Microsomal triglyceride transfer protein is necessary to maintain lipid homeostasis and retinal function

Abstract

Lipid processing by the retinal pigment epithelium (RPE) is necessary to maintain retinal health and function. Dysregulation of retinal lipid homeostasis due to normal aging or to age-related disease triggers lipid accumulation within the RPE, on Bruch’s membrane (BrM), and in the subretinal space. In its role as a hub for lipid trafficking into and out of the neural retina, the RPE packages a significant amount of lipid into lipid droplets for storage and into apolipoprotein B (APOB)-containing lipoproteins (Blps) for export. Microsomal triglyceride transfer protein (MTP), encoded by the MTTP gene, is essential for Blp assembly. Herein we test the hypothesis that MTP expression in the RPE is essential to maintain lipid balance and retinal function using the newly generated RPEΔMttp mouse model. Using non-invasive ocular imaging, electroretinography, and histochemical and biochemical analyses we show that genetic depletion of Mttp from the RPE results in intracellular lipid accumulation, increased photoreceptor–associated cholesterol deposits and photoreceptor cell death, and loss of rod but not cone function. RPE-specific reduction in Mttp had no significant effect on plasma lipids and lipoproteins. While, APOB was decreased in the RPE, most ocular retinoids remained unchanged, with the exception of the storage form of retinoid, retinyl ester. Thus suggesting that RPE MTP is critical for Blp synthesis and assembly but is not directly involved in plasma lipoprotein metabolism. These studies demonstrate that RPE-specific MTP expression is necessary to establish and maintain retinal lipid homeostasis and visual function.

Graphical Abstract

The healthy RPE (left) plays a crucial role in maintaining photoreceptor health and function, in part, via the uptake (e.g., outer segment phagocytosis, plasma lipid absorption, de novo lipid synthesis), processing (e.g., lipid oxidation, metabolism), and removal or recycling (e.g., β-lipoproteins, HDL) of prodigious amounts of lipid. However, when RPE-specific MTP expression is reduced (right), secretion of β-lipoproteins is curtailed and lipid homeostasis is disrupted, leading to lipid accumulation in the RPE and subretinal space and compromising retinal function.

Introduction

Visual function depends on the intimate structural, functional and metabolic interactions amongst the retinal pigment epithelium (RPE), neural retina (NR) and choriocapillaris (1–3). A growing body of research highlights the importance of local pathways (i.e., those operating within the eye) in the regulation of retinal lipids. In particular, the RPE is a central hub for lipid processing within the eye and acts as a gatekeeper for lipid transport into and out of the NR (1, 4–7). In this role, the RPE ingests fragments of outer segments (OS) of photoreceptors (PR) in a process historically called “phagocytosis,” but more accurately referred to as “trogocytosis,” or “nibbling,” (8). In the human retina, Volland et al. (2015) estimate that each RPE cell ingests and processes ~0.30 pmol of phospholipid per RPE cell daily (9). In addition to the processing of lipid-rich OSs, the RPE synthesizes lipids de novo, and takes up plasma lipids from the choroidal circulation through its basal membrane. The RPE also transfers lipids from these three sources to the NR through its apical membrane(3, with the recycling of the polyunsaturated fatty acids, docosahexaenoic acid (DHA) essential for function{Lakkaraju, 2020 #60, 9–11). Therefore, the RPE plays an essential homeostatic role for processing and bidirectional export of lipids, including cholesterol, for the entire RPE/NR/choroid complex(1–3, 5, 12).

Given that lipid homeostasis between the RPE and NR is critical to retinal health and, thereby, visual function, the RPE must employ mechanisms to prevent steatosis (lipid build-up) due to its considerable lipid intake. In this regard, the RPE oxidizes lipids and generates β-hydroxybutyrate (βHB), which it releases apically, providing the PR with metabolic substrate (13, 14). To facilitate lipid export to the NR and into the systemic circulation, the RPE packages lipids into APOB-containing lipoproteins (Blps) (15, 16). These Blps consist of a lipid core, rich in esterified cholesterol (EC) and triglycerides (TG), enclosed in a layer of phospholipids and apolipoproteins, including APOB. Such an arrangement facilitates en mass transport of hydrophobic lipids through the aqueous environment(17, 18). Microsomal triglyceride transfer protein (MTP), encoded by the MTTP gene, is indispensable for Blp assembly, as it lipidates nascent APOB into Blps (17). In the absence of MTP, APOB is quickly degraded(18). The role of MTP in Blp secretion has been well-studied in hepatocytes, enterocytes, and cardiac monocytes, and the loss of MTP in these cell types leads to steatosis(19). MTP is also expressed in the RPE (16), but its role in local retinal lipid homeostasis has not been well-studied in vivo.

Numerous studies encompassing in vitro systems, in vivo models, and patient data have indicated that defects in lipid processing pathways and metabolic dysregulation in the RPE contribute to aging, retinitis pigmentosa and age-related retinal disease (15–17, 20–22). Mouse models of defective phagosome maturation and of dysfunctional lipid processing pathways, have documented RPE steatosis-like lipoidal degeneration(23, 24), loss of RPE and/or PR function, and inflammation (5, 6, 20). For instance, RPE-specific ablation of the ATP-dependent transporter ABCA1, which exports cholesterol to the NR, (3, 10, 11) results in RPE lipid accumulation, loss of RPE and PR function, and retinal degeneration (2). In age-related-macular-degeneration (AMD), a critical role of lipids is highlighted by the presence of lipids in pathognomonic extracellular deposits called drusen (15, 20). The role of RPE-specific lipid processing pathways in health and disease, such as in the development of subretinal and sub-RPE deposits, remains to be fully explored.

Understanding how local lipoprotein production and secretion forestalls steatosis in the outer retina is critical for identifying therapeutic targets for retina–specific lipid accumulation. The objective of the current study is to assess the role of RPE-specific MTP expression in maintaining retinal lipid homoeostasis. We hypothesized that Blp assembly in the RPE, a local mechanism to regulate lipid processing and transport, is necessary to maintain retinal health and function. To test this hypothesis, we generated the RPEΔMttp mouse model which has an RPE-specific MTP deficiency. With a reduction of MTP expression, RPE of RPEΔMttp mice accumulated large amounts of lipids, suggesting that a major function of MTP in the RPE might be to prevent lipid accumulation by exporting lipids out of the eye and into the choroidal circulation. The inability to properly process lipids for expert, resulted in diminished rod but not cone function. Our results highlight the essential role of MTP-mediated Blp assembly in the RPE and demonstrate the pathogenic consequences of decreased lipid transport via Blps assembled in the RPE.

Materials and Methods

Animals

Mice homozygous for the floxed Mttp allele (Mttp^flox/flox) on a C57Bl/6J background(25–29)were crossed with a transgenic line that expresses Cre recombinase under control of the human bestrophin 1 (BEST1) promoter (C57BL/6-Tg(BEST1-cre)1Jdun/J, referred to as Best1-Cre; Jax stock no. 017557) (30) which were purchased from Jackson Laboratories (Bar Harbor, ME, USA). Two-generation crosses were used to produce mice that were Mttp^flox/flox and Cre transgenic (Mttpf^lox/flox; BEST1-Cre^+/−, hereafter referred to as RPEΔMttp). In all of our experiments, RPEΔMttp mice were compared to age-matched controls, which were Mttp^wt/wt and Cre transgenic (Mttp^wt/wt; BEST1-Cre^+/−, hereafter referred to as controls). Mouse lines were confirmed to be free of retinaldegeneration- related mutations, rd8 and rd10, by genotyping (Transnetyx, Cordova, TN,USA) (31, 32). The mice were genotyped for Mttp and Cre by PCR on tail genomic DNA using the following primers:

• Mttp primers:

  • MTP forward: 5’ – GAGCCTGTTGAGCAAGTACAAG – 3’
  • MTP reverse: 5’ – GGCAGCAGGACAGAGACAC – 3’
• Cre primers

  • Best1-Cre F: 5’ – TCGATGCAACGAGTGATGAGG – 3’
  • Best1-Cre R: 5’ – GGCCCAAATGTTGCTGGATAG – 3’

Maintenance of mouse colonies and all experiments involving animals were as described previously (23). Mice were housed under a 12-hr light/dark cycle and fed ad libitum (23) with both female and male mice used in these studies. All procedures involving animals were approved by the Institutional Animal Care and Use Committee at the University of Pennsylvania and were in accordance with the Association for Research in Vision and Ophthalmology (ARVO) guidelines for use of animals in research.

