MPB70 and MPB83 as Indicators of Protein Localization in Mycobacterial Cells

Abstract

Culture fluids after growth of Mycobacterium bovis BCG on Sauton medium contain actively secreted proteins and proteins released by bacterial lysis. BCG culture fluids and sonicates of Mycobacterium tuberculosis and Mycobacterium paratuberculosis were tested after separation by gel filtration and sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). The localization of marker proteins was determined by enzyme-linked immunosorbent assay and Western blotting with selected monoclonal antibodies of known specificities. Soluble secreted proteins (MPB64 and proteins of the antigen 85 complex) and three heat shock proteins (DnaK, GroEL, and GroES) were recovered in a single peak after gel filtration, indicating their occurrence as a free monomer in the culture fluid and cytosol, respectively. Other constituents eluted in two distinct peaks during gel filtration. The first peak corresponded to the void volume, indicating complex formation between several proteins or attachment to lipids in the surface layer or the cytoplasmic membrane; the second peak corresponded to the expected monomer size indicated by SDS-PAGE under conditions that separate proteins from each other during sample preparation. The two-peak group contained constituents with known lipid contents, the 19- and 38-kDa lipoproteins and lipoarabinomannan. The 26-kDa form of MPB83 behaved similarly. After extraction with Triton X-114, these constituents entered into the detergent phase, confirming the lipoprotein nature of 26-kDa MPB83. The MPB83 molecule was shown to be available on the surface of BCG Tokyo bacilli for reaction with monoclonal antibody MBS43 by flow cytometry.

A multitude of different proteins are synthesized by the mycobacterial cell. It is often valuable to consider these proteins in different broad categories based on common characteristics, such as physicochemical properties, localization in the mycobacterial cell, and active secretion during culture. In turn, these distinct properties are closely related to their functional properties and the tendency to interact with the immune system of the host after infection.

MPB70 was initially isolated by Nagai et al. (37). This secreted protein is of particular interest since it is highly species specific (13). It is consistently present in virulent Mycobacterium bovis, whereas it is synthesized and secreted in markedly different amounts by various substrains of avirulent M. bovis BCG (13, 32, 35).

We recently studied the closely related MPB83 protein and isolated three peptides derived from it after CNBr cleavage, showing that MPB70 and MPB83 are homologous but clearly distinct proteins and are therefore encoded by different genes (15). The heterogeneity between different substrains of BCG in regard to the synthesis of MPB70 and MPB83 proteins is clearly greater than previously realized (53). Another type of heterogeneity has also previously been identified; 26- and 23-kDa fractions of a protein that was presumed to be MPB70 differed markedly in carbohydrate content (9). The available data indicate that MPB83 occurs in 26- and 23-kDa forms, both glycosylated, whereas MPB70 (at 22 kDa) is nonglycosylated.

The term MPB was introduced (37) for the designation of a protein purified from M. bovis BCG, with the number after MPB denoting the relative mobility by electrophoresis on a 7.7% polyacrylamide gel run at pH 9.5. MPT is used for similar designations of proteins purified from M. tuberculosis (38). The designations mpb70 and mpt70 denote the corresponding genes.

Cloning of mpb70 (43) revealed the sequence of a polypeptide chain preceded by a signal peptide that is typical of secreted proteins. In contrast, both mpb83 from M. bovis BCG (33, 45) and mpt83 from M. tuberculosis (20) revealed a typical lipoprotein consensus element in the signal sequence. The relative concentrations of the 26- and 23-kDa components of these proteins vary markedly between sonicates and culture fluids of BCG bacilli. In sonicates, the 26-kDa component, consisting of MPB83, dominates. In culture fluids, the reverse is observed, with a markedly higher concentration of the 22- to 23-kDa components in BCG Sweden and BCG Russia (18).

The purpose of the present work was to investigate these properties at the level of native proteins in relation to the localization of these and other marker proteins in the mycobacterial cell. Using monoclonal antibody (MAb) MBS43, which reacts with MPB83 but not with MPB70 (53), permitted more precise distinctions between these proteins than previously obtained.

MATERIALS AND METHODS

Bacterial strains and culture.

