Anterior-posterior graded response to Otx2 controls proliferation and differentiation of dopaminergic progenitors in the ventral mesencephalon

The dopaminergic (DA) neurons of the ventral tegmental area (VTA),substantia nigra (SN) and retrorubral field (RRF) represent a specialized population of neurons that originate in the floor-plate region of the mesencephalon and anterior prosomeres (P1-P3)(Hökfelt et al., 1984; Björklund and Lindvall,1984; Smidt and Burbach,2007; Smits et al.,2006; Marín et al.,2005). These neurons play a crucial role in the control of motor,sensorimotor and motivated behaviors(Björklund and Lindvall,1984; Kelley and Berridge,2002; Isacson,1993). Degeneration of the DA neurons in the SN leads to the characteristic symptoms of Parkinson's disease (PD), whereas abnormal functioning of those in the VTA is involved in psychiatric and affective dysfunction (Jellinger, 2001; Egan and Weinberger, 1997; von Bohlen und Halbach et al.,2004). These disorders highlight the relevance of this population of neurons and the enormous effort to understand the molecular basis that controls their neurogenesis.

During development, neural cell-type specification is controlled by inducing signals and depends on the responsiveness of target progenitor cells(Lumsden and Krumlauf, 1996; Wurst and Bally-Cuif, 2001; Jessell, 2000; Briscoe and Ericson, 2001). Mesencephalic DA (mesDA) progenitors require the inducing properties of Fgf8 and Shh signals that are emitted at the isthmic organizer and floor-plate region, respectively, whereas competence in interpreting these inducing signals is provided by the molecular code that is defined by the expression of different transcription factors (Briscoe and Ericson, 2001; Ye et al.,1998; Simeone,2005; Smidt and Burbach,2007; Smits et al.,2006; Prakash and Wurst,2006; Liu and Joyner,2001). Several genes, including the transcription factors Pitx3,Lmx1a, Lmx1b, En, Msx1, Foxa2, Ngn2, Otx2, the orphan nuclear receptor Nurr1,the Wnt1 and Wnt5a members of the Wnt family, and retinoic acid play a relevant role in differentiation and survival of mesDA neurons(Saucedo-Cardenas et al.,1998; Semina et al.,1998; Smidt et al.,1997; Smidt et al.,2000; Smidt et al.,2004; Kele et al.,2006; Ferri et al.,2007; Andersson et al.,2006a; Andersson et al.,2006b; Ono et al.,2007; McCaffery and Dräger, 1994; van den Munckhof et al., 2003; Nunes et al., 2003; Simon et al.,2001; Zetterstrom et al.,1997; Castelo-Branco et al.,2003; Puelles et al.,2003; Puelles et al.,2004; Prakash et al.,2006). Among these, our attention has been focused on Otx2, a transcription factor containing a bicoid-like homeodomain that plays a crucial role in specification, regionalization and differentiation of forebrain and midbrain (Puelles et al.,2006; Acampora and Simeone,1999; Simeone et al.,2002; Simeone,2005). Accordingly, previous studies have shown that: (1) Otx2 together with Otx1 is required to control the positioning of Shh and Fgf8 expression, and that failure in this control generates profound alteration in the identity code of progenitor domains in the ventral mesencephalon (VM), which exhibits a dorsal expansion of the mesDA compartment at the expense of the red nucleus (RN) domain(Puelles et al., 2003); (2)Otx2 is required in the VM to suppress the transcription factor Nkx2.2 and prevents the generation of 5-HT-containing neurons in place of RN and dorsalmost mesDA neurons (Puelles et al.,2004); and (3) Wnt1 and Otx2 may be engaged in a positive-feedback loop that, in turn, is required for proper development of mesDA neurons(Prakash et al., 2006). However, although these studies have indicated that Otx2 plays a relevant role in the establishment of the mesencephalic field and molecular code of the entire VM, they have not provided direct evidence on whether Otx2 is intrinsically required for mesDA neurogenesis. Here, we have investigated this crucial aspect by analyzing mutant mice that conditionally overexpress or lack Otx2 using En1-driven Cre recombinase. Our findings indicate that Otx2 controls selectively the proliferating activity of mesDA progenitors through a dose-dependent anterior-posterior (AP) graded effect and regulates the differentiation of mesDA neurons by inducing the expression of crucial determinants required to promote the transition of Sox2⁺ mesDA progenitors into Nurr1⁺ immature mesDA neurons. In general, these findings provide new insights into the regulatory mechanism that controls mesDA neurogenesis.

