8.1Introduction

Rod and cone photoreceptor cells are photosensitive; they detect the presence of photons through the 11-cis retinal chromophore bound to opsin proteins. Light isomerizes 11-cis retinal to all-trans retinal, which then dissociates from opsin. This photoisomerization is the initial event that triggers the visual transduction pathway, activation of second order neurons, and eventually transmission of the signal to the brain. Under constant illumination, 11-cis retinal needs to be replaced and all-trans retinal needs to be removed from the vicinity of opsin so that photoreceptor cells continue to have optimum sensitivity to light. Retinol dehydrogenases (RDHs) located in photoreceptor outer segments and in retinal pigment epithelium (RPE) cells participate in these very important functions [1]. Surprisingly, several RDHs expressed in photoreceptor cells are not located in the outer segment but in the inner segment only. Inner segment RDHs are RDH11 [2, 3], RDH12 [4Ð6], and RDH13 [7]. Their physiological role(s) are not completely understood. We propose the hypothesis that inner segment RDHs play a role in the detoxiÞcation of polyunsaturated fatty acids (PUFAs) oxidation products. These toxic products are easily generated in photoreceptor cells and have to be reduced to keep the cells alive and functional.

8.2Subcellular Localization, Expression Levels in the Retina, and Substrate Specificity

8.2.1Subcellular Localization

To determine the physiological role of an enzyme in vivo, it is important to determine its localization, relative to that of its possible substrates. RDH11, 12, and 13 are located in the inner segments of rod and cone photoreceptors. RDH11 and 12 are integral membrane proteins and thus cannot be extracted from membrane fractions by alkaline treatment [2]. Both enzymes are inserted in the membrane through a stretch of ~20 amino-terminal hydrophobic residues [2, 7]. RDH13 does not contain this stretch of hydrophobic residues and is a peripheral membrane protein, as demonstrated by the fact that it can be extracted from membrane fractions by alkaline treatment [8]. In previous studies, we showed that RDH11 is localized in the Golgi apparatus in spermatocytes [2] and in various cultured cells (unpublished observation). Here, we performed a subcellular fractionation of retinal tissues through ultracentrifugation on sucrose gradient [9]. As shown in Fig. 8.1, RDH12

Fig. 8.1 Subcellular fractionation of mouse retina. Mouse retinas were pooled, homogenized, and subjected to ultracentrifugation as described [9] to prepare the total membrane pellet (Mb) and the cytosolic (C) fractions. Mb fraction was resuspended, protein concentration was measured, and 10 mg of each fraction was subjected to SDS-PAGE. The Mb fraction was further fractionated by ultracentrifugation on sucrose step gradients of 1.25, 1.1, and 0.25 M sucrose solutions as described [9]. The Golgi-enriched fraction was collected at the interphase of the 0.25 and 1.1 M sucrose solutions; the ER-enriched fraction was collected at the interphase of the 1.1 and 1.25 M sucrose solutions; and the microsome pellet (M) was at the bottom of the tube. Protein concentration was measured and 5 mg of each fraction was subjected to SDS-PAGE. Immunoblotting was performed as described [10] with anti-RDH11 and anti-RDH12 polyclonal antibodies

is found in Golgiand endoplasmic reticulum (ER)-enriched fractions and RDH11 is detected only in the Golgi-enriched fractions. Another study showed that RDH13 is a mitochondrial enzyme, localized within the intermembrane space, and associated with the inner mitochondrial membrane [8]. These distinct subcellular localizations suggest a speciÞc role for RDH13 in the mitochondria, RDH12 in the ER, and redundant functions for RDH11 and RDH12 in the Golgi.

8.2.2Expression Levels in the Retina

To determine the relative expression levels of these enzymes in the retina, we compared mRNA levels by quantitative RT-PCR. As shown in Fig. 8.2, the level of Rdh12 mRNA in BALB/c mouse retina is ~7-fold higher than the level of Rdh11 and ~200-fold higher than the level of Rdh13. In a previous study [10], using a different strain of mice, we found that the level of Rdh12 mRNA in C57BL6 was ~40fold higher than the level of Rdh11. We also compared expression levels of RDH11 and RDH12 proteins and, after calibration of our speciÞc antibodies, found that

Fig. 8.2 Quantitative expression of Rdhs in the mouse retina. Total RNAs were prepared from retinas of adult BALB/c mice raised under dim cyclic light and subjected to quantitative RT-PCR as described [10]. Each value represents the average amount of indicated mRNA, relative to the value of the housekeeping gene Rpl19, arbitrarily deÞned as 1. Error bars represent the standard error for the four samples

RDH12 was ~7-fold higher than RDH11 in C57BL6 mouse retina [10]. Taken together, these experiments show that, of these three enzymes, RDH12 is the most abundantly expressed RDH in photoreceptor inner segments.

