Conditional knockout of Ybx1 in the nervous system results in perinatal lethality and impaired cortical development

To investigate the role of the m⁵C reader Ybx1 in brain development, we first examined its spatiotemporal expression patterns within the developing mouse brains. In situ hybridization and immunofluorescence assays showed that Ybx1 is highly expressed in the whole cortex from E11.5 to E15.5 (Fig 1A and 1B). To further confirm the expression pattern of Ybx1 protein in the cortex regions at E13.5 and E15.5, we performed immunofluorescence of Ybx1 using adjacent sections and co-stained with markers of radial glial cells (RGCs, marked by Pax6) and intermediate progenitors (IPs, marked by Tbr2). The results showed that Ybx1 exhibits high expression levels in both progenitor cells and neurons at E13.5 and E15.5 (S1A Fig), which is consistent with the published single-cell RNA sequencing (scRNA-seq) data [28]. As cortical neurogenesis completes at E17.5 and postnatal day 0 (P0), Ybx1 expression diminishes in the proliferative zones but remains prominent in the cortical plate (CP) (Fig 1A and 1B). In summary, the high expression of Ybx1 in the embryonic mouse cortex suggests its potential role in the embryonic cortical neurogenesis.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Ybx1 is highly expressed in the progenitor cells of earlier embryonic mouse cortex and conditional ablation of Ybx1 causes impaired cortical development.

(A) In situ hybridization using DIG-labeled RNA probes (antisense and sense) targeting Ybx1 was performed on coronal sections of mouse brains at E11.5 to P0. Representative cortical regions were shown as magnified images. Scale bars, 500 μm. (B) Immunostaining for Ybx1 protein was done on coronal sections of mouse brains at E11.5, E13.5, E15.5, E17.5, and P0. Representative cortical regions were shown. Scale bars, 100 μm. (C) Representative images of P0 neonatal mice from the same litter with different genotypes were shown at 2 and 12 h after birth, respectively. (D) Quantitative analysis of neonatal body weight at P0 was shown as box and whisker plots. Ybx1^fl/fl, Nestin-Cre, Ybx1^+/fl, and Nestin-Cre, Ybx1^fl/fl were designated as control (Ctrl), heterozygous (Het), and cKO, respectively. Ctrl (n = 39 pups) vs. Het (n = 26 pups), p = 0.37; Ctrl vs. cKO (n = 28 pups), **p = 0.0014; Het vs. cKO, *p = 0.021; ns, not significant; all by one-way ANOVA followed by Tukey’s multiple comparison test. (E) Representative images of the P0 brains from cKO (#1), Het (#2), and Ctrl (#3) pups shown in (D). Black arrowed lines indicate the cortical length and width. (F) Quantitative analysis of cortical length and width at P0 was shown as box and whisker plots. n = 21 pups for Ctrl, n = 15 pups for Het, n = 16 pups for cKO: p = 0.65 for width, p = 0.80 for length; ns, not significant; all by one-way ANOVA followed by Tukey’s multiple comparison test. (G) Representative images of P0 coronal brain sections stained with an antibody against Ybx1 protein demonstrated efficient ablation of Ybx1 in the cKO brain. The asterisk indicates the enlarged ventricle in the cKO brain. Scale bar, 1,000 μm. (H) Quantification of lateral ventricle area of P0 coronal brain sections was presented as box and whisker plots: Ctrl (n = 21 confocal fields) vs. Het (n = 22 confocal fields), *p = 0.035; Ctrl vs. cKO (n = 21 confocal fields), ****p = 2.03E−11; Het vs. cKO, ****p = 2.03E−011; all by one-way ANOVA followed by Tukey’s multiple comparison test. (I) Quantification of P0 cortex thickness was presented as box and whisker plots: Ctrl (n = 20 confocal fields) vs. cKO (n = 21 confocal fields), ****p = 6.73-05; Het (n = 19 confocal fields) vs. cKO, **p = 0.0014; ns, not significant; all by one-way ANOVA followed by Tukey’s multiple comparison test. The data underlying all the graphs shown in the figure are included in S1 Data.

https://doi.org/10.1371/journal.pbio.3003175.g001

To explore the potential functions of Ybx1 during the onset of neurogenesis, we generated mice with floxed Ybx1 alleles (referred to as Ybx1^fl/fl) by crossing them with Nestin-Cre mice (S1B Fig). We analyzed three genotypes: Nestin-Cre,Ybx1^fl/fl (Ybx1 cKO); Nestin-Cre,Ybx1^+/fl (heterozygote, Het); and Ybx1^fl/fl (control, Ctrl) mice. All Ybx1 cKO mice died within 1–2 days after birth, with no surviving pups observed after 2 days. Ybx1 cKO mice are smaller and have reduced body weight compared to their control littermates at postnatal day 0 (P0) (Fig 1C and 1D). When examining the brains of P0 cKO pups, no obvious anatomical abnormalities were observed, and there were no differences in cortical length and width compared to control littermates (Fig 1E and 1F). Subsequent immunostaining of coronal brain slices from P0 mice confirmed the successful ablation of Ybx1 in the cKO cortex (Fig 1G). However, we noted a significant enlargement of the lateral ventricles and a decrease in cortical thickness in Ybx1 cKO pups, compared to their control littermates (Fig 1G–1I and S1C Fig). These data suggest that the removal of Ybx1 during the initiation of neurogenesis leads to defects in cortical development.

Conditional ablation of Ybx1 leads to a reduction in cortical progenitor cell pool and a decrease in cortical thickness

Within the developing neocortex, two distinct progenitor cell types are present. Radial glial cells (RGCs) situated in the ventricular zone (VZ) can be marked by Pax6. Proliferative populations derived from RGCs form the subventricular zone (SVZ) and are designated as intermediate progenitor cells (IPCs). The transcription factor Tbr2 is primarily expressed in IPCs within the SVZ [7]. To more comprehensively characterize the impact of Ybx1 deletion on cortical development, we performed immunostaining for Pax6 and Tbr2 to assess the neural progenitor cell pool (Fig 2A). The number of Pax6⁺ RGCs was reduced by over 50% in P0 Ybx1 cKO mice compared to control littermates, while Tbr2⁺ cells decreased by 30% (Fig 2B and 2C). Consistently, we observed that the thicknesses of both VZ and SVZ were significantly reduced in Ybx1 cKO pups compared to the control ones (Fig 2D and 2E), and the overall thickness of the cortex decreased by 20% (Fig 2F). All experimental results in this study (including the following figures) showed no significant differences between the heterozygous and control littermates.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Conditional ablation of Ybx1 leads to a reduction in cortical progenitor cell pool and a decrease in cortical thickness.

