The University of Texas Southwestern Medical Center’s (UTSWMC) Institutional Animal Care and Use Committee (IACUC) reviewed and approved all experiments and procedures used in this study. We followed the NIH’s guide for the care and use of Laboratory Animals, the Association for Research in Vision and Ophthalmology’s Statement for the Use of Animals in Ophthalmic Research, and other applicable international and national guidelines. For the forward genetics studies, germ line mutations were randomly induced using N-ethyl-N-nitrosourea (ENU) treatment of C57BL/6J (B6J) male mice (see below). Lipe−/− mice were generated using CRISPR/Cas9 mediated gene targeting on a wild-type B6J background as detailed below.
ENU mutagenesis, whole exome sequencing and determination of candidate genes
Our unbiased forward genetics protocol (summarized in a simplified diagram, Fig. 1) involves the generation of mutations randomly throughout the genome of male B6J mice using N-ethyl-N-nitrosourea (ENU)21,22. Germline mutations are identified in G1 male founders using whole exome sequencing. G1 male mice are bred to generate G3 mice, which are screened for phenotypes (see Fig. 1 of Wang et al.22). Each G3 mouse carries around 60 mutations. The zygosity of each mutation in G2 dams and in all G3 mice of each pedigree is determined before phenotypic screening by sequencing across pedigree-specific coding/splice site mutations using Ion Torrent AmpliSeq custom primer panels. In this work, the main phenotypic screening was fundus photography followed by scoring of yellow fundus spots using a special fundus spot scale (see below).
Automated meiotic mapping was performed by Linkage Analyzer software to detect statistically significant mutation-phenotype associations22. In brief, Linkage Analyzer, an R-based analysis program, performs automated computations of single-locus linkage for every mutation in the pedigree. The magnitude of a quantitative phenotype is correlated with genotype (REF, homozygous for reference allele; HET, heterozygous for reference allele and variant allele; or VAR, homozygous for variant allele) at each mutation site in all mice in the pedigree. In the setting of ENU-generated pedigrees we refer to mice as REF, HET and VAR to emphasize that these mice have many other mutations and that we are only referring to the zygosity of the specific allele in question. The software uses recessive, semi-dominant, and dominant linear regression models to assess linkage. The output of this automated mapping is a Manhattan plot of genotype-phenotype associations for every mutation in the pedigree. In the Manhattan plot, −log10 P values (y-axis) for an association to a given phenotype are plotted vs. the chromosomal positions of the mutations (x-axis) that had been identified in the G1 founders of each pedigree.
Genes were considered of interest if their Manhattan plot peaks met the following criteria: 1) the peak is above a horizontal line representing a threshold of P = 0.05 with Bonferroni correction, and 2) the peak is at least 3 logs higher than the second highest peak in the pedigree.
Candidate Explorer, a machine learning algorithm, was then used to obtain an estimate of the likelihood that the phenotype of interest really emanates from the gene in question35.
Fundus photography and fundus spot grading
We anesthetized mice using Ketamine/Xylazine cocktail (100 mg/kg-5 mg/kg) before performing the fundus screening. We used a mixture (1:1, v/v) of Phenylephrine Hydrochloride solution 2.5% and Tropicamide Ophthalmic Solution 1% to dilate the pupils. Before imaging, we applied GenTeal liquid gel (Novartis, East Hanover, NJ, USA) to the surface of both eyes to prevent corneal dehydration and to serve as coupling lens. Both eyes from each mouse were visualized and fundus photographs were captured using Micron IV retinal imaging microscope (Phoenix Micron, Inc. Bend, OR). Investigators masked to the genotypic data of the mice graded the photos using a modified version of a previously reported fundus spot scale33. Specifically, a fundus spot score for each eye is determined based on the amount of white/yellow fundus spots present as follows: no spots (score 0), 1 to 10 fundus spots (score 1), equivalent of one fundus-quadrant of spots (score 2), two to three fundus-quadrants of spots (score 3), and all four fundus-quadrants with spots (score 4). We then add the scores of both eyes for a final score of 0 to 8 for each mouse (9 potential values). The resulting scores were loaded into the Mutagenetix platform. Linkage Analyzer correlated this information with the known genetic information on mutations for each mouse, and generated Manhattan plots in which −log10 P values (y axis) are plotted vs. the chromosomal positions of the mutations (x axis) identified in the G1 founders of each pedigree.
