Amelotin is expressed in retinal pigment epithelium and localizes to hydroxyapatite deposits in dry age-related macular degeneration

Abstract

Deposition of hydroxyapatite (HAP) basal to the retinal pigment epithelium (RPE) is linked to the progression of age-related macular degeneration (AMD). Serum-deprivation of RPE cells in culture mimics some features of AMD. We now show that serum-deprivation also leads to the induction of amelotin (AMTN), a protein involved in hydroxyapatite mineralization in enamel. HAP is formed in our culture model and is blocked by siRNA inhibition of AMTN expression. In situ hybridization and immunofluorescence imaging of human eye tissue show that AMTN is expressed in RPE of donor eyes with geographic atrophy (“dry” AMD) in regions with soft drusen containing HAP spherules or nodules. AMTN is not found in hard drusen, normal RPE, or donor eyes diagnosed with wet AMD. These findings suggest that AMTN is involved in formation of HAP spherules or nodules in AMD, and as such provides a new therapeutic target for slowing disease progression.

INTRODUCTION

Age-related macular degeneration (AMD) is an increasingly important cause of irreversible vision loss in aging populations.^1,2 AMD falls into two broad categories: choroidal neovascularization (CNV) or “wet” AMD and geographic atrophy (GA) or “dry” AMD.^3,4 Currently, there are effective therapies for CNV, but none for GA.^5,6 GA is characterized by the progressive loss of central vision due to death of retinal pigment epithelium (RPE) and photoreceptor cells associated with buildup of basal deposits including drusen.⁴

Drusen are accumulations of extracellular particles, including lipids, proteins, and minerals, between Bruch’s membrane (BrM) and the RPE, and are the most common early sign of AMD.^7–10 Sub-RPE deposits increase in number and size with age and are believed to impair metabolic exchange between the choroidal blood circulation and the retina.^11,12 This blockage of nutrient and waste flow to and from the photoreceptors may eventually lead to AMD.^13,14 Recent work has identified many lipid, protein, and mineral components in drusen,^11,15–17 including the discovery of hydroxyapatite (HAP) (mineralized calcium phosphate) spherules with cholesterol-containing cores in sub-RPE deposits.^9,16 These HAP spherules are implicated in the formation of the sub-RPE deposits and Bruch’s membrane plaque in AMD,¹⁶ while multilobular HAP nodules are associated with progression to advanced AMD.¹⁸

It has been proposed that sub-RPE deposit formation is initiated by, and probably regulated, in the RPE.^9,12,19 Using a serum-starved transwell RPE cell culture model, we have previously shown the up-regulation of cholesterol synthesis/accumulation of unesterified cholesterol as well as accumulation of proteins including Apolipoprotein B (ApoB) and EFEMP1/Fibulin3 (Fib3) similar to that seen in AMD.^15,20 Here we show that in the same model system, RPE cells express amelotin (AMTN) a protein that is expressed during the maturation of dental enamel.²¹ AMTN has a direct influence on biomineralization by promoting HAP mineralization and has a critical role in the formation of the compact aprismatic enamel.²²

Using a combination of cell culture techniques, ex-vivo histopathology, and molecular imaging techniques, we determined that AMTN is expressed by RPE cells and promotes calcium phosphate mineralization in vitro and that AMTN is found in drusen containing HAP spherules and in large advanced drusen with crystalline calcium phosphate nodules, but not in plaque or hard drusen, suggesting that it plays an important role in AMD.

MATERIALS AND METHODS

Study design

The objectives of this work were to characterize the expression of amelotin in RPE which was originally discovered from RNASeq experiments. The initial experiments were performed using a serum deprived transwell cell culture model which mimics some features of AMD.^15,20 ARPE-19 cells were authenticated using short tandem repeat (STR) analysis by the cell line authentication service (ATCC). The cell line has been published previously and the cells were free from contamination. ARPE-19 cells were purchased from ATCC (ATCC, Manassas, VA, CRL-2302), and validated for the expression of the RPE65 and RLBP1 marker genes with PCR of cDNA using the RPE65-specific primers 5′-CCA GAT GCC TTG GAA GAA GA-3′; 5′-CTT GGC ATT CAG AAT CAG GAG-3′ (99 bp amplicon) and the RLBP-specific primers 5^∕-AGA TCT CAG GAA GAT GGT GGA C3′; 5′-TGG ATG AAG TGG ATG GCT TT-3′ (72 bp amplicon) as shown in our previous publication.²⁰ In all studies, representative results are from three independent repeats, except where specified otherwise in the figure legends.

The expression of amelotin in RPE and its association with hydroxyapatite deposition was validated in normal and AMD donor eye tissue. Human tissue samples were obtained from the National Disease

Research Interchange (NDRI) (Philadelphia, PA) under NIH OHSR exemption 11287.

In vitro RPE cell culture

ARPE-19 cells were cultured in complete Dulbecco’s Modified Eagle Medium F12 (Gibco Life Technology, Gaithersburg MD; DMEM F12) medium containing 10% fetal calf serum (FCS) and 1% penicillin/streptomycin (100-units penicillin/ml, 100 μg streptomycin/ml), (Invitrogen, San Diego, California, USA). Cells were grown in an incubator at 37°C with 5% CO₂ using six-well plates and culture chambers (Corning Primaria plastic culture ware). The culture plates used in these experiments were not coated with collagen because preliminary findings showed cells cultured in collagen coated wells showed no difference in amelotin expression by in situ hybridization and immunofluorescence (Fig. S1) to cells cultured directly on plastic. Cells were grown to a minimum of 90% confluence prior to serum deprivation experiments.

The extracellular pH of the confluent cells cultured with either serum deprived media or 10% serum supplemented media were measured by slightly tilting the six-well culture plates to pool the media and inserting a micro pH electrode (Thermo Fisher Scientific, Waltham, MA). The pH reading was performed once a day from two replicate wells for each culture condition for nine days. Intracellular pH of ARPE-19 cells cultured in 10% serum or SFM were labeled with pHrodo Green AM Intracellular pH Indicator (Thermo Fisher Scientific, Waltham, MA) following manufacture’s protocol.

Cell viability assay

ARPE-19 cells were plated in 96-well plates to 90% confluence and incubated overnight in complete media at 37°C with 5% CO₂. Cells were washed thoroughly with serum free medium before continuing incubation in SFM. Cell free wells served as blank controls. Following serum deprivation, the medium was removed from wells, and 110 μl of 10% CCK-8 reagent (Dojindo Molecular Technologies Inc., Rockville, MD) in SFM was added to all wells. The plates were incubated at 37°C for 1 hour, then the optical density (OD) at 450 nm was measured using a microplate reader. The survival percentage for each day was calculated relative to the absorbance of day 0, taking the average reading of six wells from three separate plates.

