Developmental Biology 207, 14–25 (1999) Article ID dbio.1998.9086, available online at http://www.idealibrary.com on View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Elsevier - Publisher Connector gas2 Is a Multifunctional Involved in the Regulation of and Chondrogenesis in the Developing Mouse Limb

K. K. H. Lee,*,1,2 M. K. Tang,* D. T. W. Yew,* P. H. Chow,* S. P. Yee,† C. Schneider,‡,§ and C. Brancolini‡,§,1 *Department of Anatomy, Faculty of Medicine, Chinese University of Hong Kong, Shatin, Hong Kong, People’s Republic of China; †Cancer Research Laboratories, London Regional Cancer Centre, 790 Commissioners Road East, London, Ontario, Canada; ‡Laboratorio Nazionale Consorzio Interuniversitario Biotecnologie Area, Science Park Padriciano 99, Trieste, Italy; and §Dipartimento Scienze e Tecnologie Biomediche Facolta’ di Medicina, University of Udine, p. Kolbe 4, Udine, Italy

The growth-arrest-specific 2 (gas2) gene was initially identified on account of its high level of expression in murine fibroblasts under growth arrest conditions, followed by downregulation upon reentry into the cycle (Schneider et al., Cell 54, 787–793, 1988). In this study, the expression patterns of the gas2 gene and the Gas2 peptide were established in the developing limbs of 11.5- to 14.5-day mouse embryos. It was found that gas2 was expressed in the interdigital tissues, the chondrogenic regions, and the myogenic regions. Low-density limb culture and Brdu incorporation assays revealed that gas2 might play an important role in regulating chondrocyte proliferation and differentiation. Moreover, it might play a similar role during limb myogenesis. In addition to chondrogenesis and myogeneis, gas2 is involved in the execution of the apoptotic program in hindlimb interdigital tissues—by acting as a death substrate for caspase enzymes. TUNEL analysis demonstrated that the interdigital tissues underwent apoptosis between 13.5 and 15.5 days. Exactly at these time points, the C-terminal domain of the Gas2 peptide was cleaved as revealed by Western blot analysis. Moreover, pro-caspase-3 (an enzyme that can process Gas2) was cleaved into its active form in the interdigital tissues. The addition of zVAD-fmk, a caspase enzyme inhibitor, to 12.5-day-old hindlimbs maintained in organ culture revealed that the treatment inhibited interdigital cell death. This inhibition correlated with the absence of the Gas2 peptide and pro-caspase-3 cleavage. The data suggest that Gas2 might be involved in the execution of the apoptotic process. © 1999 Academic Press Key Words: apoptosis; chondrogenesis; growth-arrest-specific-2 gene; mouse limb.

INTRODUCTION carboxyl-terminal domain by caspase enzymes. These aspartic-specific cysteine protease are the principal effec- gas2 is a member of a family of six that are highly tors of the apoptotic program, by cleaving specific cellular expressed in 3T3 fibroblasts during growth arrest (Schneider targets or death substrates (reviewed by Martin and Green, et al., 1988; Brancolini et al., 1992). The products of these 1995; Jacobson et al., 1997). Removal of the carboxyl- genes regulate both negatively and positively cell prolifera- terminal domain of Gas2 dramatically reorganizes the mi- tion and apoptosis (Del Sal et al., 1992; Fabbretti et al., crofilament system and as a result unveils a potent change 1995; Goruppi et al., 1997). The product of gas2 is a in the shape of the affected cells. Therefore, the processing component of the microfilament system and is now be- of Gas2 during apoptosis by caspase enzymes could play an lieved to be a cell death substrate (Brancolini et al., 1992, important role in mediating the complex morphological 1995). During apoptosis, the Gas2 peptide is processed at its changes that characterize cell death. Furthermore, Gas2 is a promising death substrate since its expression is also 1 Both authors contributed equally to this work. coupled to increased susceptibility to apoptosis (Brancolini 2 To whom correspondence should be addressed. Fax: (852) 2603 et al., 1997a). The relationships between microfilament 5031. E-mail: [email protected]. organization and Gas2 have also been demonstrated during

0012-1606/99 $30.00 Copyright © 1999 by Academic Press 14 All rights of reproduction in any form reserved. gas2 in Limb Development 15

