Polypeptide N-Acetylgalactosaminyltransferase 13 Contributes to Neurogenesis via Stabilizing the Mucin-type O-Glycoprotein Podoplanin*

Mucin-type O-glycosylation is initiated by an evolutionarily conserved family of polypeptide N-acetylgalactosaminyltransferases (ppGalNAc-Ts). Previously, it was reported that ppGalNAc-T13 is restrictively expressed at a high level in the brain. Here we provide evidence for the critical role of ppGalNAc-T13 in neural differentiation. In detail, we show that the expression of ppGalNAc-T13 was dramatically up-regulated during early neurogenesis in mouse embryonic brains. Similar changes were also observed in cell models of neuronal differentiation by using either primary mouse cortical neural precursor cells or murine embryonal carcinoma P19 cells. Knockout of ppGalNAc-T13 in P19 cells suppressed not only neural induction but also neuronal differentiation. These effects are at least partly mediated by the mucin-type O-glycoprotein podoplanin (PDPN), as knockdown of PDPN led to a similar inhibition of neuronal differentiation and PDPN was significantly reduced at the posttranscriptional level after ppGalNAc-T13 knockout. Further data demonstrate that PDPN acts as a substrate of ppGalNAc-T13 and that the ppGalNAc-T13-mediated O-glycosylation on PDPN is important for its stability. Taken together, this study suggests that ppGalNAc-T13 contributes to neuronal differentiation through glycosylating and stabilizing PDPN, which provides insights into the regulatory roles of O-glycosylation in mammalian neural development.

Development of the CNS involves a well ordered generation of a variety of distinct neural cell types with progressive restriction in fate potential of neural progenitors (1). This process is precisely regulated by a large set of transcriptional factors (2,3) and occurs with the reorchestration of molecule expression on the cell surface. These molecules, as indicated by many previous reports, include not only various glycoproteins or glycolipids but also the glycans on them (4 -6). Accumulating evidence shows that cell surface glycans play crucial roles in CNS development, where they do not function independently but via reg-ulating the functions of their carrier proteins or lipids. It has been reported that N-glycans modulate neural cell adhesion, axonal targeting, neural transmission, and neurite outgrowth by affecting the folding or trafficking of certain carrier glycoproteins such as synaptic vesicle protein 2 (SV2) and ionotropic glutamate receptors (7). Also, genetic inactivation of ST8Sia II and ST8Sia IV sialyltransferases, which catalyze polysialic acid structures on neural cell adhesion molecule, leads to severe defects in neurite outgrowth, synaptic plasticity, etc. (7). In contrast to the increasing evidence for the significance of sialylation and N-glycosylation, there is very limited information regarding the mechanistic roles of O-glycosylation in neural development (8).
Mucin type O-glycosylation is an evolutionarily conserved protein modification that plays important roles in protein processing, secretion, stability, and ligand binding (9,10). In mammals, it is initiated by a family of 20 UDP-GalNAc:polypeptide N-acetylgalactosaminyltransferase (ppGalNAc-Ts, 2 EC 2.4.1.41) (11)(12)(13)(14), which transfer GalNAc to the hydroxyl group of serine or threonine, forming an ␣-anomeric linkage. ppGalNAc-Ts display tissue-specific expression in mammals and, notably, unique spatial and temporal patterns of expression during eukaryotic development as well (15). The correct timing of their expression is crucial for proper development, as aberrant expression has been associated with many developmental disorders and multiple diseases (16 -27). As examples, ppGalNAc-T1-deficient mice exhibit a developmental delay of the submandibular gland and reduced secretion of basement membrane components (19). ppGalNAc-T2 has been reportedly involved in abnormalities of lipid metabolism (23,24), at least partly by modulating the proprotein processing of angiopoietin-like protein 3 (ANGPTL3) (25). Unlike the ubiquitous expression of ppGalNAc-T1 and T2, our group has reported that ppGalNAc-T13 is an enzyme restrictively expressed at a high level in the brain, although it shares 84% amino acid identity with ppGalNAc-T1 (28). This prompted us to investigate the role of ppGalNAc-T13 in neural development.
