Absence of plasmids encoding adhesion-related proteins in non-insect-transmissible strains of Spiroplasma citri

Plant-pathogenic mollicutes, phytoplasmas and spiroplasmas, are associated with hundreds of diseases that affect a wide variety of plants including ornamentals, vegetables, fruit trees and grape, leading to important economic losses (Lee et al., 2000; Seemüller et al., 2002). The bacteria multiply in the phloem sieve tubes, where they are introduced by phloem sap-feeding leafhoppers or psyllids, which therefore play a crucial role in the spread of these pathogens. Spiroplasma citri is the causal agent of stubborn disease of citrus (Saglio et al., 1973) and brittle root disease of horseradish (Fletcher, 1983). However, it infects many other plants including the Madagascar periwinkle (Catharanthus roseus), which serves as the experimental host plant for most of the plant-pathogenic mollicutes. In nature, S. citri is transmitted in a persistent manner by the leafhopper vectors Circulifer tenellus in the USA (Liu et al., 1983a) and Circulifer haematoceps in the Old World (Fos et al., 1986). S. citri can be experimentally transmitted to periwinkle plants by injection into the leafhopper vector (Foissac et al., 1996). Over the years, S. citri has become a model for studying several aspects of the relationships of plant-pathogenic mollicutes with their hosts, the plant and the vector insect (Bové et al., 2003; Fletcher et al., 1998). In particular, it has been shown that sugar metabolism, and especially fructose import, is a key factor of spiroplasmal pathogenicity (Gaurivaud et al., 2000; André et al., 2005). S. citri has also been used to study the pathway of spiroplasmas within the leafhopper vector (Kwon et al., 1999; Liu et al., 1983b). When ingested by the leafhopper, spiroplasmas must cross two physical barriers before being transmitted to the host plant. First, they cross the gut epithelium to move from the lumen to haemolymph, where they multiply, and then they migrate to the salivary glands, moving across the associated membranes to reach the salivary ducts (Liu et al., 1983b). Mainly based on electron microscopy studies, model mechanisms of spiroplasma movement through these two barriers have been proposed (Fletcher et al., 1998). However, the processes are still not well understood and the genetic determinants involved are not known. In S. citri, several surface proteins have been suggested to play a role in transmission. Among these, protein P89, later named Sarp1, has been shown to be implicated in adhesion to insect (C. tenellus) cells, suggesting that this protein is directly involved in spiroplasmainsect cell interaction (Yu et al., 2000; Berg et al., 2001). Using a reverse genetic approach, it has been shown that disruption of a gene predicted to encode a surface lipoprotein, with homology to a solute-binding protein of an ABC transporter, strongly reduces transmission of S. citri by the leafhopper C. haematoceps (Boutareaud et al., 2004). Recently, it has been shown that spiralin, the most abundant membrane protein of S. citri, acts as a lectin binding to glycoproteins from the insect vector C. haematoceps (Killiny et al., 2005). Besides being the most abundant lipoprotein at the cell surface, spiralin is highly immunogenic. Consequently, monoclonal antibodies (mAbs) raised against S. citri cells were all found to react with spiralin. In previous studies, we have shown that insect transmission of a spiralin-disrupted S. citri mutant was poorly efficient as compared to the wild-type strain, suggesting that spiralin is required for efficient transmission of the spiroplasma by its leafhopper vector (Duret et al., 2003). However, the finding that this mutant could still be transmitted also suggested that proteins other than spiralin were probably involved. With the aim of identifying new surface proteins, putatively involved in the spiroplasmavector insect interactions, we used the spiralin-defective mutant as the antigen source to produce mAbs directed against surface proteins other than spiralin. One mAb was found to react with a membrane-associated protein sharing strong similarities with adhesion-related protein Sarp1 of S. citri BR3 (Berg et al., 2001). In the S. citri wild-type strain GII-3, this protein is one of the eight members of the adhesion-related protein family which we named Scarp (for S. citri adhesion-related protein). Interestingly, the eight scarp genes are distributed on five plasmids (pSci15) of 1328 kbp (Foissac et al., 2004). Comparison of insect-transmissible and non-transmissible strains of S. citri revealed that the non-transmissible strains did not express Scarps and even did not possess scarp genes. We showed that the S. citri mutant GII3-9a2, which possesses no spiralin, also lacked pSci5, a plasmid carrying three scarp genes. Experimental transmission assays were conducted to determine whether pSci5 was required for efficient transmission of S. citri via injection into its vector insect C. haematoceps. Bacterial strains, transformation and growth conditions.
S. citri GII-3 wild-type strain was originally isolated from its leafhopper vector Circulifer haematoceps captured in Morocco (Vignault et al., 1980). S. citri strains R8A2 (Saglio et al., 1973), ASP-1 (Townsend et al., 1977) and 44 (kindly provided by Dr Hosseini Pour, University of Kerman, Iran; unpublished) were isolated from stubborn-diseased sweet orange trees in Morocco, Israel and Iran, respectively. Unlike R8A2 and ASP-1, strain 44 has not been extensively propagated in vitro. However, attempts to infect periwinkle plants via injection into the leafhopper vector C. haematoceps repeatedly failed, similar to the situation with R8A2 and ASP-1. The spiralin-disrupted mutant GII3-9a2 was obtained by gene targeting through homologous recombination (Duret et al., 2003). Spiroplasma melliferum BC3 (ATCC 33219), Spiroplasma apis B31 (ATCC 33834), Spiroplasma floricola 23.6 (ATCC 29989), Spiroplasma phoeniceum P40 (ATCC 43115) and Mesoplasma florum L1 (ATCC 33453) were also used in this study. Spiroplasmas and mesoplasmas were grown at 32 °C in SP4 medium (Whitcomb, 1983) from which fresh yeast extract was omitted. Electrotransformation of S. citri has been described previously (Stamburski et al., 1991).

