Reelin-expressing Cajal-Retzius (CR) cells are among the earliest generated cells in the mammalian cerebral cortex and are believed to be crucial for both the development and the evolution of a laminated pattern in the pallial wall of the telencephalon. LIM-homeodomain (LIM-hd) transcription factors are expressed during brain development in a highly restricted and combinatorial manner, and they specify regional and cellular identity. We have investigated the expression of the LIM-hd members Lhx1/Lhx2/Lhx5/Lhx6/Lhx9 in the reelin-expressing cells, the pallium, and the regions of origin of CR cells including the cortical hem of 3 amniote species: the mouse, the chick, and the macaque monkey. We found major differences in the combinatorial LIM-hd expression in the marginal zone as well as in the hem. 1) Lhx5 is a “preferential LIM-hd” for CR cells in mammals but not expressed by these cells in chicks. 2) Lhx2 is expressed in the hem of the chick, whereas it is excluded from this region in mouse. 3) Whereas mouse CR cells express Lhx5/Lhx1, their monkey counterparts express 4 of these factors: Lhx1/Lhx2/Lhx5/Lhx9. We discuss our findings in evolutionary terms for the specification of the midline hem and CR cell type and the emergence of the cortical lamination pattern.
In mammalian vertebrates, the cerebral cortex displays a highly specific laminar organization that results from complex specification and migration events during embryogenesis (reviewed in Marin and Rubenstein 2001; Mérot et al. 2009). One of the first-generated types of cortical cells, the Cajal-Retzius (CR) cells, play a crucial role in these processes (reviewed in Soriano and Del Rio 2005). CR cells appear at the onset of corticogenesis in the marginal zone/layer I and disappear at the end of the neuronal migration period. Although they are all glutamatergic (Hevner et al. 2003), they constitute a heterogeneous population not only in terms of morphology and molecular composition (Marin-Padilla 1998; Meyer et al. 1999; Yamazaki et al. 2004) but also in terms of embryological origins: they are born at several sites at the borders of the pallium, including the cortical hem at the telencephalic dorsal midline (Meyer et al. 2002; Takiguchi-Hayashi et al. 2004) and the anti-hem/ventral pallium at the pallio-subpallial boundary (Bielle et al. 2005). Studies in genetically hem-ablated mice have shown that the cortical hem produces the CR cells of the hippocampus and neocortex, whereas the retrobulbar area, septum, and thalamic eminence seem to produce the CR cells of the olfactory cortex and amygdala (Yoshida et al. 2006; Cabrera-Socorro et al. 2007). They are subsequently redistributed all over the cortical mantle by tangential migrations in the marginal zone. There, CR cells are strategically located to orchestrate cortical migrations, being adjacent to the pial end-feet of glia along which cortical neurons migrate to reach their layer of destination.
CR cells are considered central in corticogenesis mainly because they produce reelin, a large extracellular matrix glycoprotein that is essential for a correct cortical lamination. Indeed, in “reeler” mice (Falconer 1951) that lack functional reelin (Ogawa et al. 1995; D'Arcangelo et al. 1995; D'Arcangelo et al. 1996), cortical layers are inverted (reviewed in Tissir and Goffinet 2003) and this phenotype is rescued by providing exogenous reelin (Magdaleno et al. 2002; Jossin et al. 2004).
“Reelin” is first detected around embryonic day (E) 10 in mouse and is expressed in the marginal zone during mammalian corticogenesis (Alcantara et al. 1998). The identification of CR cells as the major reelin-producing cells in rodents is based on their location in the marginal zone, early birthday, typical bipolar horizontal morphology, and coexpression of calretinin and p73 markers. They are detected with CR-50 antibody (D'Arcangelo et al. 1997). In addition to reelin-expressing CR cells described in mouse, the rat marginal zone also contains subpial granular layer cells (SGL), a population that is more abundant and better characterized in primates: in human and monkey, reelin is expressed at early stages by large CR cells and at later stages by smaller SGL cells (Meyer and Goffinet 1998; Zecevic and Rakic 2001), which are proposed to compensate for the progressive loss of CR cells through the long period of corticogenesis in primates. In adult mouse and monkey cortex, long after CR cells have disappeared from layer I, some γ-aminobutyric acidergic (GABAergic) interneurons still express reelin (Alcantara et al. 1998; Rodriguez et al. 2002). These complex features of reelin expression suggest that it has multiple roles in brain development and function. In fact, in humans, disruption of the reelin gene leads to lissencephaly with cerebellar hypoplasia (Hong et al. 2000). Impaired reelin signaling and polymorphisms in the reelin gene were reported in autism (Fatemi et al. 2005; Serajee et al. 2006), and reelin is considered as a vulnerability factor in schizophrenia and bipolar disorders (Grayson et al. 2005; Torrey et al. 2005).
