Hypoxic Response of Mycobacterium tuberculosis Studied by Metabolic Labeling and Proteome Analysis of Cellular and Extracellular Proteins

Metabolic labeling.

A starter culture of M. tuberculosis H37Rv was grown at 37°C in modified Sauton medium to an optical density at 580 nm (OD₅₈₀) of ca. 1.0 (6). Sterile, 10-ml polystyrene tubes (Nunc, Roskilde, Denmark) or 125-ml polycarbonate Erlenmeyer flasks (Corning, Acton, Mass.) containing 6.7 or 20 ml of modified Sauton medium, respectively, were inoculated with 8 × 10⁶ bacteria per ml from the starter culture. Erlenmeyer flasks were placed in a standard shaking incubator at 37°C (normal cultures), whereas tubes with tightly screwed caps (low-oxygen cultures) were placed at 37°C under magnetic stirring at ∼100 rpm. Metabolic labeling was performed by the addition of 10 μCi of l-[³⁵S]methionine and l-[³⁵S]cysteine (Redivue Promix, Amersham Pharmacia Biotech, Buckinghamshire, United Kingdom)/ml. After 20 h, bacteria were harvested by centrifugation, and the medium was also collected. The bacterial pellet was washed once in phosphate-buffered saline (PBS; pH 7.4) and resuspended in 300 μl of a suspension containing equal volumes of 0.1-mm glass beads and PBS with 0.1% sodium dodecyl sulfate (SDS) and 1 mM phenylmethylsulfonyl fluoride. The bacteria were lysed for 5 min at maximum speed on an MS2 minishaker (IKA Works, Inc., Wilmington, N.C.). The harvested medium (culture filtrate) was sterile filtered and concentrated ca. 20 times in a Centricon-3 ultrafiltration unit (Amicon, Danvers, Mass).

Growth under defined oxygen tensions.

M. tuberculosis H37Rv was grown in Middlebrook medium in insect cell culture flasks under a continuous flow of nitrogen containing oxygen tensions of 1, 5, and 20% as previously described (31). Culturing was continued until an OD₆₅₀ of ca. 0.5 was reached. Bacteria were harvested by centrifugation and lysed by using 0.1-mm glass beads.

2-DE analysis.

During rehydration, 40 μl of each lysate or culture filtrate was applied to 13-cm IPG pH 4-7L strips (Amersham Pharmacia Biotech, Uppsala, Sweden) according to the manufacturer's instructions. Focusing started at 500 V (1 h), was increased to 1,000 V (1 h), and finally was increased to 8,000 V (2 h) in an IPGphor unit (Amersham Pharmacia Biotech). The second-dimensional separation was performed in SDS-10 to 20% polyacrylamide gel electrophoresis gradient gels in the Protean IIxi system (Bio-Rad, Richmond, Calif.). The gel was either silver stained (26) or Coomassie blue stained (Gelcode Blue stain reagent; Pierce, Rockford, Ill.) for protein identification by MS, or the gel was blotted to polyvinylidene difluoride (PVDF) membrane, and the membrane was exposed to Biomax MR film (Kodak, Rochester, N.Y.) for 3 to 42 days to make autoradiographs.

Image analysis of 2-DE gels.

Autoradiographs and silver-stained gels were scanned with a Umax Powerlook III scanner (Umax Systems GmbH, Willich, Germany) in transparency mode and analyzed with the Phoretix 2D gel analysis software (version 5.01; Non Linear Dynamics, Newcastle upon Tyne, United Kingdom). Spot volumes were normalized to the total spot volume, and spots from lysates or culture filtrates of M. tuberculosis cultured under stress conditions were matched to the unstressed control. Spots were considered to be more intense when they showed at least twofold-increased normalized volume compared to the respective control experiment. The changes in the protein levels were considered valid only if they were observed in at least two independent experiments.

Antibodies.

