BIOCHEMICAL DIFFERENTIATION OF MYCOBACTERIUM TUBERCULOSISGROWN IN VIVO AND IN VITRO" 2

WILLIAM SEGAL AND HUBERT BLOCH3Division of Tuberculosis, The Public Health Research Institute of The City of New York, Inc., New York 9,

New York

Received for publication November 17, 1955

Most of our knowledge concerning the tuberclebacillus stems from studies on bacteria grownunder standard laboratory conditions in vitro.Numerous attempts have been made to determineindirectly the relationship between the hostenvironment and the pathogenicity of the tuber-cle bacillus by a study of the comparative reac-tion of virulent and avirulent strains to variousconditions occurring in the host environment.Although some consistent correlations have beenmade, the results are open to question sinceinterpretation is based on the assumption thatvirulent tubercle bacilli grown artificially areidentical with bacilli as they multiply in thetissues of an infected animal. This assumption isimplicit in most of the work done on the tuberclebacillus. However, there have been isolatedreports in the literature indicating a lack ofidentity between in vitro and in vivo growntubercle bacilli. Anderson et al. (1943) wereunable to detect in tuberculous lung tissue severalchemical compounds (phthiocol, tuberculostearicacid, phthioic acid, and specific polysaccharide)that are characteristic constituents of humantubercle bacillus cultivated on artificial medium.These authors raised the question whether tuber-cle bacilli growing in living tissue produce thesame characteristic chemical compounds that arefound in bacilli grown in vitro. Sheehan andWhitwell (1949) concluded that a differenceexists between the lipid content of artificiallycultured tubercle bacilli and that of the organismsin pathological lesions on the basis of theirdifferential method of "sudan black B" staining.

I This investigation was supported in part by aresearch grant from the National Institutes ofHealth, U. S. Public Health Service, RG 4541.

2 Presented in part at the 55th General Meetingof the Society of American Bacteriologists held inNew York City, May 8-12, 1955.

3 Present address: Department of Microbiology,Medical School, University of Pittsburgh, Pitts-burgh, Pennsylvania.

Our understanding of the host-parasite rela-tionship in tuberculosis would be furthered by astudy of the tubercle bacillus in its naturalparasitic state, with particular reference to anydifferences that might exist between in vivo andin vitro grown bacilli. Such a study was madepossible by the development of a method forseparating tubercle bacilli in large numbersdirectly from the lungs of infected mice.The method of separating pathogen from its

diseased host has in recent years been applied tothe study of two other infectious bacteria,Mycobacterium lepraemurium and Bacillus anthra-cis. Hanks (1951a) separated the murine leprosybacillus from testicular tissue of the rat by amethod different from that reported here.Hanks (1951b) was able to obtain an estimateof the viability of this noncultivable organismby measurement of hydrogen transfer capacity.A study of the metabolism of separated murineleprosy bacilli was made by Gray (1952) as theonly available means for the study of the physio-logical requirements of this organism. The workof Hanks and Gray served as a valuable guidefor the present study. -More recently the use ofin vivo grown bacteria has been applied by Smithand Keppie (1955) to the anthrax bacillus withhighly successful results. Using bacilli separatedfrom exudates and plasma of diseased guineapigs, they were able to isolate and identifychemical substances involved directly in thepathogenicity of this organism. Their study didnot primarily concern itself with establishingdifferences between B. anthracis grown in vivo andin vitro; however, they did find that the in vivogrown bacilli were rapidly lysed in 0.16 per centammonium carbonate solution at 0 C in contrastto in vitro grown bacilli.

In the case of the study reported here on My-cobacterium tuberculosis, the primary objective ofexamining the comparability of in vivo and invitro grown bacilli was accomplished by anexamination of various biochemical characteris-

132

on October 17, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

BIOCHEMICAL DIFFERENTIATION OF M. TUBERCULOSIS

tics. The differences observed are striking andindicate that tubercle bacilli in their naturalinfectious state are dissimilar to those grown inartificial culture medium.

