Phenolic lipid synthesis by type III polyketide synthases is essential for cyst formation in Azotobacter vinelandii

Nobutaka Funa*, Hiroki Ozawa*, Aiko Hirata†, and Sueharu Horinouchi*‡

*Department of Biotechnology, Graduate School of Agriculture and Life Sciences, University of Tokyo, Bunkyo-ku, Tokyo 113-8657, Japan; and †Department of Integrate Biosciences, Graduate School of Frontier Sciences, University of Tokyo, Kashiwa, Chiba 277-8562, Japan

Edited by Arnold L. Demain, Drew University, Madison, NJ, and approved March 1, 2006 (received for review December 28, 2005) Cysts of Azotobacter vinelandii are resting cells that are sur- substrates onto a ‘‘starter substrate,’’ whereas iterative type I rounded by a protective coat, conferring resistance to various PKSs use a single ketosynthase domain repeatedly for several chemical and physical agents. The major chemical components of times of extension (4). Type II PKSs consist of a ‘‘minimal PKS,’’ the cyst coat are alkylresorcinols, which are amphiphilic molecules and their auxiliary subunits contain a single set of iteratively used possessing an aromatic ring with a long aliphatic carbon chain. active sites that are carried on separate proteins (5). Type III Although alkylresorcinols are widely distributed in bacteria, fungi, PKSs are ketosynthases with a homodimeric form that act plants, and animals, no enzyme systems for their biosynthesis are iteratively for polyketide chain extension (6). Chalcone synthase known. We report here an ars operon in A. vinelandii that is (CHS) (Fig. 1B), responsible for flavonoid synthesis in plants, is responsible for the biosynthesis of the alkylresorcinols in the cysts. a member of type III PKSs (6). Stilbene synthase (STS) and The ars operon consisted of four genes, two of which encoded a p-coumaroyltriacetic acid synthase (CTAS), both sharing Ͼ70% type III polyketide synthase, ArsB and ArsC. In vitro experiments amino acid sequence identity to CHS, catalyze the same iterative revealed that ArsB and ArsC, sharing 71% amino acid sequence condensation to yield tetraketide intermediates in the same identity, were an alkylresorcinol synthase and an alkylpyrone manner as CHS (Fig. 1B). However, the mode of ring folding synthase, respectively, indicating that ArsB and ArsC are not differs among these three type III PKSs. STS cyclizes tetraketide isozymes but enzymatically distinct polyketide synthases. In addi- intermediates via intramolecular C2-to-C7 aldol condensation, tion, ArsB and ArsC accepted several acyl-CoAs with various and CTAS catalyzes intramolecular C5 oxygen-to-C1 lactoniza- lengths of the side chain as a starter substrate and gave corre- tion (Fig. 1B). Because the products of STS differ from alkyl- sponding alkylresorcinols and alkylpyrones, respectively, which resorcinols only in their moiety derived from the starter substrate suggests that the mode of the ring folding is uninfluenced by the and because type III PKSs often produce monocyclic structure of the starter substrates. The importance of the alkylres- polyketides, we expected that a type III PKS(s) would be orcinols for encystment was confirmed by gene inactivation ex- involved in the formation of the alkylresorcinols in the cysts of periments; the lack of alkylresorcinols synthesis caused by ars A. vinelandii. mutations resulted in the formation of severely impaired cysts, as This report deals with the enzyme system responsible for the observed by electron microscopy. biosynthesis of the alkylresorcinols and alkylpyrones localized in the cysts of A. vinelandii, together with the biological importance he genus Azotobacter, comprising Gram-negative, nitrogen- of these phenolic lipids for cyst formation. As we expected, the Tfixing soil bacteria, produces phenolic lipids, which are a phenolic lipids were shown to be synthesized by type III PKSs. mixture of alkylresorcinols and alkylpyrones with various lengths We named the operon encoding the type III PKSs alkylresorcinol of their side chain (Fig. 1A). Alkylresorcinols, distributed widely synthesis (ars) (Fig. 2A). In vitro analysis of these type III PKSs in bacteria, fungi, plants, and animals, play several roles that are revealed that ArsB and ArsC encoded in the operon were an concerned with cellular biochemistry and membrane chemistry alkylresorcinol synthase and an alkylpyrone synthase, respec- (1). For example, 5-alkylresorcinols are readily incorporated into tively. To our knowledge, this is the first report on the enzymes phospholipid bilayers and biological membranes, thereby caus- responsible for alkylresorcinol synthesis. An additional impor- ing considerable changes in their structure and properties (1). tant finding in this study is that the phenolic lipids are essential They also act as modulators of oxidation of liposomal mem- for the encystment in A. vinelandii. branes and fatty acids (1). When Azotobacter vinelandii differ- entiates to form metabolically dormant cysts, the phospholipids Results and Discussion in the membrane are replaced by 5-alkylresorcinols (2). This Isolation and Identification of Phenolic Lipids from Cysts of A. vine- unique membrane matrix is presumed to contribute to the landii. The phospholipids in the membrane of the vegetative cells physiology and desiccation resistance of the cyst (2) and, there- of A. vinelandii are replaced by alkylresorcinols and alkylpyrones fore, is sometimes called a ‘‘cyst coat.’’ Because of high incor- when it forms metabolically dormant cysts (2). We prepared a poration of 14C-labeled acetate into alkylresorcinols in A. vine- lipid fraction from mature cysts of A. vinelandii OP and analyzed landii (3), they were thought to be synthesized through a it by HPLC (Fig. 2B). The major compound migrating at a polyketide biosynthetic pathway. However, no enzyme systems retention time of 8.9 min was identified to be 5-heneicosylres- for the biosynthesis of alkylresorcinols, not only in Azotobacter orcinol (4b) by proton and carbon NMR spectrometry and liquid but also in other organisms, have been elucidated. Polyketides are synthesized by polyketide synthases (PKSs) that fall into three groups, type I to type III. All three bacterial Conflict of interest statement: No conflicts declared. PKSs possess a ketosynthase activity that catalyzes the sequential This paper was submitted directly (Track II) to the PNAS office. head-to-tail incorporation of acetate building blocks into a Abbreviations: ACP, acyl carrier protein; ars, alkylresorcinol synthesis; CHS, chalcone syn- growing chain. Type I PKSs are multifunctional enzymes that are thase; LC-APCIMS, liquid chromatography–atmospheric pressure chemical ionization mass organized into modules, each of which harbors a set of distinct spectrometry; PKS, polyketide synthase; STS, stilbene synthase. catalytic domains (4). Modular type I PKSs assemble polyketides ‡To whom correspondence should be addressed. E-mail: [email protected]. by using each ketosynthase domain once to add many extender © 2006 by The National Academy of Sciences of the USA

