Improved Lipofuscin Models and Quantification of Outer Segment Phagocytosis Capacity in Highly Polarized Human Retinal Pigment Epithelial Cultures

Summary

This protocol describes a lipofuscin accumulation model in highly differentiated and polarized human retinal pigment epithelial (RPE) cultures and an improved outer segment (OS) phagocytosis assay to detect the total OS consumption/degradation capacity of the RPE. These methods overcome the limitations of previous lipofuscin models and classical pulse-chase outer segment phagocytosis assays.

Abstract

The daily phagocytosis of photoreceptor outer segments by the retinal pigment epithelium (RPE) contributes to the accumulation of an intracellular aging pigment termed lipofuscin. The toxicity of lipofuscin is well established in Stargardt's disease, the most common inherited retinal degeneration, but is more controversial in age-related macular degeneration (AMD), the leading cause of irreversible blindness in the developed world. Determining lipofuscin toxicity in humans has been difficult, and animal models of Stargardt's have limited toxicity. Thus, in vitro models that mimic human RPE in vivo are needed to better understand lipofuscin generation, clearance, and toxicity. The majority of cell culture lipofuscin models to date have been in cell lines or have involved feeding RPE a single component of the complex lipofuscin mixture rather than fragments/tips of the entire photoreceptor outer segment, which generates a more complete and physiologic lipofuscin model. Described here is a method to induce the accumulation of lipofuscin-like material (termed undigestible autofluorescence material, or UAM) in highly differentiated primary human pre-natal RPE (hfRPE) and induced pluripotent stem cell (iPSC) derived RPE. UAM accumulated in cultures by repeated feedings of ultraviolet light-treated OS fragments taken up by the RPE via phagocytosis. The key ways that UAM approximates and differs from lipofuscin in vivo are also discussed. Accompanying this model of lipofuscin-like accumulation, imaging methods to distinguish the broad autofluorescence spectrum of UAM granules from concurrent antibody staining are introduced. Finally, to assess the impact of UAM on RPE phagocytosis capacity, a new method for quantifying outer segment fragment/tips uptake and breakdown has been introduced. Termed "Total Consumptive Capacity", this method overcomes potential misinterpretations of RPE phagocytosis capacity inherent in classic outer segment "pulse-chase" assays. The models and techniques introduced here can be used to study lipofuscin generation and clearance pathways and putative toxicity.

Introduction

The retinal pigment epithelium (RPE) provides critical support for overlying photoreceptors, including the daily uptake and degradation of photoreceptor outer segment tips or fragments (throughout this protocol, the abbreviation OS stands for OS tips or fragments rather than whole outer segments). This daily uptake in the post-mitotic RPE eventually overloads phagolysosomal capacity and leads to the buildup of undigestible, autofluorescent intracellular material, termed lipofuscin. Interestingly, several studies have also demonstrated that RPE lipofuscin can accumulate without OS phagocytosis^1,2. Lipofuscin has many components, including cross-linked adducts derived from visual cycle retinoids, and can occupy nearly 20% of RPE cell volume for those over the age of 80³.

Whether lipofuscin is toxic has been hotly debated. Stargardt's disease is an autosomal recessive degeneration of the photoreceptors and RPE in which a mutation in ABCA4 triggers improper processing of visual cycle retinoids contained within photoreceptor outer segments. Improper retinoid processing leads to aberrant cross-linking and formation of bis-retinoid species, including the bis-retinoid N-retinylidene-N-retinylethanolamine (A2E). Studies have demonstrated multiple mechanisms for A2E toxicity^4,5. Lipofuscin contributes to fundus autofluorescence signals during clinical imaging, and both Stargardt's patients and animal models display increased fundus autofluorescence prior to retinal degeneration, suggesting a correlation between lipofuscin levels and toxicity^6,7. However, with age, lipofuscin accumulates in all humans without triggering an RPE degeneration. Further, in age-related macular degeneration (AMD), where RPE degeneration occurs only in elderly patients, those with early and intermediate forms of the disease have less fundus autofluorescence signals than age-matched non-diseased humans⁸. These clinical findings have been verified at the histologic level as well^9,10.

Animal models of RPE lipofuscin accumulation have also left some ambiguity about lipofuscin toxicity. The ABCA4 knockout mouse does not display retinal degeneration on a pigmented background, whereas it does on an albino background or when exposed to blue light^11,12. Further, the toxicity of lipofuscin derived via ABCA4 knockout likely differs from the more slowly accumulating lipofuscin that occurs with natural aging, as seen in AMD¹³.

In vitro models of lipofuscin accumulation provide an alternative to studying the effects of lipofuscin accumulation on RPE health. Such models allow for manipulating lipofuscin components, from feeding single retinoid components to feeding OS, and allow study in human rather than animal RPE. In the last couple of decades, multiple methods have been developed to model RPE lipofuscin in culture. Along with other groups, Dr. Boulton's group fed bovine OS daily for up to three months on passage 4 to 7 human primary RPE cells from donors aged 4 to 85 years old¹⁴. Alternatively, inhibition of autophagy has also led to lipofuscin accumulation in passages 3 to 7 primary human RPE cultures¹⁵. However, sub-lethal lysosomal inhibition in highly differentiated, passage 1, primary human pre-natal RPE (hfRPE) cultures failed to induce lipofuscin, even with the repeated addition of OS on a daily basis¹⁶.

As a more reductionist approach, others have fed single lipofuscin components to cultures, especially the bis-retinoid A2E^4,17. Such studies are valuable in that they define potential direct mechanisms of toxicity for individual lipofuscin components, implicating, for example, lysosomal cholesterol and ceramide homeostasis¹⁸. At the same time, there is debate about the toxicity of A2E¹⁹, and feeding it directly to cells circumvents the typical pathway for lipofuscin accumulation, which involves phagocytosis of photoreceptor OS. In an attempt to deliver all components of lipofuscin to RPE cultures, Boulton and Marshall purified lipofuscin from human eyes and fed this to passage 4 to 7 human primary RPE cultures derived from both fetal and elderly human donors²⁰. While innovative, this method represents a limited lipofuscin source for repeated experiments.

