|Home | About | Journals | Submit | Contact Us | Français|
As a first step toward the design of an epitope vaccine to prevent contagious agalactia, the strongly immunogenic 55-kDa protein of Mycoplasma agalactiae was studied and found to correspond to the AvgC protein encoded by the avgC gene. The avg genes of M. agalactiae, which encode four variable surface lipoproteins, display a significant homology to the vsp (variable membrane surface lipoproteins) genes of the bovine pathogen Mycoplasma bovis at their promoter region as well as their N-terminus-encoding regions. Some members of the Vsp family are known to be involved in cytoadhesion to host cells. In order to localize immunogenic peptides in the AvgC antigen, the protein sequence was submitted to epitope prediction analysis, and five sets of overlapping peptides, corresponding to five selected regions, were synthesized by Spot synthesis. Reactive peptides were selected by immunobinding assay with sera from infected sheep. The three most immunogenic epitopes were shown to be surface exposed by immunoprecipitation assays, and one of these was specifically recognized by all tested sera. Our study indicates that selected epitopes of the AvgC lipoprotein may be used to develop a peptide-based vaccine which is effective against M. agalactiae infection.
Mycoplasma agalactiae is the etiological agent of contagious agalactia, a disease affecting sheep, goats, and other small ruminants (5, 6). It primarily affects the mammary glands, joints, and eyes, causing various clinical manifestations, such as mastitis, arthritis, and conjunctivitis (3). In most cases, infections result in a rapid drop in milk production. The disease causes severe economic losses in areas where the economy is largely based on shepherding (14).
Membrane surface proteins play a fundamental role in the pathogenesis of mycoplasmas; in particular, the attachment of mycoplasmas to host cells is mediated by cytoadhesins (2, 23, 32). Mycoplasmas affect the immune system by inducing either suppression or polyclonal stimulation of B and T lymphocytes, inducing the expression of up- and down-regulating cytokines (19), and increasing the cytotoxicity of macrophages, natural killer cells, and T cells (19).
Previous studies have shown that fewer than 10 membrane proteins of M. agalactiae, with molecular masses between 18 and 80 kDa, are responsible for immunoresponses in sheep (28). Some of these proteins are exposed on the cell surface (28) and could be utilized to generate a vaccine against M. agalactiae. Vaccines currently used for contagious agalactia consist of either killed (4, 29) or attenuated (8, 9) infectious agents. Despite the utilization of those vaccines, contagious agalactia is still a problem; furthermore, development of vaccines by conventional methods is limited by several factors, namely, unwanted side effects, the presence of contaminating materials, and difficulty in storage. To overcome those limitations, synthetic peptides are an appealing alternative because they are selective, chemically defined, safer, and more stable. Peptide-based vaccines have the additional advantage that relatively short stretches of amino acid sequences or peptide epitopes are capable of eliciting a protective immune response. They can be selected in a protein, thus eliminating other epitopes potentially responsible for unwanted effects due to nonspecific or undesirable stimulation (25). Furthermore, peptide-based immunogens are more likely to resist denaturation, and they can be easily stored and transported without refrigeration (25); this is an important requisite, especially in the veterinary field.
The use of multiple antigen peptides (MAPs) has enhanced the potential of peptide vaccines. MAPs incorporate multiple epitopes in the same construct, since they are formed by a core matrix carrying a number of branched peptides. The generation of an optimal antibody response requires the recognition of the antigenic peptide fragments by T-helper cells, which in turn promote the engagement of antigen-specific B cells. An effective synthetic immunogen should therefore contain sequences known to activate both T and B lymphocytes. The incorporation in the immunogen of T and B epitopes from the same pathogen might be particularly useful for enhancing a specific immune response when infection occurs (18, 25).
The aim of the present study was to identify a strongly immunogenic protein of M. agalactiae and to identify and localize protein epitopes that could be valuable for the design of a synthetic vaccine against contagious agalactia.
Reference strain M. agalactiae PG2 was obtained from the collection of Istituto Zooprofilattico Sperimentale della Sardegna, Sassari, Italy.
Epidemic strains PCR1 and PCR2, isolated from the milk of Sardinian sheep in 1998 and 1999, respectively, were kindly provided by Istituto di Malattie Infettive, Facolta di Medicina Veterinaria, Sassari, Italy.
