5384204 Brain Segmentation and Forebrain Development in Amniotes

Embed Size (px)

Citation preview

Brain Research Bulletin, Vol. 55, No. 6, pp. 695710, 2001 Copyright 2001 Elsevier Science Inc. Printed in the USA. All rights reserved 0361-9230/01/$see front matter

PII S0361-9230(01)00588-3

Evolution of the Nervous System

Brain segmentation and forebrain development in amniotesLuis Puelles* Department of Morphological Sciences, University of Murcia, Murcia, SpainABSTRACT: This essay contains a general introduction to the segmental paradigm postulated for interpreting morphologically cellular and molecular data on the developing forebrain of vertebrates. The introduction examines the nature of the problem, indicating the role of topological analysis in conjunction with analysis of various developmental cell processes in the developing brain. Another section explains how morphological analysis in essence depends on assumptions (paradigms), which should be reasonable and well founded in other research, but must remain tentative until time reveals their necessary status as facts for evolving theories (or leads to their substitution by alternative assumptions). The chosen paradigm affects many aspects of the analysis, including the sectioning planes one wants to use and the meaning of what one sees in brain sections. Dorsoventral patterning is presented as the fundament for dening what is longitudinal, whereas less well-understood anteroposterior patterning results from transversal regionalization. The concept of neural segmentation is covered, rst historically, and then step by step, explaining the prosomeric model in basic detail, stopping at the diencephalon, the extratelencephalic secondary prosencephalon, and the telencephalon. A new pallial model for telencephalic development and evolution is presented as well, updating the proposed homologies between the sauropsidian and mammalian telencephalon. 2001 Elsevier Science Inc. KEY WORDS: Prosomeric model, Forebrain, Patterning, Segmentation, Neuromeres, Longitudinal zones, Histogenesis, Migration, Diencephalon, Telencephalon, Pallium, Subpallium.

INTRODUCTION Amniote vertebrates include reptiles, birds, and mammals. Stem reptiles clearly were ancestral to all present amniote forms, and derived from anamniote amphibian stock. It is still being discussed which is the correct evolutionary branching schema among amniotes. Turtles are thought to be closest to both stem reptiles and stem mammals, and crocodiles, jointly with extinct dinosaurs, closest to birds, which agrees with data on brain morphology (see [87] for a modern molecular viewpoint). The amniote forebrain is notoriously complex and therefore there is no easy way to approach its development. I will not deal here with neural induction, early regional patterning, and clonal

growth previous to neurulation. These fundamental processes show substantial topological, cellular, and molecular similarities when studied across vertebrates (for review see [73]). In trying to understand forebrain development and evolution, I have come to favor a threefold approach, which involves topological analysis of form and structure, causal cellular mechanisms, and molecular fundaments. The last two lines of research are pursued abundantly in current developmental literature, but topological analysis remains largely a solitary endeavor, owering exceptionally here and there. Still, I believe it is an all important line because topology of form and structure gives critical contextual (positional) meaning to cellular interactions and molecular function. It is of no use to simplify excessively or forget this aspect. Of course, in the brain, form and structure means the shape and building blocks of the walls of the neural tube, and these evolve variously in three dimensions during both ontogenesis and phylogenesis. The hollow shape of the neural tube is a topological onemany transform of the neural plate shape (Figs. 1A,B) as indicated by fate-mapping studies [16,34,56,73]. A one-many transform means that, growth being implied, each small primordium (cell group or individual cell) gives rise to a larger derivative (or clone), but the overall topology of parts does not change. Later, differential growth and differentiation processes introduce other site-specic topological transformations (i.e., evagination of the telencephalic hemispheres; Fig. 1D). In essence, regionalized distinct elds of the primitive neural tube wall each give rise to a tridimensional radial complex of neural structures in the mature brain, stretching from the ventricular surface to the pial surface (Figs. 1C,D). Eventually, some of the cell populations originated within one radial histogenetic complex migrate tangentially, colonizing other radial complexes (arrow T in Fig. 1D; [1,2,49]). This usually affects the grain or structure of the brain wall, but not so much its form, at least in the forebrain. Some tangential migrations in the hindbrain do affect its form (i.e., the rhombic lip migrations). Cell groups found at each site of the neural tube wall thus in principle have formal relationships along three dimensions: DV, anteroposterior, and radial (ventriculo-pial, or inside-out) (Fig. 1). Understanding of local brain morphogenesis therefore involves

* Address for correspondence: Luis Puelles, Department of Morphological Sciences, Faculty of Medicine, University of Murcia, 30100 Murcia, Spain. Fax: 34-968-363955; E-mail: [email protected]

695

696

PUELLES

FIG. 1. Schemata illustrating the topological radial structure and relative dorsoventral (DV) dimensions of the neural plate and tube. (A) Neural plate cross-section and direction of neurulation deformation. (B) Neural tube, overlying neural crest and skin. (C) Neural tube cross-section at a more advanced stage of development, showing differential mass of ventricular zone neuroepithelium (inverse to differentiated state) and radial migration routes of alar and basal neuroblasts into the forming mantle layer. Arrows D/V indicate dorsal and ventral directions. (D) Schema of more complex forebrain area, including the evaginating telencephalon (section plane indicated in Fig. 6). Note advanced regionalization, particularly of the alar plate, and the diverse topography in each part of the topologically invariant radial dimension (unlabeled arrows). For comparison, arrows labeled D or V indicate the deformed dorsoventral dimension and arrow T symbolizes what would represent a supercial tangential migration. Abbreviations used in gures: A, anterior; Aa, anterior archistriatum (part of avian amygdala); ac, anterior commissure; ACC, nucleus accumbens (striatal); AEP, anterior entopeduncular area; AH, anterior hypothalamus; Aid, intermedio dorsal archistriatum (part of avian amygdala); Aiv, intermedio ventral archistriatum (part of avian amygdala); AMYGD amygdala; AON, anterior olfactory nucleus; AP, alar plate; Ap posterior archistriatum (part of avian amygdala); BL, basolateral amygdaloid nucleus; BM, basomedial amygdaloid nucleus; BP, basal plate; BST, bed nucleus stria terminalis; BSTA, anterior bed nucleus stria terminalis; BSTP, posterior bed nucleus stria terminalis; C, central amygdaloid nucleus; ca, anterior commissure; Cb, cerebellum; CENT, central; cf, cephalic exure; ch, choroidal roof tissue; CLdl, dorsolateral claustrum; CLvm, ventromedial claustrum; cp, pallial/hippocampal commissure; CX, cortex; D, dorsal; DBH, diagonal band, horizontal part; DBV, diagonal band, vertical part; DH, dorsal hypothalamus; Di, diencephalon; DP, dorsal pallium; DT, dorsal thalamus; DVR, dorsal ventricular ridge; EMT, eminentia thalami; Ep, endopiriform nucleus; EP, epiphysis; m, mbria hippocampi; FP, oor plate; Hab, habenula; hc, hippocampal commissure; HP, neurohypophysis; HV, hyperstriatum ventrale (dorsal part of avian DVR LP); IC, internal capsule; IH, infundibular hypothalamus; IIIv, third ventricle; Is, isthmus; ivf, interventricular foramen; L, lateral amygdaloid nucleus; LGE/ST, lateral ganglionic eminence (striatum); lot, lateral olfactory tract; LP, lateral pallium; M, midbrain; Mam, mammillary region; Me, medial amygdaloid nucleus; MES, mesencephalon; MGE/BST, medial ganglionic eminence (supracapsular bed nucleus of stria terminalis); MGE/PAL, medial ganglionic eminence (pallidum); MP, medial pallium; mz, mantle zone; N, neostriatum (ventral part of avian DVR VP); NC, notochord; NCL, neostriatum caudolaterale (part of neostriatum); OB, olfactory bulb; och, optic chiasm; os, optic stalk; P, posterior; p1p6, prosomeres 1 6; PAL, pallidum; Palio, pallium; PAL-S, pallidal septum; PO, preoptic region; POA, anterior preoptic area; PP,

