Lipofuscin Granule Bisretinoid Oxidation in the Human Retinal Pigment Epithelium forms Cytotoxic Carbonyls

Age-related macular degeneration (AMD) is the primary cause of central blindness among the elderly. AMD is associated with progressive accumulation of lipofuscin granules in retinal pigment epithelium (RPE) cells. Lipofuscin contains bisretinoid fluorophores, which are photosensitizers and are phototoxic to RPE and neuroretinal cells. In the presence of oxygen, bisretinoids are also oxidized, forming various products, consisting primarily of aldehydes and ketones, which are also potentially cytotoxic. In a prior study, we identified that in AMD, bisretinoid oxidation products are increased in RPE lipofuscin granules. The purpose of the present study was to determine if these products were toxic to cellular structures. The physicochemical characteristics of bisretinoid oxidation products in lipofuscin, which were obtained from healthy donor eyes, were studied. Raman spectroscopy and time-of-flight secondary ion mass spectrometry (ToF–SIMS) analysis identified the presence of free-state aldehydes and ketones within the lipofuscin granules. Together, fluorescence spectroscopy, high-performance liquid chromatography, and mass spectrometry revealed that bisretinoid oxidation products have both hydrophilic and amphiphilic properties, allowing their diffusion through lipofuscin granule membrane into the RPE cell cytoplasm. These products contain cytotoxic carbonyls, which can modify cellular proteins and lipids. Therefore, bisretinoid oxidation products are a likely aggravating factor in the pathogenesis of AMD.


Introduction
Age-related macular degeneration (AMD) is a chronic eye disease characterized by damage to the macular region and progressive central vision loss [1]. AMD is a complex disease with multiple risk factors and disease pathologies [2]. Oxidative stress is central to the development of AMD [2][3][4][5]. It is characterized by increased levels of reactive oxygen species (ROS) resulting in damage or modification of cellular proteins, lipids, and DNA, impairing their physiological functions [5]. Lipofuscin granules (LGs) are one of the sources of ROS in retinal pigment epithelium (RPE) cells. Light exposure induces ROS formation in LGs, initiating oxidative stress in RPE cells [6,7]. Therefore, development of AMD is associated with progressive accumulation of LGs in the RPE [8].
LGs are formed by incomplete lyso-

