Key Points
GrB-induced intracellular cleavage and phosphorylation events do not need perforin.
These cleavages and phosphorylations require GrB enzyme activity.
DDR and IRF-3 phosphorylation are some of the functional results of these events.
Abstract
Granzyme B (GrB) is an immune protease implicated in the pathogenesis of several human diseases. In the current model of GrB activity, perforin determines whether the downstream actions of GrB occur intracellularly or extracellularly, producing apoptotic cytotoxicity or nonapoptotic effects, respectively. In the current study, we demonstrate the existence of a broad range of GrB-dependent signaling activities that 1) do not require perforin, 2) occur intracellularly, and 3) for which cell death is not the dominant outcome. In the absence of perforin, we show that GrB enzymatic activity still induces substoichiometric activation of caspases, which through nonlethal DNA damage response signals then leads to activity-associated phosphorylation of IFN regulatory factor–3. These findings illustrate an unexpected potential interface between GrB and innate immunity separate from the traditional role of GrB in perforin-dependent GrB-mediated apoptosis that could have mechanistic implications for human disease.
This article is featured in Top Reads, p.243
Introduction
Granzyme B (GrB) is a secreted immune protease implicated in the pathogenesis of a broad range of human diseases (1–7). GrB and the other human granzyme family members (granzymes A, H, K, and M) were initially described as cytotoxic mediators secreted by NK cells and CTLs into immune synapses formed with target cells (8). Crystallography studies have resolved the general structure of GrB as a homodimer of six-stranded β-barrels organized around a catalytic site defined by a triad of serine, histidine, and aspartic acid (9). The substrate specificity of GrB is unique among serine proteases. The tetrapeptide immediately N-terminal to the cleavage site invariably has aspartic acid as the P1 residue; P4 is usually isoleucine, valine, or leucine; whereas the specificity for residues in P2 and P3 is broader (10, 11). This tetrapeptide preference is distinct from that of the group II caspases-3 and -7, which preferentially recognize a consensus P4–P1 sequence of aspartic acid–glutamic acid–valine–aspartic acid (DEVD) (12). Interestingly, GrB activates group II caspases directly by cleaving their proenzyme forms and also indirectly through activating cleavages of group III caspases, which in turn also cleave the proenzyme forms of group II caspases (13, 14). Mature group II caspases and GrB often cleave the same substrate at sites in close proximity (typically 5–30 aa apart) or at shared sites (13, 15). These observations suggest that these protease families might have evolved convergently to cleave features of functional relevance in common target substrates.
The cytotoxic activities of GrB are focused on target cells by cosecretion with the lytic factor perforin (16–18). This requirement for perforin is thought to arise from its function as an essential factor enabling cytosolic and nuclear accumulation of GrB, although the exact underlying mechanisms of perforin action continue to be debated (19–23). Regardless, after entering cells, GrB-mediated apoptotic death is induced by group II caspase activation and cleavage of a diverse array of intracellular substrates, either by activated caspases or directly by GrB [the GrB degradome (24)]. It is noteworthy that activated CTL infiltrates expressing perforin and GrB are present in affected tissues in a variety of disease states, including atherosclerosis, solid transplant rejection, and a wide spectrum of autoimmune rheumatic diseases, including scleroderma skin and lung disease, Sjogren syndrome sialadenitis, rheumatoid arthritis synovium, diseased muscle in inflammatory myositis, lupus skin and kidney, and the arterial walls in giant cell arteritis (5, 25–32). The extent, however, to which lymphocyte cytotoxicity mediated by perforin–GrB cooperativity contributes to overall tissue dysfunction in such diseases remains unresolved.
GrB may also have extracellular effects in tissues in the absence of perforin. These effects could occur by at least two different mechanisms: 1) GrB may escape from normal immune synapses and diffuse to neighboring cells, or 2) GrB expression in the absence of perforin is a feature of specific immune effector cells (e.g., specialized CD4 regulatory T cells and CD4 memory cells, plasmacytoid dendritic cells, mast cells, various granulocytes, and regulatory B cells) and even some nonimmune cells, including specialized keratinocyte subsets (33–39). Such cells would secrete GrB without perforin into the extracellular space. Extracellular GrB has effects on extracellular matrix (ECM) remodeling, postsecretion activation of cytokines, and on surface receptor signaling and has been proposed to have a role in chronic inflammation and wound healing (20, 39–44).
In this manuscript, we investigate the complementary existence of GrB-induced activities that occur intracellularly, are mediated by protease activity, do not require perforin, and for which cell death is not the dominant outcome. Even without perforin, GrB enzymatic activity still promotes substoichiometric activation of group II caspases and cleavage of several intracellular proteins, causing activation of various downstream pathways in cells. As an example, we show that GrB treatment without perforin induces DNA damage in a nonlethal group II caspase-dependent manner. Subsequent DNA damage response (DDR) signaling transduced through ATM and DNA-PKcs also leads to phosphorylation of the transcription factor IFN regulatory factor (IRF)–3 in a manner associated with its activation. These findings illustrate an unexpected interface between GrB and innate immunity beyond the traditional role of GrB in perforin–GrB apoptosis that may have broad mechanistic implications for a wide range of human disease.
