Isoprenoid Biosynthesis: Manifold Chemistry Catalyzed by Similar Enzymes K Ulrich Wendt and Georg E Schulz*

View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Elsevier - Publisher Connector

Minireview 127

Isoprenoid biosynthesis: manifold chemistry catalyzed by similar enzymes K Ulrich Wendt and Georg E Schulz*

Isoprenoids comprise a family of more than 23,000 natural The volatile monoterpenes are common fragrances and products, among them the precursors of cholesterol and flavors. Sesquiterpenes and diterpenes serve as phero- taxol. The structures of three isoprenoid-cyclizing enzymes mones, defensive agents, visual pigments, antitumor drugs have recently been determined and here are placed in the and components of signal transduction networks. The greater context of isoprenoid biosynthesis. On the basis of triterpenes are important membrane constituents and pre- reaction mechanisms, a subdivision into class I and class II cursors of the steroid hormones and bile acids. Higher enzymes is proposed. The chain folds of these enzymes oligoterpenes function as photoreceptive agents, cofactor also appear to fall into the same two distinct classes, sidechains and they constitute natural polymers. suggesting that class I and class II enzymes have evolved separately. The backbone of isoprenoid biosynthesis is formed by a chain of linear polyunsaturated allylic diphosphate (previ- Address: Institut für Organische Chemie und Biochemie, Albertstr. 21, ously named pyrophosphate) esters that contain increasing D-79104 Freiburg im Breisgau, Germany. numbers of isoprene (C5) units. Beginning with dimethyl- *Corresponding author. allyl diphosphate (DMAPP), the esters are formed by sub- E-mail: [email protected] sequent 1′–4 addition of isopentenyl diphosphate (IPP) to the growing chain (Figure 1; [1]). This process is catalyzed Structure 6 15 February 1998, :127–133 by chain length-selective prenyl transferases. Further http://biomednet.com/elecref/0969212600600127 linear compounds are generated by head-to-head conden- © Current Biology Ltd ISSN 0969-2126 sations of the resulting isoprenyl diphosphates, yielding, for example, squalene. The structural variety within the Isoprenoids, also named terpenoids, are the chemically isoprenoid family is produced by numerous terpene most diversified family of low molecular mass natural prod- cyclases that transform the linear precursors into the vast ucts, consisting of more than 23,000 currently identified number of cyclic products using cationic mechanisms compounds with various biological functions (Figure 1). (Figure 2).

Figure 1

Scheme of isoprenoid biosynthesis. From top to bottom prenyl transferases add successive C5-building blocks to form linear terpenes of various lengths. Furthermore, head-to-head condensations of C15 and C20-blocks give rise to squalene and phytoene, respectively. Numerous cyclases transform the linear terpenes into a great variety of cyclic compounds, common examples of which are depicted at the right-hand side. On the basis of the initial carbocation formation, two classes of enzymes can be proposed: class I produces an allylic carbocation species by abstracting a pyrophosphate, whereas class II creates a carbocation species by protonating either an epoxy group or a double bond. The common abbreviations are used — DMAPP, IPP, GPP, FPP and GGPP — which are derived from their earlier names (e.g. DMAPP, dimethylallyl pyrophosphate).

Structure 128 Structure 1998, Vol 6 No 2

Figure 2 Table 1

Quantitative chain fold comparisons among the known structures of terpenoid synthases*. DMA PP IPP GPP

OPP + OPP OPP PLS EAS (C-term) SHC (minor) EAS (N-term) :B1 DDXXD: Mg2+ FPS 106 (11) 118 (10) – –

5-epi-aristolochene PLS – 165 (16) – – FPP B2:H SHC (major) – – 108 (23) 70 (11) SHC (minor) – – – 47 (13) OPP :B1 *Given are the number of residues in identical sequential order that DDXXD: Mg2+ superimpose within a cut-off distance of 3 Å (minimum segment length was three residues). The percentage of identical residues in the FPP H:B1 pentalenene H aligned segments is stated in parentheses. All superpositions were H :B1 initialized by a search using program DALI [19]. H OPP

