Xeno- and Feeder-Free Differentiation of Human iPSCs to Trabecular Meshwork-Like Cells by Recombinant Cytokines

Abstract

Purpose

Stem cell-based therapy has the potential to become one approach to regenerate the damaged trabecular meshwork (TM) in glaucoma. Co-culture of induced pluripotent stem cells (iPSCs) with human TM cells has been a successful approach to generate autologous TM resembling cells. However, the differentiated cells generated using this approach are still problematic for clinical usage. This study aimed to develop a clinically applicable strategy for generating TM-like cells from iPSCs.

Methods

Highly expressed receptors during iPSC differentiation were identified by AutoSOME, Gene Ontology, and reverse transcription polymerase chain reaction (RT-PCR) analysis. The recombinant cytokines that bind to these receptors were used to generate a new differentiation protocol. The resultant TM-like cells were characterized morphologically, immunohistochemically, and transcriptionally.

Results

We first determined two stages of iPSC differentiation and identified highly expressed receptors associated with the differentiation at each stage. The expression of these receptors was further confirmed by RT-PCR analysis. Exposure to the recombinant cytokines that bind to these receptors, including transforming growth factor beta 1, nerve growth factor beta, erythropoietin, prostaglandin F2 alpha, and epidermal growth factor, can efficiently differentiate iPSCs into TM-like cells, which express TM biomarkers and can form dexamethasone-inducible CLANs.

Conclusions

We successfully generated a xeno- and feeder-free differentiation protocol with recombinant cytokines to generate the TM progenitor and TM-like cells from human iPSCs.

Translational Relevance

The new approach minimizes the risks from contamination and also improves the differentiation efficiency and consistency, which are particularly crucial for clinical use of stem cells in glaucoma treatment.

Introduction

Intraocular pressure (IOP) homeostasis, which is dramatically influenced by the resistance to aqueous humor (AH) drainage,¹ is essential for ocular health. Disruption of the trabecular meshwork (TM) function can lead to increased resistance to AH outflow and sustained elevation of IOP, which is the major risk factor for glaucomatous optic neuropathy.^2,3 Multiple groups have demonstrated that stem cell-based therapy is one effective approach to regenerate the damaged TM and maintain the IOP in the eye; such approaches have included the use of adult TM stem cells, mesenchymal stem cells, and induced pluripotent stem cells (iPSCs).^4–12

To date, the most commonly used approach to generate TM-like cells from iPSCs is through co-culture of iPSCs with human TM (HTM) cells.¹³ Using cell culture medium conditioned by human TM cells, Abu-Hassan et al.¹⁰ successfully generated TM-like cells from iPSCs growing on a TM cell-secreted extracellular matrix. Likewise, our group demonstrated that iPSCs growing on Matrigel-coated plates can also be induced to differentiate into TM-like cells by co-culturing iPSCs with human immortalized TM cells.¹³ We subsequently improved differentiation by using HTM cells, a differentiation approach that appears to be more successful than the use of human immortalized TM cells according to the morphology of differentiated iPSCs.⁹

Although with these improvements, these iPSC-derived TM-like cells are still problematic for clinical usage because of contamination from both fetal bovine serum (FBS) and humans, unstable differentiation efficiency, and the required long-term culture in vitro. FBS commonly used for expending cells in vitro and the products from other human beings could carry the risk of xenoimmunization and zoonotic transmission.¹⁴ Human TM cells from different donors exhibit some differences in cell viability and function,¹⁵ which can lead to unstable efficiency for stem cell differentiation. Long-term cell culture in vitro triggers many side effects, such as genetic mutation¹⁶ and cellular senescence.¹⁷ To circumvent these disadvantages, a new differentiation approach of xeno- and feeder-free culture eliminating various contaminations is required.

Here, we first defined two stages of iPSC differentiation through morphological analysis and bioinformatic analysis of RNA-sequencing data of TM-like cells.⁹ From each stage, highly expressed receptors associated with TM development were selected through AutoSOME (http://jimcooperlab.mcdb.ucsb.edu/autosome) and Gene Ontology analysis, further confirmed by reverse transcription polymerase chain reaction (RT-PCR) analysis. Recombinant cytokines binding to these receptors were used to generate the new xeno- and feeder-free differentiation approach, which can effectively induce human iPSCs to differentiate into TM-like cells, by studying their morphological features and conducting immunohistochemistry analyses and transcriptomic assays.

Material and Methods

Human Trabecular Meshwork Cell Isolation and Culture

Ocular anterior segments from human donors 1 to 3 were obtained from the Iowa Lions Eye Bank (Iowa City, IA), and those from donors 4, 5, and 6 were obtained from Beijing Tongren Eye Hospital (Beijing, China). The protocol for collecting human tissues was approved by the institutional review boards of the University of Iowa Institutional and Beijing Tongren Eye Hospital. After dissection, the TM was visible as a brown circle in the irideocorneal angle and isolated by using a 0.5-mm curette. After rinsing with Dulbecco's modified eagle medium (DMEM; Sigma-Aldrich, St. Louis, MO), the tissue was digested in DMEM containing 4 mg/mL collagenase A (Sigma-Aldrich) and 4 mg/mL human serum albumin (Sigma-Aldrich) at 37°C for 2 hours. After centrifugation (1000 rpm for 3 minutes), the pellets were resuspended in TM cell growth medium comprised of Gibco Medium 199E containing 20% Gibco FBS, 90 µg/mL porcine heparin (Sigma-Aldrich), 20 U/mL endothelial cell growth supplement (Sigma-Aldrich), and 1.7 mM l-glutamine (Sigma-Aldrich). The digested tissue was seeded onto 1% gelatin precoated six-well plates (Thermo Fisher Scientific, Waltham, MA) and cultured in the incubator with 5% CO₂ at 37°C. As in our previous studies,⁸ cells at passages five to eight with robust expression of TM biomarkers and the capacity for dexamethasone (DEX)-induced myocilin (MYOC) secretion were used in this study.

