Sonic hedgehog expression in zebrafish forebrain identifies the teleostean pallidal signaling center and shows preglomerular complex and posterior tubercular dopamine cells to arise from shh cells

Ventralization, a major patterning process in the developing vertebrate neural tube (central nervous system, CNS), depends on Sonic hedgehog (SHH) as a main signaling morphogen. We studied the CNS of late larval and young adult zebrafish in a transgenic shh‐GFP line revealing increased neuroanatomical detail due to the progressed differentiation state compared to earlier stages. Some major findings emerge from the present study. (a) shh –GFP is still expressed along the adult zebrafish CNS neuraxis in most locations seen in larvae. (b) We newly identify a ventroposterior shh pallidal domain representing the basal telencephalic signaling center important for basal ganglia development known in other vertebrates (i.e., the anterior entopeduncular area—basal medial ganglionic eminence of mammals). (c) We further show late‐emerging shh‐GFP positive radial glia cells in the medial zone of the dorsal telencephalon (i.e., the teleostan pallial amygdala). (d) Immunostains for tyrosine hydroxylase demonstrate that there is selective colocalization in adult dopamine cells with shh‐GFP in the posterior tuberculum, including in projection cells to striatum, which represents a striking parallel to amniote mesodiencephalic dopamine cell origin from shh expressing floor plate cells. (e) There is no colocalization of shh and islet1 as shown by respective shh‐GFP and islet1‐GFP lines. (f) The only radially far migrated shh‐GFP cells are located in the preglomerular area. (g) There are no adult cerebellar and tectal shh‐GFP cells confirming their exclusive role during early development as previously reported by our laboratory.

However, there are two unresolved problems regarding sonic hedgehog expression in the zebrafish brain. One is the lack of a telencephalic (subpallial) shh expression domain comparable to what is described in amniotes as the anterior entopeduncular area (AEP) and medial ganglionic eminence (MGE; see section 4). These amniote telencephalic shh domains are crucial for correct ventral telencephalic gene expression (e.g., Dlx, Ascl1, Nkx2.1, Islet1, Lhx6/7) and, thus, for correct basal ganglia development as well as for repressing dorsal (i.e., pallial) gene expression ventrally (see section 4). A second issue is the developmental role of SHH in the generation of basal diencephalic dopamine cells. In mammals, midbrain substantia nigra/ventral tegmental dopamine cells are known to derive from sonic hedgehog expressing floor plate cells (Joksimovic et al., 2009;Blaess et al., 2011;Hayes, Zhang, Albert, Zervas, & Ahn, 2011; see section 4). Since teleosts lack midbrain dopamine cells and only possess basal diencephalic dopamine cells (posterior tuberculum)-which also contribute to the SN/VT in mammals (see section 4)-that nevertheless project to the fish basal ganglia (review Wullimann, 2014), we wanted to verify whether these zebrafish diencephalic ascending dopamine cells are produced by shh expressing cells.
Thus, we looked at early adult (3-month-old) transgenic shh-GFP zebrafish and described in neuroanatomical detail all GFP-positive CNS structures. Because the advanced differentiation state of the adult brain allows for a far more detailed identification of shh-GFP structures compared to larvae, we anticipated that the data will shed light onto both the recognition of a true pallidal shh domain as well as on the origin of dopaminergic posterior tubercular projection neurons to the teleostean striatum (see section 4.4.) from shh expressing cells from which part of the preglomerular area is also revealed to derive.

