3. Genetic ‘fingerprinting’ for clues to the pathogenesis of tuberculosis

472 TRANSACTIONS OF THE ROYAL SOCIETY OP

TROPICAL MEDICINEAKD HYGIENE(1992)86, 472-475

Aspects of tuberculosis in Africa. 3. Genetic ‘fingerprinting’ pathogenesis of tuberculosis

for clues to the

Peter Godfrey-Faussett* and Neil G. Stoker Bacterial Molecular Genetics Unit, Department of Clinical Sciences, London School of Hygiene and Tropical Medicine, Keppel Street, London, WClE 7HT, UK Abstract The recent discovery of a repetitive element within the DNA of Mycobacterium tuberculosis, which is present in variable numbers at different locations in separate strains of the organism, has led to the development of genetic ‘fingerprinting’ to distinguish between different isolates. Clusters of cases of tuberculosis have been identified in Europe and the USA in which the organisms cultured had identical ‘fingerprints’, confirming that transmission was occurring. Unrelated isolates generally have distinct ‘fingerprints’. In Africa, where transmission is more common than in Europe, there is less heterogeneity between isolates. We have typed 117 isolates of M. tuberculosis collected from continuing studies in Malawi and Kenya. Paired isolates from an individual patient produced matching ‘fingerprints’ in 22 of 25 cases. There were 18 isolates which had an identical matched pair from a separate patient; WChave not yet found any epidemiological link between these patients. These data show that there is sufficient heterogeneity amongst African isolates of M. tuberculosis to make studies of transmission feasible and to address questions of pathogenesis and epidemiology. Introduction The past decade has seen rapid advances in the field of molecular biology. The study-of mycobacterial genetics has been a major area of growth (MCFADDEN, 1990). In December 1991, the Royal Society of Tropical Medicine and Hvaiene met to discuss the future of trouical meditine. &estions were raised about the appropiiateness of these new technologies for research in the tropics. Despite the tenor of that meeting, we hope to show that it is possible to use rather basic molecular methods that do not require sophisticated laboratories in order to address questions that are both fundamental and unlikely to be answered in the near future by other means. Our current understanding of the chain of transmission of tuberculosis has been summarized by RODRIGUES & SMITH (1990). Following primary infection with tubercle bacilli, a proportion of people develop tuberculosis during the first one or 2 years but, in a substantial majority, the initial challenge is overcome without symptoms and the bacilli then lie dormant. Post-prican then result either from mary tuberculosis endogenous reactivation of the original infection following a waning of immunity, or by a further exogenous infection. The relative contribution of the 2 mechanisms is not known because the clinical result is identical, and most isolates of Mycobacterium tuberculosis cannot be distinguished from each other. PORTER & MCADAM (1992) have mentioned how inextricably human immunodeficiency virus (HIV) and tuberculosis are bound together in Africa? with up to 70% of new tuberculosis patients also being infected with HIV (ELLIOTT et al.? 1990). Most of the adult population in sub-Saharan Africa have already been infected with M. tuberculosis so that, with failing immunity induced by HIV infection, clinical disease may be reactivated. In HIV-infected drug users in New York, SELWYN et al. 119891 have shown that nrevious infection with M. tuberculosis (as shown by a reactive tuberculin skin test) is the major determinant of who develops tuberculosis and so reactivation was DroDosed as the nredominant mechanism. However, otherstudies in San Francisco (DALEY et al.. 1992) and in Italv (DI EERRI et al., 1989) have demonstrated that patients‘ immunosuppressed by HIV are also more susceptible to exogenous infection with M. tuberculosis. In sub-Saharan Africa the risk of infection from the environment is probably greater than in New York;, so it is possible that reinfection is as important as reactivation in determining who develops tuberculosis in that region. Patients infected with HIV and tuberculosis are also more likely to ‘relapse’ after treatment (NUNN et al., 1991b; A. M. Elliott, personal communication). How *Address for correspondence: Department of Medicine, University Teaching Hospital, P.O. Box 50110, Lusaka, Zambia.

