Structural Modeling and Site-Directed Mutagenesis of the Actinorhodin β-Ketoacyl-Acyl Carrier Protein Synthase

Abstract

A three-dimensional model of the Streptomyces coelicolor actinorhodin β-ketoacyl synthase (Act KS) was constructed based on the X-ray crystal structure of the related Escherichia coli fatty acid synthase condensing enzyme β-ketoacyl synthase II, revealing a similar catalytic active site organization in these two enzymes. The model was assessed by site-directed mutagenesis of five conserved amino acid residues in Act KS that are in close proximity to the Cys169 active site. Three substitutions completely abrogated polyketide biosynthesis, while two replacements resulted in significant reduction in polyketide production. ³H-cerulenin labeling of the various Act KS mutant proteins demonstrated that none of the amino acid replacements affected the formation of the active site nucleophile.

Streptomyces coelicolor produces actinorhodin (Act), an aromatic polyketide whose biosynthetic pathway is an important model for the study of type II polyketide synthase (PKS) systems (6). A key component in the Act PKS is a β-ketoacyl-acyl carrier protein (ACP) synthase (KS_α) which presumably forms a heterodimer with a similar protein, KS_β (previously referred to as chain length factor [14] and recently renamed chain initiation factor [1]). In conjunction with ACP, Act KS_αKS_β catalyzes sequential condensation of an acetyl-coenzyme A (CoA) starter unit and seven malonyl-CoA extender units to form an octaketide carbon chain. Although a growing number of studies have demonstrated the ability to manipulate Act PKS and other type II PKSs to generate new compounds (8, 10), relatively little is known about the structural framework for catalysis in these multienzyme systems. Here we report a three-dimensional structure model of Act KS_α based on the recently resolved X-ray crystal structure of Escherichia coli fatty acid synthase (FAS) β-ketoacyl synthase II (KAS II) (7) and its assessment by site-directed mutagenesis.

Structural modeling of Act KS_α.

The condensation reactions catalyzed by KS_α are conceptually the same as the chain elongation steps of fatty acid biosynthesis carried out by the condensing enzymes of bacterial FASs (15) (Fig. 1A). The crystal structure of E. coli FAS KAS II has recently been solved (7). This enzyme catalyzes the condensation reaction that leads to the elongation of palmitoleic acid (C16:1) to cis-vaccenic acid (C18:1) in fatty acid biosynthesis. Sequence alignment showed that Act KS_α and E. coli KAS II share 40% identity and 50% similarity (Fig. 1B). Assuming that the same enzymatic mechanism operates for these two proteins, this resemblance in amino acid sequence indicates that they share a similar protein folding pattern and a similar catalytic active site architecture. A three-dimensional model of Act KS_α was thus generated using the comparative protein modeling server SWISS-MODEL (5), based upon the coordinates of E. coli KAS II (accession code 1kas in the Protein Data Bank, Brookhaven National Laboratory, Upton, N.Y.) (see Fig. 3). The quality of the model has been assessed by using the 3D-1D profile verification method (13) and Prosa II (16), originally built within SWISS-MODEL, as well as by using Procheck (12). The results showed that there are no unfavorable contacts between atoms in this model and that the stereochemical quality of this model is comparable to that of the template structure of E. coli KAS II (data not shown), indicating that Act KS_α can fold like E. coli KAS II.

FIG. 1.

(A) Scheme showing the condensation reaction catalyzed by β-ketoacyl-ACP synthase. Steps are labeled as follows: 1, transfer of ACP-bound acyl group to the substrate-binding cysteine residue in KS results in a thioester; 2, a carbanion is generated through decarboxylation of the ACP-bound malonyl group; and 3, nucleophilic attack of this carbanion at the carbonyl carbon atom on the KS-bound thioester results in formation of a carbon-carbon bond. (B) Sequence alignment of Act KSα and E. coli FAS KAS II used to build the model of Act KSα. Asterisks and dots appear below identical and similar amino acid residues, respectively. Active site Cys residues and the five conserved residues are also highlighted.

FIG. 3.

