Pax6 regulates neuronal migration and cell proliferation via interacting with Wnt3a during cortical development

Abstract

The paired box 6 (Pax6) gene encodes a highly conserved transcription factor, involved in the development of eyes, brain, and endocrine glands. Homozygous loss of Pax6 resulted in neonatal death in mice, plus loss of eyes and malformation of cerebral cortex. In patients with heterozygous Pax6 mutations, a reduction in thickness of the frontoparietal cortex was detected, which was also observed in small eye mice. In this study, we found that Pax6 overexpression increased the cortical thickness, especially in the intermediate zone of the cortex, which conflicts with the report of Manuel et al. Pax6 overexpression appears to detain neurons in the intermediate zone while promoting cell proliferation. It is worth noting that the impact of Pax6 overexpression on cortical thickness and neuronal migration was temporal, explaining the differences with other reports. We postulated that the alteration of Pax6 isoform ratio by autoregulation might be responsible for this. JASPAR analysis together with the results of qPCR, Western blot, CUT&Tag, and rescue experiments revealed that Pax6 regulates neuronal migration and cell proliferation by indirectly mediating Wnt3a expression. Therefore, we propose that Pax6 participates in corticogenesis via interaction with Wnt3a in regulating neuronal migration and cell proliferation.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-88662-5.

Introduction

Pax6, a highly conserved transcription factor with two DNA-binding motifs (the paired domain and homeodomain), plays pivotal roles in various processes of brain development, including brain patterning, neuronal specification, neuronal migration, axonal guidance, and neuronal circuit formation^1,2. In mammalian brain, Pax6 is expressed throughout the lifespan in a region-specific manner which contributes to the formation of anterior-posterior and dorsal-ventral brain axes^3–5. Heterozygous loss-of-function of Pax6 (haploinsufficiency) in humans causes a variety of structural brain abnormalities, for instance, absent/abnormal pineal gland, absent/hypoplastic anterior commissure, reduced corpus callosum, and deficient auditory interhemispheric transfer^6–9, which are also mirrored in small eye (Sey) mice carrying heterozygous Pax6 mutations⁴. In rodents, homozygous Pax6 mutations (Pax6^Sey/Sey) lead to a severe malformation of cerebral cortex with other defects, and ultimately neonatal lethality^10,11. This study aims to investigate the functional role of Pax6 in cortical development.

During cortical development, neurons and glia are generated in the ventricular zone (VZ), a layer of multipotent progenitors at the inner surface of the dorsal telencephalon¹². Multipotent progenitors are thought to undergo two distinct types of divisions, symmetrical divisions and asymmetrical divisions¹³. Symmetrical divisions produce two identical daughter cells which remain within the VZ and expand the progenitor pool, whereas asymmetrical divisions generate two different daughter cells, one that remains in the VZ as a progenitor, the other that migrates away from VZ to differentiate into postmitotic cells^14–16. In mouse cortex, the majority of divisions from embryonic day (E) 10.5 to E15.5 are symmetrical; however, proportions of asymmetrically dividing progenitors increase more in Pax6^Sey/Sey mouse cortex than in wild type, with a shorter cell cycle, consequently generating more postmitotic neurons¹⁷. Even so, cortical cell number is reduced in Pax6^Sey/Sey cortex, which then results in a reduction of cortical thickness^1,18,19. Quinn et al. proposed that the reduction of neurons in Pax6^Sey/Sey cortex was due to abnormal depletion of the progenitor pool caused by high proportions of newly divided cells exiting from the cell cycle¹⁸, suggesting that Pax6 is crucial for maintaining progenitor cells. Interestingly, Gotz et al.²⁰ and Heins et al.²¹ demonstrated that Pax6 can also promote cell differentiation by analyzing the effects of Pax6 in glia cells isolated from Pax6-transfected cortex. The isoforms of Pax6 and their expression levels are thought to contribute to this dual role^1,22,23.

Neuronal migration is critical for the establishment of neural networks since the birthplace of most neurons is different from their functioning region²⁴. Disturbances of neuronal migration are implicated in cortical malformations²⁵. During corticogenesis, some neurons originate from the VZ, and migrate to the cortical plate where they form the six-layer structure in an inside-out pattern²⁶. In developing Pax6^Sey/Sey cortex (E10.5 to E18.5), cells null for pax6 accumulate at the boundary of the subventricular zone (SVZ) and intermediate zone (IZ), suggesting that Pax6 regulates neuronal migration²⁷. Recently, extracellular Pax6 was also reported to regulate neuronal migration in the developing neocortex; blocking extracellular Pax6 disrupts tangential migration of Cajal-Retzius neurons²⁸. These findings might explain why axonal connections are disrupted in Pax6^Sey/Sey cortex—neurons fail to migrate to their appropriate locations in the absence of Pax6.

However, the specific molecular mechanism of how Pax6 regulates cell proliferation/differentiation and neuronal migration during corticogenesis remains to be investigated. In this study, we found that Pax6 overexpression resulted in an increase of cortical thickness at E18.5, especially in the IZ, inconsistent with Manuel et al.’s finding in the PAX77 mice (Pax6-overexpression mice)²⁹. Pax6-overexpressing cells were detained in the IZ, while immunostaining results supporting that Pax6 overexpression promoted the proliferation of neural stem cells. Notably, we first report that Pax6 overexpression temporally influences cortical thickness and neuronal migration, explaining no significant change observed in cortical thickness of P7 PAX77 mouse cortex. We hypothesize that the alteration of Pax6 isoform ratio by autoregulation might contribute to the temporal effect of Pax6 overexpression. In addition, our data reveal that Pax6 overexpression might induce the expression of Wnt3a to inhibit neuronal migration and promote cell proliferation. Hence, we propose that Pax6 participates in corticogenesis by interacting with Wnt3a to regulate neuronal migration and cell proliferation.

Materials and methods

Animals

C57BL/6J mice aged 8 weeks were purchased from Beijing Victoria Laboratory Animal Technology Co., Ltd, and housed with a 12-h light/dark cycle (lights on from 8:00 to 20:00) at a constant temperature of 25 °C. All experimental procedures involving the animals were performed in strict accordance with the guidelines of The Ministry of Science and Technology of the People’s Republic of China [(2006)398]. Additionally, the protocols were reviewed and approved by the Animal Care Committee of Xinxiang Medical University (No. 030032).

Method of anesthesia

A multi-function animal anesthesia system (RWD Life Science Co., Ltd., China) was utilized for the anesthesia of pregnant mice. The procedure was as follows: The anesthesia air pump was activated to deliver isoflurane (supplied by RWD Life Science Co., Ltd.) into the induction chamber. Pregnant mice were then introduced into the chamber and allowed to inhale the anesthetic for several minutes until they reached a state of surgical anesthesia. Once anesthetized, the mice were carefully transferred to an anesthesia operation platform, where in utero electroporation was performed.

In utero electroporation

Pregnant female mice at embryonic day 13.5 (E13.5) or 15.5 (E15.5) were anesthetized using a multi-function animal anesthesia system. Their uterine horns were gently pulled out of the abdominal cavity to expose the embryos. A microneedle was used to inject 2 µg/µl of the following plasmids into the lateral ventricle of the embryos: pCAGGS-EGFP (Control), pCAGGS-EGFP + pCAGGS-Pax6-EGFP (1:4, Pax6 overexpression), pCAGGS-EGFP + pECMV-Wnt3a-Flag (1:4, Wnt3a overexpression), or pCAGGS-EGFP + pCAGGS-Pax6-EGFP + pTRIPZ-Wnt3a shRNA (1:4:4, Pax6 overexpression + Wnt3a knockdown). Pax6 (NM_001244200.2) and Wnt3a (NM_009522.3) gene sequences in these constructs are derived from Mus musculus. Electroporation was executed using CUY-21 EDIT electroporator (Nepa Gene, Japan) with settings of 35 V for 60 ms, delivered as 5 pulses with 100-ms intervals between each pulse. After electroporation, pregnant mice were continuously maintained 1 day (1d), 3 days (3d), 5 days (5d), or 7 days (7d) for sample collection, depending on the experimental objectives. Brain samples were fixed in 4% paraformaldehyde (PFA, G1101, Servicebio, China) at 4 °C, then dehydrated in 18% sucrose solution. The dehydrated brains were embedded in Tissue-Tec O.C.T compound (4583, Sakura Finetek, USA), solidified in liquid nitrogen, and horizontally or coronally sectioned at 20-µm thickness using a Leica 1850 cryotome (Germany) for subsequent thionine or immunofluorescent staining. n = 3 for each group.

