Dorsoventral Arrangement of Lateral Hypothalamus Populations in the Mouse Hypothalamus: a Prosomeric Genoarchitectonic Analysis

The lateral hypothalamus (LH) has a heterogeneous cytoarchitectonic organization that has not been elucidated in detail. In this work, we analyzed within the framework of the prosomeric model the differential expression pattern of 59 molecular markers along the ventrodorsal dimension of the medial forebrain bundle in the mouse, considering basal and alar plate subregions of the LH. We found five basal (LH1–LH5) and four alar (LH6–LH9) molecularly distinct sectors of the LH with neuronal cell groups that correlate in topography with previously postulated alar and basal hypothalamic progenitor domains. Most peptidergic populations were restricted to one of these LH sectors though some may have dispersed into a neighboring sector. For instance, histaminergic Hdc-positive neurons were mostly contained within the basal LH3, Nts (neurotensin)- and Tac2 (tachykinin 2)-expressing cells lie strictly within LH4, Hcrt (hypocretin/orexin)-positive and Pmch (pro-melanin-concentrating hormone)-positive neurons appeared within separate LH5 subdivisions, Pnoc (prepronociceptin)-expressing cells were mainly restricted to LH6, and Sst (somatostatin)-positive cells were identified within the LH7 sector. The alar LH9 sector, a component of the Foxg1-positive telencephalo-opto-hypothalamic border region, selectively contained Satb2-expressing cells. Published studies of rodent LH subdivisions have not described the observed pattern. Our genoarchitectonic map should aid in systematic approaches to elucidate LH connectivity and function. Supplementary Information The online version contains supplementary material available at 10.1007/s12035-022-03043-7.


Introduction
The lateral hypothalamus (LH) is a heterogenous region that is conventionally related to regulation of metabolism, feeding behavior, sleep and wakefulness, inflammatory pain regulation, and motivation [1][2][3][4][5][6][7][8][9][10][11]. It was first recognized as a feeding center in the middle of the twentieth century by Hetherington and Ranson, and Anand and Brobeck [12,13] after stimulation and lesion experiments. Later, the LH was also related to reinforcement learning [14]. Crosby and Woodburne [15] first proposed to divide the mammalian hypothalamus into periventricular, medial, and lateral formations; the limit between the cell-dense medial and the cell-sparse lateral hypothalamus interstitial to the medial forebrain bundle was marked roughly by the course of the fornix tract [16]. Puelles et al. [17] later suggested the possibility to exclude a superficial stratum from the LH that includes the cerebral peduncle (lateral forebrain bundle) and associated subthalamic, entopeduncular and subpial cell populations. These authors also proposed a bi-neuromeric schema of hypothalamic structure (peduncular (PHy) versus terminal (THy) hypothalamus as parts of hp1 and hp2 prosomeres, respectively), and concluded that the LH concept only applies to the peduncular hypothalamus. This notion results from noticing that the intrahypothalamic boundary separating PHy from THy curves around the ventromedial nucleus to reach radially the brain surface rostrally to the peduncle (a similar idea was employed de facto by Nieuwenhuys et al. [18]). Puelles et al. [17] also advanced the notion that the LH probably contains diverse alar and basal neuronal populations along its dorsoventral extent. It is to be noted that conventional columnar analysis of the hypothalamus regards the LH as a longitudinal continuum devoid of alar and basal subdivisions (e.g., [15,19]), though more recently Swanson [20] conceives the whole hypothalamus as a part of the forebrain basal plate. The respective length axis of the alternative and inconciliable prosomeric and columnar models are at a right angle.
Such remarkable functional heterogeneity suggests the possibility that the LH may hold a larger variety of cell types than is presently assumed, so that it might be subdivided into a number of distinct sectors [17].
Classical columnar studies assumed that the preoptic area (POA) belongs to the rostral hypothalamus, and thus it was held that the LH extends "rostrocaudally" from POA to the ventral tegmental area (VTA) (nowadays the POA is ascribed to the subpallial telencephalon on the basis of molecular evidence; [17,36,37]). The VTA was classically wrongly ascribed exclusively to the midbrain but is now interpreted as mesodiencephalic [17,38]. The diencephalic tegmentum that is presently contemplated in the prosomeric model was associated in classic columnar descriptions to the now outdated concept of "posterior hypothalamus." As mentioned above, we regard all LH neuronal populations as restricted to the peduncular hypothalamus (PHy), thus named precisely because it is traversed dorsoventrally by the lateral (peduncular) and medial forebrain bundles (pe, mfb; Figs. 1 and 2A; [17,26]). PHy is bounded caudally by the diencephalic prosomere 3 that includes the characteristic alar Dlx-expressing prethalamus and its corresponding basal tegmentum, whereas PHy limits rostrally with the terminal hypothalamus (THy; Fig. 1; [17,26,40]). PHy extends dorsally into the subpallio-pallial telencephalic region through a Foxg1-positive "telencephalo-opto-hypothalamic" transitional region ( [37]; TOH; Fig. 1; see "Discussion").
Along the radial dimension of the whole alar and basal PHy, LH neurons are included in an outer intermediate stratum lying superficial to the inner intermediate stratum formed by the cell-dense "medial hypothalamic" formations and deep to the superficial stratum containing the peduncle, entopeduncular nuclei, and, at basal levels, the related subthalamic nucleus. This relationship of LH with the covering peduncle occurs particularly caudally, next to the hypothalamo-diencephalic boundary. However, the peduncle does not reach rostralwards the intrahypothalamic boundary (PHy/THy) that bends pialwards around the ventromedial nucleus, so that here, the LH seems to reach the brain surface in relationship with the parasubthalamic nucleus and the nigrostriatal tract (Fig. 2D, E; [17]). The alar part of PHy is divided into paraventricular and subparaventricular progenitor areas (PPa; PSPa; the prefix "P" means "peduncular"). Consequently, the alar LH is divided into paraventricular and subparaventricular subdomains; the PPa domain further displays three dorsoventral subdivisions (dorsal, central, ventral; Figs. 1 and 2A; [17]). The basal PHy is divided dorsoventrally into a retrotuberal domain (subdivided into distinct dorsal, intermediate, and ventral subdivisions), and subjacent periretromamillary and retromamillary progenitor domains ( Fig. 1; [17]). The LH thus comprises five differentiable portions at basal levels ( Fig. 2A).
However, such potential alar and basal subregions related to the hypothalamic course of the medial forebrain bundle were not extensively mapped genoarchitectonically [17,41].
In a recent single-cell transcriptomic study of the mouse LH, fifteen glutamatergic and fifteen GABAergic cell clusters were identified [42]. This analysis apparently focused only on the retrotuberal cell populations of the basal hypothalamus, which was the only LH subregion illustrated in three coronal section levels through the dorsomedial hypothalamic nucleus ( [42]; their Fig. 1b). If this interpretation is accurate, these authors did not consider in their analysis various other alar and basal LH cell populations intermingled with the hypothalamic course of the medial forebrain bundle, as predicted by prosomeric analysis ( [17,26]; present data). The same occurs with another transcriptomic analysis by Wang et al. (2021). Other recent hypothalamus transcriptomic studies have referred only superficially to the LH, or not at all [43][44][45][46][47][48].
Our aim in the present work was to analyze systematically the diverse LH regions associated with different dorsoventral Fig. 1 Prosomeric representation of the peduncular hypothalamus (PHy; background in light green) and its relationship with neighboring regions in a schematic sagittal view, following Puelles et al. [17]. The rostral (R) and dorsal (D) spatial directions are indicated. The red dash line indicates the alar/basal limit (a/b). Characteristically, the PHy is traversed dorsoventrally by the medial and lateral forebrain bundles (mfb in blue; lfb or peduncle, pe in pink), as well as by the fornix tract (f in blue). The ascending and descending trajectory of mfb/pe fibers is indicated with arrows. The PHy contacts caudally across the hypothalamic-diencephalic border (HDB) with the diencephalic prosomere p3, representing the prethalamus (PTh). Rostrally, it limits across the intrahypothalamic border (close in front of the fornix tract) with the terminal hypothalamus (IHB; THy). The HDB and IHB boundaries are indicated by black dash lines. The prosomere hp1 is formed jointly by the PHy and the evaginated telencephalon, of which we see subpallial and pallial parts (diagonal area [Dg], pallidum [Pal], striatum [St], pallium [Pall]). The PHy reaches dorsally up to the telencephalic diagonal area domain (Dg). The THy extends into the non-evaginated subpallial preoptic area (POA), represent-ing jointly the prosomere hp2. The PHy is subdivided dorsoventrally in four alar and five basal longitudinal subdomains characterized as molecularly different progenitor areas (Puelles et al. [17]). The alar PHy contains dorsal, central, and ventral peduncular paraventricular (PPa) subregions (PPaD, PPaC, PPaV) as well as an underlying peduncular subparaventricular subdomain (PSPa). The peduncular and terminal parts of PaD have been recently identified as a transitional telencephalo-opto-hypothalamic region (TOH; Morales et al. [37]). The alar PHy subregions contact caudally prethalamic territories. The PPaD (or TOH) borders the prethalamic eminence (PThE), whereas the PPaC contacts the reticular nucleus (Rt), the PPaV limits with the zona incerta (ZI), and the PSPa contacts the rostral liminar area (RLi, [40]). The basal PHy is subdivided dorsoventrally in dorsal, intermediate, and ventral retrotuberal subdomains (RTuD, RTuI, RTuV), complemented more ventrally by the periretromamillary (PRM) and retromamillary (RM) subdomains; RM contacts the floor plate of PHy (F). These basal subdivisions are continuous caudally with the p3 tegmentum (p3tg) levels of the medial-lateral forebrain bundle along the alar and basal parts of PHy, describing characteristic aspects of their molecular profile and checking their number and topography. Each distinct genoarchitectonic level can be tentatively explained as representing cells derived from a distinct progenitor domain. This possibility implied examining the apparent topographic relationships of molecularly distinct adult LH populations with the alar and basal progenitor domains reported within PHy [17,41]. Of course, it cannot be assumed that all neuron types remain within the radial domain where they were born, so that partial dispersion or even massive migrations into other domains are possible. Nevertheless, we hypothesized that each distinct dorsoventral PHy domain may originate one or more specific LH cell subpopulations having a particular molecular profile. These would appear in the mantle either intermixed in a salt and pepper manner or stratified in a temporal sequence pattern. To that end, we examined empirically identified selective lateral hypothalamic cell markers in images downloaded from in situ hybridization data of the Allen Brain platform (© 2008 Allen Mouse Brain Atlas; Allen Developing Mouse Brain Atlas). According to our analysis, nine molecularly distinct DV tiers of the LH can be distinguished within PHy, four of them in the alar part and five within the basal part.