G0 3 G1 transition of the cell cycle. Gas2 is rapidly MATERIALS AND METHODS phosphorylated after treatment of growth-arrested cells with various growth factors and the hyperphosphorylated Embryos form of Gas2 is relocalized to the membrane ruffles formed Day-10.5 to Day-14.5 embryos were obtained from pregnant ICR at the edges of the cells (Brancolini and Schneider, 1994). In mice (Chinese University of Hong Kong). The presence of a vaginal vitro studies implicated another potential role for Gas2, in plug was designated as embryonic day 0.5. The mice were killed by the regulation of keratinocytes (Manzow et al., 1996) and cervical dislocation and the embryos were isolated from the de- muscle (Cowled et al., 1994) differentiation. cidua in prewarmed Dulbecco’s phosphate-buffered saline contain- Although a great deal is now known of the physiological ing 0.4% bovine serum albumin (PB1; Sigma, St. Louis, MO). functions of gas2 in vitro, little remains known of its Whole embryos and hindlimbs isolated from embryos were either functional role in vivo. Therefore, we have examined the fixed in 4% cold paraformaldehyde and then used for in situ hybridization studies or directly used for culture studies. expression pattern of gas2 in the mouse embryonic limb during development. The vertebrate limb has traditionally been a favorite model for studying apoptosis (Lee et al., Cell Culture 1993; Zou and Niswander, 1996; Ganan et al., 1996), The distal regions of 12.5-day forelimbs were harvested from chondrogenesis (Fell, 1935; Zimmermann et al., 1982; embryos in PB1 medium. After rinsing the limb fragments with Zanetti and Solursh, 1984; Lee et al., 1994; Goff and Tabin, PBS, they were dissociated by incubation with 0.5% trypsin and 1997), myogenesis (Lee and Sze, 1993; Bladt et al., 1995; 0.25% pancreatin in MEM–Hepes medium (Sigma) containing Webb et al., 1997), tendon development (Patel et al., 1996), 2.2% sodium bicarbonate, for 30 min at 4°C then for 10 min at and tissue patterning (reviewed in Cohn and Tickle, 1996). 37°C followed by trituration with a Pasteur pipet to mechanically disrupt the tissues. The enzymatic reaction was inhibited by the The amenability of the limb to experimental manipulation addition of 10% fetal bovine serum (FBS; Gibco BRL) and all and the vast accumulation of information currently avail- undissociated clumps of cells were removed by filtration through a able regarding the cellular and molecular basis of its devel- nylon filter (21-␮m pore size) to produce a single-cell suspension. opment have also made the limb an ideal model for explor- The dissociated cells were pelleted by centrifugation at 250g for 3 ing the developmental function of newly identified genes. min and then resuspended in Dulbecco’s modified essential me- In this study, we have focused mainly on two aspects of dium (DMEM, Sigma) containing 10% FBS. Cell concentration was limb development, interdigital cell death and chondrogen- established with an improved Neubauer hemocytometer and vi- ability was assessed with Trypan blue. The cells were plated out at esis, because gas2 has been implicated in apoptosis, cell 5 5 ϫ 10 cells/ml onto 0.1% gelatin-coated coverslips housed inside growth arrest, and differentiation in vitro. The spatial and four-well tissue culture plates. The cultures were maintained at temporal patterns of programmed cell death in the inter- 37°C and 5% CO2 for 1–7 days. Brdu was added to some of the digital regions of 12.5- to 14.5-day-old mouse autopodia are cultures, 6 h before harvest, so that the extent of cell proliferation now well established (Lee et al., 1994). Bmp-2, -4, and -7 are in these cultures could be established (Brdu cell proliferation kit, now believed to be the initiation signals for the induction of Amersham Life Sciences). Fluorescently-labeled peanut agglutinin interdigital cell death (Zou and Niswander, 1996; Ganan et (TRITC-PNA, Sigma) was used to confirm the presence of chon- drogenic cells in the cultures. PNA is the earliest known marker for al., 1996; Macias et al., 1997). However, no genes have yet chondrogenic cells (Stringa and Tuan, 1996). Briefly, the cultures been identified to be directly involved in the execution of were fixed for1hincold 70% ethanol and then stained with 1 the apoptotic program in the limb. gas2 is a good candidate ␮g/ml of TRITC-PNA for 30 min at room temperature. The for this role by acting as a death substrate that could be specimens were washed several times in PBS and viewed under a activated by caspase processing (Brancolini et al., 1995; confocal microscope. Porter et al., 1997). The development of cartilage in the limb involves a series of Organ Culture precisely timed acts that include cell condensation, cell move- ment, changes in cell shape, and programmed cell death (Fell, Intact 12.5-day hindlimbs were isolated and placed on a piece of sterile Millipore filter (0.22 ␮m, Millipore, Bedford, MA) mounted 1935; Ede and Wilby, 1981; Zimmermann et al., 1982; on a non-toxic stainless mesh inside the central well of an organ Solursh, 1989a; Lewinson and Silbemann, 1992). These mor- culture dish (Falcon, Lincon Park, NJ). BGJb was introduced into phogenetic events are accompanied by alteration in the ar- the central well while sterile distal water was placed in the outer rangement of the cell’s cytoskeleton. It has been revealed that well. Some of the limbs were cultured in the presence of 100 ␮M limb mesenchyme cultured in the presence of cytochalasin D, Z-Val-Ala-Asp-fluoromethylketone (zVAD-fmk, Calbiochem) caspase a drug that disrupts the organization of actin fibers, causes inhibitor. These limbs were incubated for 6–24 h at 37°C and cells to round up and differentiate into chondrocytes (Archer 5% CO2. et al., 1982; Zanetti and Solursh, 1984). Gas2 is a component of the microfilament network system codistributed with actin Histology fibers (Brancolini et al., 1992). Therefore, we have examined All cultures and whole limbs were freshly fixed in cold 4% gas2 expression in limb cell cultures and in the developing paraformaldehyde at 4°C for 12 h. The specimens were then limb to ascertain whether gas2 plays a role in the regulation of dehydrated in a graded series of cold ethanol, cleared in xylene, chondrogenesis. embedded in wax, and stored at 4°C. When required, the specimens

Copyright © 1999 by Academic Press. All rights of reproduction in any form reserved. 16 Lee et al.

were sectioned at 7 ␮m and mounted onto to TESPA-treated slides. All solutions used were RNase free.

TUNEL Analysis The presence of apoptotic cells in 12.5- to 14.5-day hindlimbs and 11.5-day embryos was detected using an ApopTag kit (Oncor, Gaithersburg, MD). Deparaffinized limb sections were digested in proteinase K, quenched in 2% H2O2 in PBS, and equilibrated in buffer. The sections were then reacted with terminal deoxynucleo- tidyl transferase and digoxigenin labeled. Peroxidase-conjugated digoxigenin antibody and DAB were used to view the location of apoptotic cells. The sections were finally counterstained in methyl green dye.

