FZD10 + cells behave as high potential progenitor cells in vitro and Wnt signaling enhanced the sphere-forming ability of FZD10 + cells
The spatiotemporal expression pattern of FZD10 in the mice cochleae was firstly characterized. The Frizzled 10-IRES-creERT2 mice31 were crossed with Rosa26-tdTomato mice and the double-positive (Frizzled 10/tdTomato) mice was found to have normal cochlear morphology and hearing thresholds (data not shown here). Frizzled 10/tdTomato mice were treated with tamoxifen at P0 and the cochleae were harvested at P3 for immunostaining analysis (Fig. 1A). Myosin7a, Tuj1 and Sox10 were utilized as markers for HCs, SGNs and GCs separately, and the cochlea whole mount immunofluroscence showed that the tdTm fluorescence was discovered restricted to the RC and OSL but not in the sensory HC region (Fig. 1B). High magnification images illustrated there was no co-labeling of tdTm and Myosin7a, indicating FZD10 was not expressed in HCs, and the tdTm fluorescence was mainly distributed along the nerve fibers or surrounded the soma of Tuj1-labeled SGNs while very rare Tuj1 + neurons co-expressed tdTm (Fig. 1B’-B’’’). The immunostaining of cochlear cryosections revealed that the tdTm fluorescence was found in a portion of Sox 10 positive cells, and the percentage of tdTm positive GCs (tdTm + Sox10 + cells/ Sox10 + cells) was 13.44% ± 2.50% in P3 mice cochlea (Fig. 1C, C’). We then investigated the expression of FZD10 in cochleae at different ages (Fig. 1D), and found that the percentage of tdTm positive GCs was increased significantly in P7 mice cochlea (30.42% ± 1.91%) compared to P3 mice, but there was no statistical differences between P7 group and P14 (28.35% ± 0.50%) group or P21 group (27.65% ± 1.58%) (Fig. 1E, F). Furthermore, Sox10/tdTm/EdU triple positive cells, which account for 16.10% ± 0.97% of the total tdTm + cells, were observed in the cochlea after EdU was injected to the Frizzled 10/tdTomato mice from P2 to P4 and cochleae were collected at P7 (Fig. 1G, H). Together, these data suggesting a self-proliferation may occur in cochlea FZD10 + GCs in the early postnatal mouse cochleae (P3 to P7), while the number of FZD10 + GCs reaching the peak at P7 and then keep stabilized thereafter.
Next, to determine the proliferation ability of the FZD10 + cells, neurosphere-forming analysis was performed in vitro. The FZD10 + cells were isolated via flow cytometry from P3 Frizzled 10/tdTomato mice cochlea modiolus (supplementary Fig. 1A) and the tdTm positive cells, i.e., FZD10 + cells, constituted about 3.81% of viable cells (supplementary Fig. 1B), the rest of collected modiolus cells were defined as FZD10- cells. Immunostaining of the sorted FZD10 + cells illustrated almost all tdTm positive cells co-expressed Sox10, but there was no tdTm/NeuN double positive cells. The sorted FZD10- cells were positively co-labeled with Sox10 (43.30% ± 3.52%), and NeuN (4.45% ± 0.26%) (supplementary Fig. 1D). There were higher mRNA expression levels of Frizzled 10 and Sox10, but a significant lower mRNA expression of Tubb3, in the FZD10 + cells compared to the FZD10- cells (supplementary Fig. 1C). These data suggest that the sorted FZD10 + cells were GCs of high purity with no SGNs contamination. The sorted FZD10 + cells and FZD10- cells were then cultured for sphere-forming respectively in vitro (Fig. 2A). Neurospheres grew in both FZD10 + cells group and FZD10- cells group and were proliferated for at least five generations. The number of spheres was significantly more and the diameter of neurosphere was larger in FZD10 + cells than that in the FZD10- cell group of each generation (Fig. 2B-D, Supplementary Table 1), indicating that the FZD10 + cells preserve high sphere-forming ability in vitro.