In this study, we show that ppGalNAc-T13 was dramatically up-regulated during neurogenesis in the mouse embryonic brain and persisted at a high level after birth. Knockout of ppGalNAc-T13 in murine embryonal carcinoma P19 cells not only reduced the aggregate size of neural stem cells but also inhibited neuronal differentiation. These effects of ppGalNAc-T13 are at least partly mediated by podoplanin (PDPN), a classical mucin-type O-glycoprotein, as further data show that the expression of PDPN was significantly down-regulated in ppGalNAc-T13 knockout cells; that, like ppGalNAc-T13 knockout, depletion of PDPN expression also resulted in an inhibition of neuronal differentiation; and that the ppGalNAc-T13-mediated O-glycosylation on PDPN is required for its stability. Taken together, our results clearly demonstrate that ppGalNAc-T13 contributes to neuronal differentiation by regulating the stability of PDPN, which provides a plausible explanation for its restrictive expression in the brain.

Results
The Expression of ppGalNAc-T13 Is Dramatically Up-regulated during Neurogenesis-Consistent with our previous study (28) in humans, quantitative RT-PCR (qRT-PCR) analysis with different murine tissues here also revealed high expression of ppGalNAc-T13 in the brain (Fig. 1A), suggesting a possible role of ppGalNAc-T13 in the nervous system. To test this idea, we first examined its expression during the embryonic and postnatal development of the mouse brain. In contrast to the persistently high level during the postnatal period, qRT-PCR results showed that ppGalNAc-T13 transcription in the embryonic stage was initially increased and peaked at embryonic day 17.5 (E17.5), followed by a momentary decrease at E19.5 (Fig. 1B). Because ppGalNAc-T1 shares 84% identity with ppGalNAc-T13, we also detected the expression of ppGalNAc-T1 during brain development. Interestingly, unlike the dramatic increase of ppGalNAc-T13, ppGalNAc-T1 was persistently highly expressed and had little change during the process (Fig. 1B).
To confirm the changes of ppGalNAc-T13 during brain development, an in vitro model of progressive neuronal differentiation was established using cortical neural precursor cells isolated from E17.5 mouse brain. Neurite formation and neuronal interactions were gradually up-regulated during the maturation of primary neuronal cells (Fig. 1C). Concomitant with these morphological changes, the expression of ppGalNAc-T13 increased first to a peak on day 7 and then decreased on day 9 ( Fig. 1, D and E). Double immunostaining of the primary neuronal cultures on day 7, intriguingly, showed that ppGalNAc-T13 was predominantly co-expressed with the neural stem cell marker Nestin as well as the pan-neuronal marker ␤-III-tubulin but not with the astrocyte marker GFAP (Fig. 1F), suggesting a cell type-specific distribution of ppGalNAc-T13 in neural stem cells and neurons.
The P19 embryonal carcinoma cell line is a well established model system for analyzing the factors that regulate neuronal differentiation (29 -31). P19 cells could differentiate into neurons as well as monolayer non-neuronal cells with astroglia morphology after induction with retinoic acid (ATRA) (29,32). Therefore, we also checked the expression of ppGalNAc-T13 in this system. Fig. 2A shows a schematic of ATRA-induced neuronal differentiation of P19 cells. It comprises two major stages: neural induction and neuronal differentiation. During neural induction, pluripotent P19 cells are allowed to aggregate and form embryonic bodies with increased expression of the neural stem cell marker Nestin (31). When neuronal differentiation occurs, most of the neural stem cells begin to differentiate into neurons, accompanied by an increased expression of a panneuronal marker, ␤-III-tubulin (30,31), whereas a small number of them could also differentiate into monolayer non-neuronal cells with astroglia morphology (29,32). As shown in Fig.  2B, these characteristic changes in cell morphology during neuronal differentiation were clearly observed in our system, and and ppGalNAc-T1 in mice. Gapdh was used as an internal control. The -fold changes of ppGalNAc-T13 and ppGalNAc-T1 expression were calculated based on the results of qRT-PCR (compared with E11.5; (E11.5-19.5) and postnatal days 1 (P1) to P49). To confirm the expression pattern of ppGalNAc-T13 during neuronal differentiation, primary cortical neural precursor cells from the mouse brain at E17.5 were isolated and cultured as described under "Experimental Procedures." C, representative images of primary neuronal cultures on days 1, 3, 5, 7, and 9 were captured. D, qRT-PCR using total RNA extracted from these cells was performed to analyze the mRNA levels of ppGalNAc-T13. Gapdh was used as an internal control. E, the corresponding protein levels of ppGalNAc-T13 were examined by Western blotting analysis. ␤-Actin was used as a loading control. F, to determine the cellular distribution of ppGal-NAc-T13 in the brain, double immunostaining of primary neuronal cultures on day 7 was performed using anti-ppGalNAc-T13 antibody (red) paired with antibodies (green) against Nestin (neural stem cell marker), ␤-III-tubulin (panneuronal marker), or GFAP (astrocyte marker). The quantitative data were obtained from three independent experiments. Error bars indicate standard deviation. The p values were calculated using two-tailed unpaired t test. *, p Ͻ 0.05; **, p Ͻ 0.01 (versus E11.5 or day 1).
