Introduction

Human embryonic stem cells (hESCs)1 and human induced pluripotent stem cells (hiPSCs)2 are regarded promising cell sources for transplantation in regenerative medicine. The challenges associated with the complex stem cell-derived products used in regenerative medicine require great scientific progress. In addition to the dynamic complexity of their biology, several safety concerns for human pluripotent stem cell (hPSC)-derived products have hindered their clinical translation, including the genomic instability of hPSCs and the risk of tumorigenicity3, 4.

The ability to confirm the quality and safety of cell-based therapeutic products (CTPs) in the manufacturing process will be a critical factor in the anticipated success of regenerative medicine. One of the most important issues in the development of safe hPSC-derived CTPs is ensuring that the final product does not form tumors after implantation3. There are two main causes of tumor formation from hPSC-derived CTPs. Firstly, products derived from hPSCs might contain residual undifferentiated stem cells that might proliferate and form teratomas5. Secondly, some cells in the products may transform to finally form tumors. The latter is a common issue in CTPs, regardless of the cell type used as raw material6.

To address the issue of tumorigenicity, recent publications have advocated the development of highly efficient differentiation protocols for hPSCs7,8,9 and have outlined methods for eliminating residual hPSCs in the products10, 11. Several methods for detecting a small population of residual undifferentiated cells in hPSC-derived CTPs have been reported: (1) in vivo tumorigenicity tests that detect teratoma formation in severe immunodeficient NOG mice12, (2) detection of LIN28 mRNA as an undifferentiated cell marker using quantitative reverse transcription (qRT-)PCR and droplet digital PCR13, 14, and (3) a highly efficient culture method for residual undifferentiated hiPSCs in products15. Similarly, assay methods have been developed for the detection of small numbers of transformed cells in CTPs: (1) in vivo tumorigenicity tests with NOG mice16, (2) digital analysis of soft agar colony formation17, and (3) cell growth analysis18. Although these assays are sensitive, they are time-consuming.

Cellular immortalization is widely known as a key step in the development of most human cancers, a defining property of cancer cells, and a prerequisite for cell transformation19. Therefore, we attempted to seek a novel immortalized cell marker and develop a rapid assay for detecting immortalized cells contained in CTPs. In this study, we employed retinal pigment epithelial (RPE) cells as a model of hPSC-derived CTPs because of their wide use in hPSC-derived CTPs. In the case of immortalization in primary human RPE cells, four papers have reported the establishment of immortalized RPE cell lines that spontaneously arose during in vitro culture20,21,22,23. Thus, it cannot be denied that immortalized RPE cells appear in products during the manufacturing process. Comprehensive microarray-based gene expression analysis indicated a marked upregulation of 15 transcripts in immortalized RPE cells. Our study identified a gene encoding slow skeletal muscle troponin T (TNNT1) as a suitable marker of immortalized RPE cells contained in hPSC-derived CTPs, and we propose a qRT-PCR method for the detection of this marker.

Results

Detection of Immortalized RPE Cells with Conventional Assays

To develop a detection assay of immortalized cells contaminating hPSC-derived RPE cells, we used three commercial immortalized RPE cell lines, and three lots of primary RPE cells as non-immortalized cell control samples. The following three immortalized RPE cell lines were used: ARPE-19, a spontaneously immortalized cell line21; ARPE-19/HPV-16, an ARPE-19 cell line transfected with DH-5 HPV-16; h1RPE7, a cell line derived from primary RPE cells transfected with a SV40 large T antigen24. Growth curve analysis confirmed that the growth rate of the primary RPE cells decreased as passage progressed, whereas ARPE-19 cells continued to proliferate (Fig. 1a). First, we sought to assess whether conventional assays were able to detect the immortalized RPE cells. This assay is generally used to evaluate cell transformation in vitro, demonstrated by the ability of the cells to grow in an anchorage-independent environment25. To quantify cell transformation, we used the CytoSelect 96-well Cell Transformation Assay. Unlike HeLa cells, which were used as a positive control, all immortalized RPE cell lines as well as primary RPE cells failed to grow in soft agar media when seeded at a density of 1.0 × 104 cells/well (Fig. 1b,c). These results demonstrated that the soft agar colony formation assay is not appropriate for the detection of immortalized RPE cells.

