BAC Transgenesis in Human Embryonic Stem Cells as a Novel Tool to Define the Human Neural Lineage

EMBRYONIC STEM CELLS/INDUCED PLURIPOTENT STEM CELLS BAC Transgenesis in Human Embryonic Stem Cells as a Novel Tool to Define the Human Neural Lineage DIMITRIS G. PLACANTONAKIS,a,g MARK J. TOMISHIMA,b,c FABIEN LAFAILLE,b SABRINA C. DESBORDES,b FAN JIA,h NICHOLAS D. SOCCI,d AGNES VIALE,e,f HYOJIN LEE,b NEIL HARRISON,h VIVIANE TABAR,a LORENZ STUDERa,b Department of Neurosurgery, bDevelopmental Biology Program, cSKI Stem Cell Research Facility, Computational Biology Center, eMolecular Biology Program, fGenomics Core Laboratory, Sloan-Kettering Institute for Cancer Research, New York, New York, USA; gDepartments of Neurological Surgery and h Pharmacology and Anesthesiology, Weill Medical College of Cornell University, New York Presbyterian Hospital, New York, New York, USA

a

d

ABSTRACT Human embryonic stem cells (hESCs) have enormous potential for applications in basic biology and regenerative medicine. However, harnessing the potential of hESCs toward generating homogeneous populations of specialized cells remains challenging. Here we describe a novel technology for the genetic identification of defined hESCderived neural cell types using bacterial artificial chromosome (BAC) transgenesis. We generated hESC lines stably expressing Hes5::GFP, Dll1::GFP, and HB9::GFP BACs that yield green fluorescent protein (GFP)1 neural stem cells, neuroblasts, and motor neurons, respectively. Faith-

ful reporter expression was confirmed by cell fate analysis and appropriate transgene regulation. Prospective isolation of HB9::GFP1 cells yielded purified human motor neurons with proper marker expression and electrophysiological activity. Global mRNA and microRNA analyses of Hes5::GFP1 and HB9::GFP1 populations revealed highly specific expression signatures, suggesting that BAC transgenesis will be a powerful tool for establishing expression libraries that define the human neural lineage and for accessing defined cell types in applications of human disease. STEM CELLS 2009;27:521–532

Disclosure of potential conflicts of interest is found at the end of this article.

INTRODUCTION The availability of human embryonic stem cells (hESCs) offers unprecedented access to the earliest stages of human development [1]. Development of the nervous system is of particular interest given the availability of differentiation protocols that enrich for neural fates [2–5], the need for understanding normal and abnormal human neural development and disease, and the potential of generating unlimited numbers of specialized neural cell types for regenerative medicine. Although conditions have been reported for hESCs that enrich for neural stem cells [6–8] and specific neuronal or glial cell types [3, 5, 9–14], it has become evident that under all these conditions considerable cell heterogeneity persists. Furthermore, most protocols that enrich for a given cell type of interest require long-term culture associated with selective cell

survival and proliferation and with limited understanding of critical cell intermediates (‘‘black box’’). The use of hESCs to model human neural lineage specification in a systematic manner has been hampered by the lack of appropriate genetic tools to observe, prospectively identify, and genetically characterize defined cell stages. Here we present the use of bacterial artificial chromosome (BAC) transgenesis as a novel strategy to generate cell type-specific reporter lines that read out specific neural fates including neural stem/precursor cells, neuroblasts, and motor neurons. Our data provide proof-of-concept for prospective isolation, as well as genetic and functional characterization of multiple genetically defined hESC derivatives. The techniques presented here should enable the generation of larger scale BAC transgenic libraries harnessing hESCs as a powerful genetic reporter system to decipher the human neural lineage.

Author contributions: D.G.P.: conception and design, financial support, collection of data, data analysis and interpretation, manuscript writing, final approval of manuscript; M.J.T.: conception and design, collection of data, data analysis and interpretation, manuscript writing, final approval of manuscript; D.G.P. and M.J.T. contributed equally to this work; F.L.: collection of data, data analysis and interpretation; S.C.D.: collection of data, data analysis and interpretation; F.L. and S.C.D. contributed equally to this work; F.J.: collection of data, data analysis and interpretation; N.D.S.: collection of data, data analysis and interpretation; A.V.: collection of data, data analysis and interpretation; H.L.: collection of data, data analysis and interpretation; N.H.: collection of data, data analysis and interpretation; V.T.: data analysis and interpretation, manuscript writing; L.S.: conception and design, financial support, data analysis and interpretation, manuscript writing, final approval of manuscript. Correspondence: Dimitris G. Placantonakis, M.D., Ph.D., Department of Neurological Surgery, Weill Cornell Medical College, 525 E68th Street, New York, New York 10065, USA. Telephone: 212-746-2374; Fax: 212-746-8190; e-mail: [email protected]; or Mark J. Tomishima, Ph.D., SKI Stem Cell Research Facility, Sloan-Kettering Institute for Cancer Research, 1275 York Ave, New York, New York 10065, USA. Telephone: 212-639-3913; Fax: 646-422-2062; e-mail: [email protected] Received September 6, 2008; accepted for publiC AlphaMed Press 1066-5099/2009/$30.00/0 cation November 28, 2008; first published online in STEM CELLS EXPRESS December 11, 2008. V doi: 10.1634/stemcells.2008-0884

STEM CELLS 2009;27:521–532 www.StemCells.com

BAC Transgenesis in hESCs

522

MATERIALS

AND

METHODS

Human ESC Culture, Nucleofection, BAC Engineering, Karyotyping, and Fluorescence In Situ Hybridization Analysis The H9 (WA-09) hESC line was used for all experiments and was maintained on mitotically inactivated mouse embryonic fibroblasts (MEFs) (Millipore, Billerica, MA, http://www.millipore. com) under conditions described previously [3]. Prior to nucleofection, hESCs were cultured on matrigel (BD Biosciences, San Jose, CA, http://www.bdbiosciences.com) in MEF-conditioned medium supplemented with 10 ng/ml fibroblast growth factor 2 [15]. Cells were then dissociated with Accutase (Innovative Cell Technologies, San Diego, http://www.innovativecelltech.com) and resuspended in solution V (Amaxa, Gaithersburg, MD, http:// www.amaxa.com) at a density of 5
106 cells/100 ll. The cell suspension was then mixed with 5 lg freshly prepared (Princeton Separations, Adelphia, NJ, http://prinsep.com/) BAC DNA, nucleofected under protocol B-16 (Amaxa), and seeded on MEFs in medium supplemented with 10 lM Y-27632 (Sigma-Aldrich, St. Louis, http://www.sigmaaldrich.com) at a density of 3.3
104 cells/cm2. Y-27632 was added to the medium for the first 2 days after nucleofection. Selection with G418 (Sigma-Aldrich) started on day 4 at 25 lg/ml followed by 50 lg/ml on day 14. Colonies were isolated after 3 weeks of G418 selection and maintained subsequently in 25 lg/ml G418. Mouse BACs engineered to express green fluorescent protein (GFP) were obtained from the GENSAT library at Rockefeller University. The Hes5::GFP, Dll1::GFP, and HB9::GFP BACs were engineered from the RP24-341I10, RP23-306J23, and RP24-351I23 parent BACs, respectively (http://www.gensat.org). The BACs were modified to express the neomycin resistance gene as previously described [16]. The engineered BACs were fluorescently labeled with SpectrumRed (Abbott Laboratories, Abbott Park, IL, http://www. abbott.com) and used as probes in fluorescence in situ hybridization (FISH) analysis of transgenic hESC and mouse (m)ESC lines. In the FISH study of the HB9::GFP transgenic hESC line, the endogenous HB9 gene was detected using the human RP11354K9 BAC as a probe. Karyotyping was based on 12-18 metaphases from each transgenic line.

