Export of recombinant Mycobacterium tuberculosis superoxide dismutase is dependent upon both information in the protein and mycobacterial export machinery. A model for studying export of leaderless proteins by pathogenic mycobacteria.

We have investigated the expression and extracellular release of enzymatically active superoxide dismutase, one of the 10 major extracellular proteins of Mycobacterium tuberculosis, both in its native host and in the heterologous host Mycobacterium smegmatis. We found that the M. tuberculosis superoxide dismutase gene, encoding a leaderless polypeptide of Mr approximately 23,000 representing one of the four identical subunits of the enzyme, is expressed constitutively under normal growth conditions and at a 5-fold increased level under conditions of hydrogen peroxide stress. The highly pathogenic mycobacterium M. tuberculosis expresses 93-fold more superoxide dismutase than the nonpathogenic mycobacterium M. smegmatis, and it exports a much higher proportion of expressed enzyme (76 versus 21%); taking both expression and export into consideration, M. tuberculosis exports approximately 350-fold more enzyme than M. smegmatis. In M. smegmatis, recombinant M. tuberculosis superoxide dismutase is expressed at 8.4 times the level of the endogenous enzyme and the proportion exported (66%) approaches that in the homologous host; hence M. smegmatis exports up to 26-fold more of the recombinant than endogenous enzyme. Interestingly, subunits of the M. tuberculosis and M. smegmatis enzymes readily and stoichiometrically exchange with each other, forming five different complexes of four subunits, both when the enzymes are expressed in the recombinant host and when the purified enzymes are incubated together; however, each subunit retains its characteristic metal ion, iron for M. tuberculosis and manganese for M. smegmatis. Compared with the cell-associated enzyme, the supernatant enzyme of recombinant M. smegmatis is enriched for M. tuberculosis enzyme subunits, consistent with preferential export of the M. tuberculosis enzyme. Recombinant M. tuberculosis superoxide dismutase transcomplements a superoxide dismutase-deficient Escherichia coli, resulting in a reduction of sensitivity of the strain to oxidative stress, but the enzyme is not exported from this nonmycobacterial host. Our findings indicate that the information for export of the M. tuberculosis superoxide dismutase is contained within the protein but that export additionally requires export machinery specific to mycobacteria.

Recombinant M. tuberculosis superoxide dismutase transcomplements a superoxide dismutase-deficient Escherichia coli, resulting in a reduction of sensitivity of the strain to oxidative stress, but the enzyme is not exported from this nonmycobacterial host. Our findings indicate that the information for export of the M. tuberculosis superoxide dismutase is contained within the protein but that export additionally requires export machinery specific to mycobacteria.
Mycobacterium tuberculosis secretes or otherwise releases a large number of proteins into its extracellular milieu. Approximately 11 such proteins are released into broth cultures in sufficient abundance so as to merit designation as major extracellular proteins. In all cases examined thus far, these major extracellular proteins are also released into the phagosome of M. tuberculosis in human mononuclear phagocytes. The functions of several of these proteins are known, including glutamine synthetase (subunit molecular mass 58 kDa), mycolyl transferase (30/32-kDa complex), and iron-specific superoxide dismutase (subunit molecular mass 23 kDa).
Extracellular proteins of M. tuberculosis are of great interest because of the likelihood that these proteins play critical roles in host-pathogen interaction, because these proteins are leading candidates for incorporation into subunit vaccines, and because these proteins are prime targets for the development of new antimycobacterial drugs (1)(2)(3)(4).
Most of the extracellular proteins have leader sequences, and their export presumably occurs after processing of the leader peptide by a leader peptide-specific peptidase. However, two proteins, glutamine synthetase and superoxide dismutase, do not have leader sequences. The mechanism by which these proteins or their subunits are exported is unknown.
In previous studies, we have examined the extracellular release of M. tuberculosis glutamine synthetase. These studies demonstrated that glutamine synthetase is abundantly exported by pathogenic mycobacteria, including M. tuberculosis, Mycobacterium bovis, and Mycobacterium avium but not by nonpathogenic mycobacteria, including Mycobacterium smegmatis and Mycobacterium phlei. The studies also demonstrated that the information for export was contained within the protein since recombinant M. tuberculosis glutamine synthetase expressed in M. smegmatis was abundantly (Ͼ250 g of protein per liter of M. smegmatis culture and amounting to Ͼ90% of all glutamine synthetase molecules synthesized) exported, whereas the endogenous M. smegmatis enzyme was located intracellularly (Ͼ95% of all enzyme molecules synthesized) (1,5).
These findings with respect to the export of glutamine synthetase prompted us to investigate the other leaderless major extracellular protein of M. tuberculosis, superoxide dismutase. In particular, we wished to determine if its pattern of export followed the same paradigm as glutamine synthetase. A previous report in the literature that recombinant M. tuberculosis superoxide dismutase is not exported in M. smegmatis suggested that superoxide dismutase follows a different paradigm (6).
Superoxide dismutase (EC 1.15.1.1) catalyzes the dismutation of the superoxide anion. The enzyme is believed to be an integral part of the cell's defense against toxic oxygen metabolites. Phagocytic cells, including mononuclear phagocytes, the host cells of M. tuberculosis, generate these toxic oxygen metabolites as part of their effort to fend off pathogenic invaders (7). Superoxide dismutase has been characterized at the DNA and protein level in several mycobacterial species in addition to M. tuberculosis. All of these enzymes are members of a family of homologous superoxide dismutases with similar biochemical properties and tertiary structure, combining either two or four subunits to form the active enzyme and coordinating either iron or manganese ions at the catalytic site (4,8,9). A feature of the M. tuberculosis superoxide dismutase that distinguishes it from the other mycobacterial enzymes studied is that it binds iron as its metal ion ligand instead of manganese (4).
This study presents our findings on the expression and extracellular release of endogenous and recombinant superoxide dismutases from M. tuberculosis and M. smegmatis. The experiments were designed to answer the following fundamental questions concerning the expression and secretion of this enzyme in mycobacterial species. (i) To what extent is superoxide dismutase exported in these two species? (ii) Is recombinant M. tuberculosis superoxide dismutase expressed and secreted in M. smegmatis? (iii) Is recombinant M. tuberculosis superoxide dismutase capable of transcomplementing a completely superoxide dismutase-devoid Escherichia coli mutant, a species that, when superoxide dismutase-proficient, expresses the enzyme strictly intracellularly? (iv) Most importantly, does the export of superoxide dismutase and glutamine synthetase follow a similar or dissimilar paradigm?

