Crosstalk of protein clearance, inflammasome, and Ca²⁺ channels in retinal pigment epithelium derived from age-related macular degeneration patients

Abstract

Degeneration and/or dysfunction of retinal pigment epithelium (RPE) is generally detected as the formation of intracellular and extracellular protein aggregates, called lipofuscin and drusen, respectively, in patients with age-related macular degeneration (AMD), the leading cause of blindness in the elderly population. These clinical hallmarks are linked to dysfunctional protein homeostasis and inflammation and furthermore, are both regulated by changes in intracellular Ca²⁺ concentration. While many other cellular mechanisms have been considered in the investigations of AMD-RPE, there has been relatively little work on understanding the interactions of protein clearance, inflammation, and Ca²⁺ dynamics in disease pathogenesis. Here we established induced pluripotent stem cell–derived RPE from two patients with advanced AMD and from an age- and gender-matched control subject. We studied autophagy and inflammasome activation under disturbed proteostasis in these cell lines and investigated changes in their intracellular Ca²⁺ concentration and L-type voltage-gated Ca²⁺ channels. Our work demonstrated dysregulated autophagy and inflammasome activation in AMD-RPE accompanied by reduced intracellular free Ca²⁺ levels. Interestingly, we found currents through L-type voltage-gated Ca²⁺ channels to be diminished and showed these channels to be significantly localized to intracellular compartments in AMD-RPE. Taken together, the alterations in Ca²⁺ dynamics in AMD-RPE together with dysregulated autophagy and inflammasome activation indicate an important role for Ca²⁺ signaling in AMD pathogenesis, providing new avenues for the development of therapeutic approaches.

Age-related macular degeneration (AMD) is the leading cause of blindness in the elderly population, and it has become a major public health and financial burden. The most prevalent form of AMD, dry AMD (80–85% of cases), is currently untreatable (1). Wet AMD in turn (affecting 10–15% of patients) is defined by the growth of blood vessels from choriocapillaris into the retina, a process called neovascularization. Intravitreal injections of anti-vascular endothelial growth factor and drugs (i.e. bevacizumab, ranibizumab, aflibercept, brolucizumab) have proven to be effective to some extent (2). AMD etiology is known to be multifactorial. Aging, smoking, arteriosclerosis, obesity, hypertension, hypercholesterolemia, and unhealthy diet predispose to AMD (1). In addition, the genetic factors seem comprising up to 71% of the disease severity (3, 4, 5), which is strongly linked to increased oxidative stress (4).

The increased oxidative stress may induce dysfunction of retinal pigment epithelium (RPE) leading to AMD. RPE is a tight monolayer of cells under the photoreceptors that has many important roles in maintaining retinal homeostasis and the function of photoreceptors, such as providing metabolic support, controlling ionic homeostasis, absorbing excessive light, regenerating visual pigment, and helping photoreceptor renewal through phagocytosis (5, 6, 7). Many of these functions, such as phagocytosis, constantly expose RPE cells to oxidative stress and create a need of removing damaged cellular proteins, lipids, nucleic acids, and cellular organelles, including mitochondria through cellular clearance pathways (1, 6). Two main pathways that keep the protein homeostasis in balance are the ubiquitin–proteasome and the lysosomal/autophagy pathways. In case of AMD, the lipofuscin formation in the RPE and extracellular drusen accumulation between the RPE and Bruch’s membrane demonstrate that the increased level of oxidative stress influences the normal RPE function and results in disturbed proteostasis (1). Additionally, chronic oxidative stress and inflammation are all strongly linked to AMD pathogenesis on cellular level. These factors activate and amplify each other leading to a loss of cellular homeostasis and eventually cell death (1, 8, 9).

Drusen deposits, found in AMD, contain inflammatory proteins and thus, are evidence of an imbalance of both the proteolytic and inflammatory systems (10, 11, 12). Inflammasomes are cytosolic multiprotein complexes that facilitate the maturation of pro-inflammatory cytokines that together are key components of innate immune response (9). The most studied inflammatory proteins in the RPE of AMD patients are the components of the NLRP3 (NACHT, LRR, and PYD domains-containing protein 3) inflammasome (9, 13, 14) and pro-inflammatory cytokines, such as interleukin (IL)-1β, IL-6, tumor necrosis factor-α, and IL-18 (9, 13, 15). These proteins have been found to be upregulated in advanced AMD. Additionally, IL-1β, IL-6, tumor necrosis factor-α, and IL-18 have been proposed as biomarkers of AMD (9, 13, 15).

Signaling routes of proteostasis and inflammation are carefully controlled by changes in temporal and spatial distribution of intracellular free Ca²⁺ (16, 17, 18, 19, 20). Generally, the increase in intracellular Ca²⁺ concentration can occur via two pathways: Ca²⁺ is released from intracellular stores or it travels through plasma membrane Ca²⁺ channels into the cytosol (21). Abnormalities in these calcium pathways have been associated with several neurodegenerative diseases, such as Alzheimer’s disease (22, 23) and Huntington disease (24). Interestingly, aging (25) and drusen accumulation (26), risk factors linked to AMD, have also been associated with disturbed Ca²⁺ signaling.

One important activator of Ca²⁺ signaling in RPE is the energy-carrying molecule, adenosine triphosphate (ATP). ATP is capable of binding to purinergic receptors and causing a molecular cascade where the level of cytosolic inositol (1,4,5) triphosphate increases and Ca²⁺ is released from the endoplasmic reticulum (ER). This ATP-induced Ca²⁺ signaling is essential for the functionality of RPE as it regulates ion transport and hydration of subretinal space, affecting to the communication between RPE and retina (27, 28). Interestingly, ATP levels in the subretinal space are increased following light stimulation (28), and it has been shown that stressed retinal cells utilize the release of ATP into subretinal space as a danger signal to RPE (29).

As a route for Ca²⁺ into the cytosol, RPE expresses a variety of Ca²⁺ channels including L-type voltage-gated Ca²⁺ channels (Ca_V). These channels have a key role in many important functions in the RPE, such as phagocytosis (30), secretion of growth factors (31), and differentiation (5). Furthermore, altered regulation of L-type Ca²⁺ channels has been connected to pathophysiology of degenerative diseases of the retina (32). Three subtypes of L-type Ca²⁺ channels, Ca_V1.1–Ca_V1.3, are expressed in the RPE, but Ca_V1.3 is considered to have the main role in RPE functions (5, 30, 31, 33, 34, 35, 36).

Induced pluripotent stem cells (iPSCs) offer effective tools for studying AMD pathophysiology. The iPSCs are generated from adult somatic cells that are harvested from patients with AMD providing disease-specific cell models (37, 38). The technology has been widely used in AMD modeling (e.g., (39, 40, 41)) as well as for high-throughput screening of potential novel drug candidates for AMD treatments (42).

In the present study, we established iPSC-derived RPE cells from two patients with advanced AMD and one healthy control subject to study the role of Ca²⁺ in AMD and focusing on autophagy and inflammasome activation under disturbed proteostasis. We demonstrate that iPSC-RPE cells derived from AMD patients have reduced intracellular free Ca²⁺ levels and impaired Ca_V1.3 functionality and localization. These alterations occurred in AMD-RPE that also exhibit dysregulated autophagy and inflammasome activation. Together, these observations can open new ways to understand the cellular pathophysiology of AMD paving the way for therapy development.

Results

Each iPSC-RPE cell line showed mature RPE characteristics

During the iPSC reprogramming induction, all three donor samples performed similarly with no visible differences. After the selection of iPSC colonies (at day 22–24) and further passaging of selected individual clones, generated iPSC lines propagated with similar expansion rate and morphology. For the establishment of working stem cell banks, three individual iPSC clones were selected from each donor. The iPSCs were thoroughly characterized with well-established characterization methods, and their pluripotency was confirmed based on morphology, the expression of pluripotency markers (stage-specific embryonic antigen-4 (SSEA-4), tumor-related antigen-1-60 (TRA-1-60)), with trilineage differentiation capacity towards mesoderm, ectoderm, and endoderm, and normal diploid karyotype (Fig. S1). Further mycoplasma negativity was confirmed with VenorGeM Classic (data not shown).

