The fungal genus Pneumocystis contains multiple species including the causative agent of human Pneumocystis pneumonia, which afflicts individuals with impaired immune system function, such as Acquired Immunodeficiency Syndrome (AIDS) patients (Thomas, C. F., Jr. and Limper, A. H. 2004). Studying Pneumocystis organisms is difficult because they do not proliferate well in culture (Walzer, P. D. et al. 2001). Therefore, animal models have been the main source of organisms (Armstrong, M. Y. and Cushion, M. T. 1994; Dei-Cas, E. et al. 1998; Larsen, H. H. et al. 2002).

Pneumocystis murina, the species of Pneumocystis found in laboratory mice (Keely, S. P. et al. 2004), is of interest because the laboratory mouse provides an advanced animal model of host response to Pneumocystis infection (Wright, T. W. et al. 2001; Lund, F. E. et al. 2003; An, C. L., Gigliotti, F., and Harmsen, A. G. 2003; Qureshi, M. H., Harmsen, A. G., and Garvy, B. A. 2003; McAllister, F. et al. 2004; Empey, K. M. et al. 2004; Linke, M. et al. 2005; Qureshi, M. H., Empey, K. M., and Garvy, B. A. 2005; Linke, M. et al. 2006a; Linke, M. et al. 2006b). P. murina is a close relative of P. carinii and P. wakefieldiae, both found in rats, and a more distant relative of the human pathogen, P. jirovecii (Frenkel, J. K. 1976; Frenkel, J. K. 1999; Stringer, J. R., Cushion, M. T., and Wakefield, A. E. 2001; Stringer, J. R. et al. 2002; Keely, S. P. and Stringer, J. R. 2005; Redhead, S. A. et al. 2006). All four of these species feature an abundant surface protein called Major Surface Glycoprotein (MSG) (see Table 1), which has also been observed in Pneumocystis from ferrets, where it is known as gpA (Linke, M. J., Cushion, M. T., and Walzer, P. D. 1989; Tanabe, K. et al. 1989; Haidaris, C. G. et al. 1991; Nakamura, Y. et al. 1991; Haidaris, P. J. et al. 1992; Gigliotti, F. 1992; Stringer, S. L. et al. 1993; Kovacs, J. A. et al. 1993; Garbe, T. R. and Stringer, J. R. 1994; Wright, T. W. et al. 1994; Kitada, K., Wada, M., and Nakamura, Y. 1994; Wright, T. W. et al. 1995; Haidaris, C. G. et al. 1998).

Table 1 Abbreviations referring to P. murina MSG gene family and its expression site

Studies on other Pneumocystis species, primarily P. carinii, suggest that P. murina may use a family of MSG genes to produce antigenic variation in populations of the fungus dwelling in mice. P. carinii MSG is encoded by a multigene family, members of which are arranged as head-to-tail repeats located near the telomeres of all 17 chromosomes (Wada, M. et al. 1993; Kovacs, J. A. et al. 1993; Kitada, K., Wada, M., and Nakamura, Y. 1994; Wada, M. et al. 1995; Edman, J. C. et al. 1996; Stringer, J. R. and Keely, S. P. 2001; Cornillot, E. et al. 2002; Stringer, J. R. 2003; Stringer, J. R. 2005; Keely, S. P. et al. 2005). All indications are that only one MSG gene is expressed per P. carinii organism. Control of P. carinii MSG expression involves a single-copy locus called the expression site, or ES (Table 1). The majority of organisms in P. carinii populations appear to be haploid (Wyder, M. A., Rasch, E. M., and Kaneshiro, E. S. 1998). Because only one MSG gene can occupy the expression site at a time, the expression site system restricts transcription of the family to one gene per organism (Wada, M. et al. 1995; Edman, J. C. et al. 1996; Sunkin, S. M. and Stringer, J. R. 1996; Sunkin, S. M. and Stringer, J. R. 1997; Keely, S. P., Cushion, M. T., and Stringer, J. R. 2003). A large number of different MSG genes have been observed at the P. carinii expression site, suggesting that MSG genes can be moved to this locus by DNA recombination (Sunkin, S. M. and Stringer, J. R. 1997; Keely, S. P., Cushion, M. T., and Stringer, J. R. 2003). The mechanism of recombination is not known, but the possibility of a site-specific recombinase has been raised by the presence of a 28 basepair sequence called the Conserved Recombination Junction Element (CRJE) (Table 1). There is a copy of the CRJE at the junction between the expression site and adjacent MSG coding region. In addition to this copy, every MSG gene not at the expression site begins with a copy of the CRJE. Thus it is possible that recombination between the expression site and MSG genes at other loci is mediated by interactions between two copies of the CRJE. Whatever the mechanism, switching the MSG gene at the expression site discontinues expression of the previous resident MSG gene and activates expression of the new expression-site-linked MSG gene. Studies with antibodies that bind to few isoforms of MSG have shown that the protein encoded by the MSG gene at the expression-site locus is present in or on P. carinii cells (Schaffzin, J. K. and Stringer, J. R. 2004). Other studies using antibodies detected antigen variation between clusters of P. carinii in a rat lung (Angus, C. W. et al. 1996). Hence it would appear that P. carinii populations in the lung develop antigenic variation by switching the MSG gene that is at the expression site.