Figure 1
figure 1

Immortalized RPE cells are undetectable by conventional in vitro assays. (a) Cell growth analysis of primary RPE cells and ARPE-19 cells. (b,c) A soft agar colony formation assay was carried out to detect immortalized RPE cells. Phase-contrast images of primary RPE cells, immortalized RPE cells, and HeLa cells cultured in soft agar medium for 30 days (b; Scale bars, 300 µm). Quantification of the cellular DNA is shown in a bar graph. Results were expressed as a relative fold change of the value of blank well. Data are presented as the mean ± standard deviation of three independent experiments (c). (d) The relative mRNA expression of TERT and BIRC5/Survivin in primary RPE cells, immortalized RPE cells, and hiPSCs was determined by qRT-PCR. Data are presented as the mean ± standard deviation of three independent experiments. N.D.: not detected.

Full size image

Several tumor markers have shown to be detected with varying degrees of sensitivity and specificity26, 27. Accordingly, we expected that qRT-PCR detection of the tumor markers TERT and Survivin/BIRC5 would contribute to a sensitive assay of immortalized RPE cells. Contrary to our expectation, TERT was not detected in all of the immortalized RPE cell lines, while it was detected in hiPSCs (Fig. 1d). On the other hand, Survivin/BIRC5 was highly expressed in h1RPE7 cells and hiPSCs but not in ARPE-19 or ARPE-19/HPV-16 cells as compared to primary RPE cells. Therefore, TERT and Survivin/BIRC5 are not likely to be suitable marker genes for immortalized RPE cells.

Identification of Marker Genes for Immortalized RPE Cells

We attempted to identify a novel marker gene of immortalized RPE cells. To this end, we compared the transcription profiles of the three lots of primary RPE cells and three immortalized RPE cell lines using microarray analysis. In total, 15 genes [BDNF (brain-derived neurotrophic factor), CAMK2N1 (calcium/calmodulin-dependent protein kinase II inhibitor 1), CARD6 (caspase recruitment domain family, member 6), CDH11 (cadherin 11), CLDN11 (claudin 11), GAD1 (glutamate decarboxylase 1), IL15 (interleukin 15), PAPSS2 (3′-phosphoadenosine 5′-phosphosulfate synthase 2), PLK2 (polo-like kinase 2), PSG1 (pregnancy specific beta-1-glycoprotein 1), PSG4 (pregnancy-specific beta-1-glycoprotein 4), PSG7 (pregnancy-specific beta-1-glycoprotein 7), SHOX2 (short stature homeobox 2), TNNT1 (troponin T type 1), TWIST1 (twist homolog 1)] were highly expressed in immortalized versus primary RPE cells (P < 0.05, fold change >20). We used qRT-PCR for accurate confirmation of the expression of these genes in the RPE cells and hiPSCs. In particular, the expression level of TNNT1 was equally high in all immortalized cell lines, while equally low in all three lots of primary cells. Furthermore, TNNT1 mRNA was highly expressed in hiPSCs that can self-renew and the other two immortalized RPE cell lines, hTERT RPE-1 cells and ABM-RPE cells (Fig. 2 and Supplementary Fig. 1).

Figure 2
figure 2

Identification of immortalized RPE cell marker genes. qRT-PCR analysis of immortalized RPE cell marker candidate genes in primary RPE cells, immortalized RPE cells, and hiPSCs. Bar graph represents fold gene expression relative to ARPE-19 cells. Results are means ± standard deviations (n = 5). N.D.: not detected.

Full size image

Masutomi et al. have previously reported that the expression of TERT corresponds temporally with the entrance and transit of a large percentage of these synchronized cells through S phase in normal human cells28. To determine the schedule of TNNT1 as well as TERT expression throughout the cell cycle, we monitored mRNA levels of TNNT1 in primary RPE cells during cell growth. With increasing culture time, both the population of S-phase cells and the expression of TERT and Survivin/BIRC5 decreased. In contrast, TNNT1 was constitutively expressed even after 14 days of culture (Fig. 3a and Table 1). Thus, transcripts of TNNT1 corresponded less well with the cell cycle state in primary RPE cells than did those of TERT and Survivin/BIRC5. These results suggested that TNNT1 is a stable marker gene of immortalized RPE cells without variable background signal derived from the cell-cycle state of normal RPE cells.