EXPERIMENTAL METHODS

Method for separation of in vivo grown tuberclebacilli. The source of in vivo grown tubercle bacilliwas the lungs of moribund or dead mice 18 to 23days after infection. Initially mice were infectedwith a virulent human strain of M. tuberculosis(H37Rv), but thereafter with tubercle bacilliseparated from infected lungs (LRv). The methodfor separation of tubercle bacilli from infectedlungs is based on the use of isotonic sucrose(0.25 M) for the isolation of particulate cell com-ponents by differential centrifugation (Umbreitet al., 1949). Lungs were homogenized in a solu-tion (5 ml per lungs of 1 mouse) of 0.25 M sucrosecontaining 0.25 per cent bovine albumin. Inclu-sion of 0.25 per cent albumin in the homogenizingmedium was found to stabilize the endogenousrespiration of the separated lung bacilli withoutappreciably affecting the nature and degree ofseparation. Thorough homogenization was ob-tained by use of a "virtis" homogenizer run attop speed (29,000 rpm) for 3 min. The entireseparation was carried out under sterile condi-tions and at near-freezing temperatures by meansof crushed ice baths. The homogenate wascentrifuged in 40-ml round-bottom tubes in anInternational refrigerated centrifuge at 0 C andat a speed of 2,000 rpm (450 G) for 15 min. Thesupernatant containing light tissue componentswas discarded. The sediment was resuspended inalbumin-sucrose solution (5 ml per original lungsof 1 mouse), centrifuged for 5 min at 1,500 rpm(250 G), and the supernatant removed to a 40-mlcentrifuge tube, which was stored on ice. Thislatter step was repeated twice, resulting in threeseparate supernatants. The residual sediment,consisting of heavier tissue components such asnuclei and some tubercle bacilli, was discarded.The three batches of supernatant were then cen-trifuged at 2,000 rpm (450 G) for 10 min, andthe supernatants were discarded to assure com-plete removal of light tissue components. Thethree resultant sediments were examined micro-scopically for purity of separation, using theZiehl-Neelsen acid-fast stain and prolongedcounterstaining with methylene blue. The threesediments were then combined in saline contain-

ing 0.25 per cent albumin and centrifuged at2,000 rpm (450 G) for 10 min; the sediment wasresuspended in albumin-saline and centrifugedat 3,000 rpm (1,000 G) for 10 min. The sedimentwas finally suspended in saline, stained and ex-amined microscopically for purity of separation,and adjusted to standard optical density using aColeman Universal spectrophotometer.Male albino mice of the CFl strain (Carworth

Farms, New City, New York) were injectedintravenously with 0.1 ml of saline suspensions ofLRv adjusted to 40 per cent transmittance, ap-proximately equal to 0.02 mg dry weight or 5million bacilli as determined by plate counts.Plate counts were made of colonies grown onoleic acid-albumin agar (Fenner, 1951).

Source of in vitro grown tubercle bacilli. Forcomparative purposes a 10-day "tween"-albuminsubmerged culture (Dubos and Middlebrook,1947) of the virulent human strain (H37Rv) ofM. tuberculosis was used. The cells were separatedfrom the culture medium by centrifugation,washed twice in a salt solution containing 0.05per cent "tween 80," and adjusted to standarddensity. It is of interest to note that the presenceof tween 80 is essential to obtain an even suspen-sion of H37Rv whereas the separated lung bacilliLRv suspend evenly in saline or distilled waterin the absence of tween 80.

Quantitative comparison of in vivo and in vitrogrown tubercle bacilli. Table 1 summarizes data onthe quantitative comparability of suspensions ofequal optical density of in vivo and in vitro growntubercle bacilli, as well as indicating the source,age, and yield. In order to compare the viablenumbers of LRv and H37Rv by plate counts,it was necessary to eliminate the cords and largeclumps of bacilli that occur in cultures of H37Rv.This was accomplished by centrifuging for 2min at 1,000 rpm (125 G) and discarding thesediment. The supernatant was found to containonly single bacilli and small aggregates of lessthan 30 bacilli, similar to those resulting in theseparation of lung bacilli. Quantitative estima-tion was obtained by adjustment of opticaldensity and determination of dry weight.As seen in table 1, the yield of tubercle bacilli

from infected lungs was considerable, supplyingadequate material for use in Warburg studies.Suspensions of H37Rv and LRv adjusted to thesame optical density were found to contain com-parable amounts of bacilli as determined by dry

13319561

on October 17, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

SEGAL AND BLOCH

TABLE 1Quantitative comparison of suspensions of in vitro and in vivo grown tubercle bacilli

Type Source and Age Avg Yield Dry Wt* Plate Count*

mg dry wI mg/1JO "l colonies/1O nil

H37Rv ....... 10-day Tween culture 8 per flask of 30 ml 5.6 3.2 X 109L,Rv ......... Lungs of moribund or dead mice 4 per lungs of 1 mouise 5.3 1.9 X 109

18 to 23 days after infectiont

* Both suspensions of H37Rv and LR.v were adjusted to 10 per cent transmittance in the ColemanUniversal Spectrophotometer.

t Initially mice were infected with H37Rv, but thereafter with LRv.

weight and plate counts on oleic acid-albuminagar. These results confirmed the purity of thesuspensions of lung bacilli and established a basisfor quantitative biochemical comparison oftween-grown and lung-grown tubercle bacilli.

