Neuronal differentiation of hair-follicle-bulge-derived stem cells co-cultured with mouse cochlear modiolus explants

Stem-cell-based repair of auditory neurons may represent an attractive therapeutic option to restore sensorineural hearing loss. Hair-follicle-bulge-derived stem cells (HFBSCs) are promising candidates for this type of therapy, because they (1) have migratory properties, enabling migration after transplantation, (2) can differentiate into sensory neurons and glial cells, and (3) can easily be harvested in relatively high numbers. However, HFBSCs have never been used for this purpose. We hypothesized that HFBSCs can be used for cell-based repair of the auditory nerve and we have examined their migration and incorporation into cochlear modiolus explants and their subsequent differentiation. Modiolus explants obtained from adult wild-type mice were cultured in the presence of EF1α-copGFP-transduced HFBSCs, constitutively expressing copepod green fluorescent protein (copGFP). Also, modiolus explants without hair cells were co-cultured with DCX-copGFP-transduced HFBSCs, which demonstrate copGFP upon doublecortin expression during neuronal differentiation. Velocity of HFBSC migration towards modiolus explants was calculated, and after two weeks, co-cultures were fixed and processed for immunohistochemical staining. EF1α-copGFP HFBSC migration velocity was fast: 80.5 ± 6.1 μm/h. After arrival in the explant, the cells formed a fascicular pattern and changed their phenotype into an ATOH1-positive neuronal cell type. DCX-copGFP HFBSCs became green-fluorescent after integration into the explants, confirming neuronal differentiation of the cells. These results show that HFBSC-derived neuronal progenitors are migratory and can integrate into cochlear modiolus explants, while adapting their phenotype depending on this micro-environment. Thus, HFBSCs show potential to be employed in cell-based therapies for auditory nerve repair.


Introduction
Sensorineural hearing loss (SNHL) can be induced by a variety of causes, such as genetic mutations, prolonged exposure to loud noise, ototoxic drug treatment, or simply as a result of PLOS  ageing, and is frequently the result of irreversible damage to, and subsequent loss of, hair cells and auditory neurons [1]. The degree of neuron loss is of significance for hearing-impaired patients using a cochlear implant (CI), because this device directly stimulates auditory neurons. Stem cells may counteract the effect of neuronal degeneration and the ensuing neuron loss [2]. Here, stem cells may serve different purposes: their descendants can replace spiral ganglion cells and/or glial cells, but regeneration may also be stimulated by stem-cell-driven paracrine secretion of cytokines and growth factors [3]. This could be of interest in clinical applications, because it can be surmised that in deaf patients stem cells will lack trophic support from hair cells and perhaps also from non-sensory cells.
In 2012, Chen et al. reported that human embryonic stem-cell-derived otic progenitors could be successfully incorporated into inner ear tissue in an animal model of SNHL and differentiated into neurons which subsequently innervated the hair cells [4]. Nevertheless, it would be ideal for future clinical applications to have suitable autologous stem cells in order to avoid the risk of graft rejection. Cell-based therapy using autologous neural-crest-derived stem cells (NCSCs) may represent such an attractive therapeutic option. During embryogenesis, neural crest cells delaminate from the dorsal neural tube and migrate laterally towards the otic placode, where they develop in close association with cells from the otic placode [5][6][7]. Moreover, during embryogenesis, the neural crest and the otic placode develop and differentiate in a similar way, especially since both originate from the lateral border of the neural plate [5,6,[8][9][10][11][12][13]. Hence, NCSCs may easily be directed towards an appropriate neuronal or glial phenotype, which is necessary for auditory nerve repair [5]. Another attractive characteristic of NCSCs is that they are migratory by nature. Given this property, it may be surmised that stem cells implanted into the basal cochlear turn will move from the transplantation site further into the damaged area, or will migrate to more apically located cochlear turns. In the adult body, niches with residing NCSCs, such as the hair follicle bulge, can be easily accessed. Only minimally invasive surgery is necessary to harvest hair-follicle-bulge-derived stem cells (HFBSCs), which opens the way to autologous transplantations. Taken together, HFBSCs would be well suited to study cell-based therapy in animal models of SNHL. However, HFBSCs have never been used in inner ear regeneration research. Therefore, before starting transplantation studies, it is important that a number of prerequisites are met in vitro. The current study aimed on establishing if HFBSCs demonstrate three key factors to ensure successful auditory nerve repair in CI users: (1) retention of the nature as a migrating neural crest cell, (2) their incorporation into damaged cochlear tissue (i.e., without hair cells) and (3) their capability to differentiate into auditory neuronal or glial precursor phenotypes after incorporation-without the support of growth factors from hair cells.
To mimic the situation in hearing-impaired patients, we used cochlear modiolus explantsdevoid of sensory hair cells-from adult mice and co-cultured the explants with genetically manipulated HFBSCs to investigate the aforementioned key factors [14]. We used stem cells which express copepod green fluorescent protein (copGFP) under the control of either the constitutively active elongation factor 1α (EF1α) or the promotor of doublecortin (DCX), the neuronal migration protein [15].