Materials and Methods
Cell culture
Human submandibular gland (HSG), A431, human skin fibroblast cell, and HEK293 cell lines were cultured in DMEM plus 10% FBS and maintained under standard tissue culture conditions. Cells were plated in 24-well tissue culture plates at 200,000 cells/cm2 and grown to 80–90% confluence before use.
Synthesis and purification of wild-type and mutant inactive recombinant GrB
Except for studies directly comparing the effects of wild-type and mutant inactive GrB (GrBS183A), the recombinant GrB used was a gift from Dr. N. Thornberry (Merck Research Labs, Rahway, NJ). This recombinant GrB preparation was synthesized and purified from a baculovirus expression system, and the purity of the final product was confirmed by mass spectrometry, as previously described (45, 46).
For studies directly comparing the effects of wild-type and mutant inactive GrBS183A, recombinant protease was cloned, produced, and purified as follows: the coding sequence of mature GrB (aa 21–247) was amplified by PCR using cDNA generated from human PBMCs. Primers were designed to introduce an enterokinase cleavage site upstream of GrB and to remove the stop codon. The PCR product was cloned into AbVec2.0-IGHG1 (plasmid number 80795; Addgene; provided as a gift from Dr. E. Meffre, Yale University) with the Ig gene signal peptide sequence. The sequence containing the Ig signal peptide, enterokinase, and GrB sequences was subcloned into pcDNA3.1(+) (catalog no. V79020; Invitrogen) in frame with the sequence encoding a C-terminal 6× His-tag.
The GrBS183A mutant was derived from this wild-type construct by changing the active site serine residue at position 183 for alanine using the Q5 Site-Directed Mutagenesis Kit (E0554S; New England Biolabs). Both wild-type GrB and GrBS183A
S183A were eluted with buffer containing 250 mM imidazole.Sequences of the primers for subcloning wild-type GrB from PBMCs are provided in Table I. Sequences for the mutagenesis primers used to generate GrBS183A are also provided in Table I.
Purification of human NK cell cytotoxic granule contents
Cytotoxic granule contents (GCs) were isolated from the YT human NK cell line as previously described (47). The specific preparation of YT GCs used for studies in this manuscript was described previously (48).
Cell culture treatments
Cultures were treated with recombinant GrB, GrB with purified perforin (Enzo Life Sciences) or with purified YT NK cell cytotoxic GCs using a modified version of the protocol previously described (15, 48, 49). HBSS was substituted with serum-free DMEM. Unless stated otherwise in figure legends, cells were preincubated on ice with GrB or concentrated and purified GCs in calcium-free, serum-free DMEM for 15 min to maximize contact between GrB or GCs and cells, respectively. Treatment was then initiated by adding CaCl2 and followed immediately by incubation at 37°C for the times specified in the figure legends. Final cell treatment conditions were 75 nM GrB (unless otherwise stated in specific figure legends) or a 1:5 dilution of GCs in serum-free DMEM containing 2 mM CaCl2. Reactions were terminated by the addition of gel application buffer followed by boiling of the resulting lysate.
For treatments performed with the kinase inhibitors KU-55933 and NU-7026, the GrB protease inhibitor Compound 20 or the group II caspase inhibitor Z-DEVD-FMK, cells were first preincubated with these reagents for 1 h in culture medium (DMEM with 10% FBS) to allow cell uptake before treating with GrB. Concentrations for each inhibitor are specified in the respective figure legends. GrB treatment in the presence of the respective inhibitor was then performed. Compound 20 has been described previously (50, 51). KU-55933 and NU-7026 were purchased from Sigma-Aldrich. Z-DEVD-FMK was purchased from Santa Cruz Biotechnology.
Trypsin removal of membrane-bound GrB
After treatment with GrB or GCs, cells were washed three times with PBS, followed by incubation with 0.25% trypsin/EDTA (25200056; Life Technologies) for 10 min. Cells were then washed again three times with PBS and then lysed by addition of gel application buffer followed by boiling.
DAPI live–dead analysis by flow cytometry
After cell culture treatments, DAPI live–dead analysis was performed by flow cytometry. Cells were harvested by collecting the medium and then detaching the adherent fraction with an equal volume of 0.25% trypsin in EDTA. The medium and detached cells were pooled, then diluted 1:1 with DMEM containing 10% FCS. Samples were stained with DAPI per the manufacturer’s protocol (564907; BD Pharmingen). Flow cytometry for DAPI staining was performed using a BD Pharmingen FACSAria III instrument. A minimum of 40,000 events was acquired for each sample. Data analysis was performed using FCS Express Software, version 5 (DeNovo Software).
Immunoblotting
Gel samples were electrophoresed on SDS–polyacrylamide gels followed by transfer of proteins to nitrocellulose membranes for immunoblotting. Equal amounts of lysate protein or equal cell numbers were loaded in each gel lane. Primary Abs and concentrations/dilutions used for immunoblotting are listed in Table II. Blots performed with anti-vinculin or GAPDH Ab were used as loading controls. Proteins were detected using HRP-labeled secondary Abs (Pierce Biotechnology, Rockford, IL) and chemiluminescence. Images were acquired using a ProteinSimple FluorChem M digital imager. Densitometry was performed on images using the ProteinSimple AlphaImager analysis software.
Immunofluorescence microscopy
HSG cells were fixed and stained as previously described (52). For cleaved caspase-7 staining, anti–cleaved caspase-7 primary Ab (catalog no.8438; Cell Signaling Technology) was used at a final concentration of 0.14 μg/ml. Fluorescence microscopy images were acquired using a Zeiss Axioskop 50 with a Zeiss AxioCam HRc camera and AxioVision 4.9.1 software.
