Network -Tubulin and Disorganizes the Microtubule Α NK Cell Protease

NK Cell Protease Granzyme M Targets α -Tubulin and Disorganizes the Microtubule Network

This information is current as Niels Bovenschen, Pieter J. A. de Koning, Razi Quadir, Roel of September 24, 2021. Broekhuizen, J. Mirjam A. Damen, Christopher J. Froelich, Monique Slijper and J. Alain Kummer J Immunol 2008; 180:8184-8191; ; doi: 10.4049/jimmunol.180.12.8184

http://www.jimmunol.org/content/180/12/8184 Downloaded from

References This article cites 34 articles, 13 of which you can access for free at: http://www.jimmunol.org/content/180/12/8184.full#ref-list-1 http://www.jimmunol.org/

Why The JI? Submit online.

• Rapid Reviews! 30 days* from submission to initial decision

• No Triage! Every submission reviewed by practicing scientists

• Fast Publication! 4 weeks from acceptance to publication by guest on September 24, 2021

*average

Subscription Information about subscribing to The Journal of Immunology is online at: http://jimmunol.org/subscription Permissions Submit copyright permission requests at: http://www.aai.org/About/Publications/JI/copyright.html Email Alerts Receive free email-alerts when new articles cite this article. Sign up at: http://jimmunol.org/alerts

The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 1451 Rockville Pike, Suite 650, Rockville, MD 20852 Copyright © 2008 by The American Association of Immunologists All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. The Journal of Immunology

NK Cell Protease Granzyme M Targets ␣-Tubulin and Disorganizes the Microtubule Network1

Niels Bovenschen,* Pieter J. A. de Koning,* Razi Quadir,* Roel Broekhuizen,* J. Mirjam A. Damen,† Christopher J. Froelich,‡ Monique Slijper,† and J. Alain Kummer2*

Serine protease granzyme M (GrM) is highly expressed in the cytolytic granules of NK cells, which eliminate virus-infected cells and tumor cells. The molecular mechanisms by which GrM induces cell death, however, remain poorly understood. In this study we used a proteomic approach to scan the native proteome of human tumor cells for intracellular substrates of GrM. Among other findings, this approach revealed several components of the cytoskeleton. GrM directly and efficiently cleaved the actin-plasma membrane linker ezrin and the microtubule component ␣-tubulin by using purified proteins, tumor cell lysates, and tumor cells undergoing cell death induced by perforin and GrM. These cleavage events occurred independently of caspases or other cysteine proteases. Kinetically, ␣-tubulin was more efficiently cleaved by GrM as compared with ezrin. Direct ␣-tubulin proteolysis by Downloaded from GrM is complex and occurs at multiple cleavage sites, one of them being Leu at position 269. GrM disturbed tubulin polymerization dynamics in vitro and induced microtubule network disorganization in tumor cells in vivo. We conclude that GrM targets major components of the cytoskeleton that likely contribute to NK cell-induced cell death. The Journal of Immunology, 2008, 180: 8184–8191.

ytotoxic lymphocytes, i.e., CTLs and NK cells, are key proteins leads to DNA fragmentation and mitochondrial damage, http://www.jimmunol.org/ players in the effector arm of the immune response that respectively. GrA predominantly kills by cleaving nuclear (e.g., C eliminates virus-infected cells and tumor cells (1, 2). Cy- Ku70), mitochondrial, and cytoplasmic substrates (e.g., SET com- totoxic lymphocytes predominantly destroy their targets by releas- plex components) (2, 7–9). Cleavage of these substrates results in ing the content of their cytolytic granules. These granules contain single-stranded nicking of chromosomal DNA. perforin and a family of unique structurally homologous serine In contrast to GrA and GrB, far less is known about the other proteases known as granzymes (3, 4). Although perforin facilitates human granzymes. It has been demonstrated that granzyme M the entry of granzymes into the target cell, the latter induce cell (GrM), which is speciﬁcally expressed by NK cells, mediates a death by cleaving critical intracellular substrates (1, 2).

novel major and perforin-dependent cell death pathway with by guest on September 24, 2021 In humans, five different granzymes (GrA, GrB, GrH, GrK, and unique morphological hallmarks that plays a significant role in NK GrM) are known that differ on the basis of their substrate speci- cell-induced death (10). The molecular mechanism by which GrM ficity (3, 4). Over the past few decades, it has been well established 3 induces cell death remains unclear. One study has found that GrM- that granzyme A (GrA) and granzyme B (GrB) serve as important induced cell death occurs independently of caspases, DNA frag- determinants of cellular cytotoxicity. Both granzymes induce nu- mentation, and reactive oxygen species (ROS) generation (10), clear and non-nuclear damage in target cells by cleaving distinct whereas other recent reports have demonstrated the opposite (11, nonoverlapping sets of substrates (1, 2). Two important intracel- 12). This suggests that GrM targets multiple independent cell death lular substrates of GrB include procaspase 3 (5) and the small pathways, which has also been demonstrated for GrA and GrB (1, Bcl-2 homology domain 3-only protein Bid (6). Cleavage of these 2, 5–9). In the present study, we used a proteomic approach to define potential substrates of GrM. We report that GrM targets the *Department of Pathology, University Medical Center, Utrecht; †Department of Bio- cytoskeleton in tumor cells by cleaving the actin-plasma mem- molecular Mass Spectrometry, Utrecht University, Utrecht, The Netherlands; and brane linker ezrin and the microtubule component ␣-tubulin. This ‡Department of Medicine, Evanston Northwestern Healthcare Research Institute, Evanston, IL 60201 likely contributes to the mechanism and the specific morphological Received for publication July 25, 2007. Accepted for publication April 14, 2008. changes that coincide with GrM-mediated target cell death. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. Materials and Methods 1 This work was supported by the Netherlands Organization for Scientific Research Reagents Grant 916.66.044 (to N.B.), Dutch Cancer Society Grant UMCU-2004-3047 (to J.A.K.), and National Institutes of Health Grant R01 AI044941-07 (to C.J.F.). Abs were anti-␣-tubulin clone B-5-1-2 (Sigma-Aldrich), anti-ezrin clone 2 Address correspondence and reprint requests to Dr. J. Alain Kummer, Department 3C12 (Zymed Laboratories), anti-␤-actin clone 2A2.1 (United States Bio- of Pathology, University Medical Center, Heidelberglaan 100, 3584 CX, Utrecht, The logical), anti-caspase-3 clone H-277 (Tebu-bio), anti-GST tag (Santa Cruz Netherlands. E-mail address: [email protected] Biotechnology), and anti-His tag (BD Biosciences). E64, trans-epoxysuc- 3 Abbreviations used in this paper: GrA, granzyme A; GrB, granzyme B; GrM, gran- cinyl-L-leucylamido-(4-guanidino)butane, was from Sigma-Aldrich and zyme M; GrM-SA, GrM with S195A mutation in catalytic center; HSP, heat shock benzyloxycarbonyl-Val-Ala-Asp(OMe)-fluoromethylketone (z-VAD-fmk) protein; MAP, microtubule-associated protein; MS, mass spectrometry; PI, propidium was from Biomol. The chromogenic caspase-3 substrate Ac-Asp-Glu-Val- iodide; ROS, reactive oxygen species; Ac-DEVD-pNA, acetyl-Asp-Glu-Val-Asp-p- Asp-p-nitroaniline (Ac-DEVD-pNA) was from Bachem. Purified recom- nitroaniline; E64, trans-epoxysuccinyl-L-leucylamido-(4-guanidino)butane; z-VAD- fmk, benzyloxycarbonyl-Val-Ala-Asp(OMe)-fluoromethylketone. binant human GST-k-␣-1-tubulin was purchased from Cytoskeleton. Hu- man perforin was purified as described (13). Protein was quantified by the Copyright © 2008 by The American Association of Immunologists, Inc. 0022-1767/08/$2.00 Bradford method. www.jimmunol.org The Journal of Immunology 8185