| Transgenic zebrafish strains
The transgenic line Tg(2.4shha-ABC-GFP)sb15 was originally published as Tg(2.2shh:gfpABC#15) by Shkumatava, Fischer, Müller, Strähle, and Neumann (2004). The injected construct includes the sonic hedgehog promoter (SalI/XhoI fragment) upstream of gfp as well as intronic sequences for required enhancer regions (Müller et al., 1999). The line will be referred to in the following as shh-GFP line. Our lab has used it previously to study the larval expression of shh-GFP (Biechl et al., 2016). Here, we raised zebrafish shh-GFP specimens into larval stages and up to 3 months (early adults). Fish were maintained according to standard protocols (Westerfield, 2007).
The shh-GFP transgenic zebrafish line used here has previously been characterized to faithfully represent shh expression (Biechl et al., 2016;Shkumatava et al., 2004) in zebrafish retina and brain/spinal cord and the transgenic expression patterns are furthermore well in line with known shh expression patterns in other vertebrate species (reviewed in Biechl et al., 2016 andBaeuml et al., 2019).
The transgenic islet1-GFP line Tg(isl1:GFP) was originally generated by Higashijima, Hotta, and Okamoto (2000) by fusing gfp sequences with islet-1 promoter sequences (ICP) to produce the core plasmid and adding enhancer elements (CM) for the construct that proved sufficient for specific neural expression. This line will be referred to here as islet1-GFP line. Details for the generation of these specimens, as well as the origin of brain sections depicted in this contribution, are given in a previous paper reporting on islet1-GFP expression (Baeuml et al., 2019).
All procedures involving live zebrafish were carried out according to EU guidelines and German legislation (EU Directive 2010_63, license number AZ 325.1.53/56.1-TU-BS). Transgenic animals used in this study were killed with an overdose of tricaine methanesulfonate  and fixed in paraformaldehyde (4% PFA in Sörensen's phosphate buffer, PB) at 4 C overnight. The raising and fixation of transgenic animals were performed in Prof. Reinhard Köster's lab (Technical University Braunschweig, Germany) and kindly subsequently provided to us. Therefore, the present study only involved fixed animal tissue and needed no further approval.

| Cutting procedure
Following cryoprotection in sucrose solution (30% sucrose solution at 4 C overnight), the brains (heads) of adult shh-GFP zebrafish were embedded in TissueTek (tissue freezing medium, A. Hartenstein GmbH) and cryosectioned (Leica, CM 3050S) at 30 μm in the transverse or sagittal plane before thaw mounted onto Superfrost Plus glass slides (Thermo) and coverslipped after immunoprocedures. In total, 18 zebrafish specimens were used in this study, that is, one specimen each of 3-8 days postfertilization (dpf) larvae, four 13 dpf larvae, and eight 3-month-old specimens. Additionally, various 2, 3, 4, and 5 dpf shh-GFP specimens were available from a previous study (Biechl et al., 2016).

| Immunohistochemical processing
Immunohistochemical incubations were done in a humid chamber. After washing off TissueTek in cryosections with phosphate-buffered saline (PBS) the sections were blocked with blocking buffer (2% normal goat serum, 2% bovine serum albumin, 0.2% Tween20, 0.2% TritonX-100 in PBS) for 1 hr at RT before exposure to a primary antibody against GFP diluted in blocking buffer at 4 C for 1-3 days (dilutions see Table 1).
After washing in PBT (PBS + 0.1% Tween 20), the sections were incubated with the secondary antibody (see Table 1) diluted in blocking buffer solution overnight at 4 C. Subsequently, a second primary antibody against tyrosine hydroxylase (TH; see Table 1) was applied after intermittent washing in PBT and blocking (see above for details), followed by the application of the appropriate secondary antibody (see Table 1) diluted in blocking buffer overnight, after intermittent washing in PBT and blocking (see above). Finally, sections were washed in PBT and counterstained with DAPI (4 0 -6-diamidino-2-phenylindole; Carl Roth, 1:1000) and washed in PBS. Slides were then mounted with Vectashield (Vectorlabs) or ProLong Diamond (Invitrogen/Thermo Fisher) and coverslipped. Previously, various controls and Western blot analysis for the antibody against TH have been performed (Yamamoto, Ruuskanen, Wullimann, & Vernier, 2010.
Furthermore, there were no neuroanatomical differences between the intrinsic GFP signal with the one enhanced through the use of the anti-GFP antibody.
All images were eventually slightly adapted for brightness and contrast with Corel PHOTO-PAINT 9.0 and mounted into figures with Corel DRAW 9.0 (Corel Corporation, Ottawa, Canada).

| Analysis of data
Most sections were photographed in three appropriate fluorescent spectral channels for the presence of the nuclear stain DAPI, shh-GFP or islet1-GFP, and TH. In cases where the GFP and TH label was in the same area, the ImageJ tool of synchronizing all windows was used to analyze cellular colocalization of shh-GFP with TH on a neuroanatomical background yielded by the DAPI pictures. Since the three microphotographs were identical in each case except for the fluorescence visualized, we could assign in detail to a cell nucleus seen in DAPI stain the associated cytoplasmic green GFP and red transmitterrelated enzyme stain on a cell to cell basis.

