sign in sign up
<<
Home , EN2 (gene), GATA1, HAND2, HES1, HES3, HES5

Role of beta in neural differentiation of mouse embryonic stem cells

Mukesh K. Varshneya, José Inzunzaa, Diana Lupua,b,c, Vaidheeswaran Ganapathya,b, Per Antonsona, Joëlle Rüeggb,d, Ivan Nalvartea,1,2, and Jan-Åke Gustafssona,e,1,2

aDepartment of Biosciences and Nutrition, Karolinska Institutet, 141 83 Huddinge, Sweden; bSwetox, Unit of Toxicology Sciences, Karolinska Institutet, 151 36 Södertälje, Sweden; cDepartment of Toxicology, Iuliu Hatieganu University of Medicine and Pharmacy, 400012 Cluj-Napoca, Romania; dCenter for Molecular Medicine, Department of Clinical Neuroscience, Karolinska Institutet, 171 64 Solna, Sweden; and eCenter for Nuclear Receptors and Cell Signaling, University of Houston, Houston, TX 77204

Contributed by Jan-Åke Gustafsson, October 25, 2017 (sent for review August 10, 2017; reviewed by Luis M. Garcia-Segura and Stephen Safe) The ability to propagate mature cells and tissue from pluripotent in layers II–IV of the somatosensory cortex, which persists in aged stem cells offers enormous promise for treating many diseases, mice (13, 14). However, adult mice lacking ERβ display increased including neurodegenerative diseases. Before such cells can be vulnerability to neurodegeneration (15, 16), increased anxiety-like used successfully in neurodegenerative diseases without causing behavior (17, 18), perturbed serotonin levels in several brain re- unwanted cell growth and migration, regulating growth gions, and lower dopamine levels in the caudate putamen (17), an and migration of neural stem cells need to be well characterized. areaofthemidbrainimplicatedinParkinson’s disease, as well as beta (ERβ) is essential for migration of neurons behavioral deficits related to impaired spatial learning (18, 19). and glial cells in the developing mouse brain. To examine whether These changes are not observed in ERα-KO mice (which instead ERβ influences differentiation of mouse embryonic stem cells show abnormal behavior linked to reproduction), indicating that (mESC) into neural lineages, we compared control and ERβ knock- ERβ serves as the main ER isoform in regulating neuronal devel- out (BERKO) mESCs at defined stages of neural development and opment associated with affective behaviors and cognition. Investi- examined the effects of an ERβ-selective ligand (LY3201) with a gations in mice and rats also indicate that both ERα and ERβ are combination of global and targeted -expression profiling and specifically expressed not only in neurons but also in glial cells in the expression of key pluripotency markers. We found that ERβ different brain areas during development (20). ERβ is expressed was induced in embryoid bodies (EBs) and neural precursor cells before the onset of E2 synthesis in the embryonic mouse brain (9) (NPCs) during development. Proliferation was higher in BERKO and may affect the cortical layering even though E2 is not present. NPCs and was inhibited by LY3201. was reduced in Also, estradiol-independent activation of ERβ is evidenced by the BERKO ES cells, and oligodendrogliogenesis was enhanced. BERKO fact that aromatase-knockout mice (Ar-KO), which are incapable of EBs expressed higher levels of key ectodermal and neural progen- synthesizing E2, do not show any major aberrations in brain struc- itor markers and lower levels of markers for mesoderm and endo- ture (21). Although it is expressed early in embryonal brain devel- β derm lineages. ER -regulated factors are involved in cell adhesion, opment, a detailed molecular study of the functions of ERβ during axon guidance, and signaling of Notch and GABA receptor path- neural development remains to be performed. ways, as well as factors important for the differentiation of neuro- In the present study, we used mouse ES cells (mESCs) from WT nal precursors into dopaminergic neurons ( 1) and for the β β β and ER -KO (BERKO) mice to delineate the role of ER in fate acquisition (Olig2). Our data suggest that ER neural differentiation. We used highly controlled culture conditions is an important component for differentiation into midbrain neu- rons as well as for preventing precocious oligodendrogliogenesis. Significance | | dopamine | oligodendrocyte | midbrain Controlling the proliferation and proper fate acquisition of plu- ripotent stem cells is a major challenge in regenerative therapies ifferentiation of neural precursor cells (NPCs) into specific today. Our study reveals that the estrogen receptor beta (ERβ)is neurons and glial cells is the basis for cell-replacement therapy D an important factor in maintaining the neuroepithelial and in neurological disorders such as Parkinson’s disease (1–3). Al- midbrain stem cell pools by repressing proliferation and early though successful transplantation of NPCs in animal models of ’ – nonneuronal fate acquisition. We report on the factors that Parkinson s disease has been done (1 3), their clinical use is still underlie these effects of ERβ. Further, we report that ERβ facil- limited due to challenges such as stem cell tumorigenesis. There- itates midbrain dopaminergic fate and function. The data pre- fore, more studies are necessary to understand NPC differentiation, sented in this study suggest that ERβ is a factor to be considered proliferation, and maintenance. One factor that may play an im- in designing regenerative therapies for example neurodegen- portant role here is the ovarian steroid hormone 17β-estradiol erative diseases such as Parkinson’s disease. (estrogen, E2), which has shown beneficial effects in ameliorating neurodegeneration in animal models (4, 5) and is involved in the Author contributions: I.N. conceived the study; J.I., I.N., and J.-Å.G. designed research; survival of midbrain dopaminergic neurons and neurotransmitter M.K.V., D.L., V.G., P.A., and I.N. performed research; J.I., J.R., I.N., and J.-Å.G. contributed new reagents/analytic tools; M.K.V., J.I., D.L., J.R., and I.N. analyzed data; and M.K.V., J.R., synthesis and catabolism (6). I.N., and J.-Å.G. wrote the paper. Estrogenic actions are mediated mainly through two distinct Reviewers: L.M.G.-S., Instituto Cajal; and S.S., Texas A&M University. estrogen receptor (ER) subtypes: ERα and ERβ. Both ERs are The authors declare no conflict of interest. expressed in the CNS. ERα is the predominant ER in the hypo- Published under the PNAS license. , arcuate nucleus, and preoptic area, controlling re- β Data deposition: The microarray data are accessible through Omnibus production (7), while ER influences nonreproductive processes (GEO) database, https://www.ncbi.nlm.nih.gov/geo (accession no. GSE103312). in the brain. It is the main ER expressed in the cerebral cortex, 1 – I.N. and J.-Å.G. contributed equally to this work. hippocampus, and the dorsal raphe (8 10), where it modulates 2 β To whom correspondence may be addressed. Email: [email protected] or jgustafs@ hydroxylase (Tph2) levels (11, 12). ER knockout in central.uh.edu. mice affects cortical layering and migration of newly formed This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. neurons at later stages of corticogenesis, resulting in hypocellularity 1073/pnas.1714094114/-/DCSupplemental.