For the initial RPE differentiations and characterizations, three iPSC clones from all three donors (two AMD patients and one healthy control subject, Table 1) were used. Based on the initial iPSC-RPE characterizations, one clone from the healthy control (control-RPE) and AMD1 (AMD1-RPE) and two clones from AMD2 (AMD2₍₁₎-RPE and AMD2₍₂₎-RPE) were selected for more rigorous RPE characterizations and further experiments. Two clones were selected from AMD2 patient due to the patient’s more severe visual impairment (see Table 1). Figure 1 shows mature RPE characteristics for each cell line; the presence of RPE marker cellular retinaldehyde binding protein (CRALBP), the localization of ezrin and Na⁺/K⁺-ATPase to the microvilli and apical membrane, and the expression of phagocytosis-associated marker tyrosine-protein kinase Mer (MERTK). Interestingly, ezrin staining appeared more heterogenous and patchy in the two AMD-RPE cell lines compared to the control. Concerning RPE barrier properties, all cell lines also demonstrated intact tight junctions shown as junctional localization of zonula occludens-1 (ZO-1) (Fig. 2A). However, the expression and localization of claudin-19 in the junctional area was weaker in the AMD2-RPE cell line than AMD1-RPE cell line, being the strongest and most junctional in the control RPE (Fig. 2A). When evaluating the functionality of the iPSC-RPE, the transepithelial electrical resistance (TEER) was observed to be similar between control-RPE (630 ± 114 Ωcm², n = 31) and AMD1-RPE (604 ± 66 Ωcm², n = 30), whereas AMD2-RPE (207 ± 50 Ωcm², n = 30 and 199 ± 31 Ωcm², n = 15) had significantly lower TEER values (p < 0.0001 for both, Fig. 2C). Note that AMD2-RPE was derived from the individual with the greatest visual impairment level (Table 1). Each iPSC-RPE cell line showed polarized secretion of pigment epithelium-derived factor (PEDF), but the secretion level of apical PEDF was significantly smaller for AMD1-RPE than the control (n = 3–4, p = 0.029, Fig. 2D). The phagocytosis capacity of control-RPE, AMD1-RPE, and AMD2₍₁₎-RPE was tested by 1 h room temperature (RT) + 2 h at 37 °C assay, and in all cell lines, bound and internalized photoreceptor outer segment (POS) particles can be seen in the confocal micrographs together with hexagonal RPE morphology (Fig. 2B). When quantifying the number of internalized POS particles (see Experimental procedures), statistically significant difference was found between AMD1-RPE and control-RPE (p = 0.0007), AMD1-RPE having more internalized POS particles (n = 10–11, Fig. 2E). No difference was found between AMD2₍₁₎-RPE and the control (p = 0.17). We also tested the phagosome-processing capacity of iPSC-RPE cell lines using 1 h RT + 4 h at 37 °C as the time point of the processing phase (n = 10–11). The number of POS particles declined significantly between the 2 h and 4 h time points, indicating their further processing in each cell line (p = 0.029, p = 0.004, p = 0.0001, see Fig. 2E). Overall, the mature characteristics as well as the functionality of the compared RPE lines were in most respects similar, demonstrating the consistency of our iPSC-based RPE differentiation.

Table 1.

Demographics of the patients and naming of the generated iPSC lines

iPSC-RPE line	Clinical diagnosis	Age	BMI	Smoking	GA	BP	AC	AA	Visual impairment
Control	No AMD	73	24	No	-	+	-	+	Good visual acuity
AMD1	Wet/wet	76	24	Yes	+/+	+	+	-	Moderate/blindness
AMD2	Wet/dry	71	26	Not currently	+/+	+	-	-	Severe/blindness

The slash symbol is used to divide right and left eyes.

Abbreviations: AA, anti-aggregation medication; AC, anti-coagulation medication; BMI, body mass index; BP, blood pressure medication; GA, geographic atrophy.

Figure 1.

Mature characteristics of the iPSC-RPE monolayers of control-RPE and the two AMD-RPE cell lines with z-projections and cross-sections (y-projections of 10 sections highlighted with a white bar). Nucleus (DAPI, magenta) labeled together with an RPE marker CRALPB (cellular retinaldehyde-binding protein, gray and green), microvilli-localized ezrin (gray and green), polarization marker Na⁺/K⁺-ATPase (gray and green) and, phagocytosis-associated receptor MERTK (tyrosine-protein kinase Mer, gray and green). Scale bars represent 10 μm. AMD, age-related macular degeneration; iPSC, induced pluripotent stem cell; RPE, retinal pigment epithelium.

Figure 2.

Functional characteristics of control-RPE and the two AMD-RPE cell lines.A, laser scanning confocal microscopy (LSCM) z-projections of tight junction–associated protein ZO-1 and claudin-19 of iPSC-RPE (scale bars represent 10 μm). B, phagocytosis assay performed by incubating purified porcine photoreceptor outer segment (POS) particles on the iPSC-RPE for 1 h RT + 2 h at 37 °C. LSCM z-projections (scale bars represents 25 μm) and cross-sections (y-projections of ten sections highlighted with a white bar) show staining of POS (green) together with filamentous actin (magenta). Lower panel shows in higher magnification the region highlighted with a white dashed box (scale bar represents 10 μm). C, box plot with a mean line and SD whiskers show transepithelial electrical resistance (TEER) measurements for each cell line (n = 15–31). D, a bar chart showing that the secretion of pigment epithelial-derived growth factor (PEDF) was predominantly apical in all cell lines (mean and SD whiskers, n = 3–4). E, box plot with a mean line and SD whiskers show the quantification of internalized POS particles per field (n = 10–11 fields from three inserts) after 1 h RT + 2 h at 37 °C and 1 h RT + 4 h at 37 °C phagocytosis assays. The number of POS particles decreased significantly between the 2 h and 4 h time points, indicating POS processing. AMD, age-related macular degeneration; iPSC, induced pluripotent stem cell; RPE, retinal pigment epithelium; ZO-1, zonula occludens-1; RT, room temperature.

Increased Hsp70 and decreased ubiquitin-tagged protein levels in AMD-RPE under proteasomal inhibition

The capacity for autophagic clearance of iPSC-RPE cells was tested by treating cells with proteasome inhibitor MG-132 causing accumulation of proteins and generation of protein aggregates (43). The accumulation of ubiquitin-tagged proteins, autophagy-destined protein aggregates (sequestosome-1 (SQSTM1), also known as p62), and molecular chaperone (heat shock 70 kDa protein (Hsp70)) was evident after 24 h and 48 h of proteasome inhibition (Fig. 3, A–D, n = 5, p = 0.008 for each iPSC-RPE cell line). Interestingly, the accumulation of ubiquitinated proteins and SQSTM1 were significantly higher in the control-RPE than in AMD-RPE in comparison with the respective starting levels after 24 h (n = 5, ubiquitinated proteins: p = 0.032 for both AMD-RPE lines, SQSTM1: p = 0.032 and p = 0.008, see Fig. 3, A and C). Instead, the activation of Hsp70 was significantly more elevated in the AMD1-RPE than in control-RPE after proteasome inhibition (n = 5; after 24 h, p =0.032 and after 48 h, p = 0.008, see Fig. 3, A–D). AMD2₍₁₎-RPE showed a tendency for increased Hsp70 levels after 24 h proteasome inhibition, and the increase in Hsp70 response was evident after 48 h (n = 5, p = 0.008, Fig. 3, B and D). Simultaneously, the levels of ubiquitin-tagged proteins in AMD2₍₁₎-RPE were significantly reduced after 48 h (n = 5, p = 0.016, Fig. 3, B and D), and the AMD1-RPE showed a tendency for decreased ubiquitin-tagged proteins. The accumulation of SQSTM1-marked protein aggregates showed no difference between control-RPE and AMD-RPE after 48 h proteasome inhibition. Also, the removal of proteasome inhibitor and the 24 h recovery from 24 h or 48 h proteasome dysfunction showed no difference in the degradation of protein aggregates between the cell lines (Fig. S2), indicating functional autophagy without proteasome inhibition.

Figure 3.