certain are we that these relapses are not actually reinfections? The answers to these questions will have a bearing on issues of both public health and clinical management. Studies of chemoprophylaxis, either with isoniazid given for 6 months or more, or with shorter courses of rifampicin-containing regimens, have been made a priority by the World Health Organization. The aim is to prevent reactivation, but reinfection will be affected only during the time that antituberculous drugs are being taken. Do we need to use more intensive xand more expensive) regimens to eliminate tuberculous infection in HIV-infected patients, or should the resources be spent on reducina the risk of acauirine a further infection? A difficulty in investig&ng these issues has been the naucitv of methods for distineuishina between isolates of k. tuberculosis. Virulence inguinea-pigs and phage typina will distinguish onlv a few arouns with differing distributions in different regions ‘bf the world (GRAZE et al., 1978). Using more phages helps a litte (JONES, 1990) but reproducibility becomes a problem @ADO et al., 1975). Recently, ZAINUDDIN & DALE (1989) found a repetitive element within the deoxyribonucleic acid (DNA) of M. tuberculosis, which is present in variable numbers and at different locations in separate strains of the organism. It was subsequently sequenced and shown to be an insertion element (MCADAM et al., 1990). It had also been found and sequenced by another group (THIERRY et al., 1990) and so is now called either IS986 or IS6110 (or IS987 in bacillus Calmette-Guerin (HERMANS et al., 1991). It is present in all isolates of M. tuberculosis so far studied in between 1 and 20 copies. The insertion element is a short niece of DNA whose only function is its own transposition, although it is likely- to disrupt the function of any gene into which it happens to be inserted. Because the copies of this insertion element are scattered more or less randomly through the genome it is possible to generate ‘fingerprint’ patterns (HERMANS et al., 1990b). Like all bacteria, M. tuberculosis is thought to have a single circular chromosome. Following digestion with a restriction endonuclease, which will cut the DNA onlv at a specific motif, fragments of varying size will be generated. Because thev are charged, these fragments can be separated by gel electrophor&is-and then transferred to a nylon membrane. A short piece of DNA complementary to the insertion element, labelled with horseradish peroxidase (Amersham, UK; ECL@ gene detection system) is used as a probe. This will bind to the fragments of DNA on the membrane that contain the insertion element and will then catalyse the oxidation of luminol which generates light that can be detected by exposing photographic film to the membrane in an X-ray cassette, to produce an

473 autophotograph. This system avoids the safety and storageproblems of radioactive probes. The fact that the changesin the ‘fingerprint’ of a strain were not too rapid to be useful was demonstrated by HERMANSet al. (1990b), who showed stability of the patterns during animal passageand repeated subculture. Individual isolates from clusters of casesalso had identical ‘fingerprints’. They were also able to demonstrate sample-to-sample contamination in a mycobacteria diagnostic laboratory. Stability of the ‘fingerprints’ has also been demonstrated in 5 patients with relapsed tuberculosis from whom isolates from the original episode were still stored (OTAL et al, 1991). In all 5, the isolates from both episodes of diseasewere identical, confirming that in these cases relapse rather than reinfection had occurred. These experiments all used isolates from patients living in western Europe or the USA, where the prevalence of tuberculosis is low. ‘Fingerprints’ are almost always very different, presumably because the original infections were acquired in many different regions and local transmission is limited by effective treatment and contact-tracing. The first ‘fingerprints’ from African M. tubercdosis isolates showed less heterogeneity than those from Dutch isolates (VAN SOOLINGEN et al:, 1991)! presumably becausetransmission rates were higher within a particular region. Our studies have been aiming to determine whether DNA ‘finerprinting’ of M. tuberculosiswill be useful in addressing questions about the epidemiology and pathogenesisof tuberculosis in Africa.

1991a, 1991b). Twenty-one isolates, from 13 patients, came from Dr C. Gilks’s and his colleagues’ study of morbidity among a cohort of women sex-workers in the Pumwani suburb of Nairobi. Isolates from each studv were identified only by number, and no information about their origin was available until after ‘fingerprints’ had been made and compared. Isolates from the samepatient, either sequentially or from different sites of disease,were included. Two pairs of isolates were available from Nairobi from natients who had relansed and for whom the original isolate was stored. Isolates were grown on Lowenstein-Jensen slopes. Cells were scraned off the surface and resusnended in 1 ml TE buffer-(10 mM Tris HCl, pH 8.0, 1 *mM ethylenediaminetetraacetic acid). ‘Fingerprints’ were produced using the methods described by HERMANSand colleagues (1990a, 1990b) with a probe, 198 base pairs long, directed against the IS986/6110 prepared by amplifying a segment of DNA using the polymerase chain reaction (GODFREY-FAUSSETT et al., 1991). DNA was extracted and digested on 2 separate occasions from each isolate and some of the digests were subjected to electrophoresis on 2 separate gels to check the consistency of the ‘fingerprints’. Each membrane was exposed several times with exposures varying from one minute to severalhours. ‘Fingerprints’ of the same isolate were compared visually between different autophotographs and were shown to produce a consistent pattern. The patterns from different isolates were then compared with each other.

