c-Jun NH₂-Terminal Kinase 1/2 and Endoplasmic Reticulum Stress as Interdependent and Reciprocal Causation in Diabetic Embryopathy

C57BL/6J mice, JNK1 heterozygous (jnk1^+/−), and JNK2 knockout (JNK2KO) mice in C57BL/6J background were purchased from The Jackson Laboratory (Bar Harbor, ME). Streptozotocin (STZ) from Sigma-Aldrich (St. Louis, MO) was dissolved in sterile 0.1 mol/L citrate buffer (pH 4.5). Sustained-release insulin pellets were purchased from Linplant (Linshin, Canada). Plasma insulin levels were assessed by the Rat/Mouse Insulin ELISA kit (Millipore, Billerica, MA; catalog number EZRMI-13 K).

The procedures for animal use were approved by the Institutional Animal Care and Use Committee. For decades, we (21,23,24) and others (7,25–27) have used an internationally accepted rodent model of STZ-induced diabetes in research of diabetic embryopathy. Ten-week old female wild-type (WT), JNK2KO mice, and jnk1^+/− mice were intravenously injected daily with 75 mg/kg STZ over 2 days to induce diabetes. Blood glucose levels were monitored daily by tail vein puncture and using the FreeStyle Blood Glucose Monitoring System (TheraSense; Abbot, Alameda, CA).

Diabetes was defined as 12-h fasting blood glucose levels ≥250 mg/dL, which normally occurred at 3–5 days after STZ injections. Based on a published report (28), male and female mice were paired at 3:00 p.m., and day 0.5 (E0.5) of pregnancy was established by the presence of the vaginal plug in the next morning (8:00 a.m.). On day 5.5 of pregnancy (E5.5), insulin pellets were removed to permit frank hyperglycemia (>250 mg/dL glucose level), so that the developing embryos would be exposed to a hyperglycemic environment during the critical period of closure of the neural tube (neurulation) (E8.0–10.5). Therefore, our mouse model of diabetic embryopathy specifically impacts neurulation and the period of induction of NTD. Based on our extensive studies (18,21,29–31), insulin treatment from E0.5–5.5 is essential for successful embryonic implantation and thus prevents early embryonic lethality (resorption) caused by hyperglycemic exposure at early embryonic stages (≤E5.5). WT, nondiabetic female mice with vehicle injections and sham operation of insulin pellet implants served as nondiabetic controls. On E8.75 (2:00 p.m. at E8.5), mice were killed, and conceptuses were dissected out of the uteri for analysis. Data on NTD incidences from jnk2 gene-deleted embryos were not collected, because they have been published elsewhere (21). We have extensively characterized our models in the C57BL/6J background by including surgical (anesthesia) controls and insulin-treated diabetic controls (6,21). The half-life of STZ is only 30 min, and STZ was injected 1 to 2 weeks before pregnancies were established. No residue of toxic effect of STZ on embryonic development in our model was observed in diabetic mice with continuous insulin treatment (6). In our studies, sham operation simulating removal of insulin pellets does not increase the incidence of NTD in nondiabetic mice (6). Consistent with our findings, two independent publications from other groups demonstrate the same finding in the C57BL/6J strain that maternal hyperglycemia induces >22% NTD (26,27).

Western blotting was performed as previously described (21). Briefly, E8.75 embryos from different experimental groups were sonicated in 80 µL ice-cold lysis buffer (20 mmol/L tris[hydroxymethyl]aminomethane-HCl [pH 7.5], 150 mmol/L NaCl, 1 mmol/L ethylenediaminetetraacetic acid, 10 mmol/L NaF, 2 mmol/L Na-orthovanadate, 1 mmol/L phenylmethylsulfonyl fluoride, and 1% Triton X-100) containing a protease inhibitor cocktail (Sigma-Aldrich). Equal amounts of protein and the Precision Plus Protein Standards (Bio-Rad) were resolved by SDS-PAGE gels (12% gel for caspase 3 and 8 and 10% gel for all other proteins) and transferred onto Immobilon-P or Immobilon-P^SQ (for cleaved caspase) membranes (Millipore). To prevent nonspecific bindings, all membranes were incubated with 5% nonfat milk in PBS (pH 7.0) for 45 min and then were incubated for 18 h at 4°C with the following primary antibodies at 1:1,000 to 1:2,000 dilutions in 5% nonfat milk: BiP, calnexin, phosphorylated (p-)eIF2α, p-PERK, CHOP, p-JNK1/2, JNK2, JNK1, p–Elk-1 (Ser383), p–c-Jun (Ser63), p-activating transcription factor (ATF) 2 (Thr71), and p-Foxo3a (Thr32) (Cell Signaling Technology, Beverly, MA); anticaspase 3 (Chemicon International, Billerica, MA); and rat anticaspase 8 (mouse specific) (Alexis Biochemicals, San Diego, CA), which detects both procaspase 8 and cleaved caspase 8. Membranes were exposed to goat anti-rabbit or anti-mouse (Jackson ImmunoResearch Laboratories, West Grove, PA) secondary antibodies. To ensure that equivalent amounts of protein were loaded among samples, membranes were stripped and probed with a mouse antibody against β-actin (Abcam, Cambridge, MA). Signals were detected using an Amersham ECL Advance Detection Kit (GE Healthcare, Piscataway, NJ) or a Supersignal West Pico Chemiliminescent Substrate (Thermo Scientific), and chemiluminescence emitted from the bands was directly captured using a UVP Bioimage EC3 system (UVP, Upland, CA). Densitometric analysis of chemiluminescence signals was performed by VisionWorks LS software (UVP). All experiments were repeated three times with the use of independently prepared tissue lysates.