Antibodies

Commercially available primary antibodies used are shown in Table 1. Secondary antibodies used are: goat anti-Mouse and goat anti-rabbit horseradish peroxidase (HRP)-conjugated antibodies (Invitrogen, Waltham, MA, USA) for immunoblotting, and donkey anti-mouse, anti-rabbit, anti-goat IgG Alexa Fluor 488/ 594/ 647 conjugates (Invitrogen) for immunostaining experiments.

Table 1.

list of abbreviations

Abbreviation	Description
11cisRAL	11-cis-retinal
ABL	abetalipoproteinemia
AFF	autofluorescent foci
AMD	age-related macular degeneration
APOB	apolipoprotein B
atRAL	all-trans-retinal
atRE	all-trans-retinyl ester
atROL	all-trans-retinol
BAF	blue autofluorescence
Blps	apolipoprotein B-containing lipoproteins
BrM	Bruch’s membrane
Ch	choroid
cSLO	confocal scanning laser ophthalmoscopy
DHA	docosahexaenoic acid
EC	esterified cholesterol
ELM	external limiting membrane
ERG	electroretinography
FOV	field of view
IHC	immunohistochemistry
IRDF	infrared dark-field
IS	inner segments
LD	lipid droplets
LDL	low-density lipoproteins
MTP	microsomal triglyceride transfer protein
NR	neural retina
ONL	outer nuclear layer
OS	outer segments
OTAP	osmium-tannic acid-paraphenylenediamine
PFA	paraformaldehyde
PNA	peanut agglutinin lectin
PR	photoreceptors
RPE	retinal pigment epithelium
RPE/Ch	retinal pigment epithelium/choroid
SD-OCT	spectral domain optical coherence tomography
SDD	subretinal drusenoid deposits
SEM	standard error of the mean
TEM	transmission election microscopy
TG	triglycerides
UC	unesterified cholesterol
VLDL	very low-density lipoprotein

RNA In situ Hybridization

Mttp RNA in situ hybridization was performed on fixed frozen sections using the BaseScope 2.5 HD assay kit (Advanced Cell Diagnostics, Newark, CA)(33, 34). A custom-designed BaseScope probe targeting bases 855–989 of the mouse Mttp transcript, NM_008642.3, which span exon 5 and part of exon 6 was used for the assay (BaseScope^™ Probe- BA-Mm-Mttp-3zz-st-C1; 1217461-C1; Advanced Cell Diagnostics). Briefly, mouse retinal cryosections (10 μm thick) from ~12-month-old RPEΔMttp and control mice were post-fixed in 4% paraformaldehyde (PFA) for 60 mins at 4°C, followed by dehydration in ethanol, bleaching (10% H2O2 at 60°C), target retrieval (5 min), dehydration, and Protease III treatment (15 min at 40°C). The sections were incubated with the BaseScope probe for 2 h at 40°C, followed by Hybridize BaseScope^™ v2 AMP steps 1–8 and signal detection per manufacturer’s instructions (BaseScope^™ Detection Reagent Kit v2 – RED; Document # 323900-USM; Advanced Cell Diagnostics). The sections were counterstained with Hoechst 33258 (1:10,000) and mounted with Prolong Gold Antifade Mountant (P36930, Molecular Probes, Eugene, OR, USA).

Immunoblotting.

RPE/choroid (RPE/Ch) lysates were prepared in RIPA buffer with 1% protease inhibitor mixture (P8340, Sigma-Aldrich, Burlington, MA, USA) and 1% phosphatase inhibitor cocktail 2 (1862495, Thermo Fisher Scientific, Waltham, MA, USA). Lysates of (20–30 μg of protein) were separated on NuPAGE 4–12% Bis-Tris gels (Invitrogen) under reducing conditions and transferred to PVDF membranes (Millipore, Burlington, MA, USA). Membranes were blocked with 5% milk in 0.1% Tween-20 in TBS (TBST) for 1 h at room temperature and incubated with primary antibodies diluted in blocking buffer (see Table 1) overnight at 4°C. Membranes were washed and incubated with secondary antibodies diluted 1:2,500 in blocking buffer for 1 h at room temperature. Blots were developed using ECL SuperSignal^® West Femto extended duration substrate (Thermo Fisher Scientific) and captured on Odyssey Fc (LI-COR Biosciences, Lincoln, NE, USA) and quantified as previously described (13).

Immunostaining

Immunostaining was performed on mouse retinal cryosections or RPE/Ch flat mounts as previously described (23). For retinal cryosections, mice were euthanized and enucleated. The eyes were incised at the ora serrata and fixed in 4% PFA in PBS (pH 7.4) for 18 h at 4°C, followed by washes with PBS to remove fixative. The eyes were then cryoprotected in 30% sucrose, embedded in OCT, and stored at −80°C. Cryosections were prepared by radially sectioning the blocks at 10 μm thickness. RPE/Ch flat mounts were prepared by separating the NR from the RPE, followed by fixation in 4% PFA for 30 min at room temperature. Sections and flat mounts were permeabilized and blocked with 5% BSA in 0.2% Triton X-100 in PBS (PBST) at 37°C for 1 h, incubated with primary antibodies diluted in blocking solution (see Table 1) at 4°C overnight, washed three times with PBST, incubated in appropriate secondary antibodies conjugated to Alexa Fluor dyes (Invitrogen, 1:1,000) and Hoechst 33258 (1:10,000) diluted in blocking buffer at 37°C for 1 h and washed three times with PBST. In some experiments with cryosections, Alexa Fluor 647 phalloidin (1 Unit/200μl) (A22287, Invitrogen) was included in the secondary antibody step. For immunostaining of cone matrix sheaths, fluorescein-labeled peanut agglutinin lectin (PNA, FL1071, Vector Laboratories, Newark, CA, USA) was added with the primary antibodies at a dilution of 1:250.

Lipid Staining

For cholesterol staining with filipin, retinal cryosections were air dried, rinsed in PBS, and permeabilized in 0.5% Triton X-100 in PBS for 20 min at 37°C, followed by PBS washes. The sections were then incubated in 60 μg/ml filipin (F9765, Sigma-Aldrich) for 2 h at room temperature in the dark, followed by PBS washes (23). The sections were counterstained with 5 μM DRAQ5 (4084, Cell Signaling Technology), washed in PBS, mounted in Prolong Gold Antifade and imaged. For LipidTOX^™ staining, flat mounts were incubated in HCS LipidTOX^™ Red (H34476, Thermo Fisher Scientific) diluted 1:200 in PBS for 30 min at room temperature and washed in PBS. Flat mounts were mounted in Prolong Gold Antifade Mountant and imaged.

Confocal Imaging and Quantification

Images were captured on an A1R laser scanning confocal microscope (Nikon, Minato City, Tokyo, Japan) with a 10X (NA 0.45) or 20X (NA 0.75) dry objective, a PLAN APO VC 60X (NA 1.2) water objective, or a CFI60 Plan Apochromat Lambda 100X (NA 1.45) oil objective at 18°C. For quantification of APOB staining, images were analyzed using Nikon Elements 5.30.03 AR software (23). Background subtraction was performed on the z-stack images, and the mean ROI intensity for each field of view (FOV) was determined (7 FOV from 2 mice/genotype).

