the eye is sufficient to support vision, but retinoid conversion reactions are essential for the economy of retinoid use. All the retinoids in the eye result from dietary uptake, which are stored in the liver, and a small fraction is delivered to the eye each day through serum delivery attached to the serum retinol binding protein (SRBP). Hence, the reconversion of ATR back to 11cRal is essential for the maintenance of vision on short time scales. There are two rate limiting steps that have been proposed for the visual retinoid cycle: (1) the reduction of ATR to VA, and (2) the hydrolysis of ATR from the ligand-binding pocket (bleaching of visual pigment). Work in the rod dominant mouse model has suggested that tRDH reduction of ATR to VA is the rate limiting step in visual pigment regeneration in the rod photoreceptor [57]. A more recent model proposed by Lamb and Pugh [58, 59] suggests that the coupled hydrolysis of the Schiff base and reloading with 11-cis-retinal is the rate limiting step. In either case, ATR is known to move to the outside surface (extracellular) of the rod visual pigment after hydrolysis. On the outside surface (extracellular surface or inside disk membrane surface), ATR is free to covalently react with membrane proteins or PE. ABCR must move ATR back to the cytoplasmic surface quickly to prevent covalent adduct formation of ATR with PE. Any delays, for example, due to mutations in ABCR or loss of functional WT ABCR from the outer segment environments will slow the rate of ATR transfer back into the cell and increase the probability of reaction between ATR and PE. Thus, ABCR can become rate limiting when its level of WT function (e.g. Vmax in the Michaelis–Menten enzymatic sense) in the photoreceptor is compromised by genetic mutations.

Studies have investigated the lipid composition of the photoreceptor outer segment membranes. PE is abundant in both the rod and cone outer segments. PE appears to be symmetrically distributed on both leaflets of the rod outer segment disk membranes [60–62]. As such, the probability of forming N-retinylidene-PE would appear to be equal on both sides of the membrane provided the appearance of ATR was also equivalently distributed. However, the movement of ATR out of the opsin apoprotein core after bleaching hydrolysis may be vectorial and not random. Evidence suggests that in addition to the active site of opsin where 11-cis-retinal forms a covalent link with the protein, there are additional entrance and exit sites for retinoids within rhodopsin and that the entrance of fresh

11-cis-retinal occurs independently of the exit of ATR. It appears that the exit of ATR from the protein does not involve solubilization within the hydrophobic environment of the membranes. In contrast, the opsin protein is the channel that guides ATR release from the protein­ . In the current model, both the entrance and exit sites for retinoids to and from the active site are on the cytoplasmic­ surface of the protein [63, 64]. This would facilitate the interaction with photoreceptor all-trans-retinol dehydrogenase (RDH8) which is essential to reduce ATR to vitamin A for returning back to the RPE. Under conditions of high light and excess bleaching,­ this process may become oversaturated with the ATR being released from the exit site prior to its reduction­ by tRDH. As all retinoids are highly hydrophobic, ATR would be free to diffuse in the membrane lipid bilayer and form covalent adducts with PE or with primary amines (lysines) on integral or membrane­ -associated proteins. On the other hand, if the entrance site for 11cRal is on the cytoplasmic side of the protein and the exit site for ATR is on the extracellular side of the protein, then all ATR would be deposited on the extracellular leaflet of the membrane (intradiscal in rods) and the external membrane surface of both the rods and cones, leading to an explicit and critical need for ABCR in the dark adaptation cycle. The severity of retinal degenerations (RP19) that result from a complete lack of ABCR in the photoreceptors provides impetus to explore this alternative model.Regardless of which side of the membrane upon which the visual pigment ATR is deposited upon bleaching and hydrolysis, it is clear that ABCR is a critical player in the retinoid visual cycle.