While repeated feedings of OS to RPE cultures produce lipofuscin in many systems, it fails to do so in highly differentiated primary RPE cultures¹⁶. Photo-oxidizing OS induces cross-linking reactions like bis-retinoid formation that naturally occurs during lipofuscin formation in vivo. This can accelerate lipofuscin-like granule formation in RPE culture systems, even those that are highly differentiated and resistant to lipofuscin accumulation¹⁶. Here, a method to induce lipofuscin-like granule accumulation in highly differentiated hfRPE and human iPSC-RPE is introduced, modified from Wihlmark's published protocol²¹. This method has the advantage of inducing lipofuscin-like granules employing the same source (photoreceptor OS) and pathway (phagolysosomal OS uptake) as occurs for lipofuscinogenesis in vivo. Further, it is done on human RPE cultures that are highly differentiated and validated in multiple studies to replicate human RPE in vivo^22,23,24. These lipofuscin-like granules are termed undigestible autofluorescent material (UAM), and provide data and discussion in this protocol comparing UAM to in vivo lipofuscin. Along with methods for building and evaluating UAM-laden cultures in highly differentiated human RPE, an updated method to assess RPE OS phagocytosis is also introduced. Multiple excellent pulse-chase methods for quantifying OS phagocytosis have been introduced, including Western blotting, immunocytochemistry, and FACS^25,26,27. However, early in the OS pulse-chase, conditions that lead to poor OS uptake can be conflated with conditions that promote rapid degradation of internalized OS. The method presented here measures the total amount of introduced OS that is fully consumed/degraded by the RPE ("Total Consumptive Capacity"), helping eliminate this ambiguity. It is anticipated that insights about lipofuscin toxicity utilizing these protocols, including effects on OS phagocytosis rates utilizing the "Total Consumptive Capacity" method, will be used to shed light on the toxicity of lipofuscin in vivo.

Protocol

The present protocol involving the acquisition and use of human tissue was reviewed and approved by the University of Michigan Institutional Review Board (HUM00105486).

1. Preparation of photo-oxidized outer segment tips and fragments

NOTE: Dark-adapted bovine retinas were purchased and shipped on ice (see Table of Materials). From these retinas, OS were purified following a previously published protocol²³.

1. Outer segment cross-linking with a UV lamp
   1. Immerse polytetrafluoroethylene-coated slides (see Table of Materials) completely in 70% ethanol for 10 min in a biosafety cabinet to sterilize the slides. Allow the slides to airdry in a sterile 100 mm cell culture dish.
   2. Place each slide in one new 100 mm cell culture dish and add 200 µL of serum-containing cell culture media (whatever media is typically used for the RPE cultures at hand) into each rectangle, then place a handheld 254 nm UV light (see Table of Materials) on the 100 mm cell culture dish (with no lid) with bulb directly facing the slide, and expose for 20 min. The media specifically used for this study and the specific product numbers for the components of the media have been previously published^{22, 23}. This media is termed "RPE media" throughout this protocol.
      NOTE: UV treatment of the media blocks the glass surface and prevents OS from sticking during the next steps.
   3. Thaw frozen aliquots of OS at 37 °C (in hand or via water bath). The amount of OS needed depends on the application. In general, one can treat up to 500 µL of 2 x 10⁸ OS/mL per rectangle on the slide.
   4. Once thawed, spin the OS at 2400 x g for 5 min at room temperature. Immediately aspirate the supernatant in the biosafety cabinet using a pipette rather than a vacuum to prevent accidental loss of the pellet.
   5. Gently resuspend the pellet with sterile PBS.
      NOTE: Keep track of the volume resuspended and the concentration expected. e.g., resuspending the pellet from a 1 mL tube of 1 x 10⁸ OS/mL with 500 µL PBS yields 2 x 10⁸ OS/mL. The maximum volume and concentration per rectangle on the polytetrafluoroethylene-coated slide is 500 µL of 2 x 10⁸ OS/mL.
   6. Aspirate the media from the slides and place up to 500 µL of 2 x 10⁸ OS/mL PBS solution on each rectangle of the slide. Place the handheld UV light over the cell culture dish (with no lid) with the bulb directly facing the OS. Expose the OS solution to 254 nm light for 40 min.
      NOTE: During the 40 min exposure, cover the UV lamp and dish with a towel or absorbent pad, close the biosafety cabinet sash, and turn the blower off. This will prevent any significant evaporation from occurring. If the UV lamp used in this protocol is not available to a researcher, there are two alternatives to ensure the appropriate amount of cross-linking is achieved. The first is to follow the quantitative UV exposure outlined in step 1.2, either with the device outlined in step 1.2 or another device that can provide the same quantitative radiant exposure as the said device in 1.2. Another alternative is to expose OS to UV irradiation for a time that induces the degree of cross-linking on the Coomassie stain seen in Figure 1C.
   7. Collect the treated OS PBS solution in sterile microcentrifuge tubes, rewashing the rectangle of the coated slide with 200-500 µL of PBS (2-3x), and collecting each wash in the microcentrifuge tube. Once done, look at the slide under a tissue culture room microscope to confirm that nearly all OS have been removed from the slide.
   8. Spin down the OS PBS solution at 2400 x g for 5 min at room temperature, aspirate the PBS in the biosafety cabinet with a pipettor, and resuspend in 500 µL standard media that will be used for RPE cultures (e.g. - "RPE media" defined in section 1.1.2). The resuspension volume can be adjusted based on the desired photo-oxidized OS (OxOS) concentration.
   9. Place 10 µL of the suspension in a new microcentrifuge tube, dilute 50-100x with more cell culture media, and count OS's with a hemocytometer.
   10. Based on the OS count, dilute the OxOS suspension to a final stock concentration. For feeding highly mature RPE cultures in 24-well Transwells (surface area of 0.33 cm²; see Table of Materials), a final media concentration of 2 x 10⁷ OS/mL is recommended.
       1. To achieve this, distribute the OxOS into 25 µL aliquots at a 5.6 x 107 OS/mL concentration, suspended in standard RPE culture media. After thawing a 25 µL aliquot, mix the entire aliquot with 45 µL of standard RPE culture media, providing a final media volume of 70 µL and OS concentration of 2 x 10⁷ OS/mL for each OxOS Transwell feeding.
   11. Before aliquoting the OxOS, add phagocytosis bridging ligands (Protein S and MFG-E8, see Table of Materials), which facilitate OS uptake and lipofuscin accumulation. The working concentrations of bridging ligands in a 24-well Transwell are 4 µg/mL for human purified Protein S and 1.5 µg/mL for human recombinant MFG-E8. Thus, for 25 µL aliquots of OxOS at 5.6 x 10⁷ OS/mL, the stock concentration for Protein S is 11.2 µg/mL, and for MFG-E8 is 4.2 µg/mL.
       NOTE: The bridging ligand concentrations were optimized for hfRPE cultures on 24-well Transwells, with an approximate cell count of 320,000 cells per 0.33 cm² well. Ligand concentrations may need to be adjusted for other RPE types and cell densities since ligands present in excess of phagocytosis receptor availability may paradoxically block OS uptake²⁸.
   12. Snap freeze the aliquots in liquid nitrogen.
       NOTE: Multiple freeze-thaws are highly detrimental to OS integrity.
2. Outer segment cross-linking with a UV crosslinker device
   NOTE: Follow step 1.1, with the exception of the steps below, which replace steps 1.1.2 and 1.1.6, respectively:
   1. (replaces step 1.1.2) Place each slide in one new 100 mm cell culture dish and add 200 µL of serum-containing cell culture media (e.g. -"RPE media" or any other substitute) into each rectangle, then place slides in an Ultraviolet Crosslinker device (see Table of Materials), treating with 254 nm UV at a radiant exposure of 3-6 J/cm² to block the glass surface and prevent OS sticking during the next steps.
   2. (replaces step 1.1.6) Aspirate the media from the slides and place up to 500 µL of 2 x 10⁸ OS/mL in sterile PBS on each rectangle of the slide. Place slides in the Ultraviolet Crosslinker device, set treatment radiant exposure as needed (3-9 J/cm²), and treat at 254 nm.
      NOTE: The radiant exposure needed can be carefully determined by titrating radiant exposure between 3-9 J/cm² until the degree of autofluorescence, and protein cross-linking (as seen in Figure 1B and Figure 1C) is achieved.
      CAUTION: Handling OS outside a biosafety cabinet can result in contamination. After exposure of OS to the UV Crosslinker Device, collect OS with sterile pipette tips, transfer to sterile microcentrifuge tubes, and handle all further steps in the biosafety hood.
3. Characterization of Oxidized OS
   1. Quantify autofluorescence emission spectrum by imaging
      1. Place 20-50 µL of stock solution from untreated OS and OxOS into two separate microcentrifuge tubes. Centrifuge at 2400 x g for 5 min at room temperature, resuspend pellet in buffered (e.g. PBS) 4% paraformaldehyde, and fix for 15min at room temperature.
      2. After fixation, spin down as above, wash with PBS x 2 (spinning down between washes), then resuspend in less than 30 µL of PBS.
      3. Place a few microliters of the resuspension on a microscope slide, add mounting media (see Table of Materials), and finally a coverslip.
      4. Image OxOS on a confocal microscope (see Table of Materials).
         NOTE: OxOS autofluorescence can be excited by a wide range of laser wavelengths, but a 405 nm or 488 nm laser line is typically employed. Emission is similarly broad, but a typical GFP/FITC excitation/emission filter set-up is adequate to view autofluorescence.
      5. To measure the OxOS emission spectrum, use λ-scanning on a confocal microscope. Typical λ scan settings include excitation with 405 nm or 488 nm, and detection of emission that ranges from 10 nm red-shifted from the laser excitation line all the way to approximately 800 nm, with a λ step size of 10 nm. Each microscopy system has its own directions on how to employ the λ mode, and the reader need to refer to the manual for their specific confocal.
   2. As an alternative, quantify autofluorescence by flow cytometry.
      1. Spin down 2 x 10⁷ untreated OS and separately 2 x 10⁷ OxOS in microcentrifuge tubes and resuspend in 1 mL of PBS.
         NOTE: Fixation is not required if flow cytometry is performed directly after thawing.
      2. Load untreated OS and OxOS samples onto flow cytometer (see Table of Materials). Adjust forward scatter (FSC) and side scatter (SSC) to an acceptable spread in the FSC-SSC scatter graph. Use PBS as a control to rule out any contaminating small particles. Note that OS are considerably smaller than cells.
      3. Use the flow cytometer's standard FITC channel for autofluorescence quantification. Count at least 10,000 events. Use the FITC histogram's pulse area value to represent fluorescence intensity.
         NOTE: For the present study, all data are analyzed using flow cytometer analysis software (see Table of Materials), following manufacturer's instructions.
   3. Assess the degree of cross-linking in OxOS
      1. Spin down 2 x 10⁷ untreated OS and separately 2 x 10⁷ OxOS in microcentrifuge tubes. Remove supernatant and directly lyse pellet by adding 1.2x Laemmli Sample Buffer (enough to cover; see Table of Materials). Vortex, keep at room temperature for 30 min, spin down at room temperature at equal or greater than 12000 x g for 10 min, and collect supernatant.
         NOTE: Assessing the degree of protein cross-linking in OxOS gives a sense of the adequacy of the UV treatment. Comparison is made between untreated OS and OxOS, and adequate cross-linking occurs when the monomeric rhodopsin band that dominates the Coomassie staining (Figure 1C, arrow) is just perceptible, with emerging of higher order aggregates and a protein smear at the top of the gel.
         ​CAUTION: When using Sample Buffer to lyse the OS, do not subsequently heat the lysis solution, as this can trigger rhodopsin aggregation. Also avoid cooling the lysed OS until ready for storage, as this will precipitate SDS. Unused lysates can be kept at -20 °C. If rethawing, be sure the lysates are completely thawed at room temperature before use.
      2. Quantify protein concentration using a protein assay that is tolerant to high SDS and reductant concentrations²⁹. See Table of Materials for suggestions on appropriate protein assay reagents, and follow manufacturer protocol for these reagents.
      3. Run untreated OS and OxOS samples by SDS-PAGE electrophoresis³⁰ on a 4%-15% gradient gel, employing standard tris-glycine SDS buffer. Gel is run at 80 V until samples enter the stack, then at 120 V for 50-60 min at room temperature.
      4. Stain the gel with Coomassie blue using standard protocols³¹. The Coomassie staining will demonstrate the adequacy of cross-linking induced by UV treatment.

