LacZ‐reporter mapping of Dlx5/6 expression and genoarchitectural analysis of the postnatal mouse prethalamus

Abstract We present here a thorough and complete analysis of mouse P0‐P140 prethalamic histogenetic subdivisions and corresponding nuclear derivatives, in the context of local tract landmarks. The study used as fundamental material brains from a transgenic mouse line that expresses LacZ under the control of an intragenic enhancer of Dlx5 and Dlx6 (Dlx5/6‐LacZ). Subtle shadings of LacZ signal, jointly with pan‐DLX immunoreaction, and several other ancillary protein or RNA markers, including Calb2 and Nkx2.2 ISH (for the prethalamic eminence, and derivatives of the rostral zona limitans shell domain, respectively) were mapped across the prethalamus. The resulting model of the prethalamic region postulates tetrapartite rostrocaudal and dorsoventral subdivisions, as well as a tripartite radial stratification, each cell population showing a characteristic molecular profile. Some novel nuclei are proposed, and some instances of potential tangential cell migration were noted.

superseded by the prosomeric model, where these regions jointly with the pretectum are held to be caudorostral components of the diencephalic alar plate (p1-p3 in caudorostral order; Figure 1) on the basis of substantial molecular and experimental evidence supporting a forebrain axis ending in the hypothalamus . The updated prosomeric model defines the diencephalon as a forebrain sector intercalated rostrocaudally between the hypothalamus (divided into hypothalamic prosomeres hp1 and hp2, in caudorostral order) and the midbrain (mesomeres m1 and m2, in rostrocaudal order; Figure 1d). The diencephalon is itself divided into three diencephalic prosomeres (p1-p3 in caudorostral order; note p3 contacts rostrally hp1, whereas p1 contacts caudally m1; m2 contacts the isthmic hindbrain; Figure 1d). All prosomeres display four fundamental longitudinal or dorsoventral zones: the floor, basal, alar and roof plates of His (1893a, 1893b). In the diencephalon, the alar domains of p1-p3 were given simplified novel names: pretectum (p1), thalamus (p2) and prethalamus (p3; Puelles & Rubenstein, 2003), thus evading wrong columnar axial connotations. The classic "dorsal thalamus" and "ventral thalamus" were simply renamed "thalamus" and "prethalamus," respectively. This makes our terminological usage consistent with the wellknown anteroposterior patterning effects of the interposed zona limitans intrathalamica (ZL; Figure 1d), and ties in with accepted analogous semantic usage with the "tectum" (midbrain alar plate) and "pretectum" (alar part of caudal diencephalic prosomere). The columnar term "epithalamus" is not problematic in the change to the prosomeric paradigm, since we conceive this area as lying hyperdorsally within the "thalamus" (i.e., it is restricted in extent to the middle or thalamic diencephalic prosomere, rather than being a general column of the diencephalon). A separate but comparable hyperdorsal domain appears at the top of the "prethalamus" (rostral diencephalic prosomere); this was misidentified classically as "eminentia thalami," a term making reference to its bulge at the back of the interventricular foramen. Its wrong historic ascription to the thalamus was due to initial lack of distinction between "ventral" and "dorsal" parts of thalamus. Hayes, Murray, and Jones (2003) reasonably proposed to rename this as "prethalamic eminence," an option adopted thereafter in the prosomeric model.
We first approach genoarchitectural analysis of the prethalamus by a detailed study of differential expression of members of the family of Dlx genes. Four Dlx homeobox genes (Dlx1,2,5,6) are expressed in the embryonic mouse forebrain, as was described in the early 1990s (see historic account and criticism in Bulfone et al., 1993;Price, Lemaistre, Pischetola, Di Lauro, & Duboule, 1991;Robinson, Wray, & Mahon, 1991;Simeone et al., 1994). There are important differences in the relative timing of their respective expression along the process of neuronal differentiation. In the telencephalic subpallium, Dlx gene and DLX protein expression cellular patterns appear to be largely indistinguishable. Dlx2 expression begins in scattered cells in the ventricular zone (VZ) and is followed by Dlx1; both are then most highly expressed in secondary progenitors in the subventricular zone (SVZ), where Dlx5 and Dlx6 expression is subsequently initiated, co-expressed with Dlx1 and Dlx2 (Eisenstat et al., 1999;Lindtner et al., 2019;Liu, Ghattas, Liu, Chen, & Rubenstein, 1997). Dlx/DLX expression patterns in postmitotic neurons are more diverse. For instance, parvalbumin-positive cortical interneurons distinctly express Dlx2 and Dlx5 but show much lower levels of Dlx1 (Cobos, Broccoli, & Rubenstein, 2005). In the amygdala, Dlx1 and Dlx2 are expressed by the intercalated nuclei, whereas Dlx5 and Dlx6 are expressed in the central nucleus (Wang, Lufkin, & Rubenstein, 2011).
The literature lacks a complete and widely accepted list of mammalian prethalamic derivatives. Some of them seem to have been ascribed to either thalamus or subthalamus (see critique of the latter obsolete concept in Puelles, Martinez-de-la-Torre, Bardet, et al., 2012), and both the transversal prethalamo-hypothalamic boundary and the longitudinal alar-basal boundary separating alar prethalamus from its corresponding tegmental domain are rather vaguely defined. There is indeed considerable uncertainty in the literature about where precisely some populations present in this area originate; for example, zona incerta, reticular nucleus, intergeniculate leaflet, and retropeduncular nucleus (see Section 4). Throwing light on this issue is the primary aim of the present report.
We mainly made use of mice (Mus musculus) transgenically modified to express the reporter bacterial enzyme beta-galactosidase F I G U R E 1 Dorsoventral, rostrocaudal, and radial subdivisions of the mouse prethalamus at embryonic stages. (a-c) Lateral and more medial sagittal sections of an E13.5 mouse brain carrying the Dlx5/6-LacZ construct (blue), plus PAX6 immunoreaction (brown) (a,b), and similar section at E15.5 labeled with Dlx5/6-LacZ and TBR1 immunoreaction (c). The rostral (R) and dorsal (D) spatial directions are indicated in (a). White dash lines indicate the transverse pretecto-thalamic, thalamo-prethalamic and prethalamo-hypothalamic interprosomeric boundaries. The dark blue dash line indicates the alar/basal limit (a/b). The central and subcentral prethalamic regions (PThC, PThSC) are delineated respectively with light blue and yellow dash lines; the main derived primordia are identified (see abbreviation list) (a-c). Note the Dlx5/6-LacZ-negative prethalamic eminence (PThE), lying immediately under the chorioidal roof (ch; green in d,e) expresses PAX6 and TBR1 in its ventricular and mantle zones, respectively. Dlx5/6-LacZ signal is mainly restricted to the underlying PThC and PThSC subdivisions, but appears also at the rostral shell of the zona limitans (ZLR) and its rostral liminar extension (RLi) along the alar-basal boundary (a,b). The Dlx5/6-LacZ signal is relatively weaker at rostral prethalamic levels (a-c). (d) Schema of the prosomeric model illustrating the main neuromeric transverse subdivisions and the dorsoventrally disposed, longitudinal roof, alar, basal and floor plates. The midbrain is subdivided rostrocaudally into mesomeres m1 and m2, while the diencephalon divides caudorostrally into diencephalic prosomeres p1, p2, and p3, and the secondary prosencephalon likewise into hypothalamotelencephalic prosomeres hp1 and hp2. Black dash lines indicate the transverse interneuromeric limits. The longitudinal a/b boundary is represented as a dark-blue dash line. The main subdivisions of the prethalamus are highlighted and their relationship with thalamic and hypothalamic neighbor structures are illustrated, based on the embryonic material in (a-c). The prethalamus (alar plate of p3) is divided dorsoventrally into PThE (sienna background), PThC (light orange), PThSC (yellow) subregions, and we see as well the partly caudal and partly ventral ZLR/RLi complex (strong orange). Note a rostral part of the PThE ventricular surface bulges into the interventricular foramen (ivf; dark gray background; this implies partial evagination of PThE into caudomedial wall of hemisphere, under the chorioidal fissure-not shown). PThC and PThSC are rostrocaudally subdivided into three entities. The PThC contains the reticular/retropeduncular, subgeniculate and pregeniculate radial complexes (RP/Rt, SG, PG) whereas the PThSC contains the preincertal, rostral zona incerta and caudal zona incerta radial complexes (PI, ZIR, ZIC). The prethalamic basal plate, or p3 tegmentum (p3Tg), is colored in light pink. PThE, PG and ZIC contact caudally with the ZLR (strong orange), whereas PThE, RP/Rt and PI bound rostrally with the hypothalamic (alar) paraventricular domain (Pa); instead, the RLi continues rostrally into the rostrally expanding hypothalamic subparaventricular domain (SPa). The thin black dash line in the hypothalamus marks the intrahypothalamic transverse boundary between hp1 and hp2 prosomeres; a similar dash line also marks the thalamo-pretectal interprosomeric limit between thalamus and pretectum and corresponding tegmentum (p2Tg). (e) Schematic color-coded map based on d of some of the gene patterns expressed in the prethalamus. Note Nkx2.2 is primarily expressed only at the ZLR and RLi, overlapping with Dlx5/6-LacZ signal. Nkx2.2 is also expressed at the caudal zona limitans shell area (ZLC) and its caudal liminar extension (CLi; this reaches the isthmic boundary). It is still unclear whether the Nkx2.2-positive forebrain liminar band (RLi and CLi) lies in the alar plate, in the basal plate, or halfway across the alar-basal boundary (the option tentatively illustrated here). Blue arrows indicate documented tangential migratory routes of GABAergic cells, originated in the ZLC, reaching the thalamic lateral geniculate nucleus (LG) and the posterior limitans nucleus in the caudal thalamus (Delogu et al., 2012;Golding et al., 2014). Analogous migrations from ZLC and/or ZLR eventually enter the PThC. (f-h) Three horizontal schemata in dorsoventral order based on sections through PThE (f), PThC (g) and PThSC (h) subregions of an E13.5 embryo. The three dorsoventral prethalamic subdivisions are highlighted using the same color code as in (d) (LacZ) under control of a zebrafish enhancer (Zerucha et al., 2000).
Once inserted into the mouse genome, this enhancer imitates the activity of its mouse homolog, the Dlx5/6 enhancer, driving LacZ expression only in mouse cells which normally express Dlx genes, in a pattern closely resembling that of Dlx5 and Dlx6 (Stühmer, Puelles, et al., 2002;Zerucha et al., 2000). This is consistent with the fact that the homologous mouse Dlx5/6 (I56i) and zebrafish zfdlx4/6 enhancers are ultraconserved, showing very little divergence across vertebrate evolution (Zerucha et al., 2000).
After beta-galactosidase histochemical reaction, the cells expressing (or having once expressed) Dlx5/6-LacZ continue to be labeled by betagalactosidase activity in the cytoplasm, demonstrable as a blue labeling product, which in most places remains visible until adulthood (e.g., in present results up to P140).
We essentially mapped the blue-labeled cells as far as possible during postnatal development (P0-P140), identifying in sagittal, coronal and horizontal section planes the successive changes in their distribution, as well as their eventual relationships with recognizable mature nuclei. We further correlated these mapping results with various other differential markers aiding a tridimensional systematization of prethalamic derivatives (i.e., along dorsoventral, anteroposterior and radial dimensions). In order to check comparatively the derivatives of the Dlx-positive prethalamic nuclei we studied postnatal RNA expression of Calb1, Calb2, Ecel1, Enc1, Islet1, Nkx2.2, Pax6, Six3, and Somatostatin, as well as Calbindin (CB), Neuropeptide Y (NPY), Parvalbumin (PV), and PAX6 proteins.
The Dlx5/6-LacZ reaction was also compared with pan-DLX immunohistochemistry (DLX1,2,5,6 proteins are all reactive to this antibody, if present). We made the surprising observation that differential patterns could be observed in specific sets of prethalamic derivatives (e.g., some nuclei showed only LacZ signal, both LacZ and immunoreaction, or only immunoreaction). After checking corresponding Dlx1/2/5/6 ISH data in the Allen Adult and Developing Mouse Brain Atlas, these findings were interpreted as evidence that different Dlx paralogues predominate in given prethalamic nuclei at postnatal stages, similarly as was found previously in the subpallium (as noted above). Our anatomic analysis and terminology was guided by the most recent, updated version of the prosomeric forebrain model (Puelles, 2013;Puelles, Martinez-de-la-Torre, Bardet, et al., 2012;Puelles, Martinez-de-la-Torre, Ferran, & Watson, 2012;.

| Animals
All experimental procedures with transgenic mice were approved by the Committee on Animal Research at University of California, San Francisco (CA), and mouse colonies maintained in accordance with National Institutes of Health and UCSF guidelines.
All experimental protocols, handling use and care of nontransgenic mice were conducted in compliance with the current normative standards of the European Community (86/609/EEC), the Spanish Government (Royal Decree, 1201Law 32/2007)  were collected from postnatal to adult stages (embryonic specimens were collected as well) (Table 1).

| Transgenic animals
The isolation of zebrafish Dlx5/6 forebrain enhancer elements, construction of the zfdlx5/6-LacZ transgenic vector, and the characterization of the LacZ expression pattern in transgenic mice (strain C57 Bl/6) with respect to that of the endogenous Dlx5 and Dlx6 genes, are described in Zerucha et al. (2000). The propagation of the transgene appears to be stable, as no change in the pattern or intensity of β-galactosidase expression has been noted for more than 20 generations. Heterozygous Dlx5/6-LacZ brain specimens were fixed and processed histochemically (with or without immunochemical counterstains) at stages P0, P4, P15, P40, and P140.