Material and methods Animals
Healthy, adult (> 23 days) male and female mice (strain C57Bl/6) were bred and housed in the Animal Care Facility of Leiden University Medical Center (LUMC, the Netherlands). Animal care and handling were in accordance with the guidelines and regulations as stipulated by the Dutch Experiments on Animals Act (WoD) and the European Directive on the Protection of Animals Used for Scientific Purposes (2010/63/EU) and approved by the Animal Experiments Committee of the LUMC (DEC permit 10172).

Isolation and dissection of adult mouse modiolus and quarter-turn explants
Auditory bullae were dissected from the adult mouse skull (>23 days) within 10 minutes after euthanasia of the mice and the cochleas were removed and placed in sterile DMEM/Ham's F-12 1:1 containing 1% GlutaMax™ (100x) and 1% antibiotic/antimycotic solution supplemented with 10% FBS. The bony capsule of the cochlea was gently peeled off, the lateral wall tissue was discarded and the whole modiolus taken out using a dissection microscope. The modiolus with the organ of Corti still attached was placed in sterile BGM in a humidified incubator until further processing (within two hours).
Some of the specimens were directly co-cultured with HFBSCs in order to investigate the migration of transduced HFBSCs towards whole modioli. Another portion of the specimens was used to prepare quarter-turn explants. These were prepared by gently cutting the isolated modiolus along a transverse plane and removing the central core of auditory neurons and any excess bone from the inner rim of the osseous spiral lamina followed by dissection, resulting in whole-turn explants containing the osseous spiral lamina and the organ of Corti. Next, the explants were cut into four and these quarter-turn explants were placed in PDL-coated wells.

Degeneration of hair cells
To mimic a sensory epithelium devoid of hair cells, comparable to the damaged epithelium in deaf patients, we had to ascertain the absence of hair cells under our culture conditions. Therefore, instead of culturing the explants in artificial perilymph, the modioli were kept in BGM for two hours or cultured in PDL-coated dishes for 36 hours and subsequently fixed for immunohistochemistry (Fig 1).

Migration of HFBSCs
The migratory potential of EF1α-copGFP-transduced HFBSCs (Fig 2A) was investigated by seeding the cells alongside the border of a well in a PDL-coated 12-well dish, while a whole modiolus explant was placed in the center (for a schematic set-up, see Fig 2B). Cells and modiolus explant were allowed to attach at 37˚C for 60 minutes, after which 500 μl BGM were added carefully (Fig 2B'). Cell migration velocity was calculated by dividing the migration distance by the time (hours) at various time points (Fig 2B'-2D'). Cell migration distance (in mm) was the distance between the border of the area occupied by the cells after seeding and the position of the cells towards the brim of the explant. Measurements continued until the first fluorescent cells were visible underneath the modiolus explant ( Fig 2D). Migration was monitored using an Olympus IX70 fluorescence microscope (FITC filter settings) equipped with a Leica DFC340 FX digital camera. Digital images were acquired and stored using Leica Application Suite Advanced Fluorescence (LAS AF) software (version 1.9.0). All images were subsequently processed using Adobe1 Photoshop1 CC software (version 2014.2.1).

Co-culture of HFBSCs
To start the co-culture experiment, 10 μl BGM containing 2x10 4 HFBSCs were pipetted on top of quarter-turn explants. The co-cultures were incubated at 37˚C for 60 minutes, allowing attachment of both cells and explants to the bottom of the PDL-coated well, after which 500 μl BGM were added carefully. Parallel cultures were performed with the same HFBSCs cells of the co-culture, in order to establish immunophenotypic changes of the cells. Every other day, half of the medium was refreshed. After 7 days in culture, the explants were taken out of the culture and fixed with pre-warmed (37˚C) 1% formaldehyde in PBS. To protect copGFP from photobleaching, the explants were stored in 1.5 ml amber-colored Eppendorf 1 Safe-Lock microcentrifuge tubes (Sigma-Aldrich) at 4˚C.