FIG. 1. Western blot analysis of Gas2 expression during mouse In Situ Hybridization embryonic development. Digoxigenin-labeled sense and antisense riboprobes were synthe- sized from a 500-bp mouse gas2 cDNA fragment inserted into pBSKII (Stratagene) using T3 and T7 RNA polymerase (Boehringer- caspase-3 (Santa Cruz Biotechnology), or anti-tubulin antibodies. Mannheim Biochemica), respectively. In situ hybridization was The blots were then rinsed three times with Blotto-Tween 20 and performed on wax sections (Wilkinson, 1992). Briefly, sections of reacted with peroxidase-conjugated goat anti-rabbit antibody the specimens were dewaxed, cleared in xylene, and hydrated. The (Southern Biotechnology) for 1 h. Finally, the blots were washed sections were then treated with 10 ␮g/ml of proteinase K for 10 several times in Blotto-Tween 20, rinsed in PBS, and developed min, postfixed in 4% paraformaldehyde, and incubated in hybrid- with an ECL kit (Amersham Life Sciences). ization buffer containing 1 ␮g/ml of digoxigenin-labeled antisense riboprobes. The sense probe was used as the control. The hybrid- ization temperature for gas2 was 58°C, and the incubation time Scanning Electron Microscopy was 18 h. Localization of transcripts was visualized using alkaline phosphatase-conjugated antidigoxigenin Fab fragments and NBT/ To examine the morphology of interdigital cells during apopto- BCIP as the chromagen. sis, 13.5-day hindlimbs were fixed in 3 M glutaraldehyde made up in 0.1 M sodium cacodylate for 4 h. The specimens were then washed with PBS, postfixed with 1% osmium tetroxide, dehy- Immunohistochemistry drated, and critical-point dried. A crack was made extending from the proximal to the distal regions of the limb to expose the For immunohistochemical analysis, embryo and limb sections interdigital cells. Finally, the samples were coated with gold and were dewaxed in xylene and rehydrated. The sections were then viewed under a scanning electron microscope. microwaved (400 W) for 4 min in the presence of 6 M urea. After several washes in H2O2, the sections were postfixed in 3% parafor- maldehyde and then washed with 0.1 M glycine made up in PBS, pH 7.5. Incubations with Gas2 antibody (1.5 mg/ml) were per- RESULTS formed in TBS buffer (30 mM Tris–HCl, pH 7.5, 150 mM NaCl, 1% Triton-X 100, 3% BSA) at 1/200 dilution for 12 h at room tempera- Gas2 Expression in the Developing Embryo ture inside a humid chamber. After several washes in TBS buffer, the sections were incubated with 1/2000 dilution of anti-rabbit The steady-state expression of Gas2 , at different alkaline–phosphatase conjugate (Southern Biotechnology) in TBS stages of mouse embryonic development, was examined by buffer for1hat37°C. The immunocomplexes were then localized Western blot analysis. The results revealed that, for the with NBT/BCIP (Boehringer-Mannheim) or DAB substrates. Fi- same concentration of total protein, there was a dramatic nally the sections were dehydrated, mounted onto coverslips, and increase in Gas2 expression between 10.5-, 11.5-, and 13.5- viewed with a Ziess Axiovert 35 microscope. day-old embryos (Fig. 1). The expression pattern of Gas2 in 11.5- and 13.5-day-old Immunoblotting embryos was examined by immunohistochemistry (Table 1). In the 11.5-day embryo, Gas2 was strongly expressed in For Western immunoblotting, were extracted from 12.5- the soft connective tissue of the face and trunk. Strong to 15.5-day autopodia after lysis in sample buffer and sonicated as expression was also obtained in the intervertebral tissues previously described (Brancolini et al., 1992). The proteins were but not in the vertebral cartilage (Fig. 2A). TUNEL analysis transferred to a 0.2-␮m nitrocellulose filter (Amersham Life Sci- revealed that extensive cell death was present in the soft ences) using a semidry blotting apparatus (transfer buffer, 20% methanol, 48 mM Tris, 39 mM glycine, and 0.0375% SDS). After connective tissues of the face (Fig. 2B) and the intervertebral staining with Ponceau S, the nitrocellulose sheets were saturated tissues (Fig. 2C)—in regions where Gas2 was expressed. for1hinBlotto-Tween 20 (50 mM Tris–HCl, 500 mM NaCl, 5% Low levels of Gas2 expression were also detected in the nonfat dry milk, and 0.1% Tween 20) and incubated for3hatroom brain and neural tube. In the 13.5-day embryo, Gas2 was temperature with anti-pan Gas2, anti-C-terminal Gas2, anti- again strongly expressed in the intervertebral tissues and

Copyright © 1999 by Academic Press. All rights of reproduction in any form reserved. gas2 in Limb Development 17

TABLE 1 Localization of Gas2 Expression in the Developing Embryos

Dorsal Soft Stage Neural root connective Adrenal Vertebral Intervert (days) Brain tube ganglion tissue Heart Lung gland Kidney Liver Lens cartilage tissues

11.5 ϩϩ Ϫ ϩϩϪNA NA NA Ϫ NA Ϫϩϩ 12.5 ϩϩ Ϫ ϩϩϪϪNA ϪϪNA Ϫϩϩ 13.5 ϩϩ Ϫ ϩϩϪϩϪ ϩϪϩϮa ϩϩ

Note. ϩ, Gas2 weakly expressed; ϩϩ, Gas2 strongly expressed; Ϫ, no expression; NA, not available. a Gas2 weakly expressed in vertebral cartilage located cranially.

the soft connective tissues of the head and trunk (Fig. 2D). ized. At day 14.5, gas2 was also expressed by cartilages Low levels of gas2 expression were also detected in the forming the carpals and tarsals. The expression in these developing lung, cortex of the kidney, lens of the eye, and cartilages was restricted to the chondrocytes, while no expres- vertebral cartilage located cranially. sion was found in the perichondrium (Fig. 3L). Moreover, mesenchymal cells in the process of condensing to form tendons also expressed gas2 (Fig. 3E). In the proximal myo- Gas2 Expression Pattern in the Developing Mouse genic regions of 13.5-day hindlimb, some of the myoblasts Limb were capable of expressing Gas2 (Fig. 3I) and examination of The gas2 sense riboprobe was used as the negative con- similar sites in older limbs revealed that the Gas2 protein was trol. No hybridization signal was detected in all limb localized in the myotubules (Fig. 3J). In addition to cartilage sections treated with the sense probe. gas2 antisense ribo- and skeletal muscles, the joint-forming regions strongly ex- probe revealed that, in 11.5-day hindlimbs, gas2 was weakly pressed gas2 (Fig. 3K). expressed by the mesenchymal cells surrounding the per- spective cartilage-forming regions (Fig. 3A). In 12.5-day gas2 Expression Pattern in Chondrogenic Cultures hindlimbs, gas2 was strongly expressed by cells enveloping the chondrogenic primordia of the digits, metatarsals, tibia, The relationship between gas2 expression and chondro- fibula, and femur (Fig. 3B). However, the cells inside the genesis was examined by dissociating 12.5-day forelimbs chondrogenic primordia did not transcribe gas2. The soft into single cells and maintaining them in culture for 1–7 connective tissue of the interdigital regions also expressed days. After 1 day of culture, numerous small round aggre- gas2 (Fig. 3B), but not all of the soft connective tissue cells gates were obtained and these aggregates did not express were capable of expressing gas2. Soft connective tissues gas2 (Fig. 4A). PNA staining revealed that the aggregates that were found distant from the chondrogenic elements were composed of chondrogenic cells (Fig. 4B). Brdu incor- either did not transcribe or transcribed very low levels of poration studies demonstrated that the increased numbers gas2 mRNA. No expression was found in the ectoderm and of cells found within the aggregates were not attributed to a apical ectodermal ridge. The Gas2 protein expression pat- local increase in cell proliferation (Fig. 4E). In the 3-day tern in 12.5-day hindlimbs (Fig. 3C) was essentially similar cultures, the chondrogenic aggregates were composed of to the gas2 mRNA pattern. two distinct cell types, as distinguished by their morphol- Immunohistochemical staining of 13.5-day hindlimbs with ogy. The cells at the center of the aggregates were round in Gas2 antibody revealed that Gas2 expression was maintained appearance, whereas the cells at the periphery were more in the interdigital tissues located proximally (Fig. 3D). In flattened. Moreover, gas2 was expressed in cells located on contrast, no expression was detected distally since apoptosis the periphery but not at the center (Fig. 4C). There were also had progressed into the interdigital tissues of these regions. more mitotic cells present at the center of the chondrogenic Gas2 was also produced by some of the chondrocytes in the aggregates than at the periphery (Fig. 4F). The gas2 expres- stylopod but not in the autopod and zygopod. However, the sion pattern described is similar to those found in 12.5- and mesenchymal cells surrounding the cartilage in the autopod 13.5-day hindlimbs, where it was demonstrated that the and zygopod were strongly stained for Gas2 (Fig. 3D). In the chondrogenic regions did not express gas2, while strong 13.5-day forelimb, developmentally 0.5 day more advanced expression was found in the mesenchymal cells that envel- than the hindlimb, Gas2 was found strongly expressed in oped the cartilage. In day 7 cultures, most of the chondro- pre-hypertrophic and hypertrophic regions of the humerus, genic aggregates were surrounded by several layers of peri- radius, and ulna (Fig. 3F). Gas2 expression was maintained by chondrium. The perichondrial cells did not express gas2, the hypertrophic chondrocytes in these regions well into day while strong expression was obtained from chondrocytes 14.5 (Fig. 3G) until expression was finally lost at day 15.5 (Fig. located at the center of the aggregates (Fig. 4D). Again this 3H)—when these hypertrophic cells die and become mineral- expression pattern was found in the developing cartilage of

