Histochemistry and Cell Biology

, Volume 124, Issue 3, pp 237–243

Programmed cell death and cell extrusion in rat duodenum: a study of expression and activation of caspase-3 in relation to C-jun phosphorylation, DNA fragmentation and apoptotic morphology


  • Kirsten Schauser
    • Division of Cell Biology Department of Anatomy and PhysiologyThe Royal Veterinary and Agricultural University
    • Division of Cell Biology Department of Anatomy and PhysiologyThe Royal Veterinary and Agricultural University
Original paper

DOI: 10.1007/s00418-005-0035-7

Cite this article as:
Schauser, K. & Larsson, L. Histochem Cell Biol (2005) 124: 237. doi:10.1007/s00418-005-0035-7


The small intestinal epithelium is continuously renewed through a balance between cell division and cell loss. How this balance is achieved is uncertain. Thus, it is unknown to what extent programmed cell death (PCD) contributes to intestinal epithelial cell loss. We have used a battery of techniques detecting the events associated with PCD in order to better understand its role in the turnover of the intestinal epithelium, including modified double- and triple-staining techniques for simultaneously detecting multiple markers of PCD in individual cells. Only a partial correlation between TUNEL positivity for DNA fragmentation, c-jun phosphorylation on serine-63, positivity for activated caspase-3 and apoptotic morphology was observed. Our results show that DNA fragmentation does not invariably correlate to activation of caspase-3. Moreover, many cells were found to activate caspase-3 early in the process of extrusion, but did not acquire an apoptotic nuclear morphology until late during the extrusion process. These observations show that the lack of consensus between different methods for detecting PCD may be explained both by different timing of appearance of PCD markers and, additionally, by the occurrence of different forms of PCD during the normal turnover of cells on small intestinal villi.


The small intestinal epithelium is renewed through a balance between cell proliferation and cell loss. Stem cells present in intestinal crypts proliferate to give rise to cells that migrate upwards to differentiate into cylindrical absorptive cells, goblet cells and endocrine cells or migrate downwards to differentiate into paneth cells (Chandrasekaran et al. 1996; Watson and Pritchard 2000; Wilson and Potten 2004). During upward migration, intestinal epithelial cells (IECs) continue to proliferate until they approach the zone of crypt–villus transition. Cells on villi are strictly postmitotic and migrate to the apical villus portion, from which they are lost by extrusion into the lumen (Chandrasekaran et al. 1996; Potten and Allen 1977; Wilson and Potten 2004). Accordingly, the frequency of cell renewal in crypts and of cell loss on the villi tips must be carefully balanced. Importantly, a disturbed balance characterizes pathological states like coeliac disease, bowel inflammatory disease, infectious diseases, postradiation disease and cancer (Jones and Gores 1997).

Cell turnover usually involves programmed cell death (PCD). Programmed cell death is a process by which damaged, aged, or otherwise unwanted cells are eliminated through a series of steps that result in the destruction of their genome. The form of PCD known as apoptosis is characterized by a series of morphological changes, including nuclear condensation and fragmentation, cytoplasmic blebbing and cell shrinkage (Kerr et al. 1972). These changes appear to depend upon the activation of cysteine-dependent aspartate-specific proteases (caspases), in particular the effector caspase, caspase-3 (Jänicke et al. 1998). Caspase-3 cleaves a number of cellular substrates, including lamins, fodrin, cytokeratin-18, actin, gelsolin, focal adhesion kinase (FAK), p21-activated kinase 2 (PAK-2), Rho-associated kinase-1 (ROCK-1), poly-ADP ribose polymerase (PARP) and inhibitor of caspase-activated DNase (ICAD) (Fischer et al. 2003; Jänicke et al. 1998). Cleavage of these multiple substrates is believed to contribute to the morphological hallmarks of apoptosis, including nuclear shrinkage and fragmentation (lamins), cytoplasmic alterations and blebbing (actin, fodrin, cytokeratin-18, gelsolin, PAK-2, ROCK-1) and to cleavage of the genome into nucleosomal fragments (caspase-dependent DNase) (Fischer et al. 2003). Thus, caspase-3-deficient human breast cancer (MCF-7) cells are able to undergo PCD but do not exhibit an apoptotic morphology unless transfected with constructs encoding caspase-3 (Jänicke et al. 1998). Accordingly, caspase-3 is not indispensable for PCD, but may be required for the morphological manifestations of apoptosis. Importantly, recent studies have demonstrated that cells may undergo other forms of PCD than apoptosis and that caspase-independent PCD may result in morphological manifestations different from those of apoptosis (Jäättelä 2004; Lockshin and Zakeri 2004).