8.2.3Substrate Specificity

RDH11, 12, and 13 are oxidoreductase enzymes that belong to the short-chain dehydrogenase/reductase (SDR) family [2, 7]. Their substrate and coenzyme speciÞcities have been evaluated in vitro. Similarly to RDHs located in photoreceptor outer segments, all three RDHs were found to reduce all-trans retinal and other retinaldehydes (in the cis conÞguration) to corresponding retinols, using NADPH as cofactor [7, 8]. An additional group of substrates was found for RDH11 [2] and RDH12 [11]. Various aldehyde-containing molecules, formed by a chain of 8Ð10 carbons and containing 0Ð2 unsaturated carbon bonds, 0 or 1 hydroxyl group and 1 aldehyde group were found to be reduced by RDH11 and RDH12 to the corresponding alcohols [2, 11]. These aldehydes are produced in cells by the oxidation of membrane PUFAs [12]. PUFAs are easily oxidized when attacked by reactive oxygen species (ROS) formed within the mitochondria as byproducts of the electron transport chain [13]. Each PUFA produces speciÞc oxidation products. For example, 4-hydroxynonenal (4-HNE; 9 carbons, 1 unsaturation, 1 hydroxyl group, and 1 aldehyde group) is an oxidation product of w-6 arachidonic and linoleic fatty acids [14]. 4-HNE is the most abundant and toxic end product of lipid oxidation found in tissues [15, 16]. In vitro studies showed that RDH11 and RDH12 can reduce 4-HNE [2, 11]. The enzymatic activity of RDH13 has not been directly tested with 4-HNE or other lipid oxidation products.

Indirect testing of the substrate speciÞcity for various aldehyde-containing molecules was performed by competition experiments. In these experiments, the

RDH of interest is incubated with a limiting concentration of all-trans retinal as the known substrate. The potential substrate being tested is added to the reaction in increasing concentrations. If a potential substrate competes with all-trans retinal (by binding to the same site), it is then considered to be a true substrate of the enzyme. When we and others performed this type of experiment with RDH12 (unpublished) and RDH13 [8], none of the aldehydes tested including 4-HNE were able to compete with all-trans retinal, even at high concentrations. This is surprising because the direct activity of RDH12 on 4-HNE has been demonstrated [11]. Absence of competition in this case could suggest that RDH12 has double substrate speciÞcity, involving two distinct substrate-binding sites. No conclusion can be drawn for RDH13 before its activity is directly tested with 4-HNE as substrate.

8.3Detoxification of 4-HNE in Cultured Cells

Shortand medium-chain aldehydes produced by oxidation of lipids have been shown to mediate oxidative damage in various pathological situations, especially in neurodegenerative diseases [17]. Because 4-HNE is the most abundant and most toxic end product of lipid oxidation and was found to be a substrate for RDH11 and RDH12 in vitro, it was important to determine if these enzymes were able to protect cells against the toxicity of 4-HNE. In a series of experiments, we generated stable cell lines, expressing RDH11, RDH12 (wild type or inactive mutant), or RDH13 and compared their sensitivities to 4-HNE-induced apoptosis. This assay allowed us to indirectly test the activity of RDH13 with 4-HNE in a way that did not involve competition with all-trans retinal. Apoptosis was quantiÞed with ßow cytometry using annexin-V staining.

In a Þrst study [18], we showed that with 75 mM of 4-HNE in the culture medium, nearly 100% of cells expressing either RDH11 or RDH12 were protected, while about 40% of control nonexpressing cells were annexin-V positive. This signiÞcant protection demonstrates that RDH11 and RDH12 are able to efÞciently detoxify 4-HNE in cells, most likely through their ability to reduce it to a nontoxic alcohol. Our data also showed that this protection could be overwhelmed with increasing concentrations of 4-HNE [18].

We then investigated whether the enzymatic activity of RDH12 was necessary for protection against 4-HNE-induced apoptosis using stable cell lines expressing active or inactive variants of this protein [18]. In studies of human RDH12, wild-type R161 and the common variant R161Q have been previously reported to exhibit similar all- trans RDH activities [19, 20]. On the other hand, the T49M mutant showed a dramatic reduction in the ability to produce all-trans retinol from all-trans retinal, explained by a defect in cofactor binding [19]. As expected, both the wild-type and common variant of human RDH12 signiÞcantly protected 40Ð50% of cells against 4-HNE-induced apoptosis. On the other hand, the inactive T49M mutant did not protect the cells. This is not due to a lower expression level of the mutant because all three variants were expressed similarly in our stable cell lines [18]. This experiment demonstrated that the enzymatic activity of RDH12 is essential to detoxify 4-HNE in cells.

