An epithelial cell destined for apoptosis signals its neighbors to extrude it by an actin- and myosin-dependent mechanism

BACKGROUND

Simple epithelia encase developing embryos and organs. Although these epithelia consist of only one or two layers of cells, they must provide tight barriers for the tissues that they envelop. Apoptosis occurring within these simple epithelia could compromise this barrier. How, then, does an epithelium remove apoptotic cells without disrupting its function as a barrier?

RESULTS

We show that apoptotic cells are extruded from a simple epithelium by the concerted contraction of their neighbors. A ring of actin and myosin forms both within the apoptotic cell and in the cells surrounding it, and contraction of the ring formed in the live neighbors is required for apoptotic cell extrusion, as injection of a Rho GTPase inhibitor into these cells completely blocks extrusion. Addition of apoptotic MDCK cells to an intact monolayer induces the formation of actin cables in the cells contacted, suggesting that the signal to form the cable comes from the dying cell. The signal is produced very early in the apoptotic process, before procaspase activation, cell shrinkage, or phosphatidylserine exposure. Remarkably, electrical resistance studies show that epithelial barrier function is maintained, even when large numbers of dying cells are being extruded.

CONCLUSIONS

We propose that apoptotic cell extrusion is important for the preservation of epithelial barrier function during cell death. Our results suggest that an early signal from the dying cell activates Rho in live neighbors to extrude the apoptotic cell out of the epithelium.

Extracellular control of cell size

Both cell growth (cell mass increase) and progression through the cell division cycle are required for sustained cell proliferation. Proliferating cells in culture tend to double in mass before each division, but it is not known how growth and division rates are co-ordinated to ensure that cell size is maintained. The prevailing view is that coordination is achieved because cell growth is rate-limiting for cell-cycle progression. Here, we challenge this view. We have investigated the relationship between cell growth and cell-cycle progression in purified rat Schwann cells, using two extracellular signal proteins that are known to influence these cells. We find that glial growth factor (GGF) can stimulate cell-cycle progression without promoting cell growth. We have used this restricted action of GGF to show that, for cultured Schwann cells, cell growth rate alone does not determine the rate of cell-cycle progression and that cell size at division is variable and depends on the concentrations of extracellular signal proteins that stimulate cell-cycle progression, cell growth, or both.

Lack of replicative senescence in cultured rat oligodendrocyte precursor cells

Most mammalian somatic cells are thought to have a limited proliferative capacity because they permanently stop dividing after a finite number of divisions in culture, a state termed replicative cell senescence. Here we show that most oligodendrocyte precursor cells purified from postnatal rat optic nerve can proliferate indefinitely in serum-free culture if prevented from differentiating; various cell cycle-inhibitory proteins increase, but the cells do not stop dividing. The cells maintain high telomerase activity and p53- and Rb-dependent cell cycle checkpoint responses, and serum or genotoxic drugs induce them to acquire a senescence-like phenotype. Our findings suggest that some normal rodent precursor cells have an unlimited proliferative capacity if cultured in conditions that avoid both differentiation and the activation of checkpoint responses that arrest the cell cycle.

Evidence that axon-derived neuregulin promotes oligodendrocyte survival in the developing rat optic nerve

It was previously shown that newly formed oligodendrocytes depend on axons for their survival, but the nature of the axon-derived survival signal(s) remained unknown. We show here that neuregulin (NRG) supports the survival of purified oligodendrocytes and aged oligodendrocyte precursor cells (OPCs) but not of young OPCs. We demonstrate that axons promote the survival of purified oligodendrocytes and that this effect is inhibited if NRG is neutralized. In the developing rat optic nerve, we provide evidence that delivery of NRG decreases both normal oligodendrocyte death and the extra oligodendrocyte death induced by nerve transection, whereas neutralization of endogenous NRG increases the normal death. These results suggest that NRG is an axon-associated survival signal for developing oligodendrocytes.

Differences between the clearance of apoptotic cells by professional and non-professional phagocytes

Both professional and non-professional phagocytes [1] participate in clearing the massive numbers of cells that undergo apoptosis during animal development [2], but it is not known how they divide this task. Using time-lapse recordings of cells in culture, we show that professional phagocytes (brain macrophages or microglia) are highly motile, ingest apoptotic cells immediately, and digest them quickly. Non-professionals such as BHK and lens epithelial cells are sessile, often recognize apoptotic cells as soon as they die by showing characteristic palpating movements, but delay ingestion until several hours later. By pre-ageing apoptotic cells, we show that this delay is because the apoptotic cells must undergo further changes before non-professionals can ingest them. The difference was also apparent in vivo, using immunofluorescence and electron microscopy of the developing central nervous system. This arrangement favours prompt clearance by professionals if present in adequate numbers; if they are scarce, however, non-professional bystanders will reluctantly clear the apoptotic cells.

Long-term culture of purified postnatal oligodendrocyte precursor cells. Evidence for an intrinsic maturation program that plays out over months

Oligodendrocytes myelinate axons in the vertebrate central nervous system (CNS). They develop from precursor cells (OPCs), some of which persist in the adult CNS. Adult OPCs differ in many of their properties from OPCs in the developing CNS. In this study we have purified OPCs from postnatal rat optic nerve and cultured them in serum-free medium containing platelet-derived growth factor (PDGF), the main mitogen for OPCs, but in the absence of thyroid hormone in order to inhibit their differentiation into oligodendrocytes. We find that many of the cells continue to proliferate for more than a year and progressively acquire a number of the characteristics of OPCs isolated from adult optic nerve. These findings suggest that OPCs have an intrinsic maturation program that progressively changes the cell's phenotype over many months. When we culture the postnatal OPCs in the same conditions but with the addition of basic fibroblast growth factor (bFGF), the cells acquire these mature characteristics much more slowly, suggesting that the combination of bFGF and PDGF, previously shown to inhibit OPC differentiation, also inhibits OPC maturation. The challenge now is to determine the molecular basis of such a protracted maturation program and how the program is restrained by bFGF.