For the inducible overexpression of Otx2, we designed a molecule where the expression of the full-coding cDNA (49bp of 5′UTR-coding-204bp of 3′UTR), driven by the CMV enhancer-chicken β actin promoter, was blocked by a removable loxP-neo-triple polyA-loxP `stop' cassette. In addition, an ires-GFP sequence was introduced downstream of the Otx2 cDNA to follow the transgene expression in vivo. The linearized construct was electroporated into E14Tg4a2 ES cells. DNAs extracted from 30 randomly integrated G418 resistant clones were digested with EcoRI and BamHI restriction enzymes, and hybridized with two probes(Neo and GFP) localized across the two restriction sites(Fig. 1A). Only clones detecting single EcoRI and BamHI bands longer than the minimum expected length were chosen for further analysis. Then, in order to detect potential disruption of any gene functions by the random integration of the construct, and for genotyping, we identified the chromosomal site and sequences flanking the transgene insertion in several ES clones by using the Vectorette System (Sigma). We chose, as the best candidate to generate a mouse line, a clone in which the transgenic and blocked Otx2 overexpressing(tOtx2^bov) construct was integrated in an intergenic region of chromosome 7 D2, 51 kb downstream of the Isg20 gene and 111 kb upstream of the Agc1 gene (Fig. 1A). To test the functionality and efficacy of the transgene expression, we re-transfected this clone to target the CreER-recombinase in the Rosa26 (R26) locus.

The tOtx2^bov/+; R26^CreER/+ ES clone was tested for excision of the loxP-Neo-triple polyA-loxP cassette by using specific pairs of primers (black arrows in Fig. 1A and upper panel in Fig. 1B). The PCR products were 260 bp and 163 bp long for the excised and non-excised allele, respectively(Fig. 1B).

The tOtx2^bov/+; R26^CreER/+ ES clone was also tested to quantify the transgenic Otx2 and GFP gene products by western blot of total protein extracts of ES cells treated 24 hours with 100 nM of 4-hydroxytamoxifen (Sigma) (Fig. 1B). Based on these tests, the original tOtx2^bov/+ ES cell clone was injected into C57 blastocysts and chimeras were mated for germline transmission.

Litters were genotyped by using specific primers (open arrows in Fig. 1A). Primers sequences used in this study are available upon request. The En1^Cre/+;Otx2^flox/flox mouse model to study Otx2 inactivation has been previously described (Puelles et al.,2004).

In situ hybridization on sections and whole embryos was performed as described (Simeone, 1999; Puelles et al., 2006) with GFP, Msx1, Lmx1a probes, each corresponding to a PCR fragment of variable length between 300 bp and 500 bp or with Wnt1, En1, Otx1,Gbx2 and Fgf8 probes(Puelles et al., 2004). Immunohistochemistry was performed as previously reported(Puelles et al., 2004; Prakash et al., 2006). Details of primary antibodies can be provided on request.

Cell counting of TH⁺ neurons was performed on eight E18.5 dissected brains for each genotype (wild type, En1^Cre/+,tOtx2^bov/+, En1^Cre/+; tOtx2^ov/+ and En1^Cre/+; tOtx2^ov/ov). TH cell counting was also confirmed in three brains for each genotype by immunostaining with Pitx3(data not shown). The procedure adopted for selection of sections and TH⁺ cell-counting can be provided on request.

To determine the labeling index (LI), pregnant females at E10.5 or E11.5 were intraperitoneally injected with 5′-bromo 2′-deoxyuridine(BrdU) at a concentration of 50 mg/kg of body weight and embryos collected 30 minutes after injection. To determine the Quit fraction (Qf), BrdU was administered as for LI experiments, but embryos were collected about 24 hours later. Further information on these experiments can be provided on request.

To assess the role of Otx2 in mesDA neurogenesis, we studied first the effect of Otx2 overexpression in the VM. To achieve this, we generated a transgenic mouse model that conditionally activates Otx2 by En1-driven Cre activity (see Materials and methods). The construct that conditionally overexpresses Otx2 (Fig. 1A) was transfected into ES cells to select the tOtx2^bov/+ clone that fulfils the following two requirements: containing a single copy of the construct inserted in a bona fide neutral genomic region; and overexpressing a moderate level of Otx2. The selected tOtx2^bov/+ ES cell clone contained a single insertion of the construct in an intergenic area of chromosome 7 at the chr7D2 position, 51 kb downstream of the Isg20 gene and 111 kb upstream of the Agc1 gene (Fig. 1A), and showed a moderate increase of about twofold of the Otx2 protein expressed in ES cells (Fig. 1B; see Materials and methods). In this assay, however, the level of GFP was lower than that expected on the basis of its mRNA(Fig. 1B; data not shown). This low level of GFP protein was also confirmed in vivo and not further investigated (data not shown). Based on these tests, the original tOtx2^bov/+ ES cell clone was used to generate the tOtx2^bov/+ strain. tOtx2^bov/bov mice were healthy and fertile, and did not exhibit abnormalities. Next, we analyzed whether, at E10.5 in En1^Cre/+; tOtx2^ov/+ and En1^Cre/+; tOtx2^ov/ov embryos, the tOtx2^bov transgene was properly activated by the En1-driven Cre recombinase. As revealed by Otx2 immunostaining and GFP transcription, the transgenic Otx2 protein was detected in the mesencephalon and the rostral hindbrain(Fig. 1C-K). Indeed, the GFP mRNA and, consequently, the transgenic Otx2 were activated also in those territories such as the posterior ventral pretectum and anterior mesencephalon, where En1-driven Cre recombinase and the endogenous En1 gene were not expressed at E10.5(Fig. 1C-K; see Fig. S1A-F″ in the supplementary material). This indicated that the activation of the transgenic allele occurred earlier, when En1expression was broader than at E10.5. Indeed, at E9.2, as revealed by GFP transcripts, the Otx2 overexpressing allele was already activated in early progenitors expressing En1 (see Fig. S1G-I″ in the supplementary material).