AMNIOTE FOREBRAIN DEVELOPMENT visualizing six neighborhood relationships (rostral, caudal, dorsal, ventral, inner, and outer) in causal morphospace. To this must be added known instances of tangential migration, which relocate cells from one topological locus into another (and other invasive phenomena, like vascularization and colonization by blood-borne microglia cells). Fortunately, at the beginning, important changes may affect preferentially one structural dimension, allowing us to momentarily consider the other dimensions as relatively unchanged. Dorsoventral (DV) patterning causes longitudinal zonation in the forebrain and elsewhere in the brain, which is topologically similar to a closed cylinder. This I regard as one of the earliest regionalization effects, immediately consequent to planar induction effects before and during gastrulation (Fig. 2; see [13,19,75]). Classically, planar induction has been described to affect only the anteroposterior (AP) pattern [15]. AP patterning leads to major brain parts (tagma), secondary organizers (like the isthmic organizer, or the thalamic zona limitans), segmentation inside tagma (metamera or neuromeres), and transverse regionalization in general. Note that mixed DV/AP effects may generate later some oblique patterns (see, e.g., [68]). Finally, later radial patterning comprises locus-specic neuroepithelial histogenesis and stratication of diverse sorts of postmitotic neurons in the mantle later, with progressive growth and maturation of the neural wall (Figs. 1C,D). Regulated proliferation (clonal patterning) is a shape-changing overall process superposed in varying patterns and degrees upon other major histogenetic processes which contribute quantitatively to forebrain morphogenesis (differentiation, cell migration, axonal navigation and fasciculation, formation of synaptic neuropils, and trophic versus cell death processes). Note that one fundamental differential property of the amniote brain structure as compared to the brains of anamniotes (shes and amphibia) is the trend for more marked radial translocation of periventricular postmitotic neurons into outer strata of the mantle zone. This difference is already evident in reptilian brains, and both birds and mammals substantially improve upon reptilian standards along this trend, albeit often with different nal outcomes. Many homology relationships which may seem strange at rst glance nd an easy explanation when this particular differential trend is considered (i.e., homologous neurons or whole neural centers come to occupy completely different radial positions). It is possible that such a difference also obtains with regard to tangential migrations, though there is less knowledge on this aspect. THE IMPORTANCE OF A PARADIGM: PREMORPHOLOGICAL ASSUMPTIONS As pointed out in a previous review [56], when we look at a brain part like the forebrain, we rarely are (or want to be) neutral observers, gathering in data which are assumption-free. Neuromorphology of the forebrain is far too complex to be dealt with efciently other than with morphological preconceptions, or assumptions, which help order the variety of data into seemingly meaningful parts. We normally see what we expect to see and only well prepared minds can notice what was not expected. Because morphological assumptions shall be used anyway, it is well worth it to examine them critically, recognize their advantages and dis-

697 advantages, and keep alternative options in mind for tentative use whenever problems appear. Form analysis is hypothetic-deductive, so that looking at the specimens either with the naked eye or the microscope represents nothing less than a test of various deduced statements, drawn explicitly or implicitly from recent or old morphological hypotheses and theories. Inconsistent or falsating observations (to falsate is to prove wrong a hypothesis or prediction) should alert us to unforeseen errors of our logic or theory, and make us search proper theoretic replacements. For normal daily analysis of results, we inevitably use a morphological paradigm, which is a complex set of more or less well-grounded beliefs (essentially a set of reasonable assumptions), which in the best case are (or might be) shared by many other colleagues. Paradoxically, the more a paradigm is shared, the less conscious we become of its use. In that case we tend to forget its essentially tentative nature, with the risk that it may become established as a dogmatic barrier to scientic progress (our minds will slip away from any thought that contradicts the dogmatic paradigm, rationalizing this in various illogical ways). Note that paradigms are not proven true while they hold this status because they just are current belief and essentially stand on the unknown. They are mental tools used (as viewpoints and methodological or interpretive scaffolds) as long as they seem to suggest meaningful addressing and interpretation of reality. Later, as the frontier of knowledge expands, old paradigms may become part of established fact within theories, or may be modied or rejected because they seem contradictory with meaningful progress. Acceptance of a paradigm (morphological or other) thus depends on the historical status of scientic knowledge and, particularly, on the reliability conceded to the diverse technical approaches that produce relevant data in each eld. New technical approaches generally lead sooner or later to changes in the accepted paradigm because old conceptions frequently turn out to be insufcient to deal with new sorts of data. When a long-trusted morphological paradigm starts to imply that a mass of newly accumulating data is meaningless, then it is high time to look around for a better paradigm. Phenomena in embryos and brains rarely are meaningless, due to the immense net of causal interactions tying together all that happens there. Most of the views considered true in this essay are based on a developmental topological perspective and on recent studies mapping gene expression, framed within the neuromorphological segmental paradigm described below. Unfortunately, in neuromorphology obsolete paradigms have a strong tendency to become entrenched dogmatically, camouaged as established nomenclature or facts in textbooks and scientic literature (tradition stretches back to Greek-speaking Alexandria). For instance, widely used terms such as dorsal and ventral parts of the thalamus, or epithalamus and hypothalamus, imply use of a particular morphological paradigm [27,28,41], which essentially contained a now clearly falsated belief in a straight length axis of the forebrain, stretching from telencephalon to midbrain and hindbrain (discussion in [56,60]). According to that paradigm, these thalamic parts have to be understood as DV patterns (thus the dorsal/ventral and epi-/hypo- terms). Nowadays, the vertebrate forebrain axis is by contrast thought to become curved early on at

prechordal plate; PRET, pretectum; PS, pallial septum; PT, pretectum; PV, supraopto-paraventricular area; R, rhombencephalon; r1r6 rhombomeres 1 6, `r7-r11 pseudorhombomeres 711 after Cambronero and Puelles [8]; RP, roof plate; S, septum; SP, secondary prosencephalon; st, stria terminalis; ST, ` striatum; ST-S, striatal septum; Subpalio, subpallium; T, tangential; TD, dorsal thalamus; Tel/TEL, telencephalon; TN, nucleus taenia (part of avian amygdala); TV, ventral thalamus; V, ventral; VP, ventral pallium; VT, ventral thalamus; vz, ventricular zone; zl, zona limitans intrathalamica.

698

PUELLES

FIG. 2. Schemata illustrating topology of primary processes of neural induction and concomitant dorsoventral (DV) patterning in the neural plate, leading to an overall longitudinal zonation. (A) Neural plate model with a bilateral symmetry: the oor, basal, alar, and roof longitudinal zones from both sides would be arranged as fully parallel elements ending rostrally at a free edge (this model is not supported at all by molecular or experimental data for the forebrain, though it is valid for structures caudal to it; see Fig. 3). (B) Neural plate model with a circumferential symmetry: longitudinal zones from either side meet rostrally (forebrain model supported my molecular data and fate maps; compare with Fig. 3). (C) Opposed planar induction effects postulated to occur across the early neural plate. The arrows symbolize epidermalizing and neuralizing inductive effects spreading from the neural plate boundary and the node, respectively; however, note the topological correspondence with hypothetical dorsalizing and ventralizing effects when this pattern is compared with the circumferential symmetry model in (B). Paradoxically, anteroposterior (AP) equals DV at the neural plate midline (or AP is a concept that only makes sense once the neural tube is closed). (D) Deformation of the neural primordium during two stages of neurulation, in oblique frontal view. At left, the rostral and caudal neuropores are still open. At right, the roof suture is closed and the choroidal roof of the forebrain starts to develop, together with the optic vesicles and the telencephalon. Note rostral-most portions of the longitudinal zones.

the cephalic exure [3,26,62,76]. Therefore, the same thalamic parts, irrespective of their names, which probably will not change, are now conceived largely as effects of AP patterning, rather than

of DV patterning. Otherwise, a set of early gene expression patterns, many experimental embryological observations, and various descriptive accounts of molecular markers, are all wholly mean-

AMNIOTE FOREBRAIN DEVELOPMENT ingless, or require very convoluted ad hoc interpretations. The old paradigm is indeed obsolete, because we now have abundant evidence that the embryonic and adult brain axis is not straight and no one can ostensibly support the statement that he/she still believes in a straight axis. The point is so fundamental that the resulting conclusions cannot be escaped. The old morphological observations per se obviously are as valid as ever, but the way in which we conceive what they mean and how they can be explained has changed, because we now share a different set of assumptions [56]. This does not mean that we now (or ever) know the Truth. BRAIN SEGMENTATION AS RELEVANT PARADIGM FOR FOREBRAIN ORGANIZATION The concept of brain segmentation represents a curious morphological paradigm because it was rst postulated at the end of the 19th century (reviewed by [43]). It emphasized observable transverse segments in the embryonic brain, correlative with segments in the head and body of vertebrates, and was used as the prevalent morphological paradigm for about 40 50 years. Its use was then discontinued by most researchers (with a few exceptions in Europe), as interest increased in the functional connotations of the longitudinal columns of the brain (i.e., columnar central arrangement of the functional components of the nerves). The resulting alternative columnar paradigm was elaborated during the rst quarter of the 20th century, and became prevalent worldwide after the second World War. The idea of segments obviously was not helpful for understanding columnar aspects of neural structure and function, with the consequence that segmental observations tended to be explained away as transient developmental phenomena, and came to be regarded as irrelevant for understanding the functioning mature brain. The columnar paradigm has guided most neuromorphology during the important second half of the 20th century, in which massive sets of data on anatomy, connections, chemoarchitectony, and functions were collected under this set of assumptions, congurating the ample fundament of present-day neuroscience. However, the alternative segmental paradigm continued to be found useful in some version or other by various authors (i.e., [4,5,12,18,22,38,60,62,70,83,84,86]). All these minority views recognized correctly, as it nally turned out, that transverse subdivisions were clearly compatible with, and in fact needed, an accompanying longitudinal subdivision. Likewise, proper functional, structural, and causal interpretation of longitudinal columns needs distinguishing transverse components within them [8,47]. The advent of molecular biology to the study of neural patterning rapidly produced a growing list of gene expression data and corresponding experimental results on gene functions in neural developmental signaling, which are manifestly relevant for assessing neuromorphological paradigms and theories. These emergent data in general have contradicted the standard columnar expectations for the forebrain and were found to be consistent instead with some forms of the old segmental paradigm [3,18,26,56,60, 72,73,76,77]. Curiously, the failure of the columnar paradigm did not reside in the concept of longitudinal columns, which was strongly supported by molecular data, but in the belief in a wrong longitudinal axis (see below). The longitudinal zones of the forebrain supported by these molecular studies therefore are different from those suggested by the columnar paradigm and in general are continuous with the longitudinal zones of the brainstem (Figs. 2 4). At the turn of the millenium, therefore, we have seen an important change in neuromorphological paradigm for the forebrain, springing from the search of meaningful interpretations of gene expressions and functions. This implied a reoating of the old

