Activation of the ATF6 (Activating Transcription Factor 6) Signaling Pathway in Neurons Improves Outcome After Cardiac Arrest in Mice

Animals

C57Bl/6 mice were obtained from The Jackson Laboratory (Bar Harbor, ME). The Cre‐dependent and tamoxifen‐inducible sATF6 knock‐in mouse line, Rosa26‐sATF6‐MER, was generated previously in our laboratory.¹⁵ Of note, the gene segment of sATF6 is tagged with the FLAG at the N‐terminus, and is also fused to the mutated estrogen receptor (MER) at the C‐terminus. Because of MER presence, the fusion protein sATF6‐MER is retained in the cytosol. Tamoxifen treatment can trigger the translocation of sATF6‐MER to the nucleus, where ATF6‐dependent gene expression is then induced, thereby activating the ATF6 UPR branch. To express sATF6 in neurons, Rosa26‐sATF6‐MER was cross‐bred with Emx1‐Cre mice (Jackson Laboratory 005628) to generate Rosa26‐sATF6‐MER; Emx1‐Cre (sATF6‐KI) mice. Littermate Emx1‐Cre mice were used as controls. The mouse line sATF6‐KI has been extensively characterized previously.¹⁵ In total, this study used 84 mice, including 21 C57Bl/6, 30 sATF6‐KI, and 33 littermate control mice.

CA and Cardiopulmonary Resuscitation Surgery

CA/cardiopulmonary resuscitation (CPR) surgery was performed as described previously.¹³, ¹⁹ Briefly, anesthesia was induced with 5% isoflurane. After endotracheal intubation, mice were maintained on 1.5% to 1.7% isoflurane before CA induction. Body temperature was measured with a rectal temperature probe and maintained at 36.8±0.2°C before CA. An ECG was continuously recorded. KCl (30 μL of 0.5 mol/L) was infused to induce asystole, which was verified by ECG and an absence of spontaneous respiration. Immediately after CA onset, mechanical ventilation was terminated, and pericranial temperature was maintained at 38.5±0.2°C while allowing body temperature to drop during CA. At 8.5 minutes after CA, mechanical ventilation with 100% oxygen was resumed, and a bolus of epinephrine (100 μL of 32 μg/mL) was given, followed by continuous infusion (25 μL/min) accompanied by rapid chest compression using a single finger. Return of spontaneous circulation was established by appearance of stable ECG sinus rhythms. If return of spontaneous circulation was not achieved within 3 minutes, resuscitation was abandoned, and the animal was excluded from the study. When spontaneous respiration was adequate, mice were returned to their home cages. Sham‐operated mice were subjected to the same surgical procedures until the jugular vein was exposed. Mice were then kept under isoflurane with mechanical ventilation for 15 minutes, allowing comparable exposures of sham and CA/CPR mice to isoflurane anesthesia.

Drug Administrations

Tamoxifen (Cayman; No. 13258; 20 mg/mL) was suspended in corn oil (Sigma; No. C8267). Animals were treated with 100 mg/kg tamoxifen or corn oil (vehicle) by oral gavage once daily for 5 days. Compound 147 (Tocris Bioscience; No. 6759) was suspended in 10% dimethyl sulfoxide (Sigma; No. D2650). Mice were administered compound 147 (2 mg/kg) or vehicle (10% dimethyl sulfoxide) via the tail vein 24 hours before surgery and 30 minutes after return of spontaneous circulation.

Behavioral Tests

All tests were conducted in the light phase, and the procedures were essentially the same as described preivouly.¹³, ¹⁹, ²⁰, ²¹ Briefly, a 9‐point scoring system was used to evaluate general functional deficits (9 points=normal, and 0 points=severe injury).²⁰ The duration that each mouse was able to remain on the vertical screen, horizontal rod, and horizontal rope was recorded and then a score was assigned (0–3 points for each test). The rotarod test evaluated balance, grip strength, and motor coordination using a device with an accelerating rotating rod (4–40 rpm; Med Associates). Animals were trained for 3 days before the surgery to ensure a similar baseline. Latency to fall (average of 3 trials) was recorded. The open field test assessed general health status. The animal was placed in an open field box (50×50×50 cm; CleverSys, Reston, VA), and allowed to move freely while being video recorded. The video data were then analyzed by the automated tracking system TopScan (CleverSys). The total distance traveled during the 10‐minute period of the test was calculated.