Plasmid constructs.
Plasmids pSR2, pBOG, pBSO and pSD4 have been described elsewhere (Lartigue et al., 2002; Renaudin, 2002). To construct pNE6, the spiralin gene of S. citri GII-3 was amplified from genomic DNA with primers SR26 and SR27. Then the 1·3 kbp EcoRI-restricted PCR product was inserted at the unique EcoRI site of pBOG. Genes scarp4a, 5a and 3b were amplified from genomic DNA of S. citri GII-3 with primer pairs 3B1F/4A1R, 5A1F/5A1R and 3B1F/3B1R, respectively (Table 1). After restriction with BamHI and BglII, the amplification products were inserted into the BamHI site of pSD4 to yield plasmids pNB4F/R, pNB5F/R and pNB3F/R [F and R indicate whether the scarp gene has the same (F) or the opposite (R) orientation of dnaA in the oriC fragment of pSD4].

Table 1. Primers

DNA isolation and Southern blot hybridization.
Spiroplasma genomic DNA was prepared from 10 ml cultures using the Wizard genomic DNA purification kit (Promega). After digestion with restriction enzymes, digested DNA fragments were fractionated by agarose gel electrophoresis, blotted to positively charged nylon membranes by the alkali transfer procedure, and hybridized with appropriate [digoxigenin]dUTP-labelled probes using standard stringency conditions (Sambrook et al., 1989). Hybridization signals were detected with anti-digoxigeninalkaline phosphatase conjugate and HNPP (2-hydroxy-3-naphthoic acid-2'-phenylanilide phosphate) as the substrate, following the supplier's instructions (Roche Diagnostics). Fluorescent signals were detected using a Fluor-S Multimager phosphoimager (Bio-Rad). Probes S4, specific to scarp4a, and S235, hybridizing to all scarp genes except scarp4a, were produced by PCR amplification of genomic DNA with primer pairs S4F/S4R and S235F/S235R, respectively. Probe F consisted of a 191 bp sequence of the hypothetical protein F that is present in all six plasmids (pSci16) of S. citri GII-3. It was generated by amplification of genomic DNA with primers repF and repR (Table 1).

Production of mAbs.
Spiroplasma cells from a 100 ml culture of S. citri GII3-9a2 were collected by centrifugation at 20 000 g for 20 min and washed three times in HS buffer (8 mM HEPES pH 7·4, 280 mM sucrose). The final pellet was resuspended in HS buffer to a final protein concentration of 1 mg ml¹ and used as the antigen. BALB/c mice were immunized by injection of 25 µg antigen emulsified with complete Freund's adjuvant by the footpad route. Two weeks later, an additional injection of antigen mixed with the incomplete adjuvant was performed by the subcutaneous route. Three days after the second injection, lymphocytes from ganglions and Sp210 myeloma cells were fused using the PEG procedure and the hybridomas were selected in HAT medium (Ravoet & Bazin, 1990). The hybridomas were grown in RPMI medium (Moore & Hood, 1993) and those producing antibodies specific to spiroplasma antigens were screened by ELISA and cloned by limiting dilution. Immunoglobulins G (IgG) were purified from the hybridoma supernatants by affinity chromatography on protein A-Sepharose columns according to standard procedures (Ausubel et al., 1995). Isotypes of mAbs were determined by ELISA.

Protein isolation, Western immunoblotting and colony blot immunoassay.
Spiroplasma cells from a 50 ml culture were collected by centrifugation and washed three times in HS buffer as described above. Protein concentration was determined as described by Bradford (1976) before the pelleted cells were lysed in 0·25 ml of Laemmli solubilization buffer (Sambrook et al., 1989). Protein separation by SDS-PAGE, Western immunoblotting and colony blot immunoassays were conducted as described previously (Duret et al., 2003).

