LncRNA ZNF503-AS1 promotes RPE differentiation by downregulating ZNF503 expression

Long noncoding RNAs (lncRNAs) have important roles in various biological processes. Our previous work has revealed that dedifferentiation of retinal pigment epithelium (RPE) cells contributes to the pathology of age-related macular degeneration (AMD). Herein, we show roles of lncRNAs in RPE differentiation. We used microarray to identify lncRNA expression profiles in human induced pluripotent stem cells (hiPSCs) and hiPSC-derived RPE cells. A total of 217 differentially expressed lncRNAs along with the differentiation were initially identified, among which 13 lncRNAs showed a consistent fold change of over 2. LncRNA ZNF503-AS1, located in the cytoplasm of RPE cells, was found consistently upregulated along with RPE differentiation, and downregulated in the RPE-choroid of AMD patients. In vitro study further suggested that ZNF503-AS1 insufficiency could inhibit RPE differentiation, and promote its proliferation and migration. As ZNF503-AS1 is transcribed from the antisense strand of the ZNF503 gene locus, we further revealed its regulatory role in ZNF503 expression. ZNF503-AS1 was reversely correlated with ZNF503 expression. Our results also suggested that ZNF503 could inhibit RPE differentiation, and promote its proliferation and migration. Thus, ZNF503-AS1 potentially promotes RPE differentiation through downregulation of ZNF503 expression. In addition, nuclear factor-κB was recognized as a potential upstream transcript factor for ZNF503-AS1, which might participate in promoting RPE differentiation by regulating the expression of ZNF503-AS1. Taken together, our study identifies a group of RPE differentiation relevant lncRNAs, and the potential role of ZNF503-AS1 in the pathology of atrophic AMD, which might help with the intervention of AMD patients.

Retinal pigment epithelium (RPE) is a monolayer of cuboidal, polarized, and pigmented epithelial cells located in the outer retina between photoreceptors and choroidal vessels. 1-3 RPE forms a part of the blood/retina barrier, secrets multiple growth factors, and is crucial in maintaining regular retinal functions. [2][3][4] Dysfunction and depletion of RPE cells are involved in multiple retinal degenerations, including agerelated macular degeneration (AMD). 2,5-7 AMD is a universal leading cause for irreversible vision loss in people aged over 55. [8][9][10] Clinically, AMD can be classified into two major categories, namely atrophic and exudative AMD. 10 Our group has previously recognized RPE dedifferentiation as a crucial contribution factor to the pathogenesis of atrophic AMD. 1,3 Atrophic AMD is typified by subepithelial deposits and degeneration of RPE cells involving but not limited to the macular region. 10 No efficient therapies have been developed for atrophic AMD. Therefore, better insights into its pathology and seeking for a putative therapeutic target have become the focus of current researches. On the basis of our previous findings, inhibiting RPE dedifferentiation may retard or block AMD disease course, thus helping with better control of AMD patients.
Long noncoding RNAs (lncRNAs) are transcripts longer than 200 nucleotides structurally homologous to proteincoding mRNAs, but has little or no protein-coding potential. 11 LncRNAs can modulate gene expressions as competing endogenous RNAs (ceRNAs). [11][12][13][14] The ceRNA hypothesis defines a microRNA (miRNA)-mediated posttranscriptional regulatory network. In this network, proteincoding and non-protein-coding RNAs share one or more miRNA response elements and compete for miRNA binding, further modulating each other's expressions. 12,15 LncRNAs have been reported to have regulatory roles in diverse biological processes, including cell differentiation, stem cell maintenance, and epigenetic regulation. [16][17][18] Dysregulation of lncRNAs is found involved in multiple human diseases such as cancer, neurological problems, and cardiovascular diseases. 19 Roles of lncRNAs in ocular diseases, such as diabetic retinopathy, proliferative vitreoretinopathy, glaucoma, ocular tumors, and ocular neovascularization, have also been identified. 11,14,[20][21][22][23][24][25][26] However, the association between lncRNAs and RPE differentiation is poorly understood. In this study, we aim to reveal the roles of a lncRNA, ZNF503-AS1, in RPE differentiation, and to seek for a lncRNA-based potential therapeutic target for dry AMD.

