Somatic Ablation of the Lrat Gene in the Mouse Retinal Pigment Epithelium Drastically Reduces Its Retinoid Storage

Abstract

purpose. To generate a mouse model in which the Lrat gene is selectively disrupted in the retinal pigment epithelium (RPE). To evaluate the effects on the synthesis of retinyl esters and on the expression of other proteins involved in the continuation of the visual cycle.

methods. A mouse line in which part of the first exon of the Lrat gene has been flanked by loxP sites, was generated and used in the study (Lrat ^L3/L3 mice). Heterozygous mice (Lrat ^+/L3) were crossed with mice expressing Cre-recombinase under control of the tyrosinase-related protein-1 (Tyrp1) promoter, which is active selectively in melanin-synthesizing cells such as RPE cells. Accordingly, mice obtained from these crosses should display an RPE-specific disruption of the Lrat gene (Lrat ^rpe−/−). In addition, by crossing CMV-Cre transgenic mice with Lrat ^L3/L3 animals, a germline null Lrat knockout (Lrat ^L−/L− mice) was generated. RNA and protein expression, endogenous retinoid levels, and electroretinogram (ERG) analyses were performed on Lrat ^rpe−/− and Lrat ^L−/^L− mice, to determine the effects of Lrat disruption. Retinoid levels in nonocular tissues were also analyzed for comparison.

results. Analysis of RPE tissues from Lrat ^rpe−/− mice showed absence of Lrat message, lack of Lrat protein expression and consequently a reduced light response in ERG recordings. In addition, RPE cells from Lrat ^rpe−/− showed a strong reduction in their ability to synthesize all-trans retinyl esters, whereas Lrat activity in other tissues known to process retinol was comparable to control Lrat ^L3/L3 animals. The Lrat ^L−/L− mice showed no detectable Lrat message, lack of protein expression, and barely detectable ester formation in RPE cells or several other relevant tissues analyzed.

conclusions. Three Lrat mouse lines with genetic modifications were generated. The Lrat ^L−/^L− mice displayed features similar to equivalent models previously reported by others. The second mouse line (Lrat ^rpe−/−) displayed loss of Lrat function only in the RPE. The third line possesses functional Lrat in all tissues, but part of the Lrat coding gene was flanked by loxP sites (Lrat ^L3/L3). This feature allows the disruption of this gene in any tissue of choice, by intercrossing with mice in which Cre-recombinase expression is driven by an appropriate tissue-specific promoter.

The retinal pigment epithelium (RPE) of the eye is a multifunctional component of the vertebrate retina. ¹ Anatomically, this epithelium is composed of a monolayer of cuboidal cells and is located between the large-bore capillary bed of the choroidal layer and the rod and cone photoreceptors of the neurosensory retina. One of the most important functions of the RPE in vision is the uptake of vitamin A (retinol) and the processing and transport of its derivatives (retinoids) in a cascade of enzymatic reactions known as the visual cycle. ^{1 2 3}  

In this cycle, retinol is initially delivered to the RPE cells from the liver as a complex with serum retinol-binding protein (RBP) and transthyretin. ⁴ This complex is bound to the basolateral plasma membrane of RPE cells via interaction with an RBP receptor. ⁵ Recently, a multitransmembrane protein of previously unknown function, stimulated by retinoic acid 6 (STRA6) has been suggested as a strong candidate for the RBP receptor. ⁶ This receptor binds RBP and delivers retinol into the cell where it is bound by cellular retinol binding protein 1 (CRBP1) for presentation to lecithin retinol acyltransferase (LRAT). ⁷ LRAT transfers fatty acyl groups from the sn-1 position of phosphatidylcholine (lecithin) to all-trans retinol for the generation of the all-trans retinyl esters. ^{8 9} All-trans retinyl esters are converted to 11-cis retinol via an isomerization reaction in which the RPE65 protein plays a critical role. ^{10 11 12} 11-cis Retinol is subsequently bound to cellular retinaldehyde-binding protein (CRALBP), oxidized into 11-cis retinaldehyde by 11-cis retinol dehydrogenase (11c-RDH) and then transported to the plasma membrane for its apical release from the RPE. ^{13 14} Finally, 11-cis retinaldehyde is exported to photoreceptors where it combines with opsin for the formation of rhodopsin and cone photopigments. 

In addition to its important role in the continuation of the visual cycle, LRAT is also critical in retinol processing in several other tissues, such as liver, ¹⁵ testis, ¹⁶ and small intestine. ⁸ Therefore, defects in this gene are predicted to have adverse effects on vision and retinoid metabolism. In humans, null mutations of visual cycle genes such as, RPE65, RDH5 (encoding 11c-RDH), RLBP1 (encoding CRALBP), and RGR (retinal-G-protein coupled receptor) have been related to inherited retinal degenerations. ^{17 18 19 20 21 22} In the case of LRAT, mutations in this gene have been associated with early-onset severe retinal dystrophy, the phenotype Leber congenital amaurosis (LCA). ²³ To understand more about the adverse impact on the visual cycle process caused by disruption of the Lrat gene, the generation of a knockout mouse model would be useful. Currently, two different groups have developed mouse lines bearing a germline null mutation of Lrat, and have highlighted the importance of these mutants in the study of human retinal dystrophies and as a model for vitamin A deprivation. ^{24 25} However, because of the importance of Lrat in essential processes such as reproduction, immune function, and development, ²⁶ the maintenance and propagation of such mouse lines represents a limitation—mainly due to limited male fertility. ^{24 25} Using the Cre/loxP technology, ^{27 28} we generated a mouse line in which the Lrat gene was selectively disrupted in the RPE of the eye. We demonstrate that, in addition to a drastic reduction of intracellular retinyl esters and impaired vision, selective disruption of the Lrat gene in the RPE also affects the expression of Crbp1, Rpe65, and RGR. These data confirm that Lrat is critical for the generation of all-trans retinyl esters and 11-cis retinoids and suggests its possible regulatory role on the expression of other proteins involved in the visual cycle. 