The P. carinii expression site encodes most of the 410 basepair Upstream Conserved Sequence (UCS) (Table 1), so named because it was found at the five prime ends of mRNAs encoding diverse MSGs (Wada, M. et al. 1995; Edman, J. C. et al. 1996; Sunkin, S. M. and Stringer, J. R. 1996; Sunkin, S. M. and Stringer, J. R. 1997; Sunkin, S. M. et al. 1998; Stringer, J. R. and Keely, S. P. 2001; Kutty, G., Ma, L., and Kovacs, J. A. 2001; Keely, S. P., Cushion, M. T., and Stringer, J. R. 2003; Stringer, J. R. 2003; Schaffzin, J. K. and Stringer, J. R. 2004; Stringer, J. R. 2005). The last 28 basepairs of the UCS are encoded by the CRJE. The UCS serves as the translation initiation site for production of an MSG precursor peptide (Sunkin, S. M. et al. 1998). The peptide encoded by the UCS appears to serve to send a precursor of mature MSG into the secretory pathway for transport to the cell surface (Sunkin, S. M. et al. 1998). The MSG on the surface lacks the amino acids encoded by the UCS, suggesting that these are removed by proteolysis in the endoplasmic reticulum (Sunkin, S. M. et al. 1998).

Given the similarities between species in the genus Pneumocystis, P. murina would be expected to contain a family of MSG genes that are expressed via an expression-site mechanism. However, this species might also be expected to have its own properties in this regard. Therefore, structural analysis of the expression-site locus of P. murina was necessary. Previously, a sequence that is approximately 62% identical to the P. carinii UCS was described at the 5’ end of a P. murina mRNA encoding an MSG (Haidaris, C. G. et al. 1998). Part of this sequence was shown to map to one chromosome of P. murina, suggesting that there is a single expression-site locus in this species (Haidaris, C. G. et al. 1998). The data presented herein characterize the locus encoding the P. murina UCS and provide more information about the MSG gene family in this species. These data show that the UCS locus of P. murina is present once per haploid genome and that a population of P. murina organisms can contain different MSG genes attached to the UCS locus. Between the genomic sequence encoding the UCS and the attached MSG gene, there is a sequence resembling the 28 bp CRJE of P. carinii, but sequence conservation extends substantially beyond the 28 bp CRJE-like sequence to encompass a block nearly five times longer than the CRJE of P. carinii. Multiple MSG P. murina genes were detected by hybridization, PCR and cloning. All but one chromosome contained MSG-related sequences. Head-to-tail tandem arrays of P. murina MSG genes were detected by PCR but no evidence of other tandem arrangements emerged. Sequence analysis detected 26 different MSG genes, which is only about one third as many as the number of MSG genes in P. carinii. Quantitative real-time PCR supported the hypothesis that the P. murina MSG gene family is smaller than the P. carinii MSG gene family.

Table 2 Sequences and positions of PCR primers.