our immunostaining assay revealed a high expression of ␤-IIItubulin in the cells on day 6. Meanwhile, Western blotting analysis showed that ␤-III-tubulin expression was significantly upregulated during the stage of neuronal differentiation (Fig. 2D). These results suggest the successful establishment of the neuronal differentiation model in this study. Based on analyses by qRT-PCR and Western blotting (Fig. 2, C and D), we found that the expression of ppGalNAc-T13 was markedly increased during neuronal differentiation of P19 cells, whereas ppGalNAc-T1 was persistently expressed at a high level and had a slight change. These results are consistent with the observation in vivo. Taken together, our results clearly demonstrate that, unlike the small change of ppGalNAc-T1, ppGalNAc-T13 expression is dramatically up-regulated during neurogenesis, indicating a potential role of ppGalNAc-T13 in this process.
Knockout of ppGalNAc-T13 Inhibits Neuronal Differentiation of P19 Cells-To directly assess the functional contributions of ppGalNAc-T13 in neuronal differentiation, we knocked out the endogenous ppGalNAc-T13 of P19 cells by CRISPR/Cas9 genome editing technology. Two clones (C4 and C13) with different frameshift mutations in ppGalNAc-T13 gene were obtained and verified by DNA sequencing and Western blotting analysis (Fig. 3, A and B). Upon treatment with ATRA, both of the clones showed a significant decrease in the aggregate size of neural stem cells on day 4 compared with wild-type P19 cells (Fig. 3, C and D). In detail, in contrast to control cells whose aggregate diameters mostly ranged from 70 -110 m, we found that most of the ppGalNAc-T13 knockout aggregates had a diameter of 50 -90 m (Fig. 3E), suggest-ing a critical role of ppGalNAc-T13 in neural induction. Consistent with these morphological changes, a clear reduction in the expression of Nestin was observed after the loss of ppGalNAc-T13 (Fig. 3F). In addition to the effect on neural induction, ppGalNAc-T13 knockout also led to an inhibition of neuronal differentiation. This was evidenced not only by the increased monolayer non-neuronal cells during the neuronal differentiation stage (Fig. 3, G and H) but also by the reduced number of ␤-III-tubulin-positive cells (Fig. 3, I and J) as well as ␤-III-tubulin expression in whole cell lysates (Fig. 3B). Taken together, these results demonstrate that ppGalNAc-T13 is important for both neural induction and neuronal differentiation of P19 cells.

The Regulatory Effects of ppGalNAc-T13 on Neuronal Differentiation Are Mediated by PDPN, a Typical Mucin-type O-glycoprotein-The question arises of how ppGalNAc-T13
functions in neuronal differentiation. The marked enhancement of ppGalNAc-T13 expression was reminiscent of a typical mucin-type O-glycoprotein PDPN, which has also been reported to increase dramatically during the neuronal differentiation of P19 cells (33,34). Consistently, such an increase was also observed in our neuronal differentiation model of both primary cortical neural precursor cells and P19 cells by Western blotting analysis (Fig. 4, A and B). These observations prompted us to ask whether ppGalNAc-T13 affects neuronal differentiation via PDPN. To test this, we first checked the effects of ppGalNAc-T13 on PDPN expression. Interestingly, knockout of ppGalNAc-T13 resulted in a clear decrease in PDPN expression (Fig. 4C). To further confirm the hypothesis above, we next knocked down Pdpn by RNA interference tech-FIGURE 2. ppGalNAc-T13 expression is increased during neuronal differentiation of P19 cells. A, schematic of neuronal differentiation of P19 cells. P19 cells were initially induced by 1 M ATRA to allow the formation of embryonic body-like structures. After 4 days, spherical embryoid bodies were dissociated and replated into DMEM/F12 1:1 medium containing 1% N2 to promote neurite formation and neuronal interaction. B, representative images were captured during neuronal differentiation of P19 cells. To confirm the successful establishment of this model, cells on day 6 were immunostained with anti-␤-III-tubulin antibody (red). The nuclei were visualized by staining with DAPI (blue). C, the mRNA levels of ppGalNAc-T13 and ppGalNAc-T1 at the indicated times during neuronal differentiation of P19 cells were detected by qRT-PCR. Gapdh was used as an internal control. The -fold changes were calculated based on the results of qRT-PCR (compared with day 0). D, the corresponding protein levels of ppGalNAc-T13 and ␤-III-tubulin were analyzed by Western blotting analysis. ␤-Actin was used as a loading control. The quantitative data were obtained from three independent experiments. Error bars indicate standard deviation. The p values were calculated using two-tailed unpaired t test. *, p Ͻ 0.05; **, p