THE BIvLLETIN

SECTION ON MICROBIOLOGYJANUARY 18, 1950

STUDIES ON VIRULENCE OF TUBERCLE BACILLI

a. Experimental analysis of the virulence oftubercle bacilliGardner Middlebrook (by invitation)The Rockefeller Institute for MedicalResearch

b. A component of tubercle bacilli con-

cerned in their virulenceHubert Bloch (by invitation)The Public Health Research Institute

of the City of New York, Inc.c. The application of virulence studies to

the BCG problemRend J. DubosCynthia H. Pierce (by invitation)Emanuel W. Suter (by invitation)The Rockefeller Institute for MedicalResearch

FRANK L. HORSFALL, JR.Chairman

HARRY MOSTSecretary

The Experimental Analysis of Virulence of Tubercle Bacilli

GARDNER MIDDLEBROOK*

The purpose of this symposium is to ac-

quaint you with recent experimental ob-servations concerning the virulence of tu-bercle bacilli and to summarize the theoret-ical as well as the experimental backgroundon which these observations have been based.It will be evident, I am sure, from Dr.Dubos' talk, that the problem with which we

are concerning ourselves tonight has imme-diate implications for problems of vaccina-tion.The background of these studies may be

of as much interest as the studies them-selves. They are based upon an explicit, ra-

tional procedure for studying infectiousdisease in terms of the virulence of the

pathogen. This will be the central theme of

* From the Laboratories of The Rockefeller In-stitute for Medical Research, New York City.

my talk.Let us first define our terms."The Nature of Virulence" was the title

given to a recent symposium.' An examina-tion of the communications abstracted fromthat symposium reveals the general state ofconfusion concerning the use of the termvirulence in microbiology. The problem is, Ithink, a semantic one, based upon the fal-lacy of accepting a word as something mag-

ically born from the head of Zeus and oflooking for a concrete thing or phenome-non with which to identify the word. Inhighly evolved fields of scientific inquirythis form of question: "What is the Natureof Something?" and the resulting contro-versies which only confound understandingrarely appear. It is understood that suchproblems can be resolved by the employment

498

Section on Microbiology

of arbitrary definitions which are only re-

quired to be useful as means of communica-tion in operational terms.2Inasmuch as it has not been possible for

us to find in the literature generally accept-ed and useful definitions of the terms viru-lence and pathogenicity, it appears essentialto formulate clear-cut definitions for theseterms before discussing tonight's topics.

I propose in this talk to limit the termpathogenicity, as applied to living agents,to the following meaning: pathogenicity isthat general attribute of a strain of micro-

organism or general class (usually a sub-species) of microorganism which character-izes it as being able under a specified set ofconditions to produce disease by multiply-ing in or on some living host. Disease isunderstood, of course, to mean either obvi-ous death of tissues or gross impairment ofbiological function. A strain or class ofmicroorganisms may be said, in generalterms, to be pathogenic if it has pathogen-icity for any host, and, conversely, a strainor class may be said to be non-pathogenic(i.e., saprophytic or parasitic but non-patho-genic), if it has no known pathogenicity forany host. And also, a strain or class whichis pathogenic for one host may be said to bespecifically non-pathogenic for another.Pathogenicity concerns the relations of para-

site A and hosts x, y, z, etc.By virulence I mean that measurable

property of a particular strain of micro-organism by which it can be distinguishedquantitatively from other similar strains ofthe same class, or from "variants" of thesame strain, in respect to pathogenicity forone specific susceptible host. Thus, in con-

trast to pathogenicity, virulence concerns

comparative relations of similar strains A,B, C, etc. and host x. It implies degree ofpathogenicity of a specific strain for a par-

ticular host in comparison with similarstrains of the same class for the same host,tested under the same set of experimentalconditions. Thus, variations in virulence im-ply inheritable variations in the strain ofmicroorganism such as occur in so-calleddissociation and which are probably geneticin character. These variations ideally in-volve identical variations in every individ-ual cell in the strain as tested in the same