Bacteria were grown at 37°C using a mycoplasma broth base (Oxoid, Basingstoke, England) completed with Supplement G (4:1). Identification was performed by PCR (27).
For the present study, 96 sera were used, obtained from naturally infected sheep from different areas of Sardinia. All sheep showed the typical symptoms of contagious agalactia. Each serum was analyzed by Western blotting, and 10 representative sera were pooled. The pooled sera were aliquoted, stored at −80°C, and utilized to characterize the epitopes and obtain peptide-specific antibodies. Serum from an M. agalactiae-free animal (Sigma Chemical Co., St. Louis, Mo.) was included as a negative control.
Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) was performed by the methods of Laemmli (13) and Schagger and Von Jagow (24) using a 10% polyacrylamide separating gel. Bacteria were pelleted by centrifugation, washed once in phosphate-buffered saline (PBS), pH 7.4, and then heated in the sample buffer at 95°C for 5 min under reducing conditions (from 2.5 to 15% 2-mercaptoethanol) prior to electrophoresis. Electrophoresis was performed using the Mini-Protean II system (Bio-Rad, Hercules, Calif.) following standard protocols. Protein molecular masses were determined with protein markers (10-kDa protein ladder and prestained protein molecular mass markers; Gibco-BRL) run simultaneously. Proteins separated by SDS-PAGE were visualized by staining with Coomassie blue. For immunoblotting, electrophoresed proteins were transferred to nitrocellulose membranes in a Mini Trans-Blot Electrophoretic Transfer Cell (Bio-Rad) as described by the manufacturer. Membranes were blocked with 3% (wt/vol) bovine serum albumin in TNT (Tris [100 mM], NaCl [150 mM], 0.05% Tween 20). Polyclonal sera of infected sheep diluted 1:250 served as a primary antibody, and goat anti-sheep immunoglobulin G conjugated to alkaline phosphatase (Sigma) diluted 1:25,000 served as a secondary antibody. Color was developed with the BCIP/NBT substrate system (Sigma).
A gel fragment containing the 55-kDa antigen was excised from the Tricine gel and digested in situ with 0.1 μg of modified trypsin (sequencing-grade trypsin; Boehringer Mannheim, Mannheim, Germany) essentially as described by Rosenfeld et al. (20). Tryptic peptides eluted from the gel were fractionated by reverse-phase high-performance liquid chromatography (HPLC) on a micro RPC C2/C18 SC (100 by 2.1 mm) column (Smart system; Pharmacia, Uppsala, Sweden). Automated Edman sequencing was performed either directly with the blotted bands using the blot cartridge or by loading selected fractions from the tryptic HPLC run onto a polybrene-containing glass fiber filter. A pulsed-liquid gas-phase sequencer (Procise; Applied Biosystems, Forster City, N.Y.) was used.
Primer selection was performed on the avgC gene (GenBank nucleotide sequence accession number AF205063) (7) with help from a website (www.williamstone.com/primers/calculator). The following primers were used for PCR: AvgC-F (5′-TTACTTGGATCAGTTGCTT-3′) and AvgC-R (3′-AGCGTATTTCTGAGTTTTAA-5′). PCR was performed on Hybond PCR Gradient Plus using genomic DNA of the reference strain PG2 and clinical isolates. Genomic DNA was isolated according to the method described by Ausubel et al. (1). The PCR mix (25 μl) contained the following: deoxynucleoside triphosphates, 200 μM; primers, 0.2 μM; Taq polymerase, 0.5 U; DNA, 200 ng; and MgCl2, 1.5 mM. The thermal program included the following steps: (i) initial denaturation for 5 min at 95°C; (ii) 35 cycles of denaturation at 95°C for 30 s, annealing at 52°C for 40 s (with an increment of 10 s/cycle), and extension at 72°C for 45 s; and (iii) final extension at 72°C for 10 min. The PCR product was cycle sequenced by using the same primers (LipC-F and LipC-R) on an ABI Prism 310 capillary DNA sequencer and genetic analyzer (Applied Biosystems) by the dideoxy chain termination method with a fluorescence dye terminator (Applied Biosystems, Warrington, United Kingdom). PCR and sequencing were performed in duplicate.
The algorithms of Hopp and Woods (12), Zimmerman et al. (34), and Levitt (15) (from www.expasy.ch/cgi-bin/protscale.pl) predict immunodominant sites of linear epitopes in the protein sequence. Predictions derived by each method were compared, and only sequences identified by all three methods were selected.
