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

Due to the 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 with a frameshift mutation of lysophosphatidylcholine acyltransferase 1 (Lpcat1), which produces saturated phosphatidylcholine (PC) composed of two saturated fatty acids, has been reported to cause spontaneous retinal degeneration (rd11 mice). 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 decreases saturated PC production and affects the proper cellular fatty acid flux, presumably by altering saturated fatty acyl-CoA availabilities. Furthermore, we demonstrated that Lpcat1 deletion increased mitochondrial reactive oxygen species (ROS) levels in photoreceptor cells, but not in other retinal cells without affecting the OS structure and trafficking of OS-localized proteins. These results suggest that the LPCAT1-dependent production of saturated PC is critical for metabolic adaptation during photoreceptor maturation. Our findings highlight the therapeutic potential of saturated fatty acid metabolism in photoreceptor cell degeneration-related retinal diseases.


INTRODUCTION
In addition to the de novo pathway (Kennedy pathway) of phospholipid biosynthesis (1), fatty acyl moieties of membrane phospholipids are turned over dynamically by de-acylation and re-acylation cycles, which is called the Lands' cycle (2). Appropriate regulation of phospholipid composition by Lands' cycles 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) re-acylation 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) produces saturated phosphatidylcholine (PC) using lysophosphatidylcholine (LPC) and saturated fatty acyl-CoA, such as palmitoyl-CoA (4-6). Previous studies have shown that the LPCAT1-mediated production of saturated PC is required for the proper functioning of pulmonary surfactant in the lung (6,7) and trafficking of growth factor receptors in cancer cells (8). In addition, Lpcat1 is essential for the survival of retinal photoreceptor cells (9,10). 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 (11). 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 DHA-containing phospholipids leads to the collapse of the OS disc structures and retinal degeneration (12). 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 LPCAT1mediated production of SFA-containing phospholipids contributes to photoreceptor survival. However, the molecular basis underlying retinal degeneration in rd11 mice remain 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 light-independent 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 (9). 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 thickness was observed in 6-week-old Lpcat1 KO mice compared to that in Lpcat1 wild-type (WT) and heterozygous (HZ) mice ( Figure 1A and 1B). Meanwhile, the thickness of the inner nuclear layer (INL) and ganglion cell layer (GCL) were similar among the three groups ( Figure 1A and 1B). The histological differences between Lpcat1 WT and HZ mouse retinas were not observed.