6356–6361 ͉ PNAS ͉ April 18, 2006 ͉ vol. 103 ͉ no. 16 www.pnas.org͞cgi͞doi͞10.1073͞pnas.0511227103 Downloaded by guest on September 25, 2021 Fig. 1. Reactions catalyzed by type III PKSs. (A) Synthesis of alkylresorcinol by ArsB and alkylpyrone by ArsC. (B) Reactions of plant type III PKSs. CHS, STS, and p-coumaroyltriacetic acid synthase (CTAS) use common tetraketide intermediates, which are derived from p-coumaroyl-CoA and malonyl-CoA.

chromatography-atmospheric pressure chemical ionization mass from an 8-day culture without an induction of encystment (data spectrometry (LC-APCIMS) analyses. The molecular masses of not shown). 3b and 2b were 448 and 406, respectively, on the basis of the [M Ϫ H]Ϫ ions on LC-APCIMS analysis, suggesting that 3b and 2b Gene Cluster Responsible for Biosynthesis of the Phenolic Lipids. A were ␣-pyrones. Their structures were verified as 6-heneicosyl- BLAST search in the protein database of A. vinelandii OP using 4-hydroxy-2-pyrone (2b) and 4-hydroxy-6-(2Ј-oxotricosyl)-2- the alfalfa CHS sequence (8) as a query revealed the presence pyrone (3b) by their comigration with the synthetic standards of two type III PKSs: one (named ArsB) sharing 23% amino acid (data not shown). Although 2b (a triketide pyrone), 3b (a sequence identity to the CHS and the other (ArsC) sharing 25%

tetraketide pyrone), and 4b (a resorcinol) had different struc- identity (Fig. 2A). arsB and arsC appeared to form an operon, MICROBIOLOGY tures from one another, their alkyl chains were presumed to be together with a putative type I fatty acid synthase (arsA) and a derived from a common starter substrate, n-behenyl-CoA (1b), functionally unknown protein (arsD). ArsD contains a 4Ј- for a PKS(s). In addition, we also assumed that 2a, 3a, and 4a phosphopantetheinyl transferase domain that would catalyze the were polyketides that were derived from n-heneicosyl-CoA (1a) posttranslational modification of acyl carrier protein (ACP) by as a starter, because the molecular masses of 2a, 3a, and 4a were the covalent attachment of the 4Ј-phosphopantetheine moiety of 434, 476, and 432, respectively. Comparison of their fragmen- CoA to a conserved serine (9), in addition to a functionally tation patterns with those from n-behenyl-CoA in LC-APCI unknown N-terminal domain of Ϸ400 aa in length. tandem mass analysis predicted that their carbon skeletons were We disrupted arsB by replacing the arsB sequence by a a triketide pyrone (2a), a tetraketide pyrone (3a), and a resor- kanamycin resistance gene to determine possible involvement of cinol (4a). These data show that A. vinelandii OP predominantly this gene cluster in the biosynthesis of the phenolic lipids (Fig. produces phenolic lipids as a mixture of alkylresorcinols and 2A). It seemed likely that the arsB::aphII mutation was polar on alkylpyrones with saturated side chains ranging from C21 to C23, the arcCD expression. This polar effect might result in the as was found for other Azotobacter species (2, 7). There were no inactivation of the entire gene cluster, because the 4Ј- significant differences in the amount or composition of phenolic phosphopantetheinyl transferase domain of ArsD is presumed to lipids when the lipid extract was prepared from the ␤-hydroxy- be crucial for the modification of the ACP domain of ArsA. As butyrate-induced cells and the aging cells that were harvested expected, no phenolic lipids were produced by the arsB::aphII