Materials and Methods

Targeting Vector and Homologous Recombination

A 129/SvJ mouse genomic library constructed in a vector (Lambda Fix II; Stratagene, La Jolla, CA) was screened with a gel-purified 697-bp PCR product containing the entire Lrat coding sequence. The Lrat coding sequence was amplified from RNA PolyA⁺ extracted from mouse liver using primers 5′-atgaagaacccaatgctgga-3′ and 5′-ctagccagacatcatccaca-3′. Two phages containing ∼14.8- and ∼12.9-kb genomic inserts, respectively, were isolated. The ∼12.9-kb fragment spanned the entire Lrat gene and was used for the construction of the targeting vector. A loxP site (lowercase sequence) was created by inserting a pair of complementary oligonucleotides 5′-AGACGTCataacttcgtataatgtatgctatacgaagttatCAGCT-3′ and 5′-GataacttcgtatagcatacattatacgaagttatGACGTCTAGCT-3′ into an SacI site located in exon 1, 49 nucleotides upstream of the ATG initiation codon. A NotI site was created by inserting a second pair of complementary oligonucleotides (5′-GATCAGCGGCCGCCATATGGATATCG-3′ and 5′-CTAGGGATATCCATATGGCGGCCGCT-3′) into the BamHI site located in the single intron, 35 nucleotides downstream from exon 1. A ∼1.7-kb NotI-restricted fragment containing a loxP-flanked neomycin-resistance gene (PGK-NeoA+, gift from Daniel Metzger, IGBMC; Institut de Génétique et de Biologie Moléculaire et Cellulaire) was then inserted into the created NotI site of the Lrat gene. The resultant ∼15-kb HpaI-digested purified targeting vector was electroporated into P1 embryonic stem (ES) cells as described. ²⁹ After selection with G418, 372 resistant clones were expanded. Genomic DNA was prepared from each clone, restricted with EcoRV and analyzed by Southern blot using a 5′ probe external to the targeting vector, which consisted of a 596-bp PCR fragment located ∼4-kb upstream of the ATG. An 8.4-kb fragment was obtained for the wild-type (WT) allele, whereas a 6.0-kb fragment was obtained for the targeted (L3) allele. Genomic DNA was also restricted with SacI, and analyzed by Southern blot using a 3′ probe external to the targeting vector, which consisted of an 807-bp PCR fragment located ∼4.7 kb downstream from the ATG. A 5.6-kb fragment was obtained for the WT allele, whereas a 8.0-kb fragment was obtained for the targeted allele (data not shown). One correctly targeted ES cell clone (K100-36) was injected into C57BL/6 blastocysts, and male chimeras derived from it gave germline transmission, thereby generating Lrat ^+/L3 mice. 

Animals and Genotyping

All experimental procedures were performed on 2- to 5-month-old mice (unless otherwise indicated), which were on a C57BL/6J background. All mice were maintained in a normal 12 hours on and 12 hours off light cycle and fed ad libitum with NIH-31 modified open formula, which contains 24,500 IU/kg vitamin A acetate (Harlan Teklad, Madison, WI). All experiments involving the use of mice were in compliance with the guidelines established by the National Institutes of Health and by the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Genotyping was performed by PCR amplification of genomic DNA from a tail biopsy using a combination of the primer sets described in Table 1 . The presence of the Tyrp1-Cre transgene was assessed using primers 5′-atttgcctgcattaccggtc-3′ and 5′-atcaacgttttcttttcgga-3′, which allow the amplification of a 350-bp PCR fragment. ³⁰ Note that the founder mice harboring the Tyrp1-Cre transgene were genotyped for the rd (retinal degeneration) mutation ³¹ that is present in the original FVB/N background, ³² and only mice that did not carry the rd mutation were used for breeding with Lrat ^L3/L3 mice. Also, to avoid an indirect effect on animal variability due to their rpe65 genotype, only mice homozygous for the Leu450 allele were used. 

Preparation of RPE Cells

RPE cells were collected from freshly dissected eyes under a stereomicroscope. The anterior segment, lens, and retina were first removed. The eye cup was then rinsed in calcium- and magnesium-free Hanks’ balanced salt solution (HBSS). Sheets of RPE cells were carefully peeled away from their choroidal attachment with fine forceps. Cells thus collected were transferred with a dropper pipette into fresh HBSS on ice. RPE cells were used immediately or temporarily stored at −80°C. 

Western Blot Analysis

Microsomal proteins from several freshly dissected tissues including RPE, brain, heart, liver, lung, kidney, testis, and pancreas were extracted as described before. ³³ All protein measurements were performed in a spectrophotometer (model ND-1000; NanoDrop Technologies Inc., Wilmington, DE). Protein (10 μg) from each microsomal sample was used for analysis. Native bovine RPE cells were included as positive controls for Lrat detection. Samples were boiled for 2 minutes before electrophoresis by 12.5% (wt/vol) SDS-PAGE (Bio-Rad, Hercules, CA). Proteins were transferred to nitrocellulose membrane filters (Schleicher−Schuell, Sacramento, CA). Each filter membrane was separately incubated for 1 hour at room temperature with antibodies described in Table 2 . Monoclonal and polyclonal antibodies were used at dilutions of 1:5000 and 1:1000, respectively. The specificity of these antibodies has been shown. ^{24 34 35 36 37 38 39} Protein bands were visualized by using the enhanced chemiluminescence (ECL) Western blot system (Amersham, Arlington Heights, IL). The stripping process for reprobing was accomplished by incubating the filter membranes in 0.2 M glycine (pH 2.8), at room temperature for 30 minutes followed by two, 10 minutes washes. 

Immunohistochemistry

For detection of Lrat on RPE sections, eyes were fixed by intracardiac perfusion ⁴⁰ of ice-cold 4% (wt/vol) paraformaldehyde in phosphate-buffered saline (PBS; 136 mM NaCl, 11.4 mM sodium phosphate [pH 7.4]), then kept in the same fixative for 4 hours at 4°C, washed in PBS, and frozen in OCT embedding medium. Sections (20-μm-thick) were first incubated for 30 minutes at room temperature in 20% (vol/vol) normal goat serum diluted in PBS-T buffer [PBS containing 0.1% (vol/vol) Tween-20], then incubated for 2 hours at room temperature with the Lrat mouse monoclonal antibody ²⁴ diluted at 1:50 in PBS-T and containing 20% (vol/vol) normal goat serum. Detection of bound primary antibodies was achieved by incubating the sections for 1 hour at room temperature, using a Cy3-conjugated goat anti-mouse IgG (Jackson ImmunoResearch Laboratories, West Grove, PA) diluted 1:500. Nuclei were counterstained with 4′, 6-diamidino-2-phenyl-indole (DAPI; Roche Diagnostics, Meylan, France) diluted in the mounting medium at 10 μg/mL (Vectashield; Vector Laboratories, Burlingame, CA). Sections were imaged with a confocal microscope (ST1; Leica, Wetzlar, Germany). 