Biochemical methods. An index of generalmetabolic activity was obtained by testing thehydrogen transfer capacity of bacilli in the pres-

ence of substrate. This was done by measuringin the Coleman Universal spectrophotometer thereduction of 2,3,5-triphenyl tetrazolium chlorideafter 2 hr of incubation in the presence of tween-albumin medium containing 0.5 per cent glucose(Vandiviere et al., 1952).The specific biochemical test used throughout

this study was the comparative response ofH37Rv and LRv to various substrates, as meas-

ured by oxygen consumption using standardWarburg manometry. The gas phase was air, thetemperature was 37.8 C, CO2 was absorbed by 0.2ml of 20 per cent KOH in the center well, and thetotal volume of liquid in the Warburg vessels was2.2 ml. The oxygen uptake of 5 to 6 mg dryweights of bacilli was determined over 2- or 3-hrperiods and from these values the Qo, was deter-mined. Endogenous respiration was determinedon bacilli suspended in M/15 phosphate buffer,pH 6.8, which also was the final pH of the varioussubstrates in the Warburg vessels. Comparativesubstrate response is expressed as per cent in-crease in the Qo, value over that of the endoge-nous respiration.

RESULTS

No significant differences were found to existwhen comparing in vivo and in vitro grown tuber-cle bacilli with respect to the following character-istics: morphological appearance, staining prop-

erties, colonial characteristics and the rate ofgrowth on oleic acid-albumin agar plates, and

TABLE 2Comparison of metabolic activity of H37Rv and LRv

Strain Dry wvt Hydrogen TransferCapacity*

Mg Per cent TtBlank.0 l 00H37Rv 2.8 29LRv.2.6 8(

* Substrate was complex, made ui) of t-een-albumin medium with 0.5 per cent glucose.

t Per cent transmittance of reduced 2,3,5-triphenyl tetrazolium chloride measured photo-metrically after 2-hr incubation.

pathogenicity for mice. It was, however, uponexamination of their biochemical state that strik-ing differences were revealed.A comparison of the general metabolic activity

of H37Rv and LRv, as measured by hydrogentransfer capacity, established a large differencebetween them (table 2). The reading of 80 percent transmittance shown by the lung bacillirepresents a very low state of metabolic activityparticularly in comparison to the reading of 29per cent for H37Rv. This result is of added in-terest in the light of the correlation previouslydemonstrated (Segal and Bloch, 1955) betweenvirulence of three strains of M. tuberculosis grownin vitro and degree of hydrogen transfer capacity.An avirulent strain (H37Ra) exhibited the highestcapacity of hydrogen transfer (10 per cent T), anattenuated strain (BCG) was intermediate (20per cent T), and the virulent strain (H37Rv)showed the least activity (28 per cent T). Thisdecreasing scale of metabolic activity as corre-lated with increase in virulence of the strain wouldindicate a high state of virulence for the activityshown by tubercle bacilli separated from infecte(dlungs, although this has not as yet been verifiedby direct animal test.

134 [VOL. 72

on October 17, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

BIO(iOHEMICAL DIFFERENTIATION OF M. TUBERCULOSIS

TABLE 3Comparative respiratory response of H37Rv and LRv to various substrates

H37Rv LRv

Substrate (Final Conc in Vessel) Mean Q,* Increase Mean Q02* Increaseover over

Substrate Endogenous endogenous Substrate Endogenous endogenous

Complex media:0.5 per cent tween-albumin medium. 17.9 4.2 326.2 7.0 4.4 59.1Infected lung filtratet ....... ....... 7.2 4.3 67.4 7.3 4.7 55.3Normal lung filtratet ........ ....... 4.2 4.4 0 4.5 4.5 0Yeast extract (Difco) (1 per cent) ... 7.7 4.4 75.0 4.6 4.7 0

Carbohydrate and intermediates:Glucose (0.2 M) ......... ........... 7.8 4.9 59.2 4.8 5.1 0Glycerol (0.2 M) .......... .......... 10.1 5.0 102.0 5.2 5.3 0DL-Lactic acid (0.0038 M) ........... 7.1 5.1 39.2 4.6 4.6 0Sodium acetate (0.0122 M) .......... 8.2 5.4 51.8 4.9 4.9 0Sodium pyruvate (0.01 M) .......... 7.6 4.4 72.7 4.5 4.5 0

Benzoate and related compounds:Sodium benzoate (0.0035 M) ........ 8.7 4.4 97.7 4.8 4.8 0Benzaldehyde (0.0047 M) ...... ...... 8.0 5.0 60.0 4.7 4.8 0m-Hydroxybenzoic acid (0.0018 M).. 4.6 4.7 0 4.8 4.8 0Sodium salicylate (0.0016 M) ........ 9.1 4.8 89.6 9.0 4.6 95.6

Fatty acids:n-Heptanoic acid (M/300) ........... 8.5 5.5 54.5 7.7 5.8 32.8Octanoic acid (M/300).............. 8.4 5.7 47.4 7.7 5.6 37.5Oleic acid (M/1000) ........ ......... 10.2 5.6 82.1 9.2 5.4 70.4* Each value represents the mean of between 3 and 7 parallel experiments. Statistical analysis applied

to replicate values for each substrate response and its corresponding replicate values of endogenousQo2 established significance for all positive responses above the endogenous Qo2 with the highest valuesof P being <0.05.

t Filtrates of infected and normal lungs were prepared by homogenization of the lungs of 3 mice in10 ml of phosphate buffer, pH 6.8, centrifugation at 3,000 rpm for 15 min, and Seitz filtration of thesupernatant.

