T he F ate of the T erminally D ifferentiated C hondrocyte: E vidence for M icroenvironmental R egulation of C hondrocyte A poptosis

Introduction

Cartilage, a product of the developing mesenchyme, serves several critical functions. In some fish (elasmobranchii), cartilage serves to support the major organs of the body. In primates, this function is minimal and is confined to the cartilages of the ear, nose, bronchi, larynx, and trachea. Separate but relevant functions relate to locomotion and bone growth. Long bone growth is mediated by the activities of cells contained within a strip of specialized cartilage, the endochondral growth plate. To achieve growth, endochondral chondrocytes proliferate and become hypertrophied. They secrete a matrix that undergoes mineralization; subsequently, the matrix is removed by resorbing cells and the potential space is filled by bone. The mature chondrocytes are then deleted from the new bone by a process that is partially or completely dependent on the induction of apoptosis. Without a deletion process that is equal in activity to the rate of chondrocyte proliferation and maturation, the plate would increase in volume and transform into a composite chondroid bone that would exhibit poor mechanical, physical, and structural properties.

In this critical review, considerable attention is directed at the chondrocyte deletion process in the growth plate, especially the induction of apoptosis. This review includes consideration of possible environmental apoptogens that may cause chondrocyte deletion as well as the pathways utilized by these agents to trigger the apoptotic process in terminally differentiated chondrocytes. We present the hypothesis that specific microenvironmental factors in cartilage, ions derived from the apatite lattice and fragmented peptides of the organic phase of cartilage, may direct chondrocyte apoptosis.

The Fate of the Hypertrophic Chondrocyte

The primary growth plate in a mammalian long bone is a highly organized tissue; as such, consideration of its architecture provides an interesting insight into its function. The plate itself extends from the cartilage of the secondary ossification center into the bone of the metaphysis. In a histological section of the plate, columns of cells, separated by connective tissue septa, are seen organized along the bone’s longitudinal axis. Within the cartilage, four distinct zones can be delineated (Fig. 1). The most superficial region, closest to the joint surface, is the reserve or resting cell zone. In this region, spherical chondroprogenitor cells, embedded in an extensive extracellular matrix, are present. Below the reserve cell zone is the proliferative layer. Here, the chondroprogenitors generate columns of chondrocytes (5-8 cells deep in mammalian plates; 15-30 deep in avian cartilage). These cells divide in the long axis of the plate so that flattened mother and daughter cells appear to lie on top and slightly to the side of each other. The shape of the cells suggests that they are under axial loading. After a limited number of rounds of proliferation, the cells change their shape and phenotype. These maturing chondrocytes become swollen and express type X collagen as well as type II collagen; in addition, they exhibit high alkaline phosphatase activity. From a functional viewpoint, these hypertrophic chondrocytes secrete membrane-limited vesicles (matrix vesicles) that serve to initiate mineralization of the extracellular matrix. As the cells become hypertrophic, they increase their volume five- to 12-fold and elevate three-fold the total amount of territorial extracellular matrix (Hunziker, 1994). When terminally differentiated, the increase in cell volume and the presence of a bulky extracellular matrix provide much of the space required for bone elongation. At the calcified end of the growth plate, trabeculae of woven bone are deposited on the calcified cartilage septa. It is in this architectural context that the question is raised: What is the fate of the terminally differentiated hypertrophic chondrocyte?

The intimate structural relationship among chondrocytes, osteoblasts, and the invading vascular elements at the chondro-osseous junction has confused rather than clarified arguments concerning the fate of the terminally differentiated chondrocyte. Many workers infer that, since cartilage and bone are variants of the same tissue type, the cells can switch identities. Two factors serve to reinforce this view. First, morphological investigations revealed that osteoblasts often occupy chondrocyte lacunae (Shapiro and Boyde, 1987); second, in vitro studies indicated that progenitor cells can commit to either the osteo- or chondrogenic pathways (Descalzi-Cancedda et al., 1992; Gentili et al., 1993). For these reasons, some workers have assumed that cells of one lineage, chondrocytes, are plastic and can convert to the related osteoblastic lineage: this conversion is called metaplasia or transdifferentiation (Haines and Mohuiddin, 1968).