Human iPSC Generation and Culture

Human induced pluripotent stem cells were created through reprogramming of renal urethra epithelial cells isolated from the urine of three healthy human donors, designated U1, U2, and U3 (Cellapy Biotechnology, Beijing, China).¹⁸ Renal urethra epithelial cells at passage two were infected with non-integrating Sendai virus carrying Oct4, Sox2, c-Myc, and Klf4 transcription factors (CytoTune-iPS 2.0 Sendai Reprogramming Kit; Thermo Fisher Scientific). Successfully reprogrammed cells, iPSC colonies, were generated during days 9 to 28 after infection and were transferred for expansion. Written informed consent was obtained from all donors, and the protocol to reprogram renal cells was carried out following the guidelines for stem cell clinical research in China. The reprogramming was accomplished by using the CytoTune-iPS 2.0 Sendai Reprogramming Kit, and the reprogrammed cells were cultured in mTeSR-1 Standarized Medium containing recombinant human basic fibroblast growth factor (rh-bFGF) and recombinant human transforming growth factor beta (rh-TGF-β; STEMCELL Technologies, Cambridge, MA) in 0.2% Matrigel (Corning Inc., Corning, NY) precoated six-well plates. The iPSC colonies were then isolated by using 5 mg/mL collagenase (Sigma-Aldrich) and expanded in mTeSR-1 Standardized Medium (STEMCELL Technologies) as described previously.¹⁸ The iPSCs at passages 20 to 30 were tested as the Sendai virus-free cells and used for the differentiation.

Human iPSC Differentiation by Conditioned Medium

Human iPSCs derived from the renal cells of U1, U2, and U3 were seeded on 0.2% Matrigel-coated plates and cultured in mTeSR-1 Standardized Medium until reaching 5% to 10% confluency. Human TM cells from each donor were seeded at 70% to 90% confluency in Corning Transwell cell culture plate inserts and positioned above the human iPSCs. These cells were cultured in biopsy medium comprised of Minimum Essential Medium-α (MEM-α; Thermo Fisher Scientific), 10% FBS, and 0.2% Primocin (InvivoGen, San Diego, CA). The inserts with human TM cells from different individuals were switched routinely to ensure that iPSCs were exposed to conditioned medium from all three human donors.

Enzyme-Linked Immunosorbent Assay Analysis

Human TM cells (5 × 10⁴) were seeded in one well of a 96-well plate and cultured in the TM medium until reaching 100% confluency. The supernatant was collected for enzyme-linked immunosorbent assays (ELISAs). First, the capture antibodies with specificity for TGF-β₁, nerve growth factor beta (NGF-β), erythropoietin (EPO), epidermal growth factor (EGF), or prostaglandin F2 alpha (PGF2α) were immobilized in microtiter wells (Shanghai BlueGene Biotech Co., Ltd., Shanghai, China). Conditioned medium and standard samples at 0, 3, 6, 12, 24, and 48 pg/mL were incubated with the immobilized antibodies at 37°C for 60 minutes. After washing with wash buffer (Shanghai BlueGene Biotech) five times, the second specific antibodies conjugated to horseradish peroxidase (Shanghai BlueGene Biotech) were added to react with the captured cytokine. A chromogenic substrate (Shanghai BlueGene Biotech) was then added and incubated at 37°C for 15 minutes. After termination, the optical density of the reaction was measured with a spectrophotometer (Synergy MxMulti-Mode Microplate Reader; BioTek, Winooski, VT) and compared with the readings of standard samples. The experiment was repeated three times by using the HTM cells of three donors.

Xeno- and Feeder-Free Differentiation With Recombinant Cytokines

Human iPSCs derived from the renal cells of U1, U2, and U3 were seeded in 0.2% Matrigel-coated plates and cultured in mTeSR-1 Standardized Medium until reaching 5% to 10% confluency. Human iPSCs were then cultured in stage 1 differentiation medium containing MEM-α, 10% KnockOut Serum Replacement (Thermo Fisher Scientific) and 2 ng/mL TGF-β₁, 100 ng/mL NGF-β, and 2 U/mL EPO (Sino Biological, Beijing, China) for 7 days. Subsequently, the cells were maintained for 14 days in stage 2 differentiation medium comprised of MEM-α, 10% KnockOut Serum Replacement, 2 ng/mL TGF-β₁, 100 ng/mL NGF-β, 2 U/mL EPO, 10 µM PGF2α, and 10 ng/mL EGF. KnockOut Serum Replacement is a commercial cell culture additive^19,20 that has been proved to be an effective serum substitute for a number of mammalian cells cultured in vitro. The medium was changed every 2 days. The morphology of differentiated cells was monitored by an inverted microscope (Leica Microsystems, Wetzlar, Germany).