From the BCM SEARCH LAUNCHER (http://dot.imgen.bcm.tmc.edu:9331/seq-search/struc-predict.html), the TMpred program was used to select a transmembrane region as a negative control. For each selected region, duplicate sets of overlapping peptides with lengths from 7 to 12 amino acids, which together covered the entire sequence of the region in question, were synthesized by Spot synthesis according to the method of Frank (10). Peptides were staggered at one-amino-acid intervals (offset one) to produce a completely overlapping set. Each peptide was synthesized on a separate spot using Whatman 1 Chr paper as a membrane. A piece of membrane was cut out, and the positions for the spots were marked with a pencil at a distance of about 1 cm. Prior to use, the membrane was chemically derivatized to provide free amino functionalities in small circular spots. This was achieved by two steps. (i) Fluorenylmethoxycarbonyl-β-alanine (Fmoc-β-Ala-OH) was esterified with free hydroxyl groups of cellulose by reacting the membrane for 3 h with a solution containing 0.2 M Fmoc-β-Ala-OH, 0.24 M diisopropylcarbodiimide, and 0.24 M N-methylimidazole in N,N-dimethylformamide (DMF). After cleavage of the protecting group (Fmoc deprotection), an even distribution of reactive amino functionality was obtained. (ii) Onto the positions for the spots another Fmoc-β-Ala-OH was coupled to these amino functionalities, but only a small aliquot (0.5 μl) of a solution containing 0.3 M 1-hydroxybenzotriazole-ester of Fmoc-β-Ala-OH in N-methyl-2-pyrrolidinone was dispensed onto each pencil point. After being washed, all residual amino functionalities on the sheet were blocked by acetylation. Removal of the Fmoc protecting groups generated free amino functionalities, which appeared as distinct blue spots after staining with bromophenol blue indicator dissolved in DMF. The dipeptide β-Ala-β-Ala served as an “anchor” for peptide synthesis at the spot position and as a spacer arm to improve the accessibility of the immobilized peptides. Peptides were synthesized by coupling amino acids to the anchor. Assembly of the individual peptides on the spots, numbered progressively, was carried out using Fmoc chemistry (Fmoc amino acids were from Novabiochem, Läufelfingen, Switzerland). In short, each step of amino acid coupling was performed twice, both times for 15 min at room temperature. Solutions of activated Fmoc-amino acid derivative (0.3 M) in N-methyl-2-pyrrolidinone were pipetted (0.9 μl) onto the blue spots. The completeness of the reaction was followed by the disappearance of the free amino group, monitored by bromophenol blue indicator: the color of the spots changed from blue to yellow. Fmoc groups were removed with 20% (vol/vol) piperidine in DMF in 5 min. After the final cycle, peptides were N-terminally acetylated with acetic anhydride. Side chain deprotection was carried out for 1 h with trifluoroacetic acid-dichloromethane (1/1) and 3% (vol/vol) triisobutylsilane as a scavenger. The peptides that were synthesized were used immobilized on the membrane in a solid phase antibody binding assay. Membranes were used immediately or dried and stored at −20°C until use.
Before use, frozen membranes were rinsed for 2 min in methanol and then incubated in blocking buffer for 1 h at room temperature. Individual sera and pools of polyclonal sheep sera were diluted 1:300 in blocking buffer and applied to the membrane. After incubation at room temperature for 1 h, the blots were washed five times with TNT and incubated with secondary antibodies (goat anti-sheep immunoglobulin G conjugated to alkaline phosphatase; Sigma) diluted 1:30,000 in TNT for 1 h at room temperature. After the washes, the blots were developed with the BCIP/NBT substrate system (Sigma).
Purification of peptide-specific antibodies using a highly purified immobilized antigen was performed following the protocol of Rybicki (21). Filter blots prepared as described above were saturated with 3% (wt/vol) bovine serum albumin in TNT and then incubated with shaking for 90 min with 10 pooled anti-M. agalactiae sheep sera (1:40). After washing, only antibodies against selected epitopes remained on the filter. Thereafter they were eluted using a stripping buffer (200 mM glycine [pH 2.8]) for 10 min. The filter was discarded, and solution containing antibodies was adjusted to pH 7.4 with 1 N NaOH.
