JOURNAL OF CLINICAL MICROBIOLOGY, Feb. 1993, p. 406-4090095-1137/93/020406-04$02.00/0

NOTES

Strain Identification of Mycobacterium tuberculosis by DNAFingerprinting: Recommendations for a

Standardized MethodologyJAN D. A. VAN EMBDEN,1* M. DONALD CAVE,2 JACK T. CRAWFORD,3 JEREMY W. DALE,4KATHLEEN D. EISENACH,2 BRIGIYIrE GICQUEL,s PETER HERMANS,6 CARLOS MARTIN,7

RUTH McADAM,8 THOMAS M. SHINNICK,3 AND PETER M. SMALL9Unit Molecular Microbiology, National Institute ofPublic Health and Environmental Protection, P.O. Box 1, 3720 BABilthoven, The Netherlands'; University ofArkansas for Medical Sciences, Little Rock, Arkansas 722052; Division

ofBacterial and Mycotic Diseases, National Center for Infectious Diseases, Centers for Disease Control,Atlanta, Georgia 303333; Department ofMicrobiology, University of Surrey, Guildford GU2 5XH,

United Kingdom4; Microbial Engineering Unit, Institut Pasteur, 75724 Paris, Cedex 15,Frances; Armauer Hansen Research Institute, Addis Ababa, Ethiopia6; Area

Microbiologia, Faculty ofMedicine, University of Zaragoza, Zaragoza,Spain7; Albert Einstein College ofMedicine, Yeshiva University,Bronx, New York 104618; and Beckman Center B251, Howard

Hughes Medical Institute, Stanford University,Stanford, California 94305-54259

Received 26 August 1992/Accepted 26 October 1992

DNA fingerprinting of Mycobacterium tuberculosis has been shown to be a powerful epidemiologic tool. Wepropose a standardized technique which exploits variability in both the number and genomic position of IS6110to generate strain-specific patterns. General use of this technique will permit comparison of results betweendifferent laboratories. Such comparisons will facilitate investigations into the international transmission oftuberculosis and may identify specific strains with unique properties such as high infectivity, virulence, or drugresistance.

Epidemiologic studies of tuberculosis can be greatly facil-itated by the application of strain-specific markers. Unusualantibiotic susceptibility patterns and phage typing, whichhave been used for this purpose, have significant limitations.The recently discovered transposable elements in Mycobac-terium tuberculosis have been shown to be of great potentialfor use in strain differentiation. M. tuberculosis strain typinghas already proven to be extremely useful in outbreakinvestigations (2, 7, 13) and is being applied to a variety ofepidemiologic questions in numerous laboratories.The existence of repetitive DNA elements in M. tubercu-

losis and their potential for use in fingerprinting of M.tuberculosis isolates was recognized independently byEisenach et al. (4), Zainuddin (14), and Zainuddin and Dale(15). The sequence of one of these elements, designatedIS6110, was first reported by Thierry et al. (11, 12) and wasshown to be related to the IS3 family of insertion sequenceswhich were discovered in members of the family Enterobac-teriaceae. McAdam et al. (9) independently sequenced theelement isolated by Zainuddin and Dale (15), which wasdesignated IS986. Subsequently, a related element fromMycobacterium bovis BCG was sequenced by Hermans etal. (6) and is referred to as IS987. These three sequencesdiffer in only a few base pairs and therefore can be consid-ered essentially the same element. To avoid confusion, we

* Corresponding author.

recommend the designation IS6110 for these elements, ex-cept when a specific copy is concerned.The effectiveness of this insertion sequence (IS) typing

system for epidemiological analysis of M. tuberculosis iso-lates has been demonstrated in a number of studies (1-3, 5,7, 8, 10, 13). In principle, the results obtained by testing largenumbers of strains in different laboratories could be com-pared. This would allow strains from different geographicareas to be compared and the movement of individual strainsto be tracked. Such data may provide important insights intothe global transmission of tuberculosis and identify strainswith particular properties, such as high infectivity, highvirulence, and/or multidrug resistance. Analysis of largenumbers of isolates may provide answers to long-standingquestions regarding the efficacy of BCG vaccination and thefrequency of reactivation versus reinfection, which are in-creasingly important in light of the AIDS pandemic. Theselarge-scale studies will require the use of computer-assistedanalysis and comparison of DNA fingerprints. This reportdescribes such a standard method, which will be adopted inour laboratories, and recommends its use to other laborato-ries so that the results obtained by different laboratories canbe compared.

