Abstract
There are >120 forms of non-syndromic deafness associated with identified genetic loci. In particular, mutation of the gap junction beta 2 gene (GJB2), which encodes connexin (CX)26 protein, is the most frequent cause of hereditary deafness worldwide. We previously described an induction method to develop functional CX26 gap junction-forming cells from mouse-induced pluripotent stem cells (iPSCs) and generated in vitro models for GJB2-related deafness. However, functional CX26 gap junction-forming cells derived from human iPSCs or embryonic stem cells (ESCs) have not yet been reported. In this study, we generated human iPSC-derived functional CX26 gap junction-forming cells (iCX26GJCs), which have the characteristics of cochlear supporting cells. These iCX26GJCs had gap junction plaque-like formations at cell–cell borders and co-expressed several markers that are expressed in cochlear supporting cells. Furthermore, we generated iCX26GJCs derived from iPSCs from two patients with the most common GJB2 mutation in Asia, and these cells reproduced the pathology of GJB2-related deafness. These in vitro models may be useful for establishing optimal therapies and drug screening for various mutations in GJB2-related deafness.
Introduction
Hearing loss is the most common congenital sensory impairment worldwide (1). Approximately 1 child in 1000 is born with severe hearing loss or will develop hearing loss during early childhood, which is known as prelingual deafness (2,3), and about half of such cases are attributable to genetic causes (4). To date, there are >120 known forms of non-syndromic deafness associated with identified genetic loci (available at http://hereditaryhearingloss.org), and the types of cells associated with the disease are diverse. In particular, mutations in gap junction beta 2 (GJB2), which encodes connexin (CX)26 protein, account for up to 50% of cases of non-syndromic sensorineural hearing loss in some populations (5), and >150 different GJB2 variants have been identified, including missense, nonsense and frameshift mutation (http://davinci.crg.es/deafness/). CX26 protein is observed in cochlear non-sensory cells and structures, including the supporting cells, spiral limbus, stria vascularis and spiral ligament, where it forms intercellular gap junctions (GJs) (6–11). By contrast, connexins are not expressed in cochlear hair cells (8,10–12). Connexins hetero-oligomerize to form hexameric channels called connexons (hemichannel), and a hemichannel at the plasma membrane of a cell docks with a partner hemichannel in a neighboring cell to form a GJ channel. GJs cluster into semicrystalline arrays that comprise tens to thousands of GJs to form a gap junction plaque (GJP) (13). In the mammalian cochlea, GJs facilitate the rapid removal of K+ from the base of cochlear hair cells, resulting in K+ recycling back to the endolymph to maintain cochlear homeostasis (14).
Previously, we generated mouse models of GJB2-related deafness and demonstrated that a mutation in GJB2 resulted in a drastic disruption of CX26/CX30 macromolecular complex and decreased gap junction intercellular communication (GJIC) in the cochlea (15). Several studies, including our study, have reproduced the ‘reduction in GJIC’, which represents the pathology of GJB2-related deafness in humans, using transfected HeLa, HEK293T and COS-7 cells (15–20). Thus, cochlear cells that form CX26 GJPs may thus represent the most important target for the treatment of hereditary deafness. The cochlea is, however, anatomically complex, is not readily accessible from the outside, and has a small number of target cells, making it more difficult to study than other sensory organs. In addition, because the cochlea is filled with lymph, an invasive procedure such as a biopsy or direct drug administration can cause irreversible hearing loss. Therefore, stem cells, for example embryonic stem cells (ESCs) and induced pluripotent stem cells (iPSCs), are an important tool for studying the molecular mechanisms underlying the inner-ear pathology as well as for generating cells for replacement therapies. Several groups have reported that mouse and human ESCs/iPSCs can be differentiated into inner-ear cells, including hair cells, supporting cells, progenitor cells and sensory neurons by in vitro differentiation in adherent culture and/or floating aggregation culture (21–28). Previously, we described the generation of CX26 gap junction-forming cells from iPSCs (induced CX26 gap junction-forming cells; hereafter, iCX26GJC) derived from mice by using a floating culture (serum-free culture of embryoid body-like aggregates with quick aggregation culture; hereafter, SFEBq culture) and adherent culture systems. These cells had characteristics and functions that were observed in the developing cochlea (29). However, no method has yet been reported that differentiates human iPSCs into supporting cells of the inner ear that form CX26 GJPs and carry out their related functions. In addition, patient-derived in vitro models for GJB2-related deafness have not been reported.
Here we developed a method for inducing functional CX26 gap junction-forming cells from human iPSCs that have characteristics of inner-ear supporting cells. Furthermore, we applied these induction methods to patient-derived iPSCs and reproduced the pathology of GJB2-related deafness.