(A) Immunostaining of coronal brain sections at P0 using antibodies against radial glial progenitor marker Pax6 and intermediate progenitor marker Tbr2. Representative cortical regions were shown. White dotted lines mark the boundaries. Scale bar, 100 μm. (B-F) Quantification of Pax6⁺ (B) and Tbr2⁺ (C) cell numbers, and cortical layer thicknesses (D–F) shown in (A). All statistical data are presented as box and whisker plots. For Pax6⁺ cell numbers, Ctrl (n = 26 confocal fields) vs. cKO (n = 24 confocal fields), ****p = 1.42E−05; Het (n = 23 confocal fields) vs. cKO, ****p = 9.54E−05. For Tbr2⁺ cell numbers, Ctrl (n = 26 confocal fields) vs. cKO (n = 24 confocal fields), ****p = 1.19E−05; Het (n = 23 confocal fields) vs. cKO, ***p = 1.85E−04. For VZ thickness, Ctrl (n = 26 confocal fields) vs. cKO (n = 24 confocal fields), ****p = 9.26E−05; Het (n = 23 confocal fields) vs. cKO, ****p = 1.33E−05. For SVZ thickness, Ctrl (n = 26 confocal fields) vs. cKO (n = 24 confocal fields), ***p = 9.49E−04; Het (n = 23 confocal fields) vs. cKO, **p = 0.0077. For whole cortical thickness, Ctrl (n = 26 confocal fields) vs. cKO (n = 24 confocal fields), ****p = 8.47E−06; Het (n = 23 confocal fields) vs. cKO, ****p = 2.61E−05. (G) Immunostaining of coronal brain sections at P0 for layer V marker Ctip2 and layer VI marker Tbr1. Representative cortical regions were shown. White dotted lines mark the boundaries of each layer. Scale bar, 100 μm. (H–M) Quantification of Ctip2⁺ (H) and Tbr1⁺ (I) neuron numbers, and the thickness of WM layer (J), layer VI (K), layer V (L), and layers I–IV (M) shown in (G). All statistical data are presented as box and whisker plots. For Ctip2⁺ neuron numbers, Ctrl (n = 22 confocal fields) vs. cKO (n = 20 confocal fields), ****p = 1.89E−06; Het (n = 20 confocal fields) vs. cKO, ***p = 1.62E−04. For Tbr1⁺ neuron numbers, Ctrl (n = 19 confocal fields) vs. cKO (n = 21 confocal fields), ****P = 4.62E−08; Het (n = 21 confocal fields) vs. cKO, ****P = 5.50E−05. For WM layer thickness, Ctrl (n = 22 confocal fields) vs. cKO (n = 20 confocal fields), ***p = 3.50E−04; Het (n = 20 confocal fields) vs. cKO, **p = 0.0038. For layer VI thickness, Ctrl (n = 22 confocal fields) vs. cKO (n = 20 confocal fields), ****p = 1.43E−05; Het (n = 20 confocal fields) vs. cKO, ***p = 2.56E−04. For layer V thickness, Ctrl (n = 22 confocal fields) vs. cKO (n = 20 confocal fields), ***p = 5.16E−04; Het (n = 20 confocal fields) vs. cKO, ***p = 5.71E−04. For thickness of layers, I–IV, Ctrl (n = 22 confocal fields) vs. cKO (n = 20 confocal fields), **p = 0.0093; Het (n = 20 confocal fields) vs. cKO, *p = 0.032. (N) A stacked bar chart summarizing the data in (D–F, J–M) shows the distribution of each layer in P0 cortex. At least 3 mice were analyzed for each genotype. All analyses were performed by one-way ANOVA followed by Tukey’s multiple comparison test. ns, not significant. The data underlying all the graphs shown in the figure are included in S1 Data.

https://doi.org/10.1371/journal.pbio.3003175.g002

Next we investigated whether the decrease in the neural stem cell pool due to Ybx1 loss was caused by apoptosis. TUNEL staining in the cerebral cortex showed that the number of apoptotic cells did not change in P0 Ybx1 cKO brain cortex compared to the control group (S2A and S2B Fig). Subsequently, we sought to investigate whether apoptosis starts at earlier stages. We performed immunofluorescence for cleaved caspase-3 (CC3) on cortical tissue at E13.5 and E15.5 (S2C and S2E Fig). Consistent with the data from the P0 stage, staining for CC3 revealed that loss of Ybx1 does not result in an increase in the number of apoptotic cells at E13.5 or E15.5 (S2C–S2F Fig). These results suggest that the absence of Ybx1 reduces the number of progenitor cells in the developing neocortex, which is not mediated by apoptosis.

Next, we assessed the impact of Ybx1 loss on cortical neurogenesis. In P0 mice, neurons differentiated from RGCs exist in six different cortical layers, with early-born neurons occupying the deep layers of the neocortex and later-born neurons migrating to the superficial layers, forming a normative six-layered structure in an “inside-out” manner [29]. To investigate the effect of Ybx1 loss-induced depletion of neural stem cells on the generation of neurons in different cortical layers, we conducted immunostaining of Tbr1 and Ctip2 at P0, which label cortical layers VI and V, respectively. This allowed us to distinguish layers I–IV, V, VI, and white matter (WM) (Fig 2G). The numbers of Ctip2⁺ layer V neurons and Tbr1⁺ layer VI neurons are decreased in P0 Ybx1 cKO mice compared with the control mice (Fig 2H and 2I). Although the overall laminar structure is similar between Ybx1 cKO and control mice (Fig 2G), each layer in Ybx1 cKO mice exhibits varying degrees of thickness reduction (Fig 2J–2N). These results indicate that the loss of Ybx1 impairs cortical neurogenesis.

Ybx1 is required for maintenance of the cortical progenitor pool from E13.5

To determine at which embryonic stage the reduction in the NSC pool occurs in the cKO mice, we first checked the knockout efficiency of Ybx1 in E11.5, E13.5, E15.5, and E17.5 cortex. Immunostaining showed that the Ybx1 signal in the cKO cortex was nearly absent starting from E13.5 onwards (S3A Fig). Ybx1 signals in the heterozygous mice were comparable to those in control mice at all stages studied. Pax6 immunostaining on cortical sections of E11.5 embryos showed no significant changes in cortical thickness and progenitor cell pools in Ybx1 cKO compared to control littermates (S3B–S3D Fig), which is consistent with the fact Ybx1 was not ablated yet at E11.5 (S3A and S3B Fig). By E13.5, the numbers of Pax6⁺ and Tbr2⁺ cells in the Ybx1 cKO cortex diminished by 11.6% and 10.1%, respectively, in comparison to control embryos (Fig 3A–3C). Meanwhile, we observed a slight decrease in the thickness of the VZ, SVZ, and whole cortex at E13.5 (Fig 3D–3G). At E15.5, the reduction in the numbers of Pax6⁺ and Tbr2⁺ cells in Ybx1 cKO cortex became more pronounced, counting as 18.4% and 13.1%, respectively (Fig 3H–3J). Significant decreases in the thickness of each layer were also observed in the cKO embryos (Fig 3K–3P). Similarly pronounced reductions were observed in E17.5 Ybx1 cKO brains (S3E–S3N Fig).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Ybx1 is required for maintenance of the cortical progenitor pool from E13.5.

(A) E13.5 coronal brain sections were immunostained with antibodies against the radial glial cell marker Pax6 and the intermediate progenitor marker Tbr2. Representative cortical regions were shown. White dotted lines mark the boundaries. Scale bar: 50 μm. (B–F) Quantification of Pax6⁺ (B) and Tbr2⁺ (C) cell numbers, and cortical layer thicknesses (D to F) shown in (A). All statistical data are presented as box and whisker plots. For Pax6, Ctrl (n = 21 confocal fields) vs. cKO (n = 22 confocal fields), **p = 0.0018; Het (n = 18 confocal fields) vs. cKO, *p = 0.019. For Tbr2, Ctrl (n = 21 confocal fields) vs. cKO (n = 22 confocal fields), *p = 0.025; Het (n = 18 confocal fields) vs. cKO, *p = 0.014. For whole cortical thickness, Ctrl (n = 21 confocal fields) vs. cKO (n = 22 confocal fields), **p = 0.0018; Het (n = 18 confocal fields) vs. cKO, *p = 0.016. For VZ thickness, Ctrl (n = 21 confocal fields) vs. cKO (n = 22 confocal fields), **p = 0.0023; Het (n = 18 confocal fields) vs. cKO, *p = 0.026. For SVZ thickness, Ctrl (n = 21 confocal fields) vs. cKO (n = 22 confocal fields), **p = 0.0055; Het (n = 18 confocal fields) vs. cKO, *p = 0.047. (G) A stacked bar chart summarizing the data in (D–F) shows the distribution of each layer in E13.5 cortex. (H) E15.5 coronal brain sections were immunostained with antibodies against the radial glial cell marker Pax6 and the intermediate progenitor marker Tbr2. Representative cortical regions were shown. White dotted lines mark the boundaries. Scale bar: 100 μm. (I–O) Quantification of Pax6⁺ (I) and Tbr2⁺ (J) cell numbers, and cortical layer thicknesses (K–O) shown in (H). All statistical data are presented as box and whisker plots. For Pax6, Ctrl (n = 24 confocal fields) vs. cKO (n = 22 confocal fields), ***p = 2.93E−04; Het (n = 21 confocal fields) vs. cKO, **p = 0.0071. For Tbr2, Ctrl (n = 24 confocal fields) vs. cKO (n = 22 confocal fields), ***p = 8.58E−04; Het (n = 21 confocal fields) vs. cKO, **p = 0.0042. For VZ thickness, Ctrl (n = 24 confocal fields) vs. cKO (n = 22 confocal fields), ****p = 5.60E−05; Het (n = 21 confocal fields) vs. cKO, ***p = 2.72E−04. For SVZ thickness, Ctrl (n = 24 confocal fields) vs. cKO (n = 22 confocal fields), ***p = 2.75E−04; Het (n = 21 confocal fields) vs. cKO, ***p = 8.94E−04. For CP thickness, Ctrl (n = 24 confocal fields) vs. cKO (n = 22 confocal fields), **p = 0.0062; Het (n = 21 confocal fields) vs. cKO, *p = 0.034. For IZ thickness, Ctrl (n = 24 confocal fields) vs. cKO (n = 22 confocal fields), ***p = 1.99E−04; Het (n = 24 confocal fields) vs. cKO, ***p = 5.31E−04. For whole cortical thickness, Ctrl (n = 24 confocal fields) vs. cKO (n = 22 confocal fields), ****p = 1.73E−05; Het (n = 21 confocal fields) vs. cKO, ***p = 1.68E−04. (P) A stacked bar chart summarizing the data in (K–O) shows the distribution of each layer in E15.5 cortex. At least 3 embryos were analyzed for each genotype. All analyses were performed by one-way ANOVA followed by Tukey’s multiple comparison test. ns, not significant. The data underlying all the graphs shown in the figure are included in S1 Data.