Views Concerning Chondrocyte-to-Osteoblast Conversion in the Growth Plate

Claims that epiphyseal growth plate chondrocytes transdifferentiate to endochondral osteoblasts are found in the literature going as far back as the 19th century (Van der Stricht, 1890; Brachet, 1893). Much of the critical evidence in support of this idea, however, is circumstantial. It is based on microscopic examination of chondrocyte and osteoblast populations at the chondro-osseous junction. Indeed, Erenpreisa and Roach (1996) described a transition from the chondrocytic to the osteoblastic phenotype that involved cycles of asymmetric (differential) division in sections of hypertrophic cartilage. Lending some support to the hypothesis that chondrocytes did not die but assumed a new (osteoblastic) phenotype was the observation that terminally differentiated cells were biosynthetically active. Hypertrophic chondrocytes have been reported to continue to synthesize type I collagen (Von der Mark and Von der Mark, 1977; Horton et al., 1983; Yasui et al., 1984), glycosaminoglycans (Silbermann and Frommer, 1972, 1974), and glycolytic and oxidative enzymes and phosphatases (Kuhlman and McNamee, 1970). Using in situ hybridization techniques, Oshima et al. (1989) showed that terminally differentiated hypertrophic chondrocytes expressed type II collagen, osteopontin, osteocalcin, osteonectin, and aggrecan core-binding protein mRNA. The continued expression of these extracellular matrix and house-keeping genes lends support to the view that hypertrophic chondrocytes remain active during late stages of the maturation cycle. However, these observations do not exclude the possibility that chondrocytes are commencing a death program, even when undergoing maturational changes leading to terminal differentiation.

The most direct support for the transdifferentiation hypothesis came from experiments performed with organ and cell culture models. In 1933, Dame Honor Fell observed that cartilage in culture became converted into bone and osteoid (Fell, 1933). Holtrop (1966) demonstrated that when costochondral rib was placed into organ culture and care was taken to remove other contaminating tissues, bone was formed. Attempts to repeat these experiments have met with mixed results, both positive (Holtrop, 1972) and negative (Bentley and Greer, 1970). A more rigid test of the transdifferentiation hypothesis was performed by Kahn and Simmons (1977), who transplanted growth plates from a quail into the chorioallantoic membrane of a chick. The resultant bone was entirely of quail origin, suggesting that the osteoblasts were derived from transplanted cartilage cells. Several investigators described evidence of transdifferentiation in Meckel’s cartilage, chondrosarcomas, and epiphyseal cartilage (Lufti, 1971; Silbermann and Frommer, 1972; Aigner et al., 2000). However, both electron and light microscopic examination of the chondro-osseous junction yielded surprisingly inconsistent results. Yoshioka and Yagi (1988) reported that hypertrophic chondrocytes maintained an intact plasma membrane up to the point at which there was vascular invasion. In perichondrium-stripped developing metatarsals, hypertrophic chondrocytes de-differentiated into stromal cells (Thesingh et al., 1991); a study of the fate of chondrocytes in antlers showed that while osteoblasts and chondrocytes occupied the same lacunae, there was no evidence of transdifferentiation (Szuwart et al., 1998).

