Report

Control of Hippocampal Morphogenesis and Neuronal Differentiation by the LIM Homeobox Gene Lhx5

See allHide authors and affiliations

Science  14 May 1999:
Vol. 284, Issue 5417, pp. 1155-1158
DOI: 10.1126/science.284.5417.1155

Abstract

The mammalian hippocampus contains the neural circuitry that is crucial for cognitive functions such as learning and memory. The development of such circuitry is dependent on the generation and correct placement of the appropriate number and types of neurons. Mice lacking function of the LIM homeobox gene Lhx5 showed a defect in hippocampus development. Hippocampal neural precursor cells were specified and proliferated, but many of them failed to either exit the cell cycle or to differentiate and migrate properly.Lhx5 is therefore essential for the regulation of precursor cell proliferation and the control of neuronal differentiation and migration during hippocampal development.

The development of Ammon's horn and the dentate gyrus of the mammalian hippocampal formation is a multistep process controlled by a complex genetic program. During embryonic development, hippocampal precursor cells originate from the medial wall of the telencephalic vesicles. These cells proliferate in a primary germinal layer called the ventricular zone. Thereafter, they migrate out in a highly organized manner to their appropriate target positions (1). This leads to morphogenesis of the characteristic interlocking C-shaped structures of Ammon's horn and the dentate gyrus. After migration, the postmitotic cells continue to differentiate and eventually give rise to the various types of hippocampal neurons, such as pyramidal cells, granular cells, and interneurons. Later during development, after most of the postmitotic cells have been generated, the number of proliferating cells decreases in the ventricular zone, which eventually forms an ependymal layer lining the ventricle.

Lhx5 is a member of the LIM homeobox gene family that encodes a transcription factor (2, 3). We detected Lhx5 mRNA in the hippocampal precursor cells at embryonic day 10.5 (E10.5) and E11.5 (Fig. 1, A and B). Between E13.5 and E18.5 (E15.5 is shown in Fig. 1C), the Lhx5 mRNA expression became restricted to the marginal zone of the developing Ammon's horn and dentate gyrus. This Lhx5 domain corresponds to a region that contains the earliest differentiated Cajal-Retzius cells (4). Cajal-Retzius cells guide neuronal migration (5) and innervation of afferent axons (6) during the development of the hippocampal formation.

Figure 1

Lhx5 expression in the developing hippocampus and its targeted deletion. Expression ofLhx5 mRNA in the developing hippocampus at E10.5 (A), E11.5 (B), and E15.5 (C) detected by in situ hybridization (27). (D) Through homologous recombination in ES cells, exons 2, 3, and 4 of theLhx5 gene (top, numbered black boxes) encoding the second LIM domain and the homeodomain (7) were replaced by the Neomycin (Neo) gene (bottom) derived from the targeting vector (middle). (E) Southern (DNA) blotting and polymerase chain reaction (PCR) analyses of DNA of E18.5 embryos derived from a heterozygous mating. After Nde I digestion, a 5′ probe detected a 13.5-kb band for the wild-type Lhx5 allele and a 11.0-kb band for the targeted allele (arrowheads). Hybridization of Hind III–digested DNA with a 3′ probe revealed a 12.5-kb band for the wild-type allele and a 14.6-kb band for the mutant allele (arrowheads). With the same DNA samples as templates, PCR (using a pair of primers from the third exon of Lhx5 and a pair of primers fromNeomycin) amplified a 250–base pair (bp) fragment from the wild-type allele (top) and a 500-bp fragment from the mutant allele (bottom), respectively. Lanes 1 and 7, 4 through 6, and 2 and 3 represent samples from wild-type, heterozygous, and homozygous embryos, respectively. Scale bar in (A) represents 100 μm for (A) and 240 μm for (B) and (C). B, Bam HI; Bx, Bstx I; H, Hind III; K, Kpn I; N, Nde I; R, Eco RI. The targeting vector was constructed by replacing exons 2, 3, and 4 of the Lhx5 gene with a neomycin-resistance gene flanked by 2.3 kb (5′) and 2.5 kb (3′) of homologous sequences and by the thymidine kinase gene. The vector was linearized and electroporated into the R1 line of ES cells (28). Clones resistant to G418 (350 μg/ml) and gancyclovir (2 mM) double selection were screened by Southern hybridization with both a 5′ probe and a 3′ probe outside the flanking homologous sequences. The ES cells heterozygous for the targeted Lhx5 allele were injected into 129/Sv blastocysts to generate chimeric mice that transmitted the Lhx5 deletion through the germline. Chimeric mice were mated to wild-type CD1 to generate heterozygous animals that were crossed to produce F2 offspring for analysis.