To compare the proliferation capacity of FZD10 + cells in different turns of the cochlea, the sorted FZD10 + cells from the apical, middle and basal turns of the cochlea modiolus were cultured for sphere-forming, respectively (Fig. 2E). Although the numbers of neurospheres generated from the FZD10 + cells of the three turns were similar, the sizes of neurospheres were different with the largest in apical turns, followed by the middle turn group, and the smallest in the basal turn group (Fig. 2F-H, Supplementary Table 1), suggesting an apical-basal gradient of sphere-forming ability of FZD10 + GCs in vitro.
Given that FZDs are the main receptor for Wnt signaling, we then explored the effect of Wnt signaling on the proliferation of FZD10 + cells. Isolated FZD10 + cells and FZD10- cells were cultured for sphere-forming and co-treated with the Wnt agonist Wnt3a and R-spondin1 (RS-1), or the Wnt antagonist IWP-2 (Fig. 2I). The co-administration of Wnt3a and RS-1 increased the number of spheres as well as enlarged the diameter of spheres in both FZD10 + group and FZD10- group, while the treatment of IWP-2 had the opposite effect (Fig. 2J-L, Supplementary Table 1), suggesting that Wnt signaling activation enhanced, whereas the inhibition of Wnt pathway suppressed, the proliferation of both FZD10 + cells and FZD10- cells. Particularly, the activation of Wnt signaling pathway increased the sphere number and the diameter to greater folds in FZD10 + group than these in FZD10- group, and vice versa (Fig. 2M, N). Furthermore, the mRNA expression of neural stem cell related genes showed that Wnt3a and RS-1 co-treatment up-regulated the expression levels of Fstl1, Ptn and Hmgb2 while IWP-2 treatment down-regulated them in the FZD10 + neurospheres (Fig. 2O). Collectively, the results suggest Wnt signaling pathway enhanced the sphere-forming ability of GCs with a more efficient effect on FZD10 + GCs, and that Wnt signaling might contribute to promote the stemness of progenitor GCs.
Wnt signaling pathway enhanced the neural differentiation of the FZD10 + GCs and promoted the induced-neurons to develop electrophysiological features
To evaluate the neural differentiation capacity of FZD10 + GCs as well as the role of Wnt signaling pathway during the process, the sorted FZD10 + cells and FZD10- cells were cultured for differentiation with or without Wnt3a and RS-1 treatment (Fig. 3A). Immunostaining was performed with anti-Tuj1 antibody to label the induced-neurons and EdU to label the mitotic cells. The differentiated cells showed typical bipolar neuronal morphology, and Tuj 1 positive cells and Tuj1/ EdU double positive cells were found in both FZD10 + group and FZD10- groups after the differentiation culture (Fig. 3B), suggesting that GCs are capable of both direct neural differentiation and mitotic neural differentiation in vitro. The percentage of Tuj1 + cells (Tuj1 + cells/ total cells) in FZD10 + group was significantly higher than that in the FZD10- group (28.58% ± 1.53% vs. 10.76% ± 0.66%), and it was significantly increased after the co-treatment of Wnt3a and RS-1 (44.16% ± 1.02%), but there was no statistical differences between the FZD10- cells with or without co-treatment of Wnt3a and RS-1 (Fig. 3C), suggesting that Wnt signaling promoted the neural differentiation capacity of FZD10 + cells but not that of FZD10- cells. The Wnt signaling activation also significantly increased the percentage of EdU/ Tuj1 double positive cells (Tuj1 + EdU + cells/ total Tuj1 + cells, Fig. 3D) in FZD10 + cells, which indicates that Wnt signaling also enhanced the FZD10 + progenitors to regenerate neurons by mitosis division. The mRNA expression levels of neuronal specific marker genes including Tubb3, Vglut1, Map2, Prox1 and Scrt2 in differentiated FZD10 + cells were all increased compared to the primary sorted FZD10 + cells, and activation of Wnt signaling pathway further increased them compared to the differentiated FZD10 + cells group (Fig. 3E).
