Molecular Fingerprinting of Mycobacterium tuberculosis Isolates Obtained in Havana, Cuba, by IS6110 Restriction Fragment Length Polymorphism Analysis and by the Double-Repetitive-Element PCR Method

Abstract

Mycobacterium tuberculosis sputum isolates from 38 patients, obtained in the first 6 months of 1997 in Havana, Cuba, were characterized by IS6110 restriction fragment length polymorphism (RFLP) analysis and the double-repetitive-element PCR (DRE-PCR) method. Among 41 strains from 38 patients, 24 and 25 unique patterns, and 5 and 4 cluster patterns, were found by the RFLP and DRE-PCR methods, respectively. Patients within two of these clusters were found to be epidemiologically related, while no relation was observed in patients in the other clusters. The DRE-PCR method is rapid, and it was as discriminating as IS6110 RFLP analysis in identifying an epidemiological association. Its simplicity makes the technique accessible for subtyping of M. tuberculosis strains in laboratories not equipped to perform RFLP analysis.

The World Health Organization-International Union against Tuberculosis and Lung Disease Global Surveillance Project on Drug Resistance in Tuberculosis reported that in 1996 Cuba, with a population of 11,005,866 inhabitants, had a tuberculosis incidence rate of 14.34 per 100,000. In that year, 835 sputum smear-positive cases were reported, representing 65% of all pulmonary cases and a case detection rate of 84%. The human immunodeficiency virus coinfection rate was 1.3% (25).

Cuba has a tuberculosis control program, established in 1962, and short-course chemotherapy is used in 100% of cases. An important consideration for the evaluation of tuberculosis control programs and for infection control in nosocomial and other institutional settings housing tuberculosis patients is the improved understanding of the transmission of the disease due to accurate epidemiological studies. Recent characterizations of biological markers for typing strains have greatly facilitated and improved the study of the epidemiology of infectious diseases (20). Molecular techniques are used to track specific strains of pathogens and to determine more precisely the distribution of infectious diseases in populations, providing opportunities for more-effective interventions.

In the past, evidence for recent active transmission of tuberculosis was based on outbreaks, on the appearance of strains showing the same drug susceptibility pattern, on conversion of tuberculin test results after contact with a tuberculosis patient, or on phenotypic methods such as phagotyping, serotyping, biotyping, and electrophoretic analysis of enzymes (7). The utilization of phenotypic markers requires 4 to 8 weeks of growth of Mycobacterium tuberculosis, and such studies have been limited due to the lack of sufficiently discriminating polymorphic markers able to distinguish the various bacilli infecting unrelated individuals (17). Until recently, the only method available for typing of M. tuberculosis strains was phage typing (2). However, this method has been used by few laboratories, due to the difficulty of the technique and the fact that only a limited number of mycobacteriophage types were recognized.

Recent advances in molecular biology have led to the use of molecular techniques, which are based on the principle that there are phenotypic or genotypic differences from strain to strain but not within a given strain (20). Genotypic fingerprinting utilizes slight differences in the total chromosome that are generally not related to phenotypic differences.

Restriction fragment length polymorphism (RFLP) analysis using the repetitive DNA element IS6110 in M. tuberculosis has been a powerful tool for confirming the results of standard epidemiological investigations (22). Pulsed-field gel electrophoresis (21) and different methods of strain typing by PCR, including double-repetitive-element PCR (DRE-PCR) (6), randomly amplified polymorphic DNA analysis (10), mixed-linker PCR (8), spoligotyping (12), and others, have also been reported. Many examples of epidemiologic studies illustrate the utility of these molecular techniques: the study of tuberculosis outbreaks in institutional settings, such as prisons, hospitals, and shelters (4, 5, 18), and in community settings, as in Switzerland (7), San Francisco (13), and New York (1, 6); studies in specific geographic areas to trace the migration of strains around the world (22, 26); studies to demonstrate occupational exposure among health care personnel (19); and studies of laboratory contamination (14). In most of these studies, IS6110-based RFLP analysis has been used.

