Nolz1 expression is required in dopaminergic axon guidance and striatal innervation

Midbrain dopaminergic (DA) axons make long longitudinal projections towards the striatum. Despite the importance of DA striatal innervation, processes involved in establishment of DA axonal connectivity remain largely unknown. Here we demonstrate a striatal-specific requirement of transcriptional regulator Nolz1 in establishing DA circuitry formation. DA projections are misguided and fail to innervate the striatum in both constitutive and striatal-specific Nolz1 mutant embryos. The lack of striatal Nolz1 expression results in nigral to pallidal lineage conversion of striatal projection neuron subtypes. This lineage switch alters the composition of secreted factors influencing DA axonal tract formation and renders the striatum non-permissive for dopaminergic and other forebrain tracts. Furthermore, transcriptomic analysis of wild-type and Nolz1−/− mutant striatal tissue led to the identification of several secreted factors that underlie the observed guidance defects and proteins that promote DA axonal outgrowth. Together, our data demonstrate the involvement of the striatum in orchestrating dopaminergic circuitry formation.

M idbrain dopaminergic (DA) neurons play an important role in several brain functions including locomotion, motivation and reward processes 1 . The guidance of DA axons towards their target areas is an important step in the establishment of functional circuits that are required for executing those roles. The projections of several molecularly distinct subpopulations of DA neurons have been defined [2][3][4] with substantia nigra (SN) and VTA DA neurons innervating the dorsal and ventral part of the striatum respectively, consistent with their specific roles. While the specification of DA neurons and their projection areas have been intensively studied, it remains relatively unclear how the establishment of DA axonal projections and striatal innervation are regulated during embryonic development.
After exiting the midbrain DA axons are attracted by and fasciculated within the medial forebrain bundle (MFB) 5 , which forms two rostrally oriented ipsilateral tracts within the ventral diencephalon. These axonal tracts run parallel to the ventral midline towards target areas in the forebrain including the striatum and cortex 6 . Several secreted guidance molecules with either attractive or repulsive activities are involved in the navigation of DA axons towards their target areas, including Slit/Robo, Netrin/Dcc, Ephrin, Semaphorin and Wnt signalling components 7 . Besides extrinsic factors secreted in the environment, axons can also be guided to the target areas by reciprocal axon-axon interactions as demonstrated for the innervation of the lateral habenula 8 and the establishment of thalamocortical-corticothalamic interactions 9 . However, whether establishment of DA connectivity is influenced by striatal patterning or the formation of striatal axonal extensions has not been determined.
The majority of the neurons in striatum are projection neurons, which can be subdivided into two subpopulations based on their transcriptional profile and target innervation [10][11][12] . Striatonigral projection neurons are specified by several transcription factors including Isl1, Ebf1 and Rarb and directly innervate the SN [13][14][15][16][17] . In contrast, striatopallidal neurons project to the GP and give rise to the indirect pathway. Here we investigate whether and how striatal patterning influences the guidance and target innervation of DA axons.
Nolz1 is as a transcriptional regulator expressed in the VTA DA neuronal lineage and striatal projection neurons [18][19][20][21] . Here we show that in the absence of striatal Nolz1 expression DA axons are misguided and fail to innervate the striatum. We demonstrate that the striatonigral to -pallidal switch in projection neuron subtype identity in Nolz1 −/− mutant embryos is associated with defects in establishment of DA and forebrain axonal tracts. The altered composition of guidance factors secreted from Nolz1 −/− mutant striatum provide a non-permissive environment for DA axons and other forebrain axonal tracts. Transcriptomic analysis resulted in the identification of proteins that can rescue the defects in DA axonal outgrowth. Finally, the acquired insight into mechanisms involved in DA circuitry formation will facilitate the development of approaches to improve graft outcome in cell transplantation studies.

Results
Nolz1 is required for establishment DA axonal connectivity. Previously, we have shown that Nolz1 is expressed in the VTA DA neuronal lineage 18 . To investigate the role of Nolz1 in DA neuron development, we analyzed tyrosine hydroxylase (TH) expression by iDISCO 22 in E18.5 Nolz1 −/− mutant embryos, in which the coding region of Nolz1 has been replaced by beta-Galactosidase (also referred to as Nolz1 bgal/bgal ) ( Supplementary  Fig. 1a). IDISCO analysis revealed that DA axons are misguided in Nolz1 −/− mutant embryos (Fig. 1a-d). While in wild-type embryos DA axons extend rostrally through the hypothalamus and innervate the striatum at E18.5 (Fig. 1a, c), a large proportion of TH labelled axons cross the midline in the hypothalamus (arrows in Fig. 1b, d) and terminate rostral of the striatum (arrowheads in Fig. 1b, d) in the mutant embryos. Analysis of NOLZ1 expression in relation to DA axons labelled by GlycoDAT and TH shows that NOLZ1 is expressed in regions that display the guidance phenotype e.g. the hypothalamus (arrow Fig. 1e, q, s) and striatum (arrowhead Fig. 1g, i) [19][20][21] . It further confirmed that a subset of DA axons cross the midline (Fig. 1e, f, k, l, q, r) and the remaining axons terminate rostral of the striatum (Fig. 1i, j, u, v) resulting in a lack of innervation of the rostral areas (Fig. 1g, h, m, n).
