“Cell Paralysis” as an Intermediate Stage in the Programmed Cell Death of Epiphyseal Chondrocytes During Development


  • Helmtrud I. Roach,

    Corresponding author
    1. University Orthopaedics, University of Southampton, General Hospital, Southampton, United Kingdom
    • University Orthopaedics, University of Southampton, CF86, General Hospital, Southampton, SO16 6YD, U.K.
    Search for more papers by this author
  • Nicholas M. P. Clarke

    1. University Orthopaedics, University of Southampton, General Hospital, Southampton, United Kingdom
    Search for more papers by this author


The efficient elimination of apoptotic cells depends on heterophagocytosis by other cells, which is difficult or impossible when the dying cells are embedded in an extracellular matrix. This situation is exemplified by the epiphyseal chondrocytes during the development of the chondroepiphyses of long bones. A detailed ultrastructural study identified an unusual type of epiphyseal chondrocyte which contained a very dark nucleus with irregular patches of condensed chromatin and a crenated nuclear membrane. The cytosol consisted of excessively expanded endoplasmic reticulum lumen, containing “islands” of cytoplasm and organelles. Since these cells appeared to be “in limbo,” neither viable nor dead, they are referred to as “paralyzed” cells. By studying cells of intermediate morphologies, we were able to demonstrate the sequence of events leading to cell paralysis. It is proposed that the paralysis represents an intermediate state in the physiological cell death of epiphyseal chondrocytes in which destruction is orderly and avoids a inflammatory, potentially locally destructive, reaction. The cell is rendered paralyzed in terms of function but impotent in respect of damaging consequences. Paralysis is compared and contrasted with apoptosis, autophagocytosis, and necrosis and may represent another mode of programmed cell death in situations where cells are immature and/or where phagocytosis by neighboring cells is difficult.


Cell death, which occurs during development or maintenance of the organism, is generally referred to as programmed cell death (PCD). This term refers not only to the cell death that always appears at the same place and the same stage during development, but also to the suicide of superfluous or imperfect cells where death is executed according to an intracellular program. PCD is contrasted with pathological or accidental cell death where cells die as a result of acute injury.

In the majority of cases, physiological elimination of cells is via apoptosis, a mode of death with distinct morphological and biochemical characteristics.(1–4) One of the most distinctive hallmarks of apoptosis is the condensation and fragmentation of the nucleus. Another, and perhaps the most fundamental, feature of apoptosis is that the cell corpses are rapidly phagocytosed by neighboring cells or macrophages, before there is any leakage of cytoplasmic contents and without an inflammatory response.(5,6) This latter feature distinguishes programmed from accidental cell death. While the former is an orderly process, the latter is imprecise because cell lysis causes spill of cellular content into the extracellular space. Because apoptosis depends on the rapid elimination of apoptotic bodies by heterophagocytosis, this mode of death will not be effective in eliminating cells where there is a need for rapid elimination of whole tissues, as for example in insect metamorphosis, or where cells are isolated from each other by extracellular matrix so that apoptotic bodies would be difficult to eliminate by heterophagocytosis.

In the first situation, cells partly take care of their own self-destruction by autophagocytosis.(7–9) The present study investigated the second situation, exemplified by chondrocytes in the developing chondroepiphysis of long bones. These epiphyseal chondrocytes are small, immature cells at an early stage of differentiation. They are embedded within the cartilage matrix, but cell division continues, as indicated by the frequent presence of cell doublets. These epiphyseal chondrocytes differentiate further to become hypertrophic chondrocytes around the secondary ossification center and at the physeal growth plate. Because there is an increased possibility of errors during cell divisions, the cell cycle is linked to apoptosis(10) so that defective cells can be eliminated. A similar situation arises during the transitions to different developmental stages. One would therefore expect to find some evidence of PCD within the chondroepiphysis, and not only in the terminal hypertrophic chondrocytes(11–14) or articular cartilage,(15,16) where PCD has been demonstrated. So far no studies have investigated this possibility, although some evidence for the PCD of epiphyseal chondrocytes was found at the chick growth plate.(17,18)

We carried out a detailed ultrastructural study of the chondroepiphysis of femurs and humeri from pre- and neonatal rabbits, initially with the aim of demonstrating apoptosis. We noticed that a small percentage of epiphyseal chondrocytes possessed a rather unusual morphology: most of the cell consisted of a translucent substance, containing dark inclusions, while the nucleus was similar to, yet different from, an apoptotic nucleus. The present study then concentrated on these chondrocytes and investigated how their unusual morphology originated and whether or not these cells were undergoing PCD.