Thionine staining

Brain sections were dried at 37 °C for 30 min, and then fixed with 4% PFA for 15 min. After being washed with 1 × TBS and distilled water, 5 min and 1 min, respectively, the brain sections were incubated with 1% thionine blue stain solution (G1901, Solarbio, China) at room temperature until Nissl’s Body was visibly stained. The sections were thoroughly rinsed in distilled water to terminate staining. Brain sections were then dehydrated in a graded series of ethanol solutions: 70% for 5 min, 85% for 5 min, 95% for 3 min, and finally 100% for 1 min, followed by xylene for 3 min. Finally, neutral balsam (G8590, Solarbio) was applied to mount the brain sections for microscopy. n = 3 for each group.

Cell transfection

N2a, SH-SY5Y, or HEK-293T cells were plated in 6- or 24-well plates with or without coverslips, and cultured in Minimum Essential Medium Eagle (MEME, M2279, Sigma-Aldrich, USA) supplemented with 10% fetal bovine serum (FBS, Hyclone, USA) and 2% L-glutamine (G7513, Sigma-Aldrich), SH-SY5Y complete medium (iCell-001b, iCell, China), or Dulbecco’s modified eagle medium (DMEM, ZQ-101, Zhongqiao Xinzhou Biotech, China). When the cells reached approximately 80% confluency, transfection was carried out using Lipofectamine™ 3000 transfection reagent (L3000015, Invitrogen, USA) according to the manufacturer’s protocol. The plasmids used for cell transfection included pCAGGS-EGFP (Control), pCAGGS-Pax6-EGFP (Pax6 overexpression), pECMV-Wnt3a-Flag (Wnt3a overexpression), and pTRIPZ-Wnt3a shRNA (Wnt3a knockdown). Seventy-two hours post-transfection, the cell medium was harvested and transferred to a new tube for medium western blot analysis. The cells were then washed three times with cold 1 × PBS. Cells were either collected in RIPA lysis buffer (P0013B, Beyotime, China) for cell western blot or in TRNzol universal reagent (DP424, Tiangen, China) for quantitative PCR (qPCR), or fixed with 4% PFA for immunostaining. n = 3 for each group.

Stable expression cell line

pCDH-CMV-MCS-EF1-GFP-Puro (Control) or pCDH-CMV-MCS-mPax6-EF1-GFP-Puro (Pax6 overexpression) lentiviral vectors were co-transfected with packaging plasmid (psPAX2) and envelope plasmids (pMD2.G) into HEK-293T cells using EZ Trans (AC04L092, Life-iLab, China). After transfection, the lentiviral particles were harvested from the culture medium of the packaging cells. SH-SY5Y cells were plated in 6-well plates, 1.5 × 10⁵ cells per well. When the cells reached 60% confluency, lentiviral particles were added to 6-well plates for lentiviral infection. After infection, puromycin (ST551-10, Beyotime) selection was conducted to obtain SH-SY5Y cells stably expressing GFP (GFP-SH-SY5Y) or Pax6 (Pax6 OE-SH-SY5Y) for immunostaining or Western blot.

Cell cycle analysis

GFP-SH-SY5Y or Pax6 OE-SH-SY5Y cells were initially fixed in cold 70% ethanol at 4°C for 30 min, followed by two-time wash of 1 × PBS. After 850 g centrifugation, supernatant was removed, and 50 µl RNase solution (100 µg/ml, ST578, Beyotime) was added. Finally, cells were incubated with 200 µl Propidium Iodide (PI) (50 µg/ml, ST1569-10 mg, Beyotime) for flow cytometry to analyze cell cycle.

Quantitative PCR

Total RNA was extracted from transfected N2a cells using the TRNzol universal reagent (DP424, TIANGEN, China). cDNA was transcribed from 2 µg of total RNA in a 20 µl reaction using 5 × All-In-One RT Mastermix (G486, abm, Canada). Primers for real-time PCR were designed using Primer3 software; primer sequences are listed below. The CFX Connect Real-Time PCR System (Bio-Rad, USA) was applied to perform qPCR. cDNA was amplified in a 10 µl reaction containing 0.7 µM of each primer and 5 µl of 2 × QuantiNovo SYBR PCR Master Mix (169024665, QIAGEN, USA). The amount of gene product (Pax6 and Wnt3a) in each sample was determined using the 2^−ΔΔCt method. β-actin was used as a housekeeping gene. The amount of gene product for each gene of interest was expressed relative to that of β-actin to normalize for differences in total cDNA between samples. n = 4 for each group.

Gene name	Forward primer	Reverse primer
β-actin	CATTGCTGACAGGATGCAGAAGG	TGCTGGAAGGTGGACAGTGAGG
Pax6	CTGAGGAACCAGAGAAGACAGG	CATGGAACCTGATGTGAAGGAGG
Wnt3a	AACTGCACCACCGTCAGCAACA	AGCGTGTCACTGCGAAAGCTAC

Western blot

Cell western blot

Transfected N2a or HEK-293T cells were lysed in 100 µl RIPA lysis buffer (P0013B, Beyotime) on ice for 30 min (vortex every 5 min), and then centrifuged at 12,000 rpm, 4 °C. The supernatant (protein sample) was collected in a clean tube and its protein concentration was detected by using BCA protein assay kit (P0012, Beyotime). Protein samples were separated by SDS-PAGE (CFAS any KD PAGE, PE008-2, ZHHC, China). Pax6 or Wnt3a proteins were detected using Mouse anti-Pax6 primary antibody (1:5000, ab197768, Abcam, UK) or Rabbit anti-Wnt3a primary antibody (1:5000, ab28472, Abcam) and HRP-conjugated anti-Mouse secondary antibody (1:5000, ZDR-5307, ZSGB-BIO, China) or HRP-conjugated anti-Rabbit secondary antibody (1:5000, ZDR-5306, ZSGB). Immunoblots were developed using Immobilon Western Chemiluminescent HRP substrate (WBKLS0500, Millipore, USA) and imaged using an Amersham Imager 600 (GE Health care, USA). ImageJ 1.53t was applied to quantify the gray values of each blot. The relative gray values for Pax6 and Wnt3a were calculated by normalizing the gray value of each protein’s band to that of its corresponding loading control (GAPDH, 1:5000, 60004-1-lg, proteintech, China). n = 3 for each group.

Cell western blot was conducted on GFP/Pax6 OE-SH-SY5Y cells to investigate the impact of Pax6 overexpression on apoptosis. The primary antibodies applied included Rabbit anti-BAX (1:1000, ab32503, Abcam), Rabbit anti-BCL-2 (1:1000, ab182858, Abcam), and Rabbit anti-Cleaved-caspase3 (1:1000, 9662, Cell signaling technology, USA), which were detected by HRP-conjugated anti-Rabbit secondary antibody (1:5000, ZDR-5306, ZSGB). n = 3 for each group.

Medium western blot

Medium collected from transfected N2a cell culture was centrifuged at 12,000 rpm, 4 °C for 20 min, and then the supernatant (protein sample) transferred to a clean tube. Equal volume of protein samples was applied to detect Wnt3a levels (specific procedures same with above). n = 3 for each group.

CUT and tag and data analysis

CUT&Tag-IT Assay Kit (53160, Active Motif, UAS) was applied on Pax6 OE-SH-SY5Y cells to generate a DNA library which reveals DNA binding sites of Pax6. The DNA library was sent to Biopharmaceutical Public Service Platform (Nanjing, China) for high-throughput sequencing (NCBI Accession: PRJNA1187277, ID: 1187277, https://www.ncbi.nlm.nih.gov/sra/PRJNA1187277). Pax6 binding motif analysis, Gene Ontology (GO) enrichment analysis, and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis were performed by Biopharmaceutical Public Service Platform. P value < 0.05 was considered as significantly binding motif. n = 1.