For statistical analysis of mutation-phenotype associations, we entered phenotype data as either a categorical variable or a continuous variable. Ordered categorical variables (ordinal traits) are analyzed using binomial calculations. Continuous variables are analyzed using a linear regression model to assess linkage. We found that due to the large number of values in our scale (9 values), a continuous variable analysis can be applied to our data and is more sensitive, allowing the detection of more associations.
Generation of Lipe
−/− mouse line
CRISPR-Cas9 mutagenesis was used to generate a Lipe knock out (Lipe−/−) mouse line according to protocols described for other genes before21,22. Briefly, super-ovulated female C57BL/6 J mice were mated overnight with C57BL/6J male mice and fertilized eggs were collected the following day for in vitro transcription using Lipe single guide RNA (sgRNA) and CRISPR-Cas9 technology. A total of 103 injected embryos were cultured in M16 medium (Sigma-Aldrich) at 37 °C in 5% CO2. To generate mutant mice, 91 two-cell stage embryos were transferred into the ampulla of the oviduct (10–20 embryos per oviduct) of pseudo-pregnant Hsd:ICR (CD-1) female mice (Harlan Laboratories). Twelve pups were born, and 1 homozygous male was bred further to produce a knockout line containing the mutation shown in Supplementary Fig. S1. The mutation resulted in a 14 bp deletion (bp 220-233 of exon 2) of the Lipe gene. This led to a frame-shift mutation beginning at amino acid 109 of the protein and terminating after the inclusion of 2 aberrant amino acids. Thus, the final peptide product was predicted to be 111 amino acids in length instead of the normal 802. While these mice were viable, male Lipe−/− mice have been reported to be infertile63. We generated a colony of Lipe−/− mice by breeding Lipe−/− female mice to Lipe+/− male mice. A colony of Lipe+/+ mice was generated by first generating founder mice from crossing Lipe+/− males to C57BL/6J females. Thereafter Lipe+/+ males were crossed to Lipe+/+ females from these founders to maintain littermate controls.
Preparation of retina and RPE-choroid-scleral flat mounts (RPE flat mounts), immunostaining and counting of subretinal microglia
Mice (4 Lipe−/− and 4 control) were deeply anesthetized and both eyes were enucleated and fixed in 4% PFA at room temperature10. In brief, enucleated intact eyes were fixed in 4% PFA for 30 min, followed by an additional 30 min after removing the cornea. Then the retina and the RPE-choroid-sclera were separated and fixed for an additional 1 h. After washing 3 × 5 min in PBS, the RPE and retina flat mounts were double or triple stained at 4 °C overnight followed by incubation at RT for 2 h with the appropriate Alexa Fluor conjugated secondary antibodies. For microglia and infiltrating macrophage discrimination, we used a combination of anti-F4/80, anti-TMEM119, and anti-CCR2 antibodies. For detection of microglial activation, we used a combination of anti-Iba1, anti-CD16. (See Supplementary Table S2 for a list of antibodies, dilations used and other details).
For microglia counting the RPE flat mounts (4 Lipe−/− and 4 Lipe+/+ mice) were observed under fluorescence microscopy at 20X magnification. Iba1+ cells were counted in selected fields at four quadrants (superior, inferior, nasal, and temporal) around the optic nerve head (ONH). Four 20X fields were selected in each of the central, paracentral, midperipheral, and peripheral regions. The Iba1+ cell counts from the four fields from each of the four regions of four flat mounts (see diagram in Fig. 3a) were used for subretinal microglia analysis and comparison between the two genotypes. For morphological analysis both the RPE and retina flat mounts (4 Lipe−/− and 4 Lipe+/+ mice) were imaged using a Leica TCS SP8 confocal laser scanning microscope equipped with a Leica Application Suite X, LAS X, software (Leica Microsystems Inc.). Images were taken either at low (25X) magnification or at high (63X) magnification using a sequential scanning method.