| RESULTS
The shh-GFP is generally still expressed in early adult zebrafish brains in most locations along the neuraxis as seen in larval zebrafish brains of 4/5 days postfertilization (dpf; see Baeuml et al., 2019;Biechl et al., 2016). These shh domains include classical floor plate cells defining the ventral midline of the neuraxis from spinal cord into the posterior diencephalon. However, the forebrain shows more complex shh-GFP expression patterns involving basal and alar plates. We will describe all shh-GFP expression domains from anterior to posterior levels, along with nuclear DAPI stains and, when necessary, with tyrosine hydroxylase (TH) immunostains. We use for identification of brain structures basically the Neuroanatomy of the Zebrafish Brain atlas (Wullimann, Rupp, & Reichert, 1996). In a recent study (Baeuml et al., 2019), we have detailed and justified some modifications from this atlas in the identification of the paraventricular organ, the intermediate hypothalamic nucleus and the posterior tuberal nucleus also applied here.
3.1 | Analysis of shh-GFP and tyrosine hydroxylase in transverse plane 3.1.1 | Telencephalon and preoptic region Similar to the larval brain (4-8 dpf), there are no shh-GFP cell bodies in the adult subpallium and in the parenchyma of all adult pallial divisions. Different from the larval brain, however, radial glia cells (white arrows in Figure 1  telencephalon (Dp; Figure 1d) suggesting that this convergence area is a pial surface and not a ventricular surface in contrast to that of the medial (Dm), central (Dc) and lateral (Dl) zones of the dorsal telencephalon. The latter zone (Dl) shows no shh-GFP radial glia cells, but many other markers demonstrate the nature of its ventricular surface (see section 4).
In the postcommissural telencephalon, a strong shh-GFP expression domain is present in the most ventroposterior basal pallidal part of the subpallium (BP; Figure 1c,e). This area has not been identified as a separate entity before, including the Neuroanatomy of the Zebrafish Brain adult atlas (where it lies in the area between the anterior parvocellular preoptic nucleus, PPa, and the postcommissural nucleus of the ventral F I G U R E 1 Legend on next page. telencephalon, Vp; page 40, cross-section 107 in Wullimann et al., 1996).
The spot of shh-GFP cells seen at the base of the supracommissural nucleus of the ventral telencephalon (Vs) is the most anterior extension of BP (arrowhead in Figure 1b2). We propose that this basal subpallial expression domain represents the zebrafish homolog of the mammalian pallidal shh expressing domain (Mueller & Wullimann, 2009;Mueller, Wullimann, & Guo, 2008). The shh-GFP cell somata in BP mostly do not lie directly at the ventricular lining and do not exhibit long fibers towards the pial periphery. Thus, they likely are not radial glia cells.
A fair number of shh-GFP cells is seen in the anterior parvocellular preoptic nucleus (PPa; Figure 1b

| Diencephalon
Transverse hind-and midbrain sections are leveled with respect to the caudorostral body axis and thus run horizontally through the forebrain because of the ventral bending of the neural tube's front end. As suggested previously (Herget, Wolf, Wullimann, & Ryu, 2014; their figure 1), it is reasonable to identify in such forebrain sections dorsal as posterior and ventral as anterior to avoid confusion with body axes. Thus, forebrain sections may show from "dorsal" (i.e., posterior) to "ventral" (i.e., anterior) at the same time parts of the (most posterior) prosomere 1 (pretectum; P1), prosomere 2 (thalamus, previously dorsal thalamus; P2), and prosomere 3 (prethalamus, previously ventral thalamus; P3) as well as the (most anterior) hypothalamus (Figures 2-4). It has to be noted that these prosomeres include alar as well as basal plate components (see below). This clarification of neural tube axes is critical in the present investigation that focuses on a ventrally expressed marker (shh) in order to keep attention to the true ventral versus dorsal side of the forebrain.

| Analysis of shh-GFP and tyrosine hydroxylase in sagittal plane
In order to provide additional means of verification and didactically improved visualization of data reported above, we also prepared sagittal sections of shh-GFP transgenic brains immunostained against tyrosine hydroxylase.
An overview of a shh-GFP zebrafish brain ( Overall, this sagittal analysis delivered a highly consistent and corroborating picture but also showed that certain details are veiled F I G U R E 3 Legend on next page. (double-labeled in TPp-p and PVO) whereas other facts are better visualized in sagittal sections (longitudinal distribution of floor plate cells and relationship of ZLI to thalamus and prethalamus). Also, certain most laterally placed shh-GFP cell groups were better to be grasped in transverse sections (preglomerular area, locus coeruleus, dorsal zone of periventricular hypothalamus; see above).