Before leaving this topic, we believe that it is important to comment on a very thoughtful study by Bianco and his co-workers (1998). These workers used a cell culture system to assess possible fates of the hypertrophic chondrocyte. They noted that chondrocytes in the "right microenvironment… switch from the synthesis of cartilage..and….organize a mineralizing bone matrix." Bianco et al. (1998) proposed that "all early hypertrophic chondrocytes have the inherent potential to differentiate to osteoblast-like cells and to contribute to initial bone formation." According to these investigators, only those chondrocytes that face the peripheral perichondral tissues are exposed to a microenvironment that potentiates osteogenesis. Hypertrophic chondrocytes located in the central portion of the cartilage are not exposed to the same microenvironment and therefore undergo apoptosis. If this view is correct, it would serve to bring together some of the conflicting ideas concerning chondrocyte-osteoblast interaction while at the same time strengthening the view that the local environment plays a critical role in determining cell fate.

It should be noted that the transdifferentiation hypothesis replaced an earlier notion that, in the calcified region of the growth plate, removed from vascular channels, cells are deprived of nutrients and O₂ and have no mechanism for removal of waste products. In this hostile environment, chondrocytes "starve" to death (Ham, 1952). Although, at that time, few comments were made concerning this form of death, it was understood that death was through ischemia, and that the death mechanism was closely related to necrosis. New understandings concerning the O₂ needs of chondrocytes and the fenestrated structure of the lacunae and the extracellular matrix provide no evidential support for this theory. However, within the past 15 years, in parallel with the growth in comprehension of the biology of cell death, an alternative fate for the terminally differentiated cell has been voiced. It has been proposed that epiphyseal chondrocytes end their life cycle by the induction of apoptosis, or programmed cell death.

Evidence of Epiphyseal Chondrocyte Apoptosis

Prior to reviewing evidence linking apoptosis with deletion of chondrocytes from the growth plate, it is important to recognize the importance and scope of the apoptotic process. Basically, apoptosis, or programmed cell death, is an evolutionary conserved blueprint that serves myriad biological functions, all of which are designed to remove unwanted or damaged cells from multicellular organisms. Many workers have emphasized that the control of cell death is as much a part of embryonic development as those events that control proliferation and differentiation (Ishizaki et al., 1994, 1995). Not surprisingly, apoptosis is active in organogenesis and morphogenesis; in the juvenile, maturing, and adult states, apoptosis is active in maintaining tissue size, shape, and function. The apoptotic process permits sculpting and establishment of the mature form and phenotype by the selective removal of cells from a tissue. Apoptotic activity is evident in palate formation, suture closure, removal of webbing from digits, and in the control of enamel formation in the tooth. When the apoptotic process is dysregulated, disease, malformations, and even organism death are evident. In terms of clinical disease, there is little doubt that alterations in the cell death process are linked to the pathogenesis of cancer, AIDS, and Alzheimer’s disease (Mattson, 2000; Roshal et al., 2001; Zornig et al., 2001).

While a considerable number of intracellular proteins have been isolated that participate in the cell killing process, a transduction pathway is discernible. The stimulus or the ligand for apoptosis is frequently a membrane receptor, the death receptor (Fig. 2). Currently, six different death receptors are known, including tumor necrosis factor (TNF) receptor-1, CD95 (Fas/APO-1), TNF receptor-related apoptosis-mediating protein (TRAMP), TNF-related apoptosis-inducing ligand (TRAIL) receptor-1 and -2, and death receptor-6 (DR6) (Pan et al., 1998; Petak and Houghton, 2001). The signaling pathway by which a ligand induces apoptosis requires that there is oligomerization of receptors by the death ligand, recruitment of an adapter protein, and the subsequent activation of an inducer caspase [caspases are a group of 14 cysteine proteases that recognize tetrapeptide motifs and cleave their substrates on the COO end of an aspartyl residue]. Caspase-8 is known to propagate the apoptotic signal, either by directly cleaving and activating downstream caspases, or by releasing the BH3 Bcl2-inter-acting protein, which serves as a molecular regulator of mitochondrial function.

