Molecular epidemiology of bovine tuberculosis II. Applications of ...

Rev. sci. tech. Off. int. Epiz., 2000,19 (3), 689-701

Molecular epidemiology of bovine tuberculosis II. Applications of genotyping

P.A. Durr, R.S. Cli f ton-Hadley & R.G. Hewinson

Veterinary Laboratories Agency (Weybridge), New Haw, Addlestone, Surrey KT15 3NB, United Kingdom

Submitted for publication: 5 Apri l 2000 Accepted for publication: 10 May 2000

Summary The applications of genotyping of Mycobacterium bovis are reviewed. Published research to date has been conducted predominantly within the context of validating typing methods, and few studies have been specifically epidemiological. This is contrasted with the situation in human tuberculosis, where the application of restriction fragment length polymorphism typing using insertion sequence IS6110 has successfully led to insights into the epidemiology and molecular evolution of the pathogen. Based upon the medical experience, the adoption of an integrated approach which combines epidemiology and molecular biology is recommended for future studies. Accordingly, clear identification and explanation of type clustering should be possible, which should facilitate decisions related to disease control.

Keywords Genotyping - Molecular epidemiology - Mycobacterium bovis - Restriction endonuclease analysis - Restriction fragment length polymorphism - Spoligotyping - Tuberculosis.

Introduction Bovine tuberculosis has a complex epidemiology (39, 41 ) . Problematic features include a variable incubation time, interaction with immuno-suppressive diseases and environmental stresses, and a wide potential host range. An additional difficulty was that until recently, no practical method existed of undertaking a subdivision of disease outbreaks by strains or types, an approach which is proving invaluable in understanding the intricacies of infectious disease dynamics (37) . With advances in cell biology, this particular problem has now been overcome, and a number of techniques are currently available or in development. A review of those applicable to bovine tuberculosis has been presented by Durr et al. (22).

While the purpose of undertaking Mycobacterium bovis typing is epidemiological, few examples have been published of the use of typing for control of animal disease. This can be contrasted with typing in human medicine, where the method has been more widely and successfully used. In part, this reflects the relative importance of the diseases, as human

tuberculosis is currently one of the most serious diseases, with an estimated one-third of the population of the world infected and therefore at risk of developing the disease (7) . In contrast, bovine tuberculosis has been extensively controlled fay test and slaughter policies in countries with a previously high prevalence of infection (39) . However, a greater difficulty faced by Veterinary epidemiologists has been the technical problem of the lack of a typing method that combines good discrimination with ease of use and interpretation. For human tuberculosis, restriction fragment length polymorphism (RFLP) typing, using the insertion sequence (IS) 6110 probe, was quickly established as the standard methodology. Only recently have comparable techniques become available for bovine tuberculosis.

While bovine tuberculosis is currently a relatively minor disease problem in the developed world, the disease remains problematic in several countries, particularly Great Britain, Ireland and New Zealand (11) . Even in countries where the disease has been officially 'eradicated', such as the United States of America (USA), problems remain, due to relatively frequent, but sporadic outbreak (23). Furthermore, the

690 Rev. sci. tech. Off. int. Epiz., 19 (3)

disease is an important zoonosis, with a particular threat presented in those tropical countries where the disease remains endemic, and may represent a particular risk to people infected with human immunodeficiency virus (HIV) (19).

Principles of using genotyping The basic assumption underlying the practical use of genotyping is the clonal model resulting from asexual reproduction. This proposes that when an infectious bacterium is passed on to a second host, the bacterium retains the same genetic make-up. Consequently, if two individual hosts are found to have isolates with different genotypes, then infection has not spread from one host to the other, even if the two hosts have had close contact. Similarly, if two isolates are found in the same host, then two independent infection events have occurred.

The practical use of this model can be demonstrated by a hypothetical example (Fig. 1). Assume a farm previously free of bovine tuberculosis in which infection is detected. The epidemiological investigation shows that a cow had been previously introduced from a herd that also had a breakdown. However, tuberculosis had also been detected shortly before in a wildlife reservoir sharing part of the farm. The key epidemiological focus is to determine the source of infection. If the M. bovis isolated from the cattle on the farm all have the same genotype as that of the introduced animal, and this is different from that of the wildlife reservoir, then this provides strong evidence that the infection was introduced via the purchased animal (Fig. la ) . If the herd has a subsequent breakdown which is found to be the same M. bovis type as the wildlife reservoir, this supports the hypothesis of spread from the latter, in contrast to the alternative that the infection continued in the herd via a non-detected infected cow.

Although the assumption of clonality is central to the use of genotyping, this must be treated as a biological approximation. All bacteria undergo mutation of deoxyribonucleic acid (DNA) sequences, and if these mutations do not affect the functioning of essential enzymes or structural proteins, the daughter generation should survive. If the typing technique can detect this genetic change, then a 'new' clone is identified. Alternatively, if mutation occurs in part of the genome that the technique does not recognise, then the daughter generation will be classified as the same clone. An important consequence of this is that the typing technique used must be considered in relation to the rate of genetic change of the organism.

The problems posed by using a technique that is too sensitive to genetic change are illustrated by the hypothetical example in Figure lb . In this case, mutations occur frequently in the genomic area targeted by the typing technique. As a consequence, although the typing can help support the

hypothesis of infection originating in an introduced animal, the technique cannot add insight as to the 'path of transmission' of the infection within the herd. Similarly, the technique cannot provide any support to the competing hypotheses of the cause of the second breakdown (i.e. whether the breakdown was due to the wildlife reservoir or non-disclosed latent infection).

The problem posed by excessive sensitivity of the typing technique to genetic mutations can be contrasted to the alternative, where only major genetic shifts are detected. This results in insufficient discrimination, and similarly limits the utility of genotyping to answer the questions of source of infection and paths of transmission (Fig. l c ) . The important concept is that, to be useful, the typing technique must fall between the two extremes. Alternatively, this compromise position between excessive sensitivity to mutation and insufficient discrimination may be resolved by using more than one technique, provided that the techniques used target different areas of the genome.

The above discussion serves to illustrate the point that the fundamental role of typing is to assist in choosing between competing hypotheses that arise as part of a thorough epidemiological disease investigation. In the hypothetical example, before the benefit of typing could be realised, the two critical steps were tracing the origin of the introduced animal and sampling the wildlife reservoir for evidence of infection. Without collecting this epidemiological data, typing isolates obtained from cattle at the farm where breakdown occurred would be of little use. In this way, molecular epidemiology can be viewed as an extension of, but not a substitute for, 'traditional' epidemiology.

