Helios expression coordinates the development of a subset of striatopallidal medium spiny neurons

Here, we unravel the mechanism of action of the Ikaros family zinc finger protein Helios (He) during the development of striatal medium spiny neurons (MSNs). He regulates the second wave of striatal neurogenesis involved in the generation of striatopallidal neurons, which express dopamine 2 receptor and enkephalin. To exert this effect, He is expressed in neural progenitor cells (NPCs) keeping them in the G1/G0 phase of the cell cycle. Thus, a lack of He results in an increase of S-phase entry and S-phase length of NPCs, which in turn impairs striatal neurogenesis and produces an accumulation of the number of cycling NPCs in the germinal zone (GZ), which end up dying at postnatal stages. Therefore, He−/− mice show a reduction in the number of dorso-medial striatal MSNs in the adult that produces deficits in motor skills acquisition. In addition, overexpression of He in NPCs induces misexpression of DARPP-32 when transplanted in mouse striatum. These findings demonstrate that He is involved in the correct development of a subset of striatopallidal MSNs and reveal new cellular mechanisms for neuronal development. Summary: The transcription factor Helios regulates G1-S transition to promote neuronal differentiation of a striatopallidal neuronal subpopulation involved in motor skill acquisition.


INTRODUCTION
The mammalian striatum controls body movements through a sophisticated neuronal network that is dependent on the neurogenesis of two major classes of striatal neurons: the striatal projection neurons (or medium spiny neurons; MSNs) and the interneurons. MSNs are subdivided into two subpopulations: neurons that constitute the direct (or striatonigral) pathway and preferentially express substance P (SP) and D1R (dopamine receptor 1; DRD1), and neurons of the indirect (or striatopallidal) pathway, which mainly express enkephalin (ENK) and D2R (dopamine receptor 2; DRD2) (Gerfen, 1992). These two populations are differentially distributed within the striatal compartments. Striatal patches or striosomes mainly contain SP + MSNs, but both MSN subpopulations, SP + and ENK + , are located in the matrix (Gerfen, 1992).
During embryonic development, radial glial cells (RGCs) from the ventricle wall of the lateral ganglionic eminence (LGE) undergo successive divisions to expand the pool of neural progenitor cells (NPCs), thereby increasing the volume of the germinal zone (subventricular zone; SVZ) (for reviews, see Götz and Barde, 2005;Merkle and Alvarez-Buylla, 2006). At certain developmental stages, NPCs differentiate into immature neurons that migrate radially to the mantle zone (MZ) (Götz and Barde, 2005;Merkle and Alvarez-Buylla, 2006;Mérot et al., 2009). Two waves of striatal neurogenesis segregate MSNs into two principal compartments: the patches, generated during the first neurogenic wave [starting at embryonic day (E) 12.5 in mouse]; and the matrix, developed during late striatal neurogenesis (starting at E14.5 in mouse) (Gerfen, 1992;Mason et al., 2005).
Ikaros family members are transcription factors that play essential roles during lymphocyte development (Cobb and Smale, 2005;Georgopoulos, 2002;Yoshida and Georgopoulos, 2014). Ikaros is the founder member of this family of DNA-binding proteins, which consists of Ikaros, Helios (He), Aiolos, Eos and Pegasus (Ikzf1-5, respectively -Mouse Genome Informatics) (John et al., 2009;Rebollo and Schmitt, 2003;Yoshida and Georgopoulos, 2014). In addition, Ikaros has been implicated in CNS development (Agoston et al., 2007;Alsiö et al., 2013;Martín-Ibáñez et al., 2010). We have recently described that He is also implicated in striatal development (Martín-Ibáñez et al., 2012). Within the LGE, He is expressed from E14.5 to postnatal day (P) 15 in both the GZ and the MZ, and its expression is downstream of Gsx2 and Dlx1/2 (Martín-Ibáñez et al., 2012). However, little is known about mechanisms of action of He during this developmental process.
