Effect of mutation and phosphorylation of type I keratins on their caspase-mediated degradation.

Type I keratins K18 and K19 undergo caspase-mediated degradation during apoptosis. Two known K18 caspase cleavage sites are aspartates in the consensus sequences VEVDA and DALDS, located within the rod domain and tail domain, respectively. Several K14 (another type I keratin) mutations within the caspase cleavage motif have been described in patients with epidermolysis bullosa simplex. Here we use extensive mutational analysis to show that K19 and K14 are caspase substrates and that the ability to undergo caspase-mediated digestion of K18, K19, or K14 is highly dependent on the location and nature of the mutation within the caspase cleavage motif. Caspase cleavage of K14 occurs at the aspartate of VEMDA, a consensus sequence found in type I keratins K12-17 with similar but not identical sequences in K18 and K19. For K18, apoptosis-induced cleavage occurs sequentially, first at (393)DALD and then at (234)VEVD. Hyperphosphorylation of K18 protects from caspase-3 in vitro digestion at (234)VEVD but not at (393)DALD. Hence, keratins K12-17 are likely caspase substrates during apoptosis. Keratin hyperphosphorylation, which occurs early in apoptosis, protects from caspase-mediated K18 digestion in a cleavage site-specific manner. Mutations in epidermolysis bullosa simplex patients could interfere with K14 degradation during apoptosis, depending on their location.

DALD. Hence, keratins K12-17 are likely caspase substrates during apoptosis. Keratin hyperphosphorylation, which occurs early in apoptosis, protects from caspase-mediated K18 digestion in a cleavage site-specific manner. Mutations in epidermolysis bullosa simplex patients could interfere with K14 degradation during apoptosis, depending on their location.
Keratins are the cytoplasmic intermediate filament (IF) 1 proteins of epithelial cells and consist of Ͼ20 separate gene products (1). The keratin subfamily of IF proteins is classified into two major groups, type I keratins (K9 -20) and type II keratins (K1-8), which associate as noncovalent type I-II heteropolymers in an epithelial cell type-specific manner (1)(2)(3)(4). Among cytoplasmic IF proteins, keratins and vimentin undergo caspase-mediated degradation as part of the cytoskeletal remodeling that takes place during apoptosis (5)(6)(7)(8). The nuclear lamin IF proteins also undergo degradation during apoptosis and were the first IF proteins demonstrated to undergo apoptosis-associated digestion (9 -11). The only keratins shown to undergo proteolysis during apoptosis are K18 and K19, whereas their type II partner (i.e. K8) manifests remarkable resistance to apoptotic degradation. Two known K18 caspase sites, VEVD and DALD, are located in the rod domain and tail domain, respectively (5)(6)(7). VEVD or similar consensus sequences are found in other IF proteins within the so-called linker 1-2 (L1-2) region of the rod domain ( Fig. 1), whereas DALD is a unique caspase site that is found only in the K18 tail domain. The signals, if any, that target keratin degradation and the significance of this proteolysis are unknown. To that end, the only keratin-related apoptosis-associated change after an apoptotic signal is marked early keratin hyperphosphorylation. The significance of this early keratin hyperphosphorylation in association with apoptosis is not known, but amino acid substitution of the major K18 phosphorylation sites does not alter susceptibility to caspase digestion (6).
Understanding the significance and regulation of keratin (and other IF protein) degradation during apoptosis is important from a cell biological perspective and may also have pathophysiological relevance to human disease. For example, although most keratin mutations that have been identified in patients with epidermal blistering keratin diseases are located at the N-terminal region of the rod IA subdomain (12,13), at least four K14 mutations have been described within the L1-2 region (14 -17) in close proximity to the caspase recognition motif (VEMDA, also referred to herein as the caspase box). The cause of blister formation in these patients may be attributed to keratin filament assembly defects with resultant cell fragility. However, the proximity of these mutations to the caspase digestion site raises the possibility that the phenotype of the keratin disease in these instances may also be associated with perturbations in keratin degradation. If so, this could potentially impact the disease pathophysiology in patients with epidermolysis bullosa simplex (EBS), who harbor K14 L1-2 region mutations, and may offer more directed therapeutic strategies.
