DTSs again selectively produce the same tertiary alcohol derivative 32 as identified here with SsSS. The mass spectra for 64 - 71 can be found in the Supporting Information (Fig. S50).

In most other cases all the products have undergone cyclization. For example, with 7 BjKS and EtKS yield two products, with the major one identified as syn-pimara-7,15-diene (72), and the minor product as syn-stemodene (73), both of which are known DTS products (Morrone et al., 2006; Wilderman et al., 2004). With 10, while BjKS and EtKS produce an epimeric mixture of manoyl oxide, 47 and 48, SaPS selectively produces 47, much as described for other DTSs (Mafu et al., 2015). Conversely, with the enantiomer 11, SaPS produces an epimeric mixture of ent-manoyl oxide (74), with slightly more ent-13-epi-manoyl oxide (45) than 74, while BjKS and EtKS selectively produce 45, much like the plant KSs (Mafu et al., 2015). The mass spectra for 72 - 74 can be found in the Supporting Information (Fig. S51). Finally, with 7 SmMS yields a single product that required de novo structural analysis and, thus (Figs. S52–S54 and Table S16), was identified as syn-abieta-9(11),12-diene (75), while SaPS yields two products, the minor of which was 73, while the major product also required de novo structural analysis and, hence (Figs. S55–S57 and Table S17), was identified as syn-pimara-9(11),15-diene (76).

While 75 has been previously isolated (Sakurai et al., 1999), the relevant DTS is unknown. Moreover, 76 does not appear to have been previously reported. Accordingly, the DTS activities reported here provide novel biosynthetic access to 63, 75, and 76 (albeit 63 and 76 are produced non-selectively). More generally, while the bacterial DTS catalyzing more complex reactions exhibit high substrate promiscuity, this is coupled to weak catalytic specificity.

2.4. Examining the structural basis for DTS promiscuity

With the striking exception of SsSS, the plant DTSs examined here exhibit strong substrate selectivity. Much as previously observed with other plant DTSs (Andersen-Ranberg et al., 2016; Brückner et al., 2014; Ignea et al., 2015; Mafu et al., 2015; Morrone et al., 2011; Pateraki et al., 2014; Pelot et al., 2018; Zhou et al., 2012), these only react with structurally closely related isomers. For example, while they all seem to react with the hydroxylated variant of their native (exo-)CPP substrate (or the reverse in the case of AbCAS), only AtKS reacts with the corresponding stereoisomer of 7-endo-CPP (and relatively poorly at that), while here only SmMS reacts with an alternative stereoisomer of (exo-)CPP (although, as noted above, analogous relaxed but not fully promiscuous CPP stereoselectivity has been observed with a few plant pimaradiene synthases). By contrast, the bacterial DTSs uniformly exhibited substrate promiscuity, differing only in degree, regardless of reaction complexity (Table 2). Indeed, the BjKS and EtKS that catalyze the most complex reaction exhibit greater promiscuity than does SaPS, despite its simpler reaction mechanism. Nevertheless, a degree of selectivity is observed with the three bacterial DTSs that catalyze more complex reactions, which provides some insight into their substrate binding constraints. For example, the low catalytic efficiency observed with the 20C 1 and, for the bacterial KSs the 15C 3, versus the lack of reactivity with the C25 4, indicates the importance of compactness and rigidity imparted by DTC bicyclization of 1, while the lack of reactivity of the cisoid 2 indicates that the trans configuration of the C=C allylic to the diphosphate ester also is important. More specifically, the lack of reactivity with the 5–6 bicycle 19 highlights the role of the 6–6 fused decalin core. Notably, while the DTC products examined here all contain trans-decalin bicycles, based on the previous reports of SsSS promiscuity (Andersen-Ranberg et al., 2016; Jia et al., 2016), it has been shown to react with the products of several other newly identified DTCs (Johnson et al., 2018; Pelot et al., 2018), including syn-cis-endo-KPP, which indicates that the decalin bridgehead configuration is not critical for substrate recognition by promiscuous DTSs. On the other hand, despite their promiscuity with all available (stereo)isomers of CPP (5 - 9), and their hydroxylated variants (10 - 12), the three bacterial DTSs that catalyze more complex reactions also do not react with normal or syn-KPP (13 and 15), nor normal HPP (16), although they do react with the C=C isomers of syn-HPP (17 and 18), with varying efficiency, and the bacterial KSs with ent-KPP (14). Accordingly, it appears that the consecutive methyl migrations away from the labdane configuration found in CPP (i.e., to generate HPP and then KPP) progressively decrease catalytic efficiency, which is coupled to some stereospecificity (at least for the bacterial KSs), in certain cases leading to complete loss of reactivity.