Here, we demonstrate that He is expressed by NPCs at the G 0 /G 1phase of the cell cycle and induces neuronal differentiation by decreasing the levels of cyclin E and blocking the progression of these NPCs into S phase. Consequently, in the absence of He, proliferating NPCs accumulate in the GZ and the number of Ctip2 + and DARPP-32 (PPP1R1B) + MSNs is reduced in the striatum resulting in disturbance of motor skill learning.
To assess whether He was expressed by proliferative cells in the LGE, we performed double staining for He and Ki67 (Mki67) at E16.5, BrdU or phospho-histone H3 (PH3) at E14.5. Our results showed that He + and Ki67 + areas were mainly coincident at the GZ-MZ border at E16.5 ( Fig. 2A). Within this area, He was expressed by NPCs expressing a low level of Ki67 (Fig. 2B,C) but not by cells expressing a high level of Ki67 + ( Fig. 2D; see Fig. S5 for quantification details). However, there was a lack of colocalization between He and short-pulsed BrdU NPCs (Fig. 2E,F), and He and PH3 + NPCs (Fig. 2G,H). Interestingly, He only colocalized with Ki67-expressing cells during the neurogenic period as we could not observe colocalization from E18.5 onwards (Fig. S6). (F) Representative images of Ctip2 + neurons in the E18.5 (mid-striatal primordium is shown). Scale bar: 120 µm. (G) Quantification of the density and total number of Ctip2 + cells in the whole striatal primordium reveals a significant reduction in He −/− mice compare with wt mice. Results represent the mean± s.e.m. of 4-5 mice per condition. Statistical analysis was performed using Student's t-test; *P<0.05, **P<0.005.
Analysis of the number of cycling cells at different developmental stages in He −/− and wt mice ( Fig. 2I-L) showed that the total number of proliferating cells in the GZ was significantly increased from E14.5 to P3 (Fig. 2I-K), inducing an enlargement of the proliferative area stained with Ki67 ( Fig.  S7). Interestingly, this feature reverted at P7, when the number of proliferating cells in He −/− mice decreased with respect to wt mice ( Fig. 2L; Fig. S8). To analyze whether a specific subpopulation of progenitors was more compromised than others, we counted the percentage of PH3 + basal, subapical and apical progenitors as described by Pilz et al. (2013) (Fig. S9A,B). No differences were found between He −/− and wt mice (Fig. S9B). We also analyzed by QPCR the expression of striatal progenitor markers at E16.5. No differences were found in the levels of mRNA for these markers in He −/− compared with wt mice (Fig. S9C).
To elucidate further the role of He in NPC proliferation, we performed loss-of-function (LOF) and gain-of-function (GOF) in vitro studies using a neurosphere assay (Fig. S10). There was an increase in the number of proliferating cells in the absence of He (Fig. S10A,C,E,F). Accordingly, He overexpression significantly reduced the number of proliferating NPCs with respect to the control eGFP overexpressing NPCs (Fig. S10B,D). In addition, in the absence of He, NPCs were less prone to differentiate to β-III-tubulin + neurons (Fig. S10H). In contrast, an increase in the number of neurons was observed after He overexpression ( Fig. S10I-K). Interestingly, He did not exert any change in the percentage of GFAP + cells in the LOF or in the GOF experiments (Fig. S10H,I). Consequently, He −/− mice did not present any defects in astrocyte differentiation compared with wt mice (Fig. S11A-D). In fact, we did not observe colocalization between He and GFAP (Fig. S11E). High magnification picture shows that although He + and BrdU + cells are located in the same area, they do not colocalize. (F) Unfilled arrowheads indicate BrdU + cells that have recently entered S phase as shown by the appearance of transcription units; filled arrowheads indicate cells that incorporated BrdU at more advanced cell cycle stages. Scale bars: 50 µm. (G,H) There is no coincidence between He-expressing cells and cells in M phase as detected by PH3 staining; low (G) and high (H) magnification images of DMS are shown. Scale bars: 50 µm. (I-L) Quantification of the total number of proliferating cells in the whole GZ show that lack of He induces a significant increase at E14.5 (I), E16.5 (J) and P3 (K) and a significant decrease at P7 (L) compared with wt mice. Results represent the mean±s.e.m. of 5-7 mice per condition. Statistical analysis was performed using Student's t-test; *P<0.05, ***P<0.001.