Reelin is conserved in all vertebrates, yet its expression pattern in the developing pallium of mammals (Alcantara et al. 1998), non-mammalian amniotes (Bernier et al. 1999, 2000; Cabrera-Socorro et al. 2007), teleost fish, and amphibians (Costagli et al. 2002) or lampreys (Perez-Costas et al. 2002; Perez-Costas et al. 2004) is strikingly different. The comparison of reelin patterns between amniote species showing some degree of cortical lamination (mammals and lizards) or no obvious pallial cytoarchitectonic condensation at all (turtles and birds) led to a “Reelin hypothesis” for cortical developmental evolution, with the condensation of reelin-expressing cells being a key feature for the establishment of a sophisticated laminated pattern (Bar and Goffinet 2000; Bar et al. 2000; Tissir et al. 2002). These comparative data point to the importance of the reelin pathway, hence of CR cells, in the morphogenesis and cytoarchitecture of pallial structures.
Mouse CR cells express the LIM-homeodomain (LIM-hd) transcription factor Lhx5/Lhx1 (Yamazaki et al. 2004), and Lhx5−/− mutants show severe hippocampal malformation (Zhao et al. 1999). Other cells of the developing cortex also express additional LIM-hd members, including Lhx2, Lhx9 (Rétaux et al. 1999; Bulchand et al. 2001, 2003), or Lhx6 (Lavdas et al. 1999; Alifragis et al. 2004). In the cortex and other brain regions, LIM-hd proteins act in a combinatorial manner in regional and cellular specification, and it has been proposed that variations in the LIM code may also play a role in brain evolution (Bachy et al. 2002). To investigate whether an LIM code exists in CR cells and whether this relates to cortical evolution, we have analyzed the combinatorial patterns of Lhx1/2/5/6/9 in re-elinpositive CR cells of the developing cortex/pallium of 3 amniote species: the mouse, with a well-developed laminated cortex; the monkey with a very large cortex and higher degree of lamination; and the chick with a poorly/nonlaminated pallium. Our analysis suggests that reelin-expressing cells in primate layer I show higher molecular complexity; that is, a larger repertoire of LIM-hd expression than their counterparts in mouse and chick. We discuss how this may underlie the diversification of CR cells and play a role in the evolution of pallial morphogenesis.
Materials and Methods
Tissue Samples and Processing
Following caesarian section, Cynomolgus monkey fetuses of known gestational time were deeply anaesthetized and perfused through the heart with 4% paraformaldehyde as described in Smart et al. (2002). All experiments were performed in compliance with the National and European laws as well as with institutional guidelines for animal experimentation.
After dissection, brains were placed in a fixative solution (4% paraformaldehyde [PFA] in phosphate buffered saline [PBS]) overnight at 4 °C, rinsed in PBS, and cryoprotected in sucrose. Tissue was then embedded in Tissue-Tek. Serial frozen sections (30 μm thick) were collected onto SuperFrost Plus glass slides, dried at least 10 min at room temperature, and stored with silica gel at −80 °C until in situ hybridization was performed.
Chick embryos from embryonic day 8 (E8; HH33–HH34) to hatching (P0) were used and staged according to Hamburger and Hamilton (1951). All animals were treated according to the regulations and laws of the European Union (86/609/EEC) and the Spanish Government (Royal Decree 223/1998; more recently Royal Decree 1021/2005) for care and handling of animals in research.
Upon extraction, the brains of earlier embryos (8–11 days incubation: E8–E11 or HH33–HH37) were dissected and fixed by immersion in 4% PFA in PBS at 4 °C for 24 h. Older embryos (from E12 or HH38) and P0 animals were first deeply anesthetized by either cold (embryos) or ethyl ether (P0) and perfused transcardially with 0.75% NaCl saline solution, followed by phosphate-buffered 4% PFA (pH 7.4). The brains were then dissected and postfixed overnight at 4 °C, embedded in 4% agarose in PBS, sectioned at 120 μm for in situ hybridization in the transversal, horizontal, or sagittal planes using a vibratome (Leica VT1000S), and subsequently processed as floating sections.