Methods Sample preparation and fingerprinting

Data analvsis

One hundred and seventeen isolates of M. tuberculosis were provided from 3 continuing studies in Malawi and Kenya. Fifty-six isolates, from 50 patients, came from a community-based study of vaccination against leprosy and tuberculosis in the Karonga district of Malawi (FINE & PONNIGHAUS, 1988). Forty isolates, from 29 patients, camefrom a prospective study of the interaction between HIV and tuberculosis in Nairobi city (NUNN et al.,

SimilarTty coefficients (SAB) were calculated for isolates in the Nairobi and Pumwani eroun bv the formula SAB=[number of bands shared between-A and B]/[(number of bands in A) + (number of bands in B) - (number of bands shared between A and B)]. This is a simplification of the formula used by SCHMID and colleagues (1990) and ignores differences in intensity of hybridization. Since PvuII cuts at one end of the insertion sequence, there will be only a single copy of the target on a given restric-

Fig. 1.DNA ‘fingerprints’ofM. tuberculosis fromKaronga,Malawi.Lanesarelabelledwith thereference numberof eachisolate.Markerlanesarelab&d M.

474 1613 ' 1970 1938 495

I I

?--..-I

I I

I

I-’

I

360 362 489 1585 1941

2087

;

0.1 I

0.2 I

0.3 I

0.4 I

015

0.6 I

Oi7

0.8 I

0.9 I

I '*O

SAB

Fig. 2. Dendrogram of similarity coefficient (SAB) for 31 isolates ofM. tuberculosis from Nairobi city. Identical paired isolates from individual patients are included only once. The two discordant pairs are both included. The position of the vertical bar joining any two isolates or groups of isolates represents the similarity between them. Isolates grouped together with similarity values greater than 0.25 are defined arbitrarily as clusters (A-F).

tion fragment. Assuming that different restriction fragments can be distinguished on the gel, each band should be an all or none signal. Identical isolates have an SAB value of 1.0, whereas isolates with no bands in common have a value of 0. Dendrograms were constructed manually using the method described by SCHMID and colleagues (1990). The factors affecting transposition are not yet established. Even if it is non-random, isolates with a high similarity value are likely to have evolved more recently from a common ancestor. If transposition is random the dendrogram can be transformed to a phylogenetic tree since D (a measure of phylogenetic distance) would then be directly related to the similarity value (D=-log, SAB). Results An autophotograph showing representative fingernrints from the Malawi aroun is shown in Fig. 1. There are identical pairs (e.g.: 909/910; 9131914; 294612765; 328613837) and pairs which are related but not identical (e.g., 9101912). A marker lane (M) allows comparison of isolates run on different gels. Sixty-two isolates had an identical match with one or more other isolates. Forty-two of these were pairs from 21 patients from whom 2 isolates were collected during the same episode of disease. Another pair came from a patient in whom the disease relapsed and whose original isolate was still stored. There were 3 other patients from whom we had a pair of isolates that had discordant ‘tingerprints’. Two of these patients had isolates taken from different sites or on different days during a single episode of tuberculosis, whereas for the third patient the isolates

were from a stored original culture and a new relapse. Amongst the 18 isolates with an identical match from separate patients, 7 came from the Pumwani patients, 9 from the Nairobi city patients, and 2 from the Malawi patients. We have not yet found any epidemiological link between individuals with matching patterns. An example of a dendrogram constructed using the Nairobi city isolates is shown in Fig. 2. The similarity between any 2 isolates or groups of isolates is represented by the position along the bottom axis of the vertical bar joining them. An arbitrary value of 0.25 has been used to define several clusters, labelled A to F on the Figure. Three isolates are not attached to any cluster. Within any one cluster, isolates are more closely related, and hence likely to have evolved more recently from a common ancestor, than isolates from any other cluster. Discussion The high proportion (22125) of paired isolates from individual patients that have identical ‘fingerprints’ shows both that the technique is consistent and that most patients have a unique isolate of M. tuberculosis at any one time. Two exceptions to this observation have been found in HIV-infected patients, in whom discordant patterns were found in paired isolates from a single episode of disease in a particular individual. In both cases the ‘fingerprint’ patterns were completely unrelated to each other. This raises the possibility of multiple strains coexisting within a single patient with active tuberculosis. The discordant patterns from an individual patient who relapsed suggest reinfection rather than true relapse. However, we have not yet been able to recover sufficient