However, laboratory methods for the identification of biological markers need to be simplified in order to increase their accessibility for clinical laboratories in both developed and developing countries. The DRE-PCR method is based on PCR amplification of segments located between two repetitive sequences from the M. tuberculosis genome: IS6110 and the polymorphic GC-rich repetitive sequence (6). This is a rapid subtyping method that can be performed with the primary growth of M. tuberculosis, eliminating the need to subculture. In the original paper (6) it has been shown to be as discriminative as the IS6110 RFLP method. In this study we compared the DRE-PCR method with the IS6110 RFLP method with regard to their abilities to predict epidemiological relationships among clinical strains of M. tuberculosis isolated in Havana, Cuba, in 1997.

Between January and July of 1997, 41 strains were isolated from sputum samples from 38 patients at the Instituto Pedro Kourí (IPK) in Havana. They represent consecutively obtained strains that have been isolated in the area, including nine isolates from an active surveillance study performed in correctional institutions in Havana. The samples were examined by microscopy after Ziehl-Neelsen staining and cultured in Lowenstein-Jensen medium. Identification and susceptibility testing were performed according to methods described in references 9 and 3, respectively. No resistance to rifampin, isoniazid, streptomycin, pyrazinamide, or ethambutol was observed. The patients were tested for human immunodeficiency virus infection, and all results were negative.

IS6110 RFLP analysis.

DNA fingerprinting was carried out by the standardized protocol of van Embden et al. (23). Briefly, genomic DNA was isolated and digested with PvuII (Gibco/BRL, Rockville, Md.), electrophoresed on an 0.8% agarose gel, and vacuum blotted onto nylon membranes (Hybond N+; Amersham, Little Chalfont, Buckinghamshire, United Kingdom). A mixture of lambda-HindIII and φX174-HaeIII (Gibco/BRL) was used as an external marker. No internal standards were used. Southern blotting was performed with a PCR-generated probe amplified from M. tuberculosis H37Rv DNA with the primers INS-1 and INS-2 (23). Labeling and detection were performed with the ECL Direct System (Amersham). Comparison of fingerprints was performed visually. Two or more isolates with identical RFLP patterns or with patterns differing in only one band were considered to belong to a cluster and possibly to be epidemiologically related. The subgroup of clustered strains was subjected to a second RFLP analysis, and the patterns were analyzed with GelCompare software (Applied Maths, Kortrijk, Belgium) (Fig. 1).

FIG. 1.

RFLP patterns from the strains showing similar or identical patterns by visual inspection. The RFLP patterns were analyzed and normalized by using GelCompare software. Cluster I comprises strains 9, 12, 16, 28, 38, 14, and 40; cluster II, strains 3, 7, 8, and 20; cluster III, strains 31 and 24; cluster IV, strains 1 and 13; and cluster V, strains 37 and 36. Mt, M. tuberculosis MT14323 reference strain. Strain 22 was not included in cluster III, despite pattern similarity, because it differs in more than one band.

Two to sixteen copies of IS6110 were found in each of the Cuban M. tuberculosis isolates, and 29 different patterns were observed. Of these, 17 strains were included in five cluster patterns (Fig. 1). Cluster I comprised seven strains from seven young incarcerated males, aged 18 to 33 years, whose tuberculosis cases were detected in an active surveillance investigation. Their RFLP patterns were identical and showed 9 bands in all (Fig. 1), except for one strain (strain 14) that had 10 bands. Cluster II showed three very similar patterns of 11 bands: a shift in one band in strain 3 (pattern IIa) was observed in strains 7 and 20 (IIb), and a shift in a different band in strain 3 pattern was observed in strain 8 (IIc) (Fig. 1). The patients, a female of 28 and three males of 40, 50, and 70 years, respectively, had no recognized epidemiological relationships and were diagnosed in different regions of Havana. Cluster III showed an identical nine-band pattern in strains from two young males imprisoned in the same institution as the patients in cluster I. The RFLP patterns of clusters I and III were unrelated. Another identical pattern was observed in strains from two young men who apparently were not epidemiologically related; these two strains constituted Cluster IV. Cluster V comprised two strains obtained from apparently unrelated patients, a young male and an old female diagnosed in different districts of Havana.