| Colocalization of late larval dopamine cells with shh-GFP
Because of the broad shh-GFP expression in early larval zebrafish brains (Baeuml et al., 2019;Biechl et al., 2016)  However, the amniote floor plate itself appears to extend further anteriorly into the mammillary (caudal) hypothalamus as indicated by longitudinally expressed marker genes, such as the LIM homeobox F I G U R E 3 Analysis of colocalization of shh-GFP and islet1-GFP in the adult zebrafish brain. Transverse sections of an islet1-GFP brain are shown in two left columns (a-f) and of a shh-GFP brain in two right columns (g-l), both for GFP positivity and additionally shown in nuclear DAPI stain. The islet1-GFP data stem from our previous study (Baeuml et al., 2019). (a/g) supracommissural nucleus of ventral telencephalon (Vs). (b/h) anterior parvocellular preoptic nucleus (PPa). (c/i) suprachiasmatic nucleus (SC). (d/j) prethalamus (ventral thalamus; PTh). (e/k) periventricular nucleus of posterior tuberculum (TPp) and ventral zone of periventricular hypothalamus (Hv). (f/l) dorsal zone of periventricular hypothalamus (Hd). Generally, shh-GFP cells are located more ventricularly than islet1-GFP cells, as evidenced by dashed lines in Vs, PPa, Hv, and Hd where shh-GFP cells are always within the lining and islet1-GFP cells are on the outside (i.e., the latter are more distant from the ventricle than the former). For location of shh-GFP cells in TPp see Figure 4. While such ventricularly located shh-GFP cells are also seen in the prethalamus (tier 1 in d), both shh-GFP and islet1-GFP cells exist in tier 2 (or ventromedial thalamus; see section 4). In SC, the medial islet1-cells cells are far apart from the lateral shh-GFP signal. Inset in (i1) shows enlargement of shh-GFP cells in SC. Inset in (j1)  Transverse adult serial sections reveal that most zebrafish forebrain shh-GFP cell bodies lie close to the ventricle, yet they lack the typical cytological nature of floor plate cells seen more posteriorly.
Only in a few forebrain regions are shh-GFP cell bodies observed in more migrated locations. This is expected for the zebrafish ZLI which forms a barrier between prosomeres 2 and 3 and, thus, its migrated cells are present in the diencephalic adult gray matter in this location (best seen in transverse view in Figure 2c2,d2). However, migrated shh-GFP cell bodies are also present in the area of the posterior tuberculum and, to a lesser degree, in the area of the nucleus of the medial longitudinal fascicle (Nmlf) as well as within the (alar plate) prethalamus. We will focus below on ventricularly located adult forebrain shh-GFP cells, and also consider peripherally migrated shh-GFP cells.
All adult shh-GFP positive regions just mentioned are present in larval shh-GFP zebrafish brains in a less differentiated state as will also be discussed below (see sections 4.2-4.4).
However, we will start by discussing three more surprising find-
Moreover, HNF3ß, a floor plate marker induced by notochordal SHH (Echelard et al., 1993) is  Smeets & Reiner, 1994a, 1994bSmeets & González, 2000;Wullimann, 2014). These ventral diencephalic dopamine cells include projection cells to the teleostean striatum (Mueller et al., 2008;. Our shh-GFP data show that among all dopamine neurons in the adult zebrafish brain, exclusively those in the posterior tuberculum are shh-GFP positive (compare Figure 4 and Table 2). This is the case for the parvocellular Finally, beyond these posterior tubercular dopamine groups, a second population of differentiated zebrafish brain neurons that expresses shh-GFP can clearly be identified in the preglomerular complex ( Figure 5). This large migrated diencephalic area is specific for teleosts and involved in processing ascending sensory information

| Newly identified adult shh-GFP cells in zebrafish basal pallidal region (BP) and in pallial radial glia cells (Dm; pallial amygdala)
In tetrapods, as particularly well studied in amniotes, a basal telencephalic (pallidal) sonic hedgehog (shh) expressing center is known to be instrumental for correct basal ganglia development and telencephalic   Wullimann et al., 1996;p. 40, cross-section 107). The

