JOURNAL OF BACTERIOLOGY,0021-9193/98/$04.0010

Feb. 1998, p. 801–808 Vol. 180, No. 4

Copyright © 1998, American Society for Microbiology

Mycobacterial Stationary Phase Induced by Low Oxygen Tension:Cell Wall Thickening and Localization of the

16-Kilodalton a-Crystallin HomologADAM F. CUNNINGHAM AND CLAIRE L. SPREADBURY*

Molecular Mycobacteriology Research Group, Department of Infection, The Medical School,University of Birmingham, Birmingham B15 2TT, United Kingdom

Received 25 August 1997/Accepted 16 December 1997

Most cases of tuberculosis are due to reactivation of endogenous infection which may have lain quiescent ordormant for decades. How Mycobacterium tuberculosis survives for this length of time is unknown, but it ishypothesized that reduced oxygen tension may trigger the tubercle bacillus to enter a state of dormancy.Mycobacterium bovis BCG and M. tuberculosis H37Rv were cultured under aerobic, microaerobic, and anaerobicconditions. Their ultrastructural morphology was analyzed by transmission electron microscopy (TEM), andprotein expression profiles were compared by sodium dodecyl sulfate-polyacrylamide gel electrophoresis(SDS-PAGE). TEM revealed that the microaerobically and anaerobically cultured bacilli but not the aerobi-cally cultured bacilli developed a strikingly thickened cell wall outer layer. The thickening was not observed inaerobically cultured stationary-phase bacilli or in anaerobically cultured Mycobacterium smegmatis. A highlyexpressed protein was detected by SDS-PAGE in microaerobic and anaerobic cultures and was identified as the16-kDa small heat shock protein or a-crystallin homolog. Immunolocalization by colloidal gold immunoelec-tron microscopy identified three patterns of protein distribution in M. bovis BCG cultured under low oxygentension. The 16-kDa protein was strongly associated with the cell envelope, fibrous peptidoglycan-like struc-tures, and intracellular and peripheral clusters. These results suggest that tubercle bacilli may adapt tolow-oxygen conditions by developing a thickened cell wall and that the 16-kDa protein may play a role instabilizing cell structures during long-term survival, thus helping the bacilli survive the low oxygen tension ingranulomas. As such, the cell wall thickening and the 16-kDa protein may be markers for the dormant stateof M. tuberculosis.

Mycobacterium tuberculosis, the causative agent of tubercu-losis, accounts for more deaths in the world than any otherpathogen (2, 25). One of the major contributing factors to thesuccess of M. tuberculosis as a human pathogen is its ability topersist in the face of the immune response, only to reactivatelater and cause disease. Most cases of tuberculosis are theresult of reactivation of preexisting infection rather than rein-fection (25), and one-third of the world’s population has beenestimated to be latently infected (39). The 6-month treatmenttime (so called short-term therapy) needed to achieve a curecan lead to patient noncompliance resulting in a poor outcome,further spread of the disease, and the development of drug-resistant strains. A drug which can selectively kill dormantbacilli is urgently needed. The combination of such a drug andexisting drugs that target actively replicating bacilli would rap-idly kill both populations, drastically reduce the treatmenttime, and eliminate persisters and their potential for reactiva-tion.

How the tubercle bacillus survives during the dormant stageof infection is largely unknown. Following initial infection, thebacilli typically replicate inside host macrophages until an ef-fective immune response is mounted and the bacilli becomerestricted to granulomas and the progression of the disease ishalted. The internal environment of the granuloma is consid-

ered to be a hostile one, characterized by low O2 levels andhigh CO2 levels, acidic pH, and the presence of aliphatic or-ganic acids, such that it was assumed that the bacilli died withina short period of time (11–13). However, it was later demon-strated that bacilli were not necessarily killed and could survivefor many years (44). Using an in vitro model whereby M. tu-berculosis bacilli were grown in a nonagitated culture, Waynefound that in the microaerophilic sedimented layer of the me-dium, nondividing, yet viable, bacilli were present (41). Thesenondividing-but-viable bacilli led Wayne to suggest that theymay be analogous to tubercle bacilli that persist in vivo duringquiescent tuberculosis. Microaerophilically cultured bacilliwere found to undergo an orderly metabolic shiftdown, evi-denced by an apparent shift into the glyoxylate cycle, permit-ting adaptation to the usually lethal effects of anaerobiosis(43). Tolerance of anaerobic conditions was further supportedby the finding that the dormant bacilli were partially or com-pletely resistant to isoniazid and rifampin yet, unlike aerobi-cally cultured bacilli, were susceptible to metronidazole, a drugtypically directed against anaerobic bacteria (45).

The experimental work by Wayne and coworkers clearlyimplies that low oxygen tension may play a role in mycobacte-rial dormancy and that adaptation to anaerobiosis may be afeature of persisting tubercle bacilli. In this study we sought todetermine the adaptations mycobacteria may make at both themorphological and molecular level in response to low-oxygenculture conditions. We report that there is a very markedthickening of the cell wall outer electron-opaque layer in My-cobacterium bovis BCG and M. tuberculosis when the bacteriaare cultured under either microaerobic or anaerobic condi-tions. We also identified a protein in M. bovis BCG bacilli

* Corresponding author. Mailing address: Molecular Mycobac-teriology Research Group, Department of Infection, The MedicalSchool, University of Birmingham, Birmingham B15 2TT, UnitedKingdom. Phone: 44 (0)121 414 6975. Fax: 44 (0)121 414 3454. E-mail:c.l.spreadbury@bham.ac.uk.

801

on August 23, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

which was upregulated under microaerobic and anaerobic con-ditions, and we describe the localization of this 16-kDa proteinand its patterns of distribution, as determined by immunogoldelectron microscopy. The significance of these findings to thedormant phase of M. tuberculosis and to the latency of humantuberculosis are discussed.