He controls proliferation through regulation of the G 1 -S checkpoint
To understand the cellular mechanism by which He regulates NPC proliferation and neurogenesis, we next analyzed the cell cycle. We observed that lack of He induced a significant increase in NPC Sphase length that, in turn, increased cell cycle length as measured by an accumulative exposure to BrdU (see Materials and Methods; Lange et al., 2009) (Fig. 3A,C). However, no differences were observed between the length of the G 2 /M phases in NPCs derived from He −/− compared with wt mice, as determined by analysis of the mitotic BrdU labeling index as described previously (Takahashi et al., 1995) (Fig. 3B,C; Fig. S12). Representation of the percentage of cell cycle phases respect to the total cell cycle length clearly demonstrated an elongation of S-phase length when He was knocked down (Fig. 3C). Consistently, He overexpression induced a severe reduction of S-phase length (GOF; Fig. 3D). Our results also showed that in the absence of He more NPCs entered S phase ( punctate BrdU + /EdU + ; Fig In order to demonstrate the mechanism by which He controls Sphase entry, we next analyzed the protein levels of cyclin E (Fig. 4), a key regulator of the transition from G 1 to S phase (Ohtsubo et al., 1995). NPCs derived from He −/− mice presented increased levels of PCNA, a marker of cell proliferation, and cyclin E ( Fig. 4A-D). Accordingly, He overexpression ( Fig. 4E-H) produced a reduction of PCNA and cyclin E protein levels ( Fig. 4E-H), and a drastic reduction of cyclin E mRNA levels (Fig. 4J). Similarly, in vivo analysis showed that an increased number of NPCs had entered into S phase in the GZ of He −/− compared with wt mice (Fig. 4K), which was accompanied by increased protein levels of cyclin E in the LGE (Fig. 4L,M). Chromatin immunoprecipitation experiments performed by Kim and co-workers (Kim et al., 2015) demonstrated that He binds the cyclin E gene (Ccne1) promoter site and another site downstream of the gene (Fig. 4N). However, no changes of the two cyclin E regulators E2F1 and retinoblastoma (Rb; Rb1) (Harbour, 2000;Ohtani et al., 1995) were observed in NPCs derived from He −/− mice (Fig. S14). Altogether, these results suggest that He might control cell cycle progression through regulation of cyclin E expression.
Postnatal cell death is increased in He −/− mice We next investigated whether cell death was altered in the absence of He during embryonic and postnatal stages. Cleaved caspase-3 immunohistochemistry did not reveal any differences between He −/− and wt mice at embryonic stages (E14.5, E16.5 and E18.5; data not shown). However, a significant increase in the number of apoptotic cells was detected in the GZ and the MZ at P3 in He −/− mice (  He is necessary for MSN development We next characterized the striatum of He −/− adult mice. First, we studied brain hemisphere volume and detected a slight decrease in He −/− mice compared with wt mice (Fig. S16A,C; 8.36% decrease). Interestingly, characterization of striatal volume revealed a larger and significant reduction in He −/− compared with wt mice (Fig. S16B,C; 20.17% decrease). The ratio of striatal versus hemisphere volume showed that striatal volume is selectively disturbed in He −/− mice (wt, 18.23±0.79%; He −/− , 15.45±0.60%), showing a 15.24% reduction of relative striatal volume. Stereological analysis of calbindin + and DARPP-32 + neurons revealed a significant decrease in the density (Fig. S16D,E,H,I) and total number of MSNs in the striatum of He −/− compared with wt mice (Fig. 6A,B). We also analyzed the density of DARPP-32 + neurons in different striatal areas including the dorso-medial striatum (DMS), dorso-lateral striatum (DLS), ventro-medial striatum (VMS) and ventro-lateral striatum (VLS) (Fig. 6K). These experiments demonstrated a significant decrease only in the DMS in He −/− mice compared with wt mice (Fig. 6E-H). Interestingly, a specific alteration of the ENK + population was also observed in the DMS in the absence of He (Fig. 6I). However, no differences were found for the SP + population in He −/− mice compared with wt mice (Fig. 6J). In addition, no differences were observed between genotypes in the cholinergic and parvalbumin + striatal interneurons (Fig. S16F,G; Fig. 6C,D).
In order to study the direct involvement of He in the acquisition of a mature MSN phenotype, we transplanted eGFP or Heoverexpressing NPCs into the mouse neonatal forebrain (Fig. 7A). Compared with control cells, He-overexpressing cells displayed more robust branching 2 weeks post-transplantation (total neurite tree length per neuron: GFP 168.13± 21.92 µm, He 413.66±98.84 µm, P=0.0046; number of branches per neuron: GFP 14.43±1.68, He 24.89±4.08, P=0.0089; Fig. 7B-E) and DARPP-32 expression was observed in few scattered cells adjacent to the striatum (Fig. 7G,H). Four weeks post-transplantation, several He-overexpressing cells displayed DARPP-32 expression (Fig. 7J-L), in contrast to control cells, which were all DARPP-32 negative (Fig. 7I). Quantification of DARPP-32 + neurons in GFP transplanted cells demonstrated a 150fold increase in the number of double-stained cells in He-expressing cells compared with controls. In addition, He overexpression in striatal primary cultures significantly increased the number of calbindin + , DARPP-32 + and ENK + cells (Fig. S17).