Characterization of the function of molecular chaperones, ubiquitin–proteasome system, and autophagy in iPSC-RPE.A, Western blot of Hsp70, SQSTM1, and ubiquitinated proteins with and without proteasome inhibition (MG-132, 5 μM) after 24 h. B, Western blot of Hsp70, SQSTM1, and ubiquitinated proteins with and without proteasome inhibition (MG-132 5 μM) after 48 h. C, quantitation of Hsp70, SQSTM1, and ubiquitinated proteins with and without 5 μM MG-132 after 24 h. D, quantitation of Hsp70, SQSTM1, and ubiquitinated proteins with and without 5 μM MG-132 after 48 h. E, Western blot of LC3-II and SQSTM1 with and without autophagy inhibitor bafilomycin A1 (150 nM) for 3 h. F, Western blot of LC3-II and SQSTM1 upon autophagy inhibitor bafilomycin A1 (150 nM, 3 h) with and without 24 h proteasome inhibition (MG-132, 5 μM). G, quantitation of SQSTM1 upon autophagy inhibitor bafilomycin A1 (150 nM) for 3 h. H, quantitation of LC3-II upon autophagy inhibitor bafilomycin A1 (150 nM) for 3 h. I, quantitation of SQSTM1 upon autophagy inhibitor bafilomycin A1 (150 nM) with and without proteasome inhibition (MG.132, 5 μM). J, quantitation of LC3-II upon autophagy inhibitor bafilomycin A1 (150 nM) with and without proteasome inhibition (MG.132, 5 μM, comparison has been made between bafilomycin A1 and bafilomycin A1 + MG-132 treatments). All bands were normalized against α-tubulin followed by a comparison between treated and nontreated bands resulting in relative protein levels (fold change against control). Bar charts show mean and SD whiskers and are combined from 2 to 3 independent experiments giving n = 5. iPSC, induced pluripotent stem cell; RPE, retinal pigment epithelium; SQSTM1, sequestosome-1.

Autophagy activation was further studied by using bafilomycin A1 treatment, which inhibits lysosomal acidification and thereby autophagosome degradation. Bafilomycin A1 caused accumulation of the observed autophagosomes as significantly increased LC3-II levels for each cell line (n = 5, p = 0.008 for all). When comparing the baseline autophagy without proteasomal inhibition and stress, the control-RPE and AMD-RPE showed no difference in LC3-II accumulation upon bafilomycin A1 treatment (Fig. 3, E and H). Proteasome inhibition created a need for protein clearance, and simultaneous Bafilomycin A1 treatment resulted in lower LC3-II accumulation in AMD-RPE in comparison to the control-RPE (n = 5, p = 0.032 and p = 0.016, see Fig. 3, F and J), indicating reduced formation of autophagosomes. The levels of SQSTM1 upon bafilomycin A1 treatment increased slightly in all cell lines (Fig. 3, E and G), and the proteasome inhibition resulted in similar decreasing trend in AMD-RPE cells as seen with LC3-II (Fig. 3, F and I). Thus, AMD-RPE seemed to respond to protein accumulation by inducing the molecular chaperone response rather than autophagy.

Only control-RPE showed increased caspase-1 activation after proteasomal inhibition

In order to study cell viability and inflammasome activation, iPSC-RPE cells were first primed using human recombinant IL-1α and then treated with MG-132. IL-1α mimics early inflammation in cell cultures by inducing the release of other pro-inflammatory cytokines and the production of essential components of NLRP3 inflammasome signaling and, therefore, primes RPE cells for the inflammasome activation (44). Exposure to MG-132 for 48 h increased the leakage of lactate dehydrogenase (LDH) on both the apical and basal sides in all IL-1α–primed iPSC-RPE lines (n = 5–6, p = 0.004 to p = 0.002, see Fig. 4A). Concurrently, NLRP3 levels in the basal side medium increased significantly following MG-132 exposure (n = 4, p = 0.029 for all, Fig. 4B). However, proteasome inhibition in IL-1α–primed cells caused no significant increase in IL-1β levels on either the basal or apical side of any iPSC-RPE cell line (n = 5–6, Fig. 4C). Noticeably, caspase-1 activity, a measure of inflammasome activation, was significantly increased by MG-132 exposure only in the control cell line (n = 7–14, p = 0.008), which also showed little to no baseline activity following IL-1α priming (Fig. 4, D and E). Conversely, AMD-RPE showed significantly higher baseline caspase-1 activity (p < 0.001 and p = 0.0008, see Fig. 4E) and tended to secrete higher apical levels of IL-1β following MG-132 exposure (p = 0.013, see Fig. 4C). MG-132 exposure led to a robust increase in the secretion of both IL-6 and IL-8 to the basal side of IL-1α–primed iPSC-RPE cells, and the levels of the secreted cytokines in the basal side medium tended to be higher in AMD-RPE (Fig. 4, F and G).

Figure 4.

Analysis of cell viability and inflammatory responses in iPSC-RPE cells. Cells were primed with IL-1α (4 ng/ml, 24 h) and exposed to MG-132 (5 μM, 48 h (A–C, F, G), 6 h (D, E)).A, LDH levels released from iPSC-RPE cells (n = 5–6). B, extracellular levels of NLRP3 as a level of measured absorbance at 450 nm corrected with absorbance at 620 nm (n = 4). C, levels of IL-1β released from iPSC-RPE cells (n = 5–6). D, caspase-1 activity, detected using the fluorescent FLICA probe (green) and nuclei (blue). Representative images are shown, and the arrows indicate the counted FLICA-positive cells. Scale bars represent 10 μm. E, quantitative analysis of FLICA-positive cells (n = 7–14). F, levels of IL-6 secreted from the iPSC-RPE (n = 5–6). G, secretion of IL-8 from the iPSC-RPE cells (n = 5–6). Bar charts show mean as well as SD whiskers. IL, interleukin; iPSC, induced pluripotent stem cell; LDH, lactate dehydrogenase; RPE, retinal pigment epithelium.

AMD-RPE show diminished intracellular calcium levels and ATP-induced calcium signaling

Disturbed autophagy and inflammation are known to play a role in the development of AMD. However, the underlying mechanisms that are causing these abnormalities to thrive are still unknown (1, 8, 9). Intact RPE functionality depends on precisely regulated intracellular Ca²⁺ concentration (5), and several factors activate Ca²⁺ signaling within the RPE, one important example being ATP (27). As Ca²⁺ is known to participate in the activation of autophagy and inflammation (16, 17, 18, 19, 20), we studied the free cytosolic Ca²⁺ levels and ATP-induced Ca²⁺ signaling in our RPE cell lines. The experiments were performed using Fluo-4-acetoxymethyl ester (Fluo-4 AM) Ca²⁺ indicator and fluorescence time-lapse microscopy. Spontaneous Ca²⁺ signaling was recorded for 2 min, and the average fluorescence intensities were analyzed for the duration of the recordings to compare the average free cytosolic Ca²⁺ levels between the cell lines. Interestingly, the average intensities were significantly lower in both AMD1-RPE and AMD2₍₁₎-RPE than control-RPE (n = 10–11, p = 0.006 and p < 0.0001, Fig. 5, A and B). Additionally, 100 μM ATP-induced Ca²⁺ responses were measured from each cell line (Fig. 5C), and the response properties were analyzed from the cell populations at the single cell level (Fig. 5D). We focused on maximum amplitude and response duration at the 50% threshold and determined them from the fluorescence intensity curves (Fig. 5E). The responses were qualitatively similar in all cell lines (Fig. 5D), yet the more detailed analysis revealed small but statistically significant differences between control-RPE and AMD-RPE with reduced ATP-induced Ca²⁺ responses (n = 630, p < 0.0001, Fig. 5F) and shorter response durations (n = 630, p < 0.0001, Fig. 5G) in AMD-RPE.

Figure 5.

Intracellular Ca²⁺concentration and ATP-induced Ca²⁺response in control-RPE and AMD-RPE.A, representative average intensity projection from each cell line of the 2 min time-lapse microscopy with Fluo-4 Ca²⁺ indicator; scale bar represents 100 μm. B, the corresponding box plots with a mean line and SD whiskers (n = 10–11). C, pseudocolored image time series of the 100 μM ATP-induced Ca²⁺ response in each iPSC-RPE monolayer, scale bar represents 25 μm. D, fluorescence intensity time course showing the representative ATP-induced Ca²⁺ responses in single cells in the control-RPE and both AMD-RPE cell lines. E, a schematic curve with illustrations of the parameters determined from the ATP-induced Ca²⁺ response from each cell. Amplitude is the maximum amplitude of the response measured from the baseline, and the duration of the response is the time in seconds between 50% threshold values. Box plots present mean lines and SD whiskers of (F) amplitude (normalized to the baseline, n = 630 cells from three independent experiments) and (G) duration (n = 630 cells from three independent experiments) for each cell line. AMD, age-related macular degeneration; iPSC, induced pluripotent stem cell; RPE, retinal pigment epithelium.

