The stable human induced pluripotent stem (iPS) cell line carrying doxycycline-inducible SIX1, ATOH1, POU4F3, and GFI1 promotes robust induction of hair cell transcription factors.

(A) Schematic representation of the strategy used to generate the stable human iPS cell line capable of inducible expression of hair cell transcription factors (TFs). A Tet-On vector containing polycistronic four-TF (SIX1, ATOH1, POU4F3, and GFI1) reprogramming cassette (highlighted by red dashed box) under the control of a TRE3G inducible promoter is shown at the top. The four TFs were separated by 2A self-cleaving peptides to generate multiple proteins from a single primary transcript. Epitope tags were incorporated at the 3’ end as follows: the HA tag with SIX1, the 6xHis tag with POU4F3, and the FLAG tag with ATOH1. The vector was integrated into the CLYBL safe harbor locus between the two homology arms (HA1 and HA2) using CRISPR/Cas9-mediated knock-in. Upon confirmation of correct targeting, the cell line was subsequently treated with TAT-CRE recombinase to excise the mApple screenable marker, FACS-sorted as mApple-negative cells, followed by clonal expansion. (B) Schematic of direct reprogramming of human iPSCs into hair cell-like cells (induced hair cells) using the doxycycline-inducible system. The POU4F3-tdTomato hair cell reporter becomes visible after ∼3 days of reprogramming. Representative image shown was acquired at 7 days post-doxycycline treatment. Scale bar represents 200 µm. (C-D) RT-qPCR analysis of the cell line treated with continuous doxycycline (C) or a 3-day pulse of doxycycline (D) over the indicated time points. Dashed lines indicate post-doxycycline removal. Values are normalized to GAPDH, and a ratio is calculated by dividing the uninduced control (0 hour). Fold-change values are log2-transformed. Error bars indicate SEM. n = 3 biological replicates.

The stable cell line enables highly efficient reprogramming of human iPSCs into hair cell-like cells via inducible expression of SIX1, ATOH1, POU4F3, and GFI1.

(A) Representative images of uninduced control and after 3 or 7 days of reprogramming with continuous doxycycline treatment. Immunostaining for hair cell markers POU4F3 and MYO7A coincides with the loss of pluripotency marker NANOG, indicating the conversion of iPSCs into hair cell-like cells. Scale bar represents 50 µm. (B) Quantification of human induced hair cells after 7 days of reprogramming. Cells were plated in a 96-well format and treated either with continuous doxycycline for 7 days (+Dox) or left untreated (-Dox) and doubled-labeled with DAPI nuclear stain and antibodies against the HA epitope tag, POU4F3, MYO7A, or MYO6. The HA tag, attached to the 3’ end of the SIX1 transgene in the reprogramming cassette (see Fig. 1A), serves as a representative marker to track the expression of reprogramming factors. The presence of each marker was quantified from n = 4 wells, represented as total number of cells per well (left y-axis) and as a percentage out of the total number of DAPI+ cells per well (right y-axis). Data are presented as mean ± SEM. (C) Comparison of conversion efficiency between the doxycycline-inducible system (7 days of reprogramming) and retroviral-mediated transduction (14 days of reprogramming, the standard duration used in our previous study (Menendez et al., 2020)). Compared to retroviral reprogramming, the inducible system exhibits a 19.1-fold increase in conversion efficiency to MYO7A+ cells per well of a 96-well plate. Error bars indicate SEM. (D-E) Western blot analysis of uninduced control (UI) and cells treated with continuous doxycycline for 7, 14, or 21 days. Protein abundance was quantified by normalizing the signal intensity of each target protein to β-ACTIN and graphed on a scale of 0 to 1 along the y-axis. (F-G) RT-qPCR analysis of cells treated with continuous doxycycline (F) or a 3-day pulse of doxycycline (G) over the indicated time points. Dashed lines indicate post-doxycycline removal. Values are normalized to GAPDH, and a ratio is calculated by dividing the uninduced control (0 hour). Fold-change values are log2-transformed. Error bars indicate SEM. n = 3 biological replicates.

Single-nucleus transcriptional profiling of the stable cell line demonstrates hair cell gene network activation following inducible expression of SIX1, ATOH1, POU4F3, and GFI1.