RNA Sequencing and AutoSOME Analysis

RNA sequencing was carried out in human fibroblasts, fibroblast-derived iPSCs, and conditioned medium-derived TM-like cells for 1, 2, and 3 months and in human TM cells of donors 1 to 3 on the HiSeq 2500 sequencer (Illumina, San Diego, CA). We also collected urine epithelial cell-derived iPSCs, urine-derived TM-like cells by conditioned medium (U-TM-like cells-C; 1 month differentiation), urine-derived TM-like cells by a xeno-free approach (U-TM-like cells-X), and HTM cells of donors 4 and 5 for RNA sequencing using the Illumina NovaSeq 6000 System. Total RNA was extracted using TRIzol (Thermo Fisher Scientific) and assessed using Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA) and Qubit Fluorometer (Thermo Fisher Scientific). A stranded mRNA library was prepared for sequencing mRNA and generating stranded mRNA information. The fragments per kilobase per million mapped reads (FPKM) value, which reflects the relative expression of a transcript, was compiled from the Cufflinks output (University of Washington, Seattle, WA). Genes with fold change > 4 and averaged FPKM > 1 were selected for AutoSOME analysis.²¹ Unit variance, median centering, and P threshold of <0.01 were chosen in AutoSOME for clustering genes with the same expression modulus. Cluster analysis (CapitalBio Technology Co., Ltd., Beijing, China) was performed using significantly differentially expressed genes with fold differences ≥ 2 (|log2FC|≥1, where FC is the fold change of expressions) and P < 0.05.²²

RT-PCR Analysis

Total RNA from four samples (iPSCs of U1, iPSCs of U1 differentiated for 1 month, and human TM cells of donors 4 and 5) was isolated with TRIzol. The concentration and purity of RNA were determined by measuring the absorbance at 260 and 280 nm (A260/280) using a NanoDrop spectrophotometer (Thermo Fisher Scientific). cDNAs were generated from RNAs through random primed reverse transcription reaction (Promega, Madison, WI). The target genes were amplified in triplicate using the SYBR Green system (Bio-Rad, Hercules, CA). The primers used for amplifying the target genes are listed in Table 1, and GAPDH (Gene ID: 2597, OMIM: 138400) was used as a reference gene. PCR reactions were carried out at 95°C for 15 minutes, followed by 50 cycles at 95°C for 10 seconds, at 50°C for 30 seconds, and at 72°C for 30 seconds.

Table 1.

Primers for Amplifying Receptors

Gene	Forward Primer	Reverse Primer
HEPOR	CTTGTGGTATCTGACTCT	ATAAGGCTGTTCTCATAAG
HTGFBR1	GATTACCAACTGCCTTAT	TTAGCCATTACTCTCAAG
HTGFBR2	GAATATAACACCAGCAATC	CAGTAGAAGATGATGATGA
HTGFBR3	GTCTACTATAACTCCATTGT	TCTCCTTCATCCATATCT
PTPRN	AGTGAAGGAGATTGACAT	CAAATTCAAACTGGTCCT
PTGER4	TGGTCATCTTACTCATTG	GCATTTCTCTTGTATGTATAG
PTGFR	AATGATGTTAAGTGGTGT	ATGTCTTCTGTGTTGTAG
PTGER2	ATGAATGAAACCTCTTCC	CATCTTGTGTTCTTAATGAA
PTGER1	CTTGTCGGTATCATGGTG	GTAGGATGTACACCCAAG
EDIL3	AGTTACTGGTGTGATTAC	TCATTACTGTAGGCAATT
MEGF9	GTCATCATCATTGTTGTG	CTTCATTGCCATCATCTT

EDIL3, EGF-like repeat and discoidin I-like domain-containing protein 3; HEPOR, human erythropoietin receptor; HTGFBR1, 2, and 3, human transforming growth factor-β receptor 1, 2, and 3; MEGF9, multiple epidermal growth factor-like-domains 9; PTGER1, 2, and 4, prostaglandin E receptor 1, 2, and 4; PTGFR, prostaglandin F2-alpha receptor; PTPRN, protein tyrosine phosphatase receptor-type N.

Immunohistochemistry

Cells were grown on coverslips coated with poly-d-lysine (Solarbio Science and Technology, Beijing, China). After reaching 70% to 80% confluency, cells were fixed with 4% paraformaldehyde (Thermo Fisher Scientific), rinsed in Dulbecco's PBS (145-mM NaCl, 8.1-mM Na₂HPO₄·12H₂O, and 1.9-mM NaH₂PO₄·2H₂O, pH 7.2–7.4; Thermo Fisher Scientific) for 5 minutes and incubated in the blocking solution (Dulbecco's PBS with 1% bovine serum albumin Sigma-Aldrich) for 1 hour. The sections were further incubated with the diluted primary antibodies (1:100) and the corresponding secondary antibodies (1:200). The nuclei of the tissue were stained with 4′,6-diamidino-2-phenylindole (DAPI; Santa Cruz Biotechnology, Dallas, TX). Coverslips were then mounted on the slides using Neutral Balsam (Solarbio, Beijing, China) and imaged with confocal microscopy (Nikon, Tokyo, Japan). The primary antibodies included rabbit polyclonal antibodies anti-laminin A4 (LAMA4) (Abcam, Cambridge, UK), anti-tissue inhibitor of matrix proteases 3 (TIMP3) (Abcam), anti-aquaporin 1 (AQP1) (Abcam), anti-MYOC (Abcam), mouse polyclonal antibodies anti-collagen IV (ColIV) (Abcam), and Phalloidin-iFluor 488 (Thermo Fisher Scientific). The secondary antibodies included Donkey anti-Rabbit IgG Alexa Fluor 350 and Donkey anti-Mouse IgG Alexa Fluor 488 (Thermo Fisher Scientific).

