Aquaporin Expression and Function in Human Pluripotent Stem Cell–Derived Retinal Pigmented Epithelial Cells

Abstract

Purpose.: Aquaporins (AQPs), a family of transmembrane water channel proteins, are essential for allowing passive water transport through retinal pigmented epithelial (RPE) cells. Even though human native RPE cells and immortalized human RPEs have been shown to express AQPs, the expression of AQPs during the differentiation in stem cell–derived RPE remains to be elucidated.

Methods.: In human embryonic (hESCs) and induced pluripotent stem cells (hiPSCs)–derived RPE cells, the expression of several AQPs was determined by quantitative real-time PCR and the localization of AQP1 was assessed with confocal microscopy. The functionality of AQP water channels was determined by cell volume assay in hESC-derived RPE cells.

Results.: AQP1, AQP3, AQP4, AQP5, AQP6, AQP7, AQP10, AQP11, and AQP12 were expressed in hESC- and hiPSC-derived RPE cells. Furthermore, the expression of AQP1 and AQP11 genes were significantly upregulated during the maturation of both hESC and iPSC into RPE. Confocal microscopy shows the expression of AQP1 at the apical plasma membrane of polarized cobblestone hESC- and hiPSC-derived RPE cells. Lastly, aquaporin inhibitors significantly reduced AQP functionality in hESC-RPE cells.

Conclusions.: hESC-RPE and hiPSC-RPE cells express several AQP genes, which are functional in mature hESC-derived RPE cells. The localization of AQP1 on the apical plasma membrane in mature RPE cells derived from both hESC and hiPSC suggests its functionality. These data propose that hESC- and hiPSC-derived RPE cells, grown and differentiated under serum-free conditions, resemble their native counterpart in the human eye.

Introduction

The outer blood–retinal barrier (BRB) is formed by the tight junctions between retinal pigmented epithelial (RPE) cells. ¹ This outer BRB forms a barrier that impedes in transcellular movement of ions and solutes between choroidal capillaries and neural retina. The major ionic imbalance existing between the apical and basal sides of the outer BRB is maintained by ion pumps. ² Under physiologic conditions, the presence of this transepithelial osmotic gradient leads to passive fluid absorption by RPE cells, ³ likely through aquaporins (AQPs). AQPs represent a family of transmembrane water channel proteins that are widely expressed in both animal and plant kingdoms. ⁴ To be functional, AQPs need to assemble into tetramers. ⁵ In mammals, 13 AQPs have been identified. ⁴ Based on their permeability, AQPs can be subdivided into: classical AQPs (AQP0, AQP1, AQP2, AQP4, AQP5, AQP6, and AQP8) that are permeable to water; aquaglyceroporins (AQP3, AQP7, AQP9, and AQP10) that are permeable to small solutes (e.g., glycerol), in addition to water; and unorthodox AQPs (AQP11 and AQP12) with unusual structure. ⁶  

Human native RPE cells were first reported to express AQP1. ⁷ However, more recently, human native RPE cells have been shown to express several other AQPs, including AQP0, AQP1, AQP3, AQP4, AQP5, AQP6, AQP7, AQP8, AQP9, AQP10, AQP11, and AQP12, but not AQP2. ^8,9 In addition, in a microarray study AQP3 was shown to be expressed by human stem cell–derived mature RPE cells but not by an immortalized RPE cell line. ¹⁰ In a similar microarray survey mature stem cell–derived RPE cells were shown to also express AQP1. ¹¹ However, aquaporin expression was not confirmed outside the microarray data by qRT-PCR or immunofluorescence or Western blot. 

In their mature form, RPE cells are highly polarized, and their integrity is critical for the maintenance of the functions of the outer BRB. In several retinal diseases the dysfunction of the BRB leads to an alteration of the fluid reabsorption by RPE cells. The expression and/or function of AQPs have been shown to be modified under diabetic retinopathy, ¹² retinal ischemia, ⁸ autoimmune uveitis, ¹³ and UV radiation and oxidative stress of the retina. ¹⁴  

Human pluripotent stem cells, including the embryonic stem cells (hESCs) and induced pluripotent stem cells (hiPSCs), open new horizons for transplantation therapies ¹⁵ and in vitro drug testing platform. ¹⁶ Although hESC-derived RPE cells have been shown to resemble the native RPE counterpart in the human eye, ^10,17 and mature hESC-derived RPE to express AQP1 ¹¹ and AQP3, ¹⁰ the expression of all AQPs during RPE maturation and their function remain to be elucidated. 