Fortuitously, crystal structures for BjKS have been reported (Liu et al., 2014), which enabled more detailed computational investigation of the basis for its relaxed substrate selectivity observed here. Although the structure of a substrate bound form was reported, 6 was soaked into preexisting crystals that were grown under conditions with high concentrations of tartrate, which chelates metals, so the resulting structure does not contain the requisite Mg²⁺ co-factors. Perhaps not surprisingly, this structure further did not appear to have a fully closed active site, not least as it also is missing the loop between the J and K helices (J-K loop) that normally folds over the cavity. In order to model a catalytically competent form, the BjKS with 6 structure was used as a starting point, with the well-defined substrate and Mg²⁺ bound structure for bornyl diphosphate synthase used as the template (Whittington et al., 2002). This allowed generation of a completely enclosed active site into which the necessary three Mg²⁺ could be docked (Fig. S58), enabling further docking of three potential substrates with distinct reactivity. In particular, the native (fully reactive) substrate ent-CPP isomer 6, partially reactive syn-HPP isomer 17 and the unreactive syn-KPP isomer 15. The hydrocarbon backbones of each of these appears to be bound in quite different configurations (Fig. S59), which then affects diphosphate positioning and interactions with the Mg²⁺ co-factors and other polar contacts. Of particular interest is the number of contacts formed by a highly conserved arginine (Arg204), which is part of an extension of the second characteristic TPS motif - i.e., RLx(N/D)Dxx(S/T/G)xxx(E/D) - and is important for lysis of the diphosphate ester (Liu et al., 2014), as the number of contacts is sequentially decreased between these (potential) substrates in concert with their reactivity (Fig. 8). This then provides a rationale for the somewhat selective reactivity observed here with BjKS. Similarly, modeling with the functionally analogous but phylogenetically disparate EtKS (51% sequence identity) suggests that its analogous reactivity can be rationalized in the same fashion - e.g., their active site volumes are almost exactly identical, 1720 ų versus 1714 ų, respectively (Fig. S60).

Fig. 8.

Molecular docking study of BjKS with three representative substrates highlighting diphosphate positioning (phosphorus in orange and oxygen in red), with the bicyclic olefin substituent simply represented by a methyl group for ease of visualization - ent-CPP (6) in purple (A), syn-KPP (15) in yellow (B) and syn-HPP (17) in green (C). Residues involved in polar contacts with the phosphate group are shown as aqua-blue lines, Mg²⁺ ions shown as dark blue spheres, and polar contacts shown as red dotted lines.