In this study, we use a mutagenesis approach to confirm that Asp 396 in the K18 tail domain is a caspase cleavage site in vivo. In addition, we show that sequential K18 digestion occurs at the tail (DALDS) and then at the rod (VEVDA) domains and that keratin hyperphosphorylation protects against cleavage at the rod but not the tail motif. To address the significance of the caspase box residues in keratin degradation, we generated a battery of caspase box mutations that mimicked the K14 mutations described in EBS patients and examined parallel mutations in K18 and K19. The results show that: (i) K14 VEMDA3MEMDA or VEMDD has no effect on susceptibility to caspase digestion, (ii) K14 VEMDA3 VERDA and equivalent K18 and K19 mutations altered the migration of the N-terminal fragment on gel analysis, and (iii) K14 VEMDA3 VEMGA and the equivalent K18 mutation abolished caspase digestion (boldface letters indicate residues that are mutated). Hence, pathogenic K14 mutations within the L1-2 region in patients with EBS can prevent caspase-mediated keratin degradation during apoptosis. In addition, the VEMDA caspase box, which is found in many type I keratins (K12-17) is a suitable caspase substrate, as shown here for K14.
Cell Culture and Transfection-HT29 (human colon) and BHK-21 (hamster kidney) cells (American Type Culture Collection, Manassas, VA) were cultured in media as recommended by the supplier. HT29 cells express K8, K18, and K19, whereas BHK-21 cells do not express any easily detectable keratins (data not shown). Wild-type (WT) or mutant keratin constructs, generated using a Transformer TM kit (CLONTECH), were transfected using LipofectAMINE (Life Technologies, Inc.) according to the manufacturer's instructions. Notably, this transfection method induces apoptosis and keratin degradation in BHK-21 cells as reported previously (6).
Biochemical and Immunoblotting Analysis-Keratin degradation and subsequent apoptosis were induced in HT29 cells by culturing the cells in the presence of An (10 g/ml in Me 2 SO) for 0, 0.5, 1, 2, 4, 8, 12, or 16 h. Total cell lysates in 2% SDS-containing sample buffer were resolved by SDS-polyacrylamide gel electrophoresis (PAGE) (18) and then transferred to polyvinylidene difluoride membranes, followed by immunoblotting (19). BHK-21 cells transiently transfected with K8 and/or K18 were treated with 0.1% Me 2 SO or 50 M caspase inhibitor III (BOC) overnight, followed by isolation of total cell lysates and immunoblotting.
In Vitro Caspase Digestion Assay-HT29 cells were solubilized with FIG. 1. K18 caspase cleavage sites and conservation of the VEVDA sequence within the L1-2 subdomain of IF proteins. A, IF proteins share the structural features of a central ␣-helical rod domain that is flanked by non-␣-helical N-and C-terminal head and tail domains, respectively. The rod domain is divided into subdomains that are separated by short linker (L) regions. The K18 p29, p23, and p4 fragments are labeled as a, b, and c fragments, respectively, and the amino acid location of each subdomain and the location of K18 Asp 237 and Asp 396 caspase cleavage sites are shown. Single-letter abbreviations are used to indicate amino acids. B, conserved amino acid sequences within the L1-2 domain of mammalian IF proteins are shown and represent the caspase box of K18 and K20. Note the highly conserved Asp, which is replaced by an Asn in K9 and K10. Dots indicate amino acid residues that are identical to the corresponding VEVD sequence of K18, K19, and K20. Of note, Xenopus type I keratins also have the VEMD sequence (data not shown). Boxed residues in K12-17 sequences correspond to valine, methionine, aspartate, and alanine, which are mutated in patients with epidermolysis bullosa simplex (V270M, M272R, D273G, and A274D). The arrow points to the aspartate cut site. Examination of the IF proteins sequences within the L1-2 subdomain provides the consensus motif X 1 E/DX 2 DX 3 (X 1 -X 3 , aliphatic/ uncharged amino acids), except for the type II keratins (K1-8), which appear to be relatively spared from caspase digestion and have a serine instead of E/D after the X 1 position.

FIG. 2.