The heads of femurs and humeri were dissected from rabbits that were −1, 0, 1, and 4 days old. Most of the endochondral bone below the physeal growth plate was removed so that the tissue specimen consisted of the chondroepiphysis, in which the secondary ossification center had started to develop, and the physeal growth plate (for overview, see Fig. 2a).

Figure FIG. 2..

Distribution of paralyzed cells. (a) Overview of the secondary ossification center in the head of humerus from a 4-day-old rabbit. Cartilage canals (stars) surround the ossification center. The center part consists of calcified cartilage/bone trabeculae, partially invaded by vascular elements. A spherical growth plate surrounds the ossification center, and chondrocytes differentiate along the line indicated by the long arrows. The approximate locations of the micrographs seen in (b) and (c) are indicated by solid squares. (b) Paralyzed chondrocytes were present in the transition zone to the hypertrophic state (H). (c) Paralyzed chondrocytes, present singly or in asymmetric doublets (arrows) adjacent to a cartilage canal (cc). Bar = 100 μm in (a) and 5 μm in (b) and (c).

Electron microscopy

The tissues were fixed in 2% gluteraldehyde/0.7% Ruthenium hexamine trichloride in 0.05 M sodium cacodylate for 4−18 h. The latter was added because it considerably improves the fixation of chondrocytes and stains the cartilage matrix by binding proteoglycans.(19) The samples were washed in 0.05 M Na cacodylate containing 0.329% NaCl (3 × 30 minutes); postfixed in cacodylate/1% osmium tetroxide for 1 h; washed in 0.1 M cacodylate (3 × 30 minutes), then dehydrated and embedded in Spurr resin. The femoral and humoral heads were fixed and processed whole, then halved coronally before embedding. Semithin sections were cut at 1 μm and stained with Toluidine blue. Thin sections were contrasted with lead citrate and uranyl acetate.

Light microscopy and terminal transferase mediated d-UTP nick end-labeling method

The specimen were fixed in 4% paraformaldehyde (in 0.1 M phosphate-buffered saline, pH 7.4) for 5–18 h, dehydrated in increasing concentrations of ethanols (50%, 90%, 2 × 100%, 1 h in each), cleared in chloroforms (2 × 1 h) and embedded in paraffin wax. Sections, 5–6 μm thick, were mounted onto poly L-lysine–coated slides. Cells containing fragmented DNA were localized by the terminal transferase mediated d-UTP nick end-labeling (TUNEL) method of Gavrieli,(20) using d-UTP-biotin, Extr-avidin-peroxidase, and 3-amino-9-ethycarbazole to visualize the reaction product. No proteinase K digestion was carried out since this pretreatment had given inconsistent results and often led to positive staining of viable cells. Instead, nuclear proteins were removed by a limited microwave digestion technique. Twenty slides were microwaved in 200 ml of citrate buffer (pH 6.0) for 5 minutes at 40% power in a 980 W microwave oven which caused a temperature increase to about 65°C. The slides were cooled immediately in two changes of cool water. Positive and negative controls were carried out on parallel slides. Positive controls were treated with DNAse (2 U/ml) for 1 h at room temperature (buffer = 30 mM TRIS/HCl, pH 7.2, 140 mM Na cacodylate, 4 mM MgCl2, 0.1 mM dithiothreitol). Negative controls were processed without terminal transferase.

Acid phosphatase histochemistry

To demonstrate tartrate-resistant acid phosphatase (TRAP) activity in skeletal cells other than osteoclasts requires fixation by periodate-lysine-paraformaldehyde (PLP), which preserves the much weaker TRAP activities in chondrocytes and the matrix.(21) The PLP fixative consisted of 0.01 M Na-m-periodate, 0.075 M lysine, 2% paraformaldehyde, and 0.037 M phosphate buffer with a final pH of 6.2. Acid phosphatase activity was demonstrated at pH 5.0, using Naphthol AS-B1 phosphate and hexazotized pararosaniline (see Roach(21) for details).