Time-pregnant mice (Swiss) were killed by cervical dislocation, and embryos were rapidly removed. Animals were treated according to the regulations and laws of the European Union (86/609/EEC) for care and handling of animals in research. After dissection, the brains were fixed overnight in PFA 4%, either embedded in 4% agarose in PBS and sectioned at 120 μm using a vibratome for processing as floating sections or cryoprotected and frozen on dry ice for cryostat sectioning.
cDNA Cloning and Probes
Total RNA from monkey embryo brain was reverse transcribed with random primers using AMV reverse transcriptase (Promega, France). Partial cDNA coding sequences for Lhx1 (640 bp), Lhx2 (586 bp), Lhx5(801 bp), Lhx6 (781 bp), and Lhx9 (513 bp) were amplified by polymerase chain reaction (PCR) using the following primers designed on human sequences (LHX1: NM_005568; LHX2: NM_004789; LHX5: NM_022363; LHX6: NM_014368; LHX9: NM_020204; reelin: NM_173054; and EMX2: NM_004098): Lhx1F 5′-CCACTGGCGAGGAACTCTAC-3′, Lhx1R 5′-AGGGTAGGTCCACTGGTGTC-3′; Lhx2F 5′-GGTCTTCCCTACTACAATGG-3′, Lhx2R 5′-GGTTGGTAAGAGTCGTTTG-3′; Lhx5F 5′-AATGTGTTCAGTGCTGCG-3′, Lhx5R 5′-AGTCGTAGTTGCTTCCCG-3′; Lhx6F 5′-CACGCCATCTGTCTGCTC-3′, Lhx6R 5′-CTGCTGAACGGGGTGTAG-3′; Lhx9F 5′-GCCTGAAGTGCTGTGAATG-3′, Lhx9R 5′-TCTGCGAGGGTGGATAAG-3′; EMX2F 5′-CAGCCCCATAAATCCGTT-3′, EMX2R 5′-TTGTTGCGAATCTGAGCCT-3′ Each PCR product was subcloned in pGEM-T easy vector (Promega) and fully sequenced.
The cDNAs from chick genes were cloned or purchased (cDNA ESTs purchased from ARK-genomics [Roslin Institute, Midlothian, UK] or Geneservice Limited [Cambridge, UK]). The purchased clones were obtained from the Biotechnology and Biological Sciences Research Council (BBSRC) ChickEST Database (Boardman et al. 2002) and have a corresponding Genbank accession number:
partial cLhx1 and cLhx5 cDNAs were isolated by degenerate or specific reverse transcriptase-PCR as described (Bachy et al. 2001). The starting material was total RNA extracted from hatched chick brain. Primers were as follows: Forward_cLhx1, AAYTGYTTYACITGYATGRTITG; Reverse_cLhx1, CKICKRTTYTGRAACCAIACYTG (where I is an inosine residue); Forward_cLhx5, CTGCTGAACGTCTTGGACAG; Reverse_cLhx5, GGTGCGAGATCATATCCGTG. PCR products (436 bp for Lhx1 and 993 bp for Lhx5) were cloned into pCRII-TOPO (InVitrogen, Carlsbad, CA) and sequenced. The cloned fragments encompass the sequence spanning from the LIM1 domain to the end of the homeodomain and correspond to positions bp 398–bp 834 of cLhx1 and to bp 49–bp 1042 of cLhx5, now accessible in Genbank (cLhx1: L35569; cLhx5: XM_001234552);
cLhx2 (bp 208–939; Genbank NM_204889);
cLhx9 (bp 596–1502; Genbank NM_205426); and
In Situ Hybridization and Immunohistochemistry
Digoxigenin- or fluorescein-labeled riboprobes were synthesized from PCR templates (Nguyen et al. 2001). LIM-hd factors and reelin distribution were investigated by in situ hybridization and/or immunohistochemistry on mouse brain cryostat sections at E11.5, E12.5, E13.5, and E14.5 and on monkey frontal cortex at E80. The in situ protocol was as previously described (Rétaux et al. 1999; Bachy et al. 2001) with adaptations described below. Chick free-floating sections were processed for in situ hybridization following a variation of the standard procedure using digoxigenin-labeled riboprobes (Medina et al. 2004; Abellan and Medina 2009).
For mouse double mRNA staining, cryostat sections (14 μm thick) were hybridized (65 °C, overnight) with an LIM-hd dig-labeled probe and a reelin fluorescein-labeled probe and washed under high stringency conditions. The slides were incubated overnight at 4 °C with the anti-fluorescein, alkaline phosphatase-conjugated antibody. After several washes, reelin mRNA labeling was revealed with Fast Red. The reaction was stopped with a glycine–HCl solution (pH 2.2) and followed by the second-step anti-digoxigenin alkaline phosphatase-conjugated antibody. The LIM-hd mRNA staining was revealed with NBT/BCIP (Roche, Mannheim, Germany).