Mutational analysis confirms the presence of two K18 caspase sites. K8/K18 immunoprecipitates were prepared from BHK-21 cells transfected with vector alone (Control) or with WT K8 and one of the following K18 constructs: WT, D237E, D396E, or the double mutant D237/396E. The immunoprecipitates were analyzed by SDS-PAGE and then Coomassie Blue staining (A) or immunoblotted with: (i) anti-K18 N-terminal Ab (B) that recognizes intact K18 (data not shown), K18 a, or K18 aϩb (see Fig. 1 for schematic representation of fragments); (ii) anti-K18 C-terminal Ab (C) that recognizes intact K18 (data not shown), K18 bϩc, or K18 b (the K18 c fragment is too small to be detected by the gel conditions used); or (iii) mix of K18 N-terminal and C-terminal Abs (D). Limited amounts of K18 a or K18 bϩc are noted upon mutation of K18 at D237E and/or D237/396E, which likely reflects insufficient digestion due to the mutation(s) and/or digestion of hamster K18 that may be expressed at low levels in BHK-21 cells.
1% Nonidet P-40 in phosphate-buffered saline containing a mixture of protease inhibitors (6). After 1 h, lysates were pelleted, and the supernatant was used for immunoprecipitation by incubation in the presence of Sepharose-protein A coupled to mAb L2A1. Two duplicate K8/K18 immunoprecipitates were either used as a control or incubated with calf intestine alkaline phosphatase (20 units) for 1 h at room temperature to obtain dephosphorylated K8/K18. Hyperphosphorylated K8/K18 immunoprecipitates were obtained from HT29 cells treated with 1 g/ml okadaic acid for 2 h. The immunoprecipitates (control, dephosphorylated, or hyperphosphorylated K8/K18) were incubated with buffer alone or with buffer containing recombinant human caspase-3 for 0.5, 1.5, or 3 h. The samples were then separated by SDS-PAGE and analyzed by immunoblotting.

K18 Contains Two Caspase Recognition Sites That
Are Sequentially Digested-Previous studies utilizing K18 D237A (5) or direct sequencing of apoptosis-generated K18 fragment b ( Fig. 1 and Ref. 6) showed that K18 Asp 237 is a major cleavage site in vivo. In addition, in vitro digestion of K18 using caspase-3 or caspase-7 indicated that a second caspase site is likely to be present in the tail domain of K18 (5). The second site was inferred to be K18 Asp 396 by epitope mapping using mAb M30 (7), although this was not formally tested by mutational analysis. We mutated K18 D396E and K18 D237/396E and examined caspase digestion of the mutants as compared with WT K18-transfected BHK-21 cells undergoing apoptosis. K18 D396E generated a "new" 27-kDa band (K18 bϩc fragment) that was recognized by a K18 C-terminal-specific Ab ( Fig. 1A; Fig. 2, A and C), whereas K18 D237E accumulated a 43-kDa band (K18 aϩb fragment; Fig. 2A) that was recognized by a K18 N-terminal-specific Ab ( Fig. 1A; Fig. 2, B and D). The double mutant K18 D237/396E generated only one major undigested K18 species (Fig. 2A, lane 5). Hence, both K18 Asp 237 and Asp 396 are caspase digestion sites in vivo.
Generation of the K18 fragments aϩb, a, or b was inhibited using the broad range caspase inhibitor BOC in cells transfected with WT K18 or with WT K8/K18 (Fig. 3A), thereby indicating that K18 fragmentation at Asp 237 and Asp 396 is caspase-mediated. Antibody specificity was confirmed by blotting K8-transfected cells ( Fig. 3A; K8 expression was determined by blotting with anti-K8-specific Ab; data not shown). The M30 mAb (which recognizes an epitope that becomes exposed after K18 is cut at Asp 396 ) does not detect K18 when mutated at D396E or D237/396E (Fig. 3B, lanes 4 and 5) as predicted from its reactivity with K18 synthetic peptides (7). In addition, M30 reactivity for K18 aϩb fragment increases dramatically in K18 D237E as compared with WT K18 transfectants, because K18 aϩb fragment accumulates because it can no longer be further degraded (Fig. 3B, compare lanes 2 and 3).

FIG. 3. Blocking of K18 fragmentation by caspase inhibition and time course of K18 digestion.
