Expression of a mitochondrial stress protein in the protozoan parasite Leishmania major

The protozoan parasite Leishmania major (causative agent of human cutaneous leishmaniasis) exists as an extracellular, flagellated form (the promastigote) in its sandfly vector, and as an intracellular, aflagellated form (the amastigote) within macrophage phagolysosomes in its mammalian host (Molyneux and Killick-Kendrick, 1987). The temperature and environmental changes associated with transfer of the parasite from vector to host have initiated studies on the stress proteins of Leishmania and their potential role in parasite infectivity and survival (Hunter et al., 1984; Lawrence and Robert-Gero, 1985; Van der Ploeg et al., 1985; Shapira et al., 1988; Searle et al., 1989; Smith et al., 1989).

Eukaryotic stress proteins are a group of molecules that are synthesised in response to a variety of stimuli in all types of cell (reviewed by Lindquist, 1986; Lindquist and Craig, 1988; Morimoto et al., 1990). These stresses include heat shock, metabolic disruption, nutrient deprivation, the presence of oxygen radicals and viral infection. Members of this group include a family of ATP-binding proteins related to heat shock protein 70 (hsp70), a molecule first identified as one of a small number of proteins synthesised in Drosophila cells in response to a rapid temperature increase (Tissieres et al., 1974). The genes encoding hsp70related proteins show both constitutive and inducible expression to generate 70 kDa molecules that are amongst the most highly conserved proteins in evolution. They function biologically in the assembly and disassembly of protein complexes and are also involved in protein translocation across intracellular membranes (Pelham, 1986; Morimoto et al., 1990). Proteins of this type, that are involved in the modulation of precursor protein conformation and aggregation, have been termed molecular “chaperones” (Ellis, 1987). One of the best characterised is the mitochondrial protein, mhsp70, from Saccharomyces cerevisiae that is encoded by the nuclear SSC1 gene, synthesised with a typical N-terminal targeting signal and imported into the mitochondrion (Craig et al., 1987, 1989; Scherer et al., 1990).

Proteins with similar functions are also found in prokaryotes. Of those characterised, the dnaK gene of E. coli is probably the best defined. It encodes a bacterial hsp70related protein that is 48% identical with eukaryotic homologues and has a role in protein export and folding (Cegielska and Georgopoulos, 1989; Georgopoulos et al., 1990).

In Leishmania, a number of stress proteins have been identified (Hunter et al., 1984; Lawrence and Robert-Gero, 1985; Shapira et al., 1988) and their genes characterised (Lee et al., 1988; Searle et al., 1989; Smith et al., 1989; Macfarlane and Kelly, 1990). In Leishmania major, at least two low copy number sequences with homology to eukaryotic hsp70 genes have been identified on different chromosomes (Searle et al., 1989), in addition to the tandemlyrepeated hsp70 gene family (Lee et al., 1988) that maps to a single chromosome of approx. 1 Mb (presumably homologous to chromosome 17 of the L. major karyotype; Samaras and Spithill, 1987). The functions of the proteins encoded by these genes are unknown, but the immunodominant character of some parasite hsp70-related proteins has provoked speculation as to the role of these molecules in host infection (Newport et al., 1988; Young et al., 1990).

In this paper, we describe the characterisation and expression of one of these Leishmania genes related to hsp70, Lmhsp 70.1 (Searle et al., 1989). The protein product bears an N-terminal mitochondrial leader peptide that is presumably cleaved subsequent to localisation within the parasite mitochondrion. In Leishmania and other trypanosomatid organisms, the mitochondrial membranes are continuous with the kinetoplast, a DNA-containing organelle characteristic of these species (Molyneux and Killick-Kendrick, 1987). The kinetoplast DNA is composed of two types of molecule: the protein-coding maxicircles (encoding typical mitochondrial proteins) and the guide RNA-coding minicircles (Simpson, 1987; Sturm and Simpson, 1990). The Lmhsp 70.1 gene product, mp70.1, is distributed throughout the mitochondrion of both promastigotes and amastigotes under normal growth conditions.

Parasites of the Friedlin line of L. major were maintained as previously described (Searle et al., 1989). In all heat shock experiments, promastigotes were incubated at 37°C for 2 h before use. The amastigotes used in immunofluorescence experiments were the gift of J.M. Blackwell, University of Cambridge.

