Mutation of Two Mycoplasma arthritidis Surface Lipoproteins with Divergent Functions in Cytadherence

Abstract

Mycoplasma arthritidis is a natural pathogen of rats, causing an acute polyarthritis. Previous studies identified two membrane-bound lipoproteins, Maa1 and Maa2, thought to be associated with cytadherence of M. arthritidis strain 158p10p9. We have since confirmed that Maa1 is a major adhesin, although the role of Maa2 has proven more elusive. Both proteins were capable of eliciting protective immunity in rats against challenge with the virulent strain 158p10p9, suggesting that they may be important in pathogenesis. The purpose of this study was to better understand the roles of Maa1 and Maa2 in cytadherence in vitro. Insertion mutants were created for both genes by transposon mutagenesis. In vitro adherence of the Maa1 mutant KOMaa1 to rat L2 lung cells was reduced to the level previously reported for a spontaneous low-adherence mutant of 158p10p9 in which Maa1 is truncated and nonfunctional. Surprisingly, adherence of the Maa2 mutant KOMaa2 was approximately fivefold greater than that of the wild type. Complementation of KOMaa1 and KOMaa2 with wild-type alleles of maa1 and maa2, respectively, returned adherence to wild-type levels. This work confirms our earlier observation that Maa1 is a major adhesin for M. arthritidis strain 158p10p9. Maa2, on the other hand, may play a suppressive or modulatory role, possibly serving to release organisms from microcolonies at certain stages of infection.

Survival and virulence of pathogenic mycoplasmas are generally considered to be highly dependent on very close associations with host cells. However, early reports that Mycoplasma arthritidis could neither hemadsorb nor attach to mouse fibroblasts or macrophages indicated that it might be deficient in that regard (4, 9, 11). We have since confirmed that M. arthritidis does not hemadsorb, but we also observed that it did, in fact, adhere very well to primary rabbit synovial fibroblasts (16). We subsequently showed that M. arthritidis could also attach to several other cell types, although not to kidney cells, suggesting that specific receptors may be involved. Adherence was dose dependent and reached saturation, adherent cells could be depleted from the population by serial passage over cultured L2 rat lung cells, and major adhesins were surface exposed and trypsin sensitive (17). Two monoclonal antibodies (MAbs) from a panel raised against M. arthritidis membrane antigens partially inhibited attachment. One, designated A9a, was directed against a 90-kDa protein, and the other, designated 7a, was directed against a 71-kDa protein (17). These proteins were designated Maa1 and Maa2, respectively. M. arthritidis produced both proteins during the course of infection, and both were capable of eliciting protective immunity in rats, suggesting roles in pathogenesis (20).

Further characterization revealed that Maa1 was an 86.5-kDa, basic, largely hydrophilic lipoprotein. Maa2 was a 61.4-kDa lipoprotein that was also basic and hydrophilic. Both proteins contained 29-amino-acid lipoprotein signal peptides that were 90% identical and 93% similar to each other (18, 21). Both migrated more slowly on polyacrylamide gels than their actual sizes predicted, probably because of the lipid modification. Maa1 is produced constitutively and does not vary in size. However, Maa2 varies rapidly and (presumably) spontaneously in both size and expression (19). Size variation is mediated by changes in the number of tandem 264-bp direct repeats that make up the bulk of the gene. Phase variation is mediated by changes in the length of a poly(T) tract between putative −10 and −35 sites (21). Both are caused by slipped-strand mispairing (10).

During the original adherence study, we isolated a spontaneous mutant that had lost the ability to attach to rat cells in vitro (17). This mutant, originally designated LC1, is now referred to as the low-adherence (LA) variant. Western blot analysis showed that the LA variant was missing the 90-kDa Maa1 but produced instead a single 24-kDa protein that was absent from the wild-type strain (17). We subsequently showed that the smaller protein was a truncated version of Maa1, which we now designate Maa1Δ, and that the maa1 gene in the LA variant contained a nonsense mutation halting expression about a third of the way into the coding region (18). Complementation of the LA variant with a full-sized copy of maa1 restored its adherence to wild-type levels, indicating a direct role for Maa1 in adherence (15). However, the LA variant is not completely Maa1 negative, because it continues to express the truncated protein. What role, if any, that protein may play in adherence or pathogenesis is not clear. Moreover, no such mutant was available to study the role of Maa2 in these processes. Therefore, we used transposon mutagenesis to produce isogenic M. arthritidis mutants lacking Maa1 and Maa2 expression in order to further characterize their effect on adherence.

MATERIALS AND METHODS

Bacterial strains, cultivation, and storage.

M. arthritidis strains 158p10p9 (5), 158 (3), and H606 (9) and the LA variant derived from strain 158p10p9 (17) were used in this study. Mycoplasma cultures were grown in modified Edward broth (EB) containing 2% (wt/vol) BBL mycoplasma broth base (Becton Dickinson and Company, Sparks, MD) and supplemented with 20% (vol/vol) heat-inactivated equine serum (HyClone, Logan, UT), 0.5% (wt/vol) l-arginine HCl (Sigma-Aldrich, St. Louis, MO), 0.5% (vol/vol) BBL IsoVitaleX, 0.2% (wt/vol) denatured salmon sperm DNA (ICN, Aurora, OH), 50 μg/ml ampicillin, and 0.002% (wt/vol) phenol red. For solid medium, 1.1% (wt/vol) BBL Select Agar was added (15). Kanamycin (40 μg/ml), tetracycline (5 μg/ml), or chloramphenicol (25 μg/ml) was added to the medium as needed. Frozen stocks were prepared by freezing 1-ml aliquots of overnight broth cultures at −70°C. Subcultures were started from these stocks by transferring an ice crystal from a frozen vial into fresh medium.