3. Conclusions

Here the hypothesis that DTS catalyzing simple lysis and immediate deprotonation reactions in LRD biosynthesis would prove to exhibit substrate promiscuity with catalytic specificity and, hence, enable rational combinatorial biosynthesis, was investigated. In particular, to supplement the earlier finding of such utility for the bacterial exo-ene derivative producing KgTS and plant derived SsSS that produces tertiary alcohol derivatives, the plant derived AbCAS that naturally produces a cis-endo-ene derivative and bacterial ScLS that yields trans-endo-ene derivatives were targeted here. The bacterial ScLS was found to exhibit extreme substrate promiscuity with at least moderate catalytic specificity, approximating that observed with KgTS and SsSS, including an extended palette of potential substrates, together providing novel biosynthetic access to 15 LRDs (all the DTS products can be found in Tables 3 and S18, with the corresponding chemical structures found in Figs. 9 and S61). However, the plant derived AbCAS exhibited quite strict substrate specificity. In conjunction with previous reports of plant DTS substrate specificity, this suggested the alternative hypothesis that substrate promiscuity might be more tightly correlated with phylogenetic origin than reaction mechanism. Indeed, investigation of additional DTSs, including SmMS (which is closely related to SsSS), strongly supports this alternative hypothesis, with even the investigated bacterial KSs that catalyze a particularly complex (bi)cyclization and ring rearrangement reaction exhibiting a high degree of substrate and catalytic promiscuity, contrasting with the strict specificity found for the plant KSs. Analysis of substrate reactivity relationships, alongside docking studies with the structurally defined bacterial BjKS, provided some insight into the basis for the promiscuity found here, specifically suggesting that this is based on accommodation of potential substrates in reactive configurations, and further highlights the intrinsic reactivity of these isoprenoid substrates. Regardless of such mechanistic speculation, the results reported here suggest that bacterial DTSs will generally prove to be significantly more promiscuous than those from plants. This may be a function of the universal presence of extensive LRD metabolism, derived from the requisite biosynthesis of gibberellin hormones, in plants, which contrasts with the rarity of such metabolism in bacteria. As previously suggested (Tawfik, 2014), the absence of alternative substrates presumably alleviates selective pressure against promiscuity. Whatever the underlying rationale, it will be of interest to more thoroughly investigate the full array of potential substrates with not only the now known extremely promiscuous SsSS, KgTS and ScLS, but also bacterial DTSs more generally, to mine these for novel biosynthetic capacity.

Table 3.

Overview of the DTS products obtained from this study^a.