Bacterial Cultures
M. tuberculosis strain Erdman (ATCC 35801) and M. smegmatis 1-2c (provided by Peadar O'Gaora, Imperial College School of Medicine at St. Mary's, London, UK (10)) were grown in 7H9 medium (Difco) or Sauton's medium (11) at pH 6.7. M. tuberculosis cultures were maintained in stationary flasks at 37°C and a mixture of 5% CO 2 , 95% air, growing logarithmically for approximately 3 weeks, whereas M. smegmatis cultures were maintained in an environmental shaker at 180 rpm for 3-4 days or in unshaken flasks for 6 -8 days at 37°C and a mixture of 5% CO 2 , 95% air. M. smegmatis transformants were initially grown in 7H9 medium supplemented with 2% glucose before being transferred to standard medium containing hygromycin at 50 g/ml.
E. coli DH5␣ (12) was grown in Luria-Bertani (LB) medium, and transformants were grown under the appropriate selective conditions. The superoxide dismutase-relevant E. coli strains QC771, QC772, QC773, and QC774, specifying the parent strain, single superoxide dismutase mutants, and the sodA sodB double mutant (provided by Danièle Touati, Institut Jaques Monod, Université Paris, Paris, France (13)), respectively, were cultured in selective medium as recommended (13). For the analysis of physiologic parameters, these strains and the QC774 double mutant transcomplemented with the M. tuberculosis superoxide dismutase were grown under various culture conditions that are specified in the corresponding figure legends.

Molecular Cloning and DNA Sequence Determination of M. tuberculosis and M. smegmatis Superoxide Dismutase Genes
High molecular weight genomic DNA of M. tuberculosis Erdman was isolated by hot phenol extraction and used as a template in 40 rounds of amplification at 94-55-72°C of a region containing the published DNA sequence of the structural gene for superoxide dismutase from M. tuberculosis H37Rv (6). The amplification product with the 5Ј and 3Ј primers at its respective ends (5Ј primer, 5Ј 3 3Ј GTGGCCGAATA-CACCTTGCCAGAC, specifying the first 8 amino acids of superoxide dismutase including the initiator methionine GTG; 3Ј primer, 5Ј 3 3Ј TCAGCCGAATATCAACCCCTTGGT, specifying the last 7 amino acids of the enzyme plus the stop codon) was cloned into pCR TM (Invitrogen) and used as a probe for Southern analyses with M. tuberculosis genomic DNA digested with various restriction enzymes. Southern hybridizations were performed on nitrocellulose membranes at 60 -65°C for 24 h in 5ϫ SSC (1ϫ SSC is 150 mM sodium chloride and 15 mM sodium citrate, pH 7.2) or 5ϫ SSPE (1ϫ SSPE is 150 mM sodium chloride, 10 mM sodium phosphate, 1 mM EDTA, pH 7.2) with probes labeled to specific activities of 5 ϫ 10 7 to 1 ϫ 10 8 of 32 P cpm/g. Filters were washed at hybridization temperature with 0.2ϫ SSC or SSPE, dried, and autoradiographed on Kodak X-Omat XAR5 or Fuji RX film for various times either at room temperature or at Ϫ70°C using a Cronex Lightning-Plus intensifying screen (DuPont).
BamHI restriction fragments of ϳ6 -10 kilobase pairs (kb), 1 containing the hybridizing genomic DNA fragment of ϳ8 kb, were cut from an agarose gel, eluted by centrifugation through siliconized glass wool, ethanol-precipitated, verified by Southern hybridization to contain the superoxide dismutase-encoding fragment, and digested with various restriction enzymes other than BamHI to produce smaller fragments for rapid directional cloning in the E. coli/mycobacteria shuttle plasmids pSMT3 (10) or pNBV-1 (14). Restriction fragments of ϳ4 kb generated with the enzymes BamHI and ClaI yielded unambiguous patterns in Southern and colony hybridizations performed as described above as well as in restriction site mapping analyses that verified that the isolated restriction fragment was identical to the published DNA sequence of the canonical shuttle cosmid T264 encompassing the sodA locus (15). Based on this sequence and using the same reaction conditions as described above, we employed two pairs of primers (5Ј primer 1 (primer 1 in Fig. 1): 5Ј 3 3Ј CAGT (arbitrary nucleotides) -ATCGAT (ClaI site) -GCTTAGTTGGTGAGATTGCGAAAG, annealing from 541 to 518 nucleotides upstream of the superoxide dismutase gene's initiator GTG codon, and 3Ј primer 1 (primer 2 in Fig. 1): 5Ј 3 3Ј CAGT (arbitrary nucleotides) -GGATCC (BamHI site) -GTAGCTCGAGTAAC-CAAGCATGCG, annealing to nucleotides 156 to 133 downstream of the gene's TGA stop codon; and a second pair, 5Ј primer 2 and 3Ј primer 2, with exactly the same sequences except that the restriction sites were switched) to amplify two fragments of ϳ1.3 kb that were restricted with ClaI and BamHI and cloned directly into pNBV-1 restricted with the same enzymes. Both plasmid inserts were completely sequenced by the chain termination method (16), and their sequence was identical to the published sodA locus on cosmid T264 (15).
The M. smegmatis 1-2c superoxide dismutase gene sequence was established for an amplification product whose primers were derived from M. tuberculosis and Mycobacterium fortuitum superoxide dismutase gene sequences. Two amplification products were combined to yield the entire coding region. A first pair of primers (M. smegmatis 5Ј primer 1, 5Ј 3 3Ј GTGGCCGAATACACCTTGCCAGACCTG, corresponding to the first 8 amino acids of the M. tuberculosis superoxide dismutase gene, and M. smegmatis 3Ј primer 1, 5Ј 3 3Ј CACATCGGC-CCAGTTCACGACGTTCCAAAA, corresponding to codons 190 to 183 in the M. tuberculosis enzyme) yielded an amplification product of ϳ570 bp. A second set of primers (M. smegmatis 5Ј primer 2, 5Ј 3 3Ј GAC-CAGTTCGGCTCGTTCGACAAGTTC, corresponding to codons 101-109 of the M. smegmatis enzyme, and M. smegmatis 3Ј primer 2, 5Ј 3 3Ј CTAGCCGAAGATCAGGCCGTTGGT, corresponding to the last 7 codons plus stop codon of the M. fortuitum enzyme), yielded an amplification product of ϳ325 bp. Both amplification products were sequenced completely, combined, and cloned into pCR TM .