Induction of apoptosis

Excised femoral heads from 4-day-old rabbits were placed into short-term organ culture. The explants were cultured at 37°C on metal supports at the interface between the serum-free culture medium and the gas phase, which was 5% CO2/95% air (details in Roach(22)). The tissues were precultured for 24 h to allow the explants to adapt to culture conditions, then the medium was replaced with culture medium containing 0.4 μM etoposide (Sigma Chemical Co., St. Louis, MO, U.S.A.) and the culture continued for 5 or 24 h. Control cultures did not contain etoposide. Etoposide is a DNA topoisomerase II inhibitor which has been used to induce apoptosis.(23) The explants were then processed for light and electron microscopy as above.

Incorporation of fixable fluorescent markers of live/dead cells

As an alternative method of distinguishing dying from viable cells, ethidium homodimer-1 (EH-1) and 5-chloromethylfluorescein diacetate (CMFDA; CellTracker green) were loaded into the explants by continuing the culture for 5 h in medium containing 25 μM of each probe. EH-1 is an intercalating DNA fluorescent dye which has a very low membrane permeability, i.e., the dye can only reach the nucleus if the cell membrane is no longer intact. EH-1 will therefore identify cells with a leaky or damaged membrane. CellTracker green freely diffuses into cells where its chloromethyl moieties react with intracellular thiols and their acetate groups are cleaved by cytoplasmic esterases, releasing green-fluorescent product.(24) CMFDA hence labels the cytoplasm of viable cells. These markers survive fixation and processing into paraffin wax.(25) Sections from tissues containing the fluorescent markers were viewed with a Leica TCS confocal microscope (Leica Microsystems Ltd., Milton Keynes, U.K.), using a krypton/argon laser.


Typical and “atypical” chondrocytes

As documented extensively in previous work (for review, see Hunziker and Herrmann(26)), the vast majority of chondrocytes were cells with relatively pale cytoplasm, sparse rough endoplasmic reticulum (rER), few mitochondria (m), an inconspicuous Golgi apparatus (go), and a generally pale nucleus with diffuse or flocculent chromatin (Fig. 1a). The mitochondria in these cells are generally small (m), but may also be swollen (sm). These young chondrocytes were surrounded by a pericellular rim which stained more intensely than the rest of the cartilage matrix (arrows in Fig. 1a) due to its increased concentration of proteoglycans.(26) However, a small percentage of chondrocytes (∼1%) had a radically different appearance and we initially termed these “atypical” chondrocytes (left-hand cell in Fig. 1b). There was no pericellular rim and the cells contained a strikingly dark, crenated nucleus with irregular small patches of condensed chromatin. The nucleus was surrounded by an electron-lucent material containing small, dark inclusions and sometimes secretory vesicles (sv) or Golgi apparatus, but otherwise cytoplasmic organelles were absent. Although the condensed chromatin suggested a connection with apoptosis, the morphology of the cytosol was unlike that of apoptotic, necrotic, or normal cells. Since there were no mitochondria to provide energy, no ribosomes to transcribe proteins, and since the nuclear DNA was fragmented, these cells were clearly no longer able to sustain life. Nevertheless, since cells with this morphology were the most frequently observed atypical chondrocyte, whereas the earlier stages (described below) were extremely rare, it may be inferred that such chondrocytes were stable for a reasonable time period. Thus these cells appeared to be “in limbo,” unable to live or die. Hence, we believe that the term “paralysis” would be an apt description of their state, implying immobility and loss of function, yet temporary stability.

Figure FIG. 1..

Typical and atypical epiphyseal chondrocytes. (a) In the typical epiphyseal chondrocyte, the usual cell organelles of rER, mitochondria (m) which may be swollen (sm), and Golgi apparatus (go) are present as well as a typical interphase nucleus. The cell is surrounded by a pericellular rim (short arrows). (b) A doublet of chondrocytes where the left cell is an atypical chondrocyte. Secretory vesicles (sv) can be recognized, otherwise all organelles have disappeared. Note the dark nucleus with a convoluted membrane. (a) Head of humerus from a newborn rabbit; (b) femoral head from a –1-day-old rabbit, located in the region where the secondary ossification center would have formed within 1 day. Bars = 1 μm.