https://doi.org/10.1371/journal.pbio.3003175.g003

These findings suggest that loss of the Ybx1 leads to a reduction in the progenitor cell pool in the mouse cortex from E13.5, followed by decreases in thickness of cortical layers composed of late-born neurons.

Ybx1 regulates proliferation and differentiation of cortical progenitor cells

To investigate the mechanism underlying the observed reduction in cortical NSCs due to Ybx1 loss, we performed in vivo BrdU pulse labeling experiments in mice. Pregnant mice at E19.5 received BrdU injections, and 24 h post-injection, coronal brain sections from P0 pups were prepared for BrdU and Ki67 immunostaining analysis. BrdU incorporation can be used to monitor the proliferation. We observed a significant decrease in the cumulative number of BrdU-labeled cells in the cKO cortex compared to control cortex at P0 (Fig 4A and 4B), suggesting that proliferation may be inhibited in Ybx1 cKO embryos or the neural progenitor cell pool of Ybx1 cKO embryos is simply reduced at P0. Ki67, a cell cycle marker closely associated with cell proliferation, allows for the assessment of cell cycle exit rates, as indicated by the proportion of Ki67⁻BrdU⁺ in total BrdU⁺ cells. Immunofluorescence results showed that, relative to the littermate controls, the proportion of cells exiting the cell cycle (Ki67⁻BrdU⁺/BrdU⁺) was significantly reduced in Ybx1 cKO embryos, indicating that the ablation of Ybx1 also inhibited the differentiation of cortical NSCs (Fig 4A and 4C).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Ybx1 regulates proliferation and differentiation of cortical progenitor cells.

(A) Coronal brain sections at P0 were stained with antibodies recognizing BrdU and Ki67. Pregnant mothers received a BrdU pulse 24 h before pup dissection at P0. Representative cortical regions were shown. Scale bar: 100 μm. (B and C) Quantification of BrdU⁺ cell numbers (B) and the percentage of cells exiting the cell cycle (C) at P0 shown in (A). All statistical data are presented as box and whisker plots. For BrdU⁺ cell numbers, Ctrl (n = 27 confocal fields) vs. cKO (n = 24 confocal fields), ***p = 2.11E−04; Het (n = 23 confocal fields) vs. cKO, ***p = 4.24E−04. For Ki67⁻BrdU⁺/BrdU⁺, Ctrl (n = 27 confocal fields) vs. cKO (n = 24 confocal fields), **p = 0.0015; Het (n = 23 confocal fields) vs. cKO, **p = 0.0051. (D) Coronal brain sections at E13.5 were stained with antibodies recognizing BrdU and Ki67. Pregnant mothers received a BrdU pulse 24 h before embryo collection. Representative cortical regions were shown. Scale bar, 50 μm. (E and F) Quantification of BrdU⁺ cell numbers (E) and the percentage of cells exiting the cell cycle (F) at E13.5 shown in (D). All statistical data are presented as box and whisker plots. For BrdU⁺ cell numbers, Ctrl (n = 22 confocal fields) vs. cKO (n = 25 confocal fields), **p = 0.0052; Het (n = 23 confocal fields) vs. cKO, *p = 0.012. For Ki67⁻BrdU⁺/BrdU⁺, Ctrl (n = 22 confocal fields) vs. cKO (n = 25 confocal fields), **p = 0.0028; Het (n = 23 confocal fields) vs. cKO, *p = 0.027. (G) Representative images of primary and secondary neurospheres formed by NSCs isolated at E13.5. Scale bar, 50 μm. (H–K) Quantification of the sizes (H, J) and numbers (I, K) of the primary (H, I) and secondary (J, K) neurospheres show in (G). All statistical data are presented as box and whisker plots. In H, Ctrl (n = 42 neurospheres) vs. cKO (n = 38 neurospheres), ****p = 6.65E−05; Het (n = 46 neurospheres) vs. cKO, ****p = 5.22E−05. In J, Ctrl (n = 42 neurospheres) vs. cKO (n = 38 neurospheres), ****p = 2.76E−05; Het (n = 46 neurospheres) vs. cKO, ****p = 2.26E−05. (L) Schematic drawings of the IUE experiment. Pregnant mice were injected at E14.5 with plasmids. Subsequently, a single EdU pulse was administrated at E15.5, and embryos were dissected at E16.5 for analysis. (M) Confirmation of Ybx1 knockdown in the cortex after IUE of shYbx1. Immunostaining for GFP and Ybx1 on coronal sections of E16.5 cortex. Representative cortical regions were shown. Scale bar, 50 μm. (N) Immunostaining for GFP, EdU, and Ki67 on coronal sections of E16.5 mouse cortex after knockdown of Ybx1 using IUE. Representative cortical regions were shown. Scale bar, 50 μm. (O and P) Quantification of percentage of EdU⁺GFP⁺/GFP⁺ (O) and Ki67^-EdU⁺/EdU⁺GFP⁺ (P) shown in (N). Data are presented as box and whisker plots: in O, shCtrl (n = 22 confocal fields) vs. shYbx1 (n = 19 confocal fields), ***p = 1.38E−04; in P, shCtrl (n = 22 confocal fields) vs. shYbx1 (n = 19 confocal fields), ***p = 8.30E−04. At least 3 pups or embryos were analyzed for each genotype or condition. Analyses were performed by one-way ANOVA followed by Tukey’s multiple comparison test (B, C; E, F; H–K), or by unpaired Student t test (O, P). ns, not significant. The data underlying all the graphs shown in the figure are included in S1 Data.

https://doi.org/10.1371/journal.pbio.3003175.g004

We further backtracked the earlier embryonic stages at E13.5, E15.5, and E17.5 using BrdU/Ki67 staining analysis. From E13.5 onwards, the total number of BrdU-labeled cells and the proportion of cells exiting the cell cycle in the Ybx1 cKO cortex were reduced, in comparison to control cortex (Fig 4D–4F). Similar experimental results were obtained in E15.5 and E17.5 cKO cortex (S4A–S4F Fig). Together, these results suggest that Ybx1 is required for normal proliferation and differentiation of cortical progenitor cells.

To further understand the impact of Ybx1 on the self-renewal of neural stem cells, we performed in vitro neurosphere assay using dissociated cortical NSCs from E13.5 embryos. While neural stem cells from Ybx1 cKO, heterozygous, and control littermates at E13.5 generated comparable numbers of neurospheres, the size of neurospheres from cKO was notably smaller than those from controls (Fig 4G–4I). Subsequently, we dissociated and passaged the obtained neurospheres, and consistently, the size of neurospheres in Ybx1 cKO was significantly reduced, while the number remained unchanged compared to the control group (Fig 4G, 4J, and 4K). This result aligns with the observed decrease in the mitotic capacity of progenitor cells through in vivo BrdU labeling of Ybx1 cKO embryos.