11.2.3.2  ELOVL4

ELOVL4 is an enzyme initially proposed to be involved in the synthesis of very long-chain fatty acids (VLCFAs) on the basis of bioinformatics homology to similar proteins expressed in yeast. It is expressed abundantly in both the rod and cone photoreceptors where it localizes to the inner segments [6, 7, 17] (Fig. 11.12a). ELOVL4 is also expressed less abundantly in the lens, brain, skin, and testes [65]. Fatty acids are synthesized in the smooth endoplasmic reticulum and must traffic in vesicles from the inner segments of photoreceptors to the base of the outer segment where they engage in the formation of new disk membranes. It has long been established that long-chain (C18–C22) polyunsaturated fatty acids

Fig. 11.12  ELOVL4 cellular localization and function.

(a) ELOVL4 is localized to the microsomal membranes in the inner segments of both the rod and cone photoreceptors.

(b) Molecular schematic of

ELOVL4. ELOVL4 appears to perform very long-chain fatty acid elongation through the addition of two carbon Co-A substrate species (product of the Krebs cycle) to an elongating chain enzyme-trapped product. (c) Precursor lipids for ELOVL4 appear to be EHA and possibly DHA

a

ROS

= ELOVL4

= lipid

COS

CIS

Cone

Synapse

(LCPIFA) are essential to the high levels of membrane fluidity that govern the process of phototransduction in photoreceptor outer segments [60, 66]. One of the main components of the retinal membrane lipids containing LCPUFAs is dietary docosahexaeneoic acid (DHA), a polyunsaturated long-chain fatty acid which is known to be present in abundance (33–50% of fatty acid content) in the photoreceptor outer segments, 16–22% in the general retina, and is important for membrane fluidity and visual function [66–70]. This leads to the initial hypothesis that ELOVL4 was tied to the synthesis of

membrane lipids containing DHA, or that DHA was used as a substrate in even longer chain synthesis [17]. Recent and prior studies are now beginning to demonstrate that ELOVL4 is involved in the synthesis of very long-chain (C24–C36) polyunsaturated fatty acids (VLCPUFA) (Fig. 11.12b). Complete deficiency of ELOVL4 promotes perinatal lethality on the basis of profound dehydration that results from the absence of VLCPUFAs in the skin surface [71–76]. Such lipids have wax-like qualities and would be expected to function in such a role to protect against dehydration in a

CoA CoA

CoA

CoA

COOH

membrane

dessicating environment. Two dermatological studies of ELOVL4 deficiency suggest that ELOVL4 promotes the elongation of FA chains (saturated and unsaturated) beyond the C26 chain length stage, given that fatty acids longer than C26 were absent in the dermis [71–73]. ELOVL4 deficiency was recently shown to cause a specific loss of C32–C36 acyl-phosphatidylcholines (PC) in the mouse retina [74, 75]. The formation of C32–C36 fatty acids in the retina and photoreceptors was recently shown to be dependent upon a dietary substrate called eicosapentaenoic acid (20:5(n–3)) (EPA) and not DHA (22:6(n–3)) (Fig. 11.12c) [77]. A recent study showed a statistically significant inverse relationship between the

phenotypic severity of STGD3 due to an ELOVL4 mutation­ and the level of adipose tissue EPA levels (reflecting an intake history of over 2–3 years), whereas a significant inverse relationship between the severity and the levels of both EPA and DHA in red blood cell lipids (reflecting an intake history of over 6–8 weeks) was observed [78]. Since the American diet is relatively deficient in DHA and EPA, requiring more to be synthesized in vivo, this study suggests a potential role for dietary supplementation in ELOVL4-mediated STGD3 disease. Normal dietary intake of omega fatty acids may also be an environmental factor that influences phenotypic diversity within a single pedigree with ELOVL4