Ribbon representation of the modeled three-dimensional structure of Act KS_α. The carbon backbone of the active site Cys169, the five conserved polar residues and Ser347 are in gray, and the side chain oxygen, nitrogen, and sulfur atoms are in red, blue, and yellow, respectively. The figure was prepared using SWISS-PdbViewer 5 and Setor 3.

Organization of the active site in the Act KS_α model.

Previous studies have already established that a universally conserved Cys residue in various β-ketoacyl synthases is the active site nucleophile where the nascent acyl chain is covalently linked (Fig. 1A) (4, 11, 17). In addition, there are five universally conserved residues that are presumably involved in catalysis of the condensation reaction or maintenance of a functional active site configuration (15). In the final energy-minimized Act KS_α structural model (see Fig. 3), the relative organization of the active site Cys169 residue and the five conserved polar amino acids (His309, His346, Lys341, Asp317, and Glu320) is very similar to that of E. coli KAS II, as analyzed by Swiss-PdbViewer (5). In summary, two His residues (His309 and His346) are in closest proximity to the active site Cys169, with the distances between the Nɛ atoms and the Sγ atoms being 4.5 and 3.2 Å, respectively. Lys341 is situated between the two His residues and is within the hydrogen bond distance to the backbone carbonyl oxygen of His309, but the most energy-favorable position of its amino group is pointing away from the Sγ atom of Cys169 (distance, 7.5 Å). Together with Cys169, these three basic residues are located in a solvent-accessible pocket that is lined predominantly by hydrophobic residues. Two acidic residues (Asp317 and Glu320) are located on an α helix near this active site pocket and can be involved in a network of hydrogen bonds that hold a strand containing His309, Gly310, Ser311, Gly312, and Thr313 to form part of the active site pocket.

Functional analysis of the five conserved amino acid residues in Act KS_α.

To assess this model and to examine the importance of the five conserved residues for the activity of Act KS_α, each of the five conserved residues in Act KS_α was replaced individually by a neutral amino acid. The changes include His309 to Asn, His346 to Gln, Lys341 to Gln, Asp317 to Asn, and Glu320 to Gln. An E. coli-Streptomyces shuttle vector pRM5 (14), which contains a subset of act biosynthetic genes that specify production of the polyketide aloesaponarin II (Aloe II) and its acidic form, 3,8-hydroxy-1-methyl-anthraquinone-2-carboxylic acid (DMAC), was used as the template for site-directed mutagenesis of actI-orf1 (encoding Act KS_α) by PCR (9). Replacement of selected amino acid residues was done individually with synthetic oligonucleotide primers, as follows: Asp317 was replaced by Asn (primer 1, 5′-ACCCGGCAGAACAACCGCCACGAGACAGC-3′), Glu320 was replaced by Gln (primer 2, 5′-CAGAACGACCGCCACCAGACAGCGGCGTA-3′), His309 was replaced by Asp (primer 3, 5′-ATCGACTACATCAACGCGAACGGCTCCGG-3′), His346 was replaced by Gln (primer 4, 5′-AACTCGATGGTCGGCCAGTCGCTGGGCGC-3′), and Lys341 was replaced by Gln (primer 5, 5′-TCGATCCAGTCGATGGTCGGCCACTCGCT-3′). A six-His tag was engineered at the C terminus of each protein to facilitate protein purification and provide a unique epitope that distinguished the tagged Act KS_α from its homologous protein KS_β. Control experiments demonstrated that fusion of the six-His tag at the C terminus of wild-type Act KS_α did not affect the function of this protein, as shown by the normal production of polyketide metabolites (data not shown).

