VC induces TSC differentiation and Hand1 protein
To determine the role of VC on the TSC differentiation, we treated the primary mouse TSCs grown in the differentiation medium supplemented with or without 100 µM VC for 3-d and analyzed the mRNA levels of trophoblast-specific markers. Quantitative RT-PCR results showed VC treatment robustly induced the mRNA levels of Ctsq, PL-2, PL-1, P450scc, and Plf by approximately 4.7-, 2.7-, 1.2-, 0.8- and 0.4-fold, respectively, as compared to vehicle treatment (Fig. 1a-c). In contrast, VC treatment enhanced Tpbpα mRNA levels by 1.3-fold, but affected the mRNA levels of neither SynA nor SynB over vehicle treatment (Fig. 1b and c). Thus, VC serves as a potent inducer of TSC differentiation into SpTs and TGCs, especially S-TGCs.
Since the markers of TGCs are most strikingly increased after VC treatments and all TGC subtypes require the bHLH transcription factor Hand1 for their differentiation [5], we next examined the Hand1 protein and mRNA levels in primary TSCs after VC treatments for 3-d. VC treatments did not affect the Hand1 mRNA levels but increased the Hand1 protein in a dose-dependent manner, and the maximal induction was achieved by the exposure of TSCs with 30 or 100 µM of VC (Fig. 1d and Supplemental Figure S1a). Western analyses of cytosol versus nuclear fraction of Hand1 protein in primary TSCs indicated VC treatments induced not only the cytosol but also the nuclear Hand1 levels (Fig. 1e). Likewise, VC treatments in human trophoblast-like choriocarcinoma JEG-3 cells or HEK293T cells (293T) transiently transfected with Flag-tagged Hand1 (Flag-Hand1) consistently caused a dose- and time-dependent increase of exogenous Hand1 protein levels (Fig. 1f, g and Supplemental Figure S1b). To confirm the effect of VC on Hand1 expression and distribution, we examined the subcellular localization of the Flag-Hand1 in response to VC by immunofluorescence confocal microscopy. Detection of the Flag tag revealed that the exogenous Hand1 was rarely present in either the cytoplasm or nucleus in the absence of VC, but was present at a robust level in the nuclei and at a relatively low level in the cytosol in the presence of VC (Fig. 1h). Thus, VC induces the TSC differentiation into TGCs, especially S-TGCs, by inducing the protein expression of Hand1.
VC inactivates JNK to stabilize Hand1 in regulating the TSC differentiation
VC functions as a potent antioxidant in a variety of cellular functions, we next investigated the role of oxidative stress in the TSC differentiation into S-TGC. Although treatments of TSCs with VC significantly induced the Ctsq mRNA, neither the treatments with other antioxidants including N-acetyl cysteine (NAC), α-tocopherol (VE), glutathione (GSH), and lipoic acid (LA) affected the basal Ctsq mRNA, nor the treatments with oxidants antagonized the effect of VC on the Ctsq mRNA (Supplemental Figure S2a). Basic HLH factors usually form either homodimer or heterodimer to induce the TGC differentiation, and induction of TSC differentiation into S-TGC requires the formation of Hand1 homodimer to regulate the transcription [20]. To determine the effect of VC on the formation of Hand1 homodimer, we performed the co-immunoprecipitation using a HA antibody in JEG-3 cells co-transfected with HA-Hand1 and Flag-Hand1 and treated with or without VC. Western assays using a Flag antibody indicated that VC treatment had no influence on the homodimerization of Hand1, but caused an increase in the amount of Hand1 homodimer due to the induction of Hand1 protein expression (Supplemental Figure S2b). Thus, VC induces TSC differentiation into TGC subtypes due primarily to its induction of Hand1 expression but not its antioxidative action.