699

FIG. 3. Fate map of the neural plate in the frog Xenopus (according to Puelles [56]). The oor plate is represented as a thick midline line. The basal-alar boundary appears dashed. At left, the major transverse boundaries are indicated, separating, secondary prosencephalon (with the eye eld and telencephalon) from the diencephalon, midbrain, rhombencephalon, and spinal cord. At right, prospective interprosomeric boundaries postulated in the favored model are presented, together with a projection of the subsequent expression pattern of two homeobox developmental genes, which respect to such boundaries at later stages (Dlx1/2 in gray; Gbx2 striped).

segmental ideas, but also an integration of what was solidly established with respect to longitudinal zones. This comparatively recent paradigmatic change (or molecular substantiation of an old paradigm; [56]) still has not become generalized, or even recognized as such, by all scientists, although it starts to appear at the textbook and atlas level [20,21,74,78]. This leads to considerable confusion because naive individuals inevitably try to make compatible some obsolete morphological beliefs supported in the standard literature with recent molecular data and corresponding lines of reasoning, which are partially mutually incompatible. The following paragraphs contain an introduction to the recent segmental paradigm, as a tentative conceptual scaffold for understanding vertebrate forebrain development in a modern molecular context. We call this the prosomeric model [6,48,56,60,68,72,73,76,77]. Relationship to DV Patterning of the Forebrain As mentioned above, this process begins very early during neural plate formation (Fig. 2). Early planar neural induction is triggered at short range by morphogens diffusing from the node. Topologically, this can be understood as neural (possibly differential) specication of concentric rings of epiblast lying between

700

PUELLES

FIG. 4. Schemata illustrating axial incurvation and transversal segmentation of the rostral neural tube in relation to the underlying parts of the axial mesoderm (prechordal plate and notochord). The oor plate induced and maintained by the axial mesoderm is marked as a thicker black line in the oor of the neural tube. The alar-basal boundary appears as a dash line. The thicker transverse lines represent the intertagmatic boundaries (compare Fig. 3). Thinner transverse lines mark interneuromeric boundaries in the hindbrain (rhombomeres 1 6 and pseudorhombomeres 711, according to Cambronero and Puelles [8]) and forebrain (prosomeres 1 6). Note the isthmus (Is) builds the rostral-most distinct part of the hindbrain. The cephalic exure (arrow marked cf in the middle drawing) progressively separates the midbrain and diencephalon from the rigid axial mesoderm, causing a corresponding deformation of all the longitudinal and transverse boundaries in the area. The relative topology of the eye and telencephalic evaginations is indicated as well. Note that a topologically transverse section through prosomeres 5 or 6 would give the schematic section depicted in Fig. 1D.

the node (the primary organizer) and the peripheral neuralnonneural boundary (Fig. 2C). The topology of this process across the early neural plate reects circumpherential symmetry, rather than bilateral symmetry (Figs. 2A,B). This distinction seems essential for understanding properly the forebrain in causal molecular terms. Of course, subsequent elongation of the neural plate at late neurula

stages introduces bilateral symmetry for all brain regions caudal to the rostral-most forebrain (reviewed by [56,73]). The node and the non-neural ectoderm act as mutually opposed sources of prospective ventralizer and dorsalizer signals, respectively (Fig. 2C). The non-neural ectoderm is well known as a source of dorsalizing factors [45]. Substances diffusing from the

AMNIOTE FOREBRAIN DEVELOPMENT node generally have been analyzed either as general neural inducers, or as AP signals, but the circumpherential symmetry model proposed here implies that much of what was thought early median AP pattern may be interpreted as simultaneously representing DV pattern. It can be hypothetized that at least some nodal signals may have an early ventralizing role. In any case, node-derived cells are postulated as the subsequent sources of ventralizing effects. As gastrulation and then neurulation proceed (Fig. 2D), planar induction is increasingly complemented by vertical induction stemming from the node-derived axial mesoderm (prechordal plate and notochord). The postulated concentric rings of the early neural plate simultaneously elongate due to active stretching of the neural plate midline, as well as by clonal growth and cell intercalation laterally. Data on cellular clones reviewed by Rubenstein et al. [73] reveal that the primary rostral continuity of the initially ring-shaped concentric left and right neural domains is conserved during the elongation and cylindric closure of the neural primordium (Fig. 2D). Axial mesoderm underneath the oorplate (notochord and prechordal plate; Fig. 4) continues as a source of ventralizing signals (i.e., secretion of SHH protein, which produces vertical induction effects on the overlying neural tube oor). Non-neural ectoderm, neural crest and roof structures of the closed neural tube continue signaling as dorsalizers (i.e., secretion of BMPs growth factors and Wnt family proteins). Such knowledge, presented here only in minimal detail, represents the background for the essential concept of the brain longitudinal axis postulated in our prosomeric paradigm (reviewed in [56,60,76,77]. This paradigm embodies the assumption that these earliest planar and vertical neural induction processes, and secondary ventralizing and dorsalizing effects, dene primary longitudinal zones circling around the rostral midline of the neural plate. These zones specify in a denitive way the DV pattern of the developing forebrain (Figs. 2,3). Accordingly, the forebrain can be conceived as containing a median oor plate (caused by maximal ventralizing effect) restricted to median eminence, infundibulum, mammillary, retromammillary and prerubral areas (Fig. 4). The oorplate does not extend into the midrostral border of the plate (i.e., the anterior neuropore), since that foremost space is occupied by the other progressively more peripheral (topologically more dorsal) rings of neural tissue (Figs. 2D,3; see [76,77]). The clearest illustration of the oorplate according to this concept occurs in Xenopus, where it is distinct as a central neural plate area lacking expression of N-CAM, reproducing exactly the oorplate concept of the prosomeric model [16]. The basal plate (strongly ventralized tissue, but less than the oorplate) crosses the rostral forebrain midline immediately in front of the end of the oor longitudinal domain, at prospective retrochiasmatic (postoptic) level, and encompassing the eminentia media and infundibular hypothalamus (Figs. 35). The alar and roof plates also cross the midline more peripherally (Figs. 2D,3; [46,76]); the median alar plate includes the suprachiasmatic, chiasmatic, and lamina terminalis (median preoptic) prospective areas, whereas the median roof domain maps upon the anterior commissure [9,10,63,73]. Thus the topographically rostrocaudal sequence of structures at the median plane of the rostral forebrain is postulated to be topologically (and causally) comparable to the DV structure found elsewhere along the lateral wall of the whole tube (Figs. 2,3). Development of these median forebrain domains depends on the prechordal mesoderm and the cyclops gene, and failure in its formation leads to cyclopia and holoprosencephaly (one eye, one nose, and one telencephalon across the midline of the head). The foregoing account is now a well-supported assumption, which bears profoundly on the changing collective belief of what is AP versus DV in the forebrain, and maybe is acquiring the status of established fact, by sheer accumulation of corroborating evi-

701

FIG. 5. Horizontal section through prosomeres p2p6 in a rat embryo at E12 (routine aniline stain). Note constrictions building the interneuromeric boundaries and intervening neuromeric outpouchings. The p2/p3 limit forms later the zona limitans interthalamic boundary, between the dorsal and ventral thalamus. The caudal-most pole of the telencephalic vesicle is seen at left, whereas the optic stalk appears on the right side.

dence from many lines of research. Curiously, it is in this longitudinal aspect in which the prosomeric model has had it largest impact on the scientic community. What is the consequence of all this in morphogenesis? As development proceeds, unequal distribution of proliferative maxima occurs in the neural tube wall, consequent to DV secondary regionalization effects (dorsal parts proliferate more intensely than ventral parts). This causes the neural tube to bend progressively in the midsagittal plane, generating a marked rostroventral inexion called the cephalic exure, approximately at the caudal end of the forebrain (Fig. 4; note other exures appear elsewhere). This phenomenon is a purely mechanical effect, which does not have any visible impact on the ongoing molecular regionalization of the neural tube wall. This means that as the neural tube bends as a whole, the intrinsic longitudinal domains established by previous and ongoing DV patterning processes are not interrupted or changed in any major way, and consequently bend as well, according to their relative positions. This point has been corroborated expressly in zebrash, because the early established tract of the postoptic commissure (with associated neuronal populations) is formed before the cephalic exure appears in this species, and all this meshwork of neurons and bers becomes bent congruently as the neural tube bending occurs [3,26,32,71]. One concern here is that the formation of the cephalic exure progressively separates the midbrain and caudal forebrain oor from the underlying more rigid notochord (Fig. 4). It has not been investigated yet whether this alters the pattern of primary ventral domains. However, it is known that the early ventralizing effects entrain other interactions intrinsic to the neural wall (i.e., oor plate inducing and maintaining basal plate fate), which may contribute to the long-term continuity of these domains, independently