Materials and Methods

Cell Lines

The pluripotent hESC line Regea08/017 was derived and characterized in our laboratory. ¹⁸ The pluripotent FiPS 5-7 generated from human fibroblasts ^18,19 was obtained from Professor Timo Otonkoski (University of Helsinki, Finland). These pluripotent cell lines were grown on top of mitotically inactivated (γ-irradiated with 40 Gy) human foreskin fibroblast (hFF) cells (36,500 cells/cm², CRL-2429; American Type Culture Collection, Manassas, VA; Fig. 1G) at 37°C in 5% CO₂ and in a culture medium made of Knock-Out Dulbecco's modified Eagle's medium (KO-DMEM), 20% Knock-Out serum replacement (KO-SR), 2 mM GlutaMax, 0.1 mM 2-mercaptoethanol (all from Life Technologies, Carlsbad, CA), 1% Minimum Essential Medium nonessential amino acids, 8 ng/mL human basic fibroblast growth factor (bFGF) (R&D Systems, Inc., Minneapolis, MN), and 50 U/mL penicillin/streptomycin (both from Cambrex BioScience, Walkersville, MD). The culture medium was replenished six times a week. The undifferentiated areas of pluripotent hESC colonies (Fig. 1A) and hiPSCs (Fig. 1D) were selected manually and passaged once in a week on new hFFs. Only the undifferentiated areas of 1-week-cultured stem cell colonies, here named undifferentiated hESC (Fig. 1A) and undifferentiated hiPSC (Fig. 1D) were used for analyses. 

Figure 1

 

Morphology and gene expression of cell samples. Brightfield micrographs of cell cultures showing the representative morphology of (A) undifferentiated hESC (Regea08/017), (B) fusiform hESC-RPE, (C) cobblestone hESC-RPE, (D) undifferentiated hiPSC (iPSC 5–7), (E) fusiform hIPCS-RPE, (F) cobblestone hIPCS-RPE, and (G) hFF. Scale bars: 100 μm. (H) Gene expression of 1: undifferentiated hESC, 2: fusiform hESC-RPE, 3: cobblestone hESC-RPE, 4: undifferentiated hiPSC, 5: fusiform hiPSC-RPE, 6: cobblestone hiPSC-RPE, 7: hFF, and 8: water (i.e., negative control). From the studied genes only GAPDH, MITF, and PAX6 were expressed in hFF.

The manually selected areas of 1-week-cultured undifferentiated hESCs and hiPSCs were induced to differentiate into RPE cells in floating cell aggregates by reducing the KO-SR concentration to 15% and removing the bFGF, as previously described. ¹⁷ This culture condition was maintained for 87 to 202 days (Table 1). The culture medium was replenished thrice in a week. 

Table 1. 

 

Culture Periods of Cell Samples

Cell Type	Gene Expression (PCR)		Protein Localization (IF)		Functional Assay
	Adherent Culture Passage/d, (SD)	Entire Culture Period, (SD)	Adherent Culture	Entire Culture Period	Adherent Culture	Entire Culture Period
Feeders	d7 (7)	p16, p24, p11	d7 (6–8)	p16, p10
Undifferentiated hESC	d7 (7)	p30, p57, p58, p92	d7 (7)	p16, p66
Fusiform hESC-RPE	d21 (7–76)	d193 (138–293)	d59 (16–65)	d113 (59–167)	d12 (9–15)	d130 (127–133)
Cobblestone hESC-RPE	d84 (71–91)	d247 (164–291)	d69 (61–83)	d235 (160–181)	d259 (118–372)	d359 (154–458)
Undifferentiated hIPSC	d7 (7)	p45, p75, p90	d7 (7)	p100
Fusiform hIPSC	d16 (7–47)	d136 (126–160)	d13	d206	d12 (9–15)	d125 (122–128)
Cobblestone hIPSC	d70 (50–91)	d134 (90–78)	d96 (78–114)	d224 (206–230)	d106 (91–115)	d361 (274–610)

For the experiments, the pigmented areas of floating aggregates were manually dissected, and dissociated with 1× Trypsin-EDTA and seeded on culture inserts (BD Biocoat; BD Biosciences, San Jose, CA). Adherently cultured cells underwent morphologic changes starting from a nonpigmented fusiform morphology (Figs. 1B, 1E) followed by rounding to more pigmented epithelioid cells, and finally developed a typical RPE-like cobblestone morphology (Figs. 1E, 1F). Fusiform and cobblestone cells were harvested for RNA and protein extraction, immunofluorescence labeling, and functional testing (Figs. 1B, 1C, 1E, 1F, 1I). 