Effect of Light and Superoxide Radicals on the Fluorescent Properties of Lipofuscin Granules (LGs)
The composition of LG fluorophores from human RPE cells includes approximately 20 BisRets and other all-trans retinal derivatives, which undergo photo-oxidative destruction under light exposure [31,43,44]. BisRet-OXs with increased hydrophilicity were previously thought to leave LGs and enter RPE cell cytoplasm [34], directly damaging cell structures [45].
In the present study, BisRet-OX formation was evaluated by measuring the change in the fluorescence spectra of samples before and after irradiation with visible light, as well as after exposure to superoxide radicals. During BisRet oxidation, the absorption and fluorescence spectra of fluorophores shifted to the short-wave region [31,46]. We used buffer solution with LGs (subsequently referred to as LG suspension), the sediment from the LG suspension (subsequently referred to as LG sediment), and the supernatant (subsequently referred to as supernatant).
Three wavelengths of 340, 365, and 488 nm were used to excite fluorescence. The 488 nm wavelength is traditionally used clinically to assess fundus autofluorescence [47]. Two other wavelengths (340 and 365 nm) were used for a more complete study of the spectral properties of LG fluorophores, most of which absorb in the short-wave region, especially the oxidation products.
Irradiation of the LG suspension with visible light or exposure to superoxide radicals caused increased fluorescence intensity in the short-wave region of 450-550 nm, which indicated increased BisRet-OX contents [31,46] The spectral changes under different LG fluorophore oxidation conditions varied (Figure 1). fluorescence spectra of fluorophores shifted to the short-wave region [31,46]. We used buffer solution with LGs (subsequently referred to as LG suspension), the sediment from the LG suspension (subsequently referred to as LG sediment), and the supernatant (subsequently referred to as supernatant). Three wavelengths of 340, 365, and 488 nm were used to excite fluorescence. The 488 nm wavelength is traditionally used clinically to assess fundus autofluorescence [47]. Two other wavelengths (340 and 365 nm) were used for a more complete study of the spectral properties of LG fluorophores, most of which absorb in the short-wave region, especially the oxidation products.
Irradiation of the LG suspension with visible light or exposure to superoxide radicals caused increased fluorescence intensity in the short-wave region of 450-550 nm, which indicated increased BisRet-OX contents [31,46] The spectral changes under different LG fluorophore oxidation conditions varied ( Figure 1). LG suspension before light irradiation or incubation with superoxide radicals; sediment and supernatant obtained from this suspension. 2-LG suspension after 60 min visible light irradiation; sediment and supernatant obtained from this suspension. 3-LG suspension after 60 min incubation with potassium superoxide; sediment and supernatant obtained from this suspension. The wavelengths of fluorescence excitation were 365 nm (left) and 488 nm (right).
Importantly, in the supernatant obtained from the original LG suspension, the fluorescence intensity was very low, indicating a low content of hydrophilic oxidized forms LG suspension before light irradiation or incubation with superoxide radicals; sediment and supernatant obtained from this suspension. 2-LG suspension after 60 min visible light irradiation; sediment and supernatant obtained from this suspension. 3-LG suspension after 60 min incubation with potassium superoxide; sediment and supernatant obtained from this suspension. The wavelengths of fluorescence excitation were 365 nm (left) and 488 nm (right). Importantly, in the supernatant obtained from the original LG suspension, the fluorescence intensity was very low, indicating a low content of hydrophilic oxidized forms of LG fluorophores ( Figure 1C, spectra 1 at fluorescence excitations of 365 and 488 nm). However, after oxidation of the LG suspension, the fluorescence intensity of the super-natants increased sharply in the short-wavelength region of the spectrum (450 nm, Ex. 365 nm; 550 nm, Ex. 488 nm). This was particularly apparent in LG suspension exposed to superoxide radicals ( Figure 1C, spectra 3, fluorescence excitations 365 and 488 nm).
This suggested that when the LG suspension was irradiated with visible light or exposed to superoxide radicals, BisRet-OXs were formed, some of which had hydrophilic properties and could potentially diffuse from LGs into RPE cell cytoplasm.
Not only hydrophilic compounds, but also substances with amphiphilic properties, can pass into the aqueous phase. Therefore, to study the nature of fluorophores diffusing into aqueous medium in more detail, chloroform was added to the supernatants ( Figure 1C). Figure 2 shows the fluorescence spectra of chloroform extracts from the supernatants (A) and the aqueous fractions (B) of the same samples. Both the water and chloroform fractions had fluorescent properties, suggesting that supernatants from oxidized LG suspensions contained both hydrophilic and amphiphilic fluorophore oxidation products. of LG fluorophores ( Figure 1C, spectra 1 at fluorescence excitations of 365 and 488 nm). However, after oxidation of the LG suspension, the fluorescence intensity of the supernatants increased sharply in the short-wavelength region of the spectrum (450 nm, Ex. 365 nm; 550 nm, Ex. 488 nm). This was particularly apparent in LG suspension exposed to superoxide radicals ( Figure 1C, spectra 3, fluorescence excitations 365 and 488 nm). This suggested that when the LG suspension was irradiated with visible light or exposed to superoxide radicals, BisRet-OXs were formed, some of which had hydrophilic properties and could potentially diffuse from LGs into RPE cell cytoplasm.
Not only hydrophilic compounds, but also substances with amphiphilic properties, can pass into the aqueous phase. Therefore, to study the nature of fluorophores diffusing into aqueous medium in more detail, chloroform was added to the supernatants ( Figure  1C). Figure 2 shows the fluorescence spectra of chloroform extracts from the supernatants (A) and the aqueous fractions (B) of the same samples. Both the water and chloroform fractions had fluorescent properties, suggesting that supernatants from oxidized LG suspensions contained both hydrophilic and amphiphilic fluorophore oxidation products.