mRNA was reverse-transcribed to cDNA, which was used for PCR. The PCR primers were: forward, 5′-AAACAGAGTAGCAGCGCAGACTGC-3′ and reverse, 5′-TCCTTCTGGGTAGACCTCTGGGAG-3′. If no XBP1 mRNA splicing occurred, one band at 205 bp was produced. Two bands, 205 and 179 bp (a main band), were present in PCR products when XBP1 mRNA was spliced.

The procedure of whole-embryo culture has been previously described elsewhere (21,32). C57BL/6J mice were paired overnight. The next morning was designated embryonic day E0.5 if a vaginal plug was present. Mouse embryos at E8.5 were dissected out of the uteri in PBS (Invitrogen, La Jolla, CA). The parietal yolk sac was removed using a pair of fine forceps, and the visceral yolk sac was left intact. Embryos (four per bottle) were cultured in 4 mL rat serum at 38°C in 30 revolutions/min rotation in the roller bottle system. The culture bottles were gassed with 5% O₂/5% CO₂/90% N₂ for the first 24 h and 10% O₂/5% CO₂/85% N₂ for the last 12 h. Embryos were cultured for 12 h (for Western blots) or 36 h (for NTD examination) under 5 mmol/L glucose, a value close to the blood glucose level of nondiabetic mice, or 20 mmol/L glucose, which is equivalent to the blood glucose level of diabetic mice, in the presence or absence of 1 or 2 mmol/L 4-PBA (Sigma-Aldrich). The range of 4-PBA concentrations in cell-culture studies is from 0.5–20 mmol/L (33). We started our whole-embryo culture experiments using 1 and 2 mmol/L 4-PBA and found that 2 mmol 4-PBA completely inhibited high glucose–induced NTD. Therefore, we used 2 mmol/L 4-PBA in our subsequent mechanistic studies. Because treatment of diabetic mice with 4-PBA results in normalization of hyperglycemia (33), in vivo 4-PBA administration is not suitable for assessing the direct effect of 4-PBA on maternal diabetes–induced NTD. At the end of 36-h culture, embryos were dissected from the yolk sac and examined under a Leica MZ16F stereomicroscope (Leica Microsystems, Bannockburn, IL) to identify embryonic malformations. Images of embryos were captured by a DFC420 5 MPix digital camera with software (Leica Microsystems). Normal embryos were classified as possessing a completely closed neural tube and no evidence of other malformations. Malformed embryos were classified as showing evidence of failed closure of the neural tube or NTD. The open neural tube structures of NTD embryos were verified by histological sections. Because embryos were only exposed to high glucose during a period of neurulation, NTD were the only structural defects detected in this study. Other malformations such as cardiac abnormalities, which were developed at the late embryonic stage (E15.5), were not within the scope of our study.

Abnormal ER structures were examined by transmission EM, which was done in our EM core facility. Thick sections (1 μm) were cut and visualized at ×100 and ×200 original magnifications to identify the neuroepithelia of E8.75 embryos. Thin sections (80 nm) of identified neuroepithelia were cut and viewed with an electron microscope (Jeol JEM-1200EX; Jeol, Tokyo, Japan) under high-power resolution (10,000, 12,000, and 25,000) for identification of abnormal ER structures.