Induction of Cross-Linked Actin Network Formation

Human differentiated iPSCs and TM cells were exposed to 100-nM DEX for 3 days. Dimethylsulfoxide (Sigma-Aldrich) with the same dilution ratio was used as the vehicle control. High-magnification (100×) images of cells stained with Phalloidin-iFluor 488 were taken by confocal microscopy (Nikon). As reported before,^23,24 cross-linked actin networks (CLANs) were defined as a specific cytoskeletal structure with geodesic dome-like polygonal lattices in the dome-shaped cells. CLAN-formed cells in 16 regions of each image were counted. The CLAN forming ratio was calculated by dividing the numbers of cells with the typical CLAN structure by the numbers of the nuclei stained with DAPI.

Statistical Analyses

Two-tailed t-tests and normality and log-normal tests were used for statistical analyses. Tukey's multiple comparisons test was used for statistical analysis of cell size data, RT-PCR results, and FPKM values of RNA sequencing. P < 0.05 was considered to be statistically significant. Data are expressed as the mean ± standard error of the mean.

Results

Two Distinct Stages Involved in Human iPSC Differentiation to Trabecular Meshwork Cells

As we described earlier,^9,10,13 human iPSCs reprogrammed from fibroblasts or keratinocytes can be successfully induced to differentiate into cells that resemble TM cells by using medium conditioned by human TM cells. Here, we induced human iPSCs derived from renal epithelial cells to differentiate into TM-resembling cells by using the same approach. Undifferentiated iPSCs were observed as compact colonies comprised of cells with a large nucleus and less cytoplasm. iPSCs differentiated for 1 month altered their morphology to assume a spindle-like shape reminiscent of HTM cells, although their size was much smaller than that of HTM cells (surface area, P < 0.01) (Figs. 1A, 1B). In the following 2 months, differentiated cells kept growing (surface area, 5684.0 ± 388.6 µm²; length, 221.4 ± 6.7 µm) until they reached a similar size as the HTM cells of donor 4 (surface area, 5978.0 ± 372.0 µm²; length, 200.4 ± 9.0 µm), donor 5 (surface area, 5194.0 ± 320.7 µm²; length, 158.4 ± 9.6 µm), and donor 6 (surface area, 5292.0 ± 299.5 µm²; length, 137.6 ± 4.7 µm) (Fig. 1B). We also observed cell growth rate during differentiation (Supplementary Fig. S1). From iPSCs to TM progenitors (D0 to D30), cells were dramatically enlarged at a length-increasing rate of 18.4 µm/day. In the following month (D30 to D60), the cell growth rate (length-increasing rate of 11.6 µm/day) was more reduced than during the first stage (D0 to D30). Therefore, two distinct differentiation stages can be defined in iPSC differentiation: iPSCs to TM progenitors (1 month) and progenitors to TM-like cells (2 months). Our immunohistochemistry analysis showed robust expression of the TM markers LAMA4, TIMP3, AQP1, MYOC, and ColIV in both TM-like cells and human TM cells (Fig. 1C).

Figure 1.

Characterization of human TM-like cells induced by the conditioned medium. (A) Morphological observations of TM-like cells during differentiation. Human iPSCs colonies of U1 were generated via reprogramming renal urethra epithelial cells. Through co-culturing with HTM cells, iPSCs were differentiated for 60 days. The morphological changes monitored by Nikon microscopy were compared with HTM cells of donor 1. (B) Surface area and length of TM progenitor and TM-like cell. TM progenitors (D30) were significantly smaller than TM-like cells (D60). Tukey's multiple comparisons test was used for data analysis. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. (C) Immunohistochemical characterization of TM-like cells differentiated for 60 days. Laminin A4 (LAMA4), tissue inhibitor of matrix proteases 3 (TIMP3), aquaporin 1 (AQP1), myocilin (MYOC), and collagen type IV (ColIV) were detected with robust expression in both TM-like cells and HTM cells. Scale bar: 100 µm.

Receptors With the Most Robust Expression at Each Differentiation Stage

The observation that iPSC differentiation occurs in two stages is also supported by our previous study, which suggested that differentiated iPSCs display a very stable transcriptional signature after induction for 1 month.⁹ Given the roles of the conditioned medium in iPSC differentiation, receptors were assumed to function importantly to transduce the differentiation signals. In this study, we reanalyzed the data to identify highly expressed receptors in each stage of iPSC differentiation through AutoSOME analysis (Fig. 2A). After removing genes with fold change less than 4 and mean FPKM less than 1,6300 genes were submitted to AutoSOME. Genes were grouped into one of 136 clusters based on similarity of expression modulus and were further classified into five groups based on the gene expression pattern (Fig. 2B). In each group, the black line in the figure represents the log(FPKM + 1) value, and the confidence scale for each gene was reflected by the size of the line. The colored line in each group represents the mean value of log(FPKM + 1). As shown in Figure 2B, genes in groups 2 and 3 were considered to have the highest expression in TM progenitors and human TM cells, respectively.

