Camptothecin (CPT) was purchased from Biomol Research Laboratories (Plymouth Meeting, PA). Etoposide (also known as VP-16) was purchased from Nippon Kayaku (Tokyo, Japan). Adriamycin (ADM), wortmannin, and MG132 were purchased from Merck (Darmstadt, Germany). Dimethyl sulfoxide (DMSO) was purchased from FUJIFILM Wako Pure Chemical Corporation (Osaka, Japan).

The protocols of the present study were in compliance with the 1975 Declaration of Helsinki and approved by the Local Ethical Review Board of the Chiba University, Graduate School of Medicine in Chiba, Japan (No. 2018-320, 2020-1129) and the review boards of Sawara Hospital. Sera were collected from patients after provision of written informed consent. Serum samples from healthy donors (HD) and patients with AIS or transient ischemic attack (TIA) were obtained from Sawara Hospital as described previously (15). The stroke subtypes were assessed using the Trial of Org 10172 in the Acute Stroke Treatment categorization system (16), and AIS or ischemic stroke included large-artery atherosclerosis or minor arterial occlusion (lacune). HD subjects were selected as those without prior cancer history, autoimmune disease, or cerebro- and cardiovascular disease and exhibiting no abnormalities in cranial MRI. The sera of patients with chronic kidney disease (CKD) were obtained from the Kumamoto cohort (17, 18), of which the subject information used in the correlation analysis (Table 1) was summarized in Supplementary Table S1. Each serum sample was centrifuged at 3,000g for 10 min and supernatant was stored at −80°C until use. Repeated thawing and freezing of samples was avoided.

Antigenic GST-fused GADD34 was expressed in Escherichia coli and purified as described previously (12, 15). AlphaLISA was used to measure serum antibody levels. AlphaLISA was performed in 384-well microtiter plates (white opaque OptiPlate™ from Perkin Elmer, Waltham, MA) containing 2.5 μL of 100-fold diluted sera and 2.5 μL of GST-GADD34 or control GST protein (10 μg/mL) in AlphaLISA buffer (25 mM HEPES, pH 7.4, 0.1% casein, 0.5% Triton X-100, 1 mg/mL dextran-500, and 0.05% Proclin-300). The reaction mixture was incubated at room temperature for 6–8 h, followed by incubation with anti-human IgG-conjugated acceptor beads (2.5 μL at 40 μg/mL) and glutathione-conjugated donor beads (2.5 μL at 40 μg/mL) at room temperature in the dark for 1–14 days. Emissions were read on an EnSpire Alpha microplate reader (PerkinElmer) as described previously (15, 19–21). Specific reactions were calculated by subtracting the Alpha photon counts of the GST control from the counts of GST-fusion proteins.

The human osteosarcoma cell line, U2OS, human glioblastoma cell line, U87, and human lung carcinoma cell line, A549, were cultured in Dulbecco’s modified Eagle’s minimal essential medium supplemented with 10% (v/v) fetal bovine serum (Thermo Fisher Scientific, Waltham, MA) and 100 μg/mL of kanamycin (Meiji Seika, Tokyo, Japan) at 37°C in a humidified atmosphere containing 5% CO₂.

cDNA of GADD34 was obtained from GE Healthcare Life Sciences (Buckinghamshire, United Kingdom). The expression plasmids, pCMV-p53WT (22) and pCMV-MDM2, were provided by Dr. Bert Vogelstein (Howard Hughes Medical Institute). shRNA expression plasmids were purchased from Merck (MISSION^® shRNAs; Merck, Darmstadt, Germany). siRNA duplexes against GADD34, ATM, ATR, DNA-PK, and MDM2 as well as negative control scrambled siRNA were synthesized and purified by Merck. The sequences of siRNAs used in the present study were as follows:

Cells were transfected with expression plasmids and shRNA using FuGENE 6 and X-tremeGENE HP (Roche Diagnostics, Basel, Switzerland) or transfected with siRNA using Lipofectamine RNAiMAX (Thermo Fisher Scientific) according to the manufacturer’s instructions.