In an effort to analyze the function of Lhx5, we deleted part of the Lhx5 gene encoding the second LIM domain and the homeodomain (7) in embryonic stem (ES) cells (Fig. 1, D and E). Mice heterozygous for the Lhx5 mutation appeared normal and fertile. Homozygous offspring from crosses between the heterozygous parents were born alive and appeared normal at first. However, most of them died within a few days after birth. At the time of weaning, only 45 out of a total of 838 progeny derived from heterozygous crosses were found to be homozygous mutants. Histological analysis of the mutant embryos revealed that the hippocampus was misformed, the choroid plexus of both the lateral ventricle and the third ventricle was missing, and the anterior callosal axons failed to cross the midline. Although these defects may not cause the lethality, the cyanotic appearance of dying animals suggested to us additional defects in respiratory control centers. However, we were unable to detect morphological defects either in the hindbrain or in other discrete regions of the forebrain (including the neocortex, basal ganglia, thalamus, and hypothalamus), midbrain, and spinal cord whereLhx5 is normally expressed. The cytoarchitectonic organization of all these regions appeared normal in mutant embryos as compared to wild-type controls. It is possible that functional compensation for Lhx5 is rendered, at least in part, byLhx1, a closely related LIM homeobox gene that is coexpressed with Lhx5 in these regions (2).

At E18.5, the Lhx5 homozygous mutant embryos showed histological defects in the hippocampal regions (Fig. 2, A and B). The ventricular zone was thicker than normal. Numerous cells were clustered in a region ventral to the lateral ventricle, but these cells failed to form the morphologically distinctive structures of Ammon's horn and the dentate gyrus. The fimbria and the hippocampal commissure, two major axon tracts of the hippocampal formation, were entirely missing from the mutant embryos. Consistent with the thickening of the hippocampal ventricular zone, bromodeoxyuridine (BrdU) pulse labeling at E18.5 showed that the number of proliferating cells in the mutant hippocampal ventricular zone was increased as compared to that in the wild-type control (Fig. 2, C and D). Using BrdU labeling, we also marked cells undergoing their final cell division at E13.5. By E18.5, many BrdU-labeled postmitotic cells were observed in the hippocampal region of mutant embryos (Fig. 2F). These cells migrated out of the ventricular zone, but they failed to position themselves properly to form the distinctive structures of Ammon's horn and the dentate gyrus that were observed in wild-type embryos (Fig. 2E).

Figure 2

Defects in morphogenesis and precursor cell proliferation in the developing hippocampus of E18.5Lhx5 mutant embryos (B, D, and F) as compared to wild-type controls (A, C, and E). (A and B) Hematoxylin and eosin staining of coronal sections through the hippocampal region. (C and D) Proliferating cells in the hippocampal ventricular zone labeled by anti-BrdU staining. (E andF) Distribution of postmitotic cells born at E13.5 in the hippocampal region of E18.5 embryos. Arrows in (A) through (F) point at the hippocampal ventricular zone. Scale bar in (E) represents 100 μm for (C) and (D) and 250 μm for (A), (B), (E), and (F). ah, Ammon's horn; dg, dentate gyrus; th, thalamus. To label proliferating cells, pregnant females were injected intraperitoneally with BrdU (100 μg per gram of body weight, dissolved in 0.9% NaCl) and killed 2 hours later. Embryos were dissected and fixed (27). Frozen sections (10 μm) were treated with 2M HCl and neutralized in sodium borate. BrdU was detected with an alkaline phosphatase–labeled antibody to BrdU (1 U/ml, Boehringer Mannheim). A similar procedure was followed for birth dating analysis, except that BrdU was injected at E13.5 and the embryos were collected at E18.5.

