Endoplasmic reticulum stress contributes to the decline in doublecortin expression in the immature neurons of mice with long-term obesity

Adult hippocampal neurogenesis (AHN) plays an important role in hippocampus-dependent function. The number of doublecortin (Dcx)-positive immature neurons in the dentate gyrus decreases over time, especially in the early stages of Alzheimer’s disease (AD), and is further reduced in later stages of AD. Obesity in midlife is associated with dementia later in life; however, the underlying mechanisms by which obesity results in the development of dementia later in life remain unknown. Here, we show that endoplasmic reticulum (ER) stress was activated in the hippocampus and processes of Dcx-expressing immature neurons were shortened, coexpressing CHOP in APP23 AD model mice with high-fat diet-induced long-term obesity and in aged Leprdb/db (db/db) mice. Moreover, in cells differentiating from hippocampal neurospheres, Dcx mRNA was rapidly degraded via a microRNA (miRNA) pathway after thapsigargin treatment in vitro. These results indicate that loss of Dcx mRNA induced by ER stress during AHN may cause memory impairment in obese individuals later in life.


Results
Behavior to novel object and swimming capacity were compromised in AD model mice with long-term obesity and diabetes model mice. We previously showed that the expression levels of ATF4 are increased in the cortex, hippocampus, and amygdala in the offspring of obese and diabetic APP23 AD model mice crossed with db/db mice 19 through a process known as the ISR 34 . ATF4 expression in the brain is decreased by quercetin, and this change is accompanied by an improvement in memory 19 . To examine the effect of ER stress on AHN in the brains of mice with long-term obesity, we first examined whether memory was impaired in HFD-induced obese mice. Obesity was induced by HFD feeding for 41 weeks in 66-week-old APP23 AD model mice ( Fig. 1a-d) and for 64 weeks in 74-week-old wild-type C57BL6 mice (Suppl. Fig. 2). The serum levels of fasting blood sugar (FBS) and insulin were increased in both groups of mice subjected to long-term HFD feeding compared with mice fed a standard diet ( Fig. 1b and Suppl. Fig. 2b). To examine whether long-term HFDinduced obesity affects memory, we performed the novel object location (NOL) test. Analysis of the ratio of the time spent exploring the target object to the time spent exploring all three objects revealed that the cognitive capacity of APP23 mice with long-term obesity was not significantly different from that of APP23 mice fed a standard diet. However, the time spent exploring the object at the novel location was significantly decreased in APP23 mice with long-term obesity, as they potentially exhibited neophobic behavior ( Fig. 1c and d). Next, we examined whether spatial memory was affected in eighteen-week-old Lepr db/db (db/db) mice, which had a higher body weight than wild-type and heterozygous littermates (control), even at 6 weeks of age (Suppl. Fig. 1a). In the MWM test, the latency to find the invisible platform during the acquisition phase was not changed and the number of platform crossings made by db/db mice was significantly decreased; these parameters were considerably affected by the differences in swimming speed and distance (Suppl. Fig. 1b). We next assessed aged wild- Figure 1. Memory was impaired in aged mice with long-term obesity. (a-d) APP23 mice (25 to 27 weeks old) were fed a standard chow diet (st) or a HFD (60% fat) for 43 weeks. The body weights (BWs) of APP23 (n = 7 standard diet-fed mice; n = 9 HFD-fed mice) (a) mice. The error bars represent the standard errors of the mean (SEMs). *p < 0.05. Statistical significance (p < 0.05) was determined by Student's t test. (b) Serum levels of FBS and insulin in standard diet-fed mice (n = 5) and HFD-fed mice (n = 8). Statistical significance (p < 0.05) was determined by Student's t test. (c) Performances of mice with long-term obesity and control mice in the NOL test. One of three objects was moved from the previous location (P) to a novel location (N). (d) Representative trajectories of mice exploring the objects (left panels). The total distance traveled (arbitrary units: A. U.) and the exploration