Next, the whole-cell patch-clamp recording was performed to determine the electrophysiological properties of the induced-neurons, and the native SGNs from P3 WT mice (isolated and cultured under the same conditions) were used as positive controls. The induced neurons exhibited resting membrane potential (RMP) of -56.14 ± 4.87 mV in the FZD10 + control group, -45.57 ± 3.69 mV in the Wnt3a + RS-1 group, and − 42.71 ± 1.57 mV in P3 SGNs (n = 7 cells per group, Fig. 3F, G). Notably, the FZD10 + GCs derived induced neurons were able to fire action potentials (AP), and the threshold stimulus current in the FZD10 + control group was larger than that in the Wnt3a + RS-1 group and in P3 SGNs (97.14 ± 8.08 pA, 61.43 ± 5.53 pA and 44.29 ± 2.02 pA) (Fig. 3H). The depolarization duration was longer in FZD10 + control group compared to the Wnt3a + RS-1 group and the P3 SGNs (Fig. 3I). The results suggested that the electrophysiological characteristic of the induced-neurons from FZD10 + GCs after Wnt signaling activation was quite similar to that of P3 native SGNs.
Wnt signaling improved the proliferation ability of FZD10 + GCs after ouabain treatment in cultured cochlea explants
The effect of Wnt signaling on the proliferation ability of GCs after SGNs damage was determined. First, the cultured cochlea explants were treated with ouabain, a sodium-potassium ATPase inhibitor which could selectively cause SGNs damage without direct damage of GCs or the sensory epithelium32, to cause SGNs damage (Supplementary Fig. 2A). The percentage of proliferated GCs (EdU + Sox10 + cells / total Sox10 + cells) was slightly but significantly increased after ouabain induced significant SGNs loss, compared with the control group (Supplementary Fig. 2B-D). β-Catenin was found to migrate from cytoplasm to the nucleus of GCs after ouabain treatment (supplementary Fig. 2E), and the number of β-Catenin positive GCs was increased significantly in the ouabain treated explants compared with the control group (supplementary Fig. 2F). Moreover, the mRNA expressions of Ctnb1 and Axin2, two known Wnt target genes, were increased in ouabain treated group (supplementary Fig. 2G). These results indicate the self-proliferation of GCs was triggered and the Wnt signaling was activated in GCs after SGNs were damaged by ouabain.
Next, cultured cochlea explants from WT mice were treated with ouabain for 24h and then the medium was changed with supplement of EdU with Wnt3a and RS-1 or IWP-2 (Fig. 4A). The percentage of proliferated GCs were significantly increased in Wnt3a and RS-1 administration group compared to the ouabain-only group (10.20% ± 0.44% vs. 7.40% ± 0.69%), yet the addition of IWP-2 significantly decreased the percentage (4.19% ± 0.34%) (Fig. 4B, C). More importantly, when the Frizzled 10/tdTomato mice were used and treated with the same drugs as above, the proliferated GCs derived from FZD10 + GCs lineage occupied 61.39% ± 4.09% of total proliferated GCs (Sox10+/tdTm+/EdU + cells / Sox10+/EdU + cells), which was much more than the percentage of proliferated GCs derived from FZD10- GCs (Sox10+/tdTm-/EdU + cells / Sox10+/EdU + cells, 38.61% ± 4.09%) (Fig. 4D, E). Supplement of Wnt3a and RS-1 further enhanced the proliferation ability of FZD10 + GCs, and the proportion of FZD10 + GCs-proliferated GCs increased to 75.26% ± 3.38% (Fig. 4D, E). Furthermore, the tdTm+/Tuj1 + cells, indicating the newly-regenerated-neurons from FZD10 + GCs, were found in the cultured cochlea explants after ouabain treatment, with much more tdTm+/Tuj1 + cells in the apex turns than that in the middle turns, while few such cell was found in the base turns (Fig. 4F-H). Co-treatment of Wnt3a and RS-1 significantly increased the number of tdTm+/Tuj1 + cells in the apex and middle turn cochlea, yet no such effect in the basal turn explants was observed (Fig. 4F-H).
Collectively, these data indicate that FZD10 + GCs might be a major type of GCs which function as progenitors to proliferate and different after SGNs damage as well as responding to Wnt signaling regulation.