Figure 1 Map of the P. murina expression-site locus

Real-time PCR was performed with a Cepheid thermocycler (Sunnyvale, CA) under the following conditions: 95°C for 15 seconds, annealing temperature for 15 seconds, and 72°C for 15 seconds. The optics were set to detect SYBR Green fluorescence. The concentrations of deoxyribonucleotide triphosphates were the same as those described above. The reactions contained Taq polymerase (1 unit per 25µl) (Promega USA, Madison WI), 5 mM MgCl₂, 1:1 of a 1:10,000 dilution of SYBR Green (BioWhitaker Molecular Applications) and 20 ng of each primer. Approximately 3 nanograms of P. murina genomic DNA was added to each reaction. The annealing temperatures used for primer pairs were as follows: 50°C for p2.1/p2a, p2/p3a and p3.1/p6.1, 55°C for p3/p6.1 and p4/p6.1. The performance of each primer pair was assessed by linear regression analysis of amplification kinetics of reactions containing known amounts of plasmid carrying the amplification target. For each experiment, cycle threshold (CT) was plotted versus amount of plasmid DNA. For example, Figure 5A shows a scatter plot of data from multiple trials performed with two primer pairs, p3/p6.1 and p4/p6.1. The figure shows that the CT values correlated (r² = 0.99) with the amount of plasmid over several logs of plasmid DNA concentration (Table 4). The amplification efficiency ranged from 71 to 82% (Table 4).

Figure 5 Quantitative real-time PCR of the MSG expression site

Table 4 Information about real-time PCR

To generate the MSG probe, primers p6 and p7 were utilized to produce an amplicon from approximately 3 nanograms of genomic DNA from P. murina. PCR was performed under the following conditions: 95°C for 3 minutes, then 40 cycles of incubation at 95°C for 10 seconds, 45°C 30 seconds, and 72°C for 30 seconds, then 72°C for 5 minutes, storage at 4°C. Reaction volumes were 25 microliters containing 100 mM each of dATP, dCTP, dGTP, dTTP, 1 U of Tfl polymerase (Epicenter, Madison, WI), 1.5 mM MgCl₂, and 20 ng each of primer. Amplification with primers p6 and p7 was expected to yield a 325 bp product, but a product closer to 1 kb in size was obtained. Sequencing showed that the amplicon ended at a point 716 downstream of the expected p7 priming site. Alignment of the 1 kb sequence to the P. murina MSG cDNA sequence listed under accession number AF043102 showed that the 1 kb amplicon was 94% identical to a P. murina MSG gene and identified a site where primer p7 matched 19 of24 bases. These data explained the source of the 1 kb amplicon and established that it was suitable for use as hybridization probe. The entire plasmid was made radioactive using a Prime-it II Random Primer Labeling Kit (Stratagene, La Jolla, CA), denatured with heat, and incubated with the CHEF blot in Rapid-Hybe (Amersham) at 50°C for 18 hours. After hybridization, membranes were washed three times at 50°C in 2X SSC (0.3M NaCl, 0.03M NaCitrate) and 0.1% sodium dodecyl sulfate (SDS). Bound radioactive probe was detected by autoradiography.

To generate the kex1 probe, approximately 3 nanograms of genomic DNA from P. murina were subjected to PCR amplification with primer KEX-upper (5’:GGGATAAATCTTGGAAGGAAA:3’) and primer KEX-lower (5’:GAACCCGAATATGTAGAAGCA: 3’) which bind nucleotide positions 979 to 999 and 1158 to 1178, respectively, of the sequence listed under accession no. AF093132. PCR was performed under the following conditions: 95°C for 3 minutes, then 40 cycles of incubation at 95°C for 10 seconds, 55°C 30 seconds, and 72°C for 30 seconds, then 72°C for 5 minutes, storage at 4°C. Reaction volumes were 25 microliters containing 100 mM each of dATP, dCTP, dGTP, dTTP, 1 U of Tfl polymerase (Epicenter, Madison, WI), 1.5 mM MgCl₂, and 20 ng each of primer. The amplicon was cloned into TOPO 4.0 (Invitrogen, Carlsbad, CA) and the identity of the insert in the plasmid was confirmed by sequencing. A radioactively labeled DNA probe was prepared as described above. Hybridization was performed in a solution containing 6XSSC (0.9M NaCl, 0.09M sodium citrate) 5X Denhardt's reagent (water containing 1% (weight:volume) Ficoll (Type 400) (Pharmacia), Polyvinylpyrolidone (Sigma) and BSA (Fraction V) (Sigma)) 0.5% SDS, and 100 ug of sheared herring sperm DNA per ml at 65°C for 18 hours(Cushion, M. T. et al. 1993). Unbound probe was removed by 3 washes with 2XSSC, 0.1% SDS at 60°C. Bound probe was detected by autoradiography as described (Cushion, M. T. et al. 1993).

