Monitoring of epithelial cell caspase activation via detection of durable keratin fragment formation

Abstract

Keratins 18 and 19 (K18/K19) are epithelial-specific intermediate filament proteins. Apoptosis induces caspase cleavage at the highly conserved K18 or K19 Asp237, which in K18 is preceded by cleavage at Asp396. We characterized the keratin N-terminal fragments that are generated upon caspase digestion of K18/K19 at Asp237 in order to study keratin dynamics during apoptosis. This was carried out by generating and characterizing antibodies selective to K18/K19 Asp237. K18 or K19 peptides that expose Asp237 in ²³⁴VEVD were used for rabbit immunization. The generated antibodies recognized cleaved but not intact K18/K19, exclusively, as determined by blotting or immunofluorescence staining of apoptotic human HT29 cells or livers isolated from Fas-Ab-injected mice. Antibodies to K18/K19 Asp237 recognized the common VEVD-motif as determined by immunoblotting of cells transfected with K18, K19 or K20. The K18/K19 VEVD-directed antibodies demonstrated sequential Asp396 then Asp237 K18 cleavage during apoptosis. Specific-keratin selectivity of the anti-Asp237 antibodies was confirmed by their inability to recognize K14 after UV-induced apoptosis in transfected cells. The Asp237-containing apoptotic keratin fragments are secreted into the medium of cultured HT29 cells and are stable up to 96 h after inducing apoptosis. Furthermore, the generated antibodies recognize keratin apoptotic fragments in sera of mice undergoing hepatocyte apoptosis and sera of patients with cirrhosis, and also recognize apoptotic cells in various epithelial human tumours. Therefore, the N-terminal caspase-generated K18 fragment is stable in tissues and biological fluids. The Asp237-directed antibodies provide a powerful tool to study apoptosis in human and mouse tissues, cells and serum, using a broad range of detection modalities.

Introduction

Intermediate filaments are one of three major cytoskeletal networks found in most eukaryotic cells, and are expressed in a tissue- and differentiation-specific manner [1–3]. Intermediate filaments share a common structure that contains a central α-helical ‘rod’ domain, which is flanked by non-α-helical amino-terminal and carboxy-terminal regions termed the ‘head’ and ‘tail’ domains, respectively [4,5]. The rod domain is interrupted by short stretches termed linkers (L) L1, L12 and L2 [6]. The head/tail domains include sites of posttranslational modifications (e.g. phosphorylation/glycosylation) that play numerous regulatory roles [7,8]. In epithelial cells, cytoplasmic intermediate filaments are composed of type-I (K9-K20) and type-II (K1-K8) keratins [9–11]. All epithelial cells express at least one type I and one type II keratin that exist as obligate non-covalent heteropolymers. For example, K8/K18 heterodimers are abundantly co-expressed in liver and other glandular epithelia and form heterodimers that may include variable K7/K19/K20 levels, depending on the cell-type [1,12]. Keratin networks are highly dynamic and reorganize during differentiation, mitosis and apoptosis [7,13]. K8/K18 play important cytoprotective and structural roles and protect epithelial cells from apoptosis (reviewed in [10,13–15]).

Apoptosis is an important form of cell death in normal and diseased tissues, and involves the activation of caspases that are responsible for step-wise cleavage of critical cellular proteins [16–18]. Among cytoskeletal-related caspase substrates, K18 and K19 [19–21] undergo caspase-mediated proteolysis at an L12 highly conserved Asp237 (Figure 1A). Other intermediate filaments also contain the conserved L12 caspase consensus sequence [19–24], but some differences are found in proximity to the conserved aspartate (Supplementary Table 1, available at: http://www.interscience.wiley.com/jpages/0022–3417/suppmat/path.2344.html). K18 is unique in that it includes a second caspase-cleavage site at Asp396 (Figure 1A) that is not conserved among other intermediate filaments [20–22]. Caspase cleavage at K18 Asp396 occurs early during apoptosis, preceding annexin V reactivity, DNA fragmentation, and Asp237 cleavage [21,22]. Cleavage at Asp396 generates a neoepitope that is specifically recognized by the M30 monoclonal antibody [22], which has been widely used for detecting apoptosis in cultured cells [22,25], tissues [22,26] and sera of patients with cancer [27,28] and liver disease [29–31].

Figure 1.