For the simultaneous detection of LIM-hd mRNA and reelin protein on mouse sections, a reelin antibody (monoclonal G10; Chemicon [Millipore, Billerica, MA]; 1/1000) was added together with the anti-digoxigenin alkaline phosphatase-conjugated antibody. The labeled mRNA probe was revealed using Fast Red or NBT/BCIP. The reaction was stopped by fixation, and the reelin immunohistological staining was revealed by a secondary anti-mouse Alexa-488 fluorescent antibody (Molecular Probes, Invitrogen, Cergy-Pontoise, France).
In another set of experiments, thick vibratome sections of embryonic mouse brains were processed as free-floating sections for hybridization as described for chick.
For monkey staining, on the day of the experiment, sections were defrosted, fixed in 4% PFA, for 30 min at room temperature, washed 3 times in PBS, and acetylated in TEA solution (0.1 M TEA pH 8, 0.25% acetic anhydride) for 30 min. After 3 washes in PBS, slides were incubated in prehybridization buffer (50% formamide, Denhardt, sulfate dextran, tRNA) at 65 °C for 2 h and hybridized under coverslips with digoxigenin-labeled probes for 14–16 h at 65 °C in a humid chamber. After hybridization, coverslips were removed by bathing in 2× SSC/50% formamide solution at 65 °C for 15 min. Sections were then washed twice in 2× SSC/50% formamide solution (65 °C for 30 min) and twice in MABT (room temperature for 30 min). Before incubation overnight at 4 °C with alkaline phosphatase-coupled anti-digoxigenin antibody diluted to 1:1500, sections were preincubated in blocking buffer (blocking reagent, foetal veal serum in MABT) for at least 2 h at room temperature. Sections were washed 4 times in MABT solution, and then alkaline phosphatase staining was developed with NBT/BCIP including 10% polyvinyl alcool. The staining reaction was stopped by bathing 3 times in PBS, 20 min in 4% PFA, and 3 times in PBS. Double labeling for Lhx6 and γ-aminobutyric acid (GABA) was obtained as described above, using a rabbit anti-GABA antibody (Sigma; 1/3000).
For analysis, all slides were mounted with glycerol and observed with an ApoTome microscope (Zeiss, Germany).
LIM-hd Repertoire in Mouse CR Cells
In mouse pallium, we systematically analyzed LIM-hd factor expression in CR cells using double labeling for Lhx1/2/5/9 and reelin at 3 stages of embryonic (E) development.
Lhx1 signal was readily present in the medial to dorsal and ventral aspects of the marginal zone of the cortical wall and in the eminentia thalami from E11.5 and persisted in these same regions until E14.5 and E15.5 (Fig. 1A–D) except in the dorsal pallium (Fig. 1D). Lhx1 was also expressed in the marginal part of the cortical hem (Fig. 1C). This pattern suggested that Lhx1 is expressed in CR cells originating at the borders of the cortex. Indeed, from E11.5 to E13.5 and E15.5, a significant colocalization of reelin and Lhx1 was observed in cells of the cortical marginal zone. As shown in Figure 1E–G’, not all Lhx1-positive cells were reelin positive, and vice versa, suggesting that Lhx1 is expressed both by a subpopulation of CR cells and by other cell types in the marginal zone and upper cortical plate. Moreover, the fact that Lhx1 is not expressed in the dorsal pallial marginal zone at late stages suggests that it is switched off during the migration of CR cells from their point of origin in the cortical hem to their destinations all over the cortical mantle.
Lhx5 (Lhx1 paralog) displayed a broader expression when compared with Lhx1. It was expressed not only in the cortical hem (Fig. 1H at E11.5; Fig. 1J at E14.5), dorsal septum, retrobulbar area, and eminentia thalami but also throughout the cortical marginal zone including the dorsal pallium at all stages examined (Fig. 1H–J, arrows). Lhx5 was therefore detected in the exact putative location of origin and migratory path of CR cells during corticogenesis. When examined on sections double labeled for reelin, Lhx5 was indeed present early on at E11.5 in a subpopulation of reelin-expressing CR cells (Fig. 1K–K’) and was later strongly colocalized with the CR cell marker at stages E13.5 and E15.5 (Fig. 1L–L’,M,N). Notably, at E13.5, the colocalization was virtually 100%, meaning that all Lhx5-positive cells may be reelin positive and vice versa (Fig. 1L–L’). At E15.5, the Lhx5-positive population increased and, while still containing the reelin-expressing CR cells, likely corresponded to other additional cortical populations as well (Fig. 1M,N).
