今日推送的文章是ChemBioChem发表在上的“Determinants of Product Outcome in Two Sesquiterpene Synthases from the Thermotolerant Bacterium Rubrobacter radiotolerans”，作者为曼彻斯特生物技术研究所的Joshua N. Whitehead。
Rhodobacterium radiata-tolerant nerolidol synthase (NerS) and trans-α-bergamoene synthase (BerS) were one of the first terpene synthases (TPS) to be discovered from heat-tolerant bacteria, and although they share the same substrate, terpenoids are generated using different carbon scaffolds. Here, the potential thermal stability of NerS and BerS was investigated and found to remain active at temperatures up to 55 °C. A library of 22 NerS and BerS variants was designed to explore the different reaction mechanisms of NerS and BerS, including residues that may be involved in substrate chelation, cation-π stabilization of reaction intermediates, and active site profiling.
1
Bioinformatics analysis and homology modeling of NerS and BerS
NerS and BerS have typical class I TPS metal-binding motifs. In NerS, these motifs are 82DDxxD and 223NxxxSxxxE, and in BerS, these motifs are 89DDxxD and 243NxxxSxxxE. Bound trinuclear metal clusters (usually magnesium or manganese) help to isolate substrates and are therefore critical for TPS activity. The structurally conserved bacterial effector motifs 177RxxTG and 197RxxTG are also present in NerS and BerS, respectively, where arginine "sensors" are conserved and threonine and glycine represent the most common "linker" and "effector" residues in bacterial TPS. Arginine senses the pyrophosphate (PPi) moiety of the substrate, and the effector initiates the reaction cascade by donating electron density to the π* molecular orbital of the C2=C3 double bond of the substrate, which triggers ionization and forms subsequent cationic intermediates. 1NerS and BerS are most similar to bacterial and fungal TPS when queried in the SwissProt database. Whether their main products are derived from farnesyl or neroli cations is also consistent with the best match in this database. This is in contrast to plant TPS, which generally tends to cluster according to phylogeny.
Homology models of NerS and BerS (NerS-HM and BerS-HM) were constructed using the Automodel function 18 of SWISS-MODEL based on the PDB:4OKM (selinadiene synthase in pyrophosphate complex) template. The sequence identity of template 4OKM was 32.69% and 23.96% with NerS and BerS, respectively. These are typical values for uncorrelated TPS, as shown by the best match in the SwissProt database, although NerS has higher sequence homology with characterized TPS. In both models, none of the three magnesium ions and pyrophosphate in PDB:4OKM were retained. In NerS-HM, two metal ions and pyrophosphate are retained; Only one metal ion is retained in BerS-HM. In both cases, the missing magnesium ions were added by alignment with 4OKM, thus completing the trinuclear cluster known to be present in class I TPS. 9 Pyrophosphate ligands (retained in NerS-HM and imported into BerS-HM) were later used as a reference for molecular docking. Missing metal ions in both models affect (or cause) the location of the DDxxD metal-binding motif, with D86 (NerS) and D93 (BerS) not suitable for binding to one of the introduced metal ions. This can be emphasized by comparing NerSHM and BerS-HM to similar models built with AlphaFold, where metal binding and effector motifs are well aligned with the exception of NerS-D86 and BerS-D93. This may be because the magnesium ion has a weak affinity for the third site, as shown by other examples in the literature19, and is therefore difficult to predict in homology models. Nonetheless, both models may have a good approximation of the overall structure of NerS and BerS and their key motifs and remained stable during molecular dynamics (MD) simulations.