Several pathways may activate PCD. One of these pathways involves the activation of Jun N-terminal kinases (JNKs). Jun N-terminal kinases are encoded by three different genes and are activated by a wide variety of stimuli, including UV irradiation, osmotic shock, heat shock, cytoskeletal disruption, proinflammatory cytokines and growth factors (Ip and Davis 1998). A major target for JNKs is c-jun, a component of the transcription factor activator-1 complex (AP-1) (Ip and Davis 1998; Karin 1995). JNKs phosphorylate c-jun on serine residues 63 and 73 (Derijard et al. 1994). Depending upon the cell type and experimental condition, JNK activity may either promote or inhibit apoptosis (Ip and Davis 1998). Activation of JNKs has been strongly implicated in the induction of PCD in response to removal of growth (survival) factors from neurons (Le-Niculescu et al. 1999; Xie et al. 1995). Moreover, mice deficient in JNK3 show defects in PCD-induction in specific populations of neurons (Yang et al. 1997). Interestingly, PCD can be prevented by mutation of c-jun serine residues 63 and 73 to alanine residues that cannot be phosphorylated (Le-Niculescu et al. 1999). Recent data have demonstrated increased phosphorylation of c-jun on serine-63 and increased PCD in response to Salmonella infection of porcine intestinal epithelium (Schauser et al. 2005).

It has been very contentious as to whether PCD is involved in cell loss from intestinal villi. Morphological studies have primarily revealed cells with an apoptotic morphology in crypts, where they have been considered to reflect adjustments of the size of the proliferating compartment (reviewed in Wilson and Potten 2004). On the other hand, the TUNEL method, which demonstrates DNA degradation, has revealed a large number of positive cells on the tips of the villi, as would be expected if PCD preceded cell extrusion (Gavrieli et al. 1992). Finally, studies of activated caspase-3 have indicated that caspase-3 is activated prior to cell extrusion (Groos et al. 2003; Grosssman et al. 2002).

A key issue that remains to be resolved is whether PCD is the cause or the result of extrusion of epithelial cells on villi. Ultrastructural evidence definitely demonstrates that villus epithelial cells are extruded (Potten and Allen 1977). The ensuing loss of contact with the basement membrane may result in a form of PCD known as anoikis (Frisch and Francis 1994). It is therefore possible that PCD is a consequence and not the cause of extrusion. Caspase-3 is known to cleave proteins involved with cell-matrix adhesion, including FAK and paxillin, as well as proteins involved with cell–cell adhesion, like E-cadherin (Fischer et al. 2003). Thus, activation of caspase-3 may contribute to cell extrusion and may inhibit survival signaling from cell adhesions (Groos et al. 2003; Grosssman et al. 2002). In order to resolve this issue we have used triple- and double-staining techniques to correlate the appearance of morphological markers, DNA fragmentation, caspase-3 activation and c-jun phosphorylation in epithelial cells of rat duodenum. Our studies show that caspase-3 is activated in some small intestinal epithelial cells prior to extrusion and also demonstrate that classical signs of apoptosis appear late in the process of extrusion. However, our results also show that not all cells in the process of extrusion contain activated caspase-3 and that lack of correlations between different PCD markers are likely to be due both to differences in timing of their appearance as well as to the occurrence of different forms of PCD during the normal turnover of the small intestinal epithelium.

