Report

Transcriptional Regulation of Cortical Neuron Migration by POU Domain Factors

See allHide authors and affiliations

Science  22 Feb 2002:
Vol. 295, Issue 5559, pp. 1528-1532
DOI: 10.1126/science.1067132

Abstract

The identification of pathways mediated by the kinase Cdk5 and the ligand reelin has provided a conceptual framework for exploring the molecular mechanisms underlying proper lamination of the developing mammalian cerebral cortex. In this report, we identify a component of the regulation of Cdk5-mediated cortical lamination by genetic analysis of the roles of the class III POU domain transcription factors, Brn-1 and Brn-2, expressed during the development of the forebrain and coexpressed in most layer II-V cortical neurons. Brn-1 and Brn-2 appear to critically control the initiation of radial migration, redundantly regulating the cell-autonomous expression of the p35 and p39 regulatory subunits of Cdk5 in migrating cortical neurons, withBrn-1(/−)/Brn-2(/−) mice exhibiting cortical inversion.

Development of the mammalian cerebral cortex involves both radial migration of postmitotic neurons arising from the dorsal telencephalon and tangential migration of inhibitory interneurons originating in the ganglionic eminence (1, 2). Analysis of murine mutations has partially defined two signaling pathways regulating radial migration of cortical neurons, one involving reelin, expressed in marginal zone Cajal-Retzius cells, which binds the receptors for reelin and activates the downstream effector molecule mDab1 in radially migrating neurons (3–5), and the second mediated by Cdk5 and its regulatory subunits, p35 and p39 (6, 7).

To assess the potential roles of the class III POU domain transcription factors Brn-1 and Brn-2 (8–10) in cortex development, we examined the consequences of the deletion of theBrn-1 (Web fig. 1, A to D) (11) andBrn-2 (10) genomic loci, both separately and together. Brn-1(+/−) heterozygous mice were phenotypically normal; Brn-1 gene–deleted mice were present at expected Mendelian ratios at birth but survived no longer than 36 hours postpartum, probably reflecting a defect in renal development affecting Henle's loop (12). Crosses of grossly normalBrn-1(+/−)/Brn-2 (+/−) mice generatedBrn-1(/−)/Brn-2(/−) mice at the expected 1/16 frequency, but they were not viable at birth.

In wild-type animals, Brn-1 and Brn-2 were normally expressed in the superficial cortical layers and ventricular zone of the neocortex at postpartum day 0 (P0), whereas Brn-1 alone was expressed in the hippocampal CA1 field (Fig. 1A) (13). A high-magnification view illustrating costaining of Brn-1 (red) and Brn-2 (green) in midlateral, upper cortical plate demonstrates numerous Brn-1+/Brn-2+ nuclei (yellow), with Brn-1+ nuclei (red) also evident (Fig. 1A, upper right panel). As expected, Brn-1 protein was undetected in the brain of Brn- 1(/−) mice, but the distribution of Brn-2 protein expression was unaffected. Neither protein was detected in the brains ofBrn-1(/−)/Brn-2(/−) mice (Fig. 1A). Neurons in Brn-1(/−) andBrn-1(/−)/Brn-2(/−) mice that normally would have expressed Brn-1 were marked by the presence of an mRNA transcript containing the 3′ untranslated region ofBrn-1 (Brn-1 3′UT) (Web fig. 1E) (11, 13).

Figure 1

Brn-1/Brn-2 expression and morphology. (A) Indirect immunofluorescence detection in adjacent series of wild-type (WT), Brn-1(/−), andBrn-1(/−)/Brn-2(/−) P0 coronal hemisections of Brn-1 and Brn-2 protein, counterstained with Hoechst 33258 in low-magnification views. (B) Thionin-stained wild-type and Brn-1(/−) P1 coronal sections of hippocampus. Scale bar: (A), 500 μm; (B), 125 μm. CA1, hippocampal plate; CP, cortical plate; DG, dentate gyrus; EL, ectopic cell layer; TC, transitional cortex; VZ, ventricular zone.