Figure 1
2
Molecular docking
The native sTPS substrate FPP was grafted to NerS-HM and BerS-HM (including three metal ions) using Autodock Vina20 as described in the Methods section. The selection of the optimal conformational isomer is based on three criteria: 1) alignment of the PPi moiety of the FPP with the PPi ligand retained or imported from the template PDB:4OKM; 2) PPi moiety with respect to arginine sensor residues R177 (NerS) and R197 (BerS); 3) The position and orientation of C3 of the FPP relative to the effector residues G181 (NerS) and G201 (BerS). The latter two criteria are particularly important because the ionization of FPPs by this catalytic motif is known to initiate a reaction cascade in class I TPS. 8 For both NerS and BerS, the chosen binding mode is one of the top results, and some of these results with "better" docking scores may be discarded because it is obviously impractical (e.g., the substate is flipped relative to the metal ion at the "top" of the active site). The active sites of the final models, NerS-HM-3Mg-FPP and BerS-HM-3Mg-FPP, are shown in Figure 1. The hydrophobic pockets in NerS are defined by the non-polar residues L55, Y75, M78, F79, V219, and W305 (Figure 1C). The polar residue N302 points to the tail of the FPP and corresponds to several characteristic asparagine residues in TPS16,21,22, although these enzymes produce cyclic products while NerS does not. Y75, F79, and W305 have the potential to influence reactions through cation-π interactions. F79 is the best candidate because its aromatic ring is adjacent to the C3 of the FPP at a distance of 4.6 Å. Positive charges accumulate on C3 during and after ionization and isomerization steps, which are presumed to occur during the formation of nerolidol. Y75 and W305 also have the ability to interact with cation-π, but are far from the nearest sp2 carbon (C7) in FPP, and are presumed not to undergo hybridization changes during the formation of acyclic nerolidol. Therefore, Y75 and W305 are more likely to define the active site profile along with the other non-polar residues described. In BerS, the active sites are defined by the corresponding residues L62, L85, L86, S82, F171, C239, I325, and W328 (Figure 1D). The aromatic ring of F171 is located at a distance of 5.5 Å from C3 of the FPP, in a direction that is not conducive to cation-π interactions. BerS lacks the aromatic properties corresponding to Y75 and F79 in NerS and instead has serine and leucine (S82 and L86) at these positions. It also lacks polarity at I325, which corresponds to N302 in NerS. The formation of trans-α-bergamot requires a larger substrate rearrangement than nerolidol, although BerS is primarily a (E)-β-farnesene synthase as shown in the following sections. (E)-β-farnesene is produced by the simple deprotonation of farnesyl cations after ionization. However, as discussed later, many plant (E)-β-farnesene synthases produce trans-α-bergamotenne as a minor product, as does BerS. One adaptation in this regard may be S82, whose side-chain hydroxyl group is in a favorable position with respect to the C7 of the FPP, where the positive charge accumulates after the formation of bisabol-based cations. Therefore, this residue may play a role in the formation of trans-α-bergamot limonene through cationic dipole interactions.
3
NerS、BerS 及其变体的体内产品概况
Based on the putative active sites identified above, six similar residues in NerS and BerS are selected for mutation to create a mini-library of single-point variants. These residues are (NerS/BerS): Y75/S82, M78/L85, F79/L86, G181/G201, V219/C239, and N302/I325. Variants are selected by swapping each pair of residues and/or creating larger and smaller variants. A total of 22 single-point variant enzymes were created. In vivo analysis of WT NerS, WT BerS, and variant mini-libraries was performed using the previously described "plug and play" platform. It has previously been reported that, when analyzed in vivo, NerS produces the monoterpenes β-myrcene and geraniol, as well as sesquiterpene nerolidol, α- and β-farnesene and farnesol. BerS has been reported to produce the monoterpenes β-myrcene, geraniol, and linalool, as well as sesquiterpene nerolidol, farnesol, farnesene (unspecified isomer), and trans-α-bergamonene. The product titers of these compounds were not quantified in this study. In vivo analysis of NerS and BerS in this study confirmed the presence of these monoterpenes and sesquiterpenes. Notably, since the plug-and-play platform relies on the overproduction of C5 precursors IPP and DMAPP, as well as the overexpression of GPP synthase, the excess geraniol is likely to be produced due to GPP conversion, and TPS is not able to process GPP fast enough. This is done by endogenous phosphatase in E. coli and explains why geraniol had the highest titer in most samples, including the otherwise inactive variant, which will be discussed later. The same transformation may explain some of the farnesol (from FPP) observed in vivo. While the presence of farnesol in WT NerS, WT BerS, and BerS-L86F is enzymatic demonstrated by in vitro assays, it is not possible to fully distinguish between enzymatic and endogenous farnesol by in vivo assays alone, so for the rest of the variant library, the changes produced by farnesol are not considered in detail. The main products of WT NerS and BerS are shown in Figure 2.