Light Microscopy

Mice anesthetized with carbon dioxide were fixed by intracardiac perfusion with a mixture of 2% (wt/vol) paraformaldehyde and 2.5% (wt/vol) glutaraldehyde in 0.1 M sodium phosphate buffer (pH 7.2). Light cautery was applied at the superior pole of the cornea to mark the orientation before enucleation. A window was cut in the cornea, and the eye was immersed in primary fixative and rotated at room temperature for 2 hours. The anterior segment was removed, and the remaining eye cup was refrigerated overnight in primary fixative. The eye cup was trimmed into temporal and nasal hemispheres. The tissues were immersed in 1% (wt/vol) osmium tetroxide in 0.1 M sodium phosphate buffer (pH 7.2) for 1 hour followed by dehydration in a graded series of alcohols. The temporal hemispheres were embedded in an Epon/Araldite mixture (25 parts/15 parts). Light microscope sections were cut at 1 μm and stained with 1% (wt/vol) toluidine blue and 1% (wt/vol) sodium borate, then photographed with a light microscope (63× oil lens) (LSM 210; Carl Zeiss Meditec, Dublin, CA) equipped with a digital camera (Coolsnap; Roper Scientific, Duluth, GA). 

ERG Analysis

Mice, dark-adapted overnight, were anesthetized with an intraperitoneal injection of normal saline containing ketamine (15 mg/g) and xylazine (3 mg/g body weight). ERGs were recorded from the corneal surface of one eye after pupil dilation 1% [wt/vol] atropine sulfate in saline buffer) using a gold-loop electrode referenced to a similar gold wire in the mouth. All stimuli were generated with a photic stimulator (model PS33 Plus; Grass-Telefactor, West Warwick, RI) affixed to the outside of the dome at 90° to the viewing porthole. ERGs were recorded to a series of blue (Wratten 47 A; λ max = 470 nm) flashes of increasing intensity, ranging from −4.7 to −0.44 log cd-s/m². Responses were sampled at 1 kHz, amplified (CP511 AC amplifier, ×10,000; Grass-Telefactor) and digitized using an I/O board (PCI-1200; National Instruments, Austin, TX) in a personal computer. Signal processing was performed with custom-written software. For each stimulus, responses were computer averaged with up to 50 records averaged for the weakest signals. A signal rejection window could be adjusted online to eliminate artifacts. 

Retinoid Analysis

All procedures were performed under dim red light. The mice were killed, and samples of RPE (posterior eyecup minus retina), liver, lung, testis, and kidney were flash frozen and stored at −80°C. At time of analysis, the tissue samples (0.15–0.75 g wet weight) were thawed in 1 to 2 mL PBS (pH 7.2) and homogenized with a tissue grinder (PowerGen; Fisher Scientific, Houston, TX). Retinoids in the sample extract were analyzed by normal phase high-performance liquid chromatography, as previously described. ³³  

RT-PCR Analysis

RPE cells and other tissues including brain, heart, liver, lung, kidney, testis, and pancreas were collected from mice with the indicated genotypes. Total RNA was extracted (RNeasy; Qiagen, Valencia, CA). Reverse transcription coupled to polymerase chain reaction (RT-PCR) experiments were performed with an RNA-PCR kit, according to the manufacturer’s recommendations (Applied Biosystems [ABI], Branchburg, NJ). Primer sets specific for each gene are described in Table 3 . All experiments were performed with a temperature cycler (Robocycler 40; Stratagene, La Jolla, CA). To check for identity, PCR products were cloned into a vector (pCR II; Invitrogen, San Diego, CA), and sequenced with dye termination chemistry (DYEnamic ET Terminator Kit; GE Healthcare, Piscataway, NJ) in a genetic analyzer (Prism 3100 Genetic Analyzer; ABI). 

Results

Generation of Mice Harboring a Conditional LoxP-flanked Lrat allele

To create mice in which the Lrat gene can be selectively ablated in the RPE or in any other tissue of interest, a loxP-flanked Lrat allele was engineered in ES cells through homologous recombination (Fig. 1A) . The ES cell clone K100-36, harboring the conditional (L3) Lrat allele, was used to generate a mouse line (Lrat ^+/L3 mice). By using primers P1 and P2 (Table 1) , a 376-bp PCR product was obtained for the WT allele, whereas a 423-bp fragment was obtained from the L3 allele, thereby allowing discrimination of heterozygous and homozygous progenies from Lrat ^+/L3 intercrosses (Fig. 1B) . In addition, PCR amplification of a 741-bp fragment using primers P5 and P6 consistently confirmed the presence of the loxP-flanked PGK-NeoA+ selection cassette in mice harboring the L3 allele (data not shown). Heterozygous matings (n = 12, mean of seven pups per litter) yielded 27% (n = 23) WT (Lrat ^+/+), 49% (n = 41) heterozygous (Lrat ^+/L3), and 24% (n = 20) homozygous (Lrat ^L3/L3) mice. The latter were healthy, fertile, and indistinguishable from WT littermates. Lrat mRNA and Lrat protein levels were normal in Lrat ^L3/L3 mouse tissues such as RPE and liver (data not shown), indicating that neither the presence of loxP sites nor PGK-NeoA+ in the genome of Lrat ^L3/L3 mice affects the expression of Lrat protein (see Figs. 2 7 ). 