Although the cerebral cortex of Brn- 2(/−) mice is apparently normal (10), a substantial cellular disorganization was observed at embryonic day 18 (E18) in the hippocampus and adjacent transitional cortex of Brn-1(/−) mice, which became progressively more severe by P1, extending throughout the rostrocaudal axis of the hippocampus (Fig. 1B and Web fig. 2A) (11). By E18, mRNA and protein for Brn-1, but not Brn-2, were normally present in the pyramidal neurons of the CA1 field and adjacent transitional cortex (Figs. 1 and 2A) (8,9). At this stage, Brn-1+ migrating neurons destined for the CA1 field in the wild-type brains, detected by theBrn-1 3′UT probe, were also prominently distributed between the ventricular zone and CA1 field (Fig. 2A). In contrast,Brn-1 3′UT+ neurons accumulated as an ectopically positioned layer of densely packed cell bodies in the hippocampus of Brn-1(/−) orBrn-1(/−)/Brn-2(/−) mice between the ventricular zone and the CA1 field, with little or no signal detected within the CA1 field (Fig. 2A). In adjacent sections, immunofluorescent localization of the subplate between the ectopic cell layer and CA1 field with the molecular marker chondroitin sulfate proteoglycan (CSPG) (14) demonstrates that neurons that would have normally expressed Brn-1 protein are unable to migrate through the subplate to the CA1 field in single- and double-mutant mice (Fig. 2A and Web fig. 2B) (11). This migration defect appears limited to the hippocampus in theBrn-1(/−) brains, but the combined gene deletion of both Brn-1 and Brn-2 resulted in the extension of this phenotype throughout the neocortex (Fig. 2B). Here, the distribution of Brn-1 3′ UT hybridization signal was largely concentrated between the ventricular zone and subplate, rather than in the cortical plate, as observed in wild-type andBrn-1(/−) brains (Fig. 2B), consistent with a requirement for both Brn-1 and Brn-2 in the regulation of radial migration of neurons in the neocortex.

Figure 2

Subplate position in wild-type and mutant brain. (A) Hippocampus and (B) neocortex of adjacent series of P0 wild type, Brn-1(/−), andBrn-1(/−)/Brn-2(/−) analyzed by in situ hybridization with a probe specific for Brn-1 3′UT (white), counterstained with Hoechst 33258 (blue), and by indirect immunofluorescence detection of CSPG (green). (A) CSPG panels represent fourfold magnified view of labeled region in adjacent in situ analyses. (B) Brackets highlight Brn-1 3′UT signal in VZ. Scale bar: (A and B), 250 μm. CP, cortical plate; DG, dentate gyrus; EL, ectopic layer; IZ, intermediate zone; MZ, marginal zone; SP, subplate; VZ, ventricular zone.

We next used bromodeoxyuridine (BrdU) birth-dating analyses (13) to determine the specific cortical laminae in which Brn-1+ neurons are normally located and to ascertain the relative locations of these laminae in the brains of correspondingBrn-1(/−) andBrn-1(/−)/Brn-2(/−) littermates. Wild-type neurons labeled between E11 and E16 and analyzed between E16 and E19 revealed that a majority of cortical neurons born on E13 and thereafter express Brn-1 protein, with virtually no colabeling of Brn-1 and BrdU observed in neurons born at E11 and E12 (Web fig. 3, A to C) (11). Neurons born between E13 and E16 have been shown to contribute extensively to cortical layers II-V (15), suggesting that the neurons marked by the Brn-1 3′UT hybridization signal that accumulate beneath the subplate of Brn- 1(/−)/Brn-2(/−) cortex represent this population of later-born neurons. This was confirmed by BrdU-labeling studies of wild-type, single- and double-mutant littermates conducted from E11 to E14, which demonstrate the location of early-born neurons in the superficial cortical plate and later-born neurons in the deeper layers of the Brn- 1(/−)/Brn-2(/−) cortex. A representative example analyzed at P0 demonstrates an abnormal, superficial location of early-born subplate neurons, visualized by costaining of microtubule-associated protein 2A/2B (MAP2A/2B) (14) and CSPG, in Brn- 1(/−)/Brn-2(/−) neocortex, and, in adjacent sections, a majority of later-born, E13 labeled neurons positioned below the subplate (Fig. 3A). In contrast, in both wild-type andBrn- 1(/−) brains, subplate neurons were located deep within the cortex, whereas the majority of E13 BrdU-labeled neurons were positioned above the subplate, in the cortical plate (Fig. 3A).