He loss disturbs the acquisition of motor skills
To analyze the functional implication of He loss, we assessed the performance of motor tasks in wt and He −/− mice (Fig. 8). In the simple swimming test, He −/− mice displayed significant abnormalities compared with wt mice in their swimming latency in the first testing trial (genotype: F 2,162 =4.08, P<0.05; post-hoc trial 1: P<0.01), but these disappeared over subsequent trials (Fig. 8A).
In the rotarod test, all mice reached a stable level of performance within six trials (Fig. 8C), as measured by a decrease in the number of falls in 60 s per mouse (testing trial F 5,138 =15.87, P<0.01). However, acquisition on the rotarod task was significantly delayed in He −/− compared with wt mice (genotype F 2,138 =21.03, P<0.01).

DISCUSSION
Striatal MSNs are generated from NPCs located at the GZ of the LGE. Here, we show that He regulates late striatal neurogenesis that gives rise to D2R + ENK neurons. He is expressed by NPCs in the G 1 /G 0 cell cycle phase at the GZ, impairing the G 1 -S transition by the regulation of cyclin E, which in turn induces neuronal differentiation. Consequently, lack of He produces an extended S phase and cell cycle length that increases the number of proliferating NPCs at the GZ. At the beginning of the postnatal period, the number of these NPCs is reduced due to their late aberrant neurogenesis that results in cell death. These abnormalities of embryonic development in He −/− mice produce a reduction of a specific subset of striatopallidal neurons of the dorsomedial striatum that control motor skill learning.
He is necessary for striatopallidal neurogenesis NPCs located at the GZ of the LGE become postmitotic and migrate into the MZ to acquire the MSN phenotype (Brazel et al., 2003). We have previously proposed a model for the development of striatal subpopulations in which Ikaros and He are involved in the development of striatopallidal ENK + matrix MSNs (Martín-Ibáñez et al., 2012). This hypothesis is reinforced by the localization of He in ENK + neurons that co-express D2R ( present results). Besides the apparent similar function between He and Ikaros on ENK + neurogenesis, there is much evidence that they determine different ENK + subpopulations. They are expressed by different cells (Martín-Ibáñez et al., 2012), and their expression is not modified in the reciprocal knockout mice (Martín-Ibáñez et al., 2010. These results are contrary to the role of Ikaros family members in the hematopoietic system where they directly interact (Hahm et al., 1998;John et al., 2009), suggesting specific mechanisms of action in each system.

