The Multiple cpb Cysteine Proteinase Genes ofLeishmania mexicana Encode Isoenzymes That Differ in Their Stage Regulation and Substrate Preferences*

The cpb genes ofLeishmania mexicana encode stage-regulated, cathepsin L-like cysteine proteinases that are leishmanial virulence factors. Field inversion gel electrophoresis and genomic mapping indicate that there are 19 cpb genes arranged in a tandem array. Five genes from the array have been sequenced and their expression analyzed. The first two genes, cpb1 and cpb2, differ significantly from the remaining 17 copies (cpb3–cpb19) in that: 1) they are expressed predominantly in metacyclic promastigotes (the form in the insect vector which is infective to mammalian macrophages) rather than amastigotes (the form that parasitizes mammals); 2) they encode enzymes with a truncation in the COOH-terminal extension, an unusual feature of these cysteine proteinases of trypanosomatids. Transfection ofcpb1 into a cpb null mutant resulted in expression of an active enzyme that was shown by immunogold labeling with anti-CPB antibodies to be targeted to large lysosomes. This demonstrates that the 100-amino acid COOH-terminal extension is not essential for the activation or activity of the enzyme or for its correct intracellular trafficking. Transfection into thecpb null mutant of different copies of cpb and analysis of the phenotype of the lines showed that individual isoenzymes differ in their substrate preferences and ability to restore the loss of virulence associated with the null mutant. Comparison of the predicted amino acid sequences of the isoenzymes implicates five residues located in the mature domain (Asn18, Asp60, Asn61, Ser64, and Tyr84) with differences in the activities of the encoded isoenzymes. The results suggest that the individual isoenzymes have distinct roles in the parasite’s interaction with its host. This complexity reflects the adaptation of cathepsin L-like cysteine proteinases to diverse functions in parasitic protozoa.

fied as belonging to the papain superfamily (1). They show in general a marked stage-regulated expression with there being a very low level of mRNA detected in multiplicative promastigotes, higher levels in metacyclic promastigotes, and very high mRNA levels in amastigotes (17). The encoded cysteine proteinases similarly show stage regulation in that most of the multiple enzymes are at the highest activity in amastigotes (15,16). There are, however, two CPB activities that appear to be predominantly expressed in metacyclic promastigotes (18). The amastigote's CPB activity has been categorized into four groups (A, B, C, and E) on the basis of their physical characteristics and substrate preferences (15,16). Group A cysteine proteinases are glycosylated, as assessed by their ability to bind concanavalin A, whereas groups B and C do not bind this lectin. Groups B and C enzymes differ in their elution from anion exchange resin (reflecting charge differences) and have different substrate preferences; group C enzymes are less efficient at cleaving substrates with a bulky amino acid in the substrate P1 position (15). Group E enzymes differ from the other amastigote cysteine proteinases in having slower mobilities in gelatin-SDS-polyacrylamide gels and apparently being membrane-associated (16). NH 2 -terminal sequence analysis confirmed that cysteine proteinases of groups A, B, and C are encoded by cpb genes, leading to the hypothesis that different copies of the repeat encode cysteine proteinase isoenzymes with distinct characteristics (16).
The cpb genes, together with homologs from other Leishmania species (19,20), Trypanosoma brucei (21,22) and Trypanosoma cruzi (23,24), encode enzymes of the type I class of trypanosomatid cysteine proteinases (12). Type I cysteine proteinases are cathepsin L-like in the mature domain but are characterized by the presence of an unusual, long COOH-terminal extension (CTE). CTEs have also been found on cathepsin L-like cysteine proteinases of rice and other plants, although they have no sequence homology with the CTEs of the trypanosomatid enzymes (25,26); none has been observed with vertebrates' cysteine proteinases. The two copies of cpb which have been described previously encode a 100-amino acid CTE, some residues of which are apparently processed in the production of the mature enzyme (17,27). The function of the CTE is unknown, although several roles have been postulated, including trafficking of the enzyme to the lysosome, modulating the substrate specificity of the enzyme, inhibiting the activity of the enzyme until correctly processed and targeted, and diverting the host's immune system (12).
We have described the generation, by targeted gene disruption, of L. mexicana mutants null for the cpb gene array, which showed that the CPBs are virulence factors (27). Null mutants are less able than wild type L. mexicana to survive in macrophages in vitro and to infect mice. Re-expression of one internal copy of the tandem array in the null mutant gave an active cysteine proteinase that restored the in vitro virulence phenotype (27). This approach of expressing individual copies of the cpb gene array in the null mutant provides a powerful method for determining the activity of, and functional differences between, individual isoenzymes and so allows analysis of the relationship between structure and function. We have now cloned and sequenced three more cpb genes from the array, re-expressed them individually in the null mutant, and investigated their substrate specificities and their ability to restore the virulence phenotype.