Histology

Mice were anesthetized and transcardially perfused with saline, followed by 4% paraformaldehyde. Frozen brain sections (20‐μm thick) were used for Fluoro‐Jade C (FJC) staining to identify neuronal cell death.¹³ Briefly, the slices were incubated in 0.06% potassium permanganate for 10 minutes, and then immersed in FJC staining solution (Millipore; No. AG325) for 30 minutes. Sections were imaged using a fluorescence microscope. An evaluator, who was blinded to experimental groups, counted FJC‐positive cells in both entire hippocampi of each brain section (around bregma −2.06 mm). Data are presented as a mean of total FJC‐positive cells (2 hippocampi) for each mouse.

Quantitative Reverse Transcription–Polymerase Chain Reaction

The mRNA levels of selected genes were analyzed by quantitative reverse transcription–polymerase chain reaction. In short, total RNA was extracted from brain hippocampus samples using TRIzol reagent (ThermoFisher; No. 15596026), followed by cDNA synthesis (ThermoFisher; No. 18091050). Quantitative reverse transcription–polymerase chain reaction was performed using a Lightcycler 2.0 (Roche, Mannheim, Germany). All primers used in this study are listed in Table S2.

RNA‐Seq Analysis

For the RNA‐Seq analysis, total RNA was first prepared using TRIzol, and then treated with DNase I, followed by purification using the RNeasy MinElute Cleanup kit (Qiagen; No. 74204). The integrity and concentration of the RNA samples were evaluated in an Agilent Bioanalyzer (Agilent Technologies, Santa Clara, CA) and Qubit 2.0 (ThermoFisher), respectively. RNA‐Seq analysis was performed in the Genomic Analysis and Bioinformatics Shared Resource at Duke University Medical Center, similarly to our previous study.²² Briefly, total RNA (500 ng/sample) was used for library construction with the KAPA Stranded mRNA‐Seq Kit (Kapa BioSystems, Wilmington, MA). Sequencing was performed on the Illumina HiSeq 4000 sequencing platform and de‐multiplexed using bcl2fastq2 v2.20. The raw sequencing data are available in the Gene Expression Omnibus database (National Institutes of Health, National Library of Medicine, Bethesda, MD; accession number GSE160259). For data analysis, we first cleaned up low‐quality bases and sequencing adapters with the TrimGalore toolkit, and then mapped reads to the GRCm38v68 version of the mouse genome and transcriptome using the STAR RNA‐Seq alignment tool. Gene counts were quantified using the HTSeq tool, and normalization and differential expression were analyzed using the DESeq2 Bioconductor package with the R (R Foundation for Statistical Computing, Vienna, Austria) statistical programming environment. The false discovery rate (shown as Padj) was calculated to control for multiple hypothesis testing. For the heatmap, gene expression has been z‐score normalized, and the samples and genes are clustered by correlation distance with complete linkage. A gene ontology category database was applied for biological process, molecular function, and cellar component annotation of differentially expressed genes (DEGs). Enrichment analysis of gene ontology categories was performed using DAVID online tools (https://david.ncifcrf.gov). The Ingenuity Pathway Analysis (Ingenuity Systems, Redwood City, CA) on‐line program was used to identify function and canonical pathways.

Western Blotting

Tissue samples were processed using our standard method.¹³ Briefly, the brain hippocampus was dissected on ice and snap frozen in liquid nitrogen as quickly as possible to prevent deconjugation of ubiquitinated proteins. Tissue samples were homogenized by sonication using lysis buffer supplemented with 2% SDS. Protein samples were run on 4% to 12% Bis‐Tris Plus gels (ThermoFisher), and were transferred to polyvinylidene difluoride (PVDF) membranes. After blocking, membranes were incubated with a primary antibody at 4°C overnight. After extensive washing, membranes were incubated with a secondary antibody for 1 hour at room temperature. Proteins were then detected with the enhanced chemiluminescence (ECL) analysis system (GE Healthcare). Quantification of signal intensities was performed using ImageJ (National Institutes of Health). The primary antibodies are listed in Table S3.

Statistical Analysis

Sample sizes were determined based on our previous studies using the same CA/CPR model. Data were analyzed using Prism 8 software (GraphPad, LaJolla, CA). Statistical analysis was assessed by Mann‐Whitney U test for neurologic scores, Mantel‐Cox log‐rank test for survival rates, unpaired Student t test for 2 groups, and 1‐way ANOVA with post hoc Fisher least significant difference test for >2 groups. Data are presented as mean±SEM, median, or percentage. For RNA‐Seq analysis, selection of DEGs was based on adjusted P<0.05 with fold change >1.5 or <−1.5. For other analyses, the level of significance was set at P<0.05.