Escherichia coli strain JM109 (Promega, Madison, WI) was grown at 37°C in Luria-Bertani (LB) broth (Miller formula; Fisher Scientific, Fair Lawn, NJ) with agitation at 250 rpm or on agar plates (LB medium supplemented with 1.5% [wt/vol] BBL Select Agar). Media were supplemented as needed with ampicillin at 100 μg/ml and with kanamycin, tetracycline, or chloramphenicol at the concentrations listed above. Stock cultures were created by freezing 0.5-ml overnight broth cultures in an equal volume of sterile glycerol at −70°C.

Filter cloning.

To distinguish genuine Maa2-negative mutants from “off”-switched phase variants, Maa2-negative clones isolated after M. arthritidis transformation (see below) were expanded in broth medium, filtered through a 0.45-μm-pore-size membrane, and plated onto solid medium. When growth became visible, a single colony was transferred to broth and grown until a color change became visible. This culture was similarly filtered and plated. The process was repeated a total of four times, after which the resulting colonies were subjected to colony blotting with anti-Maa2 MAb 7a as described previously (19) to determine whether they had begun to produce Maa2.

DNA preparation.

Mycoplasmal genomic DNA was prepared as described previously (15). Plasmid DNA was isolated from E. coli by a large-scale alkaline lysis procedure for use in mycoplasma transformations and by the boiling method for routine screening (1). For sequencing, plasmid DNA was isolated with Qiagen mini or midi purification kits as described by the manufacturer (Qiagen, Inc., Valencia, CA). For some cloning procedures, DNA restriction fragments were isolated from low-melting-point agarose gels with Quantum Prep Freeze-'N-Squeeze DNA gel extraction spin columns as described by the manufacturer (Bio-Rad Laboratories, Hercules, CA). DNA was quantitated spectrophotometrically (SmartSpec Plus spectrophotometer; Bio-Rad).

Polyethylene glycol-mediated transformation of M. arthritidis.

Because strain 158p10p9 contains an AluI-like restriction-modification system (13), DNA was methylated prior to transformation (15). Strain H606 does not possess the restriction-modification system, so when it was used, the methylation step was omitted. Donor DNA was introduced into M. arthritidis by a polyethylene glycol-based protocol as described previously (13, 15). Transformants were selected on medium containing kanamycin, tetracycline, or chloramphenicol alone or in combination, as appropriate. Transformant colonies were transferred from agar to 1 ml EB containing the appropriate antibiotics. Once grown, as indicated by color change, cultures were frozen at −70°C pending further analysis.

Plasmids and genetic constructs.

Constructs used in this study are listed in Table 1 and PCR primer sequences in Table 2. Maps of the most important plasmids and locations of PCR primers are shown in Fig. 1.

TABLE 1.

Constructs used in this study

Plasmid	Characteristics	Source or reference
pGEM-T Easy	PCR cloning vector	Promega
pISM2062	Mycoplasma cloning vector containing Tn4001mod	8
pIVT	pISM2062 with 5-kb HincII fragment containing tet(M) in the SmaI site of Tn4001mod (Tn4001T)	6
pIVT::maa1	pIVT with maa1 in the BamHI site of Tn4001T	This study
pIVT::maa2	pIVT with maa2 in the BamHI site of Tn4001T	This study
pIVT::aphA3	pIVT with aphA3 in the BamHI site of Tn4001T	This study
pIVC	pISM2062 with a cat gene (Cmr) from E. coli in the SmaI site of Tn4001mod	6
pIV(MA)C	pISM2062 with a chimeric cat gene (Cmr) from E. coli driven by maa1 promoter in the BamHI site of Tn4001mod	This study
pIVK	pISM2062 with aphA3 (Kanr) in the BamHI site of Tn4001mod (Tn4001K)	This study
pIVK::maa1	pIVK with maa1 in SmaI site of Tn4001K	This study

TABLE 2.

Oligonucleotide primers and probes

Gene	Name	Sequence (5′→3′)a	Target (reference)
maa2	OF-F	CAGGCTACTTTTAGTAC	Forward, 5′ end of maa2-containing fragment (21)
	MA2BF	GCCGGATCCAGGCTACTTTTAGTAC	Same as OF-F, with BamHI site
	HMPR	GCAACTCATATTTAGGGTTGC	Reverse, 3′ end of maa2-containing fragment (21)
	MA2BR	GCGGGATCCTCATATTTAGGGTTGC	Same as HMPR, with BamHI site
	OF-R	GCCAATAACAAGCTAAGTGC	Reverse, 16 nt into the maa2 coding region
	TERM-OUT	CAAAATGCAACCCATAATATG	Forward, 3′ end of maa2-containing fragment
maa1	PF4	GATAAAGCGCACTGTTGTGCTC	Forward, 5′ end of maa1-containing fragment (18)
	PF4Sma	GCTCCCGGGGATAAAGCGCACTGTTGTGC	Same as PF4, with SmaI site
	PF4Bam	GCTGGATCCGATAAAGCGCACTGTTGTGC	Same as PF4, with BamHI site
	F4Bgl	GTTAGATCTCATAATTGAAAAAGATGCCG	Same as PF4, with BglII site (15)
	PR2	CATAATTGAAAAAGATGCCGC	Reverse, 3′ end of maa1-containing fragment (18)
	R2Blg	GTTAGATCTCATAAAGCGCACTGTTGTGC	Same as PR2, with BglII site (15)
	MA1R3	CCGGAAGCTTCTCCTATTTATATATTTAAAATG	Reverse, 519 nt into maa1 coding region, with HindIII site
	MA1R4	CGGAAGCTTCTCCTATTTATATATTTAAATG	Reverse, just upstream of maa1 ATG start site, with HindIII site
aphA3	KF	GAACCATTTGAGGTGATAGG	Forward, 5′ end aphA3-containing fragment
	KFBam	CACGGATCCGAACCATTTGTGGTGATAGG	Same as KF, with BamHI site
	KR	AGTTGGGATGTACTTCAG	Reverse, 3′ end aphA3-containing fragment
	KRBam	CCAGGATCCAGTTGCGGATGTACTTCAG	Same as KR, with BamHI site
	KRC	CTGAAGTACATCCGCAAC	Forward, complement of KR
tet(M)	TF1	GAAAAGAACGGGAGTAATTGG	Forward, 413 nt into tet(M) coding region (15)
	TRI	CCATACATAACGGAAAGAGCCG	Reverse, 314 nt downstream of 3′ end of tet(M) coding region
	TRC	GCTCTTTCCGTTATGTATGG	Complement of TR
	T-OUT	GGACTGCATAACATCTTCCGCAG	Reverse, near 5′ end of tet(M)-containing fragment in pIVT
	TM5	CCCAATCCCATAGCCATACCTATC	Forward, 313 nt downstream from 5′ end of tet(M)-containing fragment in pIVT (15)
	TM12	GGACATCCAATTATTTGTTCCCGC	Reverse, 1893 nt downstream from 5′ end of tet(M)-containing fragment in pIVT (15)
	InPCRL	CACATCGAGGTCCGTCTGAAC	Forward, 725 nt into tet(M) coding region
cat	CF	GGAAAAGCTTATGCAGAAAAAAATCAG	Forward, 5′ end of cat-containing fragment, with HindIII site
	CR	CACTTCCCGGGGCGTAGCACCAGG	Reverse, 3′ end of cat-containing fragment, with SmaI site
	CRBam	CACTTGGATCCGCGTAGCACCAGG	Same as CR, with BamHI site