Construction and Screening of a L. mexicana Genomic Library in Bacteriophage -15-23-kb fragments of L. mexicana DNA partially digested with Sau3AI were purified on a 1-3 M NaCl gradient and ligated into Dash II (Stratagene). A library of 3 ϫ 10 5 independent clones was generated. The library was screened with 32 P-labeled cpb cDNA clone pSWB1a (17) and positive clones isolated by plaque purification according to standard methods (32). DNA was isolated using Wizard bacteriophage minipreparations (Promega). clones were mapped using restriction enzyme digestion and hybridization with probes able to distinguish the 5Ј-and 3Ј-flanking sequences. cpb genes were subcloned into pBluescript SKII plasmid (pBS) for sequence analysis. A 2.8-kb SalI fragment from 13, containing cpb1 (the first gene of the array), was cloned into the SalI site of pBS(SKϪ) to give plasmid clone pNE2. A 2.8-kb SalI fragment from 20, containing cpb18 (the penultimate gene in the array), was cloned into the SalI site of pBS-(SKϪ) to give plasmid pNX2. A 3.0-kb SalI fragment from 20, containing cpb19 (the last gene of the array), was cloned into the SalI site of pBS(SKϪ) to give plasmid pNX3. The nucleotide sequence covering the cpb open reading frame for each gene was determined in both orientations using a series of overlapping oligonucleotide primers about 300 bp apart. Sequencing was carried out on an ABI 373 DNA sequencer (Perkin-Elmer), and analysis was performed using the Wisconsin GCG package.
Transfection-cpb genes were cloned directionally into the pTEX vector (33). cpb1 was subcloned on a EcoRI/HindIII fragment from pNE2 into the EcoRI/HindIII of pTEX to give pTEXNE2, cpb18 on a EcoRI/HindIII fragment from pNX2 into the EcoRI/HindIII of pTEX to give pTEXNX2, and cpb19 on an HindIII/XhoI fragment from NX3 into the HindIII/XhoI site of pTEX to give pTEXNX3. The cloning of pTEXG2.8 and pTEXSWB1a has been described (27). Transfection of L. mexicana has been described previously (27,34).
Southern and Northern Blot Analyses-Wild type L. mexicana DNA was isolated using a Nucleon kit (Scotlab, Coatbridge). RNA was isolated from different life cycle stages of L. mexicana (mid-log phase promastigotes, stationary phase promastigotes, and amastigotes) using a Qiagen total RNA kit. Southern and Northern blots were carried out as described previously (28). 223-bp DNA probes corresponding to sequence in the CTE of cpb2.8 and the equivalent position in cpb1 were generated by PCR using, respectively, primers DB1 and DB2 (on plasmid template pNE2) and primers DB3 and DB4 (on plasmid template pTEXG2.8; Ref 27). A 1.4-kb EcoRI/XhoI fragment from plasmid pSWB1a was used as a cpb cDNA probe (17). DB1: ACTGATATGTACTGCACTCAG DB2: TATAGGGAGGTGGTCTTCCTC DB3: TTCGATAAGAACTGCACTCAG DB4: TAGTGACAGGTGTTCATGATC DNA fragments were labeled by random priming (Stratagene) and hybridized as described (29).
Field Inversion Gel Electrophoresis-FIGE was carried out at 10°C using a Hoefer Switchback Pulse Controller. 2.5 g of L. mexicana wild type genomic DNA was digested to completion with XhoI and run at 150 volts (at 5.2 V/cm) on a 0.9% GTG agarose gel in 0.5 ϫ TBE buffer. The run-in time was 10 min, run time was 28 h, pulse time was 0.6 to 2.0 s in reverse mode with a F/R ratio of 3:1. The gel was blotted onto Hybond N (Amersham Corp.) and probed with a 1.4-kb EcoRI/XhoI fragment from cpb cDNA clone pSWB1a as described (17).
