Functional neurons and melanocytes induced from immortal lines of postnatal neural crest-like stem cells

Materials

Tissue culture plastics (Nunc), FCS, and newborn calf serum (NBCS) were obtained from Life Technologies Europe (Ux-bridge, UK). Fibroblast growth factor 2 (FGF2) and neuregulin 1-ß1 (NRG1-ß1) were obtained from R&D Systems Europe (Abingdon, UK). Bone morphogenetic protein 2 (BMP-2) was obtained from PeproTech EC (London, UK). Other peptide factors, TPA, mitomycin C, and 5-azacytidine were obtained from Sigma-Aldrich (Gillingham, UK). NDP-MSH is an analog of aMSH that is resistant to degradation by serum enzymes. Protein factor stock solutions were prepared in PBSA (Dulbecco's PBS lacking CaCl₂ and MgCl₂) with bovine serum albumin (1 mg/ml) as carrier, and stored at -70°C.

Feeder cells

Growth-inactivated feeder cells were either XB2 murine kertinocytes (23), or SC1 dermal fibroblastoid cells (24). XB2 stocks were grown in DME medium with 10% FCS, and SC1 cells in RPMI 1640 medium with 5 or 10% FCS. Feeder cells were prepared by treatment with mitomycin C (6 µg/ml for 3 h), followed by washing, resuspension, and freezing as described previously (25). Feeder cells were thawed and plated usually 1 d before use.

Establishment and passage of NCLSCs

Primary cultures from mouse trunk skin were made as for melanoblast/melanocyte cultures (24). The skin was from neonatal misty (m/m) mice. [The misty mutation impairs proliferation of melanocytes and increases their pigmentation but has no detectable effect on melanoblasts (26), so it seems unlikely that this mutation is relevant to the presence in skin of the cell type described here.] In short, trypsin-eparated epidermal sheets were dissociated with trypsin and EDTA, and the resulting cell suspension was plated onto keratinocyte feeder cells. The basic growth medium was RPMI 1640 with penicillin, streptomycin, and glutamine (2 mM). Incubation was with 10% v/v CO₂ at 37°C throughout. Initial supplements in these cultures were FCS (5%), TPA(200 nM), cholera toxin (CT) (200 pM), FGF2 (20 pM), ethanolamine (1 µM), and phosphoethanolamine (1 µM) as used in a melanoblast growth medium (27). Fresh XB2 cells were added at 3 wk, and after this the medium was supplemented with TPA (200 nM) and CT (50 pM) only. From passage 1, cultures were plated onto SC1 fibroblastoid feeder cells (24). New feeder cells were added every 2–3 wk. Cells were subculured by rinsing gently in PBSA and then in 250 µg/ml trypsin and 200 µg/ml EDTA in PBSA; they were incubated until completely detached, and resuspended and diluted in growth medium as required. At one point cultures were treated with antibiotic G418 (75 µM, 3 d) to kill fibroblasts; this drug can select for melanocytes and melanoblasts (24, 28) and appeared effective also in selecting for NCLSCs.

NCLSC lines, once established, were routinely grown in RPMI 1640 medium with 10% NBCS and 2 nM TPA. For experiments, various supplements were added, at the following concentrations unless specified: FCS (10%), NBCS (10%), TPA (2 nM), NDP-MSH (1 nM), CT (1 nM), 5-azacytidine (300 pM), ascorbic acid 2-phosphate (50 µg/ml), TGFß1 (120 pM), NRG1-ß1 (10 nM), brain-derived neurotrophic factor (BDNF) (25 ng/ml), nerve growth factor (NGF) (50 ng/ml), neurotrophin 3 (NT3) (25 ng/ml), BMP-2 (10 ng/ml), and FGF2 (40 pM).

