Expression and purification of four different ... - Semantic Scholar

Microbiology (2000), 146, 839–849

Printed in Great Britain

Expression and purification of four different rhizobial acyl carrier proteins Isabel M. Lo! pez-Lara† and Otto Geiger† Author for correspondence : Otto Geiger. Tel : j52 73 131697. Fax : j52 73 175581. e-mail : Otto!cifn.unam.mx

Institute of Biotechnology, Technical University of Berlin, Seestrasse 13, D-13353 Berlin, Germany

In rhizobia, besides the constitutive acyl carrier protein (AcpP) involved in the biosynthesis and transfer of common fatty acids, there are at least three specialized acyl carrier proteins (ACPs) : (1) the flavonoid-inducible nodulation protein NodF ; (2) the RkpF protein, which is required for the biosynthesis of rhizobial capsular polysaccharides ; and (3) AcpXL, which transfers 27hydroxyoctacosanoic acid to a sugar backbone during lipid A biosynthesis. Whereas the nucleotide sequences encoding the three specialized ACPs are known, only the amino acid sequence of the AcpP of Sinorhizobium meliloti was available. In this study, using reverse genetics, the genes for the constitutive AcpPs of S. meliloti and of Rhizobium leguminosarum were cloned and sequenced. Previously, it had been shown that NodF and RkpF can be overproduced in Escherichia coli using the T7 polymerase expression system. Using the same system, the constitutive AcpPs of S. meliloti and of R. leguminosarum , together with the specialized ACP AcpXL, were overproduced and purified. All the known ACPs of rhizobia can be labelled in vivo during expression in E. coli with radioactive β-alanine added to the growth medium due to their modification with a 4’-phosphopantetheine prosthetic group. The availability of all functionally different ACPs should help to unravel how different fatty acids are targeted towards different biosynthetic pathways in one organism.

Keywords : Rhizobium, acyl carrier protein, acpP, acpXL

INTRODUCTION

Biosynthesis and transfer of fatty acids is of major importance during the formation of biological membranes, storage lipids, or certain amphiphilic signal molecules [i.e. Nod factors in rhizobia (De! narie! et al., 1996) or autoinducers in Gram-negative bacteria (Fuqua et al., 1996)]. In higher animals, fatty acid biosynthesis occurs on a multienzyme complex whereas in bacteria, fatty acids are formed by the catalytic activity of a number of monofunctional enzymes. In the latter organisms, a small acidic protein carries the growing fatty acyl chains via a thioester linkage of a 4hphosphopantetheine prosthetic group, which itself is .................................................................................................................................................

† Present address : CIFN – UNAM, Av. Universidad s/n, Colonia Chamilpa, Apdo postal 565-A, CP 62210, Cuernavaca, Morelos, Mexico. Abbreviations : ACP, acyl carrier protein ; iPCR, inverse PCR. The GenBank accession numbers for the nucleotide sequences reported in this paper are AF159244 and AF159243 for the acpP-containing regions of Sinorhizobium meliloti 1021 and of Rhizobium leguminosarum LPR5045, respectively. 0002-3775 # 2000 SGM

attached to a conserved serine residue. This protein is called acyl carrier protein (ACP). In bacteria, ACPs are involved not only in the synthesis of fatty acyl chains, but also in their transfer during phospholipid, lipid A or α-haemolysin biosynthesis (Issartel et al., 1991 ; Jackowski et al., 1991). In Escherichia coli, all these functions seem to be realized by a single and absolutely essential ACP which is produced constitutively from the acpP gene (Jackowski et al., 1991 ; Rawlings & Cronan, 1992) and which is now named AcpP. Some bacteria have additional ACPs which are involved in the biosynthesis of special β-ketides or fatty acids (Bibb et al., 1989 ; Shearman et al., 1986). In rhizobial species infecting legume plants of the Trifolieae, Vicieae, and Galegeae tribes, one specialized ACP, the nodulation protein NodF, is involved, together with NodE, in the synthesis of polyunsaturated fatty acids which are host-specific substituents of nodulation (Nod) factors (Yang et al., 1999). In addition to AcpP and NodF, there are at least two additional ACPs in rhizobia. The first was identified as 839

I . M . L O! P E Z -L A R A a n d O . G E I G E R

the rkpF gene product at the fix23 locus in Sinorhizobium meliloti 41 (Petrovics et al., 1993 ; Reuhs, 1996 ; Epple et al., 1998). The second novel ACP is involved in the transfer of the unusually long 27hydroxyoctacosanoic acid to a sugar backbone during lipid A biosynthesis in Rhizobium leguminosarum and has been named AcpXL (Brozek et al., 1996). The overall amino acid sequence identities of these four ACPs range from only 26 to 32 % (Brozek et al., 1996), and a well-conserved amino acid region can only be observed around the phosphopantetheine-binding site. However, the three-dimensional structure of the AcpP of E. coli has been determined (Kim & Prestegard, 1990) and an initial characterization of the secondary structure and the general tertiary fold of the NodF protein (Ghose et al., 1996) demonstrates that the overall structures of ACPs are surprisingly well conserved. In spite of the general structural similarity between NodF and AcpP of E. coli, AcpP cannot substitute for NodF in vivo in the synthesis of polyunsaturated fatty acids (Ritsema et al., 1998). Furthermore, these authors have shown that the NodF-specific functions are encoded in the C-terminal half of the protein. The presence in rhizobia of four different ACPs offers an unusual opportunity for investigation of the structure–function relationship of the ACPs. The nodF gene of R. leguminosarum bv. viciae had been cloned in the pET9a vector (Ritsema et al., 1994). With this system it is possible to overexpress NodF in E. coli to comprise 40 % of the total soluble cell protein (Ghose et al., 1996). In an analogous way, the rkpF gene of S. meliloti has been cloned and its protein product overproduced in E. coli (Epple et al., 1998). Overexpression of the different rhizobial ACPs will allow detailed structural investigations. Also, the different rhizobial ACPs will be invaluable tools for comparative biochemical studies in which acyl-ACPs are used as cosubstrates. In particular, the constitutive AcpP of S. meliloti will be required to elucidate the biosynthesis of a phosphorus-free ornithine-derived lipid recently described in this species (Geiger et al., 1999). In the present study the acpXL gene from R. leguminosarum was amplified using the PCR technique and cloned in an overexpression vector. Previously, only the amino acid sequence of the constitutive ACP of S. meliloti was known (Platt et al., 1990). Using reverse genetics, we have cloned and sequenced the acpP gene of S. meliloti and of R. leguminosarum. Furthermore, we have established the methods for the overproduction and subsequent purification of AcpXL and rhizobial AcpPs. In vivo labelling of the ACP-overproducing E. coli strains with β-alanine has shown that each of the four ACPs of rhizobia can be post-translationally modified in E. coli with a 4h-phosphopantetheine prosthetic group. METHODS