Figure 2
Nerolidol tertiary alcohol is the main product of NerS, and it also produces small amounts of linalool, farnesol, and (E)-β-farnesene. Notably, NerS produced much more nerolidol tertiary alcohol than farnesol, and (E)-β-farnesene, suggesting that NerS successfully ionized FPP to farnesyl cations and isomerized farnesyl cations to nerolidol tertiary alcohol cations (Figure 3). While nerolidol tertiary alcohol can also be obtained from farnesyl cations, the results of the variant library suggest that it is formed from nerolidol cations in NerS and BerS. In addition, since both enzymes produce farnesol and nerolidol (which have been shown to be enzymatic products), it is more likely that there is a dedicated water attack/deprotonation step in each enzyme, and that farnesol and nerolidol are from hydroxylation of non-isomerized or isomerized cations, respectively. In contrast, the main products of BerS are (E)-β farnesene and farnesol, both of which are derived from farnesyl cations. Nerolidol and trans-α-bergamotenne are derived from nerolidol cations and are only detected in BerS at very low concentrations. Therefore, although BerS was previously designated as trans-α-bergamoene synthase-10, it seems more appropriate to think of it as (E)-β-farnesene synthase, although it can make trans-α-bergamotene, and the name BerS will be retained for consistency. In vitro assays using purified proteins confirmed the product profile of WT NerS and BerS and demonstrated that the production of the main product was enzymatic. All NerS variants produce less nerolidol than WT, although this may be due to changes in soluble expression. Therefore, the focus here is on the change in the proportion of products or the emergence of new products, which may be due to enzymatic changes.
Figure 3
The product profile of NerS-M78F is similar to that of WT NerS, producing more nerolidol tertiary alcohol than (E)-β-farnesene or farnesol, although the overall titer is about 10-fold lower. However, despite the lower overall titer, the number of linalools (monoterpenoids) detected was similar to that of WT. This suggests that M78 may play a role in defining the size of the active site. M78A appears to be inactive, probably because the relaxed active site pocket does not properly sequester the substrate (M78 also forms hydrogen bonds with D82 of the DDxxD motif). However, the additional space volume of M78F may facilitate the binding of GPP (C10) to replace FPP (C15), resulting in more monoterpenoid products, but more work is needed to confirm this. In docking experiments, F79 in NerS was highlighted as potentially important because of its favorable location of the aromatic ring relative to the docking substrate C3 (Figure 1C). Positively charged cation-π stabilization on C3 will promote the formation of nerolidol on farnesyl cations during and after ionization and isomerization (Figure 3). This effect of F79 is supported by NerS-F79A data, which produces approximately 15-fold less nerolidol than WT (Figure 4). The nerolidol yield in the F79L variant is slightly higher, but still about 10-fold lower than WT, suggesting that the presence of larger leucine residues can "rescue" WT activity to some extent, but aromaticity is required for full activity, which is consistent with what has been observed in other terpene synthases. Two variants were produced by altering the effector residue of NerS (G181). Effector triads are structurally completely conserved in bacterial and fungal TPSs17, where mutations in residues often lead to enzyme inactivation due to impaired ability to sequester substrates and initiate reactions. Thus, the larger G181I variant was inactivated by the uncommon isoleucine effector of bacterial linalool synthase (bLinS)2, as expected. However, some plasticity is reported in the position. The NerS-G181A variant was also made, which produced the most neroli tertiary alcohol of any NerS variant (15.8 ± 3.1 mg L-1org vs 22.0 ± 4.8 mg L-1org in WT, Figure 4). Again, this illustrates that bacterial TPS can tolerate mutations in this location. Since the reaction is initiated by the backbone carbonyl of the effector residue, the identity of the side chain is less important, as long as the new residue is small enough and flexible enough not to disrupt the overall kinetics of the G-helix. In fact, as previously noted, the rational design of effector residues with different sidechain properties can prove to be a useful tool for constructing design TPS. Both 9V219C and N302I produced approximately 1.0 mg of nerolidol of Lorg-1, while N302Y produced nerolidol at all and only trace amounts of linalool. Both of these residues constitute the bases of the active site (Figure 1C), and the presence of macromolecular tyrosine substitutions may prevent the binding of longer FPP substrates and weaken the binding of shorter GPPs.