The comparative metabolic activity of lungand artificially cultured tubercle bacilli was fur-ther tested by measuring respiratory response tospecific substrates in the Warburg. It was foundthat both H37Rv and LRv have a similar endog-enous rate of respiration in M/15 phosphatebuffer at pH 6.8, with Qo2 values ranging from4.1 to 6.1 for H37Rv and 4.1 to 5.6 for LRv.These values apply to short range runs of 2 or 3hr. There is some evidence that endogenous re-serves are greater for lung bacilli which showhigher levels of endogenous respiration over a20-hr period. For each experiment to be reported,the endogenous Qo2 was determined routinely inorder to determine specific substrate response.In addition, the constancy of this value alsoserved as an excellent criterion for purity andintactness of the separated lung bacilli.Mean Qo, values of substrate response and of

corresponding endogenous oxygen uptake ob-tained from the results of between 3 to 7 repeatedexperiments are presented in table 3. Statisticalanalysis applied to replicate values for each sub-strate response and its corresponding replicatevalues of endogenous Qo, established significancefor all positive responses above the endogenousQo, with the highest values of P being <0.05.These values represent two types of experiments:one in which a group of related substrates wastested using aliquots of the same batch of cells;the other in which different batches of bacilli wereused to test the same substrate. The use of bacilliharvested at different times from cultures ofH37Rv and from infected lungs largely accountsfor their variation in endogenous respiration.However, this variation did not appreciably alterthe net per cent increase in substrate response.The limited variations that did occur were in the

19561 135

on October 17, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

SEGAL AND BLOCH

opposite direction to what might be expectedmathematically, that is, a lower endogeonus res-piration generally resulted in a somewhat loweredsubstrate response on a percentage basis. Whenaliquots of the same batch of bacilli were used,little variation occurred. This particularly ap-plied to suspensions of LRv which contain smallaggregates of 1 to 30 bacilli and which do notclump in water, saline, or phosphate solutions.Equal aliquots of suspensions of H37Rv wereobtained by the inclusion of 0.02 per cent tween80 in the salt solution in which H37Rv was ad-justed and from which aliquots were taken beforefinal saline washing and suspension in phosphatebuffer for use in the Warburg vessel.

Several complex media were used as substratesin an attempt to see whether the respiratoryresponse of lung bacilli could be activated asknown for H37Rv grown in vitro, and to obtainsome clue to the kind of materials to which itmight respond. The results are presented intable 3. Paralleling the low hydrogen transfercapacity shown previously for LRv, 0.5 per centtween-albumin medium gave only a limited meanresponse of 59.1 per cent increase over the endog-enous respiration as compared with the largeincrease of over 300 per cent for H37Rv. Infectedlung filtrate gave a definite response indicatingthe absence of toxic activity for both LRv andH37Rv. Normal lung filtrate and yeast extract(Difco) failed to stimulate the respiration of lungbacilli, whereas H37Rv responded to yeast ex-tract (Difco) with a mean increase in respirationof 75.0 per cent over the endogenous.On the basis of these results, the response to

specific substrates which have been reported inthe literature to stimulate the oxygen uptake ofH37Rv was measured for both H37Rv and LRv.The concentrations of the different substratestested are based on those reported in the lit-erature.The comparative effect of carbohydrate and

several intermediates of glycolysis showed astriking difference in the biochemical activity ofH37Rv and LRv as seen in table 3. The respira-tory response of H37Rv to glucose, glycerol,sodium lactate (Loebel, Shorr, and Richardson,1933), sodium acetate, and sodium pyruvate(Cutinelli, 1940) was found to be consistent withthat reported in the literature, while tuberelebacilli separated from lung tissue gave no response

above that of the endogenous to any of these sub-strates.The comparative effect of sodium benzoate and

related compounds is summarized in tahle 3.Consistent with earlier observations (Bernheim,1941), benzoate, benzaldehyde, and salicylate(o-hydroxybenzoate) all produced considerableincrease in oxygen uptake of H37Rv over theendogenous, whereas m-hydroxybenzoate failedto stimulate respiration. On the other hand, thelung bacilli failed to respond to benzoate andbenzaldehyde as well as to m-hydroxybenzoate,but for the first time showed a very marked re-sponse to salicylate, consistently above thatshown for H37Rv. The singular response of invivo grown M. tuberculosis to salicylate is espe-cially noteworthy in view of the significance thathas been attached to respiratory response to thissubstance as correlated with pathogenicity ofmycobacteria by Bernheim (1942) and Lehmann(1946), who reported a negative response fornonpathogenic strains. 1Iore recently Pope (1954)has reported that this correlation does not applywhen a large number of virulent, avirulent, andattenuated strains are tested. We have also foundthat the avirulent strain H37Ra does respond tosalicylate (0.0016 M) to the extent of 45.5 per centincrease of oxygen uptake over the endogenousrespiration. This result is in partial agreementwith the observations of Bernheim and of Leh-mann in that the avirulent strain, H37Ra,showed a considerably lower response to the sameconcentration of salicylate than that of the viru-lent strain H37Rv (table 3: 89.6 per cent).The response of H37Rv and LRv to several