al., 1990), which expresses the T7 polymerase under the control of the lac promoter, was used as a host for the overproduction of ACPs. E. coli strains were grown at 37 mC in Luria–Bertani (LB) medium (Sambrook et al., 1989), except for large-scale cultures for the overproduction of ACPs that were grown in M9 minimal medium (Miller, 1972). Antibiotics were added, when required, to obtain the following final concentrations (µg ml-") : carbenicillin, 100 ; kanamycin, 50 ; and chloramphenicol, 20. DNA techniques and DNA sequence analysis. DNA manipulations were carried out using standard procedures (Sambrook et al., 1989). Labelling of DNA probes and colorimetric detection were carried out with the DIG-DNA labelling and detection kit of Boehringer Mannheim. DNA sequencing was performed using the dideoxy chain-termination method (Sanger et al., 1977) using the T7 Sequenase 7-deaza-dGTP sequencing kit (Amersham). Plasmids for DNA sequencing were isolated using a QIAprep Spin Miniprep kit (Qiagen). When required, PCR products were purified using the QIAquick PCR purification kit (Qiagen). DNA and derived protein sequences were analysed with the  search program (Altschul et al., 1997). Alignments of DNAs or amino acid sequences were done with the program  at the Genestream network server IGH Montpellier, France (http :\\ vega.igh.cnrs.fr\bin\lalign-guess.cgi). Cloning of the acpP gene of S. meliloti. This was was carried

out in three steps. (1) Cloning of the first half of the acpP gene. Two degenerate oligonucleotides were used in a PCR to amplify and subsequently clone that part of the gene encoding the aminoterminal half of the AcpP of S. meliloti. These oligonucleotides were deduced from the amino acid sequence of different ACPs, among them the sequence of the AcpP of S. meliloti 1021 (Platt et al., 1990). The forward primer oACP4 was based on the amino terminus of constitutive ACPs and the backward primer oACP3 was based on the conserved phosphopantetheinebinding region of several ACPs (Fig. 1). Using genomic DNA of S. meliloti 1021 as a template PCR was carried out with degenerate primers (oACP3 and oACP4), Taq polymerase (MBI Fermentas) and a PCR programme especially designed for degenerate oligonucleotides. The programme started with 3 cycles consisting of 30 s at 94 mC, 30 s at 37 mC and 30 s at 72 mC, and was followed by 30 cycles with identical parameters except that the annealing temperature was 55 instead of 37 mC. The products of the PCR reaction were separated in a 3 % NuSieveR GTGR agarose (FMC Bioproducts) gel and the DNA fragment migrating according to the expected size of 140 bp was excised. The piece of agarose was melted for 10 min at 65 mC and a 10-fold dilution was used as template in a new round of PCR with degenerate oligonucleotides oACP4 and oACP3. For this second round of PCR, conditions of amplification were 30 cycles each of 30 s at 94 mC, 30 s at 65 mC and 20 s at 72 mC. The products were purified, digested with NdeI and BamHI and ligated to pET9a that had been digested with the same enzymes. Four transformants carrying pET9a with an insert of 120–140 bp were sequenced using oligonucleotides oACP4 and oACP3. Plasmid pTB5013 harbours a fragment with an internal core sequence of 67 nucleotides encoding amino acids 11– 32 of AcpP of S. meliloti (Platt et al., 1990).

Bacterial strains, plasmids and media. The strains and

(2) Amplification and cloning of inverse PCR (iPCR) products to identify the full-length coding sequence of AcpP. Essentially the

plasmids used in this study are listed in Table 1. E. coli DH5α was used for cloning purposes. E. coli BL21(DE3) (Studier et

iPCR method described by Triglia et al. (1988) was followed. A pair of divergent oligonucleotide primers (oTBC16 : BamHI,

840

Rhizobial acyl carrier proteins Table 1. Bacterial strains and plasmids used Strain/plasmid

Relevant characteristics*

Source/reference

Escherichia coli DH5α

Host used for cloning

BL21(DE3) OG7001

Host used for expression panD mutant of BL21(DE3)

Bethesda Research Laboratory Studier et al. (1990) Epple et al. (1998)

Rhizobia Sinorhizobium meliloti 1021 Rhizobium leguminosarum LPR5045

SmR mutant of SU47 wild-type RifR, symbiosis-plasmid-cured

Meade et al. (1982) Hooykaas et al. (1981)

CbR, cloning vector KnR, expression vector CmR, production of lysozyme for repression of T7 polymerase KnR, used for overexpression of NodF KnR, used for overexpression of RkpF KnR, used for overexpression of AcpXL KnR, used for overexpression of AcpP of S. meliloti KnR, used for overexpression of AcpP of R. leguminosarum KnR, used for overexpression of AcpP of E. coli

Yanisch-Perron et al. (1985) Studier et al. (1990) Studier et al. (1990)

Plasmids† pUC18 pET9a pLysS pMP2301 pTB1003 pTB5011 pTB5035 pTB5040 pTB5079

Ritsema et al. (1994) Epple et al. (1998) This work This work This work This work

* SmR, RifR, CbR, KnR and CmR : streptomycin, rifampicin, carbenicillin, kanamycin and chloramphenicol resistance, respectively. † Other plasmids constructed in this work were pCU18 derivatives carrying the fragments as shown in Fig. 2.

HindIII 5h-AAAGGATCC GCA AGC TTC ATC GAT GAC-3h ; and oTBC18 : BamHI, SalI 5h-AAAGGATCCGC GTC GAC GCC AAG ATG ATC-3h ; restriction sites are underlined and respective enzymes are given before the sequence) were designed from the internal sequenced region (67 bp) of the acpP gene cloned in pTB5013 (Fig. 2a) in such a way that they annealed to the target DNA template 18 bp apart and their 3h termini were in opposite directions. S. meliloti 1021 genomic DNA was completely digested with individual restriction enzymes, the DNA was cleaned using phenol\chloroform extraction followed by ethanol precipitation, and was then used for self-circularization by ligase treatment at low DNA concentrations (Maiti & Ghosh, 1996 ; Triglia et al., 1988). iPCR reactions were performed with the self-circularized DNA samples using the standard reaction mixture, primers oTBC16 and oTBC18 and Taq polymerase. After a preheating step (95 mC, 30 min), conditions for amplification were 50 cycles each consisting of 45 s at 94 mC, 45 s at 65 mC and 150 s at 72 mC. A specific fragment of 1n5 kb from the KpnI digest was purified on an agarose gel. After digestion of the purifed fragment with KpnI and HindIII, the DNA sequences downstream of the acpP gene were cloned in pUC18 that had been digested with the same two enzymes (Fig. 2a ; plasmid pTB5021). After digestion of the 1n5 kb iPCR

product with SalI, a SalI fragment of 214 bp carrying sequences upstream of the acpP gene was cloned in pUC18 that had been digested with SalI (Fig. 2a ; plasmid pTB5030). (3) Cloning of the acpP gene of S. meliloti in the overexpression plasmid pET9a. When the complete DNA sequence encoding the