He regulates neurogenesis through the control of the G1-S phase checkpoint
Gsx2 + radial glial cells constitute the first NPCs that appear during LGE ontogeny, which differentiate with the onset of the neurogenesis from the neuroephithelial cells (for a review, see Dimou and Götz, 2014). He-expressing cells are derived from radial glial cells, as its expression disappears in Gsx2 knockout mice (Martín-Ibáñez et al., 2012). However, He loss does not compromise the number of the radial glial cell subtypes described elsewhere (Pilz et al., 2013). Radial glial cells generate the large MSNs output by a series of intermediate NPCs to amplify specific lineages, although these striatal NPCs are still poorly characterized. He is expressed by a small number of NPCs distributed in deep SVZ. Although the localization of He is mainly at the dorsal areas, it does not seem to be defining a specific SVZ domain as it has been described for other transcription factors in the VZ (Flames et al., 2007).
Some of the NPCs that express He at the GZ co-express low levels of Ki67. Considering that Ki67 labels cells during all phases of the cell cycle except G 0 (Kanthan et al., 2010;Scholzen and Gerdes, 2000) and that G 1 is the cell cycle phase with lower Ki67 expression levels (Lopez et al., 1991), we hypothesized that He is expressed in a subset of NPCs during G 1 and G 0 phases. The lack of colocalization between He and BrdU or PH3 reinforces the idea that He is not expressed by cells at S or M phases, respectively. Within G 1 phase He impairs S-phase entry, reducing S-phase length and arresting NPCs at G 1 /G 0 phase to facilitate neuronal differentiation. Consequently, He −/− mice NPCs increase S-phase entry and continue proliferating in the striatal GZ impairing neurogenesis (see Fig. S18 for a representative scheme). Similarly, Lacomme and co-workers demonstrated that Ngn2 regulates G 1 -S phase transition, blocking S-phase entry and increasing the number of NPCs at G 1 /G 0 phase (Lacomme et al., 2012). In addition, NPCs shorten S phase on commitment to neuron production (Arai et al., 2011;Turrero García et al., 2015). Thus, cell cycle length and G 1 -S phase transition are crucial processes for neurogenesis and both are regulated by He. We hypothesize that He arrests LGE-derived NPCs into phases G 1 /G 0 to allow the accumulation of the protein machinery necessary for their differentiation to specific striatal neurons. In fact, crucial aspects of neural commitment are acquired in the final division cycle of NPCs. For example, the cortical laminar fate of NPC is acquired during the final progenitor cell division (Bohner et al., 1997;Edlund and Jessell, 1999;McConnell and Kaznowski, 1991). Similarly, during motor neuron development, NPCs become sonic hedgehog (Shh) dependent late in their final progenitor cell cycle (Ericson et al., 1996), which commits them to a motor neuronal fate (Tanabe et al., 1998).
G 1 -S phase transition is regulated by Cdk2 and cyclin E, which form a complex that participates in G 1 -S phase checkpoint (reviewed by Hardwick and Philpott, 2014;Ohtsubo and Roberts, 1993). Our results suggest that cyclin E is a key factor regulated by He that correlates with the G 1 -S phase transition impairment observed in the He −/− mice. In fact, the cyclin E gene (Ccne1) has two very strong He-binding domains (Kim et al., 2015) suggesting a direct regulation. Similar to our results, Pilaz and colleagues described that overexpression of cyclin E in cortical NPCs promotes a proliferation increase whereas downregulation of cyclin E led to a decrease in progenitor proliferation (Pilaz et al., 2009). In addition, a direct correlation between cyclin E and S-phase entry was proposed by ectopic expression of cyclin E, which shortens the G 1 interval and increases the length of S phase by advancing G 1 -S phase transition (Resnitzky et al., 1994). Furthermore, ectopic expression of cyclin E can drive G 1 cells into S phase under conditions in which Rb is not phosphorylated and E2F is not activated (Leng et al., 1997;Lukas et al., 1997). This is in agreement with our results, as we observed an increase in cyclin E but no alterations in phosphorylated RB or E2F in He −/− mice.