The Asp237 antibody specifically recognizes caspase-generated but not intact K18 in cultured human cells and mouse liver. (A) Schematic of caspase-mediated sequential digestion of K18 at Asp396 then Asp237 during apoptosis. The diagram also shows the prototype intermediate filament structure of the ‘head’, ‘rod’ and ‘tail’ domains. The α-helical rod domain includes subdomains IA, IB and II, which are separated by the linker L1 and L12 non-α-helical short 8 – 17 amino acid stretches. p43, p29 and p23 represent the caspase-cleaved K18 polypeptides that migrate on SDS–PAGE with an apparent M_r of 43, 29 and 23 kDa, respectively. (B) K18 Asp237-Ab was pre-incubated with or without the immunization peptide (0.5 mg/ml, 4 ^◦C, 16 h), then used to immunoblot lysates of HT29 cells. The cell lysates were generated after culturing cells in the presence of anisomycin (An) to induce apoptosis or DMSO (vehicle control). (C) Apoptosis was induced in HT29 cells (as in B) using An, or in mouse livers by intraperitoneal injection of Fas antibody. The cell or tissue lysates were separated on SDS–PAGE, followed by blotting using an antibody that recognizes intact (i) K18 or the K18 Asp237-Ab that was purified using a peptide column (as described in Materials and methods). Note that Asp237-Ab recognizes only the caspase-cleaved fragment generated during apoptosis, but not intact K18

Given the highly conserved nature of K18/K19 Asp237 among intermediate filaments, we tested the hypothesis that its characterization provides a powerful tool to study its dynamics and serve as a marker to monitor keratin caspase digestion during apoptosis. We demonstrate herein that the Asp237-containing K18/K19 fragment is stable and provides remarkable versatility for detecting mouse and human epithelial apoptosis in cell culture, serum and tissues.

Methods

Reagents and antibody generation

Anisomycin (An) and monoclonal antibody (mAb) Jo2 (anti-Fas) were purchased from CalBiochem and Pharmingen, respectively. For K18/K19 Asp237-Ab generation, the peptides NH₂-CSSGLTVEVD²³⁷ (K18) and NH₂-CGGQVSVEVD²³⁷ (K19) were synthesized, conjugated to keyhole-limpet-haemocyanin, then used for immunization (two rabbits/peptide). Rabbits providing the highest titre and specificity (not shown) were subsequently used. Antibodies were purified using beads conjugated to the peptides NH₂-SSGLTVEVDAPK²⁴⁰ (K18) and NH₂-GGQVS VEVDSAP²⁴⁰ (K19) that extend three amino acids beyond Asp237. Other antibodies included: mAb-L2A1 (recognizes human K18) and rabbit Ab-8592 (recognizes human/mouse K8/K18 [32]); mAb-M30 (Roche Diagnostics Corporation), which recognizes the K18 43 or 23 kDa apoptotic fragments generated after caspase cleavage at Asp396 or Asp396 + Asp237 [19,21,22]; and anti-human K18 mAb-DC10 or anti-human K20 (Lab Vision). Liver explants from patients with alcohol- or hepatitis B-related cirrhosis were used as part of an approved protocol (Stanford University’s Panel on Human Subjects).

Induction of apoptosis

Anisomycin (An) was added to near-confluent human colonic HT29 cells (10 μg/ml; 0– 96 h). Culture media were removed and saved (−80 ^◦C) for subsequent ELISA and other testing. For cell homogenate preparation, the cells were rinsed with pre-warmed PBS (37 ^◦C) containing 5 mM EDTA, then fixed for immunofluorescence staining or solubilized using 2% SDS-containing buffer (90 ^◦C), followed by SDS–PAGE analysis and immunoblotting.

Apoptosis was induced in overnight fasted (water provided ad libitum) FVB mice by intraperitoneal injection of anti-Fas-Ab (mAb-Jo2; 0.15 μg/g body weight, diluted in PBS). Control age- and sex-matched mice were injected with PBS. The mice were sacrificed by CO₂ inhalation 4 h after injection of saline/Jo2. Blood was collected by intracardiac puncture and the livers were immediately resected and snap-frozen. The animal studies were approved by the Animal Subjects Committee. Sera from patients with hepatitis C infection were collected during the year 2000 and stored (−80 ^◦C).

Cell transfection

Lipofectamine 2000 (Invitrogen Life Technologies) was used for transient transfections of human K14, K18, K19 or K20 into BHK cells. The cells were co-transfected with wild-type K5 (and K14) or K8 (and K18/K19/K20) in order to stabilize the expressed type-I keratins [33]. After 48 h, the transfected cells were treated with anisomycin (20 h), or UV-irradiation (1 min) then harvested 16 h later for immunoblotting.