Electron Microscopy/Immunocytochemistry-Axenic amastigotes were centrifuged at 1,000 ϫ g for 5 min at 20°C and the pellet fixed in freshly prepared 2% (w/v) p-formaldehyde, 0.1% glutaraldehyde in phosphate-buffered saline, pH 7.4, at 4°C for 30 min. Pellets were washed twice for 5 min in cold phosphate-buffered saline and cold processed (modified from Ref. 35) through an increasing series of alcohol concentrations: 50% ethanol, 4°C for 10 min; then 70%, 95%, and 100% ethanol steps at Ϫ20°C for 20 min each, with agitation at 10 min. The samples were subsequently treated with a 50:50 mixture of LR White Medium resin and absolute ethanol at Ϫ20°C for 30 min, then with a 75:25 resin/ethanol mixture at Ϫ20°C for 30 min, then with 100% resin for 8 h (during which time the specimens were brought up to 20°C) and subsequently allowed to infiltrate fully overnight with mild agitation. Parasite pellets were transferred to gelatin capsules in fresh LR White Medium resin and polymerized at 20°C under 365-nm indirect UV light for 2 days. Ultrathin sections were collected onto 300 mesh nickel grids. On-grid immunostaining was performed essentially as described previously (36). Blocking and conditioning steps using 0.2 M glycine and 1% bovine serum albumin in phosphate-buffered saline for 30 min each were followed by incubation for 30 min in anti-CPB antibody diluted 200-fold in phosphate-buffered saline, 1% acetylated bovine serum albumin (Aurion) and then a 30-min incubation in goat anti-rabbit antibody, 10 nm of gold conjugate (Aurion) diluted 20-fold in the same buffer. Sections were briefly counterstained in 0.5% uranyl acetate and viewed with a Zeiss 902 EFTEM at 0 eV to improve contrast.

RESULTS
Polymorphism within the cpb Gene Array-Proteinases encoded by the cpb gene array are polymorphic (16), but it was not known whether this is because of differences in gene sequence between copies of the array or because of post-translational modifications, or a combination of both. We have characterized in detail the cpb locus (Fig. 1A) using genomic mapping, FIGE (Fig. 1B), analysis of clones containing sections of the array, and sequence analysis. Southern blot analysis of genomic XhoI fragments from wild type L. mexicana DNA separated by FIGE and probed with a cpb-specific probe revealed three hybridizing fragments of 47, 4.0, and 2.8 kb (Fig. 1B). XhoI sites are located in the unique 5Ј-and 3Ј-flanking sequences of the array and also just downstream of the cpb open reading frame for the first two copies of the array, but they are not present in any other of the cpb repeat units ( Fig. 1A and Ref. 27). The 4.0-kb XhoI fragment contains cpb1, the first copy of the array, the 2.8-kb fragment the second copy, cpb2, and the 47-kb fragment the remaining copies. As all the copies occur on 2.8-kb SalI fragments (17), and taking into account the position of the XhoI site downstream of the last copy of the array (Fig. 1A), it was calculated that there are 17 gene copies present on the 47-kb XhoI fragment and hence a total of 19 copies in the whole array. The single band at 47 kb indicates that the two allelic arrays contain the same number of copies of cpb. This is in contrast to the type I cysteine proteinase genes of the X10.6 strain of T. cruzi which have 14 and 23 copies in allelic clusters (37).
A library was constructed in by partial digestion of L. mexicana genomic DNA with Sau3AI. The library was screened with the cpb cDNA, and several clones were isolated. Southern blot analysis of clones digested with diagnostic restriction enzymes and hybridized with 5Ј and 3Ј cpb cDNA probes identified fragments that contained either unique 5Ј-or 3Ј-flanking sequence with varying amounts of the cpb repeat ( Fig. 1A) and clones that contained cpb repeat units only. 14 was found to contain the complete cpb1 gene and part of cpb2, whereas 13 contained cpb1, cpb2, and cpb3. 20 contained the last two copies in the array (cpb18 and cpb19) and the 3Ј-flanking sequence. Several of the cpb genes were subcloned into pBluescript for sequence analysis. cpb1 was cloned on a 2.8-kb SalI fragment from 14 (to give pNE2), cpb18 on a 2.8-kb SalI fragment from 20 (to give pNX2), and cpb19 on a 3.0-kb SalI fragment from 20 (to give pNX3). The sequences of the cpb cDNA and internal gene cpb2. 8 have been reported previously (17,27). The nucleic acid sequence of the 2.1-kb HindIII/XhoI fragment containing cpb1 is shown in Fig. 2A. The first 453 bp do not occur within the intergenic sequence of the cpb array; and a probe specific to this region, generated by PCR with primers to the sequences indicated (underlined in Fig. 2A), recognized a single band in Southern blots (Ref. 27 and data not shown), indicating that this sequence is unique. Sequence downstream of the divergence point (uppercase in Fig. 2A) is present within the 2.8-kb cpb repeat unit. The predicted ATG start codon is located at position 804. Protein coding genes in Leishmania and other trypanosomatids are transcribed polycistronically and then processed by trans-splicing and polyadenylation (38,39). Sequence analysis of the cDNA clone pSWB1a (17)  the copy of the cpb repeat from which the cDNA was derived. An AG dinucleotide splice site is located at the same position upstream of the cpb1 start codon ( Fig. 2A, position 661, shown in bold). Other possible AG dinucleotides are located between the one used for the cpb cDNA and the ATG start codon, although the one indicated is a good candidate for the cpb1 splice site as upstream of it are a number of polypyrimidine stretches ( Fig. 2A, double underlined) which have been shown to be important for accurate splice site selection (39). No significant open reading frames were located in the sequence immediately 5Ј to the first cpb gene of the array.