MATERIALS AND METHODS

Organisms and culture conditions. M. bovis BCG (Statens Seruminstitut strainST1077), M. tuberculosis H37Rv (ATCC 9360), and Mycobacterium smegmatismc2155 were grown in Middlebrook 7H9 medium supplemented with albumin-dextrose-catalase (ADC; Difco Laboratories Ltd., West Molesey, United King-dom) and 0.05% (vol/vol) Tween 80 (Sigma, Poole, United Kingdom). In someexperiments Tween 80 was replaced, where stated, with 0.2% (vol/vol) glycerol.Either sterile polystyrene universals or 100-ml glass bottles containing 10 or 50ml of 7H9–ADC–Tween 80 broth, respectively, were inoculated with bacilli andcultured in a shaking 37°C incubator until slight turbidity was achieved (usuallyat an optical density at 600 nm [OD600] of ;0.4). The culture vessels were thenplaced in a standard nonshaking 37°C incubator (the M. tuberculosis cultureswere placed in a nonshaking CO2 incubator with 5% CO2) under either mi-croaerobic or anaerobic conditions. To obtain a microaerobic environment, aself-induced oxygen gradient was permitted to form as the bacilli settled to thebottom of the culture vessel. The screw caps were loosened to allow gaseousexchange between the environment and the media. For anaerobic conditions, theculture was overlaid with filter-sterilized mineral oil (molecular biology grade;Sigma) and the tube caps were tightened. In some experiments M. bovis BCG wascultured in an anaerobic cabinet in the absence of mineral oil. The microaerobicand anaerobic cultures were left undisturbed for periods of up to 6 monthsbefore harvesting. Controls comprised bacilli cultured aerobically with agitationto an OD600 of 0.07 to 0.4 (low-OD control) and bacilli cultured aerobically withshaking for 6 weeks to an OD600 of .1.0 (nutrient-limited stationary-phasecontrol [NLSPC]). Aerobic cultures were coprocessed with the anaerobic sam-ples. Cultures were routinely checked for contaminants and were examined byZiehl-Neelsen staining.

TEM. Cells were harvested by low-speed centrifugation, washed in sterilephosphate-buffered saline (PBS), and fixed for a minimum of 1 h in 2.5%(vol/vol) glutaraldehyde in 0.1 M sodium phosphate. Cells were then pelleted,the supernatant was removed, and the pellet was resuspended in 1% (wt/vol)osmium tetroxide for 1 h. Cells were then dehydrated through a graded ethanolseries (50, 70, 90, and 100%; each level was applied twice for 5 min each time)and propylene oxide (twice for 15 min) and infiltrated with, unless otherwisestated, agar 100 (Agar, Stansted, United Kingdom). Resin blocks were polymer-ized at 60°C for 24 h. Sections were cut on a Reichert-Jung Ultracut E ultratometo silver/gold interference, picked up on Formvar-coated grids, and stained with30% (wt/vol) uranyl acetate in 70% (vol/vol) methanol and counterstained withReynold’s lead citrate (34) before examination on a Jeol JEM-100CXII TEM atan acceleration voltage of 80 kV. Under some circumstances, where stated, oneor more of the above fixation or staining steps were omitted. Cells were alsoprepared for TEM by the method of progressive lowering of temperature (PLT)(see below), following which grids were stained with osmium tetroxide vapor byplacing them in a petri dish for 3 h with an osmium tetroxide crystal. QualitativeX-ray microprobe analysis was used to help determine which heavy metal wasresponsible for the contrast observed in the thickened cell wall. Briefly, routinelyprepared sections were analyzed in the electron microscope fitted with a high-resolution scanning attachment, a LaB6 filament, and a Link ISIS 30-mm2 Si(Li)detector, atmosphere thin window, and multichannel analyzer (Oxford Instru-ments, High Wycombe, United Kingdom).

Protein extraction and SDS-PAGE analysis. Bacilli were harvested, washed,and resuspended in 1 ml of PBS containing 0.4 ml of sterile glass beads (0.1 mmin diameter) before disruption in a Mini Beadbeater (Stratech Scientific, Luton,United Kingdom) by three 60-s bursts at 50,000 rpm. Protein concentration wasdetermined by the bicinchoninic acid assay (BCA kit; Pierce and Warriner,Chester, United Kingdom). Proteins were separated on a 12% denaturing poly-acrylamide gel by sodium dodecyl sulfate-polyacrylamide gel electrophoresis(SDS-PAGE) (26) using a Mini-Protean II minigel apparatus (Bio-Rad, HemelHempstead, United Kingdom) and were visualized by 0.5% Coomassie brilliantblue staining. A protein band of interest was excised, and the N-terminal se-quence was determined at the Glaxo Wellcome Medicines Research Centre,Stevenage, United Kingdom.

Cloning and sequencing of the M. bovis BCG 16-kDa a-crystallin homolog.Oligonucleotide primers based on the N-terminal sequence of the M. bovis BCG16-kDa protein and the nucleotide sequence of the gene encoding the M. tuber-culosis 16-kDa a-crystallin homolog (GenBank accession no., S79751) were de-signed. Two primers, 59-ATGGCCACCACCCTTCCCGT-39 and 59-CAGTTGGTGGACCGGATCTG-39, were used to amplify the M. bovis BCG 16-kDaa-crystallin gene with TaqI polymerase (Boehringer Mannheim, Lewes, UnitedKingdom) by PCR. The PCR product was purified (Geneclean Spin kit;Anachem, Luton, United Kingdom), polished with PfuI polymerase, blunt-endcloned into pCR-ScriptSK(1) (Stratagene, Cambridge, United Kingdom), andtransformed into Escherichia coli SURE cells (Stratagene). Transformants were

selected for ampicillin resistance with blue/white color selection. Plasmid DNAcontaining the appropriate insert was sequenced on an automated sequencer(ABI 373) by Alta Bioscience, University of Birmingham, United Kingdom, bythe Taq Dye Deoxy terminator cycle method (Applied Biosystems Ltd., War-rington, United Kingdom).