Such unusual chondrocytes have occasionally been observed before, in the chondroepiphysis of fetal sheep,(27) human fetal cartilage,(28) and in cultured chondroepiphyses of young rabbits,(29) although the authors could not explain the unusual morphology. Another type of chondrocyte, which is probably related to the paralyzed cells, has been termed “dark chondrocyte.” These have been described in the mandibular condyle of mice,(30) the physes of growing swine,(31) and in the growth plate of chick embryos.(32) These dark chondrocytes had the same dark, crenated nucleus as found in the present study, but also had very dark cytoplasm with abundant rER, but without the excessive expansion of the ER lumen. The dark chondrocytes were found among mature and hypertrophic chondrocytes, whereas paralyzed chondrocytes were present among the immature epiphyseal chondrocytes, suggesting that the latter might represent a subtype of dark chondrocytes which is found mainly in immature cells.

Distribution of paralyzed chondrocytes

Within the chondroepiphysis (Fig. 2a), isolated paralyzed cells were found at random locations, always adjacent to normal light chondrocytes, and frequently forming a doublet either with another paralyzed cell (Figs. 2b and 2c) or with a normal chondrocytes (Fig. 1b; fat arrows in Fig. 2c), suggesting that both types of chondrocytes had originated from the same mother cell. There were, however, some regions in which there was a greater likelihood of finding paralyzed chondrocytes (see Figs. 2a2c). These were: around the secondary ossification center in the region where epiphyseal chondrocytes enlarge to become hypertrophic cells (transition zone, Fig. 2b); and in proximity to some cartilage canals (Fig. 2c). During development, defects in cells may occur at random, but there is an increased chance when cells divide or enter a new stage of differentiation, as is the case in the transition zone. Near invading cartilage canals, factors from the vascular system are thought to induce PCD.(29) The distribution of paralyzed cells was thus consistent with the view that they were damaged, dying cells.

Development of the cytoplasmic morphology of paralyzed chondrocytes

By studying cells of intermediate phenotypes, the unusual morphology of atypical cells and the possible sequence of cellular events involved in cell paralysis could be understood.

Swollen mitochondria

The mitochondria of atypical chondrocytes were swollen (sm in Fig. 3b), suggesting disruption of the inner mitochondrial membrane. Mitochondrial swelling is a characteristic feature of cell necrosis, but recent evidence suggests that disruption of mitochondrial function might also be an essential event that commits cells to apoptosis.(33–35) Loss of mitochondrial function thus seems to be an early event during necrosis, apoptosis, and cell paralysis.

Figure FIG. 3..

Early stages in the expansion of the rER lumen. (a) Focal distension of the rER lumen surrounds organelles, such as mitochondria (arrows). (b) With further distension, regions of cytoplasm are sequestered and made into islands within a lake of ER lumen. (c) An enlargement (×70,000) of the area indicated by the square in b, illustrating the rearrangements that lead to the isolation of cytoplasmic islands (large open arrow). From the head of humerus from a newborn rabbit, bars = 1 μm.

Focal distensions of the rER lumen

In most chondrocytes, some focal distensions of the lumen of the rER were present (Fig. 3a). These might have resulted from either increased synthesis or increased retention of proteins within the lumen. The ER acts as an architectural editor and retains proteins that are abnormal in structure or assembly.(36,37) In normal chondrocytes, these focal distensions were probably reversible, but in atypical chondrocytes the cisternae of the ER continued to swell and to surround organelles, such as mitochondria or lysosomes (arrows in Fig. 3a). With continued expansion, parts of the cytoplasm became isolated by a rearrangement of the reticular membranes. The sequence of this process is illustrated in the region marked by a square in Fig. 3b and seen enlarged in Fig. 3c. Here a finger-like projection of cytoplasm was almost completely surrounded by rER lumen. There was a break in the single rER membrane which appeared to reform to enclose the cytoplasmic projection (open arrow). The end result was that parts of the cytosol now appeared as islands, bordered by ribosome-studded ER membrane (small arrows in Fig. 3c) within a “lake” of rER lumen.