The bacteriophage lambda recombinant, Lmhsp 70.1, was isolated from a bacteriophage lambda EMBL 4 genomic library, characterised and subcloned as previously described (Searle et al., 1989; Smith et al., 1989). DNA fragments from the subcloned region containing the Lmhsp 70.1 gene were propagated in the M13 vectors, tg130/131 (Amersham) and sequenced, using both specific and universal oligonucleotide primers (Searle et al., 1989; Coulson and Smith, 1990). DNA sequences were compiled and analysed using the Microgenie (Beckman) computer program. The sequence presented in Fig. 1 has been submitted to the EMBL data base and assigned the accession number X64137.

The 2.7 kb HindIII/BamHI and the 0.8 kb HaeIII fragments of pLm 70.1 (subcloned from Lmhsp 70.1 by Searle et al., 1989), were subcloned into the glutathione-S-transferase expression vector pGEX 1 to yield the constructs pGEX70.1 and pGEX70.1A (Fig. 1). Sub-cloning was achieved by blunt-end ligation into the SmaI site (pGEX70.1A), or by ligation of a BamHI-linkered, endfilled fragment into the BamHI site (pGEX70.1). These constructs were expressed using standard methods (Smith and Johnson, 1988) and the fusion proteins used to raise rabbit antibodies (Harlow and Lane, 1988). The antibody fractions were named pX70.1 and pX70.1A, respectively. After four boosts, samples of rabbit sera were passed through glutathione-S-transferase affinity columns (Harlow and Lane, 1988). The flow-throughs were tested for their negative cross-reactivity to glutathioneS-transferase before use in further experiments.

Promastigotes were collected by centrifugation (10 min at 4000 revs/min at 4°C), washed three times in phosphate-buffered saline (PBS) and stored at −70°C in small samples of PBS containing 1 mM EDTA, 2 mM phenylmethylsulphonylfluoride (PMSF), 0.3 μM aprotonin and 0.1 mM N-tosyl-L-phenylalanine chloromethylketone (TPCK).

Two-dimensional electrophoresis of proteins was carried out as described by Maizels et al. (1991). In brief, promastigotes were solubilised in an equal volume of lysis buffer containing 9.5 M urea, 2% Nonidet P-40, 2% Ampholines pH range 3-10, 0.78% dithiothreitol. Routinely, proteins from 1 × 10⁷ promastigotes were loaded per isoelectrofocusing gel and electrophoresed for 16 h at 400 V. The gels were extruded into 5 ml of equilibration buffer (containing 10% glycerol, 5% SDS, 62.5 mM Tris, pH 6.8, 5% 2-mercaptoethanol) prior to loading and electrophoresis through the second dimension.

Immunoblotting from SDS-polyacrylamide gels was carried out as described by Harlow and Lane (1988). The second antibody used in all experiments was peroxidase-conjugated, goat anti-rabbit H + L chain (Jackson Labs). Antibodies against total stationary phase promastigote products (pSTAT1) were raised in a rabbit injected with a mixture of proteins isolated from a 10-day culture of L. major, containing 85% metacyclic (infective) promastigotes.

Promastigotes and amastigotes were fixed and stained for DNA content and immunofluorescence as follows: cells were spread on to acid-washed slides, methanol-fixed and preincubated at room temperature for 45 min in PBS containing 10% fetal calf serum (FCS) and 0.1% Triton X-100. Subsequent incubations were carried out in a humidified chamber with antibodies diluted in PBS containing 3% FCS and 0.1% Triton X-100. The slides were incubated with primary antibody overnight at 4°C and washed four times in PBS containing 0.1% Triton X-100. The second antibody, fluoresceinated goat anti-rabbit H + L chain (Jackson Labs) diluted 1:100, was incubated with the slides for 4 h at room temperature. The slides were then washed twice in PBS containing 0.1% Triton X-100 and four times in PBS alone. The cells were stained for DNA content with Hoechst 33258 (Sigma), a fluorescent DNA intercalating agent that binds preferentially to A+T-rich DNA. After three further PBS washes, the samples were mounted in 85% glycerol, 2.5% n-propylgallate, prior to examination and photography.