Ͻ 0.01 (versus day 0). nology in P19 cells and examined the effects on neuronal differentiation. Fig. 4D shows efficient shRNA-mediated silencing of Pdpn. Like ppGalNAc-T13 knockout, knockdown of PDPN also reduced the aggregate size of neural stem cells (Fig. 4, E and F). The majority of cell aggregates formed by control cells had diameters of 70 -110 m, whereas the diameters of cell aggregates formed by PDPN knockdown P19 cells mainly ranged from 50 -90 m (Fig. 4G). In addition, knockdown of PDPN also inhibited neuronal differentiation, as seen not only by increased monolayer non-neuronal cells after silencing Pdpn (Fig. 4, H and I) but also by a significant decrease in the number of ␤-III-tubulin-positive cells (Fig. 4, J and K) and the expression of ␤-III-tubulin in the whole cell lysates (Fig. 4D). These results suggest that PDPN is also critical for the neuronal differentiation of P19 cells. So it is reasonable to conclude that PDPN, at least in part, mediates the effects of ppGalNAc-T13 on neuronal differentiation.
PDPN Acts as a Substrate of ppGalNAc-T13 and ppGalNAc-T1, and Some Sites on PDPN Could Be Glycosylated Only by ppGalNAc-T13-Different ppGalNAc-Ts display distinct substrate specificities (35). To determine whether ppGalNAc-T13 is responsible for the O-glycosylation of PDPN, an in vitro enzymatic activity assay was carried out using peptide fragments of PDPN with potential O-glycosylation sites (Fig. 5A). As shown in Fig. 5B, all five peptides could be glycosylated by the recombinant ppGalNAc-T13, indicating that PDPN serves as a substrate of ppGalNAc-T13. In addition to ppGalNAc-T13, we also checked the activity of ppGalNAc-T1 and ppGalNAc-T2. In contrast to the weak activity of ppGalNAc-T2, ppGalNAc-T1 also had high activities to PDPN peptides like ppGalNAc-T13. However, notably, on PDPN peptide S4, we found a specific product peak (P4) of ppGalNAc-T13 that was not produced by ppGalNAc-T1 (Fig. 5, B and C). MALDI-TOF mass analysis showed that, like the common product P3, this ppGalNAc-T13-specific product also contained two Gal-NAc moieties (Fig. 5D). These results demonstrate that both of ppGalNAc-T13 and ppGalNAc-T1 could glycosylate PDPN but that there exist some sites on PDPN that could be specifically glycosylated by ppGalNAc-T13.
O-glycosylation of PDPN Is Important for Its Stability-Next, to explore the mechanisms by which ppGalNAc-T13 regulates the expression of PDPN, we examined the transcriptional level of Pdpn in ppGalNAc-T13 knockout clones. Intriguingly, no big change was observed after ppGalNAc-T13 knockout (Fig.  6A), suggesting that ppGalNAc-T13 may regulate the expression of PDPN at a posttranscriptional level. Consistent with this hypothesis, there was little difference in the transcription of Pdpn during the neuronal differentiation of either primary cortical neural precursor cells or P19 cells (Fig. 6A), although its protein level increased dramatically as described above. Given the accumulating evidence for the importance of O-glycosylation in protein stability (36 -39), we then investigated the effects of O-glycosylation on the stability of PDPN using two CHO cell lines CHO-K1 and CHO-ldlD. CHO-ldlD is an O-glycosylation-deficient cell line because of the lack of the epimerase that converts UDP-Glc/GlcNAc to UDP-Gal/GalNAc (40). Therefore, in contrast to CHO-K1 cells, where the majority of the FLAG-tagged PDPN protein overexpressed was hyperglycosylated, the recombinant PDPN in CHO-ldlD cells was exclusively underglycosylated (Fig. 6, B and C). To compare the relative half-life of PDPN in these cells, translation was arrested by treatment with 100 g/ml cycloheximide (CHX). At various times after CHX addition, protein was extracted, and total PDPN was semiquantified by Western blotting analysis. As seen in Fig. 6B, a significant amount of hyperglycosylated PDPN was still detectable after 16 h (t1 ⁄ 2 , ϳ24 h) in CHO-K1 cells. In contrast, most of the underglycosylated PDPN was already degraded after 12 h (t1 ⁄ 2 , ϳ14 h) in the CHO-ldlD cells (Fig. 6, C, I, and J). Addition of exogenous GalNAc and Gal into the culture medium of CHO-ldlD cells clearly enhanced the level of hyperglycosylated PDPN (Fig. 6D) and prevented protein decay (t1 ⁄ 2 , ϳ18 h) (Fig. 6, I and J). These results clearly demonstrate that O-glycosylation is important for the stabilization of PDPN. Consistently, concomitant expression of either ppGalNAc-T13 or ppGalNAc-T1 in CHO-K1 or CHO-ldlD cells led to a further increase in the expression of hyperglycosylated PDPN (Fig. 6, E-J) as well as its half-life, suggesting that ppGalNAc-T13-or ppGalNAc-T1-mediated O-glycosylation could facilitate the stabilization of PDPN. These data suggest that O-glycosylation on PDPN is critical for its stability.
