|Home | About | Journals | Submit | Contact Us | Français|
Cortical malformations are important causes of neurological morbidity, but in many cases their etiology is poorly understood. Mice with Foxc1 mutations have cellular defects in meningeal development. We use hypomorphic and null alleles of Foxc1 to study the effect of meningeal defects on neocortical organization.
Embryos with loss of Foxc1 activity were generated using the hypomorphic Foxc1hith allele and the null Foxc1lacZ allele. Immunohistologic analysis was used to assess cerebral basement membrane integrity, marginal zone heterotopia formation, neuronal overmigration, meningeal defects, and changes in basement membrane composition. Dysplasia severity was quantified using two measures.
Cortical dysplasia resembling cobblestone cortex, with basement membrane breakdown and lamination defects, is seen in Foxc1 mutants. As Foxc1 activity was reduced, abnormalities in basement membrane integrity, heterotopia formation, neuronal overmigration, and meningeal development appeared earlier in gestation and were more severe. Surprisingly, the basement membrane appeared intact at early stages of development in the face of severe deficits in meningeal development. Prominent defects in basement membrane integrity appeared as development proceeded. Molecular analysis of basement membrane laminin subunits demonstrated that loss of the meninges led to changes in basement membrane composition.
Cortical dysplasia can be caused by cellular defects in the meninges. The meninges are not required for basement membrane establishment but are needed for remodeling as the brain expands. Specific changes in basement membrane composition may contribute to subsequent breakdown. Our study raises the possibility that primary meningeal defects may cortical dysplasia in some cases.
Cortical dysplasia is an important cause of neurologic morbidity. Advanced imaging techniques are increasing recognition of dysplasia and other structural brain abnormalities in patients with intellectual disability, global developmental delay, and epilepsy1. Specific treatment, such as resective surgery for focal cortical dysplasia in epilepsy2, may help some patients. However, for most there is only symptomatic treatment. Thus, understanding dysplasia pathogenesis is critical to improving diagnostic and treatment options.
Dysplasia occurs when the exquisitely regulated process of corticogenesis is perturbed at a critical point in embryonic life. The meninges are increasingly recognized as critical regulators of cortical development due in part to the unique position of the meninges adjacent to the superficial cortex with access to the forming cortical layers as well as radial glia endfeet. The meninges comprise three distinct cellular layers covering the brain: the outer dura contacts the inner table of the skull, the inner pia adheres to the surface of the brain, and the netlike fibers of the arachnoid connect the two3. The meninges produce important regulatory factors for brain and skull development including CXCL12, regulating Cajal-Retzius cell and interneuron migration4-7, TGF-β and FGF-2, regulating skull development8, 9, and all-trans retinoic acid, regulating cortical neurogenesis10.
We previously identified a hypomorphic mutation, hith, in the forkhead transcription factor, Foxc1 with defective forebrain meningeal development and neocortical malformations characterized by cerebral basement membrane (CBM) breakdown and neuronal overmigration10, 11. Foxc1 is expressed strongly by all components of the meninges and is not expressed within the brain except for perivascular pericytes. Thus, Foxc1 is thought to regulate brain development indirectly by controlling meningeal development.
Here, we investigate how defects in meningeal development promote CBM breakdown and dysplasia formation. Using Foxc1 mutants, we find that more severe meningeal defects accelerate CBM breakdown, promoting formation of cortical dysplasia during development. Extensive meningeal defects in these mutants precede CBM breakdown and dysplasia formation. Our study raises the possibility that primary meningeal pathology might play a role in dysplasia formation.
Foxc1lacZ/+ (T. Kume, Vanderbilt University) and Foxc1hith/+ animals were maintained and genotyped as described10, 11. CXCR4-EGFP BAC transgenic animals (Gene Expression Nervous System Atlas (GENSAT) Project, NINDS Contracts N01NS02331 & HHSN271200723701C to The Rockefeller University (New York, NY)) were maintained on a CD-1 background. Breeding with Foxc1lacZ/+ animals allowed production of Foxc1lacZ/lacZ;CXCR4-EGFP/+ embryos. Timed embryos were obtained by examining for vaginal plugs in mated females. The morning of plug detection was designated E0.5. Mouse husbandry was done using protocols approved by the University of California, San Francisco, Committee on Animal Research in facilities approved by the American Association of Laboratory Animal Care.