702 from their lost proximity to the notochord. This circumstance nevertheless may explain some common peculiarities of the midbrain and caudal forebrain oor and basal plate domains (i.e., origin of the ventral tegmental area and substantia nigra dopaminergic cell populations; [59,61,88]. Unfortunately, the myth exists that the cephalic exure is a transient feature, which supposedly would become straightened out later in development. It sufces that the reader examines midsagittal sections of any adult vertebrate brain to assess that this is never true (my experience here ranges from lamprey to man). During the 20th century that tradition led many morphologists following the example of Herrick [28] to accept a straight longitudinal axis of the brain, running from the olfactory bulb back into the medulla oblongata. This widely assumed axis is the primary reason why most published forebrain neuroanatomy appears in cross-sections. Topologically, these are not true transverse sections with regard to the real embryological, molecular, and causal (bent) longitudinal axis. Such sections actually are quite difcult to understand developmentally. The order of increasing difculty I recommend for students, and which I follow myself with any new brain I study, is (1) sagittal (to assess how the axis is bent), (2) horizontal (adapted to the sector of the bent axis which is of interest, to best detect AP subdivisions), and then, (3) transversal to the area of interest (to best detect DV subdivisions; see Fig. (4), and, only then, (4) cross-sections and published atlases (various sorts of obliqueness). The essentially simple error on the axis was strongly implemented in the assumptions of the columnar paradigm, and it has been burdening the progress of neuromorphology as an unnoticed dogma now for several decades [56]. This unacknowledged situation reaches the scientically scandalous point that major atlases are not consistent with current embryology on the limits of the major brain parts, and, therefore, on the origin of many brain formations [57]). However, there are signs we may be slowly recovering from this effect [20,52]. Note that different developmental stages of any species, and different species, will differ more or less importantly in the degree of their respective neural axial bending; this often makes crosscorrelation of cross-sections difcult and confusing. Neurogenetic gradients studied in the forebrain embryologic literature systematically refer data to such cross-sections. These are not aligned with the embryonic DV or AP dimensions, which provide the positional framework within which proliferation is regulated molecularly in the brain wall regions. The result is much loss of information, and in some cases straightforward absurdity [57]. We thus see how a change of paradigm affects the way in which a complex form as is the forebrain can come to be analyzed/ understood in quite different and mutually incompatible ways, and also how some of them can be deemed to be better than the others. Relationship to AP Patterning The DV patterning effects occurring similarly along the whole length of the neural tube likely establish the primary fundament (largely not yet understood) for neural metamery. Metamery is a special case of homology where separate units serially disposed along the length axis of an organism (as occurs with vertebrae and ribs) seem to be repeated in terms of shared fundamental morphological features. The transversal metameric units, called neuromeres in the brain (prosomeres in the prosencephalon; rhombomeres in the rhombencephalon), are structurally comparable or homologous one to another at least in terms of their common longitudinal zonation and related causal or consequent mechanisms. Structural homology is a complex comparative idea meaning basically that two forms are conceptually the same on the basis of topology and construction, irrespective of apparent differ-

PUELLES ences, even in genesis or function. In structural homology [40, 42,53], the site occupied by the relevant structure within the body-plan (Bauplan) and the mutual neighborhood relationships between its intrinsic fundamental structural components seem allimportant. The more modern biological homology concept in addition postulates the existence of some common developmental underpinnings (common processes or constraints, leading to common morphogenesis and morphostasis, that is, development and conservation of form). In addition, historical or taxonomic analysis of homology provides an evolutionary explanation to account for the common body plan and pattern of similarly developed fundamental components [2325]. Thus metameres or other nonmetameric homologous forms may look quite different, and even function diversely, and still be homologous to their partners in some aspect; consider, for instance, your nose and that of an elephant, or your upper and lower extremities. Note that a common DV structure in the neural tube establishes from the beginning a repeated fundamental structural pattern along the whole length axis of the neural tube. The only additional structural characteristic we need is a transversal delimitation of potentially serially homologous parts, sharing this DV structure. Such limits orthogonal to the bent axis actually start to appear at about the stage in which the telencephalic vesicles begin to evaginate (Figs. 2D,3,4). They are visible in sagittal and horizontal sections as constrictions where the proliferative activity of the neural tube wall is relatively reduced. These limits separate more strongly proliferating outpouchings of the forebrain wall, which may be identied as prosencephalic neuromeres or prosomeres (Fig. 5) [5,12,38,39,60,62,70,83,84,86]. The outpouchings are more apparent across the alar plate longitudinal domain, due to the higher proliferative activity there. Species with low proliferation rates tend not to show clearly the constrictions and outpouching phenomena of neuromery, even though the subjacent genetic specication and proliferative differences may exist [25,51,55,67,91]. Various sorts of study have analyzed the histogenetic phenomena occurring in the different prosomeres, noting that each neuromeric eld tends to develop its cohorts of basal and alar neurons in an independent and heterochronic pattern (i.e., [62,64]). The differential proliferative and neurogenetic behavior of prosomeres thus suggested underlying molecular causes. These started to accrue since the nineties in the form of genes coding for transcription factors. They turned out to be expressed in variously shaped spatial domains of the forebrain wall, usually respecting with their expression boundaries the interprosomeric limits and/or the limits between the longitudinal zones. Some genes preferentially relate to transverse limits, whereas others preferentially coincide with longitudinal zonal limits (Fig. 3; see for review [48,60; also, Puelles, Martnez, and Rubenstein, review in preparation). Such expression patterns conrmed the previous assumption that neuromeric alar or basal elds of the forebrain behave differentially in histogenesis because variously overlapping transcription factor signals combine into constellations specic for each DV and AP delimited eld. These diverse genetic contexts of transcription factors, plus diverse signals mediating intercellular communication, presumably play on available gene enhancers to select further genes to be activated or suppressed. In this way specic programs of histogenesis are initiated in each molecularly distinct part of the neural wall, which leads them to also become structurally distinct in the long run (Fig. 8A). The variety of gene patterns observed supports both the existence of signals that are common to given longitudinal domains along more or less extensive stretches of the neural tube (supporting the fundamental unifying role of DV longitudinal zones) and messages that establish more restricted positional identity at given longitudinal or segmental territories, or even

AMNIOTE FOREBRAIN DEVELOPMENT

703

FIG. 6. Schema redrawn from part of Fig. 4, indicating the main structural fates of the different alar or basal portions of the extratelencephalic forebrain wall, in the diencephalon and secondary prosencephalon (compare with Fig. 7).

generate smaller subdivisions within the primary alar or basal areas (Fig. 8A). THE FOREBRAIN IS PROGRESSIVELY REGIONALIZED The prosomeric model was postulated interactively with J. L. R. Rubenstein and other colleagues (notably M. Martinezde-la-Torre, S. Martnez, A. Bulfone, K. Shimamura, L. Medina, C. Redies, C. Verney, C. M. Trujillo, F. J. Milan, M. A. Pombal, M. Wulliman, J. C. Davila, S. Guirado, etc.) on the basis of diverse gene expression patterns and morphological correlation with heterochronic forebrain developmental elds. At initial neural tube stages, the forebrain is formed simply by the primary prosencephalon, a tagma placed rostral to the midbrain. One also may conceive that the midbrain enters into this early rostral tagma and is a caudal-most subunit specied differentially within it, as suggested, for instance, by a number of common histogenetic aspects [59,61, 88]. At later developmental stages, the prosomeric model divides the primary forebrain into two AP parts: the secondary prosencephalon and the diencephalon proper (Fig. 4). The main arguments for distinguishing these units are that the secondary prosencephalon is distinctly prechordal and participates in the formation of the eyes and the telencephalon (entities which react together in holoprosencephaly), while the diencephalon is epichordal and lies caudal to the telencephalon. Note that here the denition of the diencephalon differs notably from the usual textbook columnar formulation (and also from some previous neuromeric models), since the hypothalamus is excluded from the diencephalon and reinterpreted as the oor of the secondary prosencephalon. The diencephalic oor area is identied instead as the retromammillary and prerubral region of the tegmentum, at the apex of the cephalic exure (Fig. 4; columnar schemas regard these as parts of the midbrain; see discussion in [88]).