Evidence that Wallerian degeneration and localized axon degeneration induced by local neurotrophin deprivation do not involve caspases

The selective degeneration of an axon, without the death of the parent neuron, can occur in response to injury, in a variety of metabolic, toxic, and inflammatory disorders, and during normal development. Recent evidence suggests that some forms of axon degeneration involve an active and regulated program of self-destruction rather than a passive "wasting away" and in this respect and others resemble apoptosis. Here we investigate whether selective axon degeneration depends on some of the molecular machinery that mediates apoptosis, namely, the caspase family of cysteine proteases. We focus on two models of selective axon degeneration: Wallerian degeneration of transected axons and localized axon degeneration induced by local deprivation of neurotrophin. We show that caspase-3 is not activated in the axon during either form of degeneration, although it is activated in the dying cell body of the same neurons. Moreover, caspase inhibitors do not inhibit or retard either form of axon degeneration, although they inhibit apoptosis of the same neurons. Finally, we cannot detect cleaved substrates of caspase-3 and its close relatives immunocytochemically or caspase activity biochemically in axons undergoing Wallerian degeneration. Our results suggest that a neuron contains at least two molecularly distinct self-destruction programs, one for caspase-dependent apoptosis and another for selective axon degeneration.

An intrinsic timer that controls cell-cycle withdrawal in cultured cardiac myocytes

Developing cardiac myocytes divide a limited number of times before they stop and terminally differentiate, but the mechanism that stops their division is unknown. To help study the stopping mechanism, we defined conditions under which embryonic rat cardiac myocytes cultured in serum-free medium proliferate and exit the cell cycle on a schedule that closely resembles that seen in vivo. The culture medium contains FGF-1 and FGF-2, which stimulate cell proliferation, and thyroid hormone, which seems to be necessary for stable cell-cycle exit. Time-lapse video recording shows that the cells within a clone tend to divide a similar number of times before they stop, whereas cells in different clones divide a variable number of times before they stop. Cells cultured at 33 degrees C divide more slowly but stop dividing at around the same time as cells cultured at 37 degrees C, having undergone fewer divisions. Together, these findings suggest that an intrinsic timer helps control when cardiac myocytes withdraw from the cell cycle and that the timer does not operate by simply counting cell divisions. We provide evidence that the cyclin-dependent kinase inhibitors p18 and p27 may be part of the timer and that thyroid hormone may help developing cardiac myocytes stably withdraw from the cell cycle.

An analysis of the early events when oligodendrocyte precursor cells are triggered to differentiate by thyroid hormone, retinoic acid, or PDGF withdrawal

Oligodendrocyte precursor cells withdraw from the cell cycle and terminally differentiate after a limited number of cell divisions. The timing of cell-cycle withdrawal and differentiation is controlled by an intrinsic timer, which consists of a timing component that measures elapsed time and an effector component that arrests the cell cycle and initiates differentiation. The effector component can be triggered by either thyroid hormone (TH) or retinoic acid (RA). In this study we investigate how TH and RA act to trigger differentiation. We show the following: (1) Synthetic retinoids that can inhibit AP-1 transcription factors but do not activate gene transcription cannot trigger the effector mechanism, suggesting that TH and RA do not act only by inhibiting AP-1 activity as previously suggested. (2) Both TH and RA induce a transcriptionally dependent antigenic change in purified precursor cells within 2-4 h. (3) Unexpectedly, even before they differentiate, the precursor cells express ceramide galactosyltransferase (CGT), the enzyme that catalyzes the final step in the synthesis of galactocerebroside, an early marker of oligodendrocyte differentiation. (4) Neither TH nor RA directly activates the transcription of the CGT gene, a number of immediate early genes, or the genes that encode any of the known cyclin-dependent kinase inhibitors. (5) The withdrawal of the mitogen platelet-derived growth factor (PDGF), but not TH or RA treatment, causes a rapid decrease in c-fos, NGFI-A/Krox-24, and cyclin D2 mRNA, even though all three treatments trigger cell-cycle arrest and differentiation. (6) PDGF withdrawal and TH treatment, but not RA treatment, induce an increase in cyclin D3 mRNA within 4 h. Thus, we have not found any early changes in gene expression that occur with all three treatments that trigger oligodendrocyte differentiation.

A role for Sonic hedgehog in axon-to-astrocyte signalling in the rodent optic nerve

Retinal ganglion cell (RGC) axons have been shown to stimulate the proliferation of astrocytes in the developing rodent optic nerve, but the signals that mediate this effect have not been identified. The following findings suggest that Sonic hedgehog (Shh) is one of the signals. (1) RGCs express both Shh mRNA and protein, whereas the optic nerve contains the protein but not the mRNA. (2) Astrocytes and their precursors in the developing optic nerve express the Hedgehog (Hh) receptor gene Patched (Ptc), suggesting that they are being signalled by an Hh protein. (3) Ptc expression in the nerve is greatly decreased by either nerve transection or by treatment with neutralizing anti-Shh antibodies, suggesting that it depends on axon-derived Shh. (4) Astrocyte proliferation in the developing nerve is reduced by treatment with anti-Shh antibodies, suggesting that Shh normally helps stimulate this proliferation.