Results
Differentiation of human iPSCs into CX26-expressing cells
The method for the induction of iCX26GJCs from human iPSCs, which is a combination of SFEBq culture and adherent culture techniques, was modified from our previous method for mouse iPSCs (29,30), and the conditions required for differentiation were then assessed. In SFEBq cultures, human iPSCs were re-aggregated (9000 cells/well) and cultured in maintenance medium (StemFit AK02N) for 2 days (day −2 to day 0). Aggregates were then transferred to growth factor-free chemically defined medium (gfCDM; Supplementary Material, Table S1). At day 3 in the SFEBq culture, we added BMP4 (BMP) and/or Activin/Nodal/TGF-β pathway inhibitor (SB431542:SB) to gfCDM according to the mouse iCX26GJC induction method (Supplementary Material, Fig. S1A). However, the addition of these supplements did not increase GJB2/GJB6 mRNA expression and CX26-positive cell mass (CX26+ vesicles) production. On the contrary, the addition of SB induced a decrease in GJB2 mRNA expression and CX26+ vesicle production (Supplementary Material, Fig S1B–J). In an SFEBq/gfCDM culture, the addition of insulin to human ESC improved their survival and growth (31). In addition, several protocols, which efficiently induced ectodermal-derived tissue from human ESCs/iPSCs, added a high concentration of insulin (7 μg/ml: equivalent to insulin contained in 10% KSR) to gfCDM in the SFEBq culture (25,31–33). From these results and reports, in the following experiments we added insulin (7 μg/ml) to gfCDM from day 0. The modified SFEBq culture for the differentiation of human iPSCs into CX26-expressing cells is shown in Figures 1A (SFEBq culture) and 3A (adherent culture).
Treatment with insulin induces GJB2/GJB6 genes and otic progenitor marker genes in iPSCs. (A) Procedure for the differentiation of CX26-expressing cells from human iPSCs in the SFEBq culture. (B) Quantitative PCR analysis on undifferentiated human iPSCs (day 0) and iPSC-derived aggregate (day 7) for the expression of NANOG, GJB2, GJB6, PAX2, PAX8 and GATA3 (n = 4 from two to four independent experiments). mRNA expression levels were calculated relative to day 7 aggregate without insulin (Ins(−)). SF, StemFit AK02N; Y, Y-27632; Ins, insulin. The statistical difference was determined by Scheffe’s multiple comparison test, mean ± SE; *P < 0.05; **P < 0.01.
Screening for high CX26 expression in SFEBq cultures
In the modified SFEBq culture, aggregates were collected on day 7, and mRNAs (GJB2, GJB6, PAX2, PAX8 and GATA3) in the different culture groups were measured. The GJB6 gene encodes CX30 protein, which is co-expressed with CX26 in the cochlear supporting cells (7,10). PAX2, PAX8 and GATA3 gene sets have been used as otic progenitor markers in several studies aimed at the differentiation of ESC/iPSC into inner-ear cells (21,22,25,34). In day 7 aggregates, GJB2, GJB6, PAX2, PAX8 and GATA3 genes were more upregulated than those in undifferentiated iPSC (Day 0). Furthermore, the expressions of PAX2, PAX8 and GATA3 genes were significantly increased as compared with undifferentiated iPSC (Fig. 1B). The iPSCs cultured in gfCDM supplemented with insulin showed significantly higher expression of these mRNAs (GJB2, 20.3-fold increase; GJB6, 13.7-fold increase; PAX8, 1.7-fold increase; GATA3, 304.0-fold increase) relative to those cultured without insulin (Fig. 1B). In the day 7 aggregate without insulin, cellular debris was observed around the aggregate. In contrast, no debris was observed when insulin was added to the aggregate (Supplementary Material, Fig. S2A). In addition, the diameter of the day 7 aggregate was significantly larger with insulin (mean = 876.6 ± 6.90 μm) than that without insulin (mean = 644.8 ± 24.58 μm) (Supplementary Material, Fig. S2B). To analyze the localization of CX26 in iPSC aggregates, we performed immunohistochemistry with day 7 aggregates. In the day 7 aggregates with or without insulin, CX26 expressing cell masses (CX26+ vesicle) were observed (Fig. 2A–D). We compared the number of CX26+ vesicles among the different treatment groups. Cells treated with insulin had significantly more CX26+ vesicles (mean = 2.7 ± 0.21) than cells cultured without insulin (mean = 1.7 ± 0.37) (Fig. 2E). On the other hand, addition of SB and/or BMP to gfCDM with insulin did not increase mRNA expression and CX26+ vesicle production (Supplementary Material, Fig. S3A–J). Concerning the diameter of the CX26+ vesicles, cells treated with insulin (mean = 161.3 ± 10.83 μm) were significantly larger than those without insulin (mean = 129.2 ± 4.70 μm) (Fig. 2F). In addition, these CX26+ vesicles were composed of 281.9 ± 41.4 cells (with insulin) or 184.2 ± 12.07 cells (without insulin), respectively (Fig. 2G). The proportions of CX26 positive cells (CX26 (+)) and CX26 negative cells (CX26(−)) that consisted of aggregates were 4.18 and 95.8% for insulin (−) aggregates and 8.86 and 9.13% for insulin (+) aggregates, respectively (Fig. 2H). In the confocal analysis of the insulin-treated day 7 aggregates, CX26-expressing cells were disseminated throughout the CX26+ vesicles (Fig. 2I–N). These cells formed CX26+ GJs at their cell–cell borders (Fig. 2J, K, M and N). In the three-dimensional reconstruction of the confocal images, we observed planar CX26-containing GJPs (Fig. 2, O and P and Supplementary Material, Video 1), as reported in our previous study in mice (29).