E10428–E10437 | PNAS | Published online November 13, 2017 www.pnas.org/cgi/doi/10.1073/pnas.1714094114 Downloaded by guest on September 29, 2021 to differentiate mESCs toward defined stages of midbrain neu- differentiation day 10 (Fig. 1). Both WT and BERKO NPCs ac- PNAS PLUS rogenesis, allowing systematic investigation of the effect of loss of quired typical rosette-like morphology and expressed ERβ during this process via global and targeted gene-expression the NSC markers and (SI Appendix, Fig. S5A). At this profiling. We report that ERβ is necessary for maintaining neural stage very few cells were positive for Tubb3, a marker for differ- stem cell (NSC) proliferation and self-renewal, likely through entiated neurons (SI Appendix,Fig.S5B), confirming that both Notch–Hes signaling, as well as for decisions of dopaminergic– WT and BERKO NPCs constituted a population dominated serotonergic neuron specification and function. In addition, the by NSCs. loss of ERβ resulted in the enhanced expression of genes involved We found that proliferation was higher in BERKO NPCs than in precocious oligodendrogliogenesis. in WT cells (Fig. 2A), with no measurable differences in apo- ptosis between the genotypes (Fig. 2B). The ERβ-selective ago- Results nist LY3201 (optimal dose set to 0.5 nM) (SI Appendix, Fig. S5C) Pluripotency Characterization of BERKO mESCs and Embryoid Bodies. (22) lowered WT NPC proliferation by ∼50% but had no effect Female mESCs from WT and BERKO mice were genotyped to on BERKO cells (Fig. 2A). The increased BERKO NPC pro- confirm ERβ knockout (SI Appendix, Fig. S1 A, D, and E) and liferation was accompanied by decreased expression of the were adapted to feeder-free culture conditions in a chemically cyclin-dependent kinase (CDK) inhibitors and cell-cycle regula- defined two-inhibitor medium (2i: MEK and GSK3β inhibitors) tors p27kip1 and p21cip1 in BERKO NPCs, which promote G1–S in the presence of inhibitory factor (LIF). The mESCs phase transitions (Fig. 2C). LY3201 treatment had no effects on were then differentiated (Materials and Methods and Fig. 1) to the expression of these regulators in both genotypes, suggesting four defined stages along the midbrain neuronal lineage com- that ERβ functions independently of ligand activation or that the mitment: embryoid body (EB), NPC/NSC, midbrain precursor cells synthesize endogenous estradiol. However, the expression cell (mDPC), and midbrain neuron (MN) (Fig. 1). The absence of aromatase (Cyp19a1) in EBs and NPCs was very low, at the of ERβ in BERKO cultures was confirmed by immunocyto- limit of detection (SI Appendix, Fig. S5D and Dataset S2), which chemistry (SI Appendix, Fig. S1B), qPCR (SI Appendix, Fig. suggests that ERβ may work predominantly independently of S1C), and Western blotting (SI Appendix, Fig. S1F). ERβ was not ligand activation during early neural differentiation. detectable in ES cells, and there were no measurable differences Furthermore, we could not observe any compensatory increase between WT and BERKO ES cells in pluripotency marker ex- in ERα expression in the BERKO NPCs (SI Appendix, Fig. S1C), pression (SI Appendix, Fig. S2), formation of EBs, or differen- and there were no differences in ERα subcellular localization, tiation into the three primary germ layers, ectoderm, mesoderm, which was predominantly nuclear, whereas ERβ localization was and endoderm (SI Appendix, Fig. S3). both nuclear and cytoplasmic (SI Appendix, Fig. S1B). This ERβ was expressed in WT EBs and in all stages of neural de- strengthens the notion that ERα and ERβ function largely in- velopment as differentiation progressed. Upon differentiation to dependently and nonredundantly of each other here. the MN stage, ERβ expression was reduced. (SI Appendix,Fig.S1 F and G). A qPCR array containing a panel of genes characteristic Loss of ERβ Perturbs Notch–Hes Signaling in mESC-Derived NPCs. – of primary germ lineages showed that BERKO EBs expressed NSCs are reliant on Notch Hes signaling during development, – – higher levels of key ectodermal and neural progenitor markers and both Notch and the Hes family of basic helix loop helix [Hes family 5 (Hes5), mesothelin (Msln), and (bHLH) transcriptional repressors are critical to maintain stem- the transmembrane glycoprotein prominin1 (Prom1)] and lower cell identity and avoid depletion through premature differenti- levels of markers for mesoderm and endoderm lineages (Table 1 ation (23). Hes1, Hes3, and Hes5 are highly expressed in NSCs. and SI Appendix,Fig.S4). Thus, ERβ is important in early neural Inactivation of Hes genes leads to an increase in proneural development and has effects as early as the EB cell stage. genes, acceleration of neurogenesis, and premature depletion of NSCs. In BERKO NPCs the expression of Hes3 transcript was BERKO NPCs Exhibit Higher Proliferation. To investigate the mech- decreased by 75% (Fig. 2C); the consequence of this reduction anisms involved in ERβ regulation of NPCs, we derived a ho- should be a reduction in the proliferation of stem cells, early mogeneous population of NPCs from WT and BERKO EBs at differentiation, and depletion of the stem-cell pool. In neuro- epithelial cells persistent activation of Notch receptors is re- quired for sustained Hes signaling (24). Notch components (Notch1 receptor and its ligand Dll1) were lower in BERKO NPCs than in WT NPCs (Fig. 2C). Despite lower Notch signaling and increased proliferation, there was no difference between WT and BERKO NPC cultures in the expression of the neuronal mESC EB NPC mDPC Midbrain Neurons (MNs) WT/ markers Nestin and Tubb3 (Fig. 2C and SI Appendix, Fig. S5).

BERKO Thus, the increased proliferation appeared to increase a non- MEDICAL SCIENCES D0 D4 D10-12 D15-17 D19-21 D29-34 Neural neuronal cell pool. LIF+2i -LIF/-2i Expansion Fate Specification Dopaminergic/Serotonergic Induction neurons ITS+Fn N2+FGFb+Laminin N2+FGFb N2 ERβ Is Involved in Midbrain Fate Determination of NSCs. +FGF8b+Shh +Ascorbic acid To examine +Ascorbic acid whether ERβ affects MN fate, we induced differentiation of WT LY3201 and BERKO NPCs toward mDPCs (Fig. 1). In the BERKO Ligand 0.5 nm exposure LY3201 0.5 nm mDPC population there were more midbrain mesencephalic precursors as measured by the expression of ALCAM, indicating Fig. 1. Neural differentiation scheme. mESCs were cultured in a feeder-free 2i precocious differentiation (684.7 ± 26.5) compared with WT system, and EBs were formed in suspension for 4 d without 2i supplementation. mDPCs (441.8 ± 24.8) (Fig. 3A). At this stage Notch1, Dll1, and Neural induction was performed on 4-d-old EBs using neural lineage-selective Hes3 decreased in the BERKO culture (Fig. 3A), but there was – + medium (ITS-FN), and NPCs were separated on day 10 12. Midbrain fate was no change in the number of mature (Tub3 ) neurons. However, induced in NPCs by adding FGF8b, Shh, and ascorbic acid in expansion medium. a most interesting change was the reduction of Engrailed genes Midbrain precursor cells were obtained during the next 4–6 d, and they could be either expanded in the same medium or used for deriving a mature MN [75% reduction in Engrailed 1 (En1) and 50% reduction in En- population (dopaminergic and serotonergic neurons) by withdrawing all grailed 2 (En2)]. Engrailed genes are key factors for the pro- growth factors except ascorbic acid for next 10–15 d. The cells were treated with gramming and maintenance of dopaminergic neurons (25–27). the ERβ-selective agonist LY3201 for 4 d at 0.5 nM as indicated. Further analysis of transcription factors involved in midbrain fate

Varshney et al. PNAS | Published online November 13, 2017 | E10429 Downloaded by guest on September 29, 2021 Table 1. Absence of ERβ perturbs neuroectoderm differentiation in 14-d-old EBs Down-regulated (≤twofold) Up-regulated (≥twofold) Expression/function

Ectoderm Hes5 Ectoderm neuronal progenitors Krt10 Ectoderm terminal differentiation–keratinocytes Krt14 Ectoderm terminal differentiation–keratinocytes Msln Ectoderm limbal progenitors Neurog2 Neuroectoderm Nkx2-2 Ectoderm oligodendrocyte progenitors Prom1 Ectoderm neuronal progenitors Mesoderm Ccr5 Mesoderm terminal differentiation markers–macrophages Cd3e Mesoderm early T cell progenitors (cell surface glycoprotein) Cd79a Mesoderm early B cell progenitors (MB-1 membrane glycoprotein) Hand1 Mesoderm (heart and neural crest derivatives expressed 1) Hand2 Mesoderm cardiomyocyte progenitors Ibsp Mesoderm terminal differentiation– Myh1 Mesoderm terminal differentiation–skeletal muscle cells Ptcra Mesoderm early T cell progenitors Endoderm Gata1 Globin transcription factor 1–endoderm Tat Endoderm terminal differentiation–hepatocytes Miox Endoderm terminal differentiation–proximal tubule cells

Panel of mature and immature cell lineage markers with significant differential expression. BERKO EBs show up-regulation of key neural progenitor markers, while some of the markers for mesoderm and endoderm lineage are down-regulated.

determination revealed a small reduction in orthodenticle ho- Slc6a4) (Fig. 3B), but there was no change in the dopaminergic meobox 2 (Otx2), while NK2 2 (Nkx2-2) was slightly marker hydroxylase (Th, a rate-limiting enzyme of do- up-regulated (Fig. 3B). In line with the changes Otx2 and Nkx2-2 pamine synthesis) or Slc6a3 (a dopamine transporter) (Fig. 3B). expression, in BERKO mDPCs there was increased expression Thus, BERKO mDPCs have a perturbed midbrain neurogenesis of markers of serotonergic neurons (Tph2, the rate limiting en- and are more prone toward a serotonergic fate (midbrain– zyme for serotonin , and the serotonin transporter, hindbrain boundary cells) than a dopaminergic fate.