Furthermore, we explored another Ybx1 loss-of-function approach by knocking down Ybx1 using in utero electroporation (IUE) in the mouse embryonic cortex. We performed IUE using a plasmid driven by the CAG promoter expressing shYbx1 with GFP reporter at E14.5. Then, we conducted EdU injection and dissection at 24 and 48 h post-electroporation (i.e., E15.5 and E16.5) (Fig 4L). Ybx1 immunofluorescence confirmed efficient knockdown of Ybx1 protein in GFP⁺ cells compared to the control group (Figs 4M and S4G). We then performed immunofluorescence analysis of EdU and Ki67. The results showed a significant reduction in the percentage of EdU⁺GFP⁺/GFP⁺ compared to the control group, indicating decreased proliferation of neural progenitor cells due to Ybx1 knockdown (Fig 4N and 4O). Additionally, we observed a reduced percentage of GFP⁺ cells that were EdU⁺ but Ki67^-, suggesting fewer cells had exited the cell cycle (Fig 4N and 4P). Furthermore, the percentages of Pax6⁺GFP⁺ representing radial glial cells (RGCs) and Tbr2⁺GFP⁺ representing intermediate progenitor cells (IPCs) in all GFP⁺ cells were also reduced in the shYbx1 cortex (S4H–S4J Fig). These knockdown results were consistent with cKO analysis, further confirming that Ybx1 loss-of-function in the mouse cortex impairs the proliferation and differentiation of neural progenitor cells.

Taken together, these results indicate that Ybx1 plays an important role in regulating the proliferation and differentiation of cortical progenitor cells, and its loss impairs their self-renewal capacity and neurogenesis.

Ybx1 controls the stability of its m⁵C-modified target transcripts in neural progenitor cells

To elucidate the mechanisms by which Ybx1 regulates NSC proliferation and differentiation, we initially tried to ascertain the subcellular localization of Ybx1 in neural stem cells. Immunostaining of Ybx1 in cultured E13.5 embryonic cortical progenitor cells, marked by Pax6 and Ki67, revealed that Ybx1 protein was predominantly located in the cytoplasm, suggesting a major cytoplasmic function for Ybx1 (S5A and S5B Fig). So, we focused on identifying and elucidating its target mRNAs which Ybx1 regulates in post-transcriptional level.

We first performed transcriptome analysis of cortical tissues from both Ybx1 cKO and control littermates at E13.5 (E13.5-cKO) and E18.5 (E18.5-cKO). Additionally, in order to rule out the possible compensatory effects brought by ablation of Ybx1 in embryos, we also conducted transcriptome analysis of E13.5 embryonic neural stem cells cultured in vitro after Ybx1 knockdown (KD) using lentiviral shRNA (E13.5-KD). RNA sequencing (RNA-seq) revealed 161 upregulated genes and 215 downregulated genes in E13.5-cKO compared to the control group (S1 Table), 645 upregulated genes and 427 downregulated genes in E18.5-cKO (S2 Table), and 1886 upregulated genes and 1949 downregulated genes in E13.5-KD (S3 Table). Gene Ontology (GO) analysis of the differentially expressed genes (DEGs) across the three cKO/KD-RNAseq experiments showed significant enrichment in biological processes such as regulation of nervous system development, neural precursor cell proliferation, cerebral cortex neuron differentiation, and forebrain development (Fig 5A–5C). This aligns with the observed impact of Ybx1 ablation on cortical development. Subsequently, we performed RNA immunoprecipitation sequencing (RIP-seq) using Ybx1 antibody to fish target transcripts interacting with Ybx1 in the mouse cortex at E13.5, identifying a total of 2,567 transcripts (S4 Table). Functional annotation of Ybx1 targets revealed enrichment in cellular components such as spindle, mitotic spindle, and cyclin-dependent protein kinase holoenzyme complex; molecular functions including kinase regulator activity and mRNA binding; and biological process such as regulation of neurogenesis, neural precursor cell proliferation, and mitotic cell cycle phase transition (Fig 5D).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Ybx1 controls the stability of its m⁵C-modified target transcripts in neural progenitor cells.

(A-C) Gene Ontology (GO) analysis of transcripts with altered expression levels in E13.5 Ybx1 cKO cortex (A), E18.5 Ybx1 cKO cortex (B), and E13.5 Ybx1 KD cortical progenitors (C). GO terms in biological processes were shown. (D) GO analysis of Ybx1 target transcripts identified by anti-Ybx1 RIP-seq in the E14.5 mouse cortex. (E) Venn diagram showing the overlap of target mRNAs identified by three cKO or KD RNA-seq and anti-Ybx1 RIP-seq. GO terms in biological processes were shown. (F) Validation of m⁵C modification on Ybx1 target mRNAs by anti-m⁵C immunoprecipitation combined with RT-qPCR. In vitro transcribed GFP mRNA which has no m⁵C modification was used as a negative control. Data represent mean ± SD (n = 3 replicates): ***p = 5.99E−04 for Ccnd2; ***p = 1.37E−04 for eEF1g; **p = 0.0045 for Rpsa; ***p = 8.89E−04 for Rps5; ***p = 8.06E−04 for Uhmk1; not significant (ns) for GFP; by unpaired Student t test. (G) Ybx1 target transcripts exhibit accelerated degradation in the Ybx1 cKO cortex. Radial glial cells (RGCs) dissected from E14.5 Ybx1 cKO and control embryos were cultured, treated with actinomycin D (ActD), and collected at different time points. Ybx1 target mRNA levels were measured by RT-qPCR while β-actin mRNA was used a non-target control. Data represent mean ± SEM (n = 3 replicates): for Ccnd2, **p = 0.0017 (2.5 h), ***p = 7.50E−04 (5 h), ***p = 2.32E−04 (7.5 h), **p = 0.0018 (10 h); for eEF1g, **p = 0.0014 (2.5 h), ***p = 3.98E−04 (5 h), ***p = 3.10E−04 (7.5 h), **p = 0.0022 (10 h); for Rpsa, **p = 0.0096 (2.5 h), **p = 0.0012 (5 h), ***p = 4.09E−04 (7.5 h), ***p = 5.08E−04 (10 h); for Rps5, ***p = 5.67E−04 (2.5 h), ***p = 1.33E−04 (5 h), ***p = 1.65E−04 (7.5 h), ***p = 6.41E−04 (10 h); for Uhmk1, **p = 0.0013 (2.5 h), **p = 0.0025 (5 h), **p = 0.0011 (7.5 h), ***p = 7.91E−04 (10 h); for β-actin, ns, not significant; by unpaired Student t test. The half-lives for each mRNA in Ctrl and cKO were calculated (? indicates that the half-lives cannot be calculated). The data underlying all the graphs shown in the figure are included in S1 Data.

https://doi.org/10.1371/journal.pbio.3003175.g005

Subsequently, we juxtaposed the anti-Ybx1 RIP-seq data with the three cKO/KD-RNAseq datasets to pinpoint mRNA targets which are directly bound and regulated by Ybx1 in cortical development. By integrating RIP-seq results with DEGs from E13.5-cKO, E18.5-cKO, and E13.5-KD RNA-seq, we identified 49, 223, and 536 candidate targets, respectively. GO analysis of these candidate targets showed significant enrichment in biological processes related to neurogenesis, mitosis, and cell cycle regulation (S5C–S5E Fig). These GO terms are also in line with the phenotype of impaired proliferation and differentiation of NSCs after ablation of Ybx1. By overlapping the DEGs in E13.5-cKO, E13.5-KD, and E18.5-cKO RNAseq with anti-Ybx1 RIP-seq, we obtained 9 target mRNAs (including Ybx1 itself, a pseudogene of Ybx1, and two pseudogenes of Rpsa) that were downregulated consistently in different stages and conditions for Ybx1 loss of function (Fig 5E). We focused on the five target mRNAs (Ccnd2, eEF1g, Rpsa, Rps5, and Uhmk1) in the following analysis.

We first verified downregulation of these targets in Ybx1 loss of function. When Ybx1 was knocked down in neural stem cells cultured in vitro, the mRNA levels of these five targets were significantly reduced (S6A Fig). Further in vivo analysis of the cortices at E13.5, E15.5, and E17.5 was conducted by RT-qPCR on RNAs extracted from these tissues. The results confirmed decreases in the mRNA levels of Ccnd2, eEF1g, Rpsa, Rps5, and Uhmk1 in Ybx1 cKO cortices compared to the controls (S6B Fig). These data verified that Ybx1 regulates the target mRNA levels.

