Edinburgh Research Explorer Regulation of boundary cap neural crest stem cell differentiation after transplantation

Success of cell replacement therapies for neurological disor- ders will depend largely on the optimization of strategies to enhance viability and control the developmental fate of stem cells after transplantation. Once transplanted, stem/progen- itor cells display a tendency to maintain an undifferentiated phenotype or differentiate into inappropriate cell types. Gain and loss of function experiments have revealed key transcription factors which drive differentiation of immature stem/progenitor cells toward more mature stages and eventually to full differentiation. An attractive course of action to promote survival and direct the differentiation of transplanted stem cells to a speciﬁc cell type would there-fore be to force expression of regulatory differentiation molecules in already transplanted stem cells, using inducible gene expression systems which can be controlled from the outside. Here, we explore this hypothesis by employing a tetracycline gene regulating system (Tet-On) to drive the dif- ferentiation of boundary cap neural crest stem cells (bNCSCs) toward a sensory neuron fate after transplantation. We induced the expression of the key transcription fac- tor Runx1 in Sox10-expressing bNCSCs. Forced expression of Runx1 strongly increased transplant survival in the enriched neurotrophic environment of the dorsal root ganglion cavity, and was sufﬁcient to guide differentiation of bNCSCs toward a nonpeptidergic nociceptive sensory neu- ron phenotype both in vitro and in vivo after transplantation. These ﬁndings suggest that exogenous activation of transcription factors expression after transplantation in stem/progenitor cell grafts can be a constructive approach to control their survival as well as their differentiation to the desired type of cell and that the Tet-system is a useful tool to achieve this.


INTRODUCTION
The use of stem/progenitor cells is an attractive avenue for cell replacement strategies in the treatment of neurological disorders and traumatic injuries. However, to generate neurons, and particularly of a specific desired type, from the transplanted stem cells still poses a significant challenge. The current state of the art in neural stem cell transplantation is to differentiate stem cells in vitro into the required cell type and to subsequently transplant an enriched population of these cells into the injured or diseased nervous system. However, this might not be the optimal protocol for every application, since the time window for a successful transplantation may be narrow for some types of stem/progenitor cells and their fate after transplantation remains unpredictable. As a result, the transplanted neural stem cells survive transplantation poorly or undergo inappropriate or incomplete differentiation [1][2][3]. An alternative approach to promote differentiation of stem cells would focus on initiating this process after transplantation. The aim of this study was to investigate if the exogenous induction of a key transcription factor(s) after transplantation could be effective in inducing subtype-specific differentiation of transplanted stem cells. Here, we explore this approach in a model system of boundary cap neural crest stem cell (bNCSC) transplantation to the dorsal root ganglion (DRG) cavity.
We previously developed a method of transplantation to the DRG cavity that allows the identification of transplanted cells by their specific location, and demonstrated that the transplanted sensory neurons established functional connections with the spinal cord [4][5][6]. When transplanted in this system, mouse forebrain neural stem/progenitor cells (NSPCs) predominantly differentiated into glial cells [7], while human NSPCs differentiated into neurons, although they remained immature for up to 3 months post-transplantation [8]. In the present study, we prepared bNCSCs for transplantation from the embryonic 11,5 DRGs, including the boundary cap (b), a transient neural crest derived structure giving rise to the last wave of DRG neurons in development [9][10][11]. The bNCSCs are multipotent stem cells that have been shown to generate sensory neurons in vitro [12,13] and glial cells in vivo after transplantation [14]. Here, we explore whether conditionally induced expression of the key transcription factor Runx1 in Sox10-expressing cells from transgenic mice can guide the differentiation of such peripherally transplanted bNCSCs toward a nociceptor neuron phenotype.
Differentiation of neural stem cells is controlled by the combined action of external signals and sequentially expressed transcription factors, with some of these representing master regulators [15,16]. We chose to control Runx1 using the Sox10 pattern of expression, since Runx1 had been suggested to play a key role in the initial stage of differentiation of NCSCs into neurons [11].
Sox10 or SRY-box containing gene 10, is a high mobility group transcription factor expressed in all neural crest (NC) cells and is involved in several aspects of NC development [17,18]. Importantly, it maintains NC cell multipotency [19] and hence, its downregulation is a prerequisite for neuronal differentiation [20,21]. We employed the tetracycline gene regulating system (Tet-On) [22] for exogenous induction of Runx1 overexpression in bNCSCs in vitro and after transplantation. This approach allowed us to use Sox10-rtTA mice [23] to specifically target Sox10-expressing cells, since only they will respond with TRE-Runx1 activation. We find that exogenous activation of Runx1 expression in Sox10 expressing bNCSCs promotes survival and induces specific differentiation toward nonpeptidergic nociceptive-type sensory neurons in vitro and after transplantation.

