Early Neocortical Regionalization in the Absence of Thalamic Innervation

See allHide authors and affiliations

Science  06 Aug 1999:
Vol. 285, Issue 5429, pp. 906-909
DOI: 10.1126/science.285.5429.906


There is a long-standing controversy regarding the mechanisms that generate the functional subdivisions of the cerebral neocortex. One model proposes that thalamic axonal input specifies these subdivisions; the competing model postulates that patterning mechanisms intrinsic to the dorsal telencephalon generate neocortical regions. Gbx-2mutant mice, whose thalamic differentiation is disrupted, were investigated. Despite the lack of cortical innervation by thalamic axons, neocortical region–specific gene expression (Cadherin-6, EphA-7, Id-2, andRZR-beta) developed normally. This provides evidence that patterning mechanisms intrinsic to the neocortex specify the basic organization of its functional subdivisions.

The mammalian neocortex is organized into regionally distinct functional subdivisions. There are two proposed mechanisms for neocortical regionalization. The protocortex hypothesis postulates that thalamic afferent fibers play an important role in neocortical regional development (1). On the other hand, the protomap hypothesis (2) postulates that regionalization of the neocortex is due to molecular differences within the neocortical ventricular zones (3–7). We investigated mice with a mutation of the Gbx-2 gene (8, 9) that disrupts thalamic histogenesis, which in turn blocks formation of thalamocortical projections.

Analysis of Gbx-2, Id-4, and Lef-1expression showed that Gbx-2–deficient mice have abnormal thalamic development. Gbx-2 expression in the dorsal thalamus (DT) could be detected in the subventricular (SVZ) and mantle zones (MZ) (10). In the Gbx-2 mutant,Gbx-2 mRNA expression was maintained in the SVZ but was lost in the MZ (Fig. 1, E and F), suggesting that the mutant is unable to produce normal DT postmitotic cells (Gbx-2 transcripts were detected in the mutant because the 5′ end of this gene has not been altered). This hypothesis was supported by the expression pattern of Id-4. TheId-4 helix-loop-helix gene (11), which normally is expressed throughout the MZ of the DT, was not detectable in the DT of the Gbx-2 mutant (Fig. 1, I and J). In addition to the apparent defect in the MZ, there was ectopic expression ofLef-1 in the mutant. At embryonic day (E) 14.5, theLef-1 high mobility group box transcription factor (12) was expressed in the MZ of the pretectum (PT) and DT. In the Gbx-2 mutants, Lef-1 expression in the PT was unaffected, whereas its expression in the MZ was lost in the DT, and ectopic periventricular expression was observed (Fig. 1, M and N). Thus, by E14.5, loss of Gbx-2 function disrupted differentiation in most, or all, of the DT.

Figure 1

Gene expression patterns in the thalamus of E14.5 and P0 Gbx-2 mutant and control brains. The age and genotype of the sections are indicated on the left; the gene markers are indicated at the top of the figure. (A to D) Nissl stain reference sections. (E to H) Expression of Gbx-2. (I to L)Id-4 expression. (M to P)Lef-1 expression; the arrow in (P) highlights the ectopic periventricular expression. CM, centromedian; VLG, ventral lateral geniculate; VZ, ventricular zone. (25). Scale bars: (E), 0.75 mm [same for (F) to (L), (O), and (P)]; (M), 0.53 mm [same for (N)].

To verify the thalamic abnormalities, we studied thalamic anatomy at postnatal day (P) 0, the day when these animals die. Histological abnormalities in the P0 Gbx-2 mutant thalamus were even more pronounced. Gbx-2 expression marked the lateral dorsal (LD), centromedian (CM), dorsal lateral geniculate (DLG), and reuniens (Re) nuclei (Fig. 1G). Several other nuclei also expressed Gbx-2, including the ventral lateral, the midline, the anteromedial, the mediodorsal, and the medial geniculate (MG) nuclei (13). In the mutant, there was only weak Gbx-2 expression, which was restricted to a periventricular region (see asterisk in Fig. 1H).

