The Shark Alar Hypothalamus: Molecular Characterization of Prosomeric Subdivisions and Evolutionary Trends.

The hypothalamus is an important physiologic center of the vertebrate brain involved in the elaboration of individual and species survival responses. To better understand the ancestral organization of the alar hypothalamus we revisit previous data on ScOtp, ScDlx2/5, ScTbr1, ScNkx2.1 expression and Pax6 immunoreactivity jointly with new data on ScNeurog2, ScLhx9, ScLhx5, and ScNkx2.8 expression, in addition to immunoreactivity to serotonin (5-HT) and doublecortin (DCX) in the catshark Scyliorhinus canicula, a key species for this purpose since cartilaginous fishes are basal representatives of gnathostomes (jawed vertebrates). Our study revealed a complex genoarchitecture for the chondrichthyan alar hypothalamus. We identified terminal (rostral) and peduncular (caudal) subdivisions in the prosomeric paraventricular and subparaventricular areas (TPa/PPa and TSPa/PSPa, respectively) evidenced by the expression pattern of developmental genes like ScLhx5 (TPa) and immunoreactivity against Pax6 (PSPa) and 5-HT (PPa and PSPa). Dorso-ventral subdivisions were only evidenced in the SPa (SPaD, SPaV; respectively) by means of Pax6 and ScNkx2.8 (respectively). Interestingly, ScNkx2.8 expression overlaps over the alar-basal boundary, as Nkx2.2 does in other vertebrates. Our results reveal evidences for the existence of different groups of tangentially migrated cells expressing ScOtp, Pax6, and ScDlx2. The genoarchitectonic comparative analysis suggests alternative interpretations of the rostral-most alar plate in prosomeric terms and reveals a conserved molecular background for the vertebrate alar hypothalamus likely acquired before/during the agnathan-gnathostome transition, on which Otp, Pax6, Lhx5, and Neurog2 are expressed in the Pa while Dlx and Nkx2.2/Nkx2.8 are expressed in the SPa.

Introduction

The <span class="Disease">hypothalamus is a conserved integrative center that coordinates autonomic, endocrine, and limbic resppan>onses. Its organization is the result of complex patterning processes that converge at the rostral-most part of the neural tube, which in turn, yields a complex structure that has been difficult to systematize (Shimamura et al., 1995; Puelles and Rubenstein, 2003, 2015; Puelles et al., 2004, 2012; Medina, 2008; Szabó et al., 2009; Shimogori et al., 2010; Alvarez-Bolado et al., 2012, 2015; Croizier et al., 2015). As a result, its development has been recently a topic of <span class="Chemical">active research (Alvarez-Bolado et al., 2015).

Classically, the <span class="Species">mammalian <span class="Disease">hypothalamus has been described as consisting of a wide variety of neuronal clusters subdivided in four regions: preoptic, anterior, tuberal, and mamillary (Simerly, 2004). This organization relies on conceptions that understood the brain to be organized in functional columns (also referred as “columnar models”), in which the hypothalamus would be located ventrally to the remaining diencephalon. Alternative brain conceptions have understood the brain to be divided in transverse segments (or neuromeres) yielding segmental (or neuromeric) paradigms of brain organization (Puelles, 2009). Modern segmental paradigms, such as the “prosomeric model,” recognize the mentioned hypothalamic regions under an alternative axial concept that relies on known developmental processes and in neuromeric organization (Puelles and Rubenstein, 2003, 2015; Puelles, 2009; Puelles et al., 2012). Furthermore, the prosomeric model also considers that these segments are organized in different histogenetic territories defined by neuroepithelial transcription factor specification codes and radial units (Puelles et al., 2012; Nieuwenhuys and Puelles, 2016). This paradigm understands the hypothalamus to be ventrally located with respect to the telencephalon (thus, rostral to the remaining diencephalon) forming together a pair of hypothalamo-telencephalic segmental units (hp1, hp2) at the rostral-most part of the neural tube. Under these criteria the preoptic area is recognized as part of the subpallial telencephalon rather than belonging to the hypothalamus (Flames et al., 2007; Medina and Abellán, 2012; Puelles et al., 2012).

The updated prosomeric model recognizes the <span class="Disease">hypothalamus to be subdivided in dorso-ventrally arranged histogenetic domains. The intrahypothalamic border (<span class="Chemical">IHB) divides these histogenetic domains in rostral (hp2 or terminal) and caudal (hp1 or peduncular) portions. As a result, in the alar plate the hypothalamus presents at least four progenitor subdomains (from rostral to caudal and dorsal to ventral): terminal and peduncular paraventricular area (TPa, PPa; respectively), terminal and peduncular subparaventricular area (TSPa, PSPa; respectively). Further dorso-ventral subdivisions have also been proposed (Morales-Delgado et al., 2011, 2014; Puelles et al., 2012; Díaz et al., 2015; Ferrán et al., 2015; Puelles and Rubenstein, 2015). The relationships among these domains and hypothalamic nuclei have been pointed out: the TPa/PPa domain gives rise to magnocellular and parvocellular neurosecretory populations of the supraopto-paraventricular complex; the TSPa/PSPa domain will form mainly the suprachiasmatic nucleus, the anterior hypothalamic nucleus, and the subparaventricular zone, while the basal hypothalamus gives rise to classical tuberal and mamillary derivatives (Morales-Delgado et al., 2011; Puelles et al., 2012; Herget et al., 2014; Herget and Ryu, 2015). Finally, additional histogenic domains are recognized in the acroterminal territory, the rostral-most portion of the neural tube (see Puelles et al., 2012; Ferrán et al., 2015; Puelles and Rubenstein, 2015; Santos-Durán et al., 2015). This median region is responsible of the development of structures such as the lamina terminalis or the optic chiasm in the alar hypothalamus.

Because of their phylogenetic position as out-group to osteichthyans (the other major phylum of gnathostomes, which includes bony fish and <span class="Species">tetrapods), chondrichthyans are essential to reconstruct gnathostome ancestral characteristics through comparisons with other vertebrate models. Recently, we have carried out a preliminary study of the molecular histogenetic organization of the hypothalamus of an elasmobranch representative, the catshark Scyliorhinus canicula, and we analyzed this organization under the updated prosomeric framework (see Santos-Durán et al., 2015). Such analysis revealed a strikingly high degree in the conservation of hypothalamic histogenetic compartments between chondrichthyan and murine models. Indeed, the alar expression of ScOtp and ScDlx2/5 revealed apparently conserved paraventricular (Pa) and subparaventricular (SPa) progenitor domains. The basal expression of these and other genes lead to the identification of tuberal/retrotuberal (Tu/RTu), perimamillary/periretromamillary (PM/PRM) and mamillary/retromamillary (MM/RM) domains, apparently homologous to those described in murine models. Besides, a molecular hypothalamo-telencephalic border (HTB) and a hypothalamo-diencephalic border (HDB), matching with those described in the model, were identified in the shark, together with an IHB defined, as in mouse (Puelles et al., 2012), based on the course of ascending tracts to the telencephalon (see Santos-Durán et al., 2015).

In such previous study some histogenetic differences were also observed within particular subdomains, but their significance has not been explored so far. Moreover, although many of the boundaries and assumptions predicted by the prosomeric model were confirmed in the chondrichthyan model, further dorso-ventral subdivisions and genetic evidences of rostro-caudal segmentation, particularly concerning the <span class="Disease">alar hypothalamus, have not been previously addressed. Besides, the meaning of the results observed in the shark remains to be analyzed in evolutionary context. For all these reasons, here we have examined more deeply the molecular profile of the <span class="Disease">alar hypothalamus of S. canicula, with three aims: (i) to look for further prosomeric molecular subdivisions, (ii) to better define the molecular alar-basal boundary (ABB) and (iii) to obtain some insights on the evolution of the alar hypothalamus by comparative analysis. To address these questions, previous data on ScOtp, ScDlx2/5, ScTbr1, ScNkx2.1 expression and Pax6 immunoreactivity were revised jointly with new data on ScNeurog2, ScLhx9, ScLhx5, and ScNkx2.8 expression and serotonin (5-HT) and doublecortin (DCX) immunoreactivity. Further details on the rostro-caudal and dorso-ventral molecular organization of the alar hypothalamus obtained with these genes support partially the existence of subdomains similar to those proposed by the prosomeric model. Particularly, terminal and peduncular subdivisions were defined in the Pa and SPa domains, but a dorsoventral subdivision only could be distinguished in the SPa domain.