Cell lines and cell-free protein extracts HeLa and Jurkat cells were grown in DMEM and RPMI 1640 medium, respectively, supplemented with 10% FCS, 0.002 M glutamine, 100 U/ml penicillin, and 100 ␮g/ml streptomycin (Invitrogen). Cell-free protein extracts were generated from exponentially growing HeLa and Jurkat cells. Cells (108 cells/ml) were washed two times in a buffer containing 50 mM Tris (pH 7.4) and 150 mM NaCl, and lysed in the same buffer by three cycles of freezing/thawing. This method gently disrupts the plasma membrane and minimally affects cell compartment integrity (14). Samples were centrifuged for 10 min at 14,000 rpm at 4°C and cell-free protein extracts were stored at Ϫ80°C. Recombinant proteins The cDNA encoding mature human GrM (residues Ile26–Ala257) was am- FIGURE 1. Identification of GrM-induced cleavage events in tumor cell plified from IMAGE clone 5222281 and cloned into yeast expression vec- lysates. HeLa cell freeze/thaw lysates were incubated with GrM-SA (1 tor pPIC9 (Invitrogen). Catalytically inactive GrM-SA, in which the Ser195 ␮M) (A) or GrM (1 ␮M) (B) for 60 min at 37°C. Proteins were separated residue in the catalytic center is replaced by Ala (S195A), was generated by two-dimensional gel electrophoresis (10%) and visualized by silver by site-directed mutagenesis (Stratagene). Plasmids were transformed into staining. Proteins that are reduced in abundance after GrM incubation rep- the GS115 (his4) strain of Pichia pastoris and granzymes were expressed resent potential GrM substrates (A) and new spots that appear during GrM in conditioned medium for 72 h as described by the manufacturer (Invitro- treatment are cleavage products (B). This experiment was performed three gen). GrM and GrM-SA were purified to homogeneity by cation-exchange times with similar results and the changed protein spots (n ϭ 37) were chromatography (GE Healthcare) using a linear salt gradient for elution. Downloaded from excised from two-dimensional gels. Protein spots that could be identified GrM preparations were dialyzed against 50 mM Tris (pH 7.4) and 150 mM ϭ NaCl and stored at Ϫ80°C. GrM, but not GrM-SA, was active as deter- (n 16) by liquid chromatography-tandem MS are indicated by white mined by a synthetic chromogenic leucine substrate (Bachem) (data not circles. shown). The human k-␣-1-tubulin cDNA was amplified from IMAGE clone 3871729, cloned into the bacterial expression vector pQE80L, and expressed as recommended by the manufacturer (Invitrogen). The L269A, Tubulin polymerization assay L286A, L286A/L269A, M302A, M302A/L269A, M313A, M313A/ http://www.jimmunol.org/ L269A, and L317A/L318A tubulin mutants were generated by site-directed A tubulin polymerization assay kit (Cytoskeleton) was used to address the ␣ mutagenesis. Recombinant His-␣-tubulin protein was purified by metal- effects of GrM on tubulin polymerization dynamics. A purified bovine - ␤ chelate chromatography (Clontech), dialyzed against PBS, and stored at and -tubulin preparation (3 mg/ml) virtually free of microtubule-associ- Ϫ80°C. Ezrin cDNA was from RZPD German Resource Center for Ge- ated proteins (MAPs) (Cytoskeleton catalog no. HTS02) in 80 mM PIPES nome Research. The pGEX-GST-ezrin bacterial expression construct, in (pH 6.9), 0.5 mM EGTA, and 2 mM MgCl2 was incubated with GrM (1 ␮ ␮ ␮ which the GST tag is fused to the N terminus of human ezrin, was provided M), GrM-SA (1 M), paclitaxel (Taxol) (5 M), or buffer in 20 mM Tris by Dr. H. Rehmann (UMC Utrecht, The Netherlands). Recombinant GST- (pH 7.0). With the exception of Taxol, samples were preincubated for 2 h ezrin was expressed and purified as described above for recombinant at 30°C. Microtubule polymerization was initiated by the addition of GTP His-␣-tubulin. (1 mM) and 5% (v/v) glycerol. Changes in microtubule turbidity were measured kinetically at 340 nm at 37°C (Anthos Labtec).