To check whether Cre-mediated excision at the Lrat locus can occur in vivo, Lrat ^L3/L3 mice were crossed with CMV-Cre transgenic mice, which express Cre recombinase in germ cells. ²⁸ The resultant mice (Lrat ^+/L−) transmitting the excised allele (L−) in their germline were identified by Southern blot (data not shown). Identification of heterozygous and homozygous progenies from Lrat ^+/L− intercrosses was performed by PCR analysis of tail DNA. By using primers P3 and P4 (see Table 1 ), a 346-bp PCR product was obtained for the L− allele, whereas an 877-bp fragment was obtained from the WT allele (Fig. 1C) . Heterozygous mating (n = 32, mean of six pups per litter) yielded 25% (n = 48) WT (Lrat ^+/+), 52% (n = 101) heterozygous (Lrat ^+/L−), and 23% (n = 43) homozygous (Lrat ^L−/L−) mice. The latter were externally indistinguishable from WT littermates. To verify that excision generated a null allele, we determined whether Lrat protein could be detected in RPE tissues of Lrat ^L−/L− mice. We performed Western blot analysis on microsomal proteins from WT and Lrat ^L−/L− RPE tissues using an Lrat monoclonal antibody ²⁴ (Fig. 1D , top). A protein band with a molecular mass of ∼25 kDa was detected in RPE cells of WT mice and was comparable to the one observed in bovine RPE cells used as a positive control for Lrat detection. In contrast, no Lrat protein was detected in RPE extracts from Lrat ^L−/L−mice. A smaller protein of ∼23 kDa was also observed in WT extracts, the identity of which is currently unknown. This protein is probably related to Lrat, as it is not detected in the RPE extracts of mutant mice. The expression of actin (∼43 kDa) was subsequently determined to monitor protein loading among the samples (Fig. 1D , bottom). To demonstrate further that Lrat gene disruption is efficient, we used immunohistochemistry (Fig. 1E) . When compared with a WT littermate’s eye (Fig. 1E , left), no staining for Lrat was detected in the eye of Lrat ^L−/L− adult mice (Fig. 1E , right), indicating that these mutants lack Lrat. Thus, Cre-mediated recombination of the Lrat L3 allele into L− actually produced a null allele. 

Somatic Ablation of Lrat in Mouse RPE Tissue

To selectively inactivate the Lrat gene in the RPE, we used mice carrying a transgene made of the Cre recombinase-coding sequence placed under control of the tyrosinase-related protein-1 (Tyrp1) promoter, ⁴¹ which is active in melanin-synthesizing cells and therefore drives gene expression in the RPE. ⁴² This Tyrp1-Cre transgene is active in the RPE from embryonic day 10.5 to postnatal day 12, resulting in excision of any loxP-flanked DNA segment in the RPE from embryonic day 10.5 to adulthood. ⁴² Thus, Lrat ^L3/L3 females were crossed with males bearing both the transgene and one L3 allele of Lrat (i.e., Typr1-Cre ^tg/0/Lrat ^+/L3 mice). These crosses generated control (i.e., Lrat ^L3/L3, also designated as WT) and experimental (i.e., Tyrp1-Cre ^tg/0/Lrat ^L3/L3) mice. No Lrat protein was evidenced in RPE from Tyrp1-Cre ^tg/0/Lrat ^L3/L3 animals, which were therefore referred to as Lrat ^rpe−/− mutants (Fig. 1D , top). a protein band corresponding to Lrat was detected in other tissues of Lrat ^rpe−/− mice known to process retinol including liver, testis, lung, and kidney, as well as in others tissues such as brain, heart, and pancreas (Fig. 2 , bottom). This observation indicates that somatic ablation of Lrat actually occurred selectively in RPE. In comparison, Lrat protein could not be detected in tissues from Lrat ^L−/L− mice, as expected for a germline null mutation (Fig. 2 , middle). RT-PCR experiments performed on RNA extracted from several tissues of Lrat ^rpe−/− mice confirmed that Lrat gene ablation occurred selectively in RPE tissues (data not shown). 

PCR Analysis of Lrat Cre-Mediated Excision in Tissues from Lrat ^rpe−/− Mice

Genomic DNA was extracted from RPE, brain, heart, liver, lung, testis, and pancreas of Lrat ^rpe−/− mice in an attempt to detect Lrat Cre-mediated excision in tissues other than RPE. As shown in Figure 3 , a 423-bp band (L3 allele) was observed in all tissues including RPE of Lrat ^L3/L3 controls, using the P1-P2 primer set described in Table 1(Fig. 3 , first panel). Of note, this band was absent in the RPE of Lrat ^rpe−/− mice after PCR amplification with the same set of primers (Fig. 3 , second panel). In contrast, only the RPE and no other tissue from Lrat ^rpe−/− mice showed a 346-bp band (L− allele) when using the P3-P4 primer set (Fig. 3 , third panel). This 346-bp band was observed in all tissues, including the RPE of Lrat ^L−/L− mice used as a reference in this analysis (Fig. 3 , fourth panel). 

Effect of Somatic Ablation of Lrat in RPE on the Histologic Appearance of the Retina Outer Segment and Nuclear Layer

Comparison of retina sections from WT (Lrat ^+/+) and Lrat ^L3/L3 control animals with sections from Lrat ^L−/L− and Lrat ^rpe−/− mutants at 6 and 8 weeks and at 6 months of age revealed a pronounced shortening in rod outer segment length and a slight reduction of the number the photoreceptor nuclei (Fig. 4) . These observations suggest the importance of all-trans retinyl esters generated by Lrat in the RPE under normal conditions, and the adverse consequences on the outer segments due to the absence of Lrat in RPE cells. 

Effect of Somatic Ablation of Lrat in the RPE on the ERG Response

Representative dark-adapted ERG recordings for WT, Lrat ^L3/L3, Lrat ^L−/L−, and Lrat ^rpe−/− mice are shown in Figure 5 . Figures 5A 5B 5C 5Dshow the average response to progressively higher intensities, as indicated for the WT mouse in Figure 5A . The responses for the Lrat ^L3/L3 mouse (Fig. 5B)were well formed and were of the same order of magnitude as those from the WT mouse (Fig. 5A) . The saturated b-wave amplitude, obtained from a Naka-Rushton fit, was 558.0 and 630.0 μV for the WT and Lrat ^L3/L3 mice, respectively. ERG responses were nondetectable for the Lrat ^L−/L− mouse (Fig. 5C) , as previously described for another Lrat mutant line. ²⁴ These data further confirmed that the Cre-mediated recombination of the Lrat L3 allele into L− actually produced an Lrat loss of function. In contrast, a small residual response was observed for the Lrat ^rpe−/− mouse (Fig. 5D , asterisk), but only at the brightest stimulus intensities. To confirm the existence of this small signal, ERGs were recorded to a 10-Hz flash stimuli (blue, −0.87 log cd-s/m²) under dark-adapted conditions with significant signal averaging. Representative recordings for WT, Lrat ^L−/L−, and Lrat ^rpe−/− mice are shown in Figure 5E . A recordable signal was observed in the Lrat ^rpe−/− mice that was not present in the Lrat ^L−/L− mice. The response waveforms were analyzed by fast Fourier transform (FFT), to derive estimates of signal and noise at the fundamental stimulus frequency (10 Hz). The FFT amplitude for the Lrat ^rpe−/− mouse in Figure 5Eexceeded that of noise (defined as the amplitude at twice the fundamental frequency) but did not for the Lrat ^L−/L− mouse, thereby confirming the existence of a very small residual response in the Lrat ^rpe−/− line. 