Animals and Genotyping
Sox10-rtTA2 S -M2 mice, which contain a second-generation reverse tetracycline-controlled transactivator (rtTA2 S -M2) knocked into the genomic Sox10 locus, have been previously described along with protocols for their genotyping [23]. Transgenic heterozygous C57BL/6-b-actin (CAG)-enhanced green fluorescent protein (EGFP) mice (Jackson Laboratories, Bar Harbor, ME, http://www.jax.org) were used as previously described [7]. As control to exclude any doxycycline (DOX)-mediated effects on bNCSC differentiation in the transplants, we used neurospheres from ROSA26-rtTA/HPRT-IRES-EGFP mice (generated in the Medvinsky laboratory) in which expression of EGFP was activated by Tet-On. ROSA26-rtTA/HPRT-IRES-EGFP mice were generated from Ainv15 embryonic stem (ES) cells [24] These cells contain the rtTA transgene targeted into the ROSA26 locus and a homing site for a single LoxP targeting site upstream of the HPRT locus. cDNA targeted into the homing site is driven by a Tet-dependent promoter. The ES cell line was generated by targeting IRES-EGFP. ES clones that showed high levels of EGFP expression on addition of DOX were selected for blastocyst injections.
For experiments aimed at evaluating postgrafting bNCSC survival and extension of the transplanted cells, embryos were obtained by breeding heterozygous EGFP male and heterozygous Sox10-rtTA2 S -M2 (Sox10 þ/rtTA ) female mice. Embryos with EGFP expression were identified using fluorescence microscopy and genotyped for the presence of Sox10-rtTA2 S -M2. All procedures were approved by the Regional Ethical Committee for Research on Animals and carried out according to the guidelines of the Society for Neuroscience.

Culture of bNCSCs
bNCSCs were isolated in a semiclonal fashion from Sox10 þ/rtTA or from CAG-EGFP:Sox10 þ/rtTA embryos on embryonic day (E) 11, as described previously [12]. Briefly, the DRGs along with boundary caps were mechanically separated from the isolated spinal cord and mechanoenzymatically dissociated using Collagenase/Dispase (1 mg/ml) and DNase (0.5 mg/ml) for 30 minutes at room temperature. Cells were plated at 0.5-1 Â 10 5 cells/cm 2 in N2 medium containing B27 (Gibco, Grand Island, NY, http:// www.invitrogen.com) as well as EGF and basic fibroblast growth factor (R&D Systems, Minneapolis, http://www.rndsystems.com; 20 ng/ml, respectively). After 12 hours, nonadherent cells were removed together with half of the medium before adding fresh medium. The medium was changed every other day (50% of the medium replaced with fresh medium) until neurospheres could be observed after approximately 2 weeks of culture. Nonpassaged neurospheres between 2 and 3 weeks in culture were used for subsequent experiments.

Nucleofection of DNA and Activation of eYFP/Runx1 Expression in Transfected Cells
We used the construct described in Figure 1D. Nucleofection of 3-weeks-old Sox10 þ/rtTA neurospheres was performed with an Amaxa Biosystems Nucleofector according to the manufacturer's instructions for the Optimal Nucleofector Program A33. The level of transfection and the efficacy of the Tet-On system were assessed by analyzing the EYFP expression in DOX (5 lg/ml)treated and DOX-untreated cultures. DOX was added to cultured neurospheres 24 hours after transfection and replaced every third day when the culture medium was changed. The EYFP fluorescence was monitored in an inverted fluorescence microscope. To verify the validity of EYFP expression as a measure of Runx1 upregulation, the expression of Runx1 and EYFP in DOX-treated and nontreated cultures was assessed with RT-PCR.

