Lysophosphatidylcholine acyltransferase 1 controls mitochondrial reactive oxygen species generation and survival of retinal photoreceptor cells

Due to their high energy demands and characteristic morphology, retinal photoreceptor cells require a specialized lipid metabolism for survival and function. Accordingly, dysregulation of lipid metabolism leads to the photoreceptor cell death and retinal degeneration. Mice bearing a frameshift mutation in the gene encoding lysophosphatidylcholine acyltransferase 1 (Lpcat1), which produces saturated phosphatidylcholine (PC) composed of two saturated fatty acids, has been reported to cause spontaneous retinal degeneration in mice; however, the mechanism by which this mutation affects degeneration is unclear. In this study, we performed a detailed characterization of LPCAT1 in the retina and found that genetic deletion of Lpcat1 induces light-independent and photoreceptor-specific apoptosis in mice. Lipidomic analyses of the retina and isolated photoreceptor outer segment (OS) suggested that loss of Lpcat1 not only decreased saturated PC production but also affected membrane lipid composition, presumably by altering saturated fatty acyl-CoA availability. Furthermore, we demonstrated that Lpcat1 deletion led to increased mitochondrial reactive oxygen species levels in photoreceptor cells, but not in other retinal cells, and did not affect the OS structure or trafficking of OS-localized proteins. These results suggest that the LPCAT1-dependent production of saturated PC plays critical roles in photoreceptor maturation. Our findings highlight the therapeutic potential of saturated fatty acid metabolism in photoreceptor cell degeneration–related retinal diseases.

Due to their high energy demands and characteristic morphology, retinal photoreceptor cells require a specialized lipid metabolism for survival and function. Accordingly, dysregulation of lipid metabolism leads to the photoreceptor cell death and retinal degeneration. Mice bearing a frameshift mutation in the gene encoding lysophosphatidylcholine acyltransferase 1 (Lpcat1), which produces saturated phosphatidylcholine (PC) composed of two saturated fatty acids, has been reported to cause spontaneous retinal degeneration in mice; however, the mechanism by which this mutation affects degeneration is unclear. In this study, we performed a detailed characterization of LPCAT1 in the retina and found that genetic deletion of Lpcat1 induces light-independent and photoreceptor-specific apoptosis in mice. Lipidomic analyses of the retina and isolated photoreceptor outer segment (OS) suggested that loss of Lpcat1 not only decreased saturated PC production but also affected membrane lipid composition, presumably by altering saturated fatty acyl-CoA availability. Furthermore, we demonstrated that Lpcat1 deletion led to increased mitochondrial reactive oxygen species levels in photoreceptor cells, but not in other retinal cells, and did not affect the OS structure or trafficking of OS-localized proteins. These results suggest that the LPCAT1-dependent production of saturated PC plays critical roles in photoreceptor maturation. Our findings highlight the therapeutic potential of saturated fatty acid metabolism in photoreceptor cell degeneration-related retinal diseases.
In addition to the de novo pathway (Kennedy pathway) of phospholipid biosynthesis (1), fatty acyl moieties of membrane phospholipids are turned over dynamically by deacylation and reacylation cycles, which is called the Lands' cycle (2). Appropriate regulation of phospholipid composition by the Lands cycle is required for various cellular functions, including lipoprotein production and lipid mediator production (3). In the Lands cycle, lysophospholipid acyltransferases (LPLATs) involve fatty acid (FA) reacylation to generate membrane phospholipid diversity. Depending on their substrate (lysophospholipids and acyl-CoAs) selectivity, LPLATs produce specific types of membrane phospholipids, such as polyunsaturated FA (PUFA)-containing phospholipids and saturated FA (SFA)-containing phospholipids.
Among LPLATs, lysophosphatidylcholine acyltransferase 1 (LPCAT1, also known as LPLAT8) produces saturated phosphatidylcholine (PC) using lysophosphatidylcholine (LPC) and saturated fatty acyl-CoA, such as palmitoyl-CoA (4)(5)(6)(7). Previous studies have shown that the LPCAT1-mediated production of saturated PC is required for the proper functioning of pulmonary surfactant in the lung (7,8) and trafficking of growth factor receptors in cancer cells (9). In addition, Lpcat1 is essential for the survival of retinal photoreceptor cells (10,11). The critical role of Lpcat1 in photoreceptor cells was originally uncovered by the analysis of retinal degeneration 11 (rd11) mice, possessing a frameshift mutation in the Lpcat1 gene. In rd11 mice, the retinal outer nuclear layer (ONL) composed of photoreceptor cells is rapidly diminished, causing vision loss until 1 month of age.
Photoreceptor cells possess a unique membrane phospholipid composition (12). In the membrane of photoreceptor cells, especially in the outer segment (OS) discs, docosahexaenoic acid (DHA)-containing phospholipids are extremely enriched. Our recent study revealed that the loss of DHAcontaining phospholipids leads to the collapse of the OS disc structures and retinal degeneration (13). In parallel with this, significant levels of SFA-containing phospholipids were also present in photoreceptor cells. Therefore, the photoreceptor degeneration in rd11 suggests that the LPCAT1-mediated production of SFA-containing phospholipids contributes to photoreceptor survival. However, the molecular basis underlying retinal degeneration in rd11 mice remains unclear.
In this study, we investigated the role of LPCAT1 in the retina using the Lpcat1 knockout (KO) mice. Consistent with rd11 mice, the retina of Lpcat1 KO mice showed a rapid loss of ONL with decreased levels of dipalmitoyl PC (DPPC), the major saturated PC species in the retina. We demonstrated that retinal degeneration in Lpcat1 KO mice was caused by apoptosis of photoreceptor cells, which occurs in a lightindependent manner. Furthermore, Lpcat1 KO photoreceptor cells accumulated mitochondrial reactive oxygen species (ROS). Our study demonstrated that the roles of LPCAT1 are not only in DPPC production for maintenance of membrane integrity but also in the regulation of normal mitochondrial functions.

