Diverse regulatory roles for lysosomal proteases in the immune response

Enzymes

Endosomes and lysosomes harbour mainly cysteine and aspartic acid proteases so-named because their active site utilises either a cysteine thiol or an aspartic acid as a key part of the catalytic site. Some endosome/lysosome located serine proteases such as cathepsin G, granzymes and thymus specific serine protease (TSSP) have important roles in the immune system; however, we focus primarily here on the cysteine and aspartic proteases. Most of the lysosomal cysteine proteases are related to papain and belong to the so-called C1 family. These include cathepsins L, S, C, F, H, B, X, K, V and W. Cathepsins D and E are also found in lysosomes but are aspartic acid proteases related to pepsin. An additional cysteine protease is found in lysosomes, which is more closely related to the caspases. This is asparaginyl endopeptidase (AEP) or legumain, a member of the C13 family. Some of these enzymes are endopeptidases (cathepsins S, L, K, F, V, D, E and AEP), whereas others are either amino (cathepsins C and H) or carboxy exo-peptidases (cathepsins X and B). Depending on pH, cathepsin B can function either as a carboxypeptidase or as an endopeptidase (Table 1) 5. An excellent general resource is the MEROPS peptidase database in which all peptidases (proteases) are classified according to their mechanism and sequence similarity: http://merops.sanger.ac.uk/index.htm. The functional significance of these enzymes are being clarified by the development of both specific inhibitors and gene-targeted mice. It is important to bear in mind that one protease may regulate the configuration, activity and longevity of other proteases. Consequently, the ablation of one activity may impact on others. For example, conversion of pro-AEP involves first autoactivation but then further processing by other enzymes 6. In turn, AEP is absolutely required for conversion of cathepsins B, L and H from a single chain to the normal two-chain form 7. Thus, targeting a single enzyme may have an impact on the activity or at least the form and stability of other enzymes. However, it is not clear if the single versus two-chain forms of the above-mentioned cathepsins have different substrate preferences. Endosomal/lysosomal proteases play important roles in several important areas of the immune response as we now discuss.

Antigen processing and presentation

Lysosomal proteases introduce cleavages in endocytosed antigens, which trigger unfolding and capture of processed antigen by newly synthesised class II MHC molecules 8-10. In addition, removal of the invariant chain (Ii) chaperone requires the action of one or more lysosomal proteases 11, 12. A key issue in this area has been the extent to which specific, as opposed to generic, protease activity is required. In other words, how redundant is Ii and antigen processing? This has been well reviewed elsewhere but briefly, data in mouse and human APC indicate that the later stages of Ii processing are relatively dependent on cathepsin S (in B cells and DC) or cathepsin L (or V) (in cortical thymic epithelial cells (cTEC) (reviewed in 13, 14). Initiation of Ii processing on the other hand can be performed by several enzymes including AEP 15, 16. In principle, blockade of Ii and/or antigen processing might be broadly immunosuppressive. Chemical or genetic ablation of cathepsin S attenuated Th1-driven autoimmune disease in different mouse models (reviewed in 13). Whether this is due to perturbation of Ii processing, antigen processing or some other process is less clear. This uncertainty was not an issue in a study on viral and ovalbumin cross-presentation on class I MHC molecules, which do not require Ii processing. Cross-presentation can be TAP-dependent or less commonly, TAP-independent. Surprisingly, the absence of cathepsin S markedly compromised TAP-independent cross-presentation of both ovalbumin and influenza virus antigens consistent with a role for this enzyme in epitope generation 17.

Interestingly, a recent study has linked Ii processing to DC migration. Compared with WT DC, those lacking Ii migrated faster to draining lymph nodes whereas DC lacking cathepsin S, which accumulates Ii, migrated more slowly 18. Apparently, the presence or persistence of Ii acts as a brake on DC migration, which is only relieved by proteolytic destruction of Ii. Mechanistically, this appears to be due to sequestration of myosin II onto Ii-positive endocytic compartments 18. Other recent studies report a transient arrest of DC migration following TLR signalling that correlated with loss of actin-rich matrix metalloprotease-organising podosomes, which may also be required for DC exit from tissues 19. Taken together, these studies support the idea that DC exit from tissues is coupled to completion of the earlier stages of antigen capture and Ii processing.