DDXXD: Mg2+ With respect to the first reaction step, the enzymes of isoprenoid biosynthesis can be subdivided into two classes. squalene hopene Class I enzymes generate a comparatively stable allylic carbocation species by the release of a pyrophosphate group. This class includes the prenyl transferases, the B : B1:H 2 monoterpene and sesquiterpene cyclases, as well as many Structure diterpene cyclases (Figure 1). Enzymes of this class contain a DDXXD-sequence motif that binds the diphos- Scheme of the reactions catalyzed by the four enzymes discussed in phate moiety of the substrates via Mg2+ ions, facilitating this article. From top to bottom are the reactions catalyzed by FPS pyrophosphate release. Class II enzymes generate the (only first step of the reaction shown), EAS, PLS (all class I) and SHC carbocation species by protonating a C–C double bond or (class II). Class I uses Mg2+ ions bound to the DDXXD-sequence motif for producing the initial carbocation. The initial protonating acid (B ) the corresponding epoxide. It includes the triterpene and 1 β and the final deprotonating base (B2) of the class II enzyme are tetraterpene cyclases involved in cholesterol and - indicated. The common abbreviations are used. carotene biosynthesis.

Structural homology within class I enzymes Although the complex stereochemistry of isoprenoid The structurally characterized prenyl transferase, farnesyl biosynthesis has been intensively investigated [2–4], diphosphate synthase (FPS) from chicken, and two numerous details of these enzymatic reactions remained sesquiterpene cyclases, 5-epi-aristolochene synthase (EAS) unclear. For example, how do the terpene cyclases enforce from tobacco [7] and pentalenene synthase (PLS) from the required specific and strained substrate conformations? Streptomyces [8], belong to class I. All these enzymes How do the enzymes create, stabilize and finally quench consist predominantly of α helices, five of which surround the highly reactive carbocations? Are the resembling reac- their respective active centers. The chain fold, which is tions catalyzed by similar enzymes? With the known struc- essentially identical in these three enzymes (Figure 3), has ture of a prenyl transferase [5–6] and the recently reported been named the terpenoid synthase fold [8]. We suggest structures of two sesquiterpene cyclases [7–8] and one that this name should be changed to a more specific one, triterpene cyclase [9], structural biology has provided novel such as class I terpenoid synthase fold or, on the safer side, data in a field previously dominated by organic chemistry. FPS fold, because the chain fold of a squalene cyclase differs grossly from that of FPS (see below). Catalysis Enzymatic terpene cyclization can be subdivided into In a quantitative analysis, we superimposed the chain three phases: a reactive carbocation is generated (first folds and counted the number of equivalent residues phase), which then steps through the substrate chain within 3 Å distance, yielding a measure for a relationship. (often involving de- and reprotonation of the intermedi- The results are given in Table 1, demonstrating that the ates) and results in a terminal carbocation (second phase), structure of the catalytic C-terminal domain of EAS is which is quenched by a base (third phase). The cycliza- quite closely related to the structure of PLS, and both tion reactions catalyzed by the terpene cyclases are special EAS and PLS are less well related to FPS. Except for the cases of cationic SN1-like alkylations [1] and thus corre- common DDXXD-motif, the homology between the spond to the prenyl transferase reaction (Figure 2). three enzymes can barely be detected in their sequences Minireview Isoprenoid biosynthesis Wendt and Schulz 129

Figure 3

(a) (b) FPS

DDXXD

A DDXXD

EAS (C-term)

A DDXXD

C N

PLS

A DDXXD

Structure

The common chain fold of three enzymes of isoprenoid biosynthesis. monomeric PLS (bottom). (b) Topology diagrams of the same three (a) The three chains were superimposed and then vertically separated chain folds in the same order, viewed approximately from the top of the for clarity. The chain ends are labeled. The DDXXD-sequence motifs ribbon plot in (a). The α helices run upward (green), downward (blue) are shown in blue. The active-center pockets are marked by an A. and horizontally (brown). The common helices (criteria of Table 1) are Stereo views are shown for one subunit of the dimeric FPS (top); the marked with diagonal lines. C-terminal catalytic domain of the monomeric EAS (center); and the 130 Structure 1998, Vol 6 No 2

