The Implications of Rod-Dependent Cone Survival for Basic and Clinical Research

Many inherited retinal degenerations are characterized by an initial rapid period of rod photoreceptor death. Although we still do not understand the complete pathways leading to execution of apoptosis, their destruction is perhaps not surprising, because in most cases in which the gene mutation has been identified, it involves loss or compromise of function of rod-specific proteins. Therefore, in human retinitis pigmentosa (RP), a heterogeneous family of heritable blinding diseases for which there is currently no cure, many mutations have been identified in genes coding for functional (phototransduction) and structural proteins. ^{1 2 3 4 5} Animal models of inherited retinal degeneration show very similar causes. The retinal degeneration (rd) mouse is a naturally occurring mutant strain exhibiting mutations in the gene coding for the β subunit of rod cyclic guanosine monophosphate (cGMP) phosphodiesterase, ⁶ as is seen in some forms of human RP. In this animal, rod degeneration is rapid and practically complete within 1 month of postnatal age. ⁷ The retinal degeneration slow (rds) mouse strain exhibits defects in the gene coding for rds/peripherin, ⁸ as observed in human RP. Rod degeneration in this case is much slower than in the former mutant, occurring over many months rather than days. ⁹ Nevertheless, rods eventually die by apoptosis. ¹⁰ Many transgenic strains have been prepared in which abnormal rhodopsin genes have been inserted, resembling the different forms of human RP. These include mice, ¹¹ rats, ¹² and pigs, ¹³ and rod cells invariably degenerate and die as a result of the mutation. 

Yet, in all cases for which data are available, one observation of paramount importance has no current explanation: Subsequent to the initial phase of rod cell loss there is a second wave of cone cell death. Examination of rod and cone dysfunction in 18 different human rhodopsin mutations has demonstrated that cone loss is spatially and temporally correlated with that of rods. ¹⁴ The rd mouse exhibits delayed cone degeneration after rod death. ⁷ Transgenic mice in which rod photoreceptors are ablated with toxic transgenes show secondary cone defects. ¹⁵ Transgenic pigs containing mutant rhodopsin genes reveal cone destruction paralleling that observed in rods. ¹³ Why should these cells, which for the most part are not the population harboring the defective gene (but see later discussion) also degenerate? The answer to this question is of fundamental importance to vision research biologists and clinicians. For the biologist it raises the possibility of rod–cone interactions playing a vital role in coordinating photoreceptor development and survival. For clinicians it is the secondary cone death in RP that accounts for the most debilitating aspect of vision loss. Whereas rod breakdown results in night blindness and restriction of the visual field, cone destruction leads to disappearance of central vision and renders the affected person unable to distinguish colors or details. 

It is easy to imagine one of at least two non–mutually exclusive scenarios to account for this delayed cone loss. In the first, rod breakdown would adversely affect neighboring cones through nonspecific environmental influences. The progressive disintegration of the surrounding far more numerous rods may leave the cone outer segments exposed, say, to toxic concentrations of neurotransmitters. Although rods die by apoptosis, which normally avoids release of potentially toxic cellular metabolites that would create problems of poisoning or inflammation and therefore prevents extension of cellular breakdown to surrounding regions of tissues, it seems nevertheless possible that degeneration of the abundant rods (∼20 times more numerous than cones in many mammalian species, including man) could exert adverse effects on the adjacent cones. However if rods were releasing general toxic biproducts, other nearby cell types, such as the immediately postsynaptic partners of photoreceptors, the bipolar cells, would be expected to die as well. Although there are some modifications of inner retinal neurons after photoreceptor loss, they do not undergo widespread death. Furthermore, each cone is individually surrounded by an insoluble glycocalyx that forms a privileged structural microdomain linking each cone to a retinal pigment epithelial (RPE) cell. 

A second explanation, which if true could have far-reaching implications, is that rods produce some kind of signal that is essential for maintaining cone viability, so that the disappearance of rods for whatever reason would deprive the cones of this signal and trigger their degeneration. This idea may seem far fetched, but both circumstantial evidence and recent experimental findings suggest that not only do such rod–cone interactions exist, but also that we can intervene to limit or prevent secondary cone death. 