Promastigotes were fixed in PBS containing 2% paraformaldehyde/0.1% glutaraldehyde, stained in uranyl acetate and embedded in LR white resin before sectioning. Antigens were localised in rehydrated sections with the antibodies pX70.1 and pX70.1A, diluted 1:20 in PBS/1% bovine serum albumin/0.01% Tween 20 (PBS/BSA/T), followed by 15 nm colloidal gold-conjugated, goat anti-rabbit IgG H + L chain (BioCell) diluted 1:300 in PBS/BSA/T. Control sections were incubated in PBS/BSA/T minus primary antibody. To improve contrast, sections were postosmicated in 1% osmium tetroxide, prior to restaining in uranyl acetate and Reynolds’ lead citrate solution.

The bacteriophage lambda recombinant Lmhsp 70.1 was originally isolated as one of several molecules that hybridised to a human hsp70 probe in a genomic DNA library screen (Searle et al., 1989; Smith et al., 1989). This recombinant encoded a putative single copy gene that shared sequence identity with other eukaryotic hsp70 genes but also carried a 5′ extension sequence with the predicted coding characteristics of a mitochondrial signal peptide (Searle et al., 1989; see Fig. 1B). These data suggested that the Lm 70.1 gene product was a mitochondrial protein in Leishmania.

The complete sequence of the Lmhsp 70.1 open reading frame (ORF) has been determined and is shown in Fig. 1. The physical location of the open reading frame (ORF) for translation is shown in Fig. 1A and the DNA sequence and amino acid assignment in Fig. 1B. Analysis of the Lmhsp 70.1 DNA sequence revealed a predicted ORF of 1905 bp, encoding a protein of 635 amino acids with a molecular mass of 68 kDa. This protein is named mp70.1 hereafter. Sequence analysis of the transcribed region of Lm 70.1 upstream of the ORF in Fig. 1B revealed no alternative inframe methionine codon for the initiation of translation (Searle, 1992).

The amino acid composition of mp70.1 included 14% acidic and 11% basic residues, with an overall hydrophobicity of 28%. At the N terminus, there was an extension of 23 amino acids, not present in other related “hsp70-like” proteins, with the exception of mtp70, the mitochondrial 70 kDa protein from Trypanosoma cruzi (Engman et al., 1989; see Fig. 2) and mhsp70 from S. cerevisiae (Craig et al., 1989). This putative signal sequence contained 13% hydroxylated amino acids, only 1 acidic residue, and 3 charged arginines. By comparison with other mitochondrial signal sequences, that are punctuated with basic amino acids every 4-to-8 residues and are capable of forming amphiphilic alpha helices (Douglas et al., 1986), the distribution of the charged amino acids in the signal sequence of mp70.1 make it unlikely that any helix would display a sideness of charge. Thus, while the putative mitochondrial signal sequence of mp70.1 possesses some of the features common to a number of mitochondrial signal sequences that have been implicated in the process of targeting and transport across membranes (Hartl et al., 1989), the distribution of the amino acids could indicate fundamental differences in the mechanisms of mitochondrial/kinetoplast import in Leishmania major.

The predicted ORF in Fig. 1 also shared sequence identity with the predicted ATPand calmodulin-binding domains of other hsp70 proteins (Fig. 2, see below).

The deduced amino acid sequence of Lmhsp 70.1 was compared with the amino acid sequences of several other related proteins (Fig. 2). These include mtp70, the T. cruzi mitochondrial protein (Engman et al., 1989), the DnaK gene product of E. coli (Bardwell and Craig, 1984), the product of the tandemly-repeated, hsp70 gene of L. major (Lee et al., 1988) and a human, heat-inducible hsp70 protein (Hunt and Morimoto, 1985). A probe encoding this latter protein was used in the primary isolation of Lmhsp 70.1 (Searle et al., 1989).

Amino acid identity between these four proteins and mp70.1 was highest in the N-terminal third of the predicted ORFs, in which 80 out of 210 amino acids were shared between all five molecules (Fig. 2). This region spans the mitochondrial signal sequences of mp70.1 and mtp70 (that share 47% amino acid identity), and also includes some of the residues implicated in the mechanism of ATP binding, a feature of all hsp70 proteins so far characterised (Morimoto et al., 1990). These amino acids include a conserved aspartic acid residue at position 33 of mp70.1 (^* in Fig. 2) that has been suggested as a candidate for the catalytic proton acceptor in the ATPase mechanism of bovine hsc70 (Flaherty et al., 1990). Alternatively, mutagenesis experiments with the E. coli DnaK protein have implicated other candidate proton acceptor amino acids that are conserved at positions 196 (glutamic acid) and 199 (alanine) of mp70.1 (^* in Fig. 2), and a possible autophosphorylation site that is conserved at position 223 (threonine; Hightower, 1991).