Image-guided OCT and measuring thickness of retinal layers on OCT images
After anesthetizing mice and pupil dilation (see above), we obtained OCT images from both eyes using a Micron IV-OCT2 (Phoenix-Micron, Inc). Images obtained as described before64 were used to determine retinal thickness measurements. The thickness of several retinal layers was measured using Fiji/ImageJ (https://doi.org/10.1038/nmeth.2019). Three measurements at 100 μm intervals were taken in the center of OCT image for each of the following four OCT parameters: 1) Total Retinal Thickness (TRT), 2) Ganglion Cell Complex (GCC), 3) Outer Nuclear Layer (ONL), and 4) Outer Retinal Thickness (ORT). Using the Straight-Line tool in Fiji/ImageJ, TRT was measured from top of Bruch’s membrane (BM) to top of Internal Limiting Membrane (ILM). GCC was then measured from the top of the Inner Nuclear Layer (INL) to the top of the ILM. ORT was measured from the top of BM to the top of External Limiting Membrane (ELM). After sharpening the image to get better contrast, we then took our last measurement of ONL from the top of the ELM to the bottom of the Outer Plexiform Layer (OPL).
A separate instrument (Spectralis® OCT, Heidelberg Engineering, Heidelberg, Germany) was used for experiments trying to correlate fundus spots changes with time to OCT findings. Images were acquired according to the manufacturer protocols for registration and tracking changes in the fundus spots overtime.
Outer nuclear layer measurements on H&E-stained retinal sections prepared by either a freeze substitution or a cryosection technique
To corroborate our OCT retinal layer thickness data, we prepared histological sections of retinas from Lipe−/− and Lipe+/+ mice (n = 6 per group) as described before47. Briefly, we collected right eyes of each mouse for freeze-substitution fixation and subsequent Hematoxylin and Eosin (H&E) staining. Sequential images of the H&E sections were taken at 20x magnification on either side of the ONH using a Leica DM2000 Upright Compound microscope (Leica Microsystems, Danaher Corporation Wetzlar, Germany,) equipped with an Optronics Microfire color CCD camera (Optronics, Goleta, CA, USA). The H&E images were opened in ImageJ and the ONL thickness was measured at 300 μm interval starting from the ONH on both directions. We took an average of three measurements within a 20 μm area at each interval. The number of cells in the ONL was also counted at the 300 μm interval by using the duplicate tool in image J which covered a rectangular area making sure we included three columns of nuclei consistently on one side of the 300 μm mark. The result was divided by 3 to report an average nuclear count per column.
For ONL measurements in cryosections, eyes were enucleated from 8-month-old mice (Lipe+/+, n = 4 eyes, Lipe−/−, n = 4 eyes). The eyes were placed in the same orientation using the optic nerve head as reference in an OCT compound and immediately frozen in liquid nitrogen. The frozen samples were then processed by a routine sectioning method. The cryosections were then used for H&E staining and ONL quantitation as described above.
TUNEL assay and cone arrestin staining
Freeze-substitution fixated eyes were paraffin processed without interceding hydration. They were embedded in sagittal orientation and sectioned at 5 µm by rotary paraffin microtomy according to established procedures. Serial paraffin sections were concomitantly prepared and checked by dark-field microscopy for mid-line ocular anatomy. The resulting sections were used for Terminal deoxynucleotidyltransferase-mediated UTP End Labeling (TUNEL) and for cone arrestin staining. Positive nuclei of retinal cells possessing nicked DNA were labeled with fluorescein according to methods of first report65 and literature supplied with the DeadEnd Fluorometric TUNEL System (Promega Cat # G3250). Sections subjected to TUNEL were counterstained with propidium iodide. To check for the presence or absence of cone-photoreceptor clumping we stained retinal sections using anti-cone arrestin antibodies as a cone marker.