shh-GFP cells continue somewhat into the basal posterior (Vp) and
supracommissural (Vs) nuclei of the ventral telencephalon and maybe slightly even into the ventral part of the dorsal nucleus which represents the differentiated pallidum (Vdv; Mueller et al., 2008;Mueller & Wullimann, 2009). In any case, this shh-GFP population is a basal telencephalic domain located in the ventroposterior subpallium, that is, basal pallidum (BP). This is the first time that the telencephalic shh signaling center in the basal pallidum common to F I G U R E 6 Legend on next page. all tetrapods has unequivocally been visualized in the zebrafish brain.
F I G U R E 6 Floor plate shh-GFP and tyrosine hydroxylase expression in adult zebrafish brain.  Anterior to the midbrain, a morphologically defined floor plate is no longer seen, but Shh is still expressed in basal plate diencephalon (P1-P3) through the hypothalamus up into the (alar) preoptic area. A vertebrate typical deviation from this longitudinal course is seen at the interface of thalamus (P2) and prethalamus (P3) where Shh extends in transverse direction dorsally (Figure 10c). All along its neuraxial course from spinal cord up to preoptic levels, the Shh expression domain is closely accompanied dorsally by a thinner expression stripe of the homeobox gene Nkx2.2 (Price et al., 1992;Qiu et al., 1998;Shimamura et al., 1995). The related Nkx2.1 gene is exclusively expressed in the forebrain, largely overlapping with Shh from P3 into hypothalamus and preoptic region (Figure 10c; Lazzaro et al., 1991;Shimamura et al., 1995;Kimura et al., 1996;Qiu et al., 1998). Furthermore, ventral forebrain Islet1 expressing cells coexpress Nkx2.1 (Ericson, Muhr, Placzek, et al., 1995) in contrast to posterior islet1 cells (where Nkx2.1 is not expressed). In lateral views, the basal telencephalon appears to form an upper floor of Shh expression. However, transverse views reveal that Shh has a continuous expression in the neural wall from the preoptic area (POA) into the so-called anterior entopeduncular area (AEP) which in turn continues dorsally into the most basal division of the medial ganglionic eminence ( Figure 10B2; MGE, i.e., the future pallidum; Asbreuk et al., 2002;Bulfone et al., F I G U R E 7 Sagittal analysis: Overview of shh-GFP and tyrosine hydroxylase expression in adult zebrafish brain. Parasagittal section shown for shh-GFP immunostain (a1) and shown for nuclear DAPI stain (a2). Enlargements (frames a through d in a1) show telencephalon (a), posterior tuberculum (b1-b3), hypothalamus (c1-c2) and hindbrain (d) in these two stains plus tyrosine hydroxylase (TH) immunostains when appropriate. Note that all shh-GFP cells in pallium are radial glia cells at the wrinkled medial and dorsal surface of the medial zone of the dorsal telencephalon (compare with transverse sections in Figure 1).  (Figure 10c; Puelles et al., 2000;Shimamura et al., 1995).
F I G U R E 8 Sagittal analysis: Focus on diencephalon of adult transgenic shh-GFP zebrafish. Four parasagittal sections of a shh-GFP transgenic zebrafish brain through preoptic region, prethalamus/thalamus, pretectum, posterior tuberculum and hypothalamus showing nuclear DAPI stain (a1-d1), shh-GFP (a2-d2) and tyrosine hydroxylase (TH; a3-d3) immunostains. Note the absence of overlap of shh-GFP and TH in preoptic region, pretectum, prethalamus, and caudal/ventral zone of periventricular hypothalamus whereas in TPp-m cells these two markers colocalize (yellow arrows). For colocalization in adult PVO and TPp-p, see Figure 4 and text.  (Qiu et al., 1998). Expression of Nkx6.2 is similar to Nkx6.1 in hindbrain and midbrain, but absent in spinal cord F I G U R E 9 Legend on next page. and in P3 through hypothalamus (Qiu et al., 1998). However, Nkx6.2 is again expressed in the most dorsal MGE (Fogarty et al., 2007).
Two conclusions follow from these studies. First, the differential ventrodorsal gene expression patterns along the neuraxis share similarities into the telencephalon (e.g., Shh; nkx genes) suggesting ventrodorsal induction also there. Second, local differences in longitudinal gene expression (e.g., Nkx2.1 vs. Nkx6.1) suggest that the mechanisms of induction of ventral phenotypes along the anteroposterior axis differ (Balaskas et al., 2012;Litingtung & Chiang, 2000;Placzek & Briscoe, 2005). We will focus on data directly relevant to forebrain ventralization, in particular, the telencephalon, for which we report new results in the zebrafish.