Figure 4
Many BerS variants are capable of generating (E)-β-farnesene, albeit in reduced quantities. The two variants, L86F and S82L, produce higher product titers than WT and will be discussed separately. BerS-L85A produces about 4-fold less (E)-β-farnesene than WT-BerS, whereas L85F appears to be inactive. L85 is located near the entrance to the active site, which could explain the insufficient product observed when replacing large phenylalanine. BerS-S82Y did not detect any products. The hydroxyl side chain of S82 points to the active site and may interact favorably with C7, where the positive charge accumulates in the bisabol-based cation on the way to trans-α-bergamonene. Mutation to tyrosine significantly reduces the volume of the active site, thereby excluding the substrate, as is the case with L85F. Notably, these inactivated variants did not even detect monoterpene products from shorter GPP substrates, although this may also be due to changes in soluble expression. The effector residue in BerS is G201, and mutations at this location have the same effect as mutations in NerS. Replacing glycine with isoleucine results in enzyme inactivation, but in BerS-G201A, the detection level of the main product, (E)-β-farnesene, is close to WT levels. Again, this suggests that bacterial TPS has some plasticity at this important catalytic position.
Substrate FPP ionization yields farnesyl cations (Figure 3). By quenching this cation by deprotonation at C14, (E)β-farnesene is produced. Alternatively, farnesyl cation isomerization produces nerolidol cations through nerolidol pyrophosphate ulolidol. From there, 1,6-ring closure yields birrh-based cations, and then 2,7-loop closure and deprotonation yields trans-α-bergamot. If nerolidol is produced from nerolidol by NerS and BerS from nerolidol cations, as we hypothesize based on the above reasons, nerolidol is dominant in NerS compared to (E)-β-farnesene and farnesol for BerS, suggesting that NerS can perform the isomerization step of generating nerolidol cations more successfully. This is reflected in the ratio of (E)-β-farnesene:trans-α-bergamotenne to approximately 17:1 and farnesol:nerolidol to nerolidol in BerS at approximately 17:1. However, it is worth noting that although NerS appears to reach nerolidol cations more easily, it does not have the correct active site structure to subsequently obtain bisabolol cations. The product profiles obtained in vitro and in vivo also indicate that, unlike NerS, BerS preferentially quenches the substrate by deprotonation, as (E)-β-farnesene and trans-α-bergamot are formed in this way. In addition, as shown in Figure 3, this deprotonation occurs at the same location (C14) for (E)β-farnesene and trans-α-bergamitene. If it is assumed, as shown in the literature,8, 9 TPS substrates bind in a product-like conformation with minimal overall rearrangement, then the main difference between the two products is how far the reaction has taken before the deprotonation step takes place. Perhaps it is for this reason that (E)-β-farnesene and trans-α-bergamot are often observed as by-products, although they look different.
4
Plant (E)-β-farnesene synthetase is usual
Trans-α bergamotenne is produced as a by-product
Plant sesquiterpene synthase (sTPS) was previously analyzed by Durairaj et al. and a library of annotations was assembled to identify motifs that control product outcomes. Examination of this library revealed that several seed plant sTPS known as (E)-β-farnesene synthase also produce trans-α-bergamot as a secondary product. Köllner and colleagues explored this relationship in maize and maize plants, which are known to release volatile terpenes to defend against lepidopteran larvae, attracting their natural enemy, the parasitic wasp Cotesia marginiventris. The two most important compounds in this defense strategy are (E)-β-farnesene and trans-α-bergamot, with a specific TPS ("TPS10") responsible for the biosynthesis of these compounds in maize. For different congeners of the TPS10 enzyme, the ratio of these two products varies, depending on the individual amino acid leucine/phenylalanine conversion. For enzymes that produce equal amounts of both products, this residue is leucine. For the two enzymes that preferentially produce (E)-β-farnesene, it is phenylalanine. Reciprocal mutations of leucine and phenylalanine between these groups swapped product distribution ratios, strongly suggesting that this position controls the distribution between (E)-β-farnesene and trans-α-bergamot. The authors speculate that an increase in phenylalanine may inhibit the formation or rotation of neroli alkyl pyrophosphate on the way to the neroli alkyl cation, thereby preventing isomerization. To reinforce this, when enzymes are fed with an already isomerized unnative substrate (Z,E)-FPP, their product distribution is dominated by trans-α-bergamot.
