Author + information
- Received March 12, 1999
- Revision received May 20, 1999
- Accepted June 29, 1999
- Published online November 1, 1999.
- Christine Aimé-Sempé, PhD∗,
- Thierry Folliguet, MD§,
- Catherine Rücker-Martin, PhD‡,
- Maryla Krajewska, MD∥,
- Stanislaw Krajewski, MD∥,
- Michèle Heimburger, PhD†,
- Michel Aubier, MD, PhD∗,
- Jean-Jacques Mercadier, MD, PhD†,
- John C Reed, MD, PhD∥ and
- Stephane N Hatem, MD, PhD†,* ()
- ↵*Reprint requests and correspondence: Dr. S. Hatem, INSERM Unité 460, Faculté de Médecine Xavier Bichat, 16 rue Henri Huchard, 75018 Paris, France
The aim of the present study was to determine if myocytes can die by apoptosis in fibrillating and dilated human atria.
The cellular remodeling that occurs during atrial fibrillation (AF) may reflect a degree of dedifferentiation of the atrial myocardium, a process that may be reversible.
We examined human right atrial myocardium specimens (n = 50) for the presence of apoptotic myocytes. We used immunohistochemical and Western blotting analysis to examine the expression of a final effector of programmed cell death, caspase-3 (CASP-3) and of regulatory proteins from the BCL-2 family.
Sections from atria in AF contained a high percentage of large myocytes with a disrupted sarcomeric apparatus replaced by glycogen granules (64.4 ± 6.3% vs. 12.2 ± 5.8%). These abnormal myocytes, which also predominated in atria from hearts with decreased left ventricular ejection fraction (42.3 ± 10.1%), contained large nuclei, most of which were TUNEL positive, indicating a degree of DNA breakage. None of these abnormal myocytes expressed the proliferative antigen Ki-67. A small percentage of the enlarged nuclei (4.2 ± 0.8%) contained condensed chromatin and were strongly TUNEL positive. Both the pro- and activated forms of CASP-3 were detected in diseased myocardial samples, which also showed stronger CASP-3 expression than controls. Expression of the antiapoptotic BCL-2 protein was decreased in diseased atria, whereas that of the proapoptotic BAX protein remained unchanged.
In fibrillating and dilated atria, apoptotic death of myocytes with myolysis contributes to cellular remodeling, which may not be entirely reversible.
Atrial fibrillation (AF) is the most frequent sustained cardiac arrhythmia and causes serious deleterious effects such as impairment of cardiac function or thromboembolic events (1). This arrhythmia is due to circuits of micro-reentry of electrical impulses in the atrial wall, a process referred to as the multiple-wavelets mechanism (2). The incidence of wavelets that can coexist is determined by both the mass and the electrical vulnerability of the atrial myocardium, explaining why AF is often observed in clinical situations associated with atrial enlargement and shortening of the atrial refractory period (3,4). The other major characteristic of AF is its tendency to become sustained, suggesting that AF induces a vicious circle leading to the increase in its arrhythmogenic substrate (4).
Distinct cellular alterations occur in AF that may also contribute to its perpetuation. In human atrial biopsies or in animal models of AF, a number of myocytes lose their sarcomeric apparatus, which is replaced by accumulation of glycogen granules (5–9). Because these abnormal myocytes reexpress some fetal proteins, it has been proposed that they are dedifferentiated, reflecting a certain degree of adaptation of the atrial myocardium to changes in its working conditions, a process that may be reversible (9). It is also conceivable that as the result of the persistence of pathogenic factors such as increased pressure load or high frequency of beating, some of the myocytes with structural alterations could activate a programmed cell death (PCD) pathway.
Programmed cell death is an active cell suicide mechanism culminating in characteristic features commonly known as apoptosis. During this process, caspases, which are highly specific cysteine proteases, function in both cell disassembly and in initiating this disassembly in response to proapoptotic signals by cleaving a discrete set of proteins such as ICAD/CAD, gelsolin or proteins involved in DNA repair such as Poly (ADP ribose) polymerase (10,11). The caspases are all expressed as proenzymes that contain three domains: an NH2-terminal domain, a large and a small subunit. Activation involves proteolytic processing between domains, followed by association of the large and small subunits to form a functional heterodimer. The Caspase-3/CPP32 (CASP-3) is one of the major caspases involved in apoptosis (11). BCL-2 is the founding member of a growing family of proteins that either protect against cell death such as BCL-2 or that promote apoptosis such as BAX (12). The data available to date indicate that BCL-2 family proteins may dictate whether effector caspases become active or remain quiescent after exposure of cells to apoptotic stimuli (13). Both caspases and BCL-2 family proteins are involved in the apoptosis of cardiac myocytes, which occurs in various cardiopathies (14–17).