Figure 4

(a) (b) SHC (major)

37 306

SHC (minor)

305

EAS (N-term)

Structure

Related folding patterns in the two domains of SHC and in one domain SHC, shown with yellow inner and red outer helices. The of EAS. (a) The chains have been superimposed and then separated QW-sequence motifs are shown in green. In the related folds below, vertically for clarity. Stereo views are shown for the major domain of equivalent helices are yellow and red, respectively, whereas additional αα SHC forming an ( )6-barrel (top); the minor domain of SHC (center); helices are white. (b) Topology sketches of the same three chain folds and the N-terminal domain of EAS, the function of which is still in the same order and with the same coding as in Figure 3b. Hexagons αα) unknown (bottom). The most symmetric fold is the ( 6-barrel of denote 310 helices.

because the percentage of identical residues is 16% and Chain fold comparisons with a class II enzyme lower in the superimposed parts (Table 1). Interestingly Squalene-hopene cyclase (SHC) from Alicyclobacillus is a enough, taxadiene synthase [10], which produces the class II enzyme. It consists of two domains (Figure 4): the αα precursor of the anticancer drug taxol, has 30% major domain is a regular ( )6 barrel that resembles the sequence identity with EAS and should therefore barrels of several glucanases [11] and a protein-farnesyl assume a similar structure. transferase [12]; the minor domain assumes a similar Minireview Isoprenoid biosynthesis Wendt and Schulz 131

Figure 5

The active centers of three enzymes involved in isoprenoid biosynthesis. (a) View into the active center of PLS showing the DDXXD- sequence motif (red) and the nonpolar bottom, where the farnesyl moiety of FPP is forced into the conformation required for the initial cyclization step (reproduced with permission from [8]). (b) Geometry of the diphosphate moiety of an FPP analogue bound to EAS, showing three Mg2+ ions bound to the phosphate groups (reproduced with permission from [7]). (c) The active center of SHC [9] with the modeled product hopene. The aromatic sidechains surrounding hopene probably stabilize intermediate carbocation species. In the initial step, Asp376 protonates the 2,3-double bond of the substrate squalene. The polar sidechains (red) at the bottom hold a water molecule (not shown) that is the final deprotonating base.

αα-barrel fold. In a structural alignment of the two that is common to all triterpene cyclases [13]. These QW- domains, 108 residues superimpose within 3 Å and the motifs enchain the outer helices of the barrels by numer- aligned segments have as much as 23% sequence identity, ous interactions [9]. The N-terminal domain of the class I which points to a gene duplication. Both domains contain a enzyme EAS has a similar αα-barrel fold (Figure 4). The repetitive QW-sequence motif at their surfaces (Figure 4a) function of this domain is still unknown. The comparisons 132 Structure 1998, Vol 6 No 2