We next examined the expression patterns of another subfamily of LIM-hd factors, namely Lhx2 and Lhx9. Lhx2 was heavily expressed throughout the cortical neuroepithelium including the proliferation zone (Fig. 2A–C), at the exclusion of the cortical hem (Fig. 2C, see also Failli et al. 2002). At all stages examined however, Lhx2 was never expressed in the reelin-positive CR cells at the very margin of the cortical wall (Fig. 2D–F’). Lhx2 paralog Lhx9 showed much more restricted expression in the developing pallium, being expressed in specific layers (see also Rétaux et al. 1999) of the medial, dorsal, and ventral pallium (Fig. 2G–I). Lhx9 was not expressed in the cortical hem (Fig. 2I, see also Failli et al. 2002). Double-labeled sections revealed a possible but faint coexpression in reelin-positive cells in the marginal zone at E11.5 (Fig. 2J–J’). At later stages E13.5–E15.5, Lhx9 was no more present in reelin-positive cells (Fig. 2K–L’). Thus, from E11.5 to E15.5, mouse CR cells strongly and preferentially express Lhx1 and Lhx5 but not Lhx2 or Lhx9 (summarized in Table 1).
|In CR cells of|
|In the cortical hem of|
|In CR cells of|
|In the cortical hem of|
Note: (+), (++), and (+/−) indicate both apparent “relative” level of expression (note that in situ hybridization is not a quantitative method) and apparent number of expressing cells.
|lfb||Lateral forebrain bundle|
|NCS||Superficial part of caudal nidopallium|
|pTh||Prethalamus (previously ventral thalamus)|
|zli||Zona limitans intrathalamica|
|lfb||Lateral forebrain bundle|
|NCS||Superficial part of caudal nidopallium|
|pTh||Prethalamus (previously ventral thalamus)|
|zli||Zona limitans intrathalamica|
LIM-hd Repertoire in Macaque Monkey CR Cells
Since CR cells may be responsible for some aspects of cortical evolution in mammals and LIM-hd factors are suspected to work as a combinatorial code to govern cell and neuronal specification, we next sought to examine the LIM-hd repertoire of CR cells in primates. To this end, we isolated Lhx1/2/5/6/9 cDNAs in macaque monkey, together with reelin and another cortex-expressed gene, Emx2. We analyzed their expression in the developing cortical area 17 of E80 fetuses—a developmental stage that corresponds to ∼E15 in the mouse in terms of corticogenesis (Smart et al. 2002). The results are shown in Figure 3 and correspond to photographs of adjacent sections hybridized with the indicated probes as reelin double labeling was unsuccessful in our hands on macaque material. Reelin mRNA itself was expressed by numerous cells in the marginal zone of the monkey cortex, as expected for CR cells (Fig. 3A,A’), and gave an indication of CR cell localization, density, and approximate size in the cortical marginal zone of the monkey at E80. Emx2, a known marker of CR cells during mouse corticogenesis (Hevner et al. 2003), was expressed in a strongly compatible pattern to be also a marker of CR cells in the macaque monkey (Fig. 3B,B’). Examination of sections hybridized for Lhx1 (Fig. 3C,C’) and Lhx5 (Fig. 3D,D’) strongly suggest that monkey CR cells express these 2 LIM-hd factors, as they do in mice. In fact, Lhx1 and Lhx5 were preferentially expressed in cells of the marginal zone at E80, and Lhx1 was also very weakly expressed in the upper cortical plate (Fig. 3C). On the other hand, the 2 paralogs Lhx2 and Lhx9 were strongly expressed not only in the developing cortical plate but also in cells located in the marginal zone that exhibit features totally compatible for CR cells. Although Lhx2 signal was relatively low in these putative CR cells (Fig. 3E,E’; when compared with the expression level on the same section in cortical plate cells), Lhx9 signal was strong and comparable to that shown by Lhx1 or Lhx5 (Fig. 3F,F’). In sum, the LIM-hd repertoire of monkey CR cells is larger than in their rodent counterparts and includes the 4 LIM-hd members studied. As a negative control, we used Otx1 and found that it was not expressed by CR cells (but showed scattered expression in the medial cortical plate; data not shown).
