JP7760144B2 - Method for producing inner ear cells - Google Patents

Description

The present invention relates to a new method for producing inner ear cells that have connexin 26 gap junctions.

Hearing loss is the most common congenital sensory disability, with approximately 1 in 1000 people developing severe hearing loss at birth or during infancy, defined as prelingual deafness, about half of which is genetically determined. There are over 100 known forms of nonsyndromic hearing loss associated with identified genetic loci.
Mutations in the gap junction beta 2 gene (Gjb2), which encodes connexin 26 (CX26), account for 50% of nonsyndromic sensorineural hearing loss. CX26 and CX30, encoded by Gjb2 and Gjb6, respectively, assemble and participate in the formation of intercellular gap junctions. These connexins are two of the most abundant gap junction-forming proteins in the cochlea.
Gap junctions maintain cochlear homeostasis by facilitating the rapid removal of K+ ions from cochlear hair cells and returning them to the endolymph. CX26 and CX30 form heterogeneous and heterotypic channels in many cochlear gap junction plaques (GJPs) and in vitro experiments. Our recent research has shown that disruption of CX GJPs is associated with the development of GJB2-related hearing loss and that cochlear GJP assembly is dependent on CX26 (Non-Patent Document 1). We have also reported that cochlear gene transfer of GJB2 using adeno-associated viruses significantly improved GJP formation and hearing function (Non-Patent Document 2). We have also developed a novel strategy for inner ear cell therapy using bone marrow mesenchymal stem cells (Non-Patent Document 3).
Furthermore, the present inventors have reported that inner ear cells with CX26 gap junctions can be produced by culturing pluripotent stem cells such as mouse iPSCs in the presence of one or more additives selected from BMP, a TGF-β1 receptor inhibitor, and FGF, and then culturing the resulting embryoid bodies in the presence of cochlear-derived feeder cells (Non-Patent Document 4, Patent Document 1).

International Publication No. 2017/146035

J. Clin. Invest. 124, 1598-1607 (2014) Hum. Mol. Genet. 24, 3651-3661 (2015) Am. J. Pathol. 171, 214-226 (2007) Stem Cell Reports (2016) 7(6), 1023-1036

However, the method for producing inner ear cells with CX26 gap junctions from pluripotent stem cells previously reported by the inventors is still insufficient in terms of yield, and it is desirable to provide a more efficient method for producing inner ear cells with CX26 gap junctions.

The inventors therefore investigated additives used when inducing differentiation of pluripotent stem cells into embryoid bodies. As a result, they discovered that by culturing pluripotent stem cells in the presence of insulin instead of the conventionally used growth factors such as BMP, TGF-β1 receptor inhibitor, and FGF to form embryoid bodies, and then culturing these embryoid bodies in the presence of cochlear-derived feeder cells, they were able to obtain inner ear cells with CX26 gap junctions with high efficiency, leading to the completion of the present invention.

That is, the present invention provides the following [1] to [4].

[1] A method for producing inner ear cells having connexin 26 gap junctions, characterized by culturing embryoid bodies obtained by culturing pluripotent stem cells in the presence of insulin, in the presence of cochlear-derived feeder cells.
[2] The method for producing inner ear cells according to [1], wherein the cochlea-derived feeder cells are cells formed by trypsinizing cochlear cells and culturing them to form colonies.
[3] A method for producing inner ear cells according to [1] or [2], wherein the pluripotent stem cells are iPS cells or ES cells.
[4] The method for producing inner ear cells according to any one of [1] to [3], wherein the pluripotent stem cells are human iPS cells or human ES cells.

The inner ear cells obtained by the method of the present invention form CX26 and CX26-containing gap junction plaques, similar to those in the cochlea, and upregulate the otic progenitor cell markers GJB2, GJB6, PAX2, PAX8, and GATA3. They also express the cochlear cell markers CX3, SOX2, SPARCL1, KCC3, KIAA1199, MIA, and OTOR. Furthermore, inner ear cells with connexin26 gap junctions derived from iPSCs of hearing-loss patients were produced, and the pathology of GJB2-related hearing loss was successfully reproduced. Therefore, the inner ear cells obtained by the present invention can be used to screen for drugs for hereditary hearing loss caused by GJB2 mutations and for inner ear regenerative medicine.