Two-dimensional gel electrophoresis and spot identification by by guest on September 24, 2021 mass spectrometry Results Washed HeLa cells (108 cells/ml) in 25 mM Tris (pH 8), 30 mM NaCl, and GrM-induced cleavage events in tumor cell lysates 1 mM DTT were subjected to three rounds of freeze/thaw lysis and cell- To define potential intracellular substrates of GrM, we used a pro- free extracts (50 ␮g) were incubated with GrM (1 ␮M) or GrM-SA (1 ␮M). After1hat37°C, samples were precipitated using the Plus One two- tease-proteomic approach. Because it is difficult to deliver large dimensional Clean-up kit as recommended by the manufacturer (GE amounts of GrM to all target cells via perforin, we used freeze/ Healthcare) and solubilized in 8 M urea, 2 M thiourea, 4% CHAPS, 20 mM thaw lysis of HeLa tumor cells to gently disrupt the plasma mem- DTT, 0.2% Biolyte (pH range 3–10), and 0.2% bromophenol blue (200 ␮l) brane and minimally alter the native proteome. These protein ex- ␮ for isoelectric focusing. Samples (50 g) were rehydrated passively into tracts were incubated with purified recombinant mature human 11-cm pH 3–10 immobilized pH gradient (IPG) strips for 15 h at room temperature before isoelectric focusing in the IPGphor system (GE Health- GrM or the catalytically inactive GrM-SA mutant and cleavage care) for 20 kVh. The IPG strips were reduced for 60 min in 2% (w/v) events were analyzed by two-dimensional gel electrophoresis (Fig. DTT, 6 M urea, 2% (w/v) SDS, 20% (v/v) glycerol, and 0.375 M Tris (pH 1). Spots present in greater abundance in the control sample indi- 8.8) and alkylated for 30 min in the same buffer containing 2% (w/v) cate possible GrM substrates, whereas spots present in greater iodoacetamide instead of DTT. Strips were mounted on 10% SDS-poly- acrylamide gels, proteins were separated, and two-dimensional gels were abundance in the GrM-treated sample reflect potential cleavage stained by mass spectrometry (MS)-compatible silver staining. Gel features products. Of ϳ1500 proteins that were resolved in this proteomic were evaluated by PDQuest 7.4 software and selected spots were excised screen, ϳ15 spots clearly disappeared (Fig. 1A) and ϳ22 spots robotically with a ProteomeWorks Spot Cutter (Bio-Rad). Gel cores were clearly appeared (Fig. 1B) following the incubation with GrM. destained and subjected to in-gel tryptic digestion. Peptide mixtures were These changes were highly reproducible and could also be de- applied to liquid chromatography-tandem MS (Finnigan LTQ) and the results were analyzed by MASCOT (www.matrixscience.com). tected when samples were labeled fluorescently and analyzed on the same two-dimensional gel using fluorescence two-dimensional GrM-mediated cell death difference gel electrophoresis (fl-2D-DIGE; data not shown). Spots Washed Jurkat cells (1 ϫ 106) were treated with GrM (1 ␮M) or GrM-SA were excised that exhibited high reproducibility and displayed Ͼ (1 ␮M) in the presence or absence of a sublytic dose of perforin (40 ng/ml) 3-fold changes in abundance following GrM treatment. We were in 50 mM HEPES (pH 7.4), 150 mM NaCl, 2.5 mM CaCl , and 1% (w/v) 2 able to identify 16 of 37 excised protein spots from two-dimen- BSA for 4 or 10 h at 37°C. Cells were washed in the same buffer and used for cytospins, propidium iodide (PI) flow cytometry, or direct lysis with sional gels using tandem MS (Table I). Several spots that consis- SDS-PAGE loading buffer. Cytospins were fixed with 96% (v/v) ethanol tently changed following GrM treatment could not be identified, for 10 min and stained with Giemsa or immunostained with an Ab against most likely because protein levels were too low. The identity of ␣-tubulin (clone B-5-1-2), followed by a tetramethylrhodamine isothiocya- potential GrM substrates that could be identified include a group of nate-conjugated Ab to visualize ␣-tubulin by confocal microscopy. For flow cytometry, cells were incubated with PI (46 ␮g/ml) for 10 min at room highly homologous proteins involved in chaperone systems and temperature. Cell viability after a 24-h incubation period with GrM/per- cellular stress response (i.e., heat shock protein (HSP) 90␤, endo- forin or GrM-SA/perforin was measured by trypan blue staining. plasmin, and protein disulfide isomerase), proteins involved in 8186 GRANZYME M TARGETS THE CYTOSKELETON

Table I. Overview of GrM-induced cleavage events identiﬁed by tandem mass spectrometrya

Theoretical Molecular Swiss-Prot Mass/Isoelectric Spot No. Protein Identity Accession No. Point Details

Chaperones/cell stress response: 1 Heat shock protein 90␤ (HSP90␤) P08238 83.1/5.0 2 Endoplasmin (Grp94) P14625 92.5/4.8 3 Protein disulﬁde-isomerase (PDI) P07237 57.1/4.8

Translation machinery: 4 Heterogeneous nuclear ribonucleoproteins A2/B1 (hnRNP A2/B1) P22626 37.4/9.0 5 116-kDa U5 small nuclear ribonucleoprotein (snRNP) component Q15029 109.4/4.8 6 Elongation factor Tu (EF-Tu) P49411 49.5/7.3 7 Elongation factor 1-␣-1 (EF-1-␣-1) P68104 50.1/9.1 8 Poly(rC) binding protein 1 (hnRNP E1) Q15365 37.5/6.7

Cytoskeleton: 9 ␤-Actin P60709 41.7/5.5 10 Ezrin P15311 69.3/6.0 11 ␣-Tubulin Q71U36 50.1/4.9 Downloaded from Miscellaneous: 12 Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) P00354 35.9/6.6 13 Carbamoyl-phosphate synthase (CPSase I) P31327 164.9/6.3

Less-deﬁned function: 14 13- kDa leucine-rich protein (LRP130) P42704 145.2/5.5

a Spot numbers correspond to the numbered spots in gel images of Fig. 1. Protein details were assessed in the Swiss-Prot database. http://www.jimmunol.org/ translational machinery (i.e., heterogeneous nuclear ribonucleo- bulin by GrM in cell lysates was virtually completed after 30–60 protein (hnRNP) A2/B1, U5 small nuclear ribonuclear protein min of incubation, whereas complete cleavage of ezrin occurred (snRNP) component, EL-Tu, EL-1-␣-1, and hnRNP E1), a protein after 2–4 h. GrM-SA did not show any reactivity with these sub- of less well defined function (i.e., LRP130), some miscellaneous strates, indicating that all cleavage events are specific for GrM (i.e., GAPDH and carbamoylphosphate synthetase I), and several proteolytic activity. Similar results were obtained when Jurkat cell components that control the integrity of the cytoskeleton (i.e., lysates were used (data not shown). These data indicate that ezrin ␤-actin, ezrin, and ␣-tubulin) (Table I). and ␣-tubulin are direct or indirect substrates of human GrM and by guest on September 24, 2021 that ␣-tubulin kinetically was more efficiently cleaved in tumor GrM cleaves cytoskeleton-related proteins cell lysates as compared with ezrin. The group of proteins that control the integrity of the cytoskeleton ␣ (i.e., ␤-actin, ezrin, and ␣-tubulin) was selected for further studies GrM cleaves -tubulin and ezrin in tumor cells undergoing cell because inactivation or down-regulation of these proteins has been death causally linked to cell death and they are known to protect tumor To examine whether ␣-tubulin and ezrin are also cleaved in intact cells from apoptosis (15–19). In addition, these proteins reside in cells undergoing cell death induced by GrM, we used a cell death the cytoplasm and as such are accessible for GrM. Finally, ␣-tu- assay in which perforin was used to deliver GrM into the tumor bulin has recently been identified as a physiological substrate of cell (Fig. 3). It has been established that GrM-induced cell death the cytotoxic lymphocyte component GrB (19, 20) and is currently can be detected by typical morphological changes and by measur- considered one of the most successful targets for anticancer che- ing membrane integrity by PI flow cytometry (10). Consistent with motherapy (16, 18). To verify the cleavage of ␤-actin, ezrin, and ␣-tubulin by GrM in protein extracts, HeLa cell lysates were incubated in the presence or absence of GrM or GrM-SA and subjected to immunoblotting by using Abs against these proteins (Fig. 2). Incubation of lysates with GrM resulted in time-dependent cleavage of ezrin and ␣-tubulin, but not ␤-actin. Unlike ezrin and ␣-tubulin, the ␤-actin protein was identified by tandem MS from cleaved ␤-actin fragments that appeared during GrM cleavage (spot no. 9 in Fig. 1B). This precluded information on the effi- ciency by which intact ␤-actin is cleaved by GrM. Western blot analysis, however, now indicates that the bulk of the intact ␤-actin protein remains uncleaved (Fig. 2). Although the mAbs used could not detect cleavage products, GrM-mediated cleavage of ezrin and ␣-tubulin was illustrated by the progressive time-dependent dis- ␣ appearance of both ezrin and -tubulin protein bands. The molec- FIGURE 2. Ezrin and ␣-tubulin are cleaved by GrM in tumor cell ly- ␣ ular weights of noncleaved ezrin and -tubulin matched the mo- sates. HeLa cell lysates were incubated with GrM (1 ␮M), GrM-SA (1 lecular weights of the protein spots excised from GrM-SA-treated ␮M), or buffer for the indicated times at 37°C. Samples were immuno- lysates on two-dimensional gels (Fig. 1A) and that of the theoret- blotted using Abs against ␤-actin, ezrin, and ␣-tubulin. These experiments ical molecular mass of these proteins (Table I). Cleavage of ␣-tu- were repeated at least three times with the same results. The Journal of Immunology 8187

