THE JOURNAL OF COMPARATIVE NEUROLOGY 310:571-592 (1991)

Appearance of Putative Amino Acid Neurotransmitters During Differentiation of Neurons in Embryonic Turtle Cerebral Cortex MARK G. BLANTON AND ARNOLD R. KRIEGSTEIN Stanford University Medical Center, Stanford, California 94305

ABSTRACT Pyramidal and nonpyramidal neurons can be recognized early in the development of the cerebral cortex in both reptiles and mammals, and the neurotransmitters likely utilized by these cells, glutamate and gamma-aminobutyric acid, or GABA, have been suggested to play critical developmental roles. Information concerning the timing and topography of neurotransmitter synthesis by specific classes of cortical neurons is important for understanding developmental roles of neurotransmitters and for identifying potential zones of neurotransmitter action in the developing brain. We therefore analyzed the appearance of GABA and glutamate in the cerebral cortex of embryonic turtles using polyclonal antisera raised against GABA and glutamate. Neuronal subtypes become immunoreactive for the putative amino acid neurotransmitters GABA and glutamate early in the embryonic development of turtle cerebral cortex, with nonpyramidal cells immunoreactive for GABA and pyramidal cells immunoreactive for glutamate. The results of controls strongly suggest that the immunocytochemical staining in tissue sections by the GABA and glutamate antisera corresponds to fixed endogenous GABA and glutamate. Horizontally oriented cells in the early marginal zone (stages 15-16) that are GABA-immunoreactive (GABA-IR) resemble nonpyramidal cells in morphology and distribution. GABA-IR neurons exhibit increasingly diverse morphologies and become distributed in all cortical layers as the cortex matures. Glutamate-immunoreactive (Glu-IR) cells dominate the cellular layer throughout development and are also common in the subcellular layer at early stages, a distribution like that of pyramidal neurons and distinct from that of GABA-IR nonpyramidal cells. The early organization of embryonic turtle cortex in reptiles resembles that of embryonic mammalian cortex, and the immunocytochemical results underline several shared as well as distinguishing features. Early GABA-IR nonpyramidal cells flank the developing cortical plate, composed primarily of pyramidal cells, shown here to be Glu-IR. The earliest GABA-IR cells in turtles likely correspond to Cajal-Retzius cells, a ubiquitous and precocious cell type in vertebrate cortex. Glutamate-IR projection neurons in vertebrates may also be related. The distinctly different topographies of GABA and glutamate containing cells in reptiles and mammals indicate that even if the basic amino acid transmitter-containing cell types are conserved in higher vertebrates, the local interactions mediated by these transmitters may differ. The potential role of GABA and glutamate in nonsynaptic interactions early in cortical development is reinforced by the precocious expression of these neurotransmitters in turtles, well before they are required for synaptic transmission. The glutamate expression observed here provides a potential source of agonist for the spontaneous activation of glutamate receptors recently reported for embryonic cortical plate neurons (Blanton et al., '90: Proc. Natl. Acad. Sci. USA 87r8027-8030). The early emergence of amino acid neurotransmitter expression detailed in this study indicates a potential role for GABA- and glutamate-mediated developmental interactions in embryonic brain. Key words: pyramidal and nonpyramidal neurons, glutamate, GABA, cortical development

Accepted May 1,1991 D

1991 WILEY-LISS, INC.

M.G. BLANTON AND A.R. KRIEGSTEIN

572

In the mature cerebral cortex of reptiles and mammals, pyramidal and nonpyramidal cells differ physiologically (McCormick et al., '85; Connors and Kriegstein, '86; Kriegstein and Connors, '86) as well as morphologically. These cell classes also differ in their amino acid neurotransmitter content, with pyramidal cells utilizing excitatory amino acids (EMS, e.g., glutamate, aspartate, and related molecules, see Fagg and Foster, '83; Streit, '84; Ottersen and Storm-Mathisen, '84a; Conti et al., '89) and nonpyramidal cells utilizing gamma-aminobutyric acid (GABA, see Ribak, '78, and for review, Houser et al., '84). Neurotransmitter phenotypes are critical in determining the specific roles, whether excitatory or inhibitory, of these cell classes, both in synaptic signalling and in maintaining morphological integrity in the cortex (Mattson, '88). Imbalance of excitatory and inhibitory influences can compromise cortical function, leading to abnormal synaptic signallingand epileptogenesis (Traub et al., '87) and to morphological alterations, with neuronal circuit degeneration (Mattson and Kater, '89). Pyramidal and nonpyramidal neurons can be recognized early in the development of the cerebral cortex in both reptiles and mammals (Valverde et al., '89; Miller, '88; Goffinet, '83; see companion paper). The neurotransmitters utilized by the adult counterparts of these cells, glutamate and GABA, are also present early (Van Eden et al., '89; Erdo and Wolff, '89). These neurotransmitters have been suggested to play critical developmental roles, regulating neuronal survival (Balazs et al., '88; Brenneman et al., 'go), biochemical and cellular differentiation (Aruffo et al., '87; Meier et al., '87; Moran and Patel, '89; see Redburn and Schousboe, '87, and McDonald and Johnston, '90, for review), and patterns of neurite outgrowth (Brewer and Cotman, '89; Pearce et al., '87; Mattson et al., '88; see Mattson, '88, for review). Glutamate and GABA appear to have different trophic roles in embryonic life (Mattson and Kater, '89). Information concerning the timing and topography of amino acid neurotransmitter phenotype expression by specific classes of cortical neurons is important for understanding potential developmentalroles of neurotransmitters and for identifyingpotential zones of neurotransmitter interaction in the developingbrain.

To determine when morphologically distinct neuronal types in the cerebral cortex begin to express neurotransmitters, we analyzed the appearance of GABA and glutamate in the dorsal cerebral cortex of embryonic turtles. The cerebral cortex of turtles was chosen for study because its relative structural simplicity (Blanton et al., '89) and resistance to anoxia (Belkin, '63) facilitate analysis of its morphological and physiological development, and also because of its proposed evolutionary relationship to mammalian cortex (Ebner, '76; Northcutt, '81; Desan, '88). Data on events in the differentiation of turtle and mammalian cortex are important for understanding not only how these different cortical structures emerge during development, but possibly also how these structures emerged in evolution. The appearance of neurotransmitter phenotypes in turtle dorsal cerebral cortex was studied using polyclonal antisera raised against GABA and glutamate coupled to carrier proteins, prevnously characterized in mature mammalian neocortex (Storm-Mathisenet al., '83; Ottersen and StormMathisen, '84b). Similar antisera have been used to demonstrate the precocious appearance of GABA-utilizing neurons in mammalian neocortex (Lauder et al., '86; Van Eden et al., '891, but immunocytochemical data on the appearance of glutamate-utilizingneurons is unavailable for vertebrate cerebral cortex. A brief report of some of the GABA immunohistochemical data has appeared (Kriegstein et al., '88).

METHODS Animals Embryonic turtles were obtained from Tangi Turtles (Vidalia,LA) and incubated in ovo at 25-30°C as described previously (see companion paper). Embryos were staged by the external morphological criteria of Yntema ('68). Hatchlings (within 2 days post-hatch) and adults (15-20 cm carapace) were obtained from the same supplier. A summary of animals used appears in Table 1.

Tissue preparation Selected embryos were removed from eggs and anesthetized with hypothermia, and older animals were anestheTABLE 1. Animals Used and Immunohistochemical Analysis Performed

Abbreviations: immunocytocheniistry ASP BSA BSA-glut-* EAA GABA GABA-IR Gln Glu Glu-IR glut-" CL CP DC DVR EL IZ ML

MZ PC SL

vz

aspartate bovine serum albumin BSA-glutaraldehyde-aminoacid conjugates excitatory amino acid (glutamate, aspartate, etc.) gamma-aminohutyric acid GABA-immunoreactivity glutamine glutamate glutamate-immunoreadivity glutaraldehyde-amino acid conjugates

Abbreuiations: anatomical cellular layer cortical plate dorsal cortex dorsal ventricular ridge ependymal layer intermediate zone molecular layer marginal zone piriform cortex (lateral cortex, LC) subcellular layer ventricular zone

Stage 14 15 16

17 18 19 20 21

22 23

anti-GABA'

anti-Glu'

2

2 2 3

4

5 6

3 1 2 4

5 2 1 2

3 1

24

2 3 3

25

2

26

1

4 4 2 1

Adult

1

1

Absorption control3

Filter disc4

Nonimmune serum'

+ +

+ +

+

+ + + +

+ + + + + +

+ +

+ + + + +

+ + +

+ + + ~

+ +

~

~~

'Polyclonal antisera 25 and 26 GAEIA, supplied by Dr. 0. Ottersen and Dr. J. StormMathisen, dilutions 1:300to 1:1,200. 'Polyclond antisera 03 and 13 Glu, supplied by Dr. Ottersen and Dr. Storm-Mathisen, dilutions 1:500to 1:2,000. 3Polyclond antisera were preabsorhed for 30 minutes with homologous amino acidglutaraldehyde conjugates (i.e., GABA fiera-GABA conjugates, 200 &MI and, separately, with heterologous conjugates (i.e., GABA sera-Glu conjugates, 200 pM). 4Filter discs containing spots of different amino acids conjugated to brain protein using glutaraldehyde, processed in the same wells as tissue sections. 5Nonimmune serum: normal rabbit serum at 1:500 dilution.

APPEARANCE OFNEUROTRANSMITTERSINTURTLE CORTEX tized as described previously (Blanton et al., '87). Animals were perfused transcardially with a wash of 2% dextran in 0.1 M phosphate buffer (PB), containing heparin (50 units, Elkins-Sinn, Cherry Hill, NJ) and xylocaine (0.5 mg, Astra Pharmaceuticals, Westborough, MA). The wash was followed by 5% glutaraldehyde (electron microscopic grade, Polysciences, Warrington, PA) in PB or by 4% paraformaldehyde/O.l-1% glutaraldehyde in PB. In some cases, the wash was omitted and animals perfused with 1%paraformaldehyde/l.25% glutaraldehyde followed by a more concentrated fixative (4%/5%of each aldehyde) in PB. The entire head was post-fixed in the concentrated fixative overnight at 4"C, and the brain was removed and post-fixed an additional 2 hours. Brains were stored in 0.2% glutaraldehyde at 4°C until processed. Tissue was sectioned in the coronal plane (50-70 pm) using a vibratome.