Effect of Somatic Ablation of Lrat in RPE on Retinoid Storage in the Eye Cup

The levels of endogenous retinoids in tissues known to express Lrat—namely, RPE, liver, lung, testis, and kidney—were evaluated (Table 4)as a surrogate measure of Lrat activity. In agreement with studies previously described, ^{24 25} all-trans retinyl esters were barely detectable in nonocular tissue samples from Lrat ^L−/L− mice. In contrast, nonocular tissue samples from Lrat ^rpe−/− contained levels of all-trans retinyl esters that were comparable to its control Lrat ^L3/L3 and to WT (Lrat ^+/+) control mice, except for liver and lung which showed smaller amounts. Of note, both Lrat ^L−/L− and Lrat ^rpe−/− mice contained very low, but detectable, levels of all-trans retinyl esters in the RPE (∼9% and ∼18%, respectively, of the amount measured in corresponding control mice), indicating that some residual retinol esterification activity may exist in RPE of both mutant mice (Table 4 ; Fig. 6 ). 

Effect of Somatic Ablation of Lrat on the Expression of Other Visual Cycle Proteins in RPE

To analyze the expression of genes whose products are involved in the visual cycle, we performed RT-PCR analysis on total RNA samples extracted from RPE tissues of control and Lrat mutant mice. We additionally included the genes coding for acyl coenzyme-A, diacylglycerol acyltransferase-1 (Dgat1), and Stra6, because it has recently been demonstrated that STRA6 is the long sought RBP receptor ⁶ and that DGAT1 could contribute to retinol esterification through an acyl-CoA-dependent reaction in the absence of LRAT. ⁴³ The set of primers used during this analysis are described in Table 3 . As shown in Figure 7 , the Rpe65, Rdh5, Rlbp1, Rgr, Rbp1, Dgat1, and Stra6 mRNAs coding for Rpe65, 11c-Rdh, Cralbp, Rgr, Crbp1, Dgat1, and Stra6, respectively, were amplified both from WT, Lrat ^L−/L−, and Lrat ^rpe−/− RPE tissues. Despite the absence of Lrat transcripts in the mutants (Fig. 7 , lane 2), only slight variations in the amount of amplified products were observed for these genes. 

To further analyze their expression at the protein level, we performed Western blot analysis on RPE microsomal extracts of control and mutant mice by using the set of polyclonal and monoclonal antibodies described in Table 2 . No difference was observed in the expression of Stra6 protein in RPE samples from Lrat ^L−/L− and Lrat ^rpe−/− mutant mice, compared with the respective WT (Lrat ^+/+) and Lrat ^L3/L3 control (Fig. 8 , top, lanes 1–4). Because it is known that STRA6 is normally expressed in testis, ³⁸ we used this tissue as reference to determine whether any variation among the Lrat mutants and control animals existed. Of note, the pattern of expression seen in RPE samples was also observed in protein extracts from testis of the same set of animals (Fig. 8 , top, lanes 5–8). Similarly, no variation was observed among Lrat mutants and control animals after RPE samples were reacted against Dgat1 antibodies (Fig. 8 , bottom, lanes 1–4). In this case we used Wat (white adipose tissue) as a reference, since WAT has been reported to express Dgat1 under normal conditions. ⁴⁴ Our results showed that Dgat1 is equally expressed among Wat samples of Lrat mutants and controls and was highly expressed compared with RPE tissues (Fig. 8 , bottom, lanes 5–8). The analysis of visual cycle proteins is shown in Figure 9A . The expression of Crbp1 protein was two- to threefold reduced in the RPE of Lrat ^rpe−/− mice (Fig. 9 , compare lane 3 with lane 4). Other proteins such as Rpe65, as well as Rgr, which appears to modulate isomerohydrolase activity, ⁴⁵ were highly attenuated in the RPE of Lrat ^rpe−/− mice (Fig. 9 , compare lane 5 with lane 6, and lane 7 with lane 8, respectively). In contrast, no variation of expression was found in the case of Cralbp (Fig. 9 , compare lane 9 with lane 10). Subsequent detection with an anti-actin monoclonal antibody showed that loading of the proteins was similar in each lane (Fig. 9B) . Detection with the anti-Lrat antibody showed that Lrat ^rpe−/− mutants lack Lrat in RPE tissue. Taken together, our data therefore demonstrate that ablation of Lrat in RPE alters the steady state level of other proteins involved in the visual cycle. 

Discussion

Lrat catalyzes an enzymatic reaction that transforms dietary vitamin A (all-trans retinol) into a stable retinyl ester storage form. Lrat is present in many tissues, including liver, testis, lung, small intestine, and the RPE, where it plays a key role in retinol metabolism. The mRNA from human Lrat was initially cloned and shown to encode for an integral membrane protein of 230 residues. ⁴⁶ This finding facilitated the cloning of Lrat in other species such as rat and mouse. ⁴⁷ In the eye, all-trans retinyl esters serve as the substrate for the isomerase of the visual cycle. The role of the isomerase is to generate 11-cis retinol, which is the immediate precursor of the visual chromophore used in rod and cone photoreceptors. ⁴⁸ To examine the consequences of Lrat deficiency in the visual cycle, we generated a mouse line with selective inactivation of the Lrat gene in the RPE (Lrat ^rpe−/−). For comparison, we have also generated a mouse line with germline disruption of Lrat (i.e., Lrat ^L−/L− mutants). Neither of these mutants expresses Lrat mRNA or Lrat protein in the RPE. Consequently, both lines demonstrate a pronounced dark-adapted ERG deficit compared to control mice (Lrat ^L3/L3 and WT). As expected from previous studies, ²⁴ Lrat ^L−/L− mice showed no detectable EGR response at the brightest stimuli. More important, ERG recordings for Lrat ^rpe−/− mice had intensity responses of 15 to 20 μV, which represents less than 5% of the 630 μV for the Lrat ^L3/L3 control subjects. These observations are consistent with the morphologic differences between retina sections from Lrat mutant and control mice (Fig. 4) . We conclude therefore that ablation of Lrat selectively in RPE cells is sufficient to compromise severely the response to light. 