Triton X-114 phase partitioning.
Spiroplasma proteins were separated into hydrophobic and hydrophilic fractions by the Triton X-114 partitioning method of Bordier (1981) essentially as described by Cheng et al. (1996). Spiroplasma cells from a 150 ml culture (10¹¹10¹² c.f.u.) were collected by centrifugation and washed three times in HS buffer as described above. The pelleted cells were dispersed in 1 ml Tris-NaCl buffer (10 mM Tris/HCl pH 7·4, 154 mM NaCl) containing 1 % Triton X-114 and the mixture was maintained at 4 °C for 40 min with gentle shaking. After removal of the insoluble components by centrifugation at 15 000 g for 10 min at 4 °C, the supernatant was incubated at 37 °C for 10 min to allow condensation of the detergent phase. The aqueous and detergent phases were then separated by centrifugation at 15 000 g for 10 min at room temperature. The aqueous phase was transferred to a new tube, chilled at 4 °C, and Triton X-114 was added to a final concentration of 1 % in a total volume of 1 ml. The detergent phase was adjusted to 1 ml with Tris-NaCl buffer without adding Triton X-114. After 10 min at 4 °C the mixtures were incubated at 37 °C for 10 min and then centrifuged at 15 000 g for 10 min at room temperature. This cycle was repeated three times to ensure complete partitioning. The final aqueous and detergent phases were adjusted with Tris-NaCl buffer to 1 and 0·5 ml, respectively.

Mass spectrometry (MALDI-TOF).
After SDS-PAGE the proteins were stained with colloidal Coomassie blue and the protein bands that were shown to react with mAb 10G3 by Western immunoblotting were excised for analysis. The in-gel trypsin digestion, MALDI-TOF (Applied Biosystems, Voyager DE super STR) analyses and peptide mass fingerprinting database searches were carried out at INRA (). The masses of peptides obtained from proteolytic digestion were compared to the predicted masses deduced from the S. citri GII-3 protein database. This database was constructed from sequence data of the ongoing S. citri GII-3 genome sequencing project (Foissac et al., 2004).

In silico analyses.
DNA and protein sequence analyses were performed using the programs proposed by Infobiogen (). Multiple alignments were done with MultAlin (Corpet, 1988). The BLAST program (Altschul et al., 1997) was used to search homologies in general databases () or in the Spiroplasma kunkelii partially sequenced genome ().

Experimental transmission assay.
S. citri was transmitted to periwinkle (Catharanthus roseus) plants via injection into its leafhopper vector (Circulifer haematoceps) by the method of Foissac et al. (1996) as described previously (Duret et al., 2003). Briefly, the insects were microinjected with 0·2 µl of the spiroplasma culture (approx. 10⁵ spiroplasma cells), and the injected insects were caged on young periwinkle plants (12 insects per plant, 510 plants per spiroplasma strain). After a 2 week transmission period, the insects were removed, and the plants were kept in the greenhouse at 30 °C for symptom development. Culture of S. citri from plants and insects has been described elsewhere (Foissac et al., 1997; Gaurivaud et al., 2000; Duret et al., 2003). Midribs (∼0·2 g) were minced with a razor blade in 2 ml SP4 medium. After 30 min at room temperature, the extract was filtered (0·45 µm) and dilutions were plated on agar medium to determine the number of c.f.u. Titres of S. citri in the insects were determined at the end of the transmission period.

mAbs against S. citri GII3-9a2
With the aim of characterizing S. citri antigenic proteins other than spiralin, the spiralin-defective mutant GII3-9a2 (Duret et al., 2003) was used as the source of antigens to immunize mice. Hybridomas producing immunoglobulins specifically directed against S. citri were first screened by ELISA and then by Western immunoblotting. Two hybridomas were selected, both of which produced antibodies of the IgG2b type. One of these, mAb 11A4, was found to react with a polypeptide of 70 kDa (Fig. 1a, lane 2). Unfortunately, subsequent passaging of hybridoma cells repeatedly resulted in loss of antibody production. The other, mAb 10G3, reacted with one major polypeptide of 43 kDa and several other polypeptides with apparent molecular masses ranging from 80 to 95 kDa (Fig. 1a, lane 3). These results suggested that distinct proteins shared a common epitope, or that the various signals corresponded to processed products of a single protein. As shown in Fig. 1(b), polypeptides reacting with mAb 10G3 were present in S. citri GII-3 and GII3-9a2 (lanes 1 and 2), and in S. phoeniceum (lane 4) but not in S. melliferum (lane 3), S. floricola (lane 5), S. apis (lane 6) or M. florum (lane 7).

(17K): Fig. 1. (a) Western immunoblotting, with mAb 10G3, of total proteins from S. citri GII3-9a2 with hybridoma supernatants 4C10, used as the negative control (lane 1), mAb 11A4 (lane 2) or mAb 10G3 (lane 3). (b) Western immunoblotting of total proteins from S. citri GII-3 (lane 1), GII3-9a2 (lane 2), S. melliferum (lane 3), S. phoeniceum (lane 4), S. floricola (lane 5), S. apis (lane 6) and M. florum (lane 7).

mAb 10G3 reacts with an adhesion-like protein
In Triton-X-114 partitioning the hydrophobic, membrane-associated proteins are incorporated into the detergent phase while hydrophilic proteins are sequestered in the aqueous phase. Western immunoblotting of S. citri GII-3 proteins using mAb 10G3 indicated that polypeptides P43 and P80-95 were amphiphilic, as they were completely partitioned in the detergent phase (Fig. 2b, lane 5). No polypeptides were detected in the aqueous phase (Fig. 2b, lane 6). Interestingly, several polypeptides were repeatedly detected in the detergent phase, whereas in total protein preparations, P43 was the major if not the only (depending on the experiment) polypeptide detected (Fig. 2b, lane 4). In mutant GII3-9a2 also, P43 was the major polypeptide detected in total proteins (Fig. 2b, lane 1). In the detergent phase, however, three major polypeptides of 43, 47 and 80 kDa were detected (Fig. 2b, lane 2).