(A) Uninduced control and cells treated continuously with doxycycline for 21 days were collected and separately profiled using single-nucleus RNA sequencing (snRNA-seq). Integration of these datasets and subsequent UMAP projection after unbiased Seurat clustering revealed a clear separation between the two conditions (Uninduced and Day 21). (B) Clustering analysis identified three distinct clusters for uninduced control (C1–3), while five distinct clusters were observed for reprogrammed cells (R1–5). RNA velocity (trajectory) of reprogrammed cells (clusters R1–5) was visualized using Velocyto.R. (C) Among the reprogrammed clusters, higher activation levels of SAPG (both inducible SAPG and endogenous SAPG) were observed in cluster R5 compared to the other four reprogrammed clusters (R1–4). (D-E) Expression of markers associated with pluripotency (POU5F1, NANOG, and GDF3), or the iPSC state (DNMT3B) (Chan et al., 2009), and genes involved in cell cycle progression, was predominantly found in cluster C1. Conversely, cochlear marker genes (NR2F1/2 and LRP2) were activated at higher levels in cluster R2, while hair cell markers were expressed at higher levels in cluster R5.

SIX1, ATOH1, POU4F3, and GFI1 effectively reprogram human fibroblasts to a hair cell-like state via retroviral transduction, exhibiting activation of hair cell gene networks comparable to that observed with the inducible system.

(A) Human iPSC-derived fibroblasts were reprogrammed with a cocktail of retroviruses expressing SIX1, ATOH1, POU4F3, or GFI1 as described in (Menendez et al., 2020) for 21 days. A sorted population consisting of 50% POU4F3-tdTomato+ cells and 50% POU4F3-tdTomato-cells was profiled using single-cell RNA-seq and plotted using UMAP projection. RNA velocity of hair cell-like clusters (labeled Retrovirus-Reprogrammed clusters 1 through 3 [RV-R1–3]) was visualized using Velocyto.R, indicating a transition of reprogramming cells from cluster 1 to cluster 3. (B) Heat map of the top 30 differentially expressed (DE) genes present in each of the hair-cell like clusters (RV-R1–3). (C) Gene set enrichment analysis (GSEA) was performed on the top 200 DE gene set identified in RV-R3 against cluster R5 (Fig. 3) generated by the doxycycline-inducible system. 86.7% of the gene set identified in SAPG-transduced hair cell-like cells was also activated in hair cell-like cells reprogrammed from the stable cell line. The ranked list metric is sorted by log2(fold change)*-log10(adj p-value) (shown in thousands).

The transcriptome of human hair cell-like cells closely resembles that of fetal human hair cells and displays mixed expression patterns of different hair cell subtypes.

(A) Single-nucleus RNA-seq of human induced HCs (cluster R5 in Fig. 3) and uninduced control iPSCs (cluster C1) were integrated with the human fetal inner ear dataset published in van der Valk et al. (2023) (van der Valk et al., 2023) and plotted using UMAP projection. (B) UMAP plot of integrated datasets with cell type annotations. Induced hair cells (iHCs) overlap with developing fetal hair cells. (C) Spearman’s rank correlation matrix for cell populations present in (B). iHC cluster shows the highest correlation with fetal HC cluster. (D) Gene Ontology (GO) analysis of the top 200 DE genes in human iHCs. GO Biological Process terms are shown, ranked by p-value. (E) Human iHCs exhibit co-expression patterns of both outer hair cell (OHC) and inner hair cell (IHC) markers. scRNA-seq profiles of cluster RV-R3 highlight this co-expression phenomenon, where cells show the simultaneous expression (colored in pink) of blue OHC subtype-specifying gene INSM1 and red OHC markers (DNM3, CHRNA10, KCNQ4, LBH, SRI) or IHC markers (SLC7A14, OTOF, SLC17A8/VGLUT3, IGFB5, CTBP2). Only cells demonstrating moderate-to-high expression levels of each gene are colored.

Human induced hair cells exhibit heterogenous voltage-dependent ion currents.