In Vitro Differentiation Assay
DOX-treated EYFP-expressing neurospheres were picked up by pipette under the fluorescence microscope, and plated at a density of 1.2 Â 10 3 cells on a poly-D-lysine (50 lg/ml)/laminin (20 ng/ ml)-coated coverslip and maintained in Dulbecco's modified Eagle's medium-F12/neurobasal medium supplemented with N2, B27, 0.1 mM nonessential amino acids and 2 mM sodium pyruvate. Nontreated neurospheres were used in parallel as control. The medium was replenished and DOX was added every third day. After 1 or 2 weeks, cells were fixed with 4% phosphate buffered paraformaldehyde. Each experiment was repeated at least six times.

Transplantation and Post-Grafting Treatment
Adult C3H mice (Mollegard, Denmark) were anesthetized by intraperitoneal injection of xylazine (RompunV R vet.; http:// www.bayer.com) and ketamine (KetaminolV R vet.; http://www. intervet.com) (10 and 100 ng per gram body weight, respectively). The transplantation procedure was described previously [4][5][6][7]. Briefly, the left L4 DRG was exposed via a partial laminectomy and carefully removed, leaving the ventral root intact. Four to five Sox10 þ/rtTA bNCSC neurospheres transfected with TRE bi -EYFP-Runx1 or the same amount of neurospheres produced from ROSA26-rtTA/HPRT-IRES-EGFP mice were placed into the empty DRG cavity to an approximate total number of 10,000 cells/ transplant, the dorsal root and the spinal nerve were attached to the collection of bNCSCs and the wound closed in layers.
All recipients (n ¼ 41) received daily intraperitoneal injections of cyclosporin (Sandimmun, Novartis International, Basel, Switzerland, http://www.novartis.com; 15 mg/kg body weight). The experimental groups (n ¼ 12) with Sox10 þ/rtTA :TRE bi -EYFP-Runx1 donor cells received DOX in their drinking water (3 g/l with 50 g/l sucrose) from the first day after surgery and through the entire post-transplantation period while the control group with TRE bi -EYFP-Runx1 donor cells (n ¼ 6) received pure drinking water.
In an additional experiment to evaluate transplant size and overall survival of transplanted bNCSCs, CAG-EGFP:Sox10 þ/rtTA bNCSCs ubiquitously expressing EGFP were used (n ¼ 6 treated; n ¼ 3 control) with the same treatment.
For bromodeoxyuridine (BrdU) labeling, six mice with neurospheres harboring Sox10 þ/rtTA :TRE bi -EYFP-Runx1 transgenes were injected with BrdU intraperitoneally once every day (10 mg/ml; 200 ll) from day 1 after transplantation and for a timeperiod of 2 weeks. Three of these mice were given DOX in their drinking water from the first post-transplantion day and the other three mice were given DOX with a delay of 1 week postgrafting.

Processing of Transplant Tissue
One month after transplantation, the animals were reanesthetized and perfused with phosphate-buffered picric acid-formalin. The graft sites, together with the attached dorsal root and spinal nerve, as well as the corresponding part of the spinal cord were removed, placed in cold fixative for 4 hours, and cryoprotected overnight in phosphate-buffered saline (PBS) containing 15% sucrose. Tissue was sectioned using a cryostat at 9 lm thickness and stored at À20 C for later immunohistochemical processing.