indices in each zone at trials 3 and 4 are shown. Statistical significance (p < 0.05) was determined by Student's t test (total distance; index comparing between zones P and N) or one-way ANOVA followed by the Bonferroni post-hoc test [standard diet-fed APP23 mice: F(2,12) = 2.76, p = 0.10 (trial3), F(2,12) = 3.62, p = 0.06 (trial4); HFD-fed APP23 mice: F(2,24) = 0.54, p = 0.59 (trial3), F(2,24) = 2.07, p = 0.15 (trial4)]. Standard diet-fed mice (n = 5); HFD-fed mice (n = 9). www.nature.com/scientificreports/ HFD and the standard diet were 55.8 ± 9.7 g and 36.0 ± 3.9 g, respectively, at the age of 86 weeks. Western blot analysis showed that the expression levels of CHOP in the hippocampi of C57BL6 mice with long-term obesity were significantly increased compared with those in the hippocampi of mice fed the standard diet (Suppl. Fig. 3). Furthermore, the expression levels of CHOP in the hippocampi of db/db mice with long-term obesity were significantly higher than those in the hippocampi of control mice at 46 weeks of age (Fig. 2b). We also observed an increase in ATF4 expression in wild-type and db/db mice with long-term obesity ( Fig. 2b and Supp. Fig. 3). Since three signaling pathways, the PERK, ATF6, and IRE1 pathways, are activated during ER stress, we examined the  www.nature.com/scientificreports/ expression of Xbp-1s and ATF6 fragmentation in the hippocampi of 60-week-old db/db mice. We found that the N-terminal cleavage product of ATF6 and Xbp-1s were not detected in the hippocampi of aged db/db mice but were detected in mouse ES cells treated with 2 mM dithiothreitol (DTT) (Fig. 2c). Therefore, we examined the levels of Chop mRNA in the hippocampi of aged control and db/db mice. We found that Chop mRNA levels were not increased in db/db mice (Fig. 2d), suggesting that ATF6 is not involved in CHOP induction in aged db/db mice. Xbp-1s expression was not observed in aged db/db mice. These results suggested that long-term obesity induced ER stress, particularly the ISR, in the hippocampus.
CHOP and Dcx were coexpressed in the immature neurons of mice with long-term obesity. To investigate the impact of the ER stress on AHN, we performed immunostaining for Ki67, which is continuously produced from the S phase until the cell cycle exit 35 , in 10-week-old (Fig. 3a) and 45-week-old (Fig. 3b) db/db mice using coronal plane brain sections at a distance of −1.5 to −2.5 mm from bregma. The number of Ki67positive cells in the hippocampal DGs of db/db mice was similar to that in the DGs of control mice in both the young and aged groups (Fig. 3a, b). Additionally, the levels of mature BDNF in the hippocampi of 46-week-old control and db/db mice were similarly measured (Fig. 3c). The levels of Dcx in APP23 mice fed a HFD were significantly decreased compared with those in APP23 mice fed a standard diet (Fig. 2a), but the levels of Dcx did not differ between C57BL6 mice fed a HFD and C57BL6 mice fed a standard diet or between db/db mice and control mice ( Fig. 2b and Suppl. Fig. 3). Therefore, we performed immunohistochemistry to examine the expression of Dcx in the hippocampus. Since Dcx is sensitive to postmortem breakdown 36 , the mice were anesthetized and quickly perfused with 4% paraformaldehyde for fixation, and then the expression of Dcx in the hippocampi of control and db/db mice was then examined by immunohistochemistry. The number of Dcx-positive cells in the DG did not differ between 45-week-old control and db/db mice (Fig. 3d, f) or between 10-week-old control and db/db mice (data not shown). We observed reduction of Dcx expression in processes in z-stacks by confocal microscopy of Dcx-positive cells in the DG in 45-week-old db/db mice (Fig. 3e). Using the method described by Plumpe et al.for the characterization of Dcx-positive cells 37 , more cells with short processes were observed in 45-week-old db/db mice than in 45-week-old control mice ( Supplementary Fig. 4). These Dcx-positive cells in 45-week-old db/db mice expressed CHOP (Fig. 3g). Consistently, we observed that the Dcx-positive cells of APP23 mice with long-term obesity expressed CHOP (Fig. 3h). These results suggested that the ER stress was activated in the Dcx-positive immature neurons of mice with long-term obesity.