Treatment with insulin induces CX26 expression cells in iPSCs. (A–D) Immunostaining for CX26 (red) in day 7 aggregates. The arrows indicate CX26+ vesicles. The boxed regions in (A) and (B) are magnified in (C) and (D), respectively. (E) Number of CX26+ vesicles per mm3 of day 7 aggregate (n = 9 or 15 aggregates from two or three independent experiments). (F) Average diameter of CX26+ vesicles in day 7 aggregate (n = 8 or 20 CX26+ vesicles from three independent experiments). (G) Average number of cells composed CX26+ vesicle in day 7 aggregates (n = 12 CX26+ vesicles from three independent experiments. (H) Percentage of CX26 positive cells vs. CX26 negative cells in day 7 aggregate (n = 9 or 14 from day 7 aggregates from two to three independent experiments). White column is CX26 negative (CX26−) cells. Black column is CX26 positive (CX26+) cells. (I–N) Immunostaining for CX26 (red) and F-ACTIN (green) on CX26+ vesicles in day 7 aggregates. Nuclear counterstaining with DAPI (blue in L–N). The boxed regions in (I), (J), (L) and (M) are magnified in (J), (K), (M) and (N), respectively. (O and P) A three-dimensional image showing the GJPs was reconstructed from the image in (M). Arrowheads point to GJPs. Scale bars: 200 μm (A and B); 50 μm (C and D); 20 μm (I and L); 10 μm (J, M, and P); 5 μm (K, N, and O). The statistical difference was determined by Student’s t test, mean ± SE; *P < 0.05; **P < 0.01.
The iCX26GJC expressed typical cochlear cell markers
The regions with iCX26GJC-containing CX26+ vesicles were separated from aggregates on day 7–11 and were transferred in DMEM/Ham’s F12 supplemented with N2/B27 (DFNB medium; Supplementary Material, Table S1) to mouse cochlea-derived feeder cells (trypsin-resistant inner-ear cells, TRICs) (Fig. 3A). The transferred iCX26GJC-containing regions indeed formed colonies on the TRIC feeder cells, and the colonies contained iCX26GJCs (Fig. 3B and C). These proliferating cells formed CX26-positive GJs at their cell–cell borders (Fig. 3D–H). In the three-dimensional construction of the confocal images, we observed large planar CX26-containing GJPs (Fig. 3I–K and Supplementary Material, Video 2). To determine whether the iCX26GJCs were similar to cochlear supporting cells, we examined the expression of the following proteins (CX30, SOX2, SPARCL1, KCC3 and some cytokeratins) and genes (SOX2, SPARCL1, KIAA1199, MIA and OTOR) that were observed in cochlear supporting cells and/or fibrocytes (see Supplemental Information). We summarized the immunostaining and qPCR results for iCX26GJCs in Tables 1 and 2. In human iCX26GJC, CX30 and CX26 expressions were observed in the same cells, but these proteins did not necessarily co-assemble in gap junction plaques (Fig. 4A). In the fibrocytes of human cochlear spiral ligament, CX26/CX30 proteins form separate GJPs and do not necessarily co-assemble in GJPs (35). However, it is unclear whether these proteins form gap junction plaques in cochlear supporting cells as in the cochlear fibrocytes. On the other hand, in rodents, the expression pattern of CX26/CX30 proteins is different between cochlear supporting cells (form hexagon) and lateral wall fibroblasts (do not form polygon) (9,15). From the above text, it cannot be concluded that, in human cochlear supporting cells, which were our target, CX26/CX30 protein does not co-assemble the GJP as in the fibrocytes of the spiral ligament. Furthermore, iCX26GJC co-expressed SOX2, SPARCL1, KCC3, P-CK, CK8 and CK18 proteins (Fig. 4B–G). In contrast, immunolabeling for CK5, CK10 and CK14, which are skin markers, was not detected in iCX26GJCs (Supplementary Material, Fig. S4). Furthermore, GJB2, GJB6, KIAA1199, SPARCL1, MIA and OTOR genes were significantly upregulated compared with undifferentiated iPSC (Fig. 4H). On the other hand, the expression levels of NANOG, which is an undifferentiated marker, significantly decreased compared with undifferentiated iPSC.
Proliferation of iCX26GJCs on adherent cultures. (A) Procedure for the proliferation of iCX26GJCs from human iPSC aggregates using adherent culture. (B) Phase contrast microscope (PCM) images from adherent cultures at day 21. (C) Staining for CX26 (red) and PCM (white). Magnification of boxed region in (B). (D–G) Staining for CX26 (red) and F-ACTIN (green). The boxed regions in (D), (E) and (F) are magnified in (E), (F) and (G), respectively. (H and I) Staining for CX26 (red) and DAPI (blue). (H) and (I) are same regions in (F) and (G). The boxed regions in (H) are magnified in (I). (J and K) A three-dimensional image showing the GJPs was reconstructed from the image in (I). The arrowheads point to GJPs. Scale bars: 200 μm (B); 100 μm (D); 50 μm (C and E); 10 μm (F and H); 5 μm (G and I-K).