Next, we analyzed whether the Otx2 overexpression affected the integrity of the midbrain-hindbrain boundary (MHB) region by monitoring the expression of Otx1, Fgf8 and Gbx2. Compared with E10.5 and E12.5 tOtx2^bov/+ embryos, the expression of these three genes was essentially unaffected in En1^Cre/+;tOtx2^ov/+ embryos (see Fig. S2A-J in the supplementary material), whereas in En1^Cre/+; tOtx2^ov/ov, the posterior border of the Otx1 and Fgf8 expression domains was moderately expanded in the anterior hindbrain at E10.5 (see Fig. S2K-M in the supplementary material). However, at E12.5 in En1^Cre/+;tOtx2^ov/ov mutants, these MHB abnormalities appeared much less severe (see Fig. S2N,O in the supplementary material). Thus these data indicate that the tOtx2^bov allele was properly activated by En1-driven Cre activity and that, probably owing to the moderate level of Otx2 activation, the integrity of the MHB was not remarkably affected. A more severe phenotype was instead observed in the cerebellum at E18.5 (data not shown).

Analysis of mesDA neurons was performed at four AP anatomical levels in E18.5 wild-type, tOtx2^bov/+, En1^Cre/+ and Otx2 overexpressing mice (n=8 per genotype). No difference in mesDA neurons was detected among wild-type, tOtx2^bov/+ and En1^Cre/+ embryos (data not shown)(Sonnier et al., 2007),whereas in En1^Cre/+; tOtx2^ov/+ embryos, the number of TH⁺ neurons showed a graded AP increase(Fig. 2E-H,M-P). In particular,compared with control embryos, in En1^Cre/+;tOtx2^ov/+ mutants, increases of about 20%, 35% and 50% of TH⁺ neurons were detected, respectively, in the VTA of the anterior, intermediate and posterior mesencephalon. We then studied the phenotype of En1^Cre/+; tOtx2^ov/ov mutants(n=8) to investigate whether, compared with En1^Cre/+;tOtx2^ov/+ mutants, a more robust overexpression of Otx2 correlated with a higher number of TH⁺ neurons. We found that,compared with control brains (Fig. 2A-D), En1^Cre/+; tOtx2^ov/ov mutants generated up to about 80 and 100% more TH⁺ neurons in the intermediate and posterior mesencephalon, respectively; and up to 25% and 40%more TH⁺ neurons in the SNpc and anterior VTA, respectively(Fig. 2I-P). These findings were also confirmed by counting Pitx3⁺ cells (data not shown).

Next, we analyzed whether this AP differential response to Otx2 overexpression was detected in meso-diencephalic DA (mdDA) progenitors and in early post-mitotic mdDA neurons by analyzing in wild type and Otx2 overexpressing embryos the expression of Foxa2, Nkx6.1, Otx2 and Nkx2.2 at E10.75 and that of Pitx3, Nkx6.1, Otx2 and Nkx2.2 at E12.5. At E10.75, we analyzed whether the relative extent of the mdDA domain(Foxa2⁺-Nkx6.1^-) was gradually expanded along the AP axis and whether the dorsal Nkx6.1⁺-Foxa2⁺-Nkx2.2^- and Foxa2⁺-Nkx6.1⁺-Nkx2.2⁺ domains were affected by the Otx2 overexpression (Fig. 3; see Fig. S3 in the supplementary material). Compared with wild-type embryos, a selective and AP graded expansion of the Foxa2⁺-Nkx6.1^- mdDA domain was detected in Otx2-overexpressing mutants (Fig. S3A-F′; data not shown). In particular, this expansion was much less evident in En1^Cre/+;tOtx2^ov/+, where this expansion was detected in the posterior region of the intermediate mesencephalon and in the posterior mesencephalon(data not shown; Fig. 4). Instead, in En1^Cre/+; tOtx2^ov/ov embryos, a mild expansion of the Foxa2⁺-Nkx6.1^--Nkx2.2^-domain was detected in the pretectum and anterior mesencephalon (see Fig. S3A-B′ in the supplementary material); this expansion gradually increased, moving towards the posterior mesencephalon (see Fig. S3C-F′in the supplementary material). Interestingly, the relative extent of the Nkx6.1⁺-Foxa2⁺-Nkx2.2^- and Nkx6.1⁺-Foxa2⁺-Nkx2.2⁺ progenitor domains appeared not affected by the Otx2 overexpression along the AP axis (see Fig. S3 in the supplementary material).