Materials and Methods

Experimental Animals

Some embryos of the catshark (also known as the lesser <span class="Species">spotted dogfish; <span class="Species">S. canicula) were supplied by the Marine Biological Model Supply Service of the CNRS UPMC Roscoff Biological Station (France) and the Estación de Bioloxía Mariña da Graña of the University of Santiago de Compostela. Additional embryos were kindly provided by the Aquaria of Gijón (Asturias, Spain), O Grove (Pontevedra, Spain) and Finisterrae (A Coruña, Spain). Embryos were staged by their external features according to Ballard et al. (1993). For more information about the relationship of the embryonic stages with body size, gestation and birth, see Table 1 in Ferreiro-Galve et al. (2010). Fifty embryos from stages 18 to 32 were used in this study. Eggs from different broods were raised in seawater tanks in standard conditions of temperature (15–16°C), pH (7.5–8.5) and salinity (35 g/L). Adequate measures were taken to minimize animal pain or discomfort. All procedures conformed to the guidelines established by the European Communities Council Directive of 22 September 2010 (2010/63/UE) and by the Spanish Royal Decree 53/2013 for animal experimentation and were approved by the Ethics Committee of the University of Santiago de Compostela.

Tissue Processing

Embryos were deeply anesthetized with 0.5% <span class="Chemical">tricaine methane sulfonate (<span class="Chemical">MS-222; Sigma, St. Louis, MO, USA) in sea water and separated from the yolk before fixation in 4% Paraformaldehyde (PFA) in elasmobranch’s phosphate buffered [EPB: 0.1 M phosphate buffer (PB) containing 1,75% urea, pH 7.4] for 48–72 h depending on the stage of development. Subsequently, they were rinsed in phosphate buffered saline (PBS), cryoprotected with 30% sucrose in PB, embedded in OCT compound (Tissue Tek, Torrance, CA, USA), and frozen with liquid nitrogen-cooled isopentane. Parallel series of sections (12–20 μm thick) were obtained in transverse planes on a cryostat and mounted on Superfrost Plus (Menzel-Glasser, Madison, WI, USA) slides.

Single Immunohistochemistry (IHC) on Sections

For heat-induced epitope retrieval, sections were pre-treated with 0.01 M <span class="Chemical">citrate buffer (pH 6.0) for 30 min at 95°C and allowed to cool for 20–30 min at room temperature (RT). Sections were then rinsed twice in 0.05 M Tris-buffer saline (<span class="Chemical">TBS; pH 7.4) for 5 min each and incubated overnight with the primary antibody (rabbit anti-Pax6 polyclonal antiserum, Covance, Emeryville, CA, USA, diluted 1:400; polyclonal rabbit anti-Sonic Hedgehog [anti-Shh], Santa Cruz Biotechnology, Santa Cruz, CA, USA, diluted 1:300; polyclonal rabbit anti-doublecortin [anti-DCX] Cell Signaling; diluted 1:300–500; rabbit anti-serotonin [anti-5-HT] polyclonal antiserum, DiaSorin, Immunostar, Hudson, WI, USA, diluted 1:5000). Appropriate secondary antibody (horseradish peroxidase [HRP])-conjugated goat anti-rabbit were incubated for 2 h at RT. For immunofluorescence, appropriate secondary antibody was used (DAR546 [Alexa 546-conjugated donkey anti-rabbit] Molecular Probes, Eugene, OR, USA, diluted 1:100). Sections were rinsed in distilled water (twice for 30 min), allowed to dry for 2 h at 37°C and mounted in MOWIOL 4-88 Reagent (Calbiochem, MerkKGaA, Darmstadt, Germany). All dilutions were made with TBS containing 15% donkey normal serum (DNS; Millipore, Billerica, MA, USA), 0.2% Triton X-100 (Sigma) and 2% bovine serum albumin (BSA, Sigma).

Controls and Specificity of the Antibodies

No immunostaining was detected when primary or secondary antibodies were omitted during incubations. For details about the specificity of anti-5-HT and anti-<span class="Gene">DCX antibodies, see Pose-Méndez et al. (2014). For details about the specificity of anti-<span class="Gene">Pax6 antibody, see Quintana-Urzainqui et al. (2014). The polyclonal anti-Shh antibody (Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA) was raised in rabbit against the amino acids 41–200 of Shh human protein. The in situ hybridization (ISH) results were similar to those obtained by immunohistochemistry (IHC), and therefore validate the specificity of the anti-Shh antibody used here.

In situ Hybridization (ISH) on Sections and Whole Mounts Embryos

We applied ISH for Sc<span class="Gene">Otp (Santos-Durán et al., 2015), Sc<span class="Gene">Dlx2 (Quintana-Urzainqui et al., 2012, 2015; Compagnucci et al., 2013; Debiais-Thibaud et al., 2013; Santos-Durán et al., 2015), ScDlx5 (Compagnucci et al., 2013; Debiais-Thibaud et al., 2013; Santos-Durán et al., 2015), ScLhx9 (Pose-Méndez et al., 2015; Quintana-Urzainqui et al., 2015), ScLhx5, ScNkx2 (Quintana-Urzainqui et al., 2012, 2015; Santos-Durán et al., 2015), ScNkx2.8, ScTbr1 (Quintana-Urzainqui et al., 2015; Santos-Durán et al., 2015), and ScNeurog2 genes. These probes were selected from a collection of S. canicula embryonic cDNA library (mixed stages S9–S22), constructed in pSPORT1, and submitted to high throughput EST sequencing. cDNA fragments were cloned in pSPORT vectors. Sense and antisense digoxigenin-UTP-labeled and fluorescein-UTP-labeled probes were synthesized directly by in vitro transcription using as templates linearized recombinant plasmid DNA or cDNA fragments prepared by PCR amplification of the recombinant plasmids. ISH in whole mount and on cryostat sections was carried out following standard protocols (Coolen et al., 2009). Briefly, sections were permeabilized with proteinase K, hybridized with sense or antisense probes overnight at 65°C and incubated with the alkaline phosphatase-coupled anti-digoxigenin and anti-fluorescein antibody (1:2000, Roche Applied Science, Manheim, Germany) overnight at 4°C. The color reaction was performed in the presence of BM-Purple (Roche). Control sense probes did not produce any detectable signal. Immunohistochemistry was performed after ISH to obtain double ISH-IHC staining as described above.

Inhibition of the Shh Pathway

Inhibition of the <span class="Gene">Shh pathway was performed by in ovo injection of the pharm<span class="Chemical">acological inhibitor cyclopamine to test if the initiation of ScNkx2.8 expression in the forebrain, specifically in the ABB, is dependent on Shh, similarly to other Nkx paralogs like Nkx2.1 or Nkx2.2. First, 200 μL of a solution containing 1X PBS, 500 μM cyclopamine and 5% dimethyl sulfoxide (DMSO) were injected through the shell of stage 15–16 S. canicula eggs. This solution was replaced by the same volume of 5% DMSO in 1X PBS for control embryos. The eggs were maintained for 3 days in oxygenated sea water at 17°C, with viabilities higher than 90%. Embryos reached stage 18 in these conditions. They were dissected, fixed in PFA 4%, dehydrated and stored in methanol 100% prior to ISH.

Image Acquisition and Analysis

Light field images were obtained with an Olympus BX51 microscope equipped with an Olympus <span class="Gene">DP71 color digital camera. Fluorescent sections were photographed with an epifluorescence photomicroscope Olympus AX70 fitted with an Olympus DP70 color digital camera. Photographs were adjusted for <span class="Disease">brightness and contrast and plates were prepared using Adobe Photoshop CS4 (Adobe, San Jose, CA, USA).

Results

Genes whose expressions define broader domains (Sc<span class="Gene">Otp and Sc<span class="Gene">Dlx2/5) are described first and different subdomains are defined by analyzing the Pax6 immunoreactivity pattern and the expression pattern of the following genes: ScTbr1, ScNeurog2, ScLhx9, ScLhx5, ScNkx2.1, ScNkx2.8.

ScOtp Expression

An overview of the expression of Sc<span class="Gene">Otp in the <span class="Disease">hypothalamus of S. canicula from early to late stages of development has been previously reported (Santos-Durán et al., 2015). In the alar plate ScOtp is mainly expressed in the Pa territory, and therefore it is helpful in recognizing this histogenetic domain. To deepen in the genoarchitectonic profile of this compartment, in this study we have checked the expression of ScOtp mainly by the analysis of transverse sections through the hypothalamus of embryos from stage 29 to 32.

At stage 29, Sc<span class="Gene">Otp is expressed in the surroundin<span class="Disease">gs of the optic stalk and caudally beyond, in what mainly represents the Pa histogenetic domain, as seen in sagittal (Figure ) and transverse (Figures ) sections. ScOtp is recognized in individual cells in the TPa domain (Figure ) mostly located in the marginal zone (blue arrowhead in Figure ) but also scattered through the ventricular zone. Similarly, in the PPa domain (Figure ), ScOtp is mainly expressed in the marginal zone (blue arrowheads in Figure ), while scarce ScOtp-expressing cells are observed in the ventricular zone. Furthermore, ScOtp-expressing cells are recognized in the marginal zone of territories placed dorsal and ventral with respect to the TPa/PPa domain, forming a continuous stream with the marginal ScOtp-expressing cells of this domain. Particularly noticeable are the strings of ScOtp-expressing cells dorsally extended from the TPa domain into the subpallial territory (black arrowheads in Figures ) and from the rostral-most portion of the PPa domain into the pallium (red arrowheads in Figures ). ScOtp-expressing cells are also observed in the marginal zone ventral to the TPa/PPa domain. These cells cannot be observed at the rostral-most TSPa domain (Figure ) but they spread just caudal from this point into the remaining TSPa/PSPa domain (yellow arrowheads in Figures ). Of note, these ventral ScOtp-expressing cells are distributed into the dorsal-most marginal zone of TSPa/PSPa domains (yellow arrowhead in Figure ) but they spread through the SPa domain by late stage 29 (not shown).