Biochemical methods

Proteins were separated using 10% SDS–PAGE and analysed by immunoblotting. Blots were visualized by enhanced chemiluminescence (ECL) or using Western Lightning^™ chemiluminescence reagent plus (ECLplus; Perkin-Elmer Life Sciences).

Asp237/Asp396 fragments in cells or culture media were determined after culturing HT29 cells with An (0– 96 h). For each time-point, floater and adherent cells were combined and pelleted (510 × g, 5 min) followed by re-pelleting of the supernatant (18 800 × g, 15 min) to remove potential debris. SDS-containing buffer was then added to the supernatants and the combined cell fractions, followed by blotting using M30-Ab (1 : 50) or Asp237-Ab (1 : 1000).

Immunofluorescence and immunohistochemical staining

HT29 cells were fixed (100% methanol, −20 ^◦C, 3 min) and liver sections were acetone-fixed (−20 ^◦C, 5 min). Immunofluorescence staining was done as described [32]. For immunohistochemistry, formalin-fixed, paraffin-embedded mouse/human tissue sections were subjected to antigen retrieval, then stained. Human tumour/normal tissue arrays were purchased from BioChain Institute Inc.

ELISA analysis

U96-Maxsorp Nunc immunoplates (Fisher Scientific) were coated with purified mAb-DC10 (10 μg/ml) in 50 mM carbonate–icarbonate buffer, pH 9.6, as a capture antibody and incubated overnight (4 ^◦C). All subsequent steps were performed at room temperature with shaking. After washing, the wells were blocked (5% non-fat dried milk in PBS, 1 h), followed by adding 50 μl/well of appropriately diluted cell culture medium or mouse/human serum (2 h). The wells were washed, followed by adding Asp237-Ab (1 h), rewashing, incubating with horseradish peroxidase-conjugated goat anti-rabbit IgG (1 h), then adding the substrate. Student’s t-test was used for statistical analysis and significance was defined as a two-tailed probability (p) < 0.05.

Results

Antibodies to K18/K19 Asp237 recognize apoptosis-induced cleaved but not intact keratins

Caspase activation results in proteolysis at conserved domains of many cellular substrates. K18 includes two established caspase digestion sites (Asp237/Asp396) with Asp237 being conserved in K18 and K19 (Figure 1A). We generated rabbit anti-sera to the peptides ²²⁹SSGLTVEVD (K18) and ²²⁹GGQVSVEVD (K19). Analysis of the K18-Asp237 antibody by immunoblotting of lysates isolated from cells (± An) showed that the Ab recognized primarily the caspase-generated fragment (∼29 kDa) in An-treated cells with slight cross-reactivity to intact K18 (∼48 kDa) (Figure 1B). Antibody reactivity was abolished when incubated with the immunization peptide (Figure 1B). Identical results were obtained using the K19-Asp237 antibody (not shown). Purification of the antibody (see Materials and methods) removes reactivity to intact mouse and human K18 (Figure 1C). Therefore, highly-specific antibodies to human and mouse K18-Asp237 can be generated.

The relatively conserved caspase-cleavage site VEVD of K18/K19/K20 is replaced by VEMD in K12–K17 (Supplementary Table 1, available at: http://www.interscience.wiley.com/jpages/0022–3417/suppmat/path.2344.html). We asked whether the K18 Asp237-Ab recognizes the caspase-generated fragment of K12– K17 using K14-transfected BHK cells (which do not have endogenous keratins) that were treated with or without UV-irradiation to induce apoptosis, with K18 and K20 being used as controls. Notably, the transfection procedure alone causes some apoptosis and caspase-mediated keratin digestion (as shown previously [21] and exemplified in Figure 2A, lane 3). The K18 Asp237-Ab recognizes caspase-generated K18 and K20 but not any potential K14-related fragments (Figure 2A). Therefore, the K18 Asp237-Ab specifically recognizes the exposed Asp in VEVD but not in VEMD.

Figure 2.