Other cell lines

Mouse melanoblast lines melb-a (nonagouti, a/a) and melb-s (piebald, Ednrb^s/ Ednrb^s) were grown in RPMI 1640 medium with 10% FCS and 20 nM TPA as described previously (melanoblast medium) (24, 26); SC1 feeder cells were used for melb-a cells and omitted for the final 2 wk. Mouse melanocyte lines melan-a, melan-a2, melan-c (albino, Ty^c/Ty^c), melan-s (Ednrb^s/Ednrb^s), melan-m1 (misty, m/m), melan-p1 (pink-eyed dilution, p-null), melan-rs (recessive spotting, rs/rs), and melb-p3 (p-null) melanoblasts (24, 29–31) were grown in RPMI 1640 medium with FCS (10%), TPA (200 nM), and CT (200 pM) (melanocyte medium). Melan-s1 and melan-a2 are melanocyte lines derived by differentiation from melanoblast lines melb-s1 and melb-a, respectively (26). B16-F1 and K1735 murine melanoma cells, originally obtained from I.J. Fidler (M. D. Anderson Cancer Center, Houston, TX, USA), were grown in RPMI 1640 with FCS (10%).

Digital image acquisition

Unstained cultures for photography were usually fixed in 4% formaldehyde in PBSA, rinsed in distilled water, and air dried; they were photographed in water. TIFF images were obtained at room temperature through an Olympus IMT-2 inverted microscope and ×10 objective (numerical aperture 0.30; Olympus, Tokyo, Japan), using a Pixelink PL-A662 camera and Pixelink A6xx-Capture software (Pixelink, Ottawa, ON, Canada). Immunostained cells were mounted under coverslips in Aquamount (BDH, Poole, UK) and (except where stated,) photographed using a Zeiss Axioplan microscope with an ×40 objective (numerical aperture 0.75) and Axiovision 4.6 digital acquisition software (Carl Zeiss, Oberkochen, Germany). Montages were prepared in Adobe Photoshop (Adobe Systems, San Jose, CA, USA).

Cell proliferation assays

Triplicate cultures for growth/differentiation experiments were plated at the specified density in 3-cm dishes (2 ml/ dish) and cultured as specified. Media were renewed twice weekly. Cells were harvested separately from each plate by trypsinization. Triplicate cell counts from each suspension were made by hemocytometer.

To coat culture dishes with gelatin, the dish bases were covered with a sterile gelatin solution (10 µg/ml in water) and left to stand for 5 min. The solution was removed and the dishes were rinsed in PBSA before plating cultures.

Immunocytochemistry and histochemistry

aPEP8 and aPEP13 rabbit polyclonal antisera, against murine dopachrome tautomerase and Si/gp87, respectively, were kindly provided by Dr. Vince Hearing (U.S. National Institutes of Health, Bethesda, MD, USA). Rabbit monoclonal anti-substance P, rabbit anti-neurofilament 200 (IgG fraction), mouse monoclonal anti-GAP43 (clone GAP-7B10), and Alcian blue were obtained from Sigma-Aldrich. Rabbit polyclonal anti-GFAP and anti-nestin and FITC-conjugated goat anti-rabbit IgG were obtained from Abcam (Cambridge, UK).

Procedures for immunocytochemistry with these antibodies were as described previously (24), except that the secondary antibody conjugate for anti-GAP43, GFAP, and substance P was alkaline phosphatase from EnVision System-AP (Dako, Cambridge, UK) and was used according to the manufacturer's instructions; the secondary antibody for nestin was goat anti-rabbit IgG-FITC (1:200). Nuclei were counterstained where specified with DAPI (1 µg/ml). Primary antibodies were diluted as follows: substance P, 1:40,000; GAP43 and GFAP, 1:10,000; nestin, 1:250; others 1:1000. For neurofilament and TRPV1 staining, mixed neurofilament antibodies (NA 1297; Affiniti Research Products, Exeter, UK; 1:5000) and rabbit anti-TRPV1 antisera (GSK, Harlow, UK; 1:10,000 for GSK-C22 and 1:400 for GSK no. 2) were used. The specificity of GSK-C22 has been reported previously (32), while GSK no. 2 against a different epitope can be preabsorbed using homologous peptide antigen on tissue sections. Cells were washed in PBS, fixed in 4% w/v paraformaldehyde (30 min), permeabilized in Triton X-100 (0.2% in PBS) (30 min), and dehydrated with graded alcohols. Endogenous peroxide was blocked with 0.3% v/v H₂O₂ in methanol (30 min). Cells were rehydrated, and primary antibodies in PBS with BSA (0.5 mg/ml) and sodium azide (1 mg/ml) were applied overnight, then visualized by a nickel-enhanced avi-din-biotin-immunoperoxidase method (33). Nuclei were counterstained with neutral red (0.5 mg/ml), and cells were photographed with an Olympus BX50 photomicroscope.