Total RNA was isolated from E8.75 embryos using an RNeasy Mini Kit (Qiagen, Valencia, CA). Real-time PCR for BiP (HSPA5), calnexin, eukaryotic translation initiation factor 2α kinase 3 (eIF2ak3), eIF2a ribosome biogenesis regulator homolog (S. cerevisiae) (eIF2as1), protein disulfide isomerase family A, member 3 (PDIA3), IRE1α (ern1), and β-actin were performed using ABI TaqMan Gene Expression Assays (Applied Biosystems, Foster City, CA). RNA was reverse transcribed by using the high-capacity cDNA archive kit (Applied Biosystems). Real-time PCR and subsequent calculations were performed by the StepOnePlus Real-time PCR system (Applied Biosystems), which detected the signal emitted from fluorogenic probes during PCR.

The transferase-mediated dUTP nick-end labeling (TUNEL) assay was performed using the In Situ Cell Death Detection kit (Millipore). Five-micrometer serial coronal sections through the anterior neural tube were fixed with 4% paraformaldehyde in PBS and incubated with TUNEL reaction agents with or without Sox1 immunostaining. TUNEL-positive cells in the neural tube area of each section were counted. Based on the counting of total cell nuclei, the sizes of the neural tube areas of each section were relatively constant. Thus, apoptotic cell numbers were expressed as total TUNEL-positive cells per neural tube area. TUNEL assays were performed without the observer knowing the experimental groups.

Data are presented as means ± SE. One-way ANOVA was performed using SigmaStat 3.5 software, and a Tukey test was used to estimate the significance. Statistical significance was accepted at P < 0.05. Significant difference between groups in malformation incidences was analyzed by χ² test.

To determine whether maternal diabetes induces ER stress, levels of ER stress markers, BiP, p-PERK, and CHOP were analyzed in E8.75 embryos, a critical time point of neurulation, from nondiabetic and diabetic dams. Levels of BiP, p-PERK, CHOP, and p-JNK1/2 were significantly higher in embryos from diabetic dams than those from nondiabetic dams (Fig. 1A–D). XBP1 mRNA splicing, another ER stress indicator, was analyzed in RNA samples by PCR. Embryos exposed to maternal diabetes exhibited robust splicing manifested with two bands at 205 and 179 bp of the PCR products, whereas embryos from nondiabetic dams showed no XBP1 splicing with only one band (205 bp) of the PCR products (Fig. 1E). Because the main goal of our study was to analyze embryos during neurulation and examine NTD as the ultimate outcome, we determined whether maternal diabetes-induced ER stress mainly occurred in cells of the developing neuroepithelium. EM was employed to examine the swollen/enlarged ER lumens, which are direct morphological evidence of ER stress. Swollen/enlarged ER lumens were consistently present in neuroepithelial cells in E8.75 embryos exposed to maternal diabetes (Fig. 1F). In contrast, neuroepithelial cells in five embryos from five nondiabetic dams did not show any abnormal ER lumen structure (Fig. 1F). In our model, insulin pellets effectively restored serum insulin levels and euglycemia in diabetic mice (Supplementary Table 1). Because of the relatively short duration of our experiments, diabetes did not significantly affect body weight (Supplementary Table 2).

Because either jnk1 or jnk2 gene deletion significantly reduce maternal diabetes-induced JNK1/2 activation (Figs. 2A and 3A), E8.75 jnk1^−/− embryos from jnk1^+/− mothers mated with jnk1^+/− males, and jnk2^−/− embryos from JNK2KO mice were used to determine the causal role of JNK1/2 activation in ER stress. We found that jnk1^−/− embryos exposed to hyperglycemia had significantly lower levels of ER stress markers, BiP, p-PERK, CHOP, and p-eIF2α, than those in WT embryos from the same group of diabetic jnk1^+/− mothers (Fig. 2B–E). Similarly, jnk2^−/− embryos from diabetic JNK2KO mice had similar levels of ER stress markers, calnexin, BiP, CHOP, and p-eIF2α, as those in embryos of nondiabetic WT dams, which were significantly lower than those in embryos from diabetic WT mice (Fig. 3A–D). Moreover, hyperglycemia-induced XBP1 splicing was diminished in jnk1^−/− and jnk2^−/− embryos (Fig. 3E).

ER stress increases an array of ER chaperone gene expression. To determine whether JNK1/2 deficiency caused by jnk1 or jnk2 gene abrogates maternal diabetes-induced ER chaperone gene expression, mRNA levels of six ER chaperone genes were determined. Maternal diabetes significantly increased the expression of BiP, calnexin, eIF2ak3, eIF2as1, and PDIA3 (Fig. 4A–E). Deletion of either jnk1 or jnk2 gene inhibited maternal diabetes-increased expression of BiP, calnexin, eIF2ak3, eIF2as1, and PDIA3 to the levels in the nondiabetic WT group (Fig. 4A–E). Maternal diabetes did not increase IRE1α expression (Fig. 4F). However, either jnk1 or jnk2 gene deletion was even able to reduce IRE1α expression (Fig. 4F).