All five molecules showed decreasing similarity towards their C termini, reflecting the general observation that hsp70 proteins are less well-conserved in these domains (Lindquist, 1986). The predicted 21-amino acid “calmodulin-binding domain” (between residues 276 and 296) is highly conserved between mp70.1 and mpt70 (with only 3 mismatches) and shows 47% identity with the region of mouse hsp70 shown to bind calmodulin in vitro (Stevenson and Calderwood, 1990). The mouse protein is identical over this amino acid stretch with the human hsp70 used for the comparison in Fig. 2. In mammalian cells, this domain is proposed to affect the overall regulation of hsp70 protein activity, by the reversible binding of calmodulin as a function of intracellular Ca²⁺ concentration.

The similarity between mp70.1 and the mitochondrial protein mtp70 from T. cruzi is greater than between mp70.1 and the other hsp70-related gene from L. major (Fig. 2). The sequence similarity extends throughout the predicted ORF to give 80% amino acid identity overall. This conservation between species suggests that the two proteins may have similar functions within the mitochondria of trypanosomatid organisms. By contrast, the other hsp70related gene from Leishmania (Lm70) is more divergent at the amino acid level (50% amino acid identity; Fig. 2), suggesting a fundamentally different role in the parasite life cycle.

The DNA sequence data presented in Figs 1 and 2 suggest that the Lmhsp 70.1 gene product is a mitochondrial hsp70related protein. To confirm this prediction, antibodies were required for the intracellular localisation of mp70.1 within the Leishmania cell. Given the similarities between the different parasite hsp70 genes, it was essential to chose genespecific regions for expression to generate antibodies restricted to this single gene product. From the sequence analyses described above, two regions were chosen for subcloning in the bacterial expression vector pGEX1N. Firstly, an internal HaeIII fragment encoding 260 amino acids (of which 67% are mp70.1-specific) was used to construct pGEX70.1A (Fig. 1A). Secondly, a large segment of the ORF containing all but the first 71 amino acids of the putative protein was subcloned to generate pGEX70.1 (Fig. 1A). Expression of this region should allow production of an antibody fraction able to recognise other related hsp70 proteins in Leishmania. The fusion proteins generated were used to raise antibodies in rabbits (see Materials and Methods).

The specificity of these antibodies was assessed on twodimensional immunoblots of total promastigote proteins isolated from a 5-day culture containing 25% metacyclic forms (Fig. 3). The pX70.1 antibody recognised a number of proteins that were assumed to include members of the hsp70-related protein family in Leishmania (Fig. 3B, −HS). These proteins ranged in size between 71 kDa and 27 kDa and included molecules of 65 kDa and 62 kDa at pI 6.7 (arrowed in Fig. 3B), that were also recognised by a polyclonal antiserum raised against total stationary phase promastigote proteins, pSTAT1 (Fig. 3A, −HS). In addition, pSTAT1 recognised a range of other proteins of different size and isoelectric point.

The pX70.1A antibody recognised a single protein in non-heat-shocked promastigotes (Fig. 3C, −HS) that migrated with the same molecular mass and isoelectric point (65 kDa, pI 6.7) as one of the proteins recognised by pX70.1 and pSTAT1. This molecule was 3 kDa smaller than the 68 kDa protein predicted from the ORF analysis of Lmhsp 70.1; this size is consistent, however, with that of a protein generated by cleavage of the putative signal sequence after entry into the mitochondrion.

No significant difference in the level of the 65 kDa protein was observed when antibody pX70.1A was used on an immunoblot of heat-shocked promastigote proteins (Fig. 3C, +HS), a result that correlated with the observed lack of inducibility of the Lmhsp 70.1 gene transcript following a 37°C heat stress (Searle et al., 1989). However, an additional protein of 62 kDa, pI 6.7, was recognised at a low level by this antiserum in heat-shocked samples only (Fig. 3C, +HS). If this protein is the same molecule as that recognised by the antibody pX70.1 in the absence of heat shock (Fig. 3B), then antibody pX70.1A must recognise an epitope on the 62 kDa protein that changes conformation after stress.