Northern blot analysis

Poly(A)-enriched RNA was isolated from cell lines and from normal mouse liver and kidney, and separated by agarose gel electrophoresis (5 µg/lane); Northern blot analysis was performed as described previously (31, 34). Probe cDNA sequences were kindly provided by the following: Colin Goding (Oxford Ludwig Institute, Oxford, UK) (Brn2, Mitf, Pax3, Snai2), Ian Jackson (MRC Human Genetics Unit, Edinburgh, UK) (Tyr, Tyrp1, Dct), Mike Clemens (St. George's, University of London, London, UK) (Gapd), Roger Cone (Vollum Institute for Advanced Biomedical Research, Portland, OR, USA) (Mc1r), Christine Farr (Genetics, University of Cambridge, Cambridge, UK) (Sox4), Ed Geissler (Harvard Medical School, Boston, MA, USA) (Kit), Byoung Kwon (University of Ulsan, Ulsan, South Korea) (Si), Rick Lindberg (Amgen, Thousand Oaks, CA, USA) (Efna1), Toshi Shioda (Massachusetts General Hospital Cancer Center, Boston, MA USA) (Cited1), Richard Spritz (University of Colorado Health Sciences Center, Denver, CO, USA) (P), Jim Weston (University of Oregon, Portland, OR, USA) (Pdgfra), and David Wilkinson (National Institute for Medical Research, London, UK) (Epha4). Probes were labeled by random hexamer priming to a specific activity of ~2 × 10⁹ dpm/µg DNA. Prehybridization and hybridization were carried out at 42°C. The final wash was in 0.5× sodium-saline-phosphate EDTA buffer with 0.1% w/v sodium dodecyl sulfate, at 55°C. After washing, filters were exposed for up to3dat -70°C with 2 intensifying screens.

Ligand-responsive calcium fluxes

NC-m4 cells were plated on coverslips and grown for 14 d with FCS, TPA, BDNF, NGF, NT3, and FGF2 to induce neuronal differentiation. The cells were transferred to a perfusion chamber (RC-25;Warner Instruments, Hamden, CT, USA), in a bath solution containing 140 mM NaCl, 5 mM KCl, 2 mM MgCl₂, 1.8 mM CaCl₂, 10 mM glucose, and 10 mM HEPES at pH 7.4. (All inorganic reagents were from Sigma-Aldrich.) Cells were loaded with fluo-4 acetoxymethylester (fluo-4 AM; Molecular Probes, Eugene, OR, USA) for 1 h. This and subsequent steps were at room temperature (25±2°C). Residual fluo-4 was washed out, followed by a 15-min pause for deesterification. Capsaicin was added to the bath chamber at 10 µM final concentration from a 10 mM solution in DMSO, or cinnamaldehyde was added at a final 100 µMfroma10 mM solution in ethanol. Ca²⁺ fluctuations were imaged by exciting fluo-4 at 450–480 nm and detecting emitted fluorescence at >520 nm, using an intensified CCD (CoolView IDI camera system; Photonic Science, Sussex, UK) coupled to a Nikon TE-2000 inverted microscope (Nikon, Tokyo, Japan) and controlled by Image-Pro Plus software (Media Cybernetics, Wokingham, UK). Images were acquired at 2 frames/s with 100 ms individual frame exposure time. Data were further analyzed with Origin 5 (Microcal Software, Northampton, MA, USA). Corresponding time traces show normalized fluorescence intensity (ratio of fluorescence to initial fluorescence, F/F₀) in the area of interest.