Summary of protein expression in iCX26GJCs and mammalian cochlear cells
| Protein name | iCX26GJCs | Cochlea | ||
|---|---|---|---|---|
| SCs | HCs | |||
| CONNEXIN 26a | + | + | − | |
| CONNEXIN 30a | + | + | − | |
| SOX2a | +/− | +/− | − | |
| SPARCL1b | +/− | +/− | − | |
| SLC12A6 (KCC3)c | +/− | +/− | − | |
| CYTOKERATINd | Pan | + | N/A | N/A |
| 5 | − | − | − | |
| 8 | + | + | − | |
| 10 | − | − | − | |
| 14 | − | − | − | |
| 18 | + | + | − | |
| Protein name | iCX26GJCs | Cochlea | ||
|---|---|---|---|---|
| SCs | HCs | |||
| CONNEXIN 26a | + | + | − | |
| CONNEXIN 30a | + | + | − | |
| SOX2a | +/− | +/− | − | |
| SPARCL1b | +/− | +/− | − | |
| SLC12A6 (KCC3)c | +/− | +/− | − | |
| CYTOKERATINd | Pan | + | N/A | N/A |
| 5 | − | − | − | |
| 8 | + | + | − | |
| 10 | − | − | − | |
| 14 | − | − | − | |
| 18 | + | + | − | |
Summary of protein expression in iCX26GJCs and mammalian cochlear cells
| Protein name | iCX26GJCs | Cochlea | ||
|---|---|---|---|---|
| SCs | HCs | |||
| CONNEXIN 26a | + | + | − | |
| CONNEXIN 30a | + | + | − | |
| SOX2a | +/− | +/− | − | |
| SPARCL1b | +/− | +/− | − | |
| SLC12A6 (KCC3)c | +/− | +/− | − | |
| CYTOKERATINd | Pan | + | N/A | N/A |
| 5 | − | − | − | |
| 8 | + | + | − | |
| 10 | − | − | − | |
| 14 | − | − | − | |
| 18 | + | + | − | |
| Protein name | iCX26GJCs | Cochlea | ||
|---|---|---|---|---|
| SCs | HCs | |||
| CONNEXIN 26a | + | + | − | |
| CONNEXIN 30a | + | + | − | |
| SOX2a | +/− | +/− | − | |
| SPARCL1b | +/− | +/− | − | |
| SLC12A6 (KCC3)c | +/− | +/− | − | |
| CYTOKERATINd | Pan | + | N/A | N/A |
| 5 | − | − | − | |
| 8 | + | + | − | |
| 10 | − | − | − | |
| 14 | − | − | − | |
| 18 | + | + | − | |
Summary of gene expression in iCX26GJCs
| Gene Symbol | Gene name | Fold difference (hiPSC vs iCX26GJC) | p-value | Reference for gene or protein expression in cochlea |
|---|---|---|---|---|
| GJB2 | Gap junction protein beta-2 | 8.23 ± 1.20 | 3.23E-04 | Liu et al., 2016* (35) |
| GJB6 | Gap junction protein beta-6 | 25.43 ± 5.23 | 0.003 | Liu et al., 2016* (35) |
| NANOG | Nanog homeobox | 0.006 ± 0.0007 | 9.09E-28 | − |
| SOX2 | SRY-box transcription factor 2 | 0.67 ± 0.086 | 0.003 | Locher et al., 2013* (54) |
| KIAA1199 (CEMIP or HYBID) | Cell migration inducing hyaluronidase 1(CEMIP), Hyaluronan binding protein involved in hyaluronan depolymerization (HYBID) | 74.37 ± 23.10 | 1.15E-05 | Hosoya et al, 2016 (55) |
| SPARCL1 | SPARC-like protein 1 | 216.49 ± 43.16 | 5.43E-04 | Burns et al., 2015 (56) |
| MIA | Melanoma inhibitory activity | 5.52 ± 0.96 | 0.002 | Schrauwen et al., 2016* (50) |
| OTOR | Otoraplin | 11.42 ± 0.82 | 1.43E-05 | Schrauwen et al., 2016* (50) |
| Gene Symbol | Gene name | Fold difference (hiPSC vs iCX26GJC) | p-value | Reference for gene or protein expression in cochlea |
|---|---|---|---|---|
| GJB2 | Gap junction protein beta-2 | 8.23 ± 1.20 | 3.23E-04 | Liu et al., 2016* (35) |
| GJB6 | Gap junction protein beta-6 | 25.43 ± 5.23 | 0.003 | Liu et al., 2016* (35) |
| NANOG | Nanog homeobox | 0.006 ± 0.0007 | 9.09E-28 | − |
| SOX2 | SRY-box transcription factor 2 | 0.67 ± 0.086 | 0.003 | Locher et al., 2013* (54) |
| KIAA1199 (CEMIP or HYBID) | Cell migration inducing hyaluronidase 1(CEMIP), Hyaluronan binding protein involved in hyaluronan depolymerization (HYBID) | 74.37 ± 23.10 | 1.15E-05 | Hosoya et al, 2016 (55) |
| SPARCL1 | SPARC-like protein 1 | 216.49 ± 43.16 | 5.43E-04 | Burns et al., 2015 (56) |
| MIA | Melanoma inhibitory activity | 5.52 ± 0.96 | 0.002 | Schrauwen et al., 2016* (50) |
| OTOR | Otoraplin | 11.42 ± 0.82 | 1.43E-05 | Schrauwen et al., 2016* (50) |
*Gene or protein expression pattern has been confirmed in the human cochlea hiPSC: undifferentiated human iPSC. iCX26GJC: 2D culture at Day 21. The mRNA expression levels were calculated relative to hiPSC.