(35K): Fig. 2. Coomassie blue-stained SDS-PAGE (a) and Western immunoblotting with mAb 10G3 (b) of proteins from S. citri GII3-9a2 (lanes 13) and GII-3 (lanes 46). Lanes 1 and 4, total proteins; lanes 2 and 5, proteins from the detergent phase; lanes 3 and 6, proteins from the aqueous phase. Gel bands from which proteins were submitted to MALDI-TOF analysis are boxed.

From one-dimensional SDS-PAGE, the four protein bands P95, P80, P47 and P43 (boxed in Fig. 2a) corresponding to polypeptides recognized by mAb 10G3 were excised, and submitted to mass spectrometry (MALDI-TOF) analysis. For protein identification the masses of generated peptides were compared to the predicted masses of peptides from the protein database of S. citri (Foissac et al., 2004). The results indicated that 26 % of peptide masses from P47 and 31 % from P80 matched a unique coding sequence (CDS), sharing homology with the adhesion-related protein P89 (Sarp1) of S. citri BR3 (Berg et al., 2001), with the matched peptides covering 28 % (203/725) and 36 % (268/725) of the CDS, respectively. The P80 protein was later named Scarp4a (see below). The peptides from P47 and P80 also matched, though to a lesser extent (15 % and 9 %, respectively), the related CDS, Scarp5a. In the case of P95, 21 % of peptide masses matched the CDS Scarp3a, over 28 % (245/866) of the protein. Analysis of P43 yielded no conclusive results. In agreement with the complex Western immunoblotting pattern (mAb 10G3 recognized several polypeptides), these mass spectrometry data suggested that the epitope reacting with mAb 10G3 was probably carried by more than one protein. In addition, the finding that peptides from both P47 and P80 matched the same CDS strongly suggested that P47 was a cleavage product of P80, which was annotated as Scarp4a (theoretical molecular mass 77 040 Da) in the genome of S. citri GII-3 (Foissac et al., 2004). Indeed peptides generated from P47 matched the C-terminal moiety of the protein whereas peptides from P80 were distributed all along the protein sequence (Fig. 3). According to these data, the cleavage site would be around and upstream of lysine residue K263. Search for homologies in the databases with the BLAST program revealed that Scarp4a shared 40 % identity and 54 % similarity (19 % gap) with the adhesion-related protein P89 (Sarp1) of S. citri BR3 (Berg et al., 2001), and 44 % identity and 54 % similarity (25 % gap) with the related protein Skarp1 of S. kunkelii (Davis et al., 2005). A survey of the S. citri GII-3 genome revealed that it contained eight genes encoding Sarp1-related proteins with sizes ranging from 77 to 95 kDa. Based on their similarities, the S. citri GII-3 proteins were classified into four groups (25) and named Scarp2a, 2b, 3a, 3b, 3c, 3d, 4a and 5a. Multiple alignments of these proteins together with proteins Sarp1 and Skarp1 revealed that they shared distinctive properties. They possess a highly conserved signal peptide sequence followed by 68 tandem repeats consisting of 3942 amino acids each, a conserved central region of about 330 amino acids, a variable region of about 100 amino acids, and a transmembrane α-helix followed by a charged C-terminal end. As exceptions to the rule, Scarp4a does not possess the amino acid repeats, and Scarp2b and 5a lack the D/E-rich, charged domain at the C-terminal end (Foissac et al., 2004). When compared to Scarp4a as a whole, the cleavage product P47 was found to contain most of the conserved region (more than 75 %). Therefore it is very likely that mAb 10G3, which reacts with P80 (Scarp4a) and its cleavage product P47, as well as P95 (Scarp3a), recognizes an epitope located in this conserved region. Interestingly, in S. citri GII-3, all eight scarp genes are carried by plasmids (Foissac et al., 2004). Genes encoding Scarp3a, 3d and 3c are carried by plasmids pSci1, pSci2 and pSci3, respectively. Scarp2a and 4a are encoded by pSci4, and Scarp2b, 3b and 5a are encoded by pSci5. Sequences encoding putative C-terminal-truncated Scarp polypeptides of 157 and 429 amino acids have been found in pSci6.

(54K): Fig. 3. Amino acid sequence of Scarp4a. Peptides identified by MALDI-TOF analyses are indicated. The peptides shared by both P47 and P80 are shaded in grey and the peptides specific to P80 in black. The amino acid sequence of the putative cleavage product P47 is boxed.