Figure 2 Sequence at the P. murina expression-site locus

Comparison of amplified genomic DNA sequence to the sequence at the beginning of the previously described P. murina transcript identified a 227 bp segment present only in the genomic sequence. This segment began and ended with canonical splice donor and acceptor sites, GT and TAG, respectively (Figure 2), suggesting that the expression-site locus of P. murina contains an intron, as is the case in other species of Pneumocystis (Wada, M. et al. 1995; Sunkin, S. M. and Stringer, J. R. 1996; Schaffzin, J. K. and Stringer, J. R. 2000; Kutty, G., Ma, L., and Kovacs, J. A. 2001). Putative splice donor and acceptor sequences were similar to those of UCS introns from three other Pneumocystis species and to the consensus sequences for 19 different intron-containing genes (Table 3).

Table 3 Comparison of splice donor and splice acceptor regions in four Pneumocystis species.

Removal of the intron from the genomic UCS sequence generated an open reading frame encoding a 130 amino acid polypeptide 40–45% identical to the UCS-encoded polypeptides of P. carinii and P. wakefieldiae. Computer analysis of these three predicted peptides suggested that each of the three UCS-encoded peptides can function to send a nascent MSG polypeptide into the endoplasmic reticulum, consistent with results from previous experiments, where the P. carinii UCS-encoded peptide was shown to function as a signal peptide in insect cells (Sunkin, S. M. et al. 1998).

To confirm that the 227 bp segment functions as an intron, 8 additional UCS-MSG cDNA sequences were produced (data not shown). These data confirmed that splicing occurs. The splice donor site lies between bases 97 and 98 when base 1 is the A residue in the AUG codon that presumably initiates translation (Figure 2). This splice site is exactly the same distance from the AUG codon as it is in P. carinii, P. jirovecii and P. wakefieldiae.

The P. murina genomic UCS sequence matched that of P. carinii at 50% of positions. The coding regions were more similar (58% identity) than the introns (45% identity) (Figure 3). The 227 bp P. murina intron is longer than the P. carinii intron, which varies between 150 and 162 bp in length, but sequences present at the beginning and end of the P. murina intron were similar to those in the P. carinii intron (Figure 3). The P. murina intron is approximately the same length as the intron in the UCS locus of P. wakefieldiae (220 bp), and these two introns were approximately 50% identical (alignment not shown).

Figure 3 Comparison of expression site loci in P. murina and P. carinii

Nearly 80% of the bases in the P. murina intron were either adenine (A) or thymidine (T), and the intron contained numerous A-rich simple repeats. For example, there were fifteen tracts comprised of consecutive adenine residues (ranging from 3 to 9 bases in length). Altogether, the intron contained twenty-three mononucleotide tracts, averaging 4.5 ± 1.7 (S.E.) nucleotides per tract. These tracts comprised forty-six percent of the intron. Mononucleotide tracts tend to be polymorphic within a species (Jonsson, A. B., Nyberg, G., and Normark, S. 1991). Therefore the intron of the P. murina UCS locus might have utility as a marker for population genetics studies.

Figure 4 (lane 4) shows that only one P. murina chromosome, approximately 440 Kb in size, hybridized strongly to the exon1-plus-intron probe, suggesting that the genome of P. murina contains a single copy of UCS exon 1 and intron. Later, quantitative PCR experiments, described below, confirmed the single-copy hypothesis.