To investigate whether the DA axons that cross the midline have a different identity in comparison with the axons that keep following their trajectory towards the striatum, we analyzed the expression of GlycoDAT, which is higher expressed in SN neurons and their projections compared with VTA neurons 23 . Interestingly, mainly GlycoDAT + TH + SN DA axon bundles crosses the midline (Fig. 1k, l), while GlycoDAT − TH + axons extend towards the striatum (Fig. 1m, n), indicating that SN and VTA derived DA axons respond differentially to the absence of Nolz1 expression. Furthermore, the DA axon bundle is defasciculated and more spread out along the medial-lateral axis in Nolz1 −/− mutant embryos (Fig. 1o, p). Other axonal tracts running in the MFB are also misguided with 5HT labelled serotonergic axons crossing the midline alongside DA axons ( Supplementary Fig. 1b).
While the caudal part of the striatum remains innervated to a certain extent, the globus pallidum (GP) is totally devoid of any DA terminals in Nolz1 −/− mutant embryos (Fig. 1i, j, w, x and Supplementary Fig. 2a). The GP specification is normal as shown by CTIP2 (Fig. 2h, j) and NKX2.1 ( Supplementary Fig. 2b) expression in Nolz1 −/− mutant embryos. However, in the absence of Nolz1 we observed several aberrantly located ISL1 + positive cells ( Supplementary Fig. 2c) within in the GP, which could underlie the defect in striatal innervation by DA axons 24 .
Finally, the axon guidance phenotype observed in embryos lacking Nolz1 expression was not caused by the mislocalization or elimination of mutant cells as bGAL expression in Nolz1 bGal/bGal mutant embryos was similar to Wt and heterozygous embryos in the midbrain ( Supplementary Fig. 1c), hypothalamus ( Fig. 1q-t (arrows)) and the striatum (Figs. 1u-x, 2w).
Nolz1 is required for striatal projection neuron development. The DA axon guidance phenotype observed in Nolz1 −/− mutant embryos could not be explained by defects in DA neuron differentiation or hypothalamic patterning as no striking changes in DA neuron and hypothalamic marker expression could be observed (Supplementary Figs. 1c-e, 3a, b and Supplementary Table 1). However, we found that Nolz1 is required for the specification of striatal projection neurons. NOLZ1 is expressed by all striatal projection neurons as demonstrated by its overlap with DARPP32 ( Fig. 2a, a′) and CTIP2 (Fig. 2g, g′) expression 20,21 . In Nolz1 bgal/bgal mutant embryos striatal projection neurons, labelled by CTIP2 and bGAL, lack DARPP32, indicating that there is a defect in their maturation (Fig. 2b, b′, d, d′). DARPP32 expression appears in clusters in Nolz1 −/− mutant striatum that are positively labelled by L1 (Fig. 2f), indicating that these clusters represent abnormally fasciculated axonal tracts. There was a loss of SN innervation by DARPP32 + labelled striatal projection neurons in Nolz1 −/− mutant embryos (Fig. 2q-t). In contrast, the expression of CTIP2 25 (Fig. 2g, g′, i, i′, w) was unchanged in Nolz1 bgal/bgal mutant, suggesting that the differentiation of striatal projection   Fig. 1 Nolz1 −/− mutant embryos display DA axon guidance defects. a-d Visualization of TH + DA axon bundles in E18.5 Wt and Nolz1 −/− mutant mouse brains stained and cleared according to the iDISCO protocol. a Arrow point to DA axon bundle running in parallel to the ventral midline. b, d Arrows indicate midline crossing of DA axon bundle in Nolz1 −/− mutant embryos. b Arrowhead points to axons terminating in front of the striatum. c Arrowhead points to normal DA innervation of striatum in Wt embryos, which is disrupted Nolz1 −/− mutant embryos (arrowhead in d). Sagittal view in a, b and ventral view in c, d. e-j Immunohistochemical analysis of GlycoDAT and NOLZ1 expression in the hypothalamus (e, f) and striatum (g-j) of E18.5 mouse Wt and Nolz1 −/− mutant embryo. Arrows indicate midline crossing of DA axon bundles in the Nolz1 −/− mutant hypothalamus (f). i, j Arrows point to the GP, which lacks innervation in Nolz1 −/− mutant embryos. Arrowheads pointing to NOLZ1 positive cells in the hypothalamus (e) and striatum (g, i) of Wt embryos. k-p Immunostaining showing expression of GlycoDAT and TH in the hypothalamus (k-l), striatum (m-n) and caudal diencephalon/midbrain (o-p) in E18.5 Wt and Nolz1 −/− mutant embryos. Arrow in (l) indicates GlycoDAT + TH + DA axons crossing the midline in Nolz1 −/− mutant hypothalamus. o, p Dashed line indicates the width of the DA axon bundle extending from Nolz1 −/− mutant midbrain. q-x Expression of GlycoDAT and NOLZ1/BGAL in the diencephalon (q-t) and striatum (u-x) of Wt and Nolz1 −/− mutant embryos. Arrows indicating NOLZ1 + BGAL + labelled cells in Nolz1 bgal/+ heterozygous (q, q′, s) and bGAL labelled cells in Nolz1 bGal/bGal homozygous mutant (r, r′, t) hypothalamus. u, v DA axon bundles terminate in front of striatum in Nolz1 bGal/bGal mutant embryos. w, x GP in Nolz1 bGal/bGal mutant embryos is devoid of BGAL labelled cells and lack innervation by DA terminals. q, r, u, v sagittal and s, t, w, x coronal view. Data are representative of two (a-d) or three (e-x) independent experiments. Mb (midbrain), Str (striatum), GP (globus pallidus), Zi (Zona Incerta), Hth (hypothalamus). Scale bar 1000 μm (a-d); 200 μm (e-x). NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-16947-6 ARTICLE NATURE COMMUNICATIONS | (2020) 11:3111 | https://doi.org/10.1038/s41467-020-16947-6 | www.nature.com/naturecommunications neurons was normally initiated. In addition, striatal projection neuron markers FOXP1 and FOXP2 26,27 were downregulated in Nolz1 −/− mutant striatum ( Fig. 2k-n, w). There was no change in cell number, striatal volume and number of bGAL labelled cells between Nolz1 bGal/+ and Nolz1 bGal/bGal mutant striatum (Fig. 2u, v, w). Consistently, there was no increase in apoptosis in the absence of striatal Nolz1 expression ( Supplementary Fig. 2d). These data show that striatal Nolz1 expression is required for the generation of mature striatal projection neurons and their projections towards the midbrain.