We have previously shown that DHA, an oxidized form of VC, inactivates c-Jun N-terminal kinase (JNK), a member of mitogen-activated protein kinases (MAPKs), in regulating steroidogenesis of human trophoblast-like cells and literature has readily indicated that TSC stemness maintenance by fibroblast growth factor 4 (FGF4) requires MAPK kinase kinase (MEKK4) activation of JNK [21, 22]. To determine the role of VC in the activation of MAPKs, we performed the western analyses in primary mouse TSCs treated with VC for 30 min. VC ranging from 0 to 100 µM decreased the phosphorylated JNK (p-JNK) and P38 (p-P38) up to 30% and 40%, respectively, but did not affect the phosphorylated extracellular signal–regulated kinases 1/2 (p-ERK) (Fig. 2a). In contrast, anisomycin, an agonist of both JNK and P38, not only time-dependently negated the Hand1 protein but also dose-dependently diminished VC-induced Ctsq mRNA in the primary TSCs (Fig. 2b and c). Importantly, though SB203580, a specific P38 inhibitor, did not affect the Ctsq mRNA levels, SP600125, a JNK inhibitor, dose-dependently induced the Ctsq mRNA levels in the primary TSCs (Fig. 2d and e). Consistently, anisomycin and SP600125 markedly decreased and increased the expression of Flag-Hand1, respectively, whereas SB203580 did not affect the expression of Flag-Hand1 in JEG-3 cells (Fig. 2f and g). Thus, JNK instead of P38 and ERK1/2 mediates the VC-induced Hand1 protein expression and TSC differentiation into TGC subtypes.
To dissect further the role of JNK in the VC-induced expression of Hand1, we next overexpressed a constitutively active form of JNK1 (JNK1*) or knocked down the JNK1/2. Overexpression of JNK1* significantly negated not only the basal but also VC-induced Flag-Hand1 protein, while JNK1/2-shRNA-expressing lentiviruses, which reduced the JNK1/2 expression by approximately 80%, induced the Flag-Hand1 protein by 2.3-fold in JEG-3 cells, compared with scramble-shRNA-expressing lentiviruses (Fig. 2h-j). To assess the potential interaction between the exogenous Hand1 and JNK1*, co-immunoprecipitation experiments were performed using cell lysates from JEG-3 cells transiently transfected with or without Flag-JNK1* and Myc-Hand1. Protein complexes precipitated with a Myc antibody contained no or a little Flag-p-JNK1* in JEG-3 cells expressing Myc-Hand1 or Flag-JNK1* alone but abundant Flag-p-JNK1* in JEG-3 cells expressing both Myc-Hand1 and Flag-JNK1*, in contrast, protein complexes precipitated with a Flag antibody contained no or little Myc-Hand1 in JEG-3 cells expressing Flag-JNK1* or Myc-Hand1 alone but abundant Myc-Hand1 in JEG-3 cells expressing both Myc-Hand1 and Flag-JNK1* (Fig. 2k). Thus, JNK1* physically interacts with and negates the Hand1 protein.
To determine whether JNKs negate the Hand1 protein via the proteasomal degradation of Hand1, we treated the JEG-3 cells expressing the Flag-Hand1 and Flag-JNK1*/JNK2* with MG132, a proteasome inhibitor, and performed western analyses using a 15% SDS-PAGE to detect the potential shift of Hand1 bands after JNK-mediated phosphorylation. Although both JNK1* and JNK2* robustly decreased the Flag-Hand1 in JEG-3 cells, JNK1* appeared to be more potent than JNK2* (Fig. 2l). Notably, JNK1* and JNK2* uniformly caused an apparent upper shift of Flag-Hand1 bands, and MG132 treatments did not affect the shift of Hand1 bands but robustly reversed either JNK1*- or JNK2*-negated Hand1 protein (Fig. 2l). This result indicates that JNK1/2 could directly phosphorylate Hand1 to enhance its proteasome degradation. To confirm this notion, we used the MG132 and VC to block the proteasomal degradation and inactivate JNK, respectively, in JEG-3 cells transfected with Flag-Hand1 and HA-tagged ubiquitin (HA-Ubiquitin), and then performed the co-immunoprecipitation using a Flag antibody as well as western analyses using a HA antibody. Protein complexes from the VC-treated cells contained 50% less ubiquitinated Hand1 with the predicated molecular weight of 35 kDa (band #1) and 30% less ubiquitinated Hand1 with the predicated molecular weight of 44 kDa (band #2), compared with those from vehicle-treated cells (Fig. 2m). This result suggests that VC inactivates JNK to block the ubiquitination-mediated proteasomal degradation of Hand1. Taken together, VC inactivates JNK to stabilize Hand1 and to subsequently promote the TSC differentiation into TGC subtypes.