High-Performance Liquid Chromatography (HPLC) and Mass Spectrometry Analyses of Lipofuscin Granule BisRets and BisRet-OX
To determine the nature of supernatant compounds, we performed a comparative HPLC analysis (Figure 3) of the chloroform fractions from all samples presented in Figure  1: LG suspensions (A), LG sediments (B), and supernatants (C).
HPLC analysis of samples obtained from original non-oxidized LG suspensions revealed that almost all detectable products were present in the first group of peaks (gr1) on the chromatogram pass from LGs into the supernatant ( Figure 3C, original sample). We postulate that these products could correspond to BisRet-OX [31,33]. Moreover, in the supernatants, products belonging to the second group of peaks (gr2) were present, corresponding to BisRets and their oxidized forms, identified in prior studies [32,41,42,48].

High-Performance Liquid Chromatography (HPLC) and Mass Spectrometry Analyses of Lipofuscin Granule BisRets and BisRet-OX
To determine the nature of supernatant compounds, we performed a comparative HPLC analysis (Figure 3) of the chloroform fractions from all samples presented in Figure 1: LG suspensions (A), LG sediments (B), and supernatants (C).
Importantly, the products formed after exposure of LG suspensions to photo-oxidation or superoxide absorbed primarily in the shorter wavelength range of the spectrum (365 nm), and passed into the supernatant (Figure 3C, sample exposed to visible light, and superoxide-oxidized sample). A2E, its iso-form, and ox-A2E were detected in trace amounts at both 430 and 365 nm in chloroform extracts from the supernatants of all samples ( Figure 3C). This indicates that A2E and its slightly oxidized forms (ox-A2E) [33] practically do not diffuse from LGs into the aqueous medium of the supernatants. Mass spectrometry analysis confirmed the HPLC data ( Figure 4). A2E (m/z = 592) and its slightly oxidized forms (singly oxidized A2E m/z = 608; doubly oxidized A2E m/z = 624) [33] ( Figure 4A) were nearly absent in supernatants ( Figure 4B and C). This suggested that A2E and its minimally oxidized forms did not significantly diffuse through the LG membranes into the RPE cell cytoplasm.
Importantly, supernatant BisRet-Ox content increased after irradiation of the LG suspension ( Figure 4C). For example, the amounts of both newly formed BisRet-OX and the contents of existing BisRet-OX in the sample increased after irradiation ( Figure 4C). These BisRet-OX products had lower masses than A2E (m/z = 422, 432, 438, 448, 464, 470, 488, HPLC analysis of samples obtained from original non-oxidized LG suspensions revealed that almost all detectable products were present in the first group of peaks (gr1) on the chromatogram pass from LGs into the supernatant ( Figure 3C, original sample). We postulate that these products could correspond to BisRet-OX [31,33]. Moreover, in the supernatants, products belonging to the second group of peaks (gr2) were present, corresponding to BisRets and their oxidized forms, identified in prior studies [32,41,42,48].
Importantly, the products formed after exposure of LG suspensions to photo-oxidation or superoxide absorbed primarily in the shorter wavelength range of the spectrum (365 nm), and passed into the supernatant (Figure 3C, sample exposed to visible light, and superoxideoxidized sample). A2E, its iso-form, and ox-A2E were detected in trace amounts at both 430 and 365 nm in chloroform extracts from the supernatants of all samples ( Figure 3C). This indicates that A2E and its slightly oxidized forms (ox-A2E) [33] practically do not diffuse from LGs into the aqueous medium of the supernatants.
Mass spectrometry analysis confirmed the HPLC data ( Figure 4). A2E (m/z = 592) and its slightly oxidized forms (singly oxidized A2E m/z = 608; doubly oxidized A2E m/z = 624) [33] ( Figure 4A) were nearly absent in supernatants ( Figure 4B,C). This suggested that A2E and its minimally oxidized forms did not significantly diffuse through the LG membranes into the RPE cell cytoplasm. and 504) [38,39]. This suggested that both the oxidation and destruction of bisretinoids occurred during irradiation. The HPLC data thus revealed that most of the products in the chloroform extracts of LG suspensions detected at 365 nm were BisRet-OX, which have amphiphilic properties and can potentially diffuse through the LG membrane into the RPE cell cytoplasm.