Assays to exclude postlytic GrB-induced phosphorylation events
The cell harvest protocol described above for terminating treatment reactions was first applied to untreated cells. Gel application buffer was added directly to cells followed by boiling. This lysate was then incubated with or without GrB on ice for 15 min followed by incubation at 37°C for 90 min, identical to the treatment procedure used on intact cells. Reactions were then terminated by boiling again.
Data replication
All experiments were repeated between two and eight times with reproducible results on each occasion. Data shown in the figures are representative of reproducible findings.
Results
Treatment with GrB in the absence of perforin induces substoichiometric intracellular proteolytic events without causing cell death
To study the consequences of perforin-independent GrB effects on target cells, cultures were treated for 90 min with GrB in the absence of perforin. As expected, flow cytometry live–dead determinations performed on intact HSG cells treated with GrB showed no appreciable induction of death relative to their untreated counterparts (Fig. 1). Additionally, no evidence of karyopyknosis and karyorrhexis, the nuclear morphologic hallmarks of apoptosis, was observed by immunofluorescence microscopy of DAPI-stained HSG cells treated with GrB alone (Supplemental Fig. 1A). The purity of the specific recombinant GrB preparation used for these experiments was established in a previous study by mass spectrometry (45).
GrB without perforin does not cause cell death. DAPI live–dead quantification by flow cytometry was performed on untreated HSG cells or HSG cells treated with 75 nM GrB for 90 min. HSG cells heated to 72°C for 5 min are shown as a positive control for the detection of dead cells. Live cell percentage summary statistics (means ± SDs) are shown for n = 3 untreated and n = 3 GrB-treated biologic replicates.
Interestingly, cleavages of a broad array of intracellular proteins, some of which are known GrB and caspase substrates, were still detected by immunoblotting in cells incubated with GrB in this manner albeit at substoichiometric levels. Inhibitor of caspase-activated DNase (ICAD), caspase-7, and caspase-8, are shown as representative examples (Fig. 2A). Caspase-3, DDX21, topoisomerase-IIα, and NuMA are additional examples (Supplemental Fig. 1B). As expected, cleavage of factors associated with GrB-mediated apoptosis (like the group II caspases and PARP) occurred only at extremely low substoichiometric levels compared with cells that were treated with lethal doses of purified NK cell cytotoxic GCs, as previously described (Supplemental Fig. 1D, 1E) (15). Caspase-8 maturation under these conditions was also incomplete, with only the p43/41 intermediate form being induced by extracellular GrB exposure and no p26/25 or fully matured p18 forms being generated (Supplemental Fig. 1C).
Perforin is not required for GrB to induce cleavage of intracellular proteins in intact cells. (A) Immunoblots showing perforin-independent cleavages of caspase-8, caspase-7, and ICAD in HSG cells incubated in the absence or presence of 75 nM GrB for the specified times. Gel samples prepared from equal numbers of cells were loaded in each lane. The Abs used for caspase-8, caspase-7, and ICAD were specific for the cleavage fragments and do not recognize the intact protein. The migration of protein molecular mass marker standards (in kiloDaltons) is shown on the left. The asterisk denotes GrB-dependent cleavage fragments. (B) GrB promotes cleavage of intracellular proteins exclusively in intact cells. HSG cell lysates (“Lysate”) or intact cells (“Cells”) were incubated with or without 75 nM GrB for 90 min. Samples were then electrophoresed and immunoblotted for cleaved caspase-8, cleaved caspase-7, and cleaved ICAD, as in (A). An immunoblot for recombinant GrB treatment input is also shown. (C) Incubation of intact cells with GrB induces cleavage events in a dose-dependent manner. HSG cells were incubated with increasing concentrations of GrB for 90 min. GrB concentrations of 8.33, 25, and 75 nM, respectively, were used. Immunoblots for cleaved caspase-8, cleaved caspase-7, and cleaved ICAD are again presented. (D) Intact GrB enzymatic activity is required to induce intracellular protein cleavages. HSG cells were treated either with 75 nM of wild-type GrB, GrBS183A, or with wild-type GrB and 100 μM Compound 20 and then immunoblotted for cleaved intracellular proteins, as in (A)–(C). (E) Immunofluorescence microscopy of HSG cells incubated in the absence or presence of GrB, as in Fig. 1, followed by fixation and staining for cleaved caspase-7. Representative original magnification ×20 images are shown. Scale bar, 20 μm.
