Caspase-3-Generated Fragment of Gelsolin: Effector of Morphological Change in Apoptosis

See allHide authors and affiliations

Science  10 Oct 1997:
Vol. 278, Issue 5336, pp. 294-298
DOI: 10.1126/science.278.5336.294


The caspase-3 (CPP32, apopain, YAMA) family of cysteinyl proteases has been implicated as key mediators of apoptosis in mammalian cells. Gelsolin was identified as a substrate for caspase-3 by screening the translation products of small complementary DNA pools for sensitivity to cleavage by caspase-3. Gelsolin was cleaved in vivo in a caspase-dependent manner in cells stimulated by Fas. Caspase-cleaved gelsolin severed actin filaments in vitro in a Ca2+-independent manner. Expression of the gelsolin cleavage product in multiple cell types caused the cells to round up, detach from the plate, and undergo nuclear fragmentation. Neutrophils isolated from mice lacking gelsolin had delayed onset of both blebbing and DNA fragmentation, following apoptosis induction, compared with wild-type neutrophils. Thus, cleaved gelsolin may be one physiological effector of morphologic change during apoptosis.

A conserved family of aspartate-specific cysteinyl proteases (caspases) has been identified as critical mediators of apoptosis in Caenorhabditis elegansand mammals (1, 2). Although multiple protein substrates of caspases have been found, the functional significance of the substrates is poorly understood (3). We reasoned that an unbiased approach to determine proteins that were the best substrates of caspase-3 in vitro would yield a physiologically relevant substrate. Therefore, we constructed a protein library by translating a murine embryo cDNA library in vitro (4) and tested the translated proteins for their sensitivity to caspase-3 cleavage. To facilitate screening, we separated the cDNAs into small pools before in vitro translation and incorporated [35S]methionine into the translation mix to label the proteins. The translated pools were separated into two parts; one portion was incubated with active caspase-3, and the other portion was incubated with caspase-3 inactivated by the addition ofN-acetyl-Asp-Glu-Val-Asp–fluoromethyl ketone (DEVD-fmk) (2). The reaction products were resolved by SDS–polyacrylamide gel electrophoresis (PAGE), and protein targets of caspase-3 were identified by comparing the pattern of35S-labeled proteins in the samples treated with active and inactive caspase-3. To avoid false positives, we found it necessary to carefully titrate the amount of enzyme. We used an amount sufficient to cleave 80% of a known substrate of caspase-3, baculovirus protein p35 (Fig. 1A) (5).

Figure 1

Identification of gelsolin as a substrate of caspase-3. (A) Autoradiogram of [35S]methionine-labeled proteins treated with purified active (–) or inactivated [by addition of a caspase-3 inhibitor, DEVD-fmk (+)] recombinant caspase-3. Lanes 1 through 6 contain the translated proteins from six different representative pools of cDNAs (23). The arrows indicate a caspase-3–sensitive protein that is cleaved from an apparent size of 65 kD to 48 kD. This protein was identified as gelsolin. (B) Cleavage of bacterially expressed murine gelsolin (16) by caspase-3. Gelsolin (13.5 μg) and caspase-3 (6 ng/ml) were incubated at 37°C for the indicated times in 6 mM tris-HCl (pH 7.5), 1.2 mM CaCl2, 5 mM dithiothreitol (DTT), 1.5 mM MgCl2, and 1 mM KCl in a 25 μl volume. A volume of 2 μl was resolved on a gel and stained with Coomassie blue. Bands of apparent size of 48 and 40 kD, corresponding to actual molecular sizes of 41 and 39 kD, respectively, are seen.

One thousand cDNA pools, each containing 100 cDNA clones (100,000 cDNA clones total), were screened to identify substrates of caspase-3. In three different pools, incubation of active caspase-3 with labeled proteins reduced the intensity of a 65-kD band and generated a new band at 48 kD (Fig. 1A, pool 4). To identify the cDNA clone encoding the 65-kD protein, the DNA from the positive pools was used to transformEscherichia coli, and the DNA prepared from single colonies was screened again. DNA sequencing and polymerase chain reaction analyses of the single positive clone from all three pools identified the 65-kD protein as the partial sequence of gelsolin from residues 142 to 731.