To quantify neutral lipids, as stained by LipidTOX, in the RPE flat mounts from 8-month-old and 3–4-month-old mice, LipidTOX-positive structures were delineated on the summed LipidTOX channel z-stacks using the Threshold and Watershed tools of the FIJI distribution of ImageJ, and the area of each LipidTOX-positive structure was measured. In the 3–4-month-old mice, PLIN2-positive lipid structures were identified in five fields of view for each genotype using a process based on the method described by (35). Briefly, the maximum pixel intensity of the summed PLIN2 channel z-stacks within each LipidTOX-positive structure was measured, and LipidTOX-positive structures with a maximum PLIN2 pixel intensity above a chosen threshold were determined to be PLIN2-positive. The mean fraction of LipidTOX-positive structures that were also PLIN2-positive in each field of view based on number and area was calculated and compared between RPEΔMttp and control mice.

Images were captured on an A1R laser scanning confocal microscope (Nikon, Minato City, Tokyo, Japan) with a 10X (NA 0.45) or 20X (NA 0.75) dry objective, a PLAN APO VC 60X (NA 1.2) water objective, or a CFI60 Plan Apochromat Lambda 100X (NA 1.45) oil objective at 18°C. For quantification of APOB staining, images were analyzed using Nikon Elements 5.30.03 AR software (23). Background subtraction was performed on the z-stack images, and the mean ROI intensity for each field of view (FOV) was determined (7 FOV from 2 mice/genotype).

APOB and Cholesterol Quantification

APOB protein levels in RPE and retinal lysates were measured using an Apo B ELISA Kit (ab230932, Abcam Cambridge, UK). Briefly, tissues were homogenized in chilled 1× Cell Extraction Buffer (5X PTR; ab193970, Abcam), incubated on ice for 20 minutes, and centrifuged at 18,000-× g for 20 minutes at 4°C. Cleared supernatant was collected, and protein concentration was measured using Micro BCA^™ Protein Assay Kit (23235, Thermo Fisher Scientific). Samples were diluted to 200 μg/ml in 1x Cell Extraction Buffer, and the assay performed per the manufacturer’s specifications.

Total cholesterol (esterified and unesterified) in neural retina lysates was measured using Amplex^™ Red Cholesterol Assay Kit (A12216, Invitrogen) per the manufacturer’s instructions as we have described previously(36). Intensity measurements were performed with an Infinite 200 Pro plate reader (Tecan Group Ltd., Männedorf, Switzerland) with at λex=555nm and λem=580nm.

Transmission Electron Microscopy (TEM)

RPEΔMttp and age-matched Best1-Cre control mice were enucleated, the anterior segment removed, and eyecups immersed in fixative (2.5% glutaraldehyde and 2%PFA, in 0.1 M cacodylate buffer) overnight at 4°C, rinsed with sodium cacodylate buffer (pH 7.4), trimmed into ~2 mm pieces and post-fixed in 1% osmium tetroxide. Samples were, stained with uranyl acetate, dehydrated and processed for embedding into Epon as described (37–39). In some studies, samples were fixed using the osmium-tannic acid-paraphenylenediamine (OTAP) method to preserve neutral lipid as described {Guyton, 1988 #309;Jiang, 2015 #310. Semi-thin sections (~1μm thick) were obtained along the dorso-ventral axis, stained with toluidine blue, and imaged using Nikon Eclipse 80i microscope with 40X and 100X objectives. Ultrathin sections (60–80 nm) were imaged with a JEM-1010 transmission electron microscope (Jeol, Akishima, Tokyo, Japan). Post-fixation, embedding, and ultrathin sectioning were performed by the University of Pennsylvania Electron Microscopy Resource Laboratory.

In vivo Retinal Imaging

The retinas of RPEΔMttp and age-matched control mice were imaged at 3.5, 5 or 7-months of age using a Spectralis HRA confocal scanning laser ophthalmoscope (cSLO; Heidelberg Engineering, Inc., Franklin, MA, USA) and Bioptigen Envisu R2200 ultra-high resolution (UHR) spectral-domain optical coherence tomography system (SD-OCT; Leica Microsystems, Deerfield, IL, USA) {Dhingra, 2018 #6}. Mice were prepared for in vivo imaging by applying one drop of a 2:1 mixture of 1% Tropicamide HCl (Benzeneacetamide, N-ethyl-α-(hydroxymethyl)-N-(4-pyridinylmethyl) and 2.5% Phenylephrine (Akorn, Inc., Lake Forest, IL, USA) to both eyes several minutes prior to anesthesia. Anesthesia was achieved with an intraperitoneal injection of 95 mg/kg of Ketamine HCL (2-(2-chlorophenyl)-. 2-(methylamino)-cyclohexanone, Dechra Veterinary Products, Overland Park, KS, USA) and 10–11 mg/kg of Xylazine HCL (N-(2,6-dimethylphenyl)-5,6-dihydro-4H-1,3- thiazin-2-amine, Akorn, Inc.). Following general anesthesia, corneas received topical anesthesia with 1% Tetracaine (Alcon Laboratories, Inc., Ft. Worth, TX, USA) followed a few minutes later by application of Refresh Artificial Tears (Allergan, Irvine, CA, USA) in conjunction with protective eye shields to prevent corneal desiccation and media opacity development (40). Mice were then relocated to the Spectralis imaging platform for cSLO imaging.

Two cSLO images were collected using a Ultrawide Field (UWF, Heidelberg Engineering, Inc., Franklin, MA, USA) 105° lens that provided ~3.4 mm field of view (FOV) of the mouse fundus. The first image was a dark-field (i.e. cross-polarization) infrared IR reflectance image used to center the optic disk in the real-time field of view window and train the focus on the RPE-choroid interface. Following alignment, a blue autofluorescence (BAF) image, which has a 486 nm raster scanned laser for excitation and 500–680 nm bandpass range for emission collection, was acquired. Twenty-five frames were acquired in high-speed mode (768 × 768 pixels) from each mouse retina and automatically co-registered and averaged in real-time using the Heidelberg Eye Explorer (HEYEX 1) software automatic real-time (ART) processing feature. Averaged images were auto-normalized for best contrast between hypo- and hyper-fluorescent/reflective features. Images were exported as TIFF files.

Following cSLO, SD-OCT images with a 45° FOV (~1.4 mm) were collected with the optic nerve centrally positioned in the image FOV. Orthogonal B-scans (1400 A-scans/2 B-scans × 15 frames/B-scan) were collected at 0° and 90° to capture the horizontal and vertical meridians through the optic disk. Upon completion of the imaging, mice were administered 1.5 mg/kg of atipamezole HCL (Modern Veterinary Therapeutics, LLC., Miami, FL, USA) to induce recovery from general anesthesia. Eyes were covered with petrolatum-based eye ointment (Puralube Vet Ointment, Dechra Veterinary Products) to protect the cornea from desiccation during recovery.

BAF-cSLO images were analyzed in ImageJ. Images were converted to 8-bit grayscale and a background subtraction applied using the ImageJ built-in tool and a rolling ball radius of 20. Mean intensity was then measured for each image, and hyperreflective spots in IRDF-cSLO images and autofluorescent foci (AFF) in BAF-cSLO images were manually counted in ImageJ.

Fifteen individual SD-OCT B-scan image frames from each orthogonal B-scan were co-registered and averaged using InVivoVue software v2.1 (Leica Microsystems). Mean outer nuclear layer (ONL) thickness and inner segment/outer segment length were measured from each regional quadrant (superior, inferior, temporal and nasal) halfway from the optic disk to the edge of the image window using the straight-line measurement tool.