The Diencephalon The diencephalon proper subdivides early into three (or four) prosomeres, identied in caudorostral order as prosomeres 13 (p1p3, plus p4). Recently, we have started to think that the prosomere 4, previously thought to be part of the secondary prosencephalon, should be redened and assigned to the diencephalon (Fig. 6). As shown in Fig. 6, the alar eld of p1 is identical to the classic pretectum. Alar p2 contains the sum of dorsal thalamus and epithalamus (habenula and epiphysis). Alar p3 initially was thought to form the ventral thalamus by itself [60], but new data suggest that alar p4 participates with p3 in the mature ventral thalamus ([48]; Puelles, Stuhmer, and Rubenstein, unpublished observations). Moreover, the dorsally placed eminentia thalami area in p4 represents a longitudinal subunit best conceived as dorsal to ventral thalamus, as revealed by the course of the stria medullaris tract. Therefore its topography seems analogous to that of the epithalamus in p2, dorsal to the dorsal thalamus. Other observations leading to a changed denition of p4 come from more detailed analysis of the development and genetic make-up of the mammillary region (previously attributed to basal p4), which suggest it actually forms in basal/oor p5 (Fig. 4; Puelles, unpublished observations). The interthalamic p2/p3 transverse limit is very well-marked in most vertebrates, forming the so-called zona limitans interthalamica (a sort of glial palisade, with some intrinsic neurons, which widens and disappears as it approaches the basal plate; see Figs. 5 and 7). A number of the genes expressed in the basal plate (like shh and sim-1) typically show a spike of expression ascending dorsalward into the core of the zona limitans. We therefore conceive this important limit as an unexplained anomaly in DV patterning, such that the normal equilibrium line for basal versus alar fates becomes strongly displaced dorsalward (thus the spike of basal markers). Observations in zebrash embryos clearly show that the zona limitans develops progressively from an initial condition in which

704

PUELLES

FIG. 7. Artist drawing illustrating the complex tridimensional morphology of the mammalian forebrain. The mouse forebrain at stage E13.5 is partially reconstructed in perspective, emphasizing the topologic connectedness of various forebrain parts at the telencephalic stalk, seen here as the transition between the III ventricle (nearest the observer) and the lateral ventricle (note the connecting interventricular foramen [ivf]). Three section planes are represented: vertical stripes label a midline sagittal section across the hypothalamus, the preoptic region and the commissural plate, with schematic indication of the anterior and hippocampal commissures (ac,hc); oblique stripes mark a caudal transverse section plane through the pretectum (PT); horizontal stripes identify an intermediate horizontal section level, similar to that shown in Fig. 8A. The lateral ganglionic eminence (LGE) is represented in bulk (dark gray); the medial ganglionic eminence (MGE) (clearer gray and longitudinally striped) is dissected, leaving out the internal capsule bers, to show the stria terminalis tract and nucleus (BST), as well as the underlying globus pallidus (PAL). Both eminences bulge above the horizontal section plane, allowing the identication of a number of striatal, pallidal, and amygdaloid portions; however, they hide the claustrum and part of the cortex (CX). The bers of the internal capsule would pass from the diencephalon into the telencephalon across the MGE/LGE-locus marked IC, under the arc-shaped BST. For reference, we indicate the alar-basal boundary at the surface of the III ventricle (thick dash linenote the dorsalward deection at the zona limitans intrathalamica) and the topologically transverse boundaries between the prosomeres p1p6 (thin dash lines); this represents a slightly outdated prosomeric model in accord with present data, since we now believe the neurohypophysis enters in p6 and the mammillary area in p5 (compare Fig. 6). Simplications introduced for clarity in this drawing included not showing most of the medial wall of the hemisphere, or the choroidal roof formations.

the alar/basal boundary is straight, as schematically depicted in Fig. 4 [3]. Each of these p1-p4 territories is well-delimited molecularly

from early stages onwards in all vertebrates examined (i.e., Fig. 8A; [26,60]. Soon afterwards they start to show differential histogenetic patterns, with signs of mantle layer differentiation

AMNIOTE FOREBRAIN DEVELOPMENT

705

FIG. 8. (A) Horizontal section through the midbrain and all forebrain prosomeres in a mouse embryo at E13.5. This pseudocolor darkeld image shows the graphically superposed relative expression patterns of two genes, Dlx2 (in red) and Tbr1 (in green), determined by in situ hybridization with mRNA probes in adjacent sections. At this stage, the mantle layer of the forebrain is much more advanced than at the stage shown in Fig. 5 for the rat. Nevertheless, the image shows distinctly how different molecular markers largely respect the postulated interprosomeric boundaries, or other boundaries within the telencephalon (i.e., the pallio-subpallial boundary). This section lies roughly in the plane represented schematically from a different viewpoint in Fig. 7. (B) Schematic illustration of pallial and subpallial subdivisions of the telencephalon (lateral view at left; medial view at right), as determined by expression of the gene markers indicated; note the novel, Emx1-negative ventral pallium domain (according to data for amphibian, reptilian, avian, and mammalian embryos from Smith-Fernandez et al. [79] and Puelles et al. [66]). The pallio-subpallial boundary is drawn as a thick red dash line. Note that several subpallial and pallial domains enter caudally the amygdala complex and rostromedially the septum.

706 and more advanced molecular regionalization [62,64,68,92]. This implies genes appearing expressed only in smaller domains or layers, contained within the primitive simpler units. These restricted signals antedate and probably partly are a prerequisite for the appearance of specic nuclei with characteristic neuronal types in these prosomeric territories. While diversication is more protracted and important in the alar plate areas of the prosomeres (e.g., see [68,92]), the respective basal plate and even the underlying median oor plate areas also show analogous processes, triggered apparently by their own primary patterns of gene expression (e.g., [88]). The Secondary Prosencephalon (Excluding the Telencephalon) The secondary prosencephalon, on the other hand, was rst postulated to contain three prosomeres (p4 p6), but we conceive it now as composed only of p5 and p6, since p4 can be assigned to the diencephalon proper (see Figs. 4 7,8A). The morphological situation is more complex in this part of the forebrain, because here the telencephalic vesicle evaginates dorsolaterally (Fig. 1D), apparently across one (p5) or more (p4-p6) prosomeres, and this generates at its neck the difcult telencephalic stalk area sensu lato. Note that the optic vesicles evaginate also at earlier stages from a more ventral part of the alar plate of p6 or p5 (Figs. 2D,4,5), causing topological deformations of the surrounding preoptic and postoptic forebrain wall which are not well understood yet. A number of alternative options can be conceived as regards a detailed schema of these rostral-most neuromeric units and whether they all enter the telencephalon or not (see [6,7] for different interpretive options). Recently, experimental work with fate mapping in the chick has conrmed that the telencephalic subpallium, containing the basal ganglia, is formed topologically rostral to the pallial telencephalic components [9,10,73,79], against our earlier assumption that it was ventral to it [6]. However, it is not clear yet that the pallio-subpallial boundary represents a straightforward, transverse interprosomeric boundary, because it might be a secondary regional limit, or perhaps be due to a developmental regionalization process essentially unrelated to AP patterning of the neural tube, unique to the telencephalon (see [66,90]. In this case, its topology might be unrelated to either longitudinal or transversal primary limits imaginable within the segmental paradigm. The alar plate of at least p5 and p6 accordingly is conceived to possess two separately evaginated portions (the telencephalon and the eye), apart of an ample, non-evaginated, prethalamic region (Fig. 6). The distinction may be seen as didactic, but it is clear that both the telencephalic and eye developmental elds form distinct proliferative and histogenetic units, and necessitate accordingly separate analysis. The prethalamus, or extratelencephalic alar domain of the secondary prosencephalon, lies strictly dorsal topologically to the prosomeric hypothalamus. This region has been studied rather poorly in the past, given the problem sketched above on the confusing effect of the omnipresent cross-sections. Nevertheless, the prethalamic region clearly contains at least preoptic, supraoptic-paraventricular and peduncular areas, as well as some other areas habitually attributed to the suprachiasmatic, anterior, perifornical and dorsal parts of the hypothalamus (some of these areas are indicated in Fig. 6; see [38,39] on the heterogeneity of the classic hypothalamus). These domains approach dorsalward the hemispheric stalk, where we postulated a distinct anterior entopeduncular area (AEP in Fig. 8B; see also [6,60,66]. This has been later corroborated as the site where telencephalic oligodendrocyte precursors are generated [54,85] and is also distinct by its exclusive expression of the gene Shh in its ventricular zone. The