2. Building lipofuscin-like granules (UAM) in RPE cultures

1. OxOS feedings: amount, frequency, and phagocytosis bridging ligands
   NOTE: Feedings occur on human iPSC-RPE or hfRPE cultures at passage 1, grown according to the protocol outlined by the Bharti lab (for iPSC-RPE)³² or our previously outlined protocol for hfRPE, adapted from the Sheldon Miller lab^22,23. All calculations below are based on feeding OxOS or OS to one 24-well Transwell (6.5 mm diameter, 0.33 cm² growth area).
   1. Thaw 25 µL of 5.6 x 10⁷ OxOS/mL at 37 °C and add 45 µL of cell culture media to the OxOS aliquot, resulting in a final volume of 70 µL.
      NOTE: OxOS aliquots prepared in step 1 will contain phagocytosis bridging ligands MFG-E8 and Protein S. However, if those ligands were not added to the aliquots prior to freeze-down, they can be added at this step, ensuring the final concentration of the ligands in the media to be fed cells is 4 µg/mL for Protein S and 1.5 µg/mL for MFG-E8. As stated in step 1.1.11, concentrations of bridging ligands may need to be altered for other RPE types or cell densities.
   2. Remove apical media from Transwell and add 70 µL of 2 x 10⁷ OxOS/mL with the phagocytosis bridging ligands to the apical chamber. After 24 h, remove and replace with a new OxOS feeding. Feeding occurs daily during weekdays, skipping weekends, until 20 feedings are completed (~1 month). Change basolateral cell culture media (400-550 µL) 2-3x/week.
   3. Upon completion of 20 feedings, resume normal media changes for the well.
      NOTE: It takes several additional media changes to wash off sticky OxOS from the RPE apical surface, so experiments with the UAM-laden cultures must be done at least 1-2 weeks after OxOS feedings conclude.
      CAUTION: It is important to have proper controls for UAM buildup via OxOS feedings. As daily media changes can affect RPE biology, the following control wells are recommended: Control 1: Replace media daily during weekdays for same number of feedings as OxOS-treated group. Control 2: Feed RPE untreated OS daily during weekdays at the same concentration and volume as OxOS feedings, for the same number of feedings.
2. Monitoring health of cultures laden with lipofuscin-like granules by trans-epithelial electrical resistance (TEER) and cell death assays
   NOTE: During and after OxOS feedings, the health of RPE cultures can be measured by assessing RPE tight-junction integrity and cell death. It has been previously shown that assessment of tight-junction integrity, via measuring trans-epithelial electrical resistance (TEER), is a sensitive marker for general cell health³³. More traditional non-invasive markers of cell death, such as release of lactate dehydrogenase (LDH), can also be employed.
   1. Perform TEER
      NOTE: TEER is tested with a trans epithelial electrical resistance (TEER) meter and TEER electrode, generally following manufacturer's instructions (see Table of Materials).
      1. To sterilize the TEER electrode, use a task wiper soaked in 70% ethanol to clean the electrode and then immerse the tips of the electrode probe in 70% ethanol for 10 min.
      2. Let the electrode completely dry, then immerse the electrode in sterile media.
      3. Remove the probe from sterile media and insert the TEER electrode's two probes into the apical and basolateral chambers of a cultured Transwell; the longer probe tip fits into the basolateral chamber.
         NOTE: It is easy to scrape the bottom of the apical chamber with the TEER electrode without practice, and such scraping will dramatically alter TEER readings as the confluent RPE monolayer is disrupted. Thus, it's recommended that beginners practice on non-important Transwells prior to testing on experimental RPE cultures. Further, after each plate of RPE cultures is tested, the experimenter should check for any scraping of the RPE monolayer under a standard tissue culture microscope.
      4. Once the apical and basolateral probes are in place in the Transwell, press the read button or step on the foot switch of the meter to record the TEER. Between plates or groups of cells, wash the electrode probe with sterile media. If infection is a high concern, resterilize between plates.
      5. Calculate TEER by taking the resistance reading from the meter, subtracting the value of a blank Transwell with media but no cells (generally 100-110 Ω for a 24-well Transwell with 0.33cm² surface area), and then multiplying by the surface area of the Transwell.
         NOTE: TEER values must be reported normalized to cell surface area. Typical values for healthy hfRPE cultures range from 350-1100 Ωcm². TEER values increase as temperature decreases. Thus, when removing cultures from a 37 °C incubator to a room temperature hood, TEER values will tend to increase across the plate. To guard against this variability, either work quickly (if experienced) or wait for 10-15 min for plate temperature to equilibrate with the room temperature.
   2. Perform LDH release assay
      NOTE: Release of LDH into cell culture supernatant occurs during cell death and be measured with a standard kit (see Table of Materials).
      1. Collect the supernatant from above the cells after a 24 h incubation and dilute to 1:100 using standard media.
      2. Induce total possible LDH release, which is important for normalizing all values from step 2.2.2.1, by addition of 2 µL 10% triton X-100 to the 100 µL of apical media from control cells in a 24-well Transwell. After incubating the triton/cell supernatant mixture for 15 minutes at 37 °C, mix the media above the now-lysed cells and collect to measure total possible LDH release.
      3. Once supernatants are collected, perform the assay using manufacturer standard instructions, measuring luminescence after a 30 min incubation using the kit's buffer solution.
         NOTE: All LDH values are normalized to the total possible amount of LDH release.
3. Characterization of lipofuscin-like granule spectrum and composition
   1. Acquiring autofluorescence quantification and spectra
      1. Fix UAM-laden cultures with 4% PFA at room temperature for 15 min, followed by PBS wash 5x.
      2. Keep a small amount of PBS in the apical chamber and then flip the Transwell upside down. Using a dissecting microscope and a razor blade, cut the semi-porous membrane from the Transwell, applying cutting force at the junction of the membrane and Transwell.
         NOTE: Keep consistent about how close to the lip of the Transwell the cut is made, as the Transwell membrane has a tendency to warp and wrinkle when cuts are not consistent.
      3. Once the Transwell membrane is cut, immediately place onto a microscope slide using forceps, being careful to avoid touching parts of the Transwell membrane with cells. Wick away excess PBS with a task wipe, while avoiding touching the cells. Add mounting media and a coverslip. Ensure to keep track of which side of the Transwell is "right-side up" when coverslipping the Transwell.
      4. Acquire the autofluorescent intensity and spectra using the same settings and procedures as step 1.3.1.4 and step 1.3.1.5.