Thapsigargin reduced Dcx expression in immature neurons through Dcx mRNA degradation. To investigate the effect of the ER stress on the Dcx expression in immature neurons, we cultured neurospheres isolated from the mouse hippocampus according to a previously established protocol 38 . After neurospheres were cultured on laminin-coated dishes for 5 days in vitro (DIV) (Fig. 4a), several marker proteins, i.e., Dcx, calreticulin, Nestin, and βIII tubulin, were expressed ( Fig. 4c). Therefore, we challenged these cells with thapsigargin, a sarcoplasmic reticulum/ER Ca 2+ -ATPase inhibitor 39 , for 6 h and then cultured them in new medium to induce ER stress according to a previously published method 27 . The expression levels of Dcx corrected for GAPDH after thapsigargin treatment were much lower than those after DMSO treatment as determined by western blot (Fig. 4c); however, the differences were not significant as determined by Student's t test (p = 0.17). Consistently, immunostaining revealed that the processes in cells expressing both Dcx and CHOP was shortened (Fig. 4b), as was observed in vivo ( Fig. 3g, h). Interestingly, loss of Dcx protein expression was observed in thapsigargin-treated cells, but the levels of other marker proteins were not decreased (Fig. 4c). The proneural basic helix-loop-helix neurogenin 1 and 2 proteins directly activate Dcx expression by binding to the promoter and upregulating p35 expression 40 . p35 protein expression was also not decreased by thapsigargin treatment (Fig. 4c). To examine the involvement of the ubiquitin-proteasome system, autophagy-lysosomal pathway, or apoptosis in the loss of Dcx protein expression in thapsigargin-treated cells, we added MG132, E-64d/pepstatin A, or z-VAD-FMK to immature neurons and then treated them with thapsigargin. Since the protein levels of Dcx were not rescued by 25 µM MG132, 10 µg/mL E-64d and 10 µg/mL pepstatin A (data not shown), or 40 µM z-VAD-FMK (Fig. 4d), as determined by western blotting, we measured the levels of Dcx mRNA by semiquantitative reverse transcriptase polymerase chain reaction (RT-PCR) (Fig. 4e). We found that the mRNA levels of Dcx, but not Nestin or p35, quickly decreased during incubation in new medium after thapsigargin treatment (lower panel in Fig. 4e). The levels of Dcx mRNA were not rescued by the IRE1 inhibitor 4µ8C, indicating that Dcx mRNA is not a substrate for RIDD (Fig. 4f). Since the Dcx transcript has a long 3′ UTR, which binds the Musashi1 RNA-binding protein, mRNA stability and translation are possibly regulated by the binding proteins 11 . We found that the levels of Dcx mRNA were not rescued by zVAD (Fig. 4g), suggesting that caspase substrates are not involved in Dcx mRNA stability and that apoptosis does not decrease Dcx mRNA expression. Next, we knocked down Dicer in differentiating NSCs using a small hairpin RNA (shRNA) and small interfering RNA (siRNA) and then treated the cells with thapsigargin to examine whether the ER stress-induced reduction in Dcx mRNA expression was regulated by Dicer, as Dcx mRNA expression has been reported to be upregulated in Dicer-deficient NSCs 14 . Neurospheres cultured on laminin were infected with lentiviral particles carrying an shRNA targeting Dicer at 1 DIV, transfected with an siRNA targeting Dicer at 4 DIV, and then treated with thapsigargin at 5 DIV. Real-time RT-PCR showed that the thapsigargin-induced downregulation of Dcx mRNA expression was significantly prevented by the knockdown of Dicer in differentiating NSCs (Fig. 4h).