AcpP of S. meliloti had been determined, the ORF was amplified from genomic DNA of S. meliloti 1021 using homologous primers oTBC19 (NdeI 5h-AGGAATA CAT ATG AGC GAT ATC GCA GAA CG-3h) and oTBC20 (BamHI 5h-AAAGGATCC TCA GGC CTG GGC TTT CTC-3h), together with cloned Pfu DNA polymerase (Stratagene). The PCR product was digested with NdeI and BamHI and ligated with pET9a also digested with these enzymes. After transformation in DH5α, plasmids with an insert of the correct size were selected and sequenced. They all showed the same DNA sequence and pTB5035 was chosen for further studies. Cloning of the acpP gene of R. leguminosarum. The same specific primers (oTBC19 and oTBC20) used for amplification of the acpP gene of S. meliloti were used for the amplification and subsequent cloning in pET9a of the gene encoding the constitutive AcpP of R. leguminosarum LPR5045. Identical DNA sequences were obtained from several inserts derived from independent PCR amplifications. By this method the 841

I . M . L O! P E Z -L A R A a n d O . G E I G E R

(a) N-terminal amino acid sequences of AcpPs AcpP of E. coli

AcpP of S. meliloti

Degenerate primer oACP4 (b) Conserved amino acid sequences of diverse ACPs AcpP of E. coli NodF of R. leguminosarum bv. viciae AcpP of S. meliloti ACP gra of S. violaceoruber ACP tcm of S. glaucescens

Degenerate primer oACP3 .................................................................................................................................................................................................................................................................................................................

Fig. 1. Design of two degenerate oligonucleotide primers for cloning and sequencing part of the acpP gene of S. meliloti. (a) Nucleotide sequence of the N-terminal degenerate primer oACP4 with a NdeI restriction site. (b) Nucleotide sequence of the degenerate primer oACP3 with a BamHI restriction site. Conserved amino acids of the phosphopantetheinebinding motif of ACPs are boxed. Sequence data were obtained from the following references from top to bottom : Rawlings & Cronan (1992), Shearman et al. (1986), Platt et al. (1990), Sherman et al. (1989), and Bibb et al. (1989). Specialized ACPs gra from Streptomyces violaceoruber and tcm from Streptomyces glaucescens are involved in granaticin and tetracenomycin biosynthesis, respectively.

core DNA region encoding the AcpP of R. leguminosarum was obtained. To determine the DNA sequence flanking this internal core region, a pair of divergent primers for iPCR were selected. One of the primers used was oTBC18, which anneals 100 % in the last 13 nucleotides of the 3h end to the DNA of R. leguminosarum and adds a SalI site which is not present in the DNA of R. leguminosarum. A new primer specific for DNA sequences of acpP of R. leguminosarum was designed, oTBC25 (BamHI, 5h-AAAGGATCCTTTCGAA GAA GAA TTC GGC-3h), which annealed 80 bp apart from oTBC18 and the 3h terminus of which was in the opposite direction. Genomic DNA of R. leguminosarum LPR5045 was digested, religated and used in iPCR reactions with primers oTBC18 and oTBC25 as described for genomic DNA of S. meliloti. A specific product of 1n2 kb was obtained with a SalI digest. Making use of the restriction sites introduced by the primers, the sequences upstream of the acpP were cloned as a SalI fragment (pTB5062) and those downstream of the acpP gene as a BamHI\SalI fragment (pTB5065) (Fig. 2b). Sequencing of the iPCR fragments revealed that the oligonucleotide oTBC19 was identical to corresponding DNA sequences of the acpP of R. leguminosarum. The oligonucleotide oTBC20 differed only in one base pair (A at position 225 instead of G) from corresponding DNA sequences of the acpP of R. leguminosarum. As this change is in the third position of the respective codon it encodes K in both cases and does not lead to any difference in the amino acid sequence of AcpP of R. leguminosarum. Therefore, oTBC19 and oTBC20 were used to amplify 842

acpP of R. leguminosarum and after cloning in pET9a the plasmid obtained (pTB5040) was chosen for overproduction of AcpP of R. leguminosarum. Cloning of the acpXL gene of R. leguminosarum and of the acpP gene of E. coli in pET9a. Based on the DNA sequence of

the acpXL gene (Brozek et al., 1996 ; GenBank accession no. AF081835), a pair of homologous primers was designed for the PCR amplification. In the forward primer oTBC12 (5hGGAATACATATG ACC GCT ACA TTC GAT AAA G-3h), the GTG starting codon occurring in R. leguminosarum (Brozek et al., 1996) was changed to ATG and a NdeI restriction site was added. The backward primer oTBC13 (5hAAAGGATCC TCA GGC CTT GGC CGC-3h) adds a BamHI restriction site. As template, genomic DNA of R. leguminosarum LPR5045 was used. Design of the primers for amplification of the acpP gene of E. coli was based on the DNA sequence of the gene (Rawlings & Cronan, 1992) (GenBank accession no. M84991), the forward primer being oTBC30 (NdeI, 5h-AGGAATACAT ATG AGC ACT ATC GAA GAA CG-3h) and the backward primer oTBC31 (BamHI, 5hAAAGGATCC TTA CGC CTG GTG GCC GTT G-3h). As template, genomic DNA of E. coli DH5α was used. After purification of the amplified products, they were digested with NdeI and BamHI and ligated to pET9a that had been previously digested with the same two enzymes. After transformation of DH5α, plasmids with inserts of the correct size were selected and sequenced. Plasmid pTB5011 carries a DNA insert of identical sequence as that published for acpXL by Brozek et al. (1996) except that the starting GTG codon is

Rhizobial acyl carrier proteins (a) (1) deg PCR

(2) iPCR

(3) Pfu PCR

(b) (1) Pfu PCR

(2) iPCR

.................................................................................................................................................................................................................................................................................................................