No.b	Semi-systematic namec	with stereo-definitions	Common named	Identificatione
20			β-farnesene	Prev. (Hamano et al., 2002)
21			(Z)-α-farnesene	Prev. (Hamano et al., 2002)
22			(E)-α-farnesene	Prev. (Hamano et al., 2002)
23			(E)-nerolidol	Comp. (commercial std.)
24			β-geranylfarnesene	Comp. (Sato et al., 2013)
25			(Z)-α-geranylfarnesene	This study
26			(E)-α-geranylfarnesene	This study
27			β-nerylmyrcene	This study
28			β-springene	Prev. (Jia et al., 2016)
29	ent-labda-7,13(16),14- triene	(9S,10R)-labda-7, 13(16),14-triene		This study
30	ent-labda-7,14-dien- 13-ol	(9S,10R)-labda-7,14- dien-13-ol		This study
31	syn-halima-5 (10), 13(16),14-triene	(8R, 9R) -halima-5( 10), 13(16),14-triene		This study
32	syn-halima-5(10), 14-dien- 13-ol	(8R,9R)-halima-5(10), 14-dien-13-ol		This study
33	mutil-4(18),13(16),14- trienef			This study
34	labda-12Z,14-dien-8α-ol,	(8R,9R,10S)-labda-12Z, 14-dien-8α-ol	cis-abienol	Prev. (Zerbe et al., 2012)
35	labda-8(17), 12E, 14- triene	(9S, 10S)-labda-8(17), 12E,14-triene		Prev. (Ikeda et al., 2016; Yamada et al., 2016)
36	labda-7,12E,14-triene	(9S,10S)-labda-7,12E, 14-triene		Prev. (Ikeda et al., 2016; Yamada et al., 2016)
37			(E)-α-springene	Comp. (Jia et al., 2016)
38	labda-8(17),13(16),14- triene	(9S, 10S)-labda-8 (17), 13(16),14-triene	sclarene	Comp. (Jia et al., 2016)
39	labda-8(17),12Z,14- triene	(9S,10S)-labda-8(17), 12Z,14-triene	(Z)-biformene	This study
40	ent-labda-8(17), 13(16),14-triene	(9R,10R)-labda-8(17), 13(16),14-triene	ent-sclarene	This study
41		(8S,9R,10R,13R)-kaur- 16-ene	ent-kaurene	Comp. (Morrone et al., 2009)
42	syn-labda-8(17), 13(16),14-triene	(9R,10S)-labda- 8(17),13(16),14-triene	griseolaene	Comp. (Jia et al., 2016)
43	syn-labda-8(17),14- dien-13-ol	(9R, 10S, 13S)-labda- 8(17),14-dien-13-ol	vitexifolin A	Comp. (Jia et al., 2016)
44	ent-labda-7,12E,14-triene	(9R,10R)-labda-7,12E, 14-triene		This study
45	ent-labda-8,13R- epoxy-14-ene	(8S,9S,10R,13R)-labda- 8,13-epoxy-14-ene	ent-13-epi-manoyl oxide	Comp. (Mafu et al., 2015)
46	cleroda-3,14 -dien- 13-ol	(5S,8S,9R,10S)-cleroda- 3,14-dien-13-ol	kolavelool	Comp. (Jia et al., 2016)
47	labda-8,13R-epoxy- 14-ene	(8R,9R,10S,13R)-labda- 8,13 -epoxy-14-ene	manoyl oxide	Comp. (Mafu et al., 2015)
48	labda-8,13S-epoxy- 14-ene	(8R,9R,10S,13S)-labda- 8,13 -epoxy-14-ene	13-epi-manoyl oxide	Comp. (Mafu et al., 2015)
49	labda-12E, 14-dien-8α-ol	(8R,9R, 10S)-labda- 12E,14-dien-8α-ol	trans-abienol	This study
50	syn-labda-9,13S- epoxy-14-ene	(8R,9R,10S,13S)-labda- 9,13 -epoxy-14-ene		This study
51	syn-labda-9,13R- epoxy-14-ene	(8R,9R, 10S, 13R)-labda- 9,13 -epoxy-14-ene		This study
52	ent-cleroda-3,12E, 14- triene	(5R,8R,9S,10R)- cleroda-3,12E, 14-triene		This study
53	syn-cleroda-3,12E, 14- triene	(5S,8R,9R,10S)- cleroda-3,12E, 14 -triene		This study
54	syn-cleroda-3,13(16),14- triene	(5S,8R,9R,10S)- cleroda-3,13(16),14- triene	terpentetriene	Comp. (Jia et al., 2016)
55	halima-5,12E,14- triene	(8S, 9R, 10S) -halima- 5,12E,14-triene		This study
56	syn-halima-5, 13(16),14-triene	(8R,9R,10S)-halima- 5,13(16),14-triene		Comp. (Jia et al., 2016)
57	syn-halima-5,12E, 14-triene	(8R,9R,10S)-halima- 5,12E,14-triene		This study
58	syn-halima-5 (10), 12E,14-triene	(8R,9R)-halima-5(10), 12E,14-triene		This study
59	abieta-8,12-diene	(10S)-abieta-8,12-diene	miltiradiene	Comp. (Gao et al., 2009)
60	isopimara-8,15 -diene	(10S,13S)-isopimara- 8,15-diene		Comp. (Xu et al., 2014)
61			geranyllinalool	Comp. (Jia et al., 2016)
62	labda-8(17), 14-dien- 13-ol	(9S,10S,13R)-labda- 8(17), 14-dien- 13-ol	manool	Comp. (Jia et al., 2016)
63	pimara-7,15-diene	(9S, 10S, 13R) -pimara- 7,15-diene		This study
64	ent-labda-8(17),14- dien-13-ol	(9R, 10R,13S) -labda- 8(17), 14-dien- 13-ol	ent-manool	Comp. (Jia et al., 2016)
65	ent-pimara-8(14),15- diene	(9R, 10R, 13R)-pimara- 8(14),15-diene		Comp. (Jia and Peters, 2016)
66	labda-7,14-dien-13 -ol	(9R, 10S)-labda-7,14- dien-13-ol		Comp. (Jia et al., 2016)
67	syn-labda-14-en-9,13 -diol	(8R,9R, 10S)-labda-14- en-9,13-diol	viteagnusin D	Comp. (Jia et al., 2016)
68	syn-labda-13(16),14- dien-9-ol	(8R,9R,10S)-labda- 13(16), 14-dien-9-ol		Comp. (Jia et al., 2016)
69	ent-cleroda-3,14 -dien-13- ol	(5R,8R,9S,10R)- cleroda-3,14-dien-13 -ol		Comp. (Jia et al., 2016)
70	ent-cleroda-3,13(16), 14- triene	(5R,8R,9S,10R)- cleroda-3,13(16),14- triene		Comp. (Jia et al., 2016)
71	syn-halima-5,14- dien-13-ol	(8R,9R,10S)-halima- 5,14-dien-13-ol		Comp.. (Jia et al., 2016)
72	syn-pimara-7,15- diene	(9R, 10S, 13R)-pimara- 7,15-diene		Comp. (Wilderman et al., 2004)
73	syn-stemodene	(9R,10S,13S)-stemod- 13(17)-ene		Comp. (Morrone et al., 2006)
74	ent-labda-8,13S- epoxy-14-ene	(8S,9S,10R,13S)-labda- 8,13 -epoxy-14-ene	ent-manoyl oxide	Comp. (Mafu et al., 2015)
75	syn-abieta-9(11),12- diene	(8S,10S)-abieta-9(11), 12-diene		This study
76	syn-pimara-9(11),15-diene	(10S, 13R)-pimara- 9(11),15-diene		This study