Electroretinogram (ERG) analyses of the visual response in Lipe
−/− and Lipe
A full-field scotopic ERG system (Celeris System, Diagnosys LLC, MA, USA) was used to record the responses of retina cells to light in Lipe−/− vs. Lipe+/+ mice after dark adaption overnight for 16 h. ERG recordings were performed under a dim red light. After anesthesia and pupil dilation, each mouse was positioned on the ERG mouse platform, which is equipped with a thermal regulator. The full-field stimulators with built in electrodes to touch each eye were then placed. We recorded 2 channels with 10 sweeps to obtain 2 standard responses/parameters of scotopic ERG (a-wave, b-wave) for both eyes. The ERG was obtained in response to moderate (0.1 log cd.s.m−2) and high (1 log cd.s.m−2) flash intensities. The ERG analysis of visual response was obtained in response to low (0.1 log cd.s.m−2) and moderate (1 log cd.s.m−2) flash intensities. The inter-stimulus interval was 0.7 s and 60 s for low and high flash intensities, respectively. The flash duration was 1 msec. A similar protocol using 3 sweeps was used to measure c-waves. After 10 min of light adaptation at a setting of 3 log cd.s.m−2, the photopic ERG measurements were obtained at 3 and 10 log cd.s.m−2 flash intensities. Ten sweeps were recorded and averaged for each flash intensity. We analyzed the ERG data using Diagnosys Espion Software (Diagnosys, Inc) for comparison of retinal function between the Lipe−/− vs Lipe+/+.
Optokinetic testing of visual function
Optokinetic testing was performed using the OptoMotry system (Cerebral Mechanics, Inc., Lethbridge, AB, Canada) in order to examine changes in visual acuity in Lipe+/+ vs. Lipe−/− mice. Visual stimuli were displayed on four LCD screens placed around a central mouse stand66,67. Each mouse is tested individually while placed without restrain on the central stand. A visual stimulus consisting of a rotating vertical sine-wave grating is presented to the mouse and the optokinetic reflex is then recorded by manual tracking of head movements. To determine spatial frequency thresholds, an increasing staircase paradigm was utilized with 100% contrast. All recordings were done by a masked investigator.
Immunoblotting and detection of Lipe protein in different tissues
We performed western blot to determine the expression of Lipe protein in ocular tissues and testis (used as positive control) of Lipe−/− and Lipe+/+ mice. Mice were euthanized (with ketamine overdose) and the eyes and testis were harvested for protein extraction34. Briefly, we dissected each eye on ice to separate cornea, retina, and RPE/choroid under a dissection microscope. After combining each tissue from both eyes per mouse we homogenized the pooled tissues using a 1 ml Dounce tissue grinder in 200 µl cold T-PER tissue lysis buffer (Catalog No. 78510; ThermoFisher Scietific, Rockford, IL, USA) containing a protease inhibitor cocktail. The homogenized tissues were centrifuged at 12,000 × g at 4 °C for 10 min. The supernatants were collected and the concentration of protein in each sample was determined using a BCA kit (Catalog No. 23225, ThermoFisher Scientific). Testes were dissected from Lipe+/+ male mice to serve as positive control and homogenized similarly. For retina and RPE/choroid samples we used protein concentrators to obtain a desired protein amount (Pierce™ Protein Concentrator, 3 K MWCO, 0.5 ml). A known amount (20–40 ug) of each sample was prepared and boiled in SDS-BME for 5 min and loaded on a 4–20% Tris-Glycine gel (Cat. No. XP04205BOX, ThermoFisher Scientific). After electrophoresis for 80 min, the proteins were transferred onto a nitrocellulose membrane and blocked overnight in Intercept (PBS) blocking buffer (Cat. No. 927-70001, LI-COR Biosciences). After removing the blocking buffer and washing 3X in TBS-T, the membrane was incubated with HSL primary antibody (Cat. No. 4107S, Cell Signaling) diluted 1:1,000 in 5% BSA (in TBS with 0.02% NaN3) overnight at 4 °C. After removing the primary antibody, the membrane was washed in TBS-T 3X and incubated with HRP chemiluminescent secondary antibody (Cat. No. 926-80011, LI-COR Biosciences) diluted 1:15,000 in 5% milk solution for 40 min at room temperature. Finally, after removing the secondary antibody solution and washing 3X, western HRP substrate (Cat. No. 34094, ThermoFisher Scientific) was added and the membrane was imaged on an Amersham Imager 600 (Amersham Biosciences, Piscataway, NJ).
