A role for cathepsin E in the processing of mast-cell carboxypeptidase A

Mast cells are well known for their harmful effects during inflammatory conditions such as asthma and allergy, but they are also important for our innate defence against bacterial infections and parasites (Metcalfe et al., 1997; Stevens and Austen, 1989; Wedemeyer et al., 2000). During these conditions, the mast cell is activated (e.g. by antigen-mediated cross-linking of IgE molecules bound to their high-affinity receptor FcϵRI) and thereby undergoes degranulation. Upon degranulation, various preformed inflammatory mediators are released, including histamine, cytokines, proteoglycans and a range of proteases [tryptases, chymases and carboxypeptidase A (CPA)].

Previous studies aiming at understanding the function of the mast cell proteases have mainly focused on tryptases and chymases, whereas relatively little is known concerning the biology of CPA. Mast-cell CPA is a zinc-containing metalloprotease with exopeptidase activity (Everitt and Neurath, 1980) and is synthesized as a proenzyme with a 94-amino-acid activation peptide (Reynolds et al., 1989). The biological function of CPA is uncertain but previous studies have implicated CPA in the generation and degradation of angiotensin II (Lundequist et al., 2004), and the degradation of apolipoprotein B (Kokkonen et al., 1986), and it has been suggested that CPA could be involved in the metabolism of leukotrienes (Reddanna et al., 2003). CPA, like mast-cell tryptases and chymases, is stored and released in fully active form (i.e. with its propeptide removed). The mechanisms for processing of the pro forms of the respective mast-cell proteases into active enzymes have been studied to some extent. It was shown recently that mice lacking dipeptidyl peptidase I (DPPI; also known as cathepsin C) fail to process the pro forms of the chymases mouse mast-cell protease 4 (mMCP-4) and mMCP-5, demonstrating an essential role for DPPI in the generation of active chymases (Wolters et al., 2001). In the same study, it was observed that the levels of active tryptase mMCP-6 was reduced in DPPI^–/– mast cells. However, active mMCP-6 was clearly detectable also in DPPI^–/– cells, indicating that DPPI has a role but is not essential for the processing of protryptase into active protease (Wolters et al., 2001). A role for DPPI has also been suggested in the processing of human β-tryptase (Sakai et al., 1996).

Unlike tryptase and chymase, the activation mechanism for CPA has been poorly studied. However, it has been shown that the processing of pro-CPA to the active enzyme occurs in the mast-cell secretory granule (Rath-Wolfson, 2001; Springman et al., 1995) and studies using protease inhibitors have implicated cysteine-protease activity in the process (Springman et al., 1995). The cysteine protease family includes several lysosomal cathepsins that are widely distributed among living organisms (for a review, see Turk et al., 2000) – cathepsins B, C, F, H, K, L, O, S, V, W and X. In addition to cysteine proteases, the cathepsin family of proteases includes two aspartic proteases, cathepsins D and E. The cathepsins require a slightly acidic environment for optimal activity, a condition that is prevalent in lysosomes but also in other cellular compartments such as the secretory granule of mast cells. Traditionally, cathepsins are believed to carry out unspecific bulk proteolysis inside the lysosome (Barrett, 1992) but there is growing evidence for specific, non-redundant in vivo functions of lysosomal cathepsins (Chapman et al., 1997; Turk et al., 2000), and several knockout mouse strains have provided important insight into their biological functions (Guicciardi et al., 2000; Halangk et al., 2000; Nakagawa et al., 1998; Nakagawa et al., 1999; Potts et al., 2004; Reinheckel et al., 2001; Roth et al., 2000; Saftig et al., 1998; Wolters et al., 2001).

Because it was shown that DPPI is involved in the processing of chymases and tryptases (see above), and because the processing of pro-CPA was previously found to be dependent on cysteine-protease activity (Springman et al., 1995), we investigated the role of DPPI in the processing of CPA (Henningsson et al., 2003). However, we found that DPPI was not involved in the processing of pro-CPA into active protease. In fact, the levels of active CPA in bone-marrow-derived mast cells (BMMCs) from DPPI-deficient mice were higher than in wild-type cells, and the same tendency was seen in BMMCs from mice deficient in cathepsin S. The most likely explanation for these findings is that CPA is degraded at a higher rate in normal BMMCs than in the DPPI-null and cathepsin-S-null cells.