host, without differences in the numbers ofliving microorgariisms introduced into thetest host or other changes in the experi-mental conditions, such as growth mediumor ages of the cultures of the strain com-

pared. In summary then, the term "viru-lence" is equivalent to the phrase "degree ofpathogenicity for a specified host." It is, ofcourse, imperative for an understanding ofthe concepts to be entertained here thatvariations in the host, though they are em-

phatically of as great import for a generalview of host-parasite relationships as varia-tions in the parasite, be considered as hav-ing no relevance to virulence: a constant,specific, susceptible host is an essential as-

sumption in any problem of virulence. Var-iations in the host being well controlled or,

in the words of L. J. Henderson, "otherthings being equal (which they never are

in practice)," virulence can be fruitfullystudied.

Needless to say here, it is usually possibleto control all such variables by employinga statistically significant number of indi-vidual hosts in the testing of pathogenicstrains, and it will be seen that the termvirulence defined in these respects has use-

fulness at certain levels of analysis of many

types of host-parasite relationship.I should remind you that no single prop-

erty of a parasite can be considered a

"cause of virulence." Obviously many prop-

erties, few of which are ever actually iden-tified, are essential for the manifestationof virulence by a microorganism. Thosestrains of pneumococcus which synthesizewell-developed capsules and yet are in-capable of causing disease serve as theclassical illustration of the fact that no

one property or component of a parasite issufficient for the manifestation of viru-lence, though it may be essential. This leadsus to the scheme of analysis of infectiousdisease which served as a background forthe work to be described tonight.

A CONCEPTUAL SCHEME OF EXPERIMENTAL

ANALYSIS

The succession of discoveries that immu-nization of animals with water soluble toxicproducts of pathogenic diphtheria, tetanus,and certain (erythrogenic toxin producing)

499

50 H BLEI

streptococcus cultures could confer high de-grees of acquired resistance against thediseases caused by those microbes led toalmost tireless search for similar poisonoussubstances in the artificial cultures of otherpathogenic microbes with the view to in-ducing a similar state of immunity againstthem. Then interest was revived for theearly established use for vaccination of viru-lent, but specifically non-pathogenic, strains(cowpox vaccination against smallpox inhuman beings) and of living attenuated oravirulent strains of the same organism(Pasteur's use of attenuated anthrax cul-tures). Calmette's experiments with BCGemphasized the earlier observations.These accents in the practical field of

immunization against microbial infectionssoon echoed and now resound in the labora-tory of the microbiologist who became in-terested in variation in microorganisms asa theoretical study in general biology with-out any necessary relation to the patho-genesis of infections. Yet there appearedduring the past thirty years studies of therelationship of the capsular polysaccharidesof pneumococci to immunization against,and the manifestation of virulence in, thisclass of pathogen. These investigations es-tablished several important points pertinentto our discussion: the manifestation of viru-lence does not necessarily depend only uponthe production by a microorganism of asubstance with lethal toxicity, because, al-though the pneumococcus might produce atoxic substance as it multiplied in vivo, theloss of its ability by dissociation (geneticvariation, in all likelihood) to produce anapparently non-toxic surface polysaccha-ride, inability to accumulate a sufficientamount of the substance at the surface ofeach bacterial cell, or combination of thismaterial with specific antibody, all renderthe pathogen unable to manifest virulence.Another fact relevant to this review is theobservation that there are strains of pneu-mococci which are apparently avirulent andyet capable of inducing upon injection intoanimals high degrees of acquired resistanceagainst virulent strains, because of theirproduction of small amounts of capsularpolysaccharide, sufficient to elicit type-spe-cific "protective" antibody, but insufficient

to allow a clearly detectable manifestationof virulence. In other words, variation andvirulence are not necessarily of an all ornone nature.Analogous to this information concerning