Antisera Rabbit antisera raised against GABA and glutamate conjugated to bovine serum albumin (BSA) by glutaraldehyde and affinity purified were kindly provided by Dr. Ole Ottersen and Dr. Jon Storm-Mathisen (University of Oslo). GABA antisera used included 25 and 26 GABA, purified by immunoabsorption on Sepharose columns bearing BSAglutaraldehyde (BSA-glut) and BSA-glut-glutamate or BSAglut-taurine, and used at a dilution of 1:300 to 1:1,200. Glutamate antisera 03 and 13 Glu, purified against BSAglut and BSA-glut-GABA (Ottersen and Storm-Mathisen, '84b; Ottersen et al., '861, were used at a dilution of 1:500 to 1:2,000.

Immunocytochemical procedure A modification of the protocol previously described (Storm-Mathisen et al., '83) was used for localization of GABA-immunoreactivity (GABA-IR) and glutamate-IR (Glu-IR) in free floating tissue sections: 1) Sections were rinsed in a buffer, containing 0.3 M NaCl, 0.1% Triton X-100, and 0.2% normal goat serum (NGS) in 0.1 M Tris/HCl, pH 7.4 (SST buffer). 2 ) Sections were rinsed in 95% ethanol, followed by SST rinses and 3) were preincubated in blocking serum (0.5% NGS in SST) for 1 hour. 4) Incubation in primary antisera followed for 1hour at room temperature or up to 72 hours at 4°C. 5) The sections were then processed using the avidin-biotin-peroxidase method (Vectastain ABC kit, rabbit IgG, Vector Laboratories, Burlingame, CA) using 0.5 mg/ml3-3'-diaminobenzidine (DAB) as chromagen, as previously described (Blanton et al., '87). Staining was intensified by adding a rinse in 5% CoCl, prior to the DAB step. 6) Finally, sections were rinsed, mounted, slowly dehydrated, cleared in xylenes, and coverslipped.

Specificity controls Antisera specificity was evaluated as follows: 1) The specific primary was omitted or replaced by nonimmune normal rabbit serum. 2) To allow direct comparison of potential cross reactivities with staining in tissue sections, millipore filter discs containing spots of various amino acids conjugated by glutaraldehyde to dialyzed brain protein were placed in wells with the sections (Ottersen and Storm-Mathisen, '84b). 3) The specific primary was preabsorbed with soluble glutaraldehyde-amino acid conjugates that mimic the antigen in fixed tissue, as previously described (Storm-Mathisen et al., '831, and was tested on tissue sections and filter discs, run separately or in the same

5 73

titer-plate wells. As a positive control, 03 or 13 Glu was preabsorbed with glutaraldehyde-glutamateconjugates and 25 or 26 GABA was preabsorbed with glutaraldehydeGABA conjugates. Negative controls included glutaraldehyde-aspartate and glutaraldehyde-GABA conjugates for the Glu antisera and glutaraldehyde-glutamate for the GABA antisera.

RESULTS Lamination and neuronal types in embryonic cortex At embryonic stages 15 and 16, the portion of the cerebral vesicle that will become the dorsal cortex consists of a neuroepithelium, with a prominent ventricular zone (VZ) of proliferating cells and mitotic figures located at the ventricular surface (Fig. 1).A loosely packed layer of immature neurons, the plexiform primordium (PP), is positioned between the ventricular zone and the pial surface. Two morphological classes of neurons are distinguishable within this zone: horizontally elongate bipolar cells and very immature pyramidal cells with ascending dendrites (see companion paper and Fig. 1). Between stages 16 and 19, the ventricular zone compresses and mitotic figures become progressively less frequent, consistent with the previously observed time course of neurogenesis in dorsal cortex from stages 15 to 18 (Goffinet et al., '86). There is no subventricular zone: at no time are mitotic figures observed away from the ventricular surface. The somata of radial glial cells form the dominant cell population at the ventricular surface by stage 19. The ventricular zone becomes a glial layer as post-mitotic neurons migrate away from the ventricular surface to form a loosely packed cortical plate. As in developing mammalian cortex, a marginal zone composed of horizontally elongate and other nonpyramidal cell forms lies above the cortical plate, but in contrast to mammalian neocortex, the intermediate or migratory zone in dorsal cortex is markedly compressed. The major neuronal classes in developing turtle cortex occupy distinct radial positions: nonpyramidal cells occupy all layers of cortex, while pyramidal cells are found predominantly in the cortical plate and scattered below. As shown in Figure 1,the basic lamination achieved by stage 19 is maintained through later stages and in mature turtles. In contrast to mammalian neocortex, in which successive waves of later generated cells cause massive expansion of the cell dense layers, the cortical plate in embryonic turtles becomes more compact with development to form the cellular layer (CL). The marginal zone expands to become the mature molecular layer (ML), the predominant zone of afferent input to cortex (Desan, '88, Smith et al., '80), while the layer below the cortical plate, containing scattered pyramidal and nonpyramidal cells and cortical efferent axons, becomes the subcellular layer (SL). The cell bodies of radial ependymal glia, which persist into adulthood, form a thin ependymal layer (EL) lining the ventricle.

Distribution of GABA-immunoreactivecells in mature cortex and controls Column-purified GABA antisera 25 and 26, previously characterized by Ottersen and Storm-Mathisen ('84b), dis-

574

M.G. BLANTON AND A.R. KRIEGSTEIN

16

19

22 I I I

I 4

0

I

I I t I

Fig. 1. Summary of lamination (upper panel) and distinct neuronal types (lower panels) in differentiating reptilian cerebral cortex at embryonic stages 16, 19, and 22, based on morphological data from the companion paper. Neurons born in the ventricular zone (VZ, mitotic

figures) migrate into the plexiform primordium (PP) to form a loose cortical plate (CP) by stage 19. Distinct nonpyramidal (NP) and pyramidal (P) forms are distinguishable early in cortical development. See text for details and key for additional abbreviations.

cretely labeled some cells in all cortical layers except the ependymal layer and appeared to label all neurons in the ML in stage 24 turtle cortex (Fig. 2). Punctate structures, likely synaptic terminals, were distributed similarly and were conspicuous surrounding unstained cells, presumably pyramidal cells, in the CL. This pattern of GABAimmunoreactivity (GABA-IR) is identical to that observed in a prior study in mature turtles using a commercially available GABA antiserum (Blanton et al., '87, using Immunonuclear reagents) and matches the distribution of nonpyramidal cells and their synaptic terminals observed previously (Desan, '84; Blanton et al., '87; Ebner and Colonnier,

tissue sections (Fig. 3). The primary antisera, generated against GABA or glutamate conjugated by glutaraldehyde to bovine serum albumen (BSA),would be expected to react in tissue sections only with the amino acid fixed to brain protein by glutaraldehyde. This selectivity was modelled by conjugating various amino acids to dialyzed brain protein with glutaraldehyde and blotting the conjugates to millipore filter discs (Ottersen and Storm-Mathisen, '84b). Consistent with the result in tissue sections, nonimmune serum failed to detect the amino acid-protein complexes or the protein alone (null spot, see Fig. 3). The specificity of the GABA antisera was tested by processing tissue sections and the filter discs together in the same titer-plate wells, to determine the degree of cross reactivity of each antiserum with similar amino acids. The GABA antiserum specifically recognized the GABA-protein

'78).

In contrast to the discrete staining produced by the GABA antisera, normal rabbit serum (NRS, nonimmune) at 1:300 dilution produced no clear labeling pattern in

APPEARANCE OF NEUROTRANSMITTERS IN TURTLE CORTEX complex spot on the discs and not those for other amino acid-protein complexes, including aspartate, glutamate or glutamine (Fig. 2). Tissue sections run in the same wells exhibited the pattern of staining described above. Preabsorption of the GABA antiserum with soluble GABA-glutaraldehyde conjugates (200 KM) completely blocked specific GABA-IR in tissue sections as well as recognition of the GABA spot on filter discs (Fig. 2). In contrast, preabsorption with glutamate-glutaraldehyde complexes did not block staining in either the tissue sections or the filter disc controls. The staining with glutamate preabsorption shows a slight decrease in intensity when compared with nonabsorbed, likely reflecting slight differences in final reaction parameters. However, the pattern of staining is identical to that without Glu-preabsorption. These results indicate that GABA 25 and GABA 26 labeled GABA-containing cells in tissue sections, with little cross reactivity with the other putative amino acid neurotransmitters tested.

Distribution of glutamate-immunoreactive cells in mature cortex and controls Glutamate antisera Glu03 and Glu13, in some cases preabsorbed with glutamine-glutaraldehyde complexes to inhibit cross reactivity with glutamine, intensely stained sections of stage 24 turtle cortex with a pattern distinct from that observed with the GABA antisera (Fig. 3). Glutamate-immunoreactivity (Glu-IR) labeled the majority of neuronal somata in the cellular layer and many in the subcellular layer. Cells in the molecular layer were unstained or very lightly stained but surrounded by intense Glu-IR in the neuropil. The neuropil was itself not homogeneously stained but contained discrete unlabeled structures approximately the size of dendrites. Cells in the ependymal layer were unstained. The distribution of Glu-IR corresponds to that of the perikarya and neurites of the predominant cortical cell population, the pyramidal cells, and is complementary to the distribution of GABA-IR. It should be noted that pyramidal neurons in reptiles are so named because of their functional resemblance to pyramidal neurons in mammalian neocortex (see Desan, '84, and the companion paper); perikarya of these cells are not typically pyramidal in shape but, rather, rounded with multiple ascending apical dendrites and a descending axon initial segment. Cells with these features were Glu-IR. The glutamate antisera recognized the glutamate spot on the filter discs, with slight cross reactivity with the glutamine spot (Fig. 3). Preabsorption of the glutamate sera with glutamate complexes eliminated the characteristic pattern of Glu-IR in tissue sections and blocked the labeling of the glutamate spot on the filter discs. Preabsorption with GABA or aspartate complexes, in contrast, failed to block specific staining in tissue sections or on the filter discs. The results indicate that Glu-IR in tissue sections corresponds to glutamate. The pattern of labeling is consistent with elevated levels of glutamate in a subpopulation of cortical cells, the pyramidal cells. In summary, GABA-IR elements had the distribution and morphology characteristic of nonpyramidal cells and their terminals, while Glu-IR elements had a complementary distribution like that of pyramidal cells. The combined patterns of staining labeled most cortical components, with the exceptionof radial glial cells which were never GABA-IR or Glu-IR. Glial cells are, however, immunostained by antisera to glial fibrillary acidic protein, a glial-specific

575

marker (Kriegstein et al., '86; Blanton, unpublished). Since these morphological cell classes are distinguishable and can be traced to early developmental time points (see companion paper), we next used the antisera for GABA and glutamate that label the relatively mature neurons to trace the expression of amino acids by these cortical cell classes during embryonic development.