Because the ER chemical chaperone, 4-PBA, has been shown to block ER stress in other systems, we used 4-PBA to determine whether it could block high glucose-induced ER stress in cultured embryos. Treatment of cultured embryos with 2 mmol/L 4-PBA suppressed high glucose-increased ER stress markers, BiP, CHOP, and p-eIF2α (Fig. 5A–C). To determine whether ER stress is a causal event leading to NTD, NTD was assessed in cultured embryos with or without 1 mmol/L or 2 mmol/L 4-PBA. The NTD rate in high glucose group was 66.7%, which was significantly higher than those in the 5 mmol/L glucose group (8.0%), the high glucose plus 1 mmol/L 4-PBA group (10.0%), and the high glucose plus 2 mmol/L 4-PBA group (8.3%) (Table 1 and Fig. 5D). The NTD rates in the high glucose plus 1 mmol/L 4-PBA group and the high glucose plus 2 mmol/L 4-PBA group were not significantly different than that in the 5 mmol/L glucose group (Table 1 and Fig. 5D). 4-PBA treatment even reduced NTD rates under 5 mmol/L glucose conditions (Table 1).

To determine whether 4-PBA treatment suppresses the JNK1/2 pathway, we assessed phosphorylated levels of JNK1/2 and its downstream effectors in cultured embryos treated with 2 mmol/L 4-PBA. 4-PBA treatment completely blocked high glucose-induced phosphorylation of JNK1/2, c-Jun, Elk1, and ATF-2 (Fig. 6A–D). Foxo3a activation depends on its dephosphorylation. High glucose induced Foxo3a dephosphorylation whereas 4-PBA treatment abrogated high glucose-induced Foxo3a dephosphorylation (Fig. 6E).

Because JNK1/2 activation leads to caspase activation and apoptosis and 4-PBA suppresses JNK1/2 activation, we assessed the effect of 4-PBA on high glucose-induced caspase activation and neuroepithelial cell apoptosis. High glucose induced robust cleavage of caspase 3 and 8 (Fig. 7A). 4-PBA treatment abolished high glucose-induced caspase 3 and 8 cleavage (Fig. 7A). There were significant higher numbers of apoptotic neuroepithelial cells in embryos exposed to high glucose than cells under 5 mmol/L glucose conditions and high glucose plus 4-PBA conditions (Fig. 7B and C). Apoptotic cell numbers in the 5 mmol/L glucose group and the high glucose plus 4-PBA group were comparable (Fig. 7B and C). High glucose-induced apoptotic cells were Sox1 positive neural progenitors (Fig. 7D).

We have demonstrated in this series of experiments the reciprocal causative relationship between JNK1/2 activation and ER stress in diabetic embryopathy. Using the ER stress chemical inhibitor, 4-PBA, we determined that high glucose in vitro-induced ER stress mediates the teratogenicity of hyperglycemia through induction of caspase activation and apoptosis. Deletion of either jnk1 or jnk2 gene abrogates maternal diabetes-induced ER chaperone gene expression and consequent ER stress. Therefore, the current study provides mechanistic evidence for the interdependent relationship between JNK1/2 activation and ER stress and for the potential use of ER stress chemical inhibitors, such as 4-PBA, to prevent maternal diabetes-induced embryonic malformations.

Our previous study (11) using EM demonstrated aberrant maturational and cytoarchitectural changes in embryonic cells exposed to high glucose. However, this previous study did not specify which cellular organelle is impacted by high glucose. By assessing total cellular levels of ER stress markers and examining swollen/enlarged ER by EM, the current study clearly reveals that ERs in neuroepithelial cells of the developing neural tube are adversely impacted by maternal diabetes. Three pathways—the IRE1–XBP-1 pathway, the PERK–eIF2α pathway, and the ATF-6–CHOP pathway—collectively constitute the ER-specific UPR (13). The key intermediates of these three pathways can be used as indices of ER stress. Our study demonstrates increased expression of BiP, CHOP, XBP1 splicing, PERK, and eIF2α phosphorylation, supporting the conclusion that hyperglycemia-induced ER stress activates all three UPR responses. Activation of these UPR pathways, particularly the activation of IRE1–XBP1 and CHOP, triggers apoptosis in cells under pathophysiologic conditions (12). These ER stress–induced apoptotic effects are in line with the central apoptotic mechanism in the induction of diabetic embryopathy (5,6). Indeed, our subsequent studies show that ER stress inhibitor, 4-PBA, blocks apoptosis in cells of the developing neural tube and thus ameliorates high glucose–induced NTD.