The m⁵C modification of these mRNAs was validated by anti-m⁵C pulldown experiment (Fig 5F). Next, we further investigated whether Ybx1 regulation of its target mRNAs depends on m⁵C modification. Previous studies have shown that the only two known m⁵C RNA methyltransferases, Nsun2 and Nsun6, are closely related to mRNA methylation [30–34]. Knockout experiments targeting Nsun2 and/or Nsun6 have demonstrated that Nsun2 and Nsun6 act on the m⁵C sites of different subsets of mRNA and together account for nearly all m⁵C modifications in mRNA [33]. We first knocked down Nsun2 and Nsun6 in cultured neural stem cells (S6C Fig) to eliminate m⁵C modification. In these Nsun2/6- and thus m⁵C-deficient cells, we continued to perform Ybx1 knockdown. The results demonstrated that, in the absence of m⁵C modification, the presence or absence of Ybx1 could not affect the target transcript levels anymore (S6D Fig). These data suggest that Ybx1 regulates the target mRNA levels in an m⁵C-dependent manner. As Ybx1 can promote the stability of its m⁵C-modified targets, we continued to test the stability of those target mRNAs in Ybx1 loss of function conditions. RNA stability assays showed an accelerated degradation of these mRNAs in NSCs from Ybx1 cKO compared with control cells (Fig 5G). Taken together, all these results suggest that Ybx1 normally can enhance the stability of its m⁵C-modified target mRNAs.

Ybx1 target mRNAs regulate the proliferation and differentiation of cortical progenitor cells

Next, we investigated whether the target mRNAs Ccnd2, eEF1g, Rpsa, Rps5, and Uhmk1 could mediate the effects of Ybx1 on proliferation and differentiation of cortical progenitor cells. We generated siRNAs targeting these transcripts and validated their knockdown efficiency (S7A–S7E Fig). Following this, we assessed the influence of target mRNA knockdown on neurosphere formation capacity using these siRNAs. The results showed that knockdown of Ccnd2, eEF1g, Rpsa, Rps5, or Uhmk1 resulted in smaller neurospheres compared to the control siRNA (Fig 6A and 6B), while the numbers remained unchanged (Fig 6A and 6C). Neurosphere assay results suggest that knockdown of Ybx1 targets can inhibit the self-renewal of neural stem cells, recapitulating the neurosphere phenotypes of Ybx1 cKO.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Knockdown of Ybx1 target mRNAs impairs the proliferation and differentiation of cortical progenitor cells.

(A) Representative images of neurospheres formed after siRNA-mediated knockdown of Ccnd2, eEF1g, Rpsa, Rps5, or Uhmk1 in cultured E13.5 neural stem cells (NSCs). Scale bar, 50 μm. (B and C) Quantification of the sizes (B) and numbers (C) of the neurospheres after siRNA-mediated knockdown of Ybx1 target mRNAs show in (A). At least 3 replicates were performed. All statistical data are presented as box and whisker plots: in B, siCcnd2 (n = 55 neurospheres) vs. siCtrl (n = 51 neurospheres), ****p = 5.87E−06; sieEF1g (n = 52 neurospheres) vs. siCtrl, ****p = 5.46E−07; siRpsa (n = 57 neurospheres) vs. siCtrl, ****p = 1.57E−07; siRps5 (n = 49 neurospheres) vs. siCtrl, ****p = 6.12E−07; siUhmk1 (n = 45 neurospheres) vs. siCtrl, ****p = 9.87E−08; by one-way ANOVA followed by Tukey’s multiple comparison test; ns, not significant. (D) Immunostaining for GFP, EdU, and Ki67 on coronal sections of E16.5 mouse cortex after knockdown of Ybx1 targets in the cortex using IUE of a cocktail shRNA against all targets. Scale bar, 50 μm. (E and F) Quantification of percentage of EdU⁺GFP⁺/GFP⁺ (E) and Ki67^-EdU⁺/EdU⁺GFP⁺ (F) shown in (D). At least 3 embryos were analyzed for each condition. Data are presented as box and whisker plots: in E, shCtrl (n = 23 confocal fields) vs. shCocktail (n = 27 confocal fields), ***p = 2.10E−04; in F, shCtrl (n = 23 confocal fields) vs. shCocktail (n = 27 confocal fields), ***p = 6.86E−04; by unpaired Student t test (E, F). The data underlying all the graphs shown in the figure are included in S1 Data.

https://doi.org/10.1371/journal.pbio.3003175.g006

Furthermore, we explored in vivo functions of Ybx1 target transcripts using in utero electroporation (IUE) experiments in the embryonic cortex. Individual knockdown of Ccnd2, eEF1g, Rpsa, Rps5, or Uhmk1 using specific shRNAs impaired the proliferation and differentiation of neural progenitor cells (except that knockdown of Rps5 caused a slight but insignificant decrease in differentiation) (S7F and S7G Fig). Interestingly, knocking down all these target mRNAs together with a shRNA cocktail caused more robust reduction in the proliferation and differentiation capacity of neural progenitor cells compared to each individual shRNA (Fig 6D–6F). Together, these results suggest that knockdown of Ybx1 targets in vivo can significantly affect the proliferation and differentiation capabilities of neural stem cells.

Overexpression of Ybx1 targets rescues the cortical development defects caused by Ybx1 ablation

We continued to test whether the target mRNAs Ccnd2, eEF1g, Rpsa, Rps5, and Uhmk1 could mediate the effects of Ybx1 on proliferation and differentiation of cortical progenitor cells. Since knockdown of these targets could recapitulate the phenotypes of Ybx1 cKO, we proceeded to explore whether overexpression of Ybx1 targets could rescue the defects caused by Ybx1 cKO. We constructed two plasmids overexpressing the Ybx1 targets Ccnd2 and eEF1g, respectively. The results showed that overexpression of either Ccnd2 or eEF1g by IUE could only partially rescue the impaired proliferation and differentiation of cortical progenitor cells caused by Ybx1 cKO (S8A–S8D Fig). We continued to simultaneously overexpress the five targets Ccnd2, eEF1g, Rpsa, Rps5, and Uhmk1 all together (OE cocktail) in the mouse cortex using IUE. We observed that in the cortex of Ybx1 cKO embryos overexpressing the cocktail of targets, the proportion of Pax6⁺GFP⁺ and Tbr2⁺GFP⁺ cells among total GFP⁺ cells were equivalent or close to the values in control embryos (Fig 7A–7C). We further conducted EdU pulse labeling experiments and found that the percentage of EdU⁺GFP⁺ cells among all EdU⁺ cells and the proportion of EdU⁺GFP⁺ cells that are Ki67^- were also successfully rescued in the Ybx1 cKO cortex overexpressing the cocktail of targets by IUE (Fig 7D–7F). Additionally, we performed overexpression of the cocktail of targets in the wild type (WT) background, which promoted the proliferation and differentiation of cortical neural stem cells (Fig 7D–7F). Compared with overexpression of the five target genes in the Ybx1 cKO cortex, overexpression in WT leads to a higher proportion of EdU⁺GFP⁺ cells among EdU⁺ cells, as well as a higher proportion of Ki67⁻ cells among EdU⁺GFP⁺ cells. We believe this difference might be due to the fact that Ybx1 knockout not only alters the expression levels of these five core target genes but also induces broader transcriptional changes. The downregulation of other genes might counteract the overexpression effects of these five core target genes.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. Overexpression of Ybx1 targets rescues the cortical development defects caused by Ybx1 ablation.