An important piece of information, which assists in the reconstruction of a plausible sequence of infectious pathways, is the time-period during which a detected animal might have harboured the disease. This does not present a problem for many infectious diseases, in which the incubation period and disease course are both well established and subject to little variation. However, in mammalian tuberculosis, these disease characteristics are both highly variable, and at present, no simple methods are available to determine the length of time a particular animal may have had the disease and been infectious to others (27). Given this lack of information, a reconstruction of the sequence of infectious events must depend upon other time-based information obtainable by routine surveillance. In the hypothetical example, this is demonstrated by the need to know the disease status of the wildlife reservoir before the index case was detected. If this information is not available, then questions arise as to the direction of infection (Fig. 2) . Accordingly, if the situation in wildlife was not investigated until after the infection had been detected in the cattle, an equally plausible scenario would be that infection spread from the cattle to the wildlife.

Hev. sci. tech. OU. int. Epiz.. 19 (3) 691

Time

x : implausible path

? : plausible, but not provable, path

• : mutation in the typing région

Fig. 1

Hypothetical example of Mycobacterium bovis infection involving

introduction of infection from both a bought in cow and from an

infected wildlife réservoir

In a), the typing technique possessed sufficient strain diversity, and species

stability enabled the correct path of transmission to be identified. In bjand

cj, no conclusion could be reached due to the technique being excessively

sensitive to genotypic lability, and insufficiently discriminatory, respectively.

Mutations are indicated by the red triangle

The example used above represents the most cornmonepidemiological problem faced by disease managers duringthe latter stages of a bovine tuberculosis eradicationcampaign. In such a low incidence situation, the relative rarityof the disease canjustify the considérable expense of a detailed'case' investigation. In contrast, a broader, population-basedapproach may be more applicable in an endémie setting(Fig. 3). This approach differs in being more researchorientated and directed at developing and exploringhypothèses. An important group of such population-basedstudies are those which are primarily exploratory, attemptingto define patterns in space and time. Thèse studies serve to

generate hypothèses, the validity of which can be tested byspécifie, targeted studies.

Few population-based studies hâve been undertaken inbovine tuberculosis epidemiology. However, an importantexample can be found in M. tuberculosis molecularepidemiology, which serves to illustrate the gênerai principlesof such studies. When a résurgence of human tuberculosiswas reported in North America and Western Europe in thelate 1980s, an important factor was to détermine the extent towhich cases represented récent transmission, as opposed tore-activation of donnant infection acquired years previously(1,48). The hypothesis used for thèse studies was that isolâtesfrom a group in which récent infection occurred should be'clustered' (i.e. identical), while those from re-activatedinfections would occur as single examples, as the latter wouldhâve been acquired over a longer period of time and space.Such analyses demonstrated that, contrary to the prevailingthéories, the majority of the new cases represented récentinfection. This had important implications for public diseasecontrol. Although subséquent analyses hâve shown that the'cluster equals active transmission' hypothesis is asimplification (10, 29), the conclusion remains validnevertheless.

Resuit 1Différent types found inwildlife réservoir and cattle =récent transmission

Resuit 2Same types in wildlife

no réservoir and cattle =>threepossible interprétations

Transmission from wildliferéservoir to cattle

Time

Transmission from cattle towildlife réservoir

Transmission from secondwildlife réservoir to cattleand main wildlife réservoir

JSfx: implausible path? : possible path

Fig. 2

The difficulty of interpreting the path of infection from cross-sectional

studies demonstrating identical types

While dissimilarity has only one interprétation (that transmission has not

occurred), similarity may hâve three alternative interprétations

692 Rev. sci. tech. Off. int. Epiz., 19 (3)

Source of infection?

Reconstruction of paths of

transmission

Developing hypotheses

Patterns of types in space ("geotyping")

Patterns of types in time (mutation

dynamics)

Types versus mammalian host, pathogenicity, etc.

Confirming/testing hypotheses

Evolution of Mycobacterium bovis

?:hypotheses yet to be proposed

Fig. 3 The t w o major types of molecular ep idemio logica l studies Case-based studies general ly aim to determine the source of infect ion, and w h e n combined w i t h detai ls about probable occurrence of in fect ion, may permit paths of t ransmission to be determined in a group. Populat ion-based studies aim to provide insight into the disease by exploring the patterns of type var iat ion and the processes wh ich drive these variat ions

Applications of Mycobacterium bovis typing

Wildlife reservoirs versus cattle-to-cattle transmission in causing breakdowns By far the most common use of genotyping has been to find evidence for the role of wildlife reservoirs in outbreaks amongst cattle herds. This has been the case in several countries, particularly New Zealand and the British Isles (Great Britain and Ireland), where tuberculosis in cattle has remained a persistent problem, despite longstanding campaigns of eradication (39). In both cases, the primary problem is considered to be the persistence of infection in wildlife reservoirs, the Australian possum (Trichosurus vulpecula) and the Eurasian badger (Meles meles), respectively. The most controversial issue is the relative importance of these wildlife reservoirs as compared to cattle-to-cattle transmission.

In New Zealand, restriction endonuclease analysis (REA) has been used to study aspects of this problem. Comparison of M. bovis isolates from infected cattle herds and brushtail possums trapped in the locality usually showed identical patterns (14). Although this was taken to support the hypothesis of transfer of infection from possums to cattle, this was an inference which was based upon epidemiological rather than typing data (15). Certain outbreaks in cattle were characterised by the same REA pattern, which indicated that only a small number of infected possums were responsible for transmitting the disease to cattle, probably as a consequence of a single heavily-infected possum traversing pasture on which the cattle were grazing. A more detailed study of the role of possums, cattle and other wildlife was conducted in the MacKenzie Basin of the South Island (21) . The area was free of tuberculosis prior to 1980, but in the subsequent years, the majority of cattle and deer herds in the area became infected with M. bovis. Typing of 125 isolates by REA and IS6110-RFLP revealed two principal groups, indicating that the infection was introduced from at least two distinct

Improved disease control

Case-based (individual herd) studies

Population-based (research) studies

Rev. sci. tech. Off. int. Epiz., 19 (3) 693

sources. While several hypotheses were presented as to the source of infection, no definitive conclusion could be drawn regarding paths of transmission.

