Neuron, Vol. 5, 373-381,September,1990,Copyright © 1990by Cell Press VIP-Mediated Increase in cAMP Prevents Tetrodotoxin-lnduced Retinal Ganglion Cell Death In Vitro Peter K. Kaiser and Stuart A. Lipton Laboratory of Cellular and Molecular Neuroscience Departments of Neurology Children's Hospital Beth Israel Hospital Brigham and Women's Hospital Boston, Massachusetts 02115 Program in Neuroscience Harvard Medical School Boston, Massachusetts 02115 Summary Afferent influences on natural cell death were modeled in retinal cultures derived from neonatal rats. Tetro- dotoxin (TTX) blockade of electrical activity produced a significant reduction in surviving retinal ganglion cell (RGC) neurons during a critical period of development, similar in magnitude to the reduction observed during natural cell death in the intact retina at a similar de- velopmental stage. The addition of vasoactive intestinal peptide (VI P) protected the RGCs from the lethal action of TTX. This effect was specific, since the related pep- tides PHI-27 and secretin produced no significant in- crease in RGC survival. Radioimmunoassay of cyclic nucleotides showed that TTX decreased culture levels of cAMP and that this trend was reversed by VI P. Decreases in RGC survival associated with TTX electrical blockade were prevented by 8-bromo:cAMP or forskolin. Fur- thermore, VIPt0-28, the C-terminal fragment that in- hibits VIP stimulation of adenylate cyclase, reduced the number of surviving RGCs. Thus, our results suggest that VIP, acting by increasing cAMP, has a neurotrophic ef- fect on electrically blocked RGCs and may be an endog- enous factor modulating normal cell death in the retina. Introduction An interesting facet in the normal development of the vertebrate central nervous system (CNS) is the large proportion of neurons that die naturally (Oppen- heim, 1981). One example is found in the mammalian visual system. Overproduction of retinal ganglion cells (RGCs), neurons that project from the retina to deeper centers of the brain, appears to be a fun- damental characteristic of development (Land and Lund, 1979; Thompson, 1979; Finlay et al., 1979; Rakic and Riley, 1983, 1984; Insausti et al., 1984; Williams et al., 1986). The death of RGCs is thought to serve at least two purposes. First, the initial overproduction of RGCs ensures that the target structures receive ade- quate input and that the subsequent cell death quan- titatively matches the projecting neuronal population to the needs of their targets (for reviews see Oppen- helm, 1981; Hamburger and Oppenheim, 1982; Co- wan et al., 1984; Cowan and O'Leary, 1984). Second, it eliminates those cells whose axons project into inap- propriate regions either in the correct target fields or in entirely incorrect areas (for reviews see Clarke, 1981; Cowan et al., 1984; O'Leary, 1987). In other words, it effectively removes certain developmental miscues that cause aberrant axonal projections or neuronal targeting errors. In adult rats, essentially all of the RGCs send axons to the superior collicutus in a highly ordered projec- tion. The RGCs differentiate between days 12 and 16 of gestation, and the axons reach the contralateral su- perior colliculus as early as embryonic day 16 (E16; Morest, 1970; Bunt et al., 1983). The number of axons in the optic nerve reaches its peak at approximately E20; thereafter there is a rapid loss of RGCs (Bunt et al., 1983; Crespo et al., 1985; Williams et al., 1986). In the pigmented rat at least half of the RGCs die natu- rally, and this normal cell death proceeds for the first 1 1/2 weeks of postnatal development (Cunningham et al., 1982; Perry et al., 1983). Although the exact cel- lular sequence of events is still under debate, the loss of neurons is thought to be caused by at least two mechanisms: the intermingling and competition of RGC axonal fibers from the same eye and from the two eyes for their postsynaptic targets du ring the early stages of development (the efferent influence; Cowan et al., 1984; Fawcett et al., 1984; Rakic and Riley, 1984; O'Leary et al., 1986b; Sretavan and Shatz, 1986; Wil- liams et al., 1986), and the effect of local factors in the retina from cells presynaptic to the RGCs (the afferent influence; Perry and Linden, 1982; Cunningham, 1982; Okado and Oppenheim, 1984; Clarke, 1985; Lipton, 1986; Furber et al., 1987). Several recent findings have shown that competition, and the consequent cell death, are centered around the blockade of electrical activity during a critical period of development (Co- wan et al., 1984; Fawcett et al., 1984; Lipton, 1986; O'Leary et al., 1986b). Previous experiments in this laboratory studied the effects of electrical blockade on identified RGCs dis- sociated from retinas of neonatal rats (Lipton, 1986). On the second day in culture, 10%-15% of the RGCs were found to be solitary neurons; the remainder ex- isted in small clusters with other retinal cells. Electri- cal recordings determined that 50% of the clustered RGCs displayed spontaneous postsynaptic potentials and action potentials, whereas the solitary RGCs did not. Furthermore, the same population of clustered RGCs that exhibited spontaneous activity died within 24 hr of exposure to either tetrodotoxin (-I-rx) or Iow- Ca2+/high-Mg 2+ medium to block synaptic activity. Similar to natural cell death, these effects occurred only if the RGCs were at a particular stage of develop- ment (postnatal day 2 to 10 [P2-P10] animals, but not older) when they either possessed or were acquiring electrical activity. Cell death was not found in the soli- tary RGCs that lacked synaptic input and consequent