Excessive expansion of rER lumen

In paralyzed cells, the rER lumen replaced more and more of the cytosolic space (∼30% in Fig. 4a) until virtually the whole nonnuclear space of the cell was taken up by rER lumen (∼90% in Fig. 4b). Some cytoplasmic ground substance remained as a thin layer around the perimeter of this lumen (short arrows in Figs. 4b and 5a) and as islands within the lake of rER lumen. Now the lumen appeared more electron lucent (compare the cells of Fig. 3 with those of Fig. 5), suggesting a dilution of the luminar content by water uptake, perhaps driven by osmosis. Ribosomes were still visible, either attached to the apparently inverted rER membrane or as free ribosomes (Fig. 4c). Swollen mitochondria (sm) were surrounded individually by the expanding ER lake. Therefore, what appeared to be cytoplasm in a paralyzed cell was in fact the expanded lumen of ER. To the authors' knowledge, dilation of the ER to the extent that the majority of the nonnuclear space is occupied has not been observed before. In genetic abnormalities of cartilage proteins, dilated ER cisternae were present,(37,38) but never to the extent seen in atypical cells. Dilation of the ER is also a common feature of necrotic cells,(39) but never to the same extent as in paralyzed cells.

Figure FIG. 4..

Excessive expansion of ER lumen. The contents of the ER lumen is now less electron dense than in the cells of Fig. 2, suggesting that a dilution had taken place. More and more space is taken up by the ER lumen, about 50% of the cytoplasm in (a) and nearly 90% in (b). (a) Two large ER lakes have formed, but the cytoplasm and Golgi apparatus (go) can still be recognized. At the left-hand side (triangle) is part of a cell that has already reached the paralyzed state. (b) Nearly the whole nonnuclear space is taken up by the ER lake, a thin layer of cytoplasm still remains around the ER lake (small, fat arrows). Many vacuoles (v) are blebbing and releasing their contents into the extracellular space. The Golgi apparatus (go) and swollen mitochondria (sm) can still be recognized. The pericellular rim is no longer present. At the bottom (triangle) is part of a cell that has already reached the paralyzed state. (c) Higher magnification (×20,000) of the central part of (b), showing the presence of polyribosomes along the border of the cytoplasmic islands and free in the cytoplasm. (a) Head of humerus from a newborn rabbit; (b) and (c) femoral head of a −1-day-old rabbit. Bars = 1 μm.

Figure FIG. 5..

Digestion of cytoplasmic islands and nuclear condensation. (a) Most organelles have disappeared; the content of many cytoplasmic islands has been degraded, but a thin band of cytoplasm is present at the perimeter of the cell (arrows); the Golgi apparatus (go) has become more developed; the pericellular rim is no longer present. (b, c) Chondrocytes that have reached the paralyzed state: the cytoplasmic islands have shrunk to electron-dense, worm-like inclusions within an enormous lake of ER lumen; the Golgi apparatus (go) has hypertrophied; the nuclear chromatin has condensed into random patches; the nuclear membrane is convoluted; the cell membrane is no longer intact and extensive blebbing of vacuoles (v) is taking place. (a) and (c) From the head of humerus from a newborn rabbit; (b) from the femoral head of a 4-day-old rabbit. Bars = 1 μm.

Digestion of cytoplasmic components

In paralyzed chondrocytes, the cytoplasmic islands had coalesced into worm-like electron-dense masses (arrows in Fig. 5c), suggesting that the content had been degraded. This raises the question of how the intracellular components might be digested to form the dark inclusions seen in the final-stage paralyzed cell. Normally, intracellular components are turned over or eliminated by the process of autophagy (see review in Glaumann et al.(40)) in which small regions of cytoplasm and/or organelles are sequestered and degraded within autophagic vesicles containing lysosomal enzymes.(41,42) The sequestering of cytoplasmic components by ER membranes was obviously similar to autophagy in that a digestive lumen was created. However, in atypical chondrocytes, the membranes were always single and ribosomes remained attached to the membranes, whereas in autophagy the membranes are double and were stripped off their ribosomes. This suggested that different processes were involved in the digestion. It is possible that lytic enzymes, which are normally present in the cytoplasm in an inactive form, were activated. During the PCD of the salivary gland cells of blowfly, free acid phosphatase, normally a lysosomal enzyme, accumulated around ribosomes prior to cell autolysis.(43) Low activity of acid phosphatase could also be demonstrated in epiphyseal chondrocytes (Fig. 6a), but was present in both normal and atypical ones. It is possible that the of pH 5.0, which is the optimum for acid phosphatase activity and which was artificially supplied during the histochemical reaction, was not actually present in all chondrocytes in vivo. Since cell acidification was an essential feature of PCD in many situations,(44) acidic conditions might have been present in the digestive lumens in atypical chondrocytes, leading to the activation of acid phosphatase.

Figure FIG. 6..