Differentiation to Mesoderm, Endoderm, and Neuroectoderm Mesoderm and endoderm induction were performed as previously described [17] (supporting information).

DAPT Experiments To test the effects of the c-secretase inhibitor N-[N-(3,5-Difluorophenacetyl)-L-alanyl]-S-phenylglycine t-butyl ester (DAPT) (Sigma-Aldrich) on the Hes5::GFP and Dll1::GFP reporter lines, hESCs were cocultured with MS5 feeders for 12 days. Neural aggregates were mechanically isolated and further cultured on polyornithine/laminin-coated plates in N2 medium supplemented with brain-derived neurotrophic factor (BDNF) (20 ng/ml), ascorbic acid (AA) (0.2 mM), and glial cell-derived neurotrophic factor (GDNF) (20 ng/ml) for an additional 7 days in the presence or absence of 10 lM DAPT. At that point, cells were subjected to either fluorescence-activated cell sorting (FACS) or immunofluorescence analysis.

Human Motor Neuron Differentiation, Flow Cytometry, and Postsort Culture Motor neurons were derived from hESCs as previously described [12] (supporting information). FACS analysis was performed on a FACSCalibur (BD Biosciences), whereas cell sorting was performed on MoFlo (Dako, Carpinteria, CA, http://www.dako. com) or FACSAria (BD Biosciences) devices. After sorting, motor neurons were cocultured with hESC-derived mesenchymal

feeders [18] in N2 medium containing BDNF, AA, GDNF, and ciliary neurotrophic factor (CNTF; 50 ng/ml).

Immunofluorescence and Microscopy Standard conventional or confocal immunofluorescent analysis was performed. For detailed information, see supporting information.

Quantitative Polymerase Chain Reaction, Quantitative Reverse Transcription-Polymerase Chain Reaction, and Microarray Analyses Detailed information is provided in the supporting information.

Electrophysiological Analysis Cells were recorded in current-clamp mode at room temperature (20-22 C) using the Multiclamp 700B amplifier (Axon Instruments, Sunnyvale, CA, http://www.moleculardevices.com). The membrane potential was manually adjusted to 60 mV. The extracellular solution contained 145 mM NaCl, 3 mM KCl, 1.5 mM CaCl2, 1 mM MgCl2, 5 mM glucose, and 10 mM HEPES. The intracellular solution contained 135 mM potassium gluconate, 2 mM NaCl, 1 mM CaCl2, 1 mM MgCl2, 10 mM HEPES, 10 mM EGTA, 2 mM Mg-ATP, and 0.4 mM Na-GTP. Current steps were applied for 500 ms to evoke action potentials. Tetrodotoxin (TTX) (Sigma-Aldrich) was applied at a final concentration of 0.5 lM to block action potentials in some recordings. Data were analyzed using Clampfit (Axon Instruments).

Statistical Analysis Statistical comparisons included Student’s t-test and analysis of variance (ANOVA), followed by post hoc analysis with Dunnett’s test. The cutoff for statistical significance was set at p  .05. Statistical analysis was performed with NCSS software (Kaysville, UT, http://www.ncss.com). Data are presented as mean  standard error of the mean.

RESULTS Development of hESC BAC Transgenesis Protocol We previously demonstrated the successful use of BAC transgenesis in mESCs [16] using the GENSAT library of BACs [19], which was engineered to express GFP under the control of developmentally regulated genes relevant to neural lineages. The mESC protocol is based on electroporation and stable selection of BAC transgenic mESCs following retrofitting of BACs with a neomycin resistance cassette [16, 20]. The translation of BAC transgenesis from mESCs to hESCs has been hampered by a number of technical hurdles, including difficulties in transfecting hESCs, especially given the large size of BACs (150-250 kb). Other challenges include suboptimal dissociation protocols and low cloning efficiencies of undifferentiated hESCs. Thus, when we attempted to transfect hESCs with BACs under standard culture conditions using electroporation [21] or lipofection [22], we did not obtain stably transfected clones (data not shown). In order to improve the conditions for stable BAC transgenesis in hESCs, we developed three main modifications to our original protocol. First, we hypothesized that the transfection efficiency of hESCs dissociated into a single-cell suspension may be higher than that obtained with cell clumps [21]. Single-cell dissociation and plating was achieved via enzymatic digestion using Accutase [17] followed by treatment of the cells with the Rho-associated kinase (ROCK) inhibitor Y-27632 (10 lM). Treatment of undifferentiated hESCs with Y-27632 has been previously shown to increase their clonogenic potential [5] and resulted in a significant increase in

Placantonakis, Tomishima, Lafaille et al.

hESC survival and cloning efficiency in our protocol (supporting information Fig. S1). Second, we optimized nucleofection (Amaxa) for dissociated hESCs, confirming previous reports on the use of hESC nucleofection with plasmids [23, 24]. We obtained transient transfection rates of 75.9%  5.4% (n ¼ 3) 2 days after nucleofection with the pMAXGFP plasmid (Amaxa), and readily generated stable clones using a plasmid expressing GFP under the control of the elongation factor 1a promoter (Ef1a::GFP) (supporting information Fig. S1). Third, we adopted a modified selection strategy, based on findings that onset of gene expression from BACs is delayed [25]. Thus, we started selection with G418 on day 4 after nucleofection at 25 lg/ml followed by 50 lg/ml at the 14-day mark. Using these three modifications, we obtained BACtransgenic G418-resistant hESC colonies for the first time with an estimated efficiency of 1.6
106 cells. An outline of the protocol encompassing these modifications is presented in Figure 1A. We focused on three BACs with promoters relevant to neural lineages: the mouse Hes5::GFP, Dll1::GFP, and HB9::GFP BACs. The Hes5::GFP and Dll1::GFP BACs were expected to label cells involved in the Notch signaling pathway [16]. More specifically, Hes5 is transcriptionally upregulated in neural stem cells [26] after activation of the Notch pathway in response to Dll1 or other Notch ligands expressed on the cell surface of adjacent cells [27]. The HB9::GFP BAC was expected to label somatic motor neurons of the brainstem and spinal cord [11, 12, 28]. Short fragments of the HB9 promoter and enhancer elements have been previously used to generate HB9::GFP reporter mice and ESCs derived from these reporter strains [28–30]. In total, 15 G418-resistant clones were obtained after nucleofection with the Hes5::GFP BAC, 23 clones were obtained after nucleofection with the Dll1::GFP BAC, and 28 clones were obtained after nucleofection with the HB9::GFP BAC. After clones were expanded, they were screened for GFP expression upon directed neural differentiation (Fig. 1A). This primary screen identified two Hes5::GFP clones (13.3%), three Dll1::GFP clones (13.0%), and one HB9:GFP clone (3.6%) that expressed GFP at the appropriate developmental stages. There was no GFP expression in the remaining clones. In the case of the Hes5::GFP and Dll1::GFP clones, screening was performed by neural induction on MS5 stromal cells [3, 8] (supporting information Fig. S2A). Screening of the HB9::GFP clones was performed by differentiation to the motor neuron fate using our previously published protocol [12] (supporting information Fig. S2B). All GFP-expressing hESC clones were subjected to a battery of quality control tests (Fig. 1A) designed to determine the karyotype (Fig. 1B), site of genomic integration of the BAC via FISH analysis (Fig. 1C), expression of hESC markers (Fig. 1D), in vitro pluripotency (Fig. 1E), and GFP transgene presence in the genome via quantitative polymerase chain reaction (qPCR) in the case of the HB9::GFP clones (Fig. 1F). More specifically, representative clones for all three nucleofections showed normal karyotypes (46,XX) (Fig. 1B). In all cases, a single BAC integration site was found by FISH analysis (Fig. 1C, red arrows). The integration sites were distinct from the endogenous loci, indicating that integration did not occur via homologous recombination. hESCs of representative clones expressed Oct4 and Nanog, both nuclear markers of hESCs (Fig. 1D), as well as the surface markers stage-specific embryonic antigen (SSEA)-3, SSEA-4, Tra1-60, and Tra1-81, and the intracellular marker alkaline phosphatase (data not shown). The undifferentiated hESCs did not express GFP, except for a small number of GFPþ cells in the Dll1::GFP line (2.2%  0.6% of the cells; n ¼ 3), likely marking cells www.StemCells.com