Analysis of Thiobarbituric Acid (TBA)-Active Product Content in Lipofuscin Granule (LG) Suspensions and Supernatants
Fluorescence spectroscopy and HPLC analyses suggested that hydrophilic and amphiphilic BisRet-OX could contain cytotoxic active carbonyl compounds, especially aldehydes and ketones. This assumption is also supported by studies that identified glyoxal formation during A2E oxidation [45].
We applied Raman spectroscopy analysis (BCARS) of LG suspensions before and after visible light irradiation to determine if free-state aldehydes and ketones were present in the LGs. Raman spectra identified bands of aldehydes and ketones such as ~1415, ~1465, ~1685, and ~1730 1/cm [49] ( Figure 5A). After light irradiation, these bands were clearly present, including the ratio of C=O bands (1685 and 1730 1/cm) relative to aromatic C=C bands (~1600 1/cm, which we propose should not essentially change) [50], which has grown by about 1.5-2 times. Thus, exposure to light increased the contents of free-state aldehydes and ketones.
The formation of oxygen-containing products, such as aldehydes and ketones, was also assessed by time-of-flight secondary ion mass spectrometry (ToF-SIMS) analysis of characteristic fragment ions containing carbonyl groups (m/z = 29: CHO+ ion; m/z = 43: C2H3O+ ion; m/z = 60: C2H4O2+ ion; m/z = 69: C4H5O+ ion). These ions accumulated during light irradiation ( Figure 5B). This was consistent with prior findings [29,40]. Therefore, the BCARS and ToF-SIMS data revealed that aldehydes and ketones accumulated in LGs in the free state. Importantly, supernatant BisRet-Ox content increased after irradiation of the LG suspension ( Figure 4C). For example, the amounts of both newly formed BisRet-OX and the contents of existing BisRet-OX in the sample increased after irradiation ( Figure 4C). These BisRet-OX products had lower masses than A2E (m/z = 422, 432, 438, 448, 464, 470, 488, and 504) [38,39]. This suggested that both the oxidation and destruction of bisretinoids occurred during irradiation.
The HPLC data thus revealed that most of the products in the chloroform extracts of LG suspensions detected at 365 nm were BisRet-OX, which have amphiphilic properties and can potentially diffuse through the LG membrane into the RPE cell cytoplasm.