fatty acids was determined. Loebel et al. (1933)showed that 0.1 per cent oleic acid increased therespiration of H37Rv, and Cutinelli (1940) ob-served marked respiratory stimulation of a humanpathogenic strain by normal saturated fattyacids of chain length C2 to C12. As mentionedpreviously, sodium acetate (listed under theheading "Carbohydrate and intermediates" intable 3) did not stimulate the oxygen uptake ofLRv in contrast to the mean increase of 51.8 percent observed for H37Rv. However, the otherthree fatty acids tested significantly stimulatedrespiration of LRv to degrees similar to thoseobserved for H37Rv. The significance of thismore generalized response of lung bacilli to fattyacids will be discussed later in this paper.

In order to examine the stability of the ob-

136 [VOL. 72

on October 17, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

BIOCHEMICAL DIFFERENTIATION OF M. TUBERCULOSIS

TABLE 4Respiratory re8ponse of LRv after one passage

through tween-albumin medium (LTRv)LRv Control H37Rv Control LTRv

Substrate(inal Conc Increase Increase Increasein Vesse) QO vrQos o ver ,0 over

endog- endog- endog-enous enous enous

Phosphatebuffer. .. 4.8 5.5 5.5

Sodiumsalicyl-ate(0.0032m).... 8.2 70.8 9.0 63.6 8.7 58.2

Glycerol(0.2 m). . 5.1 6.2 11.3 105.4 12.1 120.0

Sodiumacetate(0.0122x).... 4.6 0 8.5 54.5 8.1 47.3

served biochemical characteristics of LRv,separated lung bacilli were inoculated into tween-albumin medim (Dubos and Middlebrook,1947). The rate of growth was slow and in somecases no growth resulted. However, sufficientgrowth occurred in the first passage to supplybacilli for testing in the Warburg. As seen fromthe results presented in table 4, the lung bacilliafter one passage through tween-albumin me-

dium (LTRv) behaved just as H37Rv, with fullresponse to glycerol and acetate in contrast toLRv. Also the response to salicylate was lowerthan that obtained for LRv. These results indi-cate reversion of the lung bacilli to the originalH37Rv strain after primary growth in tween-albumin medium.A series of control experiments was conducted

in order to eliminate the possible operation ofother factors in the observed effects of differen-tial substrate response by H37Rv and LRv. Theuse of moribund or dead mice, and the mode ofkilling mice (ether or bleeding) was found to haveno effect on the respiration of bacilli separatedfrom their lungs. It was important to use micewith heavily infected lungs in order to obtainadequate yields and purity of separation. Thepossible damaging effect of the separation pro-cedure as an explanation of the limited respira-tory response of LRv was discounted by applyingthe entire procedure to H37Rv homogenized with

TABLE 5Effect of the separation procedure and of the infected

lung homogenate on the respiration ofin vitro grown H57Rv

Is Vitro GrownH37Rv Homog-

Is Vilto enized in SucroseGrown H37Rv with Mildly

Substrate Infected Lungs(Final Conc in Vessel) and Separated

Increase IncreaseQo2 over en- QoN over en-

dogenous dogenous

Phosphate buffer.... 5.5 5.2Glycerol (0.2 m) .... . 11.3 105.4 11.0 111.5Sodium salicylate

(0.0032 m) ......... 9.0 63.6 8.4 61.5DL-Lactic acid

(0.0038 x).7.5 36.4Homogenate*........ 6.2 12.7DL-Lactic acid

(0.0038 m) + ho-mogenate*......... 7.9 43.6

* The mildly infected lungs of a mouse killed 14days after infection were homogenized in 25 ml ofphosphate buffer at pH 6.8, and 1 ml of homoge-nate was used in each Warburg vessel.

mildly infected lungs in isotonic sucrose contain-ing 0.25 per cent albumin. It was also necessaryto add 0.1 per cent tween 80 to the homogenizingmedium in order to prevent flocculation ofH37Rv-an effect which does not occur withlung bacilli which suspend readily in water, sa-line, or sugar solutions. As seen in table 5, theendogenous Qo, was not significantly affectedby the separation procedure. Nor was respiratoryresponse to two specific substrates, glycerol andsalicylate, appreciably altered.