The p41 isoform of the Ii, which is abundantly expressed in some DC, turned out to be a potent inhibitor of cathepsin L 20 and recently was shown to inhibit other cathepsins 21. However, the net effect of p41 on cathepsin L activity remains unresolved since it also acts as a cathepsin L chaperone and increases active cathepsin L protein levels in bone marrow macrophages 22. Nonetheless, the ability of p41 to attenuate protease activity may account, at least in part, for its ability to enhance antigen presentation given that proteases can occassionally destroy T cell epitopes as discussed below 23.

Besides the requirement for cathepsin S in cross-presentation noted above, examples of specific enzyme requirements for antigen processing in vivo have been harder to find. Another study on cross-presentation identified a requirement for cathepsin D in human DC that had taken up apoptotic or necrotic monocytes expressing vaccinia-encoded flu matrix protein 24. Cathepsin D appeared to be acting upstream of the proteasome in this system. Cathepsin D null mice die soon after birth but bone marrow chimeric mice reconstituted with cathepsin D-deficient marrow showed no differences in B- or T-cell responses following DNP-Ig or sheep red blood cell immunisations 25. At least in vitro, loss or inhibition of single enzymes can impact, positively or negatively, on CD4⁺ T-cell epitope generation and/or on the spectrum of elutable MHC class II bound peptides (reviewed in 26). Individual enzymes present in lysosomal extracts frequently dominate processing of specific antigens in vitro and blockade of those enzymes can reduce presentation of T-cell epitopes 27, 28. This requirement may be for initial “unlocking” cleavages 29, 30 or downstream processing events 28.

Loss or inhibition of cathepsin E has been reported to compromise processing and presentation of antigens such as ovalbumin, myoglobin and tetanus toxin 31, 32. However, contrasting results were seen when macrophages and DC were compared, as cathepsin E null DC showed an enhanced presentation of ovalbumin, a result attributed to increased phagocytosis and elevated expression of co-stimulatory molecules 31. Cathepsin E-deficient mice are susceptible to developing atopic dermatitis, attributed to the accumulation of both Th2 polarising cytokines and IgE. Interestingly, atopic dermatitis in humans was also linked to low levels of cathepsin E 33. In addition, mast cells lacking cathepsin E accumulate the pro-form of carboxypeptidase A, though whether this is relevant to the development of dermatitis is not clear 34.

Although not all cathepsin null mice have been tested, there is little evidence thus far that CD4⁺ T-cell immune responses to a particular antigen in vivo critically depend on a specific lysosomal protease. Clear relationships between antigens and proteases found in vitro, such as that between tetanus toxin and AEP, may be blurred in vivo because the longer time scales of antigen presentation (days rather than hours) may allow more slowly presented epitopes to “catch up”. Also as noted above, loss of one enzyme can affect others. For example, a substantial accumulation of cathepsin L is observed in lysosomes of AEP-deficient mice and this may partially compensate for the predicted shortfall in tetanus toxin processing 7, Matthews et al. (submitted for publication). In addition, there is strong evidence that the protease content of primary cells and immortalised cell lines may be very different 35, 36, Matthews et al. (submitted for publication). In fact, even different immortalised human B cell lines may differ in protease content 16. These findings make direct comparisons between in vitro and in vivo studies difficult.

Importantly, several recent studies highlight the fact that lysosomal proteases can destroy as well as generate T-cell epitopes. DC express reduced levels of lysosomal enzymes compared with macrophages, which appear to improve rather than hinder their performance as APC 37. They can also attenuate protease activity, at least in newly formed phagosomes, by raising phagosomal pH, which protected antigen for subsequent cross-presentation on class I MHC molecules 38. In the same vein, eliminating specific enzymes 39 or making antigens more protease resistant 40 can improve antigen presentation and immunogenicity. Destructive processing events may not only limit immunogenicity but might also compromise the establishment of tolerance to self-proteins 30, 41 but this remains to be proven.