minor reduction of ezrin was found. This is consistent with the observed differential cleavage kinetics of both proteins by GrM (Fig. 2). Cleavage of ␣-tubulin and ezrin, however, was nearly complete after a 10-h incubation period. In line with the results obtained in Fig. 2, no discernible hydrolysis of ␤-actin was observed. GrM-induced cell death was irreversible, as cell viability of GrM/perforin-treated Jurkat cells was 2.1 Ϯ 2.0% (mean Ϯ S.D.) as compared with 100 Ϯ 11.9% for GrM-SA/perforin after a 24 h incubation period ( p Ͻ 0.001). These results indicate that ␣-tubulin and ezrin, but not ␤-actin, are cleaved in tumor cells that undergo irreversible cell death induced by GrM in combination with perforin. GrM cleaves ␣-tubulin and ezrin in a caspase-independent manner Lu et al. (11) recently demonstrated that one way by which GrM induces cell death involves proteolytic activation of procaspase-3, whereas another study shows that GrM-mediated cell death com-

pletely occurs independently of caspases (10). Therefore, we have Downloaded from addressed the role of caspases and other cysteine proteases during GrM-mediated proteolysis of ␣-tubulin and ezrin. To this end, we used the pan-caspase inhibitor z-VAD-fmk and the broad spectrum cysteine protease inhibitor E64. Tumor cell lysates were incubated with GrM and proteolysis of ␣-tubulin and ezrin was monitored by

immunoblotting (Fig. 4A). Neither z-VAD-fmk nor E64 affected http://www.jimmunol.org/ GrM-mediated cleavage of ␣-tubulin and ezrin, indicating that neither caspases nor other cysteine proteases are involved in this pro- cess. E64 and z-VAD-fmk did not affect the activation of procaspase-3 by GrB (Fig. 4A), which is expected because GrB is a serine protease (1–4). Next, we investigated the capability of GrM to cleave and/or activate procaspase-3. In contrast to Lu et al. (11) but consistent with Kelly et al. (10), GrM did not cleave procaspase-3 (Fig. 4A). Also at higher GrM concentrations and longer incubation times GrM did not cleave this procaspase (Fig. 4B). In by guest on September 24, 2021 agreement with these findings, GrM did not activate procaspase-3 as determined by the small chromogenic caspase-3 substrate Ac- DEVD-pNA (Fig. 4C). GrB was included as positive control for cleavage and activation of procaspase-3 in these experiments. In- deed, the activity of caspase-3 was completely inhibited by the pan-caspase inhibitor z-VAD-fmk. As expected, E64 did not affect caspase-3 activity, given that E64 does not inhibit caspases (21). FIGURE 3. Ezrin and ␣-tubulin are cleaved by GrM in tumor cells un- E64 was functionally active because it efficiently blocked the ca- dergoing cell death. A–D, Jurkat cells (1 ϫ 106) were treated with a sublytic dose of perforin (40 ng/ml) and GrM (1 ␮M), perforin (40 ng/ml) and thepsin-binding properties of DCG-04, which is an active site- GrM-SA (1 ␮M), perforin (40 ng/ml) alone, or GrM (1 ␮M) alone for 4 h directed probe that is used to specifically label active cathepsin at 37°C. Cells (1 ϫ 105) were centrifuged on glass slides and visualized by proteases (data not shown). Taken together, these results indicate Giemsa staining. Bar, 20 ␮m. E–H, Cells (1 ϫ 104) were incubated with PI that GrM neither cleaves nor activates procaspase-3 and that ␣-tu- (1 ␮g/ml) for 10 min at room temperature and analyzed by flow cytometry. bulin and ezrin are not cleaved by GrM in a caspase(-3)-dependent I, At indicated time points the cells were lysed and whole cell protein or other cysteine protease-dependent manner. extracts were immunoblotted using Abs against ␣-tubulin, ezrin, and ␤-actin. These experiments were repeated at least three times with the same GrM directly cleaves ␣-tubulin and ezrin results. We investigated whether ␣-tubulin and ezrin constitute direct GrM substrates rather than being the substrates of secondary proteases other than cysteine proteases or caspases present in tumor cells. To this, GrM/perforin-treated cells demonstrated the morphological this end, GrM or GrM-SA was incubated with purified recombi- changes as described (10), such as signs of chromatin condensation nant GST-␣-tubulin or GST-ezrin. Treatment of purified GST- and the presence of large cytoplasmic vacuoles (Fig. 3C). As ex- ezrin with GrM, but not GrM-SA, resulted in the disappearance of pected (10), GrM/perforin-treated cells showed higher PI staining the expected ϳ100 kDa GST-ezrin protein band (Fig. 5). Using as compared with controls (Fig. 3, E–H). Lysates of GrM- or GrM- anti-GST (Fig. 5A) or anti-C-terminal ezrin Abs (Fig. 5B), cleav- SA-treated cells were subjected to immunoblotting and cleavage of age products were detected by immunoblotting. This indicates that ␣-tubulin and ezrin was monitored (Fig. 3I). Cleavage of ␣-tubulin ezrin is a direct substrate of GrM. Whereas GrM-SA-treated and or ezrin was not observed in cells treated with buffer, perforin untreated GST-␣-tubulin remained intact, treatment of purified alone, or perforin in combination with GrM-SA, but was detected ␣-tubulin with increasing concentrations of GrM resulted in the in cells treated with perforin and GrM. After4hofincubation a progressive disappearance of the ϳ75-kDa GST-fused ␣-tubulin modest reduction of ␣-tubulin could be observed, whereas only a protein and the appearance of two major cleavage products of ϳ52 8188 GRANZYME M TARGETS THE CYTOSKELETON Downloaded from

