Membrane and Morphological Changes in Apoptotic Cells Regulated by Caspase-Mediated Activation of PAK2

See allHide authors and affiliations

Science  06 Jun 1997:
Vol. 276, Issue 5318, pp. 1571-1574
DOI: 10.1126/science.276.5318.1571


Apoptosis of Jurkat T cells induced the caspase-mediated proteolytic cleavage of p21-activated kinase 2 (PAK2). Cleavage occurred between the amino-terminal regulatory domain and the carboxyl-terminal catalytic domain, which generated a constitutively active PAK2 fragment. Stable Jurkat cell lines that expressed a dominant-negative PAK mutant were resistant to the Fas-induced formation of apoptotic bodies, but had an enhanced externalization of phosphatidylserine at the cell surface. Thus, proteolytic activation of PAK2 represents a guanosine triphosphatase–independent mechanism of PAK regulation that allows PAK2 to regulate morphological changes that are seen in apoptotic cells.

Apoptosis, or regulated cell death, is a fundamental process in the development of multicellular organisms. Although it is initiated by many physiologic and pathologic stimuli, all apoptotic cells undergo a similar sequence of morphological and biochemical events (1). The cascade of ICE/CED-3 family cysteine proteases (termed caspases) (2) is a common and critical component of the cell death pathway (3,4). The identified targets for proteolytic cleavage by caspases are few, and the role of individual targets in mediating particular apoptotic events remains ill-defined.

p21-activated kinases (PAKs) are serine-threonine kinases whose activity is regulated by the small guanosine triphosphatases (GTPases) Rac and Cdc42 (5, 6). PAKs regulate morphological and cytoskeletal changes in a variety of cell types (7,8), implicating PAKs as downstream mediators of the effects of Rac and Cdc42 on the actin cytoskeleton. Immunoblot analysis (9) reveals that Jurkat T cells predominantly express the 62-kD PAK2 isoform. An apparent decrease in the intensity of the 62-kD PAK2 band is observed after induction of Jurkat cell death by Fas receptor cross-linking (10). A 34-kD COOH-terminal PAK2 fragment (Fig. 1A) and a 28-kD NH2-terminal PAK2 fragment (Fig. 1B) appeared after 1 hour, suggesting that PAK2 is cleaved into two defined fragments during Fas-induced apoptosis (11). In detailed time-course studies, PAK2 cleavage was detected as early as 30 min after stimulation by immunoglobulin M (IgM) antibody to Fas (anti-Fas) and the cleavage of PAK2 always correlated with the onset of apoptotic cell death. PAK2 cleavage was also observed when apoptosis was induced in Jurkat cells with C2 ceramide or in MCF-7 cells with tumor necrosis factor–α (TNF-α) (Fig. 1C), suggesting that this is a general phenomenon in apoptotic cells.

Figure 1

Proteolytic cleavage of PAK2 in Fas-stimulated Jurkat T lymphocytes is blocked by ICE inhibitors. The formation of NH2- and COOH-terminal cleavage products of PAK2 was measured at the indicated times after stimulation by Fas cross-linking. (A) Formation of the 34-kD COOH-terminal fragment (C-term.) was detected by immunoblotting with COOH-terminal–directed anti-Ste20, which cross-reacts with the highly conserved catalytic COOH-terminus of mammalian PAK2 (9). (B) The 28-kD NH2-terminal fragment (N-term.) was detected by a blot overlay with [35S]GTPγS-labeled GST-Rac1 (11). (C) Formation of the 34-kD COOH-terminal PAK2 fragment in either the MCF-7 breast cancer epithelial cell line, in which apoptosis was induced with TNF-α (20 ng/ml) plus 10 μM pyrrolidinedithiocarbamate for 12 hours, or in Jurkat T cells stimulated with 80 μM C2 ceramide for 8 hours. The COOH-terminal fragment was detected by immunoblotting as in (A). (D) PAK2 cleavage in Fas-stimulated Jurkat cells was inhibited in the presence of 300 μM DEVD-ald or 300 μM YVAD-CMK.