JNK directly phosphorylates Hand1 at Ser48 to destabilize Hand1
Both the role of JNK in Hand1 stabilization and the physical interaction between these two proteins prompted us to examine whether JNK indeed directly phosphorylates Hand1. We used the Group-based Prediction System (GPS) to predicate the potential JNK phosphorylation sites on Hand1. A search of the Hand1 protein sequence revealed that Ser33, Ser48, and Ser81 sites or Ser98, Thr107, and Ser109 sites had high or low predication scores, respectively, with all conserved among human, mouse, and rat (Fig. 3a). To assess their potential importance, we expressed Flag-Hand1 variants that harbor mutations at the consensus serine or threonine residue (Ser and Thr to Ala) individually (S33A, S48A, S81A, S98A, T107A, and S109A) and evaluated their levels in response to JNK1* in JEG-3 cells by western analyses (10% SDS-PAGE). In the absence of JNK1*, Flag-Hand1 levels of wild-type (WT) were similar to those of S33A, S81A, S98A, T107A or S109A variants, but was 50% less than that of S48A variant; in the presence of JNK1*, Flag-Hand1 levels of WT were still similar to those of S33A, S81A, S98A, T107A, or S109A variants but was 85% less than S48A variant (Supplemental Figures S3a and b). Thus, Ser48 is critical for the JNK-mediated destabilization of Hand1.
We next performed the western assays using a 15% SDS-PAGE to detect the phosphorylated and non-phosphorylated Flag-Hand1 levels after overexpression of JNK1* in JEG-3 cells. In the absence of JNK1*, the S33A and S48A variants had essentially the same levels of phosphorylated and non-phosphorylated Flag-Hand1 as WT, whereas the S81A variant unexpectedly caused the abolishment of phosphorylated Flag-Hand1 without affecting the non-phosphorylated Flag-Hand1 levels (Fig. 3b). In the presence of JNK1*, the WT, S33A, and S81A variants had almost the same ratio between phosphorylated and non-phosphorylated Flag-Hand1 levels (3:1), whereas the S48A variant caused an reversed ratio between phosphorylated and non-phosphorylated Flag-Hand1 levels (1:5). Notably, S48A and S81A variants apparently caused the abolishment of the most upper phosphorylated band of Flag-Hand1 (Fig. 3b). Thus, these findings suggest that Ser81 is a potential phosphorylation site of kinases other than JNK, while Ser48 is a critical phosphorylation site of JNK.
We next determined whether Ser48 could indeed be phosphorylated by JNK. Here, we generated and purified the recombinant glutathione S-transferase (GST)-tagged wild-type Hand1 and its S48A mutant (Supplemental Figure S3c), and performed the in vitro phosphorylation assays using a commercially available active JNK2α2, followed by the SDS-PAGE and autoradiography. The wild-type Hand1 incorporated robust levels of 32P, as expected, the S48A variant incorporated no 32P even in the presence of active JNK2α2 (Fig. 3c, top). Subsequent western blotting using the GST antibody confirmed that the 32P-labeled band corresponded to the GST-Hand1 variants, and that similar levels of GST-Hand1 variants were present among the various reactive mixtures (Fig. 3c, bottom). Thus, JNK directly phosphorylates Hand1 at Ser48. To assess the importance of Hand1 phosphorylation at Ser48, we mutated the serine residue to phosphomimetic aspartate (D) and then transfected each construct of Flag-Hand1 variants with an enhanced green fluorescent protein (EGFP, for monitoring the transfection efficiency)-expressing vector into the JEG-3 cells. Western analyses showed EGFP constructs were expressed at almost the same levels, while S48A and S48D mutants caused a 150% increase and a 50% decrease in Flag-Hand1 levels over WT, respectively (Fig. 3d). To determine the potential importance of Ser48 in the VC-stabilizing Hand1, we evaluated the levels of Flag-Hand1 in cells expressing WT, S48A and S48D variants in response to VC at 100 µM. VC increased the WT levels by 1.4-fold over the control, but VC did not apparently increased the levels of S48A or S48D variants at all (Fig. 3e). Thus, Ser48 on Hand1 is an essential site for VC to stabilize Hand1.