The aim of the present study was to examine if atrial myocytes can undergo an apoptotic-like death process in chronic AF or in clinical settings associated with a high incidence of this arrhythmia.
With approval from our Ethics Committee, specimens of the right atrial appendage were obtained from 50 patients (31 to 86 years old) undergoing heart surgery for coronary artery disease, valve disease or congenital heart defects (Table 1). Patients with chronic AF had left and right atrial dilation (>45 mm), most often associated with increased systolic pulmonary pressure (>35 mm Hg) indicating longstanding valve disease (18). Specimens of the right atrial appendage were fixed immediately after excision and maintained in 10% buffered formalin overnight or fixed 3 h in Bouin solution and embedded in paraffin or frozen in liquid nitrogen (L2N) or isopentane.
Tissue sections were deparaffinized, transferred to xylene and rehydrated in decreasing concentrations of alcohol. Then slides were stained with hematein, eosin, orcein, Masson’s trichrome or periodic acid schiff. Myocyte diameter was determined by measuring the short axis of 100 randomly chosen cells per field. To quantify the amount and the percentage of myocytes with myolysis, two sections per atrial samples were analyzed, and at least 400 cells per section were analyzed. Cells were scored by morphometry as severely myolytic if >25% of the sarcomere was absent.
Cryo-sections (5 μm) were incubated in phosphate-buffered saline (PBS) containing 5% BSA, followed by incubation with mouse anti-rabbit sarcomeric α-actinin antibody (1/400; Sigma Aldrich, St. Louis, Missouri), then incubated with goat biotinylated anti-mouse IgG secondary antibody (1/30; Vector Laboratories, Paris, France) and with streptavidin-Texas red or fluorescein (1/30; Amersham). Some observations were carried out with an MRC-1024 (Bio-Rad, UK) confocal scanning laser with microscope (Nikon Optiphot Fluorescence) using Lasersharp (Bio-Rad).
For expression and localization of CASP-3, BCL-2 and BAX proteins, deparaffinized sections were either used directly or subjected to microwave oven antigen retrieval treatment (19,20). Tissue sections were immunostained with a rabbit anti-CASP-3, anti-BCL-2 or anti-BAX antiserum. The immunoreactions were visualized with a diaminobenzidine (DAB)-based colorimetric method, as previously described (20). Slides were lightly counterstained with hematoxylin and permanently mounted with De-Pe-X (Fluka, Buchs, Germany) before being analyzed and photographed with an Olympus BH-2 microscope connected to a camera. The specificity of the staining was confirmed by using preimmune serum and/or by preabsorbing the antiserum with specific peptide or recombinant protein before performing immunostaining.
Protein lysates were prepared, normalized for total protein content (200 or 100 μg per lane) and analyzed by 12.5% or 10% SDS-PAGE, followed by electro-transfer to polyvinylidene difluoride membranes (Bio-Rad). Immunodection was performed using anti-CASP-3 (19), anti-BCL-2, anti-BAX (20)antitubulin or anti-α-actinin sarcomeric monoclonal (Sigma, St. Louis, MO) antibodies followed by horseradish peroxydase (HRPase)-conjugated secondary antibody (Amersham Life Sciences). Detection was performed using enhanced chemiluminescence detection method (ECL+; Amersham Life Science). Densitometric analysis was performed using Gel analyst system (ICONIX).
Staining of Ki-67 antigen
The staining of Ki-67 antigen was performed on paraffin-embedded sections, using the MIB-1 monoclonal antibody (1/1,000 dilution; DakoA/S, Copenhagen, Denmark). Detection was accomplished using streptavidin biotinylated horseradish peroxydase method (Dako). DAB tetrahydrochloride 0.06% in PBS containing 0.03% hydrogen peroxyde was used as a chromogen. As positive control, we used sections from squamous cell carcinomas.
Nuclear staining with DAPI
Deparaffinized tissue sections were incubated with the intercalating agent DAPI (4′, 6-diamidino-2-phenylindole, 0.3 μg/ml in PBS, 15 min) to visualize nuclear morphology.