Funa et al. PNAS ͉ April 18, 2006 ͉ vol. 103 ͉ no. 16 ͉ 6357 Downloaded by guest on September 25, 2021 Fig. 2. Requirement of the ars operon for phenolic lipid synthesis. (A) Gene organization of the ars operon in A. vinelandii OP. ArsA (accession no. ZP࿝00418324) and ArsD (accession no. ZP࿝00418327) are similar in amino acid sequence to type I PKSs. ArsB (accession no. ZP࿝00418325) and ArsC (accession Fig. 3. Radio-TLC analysis of products synthesized by ArsB (A) and ArsC (B) no. ZP࿝00418326) are homologous to type III PKSs. The arsB::aphII mutant was from various acyl-CoA starter substrates and [2-14C]malonyl-CoA. The starter constructed by replacing the arsB sequence with the kanamycin-resistance substrates used were acetyl-CoA [1l] (lane 2), butyryl-CoA [1k] (lane 4), hex- gene, aphII.(B) HPLC chromatograms of lipid extracts prepared from ␤- anoyl-CoA [1j] (lane 6), octanoyl-CoA [1i] (lane 8), decanoyl-CoA [1h] (lane 10), hydroxybutyrate-induced cells of A. vinelandii and the arsB::aphII mutant. lauroyl-CoA [1g] (lane 12), myristoyl-CoA [1f] (lane 14), palmitoyl-CoA [1e] (lane 16), stearoyl-CoA [1d] (lane 18), arachidoyl-CoA [1c] (lane 20), and behenyl-CoA [1b] (lane 22). Lane m, control incubation without a starter mutant (Fig. 2B), indicating that the ars operon is responsible for substrate. the biosynthesis of the phenolic lipids.

In Vitro Analysis of ArsB and ArsC Reactions. It was particularly Interestingly, the kcat͞Km value of ArsB for n-behenyl-CoA interesting to determine whether ArsB and ArsC are isozymes or consumption is similar to that of ArsC, although A. vinelandii OP not, because ArsB and ArsC shared 71% amino acid sequence predominantly produced alkylresorcinols in vivo. Therefore, the identity. We prepared ArsB and ArsC with a His-tag at their in vitro catalytic abilities of ArsB and ArsC do not reflect the N-termini from Escherichia coli carrying these genes under the amounts of alkylresorcinol and alkylpyrones in vivo. This differ- control of the T7 promoter in pET vector. Both ArsB and ArsC ence may result from a decrease of alkylpyrones due to further migrated at a position of Ϸ43 kDa on SDS͞polyacrylamide gel metabolism or the instability of the compound itself. electrophoresis (data not shown). We first assayed ArsB by using n-behenyl-CoA (1b) as a starter substrate, because the major Substrate Specificity of ArsB and ArsC. Type III PKSs of bacterial component of the phenolic lipids of A. vinelandii was 5-henei- and plant origins show rather promiscuous starter substrate cosylresorcinol (4b), as described above. Incubation of n- specificity (6). We examined the reactions of ArsB and ArsC 14 behenyl-CoA (1b) as a starter substrate and [2- C]malonyl-CoA toward CoA esters of C2 to C22 straight chain fatty acids. The as an extender substrate gave a single radioactive product (Fig. radio-TLC (Fig. 3), as well as the LC-APCIMS analyses (Table 3A, lane 22), which was identified as 5-heneicosylresorcinol (4b) 2) of the reaction products, revealed broad starter substrate by LC-APCIMS analysis of the product similarly synthesized specificities of ArsB and ArsC. ArsB accepted the C10 to C22 from nonradiolabeled substrates (Table 2, which is published as esters, whereas ArsC exhibited broader substrate specificity than supporting information on the PNAS web site). On the other ArsB. Neither the substrate specificities of the enzymes nor the hand, under the same reaction conditions, ArsC gave two product profiles of the reactions were affected by the change radioactive compounds different from 5-heneicosylresorcinol from pH 6 to pH 9 (data not shown). In contrast, the optimal pH (4b)intheRf values (Fig. 3B, lane 22). We identified them as was different depending on the starter substrates (results not 6-heneicosyl-4-hydroxy-2-pyrone (2b) and 4-hydroxy-6-(2Ј- shown). oxotricosyl)-2-pyrone (3b) by comparing their LC-APCIMS with Although ArsB and ArsC accepted a broad range of starter those of the synthetic standards. These results clearly show that substrates in vitro, the corresponding phenolic lipids were hardly ArsB and ArsC are type III PKSs with different catalytic detected in the lipid fraction of A. vinelandii (data not shown). properties and are responsible for the biosynthesis of alkylres- This strict selectivity is perhaps because of the availability of orcinols and alkylpyrones, respectively. starter CoA esters, which is presumably synthesized by the We measured the steady state kinetics parameters for 4b, 2b coaction of ArsA and ArsD. Very recently, Sankaranarayanan et and 3b synthesis by varying the concentration of n-behenyl-CoA al. (10) found a unique tunnel extending from the active site to (1b) (Table 1). ArsB and ArsC exhibited saturation kinetics in the surface of the protein in the crystal structure of PKS18, a response to the increasing concentration of n-behenyl-CoA (1b). type III PKS from Mycobacterium tuberculosis, which accounts