Copyright © 1999 by Academic Press. All rights of reproduction in any form reserved. 18 Lee et al.

FIG. 2. Localization of Gas2 in 11.5-day-old (A) and 13.5-day-old (D) embryos by immunohistochemistry. Bars, 500 ␮m. TUNEL analysis of 11.5-day embryos demonstrated the presence of numerous apoptotic cells in the craniofacial mesenchyme (B) and the intervertebral tissues (C). These regions strongly expressed Gas2. Bar, 100 ␮m. FB, forebrain; HB, hindbrain; CM, craniofacial mesenchyme; I, intervertebral tissues; V, vertebral cartilage; H, heart; L, liver; NT, neural tube; LU, lung; A, adrenal grand; K, kidney.

14.5-day and older limbs (Fig. 3L). Brdu incorporation tests day hindlimbs, apoptotic cells were found mainly distrib- performed at day 7 revealed that the mitotic cells were uted in the apical ectodermal ridge and distal mesenchyme mainly distributed in the perichondrium (Fig. 4G). located near digits 1 and 5. No apoptotic cells were found in the interdigital regions (Fig. 5A). On day 14, numerous apoptotic cells were detected in the interdigital zone (Fig. Apoptosis in 12.5- to 13.5-Day Hindlimbs 5B) and by day 14.5 these cells were also found in the The location of apoptotic cells during development of the prospective joints of the developing digits. Scanning elec- autopodium was established by TUNEL analysis. In 12.5- tron micrographs of cells in the interdigital regions of

Copyright © 1999 by Academic Press. All rights of reproduction in any form reserved. gas2 in Limb Development 19

Copyright © 1999 by Academic Press. All rights of reproduction in any form reserved. 20 Lee et al.

13.5-day limbs revealed that the interdigital cells preco- kDa band present (Fig. 6D). However, in the nontreated ciously round up prior to the onset of cell death (Fig. 5C). interdigital tissues both the intact (36 kDa) and cleaved (29 kDa) forms of Gas2 peptides were detected (Fig. 6D). Anal- ysis was also performed with C-terminal Gas2 antibodies. Gas2 Cleavage during Apoptosis The results demonstrated that the cleaved Gas2-positive It has been reported that cleavage of the carboxyl termi- band present in nontreated interdigital tissues was absent nal of the Gas2 peptide by the ICE-like enzyme could when stained with the C-terminal Gas2 antibody. This induce apoptosis in fibroblasts (Brancolini et al., 1995). To implies that the C-terminal domain of Gas2 was deleted determine whether this also applies to the developing limb, (Fig. 6D). We also examined the effects of the zVAD-fmk the autopodia of 12.5- to 15.5-day hindlimbs were harvested inhibitor on caspase-3 cleavage in the interdigital tissues. In and their total protein was extracted for Western blot nontreated samples, the pro- and active forms of caspase-3, analysis. The results showed that in all 12.5- to 15.5-day 32 and 17 kDa, respectively, were present. In the zVAD- specimens examined, there was one strongly stained Gas2- fmk-treated samples, both caspase-3-positive bands were positive band (approximately 36 kDa) present. In the 12.5- also visible, but the cleaved 17-kDa band was very weakly and 13.5-day samples, a smaller Gas2-positive band (ap- stained—indicating that the inhibitor had prevented pro- proximately 29 kDa) was also detected (Fig. 5D). This extra caspase-3 from being processed (Fig. 6E). band was also present in 14.5- and 15.5-day samples but more intensively stained. These results indicate that Gas2 is cleaved when apoptosis is prevalent in the autopod. DISCUSSION Western blot analysis was also performed with caspase-3 antibody to determine whether there was a correlation The gas2 gene encodes for a product that is a component between Gas2 cleavage and caspase-3 expression. An al- of the microfilament system (Brancolini et al., 1992). The most identical pattern of expression was obtained for pro- protein is colocalized with actin fibers and its function has caspase-3 (32 kDa). A pro-caspase-3-positive band was been studied extensively in fibroblast cultures. It has been found in all 12.5- to 15.5-day footplate extracts. Moreover, revealed that Gas2 is involved in the regulation of cell the cleaved active form of caspase-3 (17 kDa) was present at growth and the execution of apoptosis (Brancolini et al., all these time points (Fig. 5D). 1995, 1997a). At present, we have demonstrated that gas2 The effects of zVAD-fmk caspase inhibitor on 12.5-day may play a role in the control of apoptosis and chondrogen- hindlimbs in organ culture were examined. A striking esis in the developing mouse limb. histological difference was found between zVAD-fmk- treated and untreated limbs, after 24 h incubation (Figs. 6A Gas2 and Chondrogenesis and 6B). Limbs treated with the inhibitor possessed signifi- cantly fewer apoptotic cells in the interdigital regions (Fig. The development of the limb skeleton begins with a 6B) than those untreated (Fig. 6A). This difference was less cellular condensation process whereby loosely associated obvious between digits 1 and 2 because apoptosis had prechondrogenic mesenchymal cells aggregate together to already proceeded in these regions, prior to the addition of form a chondrogenic nodule (Fell, 1935). This process in- the caspase inhibitor. Immunohistological staining demon- volves (1) cell migration (Ede and Flint, 1975; Ede and strated the presence of numerous caspase-3-positive cells in Wilby, 1981), (2) rearrangement of the cytoskeleton (re- the interdigital regions of the 13.5-day footplate (Fig. 6C). viewed in Solursh, 1989a), and (3) changes in cell–cell and We investigated whether the inhibitory effects of zVAD- cell–matrix interactions (Zimmermann et al., 1982; Frenz fmk were associated with Gas2. The interdigital regions et al., 1989; Caroline and Kosher, 1991)—rather than di- (between digits 2–3 and 3–4) of zVAD-fmk-treated limbs rectly attributed to a local increase in cell proliferation as were harvested and used for Western blot analysis. The we currently and others (Hornbruch and Wolpert, 1970; results showed that the Gas2 peptide was not cleaved in Janners and Searls, 1970) have demonstrated. Once the zVAD-fmk-treated interdigital tissues, with only the 36- chondrogenic primordium has become established, chon-