M. bovis BCG Tokyo substrain 172, BCG Russia, and BCG Sweden were obtained from the National Institute of Infectious Disease, Tokyo, Japan. BCG Copenhagen Danish substrain 1331 and M. tuberculosis H37Rv (ATCC 27294) were obtained from Statens Seruminstitut, Copenhagen, Denmark. Bacilli were grown as a surface pellicle on liquid, wholly synthetic Sauton medium. Bacilli were removed by sequential paper filtration and sterile filtration through a 0.2-μm filter. The resulting culture fluids were concentrated by ammonium sulfate precipitation at 80% saturation. Bacilli were washed three times in phosphate-buffered saline (PBS) and then sonicated five times for 3 min in a rosette cooling cell with a 100-W sonifier (MSE, London, United Kingdom) as described previously (7).

M. paratuberculosis, strain 316F, 2E cells were kindly provided by F. Saxegaard, National Veterinary Institute, Oslo, Norway.

Proteins.

MPB70 was purified from BCG Tokyo culture fluid as originally described (37). MPB83 was purified from BCG Tokyo culture fluid as recently reported (15). Purification to homogeneity was ascertained by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and staining with Coomassie brilliant blue, with both proteins providing single bands at 22 to 23 kDa. The separation of MPB70 from MPB83 was ascertained by isoelectric focusing; the proteins gave single bands with distinctly different isoelectric points as described previously (15).

Antibodies.

Polyvalent rabbit anti-M. tuberculosis culture fluid (49) was obtained by immunizing rabbits repeatedly with 1 mg of protein from an M. tuberculosis culture fluid with a low content of cytoplasmic proteins by our standard technique (12). Anti-MPB83 or MPB83 activity in this antiserum was demonstrated in experiments with purified MPB83 by enzyme-linked immunosorbent assay (ELISA) and Western blotting. Polyclonal rabbit anti-MPB83 (K483) was obtained by immunization with isolated MPB83 as previously reported (15). Polyvalent anti-BCG immunoglobulin (code B124; lot 063B) was kindly provided by Dako Immunoglobulins, Copenhagen, Denmark.

Table 1 provides a list of the mouse monoclonal antibodies (MAbs) used for the localization of proteins after gel filtration and in Western blotting.

TABLE 1.

MAbs used for the localization of proteins after gel filtration and in Western blotting

MAb	Reactive with (reference[s]):	Reactivity in void volumea	Source
L24b4	MPT/MPB64 (2, 24)	−	SSIb
L24b5	MPT/MPB64 (2, 24)	−	SSI
HYT27	Antigen 85 complex (41, 52)	−	SSI
L7	DnaK (5)	−	W. J. Britton
HAT3	DnaK (3, 24)	−	SSI
CBA1	GroEL (24, 29)	−	SSI
HYB76-6	GroEL (25)	−	SSI
HYB76-1	GroES (25)	−	SSI
TB71	38-kDa lipoprotein (24)	+	J. Ivanyi
HYT4	19-kDa lipoprotein (41)	+	SSI
HYT6	19-kDa lipoprotein (24)	+	SSI
ML34	LAM (22)	+	J. Ivanyi
HYB76-8	ESAT-6 (16, 42)	−	SSI
CBA2	56-kDa protein (29)	−	SSI
HYB76-3	47-kDa protein and EFTu (25)	−	SSI
HYB76-4	14-kDa protein (25)	−	SSI
HAT1	DnaK and 19-kDa lipoprotein (3, 24)	+	SSI
1-5C	MPB70 (51)	−	K. Lyashchenko
MBS43	MPB83 (10, 53)	+	R. G. Hewinson
12/6/1	MPB70 and MPB83 (21, 53)	+	R. G. Hewinson

^a+, reactivity; −, no reactivity. 

^bSSI, Statens Seruminstitut. 

Gel filtration.

Gel filtration was performed on a Superdex 200 HR 10/30 column (Pharmacia Biotechnology AB, Uppsala, Sweden) by employing an LKB high-pressure liquid chromatography unit. The application volume was 250 μl of BCG Tokyo culture fluid containing 1.2 mg of protein in 30 mM Tris-HCl (pH 8.7)–0.1 M NaCl after equilibration of the column with the same buffer. The flow rate was 0.5 ml/min, with fractions of 500 μl collected. Serial fractions were diluted 1:10 prior to the coating of microtiter plates for determinations of protein content by ELISA with MAbs as indicated. M. tuberculosis and M. paratuberculosis sonicates were fractionated on a Superose 12 HR 10/30 column (Pharmacia) under similar conditions.

