Our previous work in embryonic chick retina indicated that the mechanisms of excitotoxic neuronal death in that tissue were independent of calcium and instead were dependent most critically on chloride.
19 This does not appear to be the case in adult mammalian amacrine cells. Based on our pharmacological experiments, we propose that Na
+ overload is the initial step in lethality, followed by excess Ca
2+ entry due to Na
+ extrusion/Ca
2+ influx through reverse operation of the Na
+-Ca
2+ exchanger.
There are K
+-independent and -dependent Na-Ca exchangers: the NCX and NCKX families, respectively.
25 Because the extent of protection provided by KB-R7943, a selective inhibitor of the reverse (Ca
2+ efflux) mode of NCX, was comparable to that caused by removal of extracellular Ca
2+, it seems likely that NCKX is not involved. A Na-Ca exchanger has been demonstrated to be present in amacrine cell dendrites, where it is critically important in removing calcium after spiking and thereby shaping amacrine cell responses.
26 27
Ca
2+ entry via reverse operation of the exchanger after a sodium load is an important mechanism of disease after ischemia. This has been shown in both the spinal cord,
28 29 and the cerebrum,
24 30 31 in vivo and in vitro, as well as in non-neural tissues, such as the heart
32 33 34 and the kidney.
35 36
Three different Na-Ca exchangers have been cloned, NCX1, NCX2, and NCX3.
37 38 39 Localization studies in the retina have not been performed, and so we cannot identify with certainty which one is responsible for mediating the lethal Ca
2+ entry into amacrine cells. However, indirect evidence points to NCX2. First, NCX2 is the predominant form in adult rodent brain, with mRNA expressed at a level an order of magnitude higher than NCX1 or NCX3
40 41 (although all isoforms are present and differentially distributed in the brain
42 ). Second, the pharmacology suggests NCX2. KB-R9743 has a similar, or slightly higher affinity for the NCX2 exchanger than the NCX1 exchanger expressed in BHK cells,
43 44 whereas SEA0400 has a much higher affinity for NCX1 than KB-R7943 does.
45 Because both SEA0400 and KB-R7943 are effective, but SEA0400 is not more potent than KB-R7943
(Figs. 8 9) , this is consistent with NCX2’s mediating lethal Ca
2+ entry in adult mouse retinal amacrine cells after addition of KA.
Adult mouse ganglion cells were insensitive to excitotoxic-agonist–induced death. This is an example of how cellular physiology and pathologic susceptibilities are developmentally regulated, and why neonatal cell culture is a poorly suited model for investigating the underlying bases of adult disease.
This insensitivity of adult mouse ganglion cells to excitotoxic insults is perhaps surprising but is consistent with results in adult pig cultures.
16 Intravitreal injection of glutamate agonists has been described to kill at least some ganglion cells in chicks
46 and rodents.
6 7 47 48 We cannot exclude the possibility that this insensitivity is an artifact of the dissociation. Indeed, we have observed (Luo X, Romano C, unpublished results, 2003) that the transcription factor brn3a, present in the nuclei of most ganglion cells in vivo,
49 is gradually lost from NF-positive ganglion cells in culture. An analogous loss of receptors, exchangers, or other critical molecules responsible for the in vivo sensitivity of ganglion cells to excitotoxins may occur and thereby alter the sensitivity of the cells to glutamatergic agents.
However, there is reason to believe that this developmental difference is not merely due to dissociation and culture. First, we observed this difference between identically prepared cultures from neonatal and adult retinas. If the observed difference were an artifact, this artifactual process would itself have to be subject to developmental regulation. Second, our findings are consistent with the results using intact explant cultures obtained by Izumi et al.
50 In this study, rat retinas of various ages were acutely explanted and exposed to NMDA, and toxic effects were noted histologically. Little or no toxicity was seen in the youngest retinas examined (postnatal day 0), but the toxic effect increased with development, peaking near day 9. NMDA-induced damage and then its level declined, and adult retinas showed almost no response to NMDA. These data, taken along with ours, indicates that there is a profound developmental regulation of the direct toxic action of glutamate receptor overactivation on ganglion cells.
What then is the explanation for the effectiveness of excitotoxins in vivo? Perhaps the exposure time is longer than 24 hours in these cases. Alternatively, the agonists may be having an indirect action, requiring other ocular structures to participate in the toxic mechanism, structures not present in either explant or dissociated retina cultures. The cellular and molecular differences that account for the difference between adult and immature neurons and between amacrine and ganglion cells therefore remain unknown and are topics for future investigations.