Caspase activation in the terminal differentiation of human epidermal keratinocytes

The epidermis is a multilayered squamous epithelium in which dividing basal cells withdraw from the cell cycle and progressively differentiate as they are displaced toward the skin surface. Eventually, the cells lose their nucleus and other organelles to become flattened squames, which are finally shed from the surface as bags of cross-linked keratin filaments enclosed in a cornified envelope [1]. Although keratinocytes can undergo apoptosis when stimulated by a variety of agents [2], it is not known whether their normal differentiation programme uses any components of the apoptotic biochemical machinery to produce the cornified cell. Differentiating keratinocytes have been reported to share some features with apoptotic cells, such as DNA fragmentation, but these features have not been seen consistently [3]. Apoptosis involves an intracellular proteolytic cascade, mainly mediated by members of the caspase family of cysteine proteases, which cleave one another and various key intracellular target proteins to kill the cell neatly and quickly [4]. Here, we show for the first time that caspases are activated during normal human keratinocyte differentiation and that this activation is apparently required for the normal loss of the nucleus.

Are caspases involved in the death of cells with a transcriptionally inactive nucleus? Sperm and chicken erythrocytes

We show that mouse sperm die spontaneously within 1–2 days in culture and that treatment with either staurosporine (STS) and cycloheximide (CHX) or a peptide caspase inhibitor does not accelerate or delay the cell death. Chicken erythrocytes, by contrast, are induced to die by either serum deprivation or treatment with STS and CHX, and embryonic erythrocytes are more sensitive than adult erythrocytes to both treatments. Although these erythrocyte deaths display a number of features that are characteristic of apoptosis, they are not blocked, or even delayed, by peptide caspase inhibitors, and most of the cells die without apparently activating caspases. A small proportion of the dying erythrocytes do activate caspase-3, but even these cells, which seem to be the least mature erythrocytes, die just as quickly in the presence of caspase inhibitors. Our findings raise the possibility that both mouse sperm and chicken erythrocytes have a death programme that may not depend on caspases and that chicken erythrocytes lose caspases as they mature. Chicken erythrocytes may provide a useful ‘stripped down’ cell system to try to identify the protein components of such a death programme, which may serve to back-up the conventional caspase-dependent suicide mechanism in many cell types.

Cell number control and timing in animal development: the oligodendrocyte cell lineage

Our studies of oligodendrocyte development in the rodent optic nerve provide clues as to how cell numbers and the timing of differentiation may be controlled during mammalian development. Both cell number and the timing of differentiation depend on intracellular programs and extracellular signals, which together control cell survival and cell division. As the cells seem to compete for limiting amounts of both survival signals and mitogens, the levels of these extracellular signals must be tightly regulated, but it is not known how this is achieved. The timing of cell-cycle exit, and therefore the onset of differentiation, seems to depend in part on the progressive accumulation of the intracellular Cdk inhibitor p27/Kip1, but it is still unclear how the level of this protein is controlled over time in the dividing cells. The timing of cell-cycle exit is also regulated by thyroid hormone, which, along with other hormones, seems to coordinate the timing of development in various organs, much as the timing of the multiple changes in metamorphosis in both vertebrates and invertebrates is coordinated by hormones. In this sense, one might think of mammalian development as a prolonged metamorphosis.

p27Kip1 alters the response of cells to mitogen and is part of a cell-intrinsic timer that arrests the cell cycle and initiates differentiation

BACKGROUND

In many vertebrate cell lineages, precursor cells divide a limited number of times before they arrest and terminally differentiate into postmitotic cells. It is not known what causes them to stop dividing. We have been studying the 'stopping' mechanism in the proliferating precursor cells that give rise to oligodendrocytes, the cells that make myelin in the central nervous system. We showed previously that the cyclin-dependent kinase inhibitor p27Kip1 (p27) progressively accumulates in cultured precursor cells as they proliferate and that the time course of the increase is consistent with the possibility that p27 accumulation is part of a cell-intrinsic timer that arrests the cell cycle and initiates differentiation at the appropriate time.

RESULTS

We now provide direct evidence that p27 is part of the intrinsic timer. We show that although p27-/- precursor cells stop dividing and differentiate almost as fast as wild-type cells when deprived of mitogen, when stimulated by saturating amounts of mitogen they have a normal cell-cycle time but tend to go through one or two more divisions than wild-type cells before they stop and differentiate. Cells that are p27+/- behave in an intermediate way, going through at most one extra division, indicating that the levels of p27 matter in the way the timer works. We also show that p27-/- precursor cells are more sensitive than wild-type cells to the mitogenic effect of platelet-derived growth factor.

CONCLUSIONS

These findings demonstrate that p27 is part of the normal timer that determines when oligodendrocyte precursor cells stop dividing and differentiate, at least in vitro. It seems likely that p27 plays a similar role in many other cell lineages, which could explain the phenotypes of the p27-/- and p27+/- mice.

A role for caspases in lens fiber differentiation

There is increasing evidence that programmed cell death (PCD) depends on a novel family of intracellular cysteine proteases, called caspases, that includes the Ced-3 protease in the nematode Caenorhabditis elegans and the interleukin-1beta-converting enzyme (ICE)-like proteases in mammals. Some developing cells, including lens epithelial cells, erythroblasts, and keratinocytes, lose their nucleus and other organelles when they terminally differentiate, but it is not known whether the enzymatic machinery of PCD is involved in any of these normal differentiation events. We show here that at least one CPP32 (caspase-3)-like member of the caspase family becomes activated when rodent lens epithelial cells terminally differentiate into anucleate lens fibers in vivo, and that a peptide inhibitor of these proteases blocks the denucleation process in an in vitro model of lens fiber differentiation. These findings suggest that at least part of the machinery of PCD is involved in lens fiber differentiation.