show that it resembles the major domain of SHC more gous to the family of 2,3-oxidosqualene cyclases that pro- closely than the minor domain (Table 1), but both relation- tonate the corresponding epoxide, catalyzing the key step ships are marginal. The protein-farnesyl transferase [12] in cholesterol biosynthesis. SHC is an integral membrane has not been considered, because it is not involved in iso- protein that submerges in one of the two layers of a mem- prenoid biosynthesis and also follows a different mecha- brane and is therefore named monotopic [9]. In contrast, nism using a Zn2+ ion to alkylate a thiol group. The SHC the sesquiterpene cyclases are water-soluble. The active fold is distinct from those of FPS, PLS and the C-terminal center of SHC is in a large cavity located in the middle of catalytic domain of EAS, suggesting that the class I and the molecule (Figure 4). This cavity is connected by a class II enzymes have evolved separately. nonpolar channel to the nonpolar membrane moiety, per- mitting uptake from the membrane interior where the The prenyl transferase FPS substrate squalene is dissolved. Because of the low water The reaction catalyzed by FPS is given in Figure 2. The solubility of their substrates, it can be expected that other active center of this enzyme contains a large hydrophobic triterpene and higher oligoterpene cyclases are also inte- pocket, accommodating the growing isoprenyl chain [6]. gral membrane proteins with monotopic characteristics. Two conserved DDXXD-sequence motifs [14] are located 12 Å apart on opposite walls of the active center. Binding Using mutagenesis studies, labeling experiments and studies [6] have shown that one of these motifs accommo- structure analysis, the catalytic acid residue of SHC that dates the allylic diphosphate groups of DMAPP, GPP and initiates the cyclization cascade has been identified as FPP (Figure 1) and binds Mg2+ ions that are essential for Asp376 [9,16–18]. The cascade is terminated by a polar- the initial reaction step [5]. At the second DDXXD-motif, ized water molecule [9]. The active center of SHC is the chain-extending IPP unit can be modeled in agreement depicted in Figure 5c. As observed in the sesquiterpene with the known stereochemistry of the alkylation [1]. The cyclases [7–8], several conserved aromatic residues line product chain length resulting from reactions catalyzed by the active-center cavity of SHC, where they could stabi- prenyl transferases appears to be controlled by the size of lize intermediate carbocations. the enzyme’s hydrophobic pocket, because enlarging this pocket in mutants gave rise to longer chains [6]. Perspectives In conclusion, the enzymes of isoprenoid biosynthesis cat- The sesquiterpene cyclases EAS and PLS alyze their reactions in deep hydrophobic pockets, thereby EAS [7] and PLS [8] are involved in the biosynthesis of protecting the highly reactive cationic intermediates from the antifungal agent, phytoalexin capsidiol, and of pental- premature quenching by water and also preventing inad- enolactone antibiotics, respectively. The active centers of vertent enzyme alkylation. The enzymes furthermore both enzymes are deep hydrophobic pockets as shown in enforce strained substrate conformations and channel the detail for PLS in Figure 5a. Both EAS and PLS contain a reaction by providing appropriate groups that favor the DDXXD-motif that acts in pyrophosphate release [7–8]. production of the required intermediates. The known The location of this motif corresponds to that of the enzyme structures have a number of features in common. allylic-binding site of FPS (Figure 3). The diphosphate- The new structural data open the field of enzyme engi- binding geometry at the active center of EAS is shown in neering for producing of novel compounds, which consti- Figure 5b. The arrangement contains as many as three tutes a new type of bio-organic synthesis. First Mg2+ ions. Nearby arginine residues have been suggested engineering attempts have been reported for FPS and to pull the nucleophilic pyrophosphate away from the EAS [6,7]. developing carbocation [6–8]. Acknowledgements The mechanisms of the two sesquiterpene cyclases have We thank CA Lesburg and DW Christianson for the coor- been proposed on the basis of earlier chemical data dinates of PLS, CM Starks and JP Noel for the coordi- (Figure 2). Both enzymes stabilize intermediate cations nates of EAS, and K Poralla and C Vonrhein for either by carbonyl dipoles or aromatic quadrupoles discussions. This work was supported by the Deutsche [7,8,15]. The required substrate conformation seems to be Forschungsgemeinschaft under contract SFB-388. enforced mainly by the aromatic residues in the active centers. Interestingly enough, a tryptophan residue has References been suggested to function as the catalytic base terminat- 1. Poulter, C.D & Rilling, H.C. (1978). The prenyl transferase reaction, enzymatic and mechanistic studies of the 1′–4 coupling reaction in ing the reaction catalyzed by EAS [7]. the terpene biosynthetic pathway. Acc. Chem. Res. 11, 307-313. 2. Croteau, R. (1987). Biosynthesis and catabolism of monoterpenoids. Chem. Rev. 87, 929-954. The triterpene cyclase SHC 3. Cane, D.E. (1990). Enzymatic formation of sesquiterpenes. Chem. SHC, a class II enzyme, generates the initial cation by Rev. 90, 1089-1103. 4. Abe, I., Rohmer, M. & Prestwich, G.D. (1993). Enzymatic cyclization of direct protonation of a terminal double bond in squalene squalene and oxidosqualene to sterols and triterpenes. Chem. Rev. (Figure 2). It is sequence-related and therefore homolo- 93, 2189-2206. Minireview Isoprenoid biosynthesis Wendt and Schulz 133