Distinct telencephalic compartments The mature reptilian forebrain contains two major compartments, distinct in terms of function and neurotransmitter utilization (Ulinski, '83; Reiner et al., '84; Desan, '84; Mufson et al., '84) and expression of molecular markers (Kriegstein et al., '86): 1) A cortical zone, consisting of piriform cortex (PC, or lateral cortex [LC]), thalamorecipient dorsal cortex (DC),medial or hippocampal cortex (MC), and the dorsal ventricular ridge (DVR), a cortical region with multiple sensory representations (Ulinski, '831, and 2) a subcortical zone composed of the striatum and several basal forebrain nuclei. These zones can be followed through development and showed distinct patterns of GABA-IR at the earliest age studied (stage 14, see Fig. 4). The subcortical zone contained large numbers of GABA-IR cells and a dense but diffuse GABA-IR neuropile, at a time when only a small number of isolated cells and processes were present in the cortical zone, concentrated in the ventral part that will become DVR and PC. GABA expression occurred in a large number of cells still in the VZ in the subcortex, but not in the cortical VZ. The distinct patterns of GABA-IR in cortical and subcortical areas were maintained as development proceeded. While the number of GABA-IR cells and the density of their neurites in the cortical zone increased substantially with time, the comparative discreteness of GABA-IR neurite labeling in cortex was maintained. The zone of cortex that bordered the more diffusely immunoreactive subcortex increased in radial thickness during development and protruded progressively into the lateral ventricle to become the DVR. In contrast to its discrete ventral border with the subcortex, the dorsal border of DVR with DC was unclear at early stages and had to be inferred by position and cell plate density (see Fig. 4). The exact border of DC and MC was also unclear but was assumed to be in the vicinity of the dorsal flexure of the telencephalic wall. The determination of these borders allowed comparison of specific cortical zones through development.

Lateral to medial gradient of differentiation At early stages of development (stages 14-18), the lateral portion of the cortical zone contained more postmitotic cells (cells outside the VZ) covering a larger radial extent than more medial portions (Figs. 4, 5 ) GABA-IR cells, too, were most numerous laterally and fit clearly along the lateral-tomedial differentiation gradient. At stages 14 and 15, GABA-IR cells resided primarily in the future DVR and were infrequent in dorsal cortex (Fig. 4), although GABA-IR neurites could be observed along the whole lateral to medial extent of cortex (Fig. 5 , 6). Some of these neurites could be traced to an external fascicle of GABA-IR fibers in the subcortical zone (not shown). At stage 16, GABA-IR cells were much more frequent in DVR and DC, nearly always located outside the VZ and by stages 17 and 18were located across the extent of cortex including the ventricular zone (see Figs. 5, 7, 9). In parallel with decreasing GABAergic cell density along the lateral to medial gradient, GABAergic

Fig. 2. GABA-immunoreactivity (GABA-IR) in stage 24 cerebral cortex specificallylabels elements with the characteristic distribution of nonpyramidal cells. A schematic of dorsal cortex (left panel) shows the distribution of nonpyramidal cells in all layers (shaded ovals) and pyramidal cells in the cellular layer (open ovals); a sample filter disc indicates the location of specific amino acid-conjugate spots for this figure and for Figure 3. GABA antisera 25 labels cells with a nonpyrami-

GABA ANTISERUM PREABSORB-GLU

dal-like distribution, leaving most cells in the cellular and ependymal layers unstained, and recognizes the GABA spot on the filter disc. Staining in tissue sections and of the GABA spot on the filter disc was blocked by GABA-glutaraldehyde conjugates (preabsorb-GABA) but not by Glu-glutaraldehyde conjugates (preabsorb-GLU). GABA antiserum 26 gave the same distribution of staining. See text for further details. Scale bar = 50 km.

PREABSORB-GABA

E

r-

GLU ANTISERUM

Fig. 3. Glutamate-immunoreactivity (Glu-IR) in stage 24 cerebral cortex specificallylabels elements with the characteristic distribution of pyramidal cells. Nonimmune rabbit serum produced no specific labeling in a section from a stage 25 turtle and no clear pattern on the filter disc. Glu antiserum 03 labels the majority of cells in the cellular layer and neurites in all layers, but no cells in the molecular layer, a distribution characteristic of pyramidal cells and their neurites. The Glu anti-

NONIMMUNE S E R U M PREABSORB-GABA

serum recognizes the Glu spot on the filter disc (large arrowhead), with slight cross-reactivity with the glutamine spot (small arrowhead). See Figure 2 for the filter disc key. Staining in tissue sections and of the glutamate spot on the filter disc was blocked by glutamate-glutaraldehyde conjugates (preabsorb-GLU) but not by GABA-glutaraldehyde conjugates (preabsorb-GABA).Glu antisera 13 gave similar results. See text for further details. Scale bar = 50 pm.

PREABSORB-GLU

Fig. 4. GABA-immunoreactivity reveals distinct compartments in embryonic forebrain and the emergence of cortical zones. As early as stage 14, GABA-IR cells in the cortical zone are well isolated with discrete neurite staining, a pattern clearly distinct from that in the subcortical zone (striatum, STR), in which large numbers of cells are stained and the neuropil is densely but diffusely stained. The distinct appearance of the staining of these two compartments is preserved

throughout development. Distinct cortical zones emerge during early forebrain differentiation, and the dorsal cortex (DC, parallel line) can be identified adjacent to the dorsal ventricular ridge (DVR),which bulges increasingly into the lateral ventricle, immediately dorsal to the striatum. The dark band at the ventricular surface at stage 22 is artifactual.

APPEARANCE OFNEUROTRANSMITTERSINTURTLE CORTEX

579

Fig. 5. A lateral-to-medial gradient of appearance of Glu-immunoreactive cells in the cortical zone is followed by a similar gradient of GABA-immunoreactive cells. The zones of glutamate-IR and GABA-IR cells, lying superficial to the ventricular zone, decreases in its radial extent along a lateral to medial gradient at each stage. The leading edge of cellular glutamate expression (clear arrow) lies medial to the leading

edge of cellular GABA expression (dark arrowhead). The leading edge of neurotransmitter expression extends medially from stages 15 to 17, and the cell plate at each site along the gradient increases in size and complexity. Upper panels illustrate cortical zones enlarged from the corresponding hemispheres in the lower panels. See text for further details.

neurites decreased in complexity and staining decreased in intensity (Fig. 8C). Glutamate-IR cells showed a similar gradient from lateral to medial. At stage 15, the number of glu-IR cells and their radial extent was greatest laterally and progressively decreased medially, becoming a layer one to two cells deep adjacent to the VZ at the level of DC (Fig. 5, 6). Ventricular zone cells were unstained (Fig. 6). At stage 16, Glu-IR cells were observed more medially and the radial extent of Glu-IR cells was increased, when compared to similar levels in stage 15 cortex (Fig. 5). The lateral to medial differentiation trend continued at stage 17, with cells labeled across the entire lateral to medial extent of cortex and with increased cell number at each level.

could be seen to precede that for GABA, and to show a distinctive pattern. This distribution was apparent early in the differentiation of each zone. In dorsal cortex at stage 15 (Fig. 5 ) a row of Glu-IR cells was apparent adjacent to the VZ, while GABA expression was restricted to rare neurons and isolated neurites tipped with growth cones and rare axons with varicosities (Fig. 6). Later in differentiation at stage 16 (or more laterally at stage 151, the number of Glu-IR cells increased in DC and these cells became flanked by sparse, horizontally elongate GABA-IR cells located in the MZ and in the border region between Glu-IR cells and the VZ (Figs. 7AB, 8). This structure, a cortical plate composed of Glu-IR cells, likely pyramidal cells, flanked by GABA-IR nonpyramidal cells, grossly resembles the structure of early developing mammalian neocortex. The early cortical organization of GABA-IR cells flanking a Glu-IR cortical plate in embryonic turtles quickly gave way to a radially continuous GABA-IR cell population, as some cells in the cortical plate expressed the GABAergic

Emergence of GABA-IR and Glu-IR cells When sections stained with the GABA antisera were compared with adjacent sections stained with the Glu antisera, the lateral to medial wave of glutamate expression

1M.G. BLANTON AND A.R. KRIEGSTEIN

580

Figure 6

APPEARANCE OF NEUROTRANSMITTERS IN TURTLE CORTEX

581

Fig. 7. The appearance of a cortical plate of glutamate-immunoreactive cells flanked by GABA-immunoreactive cells is quickly followed by GABA expression in the cortical plate. At stage 16, horizontally oriented GABA-IR cells (A) flank a cell plate predominantly composed

of glutamate-IR cells (B) but by stage 17 ( C ) ,numerous GABA-IR cells are also found in the cortical plate (between arrowheads) Scale bar = 50

phenotype by stage 17 (Fig. 7C). While the majority of GABA-IR neurons were still horizontally elongate, more diverse neuronal morphologies appeared. At stage 17 and as neurogenesis in dorsal cortex neared completion at stage 18 (Goffinet et al., '86), GABA-IR and Glu-IR cells became increasingly frequent in the VZ (Fig. 9). Expression of amino acid neurotransmitters in the VZ at stages 17 and 18 differed from the situation at early stages, in which expression occurred only in cells that had reached the top of the VZ, with rare exceptions (see Figs. 5-8). At least some neurons may have expressed GABA soon after leaving theVZ. GABA-IR cells with immature morphology often have trailing processes in the VZ and rarely a GABA-IR cell is seen within the VZ with a trailing process still attached to the ventricle (Fig. 8). At stage 14, GABA-IR cells are found laterally in cortex and many of these have radially oriented processes that are inferred to be trailing processes from migration that are subsequently retracted (Fig. 8A,B and see Blanton and Kriegstein, '91, for similar retraction of pyramidal cell trailing processes). As GABA expression proceeds in a lateral to medial direction with development, cells with trailing processes as well as cells in the VZ are found at the leading edge of differentiating cells (see Fig. 8B) but are more frequent in more lateral cortex. Many of the medially placed cells show a horizontal orientation with noVZ processes. These results are consistent with local generation and transmitter expression for some cells, with delayed expression or horizontal migration for others.