RNA Extraction and cDNA Synthesis

Total RNA was extracted (NucleoSpin XS-kit; Macherey-Nagel GmbH & Co., Düren, Germany) according to the manufacturer's instructions. The RNA concentration and the quality were assessed using a spectrophotometer (NanoDrop 1000; NanoDrop Technologies, Wilmington, DE). RNA (250 ng) was reverse-transcribed to complementary DNA (MultiScribe Reverse Transcriptase; Applied Biosystems, Foster City, CA) according to the manufacturer's instructions in the presence of an RNase inhibitor. Complementary DNA was used as a template in a following PCR reaction (carried out using 5 U/μL Taq DNA Polymerase; Fermentas, Thermo Fisher Scientific, Inc., Leicestershire, UK) with 5 μM primers specific for particular genes (Biomers.net GmbH, Söflinger, Germany; Table 2). The PCR reactions were carried out (PCR MasterCycler ep gradient; Eppendorf AG, Hamburg, Germany) as follows: 95°C 3 minutes, 95°C 30 seconds, annealing 30 seconds, 72°C 1 minute, 72°C 5 minutes, for 38 cycles. Annealing temperatures and primer sequences are presented in Table 2. PCR products were analyzed on 2% agarose gels with a 50-bp DNA ladder (MassRulerTM DNA Ladder Mix, Fermentas). The bands were visualized with the Quantity one 4.5.2. Basic program (Bio-Rad Laboratories, Inc., Hercules, CA). hFFs without pluripotent stem cells were used as the control sample for undifferentiated hESC and undifferentiated hiPSC samples to see the residual of hFFs in undifferentiated samples. 

Table 2. 

 

Reverse-Transcriptase–PCR Primer Sequences and Used Annealing Temperatures

Gene	Primer Sequences (5′ > 3′)		Tann
	Forward	Reverse
GAPDH	GTT CGA CAG TCA GCC GCA TC	GGA ATT TGC CAT GGG TGG A	55
OCT 3/4	CGT GAA GCT GGA GAA GGA GAA GCT G	AAG GGC CGC AGC TTA CAC ATG TTC	55
PAX6	AAC AGA CAC AGC CCT CAC AAA CA	CGG GAA CTT GAA CTG GAA CTG AC	60
RAX	CTG AAA GCC AAG GAG CAC ATC	CTC CTG GGA ATG GCC AAG TTT	55
MITF	AAG TCC TGA GCT TGC CAT GT	GGC AGA CCT TGG TTT CCA TA	52
RPE65	TCC CCA ATA CAA CTG CCA CT	CAC CACC ACA CTC AGA ACT A	52
Bestrophin	GAA TTT GCA GGT GTC CCT GT	ATC AGG AGG ACG AGG AGG AT	60
Tyrosinase	TGC CAA CGA TCC TAT CTT CC	GAC ACA GCA AGC TCA CAA GC	52

Quantitative Real-Time PCR (qPCR)

Primers were designed using National Center for Biotechnology Information RefSeq sequence entry data as previously described. ²⁰ Primer-pair efficiency was tested by performing a qPCR using a cDNA template standard curve starting at 10 ng of control tissue, diluted serially 4-fold down to 0.039 ng (i.e., five serial dilutions). Primers were verified for any potential amplification in the presence of genomic DNA by performing a qPCR in the presence of 2.5 ng genomic DNA. Primer sequences and efficiencies are shown in Table 3. 

Table 3. 

 

Quantitative Real-Time–PCR Primer Sequences and Sequence Lengths

Gene	Primer Sequences for Quantitative Real-Time PCR (5′ > 3′)		bp
	Forward	Reverse
hsATP5B	AGA GGT CCC ATC AAA ACC AAA C	AAA AGC CCA ATT TTG CCA CC	152
B2M	AGA TGA GTA TGC CTG CCG TG	TCA TCC AAT CCA AAT GCG GC	120
hsHPRT1	TGG CGT CGT GAT TAG TGA TG	CTC GAG CAA GAC GTT CAG TC	137
hsAQP1	TGG ACA CCT CCT GGC TAT TG	GGG CCA GGA TGA AGT CGT AG	164
hsAQP2	CAC CCC TGC TCT CTC CAT A	GAA GAC CCA GTG GTC ATC AAA T	139
hsAQP3	GCT GTA TTA TGA TGC AAT CTG GC	TAA GGG AGG CTG TGC CTA TG	152
hsAQP4	GAA GGC ATG AGT GAC AGA CC	ATT CCG CTG TGA CTG CTT TC	130
hsAQP5	GCC ACC TTG TCG GAA TCT AC	TAA AGC ATG GCA GCC AGG AC	148
hsAQP6	CAC CTC ATT GGG ATC CAC TT	GTT GTA GAT CAG TGA GGC CA	147
hsAQP7	ATC TCT GGA GCC CAC ATG AA	GAA GGA GCC CAG GAA CTG	111
hsAQP8	GTG CCT GTC GGT CAT TGA G	CAG GGT TGA AGT GTC CAC C	125
hsAQP9	TCT CTG AGT TCT TGG GCA CG	GGT TGA TGT GAC CAC CAG AG	174
hsAQP10	GAT AGC CAT CTA CGT GGG TG	CAC AGA AAG CAG ACA GCA AC	130
hsAQP11	TCC GAA CCA AGC TTC GTA TC	TAG CGA AAG TGC CAA AGC TG	110
hsAQP12	ACT TGT TCT TCT GGC CGT AG	CTT ACT GGA GTA CGT GCA GG	128