Scarps in transmissible and non-transmissible strains
To assess the putative role of Scarps in transmission of S. citri by its leafhopper vector C. haematoceps, we searched for the presence of Scarps in transmissible (GII-3, GII3-9a2) and non-transmissible (44, R8A2 and ASP-1) strains of S. citri by Western immunoblotting with mAb 10G3 (Fig. 4). As expected, a 43 kDa polypeptide was detected in total proteins from S. citri GII-3 and GII3-9a2 (Fig. 4a, lanes 1 and 2). In contrast, no signal was detected with non-transmissible strains 44, R8A2 and ASP-1 (lanes 35). Similar data were obtained with proteins from the Triton phase (Fig. 4b). While polypeptides of various sizes were detected in S. citri GII-3 (lane 1), none of them were detected in non-transmissible strains (lanes 35). Fig. 4 also shows that GII-3 and GII3-9a2 displayed different patterns, in that three major high molecular mass polypeptides were detected in GII-3 (lane 1) and only one in GII3-9a2 (lane 2).

(22K): Fig. 4. Western immunoblotting, with mAb 10G3, of total proteins (a) and membrane fraction proteins (b) from S. citri GII-3 (lanes 1), GII3-9a2 (lanes 2), 44 (lanes 3), R8A2 (lanes 4) and ASP-1 (lanes 5).

Expression of Scarps in S. citri ASP-1
To make sure that mAb 10G3 did react with all Scarps, three of them (Scarp3b, 4a and 5a) were expressed by cloning in S. citri ASP-1. The relevant genes were inserted into the oriC plasmid vector pSD4 downstream of the oriC fragment, and the resulting plasmids pNB3F, pNB4F and pNB5F were introduced into S. citri ASP-1 by electroporation. Spiroplasmal transformants were selected on SP4 agar plates containing tetracycline. For all three plasmids, direct immunoblotting of colonies revealed that each tetracycline-resistant colony reacted with mAb 10G3 (Fig. 5ej) similarly to those of GII-3 (Fig. 5a, b), whereas no signal was detected with spiroplasmas transformed by the insert-free vector pSD4 (Fig. 5c, d). These data clearly indicated that the epitope detected by mAb 10G3 was present in Scarp3b, 4a and 5a. Based on their similarities, it is very likely that all eight Scarps carry this epitope.

(47K): Fig. 5. Colonies (a, c, e, g, i) and direct immunoblotting, with mAb 10G3, of colonies (b, d, f, h, j) from S. citri GII-3 (a, b) and ASP-1 transformed by plasmids pNB3F (e, f), pNB4F (g, h), pNB5F (i, j) or insert-free vector pSD4 (c, d).

scarp genes in transmissible and non-transmissible strains
To determine whether the absence of Scarps in non-transmissible strains was due to the non-expression or to the absence of scarp genes, genomic DNAs of S. citri strains were analysed by Southern blot hybridization with scarp probes. The combination of DNA digestion by HincII with the use of two distinct probes, one of which is specific to scarp4a (S4) and the other (S235) to all scarp genes except scarp4a, made it possible to detect each scarp gene as a DNA fragment of specific size (Fig. 6a, b). In S. citri GII-3 (Fig. 6a, b, lanes 1), the sizes of the HincII restriction fragments corresponding to scarp2a, 2b, 3a, 3b, 3c, 3d, 4a and 5a were 2549, 3309, 763, 556, 664, 8121, 2509 and 1585 bp, respectively. No signal was detected in non-transmissible strains 44, R8A2 and ASP-1 (Fig. 6a, b, lanes 35), indicating that no recognizable scarp genes were present in these strains. Interestingly, GII3-9a2 lacked scarp2b, 3b and 5a, as indicated by the absence of the 3309, 556 and 1585 bp signals (Fig. 6a, b, lanes 2). Because these three genes are carried on a unique plasmid (pSci5), it was hypothesized that the spiralin mutant GII3-9a2 might have lost this plasmid. Indeed, hybridization of unrestricted DNA from GII-3 and GII3-9a2 with the F probe confirmed that GII3-9a2 lacked pSci5 (Fig. 6c, lane 2). None of these plasmids were detected in the non-transmissible strains 44, R8A2 and ASP-1 (data not shown).

(37K): Fig. 6. (a, b) Southern blot hybridization between HincII-restricted DNAs of S. citri GII-3 (lanes 1), GII3-9a2 (lanes 2), 44 (lanes 3), R8A2 (lanes 4) and ASP-1 (lanes 5), and probes S235 (a) or S4 (b). Names of scarp genes and sizes of the corresponding DNA fragments (in bp) are indicated. (c) Southern blot hybridization between unrestricted DNAs of S. citri GII-3 (lane 1), GII3-9a2 (lane 2), and the F probe.