In parts of Ireland, while the badger is still considered to be the predominant source of infection, a significant amount of inter-bovine spread occurs, associated in part with frequent movements of animals between farms (40). Molecular typing has been used to try to determine the relative contribution of both factors. A study in Ireland using REA, compared isolates recovered from cattle and badgers in breakdowns in several counties (16) . When an outbreak of tuberculosis occurred in cattle, the genotypes in infected badgers nearby were usually found to be the same. In a study in Northern Ireland, isolates from cattle, badgers and deer were typed using direct repeat (DR)-RFLP (46, 47 ) . Most of the badger types were the same as those previously found in cattle, but no conclusion could be drawn regarding the direction of transmission between the two species.

Observations of the association between types and geographical areas have been explored in England and Wales, using geographical information system (GIS) technology (12). Since 1996, the majority of isolates from herd breakdowns have been spoligotyped, and using the map references of the farm locations held in the national veterinary disease database, detailed mapping has been undertaken (Fig. 4 ) . Results at both a national and regional level reveal a distinct clustering of spoligotypes isolated from cattle, and often (but not always), these are the same as those isolated from badgers found nearby.

Perumaalla et al. investigated isolates recovered from cattle in Mexico and Texas using two RFLP techniques (IS6100 and DR probes) (42) . Based upon the high frequency of isolates possessing multiple copies of IS6120, a hypothesis involving wildlife reservoirs was presented. However, a subsequent investigation of potential wildlife sources failed to find any evidence to support this hypothesis (43) .

Transmission may potentially occur in the reverse direction, namely from farmed to wild animals. Whipple et al. investigated M. bovis tuberculosis in captive and free ranging deer and coyotes in an outbreak in Montana, USA, and found similar types using RFLP (IS6110 and polymorphic guanine and cytosine-rich repeat sequences [PGRS] probes) (55). Although this result could not conclusively determine the direction of transmission, in the context of the whole case investigation, infection was considered more likely to have passed from the farmed animals to the free-ranging ones (44) . The transmission of M. bovis between wild boars and cattle was suggested in a study in north-western Italy (45) , where spoligotyping and RFLP were used to show commonality of types between the host species. In this case, the inference was that transmission occurred from the cattle to the boar and not vice-versa, although this was not proven.

Associating types to geographical regions (geotyping') In many of the studies investigating wildlife reservoirs, the alternative source of infection is purchased animals or spread from infected neighbouring farms. The ability to detect and assign certain types to distinct geographical regions provides the advantage that if a particular type is isolated outside the area in which it predominates, then the investigation can focus upon the hypothesis of an introduced infection, and hence detection of the animal responsible for the introduction.

To date, no specific studies of this kind have been reported. Nevertheless, geographical clustering has been reported in a number of studies, for example, in the Republic of Ireland (16), Northern Ireland (46, 47) , New Zealand (14, 15), Argentina (25) and Great Britain (12).

The study by Clifton-Hadley et al. using a GIS (12), has been the most sophisticated demonstration of geotyping to date (Fig. 4 ) . In particular, one spoligotype (type '17') was found overwhelmingly in a confined geographical location in south-central England, while the other types were scattered throughout the country. The potential of these results for systematic disease control is currently being investigated (P.A. Durr, unpublished findings).

However, geographical clustering of types has not been invariably demonstrated. Perumaalla et al. found little geographical clustering of types in a study of isolates recovered from cattle in Mexico and Texas (42). This was thought to reflect the extensive movement of cattle within regions and between the two countries. A similar lack of geographical clustering was shown in a study using random amplified polymorphic DNA polymerase chain reaction (RAPD-PCR) to examine isolates from six geographical regions in Mexico (38).

Determining the source of Mycobacterium bovis in imported animals In countries free of bovine tuberculosis, most outbreaks are associated with imported animals. Detailed investigations are consequently undertaken to determine the source, so that lessons can be learnt to prevent repetition of the problem.

Bölske et al. used REA to undertake a thorough epidemiological investigation of an outbreak of bovine tuberculosis in several herds of farmed fallow deer (Dama dama) in Sweden (8) . Detailed tracing was able to reveal that all the infected deer had either been imported into Sweden from the United Kingdom (UK) four years previously, or had been in contact with these imported animals. The REA patterns were identical for eight isolates from five affected farms, which indicated a common source of infection. Furthermore, the patterns were distinct and only resembled others that were similarly derived from isolates obtained from

694 Rev. sci. teck Off. int. Epii, 19 (3)

100 km

Spoligotype

• 9

• 17

IM

A

0 100km

A

100 km

A

100 km

Fig.4Distribution of the most common Mycobacterium bovis spoligotypes in England and Wales for isolâtes from cattie (a and c) and badgers (6 and d),1996-1997Note the geographical similarity of types found between the two mammalian host species and the distinct clustering of types.Adapted from Clifton-Hadley étal. (12)

deer in the UK. A follow up study using IS61J0-RFLPconfirmed this conclusion (49).

In 1992, an outbreak of tuberculosis was observed amongcattie housed at three farms in the Netherlands (54). Tracingstudies showed that the probable source of infection was ayoung, diseased bull which was imported from Austria in1991 and housed at ail three of the farms. Using

IS6110-RFLP, ail the isolâtes from the infected bull and cattiewere demonstrated to be clonal, confirming the bull as thesource of infection.

Zumârraga et al. used spoligotyping to demonstrate thatisolâtes from South America could be differentiated fromthose obtained from Spain (56). If similar associations couldbe made between types and countries, systematic assignment

Rev. sci. tech. Off. int. Epiz., 19 (3) 695

of the source of infection in imported animals might be possible, thereby assisting epidemiological investigations when disease is detected in imported animals.

Paths of transmission of Mycobacterium bovis outbreaks in zoos and wildl i fe parks Mycobacterium bovis infection is a common disease of captive wild mammals, possibly associated with the stress and relative overcrowding of zoological parks (50) . Recently, typing has been utilised to establish paths of transmission between infected animals.

In Sweden, during an outbreak of M. bovis in a zoological garden involving tapirs and a gibbon, IS6110-RFLP was used to demonstrate that a single clone was involved (34). However, determination of the direction of transmission was not possible.