Immunofluorescent staining

For labelling the cortical layer, coronal brain sections were dried at 37 °C for 30 min, and then fixed with 4% PFA at 4 °C for 15 min. After three-time wash of 1 × TBS (5 min each), brain sections were rinsed with 1% Triton X-100 containing 1 × TBS for 5 min, followed by blocking with QuickBlock™ Blocking Buffer for Immunol Staining (P0260, Beyotime) for 1 h. Subsequently, they were incubated with Rabbit anti-Tbr1 (1:500, ab183032, Abcam) overnight at 4 °C. Cy3-conjugated anti-Rabbit secondary antibody (1:500, 135294, Jackson ImmunoResearch Inc., USA) was applied to detect the primary antibody. After DAPI staining, brain sections were mounted for microscopy using a Nikon C2 + confocal microscope (Nikon, Japan). n = 3 for each group.

For detecting Pax6-positive cells, neural stem cells, intermediate progenitor cells, neurons, or proliferating cells, the specific procedures were similar to labelling cortical layer. The primary antibodies applied were Mouse anti-Pax6 (1:500, ab197768, Abcam), Rabbit anti-SOX2 (1:500, ab929494, Abcam), Rabbit anti-Tbr2 (1:500, ab216870, Abcam), Rabbit anti-NeuN (1:500, ab177487, Abcam), and Rabbit anti-Ki67 (1:500, ab15580, Abcam), which were detected by Cy3-conjugated anti-Mouse secondary antibody (1:500, 134635, Jackson ImmunoResearch Inc.) or Cy3-conjugated anti-Rabbit secondary antibody (1:500, 135294, Jackson ImmunoResearch Inc.). Pax6, SOX2, Tbr2, NeuN, and Ki67 antibodies are used to label Pax6-positive cells, neural stem cells, intermediate progenitor cells, neurons, and proliferating cells, respectively.

To determine the distribution of Pax6-positive cells, we counted the number of these cells across various cortical layers and calculated their proportion as a percentage of the total cell count in each layer. The expression levels of SOX2 and Ki67 within the cortex were quantified using ImageJ software version 1.53t. Specifically, the percentage of the cortical area positive for SOX2 and Ki67 was determined by analyzing the staining intensity and distribution. Additionally, the total number of SOX2-positive cells was enumerated, and their representation as a percentage of the total cell count in each cortical layer was calculated. n = 3 for each group.

To analyze neuronal migratory trajectory, cells whose main processes are oriented toward the marginal zone indicate their vertical migration, therefore classified as exhibiting vertical migratory morphology; those with processes extending laterally are categorized as having lateral migratory morphology. The number of cells exhibiting these two morphologies in the intermediate zone was counted, and their proportions were calculated as percentages of the total cell count. n = 3 for each group.

For examining cell proliferation, SH-SY5Y cells transfected with pCAGGS-EGFP (Control), and pCAGGS-Pax6-EGFP (Pax6 overexpression) constructs were fixed with 4% PFA for 30 min at 4 °C, followed by three-time wash of 1 × TBS. 1% Triton X-100 containing 1 × TBS was applied to penetrate SH-SY5Y cells for 5 min. After blocking 1 h with goat serum solution (C0265, Beyotime) at room temperature, they were incubated with Rabbit anti-Ki67 (1:500, ab15580, Abcam) primary antibody overnight at 4 °C. On the second day, primary antibody was detected with Cy3-conjugated anti-Rabbit secondary antibody (1:500, 135294, Jackson ImmunoResearch Inc.). After DAPI staining, the SH-SY5Y-containing coverslip was mounted on the slide for microscopy by using a Nikon Eclipse 80i microscope equipped with a Leica DFC 425 C camera. The proportion of Ki67-positive SH-SY5Y cells was determined by calculating the percentage of these cells relative to the total number of SH-SY5Y cells in the sample. This quantification was performed to assess the rate of cell proliferation within the SH-SY5Y cell population. n = 3 for each group.

For analyzing cell morphology, immunostaining against Pax6 was performed on GFP-SH-SY5Y (Control) and Pax6 OE-SH-SY5Y cells (Pax6 overexpression). The specific procedures were similar to examining cell proliferation. The primary and secondary antibodies applied were Mouse anti-Pax6 (1:500, ab197768, Abcam) and Cy3-conjugated anti-Mouse (1:500, 134635, Jackson ImmunoResearch Inc.). Microscopic examinations were conducted using a Nikon C2 + confocal microscope. n = 3 for each group.

Statistical analysis

All data were presented as mean ± S.E.M. Assessment of statistical significance was performed using an unpaired Student’s t test and one-way ANOVA using GraphPad Prism 9.5.0 (GraphPad, USA). P < 0.05 was considered as a statistically significant difference.

Results

Overexpression of Pax6 increases the thickness of cortical layers

in utero electroporation was performed on the E15.5 mouse embryos to investigate the role of Pax6 during corticogenesis; the constructs, pCAGGS-EGFP (Control) and pCAGGS-EGFP + pCAGGS-Pax6-EGFP (1:4, Pax6 overexpression), were applied; brain samples were then collected three days after electroporation (E15.5 + 3d) (Fig. 1A). Green fluorescence marked the transfected side in the control (Fig. 1B) and Pax6-overexpressing brain (Fig. 1C). Interestingly, the width (w2) of cerebral hemispheres (Left hemisphere + Right hemisphere) in the Pax6 overexpression group showed a significant increase, compared with that of the control (w1) (Fig. 1D), suggesting that Pax6 overexpression enlarged brain size, inconsistent with previous reports²⁹. To further explore the specific alteration of cerebral hemispheres caused by Pax6 overexpression, Pax6-overexpressing mouse brains were sectioned horizontally or coronally. Figure 1E showed the horizontal section of cerebral hemispheres of Pax6-overexpressing brain; the cortical layer (H2) of transfected hemisphere (marked by white rectangle) was much thicker than that (H1) of the corresponding area of the non-transfected side (Fig. 1E-G). Consistent with the horizontal section results, coronal brain sections unveiled that the cortical thickness (C2) of transfected area in Pax6-overexpressing brain was expanded compared to the control (C1, non-transfected area) (Fig. 1H and J). These data together indicated that Pax6 overexpression led to an increase in cortical thickness.

Fig. 1.

Pax6 overexpression resulted in abnormalities of cortical patterning. (A) pCAGGS-EGFP (Control) or pCAGGS-Pax6-EGFP (Pax6 overexpression) constructs were transfected into E15.5 embryo cortex by in utero electroporation, and brain samples were collected three days after electroporation (E15.5 + 3d) for DAPI staining (E, H, andI), and thionine staining (K and L). (B and C) are the overview of the brain of the control and Pax6 overexpression group. (D) Statistical graph of brain size of the control and Pax6 overexpression. (E) Horizontal section of Pax6-overexpressing brain where the right side is Pax6-transfected. (F) is the amplification of the framed area of E. (G) Statistical graph of cortical thickness of the non-transfected and transfected side of horizontal Pax6-overexpressing brain section. (H and I) are the coronal section of non-transfected and transfected cortex of Pax6-overexpressing brain. (J) Statistical graph of cortical thickness of non-transfected and transfected side of coronal Pax6-overexpressing brain section. Thionine staining was performed on the coronal section of Pax6-overexpressing brain to label cortical layers; (K and L) show non-transfected and transfected cortex of coronal Pax6-overexpressing brain section, respectively. (M) Statistical graph of the thickness of each cortical layer in non-transfected and transfected cortex of coronal Pax6-overexpressing brain section. (N) Statistical graph of the percentage of each sublayer in cortical plate of non-transfected and transfected cortex of coronal Pax6-overexpressing brain section. Scale bars are 200 μm in E, 50 μm in (F, K and L), and 100 μm in (H and I). Abbreviations: OE, overexpression; VZ, ventricular zone; SVZ; subventricular zone; IZ, intermediate zone; CP, cortical plate; MZ, marginal zone; ns, no significance. n = 3. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

To analyze the increase of cortical layer thickness after Pax6 overexpression, thionine staining was conducted on coronal brain sections to mark cortical layers. In the cortex of non-transfected and transfected sides of Pax6-overexpressing brains, ventricular zone (VZ), subventricular zone (SVZ), intermediate zone (IZ), cortical plate (CP) and marginal zone (MZ) were clearly stained (Fig. 1K and L). Compared with the non-transfected (Fig. 1K), the IZ and CP of the transfected side (Fig. 1L) dramatically thickened, up to 2.8- and 1.6- fold increases, respectively, whereas no obvious change was detected in the VZ, SVZ and MZ (Fig. 1M), implicating that the thickness increase of the IZ and CP was responsible for the expansion of cortical layers after Pax6 overexpression. Previous studies have reported that six sublayers (I-VI) were formed in the developing cortical plate by E18^30,31. Here, we also observed the formation of six sublayers of the CP in the non-transfected and transfected sides of E15.5 + 3d Pax6-overexpressing brains (Fig. 1K and L). However, the percentage of each sublayer in CP showed no difference between non-transfected and transfected sides (Fig. 1N), in spite of an increase in the CP thickness in transfected side. Immunostaining for Tbr1 was also conducted to assess the impact of Pax6 overexpression on the IZ. As depicted in Supplementary Fig. 1A-H, Tbr1-positive cells are situated in the upper stratum of the IZ in both control and Pax6-overexpressing cortices. Nonetheless, no significant alterations in IZ partitioning were detected following Pax6 overexpression.