Mouse monoclonal antibodies HYB 76-5 and HBT10 specific for the M. tuberculosis proteins HspX and Ald, respectively, have been described elsewhere (1, 15). A rabbit monospecific polyclonal antibody K2050 directed against M. tuberculosis protein Rv0569 was raised by immunizing rabbits with the corresponding recombinant, His-tagged protein obtained as previously described (22).

Identification of 2-DE spots.

After PVDF membranes were exposed to films to prepare autoradiographs, the membrane was used for protein identification by immunodetection. The same blot was stained by consecutive reactions with different antibodies as previously described (23).

For identification by MS, silver- or Coomassie blue-stained spots were excised from the 2-DE gel and digested with trypsin by a protocol essentially as described previously (26). Briefly, aliquots of the generated tryptic peptide mixtures were analyzed by matrix-assisted laser desorption ionization-time of flight (MALDI-TOF) MS on a REFLEX III instrument (Bruker Daltonik, Bremen, Germany). For low-amount samples, the tryptic peptide mixtures were concentrated and desalted on a Poros R2 perfusion chromatography column and eluted with 1 μl of 2,5-dihydroxybenzoic acid matrix solution directly onto the MALDI _target. For protein identification, the peptide masses obtained by MALDI-TOF MS were used to query the H37Rv protein database and the NRDB sequence database (European Molecular Biology Laboratory). The H37Rv database contains proteins derived from the 3,924 predicted open reading frames of the M. tuberculosis H37Rv genome (5).

LC-MS.

Cells were lysed and the crude total lysate was subjected to tryptic digestion for 16 h at 37°C in 0.2 M ammonium bicarbonate at an enzyme/protein ratio of 1:20. Upon completion, the samples were dried, resuspended in 5% acetonitrile, and analyzed by LC-MS with a Finnigan LCQ mass spectrometer (Thermo Finnigan, San Jose, Calif.). The peptides resulting from the proteolytic digest were separated on a Magic C18 reversed-phase chromatography column (5μ 200A, 0.2 by 50 mm; Michrom Bioresources, Inc., Auburn, Calif.) and eluted with a gradient of 5 to 48% acetonitrile for 30 min followed by 48 to 64% for 10 min. The eluted peptides were directly analyzed by electrospray ionization MS for each growth state on a “total-ion” basis. RplL (Ribosomal protein L7/L12; Rv0652) was used as an internal standard to determine the relative abundance of the proteins of interest rather than only the total protein concentration to determine the relative abundances of proteins in complex mixtures. This technique is based on the monitoring of a predictably stable known protein and determining relative abundance changes based on its stability. Confirmation analysis of the proteins identified by 2-DE was done by LC-MS analysis by direct comparison of the 1 and 5% oxygen protein lysates with that of 20% protein lysate. In silico, trypsin digestion of the protein of interest was performed by using the Protein Prospector program MS-Digest (http://prospector.ucsf.edu/). Specifically, signature peptide analysis consisted of direct comparison of three unique peptides and their corresponding multiply charged states. We chose to extract three noncomigrating peptides to quantify each protein. The mass of each peptide was confirmed by comigration of multiply charged state ions. We selected three independent peptides to monitor for each protein since a single peptide would only uniquely identify between 96 and 97.5% of the proteome, two peptides considered together uniquely identify between 99.29 and 99.88% of the proteome, and three peptides increases the resolution to >99.9% or a single protein. The relative abundance was determined from the average of each of the representing peptides for each protein.

Steady-state levels of M. tuberculosis proteins under defined oxygen tensions.