Bacterial growth, DNA extraction, digestion of DNA, andSouthern blotting were done as described previously (13).Agarose gels were loaded with a mixture of 1 ,g of PvuII-digested genomic M. tuberculosis DNA and molecular sizemarker DNA. PvuII-digested supercoiled ladder DNA (4 ng;

406

Vol. 31, No. 2

on July 24, 2020 by guesthttp://jcm

.asm.org/

Dow

nloaded from

on July 24, 2020 by guesthttp://jcm

.asm.org/

Dow

nloaded from

on July 24, 2020 by guesthttp://jcm

.asm.org/

Dow

nloaded from

NOTES 407

(481)

PvuII SstII Bem HII I I

Bss HIl

l B~pMil

IS61 101355

IS probeFIG. 1. Physical map of the 1.35-kb M. tuberculosis insertion element IS6110 (9). The cleavage sites of several restriction enzymes are

depicted. PvuII cleaves the element at base pair 461. Therefore, any chromosomal mycobacterial DNA fragment obtained by therecommended standard typing method is larger than 0.9 kb. The closed bars represent the 28-bp inverted repeats bordering IS6110 DNA. Thelines to the left and right denote chromosomal DNA.

Bethesda Research Laboratories, Gaithersburg, Md.) and1.2 ng of HaeII-digested OX174 DNA (Clontech, Palo Alto,Calif.) were used as molecular size standards. The molecularsizes of these two reference markers ranged from 16.2 to0.603 kb, respectively (see Fig. 2B).The mycobacterial IS probe was prepared by peroxidase

labeling of a 245-bp fragment obtained by amplification bythe polymerase chain reaction (PCR) described previously(7). Briefly, the oligonucleotides INS-1 (5'-CGTGAGGGCATCGAGGTGGC) and INS-2 (5'-GCGTAGGCGTCGGTGACAAA) were used to amplify a 245-bp fragment from puri-fied chromosomal M. bovis BCG DNA by PCR. This frag-ment was purified by Sephadex G50 chromatography. TheDNA was precipitated with ethanol, and after solubilizationthe DNA was labeled with peroxidase as described previ-ously (7).

Standard method of fingerprinting. DNA typing of M.tuberculosis complex strains is based on polymorphismsgenerated by variabilities in both the copy numbers and thechromosomal positions of IS6110 among clinical isolates ofM. tuberculosis (1-3, 5-8, 10, 13, 15). The technique offingerprinting entails the growth of M. tuberculosis, extrac-tion of DNA, restriction endonuclease digestion, Southernblotting, and probing for the IS element. Only three param-eters are critical for a standardized IS6110-based DNAfingerprinting system: the specificity of the restriction en-zyme, the nature of the DNA probe, and the use of appro-priate molecular mass standards.The physical map of the IS6110 sequence (Fig. 1) indicates

that various restriction enzymes cleave within the 1,355-bpelement. BamHI, SstII, PstI, BstEII, BssHII, and PvuIIhave all been successfully used to generate restriction frag-ment length polymorphisms (1-8, 13, 14). For the standardmethod we recommend the use of PvuII, because it has beenused by the majority of laboratories and it cleaves the IS6110sequence only once. Because of this latter property, PvuIIdigestion of IS6110-containing genomic DNA leads toIS6110-hybridizing fragments of at least 0.90 or 0.46 kb,depending on the IS6110 probe that is used. Since M.tuberculosis usually contains 8 to 20 IS6110 copies (13), theuse of a DNA probe which overlaps both sides of the PvuIIsite would result in 16 to 40 bands. This large number ofbands would result in overcrowded lanes with overlappingbands. Therefore, we arbitrarily chose a DNA probe to theright of the PvuII site on the physical map, as shown in Fig.1. This reduced the number of IS6110-containing bands inthe fingerprint to half of the maximum number possible. Inexceptional cases, when the differentiation of the patterns is

