Author contributions: J.K.: conception and design, provision of study material, collection and/or assembly of data, data analysis and interpretation, manuscript writing; C.J.L.: provision of study material, data analysis and interpretation, manuscript writing; O.K.: provision of study material, collection and/or assembly of data; J.H., J.P.C., and Q.G.: provision of study materials, collection and/or assembly of data; M.A.L.: provision of study materials, collection and/or assembly of data conception and design, manuscript writing; S.W. and E.J.S.: provision of study materials, collection and/or assembly of data conception and design; D.A.F.: provision of study materials, manuscript writing; D.F. and M.D.: provision of study materials; R.J.: conception and design, data analysis and interpretation, manuscript writing.
Disclosure of potential conflicts of interest is found at the end of this article.
First published online in STEM CELLSEXPRESS April 19, 2011
Pluripotent cells can be derived from different types of somatic cells by nuclear reprogramming through the ectopic expression of four transcription factors, Oct3/4, Sox2, Klf4, and c-Myc. However, it is unclear whether postmitotic neurons are susceptible to direct reprogramming. Here, we show that postnatal cortical neurons, the vast majority of which are postmitotic, are amenable to epigenetic reprogramming. However, ectopic expression of the four canonical reprogramming factors is not sufficient to reprogram postnatal neurons. Efficient reprogramming was only achieved after forced cell proliferation by p53 suppression. Additionally, overexpression of repressor element-1 silencing transcription, a suppressor of neuronal gene activity, increased reprogramming efficiencies in combination with the reprogramming factors. Our findings indicate that terminally differentiated postnatal neurons are able to acquire the pluripotent state by direct epigenetic reprogramming, and this process is made more efficient through the suppression of lineage specific gene expression. STEM CELLS 2011;29:992–1000
The return of a somatic epigenome back to an embryonic-like pluripotent state can be achieved through the ectopic expression of the four transcription factors Oct4, Klf4, Sox2, and c-Myc (O,K,S,M), resulting in the formation of induced pluripotent stem cells (iPSCs) [1–3]. The in vitro reprogramming of somatic cells into iPSCs has been achieved in a variety of terminally differentiated somatic cell types including pancreatic β cells , keratinocytes , hepatocytes , B cells, and T cells [7, 8].
While mounting evidence demonstrates that a diverse array of somatic cells can be reprogrammed, the generation of iPSCs from terminally differentiated postmitotic cells has not been thoroughly studied due to the heterogeneous donor cell populations and ambiguous genetic markers for postmitotic cells. Nuclear reprogramming of postmitotic cells provides critical evidence for the equivalence of genomes from terminally differentiated and embryonic stem cells (ESCs) suggesting that the lack of developmental plasticity inherent in terminally differentiated cells is a result of their epigenetic states. In previous studies, the question of whether the epigenome of terminally differentiated postmitotic olfactory neurons could be reprogrammed to pluripotency has been investigated by somatic cell nuclear transfer (SCNT) experiments [9, 10]. However, as nuclear reprogramming by SCNT is a rapid process in which cell division is known not to be required, the underlying molecular mechanism as to how postmitotic cells reestablish the pluripotent state and why they are less efficient for nuclear reprogramming remains unclear .
Mature neurons epitomize a terminally differentiated somatic cell type. Neurons exit the cell cycle, do not proliferate, and are not replaced by precursors in the event of their loss. Although recent studies have described neuronal regeneration from precursor cells located in the subventricular zone (SVZ) , this remains a localized phenomenon and not a general characteristic of neurons. Cell cycle regulators and proneural genes are thought to be unnecessary in terminally differentiated noncycling mature neurons . To genetically identify terminally differentiated neurons, we isolated postnatal day 7 (P7) cortical neurons genetically marked by α-calcium-calmodulin-dependent kinase II promoter-driven Cre (CamKII-Cre-159)-mediated activation of a LoxP-Stop-LoxP (LSL)-enhanced green fluorescent protein (eGFP) reporter allele whose activity is restricted to terminally differentiated, mature neurons .