6358 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0511227103 Funa et al. Downloaded by guest on September 25, 2021 Table 1. Steady-state kinetic parameters* of the ArsB or ArsC reactions n-Behenyl-CoA (1b) Triketide pyrone (2b) Tetraketide pyrone (3b) Resorcinol (4b)

kcat kcat͞Km kcat kcat͞Km kcat kcat͞Km kcat kcat͞Km (ϫ 10Ϫ3 (ϫ secϪ1 (ϫ 10Ϫ3 (ϫ secϪ1 (ϫ 10Ϫ3 (ϫ secϪ1 (ϫ 10Ϫ3 (ϫ secϪ1 Ϫ1 Ϫ1 Ϫ1 Ϫ1 Ϫ1 Ϫ1 Ϫ1 Ϫ1 Enzyme Km (␮M) ⅐sec ) ⅐M ) Km (␮M) ⅐sec ) ⅐M ) Km (␮M) ⅐sec ) ⅐M ) Km (␮M) ⅐sec ) ⅐M )

ArsB 4.86 Ϯ 0.50 15.6 Ϯ 0.73 3,202 — — — — — — 4.86 Ϯ 0.50 15.6 Ϯ 0.73 3,202 ArsC 1.47 Ϯ 0.18 6.80 Ϯ 0.07 4,626 1.34 Ϯ 0.33 2.77 Ϯ 0.13 2,069 1.59 Ϯ 0.06 4.04 Ϯ 0.07 2,541 — — —

*Results are mean Ϯ SE (n ϭ 3).

for the binding of long-chain aliphatic substrates. PKS18 exhibits attached on ACP of ArsA. Alternatively, ArsA may contain a a broad specificity for C6 to C20 esters to produce tri- and domain that is responsible for transacylation of docosanoate tetraketide pyrones (11). The catalysis of PKS18 is assumed to from the enzyme to CoA, like the palmitoyl transferase domain be essentially the same as that of ArsC that accepts C6 to C22 of the yeast type I system (14). Taken together, we hypothesize esters (Fig. 3B), although they share only 23% identity in amino that the docosanoate bound to ACP of ArsA could be trans- acid sequence. We speculate that a tunnel similar to that in ferred hand-to-hand toward ArsB and ArsC, as illustrated in Fig. PKS18 is also present in both ArsB and ArsC. 4. Using the same tetraketide intermediate, ArsB catalyzes aldol A vivid contrast between ArsB and ArsC was that the former condensation to form alkylresorcinols, whereas ArsC catalyzes produced alkylresorcinols in all of the reactions, but the latter lactonization to form alkylpyrones (Fig. 4). produced pyrones. Therefore, the mode of ring folding is char- acteristic of the enzymes and is independent of the starter Effect of Inactivation of the ars Operon on Encystment. When the substrates incorporated in the polyketide intermediates. Al- glucose in the medium for A. vinelandii cultures is replaced by though both enzymes use the same tetraketide intermediate, the ␤-hydroxybutyrate or 1-butanol, a metabolic shift from carbo- mechanism of ring folding is different between ArsB and ArsC; hydrate to lipid occurs, leading to encystment (2). Although it is ArsB catalyzes intramolecular C2-to-C7 aldol condensation of known that Ͼ70% of lipids are replaced by alkylresorcinols and STS type to yield 5-heneicosylresorcinol (4b), whereas ArsC other phenolic lipids during encystment (3), there have been no catalyzes intramolecular C5 oxygen-to-C1 lactonization of p- studies concerning the requirement of the alkylresorcinol syn- coumaroyltriacetic acid synthase type to yield 4-hydroxy-6-(2Ј- thesis for encystment. A. vinelandii OP produced mature cysts oxotricosyl)-2-pyrone (3b) (Fig. 1 A and B). when cultured in the presence of ␤-hydroxybutyrate; the central Another difference in their reactions from n-behenyl-CoA body, intine, and exine in the cyst were distinct, as observed by (1b) is that the size of the ArsB product is strictly confined to the transmission electron microscopy (Fig. 5A). This strain accumu- tetraketide, whereas ArsC produces 6-heneicosyl-4-hydroxy-2- lated almost no poly-␤-hydroxybutyrate, in agreement with the pyrone (2b) that is synthesized by lactonization of the triketide previous observation that poly-␤-hydroxybutyrate accumulation intermediate (Fig. 1A). We assume that the hydrolysis of the is not essential for encystment (16, 17). Concerning phenolic thioester, which is linked to CoA or ArsB, takes place before the lipid production, the aged cells grown in the absence of ␤- aldol reaction (Fig. 1A), because no alkylresorcinolic acids were hydroxybutyrate and harvested from the stationary phase pro- detected by the LC-APCIMS analysis of the reaction mixture of duced the phenolic lipids in similar amounts as those grown in ArsB. It is unclear whether decarboxylation is accompanied by the presence of ␤-hydroxybutyrate, although they formed no the aldol reaction or occurs during the dehydration͞ cysts (data not shown). These observations are consistent with aromatization step after the aldol reaction. A similar catalytic the previous observation that the encystment occurs frequently property was found for the STS reaction in which stilbene in ␤-hydroxybutyrate-induced cells, whereas it occurs in Ͻ0.01% carboxylic acid was not detected (12). Furthermore, merulinic of the aged, glucose-grown cells (18). acid, an alkylresorcinolic acid supplemented to the culture of The importance of the phenolic lipids for encystment was Azotobacter chroococcum, was not incorporated into the corre- confirmed by the failure in encystment of the arsB mutant. As sponding alkylresorcinol. Therefore, enzymatic decarboxylation described earlier, the arsB::aphII mutant produced no phenolic of alkylresorcinolic acids seems unlikely (1). Recently, a novel lipids. Transmission electron microscopy showed that the mutant mechanism leading to hydrolysis of the thioester, which in turn results in an STS-type ring folding, has been proposed for the STS reaction on the basis of the crystal structure (12). The difference between ArsB and ArsC could be explained by MICROBIOLOGY the absence of the STS-like thioesterase activity in ArsC and the presence of this activity in ArsB.