not adequate, the membranes could be reprobed with labeledDNA containing only IS sequences to the left of the PvuIIsite. For the standardized method, the exact DNA se-

quences of the probe do not matter as long as the ISsequence to the right of the PvuII site is used. An illustrationof a Southern blot is given in Fig. 2A.

In order to compare fingerprints between M. tuberculosisisolates run on different gels and in different laboratories, thesize of each IS6110-hybridizing fragment must be deter-mined. This requires the use of molecular size markerswhich span the 10- to 0.9-kb range of most IS6110-hybridiz-ing fragments. We recommend use of a combination ofexternal and internal standards which provide a compromisebetween technical ease and maximal precision. Externalmolecular size markers should be run in two or three lanes ofeach gel. Furthermore, we recommend inclusion in each gela lane containing DNA from the reference strain of M.tuberculosis Mt14323, which, when digested with PvuII andprobed with IS6110, gives 10 approximately evenly spacedbands of known size (Fig. 2A). Although the use of externalmarkers is adequate for comparing small numbers of similarstrains (such as in outbreak investigations), it may notprovide sufficient precision to permit computerized compar-isons of hundreds or thousands of strains. For this reason werecommend that molecular size standards which do nothybridize with IS6110 also be added to the wells with thecleaved M. tuberculosis DNA. After hybridization withIS6110, the membrane can be reprobed with labeled molec-ular size marker standards. This results in a second autora-diograph with molecular size standards in each lane. Thesecond autoradiograph can be superimposed on the firstautoradiograph, resulting in extremely precise molecularsize determinations (Fig. 2B). We were able to obtainstandard deviations of the molecular sizes of less than 2% inthe 1- to 2-kb range and less than 5% in the 9- to 10-kb range(2a). If internal standards are used, a single reference exter-nal marker is sufficient.

Finally, to be able to compare DNA fingerprints made indifferent laboratories, a minimal resolving power of the gelsis needed. At a given agarose concentration, the resolvingpower mainly depends on the electrophoresis time. Werecommend conditions such that the distance between the0.9-kb marker (the approximate size of the smallest PvuIIIS-containing fragment) and the 10-kb marker is at least 10cm.

These recommendations will permit comparison of DNAfingerprints ofM. tuberculosis made in different laboratoriesthat can use their own optimized procedures for DNA

SapMIl StilI I

I

m

VOL. 31, 1993

on July 24, 2020 by guesthttp://jcm

.asm.org/

Dow

nloaded from

408 NOTES J. CLIN. MICROBIOL.

A 1 2 3 4 5-) G/ 11 9'.3 1 () 1t 2 ii 19i1 20 2 1. 2 2

-'

