Introduction

Top Abstract Introduction Methods Results Discussion References

An increasing number of microorganisms, many of which are putative human pathogens, produce enzymes which degrade collagen.¹ The mechanism of penetration of Mycobacterium tuberculosis into the tissues and bloodstream is poorly understood but, as with other lung diseases,^2 3 gastrointestinal infections,⁴ and necrotic conditions,⁵ it is reasonable to assume that there is a contribution from proteases capable of disrupting the extracellular matrix of the pulmonary epithelium and the blood vessels. The aim of this study was to identify and characterise collagenolytic activity in Mycobacterium reference strains and in those isolated from patients with tuberculosis.

	Methods

Top Abstract Introduction Methods Results Discussion References

MYCOBACTERIAL CULTURES
First early morning expectoration samples from two patients with pulmonary tuberculosis and two with miliary tuberculosis, pleural fluid from two patients with pulmonary tuberculosis and one with miliary tuberculosis, and the first early morning urine sample from one patient with renal tuberculosis were collected, seeded, grown, and identified as M tuberculosis by the National Tuberculosis Reference Laboratory. The following ATCC reference strains were donated by the referred laboratory and CENID Microbiología, México: M tuberculosis H37Rv and H37Ra (virulent and attenuated mycobacteria isolated from a human lung, used for susceptibility testing^6 7), M kansasii, M fortuitum, M microti, M terrae, M avium, and M gordonae. These strains were seeded in protein-free Proskauer-Beck-Youmans (PBY) medium and maintained at 37°C for 7-9 weeks before bacteria were harvested.⁸ Some cultures were maintained for as long as 12 weeks before harvesting.

CULTURE FILTRATE PROTEIN EXTRACT (CFPE)
The culture medium was separated from the bacterial mass by filtration through 0.45 µm and 0.22 µm membrane filters (Millipore Corp). Proteins were precipitated with solid ammonium sulphate crystals (80% final saturation) and centrifuged for 30 minutes at 16 000g. The precipitate, suspended in 5 ml phosphate buffered saline (PBS) containing 0.2 mg/ml phenylmethylsulphonyl fluoride (Sigma Chemicals), was dialysed extensively against PBS and 0.5 ml aliquots (6 mg/ml) were stored in liquid nitrogen until used. Some filtrates were precipitated at 40% and 60% (NH₄)₂SO₄ final saturation.

COLLAGEN
Collagen was obtained from the skin of two month old female Wistar rats according to the method described by Epstein.⁹ The specific activity of radiolabelled collagen was 1 × 10⁶ cpm/mg of protein.

COLLAGENOLYTIC ASSAY
The enzymatic activity was determined according to the method of Johnson-Wint¹⁰ using 120 µg CFPE and 12 500 cpm radioactive collagen/well. Some experiments were performed with 20 mM/well EDTA tetrasodium salt (Sigma Chemicals). Controls included buffer, 0.001% trypsin (type IX from porcine pancreas), and Clostridium histolyticum collagenase (high purity, type III fraction A).

COLLAGENASE IDENTIFICATION
The enzyme was identified using a modification of the method described by Taylor and Birkedal-Hansen¹¹ that included elimination of SDS from the gel with 3% Triton X-100 and overnight incubation at 37°C. The region with collagenolytic activity in the CFPE containing gel (-mercaptoethanol reduced and boiled sample/lane in 10% slab gels¹²) was located, purified by electroelution, dialysed, lyophilised, and kept at -20°C until used. The non-denatured state of the substrate was corroborated by the inclusion of trypsin, an enzyme incapable of breaking down collagen but highly active upon gelatin.

MOLECULAR MASS DETERMINATION
The molecular mass of the enzyme (75 000 by SDS-PAGE) was determined by Sephacryl S-300 gel filtration chromatography. 500 µg of enzyme partially purified by electroelution were suspended in 0.05 M Tris HCl buffer (pH 7.5) containing 4 mM CaCl₂, poured onto a 1.2 × 100 cm glass column previously equilibrated with the same buffer, and eluted with the Tris HCl buffer at a flow rate of 12 ml/hour. Standards included horse heart myoglobin (17 kDa), bovine serum albumin (66 kDa), purified human IgG (150 kDa), and bovine liver catalase (240 kDa). The activity of each fraction was tested as described above.