ELISA.

The ELISA was set up as described in detail previously (50). Briefly, Immunoplate MaxiSorp (catalog no. 442404; Nunc, Copenhagen, Denmark) 96-well plates were coated with 100 μl of culture fluids (providing 1 μg of total protein per well), 0.5 μg of purified native protein per well, or fractions collected after gel filtration (diluted 1:10). Blocking was carried out with PBS (pH 7.4) containing 5 mg of bovine serum albumin (BSA) per ml. The second layer consisted of mouse MAbs as indicated in the text and figure legends. The indicator system was horseradish peroxidase (HRP)-conjugated sheep anti-mouse immunoglobulin (Amersham International plc, Amersham, United Kingdom). The substrate was 2,2′-azino-di(ethylbenzthiazoline sulfate) (ABTS). Washing was carried out four times between each step with PBS containing 0.1% Tween 20. All reaction mixtures were set up in triplicate, with the median used for recording and calculations. All results were read on an MR 7000 ELISA reader (Dynatech Laboratories Inc., Chantilly, Va.).

SDS-PAGE with immunoblotting.

SDS-PAGE was performed with the Pharmacia system for horizontal electrophoresis in a Multifor II electrophoresis unit (model 2117; LKB, Bromma, Sweden) with precast polyacrylamide gels (ExcelGel [SDS gradient, 8 to 18%]; Pharmacia). One microgram of purified protein or 10 μg of protein of various culture fluids and sonicates was used in each lane. Rainbow molecular weight markers (Amersham) were used as standards. Semidry Western blotting was performed with a Novablot electrophoretic transfer kit (catalog no. 2117-250; LKB) and 0.2-μm-pore-size nitrocellulose membranes (Schleicher and Schüll GmbH, Dassel, Germany). After electrophoretic transfer, membranes were blocked in PBS with 2% BSA and 1% gelatin for 1 h. Between each step, membranes were washed four times for 10 min in PBS with 0.1% Tween 20 on a rotary shaker. All antibodies were diluted in PBS with 0.2% Tween 20 and 0.2% BSA. Primary antibodies were incubated overnight. HRP-labelled antibodies were diluted 1/1,000 and incubated for 1.5 h. Diaminobenzidine in 0.1 M sodium acetate buffer (pH 4.0) was used as the substrate.

Detergent extraction.

BCG Tokyo culture fluid (0.5 ml [containing 1 mg of total protein]) was mixed with 0.5 ml of 4% Triton X-114 and incubated for 30 min with gentle mixing at 4°C. Phase separation was obtained by incubation for 10 min in a 37°C water bath, followed by centrifugation for 20 min at 13,000 rpm in an Eppendorf centrifuge (Netheler + Hinz GmbH, Hamburg, Germany). The water phase was carefully pipetted off with a Pasteur pipette, and the detergent phase was washed twice with prewarmed PBS at 37°C prior to SDS-PAGE and Western blotting.

Flow cytometry.

Flow cytometry analysis of BCG bacilli was performed with a FACScan (model 440; Becton Dickinson Immunocytometry Systems, Mountain View, Calif.). BCG Tokyo and BCG Copenhagen cultured on Sauton medium were suspended in PBS to 1 mg (wet weight)/ml, and 75 μl of each suspension was mixed with 225 μl of MAb diluted in PBS and incubated for 1 h at 37°C under head-on-end rotation. Bacilli were pelleted by centrifugation, washed twice in PBS to remove unbound material, incubated with fluorescein isothiocyanate (FITC)-labelled MAb to mouse immunoglobulin G (IgG) (clone LO-MG7) (Monozan, Uden, The Netherlands) diluted 1:400 for 30 min at 4°C, and run directly in a flow cytometer. In experiments with primary rabbit anti-MPB83 antibody K483 diluted 1:500, bacilli were stained with FITC-conjugated goat anti-rabbit IgG (Southern Biotechnology Associates Inc., Birmingham, Ala.) diluted 1:100. All sera, buffers, and antibody solutions were sterile filtered before use to avoid any particulate matter other than bacilli.