Materials and methods

Tissue material

Six female Wistar rats (200 g average body weight) were anaesthetized with Nembutal and perfusion-fixed with 4% paraformaldehyde in 0.1 M sodium phosphate buffer pH 7.4. Specimens from the duodenum were postfixed in the same fixative overnight and were routinely embedded in paraffin. Additionally, rats were killed by decapitation under carbon dioxide anaesthesia. The duodenal mucosa was scraped off using a glass slide and the scrapings immediately frozen in liquid nitrogen.


Fluorescent TUNEL method

Dewaxed paraffin sections were treated in 0.1 M sodium citrate buffer pH 6.0, using a Polar Patent microwave (700 W) oven at 40% effect (90°C) for 5 min and were then exposed to the reagents of the “In situ Cell Death Detection Kit” (Roche) for 1 h at 37°C. Following a rinse in Tris-buffered saline, sections were mounted in antifade medium (DakoCytomation, Glostrup, Denmark). This method results in terminal deoxynucleotidyl-transferase-catalyzed incorporation of fluorescein isothiocyanate- (FITC-) labeled nucleotides into the 3′-ends of DNA. Thus, cells with fragmented DNA show intensely green-fluorescent nuclei. Controls were incubated in the absence of enzyme and were uniformly negative. Additionally, TUNEL staining was performed using pretreatment with proteinase K as described by Cao et al. (2000).

Immunofluorescence and DNA staining

Dewaxed paraffin sections were hydrated and microwaved in 0.01 M sodium citrate buffer pH 6.0 for 1 min at a setting of 80% and, subsequently for 9 min at 40%. Following preblocking with 1% bovine serum albumin (BSA), sections were exposed overnight to affinity-purified rabbit antisera recognizing caspase-3 zymogen (no. 9662, diluted 1:200), activated, cleaved caspase-3 (no. 9661, diluted 1:1600) or c-jun phosphorylated on serine-63 (no. 9261, diluted 1:50–1:100) (Cell Signaling Technology, Beverly, MA, USA). In addition, another antibody to active caspase-3 (Abcam, Cambridge, MA, USA) was tested. To minimize unspecific staining, all antibodies were preabsorbed overnight with 1.6 mg/ml low molecular weight poly-L-lysine (Sigma) (Larsson 1988). The site of antigen–antibody reaction was routinely revealed with Alexa488-labeled antirabbit Ig (Molecular Probes, Eugene, OR, USA). Sections were counterstained with bisbenzimide (Sigma, St Louis, MO, USA) for identification of nuclear morphology and mounted in antifade medium. Controls included absorption of the primary antibodies with the corresponding immunizing peptides as well as conventional staining controls (Larsson 1988) and were uniformly negative.

Double-staining using rabbit antibodies to activated caspase-3 and phospho-c-jun were carried out by a modification of the method by Tornehave et al. (2000). In brief, sections were first reacted with the first primary antibody followed by Alexa488-labeled anti-rabbit Ig, as described above. Subsequently, sections were treated with 4% paraformaldehyde in 0.01 M sodium phosphate buffer, pH 7.4 to firmly fix the antibodies to the sites they had reacted with. In this way, loss of staining during subsequent microwaving was avoided. As in the original Tornehave et al. (2000) procedure, reactive immunoglobulin sites were next blocked by three 5-min cycles of microwaving at 100% in 0.06 M sodium citrate buffer pH 6.0. After renewed preblocking with 1% BSA, sections were reacted with the second rabbit antibody, using detection with Alexa594-labeled anti-rabbit Ig combined with staining for DNA using bisbenzimide. The efficiency of microwave-induced blocking was tested by using peptide-absorbed antisera as well as normal rabbit serum in the second staining cycle. In addition, we combined TUNEL detection (FITC) with double immunofluorescence using Alexa594-labeled and amino methyl coumarin- (AMCA-) labeled second antibodies. In this way, TUNEL-positive cells were visualized by green fluorescence while activated caspase-3 and phospho-c-jun was demonstrated by blue (AMCA) and red (Alexa594) immunofluorescence, respectively.