Figure 4 Hybridization mapped the sequence encoding UCS to a single P. murina chromosome

By contrast with the results obtained with the UCS exon1-intron probe, a probe that contained a 1 kb portion of a P. murina MSG gene hybridized to 12 chromosomes (Figure 4, lane 2). Some P. murina chromosomes produced stronger MSG hybridization signals than others, suggesting variation in either the number or sequences of the MSG gene family members. Hybridization conditions were nonstringent to allow probes to bind to MSG targets that were up to 20% different from the probe sequence. While conditions were not highly stringent, they were stringent enough to prevent indiscriminate probe binding. There was no signal from the 540 kb chromosome and none from the compression zone (compare lanes 1 and 2). In addition, the P. murina MSG probe did not hybridize to P. carinii chromosomes (data not shown).

The lack of MSG signal from the 540 kb chromosome was unexpected. To confirm that this result was not an artifact, the blot was hybridized to a P. murina kex1 probe because this probe had been reported to map to this chromosome (Lee, L. H. et al. 2000). The kex1 probe hybridized strongly to the 540 kb chromosome (lane 3), showing that the lack of signal when the MSG probe was applied was not due to an artifact of the experiment such as lack of transfer of this chromosome from the gel to the membrane.

Six polymerase chain reactions were performed, each with one of three UCS-specific primers (p1, p2, or p3) paired with either p7 or p8 (Figure 1A). Twenty-eight cloned amplicons were sequenced and ten different MSG sequences were obtained. All of the MSG sequences contained an open reading frame continuous with the spliced version of the expression-site sequence. The average variation between MSG sequences was 8% (range: 1–13%).

To characterize the upstream border of the P. murina CRJE, it was necessary to analyze the regions upstream of MSG genes that are not at the expression site. PCR presented itself as a possible tool for this purpose because MSG genes are arranged as head-to-tail tandem repeats in other species of Pneumocystis (Stringer, S. L. et al. 1991; Stringer, S. L. et al. 1993; Sunkin, S. M., Stringer, S. L., and Stringer, J. R. 1994; Garbe, T. R. and Stringer, J. R. 1994; Wada, M. and Nakamura, Y. 1994; Schaffzin, J. K., Garbe, T. R., and Stringer, J. R. 1999). This arrangement allows one to amplify the regions between MSG genes using an upstream primer that binds to a site conserved at the three prime ends of MSG genes, and a downstream primer that binds to a site conserved at the five prime ends of MSG genes.

Primer p9 (Table 2) binds to a sequence expected to be conserved at the 3-prime ends of P. murina MSG genes because this part of P. carinii MSG genes is highly conserved (Figure 1A). Primer p5 binds just upstream of the sequence that resembles the P. carinii CRJE (Figure 1A). This primer pair produced amplicons containing approximately 500 bp (not shown). Six cloned amplicons were sequenced, which showed that the procedure amplified at least 6 different tandem MSG gene pairs (data not shown). Similar data were obtained using a different upstream primer, primer p10, which binds in the 3’ untranslated regions of MSG genes (Figure 1A). In this case, five cloned amplicons were sequenced and five different sequences found (data not shown).

Comparison of the 11 amplified sequences to the sequence of the expression-site locus showed that the first 23 bp immediately upstream of the p5 primer site were identical to those at the expression site (Figure 6). Beyond these 23 bp, however, the sequences in the amplicons diverged from the expression site sequence. The divergence point is adjacent to the last nucleotide at the binding site for the primer p3.1 (Figure 6), which amplified the expression site with single-copy kinetics (Figure 5B). These data mapped the upstream border of the CRJE at the nucleotide level.

Figure 6 Comparison of expression-site to sequences (ES) upstream of MSG genes not at the expression site

PCR using either primer p5 alone or primer p9 alone failed to produce amplicons, suggesting that head-to-head and tail-to-tail tandem MSG genes either do not exist in the P. murina genome, or are too far apart to be detected by PCR as performed.