A NPC proliferation assay 5 WT 1.5x10 **

2 BERKO **** 1.0x105 WT BERKO

0.5x105 DAPI Cells/cm

0 BrdU

Control LY3201 Control

Apoptotic NPCs B 18 WT Casp7 15 BERKO 12 9 DAPI 6

% Activated 3 BrdU caspase7+ cells

0 LY3201 Control LY3201 Casp7 C WT 1.5 p21 1.5 p27 1.5 Notch1 1.5 Dll1 BERKO 1.0 1.0 **** 1.0 1.0 *** **** **** **** *** ** **** 0.5 0.5 0.5 0.5 expression Fold mRNA 0.0 0.0 0.0 0.0 Control LY3201 Control LY3201 Control LY3201 Control LY3201 WT 1.5 Hes1 1.5 Hes3 1.5 Nes 1.5 Tubb3 BERKO 1.0 1.0 1.0 1.0 *** *** 0.5 0.5 **** **** 0.5 0.5 expression Fold mRNA 0.0 0.0 0.0 0.0 Control LY3201 Control LY3201 Control LY3201 Control LY3201

Fig. 2. ERβ-KO NPCs exhibit higher proliferation and perturbed Notch–Hes signaling. (A) Cell viability/proliferation assay showing the increased proliferation of BERKO (116,324 ± 1,049) compared with WT (59,399 ± 1,734) NPCs. Treatment with the ERβ ligand LY3201 (0.5 nM) decreased the proliferation of WT NPCs (34,676 ± 3,630). (B, Left) The percentage of activated Caspase7+ cells in WT (light green) and BERKO (dark green) NPCs. (Right) The expression of BrdU and Caspase7 (Casp7). Images are representative of three independent experiments. (Scale bars: 50 μm.) (C) Real-time qPCR analysis of transcript levels of the cyclin-dependent kinase inhibitors p21cip1 (p21) and p27kip1 (p27), Notch–Hes pathway components Notch1, Dll1, Hes1, and Hes3, the NSC marker Nestin, and the neural differentiation marker Tubb3 in WT (light green) and BERKO (dark green) NPCs with or without the ERβ ligand LY3201. All transcript levels are normalized to the respective RPLP0 levels in each sample, and log2 changes are shown relative to individual transcript levels in WT NPCs. Values were obtained from three independent experiments with three technical replicates in each and represent the mean ± SEM; **P < 0.01, ***P < 0.001, ****P < 0.0001, two-way ANOVA.

E10430 | www.pnas.org/cgi/doi/10.1073/pnas.1714094114 Varshney et al. Downloaded by guest on September 29, 2021 PNAS PLUS

A mDPCs 1.5 p21 1.5 p27 1.5 Tubb3 800 WT **** BERKO 600 1.0 1.0 *** 1.0 400 ** 0.5 0.5 0.5

view field 200 Fold mRNA expression

Alcam+ cells / 0 WT BERKO 0.0 0.0 0.0 Control LY3201 Control LY3201 Control LY3201

1.5 Notch1 Dll1 Hes1 Hes3 1.5 1.5 1.5 WT BERKO * 1.0 1.0 1.0 * 1.0 *** *** **** **** 0.5 0.5 0.5 0.5 **** **** Fold mRNA expression 0.0 0.0 0.0 0.0 Control LY3201 Control LY3201 Control LY3201 Control LY3201 B 1.5 En1 1.5 En2 1.5 Otx2 2.0 Nkx2-2 * **** WT 1.5 BERKO 1.0 1.0 ** 1.0 ** **** 1.0 0.5 0.5 0.5 **** **** 0.5 Fold mRNA expression 0.0 0.0 0.0 0.0 Control LY3201 Control LY3201 Control LY3201 Control LY3201

1.5 Th 1.5 Slc6a3 3.0 Tph2 2.5 WT Slc6a4 BERKO * 2.0 ** * * 1.0 1.0 2.0 1.5 1.0 0.5 0.5 1.0 0.5 Fold mRNA expression 0.0 0.0 0.0 0.0 Control LY3201 Control LY3201 Control LY3201 Control LY3201

Fig. 3. ERβ-KO mDPCs undergo Notch–Hes pathway down-regulation and exhibit perturbed midbrain fate specification. (A) BERKO mDPCs demonstrate higher numbers of Alcam+ cells (684.7 ± 26.5) compared with WT (441.8 ± 24.8) mDPCs. qPCR analysis of the transcript levels of the cyclin-dependent kinase inhibitors p21cip1 (p21) and p27kip1 (p27), the neural marker Tubb3, and the Notch–Hes signaling components Notch1, Dll1, Hes1, and Hes3 in WT (light green) and BERKO (dark green) mDPCs. (B) qPCR analysis of transcript levels of midbrain fate determinant transcription factors (En1, En2, Otx2, and Nkx2-2) and dopaminergic (Th and Slc6a3) and serotonergic (Tph2 and Slc6a4) neuronal identity markers in WT (light green) and BERKO (dark green) mDPCs. Values were obtained from three independent experiments with two technical replicates in each and represent means ± SEM; *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, Student’s t test or two-way ANOVA.

ERβ Maintains Midbrain Neurogenesis and Extracellular Dopamine stage, suggesting that early ERβ ligand-mediated effects can be Levels. Further differentiation of mDPCs into MNs (Fig. 1) discernible at later stages of differentiation. resulted in a cell population composed of a substantial number To analyze the populations of mature and immature MNs, we of dopaminergic as well as serotonergic neurons, with very few sorted the cells based on the expression of the mesencephalic pre- neural progenitors (SI Appendix, Fig. S6). The percentage of the cursor cell-surface marker Alcam. BERKO MN cultures displayed + dopaminergic neurons was significantly lower in BERKO cells a significantly higher proportion of Alcam cells (384.8 × 103 ± (33.9% ± 1.0) than in WT cultures (51.0% ± 1.7) (Fig. 4 A and 30.3 × 103) than did WT cultures (73.2 × 103 ± 2.9 × 103)(Fig.4F). + C). On the other hand, there was an increase in the percentage In addition, the BERKO Alcam cells had lower expression of the of serotonergic neurons in BERKO cells (38.4% ± 0.9) as midbrain fate-specific transcription factors Lmx1b and En1 (Fig. compared with WT cells (21.8% ± 0.8) (Fig. 4 B and C). Along 4G), suggesting that a significant proportion of BERKO mDPCs with this predominance of serotonergic over dopaminergic did not fully differentiate to MNs, resulting in a cell population with neurons, there was a marked decrease in Th transcripts (Fig. 4D) a decreased mesencephalic dopaminergic profile. and in the expression of the dopamine transporter Slc6a3 in BERKO cells (Fig. 4D). In addition, dopamine levels were higher Gene-Expression Profiling of Neural Developmental Stages. To fur- in WT than in BERKO MNs (55.6 ± 12.5 vs. 19.0 ± 7.0 pg/mL) ther explore the molecular underpinnings of the perturbed (Fig. 4E). midbrain neurogenesis in BERKO cells, we performed a global BERKO MNs expressed higher transcript levels of Tph2 than gene-profiling analysis of WT and BERKO cells from different WT cells (Fig. 4D), indicative of the shift toward serotonergic developmental stages and with or without LY3201 treatment.