STGD3 disease. Supplementation­ with DHA in a single autosomal dominant STGD3 patient with an ELOVL4 mutation demonstrated a slight positive beneficial effect with improved subjective and objective visual function parameters [79]. While a mutational impairment of ELOLV4 does not appear to play a role in DHAdependent very long-chain fatty acid synthesis, the question of how the presence of excess dietary DHA may be beneficial arises. In polyunsaturated PC acyl glycerolipids, the sn-1 position is occupied by the VLCPUFA and the sn-2 position is commonly occupied by DHA. Excess dietary DHA could help to drive the formation of the desired product that embodies DHA, in part, along with a component likely to be synthesized through ELOVL4 activity in the photoreceptors. Since the membrane fluidity can be obtained with polyunsaturated lipids of much shorter chain length, there must be a highly specific role for VLCPUFs in the photoreceptors. As a testament to its importance ELOVL4 expression is highly conserved in the vertebrate retina [80]. Prior work has shown that VLCPUFs, mostly in the form of PCs with the VLCPUFAs in the sn-1 position and DHA in the sn-2 position of glycerolipids, are tightly associated with rhodopsin and are not extracted from rhodopsin with hexane [81, 82]. The importance of such VLCPUFAs in the retina is indicated because of their abundance and the rapid rate of synthesis compared to the long half-life of retinal lipids [83]. There appears to be a local collar of PCs with VLCPUFs surrounding the rhodopsin molecules that acts to alter the local membrane biophysical environment to support rhodopsin activation or regeneration in as yet undetermined ways [84]. Studies have shown that rhodopsin activation is associated with a local membrane lipid microenvironment [61, 62, 85]. High membrane fluidity is critical to the efficient formation of the biochemically active intermediate conformation of light-activated rhodopsin called Meta-II [86–90]. A saturated lipid microenvironment hinders the formation of the Meta-II state [87].

The mechanism of the autosomal dominant mutations in ELOVL4 are now easier to understand from the perspective of their effects on the visual pigment ground state stability in the membrane, kinetics of activation and regeneration. All known mutant alleles of ELOVL4 occur in exon 6 (final) and produce protein truncations lacking the di-lysine ER-retention signal. Such mutants, even if the enzyme component were still active, would

prevent the proper cellular localization (ER) for very long-chain fatty acid synthesis. However, studies have shown that such a haploinsufficiency effect is not the cause of STGD3 (Li et al. 2007; [91]). Rather, these truncated and likely enzymatically inactive mutant proteins exert dominant negative functionality as they promote the mislocalization of the remaining functional WT protein away from the ER in the photoreceptors [92–97]. The remaining WT protein appears to be unable to function enzymatically in the mislocalization environment(s) and this promotes the deficiency of the VLCPUFAs which are essential to rhodopsin function. There is evidence in yeast that homologous enzymes (SUR4) function to generate membrane lipids that are critical to support mammalian integral membrane ion channel function [98]. Therefore, the ELOVL4 phenotypic impact is likely to extend beyond rhodopsin to include critical biophysical support of other integral membrane proteins in the photoreceptors where ELOVL4 is expressed. While considering how ELOVL4 mutations could promote the increase in LF, FAF, and A2E that is known to occur in STGD3, it is tempting to speculate that ELOVL4 is in some way supportive of ABCR function. For example, a local collar of PC containing lipids with VLCPUFAs could provide the needed local biophysical environment for flipase function. This hypothesis is directly testable in cell culture and animal model systems. In fact, an earlier study reconstituted ABCR in synthetic liposomes made from dioleoylPC (C18:0, C18:0), 1-stearoyl-2-docosahexaenoyl-PC (C18:0, C22:6), dioleoyl-PE, or 1-stearoyl-2-docosa- hexaenoyl-PE and found that the PE containing phospholipids enhanced ABCR ATPase activity while, surprisingly, the PC containing lipids were inhibitory [99]. This result suggests the hypothesis that PC-based lipids with VLCPUFAs in the sn-1 position may be required for ABCR activity. More work is needed to further elucidate the function and role of ELOVL4 in photoreceptor metabolism.

11.2.3.3  PROM1

Prominin is expressed to the base of the outer segment plasma membrane of both the rods and cone photoreceptors. There it appears to play a role in the membranous evagination process that forms the beginning of new disks. Its membrane localization, extracellular