Thorel et al. described the use of typing in outbreaks of tuberculosis at two zoos in France (52) . In the first zoo, an outbreak of tuberculosis occurred in a group of baboons; in the second, mortality occurred in three leopards and a sea-lion. Using IS6110-RFLP, the second outbreak was demonstrated to be caused by a single clone; however, the strain in the baboons contained only a single copy of IS6110, and thus the involvement of a clonal isolate could only be inferred.

Van Soolingen et al. reported the use of IS6210-RFLP to investigate an outbreak of tuberculosis in species of oryx and gazelle from two wild parks in Saudi Arabia (54). Most of the oryxes were thought to be infected during transport in a container from one wild park to another in Saudi Arabia by contact with a single diseased animal. Although not all the isolates had the same banding pattern, the hypothesis of a single point source during transport was considered a likely explanation of much of the outbreak.

Diagnosis of tuberculosis in a colony of captive seals (Australian sea-lions and New Zealand fur seals) at a marine park in Western Australia in 1986 , led to a series of detailed studies using various typing methods (26). The REA showed that all the isolates from the six infected seals produced identical DNA patterns, indicating a common source of infection (17) . The infection was considered most likely to have been introduced when one of the seals was captured from the wild, as tuberculosis was subsequently isolated from the same species of seal found dead in the area in which the seals of the marine park were originally captured (18). Furthermore, molecular typing by REA, PGRS-RFLP and IS6110-RFLP revealed identical or similar patterns (18).

Transmission dynamics of zoonotic Mycobacterium bovis In the outbreak of M. bovis among the seals described above, typing was used successfully to show that a seal trainer

contracted the same infection (51) . Three years after having left the marine park, the trainer developed clinical pulmonary tuberculosis. Although the trainer had moved to the East Coast of Australia, the path of transmission was established by the use of REA, which revealed that the isolate was the same clone as that detected in the seals.

Zoonotic M. bovis remains a problem in Argentina, where infection is principally an occupational hazard among rural and abattoir workers (5) . The hypothesis that cattle are the major reservoir for human infections is supported by a study which demonstrated that the majority of M. bovis isolates from humans had a single IS6110 copy, as did those examined from cattle (54) . In contrast, in the Netherlands, only 4 0 % of the M. bovis isolates from humans contained a single IS6110 copy, although the local cattle strains, like those in Argentina, contain only a single copy. Combined with the fact that the cattle population of the Netherlands has been free of tuberculosis for several decades, this indicates that an animal source other than cattle may be infecting humans.

However, in Spain, where M. bovis in cattle and goats remains a problem, evidence suggests that the source of most cases in humans is cattle. In a study of eighty M. bovis isolates from humans, spoligotyping was used to demonstrate that seventy-seven belonged to the bovine type (32) . However, three isolates were typical of caprine strains. As subsequent tracing showed that all three patients had contact with goats, this suggests that goats (a previously unrecognised source of infection) have the potential to transmit the disease to humans.

Until as recently as the mid-1980s, the majority opinion held that human-to-human transmission of M. bovis occurred infrequently (30) . However, several examples of this occurrence have now been documented, generally in HIV positive patients, and typing has proved essential in providing evidence for transmission. A serious case was documented in which a citizen of Canada acquired the infection after visiting a dying relation in Spain, and approximately one year after returning home was diagnosed with severe pulmonary tuberculosis (36). Spoligotyping was used to show the commonality of the isolates from Spain and Canada.

Transmission dynamics in domestic pets Tuberculosis in domestic pets has particular importance, both due to the relative value of these animals and the risk of transmission to the owners. To date, few studies have used typing in such a clinical context. An interesting investigation was undertaken over a three-month period, on a cluster of twelve cases of tuberculosis in cats in a geographically localised area (20) . On the basis of identical REA patterns, the study concluded that the cats were exposed to a single source of infection. This source could not be conclusively determined, although infection may have occurred while the cats were visiting the same veterinary clinics at which the cases were later diagnosed.

696 Rev. sci. tech. Off. int. Epiz, 19 (3)

The potential of domestic cats to act as a source of infection

for humans has been indicated. Distinct M. bovis types (as

determined by LS6110-RFLP) found in urban domestic cats

from Buenos Aires were also isolated from humans, indicating

possible transmission (54).

Clarifying the taxonomy of the Mycobacterium tuberculosis complex Discrimination between M. bovis and M. tuberculosis has

traditionally been achieved by performing a series of cultural

and biochemical tests. Nevertheless, some isolates are

intermediate in these characteristics and fall between the two

'species'. For this reason, a degree of misclassification may

have occurred in the past. This, in turn, has serious

implications for subsequent epidemiological investigations

and any consequent changes in public health policy. This is

illustrated by a case study of a multidrug-resistant isolate from

a patient with HIV-acquired immune deficiency syndrome

(AIDS) (9). The isolate was initially reported to be M. bovis, as

all the strains possessed all the indicated phenotypic

characteristics. However, subsequent genetic analysis revealed

that the isolates were in fact dysgonic, slow-growing

M. tuberculosis strains which mimicked the M. bovis

phenotype, probably as a consequence of cellular alterations

associated with the multidrug resistance (33) . Spoligotyping

and IS6110-RFLP analysis confirmed that the outbreak was

due to a single strain. As several typing techniques have been

shown to consistently discriminate between M. bovis and

M. tuberculosis, reference laboratories for human tuberculosis

should introduce molecular typing as a routine confirmatory

test on all suspected M. bovis isolates.

In Spain, both caprine and bovine tuberculosis remain serious

problems (28). One particular epidemiological question of

interest is the relation between the strains in goats and cattle,

and through typing, the hypothesis that different subspecies

were involved in the infection of the two host species was

developed. Gutiérrez et al. demonstrated that a genetic

difference was discernible in that, while the seventeen bovine

strains had only a single IS6110, those from goats were found

to carry six to eight copies of the element (31). A similar

difference was shown between isolates from the two species

when a more extensive data source (125 isolates) was

investigated by Liébana et al. (35) , even though almost half of

the bovine isolates were shown to have more than one copy of

IS6110. A third study looked at a wider range of species, using

spoligotyping, and identified two species clusters: the first

comprising cattle, cats, wild boar and deer and the second

confined to sheep and goats (2). As a result of these typing

results, a thorough microbiological study, comparing the

cattle and goat isolates, has recently been performed (4). The

study concluded that the caprine isolates from Spain are

sufficiently distinct to warrant the creation of a separate

subspecies of the M. tuberculosis complex, namely, subspecies

caprae.