The qPCR reaction setup and plate preparation were standardized and carried out using the Sybergreen technology as previously described. ²⁰ Gene expression stability analysis and matching statistics were performed using commercial software (Biogazelle qbaseplus software; Biogazelle NV, Zwijnaarde, Belgium), containing widely recognized reference gene algorithm GeNorm. ²¹ qPCR data were normalized using three reference genes according to the minimum information for publication of quantitative real-time PCR experimental guidelines. ^22,23 qPCR data were finally expressed as fold stimulation over undifferentiated hESC or undifferentiated hiPSC sample, respectively. 

Confocal Microscopy

Adherent cells were submitted to an immunofluorescent labeling as previously described. ¹⁷ Briefly, the cells were washed three × 5 minutes with PBS, fixed 10 minutes with 4% paraformaldehyde (pH 7.4; Sigma-Aldrich, St. Louis, MO), washed with PBS, permeabilized in 0.1% Triton X-100/PBS (Sigma-Aldrich), for 10 minutes, and washed three times with PBS. Nonspecific binding sites were blocked with 3% BSA (Sigma-Aldrich) in PBS for 1 hour. Primary antibody incubations (Table 4) were carried out in 0.5% BSA-PBS for 1 hour. Thereafter, cells were washed three times with PBS. The secondary antibody (Table 4) incubations were done for 1 hour in 0.5% BSA. Cells were washed three times with PBS. Nuclei were counterstained with DAPI (Vector Laboratories, Inc., Burlingame, CA). The entire labeling procedure was performed at room temperature. Confocal microscopy images were obtained with a confocal microscope (LSM 700; Carl Zeiss, Jena, Germany) using a 63× oil-immersion objective and brightfield images were obtained (Olympus BX60 microscope; Olympus, Tokyo, Japan) with a 60× oil-immersion objective with NA-1. Overlays and image processing of confocal images were done using commercial software (Zen Software, Inc.; Carl Zeiss). 

Table 4. 

 

Antibodies Used in Immunofluorescence Labeling ¹³

Antibody Name	Abbreviation	Primary/ Secondary	Host	Dilution	Manufacturer
Anti-aquaporin-1 antibody	AQP1	Primary	Rabbit	1:400	Custom made13
PAX6	PAX6	Primary	Mouse monoclonal		Atsushi Kawakami, MD
Anti-cellular retinaldehyde-binding protein	CRALBP	Primary	Mouse monoclonal	1:1000	Abcam, Cambridge, UK
ZO-1	ZO-1	Primary	Mouse monoclonal	1:250	Zymed/Invitrogen, Carlsbad, CA
Anti-Na+/K+ ATPase	NaK	Primary	Mouse monoclonal	1:50	Abcam, Cambridge, UK
Anti-rabbit IgG Alexa Fluor 488	—	Secondary	Donkey	1:1500	Molecular Probes, Life Technologies, Paisley, UK
Anti-mouse IgG Alexa Fluor 568	—	Secondary	Donkey	1:1500	Molecular Probes, Life Technologies, Paisley, UK

Cell Volume Assay

The function of AQP proteins was assessed by an intracellular fluorophore dilution method, ²⁴ whereby cells were loaded with calcein-AM, and then the decrease in fluorescence intensity was analyzed during osmotic challenge. Because the fluorescence intensity of calcein-AM is dependent on the ABC-efflux transporters that are expressed by hESC-RPE cells, ¹⁶ we also set up a fluorescence accumulation assay that is independent of any transporters. In this fluorescence accumulation assay, cells were treated with Alexa Fluor 568 nm hydrazide (10 μM; Invitrogen, Carlsbad, CA) either in isoosmotic (280 mOsm) or hypoosmotic (100 mOsm) PBS for 10 minutes at 37°C, and photographed with fluorescence microscope immediately after the test. 

The conventional fluorescence accumulation assay was carried out in parallel with conventional dilution assay where the cells were treated with 2 μM calcein-AM (Calbiochem, La Jolla, CA) in 280 mOsm for 20 minutes at 37°C, and then subjected to isoosmotic (280 mOsm) or hypoosmotic PBS (100 mOsm) treatment for 10 minutes, and photographed with a fluorescence microscope (Olympus IX) immediately after the experiment. The results revealed that dilution and accumulation assays were comparable when using polarized RPE cells (data not shown). Thus the method that was independent of any transporter was selected to be used in this study. 