PUELLES basal and oor plate parts of p5 and p6 form the retrochiasmatic, median eminence, ventral, infundibular and posterior mammillary regions of the hypothalamus proper (Figs. 6,7; hypothalamus as originally dened by His [29 31]). The Telencephalon Recently we reanalyzed the fundamental molecular and structural constituents of the telencephalon in mouse and chicken embryos [65,66] by means of several genetic markers (transcription factors) known to be causally relevant for the normal development of the telencephalon or of some of its pallial or subpallial parts (i.e., Figs. 8A,B). One major conclusion was that the palliosubpallial boundary, dened molecularly by the interface between Dlx family genes expressed in the subpallium and Tbr1 and other genes expressed in the pallium, runs all the way across the telencephalon, stretching from the amygdala, caudally, to the septum, rostromedially. This means that both amygdala and septum possess pallial and subpallial portions comparable to central telencephalic parts (cortex, claustrum, basal ganglia) in terms of primary genetic codes, independently of their secondarily specialized development (Fig. 9). The subpallium can be divided into three parallel zones, identied as anterior entopeduncular area (AEP), pallidum and striatum, respectively (Fig. 8B). While all these areas express Dlx genes [17], only AEP and pallidum express additionally Nkx2.1, and only AEP expresses Shh in its ventricular zone [66,77]. These molecular/structural subdivisions can be traced as well into the amygdala and the septum (Fig. 8B). In contrast to this evidence for common aspects of subpallial derivatives across the telencephalon, we lack markers which distinguish the diverse specializations (i.e., distinguish caudato-putamen from the nucleus accumbens and both of them from the central amygdaloid nucleus, all of which belong to the striatal subzone). Insofar as numerous anatomical, hodological, and chemoarchitectonic data validate the distinction of such parts, we may predict that such differential genetic determinants shall be found in the near future. As regards the pallium, our analysis [65,66] concluded that four parallel subdivisions can be postulated: the medial, dorsal, lateral, and ventral pallial portions (Figs. 8B,9). This implies one subdivision more than those in other pallial schemas, namely our ventral pallium. This was postulated on the basis of differential early expression of the gene Emx1, which is generally present in the other parts of the pallium, but is not expressed in the pallial region closest to the pallio-subpallial boundary (Fig. 8B). Our results in mouse and chick embryos corroborated here the work in these species of Smith-Fernandez et al. [79], who also demonstrated the same pattern in the turtle and frog telencephalon. This suggests general validity of the four-part pallial model for tetrapods [66]. EVOLUTIONARY RELATIONSHIPS WITHIN THE FOUR-PART PALLIAL MODEL OF THE TELENCEPHALON The four-part pallial model seems advantageous to understand the evolution of the reptilian and avian telencephalon, in which the lateral and ventral pallial parts contribute importantly to the socalled dorsal ventricular ridge (DVR; [65,66,69]). This is a voluminous pallial area intercalated between the conventional cortex and the striatum, which protrudes ventricularly (while the underlying subpallium does not; Fig. 9). Although the DVR receives afferents from diverse sensory nuclei in the dorsal thalamus, its structure is not cortical. It has been a matter of long dispute what is the morphological and functional meaning/equivalence of the sauropsidian DVR, as compared to mammalian telencephalic

AMNIOTE FOREBRAIN DEVELOPMENT

707

FIG. 9. Drawings comparing the differential histogenesis of the same pallial and subpallial domains in the mammalian and avian telencephalon, suggesting possible eld homologies (i.e., derivatives of the same pallial or subpallial domain are postulated to be homologous as a eld, irrespective of the numerous differences which might be detected). Subpallial subdivisions in the septum are not represented. The ventricular cavity is lled in black. The upper two drawings show a oblique caudolateral view into the rostral telencephalon of either a mammal or a bird; the olfactory bulb extends away from the observer and the lateral olfactory tract appears cut halfway along its subpial course along the ventral pallium (neostriatum [N] in birds; area much reduced into the migrated ventromedial claustrum in mammals). Note a conserved overall topology, but differential growth pattern and migration of the deep masses of the ventral and lateral pallial regions. The lower two drawings show the opposite perspective view obliquely caudalwardsinto the caudal telencephalic pole; the compared mammalian or avian amygdaloid regions show corresponding pallial or subpallial molecular characteristics -compare Fig. 8B, but differential growth and migration extent. Avian DVR cell masses (LP, VP) bulge massively in the lateral ventricle, whereas corresponding mammalian cell masses migrate far from the ventricle and aggregate close to the brain surface. This schema updates that given by Puelles et al. [66], because we now believe on the basis of further Emx1 mapping studies that the mammalian basomedial amygdaloid nucleus belongs to the ventral pallium, instead of to the striatal domain, as thought previously. The status of the medial amygdala is also being revised. (probably also pallial, at least in part).

parts. The prevalent viewpoint during the last 30 years has been Kartens conception of an evolutionary migratory transformation of the DVR sauropsidian centers into site-specic layers of the mammalian isocortex (reviewed by [58,80,81]; see also 3537]). We think that the new four-part pallial model opens new ways of thought for understanding these areas, their divergent evolution in vertebrates, and what may be their mammalian homologs. Puelles et al. [66] proposed that the DVR homolog in the mammalian brain should be sought in the claustroamygdaloid complex (Fig. 9), as suggested earlier by Holmgren [33]. A review dealing with related issues concerning evolution of thalamic centers and the thalamopallial connectivity is in press [58].

Our line of thought suggests that ancestral tetrapods already possessed these four pallial parts, as indicated by their presence in frogs [5a,79]. While sauropsidians developed enormously the ventral and lateral pallial parts (thus the DVR, particularly large in crocodiles and birds), mammals apparently diverged early on, developing a limited DVR, and tended to transfer their DVR derivatives using radial migration (a general evolutionary trend) into positions supercial to the internal capsule, building various parts of the claustroamygdaloid complex and the olfacto-amygdaloid cortex (no periventricular DVR derivatives, thus no ventricular bulge; Fig. 9). Simultaneously, mammals greatly evolved their striatum (striatal bulge into the ventricle) and their dorsal pallium.

708 The latter process clearly led to an important relative surface increment of the telencephalic cortex, as well as to the appearance of the six-layered isocortex. Parallel changes in the medial pallium led to the appearance of the mammalian hippocampus, dentate gyrus, and parahippocampal cortex; these areas also nd their apparent homologues in the submammalian medial pallium. Recent work has shown that pallial-derived neurons, particularly pyramidal neurons, are glutamatergic, while subpallial neurons are in most cases GABAergic [50]. Inhibitory interneurons in the mammalian cortex do not arise in the pallium, but originate in the subpallium and later migrate tangentially [1,2,14,44,82,89]. This process is not restricted to mammals, because essentially the same palliopetal translocation of inhibitory neurons was found in the chick [9 11] and may occur as well in Xenopus [5a] and lamprey (Puelles et al., submitted). CONCLUDING REMARKS This review of the structural and molecular underpinnings of the prosomeric model illustrate several key features that are of conceptual importance in contemplating forebrain evolution. 1. Conservation of fundamental patterning processes creating an early DV zonation along the longitudinal axis. Of great conceptual importance is the fact that these zones circumnavigate the node, thus extending primarily around the anterior limit of the forebrain. 2. Divergence in mechanical transformations of the longitudinal axis, creating different degrees of cephalic exure in different species. This can generate considerable confusion in comparative studies and necessitates a careful attention to morphogenetic detail and comparable sectioning planes in such studies. 3. Segmentation/metamerization as a common theme for primary AP regionalization of the neuraxis, including forebrain regions, across all longitudinal zones. Ulterior, more local regionalization processes, combined with species-specic radial and tangential migratory activity of neuronal populations contributes to nal complexity of neural wall. 4. Conservation of underlying prosomeric and longitudinal organization, combined with divergence in the histogenetic elaboration of specic regions (see the four-pallial concept and differential development of the constituent portions in mammals versus other amniotes).ACKNOWLEDGEMENTS