A, BHK-21 cells were transiently transfected with the indicated constructs. After 2 days, the transfected cells were treated with Me 2 SO (Ϫ) or BOC (ϩ) as described under "Experimental Procedures." Total lysates were prepared from the transfected cells and then examined by immunoblotting with anti-K18 N-terminal-domain Ab (i.e. recognizes intact K18, K18 a, or K18 aϩb) or with anti-K18 D396-cut Ab (M30) that recognizes the exposed epitope after caspase digestion at K18 Asp 396 (i.e. recognizes K18 aϩb or K18 b). BOC inhibited K18 fragmentation at both Asp 237 and Asp 396 , and co-transfection with K8 stabilized K18 and its fragments. B, BHK-21 cells were co-transfected with K8 and the indicated WT and K18 mutants. Lysates from the transfected cells were prepared and then immunoblotted with antibodies as described in A. Note that the K18 p43 fragment (i.e. K18 aϩb, which results upon cleavage at Asp 396 ) accumulates in cells transfected with K18 D237E, but not in K18 D237/396E, because it cannot be further cleaved at Asp 237 (see also Fig. 2A, arrowhead in lane 3 that also shows accumulation of p43). C, HT29 cells were treated with An (10 g/ml) for the indicated times. Total lysates were then prepared and analyzed by immunoblotting with antibodies that recognize the indicated K18 fragments. Note that p43 (i.e. K18 aϩb) was detected after 0.5 h of exposure to An, whereas p29 and p23 were detected after 2 h of exposure to An, thereby indicating that K18 caspase cleavage at Asp 396 occurs earlier than cleavage at Asp 237 .
We then tested, using An-induced apoptosis, whether caspase digestion occurs sequentially or randomly at VEVD and DALD. Immunoblotting of total lysates from HT29 cells treated with An for various time intervals with M30 showed that K18 aϩb (i.e. digestion at DALD) begins to appear after 0.5 h, whereas K18 b (i.e. digestion at VEVD) is detected after 2 h of An treatment (Fig. 3C). This indicates that upon Aninduced apoptosis, Asp 396 is initially cleaved followed by digestion at Asp 237 . Cleavage at Asp 396 does not appear to be essential for cleavage at Asp 237 because the K18 D396E mutant remains a caspase substrate at Asp 237 (Fig. 2, compare lanes 2  and 4).
Effect of Phosphorylation on Keratin Fragmentation in Vitro-We showed previously that K8 (Ser 73 and Ser 431 ) and K18 (Ser 52 ) but not K18 (Ser 33 ) hyperphosphorylation occurred within 0.5 h after An treatment and that keratin Ser3 Ala mutants at these sites remain comparable to WT K18 in terms of their susceptibility to caspase digestion (6). Here, we examined the effect of dephosphorylation or hyperphosphorylation on K18 fragmentation using in vitro digestion by caspase-3. K8/K18 immunoprecipitates that were isolated from okadaic acid-treated cells or treated with alkaline phosphatase were digested with caspase-3 in vitro and then analyzed for the formation of keratin fragments. As shown in Fig. 4A, K18 hyperphosphorylation inhibited digestion at K18 Asp 237 without any significant effect on K18 digestion at Asp 396 . In contrast, dephosphorylation did not have a significant effect on K18 Asp 237 or Asp 396 digestion (Fig. 4A), which is consistent with our previous findings using phosphorylation-mutant keratins (6). The effect of okadaic acid and alkaline phosphatase on K8/K18 phosphorylation was confirmed by immunoblotting of the K8/K18 precipitates with anti-phospho-K8 and anti-phospho-K18 antibodies (Fig. 4B).