RESULTS

Gel filtration.

Mycobacterial proteins which are present in complex form and attached to each other, retained in the lipid surface layer, or incorporated into the cytoplasmic membrane are usually separated from each other during the initial preparation for SDS-PAGE when the sample is boiled in application buffer containing SDS. As a result, they appear as distinct bands after electrophoresis at positions that correspond to their individual molecular masses.

In contrast, direct gel filtration of BCG culture fluids is a more gentle procedure. This implies less of or no tendency for the dissociation of complexes formed by noncovalent interactions between proteins or release of proteins incorporated into the cytoplasmic membrane. In our experience, BCG culture fluids consistently contain actively secreted proteins and proteins released as a result of bacterial lysis, whereas culture fluids with negligible amounts of cytoplasmic constituents can be obtained after the growth of M. tuberculosis as a surface pellicle on Sauton medium (49). Figure 1 illustrates the gel filtration results for BCG Tokyo and M. tuberculosis culture fluids on a Superdex 200 column. Figure 2 demonstrates similar gel filtration results for M. tuberculosis and M. paratuberculosis sonicates on a Superose 12 column.

FIG. 1.

Gel filtration of culture fluids of BCG Tokyo (A, B, and D) and M. tuberculosis (C) on a Superdex 200 column. (A) Optical density (OD) at 280 nm, indicating the total protein content in eluate fractions. (B, C, and D) ODs by ELISA of eluate fractions coated on the solid phase with MAbs L24b4, HYT6, and MBS43, respectively, in the second layer and HRP-labelled anti-mouse immunoglobulin in the third layer.

FIG. 2.

Gel filtration of M. tuberculosis (A and B) and M. paratuberculosis (C) sonicates on a Superose 12 column. (A) Optical density (OD) at 280 nm. (B) OD by ELISA of eluate fractions of M. tuberculosis sonicate tested for GroEL (65-kDa heat shock protein [hsp65]) content with MAb CBA1. (C) Eluate fractions of M. paratuberculosis sonicate tested for LAM content with MAb ML34.

Proteins eluted in a single peak after gel filtration.

Figure 1A shows the total protein contents in eluate fractions of BCG Tokyo culture fluid, as determined by the optical density at 280 nm. Figure 1B shows that the actively secreted protein MPB64 (14), which reacts with MAb L24b4, eluted in a single peak at a position expected for a monomeric protein according to its known molecular mass. MAb L24b4 did not react significantly with protein in the void volume. Proteins of the antigen 85 complex (46, 48), which react with MAb HYT27, also appeared in a single peak as described previously (50) and eluted slightly ahead of MPB64, as expected from their molecular masses. MAb 1-5C, which reacts with only MPB70 (51), also gave a single, sharp peak at the position of its free monomer with no reaction in the void volume. Dimers of MPB70 are frequently observed in preparations of purified protein, but not in culture fluids.

Figure 2A shows the optical densities (at 280 nm) of eluate fractions of M. tuberculosis sonicate. Figure 2B shows that GroEL, the 65-kDa heat shock protein, also appeared in a single peak after gel filtration of M. tuberculosis sonicate. The same patterns were obtained with MAb L7, which reacts with DnaK, and with MAb HYB76-1, which reacts with GroES. None of the five MAbs that react with only these three heat shock proteins gave a significant absorption value in void volume fractions by ELISA.

Proteins eluted in two individual peaks after gel filtration.

Figure 1C shows a distinctly different pattern of two separate peaks obtained with MAb (HYT6) to the 19-kDa lipoprotein of M. tuberculosis. A similar pattern of a distinct peak at the void volume position and a second peak at the expected position of monomeric, free protein was obtained with MAb (TB71) to the 38-kDa lipoprotein. Figure 2C shows that the MAb (ML34) to lipoarabinomannan (LAM) also gave two distinct peaks in reactions with eluate fractions.