Removal of Bcl2, or its replacement by pro-apoptotic factors that localize to the mitochondrion, causes apoptosis by releasing protein components from the organelle that include cytochrome c, APAF-1, and AIF. The proteins associate to form an apoptosome which then recruits, binds, and activates pro-caspase 9, an enzyme that regulates the induction of apoptosis (Jiang and Wang, 2000). Caspase-9 then recruits and activates the executioner enzyme, pro-caspase-3. Cells lacking either APAF-1 or caspase-9 are resistant to the induction of apoptosis via the mitochondrial pathway. Caspase activity is modulated by another set of proteins, the IAPs (inhibitor of apoptosis proteins). These proteins bind to and inhibit caspase-9 and thereby block the apoptotic process (Deveraux et al., 2000). Aside from cytochrome c, APAF-1, and AIF, mitochondria release yet another protein, Smac/Diablo. This protein promotes caspase activation by associating with the apoptosome and inhibiting IAP activity (Du et al., 2000; Srinivasula et al., 2001). In terms of downstream effectors, caspase-3 can remove a negative regulatory domain from a protein kinase and trigger membrane blebbing (Rudel and Bokoch, 1997). Caspase activation executes cell death by cleaving and inactivating vital cellular proteins such as DNA repair enzymes, lamin, and gelsolin. Within the nucleus, caspase-activated DNAases (CAD) fragment DNA. This enzyme is normally inactivated by binding to an inhibitor, iCAD. During apoptosis, iCAD is activated by caspases leading to the release of the active endonuclease. It is this enzyme that hydrolyzes linker DNA producing the characteristic fragmentation products (see below).

Returning to the growth plate itself, early work provides direct information relevant to the induction of the apoptotic process in chondrocytes. For example, in 1971, Lufti made an interesting observation: "prior to their liberation the cartilage cells apparently undergo some form of metamorphosis, for it was noted that whereas the nuclei of the cells in the basal part of the hypertrophic region were relatively large and pale, they became smaller and more darkly staining as the marrow cavity was approached." These distinct morphological changes in the microanatomy of the hypertrophic chondrocyte are indicative of cell death. However, hypertrophic cartilage is notoriously difficult to examine histologically. Hunziker and co-workers (1983) elegantly demonstrated that, without specialized fixation techniques, artifactual changes were apparent in cells of the hypertrophic and calcifying zones of the growth plate. When growth plate sections were optimally fixed, a morphology consistent with the activation of apoptosis was identified in terminally differentiated chondrocytes in the developing growth plate (Farnum, 1987, 1989a,b, 1990; Farnum and Wilsman, 1987, 1989a,b; Farnum et al.,. 1990; Lewinson and Silbermann, 1992; Zenmyo et al., 1996; Aizawa et al., 1997; Ohyama et al., 1997; Roach, 1997; Rath et al., 1998; Silvestrini et al., 1998, 2000; Suda et al., 1999).

Ultramicroscopic analyses of optimally fixed sections suggest that the chondrocyte death process may be unlike that seen in other tissues (Erenpreisa and Roach, 1996, 1998). The latter investigators described chondrocytes that exhibited unusual, albeit characteristic, features: While condensed chromatin was present, suggestive of apoptosis, the "morphology of the cytosol" was unlike that of necrotic, apoptotic, or normal cells. It was suggested that these cells were "in limbo, unable to live or die". The term paralysis was used to describe this state, implying "immobility and loss of function yet temporarily stable". In a study of cell death in the nematode, Hoeppner et al. (2001) suggested that a cell might survive even when death signals were activated. They hypothesized that low caspase levels activate "eat me signals", but that there is defective co-ordinated execution of subprograms and cell removal is impaired (Hoeppner et al., 2001). Loss of these molecular sub-routines could explain why some chondrocytes adopt this paralytic state. Nevertheless, there is good reason to believe that the terminally differentiated cells undergo apoptosis. Use of the transferase-mediated UDP nick end-labeling (TUNEL) procedure to stain fragmented DNA revealed that many of the terminally differentiated hypertrophic cells are end-labeled (Hatori et al., 1995; Bronckers et al., 1996; Zenmyo et al., 1996). [A critical event in apoptosis is activation of CAD and fragmentation of linker DNA; the TUNEL stain is picked up by the cut ends of the DNA.] Hatori et al. (1995) substantiated that cells were apoptotic by electrophoretically identifying fragmented DNA in chondrocyte extracts.