Differential effect of striatal genes in the absence of Nolz1. To gain further insight into the transcriptomic changes that underlie the phenotypic alteration in Nolz1 −/− mutant striatum we compared the gene expression profile of E18.5 wild-type with Nolz1 −/− mutant striatum using RNA sequencing (Fig. 3a). The differentially expressed genes (Supplementary data 1) were subjected to hierarchical clustering and displayed in a heatmap, which revealed clusters of genes that were either higher (yellow) or lower (blue) expressed in the wild-type compared with mutant striatum (Fig. 3b) downregulation of generic striatal projection neuron markers DARPP32, FOXP1 and FOXP2 as observed in Nolz1 −/− mutant embryos ( Fig. 3f and Supplementary data 1). Interestingly, analysis of the genes representing the different clusters revealed that several genes downregulated in Nolz1 −/− mutant embryos (clusters 1, 2, 5b and 5c) are selectively expressed in striatonigral projection neurons, while striatopallidal neuron specific genes were expressed higher in the mutant (clusters 3, 4, 5a and 5e) [10][11][12] . We at randomly selected 61 differentially expressed genes for further verification and we could validate the expression of 67% of these genes (Supplementary data 2). The striatonigral specific genes, including Drd1, Rarb, Ebf1 11 , Isl1, Zfp521, Tac1 and Pdyn were either downregulated or absent in the E15.5 Nolz1 −/− mutant striatum (Fig. 3d, f). In addition, striatopallidal neuron specific genes Drd2, Six3, Grg4 (Tle4), Grik3, Penk, Ptprm and Adora2a were upregulated in the mutant striatum, except for Gpr6 (Fig. 3e, f). Despite the upregulation of several pallidal markers, the projections of striatopallidal neurons towards the GP were impaired in Nolz1 −/− mutant embryos as revealed by the analysis of PENK ( Supplementary Fig. S2e). Genes involved in early striatal patterning, progenitor-, glial-and interneuron specification were not changed in Nolz1 −/− mutant striatum (Supplementary data 1). Thus, Nolz1 is selectively required for the specification of striatonigral projection neurons and in the absence of Nolz1 expression several striatopallidal markers are ectopically expressed in the striatum.
Striatonigral to pallidal switch in Nolz1 −/− mutant striatum. Next, we investigated the specification of the striatal projection neuron subtypes in Nolz1 −/− mutant embryos in more detail. In E18.5 Nolz1 bgal/+ heterozygous mutant embryos about 40% of the striatal projection neurons labelled by bGAL express the striatopallidal marker SIX3 (Fig. 4c), which is expanded to nearly all (90%) striatal projection neurons in Nolz1 bgal/bgal mutant embryos (Fig. 4c). Since there is no reduction in total number of striatal projection neurons (Fig. 2v, w), the striatopallidal selective genes are most likely upregulated at the expense of nigral-specific markers in Nolz1 −/− mutant embryos. The temporal analysis of striatonigral and pallidal markers at several embryonic stages shows that the nigral to pallidal switch occurs at the time projection neurons are born. The striatopallidal lineage specific markers Six3, Grg4 and Drd2 are upregulated in Nolz1 −/− mutant embryos compared with Wt embryos at E11.5 and E12.5 (Fig. 4a, b). Immunolabelling shows that SIX3 is expressed in all bGAL labelled striatal projection neurons in E11.5 Nolz1 bgal/bgal mutant embryos, while the expression of striatonigral specific marker EBF1 (Fig. 4a, b) is not induced at any examined stage. Together, the temporal expression analysis demonstrates that the striatonigral to pallidal lineage switch in Nolz1 −/− mutant embryos coincides with the timing of striatal projection neuron production.