Timed embryos were harvested from pregnant dams after isofluorane anesthesia and cervical dislocation. Whole embryonic heads were either fixed overnight at 4°C in 4% paraformaldehyde (PFA), cryoprotected, and frozen in OCT (Sakura Finetek), or flash frozen in OCT. 12μm cryosections were cut and stored at -80°C until needed.
For antibodies using PFA fixed tissue, tissue was boiled for 10 min in citrate buffer (except for GFP staining), blocked with 10% lamb serum and 0.3% Triton X-100, then incubated with primary and appropriate secondary antibodies in 2% lamb serum and 0.3% Triton X-100. For antibodies requiring fresh frozen sections, tissue was fixed in methanol at -20°C for 5 minutes, blocked in 1% BSA for 5 minutes, and incubated with primary and appropriate secondary antibodies in 1% BSA. Nissl staining followed antibody incubation using NeuroTrace (Invitrogen) at 1:100 dilution. Isolectin B4 (Invitrogen) was used at 1:100 dilution. These antibodies and dilutions were used on PFA fixed tissue: rabbit anti-laminin, 1:75, Sigma; rabbit anti-collagen IV, 1:200, Abcam; rat anti-CTIP2, 1:800, Abcam; mouse anti-nestin, 1:500, Chemicon; chicken anti-GFP, 1:500, Aves Labs; mouse anti-TuJ1, 1:1000, Covance; mouse monoclonal IIH6 against α-dystroglycan, 1:100, Millipore; rabbit anti-Zic, 1:5000, gift of J. Aruga, RIKEN Institute (required use of the Tyramide Signal Amplification Kit (Invitrogen)). These antibodies and dilutions were used on fresh frozen tissue: rat anti-entactin, 1:200, Abcam; rat anti-laminin α1, 1:10, Santa Cruz Biotechnology; rat anti-laminin α2, 1:50, Santa Cruz Biotechnology; rabbit anti-laminin α4, 1:400, gift of T. Sasaki, Oregon Health and Science University; rabbit anti-laminin α5, 1:500, gift of J. Miner, Washington University School of Medicine. Images were acquired with a digital CCD camera and QCapturePro software (QImaging) on a Nikon epifluorescent microscope.
Coronal sections of E18.5 embryos were immunostained with CTIP2 and counterstained with 4′,6-diamidino-2-phenylindole (DAPI). Mosaic images for counting were generated using Photoshop (Adobe). Hemispheres without significant hemorrhagic disruption were selected. DAPI allowed identification of the cortical plate (CP) and MZ. For each hemisphere, the total number of heavily labeled CTIP2 positive cells at the superior border of the CP or in the MZ was counted.
To measure the proportion of abnormal MZ, the same tissue sections as above were analyzed. Using ImageJ (NIH), the length of the neocortex and the total length of the MZ that contained frank heterotopias or abnormal clusters of cells were measured. The proportion of the MZ that appeared abnormal was calculated as (total length of abnormal MZ)/(length of cortex) × 100.
An average of 2.25 hemispheres/embryo from three embryos from different litters were analyzed. Hemispheres were at comparable anatomic levels. Mutants were compared to age-matched controls with unpaired t-tests employing SigmaPlot (Sigmaplot Software).
We hypothesized that disruption of meningeal function, caused by stepwise reduction of Foxc1 activity, would promote formation of cortical dysplasia characterized by CBM breakdown and neuronal overmigration. As a first step, we examined the timing of CBM breakdown in an allelic series of Foxc1 mutants (using combinations of the hypomorphic Foxc1hith and the null Foxc1lacZ alleles) from E12.5 to E18.5 by staining for laminin (Fig 1). Over the medial cortex, the CBM was apparently intact at E12.5 in all genotypes (Fig 1A-D). Areas without apparent CBM overlying the cortex appeared at E14.5 only in Foxc1hith/lacZ and Foxc1lacZ/lacZ mutants (Fig 1G-H) but not in the Foxc1hith/hith mutants (Fig 1F). Defects worsened in Foxc1hith/lacZ and Foxc1lacZ/lacZ mutants at E16.5 and E18.5, and at both ages were more extensive in Foxc1lacZ/lacZ mutants (Fig 1L, P) than in Foxc1hith/lacZ mutants (Fig 1K, O). In Foxc1hith/hith mutants, breaks in the CBM were absent at E16.5 (Fig 1J), and were uncommon even at E18.5 (Fig 1N). Dilated, abnormal blood vessels were also prominent (compare Fig 1O-P with M). Double-immunostaining with laminin and TuJ1 confirmed immature neurons had overmigrated through CBM breaks (Supplementary Fig 1A-F).