To evaluate specificities of stripped antibodies for the 55-kDa protein, an immunoblotting assay was performed. Whole-cell proteins were electrophoretically separated and treated as previously described. A dot blot assay was performed to evaluate specificities of eluted antibodies for M. agalactiae using other mycoplasma species and bacteria isolated from sheep in safe and pathological conditions. Mycoplasma hominis, Mycoplasma bovis, Staphylococcus aureus, Salmonella enterica serovar Abortusovis, Escherichia coli, Streptococcus agalactiae, Streptococcus bovis, and M. agalactiae (positive control) were harvested during the exponential growth phase, washed, and resuspended in PBS. One microliter of each bacterial suspension was spotted onto a nitrocellulose membrane and air dried for 10 min; the membrane was then treated as previously described for the antibody binding assay.
For protein A preparation, 500 μg of protein A-agarose (Sigma), washed five times in PBS containing Triton X-100 (1% vol/vol), was incubated in the same buffer at 4°C with gentle agitation for 6 h. Prior to use, protein A was centrifuged at 600 × g for 2 min and then resuspended in 300 μl of PBS-Triton X-100 (1% vol/vol).
Fresh log-phase culture broth of M. agalactiae samples corresponding to 300 μg of whole-cell proteins was incubated with 2 ml of eluted antibodies (as described above) for 2 h at 4°C with gentle agitation. Protein concentrations were determined using the DC protein assay (Bio-Rad). After centrifugation at 20,000 × g for 15 min, pellets were washed three times in PBS, pH 7.4. The immunocomplex was precipitated with protein A-agarose (Sigma) in PBS with 1% (vol/vol) Triton X-100 and incubated overnight at 4°C. After being washed five times in PBS-1% (vol/vol) Triton X-100, the precipitated antigens were resuspended in sample buffer at 95°C for 5 min before SDS-PAGE and immunoblotting.
SDS-PAGE profiling of whole-cell proteins was performed with M. agalactiae epidemic strains and reference strain PG2. All strains examined had the same protein pattern (Fig. (Fig.1);1); this pattern was also similar to that previously described by Tola et al. (26). Ninety-six sera obtained from naturally infected sheep from different areas of Sardinia were tested by immunoblotting; the antibody response was qualitatively homogeneous in all sera (data not shown). Several proteins were recognized by Western blotting with sera from infected sheep (Fig. (Fig.2).2). Five proteins with masses of 40, 55, 70, 80, and 110 kDa were detected in all the strains. Among these, a 55-kDa band was the most immunogenic and was constantly recognized by all tested sera. Interestingly, although the 55-kDa protein was weakly present in the SDS-PAGE profile (Fig. (Fig.1),1), it induced a strong immunoresponse in Western blotting (Fig. (Fig.2).2). This protein appear to be less represented in reference strain PG2 than in epidemic strains. In PG2 the most immunogenic protein was recognized in the region of 45 kDa. Tola et al. (28) have shown that the 55-kDa-band protein was surface exposed and accessible to antibodies developed during natural infection. Using a Tricine gel, the 55-kDa band protein was further resolved in three bands (Fig. (Fig.3).3). The upper band in the region of 50 to 60 kDa, more reactive than the other two, was excised from the Tricine gel and trypsinized. The peptide mixture was separated by reverse-phase HPLC and subjected to Edman sequencing. The N-terminal sequence (AKCGGTKEENKKPAE) and the internal sequence T27 (MAQDTDVSGADISG) were 100% homologous to M. agalactiae variable surface lipoprotein C (AvgC) (7) and they corresponded to amino acid residues 23 to 37 and 76 to 89, respectively. The T36 sequence (NSVVVSSNDFEGEVT) was 78% homologous to M. bovis proteins of the VSP family. T23 (FEVETSSGHK) and T39 (FEVY) had no significant homology to bacterial sequences from GenBank.
Based on the published sequence of the avgC gene (GenBank accession no. AF205063) (7), we generated specific primers to compare the avgC gene sequence of strain PG2 with the epidemic strains PCR1 and PCR2. A fragment of 659 nucleotides was amplified from the three strains and subjected to sequence analysis. The amplicon obtained from strain PG2 showed 100% homology to the published sequence of the M. agalactiae PG2 variable surface lipoprotein C gene (avgC) (7). Interestingly, sequence analysis of amplicons from epidemic strains PCR1 and PCR2 showed the same amino acid substitutions at position 124 (Asn instead of Asp), 176 (Met instead of Lys), and 201 (Arg instead of Lys) as with PG2 AvgC, indicating that the observed substitutions were conserved in different strains.