To study the link between Ca²⁺ and autophagy, we used a spontaneously arising human RPE cell line ARPE-19 and induced an increase in intracellular Ca²⁺ concentration by treating IL-1α–primed cells with 0.5 μM thapsigargin and 0.5 μM thapsigargin together with 1 mM 5-Aminoimidazole-4-carboxamide ribonucleotide (AICAR), an activator of AMP-activated protein kinase to attenuate ER stress. Following thapsigargin treatment, the levels of SQSTM1, LC3-I, and LC3-II increased, and these effects were attenuated in the presence of AICAR (Fig. S2).

Reduced currents and altered Ca_V1.3 localization in AMD-RPE

Following our observation of the reduced intracellular free Ca²⁺ levels in AMD-RPE, we turned our attention to the L-type voltage-gated Ca²⁺ channels and evaluated their presence and functionality in the three RPE cell lines. These channels, together with the members of transient receptor potential (TRP) channel family, are known to influence the intracellular Ca²⁺ concentration and its dynamics in RPE (45, 46). To study channel functionality, we used whole cell patch clamp technique and recorded ionic currents from intact iPSC-RPE monolayers (Fig. 6A). Observed currents were responses to 50 ms voltage steps ranging from −70 mV up to 80 mV in +10 mV steps. Each cell line showed voltage-gated inward currents with I-V curve characteristic to L-type voltage-gated Ca²⁺ channels (Fig. 6B). The recorded currents were maximally activating at relatively positive potentials (Fig. 6B) and showed slowly inactivating kinetics (Fig. 6C). The average I–V curve (n = 9–10) showed significantly larger maximum current amplitudes for control-RPE (−69.5 ± 54.8 pA) than AMD-RPE (−29.7 ± 11.9 pA and −34.3 ± 12.6 pA, p = 0.003 and p = 0.010, Fig. 6, D and F). Additionally, the current density was significantly higher for control-RPE (−3.0 ± 3.5 pA/pF) than AMD-RPE (−1.1 ± 0.3 pA/pF and −1.4 ± 0.7 pA/pF, p = 0.0003 and p = 0.022, Fig. 6F). All the recorded currents reached the half-maximal activation potential around −33 to −44 mV and peak current around 26 to 42 mV with no statistically significant difference between cell lines (Fig. 6F).

Figure 6.

Localization and functionality of Ca_V1.3 channels in control- and AMD-RPE.A, brightfield light microscopy images from iPSC-RPE cells (scale bar represents 25 μm). B, averaged current–voltage relationship with mean and SD whiskers (n = 9–10) from whole cell patch clamp recordings in each RPE cell line using depolarizing voltage pulses (50 ms pulses from −70 to 80 mV in 10 mV steps). C, representative examples of the recorded slowly inactivating currents. D, box plots present mean line and SD whiskers of maximum current amplitudes in each RPE cell line. E, example of the inhibition effect of 10 μM Nifedipine in AMD2₍₁₎-RPE shown by current–voltage curves with data points normalized to the maximum current in control solution (top) and an example of the recorded currents in the two conditions (bottom). F, summary of the patch clamp results. G, Z-projections of immunolabeled Ca_V1.3 with filamentous actin (phalloidin) in each cell line (scale bar represents 10 μm) and (H) cross-sections (projections of 3 sections), highlighted with a white dotted bar in (G), (scale bar represents 5 μm). I, schematic curve of Z-axis fluorescence profiles of Ca_V1.3 and filamentous actin together with a cross-section image (reuse of a portion of (H) “AMD2₍₁₎” image, scale bar represents 5 μm) illustrating the determination of peak-to-peak distance of the two profiles. J, quantification of the peak-to-peak distances (n = 10–11) by box plot with mean line and SD whiskers. AMD, age-related macular degeneration; iPSC, induced pluripotent stem cell; RPE, retinal pigment epithelium.

Further verification of the recorded currents being carried through L-type voltage-gated Ca²⁺ channels was conducted using an L-type Ca²⁺ channel inhibitor, nifedipine. All cell lines were sensitive to the inhibitor (n = 7–10, p < 0.007 for all), and the maximum current amplitude decreased around 59 to 64% after 5 to 10 min of the perfusate addition of 10 μM nifedipine. No alterations were seen in the activation voltages when nifedipine was applied (Fig. 6E, see also (47)), and no statistical difference could be detected between the cell lines regarding the inhibition effect of nifedipine (Fig. 6F). Taken together, the recordings show the presence of L-type Ca²⁺ channels (based on I-V curve characteristics, current kinetics, and Nifedipine sensitivity) in the three cell lines, albeit with diminished current densities in AMD-RPE.

Ca_V1.3 has been identified as the primary L-type Ca²⁺ channel in RPE (5, 6, 47). We performed immunofluorescence staining and confocal microscopy of L-type Ca²⁺ channels Ca_V1.1-1.3 in our RPE cell lines. By using identical imaging settings with all three iPSC-RPE lines, we observed the most consistent and intense staining with Ca_V1.3 (Fig. S3). Combining these findings with the literature, we focused on Ca_V1.3 in our further work. Although all iPSC-RPE of this study expressed Ca_V1.3 (Fig. 6G), cross-sections revealed differences in channel localizations (Fig. 6H). In both AMD-RPE cell lines, localization of Ca_V1.3 on the apical cell membrane underneath the microvilli was reduced compared to the control and instead, there was a more prominent intracellular pool of Ca_V1.3. To quantify the difference in channel localization, we plotted z-fluorescence profiles from the confocal stacks showing Ca_V1.3 together with cortical actin (stained with phalloidin) near the cell membrane (Fig. 6I). From the intensity profiles, we determined the peak-to-peak distance between the major peaks of Ca_V1.3 and phalloidin (n = 10–11, Fig. 6J) for each cell line. The quantification revealed that the distance of apical Ca_V1.3 peak from the cortical actin peak at the microvilli was significantly larger for AMD-RPE than for control-RPE (p < 0.0001, Fig. 6J), demonstrating that more Ca_V1.3 was localized in the intracellular compartments in the AMD-RPE cell lines.

Discussion

iPSC-derived somatic cells from patients and healthy controls have provided a valuable tool for researchers to reveal complex disease-specific mechanisms that have been challenging to study using continuous cell lines, ex vivo differentiated cell models, or animals (38). iPSC-derived cells have the capability to exhibit the genetic signature of the desired human tissue and provide sufficient material for recognizing new disease-specific mechanisms (37, 38).

In this study, we used iPSCs from two patients with advanced AMD and from an age-matched control subject. Both AMD patients had family history as an aggravating genetic factor for AMD, and contrarily, the family history of the control subject did not predispose to AMD. The iPSCs were differentiated into RPE cells that formed an appropriate tight monolayer of pigmented hexagonal cells with the expression of mature RPE-specific markers without differences between AMD and control. To further study the epithelial integrity and barrier function, we measured TEER. The TEER values were significantly lower in AMD2-RPE, which showed the most difficult clinical phenotype, than in control- or AMD1-RPE. All the measured TEER values were in the range of earlier reports from iPSC-RPE generated from AMD patients and healthy controls (48, 49). Interestingly, AMD2-RPE cell line showed also a weaker claudin-19 expression in cell-cell junctions than in control- and AMD1-RPE. These two results together suggest that the barrier properties in the clinically most sever phenotype might be compromised. We also studied PEDF secretion and phagocytosis to further evaluate the functional characteristics of the cell lines. We found that all three cell lines had properly polarized PEDF secretion. However, the level of apical secretion was significantly decreased in AMD1-RPE, suggesting a possible link to AMD as PEDF is capable of inhibiting the growth of new blood vessels (50). Phagocytosis is one of the key functions of the RPE, moreover, the dysfunction of phagocytosis has been linked to lipofuscin accumulation and AMD pathogenesis (51). However, no decline in phagocytic function in AMD-RPE could be detected in our in vitro phagocytosis assay. It is possible that the phagocytosis defects emerge only under long-term phagocytic challenge reflecting the factor of age in the disease etiology.