Microscopic Analysis
Immunolabeled cultures and sections were analyzed in a Nikon Eclipse E800 fluorescence microscope. For photography, a Nikon DXM1200F digital camera system was used. The proportion of neuronal cells expressing subtype-specific sensory markers was estimated in the following combinations: bTUB/IB4; peripherin/ CGRP, bTUB/RET. Cells were counted under the 20Â objective with the aid of a square ocular frame (side 0.11 mm). Neurite outgrowth in the in vitro differentiation assay was analyzed using a computer-based procedure [29]. Briefly, 10 photographs (20Â objective) were taken of bTUB-labeled cultures (n ¼ 6 treated; n ¼ 6 control) and placed in a grid frame on the computer screen. The number of intersections between neurites and grid lines in relation to the number of cells within the frame was analyzed and used as a measure of neurite outgrowth/cell. Every sixth slide from transplants with Sox10 þ/rtTA :TRE bi -EYFP-Runx1 and ROSA26-rtTA/HPRT-IRES-EGFP neurospheres was immunostained and cells containing nuclei were analyzed under the 20Â objective with the aid of a square ocular frame (side 0.11 mm). For the combination IB4/CGRP/RT97 the num-ber of cell bodies labeled with either one of these markers was counted. For evaluation of transplant size and bNCSC survival, every fifth section from EGFP transplants was photographed. The NIH software ImageJ (Rasband, 1997, available at http://rsb.info. nih.gov/ij) was used to measure transplant areas. The transplant volume estimate was calculated according to the formula A ¼ TK[ P (S 1 to S n )], where T is the thickness of the section (T ¼ 9 lm), K is the number of sections between the measured areas (K ¼ 5) and S is the area of the transplant on the sections from 1 to N. To generate three-dimensional images of transplants, serial sections from transplants with eGFP expressing donor cells were photographed and digitalized (see earlier), manually aligned in Photoshop, and processed with the VolView 2.0 software (KitWare, Clifton Park, NY).
For an estimate of the cell numbers in each transplant, EGFPþ cells with nuclei were counted in every fourth section. To correct for possible differences in nuclear size in different transplants, the average nuclear diameter was analyzed by measuring 30 randomly selected nuclei. The number of neurons counted was multiplied by section separation to give a total estimated number of profiles (n). This number was multiplied by section thickness (T), divided by T plus the average diameter of the nuclei (D) to give the neuronal number (N); n ¼ n Â T/(T þ D) [30,31].

Statistics
Data were analyzed using a two-tailed unpaired Student's t test.

Sox10 Expression in bNCSCs In Vitro
An overview of the experimental setup is shown in Figure 1. The bNCSC neurospheres were produced from Sox10 þ/rtTA mouse embryos. Consistent with previous results [20], immunohistochemical analysis showed that Sox10 was expressed in E11 DRGs of Sox10 þ/rtTA mice (Fig. 1B). Sox10 expression was also found in newly formed bNCSC neurospheres (Fig.  1C) and the expression with regards to passage and days in culture was assayed by RT-PCR (supporting information Fig.  1). As seen by RT-PCR Sox10 was initially highly expressed, but this expression declined after 4 to 5 weeks in culture. Thus for the experiments of this study we did not use bNCSCs older than 3 weeks. To evaluate the expression of transgenic Runx1 in DOX-activated neurospheres, we performed RT-PCR and confirmed that the expression of Runx1 in DOX-activated neurospheres paralleled the time course of EYFP fluorescence (supporting information Fig. 2).

