INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Over 180 species are now assigned to the genus Mycoplasma (25, 26). Most have been identified as infectious agents of humans or other animals (36, 38). Mycoplasma infections are rarely of the fulminant type but rather follow a chronic course, indicating a frequent failure of the host defense mechanisms to eradicate the parasites. Although the molecular basis for mycoplasma pathogenicity and chronicity remains largely elusive, it is well appreciated that variation of surface components plays a central role in establishing the chronic nature of mycoplasma infections and is an important parameter in the interaction of these small wall-less prokaryotes with their host (9, 26, 41, 42).

A common theme among pathogenic bacteria for maintaining surface variability is the utilization of clusters of variable genes undergoing random and spontaneous ON/OFF switching at a high frequency using diverse genetic mechanisms (3, 8, 14, 21, 28, 33, 34). One mechanism by which phenotypic diversity is generated involves chromosomal rearrangements that reassort coding and regulatory regions in order to activate silent genes or pseudogenes or to generate new coding sequences by chimeric gene fusions (1, 4, 5, 7, 8, 19, 22). As a result, the bacterial population is heterogeneous, displaying different antigenic phenotypes which enable efficient avoidance of host defense mechanisms (8, 33, 34, 42). Adaptive surface variation through error-prone mutational systems linked to multigene families has been proposed as a general mycoplasma strategy for generating high-frequency size and phase variations in coat proteins (3, 11, 23, 35, 43).

Lipoproteins in mycoplasma have attracted much attention in recent years due to their abundance in the single mycoplasma membrane in contrast to the limited number of lipoproteins in membranes of other eubacteria (25, 26). Furthermore, lipoproteins are the most dominant antigens in mollicutes, many of them were shown to undergo phase variation and to possess repetitive domains (26), a motif that is found in surface antigens of bacterial pathogens and is thought to be a ligand-binding domain (40).

One pathogenic mycoplasma species that extensively changes the antigenic characteristics of its surface lipoproteins is Mycoplasma bovis, recognized as the most important etiological agent of bovine mycoplasmosis in Europe and North America. M. bovis is capable of producing subacute to acute inflammation of various organs, including the udder, joints, and the respiratory or genital tracts (13, 24). Variation in the antigenic repertoire of the M. bovis cell surface is achieved by high-frequency phase as well as size variation of major lipoprotein antigens known as variable membrane surface lipoproteins (Vsps) (2, 29).

In a previous paper, we reported the identification and characterization of the vsp genomic locus of M. bovis (18). This locus of about 23 kb contains 13 single-copy vsp genes, each of which exists as a complete open reading frame (ORF) encoding a putative surface lipoprotein (18). All vsp genes encode highly conserved N-terminal domains for membrane insertion and lipoprotein processing (12), while the rest of the mature Vsp molecules display sequence divergence (18). A major portion of the vsp coding sequence is composed of in-frame tandem repeats that create a periodicity in the polypeptide structure. Eighteen distinct repetitive domains of different length and amino acid sequences were found within the various vsp genes. The repeat domain is subjected to size variation by spontaneous expansion or contraction of these repeating units (18). Each vsp structural gene is linked to highly homologous upstream regions composed of two cassettes.

Phenotypic switching of the Vsp proteins was shown to involve rearrangement events occurring at high frequencies of about 10² to 10³ per cell per generation within the vsp locus (17). The presence of multiple copies of high sequence similarity (18) allows for the modulation of the Vsp antigenic repertoire by recombination among vsp genes.

Recently, we have also shown that in addition to variations in the expression of individual Vsps, the vsp genomic repertoire is subject to changes. An intergenic recombination between closely related vsp genes (vspA and vspO) has led to the generation of a chimeric and functional vsp gene, namely, the vspC gene (19). The VspC product was shown to be accessible on the cell surface and to be a highly immunogenic antigen and to undergo an independent high-frequency phase as well as size variation (2).