Genomic DNA between primers p4 and p8 was amplified and 56 cloned amplicons were sequenced. These 56 sequences were aligned to each other and to the 28 sequences derived from amplification mediated by primers p1, p2, and p4 combined with primers p7 or p8 (described above). Figure 7 shows that there were 26 distinct sequences. The average pairwise distance between P. murina MSG gene sequences was 7% (range: 1–16%). The hypothesis that the gene family contains 26 members fit well with the results of real-time PCR experiments on the CRJE described above (Figure 5B).

Figure 7 Neighbor-joining tree of MSG sequences

Alignment also showed that the sequences downstream of primer p3.1 were nearly invariant for 132 bp (Figure 6), suggesting that the P. murina CRJE is 132 bp long.

While the P. murina CRJE was strongly conserved, this element was not invariant. For example, four cloned copies of this sequence contained three tandem copies of the 12 bp sequence 5’:CAAGCAGCAGGA: 3’, which was present only twice in the other 80 copies sequenced (Figure 6). Repeat expansion can occur during PCR, but expansion is much less frequent than contraction (Shinde, Deepali et al. 2003). Therefore, it would appear that the predominant CRJE allele has two copies of the 12 bp sequence, and that the 3-copy allele is a natural variant rather than a PCR artifact. Variation of this sort has been seen in the expression sites of other species of Pneumocystis (Sunkin, S. M. and Stringer, J. R. 1996; Ma, L. et al. 2002). Transition mutations were also observed within the P. murina CRJE, but these were very rare, generally occurring in only one sequence out of 84. PCR error can cause variation at this frequency (Tindall, K. R. and Kunkel, T. A. 1988).

Figure 8 Comparison of expression sites in four Pneumocystis species

The expression sites of P. jirovecii and P. wakefieldiae have not been as thoroughly studied as those in P. murina and P. carinii, but current data suggest that these two species also are each distinct in this respect. In P. jirovecii, DNA hybridization data have indicated that nearly all of the DNA encoding the UCS is located solely at the expression-site (Kutty, G., Ma, L., and Kovacs, J. A. 2001). To examine this issue at the sequence level, we aligned the sequence at the P. jirovecii expression site (Kutty, G., Ma, L., and Kovacs, J. A. 2001) to the sequences upstream of three P. jirovecii MSG genes (Garbe, T. R. and Stringer, J. R. 1994). The first 18 nucleotides upstream of the sequence homologous to the 28 bp P. carinii CRJE were present at both the expression site and in regions upstream of MSG genes not at the expression site (data not shown). These data suggest that the CRJE in P. jirovecii may be as large as 46 bp. However, analysis of regions upstream of additional MSG genes may show that the CRJE in P. jirovecii is smaller than it now appears. In any event, the P. jirovecii CRJE appears to be substantially smaller than the P. murina CRJE. The expression-site locus of P. wakefieldiae appears to be quite different from those in the other three species. In P. wakefieldiae, more than half of the UCS is encoded by DNA that is repeated in the genome (Schaffzin, J. K. and Stringer, J. R. 2000).

Comparison of the UCS locus and attached MSG genes in four species of Pneumocystis showed that the structure of the expression site is conserved across species, but no two species are identical in this regard. For example, each species has an intron in the unique DNA encoding the UCS, and this intron is exactly the same distance from the beginning of the UCS open reading frame in all four species. However, the introns of different species are different in length and sequence. A second example is the CRJE. All four species have a sequence nearly identical to the 28 bp CRJE of P. carinii. However, the P. carinii CRJE appears to be rather small compared to the 132 bp CRJE of P. murina. The CRJE of P. murina is also much larger than the CRJEs in P. jirovecii. The P. wakefieldiae CRJE appears to be larger still, at over 300 bp in length (Schaffzin, J. K. and Stringer, J. R. 2000).

CRJE size differences may have functional consequences. It is possible that the CRJE serves to mediate recombination events that switch the MSG sequence that is at the expression site. If such events were to be due to the action of a site-specific recombinase, then it would seem that each species may have evolved its own such enzyme. A very large CRJE, such that in P. wakefieldiae, might allow homologous recombination between CRJE copies to contribute to variation at the expression site. Sequences 300 bp in length support efficient homologous recombination in Saccharomyces cerevisiae (Jinks-Robertson, S., Michelitch, M., and Ramcharan, S. 1993).