Transformation of GII3-9a2 with wild-type spiralin gene
In previous studies, we found that insect transmission of the spiralin-defective mutant GII3-9a2 was much less efficient than that of the wild-type strain GII-3, leading to the conclusion that spiralin is probably required for efficient transmission of S. citri by its leafhopper vector (Duret et al., 2003). To test this hypothesis, the spiralin gene was introduced into GII3-9a2 by using the oriC plasmid pBOG as the vector. Transformation of GII3-9a2 cells by the recombinant plasmid pNE6 yielded gentamicin-resistant colonies which were shown to contain the plasmid (data not shown). As expected in the light of previous studies (Renaudin & Lartigue, 2005), passaging of the transformants resulted in plasmid integration into the host chromosome as revealed by Southern blot hybridization with the spiralin probe (Fig. 7a). DNA digestion by HpaI, HindIII, EcoRI plus BamHI, and PstI plus HindIII, yielded hybridizing fragments of sizes consistent with integration of pNE6 within the pBS sequences, as indicated in Fig. 7(b). In particular the 14·1 kbp HpaI fragment of GII3-9a2 (Fig. 7a, lane 2) was replaced by two fragments of 10 and 13 kbp in the pNE6 transformant (Fig. 7a, lane 3). Once integrated into the host chromosome, plasmid sequences were stably maintained, regardless of the presence or absence of selection pressure. Expression of spiralin in two of these transformants was demonstrated by direct immunoblotting of colonies (data not shown) and was further confirmed by Western immunoblotting (Fig. 8). Large amounts of spiralin, similar to that in GII-3 (Fig. 8, lane 1), were detected in the pNE6-transformed cells (lanes 4 and 5) but not in cells transformed with the insert-free pBOG (lane 3). In the experiments reported in Table 2, transmission of the GII3-9a2 pNE6 transformant, which we named GII3-9a2S, was compared to that of the spiralin-defective mutant GII3-9a2. The wild-type strain GII-3 was used as the control. In the case of GII3-9a2S, a higher proportion of plants (6/10 vs 4/10 in expt 1 and 9/10 vs 4/9 in expt 2) were infected (as compared to GII3-9a2). In addition, a majority of infected plants (4/6 in expt 1, and 5/9 in expt 2) developed severe symptoms, concurrently with plants infected by the wild-type strain GII-3, and in contrast to those infected by GII3-9a2, for which a 24 week delay was observed. However, in spite of the high-level expression of spiralin in GII3-9a2S, the transmission efficiency was not fully restored, as indicated by the observation that a few plants (4/10 in expt 1 and 1/10 in expt 2) were not infected (Table 2). The fact that, in addition to spiralin, GII3-9a2 also lacks pSci5 would explain why functional complementation could not be achieved by transformation with the spiralin gene alone. Therefore, to assess the putative role of pSci5 in transmission, attempts were made to produce a S. citri GII-3 variant lacking pSci5, in order to compare its transmission efficiency to that of GII3-9a2.

(22K): Fig. 7. (a) Southern blot hybridization between restricted DNAs from S. citri GII3-9a2 and GII3-9a2S, and the spiralin probe (SR14-SR16). Purified pNE6 (lanes 1, 4, 7 and 10), and genomic DNAs from GII3-9a2 (lanes 2, 5, 8 and 11) and the pNE6 transformant GII3-9a2S (lanes 3, 6, 9 and 12) were restricted by HpaI (lanes 13), HindIII (lanes 46), EcoRI plus BamHI (lanes 79) and PstI plus HindIII (lanes 1012). Sizes of DNA fragments are indicated in kbp. (b) Partial restriction maps and gene organization of the spiralin gene regions of S. citri GII3-9a2 and GII3-9a2S, and schematic representation of recombination between pNE6 and pBS sequences present in the chromosome of GII3-9a2. The maps are not to scale. Ori, S. citri chromosomal replication origin (oriC); O, reduced oriC fragment; p, spiralin gene promoter; spi, spiralin gene; Δ spi, 5'-truncated spiralin gene; gfp, GFP gene; GmR, gentamicin-resistance gene aacA-aphD of Tn4001; tetM, tetracycline-resistance gene tetM of Tn916; B, BamHI; E, EcoRI; H, HindII; Hp, HpaI; P, PstI.

(14K): Fig. 8. Western immunoblotting, with anti-spiralin mAb 12G9, of total proteins from S. citri GII-3 (lane 1), GII3-9a2 (lane 2), and GII3-9a2 transformed by the insert-free vector pBOG (lane 3) or by pNE6 (lanes 4 and 5).

Table 2. Transmission of S. citri GII-3 (wild-type), GII3-9a2 and GII3-9a2S to periwinkle (Catharanthus roseus) plants via injection into the leafhopper vector (Circulifer haematoceps)

Plasmid curing of S. citri GII-3
With the aim of obtaining S. citri strains lacking one or several plasmids, serial dilutions of a S. citri GII-3 culture were grown either at normal growth temperature (32 °C) or at sublethal temperature (37 °C) for 5 days. After plating the highest dilutions with visible growth, the plasmid contents of individual clones were determined by Southern blot hybridization of HincII-digested genomic DNA with the scarp probe S235. At 32 °C, 100 % (14/14) of the tested clones displayed plasmid patterns identical to that of the parental, wild-type strain GII-3. They all possessed the six plasmids pSci16. In contrast, when grown at 37 °C, 7/17 clones had plasmid patterns distinct from that of the initial strain. Five of them lacked pSci5, one lacked pSci3, and one lacked both pSci3 and pSci5 as indicated by the absence of the relevant hybridization signals (3309, 1585 and 556 bp for pSci5, and 664 bp for pSci3). One additional propagation at 37 °C yielded new variants and one particular clone lacked pSci1, pSci3 and pSci5 (Fig. 9, lane 8). The various hybridization patterns are illustrated in Fig. 9. In particular, this figure shows that the S. citri variant GII3-5 (lane 3) exhibited the same hybridization pattern as GII3-9a2 (lane 2). The absence of the 556, 1585 and 3309 bp fragments corresponding to scarp3b, 5a and 2b indicates that this variant has lost pSci5.