Details of the protein characteristics predicted from the sequences determined for the five cpb genes are shown in Table I. cpb1 has 98% nucleotide identity with cpb2.8 in the sequence that encodes the pre-pro-region (nucleotides 170 -544; see cpb2.8, accession no. Z49962) and 99% identity for the mature domain (nucleotides 545-1197), but only 70% identity for the CTE (nucleotides 1198 -1497). This divergence in the sequence encoding the CTE of cpb1 is also found with the second copy of the array, cpb2 (data not shown). This makes the first two copies of the array significantly different in this domain from the internal genes, exemplified by cpb2.8 (see Fig. 3; cpb2.8 is used as the benchmark for comparisons as it encodes an active cysteine proteinase that is able to restore virulence toward macrophages in vitro, Ref. 27). cpb1 predicts an open reading frame of 38.7 kDa. This is less than the 47.7 kDa of CPB2.8 and is due to the truncation of the CTE to 15 amino acids. This truncation is due to a single base pair insertion (relative to cpb2.8) in the cpb1 sequence which causes a frameshift and the introduction of a stop codon. This single base insertion is also present in cpb2 producing a correspondingly truncated protein (data not shown). In contrast, cpb2.8 and the cDNA encode proteins with 100 amino acid CTEs (Fig. 3).
The nucleotide sequence of cpb19, its predicted open reading frame, and the unique 3Ј-flanking sequence downstream of the gene are shown in Fig. 2B. The underlined sequences show the DNA to which oligonucleotide primers were designed for the PCR of a 590-bp fragment for use as a 3Ј-flanking probe (27). cpb19 encodes a protein of 39.5 kDa with 98% nucleotide sequence identity with cpb2.8 over the pre-pro-region and the mature domain. However, the sequences diverge totally near the start of the sequence encoding the CTE. The point of divergence is coincident with the beginning of the 3Ј-flanking sequence (lowercase, Fig. 2B), which introduces a stop codon that truncates the CTE of cpb19 to 22 amino acids (Fig. 3). No significant open reading frames were located in the sequence immediately 3Ј of the cpb array.
cpb18 encodes a 47.9-kDa protein that also has a high degree of sequence homology with CPB2.8. CPB18 has five amino acid differences in the mature domain compared with CPB2.8 (Fig.  3). Three of these changes (N60D, D61N and D64S) 2 are also found in the cDNA and CPB19, one (H84Y) is present in CPB1 and CPB19, whereas D18N is conserved only in CPB19. CPB18 encodes a full-length CTE of 100 amino acids, which has eight amino acid differences from CPB2.8. Of these changes, six are the same as those that differ between the cDNA and CPB2.8. 10 cysteine residues are present in all three gene products containing the full-length CTEs (Fig. 3). Cysteine residues are thought to be involved in the tertiary structure of the CTE of T. cruzi (40,41). One of the differences in the CTEs of CPB18 and the cDNA (T248R) abolishes a potential N-linked glycosylation site (NCT) which is found in the COOH-terminal extension of CPB2.8. This may have relevance to the glycosylation status of different CPB isoenzymes as the single N-linked glycosylation site in the mature domain (at position Asn 103 ) is conserved in all cpb genes sequenced.