Materials and Methods
Most of our analysis was based on in situ hybridization (ISH) images downloaded from the Allen Developing Mouse Brain Atlas (https:// devel oping mouse. brain-map. org) and Allen Mouse Brain Atlas (https:// mouse. brain-map. org/). Our analysis was focused on late embryonic stages (E18.5) and postnatal stages from P4 to P56 in sagittal and coronal sections when available. According to documentation given by the Allen platform, male specimens of the C57BL/6 J mouse strain were used for the E15.5 through adult stages.
To select markers, we applied the AGEA informatic tool in the Allen Mouse Brain Atlas which is based on spatial correlations of gene expression data. We had no preconceived notion about how many molecularly different parts of the lateral hypothalamus existed. We just sampled unbiasedly genes detected by AGEA in 2-3 dorsoventral points at each of the areas identified in the Allen source as either "basal peduncular hypothalamus" or "alar peduncular hypothalamus." We only assumed that the LH is restricted to the peduncular hypothalamus and that it extends through both the alar and basal regions (see concept in Puelles et al. 2012). The AGEA tool is not resolutive enough to place the search seed in a more precise way. Each search point provided a list of some 200 candidate genes. We screened each of these markers, selecting empirically those showing in sagittal sections clearcut differential expression along ventrodorsal levels of the lateral hypothalamus (LH). Two of the authors (CD, LP) analyzed and discussed the expression pattern of each gene to arrive at a consensus decision about the interest of the pattern for the present purpose. Genes with low ISH signal, widespread signal, or high background were discarded. In a later stage of the study, a few other genes of interest reported in published transcriptomic studies and present in the Allen database were added to our list (we checked all genes mentioned in transcriptomic studies).
We identified 59 genes, classified according to protein type and related functions (Table 1). They were differentially expressed along various ventrodorsal LH levels ( Table 2). The list of the selected genes illustrated in Figs. 2, 3, 4, 5, 6, 7, 8 and 9, together with the references of their respective experiments and images downloaded from the Allen database, are indicated in Supplementary Table 1.
Moreover, we include some mouse brains treated by ISH followed by immunohistochemistry (IHC) from our histological collection. A series of transverse sections through the peduncular hypothalamus were treated by ISH for Nkx2.2 and immunoreaction against tyrosine hydroxylase (TH) antibody at P4 postnatal stage. Other series of horizontal sections through the adult hypothalamus (conventional coronal sections) were hybridized for vGlut2 (Slc17a6; potentially glutamatergic cells) and for Gad67 (Gad1; potentially GABAergic cells) and further immunoreacted with a TH antibody.

Animals
All experimental procedures, experimental protocols, handling use, and care of mice were conducted in compliance with the current normative standards of the European Community (

Preparation of Tissue
Mice were anesthetized with 2% chloral hydrate/PBS, and perfused intracardially with cold 4% paraformaldehyde/PBS. Following dissection, brains were postfixed in the same solution for 3 h to overnight. All tissues were either sectioned directly or stored at -20 °C in a solution of 30% vol/vol ethylene glycol, 30% vol/vol glycerol, 0.4 × PBS. Brains stored in this way were rinsed overnight in PBS prior to sectioning.

Sectioning of Tissue
For sectioning, the embedded tissue was oriented to obtain either sagittal, horizontal, or transverse sections across the hypothalamus, and cut into series of 120-µm-thick vibratome sections obtained from 6% agarose blocks. Sections were processed for ISH (Nkx2.2, vGlut2, Gad67) followed by IHC detection of tyrosine hydroxylase (TH).

Immunohistochemistry
The immunohistochemical reaction protocol was described in detail by Ferran, et al. [69]. A tyrosine hydroxylase antibody was used (TH; Novus, Cat# NB300-109 RRID: AB 10,077,691, dilution 1:1000). The primary antibody was incubated overnight at 4 °C. After washes, the sections were incubated with biotinylated goat anti-rabbit antibody (Vector Laboratories, CA; used at 1:200 dilution) followed by a streptavidin-peroxidase complex (Vectastain-ABC kit; Vector Laboratories; 0.001% dilution), applied for 1 h at room temperature. Peroxidase activity was developed with 0.03% 3,30-diaminobenzidine (Sigma; St Louis, MO), plus 0.003% hydrogen peroxidase. After immunohistochemical and hybridization labeling, the slides were washed several times in PBS, air-dried, and cover slipped with Cytoseal 60 (Thermo Scientific, Ref.  or Mowiol (Calbiochem, Bad Soden, Germany, Ref. 475,904). We verified the specificity of the antibodies by performing parallel control experiments that omitted the primary antibody, checking that no residual immunostaining was detected (data not shown). Immunoreaction shown in Fig. 8A was obtained with a pandistalless (DLX) polyclonal antibody whose specifications are described by Puelles et al. [40]; this figure corresponds to their Fig. 8.15E.
The PCR conditions used were an initial denaturation step at 94 °C for 5 min, then 35 cycles (30 s at 94 °C, plus 1 min at Tm temperature (58 °C), and 1 min at 72 °C), followed by 20 min at 72 °C. The PCR products were cloned into the pGEM-T Easy Vector (Promega, Cat. A1360), and sequenced (SAI, University of Murcia).

In Situ Hybridization
Brain sections were processed for in situ hybridization with digoxigenin-UTP-labeled antisense riboprobes. Sense and antisense digoxigenin-labeled riboprobes for mouse Gad67, Nkx2.2, and vGlut2 were synthesized following the manufacture's recommendations (Roche Diagnostics S.L., Applied Science, Barcelona, Spain), and applying specific polymerases (Fermentas, Madrid, Spain In situ hybridization (ISH) was performed as described by Ferran et al. [69]. After hybridization, all sections were washed and incubated in a solution containing alkaline phosphatase coupled anti-digoxigenin antibody (diluted 1:3.500; Roche Diagnostics). Nitro blue tetrazolium/5-bromo-4chloro-3-indolyl phosphate (NBT/BCIP; Roche) solution was then used as chromogenic substrate for the final alkaline phosphatase reaction (Boehringer, Mannheim, Germany). No specific signal was obtained with sense probes (data not shown).

Results
Our description of the diverse LH cell populations found associated with the medial forebrain bundle is based on material fixed at an advanced embryonic stage (E18.5) or at postnatal to adult stages (P4-P56). Data on earlier embryonic stages were included exceptionally to clarify the potential sources of given cell populations. The 59 genes selected from the Allen Developing or Adult Mouse Brain Atlases are listed in Table 1, where protein type and associated functions are specified. Table 2 shows the markers grouped by shared combinatorial properties among the ventrodorsal histogenetic subdivisions defined along the length of the medial forebrain bundle.
The conventional LH, defined as lying interstitial to the medial forebrain bundle, is in fact restricted to the peduncular hypothalamus (PHy), as was previously suggested implicitly by Nieuwenhuys et al. [18], without referring to prosomeres. As shown in Fig. 1, in the prosomeric model the PHy is the hypothalamic portion of the hypothalamotelencephalic prosomere hp1, constituting the caudal found between the superficial (purple) and medial (yellow) hypothalamic strata. Note that while the classical LH in principle reaches the brain surface, Puelles et al. [17] postulated a separate superficial stratum including diverse formations not participating in the LH features: we follow this analysis (see the purple elements). The fornix (f) and nigrostriatal (ns) tracts are identified in gray; other visible tracts except the mfb are also highlighted in gray. The superficial stratum is associated with the peduncle (pe; in gray). Various ventrodorsal LH subdivisions to be documented in the following plates are identified (basal LH1-LH5 and alar LH6-LH9). The main prethalamic and terminal hypothalamic neighboring regions are illustrated as reference landmarks. For additional abbreviations, see "Abbreviations" ◂ Transcription factor (Zic family) Role in neural crest development and neural tube closure [37,68] anteroposterior part of the hypothalamus, lying next to its caudal neighbor, the prethalamic diencephalon. The PHy is notoriously traversed (dorsoventrally, according to prosomeric analysis) by numerous ascending and descending nonmyelinated or poorly myelinated fiber packets that interconnect the telencephalon and hypothalamus, as well as other diencephalic, mesencephalic and brainstem regions [18]. This abundant set of heterogeneous components constitutes the medial forebrain bundle (mfb), which courses deep to the cerebral peduncle, or lateral forebrain bundle (pe/lfb; Fig. 1) and finally largely enters the basis and the medial wall of the hemisphere. The PHy, and therefore the LH, is bounded caudally by the prethalamus and rostrally by the terminal hypothalamus (THy); the latter represents the hypothalamic portion of the hypothalamo-telencephalic prosomere hp2 (Figs. 1 and 2A; [17,41]). The PHy/THy intrahypothalamic border (IHB) courses in front of the hypothalamic trajectory of the fornix tract (f), a myelinated tract that courses dorsoventrally (transversally) through the rostral part of PHy (Fig. 1). The fornix tract originates in the hippocampal commissural area, crosses the septum, and enters the PHy behind the anterior commissure (in front of the interventricular foramen), to course dorsoventrally through its alar and basal regions, reaching finally the retromamillary commissure in the PHy floor, after giving collaterals or terminals to the mamillary region, a basal component of THy [26]. Dorsally the PHy (including the LH) contacts the substantia innominata of the evaginated telencephalon, apparently through the recently identified transitional telencephalo-opto-hypothalamic longitudinal domain (TOH; Fig. 1; [37,70]). Figure 2 shows the radial organization of the PHy in a series of topological horizontal sections parallel to the prosomeric forebrain axis (note these are conventionally described as "coronal sections" under the columnar paradigm; see also the corresponding radial glial structure in Puelles et al. [17]; their Fig. 8.12). Following the standard developmental mantle radial stratification schema of Senn [71], the PHy appears subdivided radially into periventricular, intermediate, and superficial strata ( Fig. 2B-J); the intermediate stratum is habitually divided into cell-dense medial hypothalamus (MH) and cell-sparse LH [15,17] (the MH/LH limit roughly coincides with the fornix tract). The pial surface of PHy is oriented ventrolaterally (due to bending of its radial glial processes around the VM nucleus) and is larger than the ventricular surface. This radial bending of PHy around the VM is the reason why the classic LH is actually restricted to the PHy. It should be noticed that the cerebral peduncle (jointly with the underlying subthalamic nucleus at retrotuberal levels) only covers in horizontal sections the caudal two-thirds or so of the LH; the rostral third of LH (which contains the fornix and the nigrostriatal tract) extends into the brain surface rostral to the peduncle, in the superficial area where the parasubthalamic nucleus is found (purple area in Fig. 2B-J). The subpial optic tract crosses orthogonally the subparaventricular alar domain of PHy, close to the alar-basal boundary [72], so that its topologically ventral rim serves as an approximate landmark of the alar-basal transition, relevantly for the topologic analysis of LH components. Our description of LH will be thus restricted to the outer intermediate stratum of PHy, which contains by definition the LH populations of interest, adding the rostral portion where the bent LH seems to extend into the superficial stratum. The standard LH populations that lie interstitial to the mfb at various horizontal section levels are highlighted in blue in Fig. 2B-J, including various alar and basal perifornical cell groups (labeled Pf), whereas the associated medial and superficial populations are highlighted in yellow and purple, respectively.
For simplicity, each LH population was named with the acronym LH and numbered consecutively in ventrodorsal order ( Fig. 2A). According to our analysis, the LH extends through five basal regions (LH1-LH5, corresponding to the progenitor domains RM, PRM, RTuV, RTuI, RTuD; [17]) and four alar regions (LH6-LH9 corresponding to the progenitor regions PSPa, PPaV, PPaC, PPaD; loc. cit.; Table 2 Adult expression of 59 gene markers across the ventrodorsal subdivisions of the lateral hypothalamus (LH) and their correlation with basal and alar subdivisions of the peduncular hypothalamus (PHy). Color-coded entries represent large subpopulations of positive cells for the indicated markers at the specified LH sectors. The gray boxes represent less numerous positive perifornical subpopulations at the indicated loci. Sparse positive cells, such as Otp-positive cells within adult sectors LH7-LH9, are not shown. Some markers are selectively expressed at the indicated embryonic stages rather than in the adult. Correlation of LH sectors to ventrodorsal progenitor areas of PHy is marked at the top.