Figure 5
5
BerS-S82L 阻止异构化
BerS-S82L yields a higher titer than WT BerS for the main product (E)-β-farnesene, and the overall product titer is also higher. It also exhibited an increased preference for non-isomerized products, with (E)-β-farnesene:trans-α bergamot lemonene at approximately 23:1, farnesol:nerolidol tertiary at approximately 15:1, and WT BerS at approximately 17:1 and 10:1 (Figure 6). The small hydroxyl side chain of S82 points to substrate C7 at the BerS active site (Figure 1D), which is where the bisabolol cation accumulates positive charge on its way to trans-α-bergamone (Figure 3). In the MD simulation of BerS with 3 Mg2+ ions and FPP, the S82O–C7 distance is relatively stable, with an average distance of about 6 Å and a minimum distance of 4.4 Å. Thus, this hydroxyl group may interact with the positive charge on C7, thereby stabilizing the formation of bisabol-based cations, as previously postulated. In the S82L variant, mutation of this residue to leucine disrupts this stable interaction, potentially preventing the formation of bisabol-based cations and therefore trans-bergamolin. However, since the bisabol-based cation is derived from an already isomerized nerolidol-based cation (Figure 3), this may not be the reason for the change in the product characteristics of the BerS-S82L variant, and despite the increase in overall product titers, nerolidol and linalool (also isomerized) yields were the lowest compared to WT-BerS and BerS-L86F. However, S82 may still play a role in trans-α-bergamot synthesis in WT BerS and BerS-L86F. Conversely, the effect of BerS-S82L is more likely to be similar to that of plant TPS described above, where the increased size of leucine prevents the isomerization step compared to serine. Typically, isomerization brings the C1–C2 allyl bond closer to the C6=C7 double bond, and then the 1,6 ring closes to produce a bisabol-based cation (Figure 3). The additional spatial volume of the leucine side chain reduces the space available for the isomerization step, possibly by compressing the substrate and thus preventing the necessary loop closure step, although in this case the interruption leads to the formation of an alternative cyclization pathway rather than a noncyclic product. The sensitivity of the enzyme to the spatial requirement at this location is further demonstrated by BerSS82Y inactivity, even for the formation of monoterpenoids from short-chain GPPs. Therefore, the presence of S82 in BerS WT may represent an evolutionary process towards trans-α-bergamot biosynthesis, as TPS is known to sacrifice activity to better control the formation of complex products.
Figure 6
6
BerS-S82L 阻止异构化
Interestingly, the results for BerS-L86F are contrary to those of TPS10 analogues, as the substitution of larger residues increases the proportion of products produced from neroli alkyl cations. This suggests that the mechanistic interpretation of the change in product proportion in BerS-L86F is different. Ionization and isomerization are known to be the slowest chemical steps in TPS catalysis. 9In class I TPS, the electron density of the effector carbonyl lone pair is donated into the π* molecular orbital of the C2=C3 bond of the substrate, which initiates ionization. L86 in BerS is similar to F79 in NerS. In fact, it is on this basis that it is mutated into aromatic residues: phenylalanine is able to stabilize the positive charge on C3 by cation-π interaction after ionization, and therefore can improve activity and/or isomerization.
