TEXT

Top Abstract Text References

Mycobacterium avium, long recognized as a primary pathogen of birds, behaves as an opportunistic human pathogen. In immunocompetent patients, the organism causes pulmonary infections and cervical lymphadenitis and, occasionally, soft-tissue infections. However, in a high proportion (25 to 50%) of patients with AIDS, M. avium causes severe disseminated infections (reviewed in reference 3).

M. avium can be isolated from environmental, animal, and human sources. Tap water is regarded as the main reservoir of the organism, at least for human infections (13), but the epidemiology of M. avium infections has not been completely defined. Analysis of restriction fragment length polymorphism (RFLP), based on the insertion sequence (IS) IS1245, a 1,414-bp element belonging to the IS256 IS family, has been proposed as a suitable technique for typing of M. avium isolates for epidemiological studies (1, 5, 6, 8-10) and also as a way to provide more insight into the taxonomy and evolutionary divergence of the M. avium complex (1, 6).

M. avium strains isolated from humans and typed by the IS1245-based RFLP technique almost invariably show highly polymorphic, multibanded IS1245 RFLP patterns that share a high degree of similarity with a significant proportion of isolates from pigs (1, 6, 9). In contrast, the IS1245 banding patterns of isolates from a wide variety of bird species are characterized by a very low IS number (1, 6, 9); the "bird pattern" M. avium strains are rarely found among human or pig isolates (1, 9).

The purpose of this study was to characterize, by the IS1245-based RFLP technique, 63 M. avium strains isolated from human infections in distinct geographic areas of Italy. The isolates were from 52 human immunodeficiency virus (HIV)-positive patients, of whom 47 had monoclonal infections and 5 had polyclonal infections, and from 2 HIV-negative patients. Thirty-nine isolates were from the Pisa area, and 24 isolates were from four other geographic areas (Rome, Milan, Florence, and Ancona) of Italy. Forty-one isolates were from blood, 11 were from respiratory specimens, 3 were from urine, 2 were from stool, and 1 was from cerebrospinal fluid. Five isolates were from unknown types of specimens. All isolates were passed onto Middlebrook medium plates (Becton Dickinson), and at least two colonies from each isolate, selected on the basis of different colony morphology (if any), were grown in liquid Middlebrook medium and typed by a previously described IS1245-based RFLP assay (5) that also takes into consideration the 85% homologous insertion element IS1311 (10). The assay generates fingerprints with a number of IS1245- and IS1311-specific bands that is the sum of the copy numbers of the two insertion elements (5). Briefly, 4.5 µg of genomic DNA, prepared from approximately 1.5-ml bacterial cultures, was digested overnight at 37°C with 10 U of the restriction endonuclease NruI (Amersham) in a final volume of 20 µl. The DNA fragments generated were electrophoretically separated on a 0.8% agarose gel and blotted onto a nylon filter (Hybond N-plus membrane; Amersham). A mixture of a PvuII-digested supercoiled DNA ladder (Gibco BRL) and HaeIII-digested X174 DNA (Boehringer Mannheim) was run in two lanes of each gel and served as molecular size markers ranging from 16.2 to 0.603 kb. Filters were hybridized by addition of approximately 200 ng of a peroxidase-labeled IS1245 probe, prepared from DNA of an M. avium isolate by PCR using oligonucleotides P1 5'GCCGCCGAAACGATCTAC and P2 5'AGGTGGCGTCGAGGAAGAC as primers, as previously described (5, 6), and 25 to 40 ng of each molecular size marker probe. Hybridization was then detected on autoradiographic films by the enhanced-chemiluminescence gene detection system (Hyperfilm-ECL; Amersham). The M. avium RFLP patterns were scanned with an Epson GT 8000 scanner at 200 dots/in., and the fingerprints were compared by 3.1 GelCompar software (Applied Maths). The Dice coefficients of similarity of all pairwise comparisons of patterns were calculated, and a dendrogram of pattern relatedness among the strains was constructed by using UPGMA clustering in accordance with a previously described algorithm (12). In general, the IS1245-based RFLP patterns were polymorphic and complex, as reported in other studies (1, 6, 9, 10). Assuming that those isolates yielding banding patterns with similarity coefficients of greater than 85% were essentially identical or highly related, we detected 41 distinct IS1245 fingerprints (Fig. 1, left side). Ten banding patterns were shared by more than one isolate (i.e., cluster); the 10 clusters, indicated in Fig. 1 as a through j, included 31 (49%) of the 63 isolates. Clusters a and d were the largest, as they included six and seven isolates, respectively. Clusters b, c, e, f, g, h, i, and j each included two or three isolates. A reference strain of bird origin, i.e., M. avium ATCC 35712, occurred in cluster j together with two human isolates sharing the bird-type RFLP pattern; both isolates showed the glycopeptidolipid antigen of serotype 3, as assessed by thin-layer chromatography (2, 11). This finding reinforces the view that infections with bird-type M. avium strains do occur in humans, although rarely (1, 9). Moreover, as shown in Table 1, where the fingerprints of the isolates occurring in clusters are matched with their geographic origins, clusters a, b, c, e, f, g, i, and j each contained isolates from the same geographic area, suggesting the existence of a common source of infection for patients. For example, the isolates in cluster i were from three AIDS patients hospitalized in one hospital in Rome during the same period, which suggests the possibility of nosocomial transmission of M. avium infection. Clusters d and h contained, respectively, seven and three identical or highly related isolates from different geographic areas. Isolates in cluster d, in particular, derived from as many as four distinct areas.