Lineage tracing of FZD10 + cells in vivo
The fate of FZD + cells in cochlea during mice development was tracked. Tamoxifen was treated to the Frizzled 10/tdTomato mice at P0 and EdU was i.p. injected daily from P2 to P4, and cochlea samples were collected at P7 (Fig. 5A). Utilizing another SGN marker NeuN, we found an average of 33.50 ± 1.75 tdTm+/NeuN + cells, defined as FZD10 + GCs-lineage-derived new SGNs, per cochlea in the P7 mice, and 12.50 ± 0.76 tdTm+/EdU+/NeuN + cells in each cochleae (Fig. 5B, D), indicating the SGNs differentiated from FZD10 + GCs after mitosis. To compare with the newly induced SGNs derived from all GCs, we further analyzed the Sox 10-icre/ERT2/ Rosa26-tdTomato double-positive (Sox 10/tdTomato) mice with the same Tamoxifen and EdU treatment (Fig. 5A), and there were 55.33 ± 2.08 tdTm+/NeuN + cells and 31.0 ± 3.04 tdTm+/EdU+/NeuN + cells per cochlea in P7 Sox10/tdTomato mice (Fig. 5C, E). After comparing the numbers of induced-neurons from FZD10 + GCs lineage with all GCs lineage, we concluded that FZD10 + GCs lineage-derived new SGNs account for about 60.54% of new SGNs derived from glial lineages, and about 40.32% in the mitotic differentiation type.
We next assessed how trans-differentiation of FZD10 + GCs into new SGNs would progress as mice aged. The number of tdTm+/NeuN + cells were decreased in P14 (22.33 ± 1.80 per cochlea), P21 (22.0 ± 1.98 per cochlea) and P28 (13.33 ± 1.23 per cochlea) cochleae compared with that in P7 cochleae, and we did not detect any tomato/EdU/NeuN triple-labeled cells in cochlea of above ages (supplementary Fig. 3A-F, I), suggesting that the mitotic neural differentiation ability of FZD10 + GCs was diminished after P7.
SGNs in mammalian cochlea are divided into two types, i.e., type I SGNs, account for 90%-95% of auditory neurons and primarily innervate the inner HCs, and type II SGNs account for the remaining 5%-10% of SGNs and innervate the outer HCs. Type I SGNs can be further classified into three molecularly distinct subtypes, type Ia, Ib, and Ic according to their differential transcriptional profile33, 34 at the age of P28. Here, we further investigated the subtype distribution of induced-SGNs generated from FZD10 + GCs in p28 mice cochlea in vivo. The subtypes of SGNs were distinguished with the markers of type Ia (CALB2), Ib (CALB1), Ic (POU4F1) or II SGNs (PRPH) (Fig. 5G). The triple-co-localizations of Tuj1, tdTm and CALB2, CALB1, POU4F1, or PRPH were all found in the P28 mice cochlea, respectively (Fig. 5G), indicating that the FZD10 + GCs can differentiate into each type of SGNs in vivo. The percentage of each SGN subclass generated from FZD10 + GCs was 48.20% ± 2.93%, 21.65% ± 0.83%, 24.35% ± 4.17%, and 6.42% ± 1.18% for type Ia, Ib, Ic and type II SGNs respectively (Fig. 5F), which is similar to the normal percentages of SGNs subtypes in WT mice as previously reported (type Ia: 52% ± 2%, type Ib: 26% ± 1%, type Ic: 27% ± 1%)33, 34.
Wnt signaling enhanced the proliferation and neural differentiation of FZD10 + GCs after ouabain treatment in vivo
To investigate the role of Wnt signaling in FZD10 + GCs proliferation and neural differentiation after SGN damage in vivo, ouabain was injected into P2 Frizzled 10/tdTomato mice, EdU, Wnt3a and RS-1 were treated daily from P2 to P4, cochlea were harvested at P7 (Fig. 6A). The percentage of proliferated FZD10 + GCs (tdTm + EdU + cells/ tdTm + cells) was increased after ouabain injection compared to control group (21.55 ± 0.88% vs. 15.30 ± 0.25%), while Wnt3a and RS-1 co-treatment further increased it to 29.05% ± 0.63% (Fig. 6B, C). Furthermore, the number of tomato/NeuN double positive cells per cochlea, as well as the number of tomato/NeuN/EdU triple positive cells, were both added significantly after Wnt activation compared to the ouabain-only group in P7 cochlea (60.83 ± 2.96 vs. 46.50 ± 3.06, 25.67 ± 2.19 vs. 18.17 ± 0.79, Fig. 6D-E). However, when it came to the cochlea in adult mice, which were treated with tamoxifen and ouabain at P23, EdU, Wnt3a and RS-1 at P23 to P25, no significant differences of the above NeuN positive cells were found in P28 mice cochlea compared to that in the P21 cochlea (supplementary Fig. 3G, H, J).