Regionalization of the <span class="Disease">alar hypothalamus and neighbor territories in embryos of (A–D) and Sc<span class="Gene">Dlx2/5 (E–J) expression by means of single in situ hybridization (ISH) (A–J) and/or combined with immunohistochemistry (J) on sagittal (A,D,E,H–J) or transverse (B,C,F,G) sections. Continuous red line: ABB. Dashed red line: IHB. Gray line: HTB. Continuous black line divides dorso-ventral division of the alar hypothalamus into Pa (dorsal) and SPa (ventral) territories. Dashed black line represents subdivisions inside the SPa into SPa dorsal (SPaD) and ventral (SPaV). Asterisk point the PThE (ap3). (A–D) ScOtp expression in the Pa at indicated stages. ScOtp labeling (yellow arrowheads) in the SPa corresponds to marginal cells. Black arrowheads point to ScOtp-expressing cells in the subpallium. Red arrowheads point to ScOtp-expressing cells in the pallium. Blue arrowheads point to marginal ScOtp-expressing cells in the Pa. Yellow arrowheads point to ScOtp-expressing cells ventral to the Pa. (A′) Detail of a region equivalent to that squared area in (A) to show ScOtp-expressing cells ventral to the Pa rostrally. (E–J) ScDlx2/5 expression in the subpallium, SPa and ap3 at indicated stages. Red arrowhead points to the lack of expression in the pallium. White arrowheads point to ScDlx2-expressing (F′) cells in the marginal zone of the Pa. Green arrowhead in (G) points to ScDlx2/5 expression in the RTu. (J) Detail of a region equivalent to that squared in (I). Arrows point to 5-HT-ir tracts coursing in the marginal Pa and SPa. For abbreviations, see list.

This basic pattern is maintained until later stages of development (Figure ) with minor modifications. Sc<span class="Gene">Otp is abundantly expressed in radial domains extending from the ventricular to the marginal zone at stage 32 and also in juveniles (data not shown).

ScDlx2/ScDlx5 Expression

The expression of Sc<span class="Gene">Dlx5 from stage 18 onward and the expression of Sc<span class="Gene">Dlx2 from stage 29 onward have been previously reported in Santos-Durán et al. (2015). Essentially identical results were observed with both genes in the brain of S. canicula from stage 29 onward, so we use ScDlx2/5 at these stages to refer indistinctly to both patterns. However, differences are also observed which will be commented where appropriate (see below). Here, we reexamine these data to additionally analyze the expression of ScDlx2/5 from stage 29 to 32. We further characterize the ABB and possible dorso-ventral and rostro-caudal subdomains of the alar hypothalamus, mainly by the analysis of transverse sections through the hypothalamus.

At stage 29, in the alar plate of the secondary prosencephalon and rostral diencephalon, Sc<span class="Gene">Dlx2/5 is expressed in the subpallium, the SPa domain, and in the prethalamus, while it is also expressed in some subdivision in the basal plate subdivisions (Figures ; see also Santos-Durán et al., 2015). Between the expression domains of Sc<span class="Gene">Dlx2/5 in the subpallium and SPa territory there is a negative wedge-shaped domain spreading from the optic stalk to the pallium that contains the territory of the Pa compartment and the adjacent diencephalic region that corresponds to the prethalamic eminence (PThE), the dorsal-most part of the alar prosomere 3 (asterisk in Figure ; see also Santos-Durán et al., 2015). On transverse sections, the expression of ScDlx2/5 in the SPa compartment can be recognized ventrally to the optic stalk (Figure ). Interestingly, in the subpallium, ScDlx2/5 expression occupies the whole ventricular wall while in the SPa domain it does not (Figures ). In the PSPa compartment, the expression of ScDlx2/5 expands ventrally into domains of the basal hypothalamus (green arrowhead in Figure ) although the extent of the ventricular expression differs between both territories. The ventricular expression of ScDlx2/5 between the SPa domain and rostral diencephalon also differs (data not shown).

Noteworthy, from stage 29 onward, Sc<span class="Gene">Dlx2- but not ScDlx5-expressing cells are observed in the marginal zone of the TPa domain (white arrowheads in Figure ), which is negative for ScDlx5 expression. These cells form a continuous string with the positive ScDlx2/5 domain in the subpallium.

This pattern is maintained until stage 31 (compare Figures ). At this stage, marginal Sc<span class="Gene">Dlx2-expressing cells form a continuous stripe between the subpallium and the <span class="Chemical">TSPa domain (white arrowheads in Figure ). In parasagittal sections, 5-HT-immunoreactive (ir) fibers can be recognized ascending to the telencephalon through the alar peduncular hypothalamus (Figure ; Carrera et al., 2008). From this stage onward, ScDlx2/5 expression is recognizable in the subventricular zone of the SPa (data not shown). At stage 32, ScDlx2/5 expression is maintained both in the alar and basal hypothalamus (data not shown).

ScTbr1 Expression

Sc<span class="Gene">Tbr1 expression was described at stage 25 in Santos-Durán et al. (2015), revealing that, though Sc<span class="Gene">Tbr1 is not expressed in any region of the alar hypothalamus of S. canicula at such stage, it is a useful gene to define its boundaries. Here, we examine the detailed expression of ScTbr1 from stage 29 to 32 in order to know if such usefulness remains throughout development.

At stage 29, Sc<span class="Gene">Tbr1 continues to be expressed in the pallium and the adjacent territory of the rostral alar diencephalon, the PThE (Figures ; see also Figure 6B in Santos-Durán et al., 2015). Such expression abuts the dorsal and caudal border of the PPa area but not the TPa domain (Figures ; also compare red arrowheads in Figures and ). Its expression is roughly complementary to that of ScDlx2/5 in the telencephalon and rostral diencephalon (Figure ; also compare Figures with ; see also Figure 6B in Santos-Durán et al., 2015). Note that the expression of both genes, in turn, is complementary to the ventricular domains of ScOtp in the TPa/PPa area (Figure ; also compare Figures , , with ). Together, the distinct but complementary expression of ScTbr1, ScDlx2/5, and ScOtp domains allow to define the whole alar plate of the secondary prosencephalon and rostral diencephalon, consistently with the prosomeric model.

Regionalization of the <span class="Disease">alar hypothalamus and neighbor territories in embryos of (A–C), Sc<span class="Gene">Dlx2/5 (A′) ScNeurog2 (D–G) and Pax6 immunoreactivity (H–K) by means of single ISH on sections (A–G) and single immunoenzyme staining (H–J) or immunofluorescence (K) on sagittal (A,D,E,H,K) or transverse (B,C,F,G,I,J) sections. (A–C) ScTbr1 expression in the pallium and PThE (asterisk). Red arrowhead points to rostral-most expression in the pallium. (A′) Image results from the overlapping of two parallel sections hybridized with ScTbr1 and ScDlx2/5 probes, respectively. Color for ScTbr1 was digitally converted to pink to ease comparison. The complementary expression of ScTbr1 and ScDlx2/5 defines the Pa domain. (D–G) ScNeurog2 expression in the Pa and pallium. Note that Figures (D,E) present artifacts corresponding to broken tissue. Red arrowhead points to the rostral-most pallium lacking ScNeurog2. (H–K) Pax6 immunoreactivity at indicated stages. Red arrowhead points to marginal Pax6 immunoreactivity at the rostral-most pallium. Black and white arrowheads point to marginal Pax6-ir cells in the peduncular SPa (PSPa) and PBHy. Arrows point to Pax6 immunoreactivity dorsal and ventral to the Pa. Note that the stream of Pax6-ir cells in (K) closely follows the IHB. For other labels, see legend in Figure . For abbreviations, see list.

From stage 31 this pattern is maintained in the <span class="Disease">alar hypothalamus, and the changes observed in the telencephalon were already reported (see Quintana-Urzainqui et al., 2015).

ScNeurog2 Expression

At stage 29, Sc<span class="Gene">Neurog2 is expressed in restricted domains of the alar and basal plates of the prosencephalon and rostral diencephalon (Figures ). In the alar plate, Sc<span class="Gene">Neurog2 is expressed in the ventricular zone of the hypothalamus (TPa/PPa domains; Figures ), as well as in the pallium and the PThE. ScNeurog2 is less intensely expressed in the TPa than in the PPa (compare Figures ). ScNeurog2 expression matches the ventricular domains of ScOtp in the TPa/PPa domain (compare Figures with Figures ) and abuts the ScDlx2/5-expressing TSPa/PSPa domain (compare Figures with Figures ). At odds with ScOtp, ScNeurog2-expressing cells are not observed in the marginal zones of the TPa/PPa domain or surrounding territories. This basic pattern is maintained until stage 30. At stage 31, ScNeurog2 is downregulated in the alar hypothalamus although it is maintained in the pallium (not shown).