As the cortical marginal zone is also the substrate for tangential migrations of interneurons originating in the subpallium, we used Lhx6, a selective marker of these migrating interneurons in the mouse (Lavdas et al. 1999; Alifragis et al. 2004), to compare its expression pattern with those of Lhx1/5/2/9. Macaque Lhx6 displayed strong expression in numerous and dense cell populations of both the cortical plate and the marginal zone (Fig. 3G), where it showed a very high level of colocalization with GABA (Fig. 3H–I’). Thus, in monkey also, Lhx6 is probably a marker for GABAergic cortical interneurons of subpallial origin. These data are fully compatible with the birthdating experiments reported by Zecevic and Rakic (2001) for cortical layer I in macaques. Lhx6 pattern in the marginal zone is significantly different in terms of cell body location (dispersed in the depth of the marginal zone), number (numerous), and GABA colocalization from that of Lhx1/2/5/9 (expressed in few cells of the upper marginal zone). These data reinforce the possibility that Lhx1/5 and Lhx2/9 are indeed expressed by monkey CR cells (Table 1).
LIM-hd Repertoire in Chick CR Cells
As the combinations of LIM-hd factors seem to vary between nonprimate and primate mammals, we next investigated the expression of Lhx1/5/2/9/6 in the chick dorsal telencephalon. Indeed, birds possess a mostly unlayered pallial wall, although they do have reelin-expressing subpial cells in the pallium, but these are significantly less numerous than those in mammals (Bernier et al. 2000; see also Fig. 5).
In contrast to mammals where Lhx5 seems to be a “preferential” CR cell-expressed LIM-hd factor, in chick embryos Lhx5 was only expressed in the septum, eminentia thalami, and a tiny hem previously identified in chick based on cWnt8b expression (Garda et al. 2002), but never in the marginal zone of the pallium (Fig. 4A–D), at the various stages examined from E8 to E14. Further, although Lhx1 is expressed in the mammalian medial pallium and marginal zone, it was not present in homologous structures of the avian pallium (Fig. 4E). Of note, in other telencephalic structures such as the septum or the eminentia thalami, which are additional supposed sources of CR cells, Lhx1 and Lhx5 were expressed in chick just like they were in mouse.
Concerning Lhx2 and Lhx9, a strong, specific, and unexpected expression of Lhx2 was observed in the ventricular part of the hem (Fig. 4F,F’; compare with Lhx5Fig. 4D). Lhx9 was absent from the hem (Fig. 4G). In other regions, Lhx2 pattern was similar to mammals, strongly expressed in the ventricular zone and differentiating pallium but excluded from the marginal zone (Fig. 4F). Lhx9 showed a restricted pattern in the dorsolateral, medial, and ventral pallium (Fig. 4G), also excluding the possibility of its presence in putative CR cells of the chick. Finally, chick Lhx6 pattern was assessed to be compared with the established pattern of its mouse ortholog (Grigoriou et al. 1998) and with the present results in macaque (shown in Fig. 3). Chick Lhx6 was transcribed in the pallidum, the subpallial region of origin for pallial GABA interneurons, and throughout the pallium with a salt and pepper distribution compatible with the distribution of tangentially migrated interneurons (Fig. 4H) (Abellan and Medina 2008, 2009).
We have analyzed the repertoire of LIM-hd transcription factor expression in the developing pallium/cortex of 3 amniote species, and we have focused on reelin-expressing CR cells because they are suspected of crucial importance for pallial morphogenesis and evolution. At least 4 LIM-hds, Lhx1/2/5/9, were transcribed in putative macaque monkey CR cells, whereas only 2 of these, Lhx1/5, were prominently expressed by mouse CR cells. There was no LIM-hd pattern compatible with their expression in the chick CR cells, at the exception of Lhx5 in the very small hem region of the pallium where these cells may originate in birds like in mammals. Here, we discuss the possibility that the differential combinatorial expression of LIM-hd factors, which are known to govern neuronal subtype specification events, in mammals versus nonmammalian amniotes and in primates versus nonprimate mammals, may underlie the diversification of CR cell types and subsequent specialization of pallial regions in terms of cytoarchitecture in various species.
Lhx5 and CR cells
Our data show that Lhx5 is a “preferential marker” of CR cells in mammals (primates and rodents) as it is present in the regions of origin (demonstrated only for rodents) as well as in migrating and migrated CR cells, covering the entire marginal zone (shown for both mammalian species).