Even though we did not detect significant decline in the phagocytic capacity of AMD-RPE compared to control-RPE, our study interestingly showed that AMD-RPE had reduced formation of autophagosomes under stress introduced with proteasome inhibitor MG-132. This might be evidence of reduced autophagic activity. Indeed, our observations are in line with previous studies that demonstrated dysfunctional autophagy in the eyes of human AMD donors as well as in two animal models of AMD (52). It is noteworthy that regulatory mechanisms of phagocytosis and autophagosome formation differ from each other although are linked together with lysosomal enzymes (53).

An important factor related to inhibition of proteasome by MG-132 treatment is Hsp70 as it has a major role in the regulation of proteostasis (54, 55). Hsp70 is highly upregulated during proteasomal dysfunction, simultaneously leading to increased cellular ubiquitination (43). Here, the level of Hsp70 was significantly higher in AMD-RPE cells than in control-RPE in response to MG-132 treatment, but strikingly, ubiquitinated protein conjugates were not markedly elevated. This suggests a stronger molecular chaperone response in AMD-RPE cells rather than activation of autophagy machinery. This could be explained by a high Hsp70 capacity to repair dysfunctional proteins with functional proteasomes or disturbed ubiquitination in AMD pathology. Our earlier studies propose that Hsp70 is effectively protected from degradation by autophagy even if the proteasomal clearance systems are not fully functional (56). Thus, the elevated levels of Hsp70 seem not to be due to dysfunction of autophagy but activity of a compensatory mechanism.

We have previously shown that reduced protein clearance activates inflammasome signaling in human RPE cells (55, 57) and that iPSC-RPE of AMD origin has a higher baseline inflammasome activity than healthy controls, possibly due to a dysfunctional proteolytic system (58). The current results are consistent with our previous findings. Caspase-1 activity was significantly higher in AMD-RPE than in control-RPE in response to MG-132. Similarly, IL-1β, NLRP3, IL-6, and IL-8 levels tended to be higher in the AMD-RPE (Fig. 4, B, C, F, and G). Conversely, only the control-RPE showed increased activation of caspase-1 after MG-132 exposure, while high baseline activity or a maladapted inflammasome response possibly prevented further inflammasome activation in AMD-RPE.

Interestingly, both inflammasome activation and Hsp70 regulation are influenced by changes in intracellular Ca²⁺ concentration (16, 17, 18, 19, 20), and specifically, it has been shown that mobilizing Ca²⁺ from the intracellular stores plays a role in inflammasome activation (59). Additionally, Hsp70 is phosphorylated in the presence of Ca²⁺ (60), and on the other hand, it affects inflammation by functioning as a negative regulator of the NLRP3 inflammasome (61). Recent work also reveals a connection between autophagy and Ca²⁺ and shows that Ca²⁺ is required for autophagosome formation (19, 20). Our ARPE-19 data (Fig. S2) further supports this link. When evaluating the complex equilibrium of Ca²⁺ in iPSC-RPE, we found significant alterations in the free cytosolic Ca²⁺ concentration as well as variations in ATP-induced calcium responses. AMD cell lines showed lower free cytosolic Ca²⁺ concentration than the control. Interestingly, this observation is coupled with the reduced autophagosome formation in AMD-RPE, being in line with the recent findings in literature (19, 20). The link between Ca²⁺ and inflammasome signaling in the present work is, however, less clear being possibly influenced by the strongly expressed Hsp70 in our AMD-RPE. By analyzing the sensitive ATP-induced Ca²⁺ responses in all cell lines for large cell populations at the single cell level, we found that both the maximum amplitude and the duration of the response above 50% threshold value were decreased in AMD-RPE. However, as the absolute differences were small (around 5%), the result might not reach physiological relevance. Yet recently, Sharma et al. (42) suggested that intracellular Ca²⁺ homeostasis pathways have a central role in mediating the defects leading to AMD disease phenotypes at cellular level. Remarkably, they detected reduced ATP-induced Ca²⁺ responses in iPSC-RPE cells following complement activation (42). Our study indicates that Ca²⁺ homeostasis might be impaired in AMD-RPE even prior to age-induced defects.

Intracellular Ca²⁺ concentration and its dynamics need to be carefully tuned for proper cell functioning (5, 62). In RPE, powerful Ca²⁺ concentration regulators are the different Ca²⁺ channels on the plasma membrane and the membranes of intracellular organelles (5). Specifically, TRP and L-type voltage-gated Ca²⁺ channels affect the intracellular Ca²⁺ concentration in RPE at the resting membrane potential (45, 46). Here we focused on the presence and functionality of voltage-gated Ca²⁺ channels, as these channels are considered to have important roles in the function and viability of the RPE, and they have already been linked to other degenerative disorder of the retina (32). Our patch clamp studies revealed a significantly reduced maximum current and current density through the L-type Ca²⁺ channels for AMD-RPE. When determining the localization of the primary L-type Ca²⁺ channel in RPE, Ca_V1.3, we found that Ca_V1.3 was localized more in the intracellular compartments than the cell membrane in AMD-RPE compared to control-RPE. These observations could be related to ubiquitin and its possible role in AMD pathophysiology, as it is known that ubiquitination regulates the L-type Ca²⁺ channel expression and localization on the cell membrane (63). Indeed, we found that the expression levels of ubiquitinated proteins as well as the levels of the protein aggregation marker SQSTM1 were significantly lower in AMD-RPE after a 24 h MG-123 treatment. After 48 h, the levels for both factors reached the level of control-RPE, except the level of ubiquitinated proteins for the AMD2-RPE. Furthermore, it is worth noting that the possible occurrence of alternative splicing variants of Ca_V1.3 channel could contribute to the changes in channel ubiquitination and, thus, functionality (63, 64). Another interesting candidate affecting Ca²⁺ homeostasis in AMD-RPE and the observed shift in Ca_V1.3 localization is ezrin with its more heterogenous and patchy localization at the apical microvilli in AMD cell lines. Ezrin has been suggested to regulate Ca²⁺ homeostasis in kidney in vivo, and its knockdown shifted the localization of TRPV6 Ca²⁺ channels from the apical membrane of mouse duodenum more intracellularly (65).

The role of Ca_v1.3 in the pathophysiology of AMD is largely unexplored, and, to our current knowledge, only Rosenthal et al. (31) have studied the voltage-dependent Ca²⁺ currents and Ca_v1.3 specifically, in relation to AMD. In their recordings from choroidal neovascular tissue from AMD patients, majority of the cells lacked voltage-dependent currents; yet when such currents were detected, they showed unusually large amplitudes and fast kinetics (31). This finding, together with our observations, emphasizes the need for more studies to resolve the role of voltage-dependent Ca²⁺ channels in AMD pathophysiology.

In conclusion, this study reveals mislocalization of the L-type Ca²⁺ channel Ca_V1.3 in RPE derived from two AMD patients. The shift of Ca_V1.3 from cell membrane to intracellular membranes is accompanied with smaller currents through Ca_V1.3 channels in whole cell patch clamp recordings as well as lower intracellular free Ca²⁺ concentration. Disturbed Ca²⁺ dynamics in AMD-RPE could have an effect on its autophagic machinery and inflammatory response. In fact, AMD-RPE showed dysregulated autophagy and inflammasome activation under proteasome inhibition. To proteotoxic stress, the cells instead responded by increasing the level of Hsp70, possibly as a compensatory mechanism. However, Hsp70, as well as Ca²⁺ itself, have the ability to alter autophagy and inflammasome signaling, disturbing the cellular homeostasis further. Interestingly, AMD cell lines showed signs of chronically activated inflammation at a basal state. Thus, changes in Ca²⁺ homeostasis are likely to affect the key molecular pathways associated with autophagy and inflammasomes in AMD pathology.