Transwell culture system for ARPE-19 cells

All cell culture experiments were performed on cells in passage numbers 2–8 after purchase from ATCC. Cells were cultured in complete Dulbecco’s Modified Eagle Medium F12 (Gibco Life Technology, Gaithersburg MD; DMEM F12) medium containing 10% fetal calf serum (FCS) and 1% penicillin/streptomycin (100-units penicillin/ml, 100 μg streptomycin/ml), (Invitrogen, San Diego, California, USA). Cells were housed in an incubator at 37 °C with 5% CO₂ on collagen-coated 24 mm transwell inserts (Corning Primaria plastic culture ware, Thermo Fisher Scientific, Waltham, MA) for six weeks before serum starvation experiments. The transepithelial electrical resistance (TEER) of the ARPE-19 cells in culture was measured by using the Millicell^® electrical resistance system (ERS, EMD Millipore, Billerica, MA) as described previously.²³ TEER was measured using STX3 chopstick electrodes. A minimum of three independent Transwell filters were used in each experimental condition and repeated in two separate experiments. The cells maintained a mean TEER level of 97.8 ± 0.7 Ω cm2 at 6 weeks and 82.3 ± 0.5 Ω × cm2 after 9 days serum deprivation.

For live cell imaging, the membrane immersed in culture media was dissected from the well using a dissection microscope. The membrane, with the cells facing downwards, was placed on 100 μl of media in a petri-dish and imaged using Zeiss LSM 880 microscope with Airyscan (Zeiss, USA).

For sectioning, membranes with cells were fixed with 4% PFA for 15 minutes and the membrane side (cells facing upwards) was glued to a block of 4 mm thick 5% agarose gel. Blocks of gel containing the membranes were cut out and mounted onto the stage of the Leica_VT1000S_Vibratome. Each section was cut at a thickness of 100 μm.

Serum deprivation of cultured cells

Complete culture medium (DMEM F12 medium containing 10% FCS) was removed, and the confluent RPE cells were washed thoroughly with serum free medium {(SFM) - DMEM F12 medium without 10% FCS} before re-incubating in SFM {DMEMF12 with 1% penicillin/streptomycin (100-unit penicillin/100μg streptomycin per ml)}. Day 0 in all experiments denotes the cells that remained in complete culture medium (10% serum) throughout the experiment. The media was not changed during the serum-free conditioning.

Transfection of RPE cells with siRNA

Following manufacturer’s protocols, the silencing RNAs were optimized for the experiment, diluted, and prepared in Opti-MEM^® I Reduced Serum Medium (Thermo Fisher Scientific, Waltham, MA). In a six-well plate 1–2 × 10⁶ cells were plated in 2 ml medium per well and grown to 70–75% confluence overnight, fresh DMEM with 5 % FBS without antibiotics was added and incubated for four hours. The medium was aspirated and replaced with 500 μl transfection medium containing 25 nM human Amelotin (AMTN) siRNA, siRNA GAPDH Positive Control, or siRNA Non-Targeting Control low GC (Catalog # 1299001, Catalog #4404025, Catalog #12935–200, Thermo Fisher Scientific, Waltham, MA) and 5 μl Lipofectamine 2000. Cells were incubated at 37°C overnight. The cells were then harvested for protein or RNA extraction to determine the levels of knockdown. For serum starvation experiments performed on six-well plates, after the overnight incubation, the media containing the transfection mix was removed, cells were rinsed with DNAse/RNAse free sterile water (Invitrogen, San Diego, California, USA) and fresh serum free media was added.

For serum deprivation experiments on transwell plates, transfected cells were transferred to collagen-coated 24 mm transwells and allowed to attach for sixhours. The cells were rinsed with DNAse/RNAse free sterile water (Invitrogen, San Diego, California, USA) and serum free media was added.

Western blot

ARPE-19 cells grown in six-well plates were washed with 1X phosphate-buffered saline (PBS) (catalog# RGF-3190 KD Medical, Columbia MD) and lysed in RIPA buffer with protease inhibitors (Thermo Fisher Scientific, Waltham, MA). The protein concentrations were measured using a BCA protein assay kit (Thermo Fisher Scientific, Waltham, MA). 20 μg of the total protein was loaded onto a 10% SDS-PAGE gel. Gels were run at 80V for 30 min followed by 150V for 60 min.

Proteins were transferred to the Immobilon-FL polyvinylidene difluoride (PVDF) membrane at 350 mA for 50 min in an ice-bath. Post-transfer, blots were blocked with 5% bovine serum albumin (BSA) in tris-buffer saline with Tween-20 (TBS/T) for 1h at room temperature then rinsed once in TBST. Primary antibody, Anti-Amelotin - N-terminal (Catalog# ab173661 Abcam, Cambridge, MA), was diluted 1:200 with TBS/T and incubated overnight at 4°C. After thorough washing, the membranes were incubated with HRP-conjugated secondary antibodies (Invitrogen, San Diego, California, USA) diluted 1:1000 in PBS for two hours in the dark at room temperature. The membrane was then washed in TBS/T three times before using LumiGold ECL Western Blotting Detection Kit (VerII; Signagen Laboratories, Ijamsville, MD). The membrane was scanned. The blots shown are representative of at least three biologic repeats of each experiment. The β-actin level was used to normalize the signal between samples. Quantitation of the signal was performed using ImageJ software (version 1.45; National Institutes of Health, Bethesda, MD).

Amelotin (AMTN) secretion ELISA

ARPE-19 cells were grown in T-25 flasks (Corning Primaria plastic culture ware, Thermo Fisher Scientific, Waltham, MA) with DMEM + 10% serum. When the cells were 90 % confluent, the culture medium was removed, and the cells were washed once with SFM. Next SFM media was added to the cells to begin the experiment. After nine days culture in SFM, the culture supernatant was collected and assayed. Briefly:

A 96-well PVC microtiter plate (Thermo Fisher Scientific, Waltham, MA) was coated with monoclonal AMTN capture antibody (Catalog# LS-C799006, LsBio, Seattle WA) at a concentration of 10 μg/ml in carbonate/bicarbonate buffer (pH9.6) and incubated at 4°C overnight. Excess coating solution was removed by washing the plate three times with 200 μl PBS. The remaining unbound protein-binding sites in the coated wells were blocked by adding 200 μl blocking buffer (5% nonfat dry milk) per well and incubating overnight at 4°C before washing the plate twice with PBS.

Recombinant human AMTN (Boster bio, Pleasanton, CA) solutions ranging from 0 to 100 μg/ml were used as known standards. 100μL of each standard and culture supernatant from each sample were distributed in duplicate to the AMTN-coated capture plate and incubated for 2 hours at 37°C.

Following the 2 hour incubation the samples and standards were removed, and 100 μl of biotin conjugated diluted detection polyclonal antibody (Catalog# LS-C319460, LsBio, Seattle WA) was added to each well and incubated for 60 min at 37°C. The antibody was removed, and the plate was washed four times with 200 μl PBS per well.

100 μl HRP conjugated antibody was added to each well and incubated for 60 min at 37°C.The antibody was removed the plate was washed four times with 200 μl PBS per well.

90μl of TMB Substrate (LsBio, Seattle WA) was added to each well for 20 min at 37°C followed by 50μL of Stop Solution (LsBio, Seattle WA). The absorbance was read at wavelength 450nm using a Bio-Rad 680 reader (Bio Rad laboratories, Hercules, CA).

Quantitative PCR

RNA was isolated from cells grown and treated as described in six-well tissue culture plates using Trizol^® (#15596–018, Invitrogen, Invitrogen. com), following the standard protocol using 1 ml of Trizol^® per well and resuspending the isolated RNA in 10 μl DEPC-treated water. cDNA was prepared from the isolated RNA using ProtoScript^® II First Stand cDNA Synthesis Kit (New England BioLabs Inc. Ipswich, MA) using 4 μg of the isolated RNA per sample. Each cDNA sample was diluted to 300 μl with H₂O.