He loss increases postnatal cell death
The homeostasis of NPCs in the striatum is a regulated process in which neurogenesis precedes astro-gliogenesis during development (Alvarez-Buylla et al., 2001;Ninkovic and Götz, 2013). However, contrary to the increase of astro-gliogenesis observed in Ikaros −/− mice (Martín-Ibáñez et al., 2010), we could not detect any effects on glial cells in He −/− mice. The role of He in neurogenesis through cyclin E-mediated G1-S transition without modifying astro-gliogenesis coincides with the effect of deferoxamine, a G 1 / S-phase blocker, which increases neuronal but not astrocytic NPC differentiation (Kim et al., 2006;Misumi et al., 2008).
The reduction of NPCs in He −/− mice at postnatal stages can be related to the increase in cell death during this period. Naturally occurring cell death is a crucial step in re-defining the final size of specific neuronal populations (Burek and Oppenheim, 1996;Kristiansen and Ham, 2014), which directly correlates with the time of prior exit from cell cycle and position during neuronal development (Gould et al., 1999). Our results point to the idea that the cell death observed in He −/− mice is a consequence of the delay in NPCs exiting cell cycle around E18.5, then migrating into the MZ where they become neurons and die. Therefore, lack of He produces a dysfunction in the time and position of late-generated neurons in the MZ. Dual effects have also been described for Isl1 and Ebf1, which promote differentiation of striatonigral neurons and in their absence striatal cell death is observed (Garel et al., 1999;Lu et al., 2014). Taken together, all these results indicate that He loss causes aberrant neurogenesis, which in turn induces neuronal cell death compromising striatal development.

He participates in the differentiation of a subset of MSNs that is involved in early motor learning
He-mediated regulation of the NPC cell cycle correlates with the determination of a subset of striatopallidal MSNs. The events occurring during striatal development of He −/− mice cause a specific reduction of striatal MSNs in the DMS in the adulthood. Taken together, our present findings demonstrate that He plays a direct role in the commitment of NPCs to MSNs. Accordingly, He overexpression is sufficient to differentiate NPCs transplanted into the striatum in MSNs expressing DARPP-32. Previously published works and reviews suggest that striatal motor function is involved with habit formation (Yin and Knowlton, 2006) and procedural learning (Kreitzer, 2009), which fits with what we see in our He −/− mice. The striatum has been classically divided into dorsal and ventral areas, the dorsal being the most involved in motor behavior (Durieux et al., 2012). Accumulating evidence shows anatomical and functional differences in the striatum between the external DLS and the internal DMS (Durieux et al., 2012;Graybiel, 2008;Voorn et al., 2004). Interestingly, the DMS is involved in the initial stages of motor skill learning (Jueptner and Weiller, 1998;Luft and Buitrago, 2005), whereas the DLS is required for progressive skill automatization and habit learning (Miyachi et al., 2002;Yin et al., 2004). In addition, it has been shown that the loss of D2R + neurons in the DMS produces early motor learning impairment but the animals can improve their performances to reach control levels (Durieux et al., 2012). As He −/− mice show impairments in the acquisition of motor skills, it seems plausible that He is involved in the generation of a specific subpopulation of striatopallidal D2R + MSNs in the DMS. The cerebellum is also involved in fine-tuning the motor agility found in procedural skills. Cerebellar lesions or dysfunctions produce permanent deficits in motor tasks. However, diseased animals never perform motor tasks as well as their control or wt littermates (Sausbier et al., 2004;Stroobants et al., 2013;Vinueza Veloz et al., 2012). As He −/− mice show problems in the acquisition but not the execution of motor skills it seems that an association with cerebellar deficits is not likely.

Conclusion
In conclusion, our results demonstrate a novel mechanism for He in the development of striatopallidal MSNs of the DMS that controls motor skills learning. He exerts its main effects on the commitment of NPCs to MSNs through the regulation of the G 1 -S phase transition and arrests NPCs at G 1 phase to induce neuronal differentiation. The alterations of this mechanism observed in He −/− mice produce aberrant neurogenesis leading to the cell death of late-generated neurons.