To identify the miRNAs upregulated during ER stress, we performed small RNA-seq on differentiating neurospheres. Analysis of the differential expression of microRNAs in thapsigargin treated immature neurons revealed that some microRNA, such as miR-148a-5p, miR-129b-3p, and miR-135a-2-3p, were significantly increased upon thapsigargin treatment (Fig. 5a). www.nature.com/scientificreports/    16 . However, it is unknown why it takes a long period of time for dementia to develop after body weight increases. In the present study, we demonstrated that long-term obesity induced ER stress, mainly via the ATF4-CHOP axis, in the hippocampal DG, leading to a decrease in the number of processes on Dcx-expressing immature neurons due specifically to the loss of Dcx mRNA stability. The ATF4-CHOP axis is initiated by eIF2α phosphorylation, which is mediated by four protein kinases, through a process called the ISR, which contributes to the pathogenesis of diseases, including cognitive disorders 29 . For example, PERK phosphorylation 43 and eIF2α phosphorylation 44 are observable in AD patients. Consistently, the gene expression of the ISR-related signaling molecule ATF4 is upregulated in the AD brain 25 . Since ATF4 directly binds to the promoter of the CHOP (also known as growth arrest and DNA damageinducible protein, Gadd153) gene 45 and CHOP induction is dependent on ATF4 expression, CHOP is expressed when the ISR is activated. CHOP was detected in the hippocampi of mice with long-term obesity as determined by western blotting (Fig. 2), and immunohistochemistry and confocal microscopy confirmed that CHOP was localized in Dcx-expressing immature neurons of the hippocampal DG (Fig. 3g, h). We observed significantly short processes on Dcx-positive immature neurons in the hippocampal DGs of mice with long-term obesity (Fig. 3). Our experiment showed that the pan-caspase inhibitor zVAD had no effect on the protein expression of Dcx or on Dcx mRNA stability under the ER stress condition (Fig. 4d, g). Further experiments are needed to explore the mechanisms of dendritic loss in Dcx-expressing immature neurons. ER stress is involved in the pathogeneses of obesity and diabetes 46,47 , which exacerbate cognitive dysfunction in mouse model of AD crossing db/db mice 48 . We did not observe the cleavage product of ATF6 or XBP-1s in 20 µg hippocampal lysates even after a longer exposure as determined by western blot (Fig. 2c). Although chop expression is regulated by the PERK 49 and ATF6 50 pathways through the binding of ATF4 45 and ATF6 51 to the amino acid-response element (AARE) and ER stress response element (ERSE) of the chop promoter, respectively, chop mRNA levels were not increased, and xbp-1s mRNA was not observed in the hippocampi of aged db/db mice (Fig. 2d), suggesting that ATF6 and Xbp-1 are not involved. The results presented herein demonstrated the potential activation of ER stress/ISR in the immature neurons of mice with long-term obesity. However, the mechanisms of ER stress/ISR activation in immature neurons remain unclear. Reactive oxygen species (ROS) are produced as byproducts of protein oxidation in the ER through ER oxidoreductin-1 (ERO1) and protein disulfide isomerase (PDI) 52 . ERO1α transcription is increased by CHOP 53 . Overexpression of ERO1β causes ER stress in pancreatic β cells 54 . ER stress is also elicited by ROS 55 and by the inactivation of PDI through nitrosylation 56 . In the C17.2 NSC line, oxidative stress induced by high-glucose medium activates ER stress and prevents Tuj1 and glial fibrillary acidic protein (GFAP) expression, which is rescued by the superoxide dismutase mimetic tempol and ER stress inhibitor 4-phenylbutyrate 57 . Thus, reduction-oxidation (redox) reactions and ER stress stimulate each other 58 . Although we do not know which occurs first in the brains of obese mice, we hypothesize that oxidative stress induces ER stress/ISR in Dcx-expressing immature neurons in vivo. To confirm this hypothesis, further studies are needed; for example, in vivo dynamic nuclear polarization magnetic resonance imaging (DNP-MRI) 59 may allow detection of redox conditions in the brains of young and aged obese mice.