During striatal development the timing of cell cycle exit contributes to the subtype identity of striatal projection neurons [28][29][30] and the majority of neurons born at E11.5 have striatonigral specific identities with the number of striatopallidal neurons increasing over time 29,31,32 . To analyse whether altered timing of striatal projection neuron production contributes to the fate change in Nolz1 −/− mutant embryos we administered Bromodeoxyuridine (BrdU) to pregnant females at E10.5, E11.5 or E12.5 and analyzed the embryos at E15.5. While the percentage of BrdU labelled striatal neurons increases between E10.5 and E12.5 in Wt embryos, significantly less neurons were born in E11.5 Nolz1 −/− mutant striatum (5% in mutants vs. 20% in wild-type) ( Fig. 4d and Supplementary Fig. 4e). No differences in BrdU labelling between Wt and mutant embryos were observed when BrdU was injected at E10.5 or E12.5 ( Fig. 4d and Supplementary  Fig. 4e). The majority of the neurons labelled by BrdU resulting from injections at E11.5 and E12.5 expresses FOXP1 (Fig. 4e, f), indicating that mainly projection neurons are born at these stages in both Wt and Nolz1 −/− mutant embryos. The reduced levels of BrdU labelling at E11.5 in Nolz1 −/− mutant embryos coincide with the timepoint when the majority of neurons born in Wt embryos are positive for the striatonigral specific marker EBF1 11 (Supplementary Fig. 4a-c) and is consistent with the absence of striatonigral neurons in Nolz1 −/− mutant embryos. In contrast to the Wt striatum, the majority of BrdU labelled neurons in Nolz1 −/− mutant striatum expresses the pallidal marker SIX3 ( Supplementary Fig. 4f, g). This further demonstrates that neurons born between E10.5 and E12.5 have already been directed towards the striatopallidal fate. Furthermore, there is a strong reduction of the striosomal marker MOR1 (Supplementary Fig. 4h) and an expansion of the matrix marker CALBINDIN1 (Supplementary Fig. 4i) in Nolz1 −/− mutant striatum, which is in line with the loss of early born neurons 30,32 . No changes in BrdU labelling was observed in the cortex ( Supplementary Fig. 4d, e). The absence of Nolz1 in the ventricular zone and in proliferating cells within the subventricular zone 21 (Supplementary Fig. 4j), suggests that Nolz1 regulates the temporal production of striatal projection neurons non-cell-autonomously through a yet unknown mechanism.
Despite, the striatal selective ablation of Nolz1, born Foxg1-IRES-Cre;Nolz1 fl/fl mutant pups died within 1 month after birth for unknown reasons.   Fig. 6q) indicate a cell-autonomous role of Nolz1 in regulating apoptotic marker expression. The phenotypic resemblance between the constitutive and conditional striatal-specific Nolz1 mutant mouse lines demonstrates striatal, non-cell-autonomous requirement of Nolz1 in orchestrating the attraction and guidance of DA and other axonal tracts through the striatum. Defects in formation of striatal axonal extensions has been linked to abnormalities in forebrain axonal tract formation 36 . To investigate whether the striatal outgrowth is affected in Nolz1 −/− mutant embryos we injected DiI in E15.5 wild-type and mutant striatum ( Fig. 6a-d). DiI injected in the wild-type striatum was retrogradely transported to the midbrain in Wt embryos (Fig. 6a,  b), but not in Nolz1 −/− mutant embryos (Fig. 6c, d) revealing a defect in the establishment of striatonigral projections (Fig. 6c). Consistent with the absence of striatonigral projections, the axonal length in striatal explants from Nolz1 −/− mutant embryos was significantly shorter compared with wild-type explants (Fig. 6e, f).
Nolz1 −/− mutant striatum secretes cues repulsive to DA axons. Next we investigated whether either reduced striatal axonal outgrowth or altered guidance cue expression could underlie the defects in establishment of DA neuronal connectivity observed in Nolz1 −/− mutant embryos. In the absence of Pcdh10 striatal axonal extensions failed to form 36 , a phenotype similar to that observed in Nolz1 −/− mutant embryos. The lack of striatal axonal outgrowth in Pcdh10 −/− mutant embryos caused defects in the formation thalamocortical projections as shown by L1 and NF labelling ( Supplementary Fig. 7b, c) 36 . Furthermore, Darpp32 expressing striatonigral axons failed to outgrow towards the midbrain (Fig. 6g, h) 36 . Despite the reduced striatal axonal outgrowth, the innervation of the striatum by DA axon terminals was not affected (Fig. 6i, j) in E14.5 mutant embryos. Furthermore, in Pcdh10 −/− mutant embryos both striatonigral and -pallidal projection neurons markers and genes involved in axon guidance are normally expressed (Supplementary Fig. 7a). Thus the lack of striatal outgrowth alone does not cause the aberrant DA axonal trajectory observed in Nolz1 −/− mutant embryos.
The transcriptomic analysis revealed that a subset of genes involved in axon guidance were altered in Nolz1 −/− mutant striatum, including Netrin1, Sema3a, Vegfc, PlxnD1 and Tgfa (Fig. 3g, Supplementary Fig. 3a, Supplementary Table S2 and Supplementary data 1). Therefore, we investigated whether the altered composition of secreted factors from the striatum influences the guidance of DA axons towards the striatum. There was no difference in axonal length and direction of outgrowth between wild-type and Nolz1 −/− mutant ventral midbrain explants when cultured alone ( Supplementary Fig. 8a). DA axons labelled by B-TUBULIN and TH extended from wild-type and Nolz1 −/− mutant midbrain explants were attracted by wild-type striatal explants 40 (Fig. 6k-m; P/D ratio > 1). However, when midbrain explants were cocultured with Nolz1 −/− mutant striatal explants DA axons were strongly repelled 40 (Fig. 6k-m; P/D ratio < 1). The distance between the explants was equal under all conditions (Fig. 6n). Similarly, axonal extension emanating from the thalamus were repulsed by striatal explants obtained from Nolz1 −/− mutant embryos ( Supplementary Fig. 8b, c). In contrast to midbrain explants, Nolz1 −/− mutant thalamic explants were also repulsed by wild-type striatal explants indicating a thalamic requirement of Nolz1 in mediating chemoattraction.