Besides, the electrophysiological properties of the FZD10 + GCs-lineage-derived new SGNs in vivo were detected by the fluorescent whole-cell patch-clamp recording. Ouabain was injected at P2 Frizzled 10/tdTomato mice, Wnt3a and RS-1 were treated daily from P2 to P4, single cells from cochlea modiolus at P7 were isolated and cultured overnight for recovery, the round shaped bio-polar neurons with tomato fluorescence were recognized as the induced-neurons derived from FZD10+ (Fig. 6F), and P7 native SGNs were used as positive controls. The induced-neurons were capable of firing AP, with a lager threshold stimulus current in the FZD10 + ouabain group than that in the FZD10 + ouabain + Wnt3a + RS-1 group and in P7 native SGNs (102.86 ± 4.21 pA, 68.57 ± 6.34 pA and 54.29 ± 2.02 pA, n = 7 cells per group) (Fig. 6G, H), and the depolarization duration was much longer in FZD10 + ouabain group compared to the FZD10 + ouabain + Wnt3a + RS-1 group and the P7 SGNs (Fig. 6I). Overall, the results indicate that Wnt signaling enhances the proliferation and neural differentiation of FZD10 + GCs after SGN damage in vivo in early postnatal period.
Single-cell RNA sequencing analysis of the proliferated and differentiated FZD10 + GCs
Single-cell RNA sequencing analysis was performed to identify the characteristics of cell populations after FZD10 + GCs proliferation and differentiation. Samples of 5 groups were collected, i.e., the sorted FZD10 + GCs cultured overnight (F10_ctrl group, 12838 cells collected), cultured after proliferation (F10_pro group, 14567 cells collected), cultured after proliferation with Wnt3a + RS-1 co-treatment (WNT_pro group, 16751 cells collected), cultured after differentiation with or without Wnt3a + RS-1 co-treatment (F10_diff group, 16551 cells collected and WNT_diff group, 9400 cells collected) (Fig. 7A). Conjoint analysis of RNA sequencing data showed that the total cells of 5 groups could be divided into 7 cell populations according to their markers, defined as Schwann cells, satellite cells, proliferating stem cells, neurogenesis-like cells, and non-neurogenesis-like glial cells (Fig. 7B-D, supplementary Fig. 4A). A very small amount of epithelial cells and macrophages were also detected, which were likely contaminants during FACS purification (Fig. 7B-D, supplementary Fig. 4A). The proportion of each cell population showed that the majority cells in the F10_ctrl group were Schwann cells and satellite cells, suggesting that FZD10 expressed in these two types of GCs in cochlea, which is consistent with our above result that FZD10 expressed exclusively in GCs but not in other cell types in cochlea in Fig. 1. After proliferation culture, the number of non-neurogenesis-like GCs increased dramatically and were the main cell population in the F10_pro and WNT_pro groups (Fig. 7C, supplementary Fig. 4A). It is reasonable that these cells presented molecular profiles of GC type since they are proliferated cells from the sorted and purified FZD10 + GCs. However, according to the expression of related markers, these cells cannot be clearly defined as Schwann cells or satellite cells, which might because they are still in developing stage. Notably, the cluster of proliferating stem cells increased significantly in WNT_pro group compared to the F10_pro group (Fig. 7C, supplementary Fig. 4A), indicating that the activation of Wnt signaling pathway promoted the GCs proliferation and the dedifferentiation into stem cells. We further compared the gene expression of the proliferating stem cells between F10_pro and WNT_pro group. A directed cross-talk network between DEGs enriched pathways showed that genes in the Wnt pathway affected several downstream pathways, including inflammation-related (TNF signaling, IL-17 signaling pathway, etc.) and signaling pathways regulating pluripotency of stem cells. On the other hand, the HIF-1 signaling pathway might upstream regulated the Wnt signaling, which also affected the pluripotency pathway (Fig. 7E). Furthermore, the cross-talk network showed the interaction between the Wnt pathway and the upstream and downstream pathways, as well as the key genes associated with the corresponding connections (Fig. 7F). It suggests that Hif1a, Cdkn1b, Cdkn1a and other genes in the HIF-1 signaling pathway regulate the gene expressions of Wnt signaling, and then Lef1, Bambi, and Csnk1e in Wnt pathway are affected to act on the pluripotency pathway, which was also confirmed by our Quantitative PCR analysis as shown in Fig. 7G. Meanwhile, Wnt signaling pathway also regulates the pluripotency pathway via effecting on the Hippo pathway (Fig. 7E, F).
