A New Insight into the Mycobacterial Cell Envelope Architecture by the Localization of Antigens in Ultrathin Sections

Zbl. Bakt. 275, 287-302 (1991) © Gustav Fischer Verlag, StuttgartlNew York

A New Insight into the Mycobacterial Cell Envelope Architecture by the Localization of Antigens in Ultrathin Sections

NALIN RASTOGI l , RA YMOND HELLl02, and HUGO L. DA vml

1 Unite de la Tuberculose et des Mycobacteries, and 2 Unite de Biologie des Membranes, Institut Pasteur, 75724 Paris Cedex 15, France

With 6 Figures· Received October 25, 1990 . Accepted in revised form March 6, 1991

Summary

In an attempt to have a better insight into the mycobacterial cell envelope architecture, various subcellular fractions of Mycobacterium avium were prepared and characterized chemically and ultrastructurally. The various fractions corresponding to the mycobacterial "capsular material", outer layer, cell wall skeleton, cytoplasmic membrane, and cytosol as well as intact bacteria were then used to raise antisera in rabbits. The antisera so raised were then used to immunolabel the intact bacteria prior to embedding in epon. In parallel studies, bacteria were processed by a novel gelatin-uranyl acetate-low temperature Lowicryl HM20 embedding which preserved mycobacterial antigens, permitting to immunolabel antigens on ultrathin sections. Immunolabelling of epon-embedded intact bacteria showed that in the tripartite structure of the bacterial cell envelope, the middle electron-transparent layer acted as a barrier, not permitting the antibodies to penetrate into deeper structures. Immunolabelling of ultrathin sections showed that mycobacteria were surrounded by a "capsule" containing specific surface antigens with a glycocalyx-like topography, and that the intermediate electron transparent layer which separated the surface amphiphils from the inner arabinogalactan-peptidoglycan layer, was a virtual no man's land as it only seldom contained a single gold particle irrespective of the various antisera used. Furthermore, location of various layers in the cell envelope of M. avium using antisera raised against the subcellular fractions prepared was in agreement with chemical and ultrastructural data. A cell envelope model compatible with chemical, ultrastructural and immunolabelling data is proposed and its validity discussed.

Abbreviations used in this paper (in alphabetical order): AG = arabinogalactan; BSA = bovine serum albumin; CAP = "capsule-like structure"; CM = cytoplasmic membrane; CYT = cytosol; CWS = cell wall skeleton; DAP = diaminopimelic acid; EDTA = -ethylenediamine tetraacetic acid; EM = electron microscopy; ETL = electron-transparent layer; F = plane of fracture (in freeze etching/fracture); GAR = goat antirabbit; IL = inner layer; LAM = -lipoarabinomannan; LOS = lipooligosaccharide; OL = outer layer; P = proteins; PAG = protein A-gold; PBS = phosphate buffer saline; PG = peptidoglycan; PGL = phenolglycolipid; PIM = mannose-containing phospholipids; PL = phospholipids; PTA =

phosphotungstic acid; SDS = sodium dodecyl sulphate; SEM = scanning electron microscopy; SL = sulpholipid; TDM = trehalose dimycolate; TS = tuberculostearic acid; WC = whole cell.

19 Zbl. Bakl. 275/3

288 N. Rastogi, R. Hellio, and H. L. David

Zusammenfassung

In dem Bemiihen urn einen besseren Einblick in den Aufbau der Zellhiille von Mycobakterien wurden verschiedene subzellulare Fraktionen von Mycobacterium avium hergestellt und chemisch sowie ultrastrukturell charakterisiert. Die dem "Kapselmaterial" der Mycobakterien ensprechenden verschiedenen Fraktionen, namlich aulSere Schicht, Zellwandskelett, Zytoplasmamembran und Cytosol sowie intakte Bakterien dienten zur Herstellung von Antiseren in Kaninchen. Die auf diese Weise hergestellten Antiseren wurden dann zur Immunmarkierung der intakten Bakterien vor der Einbettung in Epon verwendet. In Paralleluntersuchungen wurden die Bakterien in das neuartige Niedrigtemperatur-GelatineUranylacetat Lowicryl HM20 eingebettet, welches die Mycobakterien-Antigene erhalt und die Immunmarkierung der Antigene auf Ultradiinnschnitten erlaubt. Die Immunmarkierung von in Epon eingebetteten intakten Bakterien zeigte, daIS in der dreiteiligen Struktur der bakteriellen Zellhiille die mittlere elektronentransparente Schicht als Schranke wirkte, die ein Eindringen der Antikorper in die tieferen Strukturen nicht zulieK Die Immunmarkierung von Ultradiinnschnitten zeigte, daIS die Mycobakterien von einer "Kapsel" umgeben sind, die spezifische Oberflachenantigene mit einer Glycocalyx-artigen Topographie enthalt und daIS die elektronentransparente Zwischenschicht, die die Amphiphile der Oberflache von der inneren Arabinogalactan-Peptidoglycan-Schicht trennt, praktisch ein "Niemandsland" war, da sie nur selten einzelne Goldpartikel enthielt, obwohl verschiedene Antiseren verwendet worden waren. Dariiber hinaus stimmte bei Verwendung von gegen die subzellularen Fraktionen hergestellten Antiseren die Lokalisierung der verschiedenen Schichten in der Zellhiille von M. avium mit den chemischen und ultrastrukturellen Daten iiberein. Es wird ein zu den chemischen, ultrastrukturellen und Immunmarkierungs-Daten passendes Modell der Zellhiille vorgeschlagen und seine Giiltigkeit diskutiert.