View larger version (71K): [in this window] [in a new window]   FIG. 1.   IS1245-based RFLP analysis of Italian M. avium isolates. The left side shows the IS1245 banding patterns ordered by similarity; the corresponding dendrogram is shown at the far left. Band positions in each lane have been normalized so that the band positions of all strains are mutually comparable. The scale at the top depicts similarity coefficients. The numbers at the bottom indicate sizes (in kilobase pairs) of standard DNA fragments. Italic letters a through j indicate clusters of identical or highly related clones. The arrowheads indicate isolates from non-AIDS patients. The right side shows the similarity matrix of banding patterns of the M. avium isolates. Similarity coefficients of 55 to 95% are shown by five different gray tones at 10% intervals. The diagonal is formed by 100% similarity coefficients of corresponding strains.

To visualize more objectively the genotype relatedness among all of the isolates, a similarity matrix was generated. This matrix shows the degree of relatedness of each IS1245 banding pattern with any other in the collection. In Fig. 1 (right side), the "families" of related IS1245 banding patterns are shown in groupings of gray-shaded values. In general, all of the groupings with related fingerprints contained few strains, with the exception of one large family, including 25 (40%) of the 63 isolates with similarity values of greater than 55%. This family, which comprises clusters c, d, e, f, and g, includes isolates derived from the five different geographic areas. The genetic relatedness among numerous strains isolated from AIDS patients in different areas may support the possibility of the existence of an IS1245 RFLP pattern(s) associated with HIV-induced immunodeficiency. Conflicting evidence has been reported on the genetic characters of M. avium strains that cause disseminated disease in AIDS patients. It has been suggested that M. avium strains that infect AIDS patients form a distinct and genetically conserved group of highly similar isolates within the M. avium complex (7). This hypothesis is also reinforced by the quantitative demonstration of a high degree of relatedness of the AIDS-associated isolates, compared with isolates from non-HIV-infected individuals (4). In contrast, several studies have shown that AIDS patients are infected by unrelated, highly variable strains, at least according to RFLP analysis (1, 6, 9, 10). Our results indicate that these two conditions may coexist. AIDS patients may be infected by different, genetically unrelated strains, but at the same time, the possibility cannot be ruled out that one or more strains may be selected in or adapted to a particular environmental niche that facilitates diffusion to AIDS patients, with the consequent emergence of a predominant strain in these patients.

In conclusion, the results of RFLP typing of Italian M. avium isolates described in this paper confirm the marked polymorphism of most isolates from AIDS patients but also raise the question of the possible definition of the predominant genomic fingerprints of virulent clones for these patients.