A similar experiment performed to detect early post-mitotic mdDA Pitx3⁺ neurons showed an evident AP graded increase in the number of the Pitx3⁺ neurons and in the extent of the Nkx6.1^-mdDA domain (Fig. 3A-F″). Notably, the generation of Pitx3⁺ neurons was apparently unaffected in the pretectum and anterior mesencephalon of En1^Cre/+;tOtx2^ov/+ mutants and, similar to E10.75 embryos, the relative extents of the Nkx6.1⁺-Nkx2.2^- and Nkx6.1⁺-Nkx2.2⁺ domains were very similar in control and Otx2-overexpressing embryos (Fig. 3). Thus, these findings collectively suggest that mdDA progenitors exhibit an AP differential and dose-dependent sensitivity in their response to Otx2 overexpression through increased generation of mature mdDA neurons.

Based on previous findings, we analyzed in detail at E10.5 and E12.5 whether the identity and/or relative extent of VM progenitor domains was altered. At the anatomical level corresponding to the intermediate mesencephalon of E10.5 wild-type (data not shown) or tOtx2^bov/+ control embryos, Lmx1b was restricted to the mesDA domain (Fig. 4A); Shh expression included the Lmx1b⁺ domain and the ventral half of the Nkx6.1⁺ domain (Fig. 4B), and was adjacent to Nkx2.2(Fig. 4C); the Nkx2.2⁺ domain, in turn, partially overlapped with the dorsal region of both the Foxa2⁺ and Nkx6.1⁺ domains(Fig. 4D and compare Fig. 4B to C); and Wnt1 was co-expressed in the mesDA domain with Lmx1b with the exception of the medialmost floor-plate region(Fig. 4E). In En1^Cre/+; tOtx2^ov/+ and En1^Cre/+; tOtx2^ov/ov embryos, no obvious abnormalities were identified in the boundary relationships between the Nkx6.1⁺, Nkx2.2⁺, Shh⁺, Lmx1b⁺, Foxa2⁺ and Wnt1⁺ domains or in the extent of the Nkx6.1⁺ and Nkx2.2⁺domains (Fig. 4F-O), whereas a selective expansion of the Lmx1b⁺ and Wnt1⁺ domains was detected(Fig. 4F,K,J,O). In particular,this expansion was mild in En1^Cre/+; tOtx2^ov/+embryos and more pronounced in En1^Cre/+;tOtx2^ov/ov mutants. In addition to the analysis performed with Pitx3, Nkx6.1 and Nkx2.2 (Fig. 3; see Fig. S3 in the supplementary material), the phenotype described at E10.5 (Fig. 4) was confirmed at E12.5 by assessing also the combined expression of Lmx1b,Nkx6.1, Nkx2.2, Foxa2 and Shh (Fig. S4A-C″).

Next, we studied whether Otx2 overexpression in the VM had a specific effect on the generation of mesDA neurons or also affected other neuronal populations. To achieve this, we first compared the expression of Isl1 and Pou4f1, two post-mitotic markers of the OM and RN neurons, with that of AADC,Lmx1b and Pitx3; and then the expression of Nurr1 with that of Lim1/2, which were, respectively, expressed in early post-mitotic mesDA and RN neurons. We found that, in contrast to the increase in the mesDA neurons, the number of Isl1⁺ and Pou4f1⁺ neurons, the size of the OM and RN and the identity of RN early post-mitotic precursors were similar in Otx2-overexpressing and control embryos (compare Fig. 5A-H with 5I-X). Collectively, these findings indicate that Otx2 overexpression induces a selective expansion of both mesDA progenitors and neurons, without affecting identity and size of adjacent progenitor domains or their post-mitotic progeny.

Next, we studied whether Otx2 may control the extent of the mesDA domain by regulating selectively the proliferation of mesDA progenitors and/or their post-mitotic transition. To achieve this, we first determined the LI in the mesDA and Nkx6.1⁺ domains by providing a short pulse of BrdU (30 minutes) and measuring the percentage of BrdU⁺ cells over the total number of cells along the AP axis of the VM at E10.5 and E11.5 in control and mutant embryos. We found that the fraction of progenitors in S phase was remarkably increased in the mesDA domain of mutants at both stages analyzed,and that this increase was dose dependent and gradually more pronounced in the intermediate and posterior mesencephalon(Fig. 6A-H). Remarkably, almost no effect was detected in the proliferating activity of progenitors belonging to the Nkx6.1⁺ domain (Fig. 6A-F,I,J).