Because the cleavage of PAK2 correlated with the onset of apoptosis, we reasoned that caspases might proteolytically cleave PAK2. Jurkat T cells were incubated with the caspase inhibitors YVAD-CMK (12) or DEVD-ald (13) before Fas receptor ligation, and PAK2 processing and apoptosis were monitored (Fig. 1D). Both DEVD-ald and YVAD-CMK peptides substantially inhibited cleavage of PAK2. In a broken cell apoptotic lysate (14), DEVD-ald totally blocked cleavage of a purified histidine-tagged PAK2 fusion protein (H6PAK2H6) (15) at concentrations as low as 0.1 μM, whereas YVAD-CMK was at least 100-fold less potent (16). As DEVD peptides are inhibitors of CPP32 (caspase 3)-like proteases (13), we compared the in vitro cleavage of H6PAK2H6 using lysates of Fas-activated Jurkat cells to cleavage by recombinant CPP32 (rCPP32) expressed as a glutathione-S-transferase (GST) fusion protein in Escherichia coli (17). rCPP32 cleaved H6PAK2H6 to a 34-kD COOH-terminal fragment (Fig.2A) and a 28-kD NH2-terminal fragment (16) of the same size as those produced by activated Jurkat lysates.

Figure 2

Proteolytic cleavage of PAK2 by CPP32 and identification of the cleavage site on PAK2. (A) Recombinant His-tagged (H6) PAK2 was incubated as described (14) with either a bacterial lysate overexpressing recombinant CPP32 (rCPP32) or a cell lysate prepared from Fas-stimulated Jurkat cells undergoing apoptosis. Lanes 1 and 6: H6PAK2H6 incubated with 40 μg of an inactive control lysate; lanes 2 and 7: 40 μg of active lysate with no exogenous H6PAK2H6 added; lanes 3 to 5: H6PAK2H6 incubated with 10, 20, and 40 μg of active rCPP32, respectively; lanes 8 to 11: 5, 10, 20, and 40 μg of active Jurkat lysate, respectively. CP, the COOH-terminal PAK2 cleavage product, detected by immunoblotting with COOH-terminal–directed anti-Ste20. (B) Recombinant His-tagged (H6) PAK1 or PAK2 were incubated with lysates from Fas-stimulated Jurkat cells, as described (14). Lanes 1 and 7: PAK1H6 and H6PAK2H6, respectively, incubated with 40 μg of inactive control lysate; lanes 2 and 8: 40 μg of active Jurkat lysate alone; lanes 3 to 6 and 9 to 12: PAK1H6 and H6PAK2H6 incubated with 5, 10, 20, and 40 μg of active Jurkat lysate, respectively. (C) Wild-type (WT) H6PAK2H6 and H6PAK2H6D212N protein were incubated either with inactive control lysate (control) or active recombinant CPP32 (rCPP32), as described (15). Active rCPP32 without H6PAK2H6 (−) served as a control. Similar resistance to cleavage of the PAK2 D212N mutant was observed in apoptotic cell lysates. The partial cleavage fragment seen just below the intact PAK2 WT and D212N proteins represents a non–caspase-related proteolysis product whose formation was not increased by CPP32.

PAK2 contains several potential CPP32 cleavage sites (18), some of which are also present in PAK1. To determine if both PAK1 and PAK2 were caspase substrates, we treated purified PAK1H6 and PAK2H6 with lysates of Fas-induced Jurkat cells under identical conditions (14). PAK2 was processed to the 34-kD COOH-terminal cleavage product, whereas PAK1 was not cleaved (Fig. 2B). The CPP32 cleavage site in PAK2 was identified as the position adjacent to Asp212 by mutagenesis to asparagine (19). Thus, H6PAK2H6D212N was not proteolytically cleaved to the 34-kD COOH-terminal fragment by rCPP32 (Fig. 2C), nor in apoptotic cell lysates containing endogenous CPP32-like caspases. The predicted sizes of the PAK2 proteolytic fragments resulting from cleavage at Asp212 are consistent with what is observed (∼34 and ∼28 kD). We therefore conclude that CPP32 cleaves PAK2 after Asp212. Although PAK1 also contains an aspartic acid residue at the homologous position as PAK2, this site is surrounded by a number of features (for example, continuous proline residues) that are likely to disrupt recognition by caspases. PAK3 lacks the relevant aspartate residues and thus the consensus caspase cleavage site.

PAKs consist of two major functional domains, an NH2-terminal regulatory domain and the COOH-terminal catalytic domain (5, 7). Because the 34-kD PAK2 cleavage product apparently contains the complete catalytic domain (5), the removal of the regulatory domain by cleavage of PAK2 during Fas-induced apoptosis might be sufficient to cause PAK2 kinase activation. By use of in-gel kinase assays (20), a p47phox peptide–dependent kinase activity (21) of 34 kD appeared after Fas ligation (Fig.3) with a time course that paralleled PAK2 cleavage (Fig. 1A). The in vitro–cleaved PAK2H6 protein had the same activity pattern in the in-gel kinase assay, and recombinant PAK2 catalytic domain (amino acids 213 to 525) was also active (Fig. 3). Little activation of the intact PAK2 protein was detected.