Wif1
PPaD is equivalent to the TOH of Morales et al. [37]. Multiple molecularly different neuronal populations can be thus distinguished and classified in an embryologically relevant DV pattern [73,74].

Basal Lateral Hypothalamic Populations
Selected markers expressed in the basal LH subregions were downloaded from ISH data at the Allen Developing Mouse Brain and Allen Mouse Brain atlases, which are illustrated in Figures 3, 4, 5, 6 and 7 and summarized in Table 2. Our description follows a ventrodorsal order, that is, from LH subregion 1 (LH1) upwards to LH subregion 5 (LH5) (see Fig. 2A, B-F).

LH2: the Periretromamillary Subdomain
LH2 is the LH subregion associated with the periretromamillary (PRM) basal subdomain. This subdomain was first defined molecularly by Puelles et al. [17] as associated with selective Otp and Sim1 basal plate signals, unrelated to alar expression domains of the same markers (see also [82][83][84][85]. At early stages, Irs4 expression (encoding insulin receptor substrate 4) accompanies Otp and Sim1 transcripts in the characterization of the PRM (Fig. 4A-C); Otp-expressing cells continue to be detected in LH2 at least until P28 (see Allen Developing Mouse Brain Atlas). Classically the PRM was wrongly lumped with the neighboring diencephalic tegmentum (largely devoid of Otp expression) under the now outdated concept of "posterior hypothalamus" (e.g., [86]; his Figs. 13-18, reproduced in Puelles et al. [17]; their Fig. 8.3). The PRM appears divided into the main PRM mass (representing the medial hypothalamic stratum), and the LH2 domain interstitial to the fornix and nigrostriatal tracts among other medial forebrain bundle components (f, ns; Fig. 4; right upper part). The LH2 ends superficially rostrally to the subthalamic nucleus and the peduncle, at levels coincident with the parasubthalamic nucleus under the lateral tuberal nucleus (LH2 in blue; STh, PSTh, and TuL in purple; pe (in gray); see schema inserted at the upper right corner of Fig. 4; note also the curved radial boundary between THy and PHy).

LH3: the Ventral Retrotuberal Subdomain
Puelles et al. [17] identified the basal PHy territory stretching from the PRM to the alar-basal boundary as a retrotuberal domain (RTu), referring to the larger THy tuberal (Tu) domain counterpart (Fig. 1). Both Tu and RTu were subdivided dorsoventrally into parallel ventral, intermediate, and dorsal subdomains (RTuV, RTuI, RTuD in the case of RTu; [17]). These three subdomains contain respectively our LH3-LH5 basal components of the LH. LH3 represents the ventral retrotuberal element. RTuV is a thin subdomain limiting dorsally the PRM.
Jointly with its TuV counterpart, it was distinguished by Puelles et al. [17] as the apparent source of histaminergic Hdc-expressing neurons, which later spread partially into neighboring tuberomamillary, retromamillary, and mamillary areas (Fig. 5A-D; Table 2; Allen Developing Mouse Brain Atlas). Hdc cells are first scattered from periventricular to subpial surfaces within the ventral retrotuberal PHy (including our LH3 at intermediate radial levels) as well as in the correlative TuV parts of THy (Fig. 5C, D). Contrarily, the intermediate and dorsal components of the RTu/Tu domains, as well as the PRM, largely lack Hdc mRNA expression. However, migrated Hdc-positive cells later appear in the terminal perimamillary area rostral to PRM, and subpially in the RM/M domains (Fig. 5A, B). Both RTuV and TuV domains otherwise produce mainly GABAergic cells, probably due to local expression of the Dlx genes, which contrasts with the underlying PRM/PM domain that displays a glutamatergic profile [17].
According to previous descriptions, histaminergic cells express the genes Prph (encoding peripherin) and Wif1 (encoding WNT inhibitory factor 1) [43,87,88], two markers absent in our AGEA analysis of ISH Allen data, though they do appear in that database. We confirmed that Prph and Wif1 are expressed in LH2 cells with a similar distribution as Hdc cells (Fig. 5E, F), and they were added to Table 2.
Nts and Tac2 encode the neuropeptides neurotensin and tachykinin 1, respectively. Nts-positive cells mainly localize in the central part of LH4, contrasting with Tac2-expressing cells, which appear in more medial and lateral locations (compare Fig. 6A-C). However, a subgroup of Nts-positive cells overlaps or intermixes with Tac2-positive cells next to the hypothalamo-diencephalic border (red asterisks; Fig. 6B, C). Neuropeptidergic Cartpt-expressing cells are largely restricted to LH4, though a few positive cells are found in the suprajacent LH5 (Fig. 6E), as occurs also with Peg10 cells (not shown). Cartpt cells are mainly located in the central part of the LH4 (Fig. 6F) whereas Peg10 cells are more numerous and distribute widely throughout LH4 (Fig. 6G). Ecel1-, Nek7-, and Pdyn-expressing cells are present in LH4 (Fig. 6H-J) but are found also in the suprajacent LH5 subregion (Pdyn in Fig. 7K, L; Table 2). The Pdyn gene encodes the precursor prodynorphin neuropeptide. Penkexpressing cells represent other significant neuropeptidergic subpopulations of LH4 (Fig. 6K). As noted previously, in the adult Penk cells appear at different basal and alar LH sectors (see Table 2), though at E18.5 they are restricted to LH2 (Fig. 4D). The cerebellin family member Cbln2 is widely expressed in LH4 (Fig. 6L).

LH5: the Dorsal Retrotuberal Subdomain
This is the dorsalmost component of the basal lateral hypothalamus, located at the intermediate stratum of the dorsal retrotuberal hypothalamic subdomain (RTuD). The local medial hypothalamic stratum contains the posterobasal nucleus (PBas) and the A13 dopaminergic cell group [17,89]. Horizontal sections through LH5 no longer intersect the peduncular dorsomedial hypothalamic nucleus (DM-P) or the subthalamic nucleus (STh), while the alar ventral entopeduncular nucleus (EPV) may start to be visible within the peduncle (Fig. 7C). LH5 characteristically contains large-size Hcrt-and Pmch-positive cells produced at this level (Fig. 7A, B, G-I). Hcrt encodes the neuropeptide hypocretin HCRT (also named orexin). Some Hcrt cells apparently disperse into both peduncular and terminal parts of the SPa area (PSPa and black arrows in TSPa; Fig. 7A). Crh-, Gpx3-, Igf1, Irs4-, Nnat-, Rfx4, S100a10-, and Trhrexpressing neurons are mainly restricted to LH5 (Table 2; Fig. 7D-F); other molecular markers of LH5 that spread variously out of it include Dlk1, Gal, Pdyn, Penk, Plagl1, and Trh (Table 2). At perinatal stages, an important subpopulation of RTuD-originated cells express Pmch, which encodes pro-melanin-concentrating hormone. These largesized Pmch elements intermix with Hcrt cells and populate the PBas nucleus (medial hypothalamus) and the LH5, and partly disperse into the subjacent LH4 and even the neighboring prethalamic zona incerta at later stages (  Table 2) or Cartpt cells (Fig. 6E). Other transcripts expressed in LH5 cells encode Fig. 3 In situ hybridization expression mapping at P56 of various markers in sector 1 of the basal lateral hypothalamus (LH1; also known as "lateral retromamillary nucleus," RML). Chosen sagittal and coronal section images (identified according to characteristic landmarks consistent with Fig. 2 atlas data) were downloaded from the Allen Developing Mouse Brain Atlas. The sagittal sections indicate the restricted dorsoventral topography of the labeled LH cells, whereas the coronal sections demonstrate the corresponding either diffuse or restricted topography within the illustrated LH level as defined in Fig. 2. Some markers are shown only in coronal sections (with corresponding level landmarks). Sagittal (A, D, G) and horizontal (conventional coronal) (B, E, F, H-L) sections are shown (marked as SAG or COR at the upper left corner), while (C) illustrates a schematic coronal section through the hypothalamus taken across LH1 (in blue), extracted from Fig. 2; corresponding medial and superficial retromamillary strata relating to LH1 (yellow and purple, respectively) are also shown. In sagittal sections, dorsal direction points to the left, and rostral direction to the bottom; in horizontal/coronal sections, medial direction points to the left, and rostral direction to the bottom. LH1 is characterized by cells expressing H), and Nnat (I). The Cbln4 (J) and Erbb4 (K, L) markers also appear expressed in LH1 cells, though they are detected likewise in cells belonging to the overlying LH2 sector (Table 2). LH1 is part of the retromamillary basal hypothalamic domain (Fig. 1). Black dash lines indicate the thalamoprethalamic, hypothalamo-prethalamic, and intrahypothalamic transversal interprosomeric boundaries. A red dash line denotes the alarbasal limit. The LH1 area is surrounded with a black line, similarly as neighboring structures of the same radial domain (medial and superficial retromamillary nuclei, RMM, RMS). The mamillo-tegmental tract (mtg) separates the LH1 and RMM areas. The fornix (f) courses dorsoventrally (transected) next to the intrahypothalamic boundary; the LH subpopulation identified as "perifornical 1" (Pf1) is surrounded by a blue dash line (due to the dorsoventral course of f, perifornical populations exist at all LH subdivisions, thus needing an alphanumeric identification). Scale bar for all images, indicated in (A): 730 µm. For additional abbreviations, see "Abbreviations"   Table 2).

Alar Lateral Hypothalamic Populations
The following analysis of alar LH subpopulations ( Fig. 1  and 2G-J) is also based mainly on ISH data from the Allen Mouse Brain Atlas. Table 2 summarizes the cell phenotypes found in each alar LH subregion. The expression of selected markers in alar LH subregions is illustrated in Figs. 8, 9, and 10.