^aRestriction sites in PCR primers are in bold.

FIG. 1.

Maps of plasmid constructs used in this study and locations of PCR primers. pBluescript sequences are indicated by black boxes; IS256 arms of Tn4001, by light gray boxes; the resident aacA-aphD aminoglycoside resistance gene on Tn4001, by boxes filled with a dashed-line pattern; maa1, by boxes filled with an pattern of alternating white circles and black squares; maa2, by boxes filled with a wavy-line pattern; the 5-kb fragment containing the tetracycline resistance gene tet(M), by boxes filled with thick diagonal black lines; the 1.4-kb fragment containing the kanamycin resistance gene aphA3, by boxes filled with a black-and-white diamond pattern; and the 731-bp fragment containing the chloramphenicol resistance gene cat, by boxes filled with narrow diagonal black stripes. Locations of enlarged segments in relationship to the constructs in which they are contained are shown by broken lines. Positions of PCR primers are indicated by arrowheads, with black for “forward” primers (reading left to right with respect to the diagrams) and gray for “reverse” primers (reading right to left with respect to the diagrams). Those primers that are identical to each other except for the presence or absence of restriction sites at their 5′ ends are enclosed in boxes. The direction of transcription is indicated on the large constructs by arrowheads and on the enlarged fragments by block arrows. Restriction sites are indicated by ^.

All constructs for M. arthritidis transformation were prepared from vector pISM2062, carrying the staphylococcal transposon Tn4001mod, which has unique BamHI and SmaI restriction sites near the end of one of the IS256 arms (8). pIVT, which contains the tet(M) gene (Tn4001T), and pIVC, which contains an E. coli-derived chloramphenicol acetyltransferase (cat) gene driven by a Mycoplasma pulmonis promoter (Tn4001C), were constructed by Dybvig et al. (6). pKV98, which contains a cat gene from Staphylococcus aureus, was constructed by Hahn et al. (7). These three plasmids were provided by Kevin Dybvig, University of Alabama at Birmingham.

For pIV(MA)C, a chimeric gene designated macat was constructed, consisting of the maa1 promoter and 5′ untranslated DNA extending to just before the ATG translation start site (amplified with primer pair PF4Sma/MA1R4 [18], which placed SmaI and HindIII sites at the 5′ and 3′ ends, respectively [Fig. 1A]) and the E. coli-derived cat gene, minus its promoter (amplified from pIVC with primer pair CF/CR, which placed HindIII and SmaI sites at the 5′ and 3′ ends, respectively [Fig. 1C]). The maa1 promoter and cat amplicons were digested with HindIII, ligated, and further amplified with PF4Sma and CR. The final product, macat, was inserted into the A-T vector pGEM-T (Promega). BamHI sites were then placed at each end of macat by PCR (primer pair PF4Bam/CRBAM), and the new amplicon was inserted into the BamHI site of pISM2062, which resulted in the construct designated pIV(MA)C (Fig. 1C). Growth of E. coli transformed with pIV(MA)C in the presence of chloramphenicol demonstrated that the chimeric gene was functional.

To construct pIVK, a 1.4-kb fragment containing the kanamycin resistance gene aphA3 was amplified (primer pair KFBam/KRBam, which placed BamHI sites at each end [Fig. 1B]) from nisin expression vector pMSP3535VA (2) (provided by Keith E. Weaver, University of South Dakota). For preliminary experiments, aphA3 was inserted into the BamHI site of pIVT, resulting in construct pIVT::aphA3 (not shown). To produce a vector that did not contain tet(M), aphA3 was later placed into the BamHI site of pISM2062, to yield pIVK (Fig. 1B).

pIVT::maa1, used to complement the maa1 insertion mutant KOMaa1, was constructed in an earlier study (15).

To prepare the pIVT::maa2 constructs used to complement maa2 insertion mutant KOMaa2, a 2.1-kb fragment containing the maa2 gene plus its promoter and terminator was amplified with primer pair MA2BF/MA2BR (Fig. 1A), which placed BamHI sites at the 5′ and 3′ ends, respectively. Amplicons from three separate reactions were then inserted into the BamHI site of pIVT, yielding pIVT::maa2-6, pIVT::maa2-32, and pIVT::maa2-34 (Fig. 1A).