Asp237 antibodies bind to VEVD K18, K19 and K20 but not to VEMD K14. (A) BHK cells were co-transfected with human K5/K14 (lanes 1, 2), K8/K18 (lanes 3, 4) or K8/K20 (lanes 5, 6) cDNA. After 2 days, the cells were treated with or without 1 min of UV-irradiation, then cultured for 16 h. Total cell lysates were then prepared and separated using SDS–PAGE, followed by immunoblotting using antibodies specific to the indicated antigens. An actin blot is also shown to confirm equal protein loading. The predicted molecular weight of the putative apoptosis-induced K14 N-terminal fragment (1 – 273 amino acids) is 28.7 kDa, using the ‘Compute pI/MW’ program of ExPASy proteomics tools (http://us.expasy.org). Note that Asp237-Ab detects caspase-generated K18 and K20 but not K14 fragments. (B) BHK cells were transfected with K18 or K19 cDNA (together with K8 cDNA). After 2 days, anisomycin (An) was added, followed 24 h later by immunoblot analysis of the cell lysates, using antibodies specific to the indicated antigens. Note that both the K18 Asp237 (K18 D237) and K19 Asp237 (K19 D237) antibodies bind to the 29 kDa fragments derived from either intact K18 or K19

Since the K18 and K19 immunization peptides share four of nine residues (VEVD), we tested whether the K18 and K19 Asp237 antibodies are specific to individual K18 and K19. This was carried out by transfecting K18 or K19 separately into BHK cells, followed by induction of apoptosis. The K18-Asp237 and K19-Asp237 antibodies recognize both caspase-cleaved K18 and K19 (Figure 2B) and as such likely share the VEVD epitope. Given this epitope-sharing, all remaining data presented herein will use the K18 Asp237-Ab (termed Asp237-Ab).

Chronology of formation and stability of the apoptotic K18 Asp237- and Asp396-containing fragments

We examined the topography and chronology of caspase-induced cleavage of K18/K19 at Asp237 in situ in individual apoptotic cells. At lower magnifications, Asp237-Ab did not stain control HT29 cells but stained numerous cells after An treatment (Figure 3A). Higher magnification, coupled with nuclear co-staining (Figure 3B,4A), showed that Asp237-positive staining involved cells with condensed chromatin or fragmented nuclei consistent with apoptotic cells. Asp237-Ab positive staining appears 1–4 h after anisomycin treatment, with cleaved keratins being within filaments that sequentially reorganize into small dots that subsequently form large aggregates 16–20 h after exposure to the apoptotic stimulus (Figure 4A).

Figure 3.

Immunofluorescence staining of apoptotic cells using the K18 Asp237 antibody. (A) HT29 cells were treated with An or DMSO (vehicle control) for 20 h. The cells were then washed, fixed with methanol (−20 ^◦C, 3 min) followed by triple staining with the indicated antibodies or Toto-3 (for nuclear staining). (B) The boxed area in (Ah) is displayed as an enlargement of each triple stain component. Scale bars: (A, all panels) 50 μm; (B, all panels) 10 μm

Figure 4.

Asp237-Ab stains both filaments and aggregates in apoptotic cells. (A) HT29 cells were treated with An for 4 h (‘early’) or 16 h (‘late’), followed by double staining using Toto-3 (nuclear staining) and Asp237-Ab. At the early stage of apoptosis, Asp237-Ab primarily stains filaments, which subsequently reorganize into aggregates in association with nuclear fragmentation. (B) The cells used in (A) were triple stained using the anti-Asp396 Ab (M30), Asp237 K18 Ab and Toto-3. In (Ba), note that the cell highlighted by an asterisk and the intracellular area indicated by an arrow were only stained by Asp396 but not Asp237. In (Bd– f), note that many dots stain exclusively with Asp237-Ab. Scale bars (A, B) = 10 μm. (C) Total cell lysates or culture media were isolated from An-treated cells (0 – 96 h), as described in Materials and methods, followed by blotting using antibodies to K18 Asp237 or Asp396 caspase cleavage sites. Duplicate membranes were analysed by blotting but band detection was carried out using either the ECL or the ECLplus systems. Protein bands with asterisks are due to minimal spilling from the left-hand lane (confirmed by running separate gels, not shown). Note the different exposure (Exp) times and the type of ECL system used, thereby reflecting different detection sensitivity for the Asp237 and the Asp396 epitopes