By E18.5, the transcript of the LIM homeobox gene Lhx2was detected in both the ventricular zone and the differentiation zone of Ammon's horn and the dentate gyrus in wild-type embryos (Fig. 3A). The expression of Lhx2 in the ventricular zone is consistent with its role in the proliferation of the hippocampal precursor cells (8). The Lhx2expression in the differentiation zone suggests that this gene may also support neuronal differentiation in the developing hippocampus. InLhx5 null mutant embryos, Lhx2 mRNA was detected in the thickened ventricular zone but not in postmitotic cells that had migrated out of the ventricular zone (Fig. 3B). At this stage, different types of hippocampal neurons could be identified in wild-type embryos by specific markers. Pyramidal cells in Ammon's horn were labeled by an antibody to glutamate receptor subunit GluR1 (Fig. 3C). Interneurons were stained for glutamate decarboxylase GAD67 (Fig. 3E). Granule cells in the dentate gyrus and the Cajal-Retzius cells were visualized with an antibody specific for the calcium-binding protein calretinin (Fig. 3G). In contrast, immunostaining for GluR1 (Fig. 3D), GAD67 (Fig. 3F), and calretinin (Fig. 3H) was diminished in comparable regions of the Lhx5 mutant embryos. However, postmitotic cells in these fields were stained in both wild-type (Fig. 3, I and K) and mutant (Fig. 3, J and L) embryos by antibodies directed against general neuronal markers such as class III β-tubulin (9), the microtubule associate protein MAP2, and the glial cell marker glial fibrillary acidic protein (GFAP). Thus, these postmitotic cells were initially able to acquire certain identities of neurons or glial cells, but they did not differentiate further into the various subclasses of hippocampal neurons in the absence of Lhx5 function. In order to determine whether the disruption of differentiation of these cells was accompanied by apoptosis, TUNEL (10) staining was performed. As in wild-type embryos, very few apoptotic cells were detected in mutant hippocampal regions.

Figure 3

Impaired hippocampal neuronal differentiation in Lhx5 mutant embryos. Expression of various molecular markers in the hippocampal regions in E18.5 wild-type (A, C, E, G, I, and K) and Lhx5 mutant embryos (B, D, F, H, J, and L). (A and B) Expression of Lhx2 mRNA detected by in situ hybridization (27). (Cthrough L) Immunohistochemical staining of GluR1 (C and D). GAD67 (E and F), calretinin (G and H), MAP2 (I and J), and GFAP (K and L). Arrows in (A) and (B) point at the hippocampal ventricular zone. Arrowhead in (B) points at the postmitotic cells that migrated out of the ventricular zone. Arrowheads in (C), (E), and (G) point at labeled pyramidal cells, interneurons, and the Cajal-Retzius cells, respectively. Arrowheads in (K) and (L) point at cells positive for GFAP. Scale bar in (A) represents 380 μm for (A) and (B) and 200 μm for (C) through (L). g, granule cells of the dentate gyrus; th, thalamus. For immunohistochemistry, frozen sections (10 μm) from paraformaldehyde-fixed samples were incubated with primary antibodies. Reacting antigens were detected with the ABC elite kit (Vector Lab) or the histomouse kit (Zymed). The primary antibodies were as follows: anti-GluR1 (1 μg/ml), anti-GAD67 (1:1000, Chemicon), anti-calretinin (1:1500, RDI), anti-MAP2 (1:250, Sigma), and anti-GFAP (1:500, DAKO).