523

undergoing spontaneous differentiation. All clones maintained their pluripotency in vitro, as evidenced by their ability to differentiate to putative mesodermal cells (Brachyury positive and smooth muscle actin positive) and endodermal (Sox17þ) precursors (Fig. 1E). Because the length of the motor neuron differentiation protocol (5-6 weeks) rendered the GFP screening of the HB9::GFP clones laborious, we also developed a secondary screen based on measuring levels of the GFP transgene using qPCR of genomic DNA (Fig. 1F). The secondary screen identified nine of 27 clones (33.3%) as positive for the transgene, suggesting that only a fraction of the G418-resistant clones were transfected with the intact BAC. To ensure that the GFP expression in the above lines reflected transgene expression that was appropriately regulated at the transcriptional level, GFPþ cells from the Hes5::GFP, Dll1::GFP, and HB9::GFP lines were separated from GFP cells by FACS, and quantitative reverse transcription-polymerase chain reaction (qRT-PCR) was performed to assess the ratio of the corresponding endogenous transcript between GFPþ and GFP cells (Fig. 1G) (for culture conditions see supporting information Fig. S2A). In all samples, transcripts could be readily detected in both the GFPþ and GFP fractions. Quantiative analysis indicated that the Hes5 transcript was 8.1  1.4-fold (n ¼ 4; day 14 of culture), the Dll1 transcript was 115.0  56.8-fold (n ¼ 3; day 14 of culture), and the HB9 transcript was 31.9  19.5-fold (n ¼ 3; day 35 of culture) enriched in the GFPþ populations of the Hes5::GFP, Dll1::GFP, and HB9::GFP transgenic lines, respectively, indicating faithful transcriptional regulation of the GFP transgene.

Characterization of Hes5::GFP1 and Dll1::GFP1 hESC Progeny Upon neural induction, both the Hes5::GFP and Dll1::GFP transgenic lines exhibited GFP expression as early as day 8-10. Although both reporters expressed the transgene in neural rosettes, only the Dll1::GFP transgenic line generated GFP-labeled neuroblasts and immature neurons with welldeveloped neurites (Fig. 2A). This observation was corroborated by time-lapse fluorescent microscopy (supporting information). Selective expression of Dll1 but not Hes5 in neuroblasts is consistent with our previous findings in mESCs [16] and with the proposed roles of Hes5 and Dll1 in the Notch signaling pathway. However, onset of Dll1 expression at the neural rosette stage suggests that a Dll1þ precursor compartment could be involved in Notch signaling during early neural induction. Notch signaling during early central nervous system (CNS) development prior to the onset of neurogenesis may be important for stabilizing a neural versus non-neural fate [31]. To better understand the transcriptional regulation of the GFP transgene in the two reporter lines, we performed a qRT-PCR analysis for the Hes5 and Dll1 transcripts in GFPþ and GFP populations separated by FACS after 2 weeks of neural induction on MS5 feeders (Fig. 2B). For both Hes5::GFP (n ¼ 4) and Dll1::eGFP (n ¼ 3), the same replicate samples were used as in Figure 1G (day 14). In the Hes5::GFP line, the Hes5 and Dll1 transcripts were 8.1  1.4- and 4.7  0.2-fold enriched, respectively, in the GFPþ pool (n ¼ 4). In the case of the Dll1::GFP line, the transcripts were 38.9  22.5- and 115.0  56.8-fold enriched in GFPþ cells (n ¼ 3). Even though these data suggested overlap between the Hes5 and Dll1 transcripts at this early stage of in vitro neural development, the two reporter lines were distinguishable when analyzing the Hes5/Dll1 ratios (Fig. 2C). The Hes5/Dll1 ratio was 1.7  0.3-fold enriched in the GFPþ pool

524

BAC Transgenesis in hESCs

Figure 1. BAC nucleofection of hESCs and basic characterization of BAC transgenic lines. (A): Protocol for the stable nucleofection of hESCs with BACs and transgenic line characterization. (B–E): Hes5::GFP, Dll1::GFP, and HB9::GFP undifferentiated hESCs were assayed for range of genetic and functional markers. (B): Cytogenetic analysis showed normal karyotype in all clones. (C): FISH analysis revealed single BAC integration sites at 13q33, 20p11, and 7q31 in the Hes5::GFP, Dll1::GFP, and HB9::GFP clones, respectively (red arrows), all sites distinct from the endogenous loci at 1p36, 6q27, and 7q36 (green arrow, right). (D): Expression of the pluripotency markers Oct4 and Nanog was confirmed in all clones. Note the presence of a subset of GFPþ but Oct4 cells in the Dll1::GFP culture. (E): In vitro pluripotency assay showed induction of mesodermal (SMA and Brachyury) and endodermal (Sox17) markers in all clones. Note the spontaneous neural differentiation exemplified by GFPþ cells in the case of the Hes5::GFP line under endodermal differentiation conditions. (F): Summary of the qPCR data for GFP using genomic DNA extracted from hESC clones transfected with the HB9::GFP BAC. The data were normalized to the GFP content of the hESC line stably transfected with the Ef1a::GFP plasmid. The arrow indicates the clone that produced GFP expression in motor neurons. (G): qRT-PCR analysis of the Hes5, Dll1, and HB9 transcripts shows enrichment of the transcripts in the GFPþ pools of the Hes5::GFP, Dll1::GFP, and HB9::GFP lines, respectively. Abbreviations: BAC, bacterial artificial chromosome; FISH, fluorescence in situ hybridization; GFP, green fluorescent protein; hESC, human embryonic stem cell; MEF, mouse embryonic fibroblast; qPCR, quantitative polymerase chain reaction; qRT-PCR, quantitative reverse transcription-polymerase chain reaction; SMA, smooth muscle actin.

of the Hes5::GFP reporter line (n ¼ 4) and 0.5  0.2-fold depleted in the GFPþ portion of the Dll1::GFP transgenic line (n ¼ 3; t-test, p < .05), indicating a significant but weak segregation between Hes5-expressing and Dll1-expressing cells at the rosette stage.