In summary, cells immunoreactive for GABA or for glutamate are present very early in cortical development and differ in their distribution and gradients of expression. Glu-IR cells appear first, located at the surface of the VZ, and can be found throughout development, in the cortical plate and later in the mature cellular layer. These cells, the dominant group numerically, correspond in distribution and in their numerical predominance to the pyramidal neurons. GABA-IR cells first appeared in the MZ and below the cortical plate, and in morphology and distribution correspond to nonpyramidal cells.

Fig. 6. At stage 15, glutamate-immunoreactive cells in dorsal cortex lie above the ventricular zone at a time when GABA-immunoreactive cells are infrequent. Rare, horizontally oriented GABA-IR cells (A, arrow) and neurites (B) with varicosities (small arrows) and growth cones (larger arrowhead) are found in the primordial plexiform layer. The lateral to medial GABA expression gradient is shown in C , with lateral to the left and the location of cells and neurites in panels A and B indicated (arrow for cell in A, arrowhead for neurites in B). Glutamate-IR cells (D, arrowheads) are found at the external margin of the ventricular zone at the same lateral-to-medial level as GABA-IR neurites. Scale bar in A, for A, B, D, = 20 pm, in C = 50 pm.

Pm.

Appearance of the mature distribution of neuronal types By stage 20, the cortical plate had condensed to form a discrete cell layer (CL), composed primarily of pyramidal neurons (see companion paper) but also of isolated nonpyramidal neurons. At this stage, the majority of cells in this layer were Glu-immunoreactive, and resembled pyramidal cells; some clear examples of non-immunoreactive cells were also found (Fig. 10B). Non-Glu-IR cells, surrounded by densely Glu-IR neuropile, were the dominant neuronal type in the ML (Fig. 10B); some of these non-Glu-IR cells exhibited the horizontal morphology characteristic of some nonpyramidal cells. In their distribution across layers and frequency, these non-Glu-IR cells resemble the population of cells stained by GABA-IR (Fig. IOA). GABA-IR cells and terminals occupied all cortical layers. A similar distribution of Glu-IR and GABA-IR cells was observed through the remainder of embryonic development (Figs. 11,121. The distribution of immunoreactive punctate structures, presumed terminals, indicates the formation of cortical circuits. GABA-IR puncta along thin processes were observed, rarely, as early as stages 14 and 15 in the marginal zone (Fig. 6). The density of GABA-IR puncta in this zone increased with development (Figs. 11, 12). Thin GABA-IR presumed axons, predominantly radial and containing varicosities, were also clearly seen in the cell layer by stage 20 and may have been present earlier. Glutamate-IR puncta were also apparent. Puncta contacting non-Glu-IR (presum-

582

Fig. 8. Cells initiate neurotransmitter expression upon leaving the ventricular zone at early developmental stages. Neurons may express GABA soon after leaving the ventricular zone, as many GABA-IR cells show trailing processes in this zone (small arrowheads, A, stage 14; B, stage 16). Some of the cells with trailing processes at the medial edge of

M.G. BLANTON AND A.R. KRIEGSTEIN

the differentiation gradient are clearly less densely GABA-IR (B, arrow) than immediately adjacent, more differentiated cells (B, arrowhead). Cells on the medial edge of the GABA expression gradient (arrow) are often more weakly GABA-IR than more lateral cells (arrowhead, stage 15). Scale bar in A, for A, B = 20 wm, for C = 50 pm.

APPEARANCE OFNEUROTRANSMITTERSINTURTLE CORTEX

583

Fig. 9. Cells express neurotransmitters while still in the ventricular zone (VZ) at later developmental stages. Cells immunoreactive for GABA (arrow, A) and glutamate (arrows, B) are common in the VZ at stage 18. Glu-IR cells are often found in clusters. Scale bar = 20 km.

ably GABA-IR) somata were observed in the cellular layer by stage 20.

DISCUSSION Neurons become immunoreactive for the putative amino acid neurotransmitters GABA and glutamate early in the embryonic development of the turtle dorsal cortex. Neuronal subtypes were found to express distinguishing biochemical features, with nonpyramidal cells immunoreactive for GABA and likely pyramidal cells immunoreactive for glutamate. The precocious expression of immunoreactivity for these neurotransmitters is consistent with the proposed involvement of GABA and glutamate in the control of early developmental processes (for reviews, see Redburn and Schousboe, '87; Mattson, '88; McDonald and Johnston, '90).

Validation of the immunocytochemical approach Interpretation of the significance of our findings depends on the specificity of the immunocytochemical reagents and an understanding of neuronal glutamate metabolism. Previous studies have demonstrated that the GABA and glutamate antisera used in the present study specifically recognize their amino acid targets on filter disc controls which mimic fixation of endogenous amino acids to brain protein in situ, and that the antisera label discrete subpopulations of neurons in the cerebral cortex (Storm-Mathisen et al., '83; Ottersen and Storm-Mathisen, '84; Ottersen et al., '86). We have found, similarly, that each antiserum recognizes the target amino acid spot on filter disc controls and a subpopulation of cortical cells in sections of embryonic and mature turtle cortex. Immunostaining in sections and on filter discs was blocked appropriately by glutaraldehyde

584

M.G. BLANTON AND A.R. KRIEGSTEIN

APPEARANCE OF NEUROTRANSMITTERS IN TURTLE CORTEX conjugates of the amino acid against which each antiserum was raised, but not other amino acids (see Storm-Mathisen and Ottersen, ’86). The results of the controls strongly suggest that, under the conditions employed in this study, the immunocytochemicalstaining in tissue sections by the GABA and glutamate antisera corresponds to fixed endogenous GABA and glutamate. Evidence that neurons specificallylabeled with the GABA antisera correspond to nonpyramidal cells and that these cells indeed use GABA as a neurotransmitter is provided by numerous studies in which other markers of GABAergic function are localized in cortical nonpyramidal cells and their synaptic terminals. Immunocytochemical localization using other GABA antisera (Seguella et al., ’84; Somogyi et al., ’85) or antisera against the GABA synthetic enzyme glutamic acid decarboxylase, or GAD (Ribak, ’78; Hendrickson et al., ’81;Peters et al., ’82), histochemical localization of GABA-transaminase, a presynaptic GABA degrading enzyme (Nagai et al., ’83) and 3H-GABA accumulation methods (Hendry and Jones, ’81; Hokfelt and Ljungdahl, ’72; Somogyi et al., ’84) all label nonpyramidal cells in mammalian neocortex. Similarly, immunocytochemical (Blanton et al., ’87), ultrastructural (Smith et al., ’801, and physiological (Kriegstein and Connors, ’86) studies implicate nonpyramidal neurons in mature turtle dorsal cortex in GABA-mediatedinhibition. In embryonic turtle cerebral cortex, neurons immunoreactive for GABA also correspond to nonpyramidal cells. Horizontally oriented cells in the early marginal zone (stages 15-16) that are GABA-IR resemble nonpyramidal cells labeled with HRP in morphology and distribution (see companion paper). GABA-IR cells and HRP-labeled nonpyramidal neurons exhibit increasingly diverse morphologies and become distributed in all cortical layers over the same time course. Other markers of GABAergic function, GADimmunoreactivity and 3H-GABA uptake, also label cells with a distribution similar to that observed for GABAimmunoreactivity in embryonic turtle cortex (Blanton, unpublished). Moreover, physiological studies indicate the presence of functional GABAergic circuits in embryonic life. Excitation of nonpyramidal cells is associated with inhibitory synaptic currents in pyramidal cells that are sensitive to GABA receptor antagonists (Blanton and Kriegstein, in preparation). These results indicate that nonpyramidal cells in embryonic turtle cortex are likely to be GABAergic. Evidence suggests that pyramidal neurons use excitatory amino acids, including glutamate, as neurotransmitters (see Streit, ’84; Fagg and Foster, ’83 for review). However, interpretation of glutamate-immunoreactivity is complicated by the possibility that “metabolic” glutamate may be found within neurons that do not use glutamate as a neurotransmitter (Ottersen and Storm-Mathisen, ’84; Storm-Mathisen et al., ’86). Several laboratories have re-