Although both types are involved in transducing light energy into electrical signals relayed to the second- and third-order retinal neurons, many aspects of rod and cone biology differ. In most species cone cells are among the first retinal cell types to leave the cell cycle, whereas rods are generally among the last to do so. ¹⁶ Evidence suggests cones may organize the photoreceptor mosaic through inducing differentiation of neighboring uncommitted precursors. ¹⁷ But other data indicate rods may trigger the onset of cone functionality. Although cones exit the cell cycle early in development of the bovine retina, the transcriptional levels of two cone-specific messenger RNAs (red cone opsin and blue cone γ subunit of cGMP phosphodiesterase) remain uniformly low for many weeks. It is not until rod photoreceptor precursors stop dividing much later in embryogenesis and begin their own differentiation (increased transcription of rod opsin and rod γ subunit of cGMP phosphodiesterase mRNAs) that cones follow suit, ¹⁸ as though having waited for some inductive or permissive signal. Although such interactions may have nothing to do with rod-dependent cone survival, they suggest that cones are responsive to rod-derived stimuli. 

Two independent groups have demonstrated that photoreceptor degeneration is non–cell autonomous. Chimeric mice retina composed of both normal and rhodopsin mutant cells display uniform rather than patchy degeneration, ¹⁹ and hemizygous female rds mice in which the normal gene had been inserted into the X chromosome showed uniform degeneration rather than the expected random mosaic. ²⁰ Although cone degeneration was not addressed in these studies, both groups evoked possible trophic influences to explain the results. In another animal model of retinal degeneration, the Royal College of Surgeons (RCS) rat, photoreceptors die subsequent to a gene defect located in the RPE, which results in defective phagocytosis of shed outer segments. ²¹ Again, two independent groups working with chimeric animals composed of normal and dystrophic cells ²² and after transplantation of patches of normal RPE into RCS rat eyes ²³ suggested that additional trophic influences may have been operating because photoreceptor survival extended beyond boundaries of normal RPE. 

How can an experiment be developed to test whether rods could influence cone survival? In fact, the pattern of photoreceptor degeneration in animal models provides a ready-made system in which to test this hypothesis. In the rd mouse the rapid disappearance of rods (>99.5% die between the 10th and the 30th days after birth) and the slower time course of cone loss (∼30% remaining at 1 month) results in a time window during which this strain possesses effectively a pure-cone retina. If at this time fragments of photoreceptor layers isolated by planar vibratome sectioning from normal sighted mice (and therefore ∼97% rod pure) are transplanted into their subretinal spaces, statistically significant increases in the numbers of cones surviving in the host central retina compared with sham-procedure controls are observed. ²⁴ To determine whether such a trophic effect may be contact mediated or due to diffusible molecules, we devised a coculture system in which target rd mice retinas were separated from test populations by a semipermeable membrane. Only in cases in which rod photoreceptors were present in the connecting chamber were higher numbers of cones observed in the target retina compared with retinas of age-matched controls. ²⁵ These data thus suggest this effect is mediated by diffusible substances and that furthermore the increased contingent of host cones cannot arise by migration from the graft. 

Finally, returning to the earlier in vivo paradigm, we recently showed that isolated layers of photoreceptors but not inner retinal neurons significantly enhance cone survival throughout the host rd retina, again implicating long-range diffusible effects. ²⁶ Furthermore, the magnitude of the rescue effect—some 40% of cones that would normally die over the experimental period were preserved after transplantation of rod photoreceptors—is impressive, given the small size (<10% of the host retinal surface) of grafts. 

These simple cell biologic studies raise several questions. They argue that the cause of secondary cone death is not release of toxic factors from sick and dying rods, because transplantation involves only small tissue fragments, and toxic components would have been diluted in the culture medium. They suggest that this aspect of rod–cone interactions is controlled by diffusible factors, although at the present time we have no idea of the identity of such molecules. Obviously, the search for such substances becomes of great interest for both biology and medicine. They also demonstrate that cone death is not inexorable and that intervention during the period after massive rod loss but preceding cone degeneration could preserve or extend the useful lifetime of residual visual capacities. This will be of considerable importance for the clinic, where many cases of RP are diagnosed after visual deficits have occurred. 