Effect of Caspase Box Mutations on Susceptibility to Keratin Fragmentation during Apoptosis-The VEVDA motif in the rod domain is found in K18 and K20, whereas the VEVDS motif is found in K19. Similar rod domain motifs, such as VEMDA, are present in other IF proteins including K14 (Fig. 1B), whereas the DALD motif in the tail domain is unique to K18. Given the conserved nature of the rod domain motif (X 1 E/DX 2 DX 3 ; X 1 -X 3 , aliphatic residues with caspase cleavage occurring at the aspartate between X 2 and X 3 ; Fig. 1A) and the presence of K14 mutations at the aspartate of VEMD (D3 G) and at X 1 (V3 M), X 2 (M3 R), and X 3 (A3 D) in patients with EBS, we asked whether these mutations have an effect on type I keratin fragmentation during apoptosis. To address this question, we generated several corresponding mutations in K14, K18, and K19 and tested the mutant constructs for susceptibility to caspasemediated degradation in cell transfection systems. Mutation V236M in K18 or K19 to generate a WT K14-like caspase box (i.e. VEMD instead on VEVD) had no effect on caspase-mediated degradation of K18 (Fig. 5A) or K19 (Fig. 5C). Similarly, EBS-like mutations at the X 1 or X 2 positions of the caspase box of K14 (V270M or M272R, respectively) and of K19 (V234M or V236R, respectively) and at the X 2 position of K18 (V236R) had no effect on keratin fragmentation upon apoptosis ( Fig. 5; Table I). However, the X 2 mutation of M (in K14) or V (in K18/19) to R generates an N-terminal fragment that migrates slightly faster on SDS-PAGE gels as compared with the equivalent N-terminal fragment generated with WT or other mutant K14, K18, or K19 fragments (Fig. 5, highlighted with an asterisk). It is unlikely that the altered migration is due to the valine to arginine substitution per se because the Nterminal fragment of K18 V220R migrates similarly to the WT K18 fragment (Fig. 5B, compare lanes 1 and 2). In addition, this faster-migrating N-terminal fragment does not appear to result from exposure of other potential cryptic caspase digestion sites (i.e. K18 Asp 180 at 177 VEND or K18 Asp 189 at 186 KVID that may be exposed after the Val3 Arg mutation in K18 V236R shown). Interestingly, an arginine substitution at the X 1 position of K18 (V234R) blocks K18 cleavage at Asp 237 (Fig.  5B, lane 3), thereby indicating that a basic residue substitution at the X 1 caspase box position is likely to inhibit caspase enzyme-substrate recognition.
An EBS-like K14 VEMD3 VEMG mutation and a similar K18 mutation (VEVD3 VEVG) abolish caspase cleavage ( Table  I). As shown previously for K18 (5) and shown here for K19 (D237E) and K14 (D273E), D3 E mutations in these keratins also block caspase cleavage (Fig. 5, C and D). In addition, K18 VEVD3 VEVE generates the 43-kDa fragment (K18 aϩb in Fig. 2A) due to caspase digestion at Asp 396 in 393 DALD, a site that is not found in other non-K18 type I keratins. Another EBS-like mutation at the X 3 position in K14 (A274D) or K18 (A238D) does not affect susceptibility to caspase-mediated digestion (Table I), thereby indicating that the X 3 position is insensitive to acidic charge perturbations. DISCUSSION Apoptosis-associated Degradation of Keratins-K18 (5-7) and K19 (6,8) are the only keratins that have previously been demonstrated to undergo degradation during apoptosis. Degradation occurs primarily in type I keratins, with marked relative sparing of type II keratins as determined for K8, which is the only type II keratin studied in this context (5,6). Sparing of type II keratins may be related to differences of the context of the caspase box within the L1-2 region of the rod (Fig. 1B, note the E3 S substitution in type II keratins within the VEVD sequence), but type II keratins do posses other potential caspase recognition sequences (e.g. 77 LEVD and 253 LDMD in K8) that do not appear to be prominently cleaved. One important finding herein is that K14 is also a caspase substrate in transfected cells. This indicates that the remaining keratins of K12-17, in addition to desmin and neurofilament-L, are also likely to be caspase substrates because they all share the same VEMD motif, which differs slightly from the K18 -20 (VEVD) motif.