(54K): Fig. 9. Southern blot hybridization between probe S235 and HincII-restricted DNAs of S. citri GII-3 (lanes 1, 7 and 9), GII3-9a2 (lane 2), and of the distinct variants having lost pSci5 (GII3-5, lane 3), pSci3 (lane 6), pSci3 plus pSci5 (lane 4), pSci1 plus pSci5 (lane 5), and pSci1 plus pSci3 plus pSci5 (lane 8). Names of scarp genes and sizes of the corresponding DNA fragments (in bp) are indicated.

Experimental transmission of S. citri GII3-5
To determine whether pSci5 was required for efficient transmission, we compared the transmission efficiency of S. citri GII3-5 lacking pSci5 with those of S. citri strains GII-3 (wild-type), used as the control, and GII3-9a2, lacking both spiralin and pSci5. In these transmission tests, spiroplasmas were injected directly into the insect haemolymph, bypassing the passage through the alimentary canal. Hence, the efficiency of transmission to the host plant reflects the ability of the spiroplasmas to cross the salivary gland barrier. The results in Table 3 indicate that all strains multiplied to high titres in the insect, regardless of the presence or absence of pSci5. Interestingly, in the case of GII3-5, all plants (7/7 in expts 1 and 2) developed severe symptoms within 2 weeks post-transmission, similarly to GII-3. In contrast, symptoms in the few plants (4/9) infected by GII3-9a2 could only be observed after a 24 week delay. These results suggested that pSci5 was not essential for transmission. Taken as a whole, the data suggest that spiralin rather than pSci5 is required for efficient transmission of S. citri to periwinkle by its leafhopper vector.

Table 3. Transmission of S. citri GII-3 (wild-type), GII3-9a2 and GII3-5 to periwinkle (Catharanthus roseus) plants via injection into the leafhopper vector (Circulifer haematoceps)

In this study, we produced mAbs against a S. citri mutant lacking spiralin. mAb 10G3 was found to react with several polypeptides, indicating that these polypeptides shared a common epitope. Mass spectrometry (MALDI-TOF) analyses of the polypeptides reacting with mAb 10G3 led to the characterization of adhesion-related protein Scarp4a, which shares strong similarities with the previously described Sarp1 of S. citri BR3 (Berg et al., 2001) and Skarp1 of S. kunkelii (Davis et al., 2005). In the S. citri GII-3 genome, scarp4a is one of the eight scarp genes, all of which are carried on plasmids: scarp3a, 3d and 3c on pSci1, pSci2, and pSci3, respectively; scarp2a and 4a on pSci4; and scarp2b, 3b and 5a on pSci5. Genes arp1 of S. citri BR3 and skarp1 of S. kunkelii also are carried by plasmids, pBJS-O (accession no. NC_007101) and pSKU146 (Davis et al., 2005), respectively. In addition to genes encoding adhesion-related proteins, the S. citri and S. kunkelii plasmids carry genes encoding proteins involved in DNA partitioning and transfer such as Soj, TraE/TrsE, TraG and Mob. TraE proteins are known to be involved in pilus biogenesis, and structures that appeared to be conjugation pili were observed to connect S. kunkelii cells to each other and to insect cells (Ozbek et al., 2003). Moreover, the finding of oriT-like regions in the S. citri GII-3 plasmids, as in pSKU146, suggests that these plasmids might be involved in conjugative gene transfer between strains. An extensive description of the plasmids from S. citri GII-3 will be published elsewhere.

Search for the presence of Scarps by Western immunoblotting of proteins with mAb 10G3 in various spiroplasmas suggested that Scarps were restricted to plant-pathogenic spiroplasmas, since they were detected in S. citri, S. kunkelii (Davis et al., 2005) and S. phoeniceum P40, but not in S. melliferum, S. floricola, S. apis or M. florum. Interestingly, in spite of the ability of mAb 10G3 to recognize the various Scarps, none of them was detected in the non-transmissible strains R8A2, ASP-1 and 44 of S. citri, all of which produce spiralin. Southern blot hybridizations also revealed that these strains do not possess scarp genes and do not even carry plasmids pSci16. This apparent correlation between transmissibility and the presence of plasmids pSci16 strongly suggests that genetic determinants of transmissibility are carried by these plasmids. Spiroplasma adhesion-related proteins, in particular, might be involved in the insect transmission process by interacting with insect cell receptors. When injected into the leafhopper, all S. citri strains (insect-transmissible and non-transmissible) multiplied to equal extents in the haemolymph. Therefore, it is likely that adhesion-related proteins are required for the spiroplasmas to cross the salivary gland barrier. In agreement with this hypothesis, it has been shown that most of the Sarp1 (or P89) protein is exposed at the spiroplasma cell surface and that this protein is implicated in adherence of spiroplasmas to cells from the leafhopper vector C. tenellus (Yu et al., 2000). In this respect, the possibility that adhesion-related proteins may also be required for crossing of the midgut epithelial layer cannot be excluded.