5. Tarshis, L.C., Yan, M., Poulter C.D. & Sacchettini, J.C. (1994). Crystal structure of recombinant farnesyl diphosphate synthase at 2.6-Å resolution. Biochemistry 33, 10871-10877. 6. Tarshis, L.C., Proteau, P.J., Kellogg, B.A., Sacchettini, J.C. & Poulter, C.D. (1996). Regulation of product chain length by the isoprenyl diphosphate synthases. Proc. Natl. Acad. Sci. USA. 93, 15018- 15023. 7. Starks, C.M., Back, K., Chappel, J. & Noel, J.P. (1997). Structural basis for cyclic terpene biosynthesis by tobacco 5-epi-aristolochene synthase. Science 277, 1815-1820. 8. Lesburg, C.A., Zhai, G., Cane, D.E. & Christianson, D.W. (1997). Crystal structure of pentalenene synthase: mechanistic insights on terpenoid cyclization reactions in biology. Science 277, 1820-1824. 9. Wendt, K.U., Poralla, K. & Schulz, G.E. (1997). Structure and function of a squalene cyclase. Science 277, 1811-1815. 10. Wildung, M.R. & Croteau, R. (1996). A cDNA clone for taxadiene synthase, the diterpene cyclase, that catalyzes the committed step of taxol biosynthesis. J. Biol. Chem. 271, 9201-9204. 11. Sakon, J., Irwin, D., Wilson, D.B. & Karplus, P.A. (1997). Structure and mechanism of endo/exocellulase E4 from Thermomonospora fusca. Nat. Struct. Biol. 4, 810-818. 12. Park, H.-W., Boduluri, S.R., Moomaw, J.F., Casey, P.J. & Beese, L.S. (1997). Crystal structure of protein-farnesyl transferase at 2.25 Ångstrom resolution. Science 275, 1800-1804. 13. Poralla, K., Hewelt, A., Prestwich, G.D., Abe, I., Reipen, I. & Sprenger, G. (1994). A specific amino acid repeat in squalene and oxidosqualene cyclases. Trends Biochem. Sci. 19, 157-158. 14. Chen, A., Kroon, P.A. & Poulter, C.D. (1994). Isoprenyl diphosphate synthases: comparisons, a phylognetic tree and predictions of secondary structure. Protein Sci. 3, 600-607. 15. Dougherty, D. (1996). Cation-p interactions in chemistry and biology: A new view of benzene, Phe, Tyr, and Trp. Science 271, 163-168. 16. Abe, I. & Prestwich, G.D. (1994). Active site mapping of affinity- labeled rat oxidosqualene cyclase. J. Biol. Chem. 269, 802-804. 17. Feil, C., Süssmuth, R., Jung, G. & Poralla, K. (1996). Site-directed mutagenesis of putative active site residues in squalene-hopene cyclase. Eur. J. Biochem. 242, 51-55. 18. Corey, E.J., Cheng, H., Baker, H.C., Matsuda, S.P.T., Li, D. & Song, X. (1997). Studies on the substrate binding segments and catalytic action of lanosterol synthase. Affinity labeling with a carbocation derived from mechanism-based analogs of 2,3-oxidosqualene and site-directed mutagenesis. J. Am. Chem. Soc. 119, 1289-1296. 19. Holm, L. & Sander, C. (1993). Protein structure comparison by alignment of distance matrices. J. Mol. Biol. 233, 123-138.