Figure 1 shows that insulin treatment induces GJB2/GJB6 genes and otic progenitor marker genes in iPSCs. (A) Procedure for differentiation of CX26-expressing cells from human iPSCs in SFEBq culture. (B) Quantitative PCR analysis of undifferentiated human iPSCs (day 0) and iPSC-derived aggregates (day 7) for 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 aggregates without insulin (Ins(-)). SF: StemFit AK02N; Y: Y-27632; Ins: insulin. Statistical significance was determined by Scheffe's multiple comparison test. Mean ± standard error (SE), *p<0.05, **p<0.01. Insulin treatment induces CX26-expressing cells in iPSCs. (A-D) Immunostaining of CX26 (red) on day 7 aggregates. Arrows indicate CX26+ vesicles. Boxed areas in (A) and (B) are enlarged in (C) and (D), respectively. (E) Number of CX26+ vesicles per mm3 of day 7 aggregates (n = 9 or 15 aggregates from 2 or 3 independent experiments). (F) Average diameter of CX26+ vesicles on day 7 aggregates (n = 8 or 20 CX26+ vesicles from 3 independent experiments). (G) Average number of cells consisting of CX26+ vesicles on day 7 aggregates (n = 12 CX26+ vesicles) from three independent experiments. (H) Percentage of CX26-positive versus CX26-negative cells in day 7 aggregates (n = 9 or 14 from day 7, 2 to 3 independent experiments). White columns represent CX26-negative (CX26-) cells. Black columns represent CX26-positive (CX26+) cells. (I-N) Immunostaining for CX26 (red) and F-ACTIN (green) on day 7 CX26+ vesicles aggregates. Nuclear DAPI counterstaining (blue in L-N). Boxed areas in (I), (J), (L), and (M) are enlarged in (J), and (K) and (M) show (N) and (O,P) GJPs, respectively. 3D images were reconstructed from the image in (M). Arrows indicate GJPs. Scale bars: 200 μm (A and B); 50 μm (C and D); 20 μm (I and L); 10 μm (J, M, P); 5 μm (K, N, O). Statistical significance determined by Student's t test, mean ± SE; *p < 0.05; ****p < 0.01. Figure 1 shows the growth of iCX26GJC on adherent culture. (A) Procedure for the growth of iCX26GJC from human iPSC aggregates using adherent culture. (B) Phase-contrast microscopy (PCM) image from a 21-day adherent culture. (C) Enlarged view of the boxed area in (B) staining for CX26 (red) and PCM (white). (D-G) Staining for CX26 (red) and F-ACTIN (green). Enlarged views of the boxed areas in (D), (E), and (F). (E), (F), (G), and (G), respectively. (H and I) Staining for CX26 (red) and DAPI (blue). (H) is the same area in (F) and (G). The boxed area in (H) is enlarged in (I). (J, K) A 3D image showing GJP was reconstructed from the image in (I). Arrowheads indicate the GJP. Scale bars: (B); 100 μm, (D); 50 μm, (C and E); 10 μm, (F and H); 5 μm (G and I-K). Immunostaining and gene expression of known cochlear markers in adherently grown iCX26GJCs are shown. (A-G) Nuclei of CX26 (red), 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) were stained with DAPI (blue). (H) Quantitative PCR analysis of undifferentiated human iPSCs (hiPSCs) 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 iPSCs (hiPSCs). Scale bars (A-G) 20 μm (A: columns 2, 3, and 4) 10 μm. Statistical significance was determined by Student's t-test, mean ± SE; *p < 0.05; **p < 0.01. CX30 GJP formation and GJP length in normal or patient iPSC-derived iCX26 GJCs. (A) Family history of the individuals from whom these iCX26 GJCs were derived. Squares indicate male family members; circles indicate female family members; filled shapes indicate diagnosed family members. (B and C) Audiometric phenotypes of patient 1 (B) and patient 2 (C) indicate severe hearing loss. (D) Formation of iCX26 GJCs by normal (201B7) or patient (GP05-235delC, GP06-235delC) iPSCs. These samples were co-labeled with anti-CX26 (red) and anti-CX30 (green) antibodies. (E) Maximum GJP length along a single cell boundary (mean ± SE, n = 27, 