The explant cocultures suggest that secreted factors emanated by Nolz1 −/− mutant striatum 5 exerts a repulsive effect on DA axons. In agreement, conditioned medium collected from cultured Nolz1 −/− mutant striatal explants caused collapsed growth cones on nearly 70% of the axons derived from ventral midbrain explants (Fig. 7a, b) 41 , showing that factors secreted from the mutant striatum repulse DA axon outgrowth. Conditioned medium from Wt striatal explants did not have any effect on the growth cone morphology (Fig. 7a, b; 24% vs. 20% in controls). Of the secreted factors differentially expressed between wild-type and Nolz1 −/− mutant striatum (Fig. 3g) we found that SEMA3A caused more than 70% of the growth cones to collapse, while TGFA and NETRIN1 did not have any effect (Fig. 7a, b). To investigate further whether there is a difference in the potential to attract DA axons over longer distances between wild-  Fig. 8f). However, the axonal attraction mediated by wild-type striatal tissue was totally abolished in the presence of the ERBB inhibitor Afatinib 43 (Fig. 7i, j and Supplementary Fig. 8e). These data demonstrate that TGFA signalling activation is sufficient to attract DA axonal projections.

Discussion
The correct establishment and maintenance of DA axonal projections is important for normal brain function and defects in these processes underlie neurological and neurodegenerative diseases including Schizophrenia and Parkinson's disease 44,45 .
Here we showed how striatal patterning defects in Nolz1 −/− mutant embryos have an impact on the establishment of DA axonal projections, which led us to propose the following model as depicted in Fig. 8. (1) In both constitutive and striatal-specific conditional Nolz1 mutant embryos a subset of DA axons crosses the ventral midline in the hypothalamus. The remaining axons maintain their trajectory towards rostral brain areas, but terminate in front of the striatum. (2) Furthermore, in the absence of striatal Nolz1 expression striatopallidal projections neurons are produced at the expense of striatonigral neurons. This lineage switch resulted in reduced striatal outgrowth (3) and altered composition of growth factors secreted from the striatum (4). The guidance of DA axons towards target areas does not depend on the elongation of striatal axons (3). Instead, the impaired striatal projection neuron specification and striatonigral-pallidal subtype conversion create a repulsive environment for DA growth cones and other forebrain tracts (4). Thus, we demonstrated that defects in striatal projection neuron specification have a direct effect on DA axonal tract formation and striatal innervation. We showed that Nolz1, which is expressed in all striatal projection neurons, is selectively required for the specification of striatonigral projection neurons. In Nolz1 −/− mutant embryos the expression of several striatonigral specific genes are either reduced or absent, which is accompanied by the ectopic induction of striatopallidal-specific genes. Nolz1 acts upstream of several other transcription factors that have been previously implicated in regulating the striatonigral fate. However, the striatonigral selective transcriptional regulators Isl1, Ebf1, Foxo1 and Rarb [13][14][15][16][17]46 , regulate only a subset of striatonigral specific genes resulting in a much milder striatal phenotype. Although most of the striatopallidal markers are normally induced in the Nolz1 −/− mutant striatum, the pallidal fate is not fully executed as shown by the reduction of Darpp32, Arpp21 and Gpr6 expression and the reduced innervation of the GP by striatopallidal neurons. Thus, while mainly the generation of striatonigral neurons is affected in the absence of striatal Nolz1 expression, Nolz1 might also be required to regulate certain aspects of the striatopallidal fate. We found that in the absence of striatal Nolz1 expression, striatopallidal projection neurons are generated at the expense of striatonigral neurons. As a transcriptional repressor 19,47-49 Nolz1 might promote the specification of nigral neurons by directly repressing striatopallidal-specific genes. A partial nigral to pallidal lineage switch is also observed in Isl1 −/− mutant embryos 15 and Nolz1 might interact with Isl1 to prevent that pallidal transcriptional programmes are unsilenced in nigral neurons. The downregulation of striatonigral specific genes in the Nolz1 mutant striatum might be a consequence of the upregulated expression of pallidal genes in nigral neurons. Candidates for mediating the repression of nigral-specific gene programmes are Six3 and Tle4 50,51 . Both genes are transcriptional repressors that have been previously implicated in pallidal fate specification 52 and are ectopically expressed in Nolz1 −/− mutant striatum. In contrast to Nolz1 and Isl1, which are involved in the repression of pallidal genes, histone methyltransferase G9a is required to prevent the ectopic induction of striatonigral specific genes 53 . Thus, the specification of these two distinct projection neuron subtypes is at least to some extent dependent on the repressing activities of nigral and pallidal lineage determining factors. Our data also revealed that Nolz1 is involved in the timing of striatal projection neuron production, which could influence the subtype identity as well. BrdU birthdating experiments showed a strong decrease in neuronal birth at E11.5 in Nolz1 −/− mutant embryos, which coincides with the production of EBF1 positive striatal projection neurons. Since Nolz1 expression is restricted to post-mitotic neurons, it is not clear how Nolz1 regulates cell cycle exit in the ventricular zone. A possibility is that Nolz1 influence the temporal aspects of projection neuron production indirectly through the induction of other secreted factors.
DA axons are guided to their target areas by temporal and spatial exposure to repulsive and attractive cues. Explant studies have revealed that DA axons are initially attracted by the MFB, followed by the striatum at later stages 5,54 . The initial recruitment of DA axons towards the MFB is not affected in Nolz1 −/− mutant embryos. However, the subsequent progression of DA axonal projections into a rostral direction towards the striatum is disrupted, resulting in midline crossing of a subset of DA axons and lack of striatal innervation. Although gene mutations in Netrin, Slit2, Robo1, Dcc and Nrp2 55-59 also causes aberrant midline crossing, these guidance cues are broadly expressed and their tissue selective requirement has remained unclear. Our results show for the first time that midline crossing of DA axons in the hypothalamus is mediated by phenotypic alterations in the  16,62,63 . In addition, while graded Netrin and Sema7a signalling in the striatum are required for the topographic guidance of DA axons, it is not necessary for DA innervation 57,64 . Thus, the loss of Nolz1 expression results in a striatal phenotype that uniquely affects the guidance and innervation of DA axons. The striatonigral to -pallidal lineage switch observed in Nolz1 −/− mutant embryos demonstrates the influence of striatal projection neuron subtype specification on establishment of DA projections. Analysis of Pcdh10 −/− mutant embryos revealed that striatal axonal outgrowth is not required for directing DA projections towards forebrain target areas. Therefore, it is more likely that the altered composition of secreted factors emanated by Nolz1 −/− mutant striatum causes aberrant DA axonal tract formation and loss of striatal innervation. The striatum normally produces a permissive environment for axonal tracts that are guided through the forebrain 5,54,65 .