Electroretinographic (ERG) Analysis

Mice were dark-adapted overnight for scotopic ERG testing, prepared in the same manner as for in vivo retinal imaging, and placed on a stage maintained at 37°C. Custom-made clear plastic contact lenses with embedded platinum wires served as recording electrodes, and a platinum wire loop inserted into the animal’s mouth served as the reference electrode. ERGs were performed using an Espion E3 ColorDome ganzfeld apparatus (Diagnosys LLC., Lowell, MA, USA). For scotopic a- and b-wave responses, seven steps of increasing flash illuminance (−3.66 to −0.19 log cd·s/m²) were presented. Ten successive trials were averaged for each of the four lowest intensity flashes, and five successive trials were averaged for each of the three highest intensity flashes. b-wave amplitude was measured from the a-wave minima to the b-wave maxima at ~5 and ~40 ms following flash stimulus onset, respectively. Upon completion of dark-adapted testing, a steady 30 cd/m² adapting field was presented in the ganzfeld bowl. Following 7 min of light adaptation, cone ERGs were recorded using strobe-flash intensities of 0, 0.3 and 0.6 log cd·s/m² on a steady 30 cd/m² background. Twenty successive trials were averaged for each of the first two flashes, and 15 successive trials were averaged for the highest flash. Cone b-wave amplitude was measured from the pre-stimulus baseline to the positive peak of the waveform following the flash stimulus(23).

Retinoid Analysis

For quantitation of ocular retinoids, mouse eyes (1 eye per sample) were homogenized in PBS containing 100 mM 0-ethylhydroxylamine HCL and neutralized with 4N NaOH to pH 6.5 essentially as described (41, 42). Subsequently, 1 ml methanol was added and all-trans-retinol acetate was added as an internal standard. After solubilization with hexane and centrifugation the sample was dried under argon and dissolved in acetonitrile. The sample was then analyzed using a reverse phase column (CSH C18 column, Waters) and a Waters Acquity UPLC system with monitoring at 320 and 360 nm and the use of gradients of water (A) and acetonitrile (B) containing 0.1% of formic acid as follows: 0–5 min, 60% B; 5 to 60 min, 60–70% B; 60 to 70 min, 70 to 100% B; and 70 to 90 min, 100% B min (flow rate of 0.3 ml/min). Absorbance peaks were identified by comparison to external standards.

Plasma Lipid Analysis

Blood was collected directly from the heart using a heparin-coated syringe and plasma was separated by centrifugation. Total plasma cholesterol (23-666-202) and TG (23-666-412) concentrations were enzymatically measured using kits (Thermo Fisher Scientific) as described (26, 29).

Statistical Analyses

All graphs and statistical analyses were performed using GraphPad Prism version 9.5.1. (Graph Pad, Boston, MA, USA). t-tests were performed between samples to determine p-values with an alpha-value of 0.05.Two-way ANOVA was performed when comparing more than 2 different variables. All data were analyzed and graphed as Mean ± SEM, unless indicated otherwise in the figure legend. A value of p ≤ 0.05 was considered significant and represented as * p < 0.05, ** p < 0.01, *** p < 0.001, no asterisk: p > 0.05. SD-OCT data were analyzed using a One-way ANOVA with Tukey’s Multiple Comparisons Test in Graphpad Prism v10.1.1.

Results

Decrease in local Apo-B containing lipoprotein (Blp) in the RPEΔMttp mice

Blps are a major component of retinal degeneration-associated deposits (22). To test the hypothesis that the local assembly of lipids into Blps supports RPE and PR health and maintains visual function, we generated transgenic mice with an RPE-specific knockdown of Mttp. To generate RPEΔMttp mice, Mttp^flox/flox mice (with loxP sites flanking exon 5 and 6 of Mttp; NM 008642.3) were crossed with BEST1-Cre transgenic mice (25, 30). To confirm the excision of the Mttp gene in RPEΔMttp mice, we examined the expression of Mttp in the RPE by in situ hybridization. In control mice, Mttp BaseScope signal was detected in the RPE, while the RPE of RPEΔMttp mice produced little detectable signal (Fig. 1A). As expected, MTP protein (97kDa) was barely detectable in the RPE of RPEΔMttp mice compared to RPE of control mice (immunoblots show ~80% decrease) (Fig. 1B). Consistent with studies in other cell types showing that MTP is required for the assembly of Blps, (26–29, 43), decrease in RPE Mttp expression was associated with a decrease in Blp levels, as measured by APOB ELISA, in the RPE/Ch (Fig. 1C, right). APOB levels in the NR were not significantly different between RPEΔMttp mice and controls (Fig. 1C). In flat mounts immunostained for APOB, the RPE of RPEΔMttp mice showed ~60% lower APOB intensity compared to control mice (Fig. 1D). Of note, Blps can be delivered to the RPE via the choroidal circulation, which could partially explain decreased rather than complete loss of Blp associated with the RPE/Ch fraction. Collectively, our results confirm the reduction of Mttp expression and corresponding decrease in Blps in the RPE of RPEΔMttp mice.

Figure 1. Knockdown of Mttp in the RPE leads to decreased β-lipoproteins in the RPE.

A. RNA in situ hybridization to Mttp expression in cryosections from 12-month-old RPEΔMttp mice and controls. Signal from the Mttp transcript (red) was detected in the RPE of control mice but diminished in the RPE of RPEΔMttp mice; Hoechst nuclear stain (blue), transmitted light (upper images); white dashed lines indicate apical and basal membranes of RPE layer. OS: outer segments, RPE: retinal pigment epithelium, Ch: choroid B. Immunoblot analysis of protein lysates from the RPE/chof RPEΔMttp mice and controls probed for MTP and normalized to β-actin. C. APOB decreases in RPEΔMttp RPE but not in RPEΔMttp retina. APOB concentrations in the RPE/chand NR of RPEΔMttp mice and controls were measured by ELISA in 9-month-old mice. Values shown are mean APOB concentration (± SEM) as a percentage of control in the RPE/ch (left, t-test, p < 0.01) and neural retina (right, t-test, p > 0.05) normalized to total protein (n = 3). D. Representative immunofluorescence images showing RPE flat mounts from 4-month-old RPEΔMttp mice and controls stained for p-cadherin (P-CAD, green) and APOB (red) with Hoechst nuclear stain (cyan); white arrows indicate APOB-stained particles. Mean APOB intensity per FOV (mean± SEM) was quantified (t-test, p < 0.05).

RPE-specific MTP knockdown has no effect on systemic lipids or most ocular retinoids

To determine if reduced Mttp expression in the RPE altered systemic lipid concentrations, we measured cholesterol and TG levels in the blood plasma of RPEΔMttp mice and age-matched controls. Mean systemic cholesterol and TG concentrations were not significantly different between RPEΔMttp mice and controls (Fig. 2A). Quantification of ocular retinoids, showed no significant difference in total retinoid, all-trans-retinal (atRAL), all-trans-retinol (atROL), and 11-cis-retinal (11cisRAL),concentrations between RPEΔMttp and control eyes (Fig. 2B). In contrast, all-trans-retinyl ester (atRE), the storage form of retinoid was decreased by ~30% in the RPEΔMttp eyes compared to control (Fig. 2C). atRE is associated with lipid droplet-like structures (44). These results suggest that decreased RPE-MTP expression does not affect systemic lipid levels and that the changes we observed in the NR and RPE of RPEΔMttp mice are most likely due to the RPE-specific loss of MTP, rather than alterations to systemic lipid levels or vitamin A deficiency.

Figure 2. RPE-specific knockdown of Mttp does not modulate plasma lipid but reduces ocular retinoid storage.