In the study by Yuan et al. (31), 2-DE of M. tuberculosis lysates grown under 1.3 and 20% oxygen demonstrated much higher levels of HspX under oxygen limitation. A detailed comparison of the gels suggested that other protein spots were also more intense under hypoxia. To identify differences in protein levels between aerobically grown M. tuberculosis and that grown (under limited-oxygen conditions), bacteria were cultured in flasks with defined oxygen atmospheres of 1, 5, and 20% oxygen, respectively, and the steady-state levels of proteins were analyzed by 2-DE and silver staining (Fig. 1 ). The protein patterns obtained were analyzed by 2-DE gel evaluation software, and the investigations were focused on the more intense spots. At 5% oxygen, 21 spots showed at least a twofold increase (encircled) compared to normal conditions at 20% oxygen. The same comparison with lysate proteins from 1 and 20% oxygen cultures showed that 14 spots were more abundant at 1% oxygen. The overall 2-DE pattern observed at 1% oxygen was similar to those of the previously reported 2-DE gels from M. tuberculosis cultures grown at 1.3% oxygen (31).

Metabolic labeling of M. tuberculosis proteins under oxygen limitation.

To examine the de novo synthesis of M. tuberculosis proteins, cultures were radiolabeled with [³⁵S]methionine and [³⁵S]cysteine. Aerobic growth was obtained by shaking, and microaerophilic growth was achieved by culture in sealed, agitated tubes (i.e., the Wayne model [29]). To allow characterization of the extracellular proteins, the medium was collected after harvest of the bacteria, sterile filtered, and concentrated by ultrafiltration. Both cellular (lysates) and extracellular proteins (culture filtrates) were then analyzed by 2-DE, followed by transfer to PVDF membrane and autoradiography (Fig. 2 and 3). The investigations were again focused on the spots that were more abundant under hypoxia, and 10 spots were found to be induced in lysates. 2-DE analysis of the culture filtrate proteins showed that 14 spots were more abundant under low-oxygen conditions. It should be noted that spots identified as more intense in metabolically labeled culture filtrates may represent proteins with increased expression, but increased secretion or bacterial lysis could also explain the enhanced levels of a protein in culture filtrates during oxygen limitation.

Identification of M. tuberculosis proteins more abundant under low-oxygen conditions.

Protein spots found to be more intense under oxygen limitation were identified either by immunodetection or by tryptic digestion of excised spots from silver- or Coomasie blue-stained gels, followed by MALDI-TOF MS analysis of the resulting digestion mixture (Table 1).

(i) HspX.

Increased steady-state level of this protein in lysates was identified at 5% oxygen (eight spots) and at 1% oxygen (five spots) by immunodetection with monoclonal antibody HYB 76-5 and by MALDI-TOF MS analysis (Fig. 1). By metabolic labeling of M. tuberculosis proteins, induced protein synthesis of HspX was demonstrated in lysates (three spots, Fig. 2), and the protein was also more abundant in culture filtrates (three spots, Fig. 3). HspX is a member of the small heat shock protein family of chaperones, and as many as eight spots representing different isoforms of HspX were observed in the lysate obtained from culturing at 5% oxygen. This apparent heterogeneity of HspX was also observed in normal aerated cultures in another study (18).

Under hypoxic conditions, increased levels of HspX were found in M. tuberculosis and M. bovis BCG (4, 7, 8, 10, 25, 31). In addition, during in vitro infection of macrophages, expression of HspX is induced (19, 31). Among the induced genes identified in a microarray study of the hypoxic response of M. tuberculosis by Sherman et al. (25) is Rv3133c encoding a two-component response regulator. _targeted disruption experiments suggest a role for the Rv3133c protein in regulation of HspX expression in response to hypoxia (25). In agreement with this, Rv3133c was found to be induced in M. bovis BCG under oxygen starvation (4). In the present study, Rv3133c was not among the identified proteins, and a close inspection of the relevant section of the gels did not reveal any obvious more-intense spots in this region. Additional studies are needed to explain this apparent discrepancy.

(ii) Rv2623.