Still, the substoichiometric cleavage events detected in cells exposed to GrB without perforin were notable for the rapidity of their induction and an absolute dependence on the protease activity of GrB. Fragments of NuMA, DDX21, topoisomerase-IIα and the caspases all were detected within 15 min of treatment (Fig. 2A, Supplemental Fig. 1B). ICAD cleavage fragments were also detected shortly thereafter by 45 min of treatment (Fig. 2A, bottom panel). Importantly, these cleavage events were induced exclusively only in intact cells exposed to GrB and were not postlytic artifacts occurring in harvested cell lysates (Fig. 2B). Treatment of lysates with GrB specifically could not recapitulate the induction of these fragments. Their induction in intact cells also occurred in a clear dose-dependent manner of GrB exposure (Fig. 2C). To establish the dependence of these events specifically on the protease activity of GrB, we compared the ability of wild-type GrB and the well-characterized enzymatically inactive mutant GrBS183A (see Table I for cloning primers) to induce fragment formation (53). We also examined the ability of wild-type GrB to induce fragment formation in cells coincubated with the cell-permeable GrB inhibitor Compound 20 (50, 51). No cleavage fragments were detected in HSG cells treated with the enzymatically inactive mutant and in cells treated with GrB inhibitor (Fig. 2D). To understand whether the GrB protease was inducing these effects in a specific subset of cells or otherwise, we next performed immunofluorescence microscopy on cultures treated with GrB using staining of cleaved caspase-7 as a readout of GrB activity (Fig. 2E). GrB exposure led to intracellular caspase-7 cleavage in all cells throughout the treated culture and was not restricted to any specific discernible cell subset.
The blockade of these cleavages when cells were coincubated with Compound 20, which is cell permeable, suggested that GrB protease activity could be relevant intracellularly, even in the absence of perforin, and that cell uptake of GrB was at least one potential mechanism for these proteolytic events. Cell uptake of GrB in the absence of perforin was also suggested by the observation that trypsin treatment could not remove GrB (Supplemental Fig. 1F). It is also plausible that these cleavages could be facilitated secondarily and in parallel through some extracellular GrB cleavage events transduced intracellularly by caspase activation. Regardless, the existence of intracellular effects mediated by GrB enzymatic activity and the rapidity with which they occur prompted us to explore whether functionally significant downstream signaling occurs in a perforin-independent manner.
Sublethal group II caspase activation drives downstream DDR signaling
The cleavages of group II caspases and ICAD induced by incubation with GrB in the absence of perforin suggested that caspase-activated DNase (CAD)–mediated DNA damage could be one functional outcome of this treatment. Nonlethal DNA damage initiated by substoichiometric group II caspase activation of CAD is increasingly recognized as an important physiologic second messenger (54–56). We suspected that extracellularly applied GrB could be an extrinsic immune mechanism for activating intracellular signaling using DNA damage as a second messenger.
To explore this possibility, we assayed HSG cells exposed to GrB for evidence of perforin-independent DDR signaling. Robust representative DDR phospho-protein signals, such as γ-H2AX and phospho-Kap1 were detected after treatment with GrB without perforin. γ-H2AX levels increased >3-fold after as little as 45 min of treatment (Fig. 3A). Similarly, GrB treatment increased p-Kap1 >6-fold after 45 min (Supplemental Fig. 2A). Interestingly, although perforin is not required for these effects, its presence did not inhibit and possibly even enhanced the ability of GrB to induce such phosphorylation events (Kap-1 in particular) (Supplemental Fig. 2B). We observed a similar GrB induction of DDR phospho-protein markers in human skin fibroblasts and in A431 cells (Supplemental Fig. 2C), indicating that the effect is observed widely in different cell types. To provide additional context, we compared the induction of γ-H2AX by GrB to that induced by the genotoxins camptothecin and etoposide. γ-H2AX induction by GrB matched the level induced by 1 μM camptothecin and exceeded that induced by 5 μM etoposide after 90 min of treatment (Fig. 3B).
GrB protease activity induces DNA damage in a perforin-independent manner. (A) Time course of γ-H2AX induction in untreated HSG cells (−) or HSG cells treated with 75 nM GrB (+). (B) Comparison of γ-H2AX induction by GrB and other genotoxic agents. Gel samples were prepared from untreated HSG cells or cells treated with 75 nM GrB, 1 μM camptothecin (CPT), or 5 μM etoposide (Etopo) for 90 min. (C) Intact GrB enzymatic activity is required for γ-H2AX induction. HSG cells were treated either with 75 nM of wild-type GrB, GrBS183A, or with wild-type GrB and 100 μM Compound 20 and then immunoblotted for γ-H2AX. (D) GrB enzymatic activity induces γ-H2AX formation in a dose-dependent manner. Treated HSG cells were incubated with increasing concentrations of wild-type GrB or GrBS183A for 90 min. GrB concentrations of 8.33, 25, and 75 nM, respectively, were used. (E) Group II caspase inhibition blocks GrB induction of DNA damage. HSG cells were treated with 75 nM GrB for 90 min in the absence or presence of 20 μM Z-DEVD-FMK (irreversible group II caspase inhibitor). Gel samples made from these cells were immunoblotted for γ-H2AX. For (A)–(E), equal amounts of lysate made from each cell culture were loaded in each gel lane. Densitometry relative to the signal in the untreated sample is shown for γ-H2AX in all panels. Vinculin was immunoblotted as a loading control for all.
Like the cleavage events described in Fig. 2A, DDR phosphorylation signals were also dependent on GrB protease activity exerted on intact cells. No γ-H2AX formation was observed in cells treated with enzymatically inactive mutant GrB (GrBS183A), nor was γ-H2AX induced in cells cotreated with wild-type GrB and Compound 20 (Fig. 3C). Wild-type GrB with intact enzymatic activity induced γ-H2AX formation in a dose-dependent manner, but GrBS183A was unable to do so (Fig. 3D). GrB was able to induce this γ-H2AX formation only in intact cells and not in HSG samples treated with GrB after lysis (Supplemental Fig. 3A), confirming that this phosphorylation event was not a postlytic artifact. In aggregate, these data establish that treatment of intact cells with GrB alone has previously unrecognized intracellular effects as an inducer of sublethal genotoxic stress and as a mediator of downstream DDR signaling.