Results
Differently expressed lncRNAs during RPE differentiation. To reveal the roles of lncRNAs on RPE differentiation, we used microarrays to modulate the expression profiles of lncRNAs and mRNAs in human induced pluripotent stem cells (hiPSCs) and in hiPSC-derived RPE cells (hiPSC-RPE) at 30, 60, and 90 days post differentiation (d.p.d.), respectively. Pluripotency of the hiPSC colonies and competence of differentiation have been previously determined. 3 Figure 1a).
ZNF503-AS1 is mainly localized in cytoplasm. ZNF503-AS1 is an intergenic lncRNA located on chromosome 10q22.2, and is partly conserved among multiple species (Supplementary Figure S1). However, the biological function of ZNF503-AS1 is largely unknown. Our above results confirmed that ZNF503-AS1 was consistently upregulated along with the differentiation from hiPSC to RPE (Figure 1j). RNA-fluorescent in situ hybridization (FISH) suggested that ZNF503-AS1 was expressed in human RPE (ARPE19) cells, and was mainly localized in cytoplasm (Figure 2a).
Clinical relevance of ZNF503-AS1. Taken that RPE dedifferentiation was involved in the pathogenesis of retinal degenerative diseases, 1,3 we assumed that ZNF503-AS1 expression would be reduced in dysfunctional RPE. To test our hypothesis, we compared ZNF503-AS1 expression between macular RPE-choroid of a 70-year-old male and a 30-year-old female donor. Our previous study suggested that the male donor had RPE dysfunction. 3 As expected, ZNF503-AS1 expression was significantly reduced in the aged donor with RPE dysfunction when compared with the young female donor (Figure 2b). To further tell the role of ZNF503-AS1 in AMD pathogenesis, we compared its expression among RPE-choroid samples of 38 AMD patients and 46 healthy controls. Personal particulars and clinical diagnosis for each participant were detailed before. 27 As indicated in Figure 2c, ZNF503-AS1 was downregulated in AMD patients when compared with controls. Subgroup analysis was then performed to see with which AMD subtype ZNF503-AS1 was mostly correlated. AMD diagnostic criterion was described previously. 27 Our findings suggested that ZNF503-AS1 expression was only found reduced in RPEchoroid of dry AMD patients (n = 16), a branch of atrophic AMD (Figure 2d). Taken together, our findings suggested that dysregulation of ZNF503-AS1 is involved in RPE dysfunction, especially in atrophic AMD.
ZNF503-AS1 promotes RPE differentiation. We next used transfection assays to modulate expression levels of ZNF503-AS1 in cells to evaluate its role in RPE differentiation. Exogenous ZNF503-AS1 was significantly upregulated in cells transfected with Ac-ZNF503-AS1 plasmid (Figure 3a), and endogenous ZNF503-AS1 was downregulated in cells transfected with ZNF503-AS1 scramble small interfering RNA (siRNA; Figure 3b). Noteworthy, three pairs of siRNAs oligos targeting different regions of ZNF503-AS1 were initially designed and tested, and the siRNA showing best efficiency and stability was chosen for further studies (Supplementary Figure S2A (Figures 5a and b). However, rates of both proliferation and migration elevated in ZNF503-AS1 siRNA-transfected group (Figures 5c and d). Thus, our data indicated an inhibitory role of ZNF503-AS1 in RPE proliferation and migration. ZNF503-AS1 is regulated by nuclear factor-κB. We applied PROMO online program to predict transcription factor binding sites (TFBS) in the promoter region of ZNF503-AS1.