^aRelevant DTS and substrate can be found in Table S18.

^bProducts are numbered as defined in the text.

^cSemi-systematic names are designated for LRD derivatives by final backbone and stereochemically by the configuration of carbons 9 and 10 in the initially formed decalin bicycle (see Fig. 2), with complete stereodefinition also given.

^dCommon names are those previously reported.

^eProducts were identified based on either previous (Prev.) reports for these enzymes or GC-MS based comparison (‘Comp.’) to other previously reported DTS products (with accompanying reference(s) in these cases), or were determined in ‘This study’ by NMR based structural analysis or comparison to such a characterized enantiomer).

^fNumbering here differs from that used in the original report (Xu et al., 2018), to maintain consistency with that of the other LRD products.

Fig. 9.

Chemical structures of newly enzymatic products discovered in this study (numbering as defined in the text).

4. Methods and Materials

4.1. General

Unless otherwise noted, chemicals were purchased from Fisher Scientific and molecular biology reagents, including synthetic genes, from Invitrogen. The trans-nerolidol standard was purchased from Sigma-Aldrich. All constructs were verified by full sequencing of the inserted gene.

4.2. Recombinant constructs

The modularity of the metabolic system utilized here is based on the use of DEST cassettes that enable facile recombination via the Gateway (Invitrogen) cloning system (Cyr et al., 2007), particularly as inserted into the Duet (Novagen) series of vectors (Morrone et al., 2010). Accordingly, all DTCs and DTSs were first cloned into the pENTR/SD/D-TOPO vector, excluding the N-terminal plastidial targeting peptide sequence from the plant-derived genes. The DTSs were then typically recombined into a pDEST15 expression vector, although SsSS and SmMS were similarly inserted into pDEST14, while EtKS was directly inserted into the commercial expression vector Champion™ pET100/D-TOPO for convenience. DTCs were generally inserted into the DEST cassette found in the previously described pGG-DEST vector (Cyr et al., 2007). To screen DTSs activity against linear precursors, for 1 the DTS expression vectors were co-transformed with a previously described pGG vector (Cyr et al., 2007), for 3 with a previously described pMevT-MBIS vector (Martin et al., 2003), for 4 with a pGF vector constructed by sub-cloning a previously described GFPPS, specifically AtIDS9 (Nagel et al., 2015), into the NcoI/EcoRI restriction sites of pACYC-Duet1, while for 2 the DTSs were recombined from pENTR into a pNN-DEST vector for single plasmid co-expression as previously described for SsSS and KgTS (Jia et al., 2016). To increase metabolic flux towards terpenoids, several key genes from the endogenous isoprenoid precursor pathway were over-expressed using a previously reported pIRS plasmid (Morrone et al., 2010).