Constructs Used for the Expression of Recombinant M. tuberculosis Erdman Superoxide Dismutase in M. smegmatis
The constructs used to express recombinant M. tuberculosis superoxide dismutase in M. smegmatis were based on the aforementioned amplification products of a stretch of ϳ1.3 kb of M. tuberculosis DNA spanning the superoxide dismutase coding region plus 5Ј-and 3Ј-flanking sequences. One construct contained the amplification product inserted into the ClaI-BamHI sites of the multi-cloning region of pNBV-1 in a 5Ј 3 3Ј direction, thereby allowing transcription of the superoxide dismutase gene to proceed in a direction opposite the direction of transcription of the vector encoded ␤-galactosidase promoter; the second construct was inserted in the same restriction sites but in opposite orientation, thus aligning the superoxide dismutase and ␤-galactosidase promoters. Constructs were first established in E. coli DH5␣ and then electroporated into M. smegmatis 1-2c as described (10). Stable M. smegmatis 1-2c transformants were selected and characterized for expression of superoxide dismutase as described below.

Construct Used for the Expression of Recombinant M. tuberculosis Erdman Superoxide Dismutase in E. coli QC774
For transcomplementation of the E. coli sodA sodB double mutant QC774, an amplification product of the M. tuberculosis superoxide dismutase gene was ligated in the NcoI-HindIII sites of the high level expression vector pKK233-2 (provided by Andrew Campbell, Brown University, Providence, RI (17)). The amplification reaction was performed under standard conditions (see above) with the cloned M. tuberculosis superoxide dismutase gene as a template and the following primer pair: 5Ј primer (primer 3 in Fig. 1), 5Ј 3 3Ј GATC (arbitrary nucleotides) -CCATGG-(NcoI site, the internal ATG provides the initiator methionine, the second G is part of the alanine codon, the first amino acid of the mature protein) -CCGAATACACCTTGCCAGACCT-GGACTGGGAC (corresponding to codons 1-11) and 3Ј primer (primer 4 in Fig. 1), 5Ј 3 3Ј CTAG (arbitrary nucleotides) -AAGCTT (HindIII site) -TCAGCCGAATATCAACCCCTTGGTCTGCGA (corresponding to the last 9 codons plus stop codon). The amplification product was first cloned into pCR TM , verified by DNA sequencing, recloned into pKK233-2 as an NcoI-HindIII fragment, maintained in DH5␣, and transformed into QC774 under the appropriate selection conditions (13).

Analysis of Expression of Endogenous and Recombinant Superoxide Dismutases
Characterization of endogenous and recombinant superoxide dismutases was performed by the following analytical methods.
Polyacrylamide Gel Electrophoresis of Cell Pellets and Culture Supernatants-M. tuberculosis, M. smegmatis, and E. coli were cultured in standard medium as described above. Aliquots of the cultures were removed and separated into cell pellets and supernatants by centrifugation. Cell pellets were taken up in a small volume of 1ϫ phosphatebuffered saline (50 mM sodium phosphate and 150 mM sodium chloride, pH 7.2) and lysed either by vortexing vigorously with 60-mesh crystalline alumina beads (Fisher) for 5 min at room temperature (mycobacterial species) or by the addition of lysozyme at 100 g/ml for 20 min on ice (E. coli). Insoluble material was collected by centrifugation, and the cell pellet proteins were dialyzed extensively against 1ϫ phosphatebuffered saline, centrifuged again, and adjusted to a final volume containing the proteins of 1 ϫ 10 8 lysed cells per l. The culture supernatants were concentrated in an Amicon diaflo unit using membranes with a molecular weight cut-off of 3,000, dialyzed extensively against 1ϫ phosphate-buffered saline, centrifuged, and adjusted to a final volume such that 1 l contained the supernatant proteins of 1 ϫ 10 8 cells of the original culture. Proteins in the cell pellets and culture supernatants were analyzed by electrophoresis on standard, 10% denaturing polyacrylamide gels, followed by staining with Coomassie Brilliant Blue. Protein concentrations in the cell pellets and culture supernatants were determined by the bicinchoninic acid reagent (Pierce).
Immunoreactivity of Endogenous and Recombinant Superoxide Dismutases-Polyvalent antibodies against the purified M. tuberculosis superoxide dismutase were raised as follows. The enzyme was purified from culture supernatants of 3-week-old M. tuberculosis Erdman cultures by ammonium sulfate precipitation and chromatography on DEAE-CL6B and Superdex 75 to yield a homogeneous enzyme preparation as judged by denaturing polyacrylamide gel electrophoresis and staining with Coomassie Brilliant Blue. The superoxide dismutase band (subunit, 23 kDa) was excised and eluted from the gel. Rabbits were immunized with an emulsion of purified superoxide dismutase and Syntex adjuvant formulation (18) containing 100 g of purified enzyme and 1.5 g of N-acetylmuramyl-L-alanyl-D-isoglutamine, followed by three booster immunizations at 12-day intervals with 100 g of enzyme. Antibody preparations had reciprocal titers of Ͻ100 preimmunization and ϳ10 5 post-immunization against the native and denatured enzyme isolated from M. tuberculosis Erdman cell pellets and culture supernatants. In Western blots, superoxide dismutase was the only protein of all cellular and extracellular M. tuberculosis Erdman proteins that reacted with the antibodies.
For the detection and identification of endogenous and recombinant superoxide dismutases, proteins in cell pellets and culture supernatants were electrophoresed in 10% denaturing polyacrylamide gels, transferred to nitrocellulose membranes, and incubated first with preand post-immune rabbit anti-M. tuberculosis superoxide dismutase antibodies at dilutions of 1:100 and 1:5,000, respectively, and then with alkaline phosphatase-conjugated goat anti-rabbit antibodies. The immunoblots were stained with alkaline phosphatase-specific color development reagents (Bio-Rad) and then scanned in an Eagle Eye II TM unit (Stratagene) to quantify expression of the various superoxide dismutases. Calibration of the Western blots for the densitometric evaluation of the immunoreactivity of the polyvalent antibodies, described above, and their cognate antigen, iron-and manganese-specific superoxide dismutases, was performed with superoxide dismutase purified from supernatants of M. tuberculosis Erdman cultures.