The aim of the present study was to gain further insight into the processing mechanisms for pro-CPA. The strategy was to take advantage of existing knockout models in which various cathepsins have been targeted. We report here that mast cells deficient in cathepsin E display defective processing of pro-CPA into active enzyme.

Mice deficient in cathepsins B (Halangk et al., 2000), L (Roth et al., 2000), D (Saftig et al., 1995) and E (Tsukuba et al., 2003), and N-deacetylase/N-sulfotransferase-2 (NDST-2) (Forsberg et al., 1999; Humphries et al., 1999) were as described previously. All of the knockout strains were backcrossed into C57BL/6 background.

Peritoneal cells were obtained by peritoneal washing with 20 ml ice-cold PBS. Cells were counted and dissolved in SDS-PAGE sample buffer. Samples of these mixtures were subjected to SDS-PAGE on 12% polyacrylamide gels under reducing conditions. Proteins were subsequently blotted onto nitrocellulose membranes, followed by blocking with 5% milk powder in TBS containing 0.1% Tween 20 for 20 minutes to 1 hour at 20°C. Next, the membranes were incubated with rabbit antiserum towards CPA, mMCP-4, mMCP-5 or mMCP-6, diluted 1:200-1:1000 in 5% milk powder in TBS containing 0.1% Tween 20, at 4°C overnight. For analysis of cathepsin E, a goat anti-cathepsin-E antibody (R&D systems, Abingdon, UK) was used at a 1:500 dilution in 5% milk powder in TBS containing 0.1% Tween 20, at 4°C overnight. After washing extensively with TBS containing 0.1% Tween 20, the membranes were incubated with horseradish-peroxidase-conjugated donkey anti-rabbit secondary antibody (for CPA, mMCP-4, mMCP-5 and mMCP-6; Amersham Biosciences, Uppsala, Sweden) or horseradish-peroxidase-conjugated rabbit anti-goat secondary antibody (for cathepsin E; Sigma-Aldrich, Stockholm, Sweden) diluted 1:3000 in TBS containing 0.1% Tween 20. After 45 minutes of incubation at 20°C, the membranes were washed extensively with TBS containing 0.1% Tween 20 followed by washing with TBS without detergent. The membranes probed with anti-cathepsin-E antibody were further incubated with a third antibody, donkey anti-rabbit horseradish-peroxidase conjugate. The membranes were developed with the enhanced chemiluminescence (ECL) system (Amersham Biosciences, Uppsala, Sweden) according to the protocol provided by the manufacturer.

BMMCs were obtained as described (Henningsson et al., 2002) by culturing C57BL/6 mouse (female, 12-15 weeks old) femur and tibia bone-marrow cells in Dulbecco's modified Eagle's medium (DMEM; SVA, Uppsala, Sweden) supplemented with 10% heat-inactivated foetal calf serum (Gibco, Stockholm, Sweden), 50 μg ml^–1 gentamicin sulfate (SVA, Uppsala, Sweden), L-glutamine (SVA, Uppsala, Sweden) and 50% WEHI-3B conditioned medium (which contains interleukin 3) for 3 weeks. The cells were kept at a concentration of 500,000 cells ml^–1 and the cell-culture medium was changed every third day. To degranulate BMMCs, 1.2 ×10⁷ cells were pelleted by centrifugation at 300 g for 10 minutes and resuspended in 12 ml serum-free culture medium. 2 ml aliquots of the suspension were added to six culture wells. To three of the wells, 2 μl 2 mM solution (in H₂O) of the calcium ionophore A23187 (Sigma; 2 μM final concentration) were added and to the remaining wells an equal volume (2 μl) of H₂O was added. The cells were incubated at 37°C, 5% CO₂ for 10 minutes, 30 minutes or 120 minutes. Subsequently, the cells were pelleted (centrifugation at 300 g, 10 minutes) and solubilized in 100 μl SDS-PAGE sample buffer. The supernatants (conditioned medium) were concentrated approximately 50 times using Amicon Ultra-4 centrifugal filter devices (Millipore, Solna, Sweden) and thereafter mixed with 0.25 volumes 4 × SDS-PAGE sample buffer. Cell extracts and concentrated conditioned media were analysed for presence of cathepsin E and CPA protein by western blotting.