Sections were transferred to xylene and rehydrated in decreasing concentrations of alcohol. Slides were then incubated (10 min, room temperature) with 10 μg of proteinase K (Sigma-Aldrich, St. Louis, MO) per milliliter of PBS. Endogenous peroxydase was inactivated by immunopure peroxydase suppressor for 30 min (Pierce). Tissue sections permeabilized with 1% Triton X-100 (4°C, 2 min) were stained with an in situ cell detection (POD) system (Boehringer Mannheim, Mannheim, Germany). DNA-strand breaks were identified by labeling free 3′-OH termini with dUTP-FITC using the terminal deoxynucleotidyl transferase (Tdt; TUNEL). Incorporated fluorescein was detected by anti-fluorescein antibody Fab fragment from sheep conjugated with horseradish peroxydase (POD). After reaction with the substrate metal-enhanced DAB (Boehringer Mannheim, Germany), sections were counterstained with hematoxylin. Positive controls consisted of incubating fixed and permeabilized sections with DNase I (1 μg/ml) (10 min, room temperature). For negative controls, sections were incubated in labeling solution without Tdt. The percentage of TUNEL − positive nuclei was calculated as followed: 50 randomly chosen fields per section corresponding to approximately 700 cells were examined at high magnification (×400).
DNA gel electrophoresis
Fragments of atrial myocardium were crushed in liquid nitrogen, homogenized, fixed with 70% ethanol and incubated in 40 ml of phosphate-citrate buffer (pH 7.8). The pellet was resuspended in 0.5 ml of lysis buffer containing 75 mM NaCl, 0.5% SDS, 10 mM Tris-HCl, 10 mM EDTA, pH 8.0 and 0.15 mg/ml proteinase K (Sigma-Aldrich) and incubated at 50°C for 3 h. The lysate was then incubated with 200 μg/ml RNase A (Sigma-Aldrich) at 37°C for 1 h (25). After extraction with phenol and chlorophorm, DNA was precipitated with 2 vol of ethanol and centrifuged; the pellet was washed with 70% ethanol and resolubilized in an appropriate volume of TE (10 mM Tris [Ethylenediaminetetraacetate disodium]- HCl, 1 mM EDTA, pH 8.0). DNA was quantified by means of spectrophotometry. Equal amounts of each sample (10 μg) were loaded on 1.5% agarose gels containing 0.5 μg/ml ethidium bromide, alongside 2 μg of molecular weight DNA marker (123-bp DNA ladder; Sigma-Aldrich). DNA laddering was visualized under ultraviolet light.
Data on structural cellular changes were tested for statistical significance by a one-way analysis of variance for multiple comparison. A p value below 0.05 was considered significant after correction with Fischer post hoc statistical test. Values are expressed as mean ± standard error of the mean.
Marked structural cellular alterations in diseased atria
Figure 1shows a typical tissue section of the right atrial specimen of a patient in sinus rhythm without atrial dilation. Myocytes of relatively uniform size were regularly arranged in parallel to their long axis. Muscle bundles were surrounded by a thin connective tissue with absent-to-moderate interstitial fibrosis (Fig. 1, A and C). By contrast, tissue sections from a dilated and chronically fibrillating atrium exhibited large areas of extensive structural alterations. Myocytes were organized in large and tortuous strips, separated from each other by an important interstitial fibrosis made up of the accumulation of collagen and elastic fibers (Fig. 1, B and D). In these areas, myocytes were of irregular shape and of large size (diameter: 21.2 ± 0.3 vs. 11.5 ± 0.2 μm, in diseased and control samples respectively; p < 0.001). A number of these large myocytes showed marked structural alterations, including 1) the depletion of contractile materials limited to the vicinity of the nucleus or frequently involving the entire cytosol and 2) the cytosol depleted of myofibrillar structure was filled with glycogen granules (Fig. 1, E and F). Figure 2shows a confocal microphotograph of tissue sections labeled with anti-α-actinin antibodies. Whereas in control atria, myofibrils were organized in a well-aligned and striated network, diseased atria showed vast areas of myofibrillar disruption. However, the remaining myofibrils maintained their striated organization and showed no contraction bands, which indicated preservation of membrane integrity (Fig. 2).