Introduction

Recent reports have shown that susceptibility to drugs of the multiple drug-resistant Mycobacterium avium could be enhanced by compounds known to disrupt their cell envelope (25). Other reports have shown that diffusion of drugs through the cell envelope could be facilitated by lipophilic carriers (7, 24, 28). These findings partially supported the view that exclusion might be an important mechanism of drug-resistance in mycobacteria (6, 9, 10, 15, 23), and gave further impetus to the proposal that the development of the rationale for chemotherapy against mycobacterial infections must depend on a better understanding of the molecular structure and topology of their envelope.

To a varying degree, the molecular structure of the cell envelope may participate in mycobacterial adherence, phagocytosis, survival, multiplication and antigen processing in target macrophages (18, 19). It seems desirable to distinguish, among the components in their cell envelope, those that may contribute to favourable and protective host immune reactions as opposed to those that may contribute to unfavourable host (immunopathological?) reactions (29). Thus in general, knowledge about the molecular structure of the outer layer of the tripartite mycobacterial cell wall must also contribute to the understanding of the invasion of macrophages during infection and may provide the rationale for designing a new generation of vaccines against mycobacterial diseases such as tuberculosis and leprosy.

Recent progress in electron microscopy processing has shown that embedding of mycobacteria using the Lowicryl method preserved antigens, making it possible to locate them on ultrathin sections (4, 18,26). These findings prompted us to examine

Mycobacterial Cell Envelope 289

M. avium, using antisera raised in rabbits against various subcellular bacterial fractions. These fractions were characterized morphologically, using transmission and scanning electron microscopy, and also chemically prior to injecting them in to rabbits. It is the purpose of this report to show that the methodology used has contributed new information about the architecture of the mycobacterial cell envelope.

Material and Methods

Bacteria and growth. Mycobacterium avium ATCC 15769 (smooth colony type) used in this study was from our own culture collection. The bacteria were grown in the complete 7H9liquid medium (Difco Laboratories, Detroit, Mich.) to an optical density of 0.15 measured at 650 nm using a Coleman junior II spectrophotometer which corresponded to about 108 viable counts/ml.

Preparation of cellullar and subcellular fractions. The methods used for preparing various cellular and subcellular fractions have been described in detail elsewhere (8), and the steps have been summarized in Figs. 1 A and 1 B. Each of the subcellular fractions used was characterised in parallel using various electron microscopical methods (see below) and also various chemical markers like diaminopimelic acid (DAP), mycolates, tuberculostearic acid (TS), mycosides-C and major proteins. The chemical analysis of the mycobacterial fractions was performed as reported earlier (8).

Briefly, the whole cells were either submitted to various extraction procedures (Table lA) or ultrasonicated to isolate various subcellular structures (Table IB). In the first case, the bacteria were extracted using the same volume of 10% (w/v) solution of sodium dodecyl sulphate (SDS) in distilled water under continuous agitation for 1 h. The bacterial suspension was centrifuged at 5000 X g and the clear supernatant and a white fluffy layer on the top of the bacterial sediment were carefully removed. This SDS supernatant was then precipitated at room temperature by adding five times its volume of pure ethanol. The precipitate was washed once in ethanol, dissolved in a small volume of PBS (0.1 M, pH 7.0), and stored at -20°C until used for analysis. Alternatively, the bacteria were successively extracted using a 4% (v/v) solution of Triton-XI00 in PBS-EDTA (1 h at 37°C with agitation), a 2:1 v/v mixture of chloroform-methanol (30 min), and diethyl ether after mild saponification in 4% (w/v) KOH in methanol.

In the second case (Fig. IB), the bacteria were disintegrated by sonication for 1 h at 15 min intervals in an ice bath using an Ultra-cell sonicator (1.2 cm diameter probe set at 300 watts, Sonics & Materials Inc., USA), the broken cells were centrifuged at 5000 x g and the supernatant was recentrifuged at 25000 x g. The sediment (cell wall) was delipidated using 2:1 v/v chloroform-methanol to obtain a delipidated cell wall. The 25000 x g supernatant, on the other hand, was recentrifuged at 100000 x g to obtain the cytoplasmic membranes (sediment) and the cytosol (supernatant), respectively.

Based on the EM (Fig. 2 and 3) and chemical characterization of various fractions (Table 1), as described under the Results section, they were designated as: whole cell (WC), sodium dodecyl extract (SDS), outer layer (OL), cell wall skeleton (CWS), cytoplasmic membrane (CM) and cytosol (CYT) fractions.

Preparation of antisera. The antisera were raised in rabbits by inoculating untreated and subcellular fractions of M. avium, and characterized as reported earlier (8, 11).

Negative staining, shadow replica and scanning electron microscopy. The bacteria as well as various subcellular fractions were processed differently, depending on the electron micro

. scopical method used. The samples were fixed for 2 h at 4°C in 1.25% (w/v) paraformaldehyde and 2.5% (w/v) glutaraldehyde in cacodylate buffer (0.1 M, pH 7.2), washed in the same buffer, and processed for negative staining using phosphotungstic acid (PTA) and

. scanning electron microscopy (SEM) as reported earlier (23).