The proteins recognised by the pSTAT1 antibody after heat shock showed few quantitative or qualitative differences; the only detectable changes were in the levels of two proteins of 50 kDa and 34 kDa (Fig. 3A, +HS). At this temperature, however, a proportion of the Leishmania cells did become rounded in morphology and this change correlated with a redistribution of existing protein, recognised by pSTAT1, to a nuclear location (see below).

The mp70.1 protein was localised in promastigotes and amastigotes of L. major using the antibodies pX70.1A and pX70.1, in both indirect immunofluorescence and immunogold labelling assays. Some members of the hsp70 protein family migrate to the nucleus upon heat shock (Velasquez et al., 1980; Velasquez and Lindquist, 1984), so localisation experiments were also carried out on heat-shocked promastigotes. The results obtained are shown in Fig. 4. For each sample, the intracellular location of the kinetoplast and nucleus was established by staining with Hoechst dye, which binds preferentially to A+T-rich DNA (Fig. 4E-H, L-N). Promastigotes probed with the antibodies pX70.1A(A) and pX70.1 (B) showed strong fluorescence over both the kinetoplast and a trailing, sub-cellular organelle extending away from the kinetoplast body. This corresponds to the main body of the mitochondrion that lies along the length of the parasite cell (see Fig. 5). Heat-shocked pro-mastigotes also showed strong kinetoplast staining but in these organisms, low level staining was also visible over the whole cell with the exception of the nucleus (Fig. 4C). Although pX70.1 was used in this experiment, the same result was also obtained with pX70.1A (data not shown). The absence of cytoplasmic staining in non-stressed cells probed with pX70.1 made it unlikely that this was caused by the 62 kDa protein. Alternatively, some mp70.1 might relocalise from the mitochondrion to the cytoplasm following heat shock.

In spite of this phenomenon, the predominant molecule recognised under heat-shock conditions was mitochondrial in location. The absence of trailing fluorescence in Fig. 4C implies a possible relocalisation of the mp70.1 protein within the mitochondrion after heat shock. Alternatively, this organelle may shorten in response to stress, producing the morphology observed in Fig. 4C; the reverse process is observed during the transformation of amastigotes to promastigotes, when the mitochondrion lengthens and elaborates (Bray, 1974). In stressed mammalian cells, collapse of the intermediate filaments (which may anchor sub-cellular organelles within the cytoplasm) leads to relocalisation of the mitochondria to a perinuclear location (Welch, 1990).

Indirect immunofluorescence experiments with L. major amastigotes showed that mp70.1 also localises to the kine-toplast in these mammalian parasite forms (Fig. 4D). This expression pattern correlates with the observed presence of 70.1-specific transcripts in all stages of the Leishmania life cycle (Searle et al., 1989; Smith et al., 1989).

To distinguish between the mitochondrial staining observed with pX70.1 and pX70.1A and the pattern generated by other stress-inducible proteins, pSTAT1 (the total stationary phase promastigote protein antibody) was also used in indirect immunofluorescence assays (Fig. 4I-K). From the results shown in Fig. 3A, it was evident that this antibody fraction recognised a number of proteins in both stressed and non-stressed cells. When the localisation of these proteins was examined, a major fraction of the molecules relocalised to the promastigote nucleus following a heat shock at 37°C for 2 h (Fig. 4I,J). This redistribution correlated with a more rounded promastigote morphology (Fig. 4J). In amastigotes, the staining pattern was diffuse throughout the cytoplasm and showed no predominant nuclear staining (Fig. 4K).

These results confirm that Leishmania promastigotes possess one or more proteins that migrate to the nucleus following a moderate heat shock. These molecules are recognised by pSTAT1, but their abundance does not vary detectably in total protein extracts. mp70.1 is not one of these proteins but remains predominantly within the mitochondrion during the life cycle of the parasite, whether or not the cell undergoes a stress response.

To confirm the mitochondrial location of mp70.1, the antibodies, pX70.1 and pX70.1A, were conjugated with colloidal gold prior to promastigote labelling and examination by electron microscopy. Staining within the mitochondrion was evident with both antibodies (Fig. 5A-C), whereas no scattered particles of colloidal gold were evident in control sections probed with pre-immune serum (Fig. 5D). The mitochondrial staining was extensive throughout this organelle, including the region containing the kinetoplast, and clearly correlated with the trailing immunofluorescence seen in Fig. 4 (Fig. 5A). In some specimens, staining was observed specifically associated with the kinetoplast (k) DNA (Fig. 5C) but a more common distribution of gold particles included both the kDNA and the mitochondrial membranes (Fig. 5B). In no case was nuclear staining evident.