FIG. 3. Gas2 expression in 11.5- to 15.5-day forelimbs (FL) and hindlimbs (HL) was determined using gas2 antisense riboprobes (A, B, E, K, L) and Gas2 antibodies (C, D, F–J). (A). In 11.5-day hindlimbs, gas2 is expressed in mesenchymal cells surrounding the presumptive chondrogenic regions. (B) In 12.5-day hindlimbs, gas2 is strongly expressed in the interdigital tissues and cells surrounding the developing cartilage. (C) Immunohistological staining revealed that Gas2 protein is also expressed in the interdigital tissues of 12.5-day hindlimbs but by day 13.5 (D) expression is restricted to the proximal regions of the interdigital tissues. (F, G) Gas2 is strongly expressed by chondrocytes (arrows) in the hypertrophic regions (Hyp) of developing cartilage. Gas2 expression is lost when the hypertrophic chondrocytes die and become mineralized (H). In the myogenic regions, some of the myogenic cells (*) are capable of expressing Gas2 (I) which later become restricted to the myotubules in mature myogenic regions (J). In addition to cartilage and muscles, gas2 is expressed by mesenchymal cells in the joint-forming regions (K) and the developing tendons (E, K), but not the perichondrium (L). C, cartilage; IDZ, interdigital zone; M, myogenic region; F, femur; T, tendon; Uln, ulna; J, joint-forming region; Carp, carpals; Pc, perichondrium. Bars, 100 ␮m.

Copyright © 1999 by Academic Press. All rights of reproduction in any form reserved. gas2 in Limb Development 21 drogenesis proceeds in the center of the condensation, mineralization process through proteoglycan degradation whereas cells located in the periphery retain their mesen- (Yasuda et al., 1995). In the limbs, Gas2 is strongly ex- chymal characteristics (Zimmermann et al., 1982). These pressed by the prehypertrophic and hypertrophic chondro- morphogenetic events can also be dynamically followed in cytes, so it is possible that Gas2 may interact with limb micromass cultures and we have exploited this tech- to control the cartilage mineralization process. nique to define the role the gas2 gene plays in chondrogen- esis. It was found that gas2 was not expressed by limb Gas2 and Apoptosis mesenchymal cells during the early stages of chondrogen- esis, when cells aggregate to form chondrogenic nodules. As fibroblasts undergo apoptosis in culture, the carboxyl- However, as these aggregates developed, gas2 was expressed terminal domain of Gas2 is proteolytically processed by by cells located on the periphery but not the center of the caspase enzymes (Brancolini et al., 1995). It has been chondrogenic nodules. This expression pattern was also revealed that, in vitro, this cleavage process is triggered by found in the limb where all of the developing cartilages caspase-3 but not by caspase-2 (Brancolini et al., 1997b). were surrounded by layers of gas2-positive cells. We specu- The cleavage and removal of the C-terminal domain induce late that the early expression of gas2 may be related to the a dramatic change in cell shape, as a result of the accom- recruitment of cells required to form the chondrogenic panying reorganization of the microfilaments (Brancolini et nodules. This interpretation is conceivable because Gas2 al., 1995). It is worth noting that caspases orchestrate protein is a component of the microfilament system and is morphogenetic changes during apoptosis by processing dif- codistributed with actin fibers in nonmyogenic cells (Bran- ferent regulators of the actin cytoskeleton such as fodrin colini and Schneider, 1994; Brancolini et al., 1992, 1995). RhoGDI and beta-catenin (Cryns et al., 1996; Martin et al., Moreover, the actin cytoskeleton is implicated in many 1996). Fodrin (nonerythroid spectrin) is proteolyzed during cellular activities such as motility, cell division, and the apoptosis. Similarly, D4-GDI, a substrate of caspase-3, is regulation of cell shape—all of which are involved in also proteolyzed during Fas-induced apoptosis (Na et al., chondrogenesis (Solursh, 1989a; Benjamin et al., 1994). The 1996). The dismantling of cell–cell contacts during apopto- lack of gas2 expression in the aggregate’s center may be sis is coupled to a caspase-dependent proteolytic cleavage of related to cell proliferation since large numbers of Brdu- beta-catenin (Brancolini et al., 1997b). positive cells are often associated with the region and Gas2 In the mouse limb, apoptosis plays a crucial role in is a growth arrest gene (Brancolini and Schneider, 1994). We sculpturing the contour of the autopodium and in the have also discovered that, at more advance stages, the gas2 separation of the digits, through the elimination of excess expression pattern is reversed—with cells at the center but tissues from the interdigital regions and the anterior and not the periphery of cartilage-expressing gas2. This reversal posterior necrotic zones (Lee et al., 1993, 1994). Our scan- is not unexpected and is probably related to the way ning electron micrographs revealed that one of the preco- cartilage normally develops. Normally, differentiation pro- cious features of cells dying in the interdigital regions is ceeds in the center of the cartilage while the perichondrium that they take on a rounded morphology. This observation in the periphery continues to proliferate and contribute has also been previously reported by Hurle et al. (1995). additional cells to the center. Therefore, the activation of Brancolini et al. (1995) demonstrated that overexpression of gas2 in the chondrogenic core may be a prerequisite for Gas2 deleted at its C-terminal, but not at its wild-type chondroblasts to differentiate by arresting cell proliferation, form, can induce a dramatic change in the actin cytoskel- but cells in the periphery can continue to multiply as gas2 eton and the morphology of affected cells. A majority of is not expressed in these regions. Indeed, in our Brdu cells overexpressing the deleted form of Gas2 have a con- incorporation studies we have ascertained that chondro- tracted spherical body and numerous finger-like cytoplas- genic regions expressing gas2 generally contained fewer mic processes. At present, we have reported that gas2 mitotic cells. In the yeast hybrid system, we have recently mRNA and peptide are expressed in the interdigital regions discovered that Gas2 can bind with calpain (unpublished of 12.5- and 13.5-day hindlimbs. In addition, Western blot observation). Calpain is a Ca2ϩ-dependent cysteine protein- analysis of total proteins extracted from whole 12.5- to ase. It has been implicated in the regulation of the cartilage 15.5-day autopodia revealed that Gas2 is cleaved at 12.5–

