The identification and characterization of novel product

In order to isolate the unknown product, 70 L of mutant yeast were incubated and the non-saponifable lipids (NSL) were extracted. One of the novel products was a major compound within the culture of the mutant for ERG7^I705F. (Fig. 3.2) The products mixture which with m/z = 426 were acetylated for increasing the polarity, then their m/z became 468. (Fig. 3.3) According to the principle that an argentic ion would bind with a different double bond of the compound, each single compound was separated by AgNO3-impregenated silica gel column chromatography with 2.5 - 3% diethyl ether in hexane as the mobile phase. By the end, through the deacetylation step, the isolated single compound (Fig. 3.2) could be identified by nuclear magnetic resonance (NMR).

Figure 3.2 The GC data of ERG7^I705F. After the novel product was isolated, the complex

Novel product

mixture was purified to became a sole product shown in picture below.

(13αH)-isomalabarica-14(26),17,21-trien-3β-ol 17α-protosta-20(21),24-dien-3β-ol

Lanosterol

I705F

Novel product

m/z = 426

m/z = 468

Figure 3.3 The mass spectra of novel products from the ERG7^I705F mutant.

The unknown product (Fig. 3.4) which contains a tetracyclic scaffold with a 17α side chain and a △^20/22 double bond, and its spectroscopic assignment was determined according to the following analysis of various NMR spectra (¹H, ¹³C NMR, DEPT, ¹H-¹H COSY, HMQC, HMBC, and NOE), and the spectra in depth are shown in Appendix section.

First, the compound showed ¹H NMR spectrum with three vinylic methyl singlets (δ 1.66, 1.6, and 1.52), and five methyl singlets (δ 0.95, 0.75, 0.9, 1.11, and 0.8), and two olefinic protons (δ 5.07, and 5.07). Second, the ¹³C NMR showed the presence of two tertiary-quaternary (δ = 131.85, 124.31, and 123.91, 137.7). These results indicated the presence of a tetracyclic nucleus with a terminal hydrocarbon side chain double bond.

Correlations of the HMQC, HMBC, and NOE spectra showed the following features: (1) the vinyl proton at δ 5.07 (δc 123.91, C-22) is coupled to carbons at 13.44 ppm (C-21), 27.77 ppm (C-23), and 51.29 ppm (C-17); (2) the vinyl proton at δ 5.07 (δc 131.85, C-24) is coupled to carbons at 18.18 ppm (C-27), 26.16 ppm (C-26), and 124.31 ppm (C-25); (3) the proton at δ 2.65 (δc 27.77, C-23) is coupled to carbons at 123.91 ppm (C-22), 124.31 ppm

(C-25), 131.85 ppm (C-24), and 137.70 ppm (C-20); (4) the proton at δ 1.59 (δc 44.94, C-13) is coupled to carbons at 16.93 ppm (C-30), 23.41 ppm (C-11), 25.78 ppm (C-12), 50.79 ppm (C-14), 137.70 ppm (C-20)and 51.29 ppm (C-17); (5) the proton at δ 0.90 (δc

23.2, C-19) is coupled to carbons at 33.87 ppm (C-1), 37.66 ppm (C-10), 46.72 ppm (C-9), and 48.72 ppm (C-5). These correlations established the bond connectivity between the tetracyclic nucleus skeleton and the exocyclic hydrocarbon side chain as well as the double bond positions. The distinct evidence of NOE spectra, which was observed among Me-19/Me-29, H-5/Me-18, and H-9/Me-19, confirmed the presence of chair-boat-chair conformation of the compound, and the tetracyclic structure contains the spatial NOE interaction between Me-30/H-17, and Me-30/H-9, confirmed the 17α side chain conformation.

Figure 3.4 The structure and the NOE correlation of the novel compound

0.8

Table 3.3 NMR assignments for 17α-protosta-20(22),24-dien-3β-ol for dilute CD2Cl2

solution

This is the second time to isolate the 17α tetracyclic triterpene alcohol derivative which has the chair-boat-chair skeleton simultaneously, while all of the other chair-boat-chair products are 17β-derivatives. The novel products, 17α-protosta-20(22),24-dien-3β-ol, and 17α-protosta-20(21),24-dien-3β-ol which was the first 17α-derivative isolated in our lab⁴⁷, are simultaneously produced by the ERG7^I705F mutant. These data showed that I705 is an important residue that contributes to the formation of 17α-products.