         NOTE: When imaging UAM, an important confounder is that OxOS are autofluorescent and often stick to the RPE apical surface for days and sometimes even weeks after completion of OxOS feeding. As a result, when quantifying UAM, a method needs to be employed to ensure measured autofluorescence is coming from UAM and not OxOS. The easiest method is to co-stain lipofuscin cultures with a rhodopsin antibody. For example, anti-rhodopsin antibody 4D2 can be used at a 1:1000 dilution and with a standard PFA-fixation immunocytochemistry protocol³⁴. The secondary antibody for rhodopsin should be a far-red dye-conjugated antibody (e.g. with an excitation maxima near 647 nm), as UAM and OxOS autofluorescence tends to be weakest in this wavelength. Confocal images can then be sequentially acquired in two channels, exciting the UAM and OxOS autofluorescence with a 405 nm laser and emission from 415 nm to 550 nm, while residual rhodopsin, indicating undigested OxOS, can be excited with standard far-red dye imaging parameters in a separate channel. Post-acquisition, the rhodopsin channel can be used as a subtractive mask in a program like ImageJ to remove autofluorescence coming from sticky remaining OxOS, leaving behind just UAM autofluorescence to quantify.
         CAUTION: Even OS that are not photo-oxidized display some autofluorescence, and even with rigorous washing, some OS and OxOS stick to the RPE surface. Thus, without generating a subtractive mask using rhodopsin immunostaining, as suggested in the above NOTE, accurate quantification of UAM autofluorescence levels in cultures fed OxOs or standard OS is not possible.
   2. Perform concurrent detection of lipofuscin-like granules and other immunofluorescence markers
      NOTE: As detailed in step 2.3.1, the wide autofluorescence spectrum of lipofuscin limits choices for fluorescence co-staining. The following methods can be considered to facilitate co-staining.
      1. Utilize a far-red fluorophore. Autofluorescence from lipofuscin is weak in the near infrared. Thus, utilizing a far-red dye for fluorescence detection of an antigen of interest, combined with careful adjustments of channel settings on a confocal microscope, can usually distinguish autofluorescence from co-staining fluorescence.
      2. Take advantage of the long fluorescence emission tail of lipofuscin.
         NOTE: As lipofuscin has a very wide fluorescence emission spectrum, it can often be excited at a low nm wavelength (e.g. 405 nm laser) and have emission still detected in the orange/red portion of the spectrum (e.g. 585-635nm). This unique combination of excitation and emission wavelengths can often be tailored around another co-staining fluorophore.
         1. Detect the antigen of interest using a fluorophore that has peak excitation around 488 nm, with emission detection at 500-530 nm. Set up a separate channel for autofluorescence detection with 405 nm excitation and 585-635 nm emission. This second channel will detect only UAM, while the first channel will detect the antigen of interest plus lipofuscin.
         2. Using a free-ware program like ImageJ, utilize this second UAM-only channel as a subtractive mask to remove the UAM signal from the first channel (which contains both the antigen signal and the UAM signal).
      3. Utilize spectral unmixing. Most modern confocal microscopes contain a spectral unmixing option. This allows one to acquire the spectrum of a sample with just UAM and another sample with just the co-staining fluorophore of interest. The experimental sample, which contains both UAM and the co-staining fluorophore, can then be acquired and subjected to linear unmixing methods to compute what percent of the signal came from autofluorescence vs. co-staining.
         NOTE: Comprehensive guides to spectral unmixing are available with most modern confocal microscopes.
      4. Utilize fluorescence lifetime imaging. While the emission spectrum between UAM and a co-staining fluorophore of interest may be similar, it is likely that their fluorescence lifetimes differ significantly. In general, UAM demonstrates shorter fluorescence lifetimes than most specific fluorophores. With access to a confocal microscope that allows for lifetime imaging, one can gate fluorescence signals to detect those that are longer than the typical lipofuscin lifetime.
         NOTE: In general, gating fluorescence lifetime to signals longer than 2 ns greatly reduces UAM contamination, although it doesn't entirely eliminate the signal.
      5. Utilize of an autofluorescence suppressor. Traditionally, Sudan Black has been used to quench autofluorescence prior to specific immunofluorescence staining³⁵. Several commercially available autofluorescence quencher products report to improve on the outcomes of Sudan Black, and these products are detailed in the Table of Materials. Autofluorescence quenching, of course, will destroy the ability to detect lipofuscin.
   3. Determine the composition of lipofuscin-like granules
      1. Assess neutral lipids
         1. Fix UAM-laden RPE and control wells (fed untreated OS) in 4% PFA at room temperature for 15 min, and wash with PBS 5x.
         2. Stain for neutral lipids using either 10 µg/mL Nile Red or 3.33 µg/mL Bodipy 493/503 in a 3% BSA PBS solution at room temperature for 1 h, followed by washing with PBS for 5 min 3x.
         3. Cut out and mount Transwell as in step 2.3.1.2 and 2.3.1.3 and image. In general, use the following excitation and emission bandwidths for imaging: Nile Red - ex 543 nm, em 620-700 nm and Bodipy 493/503 - ex 488 nm, em 500-550 nm.
      2. Assess esterified and unesterified cholesterol
         1. Follow step 2.3.3.1.1
         2. Filipin is a fluorescent dye that recognizes unesterified cholesterol, but not esterified cholesterol³⁶. Thus, to assess for the amount of total cholesterol (unesterified and esterified) in the UAM, first pre-treat the sample with cholesterol esterase to convert esterified cholesterol into unesterified cholesterol. Treat the cells with 20 U/mL cholesterol esterase in 0.1 M potassium phosphate buffer (pH 7.2) (see Table of Materials) at 37 °C for 3.5 h, followed by washing with PBS for 5 min 3x.
         3. Stain with 50 µg/mL filipin (see Table of Materials) in PBS at room temperature for 1 h, and wash with PBS for 5 min 3x. Keep samples covered, away from light, as filipin photobleaches easily.
            NOTE: If only unesterified cholesterol is to be quantified, the cholesterol esterase in step 2.3.3.2.2 can be skipped. If the amount of esterified cholesterol is to be quantified, it can be deduced by the difference between the total cholesterol and the unesterified cholesterol in the sample.
         4. Cut out and mount Transwell as in steps 2.3.1.2 and 2.3.1.3 and image.
            ​NOTE: When imaging filipin, it photobleaches very quickly. One needs to minimize the intensity and duration of excitation, and expect that some photobleaching may even occur with viewing the sample through the oculars. Therefore, when imaging, one should: (1) search for an appropriate area to image using a fluorescence channel other than the filipin channel, (2) image filipin on a widefield rather than confocal microscope (to limit intensity of light exposure), and (3) avoid repeated imaging of an area. In general, use the following excitation and emission bandwidths for imaging filipin - ex 380 nm, em 480 nm.