Figure 3

Cell birth date and molecular marker analyses. (A and B) Indirect immunofluorescent detection of BrdU-labeled cells, CSPG and MAP2A/2B (A), and in situ hybridization serial analyses of molecular markers TLE3 (A) and TLE4, p75, and OTX-1 (B) in midlateral cerebral cortex of wild-type, Brn-1(/−), andBrn-1(/−)/Brn-2(/−) brain, counterstained with Hoechst 33258. The brackets in (A), illustrating BrdU-labeled neurons or TLE3-specific signal, highlight concentrations of labeled cells or in situ signal, respectively, located above the subplate (SP) in wild-type andBrn-1(/−) cortex and below the SP inBrn-1(/−)/Brn- 2(/−) cortex. The brackets in (B) illustrate in situ hybridization signal in deeper layers of cortex in wild type and Brn-1(/−) and in superficial cortical layers inBrn-1(/−)/Brn-2(/−) brains. Scale bar: (A and B), 100 μm. 2 to 6 are the cortical layers.

The cortical inversion displayed by layers II-V in the P0Brn-1(/−)/Brn-2(/−) brain was further documented with a series of lamina-specific molecular markers. Expression of TLE3 and TLE1, specific for layers II-V and II/III, respectively (16), was detected by in situ hybridization in the deeper region of the cortex inBrn-1(/−)/Brn-2(/−) brain, rather than in the superficial layers, as observed in wild-type and Brn-1(/−) littermates (Fig. 3A and Web fig. 3D) (11). In comparison, in situ hybridization signal specific for TLE4, OTX-1, and p75, which are expressed in cortical layer VI and in subplate neurons (16–18), was observed immediately adjacent to the marginal zone in double-mutant mice, whereas this signal was located deep in the cortical plate in wild-type and Brn-1(/−) brain (Fig. 3B). Distribution patterns of two additional markers further reflected abnormal organization of cortical lamination inBrn-1(/−)/Brn-2(/−) brain. The expression of EphA4 (19), which was ordinarily observed as two distinct layers corresponding to layers II/III and V/VI in wild-type and Brn-1(/−) brains (Web fig. 3D) (11), was altered in theBrn-1(/−)/Brn- 2(/−), with only one layer of signal observed in the cortical plate. Layer IV neurons, identified by the expression of ROR-β (18), were also disorganized and failed to form a cohesive layer in the upper cortical plate inBrn-1(/−)/Brn-2(/−) cortex (Web fig. 3D) (11). No differences in apoptotic cell death between wild-type and mutant cortex were detected by terminal deoxynucleotidyl transferase-mediated deoxyuridine triphosphate nick-end labeling (TUNEL) assay (12,13).

Thus, Brn-1 and Brn-2 together are required for cell-autonomous control of radial migration in neurons of cortical layers II-V. By comparison, layer VI neurons migrated into the preplate in theBrn-1(/−)/Brn- 2(/−) brain, splitting this structure to form both the marginal zone and subplate structures, indicating that these early-born neurons, in which we did not detect Brn-1 and Brn-2 expression, remain capable of normal radial migration. The inversion of cortical layers II-V observed in theBrn-1(/−)/Brn-2(/−) brain was similar to the migration defects reported in the gene deletions of CDK5/p35/p39(6, 7), whereas the phenotype reported for genetic disruption of elements of the reelin pathway (3–5) involved cortical inversion of layers II-VI.