A Proposed Route for Phenolic Lipid Synthesis. The de novo products by bacterial type II fatty acid synthases are released as ACP esters (13), and the fatty acids are directly transacylated from acyl-ACP to the membrane phospholipids by specific glycerol phosphate transacylases (14). In addition, the chain lengths of fatty acids in phospholipids of A. vinelandii are mainly C16 and C18 (15), which suggests the absence of n-behenyl-CoA (1b)in the primary metabolism. We therefore suppose that ArsA, which is similar in amino acid sequence to type I fatty acid synthases, is responsible for the synthesis of a starter substrate of ArsB and ArsC. Because ArsA contains no thioesterase domain, which cleaves a thioester linkage between fatty acids and ACP (4), Fig. 4. A proposed route for biosynthesis of the phenolic lipids in A. docosanoate, a plausible product of ArsA, perhaps remains vinelandii.

Funa et al. PNAS ͉ April 18, 2006 ͉ vol. 103 ͉ no. 16 ͉ 6359 Downloaded by guest on September 25, 2021 Isolation and Identification of 5-Heneicosylresorcinol. A. vinelandii OP was grown at 30°C for 12 days in Burk’s medium containing 2% glucose (2 liters ϫ 4). The cells were harvested by centrif- ugation, suspended in a small volume of water, and extracted with chloroform͞methanol (2:1, vol͞vol). After the cell debris had been filtered off, the filtrate was lyophilized to dryness. The crude material was dissolved in chloroform and separated by silica gel chromatography by using chloroform as a solvent. The major compound in the fraction was applied to reversed-phase preparative HPLC equipped with a Pegasil C4 column (10 ϫ 250 mm; Shenshu Scientific, Tokyo) and eluted by 95% CH3CN in water containing 0.1% trifluoroacetic acid at a flow rate of 3 ml͞min to provide 2 mg of 5-heneicosylresorcinol as a white 1 solid. H NMR (500 MHz, CDCl3) ␦ 6.22 (m, 2H, C4H, C6H), 6.15 (m, 1H, C2H), 2.46 (t, J ϭ 8 Hz, 2H, C1ЈH), 1.23 (m, 38H, methylene of C2Ј to C21Ј), 0.86 (t, J ϭ 7 Hz, 3H, C22ЈH); 13C NMR (125 MHz, CDCl3) ␦ 156.5 (C-1, C-3), 146.5 (C-5), 108.0 (C-4, C-6), 100.1 (C-2), 35.8 (C-1Ј), 31.9 (C-20Ј), 31.1 (C-2Ј), 29.7 (C4Ј to C19Ј), 29.4 (C3Ј), 22.7 (C21Ј), 14.1 (C22Ј); APCIMS: m͞z 405.3 ([M ϩ H]ϩ ion observed in a positive ion analysis), m͞z 403.2 ([M Ϫ H]Ϫ ion observed in a negative ion analysis).