LH6: the Subparaventricular Subdomain
The LH6 subpopulation belongs to the peduncular alar SPa domain (PSPa), which is continuous caudalwards with the prethalamic rostral liminar band (RLi; Fig. 8A; Puelles et al., 2021). Both PSPa (including LH6 at the appropriate radial level) and its rostral counterpart TSPa contain GABAergic Dlx5/6-and Arx-expressing cells, as shown here in a sagittal section from a postnatal Dlx5/6-LacZ transgenic mouse whose brain was secondarily immunoreacted with a Pan-Distalless antibody (Fig. 8A). In the LH, Dlx5/6-expressing cells appear in the alar LH6, as well as in the underlying basal LH5 and LH4 subdivisions (possible tangential migration). LH6 contains a mixture of blue (Dlx5/6-LacZ β-galactosidase reaction product) and brown (DLX protein) labeled cells, as occurs partially at the neighboring basal LH5, whereas LH4 contains mainly brown immunoreacted cells. According to the analysis of such signals performed by Puelles et al. [40], the blue reaction implies persistent LacZ reaction possibly maintained by low levels of Dlx5/6 enhancer activity and/or perdurance of the β-galactosidase enzyme, with low or undetectable DLX protein levels; the brown reaction represents ongoing DLX1/2/5/6 expression, and no postnatal Dlx5/6-LacZ enhancer activity. Gad1 (Gad67), Gad2 (Gad65), and Slc32a1, three markers of GABAergic neurons, are also expressed in LH6 cells (Table 2). At E18.5, LH6 selectively contains Pnoc-positive cells, mostly located superficially, at the place where the PHy bends pialwards around the THy (Fig. 8B); however, some Pnoc-positive cells appear dispersed within other LH subregions at postnatal mouse stages (not shown).

LH7: the Ventral Paraventricular Subdomain
Puelles et al. [17] proposed a dorsoventral division of PPa in dorsal, central, and ventral subareas (PPaD, PPaC, PPaV) with subtly variant molecular identities (see also [89]). We found that the whole hypothalamic PPa subdomain is characterized as a Dlx-negative (Fig. 8B) and Otp/Sim1-positive region (Fig. 9B) lying above the SPa band. Consistently with the postulated PPa subareas, we identified three alar LH subregions (LH7, LH8, LH9; Figs. 1 and 2A, H-J). Some markers appear restricted to cell subpopulations located selectively at one of these dorsoventral levels.
LH7 cells are located at the ventral part of the PPa subdomain (PPaV), coinciding with the ventral and lateral paraventricular nuclei (PaV; PaL) and related superficially with the dorsal entopeduncular nucleus (EPD; interstitial to the peduncle) (see Fig. 8E). This LH part contacts caudally the prethalamic incertal complex (ZI; Fig. 8C-E). LH7 includes a subpopulation that expresses the Sst gene, which encodes the neuropeptide somatostatin. Sst-expressing cells are restricted to LH7 (Fig. 7C) whereas other markers such as Gal, Mdga1, Pdyn, Sim1, Slc17a6, and Trh are shared with neighboring LH subpopulations (see Table 2); Gal-, Pdynand Trh-positive cells are found in perifornical position, of which Gal-, and Trh-positive cells appear only at LH7 level ( Table 2; see Allen Developing Mouse Brain Atlas). The whole PPa subdomain and its terminal counterpart TPa (e.g., anterior hypothalamic area, AH), have abundant Slc17a6/ vGlut2-positive glutamatergic cells (Fig. 8D).

LH8: the Central Paraventricular Subdomain
The LH8 subdomain corresponds to the central PPa hypothalamic area (PPaC) and is a level with the (medial) PaC nucleus and the upper part of the (superficial) EPD nucleus (Fig. 8I). LH8 contacts caudally the prethalamic reticular nucleus (Rt; Fig. 8I). Meis2-positive cells appear as a subpopulation of LH8. We noted that at prenatal stages Meis2 is expressed through all radial levels of the PPa subdomain but is mainly restricted to LH8 in the adult (Fig. 8F, G; Table 2). Meis2 encodes a homeobox transcription factor (myeloid ecotropic insertion site 2 transcription factor). The whole PPaC area essentially produces Slc17a6 (vGlut2)-positive glutamatergic cells with diverse peptides (Fig. 8H; see [17,90]). LH8 also contains a large subpopulation of cells expressing Lhx5 and relatively less numerous Tbr1 and Cac-na2d1 subpopulations (Fig. 8J, K, L; Table 2). The Tbr1 subpopulation may be intrinsic or migrated early on from the prethalamic eminence, as was suggested in chick studies [91,92]. Lhx5 and Tbr1 encode transcription factors whereas Cacna2d1 encodes a voltage-dependent calcium channel. Erbb4-expressing cells are mainly restricted to the LH8 sector whereas perifornical Pdyn-positive cells are also found in LH7 and LH9 (Table 2).

LH9: the Dorsal Paraventricular Subdomain
LH9 is the dorsal LH component within the alar PPa domain; it is noteworthy that at these section levels the fornix tract abandons its standard rostral position close to the intrahypothalamic boundary (THy/PHy) and moves progressively caudalwards within PHy until it reaches the backside of the anterior commissure and passes into the septum. As happens generally in other parts of Pa, the PPa contains glutamatergic cells expressing the Slc17a6 (vGlut2) gene (Fig. 8D, H; Table 2), supposedly jointly with some peptidergic markers, and its cells express Otp and/or Sim1 transcription factors (Fig. 9B, D, G). Recently this longitudinal band at the hypothalamotelencephalic border has been identified with a hypothalamo-amygdalar corridor (HyA) vehiculating dorsal and central paraventricular cells that migrate into the medial amygdala (Otp cells) and pallial amygdala (Sim1 cells) [93]. In addition, Morales et al. [37] postulated this corridor as a novel Foxg1-positive telencephalo-optohypothalamic (TOH) histogenetic area, with abundant glutamatergic neurons; the TOH is interpreted by us as a molecularly differential dorsal part of the hypothalamic paraventricular area rather than as 'telencephalic' (see "Discussion").
We found that LH9 contains cells expressing seven transcription factors (Fezf2, Foxg1, Otp, Satb2, Sim1, Zic5; Table 2), and other markers (Pdyn, Nos1, Tac1; Table 2). In Fig. 9, we analyze the correlation between the hypothalamic Pa area defined by Sim1-and Otp-expressing cells, the dorsal Nkx2.1-negative TOH subdomain of Pa that co-expresses Foxg1, and the overlying diagonal/POA subpallial domains, characterized by Nkx2.1-and Dlx5/6-expressing cells (see also Fig. 8A). We found that the transcription factor Satb2 is expressed specifically in a distinct neuronal subpopulation within the TOH area, which participates also of the Foxg1 signal, and lacks Nkx2.1 signal (Fig. 9A-H). As observed in sagittal and coronal sections at E18.5, Satb2 signal appears both in a dense periventricular island and in an ampler population dispersed widely within the LH9 traversed by the medial forebrain bundle (Fig. 9H, I). The dense periventricular island may correspond to a TOH subpopulation identified as the posterior BSTM (BSTMp) by Morales et al. [37].
Fezf2, Nos1, and Tac1 are also selective markers distinguishing LH9 cells from the neighboring LH populations at postnatal stages (Table 2). Fezf2-expressing cells localize next to the intrahypothalamic boundary (Fig. 10A) whereas Nos1-expressing cells appear in a dorsocaudal perifornical LH9 location, close to the anterior commissure (Fig. 10B; compare the schematic horizontal section through LH9 level in Fig. 10C). Curiously, the same perifornical LH9 subpopulation lacks Fezf2-positive cells. Fezf 2 encodes a Fez-like zinc finger family member and Nos1 encodes the enzyme nitric oxide synthase 1. The transcription factor Zic5 (Fig. 10D,  E) is restricted also to the LH9 sector.

Nkx2.2-Positive Band Derivatives in the Lateral Hypothalamus
The Nkx2.2 gene is expressed early on in a longitudinal liminar band along the forebrain alar/basal limit [17,26,94,95] ("liminar" comes from the Latin term "limen," rim or border). Note this pattern is restricted to the extended forebrain tagma (including the midbrain; [96]). This gene appears expressed instead at the floor/ basal border of the hindbrain and spinal cord, after an isthmic discontinuity. We examined recently the Nkx2.2 signal at the interthalamic zona limitans shell domains related to the interprosomeric p2/p3 border, as a continuation of which the caudal and rostral longitudinal "liminar" parts of the original band can be followed (CLi, RLi; [40,97]). These features are also recognizable at pre-or postnatal stages in other vertebrate species studied (Fig. 11A, B; [98][99][100][101]). Nkx2.2-expressing neurons derived from RLi contribute differentially to THy and PHy, including some alar and basal LH subregions ( Fig. 11; [17]). The rostral liminar band (RLi) overlaps the hypothalamic alar/basal limit, thus being composed in theory of alar and basal parts (if the boundary is linear, adimensional). At early stages, different  Among alar LH territories, the subparaventricular LH6 subdomain is largely Nkx2.2-negative and contains some TH cells (Fig. 11C-E). In contrast, the overlying LH7 (ventral paraventricular) subdomain contains Nkx2.2-expressing cells (Fig. 11E-G), possibly associated with the RLi elements migrated dorsally that populate the PaV and PaD subnuclei (Fig. 11F-H). In the basal region of PHy, Nkx2.2-expressing cells were found in the LH5 and LH4 subdomains corresponding to the RTuD and RTuI areas (Fig. 11F-H); these positive cells seem to have migrated ventralwards from the Nkx2.2-expressing liminar RLi band; LH5 Nkx2.2-positive cells appear lateral to the TH-immunolabeled dopaminergic A13 population characteristically present at this dorsoventral level of medial basal PHy (Fig. 11F, G).

GABAergic and Glutamatergic Populations in the Lateral Hypothalamus
We analyzed adult lateral hypothalamic GABAergic and glutamatergic populations labeled by vGlut2 (Slc17a6) and Gad67 (Gad1) ISH, respectively, checking their correspondence with the postulated LH sectors. Figures 12-15 show a series of adjacent sections ordered from ventral to dorsal; this material was secondarily immunoreacted for TH. Hypothalamic GABAergic cells are usually related to Dlx-expressing progenitor domains whereas glutamatergic cells correlate with Dlx-negative domains, as is observable in Dlx5/6-LacZ brains ( Fig. 8A; see also [102]).
GABAergic and glutamatergic cell populations present in adult radial domains in the hypothalamus generally are produced separately (often distinct neighboring domains) though they may intermix partially via tangential migration. We assume that the major cell population is locally originated whereas the minor cell population is migrated [17].
In the basal hypothalamus, LH1-and LH2-cells associated respectively with the RM and PRM subdomains express massively Slc17a6 (vGlut2) transcripts and show few Gad1 (Gad67)-positive GABAergic cells (Figs. 12A-D and 13A,  B); the STh and VPM nuclei migrated out of the retromamillary area also are predominantly Slc17a6 (vGlut2) positive ( Fig. 13A-D). LH3 cells are largely histaminergic (expressing Hdc and other selective markers, as shown in Fig. 5). In contrast, the basal LH4 population found lateral to the DM-P nucleus contains both GABAergic cells, expressing Gad1 (Gad67; Fig. 13C, E, F) and glutamatergic cells, expressing Slc17a6 (Fig. 13D). The same pattern characterizes the DM nucleus, where glutamatergic cells predominate in its dense core portion and GABA cells mainly populate the surrounding shell area; the glutamatergic cells are thought to be migrated (see comments and additional data in Puelles et al. [17]. Finally, most of the dorsal retrotuberal LH5 cells are glutamatergic with only sparse GABAergic cells intermixed (not shown; see [17]).
We summarize in Fig. 16 our schema of ventrodorsal subdivisions of the LH in five basal (LH1-LH5) and four alar (LH6-LH9) cell subregions. All these molecularly distinct subregions are interstitial to the course of the mfb and typically contain at most levels perifornical cell subpopulations associated with the peduncular course of the fornix tract, close to and parallel to the intrahypothalamic boundary.