Immunoelectron microscopy. Bacilli were prepared for immunogold analysisby the PLT method (35). Samples were fixed for 60 min in 2% paraformalde-hyde–0.05% glutaraldehyde in 0.1 M phosphate buffer. Samples were dehydratedthrough a graded ethanol series (from 30 to 100% ethanol) from 0 to 250°C. Thesamples were then infiltrated with Lowicryl HM20 for 24 h at 250°C prior toembedding and polymerization with ultraviolet light in fresh Lowicryl HM20,first for 48 h at 250°C and then for 24 h at 15°C. Silver/gold sections were cut asdescribed above and labelled by blocking with 1.0% bovine serum albumin inPBS for 45 min at room temperature before exposure to a pooled polyclonalmurine anti-M. tuberculosis recombinant 16-kDa a-crystallin antibody (a kind giftfrom J. Ivanyi [MRC Clinical Sciences Centre, Tuberculosis and Related Infec-tions Unit, Imperial College School of Medicine, Hammersmith Campus, Lon-don, United Kingdom]) diluted in blocking solution for 3 h at 37°C. The gridswere washed three times for 5 min in blocking solution before the addition of theimmunogold conjugate, a 1:50 dilution of 10-nm colloidal gold-labelled goatanti-mouse antibody (British Biocell International, Cardiff, United Kingdom).The grids were washed a further three times for 5 min in blocking solution andthree times for 1 min in distilled water before being stained with 30% uranylacetate in 70% methanol. Controls of preimmune murine serum and conjugate,as well as conjugate only, were incorporated to ensure that low backgroundlabelling and high specificity of labelling were achieved. Sections were visualizedwith a Jeol JEM-100CXII TEM at an acceleration voltage of 80 kV.

RESULTS

Changes in cell wall electron density in M. bovis BCG andM. tuberculosis are associated with decreased oxygen tension.Analysis by TEM of M. bovis BCG cultured under anaerobicconditions for 6 months revealed a marked thickening of thecell wall (Fig. 1A and B). The main feature of these cells wasthe extremely dense electron staining of the cell walls, which inmost cases homogeneously covered the surfaces of the bacilli.In some cases the thickened cell wall appeared to have rup-tured, to have “peeled” off some cells, and other cells appearedto be denuded of part of their thickened cell walls, probablydue to the processing of the cells for TEM. Despite the longperiod of culture under these conditions, the internal architec-ture of the cells appeared to be preserved. The same cell wallthickening was also seen in microaerobic cultures, althoughless frequently. However, this phenomenon was not observedin aerobically grown (low-OD control) bacilli (Fig. 1C) or inbacilli cultured aerobically with continuous agitation for 6weeks, which would have entered stationary phase (NLSPC)(Fig. 1D). The bacilli cultured under microaerobic and anaer-obic conditions for 6 months could be resuscitated by inocula-tion into fresh media (7H9–ADC–glycerol broth) under aero-bic growth conditions. Only a loopful of the 6-month cultureswas inoculated, and the results were merely recorded as posi-tive growth. We did not ascertain how many cells in the inoc-ulum were viable.

To determine whether this cell wall thickening could beobserved in a pathogenic mycobacterial species, M. tuberculosisbacilli were cultured under anaerobic conditions in the sameway for 1 month. TEM analysis showed that M. tuberculosisalso developed a thickened cell wall associated with a de-creased availability of oxygen (Fig. 1E and F). This was notobserved in aerobically grown M. tuberculosis. Morphologi-cally, the thickened cell wall appeared to be similar in M. tu-berculosis and M. bovis BCG. In contrast, the fast-growingsaprophytic species, M. smegmatis, did not develop the thick-ened cell wall after 17 and 35 days of anaerobic culture. In the35-day-old anaerobic culture less than 1% of the bacilli re-mained intact. Of those which were intact, none exhibited thecell wall thickening (Fig. 1H).

Confirmation of observations: elimination of influence ofmineral oil and Tween 80. It was not possible that the presence

802 CUNNINGHAM AND SPREADBURY J. BACTERIOL.

on August 23, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

of the mineral oil used to induce anaerobic conditions wasresponsible for the thickening as it was also seen in microaero-bic cultures (cultured in the absence of mineral oil). Bacillicultured in an anaerobic cabinet without mineral oil developedthe thickened cell wall from 2 weeks onwards. Moreover, whenM. bovis BCG cells were grown as described for the aerobiccultures, overlaid with mineral oil prior to being washed andprocessed for TEM analysis, and treated in the same manneras for the anaerobic cultures, thickened cell walls were notobserved (only M. bovis BCG was used in these experiments asfacilities for standard anaerobic culture of M. tuberculosis werenot available). To eliminate the possibility that this was a

Tween 80-dependent phenomenon, M. bovis BCG cells werecultured under mineral oil for 6 weeks in 7H9–ADC broth with0.2% (vol/vol) glycerol replacing Tween 80. The same thick-ened cell walls were observed, showing that Tween 80 was notnecessary for the thickened cell walls to be formed.

Osmium tetroxide postfixation and staining is primarilyresponsible for the detection of the cell wall thickening. Os-mium tetroxide was shown, by a number of experiments, to benecessary for the postfixation of the cell wall in samples pre-pared for routine TEM and for enhancing the contrast of thecell wall in samples prepared for TEM by the PLT method. Inone experiment, anaerobic cultures were prepared for routine

FIG. 1. Effect of anaerobic culture on the morphology of M. bovis BCG, M. tuberculosis, and M. smegmatis. (A) Longitudinal section of an M. bovis BCG bacilluscultured anaerobically for 6 months showing pronounced cell wall thickening (some cytoplasmic dehydration is evident). (B) Transverse sections of M. bovis BCG bacillicultured anaerobically for 6 months. (C) Lack of cell wall thickening in low-OD aerobically cultured M. bovis BCG bacilli. (D) NLSPC M. bovis BCG bacilli showingthe absence of cell wall thickening. (E and F) Cell wall thickening in M. tuberculosis following anaerobic culture for 1 month under mineral oil. (G) Aerobically culturedM. smegmatis. (H) M. smegmatis bacilli exhibiting cell lysis following 35 days of anaerobic culture. Bars, 200 nm except for panels C (100 nm) and G (400 nm). Thebar in panel A is also valid for panel F; the bar in panel D is also valid for panel H.