Figure 2.

AutoSOME clustering analysis of RNA sequencing data. (A) A clustering heat map with the values of log(FPKM + 1). The heat map is ordered from top to bottom by decreasing cluster size and hierarchically ordered from left to right by whole-transcriptome expression. Cell samples were hierarchically ordered as somatic cells used for reprogramming; iPSCs; differentiated iPSCs for 1, 2, and 3 months; and HTM cells. A total of 6300 genes were clustered into 136 clusters based on their expression modulus; 36 clusters remained after removing clusters with more than 500 or fewer than 30 genes. (B) Grouping analysis of representative clusters with the same expression pattern. The 136 clusters with the same expression pattern were categorized into five groups, designated as 1 (red), 2 (blue), 3 (green), 4 (purple), and 5 (yellow). Representative clusters in each group are shown, and the cluster numbers are labeled on the top of each image (e.g., clusters 12, 18, 2, 3, 4, 5, and 6 in group 1). The black line represents the log(FPKM + 1) of each gene, and the confidence scale for each gene is reflected by the size of the black line. The colored line represents the mean value of log(FPKM + 1) in each cluster.

Through Gene Ontology analysis (http://www.pantherdb.org/), receptors in groups 2 and 3 were selected (Table 1), and their function in stem cell differentiation or TM maintenance was investigated by searching publications in PubMed (Table 2). As shown in Table 2, TGF-β receptor 1 (TGFBR1), receptor-type tyrosine-protein phosphatase-like N (PTPRN), and erythropoietin receptor associated with differentiation at stage 1 have been described as highly relevant to multiple types of stem cell differentiation. Prostaglandin receptors (PTGER and PTGFR), epidermal growth factor receptors (multiple epidermal growth factor-like-domains 9 [MEGF9] and EGF-like repeat and discoidin I-like domain-containing protein 3 [EDIL3]), and TGF-β receptors (TGFBR2 and 3) involved in the differentiation at stage 2 have been shown to play important roles in maintaining the function of the trabecular meshwork.

Table 2.

Receptors Associated With iPSC Differentiation

	Gene ID		FPKM/Average				Literatures (TITLE/ABSTRACT must include the keywords)		Literature (IF is in the top 5)
Cell Type	Full Name	Abbreviation	iPSC	TM Progenitor	HTM	Keywords	Stem Cell Differentiation	Trabecular Meshwork	Stem Cell Differentiation	Trabecular Meshwork
TM progenitor	Receptor-type tyrosine-protein phosphatase-like N	PTPRN	3.7	49.1	0.4	Phosphatase	176	34	Chan et al. 2003,25 Cornejo et al. 2011,26 Firulli et al. 2003,27 Yasui et al. 2014,28 Yu et al. 2013.29	Buie et al. 2013,30 Hosseini et al. 2006,31 Luo et al. 2014,32 Nishida et al. 1994,33 Pattabiraman, Pecen, and Rao 2013,34 Ramachandran et al. 2011,35 Sawaguchi et al. 1994,36 Tellios et al. 2017,37 Thomson et al. 2019,38 Xue, Comes, and Borras 2007,39 Xue et al. 2006,40 Zaiden and Beit-Yannai 2015.41
	TGF-beta receptor type-1	TGFBR1	23.5	56.4	12.3	Transforming growth factor	97	136	Bonauer et al. 2009,42 Jia et al. 2017,43 Wandzioch and Zaret 2009,44 Xu et al. 2002,45 Zhang et al. 2015.46	Agarwal and Agarwal 2018,47 Gabelt and Kaufman 2005,48 Liton et al. 2005; Luna et al. 2011,49 Raghunathan et al. 2015,50 Tamm 2002,51 Wahlig, Lovatt, and Mehta 2018.52
	Erythropoietin receptor	EPOR	4.6	11.8	3.1	Erythropoietin	22	1	Suda et al. 1985,53 Nakahata et al. 1985,54 Papayannopoulou et al. 1979,55 Greenberger et al. 1983,56 Humphries et al. 1981,57 Suda et al. 1984.58	Wierzbowska et al. 2010.59

Table 2.