neurons. However, there was a marked decrease in the expres- Unsupervised hierarchical clustering of cell types from each MEDICAL SCIENCES sion of the presynaptic serotonin transporter Slc6a4 in BERKO developmental stage revealed that mESCs, NPCs, mDPCs, and cells (Fig. 4D), and there was no significant difference in sero- MNs each have distinct gene-expression patterns that cluster into tonin release between WT and BERKO-derived MNs (Fig. 4E), four distinct classes. Overall, 7,412 genes showed differential indicating suboptimal serotonergic signaling. In line with this, gene-expression patterns across all developmental stages, geno- + Tph2 BERKO neurons had fewer neurite projections and had a types, and treatments (SI Appendix, Fig. S7 and Datasets S1 and more rounded, spindle-shaped morphology (Fig. 4B). These S2). Percentages of differentially expressed genes in the WT and findings indicate a selective preference of BERKO midbrain BERKO cell types at each developmental stage are summarized precursors to acquire a serotonergic fate with defective mature in SI Appendix, Table S1. Functional enrichment analysis [gene functions. ontology (GO) analysis] of genes differentially regulated in Treatment of mDPCs with LY3201 resulted in higher expres- BERKO and WT NPCs and mDPCs revealed pathways related sion of Th (Fig. 4D) and dopamine levels (170.8 ± 6.3 pg/mL) to NSC development and proliferation, such as integrin cell- compared with untreated (55.6 ± 12.5 pg/mL) or BERKO MNs surface interactions, laminin interactions, extracellular matrix (19.4 ± 14.4 pg/mL) (Fig. 4E) but had no effect on Tph2 or Slc6a4 organization, and Wnt and Notch signaling, as well as GABA-A expression in mature MNs (Fig. 4D). Thus, LY3201 treatment receptor activation, Reelin signaling, and glutamate receptor during the NPC or mDPC stage evoked some significant differ- pathways (Fig. 5 and SI Appendix, Figs. S8–S11 and Dataset S3). ences in gene expression that were only visible later at the MN In addition, dopamine metabolic processes were enriched in

Varshney et al. PNAS | Published online November 13, 2017 | E10431 Downloaded by guest on September 29, 2021 A Th Tubb3 DAPI Merge WT

Th Tubb3 DAPI Merge C

60 BERKO 40 **** Neurons + 20

%Th 0 B Tph2 Tubb3 DAPI Merge WT BERKO

50 WT 40 ****

Neurons 30 + 20 Tph2 Tubb3 DAPI Merge 10

%Tph2 0 WT BERKO BERKO

Tph2 Tubb3 DAPI Merge BERKO

Th Slc6a3 Tph2 Slc6a4 D 2.5 1.5 3 1.5 WT 2.0 *** BERKO 1.0 2 **** * 1.0 1.5 ** 1.0 * 0.5 1 0.5 0.5 **** **** **** *** 0.0 0.0 0 0.0 Fold mRNA expression Control LY3201 Fold mRNA expression Control LY3201 Fold mRNA expression Control LY3201 Fold mRNA expression Control LY3201 200 200 E **** 175 150 150 125 WT 100 ** 100 BERKO (pg/ml) (pg/ml) 75 50 50 Serotonin levels Dopamine levels 25 0 0 Control LY3201 Control LY3201

500 1.5 Lmx1b 1.5 En1 F **** G 400 300 1.0 1.0

x1000 200 0.5 0.5 **** 100 Alcam+ mDPCs **** Fold mRNA expression 0 0.0 Fold mRNA expression 0.0 WT BERKO WT BERKO WT BERKO

Fig. 4. ERβ knockout results in altered dopaminergic and serotonergic MN levels and function. (A and B) Combined expression of Th and neuronal marker Tubb3 (A) or Tph2 and Tubb3 (B)inWT(Top) and BERKO (Middle) MNs. (Scale bars: 50 μm.) (B, Bottom) Enlarged sections (70 × 50 μm) of the middle row of B. + + + White and yellow arrows point out spindle-shaped and normally branched Tph2 neurons, respectively. (C) Percentage of Th and Tph2 neurons. (D) qPCR analysis of transcript levels of dopaminergic (Th and Slc6a3) and serotonergic (Tph2 and Slc6a4) identity markers in WT (light green) and BERKO (dark green) MNs. (E) Dopamine (Left) and serotonin (Right) levels after exposure of mDPC-derived MNs to 56 mM KCl. (F) Quantification of Alcam+ residual mDPCs sorted from the final stage of differentiation using magnetic-activated cell sorting. (G) qPCR analysis of Lmx1b and En1 transcript levels in Alcam+ WT (blue) and BERKO (brown) MNs. Values were obtained from three independent experiments with three technical replicates in each and represent means ± SEM; *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, Student’s t test (C, F, and G) or two-way ANOVA (D and E).

E10432 | www.pnas.org/cgi/doi/10.1073/pnas.1714094114 Varshney et al. Downloaded by guest on September 29, 2021 -Log10 p-Value PNAS PLUS NPC resulted in effects that were observed only at the MN stage. 0 2 4 6 8 Some genes strongly affected by early LY3201 treatment Integrin cell surface interactions Collagen biosynthesis and modifying enzymes were RORc, Tiam2, Dnmt1, Anxa3,andNuf2 (SI Appendix, Laminin interactions Fig. S11D). Non-integrin membrane-ECM interactions G2/M DNA replication checkpoint NCAM1 interactions Early Glial Fate Acquisition in ERβ-KO NSCs. Based on the differen- Ephrin signaling DSCAM interactions tial gene-expression analysis, the GABA receptors Gabra4, Mitotic Telophase/Cytokinesis Gabra2, Gabrb1, and Grm3 were highly expressed in BERKO Extracellular matrix organization L1CAM interactions cells (Fig. 6) (28). Increased expression of Gabra4 in BERKO Reelin signalling pathway cultures was confirmed by qPCR analysis and immunocyto- Netrin-1 signaling – – – GABA A receptor activation chemistry (Fig. 7 A C, E G, and I K). Expression of GABA Activated NOTCH2 Transmits Signal to the Nucleus receptors promotes neurogenesis in the later stages of develop- Activation of the AP-1 family of transcription factors EPH-Ephrin signaling ment, but their early presence participates in the transition of Activation of AMPA receptors NSCs to oligodendroglial precursors. Furthermore, the micro- Anterior HOX activation in hindbrain development DNA array data showed that oligodendrocyte-specific factors, such as GABA synthesis, release, reuptake and degradation Omg (oligodendrocyte-myelin glycoprotein) and the oligoden- Dopamine Neurotransmitter Release Cycle Trafficking of GluR2-containing AMPA receptors drocyte fate-specific transcription factor Olig2 (oligodendrocyte Notch-HLH transcription pathway transcription factor 2) were significantly up-regulated in BERKO

-Log10 p-Value cultures (Fig. 6). We could confirm the increased expression of mDPC 0 1 2 3 4 5 6 Olig2 in all BERKO cultures (Fig. 7 A, C, E, G, I, and K), further GABA A receptor activation suggesting that NSCs lacking ERβ may have difficulties in Activated NOTCH2 Transmits Signal to the Nucleus maintaining proper midbrain differentiation. Activation of AMPA receptors Unblocking of NMDA receptor, glutamate binding and Along with this preference for the oligodendrocyte over neu- GABA B receptor activation ronal differentiation pathway, as shown in Fig. 7, there were more Activated NOTCH1 Transmits Signal to the Nucleus + TP53 Regulated G1 Arrest Sox2 progenitor cells in WT cultures than in BERKO cultures, Laminin interactions + EPH-Ephrin signaling but in the BERKO cultures a larger percentage of the Sox2 cells DSCAM interactions also expressed Olig2 (Fig. 7 D and H). The percentage of cells that Non-integrin membrane-ECM interactions + Collagen biosynthesis and modifying enzymes were Olig2 was higher in BERKO than in WT MNs (Fig. 7J, Netrin mediated repulsion signals Bottom). These data indicate a precocious oligodendroglial fate Reelin signalling pathway Class C/3 (Metabotropic glutamate/pheromone receptors) acquisition in BERKO cultures, which could be due to the re- Pre-NOTCH Processing in Golgi duced Notch signaling at the NPC and mDPC stages. Netrin-1 signaling Notch-HLH transcription pathway Some of the genes identified as being strongly down-regulated Degradation of GABA Signaling by NOTCH3 in BERKO mDPCs and MNs were Roundabout guidance re- Signaling by NOTCH4 ceptor 1 (Robo1), Netrin1 (Ntn1), and the extracellular matrix -Log10 p-Value glycoprotein Reelin (Reln) (Fig. 6). Robo1 is involved in axon MN 0 2 4 6 8 guidance, while Ntn1 and Reln are important for establishment Integrin cell surface interactions of cell–cell interactions and neuronal positioning and thus are Collagen biosynthesis and modifying enzymes Laminin interactions highly enriched during neurogenesis. We confirmed the down- Extracellular matrix organization regulation of Reln not only in BERKO mDPCs (Fig. 7 E–G) but EPH-Ephrin signaling Regulation of Commissural axon pathfinding by Slit and Robo also in NPCs (Fig. 7 A–C) and MNs (Fig. 7 I–K) compared with CRMPs in Sema3A signaling WT cultures. Again, we could not observe any effect when L1CAM interactions Ephrin signaling treating the cells with LY3201 or when treating the cells differ- G0 and Early G1 Astrocytic Glutamate- Uptake And Metabolism entiating to MNs with diarylpropionitrile (DPN), a common NCAM1 interactions ERβ-selective agonist (SI Appendix, Fig. S12). These results Netrin-1 signaling Notch-HLH transcription pathway NOTCH1 Intracellular Domain Regulates Transcription Fibronectin matrix formation Reelin signalling pathway NPC mDPC MN Role of Abl in Robo-Slit signaling Sema3A PAK dependent Axon repulsion WT BERKO WT BERKO WT BERKO Sema4D induced cell migration and growth-cone collapse 16.41 Activation of Matrix Metalloproteinases TP53 Regulated G1 Cell Cycle Arrest