FIG. 4. Gas2 expression and cell proliferation during chondrogenesis in cell culture. (A) In situ hybridization showed that gas2 is not expressed by cells condensing to form chondrogenic aggregates. (B) TRITC-PNA staining confirmed that aggregates are composed of chondrogenic cells. These aggregates are formed as a result of the active recruitment of mesenchymal cells rather than the result of localized cell proliferation as revealed by Brdu incorporation analysis (E). In more mature aggregates, gas2 expression is switched on by chondrogenic cells (arrowheads) around the periphery (C). Correspondingly, there are also more mitotic cells (arrows) present in the center than in the periphery of these aggregates (F). In fully formed chondrogenic aggregates, a stratified layer of perichondrium is evident surrounding the chondrocytes. These perichondrial cells do not express gas2. In contrast, gas2 is actively expressed by chondrocytes located at the center of these chondrogenic nodules (D). In addition, there are more mitotic cells found in the perichondrium than in the chondrogenic center (G). C, chondrogenic aggregates; PC, perichondrium; arrows, Brdu-positive cells mitotic cells; arrowheads, gas2-positive cells. Bars, 100 ␮m.

Copyright © 1999 by Academic Press. All rights of reproduction in any form reserved. gas2 in Limb Development 23

FIG. 6. Effect of caspase inhibitor on interdigital cell death. (A) 12.5-day hindlimbs maintained in organ culture for 24 h produced numerous apoptotic cells (white arrows) in the interdigital regions. (B) In the presence of zVAD-fmk caspase inhibitor, interdigital cell death is largely inhibited. (C) Immunohistological staining of 13.5-day limbs revealed the presence of numerous caspase-3-positive cells (black arrows) in the interdigital regions. (D) Western blot analysis showed that treatment of interdigital tissues with the zVAD-fmk inhibitor prevented Gas2 from being cleaved. In contrast, both the intact and cleaved forms of Gas2 are present in the control (CTL). Staining with Gas2 C-terminal antibody confirmed that the C-terminal of Gas2 is cleaved. (E) zVAD-fmk treatment also inhibited pro-caspase-3 activation in the interdigital tissues. D, digits. Bars, 100 ␮m.

15.5 days. We have previously reported that the interdigital volved in the cascade of interactions that are important for the tissues of 12.5-day hindlimbs are developmentally plastic and execution of the apoptotic program in the interdigital tissues. can be diverted from the apoptotic pathway to form cartilage Moreover, we have detected the presence of the pro- and and soft connective tissues (Lee et al., 1994)—but this plas- active forms of caspase-3 in total proteins extracted from 12.5- ticity is irreversibly lost by day 13.5. It is significant that we to 15.5-day autopodia and the presence of caspase-3 in inter- have found Gas2 cleaved at day 13.5 (a stage when apoptosis is digital cells of 13.5-day hindlimbs. To determine whether clearly visible in the interdigital regions as revealed by Gas2 and caspase-3 are involved in apoptosis, we introduced TUNEL analysis) because it suggests that Gas2 may be in- zVAD-fmk (a cell-permeable protease inhibitor of different

FIG. 5. Apoptosis and Gas2 expression. TUNEL staining demonstrating the presence of apoptotic cells (arrows) in 12.5-day (A) and 14-day (B) hindlimbs. Bars, 200 ␮m. (C) Scanning electron micrograph of the interdigital space of 13.5-day hindlimbs, showing the interdigital cells (AP) rounding up during apoptosis. Bar, 3 ␮m. (D) Western blot analysis of 12.5- to 15.5-day footplates revealed that Gas2 and pro-caspase-3 are expressed and cleaved. Arrows, apoptotic cells; D, digits.