Whole-cell voltage-clamp recordings

NC-m4 or NC-m6 cells were grown on glass coverslips in the medium specified, for 2 wk. Whole-cell patch-clamp recordings were conducted at room temperature (21°C) using a MultiClamp 700B amplifier and Clampex 9.2 software (both from Molecular Devices Corp., Sunnyvale, CA, USA). Fire-polished electrodes (2.5–2.7 MOM) were fabricated from 1.0-mm borosilicate glass capillaries (Intracel Ltd, Royston, UK) using a P-2000 Puller (Sutter Instrument Co., Novato, CA, USA). In whole-cell voltage-clamp mode, the capacity current was diminished as much as possible with the amplifier circuitry, and series resistance compensation was set at 25–80%. The pipette potential was zeroed before seal formation, and voltages were not corrected for liquid junction potential. Leakage current was digitally subtracted online using hyper-polarizing control pulses, applied before the test pulse (P/N procedure). Recordings were made within 10 min after forming the whole-cell configuration. The pipette solution contained 140 mM CsF, 1 mM EGTA, 10 mM NaCl, and 10 mM HEPES, pH 7.3. Internal fluoride was used to inhibit high voltage-activated Ca²⁺ currents and help to reveal TTX-resistant currents (35, 36). Electrophysiological data were analyzed using Clampfit 9.2 (Molecular Devices) and Origin 5.0 (Microcal Software) software.

Establishment and growth of immortal neural crest-like cell lines from skin

With the initial aim of growing melanoblasts, 3 independent primary epidermal cultures were prepared from separate neonatal mouse skins and plated on keratinocyte feeder cells in a melanoblast medium, which was initially varied slightly to optimize growth (Materials and Methods). From the first passage (after ~4 wk), these cultures were plated onto SC1 fibroblast feeder cells instead, with 10% NBCS and 2 nM 12-O-tetradecanoyl phorbol acetate (TPA, a component of the melanoblast medium) and grown thus until immortal cells were obtained (see below). Three parallel m/m cultures, grown with keratinocyte feeder cells rather than SC1 cells, all gave rise to melanocyte lines (melan-m1, -m2, and -m3) (31). In all 3 independent cultures grown on SC1 cells, however, an unfamiliar cell type appeared: extremely small, unpigmented, and spindly, with thin processes and a tendency to aggregate into clumps and strands (Fig. 1). Growth was initially very slow, and frozen stocks were prepared at passage 6 after 9 mo. Three independent cell lines were established using this method and were later designated NC-m4, -m5, and -m6 (neural crest-like cells-misty, lines 4–6) (Fig. 1 and Supplemental Fig. 1A).

It was suspected that these cells were pluripotent, as the cell shape and size resembled those of neural crest cells (37), and colonies of melanoblasts, melanocytes, and other cell types (as detailed below) appeared under some conditions. One of the lines, NC-m6, was cloned by limiting cell dilution, to exclude contaminating cell populations. The same additional cell types continued to appear, however, showing that the small cell type could produce them. After experimentation (described below), growth of the small cell type appeared best in medium with NBCS and 2 nM TPA, in which apparently pure cultures with this morphology could be obtained. These supplements were adopted for routine culture. Since these small cells were found able to produce several neural-crest-derived cell types, they have been designated “neural crest-like stem cells.” The NCLSCs showed evidence of bypassing replicative senescence, in that they initially grew extremely slowly, then accelerated somewhat, and now appear immortal. We previously showed that NC-m4 cells express neither p16 nor Arf (effectors of senescence in the mouse) (38). The 2 most extensively characterized lines, NC-m4 and NC-m6, have been grown up to 31 and 30 passages, respectively (=90 population doublings).

Morphology and substrate adhesion of NCLSCs

NCLSCs in all 3 lines were extremely small, with cell-body diameters of ~6-8 µm (Fig. 1). Confluent cultures contained up to ~3 × 10⁶ cells /ml (~10 times that of a confluent culture of normal fibroblasts). At high cell density, NCLSCs cohered into clumps and netlike strands (Fig. 1), which tended to detach and float in the medium. Similar cell clumps have been described in embryonic neural crest cultures (e.g., ref. 37), and they are also reminiscent of the spheres produced by SKPs (19) and hair follicle stem cells (20). Substrate adhesion of NCLSCs appeared generally low, potentially explaining the clumping. Tests indicated that adhesion could be increased by several types of protein coating of the plastic substrate, including collagen or gelatin. In medium with NBCS, no difference in cell number was detectable, but when FCS was used, significantly more cells were obtained in cultures on gelatin (P <0.01), and cells remained attached to the plastic rather than forming clusters. Thus, a gelatin substrate was subsequently used for stock cultures grown in FCS, but not for cultures in NBCS.