DNA methylation MEDICAL SCIENCES

Fig. 5. Affected pathways in BERKO cells at different neural developmental stages. GO pathway analysis of differentially regulated genes in NPCs, mDPCs, and MNs. Graphs show selected classes of significantly enriched pathways (Reactome −log10 P value). The full list of pathways and biological processes can be found in SI Appendix and Dataset S3. Log2 average signal mDPCs (SI Appendix,Fig.S10and Dataset S3). As in NPCs and mDPCs, the pathways affected in the differentiated midbrain stage (MN) were associated with midbrain development and mature neuronal functions, including synaptic transmission and plasticity, 3.82 axon guidance, and dendrite morphogenesis (Fig. 5 and SI Ap- pendix Fig. 6. Differential expression (levels are based on log2 average microarray ,Fig.S11and Dataset S3). LY3201 had no significant ef- signal intensities) of selected genes in BERKO and WT cultures at the in- fects on global gene expression at the mESC, NPC, and mDPC dicated midbrain differentiation stages. Red and green denote high and low stages. However, LY3201 treatment at these precursor stages expression levels, respectively.

Varshney et al. PNAS | Published online November 13, 2017 | E10433 Downloaded by guest on September 29, 2021 propose that ERβ is a factor in regulating midbrain neurogenesis individual Hes expressions. In addition, our gene-profiling by modulating cell adhesion, axon guidance, and neurotransmitter analysis points out that integrin signaling pathways are deregu- processes in a largely ligand-independent manner. In addition, lated in BERKO cells throughout neural differentiation, and since ERβ counteracts precocious oligodendroglial fate acquisition. Notch signaling cooperates with integrin and growth factor (e.g., FGF) signaling in recognizing the NSC niche (41), this may impact Discussion the neural fate acquisitions. Directing the differentiation of mESCs to specific cell fates in Our study revealed that the BERKO NPCs and mDPCs had highly defined culture conditions in vitro is a powerful tool to highly increased expression of Olig2, a transcription factor in- recapitulate developmental processes and permits understanding volved in determining oligodendroglial differentiation, which of how different factors interact with these processes. Such in- suggests that BERKO NPCs favor oligodendroglial fate acqui- formation is of great value in regenerative therapies for neuro- sition. Furthermore, Nkx2-2, which is coexpressed with Olig2 to logical diseases such as Parkinson’s disease. Today, regenerative promote oligodendrogenesis (42), was increased in BERKO therapies are still in limited use due to the risk of stem cell tu- mDPCs. We also observed the overexpression of GABA-A re- morigenicity and the acquisition of alternate mature fates. ceptor subunits α2andα4(Gabra2 and Gabra4) in BERKO In the present study we used mESCs derived from WT and NPCs, mDPCs, and MNs. Oligodendrocyte precursor cells (OPCs) BERKO mice to investigate the role of ERβ at specific stages of as well as mature express GABA-A receptors to midbrain neural development in vitro: NPC, mDPC, and the ma- receive synaptic input from neurons, which in turn is important for + ture MN stages. Both ERα and ERβ are highly expressed in the oligodendrocyte survival through increasing intracellular Ca2 mouse and human developing brain (29–31). ERβ expression, which levels (28). However, excessive GABA-A receptor activation may can be measured as early as at E10.5, precedes ERα expression, cause swelling and damage in the myelinating processes (43), which is seen only after E16.5 (31), and the first availability of thereby increasing vulnerability to neurodegeneration. Mice embryonic estrogen at E18.5 (9). In the present study, ERα mRNA lacking ERβ in oligodendrocytes are more prone to myelin dam- levels were ∼20-fold lower than ERβ levels in NPCs (SI Appendix, age than WT mice in the experimental autoimmune encephalitis Fig. S1C). The NPC stage in our study represents E8.5–E9.5. model of multiple sclerosis (16). The increased expression of Analysis of differentiation along the three germ lineages (ec- Olig2, Gabra4,andGabra2, as well as of the oligodendrocyte- toderm, endoderm, and mesoderm) in WT and BERKO EBs myelin glycoprotein Omg, in BERKO cultures strongly suggests revealed the up-regulation of genes related to neuroectoderm that the lack of ERβ favors precocious commitment toward an progenitors in BERKO EBs (Table 1), suggesting either that the oligodendroglial fate, which in turn could be due to perturbed neural progenitors in those differentiated EBs did not terminally Notch–Hes signaling in the NPCs. Although no gross phenotypic differentiate or that they were expanding more rapidly. Homo- effects on myelination have been observed in mice lacking ERβ, geneously derived BERKO NPCs from BERKO EBs as well as these data warrant deeper studies on the role of ERβ in myeli- BERKO mDPCs proliferated more than the WT progeny (Figs. nation and oligodendrocyte function. 2 and 3A). Furthermore, the CDK inhibitors p21cip1 (p21) and In the present study we observed significantly lower levels of p27kip1 (p27) were down-regulated. Several studies have shown both Th and Slc6a3 and lower levels of dopamine in BERKO that ERβ can mediate antiproliferative effects through its direct MN cultures (Fig. 4). Analysis of factors involved in midbrain or indirect repression of cyclin expression and activation of development showed a marked down-regulation of the tran- p21 in human breast cancer cells (32–35). In addition, the ERβ- scription factors En1 and En2 in BERKO mDPCs. En1 and En2, selective agonist LY3201 decreased the proliferation of WT among others, are involved in the determination and patterning NPCs (Fig. 2A). NPCs were grown in a medium containing of the midbrain/hindbrain precursors and play essential roles in FGF2, a mitogen involved in the growth and proliferation of the development and maintenance of mesencephalic dopami- NSCs; thus the antiproliferative actions of ERβ could be in nergic neurons (25–27). Otx2, which was slightly down-regulated conjunction with FGF and other growth factor signaling path- in BERKO mDPCs, is involved in the earlier positioning of the ways and might be important in preventing the tumorigenicity of midbrain–hindbrain junction and is also required for the genesis stem cells. of midbrain precursors. Increased Otx2 expression represses A major signaling pathway involved in early embryogenesis is Nkx2-2, a negative regulator of the midbrain dopaminergic the Notch–Hes pathway. It mediates cell–cell communication, program, resulting in a shift to serotonergic neuron differentia- maintains stem cell renewal, and controls neural fate decisions and tion at the expense of dopaminergic neurons. It was reported cell proliferation (23). Activation of Notch by its ligands (e.g., that ERα and ERβ might mediate ligand-dependent En1 tran- Delta1/4) induces the expression of transcriptional repressor genes scription through their interaction with β-catenin and GSK3 of such as Hes1 and Hes5 (canonical) and Hes3 (noncanonical) that the Wnt signaling pathway (44). Our results suggest that ERβ repress the expression of , culminating in the in- may be a key factor in mesencephalic dopaminergic neuron de- hibition of neuronal differentiation. Thus, Notch signaling velopment through its regulation of En1 and En2.LY3201treat- maintains the NSC population, whereas inactivation or down- ment in our study significantly increased dopamine levels in WT regulation of Notch signaling results in NSC depletion and in- MNs. Thus, although we could not detect any significant tran- duces differentiation to neuronal and oligodendroglial fates (23, scriptional changes upon LY3201 treatment at the NPC or mDPC 36–39). In this study, Notch1, Dll1, Hes1, and Hes3 were all sig- stages, ligand activation of ERβ during these early stages never- nificantly lower in BERKO NPCs and mDPCs than in WT cells. theless appears to be important for the mature dopaminergic Depending on the expression dynamics, each Hes factor can neuron function in the MN culture, which also expressed the have two contradictory functions, promoting either proliferation lowest levels of ERβ protein (SI Appendix,Fig.S1F and G). Early or cell-cycle exit for differentiation. The oscillatory expression of LY3201 treatment did result in the differential expression of a few multiple Hes factors correlates with a proliferative state, whereas genes in the mature MNs, including Rorc, Dnmt1,andAnxa3; Rorc fate determination can be attributed to the sustained expression expression was recently shown to be ERβ ligand-mediated in the of a selected single Hes factor (40). Our study reveals that high prostate (55). These data further suggest that the small changes in expression of Hes is not sustained in BERKO progenitor cells, gene expression induced by ligand activation of ERβ during early suggesting that ERβ participates in maintaining Notch- neural development may have significant implications at later or dependent NSC expansion and self-renewal. Thus, loss of ERβ mature stages. may promote both proliferation and early imbalances in neural As mentioned above, our study shows that loss of ERβ down- fate acquisitions, which may correlate with oscillatory effects of regulates the expression of p21, p27, and Notch signaling