Lysosomal proteases found in APC may be complemented by enzymes taken up from exogenous sources. For example, cathepsin G, an enzyme more commonly associated with the granules of neutrophils, is accumulated from exogenous sources by primary human B cells and contributes to destructive processing of the autoantigen myelin basic protein 35. In contrast, in EBV-transformed human B cells similar destructive processing of MBP is mediated by endogenous AEP 35, 41. These studies further illustrate that the enzymes expressed in cultured APC may not be the same as those found in primary ex vivo APC.

Reciprocal relationships between cytokines and lysosomal cathepsins

Stimulation of immune cells with certain cytokines can regulate lysosomal protease activity and conversely, cathepsins participate in cytokine activation and turnover. In the latter situation, proteases may engage their substrate pericellularly, i.e. outside their normal endosome/lysosome domain (Fig. 1A). For example, TNF-α and IL-1β increased the activity of cathepsins S and B in human DC, leading to increased class II MHC dimer formation and T-cell recognition 42. Other pro- and anti-inflammatory cytokines such as IL-6 and IL-10 also modulate protease activity upwards or downwards, respectively, at least in part by altering endosomal pH 42, 43. In another study, IL-10, produced by BCG infected macrophages, reduced cathepsin S expression levels, thereby overriding the ability of IFN-γ to enhance class II MHC levels and promote T-cell recognition of infected cells 44. However, another study found that IL-10 (and IL-6) enhanced the activity of various cathepsins, including cathepsin S and that IL-6 diminished DM levels 45. Conceivably, this may have accounted for the improved presentation of DM-sensitive “cryptic” T-cell epitopes in IL-6-treated DC 43, 46.

Cathepsin activity has been shown to either directly activate or inhibit certain subsets of cytokines. For example, the activity of IL-8, a strong neutrophil chemoattractant and activator, was originally reported to be enhanced by limited N-terminal truncation of five residues by neutrophil granule proteases, such as proteinase 3 47, whereas in contrast the same enzymes were reported to degrade IL-6 48. More recently, a convertase activity for IL-8 in human fibroblasts was identified as cathepsin L 49 (Fig. 1A). Perhaps different cells activate IL-8 using different enzymes. The role of TNF-α-converting enzyme (ADAM17) in liberating active TNF-α is well described. A recent study identified an additional requirement for cathepsin B, at least for a proportion of TNF-α release in some macrophages. Ablation of cathepsin B led to accumulation of the membrane bound TNF-α pro-form in intracellular vesicles implying a possible role for cathepsin B in fusion of these vesicles with the plasma membrane 50. The relevant substrates for cathepsin B were not identified; therefore, it remains to be established whether cathepsin B is acting within the endosome/lysosome compartment or within the cytosol.

Lysosomal proteases and T-cell development

Mice lacking either cathepsin S or cathepsin L have defects in the development of NKT cells though apparently for different reasons. NKT cells are known to be positively selected by CD1d molecules expressed on double positive thymocytes and this process is defective in cathepsin L null mice. NKT cell activation in vitro was similarly defective. Honey et al., showed that CD1d trafficking appeared normal in thymocytes lacking cathepsin L, leading them to speculate that generation/loading of the appropriate glycolipid ligands known to be displayed by CD1d, was dependent on cathepsin L 51. The key substrate(s), however, have not been identified. Interestingly, in this study the use of radiation bone marrow chimeras demonstrated that thymocytes could acquire cathepsin L produced by neighbouring thymic stromal cells similar to the acquisition of cathepsin G by B cells discussed in Antigen processing and presentation. 51, Mice lacking cathepsin S also showed defects in NKT cell selection in one 52 but not in another study 51. Riese et al., suggested that thymic DC are also involved in NKT cell selection and that the accumulation of class II/Ii intermediates seen in cells lacking cathepsin S perturbs DC endosomal compartments through which CD1d molecules traffic 52.

Since conventional T-cell development in the thymus is regulated by self-peptide presentation on class II MHC molecules, lysosomal proteases are likely to be important for thymocyte development too. Consistent with a requirement for cathepsin L in Ii removal in cTEC, mice lacking this enzyme exhibit ∼70% reduction in CD4⁺ T-cell numbers, apparently as a result of accumulation of Ii fragments in cTEC. As a result, cathepsin L-deficient mice display an altered T-cell repertoire and an increase in negative selection on bone marrow-derived cells (which use cathepsin S rather than L for Ii metabolism) 53. However, mice lacking cathepsin L still selected a reduced number of CD4⁺ T cells when Ii was also missing implying that this enzyme may additionally be generating some peptides required for positive selection 54.