3. Assessing effects of lipofuscin-like granules on RPE phagocytosis: Total Consumptive Capacity

NOTE: The rationale for measuring OS phagocytosis via the OS pulse only protocol below is detailed in the representative results section. The method, which is termed "Total Consumptive Capacity," avoids ambiguities about phagocytosis efficiency that can emerge with traditional OS pulse-chase phagocytosis assays. Assays are done on 24-well Transwell plates using 50 µL of media containing 4 x 10⁶ OS/mL.

1. Calculate the number of wells needed for the experiment and then thaw an appropriate amount of regular OS, spin down at 2400 x g for 5 min at room temperature, and resuspend in standard RPE cell culture media to 4 x 10⁶ OS/mL. Add bridging ligands to facilitate phagocytosis rates.
   NOTE: Final concentration of the bridging ligands in the media to be fed cells is 4 µg/mL for Protein S and 1.5 µg/mL for MFG-E8. As stated in step 1.1.11, concentrations of bridging ligands may need to be altered for other RPE types or cell densities.
2. Remove apical media and add 50 µL 4 x 10⁶ OS/mL, ideally with appropriate concentrations of bridging ligands.
3. At various times after OS addition (e.g. - 0 h, 1 h, 4 h, and 24 h), add 16.67 µL 4x Laemmli sample buffer with protease inhibitors to lyse both cells and the overlying OS-containing supernatant. Use a P-200 pipette to scratch the Transwell surface, being careful not to puncture the Transwell membrane or detach the membrane from the Transwell, and collect the combined cell supernatant plus cell lysate together. Vortex, spin down, and leave at room temperature for 30 min for thorough denaturation.
   CAUTION: Despite using Sample Buffer to lyse the OS, do not subsequently heat the lysis solution, as this can trigger rhodopsin aggregation. Also avoid cooling the lysed OS until ready for storage, as this will precipitate protein. Freeze unused lysates at -20 °C, making sure lysates are totally thawed before use.
4. Run lysates on SDS PAGE using the same settings as in step 1.3.3, loading equal volumes of lysates per well. GAPDH, β-actin, or another housekeeping protein should be used to normalize cell number, as phagocytosis rates are dependent on cell number.
5. Probe Western blots with antibodies against either the N- or C- terminus of rhodopsin. Use standard Western blotting conditions, and antibody dilutions are listed in the Table of Materials.
   NOTE: Below the main rhodopsin band, there will be multiple rhodopsin fragments. These fragments represent partly digested rhodopsin, a process that starts prior to fusion of the phagosome with the lysosome^32,33. An increase in the number of rhodopsin fragments in the UAM-laden RPE compared to control RPE could indicate a downstream defect in phagolysosome capacity, at the level of phagosome-lysosome fusion, lysosomal acidification, and/or degradative enzyme function^16,34,35.