To confirm the effect of specific phosphorylation on the Hand1 levels, we examined the intracellular levels of the Flag-tagged Hand1 variants in response to VC at 100 µM by an immunofluorescence confocal microscopy. Detection of the Flag tag revealed that the exogenous wild-type Flag-Hand1 was present in the cytoplasm at a relatively low level and in the nucleus at an exclusively high level under basal conditions, but was detected at significantly higher levels in the both compartments in response to VC stimulation (Fig. 3f). Importantly, the variant harboring S48A mutation accumulated at higher levels than wild-type Flag-Hand1 in both the cytoplasm and nucleus, despite the presence or absence of VC stimulation (Fig. 3f). Conversely, the S48D form was detected at lower levels than wild-type Flag-Hand1 in both the cytoplasm and nucleus, despite the presence or absence of VC stimulation (Fig. 3f). These results support further the notion that Ser48 participates in the Hand1 stabilization in response to VC stimulation.
VC/JNK/Hand1 signaling induces TGC differentiation in E8.5 murine placentas
Since VC suppresses the JNK activity to stabilize the Hand1 and all subtypes of TGCs require the Hand1 for their differentiation [20, 23, 24], we next examined the roles of VC deficiency, lentivirus-mediated knockdown of JNK or overexpression of Hand1 mutants in the TGC differentiation of E8.5 placentas. We firstly generated the global L-gulono-γ-lactone oxidase (Gulo−/−) knockout mice that are genetically unable to synthesize VC and maintained with tap water containing 4 g/L of VC for normal growth [25]. We then crossed the Gulo−/− males with the Gulo−/− females to generate the Gulo−/− placentas, and deprived of VC in the pregnant Gulo−/− females, when the plugs were identified (E0.5). Determination of maternal serum VC concentrations by HPLC indicated VC deprivation from E0.5 caused a time-dependent decrease of serum VC concentrations and serum VC dropped from 53.5 µM to approximately 11.5 µM at E8.5, compared with VC supplementation at 4 g/L (Supplemental Figure S4a). The pregnant females were dissected at E8.5 and the conceptuses were then subjected to cryosectioning and in situ hybridization with the probe of Pl-1, a marker of P-TGC and a target gene of Hand1 [26]. In the presence of VC supplementation, Pl-1-positive P-TGCs were robustly lined the implantation sites and apparently separated the maternal decidua from the ectoplacental cone, in contrast, in the absence of VC supplementation, Pl-1-positive P-TGCs were faintly lined the implantation sites and fuzzily separated the maternal decidua from the ectoplacental cone (Fig. 4a-b’). Thus, VC deficiency diminishes the differentiation of P-TGCs in the implantation sites by destabilizing the Hand1.
To confirm the importance of JNK-mediated destabilization of Hand1 in the TGC differentiation, we infected the wild-type blastocysts with green fluorescent protein (GFP)-carrying and scramble-shRNA- or JNK1/2-shRNA-expressing lentiviruses and performed the blastocyst transplantation. GFP-carrying and shRNA-expressing lentiviruses were successfully introduced into trophectoderm but not inner cell mass (ICM) of E3.5 blastocysts (Supplemental Figure S4b). JNK1/2-shRNA, which knocked down the JNK2 and JNK1 expression by approximately 36% and 63%, respectively, increased the Hand1 levels by approximately 2.5-fold in E16.5 whole placentas, as compared to the scramble-shRNA (Supplementary Figure S4c). Consistently, JNK1/2-shRNA robustly increased the number of Pl-1-positive P-TGCs that apparently lined the implantation sites and clearly separated the maternal decidua from the ectoplacental cone, as compared to the scramble-shRNA (Fig. 4c-d’).