In order to localize epitopes in the AvgC antigen, the protein sequence was submitted for epitope prediction analysis using the algorithms of Hopp and Woods (12), Zimmerman et al. (34), and Levitt (15). Five regions predicted by all three methods were selected as shown in Table Table1.1. The transmembrane region LGSVASMA, used as a negative control, was obtained through the BCM SEARCH LAUNCHER. In the immunobinding assay, all peptides from regions 1, 2, 3, and 4 reacted with different intensities with a pool of sera from infected sheep. Region 4 peptides also reacted against the negative control, showing no specificity, and region 5 peptides presented low affinity. Table Table22 indicates the best peptides selected for specificity and sensitivity from regions 1, 2, and 3. The resultant reactivities of epitopes along region 2 are shown in Fig. Fig.44 as an example. An immunobinding assay performed with 32 sera tested with selected epitopes (Pep1, Pep2, and Pep3) indicated that Pep1, Pep2, and Pep3 reacted with 60, 100, and 80% of immunoblot-positive sera to the 55-kDa protein, respectively. None of above peptides reacted with immunoblot-negative sera. Peptide-specific antibodies against Pep1, Pep2, and Pep3 (α-Pep1, α-Pep2, and α-Pep3) were eluted from the corresponding peptides by the acid stripping procedure. To demonstrate their specificities for AvgC, the immunoblotting assay was performed with whole-cell proteins. All three peptide-specific antibodies recognized the 55-kDa protein in epidemic strains. A weak reaction with a protein of approximately 26 kDa was observed only with the PCR1 strain (Fig. (Fig.5).5). Electrophoretic separation was performed under reducing conditions, indicating that the 55-kDa protein was not a dimer formed by a disulfide bridge cross-linking the 26-kDa protein. To further exclude this possibility, we performed the experiment under stronger reducing conditions (up to 15% 2-mercaptoethanol). With PG2, monospecific antibodies recognized only a protein in the region of 45 kDa (Fig. (Fig.55).
To evaluate Pep1, Pep2, and Pep3 epitope specificity for M. agalactiae, dot blot analysis was performed with different bacteria tested against the eluted antibodies α-Pep1, α-Pep2, and α-Pep3. Those antibodies strongly recognized M. agalactiae, but none of them recognized M. hominis, S. aureus, S. enterica serovar Abortusovis, E. coli, or S. agalactiae. As expected, a weak reaction was observed with M. bovis. In order to evaluate whether the selected peptides Pep1, Pep2, Pep3 belonged to exocytoplasmic domains, immunoprecipitation was performed using α-Pep1, α-Pep2, and α-Pep3. Each of the three peptide-specific antibodies precipitated the 55-kDa protein; α-Pep1 yielded larger amounts of immunoprecipitate, suggesting a higher affinity to the protein than with α-Pep2 and α-Pep3 (data not shown).
In this study we have identified a 55-kDa protein, found to correspond to the AvgC protein, as one of the most constantly and intensely immunogenic proteins in M. agalactiae. The major epitopes of this protein have been selected and characterized. The presence of the protein was scarce in the SDS-PAGE profile, but it induced a strong and constant immune response in M. agalactiae-infected sheep. Others have indicated this protein to be a highly immunogenic surface molecule, expressed since the beginning of infection (28). N-terminal sequencing of the 55-kDa protein showed its correspondence with the AvgC protein, encoded by the avgC gene (7). Flitman-Tene et al. (7) have found that the chromosome of M. agalactiae contains a family of multiple avg genes with significant homologies to the corresponding region of the vsp genes in M. bovis (16). Some members of the Vsp family are known to be involved in M. bovis cytadherence to host cells (22). Variable membrane proteins of other Mycoplasma species, such Vaa of Mycoplasma hominis (33) and Mycoplasma synoviae A or B (MSPA or MSPB) (17), were also shown to be adhesive. The above analogies between M. agalactiae and M. bovis suggest that avg genes may be involved in cytadherence and indicate AvgC as a good candidate for immunization.