Continued

	Gene ID		FPKM/Average				Literatures (TITLE/ABSTRACT must include the keywords)		Literature (IF is in the top 5)
Cell Type	Full Name	Abbreviation	iPSC	TM Progenitor	HTM	Keywords	Stem Cell Differentiation	Trabecular Meshwork	Stem Cell Differentiation	Trabecular Meshwork
HTM	TGF-beta receptor type-2	TGFBR2	2.8	50.0	163.2	Transforming growth factor	97	136	Bonauer et al. 2009,42 Jia et al. 2017,43 Wandzioch and Zaret 2009,44 Xu et al. 2002,45 Zhang et al. 2015.46	Agarwal and Agarwal 2018,47 Gabelt and Kaufman 2005,48 Liton et al. 2005;Luna et al. 2011,49 Raghunathan et al. 2015,50 Tamm 2002,51 Wahlig, Lovatt, and Mehta 2018.52
	Transforming growth factor beta receptor type 3	TGFBR3	2.4	12.6	13.4		97	136
	Prostaglandin F2-alpha receptor	PTGFR	0.1	2.6	11.4	Prostaglandin	7	105	Ludin et al. 2014,60 North et al. 2007,61 Sancilio et al. 2019,62 Wong et al. 2016,63 Zhang et al. 2017.64	Barraza, McLaren, and Poeschla 2010,65 Garnock-Jones 2014,66 Hoy 2018,67 Impagnatiello et al. 2019,68 Narayanaswamy et al. 2015.69
	Prostaglandin E2 receptor EP4 subtype	PTGER4	0.3	2.5	1.0		7	105
	EGF-like repeat and discoidin I-like domain-containing protein 3	EDIL3	16.1	77.4	152.4	Integrin	80	46	Cheng et al. 2019,70 Jiang et al. 2016,71 Aoyama et al. 1999,72 Kang et al. 2017,73 Mikkola et al. 2003,74 Przybyla, Lakins, and Weaver 2016.75	Clark et al. 2013, 76 Faralli, Filla, and Peters 2019,77 Inoue and Tanihara 2013,78 Kwon and Tomarev 2011,79 Yan et al. 2020.80
	Multiple epidermal growth factor-like domains protein 9	MEGF9	2.3	4.4	11.5	epidermal growth factor	29	7	Kratchmarova et al. 2005,81 Moore et al. 1997,82 Nakajima et al. 2007,83 Prabhakaran, Venugopal, and Ramakrishna 2009,84 Zhou et al. 2016.85	Acharya et al. 2012,86 Wiederholt, Groth, and Strauss 1998,87 Wordinger et al. 1998.88

Confirmation of Expression Modulus of the Selected Receptors by RT-PCR Analysis

We next determined the mRNA levels of these receptors in iPSCs of U1, U1-derived TM progenitor (1 month), and human TM cells of donor 4 and 5 by RT-PCR analysis. GAPDH was determined to be a better reference gene compared with 18S ribosomal RNA (18S) and actin beta (ACTB) through NormFinder analysis (Supplementary Fig. S2)⁸⁹ and was used as the reference gene in this study. Overall, the expression of receptors—human erythropoietin receptor (HEPOR), human TGF-β receptor 1 (HTGFBR1), HTGFBR2, HTGFBR3, PTPRN, MEGF9, prostaglandin F2-alpha receptor (PTGFR), EDIL3, and prostaglandin E receptors 1, 2, and 4 (PTGER1, 2, and 4)—in TM progenitors and human TM cells was higher than in iPSCs (Fig. 3). As expected, receptors with the highest FPKM values (Supplementary Fig. S3) were also detected with the highest expression in HTM cells by RT-PCR. However, receptors with the highest FPKM values in TM progenitors (HTGFBR1, HEPOR, and PTPRN) were not identified as being the most robustly expressed receptors in U1-derived TM progenitors.

Figure 3.

Relative mRNA levels of receptors associated with two-stage differentiation. RT-PCR analysis was used to examine the expression of human transforming growth factor-β receptor 1 (HTGFBR1), human erythropoietin receptor (HEPOR), protein tyrosine phosphatase receptor-type N (PTPRN) for stage 1, human transforming growth factor beta receptor 2 (HTGFBR2), human transforming growth factor beta receptor 3 (HTGFBR3), multiple epidermal growth factor-like-domains 9 (MEGF9), prostaglandin F2 alpha receptor (PTGFR), EGF-like repeat and discoidin I-like domain-containing protein 3 (EDIL3), and prostaglandin E receptors 1, 2, and 4 (PTGER1, 2, and 4) for stage 2. GAPDH was used as the reference gene. Data from three independent experiments (n = 3) were statistically analyzed by Tukey's multiple comparisons. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Xeno- and Feeder-Free Differentiation With Recombinant Cytokines

To determine the effects of the selected receptors on stem cell differentiation, recombinant cytokine ligands were purchased and used for iPSC differentiation (Table 3). After multiple tests, an optimal two-stage protocol was determined based on cell viability and morphology (Fig. 4A). The first stage is to induce iPSCs to differentiate for 7 days in the presence of TGF-β₁, NGF-β, and EPO. The second stage is to generate TM-resembling cells by using PGF2α and EGF for another 14 days. We first detected the cytokine concentrations in the conditioned medium by ELISA assay. As shown in Figure 4B, the ELISA results demonstrated a detectable supernatant concentration of TGF-β₁ (37.9 ± 3.2 pg/mL), NGF-β (45.8 ± 4.9 pg/mL), EPO (40.6 ± 0.7 pg/mL), EGF (54.3 ± 1.4 pg/mL), and PGF2α (39.4 ± 7.0 pg/mL). This observation indicated that these cytokines (TGF-β₁, NGF-β, EPO, EGF, and PGF2α) are retained at detectable levels in HTM conditioned media and might be important for iPSC differentiation.

Table 3.

Xeno- and Feeder-Free Differentiation by Recombinant Cytokines

Stage	Medium	Serum	High Receptors	Cytokines	Concentration	Differentiation Period
1	MEM-α	KnockOut Serum Replacement	TGFBR1	TGF-β1	2 ng/mL	7 d
			PTPRN	NGF-β	100 ng/mL
			EPOR	Erythropoietin	2 U/mL
2	MEM-α	KnockOut Serum Replacement	TGFBR2	TGF-β1	2 ng/mL	14 d
			TGFBR3
			PTPRN	NGF-β	100 ng/mL
			EPOR	Erythropoietin	2 U/mL
			PTGER	PGF2α	10 µM
			PTGFR
			MEGF9	EGF	10 ng/mL
			EDIL3

Figure 4.