585

ported glutamate immunostaining in discrete subpopulations of cortical pyramidal cells and in excitatory synaptic terminals (Conti et al., ’87, ’89; Dori et al., ’89; Guiffrida and Rustioni, ’89). The antisera employed in the present study appeared to label a majority of the pyramidal neurons in neocortex that are thought to use glutamate as their neurotransmitter or to contain glutamate in a metabolic pool (Ottersen and Storm-Mathisen ’84ab, ’86), and the pattern of glutamate-immunostaining was shown previously to be generally complementary to the pattern of GABA immunostaining (Ottersen and Storm-Mathisen, ’84ab; see also Conti et al., ’87). Therefore glutamate-IR with this antiserum could be thought of, conservatively, as a marker of non-GABAergic cells, cells which lack the synthetic machinery for converting glutamate into GABA. In the present study, Glu-IR cells dominate the cellular layer and are also common in the subcellular layer, a distribution like that of HRP-labeled pyramidal cells and distinct from that of nonpyramidal cells, and a subset of these cells correspond to pyramidal neurons that make distant projections to thalamus (see companion paper). As early as stage 16, corticothalamic projection neurons and glutamate-IR cells are observed superficial to the ventricular zone. The description of these cells as “pyramidal” stems from their resemblance to neocortical pyramidal cells: pyramidal cells in reptiles and mammals make distant projections, have local recurrent axon collaterals, and have characteristic physiological properties (see Connors and Kriegstein, ’86, for discussion). The majority of the pyramidal or projection neurons in embryonic turtle have rounded or multipolar perikarya, shaped by a dominant organizational feature of these cells, an ascending tuft of several oblique apical dendrites. Perikarya with this shape and characteristic descending axon initial segments are glutamate immunoreactive. Therefore, neurons with the distribution, abundance, and perikaryal morphology of pyramidal (or projection) neurons in turtles are glutamateimmunoreactive. By analogy with neocortex, at least some of these glutamate-immunoreactive cells are likely to use glutamate as a neurotransmitter. The immunocytochemical results in mature brain using antisera to amino acid neurotransmitters are usually interpreted in terms of whether immunoreactive neurons correspond to those capable of releasing the particular amino acid at their synaptic terminals, and thus whether the amino acid resides in a “neurotransmitter” pool rather than a “metabolic” pool (Ottersen and Storm-Mathisen, ‘84, Storm-Mathisen and Ottersen, ’86; Storm-Mathisen et al., ’86).During early development prior to synapse formation, a distinction between neurotransmitter and metabolic pools is difficult to assess, as the concept of neurotransmission is problematic prior to synapse formation. Neurotransmitters are classically thought of in terms of synaptic function, and the roles of neuroactive substances are most easily considered and tested by comparison with events at synapses. Fig. 10. A definitive cellular layer populated mainly by glutamateOne developmentaldefinition for neurotransmitters could immunoreactive cells has formed at stage 20, and GABA-immunoreactive cells show a complementary distribution. A GABA-IR cells resem- be that the neuroactive substance, expressed precociously, ble nonpyramidal cells in morphology and distribution. The majority of would later be used as a neurotransmitter once a particular neurons in the cellular layer (between arrowheads) are unstained (*I. cell forms functional synapses. A stricter definition would B: Glu-IR cells with pyramidal-shaped somata and descending axons require a signalling role for the substance when expressed, (arrow) predominate in the cellular layer; these cells resemble pyrami- prior to synapse formation. Glutamate or other excitatory dal cells in morphology and in distribution. Unstained, non-Glu-IR cells have the same distribution and frequency as cells that are GABA-IR amino acids may fulfill these criteria in young cortical cells. Glutamate released into the extracellular space from growth (see A). GABA-IR and Glu-IR puncta, likely synaptic terminals, are cones or other compartments could potentially signal develapparent. Scale bar = 15 pm.

586

M.G. BLANTON AND A.R. KRIEGSTEIN

APPEARANCE OF NEUROTRANSMITTERS IN TURTLE CORTEX

15

15+

16

17

18

20

587

22

24

26

GLUTAMATE

Fig. 12. Summary of the embryonic differentiation of GABAimmunoreactive and glutamate-immunoreactive elements. A layer of glutamate-containing cells is present early, along with GABAergic neurites and then a flanking population of GABAergic cells by stage 16. Neurotransmitter expression in the ventricular zone at stages 17 and 18 gives rise to a loosely organized cortical plate, a radially continuous distribution of GABAergic cells and glutamate-containing cells. The cortical plate becomes condensed by stage 20, and the distinct distribu-

Fig. 11. A relatively mature distribution of GABA and glutamate immunoreactive cells and presumed terminals is apparent at stage 26, hatching. GABA-IR cells (A, arrowhead) and puncta, presumed terminals, are found in all layers, and puncta (arrows) outline unstained presumed pyramidal neurons (*) in the cellular layer. Glu-IR pyramidal cells (arrowhead) are found in the cellular layer and punctate structures, likely terminals (arrows), are found in both the ML (arrows),and on non-Glu-IR somata (*, with arrows), likely NPs, in the cellular layer. Scale bar = 15 pm.

tions of glutamate and GABA-containing cells are maintained with further differentiation. As the ependymal layer condenses at the ventricular surface, only rare immunoreactive neurons are found in this zone (not represented). The molecular layer expands during synaptogenesis during embryonic life, indicated by the increase in GABA-immunoreactive puncta with development. Glu-immunoreactive puncta are also present but not represented.

opmental information (see McDonald and Johnston, 'go), providing a potential agonist for the endogenous NMDA receptor activation observed early in the development of embryonic turtles (Blanton et al., '90) and rats (LoTurco et al., '91). Synapses with ultrastructural characteristics of excitatory contacts (Blue and Parnavelas, '83) only appear in significant numbers at stage 18 (Blanton, Parnavelas, and Kriegstein, unpublished). With the onset of synapse formation, glutamate released by pyramidal cell s o n s can

M.G. BLANTON AND A.R. KRIEGSTEIN

588 underly the classical signalling at synapses, acting on non-NMDA receptors to produce the spontaneous excitatory synaptic currents first observed at stage 18 (Blanton and Kriegstein, in press).

Detection of emerging cell classes

oping elaborate horizontal dendrites. Until combined birthdating and GABA immunocytochemical studies are performed, the timing of expression and possible migration route of these cell types will not be known with certainty. Appearance of GABAergic horizontally oriented neurons occurs early in cortical differentiation and could occur both by local radial migration and expression of GABA and by tangential migration of cells into a cortical region that expressed GABA at more ventral levels. Initially, GABA and glutamate expression in turtles occurs in postmitotic neurons once their somata have left the ventricular zone, but in later cortical development, expression of neurotransmitters, especially glutamate, occurs in some cells still in the ventricular zone. In mammalian corticogenesis, postmitotic neurons in the ventricular zone and in the process of migration have been found to express neurotransmitters (Phelps et al., '90; Specht et al., '81; Wallace and Lauder, '83), including GABA in the cortical ventricular zone (Chun and Shatz, '89; Van Eden et al., '89). Some of the neurotransmitter expressing neurons in the turtle ventricular zone are likely to migrate to join the cortical plate, while others may have attained their definitive positions, as occasional GABA-IR and glutamate-IR cells are observed at the ventricular surface in mature turtles, intermingled with ependymal cell somata.

Immunocytochemical techniques can allow identification of emerging cell classes in developing brain (Roberts et al., '87; Van Eden et al., '89). A population with distinct morphology and distribution, GABA-IR nonpyramidal cells can be detected early and followed throughout cortical development. Glutamate-IR cells, likely pyramidal cells, can also be followed throughout development, in the cortical plate and its successor, the cellular layer of adults. Glutamate-IR appeared to label the majority of neurons in this layer, most of which are known to be pyramidal (see companion paper), and glutamate-IR cells in this layer have somata shaped like those of pyramidal neurons, with characteristicdescending axons. Expression of GABA-IR and glutamate-IR occurs in distinct cell groups that have differing lateral to medial expression gradients. Early in cortical development at stage 15, glutamate-IR cells occupying the future cortical plate have a distribution distinct from that of GABA-IR cells, which flank this zone when they appear slightly later. This sequence of transmitter differentiation, like morphogenesis Emergence of cortical organization: (Bayer and Altman, 'go), does not necessarily reflect the comparative aspects relative times of origin of the cell classes: while glutamate The primary neuronal constituents of mature cerebral expression precedes GABA expression, neurons that express glutamate are not necessarily generated before those cortex are GABAergic nonpyramidal neurons and excitathat express GABA. It is at present unclear whether the tory amino acid-utilizing projection neurons (Houser et al., early collection of glutamate-IR cells is a homogenous '84; Streit, '84; Fagg and Foster, '83). Our results indicate population, consisting entirely of cells that will mature to that basic features of these neurons, including their expresbecome pyramidal neurons, or whether some of the gluta- sion of amino acid neurotransmitters, occurs very early in mate-expressing cells will change their position, migrating their development. Information on the differentiation of above or below the cell plate and, by expressing GAL), the major cell classes in vertebrate cortex, including the timing of neurotransmitter expression, may shed light on convert their glutamate into GABA. Neurons begin expressing their specific neurotransmit- the mechanisms of cortical evolution and shape questions ters after they reach their final positions in mammalian concerning evolutionary homology. The reptilian dorsal cortex is flanked medially by an cortex (Miller, '86; Cavanagh and Parnavelas, '88) or in some cases during their migration (Van Eden et al., '89). archicortex (hippocampus) and laterally by paleocortex Cells in the subplate and marginal zone in rat (Van Eden et (olfactory cortex) and by topography and connectivity is al., '89) and cat (Chun and Shatz, '89) express GABA and likely to be evolutionarily related to the mammalian neocorbased on their morphology have been hypothesized to tex and similar regions in other vertebrates (Ebner, '76; migrate from more ventral levels, with a leading process Northcutt, '81; Desan, '88; Karten and Shimizu, '89). The indicating the trajectory of movement (Van Eden et al., appearance of a cortical plate flanked by precocious horizon'89). In early cortical development in turtles, some of the tally oriented nonpyramidal cells in early mammalian most medially located horizontal nonpyramidal cells have cortical development and its gross similarity to mature less mature dendritic arbors than cells located ventrolater- reptilian brain (Marin-Padilla, '71, '78; Rickmann et al., ally and may have a process trailing into the ventricular '77) has given rise to the notion that neocortex passes zone; these cells are also usually less densely GABA-IR than through a "reptilian stage" in development, a recapitulacells located more ventrolaterally. Unless these neurons tion of its evolutionary origins (Marin-Padilla, '71, '72). become less differentiated after migration, these particular However, cortical evolution is unlikely to have proceeded cells in turtles are likely to be generated, migrate radially, merely by addition of "mammalian" steps to the end of a and elaborate neurites locally rather than migrating dorso- reptilian series (see Gould, '77 and Gottlieb, '87, for general medially in the intermediate zone. However, local differen- discussion) but rather to have utilized some of the same tiation and neurotransmitter expression for some GABAer- fundamental building blocks and molecular mechanisms to gic cell types does not rule out the possibility that these cells elaborate new and distinctive structures. The patterns of neurogenesis and migration of cortical migrate short distances tangentially, or that other cell types express GABA and migrate longer distances as proposed for cells differ radically in reptiles and mammals. In mammals, similar appearing cells in rat cortex (Van Eden et al., '89). GABAergic and peptidergic Cajal-Retzius and subplate cells Indeed, there are many cells present medially with no are generated in an outside-in sequence (outer cells first, process in the VZ that could either have migrated in or, inner cells last) and are split into upper and lower populaalternatively, could have expressed GABA only after devel- tions by intercalation of the cortical plate (Fig. 13). The