A. Concentrations of cholesterol (left, t-test, p > 0.05, ns) and triglycerides (right, t-test, p > 0.05) in blood plasma of 3 to4-month-old were not significantly different between RPEΔMttp mice and controls (n = 4). Values are mean (± SEM). B. Mean all-trans-retinal (atRAL), all-trans-retinol (atROL), 11-cis-retinal (11cisRAL), and total retinoids concentrations (± SEM) in whole eyes from 3 to 4-month-old mice were not significant different between RPEΔMttp mice and controls (n = 3, two-way ANOVA, p > 0.05, ns). C. All-trans-retinyl ester (atRE) concentrations were decreased in RPEΔMttp mice. Mean atRE concentration in whole eyes from 3 to 4-month-old RPEΔMttp mice and controls (n = 3, t-test, p < 0.01)

Retinal degenerative changes in the RPEΔMttp mice

We next examined the impact of RPE-specific MTP depletion on retinal integrity in the RPEΔMttp eye using in vivo imaging using blue autofluorescence- (BAF-) and infrared dark field- (IRDF-) cSLO (23, 45). The BAF-cSLO channel utilizes a blue excitation wavelength (~486 nm) to excite retinal bisretinoids. Bisretinoids and bisretinoid byproducts accumulate over time in lipofuscin granules within the RPE as a result of photoreceptor outer segment shedding and phagocytosis by the apical RPE (46). Lipofuscin is the most dominant fluorophore contributing to the BAF-cSLO. Increases in fluorophore concentration and granule size and distribution over time and/or due to developing pathology contribute to increased autofluorescence signal and BAF-cSLO background heterogeneity or mottling. This can be further accentuated by activated microglia that infiltrate into the sub-retinal space as a result of stress or dysfunction. IRDF-cSLO channel utilizes near infrared wavelength (~815nm) that penetrates beyond the RPE and into the choroid; it works in a cross-polarized, dark field manner generating brighter/hyper-reflective features in the presence of polarization altering structures. Both BAF-and IRDF-cSLO images indicated a change in autofluorescence and reflective heterogeneity in 5- and 7-month-old RPEΔMttp mice compared to age-matched controls (Fig. 3A). Auto-fluorescence observed in RPEΔMttp mice appeared more spatially heterogeneous than in controls (Fig. 3A, green arrows). Such profound changes often originate from cellular alterations at the RPE level, giving rise to a mottled spatial appearance(23). BAF-cSLO images were analyzed for mean autofluorescence signal and mean autofluorescence foci (AFF) counts. Assessment of mean grayscale intensity of BAF-cSLO images from 7–8-month-old RPEΔMttp mice was significantly higher relative to age-matched controls as well as to 5-month-old RPEΔMttp mice (SFig. 1A). No significant change was detected in mean AFF counts (SFig. 1B), although the mean hyperreflective spot count indicated a statistically significant increase in 7–8-month-old RPEΔMttp mice relative to 5-month-old RPEΔMttp mice (SFig. 1C).

Figure 3. Photoreceptors degenerate in RPEΔMttp mice..

In vivo ocular imaging of 5-month-old and 7-month-old RPEΔMttp mice and controls. A. Representative blue autofluorescence- (BAF-) and infrared dark field- (IRDF-) cSLO images with hyper-autofluorescent or hyperreflective foci (green arrows), respectively B. Representative SD-OCT images with a hyperreflective lesion visible in outer retina of the 7 month-old RPEΔMttp mouse (gold arrow). NFL/GCL: nerve fiber layer/ganglion cell layer, IPL: inner plexiform layer, INL: inner nuclear layer, OPL: outer plexiform layer, ELM: external limiting membrane, ONL: outer nuclear layer, IS/OS: photoreceptor inner/outer segments C. ONL thickness decreased in RPEΔMttp mice. Mean ONL thickness (± SEM) measured from SD-OCT scans of RPEΔMttp mice and controls at 5 months (t-test, p < 0.01) and 7 months (t-test, p < 0.001).

Using SD-OCT, we monitored changes in the retina and show representative B-scans from the horizontal meridian for 5- and 7-month-old RPEΔMttp eyes and age-matched controls (Fig. 3B). We highlighted a region of interest just to the nasal side of the optic nerve (Fig. 3B, gold box) that contains a hyperreflective feature indicative of intra-retinal pathology (gold arrow) in the 7-month-old RPEΔMttp mouse. This lesion appears to incorporate inner segment (IS)_ and outer segment (OS) bands of photoreceptor cell attributable OCT lamina processes observed in clinical studies(47). Similar changes were not detected in 3.5-month-old RPEΔMttp mice (S.Fig.2). To determine if outer retinal abnormalities correlated with photoreceptor degeneration, ONL thickness was measured from the same SD-OCT B-scans. At 5-months of age, the ONL thickness of RPEΔMttp mice was reduced by over 1.5 μm relative to controls, with this difference doubling by 7 months of age (Fig. 3C). Mean photoreceptor length, as measured from OCT scans, in RPEΔMttp mice at 5 months old was 98.6% that of controls while among 7–8-month-old mice, mean photoreceptor length in RPEΔMttp mice was 99.6% that of controls. The means were not significantly different between genotypes in either age group (t-test, p > 0.05, data not shown).

Decreased MTP in the RPE leads to RPE lipid imbalance and dysregulation of retinal cholesterol.

Collectively, the SLO and SD-OCT imaging studies suggested alterations in RPE that may be indicative of lipid accumulation. Staining RPE/ch flat mounts for neutral lipids with LipidTOX revealed strong lipid accumulation in RPE cells of RPEΔMttp mice(Fig. 4A and B). The mean area of LipidTOX-positive structures (lipid aggregates) in the RPEΔMttp mice was 1.7 times that observed in age-matched controls. In control RPE, neutral lipid structures with an average area of 0.92 μm² ±0.083 μm², were observed, In comparison, in the RPEΔMttp mice, the average area of neutral lipid structures was1.587 μm² ±0.182; a ~ 70% increase in size (Fig. 4A and B).

Figure 4. Lipid accumulates within PLIN2-positive structures in the RPE of RPEΔMttp mice.

A. Neutral lipids accumulate in the RPE of RPEΔMttp mice. RPE/Ch flat mounts from 8-month-old RPEΔMttp mice and controls were stained for Cre recombinase (green), neutral lipids (LipidTOX, red), and Hoechst nuclear stain (cyan). B. Mean area (± SEM, p < 0.001) of LipidTOX-positive aggregates (outlines shown) in flat mounts from 8-month-old mice. C. Neutral lipids accumulate in PLIN2-positive structures in the RPE cells of RPEΔMttp mice. RPE/Ch flat mounts from 3–4-month-old RPEΔMttp and control mice stained for PLIN2 (green), neutral lipids (LipidTOX, red), and Hoechst nuclear stain (cyan). D. Mean area (± SEM, p < 0.001) of LipidTOX-positive structures in flat mounts from 3–4-month-old mice. E. Mean fraction of the total number LipidTOX-positive structures that were also PLIN2-positive per FOV (± SEM, p < 0.001) in flat mounts from 3–4-month-old mice. F. Mean fraction of the total area of LipidTOX-positive structures that were also PLIN2-positive per FOV (± SEM, p < 0.05) in flat mounts from 3–4-month-old mice.