To confirm the CBM defect involved multiple CBM components, we examined expression of additional CBM proteins collagen IV and nidogen-112, 13 (Supplementary Fig 1G-R). Both proteins outlined intact CBM at E12.5 in all genotypes. At E18.5, they recapitulated the pattern of CBM breakdown seen with laminin staining in the mutants. The nidogen-1 antibody required fresh frozen tissue that caused artificially fractured CBM in controls (Supplementary Fig 1M, P), which was easily distinguished from the dramatic disorganization seen in mutants. Abnormal glycosylation of α-dystroglycan is associated with breakdown of the basement membrane and neuronal migration defects seen in congenital muscular dystrophies14, 15. However, in E16.5 Foxc1lacZ/lacZ mutants, there was no clear difference in staining with monoclonal antibody IIH6 (an antibody that does not recognize hypoglycosylated α-dystroglycan16) implying that the glycosylation of α-dystroglycan was probably intact in our mutant mice (Supplementary Fig 2). These data indicate that dosage dependent meningeal defects, caused by loss of Foxc1 function, promote CBM breakdown and neuronal overmigration as development proceeds, through a different biochemical pathway than the congenital muscular dystrophies.
In cobblestone cortex, CBM breakdown leads to neuronal overmigration forming dysplasia14. Because worsening meningeal defects promote CBM breakdown, we predicted formation of cortical dysplasia would accelerate as meningeal defects worsen. We therefore performed a detailed analysis of the timing and location of cortical dysplasia formation (Fig 2). Dysplastic cortex was most prominently seen in the dorsomedial cortex in E18.5 mutant embryos (Fig 2A-D). This was absent at E12.5 but small dysplastic areas appeared at E14.5 in the severely affected Foxc1lacZ/lacZ mutants (data not shown). At E16.5, dysplasia was prominent in the Foxc1hith/lacZ and Foxc1lacZ/lacZ mutants (Fig 2G-H), but was much less apparent in the Foxc1hith/hith mutant (Fig 2F). At E18.5, dysplasia was obvious in all mutants. Thus, at all ages, the cortical disorganization and severity of dysplasia worsened as Foxc1 activity decreased.
We assessed the severity of cortical dysplasia by measuring the degree of neuronal overmigration and the extent of MZ heterotopias (Fig 3). Normally, neurons expressing CTIP2 form a well-organized layer in cortical layers Vb and VI17 (Fig 3A). In the mutants, extensive neuronal overmigration occurred in areas of CBM breakdown. Many CTIP2 positive neurons were found in the upper CP and MZ, intermixed with CBM fragments (Fig 3B). In areas of apparently intact CBM, deep layer neurons either remained in place, were displaced upward mildly in the CP, or formed much smaller clusters in the MZ (data not shown). Quantification of displaced CTIP2+ neurons at the upper border of the CP or within the MZ was performed for each mutant genotype (Fig 3C-G). Displaced neurons had high expression of CTIP2 and were easily distinguished from the normal population of lightly labeled cells in the MZ (Fig 3C). In the Foxc1hith/hith mutant, CTIP2+ neurons were displaced upward in the CP but did not enter the MZ in great numbers and the heterotopias did not contain many CTIP2+ neurons (Fig 3D). In contrast, the Foxc1hith/lacZ and Foxc1lacZ/lacZ mutants had large MZ heterotopias containing many CTIP2+ neurons (Fig 3E-F). The number of CTIP2+ neurons bordering on or within the MZ per hemisphere was significantly greater in mutants compared to control. Severely affected mutants had many more displaced CTIP2+ neurons than the Foxc1hith/hith mutant (Fig 3G).