For shadow replica, 1 to 2 drops of samples were placed on a piece of split mica, the preparations were air dried and then shadowed with platinum at a 45° angle and carbon at a 90° angle in a Balzer BA-360 M vacuum evaporator. The carbon-platinum replicas were attached to mica supports and soaked overnight in 50% chromic acid. The replicas were rinsed twice in deionized water and mounted on Formvar-coated grids.

Immunolabelling of intact bacteria. For these studies, the bacteria harvested from liquid medium were washed twice with phosphate buffer saline (PBS) and incubated with various antisera raised in rabbits in excess (1 : 1 ratio) for 30 min at room temperature. The preparations were rinsed with PBS and incubated with a secondary probe coupled to colloidal gold (Goat anti-rabbit IgG antibodies coupled to 5 nm gold particles, GAR-S EM grade from Janssen Laboratories, Belgium) in excess (1: 1 ratio) for 30 min at room temperature. The bacteria were then fixed overnight at 4°C in a mixture of 1.25% (w/v) paraformaldehyde and 2.5% (w/v) glutaraldehyde in cacodylate buffer (0.1 M, pH 7.2), washed in the same buffer, and processed for transmission electron microscopy using epon embedding as reported earlier (23). The sections were mounted on formvar-coated copper grids and observed under a Philips CM-12 transmission electron microscope.

Gelatin-Lowicryl HM20 embedding of mycobacteria. The mycobacterial antigens in ultrathin sections of M. avium could be satisfactorily preserved using a recently developed gelatin-Lowicryl HM20 embedding (4, 26). The bacteria were fixed with 1.25% (w/v) glutaraldehyde in cacodylate buffer (0.1 M, pH 7.2) for 1 h at 4°C, washed in cacodylate buffer to remove the excess of aldehydes, treated for 30 min with 50 mM NH4Cl prepared in the same buffer (to block any remaining free aldehyde groups) and kept overnight in cacodylate buffer at 4°C. On the next day, the bacteria were embedded in 10% (w/v) gelatin in water, postfixed with 0.5% (w/v) uranyl acetate in veronal buffer, dehydrated in graded ethanol series, and embedded using HM20 grade Lowicryl below -30°C as described previously (32). The choice of HM20 grade instead of K4M grade Lowicryl, and pretreatment of bacteria with uranyl acetate before the dehydration step, was based on recent observations by Benichou et al. (44).

Immunolabelling of ultrathin sections. Ultrathin sections of gelatin-Lowicryl HM20-embedded M. avium were mounted on formvar-carbon-coated 200 mesh nickel grids, labelled using various antisera raised in rabbits, and the labelling was then visualized using a secondary probe. Two different secondary probes used in this study were either GAR-S or protein-A coupled to 7 nm gold particles ( PAG-7). The first probe permitted a much higher labelling of bacteria than the second one, however the visual quantification of bacteria labelled by various antisera could be more easily performed using the PAG-7 probe due to the presence of larger gold particles. Consequently, we used both the probes in parallel. Bovine serum albumin (BSA) at 0.5% or 1 % (w/v) in PBS was used at all the steps to further block free aldehydes, if any, and to avoid non-specific labelling.

For immunolabelling, the grids were floated at room temperature on drops of PBS-BSA 1 %,3"0 min; antisera 11100 or 11200 diluted in PBS-BSA 1 %, 1 h; 3 X PBS-BSA 0.5%; 11 .100 diluted PAG-7 or 1/200 diluted GAR-S in PBS-BSA, 1 h; 3 X PBS-BSA 0.5%; 3 X double distilled, deionized water, respectively; and were observed under the electron microscope after staining with uranyl acetate and lead citrate.

Results

Preliminary observations

M. avium bacteria as well as various subcellular fractions prepared (Figs. 1A and lB) were characterized chemically (Table 1) as reported earlier (8). Unlike the WC fraction which contained all four chemical markers used (DAP, mycolates, TS and mycosidesC), the SDS, CM and CIT fractions were lacking these markers, the CWS fraction contained only DAP and mycolates and traces of mycosides-C, whereas the OL frac-


~. avium, Whole cells (WC)

(Fig. 2A and 3A)

10% (w/v) SOS, Vortex Ih

5,OOOXgl

sedi~ent (di scarded)

, supernatant (prec i pi ta ted wi th ethanol, and dissolved in PBS, pH 7.0

;SOS Extract

I SOS Fraction I

sedi'ment

4%(v/v)Triton X-IOO in PBS-EOTA Ih at 37"C with agitation

l5,OOO x g

.1 sedlment (Fig. 2B,

3B & 3D)

1 Extract (Fig. 3C) Oi scarded

CHC1 3:CH30H

(2:1),30 min.

5,000 x 9 , sediment (Fig.2C)

4%(v/v)KOH i n met~ano 1 methyl cellosolve

5,000 x 9

I Extract (loosely-

bound 1 i pi ds

=outer 1 ayer,

I OL Fraction I

super~atant (Fig. 20)

Discarded after chemi ca 1 ana lys is.

Oi scarded after chemi ca 1 analysis.

[IJ ~. ~, Whole cells,Fig.2A & 3A

I WC Fractionl

Sonication, Ih 300 watts, 1. 2 cm probe 4 times 15 min.