Total RNA was isolated from mouse BMMCs using the NucleoSpin RNA II kit (Macherey Nagel, Düren, Germany) and was used for first-strand cDNA synthesis using SuperScriptII for reverse transcription PCR (RT-PCR) using a CPA-specific reverse primer (5′-GTCCTCGAGCAAGGGCAATTCATTAGGAAGTATTCTTGA-3′). A PCR approach was used to produce a construct for expression of recombinant pro-CPA in a mammalian system. The 5′ primer used (5′-CACGAATTCCACCATCACCATCACCATGACGATGACGATAAGATTGCTCCTGTCCACTTTG-3′) in the PCR introduced a 6 × His tag, to facilitate purification. In addition, an enterokinase-susceptible peptide (Asp-Asp-Asp-Asp-Lys) was inserted between the His tag and the pro-CPA, replacing the natural activation peptide. The enterokinase site enables subsequent removal of the His residues. The PCR product was ligated into the pCEP-Pu2 vector as described (Hallgren et al., 2000) and transformed into competent Escherichia coli. Preparations of plasmid DNA were made using QIAprep Spin mini-prep (Qiagen, Stockholm, Sweden) and positive clones were confirmed by sequencing using Big Dye Sequencing kit (Applied Biosystems Prism, Foster City, CA) and sequenced in a 310 Genetic Analyzer (Applied Biosystems Prism, Foster City, CA).

The human embryonic-kidney-cell line 293-EBNA expresses the gene for the Epstein-Barr nuclear antigen 1 (EBNA-1) protein. This protein is a DNA binding protein that interacts with OriP, which is the Epstein-Barr virus (EBV) origin of replication. OriP is located in the pCEP-Pu2 vector and is required for a stable episomal maintenance of the vector in the 293-EBNA cell line. The pCEP-Pu2 vector containing the pro-CPA insert was transfected into 293-EBNA cells using the Lipofectamin2000 kit (Invitrogen). The cells were kept in DMEM (SVA, Uppsala, Sweden) supplemented with 10% heat-inactivated foetal calf serum (Gibco) and 2 mM L-glutamine. Gentamicin (50 μg ml^–1) was included as antibiotic and, for selection, puromycin (0.5-5.0 μg ml^–1; Sigma-Aldrich, Stockholm, Sweden) was used. Conditioned medium was harvested, centrifuged and stored at –20°C. For purification, conditioned medium (750 ml) was added to a 10 ml Poly-Prep column containing 300 μl Ni-NTA agarose (Qiagen, Stockholm, Sweden). After washing the column, the pro-CPA fusion protein was eluted with 100 mM imidazole in eight 300 μl fractions. The fractions were analysed by SDS-PAGE and subsequent Coomassie Brilliant Blue staining or anti-CPA westernblot analysis.

To activate recombinant cathepsin E (R&D Systems, Abingdon, UK), 0.15 μg recombinant protein was incubated in 100 μl 50 mM sodium acetate, 0.1 M NaOH (pH 3.5) at 20°C for 15 minutes. Thereafter, 1.5 ng or 15 ng activated cathepsin E were mixed with ∼20 ng recombinant pro-CPA and the pH was adjusted to 5.5 with 50 mM sodium acetate, 0.1 M NaOH (pH 5.5). In addition, recombinant pro-CPA was incubated with 300 ng or 2 μg recombinant cathepsin D (Sigma). The samples were incubated at 37°C for 30 minutes and thereafter mixed with SDS-PAGE sample buffer and subjected to SDS-PAGE on a 12% polyacrylamide gel. Subsequently, western-blot analysis for pro-CPA/CPA was performed, as described above.