Discussion
Prosomeric analysis of 59 markers selected from the ISH database of the Allen Mouse Brain Atlas [103] and the Allen Developing Mouse Brain Atlas [96] leads us to propose a ventrodorsal regionalization of the lateral hypothalamus (LH) according to molecular criteria. This  [17,74,84,90,104]. The fact that the conventional lateral hypothalamus (i.e., as conceived by Nieuwenhuys et al. [18] or Swanson [19]) stretches from dorsal to ventral (in prosomeric terms) orthogonally to a series of longitudinal progenitor domains [17] already suggested that it probably was molecularly heterogeneous and, accordingly, was being treated simplistically as a unit in the literature. Swanson [105] followed the columnar approach and visualized the LH as a longitudinal hypothalamic structure; he thus distinguished a number of anteroposterior (AP) parts of LH neighboring diverse nuclear landmarks; his AP divisions of LH apparently coincide partly with our DV components (see below). Note columnar theory lacks a developmental causal explanation for AP subdivisions within a longitudinal column, so that Swanson's [105] concept of LH cannot be explained developmentally.
Among the markers used in the present study, chosen empirically for their partial labeling of the LH (see Table 2), there are neuropeptides and neurotransmitters, as well as transcription factors, synaptic proteins, vesicular transporters, and receptors ( Table 1). The transmitters are not very selective, except perhaps histamine and dopamine. Fifteen glutamatergic and 15 GABAergic otherwise different neuronal cell clusters were identified in a recent single-cell RNAseq transcriptomic classification of LH cell populations in postnatal mouse brains; each cluster was distinguished by differential expression of transcription factors and neuropeptides, among other markers [42]. However, the cited analysis was focused exclusively on what we regard as basal LH subpopulations, described columnarly as "caudal to the retrochiasmatic area and rostral to the tuberomammillary nucleus in the posterior hypothalamus." A regional differential location of the identified cell clusters was not investigated. Since several of those cell markers were included in the present study, we will try to correlate both sets of data. Another scRNA-seq analysis of the same group was centered on the mamillary/retromamillary region, also relevant for our approach [87]. Hypothalamic single-cell RNAseq transcriptomic data were reported likewise by Kim et al. [46], who included in their Supplementary data (their Supplementary Fig. 10) an analysis of the LH in E11 to E16 embryonic mice; they identified 11 glutamatergic and 2 GABAergic cell clusters. The analogous study of Chen et al. [43] did not mention LH cell populations, although some of the cell clusters identified may match with LH cell subpopulations found in the present work.
Our detailed prosomeric genoarchitectonic analysis thus reexamines cytoarchitectonic studies of LH cell populations associated topographically with the medial forebrain bundle [18,23,24] and expands a tentative genoarchitectonic dorsoventral LH subdivision proposed initially by Puelles et al. [17]. These authors argued that, as classically defined, the LH belongs strictly to the peduncular hypothalamus, understood by them to derive from the hp1 hypothalamic prosomere, which expands dorsally into the evaginated telencephalon (Fig. 16). The related transverse intrahypothalamic boundary limiting hp1 from hp2 (or peduncular from terminal hypothalamus territories; PHy/THy) passes roughly in front of the fornix tract, thus including a dorsoventral series of perifornical cell groups within the LH (Figs. 1 and 16). Tridimensional morphogenetic deformation of the PHy as marked by radial glia causes its superficial part to bend at basal levels around the VM mass to reach the lateral tuberal surface rostrally to the peduncle ( [17]; their Fig. 8.12). Moreover, Puelles et al. [17] also updated the LH concept by distinguishing the LH, understood as an intermediate stratum within PHy, from a covering superficial stratum occupied by the peduncle (lateral forebrain bundle) and associated superficial cell groups such as the STh, parasubthalamic, tuberal, and entopeduncular nuclei. Superficial populations lying rostral to the peduncle, such as the lateral tuberal and parasubthalamic nuclei, which were included by Geeraedts et al. [24] as LH cell populations, are thus ascribed by us to the superficial hypothalamic stratum and not included in our study of LH. Correspondence of cytoarchitectonic LH subdivisions [23,24] with genoarchitectonic LH subregions (present results) should be regarded as merely tentative.