S. coelicolor CH999 (14), a mutated strain of S. coelicolor with the entire act gene cluster deleted, was transformed with plasmids bearing each mutation (Table 1). Polyketide production in CH999 transformants containing the different constructs is summarized in Table 2. Briefly, replacement of two His residues (His309Asn and His346Gln) greatly impaired the function of Act KS_α, as indicated by the trace amount of Aloe II-DMAC produced by CH999/pDHS3301 (H309N) and CH999/pDHS3302 (H346Q), while replacement of the Lys residue (Lys341Gln) and two acidic residues (Asp317Asn and Glu320Gln) resulted in a completely inactive Act KS_α, as indicated by the loss of the production of Aloe II-DMAC or any other previously detected polyketide in CH999 containing pDHS3303 (K341Q), pDHS3304 (D317N), or pDHS3305 (E320Q). Immunoblotting analysis carried out with equal amounts of protein extracts from each culture with anti-six-His antibody demonstrated that each Act KS_α mutant protein was produced at a similar level (Fig. 2B), indicating that none of the mutations introduced had seriously affected the expression of the mutant Act KS_α genes or the stability of the protein. It is evident that individual replacement of the five conserved residues has a direct affect on the activity of Act KS_α, confirming the functional importance of each that is suggested by the modeled structure.

TABLE 1.

Plasmids and strains used in this study

Plasmid or strain	Description	Source or reference
Plasmid
pRM5	E. coli-Streptomyces shuttle vector containing a subset of act biosynthetic	14
pDHS3401	pRM5 derivative encoding wild type Act Act β-KSα with a six-His tag at its C-terminus	This work
pDHS3501	pDHS3401 derivative carrying gene encoding mutated Act β-KSα (His309→Asn)	This work
pDHS3502	pDSH3401 derivative carrying gene encoding mutated Act β-KSα (His346→Gln)	This work
pDHS3503	pDHS3401 derivative carrying gene encoding mutated Act β-KSα (Lys341→Gln)	This work
pDHS3504	pDHS3401 derivative carrying gene encoding mutanted Act β-KSα (Asp317→Asn)	This work
pDHS3505	pDHS3401 derivative carrying gene encoding mutated Act β-KSα (Glu320→Gln)	This work
Strain
S. coelicolor CH999	proA1 argA1 SCP1− SCP2−redE60 Δact	14

TABLE 2.

Polyketide production by S. coelicolor CH9999 containing various constructs

Expression system	Polyketide produced
CH999	—c
CH999/pRM5	Aloe II-DMACa
CH999/pDHS3401	Aloe II-DMAC
CH999/pDHS3501(His309-Asn)	Aloe II-DMACb
CH999/pDHS3502(His346-Gln)	Aloe II-DMACb
CH999/pDHS3503(Lys341-Gln)	—
CH999/pDHS3504(Asp317-Asn)	—
CH999/pDHS3505(Gln320-Gln)	—

^aDMAC is the acidic form of Aloe II. 

^bLevel of production was less than 1% of the wild-type production level. 

^cNo Aloe II-DMAC produced. 

FIG. 2.

(A) SDS–12% PAGE analysis of total protein extracts from cultures expressing wild-type and mutant Act KSα. Lane 1, prestained protein marker (from top to bottom: 43 kDa, 29 kDa, 18.4 kDa, and 14.3 kDa); lane 2, CH999/pDHS3501; lane 3, CH999/pDHS3502; lane 4, CH999/pDHS3503; lane 5, CH999/pDHS3504; lane 6, CH999/pDHS3505; lane 7, CH999/pDHS3401 (wild type); lane 8, CH999. (B) Western blot analysis of the gel used for panel A was carried out with anti-six His antibody. Lanes are as described for panel A. (C) Autoradiography of the SDS-PAGE gel of ³H-cerulenin-labeled Act KSα proteins. Lane 1, Act KSα mutant (H309N); lane 2, Act KSα mutant (H346Q); lane 3, Act KSα mutant (K341Q); lane 4, Act KSα mutant (E320Q); lane 5, Act KSα mutant (D317N); lane 6, wild-type Act KSα; lane 7, Ni-NTA column purified protein extracts from CH999.

In vitro labeling of purified Act KS_α with ³H-cerulenin.