Fig. 2. Steps for the cloning of the acpP genes of rhizobia in pET9a. (a) The three steps followed for the cloning and sequencing of the acpP of S. meliloti 1021 : (1) PCR with degenerate primers, (2) iPCR using specific primers derived from the region cloned with the degenerate primers, and (3) PCR amplification with Pfu polymerase and specific primers. (b) The two steps followed for the cloning and sequencing of the acpP of R. leguminosarum LPR5045 : (1) PCR amplification with the specific primers devised for acpP of S. meliloti, and (2) iPCR to check the sequences at the borders of acpP of R. leguminosarum. B, BamHI ; H, HindIII ; K, KpnI ; N, NdeI ; S, SalI. At the bottom of both (a) and (b) schemes of the sequenced regions are shown. RBS denotes putative ribosome-binding sites, a T above two arrowheads denotes a putative transcription terminator and inverted arrows in (a) denote the palindromic area downstream of acpP of S. meliloti. The area with homology to RIME1 in (a) is marked with a dashed line.

changed to ATG. Plasmid pTB5079 carries a DNA insert of identical sequence to the one previously published for acpP of E. coli by Rawlings & Cronan (1992). Overproduction of ACPs in E. coli. E. coli strain BL21(DE3)

(Studier et al., 1990) was individually transformed with the selected plasmid for overproduction of the respective ACP. In the case of the AcpPs, which are known to be toxic to E. coli when cloned in a high-copy-number plasmid, E. coli strain BL21(DE3) carrying the plasmid pLysS was used as host. The

presence of pLysS represses the basal activity of the T7 polymerase. Small-scale overproduction was used for establishing the best conditions for each of the ACPs produced. Cultures of 30 ml in LB medium containing the required antibiotics were grown at 37 mC to an OD of approximately 0n4, then IPTG was '#! added to a final concentration of 0n1 mM and cells were harvested by centrifugation at different time points after induction. The cell pellet was resuspended in 1n5 ml of 50 mM 843

I . M . L O! P E Z -L A R A a n d O . G E I G E R

Tris\HCl, 0n1 M KCl, pH 6n8 and kept frozen at k20 mC. After thawing, the cells were broken by a single passage through a French press at 20 000 p.s.i. (138 MPa). Broken cell suspensions were centrifuged for 5 min in an Eppendorf centrifuge at 15 000 r.p.m. and the supernatants were analysed for the presence of the overproduced protein by 20 % native PAGE (Jackowski & Rock, 1983). For large-scale overproduction of ACPs, the strains were grown at 37 mC in 2 l flasks containing 500 ml M9 minimal medium supplemented with the appropriate antibiotics. At an OD of 0n4, IPTG was added to a final concentration of '#! and cells were harvested by centrifugation 4 h later. 0n1 mM The cell pellets (1 vol.) were resuspended in 10 vols of either 50 mM Tris\HCl, 0n1 M KCl, pH 6n8 in the case of AcpXL and E. coli AcpP, or of 50 mM MOPS\KOH, 6 mM MgSO , % 1 mM DTT, pH 7n4 in the case of rhizobial AcpPs and kept frozen at k20 mC. After thawing, the cells were broken by passing the suspension twice through a French press at 20 000 p.s.i. The lysates were centrifuged at 7000 g for 20 min at 4 mC and the supernatants were used as cell-free extracts. In vivo labelling of rhizobial ACPs with β-alanine. In vivo labelling of ACPs with β-alanine, the biosynthetic precursor of 4h-phosphopantetheine, was carried out essentially as described by Epple et al. (1998). A β-alanine auxotroph, strain OG7001 (Epple et al., 1998), was transformed with pLysS and the new strain OG7001(pLysS) was individually transformed with the different plasmids for overproduction of ACPs or with pET9a. The presence of the pLysS plasmid facilitates the lysis of the cell pellets. Purification of ACPs. NodF from R. leguminosarum (Ghose et al., 1996) and RkpF from S. meliloti (Epple et al., 1998) were purified as described before with the exception that 2-propanol was removed by rotatory evaporation, as described below, instead of by dialysis.

The procedure described below was applied to extracts obtained from cells of 2 l cultures. For the purification of AcpXL, cold 2-propanol was added dropwise to a cell-free extract of the E. coli strain overproducing the ACP to a final concentration of 60 % (v\v). After 60 min at 4 mC, the precipitate was removed by centrifugation at 7000 g at 4 mC for 20 min. The supernatant was concentrated by rotatory evaporation to remove the 2-propanol and then applied to a 30 ml column of DEAE-52 cellulose (Whatman), which had been equilibrated with 50 mM Tris\HCl pH 6n8, 1 mM CHAPS. The column was washed with 90 ml buffer A (10 mM Bistris\HCl pH 6, 1 mM CHAPS). Elution was performed with a linear gradient from 0 to 1 M NaCl in buffer A in a total volume of 160 ml. Fractions of 3 ml were collected and analysed by PAGE. For the purification of AcpPs of rhizobia, cold 2-propanol was added dropwise to the cell-free extracts to a final concentration of 20 %. After 60 min at 4 mC, the precipitate was removed by centrifugation at 7000 g at 4 mC for 20 min. The supernatant was applied to a 30 ml DEAE-52 cellulose column equilibrated with buffer A containing 1 mM DTT. Washing and elution of the column were as described for the chromatography of AcpXL with the exception that all buffers contained 1 mM DTT. Fractions containing AcpP were pooled and after dialysis against 5 mM Tris\HCl pH 6n8 at 4 mC overnight, the volume was reduced by a sixth by rotatory evaporation. Because the rhizobial ACPs run distinctly ahead of the E. coli AcpP in the non-denaturing PAGE (Platt et al., 1990 ; Figs 3 and 4), the concentrated samples carrying the overproduced AcpP of rhizobia were further purified by a preparative native 844

PAGE (1n5 mm thick gels) in order to obtain separation from the E. coli AcpP. The unstained gel band containing the rhizobial AcpP (after identification by parallel staining) was sliced into small pieces and the AcpP protein was electroeluted with a Schleicher & Schuell Biotrap electroseparation system. Electroelution was carried out for 7 h at 200 V. AcpP of E. coli was purified from the overproducing strain using the same protocol as described for the purification of AcpXL.