Phenotype of LH Basal Subpopulations
Previously, Puelles et al. [17] subdivided the basal LH in a ventrobasal LH subdomain containing Cartpt-and Galpositive GABAergic cells and a dorsobasal LH subdomain with Hcrt-expressing glutamatergic neurons; periretromamillary and retromamillary LH subpopulations presently introduced by us were not contemplated. According to our present schema the basal LH is divided in five molecularly distinguishable subregions-LH1-LH5-superposed in ventrodorsal order one upon another. These are associated causally with the fundamental ventrodorsal molecular regionalization of the basal peduncular hypothalamus (basal PHy) in retromamillary, periretromamillary, and ventral, intermediate, and dorsal retrotuberal progenitor subdomains (RM, PRM, RTuV, RTuI, RTuD; [17]), representing in each case the respective LH intermediate stratum. Though it may be unclear how far the RM and PRM domains are traversed by the medial forebrain bundle (the latter is usually taken as an auxiliary criterium to define the LH, and we believe it traverses at least the PRM), we regard the structural and developmental (outside-in neurogenetic layering) classification into periventricular, medial, lateral, and superficial strata strictly applicable to these ventralmost DV tiers. The RM, PRM, and RTuD subdomains contain predominantly glutamatergic cells, as observed also in LH1, LH2, and LH5, respectively, while RTuI shows mainly GABAergic cells (as does LH4), and RTuV displays histaminergic neurons within a singular molecular profile (like LH3) ( [17]; present results; Table 1). LH5 cells are distinctly characterized by Hcrt-expression and clearly correspond to the (now obsolete) dorsobasal LH subdomain proposed by Puelles et al. [17], whereas LH4-LH3 jointly correspond to the (likewise obsolete) old ventrobasal LH subdomain. We think that primary ventrodorsal basal plate regionalization within PHy and a diversity of local LH cell subpopulations originated Fig. 11 Nkx2.2 expression in the lateral hypothalamus (LH) and other hypothalamic regions in two P4 mouse brains sectioned either sagittally (A, B) or transversally to the hypothalamus (C-H; that is, parallel to the peduncle). The Nkx2.2 hybridization (dark blue) was followed by an immunoreaction with a tyrosine hydroxylase (TH) antibody (brown). The TH-positive dopaminergic fibers of the nigrostriatal tract (ns), a component of the medial forebrain bundle, contributes to the identification of LH subpopulations. The dash black lines indicate thalamo-prethalamic, prethalamo-hypothalamic, and intrahypothalamic transverse interprosomeric boundaries. The green dash line indicates the alar/basal limit. The black line marks the hypothalamo-telencephalic boundary. (A, B) Nkx2.2-positive cells appear in sagittal sections both in the alar paraventricular area (Pa) and the basal ventromedial nucleus (VM); in between these populations, there is a thin longitudinal periventricular Nkx2.2-positive population identified as the rostral liminar band (RLi; modified from Puelles et al. [40]; see their Fig. 8f, g). Reported embryonic observations suggest that both the alar and basal Nkx2.2-positive derivatives originate via early tangential migrations from the primary RLi locus, placed roughly along the alar-basal boundary (Puelles et al. [17]).  Fig. 16) and the notochord and by secondarily added WNT8 or other effects released from the likewise longitudinal circumventricular ventricular hypothalamic organ differentiated at RTuV level (HVO; [17,73,74]). Antagonistic dorsalizing BMP/WNT effects spreading from the roof plate are possibly involved as well at initial stages, but may be weaker afterwards, due to the interposed mass of the telencephalon (see [74]).
With respect to cytoarchitectonic LH subdivisions proposed in Geeraedts et al. [24] their lateral supramamillary nucleus (our RML) clearly matches our LH1, whereas the sum of our LH2-LH3 may correspond to their part 3 and caudal part 4 of the lateral hypothalamic nucleus. In their turn, our LH4-LH5 corresponds to their part 2 and rostral part 4 of the lateral hypothalamic nucleus.
We found expression of at least twenty genes in LH1 cell subpopulations, eight of which are selective of this LH subregion (see Table 2). Most LH1 cells are glutamatergic and express Slc17a6 (also named vGlut2; [17]; present results); this gene encodes a vesicular glutamate transporter. LH1 may correlate with cluster 4 of Mickelsen et al. [42] that contained Tac1-and Pitx2-expressing glutamatergic cells. Tac1 encodes a precursor protein that produces the peptides substance P and neurokinin A of the tachykinin family, whereas Pitx2 encodes a paired homeodomain transcription  Fig. 3). This migration is severely altered in Pitx2 null mice mutants [80]. It is not yet understood what causes some RM cells to migrate tangentially away in different directions while others stay in place (see [82] on the ventral premamillary nucleus).
According to present results, LH2 cells may be discriminated by the expression of five markers (Dlk1, Irs4, Otp, Peg10, and Penk) which are not expressed in either LH1 or LH3 cells (Table 2). LH2 cell subpopulations are restricted to the dorsoventral level of the periretromamillary subdomain (PRM; [17]) within basal PHy, whose cells were previously characterized by massive Sim1/Otp expression from early embryonic stages onwards ( [17,82,84]; present results). In columnar accounts, the PRM area was included in the classical "posterior hypothalamus" (see [86]; his Figs. 16-18, reproduced in [17]; their Fig. 8.3). However, the columnar "posterior hypothalamus" included also a strictly non-hypothalamic diencephalic tegmental part that reaches the retroflex tract in most accounts (see historical review of this concept in [17]). Expression of Otp and Sim1 dissociates cleanly the hypothalamic and diencephalic parts of the old "posterior hypothalamus.' Both Otp and Sim1 genes have an essential role in the differentiation of hypothalamic neuroendocrine glutamatergic subpopulations such an the paraventricular, supraoptic, and arcuate nuclei [110][111][112][113][114][115], but their function in the differentiation of LH2 cells has not been analyzed. In the adult rat brain Irs4-expressing cells were illustrated at what we identify as the LH2 locus ( [116]; their Fig. 2F).
The glutamatergic cell clusters 8 (with Otp/Cbln2/Ebf3positive cells) and 11 (with Otp/Mdga1-positive cells) of Mickelsen et al. [42] may also correspond topographically with LH2. All the additional markers cited of these phenotypes that co-express Otp are present in LH2 cells (present results; some of them also appear within LH1, but devoid of Otp co-expression). Ebf3 and Cbln2 cells were previously illustrated during mouse development in hypothalamic retromamillary and periretromamillary subdomains [117,118]. Moreover, Litwack et al. [119] cited the presence of Mdga1positive cells in the adult rat hypothalamus (without showing them). The Ebf3 gene encodes the "early B cell factor 3," a member of the EBF transcription factor family, which has a role in neuronal differentiation and maturation [117]; the Mdga1 gene encodes a glycosylphosphatidylinositol (GPI)anchored membrane protein, a cell surface glycoprotein; and the Cbln2 gen encodes a cerebellin 2 precursor protein implicated, as other cerebellins, in the formation and/or function of synapses [120][121][122][123][124][125]. Production of histaminergic Hdc-expressing neurons characterizes the thin LH3 sector, which is part of the ventral retrotuberal subdomain (RTuV), as defined by Puelles et al. [17]. This cell type is also produced in the continuous ventral tuberal subdomain of THy (TuV), and both coincide with the hypothalamic ventricular organ (HVO), a potential secondary organizer [17,73,74]. In recent single-cell transcriptomic classifications of hypothalamic populations, a separate histaminergic cell cluster was identified, distinct from glutamatergic and GABAergic cell clusters [43,87]. Consistently, the LH3 cells do not express either vesicular glutamate (vGLTU2) or GABA transporters (Slc17a6 and Slc32a1 genes; [17,43]; present results).  [88] showed Prph expression in mouse histaminergic neurons. Hu et al. [126] detected in the postnatal hypothalamus Wif1 signal (encoding the WNT inhibitory factor 1; see also [127]) but did not show its topography. These coincident findings may imply that the thin LH3 subdomain produces a single cell type, though, of course, many more markers would need to be examined to fully establish this point. Note that the unlabeled cells seen intermixed with the labeled ones may derive by dispersion from molecularly different neighboring LH subdomains.
The larger LH4 subdomain is a component of the intermediate RTu subdomain (RTuI), characterized medially by the peduncular dorsomedial nucleus, which displays a glutamatergic dense core surrounded by a dispersed shell population of GABAergic cells (present results; [17]; the latter authors deduced that the GABAergic neurons of DM are endogenously produced, whereas the glutamatergic ones in the core possibly are migrated tangentially from a different source). There is also in THy a continuous terminal DM nucleus portion with analogous core/shell structure. LH4 cells indeed express markers related to the GABAergic phenotype such as Slc32a1, Gad1, and Gad2. Moreover, they co-express the genes Cartpt, Gal, Nts, and Tac2 encoding respectively the neuropeptides cocaine-and amphetamine-regulated transcript, galanin, neurotensin, and tachykinin [52,53]. Dlk1 is also expressed in LH4 cells; this gene encodes the delta-like non-canonical notch ligand 1, a protein implicated in cell proliferation or differentiation depending on the local context, which is also involved in hormone-producing pituitary cells [128][129][130][131][132][133][134]. These cell phenotypes match with two GABAergic clusters (1 and 3) identified by Mickelsen et al. [42]; cluster 3 is characterized by co-expression of Cartpt and Nts, and cluster 1 contains cells expressing Nts, Gal, and Dlk1 transcripts ( [42]; their Fig. 6). Other markers found in both our LH4 cells and cell cluster 3 are Gpr101, Nek7, Pdyn, and Penk ( [42]; their Fig. 6).
Trh-positive cells appeared likewise associated with LH4 (present results). Mickelsen et al. [42] found two glutamatergic Trh-expressing clusters; one of them also contains Cbln2 signal, a marker also found in LH4 cells (present results; Table 2); both cell clusters co-express Otp a characteristic marker of the paraventricular (Pa) and periretromamillary domains [17,84,111,138]. According to these findings, Trh-positive glutamatergic cells in LH4 may have migrated ventralward from the Otp-positive peduncular PaV progenitor domain, as was previously suggested [104]. PaV contains mostly locally originated glutamatergic cells ( [17]; present results).
Hcrt-and Pmch-expressing neurons characterize LH5, a component of the dorsal retrotuberal subdomain (RTuD), the dorsalmost part of the basal plate. Swanson et al. [139] also observed these cell types in the adult rat brain, describing them somewhat less precisely within their columnar paradigm as located in the "caudal part" of LH (roughly equivalent to our basal territory; note the basal notion allows a causal explanation). Hcrt and Pmch cells are distinct intermixed populations within LH5 [140]. 50% of rat hypocretin (HCRT) neurons were found to be perifornical [75]. Our data suggest that while both Hcrt and Pmch cell populations seem to have the same anatomic origin within the RTuD area, where they tend to distribute at least within the medial and lateral strata (note the medial stratum is shared also with the A13 dopaminergic neurons; [17,89]), the Pmch cells secondarily spread out more than the Hcrt cells, invading parts of the subjacent RTuI area (DM-P area) and even the caudally neighboring prethalamic zona incerta (see also [141]). The Hcrt and Pmch genes encode hipocretin/ orexin and pro-melanin-concentrating hormone, respectively; both peptides are implicated in arousal, feeding, and energy metabolism [9,27,28,[142][143][144][145][146][147][148][149]. Mouse Pmch and Hcrt subpopulations have been ratified as two molecularly distinct glutamatergic clusters by single-cell transcriptomic analysis [42]. Consistently with such data, the LH5 consists largely of Slc17a6-positive glutamatergic cells ( [17,141]; present results).
The cluster containing the large-sized Hcrt-expressing glutamatergic neurons co-expresses, among other markers, Pdyn, and Nek7 [42]; these two phenotypes are also found in LH5 (present results; see Table 2). Interestingly, the Lhx9 gene which encodes a transcription factor member of the LIM homeodomain family is co-expressed in all Hcrt-positive cells, whereas the transcription factor NKX2.1 is not expressed at all in mantle Hcrt cells [150]. Indeed, Lhx9 is expressed early on in the Nkx2.1-negative mantle layer of the RTuD subregion lying just ventral to the Lhx6/Nkx2.2positive alar liminar band [17,36]. The LHX9 transcription factor has been implicated in the specification of hypocretin cells in mice [36,151,152] and zebrafish [153,154]. Removal of this gene resulted in a 30% decrease of Hcrt neurons in the mouse [151]. Hcrt cells also co-express Nptx2 (99%), and Nek7 (84%) transcripts [42,150], both also observed by us in LH5 cells. Nptx2 encodes neuronal pentraxin-2 and Nek7 (NIMA) encodes the never in mitosis gene A, a member of the serine/threonine kinases family.
Pmch neurons produce melanin-concentrating hormone (MCH). They are mostly distinct from Hcrt cells though they coincide within LH5 and the local medial stratum (our posterobasal nucleus combined with the A13 dopaminergic cell group). Pmch neurons possibly emerge within the Nkx2.2-positive and Lhx9-negative subliminar subdomain of RTuD in both rats and mice [17]; the underlying Nkx2.2negative "main subdomain" of RTuD is where Lhx9 signal is observed, coinciding with the mantle locus where Nkx2.1 signal is absent. Croizier et al. [155,156] referred to this location as the "intrahypothalamic diagonal" of Shimogori et al. [36], without noticing that this origin of Pmch neurons is largely restricted therein to what we call peduncular hypothalamus, a partition not recognized in their hypothalamic schema. The combined longitudinal TuD/RTuD domain of the prosomeric model was wrongly interpreted as "diagonal" because Shimogori et al. [36] naively combined the orthogonal columnar and prosomeric length axes, thus obtaining arbitrarily an extra diagonal direction in morphospace ( [17]; see footnote in p.246).
Differently than the Hcrt cells, Pmch-positive glutamatergic neurons, identified as cluster 1 by Mickelsen et al. [42], co-express Nkx2.1 in the rat embryo [155]. Nkx2.1 signal was also found in 60% of MCH neurons in the adult mouse [150]. We thus propose that Pmch/MCH cells originate from the Nkx2.1-positive subliminar RTuD progenitor subdomain at early embryonic stages, while Hcrt neurons, known to be Nkx2.1-negative, probably arise within the underlying Lhx9-positive "main" RTuD progenitor subdomain, whose mantle later is devoid of Nkx2.1 signal (not so the ventricular zone). Consistently with our conclusion, Pmch/ MCH cells do not express Lhx9 [155], a transcription factor related instead to Hcrt cell differentiation [36,[151][152][153][154]. Thus, NKX2.1 seems to be continuously transcribed during MCH cell differentiation, as a characteristic of the subliminar RTuD domain, while LHX9 signal, which is expressed selectively within the underlying main RTuD subdomain in combination with downregulation of Nkx2.1 in postmitotic neurons, is clearly related to local HCRT cell differentiation. The apparent ventral displacement of some Pmch neurons beyond LH5 into LH4 may reflect the tendency of neurons derived from the RTuD band to migrate ventralwards (discussed in [17], pp.283-286).
The general expression of Nkx2.1 in the basal hypothalamus reportedly depends on early activation of GLI1 downstream of the SHH signal [94,157]. Inhibition of the SHH pathway by cyclopamine in E11 mice produces a dramatical reduction of Pmch mRNA level and no Pmch/MCH signal was detected in embryonic mutant mice lacking Shh [155].
A subpopulation of Cartpt cells was found within the same LH5 subregion (present results). Cartpt/CART is also co-expressed in a subpopulation of Pmch/MCH neurons ( [136,141,158,159]; reviewed in [135]) that specifically project to the cerebral cortex [160]. 66% of MCH neurons co-express the Cartpt transcript in the mouse [150] and 70% in the rat [136]. Both Cartpt and Pmch cells were found in LH5 and LH4 in correlation with either GABAergic or glutamatergic phenotypes. Approximately 67% CARTexpressing neurons express Gad67 [136]; we estimate that this subtype probably appears mostly located within the LH4 subregion, rather than in LH5.
Interestingly, the Dlk1 gene is likewise expressed in LH5 cells (present results), as well as in the related neuroepithelium; it codifies for DLK1, a protein expressed in virtually all hypocretin-(HCRT-or orexin-) and dynorphin-positive (DYN) neurons but not in MCH-or CART neurons in the rat LH [161]. This suggests again a dissociation in the cellular phenotypes of neurons born at the subliminar versus main RTuD subdomains. DLK1 promotes neurogenesis in progenitors by modulating Notch and BMP signaling [162,163] and may thus be implicated in generating hypocretin/ dynorphin cells. Furthermore, a small group of DLK1-positive neurons also express galanin (GAL) and neurotensin (NTS), though galanin is not expressed in hypocretin and MCH neurons [137].
According to present results, Igf1, which encodes the insulin-like growth factor 1 (IGF1), is also expressed selectively in many LH5 cells; all Pmch-expressing neurons contains also Igf1 [141]. IGF1 is an essential growth factor involved in different neural processes including neurogenesis. Igf1 overexpression produces an increase in brain size [164] and Igf1 lack of function originates a reduction in brain size in the mouse [165]. Moreover, IGF1 is involved in proliferation and migration of GABAergic and glutamatergic neurons as well as in synaptogenesis [166,167]. In the adult brain, IGF1 is also implicated functionally in the somatotropic axis, acting downstream of the growth hormone, as well as in hippocampal neurogenesis, neuronal circuit maturation, and cell neuroplasticity (see reviews in [168,169]).
Wang et at [141] recently subdivided the tuberal LH (including perifornical populations) in nine spatially and molecularly distinct subregions analyzing markers from single-cell RNA sequencing transcriptomic (scRNA-sq) datasets by Expansion-Assisted Iterative Fluorescence In Situ Hybridization (EASI-FISH) in 300-µm-thick brain sections of the adult male mouse. For scRNA-sq analysis, they used transgenic male mice (Agrp-IRES-Cre x Ai9) and the authors identified the presumed boundaries of LH as the territory targeted by AGRP-positive axon terminals. The fact that they only studied the tuberal LH suggests that this criterium leaves out the alar parts of LH and is therefore imperfect. They found 17 glutamatergic (Slc17a6expressing) and 17 GABAergic (Sclc32a1-expressing) cell clusters. Based on the expression of Meis2, Otp, Slc17a6, and Sclc32a1 markers in thick sections, Wang et al. [141] proposed nine subregions in the lateral hypothalamic area (referring actually mainly to its tuberal/retrotuberal part according to our model) arranged in the mediolateral and dorsoventral dimensions of the columnar model (note their dorsoventral dimension in coronal sections corresponds to our prosomeric caudorostral dimension). Three of these subregions were depicted adjacent to the prethalamic incertal alar region and were identified as "dorsal diagonal band," "hypocretin-positive diagonal band," and "dorsolateral diagonal" LH parts. Another LH region (subdivided into lateral and medial subregions) was named "suprafornical diagonal band." The descriptor "diagonal band" refers to the "intrahypothalamic diagonal" of Shimogori et al. [36], which corresponds to our alar longitudinal subparaventricular area [17]. However, we think that all these areas called "diagonal band", and particularly the hypocretin-positive one, lie within the basal hypothalamus, at the dorsal retrotuberal region (see our Fig. 7A), and are thus wrongly referred to the "intrahypothalamic diagonal." In addition, several mediolaterally disposed perifornical subregions -also in the basal hypothalamus, in our opinion-were distinguished, identified as "medial" and "lateral" fornical regions (the latter subdivided into medioventral, dorsolateral, and ventrolateral "lateral" parts). The "dorsal diagonal" and the three "lateral" fornical subregions contain mainly excitatory neurons (Slc17a6 + /Otp +), as does the "dorsolateral diagonal" subregion (Slc17a6 + /Meis2 +), whereas inhibitory cells largely occupy the medial and lateral "suprafornical diagonal" subregions whose cells are Slc32a1 + /Meis2 + . The "medial fornical" subregion also shows inhibitory cells expressing Slc32a1 + /Otp + . We understand the LH region analyzed by Wang et al. [141] as corresponding topographically in part to our LH5 domain (mainly containing glutamatergic cells) and in part to our LH4 domain (containing a mixture of glutamatergic and GABAergic cells) sectors. Both lie within the peduncular basal hypothalamic LH region which is classically referred to as "tuberal area," though the prosomeric model more precisely identifies it within the hp1 prosomere and PHy as "retrotuberal area." Indeed, the latter is restricted to the peduncular hypothalamus whereas "tuberal area" is a descriptor of the corresponding terminal hypothalamic basal domain, where no LH exists (Fig. 16). The coronal section illustrated in their Fig. 3A [141] largely corresponds to our Figs. 13E, F. This part of the basal LH possibly relates also to the lateral and medial "middle LH groups" described by Swanson [105] as juxtadorsomedial, perifornical, (supra-and subfornical), and lateral parvicellular LH subdivisions (see also Swanson et al. [170]; his "supporting information 9 and 10"); other LH subdivisions of Swanson, such as diverse "juxtaventromedial parts," lie topographically within our terminal hypothalamus (THy), which according to us does not contain any lateral hypothalamic populations (and contains no part of the medial forebrain bundle). The "hypocretin-positive diagonal" subregion of Wang et al. [141] relates clearly to our LH5, since it typically contains a glutamatergic Hcrt-positive subpopulation whose neurons receive AGRP afferences from the arcuate nucleus [140,171], jointly with a distinct Pmch-positive peptidergic subpopulation also detected by Wang et al. [141]. A high percentage of Pmch-expressing neurons are Cartpt-positive (77% according to Wang et al. [141]), a subpopulation which is also mainly related to our LH5 (present results). However, other peptidergic populations such as Nts-and Trh-expressing cells, and a group of Cartpt-positive cells, were associated by us with LH4. Wang et al. [141] reported Trh-positive neurons and Otp-positive cells in their fornical subregions. However, the presence of Otp/Trh-expressing neurons in the retrotuberal LH region may be due to tangential migrations reported from alar and/ or basal Otp-positive progenitor domains [37,84], i.e., the alar paraventricular area and/or the basal perimamillary/retromamillary domain, both of which produce glutamatergic cells (see e.g. [138] and our Fig. 4A). According to Wang et al. [141], in their "Hcrt-positive diagonal" territory there appear Nts-positive neurons co-expressing Grp101 and Gal signals; both markers were detected in cell populations of the LH5 sector in our present work (see Table 2). On the whole, the Wang et al. [141] analysis was handicapped in its anatomical approach by an incomplete and possibly outdated model of the LH region. Transcriptomic studies are obviously doubly sensitive to this sort of problem since it affects both the dissection of material and the interpretation of results.
In summary, we subdivided the basal LH into five regions (see Fig. 16), three containing glutamatergic cells (LH1, LH2, LH5), a fourth including mostly GABAergic cells (LH4), and the fifth containing essentially only histaminergic cells (LH3). Each of these LH subregions is characterized by unique combinations of transcription factors, cell adhesive molecules, neuropeptides, and receptors, among other markers. Whereas our topographic analysis of cellular heterogeneity in the basal LH in correlation with postulated developmental units (basal dorsoventral progenitor domains of Puelles et al. [17]) led us to the LH1-LH5 compartments, the single-cell transcriptomic LH study of Mickelsen et al. [42] distinguished 15 glutamatergic and 15 GABAergic basal neuronal clusters (17 in the study of Wang et al. [141], referring mainly to our LH4 and LH5 domains). This implies that, as discussed amply in the case of LH5, several cell clusters apparently may coexist intermixed within our developmental units (with the possible exception of the simpler histaminergic LH3 unit). A salt-and-pepper generative pattern may need to be considered in some of these cases, since a temporal sequence is not likely given the equivalent spatial stratification of the different intermixed LH cells (overall outside-in neurogenetic pattern). With respect to glutamatergic cell clusters, the Tac1-and Pitx2-positive cell cluster 4 of Mickelsen et al. [42] may be a migrated subpopulation of LH1, the Otp/Cbln2/Ebf3-positive cell cluster 8 and Otp/Mdga1positive cell cluster 11 may be subpopulations of LH2, and the Pmch/Nkx2.1/Nkx2.2/Carpt-expressing cell cluster 1 (MCH cluster) and the Hcrt/Dlk1/Lhx9/Nptx2/Nek7/ Plagl1/Pdyn-positive cell cluster 5 (hypocretin cluster) are subpopulations of LH5. In this case, we discussed data suggesting that the related progenitor domain (RTuD) probably may be subdivided into a thin dorsal subliminar Nkx2.2-positive band (the source of Pmch cells) and an underlying main subdomain that lacks this signal and combines selectively Lhx9 expression (origin of Hcrt cells). As regards GABAergic cell clusters, the cell clusters 1 (Nts/Gal/Dlk1-expressing) and 3 (Carpt/Nts/Gpr101/Nek7/ Pdyn/Penk-expressing) of Mickelsen et al. [42] may represent distinct LH4 subpopulations. A histaminergic cell cluster (Hdc/Prph/Wif1-expressing) described by Chen et al. [43] matches the LH3 unit (but note that variously displaced Hdc cells appear at all radial levels of RTuV and TuV domains (and even within RM and M), this population thus not being totally restricted to LH [17]).