FIGURE 5. GrM directly cleaves ezrin. Puriﬁed recombinant GST- ezrin (75 nM) was treated with indicated concentrations of GrM (0–600 nM) or GrM-SA (600 nM) for2hat37°C. GrM:GST-ezrin stoichiometries were 1.5:1, 2.6:1, 5.3:1, and 7.7:1. Proteins were separated by SDS-PAGE http://www.jimmunol.org/ (10%) and subjected to immunoblotting using Abs against the N-terminal GST tag (A) or the C-terminal part of ezrin (B). Full-length GST-ezrin (solid arrow) and cleavage products (dotted arrows) are indicated. ns, Nonspeciﬁc.

tant His-tagged ␣-tubulin was also cleaved by GrM, a different amino-terminal cleavage product with increased m.w. was observed (Fig. 6B). Using a highly sensitive fluorescent protein staining on GrM-cleaved wild-type and L269A mutant ␣-tubulin, the by guest on September 24, 2021 up-shift of the N-terminal proteolytic fragment of L269A ␣-tubulin mutant was again evident (Fig. 6C). Strikingly, however, under these conditions at least seven other ␣-tubulin cleavage fragments FIGURE 4. GrM cleaves ␣-tubulin and ezrin in a caspase-independent appeared when ␣-tubulin was cleaved by GrM (Fig. 6C). The ␮ manner. A, Jurkat cell lysates were incubated with GrM (1 M), GrM-SA L269A ␣-tubulin mutant kinetically was equally well cleaved by ␮ ␮ Ϫ ϩ (1 M), or GrB (1 M)for2hat37°C, in the absence ( ) or presence ( ) ␣ D ␮ ␮ GrM as compared with wild-type -tubulin (Fig. 6 ), strongly of z-VAD-fmk (10 M) or E64 (10 M). Samples were immunoblotted, ␣ ␣ suggesting that other GrM cleavage sites in -tubulin are at least using Abs against -tubulin, ezrin, or caspase-3. B, Jurkat cell lysates were ␣ incubated with GrM (2 ␮M), GrM-SA (2 ␮M), or GrB (0.5 ␮M) for 0–4 equally important. Furthermore, the N-terminal -tubulin cleavage h at 37°C. Samples were immunoblotted, using Abs against caspase-3. fragment appeared at low GrM concentrations with limited prote- These experiments were repeated at least three times with the same results. olysis at multiple sites until it completely disappeared when higher C, Jurkat cell lysates were incubated with GrM (1 ␮M), GrM-SA (1 ␮M), GrM concentrations were used (Fig. 6D). We have attempted to or GrB (1 ␮M)for4hat37°C in the absence (Ϫ) or presence (ϩ)of identify additional GrM cleavage sites by site-directed mutagene- z-VAD-fmk (10 ␮M) or E64 (10 ␮M). Samples were incubated with the sis of Leu and Met residues more C-terminal of Leu269. These chromogenic caspase-3 substrate Ac-DEVD-pNA (0.5 mM) and measured ␣-tubulin mutants include L286A, L286A/L269A, M302A, kinetically at 405 nm for 60 min. Data are presented as pmol of p-nitro- M302A/L269A, M313A, M313A/L269A, and L317A/L318A. Ex- aniline that is cleaved from the chromogenic substrate per minute as cept for L269A, we were not able to demonstrate a difference in mean Ϯ S.D. of three independent experiments. proteolysis of these mutants, either alone nor in combination with L269A (data not shown). This indicates that other Leu or Met residues in ␣-tubulin are more important GrM cleavage sites or and ϳ23 kDa (Fig. 6A). This cleavage event already occurred at that GrM cleaves ␣-tubulin after other amino acids than the pro- low nanomolar concentrations of GrM (5–20 nM) and relatively posed Leu or Met (22, 23). The latter would be consistent with a high tubulin concentrations (1 ␮M). Immunoblotting with an Ab Ser residue being a GrM cleavage site in (inhibitor of caspase- against GST revealed that the 52-kDa band represents the N-ter- activated DNase (ICAD; Ref. 11). Cleavage at alternate sites in minal ␣-tubulin moiety fused to the GST tag (data not shown). substrates has also been found for GrA and its substrates Ku70 and Higher concentrations of GrM further processed the N-terminal SET (9). Cleavage by GrM after Leu269 bisects the MAP-binding 52-kDa cleavage product, indicating that GrM cleaves ␣-tubulin at domain of ␣-tubulin (Fig. 6E) and this cleavage site is conserved least at two sites. Taking a closer look at the molecular weights of in all human ␣-tubulin isoforms except ␣-tubulin-L3 (Fig. 6F). the cleavage fragments of ␣-tubulin and knowing the proposed P1 The latter isoform, however, harbors P1 Met that can also be hy- primary and P2-P4 subsite specificities of GrM (22, 23), we mu- drolyzed by GrM (22, 23). Thus, ␣-tubulin proteolysis by GrM is tated the Leu residue at position 269 into Ala. Although this mu- direct, efficient, complex, and occurs at multiple cleavage sites. The Journal of Immunology 8189 Downloaded from http://www.jimmunol.org/

FIGURE 6. GrM directly cleaves ␣-tubulin. A, Purified recombinant GST-␣-tubulin (1 ␮M) was treated with indicated concentrations of GrM (0–500 by guest on September 24, 2021 nM) or GrM-SA (500 nM) for2hat37°C. GrM:GST-␣-tubulin stoichiometries were 1:1000, 1:200, 1:50, 1:10, and 1:2. Proteins were separated by SDS-PAGE (10%) and stained with Coomassie Brilliant Blue. Full-length GST-␣-tubulin (solid arrow) and cleavage products (dotted arrows) are indicated. B, Purified recombinant wild-type His-␣-tubulin (WT) (1 ␮M) and the His-␣-tubulin mutant in which Leu269 has been replaced by Ala (L269A) were treated with GrM (50 nM) or GrM-SA (50 nM) for2hat37°C. Samples were subjected to Western blot analysis, using an anti-His tag Ab. C, Purified recombinant wild-type (WT) and mutant L269A His-␣-tubulin (1 ␮M) were treated with (ϩ) or without (Ϫ)GrM(50nM)for2hat37°C. Proteins were separated by SDS-PAGE (10%) and stained with fluorescent Flamingo staining. D, Wild-type and mutant L269A His-␣-tubulin were treated with GrM (0–500 nM) or GrM-SA (500 nM) for2hat37°C. Proteins were separated by SDS-PAGE (10%) and stained by immunoblotting, using an anti-His tag Ab. Full-length His-␣-tubulin (solid arrow), and N-terminal (N1 and N2) cleavage products and other cleavage fragments (dotted arrows) are indicated (B–D). E, Schematic representation of ␣-tubulin domain structure, including the GrM cleavage site. F, Sequence alignment of amino acid region 246–289 of human ␣-tubulin isoforms. Amino acid identity is indicated in black, except that the GrM cleavage site is depicted in gray. A–D represent P1Ј–P4Ј and 1–4 represent P1–P4, respectively.