Expression of cpb in Wild Type L. mexicana-We reported previously on the marked stage-regulated expression of the cpb genes with high levels of mRNA being detected in amastigotes, lower levels in stationary phase (metacyclic) promastigotes, and very low levels in multiplicative promastigotes (17). To determine at which stages during the life cycle cpb1 and cpb2 are expressed, Northern blots were performed with a PCRderived probe specific to these genes (Fig. 4). The probe was derived from 223 bp of sequence just downstream of the stop codon of cpb1. On a Southern blot of L. mexicana DNA digested with XhoI, the probe detected 2.8-and 4.0-kb fragments (Fig.  4A, lane 1) corresponding to fragments containing the cpb2 and cpb1 genes, respectively (see Fig. 1, A and B). No hybridization was detected with the remaining copies of the array located on the 47-kb fragment, confirming that the probe is specific for cpb1 and cpb2. In contrast, a PCR-derived probe generated to the corresponding region of cpb2.8 detected the 47-kb fragment but not the 2.8-or 4.0-kb fragments (Fig. 4A, lane 2). This probe would hybridize to cpb2.8, cpb18, the genomic copy encoding the cDNA, and possibly all the other genes located between cpb3 and cpb18 inclusive.
Total RNA was prepared from the three life cycle stages of L. mexicana: mid-log phase (multiplicative) promastigotes, stationary phase (metacyclic) promastigotes, and amastigotes isolated from a mouse lesion. Relative hybridization intensity between life cycle stages on the Northern blots were quantified on a PhosphorImager and normalized to an ␣-tubulin control (Fig. 4E). The cpb cDNA probe, which should hybridize to all the cpb genes, detected very low levels of cpb mRNA in multiplicative promastigotes, higher levels in stationary phase (metacyclic) promastigotes, and the highest level in amastigotes (in the ratio of 1:10:25, Fig. 4B). In contrast, the cpb1/cpb2-specific probe detected a significantly different stage expression with low multiplicative promastigote expression, predominant metacyclic promastigote expression, and very low amastigote expression (the ratio was 1:6.5:0.2, Fig. 4C). The probe designed to the genes cpb3-cpb18 detected 2.2-kb mRNA predominantly present in the amastigote (ratio of 1:6:33, multiplicative promastigote:metacyclic promastigote:amastigote, Fig. 4D). The expression of cpb19 was assessed using a PCR-derived probe generated to sequence immediately downstream of the stop codon. This contains a 3Ј-flanking sequence that would be predicted to be part of a 3Ј-untranslated region for the cpb19 gene. No hybridization, however, was detected with this probe in any of the three life cycle stages (not shown). Furthermore, when cpb19 was expressed from the pTEX vector in the cpb null mutant (see below) it was not possible to detect either cysteine proteinase activity by gelatin gel electrophoresis or protein by Western blotting. Our conclusion is that cpb19 is a pseudogene created through recombination events that produced the tandem array and that the 3Ј-flanking sequence immediately downstream of the gene does not contain the sequence elements required for the correct processing of a functional cpb19 mRNA. A similar truncation in the CTE and lack of expression were observed in the last copy of the type I cysteine proteinase gene array of T. cruzi (37).
Expression of Individual cpb Genes in cpb Null Mutants-We had previously generated, via targeted gene disruption, an L. mexicana cell line null for cpb (designated ⌬cpb) (27) and also cell lines that re-expressed the cpb cDNA or cpb2.8 in the cpb null mutant background (27). To compare the cysteine protein- FIG. 2. Nucleic acid sequence and predicted protein sequence for cpb1 (A) and cpb19 (B). Underlined nucleotides show the positions of oligonucleotides used to make DNA probes specific for the 5Ј and 3Ј flanks (27). The lowercase sequence is unique to the 5Ј (panel A) or 3Ј (panel B) flank. The sequence in uppercase is found in the cpb gene array. The AG dinucleotide splice site used for spliced leader addition in the cDNA (17) is shown in bold, a polypyrimidine stretch upstream of the splice site which may have a role in splice site selection is double underlined, and the SalI site used for cloning is dot underlined. ase activities encoded by different copies of cpb and to assess their biological roles in the parasite, the genes were cloned from Bluescript into the pTEX vector and transfected into ⌬cpb to generate cell lines expressing either CPB1 (⌬cpb-[pTEXNE2]), CPB18 (⌬cpb[pTEXNX2]) or CPB19 (⌬cpb-[pTEXNX3]). Cell lines were grown at a high concentration (500 g ml Ϫ1 ) of G418 to increase expression levels. Extracts were prepared from stationary phase promastigotes of each cell line and analyzed for the presence of CPB protein by Western blotting and for cysteine proteinase activity by substrate-SDS-PAGE using gelatin and fluorogenic substrates (Figs. 5 and 6). Two main bands of cysteine proteinase activity are characteristically detected on gelatin gels with wild type L. mexicana stationary phase promastigote