Phenotype of LH Alar Subpopulations
Puelles et al. [17] tentatively subdivided the mouse alar LH in three dorsoventral domains named according to the estimated prethalamic neighbors as pre-incertal LH (with Dlx5/6-and Nkx2.2/Lhx6-positive GABAergic cells), prereticular LH (with Tbr1-expressing glutamatergic cells) and pre-eminential LH (with Fezf2-and Tbr1-positive glutamatergic cells). Our recent analysis of prethalamic genoarchitecture [40] disclosed an error in one of the cited associations: the incertal prethalamic territory is not continuous rostralward with the Dlx5/6-and Nkx2.2/Lhx6-positive subparaventricular hypothalamic domain, but with the Dlx5/6and Nkx2.2/Lhx6-negative ventral part of the paraventricular domain. This means that all three initially postulated alar LH units are related to the tri-partite paraventricular complex (PaD, PaC, PaV). The underlying and molecularly distinct hypothalamic subparaventricular domain was found to be continuous backwards instead with the rostral liminar band (RLi) defined by us just above the prethalamic alar-basal boundary; the latter is distinguished by early expression of Nkx2.2 and Lhx6 jointly with Dlx5/6 and Arx [40]. This correction of the hypothalamus model requires the proposal of a fourth alar LH component associated with the subparaventricular domain. In our present alphanumeric terminology, we thus subdivide the alar LH region into four subregions: LH6-LH9. LH6 is a component of the peduncular subparaventricular area (PSPa) which is generally characterized by Dlx/Arx-positive GABAergic cells. The LH7-LH9 units belong respectively to the ventral, central, and dorsal peduncular paraventricular area (PPa), which generally displays Otp/Sim1-positive glutamatergic cells ( [17,133]; present results).
Comparison with the cytoarchitectonic LH subdivisions of Geeraedts et al. [23,24] suggests that their lateral hypothalamic nucleus (part 1), intermediate hypothalamic area (part 1), and dorsal hypothalamic area may correspond to our LH6-LH8, whereas our LH9 sector (included in the TOH domain) may be equated to a subarea of the lateral preoptic region in their account.
Alar Lhx6 signal appears just dorsal to basal Lhx9 expression in the embryonic mouse hypothalamus [17,36], with Lhx9 signal being restricted to the TuD/RTuD area [17]. Lhx6 signal accordingly relates with the peduncular and terminal SPa area. Lhx6 mRNA is known to be expressed in Cart (Cartpt)-positive neurons at P6 [36]. However, Cartpt-expressing cells are not found in alar but in basal LH populations in the adult mouse (present results); a ventralwards migration of SPa-originated Cartpt/Lhx6 cells would explain these findings (see [90]. Recently, Kim et al. [172] found three clusters of postnatal Lhx6-eGFP hypothalamic neurons whose cells were GABAergic (expressing GABA transporter Slc32a1), and showed either peptidergic or nonpeptidergic markers (Calb1, Calb2, Cck, Gal, Nos1, Pnoc, Penk, Th, Trh; their Fig. 3). These authors did not assess the possibly differential ascription of such neurons to the two hypothalamic sites known to express Lhx6 [36], namely what the latter authors call "intrahypothalamic diagonal" (i.e., our longitudinal RLi band associated to SPa; [40]) and "tuberomammillary band" (i.e., our RTuV/TuV subdomain; [17]). Pnoc-positive cells are restricted to LH6 at E18.5, but the Pnoc-positive population is widely dispersed among hypothalamic cells at postnatal mouse stages (present results; [173]; Allen Developing Mouse Brain Atlas). Kim et al. [172] found Pnoc signal associated with postnatal Lhx6-expressing hypothalamic neurons in one of the cited clusters, which may thus correspond to the RLi/SPa progenitor domain. The Pnoc gene encodes the prepronociceptin peptide precursor. Some LH6 cells also express the Th transcript, encoding the enzyme tyrosine hydroxylase (TH) that indicates a dopaminergic cell phenotype; Th was one of the markers co-expressed in Lhx6 cells in two of the mentioned clusters ( [172]; see [89]).
LH7 is the ventralmost lateral hypothalamus component of the peduncular paraventricular alar subdomain (PPaV). Its cells are Slc17a6-positive (a vesicular glutamate transporter) and combine Otp/Sim1 expression. The LH7 subregion contacts caudalwards the GABAergic incertal prethalamic complex ( [40]; present results). According to our results, three markers are restricted to LH7 cells, Trh (encoding thyrotropin-releasing hormone), Sst (encoding somatostatin), and Mdga1 (encoding MAM domain-containing glycosylphosphatidylinositol anchor 1). A PaV source of Trh-expressing cells was previously suggested by Morales-Delgado et al. [104]. Additionally, Sst-positive cells were found in an Otpexpressing perifornical subpopulation of the PPaV area, next to the intrahypothalamic border ( [133]; their Fig. 5B, C, M, N). Mdga1 expression was not previously reported within the alar LH but was correlated instead with a basal LH cluster [42]. Nkx2.2-positive cells held to have migrated dorsalward into the Pa area apparently originate from the RLi band [17] and colonize preferentially the medial and lateral PPaV portions. Some of these cells are found in the LH7 where they were not previously identified. Numerous tangential cell migrations have been reported in the hypothalamus [17,90,104,133,174,175], so that a measure of caution is needed when interpreting postnatal and adult data on cells located in postmigratory locations more or less distant from the progenitor areas where they apparently originated.
The overlying LH8 sector displays characteristically in the adult mouse Meis2-, Lhx5, and Tbr1-positive glutamatergic cell subpopulations. The three markers have an early embryonic expression in the prethalamic eminence of the diencephalon, from where cells migrate partly into the paraventricular area and other forebrain areas. There is a progressive increase in the number of superficial and intermediate Tbr1-, Lhx5-, and Meis2-positive cells in the central paraventricular area which contrasts with a decrease of eminential Meis2-, Lhx5-, and Tbr1-positive cells (see Allen Developing Mouse Brain Atlas). Descriptive and fate-mapping results in the chick showed that paraventricular Tbr1expressing cells migrate tangentially rostralwards from the prethalamic eminence [91,92] whose glutamatergic cells initially express Tbr1/TBR1, Lhx5, and Meis2 [40,93,176]. Mouse Tbr1-, Lhx5-, and Meis2-positive LH8 cells accordingly may be likewise migrated eminential cells.
LH9 lies within the dorsal part of the PPa area (PPaD) which is characterized by Sim1-and Otp expression in its mantle layer [17], as is the rest of PPa, but also differentially expresses Foxg1 [37]. Typically, it displays next to the hypothalamo-diencephalic boundary a dorsalward Sim1-positive expansion of PPaD into a curved spike that enters the telencephalon through the floor of the interventricular foramen and follows the bottom of the terminal sulcus to reach the pallial amygdala. This telencephalic extension was recently named the hypothalamo-amygdalar corridor (HyA; [93]), but also coincides with the telencephalo-opto-hypothalamic domain -TOH-identified by Morales et al. [37] and apparently contains Sim1/Otp cells migrated from underlying parts of the paraventricular area.
A new concept affects this alar domain where the hypothalamus limits with the telencephalon. The peduncular and terminal subdivisions of the dorsal paraventricular (PPaD/TPaD) domain were recently recognized as a telencephalo-opto-hypothalamic area due to its expression of Foxg1 (TOH; [37]). This area generates glutamatergic cells co-expressing Otp, but not Nkx2.1 or Dlx genes, indicating a lack of subpallial character; these excitatory cells invade partially the BST complex and the medial amygdala [37]. These authors proposed that TOH be considered a novel telencephalic radial domain transitional between subpallium and alar hypothalamus but acknowledged that it essentially corresponds topographically with the dorsal paraventricular hypothalamic subregion of Puelles et al. [17], the domain that builds the HyA (see also [70]). We continue to ascribe the TOH/HyA domain to the alar hypothalamus, given its fundamental genoarchitectonic difference with the Dlx-expressing telencephalic subpallium, and considering that ontogenetically the whole telencephalon is a hypothalamic derivative (we are speaking of evaginated and non-evaginated parts of the same alar plate, that of the hp1 and hp2 prosomeres). There is no difficulty in assuming that Foxg1 expression extends dorsoventrally out of the topologically dorsal telencephalic alar region into the continuous hypothalamic one, coming to overlap there partly the field of paraventricular Sim1/Otp signal. This interpretation also keeps the definition of the telencephalon simple, limited to subpallium and pallium. We only must know now that the PPaD also expresses Foxg1, being thus partially distinct molecularly from the central (PPaC) and ventral (PPaV) paraventricular subregions devoid of this signal. TOH/PaD cells also lack Brn2 (Pou3f2) expression, present in the central and ventral Pa subdomains [110].
LH9 selectively contains numerous dispersed Satb2expressing cells intermingled with the medial forebrain bundle ( Fig. 9H; see also [178]), laterally to a dense Satb2positive periventricular island (see Fig. 9I). This population lying apparently at the hypothalamo-telencephalic border zone (coinciding with the dorsal part of the hypothalamic paraventricular complex) should accordingly be considered an additional LH subregion representing the end of the lateral hypothalamus next to the subpallial telencephalon (e.g., substantia innominata).