475 pairs of cultures from relapse and original episode to address the relative contribution of relapse or reinfection in an African context, which is important for implementation of effective public health programmes. Constructing dendrograms allows an objective assessment of the amount of heterogeneity in a group of isolates. Within the Nairobi city group, there are several loose clusters. Whether these reflect important differences between the isolates or the patients from whom they were cultured remains to be seen. These early studies of DNA ‘fingerprinting’ of M. tuberculosis isolated in Africa confirm that the technique is a useful new tool with which to address questions of pathogenesis and epidemiology. Although less heterogeneous than their European counterparts, there are sufficient differences in ‘fingerprints’ to make studies of transmission feasible. We are currently limited by suitable pattern recognition and matching software. The dendrogram presented was calculated manually with computer assistance. Each band in each lane has to be compared with each other, so the number of comparisons rises rapidly with increasing numbers of isolates. Further development of computer software should allow rapid comparisons to be made between multiple strains and much larger phylogenetic trees to be constructed. It is too early to confirm our impression that these methods will separate isolates from different locations. If this is indeed uossible. we should be able to study the spread of tuberculosis on a global scale and to address old questions about pathogenesis anew. At this stage we veer rapidly into speculation but at least with these new technologies many of these hypotheses will be able to be tested. Acknowledgements

We thank the microbiologists and clinicians who provided the cultures of M. tuberculosis used in this study: J. Ponnighaus, S. Oxborrow, P. Fine and P. Jenkins for the Malawi isolates; W. Githui and P. Nunn for the Nairobi city isolates; R. Brindle! B. Batchelor, J. Paul, C. Gilks and S. Ojoo for the Pumwam lsolates. We would also like to acknowledge the encouragement of Dr D. Koech, Director of the Kenya Medical Research Institute, and Professor K. McAdam, Head of the Department of Clinical Sciences of the London School of Hygiene and Tropical Medicine. P.G.-F. is funded by a Medical Research Council training fellowship; N.G.S. is partially funded by the Overseas Development Administration, UK. References Daley, C. L., Small, I’., Schecter, G., Schoolnik, G., McAdam, R., Jacobs, W. & Hopewell, P. (1992). An outbreak of tuberculosis with accelerated progression among persons infected with human immunodeficiency virus. New EnglandJournal of Medicine, 326,231-235. Di Perri, G., Danzi, M., De Checchi, G., Pizzighella, S., Solbiati, M., Bassetti, D., Cruciani, M., Luzzati, R., Malena, M., Mazzi, R. & Concia, E. (1989). Nosocomial epidemic of active tuberculosis among HIV-infected patients. Lancet, ii, 1502-1504. Elliott, A. M., Luo, N., Tembo, G., Halwiindi, B., Steenbergen, G., Machiels, L., Pobee, J., Nunn, P., Hayes, R. & McAdam,, K. P. W. J. (1990). Impact of HIV on tuberculosis in Zambia: a cross sectional study. British Medical Journal, 301,412-415. Fine, P. E. M. & Ponnighaus, J. (1988). Leprosy in Malawi. 2. Background, design and prospects of the Karonga Prevention Trial, a leprosy vaccine trial in northern Malawi. Transactions of the Royal Society of Tropical Medicine and Hygiene, 82, 8 lO817. __.

Godfrey-Faussett, I’., Khoo, S., Wilkins, E. G. L. & Stoker, N. G. (1991). Tuberculous pericarditis confirmed by DNA amplification. Lancer, 337, 176177. Grange, J. M., Aber, V., Allen, B., Mitchison, D. & Goren, M. (1978). The correlation of bacteriophage types of Mycobacte-