The type of keratin fragments generated during apoptosis may differ depending on the presence or absence of other caspase recognition motifs and their susceptibility to cleavage. In the case of K18, two well-defined cut sites occur as defined FIG. 5. Effect of type I keratin caspase box mutations on their degradation. All transfections were done in BHK-21 cells, followed by cell analysis after 3 days. A, K8/K18 immunoprecipitates were obtained from the indicated K8 and K18 co-transfectants, followed by analysis by SDS-PAGE/Coomassie Blue staining or by immunoblotting with the indicated anti-K18 antibodies. Note that the N-terminal (p29 or K18 a) fragment from K18 V236R (indicated by an asterisk) migrates slightly faster than the "equivalent" fragments generated by K18 WT or V236M, whereas the C-terminal (p23 or K18 b) fragments have a similar migration. Limited formation of p29 and p23 was noted for the K18 D237E mutant, which likely reflects minor caspase cleavage at Glu 237 and/or digestion of low levels of hamster WT K18 (which may be present in BHK-21 cells). B, total cell lysates were obtained from the indicated K8 and K18 co-transfectants, followed by immunoblotting with anti-K18 antibodies that recognize intact K18 and K18 a (p29). Note that K18 V234R (lane 3) is not a caspase substrate at Asp 237 and that simply introducing an Arg (i.e. K18 V220R) does not alter the migration of K18 a (p29), as did the Arg of V236R. C, total cell lysates were obtained from the indicated K8 and K19 co-transfectants, followed by immunoblotting with Ab 8592 (which recognizes K8, K18, and K19) or with anti-K19 mAb KA4 (which recognizes an epitope within amino acids 145-227 of rod subdomain IA). Note that the Asp mutation completely blocks caspase digestion and that the V236R mutant generates an N-terminal fragment (K19/N) that migrates slightly faster (highlighted by asterisk) than the fragments generated by the remaining K19 constructs. Lane 1 (Control), cells transfected with vector only. D, total cell lysates were obtained from the indicated K6 and K14 co-transfectants, followed by immunoblotting with anti-K6 or anti-K14 antibodies. The anti-K14 Ab recognizes intact K14 and an N-terminal K14 fragment (K14/N) that is cleaved at Asp 273 (confirmed by mutation of the conserved Asp 273 , lane 5). As noted for K18 and K19, the K14/N species that is generated by K14 M272R also migrates slightly faster (indicated by an asterisk) than the K14/N species generated by the other two K14 constructs shown in lanes 2 and 3. Lane 1 (Control), cells transfected with vector only. immunologically (7) and molecularly (Fig. 2). These two sites undergo sequential caspase-mediated digestion (Fig. 6) with release of the small K18 tail fragment (397-429) from the K8/K18 complex, followed by cleavage at K18 Asp 237 to generate two stable fragments (1-237 and 238 -396) that remain associated with K8. This apoptosis-associated keratin cleavage is accompanied by significant reorganization of the keratin cytoskeletal network (5, 20 -23). Transient transfection of the K18, K19, and K14 mutants (with WT K8) did not have any significant effect on filament organization as determined by immunofluorescence staining (data not shown).
Modulation of Caspase Cleavage by Phosphorylation and by Mutations within the Caspase Box Motif-Keratin hyperphosphorylation occurs as an early event upon exposure of cells to an apoptotic signal (6,24). Mutation of the major K8 and K18 phosphorylation sites did not affect the susceptibility of caspase-mediated cleavage of K18 at Asp 237 , thereby indicating that dephosphorylation did not affect keratin degradation during apoptosis (6). However, hyperphosphorylation does significantly inhibit caspase-3 in vitro digestibility of K18 at Asp 237 (the second sequentially cut K18 site), but not at Asp 396 (the first cut site) (Fig. 6). This raises the possibility that hyperphosphorylation of the remaining type I keratins, which are cleaved at the K18 Asp 237 -equivalent site (Fig. 1B), may also be protective. Several functional roles, acting alone or in concert, can be envisioned for keratin hyperphosphorylation during apoptosis: (i) a simple by-product of the apoptosis-associated activation of multiple kinases (e.g. Refs. 25 and 26); if so, this favors a role for keratins as a phosphate reservoir or sink (24), (ii) a facilitator, alone or in concert with keratin degradation, of keratin filament reorganization during apoptosis, or (iii) a mechanism that either protects from apoptosis-induced damage (in this case degradation of keratins) or allows for a graded sequence of apoptotic events.
