A family with Papillon-Lefèvre syndrome reveals a requirement for cathepsin C in granzyme B activation and NK cell cytolytic activity

Natural killer (NK) cells are cytolytic lymphocytes whose function is to control infection and malignancy. Granule exocytosis delivers granzymes, a family of pro-apoptotic serine proteases, to target cells leading to the induction of caspase-dependent and caspase-independent apoptosis.¹  Mouse granzyme B is required for the rapid induction of apoptosis in target cells.²  However, other granzyme molecules contribute to the overall cytolytic phenotype,³  with human cells expressing granzymes A, B, H, K, and M.^1,4 

Activation of granzymes requires the removal of an N-terminal dipeptide.⁵  Cathepsin C, a cysteine protease also known as dipeptidyl peptidase I (DPPI), has been implicated in granzyme A and B processing.^6,7  Cathepsin C-deficient mice fail to process granzymes A and B and hence lack both NK and T-cell-mediated cytolytic activity.⁷  Human cathepsin C deficiency is the cause of Papillon-Lefèvre syndrome (PLS).^8-10  This is an extremely rare, autosomal recessive condition characterized by palmoplantar keratosis and severe periodontitis. In addition, skin infections and liver abscesses have been reported in PLS,^10,11  strongly suggesting an immunodeficiency component to the disease. However, in contrast to the mouse model, interleukin-2 (IL-2)-generated lymphokine activated killer (LAK) cell activity in these patients appears to be normal, and granzyme B is active in PLS LAK cells.¹²  Here, we have tested the requirement for cathepsin C in the processing of granzyme B in unstimulated human NK cells from PLS patients. We describe an NK cell cytolytic defect and a failure of granzyme B activation.

PLS patients 1 and 2 are brother and sister from a consanguineous family, family 6 in Toomes et al.⁸  Both patients are homozygous for a mutation changing glycine 277 of cathepsin C to a serine (G277S). This nomenclature is that used in the structural study¹³  and refers to cathepsin C with the 24 amino-acid signal peptide removed. A third family member is unaffected by PLS and has one G277S allele and one wild-type allele and is referred to as the heterozygote. Blood samples from patients, the heterozygote, and healthy control donors were collected with informed consent and approval from the Leeds Teaching Hospital NHS Trust UK ethics committee.

NK cells were phenotyped in whole blood using flow cytometry and isolated from peripheral blood mononuclear cells (PBMCs) using the NK isolation kit II (Miltenyi Biotec, Bergisch Gladbach, Germany). NK cells were used directly or cultured for 10 to 14 days with 50 units/mL human recombinant IL-2 (Sigma, St Louis, MO). Whole cell lysates were assayed for cathepsin C or granzyme B activity using the colorimetric substrates Gly-Phe-pNA (Sigma) and Ac-IEPD-pNA (Calbiochem, San Diego, CA), respectively,^14,15  or used in immunoblotting experiments. Aprotinin agarose binding was performed as described.^16,17  Killing and caspase activation assays were performed using flow cytometry, using CellTracker dye (Molecular Probes, Eugene, OR)-labeled K562 target cells and either propidium iodide¹⁸  (PI) or PhiPhiLux (Calbiochem) staining.¹⁹ 

The PLS patients exhibited very low levels of cathepsin C substrate hydrolysis in unstimulated and IL-2-stimulated NK cells compared with controls (Figure 1A), consistent with the proximity of the mutated residue to the active site of the enzyme.¹³  The CD56^bright and CD56^dim NK subsets²⁰  from the patients and controls were indistinguishable (Figure 1B), and expression of CD158a, CD158b, CD94, and CD16 was also normal in the patients (Figure S1, available on the Blood website; see the Supplemental Figures link at the top of the online article). In a previous report, LAK cells from PLS patients were found to have normal cytolytic activity.¹²  Our results confirmed this since control, heterozygote, and PLS NK cells cultured in IL-2 were able to kill K562 target cells (Figure 1C). The heterozygote had an intermediate level of cathepsin C activity (Figure 1A). However, this dosage effect was not evident in the cytoxicity assay (Figure 1C) since, in multiple experiments, the cytolytic activity of the PLS LAK cells was never less than the controls. Furthermore, cell lysates from IL-2-activated PLS NK cells and control NK cells contained similar levels of granzyme B activity (Figure 1D).