Apoptosis of photoreceptor cells is a major cause of retinal degeneration (13). To assess the cause of retinal degeneration in Lpcat1 KO mice, we visualized the apoptotic cells in the retina using terminal deoxynucleotidyl transferase dUTP nick-end labeling (TUNEL). TUNEL analyses indicated that the number of apoptotic photoreceptor cells was higher in 4-week-old Lpcat1 KO mice than in HZ mice ( Figure 1C). Notably, the increase in apoptotic cells in global Lpcat1 KO mice was not observed in the lungs, liver, or other retinal cell layers ( Figure 1C and Figure S1A and S1B). 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) (Figure 2A). Consistent with this, the thickness of ONL was thinner in Lpcat1 KO mice at P8, while no difference in apoptotic cell number and ONL thickness was seen in the P6 retina (Figure 2A and 2B). Dramatic increases in the expression of rhodopsin, critical for phototransduction and OS formation (14), at this time point ( Figure 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 raised in complete darkness from P2 to P9. However, photoreceptor cell apoptosis was observed even in dark-reared Lpcat1 KO mice ( Figure 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 (15). 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 (11,12), the proportion of DHA-containing PC species, including PC 38:6, PC 40:6, and PC 44:12, increased along with retinal maturation ( Figure 3). Coinciding with the onset of photoreceptor cell apoptosis in Lpcat1 KO, DPPC, a major product of LPCAT1, was increased in the retina of WT mice between 1 and 2 weeks of age ( Figure 3). Consistent with the substrate selectivity of LPCAT1 in vitro (4,6), 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 ( Figure S2A and S2B), 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 ( Figure 4A and 4B). Gene groups that increased significantly in Lpcat1 KO retinas were analyzed by functional classification using "the Database for Annotation, Visualization, and Integrated Discovery (DAVID)." These results suggest that many of the genes upregulated in Lpcat1 KO retinas were related to immune and inflammatory responses ( Figure 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 dark-reared mice ( Figure 4C and 4D). As IEGs are rapidly upregulated by various cellular stimuli, particularly in neurons (16), we performed qPCR analyses using isolated a photoreceptor cell marker CD73-positive and -negative cells (17) from the retina to identify the cellular sources of their upregulation. The qPCR results showed increased expression of IEGs in CD73 negative cells ( Figure 4E and Figure S3A). The induction of Fosb in Lpcat1 KO retinas was slightly delayed from the onset of retinal degeneration ( Figure 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 and melanin biosynthetic process ( Figure   S3B and S3C). 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 seemed to reflect the reduced contamination of RPE cell fragments in isolated photoreceptor cells in Lpcat1 KO mice. This could be caused by the loss of cell-cell tight interactions triggered by photoreceptor death. Indeed, the qPCR analysis showed that mRNA expression of RPE65, an RPEspecific gene, was lower in isolated photoreceptor cells of Lpcat1 KO mice than in control mice ( Figure S3D).
A previous study reported that the failure of SFA-containing PC production due to loss of LPCAT1 leads to excessive polyunsaturated FA accumulation and endoplasmic reticulum (ER) stress response in the retina (18). However, under the present assay conditions, no induction of CCAATenhancer-binding protein homologous protein (Chop), an ER stress marker, was observed both in Lpcat1 KO retina and isolated photoreceptor cells ( Figure S3E).  (19,20). In parallel with the DHA-containing PCs enriched in the OS discs (12), we found that DPPC was also abundant in the OS and decreased by half in Lpcat1 KO ( Figure 5B). This result suggested that the LPCAT1-mediated production of DPPC contributes to maintaining proper PC composition in the photoreceptor OS membrane.