Early in development, several homeobox genes, includingLhx2, Emx2, and Otx1, are expressed in the developing hippocampal anlagen, and some of these genes are required for the specification of the region (8,11, 12). We examined the expression of these genes inLhx5 mutant embryos. At E12.5, all three of these genes were expressed in the hippocampal anlagen in both wild-type (Fig. 4, A through C) and mutant (Fig. 4, F through H) embryos, which indicates that the hippocampal precursor cells were specified after disruption of Lhx5. In the mutant embryos, the domains of Lhx2, Emx2, andOtx1 expression expanded ventrally into the region of the telencephalic choroid plexus, and morphogenesis of the choroid plexus was impaired. Signaling molecules of the Wnt andBmp families have been implicated in patterning the medial telencephalic wall to form the hippocampal anlagen and the choroid plexus, because these molecules are expressed at the border between these two morphologically distinctive structures (13,14) (Fig. 4, D and E). In support of this idea, expression of Wnt5a (Fig. 4I), Bmp4, andBmp7 (Fig. 4J) was diminished in this specific region inLhx5 null mutant embryos.

Figure 4

Specification of hippocampal precursor cells inLhx5 mutant embryos. Expression of Lhx2(A) and (F), Emx2 (B) and (G), Otx1 (C) and (H), Wnt5a (D) and (I), andBmp7 (E) and (J) in E12.5 wild-type (A through E) and Lhx5 homozygous (F through J) embryos detected by in situ hybridization (27). Arrowheads in (A) through (C) and (F) through (H) point at the hippocampal anlagen. Arrows in (A) through (C) and (F) through (H) point at the choroid plexus in a wild-type embryo and the ventrally expanded hippocampal domain in a mutant embryo, respectively. Arrowheads in (D) and (E) point at the regions where Wnt5a (D) and Bmp7 (E) were expressed. Scale bar in (J) represents 200 μm for all panels.

Previous experiments have shown that other members of the LIM homeobox gene family play crucial roles in the differentiation of distinct cell types in various organisms (15–18). In mice, for example, Islet1 is essential for the differentiation of motor neurons in the spinal cord (15), and Lhx3is required for the differentiation of the pituitary cell lineages (16). More recently, it has been observed thatLhx3 and Lhx4 together control the axon projection of subtypes of motor neurons as well as their exact soma position in the developing spinal cord (17). Our data suggest that Lhx5 may play an analogous role in the developing forebrain.

Defects in hippocampal development have been observed in mice carrying null mutations in a variety of genes. Functional ablation of the homeobox gene Emx2 (11) or Lhx2(8) leads to an early arrest of hippocampal development as precursor cells fail to be specified or to proliferate. Mutations inReeler (5), Mdab1(19), Cdk5 (20), P35(21), and Pafah1b1 (22) impair neuronal migration, resulting in a disorganization of cells in Ammon's horn and the dentate gyrus. Our results show that Lhx5 is required for differentiation of the various types of hippocampal neurons. Together, these studies reveal an intricate genetic program underlying the assembly of complex hippocampal structures.

Expression of Lhx5 after E13.5 in the Cajal-Retzius cells raises the possibility that the impairment in hippocampal morphogenesis observed in Lhx5 mutant embryos could result from a lack of function of those cells. The gene Reeler is expressed in these cells (5), in keeping with the possibility that this gene might be a downstream target of Lhx5.However, the defects in morphogenesis as well as in neuronal differentiation of Ammon's horn and the dentate gyrus seen in theLhx5 mutant embryos are more severe than those ofReeler (5), Mdab1 (19),Cdk5 (20), P35 (21), andPafah1b1 (22) mutants, which suggests thatLhx5 may control a different pathway. The Lhx5 knockout mice provide a model to further understand the molecular and cellular mechanisms underlying the formation of Ammon's horn and the dentate gyrus and their functions in cognition, learning, and memory.

  • * Present address: Department of Biological Sciences, University of Tokyo, Tokyo 113, Japan.

  • To whom correspondence should be addressed. E-mail: hw{at}helix.nih.gov

REFERENCES AND NOTES

View Abstract

Navigate This Article