Pax6 Immunoreactivity

In the shark, the basic segmental organization of the forebrain was examined from mid-gestation to late stages of development based on <span class="Gene">Pax6 immunoreactivity by Ferreiro-Galve et al. (2008). Besides, ScPax6 expression, which matched Pax6 immunoreactivity, was analyzed in the whole brain from early to late stages of development by Ferreiro-Galve (2010). However, none of these studies was undertaken using the updated prosomeric model (in fact, the alar hypothalamus as such was not considered in those studies). Here, we revisited these data in the context of the updated prosomeric model from stage 28 until stage 32.

At stage 28, <span class="Gene">Pax6 immunoreactivity is observed in the ventricular zone of a domain continuous between the alar hypothalamus and both the telencephalic pallium (Figures ) and the rostral diencephalon including the PThE (data not shown). Pax6 immunoreaction rather contrasts with the ScDlx2/5-expressing TSPa/PSPa domain (compare Figure with Figure ), though some overlapping exists between both developmental genes at ventricular levels in the dorsal and ventral part of this domain (see black arrows in Figures ). Besides, Pax6-ir cells are observed in the marginal zone of a restricted area of the PSPa domain (black arrowheads in Figures ).

From stage 30 onward, marginal <span class="Gene">Pax6-ir cells are abundant throughout the diencephalon, including the basal plate, tegmental areas (Figure ; see also Ferreiro-Galve et al., 2008), as well as in the peduncular alar and <span class="Disease">basal hypothalamus; in contrast, they are largely absent at the terminal hypothalamus, thus emphasizing the IHB (Figure ; see also IHB in Figure 6B in Santos-Durán et al., 2015).

This basic pattern is maintained until late stages of development but <span class="Gene">Pax6 immunore<span class="Chemical">activity is intense in cells that populate the mantle zone of alar and basal plates in the prosencephalon. As previously reported, they are especially abundant in the PPa area, rostral diencephalon and pallium (Ferreiro-Galve et al., 2008; Rodríguez-Moldes, 2009; Quintana-Urzainqui et al., 2015).

ScLhx9 Expression

Sc<span class="Gene">Lhx9 expression has been previously reported in the meso-isthmo-cerebellar region of <span class="Species">S. canicula from stage 25 to stage 27 (Pose-Méndez et al., 2015) and in the pallium (medial, ventral) of embryos from stage 31 onward (Quintana-Urzainqui et al., 2015). Here, we report that at early stages (25–28), ScLhx9 is also expressed in restricted domains of the secondary prosencephalon and rostral diencephalon. From stage 29 onward, ScLhx9 is also expressed in restricted domains of the secondary prosencephalon and rostral diencephalon, i.e., it is expressed dorsal, ventral and caudal to the alar hypothalamus but not in it, as can be seen in sagittal (Figures ) and transverse sections through the hypothalamus (Figure ). These territories include ventricular and/or ScLhx9-expressing marginal cells in restricted areas of the pallium and rostral diencephalon (Figure ), subpallium and basal hypothalamus (Figure ).

Regionalization of the <span class="Disease">alar hypothalamus and/or neighbor territories in embryos of , and Sc<span class="Gene">Lhx5 (F–J) on sagittal (A–C′,E–G,J) or transverse (D,H,I) sections. Some of them were double labeled for immunohistochemistry against Shh (E,G,H–J). (A–E) ScLhx9 expression in the limits of the alar hypothalamus, pallium, PThE (asterisk) and basal hypothalamus, at indicated stages. Black arrowhead points to ScLhx9 expression in the basal hypothalamus just ventral to the SPa. Red arrowhead points to ScLhx9-expressing cells in the pallium. (C) Section medial to (B). (C′) Detail of the squared region in (C) showing absence of ScLhx9 expression from the rostral-most alar hypothalamus. (F–J) ScLhx5-expression in different regions of the prosencephalon including the alar hypothalamus. (I) Red arrowhead points to ScLhx5-expressing cells in the pallium. (J′) Sagittal section lateral to (J). Black arrowhead points to a string of ScLhx5-expressing cells extending along the alar hypothalamus (PPa) and ap3. For other labels, see legend Figure . For abbreviations, see list.

In the alar plate, Sc<span class="Gene">Lhx9-expressing cells are only observed in the marginal zone of the pallium and the rostral diencephalon (compare Figures ). In the <span class="Disease">basal hypothalamus, ScLhx9 is expressed in a restricted longitudinal and dorsal domain that abuts the PSPa domain (black arrowheads in Figures ).

This basic pattern is maintained until late stages of development (Figure ) with minor modifications concerning marginal Sc<span class="Gene">Lhx9-expressing cells in the pallium (see also Quintana-Urzainqui et al., 2015).

ScLhx5 Expression

At stage 29, the earliest we have studied, Sc<span class="Gene">Lhx5 is expressed in many different restricted domains of the alar and basal plates of the prosencephalon including pallium, subpallium, <span class="Disease">alar and basal hypothalamus and rostral diencephalon, as observed in sagittal (Figures ) and transverse (Figures ) sections through the hypothalamus.

In the alar plate of stage 29 and 30 specimens, Sc<span class="Gene">Lhx5-expressing cells are observed in the marginal zone of pallial territories dorsal to the <span class="Gene">PPa, similar to ScLhx9 pattern (red arrowhead in Figure ; compare with ScLhx9 in Figure and also with ScTbr1 expression in Figure ). From this stage onward, ScLhx5 is also expressed in the rostral diencephalon (Figure ).

At stage 31 (Figures ), Sc<span class="Gene">Lhx5 is expressed in the subventricular zone and mantle of the Pa domain. Of note, it is more extensively expressed in the <span class="Chemical">TPa territory than in the PPa. However in the PPa territory, a noticeable string of cells continuous with that of the rostral diencephalon is observed (black arrowhead in Figure ). At stage 32, intense ScLhx5 signal was observed only in the telencephalon while it decreases in the hypothalamus (data not shown).

ScNkx2.1 Expression

Sc<span class="Gene">Nkx2.1 expression and its comparison with that of <span class="Gene">Shh were previously reported from early stages of development up to stage 29 in Santos-Durán et al. (2015). Though both genes are not observed in the alar hypothalamus, they are useful to determine boundaries with neighbor territories of the basal hypothalamus and/or the subpallium. Here, we have additionally analyzed the expression of ScNkx2.1 from stage 29 to 32 to further characterize the ABB later in development and the possible dorso-ventral and rostro-caudal subdomains of the alar hypothalamus, mainly by the analysis of transverse sections through the hypothalamus. With the same aim, we have comparatively analyzed ScNkx2.1 expression and Shh immunoreactivity at stage 29.

At stage 29, Sc<span class="Gene">Nkx2.1 is expressed ventral to the optic stalk through the <span class="Disease">basal hypothalamus except in the RM compartment (not shown), forming a sharp limit with the ABB (see continuous red line in Figures ; see also Santos-Durán et al., 2015). Furthermore, ScNkx2.1 and Shh immunoreactivity co-distribute in the Tu region (Figures ), though Shh immunoreactivity does not match the dorsal border of ScNkx2.1 expression. Notably, the dorsal border of ScNkx2.1 appears to abut the ventral border of ScDlx2/5 in the TSPa/PSPa area (Figure ). Dorsal to the alar hypothalamus, ScNkx2.1 expression and Shh immunoreactivity also co-distribute in the preoptic subpallium (data not shown; see Quintana-Urzainqui et al., 2012, 2015). Both genes abut the dorsal limit of the alar hypothalamus (Figures ).

Regionalization of the <span class="Disease">alar hypothalamus and neighbor territories in embryos of (A–C,D″) and Sc<span class="Gene">Nkx2.8 (D,E) on sagittal (A) and transverse (B,E) sections. Some sections were labeled (D′) or double labeled for immunohistochemistry against Shh (B–D′). (C′) Detail of the squared regions in Figure (C). It results from the overlapping of two parallel sections hybridized with ScDlx2/5andScNkx2.1 probes, respectively. Color for ScDlx2/5 and ScNkx2.1 was digitally converted to brown and purple to ease comparison. (A–C) ScNkx2.1 expression at indicated stages. (D,E) ScNkx2.8 expression at indicated stages. Note that Shh immunoreactivity does not reach the ABB (D′,D″). For other labels, see legend Figure . For abbreviations, see list.

This basic pattern is maintained until late stages of development.

ScNkx2.8 Expression

The expression of <span class="Gene">Nkx2.8 ortholog has not been described during brain development of other vertebrates. However, its expression pattern fairly coincides with that of other genes overlapping over the <span class="Chemical">ABB such as Nkx2.2 or Nkx2.9 (Puelles et al., 2012). To characterize ScNkx2.8 expression here we analyzed the expression patterns of ScNkx2.8 through development and, particularly, its expression at stage 25 combined with anti-doublecortin (DCX) immunohistochemistry to identify pioneering tracts.