To more fully understand lipid storage dynamics in RPEΔMttp RPE cells, we re-analyzed images of RPE/Ch flat mounts from 3–4-month-old RPEΔMttp and control mice that were stained for LipidTOX and also probed for PLIN2 (perilipin-2), a protein found on the surface of lipid droplets (Fig. 4C). We previously reported increased PLIN2 expression in the RPE/Ch of 3–4-month-old RPEΔMttp mice measured with immunoblotting and a significantly higher colocalization of LipidTOX and PLIN2 immunostaining in these flat mounts (Grubaugh, et al., 2024 in press). For further analysis of these images, we measured the mean area of LipidTOX-positive structures. As in the 8-month-old mice, the mean area of LipidTOX-positive structures was significantly greater in 3–4-month-old RPEΔMttp RPE (Fig. 4D), but the magnitude of this change was only 1.2 times that of control RPE, likely due to the younger age of the mice. We then overlayed the outlines of LipidTOX-positive structures onto the PLIN2 channel to identify PLIN2-positive structures. The fraction of LipidTOX-positive structures that were also PLIN2-positive per FOV, as measured by number of structures (Fig. 4E), and by total area (Fig. 4F) was significantly greater in RPEΔMttp RPE than in control RPE. These data suggest that reduced MTP expression increases lipid storage within PLIN2-positive lipid structures. Consistent with our LipidTOX-staining studies, we observed lipid-rich structures in the RPE of RPEΔMttp mice using OTAP-fixation by electron microscopy (Fig. 5, bottom). These lipid puncta were prevalent in the RPEΔMttp RPE and rarely observed in the control mice. Fig. 5 (top, control) shows the single case where we observed a lipid-rich structure of all of the control mice examined.

Figure 5. Lipids accumulate in the RPE of RPEΔMttp mice.

Transmission electron microscopy (TEM) images of OTAP-fixed retinal sections from a 9.5-month-old RPEΔMttp mouse showing lipid pools in RPE cells (white arrows).

MTP transfers cholesterol, TG and phospholipids in the assembly of Blps, therefore we evaluated the effects of RPE-specific MTP depletion on cholesterol homeostasis in the retina by staining with filipin which binds to the 3’-OH group of unesterified cholesterol (UC). Interestingly, intensely stained cholesterol puncta were detected within the OS and IS regions of RPEΔMttp mice compared with controls (Fig. 6A and B, yellow arrows).While filipin positive puncta were observed in the basal region of the RPE in controls, we rarely observed such puncta in the RPEΔMttp mice (Fig. 6 B white arrows). Numerous bright filipin foci were also observed within the ONL, similar foci were less abundant in controls (Fig. 6B, yellow arrows). Filipin staining was ~3-fold higher in the RPEΔMttp OS layer compared with control (Fig. 6C). Biochemical assessment of cholesterol levels in the NR of the RPEΔMttp mice, showed ~ 35% increase in cholesterol relative to control (Fig. 6D). The increase in NR-associated cholesterol correlated with increased levels of RPE cholesterol transporter, ABCA1 (Fig. 6E and F); there was on average a 165 ± 18% increase in ABCA1 in RPEΔMttp compared to age matched controls. No change was detected in, the ATP-dependent cholesterol transporter, ABCG1 (102±0.76%) between RPEΔMttp and controls. These data suggest that with decreased MTP activity, photoreceptor cells retain or accumulate more cholesterol in OS and photoreceptor cell-associated laminae. The change in ABCA1 suggests a modulation of cholesterol efflux to an MTP-independent pathway in the RPE of RPEΔMttp mice.

Figure 6. Unesterified cholesterol (UC) accumulates in the neural retina (NR) of RPEΔMttp mice.

Representative confocal A. overview and B. higher magnification images of retinal cryosections from 8-month-old control (left) and RPEΔMttp (right) mice probed for cholesterol (filipin, purple). DRAQ5 nuclear stain (cyan); subretinal and IS/OS cholesterol puncta (yellow arrows), sub-RPE cholesterol puncta (white arrows). ONL: outer nuclear layer, IS: inner segments, OS: outer segments, RPE: retinal pigment epithelium C. Mean UC intensity (± SEM) was quantified in the OS region (t-test, p < 0.01). D. Total cholesterol increases in the NR of RPEΔMttp mice. Mean cholesterol concentration (± SEM) in the NR of RPEΔMttp mice and age-matched controls (t-test, p < 0.05). E. Immunoblot analysis of protein lysates from the RPE/Ch of RPEΔMttp and control mice probed for ABCA1, ABCG1, and MTP and normalized to β-actin. F. Quantification of ABCA1, ABCG1, and MTP protein expression as measured by immunoblotting (shown in E) in RPEΔMttp and control mice.

Structural and functional consequences of disrupted lipid homeostasis in the RPEΔMttp mice.

RPEΔMttp mice exhibited retinal degenerative changes, as well as RPE and retinal lipid imbalance, thus we tested if RPE-MTP depletion affects retinal structure. Retinal cryosections from RPEΔMttp and control mice were immunostained for cone matrix sheaths with PNA and for rod photoreceptors with anti-opsin. The outer retina of RPEΔMttp mice routinely (all RPEΔMttp mice studied) exhibited sporadic structural alterations resembling tubulations or rosettes (Fig. 7A and B), similar structures were not observed in any of the control mice studied. Photoreceptor degeneration is often accompanied by immune cell infiltration and microglial action in the subretinal space (48, 49). Therefore, to detect microglia, retina cryosections were immunostained with an antibody to Iba1. RPEΔMttp tubulations were often associated with Iba 1 positive cells in the subretinal space suggestive of inflammatory microenvironment in the affected areas (SFig. 3). As previously described, autofluorescence associated with the microglia and detectable at 488 nm excitation likely derives from phagocytosed bisretinoid fluorophores that form in the photoreceptor outer segments populating the interiors of the tubulations/rosettes(50). Consistent with the OCT based measurements, in regions away from dysplasia, mean IS+OS length was not significantly different between 8 month-old RPEΔMttp and control mice (mean IS+OS length in RPEΔMttp mice was 99.5% of control; t-test, p > 0.05).

Figure 7. Structural anomalies in the retinas of RPEΔMttp mice.

Retinal cryosections from A. 8-month-old and B. 12-month-old RPEΔMttp mice and age-matched controls stained with phalloidin (F-actin, green), Hoechst nuclear stain (cyan) and probed for opsin (RHO, red). Retinal dysplasia was evident in RPEΔMttp mice but not age-matched controls. Retinal cryosections from C. & D. 8-month-old and E. & F. 12-month-old RPEΔMttp mice and age-matched controls stained with phalloidin (F-actin, green) and Hoechst nuclear stain (cyan) and either C. & E. stained with peanut agglutinin lectin (PNA, magenta) or D. & F. probed for opsin (RHO, red) showing retinal regions without overt pathology.

Toluidine blue-stained semi-thin sections of the RPEΔMttp revealed hypertrophic and hyperpigmented RPE (Fig. 8A and C). These semi-thin sections of the RPEΔMttp showed large deposits located in the sub-retinal region consisting of various size inclusions mixed with disordered outer segment fragments and was validated by TEM image from the corresponding area (Fig. 8B, yellow arrows). As shown in Fig. 8C and D, a hypertrophic RPE cell that appeared to be full of pale spherical inclusions that are relatively large, homogeneous and electron lucent (Fig. 8D, TEM of the corresponding area). In Fig. 9. we describe a series of morphological anomalies representative of those observed in the RPEΔMttp mice analyzed. Degenerating photoreceptor debris is observed in the subretinal space in Fig. 9B, with additional areas within the RPE containing vacuolar structures (blue arrows). In Fig. 9D, we show a clear displacement of RPE apical microvilli abutting what appears to be a deposit consisting of granular debris (Fig. 9B, Inset, red arrows). Similar structures were not observed in the control mice (Fig. 9A and C).

Figure 8. Morphological alterations in RPEΔMttp mouse retina.

A. Toluidine blue-stained semithin retinal sections from 8-month-old control (left) and RPEΔMttp (right) from similar regions of the retina showing A. a subretinal deposit and C. a swollen RPE cell. Transmission electron microscopy (TEM) images of retinal sections (standard fixation) with the same pathologies as in A. & C. showing B. enlarged debris and OS fragments (yellow arrows) in the subretinal deposit depicted in A. and D. spherical inclusions of varying sizes and electron density filling the engorged RPE cell depicted in B.