In the present study, the molecular basis of VspA and of VspC phase variation, as representatives of the Vsp lipoprotein family, was investigated. We provide experimental evidence demonstrating that site-specific DNA inversions mediate VspA phase variation by fusion of an active promoter to the upstream region of a silent vsp gene. We propose a model for Vsp phase variation that involves site-specific DNA inversions between specific 35-bp sequences, designated vis (vsp inversion sequence), present within the conserved upstream region of all known vsp genes.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains, plasmids, chemicals, and growth conditions. The M. bovis strain used in this study was the PG45 type strain. Its origin and growth conditions have been described elsewhere (29). Escherichia coli strain DH5MCR (Gibco BRL Life Technologies, Inc., Gaithersburg, Md.) was used as a host. Recombinant clones were constructed in the plasmid vector pKS (Strategene, La Jolla, Calif.). E. coli cultures for plasmid isolation were grown in Luria-Bertani broth (31). Restriction enzymes, T4 ligase, and T4 polynucleotide kinase were purchased from MBI Fermentas, Amherst, N.Y. 5-Bromo-4-chloro-3-indolyl--D-galactopyranoside (X-Gal), isopropyl--D-thiogalactopyranoside (IPTG), and ampicillin were purchased from Sigma Chemicals (St. Louis, Mo.). [-³²P]ATP, [-³²P]CTP, and [-³³P]deoxynucleoside triphosphate. ([-³³P]dNTP) were purchased from Amersham, Little Chalfont, United Kingdom.

Electrophoresis and Western immunoblotting. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis was performed by the method of Laemmli (16). Samples were prepared by heating at 100°C for 5 min in sample buffer (2% SDS, 5% [vol/vol] 2-mercaptoethanol, 10% [vol/vol] glycerol and 62.5 mM Tris, pH 6.8). Proteins were separated in  9% acrylamide gel and were transferred to nitrocellulose membrane filters (0.45-µm pore size; Schleicher & Schuell, Dassel, Germany) by the method of Towbin et al. (37). Blot contents were incubated for 1 h at room temperature with phosphate-buffered saline (PBS) buffer containing 3% bovine serum albumin (Sigma, St. Louis, Mo.) and were then incubated overnight at 4°C with the primary antibodies diluted in phosphate-buffered saline buffer containing 20% (vol/vol) fetal calf serum. After three washes in PBS buffer, blots were incubated for 2 h at room temperature in peroxidase-conjugated goat antiserum to mouse immunoglobulin M or to mouse immunoglobulin G (Jackson ImmunoResearch Laboratories, West Grove, Pa., and Nordic, Tilburg, The Netherlands). For detection, the enzyme substrate o-dianisidine (Sigma) was used as previously described (2, 29).

DNA preparation and manipulation. Genomic DNAs from M. bovis PG45 clonal populations were extracted and purified by the method of Marmur (20). Plasmid isolation, restriction endonuclease digestions, and gel electrophoresis of DNA or proteins were performed as previously described (17, 44).

Oligonucleotides, labeling, and hybridization conditions. Synthetic oligonucleotides were synthesized on a model 380B DNA synthesizer (Applied Biosystems, Inc., Foster City, Calif.). The sequence of the oligonucleotides that target unique sequences of cassette no. 2 of the vspA, vspO, and vspL genes (designated A₂, O₂, and L₂, respectively) are as follows: 5'-GCTTTTATTTAGTTCTTAATACTTCATATAATAAA-3' (A₂), 5'-CCTGGGTAACAGATGCAA-3', (O₂), and 5'-GCTTCTTAAGTGCAATAT-3' (L₂). The EX-1 oligonucleotide (5'-AATTTATGCCTTTTTGCA-3') was used as for primer extension analysis. The RA-1 (5'-CGCCAGGTGTTTTATTTT-3'), the RA-4 (5'-GTTAGTTCCTGCACCTTGTT-3'), and the RF-2 (5'-TGGTGCTTTAGGTGCTCC-3') oligonucleotides were used as probes in Northern blot analysis. The conditions of oligonucleotide labeling or DNA labeling and Southern blot hybridization were described elsewhere (17).