VOL. 180, 1998 CELL WALL THICKENING AND a-CRYSTALLIN LOCALIZATION 803

on August 23, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

TEM; however, half of the culture was not postfixed in osmiumtetroxide but was treated identically in every other respect tothe remainder of the culture. When sections from the nonpost-fixed specimen were examined, it was found that the thick cellwall was not visualized. This contrasted starkly with the os-mium tetroxide-postfixed cell preparation in which the thickcell wall was evident even in the absence of further stainingwith lead or uranium salts. To provide further evidence thatosmium tetroxide was responsible for the staining, X-ray mi-croprobe analysis was undertaken to compare the cell wall tothe background. This analysis confirmed the presence of os-mium in the cell wall. No common coprecipitants such asphosphate (as illustrated by the presence of phosphorus) wereidentified. To further confirm that osmium tetroxide was re-sponsible for the staining, some cells were prepared for TEMby the PLT method. This method of preparation permits struc-tural integrity to survive the preparation process but yieldspoor ultrastructural detail. Sections prepared in this mannerwere stained with osmium vapor after processing. The detec-tion of the unusual cell wall morphology in cells prepared andstained in this manner suggests that the osmium was not alter-ing the cell wall during processing for standard TEM analysis.

Cell wall thickness and cell size. Image enhancement ofelectron micrographs taken of the thickened cell wall (Fig. 2)showed that the thickening was restricted to the cell wall outerelectron-opaque layer. The other components of the cell wallappeared not to be enhanced in any observable way. The thick-ness of the cell wall was measured for anaerobically culturedbacilli. Measurements were only taken on transverse sectionsof the cell wall at the thinnest point. Anaerobic culture of

M. bovis BCG in the presence or absence of mineral oil did notalter the degree of thickening of the cell wall. The cell wall wasthicker in anaerobically cultured M. tuberculosis than in anaer-obically cultured M. bovis BCG (21.2 6 2.04 nm [n 5 9] versus16.1 6 1.03 nm [n 5 49], respectively). The internal diametersof the bacilli (i.e., excluding the cell wall) were also measuredby measuring the diameters of the cells at the narrowest axes oftransverse sections. The cell diameter was used as opposed tothe cell length as it was easier to measure and more likely toreflect a valid and reproducible measurement. The anaerobi-cally cultured M. bovis BCG cells were much thinner (255 64.35 nm [n 5 52]) than the aerobically grown controls (284 66.20 nm [n 5 29]), and these differences were highly significant(P , 0.001). The same pattern was also observed for M. tuber-culosis, where the anaerobically cultured bacilli were thinner(278 6 22.5 nm [n 5 9]) than the aerobically grown bacilli(338 6 18.9 nm [n 5 10]; P , 0.05).

Identification of the M. bovis BCG 16-kDa a-crystallin ho-molog. SDS-PAGE analysis of protein extracted from M. bovisBCG cultured microaerobically and anaerobically for 1, 2, 4,12, and 26 weeks (6 months) showed the presence of a proteinof approximately 18 kDa (data not shown). This protein ap-peared to reach maximal expression, as a proportion of totalprotein, after 2 weeks of either microaerobic or anaerobicculture, and this level of expression was maintained for up to26 weeks of culture. The protein appeared to be poorly ex-pressed in the low-OD aerobic controls, but greater expressionwas noted in the NLSPC and probably reflects its induction byoxygen deficiency due to cell clumping. The N-terminal se-quence of this protein—ATTLPVQRHPRSLFP—had 100%

FIG. 2. Cell wall architecture of anaerobically cultured M. bovis BCG (A to C) compared to that of aerobically cultured M. bovis BCG (D) showing that only theouter electron-opaque layer is thickened in response to low oxygen tension. The numbers in panel A refer to the individual cell wall envelope components, as interpretedby Brennan and Nikaido (3). From the cytoplasm outwards, the numbers indicate the following: 1 and 2, the plasma membrane; 3, the electron-dense layer(peptidoglycan); 4, the electron-transparent layer; 5, the outer electron-opaque layer. The thickened outer electron-opaque layer and cell envelope appear to exhibitplasticity (B). Note the absence of the outer electron-opaque layer where two bacilli are in direct physical contact (C). (D) Low-OD aerobically cultured bacillus. Bar,30 nm (for all panels).

804 CUNNINGHAM AND SPREADBURY J. BACTERIOL.

on August 23, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

homology with the previously described M. tuberculosis 16-kDasmall heat shock protein (smHSP) or a-crystallin homolog (27,40).

A partial nucleotide sequence of the gene encoding the M. bo-vis BCG 16-kDa a-crystallin protein is 100% identical to thatof the equivalent region in the M. tuberculosis homolog. PCRamplification of M. bovis BCG genomic DNA with primersbased on the M. tuberculosis gene sequence encoding the 16-kDa protein generated a product 434 bp in size. Excluding theprimer sequences, 394 bases were sequenced and were foundto be 100% identical to those of the equivalent region inM. tuberculosis.

Immunolocalization of the M. bovis BCG 16-kDa a-crystal-lin protein. Immunoelectron microscopy with a specific poly-clonal antiserum to the 16 kDa protein was used to determinethe distribution of the protein. An examination of aerobicallycultured low-OD bacilli revealed poor expression of the 16-kDa protein (Fig. 3A). However, a high degree of specificlabelling was observed for the microaerobic and anaerobiccultures from 2 to 26 weeks, and three patterns of 16-kDaprotein distribution were observed.