-Synthesis of 6-Heneicosyl-4-hydroxy-2-pyrone and 4-Hydroxy-6-(2؅ oxotricosyl)-2-pyrone. The details of chemical synthesis are in Fig. 6, which is published as supporting information on the PNAS web site. We first prepared 1-eicosanal (22) and 4-benzyloxy-6- methyl-2-pyrone as reported in ref. 23 and used them for synthesis of 4-benzyloxy-6-(2Ј-hydroxyheneicosyl)-2-pyrone, which was then converted into 4-benzyloxy-6-(heneicos-1Ј-enyl)- 2-pyrone. Hydrogenation of the pyrone by 10% palladium on Fig. 5. Requirement of the phenolic lipid for encystment of A. vinelandii. carbon yielded 6-heneicosyl-4-hydroxy-2-pyrone. For synthesis Electron micrographs of an ultrathin section of the cyst of A. vinelandii OP (A) of 4-hydroxy-6-(2Ј-oxotricosyl)-2-pyrone, 4-benzyloxy-6-(2Ј- and a cell of the arsB::aphII mutant (B) are shown. ⌻he central body (CB) is hydroxytricosyl)-2-pyrone was prepared from 1-docosanal and surrounded by a multilayered cyst coat (exine, Ex) and an inner amorphous 4-benzyloxy-6-methyl-2-pyrone. Dess–Martin oxidation of 4- ␮ area (intine, In). (Scale bars: 0.5 m.) benzyloxy-6-(2Ј-hydroxytricosyl)-2-pyrone yielded 4-benzyloxy- 6-(2Јoxotricosyl)-2-pyrone, followed by deprotection to give 4-hydroxy-6-(2Ј-oxotricosyl)-2-pyrone. formed no mature cysts; the central body with no intine in a cyst-like cell was surrounded by a severely impaired exine (Fig. Synthesis of Behenyl-CoA. Behenyl-CoA was synthesized by a 5B). Thus, the phenolic lipid synthesis is essential for the mature modification of the procedure reported by Blecher (24). The cyst formation in A. vinelandii. There was no morphological details of chemical synthesis are in Fig. 7, which is published as differences between the aged cells of A. vinelandii OP and the supporting information on the PNAS web site. arsB::aphII mutant grown in the absence of ␤-hydroxybutyrate, when observed by optical and transmission electron microscopy Production and Purification of ArsB and ArsC. The nucleotide se- (data not shown). quence (ATCATG) covering the ATG start codon of arsB was changed to CATATG to create an NdeI site by PCR with primer Materials and Methods I: 5Ј-CGCGAATTCCATATGAGCAGTCCCCACAACGCA- Bacterial strains, Plasmids, and Media. E. coli JM109 and plasmid GTT-3Ј (with the nucleotides to be changed shown by the italic pUC19, used for DNA manipulation, were purchased from letters and an EcoRI site shown by underlining) and primer II: TaKaRa Biochemicals (Shiga, Japan). pET16b (Novagen) was 5Ј-CGCGGATCCATCGGCCAGGACCGCGCT-3Ј (with a used for expression of His-tagged proteins in E. coli BL21 (DE3). BamHI site shown by underlining). Similarly, the nucleotide Media and growth conditions for E. coli were described by sequence (AATATG) covering the ATG start codon of arsC was Sambrook et al. (19). changed to CATATG to create an NdeI site by PCR with primer III: 5Ј-CGCGAATTCCATATGAACGACATGGCCCAC- A. vinelandii strain OP (20), obtained from R. Dixon (John Ј Innes Centre, Norwich, U.K.), was routinely cultured at 30°C for CCC-3 (with the nucleotides to be changed shown by the italic 8 days on Burk’s medium (21) containing 2% glucose. For letters and an EcoRI site shown by underlining) and primer IV: 5Ј-CGCGGATCCCGCTGTTGGTATCGAATACC-3Ј (with a preparation of cysts, the cells collected from a 5-day culture in BamHI site shown by underlining). After amplification by PCR Burk’s medium with 2% glucose were washed once with Burk’s under the standard conditions, the EcoRI–BamHI fragments, buffer (medium without a carbon source) and suspended in the ␤ excised from amplified arsB and arsC sequences, were separately same volume of Burk’s buffer containing 0.2% -hydroxybu- cloned between the EcoRI and BamHI sites of pUC19, resulting tyrate for the induction of encystment (18). The culture was in pUC19-ArsB and pUC19-ArsC, respectively. The absence of incubated at 30°C for an additional 3 days for encystment. For undesired alterations was checked by nucleotide sequencing. The microscopy studies, cells were grown first on 2% agar of Burk’s NdeI–BamHI fragments, excised from pUC19-ArsB and medium containing 2% glucose for 5 days and then on 2% agar pUC19-ArsC, were separately cloned between the NdeI and of Burk’s medium containing 0.2% ␤-hydroxybutyrate. The agar BamHI sites of pET16b, resulting in pET16b-ArsB and pET16b- plates were incubated at 30°C for an additional 3 days for ArsC, respectively. For production of histidine-tagged ArsB and encystment. ArsC, E. coli BL21 (DE3) harboring pET16b-ArsB or pET16b-