Total cellular RNA was extracted using RNeasy kits (QIAGEN). Relative mRNA expression levels were measured by RT-PCR using the SuperScript III One-Step RT-PCR System according to the manufacturer’s instructions (Cytiva, Pittsburgh, PA). Primer sequences used in the present study were as follows:

Firefly luciferase reporter plasmids, pG13-Luc (23) and PUMA-Luc (7), were provided by Dr. Bert Vogelstein (Howard Hughes Medical Institute). pGL3-p21-Luc (4) and pGL3-Bax-Luc (5) were provided by Dr. Mian Wu (University of Science and Technology of China). pGV-B2 Noxa-Luc (6) was provided by Dr. Nobuyuki Tanaka (Nippon Medical School). Decorin-Luc is a p53 reporter which contains the promoter region between positions −252 and −205 of the decorine gene (8). Cells seeded on 24-well plates were transfected with p53 expression plasmids together with firefly luciferase and a control Renilla luciferase reporter plasmid, SV40-Rluc, using FuGENE 6 and X-tremeGENE HP DNA Transfection Reagent (Roche Diagnostics). Two days after transfection, firefly and Renilla luciferase activities were determined using the Dual Luciferase Assay System (Promega) and a luminescencer (Atto, Tokyo, Japan) as previously reported (24). Firefly luciferase activities were normalized against Renilla luciferase activity.

Cells were suspended in 1% Nonidet P-40 lysis buffer [50 mM Tris–HCl (pH 7.6), 150 mM NaCl, 1% Nonidet P-40, 10 mM NaF, 1 mM Na₃VO₃, 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, 10 μg/mL leupeptin, and 10 μg/mL of pepstatin] and incubated on ice for 30 min. After centrifugation, cell lysates were immunoprecipitated with anti-GADD34 polyclonal antibody (Proteintech, Rosemont, IL) or Myc-Tag mouse monoclonal antibody (9B11, Roche Diagnostics) followed by precipitation using Protein G PLUS-Agarose Immunoprecipitation Reagent (Sepharose Bead Conjugate; Santa Cruz Biotechnology, Dallas, TX). Immunoprecipitates were washed three times with 0.1% Nonidet P-40 lysis buffer prior to use in Western blotting. Cell lysates or immunoprecipitated proteins were separated by SDS-polyacrylamide gel electrophoresis and transferred to nitrocellulose filters (Bio-Rad, Tokyo, Japan). Membranes were then incubated with anti-GADD34, anti-p53 (DO-1, Santa Cruz Biotechnology), anti-phospho-p53 (Ser-15) (Cell Signaling, Danvers, MA), anti-ATM (Santa Cruz Biotechnology), anti-MDM2 (SMP-14, Santa Cruz Biotechnology), anti-ubiquitin (Cell Signaling), anti-c-myc, anti-β-actin (Santa Cruz Biotechnology), or anti-α-tubulin (Merck) antibodies. Immunoreactive proteins were visualized using Immobilon Western Chemiluminescent Substrate (Merck).

Cells were fixed with 70% ethanol at room temperature for 30 min and stained with Giemsa solution (FUJIFILM Wako Pure Chemical Corporation).

GADD34 was identified by serological identification of antigens by recombinant cDNA expression cloning (SEREX) using sera of patients with atherosclerotic ischemic stroke and a human microvascular endothelial cell cDNA library (12). Serum anti-GADD34 antibody (s-GADD34-Ab) levels have previously been measured using ELISA; however, AlphaLISA can provide stable results with low background noise (15). s-GADD34-Ab levels were significantly higher in patients with AIS compared to HDs (p < 0.0001); however, no difference was observed between patients with TIA and HDs (Figure 1A). Mean (±SD) s-GADD34-Ab levels (Alpha counts) in HDs, patients with TIA, and patient with AIS were 992 (±552), 1,388 (±914), and 1,444 (±764), respectively. When a cutoff value representing the mean s-GADD34-Ab level in HDs plus two standard deviations was used (Alpha count = 2,095), the proportions of individuals with anti-GADD34 antibodies among HDs, patients with TIA, and patients with AIS were 6.5, 13.6, and 18.0%, respectively (Supplementary Table S2). Receiver operating curve (ROC) analyses were performed to evaluate the utility of s-GADD34-Ab levels in detecting AIS. The area under the curve (AUC) of GADD34-Abs for AIS was 0.6927 [95% confidence interval (CI), 0.6376–0.7478; Figure 1B]. When a cutoff value for s-GADD34-Ab levels of 1,247 determined according to the Youden index was used, the sensitivity and specificity of s-GADD34-Ab levels for the diagnosis of AIS were 53.1 and 76.8%, respectively.