Peptide-specific antibodies eluted from selected epitopes recognized the 55-kDa protein with both epidemic strains, PCR1 and PCR2. Interestingly, with PCR1, peptide-specific antibodies also recognized a protein in the region of 26 kDa. Based on the published sequence, the expected molecular mass of AvgC would be about 26 kDa instead of 55 kDa as determined here. SDS-PAGE was performed here under strong reducing conditions (up to 15% 2-mercaptoethanol), excluding the occurrence of S-S-bond dimers to generate the 55-kDa protein. Moreover, for PG2, peptide-specific antibodies eluted from selected epitopes strongly recognized a protein in the region of 45 kDa, the same protein recognized by sera of infected sheep in immunoblotting. This suggests that the 55-kDa protein recognized in epidemic strains and the 45-kDa protein recognized in the reference strain are the same protein. Since major lipoprotein antigens of M. bovis undergo high-frequency phase and size variations and possess extensive reiterated coding sequences (23), it is possible that the same phenomena may occur, although the same reiterated coding sequences have never been described for M. agalactiae. An alternative explanation would include posttranslational modifications of the AvgC protein in M. agalactiae. Further mass spectrometry studies could clarify the relationship between these proteins recognized by the same antibodies.
The ability of mycoplasmas to overcome host immune defenses has been attributed to the rapidly changing antigenic repertoire of surface lipoproteins (19, 30) and to disregulation of cytokine expression (19). Rapid changes in the antigenic repertoire of the cell surface described for mollicutes (11, 19, 31) and the possible rearrangement of the avg genes potentially involved in M. agalactiae antigenic variation which occurs in vivo in the natural host (7) prompted analysis of the avgC DNA sequence of an epidemic strain isolated in Sardinia. Although the DNA homogeneity of isolates from different Italian regions was previously demonstrated by pulsed-field gel electrophoresis, suggesting that only one strain of M. agalactiae is present in Italy (26), the possibility cannot be excluded that in M. agalactiae avgC variability occurs at the DNA level by analogy with vsp genes of M. bovis (16). Six hundred fifty-nine of 717 bp of the avgC gene have been sequenced, indicating that the epidemic strain has the same sequence and differs from the reference strain PG2 in three nucleotides, resulting in three amino acid substitutions. The same substitutions have been observed in another strain (PCR2) isolated in a different year and in a different area of Sardinia, demonstrating that the mutations were conserved in epidemic strains. Two of these amino acid substitutions were within two of the five regions predicted to be most antigenic, while no substitution occurred in the epitopes selected by the immunobinding assay.
The use of highly overlapping peptides ensures that epitopes are not missed by being cut at critical points. This may occur if abutting rather than overlapping sequences are utilized. Immunoassays performed with epitopes differing only by one amino acid substitution indicate that this immunological technique is accurate for description of epitope structure. Octapeptides were used here because they reacted strongly with sera of infected sheep. The same test performed with peptides of variable length, from 9 to 12 amino acid residues and containing the selected octapeptide, showed lower reactivity. The peptide-specific antibodies α-Pep1, α-Pep2, and α-Pep3 immunoprecipitated the AvgC protein, suggesting that these epitopes are surface exposed. This may be an important requirement for their utilization in peptide vaccine development and may also be useful for diagnostic assays based on microbial capture or microbial labeling. Experiments with the peptide-specific antibodies α-Pep1, α-Pep2 and α-Pep3 against other mycoplasma species and ovine bacteria demonstrated the specificity of identified epitopes and confirmed their possible use for diagnostic purposes. The weak reaction observed with M. bovis was expected in view of homologies between the vsp gene of M. bovis and the avg gene of M. agalactiae. However, it should not represent a problem for diagnostic purposes, since the two mycoplasma species infect different hosts. The reactivities of the three epitopes with different sera (Pep2, 100%; Pep1, 60%; and Pep3, 80%) suggest that all three could be used together for immunization, while only Pep2 might be useful for diagnostic purposes.
Immunogenic epitopes described here for M. agalactiae may provide new indications for the development of single-antigen peptide or MAP vaccines and for the improvement of diagnostic tools. The use of a MAP vaccine including more than one immunogenic protein would be particularly advisable if it could be demonstrated that the AvgC protein undergoes phase variation.
This work was supported by Assessorato alla Programmazione e Bilancio della Regione Autonoma Sardegna.
We thank Piero Cappuccinelli and Sergio Uzzau for assistance and critical reading of the manuscript and Maurizio Taddei for help in the choice of a peptide synthesis method.
Editor: D. L. Burns