While the mechanism that generates change at the expression site is not completely understood, such a mechanism appears to be at work in P. murina because multiple MSG genes were observed at the expression-site locus. This situation is similar to that observed in other Pneumocystis species (Wada, M. et al. 1995; Sunkin, S. M. and Stringer, J. R. 1996; Sunkin, S. M. and Stringer, J. R. 1997; Schaffzin, J. K. and Stringer, J. R. 2000; Kutty, G., Ma, L., and Kovacs, J. A. 2001; Keely, S. P., Cushion, M. T., and Stringer, J. R. 2003). Studying the switching process has been difficult because Pneumocystis does not proliferate much in culture. Progress in understanding the switching process would be facilitated by developing assays utilizing animals infected with a small number of Pneumocystis expressing one, known MSG gene. Under these conditions, organisms expressing a different MSG are expected to emerge as the population expands in the animal. The rate of appearance of these “switched” organisms would provide an indication of the frequency of switching, and the structure of the expression-site linked MSG gene before and after switching would indicate the mechanism of switching. Studies on switching in P. carinii were initially promising, but have been hampered by the tendency of laboratory rats to be colonized by this fungus (Keely, S. P., Cushion, M. T., and Stringer, J. R. 2003). Such animals are difficult to use to study the frequency and mechanism of switching. By contrast, laboratory mice free of P. murina are routinely available. Therefore, it should be easier to acquire information on switching by studying P. murina.

The ability to switch MSG genes at the expression-site locus is a means to generate surface variability, which is probably an important feature of the Pneumocystis lifestyle. Programmed surface variation mediated by differential expression of gene families is common among microbes that must avoid the host immune response in order to survive (Borst, P. 2002; Recker, M. et al. 2004; Gatton, M. L. and Cheng, Q. 2004; Andrews, T. D. and Gojobori, T. 2004; Embers, M. E., Ramamoorthy, R., and Philipp, M. T. 2004; Liang, F. T. et al. 2004). Most examples of antigenic variation come from highly virulent microbes such as African trypanosomes, which use antigenic variation to maintain a large population in the blood stream despite a strong host immune response (Vickerman, K. 1978). By contrast, P. murina is unable to cause disease in immunocompetent mice, but is able to cause transient infections in them (Gigliotti, F., Harmsen, A. G., and Wright, T. W. 2003; An, C. L., Gigliotti, F., and Harmsen, A. G. 2003; Chabe, M. et al. 2004). Disease is seen only when the immune system is debilitated. The same seems to be true in other host species, including wild and laboratory animals, and humans (Mazars, E. et al. 1997; Bishop, R. et al. 1997; Nielsen, M. H. et al. 1998; Laakkonen, J. 1998; Weisbroth, S. H. et al. 1999; Denis, C. M. et al. 2000; Palmer, R. J. et al. 2000; Vargas, S. L. et al. 2001; Demanche, C. et al. 2001; Icenhour, C. R. et al. 2001; Durand-Joly, I. et al. 2003). Therefore, the function of the MSG system is probably not to overcome the immune response and cause disease. Instead, MSG variation might be used to prolong colonization by producing cells that are not recognized by the defense mechanisms deployed against cells with the most common antigenic phenotype.

The frequent-colonization hypothesis fits with the observation that each host species has its own kind of Pneumocystis, suggesting that the microbe and host have coevolved, and that each species of Pneumocystis depends on a particular host species for survival (Guillot, J. et al. 2001). This suggestion is supported by the fastidiousness of these microbes. They do not thrive in culture, and seem to be able to propagate continuously only in the lungs of the host in which they are found (Gigliotti, F. et al. 1993; Wakefield, A. E. et al. 1998; Dei-Cas, E. 2000; Durand-Joly, I. et al. 2002). It is reasonable to speculate that the P. murina MSG system serves to foster survival of relatively low numbers of P. murina in the lungs of immunocompetent mice, and that this survival is necessary to assure transmission to other mice.

This work was supported by grant: R01AI36701 from the National Institutes of Health.