PCR.

PCRs were carried out in 0.2-ml thin-walled tubes in 50-μl reaction mixes containing 1× reaction buffer without MgCl₂, 10 mM deoxynucleoside triphosphates, 5 μM each primer, and 2.5 U (0.5 μl) Taq DNA polymerase (Promega). For primer sets TM5/TM12, OF-F/HMPR, and MA2BF/MA2BR (Fig. 1A; Table 2), 1.5 mM MgCl₂ was added to the reaction mixes. For all other reactions, 2.5 mM MgCl₂ was used. Primers were purchased from Sigma-Genosys (St. Louis, MO). Reactions were carried out in an Eppendorf Mastercycler gradient apparatus. Cycling conditions are listed in Table 3.

TABLE 3.

Thermal cycler settings

Primers	No. of cycles and step no.	Temp (°C)	Time
BigDye-2	1
	1	95	5 min
	45
	2	95	30 s
	3	52	30 s
	4	60	4 min
	1
	5	4	Hold
CF/CR, PF4Bam/CR, PF4Bam/CRBam,	1
KFBam/KRBam, KFBam/MA1R3,	1	94	5 min
KF/KR	3
	2	94	1 min
	3	37	1 min
	4	72°	1 min
	26
	5	94	1 min
	6	55	1 min
	7	72	1 min
	1
	8	72	10 min
	1
	9	4	Hold
PF4/PR2, PF4Sma/MA1R4,	1
F4Bgl/R4Bgl, PF4/KRC, PR2/KRC,	1	94	3 min
OF-F/HMPR, OF-F/KRC, MA2BF/	30
MA2BR, TRC/MAR3, TRC/TERM-	2	94	1 min
OUT, TRC/OF-R, TM5/TM12,	3	52	1 min
T-OUT/InPCRL, T-OUT/TF1	4	72	3 min
	1
	5	72	10 min
	1
	6	4	Hold

Inverse PCR was used to map transposon insertion sites in KOMaa2 clones complemented with the three pIVT::maa2 constructs and in one KOMaa1 clone complemented with pIVT::maa1. Nine micrograms of chromosomal DNA was digested with Sau3AI (for KOMaa2 clones) or NsiI (for the KOMaa1 clone) for 2.5 h at 37°C; the enzymes were heat inactivated for 20 min at 65°C. The entire reaction mix was ligated with T4 DNA ligase (Promega) in a total volume of 400 μl at 4°C overnight to circularize individual restriction fragments. Five microliters of overnight ligation mixes were used as PCR templates with primer pair T-OUT/InPCRL for KOMaa2 complemented clones or T-OUT/TF1 for the KOMaa1 complemented clone (Fig. 1A; Table 2). The resulting amplicons were inserted into pGEM-T and sequenced.

DNA sequencing.

Plasmid constructs were sequenced by the Iowa State University DNA Sequencing and Synthesis Facility.

Transposon insertion sites in two of the KOMaa1 clones complemented with wild-type maa1 were sequenced directly from chromosomal DNA by the University of South Dakota Department of Biology DNA Sequencing Facility. Sequencing samples were generated using the v3.1 cycle sequencing kit according to the manufacturer's instructions (Applied Biosystems, Foster City, CA) with primer T-OUT (Table 2; Fig. 1A). Thermal cycler conditions are given in Table 3. After PCR amplification, samples were applied to AutoSeq G-50 spin columns following the manufacturer's instructions (Amersham Biosciences). The DNA pellets were washed twice with 250 μl ice-cold ethanol and dried completely. Samples were sequenced immediately or stored overnight at −80°C.

Dot, Western, and colony immunoblotting.

Mycoplasma transformants were screened for expression of Maa1 and Maa2 by dot blotting. Suspensions of intact cells containing approximately 20 μg of mycoplasmal protein each were transferred to nitrocellulose membranes in a Bio-Dot apparatus (Bio-Rad). Membranes were processed as described for Western blotting (see below) with anti-Maa1 MAb A9a or anti-Maa2 MAb 7a (17). Ninety-four clones were screened per membrane; each membrane contained one sample of known positive wild-type strain 158p10p9 and known negative strain 158 (19). After dot blotting, selected clones were expanded, and DNA and protein stocks were prepared. Protein was quantitated with an RC DC protein assay kit (Bio-Rad).

For Western blotting, 20- to 40-μg mycoplasmal total protein samples were electrophoresed on sodium dodecyl sulfate (SDS)-polyacrylamide gels and transferred to nitrocellulose membranes as described previously (19). Membranes were blocked for a minimum of 1 h in TBS/T (20 mM Tris, 137 mM NaCl [pH 7.6], 0.1% [vol/vol] Tween 20) containing 5% (wt/vol) nonfat dry milk (TBS/T-milk). Blocked membranes were incubated for 1 h with shaking at room temperature with MAb A9a or 7a (17) diluted 1:5,000 in TBS/T-milk. Membranes were then washed with excess TBS/T and incubated with shaking at room temperature for 1 h with peroxidase-conjugated rabbit immunoglobulin G fraction anti-mouse immunoglobulin G (whole molecule-specific) (Organon Teknika, Cappel Division, Durham, NJ) diluted 1:100,000 in TBS/T-milk. After washing, the membranes were incubated for 5 min in SuperSignal West Pico chemiluminescent substrate (Pierce, Rockford, IL) following the manufacturer's instructions and scanned with a Typhoon 9710 imager (Amersham Biosciences, San Francisco, CA). Digital images were converted to TIFF format for analysis.

Colony blotting with MAb 7a was performed as previously described (19). Images were captured with a Spotflex 64-megapixel camera (version 4.6; Diagnostic Instruments, Sterling Heights, MI) mounted on an Olympus BX51 microscope with 4× and 10× objective lenses.

Adherence assays.