Of the 16 initial MAbs listed in Table 1, TB71, HYT4, HYT6, and ML34 showed significant reactivities in the void volume area after gel filtration. The only known common feature of these MAbs is their reactivities with lipid-containing materials, lipoproteins, and LAM.

A particular feature of MAbs to the 70-kDa heat shock protein DnaK was noted. MAb L7 gave a single peak corresponding to a free, monomeric protein. In contrast, MAb HAT1, previously reported to react with only DnaK (24), reacted at three positions during gel filtration with various sonicates, showing strong reactivity with void volume fractions. Pursuing this observation, we noted that the additional reactivity of this MAb at 19 kDa could be explained by its independently demonstrated cross-reactivity with the 19-kDa lipoprotein of M. tuberculosis (4).

Of MAbs reacting with MPB70 and/or MPB83, 1-5C, which reacts with only MPB70 (51), gave a single, sharp peak with a maximum at fraction 36, whereas MAb MBS43, which reacts with only MPB83 (53), gave two separate peaks corresponding to the void volume and a free, monomeric protein with a maximum at fraction 36 (Fig. 1D). A similar pattern was observed with MAb 12/6/1, which reacts with both MPB70 and MPB83, in ELISA, with protein isolated in the first layer on the solid phase.

Protein content in the void volume after gel filtration of BCG Tokyo culture fluid on a Superdex 200 column.

Figure 3 shows the SDS-PAGE and Western blotting results for the void volume fraction compared with those for a later fraction containing free protein after gel filtration. The two initial membranes show that MAbs to secreted proteins (MPB64 and the antigen 85 complex) did not react with proteins in the void volume but revealed distinct bands at different, later positions corresponding to the molecular masses of the constituents. The third membrane shows a similar pattern with a MAb (L7) to DnaK. These observations from a different technique thus confirmed the gel filtration and ELISA findings (Fig. 1 and 2; Table 1).

FIG. 3.

SDS-PAGE and Western blotting results for the void volume fraction after gel filtration of BCG Tokyo culture fluid on a Superdex 200 column compared with those for a later fraction containing free protein. Lanes V, void volume fraction; lanes F, fraction corresponding to the position of a monomeric, free protein. The MAbs used are indicated above membranes. In the membrane with anti-M. tuberculosis (a–M.tb) culture fluid, lane 1 contained BCG Tokyo culture fluid, lane 2 contained the void volume fraction, and lane 3 contained pooled fractions with 20- to 40-kDa free proteins. Horizontal lines on the left indicate molecular mass markers (97.4, 66, 46, 30, 21.5, and 14.3 kDa).

The fourth membrane shows the reaction pattern of rabbit anti-M. tuberculosis culture fluid. This polyvalent antibody reacts with a series of different proteins in BCG Tokyo culture fluid (49). In the void volume, three strong bands, corresponding to the 19-kDa lipoprotein, a 26-kDa band containing MPB83, and the 38-kDa lipoprotein, as further defined in additional experiments with MAbs to these proteins, were obtained. Fractions corresponding to the 20- to 40-kDa molecular mass area were pooled, concentrated, and tested in the third lane of this membrane. A series of distinct bands were obtained. Note in particular the strong reactivity with proteins of the antigen 85 complex at about 30 kDa. MAbs 12/6/1 and MBS43 gave similar patterns. The strongest reactivities were at 26 kDa in the void volume fraction and at a lower position with pooled fractions containing 20- to 40-kDa free proteins.

Detergent extraction with Triton X-114.

Triton X-114 was used for detergent extraction because of its low cloud point (20°C). Extraction was performed below this point (4°C) to provide a water-clear solution, with subsequent centrifugation at 37°C to provide a distinct separation of the detergent and water phases (34, 40, 56).

Figure 4 shows Western blot results after the initial separation of proteins in water and detergent phases by SDS-PAGE. The left membrane shows the reactivity of polyvalent rabbit anti-M. tuberculosis culture fluid, revealing a multitude of bands in BCG Tokyo culture fluid and its water phase after extraction with Triton X-114. In contrast, the detergent phase demonstrated a simpler pattern, with three distinct bands, corresponding to the known positions of the 19-kDa lipoprotein, a 26-kDa band containing MPB83, and a distinct but weaker band at 38 kDa. The last is the position of the 38-kDa lipoprotein (1), which is known to be present in BCG, albeit at a far lower concentration than that in M. tuberculosis (17, 47). The diffuse smear further above (Fig. 4) shows the typical reaction pattern with LAM in the blot.