N-terminal Analysis of Endogenous and Recombinant Superoxide Dismutases-Cell pellets and culture supernatants were first dialyzed against low ionic strength phosphate buffer and separated from low molecular mass components (Ͻ30 kDa) by centrifugation in centricon units (Amicon) containing a membrane with a 30-kDa cut-off and subsequently electrophoresed on duplicate 10% denaturing polyacrylamide gels and transferred to polyvinylidene difluoride membranes in 10 mM CAPS and 10% methanol, pH 11. One blot was used to identify bands containing superoxide dismutase by staining with antibodies. Corresponding bands on the duplicate blot were then subjected to N-terminal amino acid sequence analysis at the UCLA protein microsequencing facility using a Porton 2090 E amino acid sequencer.
Metal Analysis of Recombinant Superoxide Dismutase-The 5-band complex of the active recombinant M. tuberculosis superoxide dismutase enzyme was transferred from native polyacrylamide gels to nitrocellulose membranes. The top, middle, and bottom bands were excised from the membranes, solubilized in a small volume of dimethyl sulfoxide, and finally brought up in 1 ml of water that had been Chelex-treated to eliminate traces of metal ions in the distilled water. Manganese and iron atoms were measured by flame atomic absorption spectrometry.
Enzymatic Assay of Superoxide Dismutase Activity in Cell Pellets and Culture Supernatants-All bacterial cell pellets and culture supernatants were electrophoresed on 15% non-denaturing polyacrylamide gels with constant voltage of 100 V per gel at room temperature and assessed quantitatively and qualitatively for enzymatic activity by staining as described (19). The gels were incubated for 30 min in 2.5 mM nitro blue tetrazolium, followed by incubation for 20 min in 30 mM potassium phosphate, 30 mM TEMED, and 30 M riboflavin, pH 7.8, and visualization of the superoxide dismutase activity bands by illuminating the gels for 10 min on a lightbox. Areas of superoxide dismutase activity were visible as white bands against a blue background. Quantitation of superoxide dismutase activity was standardized by scanning activity gels using the two commercially available E. coli enzymes, iron superoxide dismutase and manganese superoxide dismutase. Enzyme units were first defined by the nitro blue tetrazolium reduction assay (19) and then correlated to the observed intensities of the white bands in the activity gels. Furthermore, since we had established growth curves of all described bacterial strains by determining colony-forming units at various time points during their respective incubation periods, we were able to correlate any densitometrically determined enzyme activity to the number of bacteria needed to produce that signal.
Sensitivity of E. coli QC774 and Its Transformants to Paraquat and Hydrogen Peroxide-For an assessment of the paraquat (methyl viologen, 1,1Ј-dimethyl-4,4Ј-dipyridinium) sensitivity of QC774 and its transformants, broth cultures were inoculated with 1 ϫ 10 6 cells per ml and grown in the presence of various paraquat concentrations for 6 -72 h. Colony-forming units were enumerated at various time points during the growth phase of the cultures by plating aliquots of the cultures on selective media. For comparison, the parent strain QC771 was analyzed in the same fashion. Hydrogen peroxide sensitivity of QC774 and its transformants was determined by growing broth cultures in LB medium to a cell density of 1 ϫ 10 8 cells per ml and then treating the cultures with 2.5 or 5 mM hydrogen peroxide for 30 or 60 min. Aliquots of the cultures were removed at the end of the treatment period, plated, and enumerated for colony-forming units after 24 h at 37°C. For comparison, the parent strain QC771 was also assessed for its hydrogen peroxide sensitivity under the same culture conditions.

Materials
All chemicals and enzymes were purchased from Sigma unless indicated otherwise and were of the highest grade available. All oligonucleotides were purchased from Genosys Biotechnologies, Inc.

Endogenous Superoxide Dismutase Is Abundantly Exported from M. tuberculosis but Not M. smegmatis
We first established a base line for the expression of the endogenous superoxide dismutases from both mycobacterial species by assessing cell pellets and culture supernatants of unshaken M. tuberculosis cultures and both shaken and unshaken M. smegmatis cultures for enzyme activity. The results clearly showed that in all cultures enzyme activity increased in parallel with cell density, and the protein patterns did not change significantly during the observation period, suggesting that there are few if any truly early or late released proteins. The only measurable difference was that M. smegmatis cultures in unshaken flasks grew with approximately 50% the growth rate of the shaken cultures. This result further indicated that the proteins in the culture supernatants were stable. This supposition was later confirmed by immunoblot analyses of cell pellets and culture supernatants, which showed no detectable degradation products of the superoxide dismutases. In view of these results, we selected the following time points for assessing the various cultures for the amount, distribution, and activity of expressed superoxide dismutase: 3 weeks for M. tuberculosis cultures and 3 days for M. smegmatis shaken cultures.
The two mycobacterial species differed dramatically in their superoxide dismutase expression patterns with regard to both the level of expression and distribution of the enzyme. In an analysis in which superoxide dismutase expression was quantified by densitometrically scanning immunoblots (Fig. 2 An analysis of enzyme activity in the culture supernatant confirmed these results. M. tuberculosis exported 2,778 nanounits per cell versus 7.6 nanounits per cell for M. smegmatis, a 365-fold difference (Fig. 2, bottom portion).
The presence of extracellular superoxide dismutase in both species was not due to leakage of proteins in the culture medium from dead cells. All cultures were continuously monitored for the release of the intracellular marker enzyme lactate dehydrogenase. For all bacterial cultures, the enzyme was found to be strictly associated with the cellular fraction and not the culture supernatant. Calculated on the basis of the total de-tectable lactate dehydrogenase activity, i.e. intracellular plus extracellular lactate dehydrogenase activity, the extracellularly detectable lactate dehydrogenase activity increased over time but did not amount to more than about 0.5% of the total activity by the end of the growth period for any culture including M. smegmatis expressing recombinant M. tuberculosis superoxide dismutase.