We also assessed the severity of heterotopia formation by measuring what proportion of the MZ contained heterotopic clusters of CTIP2+ neurons (Fig 3H). All mutants had a higher proportion of abnormal MZ compared to control. Paralleling results seen with CTIP2+ neuron overmigration, the severely affected mutants had higher proportions of abnormal MZ than the Foxc1hith/hith mutant. The Foxc1lacZ/lacZ mutant had a statistically lower extent of abnormal MZ than the Foxc1hith/lacZ mutant (p<0.05), likely due to its dramatically increased neocortical length and severe loss of post-mitotic neurons 10. These results indicate that primary pathology of the meninges in Foxc1 mutants promotes formation of cortical dysplasia.
Disruption of RG organization is believed to be a key factor in the development of cortical dysplasia with CBM breakdown and neuronal overmigration13, 18-21. Thus, we postulated that RG disorganization should also be present in Foxc1 mutants. The glia limitans was visualized at E16.5 by double-immunostaining with nestin22 and collagen IV (Fig 4). At this age, both an extensive network of RG fibers23 and extensive CBM breakdown occur. RG fibers were absent from CBM gaps in Foxc1hith/lacZ or Foxc1lacZ/lacZ mutants, but extended to fragments of CBM bordering gaps (Fig 4C-D). In control and Foxc1hith/hith embryos RG fibers extended to the CBM (Fig 4A-B). Disruption of RG fiber extension was more extensive in the Foxc1lacZ/lacZ mutant than the Foxc1hith/lacZ mutant and was primarily medial, corresponding with areas of CBM breakdown.
Clarifying the timing of the appearance of meningeal defects with the onset of cortical dysplasia should help establish the pathologic link between the two. We assessed the extent of meningeal defects at E12.5, before dysplasia is apparent. Double-immunostaining for Zic proteins10, 24 and laminin identified meningeal fibroblasts and the CBM, respectively (Fig 5). A well-defined layer of meningeal fibroblasts covered the CBM over the telencephalon in controls (Fig 5A-C). Fibroblast nuclei were slightly superficial to the CBM. In addition to meningeal fibroblasts, Cajal-Retzius cells also express a Zic family member and are thus stained, but are found in the superficial MZ24. Less intensely stained Zic+ cells, likely representing neural crest cells contributing to the frontal bones25, 26, overlaid the brain but lay outside the layer of meningeal fibroblasts. Foxc1hith/hith mutants had a small defect over the dorsomedial neocortex (Fig 5D-F) where meningeal cells were absent. Meningeal coverage remained robust and intact over the lateral cortex. Foxc1hith/lacZ mutants had a larger meningeal defect (Fig 5G-I) from the dorsomedial to the basolateral cortex. The Foxc1lacZ/lacZ mutants had the most severe meningeal defect (Fig 5J-L), with no meningeal fibroblasts clearly identified over the cortex and coverage by meningeal fibroblasts restricted to the basal telencephalon. Cajal-Retzius cells, revealed by the use of a CXCR4-EGFP transgene27, remained in the MZ in the Foxc1lacZ/lacZ mutants despite the absence of meningeal fibroblasts over the cortex (Supplementary Fig 3). Importantly, and somewhat surprisingly, we did not see apparent breakdown of the CBM or dysplasia formation in the mutants at this age. Our data indicated extensive meningeal defects appeared well before CBM breakdown or dysplasia arose, intact meninges were dispensable for initial establishment of the CBM, and, at early stages of development, Cajal-Retzius cells do not require meninges to localize to the MZ.
Laminin subunits α1 and α2 are known components of the embryonic CBM28. In adults, there is evidence that meningeal cells and astrocytes produce laminin α1 and α2 and endothelial cells synthesize laminin α4 and α529. We wondered whether meningeal defects cause differential changes in laminin subunit composition during development, which might be a potential mechanism for CBM instability. Therefore, we studied the composition of laminin α subunits in the CBM using specific antibodies (Fig 6). In controls, laminin α1, α2, and α5 had strong expression within the CBM. As expected, laminin α4 appeared to be expressed mainly in blood vessels30. Interestingly, we found that in mutant mice, even with large meningeal defects, the CBM had strong expression of laminin α1, α1, α2, and α5. These subunits were found in an intact CBM in all genotypes at E12.5. At E18.5, the CBM degenerated in Foxc1hith/lacZ and Foxc1lacZ/lacZ mutants, consistent with our prior results (Fig 1), leaving fragments of brightly labeled CBM (arrows, Fig 6O, R, X).