Since phenylalanine is larger than leucine, it is unlikely that the increase in isomerization products in this variant is due to an increase in substrate degrees of freedom, as is the case with the TPS10 analogues described above (as well as the previously discussed Artemisia annua-4,11-diene synthase and (E)-β-farnesene synthase13). Therefore, it is speculated that the main effect in this BerS variant is the electron effect, resulting from the increased ability of L86F to initiate and stabilize ionization and isomerization by interacting with the favorable cation-π of the newly formed carbocation. This is supported by the change in the ratio of (E)-β-farnesene:trans-α bergamot and farnesol:nerolidol tertiary alcohol in BerS-L86F. In WT BerS, these ratios are ~17:1 and ~10:1, respectively, indicating a strong preference for farnesyl cationic derivatives. In BerSL86F, these ratios are much lower, around 10:1 and 4:1, respectively, which means that BerS-L86F has a better ability to isomerize farnesyl to neroli alkyl cations, contrary to the changes observed with BerS-S82L (Figure 6A). We also observed the presence of small amounts of β-bisambonene, an isomerized cyclic sesquiterpene not observed in WT BerS, in this variant. This product was obtained by deprotonation of birrh-based cations at C15 (Figure 3), further demonstrating an improved ability of BerSL86F to reach birrh-based cations through isomerization and subsequent ring closure. An increase in catalytic activity (confirmed in vitro, see below) may also mean an improvement in the ionization step of the initiation reaction, as this is often the rate-limiting step of TPS.
7
WT BerS 和 BerS-L86F 与 GPP 和 NPP 的孵育
To further explore the reasons for the significant improvement in activity and isomerization of BerS-L86F compared to WT BerS, two enzymes were purified and incubated with the native mTPS substrate GPP as well as nerolidol pyrophosphate (NPP), which is also C10 but has been isomerized. NPP is not a natural substrate for NerS and BerS, although it is a natural substrate for some TPS. It is worth noting, however, that while GPP is a natural substrate for monoterpene biosynthesis, BerS is a sesquiterpene synthase (sTPS) whose true substrate is C15 FPP. Perhaps it is for this reason that the product of WT BerS was not detected after 30 min at 37°C when incubated with GPP. Feeding NPP to WT BerS also did not produce a detectable product, suggesting that ionization may be a rate-limiting factor or that WT BerS simply was unable to form a proper Michaelis complex with these shorter substrates, which was detected by FPP compared to FPP. As mentioned later, the turnover number of BerS relative to FPP to its main product (E)-β-farnesene is calculated to be only 37 min-1 at 0.018 °C, so it may not be surprising that no product of GPP is detected, which is not the main substrate for BerS. However, the product of BerS-L86F was detected, further confirming the improvement in activity observed in vivo (Figure 6B). Geraniol and nerolidol are produced by the hydrolysis of GPP and NPP, respectively. The absence of these compounds in the in vitro control reaction suggests that this is enzymatically catalyzed, although significantly more non-enzymatic geraniol was also observed in vivo due to overexpression of the heterologous MVA pathway and GPP synthase in the authors' plug-and-play system. Crucially, unlike BerS WT, BerS-L86F is able to convert non-isomerized GPP to detectable amounts of isomerized linalool, further suggesting that the new aromatic ring plays a role in promoting the isomerization of cation-π interactions. The presence of nerolidol (derived from NPP) also suggests that BerS-L86F has an active site profile that is more suitable for isomerization substrates, which is consistent with the increased yield of complex isomerization products such as trans-α-bergamite and the emerging β-bisabolene in FPP.
8
Circular dichroism is used to probe the potential thermal stability of NerS and BerS
Since NerS and BerS were mined from the genome of the heat-tolerant bacterium Rubrobacter radiotolerans, their potential thermal stability was tested using circular dichroism (CD). First, NerS and BerS (containing magnesium ions) were scanned from 180-260 nm to find the characteristic α-helical absorption signal. Both enzymes show strong absorbance at the expected wavelength of 210 nm. Next, a temperature slope analysis was performed while measuring the change in the CD signal at 210 nm. Fitting the CD measurement curve yields the melting temperature (TM) for each enzyme. The TM of NerS is 47.6 °C ± 0.4, while the TM of BerS is 44.4 °C ± 0.4, suggesting that NerS is more heat-resistant than BerS. However, TM may not be directly related to the optimal temperature, as enzymatic functions can be retained for some time after the onset of structural degradation. Similarly, activity may cease before the melting temperature is reached. Therefore, a temperature control determination is required to determine the optimal temperature. Since TPS undergoes significant dynamic rearrangements upon substrate binding, and these conformational changes may affect thermal stability, the CD temperature assay is repeated by adding [10x] PPi to the assay buffer. PPi mimics the PPi moiety of FPP and can bind to trinuclear magnesium clusters or complex with effector residues and other basic residues in the hydrophilic region of the active site. These induced fit changes may lead to an increase in enzyme stiffness, which improves thermal stability. However, the melting temperatures of both NerS and BerS remained constant in the presence of PPi (47.4 °C ± 0.5 and 44.2 °C ± 0.3 vs. 47.6 °C ± 0.4 and 44.4 °C ± 0.4), suggesting that the presence of PPi had no effect on protein stiffness or thermal stability. Full induction of fit may require the presence of a natural substrate or analogue, not just a PPi component.