Immunohistochemistry
The fixed quarter-turn explants and control cells were washed with phosphate-buffered saline (PBS) containing 0.05% Tween-20 (Promega, Madison, WI, USA) for 5 min and permeabilized with 0.3% Triton X-100 (Sigma-Aldrich) in PBS containing 5% normal goat serum (Dako, Glostrup, Denmark) for 60 minutes. After washing with PBS/Tween-20, the specimens were incubated in blocking solution containing 0.1% Triton X-100 and 5% normal goat serum in PBS for 30 minutes. The specimens were incubated overnight with primary antibodies at room temperature. Primary antibodies used in this study are listed in Table 1. After washing, the specimens were incubated in blocking solution for 10 minutes followed by incubation with goat anti-mouse or goat anti-rabbit fluorochrome-conjugated (Alexa1 Fluor 555, 1:300; Alexa1 Fluor 750, 1:50) secondary antibodies, at room temperature for 90 minutes. Nuclei were counterstained with DAPI (Invitrogen) at a dilution of 1:1000 in PBS for 15 minutes. The specimens were mounted in a drop of Roti1-Mount FluorCare (Carl Roth GmbH + Co. KG, Karlsruhe, Germany) under a cover glass. All specimens were examined with a Leica DM5500 B fluorescence microscope (filter settings: TXR, Cy7, FITC and DAPI), equipped with a Leica DFC365 FX digital camera. Digital images were acquired and stored using Leica Application Suite X (LAS X) software. All images were subsequently processed using Adobe1 Photo-shop1 CC software (version 2014.2.1).

Phalloidin staining
The presence of F-actin in quarter-turn explants was demonstrated with Alexa1 Fluor 555-conjugated phalloidin (Invitrogen, Carlsbad, CA, USA). F-actin is a cytoskeleton marker and used to stain extant cells in the quarter-turn explant. Fixed specimens were washed three times with PBS and incubated with phalloidin (diluted in PBS containing 0.05% Tween-20) for 60 minutes followed by three washing steps with PBS. Next, nuclei of the specimens were counterstained with DAPI (diluted 1:1000 in PBS) for 15 minutes followed by washing with PBS and covered with Roti1-Mount FluorCare and a cover glass.

Statistical analyses
Data of the migration experiment were processed in Microsoft Excel 2010 (Microsoft Corporation, Redmond, WA, USA) and expressed as mean ± SD. Statistically analysis was performed using GraphPad Prism version 6.02 software (GraphPad Software, La Jolla, CA, USA).

Adult hair cells degenerate in BGM within 2 hours
Quarter-turn explants, which were fixed in formaldehyde immediately after dissection, stain for the adult hair cell marker myosin VIIa (n = 6, Fig 1A). When cultured in BGM for 2 hours-not in artificial perilymph-only some diffuse, pectinate staining for myosin VIIa could be observed, indicating that already most hair cells had been lost during such a short period of culture (n = 3, Fig 1B). After incubation in BGM for 36 hours, the specimens were completely devoid of staining for myosin VIIa (n = 3, Fig 1C).
indicates the position of the cells located underneath the modiolus. (D') Overview of area depicted in D (boxed area) illustrating how the first cells reached the modiolus after migrating another 1.2 mm and 0.4 mm, respectively (arrow), after 92 hours. Cells migrated over a cell-free zone (gray) towards the modiolus. In control experiments, i.e. wells without a modiolus in the center, cells proliferated at the border of the well, but did not migrate towards the center of the well (data not shown).

Transduced HFBSCs migrate towards modiolus explant
EF1α-copGFP HFBSCs crossed the cell-free zone (radius: 7.5 ± 0.6 mm μm) with a velocity of 80.5 ± 6.1 μm/h (n = 4) towards the modiolus explant and arrived, often in groups, at the explant within 2-3 days (Fig 2C and 2C'). Cells located underneath the explant could clearly be distinguished from the modiolus tissue by adjusting the focus level during observation, i.e., to verify whether fluorescent cells arrived at the tissue (Fig 2D). In control experiments, i.e. wells without a modiolus explant in the center, cells proliferated at the border of the well, but did not migrate towards the center of the well (data not shown).