Experimental procedures

Donors with clinical phenotype

Patients were diagnosed in the Department of Ophthalmology of Kuopio University Hospital, Finland. Clinical examination included best-corrected visual acuity testing, biomicroscopy analysis, fundus photographs (Canon CX-1 Hybrid Retinal Camera, Canon), fluorescein angiography (Canon CX-1), and/or optical coherence tomography (SPECTRALIS OCT2, Heidelberg Engineering). The Ethics Committee of the Kuopio University Hospital (42/2014) approved the study, and the tenets of the Declaration of Helsinki were followed. All patients signed an informed consent form. As shown in Table 1, clinical examination of the patient AMD1 showed wet AMD and geographic atrophy lesions in both eyes. The patient was an active smoker and under blood pressure and anti-coagulant medications. The moderate visual impairment level was examined in the right eye, while the left eye was clinically blind due to disease. The patient AMD2 had wet AMD in the right eye and dry AMD in the left eye. GA lesions were also seen in both eyes. The right eye was blind, and in the left eye, severe visual impairment was examined. The AMD2 patient had a history of smoking and was under blood pressure medication. Both AMD patients had family history as an aggravating risk factor, and importantly, they were brothers to each other. The healthy age- and gender-matched control did not show any signs of AMD under clinical examination, and his family history did not predispose to AMD. However, the healthy control subject received blood pressure and anti-aggregation treatment.

Establishment and characterization of iPSC lines

Peripheral blood samples were collected from two AMD patients and a healthy control subject. Peripheral blood mononuclear cells isolated from the blood samples were cultured according to standard procedures as described in the manual of CytoTune-iPS 2.0 Sendai Reprogramming Kit (Thermo Fisher Scientific). The conversion of peripheral blood mononuclear cells to iPSCs was purchased from Glykos Finland Ltd and conducted according to the manufacturer’s instructions. Briefly, emerging colonies were selected at day 22 to 24. To remove the vectors from cells, colonies were individually further passaged. Reprogrammed iPSCs were cultured without feeder cells on laminin-521 (LN521, BioLamina)-coated culture plates in the Nutristem XF/FF-medium (Thermo Fisher Scientific). For long-term storage, cells were collected and frozen in 10% DMSO+Nutristem+10 μM ROCK inhibitor Y-27632 (Calbiochem) freezing buffer.

For the generation of a research-grade working cell bank of selected clones from the individual donors, cells were cultured under feeder-free conditions as previously described by Hongisto et al. (66) using Corning CellBIND 24-well culture plates (Corning) coated with human recombinant laminin-521 (BioLamina) and Essential 8 Flex Medium (Thermo Fisher Scientific) supplemented with Essential 8 Flex supplement (50 × ) and 50 U/ml penicillin-streptomycin (Gibco, Thermo Fisher Scientific). The iPSC pluripotency was verified by observing their spontaneous differentiation as embryoid bodies (EB), followed by immunofluorescence labeling for derivative cells of the three embryonic germ layers (endoderm/SOX17, mesoderm/SMA, and ectoderm/OTX2), and the quality of iPSCs was continuously monitored for attachment, growth, and morphology using a Nikon Eclipse TE2000-S phase contrast microscope (Nikon Instruments Europe B.V.), flow cytometry analyses of marker expression (SSEA-4, TRA-1-60), normal karyotype (Karyolite Bobs, PerkinElmer), and cells were further tested for mycoplasma negativity with VenorGeM Classic (Minerva biolabs) as described previously (66).

RPE differentiation and characterizations

For the RPE differentiations and characterizations, one iPSC clone from the donor with moderate/blindness visual impairment (AMD1), two clones from the donor with severe/blindness visual impairment (AMD2_{(1) & (2)}), and one from the healthy age- and gender-matched donor (control) were used (Table 1).

For the differentiation, the iPSCs were cultured on laminin-521 matrix (BioLamina) in Essential 8 Flex Medium (Thermo Fisher Scientific) at +37 °C in a cell incubator providing a humidified atmosphere enriched with 5% CO₂, as described (66). For the RPE differentiation, iPSCs were detached to single cell suspension with TrypLE Select Enzyme (TrypLE, Thermo Fisher Scientific) and transferred to ultra-low attachment plates (Corning Inc) in KnockOut Dulbecco’s modified Eagle’s Medium (KO-DMEM) supplemented with 15% KnockOut Serum Replacement, 2 mM GlutaMAX, 0.1 mM 2-mercaptoethanol (all from Thermo Fisher Scientific), 1% non-essential amino acids, and 50 U/ml penicillin-streptomycin (both from Lonza). EB formation was induced with overnight induction with 10 μM blebbistatin (Sigma-Aldrich). For the following 2 days, EBs were allowed to undergo spontaneous differentiation, and on day 4, the EBs were plated down to 0.75 μg/cm² LN521 matrix and 10 μg/cm² human placental collagen type IV (col IV; Sigma-Aldrich) coating. The medium was thereafter changed three to four times a week. After 30 to 45 days of differentiation, pigmented foci were selected with a scalpel, dissociated with TrypLE, and replated to similarly coated culture wells (RPE passage 1). Forty five days later, the iPSC-RPE was again replated (RPE passage 2), and 9 days later, the iPSC-RPE cells were frozen. For the iPSC-RPE characterizations and other downstream experiments, cells were thawed (passage 3) and plated to Matrigel-coated (44 μg/cm², Corning) polyethylene terephthalate (PET) inserts with 1 μm pore size (Millipore) at a density of 7 × 10⁴ cells/insert and cultured for 8 to 10 weeks. The cells were characterized for mature RPE phenotype as described previously (66). Briefly, key protein expression and localization in RPE was verified with immunofluorescence labeling, growth factor secretion capacity with enzyme-linked immunosorbent assay, cell-cell tight junction maturity with TEER, and phagocytotic functionality with phagocytosis assay. These methods are described separately in the following text.

Immortalized RPE cell culture

The human RPE cell line ARPE-19 was acquired from the American Type Culture Collection. For routine culture, cells were kept in DMEM/F-12 growth medium (1:1, Life Technologies) supplemented with 10% fetal bovine serum (FBS, Global Life Sciences Solutions), 100 U/ml penicillin, 100 μg/ml streptomycin (both from Lonza), and 2 mM L-glutamine (Life Technologies). Cells were kept in an incubator providing a humidified atmosphere at 5% CO₂ and a temperature of +37 °C. Routine passaging was performed every 3 to 4 days, using 0,25% Trypsin-EDTA (Life Technologies). Cells of passage numbers 25 to 37 were used for all experiments.

Immunofluorescence-labeling assays

For immunofluorescence labeling, the iPSC-RPE cells cultured on permeable cell culture inserts were washed three times with Dulbecco's phosphate-buffered saline (DPBS, Lonza) and fixed for 15 min with 4% paraformaldehyde (pH 7.4, Sigma-Aldrich) at room temperature (RT) following repeated washings with DPBS. The cells were permeabilized with 0.1% Triton X-100 in DPBS (Sigma-Aldrich) at RT for 15 min. Unspecific binding sites were blocked with 3% bovine serum albumin (BSA) in DPBS (BSA; Sigma-Aldrich) for 1 h at RT. Samples were incubated with primary antibodies diluted in 3% BSA overnight at +4 °C or 1 h at RT.

The following antibodies were used: CRALBP (1:200, ab15051, Abcam), ezrin (1:200, PAB7060, Abnova), Na⁺/K⁺-ATPase (1:200, ab7671, Abcam), MERTK (1:50, H00010461-M01, Abnova), claudin-19 (1:100, MAB6970, R&D), ZO-1 (1:200, 61-7300, Thermo Fisher Scientific), opsin (1:200, Sigma-Aldrich), and Ca_V1.3 (1:100, ACC-005, Alomone Labs).

The cells were washed three times with DPBS and labeled with secondary antibodies (1:200 in 3% BSA) for 1 h at RT: Alexa Fluor 568–conjugated donkey anti-rabbit IgG (A10042) and donkey anti-goat IgG (A-11057), and Alexa Fluor 488–conjugated donkey anti-mouse IgG (A-21202) and donkey anti-rabbit IgG (A-21206, all from Thermo Fisher Scientific). Phalloidin–Tetramethylrhodamine B isothiocyanate (1:1000, P1951, Sigma-Aldrich) or Atto 633 (1:100, 68825, Sigma-Aldrich) were used for labeling filamentous actin. Cells were washed three times with DPBS, and for some samples, nuclei were counterstained with DAPI included in ProLong Gold Antifade Mountant with DAPI (P36931, Thermo Fisher Scientific) and/or Hoechst (1:3000 in DBPS, 33342, Thermo Fisher Scientific) followed by mounting.