Studies in nackt mice, which express a truncated, apparently inactive form of cathepsin L, indicate that the enzyme may also control ECM expression in lymphoid organs. In addition, in spite of reduced thymic output, normal absolute numbers of CD4⁺ T cells were observed in lymph nodes of nackt mice 55. It is not clear if the phenotype in nackt mice is reproduced in gene-targeted cathepsin L-deficient mice.

Activation of granule serine proteases

The effector functions of CD8⁺ T cells, NK cells, neutrophils and mast cells depend on the activation of granule-localised serine proteases such as granzymes A and B (T and NK cells), elastase, cathepsin G and proteinase 3 (neutrophils) and chymase (mast cells). These enzymes are initially made as inactive zymogens and depend on the amino dipeptidase cathepsin C (also known as DPPI), which removes two N-terminal residues to trigger activation 56. Mice lacking cathepsin C fail to activate neutrophil and mast cell granule proteases 57, 58. Defects in cytotoxic T-cell function have also been reported in these mice 59. However, recent studies indicate that residual granzyme B (but not granzyme A) activity in cytotoxic T cells was observed in the absence of cathepsin C and that this was sufficient to control some viral infections 60. Similarly, human insufficiency in cathepsin C, resulting in Papillon Lefévre syndrome, is characterised by periodontitis and skin infections but not by a general T-cell immunodeficiency probably because significant granzyme activity was still present 61. Neutrophil serine protease activity was more profoundly affected and this resulted in deficient bactericidal activity in some but not all patients studied 61. Low and variable levels of residual cathepsin C activity might account for these differences but it is also possible that an additional enzyme(s) can substitute for cathepsin C in serine protease zymogen activation.

Paradoxically, lack of cathepsin C was actually protective in some murine models of inflammation and sepsis due to attenuated neutrophil function 57, 62. These data indicate that cathepsin C may be an attractive drug target to control multiple inflammatory serine proteases. Nonetheless, it appears that efficient and sustained inhibition of the enzyme will be required to fully quench neutrophil granule protease activity 63. Cathepsin C may itself be regulated by cystatin F. Cystatins are a family of proteinaceous cysteine protease inhibitors, some of which bind very tightly to cathepsin active sites 64. We recently showed that cystatin F, which is principally found in immune cells, could inhibit cathepsin C 65. Cystatin F is made initially as an inactive disulphide-linked dimer 66-68. Reduction generates an inhibitor able to target some endopeptidases but we found that carbohydrate-driven lysosomal targeting 69 resulted in monomerisation and activation by proteolysis, which extended the protease target range to enzymes such as cathepsin C 65. The presence of a cathepsin C protease inhibitor regulated by proteolysis suggests a possible autoregulatory feedback system to limit cathepsin activity.

Cathepsin W, also known as lymphopain, is selectively expressed in CD8⁺ T cells and NK cells which led to the suggestion that it might have a function in cytotoxicity. However, at least in vitro, CTL and LAK cell-mediated cytotoxicity was not affected by genetic ablation of cathepsin W 70, 71. The role of cathepsin W remains to be elucidated; its unusual localisation in the ER and golgi apparatus is likely to be significant.

Integrin activation and cell adhesion

Integrins may be activated by “inside-out” signalling, for example, following FcR ligation by immune complexes. Increased integrin affinity for its ligands then permits enhanced “outside-in” integrin activation of downstream signalling pathways, leading to the secretion of chemokines and reactive oxygen intermediates, cytoskeletal remodelling and other inflammatory events. Cathepsin G-deficient neutrophils failed to trigger normal CD11b integrin clustering which in turn compromised Rac1 activation and the production of MIP-2 and reactive oxygen intermediates 72. This phenotype could be rescued by addition of exogenous catalytically active cathepsin G. Its substrates remain to be identified, however.