Analysis of Thiobarbituric Acid (TBA)-Active Product Content in Lipofuscin Granule (LG) Suspensions and Supernatants
Fluorescence spectroscopy and HPLC analyses suggested that hydrophilic and amphiphilic BisRet-OX could contain cytotoxic active carbonyl compounds, especially aldehydes and ketones. This assumption is also supported by studies that identified glyoxal formation during A2E oxidation [45].
We applied Raman spectroscopy analysis (BCARS) of LG suspensions before and after visible light irradiation to determine if free-state aldehydes and ketones were present in the LGs. Raman spectra identified bands of aldehydes and ketones such as~1415,~1465,~1685, and~1730 1/cm [49] ( Figure 5A). After light irradiation, these bands were clearly present, including the ratio of C=O bands (1685 and 1730 1/cm) relative to aromatic C=C bands (~1600 1/cm, which we propose should not essentially change) [50], which has grown by about 1.5-2 times. Thus, exposure to light increased the contents of free-state aldehydes and ketones.
We used TBA-active product (reactive carbonyls) registration in the supernatants to determine the ability of aldehydes and ketones to diffuse through the LG membrane into the RPE cell cytoplasm ( Figure 6). Exposure of LG BisRets to visible light or superoxide radicals resulted in significant accumulation of TBA-active products in the supernatants ( Figure 6A).   The formation of oxygen-containing products, such as aldehydes and ketones, was also assessed by time-of-flight secondary ion mass spectrometry (ToF-SIMS) analysis of characteristic fragment ions containing carbonyl groups (m/z = 29: CHO+ ion; m/z = 43: C 2 H 3 O+ ion; m/z = 60: C 2 H 4 O 2 + ion; m/z = 69: C 4 H 5 O+ ion). These ions accumulated during light irradiation ( Figure 5B). This was consistent with prior findings [29,40]. Therefore, the BCARS and ToF-SIMS data revealed that aldehydes and ketones accumulated in LGs in the free state.
We used TBA-active product (reactive carbonyls) registration in the supernatants to determine the ability of aldehydes and ketones to diffuse through the LG membrane into the RPE cell cytoplasm ( Figure 6). Exposure of LG BisRets to visible light or superoxide radicals resulted in significant accumulation of TBA-active products in the supernatants ( Figure 6A).
Interestingly, the fluorescence intensities of the supernatants after LG suspension irradiation with visible light or oxidation by superoxide radicals (Figure 6B, spectra 2 and 3) did not correlate with the content of TBA-active products in these samples ( Figure 6A). This suggested that the composition of the oxidized products was not identical in the supernatants from the LG suspensions exposed to visible light irradiation or superoxide. This potentially suggested that not all fluorescent compounds were TBA-active products ( Figure 6B).
In addition, a comparative analysis of the hydrophilic and amphiphilic TBA-active product contents contained in the supernatant from the LG suspension irradiated with visible light was conducted. Aqueous and chloroform fractions of the supernatant were separately prepared, and the content of TBA-active products was measured. The TBAactive product contents were nearly equal in both the water-soluble and chloroform fractions of the supernatant ( Figure 6C, columns 2 and 3, respectively). This suggests that the TBA-active products of photo-oxidized Bis-Ret fluorophores were likely a mix-ture consisting of fully oxidized, water-soluble substances and partially oxidized, lipid affinity-preserving substances. Figure 5. (A) Averaged Raman spectra for lipofuscin granule (LG) suspensions before (solid line) and after (dotted line) visible light irradiation for 100 min. Three independent experiments were conducted. In each experiment, 25 spectra were obtained (p < 0.05). (B) Time-of-flight secondary ion mass spectrometry (ToF-SIMS) analysis of LG suspensions before irradiation (1) and after visible light irradiation for 2 (2), 10 (3), 40 (4), 100 (5), and 160 (6) min. On the abscissa axis, the numbers correspond to the mass of positive fragment ions, containing carbonyl groups (29: CHO+; 43: C2H3O+; 60: C2H4O2+; 69: C4H5O+). On the ordinate axis, the relative intensities of the corresponding positive fragment ions are plotted as relative units. Data are presented as means ± SD from nine independent experiments. * р < 0.01.

BisRet-OX-Induced Protein and Lipid Modification Products
The ability of supernatant products obtained from non-irradiated and irradiated LG suspensions to induce protein and lipid modifications in the absence of light was investigated. Photoreceptor outer segments (POSs) from bovine retinas were used as a protein and lipid substrate. The proposed modification occurs via the interaction of carbonyl products present in the supernatants with the free amino groups of proteins and lipids, especially after irradiation of the LG suspension. Fluorescent Schiff bases with emission maxima at 440-455 nm are formed through this interaction. Importantly, the LG bisretinoid fluorophores themselves could not contribute to fluorescence in this experiment, as they were separated from the modified proteins and lipids by dialysis (see Materials and Methods, Section 4.11). Figure 7 demonstrates that incubation of the POS suspension with water-soluble fractions obtained from the LG suspensions significantly increased fluorescence intensity.
A noticeable increase in fluorescence was observed in the POS suspension incubated with the supernatant from the LG suspension irradiated with visible light (Figure 7A, 3). However, in the presence of the glycation inhibitor, aminoguanidine, the water-soluble fraction of the irradiated LG suspension did not significantly increase in intensity of POS fluorescence ( Figure 7A, 4), and in fact had a similar effect of the water-soluble fraction of the non-irradiated LG suspension. Aminoguanidine reacts with dicarbonyl compounds, such as methylglyoxal and glyoxal, to form 3-amino-1,2,4-triazine derivatives and prevent the glycation process by these agents [51].
The dynamics of the increase in the POS fluorescence intensity for all experiments are shown in Figure 7B. The concentration of fluorescent products in POS samples incubated with the water-soluble fractions from irradiated LG suspensions (column 3) increased with incubation time. Importantly, in POS samples containing supernatant from a non-irradiated LG suspension, an increase in fluorescence intensity was also observed. This is because the unirradiated supernatant also contained carbonyl compounds, but in lower concentrations (Figures 5 and 6). In addition, during prolonged incubation of POS suspensions, which contain abundant polyunsaturated fatty acid residues [52], at 37 • C, these fatty acids would be expected to undergo auto-oxidation resulting in formation of carbonyl compounds, which also induce modification of POS proteins and lipids.
gated. Photoreceptor outer segments (POSs) from bovine retinas were used as a protein and lipid substrate. The proposed modification occurs via the interaction of carbonyl products present in the supernatants with the free amino groups of proteins and lipids, especially after irradiation of the LG suspension. Fluorescent Schiff bases with emission maxima at 440-455 nm are formed through this interaction. Importantly, the LG bisretinoid fluorophores themselves could not contribute to fluorescence in this experiment, as they were separated from the modified proteins and lipids by dialysis (see Materials and Methods, Section 4.11). Figure 7 demonstrates that incubation of the POS suspension with water-soluble fractions obtained from the LG suspensions significantly increased fluorescence intensity.