Figure 3
figure 3

Time course analysis of TNNT1 mRNA expression and ChIP-qPCR assay of the transcription start site of TNNT1. (a) Time course of the expression of TERT, Survivin/BIRC5, and TNNT1 mRNA in primary RPE cells during culture for 3, 7 and 14 days as analyzed by qRT-PCR. Bar graph represents fold gene expression relative to day 3. (b) ChIP-qPCR assay of H3K4me3 and H3K27me3 in −3 kb, −2 kb, −1 kb, and + 1 kb regions from the TSS of TNNT1. Positive control experiments were performed with promoter regions of GAPDH (H3K4me3) and LIN28 (H3K27me3). Results are means ± standard deviations of three independent experiments (black; ARPE-19 cells, gray; primary RPE cells). ND: not detected. *P < 0.05, **P < 0.01, Student’s t-test (n = 3).

Full size image
Table 1 Cell cycle analysis of primary RPE cells.
Full size table

To ascertain whether the upregulation of TNNT1 expression in immortalized RPE cells was due to epigenetic regulation, we explored the chromatin status at the TNNT1 locus in immortalized RPE cells. We performed ChIP-qPCR in primary RPE and ARPE-19 cells using antibodies specific for trimethylated lysine 27 of histone H3 (H3K27me3), a repressive mark, and trimethylated lysine 4 of histone H3 (H3K4me3), an active mark, to determine their mark in −3 kb, −2 kb, −1 kb and +1 kb regions from the transcription start site (TSS) of TNNT1. H3K27me3 was significantly reduced at −2 kb and +1 kb regions relative to the TSS in ARPE-19 cells when compared to primary RPE cells. On the other hand, we did not recognize any appreciable difference in H3K4me3 enrichment at the TNNT1 gene locus between primary RPE cells and ARPE-19 cells (Fig. 3b). These findings implied that loss of H3K27me3 mark at the TNNT1 locus promotes transcriptional activation of TNNT1.

Detection of Immortalized RPE Cells Using qRT-PCR

To develop and validate a qRT-PCR assay detecting TNNT1 mRNA, we spiked 5 × 104 (10%), 1.5 × 104 (3%), 5 × 103 (1%), 1.5 × 103 (0.3%), and 5 × 102 (0.1%) ARPE-19 cells into 5 × 105 primary RPE cells. Total RNA was extracted from the mixtures and subjected to qRT-PCR using TNNT1-specific primers. Relative values of TNNT1 mRNA of the 10%, 3%, 1%, 0.3%, and 0.1% ARPE-19 cell-spiked samples were 6.9 ± 0.7, 3.3 ± 0.4, 1.5 ± 0.2, 1.4 ± 0.3, and 1.2 ± 0.1 (ratio to the average level of primary RPE cells ± standard deviation), respectively. Correspondingly, to determine the lower limit of detection (LLOD) of ARPE-19 cells contained in primary RPE cells, relative values of TNNT1 mRNA were analyzed in five lots of primary RPE cells. The LLOD of the assay signal is commonly calculated as the mean plus 3.3-fold the standard deviation of the measurement of negative controls29. Based on the dispersion of signals from five lots of primary RPE cells as a negative control (1 ± 0.35 [fold over the average of five lots ± standard deviation]), the LLOD of the TNNT1 qRT-PCR assay was calculated as 2.1 (Fig. 4a). These results indicated that the TNNT1 qRT-PCR assay is able to detect 3% of ARPE-19 cells contained in primary RPE cells. Next, RNA isolated from undifferentiated hiPSCs, hiPSCs after 5 and 40 days of differentiation, and purified hiPSC-derived RPE cells was subjected to the TNNT1 qRT-PCR assay. The mRNA expression of TNNT1 gradually decreased during differentiation to a final level below the LLOD. Relative values of TNNT1 mRNA of the hiPSC, day 5, day 40, and purified hiPSC-derived RPE samples were 123.8 ± 18.0, 62.8 ± 6.0, 17.7 ± 2.2, and 1.2 ± 0.1, respectively (Fig. 4b). These results indicated that TNNT1 mRNA levels are strongly dependent on the extent of residual undifferentiated hiPSCs in differentiated RPE cells.