5,000 x 9

sedi'ment ,

Supernatant (Fig. 2E) I Oi scarded 25,OO~ x 9

1 1 Sediment Supernatant

(Cell Wall) I

Fig. 2F 10,0,000 x 9 , , Sediment Supernatant

CHll : CH30H ;Cytop 1 asmi c ;Cytosol membranes Fig. 3G

(2:1),30 min. (Fi9. 3F & 31) !CYT Fraction! 5,000 x 9

ICM Fraction! 1 I

Sediment Supernatant

(delipidated cell wall ) (Wall lipids)

;Cell wall skeleton Oi scarded

Fig. 3E & 3H

Icws Fractionl

Fig. 1. Schematic representation illustrating the steps involved in the preparation of various subcellular fractions of M. avium. Fig. 1A shows the steps involving chemical extraction of whole bacteria whereas Fig. IB represents steps involving sonication of the bacteria. Parallel chemical and ultrastructural controls are represented in Table 1 and Figs. 2 and 3.


Table 1. Characterization of M. avium and its subcellular fractions used as antigens

Chemical data

Antigenic Diamino- Mycolates Tuberculo- Mycosides- Major proteins fraction pimelic acid stearic acid C (approximate mol.wt.)

WC + + + + Several bands SDS 20 kDa

OLb Trace + + + ND CWS + + Trace ND CM ND 86, 58, 40, and 26 kD CYT ND 58, 43, 36, and 19 kD

a Glycoprotein, as this band on a SDS-PAGE gel stained both with Coomassie blue and periodic acid-silver nitrate.

b In the cell wall, the outer layer represented about 40% of the weight as compared to the whole cells where it represented about 27% on a weight basis.

ND = Not done.

tion contained mycolates, TS and mycosides-C with traces of DAP. These chemical data have been dealt with in detail in an earlier paper (8).

The ultrastructural observations using SEM and negative staining along with shadow replica preparation for cell surface characterization of intact bacteria and various fractions prepared have been summarized in Figs. 2 and 3. A detailed description of the EM images has been given in the legends to Figs. 2 and 3 and will not be repeated here. Based on the steps involved in preparing various fractions (schematic representation, Figs. 1A and lB), along with parallel chemical analysis (Table 1) and EM data (Figs. 2 and 3), it was concluded that the fractions prepared were sufficiently pure to be used for raising antisera in rabbits.

Immunogold labelling of intact bacteria

When whole cells were incubated with various antisera and the labelling was revealed using the GAR-5 secondary probe, the labelling remained only at the level of the outermost layer of the mycobacterial cell envelope (Fig. 4). All the antisera including the anti-CWS, CM or CYT fractions (Figs. 4B, 4C and 4F, respectively), were neither able to label the basal peptidoglycan (PG) layer and the intermediate electron transparent layer (ETL) in the mycobacterial cell envelope, nor deeper structures like the cytoplasmic membrane and the bacterial cytoplasm. This was in agreement with earlier findings with immunolabelled M. avium (11), M. intracellulare (30) and M. leprae (20). These observations showed that antibodies were unable to penetrate beyond the mycobacterial outer layer (OL) and that the ETL beneath the OL structure surrounded the basal PG layer without discontinuity.

As all the antisera fractions only labelled the bacterial surface, we assessed the percentage of labelled bacteria in order to determine whether differences in the relative labelling by various antisera could be established. These data have been summarized in Table 2. Labelling of the bacterial surface was lowest with anti-CYT serum, suggesting that CYT antigens were not always expressed at the mycobacterial surface. The fact that less bacteria were labelled when using antisera raised against surface antigens (OL


Fig. 2. Scanning electron microscopy of M. avium. (A): Untreated control bacteria were rodshaped with perfectly smooth outer surface; (B): Bacteria extracted with Triton X-lOO were not ruptured, but their surface became rough; (C): After chloroform-methanol treatment, bacteria formed clumps; (D): Chloroform-methanol extracted bacteria observed after saponification; (E): Sonicated bacteria and (F): Cell wall sediment. Refer to Figs. lA and IB and Materials and Methods section for steps involved in the preparation of various subcellular fractions. Bar = 1 ~m.


Fig. 3. Negative staining (A, B, C, E, F, and G) and shadow replica (D, H, and I) preparations of M. avium. (A): Untreated control bacteria were rod-shaped with a smooth surface and an electron-transparent halo which corresponded to the peripheral apolar region of the bacteria; (B): Triton X-I DO-extracted bacteria lost the halo and the surface revealed fibrillar structures; (C): The Triton supernatant obtained from the treated bacteria showed that surface amphiphils were rearranged to form bilayers; (D): In shadow replica of Tritonextracted bacteria, one could observe both the bacterial surface and the surface amphiphils emerging from it; (E): CWS fraction; (F): CM fraction; (G): CIT fraction; (H): shadow replica of CWS fraction; and (I): shadow replica of CM fraction. Refer to Figs. IA and IB and Materials and Methods section for steps involved in the preparation of various subcellular fractions. Bar = 100 nm.