M. tuberculosis and M. smegmatis Superoxide Dismutase Genes
The finding that the enzyme superoxide dismutase is abundantly released by pathogenic but not by a nonpathogenic mycobacterium (6) and that the pattern of expression and extracellular release of this enzyme mimics that of the other major leaderless extracellular enzyme of M. tuberculosis, glutamine synthetase, prompted us to study the expression of M. tuberculosis superoxide dismutase in a heterologous host. This first required that we clone and characterize the M. tuberculosis and M. smegmatis superoxide dismutase genes.
Cloning of the Superoxide Dismutase Gene from M. tuberculosis Erdman-Starting point for the analysis of the M. tuberculosis Erdman superoxide dismutase gene was an amplifica-tion product of genomic DNA, which was based on the published DNA sequence of the structural gene for superoxide dismutase from M. tuberculosis H37Rv (6) and which was used for all subsequent Southern and colony hybridizations. All cloned restriction fragments containing the superoxide dismutase gene were quite large, the smallest being a ClaI-BamHI fragment of ϳ3.8 -3.9 kb, as compared with the described 624 bp open reading frame encoding the enzyme. Based on our experience expressing recombinant M. tuberculosis proteins (2,5) and the then just released information on the mapping and sequencing of the M. tuberculosis genome (15), we decided to use this DNA fragment to amplify a region of ϳ1.3 kb that included the superoxide dismutase coding region, ϳ540 bp of 5Ј-flanking region and ϳ125 bp of 3Ј-flanking region (Fig.  1). This strategy allowed us to exclude other potential open reading frames contained in these larger fragments and to utilize the same restriction sites (ClaI and BamHI) that are present as unique sites in the E. coli/mycobacterial shuttle expression vector pNBV-1 (14) for bidirectional cloning and expression of the superoxide dismutase-containing amplification product. Cloning of the Superoxide Dismutase Gene from M. smegmatis 1-2c-We developed and pursued a similar strategy for the cloning and expression of the M. smegmatis 1-2c superoxide dismutase gene. However, since we were primarily interested in the elucidation of differences in expression and secretion between the M. tuberculosis and M. smegmatis enzymes, we first focused on establishing the DNA sequence of the coding region of the M. smegmatis enzyme, which is shown in Fig. 3 along with the aligned coding region of the M. tuberculosis enzyme. This DNA sequence was determined for an amplification product of M. smegmatis genomic DNA, which was cloned into pCR TM .
Nucleotide Analysis of the M. tuberculosis and M. smegmatis Superoxide Dismutase Genes-The results of the nucleotide analysis of the two DNA sequences coding for superoxide dismutase can be summarized as follows. (i) Both the M. tuberculosis and M. smegmatis genome most likely contain only one gene coding for an iron-specific and manganese-specific superoxide dismutase, respectively. If a second gene is present, its DNA sequence must differ significantly from the ones described here. (ii) Both coding regions are 624 bp long, utilize GTG as the initiator methionine codon, and encode a mature protein of 206 amino acids with a theoretical molecular mass of ϳ23,000 Da. This mass is consistent with the migration patterns of purified superoxide dismutases on denaturing poly-acrylamide gels (Fig. 2, middle portion). (iii) The coding regions are highly homologous to each other and to other described mycobacterial superoxide dismutases (8,9) and to a lesser degree to superoxide dismutases from other bacterial species (20,21). (iv) Although the active site residues, His-28, His-76, His-164, and Asp-160, are conserved between the two sequences, the amino acid residue 145 puts the M. tuberculosis enzyme (His-145) in the category of iron-binding enzymes and the M. smegmatis enzyme (Gln-145) in the category of manganese-binding enzymes, according to a recent structural study on metal ion specificity of superoxide dismutases (4).

Recombinant M. tuberculosis Superoxide Dismutase Is Exported Abundantly in a Heterologous Mycobacterial Host
The two recombinant pNBV-1 constructs described above contained the M. tuberculosis superoxide dismutase gene and flanking DNA sequences in two orientations (Fig. 2). The plasmids were maintained in E. coli DH5␣ where they were stably propagated without any detectable changes and without any indication of expression of recombinant superoxide dismutase, most likely because the E. coli transcription machinery is unable to recognize the mycobacterial promoter. Following transformation of the two recombinant constructs into M. smegmatis 1-2c, several transformants for each construct were assayed for expression of recombinant superoxide dismutase as described above for M. smegmatis cultures. Since the promoter region contained in the plasmid insert cannot be regulated, any expression of recombinant superoxide dismutase in these cultures is constitutive.
The recombinant M. tuberculosis superoxide dismutase was expressed from its own promoter in both pNBV-1 constructs (Fig. 2, top portion). Expression of the correct protein in M. smegmatis was verified by sequence analysis of the 15 Nterminal amino acids, which confirmed that the recombinant protein contained a tryptophan residue in position 10 (AEYTLPDLDW), whereas the endogenous M. smegmatis enzyme contained a tyrosine residue in that position (AEYTLP-DLDY). This difference is very important from an analytical point of view, because the next difference between the two enzymes occurs at position 40. Intra-and extracellular recombinant superoxide dismutases had identical N termini, indicating the lack of a leader peptide.
The recombinant enzymes were expressed at greatly different levels depending on the construct used (Fig. 2, middle  portion). The construct containing the superoxide dismutase insert in the same direction of transcription as the vectorencoded ␤-galactosidase expressed the recombinant enzyme at a level that was only approximately twice that of the endogenous enzyme and the distribution between cell pellet and culture supernatant was similar to that of the endogenous enzyme. In contrast, the construct containing the insert in the opposite direction expressed the recombinant enzyme at a level that was approximately 8.3-fold greater than the endogenous enzyme. Moreover, approximately 66% of the expressed enzyme was exported, a level of export approaching that of the endogenous M. tuberculosis enzyme.
Superoxide dismutase in the culture supernatant of the recombinant M. smegmatis appeared as a set of five bands with slightly different mobilities on the activity gels (Fig. 2,  The recombinant M. smegmatis strain, of course, expresses its own endogenous superoxide dismutase, but its contribution to the total detectable enzyme activity is minimal, since its expression level does not change dramatically when compared with the parent strain. Although we cannot prove directly that the expression level of endogenous M. smegmatis superoxide dismutase remains nearly constant, two observations provide strong support for this assumption. (i) In the activity gel analyses, the level of expression of the M. smegmatis enzyme subunits by the parent strain (1 band) was comparable to that of the recombinant strain (predominantly the lower bands of the 5-band set). (ii) More importantly, N-terminal amino acid sequence analysis of the exported recombinant enzyme (1 band with a molecular mass of ϳ23 kDa on a denaturing polyacrylamide gel and transferred to a polyvinylidene difluoride membrane) demonstrated a Ͼ100:1 ratio of M. tuberculosis: M. smegmatis subunits (presence of tryptophan versus tyrosine at position 10). Since the increase in expression overall was only 8-fold, this result indicated that the increased amount of extracellularly detectable enzyme activity in the recombinant M. smegmatis strain is due primarily to the presence of the recombinant M. tuberculosis enzyme and not to a major increase in export of the endogenous M. smegmatis enzyme.