To summarize: Studies performed by several investigators lend strong support to the notion that terminally differentiated chondrocytes exhibit morphological characteristics of apoptosis; TUNEL staining and molecular analysis confirmed that chondrocyte nuclei present fragmented DNA, a cardinal sign of the induction of apoptosis. It is also clear that from a morphological viewpoint, the dying cells display several atypical morphological characteristics. The import of these characteristics is not as yet understood.

Microenvironmental Factors, Chondrocyte Maturation, and the Induction of Apoptosis

Based on what is known of the chemistry and structure of the growth plate, the molecular composition of the microenvironment around chondrocytes, depends on the maturational stage. Thus, the hypertrophic chondrocyte resided briefly in a matrix in which the local microenvironment is influenced by the presence of type X collagen molecules and apatite. In contrast, the microenvironment around proliferating chondrocytes reflects the presence of cationic glycan molecules and type II collagen. It is also likely that the microenvironment at the tissue periphery is different from that experienced by chondrocytes residing in the core of the epiphyseal growth plate. We contend that as a chondrocyte becomes terminally differentiated, local factors in the cartilage matrix initiate the death process. Related to this maturation process, in vitro studies provide strong support for the idea that terminally differentiated cells develop an increased sensitivity to apoptogens. For example, when immature sternal chondrocytes were challenged with an apoptogen, they were far more resistant to the induction of apoptosis than were hypertrophic cells (Gibson et al., 2002). In a very recent study of the maturation process, Adams and co-workers (2002) showed that maturation of tibial chondrocytes with retinoic acid did not in itself induce apoptosis. Instead, in the terminally differentiated state, the chondrocytes were very sensitive to the presence of low concentrations of a matrix apoptogen.

Since chondrocytes are sensitive to factors present in the local microenvironment, it begs the question: What agents exist in the immediate vicinity of the terminally differentiated cell that activate apoptosis? Three different types of molecules come to mind: components of the degraded extracellular matrix, growth factors and cytokines, and the local pO₂. Considering the O₂ concentration, information from several different sources points to the importance of this molecule in regulating apoptosis, especially in tumor cells. However, it is doubtful if the local pO₂ activates chondrocyte death. The anatomy of the plate indicates that while cartilage cells occupy a poorly vascularized milieu, once the proliferative stage of the maturation cycle is complete, the cells are almost completely glycolytic. When this occurs, the cells develop a minimal need for O₂ (Rajpurohit et al., 1996). Clearly, chondrocytes are well-adapted to their environment, and for this reason, it is concluded that a local decrease in pO₂ is unlikely to cause cell death.

Turning next to growth factors, it is clear that these agents are required for chondrocyte survival. Indeed, serum withdrawal (absence of growth factors) caused apoptosis of terminally differentiated cells (Feng et al., 1998). Related to this observation, Gibson et al. (2002) clearly demonstrated that the mature chondrocyte was more sensitive to serum factors and apoptogens then were immature cells. Interestingly, the release and activation of transforming growth factor-beta 2 (TGF-β₂) are features of chondrocyte hypertrophy; TGF-β₂ administration causes an increase in intracellular caspase expression. For this reason, Gibson et al. (2002) suggested that activation of this cytokine is linked to the induction of apoptosis. In light of what is known of the importance of the class of molecule in regulating cell function, this critical observation begs the following questions: First, how are these molecules and their cognate receptors expressed in the growth plate? Second, what is the mechanism by which these bioactive agents regulate chondrocyte apoptosis and survival?