Adherence assays were carried out in 24-well tissue culture plates with the L2 rat lung epithelial cell line (ATCC CCL-149) as previously described (15, 17). Briefly, each well was infected with ∼2.5 × 10⁷ CFU mycoplasmas in 250 μl, plates were incubated for 6 h at 37°C in 5% CO₂, and wells were rinsed thoroughly to remove unbound organisms and trypsinized. Colony counts were performed on the original mycoplasma inoculum and on the trypsinized cell suspensions. An adherence index was calculated for each test sample by dividing the percentage of the original mycoplasma inoculum remaining attached to L-2 cells by the percentage calculated for wild-type 158p10p9. Each sample in a single assay was tested in triplicate, and each assay was repeated at least three times. Statistical significance was calculated by Dunnett's post hoc test.

Southern hybridization.

Three to five micrograms of chromosomal DNA and 160 ng of plasmid DNA were digested with SpeI, electrophoresed, and transferred from gels to nylon membranes (Hybond-N⁺; Amersham Pharmacia Biotech, Piscataway, NJ) by vacuum blotting (VacuGene XL; Amersham Pharmacia). Oligonucleotide probes were labeled with 3′-fluorescein-dUDP, and hybridization was performed according to the manufacturer's instructions (Amersham Gene Images 3′ oligolabeling module; GE Healthcare, United Kingdom). Signals were detected by exposure to Kodak Biomax Light-2 X-ray film (Eastman Kodak Company, Rochester, NY).

Mycoplasma growth curves.

For determination of growth rates, 7 × 10⁶ CFU were quickly thawed from frozen stocks and inoculated into 1 ml EB (with appropriate antibiotics when testing transformants). Colony counts were performed on 10-μl samples collected every 6 h for 60 h. Exponential growth was observed between 6 and 18 h postinoculation. Doubling times were calculated using the formula g = (log₁₀ N_t − log₁₀ N_o)/log₁₀ 2, where N_t and N_o = the number of cells at 6 h postinoculation and 18 h postinoculation, respectively.

RESULTS

Identification of a second antibiotic resistance marker for use in genetic complementation of insertion mutants.

Insertion mutagenesis and complementation of mutants with pISM2062-derived constructs required the use of two different antibiotic resistance markers; however, at the onset of this project, only one such marker had been identified (6). pISM2062 is a suicide vector, so gene delivery is accomplished by transposition of Tn4001mod or its derivatives into the mycoplasmal genome. The bifunctional aminoglycoside resistance gene aacA-aphD carried on Tn4001 does not function in M. arthritidis (6), but tet(M) carried on pIVT was a viable alternative for use as one of the two markers (6, 13, 15). For the second resistance marker, we determined MICs of three additional antibiotics for M. arthritidis, i.e., erythromycin, chloramphenicol, and kanamycin. M. arthritidis strains 158p10p9 and H606 both exhibited high-level resistance to erythromycin (MIC of >150 μg/ml) but were sensitive to chloramphenicol (MIC of ≤15 μg/ml) and kanamycin (MIC of ≤10 μg/ml). For the sake of convenience, we used M. arthritidis strain H606 for initial experiments, because transformation of this strain does not require methylation of donor DNA.

We first tested pISM2062-based constructs pKV98 (7) and pIVC (6), carrying cat genes from S. aureus and E. coli, respectively, with the latter driven by an M. pulmonis promoter. pKV98 functions in Mycoplasma pneumoniae (7), and pIVC functions in M. pulmonis (6). However, neither conferred selectable resistance to M. arthritidis; our results with pIVC confirmed earlier observations by Dybvig et al. (6). We next constructed a chimera of the E. coli cat gene fused to the M. arthritidis maa1 promoter and inserted it into pISM2062, yielding pIV(MA)C (Fig. 1C). This gene conferred resistance to chloramphenicol in E. coli, but again we were unable to isolate chloramphenicol-resistant M. arthritidis transformants.

Next, we transformed M. arthritidis strain H606 with pIVT::aphA3 (not shown), which carried both tetracycline and kanamycin resistance genes. We selected four dually resistant clones and showed by PCR (primer pair KFBam/KRBam [Table 2; Fig. 1B]) that all contained aphA3 (not shown). We then tested pIVK, the pISM2062-based construct containing aphA3 but not tet(M) (Fig. 1B), and again obtained kanamycin-resistant transformants containing aphA3. Moreover, the transposon was retained over 15 serial passages in the absence of antibiotic selection, as shown by Southern hybridization of SpeI-digested transformant DNA with KFBam (Table 2) (results not shown). We obtained similar results with strain 158p10p9. We then confirmed that pIVK could be used for gene delivery by transforming the LA variant with pIVK:maa1 (not shown). Two kanamycin-resistant transformants subsequently produced both full-sized Maa1 (from the insert) and the truncated protein (from the resident gene) (not shown). Finally, we demonstrated that two copies of the Tn4001 transposon, each carrying different antibiotic resistance genes, could coexist in the same genome by transforming LA variants already containing Tn4001T, constructed in an earlier study (15), with pIVK. PCR with primer pairs TM5/TM12 and KFBam/KRBam (Table 2; Fig. 1A and B, respectively) showed that dually resistant transformants carried both tet(M) and aphA3 (not shown).

Transposon mutagenesis of M. arthritidis.

pIVK was used to create an insertion library in strain 158p10p9. Approximately 5,000 clones were generated and screened by dot blotting for expression of Maa1 and Maa2 (not shown). We identified a single Maa1-negative mutant, designated KOMaa1. We identified more than 60 Maa2-negative transformants; however, since Maa2 is phase variable, it was necessary to eliminate any possible “off” variants from the pool of clones. This was accomplished by multiple rounds of filter cloning and serial passage, after which all but one clone, designated KOMaa2, showed a typical sectoring pattern on colony blotting, indicating that they were simply “off”-switched Maa2 phase variants. The absence of Maa1 and Maa2 expression in clones KOMaa1 and KOMaa2, respectively, was confirmed by Western blotting (Fig. 2A, lane 4, and B, lane 3).