We immunostained NC-m6 and NC-m4 cells with the well-established stem cell marker nestin (19). The great majority of the cells (384/417; 91.7±1.3%) showed distinct cytoplasmic immunoreactivity for nestin. Control melan-c albino mouse melanocytes demonstrated only background staining (Fig. 1 and Supplemental Fig. 1B, C).

Induced differentiation of melanoblasts and melanocytes

Melanocytic determination from cultured embryonic crest can be stimulated by agents, including TPA with a cAMP agonist (6), stem cell factor (SCF) and endothelin 3 (7, 39), and Wnt1 (8). Here we tested for promotion of melanocytic differentiation of NC-m4 cells by FCS, TPA, and the cAMP agonists CT and [4-Nle, 7-D-Phe] a-melanocyte-stimulating hormone (NDP-MSH), a stable analog of aMSH. All cultures with CT initially showed apparent cell death, as indicated by floating, phase-dark single cells and lower cell numbers relative to controls at 2–4 wk (P <0.001). In cultures with FCS and 200 nM TPA as well as CT, or sometimes with FCS and 200 nM TPA only, groups and colonies of melanoblast-like cells emerged (larger, bi-, and tripolar). Pigmented cells (melanocytes) were typically seen within the colonies by 10–14 d (Fig. 2B; compare NCLSCs in Fig. 2A). The efficiency of melanoblast/ melanocyte production varied in the long term (possibly depending on batch of FCS). These cells were immunostained for melanosomal proteins Si/Silv/gp87/Pmel (silver) and dopachrome tautomerase (Dct), expressed by melanoblasts and melanocytes (26). Most cells in the colonies showed cytoplasmic staining for both Dct and Si (Si immunostain shown as illustration in Fig. 2C). After growth of the melanoblast-like cells into lines, these and other lineage markers were corroborated by gene expression studies (see below), confirming that the unpigmented cells were melanoblasts. In comparison, NC-m4 NCLSCs expressed no melanocytic marker. Melanocytes were also obtained when NDP-MSH was used instead of CT. They tended to appear in the centers of colonies, consistent with observations that high local cell density promotes melanocytic differentiation (40, 41).

Likewise, no pigmented cells were visible in the cloned NC-m6 line when grown with NBCS and 2 nM TPA, but these appeared 1-2 wk after changing the supplements to FCS, CT, and 200 nM TPA, with or without additional melanogenic factors, showing that determination rather than selection was occurring. Long-term culture with these factors resulted in the establishment of an apparently pure melanocyte line, melan-m6.

Melanogenic factors similarly induced NC-m5 cultures to produce substantial numbers of melanoblasts; these were grown up giving an apparently pure melanoblast line, melb-m5. In turn, a pigmented melanocyte line, melan-m5, was produced from these melanoblasts by growth in melanocyte medium. A pure line of melanocytes (melan-m4) was also produced from NC-m4 cells by expansion of one pigmented colony in melanocyte medium. These lines where tested showed typical melanoblast and melanocyte gene expression (next section).