Examination of controls showed laminin α1 and α2 expression in a reticulated pattern just superior to the CBM at E12.5 (arrows, Fig 6A, D) and E18.5 (Fig 6M, P) while in Foxc1 mutants this pattern was altered with a significant decrease in laminin α1 expression and increase in laminin α2 expression (Fig 6C, F, O, R). Laminin α2 is expressed in brain capillaries31, so its increase is consistent with blood vessel overgrowth. The loss of reticulated laminin α1 expression paralleled the loss of meningeal cells, which are ideally located to regulate expression of laminin subunits in this area (see Fig 5B).
In parallel with the increased reticulated laminin α2 expression, laminin α4 expression was increased in the mutants at E12.5 and especially at E18.5. It became quite disorganized, consistent with formation of large, disorganized blood vessels (Fig 6H-I, T-U). Laminin α5 lacked the reticulated pattern superior to the CBM seen with laminin α1 and α2 (Fig 6J, V), but was expressed in some of the blood vessels within the brain and superior to the CBM in both mutants and control (data not shown and 6W).
Double-staining with laminin subunits and isolectin B4, an endothelial cell marker32, in E18.5 control and Foxc1lacZ/lacZ mutants confirmed that increased laminin α2 expression was associated with abnormal blood vessel proliferation (Fig 7). Mutants had increased numbers of endothelial cells, which generally expressed laminin α2 and not laminin α1 (Fig 7F, L), suggesting that meningeal cells produce laminin α1, and endothelial cells produce laminin α2. The reticulated pattern of staining superior to the CBM noted with both laminin subunits overlapped with isolectin staining, likely due to co-localization of meningeal and endothelial cells.
We have shown that primary meningeal defects, induced by Foxc1 mutation, cause cortical dysplasia characterized by CBM breakdown, neuronal overmigration, RG detachment, and formation of MZ heterotopias. Extensive meningeal defects are established prior to the onset of dysplasia and meningeal defects are associated with changes in the expression of laminin subunits (namely loss of α1 expression).
The dysplasia in the Foxc1 mutants resembles cobblestone cortex, in which CBM breakdown is thought to initiate a process leading to neuronal overmigration and heterotopia formation14. Regulation of CBM development and stability is likely to be dynamic and complex. The CBM, like other basement membranes (BMs), consists of thin protein sheets. Major components are laminins, collagen IV, nidogens and perlecan, a heparan sulfate proteoglycan. BMs are formed from two independent networks of laminin and collagen IV, while nidogens and perlecan provide stability. Additional minor components contribute to cross-linking of the networks and interaction with the interstitial matrix. BMs have important roles in providing tissue stability and regulating access to growth factors30, 33.
The Foxc1 mutants in our study differ significantly from previously reported mutants with cobblestone cortex, which are caused either by mutations in structural CBM components13, 34, CBM receptors 18, 20, 35, or downstream signaling molecules19, 21. Importantly, similar dysplasia is characteristic of dystroglycanopathies, caused by abnormal processing of α-dystroglycan, a molecule needed for interaction of cells with the extracellular matrix. Dystroglycanopathies include Fukuyama congenital muscular dystrophy, muscle-eye-brain disease, and Walker-Warberg syndrome14, 15. Foxc1, however, is expressed in meningeal fibroblasts and is not a known regulator of CBM structural or signaling molecules, and loss of expression does not affect α-dystroglycan glycosylation. Reduction or loss of Foxc1 likely impairs migration of meningeal cells, which are then not in position to provide a critical function needed for CBM stability and remodeling. In addition to directly impacting CBM development, loss of meningeal cells may remove critical secreted factors, analogous to SDF-1 and retinoic acid, needed to help migrating neurons maintain proper positioning. Interestingly, compensatory mechanisms must exist to partially maintain neuronal positioning for at least some cell types in the face of meningeal loss. We found that Cajal-Retzius cells were properly positioned early in development in the Foxc1 mutants, even though the meninges provide an important chemoattractant for these pioneer neurons4, 7.