Here, we assessed the specificity of Cre-mediated recombination in cortical neurons cultured ex vivo and observed that the vast majority of these genetically marked neurons are postmitotic. We observed that the ectopic expression of the four canonical reprogramming factors is not sufficient to generate iPSCs, to induce significant cell proliferation, or to suppress neuronal gene expression from eGFP+ cortical neurons. We demonstrate that accelerated cell proliferation by suppression of p53 activity coupled with transduction of the reprogramming factors results in efficient iPSC formation from eGFP+ postnatal neurons. Furthermore, we find that suppression of neuronal gene activity by overexpression of the repressor element-1 silencing transcription (REST)/neuron-restrictive silencer factor (NRSF) increases reprogramming efficiency.
MATERIALS AND METHODS
Tail tip fibroblasts (TTFs) used to derive primary iPSC lines were harvested from crosses between mice homozygous for a CamKII-Cre-159 allele  and mice carrying LSL-eGFP reporter allele. ESCs and established iPSCs were cultured on irradiated mouse embryonic fibroblasts (MEFs) in Dulbecco's modified Eagle medium containing 15% fetal bovine serum, leukemia inhibiting factor, penicillin/streptomycin, L-glutamine, β-mercaptoethanol, and nonessential amino acids. For the primary postnatal neuronal culture, the highly contributed chimera mice at P7 were anesthetized, forebrains were removed, and the cortical plates (CPs) from the cerebral cortex were dissected and dissociated with Papain Dissociation System (Worthington Biochem, Lakewood, NJ, www.worthington-biochem.com) according to the manufacture's protocol. The dissociated neuronal cells were then plated onto poly-D-lysine/laminin coated plates and cultured in neuronal culture media. To examine the genomic integrity of neurons, we analyzed metaphase spreads of neuron-derived iPSCs for karyotypic abnormalities. Cells were arrested in metaphase with colcemid (0.5 μg/ml) for 3 hours and metaphase spreads were prepared and stained with Geimsa. Analysis of 20 spreads from neuron-derived iPSC lines showed no evidence of chromosome number alterations, chromosome breaks, and chromosome gaps. Based on this analysis, we conclude that both the neurons and neuron-derived iPSCs are karyotypically normal (Supporting Information Fig. 4B).
Lentiviral preparation and infection with doxycycline (dox)-inducible lentiviruses encoding a polycistronic O,K,S,M expression cassette (pHAGE2-tetOminiCMV-STEMCCA)  or a constitutive reverse tetracycline transactivator (flap-Ub promoter-WRE (FUW)–M2rtTA) were previously described . A p53-knockdown lentiviral vector encoding a specific hairpin for the mouse p53 tumor suppressor gene described previously  and the mouse REST cDNA were cloned into the EcoRI cloning site of dox-inducible FUW lentiviral vectors. For lentiviral vector infections of primary fibroblasts, cells were seeded in six-well plates at a density of 1 × 105 cells per well and infected on 3 consecutive days. Medium changes were performed 12–24 h after infection. One day after the last infection, ESC medium containing 2 μg/ml dox was added. Fresh ESC-medium with dox was added every other day until iPSC colonies developed. iPSC colonies were picked, trypsinized, and grown on MEFs in standard ESC conditions.
For 5-bromodeoxyuridine (BrdU) analysis, we used a Click-iT EdU flow cytometry assay kit (Invitrogen, Carlsbad, CA, www.invitrogen.com). Cells were resuspended in fluorescence-activated cell sorting (FACS) buffer for analysis on a FACS LSR cell sorter. For live green fluorescent protein sorting, cells were sorted on a FACS Aria cell sorter
Injections of iPSCs into B6 X DBA F2 host blastocysts were carried out as described previously .
Alkaline phosphatase (AP) staining was performed using an AP substrate kit (Vector Laboratories, Burlingame, CA, www.vectorlabs.com) according to manufacturer's manual. For the number of AP+ colonies, equal numbers of cells were plated in the absence or presence of dox on 100 mm dishes coated with gelatin. The number of dox-independent iPSC colonies that grew after the withdrawal of dox was determined. iPSC pluripotency was further validated by immunofluorescence staining for Nanog.
iPSCs were collected and separated from feeders by sedimentation of iPSC aggregates. Cells were washed, resuspended in 500 μl mouse ESC medium, and injected subcutaneously into severe combined immunodeficiency (SCID) mice (Taconic, Hudson, NY, www.taconic.com). Four weeks after injection, tumors were removed from euthanized mice and fixed in formalin. Samples were paraffin imbedded, sectioned, and analyzed on the basis of H&E stainings.