Photoreceptor cell apoptosis in Lpcat1 KO mice
A previous study has shown that a frameshift mutation in the Lpcat1 gene in rd11 and B6-JR2845 mice leads to the spontaneous development of severe retinal degeneration (10). Therefore, to investigate whether the loss of function of LPCAT1 causes retinal degeneration, we used mice genetically deleted for the Lpcat1 gene. Consistent with rd11 and B6-JR2845 mice, a dramatic reduction in ONL and OS + inner segment thickness was observed in 6-week-old Lpcat1 KO mice compared to that in Lpcat1 wildtype (WT) and heterozygous (HZ) mice (Fig. 1, A and B). Meanwhile, the thickness of the inner nuclear layer and ganglion cell layer were similar among the three groups (Fig. 1, A and B). The histological differences between Lpcat1 WT and HZ mouse retinas were not observed.
Apoptosis of photoreceptor cells is a major cause of retinal degeneration (14). To assess the cause of retinal degeneration in Lpcat1 KO mice, we visualized the apoptotic cells in the retina using terminal deoxynucleotidyl transferase dUTP nickend labeling (TUNEL). TUNEL analyses indicated that the number of apoptotic photoreceptor cells was higher in 4-weekold Lpcat1 KO mice than in HZ mice (Fig. 1C). Notably, the increase in apoptotic cells in global Lpcat1 KO mice was not observed in the lungs, liver, or other retinal cell layers (Figs. 1C and S1, A and B). These results indicate that loss of Lpcat1 leads to photoreceptor cell apoptosis in a specific manner.

Onset of the photoreceptor apoptosis in Lpcat1 KO mice
To assess the onset of photoreceptor cell apoptosis in Lpcat1 KO mice, we performed TUNEL staining of the retina at different time points. Apoptotic cells in the retina of Lpcat1 KO mice increased greatly from postnatal day 8 (P8) ( Fig. 2A). Consistent with this, the thickness of the ONL was thinner in Lpcat1 KO mice at P8, while no difference in the apoptotic cell number and ONL thickness was seen in the P6 retina (Fig. 2, A  and B). Dramatic increases in the expression of rhodopsin, critical for phototransduction and OS formation (15), at this time point (Fig. 2C), suggests that photoreceptor apoptosis in Lpcat1 KO retinas occurred concomitantly with photoreceptor OS formation.
As OS is required for phototransduction, we hypothesized that photoreceptor cell death in Lpcat1 KO mice is related to light exposure. Thus, we investigated the apoptosis of photoreceptor cells in P9 mice, which were raised in complete darkness from P2 to P9. However, photoreceptor cell apoptosis was observed even in dark-reared Lpcat1 KO mice (Fig. 2D), suggesting that retinal degeneration in Lpcat1 KO mice was independent of light stimulation.