Peritoneal cells were collected from NDST-2^–/– and NDST-2^+/– littermates. Cytospin slides were prepared and stained with a polyclonal goat anti-mouse-cathepsin-E antibody (R&D Systems). A biotinylated horse anti-goat-IgG antibody (affinity purified; Vector Laboratories, Burlingame, CA) was used as secondary antibody. Staining was performed by a standard protocol using the biotin/avidin-based Vectastain Elite (Vector Laboratories) and DAB (3,3′-diaminobenzidine) for detection of the secondary antibody. As a negative control, affinity-purified IgG purified from goat serum (SVA, Uppsala, Sweden) was used. To visualize mast cells, slides were counterstained with May-Grünwald/Giemsa. First, the slides were fixed in methanol for 3 minutes and thereafter stained with May-Grünwald (Merck, Sollentuna, Sweden) for 15 minutes. After washing with water, the slides were stained with 5% Giemsa (Merck) in H₂O for 10 minutes.

Affinity chromatography was performed on a 10 ml Poly-Prep column containing ∼100 μl heparin-Sepharose (Amersham Biosciences, Uppsala, Sweden). The column was equilibrated with PBS, 0.05 M NaCl, pH 6.0 or pH 7.4. Recombinant cathepsin E (2.3 μg) (R&D Systems) was loaded onto the column and left in the column for 15 minutes at 20°C. Thereafter, the recombinant protein was stepwise eluted with PBS, pH 6.0 or pH 7.4, with increasing concentrations of NaCl (0.05 M, 0.14 M, 0.3 M, 0.5 M, 1.0 M). For every elution step, four 100-μl fractions were collected and 30 μl of each fraction was analysed by SDS-PAGE using 12% polyacrylamide gels followed by anti-cathepsin-E western-blot analysis.

A homology model was constructed using the coordinates for human cathepsin E (PDB code 1TZS) (Ostermann et al., 2004). The sequences of mouse and human cathepsin E were aligned using ClustalW. The sequences are 82% identical, with no gaps or insertions in the alignment. A model for mouse cathepsin E was generated using the computer program SOD (Kleywegt et al., 2001) and the molecular graphics program O (Jones et al., 1991). Side-chain rotamers were selected to correspond as closely as possible to those in the experimental models except when a different rotamer had to be chosen to avoid steric clashes. Limited energy minimization was carried out with REFMAC5 (Murshudov, 1997). Surface representations were generated using PYMOL (http://www.pymol.org/).

A previous report implicated cysteine-protease activity in the processing of pro-CPA into active protease (Springman et al., 1995), and so we assessed whether either cathepsins B or L were involved in this processing. Peritoneal cells were recovered from mice deficient in cathepsins B and L, as well as from the corresponding wild-type or heterozygous controls. Cellular extracts were prepared and were subjected to westernblot analysis using an antibody that detects both the pro and active forms of CPA (Henningsson et al., 2002). In wild-type peritoneal cells, only the active form of CPA (∼35 kDa) was detected (Fig. 1A-D). Similarly, only the active form of CPA was detected in extracts from peritoneal cells deficient in either cathepsin B or cathepsin L, indicating that neither of these cathepsins is involved in the processing of pro-CPA (Fig. 1A,B). Furthermore, we have previously established that neither cathepsin C nor cathepsin S is involved in pro-CPA processing into active enzyme (Henningsson et al., 2003).

Cathepsins are traditionally thought to be located mainly in lysosomes. However, it was recently found that one cathepsin, the aspartic-protease cathepsin D, was located in the secretory granule of rat basophilic leukemia (RBL) cells. We therefore considered the possibility that cathepsin D or another aspartic cathepsin, cathepsin E, is present in peritoneal mast cells and participates in pro-CPA processing. Western-blot analysis of cell extracts from cathepsin-D-deficient peritoneal cells did not reveal any defects in pro-CPA processing (Fig. 1C). By striking contrast, peritoneal-cell extracts from cathepsin-E-deficient animals displayed readily detectable levels of pro-CPA (∼55 kDa), indicating defective pro-CPA processing with an accompanying accumulation of the proenzyme (Fig. 1D). This clearly suggests a role for cathepsin E in pro-CPA processing. However, active CPA was also detected, although at reduced levels, indicating that cathepsin E is not essential for the processing.

Next, the possibility that cathepsin E also influenced the processing of other mast-cell proteases was addressed. Peritoneal mast cells from C57BL/6 mice produce, in addition to CPA, the chymases mMCP-4 and mMCP-5 as well as the tryptase mMCP-6. However, western-blot analysis did not demonstrate any differences in the levels of these proteases between cathepsin-E^+/+ and cathepsin-E^–/– cells, nor was there any accumulation of their respective proforms (Fig. 2). Thus, cathepsin E appears to be specifically involved in the processing of pro-CPA, without affecting the processing of secretory granule chymases or tryptases.