RNA isolation and Northern blot analysis. Total cellular RNA was extracted from mid-logarithmic-phase cultures of M. bovis PG45 clonal variants using the RNA isolation RNeasy kit (Qiagen, Hilden, Germany). Total RNAs (2 µg) were denatured for 10 min at 65°C in the presence of 65% formamide and 8% formaldehyde. RNA was fractionated by electrophoresis in a 1% agarose gel containing 6% of formaldehyde (vol/vol) in morpholinepropanesulfonic acid buffer (31). After electrophoresis, the RNA was stained with ethidium bromide and was visualized under UV light. RNA marker I (Boehringer Mannheim, Mannheim, Germany) was included. The RNA was transferred onto a nylon membrane (Schleicher & Schuell) and was baked for 2 h at 80°C. The membrane was prehybridized at 42°C for 2 h in 50% formamide, 5× Denhardt's reagent, 0.1% SDS, 200 µg of salmon sperm DNA/ml, and 5× SSPE (1× SSPE is 0.18 M NaCl, 10 mM NaH₂ PO₄, and 1 mM EDTA [pH 7.7] (31). Hybridization was performed at 42°C in 50% formamide, 2.5× Denhardt's reagent, 0.1% SDS, 100 µg of salmon sperm DNA/ml, and 5× SSPE. The membranes were washed twice for 15 min each at room temperature in 6× SSPE and 0.1% SDS and once for 15 min at 37°C in 1× SSPE and 0.1% SDS. The membranes were dried and autoradiographed using Super RX Fuji X-Ray Film (Tokyo, Japan).

Primer extension. Primer EX-1 was end labeled using T4 polynucleotide kinase and [³²P]ATP and was then purified on a G-50 minicolumn (Boehringer Mannheim). The primer extension reaction contained 2 µg of total RNA, 100 ng of labeled primer, 3.9 µl of reaction buffer (0.1 M Tris-HCl, pH 8.3, 0.14 M KCl, 10 mM MgCl₂, and 10 mM dithiothreital) in a final volume of 15 µl. The reaction mixture was incubated for 10 min at 65°C, followed by 5 min at room temperature. After annealing, a mixture of dNTPs (2.5 mM each) (Boehringer Mannheim) and 5 U of avian myeloblastosis virus reverse transcriptase (Promega) were added. The reaction mixture was incubated at 42°C for 60 min. RNAse (66 µg/ml) was then added, and the reaction mixture was incubated for 30 min at 37°C, incubation was then terminated by heat inactivation (10 min; 75°C). Primer extension products were mixed with loading buffer and were resolved on a 6% polyacrylamide sequencing gel. DNA sequence analysis was performed by the dideoxy chain termination method (32) using the Thermo Sequenase Radiolabeled Terminator Cycle Sequencing Kit (Amersham).

PCR. Reactions were carried out in 100 µl containing 100 ng of template DNA, 5 U of Vent DNA polymerase in its 1× ThermoPol Buffer (New England BioLabs), a 0.2 mM mixture of dNTPs, and 500 ng of each primer. PCR amplifications were performed using the PC-Personal cycler (Biometra, Gottingen, Germany) programmed for 31 cycles as follows: one cycle of 3 min at 95°C, 1 min at 54°C, and 90 at 72°C, followed by 30 cycles of 30 at 95°C, 30 at 54°C, and 1 min at 72°C. The reaction mixtures were then incubated for 10 min at 72°C and allowed to cool to 4°C. The amplicons were purified using High Pure Filter Columns (Boehringer Mannheim GmbH, Indianapolis, Ind.) and were directly sequenced. Synthetic oligonucleotides used as primers for the VspA variants were 5'-TGAATCTGATTCTCCCCC-3', 5'-TTTACATAGTGTTATTGTGC-3', and 5'-GCTTGTTCTCTTTGACCCAC-3' (designated P₁, P₂, and P₃, respectively). PCR primers for the VspC variants were 5'-TCATCTGCTGGTTTGTCAG-3', 5'-CACCAGAAGAGGAAAATGCTG-3', and 5'-GCCTTGATCTGTATTTTCGC-3' (designated Pe-3, Pi-3, and nt-2, respectively).