Altered phospholipid composition in Lpcat1 KO mouse retina
Photoreceptor cells undergo dynamic morphological and metabolic alterations during retinal maturation (16). Since LPCAT1 produces SFA-containing PC species, we next analyzed PC composition and LPCAT1 expression during retinal maturation. As shown in Figure 3, liquid chromatography-tandem mass spectrometry (LC-MS/MS) analysis showed a gradual alteration in retinal PC composition during maturation. Consistent with previous studies (12,13), the proportion of DHA-containing PC species, including PC 38:6, PC 40:6, and PC 44:12, increased along with retinal maturation (Fig. 3). Coinciding with the onset of photoreceptor cell apoptosis in Lpcat1 KO mice, DPPC, a major product of LPCAT1, was increased in the retina of WT mice between 1 and 2 weeks of age (Fig. 3, A and B). Consistent with the substrate selectivity of LPCAT1 in vitro (5, 7), DPPC levels were decreased by half in Lpcat1 KO retinas compared to WT retinas. Thus, DPPC production in the retina largely depends on LPCAT1 expression. However, because the induction of Lpcat1 mRNA and protein was not observed during retinal maturation (Fig. S2, A and B), post-translational modifications or the increased substrate supply for LPCAT1, such as LPC and palmitoyl-CoA, may affect the age-dependent elevation of DPPC.
Transcriptomic analysis of Lpcat1 KO retina and isolated photoreceptor cells Next, we aimed to clarify the cause of photoreceptor cell death in Lpcat1 KO mice by comparing the mRNA expression profiles with the control mice. First, we analyzed the transcriptomic differences in the 2-week-old retinas of Lpcat1 WT and KO mice (Fig. 4, A and B). Gene groups that increased significantly in Lpcat1 KO retinas were analyzed by functional classification using "the Database for Annotation, Visualization, and Integrated Discovery." These results suggest that many of the genes upregulated in Lpcat1 KO retinas were related to immune and inflammatory responses and genes downregulated in Lpcat1 KO retinas were involved in the cell projection organization and cilium movement ( Fig. 4B and Table S1). In addition, we found that the expression of genes termed immediate early genes (IEGs), such as Fosb, Fos, Egr1, and Egr2, was higher in Lpcat1 KO than in the control retina (Table S1). Among them, we found that Fosb expression in Lpcat1 KO was significantly higher than in the control retinas, even in darkreared mice (Fig. 4, C and D). As IEGs are rapidly upregulated by various cellular stimuli, particularly in neurons (17), we performed qPCR analyses using isolated a photoreceptor cell marker CD73-positive and CD73-negative cells (18) from the retina to identify the cellular sources of their upregulation (Fig. S3A). The qPCR results showed increased expression of IEGs in CD73-negative cells (Figs. 4E and S3B). The induction of Fosb in Lpcat1 KO retinas was slightly delayed from the onset of retinal degeneration (Fig. 4C), suggesting that it might be triggered by apoptosis of photoreceptor cells.
To clarify the primary cause of photoreceptor cell apoptosis in Lpcat1 KO mice, we performed microarray analysis using isolated photoreceptor cells at P8 when apoptosis of photoreceptor cells in Lpcat1 KO mice begins. As a result of functional classification analysis, no significant enrichment of the biological process altered in Lpcat1 KO photoreceptor cells was found in the gene list, except for the categories of cell adhesion (Fig. S3, C and D). Since most of the downregulated genes listed in these categories in Lpcat1 KO photoreceptor cells were reported to be highly expressed in retinal pigment epithelial (RPE) cells, it may reflect the reduced contamination of RPE cell fragments in isolated photoreceptor cells in Lpcat1 KO mice. Consistently, qPCR analysis showed that mRNA expression of RPE65, an RPE-specific gene, was lower in isolated photoreceptor cells of Lpcat1 KO mice than in control mice (Fig. S3E). Although the cause of this observation is currently unknown, transcriptomic analysis of isolated photoreceptor cells suggested that loss of Lpcat1 did not induce the obvious changes in the gene expression profiles in photoreceptor cells. The right bar graph showed the ONL thickness (P6, n = 3 for each group; P8, n = 6 for each group). Scale bars are 50 μm. C, the levels of rhodopsin mRNA in P6 and P8 of Lpcat1 HZ and littermate KO retina. The expression level was normalized by glyceraldehyde 3-phosphate dehydrogenase (Gapdh) (n = 3 for HZ, n = 4 for KO). D, the average value of TUNEL-positive cells per outer nuclear layer (ONL) areas in P9 dark-reared Lpcat1 HZ and littermate KO retina (n = 4 for HZ, n = 5 for KO). A-C, significance is based on two-way ANOVA followed by Bonferroni's multiple comparisons test (*p < 0.05, **p < 0.01, ns; no significance). Data are shown as mean + SD of independent experiments. D, significance is based on the unpaired t test (**p < 0.01). Data are shown as mean + SD of independent experiments. TUNEL, terminal deoxynucleotidyl transferase dUTP nick-end labeling.
A previous study reported that the failure of SFA-containing PC production due to loss of LPCAT1 leads to excessive PUFA accumulation and endoplasmic reticulum (ER) stress response in the retina (19). However, under the present assay conditions, no induction of CCAAT-enhancer-binding protein homologous protein, an ER stress marker, was observed both in Lpcat1 KO retina and isolated photoreceptor cells (Fig. S3F).