The last group of local environmental factors that can activate chondrocyte apoptosis is derived from components of the calcified cartilage matrix. During the replacement of calcified cartilage by bone, these matrix molecules could accumulate in the microenvironment of the terminally differentiated chondrocyte and could induce chondrocyte apoptosis. With respect to apatite, once solubilized, there could be a marked elevation in the local Ca²⁺ and phosphate (Pi) ion concentration. The impact of this anion on chondrocyte viability was examined by Mansfield et al. (1999). These investigators demonstrated that Pi was capable of activating apoptosis in chondrocytes at nearly physiological levels. Thus, treatment of hypertrophic tibial chondrocytes with Pi induced death in a dose- and time-dependent manner. Within 48 hours, 3 mM Pi increased chondrocyte apoptosis by 30%; lower concentrations of Pi induced death after 48 hours. More recently, it was shown that this effect is in large part dependent on another apatitic ion, Ca²⁺. A modest increase in the Ca²⁺ concentration from 1.9 to 2.3 mM caused a dramatic increase in chondrocyte death. At a Ca²⁺ level of 2.8 mM, a small rise in the Pi concentration promoted rapid cell death. Since Ca²⁺ alone, even at concentrations of 2.8 mM, did not influence the rate of killing, it was concluded that the concentration of the ion pair served as a primary death signal.

Aside from solubilizing the mineral phase of calcified cartilage, chondroclasts (Gerber et al., 1999) and septoclasts (Lee et al., 1995) degrade protein components of the extracellular matrix. To evaluate whether matrix fragments activated cell death, investigators constructed peptides that contained the sequence RGD (arg-gly-asp). This motif is found in several major extracellular matrix molecules and serves as a ligand for chondrocyte integrin receptors. Peptides containing the RGD sequence caused the death of adherent chick chondrocytes. Surprisingly, the RGD tripeptide itself had no effect on chondrocytes. Chondrocytes were most susceptible to GRGDSP, and somewhat less susceptible to RGDS (Perlot et al., 2002).

Related to matrix degradation and the release of protein fragments that activate apoptosis, a null mutation in the MMP-9/gelatinase B gene caused an abnormal pattern of skeletal growth (Vu et al., 1998). While terminal differentiation appeared to be normal, apoptosis was delayed, resulting in an eight-fold lengthening of the growth plate. Transplantation of wild-type bone marrow cells rescued growth in these mice, indicating that gelatinase B-expressing cells of bone marrow origin probably mediated these processes. There are two possible explanations for the defective apoptosis. First, since gelatinase-B is required for vascularization, it could be argued that failure to generate an angiogenic signal results in defective vascularization, and cells survive attached to matrix macromolecules. The alternative suggestion is that activation of proteases has been linked to the induction of apoptosis during development, neuronal death, and mammary gland involution. Hence, loss of protease activity could serve to spare chondrocytes from the induction of the death response. In a similar vein, it is likely that processes that inhibit mineral accumulation and dissolution in the growth plate would be expected to delay or prevent chondrocyte death. Whether the expanded rachitic or scorbutic growth plate is due to inhibition of chondrocyte apoptosis has yet to be determined.

To summarize the investigations reported here: First, cell culture studies have shown that apoptosis can be triggered by solubilized components of the extracellular matrix. Gene knock-out studies confirm that loss of matrix-degrading enzymes inhibits apoptosis. Thus, there is a direct relationship between matrix degradation and the induction of apoptosis. Second, it is evident that terminally differentiated cells are far more susceptible to apoptotic stimuli than are less mature chondrocytes (some of the reasons for this are summarized in Fig. 3). While both of these observations are very relevant to cells at the chondro-osseous junction, their physiological significance needs further in-depth study. Thus, data are lacking concerning the concentration of the apoptogen in the maturing cell zone, and the extent of exposure. Finally, it is not unreasonable to assume that growth factors and cytokines also play a role in the regulation of apoptosis, possibly by modulating events in survival pathways. Currently, however, there is a paucity of information concerning the contributions of these agents to the apoptotic process.