Figure 4
figure 4

Validation of the TNNT1 qRT-PCR assay. (a) qRT-PCR analysis of TNNT1 mRNA expression in primary RPE cells spiked with different amounts of ARPE19 cells and 5 lots of primary RPE cells. The LLOD (dashed line) was determined as described in the Results section. (b) Time course of the expression of TNNT1 mRNA during differentiation of hiPSCs to RPE cells. All values are expressed as mRNA levels relative to the mean of 5 lots of primary RPE cells. Results are means ± standard deviations (n = 3).

Full size image

Expression of TNNT1 in Various Cancer Tissues

To investigate the versatility of TNNT1 as a marker of immortalized cells, we examined TNNT1 expression levels in various types of human cancer tissues and normal tissues using a TissueScan array spotted with cDNAs derived from 18 different cancer or normal tissues (adrenal glands, breasts, cervix, colon, endometrium, esophagus, kidneys, liver, lungs, lymphoid tissue, ovaries, pancreas, prostate, stomach, testes, thyroid gland, urinary bladder, and uterus). The tissues displaying a statistically significant difference in TNNT1 expression are shown in Fig. 5, other tissues are shown in Supplementary Fig. 2. As shown in Fig. 5, cancer tissues of cervix, colon, lungs, ovaries, and testes showed significantly elevated levels of TNNT1 expression as compared with the corresponding normal tissues. As for ovarian cancer, these results agree with a study by Bapat et al. showing a significant upregulation of TNNT1 expression in ovarian cancer in in silico transcriptomics analysis30.

Figure 5
figure 5

Versatility of TNNT1 as a marker of immortalized cells. Expression of TNNT1 in human cancer tissues and normal tissues was quantified using the TissueScan Cancer Survey Panel. Data were obtained using the comparative CT method, with values normalized to β-actin levels. Bar graph represents fold expression of TNNT1 relative to the lowest detected sample in cancer tissue samples (black; cancer tissue, gray; normal tissue). Blank: not detected. Statistical analysis was performed using the Mann–Whitney U test.

Full size image

Overexpression of TNNT1 Enhances Cell Migration and Actin Polymerization

On the basis of the above findings, we predicted that TNNT1 might play a role in immortalization. To investigate this, we first aimed to stably transduce the TNNT1 gene into primary RPE cells via transgene integration into a zinc finger nuclease (ZFN)-targeting site using a lentiviral vector delivery. However, we were not able to generate primary RPE cells overexpressing TNNT1 that were resistant to puromycin. Instead, we established ARPE-19 cells overexpressing TNNT1 by inserting TNNT1 at an AAVS1 site cleaved with ZFN (pCMV-TNNT1 ARPE-19). As expected, the pCMV-TNNT1 ARPE-19 cells showed an approximately 70-fold increase in TNNT1 mRNA (Fig. 6a). As expression of TERT, a key player in telomere maintenance in immortalized cells, is involved in the migration of HCT116 cells31, we examined the effects of TNNT1 on cell migration by a gap closure migration assay. The pCMV-TNNT1 ARPE-19 cells showed significantly higher inward migration than wild-type ARPE-19 cells (Fig. 6b). Complementary experiments demonstrated that transient expression of TNNT1 in hTERT RPE-1 cells also enhanced their migration and supported our results (Supplementary Fig. 3). Conversely, shRNA-mediated TNNT1 knockdown in ARPE-19 cells significantly suppressed migration of knockdown cells (Supplementary Fig. 4). Moreover, we observed a distinct enhancement of actin polymerization induced by overexpression of TNNT1 in ARPE-19 cells (Fig. 6c). These results suggested that TNNT1 expression affects actin filament organization and enhances cell migration ability in immortalized RPE cells.

Figure 6
figure 6

Overexpression of TNNT1 enhances cell migration and actin polymerization. (a) qRT-PCR analysis of TNNT1 mRNA in ARPE-19 cells (WT) and pCMV-TNNT1 ARPE-19 cells (pCMV-TNNT1). Bar graph represents fold expression of TNNT1 relative to ARPE-19 cells. (b) Phase-contrast images of migration assay of ARPE-19 cells (WT) and pCMV-TNNT1 ARPE-19 cells (pCMV-TNNT1). Scale bar: 200 µm. Closure rates were calculated as described as material methods. (c) Immunocytochemistry of phalloidin-Alexa 488 of ARPE-19 and pCMV-TNNT1 ARPE-19 cells. Scale bar: 50 µm. Results are means ± standard deviations of three independent experiments. *P < 0.05, **P < 0.01, Student’s t-test (n = 3).