However, extrinsic factors secreted from the Nolz1 −/− mutant striatum impose a repulsive effect on DA axons as shown by the growth cone collapse assay and explant culture. Whether the repulsive environment is caused by the loss of attractive or by the induction of repulsive guidance factors in Nolz1 −/− mutant embryos will need to be determined. It is likely that a combination of factors secreted by the Nolz1 −/− mutant striatum causes the observed DA axon guidance phenotype. It has been shown that axonal responsiveness to guidance factors is temporally regulated and highly context dependent 66 . Also the induction of cCASP3 in thalamic axons could alter axonal responsiveness to extrinsic factors resulting in guidance defects 67 . We identified TGFA as one of the secreted signalling molecules that is downregulated in the striatum of Nolz1 −/− mutant embryos. In Tgfa −/− mutant embryos there is a slight reduction of striatal innervation 68 . Interestingly, we showed that addition of TGFA could rescue the outgrowth of DA axons towards the Nolz1 −/− mutant striatum. The TGFA receptors EGFR1 and ERBB4 are expressed by DA neurons and blocking the receptor signalling by Afatinib 43 totally abolished the axonal outgrowth. So far, TGFA signalling has been shown to regulate axonal growth indirectly through its function in astrocytes 69 . Our finding suggests a more direct role of this signalling pathway in regulating axonal growth and shows that TGFA is sufficient to attract DA axons towards the striatum. Detailed understanding of mechanisms involved in the establishment of DA axonal projections will facilitate the development of novel strategies to improve graft outcome in cell replacement studies. Our study has identified several signalling molecules altered in Nolz1 mutant striatum that could be further exploited for its involvement in DA axon guidance. For example, addition of signalling molecules including TGFA could promote the outgrowth of DA axons from intranigral grafts and accelerate target innervation.
Immunohistochemistry. Embryonic tissue was fixed in 4% PFA on ice for either 1 h (E11.5 heads), 1.5 h (E13.5 heads and E15.5 brains) or 3 h (E18.5 dissected brains). Brains were embedded in OCT (VWR) and cryosectioned at a thickness of 12 μm using a Leica cryostat CM3050S. Sections were mounted on Superfrost plus glass slides (Thermo Scientific, USA) and air dried for 30 min and washed 3 × 5 min in PBS/0.1% triton-x. Adult brains were fixed in 4% PFA overnight at 4°C and sections were cut on a freezing microtome (Leica) at a thickness of 30 μm. Immunohistochemistry was performed as described 18 . Nuclei were visualized by DAPI (Sigma). Stained cells were analyzed using a confocal microscope (LSM 510 Meta, Zeiss or LSM 880, Zeiss). The antibodies and concentrations are shown in the supplementary resource table.
BrdU administration and tissue processing. Time-mated pregnant females were injected subcutaneously with 5-bromo-2′-deoxyuridine (BrdU) 100 mg kg −1 (Sigma Aldrich, USA) at E10.5, E11.5 or E12.5. Embryos were isolated at E15.5 and dissected brain were fixed for 1.5 h in 4% PFA on ice and further processed as for immunohistochemistry. For BrdU staining, sections were incubated with 0.1 M citric acid for 20 min at 100°C, adjusted to room temperature and rinsed three times in PBS before applying the blocking solution containing 0.25% Triton X-100 (Sigma), 5% donkey serum in PBS (Jackson ImmunoResearch, UK). Sections were incubated with rat monoclonal anti-BrdU antibody (1:500, Abcam) diluted in blocking solution at overnight at 4°C. The next day, sections were washed three times in PBS before being incubated for 1 h with a goat anti-rat secondary antibody (1:500, Abcam).
In situ hybridization. Tissue was fixed and processed as described for the immunohistochemistry. Probes were selected from gene lists obtained by comparing gene expression profiles of E18.5 Wt and Nolz1 −/− mutant striatum. Primers used for generating in situ probes are shown in the supplementary Table 3. The in situ probes were produced by in vitro transcription as using cDNA from E18.5 mouse embryos as a template. The in situ hybridization was performed following standard procedures. Sections were analyzed and photographed using Axiovert 200 microscope (Zeiss) equipped with a digital camera using 10x magnification.
In vitro explant culture in collagen matrices. Ventral midbrain, striatal and thalamic explants were obtained from E13.5 embryos as described 40 . Striatal explants were embedded in a collagen matrix either with an explant derived from the ventral midbrain or thalamus as previously described 40 with some minor modifications. Special care was taken to keep a distance of 300-500 μm (1 explant size) between explants during the embedding process. Explants were cultured in explant medium containing Neurobasal A medium supplemented with B27, 1 M Hepes, L-glutamine, Pen-Strep and Beta-mercaptoethanol for 3 days. To determine the attractive/repulsive effects on midbrain and thamamic explants, the length of b-Tubulin labelled axons was measured with Zeiss software 2012. For quantification, the midbrain and thalamic explants were divided into two areas, proximal (facing towards the explant) and distal (facing away from explant). In cases where explants extended many axons, only the five largest were considered. The average axonal length for the proximal and distal compartments was calculated to obtain the P/D ratio. When the value was >1.0, the effect was considered attractive; values <1.0 were considered to be indicative of a repulsive effect 40 . The distance between two explants was measured using the Measure function in Zeiss software 2012 to control for aberrant distance-mediated effects on axonal outgrowth.