Bisulfite reactions were carried out according to the manufacturer's instructions (Epitect Bisulfite Kit, Qiagen, Valencia, CA, www.qiagen.com). Bisulfite (2–4 μl) treated DNA was used in a standard polymerase chain reaction (PCR) protocol to amplify Oct4 and Nanog promoter regions in mouse V6.5 ESCs, neurons, and iPSCs. PCR products were cloned into pCR2.1 vectors (Invitrogen) and sequenced using the M13 forward primer. Primers used to amplify the Oct4 and Nanog promoter region were described previously .
Cells were fixed in 4% paraformaldehyde in phosphate-buffered saline and immunostained according to standard protocols using the following primary antibodies: SSEA1 (Developmental Studies Hybridoma Bank, Iowa City, Iowa, www.dshb.biology.uiowa.edu), Sox2 (R&D Systems, Minneapolis, MN, www.rndsystems.com), Oct3/4 (Santa Cruz Biotechnology, Santa Cruz, CA, www.scbt.com), mouse Nanog (polyclonal rabbit; Bethyl Lab, Montgomery, TX, www.bethyl.com), NeuN (Chemicon, Billerica, MA, www.chemicon.com), eGFP (Abcam, Cambridge, MA, www.abcam.com), and appropriate fluorescent secondary antibodies (Invitrogen).
Southern Blot Analysis
Genomic DNA of 15 μg was digested with the indicated restriction enzymes. Electrophoresis and transfer was performed according to standard procedures. The blots were hybridized to radioactively labeled probes against c-Myc, Klf4, and REST. V6.5 ESCs were used as negative controls for determining background and endogenous bands.
Quantitative Reverse Transcriptase–PCR (RT-PCR)
Total RNA was isolated using an Rneasy Kit (QIAGEN). One microgram of DNase treated RNA was reverse transcribed using a First Strand Synthesis kit (Invitrogen). Quantitative RT-PCR analysis was performed in triplicate using 1/50 of the reverse transcription reaction in an ABI Prism 7000 (Applied Biosystems, Foster City, CA, www.appliedbiosystems.com) with Platinum SYBR green qPCR SuperMix-UDG with ROX (Invitrogen). Gene expression analysis for ESCs and neuronal markers was performed by RT-PCR. PCR primer sequences are available on request.
Results are given as mean ± SEM. Where appropriate, statistical analysis was performed with an analysis of variance test. The null hypothesis was rejected at the P < .05 level.
During murine brain development, neuronal precursors in the cerebral cortex are generated in a proliferative area called the SVZ. After their final cell division in the SVZ, postmitotic neurons migrate to the CP and form unique laminar structures. Postmitotic cortical neurons are located in the layer of the CPs in the postnatal mouse brain. To examine the possibility that postnatal neurons can undergo direct epigenetic reprogramming, we generated mice that harbor both a LSL-eGFP reporter allele and a CamKII-Cre allele (Fig. 1A). CamKII promoter activity is restricted to neurons with synaptic function leading to deletion of the LSL cassette and specific expression of eGFP in mature neurons of the adult forebrain and hippocampus . This system allows for the selective purification of mature postnatal neurons and for the retrospective confirmation that the cell of origin of any resulting iPSC clones was in fact a mature neuron.
Initially, we attempted to prepare primary cortical neurons from 2 or 4 weeks old murine brains but found that neurons cannot be efficiently cultured from animals of this age. Therefore, we isolated primary cortical neurons from P7 mice, which is the latest time point at which efficient ex vivo neuronal culture is possible, and examined their expression of eGFP. Figure 1B shows that most eGFP expression observed in the P7 mouse forebrain was restricted to the CP, where postmitotic neurons reside after migration from the SVZ. We also confirmed that the vast majority of eGFP marked cells in the CP are cortical neurons by costaining with an antibody specific to NeuN, a marker protein restricted to postmitotic neurons  (Fig. 1B). Importantly, no eGFP+ cells were observed in the SVZ, where mitotic neural progenitors reside (data not shown). The differentiated state of eGFP+ cortical neurons in primary ex vivo cultures was confirmed by NeuN costaining (Fig. 1C).