In tissue section from fibrillating and dilated atria, 64.4 ± 6.3% of total myocytes (n = 11 patients) showed severe myolysis, whereas only 12.2 ± 5.8% of total myocytes (n = 9 patients) consisted of these cells in myocardium from atria in sinus rhythm. However, a high percentage of myocytes with myolysis (42.3 ± 10.1% of total myocytes, n = 8 patients) was also found in samples obtained from the hearts of patients with a low left ventricular ejection fraction (LLVEF) in sinus rhythm (Fig. 3). Most of these patients showed a dilated atria and/or increased pulmonary artery pressure, suggesting a degree of hemodynamic overload of their right atria (Table 1).
Nuclear alterations in myocytes with myolysis
In control sections, the majority of the myocytes showed nuclei of uniform size with an ovoid shape and an apparent regular distribution of the heterochromatin (Fig. 4A). In sections of diseased myocardium, most of the myocytes with myolysis had markedly enlarged nuclei, with an evenly distributed heterochromatin (Fig. 4B). However, occasional nuclei were shrunken, with condensed and reorganized heterochromatin (Fig. 4D), and more rarely, nuclei appeared fragmented (Fig. 4C). To determine if some of these large nuclei might represent attempted mitosis, sections (n = 4) were stained with the antibody MIB-1 directed against the antigen Ki-67, the expression of which is associated with the reentry of cells into the division cycle (21). In both control and diseased specimens, we failed to detect the expression of this antigen.
In situ detection of DNA cleavage using the TUNEL method showed that a high percentage of myocytes with myolysis had nuclei TUNEL positive (45.2 ± 5%, n = 8), whereas almost none of the myocytes without structural alterations contained TUNEL-positive nuclei (Fig. 5A). The vast majority of nuclei with weak TUNEL staining were of large size with a uniform distribution of their heterochromatin (counterstained with hematoxylin) (Fig. 5B).
However, occasional nuclei (4.2 ± 0.8%, n = 8) were shrunken with a strong TUNEL staining, indicating that more extensive DNA cleavage was associated with these nuclear alterations (Fig. 5, C and D). The observation in tissue sections of only rare cells with strong nuclear TUNEL positivity was also consistent with the lack of clear typical ladders of regular size DNA detected after gel electrophoresis (Fig. 6).
Activation of PCD in diseased atrial myocardium
The lack of detection of DNA nucleosome ladders may be insufficient to eliminate the activation of PCD, which is a transient and fast process involving a limited number of myocytes. Indeed, immunohistochemistry analysis of the expression of CASP-3 revealed that the expression of this protease was enhanced in diseased (n = 5) compared with control samples (n = 6). This is illustrated in Figure 7, which shows that in a section from the diseased atrium (Fig. 7, Bto D), marked CASP-3 staining was observed predominantly in myocardial areas characterized by a high percentage of myocytes with sarcomeric depletion. The CASP-3 immunostaining exhibited a coarse-grained appearance, and in some myocytes the deposits overlaid the nucleus. The majority of the CASP-3-labeled myocytes showed hyperchromatic nuclei with features of shrinkage. A fine perinuclear weak staining was observed in myocytes from the control atrium (Fig. 7A). In all protein samples studied by immunoblotting, we detected the pro-form (p32) of the CASP-3 without significant difference in its level of expression among the different specimens. Only in patients with chronic AF or with LLVEF was the activated form (p17) of the CASP-3 detected (Fig. 8A). It is conceivable that the higher expression of CASP-3 detected in diseased samples compared with the control could result from the addition of both p32 and activated forms.
Further evidence that PCD is activated in diseased atrial myocardium was provided by studying the expression of the proapoptotic BAX and the antiapoptotic BCL-2 proteins. The analysis of these two proteins by immunochemistry (data not shown) and immunoblotting showed that both were expressed in control and diseased myocardium (Fig. 8B). Whereas BAX expression was not significantly different among the different samples studied, the expression of BCL-2 was downregulated in patients with chronic AF or LLVEF. Densitometric analysis confirmed that the ratio of p26-BCL-2/BAX was decreased in patients in AF or with LLVEF (1.42) compared with controls (2.15). Of note, in addition to the classical p26-kDa band representing BCL-2, we consistently detected a smaller band of approximately 23 kDa, which could indicate the presence of a cleaved form of BCL-2.
The results of this study indicate that fibrillating and hemodynamically overloaded atria contain a number of myocytes undergoing apoptosis. We found profound cyto-architectural and cellular alterations of the human right atrial myocardium. These alterations predominated in biopsy specimens from chronically fibrillating and dilated atria, but they were also seen in atrial samples from the heart in sinus rhythm with altered left ventricular function. In both groups of patients, a number of myocytes with myolysis had activated a cell suicide program resembling apoptosis.