We then compared caspase K18 digestion at Asp237 and Asp396 by double-staining with the corresponding antibodies. Early on (4h after An) some cells stained exclusively with Asp396-Ab (Figure 4B, cell with asterisk), while other cells manifested Asp237/396-Ab co-localization. During later stages (16 h after An; Figure 4B) Asp237 and Asp396 staining co-localized, but there were many K18 dots that stained exclusively with Asp237-Ab, while very few cells stained with Asp396-Ab. This suggests that the Asp237-containing epitope is more stable (or accessible) to Asp237-Ab than the Asp396-Ab epitope, or that the Asp396-containing species are soluble or diffusible into the culture medium. This was tested biochemically by immunoblotting culture supernatants or cells (± An, 0– 96 h). Caspase-cleavage at Asp237 [to generate p29 (see Figure 1A) and other NH₂-terminal truncated products] gradually increased in the cell lysates as compared with the progressive decrease in p43 and relatively stable p23 levels of Asp396 cleavage (Figure 4C), which is consistent with the cell staining results (Figure 4B). Therefore, both Asp237- and Asp396-containing K18 fragments are released from cells into the culture medium, but Asp237-cut species and p23 continue to accumulate with time. Notably, the sensitivity of Asp237-Ab is higher than the M30-Ab under our experimental conditions, since p43 and p23 are detected in supernatants using ECLplus only after long exposure (Figure 4C).

Asp237-Ab detects apoptosis in mouse and human tissues

The only available antibody to Asp396 (M30) is widely used for staining of K18 caspase cleavage in tissues and cells; however, it does not reliably detect mouse K18 digestion during apoptosis (Supplementary Figure 1, available at: http://www.interscience.wiley.com/jpages/0022–3417/suppmat/path.2344.html). In contrast to M30-Ab, Asp237-Ab recognizes both human and mouse cleaved K18 by immunoblotting (Figure 1C) and immunofluorescence staining (Figures 3–5). Asp237-Ab staining is absent in normal mouse liver but readily detectable after inducing apoptosis (Figure 5). The Asp237-positive hepatocytes had dots and fine filament-like staining (Figure 5Be) and positive staining in damaged areas that appear to have cell drop-off (Figure 5Ae).

Figure 5.

Detection of Fas-mediated apoptosis in mouse liver using the K18 Asp237 antibody. Mice were injected intraperitoneally with Fas antibody followed by isolation of the livers for immunofluorescence staining, as described in Materials and methods. Normal and injured livers were triple stained with the indicated markers. (A, B) Low and high magnification images, respectively. N, nucleus; CV, central vein; asterisks (Ad) exemplify areas of cell drop-off due to tissue damage; dotted area (Ae) highlights an area of injury and cell apoptosis. Scale bars: (Aa– f) 50 μm; (Ba– f) 10 μm

We also examined human cirrhotic liver explants from patients with alcohol or hepatitis B-related cirrhosis. Asp237-Ab stained only apoptotic but not other cells (e.g. cells with asterisks, Figure 6a–d). These cirrhotic livers with Asp237-positive staining also contained the 29 kDa K18 fragment, as noted in apoptotic HT29 cells (Supplementary Figure 2 available at: http://www.interscience.wiley.com/jpages/0022–3417/suppmat/path.2344.html). In contrast to mouse liver, where the M30-Ab does not recognize cleaved mouse K18 efficiently, both Asp237 and M30 antibodies similarly detect apoptosis in human diseased livers (Figure 6e, f).

Figure 6.

Immunofluorescence staining of apoptotic cells in human cirrhotic livers using Asp237-Ab. Alcohol- and hepatitis B-associated cirrhotic livers were double stained using antibodies to K18 Asp237 (a, c) and total K18 (b, d). Asterisks in (a–d) highlight cells that are negative for Asp237 but positive for K18 staining. The hepatitis B liver was also double stained with antibodies to Asp237/396 (e, f), which showed a similar pattern with the two antibodies. Scale bar: 10 μm for all images

The utility of Asp237-Ab in detecting apoptosis in formalin-fixed/paraffin-embedded tissues was assessed by immunohistochemical (IHC) staining of mouse liver and several human cancers. Livers from Fas-Ab-treated mice selectively stained with Asp237-Ab (Figure 7a, b), while a commonly utilized K18 antibody [32] stained all hepatocytes (Figure 7c). In support of the Asp237 staining, standard haematoxylin and eosin (H&E) staining showed few cells with condensed or fragmented nuclei (Figure 7d). The Asp237-Ab also detects apoptotic cells in formalin-fixed human adenocarcinomas, including those from the oesophagus, stomach and colon (Figure 7e–h).

Figure 7.