Overexpression of ppGalNAc-T13, but Not ppGalNAc-T1, Rescues the Neuronal Differentiation Defect of ppGalNAc-T13deficient P19 Cells-Although knockout of ppGalNAc-T13 using the CRISPR-Cas9 system inhibited neuronal differentiation of P19 cells, the influence could be due to potential offtarget effects of the designed sequence. To rule out this possibility, we overexpressed ppGalNAc-T13 in ppGalNAc-T13 knockout P19 cells. The expression level was verified by Western blotting analysis (Fig. 7E). Restoration of ppGalNAc-T13 expression rescued the defects in neuronal differentiation to a large extent, as shown by the significant decrease in monolayer non-neuronal cells (Fig. 7, A and B) and the increase in the number of ␤-III-tubulin-positive cells (Fig. 7, C and D) compared with ppGalNAc-T13 knockout P19 cells. Also, PDPN in the ppGalNAc-T13-overexpressing cells was up-regulated to a level comparable with that seen in wild-type P19 cells despite the small difference in its transcription (Fig. 7, E and F). These results further corroborated that ppGalNAc-T13 may contribute to neuronal differentiation by glycosylating and stabilizing PDPN. Given the fact that ppGalNAc-T13 shows high homology to the ubiquitous ppGalNAc-T1 and that ppGalNAc-T1 also facilitated the stabilization of PDPN in CHO cells (28), we asked whether ppGalNAc-T1 could complement the function of ppGalNAc-T13 in neurogenesis. To test this idea, we overexpressed ppGalNAc-T1 in ppGalNAc-T13 knockout cells and replicated the experiments above. As shown in Fig. 7, A-E, overexpression of ppGalNAc-T1 rescued the expression of PDPN but had little effects on the emergence of ␤-III-tubulinpositive cells and monolayer non-neuronal cells, indicating a distinct role of ppGalNAc-T13 in neuronal differentiation. Taken together, our results demonstrate that ppGalNAc-T13 promotes neurogenesis at least in part through glycosylating and stabilizing the mucin-type O-glycoprotein PDPN.

Discussion
In contrast to the growing evidence for the importance of O-glycosylation in organ development and various diseases (12,17,23,41), there is a particular dearth of information regarding its roles in neural development. Previously, our group found that ppGalNAc-T13 is restrictively expressed at a high level in the brain (28). This raised the question of whether it functions

. The classical mucin-type O-glycoprotein PDPN is involved in mediating the regulatory effects of ppGalNAc-T13 in neuronal differentiation.