Cathepsin X (also known as cathepsin Z) is expressed in monocytes, macrophages, DC and in prostrate and gastric tumours. It is a mono and di-carboxypeptidase similar to cathepsin B. Several reports link cathepsin X with integrin-dependent cell adhesion, migration and events contingent on adhesion. For example, overexpression of the enzyme in Jurkat T cells enhanced cell polarisation and migration through 3-D matrices 73 (Fig. 1B). Another report from the same group indicated that DC adhesion and maturation required cathepsin X activity 74. Although the effects of cathepsin X overexpression and inhibition are striking, several issues remain unresolved. First, the proposed substrates of cathepsin X, the cytoplasmic tails of β2 integrins, would not normally be accessible to a lysosomal enzyme and second, the primary reagent used to inhibit cathepsin X in these studies is an active-site-directed antibody which when applied to cells would be expected to inhibit extracellular enzyme but not enzymes localised to the cytosol. It will be important to obtain direct evidence in living cells that cathepsin X acts on the cytoplasmic tails of β2 integrins and that integrins, rather than other substrates, are relevant to the effects observed. Some of the effects of cathepsin X may not in fact require its enzyme activity: an earlier report showed that an RGD motif within the pro-region of cathepsin X, interacted directly with αvβ3 integrins and enhanced attachment and spreading of HUVEC cells 75. Further clarification of the immunological roles of cathepsin X will hopefully follow with the generation of cathepsin X-deficient mice. Apoptosis is another instance whereby lysosomal proteases act on substrates found within the cytosol, as we discuss in the following section.

Lysosomal-mediated apoptosis

Apoptosis is a process of orchestrated cell death resulting in minimal inflammation of the surrounding tissue. Although the prototypical apoptotic response involves a family of cytosolic aspartate-specific cysteine proteases called caspases, growing evidence suggests that in addition, or instead, apoptosis results from limited lysosomal membrane permeabilisation (LMP) and the subsequent release of cathepsins into the cytosol (Fig. 1C). During apoptosis, cathepsin-driven proteolysis results in the loss of mitochondrial outer membrane permeabilisation (MOMP) and caspase activation, among other hallmarks of the apoptotic process. Although the loss of lysosomal integrity as a consequence of apoptosis was demonstrated some time ago 76, the mechanism(s) controlling this phenomena, as well as its hierarchy within the apoptotic signalling cascade, are still in relative infancy (for an extensive review refer to 77). A number of recent studies have begun to clarify the role of cathepsins in apoptosis. Here, we briefly discuss the role of cathepsin-driven apoptosis in immune cells.

Apoptosis in immune cells is induced by a number of physiological stimuli and occurs as a mechanism to control inappropriate immune responses and during the resolution/contraction phase of normal responses. Cathepsin B 78 and cathepsin D 79, 80 have both been shown to target cytosolic substrates following LMP in immune cells. In T cells, for example, caspase-dependent apoptosis often follows TCR stimulation by a mechanism that depends on the TNF family of death receptors (activation-induced cell death) or occurs by mechanisms independent of death inducing ligands (activated T-cell autonomous death) 81. Similarly, following conditions where high doses of antigen are seen, peripheral T cells become tolerogenic and subsequently die through a mechanism involving lysosomal cathepsins. Michallet et al., recently showed that in a model of high-dose tolerance, mimicked by TCR cross-linking antibodies or PHA, cathepsins B and L were released into the cytosol resulting in DNA fragmentation 82. Neither cytochrome c release nor caspases 3 and 9 activation were required appearing to rule out, at least in this case, the possibility that LMP activated the intrinsic mitochondrial pathway of apoptosis.