Detection of apoptosis in mouse and human tissues using Asp237-Ab immunohistochemical staining. Formalin-fixed and paraffin-embedded Fas-Ab-treated mouse liver (a–d); and human oesophageal, gastric and colonic cancer tissues (e–h) were stained with: anti-Asp237 Ab (a, e–g); anti-total K18 Ab (c, h); pre-immune normal rabbit serum (b); or H&E (d). The arrows in (d) indicate cells with typical condensed or fragmented nuclei. Note the selective staining of apoptotic cells [brown colour in (a, e–g)] using Asp237-Ab (the Asp-Ab does not stain any cells in untreated normal mouse liver; not shown). Scale bars: (a–d) 50 μm; (e, f, h) 50 μm; (g) 25 μm

Detection of caspase-generated K18 fragments in sera of mice or patients with liver disease

One potential important non-invasive clinical application of antibodies directed to Asp237 is to monitor apoptosis using an ELISA assay of sera from patients with tissue-associated apoptosis. As a proof of principle, we first examined whether the K18 fragment is released by HT29 cells into the culture medium during An-induced apoptosis. The cleaved K18 Asp237-containing fragment can be easily detected by ELISA (Figure 8a) and by immunoblotting (Supplementary Figure 2 available at: http://www.interscience.wiley.com/jpages/0022–3417/suppmat/path.2344.html). Asp237-Ab also detects the K18 fragment in sera of mice injected with Fas-Ab and in patients with cirrhosis (Figure 8b, c). The lower signal obtained with human sera may be related to less ongoing apoptosis or because the samples are archival. Presence of the cleaved mouse K18 product in serum was confirmed biochemically (Supplementary Figure 2). Taken together, these finding indicate that Asp237-Ab may be useful for non-invasive detection of apoptosis in K18-expressing cells.

Figure 8.

Detection of caspase-generated Asp237 keratin fragments in cell culture supernatants and in human and mouse serum. The K18 Asp237 Ab was used in a sandwich ELISA to measure caspase-generated fragments that are released into: (a) the culture medium of HT29 cells (± An); (b) sera of mice ± Fas-Ab injection to induce hepatocyte apoptosis; or (c) sera of patients with cirrhosis or controls. The representative wellsof ELISA plates, the statistical data and the number of samples analysed for each control or apoptosis group are also shown

Discussion

Unique features of the Asp237-Ab

Aside from the study herein, few antibodies that specifically recognize apoptotic fragments of intermediate filament proteins have been described. One is M30-Ab, which recognizes K18 ³⁹³DALD [22]. Second, antibodies to two apoptotic fragments of glial fibrillary acidic protein (GFAP) recognize: (a) ²⁶³DLTD of the N-terminal GFAP fragment (located outside L12) after caspase cleavage; or (b) ²⁶⁷AAARNAEC, which represents the amino portion of the C-terminal fragment of GFAP that becomes exposed after diges tion [34]. Third are antibodies to two vimentin apoptotic fragments: anti-²⁵⁶IDVD (within L12) and anti-C-terminal-²⁶⁰VSKPDC, which represents the freed amino terminus of the tail-containing fragment [35].

In epithelial cell apoptosis, Asp237-Ab offers unique features as compared with M30-Ab: (a) it recognizes mouse and human caspase-digested K18; (b) it is highly sensitive and offers broad utility of detection, using a range of biochemical/histochemical methods;(c) it recognizes the second sequential K18 caspase-cut site, which was demonstrated directly by immunestaining. Use of Asp237-Ab also showed that caspase cleavage of K18 is likely to occur on filaments which subsequently reorganize into aggregates during later stages of apoptosis, as noted in cultured cells, mice and human tissues.