Quantitative PCR of selected genes was performed using the Roche Universal ProbeLibrary hydrolysis probe system (Roche, Mannheim Germany), the Luna Universal Probe qPCR Master Mix (New England BioLabs Inc. Ipswich, MA), and custom primers (Eurofins MWG|Operon) on an Applied Biosystems ViiA7 System using QuantStudio™ (v1.2) software (Life Technologies, Carlsbad, CA) all following manufacturer’s recommendations. The relative expression values were calculated using the RQ (relative quantity) method normalized to the genes ABCF3. RQ = 2^{(CtDay0 − CtDay)} x NF. NF (normalization factor) = (RQRef1 x RQRef2)^1/2.

RNA sequencing

RNA sequencing was performed on samples from serum deprived ARPE19 cells from days 0, 1, 3, 4, 5, 6, and 9 with three biological replicates for each time point. RNA was isolated from the confluent cultures on the appropriate day using 1.0ml Trizol™ (#15596-018, Invitrogen) as per manufacturer’s instructions. The RNA pellet was resuspended in 20μl H2O and the concentration was measured using a NanoDrop 100. RNA quality, as judged by RIN, was measured using an Agilent BioAnalyzer 2100 and RNA Nano Chips. All the samples had RINs greater than 7.0. The libraries of the RNA samples were constructed and sequenced by the NIH Intramural Sequencing Center, NISC, using the Illumina HiSeq platform.

Alignment of the sequence files was performed using Novoalign (v. 3.02.00) against the UCSF hg19 refFlat. Counts were determined using HTseq (v. 0.61) using the mode intersection-strict. DESeq (v. 1.12.1) with the default settings on R (v. 3.0.1) was used to identify differentially expressed genes. The sequence data is available from GEO (GSE129964).

Result sets used for gene set analysis or data display are either count matrices for the three biological replicates from each time point or lists of fold change ratios for each time point (mean normalized counts at the time point divided by mean day 0 normalized counts).

In vitro mineralization assay

A simulated body fluid (SBF) buffer was prepared for the in vitro mineralization study as previously described.²² The buffer contained 15 mM K⁺ (KCl 99.5%), 0.8 mM Mg²⁺ (MgCl₂ × 6H₂O), 10 mM HCO₃-(NaHCO₃), 2.8 mM Ca²⁺ (CaCl₂ × 2H₂O), 1.67 mM phosphate (Na₂HPO₄), 50 mM HEPES, 140 mM Na⁺ (NaCl reagent grade and NaHCO₃, (all chemicals, Sigma Aldrich, Allentown, PA). After adjusting the pH to 7.2 at 37°C with 1 N NaOH, the buffer was filtered through a 0.2 mm filter (Acrodisc, Sigma Aldrich, Allentown, PA). Rh-AMTN protein (Boster bio, Pleasanton, CA) was dissolved in the mineralization buffer at 0 μg/ml, 20 μg/ml, 40 μg/ml and 100 μg/ml concentrations. 300 μl buffer was added to each well of a sterile 96-well polystyrene plate. The plate was incubated at 37°C in a humidified incubator throughout the duration of the nine-day assay. The kinetics of calcium phosphate mineralization was monitored using a light-scattering assay that measures clarity versus turbidity of the solution. Turbidity indicates the onset and extent of precipitation from the mineralization buffer influenced by the dissolved protein compared with the SBF buffer alone. The light-scattering was measured in a microplate reader at 540 nm at different time points during each day of the mineralization experiment for nine days. Statistical analysis of turbidity measurements was conducted using one-way ANOVA analysis. Significance was assigned at p < 0.05. At the completion of the light-scattering measurements, the liquid was aspirated, wells containing calcium phosphate precipitates were briefly washed with deionized water and air-dried in a fume hood. The calcium phosphate precipitates were imaged using an Olympus light microscope.

Immunofluorescence staining of cells

Serum starved and AMTN knock-down ARPE-19 cells cultured in six-well chambers or transwell inserts were washed with cold PBS, then fixed with 4% paraformaldehyde (PFA) for 10 min at room temperature, and permeabilized with 0.1% Triton X-100 for 5 min. Next, the samples were blocked with 5% BSA for 30 min at room temperature. Cells were incubated with primary antibodies, rabbit polyclonal ZO-1 (Abcam, Cambridge, MA) diluted 1:100 in ICC buffer (1x PBS with 0.2% Tween 20, 0.5% BSA, and 0.05% sodium azide), Anti-Amelotin antibody (Catalog#ab122312, Abcam, Cambridge, MA) diluted 1: 100 in ICC buffer, anti-CRALBP (Thermo Fisher Scientific, Waltham, MA) diluted 1:500 in ICC buffer, or anti-mouse Vimentin (Santa Cruz) diluted 1:100 dilution in ICC buffer for four hours at room temperature. The antibody solution was removed, and the cells were washed with PBS. Next, samples were incubated with the anti-rabbit or anti-mouse Alexa 488 (Thermo Fisher Scientific, Waltham, MA) secondary antibody diluted 1:100 with PBS for two hours at room temperature. Slides were counter stained with DAPI (#D1306, Thermo Fisher Scientific, Waltham, MA) diluted 1:500 in PBS in the dark for one hour. After washing with 1X PBS, samples were mounted with Prolong Gold (Molecular Probes, Eugene, OR) mounting medium and examined by confocal microscopy using a Zeiss LSM 880 with Airyscan (Zeiss, USA).

HAP fluorescent labeling and calcium phosphate quantification assay on cultured cells

Serum starved and AMTN knock-down ARPE-19 cells cultured on transwell inserts were washed with cold PBS and fixed with 4% paraformaldehyde (PFA) for 15 minutes at room temperature. The membrane side (cells facing upwards) was glued to a 4 mm thick 5% agarose gel. The membranes affixed to agarose were cut out and mounted onto the stage of the Leica_VT1000S_Vibratome for sectioning. RPE sections were permeabilized with 0.1% Triton X-100 for 5 min and stained with 20 μM Bone-Tag 680RD (Li-Cor, Lincoln, NE) (excitation wavelength: 620 nm, emission wavelength: 680 nm) for one hour at room temperature.

ARPE-19 cells cultured on six-well plates were fixed with 4% paraformaldehyde (PFA) for 15 minutes and stained with 40 mM Alizarin Red S (ARS) at room temperature for 20–30 min with shaking. 800 μL of 10% acetic acid was added to each well and incubated for 30 minutes at room temperature. Then, the cells were collected, the temperature was raised to 85°C for 10 minutes, then cooled on ice for 5 minutes. Samples were centrifuged at 20,000 x g for 15 minutes. The samples were decanted to save the supernatant and 200 μL of 10% ammonium hydroxide was added to neutralize the acid. Alizarin Red S standard was prepared to the manufacturer’s instructions (Science Cell Research Laboratories, Carlsbad, CA). Samples and standards were plated onto 96-well plates in triplicate. Absorbance was read at 405 nm using a Bio-Rad 680 reader (Bio Rad laboratories, Hercules, CA).