extracts (Fig. 5A, lane 1). These are stage-regulated and occur predominantly in this parasite form (18,42). When CPB1 was re-expressed in the null mutant (lane 2) it had a single cysteine proteinase activity that comigrated with the slower band detected in wild type extracts, whereas the CPB2.8 re-expressor (lane 3) had activities with mobilities between those in the wild type extracts, similar to the situation with wild type amastigotes (15,18). The activity of the CPB18 re-expressor (lane 4) had a mobility very similar to that of the slower band detected in wild type extracts, but analysis of several gels using variable amounts of material and incubation times suggested that in fact it has a slightly faster mobility. The CPB2.8 re-expressor (lane 3) and to a lesser extent the CPB18 re-expressor (lane 4) also had significant activities with much slower mobilities (apparently 30 -40 kDa; but it should be noted that proteins do not migrate strictly according to molecular mass in gelatin-SDS-PAGE). Equivalent activities were not readily detected with wild type or in cell extracts prepared from the CPB1 re-expressor, although activity was detected if sufficient material was used. Western blotting of wild type stationary phase promastigote extracts with anti-CPB antiserum detected two major proteins (25 and 29 kDa, Fig. 5B, lane 1). It should be noted that the sample preparation for the activity gels and Western blots differ, the former samples are not denatured by boiling, which accounts for the different mobilities observed for the same proteins with the two procedures. The CPB1 was also 25 kDa (lane 2), whereas the main cpb products in the CPB2.8 re-expressor (lane 3) and CPB18 re-expressor (lane 4) migrated with a molecular mass of 28 kDa. The lines expressing CPB2.8 or CPB18 (lanes 3 and 4) also contained larger molecular mass proteins (about 38 kDa) detected by the antiserum. The cell lines were also analyzed for their ability to hydrolyze two fluorogenic peptide substrates (Fig. 6). Both cysteine proteinase activities detected in wild type extracts (lane 1) hydrolyzed each of the substrates, although the slower mobility activity showed greater activity toward SucLY-NHMec whereas the reverse was true for the higher mobility activity. The cysteine proteinase in each of the three cell lines expressing CPB1 (lane 2), CPB2.8 (lane 3), or CPB18 (lane 4) were almost equally proficient at hydrolyzing BzFVR-NHMec (Fig. 6A), whereas CPB18 had a significantly higher activity toward SucLY-NHMec than either CPB1 or CPB2.8 (Fig. 6B). The slower mobility activities toward gelatin (Fig. 5) detected for CPB2.8 (lane 3) or CPB18 (lane 4) did not hydrolyze either of the fluorogenic substrates to any appreciable degree (Fig. 6).  (17), CPB1, CPB18, and CPB19. Dashes represent sequence identity. In the numbering system used, the NH 2 terminus of the mature CPB2.8 enzyme is designated as residue 1, and the pro-region is assigned negative numbers decreasing toward the NH 2 terminus. The arrow shows the last residue of the mature domain. The two potential N-linked glycosylation sites are shown in bold. Stars show the conserved cysteine residues in the CTE.
One of the criteria used previously to assess the phenotype of the cpb mutants was their ability to survive in macrophages in vitro (27). Peritoneal exudate cells were removed from BALB/c mice, exposed to stationary phase promastigotes, and assessed for parasite survival after 7 days in culture (Table II). As reported previously (27), the null mutant ⌬cpb survived in just a few macrophages, whereas expressing CPB2.8 in the null mutant restored levels of infectivity almost back to wild type levels. In contrast, expression of CPB1 or CPB18 in the null mutant increased infectivity only a minor extent.
Localization of CPB Isoenzymes in L. mexicana-Immunogold labeling of wild type L. mexicana axenic amastigotes with anti-CPB antiserum (Fig. 7A) resulted in strong labeling of the large lysosomes (termed megasomes) as reported previously (43,44). No labeling was detected in the cpb null mutant axenic amastigotes (Fig. 7B) or in a negative control with no antibody (not shown but similar to Fig. 7B), confirming the specificity of the antiserum used to detect the CPB enzymes. The megasomes were also labeled in axenic amastigotes of null mutants expressing CPB2.8 and CPB1, indicating that both these isoenzymes were targeted to megasomes (Fig. 7, C and D). With each of the three cell lines expressing CPB, no significant labeling was detected outside megasomes, indicating that the traffick-ing of the proteinase was efficient in these parasites, and there was no accumulation of precursors in other compartments of the cell such as the Golgi. Some of the gold-labeled megasomes were adjacent to the flagellar pocket (see Fig. 7D), which could reflect some interaction between the two (either discharging of the megasomes into the flagellar pocket or uptake into megasomes). The lack of gold labeling in the pocket does not support the former possibility, although it is not yet possible to rule this out.