Trafficking and functions of OS-localized proteins in Lpcat1 KO mice
Based on the microarray analyses showing no obvious causative changes in photoreceptor cell death in Lpcat1 KO mice, we investigated whether the altered membrane phospholipid composition in Lpcat1 KO photoreceptor cells influences the localization and functions of proteins essential for photoreceptor survival. Consistent with the previous observation of Lpcat1 mutant mice (10), the structure of Lpcat1 KO OS discs photoreceptor cells appeared normal (Fig. 5A). We then explored the possibility that an altered PC composition of Lpcat1 KO photoreceptor OS membrane affected the localization and function of phototransduction-related proteins since the mislocalization of photoreceptor OS-localized proteins, such as rhodopsin and phosphodiesterase 6β (PDE6β), is observed in various types of retinal degeneration (20,21). In parallel with the DHAcontaining PCs enriched in the OS discs (13), we found that DPPC was also abundant in the OS and decreased by half in Lpcat1 KO (Fig. 5B). This result suggested that the LPCAT1mediated production of DPPC contributes to maintaining proper PC composition in the photoreceptor OS membrane.
As we found that Lpcat1 deletion leads to altered OS membrane PC composition, we next investigated the localization of rhodopsin and PDE6β in the retina. Unexpectedly, both rhodopsin and PDE6β were normally localized in the photoreceptor OS in Lpcat1 KO mice, despite the altered membrane PC composition in the OS (Fig. 5C). PDE6β is a rod photoreceptorspecific subunit of PDE that controls cyclic nucleotide-gated channel-mediated cation influx into photoreceptor cells (22). Mice harboring the loss-of-function mutant of PDE6β (termed rd1 mouse) showed marked cGMP accumulation and subsequent severe retinal degeneration (23,24). As rd1 mice show similar phenotypes to those of Lpcat1 KO, early-onset, and lightindependent retinal degeneration (21), we assessed whether the altered membrane lipid composition in Lpcat1 KO photoreceptor cells leads to PDE6β dysfunction. However, cGMP levels in the retina were slightly decreased rather than increased in Lpcat1 KO mice compared to control mice (Fig. 5D). Since the accumulated cGMP-dependent increase in calcium ion influx  triggers retinal degeneration in rd1 mice, the underlying mechanisms of retinal degeneration in Lpcat1 KO mice differ from those in PDE6β mutated mice (rd1).

Increased mitochondrial oxidative stress in Lpcat1 KO mice
Excess SFA induces apoptosis (25). Similarly, stearoyl-CoA desaturase (SCD), which converts saturated fatty acyl-CoA to   LPCAT1 is essential for photoreceptor survival monounsaturated fatty acyl-CoA, protects cells from SFAinduced cell death (26). Together with the significant decrease in the PC 32:1/PC 32:0 ratio (Fig. S4A) and SCD mRNA (Fig. S4B) during photoreceptor maturation, mature photoreceptor cells are presumably less potent in reducing the intracellular SFA levels by the unsaturation. We hypothesized that SFA stress is involved in photoreceptor-specific apoptosis in Lpcat1 KO mice based on these results. Several factors, including ceramide accumulation, mitochondrial ROS production, and ER stress, trigger SFA-induced apoptosis (27)(28)(29)(30). Since our transcriptomic analyses showed no sign of increased ER stress in Lpcat1 photoreceptor cells (Fig. S3E), we postulated that the altered palmitoyl-CoA flux (Fig. 6A) by Lpcat1 deletion might be related to retinal degeneration.
To this end, we determined whether the Lpcat1 deletion leads to ceramide accumulation in the retina. A slight increase in sphingomyelin (SM) suggested that loss of Lpcat1 increased palmitoyl-CoA availability for sphingolipid synthesis. However, we could not find any clear differences in ceramide levels and compositions between genotypes (Figs. 4C and 6, B and C).
Finally, we explored the possibility that the Lpcat1 deletion leads to increased mitochondrial FA β-oxidation and consequent ROS accumulation in mature photoreceptor cells. We analyzed the mitochondrial ROS accumulation in retinal cells of P21 control and Lpcat1 KO mice. Flow cytometric analyses of retinal CD73-positive (photoreceptor) cells showed that the ROS accumulated photoreceptor cells in Lpcat1 KO mice were increased compared to the control mice (Figs. 6D and S5, A-C). Consistent with TUNEL staining, we observed no differences in ROS accumulation between Lpcat1 KO and control CD73-negative cells (Fig. 6E). Additionally, intracellular ROS levels were also increased in Lpcat1 KO photoreceptor cells (Fig. S6, A-E). These results suggest that mitochondrial ROS accumulation triggered by Lpcat1 deletion is a photoreceptorspecific phenomenon contributing to apoptosis induction.