To determine the effects of Hand1 mutants on the TGC differentiation, we then infected the E3.5 wild-type blastocysts with GFP-carrying and Hand1(WT)-, Hand1(S48A)- or Hand1(S48D)-expressing lentiviruses and performed the blastocyst transplantation. GFP-carrying Hand1 variants were successfully introduced into and almost evenly expressed in the trophectoderm of E3.5 blastocysts before transplantation (Fig. 4e-h). Overexpression of Hand1(WT) moderately enhanced the number of Pl-1-positive P-TGCs, as compared to overexpression of GFP alone; though overexpression of Hand1(S48A) robustly increased the number of Pl-1-positive P-TGCs, overexpression of Hand1(S48D) markedly decreased the number of Pl-1-positive P-TGCs on the implantation sites of E8.5 conceptuses, as compared to overexpression of Hand1(WT) (Fig. 4e’-h”). These results suggest that S48A and S48D mutations in Hand1 positively and negatively affect the Hand1 levels to regulate TGC differentiation into P-TGCs in vivo. Together, VC inactivates JNK to stabilize the Hand1, resulting in the induction of TGC differentiation of placentas.
VC deficiency causes the failure of pregnant maintenance
Since VC is critical for the TGC differentiation during placentation, we next examined the role of VC in the pregnant maintenance. We crossed the Gulo−/− males with the Gulo−/− females to generate the Gulo−/− embryos and placentas, and deprived of the VC in the pregnant females when the plugs were identified (E0.5). HPLC assays of maternal serum VC indicated that VC deprivation from E0.5 caused a significant decrease of serum VC by approximately 94% at E12.5 ~ E18.5, compared with VC supplementation at 4 g/L in the tap water (Supplemental Figure S5a). We next assessed the effects of VC deficiency on the fetal and placental development in midgestational stages (E12.5 and E14.5; each n = 6) through late-gestational stages (E16.5 and E18.5; each n = 8). Analyses of the placental and fetal numbers indicated that VC deprivation (VC-) in the pregnant Gulo−/− females did not cause a significant difference in the numbers of implanted embryos and placentas, however, VC deprivation robustly increased the numbers of atrophic and absorbed placentas and embryos by approximately 16%, 33%, and 50% at E14.5, E16.5, and E18.5, respectively, and had no significant effect at E12.5, as compared to VC supplementation (VC+) (Fig. 5a and b). Likewise, VC deprivation consistently decreased the weights of not only the viable placentas (data not shown) but also the embryos by approximately 10%, 22%, and 30% at E14.5, E16.5, and E18.5, respectively (Supplemental Figure S5b-d). Thus, VC deficiency fails to maintain the pregnancy in the Gulo−/− pups-bearing matrices.
Osteogenic disorder (ODS) rats (genotype od/od) lack L-gulonolactone oxidase and have a hereditary defect in VC biosynthesis [27]. ODS rats should be maintained with tap water containing at least 1 g/L of VC for their normal growth [27]. To confirm the role of VC in pregnant maintenance, we intercrossed the ODS rats and deprived of VC, when plugs were identified. Dissecting the placentas and pups found that though maternal VC deprivation did not cause a significant difference in the numbers of implanted embryos and placentas, VC deprivation significantly increased the numbers of atrophic and absorbed placentas and embryos by approximately 63% (each n = 6) at E14.5, as compared to VC supplementation (Supplemental Figure S6a). Since VC is critical for the synthesis of ovary and placental hormones that are essential for the pregnant maintenance [10, 11], we next determined the serum progesterone (P4) and estradiol (E2) levels in the pregnant ODS rats. Maternal VC supplementation and deprivation caused no significant difference of the serum P4 and E2 levels in the pregnant ODS rats at E14.5 (Supplemental Figure S6b and c). Thus, VC deficiency fails to maintain the pregnancy in ODS rats due not to the insufficiency of P4 and E2.
We next examined the Hand1 and p-JNK1/2 levels in the VC-deficient placentas. Western analyses indicated VC deficiency decreased the placental Hand1 levels by ~ 50% and ~ 90% at E12.5 and E16.5, respectively (Fig. 5c). In line with the western results, immunohistochemistry staining of E16.5 placental sections indicated that Hand1 globally expressed in the E16.5 placentas, and VC deficiency resulted in a globally robust decrease in Hand1 expression (Fig. 5d). In contrast, VC deficiency increased the placental p-JNK1/2 levels by ~ 104% and ~ 175% at E12.5 and E16.5, respectively (Fig. 5e). Thus, consistent with the in vitro results, VC deficiency not only increases the p-JNK1/2 levels but also decreases the Hand1 levels in murine placentas.