DISCUSSION
This study has shown that the cathepsin L-like cpb cysteine proteinase genes of L. mexicana are found in allelic tandem arrays of 19 copies. We postulated previously (16) that the diversity in biochemical characteristics of the parasite's numerous stage-regulated cysteine proteinases detected using gelatin gels (15,16) was the result of small differences between the genes in the cpb array, although post-translational modifications could also be involved. Support for this working hypothesis has now been provided through the sequence analysis of five independent copies of cpb. The data generated show that although the genes have a very high level of overall nucleotide sequence identity (97-99%), there are a number of differences that alter the amino acid sequence of the encoded proteins or introduce premature stop codons. The significance of these amino acid substitutions became apparent when the individual isoenzymes were expressed in a cpb null mutant. Three of the re-expressed gene products (CPB1, CPB2.8, and CPB18) showed activity toward gelatin and fluorogenic substrates, but the enzymes differed both in their relative activities toward the different substrates and also their physical characteristics as displayed by the mobilities on gels. These differences correlated with evidence from biochemical studies which showed that isoenzymes have distinct properties (15,16). CPB1 has a lower molecular mass than either CPB2.8 or CPB18, as predicted from the gene sequences. The predicted size of the cpb1 gene product for the mature proteinase, including the 15-amino acid CTE, is 24.9 kDa (Table I), which correlates well with the Western blot data (Fig. 5B). The predicted size of the cpb2.8 and cpb18 gene products for the mature proteinase, including the 100-amino acid CTE, is 34 kDa, considerably larger than the 28-kDa protein detected by Western blotting (Fig. 5B). It is likely, therefore, that more than half of the CTE is processed to give the mature CPB2.8 and CPB18 cysteine proteinases. The finding that the antibody detected essentially a single mature cysteine proteinase (see below for discussion on the larger molecular mass forms) suggests that the CTE is processed at a single site to give a homogeneous isoenzyme encoded by each individual gene. Therefore, the cysteine proteinase isoenzymes that occur in wild type L. mexicana are unlikely to be caused by multiple processing of CTEs from one gene product but probably reflect the polymorphisms in the DNA sequence which introduce amino acid changes.
CPB1 co-migrates on gelatin gels with one of the two major activities characteristically present in stationary phase (metacyclic) promastigotes of wild type cells (Fig. 5A). The protein also co-migrates on Western blots with a protein in the stationary phase wild type promastigotes (Fig. 5B). These results correlate with the Northern blot data demonstrating that levels of mRNA transcribed from cpb1 and/or cpb2 are elevated in stationary phase promastigotes relative to multiplicative promastigotes or amastigotes (Fig. 4C). Taken together, these results suggest that some of the proteinase activity in wild type metacyclic promastigotes is encoded by cpb1. The differences observed between the relative activities of wild type parasites and CPB1 expressed in the null mutant toward the two fluorogenic substrates (Fig. 6, lanes 1 and 2) suggest, however, that the slower mobility activity in wild type cells may comprise more than one gene product. It has been shown previously that some activity bands detected using gelatin gels do indeed represent several isoenzymes (15). CPB2 is likely to account for some of this activity, although the detection of cpb mRNA in metacyclic promastigotes using the cpb3-cpb18-specific probe (Fig. 4D) indicates that there is also a low level of expression of one or more of these genes in this life cycle stage.
Overall, the data show that expression of the first two genes of the array, which are characterized by the lack of sequence encoding a full-length CTE, is stage-regulated and occurs primarily in the infective metacyclic promastigote. The stagespecific expression of cpb1 and cpb2 suggests that they play some important role at this stage of the life cycle. This could involve the interaction of the metacyclic promastigote with its sandfly vector or the process of differentiation itself, but it is attractive to consider that it may be important for the early events associated with invasion of a macrophage following inoculation into a mammal. The finding that the null mutant cell line expressing CPB1 was less able than the wild type parasites to survive in macrophages shows that CPB1 is not able to restore the infectivity lost due to the deletion of the cpb array under these circumstances, and in this way it differs from CPB2.8. This inability to restore infectivity is not due to the lack of a CTE as CPB18, although apparently expressed in amastigotes naturally, also is unable to restore infectivity to the null mutant. This indicates that the individual genes of the array indeed perform different functions, a suggestion supported by the observed differences in substrate specificities. The precise timing of expression of the different isoenzymes may be an important factor for parasite invasion and survival in the macrophage.
