The ciliary kinesin KIF7 controls the development of the cerebral cortex by acting differentially on SHH signaling in dorsal and ventral forebrain

Abstract

Mutations of KIF7, a key ciliary component of Sonic Hedgehog (SHH) pathway, are associated in humans with cerebral cortex malformations and clinical features suggestive of cortical dysfunction. KIF7 regulates the processing of GLI-A and GLI3-R transcription factors in a SHH-dependent manner both in humans and in mice. Here, we examine the embryonic cortex development of a mouse model that lacks the expression of KIF7 (Kif7^-/-). The cortex is composed of principal neurons generated locally in the dorsal telencephalon where SHH expression is low and inhibitory interneurons (cIN) generated in the ventral telencephalon where SHH expression is high. We observe a strong impact of Kif7 deletion on the dorsal cortex development whose abnormalities resemble those of GLI3-R mutants: subplate cells are absent, the intermediate progenitor layer and cortical plate do not segregate properly, and corticofugal axons do not develop timely, leading to a delayed colonization of the telencephalon by thalamocortical axons. These structural defects alter the cortical distribution of cIN, which moreover exhibit intrinsic migration defects and cortical trajectories resembling those of cyclopamine-treated cIN. Our results show that Kif7 deletion impairs the cortex development in multiple ways, exhibiting opposite effects on SHH pathway activity in the developing principal neurons and inhibitory interneurons.

Keywords: Sonic Hedgehog pathway; cortical projections; embryonic development; interneuron migration; kinesin; mouse; neuroscience.

Figure 1.. Kif7 deletion alters cortical development on latero-medial and rostro-caudal axes.

(A) Kif7^-/- embryos are microphtalmic (black arrow) and exhibit skin laxity (white arrow). (B) External examination of the brain reveals the lack of olfactory bulbs (white arrows, left panel) and the thinning of the dorsal telencephalon (black arrow, right panel). (C) DAPI staining of rostro-caudal series of coronal sections at embryonic stage 12.5 (E12.5) (C1), E14.5, (C2), and E16.5 (C3) illustrates the anatomical defects of Kif7^-/- embryonic brains quantified at E14.5 in C4. The ventricles of Kif7^-/- embryos are strongly enlarged (upper left graph; WT, n=4–6, Kif7^-/-, n=5–6 depending on the rostro-caudal level), their cortical thickness strongly decreased (upper right graph; WT, n=3; Kif7^-/-, n=4), resulting in minimal brain width (lower left graph; WT, n=5–6; n=5–6 for Kif7^-/- depending on the rostro-caudal level) and height (lower right graph; WT, n=5–6; Kif7^-/-, n=4–6 depending on the rostro-caudal level) changes. Statistical significance was tested by two-way ANOVA or mixed model (GraphPad 8.1.0). For ventricle surface and cortex thickness, the mixed model reveals a genotype effect (p<0.0001). For brain width and height, no genotype effect was observed, but a significant effect of the rostro-caudal level on brain width (p=0.0441) and height (p=0.0092). (D) The pallium-subpallium boundary identified by the limit of expression of ventral (GSH2, D1, D2) and dorsal (TBR2, D3, D4) telencephalic markers is less precisely defined and slightly shifted to the ventricular angle in E13-E14 Kif7^-/- embryos compared to wild-type embryos (epifluorescent low magnification, D1, D3; confocal high magnification images, D2, D4). Graphs in C4 represent the means and SEM. Cx, cortex; Hip, hippocampus; LGE, lateral ganglionic eminence; MGE, median ganglionic eminence; CGE, caudal ganglionic eminence; Th, thalamus. Scale bars, 2 mm (A), 1 mm (B), 500 µm (C, D1, D3), 150 µm (D2, D4).

Figure 2.. Western blot analysis on the cortex and medial ganglionic eminence (MGE) of wild-type (WT) and Kif7^-/- embryos at embryonic stage 14.5 (E14.5).

(A) The cleaved form of GLI3 (GLI3-R at 83 KDa) and the full-length GLI3 (GLI3-FL at 190 KDa) are more abundant in the cortex compared to the MGE (see actin band intensity for protein loading). (B) The ratio Gli3-R/Gli3-FL is lower in the MGE than in the cortex of WT animals and is significantly decreased only in the cortex of Kif7^-/- brains compared to control. Two-way ANOVA reveals significant interaction between brain structure and genotype (WT, n=4; Kif7^-/-, n=4; p=0.002), and multiple comparisons show a statistical difference between genotype only in the cortex (****, p<0.0001). Graph represents the means and SEM.