30, and 37 cell boundaries from 34 independent experiments). Statistical significance was determined by Scheffe's multiple comparison test. Mean ± standard error, *p < 0.05, **p < 0.01. Dye transfer after cell scrape-loading and subsequent quantitative analysis is shown. (A-O) Digital fluorescence images of cultured cells after scrape-loading. iPSC-derived adherent cultures including healthy human iPSCs (201B7) (A-C), feeder cells (TRICs) (D-F), iCX26GJCs derived from healthy human iPSCs (201B7-iCX26GJCs) (G-I), and iCX26GJCs derived from iPSCs from patient 1 (GP5-235delC-iCX26GJCs) (J-L) and patient 2 (GP6-235delC-iCX26GJCs) (M-O) as healthy controls. (A, D, G, J, M) Dye transfer using Lucifer Yellow (B, E, H, K, N). Pseudocolor images showing the transfer range from low (black) to high (red) signal intensity in the same area as above. (C, F, I, L, O) Phase-contrast microscopy images of the same area. (P) Quantitative analysis of intercellular dye transfer after scraping. Columns represent mean values. Distance of dye migration from the scrape line (TRICs and random 201B7: n = 40 from five independent experiments; 201B7-iCX26JCs, GP5-235delC-iCX26JCs, GP6-235delC-iCX26JCs: n = 40 from two or three independent experiments). Statistical significance was determined by Scheffe's multiple comparison test. Mean ± SE; different letters (a to c) indicate significant differences, p<0.01. Scale bar represents 50 m. 1 shows the procedure for differentiation of human iPSC culture into CX26-expressing cells by BMP and/or SB treatment. On day 7, BMP4 (10 ng/ml) and/or SB (10 μM) were added to gfCDM. Figure 1 shows that BMP and/or SB treatment did not promote the induction of CX26-expressing cells in SFEBq/gfCDM cultures. (B and C) qPCR analysis of the expression of GJB2 (B) and GJB6 (C) (n = 3 from three independent experiments) at day 7. (D→I) Immunostaining of CX26 (red: D→F) and F-ACTIN (green: G→I) aggregates at day 7. Arrows indicate CX26-positive vesicles. (J) Average CX26-positive vesicle aggregates at day 7 (n = 8-16 aggregates from three independent experiments). BMP, BMP4, SB, SB431542. Each bar represents the mean ± SE. Differences between samples were assessed by one-way ANOVA and Scheffe's multiple comparison test; *p < 0.05; **p < 0.01. The morphology of aggregates on day 7 of culture is shown. (A) The morphology of aggregates on day 0 or day 7. Differentiation of human iPSC aggregates in gfCDM without insulin (-) produced more cell debris than gfCDM with insulin (+). (B) The diameter of aggregates on day 7 with or without insulin (+). (n=30 aggregates from three independent experiments) Statistical significance was determined by Student's t-test, mean ± SE *p<0.05; **p<0.01. Scale bar: 200 μm (A). 1 shows the procedure for differentiation of CX26-expressing cells from human iPSC culture. On day 7, BMP4 (10 ng/ml) and/or SB (10 μM) were added to gfCDM. Figure 1 shows that treatment with BMP and/or SB does not enhance insulin-induced induction of CX26-expressing cells in SFEBq/gfCDM cultures. (B and C) qPCR analysis of the expression of GJB2 (B) and GJB6 (C) (n=4 from four independent experiments) at day 7. (D→I) Immunostaining of CX26 (red: D→F) and F-ACTIN (green: G→I) aggregates at day 7. Arrows indicate CX26-positive vesicles. (J) Average CX26-positive vesicle aggregates at day 7 (n=8-16 aggregates from three independent experiments). Insulin (Ins), BMP, BMP4, SB, SB431542. Each bar represents the mean ± SE. Significant differences between samples were assessed by one-way ANOVA and Scheffe's multiple comparison test; *p<0.05; **p<0.01. Other cytokeratins expressed in the skin are not observed in human iCX26GJC. Immunostaining of CX26 (red) and CK5, 10, and 14 (green) in adherent cultures with nuclear counterstaining by DAPI (blue) is shown. Scale bar: 10 μm.