In mouse, previous studies have shown that Lhx5 is expressed in the hippocampal primordium, the cortical hem, a crucial signaling center involved in cortical patterning (Grove et al. 1998), and CR cell precursors early in development and that this gene is essential for development of the hippocampus (Zhao et al. 1999). Our data corroborate the expression of mouse Lhx5 in these specific embryonic domains and cells and provide additional information at early stages. We show that Lhx5 is expressed in most of the proposed/shown sources of CR cells (Meyer et al. 1998, 2002; Meyer and Wahle 1999; Bielle et al. 2005; Garcia-Moreno et al. 2007) (reviewed by Soriano and Del Rio 2005), including not only the cortical hem but also the retrobulbar area, the dorsal septum, and the thalamic eminence. Moreover, our double-labeling experiments show that Lhx5 is expressed by reelin-positive cells, allowing identifying them as CR cells. We also found that Lhx1 is expressed in CR cells and in some of the sources of these cells, although less intensively than Lhx5.
In macaque monkey, although we could not perform the definitive double-labeling experiments to show colocalization with reelin, it is very likely that CR cells also express Lhx5, together with Lhx1, Lhx9, and Lhx2. To our knowledge, these findings constitute the first data on developmental characterization of this cell type in a primate species. There, we do not provide data on LIM-hd expression in the regions that are the sources of CR cells, but we present strong evidence for expression of a battery of LIM-hd transcription factors in the CR cells themselves. Again, like in mouse, Lhx5 seems to be the preferentially expressed LIM-hd factor in CR cells among the 4 studied. Of note, macaque CR cells probably also express Emx2.
Lhx5−/− mice show impaired hippocampal development, and their cortical hem and telencephalic midline are reduced or absent (as assayed by expression of Wnt5a or Bmp7, respectively), but corticogenesis itself appears surprisingly normal (Zhao et al. 1999). Interestingly, the anti-hem, a Dbx1-positive ventricular zone domain in the ventral pallium considered to be another source of CR cells (Bielle et al. 2005), does not express Lhx5 at any stage. If Lhx5 is indeed a determinant of CR cells—a hypothesis that will have to be tested functionally—normal corticogenesis in Lhx5−/− embryos may be explained by a rescue by reelin-expressing cells originating in the anti-hem.
In the chick, cLhx5 is expressed in the (very small) cortical hem, dorsal septum, and eminentia thalami. However, cLhx5 or cLhx1 is never observed in the marginal zone of the chick pallium, revealing a significant difference in the molecular attributes of CR cells between mammals and birds. Further, Lhx5 is not expressed in the medial pallium in Xenopus embryos (Bachy et al. 2001; Moreno et al. 2004). In the lamprey, Lhx15 is apparently present in the medial pallium (Osorio et al. 2005, 2006); however, the structure identified as medial pallium in lampreys probably derives from the same segmental division that produces the thalamic eminence (Pombal and Puelles 1999), suggesting that the so-called medial pallium of lampreys is not comparable to that of other vertebrates but is rather a diencephalic structure as the eminentia thalami. Thus, the true medial pallium does not appear to express Lhx5 in anamniotes, and this is strikingly correlated to an absence of condensation of reelin-expressing cells in certain areas of the pallium (see Introduction). Evolutionarily speaking, Lhx5 expression thus appears like an innovation of the amniote dorsal telencephalic midline (the hem) and may be partly linked to the emergence of a new cell type, the CR cells (Fig. 5). Additional expression of its paralog Lhx1 by CR cells themselves is a shared feature of mammals and may be linked to the condensation of this cell type.
Lhx1, Lhx5, Lhx2, and the Evolution of Cortical Hem
The cortical hem can be regarded as a lateral part of the dorsal midline of the mammalian telencephalon and as a crucial signaling center for cortical patterning and arealisation as it secretes morphogens of the Wnt and Bmp families (Grove et al. 1998; Shimogori et al. 2004). Strikingly, the LIM-hd “system” appears tightly linked to hem specification, as Lhx5 (present data), Lmx1a (Failli et al. 2002), Clim1 (LIM-hd cofactor), and Lmo2 (an LIM-only protein) (Bulchand et al. 2003) are expressed in the hem, while Lhx2 is excluded from the mammalian hem and negatively regulates hem specification (Bulchand et al. 2001). In fact, in Lhx2−/− mice, the hem is expanded at the expense of the adjacent pallium and thus covers the entire dorsal telencephalon (Bulchand et al. 2001). This phenotype results from the selector gene activity of Lhx2, which cell-autonomously specifies cortical fates at early stages of embryogenesis by suppressing hem and anti-hem fates at the medial and lateral borders of the cortex, respectively (Mangale et al. 2008). This is precisely where the vast majority of dorsal pallial CR cells originate. Complex reciprocal transcriptional regulations probably occur between LIM-hd system's members to establish hem versus pallium expression boundaries. In contrast to mammals, chick Lhx2 is expressed in the tiny hem region. Based on what we know on Lhx2 function in mammals, this “surprising” expression pattern may explain 2 features of the avian pallial development: the small size of the hem, due to Lhx2 selector function, and the absence of “fully differentiated” (or fully evolved?) CR cells, due to differences in specification of the hem and hem-originated cells.