| Preparation of tissue
Most of analyzed brains were from postnatal animals (n = 32) although few embryonic brains (n = 5) were also included in this study. For the preparation of embryonic brain tissue, timed-pregnant dams were killed by cervical dislocation, embryos removed and the brains dissected in cold phosphate-buffered saline (PBS: 137 mM NaCl, 2.7 mM KCl, 1.8 mM KH 2 PO 4 , 5.1 mM Na 2 HPO 4 , pH 7.4). The tissue was fixed in cold 4% paraformaldehyde/PBS for between 30 min (embryonic day [E] 13.5) to 2 hr (E15.5 and later stages). For the preparation of postnatal and adult tissue, animals were anesthetized with 2% chloral hydrate/ PBS, and cardially perfused with between 10 ml (postnatal day [P] 0) and 50 ml (P13 and older) cold 4% paraformaldehyde/PBS. Following dissection, brains were postfixed in the same solution for 3 hr 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
The brains from transgenic animals were parted along the midline to obtain transversal and sagittal sections from the same specimen. The rest of brains were sectioned as a whole. For sectioning, the embedded tissue was oriented to obtain either sagittal, horizontal or transverse sections across the prethalamus, and cut into series of 100 μmthick slices with a vibrating blade microtome (Leica). Sections from Dlx5/6-LacZ transgenic brains were immediately used for the X-gal staining reaction, and some of them processed for in situ hybridization or immunohistochemistry. Sections from nontransgenic brains were processed for in situ hybridization and/or immunohistochemistry.
The tissue was next rinsed in PBS, cleared for 30 min in 50% v/v glycerol in PBS, sequentially mounted on slides.
2.6 | Reverse transcriptase-polymerase chain reaction (RT-PCR) Calb1, Calb2, and Six3 cDNA fragments were obtained by reverse transcription (RT). RNA was individually extracted with Trizol reagent (Invitrogen, Carlsbad, CA, Cat. 10296-028) from freshly dissected brains of Mus musculus. The RNA was treated with DNase I (Invitrogen, Cat. 18068-015) for 15 min at room temperature (RT), and the enzyme was then inactivated at 65 C. Afterwards, RNA samples were converted to single-stranded cDNA with Superscript III reverse transcriptase (Invitrogen, Cat. 18080-044) and oligo-dT-anchored primers. The resulting first-strand cDNA (0.5 μl of the reverse transcription reaction) was used as a template for the PCR reaction, which was performed in presence of Taq polymerase (Promega, Cat. M8305) and the following gene-specific primers for Calb1, Calb2, and Six3 mRNAs. Used primers were: 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 T m 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 (ISH)
The tissues were processed for in situ hybridization with digoxigenin-UTP-labeled antisense riboprobes. Sense and antisense Summary of numbers of transgenic and nontransgenic animals at embryonic and postnatal stages, and markers used in each case
Our immunohistochemical reaction protocol was described in detail elsewhere (Ferran, Ayad, et al., 2015). The primary antibodies used are described in Table 3. After washes, the sections were incubated with biotinylated goat anti-rabbit or goat anti-mouse 2.9 | Interpretation rationale for double Dlx5/ 6-LacZ and ulterior Pan-Distalless immunoreaction (any DLX protein) We studied a sagittally cut Dlx5/6-LacZ P0 brain which was secondarily immunoreacted with the Pan-Distalless antibody (see Table 3). It is usually thought on the basis of published data on the striatum that during development Dlx-expressing cells activate sequentially the four different neural paralog Dlx forms, that is, starting with Dlx1/2 at postmitotic stages and proceeding on to express Dlx5/6 as differentiation advances, frequently with accompanying downregulation of Dlx1/ 2 (Eisenstat et al., 1999). Presence of the corresponding mRNAs and proteins would be expected at the appropriate stages, so that specific ISH reaction and some immunoreaction with the pan-distalless antibody should be a general feature where a Dlx gene is being expressed.
On the other hand, in our double-labeled material we found well delimited only blue (LacZ-positive), only brown (DLX immunoreactive) and mixed blue-brown areas. In the following lines, we explain the interpretive rationale we finally followed with this complex material, which bespeaks of a subtly differential nature of the corresponding progenitor subdomains, consistently with conclusions derived from other molecular markers studied. In this rationale, we considered the possibility that the LacZ reaction may have partially or totally quenched subsequent immunoreaction of the DLX proteins (particularly in cases where their cytoplasmic concentration was not high). We examined as well public Dlx1/2/5/6 gene in situ data available at the Allen Developing Mouse Brain Atlas which show the prethalamic topography of some of these transcripts at perinatal (E18.5-P4) stages. We arrived at the following conclusions.
We identified purely blue-labeled cells as elements that at some earlier stage were Dlx5/6-positive (since they show the Dlx5/ 6-LacZ reaction product), but secondarily significantly downregulated the amounts of both DLX5/6 and DLX1/2 protein present, so that their immunoreaction at stages E18.5-P4 was below detection level, any protein remnants possibly having been quenched by the LacZ histochemical procedure. Indeed, we confirmed at the Allen Developing Mouse Brain Atlas that these "blue  (Stühmer, Anderson, Ekker, & Rubenstein, 2002;Stühmer, Puelles, et al., 2002).

1:400
Nitric oxide synthase (NOS) A recombinant protein consisting of 195 amino acids from the N-terminal of rat nNOS protein (the exact positions are a company secret, according to Invitrogen Technical Service) ZYMED Laboratories (now Invitrogen Immunodetection), Carlsbad, CA; Z-RNN (equals Z-RNN3) and is now Cat#61-7000, AB_2313734; rabbit polyclonal; This antibody reacts with the 160 kDa nNOS protein and does not exhibit any cross-reactivity with the related eNOS or iNOS proteins. During development reactivity was confirmed with a 160 kD band on western blots of rat and mouse brain tissue lysates (20 μg).
Incstar (now Immunostar Inc) Hudson, WI; Cat t# 22940; RRID:AB_572253; rabbit polyclonal; The antibody was raised to NPY coupled to BSA with glutaraldehyde. The glutaraldehyde couples the NPY to the BSA using free amine groups. This strategy was intended to target the amino terminus of NPY, though other conformations were likely present since glutaraldehyde binds any free amines it can find, including those on amino acids such as lysine low Dlx1/2/5 or corresponding proteins).

| RESULTS
For simplicity, we will refer to "Dlx5/6-promoter-driven-LacZ expression" as "LacZ signal" or "Dlx-LacZ signal." Our description focuses on the prethalamus, with minimal attention to neighboring areas such as the prethalamic tegmentum, the interthalamic zona limitans, the alar and basal hypothalamus, or the substantia nigra. After some introductory embryonic data at stages E13.5 and E15.5 (Figures 1,2), we attend successively to stages P0, P4, P15, P40, and P140, offering images of sagittal and horizontal sections (Figures 3-21), with occasional inclusion of some other embryonic images, for specific reasons (E16.5, E18.5). We will emphasize in the text the major changes observed at each time point.

| Dorsoventral regions
Note we use the relatively recent "prethalamus" term (PTh) as representing the alar plate domain of the diencephalic prosomere 3 (the rostralmost diencephalic segmental unit; see Introduction).
This name was originally proposed in substitution of the obsolete columnar term "ventral thalamus" (Puelles & Rubenstein, 2003); it refers explicitly to the topologic position occupied by this rostral diencephalic domain in the prosomeric model ( Figure 1d). Our overall structural conclusion is that the PTh region is tetrapartite both dorsoventrally and rostrocaudally (Figure 1d,f-h). Most dorsally, it includes the Dlx-negative prethalamic eminence (PThE; Puelles and Rubenstein (2003); Hayes et al. (2003)

| Anteroposterior domains
As we illustrated schematically in Figure 1d,e,g,h our results support a shared anteroposterior (AP) tripartition of the PThC and PThSC subregions, while such subdivision is absent at the PThE and RLi domains. The ZLR represents by itself an independent fourth AP prethalamic domain. These AP details will be elaborated in the following sections.

| Radial subdivisions
Each of the progenitor domains identified as anteroposterior divisions of PThC and PThSC show subtle adult structural differences along the radial dimension (i.e., they show stratification along the ventriculopial axis, Figure 1g,h). These allow the overall distinction of periventricular,  these are singularly corroborative of our anteroposterior tripartite PTh interpretation.

A general observation made in newborns and onwards was that
LacZ-positive periventricular forebrain elements, and particularly those of p3, gradually became displaced at some distance away from the ventricular lining; that is, a Dlx-negative deep periventricular stratum became apparent ( Figure 3). This peculiar feature possibly relates to neuropil maturation within the deep periventricular stratum.

| Central prethalamus region (PThC)
In the Dlx-positive PThC region, we distinguish three independent subregions (i.e., molecularly differentially-specified progenitor domains) in caudo-rostral order; they are represented by the well-known pregeniculate, subgeniculate and reticular nuclei (PG, SG, both superficial, and Rt, at intermediate stratum level), respectively, but contain as well additional complementary formations within the radial strata not occupied by these better known nuclei.
In horizontal and sagittal sections through the P0 prethalamus ( Figures 3 and 4), the superficial pregeniculate nucleus (PG) can be recognized by its strong Dlx-LacZ expression and its well-known layered structure; its position is superficial (only covered by the optic tract), and it lies caudally within PThC, just rostral to the ZLR and the thalamus.
The retinorecipient PG formation was classically named "ventral lateral geniculate nucleus" (Nissl, 1889), consistently with assumptions in the now obsolete columnar forebrain model (Herrick, 1910). This name has now decayed in use with the modern axial paradigm change proposed in the prosomeric model (Puelles & Rubenstein, 2003, which underpins the alternative pregeniculate term. The PG is layered, as is best seen in horizontal and coronal sections. We detect four layers parallel to the overlying optic tract. Superficially there is a weakly labeled marginal PG layer (or Layer 1, Figure  Rostrally, PG limits with the moderately LacZ-positive, but similarly superficial and retinorecipient subgeniculate nucleus (SG). Some sources describing retinal projections do not mention this nucleus (Jones, 2007;pp. 1241-1314, but it does appear for instance in Morin and Studholme (2014) Huber & Crosby, 1926;Medina, Trujillo, Diaz, Martin, & Puelles, 1990). We accordingly propose to apply likewise the name tri- Figures 3h-j, 4p and 5h; checked at the Allen mouse brain atlas; not shown). Use of additional markers (see below) nevertheless allows differential definition of the periventricular units corresponding to the three PThC subregions. Note some sources misidentify the whole periventricular stratum of the PThC region as either "zona incerta," or "Forel's field" (Jones, 2007). Actually, all zona incerta parts belong only Lateromedial series of sagittal sections though the prethalamus of an E16 mouse embryo carrying the Dlx5/6-LacZ construct and immunoreacted with anti-PAX6 antibody. The tagged rostrocaudal subdivisions are evident superficially (a,b). Rostral is oriented to the bottom and dorsal to the left. Weakly immunolabelled PAX6 cells are restricted to the Dlx-positive SG/ZIR complexes, where the Dlx5/6-LacZ reaction is weaker than at the PG/ZIC complex (SG, PG, ZIR, ZIC; a,b), as well as to the periventricular stratum deep to the Rt (RtPv; c,d). Dlx-LacZ reaction is very strong at the ZLR/RI sites, rather weak at the dorsal triangular and oval nuclei, and practically absent at the Rt (ZLR, RI, T,Ov, Rt in a,b). Within the neighboring peduncular hypothalamus the dorsal entopeduncular nucleus (EPD) is PAX6-positive and LacZ-negative, whereas the ventral entopeduncular nucleus (EPV) strongly expresses Dlx-LacZ signal (a).
to the underlying PThSC region, and Forel's field was originally defined as a component of the subthalamic/thalamic tegmentum, that is, the prethalamic/thalamic basal plate (Forel, 1877). The primitive tegmental concept of "subthalamus" was later arbitrarily exported into neighboring alar subregions, including the zona incerta, and sometimes to the whole prethalamus (see historic account and criticism in Puelles, Martinez-de-la-Torre, Bardet, et al., 2012).

| Subcentral prethalamus region (PThSC)
The subcentral prethalamic region is likewise subdivided into rostral, middle and caudal subregions, which lie respectively ventral to the corresponding PThC subregions described above, but are distinct from them (Figures 3 and 4). The PThC/PTHSC boundary was not clearly detected in the section planes used in our Dlx5/6-LacZ material, but a cell-poor dividing gap is visible in true transversal sections through p3 (e.g., Figure 7k), and can be visualized as well with some alternative gene markers (see below). The histological image of the PThSC is dominated by the conventional zona incerta (ZI), a structural complex whose location is restricted to intermediate radial levels of this domain. This region is usually divided into "dorsal" and "ventral" parts, due to use of the arbitrary columnar axis. Prosomeric terminology translates this into caudal and rostral zona incerta parts; moreover, we add to the incertal complex a newly identified and rostralmost preincertal nucleus, which lies directly under the Rt (ZIC; ZIR; PI; Figures 3l-o, 4e-n and 5c-g; compare ZI parts in Puelles, 2013; Puelles, Martinez-de-la-Torre, Bardet, et al., 2012;.  Figure 3k-o). The ZICPv surrounds the mamillothalamic tract as the latter approaches its thalamic F I G U R E 3 Dorsoventral series of horizontal sections through the prethalamus of a Dlx5/6-LacZ P0 mouse brain illustrating differences in the signal in radial (ventriculo-pial), dorsoventral and rostrocaudal subdivisions. Nuclear primordia are tagged according to the abbreviations list. The midline lies to the right, caudal is oriented to the top of the panels. Dlx5/6 signal is restricted to the prethalamus, but shows evident downregulation at the intermediate reticular nucleus (Rt). Black dash lines indicate the caudal thalamo-prethalamic and rostral hypothalamo-prethalamic boundaries (note here the rostral relationship with internal capsule and cerebral peduncle; ic, pe). Note the thalamus (Th) is free of Dlx5/6-LacZ-labeling whereas the hypothalamic ventral entopeduncular, accessory entopeduncular and dorsomedial nuclei (EPV, EPA, DMH) display

