Morphology of programmed cell death in the ventral nerve cord of Caenorhabditis elegans larvae

The morphology of programmed cell death has been studied widely in the morphogenesis of mammalian (Ballard & Holt, 1968 ; Farbman, 1968 ; Matthiessen & Andersen, 1972; Schweichel & Merker, 1973; El-Shershaby & Hinchliffe, 1974, 1975), avian (Bellairs, 1961; Saunders, 1966; Saunders & Fallon, 1966; Manasek, 1969; Hammar & Mottet, 1971; Schlüter, 1973; O’Connor & Wyttenbach, 1974; Hurle & Hinchliffe, 1978), amphibian (Kerr, Harmon & Searle, 1974), reptilian (Fallon & Cameron, 1977) and insect tissues (Goldsmith, 1966; Giorgi & Deri, 1976). Cell death in these systems appears to be by apoptosis (Kerr, Wyllie & Currie, 1972; Wyllie, Kerr & Currie, 1980), a mode of death which is characterized by condensation of nuclear chromatin, compaction of organelles, budding into membrane-bounded fragments and phagocytosis by adjacent cells.

By contrast little is known about the ultrastructure of programmed cell death in the morphogenesis of lower invertebrates, but the larvae of the nematode C. elegans provide a uniquely favourable system in which to study this phenomenon. By the use of Nomarski differential interference contrast microscopy on living C. elegans, all postembryonic cell lineages have been followed (Sulston & Horvitz, 1977; Kimble & Hirsh, 1979). The precise time of division, migration and death is known for these cells with lespect to the stage of larval development, and the position of each of these cells is also accurately defined. In particular, at about the time of the first larval moult, 56 nerve cells are added to the ventral nerve cord and associated ganglia by a uniform division of 13 neuroblasts (P0 –P12) followed by a defined pattern of cell deaths (Sulston & Horvitz, 1977). We here record a transmission electron microscope study of progeny of Pll which are programmed to die (Fig. 1). The fate of these cells is traced from birth at mitosis through to death and disappearance.

Self-fertilizing hermaphrodite stocks of Caenorhabditis elegans var. Bristol (Strain N2) were grown at 20 °C on agar plates seeded with Escherichia coli OP50 as previously described (Brenner, 1974).

Our studies were restricted to cell death in the P11 lineage series of nerve cells observed during the first and second larval stages after hatching (LI and L2 respectively). These nerve cells are situated in the posterior end of the nematode approximately 20 μm from the anus. Cells whose death was known from lineage studies to be either imminent or in progress were identified and photographed by Nomarski optics (Sulston & Horvitz, 1977); the animal was then immediately fixed for electron microscopy.

The individual nematodes were fixed in 1 % OsO₄ in 0 ·1 M cacodylate buffer, pH 7 ·4, for 1 h at room temperature. The nematode was bisected with a fine razor blade and the caudal half treated en bloc in 1 % uranyl acetate in 0 ·05 M maleate, pH 6 ·1, for 45 min (Ware, Clark, Crossland & Russell, 1975), and then transferred on to a thin layer of 1 % agar and covered with a drop of 1 % agar which was then allowed to set. A block of agar containing the specimen was then excised, dehydrated and embedded in Araldite as previously described (Ward, Thomson, White & Brenner, 1975). Sections of approximately 50 nm were cut with a diamond knife on an LKB Ultratome III. Ribbons of serial section were cut transversely to the long axis of the nematode, and were picked up on formvar coated 75 mesh grids. The sections were post-stained with 5 % uranyl acetate, for 10 min at 60 °C, and lead citrate for 5 min and viewed and photographed in an AEI EM 6B microscope.

P11 precursor nerve cells undergo divisions, migrations and deaths according to the lineage already established (Sulston & Horvitz, 1977) (Fig. 1). The surviving progeny are three nerve cells –P11. aaaa, P11. apa, P11. app - and 1 ventral hypodermal cell-PH.p. Cell nuclei and large nucleoli are resolved clearly by Nomarski optics though cell boundaries cannot always be visualised (Fig. 2). Nerve cells have granular nucleoplasm and generally no visible nucleolus whereas ventral hypodermal cells have large nuclei and nucleoli. P11 .aap and P11. aaap undergo programmed death. In other regions of the ventral cord, homologues of these cells survive and become motor neurons. The deaths of Pll.aap and Pll.aaap appear identical and our results and illustrations are drawn from both.

Using Nomarski optics, morphological changes involved in programmed death can be divided into five stages (I –V) (Fig. 2), which are described ultrastructurally in the following section. Division of the parent cell occurs (Stage I) and soon afterwards the cytoplasm of the daughter cell programmed to die seems to become less refractile (Stage II). The nucleus of the dying cell then becomes less granular and more refractile and refractile beads appear around it (Stage III). After some time the entire cell becomes highly refractile (Stage IV). Sometimes Stage III –IV is episodic: the dying cell appears to be approaching Stage IV but then reverts to Stage III or even Stage II for a while before eventually proceeding to Stage IV. After a few minutes the refractility of the peiimeter diminishes; the central region remains refractile for longer but eventually shrinks (Stage V) and disappears.

The ultrastructure of the ventral nerve cord of the LI larva (Figs. 3–13) is essentially the same as previously described for the adult C. elegans (White, Southgate, Thomson & Brenner, 1976). It consists of a bundle of nerve fibres which run longitudinally down the ventral mid-line of the nematode alongside a ridge of hypodermis. The nervous and hypodermal tissues are enclosed within a basement membrane which separates them from the muscle cells.