Differentiation of Sox10 1/rtTA :TRE Bi -Runx1-EYFP bNCSCs In Vitro
We next assessed how forced Runx1 expression influences the differentiation of bNCSCs in vitro. We first analyzed the level of EYFP expression in transfected bNCSC neurospheres to determine the level of transfection and DOX-induced transcription. Sox10 þ/rtTA neurospheres were transfected with the TRE bi -Runx1-EYFP expression vector (Fig. 1D). The cells were divided in two groups and DOX was added to the medium of one of these groups 24 hours after transfection. The expression of Runx1 in DOX-treated and control neurospheres was monitored by the concomitant induction of EYFP expression from the bidirectional TRE promoter (Fig. 1E, 1F). EYFP fluorescence was observed in around 60% of the DOXtreated neurospheres, after 24-48 hours of DOX administration (supporting information Fig. 2). The percentage of EYFP-expressing cells in individual neurospheres varied between approximately 10% and 90%. In DOX-untreated Aldskogius, Berens, Kanaykina et al.
neurospheres, the expression of EYFP was extremely low and detected only in about 0.3%-1% of the total number of cells. EYFP/Runx1 expression in DOX-treated neurospheres gradually declined in intensity and disappeared after 2-3 weeks in culture which is in agreement with the cessation of Sox10 expression (data not shown).
Neurospheres were plated on D-polylysine-laminin coated coverslips 2 days after transfection in a medium without mitogens to allow for differentiation. No neurotrophic factors were added to decrease the risk of a Runx1-mediated effect being masked by differentiation directed by extrinsic factors. DOX was added in the treated group throughout the experiment and EYFP expression was monitored. The EYFP expression rapidly declined in differentiating cells (supporting information Fig. 3). However, in a few cells it was retained for up to 2 weeks, possibly due to rare and random stable integration events leading to poor inactivation of the Tet-promoter once activated. This time frame of 2 weeks allowed us to visualize the gradual differentiation of EYFP-expressing bNCSCs to cells with a typical DRG neuron morphology (supporting information Fig. 4).
Cultures were fixed at the end of the experiment (2 weeks) and labeled with markers for sensory neuron subtypes, with particular focus on DRG neurons previously reported to be Runx1-dependent. During early development, Runx1 is expressed in the largest subpopulation of DRG neurons, the small diameter nociceptive class [26,32]. In addition, Runx1 has been reported to induce neurite outgrowth from DRG neurons [32]. In agreement with these findings, we found that DOX-induced Runx1 expression increased neurite outgrowth in bTUBþ cells in culture ( Fig. 2A, 2B; quantified in c; p ¼ 3.6 Â 10 À8 ). Looking at the subtype markers, we found abundant induction of the markers IB4 and glial cell line-derived neurotrophic factor (GDNF) family coreceptor RET in the DOX-treated cultures, while they were completely absent in the control group (Fig. 2D-2G; in (e) p ¼ 3.6 Â 10 À6 and (g) p ¼ 1.8 Â 10 À4 ). These markers are characteristic for nonpeptidergic nociceptor DRG neurons. In contrast, CGRP, a marker for small peptidergic nociceptive neurons, showed no obvious difference between DOX-treated and control cultures (Fig. 2H, 2I; p ¼ .21). These data suggest that ectopic expression of Runx1 is sufficient to drive differentiation of bNCSCs toward nonpeptidergic DRG neurons in the absence of neurotrophic factors in vitro.