RPE cells from Lrat ^rpe−/− mice showed a notable reduction in the expression of proteins that have close interaction with Lrat during the visual cycle process. Based on the intensity of the actin internal control in each set of samples, Crbp1 and Rpe65 had a reduction of approximately two- to threefold compared with the control, whereas in the case of Rgr, the reduction was even higher, approximately five- to sixfold. Because it is now known that RPE65, RGR and CRALBP are all involved in the regulation of isomerization activity, ^{10 11 12 45 49} it is tempting to hypothesize that these proteins function as a complex to generate and mobilize the visual chromophore. The altered expression or the absence of one of these proteins may have an effect on the stability of the others within the complex. Based on our observations, we can speculate that, under normal conditions, Lrat could play a regulatory role in the recruitment of partner proteins, and that absence of Lrat affects their steady state levels (or their stability). 

Quantification of endogenous all-trans retinyl esters was used as a surrogate measure of Lrat activity. Small amounts of all-trans retinyl esters were detected in RPE extracts of Lrat ^rpe−/− mice. This finding could be related to a partial excision of the loxP-flanked Lrat fragment by the Cre recombinase in the RPE. However, the Tyrp1-Cre transgenic line used in the present study has been shown to efficiently remove the loxP-flanked retinoid X receptor alpha (Rxra) gene in mouse RPE. ^{41 50} In addition, nondetectable levels of both Lrat mRNA and Lrat protein in the RPE of Lrat ^rpe−/− mice argues against the possibility of an incomplete gene ablation (Figs. 1D 1E 7 9) . In contrast, Lrat Cre-mediated excision in Lrat ^rpe−/− mice occurred selectively in the RPE and not in other tissues analyzed in Figure 3 . Along these lines, a residual amount of all-trans retinyl esters is also detected in the RPE of Lrat ^L−/L− mice, in which the two Lrat alleles have been deleted in the germline (i.e., a situation in which partial excision is not possible). We reasoned therefore that an incomplete Lrat loss of function in the RPE of Lrat ^rpe−/− mice could be ruled out. However, the esters detected in Lrat ^rpe−/− (and Lrat ^L−/L−) RPE may result from an alternate retinyl ester synthase activity like acyl CoA-retinol acyltransferase (Arat). ^{51 52} Although the acyl group transferred to retinol in this reaction comes from palmitoyl-CoA, the esters generated by Arat, as those generated by Lrat, maintain an all-trans stereochemical configuration. Therefore, the possibility exists that an ARAT activity accounts for the presence of all-trans retinyl ester in Lrat ^rpe−/− extracts. However, the retinyl esters levels in the RPE of Lrat ^rpe−/− (and Lrat ^L−/L−) mice represent only ∼18% (and ∼9%) of the levels in control mice, which is significantly below the 50% expected, when considering that at least 50% of the retinyl ester synthase activity in bovine RPE should be due to Arat. ⁵² In addition, it has been suggested that Arat activity in bovine RPE ⁵² and rat liver ⁵³ will be triggered only in cases when concentrations of retinol are high. The fact that retinyl ester concentrations in Lrat ^rpe−/− and Lrat ^L−/L− mice are well below wild-type levels suggests that intracellular retinol would be equally low. Therefore, the retinyl esters present in the RPE of these mice probably cannot be attributed to Arat activity. 

Another pathway for generating retinyl esters is through the activity of DGAT1, which has recently been demonstrated to display an ARAT-like activity in vitro. ⁵⁴ This enzyme is expressed in intestine, mammary gland, liver, and mainly in WAT. ⁴⁴ Although the enzyme has not been reported in the RPE, we were able to amplify RNA message coding for Dgat1 and detect its protein expression at similar levels in RPE samples from Lrat ^rpe−/− and Lrat ^L−/L− and control mice (Figs. 7 8) . Although it appears that Dgat1 has a greater affinity for monounsaturated oleoyl-CoA over the saturated palmitoyl-CoA and all-trans retinol for the generation of retinyl esters, ^{54 55} it is conceivable that this enzyme could be responsible for the all-trans retinyl esters present in Lrat ^rpe−/− and Lrat ^L−/L− RPE. The generation of a double-mutant Lrat ^L−/L−/Dgat ^−/− would help resolve this issue. In either case, the possibility of ester formation by a yet undiscovered enzyme should not be discarded. 

The most parsimonious explanation at this time could be that, because all-trans retinyl esters are still normally produced by several other tissues in these Lrat ^rpe−/− mice (Table 4) , the esters detected in RPE extracts could be coming from the blood. It is known that retinyl esters processed in the small intestinal epithelium from retinol absorbed from the diet can be incorporated into the hydrophobic core of chylomicrons. The latter are then secreted into the lymph and ultimately enter the blood circulation via the thoracic and other lymphatic ducts. ⁵⁶ The chylomicrons could then be delivered to the basolateral surface of the RPE, which has detectable levels of LDL receptors for its incorporation. According to this scenario, our data demonstrate that ∼90% of the retinyl esters present in RPE arise from RPE cell autonomous de novo synthesis, whereas ∼10% come from the blood circulation. 

In summary, we have generated three mouse lines that can be useful for the study of retinal dystrophies and retinol metabolism. The first one harbors a germline null mutation of Lrat gene, and the corresponding Lrat ^L−/L− mice display features similar to those previously reported for equivalent mouse models. ^{24 25} The second one, bearing an Lrat loss of function only in the RPE (Lrat ^rpe−/− mice), represents an important model for the study of Lrat’s impact, not only as an Acyl-transferase but also as a palmitoyl-transferase, as has been suggested recently. ⁵⁷ Finally, mice from the third line we generated conserve a functional Lrat in all tissues, but part of their Lrat encoding gene has been flanked by loxP sites (Lrat ^L3/L3 mice). This genetic modification provides a versatile model to achieve the selective inactivation of Lrat in other tissues through the additional use of various transgenic mice in which Cre-recombinase expression is controlled by a tissue-specific promoter. 

 Supported by National Eye Institute Grants EY00444 and EY00331 (DB); Foundation Fighting Blindness (DB; SN); and the Dolly Green Chair at UCLA (DB).