FIG. 2.

Western blots of insertion mutants and their respective complemented clones. Gels (12.5% SDS-polyacrylamide) were loaded with approximately 20 μg mycoplasmal protein per lane. (A) Detection of Maa1 with anti-Maa1 MAb A9a. Positions of wild-type Maa1, the truncated Maa1Δ, and the degradation product (occasionally appearing as a double band) always seen in Maa1-containing lanes are indicated by arrows. Insertion mutant KOMaa1 (lane 4) did not produce Maa1. Wild-type 158p10p9 (lane 2) produced full-sized Maa1, while the LA variant (lane 3) produced only Maa1Δ. Maa1 production was restored in KOMaa1 complemented with wild-type alleles of maa1 in three different chromosomal locations (lanes 5 to 7). (B) Detection of Maa2 with anti-Maa2 MAb 7a; size variation occasionally results in a ladder pattern on Western blots with this MAb (21). Insertion mutant KOMaa2 (lane 3) did not produce Maa2. Wild-type 158p10p9 (lane 2) produced full-size Maa2. Maa2 production was restored in KOMaa2 complemented with three different wild-type alleles of maa2 in three different chromosomal locations (lanes 4 to 6).

Molecular characterization of KOMaa1 and KOMaa2.

The figure in the supplemental material shows the PCR products described in this section, along with further details. PCR primer sequences are listed in Table 2. The amplification limit of conventional PCR is approximately 5 kb, and insertion of Tn4001K into maa1 and maa2 would render both regions too large to amplify with primers flanking the genes. To confirm that the transposon insertions were actually in the genes of interest, we attempted to amplify those genes from the insertion mutants with primer pairs PF4/PR2 (18) for maa1 and OF-F/HMPR (21) for maa2 (Fig. 1A). No products were amplified from KOMaa1 or KOMaa2. Amplification of the 1.4-kb aphA3 fragment from KOMaa1 and KOMaa2 with primer pair KF/KR (Fig. 1B) confirmed that the transposons were present in both mutants but not in wild-type 158p10p9.

To map the insertion site in KOMaa1, we paired maa1-specific primers PF4 and PR2 with primer KRC (Fig. 1B), which reads out from the aphA3 gene at the left end of the transposon and into flanking chromosomal DNA. Depending upon the orientation of the transposon and its insertion into the correct gene, one of the two pairs should yield an amplicon. Primer pair PF4/KRC amplified a ∼550-bp fragment, indicating that the transposon had inserted into the coding region of the maa1 gene approximately 300 bp downstream from the start codon. Sequence analysis confirmed that the insertion site was between nucleotides (nt) 316 and 317.

We used a similar strategy to map the insertion in maa2, pairing maa2-specific primers OF-F and HMPR with primer KRC (Fig. 1B). The OF-F/KRC pair amplified a 400-bp fragment mapping to the 5′ end of the maa2 coding region. Sequence analysis identified the precise integration site as being between the A and T residues of the ATG translation start codon.

Functional characterization of KOMaa1 and KOMaa2.

Growth rate experiments showed little difference in the onset of exponential growth, maximum yield, or doubling times for KOMaa1, KOMaa2, and the wild-type strain (not shown).

KOMaa1 adherence to L2 cells was reduced to nearly the same level as that of the LA mutant (Fig. 3A), and those levels were both significantly less than that of wild-type 158p10p9 (P < 0.01). KOMaa2 adherence, however, was nearly fivefold greater than that of the wild type (P < 0.01) (Fig. 3B).

FIG. 3.

Adherence of KOMaa1, KOMaa2, and complemented clones to rat L2 lung cells in culture. The adherence index was calculated as the percentage of each test sample divided by the percentage of wild-type strain 158p10p9 adhering in the same experiment. Each bar represents results from at least seven replicate samples ± standard deviation (SD). Statistically significant differences (P < 0.05) in adherence of mutants compared to the wild-type strain are indicated by asterisks. (A) The maa1-negative mutant KOMaa1 and the LA mutant both attached significantly less well than the wild-type strain. Adherence was restored to wild-type levels by complementation of KOMaa1 with wild-type maa1. (B) The maa2-negative mutant KOMaa2 attached significantly better than the wild-type strain. Adherence was reduced to wild-type levels by complementation of KOMaa2 with three different samples of wild-type maa2.

In previous work, we showed that Maa1 and Maa2 are both surface exposed (17, 19). One possible explanation for the results of the adherence experiments described above is that elimination of one of these proteins could affect surface exposure of the other. However, in the dot blotting procedure used to screen the insertion library the antigen consisted of intact cells, so that only those proteins exposed on the cell surface were detected. This confirmed that Maa1 and Maa2 remained surface oriented on KOMaa2 and KOMaa1, respectively. A second possibility is that Maa2 might affect adherence by regulating the amount of Maa1 produced. Western blotting of KOMaa2 with the anti-Maa1 MAb A9a indicated that this was probably also not true, since the amount of Maa1 produced by KOMaa2 appeared to be similar to wild-type levels (Fig. 4). However, these results do not rule out more subtle interactions between the two proteins.

FIG. 4.

Western blot of wild-type strain 158p10p9, the LA variant, and Maa2 insertion mutant KOMaa2 with anti-Maa1 MAb A9a. Gels (12.5% SDS-polyacrylamide) were loaded with approximately 40 μg mycoplasmal protein per lane. Maa1 (arrow) was present at approximately equivalent amounts in the wild-type strain and KOMaa2.