Gene expression in NCLSCs compared to melanoblasts and melanocytes

Since both melanoblasts and melanocytes, including those derived from NCLSCs, can be grown as stable lines, extensive comparisons of gene expression are possible. We used Northern blot analysis and scanning densitometry to give semiquantitative comparisons of expression of a variety of genes, including regulatory transcription factors, other mRNA markers of neural crest or melanocytes, and some genes relevant to melanoma. The results are summarized in Table 1 and Fig. 2D Of the genes studied, NCLSCs expressed few mRNAs in common with melanoblasts and melanocytes (Table 1). These were 4 transcription factors: Pax3 (paired-box 3) and Snai2 (Snail2)/Slug, required for neural crest development (42, 43); Brn2/Pou3f2/N-Oct3, an octamer-binding factor found in brain, melanoma cells, melanoblasts, and melanocytes (44, 45); and Sox4 (SRY-box 4), required for cardiac neural crest development (46) although also expressed more widely, including in melanoblasts and melanocytes (Table 1). NCLSCs did not detectably express mRNA for Mitf (microphthalmia-related transcription factor), a “master” gene regulator for the melanocyte lineage (14). Mitf, Pax3, Brn2, Sox4, and Snai2 were expressed in all tested melanocytes and melanoblasts (Table 1). We previously reported 2 other transcription factors found in melanoblasts and melanocytes but not in NCLSCs (called at that time “melanoblast precursors”), namely, Tbx2 (T-box 2) (47) and Cited1 (Cbp/p300-interacting transactivator with Glu/Asp-rich carboxy-terminal domain 1) (Table 1), originally called melanocyte-specific gene 1/Msg1, being expressed in melanocytes and well-pigmented melanoma cells (48).

We examined expression of several membrane receptors and one related ligand, with emphasis on those found in melanocytes and melanoma cells. Interestingly, NCLSCs expressed transcripts for 2 receptors found in some melanoma cells but not melanocytes (49) and not detected here in melanocytes or melanoblasts, namely, Ephb3 (Sek4/HEK2), and Pdgfra (platelet-derived growth factor receptor A). It was not surprising that mRNA for receptors Kit and Mc1r were expressed by all tested melanoblast and melanocyte lines, since these cells respond to the respective ligands SCF and MSH (e.g., refs. 24, 41, 50). These mRNAs were not detected in NCLSCs. Two other Eph-family receptors, Epha2 (Eck) and Epha4 (Sek), and a ligand common to both receptors, Efna1 (ephrin A1, LERK1, B61), were all expressed in melanoblasts but not detected in melanocytes or in NCLSCs. In comparison, Epha2 and Efna1 are frequently expressed in melanoma cells but not human melanocytes, while Epha4/ Sek appears to be downregulated in some melanomas (34, 49).

We also quantitated transcripts for several proteins specific to melanosomes (matrix protein Si/Pmel and enzymes Tyr/tyrosinase, Tyrp1/tyrosinase-related protein 1, and Dct), or involved in melanosome biogenesis (Oa1/ocular albinism 1 and P protein/pinkeyed dilution), all required for normal pigmentation (reviewed in ref. 13). None of these mRNAs was expressed in NCLSCs. All were detected in melanocyte lines, and in melanoblasts, with the sole exception of P protein. Thus P appears to be one gene that is upregulated on differentiation to melanocytes. Since expression levels appeared to vary between lines of one type, general quantitative comparisons between cell types would require more lines, although mean expression for these melanoblast and melanocyte lines is shown for reference. Levels of expression could be directly compared in the 2 melanoblast lines melb-a and melb-s1 with those in the pigmented melanocyte cultures derived from these melanoblasts by differentiation, melan-a2 and melan-s1. Differences that seemed large enough in both pairs to be likely to reflect real changes on melanocytic differentiation were a fall to around 25% in the level of Sox4 and a rise (~3-fold) in transcription of the rate-limiting melanogenic enzyme Tyr (tyrosinase).

Normal mouse liver or kidney and mouse melanoma cell lines B16-F1 (lightly pigmented) and sometimes K1735 (unpigmented) were included for comparison. None of the transcripts studied except the loading control Gapd was detected in liver or kidney, while gene expression in B16-F1 melanoma cells was very similar to that in melanoblasts (including absence of P protein), except that Kit was not expressed. Kit is commonly down-regulated in advanced melanoma cells (reviewed in ref. 49).