FIG. 4.

SDS-PAGE and Western blotting to compare the water and detergent phases of BCG Tokyo culture fluid after extraction with Triton X-114. The antibodies used are indicated above membranes. Lane cf, culture fluid; lanes w, water phase after extraction; lanes d, detergent phase after extraction. The molecular mass markers are the same as those described in the legend to Fig. 3. a–M.tb, anti-M. tuberculosis culture fluid.

The other membrane shows the reactivity of the BCG Tokyo culture fluid water phase compared with that of the detergent phase in the neighboring lane developed with MAb 12/6/1, which reacts with MPB70 and MPB83 in Western blotting (53). The water phase gave the characteristic pattern of whole culture fluid, with a major doublet at 22 to 23 kDa and a weaker band at 26 kDa. In the detergent phase, a major band appeared at 26 kDa, with no reactivity in the 22- to 23-kDa region. Additional ELISA experiments for quantification after Triton X-114 extraction experiments with purified MPB70 and 23-kDa MPB83 showed that both were recovered in the water phase (data not shown).

Flow cytometry.

Figure 5A shows that MAb MBS43, which reacts with only MPB83, gave a distinctly higher fluorescence with BCG Tokyo than with BCG Copenhagen Danish strain 1331 (median channel fluorescence intensities [MCFI] of 180 and 16, respectively). These strains are the classical high- and low-producing strains of this protein, respectively (53). Polyclonal rabbit anti-MPB83 (K483) stained BCG Tokyo strongly, with an MCFI of 240. These findings indicate that MPB83 is available on the bacterial surface of BCG Tokyo for reaction with this MAb and polyclonal rabbit antibody to MPB83. Preimmune serum from the same rabbit run as a negative control in the same experiment had an MCFI of 27. MAbs of irrelevant specificities run as isotype controls were consistently negative.

FIG. 5.

(A) BCG Tokyo and BCG Copenhagen (BCG CPH) bacilli were incubated with MAb MBS43 (1:25) prior to incubation with FITC-labelled rat anti-mouse immunoglobulin (1:400) and flow cytometry. The number of events is recorded on the ordinate; the fluorescence intensity (FL1-Height) is recorded on the abscissa. BCG Tokyo bacilli were also incubated with rabbit anti-MPB83 antibody K483 (1:500) and with preimmune (preim.) serum as the corresponding negative control and stained with FITC-conjugated goat anti-rabbit IgG (1:100). (B) BCG Tokyo stained with MAb 1-5C (1:300), which reacts with MPB70 only, and the control substitution of PBS for MAb. (C) BCG Tokyo stained with MAbs MBS43, L24b4 (1:500), and 12/6/1 (1:500).

Figure 5B shows that MAb 1-5C, which reacts with MPB70 only, gave insignificant staining of BCG Tokyo (MCFI, 8.4). The lower pattern in Fig. 5B is that of the control, BCG Tokyo exposed to PBS instead of a MAb during the initial stage and then to FITC-labelled anti-mouse immunoglobulin (MCFI, 12).

Figure 5C shows the results of a third set of experiments. Again, the MAb to MPB83 (MBS43) gave strong staining of BCG Tokyo (MCFI, 390). MAb L24b4, which reacts with the soluble secreted protein MPB64 and was run as an expected negative control, showed significantly lower staining of BCG Tokyo (MCFI, 49).

MAb 12/6/1, which reacts with MPB70 and MPB83 by ELISA with the respective purified proteins on the solid phase and by Western blotting, gave an unexpected negative result (MCFI, 29), which we interpreted as indicating that the epitope on MPB83 which reacts with MAb 12/6/1 is not available on the bacterial surface.