the pneumococcus are the more limited ob-servations made in studies of other classesof bacterial pathogens. These facts have, onthe one hand, given the experimental basisfor a new interpretation of Pasteur's andsubsequent use of "avirulent" variants forimmunization against virulent strains. Onthe other hand, their pertinence to the prob-lem of the relation of properties of thebacterial cell to the pathogenesis of diseaseis further established by the following cor-relative generalization for which there arethus far no recognized exceptions: antibod-ies (serum globulins) which are protectiveagainst infectious disease are always di-rected against antigenic components of themicrobial cells which are essential in somerespect or another for the manifestation ofvirulence; for when these antigens, andthese antigens alone, are lost in dissociation(hereditary variation), virulence also islost.With a broad view of this wealth of ex-

perimental observations, Dubos3 was led tostate that it is not essential to exert a non-specific, lethal effect on a pathogenic mi-croorganism in the treatment or preventionof infectious disease, but that it is merelynecessary to interfere with the operationof only one of several properties of theparasite essential for its pathogenic career.Carrying this conceptual scheme further, tothe point of explicit formulation of a ra-tional method of analyzing infectious dis-ease, appeared to offer practical as well astheoretically significant results.The operational steps in this organized

method presented themselves as follows:1. Obtain a relatively stable. virulent

strain of the pathogenic parasite.2. Select by the use of any of the many

techniques now available (radiation treat-ment, ageing of cultures, dissociation bycolony isolation, heterologous animal pas-sage, etc.) variant cultures of the parentpathogen which have lost their virulence(for the host with which one is concerned).

3. Compare the avirulent variant strains

5 0 0 THE BULLETIN

Section on Microbiology

(dissociants) with the parent virulent strainby techniques which will allow quantitativeas well as qualitative analysis in an effortto identify at least one physiological change(immunochemical, biochemical, etc.) whichhas taken place simultaneously with thechange in virulence. This identification of a

virulence property is assured by the dem-onstration that, on the one hand, the prop-

erty is possessed by all virulent strains ofthe same class and that, on the other hand,those variant strains which lack it are

avirulent.4. Analyze the pathogenesis of the disease

caused by the pathogen in terms of thisspecific difference between virulent andavirulent variant strains, or in terms of thespecial property of the virulent variants.

5. And, finally, depending upon the na-

ture of this difference and its particularrole in the pathogenesis of the disease, aimat the prevention* and treatment of the dis-ease in terms of this difference.Examination of the literature on varia-

tion and virulence of bacteria, fungi andprotozoa reveals the following three gener-

alizations.First, independent variations in specific

properties of strains of microorganismswhich reproduce asexually occur much more

often than simultaneous variations in severalproperties. Were this not the case, it can

be imagined, how inappropriate and irra-tional the above scheme would be. Second,marked and easily recognizable variationsin certain properties (antigenic, biochem-ical, etc.) of bacterial strains may occur

without change in virulence. And third, themanifestation of virulence, as emphasizedabove, depends upon the operation of many

properties of any one microorganism; thus,there is usually a wide choice as to whatessential property is to be experimentallypursued-which, of course, can be decidedonly in terms of the special purpose of thepursuit. From the practical standpoint ofprophylaxis or therapy of infectious dis-ease, a property which promises to be mostamenable to attack with the well-established

* Ecological and epidemiological considerations inthe prevention of disease are purposely avoidedhere as they have no essential relevancy at thislimited level of analysis of host-pathogen rela-tionships.

methods would naturally be selected firstfor thorough investigation. It is perhapsnot necessary to state that limitation ofendeavor at this level, without a search fornew methods suggested by the nature ofthe role played by the essential property inthe host-parasite relationship, would leadto undesirable stagnation.Although the plan of experimental analy-

sis just described may seem obvious-simpleand logical-it must be stressed that in allinstances where effective methods of im-munization have been developed, successfulimmunization has been achieved before ac-

tual identification of any of the propertiesof the pathogen which are essential for themanifestation of virulence. In point of fact,successful immunization has itself often al-lowed elucidation of the properties of thepathogen essential for virulence. Althoughthere has been a general awareness of thesignificance of virulence studies for prob-lems of immunization, this rationally for-mulated plan had not been explicitly em-

ployed in the analysis of any infectiousdisease before its use in the studies of tu-berculosis which we are reviewing tonight.The direct application of the steps of