Further indication of the absence of inhibitorysubstances in infected lung homogenate was ob-tained by including diluted homogenate ofmildly infected lungs in the vessel with H37Rv,with and without substrate (table 5). Withoutsubstrate there was a limited increase over theendogenous respiration. With lactate the presenceof diluted homogenate had no inhibitory effecton oxygen uptake by H37Rv.The possibility that tissue enzymes carried

over in the separation of LRv might account forsubstrate response was eliminated. Normal andinfected lung filtrates resulted in no uptake ofoxygen when lactate or salicylate was included

1956] 137

on October 17, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

SEGAL AND BLOCH

in the Warburg vessels in the absence of bacilli.Furthermore, even when a diluted homogenatefrom a mildly infected lung was included withseveral substrates, two of which had stimulatedoxygen uptake of LRv (salicylate and oleic acid),no enhancement of response was obtained.The inclusion of 0.5 per cent tween 80 or 2.5

per cent albumin alone or in combination wasshown to have no effect on the negative responseof LRv to glycerol. In the case of H37Rv thepresence of 0.5 per cent tween 80 resulted in onlya partial increase in response to glycerol.The endogenous respiration and substrate

response of LRv was determined at three levelsof pH-6.0, 6.8, and 7.6-to test for the possibleeffect of pH. No significant increase in the nega-tive response of LRv to glucose, glycerol, andlactate was observed. The endogenous respirationof LRv also remained unaffected by changes inpH.

DISCUSSION

The development of a method for the separa-tion of tubercle bacilli from lungs of infected micehas enabled a comparison of these bacteria withthose of the same strain grown in artificial cul-ture medium. On the basis of the differences re-ported in this paper in respect to substratestimulation of respiration, it is quite possiblethat an intensive study will reveal basic differ-ences in the metabolism of in vivo grown bacillias compared to those grown in vitro. The impli-cations of these findings are far-reaching. Theresults presented disprove the commonly heldassumption that virulent tubercle bacilli grownartificially are identical to bacilli in the infectedanimal. Although this assumption has not beenexplicitly stated, it is implicit in most of thework on the tubercle bacillus, particularly thatwhich has attempted to relate effects obtainedwith the organism grown in vitro to the patho-genicity of the organism in vivo. The most com-mon attempt to do this has been to apply themethod of comparative biochemistry to a studyof virulent, attenuated, and avirulent strains.Differential responses to various test conditionsthat could be correlated with virulence of strainhave been considered to reflect the nature of thehost-parasite relationship and the nature ofpathogenicity.From the limited results presented in this

paper the biochemical differences that have been

shown in the literature to exist between avirulentand virulent strains of M. tuberculosis are rela-tively minor compared to the differences shownhere between a virulent strain grown in vitro andthe same strain separated from growth in vivo.The differences are sufficient in scope to justifyclassification of H37Rv and LRv as strain vari-ants. This is not meant to imply that there is anabsolute difference between them. On the basisof the reversion of LRv to H37Rv after growth inartificial culture medium (table 4), a more likelyexplanation of the observed biochemical differ-ences would seem to be adaptation to widelydiverse environments. Thus LRv might beexpected to respond, upon prolonged incubation,to substrates to which its respiratory responsehas been negative within a 3-hr period. To date,extension of the test period to as long as 20 hrhas failed to enhance the capacity for increasedrespiratory response to substrates. It is possiblethat a more complex supply of nutrients and astill longer period of incubation are necessarybefore LRv can adapt to the oxidation of foreignsubstrates in a Warburg vessel. That it does soeventually is clearly evident from its reversion toH37Rv after primary growth in artificial culturemedium. The reverse process of conversion ofH37Rv to LRv that evidently takes place in vivoshould theoretically be possible in vitro, assum-ing an adaptive mechanism for the changeover.This could be accomplished by duplication invitro of the essential environmental conditions ininfected tissue that determine the dissimilarcharacteristics of LRv. Such determinants willbe reflected by a comparative study of in vivoand in vitro grown tubercle bacilli.

Since the observed differences are biochemicalin nature, it is likely that considerable differencesin chemical composition exist. The ease withwhich LRv suspends in water, saline, or sucrosesolutions in contrast to H37Rv, which requiresthe presence of a surface active agent such astween 80, indicates an important difference insurface properties which is probably due to adifference in chemical structure. The two reportscited earlier in this paper (Anderson et at., 1943;Sheehan and Whitwell, 1949) also point to achemical difference between bacilli grown in arti-ficial culture and those growing in infected tissue.In turn, differences in immunogenic propertiesmight also be found to occur. Preliminary resultshave been obtained which show marked differ-

138 [VOL. 72

on October 17, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

BIOCHEMICAL DIFFERENTIATION OF M. TUBERCULOSIS

ences between the immunogenic properties ofphenol-killed suspensions of LRv and H37Rv.The lack of differences in growth characteris-

tics of submerged cultures as well as of colonieson agar plates is not surprising in view of thefindings shown in table 4, where after onegrowth-passage through tween-albumin medium,LRv has reverted to H37Rv in respect to bio-chemical characteristics. Differences in growthproperties would be expected to occur as a func-tion of this adaptive cycle ending in reversion.The delay of several days in the growth of LRvtransferred to artificial culture medium indicatesthe adaptive nature of its change to H37Rv.