E10434 | www.pnas.org/cgi/doi/10.1073/pnas.1714094114 Varshney et al. Downloaded by guest on September 29, 2021 PNAS PLUS B 80 A 60 **** DAPI

DAPI 40 DAPI 20 WT 0 Gabra4 levels (AU)

Gabra4 WT BERKO Olig2 Sox2 Reelin Nestin 80 60 DAPI

DAPI 40 *** DAPI 20 NPC

BERKO 0 Reelin levels (AU) WT BERKO Gabra4 Olig2 Sox2 Reelin Nestin 80 C 8 Gabra4 1.5 Reln 4 Olig2 WT D **** BERKO 60 6 **** **** 3 **** 1.0 40 **** 4 2 **** 0.5 **** **** % cells 2 1 20 ****

0 0.0 0 Fold mRNA expression 0 Fold mRNA expression Control LY3201 Fold mRNA expression Control LY3201 Control LY3201 Sox2+ Olig2+ Sox2+ + F 80 Olig2 E 60 **** DAPI DAPI 40 WT

Alcam 20 Reelin DAPI 0

Gabra4 levels (AU) WT BERKO Olig2 Sox2 Foxa2 Gabra4 150

DAPI 100 DAPI

50

Alcam **** BERKO mDPC 0 Reelin levels (AU)

Olig2 Sox2 WT BERKO Foxa2 Reelin DAPI Gabra4 80 Gabra4 Reln Olig2 WT H G 15 **** 1.5 4 **** BERKO 60 3 **** 10 1.0 **** **** 40 2

% cells **** 5 0.5 20 **** **** **** 1

Fold mRNA expression 0 Fold mRNA expression 0.0 0 0 Control LY3201 Control LY3201 Fold mRNA expression Control LY3201 Sox2+ Olig2+ Sox2+ Olig2+ 100 I J 80 **** 60

Th DAPI 40 WT 20 Olig2 Th DAPI

Th Reelin DAPI 0 Gabra4 levels (AU) Gabra4 WT BERKO 150

100 DAPI Th DAPI

MN 50 **** BERKO

Reelin levels (AU) 0 Olig2 Th DAPI Th Reelin

Gabra4 WT BERKO MEDICAL SCIENCES 15 Gabra4 1.5 Reln 5 Olig2 WT K **** *** ** 20 **** 4 BERKO 15 **** 10 1.0 3 cells in + **** 2 10 5 0.5

1 MN culture 5 0 0.0 0 % Olig2 0 Fold mRNA expression Control LY3201 Fold mRNA expression Control LY3201 Fold mRNA expression Control LY3201 WT BERKO

Fig. 7. ERβ-KO neural progenitors are vulnerable to early gliogenesis. (A) Expression of Gabra4 (Left), Reelin/Nestin (Middle), and Olig2/Sox2 (Right) in NPCs. (B) Quantification of Gabra4 (Upper) and Reelin (Lower) immunoreactivity. (C) qPCR analysis of Gabra4, Reln, and Olig2 mRNA in WT (light green) and BERKO + + + + (dark green) NPCs. (D) Percentage of Sox2 , Olig2 , and Sox2 /Olig2 cells. Checkered bars indicate BERKO NPCs. (E) Expression of Gabra4/Alcam (Left), Foxa2/ Reelin (Middle), and Olig2/Sox2 (Right) in mDPCs. (F) Quantification of Gabra4 (Upper) and Reelin (Lower) immunoreactivity. (G) qPCR analysis of Gabra4, + + + + Reln, and Olig2 in WT (light green) and BERKO (dark green) mDPCs. (H) Percentage Sox2 , Olig2 , and Sox2 /Olig2 cells. Checkered bars indicate BERKO mDPCs. (I) Gabra4/Th (Left), Reelin/Th (Middle), and Olig2/TH (Right) immunoreactivity in WT and BERKO MNs. (J) Quantification of of Gabra4 (Top)and Reelin (Middle) immunoreactivity and the percentage of Olig2+ cells (Bottom). (K) Transcript levels of Gabra4, Reln, and Olig2 in WT and BERKO MNs. (Scale bars: 50 μminA, E, and I.) Images are representative of three independent experiments. Values were obtained from three independent experiments with three technical replicates in each and represent means ± SEM; **P < 0.01, ***P < 0.001, ****P < 0.0001, Student’s t test (in B, F, and J) or two-way ANOVA (in C, D, G, H, and K). AU, arbitrary units.