RNAScope in situ Hybridization (ISH)
Probes for RNAscope ISH were designed (Advanced Cell Diagnostics, Hayward, CA USA)21 to detect the expression of Lipe RNA, Sfxn3 RNA (as a positive control/probe:Mn-Sfxn3-O1), and a negative control DapB probe (bacterial dihydrodipicolinate reductase mRNA) in Lipe+/+ mouse retina or testes. After cardiac perfusion of 4-month-old mice with 4% PFA, eyes and testes were collected. Post-fixation was done overnight in the same buffer at 4 °C, followed by routine paraffin embedding. RNAScope ISH was then performed. The probes were run with Advanced Cell Diagnostics red chromogenic & fluorescent kits. Images were taken on a Leica DM2000 microscope (Leica Microsystems, Wetzlar, Germany) at 20X magnification using a Jenoptik Gryphax CCD camera, Texas Red filter, and acquired with Progress software (V.126.96.36.199).
Transmission Electron Microscopy (TEM) imaging and analysis
We collected eyes from deeply anesthetized Lipe−/− and Lipe+/+ mice (see above) and the left eyes were processed for electron microscopy21. Eyes were fixed in 2% PFA and 2% glutaraldehyde in sodium cacodylate buffer followed by post fixation in 1% osmium tetraoxide. After trimming, dehydration and embedding in epoxy resin, 70-nm-thin sections were cut and stained with 2% aqueous uranyl acetate and lead citrate. The sections were then imaged with a JEOL 1200EX II transmission electron microscope (JEOL USA, Inc., Peabody, MA, USA)21 with the help of UTSW Electron Microscopy Core.
For quantification of melanolipofuscin and auto/phagolysosome granules (MLaPL), we opened each micrograph in Fiji/ImageJ Software and counted all aggregates of fused organelles such as melanolipufuscin, auto/phagolysosomes in RPE cells of 3 Lipe−/− and 3 Lipe+/+ mice (n = 20–26 TEM fields per group). We analyzed and reported the total numbers of these granules for each filed or averaged for each mouse. The thickness of the RPE, basal infoldings and Bruch’s membrane were also measured.
Measurement of axial length
Lipe+/+ (n = 21) and Lipe−/− (n = 18) mice were deeply anesthetized using a ketamine/xylazine cocktail. After enucleation, any residual adipose tissue around the optic nerve was removed under a dissecting microscope. Each eye was positioned on a glass plate to obtain the axial length. The length from the top of the cornea to the posterior sclera at the site of the optic nerve entry was measured three times using a digital caliper (Kynup Digital Caliper, eVatmaster Consulting GmbH, Germany). The measurements of both eyes of each mouse (6 total measurements per mouse) were averaged.
Lipid extraction and analysis from retina and RPE/choroid samples
Lipid standards such as cholesterol and cholesterol esters, triacylglycerols and diacylglycerols were purchased from MilliporeSigma (St. Louis, MO, USA), Avanti Polar Lipids (Birmingham, AL, USA) and Nu-Check Prep (Elysian, MN, USA) (see Supplementary Table S2). All other reagents used in the liquid chromatography experiments (see Supplementary Table S2) are available from MilliporeSigma, Burdick & Jackson (Muskegon, MI, USA), and ThermoFisher Scientific (Waltham, MA, USA).
Lipids were extracted from both retina and RPE-choroid-sclera samples (n = 5 for each) of Lipe+/+ and Lipe−/− mice. In brief, mice were euthanized by cervical dislocation, eyes enucleated, and then placed in cold PBS. Cornea, lens, and vitreous68 were surgically removed under a dissecting microscope. The retina was then separated from the RPE-choroid-sclera for each eye. The two retinas of each mouse were placed in a 2-mL glass HPLC sample vial with PTFE-lined cap filled with 0.5 mL of chloroform/methanol solvent mixture (2/1, vol/vol, CM2/1). The samples were stored at −20 °C until the final extraction of lipids. RPE-choroid-sclera specimens (two from each mouse) were treated in a similar fashion. Retina and RPE/choroid lipids were extracted from the respective samples at room temperature thrice, each time with 1 ml of the CM2/1 solvent mixture for 15 min. The extracts were transferred stepwise into a glass crimper-style 300 µl HPLC vial, bringing the aliquots to dryness between each step under a stream of purified nitrogen gas and keeping the vial warm at 36 °C. Finally, the combined oily residue was redissolved in 150 µl of iso-propanol in the same vial, crimped, and stored in a freezer until the analysis.