Trafficking and functions of OS localized proteins in Lpcat1 KO mice
As we found that Lpcat1 deletion leads to altered OS membrane PC composition, we next investigated the localization of rhodopsin and PDE6b in the retina. Unexpectedly, both rhodopsin and PDE6b were normally localized in the photoreceptor OS in Lpcat1 KO mice, despite the altered membrane PC composition in the OS ( Figure 5C). PDE6b is a rod photoreceptor-specific subunit of PDE that controls cyclic nucleotide-gated (CNG) channel-mediated cation influx into photoreceptor cells (21). Mice harboring the loss-of-function mutant of PDE6b (termed rd1 mouse) showed marked cGMP accumulation and subsequent severe retinal degeneration (22,23). As rd1 mice show similar phenotypes to those of Lpcat1 KO, early-onset and light-independent retinal degeneration (20), we assessed whether the altered membrane lipid composition in Lpcat1 KO photoreceptor cells leads to PDE6b dysfunction. However, cGMP levels in the retina were slightly decreased rather than increased in Lpcat1 KO mice compared to control mice ( Figure 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 PDE6b mutated mice (rd1).

Increased mitochondrial oxidative stress in Lpcat1 KO mice
Excess SFA induces apoptosis (24). Similarly, stearoyl-CoA desaturase (SCD), which converts saturated fatty acyl-CoA to monounsaturated fatty acyl-CoA, protects cells from SFAinduced cell death (25). Together with the significant decrease in the PC 32:1/PC 32:0 ratio ( Figure   S4A) and SCD mRNA ( Figure 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 reactive oxygen species (ROS) production, and ER stress, trigger SFA-induced apoptosis (26-29). Since our transcriptomic analyses showed no sign of increased ER stress in Lpcat1 photoreceptor cells ( Figure S3E), we postulated that the altered palmitoyl-CoA flux ( Figure 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 ( Figure S4C, Figure 6B and 6C).
Finally, we explored the possibility that the Lpcat1 deletion leads to increased mitochondrial between Lpcat1 KO and control CD73 negative cells ( Figure 6E). These results suggest that mitochondrial ROS accumulation triggered by Lpcat1 deletion is a photoreceptor-specific 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 (30,31). 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 (32). 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 1-acylglycerol-3-phosphate O-acyltransferase 3 (AGPAT3), also known as lysophosphatidic acid acyltransferase 3 (LPAAT3), KO retina cause the collapse of photoreceptor discs, suggesting the importance of the FA composition of photoreceptor discs in maintaining their structures (12). 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 ( Figure 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 PDE6b activity was affected by altered PC composition in 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 fatty acid oxidation in the mitochondria for ATP production (15,33). Together with the fact that mature photoreceptor cells require high levels of ATP to maintain the depolarized state in darkness using ATP-dependent channels (34), LPCAT1-mediated incorporation of saturated fatty acids into PC may be required not only for the production of desaturated PCs but also for the  (39,40), 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 (AMD) (41). Therefore, cell types that upregulate Fosb during retinal degeneration in Lpcat1 KO retinas will be clarified by singlecell 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 lipid metabolism by controlling the amount of palmitoyl-CoA in photoreceptor cells. Increasing evidence indicates that disrupted lipid metabolism is the cause of photoreceptor cell death. Therefore, our data provide further insights into the mechanisms underlying photoreceptor cell death-triggered 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 (6).
C57BL6/N mice were reported to show retinal degeneration because they harbor a mutation in the Crumbs homolog 1 (Crb1) gene, termed rd8 (23,42). 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 hours 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, Germany). 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, Japan) to detect apoptotic cells in the retina. For immunohistochemical analyses of PDE6b and rhodopsin, rabbit polyclonal PDE6b (ab5663; Abcam, UK) and 1D4 (ab3424; Abcam, UK) antibodies were used, respectively. Immunohistochemical analyses were performed using the VECTASTAIN Elite ABC kit (Vector Laboratories, UK).

Quantitative Reverse transcriptase PCR (qPCR)
Total RNA was extracted from the retina or isolated photoreceptor cells using an RNeasy Mini Kit (Qiagen, Germany). Single-strand cDNA was synthesized using SuperScript III reverse transcriptase

LC-MS/MS
For PC analyses, lipids were extracted from microsomal fractions or retinal OS using the Bligh & Dyer method (43). Subsequently, the extracted lipids were dried using a centrifugal evaporator and

Microarray
Total RNA from the whole retina or isolated photoreceptor cells was extracted using the RNeasy Mini Kit (QIAGEN, Germany). Total RNA was examined using the SurePrint G3 Mouse GE 8×60K Microarray (Agilent Technologies, USA.). Data were quantified using the Agilent Feature Extraction software (Agilent Technologies, USA.) and normalized by 75 percentile shift normalization using GeneSpring software (Agilent Technologies, USA.). The list of highly expressed genes in Lpcat1 KO than in control with fold change > 1.5 or 2 and P < 0.05, was analyzed for common functions of altered genes using gene ontology (GO) terms using DAVID.

Isolation of photoreceptor cells
Photoreceptor cells were isolated from the mouse retina as previously described, with slight modifications (17,45). Retinas were isolated from Lpcat1 HZ and KO mice and dissociated in 500 µL

Retinal outer segment (OS) isolation
Mouse retinal OS was purified as described previously (46). Twelve mouse retinas were suspended in

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).

Conflict of interest:
The Department of Lipid Signaling, National Center for Global Health and Medicine, is financially supported by ONO Pharmaceutical Co., Ltd..

Abbreviations:
The

Measurement of total sphingomyelin (SM) levels:
Retinal lipids were separated using thin layer chromatography (TLC) to determine the total SM levels. Egr2 early growth response 2, transcript variant 1 4.2