Effects of thyroid hormone on embryonic oligodendrocyte precursor cell development in vivo and in vitro

The oligodendrocyte precursor cell divides a limited number of times before terminal differentiation. The timing of differentiation depends on both intracellular mechanisms and extracellular signals, including mitogens that stimulate proliferation and signals such as thyroid hormone (TH) and retinoic acid (RA) that help trigger the cells to stop dividing and differentiate. We show here that, both in vivo and in vitro, TH is required for the normal development of rodent optic nerve oligodendrocytes, although in its absence some oligodendrocyte development still occurs, perhaps promoted by signals from axons. We also demonstrate that TH from both mother and pup plays a part in oligodendrocyte development in vivo. Finally, we show that precursors in embryonic nerve cultures differ from those in postnatal cultures in two ways: they respond much better to TH than to RA, and they respond more slowly to TH, suggesting that oligodendrocyte precursor cells mature during their early development.

Continuous observation of multipotential retinal progenitor cells in clonal density culture

All neural cell types in the vertebrate retina, except astrocytes, have been shown to develop from multipotential progenitor cells. It is not known, however, to what extent the progenitor cells are heterogeneous in their developmental potential or to what extent cell-cell interactions versus cell-autonomous factors influence the types of cells they become. To address these issues we developed a clonal-density cell culture system where mouse retinal progenitor cells can survive, divide, and differentiate. We followed the development of clones both by continuous time-lapse video microscopy and by daily microscopic observation. We show that even when cultured at clonal density in a homogeneous general environment, where they cannot contact cells outside their own clone, the retinal progenitor cells vary in proliferative capacity, cell cycle time, and in the cell types that they generate. In addition, we show that under these conditions single progenitor cells can generate both neurons and glia, in which case the neurons almost always develop before glial cells, as is the case in vivo.

Muller-cell-derived leukaemia inhibitory factor arrests rod photoreceptor differentiation at a postmitotic pre-rod stage of development

In the present study, we examine rod photoreceptor development in dissociated-cell cultures of neonatal mouse retina. We show that, although very few rhodopsin+ rods develop in the presence of 10% foetal calf serum (FCS), large numbers develop in the absence of serum, but only if the cell density in the cultures is high. The rods all develop from nondividing rhodopsin- cells, and new rods continue to develop from rhodopsin- cells for at least 6–8 days, indicating that there can be a long delay between when a precursor cell withdraws from the cell cycle and when it becomes a rhodopsin+ rod. We show that FCS arrests rod development in these cultures at a postmitotic, rhodopsin-, pre-rod stage. We present evidence that FCS acts indirectly by stimulating the proliferation of Muller cells, which arrest rod differentiation by releasing leukaemia inhibitory factor (LIF). These findings identify an inhibitory cell-cell interaction, which may help to explain the long delay that can occur both in vitro and in vivo between cell-cycle withdrawal and rhodopsin expression during rod development.

Is programmed cell death required for neural tube closure?

Programmed cell death (PCD) plays an important part in animal development. It is responsible for eliminating the cells between developing digits, for example, and is involved in hollowing out solid structures to create cavities (reviewed in [1] [2]). There are many cases, however, where PCD occurs in developing tissues but its function is unknown. Important examples are seen during the folding, pinching off, and fusion of epithelial sheets during vertebrate morphogenesis, as in the formation of the neural tube and lens vesicle [2]; PCD is an invariable accompaniment to these processes, but it is unclear whether it is required for the processes to occur or is just an unavoidable consequence of them. There is increasing evidence that PCD in animals is mediated by a family of cysteine proteases, known as caspases, which are thought to act in a proteolytic cascade, cleaving one another and key intracellular proteins to kill the cell in a controlled way [3] [4]. Inhibitors of caspases are, therefore, potential tools for studying the roles of PCD during animal development [5] [6]. Here, we show that peptide caspase inhibitors block neural tube closure in explanted chick embryos, suggesting that PCD is required for this crucial developmental process.

Retinal ganglion cell axons drive the proliferation of astrocytes in the developing rodent optic nerve

We show that the proliferation of astrocytes in the developing rodent optic nerve absolutely depends on axons and that this axonal influence depends on axonal transport but not on axonal electrical activity. We also show that purified retinal ganglion cells stimulate DNA synthesis in optic nerve astrocytes in culture and that the effect can be mimicked by fibroblast growth factor but not by neuregulins or several other growth factors. Taken together with previous findings, our present results indicate that axons promote glial cell proliferation and survival in the developing optic nerve by at least three distinct mechanisms.

Differentiation and morphogenesis in pellet cultures of developing rat retinal cells

We previously developed a reaggregate cell culture system (pellet cultures) in which retinal neuroepithelial cells proliferate and give rise to rod photoreceptor cells (rods) in vitro (Watanabe and Raff, 1990, Neuron 4:461-467). In the present study, we analyzed cell differentiation and morphogenesis in pellet cultures by using both cell-type-specific markers with immunofluorescence and electron microscopy. We demonstrated that, in addition to rods, the other major retinal cell types, including amacrine cells, bipolar cells, Müller cells, and ganglion cells were all present in the pellets, where most were able to develop from dividing precursor cells in vitro. The different cell types in the pellets became organized into two distinct structures: dark rosettes and pale rosettes. The cellular composition of these structures indicated that the dark rosettes correspond to the outer nuclear layer and the pale rosettes to the inner nuclear layer of the normal retina. Ultrastructural studies have indicated that the thin layer of neuronal processes surrounding the dark rosettes correspond to the outer plexiform layer, and the central region of the pale rosettes correspond to the inner plexiform layer of the normal retina. Other features of normal retinal development also occurred in the pellets, including programmed cell death and the formation of inner and outer rod cell segments and synapses. Thus, pellet cultures provide a convenient way to study different aspects of retinal development where one can control the size and the cellular composition of the initial reaggregate.