(A) Overexpression of Ybx1 targets (OE-Cocktail) in Ybx1 cKO mouse cortex using IUE, followed by immunostaining for GFP, Pax6, and Tbr2 on coronal sections at E16.5. Representative cortical regions were shown. Scale bar, 50 μm. (B and C) Quantification of percentage of Pax6⁺GFP⁺/GFP⁺ (B) and Tbr2⁺GFP⁺/GFP⁺ (C) shown in (A). Data are presented as box and whisker plots. In B, “Ctrl + OE-Ctrl” (n = 20 confocal fields) vs. “cKO + OE-Ctrl” (n = 18 confocal fields), ****p = 4.26E−05; “cKO + OE-Ctrl” vs. “cKO + OE-Cocktail” (n = 21 confocal fields), **p = 0.0040; “Ctrl + OE-Ctrl” vs. “cKO + OE-Cocktail”, p = 0.10. In C, “Ctrl + OE-Ctrl” (n = 20 confocal fields) vs. “cKO + OE-Ctrl” (n = 18 confocal fields), ***p = 5.69E-04; “cKO + OE-Ctrl” vs. “cKO + OE-Cocktail” (n = 21 confocal fields), *p = 0.043; “Ctrl + OE-Ctrl” vs. “cKO + OE-Cocktail”, *p = 0.015. (D) Overexpression of Ybx1 targets (OE-Cocktail) in control and Ybx1 cKO mouse cortex using IUE, followed by immunostaining for GFP, EdU, and Ki67 on coronal sections at E16.5. Representative cortical regions were shown. Scale bar, 50 μm. (E and F) Quantification of percentage of EdU⁺GFP⁺/GFP⁺ (E) and Ki67^-EdU⁺/EdU⁺GFP⁺ (F) shown in (D). Data are presented as box and whisker plots. In E, “Ctrl + OE-Ctrl” (n = 27 confocal fields) vs. “Ctrl + OE-Cocktail” (n = 25 confocal fields), **p = 0.0099; “Ctrl + OE-Ctrl” vs. “cKO + OE-Ctrl” (n = 24 confocal fields), ****p = 3.08E−10; “Ctrl + OE-Ctrl” vs. “cKO + OE-Cocktail” (n = 26 confocal fields), p = 0.19. “Ctrl + OE-Cocktail” vs. “cKO + OE-Cocktail”, ****p = 8.25E−6. “cKO + OE-Ctrl” vs. “cKO + OE-Cocktail”, ****p = 3.08E−10. In F, “Ctrl + OE-Ctrl” (n = 26 confocal fields) vs. “Ctrl + OE-Cocktail” (n = 28 confocal fields), *p = 0.015; “Ctrl + OE-Ctrl” (n = 26 confocal fields) vs. “cKO + OE-Ctrl” (n = 24 confocal fields), ****p = 3.08E−10; “Ctrl + OE-Ctrl” vs. “cKO + OE-Cocktail” (n = 24 confocal fields), p = 0.51. “Ctrl + OE-Cocktail” vs. “cKO + OE-Cocktail”, ***p = 0.00015. “cKO + OE-Ctrl” vs. “cKO +OE-Cocktail”, ****p = 3.08E−10. At least 3 embryos were analyzed for each genotype or condition. Analyses were performed by one-way ANOVA followed by Tukey’s multiple comparison test. ns, not significant. The data underlying all the graphs shown in the figure are included in S1 Data.

https://doi.org/10.1371/journal.pbio.3003175.g007

Together, these results indicate that overexpression of Ybx1 target genes can rescue the defective phenotypes of Ybx1 cKO cortex in cell proliferation and cell cycle exit.

Ybx1 regulates the G1 to S phase transition of cortical progenitor cell cycle

We further investigated the functions of the Ybx1 targets Ccnd2, eEF1g, Rpsa, Rps5, and Uhmk1. Previous studies have shown that Ccnd2 (cyclin D2), a D-type cell cycle protein, predominantly binds and activates CDK4 and CDK6 during the G1 phase. This activation leads to the phosphorylation and inactivation of the Rb protein family, releasing and activating E2F to initiate DNA synthesis, thereby enhancing the G1 to S phase transition and promoting cell division [35–38]. There is substantial evidence indicating that some ribosomal proteins, in addition to serving as structural components of the ribosome subunit, also participate in the regulation of cell proliferation and differentiation. Rps5, a ribosomal protein, has been shown to regulate the expression levels of CDK4/6, thereby controlling the G1 to S transition of the cell cycle [39,40]. Rpsa, initially identified as a 67 kD laminin-binding protein, is evolutionarily conserved and later identified as a 37 kD member of the RPS family of ribosomal components. Knockdown of RPSA promotes an increase in p21 expression levels, reinforcing inhibition of CDK4/6 and resulting in G1 phase cell cycle arrest in vitro [41–43]. Uhmk1, also known as KIS, consists of a kinase domain and a putative RNA-binding domain (RBD). Uhmk1 can phosphorylate the cyclin-dependent kinase inhibitor p27, leading to nuclear export and proteasomal degradation of p27 and relieving its inhibitory effect on CDK4/6 [44–47]. Thus, interestingly, four (Ccnd2, Rpsa, Rps5, and Uhmk1) out of the five Ybx1 target mRNAs are involved in the G1 to S phase transition in the cell cycle (Fig 8A), consistent with the GO enrichment of biological processes related to the regulation of cell cycle phase transition.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 8. Ybx1 regulates the G1–S phase transition of cortical progenitor cell cycle.

(A) Schematic drawings showing Ybx1 target mRNAs-encoding proteins are involved in regulating normal cell cycle progression. (B and C) Cell cycle flow cytometry analysis determining the percentage of cells in G0/G1, S, and G2/M phases in cortical progenitor cells from E13.5 (B) and E15.5 (C) Ybx1 cKO and littermate control embryos. Data are presented as mean ± SEM: n = 3 replicates; in B, *p = 0.031 for G0/G1 phase, *p = 0.044 for S phase, **p = 0.0012 for G2/M phase; in C, ***p = 4.33E−04 for G0/G1 phase, ***p = 9.75E−04 for S phase, ***p = 3.40E−04 for G2/M phase; by unpaired Student t test. The data underlying all the graphs shown in the figure are included in S1 Data.

https://doi.org/10.1371/journal.pbio.3003175.g008

To determine whether Ybx1 regulates the cell cycle process through modulating its targets encoding these cell cycle-related proteins, we performed cell cycle analysis based on flow cytometry using propidium iodide (PI) DNA staining. FACS detection results showed that, compared to the control cortex, the proportion of G1 phase cells increased by 6.8% in E13.5 Ybx1 cKO embryos (Figs 8B and S9A), and by 15.3% at E15.5 (Figs 8C and S9B), indicating that Ybx1 deletion causes cell cycle arrest at the G1 phase. Additionally, loss of Ybx1 in E13.5 Ybx1 cKO embryos resulted in a 2.0% reduction in the S phase cells and a 5.2% reduction in the G2/M phase cells (Figs 8B and S9A). At E15.5, these values decreased by 4.1% and 10.8%, respectively (Figs 8C and S9B). Additionally, we used the dual-labeling method, which allows for more precise measurement of the duration of different cell cycle phases (S9C Fig). We found that in the E14.5 mouse cortex, the total cell cycle duration (T_c) and the S phase duration (T_s) of control neural stem cells were 16.23 h and 4.29 h, respectively (S9D–S9F Fig). In contrast, Ybx1 cKO neural stem cells exhibited the total cell cycle duration and the S phase duration of 20.80 h and 4.63 h, respectively (S9D–S9F Fig). This indicates that, compared to control mice, Ybx1-deficient neural stem cells spend significantly more time in the cell cycle.

Collectively, these findings suggest Ybx1-deficient cortical progenitor cells exhibit a G1-to-S phase blockade in vivo, delayed cell cycle exit, and an extended cell cycle duration, ultimately resulting in the inhibition of cell cycle progression.

Taken together, Ybx1 facilitates the G1–S transition by stabilizing the expression of its target mRNAs encoding the pivotal regulatory factors, which is essential for the normal progression of the cell cycle in neural stem cells.

Ethics statement

All experiments involving mice were conducted following the approved animal protocol (SUSTech-JY202202059) by the Laboratory Animal Welfare and Ethics Committee of the Southern University of Science and Technology.

Animals

For the generation of Ybx1 conditional knockout (cKO) mice, targeted deletion was performed on exons 3–8 of the mouse Ybx1 gene, which encompasses a 745 bp coding sequence. Deletion of this region would result in the disruption of protein function. Briefly, sgRNAs were transcribed in vitro, and a donor vector was constructed. Cas9, sgRNAs, and the donor were microinjected into fertilized eggs of C57BL/6J mice. Transplanted fertilized eggs were used to obtain F₀ mice, confirmed positive through PCR and sequencing. Positive F0 mice were bred with C57BL/6J mice to obtain stable F1 mouse models. Ybx1^+/fl mice and corresponding Cre mice were used to produce Ybx1 cKO and littermate controls. Genotyping primers were as follows: the first Ybx1-loxP site, 5′-CCAATAGTGACGAAGGTGGAGTGG-3′ and 5′-AGCTTCCACACTATTGCCTGGG-3′; the second Ybx1-loxP site, 5′-GACTAATGGCTGCTAGTTGTCCG-3′ and 5′-AGGGTTTCTCTGTGTAGCTCTGGC-3′. The Nestin-Cre mice used in this study were obtained from Jackson Laboratory with stock number 003771. For timed pregnancy, embryonic day E0.5 was considered when the mating plug was observed at noon.