Nucleotide sequence accession number. Sequence data were analysed using MacVector 6.5.10 software. The nucleotide sequences of the vsp genomic regions reported in this study have been assigned the following GenBank accession numbers: the vspA ON and the vspA OFF configurations (see Fig. 3A and B), AF36969 and AF36970, respectively, and the vspC ON and the vspC OFF configurations (see Fig. 4A and B), AF36971 and AF36972, respectively.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Phase variation of VspA or of VspC involves rearrangement of vsp locus. Two Vsps (VspA and VspC) were chosen as representatives of the Vsp family to study the molecular mechanisms that mediate Vsp phenotypic switching. The genetic analysis was done using a linage of clonal isolates of M. bovis strain PG45 from successive generations of phase transition (i.e., ONOFFON) of only the VspA or of the VspC product. VspA phase variants expressing a 63-kDa product (Fig. 1A) were isolated using monoclonal antibody (MAb) 1E5 as previously described (2, 17). Isolation of VspC phase variants expressing a 75-kDa product (Fig. 1B) was done by immunostaining colonies of a VspC clonal isolate expressing the VspC product with anti-VspC MAb 2A8 to monitor variations in VspC expression (30). Individual colonies were isolated, and their progenies were plated and in turn were subjected to immunostaining. Continued switching (ON to OFF and vice versa) was confirmed by Western blot analysis.

View larger version (44K): [in this window] [in a new window]   FIG. 1.   Western blot analysis of M. bovis PG45 clonal isolates. Total cell proteins of clonal isolates representing successive generations and phase transitions of VspA (A and C) or of VspC (B and D) were subjected to SDS-polyacrylamide gel electrophoresis and were immunoblotted with the following MAbs: 1E5 (A), 2A8 (B and C), and 9F1 (D). The 63-kDa VspA, the 75-kDa VspC, the 40-kDa VspO, and the 55-kDa VspF proteins are indicated. Phenotypic transitions (ONOFFON, lanes 1 to 3, respectively) are indicated at the top with arrows.

View larger version (36K): [in this window] [in a new window]   FIG. 2.   Identification of vsp-related genomic fragments undergoing rearrangements during VspA or VspC phase variation. Chromosomal DNAs of the VspA variants (A) or of the VspC variants (B) were restricted with the HindIII restriction enzyme, subjected to Southern blot hybridization, and probed with a 2.3-kb HindIII fragment carrying the vspA gene in the OFF state (A) or with a 1.5-kb HindIII fragment carrying the vspC ON gene (B). Variable HindIII fragments present only in the VspA ON or in the VspC ON variants (A and B, lanes 1 and 3) are indicated by solid arrows. Invariant HindIII fragments (A and B) are indicated by open arrows. The names of the vsp genes corresponding to the HindIII genomic fragments are shown. Phenotypic transitions (ONOFFON, lanes 1 to 3, respectively) are indicated at the top with arrows.

View larger version (39K): [in this window] [in a new window]   FIG. 3.   Schematic representation of DNA inversion events in the M. bovis vsp locus during VspA phase variation. (A to C) The solid line represents about 17 kb of the vsp locus that underwent inversion during the transition from VspA ON to VspA OFF (A and C, respectively). The positions of the HindIII (H) restriction sites are marked. Large, yellow, open arrows indicate the location and direction of the vsp genes. The vspM, vspN, and vspO genes are shown in darker yellow. The location of other vsp genes (vspG, vspH, and vspl) is marked. The location of four non-vsp ORFs (ORF2 to ORF5) is shown by open-labeled arrows. A black block 5' to each vsp gene represents homologous cassette no. 1, while cassettes no. 2 are shown by pink blocks. Cassettes no. 2 of the vspA, vspL, and vspO genes (A2, L2, and O2, respectively) are differently colored. Several relevant HindIII fragments are indicated by parentheses with the corresponding size. Open triangles indicate the locations of the vis sites within conserved cassette no. 1. The postulated DNA inversions yielding three distinct vsp configurations (A to C) are indicated by crossed lines, and the involved vis sites (vis nos. 1 to 3) are indicated. The locations of the PCR primers (P1, P2, and P3) and their amplicons are marked by parentheses. (D) PCR amplification of M. bovis VspA variants and of the PG45-type strain. PCR primer pairs P1 and P3 (lanes 1 to 5) or P1 and P2 (lanes 6 to 10) were used to amplify a 1,500-or 490-bp region, respectively. Isolates included a PG45-type strain (lanes 1 and 6), VspA ON isolate No. 1 (lanes 2 and 7), VspA OFF no. 2 (lanes 3 and 8), VspA ON no. 3 (lanes 4 and 9), and VspA OFF no. 4 (lanes 5 and 10). The size of the PCR products is shown.