Xeno- and feeder-free differentiation with recombinant cytokines. (A) Diagram of the new differentiation approach with two distinct stages. The first stage is to induce iPSCs to differentiate for 7 days by using TGF-β1, NGF-β, and EPO. Two more recombinant cytokines were involved in the differentiation of the second stage (14 days), including PGF2α and EGF. (B) ELISA analysis of the cytokines in the medium conditioned by HTM cells. Observable levels of TGF-β₁, NGF-β, EPO, EGF, and PGF2α, as the ligands of the selected receptors, were found in the conditioned medium of HTM cells (n = 3). The experiment was repeated three times by using HTM cells of three donors.

The morphological changes induced by this new protocol are similar to those observed using the traditional approach (Fig. 1A), including the generation of TM progenitors with a spindle-like shape and a smaller size (surface area, P < 0.0001; length, P < 0.0001) (Fig. 5B), progressing toward TM resembling morphology (Fig. 5B). Additionally, human iPSCs derived from different individuals (U1, U2, or U3) all displayed the same morphological changes during the differentiation (Fig. 5B).

Figure 5.

Morphological observations of human TM-like cells induced by cytokines. (A) Morphological changes in human iPSCs during differentiation. Human iPSCs from renal urethra epithelial cells of U1, U2, and U3 were differentiated into a TM-resembling cell type using recombinant cytokines. The TM progenitor at stage 1 exhibited a spindle-like shape with a smaller cell size than human TM cells. Cells at stage 2 displayed a TM-like morphology with a similar cell size compared with human TM cells of donors 4, 5, and 6. Scale bar: 100 µm. (B) Surface area and length of TM-like and HTM cells. TM progenitors were significantly smaller than HTM cells. Tukey's multiple comparisons test was used for data analysis. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Characterization of iPSC-Derived TM Cells

As described previously,^90,91 TM cells had a high expression of LAMA4, TIMP3, AQP1, MYOC, and ColIV and a capacity to form DEX-inducible CLANs. Immunohistochemistry analysis indicated that the differentiated iPSCs of U1, U2, and U3 using recombinant cytokines expressed high levels of LAMA4, TIMP3, AQP1, MYOC, and ColIV (Fig. 6A), similar to TM-like cells generated by using the conditioned medium (Fig. 1B). Additionally, DEX-inducible CLAN formation can be observed in TM-like cells generated using the new approach. As shown in Figure 6B, a significantly higher percentage of CLAN formation can be observed in DEX-treated differentiated iPSCs of U1, U2, and U3 than controls (P = 0.0013, P = 0.0023, and P < 0.0001, respectively), which exhibited a similar percentage as compared with DEX-treated human TM cells of donor 4 (19.69% ± 1.2 %). We further compared the transcriptomics properties of iPSC TM-like cells using two approaches and TM cells. Like our principle component analysis,⁹ U1, U2, and U3 TM-like cell-C is a distinct cell type compared with U1, U2, and U3 iPSCs and exhibits a more transcriptional similarity than TM cells (Fig. 6C). However, this new methodology detected only U2 TM-like cells with a transcriptomic similar to that of TM cells. U1 and U3 TM-like cell-X retained a similar molecular property compared with iPSCs.

Figure 6.

Characterization of TM-like cells derived from the new differentiation approach. (A) Immunohistochemical characterization of U1-, U2-, and U3-derived TM-like cells using recombinant cytokines. LAMA4, TIMP3, AQP1, MYOC, and ColIV were detected with robust expression in both TM-like cells and human TM cells. (B) CLAN formation of U1-, U2-, and U3-derived TM-like cells induced by DEX. Differentiated iPSCs of U1, U2, U3, and HTM from donor 4 treated with DMSO or 100-nM DEX were stained with Phalloidin-iFluor 488 (green) and DAPI (blue). Cells forming CLANs in the representative image of each sample are indicated by the red arrows (left panel). The percentage of CLAN formation was calculated from 16 images of each sample (right panel). One image with a detailed structure of CLANs is shown at a higher magnification. Scale bar: 100 µm. Data were statistically analyzed by two-tailed t-test. **P < 0.01, ***P < 0.001. (C) RNA sequencing was carried out in human urethra epithelial cell-derived iPSCs (U1, U2, and U3), conditioned medium-derived TM-like cells (U1, U2, and U3 TM-like cell-C), TM-like cells by xeno-free approach (U1, U2, and U3 TM-like cell-X), and TM cells (HTM1 and HTM2). Cluster analysis was performed using differentially expressed genes (DEGs) with fold differences ≥ 2 (|log₂FC|≥1) and P < 0.05. TM-like cell-C is a distinct cell type relevant to iPSC that exhibits a more transcriptional similarity to TM cells. U2 TM-like cell-X was detected with a transcriptomic similar to that of TM cells, but U1 and U3 TM-like cell-X retained molecular properties similar to those of iPSCs.