Caspase-3 rapidly cleaved full-length recombinant murine gelsolin, confirming that gelsolin is a substrate for caspase-3 (Fig. 1B). Gelsolin was also cleaved upon incubation with cell extracts prepared from cells that were induced to undergo apoptosis by Fas activation (6). However, no gelsolin cleavage activity was observed in cell extracts prepared from untreated cells. Protein microsequencing by Edman degradation was used to determine the NH2-terminal sequences of the cleavage products and indicated a cleavage site between residues Asp352 and Gly353 of murine gelsolin, so that products had actual molecular masses of 39 (NH2-terminal) and 41 (COOH-terminal) kD. The sequence in this region, Asp349-Gln-Thr-Asp-Gly353, is consistent with the known requirements for efficient cleavage by caspase-3 (2) and is conserved between mouse, human, and porcine gelsolin (7). Cleavage of gelsolin also resulted in dissociation of the 39- and 41-kD cleavage products when assayed by size-exclusion chromatography (8).

To examine cleavage of gelsolin during apoptosis in vivo, initially we used a model cell system in which apoptosis is highly inducible (9). The assay used fibroblasts that expressed a chimeric receptor composed of the extracellular and transmembrane domains of murine CD4 and the cytoplasmic domain of Fas. Apoptosis was induced by antibody cross-linking of the extracellular CD4 domains. Gelsolin was cleaved into the predicted 39- and 41-kD products 30 min after apoptosis induction (Fig. 2A). Cleavage of gelsolin occurred early and was comparable to the time course of cleavage of poly–adenosine diphosphate ribose polymerase (PARP), a substrate of caspase-3 (Fig. 2A). Both gelsolin and PARP cleavage was blocked by the cell-permeable inhibitor of caspase-3,N-benzyloxycarbonyl-Val-Ala-Asp–fmk (zVAD-fmk) (10). Gelsolin cleavage was specific, because another cytoskeletal protein, filamin, was not cleaved under identical conditions (Fig. 2A). Thus, gelsolin was specifically cleaved by a caspase-3–like enzyme in vivo, and this cleavage is an early step in Fas-mediated apoptosis.

Figure 2

Gelsolin was cleaved in apoptosis induced by Fas and TNFα. (A) Apoptosis was induced by receptor cross-linking with and without 200 μM zVAD-fmk (caspase inhibitor) in murine L929 cells as described (9). The cell lysates from an equal number of cells were analyzed by SDS-PAGE and immunoblotted with anti-gelsolin [top panel; polyclonal antibody (16)], anti-PARP (middle panel; Enzyme Systems Products), and anti-filamin (bottom; Sigma). The immunoblots were visualized by ECL (Amersham). (B) Human neutrophils were isolated with the neutrophil isolation medium (Cardinal Associates) and resuspended in RPMI with 10% fetal bovine serum (FBS) (24). Equal numbers of cells were used to prepare cell lysates after incubation with FAS antibody (1 μg/ml, top) or 10 μg/ml TNF and 10 ng/ml cycloheximide (bottom). The lysates were separated by SDS-PAGE and analyzed by immunoblotting with a monoclonal antibody to gelsolin (Sigma), which detects only the COOH-terminal 41-kD fragment. (C) Human neutrophils were stained at different time points after isolation from blood, using annexin V apoptosis kit (Clontech), and the number of live cells was estimated by counting the number of annexin- and propidium iodide–negative cells (24).

Neutrophils purified from human blood express large amounts of gelsolin and undergo spontaneous apoptosis. The rate of neutrophil apoptosis can be further enhanced by cross-linking of Fas or treatment with tumor necrosis factor α (TNFα) and cycloheximide (11). To determine if gelsolin is also cleaved when neutrophils undergo apoptosis, we analyzed neutrophil lysates with a monoclonal antibody to gelsolin that recognizes an epitope in the COOH-terminal half of gelsolin. During spontaneous apoptosis in neutrophils, gelsolin amounts decreased and a 41-kD fragment appeared with time (Fig. 2B)—a size identical to fragments generated by cleavage of gelsolin by caspase-3. Increasing the rate of neutrophil apoptosis by cross-linking Fas with antibodies or treatment with TNFα and cycloheximide resulted in an increase in the rate of gelsolin cleavage and appearance of the 41-kD fragment (Fig. 2, B and C). Thus, the cleavage of gelsolin observed in vitro also occurred during apoptosis of neutrophils.

To determine the functional significance of gelsolin cleavage, we examined the activities of cleaved and native gelsolin, using pyrene-actin fluorimetry. In this assay, the conversion of filamentous (F) actin to monomeric (G) actin is monitored by the 25-fold difference in fluorescence between the two states. Native gelsolin severs actin polymers in a Ca2+-dependent manner (12, 13), whereas caspase-3–cleaved gelsolin severed actin polymers independent of Ca2+ (Fig. 3A, top). Gelsolin has both actin monomer–binding and F-actin–severing activities (12, 13). However, the cleaved gelsolin preferentially severed actin filaments even when briefly incubated with excess monomeric actin (Fig. 3A, bottom), suggesting that the cleaved gelsolin generated in cells during apoptosis may preferentially sever actin filaments rather than bind monomeric actin. Next, we used permeabilized fibroblasts to determine the ability of cleaved gelsolin to depolymerize the cytoskeletal actin filaments. Cleaved gelsolin depolymerized the actin cytoskeleton in a Ca2+-independent manner, whereas uncut gelsolin was inactive in the presence of EGTA (Fig. 3B).