Discussion
Here, we demonstrated that the loss of Lpcat1 leads to an early onset of severe retinal degeneration, triggered by light stimulus-independent photoreceptor cell apoptosis. Although photoreceptor cell damage induces various types of retinal degeneration, the molecular basis of photoreceptor cell death is not fully understood. Our present study suggests that Lpcat1 deletion-triggered disruption of palmitoyl-CoA flux leads to mitochondrial ROS accumulation and denaturation in photoreceptor cells.
FA saturation of membrane phospholipids affects membrane fluidity and flexibility through their biophysical properties (31,32). A recent study showed that PUFA-containing phospholipids, especially in DHA-containing phospholipids, are enriched in the center of photoreceptor discs. In contrast, SFA-containing species are enriched in the rim region (33). Therefore, significant levels of saturated PCs in the OS membrane suggest that the photoreceptor OS membrane and/ or OS-localized protein require the saturated PC-enriched membrane for their normal structure and functions. In the case of DHA-containing phospholipids, it has been reported that the dramatic decrease in these phospholipids in 1acylglycerol-3-phosphate O-acyltransferase 3, also known as LPLAT3 (4) maintaining their structures (13). However, in contrast, the structure of Lpcat1 KO photoreceptor discs appeared to be normal, regardless of a significant decline in DPPC and PC 30:0. Thus, an increased proportion of PC 34:0, another type of SFA-containing PC, in Lpcat1 KO photoreceptor OS, may play redundant roles, at least in the formation and/or maintenance of OS disc structures. Conversely, cGMP levels in Lpcat1 KO retinas were lower than those in WT mice (Fig. 5D). Although the decreased cGMP levels in the Lpcat1 KO retina may reflect the lower number of photoreceptor cells in these mice, it is also possible that the PDE6β activity was affected by altered PC composition in the photoreceptor OS membrane. The detailed mechanisms of this observation should be clarified in future studies. We demonstrated that mitochondrial ROS accumulation was observed in Lpcat1 KO photoreceptor cells. This suggests that they are highly susceptible to mitochondrial ROS; therefore, mitochondrial FA oxidation should be strictly regulated. A recent study revealed that photoreceptor cells depend largely on FA oxidation in the mitochondria for ATP production (16,34). Together with the fact that mature photoreceptor cells require high levels of ATP to maintain the depolarized state in darkness using ATP-dependent channels (35), LPCAT1-mediated incorporation of saturated FAs into PC may be required not only for the production of desaturated PCs but also for the proper control of mitochondrial FA oxidation. Since knockdown of Lpcat1 expression leads to tumor cell death (9), Lpcat1-dependent cell survival of mature photoreceptor cells and cancer cells may show some metabolic similarities.
SCDs are also related to controlling cellular SFA levels by changing saturated acyl-CoA to monounsaturated fatty acyl-CoA (36). It is reported to protect cells from SFA-induced mitochondrial ROS accumulation and apoptosis by changing the fate of SFA from mitochondria to lipid droplets (37,38). Therefore, a dramatic decrease in SCD expression during photoreceptor maturation, which is reflected in the decreased PC 32:1/PC 32:0 (DPPC) ratio, may also be associated with the sensitivity of Lpcat1 deletion-triggered increase in palmitoyl-CoA availability. It is also possible that altered mitochondrial membrane lipid composition directly affects mitochondrial functions, such as electron transfer, permeability, and fusion/ fission. Additionally, we could not exclude the possibility that mitochondrial denaturation and ROS accumulation result from photoreceptor apoptosis. Therefore, further studies are required to decipher the molecular mechanisms of mitochondrial ROS accumulation in Lpcat1 KO photoreceptor cells.
From the transcriptomic analyses, we identified the apparent induction of FosB mRNA in the Lpcat1 KO retina during the progression of retinal degeneration. Although the exact roles of FosB in the retina are unclear, it has been reported that Fosb is induced in the retina during retinal degeneration (39). As the onset of photoreceptor cell death was followed by the upregulation of Fosb in Lpcat1 KO mice, this might not be the cause of retinal degeneration. However, considering the role of FosB as a component of activator protein-1, a dimeric transcription factor involved in inflammation, angiogenesis, and apoptosis (40,41), our results suggest that Fosb inhibition may be involved in retinal degeneration in Lpcat1 KO mice. Importantly, the upregulation of Fosb is also found in the monocytes of patients with age-related macular degeneration (42). Therefore, cell types that upregulate Fosb during retinal degeneration in Lpcat1 KO retinas will be clarified by single-cell RNA sequencing in a future study.
In summary, we presented a detailed analysis of Lpcat1 deletion-induced retinal degeneration. Our data suggest that LPCAT1 is required not only for SFA-containing PC production but also for regulating intracellular ROS levels in photoreceptor cells. Since increasing evidence indicates that the dysregulation of FA metabolism is demonstrated to involve in the various retinal diseases (35,43), LPCAT1-dependent saturated FA metabolism may also be implicated in human retinal disorders. Therefore, our data provide further insights into the mechanisms underlying photoreceptor cell deathtriggered retinal degeneration.