Immunofluorescence staining of human retina sections

A total of twenty-eight normal and AMD-affected human eyes of donors aged between 65 and 98 years were obtained from the National Disease Research Interchange (NDRI) (Philadelphia, PA). Samples were obtained within 6 to 14 hours of death. Of each pair of eyes, one was fixed in formaldehyde and the other was frozen. Of these samples, five were normal and the remaining fifteen were previously diagnosed with macular degeneration. Samples were genotyped for the major GA associated polymorphism (Y402H) of CFH essentially as described before.²⁴ Genomic DNA was prepared from fixed eye tissue using the BioMasherII (Nippi, Tokyo, Japan) and Qiagen DNEasy^® Blood and Tissue Kit (Qiagen Inc., Valencia, CA). Polymerase chain reaction was performed using HotStarTaq^® Master Mix Kit, (Qiagen Inc., Valencia, CA), with primers specific to the CFH gene (CFHgenS:ATCTTATTTAAATATTGTAAAAATAATTGT), (CFHgenAS:TAAACATTTTGCCACAATTAATATAGATG). The amplified 375 bp product, spanning the Y402H region of the CFH gene, was purified from the PCR reaction using Qiaquick PCR purification kit, (Qiagen Inc., Valencia, CA) and sequenced by ACGT Inc. (Wheeling, IL) using sequencing primers (CFHPCRS: ACTATTTTGAGCAAATTTATG); (CFHPCRAS: GTCTTAGAATGTCATCTATG.) Sequences and traces were examined using Sequencer (Gene Codes Corp).

For sectioning the eyes were fixed with formalin, washed in PBS, and cryoprotected through a series of 5%, 10%, and 20% sucrose in PBS. Eyes were cut and sectioned through the macula. Sections were incubated in ICC buffer (0.5% BSA, 0.2% Tween-20, 0.05% sodium azide, in PBS, pH 7.3) for one hour at room temperature. Sections were incubated with Anti-Amelotin antibody (Catalog#ab122312 Abcam, Cambridge, MA) diluted 1:100 in ICC buffer, Biotinylated Peanut Agglutinin (PNA) (Vector Laboratories, Burlingame, CA) diluted 1: 200 overnight at 4°C. Following thorough washes with ICC buffer, the slides were incubated with secondary antibodies anti-rabbit Alexa 488 (Thermo Fisher Scientific, Waltham, MA) diluted 1: 200 and streptavidin conjugated with Alexa 633 (Vector labs, Burlingame, CA) diluted 1:300 in PBS for one hour at room temperature. Sections were washed extensively with ICC buffer and then mounted in Prolong Gold + DAPI (Molecular Probes, Eugene, OR) and examined using a confocal Zeiss LSM 880 with Airyscan microscope (Zeiss, USA).

HAP fluorescent labeling and immunohistochemistry of human donor tissue

Adjacent cryosections were stained with 1 mg/ml Alizarin Red S (excitation wavelength: 532 nm, emission wavelength: 620 nm) (SigmaAldrich, Allentown, PA), 20 μM Bone-Tag 680RD (excitation wavelength: 620 nm, emission wavelength: 680 nm) (Li-Cor, Lincoln, NE), and 1 mg/ml Xylenol Orange (excitation wavelength: 532 nm, emission wavelength: 570 nm) (SigmaAldrich, Allentown, PA), for 20 min at room temperature. Excess dye was blotted from the sections, then the sections were mounted in Prolong Gold (Molecular Probes, Eugene, OR), and examined using the microscope described above at the same magnification. Background autofluorescence for all images was assigned cyan blue.

To illustrate the histology of the tissue, sections were stained with hematoxylin and eosin (H&E). 10 μm sections were incubated in a vacuum overnight, dipped three times in 70% ethanol and rinsed in distilled water. The slides were stained with Gill’s haematoxylin (Vector Labs, Burlingame, CA) for five minutes, rinsed in water, dipped three times in ammonia solution, rinsed, stained with Eosin Y (Baker Analyzed) for three minutes, dehydrated with five quick dips in two changes of 95% ethanol and two changes of 100% ethanol, followed by ten dips in two changes of xylene to clear the samples. The slides were mounted using VectaMount H-5000 (Vector Labs, Burlingame, CA). and imaged using a Zeiss Imager Z1 (Zeiss, USA).

Fluorescent and colorimetric in situ hybridization to tissue sections and cultured cells

To detect mRNA in situ on tissue sections or ARPE-19 cultured cells we used RNAScope^® (Advanced Cell Diagnostics (ACD) https://acdbio.com) assay reagents, using the RNAscope^® 2.5 HD Reagent Kit-RED (Cat No: 322350) and the RNAScope^® human Amelotin probe or human RNAScope^® positive and negative controls, following manufacturer’s recommended protocol with slight modifications.

ARPE-19 cells, cultured on slides, were fixed with 10% neutral buffered formalin at room temperature for 30 minutes. The slides were washed with PBS and followed by 70% EtOH followed by 50% EtOH at room temperature for two minutes. The slides were air dried, then incubated with ProteaseK III reagent for ten minutes at 42°C. From this point we followed the manufacturer’s protocol.

The frozen sections were washed in 100% EtOH for one minute and air-dried overnight at room temperature. Before proceeding to the RNAScope^® protocol, the slides were baked for one hour at 60 °C, followed by washing for five minutes in room temperature PBS and a final bake for 30 minutes at 60 °C. Baking prior to hybridization kept the sections more firmly adhered to the slide. Target retrieval, protease digestion, and hybridization followed the manufacturer’s protocol. For counterstaining we used methyl green (Catalog #H-3402Vector Laboratories,) or hematoxylin (Catalog# H-3502, Vector Laboratories,), or a combination of methyl green or hematoxylin and eosin. For simultaneous fluorescence detection of mRNA and protein, after completing the above mentioned steps for mRNA in situ detection, instead of counterstaining with methyl green, hematoxylin or eosin, sections were incubated in ICC buffer (0.5% BSA, 0.2% Tween-20, 0.05% sodium azide, in PBS, pH 7.3) for one hour at room temperature and then incubated with Anti-Amelotin antibody (Catalog#ab122312 Abcam, Cambridge, MA) diluted 1:100 in ICC buffer overnight at 4°C. Following thorough washes with ICC buffer, the slides were incubated with secondary antibody anti-rabbit Alexa 488 (Thermo Fisher Scientific, Waltham, MA) diluted 1: 200 for one hour at room temperature. Sections were washed with ICC buffer and then mounted in Prolong Gold + DAPI (Molecular Probes, Eugene, OR) and examined using a confocal Zeiss LSM 880 with Airyscan microscope (Zeiss, USA).

Statistical analysis

Absorbance, turbidity and magnesium concentration assays were expressed as mean ± SEM with p < 0.05 deemed statistically significant unless indicated otherwise. Differences between groups were assessed using either an independent Student’s ttest (unpaired and two-tailed) or one-way analysis of variance with Dunnett’s or Tukey’s post-hoc tests.