The present invention provides a method for producing inner ear cells with CX26 gap junctions, which comprises culturing embryoid bodies obtained by culturing pluripotent stem cells in the presence of insulin, and then culturing the resulting embryoid bodies in the presence of cochlear-derived feeder cells.

Pluripotent stem cells used in the present invention include iPS cells and ES cells. Furthermore, while pluripotent stem cells derived from non-human animals such as mice can also be used, it is preferable to use iPS cells and human ES cells derived from humans. In this specification, iPS cells may be abbreviated as iPSCs. ES cells may be abbreviated as ESCs.

In the present invention, embryoid bodies can be formed from pluripotent stem cells by culturing the pluripotent stem cells in the presence of insulin. There is no need to use BMP, TGF-β1 receptor inhibitors, FGF, and the like used in the methods described in Patent Document 1 and Non-Patent Document 4.
The concentration of insulin added to the medium is preferably 1 to 20 μg/mL, more preferably 1 to 15 μg/mL.

To form embryoid bodies derived from pluripotent stem cells, it is preferable to add insulin to the pluripotent stem cells in a known medium for culturing pluripotent stem cells, such as gfCDM medium, DFNB medium, N2β27 medium, or GEM medium, and culture them at 30-40°C for 7-10 days. Embryoid body formation can be confirmed by the formation of cell aggregates and differentiation into sensory epithelial cells. Furthermore, differentiation into sensory epithelial cells can be confirmed by an increase in the mRNA levels of CX26 and CX30.

If the resulting pluripotent stem cell-derived embryoid bodies are cultured on cochlear-derived feeder cells, inner ear cells having CX26 gap junctions can be obtained.
The cochlear inner ear feeder cells are preferably cells formed by trypsinizing cochlear cells and culturing them to form colonies, for example. Preferably, 0.25% trypsin is added to cochlear labyrinth tissue including the lateral cochlear wall and the organ of Corti, followed by culturing, to obtain cells that have formed colonies.
The seeding density of cochlea-derived feeder cells used to culture the disaggregated embryoid bodies is preferably 5 x ¹⁰ cells/ ^cm to 5 x ¹⁰ cells/ ^cm . The culture medium used may be gfCDM medium, DFNB medium, or DMEM GlutaMAX + 10% fetal bovine serum. Culture is preferably carried out at 30 to 40°C for 3 to 4 days.

By the above-described culture, inner ear cells having CX26 gap junctions can be efficiently obtained. Here, CX26 expression can be confirmed by an increase in the concentration of Cx26 mRNA.
The inner ear cells obtained by the present invention form CX26 and CX26-containing gap junction plaques, similar to those in the cochlea, and upregulate the otic progenitor cell markers GJB2, GJB6, PAX2, PAX8, and GATA3. They also express the cochlear cell markers CX3, SOX2, SPARCL1, KCC3, KIAA1199, MIA, and OTOR. Furthermore, inner ear cells possessing connexin 26 gap junctions derived from iPSCs of hearing-impaired patients were produced, and the pathology of GJB2-related hearing loss was successfully reproduced. Therefore, the inner ear cells obtained by the present invention can be used for screening drugs for hereditary hearing loss caused by GJB2 mutations and for inner ear regenerative medicine.

The present invention will now be described in more detail using examples.

Example 1
(method)
(1) Culture of human iPSCs. A healthy human iPSC line (201B7) was provided by the RIKEN BioResource Center Cell Bank.
Two phenotypic iPSC lines, GPSC-235delC/235delC and GP6-235delC/235delC (i.e., GP5-235delC and GP6-235delC; family history and audiograms are shown in Fig. 5, A–C ), were generated from PBMCs from sibling patients, and two phenotypic iPSC lines (JUFMDOi005-A and JUFMDOi006, specific cell line names, respectively) were known.
These three human iPSC lines were maintained in a feeder-free culture system on iMatrix-511 (Nippi)-coated plates with StemFit AK02N (Ajinomoto Wako).

(2) Differentiation of human iPSCs. Human iPSCs were dissociated with 0.5% TrypLE Select and suspended in maintenance medium (Stem Fit) supplemented with Y-27632 (20 μM). The cells were then plated onto a 96-well low-cell-attachment V-bottom plate (Thermo Fisher Scientific) at 100 μL/well (9,000 cells).
After 2 days of incubation at 37°C and 3% CO ₂ , the aggregates were transferred to a 96-well low cell attachment V-bottom plate (Sumitomo Bakelite) in 100 μL of gfCDM (Table 1 below) containing 2% Matrigel (Corning).
On days 7-11, the aggregates were hemi-dissected using forceps.
The aggregates were transferred to adherent cultures containing cochlear-derived feeder cells (TRIC, see below) in DFNB medium (Table 1 below). After 7 days of incubation, the medium was changed to growth medium (DMEM GlutaMAX supplemented with 10% FBS; Table 1 below).