Animals
All animal experiments were approved and performed following the guidelines of the Animal Research Committee of the National Center for Global Health and Medicine (12053,13009,14045,15037,16062,17054,18046,19039,20037, and 21083).

Mice
Lpcat1 KO mice (C57BL6/N background) generated in our previous study were used in this study (7). C57BL6/N mice were reported to show retinal degeneration because they harbor a mutation in the Crumbs homolog 1 (Crb1) gene, termed rd8 (24,44). Thus, we crossed Lpcat1 KO mice with C57BL6/J mice to remove the rd8 mutation. Mice without the rd8 homozygous mutation were used in this study. Mice were reared in a normal 12-h light-dark cycle or complete darkness from P2 to P9.

Histological analysis of mouse retina
For frozen sections, enucleated eyes were fixed in 4% paraformaldehyde for 2 h and washed three times with phosphate-buffered saline (PBS) for 5 min, followed by incubation in 10% sucrose/PBS for 3 h, and then in 25% sucrose/ PBS overnight at 4 C. Afterward, the eyes were flash-frozen in a frozen section compound (Leica). Frozen tissue samples were sectioned at 12 μm using a cryostat. For paraffin sections, enucleated eyes, lungs, and livers were fixed in 4% paraformaldehyde at 4 C for <24 h and then embedded in paraffin. Paraffin-embedded tissue samples were sectioned at 3 μm using a microtome. Tissue sections were stained with hematoxylin-eosin and analyzed using ImageJ software to measure ONL thickness. TUNEL staining was performed (TaKaRa Biochemicals) to detect apoptotic cells in the retina. For immunohistochemical analyses of PDE6β and rhodopsin, rabbit polyclonal PDE6β (ab5663; Abcam) and 1D4 (ab3424; Abcam) antibodies were used, respectively. Immunohistochemical analyses were performed using the VECTASTAIN Elite ABC kit (Vector Laboratories).

Quantitative reverse transcriptase PCR
Total RNA was extracted from the retina or isolated photoreceptor cells using an RNeasy Mini Kit (Qiagen). Single-strand cDNA was synthesized using SuperScript III reverse transcriptase (Thermo Fisher Scientific). Using random primers (Thermo Fisher Scientific), qPCR was performed using the Fast SYBR Green Master Mix (Applied Biosystems) with the Ste-pOnePlus real-time PCR system (Applied Biosystems). Relative mRNA expression levels were calculated using the comparative cycle threshold method and normalized to GAPDH mRNA levels. Primers for qPCR were as follows: gene name: forward primer (5 0 -3 0 ), reverse primer (