Differentiation of NCLSCs to pre-Schwann cells

TGFß1 and the related family of bone morphogenetic proteins were reported to promote commitment of primary rat neural crest stem cells to autonomic neuronal lineages and smooth muscle-like cells in vitro (16, 51). We therefore tested the effect of TGFß1on NCLSCs. Simple substitution of FCS for NBCS had little effect on NC-m4 cell morphology in 1-2 wk (Fig. 3A). In medium with FCS, TPA, and TGFß1, however, most cells changed within a few days to a larger, flattened cell type somewhat like fibroblasts (Fig. 3B). There was little effect on cell proliferation (Fig. 3F). Immunostaining of these flat cells for the smooth muscle marker calponin proved negative, suggesting that they were not smooth myocytes. We then hypothesized that they might be related to Schwann (peripheral glial) cells.

Differentiation of glial cells in the peripheral nervous system and from cultured rat neural crest stem cells or glial precursors can be promoted by ß forms of neuregulin 1 (NRG1-ß, related to glial growth factor) (3, 11, 52, 53). NRG1 is expressed in developing nerves (3). We examined the effect of NRG1-ß1onthe NCLSC-derived flat cells. It did not noticeably change the morphology of these cells, but did increase their proliferation at 10 nM (Fig. 3C, F), indicating the presence of receptors for NRG1-ß1. A smaller effect of NRG1-ß1 at 1 nM was observed (P <0.01), and none at lower concentrations (P >0.1 for 0.4 nM). NRG1-ß1 (10 nM) also significantly increased numbers of stock NC-m4 NCLSCs (grown with TPA and NBCS) over 3 wk (P< 0.001), so these also possessed receptors.

Since most NCLSCs changed into the flatter cell type in a few days with TGFß1, we tested whether this was just a reversible effect of TGFß1 or was stable, like cell determination. NC-m4 cells were grown for 15 d with FCS, TPA (2 nM), and TGFß1, either with or without NRG1-ß 1. The cells were then returned for 7 d to medium with FCS and TPA (2 nM) only. As shown in Fig. 3D, E, the cells retained the larger, flatter morphology. The increased cell yield with NRG1-ß1 was maintained after the week in TPA only (Fig. 3F; P <0.001). The flat cells lost the responsiveness to gelatin seen in NCLSCs (P >0.1 for cell number) and no longer tended to form clumps and detach from the substrate (Fig. 3).

We now immunostained cultures of the NC-m4 flat cells for glial fibrillary acidic protein (GFAP) and GAP43 (growth-associated protein 43), markers expressed by immature and unmyelinated Schwann cells (GAP43 also in axonal growth cones) (11). TGFß1- and NRG1-ß1-treated cultures showed clear cytoplasmic expression of both markers (Fig. 3G, I)in ~90% of the cells, indicating that these were pre-Schwann cells. NCLSCs (Fig. 3H, J) and control melan-c albino melanocytes (Supplemental Fig. 1G, H) were unstained.

Cloned NC-m6 cells grown with TGFß1 and NRG1-ß1 showed a similar morphological response to that by NC-m4, indicating that the response was not due solely to selection or differentiation, but involved cell determination.

Differentiation of NCLSCs to chondrocytes

To determine whether NCLSCs can generate a mesenchymal cell type, we sought to induce differentiation into chondrocytes. BMP-2 was reported to promote commitment of adult neural crest-derived cells to chondrocytes in vitro (18, 21). We examined the effect of BMP-2 and ascorbic acid on NCLSCs in the presence of FCS. Both these factors increased cell proliferation (Fig. 3M).

We identified chondrocytes (~50% of all cells) in NC-m6 grown in the presence of FCS, TPA, FGF2, ascorbic acid, and BMP-2 for 14 d by Alcian blue staining (1% in 3% acetic acid, pH 2.6) (Fig. 3K). Undifferentiated NCLSCs and melan-c melanocytes showed only background staining (Fig. 3L and Supplemental Fig. 2C).

Spontaneous and induced differentiation of sensory neurons from NCLSCs

In pluripotent mouse mesenchymal stem cells, determination of some cells to several lineages (myoblasts, chondroblasts, preadipocytes) can be induced by growth with the DNA demethylating agent 5-azacytidine (54). We tested whether azacytidine would promote cell determination in NCLSCs. After culture of NC-m4 cells with FCS and 5-azacytidine, there was some cell death. Most surviving cells were unchanged, but we saw occasional single, very large, dendritic cells (Fig. 4A). Similar cells also sometimes appeared in very sparse NC-m4 cultures grown in their standard medium (Fig. 4B), so the effect of 5-azacytidine may have been nonspecific. These large, dendritic cells were immunoreactive for neurofilament 200 (a neuronal marker) and substance P (a specific marker of sensory neurons) (Supplemental Fig. 1D–F).