Sixteen-kilodalton protein pattern of distribution I: locali-zation to the cell periphery. The protein was frequently foundto be localized to the peripheral regions of M. bovis BCGbacilli (Fig. 3B), where the label could clearly be seen to followthe contours of the cell. The 16-kDa protein was also locatedto the cell membrane and cell wall skeleton and less frequentlythe outer regions of the cell wall (Fig. 3C), suggesting that the16-kDa protein may be associated with maintaining the struc-tural integrity of the cell.

Sixteen-kilodalton protein pattern of distribution II: local-ization to fibrous structures. The second type of label distri-bution showed that there were internal line structures alongwhich the 16-kDa protein was associated (Fig. 3D and E).These lines could be frequently visualized as structural detailsof varying electron density on the electron micrographs andprobably represent peptidoglycan located in the cell wall (1,22). These fibrous structures were most frequently seen inlongitudinal sections, probably due to the plane of sectioning.However, this pattern of label distribution was also seen in theabsence of detailed fibril ultrastructure (Fig. 3F).

Sixteen-kilodalton protein pattern of distribution III: local-ization to intracellular and peripheral clusters. The 16-kDaprotein was observed to form clusters (Fig. 3G and H) whichdid not exceed 40 nm in width. These clusters were locatedboth inside the cell and at its periphery, but which cellularcomponents the clusters were associated with could not bedetermined. However, little or no label appeared to be asso-ciated with chromosomal DNA (Fig. 3H).

DISCUSSION

This is the first time changes in mycobacterial ultrastructuralmorphology have been correlated to reduced oxygen tension.We have shown that there is an expansion of the outer elec-tron-opaque layer of the cell wall in anaerobically and mi-croaerobically cultured bacilli concomitant with a reduction inthe diameter of the cell as determined by measuring transversesections. The outer electron-opaque layer has only been clearlyidentified as a distinct entity relatively recently but was foundto be much thinner than that described here (32).

The cell wall thickening could only be observed in cellsprepared for routine TEM which had been postfixed with os-mium tetroxide, suggesting that the cell wall thickening wasextractable in organic solvents at room temperature (used inthe subsequent steps in routine TEM sample preparation).

Moreover, when cells prepared for electron microscopy by thePLT method were stained by osmium tetroxide vapor, the cellwall thickening was much more readily observed than it was inunstained cells prepared by this method. Rastogi and col-leagues, who first described the outer layer by ruthenium redstaining (32), suggested that it consisted of polysaccharidesbecause ruthenium red binds to acidic polysaccharides (16).However, ruthenium red can also bind to lipids (3). Rutheniumred staining in the absence of osmium tetroxide and the silverproteinate staining of our samples did not reveal any differ-ences between aerobically and anaerobically cultured bacilli(data not shown).

The thickened walls of M. bovis BCG and M. tuberculosiscells reported here may help the bacilli to survive in oxygen-deficient conditions in vivo. Although M. tuberculosis is knownto survive within the body for long periods, much less is knownabout M. bovis BCG. However, the report of an AIDS patientdeveloping M. bovis BCG adenitis 30 years after being vacci-nated with BCG as a child (33) indicates that M. bovis BCG canpersist for many years. Just how tubercle bacilli survive in vivoand in what form have been subject to conjecture for manyyears. The finding that M. tuberculosis has a sigma factor sim-ilar to the SigF sporulation sigma factor from Streptomycescoelicolor has led to the suggestion that M. tuberculosis mayenter a spore-like state during persistent infection (9). How-ever, it has been previously proposed that the thick nature ofthe mycobacterial wall may negate the necessity for these bac-teria to sporulate (29), and the thickened cell wall reportedhere may support that hypothesis. We did not find any evi-dence for a spore-like or coccoid morphology. Scanning elec-tron microscopy of our bacilli showed that they had a typicalrod-shaped morphology, and Ziehl-Neelsen staining showedthat they were acid fast (data not shown). However, we fre-quently noticed by TEM that the anaerobically cultured bacilliappeared shorter in length, although it was not possible toassess this quantitatively. The cell wall thickening may be themanifestation of the dormant state of M. tuberculosis, andinvestigations are under way to determine the clinical signifi-cance of these findings by TEM analysis of granulomas.

The thickened wall described in this report may act as aprotective coat or mantle and may offer protection againsthostile environments such as the toxic conditions associatedwith granulomas. This is significant when one considers thatthe oxidative stress response in M. tuberculosis is dysfunctionalwith an inactive oxyR gene (10, 38) and that the quantities oractivities of catalase and superoxide dismutase of anaerobicallycultured bacilli are decreased (43). It was particularly interest-ing to see that at the point where two bacilli were in directphysical contact (as are the two pairs of bacilli in Fig. 1B) thethickened outer layer was absent (see also Fig. 2C) but thatcells in close contact (Fig. 1B) clearly had generated their ownthickened outer layers. The direct contact between the elec-tron-transparent layers, which is considered the most hydro-phobic domain (3), would create a hydrophobic focus, therebypossibly negating the need to develop the outer layer.