LC-MS/MS
For PC analyses, lipids were extracted from microsomal fractions or retinal OS using the Bligh & Dyer method (45).
Subsequently, the extracted lipids were dried using a centrifugal evaporator and reconstituted in methanol. Lipid samples were then analyzed by LC-electrospray ionization-MS/MS with multiple reaction monitoring (MRM). LC-MS/ MS analyses were performed using a Nexera UHPLC system and triple quadrupole mass spectrometer LCMS-8050 (Shimadzu Corp). The lipid samples (5 μl/injection) were separated on an Acquity UPLC BEH C8 column (1.7 μm, 2.1 × 100 mm, Waters) at a flow rate of 0.35 ml/min with a gradient of mobile phase A (5 mM NH 4 HCO 3 /water), B (acetonitrile), and C (isopropyl alcohol). The column oven temperature was set to 47 C. The gradient setting was as follows-time ( For the analysis of ceramides, lipids were extracted from the mouse retina as previously described, with slight modifications (46). Briefly, retinas were homogenized in lysis buffer (100 mM Tris-HCl [pH 7.4], 300 mM sucrose) containing a 1× cOmplete protease inhibitor cocktail (Roche). After removing the tissue debris (800g, 10 min, 4 C), 20 μl of lysate was mixed with 30 μl of PBS, 50 μl of d18:1/12:0ceramide (50 μM in methanol), 750 μl of n-butanol, 30 μl of phosphate buffer (500 mM, pH 5.8), and 170 μl of water. The samples were then vortexed for 5 min, sonicated for 3 min in a bath sonicator, and centrifuged at 1000g for 5 min at 4 C. An upper phase (700 μl) of the first butanol extract was transferred to a new tube. Three hundred fifty microliters of ethyl acetate and 350 μl of hexane were added to the remaining lower phase for the second extraction. After vortexing for 5 min at 4 C, 700 μl of the upper phase (second extract) was combined with the first butanol extract and added to 700 μl of methanol. The combined lipid extract (210 μl,10%) was dried using a centrifugal evaporator, reconstituted in 60 μl of solvent B (isopropyl alcohol with 0.2% formic acid and 0.028% ammonia), and then mixed with 90 μl of solvent A (isopropyl alcohol/methanol/water [5:1:4] with 0.2% formic acid, 0.028% ammonia, and 5 μM phosphoric acid). The samples were analyzed by LC-MS/MS with selected reaction monitoring. LC-MS/MS analysis was performed using an AQUITY UPLC system and a TSQ Vantage triple-stage quadrupole mass spectrometer. The extracted lipids (5 μl/injection) were separated on an Acclaim PepMap 100 C18 column (

Microarray
Total RNA from the whole retina or isolated photoreceptor cells was extracted using the RNeasy Mini Kit (Qiagen). Total RNA was examined using the SurePrint G3 Mouse GE 8 × 60K Microarray (Agilent Technologies). Data were quantified using the Agilent Feature Extraction software (Agilent Technologies) and normalized by 75 percentile shift normalization using GeneSpring software (Agilent Technologies). The list of highly expressed genes in Lpcat1 KO than in control with fold change >1.2 or 1.5 and q-value <0.05 or 0.5 was analyzed for common functions of altered genes using gene ontology (GO) terms using the Database for Annotation, Visualization, and Integrated Discovery.

Isolation of photoreceptor cells
Photoreceptor cells were isolated from the mouse retina as previously described, with slight modifications (18,47). Retinas were isolated from Lpcat1 HZ and KO mice and dissociated in 500 μl of 0.25% trypsin (Nacalai Tesque) in PBS for 15 min at 37 C. Then, 500 μl of 20% fetal bovine serum (Thermo Fisher Scientific) and 1 μl DNase I (Invitrogen) were added on ice. Mechanical dissociation was performed by pipetting 20 times using a 1-ml tip. The cells were collected by centrifugation at 300g for 5 min. The retinal cell pellet was washed in 700 μl of 2% bovine serum albumin (BSA)/PBS and centrifuged at 300g for 5 min. The cell pellet was resuspended in 50 μl of 2% BSA/ PBS, and photoreceptor cells were labeled with PE rat antimouse CD73 antibody (TY/23, BD Pharmingen) for 30 min at 4 C. Cells were washed with 700 μl of 2% BSA/PBS and centrifuged at 300g for 5 min. The cell pellet was resuspended in 80 μl of 2% BSA/PBS and incubated with 20 μl of anti-PE MicroBeads UltraPure (Miltenyi Biotec) for 15 min at 4 C. After washing with 700 μl of 2% BSA/PBS and centrifugation at 300g for 5 min, the cell pellet was resuspended in 500 μl of 2% BSA/PBS and filtered through a 35-μm preseparation filter (Corning). The CD73-positive cells were collected as photoreceptor cells using an autoMACS Pro Separator (Miltenyi Biotec).