SDS-PAGE and immunoblotting

Mucosal scrapings were homogenized either in buffer A (50 mM PIPES pH 6.5, containing 2 mM EDTA, 0.1% CHAPS, 5 mM dithiothreitol, 1 mM phenylmethyl sulphonyl fluoride, 20 μg/ml leupeptin, 10 μg/ml pepstatin and 10 μg/ml aprotinin (all from Sigma)) or buffer B (50 mM HEPES pH 7.6, containing 100 mM NaCl, 10 mM EDTA, 1% Triton X-100, 4 mM sodium pyrophosphate, 2 mM sodium orthovanadate 10 mM sodium fluoride (all from Sigma), complete protease inhibitor tablet (Roche) and 20 μM Z-VAD-fmk (Promega)). SDS-PAGE and electroblotting were performed as described (Larsson 2004). Immunoblots were stained with antibodies recognizing cleaved caspase-3 or caspase-3 zymogen using chemiluminescent detection.


Staining for DNA using bisbenzimide revealed that only rare cells showed a nuclear morphology characteristic of classical apoptosis, including nuclear shrinkage and fragmentation. Such cells were detected in the epithelium of both crypts and villi . Most apoptotic cells on the villi were in the process of extrusion (Fig. 1a, b).

We next used the TUNEL method for detecting cells with fragmented DNA. Both methods of pretreatment (microwaving or proteinase K) produced identical results. Such cells were more numerous than cells exhibiting morphological signs of apoptosis and were detected at the tips of the villi (Fig. 2a) and, more rarely, in crypts. Some of the TUNEL-positive cells were in the process of being extruded from the epithelium, but most of them showed no signs of extrusion (Fig. 2a). Double-staining with the TUNEL method and bisbenzimide showed that only a subpopulation of TUNEL-positive cells possessed an apoptotic morphology. Thus, most TUNEL-positive cells failed to show evidence of nuclear condensation and fragmentation (Fig. 2a, c). Interestingly, some of the TUNEL-positive cells showed an attenuated staining for DNA (Fig. 1c, d) that was easily discernible from the nuclear staining seen in apoptotic cells. Such attenuated staining was only detected in a small subpopulation of TUNEL-positive cells.
Fig. 1

Detection of caspase-3 in relation to nuclear morphology and DNA fragmentation (TUNEL method) in rat duodenum. a, b Nuclear shrinkage and condensation (a) combined with staining for active caspase-3 (b) in a cell being extruded from the villus epithelium (arrow). c, d Double-staining demonstrates attenuated staining for DNA (d) in a TUNEL-positive cell (arrow). e, f Double-staining for active caspase-3 (green) and DNA (blue). In e is shown a cell early in the process of extrusion (note the inverted triangular shape) while f shows a cell late in the extrusion process. g, h depicts two cells positive for active caspase-3 (g). Note that one cell shows localization of active caspase-3 in the nucleus while the other cell shows cytoplasmic localization of the active enzyme. i, j Tangential section through a villus tip double-stained for DNA fragmentation (i) and active caspase-3 (j). Note that only one (arrow) out of four TUNEL-positive cells reacts for active caspase-3. k, l Double staining for caspase-3 (antibody 9662, detecting both zymogen and active caspase-3; k) and DNA. A double exposure is shown in (l). Note that one epithelial cell shows strong staining while the remaining epithelial cells show weaker staining

Fig. 2

Detection of c-jun phosphorylated on serine 63 (phospho-c-jun) in relation to active caspase-3, DNA fragmentation (TUNEL method) and nuclear morphology in rat duodenum. a, b Triple staining for DNA fragmentation (a), phospho-c-jun (b) and nuclear morphology (c). A double exposure of A and B is shown in (d). Note that nuclei of many cells close to the villi tip stain for phospho-c-jun but that TUNEL-positive cells show normal nuclear morphology and are only weakly positive for phospho-c-jun. e, f Double staining for phospho-c-jun (e) and DNA (double exposure: f). Note that many cells of the senescent compartment show nuclear staining for phospho-c-jun. Tangential section through a villus tip triple stained for DNA fragmentation (g), phospho-c-jun (h), and active caspase-3 (i) with a triple exposure shown in (j). Note one cell is positive for all three markers (arrow) while other cells are positive for only one or none of the markers