| What is the role of SHH in telencephalic gene expression induction and repression?
The differential transcription factor expression described above for amniotes presents a distinctly nested anteroposterior and ventrodorsal F I G U R E 9 Expression of shh-GFP and tyrosine hydroxylase in the late larval/juvenile (13d) zebrafish brain. Transverse sections run from alar diencephalon (a; pretectum/thalamus/prethalamus), through posterior tuberculum (b), into three levels of hypothalamus (c-e). Designations and Arabic numbers are used as established for larval zebrafish brain by Rink and Wullimann (2002). White arrows in (b1) point to tectal ventricle. Note that unlike in the adult zebrafish brain, the posterior tuberal nucleus (PTN) contains shh-GFP cells, which colocalize with tyrosine hydroxylase (TH) and, maybe also the caudal zone of the periventricular hypothalamus (Hc, around posterior recess forebrain patterning. These patterns principally result from the combined spatiotemporally dynamic activity of inductive signals (morphogens) from various sources (signaling centers), followed by cross-regulatory interactions of homeodomain and bHLH gene activity (Campbell, 2003;. The signaling centers include in addition to the ventral notochord/prechordal mesoderm and later ventral neural tube (SHH), also the anterior neural ridge (fibroblast growth factor, FGF8), the dorsal neural tube midline/cortical hem (bone morphogenetic protein 4-BMP4 and Wnt3a), the lateral mesoderm (retinoic acid, RA) and the transverse zona limitans intrathalamica (SHH) as F I G U R E 1 0 Legend on next page.
In addition to this information in amniotes, a similar situation is present in basal anamniote sarcopterygians (frogs, salamanders, and  In teleosts, general forebrain expression patterns also agree with those in tetrapods, for example regarding pallial versus subpallial gene expression (reviewed in Wullimann, 2009;Mueller & Wullimann, 2009, 2016, as well as regarding the process of tangential migration (Mueller et al., 2008;Mueller, Vernier, & Wullimann, 2006). Two paralogs of nkx2.1 (a/b) exist in zebrafish (Manoli & Driever, 2014;Rohr, Barth, Varga, & Wilson, 2001), with nkx2.1b expressed in the embryonic/larval pallidum and nkx2.1a in hypothalamus. This is confirmed by larval expression of both "pallidal" genes lhx6/lhx7 which are expressed in a ventral subdivision of the dorsal nucleus of the zebrafish ventral telencephalon (Mueller et al., 2008) and by corresponding adult pallidal islet1 expression (Vdv; Baeuml et al., 2019). As discussed in this previous paper, we interpret the adult In the cerebellum, shh expressing Purkinje cells act in transit amplification in the external granular layer. In the early mammalian isocortex, shh expression was reported in radial glia cells (Wang, Hou, & Han, 2016) and other cortical cells in intermediate zone, subplate, and deep cortical plate cells (Radonjic et al., 2016). Also, shh is more strongly expressed in gyrencephalic species (primates) than lissencephalic mammalian brains (rodents) in the developing cortical ventricular zone and apparently plays a role in the multiplication of progenitors (outer radial glia/intermediate progenitors; Han, 2016).
We demonstrated recently that shh expressing cells are also found in the larval zebrafish optic tectum and cerebellum (Biechl et al., 2016), but no such cells are seen in the larval pallial telencephalon.
However, here we show newly a shh-GFP expressing population in the adult zebrafish pallial telencephalon, that is, pallial radial glia cells.
Various studies have shown that mitotic stem cells (radial glia) exist in the adult zebrafish telencephalon along the subpallial and pallial ventricular lining (e.g., Chapouton, Jagasia, & Bally-Cuif, 2007;Diotel et al., 2015;Kaslin, Ganz, & Brand, 2007;Lillesaar, Stigloher, Tannhäuser, Wullimann, & Bally-Cuif, 2009;Lindsey, Darabie, & Tropepe, 2012;März, Schmidt, Rastegar, & Strähle, 2010;Than-Trong & Bally-Cuif, 2015). However, to the best of our knowledge, a role for shh has not been shown in adult telencephalic stem cells. Recently, zebrafish telencephalic stem cells were investigated with a transcriptomic approach (Cosacak et al., 2019) and shown to be organized into molecularly separable populations that are clearly closely correlated with earlier established neuroanatomical divisions (Wullimann et al., 