Genotyping has also been useful in clarifying the taxonomic

position of isolates recovered from seals with tuberculosis in

the south-western Atlantic (57). Initially thought to be

M. bovis, based on differences in the number of copies of

IS6110, spoligotyping and gene polymorphism, the strain has

been shown to represent a distinct subgroup of the

M. tuberculosis complex.

Association of types with pathogenicity Recent studies with M. tuberculosis have demonstrated an

association of types with virulence. The most conclusive of

these to date was the identification of a strain that was

responsible for a large outbreak of tuberculosis in a rural

community in the USA, which otherwise had a low risk of the

disease (53). The RFLP genotyping showed the strain to be

clonal, and laboratory studies with mice confirmed that it was

of high virulence.

As yet, only preliminary studies have been undertaken

regarding the association of the genetic variability of M. bovis

and virulence. In a study conducted in Northern Ireland, the

association between RFLP patterns and the nature of the

disease in several herds was investigated (46). However, it was

concluded that there was no evidence for a virulence-type

linkage. Perumaalla et al. discovered that isolates with

multiple copies of IS6110 were mainly found in animals other

than cattle (42). This suggested that these M. bovis isolates

with multiple copies of IS6110 may have become attenuated

for virulence in cattle. However, a study of isolates from Spain

found no evidence to support this hypothesis (3) .

Discussion: research needs and future directions The above review of applications presents very few examples

of the successful use of genotyping of M. bovis to resolve an

epidemiological problem and to enable firm conclusions to be

drawn. The exceptions are the study by Bölske et al., where

the role of imported deer from the UK was clearly

demonstrated (8); and the study by Thompson e£ al., which

showed that the pulmonary tuberculosis in the seal trainer

was the result of contact with infected animals (51). This

contrasts with the plethora of successful molecular

epidemiological studies conducted on M. tuberculosis.

One of the major factors responsible for the relative success of

molecular epidemiology in M. tuberculosis, compared to

M. bovis, was the early development of a successful typing

technique (IS6110-RFLP), which possessed the twin desirable

characteristics of isolate stability and population diversity.

Consequently, successful molecular epidemiological studies

of tuberculosis in humans were undertaken within a few years

of the development of IS6110-RFLP. In contrast, veterinary

molecular epidemiology has been greatly hampered by the

lack of a comparable technique, despite the fact that the

Rev. sci. tech. Off. int. Epiz., 19 (3) 697

earliest work in the genotyping of the M. tuberculosis complex was performed by veterinary researchers (13). Most research effort has therefore focused upon developing and validating new techniques, and a large proportion of scientific papers referring to the 'molecular epidemiology' of M. bovis are in fact genotyping validations (42, 46) .

This problem of a lack of a 'gold standard' genotyping technique which combines ease of use, good discrimination and isolate stability, has still to be resolved (22). However, the progress in PCR techniques over recent years has been sufficient to inspire confidence that such a technique or combination will be developed soon. Traditionally, the validation of genotyping techniques has concentrated on a comparison with existing techniques, especially in terms of discrimination. While this was adequate in the earlier pioneering stages of M. bovis genotyping, progress has now been sufficient for a more rigorous approach. In particular, a greater emphasis on epidemiological usefulness as the key attribute is required. One method of assessing the usefulness of a technique to resolve competing hypotheses is through case studies for which a substantial amount of epidemiological information has been collected.

A critical requirement in the use of genotyping to resolve epidemiological problems is the existence of appropriate databases. The key databases needed are those providing base-line farm and disease data, and a reference typing library. At present, most developed countries now have established national animal health information systems. In contrast, no standard reference system exists for typing; even for spoligotyping, which is rapidly becoming one of the most popular typing techniques in veterinary and medical genotyping of mycobacteria, no standard nomenclature exists. Once an M. bovis typing library is established, one of the immediate benefits will be for countries and regions that have already eradicated bovine tuberculosis. When dealing with imported animals that are diagnosed with the disease, the ability to establish whether the type is 'common' in the exporting country will provide supporting evidence of whether infection occurred pre- or post-importation.

A database system that may have wide potential in the molecular epidemiology of M. bovis is GIS. The demonstration of geographical clustering of spoligotypes from both cattle and the principal wildlife reservoir, the Eurasian badger, has important implications in terms of understanding the dynamics of the disease (12). Comparable results may be found in other countries where wildlife reservoirs are important in the persistence of the disease, specifically Ireland and New Zealand. Nevertheless, GIS may not be relevant in all situations, especially those involving considerable movement of animals (24).

One result of the lack of traditional epidemiological input into M. bovis genotyping studies, has been a general lack of biostatistical analysis. This was not necessarily a problem in the earlier stages of the development of the technology, as the imposition of rigorous biostatistical criteria can restrict the range and extent of experimentation. For example, many of the pioneering studies were performed on very few isolates, due to the technical difficulties that were encountered, and if publication had been delayed until an adequate sample had been achieved, progress would have been much slower. However, bovine tuberculosis molecular epidemiology is entering a stage of greater maturity, and consequently, more rigorous biostatistical sampling and analysis is required.

A current trend in M. tuberculosis molecular epidemiology is the greater integration of research, in which macro-scale epidemiological studies inform research at the molecular level and vice-versa (6) . Such a collaborative approach is particularly beneficial, as molecular biologists are now faced with a plethora of information regarding genomic sequences, but require some guidance as to which are of relevance in understanding pathogenicity, host resistance, etc. No comparable integrated, multidisciplinary M. bovis studies have yet been attempted, but such studies are undoubtedly essential if progress is to. continue in molecular epidemiology.

The history of molecular epidemiological studies of bovine tuberculosis has been mixed. In the fifteen years since genotyping methods were first introduced, the technical achievement in developing and refining these methods has been impressive. However, much less progress has been achieved in the application of genotyping to provide insight into disease dynamics and improved methods of disease control. Nevertheless, sufficient positive examples can be found in the molecular epidemiology of human tuberculosis to lend confidence to the belief that practical utilisation of typing in routine control of bovine tuberculosis should be achievable in the near future.

Acknowledgements The authors would like to thank Richard Grenville and Roger Sainsbury for helpful editorial comments. This review was funded by the Ministry of Agriculture, Fisheries and Food, UK, as part of investigations into the application of GIS and spoligotyping for the control of bovine tuberculosis in Great Britain (Project SE3001) .