These results confirm that human LAK cells, unlike those from the mouse,⁷  do not require cathepsin C for cytolytic activity or granzyme B activation. An early event in granzyme B-mediated apoptosis is the induction of the caspase cascade.¹  To verify that PLS LAK cells were capable of activating the caspase cascade, we modified the killing assay to include the fluorescent caspase 3 substrate, PhiPhiLux, in place of PI staining.¹⁹  The results indicate that IL-2-stimulated NK cells from PLS patients and controls are able to rapidly induce the caspase cascade in target cells (Figure 1E).

The proliferation and activation induced by IL-2 provides a convenient method to generate large numbers of highly cytolytic NK cells for functional studies. NK cells are activated by T-cell-derived IL-2.²¹  However, this is likely to be a late event in a primary immune response. We therefore analyzed PLS NK cells directly, without provision of exogenous IL-2. Unstimulated PLS NK cells were not cytolytic (Figure 2A). Lower levels of target cell killing by heterozygote NK cells could be increased using higher effector-target cell ratios. However, this was not the case with the PLS NK cells, which are clearly defective (Figure 2B). Furthermore, unstimulated PLS NK cells were unable to initiate the caspase cascade in target cells (Figure 2C), and lysates from the unstimulated PLS NK cells possess little granzyme B activity (Figure 2D). Lack of cytolytic activity was not due to loss of perforin expression, since this was normal in PLS NK cells (Figure S2).

A lack of cathepsin C activity in unstimulated PLS NK cells coupled with deficient granzyme B activity suggests a failure of granzyme B processing. On immunoblots, granzyme B from unstimulated PLS NK cells is slightly larger than granzyme B from unstimulated control cells, consistent with the retention of the pro-domain in unstimulated PLS NK cells (Figure 2E). In IL-2-stimulated NK cells, control samples possess the smaller species, whereas the corresponding samples from PLS patients contain both forms. The presence of both species in lysates from PLS IL-2-stimulated NK cells suggest that IL-2 activation allows the granzyme pro-domain to be removed, albeit with lower efficiency.

Removal of the dipeptide pro-domain from several serine proteases exposes the active site to inhibitors such as aprotinin, allowing processed forms to be identified by binding to aprotinin-agarose.^16,17  Granzyme B (which is inhibited by aprotinin²² ) was detected in the aprotinin-bound fraction in all the IL-2-stimulated samples (Figure 2F), consistent with the presence of enzymatically active granzyme B (Figure 1D). However, no granzyme B was detected in the aprotinin-bound fraction in unstimulated PLS NK cells. Instead, granzyme B from unstimulated PLS NK cells is found only in the unbound fraction (Figure 2F). Furthermore, in cell lysates, active granzyme B forms a sodium dodecyl sulfate (SDS)-resistant complex with the protease inhibitor PI-9.²³  This complex was observed in IL-2-stimulated NK cell lysates from patient and control samples as well as lysates from unstimulated NK cells from control samples. However, the complex was not detectable in the unstimulated PLS NK cell lysates, indicating that granzyme B was not activated in these cells (Figures S3). Collectively, these results show that although granzyme B is present in the unstimulated PLS NK cells, it is inactive because of a processing defect. However, IL-2 does not induce cathepsin C activity in human NK cells (Figure 1A). Interestingly, some granzyme C is processed in cathepsin C-deficient mice,⁷  indicating that cathepsin C-independent pathways of granzyme processing are present in mouse and human.

The cytolytic defect in PLS NK cells is likely to be due, in part, to a failure in granzyme B processing. However, cathepsin C is likely to be important for the activation of other granzymes. Thus, at present, we cannot unequivocally link the NK cytolytic defect to a failure of granzyme B activation, since other cytolytic components also might require cathepsin C-mediated activation. In addition, we have studied a single mutation, and the role of cathepsin C in human NK cell function should be confirmed in unrelated PLS patients. In PLS, neutrophil defects¹²  and impaired NK activity may increase susceptibility to particular infections. NK cell deficiencies are frequently associated with herpesvirus susceptibility,²⁴  and this family of viruses may play a role in periodontitis.²⁵  While IL-2 therapy might improve NK cell function, it is unlikely to overcome defects in other cell types.²⁶  In vivo, the cytolytic defect is presumably overcome once IL-2 (or possibly other cytokines) is produced.

We are indebted to the PLS family for participating in this study. We thank Carmel Toomes for genetic information on the PLS family, Mushtaq Ahmed for help with patient liaison, and colleagues for numerous blood samples. Finally, we thank Gina Doody and Yasser El-Sherbiny for help and advice, and Peter Selby for support and encouragement.