Varshney et al. PNAS | Published online November 13, 2017 | E10435 Downloaded by guest on September 29, 2021 components in NPCs and mDPCs (Figs. 2 and 3) and that EB Formation. EBs were generated at a normally scheduled passage by plating LY3201 treatment had no effect on the expression of these mESCs into non-tissue culture–treated dishes in EB medium without LIF to genes. It has been shown that ERβ can mediate the expression prevent attachment. EBs were used to test the capability of mESCs to dif- of p21, p27,andNotch1 in a ligand-independent manner (33– ferentiate into three germ layers. After 4 d EBs were plated on a gelatin-coated 100-mm tissue-culture dish in EB medium and were grown/differentiated for 35, 45). However, some of these studies also show that the 14 d. (See SI Appendix for details.) addition of E2 can potentiate this expression. Although the context of ERβ’s ligand-independent activation is far from β’ Selection and Expansion of NPCs from mESC-Derived EBs. Derivation and se- clear, it could be speculated that ER s known association with lection of NPCs from mESCs and further differentiation to midbrain cells was specific transcription factors, such as AP-1 (46, 47) and Steroid accomplished using an EB-based method, with modifications of a previously receptor 1 (SRC-1) (48), as well as cell type- reported protocol (54). In brief, 4-d-old EBs were plated on gelatin-coated specific effects, may be of importance here. In addition, al- dishes in EB medium supplemented with 15% ES cell-qualified FBS. The though the neural cells were cultured without serum supple- medium was replaced by insulin-transferrin-selenium (ITS)-fibronectin (FN) – ment, the culture medium did contain phenol red, which has medium, and the EBs were cultured for 6 8 d at 37 °C and 5% CO2. During weak phytoestrogenic properties, and the possibility that these this period, a monolayer of cells with rosette-like morphology (NPCs) grew properties had some agonistic effect on ERβ cannot be ruled from the attached EBs. After derivation of NPCs, LY3201 (0.5 nM; the kind gift of the Eli Lilly Company) or vehicle (ethanol) treatment was given daily out. Unfortunately, the use of phenol red-free medium was not for 4 d after the day of plating. (See SI Appendix for details.) well tolerated by the NPCs during derivation from EBs. These observations warrant further studies of the effects of different Differentiation of NPCs to MNs. Midbrain-specific dopaminergic and seroto- estrogenic ligands, including those of antiestrogens and nergic neurons were differentiated from NPCs by midbrain fate specification. endocrine-disruptive compounds, in different cellular contexts In brief, midbrain fate was induced in neural expansion medium supple- and stages of neural development. mented with FGFb (10 ng/mL) (PeproTech), FGF8b (100 ng/mL) (Miltenyi We also observed an increased number of serotonergic neu- Biotech), sonic hedgehog (Shh) (400 ng/mL) (PeproTech), and ascorbic acid rons in BERKO MNs. Although there was no change in sero- (200 μm) (Tocris). During this period NPCs acquire midbrain fate and gen- tonin levels between WT and BERKO MNs, there were more erate mDPCs. After derivation of mDPCs, LY3201 (0.5 nM) or vehicle (etha- + Tph2 cells in general with abnormal spindle-shaped morphol- nol) treatment was given daily for 4 d beginning on the day after the day of ogy and short neurite projections in BERKO MNs (Fig. 4 B–E). plating. mDPC differentiation to MNs was induced by culturing the cells for the next 10–15 d in neural expansion medium without growth factors but We have previously identified a significant reduction in normal- + + with ascorbic acid. (See SI Appendix for details.) looking Tph2 cells and an increase in Tph2 spindle-shaped β cells in the dorsal raphe of aged ER -KO mice (49). Similar Immunocytochemistry and Fluorescence Imaging. Cells were grown on either β observations in the present study imply that ER , in addition to gelatin-coated (for mESC and EBs) or poly-L-ornithine (100 μg/mL) and maintaining serotonergic neurons in the adult brain, may be Laminin (5 μg/mL, for NPCs) -coated coverslips (Nuvitro) or chambered glass important at an early stage in the development of these neurons. slides (ibidi GmbH) and were fixed in cold 4% paraformaldehyde (Santa Our attention was also drawn to two genes involved in neu- Cruz) for 15–20 min followed by three washes in Dulbecco’s phosphate- ronal migration and axon guidance, Reln and Robo1, re- buffered saline, permeabilized, and blocked with 1% BSA (Sigma-Aldrich), spectively. Both were down-regulated in BERKO mDPCs and 5% normal goat serum (Abcam) or donkey serum (Abcam), and 0.3% Triton have been proposed to have roles in neurodevelopmental dis- X-100 solution in PBS, and stained with the designated primary antibodies orders such as dyslexia (50) and (51, 52). Sex- (see below) followed by detection with Alexa Fluor secondary antibodies (Invitrogen) and DAPI (300 nM) (Molecular Probes). (See SI Appendix specific differences in the prevalence of these disorders have for details.) been established, and sex hormone receptors such as ERβ may be involved here (reviewed in ref. 53). Measurement of Dopamine and Serotonin Release. At the final stage of dif- To conclude, the present study demonstrates that the loss of ERβ ferentiation of NPCs to midbrain cells, culture supernatant was collected for perturbs the proliferation of neuroepithelial and midbrainstem ELISA-based quantification of the key neurotransmitters dopamine and se- cells, resulting in increased oligodendrogliogenesis. These effects rotonin. Cells were rinsed twice with HBSS and then were incubated with are likely propagated through Notch–Hes signaling, which is HBSS containing 56 mM KCl for 15 min and were collected to measure evoked deregulated by the loss of ERβ. Our data also show that ERβ can neurotransmitter release. Samples were stabilized by adding EDTA (final participate in corticogenesis by regulating genes involved in cell concentration 1 mM) and sodium metabisulfite (final concentration 4 mM) to adhesion and axon guidance and that ERβ is needed for proper prevent catecholamine degradation. The competitive dopamine and sero- midbrain fate development through balancing the serotonergic and tonin ELISA kits (Abnova) in microtiter plate format were used according to the manufacturer’s recommendations using cis-diol–specific affinity gel dopaminergic identity. The maintenance of the latter appears to be β dopamine/serotonin extraction followed by acylation and enzymatic deri- mediated through ligand-activated ER . These observations are of vation. The reaction was monitored at 450 nm, and quantification of sam- importance when designing regenerative therapies using NSCs in ples was achieved by comparing their absorbance with a reference curve the treatment of neurodegenerative disorders such as Parkinson’s prepared with known standard concentrations. disease, where the development, survival, and functionality of do- paminergic neurons are essential. Microarray Data Processing and Analysis. Total RNA from NPCs, mDPCs, and MNs treated with vehicle or 0.5 nM LY3201 (according to the scheme in Fig. 1) Materials and Methods was extracted using the Qiagen Allprep kit following the manufacturer’s Derivation, Culture, and Maintenance of Undifferentiated mESCs. Female ES protocol (Qiagen). The mouse MTA-1.0 microarray (Affymetrix) was used cells were derived from blastocysts obtained from time-mated females for gene-profiling experiments according to details given in SI Appendix. (C57BL/6J) at 3.5 d post conception at the Karolinska Center for Transgene The data are accessible through Gene Expression Omnibus (GEO) series ac- Technology (KCTT) Core Facility of the Karolinska Institute, Sweden. Derived cession number GSE103312 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi? + + + + − − mESC lines were genotyped and classified as / XX, / XY (WT), and / XX, acc=GSE103312). − − / XY (BERKO). The genotyping of two WT (XX) and two BERKO (XX) lines is shown in SI Appendix, Fig. S1A. In brief, mESCs were cultured on dishes with Statistical Analysis. Unless stated otherwise, statistical significance was de- irradiated mouse embryonic fibroblasts (KCTT) that had been grown for at termined using an unpaired, two-tailed Student’s t test, assuming unequal least 1 d in ES cell medium containing 1,000 U/mL of LIF (Millipore Inc. or variances for single comparisons or using one-way or two-way ANOVA fol-

PolyGene AG) in a humidified incubator at 37 °C and 5% CO2. (See SI Ap- lowed by the Tukey post hoc test for multiple comparisons. P values were pendix for details.) All animal experiments were carried out according to obtained relative to WT cells if not indicated otherwise. Differences were Swedish guidelines governing animal experimentation, and were approved considered significant if the P value was *P < 0.05, **P < 0.01, ***P < 0.001, by the Stockholm South ethics committee of the Karolinska Institutet. ****P < 0.0001 for all tests.

E10436 | www.pnas.org/cgi/doi/10.1073/pnas.1714094114 Varshney et al. Downloaded by guest on September 29, 2021 ACKNOWLEDGMENTS. We thank Stefan Jäger for excellent technical assistance; Research Committee of the Karolinska Hospital. This study was supported by the PNAS PLUS the KCTT; and the Bioinformatics and Expression Analysis Core Facility at Novum, Center for Innovative Medicine, Robert A. Welch Foundation Grant E-0004, the which is supported by the Board of Research of the Karolinska Institute and the Novo Nordisk Foundation, and the Emil and Wera Cornell Foundation.