3.1.3 Proposed cyclization/rearrangement pathways of TKW14C2 expressing ERG7^Ile705X

The various products derived from ERG7^I705X indicated that I705 is a crucial residue in the putative active site for catalytic function of oxidosqualene cyclase. Seven triterpenoid compounds include three tricyclic and four tetracyclic products. Three truncated tricyclic compounds including (13αH)-isomalabarica-14E,17E,21-trien-3β-ol, (13αH)-isomalabarica-14Z,17E,21-trien-3β-ol and (13αH)-isomalabarica-14(26),17, 21-trien-3β-ol, suggest that after oxidosqualene was pre-folded into the chair-boat-chair conformation and then following cyclization step via cation-π interactions to a 6,6,5-tricyclic C-14 cation intermediate, the three (13αH)-isomalabarica-trien-3β-ols were derived from deprotonation at either the C-15 or C-26 of the 6,6,5-tricyclic C-14 cation.

Following the anti-Markovnikov expansion of the five-membered C-ring to form a 6,6,6-tricyclic C-13 cation, and then via D-ring closure to generate the protosteryl C-20 cation, the substrates through a series of methyl and hydride shifts, deprotonation at C-8 or C-9 yield lanosterol. The mutation of I705 may influence the F699 residue so that it affects the stability of the C-17 cation intermediate at or after the protosteryl cation formation, and the other parts of substrates generate the truncated triterpene, protosta-13(17),24-dien-3β-ol.⁴⁸

Different from the previous characterization, an amazing phenomenon appeared in the I705 mutation, there are two 17α-derivatives, 17α-protosta-20(22),24-dien-3β-ol and 17α-protosta-20(21),24-dien-3β-ol, observed in the ERG7^I705F mutant simultaneously. The speculated pathway is the unnatural D-ring closure, α-orientation of the large side-chain, and then it generates the 17α− protosteryl C-20 cation. This unnatural mechanism may arise from lack of correct amino acids in the putative active site of the cyclase, without rearrangement compared to the normal type, with deprotonation at C-21 or C-22 positions, yielding 17α-protosta-20(22),24-dien-3β-ol and 17α-protosta-20(21),24-dien-3β-ol. (Fig.

3.5)

Figure 3.5 Proposed cyclization/rearrangement pathway occurred in the ERG7^I705X site-saturated mutants

In a previous study, the near absence of 17α dammarenyl cyclases in higher plants reflects the rarity of the 17β to 17α evolutionary step.⁴⁴ Therefore, no matter what protosteryl cation or dammarenyl cation is present, the 17α configuration is unusual and rare in the cyclization process. Formation of 17α-derivatives is unnatural, and the proposed pathway via the 17α−protosteryl C-20 cation could not be defined in the mechanism of the wild-type cyclases. On the other hand, the 17α configuration intermediates did appear only in bacteria or rare plants species, indicating that the 17α configuration may exist in early evolutionary periods, which could rationally explain the rarity of the 17α−derivatives.

3.1.4 Analysis of the ERG7^Ile705X mutants with the ERG7 homology modeling

Molecular modeling studies were performed to investigate the roles of important residues in the cyclization/ rearrangement cascade. Because of the lack of a high-resolution crystal structure of S. cerevisiae ERG7, the homology models of S. cerevisiae ERG7 and its mutated ERG7 proteins were derived from the human OSC X-ray structure, as the template, and complexed with lanosterol or 6,6,5-tricyclic C-14 cation or 17α-protosteryl cation, together with product profiles, in order to investigate the relationship between enzyme structure and product specificity. The homology model of ERG7 showed that Ile705 is a hydrophobic residue different from neighboring aromatic residues, and I705 is spatially proximal to C-14 and C-17 positions of lanosterol (Fig. 3.6a). The hydrophobic property of I705 is considered to have some interaction with the substrate, and also affect the first-tired residues. According to the product profiles of ERG7^Ile705X, the models complexed with 6,6,5-tricyclic C-14 cation and 17α-protosteryl cation, take into account the variation of the residue at 705 position. Furthermore, the product profile of ERG7^I705X is similar to that of ERG7^F699X mutations⁴⁷; therefore we observe the variation of F699 for investigating the function of I705 in depth. (Table 3.4)