We next examined whether GrB-induced DDR signaling proceeded through substoichiometric group II caspase activation. Evidence of ICAD cleavage was apparent in cells exposed to GrB (Fig. 2A). ICAD is cleaved by group II caspases, which results in ICAD inactivation and disinhibition of CAD (57, 58). Group II caspase inhibition, therefore, would be predicted to impair GrB induction of DDR signaling if indeed GrB were acting via the ICAD/CAD axis to induce DNA damage. Consistent with this prediction, cotreatment with the irreversible group II caspase inhibitor Z-DEVD-FMK blocked γ-H2AX induction in HSG cells treated with GrB (Fig. 3E). Specifically, GrB induction of γ-H2AX was reduced by at least ∼80%.
Cleavage and activation of DNA-PKcs and ATM are features of DDR signaling in response to perforin-independent GrB exposure
CAD induces primarily blunt-ended DNA double-strand breaks. DNA-PKcs and ATM are the PI3K-like kinase (PIKK) family members that orchestrate cell signaling in response to blunt-ended double-strand breaks. H2AX and Kap-1 are representative of the hundreds of redundant substrates for both kinases in this network (59–61). We therefore sought to understand whether and how perforin-independent GrB exposure might lead to the activation of DNA-PKcs, ATM, or both.
Because autophosphorylation is a standard marker of kinase activation for both ATM and DNA-PKcs, we immunoblotted samples from HSG cells incubated with GrB in the absence of perforin to test for autophosphorylation. GrB-dependent autophosphorylation of DNA-PKcs was clearly detected when we immunoblotted with an Ab against S2056 autophosphorylated DNA-PKcs (Fig. 4A, left panel). GrB-dependent autophosphorylation of ATM was detected using an Ab that recognizes ATM autophosphorylation at S1981 (Fig. 4B, left panel). Induced autophosphorylation of these kinases was not observed in HSG samples treated with GrB after lysis (Supplemental Fig. 3B, 3C). This latter result confirmed that these phosphorylation events, like the GrB-dependent cleavages and formation of γ-H2AX previously described, occur only in intact cells. The induction of DNA-PKcs and ATM autophosphorylation events also was entirely dependent on intact GrB protease activity in a dose-responsive manner with the enzymatically inactive GrBS183A mutant being completely unable to promote either signal (Fig. 4C, 4D).
Substoichiometric cleavages of DNA-PKcs and ATM associate with kinase activation in cells exposed to GrB without perforin. (A) GrB induces autophosphorylation of cleaved DNA-PKcs. HSG cells were incubated in the absence or presence of 75 nM GrB for 90 min. Equal protein amounts of lysates made from these cells were blotted with Abs against autophosphorylated (phospho) or total DNA-PKcs (using either an anti–N terminus (N-term) or anti–C terminus (C-term) Ab). For phospho and N-term immunoblots, samples were electrophoresed as replicates on the same gel, transferred to the same nitrocellulose membrane, and probed with the relevant Abs. (B) GrB induces autophosphorylation of cleaved ATM. Immunoblots are shown for autophosphorylated (Phospho-ATM; left panel) and total ATM (right panel) in untreated HSG cells or in cells incubated with 75 nM GrB. Samples were electrophoresed as replicates on the same gel, transferred to same nitrocellulose membrane, then probed with phospho-ATM or ATM Abs. (C) GrB protease activity induces DNA-PKcs phosphorylation in a dose-dependent manner. Treated HSG cells were incubated with increasing concentrations of wild-type GrB or GrBS183A for 90 min, as in Fig. 3D. GrB concentrations of 8.33, 25, and 75 nM, respectively, were used. (D) GrB protease activity also induces ATM phosphorylation in a dose-dependent manner. HSG cells treated with wild-type GrB or GrBS183A, as in (C) were immunoblotted for phospho-ATM. (E) GrB-induced autophosphorylation of cleaved DNA-PKcs requires group II caspase activity. HSG cells were incubated in the absence or presence of 75 nM GrB for 90 min with or without 20 μM Z-DEVD-FMK. (F) GrB-induced autophosphorylation of cleaved ATM also requires group II caspase activity. HSG cells were treated with or without GrB and Z-DEVD-FMK, as in (E). For (A)–(F), vinculin was immunoblotted as a loading control for all. DNA-PKcs and ATM fragments are denoted by the asterisk.
Interestingly, autophosphorylated DNA-PKcs and ATM migrated at lower molecular mass than expected on gel electrophoresis. DNA-PKcs has a predicted molecular mass of 460 kDa. The Ab used to detect S2056 autophosphorylated DNA-PKcs, however, showed induction of a 250 kDa band (Fig. 4A, left panel). Similarly, ATM has a predicted molecular mass of 350 kDa, but the GrB-induced species detected by the Ab specific for ATM autophosphorylation at S1981 migrated at slightly lower than 250 kDa (Fig. 4B, left panel). We suspected that these lower molecular mass autophosphorylated forms represented proteolytically cleaved forms and that cleavage could be linked mechanistically to kinase activation downstream of GrB exposure. Both DNA-PKcs and ATM are known group II caspase cleavage substrates during perforin–granzyme apoptosis (15, 62). DNA-PKcs is also directly cleaved by GrB during perforin–granzyme apoptosis (15), suggesting that DNA-PKcs could also be a direct GrB substrate during perforin-independent GrB exposure.