Approximately 12% of myocytes in the control atria showed marked sarcomeric depletion, compared with 60% in specimens of fibrillating atria and 40% in hearts with LLVEF. The cellular alterations observed here are more severe that those found in previous studies. For instance, Mary-Rabine et al. (6)found that <5% of myocytes had structural alterations in the majority of their adult right atrial biopsies and that this value exceeded 10% only in markedly dilated and fibrillating atria. The discrepancy between these two studies may reflect the fact that this process predominates in the right atrial appendage, whereas the anterior free wall was studied by Mary-Rabine et al. (6). Indeed, in a goat model of pacing-induced AF, cellular alterations similar to those observed in our study also predominated in the right appendage (9). Furthermore, the marked cellular remodeling observed in patients in sinus rhythm with LLVEF could contribute to the high incidence of AF observed during heart failure (22).
The mechanisms underlying the cellular alterations seen in diseased atria are still poorly understood (8,9). The resemblance between myolytic atrial cells and myocytes seen during cardiac development or in less specialized myocardial tissue had led to suggestions that such myocytes may have undergone a dedifferentiation process. This is also suggested by the reexpression of some fetal phenotypic features such as smooth α-actin (8)and the transition from α- to β-myosin heavy chain observed in the present study (data not shown, 23). The notion that myocytes with myolysis are in a dedifferentiated state could explain why these cells contain a large nucleus with a homogenous distribution of heterochromatin resembling that of fetal/embryonic cardiac myocytes. However, these myocytes did not express the nuclear antigen Ki-67, indicating that they had not reentered the cell cycle. The high percentage of nuclei with dUTP(deoxyuridine triphosphate)-positive labeling but no nuclear abnormalities indicates a degree of DNA breakage, which could reflect some level of transcription or defective DNA repair. In chicken myoblasts and in human lymphocytes, it has been shown that dedifferentiation is associated with the occurrence of DNA-strand breaks, a process that may signify some underlying mobility of the genome and be required to relax tightly packed chromatin for the transcription of new genes (24,25).
Evidence of activation of PCD in atrial myocardium
Although myocytes with myolysis may exhibit some features of dedifferentiation, it remains to be determined whether the cells are in a long-term viable state that eventually can be reversible or they eventually die by activating PCD. The possibility that some myolytic myocytes undergo apoptotic death was raised by the observations of shrunken nuclei with clumps of chromatin and strong TUNEL staining. Because myocytes showing these nuclear and DNA alterations had intact membranes (indicated by the persistence of striated myofibrils and the lack of contraction bands), cellular necrosis can be eliminated from consideration. Moreover, the activation and increased expression of CASP-3, a molecule involved in the final executional step of PCD, in diseased atria is also consistent with the notion that a number of myocytes are undergoing an apoptotic death. This biological process is regulated by proteins such as members of the BCL-2 family, and whereas the expression of the proapoptotic protein BAX was the same in all specimens studied, we found that the expression level of the antiapoptotic protein BCL-2 was reduced in specimens of diseased atrial myocardium. BAX expression is also unchanged in human ventricular myocardium during terminal heart failure, while BCL-2 expression is increased (16). This has been interpreted as evidence for compensatory mechanisms during apoptosis associated with heart failure. Although there is no clear explanation for these differences in the expression of BCL-2 proteins between atrial and ventricular myocardium, these results suggest that in the human heart the ratio of BCL-2 and BAX may play a key role in the balance between life and death (26,27). Recent studies have shown that BCL-2 could be a substrate for proteases such as CASP-3 (28), resulting in the cleavage of the 26-kDa protein to produce a 23-kDa band and removal of the BH4 domain of the protein, a domain necessary for its antiapoptotic function (26,29). The weak 23-kDa band consistently detected in atrial myocardium samples could correspond to the cleaved form of BCL-2. However, we cannot rule out the possibility that this lower band represents the BCL-2β, isoform or another unrecognized tissue-specific form of BCL-2 protein.
Cellular alterations characteristic of the final phase of PCD, such as cytoplasm condensation or nuclear and DNA fragmentation, were rarely detected in this study. One possible explanation is that apoptosis involved a very limited number of atrial myocytes that may rapidly be removed from the myocardium. Indeed, the chronic course of atria diseases is inconsistent with the notion that major cell loss plays an important role in their pathogenesis. An alternative explanation is that myocytes expressing a high level of activated death effectors and reduced BCL-2 protein expression are more susceptible to death signals. In this case, minor additional noxious stimuli may be sufficient to trigger the death of occasional vulnerable myocytes in random fashion. This is somewhat reminiscent of neurodegenerative diseases such as Alzheimer’s disease, wherein neurons showing DNA alterations may not be undergoing apoptotic death but are rather in a living state in which they are more vulnerable to death stimuli (30).