RESULTS Cloning of the acpP gene of S. meliloti

The DNA sequences of the genes encoding the three specialized ACPs described in rhizobia, NodF, RkpF and AcpXL, are known. However, only the amino acid sequence of the constitutive AcpP of S. meliloti has been reported (Platt et al., 1990). Our aim was to overproduce and purify the four functionally different ACPs of rhizobia using the T7 polymerase expression system (Studier et al., 1990). Initially, we tried to amplify the whole gene encoding AcpP by PCR using degenerate oligonucleotides that were based on the amino acid sequence of S. meliloti. These degenerate primers did not give amplification of the desired product in any of the various conditions assayed (data not shown). We then decided to amplify part of the acpP gene using a pair of degenerate oligonucleotides, the sequences of which were derived from the alignment of different ACPs (Fig. 1) as described in the Methods section. Amplification from genomic DNA of S. meliloti 1021 with oACP4 and oACP3 gave several unspecific products and a minor amount of a DNA fragment of the expected size of 140 bp (data not shown). This fragment was used for a second round of PCR and the fragment of the appropriate size was cloned in pET9a. Sequencing analysis of DNA from four independent clones showed that two of them, each with a DNA insert of 123 bp, had an internal region of 67 bp that differed in only one base pair. Because the two primers for PCR were degenerate, the nucleotide sequences of the two insert DNAs were slightly different in the region where the primers annealed. For plasmid pTB5013, the deduced 22 amino acid sequence corresponding to the 67 bp core region was that of amino acids 11– 32 of the AcpP of S. meliloti (Platt et al., 1990). iPCR with the two gene-specific primers oTBC16 and oTBC18 gave a specific product from the KpnI digest of 1n5 kb. To further confirm that the iPCR product was specific, a Southern DNA hybridization was carried out using as a probe the DNA insert of pTB5013. The probe hybridized strongly with the iPCR product and genomic DNA of S. meliloti digested with KpnI gave a specific band at 1n5 kb, in agreement with the size of the iPCR product (data not shown). The nucleotide sequences flanking the 67 bp core acpP region were obtained after sequencing of two fragments of the iPCR product of the KpnI digest (Fig. 2a). The

Rhizobial acyl carrier proteins

homologous primers oTBC19 and oTBC20 were then used to amplify the complete acpP gene from genomic DNA of S. meliloti. Three independent acpP inserts cloned in pET9a were sequenced and they all showed an identical DNA sequence for acpP. The deduced amino acid sequence is identical to that determined by protein sequencing (Platt et al., 1990). Plasmid pTB5035 was chosen for further studies. Cloning of the acpP gene of R. leguminosarum

The amino acid sequence of the acpP of R. leguminosarum was not known. We tried the same pair of specific primers, oTBC19 and oTBC20, designed for the amplification of the acpP of S. meliloti in a PCR reaction with genomic DNA of R. leguminosarum LPR5045. A unique and specific fragment of the expected size was obtained and subsequently cloned in pET9a. The deduced amino acid sequence differed in only three positions from that of the AcpP of S. meliloti ; these were (S. meliloti to R. leguminosarum) : E 21 D, S 24 V and G 26 S. The DNA sequences of the regions flanking the core region of acpP of R. leguminosarum were obtained using iPCR (Fig. 2b and Methods). Plasmid pTB5040 was chosen for the overproduction of AcpP of R. leguminosarum. Analysis of the nucleotide sequences of acpP genes of rhizobia

The complete nucleotide sequences encoding AcpPs of S. meliloti 1021 and of R. leguminosarum LPR5045 together with the surrounding regions were obtained. The high level of similarity between the DNA sequences encoding S. meliloti AcpP and R. leguminosarum AcpP (86n5 %) does not extend into the surrounding areas. Using the  program, we identified a partial ORF downstream of acpP of R. leguminosarum with homology to FabF (3-oxoacyl-ACP synthase II) from different organisms. The highest similarity is to FabF from E. coli (56 % identity and 71 % similarity in an overlap of 39 amino acids). In E. coli the fabF gene is also located directly downstream of acpP (Rawlings & Cronan, 1992). By comparison with the sequences of R. leguminosarum, we identified downstream of acpP of S. meliloti a partial ORF encoding the sinorhizobial FabF (Fig. 2a). However, whilst in R. leguminosarum the acpP–fabF intergenic region is 92 nucleotides, in S. meliloti the intergenic region is much larger (305 nucleotides) (Fig. 2). Downstream of both rhizobial acpP genes the typical structure of a rho-factor-independent transcription terminator was identified (Fig. 2). The same kind of transcription terminator structures have been found downstream of the acpP genes of E. coli and of Azospirillum brasilense (Rawlings & Cronan, 1992 ; Maiti & Ghosh, 1996). Further analysis of the intergenic region of S. meliloti revealed a complex secondary structure with several inverted repeats (shown by inverted arrows in Fig. 2a). A  search with the acpP–fabF intergenic region of S. meliloti revealed the

presence of a homologue of the mosaic sequence RIME 1 (Fig. 2a). RIME are Rhizobium-specific intergenic mosaic elements characterized by two large palindromes (Østera/ s et al., 1995). RIME 1 has been found in different rhizobial species while RIME 2 seems to be restricted to S. meliloti (Østera/ s et al., 1995). In the acpP–fabF intergenenic region of S. meliloti, upstream of RIME 1 there is another mosaic element consisting of two palindromes, only one of which shows homology to RIME 2. Putative RBS were identified for acpP of S. meliloti, acpP of R. leguminosarum and fabF of R. leguminosarum (Fig. 2). No significant similarity was observed between the DNA sequences upstream of the acpP genes of S. meliloti and of R. leguminosarum. Optimization of overproduction of ACPs

We have amplified and cloned in pET9a the genes encoding AcpXL of R. leguminosarum (Brozek et al., 1996) and for AcpP of E. coli (Rawlings & Cronan, 1992) using homologous primers (see Methods). AcpP of E. coli is the best studied ACP and therefore it was used in our studies as a control. The growth of strain BL21(DE3) was not affected when carrying the plasmid for overproduction of AcpXL, pTB5011. However, when carrying the plasmids expressing the constitutive AcpPs of rhizobia, pTB5035 or pTB5040, the growth rate of host strains was reduced. This result was not unexpected since the full-length acpP of Gram-negative bacteria cloned in a high-copynumber vector is unstable in the E. coli host (Rawlings & Cronan, 1992 ; Maiti & Ghosh, 1996). Although plasmids pTB5035 and pTB5040 could be maintained in BL21(DE3), the overproduction of encoded AcpPs was very poor (data not shown). Good overproduction of AcpPs was obtained only if in addition the plasmid pLysS was present (Fig. 3a, lanes 2 and 3). A similar effect has been described for the acpP gene of E. coli (Hill et al., 1995). Therefore, for overproduction of the three AcpPs, BL21(DE3) carrying pLysS was always used as the host. Using small cultures of 30 ml, the best conditions for the overproduction of each ACP were established. In all cases maximal protein overproduction occurred between 3 and 4 h after induction with IPTG (data not shown). The best yield was obtained when cultures were induced at an OD of 0n4–0n5. Induction with concentrations of IPTG higher than 0n1 mM did not give any significant increase in the protein yield (data not shown). In vivo labelling of rhizobial ACPs with radioactive βalanine

ACPs need to be substituted with their prosthetic group 4h-phosphopantetheine in order to be functional in fatty acid\β-ketide synthesis or transfer. During biosynthesis of CoA, β-alanine is a building block and can be used as a marker to demonstrate whether CoA derivatives, like 4h-phosphopantetheine, are covalently linked to 845

I . M . L O! P E Z -L A R A a n d O . G E I G E R

(a)

1 1

2

3

4

5

6

2

3

4

5

6

7

RkpF

RkpF

Apo Holo

Apo Holo

AcpPEc Apo Holo

AcpPEc Apo Holo

.................................................................................................................................................