Birthdating, proliferation and tracking experiments in vivo
Birthdating experiments were performed as described elsewhere (Fig. 1B;Martín-Ibáñez et al., 2010). To study the generation of He + cells, injections of EdU (50 mg/kg) at E13.5 or E14.5, or BrdU at E16.5 into wt pregnant mice were performed and allowed to develop until E18.5, when embryos were processed for He and BrdU immunohistochemistry or EdU detection (Life Technologies) (Fig. S4A).
To analyze in vivo proliferation in the GZ, E14.5 pregnant mice received a single dose of EdU (50 mg/kg). The proliferation analysis of E16.5, P3 and P7 was performed by Ki67 immunohistochemistry.
In order to track the origin of dead cells in the MZ, a pulse of EdU (50 mg/kg) was performed at E18.5, and immunohistochemistry was performed at P3 against EdU and cleaved caspase 3 (Cell Signaling Technology), nestin, GFAP or NeuN (Fig. 5G).
To study whether the lack of He could alter the cells entering the S phase of the cell cycle, we performed in vivo experiments with He −/− and wt mice as previously described (Lange et al., 2009) (Fig. 4K).
For further details of these methods, see the supplementary Materials and Methods.

Production of viral particles and cell transduction
To overexpress He, NPCs were transduced with the pLV-HE-IRES-eGFP plasmid or the pLV-IRES-eGFP plasmid, which encode human HE and eGFP or eGFP alone, respectively. For further details of viral particle production, see the supplementary Materials and Methods.

Neurosphere assay
LGEs from E14.5 wt or He −/− mice were dissected out and mechanically disaggregated to culture as neurosphere and differentiate to neural cells as described previously (Martín-Ibáñez et al., 2010). For further details of neurosphere cultures, see the supplementary Materials and Methods.
Loss-of-function (LOF) experiments were performed with neurospheres derived from He −/− mice whereas gain-of-function (GOF) experiments were performed by overexpressing He. The number of neurons (β-III-tubulin + ) and astrocytes (GFAP + ) were analyzed after 6 days of differentiation.

Cell cycle analysis in vitro Proliferation assays
BrdU incorporation assays were performed in wt and He −/− mice-derived neurospheres (LOF) and neurospheres overexpressing He (GOF) as described elsewhere (Urbán et al., 2010). The number of Ki67 + cells was also analyzed in wt and He −/− mice-derived neurospheres (LOF) and neurospheres overexpressing He (GOF).

Cell cycle length
An accumulative exposure to 1 µM BrdU over 36 h was performed in wt and He −/− mice-derived neurospheres (LOF) and in neurospheres overexpressing He (GOF) after 2 DIV in proliferation. Cells were fixed at different time points after 1 µM BrdU exposure (1, 3, 6, 12, 24 and 36 h) and processed for BrdU immunocytochemistry. Following regression analysis as previously described by Takahashi et al. (1992Takahashi et al. ( , 1995, the length of the cell cycle and the length of the S phase were calculated for the NPCs.

S-phase analysis
To study the cells entering and exiting the S phase of the cell cycle, we performed in vitro experiments with neurospheres derived from He −/− and wt mice as described previously (Lange et al., 2009) (Fig. 3E-H).

G 2 /M phase labeling
To study the combined length of the G 2 /M phases, an accumulative exposure to 1 µM BrdU over 5 h was performed after 2 DIV in proliferation to analyze the mitotic BrdU labeling index as described previously (Takahashi et al., 1995).

Cell cycle index
We analyzed cell cycle index as the number of cells that incorporate BrdU but leave the cell cycle (i.e. abandoned the G 1 -S-G 2 /M phases and entered into G 0 ) as previously described (Urbán et al., 2010) (Fig. S13).

Discerning high and low Ki67-expressing cells
We detected cells expressing high and low levels of Ki67 using the automatic intensity detection of the Cell Profiler software.
For further details of cell cycle analyses, see the supplementary Materials and Methods.