The caspase box motif that is found within the L1-2 region of the rod domain of cytoplasmic IF proteins is a prototype caspase recognition motif, represented in type I keratins by X 1 E/DX 2 DX 3 (with site of cleavage occurring at the D between X 2 and X 3 ; X 1 -X 3 , hydrophobic residues). Our results showed that replacement of X 1 by an Arg prevents caspase-mediated degradation (Table I), which suggests that basic residue substitutions at X 1 are likely to be incompatible with substrateenzyme recognition. In contrast, the X 2 or X 3 positions were not affected by basic residue substitutions in that M/V3 R mutations (i.e. VEMD/VEVD3 VERD in K14, K18, or K19) or A3 D mutations (VEMDA/VEVDA3 VEMDD/VEVDD in K14 or K18), which mimic K14 mutations found in EBS patients, had no measurable effect on caspase-mediated cleavage of the type I keratin. These results, using an in vivo transfection system, are similar to what has been noted with in vitro peptide substrates in that the X 2 position can tolerate a wide range of amino acids, whereas X 1 seems to dictate caspase enzyme-type specificity (27)(28)(29). In contrast, the X 3 position tolerated an A3 D substitution well, which would not have been predicted based on peptide in vitro substrate studies (28,29).
Significance of Keratin Degradation during Apoptosis-The significance of keratin degradation in modulating apoptotic progression is unclear, although the predicted and observed reorganization of the cytoskeleton during apoptosis (5, 20 -23) suggests that caspase cleavage during apoptosis is likely to facilitate apoptotic body formation. Formation of keratin frag-FIG. 6. Schematic of the fate of K18 during apoptosis. Exposure of epithelial cells to an apoptotic signal results in rapid hyperphosphorylation (P) of K18 with subsequent caspase-mediated cleavage of K18 at Asp 396 . The fate of K18 397-429 is unknown, but it is likely to be released from the K8/K18 complex because it is not detected upon SDS-PAGE analysis of K8/K18 immunoprecipitates after apoptosis (data not shown). The remaining larger K18 fragment (1-396), which remains associated with K8 (e.g. Fig. 2A, lane 3), undergoes a subsequent cleavage step to generate two major K18 fragments (1-237, termed p29 or K18 a; and 238 -396, termed p23 or K18 b; see also Fig. 1A). These two fragments also remain for the most part associated with K8 (e.g. Fig. 2A, lanes 2 and 4). ments during apoptosis has potential clinical utility because detection of such fragments has been used in a number of studies as diagnostic and prognostic markers (30 -35). These include the so-called tissue polypeptide antigen and tissue polypeptide-specific antigen, which are related to K8, K18, and K19 (36) and to a C-terminal K18 fragment (37), respectively. However, the molecular mechanisms for generating tissue polypeptide antigen and tissue polypeptide-specific-like fragments are unknown, although caspase-mediated degradation and/or other protease activation are likely mechanisms. Hence, understanding the precise molecular changes that occur to keratins during apoptosis is an important first step in determining the significance of caspase-mediated keratin fragment formation and release in tumors. The presence of keratin mutations within the caspase box in patients with epidermal diseases raises the possibility that alterations in susceptibility to caspase-mediated cleavage could impact disease pathogenesis or alter susceptibility to other skin diseases. For example, apoptosis (and presumably subsequent keratin degradation) is a feature of several skin diseases and injury (e.g. Refs. 38 and 39). Identification of the keratin cleavage sites during apoptosis will allow subsequent in vivo testing of the significance of mutations that inhibit keratin degradation on progression of apoptosis. Although keratin mutations are most common within the proximal region of the rod domain (12,13), K14 mutations within the L1-2 domain have been described (14 -17). Of the four such K14 mutations we tested, only the K14 D273G mutation prevented the caspasemediated digestibility of K14 (Table I). Another K14 mutation (M272R) generated a K14 fragment with altered mobility on SDS-PAGE gels, and a similar mutation that was introduced in K18 and K19 also altered the migration of the resultant apoptotic fragment (Fig. 5). The cause of this altered migration is unclear, but it is unlikely to be due to the arginine mutation per se because introducing an arginine (V220R) proximally did not alter fragment migration (Fig. 5). Given that the type I keratins K12-14, K16, K17, and K18 have been associated with a variety of human diseases (12,13), it is likely that additional mutations within L1-2 and the caspase box will be defined. If so, our results should facilitate the prediction of whether such mutations will interfere with caspase digestion.