Increased steady-state levels of Rv2623 in lysates were identified at 5% oxygen (two spots) and at 1% oxygen (two spots) by MALDI-TOF MS analysis (Fig. 1). By metabolic labeling of M. tuberculosis proteins synthesis of Rv2623 was found to be induced in lysates (two spots, Fig. 2), and the protein was also more abundant in culture filtrates (one spot, Fig. 3). Elevated levels of Rv2623 are also found under reduced oxygen tension when the M. bovis BCG strain is studied (4, 10) and when the human macrophage cell line THP-1 is infected with M. bovis BCG (19), suggesting that Rv2623 could be important for intracellular survival. Interestingly, Rv2623 contains two domains unique for a family of universal stress proteins. It has been suggested that UspA (for “universal stress protein A”) from Escherichia coli has a function related to the growth arrest state caused by starvation (20), and it is tempting to speculate that Rv2623 may play a similar role in M. tuberculosis.

(iii) Rv2626c.

Increased levels of Rv2626c in lysates were found in single spots at 1 and 5% oxygen by MALDI-TOF MS (Fig. 1). The protein was also more abundant in metabolically labeled M. tuberculosis lysates (one spot, Fig. 2) and culture filtrates (one spot, Fig. 3) under low-oxygen conditions. The increased levels of Rv2626c observed in the present study are in agreement with the microarray study (25) and the study of M. bovis BCG in the Wayne model (4). Rv2626c shows similarity to B. subtilis YlbB protein, a homolog of IMP dehydrogenase (5).

(iv) BfrB (Rv3841, bacterioferritin).

BrfB was identified by MALDI-TOF MS in a single spot in lysates from both experimental models (Fig. 1 and 3). Bacterioferritins are involved in intracellular storage of iron, and in the microarray study bfrB was also found to be induced at reduced oxygen tensions (25).

(v) Rv0569 (a conserved hypothetical protein).

Rv0569 was identified by MALDI-TOF MS and immunodetection as a more abundant protein in metabolically labeled lysates and culture filtrates (Fig. 2 and 3); increased levels were also found in cultures at 5% oxygen tension, whereas no significant increase was observed at 1% oxygen. The microarray study also identified Rv0569 as induced under hypoxia (25), but it was not identified in the proteome studies of M. bovis BCG under oxygen-limiting conditions (4, 10). This may reflect a difference in the hypoxic response in M. tuberculosis and M. bovis BCG or simply that low-mass proteins (<15 kDa) were not included in the M. bovis BCG studies which used 10% (10) and 12.5% (4) SDS-polyacrylamide gel electrophoresis gels.

(vi) Fba.

The level of Fba (for “fructose biphosphate aldolase”), an enzyme of the glycolytic pathway, was found to be increased only in culture filtrates of early microaerophilic cultures (one spot, Fig. 3). Interestingly, Bai et al. (2) observed increased activity of this type of aldolase sensitive to EDTA in cell extracts from M. tuberculosis H37Rv under conditions of low-oxygen tension. In Bacillus subtilis and Lactobacillus brevis, increased levels of this enzyme were also observed under anaerobic conditions (17, 24). Fba was not identified in the microarray study or in the proteome studies of the M. bovis BCG response to low-oxygen tension that did not include culture filtrates (4, 10, 25). The presence of this housekeeping enzyme in the extracellular environment of the mycobacterium is surprising, although this has been observed previously (14, 21).

(vii) Ald.

By immunodetection with monoclonal antibody HBT10, increased levels of l-alanine dehydrogenase (or 40-kDa antigen) were identified only in culture filtrates of metabolically labeled cultures (three spots, Fig. 3). In M. smegmatis, increased alanine dehydrogenase activity in extracts derived from bacteria grown in the Wayne model was also reported (11). It has been suggested that Ald is involved in peptidoglycan synthesis; its localization in the cell wall and the extracellular environment supports this hypothesis (12). However, Ald is not essential since M. bovis and M. bovis BCG strains lack this enzyme (1, 12). Importantly, material (available at http://schoolniklab.stanford.edu/projects/tb.html) supplementary to the microarray study (25) also indicates that Ald expression is induced by hypoxia.