Differential Connectivity Pattern of the LH Cell Subregions
The LH is clearly genoarchitecturally heterogenous (LH1-LH9); it is of interest to explore how far it is also hodologically and functionally diverse, irrespective that much literature refers to LH as a unitary concept. Reportedly, the LH plays a critical role in coordinating several physiological and behavioral functions, such as arousal, feeding, stress, reward, and motivated behaviors [1, 4, 7-11, 28, 179-181]. The LH has extensive reciprocal connections with multiple brain regions. Ascending LH projections are implicated in cognitive and emotional processing, in contrast with descending projections to the brainstem and spinal cord related to wakefulness, feeding, and various autonomic and neuroendocrine functions (reviewed in [28]). A major problem in the study of LH connectivity using tracer techniques (apart from the structural map of LH subdivisions considered) is the distinction of potential co-labeling of medial forebrain bundle fibers of passage.
A good number of connectivity experiments centered at different dorsoventral LH levels (according to a prosomeric interpretation; note authors using the columnar model will describe them as anteroposterior or rostrocaudal partitions) have shown differential connections that can be related with our proposal of LH genoarchitectonic regionalization (e.g. [25,75,160,[182][183][184][185][186][187][188][189][190][191][192]). For instance, Pelosi et al. [191] showed a distinct dorsoventral regionalization (described as rostrocaudal) of the efferent LH projections to different parts of the dorsal periaqueductal gray in the rat. Likewise, Hahn and Swanson [25], comparing afferents and efferents of (alar) juxtaparaventricular and (basal) suprafornical LH subregions, demonstrated that these domains have different connectivity patterns and deduced different behavior roles (defensive versus ingestive). It was also reported that olfactory input reaches specifically the basal LH [193]. Unfortunately, interpretation and systematization of such interesting results within the columnar paradigm are difficult and usually approximative. The columnar authors are limited to crude preoptic, anterior, tuberal, and posterior hypothalamic LH subregions devoid of postulated molecularly distinct progenitor domains. The columnar concept contrarily presupposes longitudinal homogeneity (the essence of Herrick's columns [194], as also conceived by Kuhlenbeck [195], an already false notion) and thus lacks from its origin a conceptual underpinning for the sort of subdivisions we deal with here (what developmental mechanism would pattern a longitudinal column into 9 subdomains?). What the alternative prosomeric model resolves as easily explainable dorsoventral pattern generated by well-established antagonism between floor-induced ventralization and roof-induced dorsalization is not easily imitated within columnar theory because no rostrocaudal patterning mechanism is admitted there and the floorplate is defined differently. The same theoretic antecedents lead to non-recognition of alar/basal subregions within the hypothalamus (the whole hypothalamus is newly postulated as a basal territory [20]). Another difference with our prosomeric approach is that the LH is conventionally thought to pertain to the whole hypothalamus, while the radial glial structure indicates that it essentially relates to our peduncular part of the hypothalamus, derived specifically from prosomere hp1 ( [17], their Fig. 12; [26]; present results). We conclude that connectivity data need to be reinterpreted or repeated within the prosomeric paradigm.
Neurodegenerative Diseases and Cancer (to L.P.). Infrastructure support provided by the UCSF Medical School, IMIB, and the Universities of Murcia and Castilla-La Mancha is also acknowledged.
Author Contribution Carmen Díaz, Luis Puelles, and John L.R. Rubenstein contributed to the study conception and design. Material preparation, data collection, and analysis were performed by Carmen Díaz, Margaret Martínez-de-la-Torre, and Luis Puelles. The manuscript was written by Carmen Díaz and Luis Puelles, and all authors commented on previous versions of the manuscript. All authors contributed to the article and approved the submitted version.
Funding Open Access funding provided thanks to the CRUE-CSIC agreement with Springer Nature. This work was supported by a Seneca Foundation (Autonomous Community of Murcia) Excellency Research contract, reference: 5672 Fundación Séneca 19904/GERM/15; project name: Genoarchitectonic Brain Development and Applications to Neurodegenerative Diseases and Cancer (to L.P.).
Data Availability Most of our analysis was based on in situ hybridization images downloaded from the Allen Developing Mouse Brain Atlas (https:// devel oping mouse. brain-map. org) and Allen Mouse Brain Atlas (https:// mouse. brain-map. org/). Other data are available from the corresponding author on reasonable request.

Declarations
Ethics approval All experimental procedures, experimental protocols, handling use, and care of mice were conducted in compliance with the current normative standards of the European Community (86/609/EEC), the Spanish Government (Royal Decree, 1201/2005; Law 32/2007) and the approval of University of Murcia Committee for Animal Experimental Ethics.

Conflict of Interest
The authors declare no competing interests.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.