In contrast to Gulo−/− pups, Gulo+/− pups from the intercross of female Gulo−/− and male Gulo+/+ mice should be able to synthesize VC by themselves, even VC is deficient in maternal Gulo−/− mice. To confirm the importance of fetal but not maternal VC on the pregnant maintenance, we crossed the wild-type males with the Gulo−/− females to generate Gulo+/− pups and deprived the matrices of VC, when plugs were identified. Expectedly, VC deprivation decreased the maternal serum VC to approximately 9 µM (versus ~ 60 µM) and decreased the VC of Gulo+/− pups to approximately 52 µM (versus ~ 59 µM) at either E16.5 or E18.5, as compared to VC supplementation (Fig. 5f and g). In line with the VC levels in the Gulo+/− pups, VC deprivation in the Gulo+/− pups-bearing matrices caused neither a significant increase in the numbers of the atrophic and adsorbed placentas and embryos nor an apparent decrease in the placental Hand1 expression, as compared to VC supplementation (Fig. 5h and Supplemental Figure S5d). Thus, fetal but not maternal VC deficiency specifically causes the failure of placentation and pregnant maintenance.
VC deficiency disrupts the trophoblastic differentiation in the murine placentas
In order to examine the roles of VC in the placentation, we performed the H&E and immunohistochemistry staining, in situ hybridization, and transmission electron microscopic examination using the residual and viable Gulo−/− placentas at E16.5 and/or E18.5. H&E staining indicated the significant changes were rarely observed in the decidua basalis but the equivalent decreases in the size of placental mass, junctional zone, and labyrinthine layer were frequently observed in the placenta at either E16.5 or E18.5 (Fig. 6a and b). Although the labyrinthine layer of VC-supplemented placentas exhibited a well-organized vascular network consisting of maternal and fetal blood vessels (MBV and FBV), the labyrinthine layer of VC-deficient placentas showed a messy and strip-shaped vascular network with the dilated maternal blood lacunae (MBL), enlarged FBV spaces, and sparse trophoblastic layers that separated the MBL from FBV (Fig. 6a and b). Immunostaining of MBV and FBV using the respective antibodies against cytokeratin (CK) and laminin further revealed an increased density of MBV and FBV in the Gulo−/− placentas at E16.5 (Fig. 6c and d). Thus, VC deficiency in the Gulo−/− placentas causes the severe defects in the placental structures, which possibly contributes to the failure of pregnant maintenance.
We next perform the in situ hybridization or immunohistochemistry staining to detect the expression of trophoblastic markers in the Gulo−/− placentas. P450scc, a maker of P-TGCs, was robustly expressed in the E14.5 placentas supplemented with VC but rarely expressed in those deprived of VC (Fig. 6e). VC deficiency decreased the Tpbpα-positive areas by 59% in the junctional zone of placentas, as compared with VC supplement (Fig. 6f). Likewise, Plf-, Pl-II-, and Ctsq-positive areas in the labyrinth of placentas were significantly decreased by 47%, 48%, and 64%, as compared to in the VC-supplemented placentas, respectively (Fig. 6g-i), consistent with the in vitro findings (Fig. 1a-c). In addition, though VC deficiency did not apparently affect the labyrinth expression of SynA, a marker of ST-I trophoblasts, VC deficiency did markedly increase the labyrinth expression of SynB, a marker of ST-II trophoblasts, as compared to VC supplement (Fig. 6j and k). Finally, we examined the tri-laminar interhemal barrier (TIB) of placental labyrinth by TEM. The labyrinth from VC-supplemented placentas showed a structure of typical TIB consisting of an S-TGC layer and two tightly adhered syncytiotrophoblast layers (ST-I and ST-II) that lined the fetal endothelium and separated the MBL from FBV (Fig. 6i). The labyrinth from VC-deficient placentas showed an almost normal fetal endothelium with a continuous basement membrane surrounding ST-II, an atrophic S-TGC layer apparently lacking the cell protrusions, an expanded ST-I layer attached to S-TGC layer and filled with dilated cytoplasmic vacuoles that were considered to reduce the intracellular transport, and an enlarged ST-II layer with excessive lipid inclusions (Fig. 6i). Thus, VC deficiency significantly negates Hand1 protein and apparently causes the developmental defects of trophoblasts, especially TGC subtypes, in mouse placentas, which possibly contributes to the failure of pregnant maintenance.