We then investigated whether the increase of mesDA neurons may also be contributed by a decreased number of early mesDA progenitors exiting the cell cycle. The percentage of cycling progenitors quitting the cell cycle within 24 hours of BrdU administration at E10.5 was calculated by measuring the fraction of BrdU⁺ cells that were Ki67^- (Qf). Our data showed that the Qfs measured in the mesDA domain, but not those in the Nkx6.1⁺ domain, were significantly reduced and, also in this case,correlated with the level of Otx2 overexpression (see Fig. S5 in the supplementary material). These findings thus suggest that the expansion of the mesDA domain may be caused by a selective enhancement in the proliferating activity of mesDA progenitors coupled to a decrease in the percentage of mesDA progenitors quitting the cell cycle. Based on these results, we studied the proliferation in the VM of En1^Cre/+;Otx2^flox/flox embryos(Puelles et al., 2004). Previously, we reported that, in this mutant, the lack of Otx2 generates ventral de-repression of Nkx2.2, loss of Nkx6.1 expression in progenitors,dorsal expansion of Shh and lack of Wnt1 expression. All these events resulted in the generation of 5-HT-containing neurons from RN and dorsalmost mesDA progenitors, and heavy reduction of mesDA neurons(Puelles et al., 2004; Prakash et al., 2006). Interestingly, these abnormalities mildly affected the pretectum and anterior mesencephalon (see Fig. S4D-I′ in the supplementary material), with increasing severity in the intermediate and posterior mesencephalon (about 80%less TH⁺ neurons) (see Fig. S4J-L′ in the supplementary material) (Puelles et al.,2004).

As previous studies have not addressed which process(es) was affected in mesDA neurogenesis of En1^Cre/+; Otx2^flox/floxmutants, we investigated whether the heavy reduction in mesDA neurons may be caused by abnormality in proliferation and/or premature post-mitotic transition and/or differentiation of their progenitors. First, we studied proliferation in the mesDA (Nkx2.2^-) and Nkx2.2⁺domains. Proliferating activity determined at E10.5 and E11.5 showed that the LI was heavily reduced in the mesDA domain of mutant embryos(Fig. 7A-J). By contrast, the LI detected in the Nkx6.1⁺ domain of control embryos and in the Nkx2.2⁺ domain of En1^Cre/+;Otx2^flox/flox mutants did not differ significantly(Fig. 7A-H,K,L). Next, we analyzed whether in these mutants mesDA progenitors quit the cell cycle prematurely. Compared with wild-type embryos, the Qfs for the intermediate and posterior (but not for the anterior) mesencephalon were dramatically increased in mutants (Fig. 7M-Q) and,remarkably, the expression of Ki67 was switched off in a relevant fraction of mesDA progenitors of the intermediate and posterior VM(Fig. 7M,N). As for the LI, the Qfs determined in the Nkx2.2⁺ domain also showed only a mild reduction when compared with those of the Nkx6.1⁺ domain of control embryos (Fig. 7M-P,R). Next, as Ki67 expression was heavily affected, we determined whether mesDA progenitors lacking Otx2 retained the expression of Sox2, which is normally transcribed by most of cycling CNS progenitors (Ki67⁺) and downregulated when they exit the cell cycle and differentiate(Graham et al., 2003; Kele et al., 2006). We found that, despite the severe lack of both Ki67⁺ and BrdU⁺cells (Figs 7, 8), the expression of Sox2 was retained (Fig. 8G,H),indicating that the Otx2-dependent impairment in proliferation did not affect Sox2 expression. Because this and previous studies showed that Wnt1expression was lost in the intermediate and posterior mesencephalon of En1^Cre/+; Otx2^flox/flox mutants(Prakash et al., 2006)(Fig. 8D,E) and expanded in Otx2-overexpressing embryos (Fig. 8F), we studied whether in the absence of Wnt1,Sox2⁺-Otx2^- progenitors expressed Cyclin D1 (CycD1), a direct target of the Wnt canonical pathway(Shtutman et al., 1999; Tetsu and McCormick, 1999). A remarkable loss of CycD1 expression (Fig. 8K) was detected in the Sox2⁺-Nkx2.2^-(Fig. 8B,H) domain of En1^Cre/+; Otx2^flox/flox embryos. Conversely,p27^kip1, which is normally expressed at high level in quiescent post-mitotic neuronal cells (Lee et al.,1996), was strongly activated in the Sox2⁺-Nkx2.2^- cells(Fig. 8H). Compared with En1^Cre/+; Otx2^flox/flox embryos, in mutants overexpressing Otx2 an essentially opposite phenotype was detected. Indeed, Wnt1 was expanded (Fig. 8F), BrdU⁺ (Fig. 8C,L) and CycD1⁺(Fig. 8L) cells remarkably increased in number, and a high level of p27^kip1 was detected only in Sox2^- post-mitotic neurons(Fig. 8I).

