Genetic Diversity of Pneumocystis carinii Isolated from Human Immunodeficiency Virus-Positive Patients in Turin, Italy

β-tub.All samples were positive for the expected 309-bp fragment of the β-tub locus. Sequence polymorphisms were all detected at positions previously reported to be variable (3, 5), except for a novel variant at nucleotide position 96, where a G-to-A change was detected (Table 1). On the basis of the variations, we divided the samples into three types: type 1, which is identical to the prototypic sequence reported by Edlind et al. (3); type 3, previously described by Hauser et al. (5); and type 4, a novel sporadic variation reported here for the first time, found in fewer than three specimens and therefore not considered a distinct type according to Lu et al. (14). Two coinfections were observed, as A and G were both found at position 96 in one sample (suggesting coinfection with types 3 and 4) and at position 282 in another (suggesting coinfection with types 1 and 3) (Table1).

26S.Sequence analysis of all samples for the 26S intron (a 426-bp fragment) revealed four types differing at several nucleotide positions (Table 2). We noted new polymorphisms: an A-to-G change at position 34 in one sample and a 2-bp insertion (GG) at position 356 in another. Sequences identical to the 26S prototype reported by Liu and Leibowitz (13), corresponding to Hauser type 1, are designated type α in Table 2. Types β and δ are novel types, and γ is a sporadic variation. Greek letters have been adopted for classification to avoid confusion with terms (Arabic numerals and Roman letters) already in use for other loci.

mt26S.In all samples we amplified mt26S (a 340-bp fragment). All sequences differed from the prototype sequence reported by Sinclair et al. (19) by a G-to-A change at position 288. On the basis of polymorphisms at nucleotides 85 and 248, we distinguished four sequence types (Table 3), all of them previously described (5, 9, 10, 11). In one specimen, two different nucleotides (C and T) were present at position 248, suggesting coinfection with types 2 and 3.

ITS-1.Sequence comparison revealed that ITS-1 sequences (a 204-bp fragment) can be classified into six types (Table4). The analysis was based on polymorphisms at nine different positions, including the poly(T) tract at positions 54 to 62. It has been reported that the number of T's may vary when one is resequencing the same sample (21). However, since we did not encounter this problem in two sequencing attempts, we used the poly(T) tract in typing. We found one novel sporadic variation, referred to as B5, and five previously identified types (5, 12) (Table 4). None of the sequences were identical to the prototype sequence reported by Lu et al. (14); types A3, B, and B2 have been described previously by Hauser et al. (5) using the same terms, and by Lee et al., who defined them as A, E, and N, respectively (12). Types B1 and B3 have been described previously by Hauser et al. (5) and by Latouche et al. (9) using the same terms. We also observed four samples with mixed types. For two of these, listed in Table 6 as belonging to patients 159 and 48, we identified one of the two ITS-1 types involved, while for the other two patients, 45 and 158, only PCR cloning could permit the discrimination and identification of the strains.

ITS-2.The ITS-2 region (a 327-bp fragment) was amplified in all samples. On the basis of polymorphisms observed in eight previously reported positions, we found six ITS-2 sequence types (Table5). Types a₁, a₂, a₄, and c₁ have been described by Latouche et al., and a₁ has been designated the prototype sequence by Lu et al. (9, 14). Types a₅ and b₂are novel sporadic variations. We also detected three coinfections, but we were unable to resolve any of the six ITS-2 types involved (Table6).

The combined results obtained from the analysis of the five loci led to a highly accurate characterization of P. carinii strains (Table 6). Of the 25 strains, complete molecular typing was possible for 18, among which 15 different combination types were demonstrated. Thirteen out of 19 patients were infected with a single strain, whereas 6 were coinfected, as indicated by demonstration of mixed genotypes in at least one of the studied loci. The coinfection percentage (31%) in our series is in agreement with that observed by other authors (2, 5); however, we do not know if it represents an underestimate. Indeed, the sensitivity of any molecular typing method depends on the number of microorganisms present in the specimen, which at the moment cannot be easily assessed by conventional methods.

We found two new distinct types and five novel sporadic variations, on the basis of new polymorphic positions identified in our study and/or of new variations at nucleotide positions previously reported as variable.

From the data shown in Table 6 it can be inferred that the most significant association genes for strain typing is ITS-1 and mt26S, whereas the 26S locus consented the identification of new variants but was not useful for strain discrimination. However, it can be supposed that this locus could be significant in a larger series of cases or as part of a different set of indicator genes.

Our results demonstrate great genetic diversity among the isolates. Only two combination types (B3/a₄/β/8/3 and B2/a₁/α/2/3) were present in more than one patient. Comparison of our sequence data with those reported by other authors revealed the presence of combination types similar to those circulating in other parts of Italy and in other countries. To date, few authors have examined the mt26S locus; therefore, the only possible comparison in these cases has been with the ITS loci. Moreover, there is at present no general agreement on strain classification terms: the strain that we designate B2/a₁/2 is identical to the strain designated B2/a₁/3 by Latouche et al. (10). Likewise, our strain B/a₂/2 is identical to Latouche strain B6/a₂/3 (9). Strain B1/a₂/8, present in one of the six coinfection samples, is identical to strain B1/a₂/1 found by Latouche et al. (10). However, taking into account the ITS type, our B2/a₁strains correspond to the N/e strains described by Lee (12) and to the B2/a₁ strains described by Margutti et al. (15) and by Tsolaki et al. (20), even if these authors did not consider the poly(T) tract in typing. The B1/a₃ strain described by Margutti et al. and Tsolaki et al. corresponds to our B/a₂ or B1/a₂. Our B/a₂ corresponds to the E/g strain described by Lee. Finally, the ITS combination A3/c₁corresponds to A/1 in the Lee classification.

We think it would be highly desirable to standardize the terms used to define sequence types and genetic loci in order to permit easier comparison of the circulating strains. In this paper we used the Hauser terms to define sequence types of β-tub, mt26S, and ITS-1; we used Latouche terms for ITS-2; and we introduced Greek letters for 26S to avoid confusion with Arabic numerals and Roman letters already in use.