(b) 1

2

4

5

6

7

Fig. 4. Purification of ACPs. Conformationally sensitive 20 % native PAGE gel showing purified ACPs AcpP of E. coli (lane 1), AcpP of S. meliloti (lane 2), AcpP of R. leguminosarum (lane 3), NodF of R. leguminosarum (lane 4), RkpF of S. meliloti (lane 5), and AcpXL of R. leguminosarum (lane 6). In each well about 5 µg of protein was loaded and after electrophoresis proteins were visualized with Coomassie blue. In lanes 1 and 5 apo- and holo-forms have different mobilities.

suggests that, for all ACPs tested, addition of the 4hphosphopantetheine prosthetic group to apo-ACPs can be performed by an enzyme activity present in E. coli, presumably by holo-ACP synthase (AcpS) (Lambalot & Walsh, 1995). Purification of rhizobial ACPs AcpPEc

.................................................................................................................................................

Fig. 3. In vivo overproduction of ACPs. (a) Optimization of the overproduction of ACPs. Analysis of cell-free extracts from E. coli strains overproducing the different ACPs was performed by conformationally sensitive 20 % native PAGE and proteins were visualized with Coomassie blue. In each lane the equivalent of 100 µl of culture was loaded. In lanes 1 and 5 apo- and holoforms have different mobilities. (b) Autoradiogram of cell extracts of E. coli OG7001(pLysS) overproducing the different ACPs after labelling with radioactive β-alanine. In (a) and (b) the overproduced ACPs are : AcpP of E. coli (lane 1), AcpP of S. meliloti (lane 2), AcpP of R. leguminosarum (lane 3), NodF of R. leguminosarum (lane 4), RkpF of S. meliloti (lane 5), AcpXL of R. leguminosarum (lane 6), and none (E. coli strain carrying pET9a) (lane 7).

proteins. We have checked for the in vivo labelling with β-alanine of ACPs overproduced in E. coli. Cell-free extracts of OG7001(pLysS) carrying each of the overproducing plasmids or the empty vector pET9a that had been grown in the presence of $H-labelled β-alanine were analysed by PAGE and subsequent autoradiography showed the labelling of proteins with similar mobility to the overproduced ACPs (Fig. 3). This result 846

A simple purification procedure allowed us to purify the NodF protein to homogeneity in two steps (a 2propanol precipitation followed by chromatography on DEAE-cellulose) (Ghose et al., 1996). NodF purified in this way could be used for NMR investigations (Ghose et al., 1996). In the case of E. coli AcpP and AcpXL of R. leguminosarum, we have established similar procedures (see Methods) that yield the respective homogeneous proteins (Fig. 4, lanes 1 and 6, respectively). For obtaining homogeneous rhizobial AcpPs, a third step of purification consisting of preparative PAGE was required in order to separate them from the AcpP of the E. coli strain used as host for the overproduction. Samples of purified AcpP of S. meliloti and of AcpP of R. leguminosarum are shown in Fig. 4, lanes 2 and 3, respectively. As previously described (Platt et al., 1990), the AcpPs of rhizobia run in this system distinctly ahead of the E. coli AcpP. Also, both rhizobial AcpPs purified from the overproducing E. coli strains migrate as doublet bands in a manner similar to that found for the AcpP isolated from S. meliloti (Platt et al., 1990). DISCUSSION

We have sequenced the genes for the constitutive AcpPs of S. meliloti and of R. leguminosarum, as well as the areas surrounding these genes. In constrast to the high

Rhizobial acyl carrier proteins

level of similarity observed within the coding regions (acpP and part of fabF), the intergenic regions are quite different in length and do not show any significant DNA similarity. Whilst in R. leguminosarum two specialized ACPs were known, AcpXL and NodF (Brozek et al., 1996 ; Shearman et al., 1986), here we report another ACP from this organism which probably is the one responsible for the biosynthesis and transfer of common fatty acids. Previous reports (Ghose et al., 1996 ; Epple et al., 1998), together with the data presented here, allow for the overproduction and purification of the four different ACPs described in rhizobia. Since AcpP of E. coli is the best studied ACP, we have also cloned and expressed this AcpP in an analogous way. The availability of the four different rhizobial ACPs, each with a different biological function, should help to unravel how the different fatty acids are directed towards different biosynthetic pathways in one organism. Of special interest will be the study of the distinct ACP substrate specificities of NodA proteins. The nodulation protein NodA seems to be involved in the central step of the biosynthetic pathway leading to lipo-chitin oligosaccharide signal molecules in rhizobia (Atkinson et al., 1994 ; Ro$ hrig et al., 1994). It is thought that NodA transfers an acyl residue from an ACP to a Ndeacetylated chitooligosaccharide. Different rhizobia seem to have different preferences as to the nature of the fatty acyl residue they transfer to the chitooligosaccharide backbone (Debelle! et al., 1996 ; Ritsema et al., 1996). NodAs from some rhizobia seem to transfer preferentially the nodFE-dependent fatty acids (R. leguminosarum and S. meliloti) or 1ω-hydroxylated fatty acids (S. meliloti). These fatty acids are known to be synthesized and carried on the specialized ACPs NodF and AcpXL, respectively. The selectivity of the NodA proteins can be explained either by a specific recognition of the respective fatty acids or, more likely, by a specific protein–protein interaction with the specialized ACP carrying such a fatty acid. It has been shown that acyl-AcpP from E. coli (but not free fatty acids) can function as an acyl donor in NodA-directed fatty acid transfer (Ritsema et al., 1997), suggesting that NodA might interact with ACP for the transfer of the fatty acid. A more detailed study of NodA specificity will now be possible using the functionally different ACPs of rhizobia. Post-translational modification with 4h-phosphopantetheine is required to render ACPs functionally active in fatty acid biosynthesis. The in vivo labelling experiment with β-alanine indicates that the post-translational modification of adding the prosthetic group to the apoform of the different rhizobial ACPs can be performed by an enzyme activity present in E. coli. However, the in vivo substitution of the overexpressed ACPs is in no way complete (Fig. 3a, lanes 1 and 5). Purified holo-ACP synthase (AcpS) of E. coli (Lambalot & Walsh, 1995) is able to transfer in vitro 4h-phosphopantetheine to the specialized rhizobial ACPs NodF (Ritsema et al., 1998) and RkpF (Epple et al., 1998). To quantify the in vitro