1. Bjorklund LM, et al. (2002) Embryonic stem cells develop into functional dopaminergic 29. Fan X, Warner M, Gustafsson JA (2006) Estrogen receptor beta expression in the neurons after transplantation in a Parkinson rat model. Proc Natl Acad Sci USA 99: embryonic brain regulates development of calretinin-immunoreactive GABAergic 2344–2349. interneurons. Proc Natl Acad Sci USA 103:19338–19343. 2. Kim JH, et al. (2002) Dopamine neurons derived from embryonic stem cells function in 30. González M, et al. (2007) Distribution patterns of and beta in an animal model of Parkinson’s disease. Nature 418:50–56. the human cortex and hippocampus during development and adulthood. J Comp 3. Zhang Q, Chen W, Tan S, Lin T (2017) Stem cells for modeling and therapy of Par- Neurol 503:790–802. kinson’s disease. Hum Gene Ther 28:85–98. 31. Lemmen JG, et al. (1999) Expression of estrogen receptor alpha and beta during 4. Bourque M, Dluzen DE, Di Paolo T (2009) Neuroprotective actions of sex steroids in mouse embryogenesis. Mech Dev 81:163–167. Parkinson’s disease. Front Neuroendocrinol 30:142–157. 32. Cheng J, Lee EJ, Madison LD, Lazennec G (2004) Expression of estrogen receptor beta 5. Simpkins JW, Singh M, Brock C, Etgen AM (2012) Neuroprotection and estrogen re- in prostate carcinoma cells inhibits invasion and proliferation and triggers apoptosis. ceptors. Neuroendocrinology 96:119–130. FEBS Lett 566:169–172. 6. Küppers E, Ivanova T, Karolczak M, Beyer C (2000) Estrogen: A multifunctional mes- 33. Lewandowski SA, et al. (2005) Opposite effects of estrogen receptors alpha and beta senger to nigrostriatal dopaminergic neurons. J Neurocytol 29:375–385. on MCF-7 sensitivity to the cytotoxic action of TNF and activity. 24: 7. Laflamme N, Nappi RE, Drolet G, Labrie C, Rivest S (1998) Expression and neuro- 4789–4798. peptidergic characterization of estrogen receptors (ERalpha and ERbeta) throughout the 34. Paruthiyil S, et al. (2004) Estrogen receptor beta inhibits human breast cancer cell rat brain: Anatomical evidence of distinct roles of each subtype. J Neurobiol 36:357–378. proliferation and tumor formation by causing a G2 cell cycle arrest. Cancer Res 64: 8. Bondesson M, Hao R, Lin CY, Williams C, Gustafsson JA (2015) Estrogen receptor 423–428. signaling during vertebrate development. Biochim Biophys Acta 1849:142–151. 35. Treeck O, et al. (2008) Effects of -deleted estrogen receptor beta transcript 9. Fan X, Xu H, Warner M, Gustafsson JA (2010) ERbeta in CNS: New roles in develop- variants on growth, apoptosis and gene expression of human breast cancer cell lines. ment and function. Prog Brain Res 181:233–250. Breast Cancer Res Treat 110:507–520. 10. Takeyama J, et al. (2001) Expression and cellular localization of estrogen receptors 36. Bertrand N, Castro DS, Guillemot F (2002) Proneural genes and the specification of alpha and beta in the human fetus. J Clin Endocrinol Metab 86:2258–2262. neural cell types. Nat Rev Neurosci 3:517–530. 11. Donner N, Handa RJ (2009) Estrogen receptor beta regulates the expression of 37. Kopan R, Ilagan MX (2009) The canonical : Unfolding the tryptophan-hydroxylase 2 mRNA within serotonergic neurons of the rat dorsal raphe activation mechanism. Cell 137:216–233. nuclei. Neuroscience 163:705–718. 38. Mizutani K, Yoon K, Dang L, Tokunaga A, Gaiano N (2007) Differential Notch sig- 12. Gundlah C, et al. (2005) Estrogen receptor-beta regulates tryptophan hydroxylase-1 nalling distinguishes neural stem cells from intermediate progenitors. Nature 449: expression in the murine midbrain raphe. Biol Psychiatry 57:938–942. 351–355. 13. Wang L, Andersson S, Warner M, Gustafsson JA (2001) Morphological abnormalities in the 39. Ross SE, Greenberg ME, Stiles CD (2003) Basic helix-loop-helix factors in cortical de- brains of estrogen receptor beta knockout mice. Proc Natl Acad Sci USA 98:2792–2796. velopment. Neuron 39:13–25. 14. Wang L, Andersson S, Warner M, Gustafsson JA (2003) Estrogen receptor (ER)beta 40. Imayoshi I, Kageyama R (2014) bHLH factors in self-renewal, multipotency, and fate knockout mice reveal a role for ERbeta in migration of cortical neurons in the de- choice of neural progenitor cells. Neuron 82:9–23. veloping brain. Proc Natl Acad Sci USA 100:703–708. 41. Campos LS, Decker L, Taylor V, Skarnes W (2006) Notch, epidermal growth factor 15. Khalaj AJ, Hasselmann J, Augello C, Moore S, Tiwari-Woodruff SK (2016) Nudging receptor, and beta1-integrin pathways are coordinated in neural stem cells. J Biol oligodendrocyte intrinsic signaling to remyelinate and repair: Estrogen receptor li- Chem 281:5300–5309. gand effects. J Steroid Biochem Mol Biol 160:43–52. 42. Zhou Q, Choi G, Anderson DJ (2001) The bHLH transcription factor Olig2 promotes 16. Khalaj AJ, et al. (2013) Estrogen receptor (ER) β expression in oligodendrocytes is oligodendrocyte differentiation in collaboration with Nkx2.2. Neuron 31:791–807. required for attenuation of clinical disease by an ERβ ligand. Proc Natl Acad Sci USA 43. Káradóttir R, Attwell D (2007) Neurotransmitter receptors in the life and death of 110:19125–19130. oligodendrocytes. Neuroscience 145:1426–1438. 17. Imwalle DB, Gustafsson JA, Rissman EF (2005) Lack of functional estrogen receptor 44. Varea O, et al. (2009) Estradiol activates beta-catenin dependent transcription in beta influences anxiety behavior and serotonin content in female mice. Physiol Behav neurons. PLoS One 4:e5153. 84:157–163. 45. Brooks YS, et al. (2014) Multifactorial ERβ and NOTCH1 control of squamous differ- 18. Rissman EF, Heck AL, Leonard JE, Shupnik MA, Gustafsson JA (2002) Disruption of entiation and cancer. J Clin Invest 124:2260–2276. estrogen receptor beta gene impairs spatial learning in female mice. Proc Natl Acad 46. Vivar OI, et al. (2010) Estrogen receptor beta binds to and regulates three distinct Sci USA 99:3996–4001. classes of target genes. J Biol Chem 285:22059–22066. 19. Liu F, et al. (2008) Activation of estrogen receptor-beta regulates hippocampal syn- 47. Zhao C, et al. (2010) Genome-wide mapping of estrogen receptor-beta-binding re- aptic plasticity and improves memory. Nat Neurosci 11:334–343. gions reveals extensive cross-talk with transcription factor protein-1. Cancer 20. Wilson ME, Westberry JM (2009) Regulation of oestrogen receptor gene expression: Res 70:5174–5183. New insights and novel mechanisms. J Neuroendocrinol 21:238–242. 48. Tremblay A, Tremblay GB, Labrie F, Giguère V (1999) Ligand-independent recruitment 21. Dalla C, Antoniou K, Papadopoulou-Daifoti Z, Balthazart J, Bakker J (2005) Male of SRC-1 to estrogen receptor beta through phosphorylation of activation function aromatase-knockout mice exhibit normal levels of activity, anxiety and “depressive- AF-1. Mol Cell 3:513–519. like” symptomatology. Behav Brain Res 163:186–193. 49. Suzuki H, et al. (2013) Involvement of estrogen receptor β in maintenance of sero- 22. Richardson TI, et al. (2007) Benzopyrans as selective estrogen receptor beta agonists tonergic neurons of the dorsal raphe. Mol Psychiatry 18:674–680. (SERBAs). Part 3: Synthesis of cyclopentanone and cyclohexanone intermediates for 50. Tammimies K, et al. (2013) Molecular networks of DYX1C1 gene show connection to C-ring modification. Bioorg Med Chem Lett 17:4824–4828. neuronal migration genes and cytoskeletal . Biol Psychiatry 73:583–590. 23. Kageyama R, Ohtsuka T, Kobayashi T (2008) Roles of Hes genes in neural develop- 51. Liu JS (2011) Molecular genetics of neuronal migration disorders. Curr Neurol ment. Dev Growth Differ 50(Suppl 1):S97–S103. Neurosci Rep 11:171–178. 24. Kageyama R, Ohtsuka T (1999) The Notch-Hes pathway in mammalian neural de- 52. Potkin SG, et al.; FBIRN (2009) A genome-wide association study of schizophrenia velopment. Cell Res 9:179–188. using brain activation as a quantitative . Schizophr Bull 35:96–108. 25. Alvarez-Fischer D, et al. (2011) Engrailed protects mouse midbrain dopaminergic 53. Varshney M, Nalvarte I (2017) Genes, gender, environment, and novel functions of

neurons against mitochondrial complex I insults. Nat Neurosci 14:1260–1266. estrogen receptor beta in the susceptibility to neurodevelopmental disorders. Brain MEDICAL SCIENCES 26. Simon HH, Saueressig H, Wurst W, Goulding MD, O’Leary DD (2001) Fate of midbrain Sci 7:E24. dopaminergic neurons controlled by the engrailed genes. J Neurosci 21:3126–3134. 54. Lee SH, Lumelsky N, Studer L, Auerbach JM, McKay RD (2000) Efficient generation of 27. Simon HH, Thuret S, Alberi L (2004) Midbrain dopaminergic neurons: Control of their midbrain and hindbrain neurons from mouse embryonic stem cells. Nat Biotechnol cell fate by the engrailed transcription factors. Cell Tissue Res 318:53–61. 18:675–679. 28. Wu C, Sun D (2015) GABA receptors in brain development, function, and injury. 55. Wu WF, et al. (2017) Estrogen receptor beta, a regulator of sig- Metab Brain Dis 30:367–379. naling in the mouse ventral prostate. Proc Natl Acad Sci USA 114:E3816–E3822.

Varshney et al. PNAS | Published online November 13, 2017 | E10437 Downloaded by guest on September 29, 2021