1996). One of these stem cell populations is in the medial zone of the pallial telencephalon (Dm; considered the pallial amygdala; Portavella et al., 2002Portavella et al., , 2004Wullimann & Mueller, 2004;Lal et al., 2018) which is characterized by marker genes pou3f1 and dmrta2 (Cosacak et al., 2019). It appears that we show here with shh-GFP specifically this population of molecularly defined radial glia cells within the medial zone of the dorsal (pallial) telencephalon (Dm; Figure 1). Thus, while there is no support for an early role for shh in the developing zebrafish pallium in the literature, a later role for shh in the adult pallium is suggested by our finding of shh-GFP positive radial glia cells in Dm and their function will be interesting to be studied in the future in the context of the known continuing proliferative activity in the zebrafish pallium (see citations above).
4.4 | Analysis of shh-GFP in comparison to islet1-GFP suggests that most forebrain shh cells remain at the ventricle and are not integrated into parenchymal tissue The analysis of transverse zebrafish brain sections reveals that shh-GFP positive cells are generally located close to the ventricular lining. This is evident for the floor plate cells of spinal cord and hindbrain which remain the only shh-GFP cells there also in the adult brain with the exception of a few locus coeruleus cells. Floor plate cells are also seen in the adult midbrain tegmentum (T; Figures 6 and 7), although in larvae many more cells appear to be present there compared to the hindbrain (Figure 9; Biechl et al., 2016;Baeuml et al., 2019). A position close to the ventricle is also seen for most shh-GFP cells in the forebrain, starting with the most caudal ones in the basal plate of P1 which partly are clearly identifiable as floor plate cells (see above).
However, there are a number of more migrated shh-GFP cells seen in this area of the nucleus of the medial longitudinal fascicle (Nmlf).
Such peripherally migrated shh-GFP positive cells become more abundant at the level of the ZLI and anterior to it in the area of the posterior tuberculum (Figure 2). Because dopamine cells in this area form various well-known brain nuclei, we consider sections additionally stained for tyrosine hydroxylase (TH) in detail in section 4.2.
However, in the hypothalamus, shh-GFP cells remain again rather close to the ventricle and this is also true for the telencephalon (see above).
We have recently analyzed in detail the expression of islet1 using a transgenic islet1-GFP line (Baeuml et al., 2019) and since islet1 expressing cells are generally considered to be influenced by ventricularly located SHH secreting cells (see above), we looked at comparable levels of islet1-GFP and shh-GFP transverse zebrafish brain sections at 3 months and evaluated qualitatively their positions with respect to the ventricle. Clearly, at telencephalic and preoptic into hypothalamic levels, islet1-GFP cells are always located more peripherally remote from the ventricle than shh-GFP cells (see Figure 3, where dashed lines enclose more ventricularly located shh- Figure 3g through Figure 3l and exclude more migrated islet-GFP cells in Figure 3a through Figure 3f). Thus, the zebrafish diencephalon is largely similar compared to the more posterior brain with respect to the ventricular position of shh-GFP cells which do not migrate into the brain parenchyma. This is in line with the working hypothesis that shh cells act through this morphogen on nearby cells to express islet1.

GFP cells in
Also in the (alar) prethalamus, the most ventricularly located layer 1 is free of islet1-GFP, whereas many shh-GFP cells are present there.
In contrast, more peripheral prethalamic layers 2 and 3 contain both GFP gene markers. However, because TH is colocalized only with islet1-GFP (Baeuml et al., 2019), but never with shh-GFP (this study), these are not the same cells. Thus, generally shh-GFP and islet1-GFP label in adult zebrafish brain structures do not colocalize on the cellular level (for the only possible exception see section 3.2 and Table 2).

ACKNOWLEDGMENTS
We thank Bea Stiening for various laboratory-related support, Alex Kaiser for helpful suggestions on the manuscript, and the Depart-

CONFLICT OF INTEREST
The authors declare that there are no conflicts of interest.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.