The expression of Sc<span class="Gene">Nkx2.8 was analyzed from stage 18 until midgestation of development in supplemental material (see Supplementary Figure ). From stage 28 onward, two dorso-ventrally arranged subdomains can be differentiated throughout the longitudinal expression of Sc<span class="Gene">Nkx2.8, the dorsal subdomain being more intense than the ventral one (see continuous red line in Figures ). ScNkx2.8 expression is observed ventral to the optic stalk (Figures ), overlapping the territory where the basal expression of ScNkx2.1 appears to abut the alar expression of ScDlx2/5 (see Figure ). The dorsal intense domain is expressed in the ventral-most TSPa/PSPa domain or supraliminal alar area (Figures ). The ventral less intense domain corresponds to the dorsal-most Tu/RTu area or subliminal basal area (Figures ). Furthermore, Shh immunoreactivity abuts the ventral portion of ScNkx2.8 but not the ABB (Figures ), at least at ventricular levels.

The mentioned pattern is maintained at later stages of development. However, the expression in the forebrain becomes downregulated (although still recognizable) from stage 31 onward (data not shown).

Discussion

We have identified terminal (rostral) and peduncular (caudal) subdivisions in the prosomeric paraventricular and subparaventricular areas by analyzing the expression pattern of various developmental genes (Sc<span class="Gene">Otp, Sc<span class="Gene">Dlx2/5, ScTbr1, ScNkx2.1, ScNeurog2, ScLhx9, ScLhx5, and ScNkx2.8) and immunoreactivity against molecular markers such as Pax6, 5-HT and DCX. While we are aware that those molecular markers do not define per se histological domains, their expressions have been described in homologous regions in different species, and then, they have been largely shown to be useful in the attempt of recognizing particular domains. However, the functional role of many of the genes considered and the biological significance of their expressions are not always deeply understood and thus we cannot discard that their expression dynamics in late embryos could not account for different subdomains but rather for functions unrelated to patterning.

Alar-Basal Boundary

The <span class="Chemical">ABB is an important landmark continuously defined in the prosomeric model (Puelles and Rubenstein, 1993, 2003, 2015; Puelles et al., 2012). In <span class="Species">mice, this linear forebrain boundary is defined between the early Pax6-positive alar plate and the Shh- and Nkx2.1-positive basal plate (Puelles et al., 2012). Therefore, it is a virtual or lineal boundary located between the plasma membrane of cells located in alar and basal plates. Moreover, it is well-known that the gene Nkx2.2 (and other genes co-expressing with it such as Nkx2.9, Ptc and Gsx) overlap over this boundary being expressed in restricted domains within alar and basal plates (Puelles et al., 2012).

Although, in a previous work we defined the shark <span class="Chemical">ABB mainly based on the expression of Sc<span class="Gene">Shh (Santos-Durán et al., 2015), our present data made us to reconsider this boundary since we found that ScDlx2/5 and ScNkx2.1 were the unique pairs of alar-basal markers that abut at a certain point (Figures ). Moreover, as described for Nkx2.2 in other vertebrates, we found that ScNkx2.8 overlap the virtual line defined by ScDlx2/5 and ScNkx2.1 (Figures ). Furthermore, in the alar plate, ScNkx2.8 abuts Pax6 immunoreactivity dorsally and co-distributes with ScDlx2/5 (Figure ) while in the basal plate, ScNkx2.8 co-distributes with ScNkx2.1 and abuts Shh immunoreactivity ventrally (Figure ), invalidating our previous definition. Therefore, we think that in the shark the abutted expression of ScDlx2/5-ScNkx2.1, which can also be followed by the expression of ScNkx2.8 in a band overlapping this linear boundary, represents best the ABB.

Schematic representation of various gene expressions in the <span class="Disease">telencephalon, hypothalamus and rostral diencephalon of (A) Domains and subdomains defined by Sc<span class="Gene">Dlx2/5, ScOtp, ScLhx9, ScNeurog2, ScTbr1, and ScNkx2.1 expression patterns. Some of these genes define subdivisions in the pallium. (B) Domains and subdomains defined by ScDlx2/5, ScOtp, ScLhx9, ScLhx5, Pax6 immunoreactivity, Shh immunoreactivity and ScNkx2.8 expression. Continuous red line represents ABB. Discontinuous black line represents IHB. Dotted line limits acroterminal region. Domains: (1) Subpallium, (2) Pallium, (3) Pa, (4) SPa, (5) PThE, (6) Alar p3, (7) Basal hypothalamus, (8) Pallium-Subpallium boundary, (9) SPaD, (10) SPaV (or supraliminal), (11) SIBHy (or subliminal). For abbreviations, see list.

Although more evidences coming from studies in other vertebrates are needed, the following observations in shark support that the <span class="Gene">Nkx2.8 ortholog is likely expressed over the <span class="Chemical">ABB: (i) ScNkx2.8 expression closely resembles the pattern of Nkx2.2 in other vertebrates over the ABB (Barth and Wilson, 1995; Sugahara et al., 2011; Puelles et al., 2012; Ware et al., 2014; Domínguez et al., 2015), (ii) it co-distributes with the course of the tract of the postoptic commisure (TPOC), which is also assumed to overlap the ABB (see Supplementary Figures ; Barth and Wilson, 1995; Shimamura et al., 1995; Puelles et al., 2012; Ware et al., 2014); (iii) it is expressed partly in the alar plate and partly in the basal plate, as shown by Nkx2.2 in the mouse (see Figures ; see also Puelles et al., 2012; Ware et al., 2014; Puelles and Rubenstein, 2015); and (iv) its expression is induced by the Shh pathway (see Supplementary Figures ) as other paralogs of Nkx family in the forebrain (see Santos-Durán et al., 2015).

Prosomeric Compartments and Subcompartments of the Alar Hypothalamus

Based on the alar complementary expression of Sc<span class="Gene">Otp and Sc<span class="Gene">Dlx2/5, we have previously identified the shark alar hypothalamus harboring a Pa and SPa domains and tentatively defined its rostro-caudal subdivision by the course of the medial forebrain bundle (mfb) through the rostral border of hp1, just caudal to the optic stalk (Santos-Durán et al., 2015). However, we did not find molecular evidence for further rostro-caudal or dorso-ventral subdivisions. Here, we deepen in the genoarchitectonic profile of the shark alar hypothalamus providing molecular insights of their dorso-ventral and rostro-caudal subdivisions, and further analyzing histogenetic implications of the genes expressed. Noteworthy, in the stages considered, the developmental control genes here studied did not reveal further details of the acroterminal territory although the updated prosomeric model (Puelles et al., 2012; Ferrán et al., 2015) proposes that the optic stalk and the optic chiasm belong to acroterminal subdivisions of the Pa and SPa, respectively.

Pa: Genoarchitectonic Profile and Further Subdivisions

The updated prosomeric model defined the Pa as a domain characterized by the expression of <span class="Gene">Otp and the lack of Dlx2/5 (Morales-Delgado et al., 2011; Puelles et al., 2012; Puelles and Rubenstein, 2015). In this territory, Otp expression co-distributes with that of Pax6, Neurog2, Lhx5 or Tbr1 genes (Stoykova et al., 1996; Medina, 2008; Abellán et al., 2010; Osório et al., 2010; Puelles et al., 2012; Ferrán et al., 2015) among others. Restricted expression patterns and certain progenitor subdomains led the authors of the model to divide the Pa into terminal or peduncular (Figure ; TPa, PPa; respectively) and different dorso-ventral subdomains (Morales-Delgado et al., 2011, 2014; Puelles et al., 2012; Ferrán et al., 2015; Puelles and Rubenstein, 2015). Moreover, the course of ascending tracts to the telencephalon through the rostral-most portion of the peduncular hypothalamus supports the existence of rostro-caudal subdivisions (Puelles et al., 2012; Puelles and Rubenstein, 2015).

Schematic representation of various gene expression patterns in the <span class="Disease">telencephalon, hypothalamus and rostral diencephalon of (A) developing <span class="Species">mouse adapted from Puelles et al. (2012); (B) S. canicula at stage 31; (C) lamprey adapted from Martínez-de-la-Torre et al. (2011). Domains and subdomains defined by Otp, Dlx2, Pax6, Neurog2, Lhx9, Lhx5, Tbr1, Nkx2.8, Nkx2.2, Shh, Hh, and Nkx2.1 orthologs. Continuous red line represents ABB. Discontinuous black line represents IHB. Dotted line limits acroterminal region. For abbreviations, see list.