Some thalamic regions of the Gbx-2 mutants retainedId-4 and Lef-1 expression. Id-4expression marked most nuclei in the DT and the reticular nucleus (RT) of the ventral thalamus (Fig. 1K). In contrast, the Gbx-2mutant had greatly reduced Id-4 expression in most areas of the DT (Fig. 1L) but retained expression in the RT, which does not express Gbx-2 (10). Limited Id-4expression was detectable in some dorsal thalamic areas that, on the basis of their location, may correspond to the ventrobasal complex (VB) and DLG nuclei (Fig. 1L). Lef-1 expression marked specific DT subdivisions including the DLG, VB, ventromedial (VM), and posterior (Po) nuclei (Fig. 1O). In the Gbx-2 mutant, Lef-1expression was lost from most of the MZ, but unlike the wild type, its expression was found near the ventricle (arrow in Fig. 1P). SomeLef-1 expression was observed in areas that we interpret as corresponding to the VB, DLG, and Po nuclei (Fig. 1P) (13). Thus, although most of the thalamus was defective in theGbx-2 mutant, there remained some areas that may correspond to thalamic nuclei that normally have reciprocal connections with the neocortex (for example, DLG, VB, Po, and MG).

To determine whether thalamocortical connections were intact in the mutant, we used 1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate (DiI) and immunohistochemical staining methods. Placement of a DiI crystal into the DT of E14.5 normal (Gbx-2 +/+ or Gbx-2+/ ) embryos labeled cell bodies and axons that project into the internal capsule (IC) (n = 5) (Fig. 2A). In E14.5Gbx-2 −/− embryos, the number of axons entering the IC was severely reduced (n = 5) (Fig. 2B). Implanted DiI in the DT at E16.5 labeled thalamocortical fibers that reach the neocortical subplate in normal animals (n = 7) (Fig. 2C), whereas thalamic fibers stopped in the IC of Gbx-2 mutants (n = 5) (Fig. 2D). Placement of DiI in the DT of normal mice at E18.5 (Fig. 2E) or P0 (14) labeled thalamocortical fibers anterogradely and corticothalamic fibers and their cell bodies retrogradely (n = 4). In most mutant cases (n = 5), no labeled axons were observed in the cortex (Fig. 2F). In only one mutant (E18.5) did a few thalamocortical fibers innervate the subplate; this sparse innervation was detected in both hemispheres (14).

Figure 2

Loss of neocortical innervation by the thalamus in Gbx-2 mutants. DiI fluorescence is in pink; the tissue is counterstained in blue with DAPI (4,6 diamidino-2-phenylindole). All panels are oriented with medial to the left. (Left) Gbx-2+/+ or Gbx-2+/ . (Right) Gbx-2−/−. (A and B) DiI injections in the DT at E14.5 label thalamocortical fibers that reach the striatopallial angle in normal mice [arrow in (A)]; inGbx-2 mutants, these fibers extend only into the midportion of the IC [arrow in (B)]. (C and D) DT injections of DiI at E16.5 retrogradely label some neocortical cells [arrow in (C)]. In Gbx-2 mutants, thalamocortical fibers extend no farther than the lateral IC [arrow in (D)]. DiI labeling in the cortex (Ctx) is nonspecific. (E and F) DT injections of DiI at E18.5; in controls (E), thalamic fibers that innervated the cortex are labeled, and neocortical pyramidal neurons are retrogradely labeled [inset in (E)]. In Gbx-2 mutants (F), most thalamic fibers remain within the IC [long arrow in (F)], although a few approach the cortex [short arrow in (F)]. The asterisk in (F) indicates an aberrant extension of the lateral ventricle (LV) observed in some Gbx-2 −/−forebrains. (G and H) Labeling of neurons in the DT from DiI injections into the cortex at P0. Retrogradely labeled cells are visible in the normal DT [arrows and inset in (G)] but not in the mutant DT. In the mutant, corticothalamic fibers did not reach that thalamus (14), whereas the cerebral peduncle (cp) was present (G and H). fr, fasciculus retroflexus; Hp, hippocampus; LGE, lateral ganglionic eminence; St, striatum. Scale bars: (A), 500 mm (all panels shown at the same scale); inset in (E), 50 mm (both insets shown at the same scale) (26).