Stem cells form a distinct fascicular pattern
In all cultures of the migration experiment, fluorescent HFBSCs invaded the explants within 2 days (2.2 ± 1.6 days) after arrival. Within the next 5 days (5.2 ± 1.6 days), the cells formed a distinct fascicular pattern (n = 10). To shorten culture time, the protocol was adapted by seeding 2x10 4 EF1α-copGFP HFBSCs directly onto quarter-turn explants (n = 50). We did not observe differences in fascicular organization of HFBSCs in the explants between both approaches. The time course of the fascicular pattern formation by HFBSCs is depicted in Fig 3. At day 1, cells are visible underneath the tissue, attached to the bottom of the well and invaded the explant after 2 days of co-culture. At day 3, the number of HFBSCs increased within the tissue and cells formed a distinct fascicular pattern within 5 days. This pattern became more enhanced over the next two days, up to day 7.

Similar distinct fascicular pattern formation of HFBSCs transduced with EF1α-copGFP and DCX-copGFP
After obtaining a fascicular pattern of EF1α-copGFP HFBSCs in quarter-turn explants of adult mice (Fig 4A), we investigated the role of the neuronal migration protein DCX in the process of pattern formation and seeded DCX-copGFP HFBSCs onto quarter-turn explants. During the next 4 days (4.1 ± 0.9 days), different cells became fluorescent and at day 5 (5.3 ± 0.9 days after seeding; n = 7), the fluorescent HFBSCs were organized in a distinct fascicular pattern (Fig 4B), similar to the pattern displayed by EF1α-copGFP HFBSCs. Therefore, the formation of a fascicular pattern is not related to the use of a specific promoter (p = 0.862, two-tailed, unpaired t-test, 95% confidence interval).  Fluorescent EF1α-copGFP HFBSCs, in the fascicular pattern, immunostain for (auditory) neuronal markers Fluorescent (copGFP-containing) cells in the modiolus explants showed immunostaining for the NCSC marker SOX9 (Fig 5A). Nestin, a marker for NCSCs and neural progenitors, was weakly positive in most copGFP-expressing and SOX9-positive HFBSCs (Fig 5A). TUBB3, a marker for young neurons, was weakly positive, not only in fluorescent HFBSCs but also in some non-fluorescent (endogenous) cells (Fig 5B). Medium-chain neurofilament protein (NF-M), generally expressed early during neuronal maturation, was exclusively positive in fluorescent HFBSCs (Fig 5C), while the marker for mature neurons, heavy-chain neurofilament protein (NF-H), was negative ( Fig 5D). Most copGFP-positive cells arranged in the fascicular pattern immunostained for ATOH1, a marker for neurons and hair cells in the developing cochlea (Fig 5C). The majority of ATOH1-positive cells was also positive for NF-M. Since hair cells are not present in the explant, the single ATOH1-positive, copGFP-negative cell is most likely a HFBSC that has not been successfully transduced (transduction efficiency is ± 84%). The immunostaining for SOX2, a marker for developing hair cells, was negative. In the pattern, some copGFP-positive HFBSCs, but also many non-fluorescent cells, were immunopositive for F-actin (Fig 5E). The non-GFP-fluorescent, F-actin-positive cells demonstrate a morphology reminiscent of interdental cells [19].
Immunostaining for the adult hair cell marker myosin VIIa was negative, both in the hair cell survival experiment (Fig 1) and in the co-cultures, indicating that the outer and inner hair cells were lost from the explants, even in short-term cultures. None of the glial markers (GFAP, S100, KROX-20 and MPZ) were immunopositive in copGFP-positive HFBSCs. The immunoprofile of the HFBSCs in the pattern compared to non-co-cultured HFBSCs are listed in Table 2. This table shows that HFBSCs changed their phenotype after incorporation in the explants and that they became positive for several (young) neuronal markers and negative for all glial markers. All pertinent positive and negative controls showed appropriate results.