	Results

Top Abstract Introduction Methods Results Discussion References

The collagenolytic activity of CFPE from H37Rv was four times higher than that from H37Ra (43.4% and 9.1%, respectively). Addition of EDTA to the H37Rv CFPE blocked the enzymatic activity by 88%, thus establishing that the enzyme in the virulent strain was dependent on divalent cations (fig 1). Similar results were obtained with replicate cultures. The enzymatic activity was secondary to a 75 kDa protein which was clearly established by the degradation of collagen containing gels and molecular mass determination by Sephacryl S-300 gel filtration chromatography.

View larger version (7K): [in this window] [in a new window]   Figure 1   Collagenolytic activity of M tuberculosis H37Rv and H37Ra culture filtrate protein extract in the presence of calcium or EDTA as calcium chelant. Results are expressed as mean (SE). Background counts per minute (cpm) have already been subtracted. Maximum activity, calculated assuming that all the radioactivity (12 500 cpm/well) was liberated, corresponded to Cl histolyticum (75.5).

The collagenolytic activity was not changed by the use of ammonium sulphate saturation to precipitate the CFPE, the amount of H37Rv CFPE precipitated with a final ammonium sulphate saturation of 40%, 60% and 80% being 4380 (304) cpm, 4224 (215) cpm, and 4146 (182) cpm, respectively. To determine the influence of culture time on enzymatic activity, five H37Rv cultures were seeded and harvested after 32, 38, 46, 70, and 103 days of culture and had CFPE activity of 20%, 29%, 34%, 33%, and 15%, respectively, indicating that the collagenolytic activity decays with time.

The collagenolytic activity of CFPE obtained from mycobacteria other than M tuberculosis was also tested. M microti showed the highest level of activity followed by M avium, M kansasii, M terrae, M gordonae, and M fortuitum which had the least activity (table 1). Enzymatic variability was also observed in culture filtrates obtained from mycobacteria isolated from patients with tuberculosis (table 2).

	Discussion

Top Abstract Introduction Methods Results Discussion References

The presence of collagenolytic activity in the CFPE of M tuberculosis H37Rv strain was secondary to a 75 kDa enzyme whose activity was dependent on divalent cations and was blocked by EDTA, suggesting a mechanism of non-competitive inhibition. These characteristics are similar to those reported for Clostridium histolyticum collagenase¹³ and the majority of collagenases.¹⁴ Even though we recognise the possible influence of collagenase degradation before harvesting, the differences between the CFPE samples of H37Rv virulent and H37Ra avirulent mycobacterial strains were striking and were independent of batch to batch variations, source of seed, and time of culture.

None of the mycobacteria we tested had activity greater than M tuberculosis H37Rv. As in the case of catalase activity, once considered a virulence factor,¹⁵ collagen degrading activity is important but not vital in the pathogenesis of the disease since there was no uniform activity in M avium and M kansasii, both of which induce tuberculosis. M fortuitum and the non-disease inducers M terrae and M gordonae had very poor enzyme activity. Although tissue destruction is secondary to an enhanced cellular immune response, mycobacterial metalloproteinases might be important for bacterial penetration^1 16 as has been shown with other infectious agents.^4 17 Despite the above considerations, it was interesting to observe that M tuberculosis isolated from tuberculosis patients with different clinical syndromes (renal, miliary, pulmonary) had greater activity than the avirulent reference strain H37Ra.

Our results confirm the presence of collagen degrading enzymatic activity in the culture filtrate protein extract of all the mycobacteria tested. Extracellular proteases may also play a role in the pathogenesis of mycobacterial infection. However, the recent resolution of the genome for M tuberculosis H37Rv (website: www.sanger.ac.uk) predicts at least 38 genes coding for proteins of virulence and more than 250 macromolecular and/or micromolecular degrading proteins. One of these, the product of gene Rv0198c, a zinc metalloprotease with a molecular weight of 73.8 kDa, is similar to the enzyme we describe in this work.