Immunofluorescence microscopy with the neuronal precursor marker b-tubulin (Tuj1) corroborated differences between the Hes5::GFP and Dll1::GFP reporter lines (Fig. 2D, 2E). The neuroblasts/neurons that stained brightly for Tuj1 (Fig. 2D, arrows) were GFP in the Hes5::GFP line, but GFPþ in

Placantonakis, Tomishima, Lafaille et al.

525

Figure 2. Basic properties of the Hes::GFP and Dll1::GFP transgenic hESC lines. (A): Fluorescence microscopy on living cells cocultured with MS5 feeders shows that both reporters label neural rosettes (dashed areas), whereas only the Dll1::GFP reporter marks neuroblasts. (B, C): Quantitative reverse transcription-polymerase chain reaction data for the Hes5 and Dll1 transcripts in the two lines shows overlap in their expression. However, the Hes5/Dll1 ratio is significantly higher in the Hes5::GFP line, indicating segregation between the two transcripts (*p < .05). (D): Immunofluorescence microscopy shows that neuroblasts brightly stained with Tuj1 antibody are GFPþ in the case of the Dll1::GFP line, but GFP in the Hes5::GFP line. (E): Higher magnification images show the phenomenon described in (D) in greater detail. (F): Example of fluorescence-activated cell sorting analysis performed on Hes5::GFP cells grown under control conditions or in the presence of DAPT. (G): Cumulative statistics on the effect of DAPT on the percentage of GFPþ cells in the case of the Hes5::GFP line (*p < .001). (H): Cumulative statistics on the percentage of GFPþ cells in the undifferentiated state, without DAPT or with DAPT in the case of the Dll1::GFP line. All groups significantly differed from each other by analysis of variance (*significant post hoc Dunnett’s test). (I): Immunofluorescence analysis if the effects of DAPT on neural cells of the two reporter lines. Abbreviations: DAPT, N-[N-(3,5-Difluorophenacetyl)-L-alanyl]-S-phenylglycine t-butyl ester; GFP, green fluorescent protein; hESC, human embryonic stem cell.

the Dll1::GFP line, suggesting that Dll1 but not Hes5 marks the neuroblast/neuronal compartment. Due to lack of specific antibodies against the Hes5 and Dll1 proteins, we were unable to test the colocalization of GFP with Hes5 and Dll1 in the two lines.

Regulation of Hes5::GFP and Dll1::GFP Reporters by Chemical Inhibition of the Notch Pathway We previously demonstrated in mESC-derived neural progeny that blocking Notch signaling via exposure to a c-secretase inhibitor leads to a decrease in the proportion of Hes5þ cells www.StemCells.com

with a concomitant increase in Dll1þ cells [16]. Here we used this assay to probe for specificity of transgene expression in the human Hes5::GFP and Dll1::GFP reporter lines. Cells were exposed to the c-secretase inhibitor DAPT (10 lM) followed by FACS analysis for GFPþ cells (Fig. 2F). As mentioned above, Hes5::GFP undifferentiated cells exhibited no GFP signal (n ¼ 3) (Fig. 2G), whereas Dll1::GFP undifferentiated hESCs displayed little fluorescence (2.2%  0.6%; n ¼ 3) restricted to Oct4 cells (Fig. 2H). After neural induction (12 days) and further culture in N2 medium containing BDNF, AA, and GDNF (7 days), 79.2%  3.4% and 25.0%  1.6% of the cells (n ¼ 3 for each) were GFPþ in

526

BAC Transgenesis in hESCs

the Hes5::GFP and Dll1::GFP lines, respectively (Fig. 2G, 2H). Exposure to 7 days of DAPT treatment (days 12-19) in parallel cultures decreased the percentage of GFPþ cells to 15.9%  3.0% in the Hes5::GFP line (n ¼ 3; t-test, p < .001) (Fig. 2G), but increased the percentage to 37.7%  2.7% in the Dll1::GFP line (n ¼ 3; ANOVA F2,8 ¼ 96.69, p < .001, with a significant post hoc Dunnett’s test for all comparisons between undifferentiated, untreated, and DAPT-treated cells) (Fig. 2H). Using antibodies against GFP, Tuj1, and the neural stem cell marker nestin, we confirmed the dramatic decrease in GFPþ cells in the Hes5::GFP line after DAPT treatment (Fig. 2I). Moreover, in the case of the Dll1::GFP line, DAPT increased the numbers of GFPþ cells. These GFPþ cells costained with Tuj1 but not nestin, indicating that they represent neuroblasts and immature neurons but not stem cells (Fig. 2I). These data were consistent with our previous observations in mESCs and suggested appropriate transcriptional regulation of the BAC reporter construct in hESC progeny.

Hes5 and Dll1 Reporter Transgene Regulation During Differentiation to Dopaminergic and Motor Neuron Fates The DAPT studies suggested that neuronal differentiation conditions enhance segregation of the Hes5-expressing from the Dll1-expressing cells. We therefore tested the behavior of Hes5::GFP and Dll1::GFP upon directed differentiation towards motor neurons and dopaminergic neurons [3, 12] (supporting information Fig. S2B, S2C). Distinct GFP expression patterns became apparent during rosette expansion (supporting information Fig. S2D). However, further differentiation towards the dopaminergic fate revealed more dramatic differences. In the case of the Hes5::GFP line, GFPþ cells were confined to nestin-positive aggregates (Fig. 3A), marking contaminating undifferentiated rosette clusters. Tyrosine hydroxylase (TH)þ neurons were negative for GFP (Fig. 3A, right, white arrows). In contrast, the Dll1::GFP line yielded GFPþ cells negative for nestin but positive for Tuj1 (Fig. 3B, left and middle, white arrows), with all THþ dopaminergic neurons (Fig. 3B, right, white arrows) expressing GFP. These findings indicate that dopaminergic differentiation produces Dll1þ but Hes5 dopamine neurons. These observations were confirmed during motor neuron differentiation [12]. GFP expression in the Hes5::GFP line was again restricted to areas marking undifferentiated neural rosettes (Fig. 3C). In contrast, the Dll1::GFP line yielded GFPþ neurons expressing Tuj1 (Fig. 3C, middle, white arrows) outside of rosettes (supporting information Fig. S3). Moreover, all HB9þ motor neurons were positive for GFP (Fig. 3C, right, white arrows). Quantitative analysis revealed that 98.6% of Tuj1þ cells coexpressed GFP in the Dll1::GFP line (146 cells, three fields), whereas 21.9% of Tuj1þ cells expressed GFP in the Hes5::GFP line (64 cells, eight fields) (Fig. 3D). During dopaminergic differentiation, 98.7% of the THþ cells expressed GFP in the Dll1::GFP reporter line (153 cells, five fields), whereas only 9.8% of THþ cells colabeled with GFP in the Hes5::GFP line (61 cells, six fields). Gene expression analysis by qRT-PCR under motor neuron differentiation conditions showed a robust enrichment of the Dll1 transcript in the Dll1::GFPþ population but not in the Hes5::GFPþ pool (Fig. 3E). Similarly, the HB9 transcript was also highly enriched in the Dll1::GFPþ cells (49.1  20.8-fold enriched) but not in the Hes5::GFPþ cells (2.2  0.3-fold; p < .05, t-test) (Fig. 3E). Interestingly the Hes5/Dll1 transcript ratio showed a much greater (46-fold) difference in the Hes5::GFP line (1.4  0.1-fold enrichment; n ¼ 3) than in the Dll1::GFP line