 Submitted for publication June 5, 2007; revised August 8, 2007; accepted September 25, 2007.

 Disclosure: A. Ruiz, None; N.B. Ghyselinck, None; N. Mata, Sirion Therapeutics (F); S. Nusinowitz, None; M. Lloyd, None; C. Dennefeld, None; P. Chambon, None; D. Bok, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Table 1.

 

Sequence of Primers Used for Genotype Analysis

Name	Sequence	Gene	Expected Size of Allele (bp)
						WT(+)	L3	L−
P1	5′-aggagaagtcagggtgcaga-3′	Lrat	376	423	NA
P2	5′-tcatcctaaggcaaccaacc-3′
P3	5′-gcactttggcttctcctctg-3′	Lrat	877	—	346
P4	5′-gtggccacacccttttattg-3′
P5	5′-ggttctccggccgcttgggt-3′	PGK-NeoA+	NA*	741	NA
P6	5′-gaaggcgatgcgctgcgaat-3′

 The position of the primers along the Lrat loci are indicated in Figure 1 . NA, not amplifiable.

Table 2.

 

Antibodies Used for Western Blot Analysis

Protein	Calculated MW (kDa)	Antibodies	Antigen Origin	Source	Reference
ACTIN	43	Monoclonal	Rat	Lin	36
CRALBP	36	Polyclonal	Bovine	Bok	Unpublished (1984)
CRBP1	16	Polyclonal	Rat	Bok et al.	35
DGAT1	57	Polyclonal	Mouse	Chen et al.	39
LRAT	25	Monoclonal	Mouse	Batten et al.	24
RGR	32	Polyclonal	Mouse	Tao et al.	34
RPE65	61	Polyclonal	Bovine	Redmond & Hamel	37
STRA6	74	Polyclonal	Mouse	Bouillet et al.	38

Table 3.

 

Sequence of Primers Used for RT-PCR Analysis

Gene	Protein	Accession Number	Sequence	Position	Size (bp)
Dgat1	DGAT1	AF078752	5′-gcgtctcttaaagctggcgg-3′	1091–1110	501
			5′-tcatacccccactggggcat-3′	1573–1592
Lrat	LRAT	AF255061	5′-atctaatgcctgacatcctg-3′	481–500	462
			5′-ctagccagacatcatccaca-3′	943–962
R1	CR1	NM_011254	5′-atgcctgtggacttcaacgg-3′	2017–2037	408
			5′-tcagtgtactttcttaaaca-3′	2406–2425
Rdh5	11cRDH	BC021372	5′-caggccaggggtcgggtggt-3′	551–570	500
			5′-tcaggagactgactgggcgg-3′	1032–1051
Rgr	RGR	NM_021340	5′-atgggacacggccatccctc-3′	411–430	504
			5′-tcactgggttcggtccttct-3′	896–915
Rl1	CRAL	NM_020599	5′-acaagtatggtcgagtggtt-3′	681–700	499
			5′-tcataaggctgtgttctcaa-3′	1161–1180
Rpe65	RPE65	NM_029987	5′-atatgtacttcctttgacaa-3′	1100–1120	502
			5′-tcaggatcttttgaacagtc-3′	1583–1602
Stra6	STRA6	NM_009291	5′-ctcttccccatcaacatgctggt-3′	1900–1923	450
			5′-gccattggccttggcactagtaag-3′	2326–2350

Figure 1.

 

Homologous recombination at the mouse Lrat locus produces a conditional allele and evidence that Cre-mediated recombination produces a null allele. (A, top) The mouse wild-type (WT, +) locus consists of two exons (boxes) separated by a single intron. Solid areas in exons: coding sequences; shaded areas: untranslated regions. (A, middle) In the targeted, conditional (L3) allele of the Lrat gene, the first loxP site is inserted in exon 1. The loxP-flanked PKG-NeoA+ cassette is inserted in the intron. (A, bottom) The excised (L−) allele, in which the Cre-mediated excision deletes the initiation codon ATG and the splice donor site of exon 1, as well as the entire PGK-NeoA+ cassette, making it nonfunctional (i.e., a null allele). Arrows: locations of primers P1 to P6 used for genotyping. The size of the amplified fragment is indicated for each pair of primers. (B) Genotyping of progeny from Lrat ^+/L3 intercrosses using primers P1 and P2. (C) Genotyping of progeny from Lrat ^+/L− intercrosses using primers P3 and P4. A DNA ladder (Invitrogen, San Diego, CA) is included to indicate the size of the amplified fragments (Markers). Right: the identity of the amplified fragments. (D) Western blot analysis of proteins from RPE tissues with the indicated genotypes using an anti-Lrat monoclonal antibody (top), and an anti-actin antibody (bottom). A bovine RPE extract (lane 1, RPE) is used as a positive control for Lrat immunoreactivity. Lrat is detected in WT (lane 2), but not in Lrat ^L−/L− (lane 3) and Lrat ^rpe−/− (lane 4) RPE extracts. (E) Immunohistochemical detection of Lrat on eye sections from adult female mice with the indicated genotypes. Lrat is present in the RPE cell layer of WT (left) but not of Lrat ^L−/L− (right) mice. The insets represent a higher magnification of an RPE cell. Note that some nonspecific staining is observed in choroidal cells of both WT and Lrat ^L−/L− mice. CH, choroid layer; PL photoreceptor layer; RPE, retinal pigment epithelium. Bar, 15 μm.

Figure 2.

 

Lrat protein detection in nonocular tissues of Lrat mutants. Western blot analysis of microsomal extracts from nonocular tissues from WT control, Lrat ^L−/L− and Lrat ^rpe−/− mice using an anti-mouse Lrat monoclonal antibody. A bovine RPE extract was included as a positive control for Lrat immunodetection. A protein band with the same molecular mass as bovine RPE is observed in all extracts from WT mice (top). The same pattern of distribution was observed in Lrat ^rpe−/− extracts (bottom). In contrast, no Lrat was detected in extracts from Lrat ^L−/L− samples (middle).

Figure 3.