Indeed, immunoblotting with an Ab against the N terminus of DNA-PKcs (Table II) detected a 250 kDa DNA-PKcs fragment that comigrated precisely with the induced autophosphorylated band (Fig. 4A, middle panel). Formation of this fragment also required intact GrB protease activity (Supplemental Fig. 4A, top panel). A second DNA-PKcs Ab directed against the C terminus of the kinase (Table II) detected a reciprocal 150 kDa cleavage fragment (Fig. 4A, right panel). To definitively establish that autophosphorylation was also mechanistically linked to GrB-induced caspase activity, we examined the roles of GrB enzymatic activity and group II caspase activity on the induction of both. No DNA-PKcs fragment formation occurred in HSG cells incubated with GrBS183A (Supplemental Fig. 4A, top panel). Similarly, Z-DEVD-FMK treatment significantly reduced both GrB-dependent induction of the 250 kDa phosphorylated DNA-PKcs fragment (Fig. 4E) and cleavage (Supplemental Fig. 4B, top panel). Cleaved, autophosphorylated DNA-PKcs formation, therefore, required both GrB and group II caspase enzymatic activities.
Similar analyses for ATM in HSG cells treated with GrB showed that ATM activation was also functionally coupled to cleavage status. The ∼250 kDa ATM species autophosphorylated at S1981 comigrated with a proteolytic fragment detected by another Ab recognizing ATM regardless of phosphorylation status (Fig. 4B). Here too, autophosphorylation, GrB protease activity, and caspase-mediated cleavage appeared to be linked mechanistically. ATM cleavage was absent in HSGs treated with the GrBS183A mutant (Supplemental Fig. 4A, bottom panel). Induction of both ATM autophosphorylation (Fig. 4F) and fragment formation (Supplemental Fig. 4B, bottom panel) was also inhibited by cotreatment with Z-DEVD-FMK, consistent with both ATM cleavage and kinase activation requiring GrB-induced group II caspase activation. In contrast, baseline pretreatment autophosphorylation levels of full-length 350 kDa ATM were unaffected by GrB exposure.
Prior studies of GrB and group II caspase interactions with PIKK family members have reported conflicting results on whether cleavage promotes or inhibits ATM and DNA-PKcs kinase activities (15, 62–65). To further clarify the mechanistic relationship between PIKK cleavage and kinase activity during perforin-independent GrB exposure, we examined the ordering of PIKK cleavage and the induction of DDR phospho-protein signals in GrB-treated HSG cells. We reasoned that if PIKK cleavage inhibited kinase activity in GrB-treated cells, then fragment formation would be expected to occur after kinase autophosphorylation, and γ-H2AX and phospho–Kap-1 levels should decrease as ATM and DNA-PKcs fragments accumulated in GrB-treated cultures.
We found, however, that cleavage did not preclude kinase activation or the phosphorylation of downstream PIKK substrates. Instead, DNA-PKcs cleavage was detected prior to DNA-PKcs autohosphorylation and increasing γ-H2AX formation (Supplemental Fig. 4C). ATM cleavage and phosphorylation were coincident with increasing γ-H2AX formation (Supplemental Fig. 4D). These observations, together with the clear functional association between autophosphorylation and cleavage status, support proteolysis being a mechanism for activating PIKKs downstream of GrB exposure.
DNA-PKcs and ATM activation downstream of GrB and group II caspases leads to IRF-3 phosphorylation
DNA damage induces innate immune responses in epithelial cells via PIKK-dependent recruitment and activation of transcription factors such as IRF-3 (66–68). We wondered whether group II caspases, together with ATM and DNA-PKcs, could act as second messengers in a perforin-independent pathway by which GrB could promote IRF-3 activation. Consistent with this hypothesis, HSG cells treated with GrB, independent of perforin, showed prominent phosphorylation of IRF-3 at S386, a standard marker of DNA damage-induced IRF-3 activation (Fig. 5A). Like the other phosphorylation events already described, we found that this IRF-3 phosphorylation also required intact GrB protease activity. Neither GrBS183A nor wild-type GrB treated with the cell-permeable inhibitor Compound 20 could catalyze this posttranslational signaling modification of IRF-3. Also, like the other GrB-induced phosphorylation events that we characterized, GrB could only promote IRF-3 phosphorylation in intact cells (Supplemental Fig. 3D). Although the activation of this pathway is clearly perforin-independent, the presence of perforin enhanced GrB induction of IRF-3 phosphorylation (Supplemental Fig. 3E).
GrB mobilizes group II caspases and PIKK kinase activity to induce IRF-3 phosphorylation in a perforin-independent manner. (A) GrB induces IRF-3 phosphorylation in a manner dependent on its protease activity. HSG cells were treated with wild-type GrB, GrBS183A, or wild-type GrB and Compound 20 as in Figs. 2D and 3C followed by immunoblotting for IRF-3 phosphorylation at S386 (p-IRF-3). (B) Group II caspase inhibition blocks GrB induction of IRF-3 phosphorylation. HSG cells were treated with GrB in the absence or presence of 20 μM Z-DEVD-FMK as in Fig. 3E. Gel samples made from these cells were immunoblotted for p-IRF-3 as in (A). (C) Cooperative induction of IRF-3 phosphorylation by GrB and group II caspases requires PIKK kinase activity. Immunoblot is shown of IRF-3 phosphorylation (S386) in HSG cells incubated with 75 nM GrB for 90 min in the absence or presence of inhibitors for ATM kinase KU-55933 (ATMi) or the DNA-PKcs kinase NU-7026 (PKi).