Transcript counts were determined using HTseq (v.0.61) using the mode intersection-strict. DESeq (v.1.12.1) with the default setting R (v.3.0.1) was used to identify differentially expressed genes. Results presented as counts are mean normalized transcripts per million reads (TPM) (https://www.biostarts.org/p/273537/). Fold change ratios for each point represent mean normalized counts at the time point divided by mean day 0 normalized counts.

RESULTS

Serum-deprived ARPE-19 cells express Amelotin (AMTN)

Our earlier findings showed that ARPE-19 cell monolayers cultured in serum deprived conditions mimic certain responses relevant to AMD.^15,20 We extended this analysis by performing RNASeq on the serum deprived cell culture model over a 9-day time course.²⁵ Cell viability assays confirmed that at 9days, the percentage cell survival remained similar to our previous findings at 7-days in serum deprivation conditions (Fig. S2), therefore enabling us to extend experiment time. mRNA was prepared for cells at day 0 (10% serum) and days 1, 3, 4, 5, 6 and 9 of serum deprivation. This RNASeq data confirmed previous patterns of expression for cholesterol and zinc-binding protein genes.²⁰ We also confirmed selected results from ARPE-19 monolayers cultured in transwells for known RPE-expressed proteins and morphology (Fig. 1A–B).

Fig. 1.

Serum deprived ARPE-19 monolayers express Amelotin (AMTN) mRNA and protein at day nine. A. Diagram of the transwell cell culture system of ARPE-19 cell monolayers to mimic RPE in vivo. B. Confluent ARPE-19 monolayers showing polygonal cells with organized packing geometry, and localized immunofluorescence for known RPE-expressed proteins. C. Mean counts of total AMTN mRNA expression from RNA sequencing at day 0 (10% serum) and days 1, 3, 4, 5, 6 and 9 of serum deprivation. Data displayed are normalized mean counts for the 3 biological replicates from each time and n = 3 technical replicates. D. Representative western blot of AMTN (22 kDa) expression from cell lysate from ARPE-19 cultured in 6-well plates at day 0 (10% serum) and days 3, 5 and 9 of serum deprivation with 2 biological replicates from each time. Blots are representative of n = 3. E. In situ hybridization image of an ARPE-19 monolayers cultured in chambers at days 0 with 10% serum showing sparse AMTN mRNA detection, n = 3. F. In situ hybridization image of an ARPE-19 monolayers cultured in chambers at day 9 in serum free media showing abundant AMTN mRNA detection, n = 3. G. Changes in gene expression for genes annotated with bone and mineral associated genes with P value < 0.05 during serum deprivation time course of ARPE-19 cells as determined by RNA-Seq. Each tile represents a gene. The columns are ordered by days of serum deprivation. Genes with increased expression appears above the 0-reference line in Orange. Genes with decreased expression appears below the 0-reference line in Blue. The saturation of the color represents the magnitude of the change.

The most striking new result from the RNASeq data was the up-regulation of AMTN mRNA late in the time course. From low levels at day 0 (mean of normalized counts = 1,845), AMTN mRNA increased substantially (mean of normalized counts = 107,245) by day 9 in serum free media (SFM). (Fig. 1C). We find that AMTN protein expression is also increased at day 9 in ARPE-19 cell monolayers cultured under the same SFM conditions compared to days 0, 3 and 5 (Fig. 1D). These observations were reinforced by probing cell monolayers at day 0 and day 9 in situ with a human AMTN mRNA probe (Fig. E-F). These results indicate a clear increase in AMTN expression in cultured ARPE-19 cells under serum deprivation conditions.

At day 9 of serum deprivation we also saw increased expression of other genes reported to be involved in osteoblast differentiation, bone formation, and oculo-skeletal development (BMP4, GDF6);^26–29 maturation, development and mineralization of bone (NOG, COL1A1);^30,31 osteogenic differentiation of bone marrow stromal cells (BST2, ALOX5)^32,33 hydroxyapatite mineralization; tooth enamel calcification (AMTN, CA11);^34,35 and regulators of bone formation, by promoting recruitment and survival of osteoblasts (PTHLH) (Boileau et al., 2001).³⁶ By contrast, genes for proteins that breakdown bone and cartilage (ZFP36L1),³⁷ promoters of osteoclasts (SQSTM1),³⁸ an anchor of osteoclasts to the bone remodeling matrix (SPP1)³⁹ an inhibitor of BMP activity (SMAD9),⁴⁰ inhibitors of the b-catenin-dependent Wnt-pathway, which is central to bone development (DKK1),⁴¹ and genes associated with bone mineral density (ESR1)⁴² decreased after 9 days of serum deprivation (Fig. 1G).

Since AMTN is involved in the mineralization of HAP in tooth enamel, we investigated HAP mineralization in RPE cells. Using a calcium phosphate specific fluorescence dye (Alizarin red), we saw that serum deprived RPE cells accumulated calcium phosphate by day 9 (Fig. 2A). This was confirmed using a calcium phosphate quantification assay measuring Alizarin red absorbance (Fig. 2B). The mean concentration (measured by ELISA) of AMTN secreted in serum-deprived cell culture media was 5.6 μg/ml (Fig. S3). To show that AMTN itself can promote calcium phosphate mineralization under conditions of serum deprivation, 50 mg/ml purified recombinant human AMTN (rh-AMTN) was added to the cultures. AMTN promotes calcium phosphate mineralization under these conditions. Compared to cells cultured in 10% serum, cells cultured in SFM for 9 days formed significantly more calcium nodules. When 50 mg/ml purified recombinant human AMTN (rh-AMTN) was added to the 9-day cultures the number of nodules increased further. (Fig. 2C–D). This was confirmed by light-scattering measurements. Wells containing 100 μg/ml rh-AMTN produced extensive calcium phosphate precipitation after 4 days of incubation at 37°C compared to 0 μg/ ml, 5 μg/ml, 20 μg/ml and 40 μg/ml (Fig. 2E). Higher concentrations of AMTN increased light-scattering, indicating dose-dependent calcium phosphate mineralization. High-resolution light microscopy imaging of the precipitates from the culture wells revealed mineral structures consisting of needlelike crystallites (Fig. 2FG). Staining with calcium phosphate specific fluorescent dye Alizarin red further confirmed the presence of HAP precipitates (Fig. 2H), similar to previous findings of HAP crystals.²² Whereas wells containing buffer only (0 μg/ml rh-AMTN) did not precipitate (Fig. 2I–J).

Fig. 2.