Summary of gene expression in iCX26GJCs
| Gene Symbol | Gene name | Fold difference (hiPSC vs iCX26GJC) | p-value | Reference for gene or protein expression in cochlea |
|---|---|---|---|---|
| GJB2 | Gap junction protein beta-2 | 8.23 ± 1.20 | 3.23E-04 | Liu et al., 2016* (35) |
| GJB6 | Gap junction protein beta-6 | 25.43 ± 5.23 | 0.003 | Liu et al., 2016* (35) |
| NANOG | Nanog homeobox | 0.006 ± 0.0007 | 9.09E-28 | − |
| SOX2 | SRY-box transcription factor 2 | 0.67 ± 0.086 | 0.003 | Locher et al., 2013* (54) |
| KIAA1199 (CEMIP or HYBID) | Cell migration inducing hyaluronidase 1(CEMIP), Hyaluronan binding protein involved in hyaluronan depolymerization (HYBID) | 74.37 ± 23.10 | 1.15E-05 | Hosoya et al, 2016 (55) |
| SPARCL1 | SPARC-like protein 1 | 216.49 ± 43.16 | 5.43E-04 | Burns et al., 2015 (56) |
| MIA | Melanoma inhibitory activity | 5.52 ± 0.96 | 0.002 | Schrauwen et al., 2016* (50) |
| OTOR | Otoraplin | 11.42 ± 0.82 | 1.43E-05 | Schrauwen et al., 2016* (50) |
| Gene Symbol | Gene name | Fold difference (hiPSC vs iCX26GJC) | p-value | Reference for gene or protein expression in cochlea |
|---|---|---|---|---|
| GJB2 | Gap junction protein beta-2 | 8.23 ± 1.20 | 3.23E-04 | Liu et al., 2016* (35) |
| GJB6 | Gap junction protein beta-6 | 25.43 ± 5.23 | 0.003 | Liu et al., 2016* (35) |
| NANOG | Nanog homeobox | 0.006 ± 0.0007 | 9.09E-28 | − |
| SOX2 | SRY-box transcription factor 2 | 0.67 ± 0.086 | 0.003 | Locher et al., 2013* (54) |
| KIAA1199 (CEMIP or HYBID) | Cell migration inducing hyaluronidase 1(CEMIP), Hyaluronan binding protein involved in hyaluronan depolymerization (HYBID) | 74.37 ± 23.10 | 1.15E-05 | Hosoya et al, 2016 (55) |
| SPARCL1 | SPARC-like protein 1 | 216.49 ± 43.16 | 5.43E-04 | Burns et al., 2015 (56) |
| MIA | Melanoma inhibitory activity | 5.52 ± 0.96 | 0.002 | Schrauwen et al., 2016* (50) |
| OTOR | Otoraplin | 11.42 ± 0.82 | 1.43E-05 | Schrauwen et al., 2016* (50) |
*Gene or protein expression pattern has been confirmed in the human cochlea hiPSC: undifferentiated human iPSC. iCX26GJC: 2D culture at Day 21. The mRNA expression levels were calculated relative to hiPSC.
Immunolabeling and gene expression of known cochlear markers in iCX26GJCs grown in adherent cultures. (A–G) Staining for CX26 (red) and CX30 (A, green), SOX2 (B, green), SPARCL1 (C, green), SLC12A6 (KCC3; D, green), Pan-cytokeratin (P-CK; E, green), CK8 (F, green) and CK18 (G, green). Nuclei were counterstained with DAPI (blue). (H) Quantitative PCR analysis on undifferentiated human iPSCs (hiPSC) and iCX26GJCs for the expression of NANOG, SOX2, KIAA1199, SPARCl1, MIA and OTOR (n = 4 from two to three independent experiments). mRNA expression levels were calculated relative to undifferentiated human iPSC (hiPSC). Scale bars: (A–G) 20 μm, (A: second, third and fourth column) 10 μm. The statistical difference was determined by Student’s t test, mean ± SE; *P < 0.05; **P < 0.01.
Although the expression level of the SOX2 gene, which is an undifferentiated marker the same as NANOG, decreased, it was not as drastically reduced as NANOG. It is known that cochlear non-sensory cells include cells that express SOX2 (inner phalangeal cells, inner pillar cells, outer pillar cells, Deiters’ cells and Hensens’ cells) (36) and those that do not express SOX2 (inner-, Outer sulcus cells, fibrocytes). In addition, in the mouse iPSC-derived iCX26GJC that we examined in our previous study, we had observed SOX2 expressing or not expressing cells, as in the cochlear supporting cells (29). From the above text, we considered that a significant decrease in mRNA expression like NANOG would not be observed in human iPSC-derived iCX26GJC.
The formation of CX26/CX30 gap junction plaque in iCX26GJCs derived from healthy and diseased iPSCs
We generated the disease-specific iCX26GJCs from two iPSC lines derived from PBMCs from sibling patients carrying a homozygous 235delC mutation in GJB2 (Fig. 5A). The 235delC mutation is the most common GJB2 mutation in Asia, including Japan, and results in an audiometric phenotype of severe-to-profound hearing loss (37–39). The two human iPSC lines, GP5-235delC/235delC GP6-235delC/235delC (hereafter GP5-235delC and GP6-235delC, respectively; audiometric phenotypes of profound hearing loss are shown in Fig. 5B and C) have been characterized previously (specific cell line names: JUFMDOi005-A and JUFMDOi006, respectively) (40).
CX30 GJP formation and GJP length in normal or patient iPSC-derived iCX26GJCs. (A) A family history of the individuals from whom these iCX26GJCs were derived. The squares indicate male family members; the circle indicates the female member; filled shapes indicate family members diagnosed with GJB2-related deafness. (B and C) Audiometric phenotypes for Patient 1 (B) and Patient 2 (C) show profound hearing loss. (D) GJP formation in normal (201B7) or patients (GP05-235delC, GP06-235delC) iPSC-derived iCX26GJC. These samples were co-labeled with anti-CX26 (red) and anti-CX30 (green) antibodies. (E) Length of the largest GJPs along a single cell border (mean ± SE, n = 27, 30, 37 cell borders from three to four independent experiments). The statistical difference was determined by Scheffe’s multiple comparison test, mean ± SE; *P < 0.05; **P < 0.01.