Figure 3. Lineage segregation of the Hes5 and Dll1 transcript expression during neuronal differentiation. Differentiation of the Hes5::GFP (A) and Dll1::GFP (B) lines to a dopaminergic fate shows clear differences between the two reporter systems in the labeling of dopaminergic neurons. The insets in (A) on the left show the staining for GFP (green) and nestin or Tuj1, respectively (red). (C): The differences were replicated using the motor neuron differentiation paradigm. (D): Almost all Tuj1þ and THþ dopaminergic neurons are GFPþ in the Dll1::GFP lines, with only a small minority being positive in the Hes5::GFP line. (E, F): Quantitative reverse transcriptionpolymerase chain reaction analysis of the Dll1::GFP and Hes5::GFP lines differentiated to the motor neuron fate (*p < .01, t-test). (G): Schematic showing the abundance of the Hes5 and Dll1 transcripts during maturation of the neural lineage. Abbreviations: DAPI, 40 ,6diamidino-2-phenylindole; GFP, green fluorescent protein; hESC, human embryonic stem cell; TH, tyrosine hydroxylase.

(0.03  0.00-fold enrichment; n ¼ 3; t-test, p < .01) (Fig. 3F) at the motor neuron stage, compared with a 3.4-fold difference in the Hes5/Dll1 ratio at the neural precursor stage (Fig. 2C). These data indicate that, in the human neural lineage, Hes5 is most highly expressed at the neural stem/precursor stage and downregulated during neuronal differentiation. Dll1 is expressed at low levels in the neural stem/precursor cell stage and upregulated at the subsequent neuroblast and neuronal differentiation stages (Fig. 3G).

Placantonakis, Tomishima, Lafaille et al.

527

Figure 4. Basic properties of the HB9::GFP transgenic hESC line. (A): Early in the patterning (day 1) of Pax6þ neural precursors, there is no GFP expression (left). Fluorescence microscopy shows the onset and evolution of GFP expression in living motor neurons after 8 days (middle) and 14 days (right) of RA/SHH exposure. (B): Fluorescence-activated cell sorting analysis of transgenic cells patterned with (125 ng/ml) or without C25II SHH. (C): Summary statistics on the yield of GFPþ cells with and without SHH (*p < .001; n ¼ 3 each). (D): Confocal microscopic images of transgenic motor neurons showing the colocalization of GFP with the endogenous HB9. (E, F): Immunofluorescence analysis of the GFPþ motor neurons. The GFP immunoreactivity colocalized with Isl1, Lim3, and Nkx6.1, but not Nkx2.2, nestin, or BrdU. (G): Quantification of GFP colocalization. Abbreviations: BrdU, 5bromo-20 -deoxyuridine; DAPI, 40 ,6-diamidino-2-phenylindole; GFP, green fluorescent protein; hESC, human embryonic stem cell; RA, retinoic acid; SHH, sonic hedgehog homolog.

Characterization of HB9::GFP Transgenic Motor Neurons Our preliminary experiments showed that the HB9::GFP BAC functions appropriately in mESCs and labels HB9þ motor neurons that also express the transcription factors Isl1, Lim3, and Nkx6.1 [32] (supporting information Fig. S3). Motor neuron differentiation in hESCs was carried out according to our published protocol [12], with minor modifications. Pax6þ neural rosettes [8], which were negative for GFP (Fig. 4A, left) and HB9 (data not shown), were exposed to retinoic acid (RA) and sonic hedgehog homolog (SHH). We first detected GFPþ cells in rosette progeny after 8 days of RA/SHH treatment (Fig. 4A, middle). After 14 days of exwww.StemCells.com

posure to RA/SHH, the number and fluorescence intensity of GFPþ cells increased (Fig. 4A, right). FACS analysis on days 11-15 of RA/SHH exposure showed that the induction of GFP fluorescence was dependent on SHH (t-test, p < .001) (Fig. 4B, 4C). No GFP or HB9 expression was seen when cells were directed to a dopaminergic fate (data not shown). Confocal microscopic analysis of the GFPþ cells showed 66.1%  3.0% (n ¼ 6 fields) colocalization with the endogenous HB9 immunoreactivity (Fig. 4D). Conversely, 68.1%  5.7% (n ¼ 4 fields) of HB9-immunoreactive cells were also GFPþ. The transcription factor profile of the GFPþ motor neurons was reminiscent of their murine counterparts (supporting information Fig. S3), as 85.5%  5.7% (n ¼ 3 fields) of them colocalized with Isl1, 80.8%  2.4% (n ¼ 4 fields) colocalized

528

BAC Transgenesis in hESCs

Figure 5. Enriched motor neuron populations. (A) Sorting of the GFPþ and GFP pools. (B): Enrichment of HB9, Isl1, Lim3, and ChAT transcripts in the GFPþ versus the GFP pool after cell sorting. (C): Plating of cells on polyornithine/laminin dishes immediately after cell sorting. (D): Progressive downregulation of the GFP transgene after sorting, as evidenced by serial fluorescent images of living cells. (E): Motor neurons cultured on hESC-derived mesenchymal feeders for 1 week after cell sorting express ChAT and Tuj1. (F): Motor neuron cluster cultured on hESC-derived feeders for 2 weeks after sorting. The cells express ChAT and NF (upper panel) and extend growth cone-like processes (lower panel). (G–I): Electrophysiological analysis of human motor neurons 1 week after sorting. (G): Motor neurons were visualized with Nomarski (DIC) optics and filled through the patch electrode with AlexaFluor 594 for confirmation of their neuronal character. This technique allowed visualization of neuronal processes and even spines at stages when GFP was downregulated. (H): Motor neurons responded to current injection with tonic AP firing that was blocked by TTX. (I): The firing frequency was linearly correlated with the current pulse amplitude (n ¼ 10). Abbreviations: AP, action potential; ChAT, choline acetyltransferase; DAPI, 40 ,6-diamidino-2-phenylindole; DIC, differential interference contrast; GFP, green fluorescent protein; hESC, human embryonic stem cell; NF, neurofilament; TTX, tetrodotoxin.