APPEARANCEOFNEUROTRANSMITTERSINTURTLE CORTEX

Mammalian

IC

Reptilian

1

PP

CP

C

2

3

*

Development --t-

$f

Cajal-Retzius subplate cortical plate

Fig. 13. Potential homologies of GABA and glutamate-containing neurons in reptiles and mammals, and developmental histories. The upper panel illustrates mammalian corticogenesis in which a cortical plate (CP) forms within the primordial plexiform layer, splitting it into superficial Cajal-Retzius cells and deep subplate cells (SP), most of which are transient. The cortical plate-derived cells populate mature layers 11-VI. Horizontally oriented Cajal-Retzius neurons that are GAEJA-immunoreactive are found early in cortical development across species and are likely homologous in reptiles and mammals. The relationship of reptilian and mammalian excitatory amino acid utilizing neurons that project to thalamus (Th), contralateral hemisphere (CH), and cortex (IC) is unclear. Some possibilities include (1)Glu-IR cells in reptiles could Correspond to some SP cells, which have a similar outside-in gradient, with cortical plate pyramidal cells evolving only in mammals (or being lost in reptiles). Cellular layer (CL) pyramidal cells in reptiles would then correspond to subplate cells. (2) Alternatively, these turtle cells could correspond to both SP and CP pyramidal cells, but with CP cells evolving the capacity for migrating past earlier-born cells in mammals and not reptiles. (3) Turtle projection cells could correspond to CP pyramidal cells, with transient SP cells a distinct mammalian type. See text for further discussion.

589

cortical plate, consisting predominantly of pyramidal cells, then develops in an inside-out sequence (Luskin and Shatz, '85; Shatz et al., '88; Valverde et al., '89; Van Eden et al., '89; Bayer and Altman, '90). In contrast, the entire turtle cortex develops in an outside-in manner (Goffinet et al., '86). Although their order of neurogenesis is unknown, we show that in turtles, cells immunoreactive for glutamate, likely pyramidal cells, appear first in the cell plate, followed by cells immunoreactive for GABA, including Cajal-Retzius cells, in a flanking distribution. Until the glutamate content of the earliest appearing preplate neurons of mammals is determined and then compared with the pattern in turtles, the degree of similarity for early organizational events in reptiles and mammals will be unclear. In addition to the differences in order and topography of appearance of cell types, cell density and neurite elaboration within the cortical plate also differ in reptiles and mammals (Goffinet, '83; see companion paper). Thus, despite a superficial resemblance of the cortical plate in reptiles and mammals, fundamental structural differences are apparent early, foreshadowingthe distinctly different mature structures. While the ultimate structures they compose differ markedly, the basic cellular elements of vertebrate cortex are similar (Connors and Kriegstein, '86; Kriegstein and Connors, '86; Desan, '84; Blanton et al., '87). However, proof of homology for the various similar-appearing cell populations that comprise the cerebral cortices of reptiles and mammals is difficult (see Fig. 13). One likely homologous group is composed of the horizontally oriented Cajal-Retzius cells that are among the earliest differentiating cells in the cortex (Rambn y Cajal, '11; Marin-Padilla, '72; Raedler and Sievers, '76; Bradford et al., '77; Goffinet, '83; and see companion paper). These cells, with their characteristic morphology, location, and early generation, are likely to correspond to the early horizontal GABA-IR cells observed in developing turtles (this study) and mammals (Van Eden et al., '89; Chun and Shatz, '89; Wolff et al., '84b). The ubiquity of Cajal-Retzius cells in embryonic vertebrate cortex suggests that they may play similar roles in different species. The relationships of other GABAergic neuronal types in reptiles to those in mammals remain to be determined, as these nonpyramidal local circuit neurons are present both in the subplate (Mrzljak et al., '88; Chun and Shatz, '89; Antonini and Shatz, '90) and in cortical platederived layers in mammals. Cortical projection neurons from different vertebrate species also share characteristic features, including the use of excitatory amino acids as neurotransmitters (Streit, '84; Fagg and Foster, '83; this study), pyramidal cell morphology, similar membrane properties (Connors and Kriegstein, '86), and both distant and local projections (Desan, '84). Projection neurons in turtles could correspond to either or both of the major projection cell classes in neocortex, the transient subplate projection cells or the mature neocortical pyramidal cells, each of which likely use amino acids as their neurotransmitters (Streit, '84; Fagg and Foster, '83; Antonini and Shatz, '90). Subplate projection cells elaborate long axons early (McConnell et al., '89), and like some projection neurons in turtles, receive synaptic input early (Chun and Shatz, '88; Friauf et al., '90; Blanton and Kriegstein, in press). Unlike turtle cells, most subplate cells die once synaptic inputs are rearranged later in development (Chun and Shatz, '88, and see Shatz et al., '88, for review). Neocortical pyramidal cells, unlike turtle projec-

590

tion neurons, migrate past earlier generated cells during development to form a multilaminate cortex (Angevine and Sidman, '61). Thus, projection neurons in turtles and pyramidal and some subplate cells in mammals share certain general features and possibly an evolutionary relationship, but each type has a characteristic developmental pattern. The distinctive characteristics of the cortical cell types of different vertebrates likely arose in evolution by a variety of processes that are observed in development. Neuronal properties may have arisen by elimination of specific features from more primitive cells, mirrored in developmental pruning of exuberant cortical projections (see O'Leary and Stansfield, '89) or in loss of other biochemical or physiological traits (parcellation theory, Ebbesson, '84). Neuronal differences could also reflect changes in timing of differentiation of the same cell types in different species, with organizational sequelae (heterochrony,see Gould, '77; Gottlieb, '87; Edelman, '87; Finlay et al., '87). New traits could also have appeared de novo. An understanding of the evolutionary relationships of cortical cell types thus depends on unravelling the expression of individual traits of cells. The early expression of amino acid neurotransmitter phenotypes in neurons of reptiles and mammals indicates that these GABA and glutamate-utilizing cell types are basic units of cortical structure upon which evolutionary processes may have acted to produce a variety of cortical cell subtypes.

Roles of early expressed neurotransmitters The cellular elements that comprise the cerebral cortex of mature reptiles and mammals form distinctly different structures but use the same neurotransmitters in similar synaptic circuitry (Connors and Kriegstein, '85; Kriegstein and Connors, '85; Blanton et al., '87). During the development of synaptic circuitry in a variety of brain regions in higher vertebrates, electrical activity and synaptic activation of amino acid receptors influence the orientation and topography of dendrites of postsynaptic neurons (Steffen and van der Loos, '80; Harris and Woolsey, '81; Katz and Constantine-Paton, '88). Early expression of the neurotransmitter phenotypes underlying mature synaptic transmission may indicate that prior to synapse formation, cortical cells use GABA and glutamate in nonsynaptic interactions (Woodward et al., '71; Wolff et al., '84a), for example in the regulation of neuronal survival and neurite outgrowth. The presence of neurotransmitters so early in cortical development is a prerequisite for mediation of various developmental processes inferred to occur in immature brain as a consequence of neurotransmitter action (Balazs et al., '88; AruRo et al., '87; Pearce et a]., '87; Mattson et al., '88; Brenneman et al., '90; Brewer and Cotman, '89). The mere presence of neurotransmitters or their synthetic apparatus has been taken as evidence for developmental regulation of morphology in vivo (WolR et al., '84a; Lauder et al., '87; Van Eden et al., '89). While no direct evidence for glutamate and GABA involvement in nonsynaptic regulation of cortical development in vivo is available as yet, recent work demonstrates the capacity of early cortical neurons to respond to endogenous neurotransmitters (Blanton et al., '90; LoTurco et al., '911, and the early neuronal synthesis of amino acid transmitters reported here supports the idea that neurotransmitters may have important roles in early cortical development.

M.G. BLANTON AND A.R. KRIEGSTEIN

ACKNOWLEDGMENTS We are indebted to Dr. Ole Ottersen and Dr. Jon StormMathisen of the University of Oslo for the gift of the GABA and glutamate antisera and the raw filter disc control blots, and for many helpful suggestions. We acknowledge the expert technical assistance of John Avilla in many of the immunostaining experiments and in the photography. We also thank Dr. Paul Desan, Dr. Joe LoTurco, Dr. Istvan Mody, Dr. Carla Shatz, and anonymous reviewers for helpful comments on an earlier version of the manuscript. This work was supported by NIH grant NS21223; M.G.B. was supported by Medical Scientist Training Program grant GM07365.