We then investigated whether abnormalities in Wnt1 expression correlated with the mesencephalic territory primarily affected in cell proliferation by Otx2 (overexpression or inactivation). We found that,although in the caudal mesencephalic area the expression of Wnt1 was expanded in Otx2 overexpressing mutants and lost in embryos lacking Otx2 (see Fig. S6C-D″ in the supplementary material), in the rostral mesencephalon of both En1^Cre/+; tOtx2^ov/ov and En1^Cre/+; Otx2^flox/flox mutants, the expression of Wnt1 was slightly affected (see Fig. S6A-A″ in the supplementary material), thus suggesting a close correlation between the territory requiring Otx2 for proliferation of mesDA progenitors and abnormalities in Wnt1 expression. Collectively, these findings indicate that Otx2 selectively controls the proliferating activity of intermediate and posterior mesDA progenitors and prevents their premature post-mitotic transition, possibly, through activation/maintenance of the Wnt canonical pathway.

Relevant studies have shown that Shh-dependent induction of Lmx1a is required for Msx1 expression, and that Msx1 is necessary for Ngn2 activation,which, in turn, promotes the differentiation of Sox2⁺ mesDA progenitors into Nurr1⁺ post-mitotic DA precursors(Kele et al., 2006; Andersson et al., 2006a; Andersson et al., 2006b; Ono et al., 2007). On this basis, we first analyzed the expression of Otx2 in relation to several markers at E11.5 and E12.5 in wild-type embryos. This analysis showed that in the Nkx6.1^-Lmx1b⁺ mesDA domain (see Fig. S7A,A′ in the supplementary material), Otx2 was co-expressed in progenitors with Sox2,Mash1, Ngn2, Lmx1a, Msx1 and Lmx1b (see Fig. S7B-D′ in the supplementary material; Fig. 9C,D,E,F,O,P), in early post-mitotic precursors with Ngn2, Nurr1 and Pitx3 (see Fig. S7D-F′ in the supplementary material), and in more mature mesDA neurons with a large subset of Pitx3⁺ cells (see Fig. S7F,F′ in the supplementary material). We then investigated whether the expression of these genes and, consequently, the differentiation of mesDA progenitor subsets was altered in response to increased level or lack of Otx2. These experiments showed that, compared with wild-type embryos, in Otx2 overexpressing mutants Lmx1a and Msx1 expression was expanded and the number of Ngn2⁺-Sox2⁺ and Mash1⁺-Sox2⁺ progenitors was increased, whereas, in En1^Cre/+; Otx2^flox/flox embryos, Lmx1aand Msx1 transcription was silenced and the expression of Ngn2 and Mash1 was confined to sporadic mesDA Sox2⁺ progenitors(Fig. 9C-J″). In contrast to these genes, the expression of Lmx1b was retained in En1^Cre/+; Otx2^flox/flox mutants, where it was also expanded dorsally within the Nkx2.2⁺ domain from which 5-HT⁺ neurons were generated(Fig. 9O-P″)(Puelles et al., 2004). Finally, we analyzed the generation of Ngn2⁺-Nurr1⁺ and Nurr1⁺-Pitx3⁺ subpopulations of post-mitotic mesDA neurons, which, as expected, were remarkably expanded in En1^Cre/+; tOtx2^ov/ov mutants and virtually absent at E11.5 or heavily reduced at E12.5 in En1^Cre/+;Otx2^flox/flox embryos (Fig. 9K-N″). These findings show that overexpression of Otx2 induces increased generation of mesDA neurons that correlates with a corresponding expansion of the subpopulations of differentiating progenitors,while lack of Otx2 results in the general failure of the mesDA differentiation program.

Previous studies have indicated that the specification of mesDA neurons is characterized by sequential steps that determine, first, the mesDA field and,subsequently, the progressive maturation of mesDA progenitors into post-mitotic immature precursors and, then, mature neurons expressing the TH neurotransmitter gene (reviewed by Smidt and Burbach, 2007; Smits et al., 2006; Prakash and Wurst,2006). In this process, Otx2 is early required to control the positioning of Fgf8 and Shh expression, to regulate Wnt1 expression and to maintain the integrity of progenitor domains in the whole VM (Puelles et al.,2003; Puelles et al.,2004; Simeone et al.,2002; Prakash et al.,2006; Liu and Joyner,2001; Smidt and Burbach,2007). In this study, we have investigated for the first time the potential role of Otx2 as an intrinsic determinant of the developmental program of mesDA neurogenesis, and have discovered that Otx2 controls selectively the proliferating activity of mesDA progenitors and is required to promote their differentiation by directly or indirectly activating the Lmx1a-Msx1-Ngn2 genetic cascade (Kele et al., 2006; Andersson et al.,2006a; Andersson et al.,2006b; Ono et al.,2007).