16.21014.174l2.l38 w Z2 -I.098 A3_8.077.04j W.-

6.03 eSf11 -//r w _ amlF___ c mm .-.m.........................4.00 __ _____=2.9)7 . n m

~~~

_ mm am a

I-33 --

OX71-,1Z1.353.v ', -

.t~ejt;-/**.* 0

0.8'72. Y

1 2 1 5 fi 113 9 10 11 12 13 14 I5 16 1/ I0 19 20 21 22B -_

16.2 101't. 17412.138 m m=10.098-4a_--aa -na .- a_am8.0-7 --- a-__-aa a a -_---a-aI ma - a1__- a - a - - - -.. - a< a

6.03() ' , a a a_ _ _ _ _ u_ P .,___. a_a5.0 1 a -a aa~-4.002.9 aa a a a a__._ _ -e_r.w , a a

2.07 we-- a a a- e a . a

- ear a m a a aflm_ e m meem~~~ ~~~

1.078.v/_a - a a _- a- a a a a__en

(0872 Wa.

01.603 .7

FIG. 2. Fingerprints of M. tuberculosis strains obtained by the recommended standard method. Chromosomal DNAs of 22 differentmycobacterial isolates were digested with Pvull, and after mixing with marker DNA, the fragments were separated by overnightelectrophoresis. The fragments were transferred to filters and hybridized with peroxidase-labeled IS6110 DNA (A) and peroxidase-labeledmarker DNA (B). Lane 22, reference M. tuberculosis Mt14323, which is available on request; lanes 2 to 5, strains from an outbreak oftuberculosis in Amsterdam during 1991. All other lanes contain DNAs from M. tuberculosis isolated from epidemiologically unrelated cases.The strains corresponding to lanes 16 to 21 were selected from a collection of about 200 randomly chosen strains from patients in TheNetherlands. The internal DNA markers (B) were PvuII-digested supercoiled ladder DNAs with molecular sizes of 16.2, 14.2, 12.1, 10.1, 8.07,7.04, 6.03, 5.01, 4.00, 2.97, and 2.07 kb and HaeIII-digested 4X174 with molecular sizes of 1,353, 1,078, 872, and 603 bp, respectively. Adetailed protocol on the procedures for DNA isolation and fingerprinting is available from one of us (J.D.A.V.).

on July 24, 2020 by guesthttp://jcm

.asm.org/

Dow

nloaded from

NOTES 409

electrophoresis, blotting, and hybridization, provided thatthe resolving power of the electrophoresis procedure iswithin the DNA fragment size range of 0.9 to 10 kb.

Dick van Soolingen and Petra de Haas are acknowledged for ideasand technical assistance. Janetta Top and Wilma van der Roest areacknowledged for secretarial help.

This study was financially supported by the World Health Orga-nization Program for Vaccine Development and the EuropeanCommunity Program on Science Technology and Development.

REFERENCES1. Cave, M. D., K. D. Eisenach, P. F. McDermott, J. H. Bates, and

J. T. Crawford. 1991. IS6110: conservation of sequence in theMycobacterinum tuberculosis complex and its utilization in DNAfingerprinting. Mol. Cell. Probes 5:73-80.

2. Daley, C. L., P. M. Small, G. F. Schecter, G. K. Schoolnik, R. A.McAdam, W. R. Jacobs, Jr., and P. C. Hopeweli. 1992. Anoutbreak of tuberculosis with accelerated progression amongpersons infected with the human imnunodeficiency virus: ananalysis using restriction fragment length polymorphisms. N.Engl. J. Med. 326:231-235.

2a.de Haas, P. Unpublished data.3. Edlin, B. R., J. I. Tokars, M. H. Grieco, J. T. Crawford, J.

Williams, E. M. Sordillo, K. R Ong, J. 0. Kilburn, S. W.Dooley, K. G. Castro, W. R. Jarvis, and S. D. Holmberg. 1992.An outbreak of multidrug-resistant tuberculosis among hospital-ized patients with the acquired immunodeficiency syndrome. N.Engl. J. Med. 326:1514-1521.

4. Eisenach, K. D., J. T. Crawford, and J. H. Bates. 1988.Repetitive DNA sequences as probes for Mycobacterium tuber-culosis. J. Clin. Microbiol. 26:2240-2245.

5. Fomukong, N. G., J. W. Dale, T. W. Osborn, and J. M. Grange.1992. Use of gene probes based on the insertion sequence IS986to differentiate between BCG vaccine strains. J. Appl. Micro-biol. 72:126-133.

6. Hermans, P. W. M., D. van Soolingen, E. M. Bik, P. E. W. deHaas, J. W. Dale, and J. D. A. van Embden. 1991. The insertionelement IS987 from M. bovis BCG is located in a hot spotintegration region for insertion elements in M. tuberculosiscomplex strains. Infect. Immun. 59:2695-2705.