We next examined the dependence of GrB activation of IRF-3 on group II caspase and PIKK family member activities. Using HSG cells cotreated with GrB and Z-DEVD-FMK, we found that group II caspase inhibition blocked GrB induction of IRF-3 phosphorylation (Fig. 5B). Treatment with kinase inhibitors of ATM (KU-55933) and DNA-PKcs (NU-7026) also blocked induction of IRF-3 phosphorylation downstream of GrB (Fig. 5C). The effects observed with both kinase inhibitors thereby indicated that IRF-3 phosphorylation was a DDR-associated process dependent on both GrB-induced ATM and DNA-PKcs kinase activities. Taken together, our observations are consistent with the existence of a sophisticated pathway by which GrB can interface with an innate immune response program in a perforin-independent manner. This pathway is initiated by GrB enzymatic activity, proceeds through substoichiometric group II caspase activation, and uses apical DDR signaling events, such as ATM and DNA-PKcs activation, as unexpected second messengers.
Discussion
GrB studies historically have focused primarily on its perforin-dependent role in the granule-induced apoptotic pathway. More recent evidence indicates that GrB also exerts perforin-independent functions unrelated to cell death that are mediated by GrB proteolytic events in the ECM, cell surface, or basement membrane (20, 39–44, 69, 70). The current work is a proof-of-concept study that complements these findings and extends the range of known perforin-independent activities by demonstrating that GrB also has the ability to regulate intracellular signaling through broad perforin-independent intracellular proteolytic and phosphorylation events.
Consistent with previous reports, we did not find cell death to be a prominent outcome in cell cultures treated with GrB in the absence of perforin (Fig. 1) (16, 17, 41, 71). However, immunoblotting data showed clearly that GrB exposure still promoted cleavage of several intracellular proteins (e.g., group II caspases, caspase-8, ICAD, NuMA, DDX21, and topoisomerase-IIα, ATM, and DNA-PKcs) (Figs. 2A, 4A, 4B, Supplemental Fig. 1B) in treated cell cultures. Although the mechanisms mediating these perforin-independent proteolytic effects of GrB in cells are still not fully understood, our data exclude their occurrence as postlysis sampling artifacts. Specifically, GrB treatment of HSG cells after lysis could not recapitulate the induction of these fragments, which could only be induced when intact cells were incubated with GrB (Fig. 2B). GrB, specifically, was responsible for promoting these cleavage events, with intact GrB protease activity being absolutely required for their occurrence (Fig. 2D, Supplemental Fig. 4A).
GrB protease-dependent cleavage of group II caspases, specifically, was directly related functionally to the induction of several coordinated downstream signaling phosphorylation events occurring as part of a DDR in intact cells. Cotreatment with a group II caspase inhibitor that binds irreversibly to the active forms of caspase-3 and -7 (Z-DEVD-FMK) blocked GrB induction of ATM and DNA-PKcs autophosphorylation (Fig. 4E, 4F) as well as the direct downstream phosphorylation of representative PIKK substrates such as H2AX (Fig. 3E). Pharmacologic inhibition of group II caspase activity also blocked the GrB induction of PIKK-dependent IRF-3 phosphorylation (Fig. 5B). Similar to what was observed with upstream substoichiometric cleavage events, treatment of lysates directly with GrB could not recapitulate induction of any of these phosphorylation signaling events, confirming that GrB promotes these specifically only in intact cells (Supplemental Fig. 3).
The possibility of GrB regulating intracellular signaling in a perforin-independent manner without promoting cell death has been introduced previously. Merkulova and colleagues (41) showed that perforin-independent GrB treatment alters epidermal growth factor receptor (EGFR) kinase-dependent signaling in keratinocytes. Extracellularly applied GrB modified this signaling without directly cleaving either the receptor or its ligand, and the mechanism of these perforin-independent GrB signaling effects remains unknown. In yet another example, extracellular GrB was recently found to regulate airway epithelia IL-25 expression via a similar cell surface interaction with protease-activated receptor 2 (70). Notably, however, these studies did not investigate roles for intracellular substrate proteolysis in perforin-independent GrB-mediated effects.
In these previous works as well as in the current study, GrB could be inducing perforin-independent proteolytically mediated intracellular signaling via at least two different mechanisms that are not necessarily mutually exclusive. Because group II caspases and GrB often cleave the same substrate at sites in close proximity (typically 5–30 aa apart) or even at the same sites (15), we were unable to determine from fragment size alone whether the cleavages we detected during perforin-independent treatment were due to direct GrB proteolysis or secondary caspase-mediated proteolysis. Yet, despite the long-held view that perforin is essential for functionally relevant intracellular GrB accumulation (19–23), GrB could still be accumulating intracellularly via endocytosis or some other unknown perforin-independent mechanism(s) of cell entry (72, 73). Notably, the cell-permeable GrB inhibitor Compound 20 inhibited all intracellular cleavage events as well as the induction of all downstream DDR signaling and IRF-3 phosphorylation (Figs. 2D, 3C, 5A). Furthermore, adding perforin to GrB to facilitate GrB cell entry (Supplemental Figs. 2B, 3E) only enhanced DDR signaling and IRF-3 phosphorylation. Finally, GrB was resistant to removal from cells by trypsinization, which has previously been shown to remove externally bound GrB (Supplemental Fig. 1F) (72). Taken together, these observations provide support for GrB cell entry, perforin-mediated or otherwise, as at least one plausible mechanism for GrB mediating these signaling effects.