Other questions arise in attempting to explainthe differences between LRv and H37Rv. Thefactor of permeability might be operative, butthis would not readily explain the specific differ-ential response to salicylate. In addition, thepresence of tween 80 had no effect on negativeresponses of LRv to glycerol. The explanationthat LRv does not react due to an over supply ofendogenous food reserves is relevant. Althoughevidence to date indicates that endogenous res-piration is more stable over a 20-hr period forlung bacilli, this need not necessarily result innegative substrate response. Again the differen-tial responses to salicylate and the fatty acids aswell as response to infected lung filtrate suggestthat high endogenous reserves are not a factor. Afew trials have been made to test response afterpreliminary starvation but with no success.However, neither of these possible explanationshas been ruled out and further studies using thestarvation technique and cell-free extracts havebeen initiated toward this end.The possible operation of such factors as

damage, inhibition, or stimulation in determiningrespiratory response of LRv to substrates as aresult of the method of separation or due to thepossible presence of traces of tissue material, hasbeen eliminated by the various control experi-ments described earlier in this paper. Lastly, pHwas also shown not to be a determining factor.The nature of the positive responses of LRv to

salicylate and higher fatty acids remains to bedetermined. The question whether the stimula-tion of oxygen uptake by these compounds is onlyindirect or whether these substances are utilizedand actively metabolized is primary. Fitzgeraldand Bernheim (1948) concluded that salicylicacid is not oxidized by pathogenic tubercle bacilli

(H37Rv), and attributed its stimulating effecton oxygen uptake to catalysis of endogenous re-actions. A more extensive testing of compoundsof biological importance might offer more cluesregarding the nature of the specific responsesthus far observed.

Related to the question of high endogenousfood reserves is the impression given by theresults that lung bacilli present in infected tissuesexist in a reduced state of activity and uponseparation appear sluggish and nonreactive.However, the very marked responses to fattyacids and particularly salicylate contradict this.This apparent limited range of activity is quiteconsistent with the well-established correlationthat bacteria are increasingly fastidious in theirnutritional requirements when they are parasiticin their natural environment and as their para-sitism is more obligatory. The mycobacteria fitthe pattern well. Franke and Schillinger (1944)have indicated that the range of substrates oxi-dized by mycobacteria decreases sharply as acidfastness and pathogenicity increase. Hanks andGray (1954) have classified the mycobacteriaaccording to their metabolic properties, illus-trating that enhancement of respiratory responsebecomes more and more limited in a scale ex-tending from the various saprophytes throughthe tubercle bacillus to M. johnei, M. lepraemu-rium, and M. leprae, the most parasitic of all.The negative response of LRv to carbohydratesand its positive response to fatty acids places ithigh in the scale of pathogenicity according tothis scheme of Hanks and Gray, actually beyondthe level of listing of the tubercle bacillus, whichhas been based on studies using in vitro grownbacilli.Whatever the eventual elucidation of mecha-

nisms involved in producing these differences ofin vivo and in vitro grown tubercle bacillus, theirexistence is definite. By means of this approacha truer picture of the tubercle bacillus in itsnatural parasitic state can now be obtained.

ACKNOWLEDGMENT

The authors are grateful to Miss ArankaSchupler for her valuable technical assistance.

SUMMARY

The development of a method for the separa-tion of Mycobacterium tuberculosis in large num-bers directly from the lungs of infected mice has

1956] 139

on October 17, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

SEGAL AND BLOCH

enabled a comparative biochemical study of invivo and in vitro grown bacilli of the same strain.

Saline suspensions of tubercle bacilli separatedfrom infected lungs (LRv) were shown to bequantitatively comparable, as to dry weight andplate counts, to equally dense suspensions ofvirulent tubercle bacilli of the same strain grownon artificial culture medium (H37Rv). This find-ing confirmed the purity of separation of bacillifrom tissue, as determined by microscopic exami-nation, and established a quantitative basis forbiochemical comparison of LRv and H37Rv.A large difference in hydrogen transfer capacity

was shown to exist between LRv and H37Rv,with the lung bacilli exhibiting a much lowerstate of metabolic activity in the presence of sub-strate. Paralleling the low hydrogen transfercapacity of LRv, complex culture media had littleor no stimulating effect on its endogenous res-

piration, whereas they produced a large increasein oxygen uptake over the endogenous respirationof H37Rv. The endogenous rate of respirationwas found to be similar for both types of bacilli.