Full size image

Discussion

Tumorigenicity, which is attributed to residual undifferentiated cells and transformation of cells, is one of the major concerns in developing hESC/hiPSC-derived CTPs. Here, we introduced a novel in vitro assay for detecting immortalized RPE cells and showed its efficacy using hiPSC-derived RPE cells. Cellular immortalization is widely known as a key step in the development of most human cancers, and the importance of an immortalized marker is illustrated by the fact that four immortalized RPE cell lines spontaneously arose from primary human RPE cells during in vitro culture20,21,22,23.

Jiang et al. previously reported that TERT-expressing immortalized RPE cells do not form tumors in nude mice after subcutaneous transplantation32. In line with this report, we found that the soft agar colony formation assay failed to detect colonies of immortalized RPE cell lines. Contrary to our expectation, TERT and Survivin/BIRC5, which are highly expressed in various cancer cells, appeared not suitable as marker genes of immortalized RPE cells. Thus, such assays detecting transformed cells cannot be applied for detecting immortalized RPE cells. To develop an in vitro assay system detecting immortalized RPE cells, we comprehensively screened genes expressed in immortalized RPE cells and successfully identified 15 candidate marker genes, the expression of which was 20 times higher in immortalized than in primary RPE cells. Some promising genes have been reported to be related to cancer. CAMK2N1 and CDH11 have been shown to encode tumor suppressors33, 34. CARD6 is significantly expressed in gastric, colorectal, and esophageal cancers35. Other promising genes include the pregnancy-specific glycoprotein family genes (PSG1, PSG4, and PSG7) and BDNF, CLDN11, GAD1, IL15, PAPSS2, PLK2, SHOX2, TNNT1, and TWIST1; these were previously identified without any known immortalization-associated functionalities. Among the 15 candidate genes, only TNNT1 expression exhibited more than 100-fold increase in hiPSCs and immortalized RPE cell lines as compared to primary RPE cells. Therefore, we focused on TNNT1 as an immortalized RPE cell marker and developed a qRT-PCR assay for TNNT1 transcript detection. A validation study revealed that the TNNT1 qRT-PCR assay can detect as few as 3% of ARPE-19 cells spiked into primary RPE cells.

H3K27me3 but not H3K4me3 was strongly enriched in the TNNT1 gene locus in primary RPE cells than in ARPE-19 cells, indicating that TNNT1 expression of normal RPE cells was possibly repressed under epigenetic control. The loss of H3K27me3 at the TNNT1 locus could explain the marked difference in its expression between ARPE-19 and primary RPE cells. However, it is still unknown whether loss of the H3K27me3 mark in the TNNT1 gene locus occurs in the process of cell immortalization.

The troponin complex forms the Ca2+-sensitive molecular switch that regulates striated muscle contraction in response to alterations in the intracellular calcium concentration. Troponin T binds tropomyosin to anchor the troponin complex to a specific location of thin filaments. In vertebrates, three homologous genes have evolved to encode three muscle type-specific troponin T isoforms. TNNT1 is known to be a slow skeletal muscle-specific troponin T36. shRNA-mediated TNNT1 knockdown did not affect the immortalization of ARPE-19 cells even if cultured for more than 10 passages (Kuroda et al., unpublished observation). However, in the current study, we found that TNNT1 overexpression increased migration and enhanced actin polymerization in ARPE-19 cells. We also repeated migration assay with hTERT RPE-1 transiently expressing TNNT1 and obtained the similar results that TNNT1 enhanced migration of its overexpressed cells. As microinjection of tropomyosin into epithelial cells has been reported to induce rapid cell migration37, TNNT1 likely contributes to the interaction of actin and tropomyosin, which regulates cell migration and invasion38. In the context of cancer, we demonstrated substantial expression of TNNT1 in cancer tissues of the cervix, colon, lungs, ovaries, and testes. Similarly, Gu et al. reported that postoperative relapse of gallbladder carcinoma is positively related to TNNT1 expression39. According to our results and previous reports, TNNT1 seems to play multiple roles in cancer. TNNT1 may partly but not essentially function in cell immortalization, the mechanism of which remains to be uncovered.