Figure 9. Morphological alterations in the RPEΔMttp mouse retina.

Representative transmission electron microscopy (TEM) images of retinal sections (standard fixation) from 9.5-month-old A. control and B. RPEΔMttp mice. A large subretinal deposit in the RPEΔMttp retina in B. is shown at higher magnification (Inset, right), and additional debris is visible elsewhere in the RPE (blue arrows). Representative TEM images of retinal sections (standard fixation) from 8-month-old C. control and D. RPEΔMttp mice. A deposit between the photoreceptor outer segments and RPE of the RPEΔMttp retina that displaces apical RPE microvilli (Inset,red arrows) in D. is shown at higher magnification (right). Orange boxes indicate areas that are enlarged to show detail in insets.

Visual Function

To gain insight into whether retinal lipid dysregulation impacts vision, luminance-response functions of the scotopic and photopic ERGs were compared between 5-month-old RPEΔMttp mice and controls. Dark-adapted (rod photoreceptor driven) a-wave amplitudes were reduced by ~50% of control in RPEΔMttp mice (Fig. 10A). Dark-adapted b-waves were reduced in a similar manner, likely due to diminished activation of bipolar neurons (Fig. 10B). No significant change in photopic b-wave was detected by 5 months (Fig. 10C). Taken together these results point to a functional decline associated with the rod photoreceptor pathway in the RPEΔMttp retina.

Figure 10. Decrease in RPE-specific MTP reduces rod ERG response but does not affect cone function.

Scotopic ERG A. a-wave and B. b-wave amplitudes and C. photopic ERG b-wave (cone b-wave) amplitudes produced in response to flashes of increasing intensity by 5-month-old RPEΔMttp mice and controls. Data points indicate means (± SEM) of 3 RPEΔMttp mice and 3 controls (6 eyes/genotype), which were compared with t-tests (* p < 0.05, ** p < 0.01, *** p < 0.001, no asterisk: p > 0.05).

Discussion

The ability of the RPE to maintain lipid balance and regulate the fate of its intracellular lipid pool is central not only to RPE fucntion but also photoreceptor function. (8, 51)The RPE intracellular lipid pool is fed by the daily phagocytosis and processing of polyunsaturated lipid rich OS and is augmented with the de novo synthesis of lipids as well as the uptake of Blps from the systemic circulation as illustrated in Fig. 11. The studies herein point to MTP-mediated Blp assembly in the RPE as an essential pathway in maintaining a healthy lipid balance between the RPE and PR and, hence, cell function (for review (52)).(52)

Figure 11. MTP activity is necessary to maintain RPE lipid homeostasis.

Schematic representation of the role of MTP-mediated Blp assembly in maintaining lipid balance in healthy RPE. Chol: cholesterol, FA: fatty acids, APOB: apolipoprotein B, Blps: APOB-containing lipoproteins, MTP: microsomal triglyceride transport protein, ABCA1 & ABCG1: cholesterol transporters, HDL: high-density lipoproteins, RPE: retinal pigment epithelium, Ch: choroid.

We generated and characterized the RPEΔMttp mouse model, in which the Mttp gene is knocked down in the RPE to test the hypothesis that local Blps (i.e., those assembled in the RPE) are necessary to maintain retinal health and function. These studies fill a particularly understudied niche in retinal lipid metabolism. It was particularly important to develop a mouse model that could distinguish the role of local, RPE-derived Blps from that of systemic, plasma-derived Blps in retinal lipid homeostasis because intra-ocular Blps are distinct from those found in plasma. Compared to those found in plasma, intra-ocular Blps contain elevated levels of esterified cholesterol (EC; up 16- to 40-fold higher), carry lower levels of phosphatidylcholine, and display a different morphology (intra-ocular Blps are larger) (16, 22). In the RPEΔMttp mice, local Blp assembly was reduced while systemic lipids were unaffected. Consistent with MTP’s role in co-translational regulation of APOB expression, the RPE of RPEΔMttp mice contained less APOB than age-matched controls. As expected, RPE-specific knockdown of the Mttp gene did not affect TG or cholesterol concentrations in the blood plasma of RPEΔMttp mice, as systemic Blps are assembled primarily in the liver and intestine (18).

The mouse retina is mostly rod-dominant with a rod:cone ratio of ~35:1 in C57Bl6/J mice (53). The pathology that we note in Fig. 3B is within 300 μm of the optic nerve. Based on published literature, we estimated the rod:cone ratio in this region to be 35.1:1, which is very similar to the average cone:rod ratio across the mouse retina. Moreover, the PR:RPE cell ratio in this area is ~200:1 (9). Collectively, these results suggest that the pathology in Fig. 3B may encompass a hypertrophic RPE cell in contact with ~6 cones and >190 rods. This ratio is similar to that of the peripheral region of the human macula, an area implicated in macular degeneration (9).

RPE-MTP implications for systemic Blps and ABL

Mutations in the MTTP gene that compromise MTP activity cause abetalipoproteinemia (ABL), which is characterized by low levels of plasma lipids and lipoproteins due to a compromised ability to synthesize chylomicrons and very low-density lipoproteins (VLDL) in the intestine and liver, respectively. ABL patients exhibit significant ophthalmological complications, including angioid streaks, retinitis pigmentosa and loss of vision (54–57), which is generally attributed to reduced absorption of dietary fat-soluble vitamins, particularly vitamins A and E, by enterocytes. Vitamin A is necessary for phototransduction, while vitamin E is neuroprotective with potent anti-oxidant properties (6, 58). The highest levels of vitamin E in the eye are found in the RPE (58, 59); in vitro, RPE treated with vitamin E have less complement activation (60) and reduced bisretinoid lipofuscin photooxidation (46). Despite treatment with mega doses of fat-soluble vitamins, ABL patient progressively lose eyesight with age (61, 62),

Here, for the first time, we have created a mouse model in which systemic lipid metabolism is normal but MTP expression in the RPE is deficient. Despite normal systemic lipid levels, RPEΔMttp mice show significant AMD-like ocular abnormalities, such as retinal degeneration, hyperreflective foci, increased lipid deposition and reduced visual function as determined by ERG. These results, coupled with the clinical observation that dietary fat-soluble vitamin supplementation only delays the onset of vision loss in ABL patients, indicate that RPE MTP plays a crucial role retinal lipid metabolism and that some of the pathologies seen in ABL patients might be related to RPE-specific MTP deficiency rather than systemic deficiency of fat-soluble vitamins.

RPE-MTP in retinal metabolic synergy

The retina relies on metabolic synergy for visual function; the RPE and retina have complementary metabolic roles such that they depend on each other for survival and individualized function. A critical aspect of this inter-dependence is the RPE’s metabolic flexibility. It relies on oxidative substrates; fatty acids, lactate and amino acids (proline) to provide TCA cycle intermediates thereby sparing glucose for the use by the NR(6, 63–66). Fatty acids are one of these flexible substrates, provided by the daily phagocytosis of lipid-rich OS, they generate fuel for RPE and NR function, through fatty acid oxidation (FAO)(14). We predict that RPE specific MTP-activity is necessary for oxidative metabolism to maintain RPE differentiation and function, which in turn regulates NR health.