The results above indicate that pro-CPA processing into active protease is dependent on cathepsin E. This dependence could be either direct (i.e. pro-CPA is a substrate for cathepsin E) or indirect. To investigate whether pro-CPA processing can actually be catalysed by cathepsin E, we expressed recombinant pro-CPA. To ensure proper folding and glycosylation conditions, we expressed it in a mammalian system. The recombinant protein was incubated with recombinant mouse cathepsin E at pH 5.5, the pH that prevails in mast-cell granules (De Young et al., 1987). Indeed, westernblot analysis showed that the recombinant cathepsin E processed recombinant ∼55 kDa pro-CPA into the ∼35 kDa active form (Fig. 3A). Extracts from BMMCs (derived from C57BL/6 mice) were included in the western blot as a marker for the pro and active forms of CPA; these cells always contain both the pro form and the active form of CPA (Fig. 3A, right) (Henningsson et al., 2002). In addition, the recombinant pro-CPA was incubated with recombinant cathepsin D in order to assess whether this aspartic protease could also catalyse the processing of pro-CPA into active enzyme. However, there was no clear increase in the level of active CPA after incubation of pro-CPA with cathepsin D (Fig. 3B).

The data presented above indicate that cathepsin E has a role in the processing of pro-CPA, implying its presence in mast cells. Further, the processing of mast-cell pro-CPA has previously been suggested to occur within the mast-cell secretory granule, suggesting that cathepsin E might actually be located in this cellular compartment. To clarify these issues, we stained peritoneal cells with an anti-cathepsin-E antibody. As seen in Fig. 4A, cathepsin E was readily detected in cells with a mast-cell-like morphology, indicating that cathepsin E is indeed present in mast cells. A control antibody did not produce any detectable staining in the peritoneal-cell population (not shown). Intense staining was observed in granular structures, supporting a location within the secretory granule. To further address the possibility that cathepsin E is located in the secretory granule, we used peritoneal cells taken from a mouse strain deficient in NDST-2. The knockout of NDST-2 results in impaired synthesis of heparin in mast cells, accompanied by a dramatic defect in formation of mature mast-cell secretory granule (Forsberg et al., 1999; Humphries et al., 1999). As shown in Fig. 4C, no staining for cathepsin E was seen in the NDST-2-null peritoneal cells, giving further support for cathepsin E being located in the heparin-containing secretory granule of peritoneal mast cells (Fig. 4C). If cathepsin E were indeed located in mast-cell secretory granule, mast-cell degranulation would be expected to result in release of cathepsin E together with other secretorygranule components. To test this possibility, BMMCs were activated by a mast-cell-degranulating agent, the calcium ionophore A23187. Conditioned media were collected after various times and analysed for the presence of cathepsin E and CPA. In the absence of added A23187, only low levels of cathepsin E and CPA were detected in the culture medium (Fig. 5). However, addition of A23187 resulted in time-dependent release of both cathepsin E and CPA into the cell culture medium. Importantly, the release of both components appeared to follow similar kinetics (Fig. 5). Thus, mast-cell degranulation is accompanied by the release of both CPA and cathepsin E, further supporting a location of cathepsin E within mast-cell secretory granules.

Peritoneal-cell populations normally contain 2-4% mast cells and these can be visualized using May-Grünwald/Giemsa staining, which stains the negatively charged proteoglycans in the mast-cell granules. As can be seen in Fig. 4B, NDST-2^+/– cells of similar morphology to those that were positive for cathepsin E also showed strong staining with May-Grünwald/Giemsa, further supporting that the cathepsin-E-positive cells were indeed mast cells (Fig. 4B). By contrast, the NDST-2^–/– peritoneal cells displayed a complete lack of staining with May-Grünwald/Giemsa, in agreement with a defective proteoglycan synthesis (Fig. 4D).