Nuclear factor-κB (NF-κB) was revealed as its putative transcript factor with three potential TFBS in its promoter region (Figure 6a). We then used ammonium ZNF503-AS1 in RPE differentiation X Chen et al pyrrolidinedithiocarbamate (PDTC) to inhibit the activation of NF-κB, and tumor necrosis factor-α (TNF-α) to stimulate its activation in ARPE19 cells. Expression of NF-κB p65 (activated form of NF-κB) was upregulated in cells treated with PDTC (Figures 6c and d), and was downregulated in cells treated with TNF-α (Figures 6b and d). To better understand the regulatory role of NF-κB, we next tested ZNF503-AS1 expression in different treated groups. Our data supported that expression of ZNF503-AS1 was reduced in the PDTC-treated group (Figure 6e) and was elevated in the TNF-α-treated group (Figure 6f).
ZNF503-AS1 regulates ZNF503 expression. ZNF503-AS1 is transcribed from the antisense strand of the ZNF503 gene locus. We then assessed whether ZNF503-AS1 could affect the expression of ZNF503. We observed that ZNF503 mRNA was decreased in cells overexpressing ZNF503-AS1 (Figure 7a), and was increased in cells with ZNF503-AS1 knocked down (Figure 7b). Our data suggested a potential regulatory role of ZNF503-AS1 in ZNF503 expression.
ZNF503 inhibits RPE differentiation, and promotes RPE proliferation and migration. As ZNF503-AS1 regulates ZNF503 expression, we next detected whether ZNF503 insufficiency would improve the regular function of RPE. Endogenous expression of ZNF503 was remarkably decreased in cells transfected with ZNF503 siRNA (Figure 7c). Three pairs of siRNA oligos targeting different regions of ZNF503 were initially designed and tested. We only selected the siRNA showing best efficiency and stability for further investigations (Supplementary Figure S2B). In ARPE19 cells transfected with ZNF503 siRNA, mRNA expressions of RPE differentiation relevant markers, including RPE65, RLBP1, LRAT, KRT18, and CTNNB1, were elevated ( Figure 7d). Immunoblotting further indicated that ZNF503 insufficiency in ARPE19 cells would increase the protein expressions of Mertk, cytokeratin-18, and β-catenin (Figures 7e and f). Enhanced expressions of ZO-1 and β-catenin were also revealed in cells transfected with ZNF503 siRNA by immunofluorescence (Figures 7g and h). No obvious protein mislocalization was found. Thus, our data indicated an inhibitory role of ZNF503 in RPE differentiation. Impacts of ZNF503 on cell proliferation and migration were also monitored. Our data demonstrated that rates of both proliferation and migration elevated in cells transfected with ZNF503 siRNA (Figures 5e and f), suggesting that ZNF503 could promote RPE proliferation and migration. Taken together, our findings indicated that ZNF503-AS1 promoted RPE differentiation and inhibited its proliferation and migration by interfering with ZNF503 expression.

Discussion
RPE dedifferentiation has been typified as a crucial contributing factor to the pathology of atrophic AMD. 1,3 Therefore, blocking RPE dedifferentiation is a promising strategy in the treatment of dry AMD. Our group has previously demonstrated that AKT2/mTOR pathway and a group of miRNAs are potentially involved in RPE differentiation. 1,3 However, the molecular mechanism underlying RPE differentiation is not fully understood. LncRNAs are important regulators in gene expression. Roles of lncRNA dysregulation in ocular diseases have been well established. 11 However, its role in RPE differentiation or pathology of atrophic AMD has never been reported. In the present study, we used microarray to obtain lncRNA expression profiles in hiPSC and hiPSC-RPE at different stages. A total of 217 differentially expressed lncRNAs along with the differentiation were initially sorted out. Among all identified lncRNAs, 13 presented a consistent fold change of over 2, including 8 upregulated and 5 downregulated lncRNAs.