This report describes the characterisation of a mitochondrial hsp70-related protein, mp70.1, from Leishmania major. The predicted amino acid sequence is related to other hsp70-like proteins from both prokaryotes and eukaryotes, and is conserved in those amino acids implicated in ATPand calmodulin-binding (Fig. 2). This conservation predicts that the Leishmania protein is also an ATP-binding protein and may be regulated by calmodulin.

Of the open reading frames compared, mp70.1 is more closely related to mtp70, a mitochondrial hsp70 of T. cruzi, than to other non-mitochondrial, stress-related proteins of L. major. Both have N-terminal 23 amino acid sequences with structural features in common with other mitochondrial targeting peptides. The immunoblotting results presented in Fig. 3 are consistent with the cleavage of mp70.1 from a 68 kDa preprotein to a 65 kDa molecule by removal of a 3 kDa signal sequence.

Initial sequence data suggested that mp70.1 was the Leishmania homologue of mtp70 from T. cruzi (Engman et al., 1989; Searle et al., 1989) and that both molecules might therefore have comparable functions within the same subcellular compartment in trypanosomatid species. The T. cruzi protein has been localised only to the region of the kinetoplast associated with the kDNA, however, whereas mp70.1 is found throughout the mitochondrion (Figs 4, 5). These observations suggest that at least two, highly-related mitochondrial proteins may be present in trypanosomatid organisms. This interpretation is supported by DNA blotting and hybridisation data: the Leishmania mp70.1 gene has sequence similarity with, but is physically distinct from, the T. cruzi mtp70 gene (Searle, 1992).

Most proteins are imported into mitochondria as precursor molecules that include a positively charged N-terminal signal sequence for correct targeting (Douglas et al., 1986; Van Loon et al., 1986; Glick and Schatz, 1991). This alpha helical peptide is thought to facilitate insertion of the protein into the outer mitochondrial membrane, while the inner membrane exerts an electrophoretic effect on the positively charged sequence (Horwich et al., 1986; Von Heijne, 1986; Vasarotti et al., 1987; Hartl et al., 1989). In yeast, the mitochondrial ATP-binding hsp70 (mhsp70) is involved in the import of precursor proteins into these organelles, the energy for membrane translocation being provided by ATP hydrolysis (Scherer et al., 1990). Transportation may require the binding of mhsp70 to the precursor polypeptide as it emerges into the mitochondrial matrix as the driving force for this reaction. In addition, the protein may have a role in facilitating the unfolding of the precursor on the cytosolic side of the mitochondrial membrane (Kang et al., 1990).

The intracellular distribution of some of the proteins recognised by the antibody pSTAT1 exhibited a classic heat-shock response (Fig. 4). These proteins migrated to the nucleus after stress, where it is known that hsp70 proteins bind to partially assembled ribosomes. By contrast, the localisation of the hsp70-related protein mp70.1, as determined by indirect immunofluorescence and immuno-gold electron microscopy, remained predominantly mitochondrial, even after a period of heat stress. This staining pattern was observed in both promastigotes and amastigotes, suggesting a fundamental role for the protein within this organelle. A stress-induced change in morphology of the parasite mitochondrion is the most likely explanation for the observed alteration in fluorescent staining after stress (Fig. 4). These observations are consistent with our previous data demonstrating constitutive transcription of the Lmhsp 70 gene and steady state RNA levels unaffected by heat shock (Searle et al., 1989; Smith et al., 1989).

The specific role of the Leishmania mp70.1 protein has not yet been established. By analogy with other mitochondrial hsp70-related proteins, however, we suggest that this protein may function as a molecular chaperone for the transport of proteins into the parasite mitochondrion. Genetic transformation experiments will verify this prediction.

We thank David Engman for the T. cruzi mtp70 antibodies and gene probe, Will Whitfield for advice on protein purification and antibody production, Tano Gonzalez for assistance with immunofluorescence experiments, Mark Blaxter for demonstration of twodimensional immunoblotting, Jennie Blackwell for provision of amastigotes, Bernadette Connolly for discussion and critical reading of this manuscript. This work was supported by the Medical Research Council and the Wellcome Trust.