This disproportionate increase in M. tuberculosis enzyme subunits in the extracellular medium of recombinant M. smegmatis revealed by N-terminal amino acid analysis indicated that M. tuberculosis superoxide dismutase is preferentially exported. To confirm this, we compared the banding pattern of the enzyme in the supernatant and pellet fractions of recombinant M. smegmatis (Fig. 4). We reasoned that if the M. tuber- The subunits of the M. tuberculosis and M. smegmatis enzymes not only mix to form five different complexes of subunit enzymes during expression of the recombinant M. tuberculosis enzyme in M. smegmatis but also when the two purified enzymes are incubated together for a short period (2 h) at room temperature (Fig. 5). Mixing experiments suggested that homologous or heterologous subunits associated with each other with comparable affinity. Moreover, when the middle of the five bands, containing two subunits of each enzyme, is eluted from the gel and incubated overnight in solution, the subunits redistribute into approximately equal amounts of the five different complexes (Fig. 5).

Recombinant M. tuberculosis Superoxide Dismutase Transcomplements a Superoxide-deficient Heterologous Host
To characterize further the functional activity and export capacity of the recombinant M. tuberculosis superoxide dismutase, we studied its capacity to complement a superoxide dismutase-deficient mutant E. coli. The superoxide dismutase proficient parent strain QC771 and the sodA sodB double mutant QC774 were first grown in aerated broth cultures under various conditions to establish critical base-line values required for the evaluation of a successful transcomplementation of the mutant strain by the mycobacterial enzyme. The parent strain QC771 grew very well in LB and in glucose containing minimal medium (Fig. 6A). The mutant QC774 grew in LB medium, but more slowly than the wild type, and it grew in minimal medium only when the medium was supplemented with 20 amino acids; even then, growth was very slow (Fig. 6A). The superoxide generator paraquat slightly inhibited growth of the wild type, but more markedly inhibited growth of the mutant (Fig. 6B), as previously reported (13).
Both the wild type and mutant were assessed for the expression of superoxide dismutase by immunoblotting and activity assays (Fig. 7, lower portion). The wild type (QC771) showed the presence in the cell pellet but not in the culture supernatant of two very faint bands containing enzymatic activity and corresponding to the iron-(M r ϳ21,000) and manganese (M r ϳ23,000)-binding superoxide dismutases. The level of expression was far below that measured even for M. smegmatis, amounting to ϳ0.36 nanounits per cell, consistent with the results of earlier studies of superoxide dismutase from E. coli B (22). The mutant strain (QC774) was completely devoid of any enzyme activity.
We explored the capacity of a mycobacterial enzyme to substitute for the E. coli enzymes by transforming the recombinant plasmid pKK233-2, containing the gene of the M. tuberculosis Erdman superoxide dismutase, into E. coli QC774 (Fig. 7, upper portion). The plasmid pKK233-2 was chosen for its very strong promoter, its optimal placement of regulatory regions, allowing the expression of recombinant enzyme by itself rather than as a fusion protein, and the presence of the initiator ATG codon, overlapping the NcoI cloning site.
Upon transformation, the recombinant strain remained stable, maintaining the introduced plasmid over many cell generations, although the strain did not plate uniformly, an issue that could not be resolved satisfactorily. Presence of the plasmid was also verified by isolating the recombinant pKK233-2 construct and DNA sequencing across the vector insert junctions with primers annealing to the cloned M. tuberculosis superoxide dismutase gene. After induction of recombinant superoxide dismutase expression, strain QC774 grew very well in LB medium and exhibited a growth pattern in minimal medium that approached that of the noncomplemented mutant in minimal medium supplemented with all 20 amino acids (Fig.  6A). The addition of paraquat at a concentration of Յ50 M to LB broth cultures had no growth inhibitory effect on the transcomplemented strain. At 50 M, the difference in colonyforming units between QC774 with and without the recombinant pKK233-2 plasmid amounted to almost 1 log unit by the end of the growth phase (Fig. 6B). At concentrations higher than 100 M, both the wild type and the transcomplemented strains showed significant growth inhibition.
Whereas treatment with 2.5 and 5 mM hydrogen peroxide had only a minimal effect on the wild-type QC771, it exerted a fast-acting, detrimental effect on the QC774 mutant, whose viability at the end of the 60-min treatment period dropped by 2 log units (2.5 mM hydrogen peroxide) and Ͼ3 log units (5 mM hydrogen peroxide), respectively (Fig. 6C). Transcomplementation of the mutant with the M. tuberculosis superoxide dismutase gene substantially protected the strain from oxidative stress, indicating that the recombinant enzyme was active in the heterologous host but could not fully substitute for its endogenous enzymes. Even under the harshest conditions, 5  s. SOD, ϳ50 units); and the mixing of subunits of both enzymes after being incubated together for 2 h at room temperature in phosphatebuffered saline. Enzymes were electrophoresed on a native 15% polyacrylamide gel and stained for enzyme activity (19). The three lanes on the right show, from left to right, recombinant M. tuberculosis superoxide dismutase (rM.tb. SOD, ϳ100 units), expressed in M. smegmatis as a tetramer exhibiting the five different possible combinations of M. tuberculosis and M. smegmatis enzyme subunits; the middle band (Band 3) of the five bands, containing two subunits of each enzyme, after excision from a polyacrylamide gel and immediate re-electrophoresis; and the middle band after excision from a polyacrylamide gel, elution from the gel with ammonium acetate and magnesium acetate, incubation for 24 h in phosphatebuffered saline, and re-electrophoresis, revealing redistribution of the subunits into the five different complexes. mM hydrogen peroxide, the transcomplemented strain's viability differed by Ͻ1 log unit from that of the wild type (Fig. 6C).