(3) Preparation of cochlear-derived feeder cells (trypsin-resistant inner ear cells: TRIC) To prepare TRIC, cochlear tissue, including the organ of Corti, basilar membrane, and lateral wall, and consisting mainly of supporting cells of the basilar membrane, hair cells, cochlear fibrocytes, and other cells, was obtained from 10-week-old mice (CLEA Japan, Inc.).
TRICs were generated by exposing cochlear tissue to trypsin and screening for trypsin-resistant cells, which were maintained under growth medium.
This cell line was used as an inner ear-derived feeder cell for expanding otic progenitor cells. For the feeder cell layer, 3 × 10 ⁵ TRICs/cm ² were seeded onto gelatin-coated wells of a 24-well culture plate after 3 hours of mitomycin C (10 mg/mL) treatment.

(4) Quantitative reverse transcription PCR of GJB2 and GJB6 mRNA expression
The 7-day-old aggregates were harvested and washed with DPBS, after which total RNA was isolated using reagents from the Neasy Plus Mini Kit (Qiagen) and reverse transcribed into cDNA using reagents from the PrimeScript II First-Strand cDNA Synthesis Kit (Takara).
Real-time PCR was performed using the reverse transcription products, TaqMan Fast Advanced Master Mix reagent (Applied Biosystems), and gene-specific TaqMan probes (see below; Applied Biosystems) on a StepOne Real-Time PCR system (Applied Biosystems). Each sample was run in triplicate. The Ct values of each mRNA were analyzed using StepOne software (Applied Biosystems), and their expression was normalized to the endogenous control, actin β mRNA.
TaqMan probes (assay ID; Applied Biosystems) were used to detect 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), and ACTB. (Hs99999903_M1), and 18S (Hs99999901_S1) mRNA expression was detected.

(5) Immunostaining and Imaging. Aggregates were fixed with 4% (w/v) paraformaldehyde in 0.01 M PBS at room temperature for 1 hour. For whole mount imaging, aggregates were permeabilized with 0.5% (w/v) Triton X-100 (Sigma-Aldrich) in 0.01 M PBS for 30 minutes. The samples were then washed twice with 0.01 M PBS and blocked with 2% (w/v) BSA in 0.01 M PBS for 30 minutes. Cells from adherent cultures were fixed with 4% (w/v) paraformaldehyde in 0.01 M PBS for 15 minutes at room temperature and then permeabilized with 0.5% (w/v) Triton X-100 in 0.01 M PBS for 5 minutes. Samples were washed twice with 0.01 M PBS and blocked with 2% (w/v) BSA for 30 min in 0.01 M PBS.
For immunofluorescence staining, primary and secondary antibody solutions were diluted with 1% (w/v) BSA in 0.01 M PBS. The primary antibody solutions used were CX26 (rabbit IgG, 71-500; 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 Systems), and SLC12A6 (rabbit IgG, KCC3). 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, 12379).
The specimens were washed twice with 0.01 M PBS and mounted with a mounting medium (Vector DAPI-added VECTASHIELD Mounting Medium).
Fluorescence confocal images were obtained with an LSM780 confocal microscope (Zeiss). Images were collected at 0.5 μm intervals (z-stacks), and single image stacks were constructed using the LSM Image Browser (Zeiss). Three-dimensional images were constructed from z-stacked confocal images using IMARIS (Bitplane).

(6) Scrape-loading/dye transfer (SL/DT) assay. The SL/DT assay was performed according to Mol. Med., 17, 550-556 (2011) and J. Hum. Genet., 50, 76-83 (2005). iCX26GJC-containing expanded cells were grown for 7 to 14 days after transfer onto TRIC (feeder cells). Undifferentiated iPSCs and TRIC were grown to confluence on dishes as controls.
The medium was changed to HBSS + 0.1% Lucifer Yellow CH (L453, Invitrogen). Many parallel lines were cut into the dish with a razor blade, and 15 minutes after loading with Lucifer Yellow, the cells 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 background intensity. Images were processed and analyzed with NIH ImageJ software, and the average distance was calculated using MicroWeb Excel software.

(7) Statistics The data were analyzed using MicroWeb Excel software and presented as mean ± standard error (SE). Comparisons of mRNA levels, dye migration distance, and GJP length were performed using one-way analysis of variance and Scheffe's multiple comparison test or two-tailed Student's t-test, with p<0.05 as the significance criterion.