ZNF503-AS1 is an intergenic lncRNA with unrecognized biological interplay. Our data suggest that ZNF503-AS1 is expressed in the cytoplasm of RPE cells. We have demonstrated that ZNF503-AS1 is consistently upregulated along with RPE differentiation, and is downregulated in RPE-choroid of atrophic AMD patients and a senior donor with RPE dysfunction. In vitro data have also indicated the role of ZNF503-AS1 in promoting RPE differentiation, and inhibiting RPE proliferation and migration. ZNF503-AS1 is transcribed from the antisense strand of the ZNF503 gene locus. ZNF503 has been recognized as a transcriptional repressor that promotes mammary epithelial cell proliferation and migration, while its role in maintaining RPE function has not been fully elucidated. 30,31 RPE cell is a type of mammary epithelial cell. Similar to previous findings, our data indicated that ZNF503 could inhibit RPE differentiation, and promote its proliferation and migration. We have also revealed that ZNF503-AS1 expression is reversely correlated with the expression of ZNF503. Thus, our findings suggest that ZNF503-AS1 potentially promotes RPE differentiation by downregulation of ZNF503 expression. However, how ZNF503 dysregulation affect RPE function is still poorly understood. Therefore, more investigations are warranted. Thus, our results indicate that ZNF503-AS1 dysregulation is involved in RPE dedifferentiation and pathology of atrophic AMD, implying its application as a biomarker and therapeutic target for atrophic AMD.
The NF-κB pathway is involved in many cellular processes, including immunity and inflammation. Roles of NF-κB in inflammatory signaling and AMD pathogenesis have been well established. [32][33][34] Activation of the NF-κB pathway has been previously suggested to stimulate pigmentation of cultured (g and h) Immunofluorescent staining suggested that ZO-1 (g) and β-catenin (h) expressions were elevated in ARPE19 cells transfected with ZNF503 siRNA when compared to cells transfected with scramble siRNA. No obvious mislocalization of proteins was identified in all transfected groups ZNF503-AS1 in RPE differentiation X Chen et al primary hRPE cells. 35 In this study, NF-κB is recognized as a potential upstream transcript factor for ZNF503-AS1, which may participate in promoting RPE differentiation by regulating the expression of ZNF503-AS1.
Taken together, our study identifies a group of RPE differentiation relevant lncRNAs. We also reveal a potential role of ZNF503-AS1 in AMD pathogenesis. ZNF503-AS1 may become a biomarker and a potential therapeutic target for atrophic AMD. However, more investigations are still needed to better illustrate the roles of lncRNAs in RPE differentiation and AMD pathology.

Materials and Methods
Samples. Postmortem specimens of a 30-year-old female donor and a 70-yearold male donor were provided by Lions New South Wales Eye Bank through Save Sight Institute, the University of Sydney, Australia. 3 All procedures followed standard procedures of eye donation for research and were approved by the institutional ethical committees conformed to Declaration of Helsinki. Written informed consents were obtained from all donors before their donations.
Culture and differentiation of hiPSC-RPE. HiPSC (IMR90-57) were cultured on mouse embryonic fibroblasts (SiDan-Sai Biotechnology Co., Ltd, Shanghai, China) in six-well tissue culture plates as detailed elsewhere. 3 The hiPSCs were then differentiated into RPE cells according to the SFEB/CS method using low-molecular-weight compounds CKI-7 (5 μM) and SB-431542 (5 μM). 36 Microarray profiling and computational analysis. Total RNA was isolated from hiPSC-RPE cells at different stages using TRIzol reagent (Invitrogen, Carlsbad, CA, USA). 4,37 Qualities and concentrations of RNA samples were determined using Nano-Drop ND-1000 spectrophotometer (Nano-Drop Technologies, Wilmington, DE, USA). Agilent Sureprint G3 Human GE 8 × 60 K Microarray (Agilent Co., Palo Alto, CA, USA) covering 27958 Entrez Gene RNAs and 7419 human lncRNAs was used to generate the lncRNA expression profiles from the following four groups in duet: hiPSC; and hiPSC-RPE at 30, 60, and 90 d.p.d. An Aglient G3 scanner was then used to determine the immunofluorescence intensities of the arrays. Data were further analyzed using feature extraction software. Multiples of differentially expressed genes were calculated. miRNA expression profiles of the four groups were generated as described before. 3 In addition, microarray data (GSE29801) and sample information (GSM738433-GSM738607) of 175 independent RPE-choroid samples were downloaded from Gene Expression Omnibus data sets and analyzed as indicated previously. 3,27 Among the 175 samples, 88 were obtained from normal individuals, including 50 macular and 38 extramacular samples, and 87 were from AMD patients, including 41 macular and 46 extramacular samples.