Differentiating NSCs were treated with thapsigargin for 6 h, washed, and cultured in medium without thapsigargin. ER stress was activated in these cells, as indicated by the induction of 78-kDa glucose-regulated protein (Grp78)/immunoglobulin heavy chain-binding protein (BiP) and CHOP expression. The Dcx protein was absent in thapsigargin-treated cell lysates, although the proteins Nestin and βIII Tubulin were present www.nature.com/scientificreports/ (Fig. 4c). Interestingly, we found that loss of the Dcx protein induced by ER stress was mainly caused by the rapid elimination of Dcx mRNA (Fig. 4e) because the protein expression of Dcx was not rescued by inhibitors of the proteasome, autophagy, and caspase. Dcx is a microtubule-associated protein that leads to microtubule polymerization 60,61 , and its mRNA is not a substrate for RIDD (Fig. 4f), probably because Dcx mRNA is not an ER-localized mRNA. The Dcx transcript has a long 3′ UTR 11 , which binds Musashi1 62 and miRNAs 13,14,[63][64][65][66] to repress Dcx mRNA translation or to regulate Dcx mRNA stability. The mRNA levels of Dcx were decreased in differentiating NSCs at 6 h after thapsigargin treatment for 6 h (Fig. 4e), indicating ER stress induced the loss of Dcx mRNA. The mRNA expression of Dcx in differentiating NSCs was significantly rescued by Dicer knockdown after thapsigargin treatment (Fig. 4h). Consistently, the Dcx transcript level is increased in unstressed adult NSCs from Dicer knockout mice 14 . The Dcx 3′ UTR is targeted by miR-128, resulting in the downregulation of Dcx protein expression in SH-SY5Y neuroblastoma cells 13 and in Neuro2A cells 14 , but the mechanism by which Dcx mRNA is regulated is controversial 13,14 . Moreover, it may be interesting to investigate the serum levels of miR-128 in individuals with preclinical AD or MCI and long-term obesity since the circulatory levels of miR-128 are significantly increased in patients with type 2 diabetes and depression compared to those in patients with type 2 diabetes 67 . It has been suggested that miRNA biogenesis is regulated upon ER stress 68 71 . We found that the expression levels of miR-148a-5p, miR-129b-3p, and miR-135a-2-3p were significantly increased [log 2 fold change > 1, false discovery rate (FDR) < 0.05] in differentiating neurospheres under the condition of ER stress (Fig. 5a). These evidences and our results demonstrate that ER stress may stimulate a miRNA pathway to regulate Dcx mRNA stability, resulting in the loss of Dcx in the immature neurons of the hippocampal DG (Fig. 5b). Although Dcx is predicted to be a target of miR-129b-3p by miRDB 72 , more experiments are needed to explore the mechanism of Dcx mRNA degradation.
In conclusion, ER stress and the ISR are activated in the hippocampal DG immature neurons of mice with long-term obesity. Loss of Dcx mRNA by ER stress/ISR during AHN may underlie the memory impairment that occurs later in the lives of obese subjects.