rium tuberculosis with guinea-pig virulence and in-vitro indicators of virulence. Journal of General Microbiology, 108, l-7. Hermans, I’. W. M., Schuitema, A. R. J., Van Soolingen, D., Verstvnen, C. I’. H. I., Bik, E. M., Thole, I. E. R., Kolk, A. H. J.-& van Embde&‘J. D: A. (1990a). S’p&ific detection of Mycobacterium tuberculosis complex strains by polymerase chain reaction. Journal of Clinical Microbiology, 28, 120C 1213. Hermans. I’. W. M.. van Soolinaen. D.. Dale. I. W.. Schitema. A. R. j., McAdam, R. A., C&y, D: & vii Emdden, J. D: A. (1990b). Insertion element IS986 from Mycobacterium tuberculosis: a useful tool for diagnosis and epidemiology of tuberculosis.Journal of ClinicalMicrobiology, 28,2052-2058. Hermans, P. W. M., van Soolingen, D., Bik, E. M., de Haas, I’., Dale, I. W. & van Embden. 1. D. A. (1991). The insertioi elen&t IS987 from Mycobacterium bo&s BeG is located in a hot-spot integration region for insertion elements in Mycobacterium tuberculosis complex strains. Infection and Immunity, 59,2695-2705. Jones, W. D., jr (1990). Geographic distribution of phage types among cultures of Mycobacterium tuberculosis. American Review of Respiratory Disease, 142, 1000-1003. McAdam, R. A., Hermans, I’. W. M., van Soolingen, D., Zainuddin, Z., Catty, D., van Embden, J. D. A. & Dale, J. W. (1990). Characterrzation of a Mycobacterium tuberculosis insertion sequence belonging to the IS3 family. Molecular Microbiology, 4,1607-1613. McFadden, J. J., editor (1990). Molecular Biology of the Mycobacteria. London: Academic Press. Nunn, P. P., Kibuga, D., Gathua, S., Brindle, R., Imalingat, A., Wasunna, K., Lucas, S., Gilks, C., Omwega, M., Were, J. & McAdam, K. (1991a). Cutaneous hypersensitivity reacsons due to thiac&azo&e in HIV-l &opositive patients treated for tuberculosis. Lancet, 337,627-630. Nunn, P. l’., Porter, J., Githui, W. & Odhiambo, J. (1991b). Treating tuberculosis in HIV positive Africans. Lancet, 338, 1141. Otal, I., Martin, C., Vincent-Levy-Frebault, V., Thierry, D. & Gicquel, B. (1991). Restriction fragment length polymorphism analysis using IS6200 as an epidemiological marker in tuberculosis.Journal of ClinicalMicrobiology, 29, 1252-1254. Porter, J. D. H. & McAdam, K. I’. W. J. (1992). Aspects of tuberculosis in Africa. 1. Tuberculosis in Africa in the AIDS era-the role of chemoprophylaxis. Transactions of the Royal Societv of Trobical Medicine and Hygiene. 86. Rado, T. A., B&es, J., Engel, H., Kankiwicz, E., Murohashi, T., Mizuguchi, Y. & Sula, L. (1975). World Health Organization studies on bacteriouhage tvoing of mvcobacteria. Subdivision of the species kyc~bacieri~m tuberculosis. American Review of Respiratory Disease, 111,459-468. Rodrigues, L. C. & Smith, I’. G. (1990). Tuberculosis in developing countries and methods for its control. Transactions of the Rqal Society of Tropical Medicine and Hygiene, 84, 739744

Schmid, J., Voss, E. & Soll, D. R. (1990). Computer assisted methods for assessing strain relatedness in Candida albicans by fingerprinting with the moderately repetitive sequence Ca3. Journal of ClinicalMicrobiology, 28, 1236-1243. Selwyn, P. A., Hartel, D., Lewis, V. A., Schoenbaum, E. E., Vermund, S. H., Klein, R. S., Walker, A. T. & Fnedland, G. H. (1989). A prospective study of the risk of tuberculosis among intravenous drug users with HIV infection. New EnglandJournal of Medicine, 320,545-550. Thierrv. D.. Cave. M.. Eisenbach. K.. Crawford. 1.. Bates. I.. Gicc$el, i. & duesion, J. (1996). Is6110, an Iii&e elen&n; of Mycobacterium tuberculosis comvlex. Nucleic Acids Research, 18. i88. Van Soolingen, D., Hermans, I’. W. M., de Haas, I’., Soil, D. & van Embden, J. D. A. (1991). Occurrence and stabilitv of insertion sequences in hiycobhcterium tuberculosis complex strains: evaluation of an insertion sequence-dependent DNA polymorphism as a tool in the epidemiology of tuberculosis. Journal of ClinicalMicrobiology, 29,2578-2586. Zainuddin, Z. F. & Dale, J. W. (1989). Polymorphic repetitive DNA sequences in Mycobacterium tuberculosis detected with a gene probe from a Mycobacterium fortuitum plasmid. Journal of GeneralMicrobiology, 135,2347-2355.