Reverse transcription-PCR and real-time PCR. Reverse transcription-PCR (RT-PCR) and real-time PCR were performed using a previously defined protocol. 4,37 One microgram of the extracted total RNA was used for cDNA synthesis using PrimeScript RT Kit (Takara, Otsu, Japan). Real-time PCR was conducted using FastStart Universal SYBR Green Master (ROX; Roche, Basel, Switzerland) with StepOne Plus Real-time PCR System (Applied Biosystems, Darmstadt, Germany). Information of primers was provided in Supplementary Table S2.
Reagents and cell transfection. ZNF503-AS1 and ZNF503 siRNAs were purchased from RiboBio Co. Ltd. Open reading frame sequence of ZNF503-AS1 was synthesized, amplified, and inserted into the expression vector pcDNA3.1 (Sigma, St. Louis, MO, USA) using XhoI and KpnI restriction sites to generate the Ac-ZNF503-AS1 plasmids. Sequences of the constructed plasmids were confirmed using Sanger sequencing.
For transfection assay, cells were seeded into six-well plates and transfected with 100 pmol siRNA or 4 μg expression vector using Lipofectamine 2000 transfection reagent (Invitrogen) per the manufacturer's protocol. Cells were collected at 48 h post transfection for RNA isolation, and at 72 h post transfection for protein extraction and immunofluorescent staining. PDTC (Beyotime, Shanghai, China; concentration: 100 μM) 44 and TNF-α (Sigma; concentration: 100 ng/μl) were used for cell stimulation. Cells were collected at 48 h after treatment.
Immunoblotting. Immunoblotting was performed per previously described protocols. 3,45 Briefly, cells were collected at 72 h post transfection in ice-cold RIPA buffer (Beyotime) containing protease inhibitors cocktail (Roche) for protein extraction. Extracted proteins were separated by 10% SDS-polyacrylamide gel electrophoresis and transferred to a polyvinylidene fluoride membrane (Millipore, Billerica, MA, USA). Membranes were then blocked, incubated with primary antibodies at 4°C overnight (Supplementary Table S3), washed, and probed with corresponding horseradish peroxidase-conjugated secondary antibodies (1:10 000 diluted in 1 × PBS; ICL Inc., Newberg, OR, USA) for 1 h at room temperature. Autoradiography with the ECL-Western blotting system (BioRad, Hercules, CA, USA) was then applied to develop the blots per manufacturers' protocols. ImageJ software (http://rsb.info.nih.gov/ij/index.html) was used to determine and quantify protein expressions.
Immunofluorescent staining. Immunofluorescence was performed per previously described protocols. 3,45 Cells were collected, fixed with 4% paraformaldehyde, permeabilized with 0.5% Triton X-100, blocked in bovine serum albumin, and incubated with primary antibodies (Supplementary Table S3) at 4°C overnight. Cells were then washed and incubated with corresponding fluorescence-conjugated secondary antibodies (1:1000 diluted in 1 × PBS; Invitrogen) for 1 h at room temperature. Cell nuclei were counterstained by DAPI (Sigma). Images were collected using a confocal microscope (LSM 510).
Monitoring cell proliferation and migration. Rates of cell proliferation and migration were detected in real-time using xCELLigence system E-Plate (Roche) according to the manufacturer's protocol. To monitor cell proliferative rates, 5000 cells were seeded in each well of the E-Plate. Transfection was performed at 24 h post plantation. For migration assay, 40 000 post-transfection cells were planted into each well. Cells were culture with fresh DME/F12 medium. Impedance value for each well was automatically determined by the xCELLigence system and expressed as a CI value. Rates of cell proliferation and migration were determined by calculating the slope of the line between two given time points.
Statistics. We used GraphPad Prism (version 4.0; GraphPad Software, San Diego, CA, USA) for statistical analysis. Student's t-test was used for comparisons between groups. All presented data were based on biological triplicates. Data were shown as mean ± S.D., and P-value o0.05 was considered as statistically significant.
Education Institutions (PAPD, JX10231801). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.