Figure 3.. Histological alterations in the developing cortex of Kif7^-/- embryos.

(A–C) Immunostaining of cortical post-mitotic layers in embryonic stage 14.5 (E14.5) (A), E16.5 (B), and E18.5 (C) in median coronal sections representative of five wild-type (WT) and Kif7^-/- embryos imaged on epifluorescence (A1, B, C) or confocal (A2) microscopes. At E14.5 (A1, A2), the TBR1(+) staining (red) of the cortical plate is more clustered in Kif7^-/- than in WT embryos, and the MAP2(+) staining (green) of the subplate is absent in the dorsal cortex of E14.5 Kif7^-/- embryos (white arrow, right column). The post-mitotic layers remain thinner in Kif7^-/- embryos at later embryonic stages as illustrated by TBR1 staining at E16.5 (B) and CTIP1 staining at E18.5 (C) that specifically labels the deeper cortical layers (V–VI). Moreover, the hippocampus is underdeveloped in the mutant (white arrow) (C). (D–F) Immunostaining of TBR2(+) proliferative layer. At E14.5 (D1, D2), the TBR2(+) layer (green) of secondary progenitors appears disorganized in the lateral cortex of the Kif7^-/- embryos (white arrowhead in D1) and reaches the brain surface in the dorsal cortex of Kif7^-/- embryo where it intermingles with post-mitotic TBR1(+) cells (D2, red) (white arrows). In E12.5 Kif7^-/- embryo (E), the TBR2(+) layer is thin and clustered; however, its localization in the thickness of the cortex is normal. At the E16.5 stage (F), the TBR2(+) layer is normally positioned in the deeper layer of the cortex in both WT and Kif7^-/- embryos. Hip, hippocampus; V, ventricle; VZ, ventricular zone; SVZ, subventricular zone; IZ, intermediate zone; CP, cortical plate; MZ, marginal zone. Scale bars: 250 µm (A1, B, F), 100 µm (A2), 200 µm (D, E), 500 µm (C).

Figure 3—figure supplement 1.. Kif7 deletion is associated with cortical heterotopia at embryonic stage 14.5 (E14.5).

(A) On coronal sections of the telencephalon of wild-type (WT) animals, TBR2(+) cells form a well-defined layer restricted to the cerebral cortex. Most TBR2(+) cells are densely packed in the subventricular zone (SVZ), whereas some TBR2(+) cells are dispersed in the ventricular zone. (B) In about 20% of Kif7^-/- embryos, heterotopia identified by a disorganization of the TBR2(+) layer was observed in various regions of the cerebral cortex, either dorsal (B1) or lateral (B2, B3). Depending on heterotopia, TBR2(+) were either displaced toward the brain surface (B2) or toward the ventricle (B3). Scale bar: 200 µm.

Figure 4.. Kif7 deletion disrupts the connectivity between the cortex and the thalamus.