Cerulenin, (2S,3R)-2,3-epoxy-4-oxo-7,10-dodecadienoylamide, is a mycotoxin produced by Cephalosporium caerulens that irreversibly inactivates various β-ketoacyl-ACP synthases by alkylating the substrate-binding Cys active site residue (4). The reaction between the epoxide group of cerulenin and the nucleophilic thiol of the substrate-binding Cys results in the covalent binding of cerulenin to β-ketoacyl synthase (KS). Thus, we used an in vitro ³H-cerulenin labeling assay to address the question of whether any of the mutations introduced into Act KS_α had affected the reactivity of the Cys169 active site residue. A 1-μg sample of each Ni-nitrilotriacetic acid (NTA) column-purified Act KS_α protein (the wild-type protein and each mutant protein) was incubated with 1 μCi of ³H-cerulenin at room temperature for 30 min, separated on a sodium dodecyl sulfide (SDS)–12% polyacrylamide gel electrophoresis (PAGE) gel, and then exposed to film. The results demonstrated that each of the mutant Act KS_α proteins can be labeled by ³H-cerulenin with efficiency equal to that of the native enzyme (Fig. 2C), indicating that none of the mutations affected the formation of the nucleophile in the substrate-binding Cys169 residue. As a negative control, protein extracts from CH999 [after running through Ni-NTA to remove WhiE KS_α, involved in spore pigment production (2)] revealed no labeling band.

Discussion and conclusion.

In the modeled structure of Act KS_α, the side chain of the active site Cys169 is very close to the side chain of two basic residues: His309 and His346. This organization implies that either one or both of these residues can serve as the base that abstracts a proton from the Sγ atom in the active site residue and thus enhances its nucleophilicity. Accordingly, mutational analysis showed that replacement of either His309 or His346 in Act KS_α caused a dramatic loss of polyketide production. However, the strains carrying either replacement still produce trace amounts of Aloe II, suggesting that these two residues can at least partially complement each other in function. Furthermore, Act KS_α bearing the His309Asn or His346Gln mutation can still bind ³H-cerulenin, indicating that formation of the nucleophile in Cys169 has not been disrupted in either mutant. It is thus likely that these two residues can complement each other in aiding the formation of the nucleophile substrate-binding thiol. However, cerulenin is an irreversible inhibitor and the assay can not reveal any potential changes in the cerulenin-binding rate caused by the introduced mutation.

Interestingly, the modeled structure suggests that Lys341, Asp317, and Glu320 are not directly involved in the function of Cys169. In agreement with this prediction, none of the replacements involving these three residues affected the ability of the active site Cys169 to react with cerulenin. However, the modeled Act KS_α structure does indicate that each of the three residues can be involved in a hydrogen bond network that appears to be critical for securing the active site geometry. Modeling of the mutated enzyme indicated that replacement of Lys341, Asp317, and Glu320 with the residues chosen in this study would either disrupt or reduce these hydrogen bonds (data not shown) and thus result in inactive proteins, as supported by the fact that the corresponding strains of S. coelicolor completely lost polyketide production. However, the precise identification of the roles of these residues to the structural integrity of the active site pocket will have to await further structural analysis of the wild-type and the mutant KS_α proteins.

It is worth noting that a partially conserved Ser347 is also located near the predicted active site pocket in this modeled structure; however, its side chain is pointing away from Cys169 in the active site pocket and does not appear to have any significant functional or structural role (Fig. 3). This configuration is consistent with previous site-directed mutagenesis studies, which showed that replacement of Ser347 with Leu resulted in only slightly reduced polyketide production (11). This result, taken together with the studies described above, provides strong support for the modeled active site structure of Act KS_α.

The results presented here have confirmed the functional importance of five amino acids that are uniformly conserved in various β-ketoacyl-ACP synthases, as originally inferred from multiple sequence alignments of various KSs and the crystal structure of E. coli KAS II. Our data also strongly support the claim that β-ketoacyl-ACP synthases from different types of FASs and PKSs may share a common protein folding pattern and a common molecular architecture for catalysis. It is evident that, in the future, the structure of various type II β-ketoacyl-ACP synthases producing different polyketides could be modeled in a similar way. Comparison of these structures might reveal the subtle structural differences around the substrate binding pocket that contribute to the creation of structural diversity in polyketide products and contribute to the rational design of enzyme structures for the production of novel polyketides.