Representative Results

The set-up for photo-oxidation of OS is demonstrated in Figure 1Ai. The polytetrafluoroethylene-coated slides allow for a large volume of OS in solution to be loaded per open rectangle without spreading across the rest of the slide. The slide with OS is contained within a sterile Petri dish with the lid off, and a UV lamp is placed over the slide as shown in Figure 1Aii. Alternatively, the slide can be placed in a UV Crosslinker device, as shown in Figure 1Aiii. After photo-oxidation, OS autofluorescence increases significantly, as assessed by both microscopy and flow cytometry (Figure 1B). The degree of protein cross-linking after photo-oxidation can be assessed by SDS-PAGE with a Coomassie stain, shown as a conversion of the dominant protein band (monomeric rhodopsin) into higher order aggregates and smear (Figure 1C). Comparing photo-oxidation with a hand-held UV lamp (Figure 1Aii) versus the UV Crosslinker device (Figure 1Aiii) demonstrates that the hand-held UV lamp produces OxOS with as much autofluorescence as a 3 J/cm² treatment from the UV Crosslinker device (Figure 1B), and as much protein cross-linking as a 6 J/cm² treatment from the UV Crosslinker device (Figure 1C). If the UV lamp available to a researcher differs from the one used in this protocol, we suggest titrating the exposure time to achieve a level of cross-linking seen in the UV lamp lane in Figure 1C. The autofluorescence spectrum of OxOS is mildly blue-shifted compared to the spectrum of untreated OS, in both the protein-isolated fraction of the OS and the lipid-isolated fraction of OS¹⁶.

The amount of UAM buildup in RPE cultures depends on the number of OxOS feedings (Figure 2A), and 20 feedings over the course of approximately 4 weeks provides a robust amount of UAM accumulation. To distinguish UAM autofluorescence from autofluorescence of residual OxOS that remain stuck to the RPE apical surface even days to weeks after wash-off, we stain with a rhodopsin antibody, detected with a far-red dye. By using this rhodopsin fluorescence channel as a mask, we can subtract out OxOS autofluorescence from the UAM channel, ensuring we are truly quantifying autofluorescence from UAM only¹⁶. Utilizing staining protocols outlined above, accumulated UAM in this model has abundant neutral lipids, as assessed by Nile Red staining¹⁶, similar to native lipofuscin. However, we find little to no esterified or unesterified cholesterol accumulation (Figure 2B), which is consistent with the lack of cholesterol accumulation we see in native lipofuscin from healthy eyes (data not shown).

Following fluorescent markers of interest concurrent with lipofuscin autofluorescence is difficult, given the very broad fluorescence emission spectrum of lipofuscin. In the methods section above, several methods are detailed to overcome this problem. The method outlined in step 2.3.2.1 above takes advantage of the low autofluorescence emission intensity for lipofuscin in the far-red. In Figure 2Ci, the colocalization of UAM and the autophagy marker LC3 is assessed by staining LC3 with an Alexa 647 dye. Despite some bleedthrough of UAM into the LC3 channel, it is apparent where lipofuscin and LC3 are located in these images. In the method outlined in step 2.3.2.2 above, the very long fluorescence emission spectra of lipofuscin allows design of an emission channel that contains fluorescence from lipofuscin only. In Figure 2Cii, UAM is excited with a 488 nm laser, while emission is detected at both 500-535 nm and at 600-645 nm. The sample is co-stained with LC3, detected with Alexa 488 dye. The Alexa 488 channel (excitation: 488 nm, emission: 500-535 nm) contains both LC3 and UAM signal. However, the second channel, with 488 nm excitation and 600-645 nm emission, contains only UAM. This second channel can be used as a mask, applied to the Alexa 488/LC3 channel, to subtract out unwanted UAM signal from the LC3 signal. In the method outlined in step 2.3.2.4 above, the shorter fluorescence lifetime of autofluorescent lipofuscin allows one to distinguish the lipofuscin from co-staining fluorescence. In Figure 2Ciii, all fluorescence signals arriving to a photon-counting detector before 2 ns are eliminated, which gates out much of the UAM autofluorescence signal while largely preserving the co-staining signal from LC3. A combination of methods may be necessary to distinguish lipofuscin from co-staining fluorescence if there is strong co-localization of the lipofuscin and co-staining fluorophore.

As multiple studies have postulated an impact of RPE lipofuscin accumulation on OS phagocytosis rates^{5,36,37,38,39}, OS phagocytosis capacity in UAM-laden cultures was assessed. Typical phagocytosis assays involve a brief incubation of cells with OS ("pulse") followed by wash-off of unbound OS and replacement of media without OS for a "chase" period. During the chase, as OS are degraded, there is a loss of rhodopsin, the most abundant protein in OS, within the RPE. At various timepoints during the chase, the OS-free media is removed, followed by cell lysis and SDS-PAGE for rhodopsin remaining within the RPE lysates. However, as the OS pulse period consists of both uptake and degradation of OS, RPE with high uptake and degradation ("phagocytosis efficient" cells) may have the same rhodopsin levels early in the chase period as RPE with low uptake and degradation ("phagocytosis inefficient" cells). This is schematically represented in the study by Zhang et al²³. To overcome this ambiguity, a "pulse-only" assay was employed here. In this method, OS are pulsed onto RPE but not washed off. At various times after OS addition, the media and the cell lysate are collected together. The media contains undigested OS and the cell lysate contains a combination of intact OS on the cell surface, partly digested OS that have been internalized, and fully digested OS that have successfully made it through the lysosome. As a control, cells are exposed to OS but then the cell lysate and OS-containing supernatant are immediately collected. This control reveals how much total rhodopsin was introduced to the cells. As the lysate plus supernatant is collected at various points after OS introduction, one can assess for what fraction of the total rhodopsin signal has disappeared, using this as a marker for amount of all introduced/starting OS that are now completely degraded. The readout for this assay is termed, "Total Consumptive Capacity", since it accurately measures all OS degradation and is not subject to the confounding interpretations of a pulse-chase experiment described above.