The fact that the aerobically grown stationary-phase culturesof M. bovis BCG did not develop the thick cell wall underlinesthe importance of oxygen deficiency as the trigger for its syn-thesis. However, it is possible that lower oxygen tension is onlya proxy for the stimulus that causes the thickened cell wall todevelop. As most tuberculosis is due to reactivation of infec-tion this cell wall thickening could be an important virulenceand survival determinant. The facts that M. smegmatis survivespoorly under low oxygen tension and does not develop a thickcell wall suggest that such a morphological adaptation is im-portant in anaerobic survival. It is intriguing that M. smegmatis

VOL. 180, 1998 CELL WALL THICKENING AND a-CRYSTALLIN LOCALIZATION 805

on August 23, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

806 CUNNINGHAM AND SPREADBURY J. BACTERIOL.

on August 23, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

does not possess the 16-kDa a-crystallin homolog (46), and itwill be interesting to determine whether this has some bearingon the inability of M. smegmatis to survive prolonged anaerobicculture and generate the thickened outer layer. The ability tosurvive anaerobiosis may only be a feature of the slow-growingpathogenic mycobacteria and may reflect their successful evo-lution in adapting to the low oxygen potential of host tissues.Previous studies have suggested that the ability to tolerateanaerobic conditions may be pivotal to the pathogenicity ofM. tuberculosis. M. tuberculosis H37Rv has a lower respiratoryrate than avirulent strain M. tuberculosis H37Ra, and the rateof respiration is less inhibited by lower oxygen tension inH37Rv than in H37Ra (20). It could also be argued that alower respiratory rate may provide the bacilli with sufficienttime to adapt to lower oxygen tension in preparation for en-tering a state of dormancy, and the ordered metabolic shut-down observed by Wayne and Lin (43) probably reflects this.Moreover, during granuloma formation the rate of oxygendepletion is likely to be slow enough to permit adaptation to ananaerobic environment.

Comparing protein expression profiles between the threeculture states in M. bovis BCG revealed the upregulation of the16-kDa smHSP or a-crystallin homolog in response to low oxy-gen tension. During the course of our work Yuan et al. pub-lished a report which showed that the 16-kDa protein was up-regulated by M. tuberculosis during stationary phase or underreduced oxygen tension (46). M. bovis BCG and M. tuberculosis16-kDa smHSPs are members of the smHSP superfamily (5)and have homology in an 80-amino-acid region at the C ter-minus known as the a-crystallin domain. They also have atendency to aggregate; Chang and coworkers (6) showed thatthe 16-kDa protein aggregates as an oligomer with a molecularmass of 149 6 8 kDa (consisting of nine monomers, in atrimer-of-trimers organization). These proteins also preventthe thermal aggregation of other proteins (6, 21, 23, 46), butthey do not prevent the loss of enzymic activity. The smHSPsact in an ATP-independent manner unlike the GroEL andDnaK HSPs (4, 6, 15, 23, 30). This is significant because theyare frequently developmentally regulated, usually when andwhere metabolic activity is minimal such as in the eye lens (4)and, intriguingly, they are increasingly being discovered insporulating microorganisms (17, 19, 36, 37). Interestingly,there appears to be an association between the expression ofthese proteins and quiescence induced by oxygen limitation.For example, the M. tuberculosis 16-kDa protein was found tohave ;41% similarity at the amino acid level (24) with the21-kDa spore protein (SP21) or the low-molecular-weight HSPof Stigmatella aurantiaca (17), which is synthesized during fruit-ing body and spore formation but which is not found in vege-tative cells. Spore production could be induced by anoxia (14),and oxygen limitation was found to induce the synthesis of thisprotein (18). The encysted embryos of the brine shrimp Ar-temia franciscana, which can survive continuous anoxia formore than 4 years and which have an undetectable metabolicrate during that time, also highly express a smHSP (7).

The three patterns of 16-kDa protein localization identified

suggest that this protein is likely to have multiple targetsthroughout the cell and that it is integral to the stabilization ofcellular structure during the dormant state. As the 16-kDaprotein was localized to the cell wall, it is possible that somemoieties the 16-kDa protein chaperones are also localized tothis region. The only other prokaryotic smHSP to be studied interms of immunoelectron microscopy localization is the SP21protein of S. aurantiaca (28). This protein has many of thepatterns of distribution reported here, although what the SP21protein binds to was not identified. It is likely that the 16-kDaprotein interacts in some way with the peptidoglycan layer ofthe mycobacterial cell, as indicated by the fact that it wasobserved to localize to fibrous structures. The clustering in thecytoplasm we observed was similar to the cytoplasmic cluster-ing seen in S. aurantiaca (28). It has been proposed that theSP21 protein of S. aurantiaca may protect RNA species re-quired for germination from being degraded (18), and as thesmHSPs may be able to bind RNA (31), one could speculatethat the mycobacterial 16-kDa protein may play a role in theprotection of RNA. For example, Wayne (42) described howmicroaerobically cultured bacilli were able to multiply in asynchronous manner 8 h after reaeration of the culture; thissuggests that much of the machinery required for this round ofreplication, including specific RNA and proteins, must alreadybe in place. These would require protection from damageduring dormancy and this protection could involve the chap-erone function of the 16-kDa protein. However, the findingthat there was little or no association of clusters with chromo-somal material suggests that the protein does not play a role inprotecting DNA during dormancy.

Yuan et al.’s data and our findings strongly suggest thatoxygen deficiency is an important trigger for the synthesis ofthe 16-kDa protein. Therefore, its expression is not likely to bepart of a general stress response. This agrees with our reverse-transcription PCR findings, which show that whereas the my-cobacterial groEL and dnaK analogs are poorly transcribedafter 2 weeks of microaerobic and anaerobic culture (withexpression levels lowest in the anaerobic culture), the 16-kDaprotein gene was highly expressed (8).

These results suggest that tubercle bacilli may adapt to lowoxygen conditions by developing a thickened cell wall and thatthe 16-kDa protein confers an advantage on the bacilli duringits dormant phase by stabilizing and protecting cell structures.The fact that both these phenomena were detected after 2weeks of culture under low oxygen suggests that they may bedevelopmentally regulated by similar mechanisms. Experi-ments to determine the biochemical components of the thick-ened outer layer and the cellular components the 16-kDa pro-tein binds to are in progress. Understanding the roles that cellwall thickening and the expression of the 16-kDa protein playin mycobacterial dormancy may lead to novel strategies to killpersisters or prevent the reactivation of disease.