A and B, cell lysates prepared from primary neuronal cultures (A) and P19 cells (B) at the indicated times of neuronal differentiation were immunoblotted with anti-PDPN antibody. C, to determine the effect of ppGalNAc-T13 knockout on PDPN expression, the protein levels of PDPN in wild-type and ppGalNAc-T13 knockout P19 cells at the indicated times were examined by Western blotting analysis. ␤-Actin was used as a loading control. To check whether PDPN was also involved in neuronal differentiation, scramble (NC) or Pdpn-targeted shRNA (1#, 3#) was transfected into P19 cells. D, after 6 -12 h, the cells were induced to neuronal differentiation by ATRA, and cell lysates from the indicated cells on day 8 were immunoblotted with anti-PDPN and ␤-III-tubulin antibodies to confirm the knockdown efficiency of PDPN. ␤-Actin was used as a loading control. The density of the band was semiquantified by QuantityOne software. E, to check the effect of PDPN knockdown on neural induction, representative images of the cell aggregation of shRNA-transfected P19 cells on day 4 were captured. F, their diameter was statistically analyzed and is presented by median/scatter dot plot. G, the percentage of cell aggregates with different diameters (using eight fields containing at least 50 aggregates) was statistically analyzed. H, to check the effect of PDPN knockdown on neuronal differentiation, representative images of the neuronal morphologies of the indicated cells on day 8 were captured. Typical monolayer non-neuronal cells are indicated by red triangles. I, the number of these cells from five different fields was statistically analyzed. J, immunostaining of differentiated P19 cells on day 6 was carried out with the antibody against ␤-III-tubulin (red). The nuclei were visualized by staining with DAPI (blue). ␤-III-tubulin-positive cells are indicated by white triangles. K, the percentages of ␤-III-tubulin-positive cells were determined by the number of ␤-III-tubulin-positive cells within more than 1000 cells and 10 fields. The quantitative data were obtained from three independent experiments. Error bars indicate standard deviation. The p values were calculated using two-tailed unpaired t test. *, p Ͻ 0.05; **, p Ͻ 0.01 (versus NC). NOVEMBER 4, 2016 • VOLUME 291 • NUMBER 45 in brain development. In this study, we demonstrate that high expression of ppGalNAc-T13 contributes to neuronal differentiation by glycosylating and stabilizing the mucin-type O-glycoprotein PDPN.

The Significance of ppGalNAc-T13 and PDPN in Neurogenesis
CNS development is a complex process that is precisely regulated by a large set of transcription factors (42)(43)(44). Our results showed that the expression of ppGalNAc-T13 was markedly up-regulated during neurogenesis in the mouse embryonic brain and sustained a high level at the postnatal stage. This could be partly mediated by PAX7, a neuron-specific transcription factor that also occurs from early development and persists in restricted regions, including neurons, throughout adulthood (45), because PAX7 has been reportedly involved in the regulation of ppGalNAc-T13 transcription (46), FIGURE 5. PDPN is a substrate of ppGalNAc-T13 and ppGalNAc-T1, whereas some sites on PDPN could be glycosylated only by ppGalNAc-T13. A, to test whether ppGalNAc-T13 could glycosylate PDPN, an in vitro enzymatic activity assay was performed using the recombinant ppGalNAc-Ts and five peptide fragments of PDPN with potential O-GalNAc glycosylation sites. The potential glycosylation sites were predicted with NetOGlyc 4.0 and are indicated by black dots. B, the purity of the recombinant ppGalNAc-Ts was checked by silver staining analysis. BSA served as the standard for quantization. C, the ability of ppGalNAc-T1, T2, and T13 to glycosylate five PDPN peptides was determined by reverse-phase HPLC. Muc5Ac served as the positive control. The peaks of glycosylated peptides were shifted to the left. S, substrates; P, products. The asterisks indicates a contaminant. The black arrow indicates the specific product peak of ppGalNAc-T13 (P4 of PDPN-S4). D, to clarify the difference in substrate specificities between ppGalNAc-T1 and T13, this ppGalNAc-T13-specific peak, together with other common peaks derived from ppGalNAc-T13-or ppGalNAc-T1-treated PDPN-S4, was isolated and subjected to MALDI-TOF mass analysis. and overexpression of PAX7 is capable of promoting the neuronal differentiation of P19 cells (47). However, during brain development, a momentary decrease in ppGalNAc-T13 expression was observed at E19.5. In general, this time point in mouse brain development corresponds to the period of gliogenesis, particularly the generation of astrocytes (48). Given our immunostaining result that ppGalNAc-T13 was specifically expressed in neural stem cells and neurons but not in astrocytes, it is reasonable to attribute the reduction at E19.5 to the enhanced proportion of astrocytes in the mouse brain we isolated. Consistent with this hypothesis, a similar decrease in ppGalNAc-T13 expression was observed, accompanied by a dramatic increase in expression of the astrocyte marker Gfap (data not shown) on day 9 of the primary neuronal cultures. In contrast, ppGalNAc-T13 expression was consistently up-regulated during the neuronal differentiation of P19 cells, where astrocytes were rarely generated (49,50). Collectively, these results indicate that ppGalNAc-T13 could serve as a marker for neural stem cells and neurons.