Figure 3A and Supplementary Figure 1 demonstrate a Western blot of remaining rhodopsin using the Total Consumptive Capacity assay. Amount of OS fed to cells is measured by the rhodopsin band at 0 h. The main rhodopsin band is degraded with equal efficiency between the control and UAM-laden RPE samples (single arrow at 4 h and 24 h). However, there is a difference between the groups when looking at partly degraded fragments of rhodopsin, which appear below the main rhodopsin band. The differences in fragment abundance imply reduced lysosomal capacity in the UAM group, since proteolytically cleaved fragments of rhodopsin should be rapidly degraded with normal lysosomal function. Increased abundance of these fragments without increased abundance of the main rhodopsin fragment suggests more subtle defects in lysosomal degradative capacity in the UAM-laden cultures. Prior studies have suggested that lipofuscin can indeed induce defects in OS phagocytosis⁴⁴, but no prior study has examined the effects of lipofuscin specifically on rhodopsin fragments, which allows a more precise pinpointing of the dysfunction to the end-stages of phagolysosomal degradation. Figure 3B demonstrates Western blotting of the main rhodopsin band plus rhodopsin fragments using two different anti-rhodopsin antibodies. Antibody 4D2 recognizes the N-terminus of rhodopsin, and N-terminal fragments are the last part of rhodopsin to be degraded in the lysosome. In contrast, antibody 1D4 recognizes the C-terminus of rhodopsin, which is degraded even prior to phagosome-lysosome fusion. Thus, understanding which fragments stain with which antibody provides a sense of rhodopsin processing defects that are pre-lysosomal vs. lysosomal^32,40,41.

Figure 1: Photo-oxidized outer segment (OxOS) treatment and characterization. (A) Depiction of OS on a polytetrafluoroethylene-coated slide (i) treated with either a 254 nm UV handheld lamp (ii) or a UV Crosslinker device (iii). (B) Compared to untreated (regular) OS (RegOS), treated OS (OxOS) had increased autofluorescence as shown by confocal imaging (left) (scale bar = 10 µm) and flow cytometry (right). Autofluorescent intensity of OxOS treated with the handheld UV lamp was similar to OS treated with the UV Crosslinker Device at a radiant exposure of 3 J/cm² (error bar = S.E.M.). (C) SDS-PAGE demonstrating the crosslinking of OS protein induced by UV exposure. OxOS crosslinking via the handheld UV lamp was similar to OS treated with the UV Crosslinker Device at a radiant exposure of 6 J/cm2. Monomeric rhodopsin is highlighted with an arrow. Please click here to view a larger version of this figure.

Figure 2: Imaging OxOS-induced lipofuscin-like granules in hfRPE cultures. (A) Autofluorescent granules (green) induced by OxOS accumulated significantly more after 20 feedings compared to 5 feedings. Residual OxOS stain with rhodopsin antibody (magenta). Scale bar = 10 µm. (B) UAM induced by OxOS (green) was not enriched with free cholesterol or cholesterol ester, as assessed by filipin staining (red). Scele bar = 10 µm. (C) Imaging methods for fluorescence co-staining in the presence of lipofuscin. (i) Use of far-red fluorophore (Alexa 647) to stain autophagy marker LC3 (magenta), which minimizes UAM bleedthrough. Pure UAM (green) imaged with a standard green channel excitation and emission filter set-up. Scale bar = 2 µm. (ii) Use of ratiometric imaging helps decipher UAM autofluorescence from LC3 staining labeled with Alexa 488. Excitation is 488 nm, green channel emission bandpass is 500-535 nm while red channel emission bandpass is 600-645 nm. Red channel can be applied as a subtractive mask to isolate LC3 signal from the green channel. Scale bar = 5 µm. (iii) As fluorescence lifetime of lipofuscin/UAM is typically shorter than specific dyes, UAM autofluorescence can be reduced by gating out fluorescence signals arriving at a photon-counting detector in less than 2 ns. Top image is without gating. Bottom image is with gating, keeping only fluorescence signals with a lifetime longer than 2 ns. There is a greater relative reduction in UAM signal compared to specific LC3 signal (labeled with Alexa 488) between top and bottom images. Scale bar = 5 µm. Please click here to view a larger version of this figure.

Figure 3: "Total Consumptive Capacity" method for measuring OS phagocytosis. (A) Total remaining rhodopsin protein in both cell lysate and supernatant after introduction of OS to RPE culture, as assayed by western blot. After feeding regular OS in 50 µL media, both the conditioned media/supernatant and cells were lysed together at 0, 4, and 24 h. The 0 h timepoint serves as a control, indicating the total amount of OS fed to each Transwell. The intact rhodopsin band (single arrow) was not different between control and UAM-laden RPE cells, suggesting no large effect of UAM on RPE phagocytosis. However, the cleavage products of rhodopsin, indicated by the double arrows, were higher in the UAM group, suggesting some mild degradative dysfunction in the phagolysosomal system. Equal levels of GAPDH indicate that cell count between wells, which can affect phagocytosis rates, was equal. (B) Different rhodopsin antibodies recognize different degradation fragments. 4D2 recognizes the N-terminus of rhodopsin, which is intact until the very last steps of lysosomal degradation. The fragments it recognizes are therefore small (double arrows). In contrast, 1D4 recognizes the C-terminus of rhodopsin, which is degraded earlier in the phagolysosomal process. This antibody therefore recognizes rhodopsin fragments of a higher molecular weight (double arrows). Both antibodies recognize intact rhodopsin (single arrow). Please click here to view a larger version of this figure.

Supplementary Figure 1: Unedited blot corresponding to Figure 3A. Please click here to download this File.

Discussion

While RPE lipofuscin has been studied for decades, its toxicity is debated^2,9,16,42. Given ambiguity about the toxicity of lipofuscin from animal models¹¹, in vitro models using human RPE are valuable. A range of in vitro lipofuscin accumulation models have been described, but none have utilized both OS feeding and highly mature and differentiated human RPE cultures. This combination represents an ideal model, as OS feeding recapitulates the method for lipofuscin accumulation in vivo and highly mature human RPE cultures best replicate RPE behavior in vivo. Interestingly, when utilizing highly differentiated hfRPE cultures, it was discovered that routine OS exposure was insufficient to induce lipofuscin-like material, even after repeated feedings¹⁶. Thus, this protocol accelerates the lipofuscin accumulation process by photo-oxidizing OS (OxOS) in a controlled fashion. Feeding OxOS to both human iPSC-RPE and hfRPE cultures resulted in robust accumulation of lipofuscin-like granules, which is termed undigestible autofluorescent material (UAM). UAM accumulation allows for modeling of lipofuscin in culture systems that mimic healthy human RPE in vivo^{22,23,29,43,44}. In both a prior publication and in this study, extensive characterization of UAM compared to native lipofuscin from healthy older adults demonstrates both similarities and differences. UAM mimics the ultrastructure of lipofuscin in vivo, including the tendency for lipofuscin granules to fuse with melanin granules to form melanolipofuscin¹⁶. Both UAM and lipofuscin contain substantial neutral lipids, no rhodopsin, and no significant enrichment in cholesterol (Figure 2A,B from our prior publication¹⁶, Figure 2A,B above, and unpublished data). We also compared native lipofuscin granule size and spectrum to those from UAM (Figure 3 from our prior publication¹⁶). UAM granules are initially slightly larger and blue-shifted compared to native lipofuscin, but over significant periods of time in culture, the UAM compact to a smaller size and red-shift in their emission spectrum, becoming much more similar to the size and spectra profile of native lipofuscin.