with Lim3, and 46.1%  13.2% (n ¼ 5 fields) colocalized with Nkx6.1 (Fig. 4E, 4G). Moreover, GFP showed little colocalization with Nkx2.2 (1.7%  1.0%; n ¼ 3 fields) (Fig. 4F, FG), a marker of the V3 spinal interneuron domain [32]. No Oct4 expression was observed (0%; n ¼ 3 fields) (Fig. 4G). The GFPþ motor neurons expressed neuronal b-tubulin (Tuj1), but not nestin (Fig. 4F), confirming their neuronal character. Likewise, most GFPþ cells did not stain with 5-bromo-20 -deoxyuridine (2.8%  1.1%; n ¼ 10 fields), indicating that HB9 marks postmitotic motor neuron progeny in hESCs (Fig. 4F, FG). We next tested for maintenance of GFP expression following terminal differentiation of motor neurons. Upon RA/SHH withdrawal and addition of GDNF and CNTF to the medium, a decrease in the fraction of GFPþ neurons was observed (supporting information Fig. S4) with a concomitant increase in choline acetyltransferase (ChAT)þ neurons. The GFPþ motor neurons could be separated from GFP cells by flow cytometry with yields averaging 2.6%  0.7% of the sorted cells (n ¼ 6) (Fig. 5A). The enriched motor neuron population was compared with the GFP pool by qRTPCR for the HB9, Isl1, Lim3, and ChAT transcripts (Fig. 5B). The same replicate samples were used for qRT-PCR analysis

as in Figure 1G (n ¼ 3). These transcripts were absent in undifferentiated hESCs and rosettes before exposure to RA/SHH (data not shown). We found robust enrichment (31.9  19.5fold; n ¼ 3) for HB9 in the GFPþ pool compared with its GFP counterpart, suggesting coordinate transcriptional regulation of the endogenous HB9 gene and the GFP transgene. Transcripts for Isl1, Lim3, and ChAT were also enriched in the motor neuron pool by 4.3-, 20.8-, and 10.7-fold, respectively (n ¼ 2 each). The sorted motor neurons (Fig. 5C) could be cultured long term on a layer of hESC-derived mesenchymal feeders [18] (supporting information Fig. S4). Serial fluorescent microscopic analysis confirmed that motor neurons progressively downregulate the GFP transgene (Fig. 5D) while exhibiting increased expression of ChAT and elaboration of dendritic and axonal processes (Fig. 5E, 5F). After 1 month of coculture, only a few cells retained GFP expression. The GFP fraction maintained under identical conditions yielded large numbers of nestin-positive precursor cells and only very few ChATþ neurons. Survival of the enriched motor neuron populations was further confirmed in coculture with primary neonatal rat brain astrocytes (supporting information Fig. S4).

Placantonakis, Tomishima, Lafaille et al.

Electrophysiological Analysis of Human Motor Neurons hESC-derived motor neurons were isolated by FACS following our published [12] or an abbreviated (supporting information Fig. S5) motor neuron induction protocol. GFPþ purified motor neurons were cocultured on human mesenchymal feeders for 1 week prior to electrophysiological analysis using whole-cell patch recordings (supporting information Table S1). We chose the hESC-derived feeders, as opposed to the primary rat astrocytes, to eliminate the risk of false recordings from contaminating rat neurons within primary astrocyte cultures. Motor neurons were targeted using Nomarski (differential interference contrast) optics and filled with AlexaFluor 594 for easy visualization (Fig. 5G). When the membrane potential was held at 60 mV, neurons responded to 500-ms depolarizing current pulses with action potential (AP) firing (Fig. 5H). In 10 of 16 recorded neurons, trains of APs were evoked whose frequency increased with the amplitude of the current pulse (Fig. 5I). The AP trains were characterized by tonic firing without any obvious accommodation and were blocked by 0.5 lM TTX (n ¼ 4), indicating that they are sodium current dependent. There were no rebound calcium spikes elicited after hyperpolarization (data not shown). In summary, these recordings indicated that the transgenic human motor neurons are capable of tonic AP firing in vitro.

Global mRNA and MicroRNA Analyses in Genetically Purified hESC Progeny The isolation of genetically defined cell types along the human neural lineage presents a tool for systematic molecular characterizations and the establishment of cell type-specific expression libraries. As a proof-of-concept, we defined the global mRNA and micro (mi)RNA profile in FACS-purified Hes5::GFPþ neural precursors at day 14 of differentiation, and FACS-purified HB9::GFPþ human motor neurons generated using our novel 4-week differentiation protocol (Fig. 6A). We focused on Hes5::GFPþ rather than Dll1::GFPþ cells as Dll1 expression partially overlaps with both the neural stem cell and motor neuron compartments (Fig. 6A). To compare the global mRNA profiles, we used the Illumina HumanWG-6 BeadChip technology, which covers >48,000 human transcripts. For the miRNA comparisons, we used the Agilent human miRNA platform, which covers 723 human and 76 human viral miRNAs. Data were obtained from biological replicates of the Hes5::GFP and HB9::GFP GFPþ populations. The analysis identified 1,246 mRNA transcripts upregulated in neural stem cells, 1,327 mRNA transcripts upregulated in motor neurons, and 1,231 mRNA transcripts shared between the two cell types (Fig. 6B, supporting information Table S2). The miRNA analysis identified 54 miRNAs upregulated in neural stem cells, 121 miRNAs upregulated in motor neurons, and 115 miRNAs shared between the two populations (Fig. 6C, supporting information Table S3). The top 50 differentially expressed mRNAs and miRNAs for each cell type are displayed in Figure 6D, 6E. Importantly, the pool of mRNAs enriched in motor neurons included Mnx1 (HB9) (shown in red in Fig. 6D) and Isl1, thus validating the results. The motor neuron-specific transcriptome contained numerous Hox transcripts, including several HoxB genes known to be regulated by RA [33]. The analysis of the differentially expressed miRNAs showed robust upregulation of the hsa-miR-302 cluster in neural stem cells and of the hsa-miR-10 cluster in motor neurons. Although the miR-302 cluster has been reported to exclusively mark undifferentiated hESCs [34], our data show persistent expression in rosette stage precursors. One remarkable feature of our miRNA datasets derived from FACS-puriwww.StemCells.com

529

fied populations was the unusually large, up to 2,800-fold, increases observed for miRNAs enriched in HB9::GFPþ cells. This suggests that miRNA fold-changes in expression profiling will be particularly sensitive to the use of purified cell populations. Interestingly, miR-10 expression has been reported in the zebrafish spinal cord regulating expression of HoxB1a and HoxB3a [35]. Additional data on potential crossregulation of mRNA and miRNA targets and gene ontology analyses (supporting information Fig. S6) are available in the supporting information.