LITERATURE CITED Angevine, J.B., Jr., and R.L. Sidman (1961) Autoradiographic study of cell migration during histogenesis of cerebral cortex in the mouse. Nature 192:766-768. Antonini, A., and C.J. Shatz (1990) Relationship between putative transmitter phenotypes and connectivity of subplate neurons during cerebral cortical development. Eur. J. Neurosci. 2344-761. Aruffo, C., R. Ferszt, A.G. Hildebrandt, and J. Cervos-Navarro (1987) Low doses of L-monosodium glutamate promote neuronal growth and differentiation in uitro.Dev. Neurosci. 9228-239. Balazs, R., N. Hack, and O.S. Jorgensen (1988) Stimulation of the N-methylD-aspartate receptor has a trophic effect on differentiating cerebellar granule cells. 1Veurosci.Lett. 87:80-86. Bayer, S.A., and J. Altman (1990) Development of layer 1and the subplate in the rat neocortex. Exp. Neurol. 107:48-62. Belkin, D.A. (1963) Anoxia: Tolerance in reptiles. Science 139:492-493. Blanton, M.G., and A.R. Kriegstein (1991) Morphological differentiation of distinct neuronal classes in embryonic turtle cerebral cortex. J. Comp. Neurol. 310.550-5 70. Blanton, M.G., and A.R. Kriegstein (in press) Spontaneous action potential activity and synaptic currents in the embryonic turtle cerebral cortex. J. Neurosci. Blanton, M.G., d.M. Shen, and A.R. Kriegstein (1987) Evidence for the inhibitory neurotransmitter gamma-aminobutyric acid in aspiny and sparsely spiny nonpyramidal neurons of turtle dorsal cortex. J. Comp. Neurol. 259277-297, Blanton, M.G., J.J. Lo Turco, and A.R. Kriegstein (1989) Whole cell recording from neurons in slices of reptilian and mammalian cerebral cortex. J. Neurosci. Meth. 312203-210. Blanton, M.G., J.J. LoTurco, and A.R. Kriegstein (1990) Endogenous neurotransmitter activates N-methyl-D-aspartate receptors on differentiating neurons in embryonic cortex. Proc. Natl. Acad. Sci. U S A . 87;8027-8030. Blue, M.E., and J.G. Parnavelas (1983) The formation and maturation of synapses in the visual cortex of the rat. I. Qualitative analysis. J. Neurocytol. 12r599-616. Bradford, R., J.G. Parnavelas, and A.R. Leiberman (1977) Neurons in Layer 1 of the developing occipital cortex of the rat. J. Comp. Neurol. I76:121-131. Brenneman, D.E., I.D. Forsythe, T. Nicol, and P.G. Nelson (1990)N-methylD-aspartate receptors influence neuronal survival in developing spinal cord cultures. Dev. Brain Res. 51t6348. Brewer, G.J., and C.W. Cotman (1989) NMDA receptor regulation of neuronal morphology in cultured hippocampal neurons. Neurosci. Lett. 99:268-273. Cavanagh, M.E., and J.G. Parnavelas (1988) Development of somatostatin immunoreactiveneurons in the rat occipital cortex: A combined immunocytochemical-autoradiographic study. J. Comp. Neurol. 268:l-12. Chun, J.J.M., and C.J. Shatz (1988) Redistribution of synaptic vesicle antigens is correlated with the disappearance of a transient synaptic zone in the developing cerebral cortex, Neuron 1:297-310. Chun, J.J.M., and C.J. Shatz (1989) The earliest-generated neurons of the cat cerebral cortex: Characterization by MAP2 and neurotransmitter immunohistochemistry during fetal life. J. Neurosci. 9:1648-1667. Connors, B.W., and A.R. Kriegstein (1986)Cellular physiology of the turtle visual cortex: Distinctive properties of pyramidal and stellate neurons. J. Neurosci. 6:164-177.

APPEARANCE OFNEUROTRANSMITTERSINTURTLE CORTEX Conti, F., J. DeFelipe, I. Farinas, and T. Manzoni (1989) Glutamate-positive neurons and axon terminals in cat sensory cortex: A correlative light and electron microscopic study. J. Comp. Neurol. 290:141-153. Conti, F., A. Rustioni, P. Petrusz, and A.C. Towle (1987) Glutamate-positive neurons in the somatic sensory cortex of rats and monkeys. J. Neurosci. 7: 1887-1901. Desan, P. (1984) The Organization of the Cerebral Cortex of the Pond Turtle, Pseudemys scripta eleguns. Ph.D. thesis, Harvard University. Desan, P. (1988) Organization of the cerebral cortex in turtle. In Schwerdtfeger, Smcets (eds): The Forebrain of Reptiles. Int. Symp. Recent Advances in Understanding the Structure and Function of the Forebrain in Reptiles, Frankfurt, pp. 1-11. Dori, I., M. Petrou, and J.G. Parnavelas (1989) Excitatory transmitter acid-containing neurons in the rat visual cortex: A light and electron microscopic immunocytochemistry study. J. Comp. Neurol. 290:169184. Ebbesson S.O.E. (1984) Evolution and ontogeny of neuronal circuits. Behav. Brain Sci. 7:321-366. Ebner, F.F. (1976) The forebrain of reptiles and mammals. In R.B. Masterton, M.E. Bitterman, D.B.G. Campbell, and N. Hotton (eds): Evolution of Brain and Behavior in Vertebrates. New York: John Wiley and Sons, pp. 147-167. Ebner, F.F., and M. Colonnier (1978) A quantitative study of synaptic patterns in turtle visual cortex. J. Comp. Neurol. 179263-276. Edelman, G.M. (1987) Neural Darwinism. NewYork: Basic Books, 371 pp. Erdo, S.L., and J.R. Wolff (1989) A comparison of the postnatal changes in aspartate and glutamate levels in cerebral cortex of the rat. Neurosci. Res. Comm. 4:51-56. Fagg, G.E., and A.C. Foster (1983) Amino-acid neurotransmitters and their pathways in the mammalian central nervous system. Neuroscience 9:701-719. Finlay, B.L., K.C. Wikler, and D.R. Sengelaub (1987) Regressive events in brain development and scenarios for vertebrate brain evoluation. Brain Behav. Evol. 30:102-117. Friauf, A., McConnell, S.K., and C.J. Shatz (1990) Functional synaptic circuits in the subplate during fetal and early postnatal development of cat visual cortex. J. Neurosci. 10:2601-2613. Goffinet, A.M. (1983) The embryonic development of the cortical plate in reptiles: A comparative study in Emys orbiculuris and Lucerta agilis. J. Comp. Neurol. 215:437452. Goffinet,A.M., Ch. Daumerie, B. Langenverf, and C. Pieau (1986) Neurogenesis in reptilian cortical structures: 3H-thymidine autoradiographic analysis. 3. Comp. Neurol. 243:10&116. Giuffrida, R., and A. Rustioni (1989) Glutamate and aspartate immunoreactivity in cortico-cortical neurons of the sensorimotor cortex of rats. Exp. Brain Res. 74:4146. Gottlieb, G. (1987) The developmental basis of evolutionary change. J. Comp. Psychology 101:262-271. Gould, S.J. (1977) Ontogeny and Phylogeny. Cambridge: Belknap Press of Harvard University Press. Harris, R.M., and T.A. Woolsey (1981) Dendritic plasticity in mouse barrel cortex following postnatal vibrissa follicle damage. J. Comp. Neurol. 196t357-376. Hendrickson, A.E., S.P. Hunt, and J.-Y. Wu (1981) Immunocytochemical localization of glutamic acid decarboxylase in monkey striate cortex. Nature 292:605-607. Hendry, S.H.C., and E.G. Jones (1981) Sizes and distribution of intrinsic neurons incorporating tritiated GABA in monkey sensory-motor cortex. J. Neurosci. 1.390408. Hokfelt, T., and A. Ljungdahl (1972) Autoradiographic identification of cerebral and cerebellar cortical neurons accumulating gamma-aminobutyric acid (3H-GA13A).Exp Brain Res. 14:354-362. Homer, C.R., J.E. Vaughn, S.H.C. Hendry, E.G. Jones, and A. Peters (1984) GABA neurons in the cerebral cortex. In E.G. Jones and A. Peters, (eds): Cerebral Cortex, Vol. 2. New York: Plenum Press, pp. 63-89. Karten, H.J., and T. Shimizu (1989) The origins of neocortex: Connections and lamination as distinct events in evolution. J. Cogn. Neurosci. lr29 1-301. Katz, L.C., and M. ConstantinelPaton (1988) Relationships between segregated afferents and postsynaptic neurons in the optic tectum of threeeyed frogs. J. Neurosci. 8t3160-3180. Kriegstein, A.R., and B.W. Connors (1986) Cellular physiology of the turtle visual cortex: Synaptic properties and intrinsic circuitry. J. Neurosci. 6:178-191. Kriegstein, A.R., J.M. Shen, and N. Eshhar (1986) Monoclonal antibodies to