One cleavage site includes Leu269 and at least one other cleavage GrM disorganizes the microtubule network during killing of site is positioned slightly more C-terminal thereof. tumor cells

GrM de-regulates tubulin polymerization dynamics To evaluate the physiological effects of GrM-mediated cleavage of ␣-tubulin, perforin was used to load Jurkat cells with GrM or To investigate the effect of GrM on tubulin polymerization rates, GrM-SA for 4 h. GrM-induced cell dead was verified by PI flow we preincubated GrM with a mixture of purified, MAP-depleted cytometry (Fig. 3, E–H) and visual morphological inspection (Fig. bovine ␣- and ␤-tubulin. Following the addition of GTP, micro- 3, A–D) (10). To visualize microtubules, cells were stained with an tubule formation was measured kinetically at an absorbance of 340 ␣ nm. The kinetics of tubulin polymerization was markedly en- Ab against -tubulin and analyzed by confocal microscopy (Fig. hanced in the presence of GrM as compared with GrM-SA or 8). The microtubule network of control perforin/GrM-SA-treated buffer (Fig. 7A). This effect of GrM was comparable to the effect cells appeared as a normal fine structured filamentous tubule net- induced by the well-established anti-microtubule, anti-cancer drug work (Fig. 8, A and B) similar to that of untreated cells and cells paclitaxel (Taxol) (Fig. 7A) (16, 18). GrM cleaved bovine ␣-tu- that were incubated with perforin alone (data not shown). In con- bulin during the time course of tubulin polymerization (Fig. 7B). trast, many cells that received perforin in combination with GrM This is consistent with the GrM cleavage site (at least P4-P4Ј) displayed an aberrant microtubule network in that ␣-tubulin struc- being completely conserved in bovine ␣-tubulin and multiple other tures were less organized and appeared more diffuse (Fig. 8, C and mammalian species. Thus, GrM deregulates the polymerization D). Interestingly, perforin/GrM-treated cells were more flattened dynamics of the microtubule network. and displayed an aberrant shape as compared with controls. The 8190 GRANZYME M TARGETS THE CYTOSKELETON

Discussion Little is known about the molecular mechanisms by which GrM kills its target cells. It has been postulated that GrM uses multiple pathways to kill, either via caspase-dependent routes that lead to DNA fragmentation and ROS production or via caspase-independent pathways that do not result in fragmentation of DNA and production of ROS (10–12). Targeting of multiple independent cell death pathways has also been demonstrated for GrA and GrB (1, 2, 5–9). In the present study, we have demonstrated that GrM neither cleaves nor activates procaspase-3 (Fig. 4), which is in contrast to Lu et al. (11) but consistent with Kelly et al. (10). We have identiﬁed several novel potential substrates of GrM. We have shown that GrM directly cleaves the actin-plasma membrane linker ezrin and the microtubule network protein ␣-tubulin in tumor cells that are attacked by perforin and GrM (Figs. 3, 5, 6). Ezrin and ␣-tubulin are not cleaved by GrM in a caspase(-3)- or cysteine protease-dependent manner (Fig. 4). Cleavage of ␣-tubulin by GrM deregulates ␣-tubulin function and leads to disorganization of the microtubule network (Figs. 7 and 8).

FIGURE 7. GrM disturbs tubulin polymerization dynamics. A, Purified Downloaded from bovine tubulin (40 ␮M) was incubated with GrM (1 ␮M) (open circles), Therefore, tubulin proteolysis by GrM is likely to be a critical event GrM-SA (1 ␮M) (open diamonds), Paclitaxel (Taxol) (5 ␮M) (closed tri- during NK cell-mediated killing. angles), or buffer alone (closed circles). Except for Taxol, treated samples Microtubules are responsible for cell survival, mitosis, motility, were preincubated for2hat30°C. Microtubule polymerization was then maintenance of cell shape, cell signaling, and intracellular trafficking initiated by the addition of GTP (1 mM) and 5% glycerol and measured of macromolecules, vesicles, and organelles (16, 18). The highly dy- kinetically at 340 nm at 37°C. Data represent the mean of three to four namic behavior of microtubules is greatly affected by well-known independent experiments. B, After 60 min of measurement, cleavage of http://www.jimmunol.org/ anti-cancer drugs, like vinblastine, vincristine, and Taxol, which all bovine ␣-tubulin by GrM was verified by immunoblotting using an anti- ␣-tubulin Ab. induce abnormal mitosis and cell death (16, 18). Furthermore, down- regulation of ␣-tubulin by RNA interference results in the death of tumor cells and limits their mitotic potential (19). We have found that GrM cleaves off the C-terminal part of the ␣-tubulin MAP-binding thickness of representative cells (mean Ϯ S.D.) was 6.6 Ϯ 0.7 ␮m domain (Fig. 6), which regulates microtubule polymerization dynam- and 3.6 Ϯ 0.3 ␮m(p Ͻ 0.001, n ϭ 7) and the diameter was ics and microtubule motor activity (16, 18). Indeed, GrM, like Taxol, 12.3 (Ϯ 1.1) ␮m and 17.7 (Ϯ 1.4) ␮m(p Ͻ 0.001, n ϭ 7) for shifted the balance of microtubule dynamics toward polymerization GrM-SA- and GrM-treated cells, respectively. These data indi-