FigA. Whole intact M. avium bacteria immunolabelled with various antisera prior to embedding in epon resin. The antigen-antibody complexes were revealed using the GAR-5 probe. The antibodies in this method only labelled the bacterial OL in varying proportions which are represented in Table 2. The figure shows bacteria labelled using antisera raised against, respectively; (A): WC control; (B): CWS fraction; (C): CM fraction; (D): SDS fraction; (E): OL fraction; and (F): CYT fraction. Bar = 100 nm.


and SDS fractions) than CM antigens indicated two distinct facts. Firstly, it reflected the differences in the cell surface within the bacterial population. Secondly, it suggested that a significant proportion of surface antigens (but not those common to the OL or SDS antigenic epitopes) were CM antigens transported to the bacterial surface and did not belong to the OL and SDS antigenic fractions. They were possibly proteins which had partly lost their immunogenicity due to denaturation during the drastic steps involved in their extraction.

The above results along with our earlier data concerning the number of precipitation lines observed between SDS, CM and CIT antigens and various antisera fractions analyzed by immunoelectrophoresis (11; Table 2) indicated that some intracellular antigens were indeed exported to the outermost layer of M. avium (as observed by our immunolabelling experiments using the antisera raised against the WC,CIT, and CM fractions). Others were wall-specific antigens, probably assembled outside the cytoplasmic membrane.

Table 2. Immunological and ultrastructural characterization of M. avium. The Table represents the number of precipitin lines obtained during immunoelectrophoresis performed using various antisera fractions raised in rabbits against SDS, CM and CIT antigenic fractions. It also represents the percentage of labelled bacteria embedded in epon, and the cell structures immunolabelled in the sections of the bacteria embedded in gelatin-Lowicryl HM20 using various antisera fractions

Anti- Number of precipitin % of Major cell structures labelled serum lines observed with labelled on gelatin-Lowicryl sections

antigenic fractions bacteria in epon "Cap-

SDS CM CIT sule" OL ETL PG CM CIT

WC 2 6 9 90 +++ ++ ± ++ ++ +++ OL 1 ND 2 30 ++ +++ + ± ± SDS 1 2 2 34 +++ + + ± ± CWS 1 ND S 70 ± ± ± ++ ± ++ CM 1 3 7 90 + + ± + +++ +++ CYT 2 7 14 36 + + ± + ++ +++

ND = Not done.

Immunogold labelling of ultrathin sections

The gelatin-Lowicryl HM20 embedding method had a distinct advantage over mycobacterial embedding in epon as the former also permitted to observe the mycobacterial "capsule". In well contrasted electron micrographs, this method revealed a looselybound capsular structure around mycobacteria (Fig. SF, for example). This "capsule" was sensitive to drastic dehydration steps during epon embedding, and it has recently been shown to contain mycobacterial antigens arranged in a "glycocalyx-like" topography (18, 26). Consequently, for quantifying labelling on ultrathin sections, various bacterial structures were observed (from inside to the outside): bacterial cytoplasm (CIT), cytoplasmic membrane (CM), basal peptidoglycan layer (PG), intermediate electron-transparent layer (ETL), bacterial outer layer (OL), and the "capsule".


Fig. 5. Immunolabelling of ultrathin sections of M. avium embedded in gelatin-Lowicryl. In this method, the immunolabelling of the bacteria was observed using PAG-7 or GAR-S secondary probes. The antisera raised against various subcellular fractions of the bacteria resulted in distinct profiles of labelling. Assessed in 300-500 bacteria in each case, the various bacterial structures immunolabelled by various antisera have been summarized in Table 2. Typical electron micrographs illustrated here represent ultrathin sections labelled using antisera raised against; (A): WC control; (B): CYT fraction; (C): CM fraction; (D): CWS fraction; (E): OL fraction; and (F): SDS fraction. An almost complete absence of grains in the bacterial ETL by any of the antisera used, and the relatively high labelling of the bacterial "capsule" by anti-SDS serum (Fig. SF) are noteworthy. Secondary probe in A, C and E is PAG-7 and GAR-S in B, D and F. Bar = 100 nm.


Although the OL could not always be easily distinguished in gelatin-Lowicryl HM20 ultrathin sections, we presumed that it corresponded to a defined zone within the "capsular matrix" permitting its attachment to the bacterial cell wall through lipidlipid interactions. Indeed, in our earlier paper in which a "capsule-like substance" was first observed when using antibody-coating of intact bacteria during phagocytosis by bone-marrow derived macrophages (epon embedding), the OL could be located within the M. avium "capsule" (12). Consequently, the OL was supposed to be placed within the "capsule", located at about 12 nm distance from the ETL layer.

Using antisera raised against various antigenic fractions, the immunolabelling of various bacterial structures has been summarized in Table 2, and typical electron micrographs are shown in Fig. S. The bacteria were homogenously labelled as expected when using WC and CYT antisera (Fig. SA and SB). When, however, using antisera against CM, CWS, OL and the SDS antigenic fractions, the immunolabelling was not uniformly distributed as above but had distinct patterns. In cells labelled with CM antiserum, the labelling remained peripheral to the cytosol or was found towards the cell envelope (Fig. SC); in cells labelled with anti-CWS, the PG was distinctly labelled but there was some labelling in the CYT and in the OL (Fig. SD); in bacteria labelled with anti-OL, most of the gold particles were located in the cell envelope beyond ETL and only rarely inside the cells (Fig. SE); and in cells labelled with anti-SDS, all labelling was found in the bacterial "capsule" (Fig. SF). The examination of more than 1000 sectioned bacteria during this study showed that gold particles were virtually nonexistent in the bacterial ETL, and were only seldom observed as a single gold particle per section.