The Asp237-Ab epitope

Dual binding to caspase-cleaved K18 and K19 by the Asp237-Abs indicates that the dominant epitope is shared between K18 NH₂-SSGLTVEVD²³⁷ and K19 NH₂-GGQVSVEVD²³⁷. This implicates the major common epitope as (T/S)VEVD, with Aspproximal valine being essential for Ab recognition. In support of this epitope assignment, Asp237-Ab does not recognize VEMD that is generated after K14 digestion but recognizes K20 NVEVD (K20 is also cleaved during apoptosis [36]). Within intermediate filament proteins, VEVD is present in K18/K19/K20 and in lamins B1/B2 (Supplementary Table 1). The Asp237-Ab does not appear to recognize the lamin B1/2 caspase cleavage sequence RLVEVD (Supplementary Figure 3, available at: http://www.interscience.wiley.com/jpages/0022–3417/suppmat/path.2344.html) that is generated during apoptosis [37]. This suggests that the environment adjacent to VEVD is also important for recognition by Asp237-Ab. However, the conclusions regarding lack of reactivity of Asp237-Ab to K14 and lamin B1/B2 are indirect, since they are supported by the decrease in intact proteins after induction of apoptosis without being able to demonstrate fragment formation (due to the lack of availability of appropriate antibodies). Interestingly, many non-intermediate filament cellular proteins harbour VEVD sequences, yet Asp237-Ab is highly specific for K18/K19/K20. This suggests that VEVD sequences within non-keratin proteins are non-accessible to caspases and therefore such sequences remain masked during apoptosis [18]. However, non-keratin proteins may have limited abundance to preclude detection and it is possible that tissues/cells we did not test may be recognized by Asp237-Ab.

K18 Asp396, within DALD, is a unique caspase-cleavage site found only in K18 but not other type-I keratins, although K18 DALD is conserved in humans and mice. Lack of convincing reactivity of M30-Ab with mouse K18 is likely related to sequence differences that are immediately proximal to DALD. For example, murine K18 has GEDFSLNDALD, while the corresponding human sequence is GEDFNLGDALD (underlined sequences are identical). The significance of the previously-reported M30-reactivity with mouse tempomandibular joint [38] is unclear, due to the likely absence of keratins in such tissue and the lack of biochemical confirmation.

The Asp237-Ab stains many more aggregates than M30 during late apoptosis and manifests higher biochemical detection-sensitivity, which implies several scenarios. For example, the cleaved Asp237-containing fragment might be more resistant to further protease degradation as compared with the Asp396-containing C-terminal fragment. Alternatively, the Asp396-containing species (p43/p23; Figure 1A) may be more soluble and hence less readily detectable as aggregates by immunofluorescence staining as compared with K18 p29. Biochemical analysis (schematically summarized in Supplementary Figure 4, available at: http://www.interscience.wiley.com/jpages/0022–3417/suppmat/path.2344.html) indicates that p43 decreases with time (likely, at least in part, due to conversion to p23), while the Asp237-containing fragments accumulate in both cells and culture medium.

Potential utility of the Asp237-Ab for non-invasive testing of epithelial cell apoptosis

The presence of non-apoptotic K8/K18/K19 as serological markers of epithelial malignancies has been described [39–41]. Keratin serum markers include tissue polypeptide antigen (represents total K8/K18/K19), tissue polypeptide-specific antigen (derived from K18), and CYFRA 21–1 (derived from K19). Furthermore, monitoring of caspase-cleavage at K18 Asp396, using M30-Ab, has been useful for detecting apoptosis in human epithelial tissues. For example, hepatocyte caspase activation was measured non-invasively in blood using an M30-Ab-based ELISA as a predictor of non-alcoholic fatty liver disease [31], and in patients with hepatitis C to predict those who may respond to antiviral therapy [30]. The clinical significance of Asp396 fragments in patient sera may be valuable in the early detection of recurrence and the rapid assessment of therapy responses.

In contrast to M30-Ab, Asp237-Ab demonstrated dramatic detection sensitivity in mouse sera and cultured cell supernatants. The somewhat limited sensitivity of detecting the Asp237-fragment by ELISA in the small group of tested patients as compared with sera of mice may be related to the extent of apoptosis in the human liver specimen, age of the human sera, or the repeated freeze–thawing, which can interfere with ELISA detection [42,43]. Our initial findings suggest that the Asp-237 epitope accumulates in cells undergoing apoptosis more readily than the Asp396 epitope, which may be related to fragment stability or release into the extracellular milieu, but this will require formal testing in different contexts. It is possible that different human clinical settings can be associated with release and stability of one epitope over another, which may give an advantage to using reagents to one epitope over another or to using reagents to both epitopes.

Taken together, the novel Asp237-Ab provides a novel tool for studying keratin caspase cleavage during apoptosis in human and mouse tissues that involve acute injury (e.g. Fas-Ab-induced liver injury in mice), chronic disease (e.g. liver disease) and human cancers. The N-terminal apoptotic keratin fragment is stable and offers a potential tool for non-invasive and rapid assessment of apoptosis in K18/K19/K20-expressing tissues or cell culture systems.