Features of paralyzed cells. (a) Acid phosphatase histochemistry in the spherical growth plate around the ossification center. Osteoclasts (bottom left corner) are strongly positive, as expected, and nearly all epiphyseal chondrocytes contain acid phosphatase. Activity is less in hypertrophic chondrocytes (H) which instead are positive for alkaline phosphatase (data not shown). (b) Overview of the edge of the ossification center in the femoral head of a 4-day-old rabbit, cultured for 5 h with etopside and loaded with EH-1 (red) and CMFDA (green). The red fluorescence is only present in cells whose cell membrane is no longer intact. A cartilage canal is present in the lower right corner, and the secondary ossification center is in the middle. There is nonspecific green autofluorescence associated with the cartilage matrix, especially the mineralized part in the center. EH-1–labeled cells are present in the transition zone which surrounds the ossification center, along the edge of the cartilage canal, and within the central hypertrophic region. The green fluorescence in viable chondrocytes is weak, but detectable at higher magnification. (c, d, and e) TUNEL staining of sections from the same block as (b), but about 50 μm away from the view seen in (b). Parallel sections of negative control (c), TUNEL method (d), and positive control (e). The dashed lines indicate the approximate start of the hypertrophic zone (H). No TUNEL staining was found in the regions where EH-1–labeled cells were present, but a few hypertrophic cells (small arrows) and a small cluster of transitional cells near a cartilage canal (large, open arrow) showed positive TUNEL staining. (f) Positive TUNEL staining in transitional chondrocytes after 24 h with etoposide. The patches of diffuse staining which is not associated with the cells, are artefacts. cc, cartilage canal; H, hypertrophic region. Bars = 10 μm in (a) and (f) and 50 μm in (b–e).

Hypertrophy of Golgi apparatus

The Golgi apparatus in normal chondrocytes was generally not well developed (Fig. 1a), but an enlarged Golgi became a prominent feature in paralyzed cells (Figs. 4a and 4b; Figs. 5a5c). In fact, when other cellular organelles were no longer recognizable, the Golgi apparatus with its associated secretory vesicles continued to hypertrophy excessively (Figs. 5b and 5c). Such a large Golgi is generally only a feature of specialized cells that elaborate large quantities of secretory product, such as intestinal epithelial cells or pancreatic acinar cells.(45) In atypical chondrocytes, the hypertrophied Golgi may be a consequence of the enlarged ER lumen, since secretory products are normally passed from the ER lumen to the Golgi.(45) The nature of these products is unknown.

Cell membrane changes, vacuolation, and disappearance of the pericellular rim

The more intensely stained pericellular rim, which still surrounded the cell shown in Fig. 4a, was no longer present around the cells seen in Figs. 4b, 5b, and 5c. Instead, numerous vacuoles (v) were blebbing from the cells, suggesting release or spillage of cellular material from the cell. The cell membrane was no longer intact, as indicated by open vacuoles (fat arrows in Fig. 5b) and release of digested cytoplasmic islands (arrows in Figs. 5b and 5c). Both blebbing and membrane damage occur early in pathological cell death. In apoptotic cells, the integrity of the cell membrane remains intact until the late stages, although “flip-flop” of some membrane phospholipids, such as phosphatidylserine, accompanies apoptosis.(46) Even apoptotic bodies were still bound by an intact membrane.(1,5) In atypical chondrocytes, membrane damage occurred at an intermediate stage in the process, after excessive dilation of the ER. This membrane damage together with the blebbing of vacuoles and digested cytoplasmic remnants probably represented the irreversible commitment to death.

Changes in nuclear morphology

While these cytoplasmic changes were taking place, the morphology of the nucleus remained normal (Figs. 3 and 4). As long as ribosomes were still visible around these islands, the morphology of the nucleus did not change significantly (Fig. 5a). Only when the majority of the cytoplasm had been digested and cell membrane damage was apparent did the nuclear chromatin condense (compare Fig. 5a with Figs. 5b and 5c). This contrasts the sequence of events in classical apoptosis, where the chromatin condensations in the nucleus are a relatively early event.(6)

The pattern of chromatin condensations in paralyzed chondrocytes (small clumps, randomly distributed throughout the nucleus) also contrasted the pattern of chromatin condensation in apoptotic or necrotic cells. In apoptosis, the nuclear chromatin contracts into round or crescent-shaped uniformly dense masses which are aligned with the nuclear envelope.(5) In necrosis, the whole nucleus becomes dense or pyknotic, but generally remains round, whereas in paralyzed chondrocytes a highly convoluted, crenated nuclear membrane was always present.