PDPN is a classical mucin-type glycoprotein that has been not only well known for its functions in the activation of platelet aggregation and maintenance of the normal development of lymphatic vessels (51) but associated with the progression of multiple types of carcinomas (52,53). Our results demonstrate for the first time that PDPN is also involved in neuronal differentiation, as knockdown of PDPN in P19 cells resulted in a significant inhibition of neuronal differentiation-related morphological and molecular changes. Considering also that FIGURE 6. The O-glycosylation modification on PDPN is important for its stability. A, to confirm whether ppGalNAc-T13 regulates PDPN at the transcriptional level, the mRNA levels of Pdpn were examined by RT-PCR using total RNA extracted from wild-type and ppGalNAc-T13 knockout P19 cells at the indicated times. Also, RT-PCR was carried out to check the transcription of Pdpn during the neuronal differentiation of P19 cells and primary cortical neural precursor cells. Gapdh was used as a loading control. The density of the band was semiquantified by QuantityOne software. B and C, to examine the influence of O-glycosylation modification on PDPN stability, CHO-K1 (B) and CHO-ldlD (C) cells were transfected with PDPN-FLAG, and translation was arrested by treatment with 100 g/ml CHX. At various times after CHX addition, protein was extracted, and then total PDPN was semiquantified by Western blotting analysis with anti-FLAG and anti-␣-tubulin antibodies. ppGalNAc-T13 deficiency clearly reduced PDPN expression in a posttranscriptional manner and that increasing the O-glycosylation level of PDPN by forced expression of ppGalNAc-T13 led to an enhancement in the half-life of PDPN, it is reasonable to conclude that ppGalNAc-T13 promotes neurogenesis at least in part by conferring stability to PDPN. However, paradoxically, a recent study reported that lack of core 1 O-glycosylation does not affect intracellular trafficking or protein turnover of PDPN (54). One possible interpretation for these differences is that the stability of PDPN depends on whether the important site is glycosylated instead of how complex the glycan structure is. Despite clear evidence for the involvement of ppGalNAc-T13 in PDPN stability, we still could not rule out the possibility that ppGalNAc-T13 may also contribute to neuronal differentiation by facilitating the interaction between PDPN and its receptors because the O-glycosylation on the platelet aggregation-stimulating domain of PDPN has been reported to be important for the interaction with its receptor CLEC-2 (55)(56)(57). Consistent with this hypothesis, our in vitro enzymatic activity assay showed that ppGalNAc-T13 had a high preference for the peptide substrate (S1) covering the platelet aggregation-stimulating domain of PDPN. It should be noted that, in addition to S1, the peptide fragments S3 and S4 were also liable to be glycosylated by ppGalNAc-T13. Considering its high preference for the triple T antigen nouvelle (Tn) epitope site (28), such substrate specificities of ppGalNAc-T13 are conceivably due to the existence of consecutive Ser/Thr residues in these three peptide fragments, and whether these triple T antigen nouvelle (Tn) epitope sites on PDPN are involved in the interaction with its ligands in neurogenesis remains to be determined.
Although ppGalNAc-T1 shares high homology with ppGalNAc-T13, our data demonstrate a distinctive role of ppGalNAc-T13 in neurogenesis because overexpression of ppGalNAc-T13, but not ppGalNAc-T1, rescued the defects in neuronal differentiation of ppGalNAc-T13 knockout P19 cells. However, the expression of PDPN was restored in either ppGalNAc-T13 or ppGalNAc-T1-overexpressing cells. Considering also our result that ppGalNAc-T1-mediated O-glycosylation facilitated the stabilization of PDPN, it is conceivable that the O-glycosylation level of PDPN is important for its stability, whether it is catalyzed by ppGalNAc-T13 or not. But why did the restored PDPN after ppGalNAc-T1 overexpression fail to rescue the defects in the differentiation into ␤-III-tubulin positive cells? One possibility is that there is/are glycosylation site(s) on PDPN that is/are important for its functional contribution in neurogenesis and that these site(s) could be glycosylated by ppGalNAc-T13 but not ppGalNAc-T1, as several groups, including us, have revealed a significant difference in the preference for glycopeptides between ppGalNAc-T13 and ppGalNAc-T1 (28, 58, 59). In agreement with this, our in vitro study using different peptide fragments of PDPN as substrates revealed a specific product peak of ppGalNAc-T13 on peptide S4, and this peak should not be random, as it was always specifically produced upon treatment with ppGalNAc-T13 when we changed the reaction time and enzyme amount. Furthermore, O-glycosylation prediction using the isoform-specific prediction tool ISOGlyP (60) showed that there are some potential sites on PDPN that could be glycosylated by ppGalNAc-T13 but not ppGalNAc-T1, which also supported our in vitro enzymatic activity results. In addition to the hypothesis above, it is also possible that there may exist some other proteins that mediate the functions of ppGalNAc-T13 in neurogenesis and that these proteins could be specifically glycosylated by ppGalNAc-T13. Besides PDPN, both sydecan-3 and syndecan-1 have been found to be the substrates of ppGalNAc-T13 (28,61,62), and sydecan-3 has been reportedly involved in axon guidance, neuralmigration,andneuronalplasticity (63,64).Meanwhile,O-glycosylation modifications have also been observed on several neuron-specific transmembrane or matrix proteins such as p75 (neurotrophin receptor), amyloid precursor protein, neural cell adhesion molecule, and testicans (65)(66)(67)(68), although little is known about their functions in neural development. Clearly, further study is needed to clarify the mechanisms of the distinct roles of ppGalNAc-T13 in neural differentiation.