Recombinant Superoxide Dismutase Is Not Abundantly Exported from a Nonmycobacterial Host
In the mycobacterial host M. smegmatis, the recombinant M. tuberculosis superoxide dismutase was not only expressed, it was abundantly exported, as in M. tuberculosis. In this case, the hosts were relatively closely related and the enzymes relatively similar. To determine if the recombinant M. tuberculosis enzyme could be abundantly exported in a more distantly related nonmycobacterial host, and one whose endogenous enzymes are even more dissimilar, we investigated directly the expression and export of the recombinant M. tuberculosis superoxide dismutase in E. coli.
For the expression of the mycobacterial enzyme, there were two requirements. First, there was an absolute requirement for the presence of an E. coli promoter. The two constructs used to express the M. tuberculosis enzyme in M. smegmatis are maintained in E. coli QC774 under proper selection but did not express any superoxide dismutase molecules at all. Second, there was a requirement for the presence within the plasmid's NcoI site of an ATG start codon. The M. tuberculosis superoxide dismutase gene uses GTG as its initiator codon, as does the M. tuberculosis glutamine synthetase, but this codon is not accepted as the initiator codon in this plasmid in E. coli (5).
Both immunoblotting and activity assays showed that the recombinant M. tuberculosis enzyme is strictly expressed intracellularly. No protein or activity was detected in the culture supernatant (Fig. 7, lower portion). The expression level was well below that observed for M. smegmatis and M. tuberculosis cultures but ϳ5.3-fold greater than that measured for wildtype E. coli QC771, amounting to ϳ1.9 nanounits of superoxide dismutase activity per cell. This is approximately 1,500-fold below its activity in M. tuberculosis. The recombinant enzyme was of the correct size. It migrated as one band with a molecular mass of ϳ23,000 Da on denaturing polyacrylamide gels, and it migrated with the same mobility as the endogenous M. tuberculosis enzyme on native activity gels. Analysis of the first 7 N-terminal amino acid residues revealed the sequence AEYTLPD, identical to the sequence of the endogenous M. tuberculosis enzyme.

DISCUSSION
Export of Endogenous Superoxide Dismutase-Our study demonstrates that the enzyme superoxide dismutase is abundantly exported in active form from a pathogenic mycobacte-(QC771) or every 12 h (all other cultures) and plating on agar medium. B shows the growth curves of E. coli QC771 and QC774, without and with transcomplementation by recombinant M. tuberculosis superoxide dismutase, in LB medium in the presence of the superoxide generator paraquat. All broth cultures were started at a cell density of 1 ϫ 10 6 cells per ml and grown at 37°C for 12 h with vigorous shaking. Paraquat was added after 2 h of cell growth at various concentrations; shown here are only the growth curves in the presence of the highest concentration examined in detail, 50 M. Changes in cell density were monitored by removing aliquots of the cultures every 2 h and plating on agar medium. C shows the hydrogen sensitivity of E. coli QC771 and QC774, without and with transcomplementation by recombinant M. tuberculosis superoxide dismutase. All broth cultures were started at a cell density of 1 ϫ 10 6 cells per ml and grown at 37°C for 4 -6 h with vigorous shaking until they had reached a cell density of 1 ϫ 10 8 cells per ml. At that time point, cultures were treated with 2.5 or 5 mM hydrogen peroxide for 30 or 60 min, as indicated. At the end of the hydrogen peroxide treatment, aliquots of the cultures were removed and plated on agar medium to monitor viability of the bacteria in each culture. The parent strain QC771, which had been characterized extensively before (13), was included in these studies as a standard for a wild-type E. coli strain. Why superoxide dismutase, a typically intracellular enzyme, is abundantly exported by M. tuberculosis is not clear. Other major extracellular enzymes exported by M. tuberculosis have been implicated in cell wall construction. These include the mycolyl transferases of the 30/32-kDa (Antigen 85) complex and glutamine synthetase (1,3). Superoxide dismutase, however, has not been implicated in cell wall construction. Rather, this enzyme is typically involved in protection against oxidative stress. Superoxide may not only cause damage by itself, it can participate in a chemical reaction with hydrogen peroxide and iron (the Fenton mechanism) that can produce an even more toxic molecule, hydroxyl radical. The host cells of M. tuberculosis are known to produce toxic oxygen metabolites, and it is tempting to speculate that M. tuberculosis releases superoxide dismutase to neutralize these metabolites in the phagosome or lung cavity, the two sites in which the pathogen multiplies in the host. In this regard, superoxide dismutase has been documented in the phagosome of M. tuberculosis in human mononuclear phagocytes. 2 The release of superoxide dismutase in the vicinity of M. tuberculosis might have the advantage of neutralizing toxic oxygen molecules even before they reach the outer cell wall of the organism. Exported superoxide dismutase might thus protect external cell wall structures of the pathogen that are not protected by its intracellular superoxide dismutase. If exported superoxide dismutase serves this purpose, this would constitute the second mechanism by which M. tuberculosis protects itself from toxic oxygen metabolites. By entering mononuclear phagocytes via complement receptors (23), which when ligated by C3 do not trigger an oxidative burst (24), M. tuberculosis avoids much of the toxic consequences of this burst.
It is of interest that the M. tuberculosis superoxide dismutase enzyme binds iron. This may reflect the greater availability of iron than manganese in the human host. It may also be pertinent that M. tuberculosis has evolved an elaborate system for acquiring iron involving the abundant export of an extracellular high affinity iron-binding siderophore, the exochelins (25). The binding of iron by superoxide dismutase may allow the enzyme to counteract hydroxyl radical production via the Fenton mechanism in two ways, by scavenging iron and by neutralizing superoxide. In this regard, it is noteworthy that exochelins themselves block hydroxyl radical production via the Fenton mechanism both in tissue culture in vitro and in isolated rabbit hearts (26).
Despite differences in their expression and export, the superoxide dismutases of M. tuberculosis and M. smegmatis display a high degree of identity or similarity at the nucleotide and amino acid level. These enzymes are also highly homologous to the superoxide dismutases of other mycobacteria including Mycobacterium marinum, Mycobacterium kansasii, and M. fortuitum (8,9,27,28).