Figure 3

Caspase-3–cleaved gelsolin is activated and severs actin filaments in a Ca2+-independent manner. (A) (Top) Ca2+-independent depolymerization of actin filaments in vitro. We incubated 0.5 μM polymerized pyrene-labeled actin and 50 nM gelsolin (circles) or caspase-3–cleaved gelsolin (triangles) in 2 mM tris-HCl (pH 7.5), 0.5 mM adenosine triphosphate, 0.2 mM DTT, 2 mM MgCl2, and 150 mM KCl, with CaCl2 (0.2 mM; open circles and triangles) or with EGTA (1 mM; solid circles and triangles). The change in pyrene fluorescence with time is shown. (Bottom) Cleaved gelsolin severs F-actin faster than a complex with G-actin is formed. Cleaved gelsolin (20 nM final concentration) was directly added to 0.5 mM F-actin (triangles), or mixed with a twofold molar ratio of G-actin and immediately added (open diamonds) or incubated for 10 min (closed diamonds) and then added to 300 μl of 0.5 μM F-actin. (B) Cleaved gelsolin severs the actin cytoskeleton. Permeabilized embryonic fibroblasts (Gsn–/–) were incubated with gelsolin or cleaved gelsolin and the actin filaments were visualized by staining with TRITC-phalloidin, as described (25). (C) The NH2-terminal gelsolin cleavage fragment induces breakdown of the cytoskeleton. The cells were injected with DNA encoding the NH2-terminal (top) or COOH-terminal (bottom) fragments of gelsolin and with a plasmid encoding GFP, and were stained for actin filaments with TRITC-phalloidin 5 hours later. The arrows point to the injected cells. The right panel indicates GFP expression, and the left panel indicates actin staining (TRITC-phalloidin) of the same field (26). (D) Adenovirus expressing the NH2-terminal gelsolin fragment induces apoptosis. A7 cells (a human melanoma cell line) were infected with adenovirus, prepared as described (27), expressing the NH2-terminal caspase-3 cleavage fragment of gelsolin (residues 1 through 352, top) or full-length gelsolin (1 through 731, bottom). (E) The NH2-terminal gelsolin cleavage fragment induces changes in cell morphology and acts downstream of caspases. The cells were injected with DNA as in (C) and were incubated with dimethyl sulfoxide (left panels) or 100 μM zVAD-fmk (right panels). The injected cells appear green because of GFP expression.

The actin-severing activity of gelsolin resides in the NH2-terminal region, residues 1 to 160 (12). To test if the NH2-terminal fragment of gelsolin generated by caspase-3 cleavage could depolymerize actin in vivo, we microinjected DNA encoding the NH2-terminal fragment (1 to 352) or the COOH-terminal fragment (353 to 731) into fibroblasts. Expression of the NH2-terminal fragment caused a rapid depolymerization of the actin cytoskeleton, whereas the COOH-terminal fragment had no effect on actin filaments (Fig. 3C). The injection of full-length gelsolin also had no effect on the cell morphology. To determine if gelsolin acts directly to depolymerize actin filaments or indirectly by activating caspases, we incubated the microinjected cells with a caspase inhibitor, zVAD-fmk. Inhibition of caspases did not block the changes in cell morphology induced by the expression of the NH2-terminal fragment of gelsolin (Fig. 3E). We also constructed adenoviral vectors expressing the NH2-terminal fragment (1 to 352) or full-length gelsolin. The adenovirus vector expressing the NH2-terminal fragment caused rapid cell death in A7 melanoma cells, with both morphologic changes and nuclear fragmentation in the majority of cells within 48 hours of application [determined by terminal deoxytransferase (TdT)–mediated deoxyuridine triphosphate (dUTP) nick-end labeling (TUNEL) staining (14)], whereas the full-length construct had no effect (Fig. 3D). A similar result was observed in other cell types, including M2 melanoma cells and NIH 3T3 cells. Both p53+/+ and p53–/– murine embryo fibroblasts (15) also displayed morphologic changes of apoptosis within 48 hours of infection by the gelsolin NH2-terminal fragment-expressing adenovirus. Thus, cleavage of gelsolin generated an NH2-terminal fragment that depolymerizes the actin cytoskeleton in a Ca2+-independent manner, induces cell death, and may be a downstream effector of the morphological changes that are observed during apoptosis.