Periventricular cells lying deep to the intermediate PI nucleus also
F I G U R E 4 Lateromedial series (a-p) of sagittal sections showing Dlx5/6-LacZ reaction and dorsoventral and anteroposterior subdivisions in a P0 mouse brain prethalamus. Black dash lines indicate the caudal thalamoprethalamic and rostral hypothalamo-prethalamic boundaries. Caudal is oriented to the top right, dorsal to the top left (part of telencephalic subpallium is visible for reference). The asterisks in l-p indicate dispersed Dlx cells in the basal p3 and p2 tegmentum (p3Tg; p2Tg). Strong Dlx5/6-LacZ signal is also present in hypothalamic territories, excepting at the alar paraventricular subdomain (Pa), retrotuberal subthalamic nucleus and retromamillary area ( Finally, some groups of LacZ-positive cells appear dispersed in the basal plate of p3 and p2, mostly lateral to the course of the mamillotegmental tract, possibly within the tegmental area originally known as "Forel's field" (p3Tg, p2Tg, F; Figures 3q and 4l-p). We interpret these cells as migrated from the p3 alar plate into p3 tegmentum, with partial subsequent dispersion into the p2 tegmentum. In general, the anteroposterior grouping of differential Dlx5/6-LacZ and IHC reactions seems to extend relatively unchanged ventralwards from the PThC into the PThSC, as we examine below. These patterns support the conclusion that there exists at least at perinatal stages a marked anteroposterior heterogeneity in the regulated co-expression of distinct F I G U R E 5 Lateromedial series (a-h) of sagittal sections through the prethalamus of a P0 mouse brain carrying the Dlx5/6-LacZ construct, which were counterstained with pan-distalless antibody (DLX; this antibody recognizes all DLX forms). Black dash lines indicate the prethalamo-thalamic (intrathalamic) and hypothalamoprethalamic boundaries. Caudal is oriented to the top right, dorsal to the top left. Blue cells represent exclusive Dlx5/6-LacZ signal, brown cells indicate exclusive DLX-immunostaining, and double-labeled cells show both reactions; see our interpretation criteria for this unusual labeling in Section 2. Combined labeling reveals rostrocaudal, dorsoventral and radial prethalamic components nondistinguishable with single Dlx5/6-LacZ labeling (compare with Figure 4). Remarkably, brown DLX-immunoreactive cells are restricted to locations devoid of LacZ signal, mainly the reticular nucleus, Rt (b-g), its periventricular stratum, RtPv (H) and the entire preincertal complex (PIS, PI, PIPv in (b-g)). Superficial and dorsal components of the reticular region, that is, the retropeduncular and triangular nuclei (RP, T), have blue-labeled cells, in contrast with the remaining brown-labeled Rt/RtPv region. The arrow in (a) points to apparent partial retropeduncular cell migration (RPm) into the peduncular hypothalamus. Note also differences of labeling between the superficial subgeniculate and pregeniculate nuclei, blue and double-labeled, respectively, at the central prethalamus (SG, PG, in (a,b)). Similar labeling differences are observable between the rostral and caudal zona incerta components at the subcentral prethalamus (ZIR, ZIC in (c-e)). Note in b-d selective DLX immunoreaction characterizing the superficial subcentral components (  formations. We observed that the double labeling at the PG is particularly evident at its internal, or parvocellular Layer 4 (PG; Figure 5b), whereas more superficial sections through its external or magnocellular layer (Layer 2) tend to a pure blue labeling (PG; Figure 5a). However, sagittal sections are not optimal to resolve this point.  In the next section of Results (and in F I G U R E 6 Expression of Dlx1, Dlx2 and Dlx6 in the prethalamus at perinatal stages. All images were downloaded from the Allen Developing Mouse Brain Atlas. Caudal is oriented to the top right and dorsal to the top left. (a-c) Lateromedial sagittal sections of a P4 mouse brain HIS reacted for Dlx1. Strong labelling is mainly found at the rostral part of the central prethalamus (Rt) and subcentral prethalamus (PI), although Dlx1 is also present in SG, PG and ZIC. (d,e) Two sagittal sections of a brain at P4, ordered from lateral to medial, which show that Dlx2 expression in the prethalamus is rather similar to the relatively stronger Dlx1 labelling. (f-h) Sagittal sections showing Dlx6 ISH expression in E18.5 (f,g) and P4 (h) mouse brains. Section f is lateral to g. Section h represents a level equivalent to g. Dlx6 labelling is mainly restricted to the caudal part of the central prethalamus (PG) and subcentral prethalamus (ZIC)  (b) and periventricular (C) levels, interpreted in the context of correlative ISH data on Dlx forms found at the Allen Developing Mouse Brain Atlas (see Figure 6; resulting color-code at bottom; following the rationale explained in Material and Methods, the conclusion is that only Dlx1/2 and Dlx6 RNA signals persist significantly in the prethalamus at P0, each with characteristic topography, irrespective of the conservation or not of the Dlx5/6-LacZ reaction product). The prethalamus appears enclosed by black linear boundaries. Black dash lines indicate the thalamo/pretectal (Th/PT) and intrahypothalamic (THy/PHy) borders. The longitudinal alar/basal boundary (a/b) is represented as a blue dash line. The rostral (R) and dorsal (D) spatial directions are indicated in a. Blue asterisks and a black arrow in a mark apparently marginally migrated retropeduncular nucleus cells (RPm, RP) found superficial to the cerebral peduncle in the hypothalamus. Yellow asterisks accompanied by a black arrow in b indicate dispersed cells of the perireticular nucleus (PRt) present within the lateral hypothalamus (superficial to paraventricular nucleus within alar PHy), showing a brown immunoreaction pattern interpreted by us -rationale in Material and Methods-as consistent with postnatal-persisting Dlx1 RNA signal, similarly as occurs in the Rt nucleus (see Fig.6). Blue asterisks represented at p3Tg and p2Tg in b,c refer to Dlx5/6-LacZ-labelled elements presumably migrated into the basal plate.

| Comparison of
matter tracts in the brain (e.g., unlabeled in Figure

| Complementary molecular characterization of prethalamic populations
In this section, we will describe the perinatal prethalamic distribution of other informative markers in material counterstained or not with Dlx5/6-LacZ reaction. These data corroborate the prethalamic subdivisions reported above, but also partly highlight some local differences not noticed in our Dlx analysis, or even identify singularly some cell populations not recognized previously. This material is illustrated in Interestingly, while Dlx5/6 expression patterns were practically continuous in the dorsoventral transition from the PThC into the PThSC, the PV expression pattern is strictly restricted to the mentioned parts of the PThC, leaving the incertal complex largely unlabeled, that is, PV negative (ZIR, ZIC; PI; Figure 8d,e,g-i). PV is therefore a differential marker between PThC and PThSC.
F I G U R E 8 Nkx2.2-expressing prethalamic derivatives at embryonic stages and P4. (a-c) Sagittal sections through the early diencephalohypothalamic mantle zone of embryos carrying an Otp-LacZ construct at E12.5, E13.5 and E14.5. The sections were reacted for Nkx2.2 ISH (blue), while Otp-LacZ cells were visualized with an antibody against β-galactosidase (brown). These Figures are modified from Figure 8.26 of Puelles, Martinez-de-la-Torre, Bardet, et al. (2012), where the emphasis was placed on hypothalamic details (Hy; e.g., dorsal versus ventral tangential migration of hypothalamic Nkx2.2-positive cells into VPa and VMH, respectively; compare with corresponding adult data in g; the hypothalamoprethalamic interprosomeric border is marked by a black dash line, typically passing just caudal to Pa, the paraventricular nucleus). Here we focus on Nkx2.2 signal restricted to the transverse rostral and caudal shells of the zona limitans (ZLR, ZLC) at the prethalamus/thalamus border and the longitudinally continuing rostral and caudal liminar bands (RLi, CLi), which follow the alar/basal boundary. Note the Shh-expressing ZLCo (Shh not shown) remains essentially Nkx2.2-negative (a,b). As the dorsal tip of the ZL is approached, the ZLR and ZLC seem to fuse together (a,b). Dorsal (D) and rostral (R) spatial orientations are indicated in a. (d-g) Lateromedial series of sagittal sections through the prethalamus of a postnatal P4 brain hybridized for Nkx2.2 (blue) and immunostained for tyrosine hydroxylase (TH; brown). Note in (d,e) the presence dorsally, in front of the LG, of superficial Nkx2.2-expressing cells in the intergeniculate leaflet (IGL), an apparently fused derivative of ZLR+ZLC, as well as in the pregeniculate nucleus (PG); the latter elements probably have migrated tangentially from the IGL, since they were not seen at early embryonic stages (a-c). More ventrally, the IGL band bifurcates in an acute angle ventralwards; we identified the rostral branch as the ventral part of the ZLR, passing underneath the pregeniculate nucleus (PG) into the RLi (e-g). In contrast, the caudal, slightly more diffuse IGL branch apparently represents a ventral part of ZLC in the form of its derivative, the conventional peripeduncular nucleus (PP), typically found in front of the MG. A black dash line illustrates the transverse rostral border of the prethalamus with the hypothalamus, while a white dash line marks the prethalamo-thalamic border. (h-j) Three conventional horizontal sections through the prethalamus at P4 (dorsoventral order), illustrating derivatives of the Nkx2.2-positive ZLR and ZLC combined with calbindin (CB) immunoreaction (brown). The transition of dorsally placed IGL into the more ventral ZLR and PP Nkx2.2-positive derivatives can be followed relative to the LG and MG nuclei; see also labelled cells migrated into PG (h). The midline lies to the right and caudal is oriented up.  Figures 8b and 9b). The ZI complex is largely Isl1-negative (Figure 9d,e,f-h). Finally, the ZLR complex also express distinctly Isl1 (Figure 9). Accordingly, Isl1 is also a differential marker for PThC and PThSC, though it shows some differences with the PV marker.
Parvalbumin-immunoreaction (brown) in the prethalamus of a transgenic P0 mouse brain carrying the Dlx5/6-LacZ construct (blue). (a-e) Dorsoventral series of horizontal sections through the prethalamus. Caudal is oriented to the top and the midline lies to the right. Note brown parvalbumin (PV) labelling is restricted to the central prethalamus, specifically to its rostral (retropeduncular, reticular, triangular; RP, Rt, T) and middle (subgeniculate nucleus -SG, and its intermediate and periventricular On the basis of its relationships studied in horizontal sections, we tentatively interpret this cell group as a basally displaced derivative of the alar prethalamic incertal periventricular stratum (thus Pax6-positive), which may be conveniently named "nucleus of the mamillothalamic tract" (MTh; Figure 10l,o). Pax6 signal also appears distinctly at the subcentral intermediate stratum, particularly at the rostral zona incerta subdivision F I G U R E 1 0 Isl1 ISH signal in the prethalamus of a P1 mouse brain. (a-e) Dorsoventral series of horizontal sections through the prethalamus downloaded from the Allen Developing Mouse Brain Atlas. Caudal is at the top and medial to right. The Isl1 labelling in the prethalamus is rather similar to PV immunoreaction showed in Figure 9. The Isl1 expression is basically restricted to the central prethalamic subregion, though the ZLR/RLi complex and the PGPv are also labelled. Potential retropeduncular cells migrated subpially to the peduncular paraventricular hypothalamus are identified (RPm in c,d). (f-g) Sagittal schemata summarizing the Isl1 expression (blue) in the prethalamus as found at superficial (f), intermediate (g) and periventricular (h) strata. The rostral (R) and dorsal (D) spatial directions are indicated in h F I G U R E 1 1 Pax6 expression in the prethalamus at perinatal stages. All microphotographs were downloaded from the Allen Developing Mouse Brain Atlas. (a-c) Three lateromedial sagittal sections of a mouse brain at E18.5. Superficially Pax6 signal is restricted to SG/ZIRS (a) and outer part of ZIR (not shown here; compare i,j), and a line of patches along the ZLRC (a-c). More medially there is little Pax6 signal at the inner part of ZIR and ZIC complex (b; compare i,j). Periventricularly there appears a densely Pax6-positive calyx-shaped cell mass which seems to include all rostrocaudal central parts; we labelled it here as the 'central periventricular stratum' (CPv; C). Underneath it there appears a sparsely labelled 'subcentral periventricular stratum' above the p3 tegmental field (SCPv; c). (d-l) Dorsoventral series of horizontal sections through the prethalamus of a P4 mouse brain. Caudal is at the top and medial to right. The Pax6 labelling is mainly restricted to the middle stratum components of the central and subcentral prethalamic subregions, i.e. the subgeniculate complex (SG, SGI, CPv) and the ventrally located rostral zona incerta (ZIRS, ZIR, SCPv). Note the Pax6 expression extends to the whole periventricular stratum of the prethalamus, as occurs at early embryonic stages, with the difference that central periventricular elements are fused into a compact CPv locus which reaches the ventricle (CPv; e-h), whereas the subcentral stratum components jointly form a disaggregated population found at some distance from the ventricle (SCPv; i-k; compare according to data available at the Allen Developing Mouse Brain Atlas. F I G U R E 1 3 Expression of Sst and Enc1 in the prethalamus at perinatal stages. (a-f) Lateromedial series of sagittal sections of a P4 mouse brain ISH reacted for Sst, downloaded from the Allen Developing Mouse Brain Atlas. Reference axes are indicated in k. Sst expression in mainly found at the rostral part of the central prethalamus (RP, Rt, T) and subcentral prethalamus (PIS, PI), but there are also Sst-positive cells at the magnocellular stratum of the PG nucleus (PGmc; a), the SGL caudal lamina (b), the oval nucleus (Ov; c-e), and the ZLR/RLi complex (b-e). (g,h) A sagittal section (g) and a horizontal section (h) showing Enc1 ISH expression (dark blue) combined with immunoreaction against calbindin (brown, CB) in an E18.5 embryonic mouse brain. Note strong and discrete Enc1 labelling at the subgeniculate nucleus, the internal cell layer of the PG, and the intermediate/periventricular PG strata of PThC (SG, PGi, PGI; g,h). The Enc1 labelling also extends ventralwards into the rostral and caudal zona incerta (ZIR, ZIC; not shown). The perireticular nucleus, neighboring the unlabelled reticular nucleus, is moderately Enc1-positive (h). The thalamus (Th) is generally strongly Enc1-positive (g,h). (i-k) Schematic sagittal representation of the Sst and Enc1 in situ hybridization patterns in the prethalamus superficial (i), intermediate (j) and periventricular (k) strata. Sst labelled areas in violet, Enc1 areas in blue, and double-labelled areas in violet with blue circles or blue with violet circles (i). The rostral (R) and dorsal (D) spatial directions are indicated in k F I G U R E 1 4 Calbindin (CB) immunoreaction in the prethalamus at postnatal stages. (a-f) Assorted sagittal sections of Dlx5/6-LacZ-reacted and CB counterstained (IMR) mouse brains at indicated postnatal stages, ordered from medial to lateral. Spatial directions are given below at l. At periventricular level there appears CB reaction restricted to the SGPv and ZIRPv elements, leaving RtPv unlabelled behind the strongly positive hypothalamic PPa nucleus (a). Some fibre packets such as the retroflexus, mamillothalamic, mamillotegmental and ansa lenticularis tracts are CB-positive (rf, mth, mtg, al). The mamillothalamic tract crosses the ZIC and ZLR at the level of the retroincertal nucleus (ZIC/ZLR, RI) to reach the CB-positive anterior thalamic complex (b,c). Dispersed blue Dlx5/6-LacZ cells are found in the thalamic tegmental area, caudally to the mth tract (arrows; b,c). The Rt shows a CB-positive neuropile (d). The Dlx5/6-LacZ-positive EPV and EPa nuclei appear interstitial to the descending cerebral peduncle, separated by the negative STh nucleus (e,f); note as well thin LacZ-negative limits separating ZIR, ZIC and RI (e), and Dlx5/6-LacZpositive lateral part of SNR next to SNL (e,f). The caudal thalamic zona limitans (ZLC) shell component of the superficial IGL is CB-positive, whereas the ZLR moiety is only Dlx-LacZ-positive (f). (g) Horizontal section of a Dlx5/6-LacZ P8 mouse brain combined with CB-immunostaining. The framed area is magnified in g´, showing CB-immunopositive cells spread in layer 2 and 3 of the pregeniculate nucleus (PG); CB cells aggregated at the Dlx5/6-LacZ-negative part of the intergeniculate leaflet are distinguishable (IGL, g, g´). (h, h´) Horizontal section at P6 with ISH expression of the Calb1 gene (calbindin) counterstained with calretinin (CR) immunoreaction, and magnified detail through framed PG area in h'; note enhanced labelling of PG strata 1-4 next to IGL.