Pax6 overexpression disrupts neuronal migration

To elucidate the impact of Pax6 overexpression on cortical thickness, immunofluorescent staining against Pax6 was performed on E15.5 + 3d coronal brain sections to trace the distribution of Pax6-overexpressing cells (Fig. 2A). Pax6 is restricted in the VZ of the developing cortex, which demarcates VZ from the CP³². Astonishingly, exogenous Pax6 proteins instead of endogenous proteins were detected by our anti-Pax6 antibody, as shown in Fig. 2F and G. In order to track neuronal migration during cortical development, the pCAGGS-EGFP construct (Control) was transfected into the E15.5 embryo cortex to mark cells (GFP⁺ cells). The pCAGGS-Pax6-EGFP construct was applied to overexpress Pax6 in the Pax6 overexpression group. However, the pCAGGS-Pax6-EGFP construct encodes a fused protein of Pax6 and GFP, and this fused protein is particularly localized in nucleus, which fails to allow observation of axonal guidance and migration trajectory. Therefore, the pCAGGS-EGFP and pCAGGS-Pax6-EGFP constructs were co-electroporated at a ratio of 1:4. Theoretically, those GFP⁺ cells should be Pax6 positive. However, immunostaining against Pax6 showed that not all GFP and Pax6 signals are colocalized, indicating that some GFP⁺ cells do not overexpress Pax6. Thus, Pax6 staining shows the localization of Pax6-overexpressing cells while their GFP labeling displays axonal guidance and migration trajectory.

Fig. 2.

Neurons were detained in the intermediate zone of cortex after Pax6 overexpression. Three days after the transfection of pCAGGS-EGFP (Control), and pCAGGS-Pax6-EGFP (Pax6 overexpression) constructs into E15.5 embryo cortex, brain samples (E15.5 + 3d) were coronally sectioned for subsequent examinations (A). Immunostainings against Pax6 were performed to track Pax6-overexpressing cells: DAPI (Blue in B and C), GFP (Green in D and E), Pax6 (Red in F and G), and Merge (H and I). (J) Statistical graph of percentage of GFP⁺ cells in different cortical regions of the control and Pax6 overexpression group. (K) Pie chart of the proportion of GFP⁺/Pax6⁺ cells in Pax6-overexpressing cortex. L) Statistical graph of percentage of GFP⁺ or GFP⁺/Pax6⁺ cells in different cortical regions of the control and Pax6 overexpression group. (M) Schematic diagrams of neuronal migration in Pax6-overexpressing cortex at E15.5 + 3d. GFP⁺/Pax6⁺ cells are pointed by white arrowheads in (G and I). Scale bar = 50 μm. Abbreviations: OE, overexpression; VZ, ventricular zone; SVZ, Subventricular zone; IZ, intermediate zone; CP, cortical plate; MZ, marginal zone. n = 3. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

The pCAGGS-EGFP construct demonstrates a higher expression efficiency compared to the pCAGGS-Pax6-EGFP construct. In the Pax6 overexpression group, the quantity of the construct mixture varies among transfected cells after in utero electroporation, leading to a gradient in Pax6 expression levels. Cells with lower amounts of the construct mixture may express high levels of GFP but insufficient Pax6 to be detected by immunostaining, and are thus classified as GFP⁺/Pax6⁻. Conversely, cells with higher amounts of the construct mixture express high levels of both GFP and Pax6, making them detectable by immunostaining and thus classified as GFP⁺/Pax6⁺.

Three days after electroporation, the majority of GFP⁺ cells in the control (Green in Fig. 2D) migrated upward from the VZ to the MZ in the cortex. Around 51% of them reached the CP, 23% to the IZ, 18% to the SVZ, whereas only 8% remained in the VZ (Fig. 2B, D and H, and 2J). Under conditions of Pax6 overexpression, the distribution of GFP⁺ cells varied considerably, 28% in the CP, 42% in the IZ, 16% in the SVZ, and 14% the in VZ (Fig. 2C, E and I, and 2J). These findings revealed that Pax6 overexpression resulted in an increase in GFP⁺ cells in the IZ with a concomitant decrease in the CP (Fig. 2J), supporting the notion that Pax6 overexpression inhibits neuronal migration. Immunostaining results showed that only 24% of GFP⁺ cells were also positive for Pax6 in Pax6-overexpressing brain (pointed out by white arrowhead in Fig. 2G and I, and 2K). Additionally, GFP⁺/Pax6⁺ cells failed to migrate to the CP (Yellow in Fig. 2I); 56% of GFP⁺/Pax6⁺ cells were distributed in the IZ, 20% in the SVZ and 24% in the VZ (Fig. 2G and I, and 2L). The percentage of GFP⁺/Pax6⁺ cells in the IZ of Pax6-overexpressing cortex significantly exceeded that of GFP⁺ cells in the control (Fig. 2L). These findings together support that Pax6 overexpression inhibits neurons from crossing the IZ to the CP (Fig. 2M). Therefore, we propose that Pax6 overexpression led to the detention of neurons in the IZ, which then increased the thickness of the IZ. Notably, a dramatic increase in the percentage of GFP⁺/Pax6⁺ cells was also detected in the VZ (Fig. 2L), suggesting that Pax6 overexpression might promote cell proliferation (Discussed in Sect. 3.5).

Pax6 overexpression temporally inhibits neuronal migration

To examine the effect of Pax6 overexpression on neuronal migration over time, we subsequently performed in utero electroporation on E13.5 mouse embryos, and collected samples 1 day, 3 days, 5 days, and 7 days after electroporation, respectively, for immunostaining against Pax6 (performed on coronal brain sections). One day after electroporation (E13.5 + 1d), GFP⁺ cells in the control (green in Fig. 3B) started migrating outwardly from the VZ, especially some reaching the IZ (Fig. 3B and D), which was also observed in Pax6-overexpressing cortex (green in Fig. 3F and H). The distribution of GFP⁺/Pax6⁺ cells in the VZ/SVZ and IZ of Pax6-overexpressing cortex showed no significant change compared to that of GFP⁺ cells in the control (Fig. 3C, D, G and H, and 3g), suggesting that overexpression of Pax6 had no effect on neuronal migration during this process. No alterations in cortical thickness were detected between control and Pax6 overexpression groups (Fig. 3A and E, and Supplementary Fig. 2A).

Fig. 3.

Pax6 overexpression temporally disrupted neuronal migration during corticogenesis. pCAGGS-EGFP (Control) or pCAGGS-Pax6-EGFP (Pax6 overexpression) constructs were transfected into E13.5 embryo cortex, and brain samples were collected respectively at 1 day (E13.5 + 1d, A-H), 3 days (E13.5 + 3d, I-P), 5 days (E13.5 + 5d, Q-X), and 7 days (E13.5 + 7d, Y-f) after electroporation for immunostainings against Pax6. Immunostaining shows DAPI (Blue in A, E, I, M, Q, U, Y, and c), GFP (Green in B, F, J, N, R, V, Z, and d), Pax6 (Red in C, G, K, O, S, W, a, and e), and Merge (D, H, L, P, T, X, b, and f). White arrowheads highlight GFP⁺/Pax6⁺ cells in Pax6 overexpression group. (g-j) Statistical graph of percentage of GFP⁺ or GFP⁺/Pax6⁺ cells in different cortical regions of the control and Pax6 overexpression at 1 day (g), 3 days (h), 5 days (i) and 7 days (j) after electroporation. (k) Schematic diagrams of neuronal migration in Pax6 overexpressing cortex at E13.5 + 1d, E13.5 + 3d, E13.5 + 5d and E13.5 + 7d. Scale bar = 50 μm. Abbreviations: OE, overexpression; VZ/SVZ, ventricular zone/subventricular zone; IZ, intermediate zone; CP, cortical plate; MZ, marginal zone; ns, no significance. n = 3. *p < 0.05, ***p < 0.001, ****p < 0.0001.