(result)
(1) Differentiation of human iPSCs into CX26-expressing cells. The method for inducing iCX26GJCs from human iPSCs, which combines SFEBq culture and adherent culture techniques, was modified from our previous method for mouse iPSCs (Patent Document 1), and the conditions required for differentiation were then evaluated.
For SFEBq culture, human iPSCs were reaggregated (9,000 cells/well) and cultured in maintenance medium (StemFit AK02N) for 2 days (day #2 to day 0). The aggregates were then transferred to a chemically defined medium (gfCDM; Table 1) lacking growth factors.
On day 3 of SFEBq culture, BMP4 (BMP) and/or an activin/Nodal/TGF-β pathway inhibitor (SB431542: SB) was added to gfCDM according to the mouse iCX26GJC induction method (Figure 7).
However, the addition of these supplements did not increase GJB2/GJB6 mRNA expression or CX26-positive cell mass (CX26+ vesicles) production.
Conversely, addition of SB induced a decrease in GJB2 mRNA expression and CX26+ vesicle production (Fig. 8).
Based on these results, in the following experiments, insulin (7 μg/mL) was added to gfCDM from day 0.
The modified SFEBq culture for differentiation of human iPSCs into CX26-expressing cells is shown in Figures 1A (SFEBq culture) and 3A (adherent culture).

(2) Screening for high CX26 expression in SFEBq cultures In the improved SFEBq cultures, aggregates were collected on day 7, and mRNA (GJB2, GJB6, PAX2, PAX8, GATA3) in each culture group was measured.
The GJB6 gene encodes the CX30 protein, which is co-expressed with CX26 in cochlear supporting cells (J. Comp. Neurol., 467, 207-231 (2003), J. Comp. Neurol., 499, 506-518 (2006)).
The PAX2, PAX8, and GATA3 gene set has been used as an otic progenitor cell marker in several studies aimed at differentiating ESC/iPSC into inner ear cells (Cell 141, 704-716 (2010), Nature, 490, 278-282 (2012), etc.).
In the day 7 aggregates, the GJB2, GJB6, PAX2, PAX8, and GATA3 genes were upregulated compared to those in undifferentiated iPSCs (day 0). Furthermore, the expression of PAX2, PAX8, and GATA3 genes was significantly increased compared to undifferentiated iPSCs (Figure 1B).
iPSCs cultured in insulin-supplemented gfCDM 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) compared with those cultured in insulin-free medium (Figure 1B). Cell debris was observed around the aggregates in day 7 aggregates without insulin. In contrast, no debris was observed when insulin was added to the aggregates (Figure 9A).
Furthermore, the diameter of the aggregates on day 7 was significantly larger with insulin (mean = 876.6 ± 6.90 μm) than without insulin (mean = 644.8 ± 24.58 μm) (Figure 9B).

To analyze the localization of CX26 in iPSC aggregates, immunohistochemistry was performed on day 7 aggregates. CX26-expressing cell masses (CX26 + vesicles) were observed in day 7 aggregates, regardless of the presence or absence of insulin (Fig. 2A-D).
Cells treated with insulin had significantly more CX26+ vesicles (mean = 2.7 ± 0.21) compared with cells cultured without insulin (mean = 1.7 ± 0.37) (Fig. 2E).
On the other hand, addition of SB and/or BMP to insulin-supplemented gfCDM did not increase mRNA expression or CX26+ vesicle production (Figs. 10 and 11).
The diameter of CX26+ vesicles was significantly larger in cells treated with insulin (mean = 161.3 ± 10.83 μm) than in cells without insulin (mean = 129.2 ± 4.70 μm) (Figure 2F).
Furthermore, 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 comprised the aggregates were 4.18% and 95.8%, respectively, for insulin(-) aggregates, and 8.86% and 9.13%, respectively, for insulin(+) aggregates (Fig. 2H).
Confocal analysis of the 7-day insulin-treated aggregates revealed that CX26-expressing cells were seeded throughout the CX26+ vesicles (Fig. 2, I–N). These cells formed CX26+ GJs at cell-cell boundaries (Fig. 2, J, K, M, and N). Three-dimensional reconstruction of the confocal images revealed planar CX26-containing GJPs (Fig. 2, O and P).

iCX26GJCs expressed typical cochlear cell markers. Regions containing iCX26GJC-containing CX26+ vesicles were isolated from the aggregates on days 7–11 and transferred to mouse cochlear feeder cells (trypsin-resistant inner ear cells, TRIC) in DMEM/Ham's F12 (DFNB medium) supplemented with N2/B27 (Figure 3A). The transferred iCX26GJC-containing regions actually formed colonies on TRIC feeder cells, and the colonies contained iCX26GJCs (Figure 3, B and C). These proliferating cells formed CX26-positive GJs at the cell-cell boundaries (Figure 3, D–H). Three-dimensional reconstructions of confocal images revealed large, planar CX26-containing GJPs (Figure 3, I–K).