| Calbindin IR
It is interesting to examine sagittal sections through the prethalamus in P0, P1, P3 and P8 mice, which carry the Dlx5/6-LacZ construct and are also immunoreacted against calbindin (CB) (Figure 13). CB immunoreaction labels landmark tracts such as the mamillothalamic tract, the ansa lenticularis and the stria terminalis in the prethalamic neighborhood (but not the stria medullaris, which is calretinin-positive) LacZ material immunoreacted for NPY showed an IGL/PG pattern that was very similar to that of CB (Figure 13i,j).
The CB-positive mamillothalamic tract contains collaterals from the longitudinal basal mamillotegmental tract, which target the F I G U R E 1 6 In situ hybridization expression of Calb2 (calretinin; a PThE marker) combined with immunoreaction against NOS, TH and CB (calbindin) in horizontal alternated series of a P40 mouse brain. Caudal is at the top and the midline is oriented to the right. Most thalamic nuclei and the habenula are Calb2-positive, with exceptions (not considered here). In the prethalamus, the Calb2 signal (dark blue) is restricted to the prethalamic eminence (PThE; b-h) and its prethalamic derivatives, the bed nucleus of the stria medullaris (BSM; h,i) and the paraxiphoid nucleus (PaXi; l,m), and potentially migrated cell populations such as the bed nucleus of the anterior commissure (BAC; h-j) and triangular septal nucleus (TS; a-e). Note also the relation of these PThE Calb2-expressing elements with CB-positive tracts such as the fimbria, stria terminalis and stria medullaris (fi, st sm). Cells of the reticular nucleus are also CB-positive (brown in f and i), differentially with regard to the caudally adjacent SGI/PGI nucleus. Note also migrated Calb2-positive cells within the PG nucleus and the IGL (g-m) anterior thalamic complex (Skidmore et al., 2012;Szabó et al., 2011;Valverde et al., 2000;mth, mtg (same figures). Surprisingly, the Ov nucleus, which we expected to appear positive dorsally to PG was Ecel1-negative, perhaps implying it should be ascribed instead to the middle PThC sector (e.g., Ov in Figure 14c,d). It was unclear in the examined material whether the ZLR/RLi complex shows any Ecel1 labeling, a pattern we have assumed means there probably is no such labeling.

| Adult prethalamic cell populations at P40 and P140
In order to provide a full treatment of all prethalamic regions at these postnatal stages we will successively describe first the mature PThE on the basis of Calb2 (calretinin) ISH at P40 (this being a selective At P15 and P40, that is, before the mating age starts (Dutta & Sengupta, 2016), the expression of Dlx5/6-LacZ in the prethalamic diencephalon (PTh) is still largely comparable to the postnatal pattern, though some changes prelude the adult appearance studied here at P140. For simplicity, only P40 and P140 data are shown (compare Figures 17 and 18 with Figures 19 and 20). The LacZ signal is still well visible at these stages, though it is somewhat paler than earlier.
The standard derivatives of the original LacZ-positive prethalamic domain are still clearly identifiable, though they are increasingly stretched and bent dorsoventrally and flattened anteroposteriorly, no doubt due to forces created mainly by the growing thalamus. Accordingly, some prethalamic boundaries, particularly those of the AP subregions, become less distinct. The distance between the Dlx-positive "periventricular" PTh formations and the unlabeled diencephalic ependyma increases slightly during this period, so that most blue cells are found external to (i.e., more superficial than) the periventricular limiting plane defined by transverse landmark fiber tracts, such as the retroflex, mamillothalamic and fornix tracts (rf, mth, f; Figures 17   and 19).  (Figures 17   and 19). This seems a result of the disproportionate growth and varying rostral protrusion of diverse nuclear parts of the thalamus (Puelles, Martinez-de-la-Torre, Ferran, et al., 2012). Moreover, at these stages, unstained myelinated landmark tracts become increasingly visible as dark gray masses in our material, due to light diffraction.

| Prethalamic eminence (PThE)
We had not given previously any details about the PThE because of the absolute lack of Dlx expression, LacZ reaction and pan-distalless immunoreaction at this locus. This hyperdorsal regional constituent of the prethalamus remains negative for Dlx5/6-LacZ at P40 and P140, as it was before (PThE; Figures 16a, 17j-l, 19a and 20i-l). Recognition of this apparently progressively reduced domain in the adult rodent brain is classically difficult, leading often to its misidentification as either a part of the "bed nucleus striae terminalis complex," or, more often, as "reticular nucleus." For instance, the PThE is not identified as either "thalamic eminence" (the misleading classic name) or the newly orthodox "prethalamic eminence" in the mouse brain atlases of Hof, Young, Bloom, Belichenko, and Celio (2000), Watson and Paxinos (2010), or Paxinos and Franklin (2013). These authors nevertheless do recognize in this area the stria medullaris tract (sm), known to course longitudinally through the PThE area before it enters the habenula (Puelles, Martinez-de-la-Torre, Bardet, et al., 2012; fig. 8.34B). Next to the sm tract a "bed nucleus of the stria medullaris" (BSM) is sometimes identified in rodent brain atlases. This nucleus may well be the chief homotopic derivative of the embryonic PThE.
However, there is evidence that some other PThE derivatives result ectopically placed, due to their embryonic tangential migration out of the PThE primordium in rostralward direction, that is, into F I G U R E 1 7 Dorsoventral horizontal section series through the prethalamus of a Dlx5/6-LacZ P40 mouse brain. Midline is to the right and caudal is oriented up. (a-l) The Dlx5/6-LacZ-positive prethalamus appears compressed between the Dlx-negative thalamus and partially Dlxpositive hypothalamus, adopting a characteristic sigmoid form. Note the difference between the small and unlayered oval nucleus (Ov; (b-f)) and the larger and layered pregeniculate nucleus (PG; (g-k)). The Dlx5/6-LacZ-positive periventricular stratum of all prethalamic formations lies farther apart from the ependyma when compared with the P0 stage (see Fig. 2). Myelinated landmark tracts are identifiable as unstained dark gray masses. (m-q) The ventralmost sections show a packet of Dlx5/6-LacZ-labelled fibers of uncertain origin at the lateralmost part of the cerebral peduncle; these fibers seem to end along the lateral part of the substantia nigra (SNL; (n-q)). The blue hypothalamic cells surrounding rostrally and partly caudally the mamillary body correspond in position and number to described histaminergic tuberomamillary neurons (TM in m,n; Puelles, Martinez-de-la-Torre, Bardet, et al., 2012) telencephalic and hypothalamic loci (Alonso, Trujillo, & Puelles, 2020;Watanabe, Irie, Hanashima, Takebayashi, & Sato, 2018).
At embryonic stages, the marker calretinin appears to characterize rather selectively the PThE mantle (Abbott & Jacobowitz, 1999), in contrast with the neighboring intratelencephalic BST complex, whose cells instead express strongly calbindin (Jacobowitz & Abbott, 1997;Medina & Abellán, 2012; fig. 7.10G). The underlying prethalamic Rt shows no calretinin and little calbindin staining (Jacobowitz & Abbott, 1997;Puelles, Sánchez, Spreafico, & Fairén, 1992). In order to delineate potentially dispersed PThE remnants, we examined a P40 mouse brain which was ISH reacted for calretinin (Calb2) in floating horizontal sections (cut parallel to the optic tract), of which alternate sections were counterstained by immunoreaction for CB, NOS or TH. This material is illustrated in Figure 16.
Eminential calretinin ISH signal was first observed at a dorsal horizontal level passing just underneath the hippocampal fimbria, where this tract arches medialwards into the hippocampal commissure (fi, PThE; Figure 15a-c). The PThE is represented by one or two elongated patches of strong calretinin signal found within supracapsular (=above the internal capsule) prethalamic territory, and found rostrolateral to the clearly delimited habenula and thalamus regions; the latter includes the anterior periventricular, parataenial, anterior and laterodorsal thalamic nuclei (PThE, Hb, PVA, PT, AD, AV, LD; Figure 15a-e). As the horizontal sections penetrate more deeply into the prethalamus, the dense eminential calretinin ISH signal becomes restricted to the apparent frontal and caudal ends of the supracapsular prethalamic territory (in our interpretation, these positive ends represent morphogenetically deformed medial/deep versus lateral/superficial eminential parts). The

F I G U R E 1 8
Lateromedial sagittal section series of a P40 mouse brain showing characteristic Dlx5/6-LacZ labelling in the prethalamus, excepting the prethalamic eminence (PThE). (a-l) The upper (dorsal) PThC portion including the reticular nucleus (Rt) appears flattened and deformed due to rostral protrusion of thalamic elements (Th), whereas the lower (ventral) PThC plus the PTHSC portion forms a thicker complex apparently lying under the thalamus, though its topologic position continues to be prethalamic (see dash lines indicating the interprosomeric boundaries). The series ends at the significantly rostrocaudally compressed periventricular stratum of the prethalamus (i-l). Some Dlx5/6-LacZpositive cells appear dispersed in patches within the p3 and the p2 tegmentum, medially to the substantia nigra pars reticulata (asterisks in h-l). Dlx signal is also variously distinguished in the secondary prosencephalon (hypothalamus and subpallium), contrasting with the wholly unlabelled thalamus and pretectum (Th, PT). Unlabelled myelinated tracts are distinguishable as dark gray masses. Note the passage of the stria medullaris tract characterizes the unlabelled PThE region (sm, PThE; (i-l)) separation of extreme deep and superficial PThE parts leaves an intermediate space occupied by the underlying prethalamic triangular nucleus, which caps the reticular nucleus; the newly recognized oval nucleus soon appears next to the superficial (laterocaudal) tip of the Dorsoventral series of horizontal sections through the prethalamus of a P140 mouse brain. Unstained myelinated tracts identifiable in dark gray are useful as landmarks. (a-m) The sections show Dlx5/6-LacZ labelling in the prethalamus and hypothalamus, as well as the unlabelled thalamus (Th). Note small accessory entopeduncular nucleus (EPA; (l)), which shows LacZ signal, similarly as the ventral entopeduncular nucleus (EPV in G,H), contrasting with the Dlx5/6-LacZ-negative dorsal entopeduncular nucleus (EPD in F). (l-q) Ventral sections illustrating Dlx5/6-LacZ-positive fibres at the dorsolateral part of the cerebral peduncle, which seem to end next to the lateral substantia nigra (SNL; (n-q))

| Central prethalamic region (PThC)
In dorsal horizontal sections passing just under the fimbria and the PThE proper, the oval nucleus and the triangular nucleus appear lateral to the stria medullaris at the point where this tract passes from the PThE into the habenula. The Ov and T cover here the dorsal ends of the Rt, SG and PG nuclei (fi, PThE, Ov, T, sm, Hb; Figures 16a-f, 18a-g and 22a).
The position of this dorsal central prethalamic complex found under F I G U R E 2 0 (a-l) Lateromedial series of sagittal sections through the prethalamus of a P140 mouse brain showing Dlx5/6-LacZ labelling in blue and myelinated tracts in dark gray. The transversal interthalamic boundary is marked by a dash line. The lateralmost sections are tangent to the PG-SG-RP complex, which partly caps laterally the underlying unlabelled Th, due to overall morphogenetic deformation of the diencephalon (a-c). From there the series expands both into the dorsal, compressed and deformed upper part of the PThC region (Rt, T, Ov, PGI/SGI; (d-h)), and the thick transition between PG/SG and the incertal complex (PI, ZIR, ZIC; (e-h)). Note the outer mantle zone of the Dlx5/6-LacZ-positive prethalamus is compressed between the thalamus (Th) and the peduncular hypothalamus (PPa), the latter traversed dorsoventrally by the cerebral peduncle, before bending caudalwards past the hypothalamic subthalamic nucleus along the diencephalic substantia nigra (pe, STh, SNR; (b-g)). The incertal intermediate stratum of the subcentral prethalamus is distinctly wider than the correlative deeper periventricular prethalamic stratum (PIPv, ZiPv, RtPv, PGPv/SGPv; (i-l)); the latter bounds rostrally with the peduncular paraventricular hypothalamus (PPa; (i-l)) the fimbria can be verified in sagittal sections (fi, Ov, T, Rt, SG, PG;  16f,h and 18g,i). Indeed, the pregeniculate nucleus, found more ventrally, rostral to the thalamic geniculate nucleus and the intergeniculate leaflet, clearly shows its definitive superficial four-layered structure, with cells predominating in the denser magnocellular (external o Layer 2) and parvocellular (internal or Layer 4) layers, which are separated by the cell-poor Layer 3 (PG2-4, LG,