In the goat model of chronic AF, Ausma et al. (8)found no evidence of nuclear alterations resembling an apoptotic process. Beside the difference in species, the discrepancy between these two studies may be explained by the relatively short period of AF in the goat model (a few weeks vs. a few years in the present study), which may be insufficient to activate an apoptotic process. Moreover, most of the patients with chronic AF in the present study had marked atrial dilation, suggesting chronic hemodynamic overload of their atria, which may be an important factor in triggering a PCD pathway. For instance, abnormal levels of resting tension have been shown to induce apoptosis of ventricular myocytes (31). This mechanism may be particularly important for myocytes of thin atrial walls, in which even moderate hemodynamic overload could induce substantial overstretching. This may also explain why atrial samples from patients in sinus rhythm with a LLVEF (which is often accompanied by a degree of hemodynamic overload of the atria) also had marked histological abnormalities and apoptotic cells. Our results raise questions as to the nature of the pathogenic stimuli, i.e., altered hemodynamic load or a high beating rate, that trigger the cell suicide program in atrial myocardium.
Our observation that atrial myocytes with myolysis can undergo apoptotic death implies that the cellular remodeling that occurs in diseased atrial myocardium may be partially irreversible. This could contribute to the difficulty of restoring sinus rhythm and to the self-perpetuation of AF.
We thank the Department of Pathology of Hôpital Marie Lannelongue for their technical assistance.
☆ This work was supported by grants from the Institut National de la Santé et de la Recherche Médicale (INSERM), the Association Française contre les Myopathies (A.F.M.) and from the Assistance Public-Hôpitaux de Paris (AOB94038). Christine Aimé-Sempé was supported by a grant from the Fondation pour la Recherche Médicale (FRM).
- atrial fibrillation
- caspase 3
- caspase-activated deoxyribonuclease inhibitor/caspase-activated deoxyribonuclease
- low left ventricular ejection fraction
- phosphate-buffered saline
- programmed cell death
- Received March 12, 1999.
- Revision received May 20, 1999.
- Accepted June 29, 1999.
- American College of Cardiology
- National Heart, Lung, and Blood Institute Working Group on Atrial Fibrillation
- Moe G.K.
- Allessie M.A.,
- Rensma P.L.,
- Brugada J.,
- Smeets J.L.R.M.,
- Penn O.,
- Kirchof C.J.H.J.
- Mary-Rabine L.,
- Albert A.,
- Pham T.D.,
- Hordof A.,
- et al.
- Frustaci A.,
- Chimenti C.,
- Belloci F.,
- Morgante E.,
- Russo M.,
- Maseri A.
- Ausma J.,
- Wijffels M.,
- Thoné F.,
- Wouters L.,
- Allessie M.,
- Borgers M.
- Thornberry N.A.,
- Lazebnik Y.
- Adams J.M.,
- Cory S.
- Reed J.C.,
- Zha H.,
- Aimé-Sempé C.,
- Takayama S.,
- Wang H.V.
- Misao J.,
- Hayakawa Y.,
- Ohno M.,
- Kato S.,
- Fujiwara T.,
- Fujiwara H.
- Henry W.L.,
- Moranroth J.,
- Pearlman A.S.,
- et al.
- Krajewska M.,
- Wang H.G.,
- Krajewski S.,
- et al.
- Nakopoulou L.,
- Vourlaki C.,
- Zervas A.,
- Tzonou A.,
- Gakiopoulou H.,
- Dimopoulos M.A.
- Mercadier J.J.,
- De La Bastie D.,
- Menasche P.,
- et al.
- Hanada M.,
- Aimé-Sempé C.,
- Sato T.,
- Reed J.C.
- ↵Cheng EHY, Kirsch DG, R.J. Clem RJ et al. Conversion of Bcl-2 to a Bax-like death effector by caspases. Science 1997;278:1966–8.
- Hunter J.J.,
- Bond B.L.,
- Parslow T.G.
- Stadelmann C.,
- Brück W.,
- Bancher C.,
- Jellinger K.,
- Lassmann H.