Tumorigenicity, which is attributed to residual undifferentiated cells and transformation of cells, is one of the major concerns in the development of hESC/hiPSC-derived CTPs. Here, we introduced a novel in vitro assay for detecting immortalized RPE cells and showed its efficacy in hiPSC-derived RPE cells. Our TNNT1 qRT-PCR assay is rapid and simple, and will hopefully contribute to process validation and quality control of hPSC-derived CTPs in future.

Materials and Methods

Cell Culture

Human fetal RPE cells were obtained from Lonza (lot. A, lot-0F3292; lot. B, lot-0F3237; lot. C, lot-0000182939; lot. E, lot-0000249852) or ScienCell Research Laboratories (lot.D, 6011) and maintained in Retinal Pigment Epithelial Cell Basal Medium (Lonza) containing supplements (L-glutamine, GA-1000, and bFGF; Lonza). ARPE-19 and ARPE-19/HPV-16 cells were obtained from ATCC and maintained in DMEM/F12 containing 20% fetal bovine serum (FBS) and 1% penicillin/streptomycin (all from Gibco). h1RPE7 cells were obtained from ECACC and maintained in Ham’s F-10 medium containing 20% FBS, 1 μg/mL puromycin, and 1% penicillin/streptomycin. hTERT RPE-1 cells were obtained from ATCC and maintained in DMEM/F12 containing 10% FBS and 1% penicillin/streptomycin. ABM-RPE cells were obtained from Applied Biological Materials (ABM) Inc. and maintained in Prigrow medium (ABM) containing 10% FBS and 1% penicillin/streptomycin. The hiPSC line 201B7 was obtained from RIKEN and maintained in mTeSR1 medium (STEMCELL Technologies) on Matrigel-coated dishes. Differentiation of 201B7 into RPE cells was stimulated as previously described13. HeLa cells were obtained from ATCC and maintained in DMEM containing 10% FBS and 1% penicillin/streptomycin. All cells were cultured at 37 °C in a humidified atmosphere of 5% CO2 and 95% air.

Soft Agar Colony Formation Assay

A soft agar colony formation assay was carried out using the CytoSelect 96-well Cell Transformation Assay kit (Cell Biolabs) as described previously13. Cells were cultured for 30 days. Colonies were lysed and quantified with CyQuant GR dye using a fluorometer equipped with a 485/520 nm filter set (Wallac 1420 ARVOsx multilabel counter, PerkinElmer).

qRT-PCR

Total RNA was isolated using RNeasy Mini Kit (Qiagen) and treated with DNase I according to the manufacturer’s instructions. qRT-PCR was performed with the QuantiTect Probe RT-PCR Kit (Qiagen) on a 7300 Real-Time PCR System (Applied Biosystems). Target gene expression was normalized to GAPDH mRNA, which was quantified using TaqMan human GAPDH control reagents (Applied Biosystems). Ct values of greater than 35 were regarded as “not detected” (“N.D.”). The sequences of the primers and probes used in the present study are listed in Table S1. Oligonucleotides were obtained from Sigma-Aldrich.

GeneChip and Biostatistical Analysis

Total RNA was isolated from three lots of primary RPE cells and three immortalized RPE cell lines using an RNeasy Mini Kit (Qiagen) and treated with DNase I according to the manufacturer’s instructions. Samples were converted into biotinylated cRNA using Two-Cycle Target Labeling and Control Reagents (Affymetrix). Labeled RNA was processed for microarray hybridization to Human Genome U133 Plus 2.0 GeneChips (Affymetrix). Five samples of each RPE cell strain were used in this experiment. An Affymetrix GeneChip Fluidics Station was used to perform streptavidin/phycoerythrin staining. The hybridization signals on the microarray were scanned using a GeneChip Scanner 3000 (Affymetrix) and analyzed using Expression console software (Affymetrix). Normalization was done by global scaling with the arrays scaled to a trimmed average intensity of 500 after excluding the 2% of probe sets with the highest and the lowest values. The National Center for Biotechnology Information Gene Expression Omnibus (NCBI GEO) accession number for the microarray data is GSE80985. Informational probe sets were filtered from the data set in three steps. First, probe sets were regarded as “present” when indicated as “present” by “absolute analysis” using Expression console software in all five samples from one strain. Probe sets regarded as “absent” in more than one immortalized RPE line were eliminated. Then, probe sets for which no significant difference was observed among the cell lines by ANOVA (P < 0.05) were eliminated. Finally, probe sets for which the difference between the maximum and minimum mean values of probe sets in the cell lines was equal to or more than 20-fold were used for further analysis.