F I G U R E 2 1
Color-coded schemata summarizing dorsoventral, rostrocaudal y radial subdivisions of the mouse prethalamus in the context of its nearest hypothalamic and thalamic neighbours. The superficial (a), intermediate (b) and periventricular (c) strata are illustrated separately to show all prethalamic subdivisions and nuclei. The rostral (R) and dorsal (D) spatial directions are indicated in c. Interprosomeric limits are marked as thin black dash-lines. The longitudinal alar/basal boundary (a/b) is indicated as a thick blue dash line; note its relationship with the pink band where the Nkx2.2 gene is expressed throughout the forebrain, as well as with the associated zona limitans organizer (ZL core and ZLR/ZLC shell portions). The Dlx5/6-LacZ-negative prethalamic eminence (PThE) forms the dorsalmost prethalamic subdivision. The underlying Dlx5/6-LacZpositive prethalamus is subdivided dorsoventrally into central and subcentral subregions (PThC, PThSC), and further includes an alar component of the rostral liminar band (RLi), which co-expresses Dlx5/6-LacZ and Nkx2.2 (in red). Leaving aside the ZLR (in red, continuous ventrally with the RLi), the PThC is rostrocaudally subdivided in rostral, middle and caudal progenitor areas, all of them with radially stratified derivatives; the main rostral derivative is the intermediate reticular nucleus (Rt), whereas the superficial subgeniculate and pregeniculate nuclei characterize particularly the middle and caudal PThC areas (SG, PG). PThSC shows also a tripartite division in preincertal, and rostral/caudal zona incerta subregions (PI, ZIR, ZIC). The most caudal prethalamic subregion corresponds to the rostral Nkx2.2-expressing shell of the zona limitans (ZLRC, ZLRI, ZLRPv; in red). Black arrows and color-coded asterisks in a-c indicate apparent tangential migrations of Dlx-positive cells into either the alar peduncular hypothalamus or the p2 and p3 tegmentum (basal plate). Figures 16g-k and 18i-k). The PG Layer 1, or marginal layer, represents the retinorecipient neuropile, which only has scattered blue cells; authors often include this layer in the magnocellular outer Layer 2 (PG1; Figures 16g-k and 18j,k). The PG as a whole is moderately labeled at P40, contrasting with denser LacZ signal at the unlayered SG (SG,PG;. This pattern is just the opposite of what was observed at earlier stages examined. The intermediate stratum of the PThC region is considerably compressed between the thalamus and the hypothalamus. This stratum is mainly represented by the Rt, which forms a thick convex cap around the rostrolateral part of the thalamus, while SGI and PGI are apparently reduced to intercalated thin-perhaps discontinuous-rows of cells difficult to distinguish from Rt in this stretched material (Rt,SGI,PGI;20b and 22b). In the periventricular PThC, LacZ signal is stronger in the ovoid SGPv stratum than in the RtPV and PGPv, this being best observed in horizontal sections (RtPv,SGPv,PGPv;). This distinct labeling reminds of previous observations in Dlx-LacZ sections counterstained with PV and Isl1 labeling at earlier postnatal stages (PV, Figure 8; Isl1, Figure 9).

| Subcentral prethalamic region (PThSC)
The preincertal subcentral formation (PIPv; PI; PIS), jointly with the rostral and caudal parts of the zona incerta (ZIRPv; ZIR; ZIRS and ZICPv; ZIC; ZICS), represent additional radial prethalamic complexes which form a separate rostrocaudal series at the ventral end of the prethalamic alar domain, under the central prethalamus 19dk,. In sagittal sections, the whole incertal region seems to lie topographically caudal to the PThC, and ventral to the thalamic mass.
As a whole, these subcentral formations appear limited rostrally and ventrally by the peduncle, before the latter turns from its dorsoventral peduncular hypothalamic course (rostral to PThE, PThC and PThSC) into its longitudinal course within the basal diencephalic tegmentum (Figure 1d, e); the basal hypothalamic subthalamic nucleus (STh) appears deep to the peduncle precisely at its topologic knee, a point which tends to be disregarded in the literature (Puelles, Martinez-de-la-Torre, Bardet, et al., 2012;Puelles & Rubenstein, 2003. The STh accordingly is also a rostral neighbor of the incertal complex (STh;17dg,20b and 22c). Caudally the subcentral prethalamic region contacts the ZLR-derived, LacZ-and Nkx2.2-positive retroincertal nucleus (RI;Figures 16j,k,18l,m,20b and 22d,e). The latter lies close to the inflection of the medial lemniscus, where it departs from its prior longitudinal tegmental course to enter ventrodorsally across the CLi domain the ventrobasal thalamic mass, just caudally to the zona limitans (ZIR,ZIC,RI,ml,ZL,;20b and 22d,e).  19e-k, 20b,c and 22c,d). The weakly LacZ-positive preincertal F I G U R E 2 2 Color-coded schemata of a dorsoventral series of horizontal sections through the prethalamus in the adult mouse. The schemata are based on a Gad67 ISH brain series, whose labelling is mainly restricted to the prethalamus in the area of interest. Dorsoventral, rostrocaudal y radial subdivisions of the central prethalamus (PThC) and subcentral prethalamus (PThSC) are highlighted in the context of the hypothalamic and thalamic neighbours. The color code is the same as in Figure 21. The midline lies to the right; caudal is oriented to the top. Interprosomeric limits are marked as thin black dash-lines nucleus (PI) lies in front of the ZIR and ventral to the reticular complex (ZIR;PI;Rt;. The subcentral superficial PI nucleus (PIS), located at the caudal edge of the peduncle (under the central retropeduncular nucleus, RP) has a particularly weak LacZ signal (PIS,17b,c,18j,k,19d,20a and 22d,e). The ZIRS, similarly as the ZIR, is compact and rather strongly labeled (ZIRS ; Figures 16l, 17b,c, 18l and 19d), whereas ZICS, like the ZIC, is relatively less compact and shows a moderate LacZ signal (ZICS: Figure 18l,m). The respective periventricular subcentral strata show in general moderate LacZ signal, possibly somewhat stronger at the ZICPv, which appears fragmented by the passage of the mamillothalamic tract (PIPv;ZIRPv;ZICPv;mth;MTh;20c and 22c,d).
In addition, scattered LacZ-positive neurons are still observed within the p3 tegmentum, dorsally to the substantia nigra and the ventral tegmental area. This topography corresponds to the classic tegmental fields H1 and H2 of Forel ("H" stands for "Haube," "tegmentum" in German) Figures 16l, 17k and 18k,l;Forel, 1877). Some dispersed LacZ cells are also found in the p2 tegmental region (Figure 17h-l).

| Zona limitans, rostral shell of zona limitans (ZLR) and RLi
The superficial small patch of intensely LacZ-positive neurons ascribed to the zona limitans rostral shell (ZLR) which was previously constantly identified as ZLRC in the neighborhood of the pregeniculate nucleus (PG) becomes less distinct at late postnatal stages. Some of its cells apparently persist within the adult intergeniculate leaflet (IGL), which we accordingly believe contains fused ZLR (prethalamic) and ZLC (thalamic) derivatives, consistently with our Nkx2.2 mappings described above (Figure 7). Part of these cells seem to incorporate after a short range tangential migration into the PG nucleus, possibly representing as well the local dispersed calbindin-positive cell population (Figure 15f,i).
CB-positive cells were found at earlier stages within the IGL and PG (IGL, PG; Figure 13g,g 0 ,h,h 0 ). Sections passing through the subcentral zona incerta show a larger and deeper ventromedial derivative of the ZLR, the retroincertal nucleus (RI), which expands in front of the thalamic entry locus of the medial lemniscus, separated from the ZIC by a cellpoor gap (RI,ZIC;ml;Figures 16j,k,18l,m,e). The molecularly and histologically distinct retroincertal nucleus appears to have been lumped classically with the caudal zona incerta (the conventional "dorsal zona incerta"). It apparently also has been often misinterpreted as either the "peripeduncular nucleus" (a derivative of ZLC, as shown above with Nkx2.2 ISH), or the "lateral terminal nucleus" of the basal optic pathway, both identified in rodent atlases just rostroventrally to the medial geniculate body (see Section 4). A packet of longitudinally descending Dlx-LacZ-positive fibers, which apparently originates from the RI and courses caudalwards close to the alar-basal boundary, was visible already at early embryonic stages in sagittal sections. These fibers seem to terminate in a weakly LacZ-positive neuropile within the so-called "lateral part" of the substantia nigra compacta (Figures 16m-q and 18n-q).
The prethalamic diencephalic territory has traditionally attracted less scientific attention than its thalamic and hypothalamic neighbors, leading at best to fragmentary anatomical and functional knowledge (concentrated on the reticular nucleus, zona incerta and some superficial visual nuclei), and at worst to considerable confusion about prethalamic boundaries and inner partitions, particularly when it was confused with the supposedly tegmental "subthalamus" (Forel, 1877). This latter trend caused erroneous ascription of some strictly prethalamic parts, when not the whole territory, to either the hypothalamus or the thalamus (see comments about the obsolete and confusing "subthalamus" concept in Puelles, Martinez-de-la-Torre, Bardet, et al., 2012).
We felt that a recapitulative study of prethalamic structure performed with genoarchitectural markers might establish on a stronger basis a precise neuromeric model for this forebrain territory, hopefully, illuminating some misconceptions arisen in the wake of the now obsolete columnar approach (Puelles, 2018;Puelles, 2019;. Our present genoarchitectural study examines in detail postnatal and adult structure of the mouse prethalamus, starting by an analysis of subtle differences in Dlx5/6-LacZ expression, and the surprisingly very informative correlation of such data with Dlx1/2/5/6 immunochemical and in situ signals. We compared next the Dlx pattern with a handful of other prethalamic gene markers chosen because of their differential expression pattern within the area of interest. Irrespective that this report focuses on postnatal structure for the sake of manageability and general usefulness, our interpretations have been constantly informed by correlative prenatal developmental data at our disposal (our own collection of preparates, as well as the Allen Developing Mouse Brain Atlas, which we frequently consulted). In this report we used developmental material only at a few selected points of interest, for clarifying purposes.
Our results basically corroborate our earlier neuromeric studies of this forebrain region (Bulfone et al., 1993;Rubenstein, Martinez, Shimamura, & Puelles, 1994), but now we suggest an expanded genoarchitectural parcellation of the mouse prethalamus, contemplated in a model with partial intersection of four rostrocaudal and four dorsoventral domains (Figures 1, 6, 20 and 22).
Due to the fact that three primary histogenetic compartments, namely PThE, RLi and ZLR, seem homogeneous (undivided), we distinguish on the whole 9 distinct progenitor domains in the prethalamus (3 central PTh + 3 subcentral PTh domains + PThE + RLi + ZLR). We did not explore the underlying prethalamic tegmentum. The central PTh is the largest prethalamic region and contains the best known large prethalamic structures, such as the reticular nucleus and the visual pregeniculate and subgeniculate nuclei. The subcentral PTh largely corresponds to the classic zona incerta, to which we added a third rostral component. Our study also underlines a prethalamic radial structure, previously scarcely visualized, which allows distinguishing between periventricular, intermediate and superficial strata, particularly within PThC and PThSC.