We have previously characterized the shark Pa domain by the expression of Sc<span class="Gene">Otp and we distinguished its rostro-caudal subdivisions by the course of the mfb caudal to the rostral border of hp1 (Santos-Durán et al., 2015). Here we show that in this domain, similarly to what was reported in mice and other tetrapods (Stoykova et al., 1996; Medina, 2008; Abellán et al., 2010; Osório et al., 2010; Puelles et al., 2012; Domínguez et al., 2013, 2015), ScOtp co-distributes with Pax6, ScNeurog2 and ScLhx5 (see Figures ). We have not found evidences supporting dorso-ventral subdivision. Terminal-peduncular subdomains can be sketched by the restricted expression of ScLhx5, but not based on Tbr1-expressing cells, as in mice (Figure ; Puelles et al., 2012; Ferrán et al., 2015), neither on Nkx2.2 expression, as suggested in other vertebrates (Domínguez et al., 2015). In the shark, from stage 30 onward, ScLhx5 seems to be more abundantly expressed in the TPa than in the PPa (Figure ). It forms an evident border (see Figures ; see also Figure ) located rostrally with respect to the 5-HT-ir tracts that course along the mfb (compare Figures with ) in hp1. These facts support the existence of molecular rostro-caudal differences in the Pa territory (Figure ; see also Figure 6B in Santos-Durán et al., 2015; see also Puelles et al., 2012) but also of the IHB as suggested by the prosomeric model. These data also evidence a striking high degree of conservation between the expression patterns found in chondrichthyans and mammals.

SPa: Further Prosomeric Subdivisions

The updated prosomeric model defined the <span class="Chemical">SPa as a domain char<span class="Chemical">acterized by the expression of Dlx2/5 adjacent to the ABB (Morales-Delgado et al., 2011; Puelles et al., 2012; Puelles and Rubenstein, 2015). Gene markers like Lhx1, Nkx2.1 or Nkx2.2, among others are co-expressed with Dlx2/5 in this territory (Shimogori et al., 2010; Puelles et al., 2012; Ferrán et al., 2015). As for the Pa, restricted expression patterns and other evidences suggest that the SPa has a rostro-caudal (terminal TSPa and peduncular PSP) and a dorso-ventral (dorsal SPa [SPaD] and ventral SPa [SPaV] or supraliminal) regionalization (Morales-Delgado et al., 2011; Puelles et al., 2012; Díaz et al., 2015; see also Figure ).

In the shark, we have previously identified the <span class="Chemical">SPa domain by the expression of Sc<span class="Gene">Dlx2/5, the lack of ScOtp, and characterized its rostro-caudal subdivisions by the course of the mfb caudal to the rostral border of hp1 (Santos-Durán et al., 2015). Our present results reveal that Pax6, ScNkx2.8 and ScOtp co-distribute with ScDlx2/5 in the SPa (Figure ). However, these makers are not expressed homogenously through this area, which lead us to recognize four microzones as predicted by the prosomeric model (Morales-Delgado et al., 2011; Puelles et al., 2012; Puelles and Rubenstein, 2015).

A <span class="Chemical">SPaD can be defined by the co-distribution of Sc<span class="Gene">Dlx2/5 and Pax6, the latter extending ventralward beyond Pa into SPaD (see continuous and dashed black lines in Figures and , respectively; see also Figure ). In this subdomain, Pax6 immunoreactivity is not detected through the whole ventricular surface and characteristically shows a weak labeling (see black arrow in Figures ; see also Figure ), which appears to be common in regions where Pax6 and Dlx co-distribute (like the pallium-subpallium boundary; see below). Moreover, a SPaV (or supraliminal) territory can be defined in shark by the ventricular co-distribution of ScDlx2/5 and ScNkx2.8 (compare Figures with ; see also Figures and ). As discussed above, ScNkx2.8 overlaps the ABB, being expressed in the alar and basal plates. So it abuts Pax6 immunoreactivity dorsally (compare Figures with ; see also Figures and ) and Shh immunoreactivity ventrally in the rostral basal hypothalamus (Figures ; see also Figures and ). Taking these data together, the dorso-ventral compartmentalization in the shark SPa territory seems evident, but slight differences with respect to that described in mammals are noted, especially in the SPaV compartment (compare Figures ). Moreover, a rostro-caudal regionalization based on the differential distribution of presumed tangentially migrated cells is consistent with the model proposal of terminal (TSPa) and peduncular (PSPa) subdomains by the intersection of dorso-ventral and rostro-caudal differences (TSPaD and TSPaV; PSPaD, and PSPaV, respectively; see below).

Tangential Migrations Involving the SPa and Basal Hypothalamus

In the <span class="Species">mouse, tangential migrations involving the <span class="Disease">hypothalamus as source and/or recipient territory have been characterized by means of immunohistochemistry, autoradiography and ISH techniques (Morales-Delgado et al., 2011; Puelles et al., 2012; Morales-Delgado et al., 2014; Croizier et al., 2015; Díaz et al., 2015). The study of some of these intra-hypothalamic migrations led to the identification of three dorso-ventral progenitor subdomains (with their respective rostro-caudal subdivisions) in the Pa of the mouse (Morales-Delgado et al., 2011, 2014; Puelles et al., 2012). Furthermore, tangentially migrated cells derived from the Pa have been also described as contributing to extra-hypothalamic territories such as the amygdala (García-Moreno et al., 2010).

Although further studies using cell-tr<span class="Chemical">acking/tr<span class="Chemical">acing techniques should be needed to confirm it, in the present study we have identified putative pathways of tangential migrations based on the spatio-temporal patterns of different genes widely used to identify tangentially migrating neurons, including cells expressing ScDlx2, ScOtp and cells showing Pax6 immunoreactivity. Although, these cells seem to involve the hypothalamus as source or recipient territories, here, we consider those restricted inside the hypothalamic territory leading us to further characterize the SPa but not the Pa domain. Because the specification process that leads to the expression of these genes is different from those of the recipient territories, we interpret that these cells in fact could have migrated tangentially as also suggested by Morales-Delgado et al. (2011) in mouse.

In the Pa territory, Sc<span class="Gene">Otp expression is continuous with Sc<span class="Gene">Otp-expressing cells that spread into the marginal zone of the SPaD region (see yellow arrowheads in Figures ; see also Figure ) excepting at the rostral-most portion of the TSPa region (see Figure ) a fact compatible with the idea of acroterminal subdivisions (Ferrán et al., 2015). However, by late stage 29 ScOtp-expressing cells are observed in the subventricular zone of the ScDlx2/5-positive SPaD/SPaV domain (data not shown). Taking into account their spatio-temporal distribution, as argued above, we consider that such Otp-expressing cells may have migrated tangentially from the Pa to the SPa territories. Similar Otp-positive cells have been also observed in the Spa region of the mouse (Bardet et al., 2008; Morales-Delgado et al., 2011). Moreover, Otp has been involved in the development of the catecholaminergic phenotype in the zebrafish hypothalamus (Del Giacco et al., 2006, 2008; Blechman et al., 2007; Ryu et al., 2007), suggesting a correlation among ScOtp-expressing cells and TH-ir cells in the GAD-ir SPa area of the shark (see Figure 1A in Ferreiro-Galve et al., 2008; note that the alar hypothalamus is not described in prosomeric terms).

Furthermore, a rostro-caudal regionalization in the <span class="Chemical">SPa domain can be distinguished based on cells showing <span class="Gene">Pax6 immunoreactivity presumed to be tangentially migrated cells. These cells are observed in the mantle zone of Pax6- and ScDlx2/5-positive domains in the PSPaD territory (black arrowheads in Figures ; compare with Figures ; see also Figure ). Similar Pax6-positive cells can be observed at the pallial-subpallial boundary, in the mantle of the dorsal lateral ganglionic eminence (LGE) and other Dlx territories pointing that the presence of Pax6-ir cells in the mantle is common in regions where Pax6 and Dlx co-distribute (Puelles et al., 2000; Flames et al., 2007; Ferreiro-Galve et al., 2008; Medina et al., 2011; Quintana-Urzainqui et al., 2015). Although at stage 28, these cells seem to occupy both the PSPaD and the PSPaV territories (Figure ) they are only observed caudal to the TSPa domain forming a continuous stream with diencephalic marginal Pax6-ir cells (black arrowheads in Figures ; agreeing with previous descriptions in Ferreiro-Galve (2010) although using a different terminology). Notably, similar cells have been found in the mouse addressed to the “posterior entopeduncular area” which parsimoniously fit with the updated hp1 prosomere (see Figures ) (Stoykova et al., 1996; Puelles et al., 2012). From stage 30 onward these cells spread into the hypothalamic basal plate (white arrowheads in Figure ) forming a striking stream that closely abuts the IHB recognized in Santos-Durán et al., 2015). Besides, these cells apparently reach the MM/RM border, and thus, the hp2/hp1 limit (Puelles et al., 2012; Santos-Durán et al., 2015), which again fits with the model. Thus we interpret that these Pax6-ir cells are caudal to the IHB likely accompanying tracts coursing by the rostral border of hp1 (of note, the updated model provided an explanatory framework for this observation). In fact, Pax6 expression has been involved in the correct pathfinding of different tract systems through the prosencephalon by means of both, indirect and local mechanisms, which supports this idea (Mastick et al., 1997; Vitalis et al., 2000; Jones et al., 2002; Prestoz et al., 2012). Taking together these observations, we interpret that TSPa can be differentiated from PSPa based on these cells both in the mouse and the shark (see Figure ; see also Stoykova et al., 1996; Puelles et al., 2012).