Three days after electroporation (E13.5 + 3d), Pax6 overexpression caused abnormalities in neuronal migration (Fig. 3J and N). In the control, roughly 52% of GFP⁺ cells (green in Fig. 3J-L) could reach the CP, while 19% and 29% were located in the IZ and VZ/SVZ, respectively (Fig. 3h). Immunostaining results against Pax6 revealed that 58% of GFP⁺/Pax6⁺ cells in Pax6-overexpressing cortex were distributed in the IZ, and the rest in the VZ/SVZ (Fig. 3N-P, and 3h), demonstrating that Pax6 overexpression inhibited neuronal migration. In addition, GFP⁺ cells of the control projected axons laterally in the IZ (Fig. 3J), whereas no axonal projection was seen in the IZ of Pax6-overexpressing cortex (Fig. 3N). However, cortical thickness remained unaffected after Pax6 overexpression at this stage (Fig. 3I and M, and Supplementary Fig. 2A).

With the passage of time (five days after electroporation, E13.5 + 5d), GFP⁺ cells in the control continuously migrated outward, and 97% of them arrived at the CP (Fig. 3R-T, and 3i), which contributed to the expansion of the CP but reduction of the IZ (Fig. 3Q). Although 73% of GFP⁺ cells reached the CP in Pax6-overexpressing cortex (Fig. 3V), all GFP⁺/Pax6⁺ cells were unable to migrate to the CP and more than half of them (approximately 53%) remained resident in the IZ (Fig. 3W and X, and 3i), reiterating the inhibition effect of Pax6 overexpression on neuronal migration. It is worth noting that the cortical thickness of Pax6-overexpressing brains strikingly increased while the IZ became thicker (Fig. 3Q and U, and Supplementary Fig. 2A), resembling the results of E15.5 + 3d (Fig. 2B and C).

At postnatal day 1 (P1, seven days after electroporation, E13.5 + 7d), all GFP⁺ cells in the control arrived at the upper CP (Fig. 3Z-b, and 3j), and the neurofilament network was well established in the IZ and CP (Fig. 3Z), which was not observed in Pax6-overexpressing cortex (Fig. 3d). Surprisingly, 80% of GFP⁺/Pax6⁺ cells in Pax6-overexpressing cortex climbed up to the upper CP, 16% remained in the IZ, and 4% in the VZ (Fig. 3d-f, and 3j), implying that Pax6 overexpression failed to inhibit neuronal migration. Furthermore, there was no obvious difference in the cortical thickness between the control and Pax6 overexpression group, but the sublayer arrangement of the CP in Pax6-overexpressing cortex was not yet completed, compared with the control (Fig. 3Y and c, and Supplementary Fig. 2A). By analyzing the percentage of GFP⁺ or GFP⁺/Pax6⁺ cells in distinct regions at different times, we found that GFP⁺ cells in the control migrated upward from the VZ to the MZ over time (Fig. 3B, J, R and Z), whereas Pax6-overexpressing cells initially failed to cross IZ to CP and then restarted climbing up to CP (Fig. 3F, N, V and d, and 3k). These collective results demonstrated that Pax6 overexpression has a temporal inhibitory effect on neuronal migration and the modulation of cortical thickness.

Pax6 overexpression disrupts neuronal migration via affecting cell morphology

Given that Pax6-overexpressing cells failed to cross the IZ to the CP at E15.5 + 3d, we therefore analyzed neuronal migratory trajectory and cell morphology in the IZ to elucidate the specific effects of Pax6 on neuronal migration. In the control, 93% of GFP⁺ cells exhibited vertical migratory morphology, while the remaining cells presented lateral migratory morphology (Fig. 4A, C, E and G, and 4I). However, the migrating direction of GFP⁺ cells after Pax6 overexpression was disrupted, especially those Pax6-overexpressing cells (Fig. 4B, D, F and H). Although the majority of GFP⁺/Pax6⁺ cells maintained vertical migratory morphology, 29% were observed as lateral migratory morphology (Fig. 4I). Furthermore, the overexpression of Pax6 induced alterations in neuronal morphology. In this study, we categorized neuronal morphology in the IZ into three primary types: neurons with two processes (bipolar neuron), neurons with one process (unipolar neuron), and no-process neurons. In the control, 73% of neurons were unipolar neurons, which decreased to 52% after Pax6 overexpression (Fig. 4J and L). The proportion of bipolar (33%) and no-process (10%) neurons significantly increased in Pax6 overexpression group, compared to the control (16% and 7%, respectively). Interestingly, the morphology of Pax6-overexpressing cells from E13.5 + 1d to E13.5 + 7d was correlated to their migratory changes (Supplementary Fig. 2B-g): Pax6-overexpressing cells showed no change in cell morphology at E13.5 + 1d (Supplementary Fig. 2E and I); then their axons were retracted at E13.5 + 3d and E13.5 + 5d (Supplementary Fig. 2M, Q, U and Y), and recovered by E13.5 + 7d (Supplementary Fig. 2c and g).

Fig. 4.

Axonal outgrowth was affected by Pax6 overexpression. Immunostaining against Pax6 was performed on E15.5 + 3d brain sections (A-H): DAPI (Blue in A and B), GFP (Green in C and D), Pax6 (Red in E and F), and Merge (G and H). (I) Statistical graph of GFP⁺ and GFP⁺/Pax6⁺ cells with vertical or lateral migratory morphology in the IZ of control and Pax6 overexpression group, respectively. (J and K) are the amplification of the framed area in G and H. White arrowheads in J and K indicate GFP⁺ and GFP⁺/Pax6⁺ cells exhibiting distinct morphologies in the IZ of control and Pax6 overexpression group, respectively. (L) Statistical graph of neuronal cells of different morphologies in the IZ of the control and Pax6 overexpression group. GFP-SH-SY5Y and Pax6 OE-SH-SY5Y cell lines were generated by lentiviral transduction, followed by immunostaining against Pax6: DAPI in blue, GFP in green, and Pax6 in red (M and N). White arrowheads in M and N highlight GFP-SH-SY5Y and Pax6 OE-SH-SY5Y cells with different shapes, respectively. (O) Statistical graph of SH-SY5Y cells of different shapes in the control and Pax6 overexpression group. Scale bars are 50 μm in A-H, M, and N; 20 μm in J and K. Abbreviations: OE, overexpression; VZ, ventricular zone; SVZ, Subventricular zone; IZ, intermediate zone; CP, cortical plate; MZ, marginal zone. n = 3. *p < 0.05, **p < 0.01.

To further demonstrate the morphological change of Pax6-overexpressing cells, immunostaining against Pax6 was conducted using GFP-SH-SY5Y and Pax6 OE-SH-SY5Y cells (Fig. 4M and N). In the control group, 87% of SH-SY5Y cells displayed an irregular morphology, 4% were spherical, and 9% exhibited a rod-shaped or fusiform configuration (Fig. 4M and O). Following Pax6 overexpression, the distribution of these morphological types shifted significantly, with 69% showing an irregular shape, 12% becoming spherical, and 19% adopting a rod-shaped or fusiform morphology (Fig. 4N and O). Pax6 OE-SH-SY5Y cells were also utilized for CUT&Tag assay. Subsequent GO analysis revealed that Pax6 participates in axonogenesis and axon development (Supplementary Fig. 3A). Therefore, we hypothesized that Pax6 overexpression perturbs neuronal migration by altering cell morphology through the changes in the actin cytoskeleton and/or microtubular cytoarchitecture.