The expression of the msp genes of L. chagasi has some similarities with the expression of cpb genes of L. mexicana. The msp genes encode a metalloproteinase known as gp63. It is the major surface proteinase of Leishmania and in L. chagasi is encoded by 18 msp genes arranged in a tandem array. These genes have been grouped into three classes based on their differential gene expression (45,46). RNAs from the multiple mspL genes are present predominantly in multiplicative promastigotes, whereas RNAs from the five mspS genes are present predominantly in stationary phase promastigotes (46). The regulation of the msp genes occurs at a post-transcriptional level, possibly via sequences located in the 3Ј-untranslated region of the mRNAs. This may also be the case for the cpb genes since there are differences in the noncoding DNA sequence between the cpb1 and cpb2.8 genes 3 which may be important for the developmentally regulated gene expression observed. Heterogeneity observed for the msp gene products on gels has been attributed to amino acid variation between different copies of the array (47), but these have not been correlated with differences in enzymatic activity of the metalloproteinase or function in different life cycle stages. The finding that cpb isogenes encode enzymes differing both in amino acid sequences and substrate specificity suggests that this may also be the case for the msp genes.
The null mutants re-expressing CPB2.8 or CPB18 exhibit a number of lower mobility bands in gelatin gels (30)(31)(32)(33)(34)(35)(36)(37)(38)(39)(40) and also larger molecular mass species in Western blots (about 38 kDa) (Fig. 5). By contrast, similar larger molecular mass proteins were not readily seen with the null mutant line reexpressing CPB1, which exhibited just the fully processed protein, or with wild type parasites unless a much larger amount of lysate was applied. These slower mobility activities in the 3 D. R. Brooks, G. H. Coombs, and J. C. Mottram, unpublished data. mutants re-expressing CPB2.8 or CPB18 may be precursors of the mature cysteine proteinase which are activated in situ following gel electrophoresis. Noncovalent complexes between lysosomal cathepsin B and its pro-peptide which are formed during autocatalytic maturation of the precursor have been described (48,49). These complexes are believed to keep the proteinase in an inactive state until it is secreted, when it is activated by local acidification. It is possible that either the CPB pro-region or CTE (or both) could be acting as an inhibitory peptide in a similar fashion, and the processing of the gel subsequent to electrophoresis provides conditions in which the inhibitory peptide is removed. The finding that CPB1 lacks a full CTE and does not appear to produce these putative precursor forms that are activated in situ provides some evidence that it may be this domain that is important with regard to any such processing events. Indeed, processing of the CTE may be a prerequisite for correct and efficient processing of the pro-domain.
Previous speculations on the function of the CTE have included a role in the targeting of the protein to lysosomes, a theory strengthened by the finding that a trypanosome cysteine proteinase apparently lacks mannose 6-phosphate, which therefore cannot play a role in the enzyme's trafficking (50). The finding using immunoelectron microscopy that the mature CPB1 enzyme, which has a truncated CTE, is located in lysosomes that are typical of the amastigotes of L. mexicana shows that a full-length CTE is not essential for successful intracellular trafficking. It is clearly also not essential for activity of these cysteine proteinases, as CPB1 is highly active toward gelatin (Fig. 5A) and other substrates (Fig. 6). This is consistent with previous reports that demonstrated that recombinant cysteine proteinases from T. cruzi and T. brucei lacking the CTE can be expressed as active enzymes (51,52). All of the CTEs, irrespective of size, have a serine/threonine-rich region bounded by proline residues (Fig. 3). It is not known whether this region has any role in targeting or processing of the enzyme, but this deserves study.
When the cpb cDNA was expressed in the null mutant cell line ⌬cpb it produced a protein of the predicted molecular size, indicating correct processing, but it was inactive toward gelatin under the standard conditions tested (27). This contrasts with the results reported here for the cpb1, cpb2.8, and cpb18 genes, which are expressed in the null mutant to produce active cysteine proteinases. The cpb cDNA encodes a protein that differs within the mature domain from CPB2.8 in only three amino acids (N60D; D61N, and D64S; see Fig. 3). A further six amino acid differences are present in the CTE. As more than half of the CTE would appear to be cleaved off to produce the mature protein, it is difficult to predict the effect that these amino acid variations in the CTE could have on the enzyme activity. The differences in the mature regions between CPB1, CPB2.8, and CPB18 must also explain the differences in relative activity toward the fluorogenic substrates observed in this study. More detailed studies on the substrate specificities of the individual isoenzymes are required, but it is already clear that the ability to express individual cpb genes in the null mutant (and also genes modified via site-directed mutagenesis) and subsequently analyze the expressed cysteine proteinase biochemically and functionally will provide a powerful approach for investigating the function of these proteinases.