To determine whether iCX26GJC resembles cochlear supporting cells, we examined the expression of the following proteins (CX30, SOX2, SPARCL1, KCC3, and several cytokeratins) and genes (SOX2, SPARCL1, KIAA1199, MIA, and OTOR) observed in cochlear supporting cells and/or fibrocytes.
The results of immunostaining and qPCR for iCX26GJC are summarized in Tables 2 and 3.

In human iCX26GJC, CX26 and CX30 expression was observed in the same cells, but these proteins did not necessarily co-assemble in gap junction plaques (Figure 4A). In human cochlear spiral ligament fibrocytes, CX26/CX30 proteins form separate GJPs and do not necessarily co-assemble in GJPs (Cell Tissue Res., 365, 13-27 (2016)). However, it is unclear whether these proteins form gap junction plaques in cochlear supporting cells as well as in cochlear fibrocytes. On the other hand, in rodents, the expression patterns of CX26/CX30 proteins differ between cochlear supporting cells (which form hexagons) and lateral wall fibroblasts (which do not form polygons) (Cell Tissue Res., 333, 395-403 (2008), J. Clin. Invest., 124, 1598-1607 (2014)).
From the above, we cannot conclude that in our target, human cochlear supporting cells, CX26/CX30 proteins do not coassemble GJPs as in fibrocytes of the spiral ligament.

Furthermore, iCX26GJC coexpressed SOX2, SPARCL1, KCC3, p-CK, CK8, and CK18 proteins ( Fig. 4, B–G ).
In contrast, immunolabeling of the skin markers CK5, CK10, and CK14 was not detected in iCX26GJCs (Figure 12). Furthermore, the GJB2, GJB6, KIAA1199, SPARCL1, MIA, and OTOR genes were significantly upregulated compared to undifferentiated iPSCs (Figure 4H). On the other hand, the expression level of the undifferentiated marker NANOG was significantly decreased compared to undifferentiated iPSCs. The expression level of SOX2, another undifferentiated marker, was also decreased, but not as dramatically 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) (Proc. Natl. Acad. Sci. USA, 105, 18396-18401 (2008)) and cells that do not express SOX2 (inner sulcus cells, outer sulcus cells, and fibrocytes).
Furthermore, in the mouse iPSC-derived iCX26GJCs reported by the present inventors in Patent Document 1, the presence or absence of SOX2 expression was observed, as in the case of cochlear supporting cells.
Based on the above, it was considered that no significant decrease in mRNA expression, such as that of NANOG, was observed in human iPSC-derived iCX26GJC.

(3) Formation of CX26/CX30 gap junction plaques in iCX26GJC derived from healthy and diseased iPSCs. Disease-specific iCX26GJCs were generated from two iPSC lines generated from sibling patients with a homozygous 235delC mutation in GJB2 (Figure 5A).
The 235delC mutation is the most common GJB2 mutation in Asia, including Japan, and results in a profoundly deaf audiogram.
Two human iPSC lines, GP5-235delC/235delC and GP6-235delC/235delC (hereafter referred to as GP5-235delC and GP6-235delC, respectively; audiograms of patients with profound hearing loss are shown in Fig. 5, B and C), have been previously characterized (specialized cell line names: JUFMDOi005-A and JUFMDOi006, respectively).
To examine whether disruption of the CX26/CX30 macromolecular complex (15, 29), a pathological condition of GJB2 hearing loss observed in mouse models, was also observed in patient iPSC-derived iCX26GJCs, we examined CX26 and CX30 by immunohistochemistry.
iCX26GJCs derived from healthy iPSCs (201B7) displayed large, planar CX30 GJPs at the cell boundaries (Fig. 5D, left column).In contrast, iCX26JCs derived from patient iPSCs (GP5-235delC, GP6-235delC) partially shortened GJPs (Fig. 5D, middle, right column).
As shown in Figure 5E , the GJP length in diseased iPSC-derived iCX26GJCs (GP5-235delC-iCX26GJC: 6.13 ± 0.35 μm; GP6-235delC-iCX26GJC: 5.91 ± 0.37 μm) was significantly shorter than that in normal iPSC-derived iCX26GJCs (8.62 ± 0.31 μm).