4.3. Metabolic Engineering

All metabolic engineering was carried out using the E. coli OverExpress C41 strain (Lucigen), and included pIRS as well as the relevant expression constructs (i.e., for the production of 1, 2, 3 or 4, and/or a DTC, along with a DTS). For initial activity screening purpose, recombinant cultures were grown in 50 mL TB medium (pH = 7.0), with appropriate antibiotics, in 250 mL Erlenmeyer flasks. These cultures were first grown at 37 °C to mid-log phase (OD₆₀₀ ~0.7), then the temperature dropped to 16 °C for 0.5 h before induction with 1 mM isopropylthiogalactoside (IPTG) and supplementation with 40 mM pyruvate and 1 mM MgCl₂. The induced cultures were grown for an additional 72 h before extraction with an equal volume of hexanes, with the organic phase then separated, and concentrated under N₂ when necessary for analysis.

4.4. Diterpene product analysis by GC-MS chromatography

GC-MS analyses were carried out using a Varian 3900 GC with a Saturn 2100T ion trap mass spectrometer in electron ionization (70 eV) mode, with an Agilent HP-5MS column (Agilent, 19091S-433), run with a 1.2 mL/min helium flow rate. Samples (1 uL) were injected in splitless mode by an 8400 autosampler with the injection port at 250 °C. The following temperature program was used: the oven temperature initially started at 50 °C, which was maintained for 3 min, and then increased at a rate of 15 °C/min to 300 °C, where it was held for another 3 min. Mass data was recorded by mass-to-charge ratio (m/z) values in a range from 90 to 650, starting from 13 min after sample injection until the end of the run. For detection of 15-carbon products, the helium flow rate was reduced to 1.0 mL/min, and the temperature program was modified to have a smaller increasing rate of 8 °C/min to 250 °C, and mass data collected from 6 to 30 min, with the range extended down to 40 (i.e., m/z from 40 to 650), while all other parameters were kept the same.

4.5. Diterpene production and purification

To obtain sufficient amount of new enzymatic products for NMR analysis, the bacterial cultures described above were simply scaled up to 1 L in 2.8 L Fernbach flasks. All other procedures were identical except that the extraction was repeated twice to ensure full yield. The separated organic phase was pooled and dried by rotary evaporation under vacuum, and the residue was re-suspended in 5 mL hexane for subsequent fractionation via flash chromatography over a 4 g-silica column (Grace) using a Grace Reveleris flash chromatography system with UV detection and automated injector and fraction collector, run at 15 mL/min. Briefly, the column was pre-equilibrated with hexanes and the sample injected, followed by 100% hexane (0–4 min), 0–100% acetone (4–5 min), 100% acetone (5–8 min), with peak-based fraction collection (15 mL maximum per tube). Generally, non-oxygen containing products eluted in the 100% hexane fraction; otherwise, the products were found in the 100% acetone fractions. Fractions containing the diterpene product, as identified by GC-MS analysis, were dried under N₂, re-suspended in 2 mL methanol, and filtered through 0.2 um cellulose filter (Thermo Scientific). These fractions were further separated using an Agilent 1200 series HPLC instrument equipped with a diode array UV detector and automated injector and fraction collector, over a semi-preparative C-8 column (ZORBAX Eclipse XDB-C8, 25 × 0.94 cm) run at 4 mL/min. The column was pre-equilibrated with acetonitrile/water (1:1 for oxygenated products, 4:1 for olefins), the sample injected, followed by washing (0 – 2 min) with same acetonitrile/water mix (i.e., depending on the targeted compound), then the percentage of acetonitrile increased to 100% (2–10 min), and final elution with 100% acetonitrile (10–30 min). Peaks were collected and analyzed by GC-MS and, if necessary, the diterpene further purified by another round of HPLC separation over an analytical C-8 column (Kromasil® C8, 150 × 4.6 mm) run at 0.5 mL/min, and using the same elution program described above. Fractions containing pure compounds were dried under N₂, and the compounds then dissolved in 0.5 mL deuterated CDCl₃ (Aldrich) for NMR analysis.