(A) Panels illustrate the corticofugal projections labeled by DiI crystal (red dots) positioned in the dorsal (A1) and lateral (A2, A3) cortex of embryonic stage 14.5 (E14.5) wild-type (WT) (left) and Kif7^-/- (right) embryos on vibratome sections performed 30 days after DiI placement and imaged on a macroscope. In Kif7^-/- embryos, fewer axons project from the dorsal (A1) and lateral (A2) cortex to the subpallium than in WT (compare enlarged views of the projections below A1 and A2). Cortical injections in Kif7^-/- embryos do not label thalamic axons (compare left and right panels in A3) but label a ventral projection (A3, white arrow). (B) Rostro-caudal series of coronal sections (B1–B5) immunostained with anti-Netrin G1a (NG1a) antibodies compare the trajectory of thalamocortical axons (TCA) in an E14.5 WT (left) and in a Kif7^-/- (right) embryo imaged on a macroscope. TCA reach the pallium-subpallium boundary of the WT embryo, whereas they are lost in the ventral forebrain of the mutant (B4, white arrow). (C) Representative three-dimensional reconstructions (C1, lateral; C2, horizontal; C3, coronal views) of WT and Kif7^-/- E14.5 brains immunostained as a whole with NG1a (red) and TBR1 (green) antibodies that label respectively the TCA and cortical plate cells before transparization and imaging with a light sheet microscope (see Figure 4—videos 1 and 2). While all labeled TCA extend in the internal capsule (IC) and a significant proportion of them enter the cerebral cortex in the WT brain, TCA in the Kif7^-/- brain split in two bundles in the basal telencephalon. A bundle stops shortly after entering the IC (white arrowheads), whereas the second bundle extends ventrally (white arrows). (D) At E16.5, TCA are immunostained with NG1a in WT (left) and Kif7^-/- (right) embryos in whole brain with TBR1 antibodies before transparization and imaging (D1) and on coronal section with TBR2 (D2). The TCA extend in the cortex in Kif7^-/- brain; however, fiber density is reduced in the median and caudal brain compared to WT (D1, white arrowhead). TCA invade the post-mitotic layers in the cortex (above TBR2 layer) to a lower extent in Kif7^-/- brain (D2). In the Kif7^-/- brain, thick bundles of NG1a(+) fibers project ventrally from the caudal telencephalon, a projection never observed in control brains (D1, white arrow). (E) At E18.5, rostro-caudal series of coronal sections (E1–E4) immunostained with NG1a antibodies compare the trajectory of TCA in WT (left) and Kif7^-/- (right) embryos imaged on a macroscope. TCA reach the dorsal cortex in both WT and Kif7^-/- embryos. However, the TCA projection in Kif7^-/- (arrowheads) becomes thinner in median sections (E2) and almost disappear in caudal sections (E3, E4). Fiber trajectories in the lateral striatum are abnormal (E2, E3, arrows). (F) Panels illustrate the cortical projections immunolabeled by Smi32 antibodies in E18.5 WT and Kif7^-/- brain. In the Kif7^-/- brain, the number of fibers connecting the cortex with other brain structures is strongly reduced (arrow). Cx, cortex; dTh, dorsal thalamus; Hip, hippocampus; Hy, hypothalamus; St, striatum; IC, internal capsule; LGE, lateral ganglionic eminence; MGE, median ganglionic eminence; PSB, pallium-subpallium boundary. Scale bars: 250 µm (A–D), 500 µm (E, F).

Figure 4—figure supplement 1.. Alterations of the thalamocortical projection at embryonic stage 14.5 (E14.5) in Kif7^-/- brains.

Immunostaining of coronal brain sections at a caudal level with NG1a (red) and PAX6 (green) antibodies shows that thalamocortical axons (TCA) mistargeting in the ventral telencephalon is not related to abnormal PAX6 expression in the zona incerta between the dorsal and ventral thalamus. Scale bar: 200 µm.

Figure 5.. Abnormal cortical distribution of cIN and Cxcl12 transcript expression in embryonic stage 14.5 (E14.5) in Kif7^-/- brains.

(A) The cortical distribution of cIN is visualized in wild-type (WT) and Kif7^-/- mouse embryos crossed with the Nkx2.1-Cre/R26R-tdTomato strain in which medial ganglionic eminence (MGE)-derived cIN express the fluorescent marker tdTomato. Panels in A1 compare the distribution of tdTomato (+) cIN in WT and Kif7^-/- cortical sections prepared at the same rostro-caudal level and in which SVZ is immunostained with TBR2 antibodies (green). Pictures show that the deep migratory stream of cIN terminates in Kif7^-/- brains in the cortical region where the TBR2(+) layer reaches the cortical surface. Quantitative analysis of the mean length of the deep and superficial migratory streams measured from the entry in the pallium to the last detected cIN in the cortex is illustrated in graph A2. Statistical significance is assessed using two-way ANOVA (layers [WT, n=4; Kif7^-/-, n=3; ***, p=0.001] and genotype [*, p=0.0157]). (B) Representative pictures (B1) of the deep and superficial tangential migratory streams of cIN in the lateral cortex of WT and Kif7^-/- embryos. Pictures illustrate the decreased thickness of the superficial stream and the reduced distance between the deep-superficial streams in Kif7^-/- embryos. Graph in B2 (WT, n=4; Kif7^-/-, n=4) compares the distribution of the fluorescence intensity along a ventricle/MZ axis (see gray rectangles in B1) using the plot profile function of Fiji. Curves show no change in the distance between the ventricular wall and the deep cIN, but a significant reduction of the distance between the two streams in the Kif7^-/- cortical sections as quantified on the graph B3 (WT, n=4; Kif7^-/-, n=4; two-way ANOVA reveals a significant interaction between genotype and layer [p=0.0233] and multiple comparisons, a statistical difference between genotype only for the distance between the cIN streams [**, p=0.0051]). (C) Panels compare the distribution of Cxcl12 mRNA in WT (left panel) and Kif7^-/- (right panel) forebrain coronal sections at E14.5. The WT section shows Cxcl12 transcript enrichment in a deep cortical layer already identified as the SVZ. In the Kif7^-/- cortical section, the expression of Cxcl12 transcripts is reduced to the lateral part of the SVZ. (D) Z-projections of 30 frames acquired during 12 hr in the dorsal cortex of living organoptypic forebrain slices representative of E14.5 WT and Kif7^-/- embryos with tdTomato expressing cIN. Kif7^-/- cIN were able to migrate dorsally but followed preferentially radially oriented trajectories. Scale bars: 200 µm.