(A) The voltage clamp protocol used to measure the inward and outward currents for uninduced and induced (7-day reprogrammed) cells involved voltage steps ranging from -120 to 30 mV. The voltage steps were less extreme in the uninduced cells, which uniformly had trouble holding a more negative potential. (B-I) Whole-cell currents were measured in response to a family of voltage steps (as described in A) in four example uninduced cells (B-E, representing the four most negative resting potentials) or in four example induced cells (F-I). A dashed red line is drawn at I=0. Sodium currents are labeled with arrows as INa. In H, an inset shows INa at a higher resolution. (J-K) The current vs. voltage input/output functions summarize the net currents measured approximately 400 ms into the voltage step in 13 uninduced cells (J) and 11 induced cells (K). (L) Comparison of the resting potential of uninduced and induced cells.

The POU4F3-tdTomato hair cell reporter becomes visible after ∼3 days of reprogramming.

Representative images were acquired at 0-, 3- or 7-days post-doxycycline treatment. Scale bar represents 200 µm.

RT-qPCR analysis of the cell line treated with continuous doxycycline or a 7-day pulse of doxycycline over the indicated time points.

Dashed lines indicate post-doxycycline removal. Values are normalized to GAPDH, and a ratio is calculated by dividing the uninduced control (0 hour). Fold-change values are log2-transformed. Error bars indicate SEM. N = 3 biological replicates.

Representative pictures of cultures treated with continuous doxycycline for 7 days (+Dox) or left untreated (-Dox).

Western blot analysis of uninduced control and cells treated with continuous doxycycline for 7, 14, or 21 days.

Protein abundance was quantified by normalizing the signal intensity of each target protein to β-ACTIN.

Dot plot of the top 50 DE genes in P1 supporting cells as identified by Kolla et al. (2020) in our control and reprogrammed clusters.

HC-like clusters show reduced expression of NOTCH1/2/3 and increased expression of hair cell-specific NOTCH ligand genes DLL3, JAG2, and DLK2 relative to control clusters.

(A) UMAP projection of Day 21 reprogrammed RV-R1–3 HCs and residual fibroblasts. (B) Dot plot of Day 21 reprogrammed R5 HCs and C1 iPSCs.

Scatterplots showing relative gene expression levels based on Monocle (pseudotime).

Cells are colored by hair cell-like clusters 1–3 (RV-R1–3). Top: expression of fibroblast-enriched gene examples (GYPC, CPXM1, COL5A2, SDC2) over pseudotime. Bottom: expression of hair cell-enriched gene examples (TNC, MREG, COL9A2, DLK2).

Human and mouse iHCs show highly significant enrichment of each other’s top DEGs.

GSEA was used to analyze the top 100 DEGs in human iHC against mouse iHC, and the top 100 DEGs in mouse iHC against human iHC. FDR (false discovery rate) of < 25% is considered significant.

Gene Ontology (GO) analysis of the top 200 DE genes in RV-R3.

GO Biological Process terms are shown, ranked by p-value.

INSM1, but not IKZF2 (HELIOS) or TBX2, is expressed in HC-like clusters.

(A) UMAP projection of Day 21 reprogrammed RV-R1–3 HCs and residual fibroblasts. (B) Dot plot of Day 21 reprogrammed R5 HCs and C1 iPSCs.

RV-R3 HCs shows expression of vestibular hair cell-specific genes SOX2 and FOXJ1.

SOX2, required for maintaining Type II vestibular HC identity, marks HC progenitors of both cochlear and vestibular nature but is absent from neonatal cochlear HCs. FOXJ1, involved in ciliogenesis, is expressed in vestibular HCs and neonatal cochlear HCs but not mature cochlear HCs.

RT-qPCR analysis of the cell line treated with continuous doxycycline for 7 days.

Cells were treated with doxycycline in mTeSR for 0, 1, or 3 days and moved to hair cell media + doxycycline for the balance of the 7-day reprogramming. Values are normalized to GAPDH, and a ratio is calculated by dividing the uninduced control (0 hour). Fold-change values are log2-transformed. Error bars indicate SEM.

Human induced hair cells (iHCs) are sensitive to cisplatin in a dose-dependent manner.

Cisplatin ototoxicity profile in human iHCs. Error bars indicate SEM.