Serum deprivation increases calcium phosphate deposition in ARPE-19 cells. A. Images showing changes in Alizarin red (calcium phosphate) staining in ARPE-19 cells cultured in 6 well plates at day 0 in 10% serum and days 6, and 9 with serum deprivation, n = 3. B. Calcium phosphate quantification assay measuring the Alizarin red absorbance correlated to calcium deposition in cells at day 0 in 10% serum and serum deprived cells at days 6, and 9. One-way ANOVA with post-hoc Dunnett’s test. Mean ± SD. Day 0 vs Day 6 adjusted P = 0.0008, Day 0 vs. Day 9 adjusted P < 0.0001; n = 3. C. Alizarin red staining of cells showing higher levels of positive nodules for nine days serum free media (SFM) and SFM + 50 mg/ml rh-AMTN compared to cells cultured in 10% serum. D. Quantification of Alizarin red staining positive nodules after nine days SFM or SFM + 50 mg/ml rh-AMTN compared to cells cultured in 10% serum. One-way ANOVA with post hoc Tukey’s multiple comparisons test. Mean ± SD. 10% Serum vs. Day 9 SFM adjusted P = 0.0007, 10% Serum vs. Day 9 SFM + 50 mg/ml rh-AMTN adjusted P = 0.0003, Day 9 SFM vs. Day 9 SFM + 50 mg/ml rh-AMTN adjusted P = 0.0186; n = 3. E. Light-scatter plot of mineralization buffers containing 0 to 100 μg/ml rh-AMTN. Buffer vs Days shows a correlation of +0.6938 and P = 0.0179; Days vs 5μg/ml shows a correlation of +0.9821 and P <0.0001; Days vs 20mg/ml shows a correlation of +0.8766 and P = 0.0010; Days vs. 40 μg/ml shows a correlation of +0.9518 and P < 0.0001; Days vs 100 μg/ml shows a correlation of +0.9379 and P < 0.0001. The mineral-promoting effect of rh-AMTN is dose-dependent. F-G. Mineral precipitates from the mineralization buffer containing 100 mg/ml rh-AMTN after 4 days of incubation at 37°C. H. Mineral precipitates from the mineralization buffer containing 100 mg/ml rh-AMTN after 4 days of incubation at 37°C stained with Alizarin red. I-J. Buffer only wells after 4 days of incubation at 37°C. High-resolution light microscopy imaging of HAP mineral structure consisting of needle-like crystallites. K. Confocal images showing AMTN protein expression (green) in cross-sections of ARPE-19 transwell culture monolayers for 6 weeks in 10% serum and nine days SFM, n = 3. L. Confocal images showing accumulation of HAP (bone-tag680RD-magenta) positive sub-RPE deposits in cross-sections of ARPE-19 transwell culture monolayers for 6 weeks in 10% serum and nine days SFM, n = 3. (Color version of figure is available online.)

HAP mineralization in tooth formation requires elevated extracellular calcium and magnesium concentrations and an acidic pH.^43,44 After 9 days in SFM, we found that ARPE-19 cultures had increased extracellular magnesium concentration compared to cells in 10% medium (Fig S4A), and both the intracellular and extracellular pH decreased from 7.5 to 6.4 (Fig S4BD), consistent with conditions permissive for mineralization. To examine AMTN and HAP expression we used AMTN-antibody and bone-tag680RD, which detects HAP, to probe vibratome cross-sections of ARPE-19 monolayers cultured in transwells for six weeks. Compared to transwell cultures of cells grown continually in 10% serum supplemented media, transwell cultures serum deprived for 9 days had a strong positive signal for AMTN (Fig. 2K). Similarly, HAP positive sub-RPE deposits were increased in serum deprived cultures compared to serum supplemented cultures (Fig. 2L).

AMTN is directly involved in HAP mineral formation in serum deprived ARPE-19 cells

Next, we investigated the effect of AMTN knock-down on serum deprived ARPE-19 cells using commercially sourced siRNA. AMTN knock-down by siRNA in serum deprived cells was confirmed by measuring the levels of AMTN mRNA and protein after transfection with AMTN siRNA, GAPDH siRNA or non-targeting (NT) siRNA. Treatment with siRNA specific for AMTN reduced levels of AMTN transcript to very low levels (Fig 3A). A control siRNA for GAPDH had much less effect, showing that suppression of AMTN does not reflect a general loss of transcripts. Surprisingly, the second control (NT) siRNA suppressed, although it did not abolish AMTN transcripts. This might be due to overlapping sequence of the low GC-content non-targeting siRNA with the low GC-content human AMTN sequence; unfortunately, this cannot currently be verified as the NT sequence is proprietary. Western blots confirmed that AMTN protein levels decreased significantly (>90%) following AMTN siRNA transfection, while the GAPDH control had no obvious effect (Fig. 3B). Dual labeling for AMTN mRNA (by ISH) and protein (by IF) confirmed the knock-down effect of AMTN siRNA, and the lack of effect of GAPDH siRNA (Fig. 3C).

Fig. 3.

Serum-deprived ARPE-19 cells form HAP basal deposits which are blocked by siRNA knock-down AMTN (AMTN-KD). A. Quantitative PCR analysis of AMTN gene expression from 9 days SFM, 9 days SFM + AMTN siRNA, 9 days SFM + GAPDH siRNA, and 9 days SFM + nontargeting siRNAs. Bars represent the ratio of the gene expression level compared to the levels in cells cultured in nine days SFM. All measures normalized to the average values of ABCF3 genes. Error bars represent standard deviation. n = 3. B. Representative western blots and quantification of AMTN expression from ARPE-19 cell lysates from cells cultured in 6-well plates with 10% serum, 9 days SFM and both 10% serum and 9 days SFM cells transfected with AMTN siRNA or GAPDH siRNA. Bands were detected for AMTN at 22 KDa and b-actin at 40 KDa. Western blot signals were quantified using ImageJ software (version 1.45; National Institutes of Health, Bethesda, MD). *, P < 0.05; **, P < 0.01. One-way ANOVA followed by Tukey’s multiple comparison test, n = 3. C. In situ hybridization images showing change in AMTN mRNA (green) and protein (blue) expression in 10% serum, nine days SFM, nine days SFM + AMTN-KD cells and 9 days SFM + GAPDH siRNA. D. Cross-section image showing AMTN protein (green) expression in of ARPE-19 cells grown in 9 days SFM and AMTN-KD ARPE-19 cells grown in 9 days SFM with diminished AMTN protein expression (green), n = 3. E. Cross-section images showing accumulated HAP deposit (bone-tag680RD-magenta) in of ARPE-19 cells grown in 9 days SFM and AMTN-KD ARPE-19 cells grown in 9 days SFM with diminished HAP accumulation, n = 3. F. Cross-section images showing EFEMP-1/Fibulin3 protein (green) expression in ARPE-19 cells in serum free media and in SFM + AMTN-KD. G. Cross-section images showing ApoB protein (red) expression in of ARPE-19 cells in serum free media and in SFM + AMTN-KD. (Color version of figure is available online.)

To aid visualization of the effects of AMTN siRNA we transferred cells transfected in 6 well plates to transwells, since the transwell membrane provides a supporting layer to aid in cutting vibratome cross-sections. Figure 3D shows diminished expression of AMTN in AMTN siRNA-transfected cells after nine days in SFM compared to untransfected cells after nine days in SFM: similar to the findings in Fig. 3C. We also observed diminished accumulation of bonetag680RD positive sub-RPE deposits in AMTN knockdown serum-deprived ARPE-19 cells compared to serum-deprived ARPE-19 cells (Fig. 3E). These findings confirm that AMTN is expressed and has a role in sub-RPE mineralized deposits during serum deprivation. In the same transwell system, we examined the expression of two other AMD-related proteins that we have previously shown to be deposited in the serum deprivation model,¹⁵ EFEMP1/Fibulin3 and ApoB, and saw no effect of the siRNA treatment, confirming the specificity of the treatment (Fig 3F,G).