Pax6 overexpression triggers the proliferation of neural stem cells

The aforementioned data demonstrated that Pax6 overexpression resulted in the detention of neurons in the IZ, which then increased the thickness of cortical layers. However, if the major role of Pax6 overexpression was to inhibit neuronal migration, it merely caused the expansion of the IZ with a reduction of the CP instead of an increase in cortical thickness. Therefore, we postulated that Pax6 overexpression should promote the proliferation of neural stem cells. To verify our postulation, immunostaining against SOX2 was performed on the coronal sections of E15.5 + 3d Pax6-overexpressing brain to assess the effect of Pax6 overexpression on neural stem cells (Fig. 5A and H). The number of SOX2-positive (SOX2⁺) cells, indicative of neural stem cells, was significantly higher compared to that of the control group (Fig. 5K). Analysis of SOX2⁺ cells across cortical layers revealed a decrease in the SVZ and an increase in the CP after Pax6 overexpression (Fig. 5L). Although a slight increase in the IZ was observed, it did not reach statistical significance. Furthermore, the majority of SOX2⁺ cells did not colocalize with GFP⁺ cells, suggesting that Pax6-overexpressing cells are distinct from neural stem cells (Fig. 5I and J).

Fig. 5.

Pax6 overexpression promoted the proliferation of neural stem cells. Immunostaining against SOX2 was performed on coronal sections of Pax6-overexpressing brain to test the change of neural stem cells: DAPI (Blue in A and E), GFP (Green in B and F), SOX2 (Red in C and G), and Merge (D and H). (I and J) are the amplification of the framed area in (D and H), respectively. (K) Statistical graph of SOX2-postive cells in the cortex of control and Pax6 overexpression group. (L) Statistical graph of percentage of SOX2-positive cells in different layers of cortex. To further investigate whether Pax6 promoted cell proliferation, immunostaining against Ki67 was conducted on coronal Pax6-overexpressing brain sections (M-T): DAPI (Blue in M and Q), GFP (Green in N and R), Ki67 (Red in O and S), and Merge (P and T). (U and V) are the amplification of the framed area in P and T, respectively. (W) Statistical graph of Ki67-positive cells in cortex of control and Pax6 overexpression group. We also conducted immunostaining against Ki67 on SH-SY5Y cells transfected with pCAGGS-EGFP (control) or pCAGGS-Pax6-EGFP constructs (Pax6 overexpression) (X-e): DAPI (Blue in X and b), GFP (Green in Y and c), Ki67 (Red in Z and d), Merge (a and e). (f) Statistical graph of percentage of Ki67 positive cells in control and Pax6 overexpression group. Scale bars are 100 μm in (M-T); 50 μm in (A-H); 25 μm in (I, J, U, and V); 20 μm in X-e. Abbreviations: OE, overexpression; VZ, ventricular zone; IZ, intermediate zone; CP, cortical plate; MZ, marginal zone; ns, no significance. n = 3. *p < 0.05, ****p < 0.0001.

Immunostaining against Tbr2 or NeuN were performed to identify the cell type of Pax6-overexpressing cells (Supplementary Fig. 4A-T). In the control cortex, no Tbr2-positive (Tbr2⁺) cells were observed (Supplementary Fig. 4C); however, in the Pax6-overexpressing cortex, Tbr2⁺ cells were detected in the VZ, SVZ, and IZ (Supplementary Fig. 4G), suggesting that Pax6 overexpression may promote the generation of intermediate progenitor cells. Despite this, the majority of GFP⁺ cells did not co-label with Tbr2 (Supplementary Fig. 4D and H–J), implying that the Pax6-overexpressing cells are not intermediate progenitor cells. Furthermore, immunostaining against NeuN revealed that the majority of GFP⁺ cells in the Pax6-overexpressing cortex were NeuN-positive (Supplementary Fig. 4N and R-T), indicating that these cells are indeed neurons.

To further substantiate the impact of Pax6 overexpression on cell proliferation, we conducted immunostaining against Ki67 (Fig. 5M and T). In the control, Ki67-positive (Ki67⁺) cells were predominantly found in the VZ, aligning with the established view that cortical neurogenesis primarily occurs in the VZ (Fig. 5O). In contrast, the majority of Ki67⁺ cells were detected in the VZ and SVZ of Pax6-overexpressing cortex (Fig. 5S). Notably, Ki67⁺ proliferating cells of Pax6-overexpressing cortex significantly outnumbered those of the control (Fig. 5W), robustly supporting our hypothesis. Furthermore, most cells in the cortex of both control and Pax6 overexpression group did not exhibit co-labelling with GFP and Ki67 (Fig. 5U and V). Immunostaining against Ki67 were also performed on SH-SY5Y cells transfected with either pCAGGS-EGFP (Control) or pCAGGS-Pax6-EGFP constructs (Pax6 overexpression) (Fig. 5X and e). The results revealed that only 23% of SH-SY5Y cells were Ki67-positive in the control group (Fig. 5Z and f), whereas approximately 88% of SH-SY5Y cells in the Pax6-overexpression group, including both Pax6-overexpressing cells and their neighboring cells, exhibited Ki67 positivity (Fig. 5d and f). Cell cycle analysis, conducted using flow cytometry after PI staining, demonstrated a significant increase in the S phase population within Pax6 OE-SH-SY5Y cells (Supplementary Fig. 4U). These in vivo and in vitro results together support that Pax6 overexpression promoted cell proliferation. Given that the vast majority of Pax6 overexpressing cells were not labelled by Ki67, we propose that Pax6 might regulate cell proliferation in a non-autonomous manner.

By further analyzing cell cycle results, we found that the SubG1 peak (2.93) in Pax6 OE-SH-SY5Y cells was relatively lower than that in the control (3.15) (Supplementary Fig. 4U), suggesting that Pax6 overexpression might have a slight inhibitory effect on apoptosis. To ascertain if the observed increase in cortical thickness following Pax6 overexpression was attributable to the inhibition of apoptosis, we conducted Western blot analysis targeting BAX, BCL2, and cleaved-caspase-3 in Pax6 OE-SH-SY5Y cells. BCL2 is an anti-apoptotic protein, while BAX is a pro-apoptotic protein³³. The BCL2/BAX ratio is a crucial indicator in the regulation of apoptosis; A high BCL2/BAX ratio indicates that anti-apoptotic signals are dominant, whereas a low BCL2/BAX ratio suggests that pro-apoptotic signals are dominant³⁴. Cleaved caspase-3 is the active form of caspase-3, an enzyme that plays a critical role in the execution phase of apoptosis³⁵. The analysis revealed no significant alterations in the BCL2/BAX ratio or in the levels of cleaved-caspase-3 (Supplementary Fig. 4V and W), indicating that Pax6 overexpression does not exert a suppressive effect on apoptotic pathways.

Pax6 participates in the regulation of Wnt3a in an indirectly manner

Thousands of Pax6 downstream genes that participate in cortical development were identified by genome-wide transcriptomic analyses³². Our analysis of Pax6 in the STRING Consortium 2024 database revealed that Wnt3a is one of the key interactors of Pax6 in regulating stem cell pluripotency³⁶. It was reported that Wnt3a overexpression leads to an increase in neuronal accumulation and promotes neuronal proliferation in chicken spinal cord³⁷. Furthermore, JASPAR analysis identified 5 potential binding sites within the regulatory sequence of the mouse Wnt3a gene, as depicted in Fig. 6A³⁸. Based on these findings, we hypothesize that Pax6 may interact with Wnt3a to regulate neuronal migration and enhance neuronal proliferation. Immunostaining for Wnt3a was conducted to assess the impact of Pax6 overexpression on Wnt3a expression levels (Fig. 6B and I). No obvious signals of Wnt3a were detected in the VZ, SVZ, and IZ of the control (Fig. 6D and J), whereas in the Pax6-overexpression group we clearly observed Wnt3a positive cells in the corresponding area (Fig. 6H and K), supporting that Pax6 overexpression significantly upregulated Wnt3a expression.

Fig. 6.

Pax6 overexpression induced the expression of Wnt3a. (A) Table of 5 Pax6 potential binding sites within the regulatory sequence of Wnt3a. Immunostaining against Wnt3a was conducted on coronal sections of Pax6-overexpressing brain (B-I): DAPI (Blue in B and F), GFP (Green in C and G), Wnt3a (Red in D and H), and Merge (E and I). J and K are the amplification of the framed area in E and I, respectively. (L) Quantitative PCR was performed to detect the alteration of Pax6 and Wnt3a at the mRNA level in Pax6-overexpressing N2a cells. The level of Pax6/Wnt3a protein in Pax6-overexpressing N2a cells (M) and their medium (O) was examined by Western blot. (N and P) are the statistical graph of (M and O). Scale bars are 50 μm in (B-I), 25 μm in (J and K). Abbreviations: Con, control; OE, overexpression; VZ, ventricular zone; SVZ, Subventricular zone; IZ, intermediate zone; CP, cortical plate; MZ, marginal zone. n = 3 in B-I and M-P; n = 4 in L. *p < 0.05, **p < 0.01, ****p < 0.0001.