Cells, grown on culture inserts (BD Biocoat), were cut into four pieces and tightly clamped to a P2307 slider (Physiologic Instruments, San Diego, CA), and assembled to a custom-made Teflon chamber. In fluorescence accumulation assay the cell sheets were exposed for 10 minutes to the hypoosmotic (100 mOsm) treatment in the presence or absence of known aquaporin inhibitors. The used inhibitors were 10 μM HgIICl (Sigma-Aldrich), a nonspecific and toxic AQP inhibitor, ²⁵ 10 mM tetraethyl ammonium (TEA; Sigma-Aldrich) a nontoxic and AQP1-specific inhibitor, ²⁶ and 2 mM acetazolamide (Diamox, Goldhield, Croydon, UK [ACE]) an AQP1-specific cationic anhydrase inhibitor. ²⁷ The samples were photographed immediately after the experiment under a fluorescence microscope (Olympus IX) with 10× lens using the same exposure setting for all samples. The intensity of the microfluorographs was analyzed with ImageJ software (NIH Image software, developed by Wayne Rasband, National Institutes of Health, Bethesda, MD; available at http://rsbweb.nih.gov/ij/index.html). The results are reported as a percentage of the inhibition of fluorescence intensity, where percentage of inhibition was calculated as follows:  

 

 

Transepithelial Resistance

Because cell volume influences the transepithelial resistance (TEER), ²⁸ the TEER of highly polarized hESC-RPE cell sheets was monitored in isoosmotic and hypoosmotic solutions in the absence or presence of AQP inhibitors (10 μM HgIICl, 10 mM TEA, or 2 mM ACE) in the onset (0 minute), in the middle (5 minutes), and at the end (10 minutes) of the experiment using an electrical resistance measuring system (Millicell, volt–ohm meter; Millipore, Billerica, MA). The cell sheets were treated similarly as in the fluorescence accumulation test. TEER values were normalized to the TEER value at time 0 (0 minute). 

Cell Viability Test

Cell viability of mature hESC-derived RPE and hiPSC-derived RPE cells was assessed in the absence or presence of AQP inhibitors (Live/Dead Viability/Cytotoxity kit for Mammalian cells; Invitrogen) as previously described. ¹⁶ Briefly, the cells were rinsed with DPBS and incubated at room temperature for 40 minutes with a mixture of 0.25 μM calcein-AM (green fluorescence) and 0.5 μM Ethidium homodimer-1 (red fluorescence, EthD-1). A fluorescence microscope (Olympus IX) was used to image the viable cells (green fluorescence) and dead cells (red fluorescence) with 10× long working distance objective. 

Statistical Analyses

Statistical analysis of the qRT-PCR data was performed using ANOVA. Bonferroni correction was applied to correct for use of multiple tests. The data from cell volume measurements and the TEER data were analyzed with the Student's t-test (with PASW Statistics, version 18; Hong Kong). 

Ethical Issues

The National Authority for Medicolegal Affairs Finland has approved our research with human embryos (Dnro 1426/32/300/05). We also have a supportive statement from the local ethics committee of the Pirkanmaa hospital district Finland to derive and expand hESC lines from surplus embryos not used in the treatment of infertility by the donating couples, and to use these cell lines for research purposes (R05116). No new cell lines were derived in this study. 

Results

hESC- and hIPSC-Derived RPE Cells Express Eye-Specific Genes

RT-PCR analysis was used to assess the cell differentiation status of both hESC and hIPSC. Pluripotency gene Oct3/4, a marker of undifferentiated hESC and hIPSC, was expressed only by undifferentiated hESC and hIPSC, as expected (Fig. 1H). Neuroectodermal marker gene, PAX6, also a marker of eye-specific lineage, was expressed at all maturation stages (i.e., undifferentiated, fusiform, and cobblestone hESC-derived RPE and hiPSC-derived RPE cells). The eye-specific gene RAX was found to be expressed only by hESC-RPE cells with cobblestone morphology. RPE-specific genes, such as MITF, bestrophin (BEST1), and RPE65, were not expressed by undifferentiated hESC cells. Both fusiform and cobblestone hESC-derived RPE expressed MITF, bestrophin (BEST1), and RPE65. MITF and bestrophin (BEST1), but not RPE65, were expressed by undifferentiated hiPSC. Both fusiform and cobblestone hiPSC-derived RPE expressed MITF and bestrophin (BEST1) at a higher level than that of the undifferentiated hIPSC cells. Only the cobblestone hiPSC-derived RPE cells expressed RPE65. The tyrosinase gene, which is essential for melanin synthesis, was only expressed in both fusiform and cobblestone hESC- and hiPSC-derived RPE, but not in undifferentiated hESC and hiPSC cells. 