Ciliary Neurotrophic Factor Enhances the Rate of Oligodendrocyte Generation

Although ciliary neurotrophic factor (CNTF) is a potent survival factor for many types of neurons and glial cells in vitro, there is currently no evidence that it participates in normal development. Here we show that CNTF greatly enhances the rate of oligodendrocyte generation. Proliferation of oligodendrocyte precursor cells purified from rodent optic nerves and cultured in platelet-derived growth factor-containing medium is significantly increased by CNTF. Similarly, the number of proliferating oligodendrocyte precursor cells in developing optic nerves of transgenic mice lacking CNTF is decreased by up to threefold and the number of oligodendrocytes is transiently decreased; proliferation is restored to normal by the delivery of exogenous CNTF into the developing optic nerve. Both oligodendrocyte number and myelination ultimately attain wild-type values in CNTF-deficient adult mice, indicating that CNTF is not necessary for either oligodendrocyte differentiation or myelination, although it normally accelerates oligodendrocyte development by enhancing the proliferation of oligodendrocyte precursor cells.

Constitutive expression of the machinery for programmed cell death

In the presence of cycloheximide (CHX) to inhibit protein synthesis, a high concentration of staurosporine (STS) induces almost all cells in explant cultures of 8/8 types of newborn mouse organs and 3/3 types of adult mouse organs to die with the characteristic features of apoptosis. Eggs and blastomeres also die in this way when treated with STS and CHX, although they are less sensitive to this treatment than trophectoderm or inner cell mass cells whose sensitivity resembles that of other developing cells. Human red blood cells are exceptional in being completely resistant to treatment with STS and CHX. As (STS plus CHX)-induced cell deaths have been shown to display the characteristic features of programmed cell death (PCD), we conclude that all mammalian nucleated cells are capable of undergoing PCD and constitutively express all the proteins required to do so. It seems that the machinery for PCD is in place and ready to run, even though its activation often depends on new RNA and protein synthesis.

Role of Ced-3/ICE-family proteases in staurosporine-induced programmed cell death

In the accompanying paper by Weil et al. (1996) we show that staurosporine (STS), in the presence of cycloheximide (CHX) to inhibit protein synthesis, induces apoptotic cell death in a large variety of nucleated mammalian cell types, suggesting that all nucleated mammalian cells constitutively express all of the proteins required to undergo programmed cell death (PCD). The reliability of that conclusion depends on the evidence that STS-induced, and (STS + CHS)-induced, cell deaths are bona fide examples of PCD. There is rapidly accumulating evidence that some members of the Ced-3/Interleukin-1 beta converting enzyme (ICE) family of cysteine proteases are part of the basic machinery of PCD. Here we show that Z-Val-Ala-Asp-fluoromethylketone (zVAD-fmk), a cell-permeable, irreversible, tripeptide inhibitor of some of these proteases, suppresses STS-induced and (STS + CHX)-induced cell death in a wide variety of mammalian cell types, including anucleate cytoplasts, providing strong evidence that these are all bona fide examples of PCD. We show that the Ced-3/ICE family member CPP32 becomes activated in STS-induced PCD, and that Bcl-2 inhibits this activation. Most important, we show that, in some cells at least, one or more CPP32-family members, but not ICE itself, is required for STS-induced PCD. Finally, we show that zVAD-fmk suppresses PCD in the interdigital webs in developing mouse paws and blocks the removal of web tissue during digit development, suggesting that this inhibition will be a useful tool for investigating the roles of PCD in various developmental processes.

Glial cells are increased proportionally in transgenic optic nerves with increased numbers of axons

To study how an increase in axon number influences the number of glial cells in the mammalian optic nerve, we have analyzed a previously described transgenic mouse that expresses the human bcl-2 gene from a neuron-specific enolase promoter. In these mice, the normal postnatal loss of retinal ganglion cell axons is greatly decreased and, as a consequence, the number of axons in the optic nerve is increased by approximately 80% compared with wild-type mice. Remarkably, the numbers of oligodendrocytes, astrocytes, and microglial cells are all increased proportionally in the transgenic optic nerve. The increase in oligodendrocytes apparently results from both a decrease in normal oligodendrocyte death and an increase in oligodendrocyte precursor cell proliferation, whereas the increase in astrocytes apparently results from an increase in the proliferation of astrocyte lineage cells. Unexpectedly, the transgene is expressed in oligodendrocytes and astrocytes, but this does not seem to be responsible for the increased numbers of these cells. These findings indicate that developing neurons and glial cells can interact to adjust glial cell numbers appropriately when neuronal numbers are increased. We also show that the expression of the bcl-2 transgene in retinal ganglion cells protects the cell body from programmed cell death when the axon is cut, but it does not protect the isolated axon from Wallerian degeneration, even though the transgene-encoded protein is present in the axon.

Quantification of normal cell death in the rat retina: implications for clone composition in cell lineage analysis

Naturally occurring cell death complicates the analysis of cell lineage studies by making the surviving members of a clone appear more closely related than they actually are. Here we ask how much normal cell death occurs during rat retinal development, and whether that amount of death is sufficient to confuse the analysis of cell lineage relationships. We measure total cell death in the retina by combining relative counts of dead cells with absolute measurements of total cell loss. For most cell types, but not rods, we find that half of the cells generated die during normal retinal development. We use a computer model to quantify the effects of different amounts of cell death in a simulated lineage study. The simulation indicates that 50% cell death means that clonal variability analysed after the cell death period is not necessarily a good indicator of how much variability actually occurs in the underlying lineage.

Programmed cell death by default in embryonic cells, fibroblasts, and cancer cells

We recently proposed that most mammalian cells constitutively express all of the proteins required to undergo programmed cell death (PCD) and undergo PCD unless continuously signaled by other cells not to. Although some cells have been shown to work this way, the vast majority of cell types remain to be tested. Here we tested purified fibroblasts isolated from developing or adult rat sciatic nerve, a mixture of cell types isolated from normal or p53-null mouse embryos, an immortalized rat fibroblast cell line, and a number of cancer cell lines. We found the following: 1) All of these cells undergo PCD when cultured at low cell density in the absence of serum and exogenous signaling molecules but can be rescued by serum or specific growth factors, suggesting that they need extracellular signals to avoid PCD. (2) The mixed cell types dissociated from normal mouse embryos can only support one another's survival in culture if they are in aggregates, suggesting that cell survival in embryos may depend on short-range signals. (3) Some cancer cells secrete factors that support their own survival. (4) The survival requirements of a human leukemia cell line change when the cells differentiate. (5) All of the cells studied can undergo PCD in the presence of cycloheximide, suggesting that they constitutively express all of the protein components required to execute the death program.