Figure 3

Constitutive activation of PAK2 catalytic domain after CPP32-mediated proteolysis. The kinase activity of PAK2 was assessed by use of an in-gel kinase assay with p47phox peptide as the substrate (20,21). Autodioradiograph of the gel (A) without peptide substrate and (B) with peptide substrate incorporated. Recombinant purified PAK2H6 was also examined for kinase activity after treatment in the presence or absence of rCPP32, as indicated. Because of six histidines fused to the COOH-terminus of PAK2H6, the resulting CPP32 cleavage product migrates at a slightly higher molecular mass than the endogenous PAK2 cleavage product. The last lane contains recombinant PAK2 truncated at the NH2-terminal (lacking amino acids 1 to 212) to generate the COOH-terminal fragment resulting from CPP32-mediated cleavage of intact PAK2. This construct was tagged with six histidines and the pET28 linker peptide at the NH2-terminus and an additional six histidines at the COOH-terminus and consequently migrates at a substantially slower rate than the untagged cleavage product.

To investigate the role of PAK in the apoptotic program, we generated two independently derived Jurkat T cell lines (TD8 and TA12) stably expressing the PAK(H83L,H86L,K299R) dominant-negative mutant kinase (22) under an isopropyl-β-d-thiogalactopyranoside (IPTG)–inducible promoter (23). Expression of PAK(H83L,H86L,K299R) was induced about threefold and fivefold in the TD8 and TA12 cell lines, respectively, in the presence of IPTG, as determined by immunoblotting. Although Fas ligation still induced cell death in Jurkat cells expressing dominant-negative PAK(H83L,H86L,K299R), as indicated by DNA fragmentation, these apoptotic cells had abnormal morphological changes (24). Cells expressing dominant-negative PAK did not form apoptotic bodies during Fas-induced killing (>90% inhibition as compared to control Jurkat cells), but rather remained as intact rounded cells in which DNA fragmentation could still be detected by terminal deoxynucleotidyl transferase–mediated dUTP nick-end labeling (TUNEL) assay (Fig.4A). One explanation for the inhibition of apoptotic body formation in TA12 cells could be that the onset of apoptosis is retarded in these cell lines. However, the normal kinetics of nuclear fragmentation and proteolytic cleavage of the CPP32 substrate D4 guanosine diphosphate dissociation inhibitor (25) do not support this hypothesis (16).

Figure 4

Dominant-negative PAK regulates morphological and membrane changes during Fas-induced apoptosis. (A) Control Jurkat T cells or the TA12 cell line after induction with IPTG were visualized for cell morphology (left) or DNA degradation in a TUNEL assay (right) after treatment as indicated. Anti-Fas IgM was added for 4 hours before analysis. (B) Annexin-V–FITC binding to control cells (□, –IPTG; ◊, + IPTG) or the TA12 stable cell line (○, −IPTG; ▵, +IPTG) was determined as described (28). Cells that bound substantial amounts of annexin-V–FITC, but did not stain for PI, were quantified. Shown are the representative results of three similar experiments.

The process of phosphatidylserine (PS) export to the outer leaflet of the plasma membrane is an early and caspase-dependent event during apoptosis of cells from numerous lineages (26,27). When we examined the exposure of PS at the cell surface, as measured by the binding of annexin-V–fluorescein isothiocyanate (FITC) (28), we again found that the dominant-negative PAK had a pronounced effect. At early times, the induced TA12 cells, which express higher amounts of dominant-negative PAK, had nearly twice the annexin binding as did the noninduced and control cells (Fig. 4B). At later times, the “leaky” noninduced TA12 cells reached a level of annexin binding similar to that of the induced cells, which plateaued at a level about twice that of control Jurkat cells (Fig. 4B). Thus, blocking PAK function during Fas-induced apoptosis inhibited the morphological changes but accelerated the PS externalization in the membrane, whereas the nuclear modifications were unaffected.

We have identified a different mechanism through which PAK2 becomes activated—its proteolytic cleavage in response to activation of the caspase protease cascade by apoptotic stimuli. Because PAK2 is ubiquitously distributed in mammalian tissues, it is likely to play a role in caspase-mediated apoptotic events in a variety of systems. Our data suggest that the proteolytic release of intact PAK2 NH2- and COOH-terminal fragments affects the actinomyosin system of the apoptotic cell. The caspase cascade has been termed the “executioner” of cell death (3): PAK2 appears to be one of its many swords.

  • * Present address: Max-Planck-Institut für Infektionsbiologie, Abt. Molekulare Biologie, Mobijoustrassse 2, D-10117 Berlin, Germany.


View Abstract

Stay Connected to Science

Navigate This Article