A substoichiometric proteolytic interaction with gasdermin E, an endogenous perforin-like molecule originally described as a key pyroptotic factor, may contribute to facilitating GrB entry when perforin is absent. This possibility will be the focus of future investigations. Perforin–GrB–secreting CTLs were recently shown to induce target cell pyroptosis during antitumor immune responses via GrB and group II caspase cleavage of gasdermin E, leading to polymerization and lethal pore formation in the cell membrane in a manner similar to perforin (74). In the absence of perforin, GrB may still potentially facilitate substoichiometric cleavage of gasdermin E and other family members. This could lead to the formation of pores or arcs below a lethal threshold but still sufficient to facilitate substoichiometric GrB cleavage and activation of group II caspases.
GrB also could be simultaneously proteolyzing one or more extracellular or cell surface substrates, which could then activate group II caspases as intracellular second messengers. GrB has been shown to induce intracellular caspase activation without requiring accumulation in intracellular compartments during anoikis, a type of apoptosis induced by detachment from the ECM because of cleavage of ECM component substrates (40, 75). Although cell death is not the predominant outcome of perforin-independent GrB treatment, low-level cleavage of ECM components below the lethal anoikis threshold could still conceivably contribute to inducing substoichiometric group II caspase activity and produce functionally significant intracellular signaling effects. Exploring such potential mechanisms will also be the topic of future studies.
Regardless, perforin-independent proteolytically mediated intracellular signaling pathways could function simultaneously in vivo with extracellular GrB activities to prime and enhance inflammatory responses in the vicinity of perforin–granzyme killing events. Whereas granule-mediated apoptosis exerts acute effects within minutes, nonlethal perforin-independent GrB activities conceivably may superimpose more chronic and long-lasting cell and even tissue-level effects. GrB can persist in a stable bioactive form even after the rapid extinction of perforin activity (which is exquisitely calcium-dependent) after CTL or NK cell degranulation (20, 39–44). Several other immune cell types express and secrete GrB without perforin, including specialized CD4 regulatory T cells and CD4 memory cells, plasmacytoid dendritic cells, various granulocytes, mast cells, and regulatory B cells (33–36, 39, 69, 76). GrB secretion by these effector cells could also have functional significance in tissues.
Additional studies are required to identify other intracellular signaling pathways regulated by GrB protease activity independent of perforin and to understand the potential relevance of such pathways in vivo. Many of these pathways, such as IRF-3 activation, are likely to be downstream of group II caspase and PIKK activation. Substoichiometric caspase activity and DDR signaling at sublethal levels has previously been linked to regulation of epithelial cell migration/motility and compensatory cell proliferation. Group II caspases and DDR signaling also have been implicated in preserving and even enhancing the pluripotency of several stem cell types, including myoblasts, melanocyte stem cells, and hematopoietic stem cells (77, 78). DNA damage has also been shown to induce increased mesenchymal and epithelial cell motility (79). In future studies, it will be important to elucidate whether GrB can function as an extrinsic regulator of these effects and others. If these effects occur in vivo, it will be relevant to define whether and how GrB expression could be used as a biomarker for grading and subtyping disease activities and the potential use of GrB as a broadly relevant therapeutic target.
Disclosures
The authors have no financial conflicts of interest.
Acknowledgments
We thank Dr. Eric Meffre (Yale University) for providing plasmid AbVec2.0-IGHG1 as a gift. We thank Dr. Nancy Thornberry (Merck Research Labs, Rahway, NJ) for providing recombinant GrB as a gift.
Footnotes
This work was supported in part by National Institutes of Health, Office of Extramural Research Grants R01 DE 12354-15A1 and R01AR069569 and National Institutes of Health Grant T32AR048522 and by the Rheumatology Research Foundation. D.J.G. was supported by grants-in-aid from the Canadian Institutes for Health Research and the Michael Smith Foundation for Health Research. E.J.G. and E.T. are Jerome L. Greene Foundation Scholars. E.J.G. was also supported in part by a Vasculitis Foundation research award sponsored in loving memory by the family of the late Dr. Darwin James Liao. L.C.-R. was supported in part by the Stabler Foundation. The manuscript content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health or the Canadian Institutes for Health and Research.
The online version of this article contains supplemental material.
Abbreviations used in this article:
- CAD
- caspase-activated DNase
- DDR
- DNA damage response
- DEVD
- aspartic acid–glutamic acid–valine–aspartic acid
- ECM
- extracellular matrix
- GC
- granule content
- GrB
- granzyme B
- HSG
- human submandibular gland
- ICAD
- inhibitor of caspase-activated DNase
- IRF
- IFN regulatory factor
- PIKK
- PI3K-like kinase.
- Received May 13, 2020.
- Accepted November 11, 2020.
- Copyright © 2021 by The American Association of Immunologists, Inc.