Evidence for large-scale astrocyte death in the developing cerebellum

There is increasing evidence that some glial cells die during normal vertebrate development, but the extent of the death and the types of glial cells that die remain uncertain. We have analyzed pyknotic cells in the developing postnatal rat cerebellum. During the first postnatal week, the majority of pyknotic cells are in the developing white matter where their number peaks at about postnatal day 7 (P7) and then declines sharply. Pyknotic cells in the internal granule cell layer peak at P10, while those in the molecular and external granule cell layers peak later. Both electron microscopy and in situ end labeling of DNA catalyzed by terminal deoxynucleotidyl transferase confirm that the pyknotic cells are undergoing apoptosis. Immunohistochemical staining suggests that 50-70% of the pyknotic cells in the white matter and internal granule cell layer are astrocytes. We estimate that at P7, as many as 50% of the white matter cells die and, of these, more than half appear to be astrocytes.

Programmed cell death and Bcl-2 protection in very low oxygen

Programmed cell death (PCD) is a fundamental feature of animal cells, but the mechanism remains unknown. Similarly, the Bcl-2 oncoprotein can suppress PCD in a variety of cell types and circumstances, but it is not known how it does so. It has been suggested that PCD involves the generation of reactive oxygen species (ROS) and that Bcl-2 protects against PCD by inhibiting the generation or action of ROS. To determine whether ROS are required for PCD, we cultured cells in a near-anaerobic atmosphere where the generation of ROS would be expected not to occur, or at least to be greatly reduced. We find that these conditions inhibit PCD induced by ROS-generating agents but do not inhibit PCD induced by other means. Furthermore, we show that Bcl-2 can protect cells from PCD in these anaerobic conditions. These results suggest that ROS are not required for PCD, and that Bcl-2 protects against PCD in ways that do not depend on the inhibition of ROS production or activity.

Programmed cell death and the control of cell survival

We draw the following tentative conclusions from our studies on programmed cell death (PCD): (i) the amount of normal cell death in mammalian development is still underestimated; (ii) most mammalian cells constitutively express the proteins required to undergo PCD; (iii) the death programme operates by default when a mammalian cell is deprived of signals from other cells; (iv) many normal cell deaths may occur because cells fail to obtain the extracellular signals they need to suppress the death programme; and (v) neither the nucleus nor mitochondrial respiration is required for PCD (or Bcl-2 protection from PCD), raising the possibility that the death programme, like mitosis, is orchestrated by a cytosolic regulator that acts on multiple organelles in parallel.

Autocrine signals enable chondrocytes to survive in culture

We recently proposed that most mammalian cells other than blastomeres may be programmed to kill themselves unless continuously signaled by other cells not to. Many observations indicate that some mammalian cells are programmed in this way, but is it the case for most mammalian cells? As it is impractical to test all of the hundreds of types of mammalian cells, we have focused on two tissues--lens and cartilage--which each contain only a single cell type: if there are cells that do not require signals from other cells to avoid programmed cell death (PCD), lens epithelial cells and cartilage cells (chondrocytes) might be expected to be among them. We have previously shown that rat lens epithelial cells can survive in serum-free culture without signals from other cell types but seem to require signals from other lens epithelial cells to survive: without such signals they undergo PCD. We show here that the same is true for rat (and chick) chondrocytes. They can survive for weeks in culture at high cell density in the absence of other cell types, serum, or exogenous proteins or signaling molecules, but they die with the morphological features of apoptosis in these conditions at low cell density. Medium from high density cultures, FCS, or a combination of known growth factors, all support prolonged chondrocyte survival in low density cultures, as long as antioxidants are also present. Moreover, medium from high density chondrocyte cultures promotes the survival of lens epithelial cells in low density cultures and vice versa. Chondrocytes isolated from adult rats behave similarly to those isolated from developing rats. These findings support the hypothesis that most mammalian cells require signals from other cells to avoid PCD, although the signals can sometimes be provided by cells of the same type, at least in tissues that contain only one cell type.

A novel role for thyroid hormone, glucocorticoids and retinoic acid in timing oligodendrocyte development

The timing of oligodendrocyte differentiation is thought to depend on an intrinsic clock in oligodendrocyte precursor cells that counts time or cell divisions and limits precursor cell proliferation. We show here that this clock mechanism can be separated into a counting component and an effector component that stops cell proliferation: whereas the counting mechanism is driven by mitogens that activate cell-surface receptors, the effector mechanism depends on hydrophobic signals that activate intracellular receptors, such as thyroid hormones, glucocorticoids and retinoic acid. When purified oligodendrocyte precursor cells are cultured at clonal density in serum-free medium in the presence of mitogens but in the absence of these hydrophobic signals, the cells divide indefinitely and do not differentiate into postmitotic oligodendrocytes. In the absence of mitogens, the precursor cells stop dividing and differentiate prematurely into oligodendrocytes even in the absence of these hydrophobic signals, indicating that these signals are not required for differentiation. The levels of these signals in vivo may normally regulate the timing of oligodendrocyte differentiation, as the maximum number of precursor cell divisions in culture depends on the concentration of such signals and injections of thyroid hormone into newborn rats accelerates oligodendrocyte development. As thyroid hormone, glucocorticoids and retinoic acid have been shown to promote the differentiation of many types of vertebrate cells, it is possible that they help coordinate the timing of differentiation by signalling clocks in precursor cells throughout a developing animal.