PUELLES5a. Brox, A.; Ferreiro, B.; Puelles, L.; Medina, L. Observations on the telencephalon of Xenopus based on calretinin immunostaining and gene expression patterns. Brain Res. Bull. In press. 6. Bulfone, A.; Puelles, L.; Porteus, M. H.; Frohman, M. A.; Martin, G. R.; Rubenstein, J. L. R. Spatially restricted expression of Dlx-1, Dlx-2 (Tes-1), Gbx-2, and Wnt-3 in the embryonic day 12.5 mouse forebrain denes potential transverse and longitudinal segmental boundaries. J. Neurosci. 13:31553172; 1993. 7. Bulfone, A.; Smiga, S. M.; Shimamura, K.; Puelles, L.; Peterson, A.; Rubenstein. J. L. R. T-Brain-1 (Tbr-1): A homologue of Brachyury whose expression denes molecularly distinct domains within the cerebral cortex. Neuron 15:6378; 1995. 8. Cambronero, F.; Puelles, L. Rostrocaudal nuclear relationships in the avian medulla oblongata: Fate-map with quail-chick chimeras. J. Comp. Neurol. 427:522545; 2000. 9. Cobos, I. Morfogenesis e histogenesis del telencefalo: Estudio exper imental en embriones de pollo. Doctoral dissertation, University of Murcia, Spain; 2000. 10. Cobos, I.; Shimamura, K.; Rubenstein, J. L. R.; Martnez, S.; Puelles, L. Fate map of the avian anterior forebrain at the 4 somite stage, based on the analysis of quail-chick chimeras. Devel. Biol. In press. 11. Cobos, I.; Puelles, L.; Martnez, S. The avian telencephalic subpallium originates tangentially migrating inhibitory neurons that invade the dorsal ventricular ridge and the cortical areas. Devel. Biol. In press. 12. Coggeshall, R. E. A study of diencephalic development in the albino rat. J. Comp. Neurol.122:241269; 1964. 13. Dale, J. K.; Vesque, C.; Lints, T. J.; Sampath, T. K.; Furley, A.; Dodd, J.; Placzek, M. Cooperation of BMP7 and SHH in the induction of forebrain ventral midline cells by prechordal mesoderm. Cell 90:257 269; 1997. 14. de Carlos, J. A.; Lopez-Mascaraque, L.; Valverde, F. Dynamics of cell migration from the lateral ganglionic eminence in the rat. J. Neurosci. 16:6146 6156; 1996. 15. Doniach, T.; Phillips, C. R.; Gerhart, J. C. Planar induction of anteroposterior pattern in the developing central nervous system of Xenopus laevis. Science 257:542545; 1992. 16. Eagleson, G. W.; Ferreiro, B.; Harris, W. A. Fate of the anterior neural ridge and the morphogenesis of the Xenopus forebrain. J. Neurobiol. 28:146 158; 1995. 17. Eisenstat, D.; Liu, J. K.; Mione, M.; Zhong, W.; Yu, G.; Anderson, S. A.; Ghattas, I.; Puelles, L.; Rubenstein, J. L. R. Dlx-1, Dlx-2 and Dlx-5 expression dene distinct stages of basal forebrain differentiation. J. Comp. Neurol. 414:217237; 1999. 18. Figdor, M. C.; Stern, C. D. Segmental organization of the embryonic diencephalon. Nature 363:630 634; 1993. 19. Foley, A. C.; Skromme, I.; Stern, C. D. Reconciling different models of forebrain induction and patterning: A dual role for the hypoblast. Development 127:3839 3854; 2000. 20. Foster, G. A. Chemical neuroanatomy of the prenatal rat brain. A developmental atlas. Oxford: Oxford University Press; 1998. 21. Gilbert, S. F. Developmental biology, 6th ed. Sunderland, MA: Sinauer Associates; 2000. 22. Gribnau, A. A. M.; Geijsberts, L. G. M. Morphogenesis of the brain in staged Rhesus monkey embryos. Adv. Anat. Embryol. Cell Biol. 91:1 69; 1985. 23. Hall, B. K. Evolutionary developmental biology. London: Chapman & Hall; 1992. 24. Hall, B. K., ed. Homology: The hierarchical basis of comparative biology. San Diego: Academic Press; 1994. 25. Hall, B. K. Homology. Novartis Foundation Symposium 222. Chichester: Wiley; 1999. 26. Hauptmann, G.; Gerster, T. Regulatory gene expression patterns reveal transverse and longitudinal subdivisions of the embryonic zebrash forebrain. Mech. Dev. 91:105118; 2000. 27. Herrick, C. J. Morphogenesis of the brain. J. Morphol. 54:233258; 1933. 28. Herrick, C. J. The brain of the tiger salamander, Amblystoma tigrinum. Chicago: The University of Chicago Press; 1948. 29. His, W. Uber das frontale Ende des Gehirnrohres. Arch. Anat. Physiol. Anat. Abst. 3/4:157171; 1893. 30. His, W. Vorschlage zur Eintheilung des Gehirns. Arch. Anat. Physiol. Anat. Abst. 3/4:173179; 1893.

Work supported by DGICYT grant PB98-0397, SENECA Foundation grant PB97-FS14; and European Community contract BIO4-CT96-0042.

REFERENCES1. Anderson, S. A.; Eisenstat, D. D.; Shi, L.; Rubenstein, J. L. Interneuron migration from basal forebrain to neocortex: Dependence on Dlx genes. Science 278:474 476; 1997. 2. Anderson, S. A.; Marn, O.; Horn, C.; Jennings, K.; Rubenstein, J. L. R. Distinct cortical migrations from the medial and lateral ganglionic eminences. Development 128:353363; 2001. 3. Barth, K. A.; Wilson, S. W. Expression of zebrash nk2.2 is inuenced by sonic hedgehog/vertebrate hedgehog-1 and demarcates a zone of neuronal differentiation in the embryonic forebrain. Development 121: 17551768; 1995. 4. Bergquist, H. Zur Morphologie des Zwischenhirns bei niederen Vertebraten. Acta Zool. 13:57303; 1932. 5. Bergquist, H.; Kallen, B. Notes on the early histogenesis and morpho genesis of the central nervous system in vertebrates. J. Comp. Neurol. 100:627 659; 1954.

AMNIOTE FOREBRAIN DEVELOPMENT31. His, W. Die Entwicklung des menschlichen Gehirns wahrend der ersten Monate. Hirzel, Leipzig; 1904. 32. Hjorth, J. T.; Key, B. Are pioneer axons guided by regulatory gene expression domains in the zebrash forebrain? High-resolution analysis of the patterning of the zebrash brain during axon tract formation. Dev. Biol. 229:271286; 2001. 33. Holmgren, N. Points of view concerning forebrain morphology in higher vertebrates. Acta Zool. (Stockholm) 6:413 477; 1925. 34. Inoue, T.; Nakamura, S.; Osumi, N. Fate mapping of the mouse prosencephalic neural plate. Dev. Biol. 219:373383; 2000. 35. Karten, H. J. The organization of the avian telencephalon and some speculations on the phylogeny of the amniote telencephalon. Ann. N.Y. Acad. Sci. 167:164 179; 1969. 36. Karten, H. J. Homology and evolutionary origins of the neocortex. Brain Behav. Evol. 38:264 272; 1991. 37. Karten, H. J. Evolutionary developmental biology meets the brain: The origins of the mammalian cortex. Proc. Natl. Acad. Sci. USA 94: 2800 2804; 1997. 38. Keyser, A. The development of the diencephalon of the Chinese hamster. Acta Anat. 83(suppl. 59):1181; 1972. 39. Keyser, A. Development of the hypothalamus in mammals: An investigation of its morphological position during ontogenesis. In: Morgane, P. J.; Panksepp, J., eds. Handbook of the hypothalamus. New York: Marcel Dekker; 1979:65136. 40. Kuhlenbeck, H. The central nervous system of vertebrates. Vol. 1: Propaedeutics to comparative neurology. Basel: Karger; 1967:159 298. 41. Kuhlenbeck, H. The central nervous system of vertebrates. Vol. 3, part II: Overall morphological pattern. Basel: Karger; 1973. 42. Kuhlenbeck, H. The central nervous system of vertebrates. Vol. 5, part I: Derivatives of the prosencephalon: Diencephalon and telencephalon. XIII. The telencephalon (birds). Basel: Karger; 1977:650 686. 43. Kupffer von, K. Die morphogenie des centralnervensystems. In: Hertwig, O., ed. Handbuch der vergleichenden und experimentellen Entwicklungslehere der Wirbeltiere. Vol. 2, part III. Fischer: Jena; 1906:1272. 44. Lavdas, A. A.; Grigoriou, M.; Pachnis, V.; Parnavelas, J. G. The medial ganglionic eminence gives rise to a population of early neurons in the developing cerebral cortex. J. Neurosci. 19:78817888; 1999. 45. Lee, K. J.; Jessell, T. M. The specication of dorsal cell fates in the vertebrate central nervous system. Ann. Rev. Neurosci. 22:261294; 1999. 46. Marcus, R. C.; Shimamura, K.; Sretavan, D.; Lai, E.; Rubenstein, J. L. R.; Mason, C. A. Domains of regulatory gene expression and the developing optic chiasm: Correspondence with retinal axon paths and candidate signaling cells. J. Comp. Neurol. 403:346 358; 1999. 47. Marn, F.; Puelles, L. Morphological fate of rhombomeres in quail/ chick chimeras: A segmental analysis of hindbrain nuclei. Eur. J. Neurosci. 7:1714 1738; 1995. 48. Martnez, S.; Puelles, L. Neurogenetic compartments of the mouse diencephalon and some characteristic gene expression patterns. In: Gofnet, A. M.; Rakic, P., eds. Results and problems in cell differentiation, vol. 30. The mouse brain development. Berlin/Heidelberg: Springer Verlag; 2000:91106. 49. Martnez, S.; Puelles, L.; Alvarado-Mallart, R. M. Tangential neuronal migration in the avian tectum: Cell type identication and mapping of regional differences with quail/chick homotopic transplants. Dev. Brain Res. 66:153163; 1992. 50. Mione, M. C.; Danevic, C.; Boardman, P.; Harris, B.; Parnavelas, J. G. Lineage analysis reveals neurotransmitter (GABA and glutamate) but not calcium-binding protein heterogeneity in clonally related cortical neurons. J. Neurosci. 14:107123; 1994. 51. Milan, J.; Puelles, L. Calretinin, calbindin and tyrosine-hydroxylase immunoreactivity patterns are consistent with the segmental map of the frog diencephalon. J. Comp. Neurol. 419:96 121; 2000. 52. Nieuwenhuys, R. Morphogenesis and general structure. In: Nieuwenhuys, R.; ten Donkelaar, H. J.; Nicholson, C. The central nervous system of vertebrates, vol. 1. Berlin: Springer; 1998:159 228. 53. Owen, R. On the archetype and homologies of the vertebrate skeleton. London: John van Voorst; 1948. 54. Perez-Villegas, E. M.; Olivier, C.; Spassky, N.; Poncet, C.; Cochard, P.; Zalc, B.; Thomas, J.-L.; Martnez, S. Early specication of oligo