Analysis of He-binding sites at the Ccne1 promoter
We obtained and analyzed the Big Wig file deposited in Gene Expression Omnibus by Kim et al. (2015), and visualized it in the Integrative Genome Viewer with the files provided aligned to the Ensembl Mouse Genome. Details of database used can be found in the supplementary Materials and Methods.

Western blots
We performed western blot analyses for cyclin E and PCNA as described elsewhere  in wt and He −/− mice-derived neurospheres (LOF) and neurospheres overexpressing He (GOF). E2F1 and retinoblastoma (Rb) were detected in LOF experiments. For further details of western blot procedure, see the supplementary Materials and Methods.

In situ hybridization
To assess which striatal subpopulation of MSNs express He, we performed double in situ hybridization for ENK or tachykinin A (Tac1, a precursor of SP), the precursor of SP, and immunohistochemistry for He as described previously (Martín-Ibáñez et al., 2010). For further details of in situ procedures, see the supplementary Materials and Methods.

Immunolabeling
For histological preparations, embryonic or postnatal brains were removed at specific developmental stages and were frozen in dry ice-cooled methylbutane or cryoprotected depending on the procedure. Immunolabeling was performed according to the protocols described by Bosch et al. (2004) and Canals et al. (2004). For further details of the antibodies used and immunostaining procedures, see the supplementary Materials and Methods.

Measurement of volumes and in vivo cell counts
The volumes of certain brain regions were measured using ImageJ v1.33 as described previously . All cell counts [EdU and Ki67 for GZ proliferation; BrdU for birthdating experiments; cleaved caspase 3 for cell death; Ctip2, calbindin, DARPP-32, choline acetyl transferase (ChAT) and parvalbumin for striatal cell population] were performed blind to genotype. Unbiased stereological counts were performed for all striatal areas for each animal. DMS, DLS, VMS and VLS were specifically counted for DARPP-32-, ENK-and SP-positive cells.
The distribution of mitosis in He −/− and wt striatum at E16.5 was analyzed as described by Pilz et al. (2013) and counted using CAST software.
Automated quantification of branches, and neurite length was performed using Cell Profiler v2.8 software.
For further details of cell counts, see the supplementary Materials and Methods.

Q-PCR
Gene expression was evaluated by Q-PCR assays as previously described by Martín-Ibáñez et al. (2010). For further details of the probes used and PCR procedures, see the supplementary Materials and Methods.
Primary striatal culture and transfection E14.5 fetal LGEs were dissected and cultured as previously described (Martín-Ibáñez et al., 2010). For He overexpression studies, cells were transfected with the MSCV-He-IRES-eGFP plasmid, or with the MSCV-IRES-eGFP plasmid as a control (Zhang et al., 2007). We counted the number of He or eGFP overexpressing cells that colocalized with calbindin, DARPP-32 or ENK. For further details of primary culture methods, see the supplementary Materials and Methods.

Cell transplants
Unilateral striatal injections of He-overexpressing cells were performed using a stereotaxic apparatus (Davis Kopf Instruments, Tujunga, CA, USA); coordinates (mm): AP, +2.3, L, +1.4 from lambda and DV, −1.8 from dura. For further details of cell transplants, see the supplementary Materials and Methods.

Mouse behavior Swimming task
The mice were placed at the end of a transparent perspex extended swimming tank facing away from a visible escape platform at one end of the tank and the time taken to reach the platform was recorded.

Balance beam
Animals were allowed to walk along a horizontally placed beam of a long steel cylinder (50 cm) with 15 mm diameter. Latency to fall and number of falls were measured.

Rotarod
Acquisition of a motor coordination task was further evaluated on the rotarod apparatus (24 rpm). Latency to fall and the number of falls during 60 s was recorded.
For further details of mouse behavior analyses, see the supplementary Materials and Methods.

Statistical analysis
All results are expressed as the mean of independent experiments±s.e.m. Results were analyzed using Student's t-test or one-way or two-way ANOVA, followed by the Bonferroni post-hoc test.