Methods
Western blot analysis and chemicals. Western blotting was performed as described previously 32 . Behavioral tests. The NOL test was performed according to the method described by Roy et al. 74 . Briefly, each mouse was habituated to a cage without objects for 15 min on day 1. The mice were exposed to three different objects, i.e., conical (diameter x height: 5 × 11.5 cm), cylindrical (6.5 × 10.5 cm), and reagent (5 × 13.2 cm) bottles, for 5 min three times at 2-min intervals on day 2. The mice were placed in their home cages for a retention interval (two minutes). Then, one of the objects was moved to the opposite corner. The behavior of each mouse was monitored using video recording software and an automated tracking system (SMART v3.0 software; Panlab, Barcelona, Spain) and a video camera (HDC-HS350; Panasonic; Osaka, Japan). Three zones (trial3: 3, 4, P; trial4: 3, 4, N) were set at a distance of 4 cm from the object in the far corners of the arena. The exploration www.nature.com/scientificreports/ index was calculated as the time spent exploring the object placed at the previous location (P), at a novel location (N), at a zone 3 or at a zone 4 divided by the total time spent in three zones [index P: P/(3 + 4 + P); index N: N/(3 + 4 + N); index 3: 3/(3 + 4 + P or N); index 4: 4/(3 + 4 + P or N)]. When the distance between the nose of the mouse and an object was less than 2 cm or when the mouse sniffed or touched the object with its snout, the mouse was considered to be exploring the object, as previously described 75 . The cages and objects were cleaned with 70% ethanol and 1% acetic acid solution before each trial to eliminate dominant odors. The MWM test was performed according to a previously described protocol 76 . Briefly, the apparatus was a 100-cm diameter tank containing water at a temperature of approximately 22 °C water, skim milk and a submerged platform. Four acquisition trials from each of the five starting positions were performed each day. The time limit for each trial was 60 s. Mice that did not reach the platform were guided to the platform and left there for 30 s. On day 5, the platform was removed for the probe test. Each trial was recorded and analyzed with the SMART v3.0 automated tracking system.
Immunostaining. Immunostaining was performed as described previously 27 and analyzed by fluorescence microscopy (BZ-9000, Keyence; Osaka, Japan) and confocal microscopy (LSM710, Carl Zeiss; Göttingen, Germany). Briefly, mice were anesthetized and perfused with phosphate-buffered saline (PBS) and fixed with 4% paraformaldehyde in 0.1 M phosphate buffer (PB). The mouse brains were postfixed for 2 h in the same fixative, which was then replaced with 15% sucrose in 0.1 M PB. The brains were embedded into the optimal critical temperature (OCT) compound (Sakura Finetech USA, Inc., Torrance, CA), and 14-µm-thick sections (coronal plane at a distance of −1.5 to −2.5 mm from bregma: The Mouse Brain in Stereotaxic Coordinates, second edition, Academic Press) were obtained with a cryostat (HM 550, Carl Zeiss). The secttions (14 µm) were incubated in PBS supplemented with 10% normal goat serum (Jackson ImmunoResearch Laboratories, Inc., West Grove, PA) and 0.1% Triton X-100 at room temperature for 1 h. The sections were then incubated with an anti-Dcx antibody and an anti-CHOP antibody or an anti-Ki67 antibody (Thermo Fisher Scientific) in PBS containing 1% normal goat serum and 0.1% Triton X-100 at 4 °C for 12 h. For the detection of Ki67, the sections were incubated in 10 mM sodium citrate (pH 6) at 80 °C for 30 min and cooled to room temperature before the antibody was added. The sections were incubated with Alexa Fluor 488-conjugated anti-rabbit IgG (H + L) and Alexa Fluor 546-conjugated anti-mouse IgG (H + L) antibodies (Thermo Fisher Scientific) and Hoechst [1 µg/ mL bis-benzimide (Sigma-Aldrich Co. LLC)] to detect fluorescence signals and nuclei. The fluorescence intensity profiles were analyzed with Zen software (Carl Zeiss). The count of Ki67-positiveand Dcx-positive cells were counted by two people using images obtained by fluorescence microscopy (BZ-9000) and confocal microscopy (LSM710), respectively.