DISCUSSION Our study is the first to demonstrate the use of BAC transgenesis in hESCs as a tool to molecularly and functionally define specific cell types along the human neural lineage. BAC transgenesis has become a routine technology for generating transgenic mice, and a library of reporter strains has been generated specific to >500 neurally related genes (http:// www.gensat.org). The technology presented here should enable the establishment of large-scale hESC-based BAC transgenic reporter libraries as a novel genetic model system of human development. The use of BACs is expected to confer a high fidelity of transgene expression compared with the use of short promoter constructs. Furthermore, BAC transgenesis does not require prior mapping of promoter or enhancer regions of a given gene. Although our current study demonstrates the successful use of BACs for generating hESC reporter lines, we have not yet directly compared fidelity of expression or efficiency of establishing transgenic lines with any alternative technique. However, given the paucity of available hESC reporter lines and the difficulties in adapting techniques developed for mouse genetics for use in hESCs [36], our study will greatly enhance the genetic toolset available for hESC studies. The Hes5::GFP and Dll1::GFP BAC transgenic hESC lines serve as an example of monitoring both cell fate and cell signaling during hESC differentiation. Whereas we provide ample evidence that the Dll1::GFP reporter line marks neuroblast and neuronal progeny, we also demonstrate an early phase of Dll1 expression at the neural rosette stage not previously reported. It is tempting to speculate that early Dll1 expression at the rosette stage marks a subset of precursors providing Notch ligand to adjacent precursors at the earliest stages of neural commitment prior to neurogenesis. Previous studies have suggested a role for Notch signaling in stabilizing early neural fate during ESC differentiation [31]. Future studies will be required to directly observe the interaction of Dll1þ and Hes5þ cells during early rosette formation. The derivation of double reporter lines allowing dynamic analysis of expression of both genes simultaneously would be of particular interest. At later differentiation stages, the Dll1:GFP line marks the neuroblast and neuronal compartment selectively and may serve as a universal tool to isolate neuronally committed progeny of various regional identities. The use of the Hes5::GFP line to monitor Notch signaling may serve as a paradigm for generating additional hESC reporter lines specific to other key developmental pathways. Somatic motor neurons are critical in the execution of motor behaviors and a key target of many debilitating neurological disorders such as amyotrophic lateral sclerosis and spinal muscular atrophy. The HB9::GFP hESC reporter line should facilitate studies on disease-related genes in purified human motor neuron progeny. Such studies could address autonomous versus nonautonomous disease mechanisms [29, 30] and

530

BAC Transgenesis in hESCs

Figure 6. mRNA and miRNA comparison of neural stem cells and motor neurons. (A): Schematic illustration of the developmental profile of Hes5, Dll1, and HB9 transcripts. (B,C): Venn diagrams showing differentially expressed mRNAs and miRNAs between the Hes5::GFPþ and HB9::GFPþ populations. (D, E): Diagrams showing the top 50 mRNAs and miRNAs differentially expressed in neural stem cells and motor neurons (see supporting information Tables S3 and S4 for full lists). (F): Functional annotation analysis of differentially expressed transcripts using the Database for Annotation, Visualization, and Integrated Discovery (http://david.abcc.ncifcrf.gov [41]). Left panel: Transcripts enriched in Hes5::GFPþ cells were related to cell cycle, development/differentiation, and cell death. Right panel: Transcripts enriched in HB9::GFPþ were related to neuronal differentiation, homeobox genes, axonal growth, and synapse formation, among others. Only nonredundant categories are shown. Data are sorted for significance of p-value to emphasize the categories most robustly enriched. The complete list of the top 50 categories each, including fold-enrichment and redundant categories, is provided as supporting information Figure S6. Abbreviations: GFP, green fluorescent protein; hESC, human embryonic stem cell; miRNA, microRNA.

Placantonakis, Tomishima, Lafaille et al.

identify novel toxic or protective factors acting directly on human motor neurons. Our study showed strong enrichment for HB9 mRNA in GFPþ versus GFP cells, and HB9 antibody expression in most but not all GFPþ cells. Incomplete colocalization may reflect dynamic differences between transcription and protein expression, incomplete specificity of the mouse HB9 BAC within hESCs, or artificial activation/silencing under in vitro culture conditions. Our global mRNA expression data showed Hox expression patterns compatible with cervical and brachial motor neuron identity. One interesting application of the HB9 reporter line will be the FACS-based isolation of regionally distinct motor neuron types following exposure to alternative caudalizing protocols. The gene expression data also provide the basis for the identification of suitable surface markers that could be used alone or in combination with GFP for the isolation of motor neurons and motor neuron subtypes. Examples of candidate surface markers identified in our study with >10-fold enrichment (supporting information Table S3) include CDH10, a member of type II cadherins expressed in selective motor neuron pools [37], SEMA3C, another surface marker known to be expressed on motor neurons [38], and ISLR2, a immunoglobulin superfamily containing leucine-rich repeat 2. Expression of ISLR2 has not yet been studied in motor neurons, but strong expression can be detected in the ventral horn of ISLR2 BAC transgenic mice (http://www.gensat.org). Global gene expression analysis in the Hes5::GFPþ cells allowed comparisons with data obtained in rosette neural stem cells (R-NSCs) [8], as recently described by our group. Similarly to R-NSCs, Hes5::GFPþ cells were significantly enriched in transcripts related to SHH and Wnt signaling pathways. However, Hes5::GFP cells were enriched in only a subset of classic R-NSC markers, such as Plagl1 and Lef1, while retaining expression of earlier markers found in ESCs, such as mir-302 and LIN28. Future studies will have to address whether Hes5::GFP cells at this early stage (day 14 of differentiation) do represent a distinct earlier neural intermediate preceding formation of classic R-NSCs. Whereas our study demonstrated the survival and function of hESC-derived motor neuron progeny in vitro, future studies should address the potential of grafting purified human motor neurons in vivo in the developing and adult CNS in normal hosts and in relevant animal models of disease. The enrichment of specific neuronal populations will be a critical tool

REFERENCES 1 2 3 4 5 6 7 8

Murry CE, Keller G. Differentiation of embryonic stem cells to clinically relevant populations: Lessons from embryonic development. Cell 2008;132:661–680. Zhang SC, Wernig M, Duncan ID et al. In vitro differentiation of transplantable neural precursors from human embryonic stem cells. Nat Biotechnol 2001;19:1129–1133. Perrier AL, Tabar V, Barberi T et al. Derivation of midbrain dopamine neurons from human embryonic stem cells. Proc Natl Acad Sci U S A 2004;101:12543–12548. Reubinoff BE, Itsykson P, Turetsky T et al. Neural progenitors from human embryonic stem cells. Nat Biotechnol 2001;19:1134–1140. Watanabe K, Ueno M, Kamiya D et al. A ROCK inhibitor permits survival of dissociated human embryonic stem cells. Nat Biotechnol 2007;25:681–686. Tabar V, Panagiotakos G, Greenberg ED et al. Migration and differentiation of neural precursors derived from human embryonic stem cells in the rat brain. Nat Biotechnol 2005;23:601–606. Conti L, Pollard SM, Gorba T et al. Niche-independent symmetrical self-renewal of a mammalian tissue stem cell. PLoS Biol 2005;3: 1594–1606. Elkabetz Y, Panagiotakos G, Al Shamy G et al. Human ES cellderived neural rosettes reveal a functionally distinct early neural stem cell stage. Genes Dev 2008;22:152–165.

www.StemCells.com

531

for the development of cell-based therapies beyond motor neuron disorders. It has been well established that contaminating undifferentiated ESCs can induce teratoma formation in CNS grafting [39]. Furthermore neural rosette-stage precursor cells pose a considerable risk for neural overgrowth [8, 10, 40]. Therefore, purification of specific neuronal precursor populations prior to transplantation will be critical for future preclinical development. Finally, BAC transgenic hESC libraries may become a critical tool for deciphering the human neural lineage at the molecular and functional level and establishing hESCs as a versatile genetic modeling system.