We confirmed for CX26 and CX30 by immunostaining to determine whether disruption of the CX26/CX30 macromolecular complex (15,29), a pathology of GJB2 deafness that was observed in a mouse model, is also observed in patient iPSC-derived iCX26GJCs. Healthy iPSC (201B7) derived iCX26GJC showed large, planar CX30 GJPs at the cell border (Fig. 5D, left column). In contrast, in patient iPSC (GP5-235delC, GP6-235delC) derived iCX26GJCs showed partially shortened GJPs (Fig. 5D, middle and right column). As shown in Figure 5E, the GJP lengths in diseased iPSC-derived iCX26GJCs (GP5-235delC-iCX26GJC: 6.13 ± 0.35 μm; GP6-235delC-iCX26GJC: 5.91 ± 0.37 μm) were significantly shorter than those of normal iPSC-derived iCX26GJC (8.62 ± 0.31 μm).
Functional evaluation of gap junction intercellular communication (GJIC) in iCX26GJCs derived from healthy and diseased iPSCs
To investigate whether there are differences in the function of GJIC networks between healthy iCX26GJCs and diseased iCX26GJCs, we performed scrape loading/dye transfer (SL/DT) with Lucifer Yellow (29,41). SL/DT was performed on iCX26GJCs derived from iPSCs from a healthy individual (201B7-iCX26GJC) and on GP5-235delC-iCX26GJC and GP6-235delC-iCX26GJC. Undifferentiated iPSCs (201B7) and feeder cells (TRICs) were included for comparison. We quantified the extent of dye transfer by measuring the distance from the scrape line to the point at which the fluorescence intensity dropped to the background fluorescence intensity. In these iCX26GJC cultures, we observed that Lucifer Yellow diffused beyond the wounded parental cells (Fig. 6G, H, J, K, M and N), indicating the presence of GJIC. In contrast, dye transfer to this extent was not observed in undifferentiated iPSCs or TRIC feeder cells (Fig. 6A, B, D and E). As shown in Figure 6P, the quantitative distance of dye transfer in 201B7-iCX26GJCs (114.3 ± 4.05 μm), GP5-235delC-iCX26GJCs (36.4 ± 1.55 μm) and GP6-235delC-iCX26GJCs (33.3 ± 0.69 μm) was significantly longer than that in undifferentiated iPSCs (22.4 ± 0.65 μm) or TRIC feeder cells (23.8 ± 0.99 μm). Furthermore, in both patient-derived iCX26GJC (GP5-235delC-iCX26GJC and GP06-235delC-iCX26GJC), the distance of dye transfer was significantly shorter than that of iCX26GJCs derived from iPSCs without the GJB2 mutation (201B7-iCX26GJC).
Dye Transfer after Scrape-Loading of Cells and Subsequent Quantitative Analysis. (A–O) Digital fluorescence images of cultured cells after scrape loading. The undifferentiated healthy human iPSCs (201B7) (A–C), feeder cells (TRICs) (D–F), iPSC-derived adherent cultures containing iCX26GJCs derived from healthy human iPSCs (201B7-iCX26GJCs) (G–I) as the healthy control, and iCX26GJCs derived from Patient 1 iPSCs (GP5-235delC-iCX26GJCs) (J-L) and from Patient 2 iPSCs (GP6-235delC-iCX26GJCs) (M–O). (A, D, G, J, M) Dye transfer using Lucifer Yellow. (B, E, H, K, N) Pseudocolor images indicating the range of transfer from low (black) to high (red) signal intensity in the same region as shown in the above images. (C, F, I, L, O) Phase contrast microscopy of the same region as shown in the above images. (P) Quantitative analysis of intercellular dye transfer after scrape loading. Columns represent the mean distance of dye transfer from the scrape line (TRICs and undifferentiated 201B7: n = 40 from five independent experiments; 201B7-iCX26GJCs, GP5-235delC-iCX26GJCs and GP6-235delC-iCX26GJCs: n = 40 from two or three independent experiments). The statistical difference was determined by Scheffe’s multiple comparison test, mean ± SE; Different letters (a–c) represent significant differences, P < 0.01. Scale bars represent 50 μm.
Discussion
Efficient method for differentiation of human iPS cells into iCX26GJC
In this study, we generated CX26 gap junction-forming cells (iCX26GJC), which have the characteristics of cochlear supporting cells, from human iPSCs by using the SFEBq culture and adherent culture as previously described (29). In addition, we applied this induction method to patient-derived iPSCs and reproduced the pathology of GJB2-related deafness.
The SFEBq culture system is the most suitable method for inducing various ectoderm-derived tissues, for example, forebrain, midbrain, hindbrain, optic cup and the otic cup from ESCs/iPSCs (25,31,42,43). The use of gfCDM and the addition of insulin in SFEBq cultures promote differentiation into the midbrain and hindbrain regions and increases these regions contained in the embryoid body (31,43–45). The inner ear, which was our target, is derived from the otic placode, which is part of the surface of the non-neuroectoderm (NNE) adjacent to part of the hind brain (46–48). From the above text, we hypothesized that these conditions in SFEBq culture are suitable for otic induction.
Several reports on the differentiation of ESC/iPSC into otic progenitor cells (OPC) have used PAX2, PAX8 and GATA3 as OPC markers (21,22,25,26,34). In this study, in addition to these gene sets, we used GJB2 and GJB6 as indicators for the culture conditions. The GJB6 gene encodes CX30 protein, which is co-expressed with CX26 in the cochlear supporting cells (7,10). We confirmed that these marker genes are upregulated by the addition of insulin in the SFEBq culture. These results suggested that adding insulin to gfCDM in the SFEBq culture would be the most suitable method for inducing the iCX26GJCs.