In summary, this study convincingly demonstrates that ppGalNAc-T13 promotes neurogenesis at least in part by stabilizing PDPN, which not only gives a plausible explanation for the restrictive expression of ppGalNAc-T13 in the brain but also provides insights into the regulatory roles of O-glycosylation in mammalian neural development.

Experimental Procedures
Mice-Pregnant and postnatal C57BL/6J mice were purchased from the Model Animal Research Center of Nanjing University (Nanjing, China). The animal protocols were approved by the Institutional Animal Care and Use Committee of Shanghai Jiao Tong University.
Primary Cortical Neuronal Cultures-The neuronal differentiation of primary cortical neural precursor cells was performed as described previously (69). In brief, 17.5-day pregnant mice were sacrificed, and brain cortices from the fetuses were isolated. The cortices then were dispersed with 0.125% trypsin at 37°C for 15 min, passed through screen mesh, and seeded at a density of 2.0 ϫ 10 5 cells/cm 2 in DMEM/F12 1:1 supplemented with 10% FBS and 1% penicillin-streptomycin. After 4 h, the supernatants were removed, and cells were cultured in serum-free Neurobasal medium supplemented with 2% B27, 2 mM L-glutamine, and 1% penicillin-streptomycin. The regents used above were purchased from Invitrogen.
Cell Lines and Cell Culture-P19C6, a subclone of the mouse embryonal carcinoma P19 cell line, was provided by Dr. Naihe Jing (Institute of Biochemistry and Cell Biology, Chinese Academy of Sciences). P19 cells, CHO epitheloid cells (CHO-K1), and mutant cells (CHO-ldlD) (40) were maintained in DMEM/ F12 1:1 medium containing 10% FBS and cultured under a humidified atmosphere containing 5% CO 2 at 37°C.
Neuronal Differentiation of P19 Cells-Neuronal differentiation of P19 cells was induced by all-trans-retinoid acid (ATRA) as reported previously (29). In brief, cells were plated at 10 6 cells/bacteriological 10-cm Petri plate in ␣-minimal essential medium containing 10% FBS and 1 M ATRA (Sigma) to allow the formation of embryoid body-like structures. After 4 days, spherical embryoid bodies were harvested, dissociated, and replated onto poly-L-lysine-and 1 g/ml fibronectin-coated tissue culture dishes and cultured for another 4 days with DMEM/F12 1:1 medium containing 1% N2 (Invitrogen). The culture medium was changed every 2 days.
Establishment of the ppGalNAc-T13 Knockout P19 Cell Line-The single guide RNAs (forward, 5Ј-CACCCGCACTT-AACCGAAGTCTGC-3Ј; reverse, 5Ј-AAACGCAGACTTCG-GTTAAGTGCGC-3Ј) targeting exon 4 of the mouse ppGalNAc-T13 gene were designed with an online tool and inserted into the pX330 plasmid. The constructed plasmids were then transfected into P19 cells using Lipofectamine 2000 transfection reagent (Invitrogen) according to the instructions of the manufacturer. After 24 h, the cells were suspended and diluted onto 96-well plates. Single colonies were selected, passaged, and genotyped. To verify the mutation in the ppGalNAc-T13 gene, the genomic region around the target site was amplified by PCR using the primers 5Ј-TGCATAGAGAGTTTTCTTGTC-TGA-3Ј (forward) and 5Ј-AGAAAACAATTCATA TTTCTT-AGCA-3Ј (reverse) and subjected to digestion with T7E1 enzyme. To further confirm the indels, the products were cloned into the pGEM-T vector for sequencing.
RNA Interference-To knock down endogenous PDPN expression in P19 cells, the target sequences of Pdpn (5Ј-GCG-