ACKNOWLEDGMENTS

This work was funded by the Glaxo Wellcome Action TB initiative.We thank Paul Stanley and Lesley Tompkins for preparing the

FIG. 3. Immunogold electron microscopy localization of the 16-kDa a-crystallin protein of M. bovis BCG. (A) Low level of prevalence of the 16-kDa protein in thelow-OD aerobic control. (B) Bacillus from a 12-week anaerobic culture showing localization to the periphery of the cell. (C) Twenty-six-week anaerobically culturedbacillus showing localization of the 16-kDa protein to the cell wall (marked by the arrowheads). (D) Twenty-six-week anaerobically cultured bacillus showing the 16-kDaprotein associated with a splayed arrangement of fibrous structures. (E) Bacillus cultured for 26 weeks microaerobically showing localization of the protein to fibrilsthought to be peptidoglycan (indicated by arrowheads). (F) Four-week anaerobically cultured bacillus; lines of colloidal gold labelling are seen despite the absence ofdetailed fibril ultrastructure. (G) Four-week anaerobically cultured bacillus showing clusters both associated with the intracellular environment and at the cell periphery.(H) Bacillus cultured for 2 weeks anaerobically. Note the lack of localization to the chromosome in the center of the cell. Bars, 100 nm except for panel A (300 nm)and panels C and H (200 nm).

VOL. 180, 1998 CELL WALL THICKENING AND a-CRYSTALLIN LOCALIZATION 807

on August 23, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

samples for electron microscopy and for their expert technical advice.We also thank Paul Stanley and Peter Whittle for preparing the elec-tron micrographs. We thank Mike Osborne and Kerstin Williams (Bir-mingham) for their constructive comments regarding the TEM work,and we particularly thank Philip Draper (NIMR, London) for hishelpful advice and suggestions. We also thank Neil Freeman (GlaxoWellcome) for performing the N-terminal sequencing and Ken Dun-can, Karen Kempsell, and Pauline Lukey (Glaxo Wellcome, Steven-age) for their support, many discussions, and helpful advice.

REFERENCES

1. Barksdale, L., and K.-S. Kim. 1977. Mycobacterium. Bacteriol. Rev. 41:217–372.

2. Bloom, B. R., and J. L. Murray. 1992. Tuberculosis: commentary on areemergent killer. Science 257:1055–1064.

3. Brennan, P., and H. Nikaido. 1995. The envelope of mycobacteria. Annu.Rev. Biochem. 64:29–63.

4. Buchner, J. 1996. Supervising the fold: functional principles of molecularchaperones. FASEB J. 10:10–19.

5. Caspers, G.-J., J. A. M. Leunissen, and W. W. de Jong. 1995. The expandingsmall heat shock family, and structure predictions of the conserved “a-crystallin domain”. J. Mol. Evol. 40:238–248.

6. Chang, Z., T. P. Primm, J. Jakana, I. H. Lee, I. Serysheva, W. Chiu, H. F.Gilbert, and F. A. Quiocho. 1996. Mycobacterium tuberculosis 16-kDa antigen(hsp16.3) functions as an oligomeric structure in vitro to suppress thermalaggregation. J. Biol. Chem. 271:7218–7223.

7. Clegg, J. 1997. Embryos of Artemia franciscana survive four years of contin-uous anoxia: the case for complete metabolic rate depression. J. Exp. Biol.200:467–475.

8. Cunningham, A. F., and C. L. Spreadbury. Unpublished data.9. DeMaio, J., Y. Zhang, C. Ko, D. B. Zhang, and W. R. Bishai. 1996. A

stationary phase, stress response sigma factor from Mycobacterium tubercu-losis. Proc. Natl. Acad. Sci. USA 93:2790–2794.

10. Deretic, V., W. Philipp, S. Dhandayuthapani, M. H. Mudd, R. Curcic, T.Garbe, B. Heym, L. E. Via, and S. T. Cole. 1995. Mycobacterium tuberculosisis a natural mutant with an inactivated oxidative-stress regulatory gene:implications for sensitivity to isoniazid. Mol. Microbiol. 17:889–900.

11. Dubos, R. J. 1950. The effect of organic acids on mammalian tubercle bacilli.J. Exp. Med. 92:319.

12. Dubos, R. J. 1953. Effect of the composition of the gaseous and aqueousenvironment on the survival of tubercle bacilli in vitro. J. Exp. Med. 97:357.

13. Dubos, R. J. 1953. Effect of ketone bodies and other metabolites on thesurvival and multiplication of staphylococci and tubercle bacilli. J. Exp. Med.98:145.

14. Gerth, K., R. Metzger, and H. Reichenbach. 1993. Induction of myxosporesin Stigmatella aurantiaca (Myxobacteria)—inducers and inhibitors of myxo-spore formation, and mutants with a changed sporulation behaviour. J. Gen.Microbiol. 139:865–871.

15. Groenen, P. J. T. A., K. B. Merck, W. W. de Jong, and H. Bloemendal. 1994.Structure and modification of the junior chaperone a-crystallin—from lenstransparency to molecular pathology. Eur. J. Biochem. 225:1–19.

16. Hayat, M. A. 1989. Principles and techniques of electron microscopy, 3rd ed.Macmillan, Basingstoke, United Kingdom.

17. Heidelbach, M., H. Skladny, and H. U. Schairer. 1993. Purification andcharacterization of SP21, a development-specific protein of the myxobacte-rium Stigmatella aurantiaca. J. Bacteriol. 175:905–908.

18. Heidelbach, M., H. Skladny, and H. U. Schairer. 1993. Heat shock anddevelopment induce synthesis of a low-molecular-weight stress-responsiveprotein in the myxobacterium Stigmatella aurantiaca. J. Bacteriol. 175:7479–7482.

19. Henriques, A. O., B. W. Beall, and C. P. Moran, Jr. 1997. CotM of Bacillussubtilis, a member of the a-crystallin family of stress proteins, is inducedduring development and participates in spore outer coat formation. J. Bac-teriol. 179:1887–1897.

