Philip Draper

National Institute for Medical Research, Mill Hill, London NW7 1AA, England

Received 9/25/98 Accepted 10/9/98

4. Porins in mycobacteria

Comparison of rates of permeation of the envelope of M.chelonae by more- or less-hydrophilic beta-lactams led to a further conclusion: permeation by hydrophilic molecules has properties consistent with the presence of water-filled pores in the permeability barrier (21). This sort of pore was already well recognized and studied in Gram-negative organisms, where it is formed by specialized outer-membrane proteins known as porins (23,24). Though there are examples of porins with considerable specificity for particular types of molecule, the common sort is relatively non-specific and allows passage of small hydrophilic molecules up to 5-600 Da in mass. Thus Gram-negative bacteria are protected by their outer membranes from large noxious molecules ? enzymes, antibodies or toxins ? but still have access to the small molecules needed for nutrition. A search was made for analogous pore-forming proteins in mycobacteria, and a protein with appropriate properties was isolated from highly purified walls, devoid of plasma membrane, of M.chelonae (25). This was able to form pores in artificial membranes in the form of liposomes or planar bilayers. The pores were larger than those produced by the most abundant porins of, for example, E.coli, but specific activity of the protein was lower. In these respects the mycobacterial protein resembled the porin of Ps.aeruginosa, OprF.

Protein with pore-forming ability was subsequently extracted from envelopes of M.smegmatis (26). This formed even larger pores than the protein from M.chelonae, which it resembled in having some specificity for cations. A pore-forming protein has since been purified from M.smegmatis and its size and amino-terminal sequence determined (27). Its reported pore size differs from that noted in (26), though the methods of measurement were different, and it is not clear whether these are the same or different proteins. The material investigated in (26) included only one active species, as did that from M.chelonae (28). A mutant strain of M.smegmatis has recently been obtained with modified permeability (29). The exact nature of the defect is not yet known, but one possibility is that this strain has a mutated porin gene.

4.1. Porin-like protein of M.tuberculosis

Both M.chelonae and M.smegmatis are rapid-growing species which have rarely been associated with infection in humans, so the question remained whether the slow-growing species possessed similar pore-forming proteins, and particularly what sort of porin would occur in the major human pathogen M.tuberculosis. An investigation of wall associated proteins of M.tuberculosis detected a protein with some sequence resemblance to outer membrane proteins of Gram-negative bacteria (30), but nothing was known of the physico-chemical properties of this protein. We attempted to extract pore-forming proteins from purified envelopes of Mycobacterium microti, a species very closely related to M.tuberculosis but not regarded as a human pathogen. Extracts were able to form pores in liposomes, but we were not able, using cultures grown on a practicable scale, to obtain sufficient active protein to be visible as a stained band on sodium dodecyl sulphate-polyacrylamide electrophoresis (31).

Interim results from the sequencing of the genome of M.tuberculosis H37Rv at the Sanger Centre, Cambridge, UK, and preliminary analyses of the data obtained were regularly released on the World Wide Web, and an open reading frame MTCY31.27 was provisionally identified as having homology with ?outer membrane proteins?. We cloned and expressed this gene in E.coli (32). The protein obtained was able for form pores in liposomes and planar lipid bilayers, and it was possible to measure several relevant physical properties; the pore size was apparently a little smaller than that of the M.chelonae porin. The gene contained a possible signal sequence at its amino-terminus, with a length and a putative cleavage site rather similar to those of known secreted mycobacterial proteins (the antigen 85 group, which are mycoloyl transferases) (33). Expression of the gene lacking the proposed signal sequence produced an inactive protein; it is not yet known whether the sequence is cleaved from the protein when it is expressed normally by M.tuberculosis.

Comparison of the M.tuberculosis gene with sequences in the published database showed that it possessed a region at its carboxyl-terminus with strong homology to the corresponding region of a number of known proteins, most of which were porins of the ?OmpA family? from Gram-negative bacteria. These proteins differ in several respects from the major porins of E.coli; they occur in that species, and are the major porins of Ps.aeruginosa (24). Typically they form larger pores but have considerably lower specific activities than the classic porins of E.coli, and differ from the latter by occurring as monomers instead of trimers. We have proposed the name OmpATb for the porin-like protein of M.tuberculosis to draw attention to the relationship. The gene is identified as ompA in the published complete sequence of the M.tuberculosis genome (1), and has been given the formal number Rv0899. From the published sequences and from the relative molecular masses it is clear that OmpATb is not the same as the wall-associated protein described in (30).