Although the structure of the bacteria using the gelatin-Lowicryl procedure was not as well preserved as with the classical epon embedding, the distinction of the immunolabelling corresponded well to the anatomy of the mycobacterial cell as described in the literature. However, there was no regular and uniform labelling of any of the subcellular structures in all of the sectioned bacteria. For instance, the CWS was distinctly labelled as patches instead of a continuous label around the cell as it was described to be present for example in Bacillus subtilis immunolabelled using a teichoic acid antiserum (4).

Discussion

The combined use of chemical analysis and various cytochemical methods of electron microscopy led to the view that the mycobacterial cell wall was a tripartite structure (10, 22, 23) composed of an inner layer made of the arabinogalactan -peptidoglycan (4 + 9 nm), and a 12 nm thick outermost layer (OL) containing the more polar regions of the surface amphiphils, the two being separated by an intermediate, 8 nm thick, electron transparent layer (ETL). The ETL was proposed to be composed of the fatty acids protruding from the arabinogalactan (mycolic acids) which formed by lipid-lipid interactions with the surface amphiphils (10).

In the present study as well as an earlier reports (11,20, 30), when intact cells were treated with specific antisera, the antigen-antibody complexes were revealed using antirabbit IgG conjugates and the bacterial sections were observed after embedding in epon, the OL, ETL and PG layers were clearly seen but the labelling remained limited to the OL. A subsequently developed mild fixation technique followed by a gelatinLowicryl embedding procedure permitted to observe a "capsule-like structure" in addi-


tion to the above tripartite structure of the mycobacterial cell wall already described (4, 12,26). The latter method, however, did not permit a clear-cut localisation of the OL within the mycobacterial "capsule".

Models which schematically describe the mycobacterial cell envelopes, so far have been inferred either from electron microscopic studies (3), chemical data (13, 17), or the combined use of chemical extractions and electron microscopic observations (10, 16). These models often have conflicting propositions in relation to the location and attachment of some of the antigenic determinants, e.g. the attachment of the lipoarabinomannan to the cytoplasmic membrane in M. leprae (13) and M. tuberculosis (14) and attachment of the trehalose moiety of cord factor to the arabinogalactan-peptidoglycan (17) as opposed to David et al. (10) who proposed that all these amphipathic surface antigens were rather attached to the mycobacterial cells at the level of ETL through lipid-lipid interactions.

In the present study, the immunolabelling of ultrathin sections using the novel gelatin-uranyl acetate-low temperature Lowicryl HM20 embedding method with antisera raised in rabbits against distinct subcellular fractions of M. avium has resolved some of the above discrepancies and permitted observations which reconciled the chemical, ultrastructural and immunolabelling data. A model based on our recent findings is represented in Fig. 6 and its validity is discussed below.

Two of most striking findings in our study (Figs. Sand 6) relate to the immunolabelling of the ETL and the "capsule". In the case of ETL, gold particles were seldom observed indicating that it did not contain structural antigens. This was in agreement with earlier propositions which described this region essentially as a region of lipidlipid interactions (10, 17), but in disagreement with proposals that some of the lipids (lipoarabinomannan, LAM and lipooligosaccharide, LOS) were anchored by their fatty acid ends to the CM (13, 14). Furthermore, in our CM preparations which were characterized both chemically and ultrastructurally, we were unable to detect TS (8; and Table 1) which was the predominant fatty acid in LAM (14). We consequently placed these molecules at the level of the lipid-lipid interaction region in the bacterial ETL with their antigenic ends protruding outwards (Fig. 6). Finally, the mycobacterial "capsule" was found to be clearly immunolabelled with the anti-SDS serum confirming our previous report about the occurence of a "capsule-like" structure around these bacteria (10, 23).

A comparison of the bacteria labelled with anti-OL and anti-SDS serum showed that both the capsule and the OL were labelled by the immunosera, but with a distinct profile (Figs. SE and SF). The anti-OL labelling remained closer to the cell surface as opposed to the diffused capsular labelling with anti-SDS serum, indicating that although the bulk of the structural antigens of the outer layer remained close to the bacterial cell, some of them protruded outwards (probably because of their size) forming a glycocalyx-like arrangement of the epitopes inside this "capsule".

There are, however, other findings which indicate that the mycobacterial capsule may not be a capsule sensu stricto as, contrary to encapsulated pathogens which need to be opsonized using specific anticapsular antibodies prior to phagocytosis (1), mycobacteria are easily phagocytized without opsonization. This indeed depends on the surface forces as it has been shown by contact angle (8) determinations using saline water by means of a goniometer (the higher the contact angle, the higher the hydrophobicity) that phagocytic cells which have a 8 around 18°, are able to phagocytize only particles with a higher 8 value (31). Mycobacteria (e.g. M. bovis) have a 8 value as high as 70° as compared to encapsulated Escherichia coli and Streptococcus


pneumoniae (a e value of less than 18°; 31; C.}. Van Oss, personal communication), and consequently are not antibody-coated prior to engulfment by host macrophages. These observations strongly suggest that the mycobacterial capsule is not a true hydrophilic capsule. It may instead contain a matrix formed of polymethylated saccharide moieties of the lipopolysaccharides anchored in the bacterial ETL.