View larger version (34K): [in this window] [in a new window]   FIG. 4.   Schematic representation of DNA inversion events in the M. bovis vsp locus during VspC phase variation. (A to C) The solid line represents the portion of the vsp locus that underwent inversion during the transition from VspC ON to VspC OFF (A and C, respectively). The positions of the HindIII (H) restriction sites are marked. The HindIII site that underwent a change in its position is marked by an asterisk. Large open-labeled arrows indicate the location and direction of three vsp genes. A black block 5' to each vsp gene represents homologous cassette no. 1. The A2, E2, and F2 cassettes no. 2 are differently colored. Several relevant HindIII fragments are indicated by parentheses with the corresponding size. Open triangles indicate the locations of the vis sites within conserved cassette no. 1. The postulated DNA inversions yielding three distinct vsp configurations (A to C) are indicated by crossed and broken lines, and the involved vis sites (vis nos. 1, 4, and 5) are indicated. The locations of the PCR primers (Pi-3, Pe-3, and nt-2) and their amplicons are marked by broken parentheses. (D) PCR amplification of M. bovis VspC phase variants. PCR primers Pe-3 and nt-2 were used to amplify a 1.3-bp region from VspC ON and VspC OFF variants (lanes 1 and 2, respectively). (E) Identification of a PCR amplicon corresponds to the middle VspC configuration. PCR primers Pi-3 and nt-2 were used to amplify a 1.6-bp region of the VspC ON or VspC OFF variant (the 1.6-kb amplicon of the VspC ON is shown in lane 2). An authentic 0.7-kb HindIII genomic fragment obtained after HindIII digestion of the 1.6-kb PCR product of the VspC OFF variant is shown (lane 3, indicated by arrowhead).

DNA inversions within vsp locus change organization of vsp upstream regions. Each vsp gene member possesses a conserved 5' noncoding sequence divided into two cassettes (18). The first (cassette no. 1) is a 71-bp region 5' upstream of the ATG initiation codon and contains a putative ribosome binding site and exhibits 98% homology among all known vsp genes. The second (cassette no. 2), ranges in size from 50 to 180 bp (depending on the gene) and is more divergent.

View larger version (69K): [in this window] [in a new window]   FIG. 5.   Nucleotide sequence alignment of the upstream regions of the vspA OFF, vspA ON, vspO ON, vspC ON, vspF ON, and vspC OFF genes. Identical nucleotides are shown by dark, shaded boxes. The name of each vsp gene and its expression state (ON or OFF) are shown on the left. Numbers above sequences indicate nucleotide position relative to the initiation codon. The position of the putative transcriptional start sites (+1) is indicated by an arrow. A prokaryotic 70 -dependent consensus sequence (10) and a ribosome binding site (SD) are overlined. The positions of the two-vsp cassettes (nos. 1 and 2) are bracketed on the right. The arrow at position 71 marks the boundary between these two vsp cassettes. The position of the 35-bp region vis is indicated by a broken line. The binding site for the EX-I is marked by a solid line.

View larger version (23K): [in this window] [in a new window]   FIG. 6.   Detection of the change in the organization of the vsp cassette no. 2 during VspA or VspC phase variation. Chromosomal DNAs of the VspA (A to C) or of the VspC (D) phase variants were digested with the HindIII restriction enzyme and were subjected to Southern blot hybridization. Three oligonucleotides complementary to unique sequences of cassette no. 2 of the vspA and of the vspC (A2), vspL (L2), and vspO (O2) genes were used as probes as follows: A2 (A and D), L2 (B), and O2 (C). The HindIII genomic fragments carrying the corresponding cassettes are marked by labeled arrows. Molecular-size markers (in kilobases) are shown on the left. Phenotypic transitions (ONOFFON, lanes 1 to 3, respectively) are indicated at the top with arrows.