We next examined the possibility that Brn-1 and Brn-2 might transcriptionally regulate components of these pathways. In situ hybridization analyses at E18 of elements of the reelin pathway (Web fig. 4A) (11) and immunohistochemical analysis of CDK5 expression (12) did not reveal deficiencies in the expression of these genes that are likely to be responsible for the observed phenotypes in the single or double gene-deleted mice. In contrast, in situ hybridization analyses demonstrated the expression of p35 to be markedly decreased, and that of p39 to be nearly absent, in the cortex of Brn-1(/−)/Brn- 2(/−) double-mutant mice, although the relative cellular densities assessed by thionin staining were similar (Fig. 4A and Web fig. 4A) (11). The expression of p35 and p39 mRNA was also nearly absent in the hippocampal CA1 fields of Brn-1(/−) andBrn-1(/−)/Brn- 2(/−) brain when compared with wild-type hippocampus (Fig. 4A). p39 and p35 remained expressed in areas where neither Brn-1 nor Brn-2 is normally expressed, including the hippocampal CA3 field of both mutant mice strains (Fig. 4A). The expression of p35 was also markedly reduced, to background levels, in the mitral cells of theBrn-1(/−)/Brn-2(/−) olfactory bulb (Web fig. 4B) (11), which normally express Brn-1 and Brn-2 (8, 9), and can be visualized here by the expression of reelin mRNA (20).

Figure 4

p35 and p39 expression and regulation. (A) p35 and p39 expression in E18 wild type,Brn-1(/−), andBrn-1(/−)/Brn-2(/−) midsagital cortex and hippocampus. Scale bar: 250 μm. (B) Transient cotransfection assay analyses of Brn-1 and Brn-2 expression plasmids with reporter plasmids containing 2x response elements or 285 base pairs of p39 5′ flanking sequence. PCR-mediated detection of p35 and p39 5′ flanking sequences in αBrn-2–immunoprecipitated chromatin prepared from Brn-2–expressing p19 cells.

Inspection of 5′ and 3′ sequences flanking both thep35 and p39 genes revealed a number of consensus Brn-1 and Brn-2 DNA-binding sites (Web fig. 4C) (11) (21). Brn-1 and Brn-2 were each capable of activating transcription in transient cotransfection assays (10) >10-fold on representative examples of these response elements, and both Brn-1 and Brn-2 activated transcription three- to fourfold from the 285–base pair 5′ flanking region of the p39 gene that encompasses two Brn-1/Brn-2 consensus binding sites (Fig. 4B). We analyzed potential occupancy of these sites in the P19 cell line, which expresses endogenous Brn-2 protein (12, 22), by chromatin immunoprecipitation (Fig. 4B) (23,24), finding that Brn-2 was specifically bound to both the p35 and the p39 regulatory regions, whereas amplification with oligomers specific for unrelated genes, such as Brn-3.2, failed to yield a polymerase chain reaction (PCR) product, although all pairs of oligomers amplified DNA product with equivalent ranges of sensitivity (Fig. 4B).

Additional phenotypes consistent with deficient CDK5 kinase activity (6, 7) were observed in theBrn-1(/−)/Brn-2(/−) cortex. One hundred sixty-kilodalton neurofilament and phosphorylated 200-kD neurofilament were not detected in axonal bundles in either Brn-1(/−) orBrn-1(/−)/Brn-2(/−) hippocampus, although phosphorylated 200-kD neurofilament was abundant in cell bodies in the CA1 fields of these brains (Web fig. 4D) (11) (12). Abnormalities in axonal outgrowth and projection were also observed associated with theBrn-1(/−)/Brn-2(/−) forebrain, as well as abnormalities in migration in other regions of the CNS, including a failure of normal olfactory bulb and cerebellar development (25).