AMTN is expressed in RPE in AMD

Following the confirmation of AMTN expression in cultured human-derived RPE cells in culture we investigated the expression of AMTN in normal and AMD human cadaver eyes. Previous studies have shown that HAP is present in spherules and nodules found within soft drusen in GA AMD eyes.^16,18 From (anonymous) diagnosis information obtained through National Disease Research Interchange (NDRI) and by imaging the deceased donor eyes in our collection we identified AMD eyes with lesions (Fig. S5). These were categorized for types of soft drusen. Normal donor eyes were used for comparison (Fig. S6A–B). In AMD donors we found eyes with regions or clusters of blister-like soft drusen containing spherule structures with intact RPE surrounding the druse (Fig. S6C–D). We also found large nodules with diminished RPE coverage or subducted RPE (Fig. S6E–F). The two drusen in Fig. S6G have both nodules and spherules adjacent to intact RPE and a thick layer of basal laminar deposits (BLamD). In contrast, the druse in Fig. S6H has completely lost its overlaying RPE and has a thin layer of BLamD.

Using BaseScope^® in situ hybridization probes for AMTN mRNA, we found that AMTN is expressed by human RPE in regions surrounding soft drusen (Fig. 4A–B). AMTN signal was found mostly in RPE in regions with soft drusen and not in regions with hard drusen or basal RPE plaque (Fig. S7). We do not see mRNA signal in drusen themselves, although leakage from damaged RPE, due to AMD or post-mortem, could occur. Only a few RPE regions in GA AMD sections without drusen had positive AMTN mRNA signal (Fig. 4C) while most regions without soft drusen had no AMTN mRNA expression. Sections from normal donor eyes and eyes from wet AMD donors were negative for AMTN mRNA (Fig 4D–E). As a positive control for ISH labelling, an AMD donor eye showing mRNA signal for EFEMP1 (Fibulun3) is shown for comparison (Fig. 4F). After identifying tissue sections positive for AMTN mRNA, we probed adjacent sections using fluorescently labeled antibodies to detect AMTN protein expression. This showed that the RPE surrounding soft drusen and some spherule structures within the drusen were positive for AMTN protein (Fig. 4G–H). There was no AMTN immunostaining in any normal donor eye sections (Fig. 4I). Our results confirm that AMTN is expressed in diseased human RPE, specifically in GA AMD with soft drusen.

Fig. 4.

AMTN is expressed in RPE near soft drusen in AMD eyes. A-B. Images showing AMTN mRNA (red) in regions of RPE surrounding drusen. Blue * indicate drusen. C. Image showing AMTN mRNA (red) in a region of RPE without drusen in an AMD eye. D. Image of a human cadaver eye diagnosed with wet AMD negative for AMTN mRNA staining. E. Image of a normal human cadaver eye negative for AMTN mRNA staining. F. Positive control image of a human cadaver AMD eye hybridized with EFEMP1 mRNA (red) in the RPE. Black arrows indicate regions with positive in situ mRNA labeling. G. Image of an AMD eye with drusen and RPE positive for AMTN protein (green) staining. H. Image of another AMD eye with drusen and RPE positive for AMTN protein (green) staining. White * indicates drusen. I. Image of a normal eye negative for AMTN protein (green) staining in RPE. White arrows indicate regions of positive AMTN protein labeling. Cryosections were treated with TrueBlack Lipofuscin Autofluorescence Quencher (Biotium, Inc, Fermont CA) prior to immunofluorescence staining to obtain true-positive fluorescence signal in RPE. Cryosections stained with; nuclei: DAPIblue; photoreceptors: PAN-magenta; AMTN-green; autofluorescence-assigned cyan. PR, photoreceptors; RPE, Retinal pigment epithelium. (Color version of figure is available online.)

AMTN is found within drusen with calcified spherules and nodules

Next, we used HAP-specific fluorescent staining and AMTN immunofluorescence (IF) with superresolution Airyscan confocal imaging techniques to investigate structures in large soft drusen. Xylenol Orange and Alizarin Red S labeling displayed very similar staining patterns for HAP and showed that, as previously reported,^16,18 the number of HAP spherules varied among sub-RPE deposits (Fig 5A–H). IF detected AMTN throughout the soft drusen but also strongly localized in ring-like structures (Fig. 5I–J). Z-series images of the drusen, reveals spherules associated with AMTN. With the phase-contrast background, the spherule structures and AMTN localization around them were clearly visible (Fig. 5K–L).

Fig. 5.

Colocalization of HAP spherules and nodules and AMTN. A, E. Autofluorescence images of individual druse (due to overlapping emission spectra, and autofluorescence was assigned cyan for clarity of images). B, F. Xylenol Orange labeling. C, G. Alizarin Red S labeling. D, H. Combined images. I, J. AMTN labeling of HAP spherules of another druse in 2 different slices of a Z-image series with black background. Autofluorescence was assigned blue. K, L. AMTN labeling of HAP spherules of a druse in 2 different slices of a Z-image series with phase-contrast background. M. Bone-Tag 680 RD labeling of a large drusen with HAP spherules and nodules. N. AMTN labeling of the large drusen. O. Combined image of AMTN and Bone-Tag HAP labeling. P-R. Magnified images of AMTN and Bone-Tag 680 RD co-labeling of HAP spherules and nodules isolated from soft drusen. RPE, Retinal pigment epithelium; BrM, bruch’s membrane. (Color version of figure is available online.)

The binding of different proteins to the HAP surface has been suggested as a potential mechanism for the formation of the macroscopic sub-RPE deposits observed in patient fundus and OCT imaging.^16,18 We isolated drusen with large calcified spherules and nodules from the tissue cross-sections. Cross-sections of a drusen with calcified nodules showed heterogeneous HAP staining, with intense staining of the crusts (Fig. 5M), whereas AMTN was localized in a ring around the HAP structures (Fig. 5N–O). Co-labeling of AMTN and HAP from numerous large calcified soft drusen displayed similar patterns (Fig. 5P–R). The OCT image of an eye from an 80-year-old Caucasian female with advanced AMD taken 2 months prior to her death clearly shows a large calcified drusen (Fig. 6A–G). Three-dimensional images generated from Z-series imaging of the druse corresponding to the OCT showed that AMTN ring-like localization continued throughout the druse (Fig. 6H–I). OCT images of a 79-year-old Caucasian female and an 89-year-old Caucasian male with advanced AMD showed large calcified lesions, and the corresponding confocal images of both drusen showed AMTN positive calcified nodules (Fig. 6J–M).

Fig. 6.

OCT images of heterogeneous internal reflectivity in drusen and 3-dimensional confocal images of AMTN labelling of nodules in the calcified druse. OCT imaging of an eye from an 80-year-old Causcasian female with advanced AMD imaged in vivo2 months prior to death and super-resolution confocal imaging of the druse corresponding to the OCT postmortem. A. Near-infrared (NIR) image shows hyper-reflective large drusen. The Yellow line showing the level of OCT scan. B-E. OCT images corresponding to 1st, 2nd, 4th and 5th Yellow lines from the NIR image. F. OCT image showing heterogeneous internal reflectivity in drusen. Calcified lesions in OCT are highlighted with red frames. G. Magnified area of the calcified druse appear hyperreflective on NIR, loss of hyper-reflective RPE and surrounding hyper-reflective dots on OCT. H-I. Druse corresponding to the OCT. Three-dimensional volume rendered super-resolution confocal images of AMTN positive labeling of the calcified druse. Calcified lesions in OCT are highlighted with red frames. J. OCT of the eye of a 79-year old Caucasian female with advanced AMD imaged in vivo1 year and 10-months prior death. K. AMTN labeling (green) of the large calcified drusen from the same eye postmortem. L. OCT of the eye of an 89-yearold Caucasian male with advanced AMD imaged in vivo 5 years prior death. M. AMTN labeling (green) of the large calcified drusen from the same eye postmortem. (Color version of figure is available online.)