Because of the low infection efficiency of postnatal neurons and the heterogeneity of cell types in the primary neuronal cultures, we used a secondary reprogramming system to examine the ability of terminally differentiated postnatal neurons to undergo epigenetic reprogramming by the ectopic expression of the O,K,S,M reprogramming factors (Fig. 1D). In this system, somatic fibroblasts are initially transduced with dox-inducible lentiviral vectors encoding the O,K,S,M reprogramming factors to induce pluripotency [7, 19]. The resulting primary iPSCs become independent of ectopic O,K,S,M expression and on withdrawal of dox can generate chimeras after blastocyst injection. Such chimeras contain populations of “secondary” somatic cells carrying the identical proviral insertions that generated the primary iPSCs. It has previously been shown that subsequent treatment of such secondary somatic cells with dox can reactivate the reprogramming factors resulting in formation of secondary iPSCs with increased efficiencies .
We transduced CamKII-Cre/LSL-eGFP mouse TTFs with the polycistronic dox-inducible lentiviral vector harboring the O,K,S,M reprogramming factors [15, 20] along with a lentiviral vector constitutively expressing the M2rtTA. After infection, dox was introduced into the culture medium and 3 weeks later iPSC clones were isolated. These primary iPSCs were eGFP negative, consistent with the exclusion of postnatal neurons from the fibroblast preparation (Fig. 2A). Ten iPSC colonies were expanded in the absence of dox, indicating that these cultures had reactivated their endogenous pluripotency regulatory networks, consistent with bisulfite sequence analysis revealing complete demethylation of the endogenous Oct4 and Nanog gene promoters (Supporting Information Fig. 1A). To verify the pluripotency of these iPSC lines, we examined the expression of Oct4, Nanog, and SSEA-1 and found that all dox-independent iPSC lines expressed these markers of pluripotency (Fig. 2A). We further verified the developmental potency of the primary iPSC lines by SCID mice. Four weeks after injection, teratomas were readily visible and histological analysis confirmed the presence of cell types derived from all three embryonic germ layers validating the pluripotency of these primary iPSCs (Fig. 2B).
We next injected three primary iPSC clones (#3, #7, and #9) into blastocysts and generated chimeras comprising secondary somatic cells as well as host blastocyst-derived cells. Primary cortical neuron cultures were derived from the CP of these chimeras' forebrains at P7 and were immediately exposed to dox for 1 week after which time we purified eGFP+ cells by FACS to remove host blastocyst-derived cells and contaminating cell types (Fig. 2C). Immunostaining confirmed reactivation of the polycistronic reprogramming cassette in eGFP+ secondary postnatal neurons after dox exposure (Supporting Information Fig. 1B), with those derived from iPSC line #7 showing the most robust activation (Supporting Information Fig. 1C). Therefore, we used secondary postnatal neurons from iPSC line #7 for subsequent experiments.
We initially confirmed that the eGFP+ secondary postnatal cortical neurons were postmitotic after isolation and culture by BrdU incorporation. Approximately, 2.43% of total cells in the neuronal culture were negative for BrdU incorporation and positive for eGFP with no appreciable double positive population, indicating their postmitotic status (Fig. 2C, left panel). On addition of dox to the cultures, a small subset of eGFP+ cells in the secondary system neuronal culture began cycling in response to expression of the four reprogramming factors, approximately 0.43% and 0.57% on days 5 and 9 of dox exposure, respectively (Fig. 2C, middle and right panels and Supporting Information Fig. 2). Thus, these data show that the CamKII-Cre/LSL-eGFP system induces eGFP expression faithfully in the terminally differentiated postmitotic neurons of the cerebral cortex and that few postmitotic cortical neurons can re-enter the cell cycle upon exposure to the reprogramming factors using the dox-inducible secondary system.