4.6. Chemical structure identification by NMR analysis

NMR spectra were acquired on a Bruker AVIII-800 spectrometer equipped with a 5-mm HCN cryogenic probe, set at 25 °C, using TopSpin 3.2 software. Chemical shifts were calculated by reference to those known for CDCl₃ (¹³C 77.23 ppm, ¹H 7.24 ppm) signals offset from TMS. All spectra were acquired using standard programs from the TopSpin 3.2 software, with collection of 1D ¹H-NMR, and 2D double-quantum filtered correlation spectroscopy (DQF-COSY), heteronuclear single-quantum coherence (HSQC), heteronuclear multiple-bond correlation (HMBC), HMQC-COSY and NOESY (800 MHz), as well as 1D ¹³C-NMR (201 MHz) spectra. Observed HMBC correlations were used to propose a partial structure, while COSY correlations between protonated carbons were used to complete the structure, which was further verified by HSQC correlations. Observed correlations from NOESY spectrum were used to assign the relative stereochemistry of chiral carbons and also the configuration of double bonds, where applicable. Absolute stereochemistry was assigned based on the known configuration of the upstream DTC product. The resulting structures were searched against the SciFinder database (https://scifinder.cas.org) to determine precedent.

4.7. Computational biology

The reported crystal structure of BjKS bound with substrate ent-CPP (PDB: 4XLY) is missing the J-K loop (residues 212–220) and also lacking the three Mg²⁺ ions required for catalysis. Therefore, the fully closed structure of BjKS was first modeled by addition of these two missing elements using the substrate analog and Mg²⁺ containing structure of bornyl diphosphate synthase (PDB: 1N20) as the template. From the structural alignment of 4XLY and 1N20 using Pymol, it was found that residues 501–509 from 1N20 aligned quite well with the missing loop in 4XLY. The PDB coordinates for this region (residues 501–509, 1N20) were obtained from the template PDB file, and Modeller (Sali et al., 1997; Šali et al., 2014) used to build back in the missing functional loop of BjKS. This initial model was optimized using the Chiron energy minimization server (Ramachandran et al., 2011), and the resulting structure used for subsequent docking studies. This was initiated by docking of the three catalytically requisite Mg²⁺ ions using AutoDock and AutoDock Vina (Morris et al., 2009; Trott and Olson, 2010), specifically so that Mg²⁺_a and Mg²⁺_c interact with the DDXXD motif, particularly the first and last aspartates (i.e., D75 and D79), while Mg²⁺_b interacts with the (N/D)Dxx(S/T/G)xxx(E/D) motif, particularly the first and last residues (i.e., N207 and D215), as almost universally found in active forms of TPSs (Christianson, 2008). The docking ligands were prepared in chem3D 15.1 with energy minimization using the MM2 molecular mechanics force field with all parameters set to their default values. The ligands then were docked individually into the modeled BjKS+Mg structure using AutoDock Vina. Twenty poses for each ligand were generated, with the lowest energy pose chosen for comparative purposes. Polar contacts to the diphosphate from both BjKS and the Mg²⁺ ions were identified using Pymol. Both the number and the type of polar contacts found in each representative pose were used to estimate the strength of interaction between the ligand and the enzyme. A structure for the (distantly) related EtKS was generated based on the original BjKS structure (PDB: 4XLY) using Modeller, and the closed structure constructed, again based on bornyl diphosphate synthase (PDB: 1N20), with final energy minimization via Chiron, as described above. This EtKS structure had a root mean-squared deviation (RMSD) of 1.2 Å with the modeled BjKS structure when considering all the residues. The volume of the active site pocket was then calculated using Chimera (Pettersen et al., 2004), based on key active site residues in BjKS (F72, L68, I36, Y168, I166, A167, L71, Y136, D79, D76, D75, D208, R204, N207) and EtKS (F72, M68, I36, Y168, I166, G167, L71, Y136, D79, D76, D75, D208, R204, N207).