Since caspase-activated DNase requires caspase-3 for activation (Jänicke et al. 1998), we next examined whether the distribution of activated caspase-3 correlated to TUNEL positivity. Staining using an antibody specific for activated (cleaved) caspase-3 (9661) resulted in crisp staining of scattered epithelial cells that exclusively occurred on the villi (Fig. 1b, e–h). Absorptions against the immunizing peptide eliminated all staining and immunoblottings showed that the antibody detected two bands of 17- and 19-kDa, corresponding to the sizes expected from the large fragments of activated caspase-3 (Fig. 3). Additionally, an antibody to activated caspase-3 from another source produced identical results. Cells positive for active caspase-3 showed cytoplasmic and/or nuclear staining (Fig. 1b, e–h). However, cells showing only nuclear staining were very rare. Three classes of immunopositive cells were detected. The first class comprised cells with a broad base facing the basement membrane, which showed no morphological signs of apoptosis (Fig. 1g–h). The second class comprised cells having the shape of an inverted triangle with the apex facing the basement membrane and showing no morphological signs of apoptosis (Fig. 1e). The third class comprised cells that were only loosely attached to the epithelium (Fig. 1b, f). Most of the latter cells showed an apoptotic nuclear morphology (Fig. 1a, b). Use of an antibody detecting caspase-3 irrespective of activation (9662) revealed strong staining of scattered epithelial cells that corresponded in number and morphology to the cells detected with the antibodies to active caspase-3 (Fig. 1k, l). In addition, weaker staining of all epithelial cells on the villi was observed (Fig. 1k). Absorptions against the immunizing peptide removed all staining of both strongly and weakly stained epithelial cells. Immunoblottings demonstrated that this antibody detected caspase-3 zymogen of 35-kDa as well as the 17 and 19-kDa forms of active caspase-3 (Fig. 3). Identical results were obtained whether or not a caspase inhibitor (Z-VAD-fmk) was included in the extraction medium. This result shows that the majority of villus epithelial cells contain low amounts of caspase-3 zymogen. Additionally, we performed double-stainings, combining the TUNEL method with immunostaining for active caspase-3. Since the antibody recognized proteolytically cleaved caspase-3, we used pretreatment with microwaving instead of proteinase K in the TUNEL method. The results showed that only partial correspondence between the methods existed and quantitations showed that only 21% of all TUNEL-positive cells were positive for active caspase-3 and that only 29% of cells positive for active caspase-3 were TUNEL-positive (Figs. 1i, j, 2g–j).
Fig. 3

Immunoblotting of extracts from rat duodenal mucosa demonstrating that the antibody to active caspase-3 (9661) detects the 17 and 19 kDa components of the large fragment of activated caspase-3 and that the antibody raised to caspase-3 zymogen (9662) in addition detects the non-cleaved 35 kDa form

These results show that three different methods that are widely used for detecting apoptotic cells produce highly divergent results. Cells with classical apoptotic morphology were rare but cells showing evidence of caspase-3 activation or DNA fragmentation were more abundant. Importantly, some cells containing fragmented DNA were negative for caspase-3.