(Fig. 7), which has also been demonstrated for GrB (19). In this con- by guest on September 24, 2021 cate that GrM disorganizes the microtubule network and cell text, however, we cannot fully exclude the following: 1) that GrM shape in tumor cells that are attacked by perforin and GrM. affects in vitro tubulin polymerization by cleaving ␤-tubulin or trace amounts of MAPs that may accompany tubulins in the tubulin polymerization assay; and/or 2) that GrM-cleaved ␣-tubulin is irrelevant to microtubule polymerization in that it is more prone to aggregation as compared with uncleaved ␣-tubulin. Nevertheless, GrM disorganizes the microtubule network and cell shape in tumor cells that are attacked by perforin and GrM (Fig. 8). Therefore, we hypothesize that GrM- induced cleavage of ␣-tubulin in tumor cells contributes to cell death induced by NK cells and/or that it enhances the NK cell function to kill. The latter possibility would be compatible with the finding that Taxol pretreatment amplifies NK cell-mediated lysis of tumor targets (24). Alternatively, it has been well established that host cell microtubules are indispensable for viral entry, replication, and exit (25). GrM plays a significant role in the elimination of virus-infected cells in vivo (26). This opens the possibility that GrM-mediated disruption of microtubule function terminates viral production in infected cells during NK cell attack. GrM may act in concert with GrB, because the latter also cleaves ␣-tubulin and resides in the same NK cell granules (19, 20). GrM cleaved the actin-plasma membrane linker ezrin by using purified proteins (Fig. 5), tumor cell lysates (Fig. 2), and tumor cells undergoing cell death induced by GrM and perforin (Fig. 3). Ezrin is (over)expressed in a variety of cancers, some of which are associated with poor clinical outcome (27). Linkage of the plasma membrane to the actin cytoskeleton by ezrin allows a cell to interact directly with its FIGURE 8. GrM disorganizes the microtubule network in tumor cells undergoing cell death. A–D, Jurkat cells (1 ϫ 106) were treated with a sublytic microenvironment (27). Ezrin also facilitates several signal transduc- dose of perforin (40 ng/ml) and GrM-SA (1 ␮M) (A and B) or GrM (1 ␮M) tion pathways, like that of the protein kinases AKT and MAPK (C and D)for4hat37°C. ␣-Tubulin is visualized by ϫ100 (original magni- (MEK/ERK) that protect cells against apoptosis (15). Therefore, fication) fluorescent immunostaining and confocal microscopy. Bar, 5 ␮m. GrM-dependent cleavage of ezrin may impair activation of AKT and The Journal of Immunology 8191

MAPK survival pathways and thus may contribute to target cell death. 4. Grossman, W. J., P. A. Revell, Z. H. Lu, H. Johnson, A. J. Bredemeyer, and Another possibility may be that GrM inhibits tumor metastatic pro- T. J. Ley. 2003. The orphan granzymes of humans and mice. Curr. Opin. Im- munol. 15: 544–552. gression by inactivation of ezrin. This would be consistent with the 5. Metkar, S. S., B. Wang, M. L. Ebbs, J. H. Kim, Y. J. Lee, S. M. Raja, and findings that ezrin is necessary for metastatic progression and that it is C. J. Froelich. 2003. Granzyme B activates procaspase-3 which signals a mitochondrial amplification loop for maximal apoptosis. J. Cell Biol. 160: 875–885. required for early metastatic survival of tumor cells in vivo (27). Fur- 6. Waterhouse, N. J., K. A. Sedelies, K. A. Browne, M. E. Wowk, A. Newbold, ther studies are required to distinguish between these possibilities. V. R. Sutton, C. J. Clarke, J. Oliaro, R. K. Lindemann, P. I. Bird, et al. 2005. A central Fourteen novel potential substrates of GrM were identified (Table role for Bid in granzyme B-induced apoptosis. J. Biol. Chem. 280: 4476–4482. ␣ 7. Fan, Z., P. J. Beresford, D. Y. Oh, D. Zhang, and J. Lieberman. 2003. Tumor sup- I). Apart from -tubulin and ezrin, however, direct processing by pressor NM23–H1 is a granzyme A-activated DNase during CTL-mediated apopto- GrM of these proteins and the precise role thereof remains to be in- sis, and the nucleosome assembly protein SET is its inhibitor. Cell 112: 659–672. vestigated. Some potential GrM substrates may play a role in the 8. Martinvalet, D., P. Zhu, and J. Lieberman. 2005. Granzyme A induces caspase- independent mitochondrial damage, a required first step for apoptosis. Immunity 22: mechanism by which GrM induces cell death, for instance proteins 355–370. involved in chaperone and cell stress response (Table I). HSP90␤ 9. Zhu, P., D. Zhang, D. Chowdhury, D. Martinvalet, D. Keefe, L. Shi, and plays an essential role in maintaining stability and activity of its client J. Lieberman. 2006. Granzyme A, which causes single-stranded DNA damage, targets the double-strand break repair protein Ku70. EMBO Rep. 7: 431–437. proteins, including a set of signaling proteins that regulate key path- 10. Kelly, J. M., N. J. Waterhouse, E. Cretney, K. A. Browne, S. Ellis, J. A. Trapani, ways in cell survival and oncogenesis (28). HSP90␤ is frequently and M. J. Smyth. 2004. Granzyme M mediates a novel form of perforin-depen- overexpressed in cancer cells and, more importantly, synthetic dent cell death. J. Biol. Chem. 279: 22236–22242. ␤ 11. Lu, H., Q. Hou, T. Zhao, H. Zhang, Q. Zhang, L. Wu, and Z. Fan. 2006. Gran- HSP90 inhibitors have successfully been evaluated in multiple phase zyme M directly cleaves inhibitor of caspase-activated DNase (CAD) to unleash II anticancer clinical trials (27). If GrM indeed inactivates HSP90␤, CAD leading to DNA fragmentation. J. Immunol. 177: 1171–1178. this may represent a novel mechanism by which GrM induces target 12. Hua, G., Q. Zhang, and Z. Fan. 2007. Heat shock protein 75 (TRAP1) antago- Downloaded from nizes reactive oxygen spiecies generation and protect cells from granzyme M- cell death. We are currently addressing this possibility. Interestingly, mediated apoptosis. J. Biol. Chem. 282: 20553–20560. the HSP90 cochaperones Hop and Hip have recently been identified 13. Froelich, C. J., J. Turbov, and W. Hanna. 1996. Human perforin: rapid enrich- ment by immobilized metal affinity chromatography (IMAC) for whole cell cy- as novel substrates of GrB (29, 30). Whether or not GrM also plays a totoxicity assays. Biochem. Biophys. Res. Commun. 229: 44–49. role in other cellular processes than cell death remains an intriguing 14. Prento, P. 1997. The effects of freezing, storage, and thawing on cell compart- question that deserves further study. ment integrity and ultrastructure. Histochem. Cell Biol. 108: 543–547. 15. Gautreau, A., P. Poullet, D. Louvard, and M. Arpin. 1999. Ezrin, a plasma mem- Our proteomic approach has identified a limited set of potential brane-microfilament linker, signals cell survival through the phosphatidylinositol http://www.jimmunol.org/ GrM substrates (Fig. 1, Table I). The relatively small number of 3-kinase/Akt pathway. Proc. Natl. Acad. Sci. USA 96: 7300–7305. cleavage events detected suggests that GrM substrate specificity de- 16. Jordan, M. A., and L. Wilson. 2004. Microtubules as a target for anticancer drugs. Nat. Rev. Cancer 4: 253–265. pends on extended binding site(s) on folded proteins rather than short 17. Gourlay, C. W., and K. R. Ayscough. 2005. The actin cytoskeleton: a key reg- linear peptides that represent the cleavage site (P1), i.e., methionine or ulator of apoptosis and ageing? Nat. Rev. Mol. Cell Biol. 6: 583–589. leucine (22, 23). This is consistent with the high specificities of GrA 18. Honore, S., E. Pasquier, and D. Braguer. 2005. Understanding microtubule dynamics for improved cancer therapy. Cell Mol. Life Sci. 62: 3039–3056. and GrB, which also fully depend on secondary structures of folded 19. Adrain, C., P. J. Duriez, G. Brumatti, P. Delivani, and S. J. Martin. 2006. The substrates (31, 32). Remarkably, our proteomic screen did not detect cytotoxic lymphocyte protease, granzyme B, targets the cytoskeleton and perturbs one of the known GrM substrates, i.e., inhibitor of caspase-dependent microtubule polymerization dynamics. J. Biol. Chem. 281: 8118–8125.