20. Heplar, J. Q., C. E. Clifton, S. Raffel, and C. M. Futrelle. 1954. Virulence ofthe tubercle bacillus. I. Effect of oxygen tension upon respiration of virulentand avirulent bacilli. J. Infect. Dis. 94:90–98.

21. Horwitz, J. 1992. a-Crystallin can act as a molecular chaperone. Proc. Natl.

Acad. Sci. USA 89:10449–10453.22. Imaeda, T., F. Kanetsuna, and B. Galindo. 1968. Ultrastructure of cell walls

of genus Mycobacterium. J. Ultrastruct. Res. 25:46–63.23. Jakob, U., M. Gaestel, K. Engel, and J. Buchner. 1992. Small heat shock

proteins are molecular chaperones. J. Biol. Chem. 268:1517–1520.24. Kempsell, K. E. Personal communication.25. Kochi, A. 1991. The global tuberculosis situation and the new control strat-

egy of the World Health Organization. Tubercle 72:1–6.26. Laemmli, U. K. 1970. Cleavage of structural proteins during the assembly of

the head of bacteriophage T4. Nature (London) 227:680–685.27. Lee, B.-Y., S. A. Hefta, and P. J. Brennan. 1992. Characterization of the

major membrane protein of virulent Mycobacterium tuberculosis. Infect. Im-mun. 60:2066–2074.

28. Lunsdorf, H., H. U. Schairer, and M. Heidelbach. 1995. Localization of thestress protein SP21 in indole-induced spores, fruiting bodies, and heat-shocked cells of Stigmatella aurantiaca. J. Bacteriol. 177:7092–7099.

29. Minnikin, D. E. 1982. Lipids: complex lipids, their chemistry, biosynthesisand roles, p. 95–184. In C. Ratledge and J. Stanford (ed.), The biology of themycobacteria, vol. 1. Physiology, identification and classification. AcademicPress, Inc., London, United Kingdom.

30. Nicholl, I. D., and R. A. Quinlan. 1994. Chaperone activity of a-crystallinmodulates intermediate filament assembly. EMBO J. 13:945–953.

31. Nover, L., K. D. Scharf, and D. Neumann. 1989. Cytoplasmic heat shockgranules are formed from precursor particles and are associated with aparticular set of mRNA. Mol. Cell. Biol. 9:1298–1308.

32. Rastogi, N., C. Frehel, and H. L. David. 1986. Triple-layered structure ofmycobacterial cell wall: evidence for the existence of a polysaccharide-richouter layer in 18 mycobacterial species. Curr. Microbiol. 13:237–242.

33. Reynes, J., C. Perez, I. Lamaury, F. Janbon, and A. Bertrand. 1989. BacilleCalmette-Guerin adenitis 30 years after immunization in a patient withAIDS. J. Infect. Dis. 160:727.

34. Reynolds, E. S. 1963. The use of lead citrate at high pH as an electronopaque stain in electron microscopy. J. Cell Biol. 17:208–212.

35. Robertson, D., P. Monaghan, C. Clarke, and A. J. Atherton. 1992. Anappraisal of low-temperature embedding by progressive lowering of temper-ature into Lowicryl HM20 for immunocytochemical studies. J. Microsc. 168:85–100.

36. Sauer, U., and P. Durre. 1993. Sequence and molecular characterization ofa DNA region encoding a small heat shock protein of Clostridium acetobu-tylicum. J. Bacteriol. 175:3394–3400.

37. Servant, P., and P. Mazodier. 1995. Characterization of Streptomyces albus18-kilodalton heat shock-responsive protein. J. Bacteriol. 177:2998–3003.

38. Sherman, D. R., P. J. Sabo, M. J. Hickey, T. M. Arain, G. G. Mahairas, Y.Yuan, C. E. Barry III, and C. K. Stover. 1995. Disparate responses tooxidative stress in saprophytic and pathogenic mycobacteria. Proc. Natl.Acad. Sci. USA 92:6625–6629.

39. Sudre, P., G. ten Dam, and A. Kochi. 1992. Tuberculosis: a global review ofthe situation today. Bull. W. H. O. 70:149–159.

40. Verbon, A., R. A. Hartskeerl, A. Schuitema, A. H. J. Kolk, D. B. Young, andR. Lathigra. 1992. The 14,000-molecular weight antigen of Mycobacteriumtuberculosis is related to the alpha-crystallin family of low-molecular-weightheat shock proteins. J. Bacteriol. 174:1352–1359.

41. Wayne, L. G. 1976. Dynamics of submerged growth of Mycobacterium tuber-culosis under aerobic and microaerophilic conditions. Am. Rev. Respir. Dis.114:807–811.

42. Wayne, L. G. 1977. Synchronized replication of Mycobacterium tuberculosis.Infect. Immun. 17:528–530.

43. Wayne, L. G., and K. Lin. 1982. Glyoxylate metabolism and adaptation ofMycobacterium tuberculosis to survival under anaerobic conditions. Infect.Immun. 37:1042–1049.

44. Wayne, L. G., and D. Salkin. 1956. The bacteriology of resected tuberculouspulmonary lesions. I. The effect of interval between reversal of infectiousnessand subsequent surgery. Am. Rev. Tuberc. Pulm. Dis. 74:376–387.

45. Wayne, L. G., and H. A. Sramek. 1994. Metronidazole is bactericidal todormant cells of Mycobacterium tuberculosis. Antimicrob. Agents Chemo-ther. 38:2054–2058.

46. Yuan, Y., D. D. Crane, and C. E. Barry III. 1996. Stationary phase-associatedprotein expression in Mycobacterium tuberculosis: function of the mycobac-terial a-crystallin homolog. J. Bacteriol. 178:4484–4492.

808 CUNNINGHAM AND SPREADBURY J. BACTERIOL.

on August 23, 2020 by guest

http://jb.asm.org/

Dow

nloaded from