Programmed cell death and Bcl-2 protection in the absence of a nucleus

The molecular basis of programmed cell death (PCD) is unknown. An important clue is provided by the Bcl-2 protein, which can protect many cell types from PCD, although it is not known where or how it acts. Nuclear condensation, DNA fragmentation and a requirement for new RNA and protein synthesis are often considered hallmarks of PCD. We show here, however, that anucleate cytoplasts can undergo PCD and that Bcl-2 and extracellular survival signals can protect them, indicating that, in some cases at least, the nucleus is not required for PCD or for Bcl-2 or survival factor protection. We propose that PCD, like the cell cycle, is orchestrated by a cytoplasmic regulator that has multiple intracellular targets.

A crucial role for neurotrophin-3 in oligodendrocyte development

The neurotrophins nerve growth factor, brain-derived neurotrophic factor, neurotrophin-3 and neurotrophin-4/5 promote the survival of subpopulations of vertebrate neurons in vitro, but so far only nerve growth factor has been demonstrated to be essential for normal neuronal development; no neurotrophin has previously been shown to function in normal glial cell development. We found recently that neurotrophin-3 promotes the survival of pure oligodendrocyte precursor cells in vitro, and, although by itself it induces only a small percentage of these cells to synthesize DNA, in combination with platelet-derived growth factor it induces the majority of them to do so. Neither of these factors, however, has been shown to contribute to oligodendrocyte precursor cell proliferation in vivo or to stimulate pure populations of these cells to proliferate (as opposed to synthesize DNA) in vitro. Here we show that neurotrophin-3 and platelet-derived growth factor collaborate to promote clonal expansion of oligodendrocyte precursor cells in vitro and to drive the intrinsic clock that times oligodendrocyte development. We also show that neurotrophin-3 helps stimulate the proliferation of oligodendrocyte precursor cells in vivo and is thus required for normal oligodendrocyte development.

Programmed cell death and the control of cell survival: lessons from the nervous system

During the development of the vertebrate nervous system, up to 50 percent or more of many types of neurons normally die soon after they form synaptic connections with their target cells. This massive cell death is thought to reflect the failure of these neurons to obtain adequate amounts of specific neurotrophic factors that are produced by the target cells and that are required for the neurons to survive. This neurotrophic strategy for the regulation of neuronal numbers may be only one example of a general mechanism that helps to regulate the numbers of many other vertebrate cell types, which also require signals from other cells to survive. These survival signals seem to act by suppressing an intrinsic cell suicide program, the protein components of which are apparently expressed constitutively in most cell types.

Does oligodendrocyte survival depend on axons?

BACKGROUND

We have shown previously that oligodendrocytes and their precursors require signals from other cells in order to survive in culture. In addition, we have shown that about 50% of the oligodendrocytes produced in the developing rat optic nerve normally die, apparently in a competition for the limiting amounts of survival factors. We have hypothesized that axons may control the levels of such oligodendrocyte survival factors and that the competition-dependent death of oligodendrocytes serves to match their numbers to the number of axons that they myelinate. Here we test one prediction of this hypothesis - that the survival of developing oligodendrocytes depends on axons.

RESULTS

We show that oligodendrocyte death occurs selectively in transected nerves in which the axons degenerate. This cell death is prevented by the delivery of exogenous ciliary neurotrophic factor (CNTF) or insulin-like growth factor I (IGF-1), both of which have been shown to promote oligodendrocyte survival in vitro. We also show that purified neurons promote the survival of purified oligodendrocytes in vitro.

CONCLUSION

These results strongly suggest that oligodendrocyte survival depends upon the presence of axons; they also support the hypothesis that a competition for axon-dependent survival signals normally helps adjust the number of oligodendrocytes to the number of axons that require myelination. The identities of these signals remain to be determined.

Large-scale normal cell death in the developing rat kidney and its reduction by epidermal growth factor

Although normal cell death is known to occur in many developing vertebrate organs, it has not been thought to play an important part in the development of the mammalian kidney. We show here that normal cell death is found in both the nephrogenic region and medullary papilla of the developing rat kidney and, in each of these areas, it follows a distinct developmental time course. As many as 3% of the cells in these areas have a typical apoptotic morphology and the dead cells seem to be cleared rapidly (within 1–2 hours) by phagocytosis by neighbouring parenchymal cells. These values are similar to those in vertebrate neural tissues where 50% or more of the cells die during normal development, suggesting that large-scale death is a normal feature of kidney development. We also show that in vivo treatment with epidermal growth factor inhibits cell death in the developing kidney, consistent with the possibility that the cells normally die because they lack sufficient survival factors. Our findings suggest that the extent of normal cell death in developing animals is still greatly underestimated and they raise the possibility that many of these cell deaths may reflect limiting amounts of survival factors.

Control of lens epithelial cell survival

We have studied the survival requirements of developing lens epithelial cells to test the hypothesis that most cells are programmed to kill themselves unless they are continuously signaled by other cells not to do so. The lens cells survived for weeks in both explant cultures and high-density dissociated cell cultures in the absence of other cells or added serum or protein, suggesting that they do not require signals from other cell types to survive. When cultured at low density, however, they died by apoptosis, suggesting that they depend on other lens epithelial cells for their survival. Lens epithelial cells cultured at high density in agarose gels also survived for weeks, even though they were not in direct contact with one another, suggesting that they can promote one another's survival in the absence of cell-cell contact. Conditioned medium from high density cultures promoted the survival of cells cultured at low density, suggesting that lens epithelial cells support one another's survival by secreting survival factors. We show for the first time that normal cell death occurs within the anterior epithelium in the mature lens, but this death is strictly confined to the region of the anterior suture.