591

the turtle cortex reveal neuronal subsets, antigenic cross-reactivity with the mammalian neocortex, and forebrain structures sharing a pallid derivation. J. Comp. Neurol. 254:330-340. Kriegstein, A.R., J.M. Shen, G.M. Lee, and M.G. Blanton (1988) Development of inhibitory function during corticogcnesis in the turtle. In W.K. Schwerdtfeger and W.J.A.J. Smcets (eds): The Forebrain of Reptiles. Basel: Karger, pp. 131-141. Lauder, J.M., V.K.M. Han, P. Henderson, T. Verdoorn, and A.C. Towle (1986) Prenatal ontogeny of the GABAergic system in the rat brain: an immunocytochemical study. Neuroscience 19:465493. LoTurco, J.J., M.G. Blanton, and A.R. Kriegstein (1991) Initial expression and endogenous activation of NMDA channels in early neocortical development. J. Neurosci. 11:792-799. Luskin, M.B., and C.J. Shatz (1985) Studies of the earliest generated cells of the cat’s visual cortex: Cogeneration of subplate and marginal zones. J. Neurosci. 5:1062-1075. Marin-Padilla, M. (1971) Early prenatal ontogenesis of the cerebral cartex (neocortex) of the cat (Felis domesticu). A Golgi study. I. The primordial neocortical organization. Z. Anat. Entwickl. Gesch 134:117-145. Marin-Padilla,M. (1972) Prenatal ontogenic history of the principal neurons of the neocortex of the cat (Felis domesticu). A Golgi study. 2. Anat. Entwickl. Gesch. 136:125-142. Marin-Padilla, M. (1978) Dual origin of the mammalian neocortex and evolution-ofthe cortical plate. Anat. Embryol. 152:109-126. Mattson, M.P. (1988) Neurotransmitters in the regulation of neuronal cytoarchitecture. Brain Res. Rev. 13:179-212. Mattson, M.P., P. Dou, and S.B. Kater (1988) Outgrowth-regulating actions of glutamate in isolated hippocampal pyramidal neurons. J. Neurosci. 82087-2100. Mattson, M.P., and S.B. Kater (1989) Excitatory and inhibitory neurotransmitters in the generation and degeneration of hippocampal neuroarchitecture. Brain Res. 478t337-348. McConnell, S.K., A. Ghosh, and C.J. Shatz (1989) Subplate neurons pioneer the first axon pathway from the cerebral cortex. Science 245978-982. McCormick, D.A., B.W. Connors, J.W. Lighthall, and D.A. Prince (1985) Comparative electrophysiology of pyramidal and sparsely spiny stellate neurons of the neocortex. J. Neurophysiol. 54:782-806. McDonald, J.W., and M.V. Johnston (1990) Physiological and pathophysiological roles of excitatory amino acids during nervous system development. Brain Res. Rev. 15:41-70. Meier, E., B. Belhage, J. Drejer, and A. Schousboe (1987) The expression of GABA receptors on cultured cerebellar granule cells is influenced by GABA. I n D.A. Redburn and A. Schousboe (eds): Neurotrophic Activity of GABA During Development. New York: A.R. Liss, Inc. pp. 139-159. Miller, M.W. (1986) The migration and ncurochemical differentiation of gamma-aminobutyric acid (GABA)-immunoreactive neurons in rat visual cortex as demonstrated by a combined immunocytochemicalautoradiographic technique. Dev. Brain Res. 28t41-46. Miller, M.W. (1988) Development of projection and local circuit neurons in neocortex. In A. Peters, E.G. Jones (eds): Cerebral Cortex, Vol. 7 (Plenum Press, New York) pp. 133-175. Moran, J., and A.J. Pate1 (1989) Stimulation of the N-methyl-D-aspartate receptor promotes the biochemical differentiation of cerebellar granule neurons and not astrocytes. Brain Res. 486:15-25. Mrzljak, L., H.B.M. Uylings, I. Kostovic, and C.G. Van Eden (1988)Prenatal development of neurons in the human prefrontal cortex: I. A qualitative golgi study. J. Comp. Neurol. 272:355-386. Mufson, E.J., P.H. Desan, M.M. Mesularn, B.H. Wainer, and A.I. Levey (1984) Choline acetyltransferase-likeimmunoreactivity in the forebrain of the redeared pond turtle (Pseudemys scripta eleguns). Brain Res. 323:103-108. Nagai, T., P.L. McGeer, and E.G. McGeer (1983) Distribution of GABA-Tintensive neurons in the rat forebrain and midbrain. J. Comp. Neurol. 218220-238. Northcutt, R.G. (1981) Evolution of the telcncephalon in mammals. Annu. Rev. Neurosci. 4:301-350. O’Leary, D.D.M., and B.B. Stanfield (1989) Selective elimination of axons extended by developing cortical neurons is dependent on regional locale: Experiments utilizing fetal cortical transplants. J. Neurosci. 9:22302246. Ottersen, O.P., and J. Storm-Mathisen (1984a) Neurons containing or accumulating transmitter amino acids. In A. Bjorklund, T. Hokfelt and M.J. Kuhar (eds): Handbook of Chemical Neuroanatomy, Vol. 3: Classical Transmitters and Transmitter Receptors in the CNS, Part 11. Amsterdam: Elsevier Science Publishers, pp. 141-246.

592 Ottersen, O.P., and J. Storm-Mathisen (198413) Glutamate- and GABAcontaining neurons in the mouse and rat brain, as demonstrated with a new immunocytochemical technique. J. Comp. Neurol. 229:374-392. Ottersen, O.P., J. Storm-Mathisen, S. Madsen, S. Skumlien, and J. Stromhaug (1986) Evaluation of the immunocytochemical method for amino acids. Med. Biol. 64:147-158. Pearce, I.A., M.A. Cambray-Deakin, and R.D. Burgoyne (1987) Glutamate acting on NMDA receptors stimulates neurite outgrowth from cerebellar granule cells. FEBS Lett. 223t143-147. Peters, A., C.C. Proskauer, and C.E. Ribak (1982) Chandelier cells in rat visual cortex. J. Comp. Neurol. 206t397416. Phelps, P.E., R.R. Barber, L.A. Brennan, V.M. Maines, P.M. Salvaterra, and J.E. Vaughn (1990) Embryonic development of four different subsets of cholinergic neurons in rat cervical spinal cord. J. Comp. Neurol. 291~926. Raedler, A,, and J. Sievers (1976) Light and electron microscopical studies on specific cells-of the marginal zone in the developing rat cerebral cortex. Anat. Emybrol. 149:173-181. Ram6n y Cajal, S. (1911) Histologie du SystGme Nerveux de 1’Homme et des Vertkbres (translated by L. Azoulay), Vol. 2. Paris: Maloine, pp. 519-646. Redburn, D.A., and A. Schousboe (1987) Neurotrophic Activity of GABA During Development. New York A.R. Liss, 227 pp. Reiner, A., S.E. Brauth, and H.J. Karten (1984) Evolution of the amniote basal ganglia. Trends Neurosci. 7:320-325. Ribak, C.E. (1978) Aspinous and sparsely-spinous stellate neurons in the visual cortex of rats contain glutamic acid decarboxylase. J. Neurocytol. 7:461478. Rickmann, M.B., M. Chronwall, and J.R. Wolff (1977) On the development of nonpyramidal neurons and axons outside the cortical plate: The early marginal zone as a pallial anlage. Anat. Embryol. 151285-397. Roberts, A,, N. Dale, O.P. Ottersen, and J. Storm-Mathisen (1987) The early development of neurons with GABA immunoreactivity in the CNS of Xenopus Zaeuis embryos. J. Comp. Neurol. 261:435449. Seguela, P., M. Geffard, R.M. Buijs, and M. Le Moal (1984) Antibodies against gamma-aminobutyric acid: Specificity studies and immunocytochemical results. Proc. Natl. Acad. Sci. U.S.A. 81:3888-3892. Shatz, C.J., J.M. Chun, and M.B. Luskin (1988) The role of the subplate in the development of the mammalian telencephalon. In A. Peters and E.G. Jones (eds): Cerebral Cortex, Vol. 7. New York: Plenum Publishing, pp. 35-57. Shatz, C.J., and M.B. Luskin (1986) The relationship between the geniculocortical af€erents and their cortical target cells during development of the cat’s primary visual cortex. J. Neurosci. 6t3655-3668. Smith, L.M., F.F. Ebner, and M. Colonnier (1980) The thalamocortical projection in Pseudemys turtles: A quantitative electron microscopic study. J. Comp. Neurol. 190:445461. Somogyi, P., T.F. Freund, and Z.F. Kisvarday (1984) Different types of 3H-GABAaccumulating neurons in the visual cortex of the rat: Characterization by combined autoradiography and Golgi impregnation. Exp. Brain Res. 54:45-56. Somogyi, P., A.J. Hodgson, I.W. Chubh, B. Penke, and A. Erdei (1985)

1CI.G. BLANTON AND A.R. KRIEGSTEIN Antisera to gamma-aminobutyric acid. J. Histochem. Cytochem. 33:240248. Specht, L.A., V.M. Pickel, T.H. Joh, and D.J. Reis (1981) Light-microscopic immunocytochemical localization of tyrosine hydroxylase in prenatal rat brain. I. Early onitogeny. J. Comp. Neurol. 199:233-253. Steffen, H., and J. Van der Loos (1980) Early lesions of mouse vibrissal follicles: Their influence on dendrite orientation in the cortical barrelfield. Exp. Brain Res. 40:419-431. Storm-Mathisen, A.K. Leknes, A.T. Bore, J.L. Vaaland, P . Edminson, S. Haug, and O.P. Ottersen (1983) First visualization of glutamate and GABA in neurones by immunocytochemistry. Nature 301:517-520. Storm-Mathisen, J . , and O.P. Ottersen (1986) Antibodies against amino acid neurotransmitters. In P. Panula, H. Paivarinta, and S. Soinila (eds): Neurohistocbemistry: Modern Methods and Applications. New York A.R. Liss, Inc., pp~.107-136. Storm-Mathisen, J., 10.P. Ottersen, T. Fu-Long, V. Gundersen, J.H. Laake, and G. Nordbo (1986) Metabolism and transport of amino acids studied by immunocytochemistry. Med. Biol. 64:127-132. Streit, P. (1984) Glutamate and aspartate as transmitter candidates for systems of the cerebral cortex. In E.G. Jones and A. Peters (eds): Cerebral Cortex, Vol. 2: Functional Properties of Cortical Cells. New York Plenum Press, pp. 119-143. Traub, R.D., R. Miles, and R.K.S. Wong (1987) Models of synchronized hippocampal bursts in the presence of inhibition. I. Single population events. J. Neurophysiol. 58:739-75 1. Ulinski, P.S. (1983) Dorsal Ventricular Ridge: A Treatize on Forebrain Organization in Fleptiles and Birds. New York Wiley, 283 pp. Valverde, F., M.V. Facal-Valverde, M. Santacana, and M. Heredia (1989) Development and differentiation of early generated cells of sublayer VIb in the somatosensory cortex of the rat: A correlated Golgi and autoradiographic study. J. Comp. Neurol. 29Ot118-140. Van Eden, C.G., L. Mrzljak, P. Voorn, and H.B.M. Uylings (1989) Prenatal development of GABAergic neurons in the neocortex of the rat. J. Comp. Neurol. 289:213-227. Wallace, J.A., and J .M. Lauder (1983) Development of the serotonergic system in the rat embryo: An immunocytochemical study. Brain Res. Bull. 10:459-479. Woodward, D.J., B.J. Hoffer, G.R. Siggins, and F.E. Bloom (1971) The ontogenic development of synaptic junctions, synaptic activation and responsiveness to neurotransmitter substances in rat cerebellar Purkinje cells. Brain Res. 34t73-97. Wolff, J.R., V.J. Balcar, T. Zetzche, H. Bottcher, D.E. Schmechel, and B.M. Cromwall (1984a) Development of GABAergic system in rat visual cortex. In J.M. ILauder and P.G. Nelson (eds): Gene Expression and Cell-Cell Interactions in the Developing Nervous System. Adv. Exp. Med. Biol. New York: Plenum Press, pp, 215-239. Wolff, J.R., H. Bottcher, T. Zetsche, W.H. Oertel, and B.M. Chronwall (1984b) Development of GABAergic neurons in rat visual cortex as identified by glutamate decarboxylase-like immunoreactivity. Neurosci. Lett. 47t207-212. Yntema, C. (1968) A series of stages in the embryonic development of Chelydra Serpentina. J. Morphol. 125:219-252.