reaction between apo-ACPs and AcpS, we are presently determining the kinetic constants (KM, kcat) for each one of the rhizobial ACPs as substrates. To use ACPs in biochemical studies, the holo-ACPs should be thioesterified to particular fatty acyl residues (or modified fatty acyl residues). We are presently carrying out different studies in order to establish a feasible and (if possible) general method for the acylation of all the ACPs described here. A chemical method for acylation of ACPs has been described (Cronan & Klages, 1981), but preferably, simpler enzymic methods should be used. Acyl-ACP synthetase (Aas) of E. coli is the enzyme that can acylate E. coli AcpP (Ray & Cronan, 1976). AcpP from S. meliloti functions as effective substrate for E. coli Aas (Platt et al., 1990), whilst NodF is not a good substrate for the E. coli Aas (Ritsema et al., 1998). Therefore, it seems that E. coli acyl-ACP synthetase has a substrate specificity that is narrower than that of E. coli AcpS and can only be used for acylation of constitutive ACPs. The Vibrio harveyi acyl-ACP synthetase seems to have a broader specificity than its E. coli counterpart (Fice et al., 1993 ; Shen et al., 1992). It will be interesting to see if the V. harveyi enzyme can acylate the specialized ACPs of rhizobia. Another enzymic approach might be to use E. coli AcpS for loading altered phosphopantetheine moieties, from CoA analogues, directly to the apo-ACP. Such a method has been successful in the preparation of acylated derivatives of Streptomyces ACPs (Gehring et al., 1997). ACKNOWLEDGEMENTS We thank Viola Ro$ hrs for excellent technical assistance. I. M. L. L. was the recipient of a long-term EMBO fellowship. This work was supported by grant Ge556\3-1 from the Deutsche Forschungsgemeinschaft.

REFERENCES Altschul, S. F., Madden, T. L., Scha$ ffer, A. A., Zhang, J., Zhang, Z., Miller, W. & Lipman, D. J. (1997). Gapped  and - : a

new generation of protein database search programs. Nucleic Acids Res 25, 3389–3402. Atkinson, E. M., Palcic, M. M., Hindsgaul, O. & Long, S. R. (1994).

Biosynthesis of Rhizobium meliloti lipooligosaccharide Nod factors : NodA is required for an N-acyltransferase activity. Proc Natl Acad Sci USA 91, 8418–8422. Bibb, M. J., Biro, S., Motamedi, H., Collins, J. F. & Hutchinson, C. R. (1989). Analysis of the nucleotide sequence of the Strepto-

myces glaucescens tcmI genes provides key information about the enzymology of polyketide antibiotic biosynthesis. EMBO J 8, 2727–2736. Brozek, K. A., Carlson, R. W. & Raetz, C. H. R. (1996). A special acyl carrier protein for transferring long hydroxylated fatty acids to lipid A in Rhizobium. J Biol Chem 271, 32126–32136. Cronan, J. E. & Klages, A. L. (1981). Chemical synthesis of acyl thioester of acyl carrier protein with native structure. Proc Natl Acad Sci USA 78, 5440–5444. Debelle! , F., Plazanet, C., Roche, P., Pujol, C., Savagnac, A., 847

I . M . L O! P E Z -L A R A a n d O . G E I G E R Rosenberg, C., Prome! , J. C. & De! narie! , J. (1996). The NodA

F. M. (1982). Physical and genetic characterization of symbiotic

proteins of Rhizobium meliloti and Rhizobium tropici specify the N-acylation of Nod factors by different fatty acids. Mol Microbiol 22, 303–314. De! narie! , J., Debelle! , F. & Prome! , J. C. (1996). Rhizobium lipochitooligosaccharide nodulation factors : signaling molecules mediating recognition and morphogenesis. Annu Rev Biochem 65, 503–535.

and auxotrophic mutants of Rhizobium meliloti induced by transposon Tn5 mutagenesis. J Bacteriol 149, 114–122. Miller, J. H. (1972). Experiments in Molecular Genetics. Cold Spring Harbor, NY : Cold Spring Harbor Laboratory. Østera/ s, M., Stanley, J. & Finan, T. M. (1995). Identification of Rhizobium-specific intergenic mosaic elements within an essential two-component regulatory system of Rhizobium species. J Bacteriol 177, 5485–5494.

Epple, G., van der Drift, K. M. G. M., Thomas-Oates, J. E. & Geiger, O. (1998). Characterization of a novel acyl carrier protein,

RkpF, encoded by an operon involved in capsular polysaccharide biosynthesis in Sinorhizobium meliloti. J Bacteriol 180, 4950–4954. Fice, D., Shen, Z. & Byers, D. (1993). Purification and characterization of fatty acyl-acyl carrier protein synthetase from Vibrio harveyi. J Bacteriol 175, 1865–1870. Fuqua, C., Winans, S. C. & Greenberg, E. (1996). Census and consensus in bacterial ecosystems : the LuxR-LuxI family of quorum sensing transcriptional regulators. Annu Rev Microbiol 50, 727–751. Gehring, A. M., Lambalot, R. H., Vogel, K. W., Drueckhammer, D. G. & Walsh, C. T. (1997). Ability of Streptomyces spp. acyl

carrier proteins and coenzyme A analogs to serve as substrates in vitro for E. coli holo-ACP synthase. Chem Biol 4, 17–24. Geiger, O., Ro$ hrs, V., Weissenmayer, B., Finan, T. M. & ThomasOates, J. E. (1999). The regulator gene phoB mediates phosphate

stress-controlled synthesis of the membrane lipid diacylglycerylN,N,N-trimethylhomoserine in Rhizobium (Sinorhizobium) meliloti. Mol Microbiol 32, 63–73. Ghose, R., Geiger, O. & Prestegard, J. H. (1996). NMR investigations of the structural properties of the nodulation protein, NodF, from Rhizobium leguminosarum and its homology with Escherichia coli acyl carrier protein. FEBS Lett 388, 66–72. Hill, R. B., MacKenzie, K. R., Flanagan, J. M., Cronan, J. E. & Prestegard, J. H. (1995). Overexpression, purification, and

characterization of Escherichia coli acyl carrier protein and two mutant proteins. Protein Expr Purif 6, 394–400. Hooykaas, P. J. J., van Brussel, A. N. N., den Dulk-Raas, H., van Slogteren, G. M. S. & Schilperoort, R. A. (1981). Sym plasmid of