The results above indicate that cathepsin E is stored in mast-cell granules and that this storage is heparin dependent. Further experiments were therefore conducted to investigate the possible heparin-binding properties of cathepsin E. To this end, recombinant cathepsin E was subjected to affinity chromatography on heparin-Sepharose. When the affinity chromatography was performed at pH 6.0 (a similar pH to that found in the mast-cell secretory granule), cathepsin E bound with remarkably high affinity, with 0.5 M NaCl being required for elution from the column (Fig. 6A). By striking contrast, very weak binding to heparin-Sepharose was seen when the chromatography was performed at neutral pH, with most of the cathepsin E being eluted either in the flow-through fraction or at 0.05 M NaCl (Fig. 6B). Because native cathepsin E is a disulfide-linked dimer (Tatnell et al., 1997), we investigated the possibility that the dimerization contributes to the high affinity for heparin (e.g. through multivalent interactions). However, monomerization of the cathepsin-E dimer by reduction with dithiothreitol did not result in any reduction in the affinity for heparin (data not shown).

Mast-cell CPA is the most anonymous of the mast-cell proteases and there are relatively few reports describing its basic biological properties. It has been known for a considerable time that CPA is stored as an active fully processed form, with its 94-amino-acid propeptide removed (Dikov et al., 1994), but the mechanism(s) involved in this processing have been poorly characterized. In a previous study, it was shown that addition of the cysteine-protease inhibitor e-64d to the mast-cell line KiSV-MC14 caused an accumulation of pro-CPA, indicating that cysteine-protease activity might be involved in its processing (Springman et al., 1995). However, the identity of this cysteine protease was not revealed. In the present study, we assessed the role of two specific cysteine proteases, cathepsins B and L, in the processing of pro-CPA, but neither of these cysteine proteases appeared to have any significant role in pro-CPA processing. Furthermore, we have shown previously that neither of two other cysteine proteases [cathepsin C (DPPI) and cathepsin S] is involved in the processing of pro-CPA (Henningsson et al., 2003). These studies were thus unable to provide any further support for a major role for cysteine proteases in the processing of pro-CPA. Importantly, however, we cannot exclude an involvement of other cysteine proteases, such as cathepsins K, O, F, V and X.

In the present study, we instead present evidence that the aspartic protease cathepsin E is involved in the processing of pro-CPA. This notion is supported both by in vivo experiments demonstrating that pro-CPA processing was defective in mice lacking cathepsin E and by in vitro experiments in which purified cathepsin E was found to process recombinant pro-CPA. Interestingly, the knockout of another aspartic protease, cathepsin D, did not affect pro-CPA processing, indicating that the ability to process pro-CPA is not a general feature of aspartic proteases but rather a specific property of cathepsin E. It is important to realize that the lack of cathepsin E did not cause a complete inability of the mast cells to produce active CPA. This could imply that another, as yet unidentified, protease contributes to the pro-CPA processing in normal mast cells. Alternatively, the knockout of cathepsin E could unleash compensating mechanisms that partly rescue the inability to process pro-CPA into active enzyme in cathepsin-E-null cells.

Cathepsin E has previously been identified in various locations, such as lymphoid tissue, the gastrointestinal tract and urinary organs (Muto et al., 1988; Sakai et al., 1989), but this is to our knowledge the first report demonstrating the presence of immunoreactive cathepsin E in mast cells. The subcellular location of cathepsin E in the mast-cell is intriguing. Cathepsin E, like all cathepsins, is normally confined to lysosomes, but the immunohistochemical staining performed here clearly pointed to a location within the secretory granule of the mast cells. A granular location for cathepsin E was also strongly supported by the lack of cathepsin-E staining in peritoneal cells lacking NDST-2. Targeting of NDST-2 was previously shown to interfere specifically with the biosynthesis of mast-cell heparin, one of the compounds present specifically in mast-cell secretory granules, without affecting any other cell type (Forsberg et al., 1999; Humphries et al., 1999). As a consequence, NDST-2-deficient mast cells display an almost complete inability to assemble normal densely packed granules while not being affected as regards transcription of mast-cell-specific markers. Clearly, the lack of cathepsin-E staining in the NDST-2^–/– peritoneal cells is therefore in strong agreement with a predominant location of cathepsin E in the secretory granule of mast cells. This is also strongly supported by our finding that mast-cell degranulation is accompanied by the release of cathepsin E. In further support for the possibility that cathepsins localize to secretory compartments, previous studies have indicated that cathepsin D and cathepsin C might be present in the secretory granules of RBL cells (Dragonetti et al., 2000) and BMMCs (Wolters et al., 2001) respectively.