Highlights.

1. Extended studies of the extreme promiscuity exhibited by the bacterial exo-ene producing KgTS and plant derived tertiary alcohol producing SsSS with additional potential substrates.

2. Investigation of endo-ene yielding diterpene synthases identifies the highly promiscuous bacterial trans-endo-ene producing ScLS.

3. Further investigation of the basis for substrate promiscuity reveals that this is more correlated with phylogenetic origin than reaction mechanism, with bacterial diterpene synthases exhibiting significantly less selectivity than those from plants.

4. The substrate promiscuity found here enables combinatorial biosynthesis, providing novel access to almost 19 labdane-related diterpenes.

Abbreviations

LRDs

labdane-related diterpenes

DTCs

diterpene cyclases

DTSs

diterpene synthases

IDS

isoprenyl diphosphate synthases

KgTS

terpentetriene synthase from Kitasatospora griseola

SsSS

sclareol synthase from Salvia sclarea

AbCAS

cis-abienol synthase from Abies balsamea

SmMS

miltiradiene synthase from Salvia miltiorrhiza

SaPS

diterpene synthase from marine bacterium Salinispora arenicola

BjKS

ent-kaurene synthase from Bradyrhizobium japonicum

ScLS

labda-8(17),12E,14-triene synthase from Streptomyces cyslabdanicus K04–0144

EtKS

ent-kaurene synthase from Erwinia tracheiphila

OsKS

ent-kaurene synthase from Oryza sativa

AtKS

ent-kaurene synthase from Arabidopsis thaliana

GGPP

(E,E,E)-geranylgeranyl diphosphate

NNPP

(Z,Z,Z)-nerylneryl diphosphate

GGPPS

GGPP synthase

NNPPS

NNPP synthase

FPP

(E,E)-farnesyl diphosphate

GFPP

(E,E,E,E)-geranylfarnesyl diphosphate

FPPS

FPP synthase

GFPPS

GFPP synthase

CPP

copalyl diphosphate

AgAS

abietaenol synthase from Abies grandis

SmCPS/KSL1

labda-7,13E-dien-15-ol synthase from Selaginella moellendorffii

NgCLS

8α-hydroxy-CPP synthase from Nicotiana glutinosa

MvCPS1

peregrinol diphosphate synthase from Marrubium vulgare

An2/ZmCPS2

ent-CPP synthase from Zea mays

AtCPS

ent-CPP synthase from Arabidopsis thaliana

OsCPS4

syn-CPP synthase from Oryza sativa

Haur_2145

kolavenyl diphosphate synthase from Herpetosiphon aurantiacus

KgTPS

terpentedienyl diphosphate synthase from Kitasatospora griseola

MtHPS

tuberculosinyl/halimadienyl diphosphate synthase from Mycobacterium tuberculosis

KPP

kolavenyl diphosphate

syn-HPP

syn-halimadienyl diphosphate

IPTG

isopropylthiogalactoside

RT

retention time

MS

mass spectra