Figure 5—figure supplement 1.. cIN migration up to birth.

(A) At embryonic stage 14.5 (E16.5), the delay observed in the ability of cIN to migrate toward the dorsal cortex observed at E14.5 persisted; however, some cIN were able to migrate forward but no longer following the wild-type (WT) pattern of tangential migration. (B, C) At birth (P0), cIN had colonized to dorsal cortex in Kif7^-/- embryos (B) but are abnormally distributed in the Kif7^-/- cortex: their density is strongly increased in the supragranular cortical layers that do not express CTIP(+) principal neurons (green) (C). Scale bar: 500 µm (A, B), 100 µm (C).

Figure 6.. Dynamic behavior of migrating cIN in co-culture and organotypic cortical slices.

(A) To prepare co-cultures, medial ganglionic eminence (MGE) explants were dissected out of telencephalic vesicles at embryonic stage 14.5 (E14.5) from wild-type (WT) or Kif7^-/- Nkx2.1-Cre;Rosa26-tdTomato embryos and placed on a substrate of WT E14.5 dissociated cortical cells (A1). After 24 hr in culture, MGE cells migrated centrifugally away from MGE explants on the substrate of cortical cells and were recorded for 7 hr. WT (n=160) and Kif7^-/- (n=126) MGE cells were tracked manually using the MTrackJ plugin allowing to analyze migratory parameters (A2). Box and whisker plots indicate the speed (A3), the time without nuclear movement (A4), the frequency (A5), and speed of nuclear movements (A6). Statistical significance is assessed by Mann-Whitney or unpaired tests. ***, p<0.001 (A4); *, p=0.0308 (A3), p=0.0123 (A5). Directionality persistence was calculated as the ratio of the cell displacement (distance between the first and last positions of the cell) on the cell trajectory (A7, scheme on the right). The graph represented the mean directionality ratios at each time point over the first 5 hr of recording for the recorded cells (A7). Statistical significance is assessed using two-way ANOVA and reveals a significant effect of time and genotype (****, p<0.0001). Scale bar: 50 µm (A2). (B) tdTomato(+) cIN migrating in living cortical slices from Nkx2.1-Cre;Rosa26-tdTomato embryos were tracked manually using the MTrackJ plugin and their trajectories color-coded as shown in legend (B2) to characterize their preferred direction (tangential, oblique, radial, immobile) and cortical layer localization (VZ-SVZ, IZ, CP). The picture in B1 illustrates the z-projection of trajectories reconstructed in a control slice, superimposed to the last picture of the video. Cortical interneurons migrating tangentially in the superficial migratory stream (MZ) could not be tracked because of high density. Graphs in B2 compare the percentage of each kind of trajectory recorded in the lateral cortex of control slices (control, see also Figure 6—video 1), Kif7^-/- slices (Kif7^-/-, see also Figure 6—video 2), control slices treated acutely with either murine SHH (SHH, see also Figure 6—video 3) or cyclopamine (Cyclo, see also Figure 6—video 4). Significance of the differences between the four distributions was assessed by a Chi-square test, χ² (24, n=1224), p=0.0004998, ***. All experimental conditions differed from the control (Fisher’s test, p=0.0004998, ***). Slices from three WT animals in control condition or treated with drugs and from three Kif7^-/- animals were analyzed; the number of analyzed cells is indicated above bars. Schemes in B3 summarize the main results observed in each experimental condition. Trajectories are represented with the same color code as in B1, and line thickness is proportional to the percentage of cells exhibiting each type of trajectory. Immobile cells are figured by a coil. Box and whisker plots indicate the mean speed (B4), the frequency of period without nuclear movement (number/hr) (B5) and the mean duration of period without nuclear movement (hour) (B6) for cIN migrating tangentially in the deep stream (red box and whisker plots, left) or to the cortical plate (green box and whisker plots, right). Statistical significance assessed by Kruskal-Wallis tests in each cluster. ****, p<0.0001; ***, B5 left p=0.0005; **, B4 right p=0.0088, B5 right, p=0.0037, B6 left p=0.0034; *, B4 left p=0.0233, B5 right p=0.0470, B6 left p=0.0134, B6 right p=0.0394. Number of analyzed cells is indicated on plots. VZ, ventricular zone; SVZ, subventricular zone; IZ, intermediate zone; CP, cortical plate; MZ, marginal zone. Scale bar: 300 µm.