Finally, placement of DiI in the normal neocortex at E18.5 (14) or P0 (Fig. 2G) labeled thalamocortical fibers and their cell bodies retrogradely (inset in Fig. 2G) and cerebral peduncle fibers (Fig. 2G) and corticothalamic fibers (14) anterogradely (n = 4). In contrast, no retrograde labeling of thalamic neurons was observed after DiI injections into the Gbx-2 mutant neocortex in six of seven cases (14). In the single exceptional case, both corticothalamic and thalamocortical projections were present in both hemispheres, showing that this animal did not have a fully penetrant defect in its thalamus.

To confirm the DiI results, we hemisected some of the P0 brains; one hemisphere was used for DiI tracing, and the other was used to study the expression of thalamocortical axonal markers [that is, serotonin (5HT), P450c17, and calretinin]. In each case, the immunohistochemical and DiI results were concordant. 5HT immunohistochemistry (15) revealed that only the control animal had staining of the SP/CP (subplate/cortical plate) (Fig. 3, A and B). Immunohistochemical analysis of P450c17, an enzyme that converts pregnenolone to DHEA (dehydroepiandrosterone), showed thalamocortical fibers in the intermediate zone (IZ) and SP in normal mice (16) but not in Gbx-2 mutants (Fig. 3, C and D). Whereas the thalamocortical pathway was disrupted, other P450c17-labeled cortical axons, such as the angular bundle (AB), appeared unaffected (Fig. 3D) (13).

Figure 3

Immunohistochemical analysis of thalamocortical projections in Gbx-2 mutants: expression of 5HT and P450c17 at P0 and calretinin at P0 and E14.5. The genotypes are indicated on the left-hand side; the markers and age of the brains are listed at the top. (A and B) 5HT expression in saggital sections of P0 brains. 5HT expression was lost in the SP/CP of the mutant. (C and D) P450c17 expression at P0 in coronal sections; expression was lost in the fibers in the IZ/SP of the mutant, whereas expression in the AB in the subiculum (SUB) was preserved (D). (E and F) Calretinin expression at E14.5 in coronal sections revealed that labeled fibers in the IC were absent in the mutant. (G and H) Calretinin expression at P0 in coronal sections revealed that labeled fibers in the SP were absent in the mutant. NCX, neocortex (27). Scale bars: (A), 1.3 mm; (C), 1.0 mm [same for (D)]; (E), 0.83 mm [same for (F)]; and (H), 1.2 mm [same for (G)].

Another marker of thalamocortical axons is calretinin, a calcium-binding protein that is expressed in the midline and intralaminar thalamic nuclei and their axons (17). At E14.5, calretinin-positive thalamocortical axons in normal mice extended to the IC (Fig. 3E), and by P0, they innervated some cortical areas (Fig. 3G). In E14.5 mutants, calretinin-positive axons were severely reduced or absent from the IC (Fig. 3F), and by P0, there were no detectable calretinin-positive fibers in the mutant cortex (Fig. 3H). The mechanism underlying the defective thalamocortical tract is unknown.

The cerebral cortex of Gbx-2 mutants, which lacked thalamocortical innervation, proved to be an excellent system to study the role of thalamocortical fibers on neocortical regionalization and differentiation. P0 brains were hemisected; one hemisphere was used for DiI analysis to verify that thalamocortical innervation was defective, and the other hemisphere was used to study neocortical development. Nissl-stained specimens did not reveal obvious defects in lamination and regionalization (Fig. 4, D and E), although in ∼20% of the animals, morphological defects, such as enlarged ventricles and tears in the telencephalic wall at the cortical/subcortical boundary, were observed (asterisk in Fig. 2F). We restricted our molecular analysis of the Gbx-2 mutant neocortex to animals that lacked obvious morphological defects.