analysis based upon the theoretical schemewhich I have presented is straightforwardfor the study of host-pathogen relationshipsin which the pathogen is a microorganismreadily cultivatable in vitro. An avirulentvariant can often, in such instances, also begrown in vitro. It is at this level that sucha study is to be initiated until evidence in-dicates that the significant difference be-tween the virulent and avirulent variants ismanifest only under conditions for mul-tiplication existing in vivo and not yet imi-table in vitro. On the other hand, in thecase of virus and rickettsial infections theapplication of this conceptual scheme wouldbe necessarily more challenging experimen-tally but perhaps approachable even at thepresent time in those instances in whichmarked loss of virulence occurs on hetero-logous host passage in the case of plant as

well as animal viruses. More thorough dis-cussion here on these points would be outof place, but the pertinency of this methodof analysis to the several types of host-pathogen relations is plain.

5 01

502 THE BULLETIN

VIRULENCE OF TUBERCLE BACILLI

AND TUBERCULOSIS

Although it is nearly a quarter of a

century since the pioneer work of Petroffand his associates4 on dissociation of tu-bercle bacilli with the isolation of "pure"avirulent variant strains of mammalian (andavian) types, extension of these studies hasbeen slow. More recently, however, the grad-ual emergence and explicit formulation ofthe conceptual scheme of relations betweenthe virulence of pathogens and the patho-genesis of infectious processes, as outlinedabove, has stimulated novel investigationsof tuberculous infection and disease.The first and second steps of the analysis

had been accomplished by several groups ofinvestigators subsequent to Petroff's earlyobservations, but many problems presentedthemselves relating to techniques of cultiva-tion, standardization and control of infec-tive inocula, the use of that convenient ex-

perimental host, the mouse, the evaluationof the mouse as a host in the study of thevirulence of the parasite, and the probabil-ity that observations on that host wouldreveal significant information relating viru-lence and the disease, tuberculosis, as itoccurs naturally in other hosts, especiallyin man.

Certain biological properties of mamma-

lian type (human and bovine) tuberclebacilli, it soon became evident, made themsuitable for virulence studies. Most strainswere easily cultivated in vitro, both the viru-lent and the avirulent variants, especiallyin the media newly devised by Dr. Dubosand his associates. Many strains were quitestable with respect to virulence, and virulent(H37Rv) and avirulent (H37Ra) variantsof the same well-known strain appeared tobe especially appropriate pure cultures withwhich to begin the third step of analysis.'The pathology of the disease, althoughmarkedly different in different hosts, didnot vary qualitatively in its several aspectsin the mouse any more than in the guineapig or other hosts with different strains ofthe same subspecies of tubercle bacilli (e.g.,human type), provided that equal doses andthe same route were used for infection.Various techniques of testing virulence,such as using different routes of inocula-

tion, did not reveal different orders of viru-lence of several strains of the same sub-species of tubercle bacilli in the mouse; andthe order of virulence of strains of thesame subspecies appeared to be approxi-mately the same in races of mice whichdiffered widely with respect to susceptibil-ity; this order, furthermore, proved to beidentical with that which had been observedfor the same strains in the guinea pig and,probably, in man.6 This latter circumstancefavored the possibility that the propertiesof human type tubercle bacilli which condi-tion virulence are, at least qualitatively,the same for many hosts.Two additional important problems relat-

ing to the pathogenesis of tuberculosis wereconsidered in these investigations. One isconcerned with the possible direct relevanceto virulence of the hypersensitivity to tu-berculin which develops during the courseof the disease in many susceptible hosts. Itwas clear that, although hypersensitivity cer-tainly plays a role in the pathology-onemay say, the pathogenesis-of tuberculosisin certain hosts, including man, it is notinvolved in the variations in virulence oftubercle bacilli with which we have becomeacquainted. This was demonstrated by sev-eral facts:

1. The virulent and avirulent variantswith which we are acquainted are equallyactive producers of tuberculin in vitro.7

2. Both, when killed, are effective in in-ducing the state of hypersensitivity totuberculin.

3. The avirulent variants with which weare concerned are not capable of establish-ing progressive disease in animals whichhave been previously rendered hypersensi-tive.