We next measured s-GADD34-Ab levels in the sera of patients with CKD, which is closely associated with atherosclerosis. CKD was classified into three groups as follows: type 1, diabetic kidney disease; type 2, nephrosclerosis; and type 3, glomerulonephritis. Samples from patients with CKD were obtained from the Kumamoto cohort. Samples from HDs were obtained from Chiba University. All three CKD groups had significantly higher serum s-GADD34-Abs levels than HDs (Figure 1C). The proportions of individuals with detectable s-GADD34-Ab levels among HDs and patients with CKD types 1, 2, and 3 were 7.3, 27.6, 18.8, and 16.3%, respectively (Supplementary Table S3). ROC analyses demonstrated that AUC values for type-1, type-2, and type-3 CKD were 0.7357 (Figure 1D), 0.7723 (Figure 1E), and 0.6605 (Figure 1F), respectively.

We then performed Spearman correlation analyses to determine the correlation between s-GADD34-Ab levels and demographics in the Kumamoto cohort including age, sex, body height, weight, body mass index (BMI), degree of artery stenosis [plaque score, maximum intima–media thickness (max IMT), ankle brachial pressure index (ABI), and cardio ankle vascular index (CAVI)], dialysis duration, smoking, and serological parameters. Subject information was summarized in Supplementary Table S1. s-GADD34-Ab levels were significantly correlated with plaque score (p = 0.058), max IMT (p = 0.076), and CAVI (p = 0.077; Table 1), indicating that s-GADD34-Ab levels are associated with atherosclerosis. s-GADD34-Ab levels had the closest correlation with ferritin levels (p < 0.0001) but no correlation with serum iron concentrations or receipt of iron supplementation. s-GADD34-Ab levels also correlated with serum aspartate aminotransferase (p = 0.0142) and alkaline phosphatase (p = 0.044) levels and percutaneous transluminal angioplasty (PTA; p = 0.0023). Inverse correlations were observed between s-GADD34-Ab levels and serum platelet counts (p = 0.0013) and creatinine level (p = 0.0439). Serum HbA1c (glycated hemoglobin) levels, a surrogate of serum glucose levels, was not significantly correlated with s-GADD34-Ab levels.

GADD34 is known to be involved in cell proliferation (25). Endogenous GADD34 and MDM2 levels were down-regulated to investigate the effects of MDM2 and GADD34 on cell proliferation. Endogenous GADD34 and MDM2 were knocked down by transfection with GADD34 siRNA and/or MDM2 siRNA in the human osteosarcoma cell line, U2OS, followed by culture for 2 weeks. Cells fixed and stained with Giemsa demonstrated increased cell numbers after transfection with GADD34 siRNA compared to control siRNA (Figure 2). Conversely, transfection with MDM2 siRNA resulted in a marked reduction in the number of cells. Transfection with both GADD34 siRNA and MDM2 siRNA resulted in lower cell numbers compared to control siRNA or GADD siRNA, indicating that GADD34 siRNA increases proliferation while co-transfection with MDM2 and GADD34 siRNA markedly reduces proliferation. This suggests that GADD34 and MDM2 are interacted with each other in the regulation of cell proliferation.

As p53 plays important roles in cell proliferation and apoptosis, the transactivation ability of p53 was examined using luciferase reporter assays. To achieve the enforced expression of GADD34, GADD34 cDNA was inserted into the eukaryotic expression vector, pME-18S (designated as pME-GADD34), and a control empty vector, pME-18S (designated as pME). Transfection of pME-GADD34 induced the activation of p53-responsive reporters including pG13-Luc, Noxa-Luc, and PUMA-Luc but not Decorin-Luc or p21-Luc (Figure 3A). Treatment with a genotoxic anticancer drug, CPT, also activated these reporters, which was markedly enhanced in the presence of GADD34. On the other hand, CPT treatment has no effect on Decorin-Luc or p21-Luc.

A similar synergistic effect between GADD34 and CPT was observed in U87 glioblastoma cells (Figure 3B). When U2OS and U87 cells were treated with etoposide, an anticancer drug, pG13-Luc, Noxa-Luc and PUMA-Luc were activated and this effect was further augmented by the presence of GADD34 (Supplementary Figures S1A,B). Etoposide or GADD34 overexpression alone were unable to activate p21-Luc or the empty reporter, pGL3.

The effects of p53 activation in response to treatment with CPT or GADD34 overexpression was confirmed by knockdown of p53. After co-transfection of a p53 shRNA expression vector, a reporter plasmid, and a GADD34 expression plasmid, CPT treatment and/or GADD34 expression markedly attenuated the activation of pG13-Luc, Noxa-Luc, and PUMA-Luc (Figure 3C). No activation of luciferase reports was observed after transfection with scrambled shRNA as a control. Similar results were observed in response to treatment with etoposide (Supplementary Figure S1C). p53 shRNA alone reduced basal luciferase activity compared to scrambled shRNA, indicating that p53 retains some transcription activity in the absence of genotoxic stress.