In several reports, the cochlear cell-like cells that had been induced from ESC/iPSC were characterized using multiple markers (21,22,25,29). By using immunostaining (CX30, SOX2, SPARCL1, KCC3 and some cytokeratins) and qPCR (SOX2, KIAA1199, SPARCL1, MIA and OTOR), we confirmed that these markers were expressed in proliferated human iCX26GJCs in the adherent culture, which was after the SFEBq culture. These protein or gene markers were reported to have been expressed in cochlea supporting cells or fibrocytes by several studies (25,26,49,50). With these characteristics, iCX26GJC is reminiscent of mammalian cochlear supporting cells or fibrocytes.
Patient iPSC-derived iCX26GJC reproduced pathology of GJB2-related deafness
In mammalian cochlea, connexin gap junction mediated intercellular ion transfer maintains the cochlear homeostasis (14). Previous studies using transfected COS-7 cells have suggested that the abnormal subcellular localization of mutated CX26 protein with 235delC leads to loss of function, resulting in serious hearing impairment (20). In addition, it has been reported that mutations in the GJB2 gene decrease GJIC (19). On the other hand, our previous study using GJB2 conditional knock out mice (Cx26f/f P0-Cre) suggested that drastic disruption of CX26/CX30 macromolecular complex and decreased GJIC in the cochlear supporting cells resulted in hearing loss (15). In this study, we confirmed CX26 deficiency and shortening of GJP length in patient iPSC-derived iCX26GJC using immunostaining. Furthermore, in agreement with several reports showing that GJB2 mutations cause a reduction of GJIC (15–20), our results show that patient iPSC derived iCX26GJCs have significantly lower performance of GJ than wild-type cells. From the above text, we speculated that, in human GJB2-deafness, hearing impairment as a result of the decrease in GJIC was associated with disruption of CX26/CX30 macromolecular complex, as in the GJB2 mutant mice (Cx26f/f P0-Cre) study (15).
By applying our method to iPSCs derived from patients that have other GJB2 mutations (for example, V37I or G45E/Y136X, which are, respectively, the second and third most prevalent mutations in GJB2 in individuals in Japan (37,39)), we expect to clarify the relationship between differences in hearing function due to mutations and GJIC. Such iPSC-derived cells should be particularly useful for drug screening and inner-ear cell therapies with genome editing targeting GJB2-related hearing loss. In addition, this system should contribute to the development of optimal treatment methods for each mutation.
Methods
Human iPSC culture
The healthy human iPSC line (201B7) was provided by the RIKEN Bio Resource Center Cell Bank. The two human iPSC lines GP5-235delC/235delC and GP6-235delC/235delC (i.e. GP5-235delC and GP6-235delC, respectively; family history and audiometric phenotypes are shown in Figure 5A–C), which were generated from PBMCs from sibling patients, have been characterized previously (specific cell line names: JUFMDOi005-A and JUFMDOi006, respectively) (40). These three human iPSC lines were maintained on iMatrix-511 (Nippi)-coated plates with StemFit AK02N (AjinomotoWako) under a feeder-free culture system as described (51,52).
Differentiation of human iPSCs
The human iPSCs were dissociated with 0.5× TrypLE select, suspended in maintenance medium (Stem Fit) supplemented with Y-27632 (20 μM), and then plated at 100 μl/well (9000 cells) in 96-well low-cell-attachment V-bottom plates (Thermo Fisher Scientific). After a 2-day incubation at 37°C, 3% CO2, the aggregates were transferred to 96-well low-cell-attachment V-bottom plates (Sumitomo Bakelite) in 100 μl of gfCDM (Supplementary Material, Table S1) containing 2% Matrigel (Corning). On days 7–11, the aggregates were semi-dissected using forceps. Aggregates were transferred into adherent cultures containing TRICs (see below) in DFNB medium (Supplementary Material, Table S1). After a 7-day incubation, the medium was changed to growth medium (DMEM glutaMAX with 10% FBS; Supplementary Material, Table S1).
To prepare TRICs, cochlear tissue was obtained from 10-week-old mice (obtained from CLEA Japan, Inc) that included the organ of Corti, basilar membrane and lateral wall and mainly comprised supporting cells, hair cells, cochlear fibrocytes, and other cells in the basilar membrane. TRICs were generated by exposing the cochlear tissue to trypsin and screening for trypsin-resistant cells, and these cells were maintained under growth medium. This cell line was used as inner ear-derived feeder cells on which to proliferate the OPC. For the feeder cell layer, 3 × 105 TRICs/cm2 were seeded into gelatin-coated wells of 24-well culture plates after mitomycin C (10 mg/ml) treatment for 3 h.
Quantitative reverse transcription-PCR of GJB2 and GJB6 mRNA expression
Day 7 aggregates were collected and washed with DPBS and then total RNA was isolated. Total RNA was isolated using reagents from an RNeasy Plus Mini kit (Qiagen) and reverse transcribed into cDNA using reagents from a PrimeScript II first strand cDNA synthesis kit (Takara). Real-time PCR was performed with the reverse transcription products, TaqMan Fast Advanced Master Mix reagents (Applied Biosystems) and a gene-specific TaqMan Probe (see below; Applied Biosystems) on a StepOne Real-Time PCR system (Applied Biosystems). Each sample was run in triplicate. StepOne software (Applied Biosystems) was used to analyze the Ct values of the different mRNAs, with their expression normalized to the endogenous control, actin beta mRNA. TaqMan Probes (Assay ID; Applied Biosystems) were used to detect the expression of human GJB2 (Hs00269615_s1), GJB6 (Hs00922742_s1), NANOG (Hs02387400_g1), PAX2 (Hs01057416_m1), PAX8 (Hs00247586_m1), GATA3 (Hs00231122_m1), OTOR (Hs00375304_m1), MIA (Hs00197954_m1), SOX2 (Hs01053049_s1), SPARCL1 (Hs00949886_m1), ACTB (Hs99999903_m1) and 18S (Hs99999901_s1) mRNAs.