Discussion
In the present study, the features of LG BisRet-OX from RPE cells were investigated. BisRet-OXs had both hydrophilic and amphiphilic properties, allowing their diffusion from the LG into the RPE cell cytoplasm. These products can be formed not only during BisRet photo-oxidation, but also during the oxidative destruction of BisRets in the dark via non-lipofuscin ROS.
The experimental data indicate that the hydrophilic and amphiphilic BisRet-OX compounds are TBA-active. Reactive carbonyls are highly cytotoxic, and can cause carbonyl stress [53,54]. It is important to note that these carbonyl products contain water-soluble amphiphilic compounds that can damage not only the water-soluble proteins of the cytoplasm, but also various membrane structures of the cell. Since carbonyl products are long-lived, they can bind tightly with long-lived proteins, such as collagen [55] or hemoglobin [56], under physiological conditions, resulting in the formation of advanced glycation end products. Protein glycation caused by these substances can activate inflammatory processes, and thus is of great importance in the pathogenesis of many eye diseases, including AMD, in which LGs accumulate in RPE cells.
Therefore, despite the fact that the BisRet-OX species absorb in the short-wavelength region of the spectrum (UV) and cannot be photoactive in vivo, they are nevertheless toxic to RPE cells. Almost all detectable BisRet-OX species have the potential to exit LGs and enter the RPE cell cytoplasm and can thus damage cellular organelles and macromolecules. Therefore, formation of BisRet-OX-containing active carbonyls, which accumulate in the LGs of RPE cells, may play an important role in the pathogenesis of eye disease, including AMD.

Tissues and Reagents
Experiments on human cadaver eye tissues were performed in compliance with officially accepted procedures, specifically the Russian Federation law N 4180-I dated 22 December 1992, "On human organs or tissue transplantation" (with the most recent modifications and additions dated 8 December 2020). Human cadaver eyes without ophthalmologic disease were obtained from the Eye Tissue Bank of the S.N. Fyodorov Eye Microsurgery Complex (Moscow, Russia) within 10 h of donor death. Donor age ranged from 50-75 years. Collection was conducted in accordance with local ethics requirements, as described in detail previously [31]. Permission was obtained from the chief medical officer of the S.N. Fyodorov Eye Microsurgery Complex, under a scientific collaboration agreement between the Complex and the Emanuel Institute of Biochemical Physics dated 11 January, 2011 to perform scientific research in the Laboratory of Physical and Chemical Bases of Vision at the Institute with RPE from cadaver eyes.
Fresh bovine eyes (Bos taurus) were purchased from a meat processing factory.