Differentiation of Sox10 1/rtTA :TRE Bi -Runx1-EYFP bNCSCs After Transplantation
After determining that forced overexpression of Runx1 in Sox10-expressing bNCSCs in vitro can induce differentiation toward the nonpeptidergic nociceptor phenotype, we examined whether this could also influence their differentiation in vivo after transplantation. To this end, we used TRE bi -Runx1-EYFP transfected neurospheres, taken from cultures in which DOX-induced EYFP expression had been confirmed in samples (the grafted cells however, had not been treated with DOX prior to transplantation), in a transplantation experiment. The left L4 DRG was removed and the transfected bNCSCs were grafted in its place with nerve root endings being reattached to the graft. One group of graft recipients received DOX in their drinking water whereas the control group received tap water. One month later, the L4 cavity containing the grafted cells was harvested for further investigation.
We first estimated the relative extent of neuron versus glial differentiation in DOX-treated and control transplants, using immunohistochemistry for bTUB and GFAP in combination with nuclear counterstaining (Fig. 3A-F). The proportion of bTUBþ cells was higher in DOX-treated transplants (although not significantly, p ¼ .08) whereas that of GFAPþ cells was significantly higher compared with DOX-untreated transplants (p ¼ 4.7 Â 10 À4 ). In DOX-untreated transplants, in contrast, the proportion of bTUB/GFAP negative cells was greater (Fig. 3I, p ¼ 1.6 Â 10 À5 ).
To assess the influence of DOX treatment on neurite outgrowth from the transplanted cells, we analyzed transplants, dorsal roots associated with the transplanted tissue, and corresponding areas (L4 and L5) in the spinal cord of DOX-treated and DOX-untreated EGFP/Sox10-rtTA2 S -M2 (Sox10þ/rtTA) mice DOX-treated Runx1-induced EGFPþ cells displayed a more mature morphology than DOX-untreated cells (Fig. 3G,  3H) and showed extensive neurite outgrowth into the host dorsal roots (Fig. 3I, 3J). However, neurites extending from DOX-treated transplants did not fully reach the host spinal cord.
We next characterized the neuronal phenotypes in transplants from DOX-treated and nontreated recipients, using sensory neuron subtype-specific markers. Transplants from DOXtreated mice were characterized by an increase in IB4-labeled (nonpeptidergic) cells (Fig. 3G, 3H, 3J; p ¼ .005). The percentage of cells immunoreactive for RT97 (antibody labeling the phosphorylated form of neurofilament heavy-chain specific for mechanosensitive cells) was significantly decreased (Fig.  3J; p ¼ .002). The percentage of CGRPþ cells also showed a decreasing trend, but was not shown to be significant ( Fig. 3J; p ¼ .31). We also investigated the expression of TrkA and RET in the bTUBþ cells in transplants from DOX-treated and control mice (Fig. 4). The amount of cells expressing both markers was significantly increased in DOX-treated transplants (Fig. 4 C and 4F; p ¼ .003 and p ¼ 3.4 Â 10 À7 , respectively). Extensive expression of RET in DOX-treated transplants was confirmed by in situ hybridization for RET mRNA (Fig. 4D). However, while TrkA was mostly expressed in non-neuronal bTUB-negative cells (Fig. 4A, 4B) the majority of RET-expressing cells were bTUBþ (Fig. 4D,  4E). In addition, we assayed for the presence of the ATP receptor P2X3 which is expressed in a subset of the non-peptidergic nociceptive neurons [33] that expresses Mas-related G protein-coupled receptor D and peripherin and innervates the superficial layers of the skin [34]. P2X3 was increased in the neuronal (bTUBþ) population by the induction of Runx1 expression ( Fig. 5A-C; p ¼ 1.3 Â 10 À4 ). The relative proportion of P2X3-expressing cells was smaller than the fraction of RET-expressing cells suggesting that, similar to the normal DRG, only a subset of the RETþ cells express P2X3 (Figs. 4F, 5C). These data suggest that using the DOX-inducible system to overexpress Runx1 in Sox10þ bNCSCs increases the differentiation into a non-peptidergic nociceptive phenotype, possibly at the expense of other sensory neuronal cell types. After BrdU pulse labeling in the early post-transplantation period, some RET-expressing neurons were also BrdUþ indicating that this population indeed originated from proliferating bNCSCs (Fig. 4D).
To asses the effect of DOX treatment on transplant survival and differentiation, we transplanted nontransfected CAG-EGFP/Sox10-rtTA, Sox10-rtTA, and ROSA26-rtTA/ HPRT-IRES-EGFP bNCSC neurospheres. We found no difference in survival and differentiation between DOX-treated nontransfected transplants and ROSA26-rtTA/HPRT-IRES-EGFP treated with DOX compared with the DOX-untreated Runx1 transfected group (supporting information Fig. 8) Particularly striking was the complete absence of RET (supporting information Fig. 8) and P2X3 expressing cells in the control groups.
Survival of CAG-EGFP:Sox10 1/rtTA : TRE Bi -Runx1-EYFP bNCSCs After Transplantation To evaluate transplant size and overall survival of transplanted bNCSCs as well as to rule out any contribution of endogenous cells to sensory neurogenesis, we generated bNCSCs from CAG-EGFP:Sox10 þ/rtTA mouse embryos (ubiquitously expressing EGFP in all cells), transfected them with the TRE bi -Runx1-EYFP construct and transplanted them to the L4 DRG cavity as described earlier. The mean volume of DOX-treated transplants was 2.3 times larger than control (Fig. 6A, 6B; p ¼ 5.5 Â 10 À4 ) and the total number of EGFP expressing cells in DOX-treated transplants was increased by 204% compared with control transplants (Fig. 6C; p ¼ 1.4 Â 10 À4 ).