DISCUSSION

Differences in the localization of individual proteins in relation to the mycobacterial cell are expected to be closely correlated to their functional properties and influence on the interaction between multiplying bacilli and the infected host. In recent years, considerable attention has been given to the group of secreted proteins which probably are available for presentation to the immune system prior to cytoplasmic constituents, which depend on lysis from the bacterial cell for presentation (23). Secreted proteins are therefore expected to be of particular importance for immune responses after infection, and the study of mechanisms of release and the development of techniques for similar releases from recombinant microorganisms are of great interest in the development of new vaccines against tuberculosis (11).

After growth of mycobacteria on synthetic media, culture fluids contain proteins actively secreted from mycobacterial cells and cytoplasmic constituents released through bacterial lysis, which occurs even during logarithmic growth. In our experience, short-term culture of M. tuberculosis is less important than is establishing conditions with minimal stress in stationary cultures to obtain culture fluids that contain actively secreted proteins with minimal contamination of cytoplasmic constituents. Further, the demonstration of actively secreted proteins is facilitated by a direct comparison of culture fluid and a corresponding sonicate of washed bacilli (49).

In a study of the closely related MPB70 and MPB83 proteins (15), the availability of MAbs reacting with both proteins (such as 12/6/1) and selectively with either of these proteins (such as 1-5C and MBS43, respectively) provided significant new information about the localization of these proteins in the mycobacterial cell. The reactivity of MAbs 12/6/1 and MBS43 with MPB83 at 26 kDa is a feature of the bacterial sonicate, whereas reactivity at 22 to 23 kDa is mainly a feature of culture fluid (18).

In flow cytometry, MAb MBS43, which reacts with only MPB83 (53), stained BCG Tokyo bacilli, whereas BCG Copenhagen, a low-producing strain, was negative (Fig. 5A). Polyclonal rabbit antibody to MPB83 (K483) also stained BCG Tokyo, whereas preimmune serum from this rabbit was negative with the same cells. These findings show that MPB83 is present on the bacillary surface for reaction with MAbs and polyclonal antibodies.

Because of this observation, the lack of significant staining with MAb 12/6/1 was surprising since this MAb reacts more strongly with MPB83 than with MPB70 by Western blotting (53). Both the localization of reactive epitopes on the MPB83 molecule and the configuration of the molecule on the bacterial surface may be of significance in this regard. By ELISAs with overlapping peptides of the entire MPB70 sequence, we localized the MAb 1-5C-reactive epitope to residues 109 to 118 (the numbering starts at Met 1 of the signal peptide) (51). MAb 12/6/1 did not react with any of the overlapping peptides, indicating reactivity with a conformational epitope whose localization remains to be determined. It may be localized in an area of the polypeptide chain that is inaccessible on the bacterial surface; alternatively, the protein configuration may be sufficiently different to account for the lack of interaction with 12/6/1 under these conditions. The reactivity of MAb 12/6/1 with MPB83 is strikingly dependent on the configuration of the protein, with a strong reaction to MPB83 by Western blotting and by ELISA with the isolated protein in the first layer coated onto the solid phase but with no reaction by capture ELISA with the MAb at the solid phase and native MPB83 in the second layer (51).

It is therefore appreciated that the lack of staining with MAb 1-5C in flow cytometry (Fig. 5B) may be a false-negative reaction and may in itself be no proof of the absence of MPB70 on the bacterial surface.

The paucity of publications with an analysis of mycobacteria by flow cytometry, apart from studies of recombinant organisms labelled with the green fluorescent protein of the jellyfish Aequorea victoria (8, 27), is striking. We recently employed flow cytometry in studies of surface antigens on mycobacteria, demonstrating surface deposition of complement activation products on M. tuberculosis (19).

By immunoelectron microscopy, Lounatmaa and Brander demonstrated an antigen, then presumed to be MPB70, in the crystalline cell surface layer of M. bovis BCG (30, 31). We have previously shown that their MAb 4C3/17 (SB10 [54]) reacts with MPB83 and MPB70 (53). Recently, Vosloo et al. (45) showed by immunoelectron microscopy that their MAb raised against sonicated BCG Tokyo, b3C4, bound to the surfaces of intact BCG bacilli. The extent of reactivity with MPB83 and MPB70 was not specified in their published data. In experiments of this kind, it is essential to employ MAbs that react selectively with one of the proteins, such as the specificity of MBS43 for MPB83 (53). Taken together with the flow cytometry data, we conclude that MPB83 is probably the constituent demonstrated by immunoelectron microscopy. The attachment of MPB83 at the bacillary surface through lipidation was further substantiated by the site-directed mutagenesis experiments of Vosloo et al. (45). After mutation of the cysteine residue of LAGC to serine, the modified recombinant MPB83 lost essential lipoprotein characteristics and appeared in the water phase after extraction with Triton X-114.