We showed by reverse transcription-polymerase chain reaction of mRNA extracted from M.tuberculosis that ompA is expressed in growing cells (32), and also that, insofar as its properties could be measured, the activity extracted from walls of M.microti resembled the recombinant protein. This material reacts with rabbit antiserum to OmpATb (34). It is still not clear whether M.tuberculosis possesses only one type of porin, or whether it has several like E.coli. However, the conclusion that the mycobacterial pathogen makes use of a porin-like passive transport mechanism to convey small hydrophilic molecules across its outer lipid permeability barrier is not affected by the question of whether there are several or one species of pore involved.

Meanwhile a pore-forming protein has been extracted from whole cells of Nocardia farcinica (35); nocardias are closely related to mycobacteria and also have mycolic acids in their envelopes, though of smaller size than those of mycobacteria. If this extraction technique is applicable to mycobacteria this is an important finding, because it greatly simplifies the preparation of porins from bacteria, making it easier to investigate them.

Properties of mycobacterial proteins so far identified are summarized in table 1, which also includes data for OprF, the major porin of Ps.aeruginosa, which also belongs to the OmpA family, and for OmpF and OmpC of E.coli, which belong to a different family operating as trimers. It is not clear that the porin activity identified in M.smegmatis as active in planar lipid bilayers (26) is the same as the 40 kDa polypeptide active in the liposome swelling assay (27).

Properties of the mycobacterial porins so far identified are compared with OprF (a typical porin of the OmpA family) and major porins of E.coli. Mass is measured by SDS-PAGE. 1Diameters measured using diffusion of neutral hydrophilic substances into proteoliposomes. 2Diameters measured by diffusion of ions in planar lipid bilayers. 3Porins of this type exist, and are only active, as stable trimers.

Data taken from (7). Amounts are given as percent of the total recovered capsule, which amounted to 2-3% of the dry mass of the bacteria.1 Individual components as percent of total polysaccharide. The protein consists of a large number of molecular species.

4.2. Porins in other slow-growing mycobacteria

The availability of DNA probes for ompA of M.tuberculosis has permitted a search for the gene in other mycobacteria by polymerase chain reaction (PCR) (36). It was present in four members of the M.tuberculosis group: M.tuberculosis, M.microti, M.bovis and M.bovis BCG and also in M.avium and Mycobacterium intracellulare, but not in another slow-growing pathogen, Mycobacterium kansasii, nor in several rapid-growing species, including M.smegmatis and M.chelonae. The PCR reaction requires a very close match between probe and DNA to be amplified, so this data does not rule out homologies between porin genes of various mycobacteria.

4.3. Access of molecules to mycobacteria

It is probable that lipophilic molecules are able to diffuse across the outer permeability barrier of mycobacteria. Small hydrophilic substances can pass through the porin pores, whereas larger hydrophilic molecules are unable to obtain access to the inside of the mycobacterial cell. So an ideal antituberculosis drug should either be a rather small molecule or a rather lipophilic one. Surprisingly, even the small and relatively polar drug isoniazid passes so readily through lipid bilayers in liposomes that it is not possible to measure any difference in diffusion rate when OmpATb is incorporated into the liposomes (37). A bilayer of which one leaflet was mycolate would be a substantially thicker than a bilayer prepared from natural phospholipid, so rates of diffusion of lipophilic substance would be slower through the mycobacterial envelope than into liposomes. At present the nature of the ?other half? of the outer lipid permeability barrier is unknown, so calculations of the thickness of the barrier are not possible. Though monolayers of trehalose dimycolate have been prepared, these were on air-water surfaces and not suitable for measurement of diffusion (16,17). So theoretical or experimental determination of the permeability of the mycobacterial outer lipid barrier is not possible except insofar as it can be measured in whole cells. Data obtained with beta-lactam drugs of varying degrees of hydrophobicity have been mentioned above (21,22) and have been reviewed (38). An important point is that for small molecules the outer permeability barrier alone is an insufficient protection since some diffusion occurs, albeit at a low rate, and other mechanisms such as drug-metabolizing enzymes or efflux pumps are needed. The relative importance of diffusion through the outer permeability barrier and through porin channels is unknown for the majority of solutes; porin-negative mutants of mycobacteria (if viable) would be useful tools in exploring this important area.