Identical or highly related patterns were found much more frequently in younger patients than in older patients. The mean age of patients with cluster pattern strains was 36.3 ± 16.2 years, while the mean age of patients with unrelated strains was 56.5 ± 20.6 years. Men were found to be infected by related or identical strains more often than women (15 men and 2 women). This observation was not related to the number of cluster pattern strains obtained from imprisoned male patients in the surveillance study.

DRE-PCR.

The procedure reported by Friedman et al. (6) was slightly modified (11a). Briefly, a loopful of each culture on a Lowenstein-Jensen slant was diluted in 1 ml of distilled water and boiled for 10 min with no further DNA purification. The PCR amplification mixture contained 20 mM Tris (pH 8.8)–50 mM KCl (1× reaction buffer; Gibco/BRL), 2.5 mM MgCl₂ (Gibco), 200 μM each deoxynucleoside triphosphate (Gibco), 6% dimethyl sulfoxide, 50 pmol of each of the four primers, and 1 U of Taq polymerase (Gibco). The primers and their sequences are described in reference 6. Ten microliters of DNA solution was used in the reaction. The PCR mixture was subjected to denaturation at 95°C for 10 min, followed by 30 cycles of denaturation at 94°C for 1 min, primer annealing at 56°C for 2 min, and primer extension at 72°C for 3 min. The amplification products were analyzed by electrophoresis in 2% agarose gels stained with ethidium bromide and visualized under UV light (Fig. 2). The modifications introduced were the inclusion of 6% dimethyl sulfoxide in the reaction mixture and the extension step of 3 min.

FIG. 2.

DRE-PCR patterns of strains from the 38 patients in this study. Numbers correspond to the sample numbers listed in Table 1. M, 1-kb DNA ladder (Gibco).

Thirty different DRE-PCR patterns were obtained, and four clusters were observed. Cluster I comprised the strains of the seven young imprisoned male patients. One strain showed a single-band difference (the same strain that showed a band difference by RFLP analysis). Cluster II was a pattern observed in four strains, one of which showed a one-band difference. Cluster III contained two strains, one showing an extra band. Cluster IV comprised two strains.

The patterns generated by the DRE-PCR method were compared to the IS6110 RFLP patterns (Table 1). With the RFLP method, 29 distinct banding patterns were observed among 41 isolates from 38 patients. Five of these were cluster patterns that included 17 patients (45% of the patients). The DRE-PCR method produced 30 different patterns among 38 unique isolates. Four were cluster patterns that included 15 patients (39%). Concordance between the two methods was observed in 36 strains. The two discordant results corresponded to two strains that had a cluster pattern with six distinct bands by RFLP and nonclustered patterns with two and four bands by DRE-PCR. Data obtained from these patients suggested that they were not epidemiologically related.

TABLE 1.

Patient characteristics, including RFLP and DRE-PCR patterns

Sample no.a	RFLP cluster	DRE-PCR cluster	Sexb	Age (yrs)	Health unit/areac
9	I	I	M	33	HNR/HE
12	I	I	M	23	HNR/HE
16	I	I	M	18	HNR/HE
28	I	I	M	24	HNR/HE
38	I	I	M	24	HNR/HE
40	I	I	M	24	HNR/HE
14	Ia	Ia	M	24	HNR/HE
3	IIa	II	F	28	Policlínico/HE
7	IIb	II	M	40	Hospital/A. Naranjo
20	IIb	II	M	70	Policlínico/Playa
8	IIc	IIa	M	50	Policlínico/Boyeros
24	III	III	M	29	HNR/HE
31	III	IIIa	M	48	HNR/HE
1	IV	IV	M	34	Policlínico/HV
13	IV	IV	M	33	Policlínico/Regla
36	V		F	74	Hospital/CH
37	V		M	41	Hospital/Boyeros
10 and 6			F	84	Policlínico/Marianao
34 and 4			M	60	Policlínico/Cerro
39 and 35			M	39	Policlínico/SM
2			M	74	Policlínico/Boyeros
5			M	84	Policlínico/Cotorro
11			M	49	Policlínico/Boyeros
15			M	76	Policlínico/10 Octubre
17			M	33	Hospital/Guanabacoa
18			M	64	Policlínico/Cotorro
19			M	56	Hospital/Marianao
21			F	17	Policlínico/A. Naranjo
22			M	25	Hospital/CH
23			M	80	Hospital/10 Octubre
25			F	50	Hospital
26			M	52	Policlínico/Plaza
27			M	91	Policlínico/CH
29			F	49	Hospital/A. Naranjo
30			M	72	Policlínico/Playa
32			M	32	Hospital
33			M	48	Hospital/A. Naranjo
41			M	53	Policlínico/Naranjo

^aWhere two numbers are given together, two samples were taken from the same patient. 