After the differentiation culture, the vast majority of cell type became the neurogenesis-like cells in the F10_diff and WNT_ diff groups (Fig. 7B, C and supplementary Fig. 4A). These cells expressed markers associated with neurogenesis and neural development, such as Ptx3 35 and Col8a136. To clarify the characteristics of these cells, we compared the neurogenesis-like cells with the non-neurogenesis-like GCs populations. The two clusters were further divided into six sub-clusters according to their expression profiles, with sub-cluster 1, 2 and 3 mainly in the neurogenesis-like cells, and cluster 4, 5 and 6 mainly in the non-neurogenesis-like GCs population (Fig. 8A, B and supplementary Fig. 4B). Interestingly, type I SGN marker genes Slc4a4, Mfap4, Fzd2, Ntrk3, Fgf1833, 37, and type II SGN marker genes Chd3, Kctd12, etc.33, were all robustly expressed in cluster 1–3, with very low expression levels in clusters 4–6 (Fig. 8C). Voltage-gated sodium channels initiate and propagate APs in neurons, which is one of the specific characteristics of neuronal physiology. Here, the genes related to neuron-specific sodium channels, such as Scn2a, Scn2b, and Scn3b were all pronouncedly expressed in clusters 1, 2 or 3, whereas barely detected in clusters 4–6 (Fig. 8C). Immunostaining results confirmed the expression of sodium channel protein Nav1.2, encoded by the Scn2a gene, in SGNs but not in the cochlea GCs (Supplementary Fig. 4C). These results together indicate the generation of SGN-like cells in sub-clusters1-3. Then we used ‘AUCell’ to quantify the expression activities of GC differentiation related molecular functions and biological processes. The activities of GC differentiation, positive regulation of neuron migration, regulation of GC differentiation, regulation of synapse assembly, etc., were all much higher in cluster 1, 2 and 3 compared to these in clusters 4–6, correspondingly, the highly expressed genes associated with these functions include Dner, Bmp4, Htra2 and Nrp1, etc. (Fig. 8D). Transcription factors (TFs) play important roles in the process of neuronal differentiation, we then inferred the regulatory activity of the TF in the 6 sub-clusters by the SCENIC algorithm. The TFs with high regulon activities were shown in the heat map, and the top three TFs, i.e. Pou3f4, Maf and Foxp1, presented obviously much stronger regulatory activities in sub-cluster 1–3 compared to clusters 4–6 (Fig. 8E), which means these three TFs might play key roles in regulating the expression of genes related to SGN differentiation in these sub-cells. Finally, cell differentiation trajectory analysis via monocle3 showed that cells in sub-cluster 1, 2 and 3 were differentiated from cells in cluster 4, and the differentiation direction was cluster 4- 1- 3 − 2 (Fig. 8F). This result together with the finding that, as shown in Fig. 8B, the end point of FZD10 + GCs differentiation, i.e. cluster 2 cells, account for a higher proportion in WNT_diff group than in F10_diff group, further indicate that Wnt activation might promote the neural differentiation process of FZD10 + GCs.