Finally, because Maa2 appeared to have a negative effect on adherence, we predicted that the mycoplasma population recovered from cells (i.e., the adherent population) might contain a higher proportion of “off”-switched variants than the original inoculum. To test this, we performed colony blotting on plates used in adherence assays for counting colonies from inoculum and adherent populations. In one such experiment, of 339 colonies in the original inoculum, 22 were completely Maa2 negative (6.5%). In the adherent population, of 262 colonies, 40 were completely Maa2 negative (15.3%). The difference is highly significant (Fisher's exact test, P = 0.0019).

Complementation of KOMaa1 and KOMaa2 with wild-type genes.

To confirm that loss of Maa1 and Maa2 expression and the observed changes in adherence were due to gene inactivation by transposon insertion, KOMaa1 and KOMaa2 were complemented with full-length maa1 and maa2 genes, respectively.

KOMaa1 was transformed with the pIVT::maa1 construct (Fig. 1A) used to restore cytadherence to the LA mutant in an earlier study (15). We selected three dually resistant clones (resistant to tetracycline and kanamycin) for further study and designated them KOMaa1-C1, KOMaa1-C2, and KOMaa1-C3.

To complement KOMaa2, we prepared three complementation constructs, pIVT::maa2-6, pIVT::maa2-32, and pIVT::maa2-34, using three separate maa2 PCR amplicons. maa2-6 was identical to the wild-type gene. maa2-32 had a total of six PCR-induced point mutations, including a single base change from T to C at the 11th residue of the 16-nt poly(T) tract in the promoter. Two changes within the coding region resulted in amino acid substitutions, V351D and K562E. maa2-34 had four nucleotide changes, three of which were in the coding region, resulting in one conservative amino acid substitution, D168N. None of the substitutions were predicted to cause structural changes. We transformed KOMaa2 with each of these constructs and randomly chose eight dually resistant subclones for further characterization, three from the pool transformed with pIVT::maa2-6 (KOMaa2-C6), two from the pool transformed with pIVT::maa2-32 (KOMaa2-C32), and three from the pool transformed with pIVT::maa2-34 (KOMaa2-C34). All clones within each group were identical by the assays described below, so results from just one each are described.

A series of PCR experiments confirmed that the complemented mutants actually contained the maa1- and maa2-carrying transposons and that antibiotic resistance was not due to spontaneous mutations. PCR primer sequences are listed in Table 2. The figure in the supplemental material shows the PCR products described in this section, along with further details.

As described above, maa1 and maa2 could not be amplified by conventional PCR from the two insertion mutants KOMaa1 and KOMaa2 because of the presence of the large transposon in their coding regions. Complementation restored the wild-type 2.4-kb maa1 (primer pair PF4/PR2) and 2.1-kb maa2 (primer pair OF-F/HMPR) products to KOMaa1 and KOMaa2, respectively, confirming that wild-type genes had been placed into the chromosomes of the insertion mutants. aphA3 (primer pair KF/KR) was detected in all six complemented clones, verifying that the original Tn4001K transposons were still present. The 550-bp fragment overlapping aphA3 from Tn4001K and the resident maa1 gene (primer pair PF4/KRC) was detected in the three KOMaa1-complemented clones, and the 400-bp fragment overlapping aphA3 and the resident maa2 gene (primer pair OF-F/KRC) was detected in the three KOMaa2-complemented clones. This confirmed that Tn4001K remained in the same location in the complemented clones as in the original insertion mutants. The presence of Tn4001T in the complemented clones was confirmed by amplification of a 2.7-kb fragment containing tet(M) from all six complemented clones (primer pair TF/TR). Finally, a 1.3-kb region overlapping the 3′ end of tet(M) and the 5′ end of the cloned maa1 unique to the Tn4001T::maa1 construct (primer pair TRC/MA1R3) was detected in KOMaa1-C1, -C2, and -C3. Similarly, a 600-bp region unique to Tn4001T::maa2-C32 and -C34 (primer pair TRC/TERM-OUT) was detected in clones KOMaa2-C32 and -C34, and a 1.1-kb region unique to Tn4001T::maa2-C6 (primer pair TRC/OF-R) was detected in KOMaa2-C6, in which maa2 was in the opposite orientation (Fig. 1A).

These experiments demonstrated that the KOMaa1 and KOMaa2 complemented clones had the same Tn4001K insertions as their parent strains and that each also contained at least one copy of Tn4001T::maa1 or Tn4001::maa2, respectively.

Identification of transposon insertion sites in complemented clones.

Sequence analysis from whole-genome templates identified Tn4001T integration sites in two of the three KOMaa1 complemented clones, while sequence analysis of inverse PCR products provided the same information for the third KOMaa1 complemented clone and all three of the KOMaa2 complemented clones. Sequences were compared with the unpublished genome of M. arthritidis strain 158L3, a MAV1 lysogen originally constructed in our laboratory (14), by Trace Archive database Mega BLAST search (database, Mycoplasma_arthritidis_158l3-1_WGS). The insertion site in KOMaa1-C1 was the only one that could not be matched to a 158L3 sequence. Nearby putative open reading frames (ORFs) were subjected to BLASTP search (1). Three of the integration sites were in ORFs, while the other three were in noncoding regions. Whether transposon insertion into any of the noncoding regions disrupted nearby genes is not known. A more detailed description of the insertion sites is presented in the supplemental material.

Functional analysis of complemented clones.

Complementation of KOMaa1 and KOMaa2 had little effect on exponential growth rate or maximum yield compared to those of the parent strains (not shown), although exponential growth was delayed for 12 h for KOMaa2-C32. The normal lag time is 8 to 9 h. Complementation in all cases restored adherence to near-wild-type levels (Fig. 3).

We predicted that the T-to-C substitution in the poly(T) tract of maa2 complementation construct 32 might affect maa2 phase variation. All three KOMaa2 complemented clones were tested by colony blotting after six subcultures on consecutive days to allow phase variants to emerge. KOMaa2-C6 and KOMaa2-C34 showed sectoring patterns indicative of phase variation; however, maa2 was constitutively expressed by clone KOMaa2-C32 (Fig. 5).