T-cell contraction at the end of an immune response also results in massive apoptosis with a small subset of highly specific memory cells surviving. Commitment to T-cell memory involves the upregulation of anti-apoptotic proteins. Spi2A is cytosolic cysteine protease inhibitor that has recently been shown to be upregulated in memory CD8⁺ T cells and was shown to suppress cytosolic cathepsin B activity. Importantly, overexpression of Spi2A in LCMV-specific CD8⁺ T cells resulted in an increased memory pool, whereas anti-sense mediated knockdown of Spi2A resulted in a more severe contraction phase and reduced anti-LCMV CD8⁺ T cells 83. These data demonstrate the importance of the endogenous regulation of cathepsin activity within the cytosol. Like T cells, germinal centre B cells are subject to cathepsin-mediated apoptosis through their antigen receptors 84. Interactions with follicular DC and CD40 ligation were shown to rescue B cells from LMP and apoptosis 85. Further studies will be necessary to pin down the role of lysosomal enzymes and the mechanism of their escape from lysosomes during these processes.

Cathepsin-mediated apoptosis can also occur as a consequence of intracellular pathogens. Laforge et al., recently showed that CD4⁺ T cells harbouring HIV are susceptible to apoptosis via a mechanism involving the virally encoded protein Nef and release of the lysosomal cysteine and aspartyl proteases cathepsin B and cathepsin D, respectively. Nef is known to perform a key role in HIV pathogenicity by modulating membrane receptor levels, among other functions. Their data further demonstrated that Nef alone was sufficient for limited LMP, although how Nef mediates cathepsin trafficking to the cytosol remains unclear 79. Cytosolic cathepsins were shown to induce changes in MOMP by targeting the anti-apoptotic Bcl-2 family member, Bax. Lysosomal destabilization within immune cells by viral Nef appears to be another example of the effective pathogenesis of HIV 86.

Cathepsins in the cytosol are likely to exhibit limited proteolytic activity at least in some cases due to structural instability at neutral pH. Nonetheless, there is clear evidence that some activity persists most likely against a limited range of substrates 87-89. Identification of the relevant physiological targets of cytosolic cathepsins has been complicated by the fact that many apoptotic signalling cascades occur in parallel. Furthermore, whether LMP occurs as an early or late event within apoptosis varies according to cell type and the primary trigger of apoptosis.

How lysosomal permeabilisation is regulated is still under investigation. Lysosomal permeabilisation appears to be inducible by diverse pathways including lysosome-proximal generation of ROS enhanced by intra-lysosomal iron loading and the accumulation of lysosomotropic lipids such as sphingosine. However, these are just two of the number of lysosome-destabilising mechanisms that have been described and the reader is referred to the recent review of Boya and Kroemer for the full discussion 77. What seems to be clear is that the precise mechanism will depend on the initial triggers of LMP. Of interest are reports indicating that, in addition to their role in MOMP, Bcl2 family members such as Bim and Bax, also appear to be involved in LMP 90. Caspases and even cathepsins themselves may be involved in amplifying the LMP pathway of apoptosis. For example, caspase 9 (activated downstream of caspase 8) contributed to TNF-induced cell death in MEF by inducing LMP 91.

It is possible that the presence of naturally occurring cathepsin inhibitors, such as serpins and cystatins, attenuate the cytosolic activity of cathepsins and hence LMP-driven apoptosis 78, 83, 92 (Fig. 1C). Attenuation of the levels of these inhibitors might be pro-apoptotic and therefore therapeutically useful, for example, in autoimmune lymphocyte populations or tumour cells.

TLR signalling

An unexpected role for endosomal proteases has emerged in innate responses to nucleic acids. TLR 3, 7, 8 and 9 are known to signal from endosomes rather than the cell surface, a restriction which limits activation to pathogen derived material. Asagiri et al., reported that chemical or genetic ablation of cathepsin K attenuated inflammatory autoimmune disease apparently via a marked reduction in TLR9-induced signalling leading to the production of IL-6 and IL-12 93. Three recent studies confirmed a requirement for endosomal proteolysis in TLR9 signalling and went further by showing that a key protease target is TLR9 itself 1, 2, 94. Apparently, cleavage in the extracellular domain of TLR9 increases binding of CpG containing oligonucleotides and crucially, cleaved TLR9 recruited the signalling adaptor MyD88 much more efficiently than intact TLR9 1. However, none of these studies confirmed a role for cathepsin K as the sole TLR9 convertase. Instead, several different lysosomal cysteine proteases appeared capable of TLR9 processing. The reason for this discrepancy is not yet clear. Conceivably, cathepsin K is required for an unknown processing event important for the cytokine readouts and disease models employed by Asagiri et al. Together, these studies identify a role for the lysosomal-processing system in innate immune responses additional to that in adaptive immunity. What could be the advantage of limiting TLR9 signalling to a processed form of the receptor? A requirement for processing might help ensure that nucleic acid-sensing TLR only respond to nucleic acids released in endosomes from viruses and bacteria. Whether proteolysis of the other nucleic acid-sensing TLR is required is not yet clear. In the studies mentioned above, one demonstrated that TLR7 proteolysis occurred 1, but no cleavage of TLR7 was detected in another study 2.