ERG7^I705F and ERG7^I705G mutants both produce 17α-protosta-20(22),24-dien-3β-ol, which is the novel product in this study. According to the homology models, the distances between ERG7^I705F and C-14 and C-17 of lanosterol are closer than wild-type, and the distance between I705F and C-17 of 17α-protosteryl cation is also closer than ERG7^I705. (Fig. 3.6b) The aromatic Phe705 stabilized the 17α-protosteryl cation with π-cation interactions, therefore producing both 17α-protosta-20(22),24-dien-3β-ol and 17α-protosta-20(21),24-dien-3β-ol. On the other hand, substitution of Gly with a small aliphatic side chain possibly enlarged the cavity of the active site to cause the substrate to be more flexible, and then could not form lanosterol accurately. There are many truncated products discovered in the ERG7^I705G mutant, including a small amount of

17α-protosta-20(22),24-dien-3β-ol. Except for the 17α-product, the other truncated products are considered to form by variation of the F699 residue, where the flexible Gly possibly influences the F699 residue in this case. Formation of the 17α-protosteryl cation is rare and it was observed in F699X mutations before, so the 17α-protosteryl cation formation may be caused by the effect of F699.

Substitution of I705 into an aliphatic group such as Ala or Val produce truncated products, whereas Leu only produces lanosterol. The ratio of lanosterol decreases when the size of the side chain becomes smaller. Because Leu has a similar bulky size as Ile, the ERG7^I705L mutant only yielded lanosterol as was observed for the wild-type. The percentage of lanosterol for ERG7^I705A and ERG7^I705V are about 75 - 78%, and even the ERG7^I705G mutant only has about 35% of lanosterol shown in Table 3.2. This is the evidence that the steric bulk size at the 705 position is an important factor during catalytic cyclization. The change of the distance between I705A/V of lanosterol are both decreased, and the variation of the F699 residue within I705A/V mutants, complexed with 6,6,5-tricyclic C-14 cation, are closer than wild-type ERG7.

The obvious phenomenon within the acidic, amide and basic group such as I705D/

N/ Q/ K mutants represent smaller amounts of lanosterol compared to the aliphatic group, even without any other products yielded in the ERG7I705E/ H/ R mutations. The balance environments destroyed by substitution of hydrophobic residue into acidic, amide and basic group within the putative active site, leading to tendency of lanosterol formation reduce and generated some truncated products by the F699 effect. In addition, the mutated ERG7I705S/ T/ C/ M/ P proteins also yielded lanosterol and truncated products, similar to that observed for most ERG7^I705X mutations. Comparing the product profiles, we discovered the side chain polarity that could affect the proportion of lanosterol. Substitutions of I705 into Cys, Met and Pro with a nonpolar side chain have higher lanosterol amounts, while Ser and Thr residues with a polar side chain have a small amount of lanosterol. The analysis of the

lanosterol formation tendency show that the acidity and polarity of the residue at the 705 position influence the environment around the substrate within the active site of the oxidosqualene-lanosterol cyclase. But the distance between I705X (X=D, N, Q, K, E, H, R, S, T, C, M, P) and lanosterol, and the variation of the F699 residue in these mutants complexed with 6,6,5-tricyclic C-14 cation are irregular. The irregularly homological results perhaps are the results of a deviation arising from the computational calculation.

However, it also provided the information that the relative positions between the substrate and the cyclase were altered during the catalytic cyclization and the distance may become farther or nearer, when the cation intermediate formed.

Finally, there is no product observed in the ERG7^{I705Y/ W} mutants. Although Tyr and Trp are aromatic residues like Phe, the bulky size makes these two mutations difficult to complement yeast viability, causing the loss of cyclase activity. The steric bulk size may affect the F699 residue strongly, and the stabilization around F699 and the substrate destroyed, so that the catalytic processes were abolished. In conclusion, the key factors which influence the diverse product profiles and yeast viability of the I705X mutations are steric effect, acidity and side chain polarity. Variation at the I705 position certainly influences the substrate and the cyclase in significant ways.

a. b.

Figure 3.6 Homology models. (a) The homology model of wild-type ERG7 complexed with lanosterol, and the distance between I705 to C-14 and C-17 positions of lanosterol is 5.61 Å and 6.14 Å. (b) The homology model of ERG7^I705F mutant complexed with the

17α-protosteryl cation.

Amino acid substitution

Distance to C-17 of lanosterol (Å)

Distance to C-14 of lanosterol (Å)

Distance to C-17 of 17α-protosteryl cation

Table 3.4 The distance of I705 to C-14 and C-17 complexed with different ligands in the homology models, and observation of the variation of F699 within the ERG7^I705X mutants, using the homology model complexed with the 6,6,5-tricyclic C-14 cation.