Depletion of RPE-MTP contributes to lipid accumulation in RPE and adjacent PR

The RPE shares metabolic similarity with cardiomyocytes; not only do both utilize fatty acids as an energy source (13, 67, 68), but they both also secrete unique EC rich Blps(69–76). Several studies suggest that MTP-mediated secretion of Blps protects cardiomyocytes from lipid overload (77–80). iPSC derived cardiomyocytes from ABL patients fail to secrete APOB, accumulate intracellular lipids, and respond poorly to stress (81). Deficiency in RPE-MTP resulted in a similar intracellular accumulation of lipids. Given their particular lipid composition and role in intra-ocular lipid trafficking, RPE-derived Blps are proposed to play a role in the formation of drusen and other RPE-associated lipid accumulations (82). Drusen contain high concentrations of esterified cholesterol, unesterified cholesterol and TG but relatively little docosahexaenoic acid, a fatty acid abundant in PR, and also contain APOB, suggesting that drusen likely consist of RPE-processed material, rather than plasma-derived Blps or unprocessed phagocytosed OS(5, 21, 70, 75, 82). Subretinal drusenoid deposits (SDDs) that accumulate on the apical face of the RPE contain UC and the degraded remnants of shed OS (83, 84) but, unlike drusen, they contain very little EC (84). While the origin of SDDs is not completely clear, dysregulation of lipid transport between the RPE and PR is implicated in their biogenesis (84).

Decreased RPE-specific MTP expression led to pathology in RPEΔMttp mice. In vivo imaging of 5- month-old RPEΔMttp mice revealed signs of retinal degeneration, such as hyperreflective foci, which are suggestive of lesions in the RPE, as well as thinning of adjacent ONL, both of which were exacerbated with age. Lipids accumulated in PLIN2-positive aggregates of neutral lipids in the RPE and in cholesterol aggregates in the retina of RPEΔMttp mice.

Intracellular lipid aggregates like those that we detected in the RPE are often called lipid droplets (LD), but this term encompasses a heterogeneous collection of lipid aggregates, including cytosolic, lumenal, and nuclear LD (85, 86). In addition to lipid storage, LD are involved in many cellular processes, including the maintenance of calcium homeostasis, the unfolded protein response, the storage of fat-soluble vitamins, and the regulation of intracellular lipid dynamics (87–89). However, accumulation of lipid droplets has the potential to induce cellular dysfunction and even toxicity (86, 90), including the inhibition of phagocytosis in RPE and microglia (91, 92). The increase in the portion of PLIN2-positive lipid aggregates in the RPE of RPEΔMttp mice is consistent with results from mice with liver-specific knockout of Mttp showing a reciprocal increase in PLIN2 protein expression (93). Therefore, MTP may regulate intracellular lipid storage dynamics, the mechanism and effects of which require further investigation.

Numerous models of dysfunctional lipid processing within the NR/RPE/choroid complex showed similar pathology to that observed here in the RPEΔMttp. Transgenic mice with ocular expression of a pathological mutant allele of human ELOVL4, an enzyme involved in the synthesis of very long-chain fatty acids, exhibited deterioration of structure and function in PR, steatosis and reduced OS processing, and eventual PR death (24). In the Map1lc3b^−/− mice, encoding for LC3B, an autophagy-associated protein essential for efficient processing of phagocytosed OS, is deleted, leading to structural abnormalities in the NR and RPE, drusen-like basal lipid deposits, neutral lipid accumulation within the RPE, delayed processing of phagocytosed OS, and recruitment of immune cells (23). More specific to Blp dysfunction, two mouse models with the ablation of lipoprotein receptors that facilitate the internalization of Blp display pathology similar to that observed in the photoreceptors and BrM of AMD patients(94, 95). Vldlr^−/− mice, which lack the receptor for very low-density lipoproteins (VLDL), display altered PR metabolism and PR secretion of the angiogenic protein VEGFA, leading to neovascularization and recruitment of Iba1⁺ cells (94). Lipids accumulate in the BrM of mice lacking the low-density lipoprotein (LDL) receptor (95). A recent bioinformatics analysis on dry AMD versus wet AMD identified Mttp as one of the top 41 most significantly downregulated genes differentially expressed in dry AMD (96).

Ocular retinoids and local (RPE-mediated) Blp assembly and secretion

Total ocular retinoid concentrations were not significantly different between RPEΔMttp mice and controls. In contrast, the loss of function mutations in MTTP that cause ABL are associated with an absence of plasma Blps, leading to a decrease in the transport and availability of vitamin A and vitamin E to peripheral tissues, including the eye (5, 72, 75, 97). These observations suggest that intestine-specific MTP activity plays a role in vision, a hypothesis that can be tested using conditional intestine-specific MTP deficient mice, ENTΔMttp^IND. While plasma Blps are required to transport fat-soluble vitamins out of the intestine and to the liver for storage, vitamin A is transported from the liver to eye via retinoid binding protein 4 (RBP4), encoded by the RBP4 gene (41). However, in Rbp4-deficient mice, RPE find a way to acquire retinol (vitamin A) by RBP4-independent mechanisms that could involve lipoprotein(41).

In contrast to the other retinoids quantified in this study, atRE was significantly lower in the eyes of RPEΔMttp mice than in controls. In the RPE, retinoids are stored as retinyl esters in retinosomes, particles similar to intracellular LD (44). Given that MTP also regulates lipid droplet formation (98 34), our future studies will test the hypothesis that MTP activity is necessary for retinyl ester storage in the RPE.

Conclusion

In this study, we demonstrate that RPE-specific MTP activity is critical to maintain intracellular RPE lipid pools, likely by serving as a conduit to prevent lipid steatosis. When we reduced MTP expression in the RPE, we observed cholesterol accumulation in the OS region and decreased rod cell function. On a molecular level, the accumulation of cholesterol and decreased DHA in PR are predicted to slow the plasma membrane-delimited enzyme interactions governing phototransduction(99–101), the latter of which we predict is depleted due to decreased apical RPE-Blp secretion. Cone cell function appears to be less dependent on MTP-mediated lipid processing. These studies have far-reaching implications for understanding how MTP expression and activity is regulated (102)in disease processes as diverse as iron accumulation-induced lipotoxicity of lysosomes (103) to the dysregulation of aerobic glycolysis/lipid oxidative phosphorylation metabolic synergy in retinitis pigmentosa (104, 105)as well as diseases of the aging retina (5, 64). Moreover, the value of the RPEΔMttp model is that the regulation of the RPE-Blp pathway can now be studied in vivo. This feature is of particular importance, due to the unique features of ocular Blps relative to other, well-studied Blps from liver, intestine, and heart as well as the singular feature of lipid handling in the RPE due to OS phagocytosis. Collectively, these studies strongly suggest that lipoprotein assembly, a key pathway associated with the biogenesis of AMD drusen, is essential for outer retinal health.

Table 2 –

List of antibodies used

Antibody	Source	Catalog #	Host species	Dilution (application)
anti-ABCA1	Novus Biologicals (Centennial, CO, USA)	NB400-105	rabbit	1:500 (Immunoblotting)
anti-ABCG1 (EP1366Y)	Abcam (Cambridge, UK)	ab52617	rabbit	1:1,000 (Immunoblotting)
anti-APOB	Abcam	ab20737	rabbit	1:100 (Immunostaining)
anti-β-Actin (AC-74)	Sigma-Aldrich (St. Louis, MO, USA)	A2228	mouse	1:5,000 (Immunoblotting)
anti-Cre recombinase (D7L7L)	Cell Signaling Technology (Danvers, MA, USA)	15036	rabbit	1:100 (Immunostaining)
anti-Iba1	Fujifilm (Osaka, Japan)	019-19741	rabbit	1:500 (Immunostaining)
anti-MTP (D-4)	Santa-Cruz Biotechnology (Dallas, TX, USA)	sc-515743	mouse	1:1,000 (Immunoblotting)
anti-P-Cadherin	R&D Systems (Minneapolis, MN, USA)	AF761	goat	1:200 (Immunostaining)
anti-PLIN2	Proteintech (Rosemont, IL, USA)	15294-1-AP	rabbit	1:1,000 (Immunoblotting)
anti-rhodopsin (4D2)	Sigma-Aldrich	MABN15	mouse	1:500 (Immunostaining)

Data Availability Statement;

Included in article. The data that supports the findings of this study are available in the Methods and or supplementary material of this article.