6360 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0511227103 Funa et al. Downloaded by guest on September 25, 2021 ArsC was grown overnight in Luria broth containing 100 ␮g͞ml of ArsB was amplified with primer V: 5Ј-CGCAAGCTTCC- ampicillin. Cells were harvested by centrifugation and resus- TCGGCCGGCCTGCCGGTC-3Ј (with the Ser-1993 codon of pended in 10 mM Tris⅐HCl (pH 8.0) and 145 mM NaCl. After arsA in bold and a HindIII site in italic) and primer VI: sonication, cell debris was removed by centrifugation, and the 5Ј-CGCAAGCTTGGATCCTGTGAAGCCGGTGAGAA- cleared lysate was applied to a column with His-bind metal CTG-3Ј (with the Thr-13 codon of arsB in bold and a BamHI chelation resin (Novagen). The histidine-tagged proteins were site in italic). The amplified fragment was cloned between the purified according to the manual from the manufacturer, except HindIII and BamHI sites of pUC19, resulting in pUC19-N- for adding 10% glycerol to each buffer. The purified histidine- BamHI. The 1.6-kb DNA fragment containing the region from tagged protein was dialyzed against 10 mM Tris⅐HCl (pH 8.0), Glu-404 of ArsB to His-120 of ArsD was amplified with primer 500 mM NaCl, and 10% glycerol. VII: 5Ј-CGCGGATCCGAGAAAATATGAACGACATG-3Ј (with the Glu-404 codon of arsB in bold, the stop codon of arsB PKS Assay. The standard reaction mixture contained 100 ␮M shown by the underline, and a BamHI site in italic) and primer [2-14C]malonyl-CoA, 100 ␮M starter CoA ester, 100 mM VIII: 5Ј-CGCGAATTCGCTTGACCCCGGGGGCATGG-3Ј Tris⅐HCl (pH 8.0), and 61 ␮g of ArsB or ArsC in a total volume (with the His-120 codon of arsD in bold and an EcoRI site in of 100 ␮l. Reactions were incubated at 30°C for 30 min before italic). The amplified fragment was cloned between the BamHI being quenched with 20 ␮l of 6 M HCl. The products were and HindIII sites of pUC19, resulting in pUC19-C-BamHI. extracted with ethylacetate, and the organic layer was evapo- After confirmation of the correct sequence, the HindIII– rated to dryness. The residual material was dissolved in 15 ␮lof BamHI fragment from pUC19-N-BamHI and the BamHI– methanol for TLC and LC-APCIMS analyses. Silica gel 60 WF254 EcoRI fragment from pUC19-C-BamHI were ligated via the TLC plates (Merck) were developed in benzene͞acetone͞acetic common BamHI site and cloned between the HindIII and acid (85:15:1, vol͞vol͞vol), and the 14C-labeled compounds were EcoRI sites of pUC19. The 1.9-kb BamHI fragment containing detected by using a BAS-MS imaging plate (Fuji). LC-APCIMS aphII from Tn5 was inserted into the BamHI site of the ⌬ analysis was carried out by using the esquire high-capacity trap resultant plasmid, resulting in pUC19- ArsB, in which the plus (HCT) system (Bruker Daltonics, Bremen, Germany) sequence encoding from Pro-14 to Leu-403 of ArsB was ⌬ equipped with a Pegasil-B C4 reversed-phase HPLC column replaced by the aphII sequence. pUC19- ArsB was linearized (4.6 ϫ 250 mm; Shenshu Scientific) and acetonitrile͞water͞ with DraI, denatured with 0.1 M NaOH, and introduced by acetic acid [900:100:1, vol͞vol͞vol (solvent A); 800:200:1, vol͞ transformation into A. vinelandii OP by the method of Page vol͞vol (solvent B); or 500:500:1, vol͞vol͞vol (solvent C)] as an and von Tigerstrom (25). Kanamycin-resistant transformants eluant at a flow rate of 1 ml͞min. UV spectra were detected on were selected, and a gene replacement, as a result of double an Agilent 1100 series diode array detector. Spectral data are crossover, was confirmed by Southern blot analysis (data not summarized in Table 2. shown). Electron Microscopy. A. vinelandii cells were mounted on a copper Determination of Kinetic Parameters of ArsB and ArsC. The reactions, containing 100 mM Tris⅐HCl (pH 8.0), 100 mM NaCl, 100 ␮M grid to form a thin layer and immersed in liquid propane cooled malonyl-CoA, and 9.6 ␮g (0.674 ␮M) of ArsB or 9.2 ␮g (0.646 with liquid nitrogen (Leica EM CPC and Leica Mikrosystems). The frozen cells were transferred to 2% OsO in dry acetone and ␮M) of ArsC, were performed in a total volume of 300 ␮l. The 4 kept at Ϫ80°C for 3 days. The samples were warmed gradually concentration of behenyl-CoA was varied between 1 and 5 ␮M. from Ϫ80°C to 0°C over 5 h, held for1hat0°C, and warmed from After the reaction mixture had been preincubated at 30°C for 5 0°C to 23°C for 2 h (Leica EM AFS and Leica Mikrosystems). min, the reactions were initiated by adding the substrate and After the samples had been washed three times with dry acetone, continued for 60 s. The reactions were stopped with 60 ␮lof6 they were infiltrated with increasing concentrations of Spurr’s M HCl, and the material in the mixture was extracted with resin in dry acetone and finally with 100% Spurr’s resin. After ethylacetate. The organic layer was collected and evaporated. polymerization, ultrathin sections were cut on an ultramic- The residual material was dissolved in 20 ␮l of methanol for Ј rotome (Leica Ultracut UCT and Leica Mikrosystems) and HPLC analysis. 5-n-heneicosylresorcinol, 4-hydroxy-6-(2 - stained with uranyl acetate and lead citrate. The sections were oxotricosyl)-2-pyrone, and 6-heneicosyl-4-hydroxy-2-pyrone viewed on an H-7600 electron microscope (Hitachi, Tokyo) at were used to generate the standard curves for the quantification 100 kV. of the products. Steady-state parameters were determined by Hanes plot. We thank Ray Dixon (John Innes Centre, Norwich, U.K.) for providing us with the A. vinelandii strain. This work was supported by a grant-in-aid Inactivation of ars Operon in A. vinelandii. The 1.7-kb DNA frag- for Scientific Research on Priority Areas from Monkasho (to S.H.) and