Cell culture and ER stress treatment. Neurospheres were cultured and isolated from the hippocampi of 10-day-old C57BL/6J mice according to a previously described protocol 38 . Neurospheres were grown in neurobasal medium supplemented with B-27 without vitamin A, 20 ng/mL basic FGF, 20 ng/mL EGF, GlutaMAX, and gentamicin (Thermo Fisher Scientific). For differentiation, the cells were plated on dishes coated with natural mouse laminin (Thermo Fisher Scientific); the next day, the medium was replaced with medium A, which was comprised of Dulbecco's modified Eagle's medium and Ham's F-12 (DMEM/F12, Wako Pure Chemical Industries, Ltd.), MACS® NeuroBrew®-21 (Miltenyi Biotec, Bergisch Gladbach, Germany), 0.5 × N-2 supplement (Thermo Fisher Scientific), 20 ng/mL basic FGF, and gentamicin. On the following days, the medium A was changed in the morning and replaced with medium B, which consisted of DMEM/F12, MACS® NeuroBrew®-21, 0.5 × N-2 supplement, and gentamicin, in the afternoon. Then, the medium was changed every two days. The cells were maintained at 37 °C in an atmosphere containing 5% CO 2 . Thapsigargin was added to the medium at a final concentration of 0.23-0.69 µM, after which the cells were incubated for 6 h and washed with medium. The medium was replaced with new medium, and the cells were incubated for the indicated amounts of times. Knockdown of Dicer was induced by an shRNA-expressing lentivirus (sc-4090-V, Santa Cruz) and an siRNA (s101206, Thermo Fisher Scientific). Differentiating cells were infected with a lentivirus expressing an shRNA targeting Dicer or a control shRNA (sc-108080, Santa Cruz) with 5 µg/mL polybrene at 1 DIV and then transfected with an siRNA targeting Dicer or a control siRNAs at 4 DIV using TransIT-X2 (Takara Bio Inc., Shiga, Japan). The following day, the cells were treated with thapsigargin. The sequences of the siRNAs targeting Dicer were as follows: 5′-GCC GAU CUC UAA UUA CGU Att-3′ and 5′-UAC GUA AUU AGA GAG AUC GGCgc-3′. For immunostaining, the cells were fixed with 4% paraformaldehyde in 0.1 M PB 4 °C for 15 min and then incubated in PBS containing 10% normal goat serum and 0.3% Triton X-100 at room temperature for 1 h. The cells were incubated with an anti-Dcx antibody, an anti-CHOP antibody and an anti-GFAP antibody (Thermo Fisher Scientific) in PBS containing 1% normal goat serum and 0.1% Triton X-100 at 4 °C for 12 h. The cells were then incubated with Alexa Fluor 488-conjugated anti-rabbit IgG (H + L) and Alexa Fluor 546-conjugated anti-mouse IgG (H + L) and Alexa Fluor 647-conjugated anti-rat IgG (H + L) antibodies (Thermo Fisher Scientific).
Mouse embryonic stem (ES) cells were grown in DMEM supplemented with 15% normal bovine serum, MEM nonessential amino acids (Thermo Fisher Scientific), 0.1 mM 2-mercaptoethanol (Sigma-Aldrich Co., LLC.), and 1 mM sodium pyruvate (Sigma-Aldrich Co., LLC.). Mouse ES cells were treated with 2 mM DTT for 1 or 30 min, and cells were then lysed by RIPA for western blot analysis.
Small RNA sequencing. Total RNA was purified using a miRNeasy mini kit (QIAGEN, Hilden, Germany) from differentiating neurospheres cultured for 6 h after treatment with DMSO or 0.23 µM of thapsigargin for 6 h. Six small RNA-seq libraries (DMSO: n = 3; thapsigargin: n = 3) were generated from the purified total RNA using NEB Next® Multiplex Small RNA Library Prep Set for Illumina® (Set 1) (New England Biolabs, Ipswich, MA) according to the manufacturer's instructions. Fifty bp single-end sequencing was performed with NovaSeq www.nature.com/scientificreports/