20. Goping, I. S., T. Sawchuk, D. A. Underhill, and R. C. Bleackley. 2006. Identi- by guest on September 24, 2021 DNase (ICAD), poly(ADP-ribose) polymerase (PARP), HSP75 fication of ␣-tubulin as a granzyme B substrate during CTL-mediated apoptosis. (TRAP1), or procaspase-3 (11, 12). Absence of the latter is consistent J. Cell Sci. 119: 858–865. with our finding that GrM neither cleaves nor activates procaspase-3 21. Rozman-Pungercar, J., N. Kopitar-Jerala, M. Bogyo, D. Turk, O. Vasiljeva, I. Stefe, P. Vandenabeele, D. Bro¨mme, V. Puizdar, M. Fonovic´, et al. 2003. Inhibition of in tumor cells (Fig. 4). Although two-dimensional gel electrophoresis papain-like cysteine proteases and legumain by caspase-specific inhibitors: when re- is capable of resolving Ͼ1000 individual protein spots on a single gel, action mechanism is more important than specificity. Cell Death Differ. 10: 881–888. not all proteins could be visualized because of low abundance or ex- 22. Mahrus, S., W. Kisiel, and C. S. Craik. 2004. Granzyme M is a regulatory protease that inactivates proteinase inhibitor 9, an endogenous inhibitor of granzyme tremes of m.w. or charge. In addition, some of the known GrM sub- B. J. Biol. Chem. 279: 54275–54282. strates may have been detected on the gels but could not be identified 23. Rukamp, B. J., C. M. Kam, S. Natarajan, B. W. Bolton, M. J. Smyth, J. M. Kelly, and by tandem MS. We were able to definitively identify 16 of 37 excised J. C. Powers. 2004. Subsite specificities of granzyme M: a study of inhibitors and newly synthesized thiobenzyl ester substrates. Arch. Biochem. Biophys. 422: 9–22. protein spots. Because of the nonquantitative nature of silver staining, 24. Mehta, S., D. Blackinton, M. Manfredi, D. Rajaratnam, N. Kouttab, and it remains difficult to address the cellular abundance of our identified H. Wanebo. 1997. Taxol pretreatment of tumor targets amplifies natural killer cell potential GrM substrates. mediated lysis. Leuk. Lymphoma 26: 67–76. 25. Dohner, K., C. H. Nagel, and B. Sodeik. 2005. Viral stop-and-go along micro- GrM is highly expressed by NK cells but not in CTLs (33). NK tubules: taking a ride with dynein and kinesins. Trends Microbiol. 13: 320–327. cells play a major role in the innate immune response that forms 26. Pao, L. I., N. Sumaria, J. M. Kelly, S. van Dommelen, E. Cretney, M. E. Wallace, D. A. Anthony, A. P. Uldrich, D. I. Godfrey, J. M. Papadimitriou, et al. 2005. the first line of defense against tumor cells and virus-infected cells, Functional analysis of granzyme M and its role in immunity to infection. J. Im- and they have broad applications in immunotherapy of cancer (34). munol. 175: 3235–3243. Knowledge of the precise mechanisms by which NK cells kill 27. Hunter, K. W. 2004. Ezrin, a key component in tumor metastasis. Trends Mol. Med. 10: 201–204. tumor cells may lead to further optimization of immunotherapy 28. Whitesell, L., and S. L. Lindquist. 2005. HSP90 and the chaperoning of cancer. and/or other proapoptotic anticancer therapies. Nat. Rev. Cancer 5: 761–772. 29. Bredemeyer, A. J., P. E. Carrigan, T. A. Fehniger, D. F. Smith, and T. J. Ley. Acknowledgments 2006. Hop cleavage and function in granzyme B-induced apoptosis. J. Biol. Chem. 281: 37130–37141. We thank Dr. M. A. G. G. Vooijs for critical reading of the manuscript. 30. Hostetter, D. R., C. R. K. Loeb, F. Chu, and C. S. Craik. 2007. Hip is a pro- survival substrate of granzyme B. J. Biol. Chem. 282: 27865–27874. Disclosures 31. Waugh, S. M., J. L. Harris, R. Fletterick, and C. S. Craik. 2000. The structure of The authors have no financial conflict of interest. the pro-apoptotic protease granzyme B reveals the molecular determinants of its specificity. Nat. Struct. Biol. 7: 762–765. 32. Bell, J. K., D. H. Goetz, S. Mahrus, J. L. Harris, R. J. Fletterick, and C. S. Craik. References 2003. The oligomeric structure of human granzyme A is a determinant of its 1. Barry, M., and R. C. Bleackley. 2002. Cytotoxic T lymphocytes: all roads lead to extended substrate specificity. Nat. Struct. Biol. 10: 527–534. death. Nat. Rev. Immunol. 2: 401–409. 33. Sayers, T. J., A. D. Brooks, J. M. Ward, T. Hoshino, W. E. Bere, G. W. Wiegand, 2. Lieberman, J. 2003. The ABCs of granule-mediated cytotoxicity: new weapons in J. M. Kelly, and M. J. Smyth. 2001. The restricted expression of granzyme M in the arsenal. Nat. Rev. Immunol. 3: 361–370. human lymphocytes. J. Immunol. 166: 765–771. 3. Smyth, M. J., and J. A. Trapani. 1995. Granzymes: exogenous proteinases that 34. Farag, S. S., and M. A. Caligiuri. 2006. Human natural killer cell development induce target cell apoptosis. Immunol. Today 16: 202–206. and biology. Blood Rev. 20: 123–137.