Export of Recombinant Superoxide Dismutase-Three lines of evidence support the concept that the information for abundant export of M. tuberculosis superoxide dismutase is contained within the protein. First, an unusually large amount of the leaderless protein is exported by M. tuberculosis, whereas other abundantly expressed proteins, e.g. certain heat shock proteins, remain cell-associated. Second, the recombinant enzyme is abundantly exported from M. smegmatis, whereas the homologous endogenous enzyme in M. smegmatis is not abundantly exported. Although the M. tuberculosis enzyme is not expressed as abundantly in the heterologous host as in the homologous host, possibly reflecting different "set points" for expression of this enzyme, expression of the recombinant enzyme in M. smegmatis is nevertheless 8-fold that of the M. smegmatis endogenous superoxide dismutase. Moreover, 66% of the recombinant enzyme is exported versus 21% for the endogenous enzyme. The net result is that M. smegmatis exports 26-fold more of the recombinant enzyme than its own endogenous enzyme. Third, N-terminal amino acid analysis of exported recombinant superoxide dismutase and a comparison of the banding patterns of exported and retained recombinant superoxide dismutase indicated that recombinant M. smegmatis preferentially exports M. tuberculosis superoxide dismutase. A theoretically possible alternative explanation for these results is that M. tuberculosis superoxide dismutase is more stable than M. smegmatis superoxide dismutase extracellularly, whereas M. smegmatis superoxide dismutase is more stable than M. tuberculosis superoxide dismutase intracellularly. However, we know of no reason why one of these highly related enzymes whose subunits indiscriminately complex with each other would be more stable in one setting and less stable in the other. Taken together, the evidence strongly indicates that an intrinsic characteristic of M. tuberculosis superoxide dismutase instructs its export.
The high correlation between expression and export of superoxide dismutase raises the possibility that export is additionally dependent on the level of expressed enzyme. However, while expression dependence could be a factor in the high level export of M. tuberculosis superoxide dismutase both in the endogenous host and in recombinant M. smegmatis and the low level export of M. smegmatis superoxide dismutase in the parent strain, this mechanism would not account for either (a) the abundant export of superoxide dismutase by M. tuberculosis but not other proteins expressed in even greater abundance or (b) the preferential export of M. tuberculosis superoxide dismutase versus the endogenous superoxide dismutase in recombinant M. smegmatis.
We do not have an explanation for the different ratio of intracellular to exported recombinant superoxide dismutase in the M. smegmatis strains with the unidirectional orientation of the superoxide dismutase and ␤-galactosidase promoters. One possibility, alluded to in the previous paragraph, is that a threshold level of superoxide dismutase must be reached before the bacterial cell is able to export significant amounts of the enzyme.
Our finding that the recombinant M. tuberculosis superoxide dismutase is abundantly exported from M. smegmatis is in contrast to the result of a previously published study (6). Although the investigators detected superoxide dismutase in the culture supernatant of M. tuberculosis by immunoblotting and activity assays, they could not detect any enzyme activity in supernatants of M. smegmatis nor in M. smegmatis cultures transformed with the M. tuberculosis superoxide dismutase gene.
Interestingly, some M. smegmatis clones that express recombinant M. tuberculosis superoxide dismutase export more endogenous enzyme than the parent M. smegmatis strain. Since M. smegmatis and M. tuberculosis subunits freely exchange with each other, one possible explanation for this observation is that endogenous enzyme subunits are exported in association with recombinant enzyme subunits, i.e. the more export-prone recombinant subunits in effect carry the endogenous subunits across the wall with them.
The recombinant M. tuberculosis superoxide dismutase successfully transcomplemented a superoxide dismutase-deficient E. coli, proving that the recombinant enzyme is active. As in the case of the heterologous mycobacterial host M. smegmatis, the level of expression of the recombinant enzyme in E. coli was much lower than in its homologous host but still 5-fold greater than the level of expression of the endogenous superoxide dis-mutase in the parent E. coli strain. Again, the lower expression of recombinant enzyme in the heterologous than homologous host may reflect a lower set point for superoxide dismutase expression in E. coli. Additionally, it may reflect differences between the two species in codon usage, protein folding, or association of subunits.
Although the recombinant M. tuberculosis superoxide dismutase is abundantly exported from a mycobacterial host, it is not abundantly exported from the nonmycobacterial host E. coli. One possible explanation for this is that the E. coli export machinery does not recognize the mycobacterial protein. Although the superoxide dismutases from E. coli exhibit the same 4 metal ion coordinating amino acid residues at His-26, His-73/81 (iron/manganese-binding enzyme), His-160/171, and Asp-156/167 (20) as the mycobacterial enzymes, the variation between the E. coli and the M. tuberculosis enzymes is greater than that between different mycobacterial species. Another possible explanation for the lack of export in E. coli is that E. coli lacks a nonclassical export pathway present in mycobacteria.
Superoxide Dismutase and Glutamine Synthetase: Parallel Models for Study of Export of Leaderless Multimeric Proteins-A common theme appears to govern the expression and export of superoxide dismutase and glutamine synthetase. Both are leaderless proteins that are abundantly expressed and exported by M. tuberculosis. Indeed, these two enzymes are among the 10 major extracellular proteins in M. tuberculosis. Both enzymes are exported abundantly in a heterologous mycobacterial host, M. smegmatis, whereas the endogenous counterparts in M. smegmatis are not abundantly exported. Finally, both enzymes are expressed in active form but not exported from the nonmycobacterial host E. coli. The similarities with respect to the expression and export of M. tuberculosis superoxide dismutase and glutamine synthetase in the homologous and heterologous host make these two enzymes excellent models for studying export mechanisms of M. tuberculosis proteins across its complex cell wall, including differences in export machinery between pathogenic and nonpathogenic mycobacteria. Possible export mechanisms specific to pathogenic mycobacteria include chaperone-guided docking to a membranebound transporter or cell exit via specific porin tunnels.
With respect to dissecting such mechanisms of export, another pertinent similarity between superoxide dismutase and glutamine synthetase is that both enzymes are multimeric complexes of identical subunits, 4 in the case of superoxide dismutase and 12 in the case of glutamine synthetase. However, one interesting difference between the two enzymes is that in the case of superoxide dismutase but not glutamine synthetase, heterologous subunits of the enzyme from M. tuberculosis and M. smegmatis readily exchange with each other. In either case, whether these proteins are exported as a multimeric complex or whether the individual subunits are first exported and then assembled on the extracellular aspect of the cell surface remains to be determined.