Aquaporin Gene Expression During RPE Cell Differentiation

The relative expression of AQPs (AQP1–AQP12) was examined by qRT-PCR in undifferentiated hESC and hiPSC, fusiform and cobblestone hESC and hiPSC, and hFF cells. No expression of AQP2, AQP8, and AQP9 could be detected in any of the studied samples. The AQP1 mRNA expression increased significantly when cells matured from undifferentiated hESC to cobblestone hESC-derived RPE (P = 0.002). Similar results were observed when cells matured from undifferentiated hiPSC to cobblestone hiPSC-derived RPE (P = 0.006), and from fusiform hiPSC-derived RPE to cobblestone hiPSC-derived RPE (P = 0.020) (Fig. 2). The AQP11 mRNA expression increased significantly between undifferentiated hESC and cobblestone hESC-derived RPE (P = 0.001) and between fusiform hESC-derived RPE and cobblestone hESC-derived RPE (P = 0.005), as well as between undifferentiated hiPSC and cobblestone hiPSC-derived RPE (P = 0.003) and between fusiform hiPSC-derived RPE and cobblestone hIPSC-derived RPE (P = 0.046). No statistically significant change in AQP3, AQP4, AQP5, AQP6, AQP7, AQP10, and AQP12 mRNA expression could be detected during hESC and hIPSC maturation into RPE. 

Figure 2

 

Expression of AQP genes. Relative expression of AQP1, AQP3, AQP4, AQP5, AQP6, AQP7, AQP10, AQP11, and AQP12 mRNA. Data are expressed as mean ± SD, *P < 0.05, **P < 0.01, ***P < 0.001. For better visualization, fold-change is represented on a logarithmic scale. SD values of fold-change from experiments are presented as error bars.

Aquaporin-1 Protein Localization Changes During RPE Cell Differentiation

Since AQP1 mRNA expression encountered the most prominent increase during the maturation of RPE cells, we investigated the localization of AQP1 protein by confocal microscopy during RPE maturation. 

The hFFs expressed AQP1 in the cytoplasm, but not at the plasma membrane (Fig. 3A). The undifferentiated hESCs showed intense neuroectodermal marker (PAX6) labeling in the nucleus but no AQP1 staining (Fig. 3B). Some fusiform hESC-derived RPE (Fig. 3C) and some fusiform hiPSC-derived RPE (Fig. 3D) expressed the RPE-specific CRALBP as well as AQP1. In hESC-derived RPE cells with cobblestone morphology AQP1 localized on the apical (Fig. 3E) and upper lateral membranes (Fig. 3F) above the nucleus, but not on the nuclear level (Fig. 3G). In the orthoprojection (Fig. 3H) made from images (Figs. 3E–G), the apical localization of AQP1 is clearly seen. CRALBP that is known to localize on the lateral and apical membranes colocalized partially with AQP1 in polarized hESC-derived RPE (Fig. 3I) and hiPSC-derived RPE (Fig. 3K) cells. Na⁺/K⁺ ATPase that is known to be present on the apical side in polarized RPE cells colocalized extensively with AQP1 in hESC-RPE cells (Fig. 3J) and also colocalized but not as extensively as in hiPSC-RPE cells (Fig. 3L). The tight junction marker ZO-1 colocalized only very moderately with AQP1 hiPSC-RPE cells (Fig. 3M). 

Figure 3

 

Expression and localization of AQP1. Confocal micrographs after indirect immunofluorescence labeling with AQP1 protein (green), neuroectodermal marker Pax6 (red), eye-specific protein cellular retinaldehyde-binding protein (CRALBP, red), or the polarization marker Na⁺/K⁺ ATPase (red), and the nuclear label 4′,6-diamidino-2-phenylindole (blue). (A) hFF, (B) undifferentiated hESC, (C) fusiform hESC-RPE, (D) fusiform hiPSC-RPE, (E–H) the localization of AQP1 in mature, cobblestone hESC-RPE in apical part (E), middle (F), and nuclear (G) part of cell, and (H) the ortho projection of the image. (I, J) cobblestone hESC-RPE and (K–M) cobblestone hiPSC-RPE. Scale bars: (A, B) 50 μm; (C–M) 10 μm.

AQP Functionality

A fluorescence accumulation test was done in the absence or presence of aquaporin inhibitors to address the functionality of AQPs in highly polarized hESC-derived RPE cells. ACE (2 mM), an AQP1 and carbonic anhydrase inhibitor, inhibited significantly by 55% (P = 0.0004) the fluorescence accumulation under hypoosmotic condition (Fig. 4A). HgIICl (10 μM) induced a significant 38% (P = 0.046) and 10 mM TEA induced 31%, an insignificant (P = 0.27) inhibition of the hypoosmotic-induced fluorescence accumulation. 