Site-specific DNA inversions occur during Vsp phase variation. Since cassette no. 1 exhibits 98% homology among vsp genes (18, 19), the exact boundary between the two cassettes in which the inversion event took place cannot be precisely determined. However, differences of a few nucleotides within cassette no. 1 of the involved vsp genes (vspA, vspC, vspF, and vspO) on the one hand and the divergent sequence of cassette no. 2 on the other hand, allow us to use these differences as distinctive fingerprints and to identify a 35-bp sequence, located at nucleotides 37 to 71 of cassette no. 1 of all known vsp genes, as the potential sequence for the DNA inversion events. This site was designated vis (vsp inversion sequence) (Fig. 5, indicated by a broken line; see also Fig. 3 and 4).

The vsp upstream A₂ cassette serves as active promoter. Comparison of the nucleotide sequences of the vspA and vspC genes and their flanking regions, during phase variation, revealed that the only sequence difference detected between the ON and the OFF expression states was the type of cassette no. 2. The expressed vspA and vspC genes possess the A₂ cassette in their upstream region, while in the silent form of these genes, the A₂ cassette has been replaced with the L₂ cassette and with the E₂ cassette, respectively (Fig. 3 and 4). The location of the A₂ cassette upstream of expressed vsp genes raised the possibility that the A₂ cassette is an active promoter and that acquisition of this element by other members of the vsp gene family will activate the recipient gene. Notably, the A₂ cassette is a unique sequence present as a single chromosomal copy. We monitored the presence of the vspA mRNA in the three VspA phase variants. Total cellular RNA was extracted and subjected to Northern blot analysis using an oligonucleotide specific for the repetitive domain RA-4 of the vspA gene (18). A single transcript of 1.24 kb hybridized with the RA-4 probe was detected in all VspA ON variants (Fig. 7A, lanes 1 and 3). However, no vspA-related mRNA was observed in the VspA OFF variant (Fig. 7A, lane 2).

View larger version (45K): [in this window] [in a new window]   FIG. 7.   Transcription analysis of the VspA and VspC phase variants. (A to D) Northern blot analysis of VspA and of VspC phase variants. Total cellular RNAs were extracted from the VspA (A and B) or from the VspC (C and D) phase variants in either the ON or the OFF state and were probed with the RA-4 oligonucleotide (A), the vspO structural gene (B), the RA-1 oligonucleotide (C), and the vspF-specific oligonucleotide (RF-2) (D). Transcripts corresponding to the vspA, vspO, vspC, and vspF genes are indicated by labeled arrows. The length of the mRNA is indicated. Phenotypic transitions of VspA are indicated at the top of the panels by arrows. (E to G) Identification of the transcription start sites of the vspA, vspO, vspC, and vspF genes. The autoradiogram of a 6% polyacrylamide gel used to analyze the extension products is shown. Primer extension analysis was performed using cellular RNA extracted from the Vsp phase variants. The letters above the lanes indicate which dideoxynucleotide was used to terminate the sequencing reaction. The resultant primer extension products of two VspA ON variants (E, lanes 1 and 2), VspO ON (F, lane 1) VspC ON (G, lane 1), and VspF ON (panel G, lane 2) are indicated by an arrow. Part of the nucleotide sequence deduced from the sequencing lanes is shown on the left. The transcriptional start site (+1) (arrowhead) and the putative 10 sequence (brackets) are marked.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

This study provides compelling evidence that Vsp high-frequency phenotypic switching is mediated by site-specific DNA inversions occurring within the vsp locus of M. bovis. These inversions occur at a high frequency, as evidenced by the ease at which Vsp clonal variants were isolated (2, 17). The target sites for the site-specific DNA inversions were identified within a 35-bp sequence (vis) present 37 bp upstream of the initiation codon of each vsp gene. The fact that the vsp genes are not all similarly oriented within a chromosomal locus allows DNA inversions to occur between any two vis copies that are oppositely oriented (10, 18).