698 Rev. sci. tech. Off. int. Epiz., 19 (3)

Épidémiologie moléculaire de la tuberculose bovine II . Applications du typage génomique

P.A. Durr, R.S. Clif ton-Hadley & R.G. Hewinson

Résumé Les auteurs examinent les applications du typage génomique de Mycobacterium bovis. Les travaux de recherche publiés à ce jour ont été effectués essentiellement dans le cadre de la validation de méthodes de typage et peu d'études ont été menées à des fins purement épidémiologiques. Il en va tout autrement de la tuberculose humaine pour laquelle l'application du typage par polymorphisme de taille des fragments de restriction utilisant la séquence d'insertion IS6110 a permis de mieux comprendre l'épidémiologie et l'évolution moléculaire de l'agent pathogène. L'expérience de la médecine humaine montre l'intérêt, pour les études à venir, d'une méthode intégrée alliant épidémiologie et biologie moléculaire. Il devrait donc être possible d'identifier avec précision les groupements par type et de les expliquer, ce qui devrait faciliter les décisions en matière de prophylaxie.

Mots-clés Analyse par enzymes de restriction (endonucléases) - Épidémiologie moléculaire -Mycobacterium bovis - Polymorphisme de taille des fragments de restriction -Spoligotypage - Tuberculose - Typage génomique.

•

Epidemiología molecular de la tuberculosis bovina II . Aplicaciones de la t ipif icación genética

P.A. Durr, R.S. Clif ton-Hadley & R.G. Hewinson

Resumen Los autores pasan revista a las aplicaciones de la tipificación genética de Mycobacterium bovis. Las investigaciones publicadas hasta ahora estaban destinadas principalmente a validar métodos de tipificación, por lo que de momento existen pocos estudios de carácter específicamente epidemiológico. Muy distinta es la situación en el campo de la tuberculosis humana, donde, gracias al uso de la secuencia de inserción IS6110 para analizar el polimorfismo de longitud de los fragmentos de restricción, se empieza a comprender la epidemiología y la evolución molecular del agente patógeno. Basándose en estos antecedentes, los autores recomiendan que de ahora en adelante se aborde la cuestión de forma integral, combinando estudios de epidemiología y de biología molecular. Ello permitiría identificar claramente los tipos bacterianos y determinar sus respectivas afinidades y relaciones, lo que sin duda facilitaría la adopción de decisiones relativas al control de la enfermedad.

Palabras clave Análisis por endonucleasas de restricción - Epidemiología molecular - Mycobacterium bovis - Polimorfismo de longitud de fragmentos de restricción - Spoligotipificación -Tipificación genética - Tuberculosis.

Rev. sci. tech. Off. int. Epiz., 19 (3) 699

References

1. Alland D., Kalkut G.E., Moss A.R., McAdam RA., Hahn J.A., Bosworth W., Drucker E. & Bloom B.R. (1994). -Transmission of tuberculosis in New York City - an analysis by DNA fingerprinting and conventional epidemiologic methods. New Engl. J. Med., 330, 1710-1716.

2. Aranaz A., Liébana E., Mateos A., Domínguez L., Vidal D., Domingo M., Gonzolez O., Rodriguez-Ferri E.F., Bunschoten A.E., Van Embden J.D.A. & Cousins D. (1996). - Spacer oligonucleotide typing of Mycobacterium bovis strains from cattle and other animals: a tool for studying epidemiology of tuberculosis. J . clin. Microbiol., 34, 2734-2740.

3. Aranaz A., Liébana E., Mateos A., Domínguez L. & Cousins D. (1998). - Restriction fragment length polymorphism and spacer oligonucleotide typing: a comparative analysis of fingerprinting strategies for Mycobacterium bovis. Vet. Microbiol, 61, 311-324.

4. Aranaz A., Liébana E., Gómez-Mampaso E., Galán J.C., Cousins D., Ortega A., Blázquez J . , Baquero F., Mateos A., Súarez G. & Domínguez L. (1999). - Mycobacterium tuberculosis subsp. caprae subsp. nov.: a taxonomic study of a new member of the Mycobacterium tuberculosis complex isolated from goats in Spain. Int. J. syst. Bacteriol, 49 , 1263-1273.

5. Barrera L. & de Kantor I.N. (1987). - Non tuberculous mycobacteria and Mycobacterium bovis as a cause of human disease in Argentina. Trop. geogr. Med., 39, 222-227.

6. Bifani P.J., Mathema B., Liu Z.Y., Moghazeh S.L., Shopsin B., Tempalski B., Driscoll J . , Frothingham R., Musser J.M., Alcabes P. & Kreiswirth B.N. (1999). - Identification of a W variant outbreak of Mycobacterium tuberculosis via population-based molecular epidemiology. J . Am. med. Assoc., 282, 2321-2327.

7. Bloom B.R. & Murray C. (1992). - Tuberculosis: commentary on a reemergent killer. Science, 257, 1055-1064.

8. Bölske G., Englund L., Wahlstrom H., de Lisle G.W., Collins D.M. & Croston P.S. (1995). - Bovine tuberculosis in Swedish deer farms: epidemiological investigations and tracing using restriction fragment analysis. Vet. Rec., 136, 414-417.

9. Bouvet E., Casalino E., Mendoza-Sassi G., Lariven S., Vallée E., Pernet M., Gottot S. & Vachon F. (1993). - A nosocomial outbreak of multidrug-resistant Mycobacterium bovis among HIV-infected patients. A case-control study. AIDS, 7, 1453-1460.

10. Braden C.R., Templeton G.L., Cave M.D., Valway S., Onorato I.M., Castro K.G., Moers D., Yang Z., Stead W.W. & Bates J.H. (1997). - Interpretation of restriction fragment length polymorphism analysis of Mycobacterium tuberculosis isolates from a state with a large rural population. J. inject. Dis., 175, 1446-1452.

11. Clifton-Hadley R.S. & Wilesmith J.W. (1995). - An epidemiological outlook on bovine tuberculosis in the developed world. In Proc. 2nd International Conference on Mycobacterium bovis (F. Griffin & G. de Lisle, eds), 28 August-2 September, Dunedin, New Zealand. University of Otago Press, Otago, 178-182.