Cytosolic and nuclear substrates of lysosomal proteases

Recent studies indicate that lysosomal proteases may act on cytosolic and nuclear substates, apart from those that are involved in the LMP-driven apoptosis. Growing evidence indicates regulation of several important cell biological phenomena by what might be termed, compartmentally reassigned lysosomal proteases (Fig. 1). For example, AEP acts on the nuclear phosphoprotein SET, which promotes DNA-nicking activity and neuronal cell death 95. How AEP or other endocytic proteases might access the cytosol/nuclear compartments is not known. Cathepsin L is also detected in the nucleus of some cells at the G1/S cell cycle transition and processes the transcriptional regulator CCAAT displacement protein (CDP)/Cux into a form with enhanced DNA binding and that promotes cell cycle progression 89. Compartmental reassignment of cathepsin L is achieved through translation initiation at alternative start codons downstream of the normal signal sequence (Fig. 1D). Interestingly, nuclear fractions appeared to contain not only the single chain but also the heavy chain of the two-chain form of cathepsin L 89. How this signal sequence deficient form of cathepsin L is processed to an apparent two-chain form is not clear. This unexpected relationship between cathepsin L and the CDP transcriptional regulator appears to be exploited by the gram negative pathogen Anaplasma phagocytophilum, which causes human granulocytic anaplasmosis and is harboured within neutrophils and compromises their function. Thomas et al., showed that this organism stimulates the expression of cathepsin L, leading to enhanced CDP cleavage and the repression of CDP-regulated genes needed for effective neutrophil function 96.

Concluding remarks

Here, we have tried to outline the diverse situations in which proteases in the endosome/lysosome compartment regulate immune phenomena. Although there is undoubtedly redundancy, it is remarkable, given the apparent broad substrate specificity of many of these enzymes that clear phenotypes are observed following ablation of individual enzymes. Although topologically “outside” the cytosol, it is clear that processing events in endosomes can have profound influence on signalling systems that operate in the cytosol. The recent work on the link between Ii processing and DC migration and on TLR9 processing and its signalling properties are two examples. In other cases, there are clear phenotypes following specific protease ablation, but the key substrates remain to be identified. For example, what is the key substrate(s) of cathepsin L that determines normal NKT cell development? Other emerging aspects of the lysosomal protease system deserve highlighting. It is becoming clear that the protease complement of immortalised cell lines may be very different to the equivalent cell types in vivo. Also, cells can and do acquire enzymes (and their endogenous inhibitors such as cystatins) passively, by endocytosis. The actual activity of a given enzyme is therefore controlled by various factors including its conversion from an inactive precursor form and the extent to which it is neutralised by endogenous inhibitors. The use of active-site-directed probes remains the best way to compare activity in different cells 97. Surprisingly, and notwithstanding their acidic pH optima, there is growing evidence that some lysosomal enzymes may have important substrates in the cytosolic and nuclear compartments, as well as the extracellular space. How they access these compartments, how they remain proteolytically stable and whether their trafficking and activity can be regulated will be important to determine in future studies. Finally, the recent success of protease targeting drugs such as angiotensin-converting enzyme inhibitors and HIV protease inhibitors demonstrates that proteases are, in principle, “drugable” as an enzyme class 98. Protease targets continue to be pursued in other areas, for example, as anti-infectives (e.g. malaria) and in cancer 99, 100. The new roles that have emerged for proteases in activation of innate and adaptive immune responses which, when normal regulatory mechanisms are bypassed, can result in chronic inflammation, autoimmunity and allergy will, it seems, continue to make them attractive as therapeutic targets.