This study shows that Otx2 overexpression causes increased generation of mesDA neurons. This increase correlated with the level of overexpressed Otx2 protein and was more dramatic in the posterior and intermediate mesencephalon than in the anterior mesencephalon and pretectum. Consistently, embryos that lack Otx2 exhibit heavy reduction of mesDA neurons in the intermediate and posterior mesencephalon, whereas in the anterior mesencephalon and pretectum their generation is less affected (Puelles et al., 2004) (see Fig. S4 in the supplementary material). These data, therefore, indicate that, depending on the position occupied along the AP axis of VM, mesDA progenitors exhibit a differential response to Otx2. A second finding of this study is that although Otx2 is overexpressed even in the VM progenitor domains adjacent to the mesDA compartment, in these domains,neurons generation is not increased. Indeed, boundary relationships among Nkx6.1, Nkx2.2, Shh and Foxa2, as well as the number and identity of OM and RN neurons are unaffected by Otx2 overexpression. This suggests that Otx2 may exert a selective control on the generation of mesDA neurons by modulating their number along the AP axis.

The selective effect on the generation of mesDA neurons has been investigated by analyzing whether Otx2 is required to regulate the proliferating activity and post-mitotic transition of mesDA progenitors. This analysis has shown that Otx2 plays a major role in controlling proliferation of mesDA progenitors. Indeed, in Otx2 overexpressing embryos, the LI is significantly increased and the Qf is reduced. Conversely, in mutants that lack Otx2, the LI exhibits a drastic reduction and the Qf a dramatic increase. Furthermore, in En1^Cre/+; Otx2^flox/floxembryos, a large fraction of Sox2⁺ progenitors in the intermediate and posterior mesencephalon switch off Ki67 and induce high level of p27^kip1, suggesting that they represent a type of `frozen'progenitor that prematurely exits the cell cycle. Remarkably, these abnormalities are restricted to mesDA progenitors, as, in mutants that overexpress or lacking Otx2, the adjacent Nkx6.1⁺ or Nkx2.2⁺ domains are apparently unaffected or exhibit mild impairments in LI, Qf and in expression of Ki67 and p27^kip1. Thus,our data provide the first evidence that (1) Otx2 may regulate selectively the generation of mesDA neurons by controlling the proliferating activity of their progenitors; and (2) this control exhibits an AP graded effect. This and previous studies have shown that Wnt1 expression is lost in the VM of En1^Cre/+; Otx2^flox/flox mice and that Wnt molecules are differentially required, being indeed involved in promoting proliferation (Wnt1) or mesDA differentiation (Wnt5a)(Panhuysen et al., 2004; Castelo-Branco et al., 2003; Castelo-Branco et al., 2004). Our data suggest that Otx2 may control mesDA proliferation through the maintenance and/or activation of the Wnt/β-catenin pathway. In this context, active Wnt pathway through the TCF/LEF/β-catenin nuclear complex may modulate the transcription of a broad range of target genes, including cyclins and, in particular, Cyclin D1, a major regulator of the cell cycle progression (Fodde and Brabletz,2007; Shtutman et al.,1999; Tetsu and McCormick,1999; Lin et al.,2000; Arber et al.,1997). We show that CycD1 expression is suppressed specifically in the mesDA domain of mutants that lack Wnt1 expression, whereas in embryos overexpressing Otx2, the expanded expression of Wnt1correlates with a higher number of CycD1⁺ cells. Therefore, these findings strongly support the possibility that Otx2 may control the proliferation of mesDA progenitors through the maintenance/activation of the Wnt canonical pathway by regulating the expression of Wnt1. In view of the selective effect on mesDA progenitors, our data indicate propagation of the Wnt1 proliferative signal over a very short range.