Fig. 6. Generalized view of the mycobacterial cell envelope architecture. In the cytoplasmic membrane (CM = 7 nm) the mannose-containing phospholipids (PIM) are placed in its outer layer (OL = 3 nm), which is thicker than the inner layer (IL = 2 nm) and reacts positively by the cytochemical method for saccharides according to Thiery (21, 27). A transparent layer (TL) clearly separates the two bilayers in the CM, and represents the plane of fracture (F). In the cell wall skeleton (CWS = 13 nm) made up of arabinogalactan (AG = 4 nm)-peptidoglycan (PG = 9 nm), the PG is shown by layers but their number has not yet been established and the AG drawn according to a recent report by Daffe et al. (5). The total thickness of the CWS was established by the periodic acid-phosphotungstic acid cytochemical staining according to Rambourg (10). The mycolic acids in the electron transparent layer (ETL = 8 nm) are shown esterifying the AG, or upside down in the lipid-lipid interaction region (= plane of fracture F; 2) of the ETL. In the wall outer layer (OL = 12 nm), a matrix of phospholipids (PL and PIM) and some complex amphiphils (sulpholipid, SL; phenoglycolipid, PGL; trehalose dimycolate, TDM, etc.) are represented with their fatty acid moieties facing the plane of fracture (8, 10, 17). The wall OL reacts characteristically by the ruthenium red method (21, 22, 23), but contrary to the OL in the CM, it does not stain when using the Thiery method except if the bacteria are autolyzing (21, 27). Although not represented in the drawing, the PL in the wall OL are esterified by tuberculostearic acid (TS), which is not the case with PL in the CM (8). The "capsule-like structure" (26; present study; CAP) is represented as protrusions of the sugar moieties of the long-sized amphiphils (e.g. lipoarabinomannan, LAM and mannose-derived lipooligosaccharides, LOS; 14) anchored by their fatty acid ends to the ETL. Since little information is available about molecular size and structure of wall proteins (P; 8, 10, 15), these were represented arbitrarily as being embedded in the wall OL matrix.


It is for the first time that a method has become available to investigate various subcellular antigens on ultrathin sections of mycobacteria. However, the partial immunolabelling of subcellular structures in this study may still represent a need to further improve this methodology. This may be decisive for a better understanding of mycobacteria-macrophage interactions at the molecular level which are supposed to be of prime importance in cell-mediated immunity. How various mycobacterial antigens are processed and transported to the surface of infected macrophages can now be satisfactorily investigated.

Acknowledgments. We thank S. Cadou for helping with Lowicryl embedding of the bacteria, to M.-C. Prevost for shadow replica preparations and M. Lesourd for scanning electron microscopy preparations. We are grateful to M.-F. Thorel for kindly providing antisera used in this investigation.

References

1. Absolom, D. R., C. J. van Oss, W. Zingg, and A. W. Neumann: Phagocytosis as a surface phenomenon: opsonization by aspecific adsorption of IgG as a function of bacterial hydrophobicity. J. Reticuloendothelial Soc. 31 (1982) 59-70

2. Benedetti, E. L., I. Dunia, M. A. Ludosky, N. Van Man, D. D. Trach, N. Rastogi, and H. L. David: Freeze etching and freeze fracture structural features of cell envelopes in mycobacteria and leprosy derived corynebacteria. Acta leprol. (Geneva). 2 (1984) 237-248

3. Barksdale, L. and K. S. Kim: Mycobacterium. Bact. Rev. 41 (1977) 217-372 4. Benichou, f.-C., C. Frehel, and A. Ryter: Improved sectioning and ultrastructure of

bacteria and animal cells embedded in Lowicryl. J. Elec. Microsc. Tech. 14 (1990) 289-297

5. Daffe, M., P. J. Brennan, and M. McNeil: Predominant structural features of the cell wall arabinogalactan of Mycobacterium tuberculosis as revealed through characterization of oligoglycosyl alditol fragments by gas chromatography/ma~s spectrometry and by IH and 13C NMR analysis. J. BioI. Chern. 265 (1990) 6734-6743

6. David, H. L.: Basis for the lack of drug-susceptibility of atypical mycobacteria. Rev. Infect. Dis. 3 (1981) 878-884

7. David, H. L., S. Clavel-Seres, F. Clement, and K. S. Goh: Uptake of selected antibacterial drugs in Mycobacterium avium. Zbl. Bakt. Hyg. A 265 (1987) 385-392

8. David, H. L., V. Levy-Frebault, and M. F. Thorel: Characterization of distinct layers of the Mycobacterium avium envelope in respect of their composition by fatty acids, proteins, oligosaccharides and antigens. Zbl. Bakt. Hyg. A 268 (1988) 193-208

9. David, H. L. and N. Rastogi: Antibacterial action of colistin (polymyxin E) against Mycobacterium aurum. Antimicrob. Agents Chemother. 27 (1985) 701-707

10. David, H. L., N. Rastogi, S. Clavel-Seres, F. Clement, and M. F. Thorel: Structure of the cell envelope of Mycobacterium avium. Zbl. Bakt. Hyg. A 264 (1987) 49-66

11. David, H. L., M. F. Thorel, C. Frehel, and N. Rastogi: Serologic and immunocytochemical analysis of the Mycobacterium avium cell envelope. Acta leprol. (Geneva). 7, Suppl. 1 (1989) 55-58

12. Frehel, c., N. Rastogi, f.-C. Benichou, and A. Ryter: Do test tube-grown pathogenic mycobacteria possess a protective capsule? FEMS Microbiol. Lett. 56 (1988) 225-230