| Boundaries and morphogenetic deformation of the mouse prethalamus
The position of the prethalamus within the prosomeric model encompasses the whole alar plate of prosomere 3 (Figure 1; Puelles, 2018;. As occurs with any welldelimited, tridimensionally developing neuroepithelial sector, the adult prethalamus (PTh) represents a deformed cuboidal portion of the rostrodorsolateral diencephalic brain wall. The PTh displays free ventricular and pial surfaces which are topologically parallel to each other, irrespective of their respective final positions, much deformed during morphogenesis at the interface between the telencephalon and the thalamus. Classical neuroanatomy tended to disregard these inner and outer PTh surfaces, wrongly ascribing them either to the telencephalon or to the thalamus (figs. 10 and 11 in Puelles, 2019;and fig. 9 in Puelles, Martínez-Marin, et al., 2019, aim to explain the sizeable morphogenetic deformations that cause this fundamental error, particularly in the human brain; such deformations are less marked in the mouse). We will explain below a minimum of needed facts about these free surfaces. There are in addition four other surfaces of the same cuboid; these limit with neighboring neuroepithelial domains in the remaining four directions of topological space (dorsal, ventral, rostral, and caudal).
The ventricular and pial PTh surfaces are strictly parallel to each other only at early neural tube stages, characterized by a simple neuroepithelial structure of the PTh (alar p3). During subsequent morphogenesis, the ventricular surface is importantly compressed rostrocaudally between the bulging thalamic mass at the back, and the peduncular hypothalamus (containing the massive peduncle) plus the associated telencephalon at the front (Figure 1d). Progressive differential growth of the adjacent hemisphere (enlargement of basal ganglia, plus formation of occipital and temporal poles) and correlative expansion of the thalamus and thalamo-telencephalic connections (the latter necessarily have to navigate through the interposed PTh into the hemispheric stalk) jointly cause the prethalamic pial surface to become diverted caudalwards together with the thalamic and pretectal pial surfaces. In the human brain all of these lateral diencephalic regions finish facing the rostral midbrain under the pulvinar (Hochstetter, 1919); however, in the mouse the thalamic "lateral" and "medial" geniculate bodies still bulge laterally as rostrodorsal and rostroventral thalamic elements, respectively, even if hidden by the covering hemisphere; the pial face of the PTh is to be found immediately rostral to these landmarks. Accompanying this partial deformation, the superficial prethalamic derivatives partly embrace the alar hypothalamic sector of the cerebral peduncle (see fig. 8.12 in Puelles, Martinez-de-la-Torre, Bardet, et al., 2012) thus allowing space for the flattened optic tract, which always courses caudalwards along the marginal alar PTh in its approach to the thalamus and pretectum (Puelles, Martinez-de-la-Torre, Bardet, et al., 2012;fig. 8.34B;Wagner, McCaffery, & Dräger, 2000;Yonehara et al., 2009). 1 In all adult mammals, the much reduced ventricular surface of PTh still separates the thalamus from the hypothalamo-telencephalic complex; the ventricular PTh remnant lies topographically at or next to the prominent caudal limit of the interventricular foramen, where both PThE and PThC participate. Actually, depending on the species, a varying rostral part of PThE evaginates during development into the caudomedial wall of the hemisphere, where the true rostral boundary of PThE opposes intraventricularly the subpallial stria terminalis complex across the sulcus terminalis (Alonso et al., 2020;Puelles, 2019). This complicating aspect is not represented in Figure 1d Figure 1d). This finding implies that classic authors ascribed the cryptic prethalamic roof area to the telencephalon, since they generally believed the velum transversum to be a diencephalo-telencephalic limiting landmark (Kuhlenbeck, 1973). The longitudinal boundary between the prethalamic roof and alar plates is visualized as a "taenia," where the chorioidal tela attaches to the upper rim of the PThE. A longitudinal tract or a commissure may be associated to a taenia. In fact, the stria medullaris tract courses subpially through PThE next to the cited alar-roof limit (see fig. 8.34B in Puelles, Martinez-de-la-Torre, Bardet, et al., 2012); in some older works the stria medullaris itself was known as the "taenia tract" (Déjerine, 1895;Riley, 1960).
The opposed ventral limiting surface of the PTh cuboid represents in our model the longitudinal boundary of the p3 alar plate with the p3 basal plate (p3 tegmentum), that is, the alar-basal boundary (Figure 1d).
In the adult, this longitudinal border is found roughly in a coronal section plane, due to the axial bending occurred developmentally at the The topologically rostral limiting surface of the PTh cuboid (which results deformed into a nearly lateral position, as mentioned above) bounds with the caudal telencephalon at levels through the interventricular foramen, as well as with the underlying alar peduncular hypothalamus at levels through the paraventricular area (PHy; Pa; Figure 1d). This is the surface which thalamo-telencephalic fibers need to cross in order to connect with the internal capsule, forming the upper (alar) root of the cerebral peduncle (Puelles & Rubenstein, 2003. The peduncular hypothalamus contains in its intermediate and superficial strata the topologically dorsoventral hypothalamic course of the cerebral peduncle, that is, the medial and lateral forebrain bundles, respectively (Puelles, Martinez-de-la-Torre, Bardet, et al., 2012;figs. 8.12 and 8.34). The medial forebrain bundle courses through the lateral hypothalamus (an intermediate hypothalamic stratum across both alar and basal plates of PHy), and this stratum is covered superficially by the lateral forebrain bundle  fig. 12). Obviously, the alar PTh relates only to the alar components of this hypothalamic territory (loc.cit.; and PTh; PHy; Figure 1d). The main periventricular rostral neighbor of PTh is the hypothalamic paraventricular nucleus, which is contacted by the PThE, the PThC and the PThSC (Puelles, Martinez-de-la-Torre, Bardet, et al., 2012;figs. 8.09, 8.12, 8.13, 8.15, 8.17, 8.18, 8.20, 8.21, 8.24, 8.27A, 8.30, 8.31). The underlying subparaventricular area is caudally contiguous with the likewise Dlx-positive RLi and ZLR complex (Figure 1d,e; we had to modify this notion, since in earlier accounts we wrongly interpreted that the incertal prethalamus was continuous with the subparaventricular hypothalamus). Dorsally to the paraventricular nucleus, the PThE limits rostrally with amygdalar and hippocampal telencephalic domains (Alonso et al., 2020;Puelles, Martinez-de-la-Torre, Martinez, Watson, & Paxinos, 2019;figs. 4, 11 and 12).
Finally, the caudal limiting surface of the PTh cuboid bounds with the adult thalamus across the transverse glial palisade that develops at the site of the embryonic zona limitans (the latter roughly coincides with the classic "external medullary lamina"; e.g., Keyser, 1972).
Separate Tal1-derived IGL and PG fluorescent cell patches are distinguished, while no prethalamic progenitors express Tal1). The ZL core domain forms itself the transverse palisade of radial glia cells mentioned above, which is crossed orthogonally by all thalamotelencephalic fibers (and viceversa), and may produce as well some mantle derivatives, mainly periventricular or intermediate . It may be accordingly understood that, in embryos, the caudal contact of PTh with the thalamus occurs across the glial ZL border and implies actually two transient parallel borders: the PThC-PThSC versus ZLR limit, and the limit between the ZLR and the ZL core (something similar happens for the thalamus caudally to the ZL). This structural state should be kept in mind for patterning studies, but possibly is not as relevant for adult PTh structural analysis, other than as an explanation of the existence of some peculiar migrated cell types within the neighboring PTh or thalamus (thalamic GABAergic interneurons may originate from the Ptc-and Nkx2.2-positive ZLC and CLi domains, unless they come really from the analogous, but Dlx-positive ZLR; this point is not yet clear). It has been so far a matter of convention whether the ZL, ZLR and ZLC formations are ascribed or not to the PTh and Th; we propose ascribing ZLR to PTh, since this whole field is Dlx5/6-positive and reacts to ZLCo-derived rostrally directed gradiental SHH signals.
As mentioned, the real shape of the deformed adult mammalian PTh is further complicated by the fact that a sizeable rostral part of the mammalian PThE results incorporated to the evaginating hemisphere across the interventricular foramen, bending around the local pial hemispheric sulcus as a fulcrum. This evaginated PThE portion ends in the form of a tapering flap that participates in the medial wall of the hemisphere, next to the sulcus terminalis (asterisk in Figure 23, modified from Puelles, 2001a; note the PThE flap has a Tbr1-positive mantle, which is also characteristically calretinin-positive, both of them being PThE markers) (see also Abbott & Jacobowitz, 1999). The evaginated part of PThE is usually misidentified in human textbook neuroanatomy as the lamina affixa, falsely believed to be adhered to the thalamus pial surface (see Puelles, 2019). Its ventricular surface within the lateral ventricle contacts the medial ganglionic eminence and eventually the amygdala and hippocampus along the sulcus terminalis, as well as the roof plate, which it must reach due to the hyperdorsal nature of the PThE, represented by the chorioidal fissure of the lateral ventricle (Puelles, 2019;Puelles, Martínez-Marin, et al., 2019). The unevaginated part of the PThE becomes visible at the caudal contour of the interventricular foramen (that is where the "eminentia" makes its bulge) and behind it (Figures 1a,b,d,e,f and 23).

| General genoarchitectural of the prethalamus
We turn now to overall molecular markers found at the PTh. The whole alar prethalamus, including the PThE, shares initially the expression of F I G U R E 2 3 Graphically superposed adjacent horizontal sections through the diencephalon of an E13.5 mouse embryo. This pseudocolor darkfield image shows the relative expression patterns of two genes, Dlx2 (in red; subpallium and noneminential prethalamus, p3) and Tbr1 (in green; pallium and eminential prethalamus, PThE), obtained by separate in situ hybridization with mRNA probes. Note the Tbr1-positive evaginated part of the PThE (asterisk) extends beyond the interventricular foramen as a flap added to the caudomedial wall of the evaginated telencephalic hemisphere (it covers the pial surface of p3). The flap ends at the unlabeled thin chorioidal fissure of the lateral ventricle opposite the fimbrial end of the hippocampal primordium. The evaginated PThE flap is separated from the Dlx-positive subpallium by the compacted sulcus terminalis (also marked by the asterisk). Modified from Puelles (2001a) genes such as Zic1 and Zic5, two zinc-finger genes (Ferran, Puelles, & Rubenstein, 2015), or Arx (the PThE is Arx-positive only at early stages, between E10.5 and E13.5, and restricted to its ventricular zone) (Allen Developing Mouse Brain Atlas; Cobos et al., 2005;. In contrast, the basal prethalamus expresses, for instance, Tcf4, Foxa1, Tle4, Lmx1b, and Ptx2 (Martinez-Ferre & Martinez, 2012).
Various gene markers delineate the rostral and caudal boundaries of the prethalamus. We corroborated that the expression of Dlx genes (also Arx) ends caudally at the interthalamic ZL boundary, specifically at the limit between the Dlx-positive ZLR and the Shh-positive ZL core domain (Cobos et al., 2005;Jones & Rubenstein, 2004;; Arx is distinctly expressed also in the ZLR (Puelles, Amat, & Martinez-de-la-Torre, 1987). This caudal PTh limit is also clearly recognized by gene markers which are not expressed in the prethalamus, but in the adjacent thalamus, such as Tcf7l2 and Lef1.
Specific genes of the Dlx family are expressed selectively within PThC and/or PThSC domains and subdomains (see Figure 6), where they probably are related to the differentiation of specific subsets of GABAergic neurons. Shimogori et al. (2010; fig. 3) illustrated Olig2 signal apparently restricted to PThSC, but correlative Olig2 material found at the Allen Developing Mouse Brain Atlas suggests this expression extends also into PThC. As development advances, Arx expression tends to become restricted to the ZLR and PThSC subregions, in a pattern similar to that of Pax6.
Gain-or loss-of-function experiments have shown that the neurally expressed Dlx genes (paralogues 1,2,5,6) are required for the differentiation of GABAergic interneurons in the mouse telencephalon (Anderson, Eisenstat, Shi, & Rubenstein, 1997;Anderson, Qiu, et al., 1997;Long, Cobos, Potter, & Rubenstein, 2009;Long et al., 2007;Stühmer, Anderson, et al., 2002;Stühmer, Puelles, et al., 2002;Wang et al., 2011). As would be expected according to the observed Dlx patterns, in rodents the adult PThC and PThSC regions express both Gad2 (Gad65) and Gad1 (Gad67), which encode two isoforms of glutamic acid decarboxylase, the enzyme responsible for the production of γ-amino butyric acid (GABA); they also express vesicular GABA transporters vGat and Slc32a1 (Nagalski et al., 2016;Yuge et al., 2011). The thalamus instead contains relatively few GABAergic cells, except some produced at the midbrain (Jager et al., 2016), and others generated either at the ZLC, the Dlx/Pv-positive Rt primordium, or another more distant Dlx5/Pv-expressing progenitor domain, from where they migrate tangentially into the Th (Jager et al., 2016; this report prefers for reasons unclear to us a subpallial origin of the thalamis Dlx/Pv interneuronal lineage). Of course, the thalamus also expresses massively vGlut2, which encodes an isoform of the vesicular glutamate transporter (Puelles, Martinez-de-la-Torre, Ferran, et al., 2012;Yuge et al., 2011). DLX and GAD65 proteins have nearly overlapping patterns in the prethalamus at embryonic stages (see fig. 2 in Stühmer, Puelles, et al., 2002). Dlx1/Dlx2 double knockout mice have reduced prethalamic signal of Gad65 and Gad67 mRNA isoforms (figs. 6i C,F and 6iii c,d,g in Le et al., 2017). Moreover, the molecular association of Dlx genes with prethalamus is evolutionarily ancient. An homotopic lamprey prethalamus homolog could be first band, as a phenomenon typically induced by very high levels of SHH morphogen, which occurs either at the ZLR, in front of the ZL core, or at the RLi, along the alar-basal boundary (Puelles, 2013;Puelles, Martinez-de-la-Torre, Bardet, et al., 2012; (Dávila, Guirado, & Puelles, 2000;Díaz et al., 1994;González, Puelles, & Medina, 2002;Martínez-de-la-Torre et al., 2002;Puelles, 2001aPuelles, , 2013Puelles, Martinez-de-la-Torre, et al., 2019;Redies et al., 2000). Moreover, the alar hypothalamus displays a dorsoventral alar organization into only two contrasting longitudinal domains, the paraventricular and subparaventricular areas (Morales-Delgado et al., 2011;Puelles, Martinez-de-la-Torre, Bardet, et al., 2012;Puelles, Martinez-dela-Torre, et al., 2019). As indicated above, the PThE mantle massively expresses TBR1, which also appears expressed less densely, though slightly more ventrally, in the central and ventral parts of the alar hypothalamic paraventricular area (Figure 1c). However, the chicken data of Alonso et al. (2020;also unpublished experimental fate-mapping data) suggest that the paraventricular Tbr1-positive cells are tangentially migrated eminential ones. Experimental studies have shown that diencephalic dorsoventral regionalization is generated by antagonistic effects of dorsalizing and ventralizing inductive signals diffusing simultaneously from the roof plate (mainly TGF-β-related morphogens such as BMP4 and BMP7, and WNT family members) and the floor plate (SHH), respectively (Basler, Edlund, Jessell, & Yamada, 1993;Dickinson, Selleck, McMahon, & Bronner-Fraser, 1995;Echelard et al., 1993;K. J. Lee & Jessell, 1999;Liem, Tremml, Roelink, & Jessell, 1995;Shimamura et al., 1995;Shimamura, Martinez, Puelles, & Rubenstein, 1997). Some peculiarities in the direct or indirect effects mediated by these or ancillary molecules which are causally related to segmental (AP) differential molecular identity may cause the singular emergence and increased relative growth of the PThC and PThSC histogenetic fields, and, ulteriorly, further expansion of its mature nuclear derivatives. This hypothesis needs to be explored experimentally. The roof plate atop the PThE was recently proposed as a forebrain signaling center based on the expression of BMPs, WNTs, and FGFs morphogens. "PThE" transplanted heterotopically next to ventral telencephalic cells led to an ectopic expression of Lef1, a target gene of the WNT/β-catenin pathway (Adutwum-Ofosu, Magnani, Theil, Price, & Fotaki, 2016). Curiously, we have not observed systematic differences in the molecular profile of the PTHC and PThSC regions; on the contrary, molecular AP divisions observed at PThC tended to continue ventrally into corresponding AP divisions of PThSC (see Table 4). Nevertheless, the nuclear derivatives of these fields are structurally characteristic in many ways, as evidenced by classical anatomic schemata, all of which differentiate the zona incerta from the reticular and pregeniculate nuclei (the intercalated subgeniculate central domain is less well known).
This suggests that we may not yet have identified enough of the genes that code for dorsoventral prethalamic architectonic differences within the Dlx-positive territory.  (Figures 20 and 22); remarkably, the characteristic rostral Rt element of PThC was arbitrarily characterized developmentally as a caudal entity in the rat by Altman and Bayer (1979a, 1979b, 1979c, 1988a, 1988b, 1995, who interpreted Rt as a direct derivative of the zona limitans interthalamic border; indeed, they repeatedly identified our ZL as "reticular eminence,"