mouse retinas [18], after exposure to bright light. Exposure to bright light causes photoreceptor apoptosis, which can be blocked by various types of antioxidants demonstrating that oxidative damage mediates light-induced photoreceptor cell death [22Ð25]. ROS can directly attack PUFAs and initiate an autoampliÞed chain reaction of lipid oxidation in cellular membranes. This causes a nonenzymatic PUFA degradation into a variety of oxidized products, including shortand mediumchain reactive aldehydes such as 4-HNE [14]. 4-HNE can then react readily with histidine, cysteine, or lysine residues of proteins forming MichaelÕs adducts [26]. This reaction leads to a variety of effects such as inhibition of enzyme activity; targeting of modiÞed proteins for degradation; inhibition of protein, RNA, and DNA synthesis; cell cycle arrest; and apoptosis [12, 14, 15, 27]. Reactive aldehydes exert cytotoxicity largely because of their facile reactivity with proteins. A study by Tanito et al. [21] assessed the formation of 4-HNE-modiÞed proteins in lightexposed rat retinas by using speciÞc antibodies against 4-HNE-protein Michael adducts. This study showed that exposure to intense light increases 4-HNE protein modiÞcation in the retina and that this effect is reversed by prior injection of the antioxidant phenyl-N-tert-butylnitrone (PBN). The study also showed that protein modiÞcations by 4-HNE are early events that precede apoptosis and subsequent photoreceptor cell death. A more detailed analysis has been performed by immunohistochemistry and cell fractionation to investigate the localization of protein modiÞcations by 4-HNE [21]. 4-HNE-protein adducts accumulated signiÞcantly in rod inner segments and photoreceptor nuclei [21]. We have recently investigated the localization of light-induced 4-HNE adducts in BALB/c mouse retina [18]. We found that light-dependent accumulation of adduct occurs in the RPE, the photoreceptor cells, and the interneurons. In photoreceptor cells, adducts accumulate in the inner segments, cell bodies, and synaptic termini, but not in the outer segments [18]. The overlapping localization of RDH11, 12, and 13 with 4-HNE in photoreceptor inner segments suggests that 4-HNE could be a physiological substrate of these enzymes in the retina.

8.4.2Retinal Levels of 4-HNE in Rdh11 and Rdh12 Knockout Mice

Since 4-HNE colocalizes with RDHs in photoreceptor inner segments, we investigated whether these enzymes could detoxify endogenous 4-HNE, by measuring retinal levels of 4-HNE-protein adducts in existing mouse lines with disrupted Rdh11 or Rdh12 genes [18]. We Þrst quantiÞed 4-HNE-protein adduct in retinal homogenates of albino BALB/c wild-type, Rdh11 and Rdh12 knockout mice raised under dim cyclic light. The level of 4-HNE-modiÞed proteins was signiÞcantly higher (60% increase) in the Rdh12 knockout retina when compared to the wild-type retina [18]. As shown in Fig. 8.4, this difference between wild-type and knockout was further increased when we compared microsomal fractions (enriched in RDH12) instead of comparing retinal homogenates (60% increase in retinal

Fig. 8.4 QuantiÞcation of 4-HNE-protein adduct in mouse retina. BALB/c wild-type and Rdh12 knockout mice were raised in dim cyclic light for 8Ð12 weeks. Microsomal fractions were prepared as described [18], and equal aliquots

(10 mg) of retinal microsomes were analyzed by dot blot as described [18]. Six mice were used in each group and the mean and standard error are plotted. Results were compared using the StudentÕs t test for signiÞcance.

* = p < 0.05; ** = p < 0.001; and *** = p < 0.0001

homogenates vs. 80% increase in retinal microsomes). By contrast, in the Rdh11 knockout retina, the basal level of adducts is similar to that of the wild-type [18]. This result might simply reßect the fact that the expression level of RDH11 is much lower than that of RDH12, making the relative contribution of RDH11 to the detoxiÞcation of 4-HNE negligible in presence of RDH12. In the future, experiments comparing the level of adduct in retinas of Rdh12 knockout and Rdh11/Rdh12 double knockout mice might uncover a redundant detoxiÞcation role for RDH11.

Surprisingly, after exposure to bright light, adduct accumulation reached a maximum level that was similar in all three mouse lines [18]. This result suggests that neither RDH11 nor RDH12 are involved in detoxiÞcation during acute stress. It is possible that their enzymatic activities are overwhelmed at this level of stress. Taken together, these results suggest that RDH12 has a housekeeping detoxiÞcation role in photoreceptor cells, reducing 4-HNE produced by a basal level of lipid oxidation taking place constantly in mouse retina. In this case, RDH12 would exert its physiological function primarily in the dark or in moderate lighting, keeping 4-HNE at low levels. Because Rdh13 knockout mice are not available, the relative contribution of RDH13 to 4-HNE detoxiÞcation in the retina could not be tested. The physiological role of RDH13 could be to reduce reactive aldehydes produced by oxidation of PUFAs within mitochondrial inner membrane, following local production of ROS. This function could be very important to stop the propagation of oxidative damage outside the mitochondria into other compartments of the cell. To further determine the relative contributions and redundancy of these RDHs for detoxiÞcation of 4-HNE and other lipid oxidation products, various combinations of double knockouts and the triple knockout mouse lines will be needed.