Rhizobium trifolii expressed in different rhizobial species and Agrobacterium tumefaciens. Nature 291, 351–353. Issartel, J.-P., Koronakis, V. & Hughes, C. (1991). Activation of Escherichia coli prehaemolysin to the mature toxin by acyl carrier protein-dependent fatty acylation. Nature 351, 759–761. Jackowski, S. & Rock, C. O. (1983). Ratio of active to inactive forms of acyl carrier protein in Escherichia coli. J Biol Chem 258, 15186–15191. Jackowski, S., Cronan, J. E., Jr & Rock, C. O. (1991). Lipid metabolism in procaryotes. In Biochemistry of Lipids, Lipoproteins and Membranes, pp. 43–85. Edited by D. E. Vance & J. Vance. Amsterdam : Elsevier. Kim, Y. & Prestegard, J. H. (1990). Refinement of the NMR structures for acyl carrier protein with scalar coupling data. Proteins 8, 377–385. Lambalot, R. H. & Walsh, C. T. (1995). Cloning, overproduction, and characterization of the Escherichia coli holo-acyl carrier protein synthase. J Biol Chem 270, 24658–24661. Maiti, M. K. & Ghosh, S. (1996). Acyl carrier protein of Azospirillum brasilense : properties of the purified protein and sequencing of the corresponding gene, acpP. Microbiology 142, 2097–2103. Meade, H. M., Long, S. R., Ruvkun, G. B., Brown, S. E. & Ausubel,

848

Petrovics, G., Putnoky, P., Reuhs, B., Kim, J., Thorp, T. A., Noel, K. D., Carlson, R. W. & Kondorosi, A. (1993). The presence of a

novel type of surface polysaccharide in Rhizobium meliloti requires a new fatty acid synthase-like gene cluster involved in symbiotic nodule development. Mol Microbiol 8, 1083–1094. Platt, M. W., Miller, K. J., Lane, W. S. & Kennedy, E. P. (1990).

Isolation and characterization of the constitutive acyl carrier protein from Rhizobium meliloti. J Bacteriol 172, 5440–5444. Rawlings, M. & Cronan, J. E. (1992). The gene encoding Escherichia coli acyl carrier protein lies within a cluster of fatty acid biosynthetic genes. J Biol Chem 267, 5751–5754. Ray, T. K. & Cronan, J. E. (1976). Activation of long chain fatty acyl carrier protein : demonstration of a new enzyme, acyl-acyl carrier protein synthetase, in Escherichia coli. Proc Natl Acad Sci USA 73, 4374–4378. Reuhs, B. L. (1996). Acidic capsular polysaccharides (K antigens) of Rhizobium. In Biology of Plant–Microbe Interactions, pp. 331–336. Edited by G. Stacey, B. Mullin & P. M. Gresshoff. St Paul, MN : International Society for Molecular Plant–Microbe Interactions. Ritsema, T., Geiger, O., van Dillewijn, P., Lugtenberg, B. J. J. & Spaink, H. P. (1994). Serine residue 45 of nodulation protein

NodF from Rhizobium leguminosarum bv. viciae is essential for its biological function. J Bacteriol 176, 7740–7743. Ritsema, T., Wijfjes, A. H. M., Lugtenberg, B. J. J. & Spaink, H. P. (1996). Rhizobium nodulation protein NodA is a host-specific

determinant of the transfer of fatty acids in Nod factor biosynthesis. Mol Gen Genet 251, 44–51. Ritsema, T., Lugtenberg, B. J. J. & Spaink, H. P. (1997). Acyl-acyl carrier protein is a donor of fatty acids in the NodA-dependent step in biosynthesis of lipochitin oligosaccharides by rhizobia. J Bacteriol 179, 4053–4055. Ritsema, T., Gehring, A. M., Stuitje, A. R. & 7 other authors (1998). Functional analysis of an interspecies chimera of acyl

carrier proteins indicates a specialized domain for protein recognition. Mol Gen Genet 257, 641–648. Ro$ hrig, H., Schmidt, J., Wieneke, U., Kondorosi, E., Barlier, I., Schell, J. & John, M. (1994). Biosynthesis of lipooligosaccharide

nodulation factors : Rhizobium NodA protein is involved in Nacylation of the chitooligosaccharide backbone. Proc Natl Acad Sci USA 91, 3122–3126. Sambrook, J., Fritsch, E. F. & Maniatis, T. (1989). Molecular Cloning : a Laboratory Manual, 2nd edn. Cold Spring Harbor, NY : Cold Spring Harbor Laboratory. Sanger, F., Nicklen, S. & Coulson, A. R. (1977). DNA sequencing with chain-terminating inhibitors. Proc Natl Acad Sci USA 74, 5463–5467. Shearman, C. A., Rossen, L., Johnston, A. W. B. & Downie, J. A. (1986). The Rhizobium leguminosarum nodulation gene nodF

encodes a polypeptide similar to acyl carrier protein and is regulated by nodD plus a factor in pea root exudate. EMBO J 5, 647–652. Shen, Z., Fice, D. & Byers, D. M. (1992). Preparation of fatty-

Rhizobial acyl carrier proteins acylated derivatives of acyl carrier protein using Vibrio harveyi acyl-ACP synthetase. Anal Biochem 204, 34–39. Sherman, D. H., Malpartida, F., Bibb, M. J., Kieser, H. M., Bibb, M. J. & Hopwood, D. A. (1989). Structure and deduced function

of the granaticin-producing polyketide synthase gene cluster of Streptomyces violaceoruber Tu22. EMBO J 8, 2717–2725. Studier, F. W., Rosenberg, A. H., Dunn, J. J. & Dubendorff, J. W. (1990). Use of T7 RNA polymerase to direct expression of cloned

Yang, G.-P., Debelle! , F., Savagnac, A. & 9 other authors (1999).

Structure of the Mesorhizobium huakuii and Rhizobium galegae Nod factors : a cluster of phylogenetically related legumes are nodulated by rhizobia producing Nod factors with α,β-unsaturated N-acyl substitutions. Mol Microbiol 34, 227–237. Yanisch-Perron, C., Vieira, J. & Messing, J. (1985). Improved M13 phage cloning vectors and host strains : nucleotide sequences of the M13mp18 and pUC19 vectors. Gene 33, 103–119.

genes. Methods Enzymol 185, 60–89. Triglia, T., Peterson, M. G. & Kemp, D. J. (1988). A procedure for

.................................................................................................................................................

in vitro amplification of DNA segments that lie outside the boundaries of known sequences. Nucleic Acids Res 16, 8186.

Received 23 September 1999 ; revised 4 January 2000 ; accepted 13 January 2000.

849