|
irrespective that the ZL has no direct developmental relationship whatsoever with the Rt primordium (other than as a source of morphogens).
Comparative data on Rt shown by Díaz et al. (1994) in a lizard and Puelles, Martinez-de-la-Torre, et al. (2019) in the chick corroborate a rostral separate position of Rt relative to the ZL, as observed here in the mouse; this error of Altman and Bayer probably causes persistent confusion in students of the interthalamic region.
The subjacent PThSC region is likewise subdivided in rostral, middle and caudal subregions, which represent tri-stratified preincertal (PI), ZIR, and ZIC radial complexes, which are completed caudally by the corresponding nonfused ventral part of the ZLR, identified by us as retroincertal nucleus (RI) (Figures 20 and 22). As mentioned, these incertal subregions tend to share molecular properties with the respective overlying PThC units, including, for example, the Dlx forms expressed postnatally (see Figure 6 and Table 4). We nevertheless deduce from their known anatomical and hodological differences that PThC and PTHSC must be differentiated by as yet undiscovered genetic determinants.
According to present results, in situ expression of Six3 appears restricted largely to the rostral central and subcentral prethalamus (i.e., mainly at intermediate radial levels, at the reticular and preincertal nuclei; Figure 11; Table 4). The rostral and caudal central and subcentral domains are selectively distinguished from the corresponding middle domains by Ecel1 signal (Figure 14; Table 4). This contrasts with Pax6, which is expressed superficially mainly at the subgeniculate nucleus and the underlying ZIR, which belong to the middle PThC domain, as observed in material from the Allen Developing Mouse Brain Atlas from embryonic stage E13.5 onwards (Figures 10 and 21; Table 4 Table 4).
Experimental studies have shown that some molecular anteroposterior regionalization of the prethalamus depends on morphogens secreted by the zona limitans, the mid-diencephalic secondary organizer (Echevarría, Vieira, Gimeno, & Martínez, 2003;Kiecker & Lumsden, 2004;Scholpp et al., 2006;Szabo et al., 2009;Vieira et al., 2005;Vieira & Martinez, 2006;Vue et al., 2009;Zeltser, 2005). SHH and WNT morphogens (WNT8b, WNT3a) are released by the ZL core, and these reportedly diffuse rostrally and caudally in the neighboring alar diencephalon to establish via a position-dependent reaction pattern the differential AP molecular profiles of the prethalamus and thalamus. The concentration gradients of these morphogens (jointly with intrinsic sensitivity properties of the genome) presumedly enable local tissue to select differential AP fates Parr, Shea, Vassileva, & McMahon, 1993;Puelles, 2017;Puelles & Martinez, 2013;Roelink & Nusse, 1991;Scholpp & Lumsden, 2010;Vieira et al., 2005). Our present data suggest that such secondary AP patterning actually occurs in 3-4 analogous steps across the PThC and PThSC parts of the prethalamic alar plate (4 steps counting the ZLR; the SGL would represent an added AP singularity). A possible modulation of this pattern by differential DV signals, including ventralizing ones from the basal/floor plates, or by other AP signals theoretically diffusing from the hypothalamus, has not been explored yet. and Ptc, among other markers, some of which are selective for ZLR or ZLC (Delogu et al., 2012;Echevarría et al., 2001;Gimeno et al., 2002;Golding et al., 2014;Kitamura et al., 1997;Martínez-de-la-Torre et al., 2002;Price et al., 1992;Puelles et al., 2004;Shimamura et al., 1995;Virolainen, Achim, Peltopuro, Salminen, & Partanen, 2012). These thin histogenetic centers are associated to the development of various thalamic or prethalamic derivatives, including the so-called "intergeniculate leaflet," a retinorecipient entity, some of whose cells migrate tangentially either caudally into the thalamic posterior limitans nucleus or rostrally into the prethalamic PG nucleus; (Delogu et al., 2012;Jeong et al., 2011); other deeper ZLR/ZLC derivatives were tentatively identified here on the basis of selective Nkx2.2 expression as the prethalamic RLi and RI, and the thalamic PP and CLi (Figure 7). The mature cell populations selectively derived from the prethalamic ZLR are still poorly known, in contrast to the better studied ZLC derivatives (Delogu et al., 2012;Jeong et al., 2011). We have observed Nkx2.2-positive and CB/Calb1-positive elements within PG. These surely include those that migrate tangentially from the thalamic intergeniculate leaflet (ZLC), but may contain as well prethalamic derivatives of the similarly Nkx2.2positive ZLR ( fig. 6 in Kitamura et al., 1997;present results; see our Figure 13g 0 ,h). The ZLR expresses selectively Dlx family genes, as well as Pax6, Arx, Lhx1 and Dbx1, whereas differential genes such as Tal1, Sox14, Six3, Npy and Calb2 appear selectively at the ZLC (present results; Delogu et al., 2012;Jeong et al., 2011;. We think that the ventralmost (and broadest) part of ZLR forms the retroincertal nucleus (RI), a deep aggregate lying next to the medial lemniscus tract as it penetrates the ventrobasal thalamus; the RI is a newly defined entity (RI in our Figures 3-7 and 16-19). This band is identified here as the ventralmost alar prethalamus as the rostral liminar band (RLi), and it shows like the ZLR/ZLC complex a Nkx2.2/Ptc-positive molecular profile (Puelles, Martinez-de-la-Torre, Bardet, et al., 2012). RLi is held to be due to analogous patterning conditions as the ZL-shell bands, namely the local existence of particularly high SHH levels (secreted by both basal and floor plates); these bands disappear entirely when notochordal ventralization of the neural tube is compromised (Andreu-Cervera et al., 2019). Practically nothing is known about adult RLi derivatives, though we have found at least two candidates, which seem distinct from both the incertal/ preincertal complex (PThSC) and the RI nucleus, and are restricted topographically to the ventralmost part of PTh. One of them is represented by a population of partly disaggregated Dlx-and Pax6-positive cells, which surround the origin of the mamillothalamic tract out of the princeps mamillotegmental tract. This Pax6-positive population was previously identified in the literature as the "nucleus of the mamillothalamic tract" (Skidmore et al., 2012;Szabó et al., 2011;Valverde et al., 2000).
Another neighboring aggregate of Dlx-positive cells found near the prethalamic alar-basal border, as well as near the hypothalamoprethalamic limit, lies practically in an entopeduncular position precisely where the cerebral peduncle bends from its dorsoventral hypothalamic course into its longitudinal tegmental diencephalic trajectory. This small cell group was identified by Ramón y Cajal (1903) as the "accessory subthalamic" nucleus. However, it does not share molecular markers with the subthalamic nucleus proper (which expresses, e.g., Calb2 and glutamatergic markers), whereas the aggregate in question expresses strongly Dlx-LacZ, Sst, Ecel1, and Arx (present results, Allen Developing Mouse Brain Atlas), and probably contains accordingly GABAergic neurons.
Given its entopeduncular hypothalamic position and its gene markers, similar to those of the ventral entopeduncular nucleus, we have chosen to name it the "accessory entopeduncular nucleus." We ignore, however, whether this mass is hodologically and functionally related to the main dorsal and ventral entopeduncular nuclei of the peduncular hypothalamus (Wallace et al., 2017).
As a consequence of its partial evagination, the PThE pial surface is bisected longitudinally by the hemispheric sulcus, which therefore is not a proper limiting sulcus (as opposed to the interpretation of Kuhlenbeck, 1973, who regards it as the external tel-diencephalic limit).