The effect of NDST-2 deficiency on cathepsin-E staining in mast cells indicates that cathepsin-E storage is dependent on heparin. This dependence could be either direct (cathepsin interacts with heparin proteoglycan) or indirect (e.g. that cathepsin is bound to a compound that in turn interacts with heparin). To address these possibilities, we therefore investigated whether cathepsin E interacts with heparin. Indeed, cathepsin E bound strongly to heparin-Sepharose, indicating that cathepsin-E storage in mast-cell granules requires that the cathepsin is associated with the proteoglycan. Interestingly, the interaction of cathepsin E with heparin showed a strong pH dependence, with strong binding at low pH and essentially complete loss of affinity for heparin at neutral pH. The reason for this strong pH dependence is not clear, but one possibility is that the high-affinity interaction to heparin involves His residues. His residues are deprotonated and thereby uncharged at neutral pH but are protonated and positively charged below pH ∼6.5. To search for potential heparin-binding regions and, in particular, to search for surface-exposed His residues that might be involved in pH-dependent interaction with heparin, we constructed a three-dimensional model of murine cathepsin E based on the refined 2.35 Å X-ray structure for human cathepsin E (Ostermann et al., 2004). Because heparin binding was not dependent on dimerization and dimerization does not appear to involve an extensive subunit-subunit interface (Ostermann et al., 2004), only the monomer was modelled. The model of mouse cathepsin E (Fig. 7) identified five surface-exposed His residues that might be interesting candidates for heparin binding. Two of the His residues have negatively charged residues nearby that might compensate for their negative charge. However, three His residues (His61, His65 and His78) and two Lys residues (Lys59 and Lys114) are found within a relatively limited region of the N-terminal domain of the molecule, where they might contribute to a heparin-binding site. It is interesting that the binding of two other mast-cell secretory-granule compounds, mMCP-6 (Hallgren et al., 2004) and mMCP-7 (Matsumoto et al., 1995), also involves surface-exposed His residues. A functional consequence of this mode of interaction would be that the binding to heparin is strong within the mast-cell granule (acidic pH), whereas exposure to neutral pH (e.g. following exocytosis) will cause His deprotonation and dissociation from the heparin proteoglycan.

Similar to cathepsin E, CPA is tightly associated with heparin proteoglycan within the mast-cell secretory granule and its storage is severely affected in the NDST-2-null mast cells (Forsberg et al., 1999; Humphries et al., 1999). It is thus possible that cathepsin E and CPA show a physical colocalization through the interaction with heparin proteoglycan. Such colocalization might be functionally important by bringing cathepsin E into close contact with pro-CPA, thus facilitating the proteolytic action of cathepsin E on pro-CPA. This would be analogous to the effect of heparin proteoglycan on mast-cell chymase, for which it has been shown that the heparin part of the heparin-chymase complex attracts heparin-binding substrate proteins and thus presents them to the chymase for efficient proteolysis (Tchougounova and Pejler, 2001).

Although cathepsin E has been implicated in several physiological and pathological conditions, the true biological function of cathepsin E is not known (Tsukuba et al., 2000). However, important clues to the function of this protease were obtained through the recent genetic targeting of cathepsin E (Tsukuba et al., 2003). It was found that the knockout of cathepsin E resulted in mice that were markedly prone to spontaneously developing atopic dermatitis, a condition that is strongly dependent on T-helper-2 cell responses. Notably, mast cells are strongly implicated in allergic diseases, including atopic dermatitis (Irani et al., 1989), and the current identification of cathepsin E in the secretory granule of mast cells might thus point to a potential role of mast-cell cathepsin E in allergic disease. However, an identification of the exact role of mast-cell cathepsin E, in comparison to cathepsin E from other cellular sources, will require substantial further investigations.

We thank Nils Ostermann for providing us with cathepsin E coordinates prior to public release. This work was supported by grants from the Swedish Research Council, Vårdalstiftelsen, Formas and King Gustaf V's 80th Anniversary Fund.