Axonal growth chambers in microfluidic devices. Polydimethylsiloxane microfluidic devices were purchased from Xona Microfluidics (Standard Neuron Device Cat.No: SND900) and were used according to the manufacturer's protocol. Briefly, they were attached by pressure to a 10 cm dish precoated with poly-D-lysine (50 μg mL −1 ). Wt ventral Midbrain tissue was dissected from E13.5 mouse embryos, kept in 5% FCS + DMEM solution and dissociated with TrypLE Express (Gibco) to obtain a single-cell suspension of DA neurons. Cells were then plated (50,000 cells μL −1 ) in the cell compartment of the microfluidic device. After 10 min, 150 μL of explant medium was added. The other side of the chamber was filled with 200 μL of explant medium with striatum explants from Wt (positive control) or Nolz1 −/− mutant embryos. Nolz1 −/− mutant striatal tissue was treated with 50 ng ml −1 TGFa, while 50 nM Afatinib was added to wild-type striatal explants. Afatinib was dissolved in DMSO at a stock concentration of 100 mM and diluted to 50 nM using serial dilutions. DMSO was added to wild-type striatal explants as a control at a similar concentration. Devices were then maintained at 37°C in humidified 5% CO 2 /95% air for 9 days and medium was changed regularly every 2-3 days. Cells were fixed by adding 200 μl of pre-warmed 4% PFA with 10% sucrose and stored at 4°C for immunohistochemistry. Importantly, while cells were seeded at the same initial density, each chamber contained a variable number of axons after 9 days. Axonal length in the devices was measured using Zeiss ZEN black 2012 software by drawing a segmented line along each individual axon projecting out of the microgroove in the distal compartment.
Striatal conditioned medium. The striatum was dissected from E13.5 mouse embryos in 5% FCS + DMEM solution and dissociated with TrypLE Express (Gibco) to obtain single cells DA neurons. Cells were plated (1 striatum/well) in 24 wells plate coated with poly-D-lysine (50 μg mL −1 ). One millilitre of explant medium previously describe was added. Striatal conditioned medium was obtained after 7 days of striatal primary culture from striatum Wt or Nolz1 −/− mutant. The media were collected and immediately placed on dry ice and then stored at −80°C.
Growth cone collapse assay. Wells were coated overnight with poly-D-lysine (50 μg mL −1 , Sigma) at 4°C, subsequently washed with sterile PBS three times before being coated with laminin (1:10 in PBS, Invitrogen, UK) and incubated for 2 h at 37°C. In the meantime, ventral midbrain tissue was dissected from E13.5 WT mouse embryos, placed in ice-cold, Dulbecco's modified Eagle's medium (DMEM; Gibco-BRL), sectioned into small pieces and then placed into the centre of each laminin-coated well. Before seeding, wells were washed two times with sterile PBS and then supplemented with explant medium as previously described 40 . Explants were then cultured at 37°C for 72 h to the laminin-coated wells. For control experiments, the explant medium was removed and replaced with fresh medium. Either conditioned medium derived from Wt and Nolz1 −/− mutant striatal tissue or recombinant proteins (R&D) of the axon guidance cues NETRIN1, SEMA3A and TGFa were added to the cell medium at a concentration of 300 ng mL −1 , 300 ng mL −1 and 50 ng mL −1 , respectively. Following addition of appropriate medium or recombinant protein, plates were incubated at 37°C for 30 min. After the fixation (1 h in pre-warmed 4% PFA with 10% sucrose), explants were washed with PBS and stored at 4°C. For quantification, growth cones were scored as either collapsed or uncollapsed using an inverted phase contrast microscope with a ×40 objective. According to previous published methods 41 , tips of axons were classified as uncollapsed if prominently spread growth cones with flattened lamellipodia and/ or two or more filopodia were identified. Bullet-shaped neurite tips with less than two filopodia and/or no lamellipodium present were scored as collapsed growth cones.
Image acquisition and quantification. Images were acquired on an Axiovert 200 M inverted microscope, Zeiss LSM 880 or Zeiss LSM 510 confocal microscopes and captured at ×10, ×20 or ×25 objective depending on staining procedure, sample size and area of interest. Images were processed using Adobe Photoshop CC2018 and Adobe Illustrator CC2018 software. For quantification, high-definition tilescan confocal images were taken at each level of the striatum in three different animals per condition using the lateral ventricle as an anatomical landmark. Images were opened with ImageJ and converted to 8-bit. Positively labelled cells were identified using built-in threshold algorithms to best determine the intensity of the background against the positive staining. After applying the threshold settings, a specified region of interest was traced along the contours of the striatum and positively labelled cells were quantified semi-automatically in ImageJ. Similarly, to quantify the total area of each striatum, the striatum was manually delineated in each singleplane section using the line selection tool and the total area was automatically computed across all rostro-caudal sections of the striatum. Results were converted from pixels 2 to mm 2 by spatially recalibrating each image. Microsoft Excel was used for the organization and statistical analysis of the data.