ER stress induces an array of proapoptotic signaling, which centers on JNK1/2 activation (22,34). We show that both ER stress and JNK1/2 activation are present in diabetic embryopathy. Because 4-PBA has been shown to attenuate diabetes-induced ER stress (33), in the current study, it was used to test whether amelioration of ER stress can block high glucose–induced proapoptotic signaling. 4-PBA treatment blocks high glucose–induced activation of JNK1/2 and its downstream effectors. These findings strongly support that ER stress is directly linked to or enhances the activation of the proapoptotic JNK1/2 pathway, which leads to caspase activation and apoptosis in diabetic embryopathy.

ER stress also induces oxidative stress (35), which may directly mediate the proapoptotic effect of high glucose on neuroepithelial cells. It is known that oxidative stress increases the expression of Bax (36), a proapoptotic Bcl-2 family member, and suppresses the expression of the antiapoptotic protein, Bcl-2 (37). Oxidative stress plays a central role in the pathogenesis of diabetic embryopathy, which also leads to Bax upregulation and Bcl-2 downregulation (38). Thus, it is possible that high glucose–induced ER stress results in oxidative stress, which, in turn, causes neuroepithelial cell apoptosis via altering the expression of Bcl-2 family members.

Previous studies have demonstrated that high glucose in vitro can rapidly activate JNK1/2 (6,31). In addition, deletion of either jnk1 or jnk2 (21) significantly reduces maternal diabetes–induced NTD. These findings, which suggest that JNK1/2 activation may occur before cells undergo ER stress, prompted us to examine whether JNK1/2 activation causes ER stress. JNK1/2 deficiency caused by either jnk1 or jnk2 gene deletion abrogates maternal diabetes–increased ER stress markers and ER chaperone gene expression, supporting the conclusion that JNK1/2 activation triggers the UPR response and subsequently induces ER stress.

Our study presents for the first time strong evidence that JNK1/2 activation leads to ER stress. In future studies, it will be of interest to delineate the mechanisms underlying JNK1/2-induced ER stress. Factors that disrupt ER function and thus induce ER stress include accumulation of protein aggregates, oxidative stress, and alterations in calcium stores in the ER lumen (34). Because JNK1/2 activation can directly increase reactive oxygen species production (39), JNK1/2 may trigger ER stress via oxidative stress. JNK1/2 activation also contributes to the formation of intracellular protein aggregates (40), which may result in ER stress. JNK1/2 activates Bid (41), a proapoptotic Bcl-2 family member, which controls ER calcium homeostasis (42). Therefore, it is possible that JNK1/2 induces ER stress by disrupting ER calcium homeostasis. The mechanisms underlying JNK1/2-induced ER stress are likely multifactorial.

JNK1^+/− and JNK2^−/− mice are viable, normal in size, do not display any gross physical or behavioral abnormalities, and display normal embryonic development (43). Indeed, embryos from either nondiabetic JNK1^+/− or nondiabetic JNK2^−/− dams are morphologically indistinct from those from nondiabetic WT mice (21,43,44). We have successfully established the causal link between JNK1/2 activation and ER stress by using jnk2^−/− and jnk1^+/− mice. Using a jnk1^+/− and jnk1^+/− mating scheme to produce JNK1-null embryos avoids any potential maternal influences due to gene deletion. However, any potential maternal influence due to jnk2 gene deletion in diabetic embryopathy is negligible because diabetic JNK2KO dams have similar high glucose levels as those in diabetic WT mice (21). A recent published report (26) also used homozygous knockout mating in diabetes-induced NTD. Therefore, the JNK2 homozygous mating scheme is a suitable approach in our study.

Either jnk1 or jnk2 gene deletion blocks maternal diabetes–induced ER chaperone gene expression and consequent ER stress. Mitigating ER stress by 4-PBA treatment, in turn, blocks JNK1/2 activation, its downstream apoptotic signaling, and NTD. Our study establishes the reciprocal and interdependent link between the two causative events, JNK1/2 activation and ER stress, in diabetic embryopathy.

This study is supported by National Institutes of Health grants R01-DK-083243 and R01-DK-083770.

No potential conflicts of interest relevant to this article were reported.

X.L. and C.X researched data and approved the final version of the manuscript. P.Y. conceived the project, designed the experiments, researched data, and wrote and approved the final version of the manuscript. P.Y. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

The authors thank Dr. E. Albert Reece at the University of Maryland School of Medicine for editing the manuscript and Ms. Hua Li at the University of Maryland School of Medicine for technical support.