Imaging of the localization of Ca_V1.3 was performed using confocal microscopy with Zeiss LSM780 laser scanning microscopy (LSCM) on inverted Zeiss Cell Observer microscope (Zeiss) using Apochromat 40x/1.4 oil immersion objective. 1024 × 1024 pixel (70.8 × 70.8 μm) stacks of 130 slices were acquired with line average of 2. Imaging of phagocytosis assay was performed using the same Zeiss LSM780 LSCM with Plan-Apochromat 63×/1.4 oil immersion objective. 1024 × 1024 pixel (210 × 210 μm) stacks of 95 slices were acquired with line average of 2. Alexa Fluor 488 was excited with 488 nm from Argon laser, Alexa Fluor 568 with 562 nm from InTune laser, Phalloidin–Tetramethylrhodamine B isothiocyanate, and Atto 633 with 633 nm from HeNe laser. Laser powers were minimized to avoid bleaching.

To quantify the difference in Ca_V1.3 channel localization, the z-fluorescence profiles of Ca_V1.3 and actin filaments (phalloidin) were plotted from the stacks using ImageJ. The intensity profiles for phalloidin and Ca_V1.3 showed a major peak near the apical membrane. The secondary peaks near the basal membrane were significantly reduced due to high pigmentation in all three RPE cell lines. The peak-to-peak distance between Ca_V1.3 intensity peak and the major phalloidin peak (apical membrane) was determined for each cell line.

Other immunostained samples were imaged by confocal microscopy with Zeiss LSM800 LSCM using inverted Plan-Apochromat 63×/1.40 Oil objective. 1550 × 1550 pixel (101.4 × 101.4 μm) stacks of 80 slices were acquired using ZEN 2.3 Software (blue edition, Zeiss). Image brightness and contrast were adjusted linearly avoiding oversaturation with ImageJ. Final figures were assembled using Illustrator 10 (Adobe Systems).

Transepithelial electrical resistance

TEER was measured with Millicell electrical resistance system volt-ohm meter (Merck Millipore) from cell culture inserts. Before measuring, the inserts were at RT for 15 min. The TEER value (Ω) was measured twice from every insert and average values were used for final values. The final TEER value (Ω∗cm²) was calculated by subtracting the TEER value of empty inserts in culture medium and by multiplying with the surface area of the insert (0.3 cm²).

In vitro phagocytosis assay

For phagocytosis assay, POS particles were isolated from porcine eyes which were obtained from a local slaughterhouse. The eyes were hemisected, and retinas detached from the underlying RPE and sclera under dim red light. As described previously in more detail (67, 68), retinal cell layers were disrupted, and POS fragmented by shaking the retinas in 73 mM sucrose phosphate buffer followed by filtering and separation in sucrose gradient using an ultracentrifuge (Optima ultracentrifuge, Beckman Coulter, Inc) at 112,400g for 1 h at +4 °C. POS layer was collected and pellets obtained by centrifugation (3000g for 10 min at +4 °C), and resuspended samples stored in 73 mM sucrose phosphate buffer at −80 °C. These POS particles were fed to the iPSC-RPE cells in KO-DMEM medium supplemented with 10% FBS. POS particles were fed for 1 h at +37 °C and then continued to pulse-chase assay. At the pulse-chase assay, the medium was changed to KO-DMEM 10% FBS without POS particles and incubated for 2 h at +37 °C. The iPSC-RPE cells were fixed and stained using the above-described immunofluorescence labeling method. Internalized POS particles were quantified as previously described (69) using ImageJ. Briefly, confocal microscope (Zeiss LSM780 LSCM) was used to obtain five 512 × 512 pixel Z-stacks with slices acquired in 100 nm interval. ImageJ was used to reslice the Z-stacks to 512 xz-slices. Internalized POS particles were manually calculated from each xz-projections consisting of 20 xz-slices.

Calcium imaging

Intracellular Ca²⁺ dynamics in iPSC-RPE cells was followed using the Ca²⁺ -sensitive dye Fluo-4 AM (Molecular Probes, Thermo Fischer Scientific) as previously described (70). Briefly, samples were loaded with 4 μM Fluo-4 AM in Elliot buffer (pH 7.4, 330 mOsm, containing in mM 137 NaCl, 5 KCl, 0.44 KH₂PO₄, 20 Hepes, 4.2 NaHCO₃, 5 glucose, 1.2 MgCl₂, and 2 CaCl₂) for 45 min at RT. During imaging, samples were perfused at RT with Elliot buffer or 100 μM ATP (Sigma-Aldrich) in Elliot buffer using a gravity-fed solution exchange system (AutoMate Scientific). Imaging was performed with Nikon CrestOptics spinning disk upright fluorescence microscope using an APO LWD 25×/1.10 water immersion objective. 1024 × 1024 images were detected with Andor Live 888EMCDD and captured every 500 ms with no binning and exposure time of 50 ms using Nikon Nis Elements Imaging Software (version 5.02, Nikon). Laser powers were minimized to avoid bleaching.

Spontaneous Ca²⁺ signaling was recorded for 2 min, and the average fluorescence intensities were analyzed for the duration of the recordings to compare the average free cytosolic Ca²⁺ between the cell lines using ImageJ. Additionally, 10 min ATP-induced Ca²⁺ response recordings were performed. One 10 min recording included 2 min of baseline imaging, after which the cells were challenged with ATP for 2 min and let to recover for 6 min. For the data analysis of ATP-induced calcium signaling, three 200 × 200 pixel (104 × 104 μm) regions were selected from every imaged area and 70 individual cells were outlined using ImageJ. The average intensity value as a function of time was extracted from each cell and further analyzed using a self-developed MATLAB script (70) (MATLAB R2017b, The MathWorks Inc). The determined parameters were maximum relative response amplitude and response duration between 50% threshold values.

Cell treatments

To study cell clearance–related mechanisms, mature iPSC-RPE cells were treated with proteasome inhibitor MG-132 (5 μM, Calbiochem) diluted in growth medium for 24 h and 48 h. This was followed by a recovery period of 24 h with the removal of proteasome inhibitor by washing the cells twice with growth medium and the continuation of cultivation in the growth medium. Additionally, autophagy inhibitor bafilomycin A1 (Sigma-Aldrich) was added directly to the growth medium at a final concentration of 150 nM with and without MG-132 after 21 h from the start of the treatments for 3 h. After desired treatment times, the cells were washed once with PBS and the proteins were collected to 150 μl of 2x Laemmli Sample Buffer (Bio-Rad Laboratories) containing 5% of β-mercaptoethanol (Sigma-Aldrich). The cells were lysed for 5 min at RT before collection and stored at −70 °C.

For studies measuring inflammasome activation and the release of pro-inflammatory cytokines, mature iPSC-RPE cells cultured on Matrigel-coated inserts were washed once in IL-1α containing culture medium, before being stimulated with IL-1α (4 ng/ml, R&D systems) from both the apical and basal sides. After 24 h, cells were exposed to 5 μM MG-132, which served as the activation signal for inflammasomes, from both the apical and basal sides. Cells were incubated for a further 48 h (ELISA, LDH measurements) or for 6 h (caspase-1 activity assay).

For the studies performed in immortalized human RPE cells, ARPE-19 cells were split to 12-well plates at a density of 200,000 cells/well. Following a 72-h incubation to assure full confluency, cells were exposed to 4 ng/ml IL-1α (R&D Systems). Another 24 h later, 0.5 μM thapsigargin (Sigma-Aldrich) either alone or in combination with 1 mM AICAR (1 mM, Toronto Research Chemicals Inc) was added to the wells. Medium samples were collected to clean prelabeled tubes and centrifuged at 2000 rpm for 10 min before being stored at −20 °C until analysis. Cell lysates were collected following a wash with PBS followed by lysis with mammalian protein extraction reagent M-PER (Thermo Fisher Scientific). Samples were centrifuged at 13,000 rpm for 10 min and stored at −80 °C until analysis by Western blotting.