Cell Cycle Analysis

Cell cycle analysis was performed by fluorescence-based cell sorting as previously described40. Briefly, after collection, the cells were fixed with 70% ethanol for 2 h at 4 °C, and were treated with 0.25 mg/mL RNase solution for 30 min at 4 °C and stained with a PI solution (50 μg/mL), and analyzed on a FACS Calibur (BD). Cell cycle distribution was analyzed with the FlowJo cell cycle Watson (Pragmatic) model.

ChIP-qPCR Assay

Primary RPE cells and ARPE-19 cells were treated with trypsin/EDTA (0.25% trypsin, 0.5 mM EDTA in phosphate-buffered saline (PBS) for 5 min at room temperature. Dissociated cells were treated with 1% paraformaldehyde for 8 min at room temperature. After quenching the PFA crosslinking reaction with 125 mM glycine, the fixed cells were washed with PBS. The cells were suspended in lysis buffer (50 mM Tris-HCl (pH 8), 10 mM EDTA, 1% SDS, and Complete protease inhibitor cocktail (Roche)) and sonicated to an average fragment size of 200–1000 bp. Solubilized chromatin was diluted with ChIP dilution buffer (16.7 mM Tris-HCl (pH 8.1), 167 mM NaCl, 1.2 mM EDTA, 0.01% SDS, 1% Triton X-100 and Complete protease inhibitor cocktail) and clarified by centrifugation at 13,000 × g for 10 min at 4 °C. The supernatant was pre-cleared with protein G-Dynabeads (Life Technologies), which were pre-blocked with rabbit control IgG (ab46540, Abcam) at 4 °C for 1 h. The pre-cleared chromatin was incubated with anti-H3K4me3 (ab8580, Abcam) and anti-H3K27me3 (17–622, Merck Millipore) antibodies overnight at 4 °C. Immune complexes were bound to protein G-Dynabeads for 2 h at 4 °C. The beads were sequentially washed with low-salt buffer (20 mM Tris-HCl (pH 8.1), 150 mM NaCl, 2 mM EDTA, 0.1% SDS and 1% Triton X-100), high-salt buffer (20 mM Tris-HCl (pH 8.1), 500 mM NaCl, 2 mM EDTA, 0.1% SDS and 1% Triton X-100), LiCl wash buffer (10 mM Tris-HCl (pH 8.1), 250 mM LiCl, 1 mM EDTA, 1% IGEPAL-CA630, 1% Na deoxycholate), and TE buffer (10 mM Tris-HCl (pH 8.0), 1 mM EDTA (pH 8.0)). Immune complexes bound to protein G-Dynabeads were suspended in Complete elution buffer (20 mM Tris-HCl (pH 7.5), 5 mM EDTA, 50 mM NaCl, 1% SDS, and 50 μg/mL proteinase K) and incubated for 2 h at 68 °C. After incubation, DNA was purified with ChIP DNA Clean & Concentrator Kit (Zymo Research) and analyzed by qPCR using the following gene-specific primers: LIN28 forward, GGGTTGGGTCATTGTCTTTTAG; reverse, AAAGGGTTGGTTCGGAGAAG, GAPDH forward, CGGGATTBTCTGCCCTAATTAT; reverse, GCACGGAAGGTCACGATGT, TNNT1 −3 kb, GPH1021012 (-)03 A; TNNT1 −2 kb, GPH1021012 (−)02 A; TNNT1 −1 kb, GPH1021012 (−)01 A; TNNT1 +1 kb, GPH1021012 (+)01 A. The piTect ChIP qPCR primers GPH1021012 (+)01 A, (−)01 A, (−)02 A, and (−)03 A were obtained from Qiagen. qPCR data were normalized to 5% of the purified input DNA, which was used as a measure of the total amount of chromatin present in the sample.