Cell counting and fibre density analysis. To determine the percentage of Aldh1a1 expressing TH neurons in the midbrain the number of ALDH1A1 and TH positive cells was counted manually in every 4th section of each E18.5 embryo in at least three embryos of each genotype. Statistical significance was calculated by the unpaired student's t test and data is presented as mean ± sd. Striatal fibre density of GlycoDAT positively labelled axonal projections was measured by densitometry using ImageJ software. Striatal sections were fluorescently labelled with GlycoDAT and images were taken using confocal microscopy. Fibre density of whole striatal area was measured and the measured values were corrected for non-specific background staining by subtracting values obtained from the cortex. Of each embryo every 8th section was analyzed of at least three different embryos.
iDISCO. Dissected brains from E18.5 WT and Nolz1 −/− mutant embryos were processed as previously described 22 . Briefly, whole brains were dehydrated in methanol in PBS and bleached overnight at 4°C. Tissue was then gradually rehydrated in PBS by removing. Detergent washing was then performed in PBS with 0.2% Triton X-100 (2 × 1 hr). Tissue was incubated overnight at 37°C in PBS with 0.2% Triton X-100 and 0.3 M glycine, followed by blocking in PBS with 0.2% Triton X-100 and 6% normal donkey serum for 2 days. Following blocking, the tissue was washed for 1 hr twice in PBS with 0.2% Tween-20 and 10 μg mL −1 heparin (PTwH). Brains were incubated with a primary rabbit-anti-TH (1:400) antibody diluted in PTwH/5%DMSO/3% Donkey serum at 37°C for 4 days. Excess primary antibody was washed for 1 day in PTwH with periodic solution changes. Secondary antibody donkey anti rabbit Alexa 647 (1/250) was applied in PTwH/3% Donkey serum at 37°C for 4 days. After incubation with the secondary antibody, samples were washed in PTwH for 5 days with periodic solution changes. After washing away excess secondary antibodies, optical clearing of iDISCO samples was performed as described in ref. 22 . Tissue was gradually dehydrated in resistant glassware with tetrahydrofuran in water. Remaining lipids were extracted with dichloromethane for 1 hr and dibenzyl ether (DBE) was used for refractive index matching. Samples were kept in a full vial of DBE.
RNA extraction and quantitative real-time PCR. Total RNA was extracted using a RNeasy micro kit (Qiagen) according to the manufacturers' protocols. Biological replicates were obtained from at least three independent experiments. The RNA quality was assessed using a Nanodrop 2000 spectrophotometer (Thermo Scientific) prior to amplification. cDNA was synthesized from total RNA via reverse transcription using the reverse transcriptase SuperScriptIII (Invitrogen) and Oli-godT as primer (Invitrogen). Real-time qPCR was performed on a QuantStudio 6 Flex system (Thermo Fisher) using Fast SYBR Green Mastermix (Thermo Fisher). Gene expression values were normalized against RPL19 and fold change was calculated using the 2 −ΔΔCT method. The primer sequences used are shown in the supplementary resource table. Measurement were taken from distinct samples. Significance was determined by the two-tailed student's t test.
RNA sequencing and bioinformatic analysis. Striatal tissue was dissected from both hemispheres of E18.5 Wt and Nolz1 −/− mutant embryos, frozen on dry ice and stored at −80°C. Each sample containing both hemispheres from the same embryos was further processed for RNA extraction using RNeasy Micro Kit (Qiagen Inc., UK). Three replicates consisting both hemispheres were used for each biological condition.
Library preparation. Total RNA was first examined using the Agilent RNA 6000 Nano kit. All samples passed QC with a RIN value > 8. To remove ribosomal RNA 2.5 μg total RNA was treated with the Illumina RiboZero Gold (Human/Mouse/Rat) kit, according to the manufacturer's protocol. From 100-200 ng of the ribo-depleted RNA sequencing libraries were prepared using the Illumina TruSeq Stranded mRNA kit, according to the manufacturer's protocol. Sequencing libraries were examined using the Agilent High Sensitivity DNA kit, and library concentrations determined. Sequencing was carried out by the Earlham Institute, 150PE using the HiSeq4000. Sequenced libraries were verified using FastQC (version 0.11.5).
RNA-seq data analysis. Adapters were trimmed using Cutadapt (version 1.16). Reads were mapped to the genome in a transcriptome-aware manner using HISAT2 (version 2.1.0), using the mouse HISAT2 genome tran index (Ensembl, GRCm38). Features were counted using the featureCounts function from the Rsubread package (version 1.28.1), using the annotation in the relevant Ensembl gtf file (GRCm38 release 92).
Bioinformatics. The differential expression between WT and mutant was analyzed using the DESeq2 (version 1.18.1) package. Genes were identified as being differentially expressed at an adjusted p value of <0.05 for the DESeq2 results and with a fold change of at least 1.5. Genes passing these thresholds, in either method, were used in subsequent analyses. Using these criteria there were 166 genes downregulated and 139 genes upregulated in the Nolz1 −/− mutant striatum. The heatmap was created using the heatmap.2 function within the gplots R package using significantly changing genes from the DESeq2 analysis with a fold change of at least 2. Clustering was done within heatmap.2 using complete-linkage.
Reporting summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability
The transcriptomic data have been deposited to ArrayExpress with the data set identifier E-MTAB-8240 (https://www.ebi.ac.uk/arrayexpress/experiments/E-MTAB-8240/). All data are included in this article and supplemental data files and are available from the corresponding author upon reasonable request. The source data underlying Figs. 2u, v, w; 3f; 4c, d, e; 5g; 6f, m, n; 7b, j and Supplementary Figs. 1d, e; 2c; 4b, c, g; 8c can be found in the source data file.