Finally, we also have observed a group of Sc<span class="Gene">Dlx2-expressing cells continuous with the subpallium in the marginal zone of the <span class="Chemical">TPa region (arrows Figure ). As discussed above, since Dlx and Otp expression are the result of different developmental processes we interpret that these cells migrate tangentially. Of note, at later stages and in an equivalent position, ScDlx2-expressing cells form a continuous string between the ScDlx2/5 expressing subpallium and the TSPa domain suggesting the existence of a telencephalic-hypothalamic stream (arrowheads in Figure ; see Figure ). Interestingly, Tripodi et al. (2004) and Shimogori et al. (2010) respectively detected COUP-TF-expressing and Foxg1-expressing cells likely emanating from the ventral telencephalon into the anterior hypothalamus. Furthermore, one of the migrating streams described by Tripodi et al. (2004) seems to course from the subpallium to the suprachiasmatic region (equivalent to the SPa) through the marginal anterior hypothalamus (equivalent to the Pa) resembling what we observed in the shark (see their Figure 6B). Thus, the existence of a tangential telencephalic-hypothalamic stream could be conserved among vertebrates similarly to other migrations that have been recently identified in the telencephalon of the shark (Quintana-Urzainqui et al., 2015).

Alternative Interpretations of the Alar Hypothalamus

In a previous work, we defined the HTB based on the complementary expression of Sc<span class="Gene">Otp and ScFoxg1a and the HDB based on the complementary expression of ScOtp and ScTbr1 (see Santos-Durán et al., 2015). Here, we show that additional developmental control genes can also define these borders. The abutted expression of ScOtp in the hypothalamus with ScDlx2/5 in the subpallium and ScTbr1/ScLhx9 in the pallium outlines the HTB (Figures ). The HDB can be also sketched based on the expression of ScOtp in the hypothalamus abutting ScLhx9 or ScLhx5 in the alar p3 (Figures ). These results further support the genoarchitectonic subdivisions proposed by the prosomeric model.

However, an alternative interpretation of these data could yield different conclusions concerning the HTB, the HDB and the genoarchitectonic organization of the alar and rostral-most prosencephalon. Firstly, although the abutted expression of <span class="Gene">Foxg1/<span class="Gene">Otp is assumed to define the HTB, Foxg1 has also been described beyond the telencephalic stalk (Pratt et al., 2004; Affaticati et al., 2015; Santos-Durán et al., 2015) questioning its telencephalon-restricted identity and thus the existence of the HTB at this place. Besides, in Foxg1 null mutant mice the ventral, but not the dorsal, telencephalon is lost (Xuan et al., 1995; Martynoga et al., 2005), which suggests it is required for the development of the subpallium but maybe not for the whole telencephalon.

In this line of thought, the continuous ventricular expression of <span class="Gene">Pax6 and <span class="Gene">Neurog2 through the alar hypothalamus, pallium and rostral diencephalon (Figure ; see also Stoykova et al., 1996; Osório et al., 2010) also questions the expression of Foxg1 defining the HTB and also the HDB. Strikingly, the fact that in the updated prosomeric model neither the Pa nor the HTB reach the HDB at the roof plate (see Figure 8.5B in Puelles et al., 2012), implicitly supports a continuity between hypothalamus, pallium and rostral diencephalon.

Besides, having taken into <span class="Chemical">account data from <span class="Species">mouse lacking neural Shh (Szabó et al., 2009), the continuous expression of Lhx5 through the Pa-rostral diencephalon (PThE) and the continuous lack of Dlx2 through the SPa-rostral diencephalon (excepting PThE) suggest: (i) a continuum between the alar hypothalamus and the alar rostral diencephalon; (ii) a dorso-ventral organization of this territory into Pax6 (dorsal) and Dlx2 (ventral) domains; and (iii) the lack of the HDB.

Together, these evidences suggest an alternative interpretation of the <span class="Disease">alar prosencephalon (up to the zli) organized in a <span class="Gene">Pax6-ir domain (Pa, pallium and PThE) separating two Dlx-expressing domains (subpallium, and SPa besides the remaining rostral-most diencephalon; see Figure ) without a HTB and a HDB. Noteworthy, recently Affaticati et al. (2015) propose the orr (optic recess region) as a morphogenetic entity including Dlx2 subdomains dorsal and ventral to Otp expression in the alar hypothalamus. These Dlx-expresing subdomains likely co-distribute with Pax6 as has been described in the pallium-subpallium boundary (Flames et al., 2007; Ferreiro-Galve et al., 2008) and in the SPaD (present results) pointing to Pax6 as an important factor patterning the alar and rostral-most prosencephalon. Nevertheless, Pax6 downregulation in the Pa territory of amphibians (Moreno et al., 2008) contradicts these observations although it could be an acquired trait since it is likely expressed in other anamniotes (see below). However, our unorthodox proposal should be considered as an open question to be pursued further in future studies.

Evo-Devo Considerations Concerning the Alar Hypothalamus

Our analysis reveals that the data described for the chondrychthyan model, <span class="Species">S. canicula, globally fit with the general assumptions and further details of the prosomeric model together with an important part of the data described in <span class="Species">mouse (see above; see also Puelles et al., 2012; Santos-Durán et al., 2015). These facts support the early acquisition of these traits in development and evolution, which explain the similarities observed across vertebrates at a certain level of analysis, being also the base of the establishment of homologies (Puelles and Medina, 2002; Nieuwenhuys and Puelles, 2016). Differences are also observed but they mainly correspond to terminal phenotype and not the organization (rostro-caudal; ventro-dorsal) of subcompartments, which could explain local differences in proliferation and differentiation patterns across organisms, yielding different morphologies and neuronal subtypes.

To better understand the evolutionary meaning of these observations, here we have comparatively reviewed our findin<span class="Disease">gs with data described in other vertebrates. A similar analysis was made by other authors in the context of precedent prosomeric conceptions (Wullimann et al., 2005; van den Akker et al., 2008; Osório et al., 2010; Moreno and González, 2011; Moreno et al., 2012; Domínguez et al., 2013) and a recent work reviewed data on the anamniote-amniote transition under the updated paradigm (Domínguez et al., 2015). The analysis of our data in shark, based on comparing the presence/absence of expression of certain genes in equivalent topological regions, disagree with some of their interpretations concerning some <span class="Chemical">acquisitions at the anamniote-amniote transition. To shed light on this matter, we have performed a further review of available results in bony fishes and agnathans to better know the anamniote scenario. We are aware of the difficulties to carry out such comparison (mainly because of the scarce number of detailed works specifically addressing the alar hypothalamus, the misleading hypothalamic nomenclature found in the literature, or interpretative differences of the model among authors, to name some).

Vertebrate Pa

The available information about genes expressed in the Pa of vertebrates as <span class="Species">mouse, <span class="Species">chick, Xenopus, zebrafish and lamprey reveals the following scenario: Otp is suggested to be expressed in the Pa territory of all the vertebrates studied so far, Lhx5 seems to be expressed only in tetrapods, and Pax6 and Neurog2 seem to be an amniote acquisition (Frowein et al., 2002; Wullimann et al., 2005; Joly et al., 2007; Osório et al., 2010; Moreno and González, 2011; Moreno et al., 2012; Domínguez et al., 2013, 2015). Our present data in a shark representative of basal gnathostomes support the early acquisition of Otp and also reveal an acquisition of Neurog2, Pax6 and Lhx5 earlier than previously suggested.

Our data clearly support that the expression of <span class="Gene">Lhx5 is present in early gnathostomes besides mammals (Shimogori et al., 2010), birds (Abellán et al., 2010), and amphibians (see “rostral supraoptoparaventricular area” in Domínguez et al., 2013, 2015). Moreover, <span class="Gene">Lhx5 expression has been also described in the alar hypothalamus of other fishes (see “preoptic supraoptoparaventricular area” in Manoli et al., 2014). However, its expression appears to be already present in agnathans since Lhx1/5 has been clearly described in the magnocellular preoptic nucleus of Petromyzon marinus (Osório et al., 2006) and Lampetra fluviatilis (see Figure 5B in Osorio et al., 2005). Taking together these observations, we interpret that, likely, Lhx5-like expression was early acquired in the Pa domain of vertebrates.