Figure 4

 

Functional testing of AQPs. The efficacy (= percentage of inhibition) of aquaporin inhibitors (2 mM ACE, 10 μM HgIICl, and 10 mM TEA) in inhibiting the flux of hypotonic (100 mOsm) saline solution to mature hESC-RPE cells during 10 minutes (A). The percentage of inhibition is calculated from relative to control (control = 100). The effect of inhibitors on transepithelial resistance (TEER) during 10 minutes of time in isoosmotic and hypoosmotic solutions (B). The studies were repeated at least three times. Data are expressed as mean ± SEM, *P < 0.05, **P < 0.01, ***P < 0.001. (C–K) The effect of inhibitors on the viability of cells. Viable (green fluorescence) hESC-RPE cells after 40-minute treatment with inhibitors. (C–F) Cells in isoosmotic PBS, (G–K) in hypoosmotic (100 mOsm) PBS, (C, G) no inhibitors, (D, H) 2 mM ACE, (E, I) 10 μM HgIICl; (F, K) 10 mM TEA. Scale bar: 50 μm.

The swelling of highly polarized hESC-derived RPE was also monitored by measuring TEER in isoosmotic and hypoosmotic solutions in the absence or presence of AQP inhibitors (Fig. 4B). In hypoosmotic solution, TEER increased 17% during a 5-minute follow-up, and remained at this level after 10 minutes in hypoosmotic solution. The difference between isoosmotic and hypoosmotic solution after 10 minutes was statistically significant (P = 0.03). In the presence of ACE, TEER stayed at the same level (no change at 5 minutes and 1% decrease at 10 minutes) in hypoosmotic solution. The protecting effect of ACE was statistically significant (P = 0.05) after 5 minutes but not after 10 minutes. With the HgIICl the hypoosmotic solution induced a 16% increase in TEER. TEA in turn induced a drop in TEER after 5 minutes, decreasing the value by 11%. However, this drop was transient and TEER returned back to the starting value after 10 minutes in TEA solution. 

Cell Viability

The microscopic observation revealed that the hypoosmotic solution did not have an effect on the cell viability (Fig. 4G) when compared with isoosmotic solution (Fig. 4C). The AQP inhibitors ACE (Figs. 4D, 4H) and TEA (Figs. 4F, 4K) did not have an effect on cell viability, whereas HgIICl had markedly lower intensity in calcein-AM both in isoosmotic (Fig. 4E) and hypoosmotic (Fig. 4I) solutions. Aquaporin inhibitors in isoosmotic solution did not increase the number of dead cells, but in the presence of pure hypoosmotic solution (Fig. 4G) or hypoosmotic solution and ACE (Fig. 4H) they increased the overall intensity of red, which might be a sign of ruptured cells. 

Discussion

In normal retina the intraocular pressure and choroidal osmotic pressure support the fluid absorption by RPE cells. This flow impedes fluid accumulation into the subretinal space and is responsible for the correct attachment of the neuroretina on RPE cells. ³ AQPs, which are integral transmembrane proteins permeable to water, are likely to play a role in fluid absorption by RPE cells under physiologic conditions. ³ Under some pathophysiologic conditions, the normal function of the blood–retinal barrier (BRB) is lost and proteins leak from the blood to the subretinal space, changing the osmotic environment. ^3,7 The macular edema, associated, for example, with diabetic retinopathy and age-dependent macular degeneration, is caused by such increased permeability of BRB. ^29,30  

The present screening of AQP mRNA expression (from AQP1 to AQP12) by quantitative real-time PCR revealed the presence of AQP1, AQP3, AQP4, AQP5, AQP6, AQP7, AQP10, AQP11, and AQP12 mRNA, and the absence of AQP2, AQP8, and AQP9 in undifferentiated hESC and hiPSC, as well as hESC- and hiPSC-derived RPE. 

AQP1, a water channel protein that is expressed solely in fluid absorbing and fluid secreting tissues, ⁷ is involved in retinal transepithelial water movement. ⁷ The AQP1 pore can be regulated by protein kinase C ³¹ and by cationic ions. ³² Previously, native human RPE cells, ⁸ cultured human adult, ⁸ and fetal RPE cells, ⁷ as well as immortalized human RPE (ARPE-19) cell line ¹⁴ have been shown to express AQP1 mRNA. Mature hESC-derived RPE cells have also been shown to express the AQP1 gene. ¹¹ Nevertheless, our present study shows for the first time that AQP1 mRNA expression was significantly increased, whereas the hESC- and hiPSC-derived RPE cells matured to cobblestone morphology. Moreover, AQP11 mRNA expression was similarly significantly increased as the AQP1 gene when hESC and hiPSC matured to cobblestone morphology. 