In situ hybridization

In situ hybridization (ISH) with DIG-labeled RNA probes was performed on mouse tissue sections following a previously reported protocol [63]. The PCR primers used to clone the template for the mouse Ybx1 ISH probe were 5′-GGTCCTCCACGCAATTACCA-3′ and 5′-TGTTGGATGACCAAACCGGA-3′. RNA probes were transcribed in vitro using the DIG RNA labeling Kit (SP6/T7) (Roche). Tissue section preparation for ISH followed a similar procedure to immunofluorescence, but the entire process used RNase-free reagents. Briefly, the procedure involved fixing tissue sections with 4% paraformaldehyde (PFA) followed by treatment with Proteinase K. The sections were then acetylated with acetic anhydride. Subsequently, the sections were pre-hybridized at 65 °C for 2 h in hybridization buffer, followed by overnight incubation at 65 °C with fresh hybridization buffer containing the probes. The next day, the sections were washed with 2× SSC and treated with 10 μg/mL RNase A at 37 °C for 8–10 min. After sequential washes with 2× SSC, 0.2× SSC, and PBT buffers, the sections were blocked with 10% heat-inactivated sheep serum at room temperature for 1 h. The sections were then incubated overnight at 4 °C with anti-digoxigenin diluted in 10% sheep serum. On the third day, after washing with PBT and alkaline phosphatase buffer, color reaction was performed using NBT/BCIP diluted in alkaline phosphatase buffer in the dark. The color reaction was stopped with PBS when the target gene signal appeared, and the sections were mounted for observation. Both anti-digoxigenin-AP and NBT/BCIP stock solutions were obtained from Roche. In situ hybridization images were collected using the TissueFAXS microscope (TissueGnostics), with the same settings for each group in the experiment.

Immunostaining

For mouse brain tissues obtained from embryos, fixation was carried out overnight in 4% paraformaldehyde (PFA) at 4 °C. After washing with PBS, mouse brains were dehydrated in 30% sucrose at 4 °C for one day, embedded in O.C.T. (SAKURA), and frozen-sectioned at a thickness of 12 μm using a Leica CM1950 cryostat. When antigen retrieval was required for certain antibodies, brain sections were treated with sodium citrate buffer (pH 6.0) at 90 °C for 20 min and then cooled to room temperature. For immunostaining, sections were first blocked in blocking buffer (10% donkey serum, 5% BSA, and 0.25% Triton X-100) at room temperature for 1–2 h, followed by overnight incubation with primary antibodies (diluted in blocking buffer) at 4 °C. After washing 3 × 10 min in PBS, secondary antibodies such as Alexa Fluor 488, 568, or 647 (Invitrogen) were added (1:500) and incubated at room temperature for 2 h. Following two PBS washes, sections were mounted using anti-fade mounting medium with DAPI (Beyotime). For cultured neurons, subsequent to two PBS washes, cells were fixed using 4% PFA for 15 min at room temperature, followed by two additional PBS washes and a 30-minute blocking at room temperature in blocking buffer. Subsequent antibody incubation conditions were similar to tissue sections. All images were acquired using Zeiss LSM 800 confocal microscope or TissueFAXS microscope. In the same experiment, each group had the same settings. For tissues requiring fluorescence intensity analysis, ImageJ software was used to measure immunofluorescence intensity, subtracting background intensity. Antibody sources and dilutions were as follows: Ybx1 (Abcam, ab76149, 1:500), Pax6 (BD Biosciences, 561,462, 1:500), Tbr2 (Abcam, ab183991, 1:500), Ctip2 (Abcam, ab18465, 1:500), Tbr1 (Abcam, ab31940, 1:500), Cleaved caspase-3 (Cell Signaling, 9661S, 1:500), BrdU (Abcam, ab6326, 1:500), BrdU (Thermo, B35128, 1:250), Ki67 (Cell Signaling, 12202S, 1:500), GFP (Abcam, ab13970, 1:500).

TUNEL assay

Apoptosis was detected using the TUNEL Apoptosis Detection Kit (Roche), following the manufacturer’s protocol. In brief, sections were fixed with 4% PFA, followed by blocking at room temperature for 1 h in blocking buffer (5% BSA and 0.25% Triton X-100). The TUNEL reaction mixture was added and incubated at 37 °C for 30 min. Subsequently, samples were stained with DAPI, and imaged using a Zeiss LSM 800 confocal microscope.

BrdU pulse labeling assay

Pregnant mice were intraperitoneally injected with BrdU dissolved in PBS (100 mg/kg), and embryonic brain tissues were harvested 24 h post-injection. For BrdU detection, antigen retrieval was performed using sodium citrate buffer (pH 6.0). Following thawing, slides were washed in PBS for 5 min at room temperature. Samples were then incubated in 2N HCl at room temperature for 30 min, followed by neutralization in 0.1 M borate buffer for 10 min. After washing with PBS, the samples underwent the same blocking and antibody incubation process as in immunofluorescence staining.

Neurosphere assay

Cortical tissues from E13.5 embryos were dissected and dissociated using Accutase (Sigma) at 37 °C. Cells were then cultured in NeuroCult medium (STEMCELL Tech) supplemented with EGF (20 ng/mL, Gibco) and bFGF (10 ng/mL, Gibco). Cultures were refreshed every two days with 100 μL of medium, and neurospheres were imaged on day 7. For clonal analysis, primary neurospheres were dissociated into single cells using Accutase for 5 min and suspended in fresh culture medium. After 5 days, newly formed secondary neurospheres were photographed and quantitatively analyzed.

Cortical culture

Cortical tissues from mouse embryos were isolated using Leibovitz’s L-15 medium, followed by dissociation in HBSS supplemented with trypsin (Sigma) and DNase I (Sigma). Dissociated cortical cells were plated on glass coverslips or culture dishes pre-coated with poly-D-lysine (100 μg/ml, Trevigen) and laminin (3.3 μg/ml, Trevigen). Cultures were maintained in neurobasal medium (Gibco) supplemented with B27 (Thermo Fisher), GlutaMAX supplement (Thermo Fisher), and penicillin–streptomycin (Thermo Fisher). The medium was changed every two days with fresh culture medium.

Gene knockdown using lentiviral shRNA or siRNA

Lentiviral vectors for gene knockdown were prepared using the pLKO.1-Puro system (Addgene). Detailed procedures for lentivirus preparation are described in previous literature [13,18]. Target sequences for shRNA were as follows:

• shYbx1 #1: 5′‐GGTATCGCCGAAACTTCAATT‐3′
• shYbx1 #2: 5′‐GTATCGCCGAAACTTCAATTA‐3′
• shCtrl: 5′‐GCATAAACCCGCCACTCATCT‐3′

To select positively infected neurons, puromycin (1 μg/mL, Sigma) was added on the second day post-infection and treated for 24 h before sample collection for analysis.

Gene silencing experiments, employing siRNA-mediated knockdown of Ccnd2, eEF1g, Rpsa, Rps5, and Uhmk1, were performed using GP-transfect-Mate transfection reagent (Genepharma) following the manufacturer’s manual. The target sequences for siRNA were as follows:

• siCtrl: 5′-UUCUCCGAACGUGUCACGUTT-3′
• siYbx1 #1: 5′-GUAUCGCCGAAACUUCAAUUATT-3′
• siYbx1 #2: 5′-GGUAUCGCCGAAACUUCAAUUTT-3′
• siNsun2: 5′-GCUCACAACACUGAGAACATT-3′
• siNsun6: 5′-GCGAAGGUUUCUCAGAAAUTT-3′
• siCcnd2 #1: 5′-CGACUUCAAGUUUGCCAUGUATT-3′
• siCcnd2 #2: 5′-CAUUGAGCACAUCCUUCGCAATT-3′
• sieEF1g #1: 5′-GAGUGACACUGGCUGAUAUUATT-3′
• sieEF1g #2: 5′-GUGGUUCCUUACCUGCAUUAATT-3′
• siRpsa #1: 5′-AGUGACGGUAUCUACAUCAUATT-3′
• siRpsa #2: 5′-CCUGAUCUUUACUUCUACAGATT-3′
• siRps5 #1: 5′-GCGCCUUACUAACUCCAUGAUTT-3′
• siRps5 #2: 5′-GCCUUACUAACUCCAUGAUGATT-3′
• siUhmk1 #1: 5′-UAUCACCUCAGAGACCUUAUCTT-3′
• siUhmk1 #2: 5′-GAGUGCGGAGAAUGAGUGUUUTT-3′

Cells were incubated with the transfection solution for 4–6 h before replacing the medium and then samples were collected after 48 h for subsequent analysis.