13. Gaylord, H. and P. J. Brennan: Leprosy and the leprosy bacillus: recent developments in characterization of antigens and immunology of the disease. Ann. Rev. Microbiol. 41 (1987) 645-675

14. Hunter, S. W. and P. f. Brennan: Evidence for the presence of a phosphatidylinositol anchor on the lipoarabinomannan and lipomannan of Mycobacterium tuberculosis. J. BioI. Chern. 265 (1990) 9272-9279


15. farlier, V. and H. Nikaido: Permeability barrier to hydrophilic solutes in Mycobacterium chelonei. J. Bact. 172 (1990) 1418-1423

16. Imaeda, T., F. Kanetsuna, and B. Galindo: Ultrastructure of cell walls of genus Mycobacterium. J. Ultrastruct. Res. 25 (1968) 46-63

17. Minnikin, D. E.: Lipids: complex lipids, their chemistry, biosynthesis and roles. In: C. Ratledge and f. Stanford (eds.), The Biology of Mycobacteria, Vol. 1, pp. 95-184. Academic Press, London (1982)

18. Rastogi, N.: Killing intracellular mycobacteria in in vitro macrophage systems: what may be the role of known host microbicidal mechanisms? In: N. Rastogi (ed.), 5th Forum in Microbiology - "Killing Intracellular Mycobacteria: Dogmas and Realities". Res. Microbiol. 141 (1990) 217-230

19. Rastogi, N. and H. L. David: Mechanisms of pathogenicity in mycobacteria. Biochimie 70 (1988) 1101-1120

20. Rastogi, N. and C. Frehel: Evidence that coating of Mycobacterium leprae surface antigens reduces its ability to hinder host microbicidal functions. Zbl. Bakt. 272 (1990) 337-346

21. Rastogi, N., C. Frehel, and H. L. David: Evidence for taxonomic utility of periodic acidthiocarbohydrazide-silver proteinate cytochemical staining for electron microscopy. Int. J. System. Bact. 34 (1984) 293-299

22. Rastogi, N., C. Frehel, and H. L. David: Triple-layered structure of the mycobacterial cell wall: evidence for the existence of a polysaccharide-rich outer layer in 18 mycobacterial species. Curr. Microbiol. 13 (1986) 237-242

23. Rastogi, N., C. Frehel, A. Ryter, H. Ohayon, M. Lesourd, and H. L. David: Multiple drug resistance in Mycobacterium avium: is the wall architecture responsible for the exclusion of antimicrobial agents? Antimicrob. Agents Chemother. 20 (1981) 666-677

24. Rastogi, N. and K. S. Goh: Antibacterial action of l-isonicotinyl-2-palmitoyl hydrazine against the Mycobacterium avium complex and the enhancement of its activity by mfluoro-phenylalanine. Antimicrob. Agents Chemother. 34 (1990) 2061-2064

25. Rastogi, N., K. S. Goh, and H. L. David: Enhancement of drug susceptibility of Mycobacterium avium by inhibitors of cell envelope synthesis. Antimicrob. Agents Chemother. 34 (1990) 759-764

26. Rastogi, N. and R. Hellio: Evidence that the capsule around mycobacteria grown in axenic media contains mycobacterial antigens: implications at the level of cell envelope architecture. FEMS Microbiol. Lett. 70 (1990) 161-166

27. Rastogi, N., f. Moniz-Pereira, C. Frehel, and H. L. David: Ultrastructural evidence for the accumulation of a polysaccharide-like substance during mycobacteriophage D29 replication in Mycobacterium smegmatis Ann. Virol (Inst. Pasteur) 134E (1983) 251-266

28. Rastogi, N., B. Moreau, M.-L. Capmau, K. S. Goh, and H. L. David: Antibacterial action of amphipathic derivatives of isoniazid against the Mycobacterium avium complex. Zbl. Bakt. Hyg. A 268 (1988) 456-462

29. Rook, G. A. W.: The role of activated macrophages in protection and immunopathology in tuberculosis. In: N. Rastogi (ed.) 5th Forum in Microbiology - "Killing Intracellular Mycobacteria: Dogmas and Realities". Res. Microbiol. 141 (1990) 253-256

30. Tereletsky, M. f. and W. W. Barrow: Postphagocytic detection of glycolipids associated with the superficial L1layer of Mycobacterium intracellulare. Infect. Immun. 41 (1983) 1312-1321

31. van ass, c. f., D. R. Absolom, and W. Neumann: Surface forces in phagocytosis. In: S. M. Reichard and P. Filkins (eds.), The Reticuloendothelial System, Vol. 7A, pp. 3-35. Plenum Publishing Corporation, New YorklUSA (1984)

32. Whitehouse, R. L., f.-c. Benichou, E. Couture-Tosi, S. Schenckman, and A. Ryter: Immunolabelling of bacteriophage A receptor protein (lamB) on thin sections of Escherichia coli embedded in Lowicryl. BioI. Cell. 51 (1984) 389-394

Dr. Nalin Rastogi, Unite de la Tuberculose et des Mycobacteries, Institut Pasteur, 25 Rue du Dr. Roux, F-75724 Paris Cedex 15, France

Documents

A New Insight into the Mycobacterial Cell Envelope Architecture by the Localization of Antigens in Ultrathin Sections