To further demonstrate whether Pax6 regulates Wnt3a expression, quantitative PCR (qPCR) and Western blot were utilized to detect the alteration of Wnt3a in Pax6-overexpressing N2a cells at the mRNA and protein levels. The mRNA level of Pax6 in Pax6-overexpressing N2a cells was on average 6000-fold higher than that in the control (Fig. 6L), indicating that the endogenous expression of Pax6 was kept at a very low level in normal N2a cells. A striking increase of Wnt3a mRNA level was detected after Pax6 overexpression, up to 5000-fold compared to the control, supporting that Pax6 overexpression promotes Wnt3a transcription in vitro. It was also supported by the results of Western blot that Wnt3a was significantly upregulated in Pax6-overexpressing N2a cells (Fig. 6M and N). Given that Wnt3a is a secreted protein, the Wnt3a level in the medium of Pax6-overexpressing N2a cells was examined by Western blot to further support that Pax6 promotes the expression of Wnt3a protein. Consistent with our expectation, Wnt3a protein was dramatically elevated in the medium (Fig. 6O and P). Thus, we concluded that Pax6 regulates Wnt3a expression.

To determine whether Pax6 directly modulates Wnt3a transcription, we conducted CUT&Tag assay on Pax6 OE-SH-SY5Y cells to create a DNA library, which was then subjected to DNA sequencing. Despite identifying 58 potential Pax6 interactors from the sequencing data, Wnt3a was notably absent (highlighted in yellow in Supplementary Table 1), implying that Pax6’s regulation of Wnt3a transcription is likely indirect.

Wnt3a participates in the Pax6 overexpression-causing defects

in utero electroporation was conducted on E13.5 and E15.5 mouse embryos to investigate whether overexpressing Wnt3a generated Pax6 overexpression-like defects (Fig. 7A). Surprisingly, an obvious increase of cortical thickness was observed in the coronal sections of E13.5 + 3d Wnt3a-overexpressing brain (Fig. 7B) and E15.5 + 3d Wnt3a-overexpressing brain (Supplementary Fig. 5A). Three days after electroporation (E13.5 + 3d), we found that the transfected area in the Wnt3a-overexpression cortex protruded inwardly and formed a bulge, which did not occur at the non-transfected side (Fig. 7B), implying that Wnt3a overexpression led to the formation of the bulge. Compared with the non-transfected side (C1), the cortical thickness of Wnt3a-transfected side (C2) was significantly elevated (Fig. 7C and E), suggesting that Wnt3a promotes cell proliferation. An exception was the accumulation of neurons in the IZ after Wnt3a overexpression (Fig. 7D), indicating that Wnt3a overexpression might affect neuronal migration. Thionine staining revealed a significant decrease in cortical thickness in the VZ, SVZ and CP of Wnt3a-overexpressing cortex with a considerable increase in the IZ, compared with the control (Fig. 7F and H). In terms of neuronal migration, 30% of GFP⁺ cells in Wnt3a-overexpressing cortex migrated to the CP, 57% to the IZ and 13% to the VZ (Fig. 7I and K). Compared with the control, the percentage of GFP⁺ cells of Wnt3a-overexpressing cortex dramatically increased in the IZ, but decreased in the VZ and CP (Fig. 7I and K), supporting that Wnt3a overexpression inhibited neuronal migration. In addition, some GFP⁺ cells of Wnt3a-overexpressing cortex accumulated in the IZ (Fig. 7J), explaining the accumulation of neurons observed in IZ (Fig. 7D).

Fig. 7.

Wnt3a participates in the Pax6-overexpression-causing defects. pCAGGS-EGFP (Control), or pCAGGS-EGFP + pECMV-Wnt3a-Flag (Wnt3a overexpression) constructs were transfected into E13.5 embryo cortex, and brain samples were collected three days after electroporation (E13.5 + 3d) for DAPI staining (B-D, I and J) or Thionine staining (F-G). (A) Schematic graph of coronal section of Wnt3a-overexpressing brain. (B) Coronal overview of Wnt3a-overexpressing coronal brain section. (C and D) are the amplification of non-transfected and transfected cortex of Wnt3a-overexpressing brain. (E) Statistical graph of cortical thickness of non-transfected and transfected cortex in Wnt3a-overexpressing brain. Thionine staining was performed on the coronal brain sections of control (F) and Wnt3a overexpression (G) to label cortical layers. (H) Statistical graph of the thickness of each layer in the control and Wnt3a overexpression. (I and J) show DAPI staining of control and Wnt3a-overexpressing cortex: DAPI in Blue, and GFP in Green. (K) Statistical graph of percentage of GFP⁺ cells in each layer of the control and Wnt3a-overexpressing cortex. To investigate whether Wnt3a participates in the Pax6-overexpression-causing defects, Wnt3a was co-suppressed with Pax6 overexpression in E15.5 embryo cortex, and brain samples were collected in 3 days (E15.5 + 3d) for immunostaining against Pax6. (L) Schematic graph of in utero electroporation and coronal brain section. Immunostaining against Pax6 was performed in the control (M, P, S, and V), Pax6 overexpression (N, Q, T, and W) and Pax6 overexpression + Wnt3a knockdown (O, R, U, and X): DAPI (Blue in M-O), GFP (Green in P-R), Pax6 (Red in S-U), and Merge (V-X). White arrowheads highlight GFP⁺/Pax6⁺ cells in Pax6 overexpression and Pax6 overexpression + Wnt3a knockdown group. (Y) Statistical graph of percentage of GFP⁺ or GFP⁺/Pax6⁺ cells in different cortical regions of the control, Pax6 overexpression and Pax6 overexpression + Wnt3a knockdown. Scar bars are 200 μm in (B); 50 μm in (C, D, I, J, and M-X); 30 μm in (F and G). Abbreviations: A, anterior; P, Posterior; OE, overexpression; KD, knockdown; VZ, ventricular zone; SVZ; subventricular zone; IZ, intermediate zone; CP, cortical plate; MZ, marginal zone; ns, no significance. n = 3. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

To further explore whether Wnt3a participated in the Pax6 overexpression-causing defects, co-suppression of Wnt3a using Wnt3a shRNA with Pax6 overexpression was performed on E15.5 mouse embryos (Fig. 7L). The knockdown efficacy of Wnt3a shRNA was validated by Western blot analysis (Supplementary Fig. 5B and C). Fortunately, the detention of Pax6-overexpressing cells in the IZ was rescued by Wnt3a knockdown (Fig. 7U). In Pax6-overexpressing cortex, 39% of GFP⁺/Pax6⁺ cells were observed in the IZ, 22% in the SVZ and 39% in the VZ (Fig. 7T and Y). However, 21% of GFP⁺/Pax6⁺ cells could reach the CP in the rescue group (Pax6 overexpression + Wnt3a knockdown) while GFP⁺/Pax6⁺ cells in the IZ decreased to 29% (Fig. 7U and Y), supporting that Wnt3a knockdown may have restarted the migration of Pax6-overexpressing cells. These together demonstrated that Wnt3a might participate in the Pax6 overexpression-causing events. Furthermore, the rescue effect of Wnt3a was limited since the cortical layers of Pax6 overexpression + Wnt3a knockdown remained thicker than the control (Fig. 7N and O, and Supplementary Fig. 5D), suggesting that Wnt3a is only one of the regulators acting downstream of Pax6 in this process.

Discussion

Pax6 is critical for the development of eyes, brain, and endocrine glands³⁹. Absence of Pax6 causes neonatal death in mice, as well as loss of eyes and malformation of the cerebral cortex⁴⁰. Heterozygous Pax6 mutations generate aniridia in humans, and the small eye phenotype in mice^41,42. A significant effort has been invested in studying Pax6’s functions in eyes, whereas researchers started to investigate the role of Pax6 in brain development in recent years. Pax6 plays a vital role in forebrain development². Sey rodents display a reduction of forebrain and cortical thickness¹⁹; humans with Pax6 haploinsufficiency suffer from a reduction in thickness of the frontoparietal cortex⁴³, suggesting that Pax6 is required for the proliferation of cortical progenitors. Therefore, it was postulated that Pax6 overexpression might increase cortical thickness. However, Manuel et al. reported no significant change in cortical thickness or brain size of the P7 PAX77 mouse (Pax6-overexpression mouse) brain, and suggested that Pax6 autoregulation contributes to it²⁹. Intriguingly, this study revealed that overexpression of Pax6 in the cortex of E18.5 mouse embryo increased cortical thickness. What is more intriguing is that this change disappeared in P1 Pax6-overexpressing cortex, consistent with the observations of Manuel et al.. Of how these changes occurred will be discussed later.