We used gelsolin null cells derived from Gsn–/–mice (16) to examine the importance of gelsolin in apoptosis occurring during physiological stimuli. The course of apoptosis in peritoneal neutrophils treated with TNF plus cycloheximide (TNF+CHX) was assessed by videomicroscopy. Wild-type neutrophils developed blebs as early as 48 min after TNF+CHX, and 52% of cells (32 of 61) had begun blebbing by 6 hours (Fig. 4A). In contrast (Fig. 4B), Gsn–/– neutrophils did not develop blebs until 2 hours 2 min after TNF+CHX, and at 6 hours, only 12% of cells (10 of 83, P < 0.001) had blebs. This delay in the progression of apoptosis appeared to contribute directly to a delay in DNA fragmentation in Gsn–/– cells compared with wild-type cells, as assessed in TUNEL assays (Fig. 4C) and DNA analysis by electrophoresis (17). No difference in caspase-3 activation was observed when Gsn–/– and wild-type neutrophils were compared, as assessed with the peptide substrate DEVD–7-amino-4-trifluoromethylcoumarin (DEVD-AFC) (18), suggesting that the differences seen were from a defect in apoptotic progression downstream of caspase-3 activation. Similar differences in apoptotic progression were seen in Gsn–/– neutrophils in response to treatment with a monoclonal antibody to Fas. Our results may explain the previously observed moderate neutrophilia of Gsn–/– mice, which have approximately twice the number of circulating neutrophils as wild-type mice (16).

Figure 4

Gelsolin null neutrophils have a reduced rate of cell death. Neutrophils were isolated from the peritoneal exudates of wild-type and Gsn–/– mice 5 hours after thioglycollate administration (16). Neutrophils were washed in PBS and then minimum essential medium, then were suspended in RPMI with 10% FBS at 106 cells/ml. mTNFα and cycloheximide (10 μg/ml) were used to induce apoptosis. (A and B) Comparison of blebbing onset in wild-type (A) versus Gsn–/– (B) neutrophils. Arrows indicate blebbing cells. Videomicroscopy was performed with differential interference contrast optics on a Zeiss Axiovert 405M inverted microscope. Images were collected with a Hamamatsu C2400 video camera (Photonic Microscopy, Bridgewater, New Jersey) and recorded on a Panasonic TQ-3038F video recorder. (C) Comparison of DNA fragmentation in Gsn–/– and wild-type neutrophils. The fraction of apoptotic cells as determined by FACS analysis in a TUNEL assay are shown (14). n = 3 with error bars.

To confirm the role of cleavage of gelsolin in the morphologic changes of apoptosis, we used the human malignant cell line HeLa, which does not express gelsolin and is relatively resistant to apoptosis (19). Gelsolin-expressing, stably transfected HeLa cell lines were more sensitive to human TNFα than were HeLa sublines transfected with control vector alone. Some 72%, 52%, and 38% of the cells from three different gelsolin-expressing HeLa sublines displayed cell rounding and blebbing after treatment with TNF+CHX for 12 hours, while only 8% of the cells from a control line showed these changes.

Caspase or caspase-like cleavage of multiple proteins has been described (3, 20), but few of these are known to have direct physiological significance in the morphologic changes and nuclear degradation that are hallmarks of apoptosis. Using an unbiased approach, we found that the actin-modulating protein gelsolin is the most prominent direct substrate of caspase-3 in murine embryos. Our data also indicate that gelsolin is a probable in vivo target of the apoptotic caspase-initiated cascade and that the gelsolin fragment mediates, in part, the morphologic changes of apoptosis. Blockade or enhancement of gelsolin cleavage might retard or enhance apoptosis in multiple cell types. Gelsolin is the founding member of an evolutionarily conserved family of proteins that extends toDictyostelium and Drosophila, and in humans consists of at least six proteins, whose expression is tissue-specific (21). Apoptotic cleavage of other gelsolin family members may also occur, and it is possible that these gelsolin homologs have similar roles in mediating apoptotic cytoskeletal changes in specific cell types and tissues. Gelsolin itself is widely expressed in adult mammalian tissues (13, 16), and its expression is specifically down-regulated in many human neoplastic lesions, including bladder, breast, and colon cancer (22). These observations on the role of gelsolin in apoptosis suggest that gelsolin down-regulation in tumors may be one mechanism by which tumors evade apoptotic signaling pathways.

  • * These authors contributed equally to this work.

  • To whom correspondence should be addressed. E-mail: kwiatkowski{at}


View Abstract

Navigate This Article