4. The bacilli of the completely avirulentvariant strains multiply little, if at all,even within the first few days after inocu-lation into susceptible hosts, in contrast tothe unharnessed multiplication of virulenttubercle bacilli-that is, before hypersen-sitivity appears.8 In this connection I wouldemphasize another point of importance inthe differential properties of our virulentand avirulent variants. They behave differ-ently in the susceptible host even beforehistologic lesions are recognizably charac-

THE BULLETIN5 O 2

Section on Microbiology 5 0 3

iH37 Rv Ii37Ra B3C0Vivulent Avivulent Attenuated

Fig. 1H37Rv (virulent) and H37Ra (avirulent) X97

Fig. 2BCG 317 (attenuated) X97

These photomicrographs illustrate the colonial morphological differences between typical strains ofmammalian tubercle bacilli possessing different degrees of virulence.

Fig. 1 shows two colonies growing side by side on the surface of an agar medium (oleic acid-albumin medium). To this medium was added before inoculation 0.004% of a non-ionic, non-toxicwetting agent, Triton A20, which, it has been shown (Dubos, R. J. and Middlebrook, G. J. Exper. Med.,1948, 88; 81), disperses the bacilli of non-cord-forming, avirulent variants of mammalian strains, butdoes not inhibit the cord formation of virulent strains. Clearly evident are the serpentine pattern of thecolony of the virulent strain and the non-oriented, almost "smooth" structure of the avirulent variantstrain's colony.

Fig. 2 shows the morphological appearance of a colony of an attenuated but not completely avirulentBCG strain (317); it is evidently intermediate in structure between the fully virulent strain and thenon-cord-forming avirulent strain illustrated in Fig. 1.

The use of the lipophilic wetting agent, Triton A20, facilitates visualization of colony differenceswhich are, nonetheless, visible on the same medium without this adjuvant.

teristic of progressive tuberculous infection.This indicates that at least one of the mostimportant properties of the virulent patho-gen in the host-pathogen relationship intuberculosis manifests itself before the es-tablishment of the characteristic anatomicallesion. This assumption, it will become evi-dent to you this evening, greatly influencedthe types of questions which we asked andthe particular experiments which were

performed.The other important problem which we

faced was the question as to whether thesignificant biological differences betweenvirulent and avirulent variants would bemanifest as they were cultivated in vitro.An abundance of previous observations, itis true, had shown that, unlike acquiredresistance in anthrax and perhaps in someother infectious diseases, there is no con-

vincing evidence that actual infection withvirulent or attenuated tubercle bacilli con-fers a type of acquired resistance againsttuberculous infection or disease which isqualitatively different from immunity con-ferred by virulent tubercle bacilli grownin vitro and killed by various techniques.Certainly then, tubercle bacilli, multiplyingin vitro, do produce at least some of theantigenic chemical components which theyproduce in vivo-as far as protective anti-gens are concerned(. There was, however, noassurance that the differences between ourvirulent and avirulent variants would haveany bearing on acquired resistance as weknow it in tuberculosis.At any rate, with purposeful disregard of

hypersensitivity at this stage of analysis,and confidence, perhaps insufficiently as-sured, in artificial cultures of tubercle ba-

504 THE BULLETIN

Representation of Effect of "Tween 80" on Colony Morphologyof Typical Mammalian Strains

SCHEMA

(Acknowledgment: From The Journal of Experimental Medicine, Vol. 82, p. 181, 1947.)

cilli, study was aimed at the identificationof at least one property of the virulentvariant of the H37 strain which was con-sistently lacking in the avirulent variant ofthe same strain.