The cytoplasm of nerve cells is granular and contains mitochondria and sparse endoplasmic reticulum. There are scanty aggregates of chromatin distributed throughout the nucleoplasm and there is a small nucleolus. The hypodermal cytoplasm is granular, rich in mitochondria and endoplasmic reticulum, and shows a prominent Golgi complex; the nucleus is large and contains a large nucleolus with scanty aggregates of chromatin throughout the nucleoplasm.

Figures 3 and 4 show the birth of anterior/posterior daughter cells respectively from the parent cell, which is in telophase (Stage I); serial sectioning reveals an intercellular bridge containing spindle tubules still connecting the daughters (Fig. 5). The internal ultrastructure of the two daughter cells appears normal, the only difference being that an arm of hypodermis is extending around the daughter cell which is programmed to die (Fig. 4).

After division (Stage II), the two daughter cells (Figs. 6 and 7) still show no difference in the structure of the nucleus and cytoplasmic organelles ; however, there are now thin parallel processes of hypodermal cytoplasm surrounding the dying cell (Fig. 7). The anterior daughter cell (which in this case is a putative motor neuron) can be seen extending an arm of cytoplasm into the region of the nerve fibres - presumably the formation of a nerve axon; this is not seen in the dying cell.

During Stage III (Fig. 8) the cytoplasm of the dying cell appears condensed and the nuclear envelope is dilated. Nuclear chromatin forms granular aggregates, notably underlying the nuclear membrane, and in addition, a cluster of electron-dense coarse particles appears in the centre of the nucleus, probably representing an altered nucleolus. Sections through the posterior part of the dying cell (Figs. 9–12) reveal that it splits into membrane-bounded fragments, surrounded by what appears to be an arm of hypodermal cytoplasm separate from that which surrounds the major portion of the dying cell.

The dying cell in Stages IV-V (Fig. 13) shows an increase in whorling of internal and plasma membranes, accompanied by irregularity of the cellular outline. The mitochondria appear distorted and are frequently within autophagic vacuoles, and there is a reduction of cytoplasmic granularity. The nuclear membrane becomes intensely convoluted and it is difficult to be certain of its continuity in serial sections. However, some sections demonstrate prolific budding of structures bounded by nuclear membrane and in addition, dense chromatin-like material is evident. Again, membrane-bounded fragments are seen in the posterior part of the dying cell and these are enclosed within the same arm of hypodermis as the rest of the dying cell.

As the dying cell shrinks (Stage V) there is complete breakdown of the nuclear membrane and condensed chromatin-like fragments are still seen within the cytoplasm. The latest recognisable appearance of the dying cell is one or more whorls of membranes, sometimes interspersed with fragments of electron-dense material, within vacuoles in the hypodermis.

We have described ultrastructural changes occurring in programmed cell death in the developing nerve cord of C. elegans. The morphological sequence of events implies that the hypodermis recognizes and surrounds the daughter cell programmed to die almost immediately after birth of such cells at mitosis, from the time the parent cell is in telophase and thereafter; either one or several parallel arms of hypodermis extend around the dying cell and any membrane-bounded fragments which are formed. Ingestion then occurs until all that remains within the hypodermis is a collection of membranous whorls interspersed with condensed chromatin-like remnants. The surviving daughter cell shows no such recognition by the hypodermis and after division has normal ultrastructure. In transverse sections through the whole nematode, membranous whorls are also seen within the lateral hypodermis; these could be remnants from other dying cells. It seems likely therefore that the hypodermis is specialized in phagocytic activity, in concordance with published descriptions of its content of esterase and acid phosphatase and its high metabolic activity (Bird, 1971).

Certain ultrastructural features observed in cell death in the nematode are common to apoptosis (Kerr et al. 1972; Wyllie et al. 1980) the characteristic mode of programmed cell death so far described in living tissues in mammalian, avian, amphibian, reptilian and insect species studied; such features include initial condensation of cytoplasm, nuclear chromatin aggregation, and fragmentation of the dying cell into membrane-bound fragments and the early recognition by neighbouring phagocytic cells. Other morphological features of cell death in the nematode, such as whorling of internal and plasma membranes and autophagic vacuolation are atypical of apoptosis, in which there is also a more pronounced degree of chromatin condensation. It is perhaps relevant that cell deletion in the normal adult planarian (Bowen & Ryder, 1974) also involves membrane whorling in dying cells though to a lesser extent than in the nematode system ; it is thus probable that lower invertebrates show a pattern of programmed cell death differing in detail from that described in living tissues in higher animals.

Neuronal cell death has been observed in other systems, for example with normally degenerating cervical motor neurons and induced cell death in peripherally deprived lumbosacral neurons in the embryonic chick spinal cord (O’Connor & Wyttenbach, 1974) and embryonic neuron cell death in peripherally-deprived chick ciliary ganglia (Pilar & Landmesser, 1976). In these systems, neurons appear to compete for synaptic sites on peripheral target organs; those that fail to make synapses subsequently die. This, however, cannot be the selection mechanism involved in the nematode as the dying cell does not appear to form a nerve process. Thus the trigger of programmed cell deletion in the developing ventral nerve cord of C. elegans has still to be determined. It may involve contacts with hypodermal cells oi interneurons that touch the cell body directly, or may be an intrinsic property of the deleted daughter cell which is specified earlier in development.

A.M.G.R. was supported by a grant from the Cancer Research Campaign to Professor Sir Alastair Currie. We thank Dr J. Sulston for advice, Dr R. Horvitz and Dr A. Wyllie for helpful discussion, and Mr S. Mackenzie for technical assistance. Transmission electron micrographs were taken in the MRC Clinical and Population Cytogenetics Unit, Western General Hospital, Edinburgh, and we acknowledge the help of Professor H. J. Evans and Mr A. Ross.