7. Hermans, P. W. M., D. van Soolingen, J. W. Dale, A. R.Schuitema, R. A. McAdam, D. Catty, and J. D. A. van Embden.1990. Insertion element IS986 from Mycobacterium tuberculo-sis: a useful tool for diagnosis and epidemiology of tuberculosis.J. Clin. Microbiol. 28:2051-2058.

8. Mazurek, G. H., M. D. Cave, K. D. Eisenach, R. J. Wallace, Jr.,J. H. Bates, and J. T. Crawford. 1991. Chromosomal DNAfingerprint patterns produced with IS6110 as strain specificmarkers for epidemiologic study of tuberculosis. J. Clin. Micro-biol. 29:2030-2033.

9. McAdam, R A., P. W. M. Hermans, D. van Soolingen, Z. F.Zainuddin, D. Catty, J. D. A. van Embden, and J. W. Dale. 1990.Characterization of a Mycobacterium tuberculosis insertionsequence belonging to the IS3 family. Mol. Microbiol. 4:1607-1613.

10. Otal, I., C. Martin, V. Vincent-LUvy-Fr6bault, D. Thierry, andB. Gicquel. 1991. Restriction fragment length polymorphismanalysis using IS6110 as an epidemiological marker in tubercu-losis. J. Clin. Microbiol. 29:1252-1254.

11. Thierry, D., A. Brisson-Nod, V. Vincent-Levy-FrWbault, S.Nguyen, J. Guesdon, and B. Gicquel. 1990. Characterization of aMycobactenum tuberculosis insertion sequence, IS6110, and itsapplication in diagnosis. J. Clin. Microbiol. 28:2668-2673.

12. Thierry, D., M. D. Cave, K. D. Eisenach, J. T. Crawford, J. H.Bates, B. Gicquel, and J. L. Guesdon. 1990. IS6110, an IS-likeelement of Mycobacterusm tuberculosis complex. Nucleic Ac-ids Res. 18:188.

13. Van Soolingen, D., P. W. M. Hermans, P. E. W. de Haas, D. R.Soll, and J. D. A. van Embden. 1991. The occurrence andstability of insertion sequences in Mycobacteinum tuberculosiscomplex strains; evaluation of IS-dependent DNA polymor-phism as a tool in the epidemiology of tuberculosis. J. Clin.Microbiol. 29:2578-2586.

14. Zainuddin, Z. F. 1988. Mycobacterial plasmids and relatedDNA sequences. Ph.D. thesis. University of Surrey, Surrey,United Kingdom.

15. Zainuddin, Z. F., and J. W. Dale. 1989. Polymorphic repetitiveDNA sequences in Mycobacterium tuberculosis detected with agene probe from a Mycobacterinum fortum plasmid. J. Gen.Microbiol. 135:2347-2355.

VOL. 31, 1993

on July 24, 2020 by guesthttp://jcm

.asm.org/

Dow

nloaded from

LETTERS TO THE EDITOR 1959

Usefulness of DNA Fingerprinting in Combating Tuberculosis

van Embden et al. have proposed DNA fingerprinting ofall Mycobacterium tuberculosis strains to perform globalepidemiological studies on tuberculosis (1). With a view to aDNA pattern library, restriction fragment length polymor-phism (RFLP) analysis could make it possible to gain impor-tant insights into the global transmission of tuberculosis. Wedo believe that these new typing techniques can be veryuseful in clearing up epidemics and thus should be madeavailable to public health officers as a valuable tool in thefight against tuberculosis. On the other hand, detection ofchains of infection will be of no practical consequence in themajority of cases, because of the long time span betweeninfection and clinical manifestation of this disease in mostafflicted patients. We have subjected 36 strains of M. tuber-culosis isolated in 1992 from 31 Austrian patients to RFLPanalysis with PvuII and an INS-986 probe. All but twoisolates were found to have unique patterns. Our investiga-tion revealed no dominating strains except in the case of amarried couple; there, however, the chain of infection ap-peared to be obvious even without molecular epidemiologi-cal techniques (due to the temporal association of the onsetof disease). The renewed rise in the incidence of tuberculosisand the increasing spread of multidrug-resistant strains ofM.tuberculosis, which are due to lack of compliance withprescribed regimens and poor social conditions, call forintensified efforts by the public health sector to contain thisendemic infectious disease. In the fight against tuberculosis,it is still essential to guarantee that the necessary diagnosticand therapeutic means are available for all patients withtuberculosis; strain identification ofM. tuberculosis by DNAfingerprinting will have only limited practical importance inthis context.