Copyright © 1999 by Academic Press. All rights of reproduction in any form reserved. 24 Lee et al. caspases) into 12.5-day hindlimb cultures. We found that REFERENCES treatment can prevent the onset of apoptosis in the interdigi- tal regions and also Gas2 from being proteolytically processed. Archer, C. W., Rooney, P., and Wolpert, L. (1982). Cell shape and In addition, the inhibitor blocked the cleavage of pro-caspase-3 cartilage differentiation of limb mesenchyme in cell culture. Cell into its active form as revealed by Western blot analysis. Differ. 11, 245–251. These results suggest that the execution of programmed Benjamin, M., Archer, C. W., and Ralphs, J. R. (1994). Cytoskeleton interdigital cell death is orchestrated by a proteolytic cascade of cartilage cells. Microsc. Res. Technol. 28, 372–377. involving caspase-3 and Gas2. However, we have not provided Bladt, F., Riethmecher, D., Isenmann, S., Aguzzi, A., and Birch- direct evidence that caspase-3 can cut Gas2 in the present meier, C. (1995). Essential role for the c-met receptor in the case, but in cultured fibroblasts Gas2 cleavage can be triggered migration of myogenic precursor cells into the limb bud. Nature by caspase-3 (Brancolini et al., 1997b). Recent gene knockout 376, 768–771. Brancolini, C., and Schneider, C. (1994). Phosphorylation of the studies seem to indicate that the caspase-1 (Li et al., 1995) and growth arrest-specific protein Gas2 is coupled to actin rearrange- caspase-3 (Kuida et al., 1996) genes are not required to produce ments during G0 3 G1 transition in NIH 3T3 cells. J. Cell Biol. interdigital cell death because targeted disruption of these 124, 743–756. genes produced mice with normal limbs. However, there are Brancolini, C., Bottega, S., and Schneider, C. (1992). Gas2, a growth at least 11 members of the caspase (Ced-3/ICE) family in arrest-specific protein, is a component of the microfilament mammals (reviewed in Chinnaiyan and Dixit, 1996) and some network system. J. Cell Biol. 117, 1251–1261. of them could play a redundant role in interdigital cell death. Brancolini, C., Benedetti, M., and Schneider, C. (1995). Microfila- The effectiveness of caspase inhibitors in blocking interdigital ment reorganization during apoptosis: The role of Gas2, a pos- cell death has also been reported by other workers in the chick sible substrate for ICE-like proteases. EMBO J. 14, 5179–5190. (Milligan et al., 1995) and the mouse (Jacobson et al., 1996). Brancolini, C., Marzinotto, S., and Schneider, C. (1997a). Suscepti- bility to dependent apoptosis correlates with increased levels Our view that Gas2 is involved in apoptosis during develop- of Gas2 and Gas3 proteins. Cell Death Diff. 4, 247–253. ment is further supported by its expression pattern in the Brancolini, C., Lazarevic, D., Rodriguez, J., and Schneider, C. developing embryo. Gas2 was found strongly expressed in the (1997b). Dismantling cell–cell contacts during apoptosis is intervertebral tissues and the craniofacial mesenchyme— coupled to a caspase dependent proteolytic cleavage of beta- regions that normally undergo extensive apoptosis. In the catenin. J. Cell Biol. 139, 759–771. cartilage of the limbs, the hypertrophic chondrocytes also Caroline, C. N. D., and Kosher, R. A. (1991). Gap junctional undergo apoptosis (Lewinson and Silbemann, 1992) and these communication during limb cartilage differentiation. Dev. Biol. cells strongly express Gas2 and caspase-3 enzyme (unpub- 144, 47–53. lished observation). Nevertheless, there are other regions in Chinnaiyan, A., and Dixit, V. (1996). The cell-death machine. Curr. the embryo, such as the heart, which also experience apopto- Biol. 6, 555–562. Cohn, M. J., and Tickle, C. (1996). Limbs: A model for pattern sis but do not express Gas2. formation within the vertebrate body plan. Trends Genet. 12, 153–157. Gas2 and Myogenesis Cowled, P. A., Ciccarelli, C., Loccia, E., Philipson, L., and Sor- rentino, V. (1994). Expression of the growth arrest-specific (gas) In addition to cartilage, Gas2 is expressed in the myo- gene in senescent murine cells. Exp. Cell Res. 221, 197–202. genic regions. The expression pattern seem to suggest that Cryns, V. L., Bergeron, L., Zhu, H., Li, H., and Yuan, J. (1996). Gas2 may be involved in the fusion of myoblasts to form Specific cleavage of alpha-fodrin during Fas- and tumor necrosis myotubules. Interestingly, Elamrani et al. (1995) proposed factor-induced apoptosis is mediated by an interleukin-1 beta- that myoblast fusion involves calpain cleaving desmin and converting enzyme/ced2 protease distinct from the poly (ADP- vimentin intermediate filaments. In a similar analogy, ribose) polymerase protease. J. Biol. Chem. 271, 31277–31282. perhaps the cleavage of Gas2 by caspase-3 is also involved Del Sal, G., Ruaro, M. E., Philipson, L., and Schneider, C. (1992). in myoblast fusion. This is possible because we have The growth arrest specific gene gas 1 is involved in growth recently detected the presence of caspase-3 in the myogenic suppression. Cell 70, 595–607. regions (unpublished observation). Ede, D. A., and Flint, O. P. (1975). Cell movement and adhesion in the developing chick wing bud: Studies on cultured mesenchyme In conclusion, Gas2 plays a multifunctional role in the cells from normal and talpid3 mutant embryos. J. Cell Sci. 18, development of the limb. The peptide is involved in differ- 301–313. entiation of cartilage and skeletal muscles. Moreover, Gas2 Ede, D. A., and Wilby, O. K. (1981). Golgi orientation and cell is involved in the execution of interdigital programmed cell behaviour in the developing pattern of chondrogenic condensa- death by acting as a death substrate for the cleavage actions tions in chick limb bud mesenchyme. Histochem. J. 13, 615–630. of caspase enzymes. Elamrani, N., Brustis, J. J., Dourdin, N., Balcerzak, D., Poussard, S., Cottin, P., and Ducastaing, A. (1995). Desmin degradation and Ca2ϩ-dependent proteolysis during myoblast fusion. Biol. Cell ACKNOWLEDGMENTS 85, 177–183. Fabbretti, E., Edomi, P., Brancolini, C., and Schneider, C. (1995). We thank L. S. Kung, P. F Kwok, and K. C. Leung for their Apoptotic phenotype induced by over expression of wild-type technical assistance. This work was supported by UPGC Direct gas3/PMP22: Its relation to the demyelinating peripheral neurop- Grant 87/022/DG and Mainline Grant 44M4025. athy CMT1A. Genes Dev. 9, 1846–1856.

Copyright © 1999 by Academic Press. All rights of reproduction in any form reserved. gas2 in Limb Development 25