Discussion
The results show that our population of HFBSCs meets three important requirements to enable their contribution to successful inner ear regeneration: they are migratory, they integrate into cochlear tissue, and have the capability to adapt a neuronal phenotype depending on the microenvironment.
The cell migration velocity of HFBSCs was relatively fast (80.5 ± 6.1 μm/h) as compared to other cell types as determined in wound-healing migration assays [20][21][22][23], indicating that they conserved the migratory nature of NCSCs. Migration of cells is to a large extent dependent on the type of extracellular matrix (ECM) and the presence of chemical attractants, such as growth factors. The cells in this experiment were seeded on a poly-D-lysine matrix, which allows migration of neural crest descendants [24,25]. On the other hand, we considered it unlikely that chemotactic factors were present in the tissue, because the cultured explants were devoid of hair cells. However, the remaining non-sensory cells may have expressed growth factors, which could function as a chemoattractant for HFBSCs [26]. In addition, it is generally known that certain growth factors can directly bind to ECM proteins [27]. Hence, it is possible that neural growth factors are present in the ECM of the explants, i.e. in the spiral limbus, the osseous spiral lamina and the basilar membrane. In the co-culture experiment, growth factors from the non-sensory cells and the ECM may therefore have attracted the HFBSCs. Within five days upon arrival into the explant, the HFBSCs formed a distinctly fascicular pattern while co-localizing with F-actin-positive, non-GFP-fluorescent cells (Fig 3E). Within seven days after their settlement, HFBSCs developed a neuronal phenotype corresponding to the expression of copGFP under regulation of the DCX promoter. This is an unusually rapid neuronal differentiation, for it has been reported that neuronal progenitors generally need three weeks to achieve NF-M positivity [28]. We hypothesize that there are several biochemical cues present in the explants, probably working in synergy, promoting a neuronal phenotype of the locally present HFBSCs. Firstly, although hair cells have disappeared, neural growth factors could still be present in supporting cells or be bound to ECM proteins. Secondly, the type of ECM proteins may be of importance: among the different classes of ECM molecules that are present in the cochlea, laminin has been shown to be particularly effective in stimulating neuronal differentiation [25,[29][30][31]. Thirdly, given the fasciscular pattern of HFBSCs within the explants, the cells are probably dwelling in or near Huschke's teeth, i.e, in the upper region of the spiral limbus. This seems a plausible assumption, for the non-copGFP-fluorescent cells near the F-actin-positive, copGFP-fluorescent HFBSCs (Fig 3E) are morphologically similar to interdental cells. It is known that a columnar-shaped ECM stimulates neuronal differentiation of cells [32,33]. Hence, the columnar plates of the spiral limbus matrix (Huschke's teeth) [34] may well have stimulated HFBSCs to differentiate into a neuronal phenotype.
In our opinion, our experiments show that HFBSCs are tolerated in inner ear tissue from adult animals. Indeed, HFBSCs do intermingle with extant cells of the modiolus and can subsequently adapt to the microenvironment. Furthermore, in spite of the absence of trophic factors from hair cells, HFBSCs changed their phenotype into that of young neuronal cells ( Fig  5). Various studies have shown that stems cells isolated from the bulge area of mouse and human hair follicles express nestin, a marker for NCSCs and neural progenitors [35][36][37][38][39][40][41][42][43][44][45]. In these reports, it was established that nestin-positive HFBSCs can differentiate towards a variety of cells such as neurons, glia cells, melanocytes, keratinocytes as well as smooth-and cardiac muscle cells. This broad differentiation capacity illustrates the potential of HFBSCs to serve many regenerative purposes. Our finding that nestin-positive HFBSCs differentiate towards a neuronal lineage, corroborates the pertinent above-mentioned reports (Table 2).
Hence, our report presents clear evidence that these cells may be good candidates for in vivo regenerative experiments in an animal model of SNHL. In upcoming studies we will use HFBSCs in CI-implanted guinea pigs, deafened by means of concomitant administration of kanamycin and furosemide. In this model no hair cells are present, similar to deaf patients and to our in vitro experiments.

Conclusion
In conclusion, the present observations show that HFBSCs are migratory and can successfully integrate into inner ear tissue, i.e., into the modiolus, while adapting a neuronal phenotype.
Hence, the results of this study show the potential of HFBSCs for auditory nerve repair. Our next step will be to investigate in vivo if HFBSCs after transplantation into the spiral ganglion receive the appropriate biochemical stimuli to differentiate into auditory neurons and/or glial cells.