It was soon observed9 that when the twostrains, H37Rv (virulent) and H37Ra(avirulent), were grown on the surface orin the depth of many types of artificialmedia, there was a consistent microscopical-ly demonstrable difference between the twostrains in their morphological properties.The avirulent variant, H37Ra, was ob-served to present a non-oriented micro-scopical arrangement of bacilli in clumps.On the contrary, the virulent variant,H37Rv, grew in an impressively orientedpattern of what I call "cords" of bacilli,

in which the rods arrange themselves inparallel along the long axis of the cord. Itis interesting that Robert Koch,10 in one ofhis first papers on the cause of tuberculosisobserved, and even pictured, the cord pat-tern of growth of virulent tubercle bacilli.Needless to say, however, he did not attachany special significance to it (Figure 1).As I have pointed out earlier, a property

significant for virulence must satisfy thefollowing requirements: it must be pos-sessed by all the virulent strains of thesame class; and, all strains which lack itmust be avirulent.

Therefore, extension of these observationson morphology was made to other eugonicstrains of tubercle bacilli, and thus far, ithas invariably been noted that those var-

THE BULLETIN5 0 4

Microbialogy 505

iants of eugonic mammalian strains whichfail to form cords are avirulent, and thatall virulent strains form cords. In addition,the BCG strains of which Dr. Dubos willspeak, afford examples of the intermediategroup of tubercle bacilli-intermediate bothin respect to virulence and to cord forma-tion (Figure 2). They form an intriguinggroup with which much work remains to bedone, especially with respect to the refine-ment of quantitative methods for testingvirulence. I venture the opinion that theBCG group of strains of mammalian tu-bercle bacilli,,as represented by the cultureswhich I have studied, is analogous to thosepneumococcus variants, mentioned before,which produce small amounts of capsularpolysaccharide adequate to allow them tomanifest a slight virulence in susceptiblehosts, but insufficient to allow them to causesevere infection or disease.A correlation was thus established be-

tween the virulence of mammalian tuberclebacilli and their tendency to orient them-selves according to a particular patternduring multiplication in many types ofartificial media. Although this correlationgives no direct information at the physio-logical level which could be of immediatevalue in the analysis of the pathogenesis ofthe infection, it did give morphological evi-dence for the opinion that some unidentifiedstructural component of the virulent tu-bercle bacillus plays a dominant, essentialrole in its ability to manifest virulence.Indeed, more than that, the fact that thishypothetical component affects the mode ofdivision and orientation of the bacilli dur-ing growth gave some confidence that thesubstance is at, or very close to, the surfaceof each bacterial cell.Now let us consider the effects of the

surface active agents which Dr. Dubosfound useful for obtaining dispersed growthof tubercle bacilli in artificial media. Theyhave effects of some interest on the colonialmorphology of cultures. As illustrated inthe accompanying schema, the wetting agentTween 80 can, in sufficient concentration,render colonies of all strains of tu-bercle bacilli indistinguishable. They appearsmooth, glistening, and hemispherical onagar, and in the depth of liquid media they

show no evidence of cord formation-thebacilli are, for the most part, isolated onefrom another in dispersed state. Whengrown in the presence of a somewhat lowerconcentration of Tween 80, the virulent or-ganisms show evidences of cord formationwhile the somewhat attenuated BCG strainand the avirulent strain show no evidenceof this pattern of orientation.'

Virulent tubercle bacilli grown in thepresence of concentrations of Tween 80which entirely inhibit cord formation are noless infective for experimental animals. Itwas presumed, therefore, that the Tweenwhich is rapidly hydrolyzed by animal tis-sues and fluids, as shown by Dr. Davis,merely masks some component of the surfaceof the virulent bacterial cells rather thanpreventing its synthesis or removing it fromthe surface of the bacterial cells. The lipo-philic character of the Tween 80 moleculesuggested further that the hypothetical sub-stance directly responsible for cord forma-tion might be, shall I say, a lipoidal sub-stance. The implications of this hypothesiswill be made more evident by Dr. Bloch.Of equally great import for the third