The OxOS UAM model has been utilized for a range of applications, including the effects of autophagy inducers on preventing and removing lipofuscin granules⁵⁰ and the effects of lipofuscin on RPE polarity and metabolism¹⁶. Further, these granules have been shown to persist in cultured RPE for more than a year, allowing one to study the evolution of lipofuscin spectra, morphology, and behavior over prolonged periods of time¹⁶. Other applications for the model include the study of lipofuscin accumulation on complement activation⁵¹, lysosomal stability and inflammatory responses⁵², and mitochondrial compromise⁵³.

There are a few critical steps to this protocol. First, RPE cultures have to be highly differentiated and mature. OxOS feedings should only start after RPE has been on Transwells for at least 8 weeks and has obtained all the qualities characteristic of highly mature RPE (the 5 'P's elaborated by Finnemann et al. - polygonal in morphology, postmitotic, pigmented, polarized, and phagocytic⁵⁴). Second, OS are highly fragile and should be handled with care during pipetting. Excessive pipetting or freeze-thaws will break apart the OS. Lastly, the ratio of OS to total UV flux exposure is critical for generating OxOS that remain intact but can still induce UAM; this ratio has been refined in this protocol. Too much UV light for a given number of OS causes excessive cross-linking and destroys critical chemical structures in the OS. Too little UV light for a given number of OS will fail to induce UAM in culture.

Modifications to the protocol largely involve altering OxOS feeding duration or concentration. Empirically, 20 feedings over the course of approximately 4 weeks produces a robust amount of UAM accumulation. Feedings beyond this may incrementally add to UAM accumulation, while fewer feedings are likely to cause only sporadic UAM accumulation. The number of feedings can be adjusted based on the purpose, needs, and resources of the experiment and experimenter. In less differentiated RPE cultures (higher passage number, cell lines, or cultures demonstrating epithelial-mesenchymal transition properties), fewer feedings and lower OxOS concentrations are needed for robust UAM induction.

The protocol has several limitations. Utilizing a handheld UV lamp for photo-oxidation of OS prevents precise quantification of the total radiant exposure delivered to OS. However, photo-oxidation of OS using the hand-held lamp was correlated in this study to a much more expensive (but quantitative) UV Crosslinker Device in Figure 1 to help provide parameters for radiant exposure. It is recommended that experimenters alter the duration of UV exposure until the cross-linking/rhodopsin smearing pattern mimics that seen in the UV lamp column of the Western blot in Figure 1C.

A second limitation of the method is that it likely lacks many of the features of lipofuscin accumulation in vivo. Indeed, throughout this protocol, the lipofuscin-like granules are referred to as undigestible autofluroescent material (UAM) to make this distinction clear. Photo-oxidizing OS recapitulates some of the natural chemical cross-linking that happens in photoreceptor outer segments in disease states, but the cross-linking is greatly accelerated and likely more non-specific in the UAM model. Further, despite nearly a month of OxOS feedings, the UAM model still represents an "acute" exposure to lipofuscin-inducing OS, compared to the decades it takes for lipofuscin to accumulate in humans. With a more prolonged lipofuscin accumulation in culture, there may be fewer phenotypic effects. Nevertheless, because UAM granules persist for more than a year in the culture system presented here, there is opportunity to study their long-term effects on RPE health¹⁶.

A limitation of the "Total Consumptive Capacity" method presented here to assess for RPE phagocytosis capacity is that it fails to distinguish whether phagocytosis defects may be due to OS binding vs. uptake vs. degradation. Indeed, when the goal of the phagocytosis assay is mechanistic insight into dysfunction of specific steps of the phagocytosis process, traditional OS pulse-chase assays are more appropriate. Measurement of "Total Consumptive Capacity" should be employed when the goal is simply to unambiguously measure the OS phagocytosis competence of the RPE culture under control versus experimental conditions.

In conclusion, a protocol is presented for lipofuscin-like granule accumulation in highly differentiated human RPE cultures. This method merges the physiologic process for lipofuscin accumulation - repeated feedings of photo-oxidized OS - with RPE culture models that have been shown in repeated studies to strongly recapitulate human RPE in vivo. The resulting cultures laden with lipofuscin-like granules can be used for myriad applications, ranging from assessment of lipofuscin effects on RPE biology to tests of small molecules, genes, and signaling pathways important for modulating lipofuscin accumulation.

Materials

Name	Company	Catalog Number	Comments
100 mm cell culture dish	Corning	#353003	Others also work
24-well Transwells	Corning	#3470
Anti-LC3 antibody	Cell Signaling Technology	#4801S	1:1000 dilution
Anti-rhodopsin antibody 1D4	Abcam	#5417	1:1000 dilution. Epitope is C-terminal.
Anti-rhodopsin antibody 4D2	EnCor Biotech	MCA-B630	1:5000 dilution for western blot, 1:1000 dilution for immunostaining. Epitope is N-terminal.
Autofluorescence quencher	Biotium	#23007	TrueBlack Lipofuscin Autofluorescence Quencher
Autofluorescence quencher	Vector Laboratories	SP-8400	Vector TrueVIEW Autofluorescence Quenching Kit
Bodipy 493/503	Life Technologies	D3922
Cholesterol esterase	Life Technologies	From A12216 kit
Confocal microscope	Leica	Leica Stellaris SP8 with FALCON module
Dark-adapted bovine retinas	W. L. Lawson Company	Dark-adapted bovine retinas (pre-dissected)	Contact information: https://wllawsoncompany.com/ (402) 499-3161 stacy@wllawsoncompany.com
Filipin	Sigma-Aldrich	F4767
Flow cytometer	Thermo Fisher	Attune NxT
Flow cytometer analysis software	BD	FlowJo
Handheld UV light	Analytik Jena US	UVGL-55
Human MFG-E8	Sino Biological	10853-H08B
Human purified Protein S	Enzyme Research Laboratories	HPS
Laemmli sample buffer	Thermo Fisher	J60015-AD
LDH assay	Promega	J2380	LDH-Glo Cytotoxicity Assay
Mounting media	Invitrogen	P36930	Prolong Gold antifade reagent
Nile red	Sigma-Aldrich	#72485
Polytetrafluoroethylene-coated slides	Tekdon	Customized	Customized specifications: PTFE mask with the following "cut-outs" -  3 glass rectangles, each measuring 17 mm x 9 mm, oriented so that the 17 mm side is 4 mm from the top of the slide and 4 mm from the bottom of the slide, assuming a standard microscope slide of 25 mm x 75 mm. Each rectangle is spaced at least 6 mm away from other rectangles and the edges of the slide. Print PTFE mask on a slide with frosted glass on one side to allow for labeling of the slide.
Protease inhibitors	Cell Signaling Technology	#5872
Protein assay	Bio-Rad	#5000122 RC DC protein assay
TEER electrode	World Precision Instruments	STX3
Trans-epithelial electrical resistance (TEER) meter	World Precision Instruments	EVOM3
Ultraviolet crosslinker device	Analytik Jena US	UVP CL-1000