ACKNOWLEDGMENTS We thank S. Gong and N. Heintz (GENSAT library at Rockefeller University) for BAC reagents and R. McKay (NINDS, NIH) for nestin antibody; J. Hendrikx, P. Anderson, and D. Domingo (MSKCC, Flow Cytometry facility) for FACS support; M. Leversha (MSKCC, Cytogenetics facility) for karyotyping and FISH analysis; K. Manova and T. Tong (MSKCC, Molecular Cytology facility) for confocal microscopic analysis; and J. Zhao (MSKCC, Genomics Core Laboratory) for technical help in microarray data analysis. We also would like to thank G. Al Shamy, C. Fasano, I. Lipchina, Y. Elkabetz, G. Lee, Y. Ganat, and A. Maroof for technical assistance and helpful discussions on the manuscript. This work was supported by the Starr Foundation, Project ALS, ALS Association of America, and by an NREF grant from the American Association of Neurological Surgeons to D.G.P. Sabrina C. Desbordes is currently affiliated with the Differentiation and Cancer Program, CRG-Centre de Regulacio´ Geno`mica, Barcelona, Spain. Hyojin Lee is currently affiliated with the Burnham Institute for Medical Research, La Jolla, California, USA.

DISCLOSURE

OF OF

POTENTIAL CONFLICTS INTEREST

The authors indicate no potential conflicts of interest.

9 10 11 12 13 14 15 16 17

Yan Y, Yang D, Zarnowska ED et al. Directed differentiation of dopaminergic neuronal subtypes from human embryonic stem cells. Stem Cells 2005;23:781–790. Roy NS, Cleren C, Singh SK et al. Functional engraftment of human ES cell-derived dopaminergic neurons enriched by coculture with telomerase-immortalized midbrain astrocytes. Nat Med 2006;12:1259–1268. Li XJ, Du ZW, Zarnowska ED et al. Specification of motoneurons from human embryonic stem cells. Nat Biotechnol 2005;23:215–221. Lee H, Al Shamy G, Elkabetz Y et al. Directed differentiation and transplantation of human embryonic stem cell derived motoneurons. Stem Cells 2007;25:1931–1939. Lee G, Kim H, Elkabetz Y et al. Isolation and directed differentiation of neural crest stem cells derived from human embryonic stem cells. Nat Biotechnol 2007;25:1468–1475. Nistor GI, Totoiu MO, Haque N et al. Human embryonic stem cells differentiate into oligodendrocytes in high purity and myelinate after spinal cord transplantation. Glia 2005;49:385–396. Xu C, Inokuma MS, Denham J et al. Feeder-free growth of undifferentiated human embryonic stem cells. Nat Biotechnol 2001;19: 971–974. Tomishima MJ, Hadjantonakis AK, Gong S et al. Production of green fluorescent protein transgenic embryonic stem cells using the GENSAT bacterial artificial chromosome library. Stem Cells 2007;25:39–45. Desbordes SC, Placantonakis DG, Ciro A et al. High-throughput screening assay for the identification of compounds regulating selfrenewal and differentiation in human embryonic stem cells. Cell Stem Cell 2008;2:602–612.

BAC Transgenesis in hESCs

532

18 Barberi T, Bradbury M, Dincer Z et al. Derivation of engraftable skeletal myoblasts from human embryonic stem cells. Nat Med 2007;13: 642–648. 19 Gong S, Zheng C, Doughty ML et al. A gene expression atlas of the central nervous system based on bacterial artificial chromosomes. Nature 2003;425:917–925. 20 Wang Z, Engler P, Longacre A et al. An efficient method for highfidelity BAC/PAC retrofitting with a selectable marker for mammalian cell transfection. Genome Res 2001;11:137–142. 21 Zwaka TP, Thomson JA. Homologous recombination in human embryonic stem cells. Nat Biotechnol 2003;21:319–321. 22 Eiges R, Schuldiner M, Drukker M et al. Establishment of human embryonic stem cell-transfected clones carrying a marker for undifferentiated cells. Curr Biol 2001;11:514–518. 23 Siemen H, Nix M, Endl E et al. Nucleofection of human embryonic stem cells. Stem Cells Dev 2005;14:378–383. 24 Hohenstein KA, Pyle AD, Chern JY et al. Nucleofection mediates high-efficiency stable gene knockdown and transgene expression in human embryonic stem cells. Stem Cells 2008;26:1436–1443. 25 Montigny WJ, Phelps SF, Illenye S et al. Parameters influencing highefficiency transfection of bacterial artificial chromosomes into cultured mammalian cells. Biotechniques 2003;35:796–807. 26 Ohtsuka T, Imayoshi I, Shimojo H et al. Visualization of embryonic neural stem cells using Hes promoters in transgenic mice. Mol Cell Neurosci 2006;31:109–122. 27 Yoon K, Gaiano N. Notch signaling in the mammalian central nervous system: Insights from mouse mutants. Nat Neurosci 2005;8:709–715. 28 Wichterle H, Lieberam I, Porter JA et al. Directed differentiation of embryonic stem cells into motor neurons. Cell 2002;110:385–397. 29 Nagai M, Re DB, Nagata T et al. Astrocytes expressing ALS-linked mutated SOD1 release factors selectively toxic to motor neurons. Nat Neurosci 2007;10:615–622.

30 Di Giorgio FP, Carrasco MA, Siao MC et al. Non-cell autonomous effect of glia on motor neurons in an embryonic stem cell-based ALS model. Nat Neurosci 2007;10:608–614. 31 Lowell S, Benchoua A, Heavey B et al. Notch promotes neural lineage entry by pluripotent embryonic stem cells. PLoS Biol 2006;4:e121. 32 Shirasaki R, Pfaff SL. Transcriptional codes and the control of neuronal identity. Annu Rev Neurosci 2002;25:251–281. 33 Oosterveen T, Niederreither K, Dolle P et al. Retinoids regulate the anterior expression boundaries of 50 Hoxb genes in posterior hindbrain. EMBO J 2003;22:262–269. 34 Suh MR, Lee Y, Kim JY et al. Human embryonic stem cells express a unique set of microRNAs. Dev Biol 2004;270:488–498. 35 Woltering JM, Durston AJ. MiR-10 represses HoxB1a and HoxB3a in zebrafish. PLoS ONE 2008;3:e1396. 36 Giudice A, Trounson A. Genetic modification of human embryonic stem cells for derivation of target cells. Cell Stem Cell 2008;2:422– 433. 37 Price SR, De Marco Garcia NV, Ranscht B et al. Regulation of motor neuron pool sorting by differential expression of type II cadherins. Cell 2002;109:205–216. 38 Zou Y, Stoeckli E, Chen H et al. Squeezing axons out of the gray matter: A role for slit and semaphorin proteins from midline and ventral spinal cord. Cell 2000;102:363–375. 39 Bjorklund LM, Sanchez-Pernaute R, Chung S et al. Embryonic stem cells develop into functional dopaminergic neurons after transplantation in a Parkinson rat model. Proc Natl Acad Sci U S A 2002;99: 2344–2349. 40 Tabar V, Tomishima M, Panagiotakos G et al. Therapeutic cloning in individual Parkinsonian mice. Nat Med 2008;14:379–381. 41 Dennis G Jr, Sherman BT, Hosack DA et al. DAVID: Database for Annotation, Visualization, and Integrated Discovery. Genome Biol 2003;4:P3.

See www.StemCells.com for supporting information available online.