In cultured human fetal RPE cells, AQPs have been localized on the apical surface. ⁷ However, the localization of AQP1 protein in human stem cell–derived RPE cells, that had remained unknown so far, was for the first time documented by our study in both hESC- and hiPSC-derived RPE cells. The undifferentiated hESCs, already expressing the early eye lineage marker PAX6, were AQP1 negative, whereas some fusiform hESC-derived RPE and fusiform hiPSC-derived RPE cells expressed AQP1 protein in the cytoplasm and the more confluent fusiform hiPSC-derived RPE cells also on the plasma membrane. The cobblestone hESC- and hiPSC-derived RPE showed AQP1 protein at the apical plasma membrane colocalizing with the apical Na⁺/K⁺ ATPase marker. The apical membrane localization of AQP1 in polarized cobblestone hESC-RPE and hiPSC-RPE cells suggests that the AQP1 is functional on those cells. 

To test the AQP1 functionality, the volume of hESC-derived RPE cells with cobblestone morphology was monitored under hypoosmotic challenge in the presence of AQP inhibitors. Cell volume assays were not performed using fusiform hESC-derived RPE, fusiform hiPSC-derived RPE, and cobblestone hiPSC-derived RPE, because those cell populations were rather heterogeneous. The ACE, a pan carbonic anhydrase inhibitor, has been shown to be a specific AQP1 inhibitor ²⁷ but induce 80% inhibition also in AQP1 homolog AQP4. ^33,34 In Xenopus oocytes ACE induced >80% inhibition, ²⁷ and in HEK cells only 40% inhibition. ³⁵ In this study ACE induced a significant 55% inhibition in fluorescence accumulation assay. Because the cobblestone hESC-derived RPE express AQPs other than AQP1 and AQP4, it is therefore logical that acetazolamide did not induce a full inhibition. HgIICl, an unspecific but also rather toxic agent, ²⁵ can induce almost 100% inhibition in AQP functional assays. ²⁵ In cobblestone hESC-derived RPE, HgIICl induced a significant 38% inhibition of fluorescent accumulation. Nevertheless, cell viability assay indicated that the viability of hESC-derived RPE was decreased in the presence of HgIICl. Therefore, the effect of HgIICl inhibition might have been consequently partially lost. TEA, a blocker of voltage-gated potassium and cationic potassium channels, has been reported to inhibit AQP1 in some studies ^24,26 or to be ineffective. ²⁵ TEA induces 40% inhibition of AQP1-mediated water flux in oocytes ²⁶ and in MDCK cells.³⁶ In our hESC-derived RPE, TEA induced 31% statistically insignificant inhibition of fluorescent accumulation that was in the same percentage range of previous studies. ^26,36  

Because an increase of cell volume is known to increase TEER, ²⁸ this phenomenon was investigated for the first time in stem cells–derived RPE. Hypotonic challenge of cobblestone hESC-derived RPE increased TEER as expected, and ACE prevented the rise of TEER during the entire 10-minute period. The HgIICl did not prevent the rise in TEER, but this might be due to the decreased viability of the cells following HgIICl treatment. TEA decreased the TEER even below the isotonic level, but after 10 minutes the effect was subdued. Overall, our results show that AQP1 and possibly other AQPs are likely to be functional in hESC-derived RPE. 

In conclusion, the findings of the present study clearly demonstrated that hESC-derived RPE and hiPSC-derived RPE cells express AQP1, AQP3, AQP4, AQP5, AQP6, AQP7, AQP10, AQP11, and AQP12 mRNA, but not AQP2, AQP8, and AQP9 mRNA, at different maturation stages. The expression of AQP1 and AQP11 genes increased significantly during hESC-RPE and hiPSC-RPE cell maturation. AQP1 protein was localized at the apical membrane of both cobblestone hESC- and hiPSC-derived RPE. Furthermore, AQP1, and possibly other AQPs, are likely to be functional in hESC-derived RPE cells as assessed by the functional assays in the presence of AQP inhibitors. 

Acknowledgments

The authors thank Outi Heikkilä, Outi Melin, Hanna Koskenaho, and Elina Konsén for their technical assistance. 

Presented in part at the annual meeting of the Association for Research in Vision and Ophthalmology, Seattle, Washington, May 5–9, 2013. 

Supported by Tekes (the Finnish Funding Agency for Technology and Innovation); the Academy of Finland Grants 137801, 218050, 260375, and 253134; Fund for Medical Scientific Research Grant 3.4502.09; a grant from the International Doctoral Programme in Biomedical Engineering and Medical Physics Graduate School; and the Evald and Hilda Nissi Foundation. The authors alone are responsible for the content and writing of the paper. 

Disclosure: K. Juuti-Uusitalo, None; C. Delporte, None; F. Grégoire, None; J. Perret, None; H. Huhtala, None; V. Savolainen, None; S. Nymark, None; J. Hyttinen, None; H. Uusitalo, None; F. Willermain, None; H. Skottman, None