^bM, male; F, female. 

^cHNR, Hospital Nacional de Reclusos; HE, Habana del Este; HV, Habana Vieja; CH, Centro Habana; SM, San Miguel del Padrón. 

The epidemiological information about the patients was reevaluated in the light of the data obtained by the IS6110 RFLP and DRE-PCR methods. Clusters I and III corresponded to strains from patients who were institutionalized in the same building, and therefore epidemiological associations could be confirmed. Cluster II (four strains) included two strains with identical RFLP patterns (strains 7 and 20) and two with distinct but very similar patterns, differing in the position of a single band (strains 3 and 8). DRE-PCR results showed identical patterns in three strains (strains 3, 7, and 20), with a single band missing in the fourth (strain 8). The four patients were not epidemiologically related. The patient infected with strain 3 had been treated for lung tuberculosis previously, in 1996. Data from patients in clusters IV and V showed that they were not epidemiologically related. Considering that the two strains in cluster V were found to differ in only one band by RFLP analysis but had completely different DRE-PCR patterns, we concluded that DRE-PCR may be even more discriminating than IS6110 RFLP analysis for epidemiologic assessment.

Several strain-typing studies in community settings have been performed by the IS6110 RFLP method. In the canton of Berne, 45 of 163 patients (27.6%) showed clustered patterns—the largest group included drug addicts, homeless persons, and alcoholics (7)—and in Austria, only 2 of 31 patients (6.4%) showed the same RFLP pattern (24). In developing countries, community-based studies showed a higher proportion of clusters, suggesting that recent transmission of M. tuberculosis is more frequent. In Guadeloupe, 17 of 51 patients (33.4%) showed clustered patterns (15), and in Honduras, 21 of 84 patients (25%) were infected with cluster pattern strains (11). The present study in Havana found 17 of 38 patients (44.7%) infected with cluster pattern strains by RFLP analysis and 15 of 38 patients (39.4%) infected with cluster pattern strains by DRE-PCR. Active surveillance for tuberculosis had been performed in the prison where two clusters were found, including a large cluster comprising seven patients. The fact that the cluster pattern strains were isolated from young incarcerated patients suggests that these infections resulted from recent transmissions.

The RFLP and DRE-PCR methods identified the same clusters of tuberculosis patients who were clearly linked epidemiologically. While it is possible that the RFLP method identifies more strains with different patterns in a collection of M. tuberculosis isolates, in this study it did not offer any advantage over the DRE-PCR method in assessing an epidemiological situation. The RFLP method requires expensive laboratory equipment and is time-consuming. The amount of extracted DNA needed to perform the procedure requires that the primary culture have abundant growth. The entire procedure may take several days or even weeks to obtain the final results. In a recent publication, Sola et al. compared the discriminatory powers of DRE-PCR, spoligotyping, IS6110 RFLP analysis, and direct-repeat RFLP analysis in assessing epidemiological relatedness (16). They found that spoligotyping plus DRE-PCR could give the same information as could be obtained by IS6110 RFLP analysis. Spoligotyping requires less DNA and is less time-consuming than the RFLP method, but a specially prepared membrane has to be provided. The DRE-PCR method is clearly simpler, less expensive, and faster than RFLP analysis and spoligotyping. The modification of the DRE-PCR protocol introduced here resulted in the amplification of more bands and therefore improved the discriminatory power of this technique. This study showed that this modified strain-typing method, when used alone, alone is accessible and produces the same epidemiological information as does the RFLP method. This could be very useful in settings where the latter method cannot be used routinely.