Figure 4

In situ hybridization shows that intraneocortical gene expression boundaries and neocortical lamination are not affected by the lack of thalamic input in Gbx-2 mutants. Genotypes are indicated on the left and gene markers on top. Nissl-stained sections are shown (Dand E). Expression of Id-2, EphA-7, and RZR-beta at P6 (A, F, andK) and P0 (B, G, and L) respects two intraneocortical boundaries: i and ii. These boundaries appear normal in P0 Gbx-2 mutants (C,H, and M). Cadherin-6 expression (P0) in parietal neocortical areas 1 and 2 (arrowheads) in the mutant matches normal expression (N and O).Tbr-1 expression (P0) shows its normal strong expression in the motor cortex (MC) and layer 6 in the mutant (28) (I and J). The asterisk in (K) indicates an area where low RZR-betaexpression is detectable. Scale bars: (O), 0.88 mm [same for (N)]; (J), 0.7 mm [same for (B) to (E), (G) to (I), (K), and (L)]; and (A), 1.2 mm [same for (F) and (K)].

We studied neocortical regionalization and laminar organization using the expression of Id-2, EphA-7,RZR-beta, Cadherin-6, and Tbr-1 at P0 and found no obvious differences in the cortical expression of these genes between control (Fig. 4, B, G, I, L, and N) and Gbx-2mutants (Fig. 4, C, H, J, M, and O). Expression of Id-2(Fig. 4, A to C), which encodes a helix-loop-helix protein, demarcates functionally important regional boundaries and layers in the cortex (5). For instance, layer 5 expression has a rostral boundary at the sensory-motor limit at P0 and P6 (18) (see boundary “i” in Fig. 4A). Expression of EphA-7 (Fig. 4, F to H), a receptor protein tyrosine kinase that is involved in axon guidance (19), has two neocortical expression boundaries at P6 and P0: Boundary i approximates the sensory-motor limit (18); boundary “ii” is within the sensory neocortex (Fig. 4, F to H). Boundary ii has not been mapped with respect to neocortical subdivisions. Expression of RZR-beta, which encodes an orphan nuclear receptor (20), has a limit that also approximates boundary ii (Fig. 4, K to M). Expression ofCadherin-6 (Fig. 4, N and O), which encodes a cell-surface adhesion protein, demarcates parietal areas 1 and 2 and temporal area 1 (21). Finally, laminar development was further assessed by studying Tbr-1, a T-box gene, which is expressed in layer 6 (5) (Fig. 4, I and J).

In summary, Gbx-2 mutant mice have a block in DT differentiation that affects most thalamic regions and that disrupts growth of thalamocortical axons. Despite the absence of thalamic afferents, neocortical regionalization and histogenesis were indistinguishable from control mice at P0. This provides strong evidence that regional specification within the neocortex is controlled at least in part by patterning mechanisms that are intrinsic to the telencephalon. The molecular regulation of patterning in the telencephalon is poorly understood. Candidate cortical patterning centers include the dorsal midline, which expresses bone morphogenetic proteins and Wnts, and the rostral midline, which expresses fibroblast growth factor–8 (22). These tissues may control the expression of transcription factors that are implicated in regulating cortical regionalization (for example, Emx-1,Emx-2, Otx-1, Bf-1, Gli-3,Tlx, and Tbr-1) (18, 23). Although some early steps in neocortical molecular regionalization (even within the sensory neocortex) and lamination may be largely independent of thalamic influence, it is likely that thalamic input regulates later steps in neocortical development, such as neuronal maturation, and the formation of neocortical modules (for example, barrels) and association cortices. This latter hypothesis is based on the evidence that afferent fibers can regulate the architecture and gene expression of thalamic and cortical tissues (24). Unfortunately, because Gbx-2mutants die on P0, we need to devise methods to study the mutant neocortex in more mature animals to investigate the role of thalamic afferents on postnatal neocortical development.

  • * These authors contributed equally to this work.

  • To whom correspondence should be addressed. E-mail: jlrr{at}

  • Present address: Cell Biology and Metabolism Branch, National Institute of Child Health and Human Development, National Institutes of Health Building 18T, Room 101, 9000 Rockville Pike, Bethesda, MD 20892, USA.


Stay Connected to Science

Navigate This Article