The stress-induced (JNK) protein kinase cascade has been found both to protect and to inhibit PCD by mechanisms that involve JNK-catalyzed phosphorylation of c-Jun on serines 63 and 73 (Le-Niculescu et al. 1999). Immunostaining with an antibody specific to c-Jun phosphorylated on serine 63 revealed strong nuclear staining of many cells present at and around the villi tips (Fig. 2b, e). Most of these cells were TUNEL-negative and showed normal nuclear morphology. In order to correlate staining for phospho-c-jun to activation of caspase-3, double immunofluorescence was used. Since the antibodies to active caspase-3 and phospho-c-jun were derived from rabbits we used the double-staining method of Tornehave et al. (2000), which employs microwaving to inhibit crosstalk between subsequent staining cycles (cf. Materials and Methods). Two modifications of the published procedure were introduced. Firstly, sections were fixed in paraformaldehyde prior to microwaving. This fixed the antibodies to the sites with which they had reacted and, thus, prevented elution of antibodies during microwaving. This greatly improved the results obtained in the Tornehave et al. (2000) procedure by yielding more intense immunofluorescent staining. Secondly, the site of antigen–antibody reaction in the first staining cycle was revealed using Alexa488-labeled anti-rabbit Ig antibodies in lieu of fluorescein isothiocyanate- (FITC-) labeled antibodies. The Alexa488-labeled antibodies tolerated microwaving well and were much more photostable than FITC-labeled antibodies. Controls included use of antigen-preabsorbed primary antibodies as well as substitution of the primary antibodies with normal rabbit serum and showed no crosstalk between the staining cycles. By this approach we could triple-stain for DNA fragmentation, phospho-c-jun and active caspase-3 and the results obtained showed that only very few cells were positive for all three markers (Fig. 2g–j). Thus, the presence of nuclear staining for phospho-c-jun marked many cells present in the senescent villus compartment, but most of these cells showed a non-apoptotic nuclear morphology and were negative for caspase-3 activation or DNA fragmentation.


Our data show that three widely used methods for detecting PCD produce highly divergent results in rat intestinal epithelial cells. To some extent these divergences may reflect differences in timing of PCD events. Thus, activated caspase-3 was detected both in epithelial cells with a normal morphology and in cells that were in the process of being extruded into the lumen. These results suggest that activation of caspase-3 may be an event preceding cell extrusion. A similar mechanism has been proposed in other studies (Groos et al. 2003; Grossman et al. 2002). Importantly, most cells positive for active caspase-3 did not display signs of apoptosis. Again, this could reflect that activation of caspase-3 precedes extrusion and onset of apoptosis. Use of double and triple-staining methods made it possible to correlate the occurrence of different markers in the same cells. This approach showed that TUNEL staining only partially correlated to staining for activated caspase-3. Thus, many cells that were positive for fragmented DNA did not stain for activated caspase-3, suggesting that fragmentation of DNA in such cells occurred by caspase-3 independent pathways. The distribution of staining using the TUNEL method definitely shows that this method detects a subpopulation of cells displaying classical morphological signs of apoptosis. In addition, many TUNEL-positive cells failed to show morphological signs of apoptosis and were negative for active caspase-3. This agrees with the findings showing that caspase-3 is not indispensible for PCD (Jänicke et al. 1998) and with findings made in intestinal pathological conditions (An et al. 2005; Schauser and Larsson 2005). Interestingly, some of the TUNEL-positive cells observed in the present study displayed an attenuated staining for DNA that seems not to have been noted before. These observations suggest that different mechanisms may lead to DNA fragmentation in rat small intestinal epithelial cells and that only a small fraction of cells with fragmented DNA display a nuclear morphology indicative of apoptosis.

In addition, our studies reveal an interesting distribution of nuclear staining for c-Jun phosphorylated on serine 63. Cells staining for this transcription factor occur on and close to the villi tips and show a variable reactivity for activated caspase-3 and TUNEL positivity. This distribution indicates that senescent epithelial cells show increased c-Jun phosphorylation. Interestingly, increased c-jun phosphorylation and increased PCD was also detected in pig intestines in response to Salmonella typhimurium infection (Schauser et al. 2005). However, the observation that many cells containing active caspase-3 and/or fragmented DNA were negative for phospho-c-jun indicates that c-jun phosphorylation is not obligatory for PCD in the intestinal epithelium. An alternative possibility is that c-jun is cleaved by caspases leading to its absence in most cells positive for active caspase-3. Our data indicate that differences in results using different markers for PCD reflect differences both in the timing of appearance of these markers in individual cells as well as the involvement of different mechanisms for PCD in the normal turnover of the small intestinal epithelium. Muti-labeling approaches for the simultaneous detection of multiple PCD markers are essential for dissecting the obviously very complex mechanisms of small intestinal cell loss.


Grant support was from the Danish MRC, FTP, Cancer Society and the Lundbeck foundation.

Copyright information

© Springer-Verlag 2005