Isolation of Lipofuscin Granules (LGs)
LGs were isolated from the RPE of 40 cadaver eyes as described previously [6]. The initial LG concentration was 5 × 10 8 granules/mL. Water-soluble fractions from LG suspension were obtained as follows. For experiments, LGs were resuspended in 0.1 M potassium phosphate buffer (pH 7.3) to the concentration of 2.0 × 10 8 granules/mL. Then suspension was divided into two samples of equal volume. The experimental sample was irradiated for 30-160 min (depending on the experiment) with visible light (400-700 nm) of a KGM 24-150 lamp (80 mW/cm 2 ) at room temperature and constant stirring. The control sample remained all this time in complete darkness. Thereafter, the samples were centrifuged at 15,000× g in a Beckman Allegra 64R centrifuge for 15 min, and the obtained supernatants were used in the experiments. The sediments containing LGs were homogenized in 1 mL 0.1 M potassium phosphate buffer.
All stages of sample preparation were conducted under subdued lighting.

Preparation of Photoreceptor Outer Segments (POSs) Containing Rhodopsin
POSs from bovine retina tissue, which was isolated within 3 h after slaughter, were prepared as described previously [57]. Retinas were added to 50% sucrose solution in Buffer A comprising 10 mM MOPS pH 7.4, 30 mM NaCl, 60 mM KCl, 2 mM MgCl 2 , 0.1 mM phenylmethylsulfonyl fluoride (PMSF), 1 mM dithiothreitol, and 0.01% NaN 3 (1 mL per each retina). The suspension was vigorously shaken for 3 min and centrifuged at 2000× g for 40 min at 4 • C. The supernatant was diluted 2.5-fold with Buffer A and centrifuged at 2000× g for 60 min at 4 • C. The pellet was resuspended in 25 mL of 40% sucrose in Buffer A, and 10 mL of Buffer A was layered on top following centrifugation at 25,000× g for 60 min at 4 • C on a Beckman Coulter Avanti J30I centrifuge equipped with a JS 24.38 rotor. The POS fraction was sampled at the buffer-sucrose interface and diluted with Buffer A to a density of 1.05 g/cm 3 . POSs were then sedimented by centrifugation at 41,400× g for 30 min at 4 • C on a Beckman Coulter Allegra 64R centrifuge equipped with a F0650 rotor and resuspended with the required amount of 0.1 M phosphate buffer, pH 7.4. All manipulation of samples containing rhodopsin was carried out under dim red illumination.
For chromatographic analysis, chloroform extracts from LG suspension or supernatant samples were prepared using the Folch method [59], followed by drying and subsequently resuspending in 200 µL methanol, as described previously [58]. Precision for each sample was determined from two separately measured chromatograms for each individual sample at each wavelength.

Sample Oxidation
For photo-oxidative destruction, the LG suspension in phosphate buffer or A2E in acetonitrile were irradiated at room temperature under constant stirring using a 150 W incandescent lamp with a heat filter (KGM 24-150, 400-700 nm). The luminous flux density irradiating the sample was 80 mWm−2 for visible light (400-700 nm), as determined by a photometer (Spectra-Physics 407A, Milpitas, CA, USA).
For oxidative destruction, reaction of the samples with superoxide radicals was conducted in the dark by adding an equal mass of dry potassium peroxide (KO 2 ) to each sample (10-25 mg). Sample incubation with superoxide was conducted for 1 h with periodic stirring at room temperature.

A2E Synthesis
A2E was prepared from all-trans retinal and ethanolamine in acetic acid and ethanol, as described previously [60]. A2E and its photo-oxidation products were identified using a 7T LTQ FT mass spectrometer (Thermo Electron Corp., Bremen, Germany) equipped with an electrospray ion source, as described previously [32]. To reliably identify A2E, we performed MC/MC studies in a linear quadrupole trap, in which molecular ions with a certain mass-to-charge ratio (M/Z) were isolated from other ionization products and subjected to collision-induced dissociation (ESI-CID) with subsequent mass spectrometry analysis of the fragmentation products [39]. Mass spectra were processed and analyzed using Qual Browser 1.4 software.