Western blot

The iPSC-RPE lysates were thawed and centrifuged at 12,000g for 20 min at RT. The protein concentration of ARPE-19 cell lysates was measured using the Bradford protein assay. Equal volumes (20 μl, iPSC-RPE) or equal protein amounts (ARPE-19) of the supernatants were run into 15% SDS-PAGE gels and transferred onto nitrocellulose membranes (10600003, GE Healthcare) in an overnight wet-blot. The molecular weight of proteins was estimated using Precision Plus Protein Dual Color Standards (1610374, Bio-Rad). Ponceau S (P7170, Sigma-Aldrich) staining was performed on the membranes to ensure the quality of the protein transfer. The membranes were cut in half above 25 kDa bands and blocked in 3% milk, 0.3% Tween 20 (822184, MilliporeSigma)–PBS (T-PBS) (upper half) or 3% milk, 0.1% Tween 20–tris buffered saline (TBS, T-TBS) (lower half) solutions for 1 or 2 h (upper and lower halves, respectively). The membranes containing ARPE-19–derived proteins were blocked in 3% milk in 0.3% T-PBS solution for 1.5 h in RT. The upper halves of the membranes were then treated with primary antibodies for Hsp70 (1:5000 in 0.5% BSA, 0.3% T-PBS; C92F3A-5, Enzo Life Sciences) for 1 h at RT, tubulin (1:8000 in 1% milk, 0.05% T-PBS; T5168, Sigma-Aldrich) for 0.5 h at RT, SQSTM1/p62 (1:4000 in 5% BSA, 0.1% T-TBS; 5114S, Cell Signaling Technology) overnight at +4 °C, or ubiquitin (1:1000 in 0.5% BSA, 0.3% T-PBS; sc-8017, Santa Cruz Biotechnology) for 1 h at RT, after which they were washed for 3 × 5 min with their respective washing buffers. The lower halves were treated with LC3 primary antibody (1:1000 in 5% BSA, 0.1% T-TBS; 3868S, Cell Signaling Technology) overnight at +4 °C and washed for 3 × 5 min. Uncut membranes containing the ARPE-19 proteins were probed for SQSTM1, LC3 (conditions as above for both) and GAPDH (1:15,000 in 0.1% T-PBS, ab8245, Abcam) for 1 h at RT. Horseradish peroxidase–conjugated anti-mouse (NA931, MilliporeSigma) or anti-rabbit (A16104, Thermo Fisher Scientific) IgG secondary antibodies were then applied: anti-mouse 1:20,000 in 3% milk, 0.3% T-PBS (Hsp70) for 1 h; anti-mouse 1:10,000 in 1% milk, 0.05% T-PBS (tubulin) for 0.5 h; anti-rabbit 1:10,000 in 5% milk, 0.1% T-TBS (SQSTM1/p62) for 1 h; anti-mouse 1:10,000 in 3% milk, 0.3% T-PBS (ubiquitin) for 1 h; anti-rabbit 1:10,000 in 3% milk, 0.1% T-TBS (LC3) for 2 h; or anti-mouse 1:12,000 in 0.1% T-PBS (GAPDH) for 1 h, all at RT. The membranes were washed again as before, after which they were treated with Immobilon Western Chemiluminescent HRP Substrate (WBKLS0100, MilliporeSigma) for 5 min. The protein bands were detected using ImageQuant RT ECL Imager (GE Healthcare), and the results were quantified using the ImageJ program (U.S. National Institutes of Health; http://imagej.nih.gov/ij/). The quantified protein levels were normalized against α-tubulin, and the treated samples were compared against nontreated controls resulting in relative protein levels (fold change against control).

LDH measurement

The levels of leaked LDH in cell culture medium either on the basal or apical side was determined after 48 h using the CytoTox 96 Non-Radioactive Cytotoxicity Assay (G1780, Promega). Briefly, 50 μl of medium was incubated with 50 μl of the kit’s assay reagent for 30 min protected from light after which another 50 μl of stop solution was added. The absorbance values of wells were measured at the wavelength of 490 nm. Fresh culture medium served as a blank and all values were normalized to IL-1α–stimulated control.

Enzyme-linked immunosorbent assays

PEDF secretion was measured using ELISA with a commercially available DuoSet ELISA Ancillary Reagent Kit 2 (DY008, R&D Systems) and antibody Human Serpin F1/PEDF DuoSet ELISA kit (DY1177-05, R&D Systems) according to the manufacturer’s instructions. Medium for this analysis was collected separately from both apical and basolateral sides of the insert after 24 to 25 h of incubation.

For the determination of secreted cytokines (IL-1β, IL-6, and IL-8), medium samples were collected from both the basal and apical sides and were analyzed with BD OptEIA human ELISA kits following the manufacturer's instructions (IL-1β; 557953, IL-6; 555220, IL-8; 555244, BD Biosciences) with some modifications (37). The levels of secreted NLRP3 were measured using Human NACHT, LRR, and PYD domains-containing protein 3 (NLRP3/C1orf7/CIAS1/NALP3/PYPAF1) ELISA kit, according to the manufacturer's instructions with a modified reference wavelength of 620 nm (CSB-E15885h, CusaBio).

Determination of caspase-1 activity

Caspase-1 activity was determined in iPSC-RPE cells cultured on permeable cell culture inserts after staining with the commercial FAM-FLICA Caspase-1 (YVAD) Assay Kit (Catalog 98, ImmunoChemistry Technologies). Briefly, cells were washed with culture medium before being stained with the FLICA probe for 1 h at +37 °C. Cells were washed 3 times for 10 min each using the kit’s own wash buffer before being stained with Hoechst 3342 (received from the kit, 1 μg/ml) for 20 min at +37 °C. Cells were fixed in 4% paraformaldehyde for 15 min at RT, washed three times, and mounted under a glass cover slide for examination under a fluorescence microscope (Zeiss Axio Imager M2) using Plan-Apochromat 63x/1.30 Oil DIC objective, AxioCamMR3 camera, and Zen 2012 Software (blue edition, Zeiss). At least five images (size 1388 × 1040 pixels) per insert were taken. Two researchers quantified the number of FLICA-positive cells and the total number of nuclei per image in an independent and blinded manner. The ratio between FLICA-positive cells and total number of cells per image were then calculated and presented in a bar chart diagram. Image brightness and contrast were adjusted linearly to avoid oversaturation using ImageJ.

Patch clamp recordings

Ionic currents were recorded from cells intact in iPSC-RPE monolayers with the standard patch clamp method using whole cell configuration. During the measurements, the monolayer was perfused with extracellular solution (NaCl 120 mM, TeaCl 5 mM, Hepes 10 mM, MgCl₂ 1.2 mM, CaCl₂ 1.1 mM, glucose 5 mM, BaCl₂ 10 mM, adjusted to pH ∼7.4, and osmolarity ∼305 mOsm (Gonotec, Osmomat 030, Labo Line Oy)) at RT. Patch pipettes with a resistance of 6 to 9 MΩ were filled with Na⁺-free Cs intracellular solution (CsCH₃SO₃ 83 mM, CsCl 25 mM, TeaCl 5 mM, EGTA 5.5 mM, CaCl₂ 0.5 mM, ATP-Mg 4 mM, Hepes 10 mM with pH ∼7.2, and osmolarity ∼271 mOsm). The pharmacological confirmation of L-type Ca²⁺ channels was performed by adding L-type Ca²⁺ channel inhibitor nifedipine 10 μM (Sigma-Aldrich) to the extracellular solution in perfusion. The effect of inhibition was recorded after 5 to 10 min of drug application. All the recordings were established in voltage clamp mode with pClamp 10.2 software (Molecular Devices) using the Axopatch 200B patch clamp amplifier connected to an acquisition computer via AD/DA Digidata 1440 (Molecular Devices). During the recordings, the access resistance was below 30 MΩ and membrane resistance above 300 MΩ. In data analysis, holding potentials were corrected for a −6.6 mV liquid junction potential. Current–voltage relationships (I-V curves) were obtained from the maximum value of the current at the plateau phase of the response at given voltage using Clampfit software (Molecular Devices). Maximal current, peak current voltage, and half-maximal activation potential were determined from the I-V curves using Excel (Microsoft) and Origin (OriginLab) software. The averaging was performed in Excel and statistical analysis in SPSS (IBM Corp.). Plotting the data was performed using Origin.

Statistical analysis

The results are presented as mean ± SD. Mann-Whitney U test was used to determine the statistical significance between the healthy control and the AMD cell lines. Statistical difference was concluded if results showed a p-value < 0.05.

Data availability

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Supporting information

This article contains supporting information.

Conflict of interest

The authors declare that they have no conflicts of interest with the contents of this article.

Reviewed by members of the JBC Editorial Board. Edited by Kirill Martemyanov