Discussion

Because the iPSC-derived TM-like cells are a promising cell type to restore IOP hemostasis and AH outflow in glaucoma,^8,92 it is crucial to develop a new differentiation approach to generate TM-like cells that are more feasible for clinical usage. According to our RNA sequencing data (Fig. 2), numerous stem cell fate-related signals are changed (data not shown). We observed that the ligands relevant to these signals would eventually function through binding to the receptors, and the receptors would be controlled by multiple feedback mechanisms to ensure an appropriate level during iPSC differentiation.⁹³ We thus selected the highly expressed receptors relevant to stem cell differentiation (Table 2) to establish the new methodology. Although RT-PCR results using U1 iPSCs showed some difference compared with previous RNA sequencing data (Fig. 3; Supplementary Fig. S3), new RNA sequencing data using urine-derived iPSCs also exhibited robust expression of HTGFBR1, HEPOR, and PTPRN in TM progenitors and high expression of HTGFBR, MEGF, PTGFR, PTGFR, EDIL3, and PTGER in TM cells (Supplementary Figs. S3 and S4). We thus have established a new methodology with the ligands relative to the above receptors (Fig. 4).

Compared with the traditional approach,¹³ we first improved the origin of iPSCs, renal urethra epithelial cells. Compared to dermal fibroblast cells from skin used in our previous study,^8,9 renal urethra epithelial cells are easier to collect from urine and more feasible for use in clinical settings. Additionally, iPSCs are generated through an integration-free approach by Sendai virus carrying Oct4, Sox2, c-Myc, and Klf4 transcription factors. This is particularly important for clinic usage, as this approach avoids the harmful effects of leaky transgene expression and insertional mutagenesis.⁹⁴

Another benefit of this new approach is a shorter differentiation time. Within 1 week undifferentiated iPSCs start to exhibit a TM-resembling morphology with a spindle-like shape (Fig. 5), which is the most critical period for changing cell fate and morphology (Fig. 1).⁹ The subsequent stage of differentiation (2 weeks) is mainly responsible for cell growth until the cells reach the same size as HTM cells (Fig. 5). Compared to 2-month differentiation using a conditioned medium, the new approach is much more efficient for generating TM-resembling cells. The reason for the accelerated development might be a higher concentration of recombinant cytokines achieved in the new approach compared with the cytokines secreted by human TM cells. Shortening the differentiation period is particularly critical for avoiding the side effects of long-term cell culture, such as cell senescence and genetic mutation.^16,17

The third vital improvement offered by this new approach is avoiding contamination from bovine serum and human cell cultures. In this study, we used the KnockOut Serum Replacement, a substitute for FBS that contains the essential supplements for iPSCs growth.¹⁹ Additionally, recombinant cytokine use avoids complications resulting from the use of human donor materials, such as media conditioned from human beings. These improvements potentially avoid the variabilities of serum obtained from different batches or among different human donors.

Most importantly, our characterization indicates that TM-like cells are a TM-resembling cell type (Figs. 5 and 6). We also found that the reported TM biomarkers⁹⁵ SERPINA3, PTGDS, MYOC, FCGBP, HBB, and HLA-DRA were not only robustly expressed in TM cells but also highly expressed in TM-like cells, whether by the traditional approach or using the cytokine-based approach (Supplementary Fig. S5). The reported TM progenitor biomarkers⁹⁵ RPLP0, RAB3B, MARCH4, FOXD1, TSPOAP1, and WDR1 were still detected, with the most robust expression occurring in our TM-like cells. Taken together, our TM-like cells are not progenitor-resembling cells but have greater similarity with TM cells.

We are aware of some possible issues associated with large-scale production of TM-like cells using this approach. Apoptosis during the iPSC differentiation remains a fundamental issue regardless of whether conditioned medium or recombinant cytokines are used. One reason for this may be the withdrawal of leukemia inhibitory factor during differentiation, which has been reported to cause the death of more than 30% of stem cells.^96,97 These apoptotic cells are cells that fail to exit self-renewal.⁹⁸ Apoptosis is necessary and benefits stem cell differentiation, which is essential to promoting differentiation efficiency.⁹⁹ However, disruption of the balance between survival and death of iPSCs during differentiation can lead to excessive cell viability loss, which can be problematic for large-scale production of TM-like cells. Another issue is the transcriptional difference between urine-derived TM-like cells and HTM cells, despite the success of one donor-derived iPSC differentiation (Fig. 6). This is surprising, as the TM-like cells exhibited a distinct cell morphology compared with stem cell colonies. In view of the differentiation, this might be due to different iPSC viabilities. At the early stage of differentiation, U1 and U3 iPSCs require more time to reach 100% confluency compared with U2. As shown in Supplementary Figure S6, after two-stage differentiation, U1 and U3 TM-like cells exhibited lower confluency than U2-derived cells. Our results suggest that cell viability would be necessary for improving differentiation efficiency using the cytokine-based approach.

Conclusions

This study identified a practical iPSC differentiation approach to generating TM-like cells that is more feasible for clinic usage. We first defined two distinct stages of iPSC differentiation and selected highly expressed receptors at each stage of differentiation, and we have developed a new type of differentiation using the corresponding recombinant cytokines. By using this approach, human iPSCs from three individuals were all successfully differentiated into a TM-like cell type with robust expression of TM markers and the ability for DEX-induced CLAN formation.