The identification of a substantial population of Brn-1+, Brn-2 cortical neurons (Fig. 1A) raised the possibility that the deletion of Brn-1 alone might result in a partial disruption of cortical migration. Indeed, the reduced distribution of TLE3-specific signal in the cortical plate, together with decreased thickness of layer II/III EphA4 staining, and a more superficial position of ROR-β+ expression inBrn-1(/−) cortex, suggest a partial defect in the radial migration of a subset of neurons in cortical layers II/III (Fig. 3B and Web fig. 3D) (11). In accord with this interpretation, BrdU-labeled neurons demonstrate reduced numbers of E14 labeled neurons in the upper cortex at E17 inBrn-1(/−) brain but do not reveal a gross cortical inversion within the earlier-generated neurons (Fig. 5A and Web fig. 5A) (11).

Figure 5

Cortical migration defects inBrn-1(/−) brain. (A andB) BrdU-labeled (A) and immmunostaining of Tst-1+ (B) neurons in midlateral cortical sections of wild-type and Brn-1(/−) brains. (Cand D) Indirect immunofluorescent colocalization of Brn-1 and Tst-1 (C) or reelin (D) in wild-type P0 midlateral coronal cortex. (E) Reelin-specific in situ hybridization signal in high and low magnifications of P0 wild-type andBrn-1(/−) coronal cortex; arrows in low-magnification view of Brn-1(/−) indicate increased reelin expression in basal ganglia relative to wild type. (F) Immunoperoxidase detection of mDab1 expression in midlateral coronal wild-type P1 Brn- 1(/−) cortex. Scale bars: (B), 40 μm; (C), 50 μm; (A), 100 μm; (F), 125 μm; and (E), 250 μm.

Also consistent with a reduced layer II/III migration, E12 and E13 BrdU-labeled neurons analyzed at P0 distributed in more superficial positions in Brn-1(/−) cortex than in comparable wild-type regions (Web fig. 5B) (11), as did Lhx-6+ and Dlx-2+ interneurons (Web fig. 5D) (11) (2). Furthermore, although layer V neurons expressing Tst-1/Oct-6/SCIP (Tst-1) (8,9) were distributed similarly in the upper cortical plate of E18 Brn-1 (/−) and wild-type cortex (Web fig. 5C) (11), by P1, Tst-1+ neurons remained in a dorsal position in Brn- 1(/−) brains, partially extending into the marginal zone, rather than deeper in the cortical plate, as observed in the wild-type cortex (Fig. 5B). Because analysis of wild-type cortex revealed colocalization of Brn-1 with Tst-1 in layer V cortical neurons (Fig. 5C), both the observed superficial location of Tst-1 neurons inBrn-1(/−) cortex and the reduced migration of layer II/III neurons might reflect influences of a Brn-1–dependent, cell-autonomous mechanism analogous to that described in theBrn-1(/−)/Brn-2(/−) cortex.

Unexpectedly, the expression of reelin in a subpopulation of cortical plate neurons that colocalize with Brn-1 at P0 (Fig. 5D), and reportedly with markers of inhibitory interneurons in postnatal brain (20), was absent in the Brn- 1(/−) cortex, whereas reelin expression was unaffected in Cajal-Retzius neurons and even increased in basal ganglia areas (Fig. 5E), where Brn-1 expression is also normally observed (8, 9, 12). Potential roles, if any, for these Brn-1+, reelin-secreting neurons in cortical development remain to be defined, but we reproducibly found a detectable increase in mDab1-specific immunohistochemical staining in the Brn-1(/−) in the neocortical plate (Fig. 5F), consistent with the observation that the intracellular levels of mDab1 protein increase in the absence of reelin signaling (14).

Together, these studies link Brn-1 and Brn-2 to the regulation of radial migration of layers II-V, at least in part, through regulation of the p39/p35 regulatory components of CDK5, resulting in cortical inversion inBrn-1(/−)/Brn-2(/−) double gene-deleted mice and disorganized cortical lamination inBrn-1(/−) mice, where radial migration of layers II/III is delayed.

  • * To whom correspondence should be addressed. E-mail: rmcevilly{at}ucsd.edu, mrosenfeld{at}ucsd.edu

REFERENCES AND NOTES

View Abstract

Navigate This Article