Final fate of paralyzed cells

The final fate might be illustrated by one of the cells in Fig. 7a. At the bottom of this micrograph are two paralyzed cells, but the top cell consists of remnants of nucleus and cytoplasm only. This morphology is consistent with extrusion of the watery contents from the expanded ER lumen of a paralyzed chondrocyte together with further collapse of the nucleus. Such inert cell debris could remain in the cartilage until such time when the cartilage containing the debris is resorbed during the development of the epiphysis. Notably, the former lacuna had filled in with cartilaginous matrix, although the mechanism of this process is unknown.

Figure FIG. 7..

Final fate of “paralyzed” cells compared with classical apoptosis. (a) The cell at the top (arrow) consists of shrunken cellular remnants, while the two chondrocytes are still in a paralyzed state. (b) Two apoptotic bodies containing uniformly dark chromatin, condensed into circular masses, while cellular organelles are generally intact. (a) From the head of humerus from a −1-day-old rabbit; (b) from the femoral head of a 4-day-old rabbit, cultured for48 h. Bars = 1 μm.

Classical apoptosis in epiphyseal chondrocytes

Despite searching many ultramicrographs for chondrocytes with classical apoptotic morphology, we have so far only found one example in an epiphyseal explant which had been cultured for 2 days (Fig. 7b). Here the chondrocyte had broken up into two apoptotic bodies, each containing the classical rounded, uniformly dark condensed chromatin and intact organelles. Clearly chondrocytes are capable of undergoing classical apoptosis. However, it is a rare phenomenon, so far only identified in culture, and apoptosis does not appear to be the usual mode of PCD in chondrocytes in vivo.

Distribution of paralyzed chondrocytes after induction of apoptosis with etoposide

Since paralyzed cells appeared to undergo a PCD, we reasoned that apoptosis-inducing agents, such as etoposide, should increase the number of atypical cells and hence facilitate their characterization. Indeed, an increased number of paralyzed cells were found in EM sections of epiphyses cultured with etoposide, together with some necrotic cells. Since membrane damage was a feature of final-stage paralyzed cells, cultures were loaded with EH-1. After 5 h with etoposide, EH-1–labeled chondrocytes were present at random locations throughout the chondroepiphysis and were concentrated in the transition zone of the secondary ossification center, close to cartilage canals and within the hypertrophic region. (Fig. 6b). The latter region is an area where dying cells would be expected since hypertrophic chondrocytes are thought to undergo apoptosis,(11,13,18) while the first two regions corresponded exactly to those sites where the incidence of paralyzed chondrocytes had been higher in vivo (Fig. 2). Etoposide thus seemed to precipitate cell death mainly in those chondrocytes that were already primed for the process.

To determine whether the nuclei of EH-1–labeled chondrocytes contained DNA breaks, sections from the same block were stained with the TUNEL method. Because the TUNEL technique can give ambiguous results, negative and positive controls were done on parallel sections (Figs. 6c6e). There was no staining in negative controls (Fig. 6c). With the TUNEL method (Fig. 6d), no chondrocytes were stained in the transition zone, except for a small group of cells at a location where transitional chondrocytes were close to a cartilage canal (large open arrow). The occasional hypertrophic cells (small arrows) were also stained. The lack of TUNEL staining was not due to inaccessibility of DNA breaks to the terminal transferase, because in the positive control every cell was stained (Fig. 6e). After 24 h with etoposide, more chondrocytes were stained with the TUNEL method (Fig. 6f), and this time TUNEL-positive cells were present in the transition zone as well as around cartilage canals (cc) and in the hypertrophic zone (H), i.e., the zones where EH-1 positive cells had been present after 5 h. This suggested that cell membrane damage was present before nuclear fragmentation. The light microscopic data thus supported the morphological evidence obtained with the electron microscope.