The localization index (49) for MPB70 is higher than that for any other mycobacterial antigen we have studied, indicating high efficiency in the secretion process. This finding previously appeared to be in conflict with the electron microscopy data (30, 31) but has been resolved through the use of selected MAbs of high specificity, with MPB83 having been demonstrated on the bacterial surface and with MPB70 as a soluble secreted protein not demonstrable in association with the bacterial cell.

After extraction with Triton X-114, Western blotting with polyvalent rabbit anti-M. tuberculosis antibody revealed a strong band at 26 kDa (Fig. 3 and 4), the position of MPB83. However, additional lipoproteins may be present in this region. In their detergent-phase separation and metabolic labelling experiments, Young and Garbe (56) demonstrated 27- and 26-kDa lipoproteins in M. tuberculosis that were clearly separable by two-dimensional electrophoresis. The 26-kDa protein may well represent MPB83. We recently identified an M. tuberculosis gene, MTCY24G1.04 in the Sanger Centre anonymous FTP site for M. tuberculosis, that codes for a polypeptide chain with a lipoprotein consensus element (LWLGC) and a deduced molecular mass of 24.1 kDa without lipidation. This lipoprotein contains a T-cell epitope that was originally demonstrated in M. leprae and is also present in M. tuberculosis and M. bovis (36, 39).

Previous fractionation experiments have provided conflicting information for the cellular localization of heat shock proteins in prokaryotes. In Escherichia coli, DnaK has previously been observed to cofractionate with the inner membrane fraction of minicells (26), but DnaK may be particularly prone to artifactual association with membranes during fractionation since as a chaperon it shows a marked tendency to hydrophobic protein-protein interactions (28). In immunogold electron microscopy assays, Bukau et al. (6) found gold particles predominantly in the cytoplasm, indicating that the majority of DnaK molecules are cytoplasmic. However, a fraction of gold particles was located close to membranes, raising the possibility that a subpopulation of DnaK proteins is membrane associated.

After gel filtration on Superdex 200 and Superose 12 columns with high exclusion volumes, we consistently observed no significant reaction of MAbs to the three heat shock proteins examined (DnaK, GroEL, and GroES) with proteins in the void volume, indicating that the bulk of these proteins are present as free proteins in the cytosol. The occurrence of these proteins in or associated with the cytoplasmic membrane would be expected to result in significant reactions in the void volume fraction after gel filtration; therefore, our findings are somewhat different than those reported by Mehrotra et al. (34), who in studies of M. fortuitum observed less reactivity for MAb TB78 with GroEL in the cytosol compared with that in the membrane fraction. The subcellular localization of heat shock proteins may also vary among mycobacterial species.

The demonstration of a lipoprotein consensus element (FLAGC) (44, 55, 56) in the amino acid sequences derived from the mpb83 (33, 45) and mpt83 (20) genes which is absent in MPB70 (43) fits well with our present observations at the protein level, with 26-kDa MPB83 recovered in the detergent phase after extraction with Triton X-114 (Fig. 4). In attempts to determine the N-terminal amino acid sequence of MPB83, we found the N terminus to be blocked. This was expected for the 26-kDa form of MPB83 recovered in the detergent phase after Triton X-114 extraction (Fig. 4) since the lipid is attached to the N-terminal cysteine residue (55). However, blocking of the N terminus of the 23-kDa form of MPB83 still remains to be explained since no evidence for lipidation of this protein has been obtained so far.

To our knowledge, MPB70 and MPB83 represent the first example of two distinct but highly homologous proteins in the M. tuberculosis complex with differences in localization in the mycobacterial cell. This indicates, in turn, that they have different functions which remain to be defined. Additional examples of such pairs of homologous proteins should be sought to enlarge our knowledge of the interactions between the mycobacterial cell and the infected host.