step of analysis is the discovery by Dr.Dubos of a striking cytochemical reactionwhich clearly distinguishes the cord-forming,virulent strains of tubercle bacilli from thenon-cord-forming, avirulent variants.", 12This reaction involves the adsorption by thecord-forming strains of the dye neutral redat alkaline pH. In aqueous solution at alka-line reaction neutral red is yellow in color;it is red at acid reactions. The cord-formingstrains of tubercle bacilli living or killedwith phenol are stained red by the dyewhile the dye in the surrounding fluid isyellow; the non-cord-forming strains arestained yellow under these conditions. It isknown that this reaction is given by severalacidic lipoidal substances-long-chain fattyacids, in particular. Although this reaction,therefore, is not chemically specific in thesense of an immunological reaction, it didemphasize the more clearly that the chemicalstructure of the surface of all virulentstrains of mammalian tubercle bacilli differsin a special way from that of the non-cord-forming avirulent strains of the same sub-species of organisms.

Section on Microbiology 5 0 5

5 0 6 THE BULLETIN

It seems possible that further applicationof the scheme of analysis which I outlinedearlier will vield fruits of significance forour understanding of pathogenesis and im-munity in tuberculosis.

R E FE R E N C E S

1. Nature of virulence, a symposium atthe Fourth General Meeting of the So-ciety for General Microbiology, J. Gen.Microbiol., 1947, 1:i.

2. Bridgman, P. W. The logic of modermphysics. New York, Macmillan Co.,1927, chapt. I.

3. Dubos, R. J. Antimicrobial agents ofbiologic origin, J.A.M.A., 1944, 124:633.

4. Petroff, S. A. and Steenken, W., Jr.Biological studies of the tubercle bacillus; instability of the organism-mi-crobic dissociation, J. Exper. Med.,1930, 51:831.

5. Steenken, W., Jr. and Gardner, L. U.History of H37 strain of tuberclebacillus, Am. Rev. Tuberc., 1946, 54:62.

6. Pierce, C. H., Dubos, R. J. and Middle-brook, G. Infection of mice with maiP-malian tubercle bacilli grown in Tween-

albumin liquid medium, J. Exper. Med.,1947, 8.2:149.

7. Steenken, W., Jr. The isolation of puri-fied protein derivatives and carbohy-drates from tuberculin and their bio-logical properties, J. Biol. Chem., 1941,141:91.

8. Pierce, C. H., and Middlebrook, G. Unt-published observations.

9. Middlebrook, G., Dubos, R. J. andPierce, C. H. Virulence and morpho-logical characteristics of mammalian tu-bercle bacilli, J. Exper. Med., 1947, 82:175.

10. Koch, R. Die Aetiologie der Tuberku-lose, Mittl. a. d. k. Gsaidtsamte. Re-printed in: Koch, R. Gesamimelte lVerke,Leipzig, George Tlhieniie, 1912, v. 1,p. 467.

11. Dubos, R. J. and Middlebrook, G.Cytochemical reaction of virulent tu-bercle bacilli, Am. Rev. Tubere., 1948,58:698.

12. Hauduroy, P. and Postenak, Y. Surune reaction permettant de distinguerles mycobacteries virulentes des myco-bacteries avirulentes, Comnpt. rend.Acad. sc., 1949, 228:781.

A Component of Tubercle Bacilli Concerned 'with

Their Virulence (ABSTRACT)

HUBERT BLOCHThe Public Health Research Institute of the City of New York, Inc.

The treatment of virulent tubercle bacilliwitlh paraffin oil or lighter hydrocarbons re-sults in the disruption of the bacillary"cords," a growth pattern characteristic forvirulent strains of tubercle bacilli. In thesame time a lipid material is obtained fromyoung virulent bacilli extracted with thesehydrocarbons. The yield is small and nosuch material can be extracted from aviru-lent variants of tubercle bacilli. Extractingthe bacilli with paraffin oil, petroleum etheror similar compounds does not kill the or-

ganisms; after extraction, they are stillable to grow on culture media. However,their biological behavior is changed. They dono longer inhibit the migration of leuko-cytes in vitro, and their state of pathogen-icity for experimental animals is reduced.The lipid material obtained by extractingthe cells exerts by itself, an inhibitory effecton leukocyte migration and is toxic for miceupon repeated injections of small doses. Thelipid material is located at the surface ofthe bacillarv cell, and extraction with petro-