Comparison of cell paralysis with other modes of cell death

In Table 1, the cellular events of cell paralysis are compared with those of apoptosis, autophagocytosis, and necrosis. One early event in all cell deaths was the disruption of mitochondrial membrane potential. The resulting permeability transition would cause release of stored Ca2+, reactive oxygen species,(47) and cytochrome c.(48) In addition, loss of oxidative phosphorylation would lead to loss of ATP. The differences between the modes of cell death are probably related to which aspect of mitochondrial damage affects the cell first. In accidental cell death, there is an immediate loss of energy that prevents cell membrane pumps from maintaining intracellular solute concentrations, resulting in an influx of water and extracellular calcium. Release of calcium and free radicals from mitochondrial stores would merely exacerbate the situation. The morphologic manifestations of these events are cytoplasmic swelling and blebbing. During apoptosis, the loss of ATP is more gradual, and it is the cytochrome c efflux which is the most important. This activates the caspases, the main executors of apoptosis,(49) which kill the cell before any other aspect of mitochondrial damage can take effect.(48,50) During the process leading to cell paralysis, protein synthesis was initially maintained and maybe even increased, as indicated by the presence of intact ribosomes and increased ER lumen. Since protein synthesis requires ATP, loss of ATP during paralysis was presumably more gradual.

Table Table 1.. Comparison of Cell Paralysis with Other Modes of Cell Death
original image

Nuclear changes were secondary to cytoplasmic changes in cell paralysis, autophagocytosis, and necrosis. Recent evidence also suggests that DNA fragmentation during apoptosis may also be a consequence rather than the cause, since e-nucleated cells can die with the same kinetics and cytoplasmic changes as their nucleated parent cells.(51) This suggests that all cell deaths might be primarily cytoplasmic rather than nuclear processes.(52) The availability of ATP might be the crucial factor determining the type of nuclear fragmentation observed in the various modes of cell death. ATP was necessary for the movement of chromatin to the nuclear periphery, but not for chromatin cleavage.(53) During apoptosis, there seemed to be just enough ATP available for chromatin margination, whereas in necrosis and in paralyzed chondrocytes, where the nuclear changes took place after mitochondria had disappeared, no ATP would be available. This may explain why the clumped chromatin was distributed throughout the nucleus rather than marginated at the inner nuclear membrane.

Hypothesis: Cell paralysis, another mode of physiological death?

Two main criteria distinguish physiological from pathological cell death: the former occurs during normal development or maintenance (and may be induced by subnecrotic pathological factors), while the latter is always pathological, resulting from injury, such as high toxicity or ischaemia; and the former is orderly with no inflammatory reaction, the latter is messy and inflammatory cytokines are released. By both criteria, cell paralysis is clearly a physiological type of cell death. Whether the process is also “programmed” would depend on demonstrating that the death is executed according to an intracellular program. We hypothesize that the hallmark of cell paralysis, i.e., the digestion of cellular organelles and cytoplasm by forming cytoplasmic islands within a lake of expanded ER lumen, is an orderly process, executed according to some program. It seems as if the normal cellular process of retaining proteins within the ER lumen is increased to such an extent that it serves a different purpose, just as the normal process of autophagy is increased in autophagocytosis to assist in the cell death process. Since the initial and subsequent morphological changes associated with cell paralysis are markedly different from apoptosis, the process cannot be interpreted as an abortive or aberrant form of apoptosis. Cell paralysis might therefore be regarded as an intermediate state in the PCD of epiphyseal chondrocytes.

Such an intermediate paralyzed state would be advantageous in situations where the elimination of apoptotic bodies by phagocytosis represents a problem. Epiphyseal cells are isolated within their lacuna which would make phagocytosis by neighboring cells difficult (although not impossible, since we have observed one instance of a chick chondrocyte sending a pseudopod into a neighboring lacuna to engulf an apoptotic body(54)). If apoptotic bodies are not phagocytosed quickly, they undergo secondary necrosis(6) with all the disadvantages of uncontrolled release of lysosomal enzymes. Hence, a state of paralysis, in which cytoplasmic organelles (including lysosomes) have been digested, could inactivate those chondrocytes that are defective or superfluous. Eventually, the cartilage tissue containing these cells will be resorbed as a consequence of growth mechanisms. Cell paralysis as an intermediate stage might represent one solution to the problem of inactivating or neutralizing defective cells in situation where heterophagocytosis is difficult.


The authors thank Mrs. Nadine Hill (University Orthopaedics), and Mr. Nick Barnett and Miss Sue Cox (Biomedical Imaging Unit) for their highly skilled technical assistance. The work was supported by the Oliver Bird Fund (Nuffield Foundation) and the Wishbone Trust.