Naturally, many questions are currently unanswered, and this apparently simple scheme is doubtless far more complicated. Are the additional surviving cones functional? How long does the survival-inducing effect last? Could such an approach be valid in other species and other types of gene mutation? Is the pattern of rod and cone death really compatible with the notion of diffusible factors? Is the survival-promoting activity derived directly from rods or through their stimulating another retinal cell type? How can cone death be explained in other forms of retinal degeneration in which cones die either exclusively or before rods? How can cone survival in cone-dominant species be explained? 

The answers to the first three questions are all directly verifiable by experimentation. Similar studies in large animal models such as rats, dogs, or pigs would reduce surgical damage and facilitate functional and behavioral testing. Although they will gradually die, many cones survive for extended periods subsequent to the disappearance of rods. Is such a scenario compatible with the classic notion of trophic deprivation, which generally leads to rapid death? Without prior identification of the active substance(s), or of knowledge of the nature of cone death (it is not yet known whether cones actually die by apoptosis), this is difficult to answer for the present. One potentially important difference between rod–cone interactions under consideration here and those paradigms in which trophic deprivation and neuronal death have been most well studied (e.g., withdrawal of nerve growth factor from sympathetic neurons ²⁷ ) concerns the developmental age. Apoptosis has been intensely studied with respect to elimination of surplus cells during normal development, but adult neurons are thought to be more resistant, related in part to changes in regulation or expression of pro- and/or antiapoptotic proteins. ^{27 28} By analogy, retinal ganglion cell death in glaucoma also occurs with a slow time course but is proposed to be due in part to trophic factor deprivation. ²⁹ The fifth question concerns the mechanism of action: Is the effect direct or indirect? We have no current evidence that the cone-survival effect is mediated directly by rods, although cultures of purified rod photoreceptors ³⁰ are undergoing testing in this sense. Because the nonneuronal types bordering the photoreceptors, the RPE, ³¹ and Müller glia ³² are both known to influence photoreceptor survival, such an indirect route is certainly possible. Regarding the sixth point, there are of course other ways in which cones can die. When they are the targets of gene mutations themselves, such as in cone-rod dystrophies ³³ or Leber’s congenital amaurosis, ³⁴ provision of normal rods may not be of any use. The same would be true for diseases in which the RPE is the seat of the gene defect, ³⁵ so that rod transplantation would be ineffective (but transplantation of normal RPE could be beneficial). Finally how could cones survive in cone-dominant species such as ground squirrels or chickens? Maybe in some species cones make their own survival factors, or these are provided by the RPE or Müller glia. 

In conclusion, the evidence is still far from complete, but our findings constitute one of the only proofs that transplantation of photoreceptors into diseased retinas may actually be of any benefit. In neural transplantation in general, such procedures fall into one of two categories: one in which the graft should integrate precisely with the recipient tissue, re-establishing synaptic contacts and reconstructing functional circuits, and one in which transplants represent reservoirs of neurotransmitters or trophic factors absent from the recipient. Our results support the second approach to retinal degeneration, in which these treatments are stop-gap procedures, a means to an end in awaiting the eventual purification and identification of the active component(s). 

There are additional implications. Because neuroprotection of rods in diseased retinas assures a continued supply of trophic factors to the cones, therapies currently intended to limit or preventing rod loss through gene therapy, ¹² growth factor application, ^{36 37} or pharmacology ³⁸ would have indirect beneficial effects on cone survival. Finally, some data suggest rods may be causally involved in age-related macular degeneration, the leading cause of blindness in the aging population in the Western world and a major cause for health concern. Histopathologic studies reveal that initial cell losses occur in the rod-rich perifovea rather than in the macula itself, ³⁹ and mutations in the gene coding for the rod-specific integral membrane protein ABCR ⁴⁰ have been described in some patients with age-related macular degeneration. ⁴¹ Therefore, identification of rod-derived cone survival factors could be important in this condition as well. 

 Submitted for publication June 1, 1999; revised July 13, 1999; accepted July 15, 1999.

 Commercial relationships policy: N.