Retinal OS isolation
Mouse retinal OS was purified as described previously (48). Twelve mouse retinas were suspended in 120 μl of 8% OptiPrep (Abbott Diagnostics Technologies AS) in Ringer's buffer (130 mM NaCl, 3.6 mM KCl, 2.4 mM MgCl 2 , 1.2 mM CaCl 2 , 0.02 mM EDTA, 10 mM Hepes-NaOH, pH7.4). Retinal OS fractions were separated from the retina by vortexing for 1 min. The samples were then centrifuged at 200g for 1 min. The vortexing and sedimentation sequence was repeated six times to collect the remaining retinal OS fraction in the retinal pellets. The collected supernatants (retinal OS fractions) were combined (600 μl) and overlaid onto a discontinuous gradient of OptiPrep (Veritas) (10%, 18%, 24%, and 30%) in Ringer's buffer (1.8 ml each) and centrifuged at 25,900g for 30 min. After centrifugation, the retinal OS fraction (approximately two-thirds of the way from the top of the gradient) was harvested and diluted three times with Ringer's buffer. The contaminated retinal fractions were removed by centrifugation at 500g for 3 min. The supernatant (pure retinal OS fraction) was precipitated by centrifugation at 26,500g for 30 min. The resultant retinal OS pellets (microsomal fractions) were resuspended in ice-cold buffer containing 20 mM Tris-HCl (pH 7.4), 300 mM sucrose, and 1 mM EDTA.

Determination of cGMP levels
P10 mice were dark-adapted for 16 h, and retinas were dissected under dim red light and frozen in liquid nitrogen. Two frozen retinas were homogenized in 200 μl of 5% trichloroacetic acid/water under dim red light and centrifuged at 1500g for 10 min. The resultant supernatant was used for cGMP measurements. cGMP levels were determined using a cyclic GMP EIA kit (Cayman) according to the manufacturer's instructions.

Transmitted electron microscopy
Enucleated eyes were immediately cut into hemispheres in a fixative containing 2% paraformaldehyde and 2.5% glutaraldehyde. The samples were then fixed in the same fixative for 2 h, followed by an additional fixation in the buffer containing 2% OsO4. The samples were dehydrated in a graded series of ethanol (70%-90%-100%) and embedded in an Epon 812 resin mixture (TAAB Laboratories Equipment Ltd). Ultrathin sections were created using an ultramicrotome, stained with uranyl acetate and lead citrate, and observed under an electron microscope (Hitachi H-7500).

Detection of mitochondrial ROS
ROS production was measured by flow cytometry using MitoSOX Red (Invitrogen). Retina was isolated from 3-weekold Lpcat1 WT and KO mice and dissociated in 500 μl of 0.25% trypsin (Nacalai Tesque) in PBS for 15 min at 37 C. Then, 500 μl of 20% fetal bovine serum (Thermo Fisher Scientific) and 1 μl DNase I (Invitrogen) were added on ice. Mechanical dissociation was performed by pipetting 20 times using a 1-ml tip. The cells were collected by centrifugation at 300g for 5 min. The retinal cell pellet was washed in 700 μl of 2% BSA/PBS and centrifuged at 300g for 5 min. The cell pellets were resuspended in 2% BSA/PBS. Photoreceptor cells were labeled with FITC anti-mouse CD73 antibody (TY/11.8, Bio-Legend). Dead cells were stained with LIVE/DEAD Fixable Far Red Dead Cell Stain kit (Invitrogen), and retinal cells were stained with 5 μM MitoSOX Red for 15 min at 37 C. The cells were centrifuged at 300g for 5 min. The cell pellet was resuspended in 500 μl of 2% BSA/PBS and filtered through a 35-μm preseparation filter (Corning). The fluorescence intensity was measured using a BD Accuri flow cytometer (BD Biosciences).

Statistical analysis
Unpaired t-tests were used when the two groups were compared. When two factors were present, a two-way ANOVA test was performed. Bonferroni's post hoc test was used when the ANOVA showed significance. All statistical analyses were performed using GraphPad Prism9 (version 9.2.0).

Data availability
The microarray data are deposited at GEO: GSE184815 (2week-old retina). The microarray data are deposited at GEO: GSE184817 (P8 photoreceptor cell).