We then tested the effects on NC-m4 cells of a combination of 4 neurotrophic factors, each previously reported to promote determination and/or differentiation of sensory neurons: NGF, BDNF, NT3, and FGF2 (55–57). In 2 wk, this cocktail produced a substantial fraction of large cells with morphology suggestive of neurons, either bipolar or stellate, with very long, dendrite-like processes (Fig. 4C). The majority of these cells (~90%) showed cytoplasmic immunostaining for substance P, similar to that seen in control primary rat sensory neurons and absent from NC-m4 NCLSCs (Fig. 4D–F). They were also immunostained by mixed neurofilament antibodies (Fig. 4G), and a proportion of cells by either of 2 antibodies to transient receptor potential channel V1 (TRPV1; see next section) (Fig. 4H, I). Untreated NCLSCs were negative for these markers (e.g., Fig. 4J).

We then investigated aspects of sensory neuronal function. First, we tested for intracellular calcium responses mediated by TRP channels, ligand-gated ion channels that mediate thermosensation by peripheral sensory neurons (58). TRPV1 is activated by heat and by vanilloids such as capsaicin (59), and TRPA1 by cold and by pungent ligands including cinnamaldehyde (60). NC-m4 cells were treated with the 4 neurotrophic factors for 2 wk as above, loaded with calcium indicator Fluo-4 and fluorescence videos made during perfusion with capsaicin or cinnamaldehyde (separate cultures). Increased fluorescence was observed in some dendritic cells with either agonist, and not with vehicle only (Fig. 5A, D and Supplemental Videos 1 and 2). The high fluorescence was transient, and could be seen in some areas moving along dendrites (Supplemental Video 2), again with either agonist. Time courses of intracellular calcium concentration in individual cells are shown in Fig. 5B, E and resemble those seen in normal sensory neurons. About 5% of all cells showed this functional activity with capsaicin, and ~2% with cinnamaldehyde (Fig. 5C, F). No such calcium responses (0) were detected in untreated NCLSCs or in control melan-c melanocytes.

Whole-cell patch-clamp recordings were then used to test for voltage-dependent inward currents, in NCLSCs induced to differentiate either with neurotrophic factors or with TGFß1 and NRG1-ß1 (Fig. 5G–I). No such currents were observed in melan-a (0/3) or melan-m5 (0/10) melanocytes or untreated NC-m6 NCLSCs (0/ 12), and only small currents in a few untreated NC-m4 NCLSCs (5/29; 17.2%). Fast-inactivating inward currents blocked by tetrodotoxin (TTX), indicating TTX-sensitive sodium channels, were recorded only from NC-m6 cells grown with TGFß1 and NRG1-ß1 (27/28; 96.4%) (Fig. 5G, H). TTX-resistant inward currents were also detected in these same cultures (Fig. 5H). These were Na⁺ rather than Ca²⁺ currents, because, although TTX-resistant, they were not sustained. These findings are consistent with cell differentiation under these conditions toward either Schwann cells or neurons (see Discussion). Conversely, the majority of NC-m4 and NC-m6 cells treated with the 4 neurotrophic factors [72/79 (91.1%) and 6/8 (75%), respectively] showed TTX-resistant currents that were not inactivated during the 100-ms depolarization at negative test potentials (Fig. 5I). Similar currents were previously recorded in explanted sensory neurons (61) and are referred to as persistent, TTX-resistant currents: the third known class of voltage-gated Na⁺ current. They have a negative threshold for activation and a depolarized midpoint of inactivation (61). These currents were activated between about -80 and -70 mV (Fig. 5J), showing slow activation and prominent ultraslow inactivation (Fig. 5I). They appeared to be Na⁺ currents because of resistance to Cd²⁺ (a Ca²⁺ channel blocker) and fluoride (see Fig. 5 legend).