Striking differences in the biochemical activityof H37Rv and LRv were observed with respectto substrate response. The positive respiratoryresponses of H37Rv to glucose, glycerol, lacticacid, sodium acetate, sodium pyruvate, sodiumbenzoate, benzaldehyde, sodium salicylate, n-hep-tanoic, octanoic, and oleic acids were found to beconsistent with reports in the literature. In con-

trast, only sodium salicylate and the latter threefatty acids were found capable of stimulatingthe endogenous respiration of LRv.The adaptive nature of the observed biochemi-

cal differences was indicated by reversion of LRvto H37Rv with respect to response to glyceroland sodium acetate after primary growth inartificial culture medium.These results indicate that tubercle bacilli

grown artificially are dissimilar to bacilli of thesame strain as they multiply in the lungs of an

infected animal.

ADDENDUM

More recently other biological differences havebeen found to exist between in vivo and in vitrogrown tubercle bacilli. LRv exhibits a greaterdegree of pathogenicity for mice, but a lesserimmunizing capacity than H37Rv. These resultswill be published in the near future.

REFERENCESANDERSON, R. J., REEVES, R. E., CREIGHTON,

M. M., AND LOTHROP, W. C. 1943 The chem-istry of the lipids of tubercle bacilli. LXV.An investigation of tuberculous lung tissue.Am. Rev. Tuberc., 48, 65-75.

BERNHEIM, F. 1941 The effect of various sub-stances on the oxygen uptake of the tuberclebacillus. J. Bacteriol., 41, 387-395.

BERNHEIM, F. 1942 The oxidation of benzoicacid and related substances by certain myco-bacteria. J. Biol. Chem., 143, 383-389.

CUTINELLI, C. 1940 La ossidazione degli acidigrassi da parte del bacillo di Koch. Boll. 1stsieroterap. milan., 19, 88-94.

Du-Bos, R. J. AND MIDDLEBROOK, G. 1947 Mediafor tubercle bacilli. Am. Rev. Tuberc., 56,334-345.

FENNER, F. 1951 The enumeration of viabletubercle bacilli by surface plate counts. Am.Rev. Tuberc., 64, 353-380.

FITZGERALD, R. J. AND BERNHEIM, F. 1948 Theeffect of salicylic acid on the growth, mor-phology and virulence of M. tuberculo8is.Am. Rev. Tuberc., 58, 210-214.

FRANKE, W. AND SCHILLINGER, A. 1944 ZumStoffwechsel der saurefesten Bakterien. I.Orientierende aerobe Reihenversuche. Bio-chem. Z., 316, 313-334.

GRAY, C. T. 1952 The respiratory metabolismof murine leprosy bacilli. J. Bacteriol., 64,305-313.

HANKS, J. H. 1951a Measurement of the hydro-gen transfer capacity of mycobacteria. J.Bacteriol., 62, 521-528.

HANKS, J. H. 1951b The biological significanceof the hydrogen transfer capacity of murineleprosy bacilli. J. Bacteriol., 62, 529-537.

HANKS, J. H. AND GRAY, C. T. 1954 The appli-cation of metabolic studies to leprosy re-search. Intern. J. Leprosy, 22, 147-161.

LEHMANN, J. 1946 Determination of pathoge-nicity of tubercule bacilli by their intermedi-ate metabolism. Lancet, 250, 14-15.

LOEBEL, R. O., SHORR, E., AND RICHARDSON, H. B.1933 The influence of foodstuffs upon therespiratory metabolism and growth of humantubercle bacilli. J. Bacteriol., 26, 139-166.

POPE, H. 1954 The effect of benzoate and sali-cylate on the oxidative metabolism of tuberclebacilli. Am. Rev. Tuberc., 69, 705-709.

SEGAL, W. AND BLOCH, H. 1955 Effects ofsunlamp irradiation on the viability, viru-lence, and biochemical activity of Mycobac-terium tuberculosis. Am. Rev. Tuberc., 71,112-125.

140 [VOL. 72

on October 17, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

BIOCHEMICAL DIFFERENTIATION OF M. TUBERCULOSIS

SEMB!HAN, H. L. AND WHirWELL, F. 1949 Thestaining of tubercle bacilli with sudan blackB. J. Pathol. Bacteriol., 61, 269-271.

SMITH, H. AND KEPPIE, J. 1955 Studies on thechemical basis of the pathogenicity of Bacillwanthracis using organisms grown in vivo. InMechanism. of microbial pathogenicity, pp.126-151. Fifth Symposium of the Society forGeneral Microbiology. University Press, Cam-bridge, England.

UM3EREIT, W. W., BuRis, R. H., AND STAuFFUR,J. F. 1949 Methods for the isolation of par-ticulate components of the cell. Manometrictechniques and ti8sue metabolism, ChapterXII, 2nd ed. Burgess Publishing Co., Minne-apolis, Minn.

VANDIVIERE, H. M., GENTRY, W. H., AND WILMS,H. 5. 1952 A rapid chemical test of totalviability for suspensions of tubercle bacilli.Am. Rev. Tuberc., 66, 95-98.

19561 141

on October 17, 2020 by guest

http://jb.asm.org/

Dow

nloaded from