DISCUSSION
In this study, we have shown that differentiation of bNCSCs towards subtype-specific sensory neuron phenotypes can be guided in vitro and after their engraftment into the DRG cavity with the Tet-On system.  For producing bNCSCs, we used DRGs and dorsal roots of 11.5-day old mouse embryos. At this stage, contaminating DRG progenitor cells are still present in the DRGs, but they cease to proliferate around E12 [35,36]. The boundary cap cells proliferate throughout the entire embryogenesis [37], migrate to the DRGs and contribute to some trunk sensory neurons, satellite cells and Schwann cells [10,12] It was shown previously that a rapidly amplifying population that could not be sub-cloned, (i.e., the non-stem cell DRG progenitor cells mentioned above) is lost at around 6-8 days of culture of boundary cap cells whereas after about 2 weeks of culture a more slowly dividing cell population appears that forms neurospheres and can be repeatedly subcloned without affecting its pluripotency [12]. Furthermore, it was shown that the DRG itself does not significantly contribute to neurosphere formation and that the bNCSCs (as well as the actual boundary cap cells in vivo) express Krox20, Sox10, and the multipotency marker SSEA-1, which is not the case for surrounding cell types [12,13]. The BrdU labeling of neurons expressing the nociceptor-specific marker RET shows the presence of stem/progenitor boundary cap cells that continue to divide after transplantation. As the experiments in this study were performed with cells isolated in semiclonal condi-tions, however, it cannot be entirely excluded that a minority of cells in the bNCSC-population are indeed nonproliferative committed progenitors and that a small part of the effect might be attributed to selection in addition to the instructive role of Runx1.
In contrast to previously transplanted neurospheres derived from mouse [7] and human [8] forebrain, which did not express any specific neuronal markers after transplantation, in the experiments described here, we achieved  differentiation of transplanted cells into neurons with extensive neurite outgrowth into the host dorsal roots and the expression of neuronal markers specific for the nociceptor neuron phenotype. Furthermore, DOX-treated recipients displayed increased graft size and cell survival. In vivo, a combination of intrinsic and extrinsic factors influences survival and differentiation of transplanted cells. Survival and differentiation of TrkA-expressing cells are dependent on nerve growth factor (NGF), whereas RET-expressing cells rely instead on members of the GDNF family [38,39]. Differentiating immature sensory neurons may provide paracrine neurotrophic support in the transplant [40]. Non-neuronal cells, particularly from the host, are likely to be a major source of growth promoting factors, including NGF and GDNF [41,42]. In collaboration with the Runx1-induced ectopic expression of the growth factor receptors TrkA and RET, these extrinsic factors may significantly contribute to survival and possibly also to differentiation of the DOX-treated, transplanted bNCSCs. Furthermore, recent studies have demonstrated that NGF promotes differentiation and maturation of nonpeptidergic neurons through both RET-dependent and RET-independent signaling mechanisms, possibly via Runx1 expression [39]. Specifically, we could observe upregulation of P2X3 in the transplants arguing that RET-independent markers are induced via the Runx1 pathway. This, in conjunction with the strong upregulation of RET would suggest that both types of nonpeptidergic genes are controlled by the Runx1 overexpression.
Compared with control transplants, we observed a significant increase in the number of TrkAþ cells as well as a reduction in the numbers of cells not expressing bTUB or GFAP in DOX-treated transplants, suggesting that Runx1 overexpression might be sufficient to drive initial differentiation. However, in the absence of a robust increase in the proportion of bTUBþ cells, our data suggest that this differentiation of grafted bNCSCs is to a large part glial. This is in accordance with previous in vitro data on Runx1 overexpression in bNCSCs promoting survival and neuronal maturation but not neurogenesis per se [32]. The same study showed that Runx1-induced cell survival of bNCSC-derived neurons is NGF-independent arguing for a more general effect of Runx1 on cell survival. However, since the grafted cells differentiate into more than one cell type it is not possible to discern between cell-autonomous effects of Runx1 and those involving signaling between the various differentiating cells in the transplant or with host cells. It is also possible that the increased amount of GFAPþ cells in the Runx1overexpressing grafts may contribute to maintaining survival of neighboring neuronal precursors, reminiscent of the relationship between satellite cells and neurons in intact DRGs [43].