The activation of p53 by GADD34 was further examined by Western blotting. p53 expression levels were almost undetectable after transfection with the control empty vector, pME. Increased p53 levels were observed after treatment with CPT (Figure 4A). ATM, ATR, and DNP-PK are well-known upstream activators of p53 (9). Increases in p53 levels were attenuated by pre-treatment with wortmannin, which inhibits ATM and DNA-PK. Enforced expression of GADD34 augmented p53 expression levels in non-treated, CPT-treated, and wortmannin plus CPT-treated cells. GADD34 levels were almost undetectable after transfection with pME but were clearly observed after transfection with pME-GADD34.

The effects of GADD34 on p53 expression were further examined by knockdown of GADD34. RT-PCR demonstrated GADD34 mRNA expression levels were reduced to <50% after transfection with siRNA-GADD34 (Figure 4B). GADD34 siRNA2 and GADD34 siRNA3 resulted in more potent knockdown of GADD34 mRNA levels than GADD34 siRNA1. This was further confirmed by Western blotting, with almost undetectable GADD34 protein levels in the presence of GADD34 siRNA2 or siRNA3 irrespective of treatment with CPT or ADM (Figure 4C). p53 expression levels in non-treated, CPT-treated, and ADM-treated cells were reduced in the presence of GADD34 siRNA. Thus, GADD34 promotes p53-mediated transcription by increasing p53 protein levels.

The involvement of ATM in the upstream signaling pathway was further examined by Western blotting. p53 levels were synergistically increased by treatment with CPT and GADD34 overexpression (Figure 4D). Co-transfection with ATM siRNA reduced ATM protein levels to almost undetectable levels and decreased protein levels of p53 and GADD34. This finding implies that the protein levels of p53 and GADD34 may be regulated by an ATM-mediated pathway.

As p53 is rapidly degraded by the ubiquitin-proteasome system (26), the effects of a proteasome inhibitor, MG132, were examined. Surprisingly, knockdown of ATM markedly increased protein levels of GADD34 in the presence of MG132, whereas GADD34 protein levels were reduced in the absence of MG132 (Figure 5A). Protein levels of p53 and MDM2 increased after treatment with MG132. Thus, GADD34 protein levels may be regulated by the ubiquitin-proteasome system.

To investigate the effect of ATM on p53 activation, U2OS cells were transfected with ATM siRNA or scrambled siRNA and then treated with ADM. Western blotting demonstrated that ADM treatment increased protein levels of GADD34 and this effect was suppressed by knockdown of ATM (Figure 5B). The expression of ATM was efficiently reduced as expected. Pre-treatment with the proteasome inhibitor, MG132, partly increased protein levels of MDM2 and GADD34. Thus, ATM may function upstream of GADD34 and MDM2, possibly via proteasomal degradation.

To further investigate the effect of MDM2 on GADD34 expression, we measured protein levels using Western blotting. U2OS cells were transfected with scrambled siRNA or two different MDM2 siRNAs and then treated with the DNA-damaging agents, ADM or MG132. In control experiments, GADD34 levels increased after treatment with ADM or MG132. In contrast, MDM2 depletion attenuated the increase in GADD34 protein levels in response to ADM or MG132 (Figures 5C,D). These findings indicate that MDM2 expression results in GADD34 stabilization but not degradation.

We then examined ubiquitination of GADD34. Myc-tag GADD34 proteins in U2OS cell lysates were immunoprecipitated with anti-Myc antibody followed by Western blotting using anti-ubiquitin, anti-Myc, and anti-MDM2 antibodies. Ubiquitinated GADD34 was detectable in immunoprecipitates and was more clearly observed following treatment with MG132 (Figure 6A). Levels of ubiquitinated MDM2 were increased whereas levels of non-ubiquitinated GADD34 were decreased in response to MG132 treatment.

GADD34 has been shown to be ubiquitinated in cultured cells (27). To investigate the effect of MDM2 on GADD34 ubiqutination, U2OS cells were transfected with scrambled siRNA or MDM2 siRNA followed by treatment with MG132. GADD34 protein was immunoprecipitated with anti-GADD34 antibodies and then probed with anti-ubiquitin or anti-MDM2 antibodies. The detection of MDM2 in immunoprecipitates indicates binding between GADD34 and MDM2. Knockdown of MDM2 resulted in a marked reduction in GADD34 ubiquitination (Figure 6B). This finding indicates that the ubiquitination and expression levels of GADD34 are regulated by MDM2.