RNA sequencing and data analysis

For RNA sequencing, cortices from E13.5 or E18.5 mouse embryos were dissected, and genotyping was conducted to differentiate between Ybx1 cKO and wild type control littermates. Tissues were lysed using Trizol reagent. Cortical cultures with Ybx1 knockdown were obtained as described earlier, and total RNA was extracted from cultured cortical cells infected with shRNA lentivirus after puromycin selection using Trizol reagent. Subsequently, mRNA purification, cDNA library construction, sequencing, and analysis were conducted. RNA sequencing was performed using the DNBSEQ platform from BGI, following standard procedures. The results of the gene expression profiles were derived from three independent replicate experiments.

RNA immunoprecipitation and sequencing (RIP-seq)

RIP experiments were conducted using the EZ-Magna RIP RNA-Binding Protein Immunoprecipitation Kit (Millipore), in accordance with a previously established protocol [16]. In brief, 2 × 10⁷ cells were lysed, and the lysates were incubated overnight at 4 °C with Ybx1 antibody (Abcam). Subsequently, immunoprecipitated RNA samples were used for cDNA library construction, sequencing, and analysis. Sequencing results for each group were derived from two independent replicate experiments. All enriched genes were subjected to Gene Ontology (GO) analysis, and gene length bias correction was performed using the GOseq R package. GO terms with corrected P values less than 0.05 were considered significantly enriched.

m⁵C mRNA immunoprecipitation

m⁵C mRNA immunoprecipitation was performed using the GenSeq m⁵C MeRIP Kit (GenSeq), following the manufacturer’s instructions. In summary, cortical RNA was extracted from E14.5 wild type mouse pups. The experiment involved the use of m⁵C antibody and corresponding control IgG (both included in the kit). GFP mRNA was in vitro transcribed using the DIG RNA labeling Kit (Roche) and used as a negative control to ensure the absence of m⁵C modifications. The immunoprecipitated RNA was then subjected to cDNA synthesis and quantitative real-time PCR based on SYBR Green.

RT-qPCR analysis

Total RNA from cortical tissues or cultured cortical cells was extracted using Trizol Reagent (Life). cDNA was synthesized with the PrimeScript RT Master Mix (Takara), and real-time quantitative PCR (qPCR) was performed using the ChamQ Universal SYBR qPCR Master Mix (Vazyme) on a BioRad qPCR CFX 96 instrument. qPCR results were quantified using the 2^−ΔΔCt method, calculating the difference in threshold cycle (Ct) values between the target gene and an internal control. The primer sequences used for qPCR were as follows: Ybx1: 5′‐GAGAGGATGGCAATGAAGAGG‐3′ and 5′‐TTGTGGTTTAGGGTTCTCTGG‐3′; Ccnd2: 5′‐GAGTGGGAACTGGTAGTGTTG‐3′ and 5′‐CGCACAGAGCGATGAAGGT‐3′; eEF1g: 5′‐ACCGCACCCCTGAATTTCTC‐3′ and 5′‐CTGGCGTACTTCCTCGCAG‐3′; Rpsa: 5′‐TGCGGGAACCCACTTAGGT‐3′ and 5′‐AGGATTCTCGATGGCAACAATAG‐3′; Rps5: 5′‐TGGCAGAGACCCCTGACAT‐3′ and 5′‐GGGCAGGTACTTGGCATACT‐3′; Uhmk1: 5′‐CTCTCCCAATGTGCCATCAC‐3′ and 5′‐TTCGTGGTTTGAGGTCTGCAT‐3′; β-actin: 5′‐GCGAGCACAGCTTCTTTGC‐3′ and 5′‐TCGTCATCCATGGCGAACT‐3′; GAPDH: 5′‐CAAGGAGTAAGAAACCCTGGAC‐3′ and 5′‐GGATGGAAATTGTGAGGGAG‐3′.

RNA stability assay

Cortical cells, isolated from mouse brain tissue, were plated in equal numbers per well and allowed to adhere. Cultured cells were treated with actinomycin D (ActD, ApexBio Technology) at a concentration of 5 μg/ml. Cells were collected at 0, 2.5, 5, 7.5, and 10 h, and RNA was isolated for qPCR analysis.

In utero electroporation and EdU assay

Plasmid transfection via intraventricular injection followed by in utero electroporation was performed as previously described [64]. In brief, the plasmids were diluted in Fast Green (Sigma–Aldrich) at a ratio of 1:20 to achieve a final concentration of 1 μg/μl. Pregnant E14.5 mice were anesthetized, and the uterine horns were exposed. A pulled glass micropipette was used to inject 1 μl of DNA solution into the lateral ventricle of the embryos. Tweezers (5 mm platinum, BTX, Harvard Apparatus) were positioned at the injection site, and the brain was electroporated using the BTX-EC830 electroporator (BTX, Harvard Apparatus) with five pulses at 40 V, 50 ms duration, and 950 ms intervals to introduce DNA into the cortex. Embryos were quickly returned to the abdominal cavity to minimize temperature loss. The abdominal wall and skin were sutured, allowing normal fetal development. To assess cell cycle exit, pregnant mice received an injection of EdU at a dose of 25 mg/kg 24 h after electroporation. Brains were collected and processed 24 h after injection, and EdU incorporation was detected using the Click-iT EdU Alexa Fluor Imaging Kit (Invitrogen).

Flow cytometry

For cell cycle analysis, neural stem cells were obtained from the mouse cortex and fixed overnight in 70% ethanol at −20 °C. Cells were washed with cold PBS, and the pellet was subsequently resuspended in FxCycle PI/RNase staining solution (Thermo Fisher Scientific), followed by a 30 min incubation at 37 °C. The cell suspension was filtered through a sterile cell strainer (40 μm) for analysis using the BD FACSAria flow cytometer, and the cell cycle distribution of neural stem cells was analyzed using FlowJo software.

Calculation of cell cycle kinetics

For the analysis of cell cycle dynamics in the embryonic cortex, a BrdU pulse (100 mg/kg body weight) was administered intraperitoneally to the pregnant dam on the day of the experiment. After 1.5 h, bromodeoxyuridine (EdU, 25 mg/kg body weight) was injected, and the embryos were sacrificed 30 min later. Detection of BrdU and EdU in the double-labeling experiments was performed according to the method described above.

Cell cycle duration was calculated using the following paradigm [65]. During the interval between BrdU and EdU (T_i = 1.5 h), cells in the initial BrdU-labeled cohort that exit S phase are designated as the leaving fraction (L cells), which will be labeled with BrdU but not EdU. Cells labeled with both BrdU and EdU are designated as S cells. The ratio of the duration of any phase of the cell cycle to another phase is equal to the ratio of the number of cells in that phase to the number of cells in the other phase [66]. Therefore, the ratio between the number of L cells and S cells is equal to the ratio of T_i (1.5 h) to T_s [67]. S phase duration (T_s) can be calculated using the following formula: T_S(1.5) = (S cells)/(L cells). The total cell cycle duration (T_c) can be estimated using the following formula: T_C/T_S = (P cells)/(S cells), where P cells represent the total number of proliferating cells in the sampling area.

Quantification and statistical analysis

Unless otherwise specified, all values are presented as the mean ± SEM. Statistical analyses were conducted utilizing GraphPad Prism version 8.0. For the comparison of means between two groups, unpaired Student t test was employed based on the experimental design. When comparing means among more than two groups, one-way analysis of variance (ANOVA) was conducted, followed by Tukey’s multiple comparison test. A p value less than 0.05 was considered statistically significant: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.