We assessed the reprogramming activity of eGFP+ secondary neurons during 2 weeks of dox treatment. Initially, secondary CamKII-eGFP+ neurons were treated for 9 days with dox, then FACS purified by gating on eGFP expression, replated and cultured in the presence or absence of dox. During this period, FACS-purified secondary neurons underwent little or no proliferation in either the presence or absence of dox, whereas secondary fibroblasts derived from the same mice underwent robust growth in response to dox treatment (Fig. 3A). CamKII-eGFP+ postnatal neurons showed no repression of neuronal genes after 14 days of dox treatment (Fig. 3B) and no induction of pluripotency associated markers such as AP or endogenous Oct4 (Fig. 3B). We maintained the eGFP+ secondary neurons up to 16 weeks with dox and observed very little proliferation or cell death in these cultures, and no induction of reprogramming (data not shown). These results suggest that the activity of the canonical O,K,S,M reprogramming factors is insufficient to mediate nuclear reprogramming in mature neurons. In contrast to CamKII-eGFP+ postnatal neurons, secondary fibroblasts were able to undergo reprogramming in the presence of dox alone (Supporting Information Fig. 3A, 3B). Therefore, we considered that additional factors might be necessary for the reprogramming of postnatal neurons, possibly those acting to inhibit neuronal identity. It has been shown that terminally differentiated B lymphocytes can be reprogrammed efficiently after suppression of B cell identity through CCAAT/enhancer-binding protein alpha overexpression ; however, mature pancreatic beta cells can be reprogrammed by the ectoptic expression of only four reprogramming factors . Thus, it remains unclear whether terminally differentiated cells can be efficiently reprogrammed to pluripotency by the four factors. For the reprogramming of postnatal neurons, we chose additional manipulations including p53 inhibition, which is known to accelerate epigenetic reprogramming, and REST overexpression. REST is a well-known negative regulator of neuronal genes that is expressed in non-neuron cells, but turned off in postmitotic neurons, indicating that it acts as a negative regulator of neuronal identity. Therefore, we reasoned that overexpression of REST may aid in reprogramming through inhibition of neuronal gene expression, which might then allow the four factors to establish the pluripotent state.
We asked whether reprogramming could be induced either through accelerated proliferation by inhibition of the p53 tumor suppressor or inhibition of neuronal gene activity by overexpression of REST. Lentiviruses encoding a p53 shRNA construct, and/or REST were introduced into CamKII-eGFP+ cells 1 day after sorting. We found that overexpression of O,K,S,M reprogramming factors alone in secondary neurons had no effect on the reprogramming process; however, overexpression of the O,K,S,M reprogramming factors in combination with inhibition of p53 and overexpression of REST resulted in efficient iPSC formation (Fig. 3C, 3D). On further examination, we found that overexpression of REST alone with O,K,S,M reprogramming factors was insufficient for iPSC formation whereas inhibition of p53 alone with O,K,S,M reprogramming factors was sufficient for iPSC formation (Fig. 3D). Overexpression of REST in CamKII-eGFP+ cells infected with p53 shRNA resulted in an approximate twofold increase in the number of AP+ colonies (Fig. 3D). This observation indicates that (a) increased proliferation driven by inhibition of p53 in combination with the O,K,S,M reprogramming factors is required for the reprogramming of mature postnatal neurons and (b) REST activity can significantly enhance the efficiency of this process, presumably by suppression of neuronal genes. REST overexpression in the secondary fibroblasts did not significantly affect the reprogramming efficiency as expected based on its known role as a suppressor of neuronal identity, while p53 inhibition accelerated reprogramming in secondary fibroblasts as reported previously  (Supporting Information Fig. 2A).
Consistent with this idea, analysis of cell proliferation during the reprogramming period confirmed that p53 inhibition coupled with O,K,S,M activity was sufficient to induce proliferation in eGFP+ secondary neurons (Fig. 3E), and neuronal gene expression was efficiently repressed by the overexpression of REST during reprogramming (Fig. 3F). Taken together, these results demonstrate that inhibition of lineage specific gene expression can positively influence the efficiency of the reprogramming process, but only in the context of active proliferation.
The neuronal origin of the iPSCs resulting from these experiments was confirmed by the expression of eGFP (Fig. 4A). The pluripotent state of both eGFP+ secondary neuron-derived iPSCs as well as secondary fibroblast-derived iPSCs was confirmed by their expression of pluripotency markers Oct4, Nanog, and SSEA1 by loss of methylation on the Oct4 and Nanog gene promoters in the absence of dox, by their ability to form differentiated teratomas upon subcutaneous injection into SCID mice, and by their ability to contribute to chimeras upon injection into blastocysts (Fig. 4B–4E). Neuron-derived secondary iPSCs formed chimeras with similar efficiencies as the primary iPSCs derived from TTFs or secondary iPSCs derived from secondary TTFs (Fig. 4D). We obtained four chimeric mice from neuron-derived iPSCs, but we were not able to test germ-line for transmission of these chimeras due to early formation of tumors. This accelerated tumorigenesis is likely due to p53 loss rather than an intrinsic property of the iPSCs [17, 21]. Southern blot analysis showed that an eGFP+ secondary neuron-derived iPSC carried four copies of the O,K,S,M lentiviral vector and one copy of the REST lentiviral vector (Supporting Information Fig. 4A). Therefore, our results are consistent with the notion that forced cell proliferation is required for the formation of iPSCs from postnatal neurons in response to O,K,S,M expression and that reprogramming is enhanced by inhibition of neuronal gene activity by the REST/NRSF.