Sequence analysis of the vsp locus from several VspA or VspC phase variants has shown that the phase-variable genes possess complete and uninterrupted ORFs regardless of the expression state. However, comparison of the nucleotide sequences upstream of the expressed vsp gene with all of the other vsp genes revealed that during phase variation, rearrangement of the vsp upstream regions involving three vsp genes had occurred. A region of 121 bp (designated A₂; Fig. 3 and 4) was found upstream of an expressed vsp gene.

The molecular basis of M. bovis phenotype switching appears to be mediated by specific sequences that are necessary for site-specific DNA inversions (vis) and for expression (A₂ cassette). The A₂ cassette was shown by primer extension experiments to possess the transcriptional start site for an expressed vsp gene and thus serves as an active promoter. The generality of the A₂ cassette as a promoter for transcription initiation of the vsp genes was demonstrated for the vspA, vspC, vspO, and vspF genes.

An interesting feature of the vsp system that emerged when the nucleotide sequences of the vsp loci of several VspA or VspC phase variants were compared was that the vsp configuration identified in the VspA OFF or VspC OFF variants could not be explained by a single DNA inversion event. We propose that VspA as well as VspC phase variation involves two site-specific DNA inversions (Fig. 3 and 4). In VspA phase variation, the first inversion of the 13.4 kb fragment occurred between two inverted vis copies of the vspA and the vspO genes (vis no. 1 and 2, respectively; Fig. 3A), generating a middle configuration shown in Fig. 3B. This configuration underwent a second inversion of a 9.5-kb fragment between two inverted vis copies of the vspA and vspL genes (vis no. 1 and 3, respectively; Fig. 3B), which led to the configuration shown in Fig. 3C of the VspA OFF variant. In VspC phase variation, the first inversion of 545 bp occurred between two adjacent, inverted vis sites of the vspE and vspC genes (Fig. 4A). The second inversion involved a DNA segment of 1,820 bp and occurred between the vis site of the vspE gene and the vis site of the vspF gene (Fig. 4B). PCR amplification and sequence analysis have confirmed that cells that underwent only a single DNA inversion and possess the locus as configured for Fig. 3B and 4B exist within the populations of the VspA or the VspC phase variants, as well as within the population of M. bovis PG45 type strain (Fig. 3 and 4).

Examination of the VspA variants revealed that in the two vsp configurations (Fig. 3B and C), the recipient of the A₂ promoter was the vspO gene. This raises an intriguing question: why are two site-specific DNA inversions needed to achieve the same goal? One speculative possibility may relate to the orientation of the vis sites. Since the DNA inversions occur between two vis copies that must be in their inverted orientations, it is not possible to express each of the genes by a single inversion. The occurrence of multiple DNA inversions within the vsp locus enables other silent vsp genes to be repositioned in the correct orientation with respect to the expressed vsp gene. Additional inversions will allow juxtaposition of the vsp promoter with different silent vsp genes and the expression of new variable lipoproteins. It should be noted, however, that the DNA inversions might occur concurrently. A growing population contains apparently distinct subpopulations that have undergone single or multiple inversions. The population as a whole therefore possesses a wide spectrum of vsp configurations, allowing an increased capability to diversify the antigenic repertoire of the mycoplasma cell surface and to ensure the presence of a desirable variant needed for survival in the case of a change in the host environment.