Of note, CR cell abundance is relatively modest in other sauropsids such as crocodiles (Tissir et al. 2003) and this abundance correlates with the size of the cortical hem, which is minimal in sauropsids (lizards and chicks) and maximal in humans (Cabrera-Socorro et al. 2007). Our Lhx5 data suggest that the cortical hem is indeed very small in chicken, which correlates with CR cells being very scarce in the pallial regions of this animal. The varying size of the cortical hem in lizards, chicks, mice, and primates may influence quantitatively both the strength and extent of its influence as a signaling center and the number of CR cells it is able to generate and therefore may be one of the relevant factors of the evolution of cortical/pallial regions across vertebrates (Fig. 5).
An LIM Code for the Diversity of CR Cells and Cortical Evolution?
There are 13 LIM-hd family members in jawed vertebrates, all of them having crucial roles in regional and neuronal specification events during brain development. A so-called “LIM code” has been proposed, especially for the development of spinal motor neurons (reviewed in Shirasaki and Pfaff 2002), after the initial observation that different subtypes of motor neurons express different combinations of LIM-hd factors during embryogenesis (Tsuchida et al. 1994). The existence of such an LIM code is molecularly rendered possible by the assembly of hetero- or homo-multimeric LIM-hd protein complexes through interactions with the CLIM cofactors (Bach et al. 1997; Rétaux and Bachy 2002). Further, we have suggested that such LIM code may be responsible not only for the development and the generation of multiple cell types but also for the evolution of the nervous system (Bachy et al. 2002). Although very difficult to test experimentally, the present data further exemplify that changes in combinatorial expression of LIM-hd factors in a developing brain region (the hem) and in a developmentally important cell type born from this region (CR cells) are strikingly linked to changes in neuroanatomical and functional features of the pallium (Fig. 5).
When comparing pallial evolution between birds and rodents and primates, there are considerable variations in size and shape and connectivity and cytoarchitecture (Molnar et al. 2006; Medina and Abellan 2009). When considering the latter, CR cells and the reelin they produce have long been thought to be crucial for emergence of lamination, a key feature of the mammalian cortex (see Introduction). Although CR cells are present in all amniotes, the reelin signal in CR cells increases in mammals and is even higher in primates including humans, due to the structural differentiation of a dense and compact axonal plexus between the cortical plate and the marginal zone, which may serve as a reelin “reservoir” (Meyer and Gonzalez-Hernandez 1993; Marin-Padilla 1998). Also, in human and monkey, there are at least 2 types of reelin-producing cells, including the first-generated large CR cells and the later-generated and smaller SGL cells (Zecevic and Rakic 2001). The diversification, differentiation, and morphological complexity of CR cells is thus maximal in primates, and this correlates with their expression of a larger repertoire of LIM-hd transcription factors, supposed to convey higher molecular diversity and the possibility to promote emergence of novelties. In this respect, the findings of Pollard et al. (2006) that a region of the human genome showing significant evolutionary acceleration contains an RNA gene (HAR1F) that is coexpressed with reelin in human CR cells further illustrate that the changing molecular attributes of these very special cells probably have a deep impact on the evolution of the cerebral cortex.
This work was supported by Centre National de la Recherche Scientifique (CNRS), Agence Nationale pour la Recherche (ANR-Neuro [MIDLINE]) to S.R.; Agence Nationale pour la Recherche (ANR [06-NEUR-010-01]), SECO (EU FP7-216593) to C.D.; Spanish Ministry of Education and Science and FEDER (DGICYT-FEDER grant no. BFU2006-14804-C02-02/BFI) to L.M. A.M. was a postdoctoral fellow supported by CNRS.
Thanks to Philippe Vernier for his interest in the early stages of the monkey project and to Yohann Mérot for critical reading of the manuscript. Conflict of interest: None declared.