REFERENCE1. van Embden, J. D. A., M. D. Cave, J. T. Crawford, J. W. Dale,

K. D. Eisenach, B. Gicquel, P. Hermans, C. Martin, R. McAdam,T. M. Shinnick, and P. M. Small. 1993. Strain identification ofMycobactenum tuberculosis by DNA fingerprinting: recommen-dations for a standardized methodology. J. Clin. Microbiol.31:406-409.

Franz AllerbergerWerner VogetsederManfred FilleManfred DierichFederal Public Health LaboratorySchoepfstrasse 41A-6020 InnsbruckAustria

Authors' ReplyIn their comment on the recommendations for a standard-

ized method of fingerprinting Mycobacterium tuberculosis,Allerberger et al. state that the identification of sources ofinfection by DNA typing will have no practical value in themajority of cases, because of the long time span between

infection and clinical manifestations of the disease. How-ever, it is known that in the vast majority of individuals whodevelop disease, the incubation period is less than 2 yearsand often only weeks to a few months. Styblo (1) showedthat 42% of the smear-positive patients had been coughingfor 1 to 6 months. Therefore, it is to be expected that byrapidly finding strains with identical fingerprints, the puta-tive sources of infection can be tracked more reliably andeasily than by the present day practice of contact tracingusing skin testing with tuberculin. Recently, various undiag-nosed individuals with pulmonary tuberculosis were recog-nized as the source of infection by using contact tracing;these sources could have been identified months before theresults of the contact tracing studies were available if finger-print results had been available more promptly. The propercontainment of such sources is of critical importance in thelong-term control of tuberculosis.Recent observations by various laboratories have demon-

strated the usefulness of fingerprinting in the epidemiologyof tuberculosis; many small epidemics, nosocomical infec-tions, and outbreaks of drug resistant M. tuberculosis havebeen traced by DNA fingerprinting alone. The fingerprintingof a mere 36 strains from 31 Austrian patients as describedby Allerberger et al. is not the way to draw conclusionsabout the usefulness of the establishment of fingerprintlibraries. The great diversity among these 36 isolates is to beexpected from a country in which there is a low incidence oftuberculosis. If more strains had been fingerprinted over alonger time span, the authors likely would have been able totrace epidemiological relationships among previously unex-pected epidemiologically related cases. In The Netherlands,such outbreaks have been established by fingerprintingalone, even without systematic typing of all isolates. Exam-ples of sources of infections indentified in this mannerincluded a pub, bronchoscopes, a discotheque, and anAustrian bull that was imported into The Netherlands lastyear. This last source caused an outbreak in cattle on variousfarms, the first such outbreak since bovine tuberculosis waseradicated in the Netherlands 25 years ago.

REFERENCE

1. Styblo, K. 1991. Epidemiology of tuberculosis, vol. 24, p. 55-70.Royal Netherlands Tuberculosis Association.

J. D. A. van EmbdenDick van SoolingenPetra E. W. de HassUnit Molecular Microbiology and LaboratoryofBacteriology and Antimicrobial Agents

National Institute ofPublic Health andEnvironmental Protection

P.O. Boax 13720BA BilthovenThe Netherlands

VOL. 31, 1993