Extensive meningeal defects are seen in the Foxc1 mutants prior to CBM breakdown and dysplasia formation. Therefore, intact meninges are dispensable for initial formation of apparently intact CBM, and are likely more important in maintenance and stability of the CBM. An attractive idea is that the meninges regulate CBM remodeling, a process which requires controlled degradation and synthesis of extracellular matrix36, as the cortex grows dramatically during embryonic life. The meninges contribute components to already established CBM37, 38, suggesting they participate in a dynamic turnover of CBM components, a key feature in the development of epithelial organs39. The meninges may produce enzymes implicated in BM remodeling, such as membrane-tethered matrix metalloproteinases40, but little is known about specific proteases expressed by embryonic meninges. The decrease in laminin α1 expression with meningeal loss is intriguing given deficits in BM stability seen with loss of other subunits in experimental models and human disease41, and may promote CBM instability independent of alterations in remodeling. Both disturbances in CBM remodeling and alteration of CBM composition might contribute to instability and breakdown of the CBM as the underlying brain rapidly expands.
In humans, FOXC1 mutations cause Axenfeld-Rieger syndrome (ARS), a disorder of neural crest migration with characteristic ocular and systemic manifestations42. Patients with ARS and the 6p25 deletion syndrome, which includes FOXC1 and other loci, can have psychiatric and neurologic symptoms43-47, although comprehensive brain imaging results are lacking. A recent study found human FOXC1 mutations cause abnormal cerebellar development and meningeal defects48. This study did not find cobblestone cortex on imaging, but small cortical dysplasia or abnormal organization of the neocortex cannot be excluded.
Most often cortical dysplasia is found sporadically, rather than as a part of a genetic syndrome. Since dysplasia can be acquired in association with prenatal insults, such as cytomegalovirus infection49, based on our study we can suggest that meningeal injury at a critical point in development might be responsible for some cases of sporadic, localized cortical dysplasia. Future pathologic studies on human brain samples, especially dysplasia resected during epilepsy surgery, will be needed to test this idea.
Supplementary Figure 1. Meningeal defects promote overmigration of immature neurons, and induce a general CBM breakdown affecting multiple proteins.
(A-F) Laminin (green) TuJ1 (red) double-immunostaining of the dorsomedial neocortex at E16.5 in control (A-C) and Foxc1lacZ/lacZ mutants (D-F). Arrow (E) indicates likely peripheral neuron. Arrow (C) indicates normal level of CBM. Arrows (F) indicate masses of TuJ1 positive immature neurons migrating through CBM breaks. In all panels dorsal is to the top.
(G-R) Collagen IV (G-L) and nidogen-1 (M-R) staining of the CBM at E12.5 (G-I, J-L) and E18.5 (J-L, P-R) in Foxc1 mutants. In all panels, dorsal is to the top and medial is to the left. Micrographs are of the CBM overlying the dorsomedial cortex where breaks are most prominent.
Supplementary Figure 2. Foxc1lacZ/lacZ mutants contain normally glycosylated α-dystroglycan.
E16.5 control (A) and Foxc1lacZ/lacZ (B) embryos stained with monoclonal antibody IIH6, which does not recognize hypoglycosylated α-dystroglycan. Note there is little difference in the staining of cell surfaces, although the cells appear disorganized in the mutants.
Supplementary Figure 3. Cajal-Retzius cells do not require the meninges for MZ positioning at E12.5.
Cajal-Retzius cells are visualized by immunofluorescence with an anti-GFP antibody in control (A) and Foxc1lacZ/lacZ embryos (B) each heterozygous for the CXCR4-EGFP transgene. The micrographs are from locations in the dorsal telencephalon where meningeal fibroblasts are absent in the Foxc1lacZ/lacZ mutant. The location of the MZ is indicated on the left. In all panels, dorsal is on the top.
This work was supported by the NIH (K12 NS01692, J.H.H., R01 DA017627, S.J.P.), Autism Speaks (J.A.S.) and the AHA/AAN (J.A.S.).