Multiple extracellular signals are required for long-term oligodendrocyte survival

We showed previously that oligodendrocytes and their precursors require continuous signalling by protein trophic factors to avoid programmed cell death in culture. Here we show that three classes of such trophic factors promote oligodendrocyte survival in vitro: (1) insulin and insulin-like growth factors (IGFs), (2) neurotrophins, particularly neurotrophin-3 (NT-3), and (3) ciliary-neurotrophic factor (CNTF), leukemia inhibitory factor (LIF) and interleukin 6 (IL-6). A single factor, or combinations of factors within the same class, promote only short-term survival of oligodendrocytes and their precursors, while combinations of factors from different classes promote survival additively. Long-term survival of oligodendrocytes in vitro requires at least one factor from each class, suggesting that multiple signals may be required for long-term oligodendrocyte survival in vivo. We also show that CNTF promotes oligodendrocyte survival in vivo, that platelet-derived growth factor (PDGF) can promote the survival of oligodendrocyte precursors in vitro by acting on a novel, very high affinity PDGF receptor, and that, in addition to its effect on survival, NT-3 is a potent mitogen for oligodendrocyte precursor cells.

Bcl-2 blocks apoptosis in cells lacking mitochondrial DNA

When the mammalian proto-oncogene bcl-2 is overexpressed it can protect various types of cells both from normal and from experimentally induced apoptosis, but the molecular mechanisms involved are unknown. Although the Bcl-2 protein is membrane-associated, its subcellular location is controversial: two studies have suggested that it is mainly associated with the nuclear envelope and endoplasmic reticulum, whereas another study has suggested that it is mainly located in the inner mitochondrial membrane. The latter study has suggested that Bcl-2 might protect cells from apoptosis by altering mitochondrial function and that mitochondria may be involved in apoptosis. Here we report that human mutant cell lines that lack mitochondrial DNA (mtDNA), and therefore do not have a functional respiratory chain, can still be induced to die by apoptosis, and that they can be protected from apoptosis by the overexpression of bcl-2, suggesting that neither apoptosis nor the protective effect of bcl-2 depends on mitochondrial respiration. We also show that the Bcl-2 protein in overexpressing cells is associated with the nuclear envelope and endoplasmic reticulum, as well as with mitochondria.

Proliferation of oligodendrocyte precursor cells depends on electrical activity in axons

Oligodendrocytes myelinate axons in the vertebrate central nervous system. It would, therefore, make sense if axons played a part in controlling the number of oligodendrocytes that develop in a myelinated tract. Although oligodendrocytes themselves normally do not divide, the precursor cells that give rise to them do. Here we show that the proliferation of oligodendrocyte precursor cells in the developing rat optic nerve depends on electrical activity in neighbouring axons, and that this activity-dependence can be circumvented by experimentally increasing the concentration of platelet-derived growth factor, which is present in the optic nerve and stimulates these cells to proliferate in culture. These findings suggest that axonal electrical activity normally controls the production and/or release of the growth factors that are responsible for proliferation of oligodendrocyte precursor cells and thereby helps to control the number of oligodendrocytes that develop in the region.

Visualization of O-2A progenitor cells in developing and adult rat optic nerve by quisqualate-stimulated cobalt uptake

Some macroglial cells of the O-2A lineage express glutamate receptor channels of the quisqualate/kainate type and take up extracellular cobalt when activated by glutamate agonists. These cells can be identified both in vitro and in situ following precipitation and intensification of the intracellular cobalt. We have used this technique to characterize these cells in the developing and adult rat optic nerve. In purified cultures of optic nerve cells, O-2A progenitor cells and type 2 astrocytes took up cobalt in the presence of quisqualate, while oligodendrocytes, type 1 astrocytes, and microglial cells did not. When whole optic nerves of various postnatal ages were exposed to quisqualate and cobalt, a subpopulation of glial cells took up cobalt. Cobalt uptake in vitro and in situ was blocked by 6-cyano-7-nitroquinoxaline-2,3-dione. The number, morphology, and spatial distribution of cobalt-filled cells in situ varied with age. In perinatal nerves, 9% of glial cells took up cobalt. These cells had a simple unipolar or bipolar morphology and were two to three times more concentrated at the chiasm end than at the eye end of the nerve. During subsequent development, this gradient disappeared and the cobalt-filled cells became progressively more complex in morphology and increased in number and density, reaching a peak toward the end of the second postnatal week. The number subsequently declined to about 16,000 (7%) in the adult nerve. The processes of some cobalt-filled cells appeared to contact nodes of Ranvier. All cobalt-filled cells in 2 1/2-week-old optic nerves had a similar ultrastructural appearance and did not resemble either mature oligodendrocytes or astrocytes. Our results suggest that the cells stimulated by quisqualate to take up cobalt in the optic nerve are the in vivo counterpart of O-2A progenitor cells. We found no evidence that any of these cells are type 2 astrocytes.

Cell death in the oligodendrocyte lineage

We have recently found that about 50% of newly formed oligodendrocytes normally die in the developing rat optic nerve. When purified oligodendrocytes or their precursors are cultured in the absence of serum or added signalling molecules, they die rapidly with the characteristics of programmed cell death. This death is prevented either by the addition of medium conditioned by cultures of their normal neighboring cells in the developing optic nerve, or by the addition of platelet-derived growth factor (PDGF) or insulin-like growth factors (IGFs). Increasing PDGF in the developing optic nerve decreases normal oligodendrocyte death by up to 90% and doubles the number of oligodendrocytes, suggesting that this normally occurring glial cell death might result from a competition for limiting amounts of survival signals. These results suggest that competition for limiting amounts of survival factors is not confined to developing neurons, and raise the possibility that a similar mechanism may be responsible for some naturally occurring cell deaths in nonneural tissues.