12. Clifton-Hadley R.S., Inwald J . , Archer J . , Hughes S., Palmer N., Sayers A.R., Sweeney K., Van Embden J.D.A. & Hewinson R.G. (1998). - DNA fingerprinting of Mycobacterium bovis isolates using spoligotyping -epidemiological issues. In Proc. Society for Veterinary Epidemiology and Preventive Medicine (SVEPM) (M.V. Thrusfield & E.A. Goodall, eds), 25-27 March, West County Hotel, Co. Clare. SVEPM, Dublin, 15-27.

13. Collins D.M. & de Lisle G.W. (1984). - DNA restriction endonuclease analysis of Mycobacterium tuberculosis and Mycobacterium bovis BCG.J. gen. Microbiol, 130, 1019-1021.

14. Collins D.M., de Lisle G.W. & Gabric D.M. (1986). -Geographic distribution of restriction types of Mycobacterium bovis isolates from brush-tailed possums (Trichosurus vulpecula) in New Zealand. J. Hyg. (Camb.), 96, 431-438.

15. Collins D.M., Gabric D.M. & de Lisle G.W. (1988). - Typing of Mycobacterium bovis isolates from cattle and other animals in the same locality. (Letter.) N.Z. vet. J., 36, 45-46.

16. Collins D.M., de Lisle G.W., Collins J.D. & Costello E. (1994). - DNA restriction fragment typing of Mycobacterium bovis isolates from cattle and badgers in Ireland. Vet. Rec., 134, 681-682.

17. Cousins D.V., Francis B.R., Gow B.L., Collins D.M., McGlashan C.H., Gregory A. & MacKenzie R.M. (1990). -Tuberculosis in captive seals: bacteriological studies on an isolate belonging to the Mycobacterium tuberculosis complex. Res. vet. Sci., 48, 196-200.

18. Cousins D.V., Williams S.N., Reuter R., Forshaw D., Chadwick B., Coughran D., Collins P. & Gales N. (1993). -Tuberculosis in wild seals and characterisation of the seal bacillus. Aust. vet.J., 70, 92-97.

19. Daborn C.J., Grange J.M. & Kazwala R.R. (1996). - The bovine tuberculosis cycle - an African perspective. J. appl. Bacteriol., 81 , S27-S32.

20. De Lisle G.W., Collins D.M., Loveday A.S., Young W.A. & Julian A.F. (1990). - A report of tuberculosis in cats in New Zealand, and the examination of strains of Mycobacterium bovis by DNA restriction endonuclease analysis. N.Z. vet. J., 38, 10-13.

21. De Lisle G.W., Yates G.E., Collins D.M., MacKenzie R.W., Crewe K.B. & Walker R. (1995). - A study of bovine tuberculosis in domestic animals and wildlife in the MacKenzie Baisin and surrounding areas using DNA fingerprinting. N.Z. vet.J., 43, 266-271.

700 Rev. sci. tech. Off. int. Epiz., 19 (3)

22. Durr P.A., Hewinson R.G. & Clifton-Hadley R.S. (2000). -Molecular epidemiology of bovine tuberculosis. I: M. bovis genotyping. Rev. sci. tech. Off. int. Epiz., 19 (3), 675-688.

23. Essey M.A. & Koller M.A. (1994). - Status of bovine tuberculosis in North America. Vet. Microbiol, 40, 15-22.

24. Ficht T.A., Whipple D., Perumaalla V., Chacón O., Alford P., Slater M., Baca D., Hernández J . , Payeur J . , Jarnagin J . , Suárez F., Turcotte C , Rohonczy E. & Adams L.G. (1997). -Molecular epidemiologic and geographic information system analyses of Mycobacterium bovis isolates from North America. In Proc. 101st Annual Meeting of the United States Animal Health Association (USAHA), 18-24 October, Louisville, Kentucky. USAHA, Richmond, Virginia, 534-542.

25. Fisanotti J.C., Alito A., Bigi F., Latini O., Roxo E., Cicuta E., Zumarraga M.J., Cataldi A. & Romano M.I. (1998). -Molecular epidemiology of Mycobacterium bovis isolates from South America. Vet. Microbiol, 60, 251-257.

26. Forshaw D. & Phelps G.R. (1991). - Tuberculosis in a captive colony of pinnipeds. J. Wildl. Dis., 27, 288-295.

27. Francis J. (1958). - Tuberculosis in animals and man. Cassel & Co., London, 11-43.

28. Garcia Marin J.F. & Gutiérrez M. (1996). - Aspectos epidemiológicos y clínicos de la tuberculosis caprina [Epidemiological and clinical aspects of caprine tuberculosis]. Ovis, 46, 45-60.

29. Glynn J.R., Vynnycky E. & Fine P.E.M. (1999). - Influence of sampling on estimates of clustering and recent transmission of Mycobacterium tuberculosis derived from DNA fingerprinting techniques. Am. J. Epidemiol, 149, 366-371.

30. Grange J.M. & Collins C.H. (1987). - Bovine tubercle bacilli and disease in animals and man. Epidemiol Infect., 92, 221-234.

31. Gutiérrez M., Samper S., Gavigan J.-A., Garcia Marin J.F. & Martin C. (1995). - Differentiation by molecular typing of Mycobacterium bovis strains causing tuberculosis in cattle and goats. J. clin. Microbiol, 33, 2953-2956.

32. Gutiérrez M., Samper S., Jiménez S., van Embden J.D.A., Garcia Marin J.P. & Martin C. (1997). - Identification by spoligotyping of a caprine genotype in Mycobacterium bovis strains causing human tuberculosis. J. clin. Microbiol., 35, 3328-3330.

33. Gutiérrez M.C., Galan J.C., Blazquez J . , Bouvet E. & Vincent V. (1999). - Molecular markers demonstrate that the first described multidrug-resistant Mycobacterium bovis outbreak was due to Mycobacterium tuberculosis. J . clin. Microbiol., 37, 971-975.

34. Källenius G., Bölske G., Irmerstedt A., Ramberg M.R.O. & Svenson S.B. (1993). - Smittade gibbonapan tapiren eller tvärtom? Ny teknik för smittspaming av tubkulos [Did the tapir infect the ape or vice versa? A new technique for tracing tuberculosis]. Läkartidningen, 90, 4658-4659.

35. Liébana E., Aranaz A., Dominguez L., Mateos A., González-Llamazares P., Rodriguez-Ferri E.F., Domingo M., Vidal D. & Cousins D. (1997). - The insertion element IS6110 is a useful tool for DNA fingerprinting of Mycobacterium bovis isolates from cattle and goats in Spain. Vet. Microbiol., 54, 223-233.