3.1.5 Product analysis of the double mutant of ERG7I705F/F699X

The product profile of ERG7^I705X is similar to that of ERG7^F699X mutations investigated before in our lab, and F699 had been identified to be an important plasticity

residue with its product diversity⁴⁷. In order to prove that the hydrophobic residue at this position plays an important role that influences the first-tired residue in the putative enzymatic active site, some double mutants were constructed for study in depth. The various products of ERG7^I705X, 17α-protosta-20(22),24-dien-3β-ol is the only compound that ERG7^F699X mutations had not discovered before. The novel compound, 17α-protosta-20(22),24-dien-3β-ol, was only observed for the ERG7^I705F mutation, so that the combination of I705F/F699X double mutants were constructed and their products analysis will be discussed later.

In the experiment with ERG7F699C/I705F double mutants, as Table 3.4 describes, the single F699C mutation loses the activity of the oxidosqualene-lanosterol cyclase, causing the failed result in the absence of exogenous ergosterol, and the result of double mutants is the same as that observed for the single mutant. These data indicate that the F699 is the crucial residue while the I705 residue is critical as an “assistant＂ to affect the first-tired residue F699. Furthermore, in the single mutant F699M, the products yielded include lanosterol, protosta-13(17),24-dien-3β-ol, protosta-17(20),24-dien-3β-ol, (13αH)-isomalabarica-14E,17E,21-trien-3β-ol, (13αH) -isomalabarica-14Z,17E,21-trien-3β-ol, the chair-chair 6-6-5 tricyclic malabarica-14(15)E,17,21-trien-3β-ol and 17α-protosta-20(21),24-dien-3β-ol. Double mutants of I705F/F699M yield 17α-protosta-20(22),24-dien-3β-ol without 17α-protosta-20(21),24-dien-3β-ol, because F699 is crucial within the active site of cyclase, where the 17α-products might be formed by mutation of F699. With different orientations, the I705 affects the F699 residue and has a tendency to deprotonate the proton at C-22 simultaneously. Moreover, interesting data was discovered in the ERG7 I705F/F699T mutation shown in Table 3.5 that except for the formation of little amount of lanosterol, the single mutant of ERG7^F699T produced protosta-13(17),24-dien-3β-ol as the major compound. The generation of protosta-13(17),24-dien-3β-ol, 17α-protosta-20(21),24-dien-3β-ol and

17α-protosta-20(22),24-dien-3β-ol also appeared in the double mutation of I705F and F699T. This result showed that I705 is not only an important assistant residue to the first-tired residue F699, but it is also essential for contribution towards the forming of 17α-products. Furthermore, the results of homology models show the relative distance between F699M/I705F, F699T/I705F to the C-17 of lanosterol are shorter than the wild-type ERG7. However, the modeling data provide us with strong information that the I705F is influential at the C-17 position. (Fig. 3.7)

45

Table 3.5 The products analysis of double mutants between I705F and F699X

^{699 3.67 3.46 4.46}

Figure 3.7 The homology models of the double mutants ERG7F699M/I705F (gray) and ERG7F699T/I705F (purple) complexed with lanosterol.

3.2 Functional analysis of PSY

within P. sativum

The construction of saturated mutations of Leu734 within the β-amyrin synthase from Pisum sativum were done by utilizing the QuikChange Site-Directed Mutagenesis kit, then by using the restriction enzyme (Nru I) mapping check. The mutated plasmids were digested into 4.6 and 3.5 kbp fragments compared to the wide-type plasmid that was digested into one fragment. Similar to the construction of Ile705X, the mutated plasmid sequences were also confirmed via the ABI PRISM 3100 DNA sequencer.

The recombinant plasmids were confirmed and transformed into TKW14C2 by the same strategies as previously described in Section 3.1.1., and the genetic selection of the TKW14C2[pPSYL734X] mutants are shown in Table 3.6. Because the plant PSY gene was expressed by the yeast system, as per the previous speculation, the exogenous ergosterol was necessary for growth. After incubating them in large amounts of culture mediums, the results were represented following lipid extraction and column chromatography. Using GC-MS instrumentation for all products analysis, all of the mutations of L734X did not generate any triterpenoid structures with molecule mass of m/z = 426. Without any truncated compounds and β-amyrin, the result reveals the importance of the L734 residue within the putative active site of P. sativum βAS.