Fell, H. B. (1935). Experiments of the development in vitro of the tion during CD95-induced apoptosis of cells and cytoplasts avian knee-joint. Proc. R. Soc. London B 116, 316–351. requires ICE/CED-3 protease activity. J. Biol. Chem. 271, 28753– Frenz, D. A., Jaikaria, N. s., and Newman, S. A. (1989). The 28756. mechanism of precartilage mesenchymal condensation: A major Macias, D., Ganan, Y., Sampath, T. K., Piedra, M. E., Ros, M. A., role for interaction of the cell surface with the amino-terminal and Hurle, J. M. (1997). Role of Bmp-2 and OP-1 (Bmp-7) in heparin-binding domain of fibronectin. Dev. Biol. 136, 97–103. programmed cell death and skeletogenesis during chick limb Ganan, Y., Macias, D., Duterque-Coquillaud, M., Ros, M. A., and development. Development 124, 1109–1117. Hurle, J. M. (1996). Role of TGF␤s and BMPs as signals control- Manzow, S., Brancolini, C., Marks, F., and Richter, K. H. (1996). ling the position of the digits and the areas of interdigital cell Expression of growth arrest-specific (Gas) genes in murine kera- death in the developing chick limb autopod. Development 122, tinocytes: Gas2 is specifically regulated. Exp. Cell Res. 224, 2349–2357. 200–203. Goff, D. J., and Tabin, C. J. (1997). Analysis of Hoxd-13 and Milligan, C. E., Prevette, D., Yaginuma, H., Homma, S., Cardwell, Hoxd-11 misexpression in chick limb buds reveals that Hox C., Fritz L. C., Tomaselli, K. J., Oppenheim, R. W., and Schwartz, genes affect both bone condensation and growth. Development L. M. (1995). Peptide inhibitor of the ICE protease family arrests 124, 627–636. programmed cell death of motoneurons in vivo and in vitro. Goruppi, S., Ruaro, E., Varnum, B., and Schneider, C. (1997). Neurons 15, 385–393. Requirement of the phosphatidylinositol 3-kinase-dependent Na, S., Chuang, T. H., Cunningham, A., Turi, T. G., Hanke, J. H., pathway and Src for Gas6-Axl mitogenic and survival activities Bokoch, G. M., and Danley, D. E. (1996). D4-GDI, a substrate of in NIH 3T3 fibroblasts. Mol. Cell. Biol. 17, 4442–4453. CPP32, is proteolyzed during Fas-induced apoptosis. J. Biol. Hornburch, A., and Wolpert, L. (1970). Cell division in the early Chem. 271, 11209–11213. growth and morphogenesis of the chick limb. Nature 226, Patel, K., Nittenberg, R., D’Souza, D., Irving, C., Burt, D., Wilkin- 746–766. son, D. G., and Tickle, C. (1996). Expression and regulation of Hurle, J. M., Ros, M. A., Garcia-Martinez, V., and Ganan, Y. (1995). Cek-8, a cell to cell signalling receptor in developing chick limb Cell death in the embryonic developing limb. Scan. Microsc. 9, buds. Development 122, 1147–1155. 519–534. Porter, A. G., Ng, P., and Janicke, R. V. (1997). Death substrates Jacobson, M. D., Weil, M., and Raff, M. (1996). Role of Ced/ICE- come alive. Bioassays 19, 501–507. family proteases in staurosporine-induced programmed cell Schneider, C., King, R. M., and Philipson, L. (1988). Genes specifi- death. J. Cell Biol. 133, 1041–1051. cally expressed at growth arrest of mammalian cells. Cell 54, Jacobson, M. D., Weil, M., and Raff, M. (1997). Programmed cell 787–793. death in animal development. Cell 88, 347–354. Janners, M. Y., and Searls, R. (1970). Changes in rate of cellular Solursh, M. (1989a). Cartilage stem cells: Regulation of differentia- proliferation during the differentiation of cartilage and muscle in tion. Connect. Tissue Res. 20, 81–89. the mesenchyme of the embryonic chick wing. Dev. Biol. 23, Solursh, M. (1989b). Extracellular matrix and cell surface as deter- 136–165. minants of connective tissue differentiation. Am. J. Med. Genet. Kuida, K., Zheng, T. S., Kuan, C.-Y., Yang, D., Karasuyama, H., 34, 30–34. Rakic, P., and Flavell, R. A. (1996). Decreased apoptosis in the Stringa, E., and Tuan, R. S. (1996). Chondrogenic cell subpopulation brain and premature lethality in CPP32-deficient mice. Nature of chick embryonic calvarium: Isolation by peanut agglutinin 384, 368–372. affinity chromatography and in vitro characterization. Anat. Lee, K. K. H. and Sze, L. Y. (1993). The role of branchial somites in Embryol. 194, 427–437. the development of the appendicular musculature in rat em- Webb, S. E., Lee, K. K. H., Tang, M. K., and Ede, D. A. (1997). bryos. Dev. Dyn. 198, 86–96. Fibroblast growth factors 2 and 4 stimulate migration of mouse Lee, K. K. H., Chan, W. Y., and Sze, L. Y. (1993). Histogenetic embryonic limb myogenic cells. Dev. Biol. 209, 206–216. potential of rat hind-limb interdigital tissues prior to and during Wilkinson, D. G. (1992). “In Situ Hybridization: A Practical Ap- the onset of programmed cell death. Anat. Rec. 236, 568–572. proach.” IRL Press at Oxford Univ. Press, Oxford, New York. Lee, K. K. H., Li, F. C. H., Yung, W. T., Kung, J. L. S., Ng, J. L., and Yasuda, T., Shimizu, K., Nakagawa, Y., Yamamoto, S., Niibayashi, Cheah, K. S. E. (1994). Influence of digits, ectoderm and retinoic H., and Yamamuro, T. (1995). m-calpain in rat growth plate acid on chondrogenesis by mouse interdigital mesoderm in chondrocyte cultures: Its involvement in the matrix mineraliza- culture. Dev. Dyn. 201, 297–309. tion process. Dev. Biol. 170, 159–168. Lewinson, D., and Silbermann, M. (1992). Chondroclasts and Zanetti, N. C., and Solursh, M. (1984). Induction of chondrogenesis endothelial cells collaborate in the process of cartilage resorp- in the limb mesenchymal cultures by disruption of the actin tion. Anat. Rec. 233, 504–514. cytoskeleton. J. Cell Biol. 99, 115–123. Li, P., Allen, H., Banerjee, S., Franklin, S., Herzog, L., Johnston, C., Zimmerman, E., Scharlach, E., and Kaatz, R. (1982). Cell contact McDowell, J., Paskind, M., Rodman, L., Salfeld, J., Towne, E., and surface coat alterations of limb-bud mesenchymal cells Tracey, D., Wardweill, S., Wei, F.-Y., Wong, W., Kaman, R., and during differentiation. J. Embryol. Exp. Morphol. 72, 1–18. Seshadriu, T. (1995). Mice deficient in IL-1␤-converting enzyme Zou, H., and Niswander, L. (1996). Requirement for BMP signalling are defective in production of mature IL-1␤ and resistant to in interdigital apoptosis and scale formation. Science 272, 738– endotoxic shock. Cell 80, 401–411. 741. Martin, S. J., and Green, D. R. (1995). Protease activation during apoptosis: Death by a thousand cuts? Cell 82, 349–352. Received for publication July 15, 1998 Martin, S. J., Finucane, D. M., Amarante-Mendes, G. P., O’Brien, Revised September 9, 1998 G. A., and Green, D. R. (1996). Phosphatidylserine externaliza- Accepted September 9, 1998

Copyright © 1999 by Academic Press. All rights of reproduction in any form reserved.