In vitro reprogramming by the ectopic expression of defined factors has been achieved in a variety of somatic cell types. However, it is not clear from these studies whether reprogramming can be achieved in terminally differentiated neurons. Here, we report that the genome of postnatal neurons from the cerebral cortex that have undergone epigenetic changes associated with terminal differentiation remains amenable to direct epigenetic reprogramming.
We purified and characterized postnatal cortical neurons employing a Lox-Stop-Lox-eGFP:CamKII-Cre transgenic mouse strain for genetic marking. Using BrdU incorporation coupled with flow cytometric analysis, we determined that the vast majority of the marked cells in postnatal cortical cultures derived from these mice were postmitotic and that about 0.57% of eGFP+ cells became mitotic in response to expression of the four reprogramming factors. Our data also showed that 0.015% of the marked cells were BrdU positive in the absence of reprogramming factor activity, raising the possibility that either a very small number of CamKII-eGFP+ neurons continue to cycle or that non-neuronal cells had been labeled by transgenic expression. This possibility seems unlikely as no iPSC colonies could be generated from cortical neurons through expression of the four canonical reprogramming factors alone. To resolve this issue, it would be desirable to prepare neuronal cultures from adult brain. However, at present, no in vitro neuronal culture method for adult neurons exists. Our result suggests that P7 postnatal neurons are amenable to epigenetic reprogramming only after entering a proliferative state mediated by p53 inhibition along with expression of the four canonical reprogramming factors.
We demonstrate that overexpression of the O,K,S,M reprogramming factors is not sufficient to reprogram the genome of postnatal neurons, while most somatic cell types analyzed to date have been amenable to reprogramming by the O,K,S,M factors alone. Recent studies have shown that inhibition of p53 increases the apparent efficiency of generating iPSCs [17, 21–24]. Importantly, Hanna et al.  reported that almost all pro-B cells are capable of becoming iPSCs and that p53 inhibition accelerates the process via increasing the rate of cellular proliferation. Thus, our findings support a model in which inhibition of p53 improves reprogramming primarily by accelerating the cell cycle.
Additionally, we found that overexpression of REST/NRSF increased the efficiency of the neuronal cell reprogramming without increasing cell proliferation. It has been shown previously that repression of lineage specifying genes can facilitate the reprogramming process . Given that REST inhibits neuronal gene expression , our data suggest that iPSC formation not only requires reactivation of pluripotency genes, but can be enhanced by additional repression of lineage specific genes. However, there is also the possibility that REST/NRSF directly participates in the reprogramming process as REST can be directly activated by Wnt signaling in human cancer cell lines and in spinal cord neurons [27, 28]. As Wnt signaling is closely connected to the core circuitry of pluripotency and the reprogramming process , overexpression of REST in neurons could directly potentiate the reprogramming process. Thus, it will be of interest to understand the underlying molecular mechanism of REST/NRSF effects on the reprogramming process in the future. Eventually, understanding the molecular barriers underlying nuclear reprogramming in postnatal neurons would provide more efficient reprogramming strategies and is critical for pluripotent cell-based regenerative medicine.
In this study, we report that terminally differentiated postnatal neurons can be reprogrammed to a pluripotent state by direct reprogramming, but that the canonical four reprogramming factors are not sufficient to induce this process. Rather, inhibition of p53 is required to reprogram postnatal neurons, and REST inhibition can increase the efficiency of this process. Although reprogramming of a variety of somatic cells has been shown, this is the first study to investigate nuclear reprogramming of terminally differentiated postmitotic cells by direct epigenetic reprogramming, and, thus, this study provides intriguing data proving that the epigenetic state associated with terminally differentiated cells remain amenable to reprogramming.
We thank R. Flannery for veterinary assistance; B. Carey, D. Hockemeyer, F.Soldner, and G. Kemske for comments; G. Welstead for technical assistance and comments; and T. Dicesare for graphics support. This work was supported by grants from the National Institutes of Health (NIH HD045022, NIH 5R37CA084198) and the Howard Hughes Medical Institute.
DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
The authors indicate no potential conflicts of interest.