Six normal and 22 AMD eyes from male and female donors with an age range of 6596 years were examined in the present study. Of the total (n= 28), 13 eyes contained soft drusen with spherules, nodules or both. AMTN was detected in 10 out of the 13 aforementioned eyes, specifically in the eyes with larger soft drusen with spherules and in six eyes which also had large calcified nodules. Donor eyes with Y/Y genotype, did not contain large soft drusen with HAP spherules or nodules and we did not find AMTN positive signal (Tables S1, S2). In contrast, donor eyes with H/H and H/Y genotypes, had soft drusen with HAP spherules and/or nodules and some which stained positive for AMTN mRNA and protein (Tables S1, S2). Our findings suggest that AMTN may be a key protein in the organization of HAP mineralization and deposition of sub-RPE deposits implicated in clinical AMD progression.

DISCUSSION

Accumulation of protein- and lipid-containing deposits basal to the RPE is common in the aging eye and is a hallmark of AMD.^17,24 Calcification of the elastin layer of Bruch’s membrane is also a sign of aging and, recently, calcified HAP spherules were discovered on the surface of lipid droplets in the inner aspect of Bruch’s membrane.^14,16

We have described a cell-culture model in which serum-deprivation of human-derived ARPE-19 cells results in responses similar to processes involved in AMD progression.^15,20 Our RNAseq results confirmed previous results and revealed another response. Late in the time course, amelotin, a protein mainly known for the formation of tooth enamel through organization of HAP mineralization^21,22 was strongly activated. Some other genes associated with development and mineralization of bone also increased by nine days in SFM. In addition to these genes with general roles in bone, we also observed that Odontogenic, Ameloblast Associated (ODAM), another gene implicated in enamel formation,⁴⁵ was induced by day 9, although at a much lower level than AMTN. No other genes for proteins specifically related to tooth formation including enamelin, amelogenin, and ameloblastin were expressed.

Furthermore, the RPE cells that formed HAP after nine days in SFM had increased extracellular magnesium concentration compared to cells in 10% medium as well as an acidified intracellular and extracellular pH, conditions conducive for mineralization as previously demonstrated in HAP formation in human dental enamel.^43,44 These findings support previous speculations that high overall extracellular calcium and magnesium concentrations and acidic pH promote HAP spherule formation and a lower magnesium concentration and neutral pH promotes HAP nodule formation in the sub-RPE space.^18,46

Although several genes related to calcification were upregulated at the same time, AMTN appears to have a critical role in the formation of HAP deposits in RPE cells in culture. Addition of recombinant AMTN alone increases HAP deposition and, most significantly, knock-down of AMTN by targeted siRNA reduced AMTN protein levels and diminished HAP deposition. At the same time, knock-down of AMTN, had no major effect on other important proteins that are deposited both in the serum deprivation cell culture model and in AMD, EFEMP1/Fibulin3 and ApoB.

Thus, in addition to induction of lipid pathways and the deposition of key AMD-related proteins,¹⁵ the serum-deprivation model also shows that under stress conditions RPE cells are capable of responses that promote biomineralization. In other cell types, calcification involves membrane vesicles.⁴⁷ Vesicles from bone and cartilage and human aortic smooth muscle cells cultured in serum free media are associated with extracellular matrix mineralization.^48–50 The vesicles concentrate calcium and phosphate in a protected microenvironment, thereby promoting mineral nucleation events.^48,50 We have previously shown that RPE cells cultured in serum-free media increased expression of membrane bound vesicles; it will be interesting to see whether these vesicles have a similar role in calcification.⁴⁶

It has been proposed that drusen containing HAP spherules and nodules participate in the development and progression of GA.¹⁶ Although it has been documented that sub-RPE deposit growth is associated with, and possibly even initiated by, the precipitation of HAP and the binding of proteins to the HAP surface,¹⁸ the proteins involved in the mineralization process have not previously been identified. Our results show that AMTN is expressed in the serum deprivation RPE cell-culture model, is also expressed in human RPE specifically associated with soft drusen (the form of drusen most closely associated with AMD) that contain HAP spherules and nodules in donor eyes with dry AMD, but not in normal eyes or those with wet AMD. HAP spherules are also found in hard drusen, but amelotin is not, suggesting that hard and soft drusen may have different calcium. Although our cell culture model induces AMTN through serum deprivation, it is possible that other forms of stress, which remain to be tested, could have similar effects.

We note that SAGE analysis of pooled adult RPE several years ago⁵¹ also detected transcripts of AMTN (https://neibank.nei.nih.gov/cgi-bin/EyeSAGE/showEyeSageGenes.cgi?searchField=unigene&searchTerm=453069), although the significance of this was not realized at the time. Collectively, these findings suggest that AMTN expression can occur in human RPE, triggered by aging, stress or disease. In the presence of calcium and phosphate during pathological aging,⁵² AMTN can play a key role in HAP mineralization and nodule proliferation, conditions that are increasingly seen as an important part of disease progression in dry AMD.¹⁸

Why trigger this program in RPE? While the outer retina depends on the blood supply of the choroid for its health and function, it is also vulnerable to incursion from the vasculature, as seen in CNV, and from immune attack.⁵³ The RPE/BrM complex has evolved to allow transport of molecules in and out of the retina, but it also plays a key part in maintaining the blood-retina barrier.^54,55 If local damage occurs to RPE/BrM, it could allow invasion by the vasculature or immune cells that would rapidly destroy the retina. Possibly, RPE maintains a defensive program to patch vulnerable areas of the blood-retina barrier using basal deposits of lipids, proteins and HAP. For small areas of RPE, this might allow repair to occur. In AMD, a more widespread response might become pathological, leading to increasing loss of photoreceptors. As such, local inhibition of AMTN-induced HAP mineralization in RPE may provide a new opportunity for therapeutic intervention in this major health problem.

AT A GLANCE COMMENTARY.

Wistow G, et al.

Background

Deposition of hydroxyapatite (HAP) basal to retinal pigment epithelium (RPE) is linked to the progression of dry age-related macular degeneration (AMD). However, the mechanisms of HAP deposition is unclear.

Translational Significance

Using a serum deprived RPE cell culture model we demonstrated HAP deposition and amelotin (AMTN), a promoter of HAP mineralization during tooth enamel formation is also expressed in RPE. Both are blocked by siRNA inhibition of AMTN. Furthermore, in human donor eyes, AMTN is expressed in RPE associated with soft drusen and localizes to HAP deposits in dry AMD. Thus, AMTN may be a new target for therapeutic intervention in dry AMD.

Abbreviations

AMD

age-related macular degeneration

AMTN

Amelotin

CNV

neovascularization

GA

geographic atrophy

HAP

hydroxyapatite

RPE

Retinal pigment epithelium