Analysis of VspC phase variation revealed that unlike the VspA phase variation, in each of the three vspC configurations, the A₂ promoter was ligated to a different vsp gene (vspC, vspE, and vspF; Fig. 4A to C, respectively). In addition, colonies exhibiting the VspC OFF phenotype that were picked from a plate using the colony immunoblot assay were shown to possess the third configuration in which the promoter was upstream of the vspF gene. In other words, the second DNA inversion event appears to be phenotypically silent. Why are two inversions needed for vspC phase variation? One intriguing speculation may relate to the size and the repetitive nature of the involved lipoproteins. An interesting study in Mycoplasma hyorhinis has shown that in antibody selected populations, the emergence of a prevalent, protective Vlp phenotype is not due to expression of a particular Vlp protein but rather can result from optional mutational pathways leading to expression of any long Vlp protein (6). The choice of pathways appears to be determined by the most favorable (high-frequency) pathway available to generate a long Vlp protein which facilitates escape from antibody mediated damage. In Mycoplasma pulmonis growth properties are affected by length of repetitive domain of the Vsa proteins (3). Interestingly, when one looks at the three promoter-recipient vsp genes during VspC phase variation, the vspE gene of the middle configuration is the smallest vsp gene, only 393 bp long, and possesses one repetitive domain (18, 19). In contrast, the expressed vspC gene (Fig. 4A) is 1,098 bp long and contains four distinct, repetitive domains representing more than 80% of the entire gene (19). Similarly, the expressed vspF gene (Fig. 4C) is 1,110 bp long and possesses two large, repetitive domains (18, 19). Although VspC phase variants, investigated in this study, were not isolated from M. bovis populations that were subjected to pressure posed by host antibodies, it is an intriguing issue whether spontaneously occurring, phenotypically silent inversions represent the most efficient pathway to generate in vivo a long surface lipoprotein. In that case, long Vsps may provide a shield from host antibodies capable of binding vital surface antigens of M. bovis as elegantly as was shown for the Vlp proteins of M. hyorhinis (6).

Chromosomal rearrangements have been shown to be associated with phenotypic switching of surface antigens in many bacterial pathogens. Homologous recombination, gene conversions, gene duplications, additions or deletions of tandem repetitive units, movement of transposable elements within the chromosome, and DNA inversions are frequently employed mechanisms regulating the expression of genes encoding surface antigens in other bacterial systems (8, 28, 33, 34). In some cases, rearrangement events enabling the activation of silent genes were reported (1, 3, 7, 14, 22, 27, 35). The maintenance of a silent gene during evolution suggests that expression of such a gene provides a selective advantage to the bacterium under certain environmental conditions. One of the commonly known mechanisms of activation of silent genes is translocation of the gene from its silent site to an expression site via gene conversions (8, 10, 27, 34). For example, the silent vmp genes of Borrelia hermsii can be activated by gene conversion events that place the silent gene downstream of a promoter (14). The silent vsa genes of M. pulmonis lack the 5' end region, and site-specific DNA inversions regulate their expression by reassorting the 5' end region from an expressed vsa gene with the 3' end region from a silent gene (3, 35). Although there are numerous examples of site-specific DNA inversion systems that regulate phase variation of surface antigens (4, 10, 22, 35), most of them are relatively simple, consisting of two recombination sites and a site-specific recombinase gene that regulate the expression of one or two genes but not of a large gene family. In comparison with other bacterial DNA inversion systems, the vsa system of M. pulmonis (3, 35) and the vsp system of M. bovis are more complex. Both systems utilize multiple recombination sites and DNA inversions to regulate expression of a large gene family. In this respect, the two-mycoplasma systems resemble most closely the R64-type shufflon system found in enteric bacterial plasmids within the IncI incompatibility group (15).

Most site-specific DNA inversions are catalyzed by members of either the invertase family (the Hin enzyme of Salmonella) or the bacteriophage integrase family of site-specific recombinases (8, 10). Comparison of the vis sequence with bacterial invertase target sites as well as to the vrs box of M. pulmonis (3, 35) did not reveal any homology. It is possible that different site-specific recombinases may play a role in the mycoplasmas.

The Vsp variants analyzed were shown to oscillate between ON and OFF expression at a high frequency of 10² to 10³ per cell per generation (2, 17, 29). The DNA inversions observed within the vsp locus occur therefore at high frequencies similar to those measured for the vsa genes of M. pulmonis (3, 35). Variation in the antigenic repertoire is a combination of the frequency of the genetic switch and the selective pressure of the environment for a particular surface protein. The selective pressure for an exposed surface protein of a bacterial pathogen can be very high if those proteins are recognized by the host immune system. For example, the host immune system was shown to provide the selective pressure for the variation of the S layer of the pathogenic bacterium Campylobacter fetus (39). The Vsp proteins may represent a similar case. Further studies are needed to determine in vivo whether certain site-specific DNA inversion patterns are preferable in the host as a result of the selective pressure provided by the immune system.