Mass Spectrometry Analysis
Sample mass spectra were obtained using a qTOF mass spectrometer MaXis 4G (Bruker Daltonics, Bremen, Germany) with an electrospray ion source (ESI) in positive mode. ESI source parameters were the following: capillary 3.4 kV, nebulizing gas pressure 0.7 bar, drying gas flow rate 6 L/min, and drying gas temperature 200 • C. Samples were dissolved into a mixture of acetonitrile and water (1:1) and sprayed at a flow rate of 3 µL/min. Measurements were conducted in the range of m/z 100-1500. Mass spectra were processed and analyzed using data analysis software (Bruker Daltonics, Bremen, Germany).
To detect BisRet-OXs, such as aldehydes and ketones, dried LG samples were analyzed by time-of-flight secondary ion mass spectrometry (ToF-SIMS) (ION-ToF, Münster, Germany) using 30 keV Bi 3 + primary ions. The analysis area was 300 × 300 µm 2 with a primary ion dose density of~4 × 10 11 ions/cm 2 . For each sample, at least nine such areas were analyzed. All spectra were recorded in both positive and negative ion modes. An electron flood gun was activated to avoid the charging effect during analysis. Mass spectra were processed and analyzed by SurfaceLab software (ION-TOF, Münster, Germany). Ion yields were calculated as the number of ions with specific masses divided by total ion counts.

Raman Spectroscopy Analysis
Raman spectroscopy analysis was conducted using a Broadband coherent anti-Stokes Raman scattering (BCARS) microspectrometer implemented as part of a femtosecond laser complex [61] representing a two-pulse collinear scheme for CARS generation [62]. Related spectra were transformed to Raman bands using the CARS-variation of MEM [63,64].
LG sample preparation and experimental conditions have been described in detail in prior studies [50].

Determination of Carbonyls
The relative toxicities of products that accumulated from oxidative and photo-oxidative degradation of BisRets in LG suspensions, LG sediments, and supernatants were assessed by the content of carbonyls reacting with TBA-active products [65]. The content of TBAactive products in chloroform extracts was determined after vacuum pump evaporation of the chloroform (Vacuubrand MZ2C NT+AK+M+D, Wertheim, Germany), followed by sediment solubilization in 0.1 M potassium phosphate buffer (pH 7.3). TBA-active product concentration was determined spectrally in photo-oxidized and superoxide-oxidized LG samples at 532 nm [66] using a Shimadzu UV-1700 spectrophotometer (Shimadzu, Kyoto, Japan). The original non-oxidized granules were used as controls.

Determination of Protein and Lipid Modification Products
Protein and lipid modifications were assessed by the formation of fluorescent Schiff bases in the reaction of aldehydes with the free amino groups of proteins and lipids. POS suspensions were used as a protein and lipid substrate (see Section 4.3). The incubation medium contained 0.1 M sterile potassium phosphate buffer, pH 7.4; 0.3-0.6 mg/mL POS protein, 3-4 mM sodium azide, and 0.4-0.7 mL supernatants from irradiated and nonirradiated (control) LG suspensions. As additional control samples, we used samples containing only POS suspensions without supernatants, supernatants without added proteins, and samples containing a glycation inhibitor, aminoguanidine (4 mM). The samples were incubated at 37 • C in the dark with constant stirring for 24, 48, or 72 h.
After incubation, sample aliquots were dialyzed against phosphate buffer to remove unreacted low molecular weight molecules. For dialysis we used a Float-A-Lyser celluloseether membrane (SPECTRUM Labs, New Brunswick, NJ, USA), which is permeable to molecules <3.5 kDa. Dialysis was conducted for 25 h at 6 • C. After dialysis, the fluorescence spectra of the modified proteins and lipids were measured at an excitation wavelength of 365 nm. The modified protein and lipid content was estimated by the magnitude of the fluorescence amplitude, measured at the emission maximum of 440-455 nm.

Statistical Analysis
Statistical significance of experimental results was analyzed using a Student's t-test. Results with p < 0.05 were considered statistically significant. Data are presented as mean ± standard deviation (SD).

Data Availability Statement:
The data presented in this study are available in the article and on request from the corresponding author.