Figure 7.. Expression of Shh transcripts.

(A–C) and distribution of Sonic Hedgehog (SHH) protein (D) in the embryonic stage 13.5 (E13.5)-E14.5 forebrain. (A, B) Distribution in median (A) and caudal (B) coronal sections of Shh mRNA detected by in situ hybridization with an antisense Shh probe at E13.5 (A1, B1) and by RNAscope at E14.5 (A2-3, B2). Shh transcripts are strongly expressed in the medial ventral forebrain (SVZ and mantle zone of the MGE and septum, A1, A2), in the mantle zone of the CGE (B1, B2), in the zona limitans intrathalamica (ZLI in B1) and in the ventral midline of the third ventricle (V3 in B1). RNAscope further confirmed the strong expression of Shh mRNA (A3, green) in MGE and septum regions that strongly express Lhx-6 mRNA (A2, red). Confocal observations in the SVZ and mantle zone of the MGE showed that Shh mRNA (green) is co-expressed with the Lhx6 mRNA (red) in a significant number of cIN (yellow cells in A4). (C) Confocal analyses at higher magnification of the double detection by RNAscope of Shh and Lhx-6 mRNA. Cells were identified on stacked images (△z=1 µm) using Nomarski optic. In the lateral cortex close to the PSB (C1, z-projection of 10 confocal planes) and in the dorsal cortex (C2, z-projection of 10 confocal planes), a very small proportion of cells expressing Lhx-6 mRNA also express Shh mRNA (white arrows in C1,C2). Counting in the deep stream (SVZ-IZ) and in the MZ is shown in graph C3 (9–17 fields in three sections). A few progenitors in the cortical VZ express Shh mRNA at very low levels (arrowhead, C1). (D) SHH-Nter and TBR2 co-immunostaining of Nkx2.1-Cre/R26R-tdTomato brain coronal sections at E14.5. Representative confocal merged stacked images (△z=0.2 µm; 48 images) in the pallium-subpallium boundary (PSB, D1), lateral cortex (D2) and dorsal (D3) cortex revealing SHH-Nter immunostaining in blood vessels and the presence of numerous bright dots all over the cortical neuropile. On the ventricular side of the PSB and of the lateral-most part of the LGE, SHH-Nter(+) bright elements are aligned radially. In the lateral cortical neuropile, smaller bright dots align radially in the VZ, tangentially in the SVZ-IZ, and radially in the CP (see higher magnification on the right panel in which SHH-Nter immunostaining is shown in white and the contrast is increased). Cx, cortex; LGE, MGE, and CGE, lateral, medial, and caudal ganglionic eminence; CP, cortical plate; Hyp, hypothalamus; PSB, pallium-subpallium boundary; Sp, septum; Th, thalamus; V3, third ventricle; VZ, ventricular zone; SVZ, subventricular zone; IZ, intermediate zone; MZ, marginal zone. Scale bars: 500 µm (A, B), 20 µm (C), 100 µm (D).

Figure 7—figure supplement 1.. SHH-Nter immunostaining specificity.

(A, B) Representative pictures of SHH-Nter immunostaining of C57Bl/6 mice brain coronal sections at embryonic stage 12.5 (E12.5) (A) and E14.5 (B) imaged with a macroscope. High signal is observed along the third ventricle (A1, arrow) and in the ZLI (A2, arrow) at E12.5 and in the choroid plexus and the septum at E14.5 (B), whereas a faint signal is observed in the cortex. (C) Representative picture of SHH-Nter immunostaining in the cortex at E12.5. Z stacks of 48 images confocal images (△z=0.2 µm) reveal the presence of numerous bright dots in the cortical neuropile with a gradient of density from the ventricular surface to the surface of the brain. (D) Coronal sections of E14.5 brain are labeled with SHH-Nter antibodies (D1) or only with the secondary antibodies (D2). Confocal Z stacks of 48 images (△z=0.2 µm) reveal a punctiform signal all over the cortical neuropile and in some cells in the cortical plate (D1), whereas no signal is observed in sections immunolabeled only with the secondary fluorescent antibody (D2). CP, choroid plexus; Cx, cortex; Sp, septum; V3, third ventricle; ZLI, zona intra-thalamica. Scale bar: 250 µm (A, B), 100 µm (C, D).