Previously, we showed that the spiralin-disrupted S. citri mutant GII3-9a2 was poorly transmitted, leading to the conclusion that spiralin was required for efficient transmission of S. citri by the leafhopper C. haematoceps (Duret et al., 2003). The putative role of spiralin in the insect transmission process was further documented by the finding that, in vitro, this protein binds to glycoproteins from the leafhopper (Killiny et al., 2005). Thus, spiralin seems more likely to be involved in specific spiroplasmainsect cell interactions rather than playing a role in protecting the spiroplasma cell membrane within the insect (Castano et al., 2002). However, in spite of the increased transmission efficiency, transformation of GII3-9a2 with the wild-type spiralin gene failed to fully restore the wild-type phenotype. Therefore, the finding that the spiralin mutant GII3-9a2 had lost pSci5, encoding Scarp2b, 3b and 5a, suggested that, in addition to spiralin, pSci5 might also be required for efficient transmission. Unexpectedly, the transmission efficiency of S. citri GII3-5, a S. citri GII-3 strain having lost pSci5, was found to be quite similar to that of the wild-type strain GII-3, indicating that pSci5 was not essential for transmission. One possibility to explain the lack of an effect of pSci5 curing is that function(s) encoded by pSci5 could be complemented by pSci14 and/or pSci6. Indeed, most of the CDS of pSci5, including Scarps, are also present in at least one of the other plasmids. Thus, the reason why introduction of the wild-type spiralin gene into GII3-9a2 did not completely restore the transmission efficiency remains unclear. However, it should be mentioned that the spiralin mutant GII3-9a2 was obtained through two successive recombination events, leading to the insertion of 10 kbp foreign sequences into the spiroplasmal chromosome (Duret et al., 2003). Consequently, the possibility cannot be excluded that, in this mutant, expression of genes other than, and in the vicinity of, spiralin might be altered.

Hitherto, all S. citri plasmids were cryptic in that they were not associated with a given phenotypic character. The finding that non-insect-transmissible strains do not possess plasmids pSci16 suggested that these plasmids encode genetic determinants of transmissibility. In this respect, plasmid curing of S. citri GII-3 would be of primary importance to assess the role of plasmid-encoded genes in a given chromosomal background. In this study, growing S. citri GII-3 at sublethal temperature (37 °C) led to spiroplasma strains lacking one, two or three plasmids (pSci1, pSci3 and pSci5). However, in spite of multiple propagations at 37 °C, we were unable to produce a S. citri GII-3 strain having no plasmids. Unexpectedly, the use of curing agents such as acridine orange, acriflavine and novobiocin did not improve plasmid loss. Such results have been previously reported in chlamydiae, in which curing agents led to a paradoxical increase in plasmid copy number (Pickett et al., 2005).

The presumed implication of plasmid-encoded genes in insect transmission has also been hypothesized in the case of the plant mollicute Candidatus Phytoplasma asteris' (Nishigawa et al., 2002). Comparison of plasmids pOYM, from an insect-transmissible line, and pOYNIM, from a non-insect-transmissible line of the onion yellows phytoplasma, revealed that pOYNIM lacked two ORFs that exist in pOYM. In this case, it was thought that the two ORFs encoding, respectively, an SSB protein and an unknown protein might confer some selective advantages to the plasmid or to the phytoplasma in the insect cell environment.

In summary, our data suggest a correlation between the presence of plasmids pSci16 and the ability of S. citri strains to be experimentally transmitted via injection into the leafhopper vector C. haematoceps. In agreement with this hypothesis, plasmids related to pSci16 were also detected in all three transmissible S. citri strains (other than GII-3) that were tested in our laboratory. In addition, the insect-transmissible strain BR3-3X has also been found to carry a plasmid (pBJS-O; accession no. NC_007101) related to pSci16. However, the possibility that loss of plasmids and loss of transmissibility could be independent events cannot be fully rejected. Obviously, further experiments, such as plasmid transfer between strains, are needed to further confirm the correlation between the presence of plasmids pSci16 and the spiroplasmal transmissibility. The scarp genes represent putative genetic determinants of transmissibility. However S. citri plasmids encode many other proteins of unknown function. Genetic studies are now required to further investigate the role of the various plasmid-encoded genes in the biology of S. citri.

We are grateful to T. Delaunay for performing immunization of mice and producing hybridomas. We thank our colleagues J. L. Danet for injecting spiroplasma cultures into the insects, P. Bonnet and F. Ferrer for growing plants and insects, and E. Daguerre for excellent technical assistance. Support for N. Berho was provided by the Ministère de l'Enseignement Supérieur et de la Recherche.