 

PCR analysis of Lrat Cre-mediated excision. PCR amplification was performed on genomic DNA extracted from RPE, brain, heart, liver, lung, kidney, testis, and pancreas of Lrat ^L3/L3, Lrat ^rpe−/−, and Lrat ^L−/L− mice. A set of primers P1-P2 for detection of L3 allele (423 bp) and P3-P4 for detection of the L− allele (346 bp) were used during the amplification. Because of Cre excision of the loxP-flanked region exclusively in the RPE of Lrat ^rpe−/− mice, the P3-P4 PCR product is the same that from Lrat ^L−/L− mice in which excision has taken place in all tissues.

Figure 4.

 

Histologic features of the retina in Lrat mutants. One-micrometer sections of retina from mice with the indicated genotypes are shown. Shortening of the rod outer segments and a slight reduction in the number of photoreceptor nuclei is observed in the retina of mice lacking Lrat (i.e., Lrat ^L−/L− and Lrat ^rpe−/− mutants). IS, inner segment; ONL, outer nuclear layer; OS, outer segment; RPE, retinal pigment epithelium. Bar, 20 μm.

Figure 5.

 

ERG analysis of Lrat mutants. ERG responses to blue-flash stimuli recorded in dark-adapted conditions are shown for each group of mice. (A, B) Average response to intensities ranging from −4.7 to −0.44 log cd-s/m², as indicated for the WT mouse (A). (C, D) Records for only the brightest stimuli. ERG responses were similar for the WT (A) and Lrat ^L3/L3 (B) mice with parameters that are within normal limits. ERG responses were nondetectable for the representative Lrat ^L−/L− mouse (C). However, a barely detectable response was recordable in the Lrat ^rpe−/− mutants at the brightest flash intensities (D, asterisk). To confirm the existence of this signal, ERGs were recorded in response to 10-Hz flashed stimuli (blue, −0.87 log cd-s/m²) and analyzed by FFT, to derive the amplitude of signal and noise at the stimulus frequency. Representative waveforms are shown in (E) for WT, Lrat ^L−/L−, and Lrat ^rpe−/− mice. The FFT amplitude for the Lrat ^rpe−/− mouse exceeded that of noise but did not for the Lrat ^L−/L−, thereby confirming the existence of a very small residual response in the Lrat ^rpe−/− line.

Table 4.

 

Retinoid Contents in Tissues from Lrat Mutants

Tissue	Retinoid (nM/g)	Lrat Genotype
				WT (+/+)	L−/L−	L3/L3	RPE−/−
RPE	all-trans RE	0.21	0.02	0.48	0.09
	all-trans ROL	0.10	0.07	0.14	0.05
Liver	all-trans RE	5266	1.64	3257	1642
	all-trans ROL	85.96	1.01	39.76	14.71
Lung	all-trans RE	1000	0.93	736	577
	all-trans ROL	8.20	0.53	3.40	5.68
Testis	all-trans RE	0.88	ND	0.57	0.55
	all-trans ROL	0.24	0.54	0.27	0.46
Kidney	all-trans RE	0.48	0.17	0.28	0.31
	all-trans ROL	0.69	0.57	0.86	1.76

 Each value represents the average of duplicates of the content measured in mice of 8 to 12 weeks of age fed a diet containing 24,500 IU/kg of vitamin A. ND, not detectable.

Figure 6.

 

HPLC analysis of retinoid extracts from Lrat mutants. Chromatographic separation by HPLC of nonpolar retinoids from RPE extracts of Lrat ^L−/L− and Lrat ^rpe−/− mice compared with that of WT and Lrat ^L3/L3 controls, respectively. Peaks corresponding to all-trans retinyl esters and all-trans retinol are indicated. All-trans retinyl esters were not observed in Lrat ^L−/L− RPE tissues, whereas a minor ester peak was detected in RPE tissues from Lrat ^rpe−/− mice.

Figure 7.

 

Analysis of RNA transcripts encoding genes involved in the visual cycle. RT-PCR amplification of Rpe65, Lrat, Rdh5, Rlbp1, Rgr, Rbp1, Dgat1, and Stra6 RNA transcripts in WT, Lrat ^L−/L−, and Lrat ^rpe−/− RPE tissues. Amplification was performed using the set of primers described in Table 3 .

Figure 8.

 

Western blot analysis of Stra6 and Dgat1 in RPE tissues from control mice and Lrat mutants. Top: expression of Stra6 protein in RPE samples from WT (Lrat ^+/+), Lrat ^L−/L−, Lrat ^L3/L3, and Lrat ^rpe−/− mice. A single band of ∼74 kDa was observed in all samples analyzed. No difference in the level of expression was observed in Lrat mutants and control samples (lanes 1–4). Similar results were found in protein extracts from testis of the same set of animals and are included for comparison (lanes 5–8). Bottom: single band of ∼57 kDa was obtained in all samples analyzed. This analysis showed similar levels of expression of Dgat1 protein in RPE samples from control mice and Lrat mutants (lanes 1–4). Also, no difference was found in the levels of Dgat1 in WAT samples from Lrat mutants and control mice (lanes 5–8). However, despite loading equal amounts of protein in each track, WAT samples had stronger Dgat1 signal compared with RPE samples.

Figure 9.

 

Western blot analysis of proteins involved in the visual cycle in RPE tissue from control mice and Lrat ^rpe−/− mutants. (A) Comparison of the expression levels of Lrat, Crbp1, Rpe65, Rgr, and Cralbp in RPE between control (Lrat ^L3/L3, left lanes) and mutant (Lrat ^rpe−/−, right lanes) mice. The sizes of the immunodetected proteins are indicated in parentheses. (B) After they were stripped, the membranes were reacted with an anti-actin monoclonal antibody to confirm an equal amount of protein loaded in both groups of mice.

The authors thank Krzysztof Palczewski and Alexander R. Moise (Case Western University, Cleveland, OH) for providing the monoclonal anti-Lrat antibody; T. Michael Redmond (Laboratory of Retinal Cell and Molecular Biology National Eye Institute, Bethesda, MD) for the anti-RPE65 antibody; David Ong (Vanderbilt University School of Medicine, Nashville, TN) for the anti-CRBP1 antibody; Henry Fong (University of Southern California, Los Angeles, CA) for the anti-RGR antibody; Robert V. Farese, Jr (University of California, San Francisco, CA) for the anti-DGAT1 antibody; Pamela Stiles, Michael Chan, Ginda Phongadith (Jules Stein Eye Institute, UCLA) and Betty Feret (Institut de Génétique et de Biologie Moléculaire et Cellulaire, Université Louis Pasteur, Strasbourg, France) for excellent technical assistance.