Recent studies have indicated that the differentiation of mesDA progenitors and their transition to post-mitotic mesDA neurons require an intricate pathway of transcription factors (reviewed by Smidt and Burbach, 2007). Indeed, experiments performed in chick and analysis of mouse mutants collectively suggest that early in mesDA differentiation Shh-dependent expression of Lmx1a activates Msx1, which induces Ngn2; Ngn2, in turn, is required for the differentiation of Sox2⁺ progenitors into Nurr1⁺ post-mitotic young mesDA neurons(Andersson et al., 2006a; Andersson et al., 2006b; Kele et al., 2006; Ono et al., 2007). In this context, however, although in chick embryos the silencing of Lmx1a abolishes Nurr1 expression, mouse Lmx1a-null mutants exhibit only 30% fewer TH neurons and, similarly, Msx1-null embryos lack 40% of Ngn2⁺ progenitors and Nurr1⁺ mesDA neurons(Ono et al., 2007; Andersson et al., 2006b). As reported, this might suggest the existence of compensatory functions and/or,in addition to their sequential requirement, an Lmx1a-Msx1 synergistic action(Andersson et al., 2006b). Our data show that, compared with control embryos, in Otx2-overexpressing embryos,the expression of Lmx1a, Lmx1b, Msx1, Ngn2 and Mash1 is expanded in a higher number of mesDA progenitors, which generate more Ngn2⁺-Nurr1⁺ immature precursors and Pitx3⁺mature mesDA neurons; conversely, in embryos lacking Otx2, the expression of Lmx1a, Msx1, Ngn2 and Mash1 is lost or severely affected in Sox2⁺ mesDA progenitors, which are greatly impaired in their ability to generate Nurr1⁺ post-mitotic mesDA neurons.

Thus, these findings indicate that Otx2 is required to activate the genetic pathway leading to mesDA neuron generation. In this context, Otx2 might directly control the activation of Lmx1a and, consequently, the subsequent steps of mesDA differentiation. Supporting this possibility is the finding that ectopic expression of Otx2 in the ventral hindbrain is sufficient to induce Lmx1a, Msx1 and proneural gene expression, and the generation of TH⁺-Pitx3⁺ DA neurons(Ono et al., 2007). A second possibility is that Otx2 is indirectly required for Lmx1a activation by providing early mesDA progenitors with competence to respond to Shh-mediated induction of Lmx1a. In molecular terms, Otx2 might be necessary to repress a Shh-independent repressor of Lmx1a. A third possibility is that Shh might activate in mesDA progenitors an Otx2 co-activator required for Otx2-mediated induction of Lmx1a. The analysis of the mesDA genetic cascade in conditional mutants lacking Otx2 or Shh or both functions might shed light on this important aspect in the future. However, although our data suggest that Otx2-dependent expression of Wnt1 is more likely to be required to control proliferation of mesDA progenitors and prevent their premature exit from cell cycle, they do not exclude a priori that Wnt1 signaling may be necessary for one or more steps of mesDA differentiation, including competence of mesDA progenitors to Shh-inducing activity. A final finding of this study is that in the mesDA domain of En1^Cre/+; Otx2^flox/flox mutants most of the Sox2⁺ progenitors are Ki67^-p27^kip1+and that these `post-mitotic' progenitors do not activate any of the genes required for post-mitotic maturation of mesDA neurons (e.g. Nurr1). This suggests that, at least in the En1^Cre/+;Otx2^flox/flox mouse model, Sox2 expression does not require cell proliferation activity and that post-mitotic maturation of mesDA progenitors is not activated only by a block in proliferating activity and/or cell cycle exit, but, rather, depends on the correct progression of the differentiation process culminating with the transition of Sox2⁺-Ngn2⁺ progenitors into Sox2^-Ngn2⁺-Nurr1⁺ post-mitotic immature mesDA neurons. Interestingly, it has been recently shown that Otx2 and Sox2 physically interact to activate Rax1 expression in the retina(Danno et al., 2008). Whether Otx2 and Sox2 may interact also in mesDA progenitors and regulate the expression of target gene(s) remains to be investigated.

This and previous studies show that Otx2 is required for multiple steps of mesDA neuron development. Indeed Otx2 controls early specification of VM by both positioning Shh and Fgf8 signals, and maintaining the identity of progenitors domains (Puelles et al.,2003; Puelles et al.,2004; Prakash et al.,2006). Here, we have provided evidence that Otx2 exerts a crucial influence over mesDA neurogenesis by regulating the proliferating activity and differentiation of mesDA progenitors. Collectively, these and previous findings suggest that Otx2 represents a potentially relevant genetic determinant in future ES cell- or mesDA progenitor-based studies that are focused on improving the generation of authentic mesDA neurons and providing potential tools for the treatment of Parkinson's disease.

We thank D. Grieco, C. Missero and T. Russo for helpful discussions and criticisms on the manuscript, and the staff of the CEINGE animal house for excellent animal care. We are also indebted to G. Corte for the Sox2 and Otx2 antibodies, and to D. J. Anderson for the Ngn2 antibody. This work was supported by the Italian Association for Cancer Research (AIRC), the FP6 for the EuTRACC Integrate Project (LSHG-CT-2007-037445) and the `Fondazione Cassa di Risparmio' of Rome to A.S., and by the Deutsche Forschungsgemeinschaft, the Bundesministerium für Bildung und Forschung and the European Communion to W.W.