FIG. 5.

Colony blots of KOMaa2 complemented clones KOMaa2-C6, -C32, and -C34 with anti-Maa2 MAb 7a. The “sectoring” pattern indicates phase variation. KOMaa2-C6 (A) and KOMaa2-C34 (C) show phase variation comparable to that of the wild-type strain (not shown). KOMaa2-C32 (B), which contains a mutation in the poly(T) tract in the maa2 promoter region, constitutively produces Maa2.

DISCUSSION

In this paper we report the construction and initial characterization of M. arthritidis strain 158p10p9 mutants failing to express Maa1 or Maa2. Tn4001K insertion into the maa1 and maa2 coding regions resulted in complete loss of these proteins in KOMaa1 and KOMaa2, respectively, and complementation with wild-type alleles restored them both. We undertook extensive molecular analysis of the mutants and their respective complemented clones to show that the original Tn4001K inserts remained in the genomes and at the original sites after complementation with wild-type alleles and that antibiotic resistance expressed by these mutants was due to insertion of the transposons and not to spontaneous mutations. We mapped the insertion sites of Tn4001T carrying wild-type alleles of maa1 and maa2 in the complemented clones and found that three insertions were within ORFs while three were in noncoding regions. Clearly those genes that were interrupted were not essential for in vitro growth, although one of the maa2-complemented clones grew more slowly immediately after removal from the freezer. This was overcome during exponential growth, and doubling times and final yields were not affected.

As we expected from earlier work (15), complete elimination of Maa1 in the KOMaa1 mutant significantly decreased its ability to adhere to rat cells in vitro, and complementation with the wild-type allele restored that ability to wild-type levels, thus confirming that Maa1 is a major adhesin for strain 158p10p9 (15). What effect loss of Maa1 will have on virulence for rats is presently under investigation. However, the results described here are consistent with our hypothesis that Maa1 plays a role in disease production. The ability of the anti-Maa1 MAb A9a to provide almost complete protection to rats against challenge with strain 158p10p9 also supports this hypothesis (20).

Our original adherence study indicated that Maa2 might also function as an adhesin, because anti-Maa2 MAb 7a partially inhibited attachment of M. arthritidis to rat cells in vitro (17). MAb 7a also protected rats against challenge with M. arthritidis, although to a lesser extent than anti-Maa1 MAb A9a (20). However, the role of Maa2 in adherence could not be directly assessed until we had created a mutant that failed to express it. To our surprise, loss of Maa2 actually enhanced adherence to rat cells in culture. Complementation with the wild-type maa2 reduced adherence to wild-type levels, indicating that enhanced adherence was not due to a polar effect or to an additional, unrelated mutation in KOMaa2.

It is unclear whether Maa2 phase variation occurs in vivo or, if so, whether it is random, as it is in vitro. The default setting in vitro appears to be “on,” because it is expressed by a majority of isogenic clones derived from 158p10p9 (19). If antigenic variation occurs in vivo, as is likely, this could promote systemic spread of a part of the population within the host, while the rest form microcolonies on mucosal surfaces. This could result in a situation functionally similar to “on/off” switching in other mucosal pathogens, such as Neisseria meningitidis, in which pilus production is transiently upregulated early in the course of infection and then switched off in favor of capsule biosynthesis as organisms invade the bloodstream (12). Our observation that Maa2 expression was switched off in a larger proportion of organisms recovered from rat cells than in the original inoculum is consistent with this idea.

Another intriguing possibility involves epitope masking of Maa1 by Maa2, as has been shown for the P56 and P120 antigens of Mycoplasma hominis (22). In contrast to the situation with P56 and P120, surface exposure of the Maa1 epitope recognized by MAb A9a does not change in conjunction with Maa2 phase variation (19), and we confirmed here that mutating maa2 affected neither surface exposure of Maa1 nor the amount produced. However, in the case of Maa1, the adherence epitope and the epitope recognized by the MAb apparently are located on different parts of the molecule. Evidence for this is provided by the fact that Maa1Δ, produced by the LA variant, is processed, inserted into the membrane, and recognized by MAb A9a, but it does not promote adherence (17). The anti-Maa1 MAb A9a does partially inhibit M. arthritidis adherence (17), but it may not do so by binding directly to the adhesin portion of the molecule, which may also explain why adherence inhibition is not complete. In the intact mycoplasma, Maa2 may associate with the Maa1 adherence epitope, perhaps partially blocking it or reducing its affinity for host cell receptors.

A similar explanation may be applied to the apparently conflicting observations that the anti-Maa2 MAb 7a (17) and the presence of Maa2 (this study) itself both suppress adherence. The adherent population is enriched for “off”-switched phase variants, and their adherence should not have been affected by the MAb. However, the majority of the adherent population still expresses Maa2. It is possible that binding of the anti-Maa2 MAb to “on”-switched phase variants sterically blocks the interaction of Maa1 or other adhesins with host cell receptors. Further studies will be required to solve this problem.

This study perhaps has raised as many questions as it has answered. We still know nothing about the nature of the host cell receptors for M. arthritidis, although they are likely to be specific, because they are not universally distributed in all cell types (17). While the role of Maa1 appears to be fairly straightforward, that of Maa2 is more complicated, and the nature of the interaction between the two, if any, remains unclear. The role of size variation also remains unclear. When maa2 is amplified from M. arthritidis, the full-size product is consistently more abundant than the smaller variants (21) (this can also be seen in panel 1, lanes 2, 5, 7, and 9, of the figure in the supplemental material). Because of this, we speculated previously that there may be some functional constraint favoring the larger product (21). The role of Maa2 phase variation will be a particularly interesting avenue to pursue, and the “locked-on” KOMaa2-C32 may prove to be a useful tool in that regard.