Immunostaining and image acquisition
Aggregates were fixed with 4% (w/v) paraformaldehyde in 0.01 M PBS for 1 h at room temperature. For whole mounts, the aggregates were permeabilized with 0.5% (w/v) Triton X-100 (Sigma-Aldrich) in 0.01 M PBS for 30 min. Then, the samples were washed twice with 0.01 M PBS and blocked with 2% (w/v) BSA in 0.01 M PBS for 30 min.
Cells from adherent cultures were fixed with 4% (w/v) paraformaldehyde in 0.01 M PBS for 15 min at room temperature and then were permeabilized with 0.5% (w/v) Triton X-100 in 0.01 M PBS for 5 min. Samples were washed twice with 0.01 M PBS and blocked with 2% (w/v) BSA in 0.01 M PBS for 30 min. For immunofluorescence staining, 1% (w/v) BSA in 0.01 M PBS was used to dilute the primary and secondary antibody solutions. Each sample was incubated in a primary antibody solution—CX26 (rabbit IgG, 71–0500; mouse IgG, 33–5800, Life Technologies), CX30 (rabbit IgG, 71–2200, Life Technologies), Pan-Cytokeratin (mouse IgG, C2562, Sigma-Aldrich), Cytokeratin 8 (mouse IgG, MA5–14428, Invitrogen), Cytokeratin 18 (mouse IgG, MA5–12104, Invitrogen), SOX2 (Goat IgG, SC-17320, Santa Cruz), SPARC-like1 (mouse IgG, AF2728, R&D System) and SLC12A6 (KCC3, rabbit IgG, ab92951, abcam)—for 1 h after blocking. The secondary antibodies were Alexa Fluor 488–conjugated anti-mouse IgG or anti-goat IgG, Cy3–conjugated anti-rabbit IgG (Invitrogen, A11070) and phalloidin FITC staining for F-ACTIN (Invitrogen, A12379). Samples were washed twice with 0.01 M PBS and mounted with mounting medium (VECTASHIELD Mounting Medium with DAPI, Vector). Fluorescence confocal images were obtained with an LSM780 confocal microscope (Zeiss). Images were collected at 0.5-μm intervals (z-stacks), and the single-image stacks were constructed using LSM Image Browser (Zeiss). Three-dimensional images were constructed from z-stacked confocal images using IMARIS (Bitplane).
Scrape loading/dye transfer assay
The SL/DT assay was performed as described (29,30). iCX26GJC-containing proliferating cells were grown for 7–14 days after transfer onto TRICs (feeder cells). Undifferentiated iPSCs and TRICs were grown to confluency on dishes as a control. The medium was changed to HBSS plus 0.1% Lucifer yellow CH (L453, Invitrogen). Many parallel lines were cut into the dish with a razor blade, and after 15 min cells scrape-loaded with Lucifer yellow were washed three times with HBSS and imaged. Scrape loading was quantified by measuring the distance from the scrape line to the point where the fluorescence intensity dropped to the background intensity. The images were processed and analyzed with NIH ImageJ software, and the mean distance was calculated using Microsoft Excel software.
Statistics
The data were analyzed using Microsoft Excel software and are presented as the mean ± standard error. A two-tailed Student’s t-test with a significance criterion of P < 0.05 was used to compare the number of CX26+ vesicles per mm3 of aggregate, diameter and number of cells. One-way ANOVA and Scheffe’s multiple comparison test or two-tailed Student’s t-test with a significance criterion of P < 0.05 was used to compare mRNA levels, the distances of dye transfer, and GJP length.
Ethics Statement
All the experimental protocols using human iPSCs were approved by the ethics committee of Juntendo University (2016080). Written informed consent was given by both patients. All experimental protocols using mouse tissues were approved by the Institutional Animal Care and Use Committee at Juntendo University School of Medicine and were conducted in accordance with the US National Institutes of Health Guidelines for the Care and Use of Laboratory Animals. All methods were carried out in accordance with relevant guidelines and regulations.
Author contributions
KK: Conception and design, Manuscript writing, Final approval of manuscript. IF: performed experiments, data analysis and interpretation, manuscript writing. YO, KD, SO, CC, KS, and AK: performed the experiments. KI: contributed reagents/materials/analysis tools. All authors reviewed the manuscript.
Acknowledgments
We thank Wado Akamatsu (Juntendo University) for advice, materials and reading the manuscript. This work was supported by grants from the JSPS KAKENHI (number 17H04348, 16 K15725, 18H02953 and 20K21662 to K. Kamiya, and number 19 K09914, 17 K16948 and 15 K20229 to I.Fukunaga), Subsidies to Private Schools (to K. Kamiya and I.Fukunaga), Japan Agency for Medical Research and Development, (AMED, number 15ek0109125h0001, 19ae0101050h0002 and 19ek0109401h0002 to K.Kamiya), and the Takeda Science Foundation (to K.Kamiya).
Conflict of interest statement. The authors have declared that no conflict of interest exists.