In E18.5 Pax6^Sey/Sey cortex, Pax6-null cells reside at the SVZ/IZ boundary, supporting that Pax6 is required for neuronal migration²⁷. Interestingly, we found that Pax6-overexpressing cells were detained in the IZ of E18.5 Pax6-overexpressing cortex, suggesting that Pax6 inhibits neuronal migration. Therefore, we postulated that neuronal migration is sensitive to the expression level of Pax6—too much or too little both inhibit neuronal migration. To our surprise, the impact of Pax6 overexpression on neuronal migration was temporal, which was not reported previously. Three days after electroporation (E16.5), Pax6-transfected cells failed to climb up to the CP and MZ, and remained in the IZ, which persisted to five days after electroporation (E18.5). What unexpectedly occurred in two days (P1) was that Pax6-overexpressing cells resumed migrating outward and some even reached the upper CP. Mammalian Pax6 encodes at least three isoforms: Pax6, Pax6(5a), and Pax6(ΔPD)¹. It is hypothesized that these isoforms may collaborate to regulate normal development through balancing the ratio of these isoforms⁴⁴. The ratio of three Pax6 isoforms in wild-type N2a exhibits Pax6 > Pax6(5a) > Pax6(ΔPD), while overexpression of Pax6 alters it as Pax6(ΔPD) > Pax6 > Pax6(5a)²³. In addition, Manuel et al. reported that Pax6 levels are stabilized by autoregulation²⁹. Therefore, we proposed that restarting neuronal migration in P1 Pax6-overexpressing cortex might be due to the change in the ratio of these three Pax6 isoforms by autoregulation. However, what ratio it might turn to remains to be investigated.

During cortical development, several modes of neuronal migration have been described, including multipolar migration, somal translocation, glia-guided locomotion, and tangential migration^26,45. Early-born neurons adopt somal translocation characterized by neurons initially extending the leading process to the pial surface, followed by the advance of soma (nucleokinesis)⁴⁵. However, late-born neurons undergo multipolar migration in the VZ/SVZ, then turn to glia-guided locomotion in the IZ, and finally choose tangential migration to reach the functioning position^24,45. Pax6 was postulated to regulate radial migration (somal translocation and glia-guided locomotion) via mediating cell properties during corticogenesis^27,46. This study aimed to investigate how Pax6 regulates glia-guided locomotion. Neurons undergoing glia-guided locomotion have a free motile leading process and glia cells act as a scaffold in neuronal movement⁴⁵. In the IZ of E18.5 control cortex, GFP⁺ cells had a long process and migrated vertically, suggesting glia-guided locomotion, whereas GFP⁺/Pax6⁺ cells in Pax6-overexpressing cortex migrated tangentially, indicating that overexpression of Pax6 switches glia-guided locomotion to tangential migration. Meanwhile, cell morphology was altered after Pax6 overexpression, implying that Pax6 overexpression changed the migratory mode through cell shape. These explain in part why Pax6-overexpressing cells failed to cross the IZ to the CP. Additionally, the migratory direction of GFP⁺/Pax6⁻ cells in Pax6-overexpressing cortex was also disrupted, hinting that Pax6 overexpression might induce signaling molecules to affect neuronal migration.

Pax6 plays a critical role in maintaining progenitor cells during cortical development. Sakayori et al. reported a reduction in the proliferative activity of neural stem/progenitor cells derived from heterozygous Pax6 mutant (rSey(2)/+) rats⁴⁷. Consistent with it, we found that Pax6 overexpression induced a significant increase of neural stem cells and intermediate progenitor cells in E18.5 cortex, as well as elevation of proliferating cells, reiterating that Pax6 promotes cell proliferation. Interestingly, Pax6-overexpressing cells affected the proliferative activity of their neighboring cells, suggesting that Pax6 might induce secreted proteins to trigger cell proliferation. Given that Pax6-overexpressing cells stopped at the IZ, this explained the dramatic increase of IZ thickness in Pax6-overexpressing cortex. However, the CP also became thicker after Pax6 overexpression. One explanation is that Pax6-transfected cells in the IZ secrete signaling proteins which induce proliferation in non-transfected cells, and these non-transfected cells could subsequently migrate to the CP, ultimately increasing the CP thickness. Regarding the recovery of cortical thickness in P1 Pax6-overexpressing brain, we postulated that it is associated with the alteration in the ratio of Pax6 isoforms. Pax6 and Pax6(5a) both reportedly inhibit progenitor proliferation in the developing cortex⁴⁸, whereas Kim et al. proposed that Pax6(ΔPD) might antagonize Pax6 function in the eye⁴⁹. The ratio of Pax6 isoforms presented Pax6(ΔPD) > Pax6 > Pax6(5a) in Pax6-overexpressing N2a cells, suggesting that Pax6(ΔPD) might be the key factor for promoting cell proliferation in E18.5 Pax6-overexpressing cortex. Therefore, we hypothesize that the ratio of Pax6(ΔPD) was reduced through negative autoregulation in P1 Pax6-overexpressing cortex, which then recovered cortical thickness.

Pax6 function is partly based on its ability to interact with various transcription factors and synergistically regulate target gene expression. Many genes have been identified as the downstream targets of Pax6 in cortical development, such as FABP7, Ngn2, p27¹, and Wnt7b^1,50–52. Mus musculus datasets of the STRING Consortium 2024 display that Wnt3a is one of the Pax6 interactors in regulating pluripotency of stem cells. A growing body of evidence demonstrates that Wnt3a regulates cell proliferation and migration in different types of cells^53–60. In our previous study, overexpressing Wnt3a in the chicken spinal cord had an influence on neuronal migration while promoting neuronal proliferation³⁷. Additionally, JASPAR analysis reveals five potential Pax6 binding sites within the Wnt3a regulatory sequence. These together imply that Wnt3a participates in Pax6 overexpression-causing defects. Here, we detected a dramatic increase of Wnt3a level in the IZ of Pax6-overexpressing cortex, indicating that Pax6 might induce Wnt3a. qPCR and Western blot results further confirmed that Pax6 participated in the regulation of Wnt3a expression. This result may also explain why Pax6-overexpressing SH-SY5Y cells promote the proliferation of their neighboring cells. Our hypothesis is that, Pax6-overexpressing cells induced the secretion of Wnt3a into the medium, stimulating their surrounding cells to multiply. Furthermore, Wnt3a overexpression produced Pax6 overexpression-like defects while Pax6 overexpression-causing defects were partially rescued by Wnt3a knockdown. These support a proposition that Pax6 participates in corticogenesis via interacting with Wnt3a on neuronal migration and cell proliferation. CUT&Tag assay results indicated that Pax6 proteins did not bind to the Wnt3a sequence, suggesting that Pax6 may not directly regulate Wnt3a transcription. However, this finding requires further validation given that the CUT&Tag assay was conducted on SH-SY5Y cells overexpressing murine Pax6 proteins. KEGG pathway analysis of these Pax6 OE-SH-SY5Y cells indicated potential Pax6 regulation of the MAPK signaling pathway and the actin cytoskeleton (Supplementary Fig. 3B). Five downstream genes of Pax6 (FGFR2, PAK1, MAP2K1, PAK2, and FGF13) were identified through gene interaction (Supplementary Fig. 3C). Considering the ERK pathway’s role in Wnt3a-mediated proliferation and motility^61,62, our subsequent studies will explore whether Pax6 might modulate cortical development through the Wnt3a/MAPK/ERK signaling cascade.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Author contributions

B.Z. prepared figures and wrote the manuscript. M.H., J.H. and Y.L. conducted experiments. All authors reviewed the manuscript.

Data availability

All data supporting the findings in this study are available within the manuscript, as well as supplementary information. Any further details will be available upon request. Please contact Dr. Bichao Zhang, Bichao.Zhang@xxmu.edu.cn.

Declarations

Competing interests

The authors declare no competing interests.