<span class="Gene">Pax6 has been described in the <span class="Disease">alar hypothalamus of amniotes (Li et al., 1994; Stoykova and Gruss, 1994; Stoykova et al., 1996; Puelles et al., 2000; Moreno et al., 2012; Robertshaw et al., 2013 and their supplemental material) but not in Xenopus (Bachy et al., 2002; Moreno et al., 2008). The expression of Pax6 in shark (Ferreiro-Galve et al., 2008; present results) supports an early acquisition in the alar hypothalamus of basal gnathostomes. The distribution of Pax6 data in zebrafish suggest that its distribution in the caudal-most preoptic area (Wullimann and Rink, 2001; Wullimann and Mueller, 2004), likely corresponds to the updated Pa domain and, similarly, the distribution of cells showing Pax6 immunoreactivity in the trout Salmo trutta fario (unpublished observations) is compatible with this idea. In lampreys, Pax6 is likely to be expressed in the Pa domain (Figure ) (Murakami et al., 2001; Derobert et al., 2002; Uchida et al., 2003; Osorio et al., 2005; Sugahara et al., 2011; Suzuki et al., 2015). Its expression has not been necessarily described in the alar hypothalamus but can be recognized by the abutted (dorsal and ventral) expression of Dlx1/6 (see Murakami et al., 2001). Moreover, the developing SOT (equivalent to the mfb as suggested by Puelles et al., 2012) divide the mentioned Pax6 domain in rostral and caudal portions (see Figures 2B and 4C2,C3 in Suzuki et al., 2015; for more data concerning the course of this tract in lamprey see Barreiro-Iglesias et al., 2008), as described for the shark (present results). Sugahara et al. (2011) also demonstrated that Pax6 expression in the rostral secondary prosencephalon of Lethenteron japonicum seems to be regulated by the Shh and Fgf pathways. Noteworthy, the inhibition of such pathways performed by Sugahara et al. (2011) yield a Pax6 expression pattern similar to that found in Lampetra japonica by Murakami et al. (2001) (compare Figures 5E and 7E in the first, with 7E in the second). Thus, Pax6 seems to be acquired but differentially expressed in lampreys. Together, all these data point that a Pax6 expressing Pa domain seems to be present in early vertebrates being an early conserved trait of vertebrates.

Members of the neurogenin family have been reported in the forebrain of different vertebrate groups, including lampreys (Wullimann et al., 2005; Guérin et al., 2009; Nieber et al., 2009; Osório et al., 2010; Robertshaw et al., 2013). <span class="Gene">Neurog2 has been described in the pallium, prethalamic eminence and Pa territory of <span class="Species">amniotes (see “supraoptic/paraventricular” and anterior hypothalamus in Medina et al., 2004; Osório et al., 2010; see also Puelles et al., 2012). In chick, there are no detailed data though it has been described in an equivalent region to the Pa domain during a short developmental window (see Robertshaw et al., 2013, and their supplemental material). In Xenopus, Ngnr1 (neurogenin related 1), the closest to mammalian Neurog2 (Nieber et al., 2009; Osório et al., 2010), is not expressed in the Pa domain (see “preoptic area” in Wullimann et al., 2005; Osório et al., 2010), a fact likely related to the absence of Pax6 (Medina, 2008; Moreno et al., 2008, 2009; Domínguez et al., 2013). In zebrafish, only one member of the neurogenin family (Neurog1) has been identified in the pallium and prethalamic eminence but not in the Pa territory (see “preoptic area” in Wullimann and Mueller, 2004; Jeong et al., 2006; Osório et al., 2010). In agnathans, there are no data about Ngn2. However, as other members of the neurogenin family are abundantly expressed through larval stage (Guérin et al., 2009), its expression in the alar hypothalamus must not be disregarded. Together these facts suggest that the presence of neurogenin in the Pa territory could be an early acquisition of gnathostomes.

Vertebrate SPa

The expression of <span class="Gene">Dlx, <span class="Gene">Arx, and Islet genes in the SPa territory of vertebrates seems to be a common and early acquired trait since such expression has been documented in all vertebrates studied so far (Martínez-de-la-Torre et al., 2011; Moreno et al., 2012; Domínguez et al., 2013, 2015; Herget et al., 2014). Variability across vertebrate groups involves the co-expression, or not, with other genes such as Nkx2.2, Shh, and Nkx2.1 in the SPa domain (Moreno and González, 2011; Moreno et al., 2012; Domínguez et al., 2013).

In lampreys, <span class="Gene">Nkx2.1 and <span class="Gene">Nkx2.2 are likely to co-distribute with Dlx in the SPa territory while not with Hh (homologs of Shh) (Myojin et al., 2001; Sugahara et al., 2011; also reviewed in Moreno and González, 2011; Domínguez et al., 2015). In zebrafish Nkx2.1, Shh and Nkx2.2 are also likely to co-distribute without subdividing the SPa domain (Barth and Wilson, 1995; Rohr et al., 2001; van den Akker et al., 2008; Domínguez et al., 2013). However, data on zebrafish Nkx2.1 expression must be taken with care since, recently, the orthologs previously defined as Nkx2.1a and Nkx2.1b have been renamed as Nkx2.4b and Nkx2.1 respectively (Manoli et al., 2014).

Furthermore, two subcompartments have also been described in the <span class="Gene">Dlx-expressing <span class="Disease">alar hypothalamus of different tetrapods based on Nkx2.2, Nkx2.1 and Shh expression (Moreno et al., 2012; Domínguez et al., 2013, 2015). One subdomain seems to co-express Nkx2.1, Nkx2.2 and Shh while the other not. From amphibians to mammals, the domain expressing these genes seems to be almost reduced in favor of the other (Moreno et al., 2012; Domínguez et al., 2015). In amphibians and reptiles these subdomains were defined as rostro-caudal (Domínguez et al., 2015), however, a dorso-ventral interpretation was made in other vertebrates. Taking these into account, it can be concluded that in tetrapods the SPa territory is divided in dorso-ventral subdomains (SPaD and SPaV) as the updated prosomeric model proposes (Puelles et al., 2012; Domínguez et al., 2013, 2015). With all, there seems to be a tendency to the reduction of Nkx2.2 and Nkx2.1 from the SPa compartment in favor of the formation and or reduction of dorso-ventral compartments from fishes to mammals.

Strikingly, the data obtained in the shark a<span class="Gene">pparently contradict this tendency. On one hand, only Sc<span class="Gene">Nkx2.8 expression (likely to match Nkx2.2 expression in mammals) seems to be expressed in the SPa compartment while ScNkx2.1 and ScShh/Shh immunoreactivity seem to be absent, resembling a situation closer to mammals than to other vertebrate groups. In mammals, Nkx2.1 seems to be expressed only in mantle cells of the alar hypothalamus (Shimogori et al., 2010; Puelles et al., 2012) further supporting this idea. This fact can be explained by the size of the shark pallium compared with that of other fishes. The balance among Pax6/Nkx2.1 is believed to regulate the size of alar/basal compartments (van den Akker et al., 2008; Moreno et al., 2012; Puelles et al., 2012; Domínguez et al., 2013). Thus, the expanded expression of Pax6 in the shark could determine a bigger pallium over the basal hypothalamus and, consequently, a ScNkx2.1 restriction. Noteworthy, the basal restriction of Nkx2.1 and the co-distribution of Dlx/Pax6 in a hypothetic SPa compartment could also be an ancestral trait as claimed by Murakami et al. (2001).

On the other hand, our data also support the existence of dorso-ventral subcompartments based on <span class="Gene">Pax6 immunoreactivity or ScNkx2.8 co-distribution with ScDlx2/5 but not with ScShh or ScNkx2.1. This fact raises the possibility that dorso-ventral compartments could also be addressed in fishes.

Moreover, as in shark, a population of <span class="Gene">Pax6-positive cells restricted to the <span class="Gene">PSPa compartment and continuous into the hypothalamus and the diencephalon has also been described in mouse (Stoykova et al., 1996). Populations of Pax6-positive cells have also been described in zebrafish under other prosomeric conceptions (Wullimann and Rink, 2001; Wullimann and Mueller, 2004) that could also support a rostro-caudal regionalization of the SPa territory in bony fishes. Together this data suggest that the presence of these cells is a conserved trait in vertebrates.

Conclusion

Our present data suggest the existence of molecular rostro-caudal and dorso-ventral subdivisions in the <span class="Disease">alar hypothalamus of the shark equivalent to those proposed by the prosomeric model and observed in the <span class="Species">mouse. A detailed comparative review of data among different vertebrates reveals a striking degree of conservation for the genes studied here, although there are minimal differences, which involve the phenotype of the compartments but not their organization. Concerning the shark ABB, it may be defined by the alar or basal expression of ScDlx2/5 or ScNkx2.1, respectively. Besides, ScNkx2.8-expression overlaps over the ABB as markers like Nkx2.2 does in other vertebrates. This revisited comparative analysis also supports that a Pa compartment positive for Otp, Neurog2, Pax6, Lhx5 and a SPa territory positive for Dlx, Nkx2.8/Nkx2.2 were already present before the agnathan-gnathostome transition.

Author Contributions

<span class="Disease">GS-D, IR-M, EC designed the study; SM contributed to data <span class="Chemical">acquisition; GS-D, AM, IQ-U, SF-G performed the experiments; GS-D, SM, IR-M, EC analyzed the data; GS-D wrote the manuscript with inputs from all authors.

Conflict of Interest Statement

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.