Extracted lipids were analyzed using an Acquity M-Class ultra-high performance liquid chromatograph (UPLC) and a Waters Synapt G2-Si high resolution quadrupole Time-of-Flight mass spectrometer (MS) equipped with an atmospheric pressure chemical ionization (APCI) IonSabre-II ion source with a Zspray/LockSpray housing (all from Waters Corporation; Milford, MA, USA).
The chromatograph was operated in the reverse phase UPLC mode (RP-UPLC). Briefly, each lipid sample was analyzed in two different experiments – initially on a C8 (2.1 mm × 100 mm, 1.7 μm) RP-UPLC column, and then on a C18 BEH (1.0 mm × 100 mm, 1.7 μm) Acquity RP-UPLC column (both from Waters, Corp.) using acetonitrile/iso-propanol solvent mixtures exactly as described earlier for other mouse ocular tissues69. The C8 column was used in isocratic elution experiments, while the C18 column was used in gradient elution experiments.
Lipid analytes (such as cholesterol, cholesteryl esters, tri-and di-acylglycerols) were detected in positive ion mode using the APCI technique70,71,72. Identification of selected major lipids was performed using their elemental composition derived from exact m/z values (+/−5 mDa) in the EleComp routine of the MassLynx software (from Waters Corp./Nonlinear Dynamics; Milford, MA, USA), and UPLC retention times, which were compared with the retention times of authentic lipid standards.
The results of lipidomic experiments were analyzed using a Progenesis QI (from Waters/Nonlinear Dynamics; Milford, MA, USA), EZinfo (v.188.8.131.52; from Waters/Umetrics), and SigmaStat (v.3.5; from Systat Software, Inc., San Jose, CA, USA) software packages. For untargeted (i.e. unsupervised) analysis, the raw UPLC-MS data were processed in Progenesis QI using its Principal Component Analysis (PCA) feature. The data were normalized using total ion abundances determined for each run separately. Then, the data were imported into EZinfo for subsequent processing and analysis. The EZinfo’s Orthogonal Projections to Latent Structures Discriminant Analysis (OPLS-DA) model implemented Pareto scaling. There were more than 200 unique variables (i.e. analytes with unique combinations of m/z values and LC retention times, RTs) detected, exact identification of which goes beyond the scope of this manuscript and is to be reported separately.
Serum lipid analysis
Prior to euthanasia, 10–11-month-old Lipe−/− and Lipe+/+ mice (n = 6, per group) were anesthetized, eyes were enucleated and blood was collected in 2 ml micro centrifuge tubes from the orbital sinus. Blood samples were allowed to clot at room temperature for 30 min. The samples were centrifuged at 1000 × g for 10 min to separate the clot and serum was transferred into new microcentrifuge tubes for lipid analysis with the assistance of metabolic-phenotyping core of UT Southwestern Medical Center (http://touchstonelabs.org/metabolic-phenotyping-core/). Total cholesterol (TC), high-density lipoprotein cholesterol (HDL-C), and triacylglycerols (TAGs) were quantitated for comparisons of Lipe−/− versus age-and-gender matched Lipe+/+ mice.
Statistics and reproducibility
Graphs and Statistical analysis were done using GraphPad Prism 9.4.1 and Microsoft Excel 16.63.1. Data are presented as the mean ± standard error of mean (SEM), except for Fig. 1 in which mean ± standard deviation is reported. For all figures, groups of measurements were taken from distinct samples, rather than from repeated measurements. Samples sizes are included in each figure legend. Normal distribution of the data was assumed. All comparisons were done between two groups using a two-tailed unpaired Student’s t test. A linear regression analysis was performed to check the trend of changes in the fundus spots and OCT parameters with age over time. A p value < 0.05 was considered statistically significant with the null hypothesis that there are no differences between the two groups. For the analysis of Iba1+ cells in flat mounts we provide the p-value both including all values (black asterisks) and also excluding outliers (blue asterisks). We used the interquartile range (IQR) to determine outliers, which were defined as those outside of a range going from Q1–1.5*IQR to Q3 + 1.5*IQR, where Q1 and Q3 are the first and third quartiles, respectively.
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.