| Central prethalamus (PThC)
We will comment on its superficial, intermediate and deep periventricular derivatives, each of which may be ascribed to either rostral, middle or caudal molecularly distinct compartments.
Central superficial stratum (RP, SG, PG and Ov): The three major superficial elements of the PThC, that is, the retropeduncular (RP), subgeniculate (SG) and pregeniculate (PG) nuclei are disposed in a rostrocaudal row under the PThE, deep to the optic tract (Figures 20   and 22). The RP, SG and PG share retinal input, as was recently illustrated by anterograde labeling studies with cholera toxin B subunit in the mouse (Gaillard, Karten, & Sauvé, 2013; review in Monavarfeshani, Sabbagh, & Fox, 2017;Morin & Studholme, 2014, e.g., fig. 1K-note the RP was wrongly identified as "PP" or "peripeduncular nucleus" by these authors). Confusingly, these experimental data newly establishing RP and SG as prethalamic retinorecipient nuclei lying rostral to the PG were not acknowledged by either Sefton, Dreher, Harvey, and Martin (2015) or Sabbagh et al. (2018). Adding to the confusion, Monavarfeshani et al. (2017) depicted the PP proper as retinorecipient in fig. 1, but did not include the RP in the list of retinorecipient structures.
The thalamic PP, possibly first identified in front of the medial geniculate nucleus by Saper, Swanson, and Cowan (1976), and later shown at this site in rodent brain atlases by Hof et al. (2000), Watson and Paxinos (2010), and Paxinos and Franklin (2013), needs to be distinguished strictly from the prethalamic RP, a wholly different entity which lies just caudal to the peduncle at levels above the zona incerta complex. Faull and Mehler (1985) held that PP belongs to the PTh, without giving its exact derivation. Paxinos and Franklin (2013) do not identify the RP at this position, though it is visible as an AChEnegative subpial gray formation lying just rostral to the AChE-positive 121,146). The conventional PP is commonly mapped more ventrally and caudally than the RP, typically rostral to the medial geniculate body, and close to the tegmental substantia nigra. We conclude that the RP relates to the peduncle next to the alar hypothalamus whereas the PP, which we think derives from the thalamic ZLC, relates to the peduncle as it passes through the thalamic tegmentum (that is, PP is found after the peduncle turns in a right angle around the tegmental subthalamic nucleus; RP lies above this turning point; Figure 20; Puelles, Martinez-de-la-Torre, Bardet, et al., 2012;. We accordingly estimate that both Nakagawa and O'Leary (2001) and Nagalski et al. (2016) referred to the parabasal PP identified within prosomere 2 (thalamus) in their molecular mappings, rather than to the more dorsal alar prethalamic RP within rostral PThC.
It is unclear whether PP also receives retinal projections (Morin & Studholme, 2014). A different retinorecipient entity, the lateral terminal nucleus ("LT") is sometimes mapped next to the PP (Morin & Studholme, 2014;Paxinos & Franklin, 2013), though we regard this topographic ascription as probably erroneous, since we conceive the true LT as a pretectal formation associated to the transverse peduncular tract. This name seems to have been carelessly applied to some other retinorecipient spot next to the zona limitans, without precise indication whether it belongs to PTh or Th.
Comparative considerations suggest that a small and normally cryptic oval nucleus (Ov) may be added to the list of superficial retinorecipient prethalamic grisea present in mammals (see Morin & Studholme, 2014; fig. 1G,H; labeling dorsal to the lateral geniculate and PG projections, or superficial to Rt). Ours results are the first genoarchitectural demonstration of Ov as a distinct Sst-positive entity in the mouse. We propose to apply the descriptive reptilian name "oval" to mammals (as we did before for the avian positional homolog; Puelles, Martinez-de-la-Torre, et al., 2019;Puelles, Martinez-de-la-Torre, Paxinos, Watson, & Martinez, 2007). Topologic homology is complemented in this case by existence of a small separate retinorecipient prethalamic nonlayered formation (not projecting to telencephalon), which lies in all cases dorsal to the PG and the optic tract, as previously recognized in amphibians (the nucleus of Bellonci; see Puelles, Milán, & Martínez-de-la-Torre, 1996), reptiles (original Ov concept; Huber & Crosby, 1926) and birds (classic "lateral anterior nucleus," often wrongly ascribed to the thalamus; Ehrlich & Mark, 1984b;Huber & Crosby, 1929;Repérant, 1973;Webster, 1974). An old account postulating the possible existence of this entity in mammals was recorded by Addens (1938). The Ov nucleus is normally elusive in mammals because, excepting Sst, it does not stain differently than the PG or other prethalamic nuclei with most markers studied. It corresponds in such material to the small subpial stratum where the Rt nucleus seems to reach the brain surface, dorsally to the PG.  fig. 1G,H, though the locus was not identified by these authors; see also review and discussion of Ov concept in Puelles, Martinez-de-la-Torre, et al., 2019;Puelles et al., 2007). The avian Ov also shows selective Sst expression (Ferran, Puelles; unpublished observations).
The comparative literature on the prethalamus in reptiles suggests an antecedent of our RP in the reptilian "suprapeduncular nucleus" described by Frederikse (1931) and Papez (Papez, 1935). Huber and Crosby (1926) distributed much the same territory among a pair of prethalamic nuclei, which they named "ventromedial" and "ventrolateral" nuclei, respectively. The "suprapeduncular" name refers, in columnar interpretation, to a position "dorsal" to the hypothalamic course of the peduncle, which was wrongly thought to be "ventral" in the old columnar model (the relationship is actually caudorostral); this nucleus would thus correspond to our "retropeduncular" element in prosomeric interpretation. The alternative names "ventromedial + ventrolateral nuclei" also refer to apparent columnar position in coronal sections, this time relative to the supposedly "dorsal" (actually "caudal") thalamic mass (Butler & Northcutt, 1973;Cruce, 1974;Huber & Crosby, 1926;Senn, 1968;Shanklin, 1930). These authors did not identify specifically a Rt-homolog population within the "ventral thalamic" (prethalamic) diencephalic area. This notion was first introduced by Pritz and Stritzel (1990). Diaz et al. (1994) later discussed and provided additional hodological data on the reptilian Rt homolog, which proved to be associated to a radially complete rostral prethalamic territory (like our rostral PThC). This complex was composed of a superficial suprapeduncular nucleus (our RP); this element was clearly distinct from the adjacent, but more caudal, superficial ventrolateral nucleus (corresponding to our present SG nucleus). Deep to the SP/RP there was the classic ventromedial nucleus (correlative to our Rt), which seemed to end in a periventricular formation, the dorsal hypothalamic nucleus (corresponding to our present RtPv). Diaz et al. (1994) defined systematically reptilian Rt-like neurons by their characteristic bidirectional and topographically ordered Rt-Th connectivity. The whole mediolateral Rt prethalamic complex (equivalent topologically to our present RtPV, Rt proper and RP in mouse) projected in a topographically ordered fashion upon the three dorsoventral tiers of thalamic structure (Díaz et al., 1994) (compare mappings of these tiers in Dávila et al., 2000;Martínez-de-la-Torre et al., 2002;Puelles, 2001b;Redies et al., 2000). Dávila et al. (2000) subsequently also illustrated some lizard suprapeduncular neurons labeled retrogradely from the thalamus, which clearly mapped just caudal to the alar hypothalamic sector of the peduncle in a sagittal section ( fig. 5E,F) fig. 1) which must be homologous with our present RP, SG and PG superficial units within mouse PThC. We substituted the RP (retropeduncular) term for the earlier SP name following our standing program to abandon obsolete columnar references whenever possible, substituting descriptive terms consistent with the prosomeric forebrain axis. The "retropeduncular" descriptor is more precise than SP, which might be confused with the parabasal "peripeduncular" (PP) term.
The RP within rostral PThC shows in our hands selective PV-, Isl1-, Six3-, Ecel1-, and Sst-expression, a profile that is shared with the deeper Rt cell population. The RP cells differ from Rt cells by their higher expression of Dlx5/6-LacZ and lack of calbindin (CB) ( Table 4).
In the cat, a RP nucleus located consistently with our definition was identified as a superficial extension of the Wisteria floribunda agglutinin (WAF)-positive and SMI32-positive Rt population (Baldauf, 2010; figs. 1B, D and 3A-C).
The superficial SG nucleus was apparently first recognized in the rat as a separate AChE-positive entity by Paxinos and Watson (1986; these authors also coined this term). As mentioned above, the SG was identified as a retinorecipient nucleus by Morin and Studholme (2014), but not by Sefton et al. (2015). Selective AChE labeling distinguishes the SG not only from the PG, but also from our RP (see our discussion above), which is also AChE negative (Paxinos & Franklin, 2013). The SG also stands out among neighboring superficial prethalamic nuclei in the mouse by its relatively strong Enc1 and Pax6 staining, and it is selective lack of Ecel1 expression (Allen Developing Mouse Brain Atlas; present results: Figures 10, 12 and 14 Table 4).
It accordingly was duly proposed to use this term also in birds (implicitly likewise in reptiles) to replace the previous "ventrolateral nucleus" name, which has clearcut wrong columnar connotations (Puelles, Martinez-de-la-Torre, et al., 2019;Puelles et al., 2007); of course, "subgeniculate" is also columnar-inspired-and strictly wrong, topologically, since it actually is a "pre-pregeniculate" entity rather than "subgeniculate," but this term is clumsy, and we thought it preferable to keep the known topographic SG name for the sake of clarity. Consistently with retinal input to the SG in rodents (Gaillard et al., 2013;Morin & Blanchard, 1999;Morin & Studholme, 2014) a prethalamic SG (ventrolateral nucleus) receives also retinal projections in birds (Ehrlich & Mark, 1984a, 1984bNorgren & Silver, 1989) and reptiles (Bass & Northcutt, 1981;Butler & Northcutt, 1978).
In the hamster, SG receives also afferent inputs from the thalamic intergeniculate leaflet and projects ipsilaterally to its prethalamic PG neighbor (old ventral lateral geniculate nucleus) and bilaterally to the intergeniculate leaflet (Morin & Blanchard, 1999).
The PG, jointly with the correlative dorsal oval nucleus (Ov), forms the caudal major superficial component of the central prethalamic subregion. The prethalamic laminar PG was previously known as "ventral lateral geniculate nucleus" (Hines, 1929) based on its location with respect to the thalamic dorsal lateral geniculate nucleus, with the intercalated IGL formation, and its retinorecipient connectivity. The terms ventral and dorsal derived from the obsolete columnar model, and correspond to topological prosomeric rostral and caudal locations. Accordingly, the names pregeniculate (PG) and lateral geniculate (LG) nuclei seem appropriate for these two entities. Both nuclei have differential development, cytology, immunohistochemistry, and pattern of connections (review in Monavarfeshani et al., 2017). In addition to a different prosomeric origin (PG derives from p3 versus LG from p2), the PG has a Dlx-positive profile reflecting its GABAergic character, whereas the LG contains mainly glutamatergic neurons and only few GABAergic cells (Gabbott & Bacon, 1994;Harrington, 1997;Inamura, Ono, Takebayashi, Zalc, & Ikenaka, 2011;Puelles, Martinez-de-la-Torre, Ferran, et al., 2012;Sabbagh et al., 2018;Yuge et al., 2011). Moreover Htr2c (a serotonin receptor) and Cdh6 are restricted to PG, whereas Sert, Nr1d1, Gfrα1 and p57kip2 markers label specifically LG (Yuge et al., 2011).
Respect to their connections, there are differences in the subtypes of retinal ganglionic cells that innervate PG and LG nuclei and in the properties of their retinal synapses (Monavarfeshani et al., 2017). Retinal inputs innervate massively the magnocellular external PG layer, which is Enc1-and Htr2c-negative in contrast with the nonretinal inner layer, which is Enc1-and Htr2c-positive (Yuge et al., 2011;present results; Table 4). Retinal and nonretinal layer cells of PG also express differentially Cat315-and WFA-extracellular matrix proteins, respectively, in contrast with LG cells, which do not express either of these molecules (Sabbagh et al., 2018). Nonretinal inputs to the PG are more diverse and mainly different that those to LG, including sources such as the superior colliculus, visual cortex, and several pretectal and rhombencephalic nuclei (review in Monavarfeshani et al., 2017;see figs. 3 and 4). Visual cortical cells project to both PG and LG, but layer V cells innervate PG whereas layer VI cells project to LG in rodents (Bourassa & Deschênes, 1995;Cosenza & Moore, 1984;Hammer et al., 2014;Jacobs et al., 2007;Seabrook, El-Danaf, Krahe, Fox, & Guido, 2013). PG efferents are also more diverse that those of LG, innervating regions related with visuomotor function, eye movement, vestibular function, and circadian function, including the superior colliculus and the hypothalamic suprachiasmatic nucleus (Matute & Streit, 1985;Moore, Weis, & Moga, 2000;Taylor, Jeffery, & Lieberman, 1986). In sharp contrast to the LG, the PG does not project to the visual cortex, or any other cortical region (Harrington, 1997). The Rt domain is labeled selectively by Six3 at perinatal and postnatal stages. The middle and caudal subregions of the PThC, containing the SGI and PGI cell populations, separate the Rt from the thalamus. The homologous reticular nucleus population in lizards and crocodilians, which suffers less morphogenetic deformation than in mammals, has been clearly illustrated as a rostral PTh subregion (Díaz et al., 1994;Pritz & Stritzel, 1990), which consistently shows GAD and PV expression (Pritz, 2018).
The term "nucleus reticularis," previously named "Gitterschicht" ("grid layer") by Nissl (1889), was introduced by Münzer and Wiener (1902) and corresponds to our Rt proper (possibly including as well our RP, T and RtPv), also known as the "main body of the reticular nucleus" (Clemence & Mitrofanis, 1992). These authors distinguished three cytoarchitectonic subregions in the reticular nucleus of cats and two in ferrets: a main body, an inner small-celled part and a perireticular nucleus. The main body of their Rt corresponds largely to our Rt proper, whose cells are GABAergic and mostly immunoreactive to PV in all studied mammals (review in Jones, 2007;Mikula, Manger, & Jones, 2008). Moreover, the Rt contains somatostatin and calbindinpositive cells (SST/Sst and CB/Calb1) (Clemence & Mitrofanis, 1992;Mitrofanis, 1992; present results; Table 4). The expression of genes such as Arc, Kcnj4, Lancl3, Rnf144b, Tiam2, Trh, Fgd5 (Nagalski et al., 2016), Meis2 and Deinh2 (Zeisel et al., 2018) appears also restricted to the Rt.
The disperse cells of the perireticular nucleus are immersed in the hypothalamic course of the peduncle, that is, lie within the Dlx-negative paraventricular peduncular hypothalamic subdomain, which lies just rostral to the Dlx-positive prethalamic RP/Rt complex. The perireticular cells are also GABAergic (GABA-and DLX-positive elements) and share with the prethalamic RP/Rt complex PV and SST (Clemence & Mitrofanis, 1992;present results), and other markers such as Isl1, Six3 and Ecel1 (present results). These data might support a prethalamic origin of these cells, assuming a subsequent migration to the paraventricular hypothalamic subdomain, as was previously suggested (Earle & Mitrofanis, 1996;Mitrofanis, 1994;Puelles, Martinez-de-la-Torre, Bardet, et al., 2012).
The Rt has an important role as a modulator of the excitatory interactions between the cerebral cortex and the thalamus. Subregions of this GABAergic nucleus project, mainly ipsilaterally, to distinct thalamic nuclei. Moreover, the Rt receives topographically ordered innervation from collaterals of traversing glutamatergic thalamocortical and corticothalamic axons (Coleman & Mitrofanis, 1996;Crabtree, 1998;Guillery, Feig, & Lozsádi, 1998;Pinault et al., 1995aPinault et al., , 1995bPinault & Deschênes, 1998). Based on electrophysiological studies, the mammalian Rt may be subdivided in at least seven segregated sectors, five of them sensory, one motor and one limbic, with some overlap between these sectors (review in Pinault, 2004;Sokhadze, Campbell, & Guido, 2019). Electrophysiological properties of neurons obviously vary depending on their location in the Rt (S. H. Lee, Govindaiah, & Cox, 2007). Neuronal typological heterogeneity has been noted also in the Rt (e.g., a majority of PV-expressing cells vs. minor subpopulations of CB-and STT-expressing cells).
The neighboring narrow intermediate strata of the SG-and PGrelated radial complexes (SGI, PGI) are difficult to differentiate from the Rt in Dlx5/6-LacZ material, but Enc1-and Ecel1-expression distinguishes PGI from its Isl1-and PV-positive SGI neighbor, and both Six3 and Sst label transiently a thin caudal lamina of SGI (the SGL) abutting on the unlabeled PGI (present results; Table 4). These entities are not identified in mammalian neuroanatomical literature, where they usually are all ascribed to the Rt. In contrast, a sparsely populated PTh sector lying caudal to the Rt and rostral to the interthalamic limit, was clearly recognized in lizards as an intercalate area (Díaz et al., 1994); it may correspond to the sum of our SGI and PGI.  Table 4). In double Dlx5/6-LacZ reaction and pan-DLX immunoreaction, the PI likewise appears as a brown-reactive mass, indicative of selective Dlx1 expression. Due to its rostral position, the PI limits directly with the subparaventricular peduncular alar hypothalamus.
The zona incerta (ZI) was originally defined in myelin-stained preparations as a scarcely stained area situated between large tegmental fiber tracts (Forel, 1877). Later several authors incorporated the ZI under the concept of "subthalamus," jointly with other formations such as the subthalamic nucleus and the dorsal hypothalamic nucleus (review in Puelles, Martinez-de-la-Torre, Bardet, et al., 2012; here reasons were considered to regard the confuse "subthalamus" concept as obsolete). Jones and Burton (1976) suggested that this neuronal population should be named "the nucleus of the zona incerta," but this name has not been used. The whole domain is GABAergic and cyto-and chemo-architectonically heterogeneous, with extensive afferent connections mainly from the thalamus, hypothalamus, brainstem, and spinal cord (review in Mitrofanis, 2005). The classic ZI is presently subdivided in three rostrocaudally ordered aggregates (PI/ZIR/ZIC). ZIR and ZIC were conventionally identified under columnar assumptions as "dorsal" and "ventral" parts of the zona incerta, although they are respectively rostral and caudal parts in a neuromeric topological conception (ZIC, ZIR). We added here the rostral preincertal nucleus (PI) as a further somewhat cryptic component of the PThSC (see above; Table 4). Kawana and Watanabe (1981)  Subcentral periventricular stratum (PIPv, ZIRPv, ZICPv): The changing molecular profiles of this rather small prethalamic region, not studied by most earlier authors, are presented in Table 4.

| Rostral shell of the zona limitans (ZLR) and RLi
We already commented above about the conceptual problems posed by this peculiar prethalamic primordium and its derivatives, in the context of relevant literature.

| Conclusions
We think that our updated molecular and topologic conception of the prethalamus may be instructive for causal navigational explanations of the thalamo-telencephalic connections, since the latter necessarily must be guided through the heterogeneous prethalamus (namely, its three AP parts, aside of its DV subdivisions). Prethalamic molecular heterogeneity suggests a more complex scenario for prethalamic patterning than has been contemplated so far. Our new model also provides new light upon the still insufficiently understood diencephalo-telencephalic boundary (Alonso et al., 2020;Puelles, 2019). Another scenario for further prethalamic studies includes exploring the role of the diverse retinorecipient grisea distinguished in our prethalamic schema (visuomotor or other functions), in comparison, for instance, with those of pretectal and midbrain visual centers.

ACKNOWLEDGMENTS
This work was funded by R01 MH049428 from NIMH to JLR and the Séneca Foundation (Fundación Séneca) contract 19904/GERM/15 to LP. Infrastructure support provided by the UCSF Medical School, IMIB and the University of Murcia is also acknowledged.

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

Luis
Puelles https://orcid.org/0000-0002-9541-7073 Carmen Diaz https://orcid.org/0000-0002-4925-0858 ENDNOTE 1 Note the clearcut dorsoventral retinal projection pattern upon the diencephalon demonstrated by means of transport of a LacZ-labeled retinoic-acid reporter by Wagner et al. (2000); by means of transport of a LacZ-labeled retinoic-acid reporter shows throughout PTh, thalamus and pretectum an inverted DV topography which is only consistent with the prosomeric model (and our present prethalamus model), but cannot be explained within the antagonic diencephalic columnar model. One of the strongest points of the prosomeric model which is not fully realized in the field is that it makes sense of the spatial organization of retinotopic projections (see also Marín, Blanco, & Nieto, 2001). This implies that growing axons somehow detect and react to serial neuromeric fields and boundaries when they form synaptic connections.