In the chick, it has been suggested that Runx1 directly activates TrkA transcription and that Runx1 overexpression in progenitors in vivo induces ectopic expression of TrkA [32]. In the same study, siRNA knock-down of early Runx1 expression as well as overexpression of a general dominant-negative Runx protein was shown to down-regulate TrkA expression and cause the subsequent death of the neurons. This suggests that Runx1 might have an effect on early progenitors by inducing TrkA expression in DRG precursors. There seems to be conflicting data with regards to an early role of Runx1 in the development of nociceptive sensory neurons since there is no loss of TrkAþ neurons in mice that lack Runx1 expression in DRG cells [26,44]. Our findings indicate that ectopic Runx1 expression indirectly influences the survival of bNCSCs grafted into the DRG cavity.
The role of Runx1 in later stages of differentiation of DRG neurons has been convincingly demonstrated. Mice that lack Runx1 function selectively in the peripheral nervous system have an impaired perinatal switch from TrkA to RET of non-peptidergic nociceptors [45]. Indeed, in the mouse, from E14.5 to parturition, Runx1 expression is restricted to TrkAþ neurons, but by postnatal day 30, Runx1þ neurons are TrkAnegative and RETþ [26]. These two studies also show that Runx1 controls, via both repressing and activating gene expression, several aspects of maturation of nonpeptidergic neurons such as the concerted expression of specific ion channels and receptors. This includes the expression of the ATPreceptor P2X3. Indeed, after forced Runx1 expression, we observed a selective up-regulation of RET, IB4-binding, and P2X3 in differentiated bTUBþ neurons, as well as a decrease of RT97 and CGRP expression in the transplants.
The possibility that increased survival and differentiation to the specific type of neurons in the transplants was rather due to the DOX-treatment or to the electroporation procedure itself [46] was ruled out by the fact that none of the control transplants displayed any presence of nociceptor-specific markers. The latter appeared exclusively in Sox10-expressing cells that were transfected with an inducible Runx1 construct and treated with DOX. In agreement with this more mature and differentiated phenotype, DOX-treated Runx1-induced EGFPþ transplants displayed extensive neurite outgrowth to the dorsal roots towards the spinal cord.
Our data show that exogenously induced Runx1 expression does not affect the relative neuron/glia proportion in the transplants. There was a trend towards neuronal differentiation in DOX-treated transplants but the difference was not significant compared to DOX-untreated recipients. This finding suggests that the decision of neural progenitors to become neuronal or glial is regulated by other transcription factors in bNCSCs when they still express Sox10. The large proportion of RETþ cells is a bit surprising but might be due to a combination of RET being directly downstream of Runx1 [39] and that there is positive selection for bNCSCs that have been transfected and express Runx1.
The Sox10þ/rtTA/TRE-Runx1 system used here appears to be sufficient to ensure advanced cell differentiation towards nociceptor subtype in vitro and in the transplants. The mechanism for this might be due to the specific targeting of Sox10expressing cells for Runx1 overexpression. It was achieved by a new approach using the Tet-On system. The separation of the rtTA2 S -M2 and TRE expressing vectors might serve as a useful tool to specifically target certain cell types for expression of the desired key transcription factor. It also gave us the possibility to mimic the normal cellular development pattern, in which the expression of the transcription factor Sox10 precedes that of Runx1, a temporal sequence which seems to be crucial for the guiding the differentiation of bNCSCs toward nociceptive-type sensory neurons.

CONCLUSION
Our results demonstrate that using the Tet-system to exogenously control Runx1 expression in grafted bNCSCs resulted in their long-term survival, increased neurite outgrowth, and selective differentiation of the neuronal population toward a nonpeptidergic nociceptive neuronal phenotype. These results are achieved by cooperation between Runx1-induced intrinsic factors and extrinsic factors from the host environment. Our data suggest that the Tet-system can be successfully used in vivo to manipulate gene activation in transplanted cells. Hence, it is reasonable to speculate that timely activation to appropriate expression levels of just a few key transcription Aldskogius, Berens, Kanaykina et al.
factors, specific for a certain type of stem/progenitor cell, would be sufficient to control survival and to guide the differentiation of transplanted cells to a desired cell type. Combining different Tet-system transactivators could allow the inducible expression of two genes [47] or, alternatively, the Tet-system could be used in combination with other conditional gene expression systems in mice [48,49]. Concomitant or sequential expression of one or more key proteins can thus be implemented for directed differentiation of stem/progenitor cells, thus offering exciting new possibilities in understanding development and for treating diseases.