9
Enzyme kinetics and temperature-controlled assays for NerS and BerS
The turnover number of NerS and BerS (relative to the conversion of FPP to nerolidol or (E)-β-farnesene) was determined using a purified enzyme assay of excess FPP. To further explore the potential thermal stability of these enzymes, these turnover assays were performed at increasing temperatures. As shown in Figure 7A, the turnaround of NerS at 50 °C is 0.38 min-1, which is more than three times faster than 0.11 min-1 at 30 °C. However, at 37 °C, the turnaround is 0.31 min-1, which accounts for almost three-quarters of this difference. The optimal temperature at which FPP can be converted to nerolidol appears to be around 50 °C, with activity decreasing at 55 °C and undetectable at 60 °C. This is consistent with the melting temperature of 47.6 °C for NerS as determined by circular dichroism. Although NerS may begin to lose some structural organization at 45 °C and above, it must retain some catalytic function. The increase in kinetic energy in the system may also lead to a higher overall turnover. This is prescient for TPS, as product release is considered to be the slowest non-chemical step. 8 Thus, the increase in kinetic energy in the enzyme:product complex may contribute to product release. As shown in Figure 7B, the BerS had a much lower turnover than the NerS under the assay conditions tested. At 30 °C, the product detected at the expected retention time of (E)-β-farnesene was at the threshold of the detectable limit of the GCMS, making quantification too difficult. There was an increase from 37 °C to 45 °C, but no activity was observed at 50 °C. Even without considering the increase in trans-α-bergamite production, the turnover of (E)-β-farnesene formation at 37 °C for the BerS-L86F variant was higher than that of WT BerS at 37 °C (0.018 min-1) or 45 °C (0.024 min-1). The improvement in activity of BerS-L86F with FPP is consistent with the results observed in both in vivo and in vitro GPP and NPP assays.
10
NerS has moderate heat resistance
As a result, NerS appears to have some heat resistance, exhibiting optimal catalytic activity at ~50 °C and remaining active at temperatures up to 55 °C. This is much higher than the reported average optimum temperature for TPS, although lower than the previously reported thermostable TPS12 and TPS32, which are specifically designed to enhance thermal stability, both of which exhibit activity up to 65 °C. The reaction rate of NerS at higher temperatures reaches a steady state after the same period of time as at lower temperatures, most likely due to enzyme adsorption to the test tube. If the protein is unstable at these higher temperatures, one may see a sharp increase in initial product formation followed by a rapid decline, but this is not the case. Similarly, the control reaction ruled out the possibility of increased product formation due to increased decomposition of the FPP solvent at higher temperatures, while the same analysis of bLinS, which also converted FPP to nerolidol resulted in a significant loss of activity above 37 °C. For these reasons, NerS does have some degree of heat resistance compared to the typical optimum temperature of the enzyme close to (or below) 37°C. As mentioned above, protein conformational changes caused by substrate sequestration may have a positive effect on the heat tolerance of NerS. The second round of circular dichroism analysis measured only the effects of PPi, not the entire substrate. Since temperature-controlled kinetic analysis is performed using the native substrate FPP, these dynamic effects may be present and may explain the heat resistance of NerS, but more work is needed to determine if other factors are at play. In contrast, CD experiments and temperature-controlled analysis have shown that BerS does not possess any meaningful heat resistance.
Figure 7