TissueScan™ Cancer and Normal Tissue cDNA Array Analysis

The TissueScan Cancer Survey Panel 4 × 96-III (Origene) was used according to the manufacturer’s protocol. The panel consisted of 381 cDNA samples covering 18 different cancers. β-Actin expression was used to normalize relative TNNT1 expression in tissue samples and was quantified using TaqMan human β-actin control reagents (Applied Biosystems). For statistical analysis, we regarded not detected as zero. Some samples were excluded from following statistical analysis, because the number of samples contained in one group was less than three (breast, endometrium, bladder, and uterus).

Generation of a TNNT1-overexpressing ARPE-19 Cell Line

For TNNT1 overexpression, a transgene encoding TNNT1 was inserted into the AAVS1 locus in the ARPE-19 cell using the CompoZr Targeted Integration-AAVS1 kit (Sigma-Aldrich) per the manufacturer’s protocol. The base sequence of pCMV promoter and human TNNT1 ORF (NM_001126133) was de novo synthesized (GenScript) and cloned into pZDonor-AAVS1 Puromycin Vector (Sigma-Aldrich). ARPE-19 cells were transduced with 30 μg targeting vector and 5 μL zinc finger nuclease (ZFN) mRNA using a Nepa21 electroporator (Nepa Gene). Then, the cells were treated with 2–5 μg/mL puromycin for 1 day. TNNT1 mRNA expression was confirmed by qRT-PCR.

Transient transfection of TNNT1

For transient expression of TNNT1, hTERT PRE-1 cells plated in 12-well dishes were transiently transfected with pCMV-TNNT1 (RG225325, ORIGENE) or pCMV-Entry vector (PS100001, ORIGENE) using TransIT-X2 Reagent (Mirus Bio LLC) according to the manufacturer’s instructions. The total amount of DNA was adjusted to 1 μg. Twenty-four hours after transfection, cells were cultured for 2 days in normal media, and then subjected to migration assay.

Lentivirus-derived RNAi and generation of TNNT1 knockdown cell line

ARPE-19 TNNT1-knockdown cells were generating by transducing ARPE-19 with lentivirus particles that expressed a TNNT1-targeting set of shRNA. Briefly, Lenti-X 293 T cells were transfected with individual clones from Sigma MISSION shRNA targeting set (TRCN0000118897) or the control shRNA plasmid along with a MISSION lentivirus packaging Mix (Sigma-Aldrich) according to the manufacture’s procedures. Media containing viruses were collected 48 hours after transfection. ARPE-19 cells were transduced with the viruses in the presence of 8 µg/ml polybrene (Sigma-Aldrich) for 24 hours, and then subjected to selection with 2 µg/ml puromycin for 48 hours.

Migration Assay

Cell migration was measured using a Radius 24-Well Cell Migration Assay Kit (Cell Biolabs) according to the manufacturer’s instructions. Each well contains a circular 680-μm-diameter “gel spot” to which cells do not attach. Before the experiment, cells were cultured to confluence. At the start of the experiment, the gel spot was removed, after which cells gradually populated the circular void space. After a 10-h incubation at 37 °C, migrated cells were captured using phase-contrast microscopy and the migration of cells into the initially cell-free area was analyzed using Keyence image analysis software.

Actin Filament Staining

Cells were fixed with 4% paraformaldehyde in PBS for 20 min at room temperature. After washing with PBS, the cells were permeabilized with 0.2% Triton-X100 in PBS for 15 min and blocked with 2% bovine serum albumin in PBS for 30 min. Samples were incubated for 20 min with 5 units/mL of Alexa Fluor 488 phalloidin (A12379, Thermo Fisher). Nuclei were stained with 10 μg/mL DAPI (Invitrogen) and examined under a BZ-X710 fluorescence microscope (Keyence).

Statistical Analysis

Statistical analysis was performed using Systat 13 Software (Systat Software Inc., CA). Comparisons were made using the Student t-test (for two groups) and the Mann–Whitney U test (for normal vs. cancer tissues, Fig. 5 and Supplementary Fig. 1). P-values < 0.05 were considered significant.