PSY^mut Restriction enzyme

Table 3.6 The site-saturated mutants of P. sativum PSY^L734X and their genetic selection and products analysis.

3.2.2 Experimental results of PSY^L734X mutants and their phenomena

β-amyrin synthase is one of the oxidosqualene cyclases that convert oxidosqualene into pentacyclic 6,6,6,6,6-fused β-amyrin, and there are many essential amino acids in the putative active site and act their specific roles. Substitution at the L734 position, a hydrophobic residue within P. sativum βAS, causes the production of β-amyrin to fail. As per the previous description, there is no triterpenoid compound with molecule mass of m/z

= 426 observed for the L734 mutation. Except for the formation of β-amyrin similar to that observed for wild type, generation of truncated products provide the evidence that each residue plays a specific functional role. In this case, the phenomenon of no truncated products and β-amyrin, it could be presumed that PSY^L734 is crucial for cyclase. Therefore, the possibility of the disrupted catalytic reaction will be discussed later.

The resolution of human OSC structure was published in Nature, Thoma and co-workers (2004) considered that the human OSC has a channel which is supposed to lead the substrate oxidosqualene into the hydrophobic active site. There are some residues such as Tyr237, Cys233 and Ile524 near the substrate passageway, and by changing the conformation of their side chains, the passage of the substrate subsequently could be achieved.²⁶ In the study of the F528 residue within SceERG7, F528 is located in the substrate entrance channel and probably influences the enzymatic activity through substrate binding. Some observations have shown that the ERG7^F528X mutations cannot complement the cyclase activity due to the change of pH scale or polarity in the substrate passageway, causing disruption of triterpenoids formation.⁴⁹ Nevertheless, this supposition could not be established in the case of PSY^L734. In the modeling picture shown in Fig. 3.8, C260 and I555 residues within P. sativum βAS, which correspond to Tyr237 and Ile524 in human OSC, are assumed to be the crucial amino acids that affect the substrate passageway.

Different from C260 and I555, the L734 residue is located below the substrate and the distance between C260, I555 and L734 are so far, that L734 may not be one of the

important amino acids that control the substrate entrance channel.

Figure 3.8 The homology model of wild-type PSY complexed with β-amyrin. C260 and I555 residues are assumed to affect the substrate passageway within the cyclase.

On the other hand, the distance between L734 and the substrate within βAS is closer than that of I705 with lanosterol, and therefore the effect of L734 is stronger than I705 for cyclases. As shown previously in Section 3.1, the products of ERG7^I705X mutations are diverse and special. Except for lanosterol, the formation of truncated products provided strong evidence for the importance of SceERG7^I705. I705 is not only a crucial residue for catalytic function of the oxidosqualene-lanosterol cyclase, but also spatially close to its neighbor first-tired residues, especially F699. Because of the stronger effect in L734X mutations, the substitution of other amino acids may influence the conformation of the substrate oxidosqualene before the initial epoxide ring opening. Incorrect substrate conformation may halt catalytic cyclization, and that is the rational explanation for losing the activity of P. sativum βAS. In addition, the functional roles of the other amino acids

such as F728, Y736, W418, W612, F474, W257, and Y259, where their corresponding residues all have crucial roles within SceERG7, have not been confirmed by experimental results. Therefore, the direct effect for causing the loss of the activity of P. sativum βAS by the L734 mutations could not be explained in depth, and it is likely the case that the disruption of catalytic cyclization was influenced by the L734 mutations indirectly. Finally, we could only speculate that PSY^L734 may stabilize the substrate conformation, but the detailed function should be investigated in the future. (Fig. 3.9)

Figure 3.9 The minimum distance of Leu734 to β-amyrin is 4 Å. The possible crucial residues also shown in this modeling picture.

3.2.3 Analysis of the PSY^L734X mutants with the PSY homology modeling

We expected the function of L734 within the β-amyrin synthase may the same as I705 within SceERG7, because of their similar hydrophobic properties, similar sizes and relative positions between substrates and cyclases. On the contrary, no truncated products and β-amyrin were generated in L734X mutations. It is unusual that site-saturated mutagenesis of an aliphatic residue would lead to losing activities of the all mutated

We expected the function of L734 within the β-amyrin synthase may the same as I705 within SceERG7, because of their similar hydrophobic properties, similar sizes and relative positions between substrates and cyclases. On the contrary, no truncated products and β-amyrin were generated in L734X mutations. It is unusual that site-saturated mutagenesis of an aliphatic residue would lead to losing activities of the all mutated