(4) Functional evaluation of gap junction intercellular communication (GJIC) in iCX26GJCs derived from healthy and diseased iPSCs. To examine whether there were differences in the function of the GJIC network between healthy and diseased iCX26GJCs, a scrape-loading/dye permeability test (SL/DT) using Lucifer Yellow (Stem Cell Reports, 7, 1023-1036 (2016), Am. J. Physiol. Cell Physiol., 293, C1032-1048 (2007)) was performed.
SL/DT was performed on iCX26GJC, GP5-235delC-iCX26GJC, and GP6-235delC-iCX26GJC derived from iPSCs from a healthy donor (201B7-iCX26GJC). For comparison, undifferentiated iPSCs (201B7) and feeder cells (TRIC) were included.
The extent of dye transfer was quantified by measuring the distance from the scrape line to the point where the fluorescence intensity dropped to background fluorescence intensity.
In these iCX26GJC cultures, we observed Lucifer Yellow diffusing beyond the injured parental cells (Figure 6, G, H, J, K, M, and N), indicating the presence of GJIC. In contrast, this degree of dye permeability was not observed in undifferentiated iPSCs or TRIC feeder cells (Figure 6, A, B, D, and E). As shown in Figure 6P, the quantitative distances of dye permeability 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) were significantly longer than those in undifferentiated iPSCs (22.4 ± 0.65 μm) or TRIC feeder cells (23.8 ± 0.99 μm). Furthermore, the dye penetration distance of both patient-derived iCX26JC (GP5-235delC-iCX26JC and GP06-235delC-iCX26JC) was significantly shorter than that of iCX26JC generated from iPSCs without the GJB2 mutation (201B7-iCX26JC).

(Consideration)
(1) Efficient method for differentiating human iPS cells into iCX26GJCs In this study, CX26 gap junction-forming cells (iCX26GJCs) with the properties of cochlear supporting cells were generated from human iPSCs using SFEBq culture and adherent culture as described in Patent Document 1. Furthermore, this induction method was applied to patient-derived iPSCs to reproduce the pathology of GJB2-related hearing loss.
The SFEBq culture system is the most suitable method for inducing various ectoderm-derived tissues, such as the forebrain, midbrain, hindbrain, optic cup, and ear cup from ESCs/iPSCs. The use of gfCDM and the addition of insulin in SFEBq culture promotes differentiation into the midbrain and hindbrain regions and increases the number of these regions contained in embryoid bodies.
Our target, the inner ear, originates from the otic placode, a superficial part of the non-neural ectoderm (NNE) adjacent to part of the hindbrain.
Based on the above, we hypothesized that these conditions for SFEBq culture are suitable for ear induction.
Several reports on the differentiation of ESCs/iPSCs into otic progenitor cells (OPCs) have used PAX2, PAX8, and GATA3 as OPC markers. In addition to these gene sets, this study also used GJB2 and GJB6 as indicators of culture conditions. The GJB6 gene encodes the CX30 protein and is co-expressed with CX26 in cochlear supporting cells. These marker genes were confirmed to be upregulated by the addition of insulin in SFEBq cultures.
These results suggest that adding insulin to gfCDM in SFEBq culture is the most suitable method for inducing iCX26GJCs. Several reports have characterized cochlear-like cells derived from ESCs/iPSCs using multiple markers. 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 human iCX26GJCs expanded in adherent culture after SFEBq culture. Several studies have reported that these protein or gene markers are expressed in cochlear supporting cells or fibrocytes. Based on these characteristics, iCX26GJCs are considered to be mammalian cochlear supporting cells or fibrocytes.

(2) Patient iPSC-derived iCX26GJC reproduced the pathology of GJB2-related hearing loss.
In the mammalian cochlea, intercellular ion movement via connexin gap junctions maintains cochlear homeostasis. Previous studies using transfected COS-7 cells suggest that abnormal subcellular localization of the 235delC-altered CX26 protein leads to loss of function and severe hearing impairment.

Claims (4)

A method for producing inner ear cells having connexin 26 gap junctions, comprising culturing pluripotent stem cells in a medium supplemented with 1 to 20 μg/mL of insulin, and culturing the resulting embryoid bodies in the presence of cochlear-derived feeder cells. The method for producing inner ear cells according to claim 1, wherein the cochlear-derived feeder cells are cells that form colonies by culturing cochlear cells after trypsinization. The method for producing inner ear cells according to claim 1 or 2, wherein the pluripotent stem cells are iPS cells or ES cells. The method for producing inner ear cells according to any one of claims 1 to 3, wherein the pluripotent stem cells are human iPS cells or human ES cells.