36. Long R., Nobert E., Chomyc S., van Embden J . , McNamee C, Duran R.R., Talbot J . & Fanning A. (1999). -Transcontinental spread of multidrug-resistant Mycobacterium bovis. Am. J. resp. c r i t . Care Med., 159, 2014-2017.

37. Maslow J.N., Mulligan M.E. & Arbeit R.D. (1993). -Molecular epidemiology: application of contemporary techniques to the typing of microorganisms. Clin. infect. Dis., 17, 153-164.

38. Milian-Suazo F., Salman M.D., Black W.C., Triantis J.M., Ramirez C , Payeur J.B. & Torres M.C. (2000). - Molecular epidemiologic analysis of Mycobacterium bovis isolates from Mexico. Am. J. vet. Res., 61 , 90-95.

39. Morris R.S., Pfeiffer D.U. & Jackson R. (1994). - The epidemiology of Mycobacterium bovis infections. Vet. Microbiol, 40, 153-177.

40. Neill S.D., Pollock J.M., Bryson D.B. & Hanna J . (1994). -Pathogenesis of Mycobacterium bovis infection in cattle. Vet. Microbiol, 40, 41-52.

41. O'Reilly L.M. & Daborn C.J. (1995). - The epidemiology of Mycobacterium bovis infections in animals and man: a review. Tubercle Lung Dis., 76, 1-46.

42. Perumaalla V.S., Adams L.G., Payeur J.B., Jarnagin J.L., Baca D.R., Guemes F.S. & Ficht T.A. (1996). - Molecular epidemiology of Mycobacterium bovis in Texas and Mexico.

J . din. Microbiol, 34, 2066-2071.

43. Pillai S.D., Widmer K.W., Ivey L.J., Newman E., Lingsweiler S., Baca D., Kelley M., Davis D.S., Silvey N.J. & Adams L.G. (2000). - Failure to identify non-bovine reservoirs of Mycobacterium bovis in a region with a history of infected dairy-cattle herds. Prev. vet. Med., 43, 53-62.

44. Rhyan J . , Aune K., Hood B., Clarke R., Payeur J . , Jarnagin J . & Stackhouse L. (1995). - Bovine tuberculosis in a free-ranging mule deer (Odocoileus hemionus) from Montana. J. Wildl. Dis., 31 , 432-435.

45. Serraino A., Marchetti G., Sanguinetti V., Rossi M.C., Zanoni R.C., Catozzi L., Bandera A., Dini W., Mignone W., Franzetti F. & Gori A. (1999). - Monitoring of transmission of tuberculosis between wild boars and cattle: genotypical analysis of strains by molecular epidemiology techniques. J. clin. Microbiol., 37, 2766-2771.

46. Skuce R.A., Brittain D., Hughes M.S., Beck L.-A. & Neill S.D. (1994). - Genomic fingerprinting of Mycobacterium bovis from cattle by restriction fragment length polymorphism analysis. J . din. Microbiol., 32, 2387-2392.

47. Skuce R.A., Brittain D., Hughes M.S. & Neill S.D. (1996). -Differentiation of Mycobacterium bovis isolates from animals by DNA typing. J. clin. Microbiol, 34, 2469-2474.

Rev. sci. tech. Off. int. Epiz., 19 (3) 701

48. Small P.M., Hopewell P.C., Singh S.P., Paz A., Parsonnet J . , Ruston D.C., Schecter G.F., Daley C.L. & Schoolnik G.K. (1994). - The epidemiology of tuberculosis in San Francisco - a population-based study using conventional and molecular methods. New Engl J . Med., 330, 1703-1709.

49. Szewzyk R., Svenson S.B., Hoffner S.E., Bolske G., Wahlstrom H., Englund L., Engvall A. & Kallenius G. (1995). - Molecular epidemiologic studies of Mycobacterium bovis infections in humans and animals in Sweden. J. clin. Microbiol, 33, 3183-3185.

50. Thoen CO., Schliesser T. & Kormendy B. (1995). -Tuberculosis in captive wild animals. In Mycobacterium bovis infection in animals and humans (CO. Thoen & J.H. Steele, eds). Iowa State University Press, Ames, 93-104.

51. Thompson P.J., Cousins D.V., Gow B.L., Collins D.M., Williamson B.H. & Dagnia H.T. (1993). - Seals, seal trainers and mycobacterial infection. Am. Rev. respir. Dis., 147, 164-167.

52. Thorel M.-F., Karoui C , Varnerot A., Fleury C. & Vincent V. (1998). - Isolation of Mycobacterium bovis from baboons, leopards and a sea-lion. Vet. Res., 29, 207-212.

53. Valway S.E., Sanchez M.P.C., Shinnick T.F., Orme I., Agerton T., Hoy D., Jones J.S., Westmoreland H. & Onorato I.M. (1998). - An outbreak involving extensive transmission of a virulent strain of Mycobacterium tuberculosis. New Engl. J. Med., 338, 633-639; 1783 (erratum).

54. Van Soolingen D., de Haas P.E.W., Haagsma J . , Eger T., Hermans P.W.M., Ritacco V., Alito A. & Van Embden J.D.A. (1994). - Use of various genetic markers in differentiation of Mycobacterium bovis strains from animals and humans and for studying epidemiology of bovine tuberculosis. J. clin. Microbiol, 32, 2425-2433.

55. Whipple D.L., Clarke P.R., Jarnagin J.L. & Payeur J.B.

(1997). - Restriction fragment length polymorphism analysis

of Mycobacterium bovis isolates from captive and free-ranging

animals. J . ve£. diagn. Invest., 9, 381-386.

56. Zumárraga M.J., Martin C , Samper S., Alito A., Latini O.,

Bigi F., Roxo E., Cicuta M.E., Errico F., Ramos M.C,

Cataldi A., van Soolingen D. & Romano M.I. (1999). -

Usefulness of spoligotyping in molecular epidemiology of

Mycobacterium bovis-related infections in South America.

J . clin. Microbiol, 37, 296-303.

57. Zumárraga M.J., Bemardelli A., Bastida R., Quse V.,

Loureiro J . , Cataldi A., Bigi F., Alito A., Ramos M.C.,

Samper S., Otal I., Martin C & Romano M.I. (1999). -

Molecular characterization of mycobacteria isolated from

seals. Microbiology, 145, 2519-2526.