ment containing the region from Ser-1993 of ArsA to Thr-13 the Waksman Foundation of Japan (to N.F.). MICROBIOLOGY

1. Kozubek, A. & Tyman, J. H. P. (1999) Chem. Rev. (Washington, D.C.) 99, 1–25. 13. Rock, C. O. & Cronan, J. E. (1996) Biochim. Biophys. Acta 1302, 1–16. 2. Reusch, R. N. & Sadoff, H. L. (1983) Nature 302, 268–270. 14. Schweizer, E. & Hofmann, J. (2004) Microbiol. Mol. Biol. Rev. 68, 501–517. 3. Su, C.-J. & Sadoff, H. L. (1981) J. Bacteriol. 147, 91–96. 15. Su, C.-J., Reusch, R. & Sadoff, H. L. (1979) J. Bacteriol. 137, 1434–1436. 4. Rawlings, J. B. (2001) Nat. Prod. Rep. 18, 190–227. 16. Pal, S., Manna, A. & Paul, A. K. (1997) Curr. Microbiol. 35, 327–330. 5. Shen, B. (2000) Top. Curr. Chem. 209, 1–51. 17. Segura, D., Cruz, T. & Espı´n, G. (2003) Arch. Microbiol. 179, 437–443. 6. Austin, M. B. & Noel, J. P. (2003) Nat. Prod. Rep. 20, 79–110. 18. Lin, L. P. & Sadoff, H. L. (1968) J. Bacteriol. 95, 2336–2343. 7. Kozubek, A., Pietr, S. & Czerwonka, A. (1996) J. Bacteriol. 178, 4027–4030. 19. Sambrook, J., Fritsch, E. F. & Maniatis, T. (1989) in Molecular Cloning: A 8. Ferrer, J.-L., Jez, J. M., Bowman, M. E., Dixon, R. A. & Noel, J. P. (1999) Nat. Laboratory Manual, ed. Nolan, C. (Cold Spring Harbor Lab. Press, Woodbury, Struct. Biol. 6, 775–784. NY). 9. Lambalot, R. H., Gehring, A. M., Flugel, R. S., Zuber, P., La Celle, M., Marahiel, 20. Bush, J. A. & Wilson, P. W. (1959) Nature 184, 381. M. A., Reid, R., Khosla, C. & Walsh, C. T. (1996) Chem. Biol. 3, 923–936. 21. Kennedy, C., Gamal, R., Humphrey, R., Ramos, J., Brigle, K. & Dean, D. 10. Sankaranarayanan, R., Saxena, P., Marathe, U. B., Gokhale, R. S., Shan- (1986) Mol. Gen. Genet. 205, 318–325. mugam, V. M. & Rukmini, R. (2004) Nat. Struct. Mol. Biol. 11, 894–900. 22. Schwink, L. & Knochel, P. (1994) Tetrahedron Lett. 35, 9007–9010. 11. Saxena, P., Yadav, G., Mohanty, D. & Gokhale, R. S. (2003) J. Biol. Chem. 278, 23. Funa, N., Ohnishi, Y., Ebizuka, Y. & Horinouchi, S. (2002) J. Biol. Chem. 277, 44780–44790. 4628–4635. 12. Austin, M. B., Bowman, M. E., Ferrer, J.-L., Schro¨der, J. & Noel, J. P. (2004) 24. Blecher, M. (1981) Methods Enzymol. 72, 404–408. Chem. Biol. 11, 1179–1194. 25. Page, W. J. & von Tigerstrom, M. (1979) J. Bacteriol. 139, 1058–1061.

Funa et al. PNAS ͉ April 18, 2006 ͉ vol. 103 ͉ no. 16 ͉ 6361 Downloaded by guest on September 25, 2021