709dendrocytes in the chick embryonic brain. Dev. Biol. 216:98 113; 1999. Pombal, M. A.; Puelles, L. A prosomeric map of the lamprey forebrain based on calretinin immunocytochemistry, Nissl stain and ancillary markers. J. Comp. Neurol. 414:391 422; 1999. Puelles, L. A segmental morphological paradigm for understanding vertebrate forebrains. Brain Behav. Evol. 46:319 337; 1995. Puelles, L. Atlas of prenatal rat brain development (book review). TINS 19:116 117; 1996. Puelles, L. Thoughts on the development, structure and evolution of the mammalian and avian telencephalic pallium. Philos. Trans. R. Soc. Ser. B Lond. In press. Puelles, L.; Medina, L. Development of neurons expressing tyrosine hydroxylase and dopamine in the chicken brain: A comparative segmental analysis. In: Reiner, A.; Smeets, W. J. A. J., eds. Phylogeny and development of catecholamine systems in the CNS of vertebrates. Cambridge: Cambridge University Press; 1994:381 406. Puelles, L.; Rubenstein, J. L. R. Expression patterns of homeobox and other putative regulatory genes in the embryonic mouse forebrain suggest a neuromeric organization. TINS 16:472 479; 1993. Puelles, L.; Verney, C. Early neuromeric distribution of tyrosinehydroxylase and dopamine- -hydroxylase immunoreactive neurons in human embryos. J. Comp. Neurol. 394:283308; 1998. Puelles, L.; Amat, J. A.; Martnez de la Torre, M. Segment-related, mosaic neurogenetic pattern in the forebrain and mesencephalon of early chick embryos. I. Topography of AChE-positive neuroblasts up to stage HH18. J. Comp. Neurol. 266:147268; 1987. Puelles, L.; Domenech-Ratto, G.; Martnez-de-la-Torre, M. Location of the rostral end of the longitudinal brain axis: Review of an old topic in the light of marking experiments on the closing rostral neuropore. J. Morphol. 194:163171; 1987. Puelles, L.; Guillen, M.; Martnez-de-la-Torre, M. Observations on the fate of nucleus supercialis magnocellularis of Rendahl in the avian diencephalon, bearing on the organization and nomenclature of neighboring retinorecipient nuclei. Anat. Embryol. 183:221233; 1991. Puelles, L.; Kuwana, E.; Puelles, E.; Rubenstein, J. L. R. Comparison of the mammalian and avian telencephalon from the perspective of gene expression data. Eur. J. Morphol. 37:139 150; 1999. Puelles, L.; Kuwana, E.; Puelles, E.; Bulfone, A.; Shimamura, K.; Keleher, J.; Smiga, S.; Rubenstein, J. L. R. Pallial and subpallial derivatives in the embryonic chick and mouse telencephalon, traced by the expression of the genes Dlx-2, Emx-1, Nkx-2.1, Pax-6 and Tbr-1. J. Comp. Neurol. 424:409 438; 2000. Puelles, L.; Milan, F. J.; Martnez-de-la-Torre, M. A segmental map of subdivisions in the diencephalon of the frog Rana perezi: Acetylcholinesterase-histochemical observations. Brain Behav. Evol. 47:279 310; 1996. Redies, C.; Ast, M.; Nakagawa, S.; Takeichi, M.; Martnez-de-la Torre, M.; Puelles, L. Morphologic fate of diencephalic prosomeres and their subdivisions revealed by mapping cadherin expression. J. Comp. Neurol. 421:481514; 2000. Redies, C.; Medina, L.; Puelles, L. Cadherin expression by embryonic subdivisions and derived gray matter structures in the telencephalon of the chicken. J. Comp. Neurol. In press. Rendahl, H. Embryologische und morphologische Studien uber das Zwischenhirn beim Huhn. Acta Zool. Stockh. 5:241344; 1924. Ross, L. S.; Parrett, T.; Easter, S. S. Jr. Axonogenesis and morphogenesis in the embryonic zebrash brain. J. Neurosci. 12:467 482; 1992. Rubenstein, J. L. R.; Martnez, S.; Shimamura, K.; Puelles, L. The prosomeric model: A proposal for the organization of the embryonic forebrain. Science 266:578 560; 1994. Rubenstein, J. L. R.; Shimamura, K.; Martnez, S.; Puelles, L. Region alization of the prosencephalic neural plate. Annu. Rev. Neurosci. 21:445 478; 1998. Sanes, D. H.; Reh, T. A.; Harris, W. A. Development of the nervous system. San Diego: Academic Press; 2000. Shimamura, K.; Rubenstein, J. L. R. Inductive interactions direct early regionalization of the mouse brain. Development 124:2709 2718; 1997. Shimamura, K.; Hartigan, D. J.; Martnez, S.; Puelles, L.; Rubenstein,

55.

56. 57. 58.

59.

60.

61.

62.

63.

64.

65.

66.

67.

68.

69.

70. 71.

72.

73.

74. 75.

76.

710J. L. R. Longitudinal organization of the anterior neural plate and neural tube. Development 121:39233933; 1995. Shimamura, K.; Martnez, S.; Puelles, L.; Rubenstein, J. L. R. Patterns of gene expression in the neural plate and neural tube subdivide the embryonic forebrain into transverse and longitudinal domains. Dev. Neurosci. 19:88 96; 1997. Slack, J. Essential developmental biology. Oxford: Blackwell; 2001. Smith-Fernandez, A.; Pieau, C.; Reperant, J.; Boncinelli, E.; Was sef, M. Expression of the Emx-1 and Dlx-1 homeobox genes dene three molecularly distinct domains in the telencephalon of mouse, chick, turtle and frog embryos: Implications for the evolution of telencephalic subdivisions in amniotes. Development 125:2099 2111; 1998. Striedter, G. F. The telencephalon of tetrapods in evolution. Brain Behav. Evol. 49:179 213; 1997. Striedter, G. F. Progress in the study of brain evolution: From speculative theories to testable hypotheses. Anat. Rec. (New Anat.) 253: 105112; 1998. Tamamaki, N.; Fujimori, E.; Takauji, R. Origin and route of tangentially migrating neurons in the developing neocortical intermediate zone. J. Neurosci. 17:8313 8323; 1997. Tello, J. F. Les differenciations neuronales dans lembryon du poulet, pendant les premiers jours de lincubation. Trab. Lab. Rech. Biol. Univ. Madrid 21:193; 1923. Tello, J. F. Les differenciations neurobrillaires dans le prosencephale de la souris de 4 a 15 millimetres. Trab. Lab. Rech. Biol. Univ. Madrid 29:339 395; 1934. Timsit, S.; Martnez, S.; Allinquant, B.; Peyron, F.; Puelles, L.; Zalc,

PUELLESB. Oligodendrocytes originate in a restricted zone of the embryonic ventral neural tube dened by DM-20 mRNA expression. J. Neurosci. 15:10121024; 1995. Vaage, S. The segmentation of the primitive neural tube in chick embryos (Gallus domesticus). A morphological, histochemical and autoradiographical investigation. Ergebn. Anat. Entwgesch. 41:1 88; 1969. Van de Peer, V.; Baldauf, S. L.; Doolittle, W. F.; Meyer, A. An updated and comprehensive rRNA phylogeny of (crown) eukaryotes based on rate-calibrated evolutionary distances. J. Mol. Evol. 51:565 576; 2000. Verney, C.; Zecevic, N.; Puelles, L. Structure of longitudinal brain zones that originate the substantia nigra and ventral tegmental area in human embryos, as revealed by cytoarchitecture and tyrosine-hydroxylase-, calbindin-, calretinin- and GABA-immunoreaction. J. Comp. Neurol. 429:22 44; 2001. Wichterle, H.; Garcia-Verdugo, J. M.; Herrera, D. G.; Alvarez-Buylla, A. Young neurons from medial ganglionic eminence disperse in adult and embryonic brain. Nat. Neurosci. 2:461 466; 1999. Wilson, S. W.; Rubenstein, J. L. R. Induction and dorsoventral patterning of the telencephalon. Neuron 28:641 651; 2000. Wullimann, M. F.; Puelles, L. Postembryonic neural proliferation in the zebrash forebrain and its relationship to prosomeric domains. Anat. Embryol. 329:329 348; 1999. Yoon, M.-S.; Puelles, L.; Redies, C. Formation of cadherin-expressing brain nuclei in diencephalic alar plate divisions. J. Comp. Neurol. 421:461 480; 2000.

77.

86.

78. 79.

87.

88.

80. 81. 82. 83. 84. 85.

89.

90. 91.

92.