CA Outcome Was Improved After Neuron‐Specific Activation of the ATF6 UPR Branch

We previously generated a conditional and tamoxifen‐inducible sATF6 knock‐in mouse line, which is a valuable tool particularly for studying the ATF6 UPR branch in various cell types.¹⁵ Here, to clarify the effect of activating the ATF6 branch in neurons after CA/CPR, we used sATF6‐KI mice, in which transgene sATF6 is expressed predominantly in forebrain neurons. First, both control and sATF6‐KI mice received tamoxifen dosing for 5 days to induce the nuclear translocation of sATF6 and thus activate downstream ATF6 signaling in sATF6‐KI mice. After 3 days of recovery to clear tamoxifen, mice were subjected to 8.5 minutes CA using our CA/CPR model. Compared with control mice, sATF6‐KI mice appeared to require less CPR time (Figure 1A). Also, sATF6‐KI mice trended toward less body weight loss on day 3 after CA/CPR (Figure 1B). To evaluate functional recovery, rotarod test and neurologic scoring were performed on day 1 and day 3 after CA/CPR. On day 1 after CA/CPR, sATF6‐KI mice tended to perform better in both tests compared with control mice, and this improvement reached statistical significance on day 3 (Figure 1C and 1D). Finally, we examined neuronal death in the hippocampus on day 3 after CA/CPR using FJC staining. Consistent with behavioral results, sATF6‐KI mice had significantly fewer FJC‐positive cells, indicating more surviving neurons after CA/CPR (Figure 1E). Based on these data, we then attempted to determine the extent to which sATF6‐KI mice exhibited improved long‐term outcome (ie, 14 days after CA/CPR) (Figure S1). However, this experiment suffered a high mortality rate in the control group versus sATF6‐KI group (55.6% versus 14.3% on day 14 after CA, respectively), which confounded long‐term functional assessments. However, even considering the potential survival bias and the small mouse number in the control group, sATF6‐KI mice appeared to have a better outcome than control mice (Figure S1). Collectively, our data indicate that activation of the ATF6 branch in neurons was sufficient to reduce neuronal injury and improve functional recovery after CA/CPR.

Pro‐Proteostatic Genes, Many Related to the ERAD Pathway, Were Upregulated After Neuron‐Specific Activation of the ATF6 UPR Branch

As the first step toward understanding the mechanisms responsible for improved CA outcome in sATF6‐KI mice, we examined another UPR pathway (ie, the PERK branch) using eukaryotic initiation factor 2α phosphorylation as an indicator. As expected, GRP78 protein levels were considerably higher in sATF6‐KI mice, compared with control mice (Figure 2). Interestingly, PERK protein levels were also significantly increased in the sATF6‐KI mouse brains. However, levels of CA‐induced eukaryotic initiation factor 2α phosphorylation were not different between control and sATF6‐KI mice (Figure 2), suggesting a similar activation of the PERK branch. This result is consistent with our previous finding in an ischemic stroke model.¹⁵

Next, considering that sATF6 is a transcriptional factor, we speculated that the protective effects of the ATF6 UPR branch may be because of prosurvival genes and related processes upregulated by sATF6. However, no study has yet systematically analyzed ATF6‐regulated genes in the brain. Thus, we performed an RNA‐Seq analysis to characterize the gene expression profile in the sATF6‐KI mouse brain (Figure 3). Hippocampus tissue samples were collected from control and sATF6‐KI mice on day 3 after 5 days of tamoxifen administration. The volcano plot showed that most of the DEGs were upregulated in sATF6‐KI brains (Figure 3A). After filtering the data with our selection criteria (fold change >1.5 or <−1.5; Padj<0.05), we found that 195 genes were upregulated, and only 11 genes were downregulated (Figure 3A and 3B). We have listed the top DEGs (top 15 from upregulated genes and all 11 downregulated genes) in Table. Then, we subjected DEGs to gene ontology analysis and Ingenuity Pathway Analysis (Figure 3C and 3D and Figure S2). As expected, enriched biological processes, as well as the cell component and molecular function, were primarily associated with adaptive ER stress response, protein folding and degradation, and redox homeostasis (Figure 3C, Figure S2A). Consistently, the Ingenuity Pathway Analysis identified 4 activated pathways including the unfolded protein response, ER stress pathway, GP6 (glycoprotein 6) signaling pathway, and NRF2 (nuclear factor erythroid 2‐related factor 2)‐mediated oxidative stress response (z score >2) (Figure S2B). Further analysis indicated that significantly upregulated function categories are associated with molecular transport, cell movement, cell function and maintenance, cell death/survival, and carbohydrate metabolism (Figure 3D). Interestingly, increased functions (indicated by positive z scores) include transport of molecule, cellular homeostasis, and cell survival, whereas decreased functions (indicated by negative z scores) are associated with cytolysis and apoptosis (Figure 3D). Notably, at least 9 ERAD core components (Padj<0.001) were upregulated (Table S4).²³ Among the DEGs, the most upregulated gene was Derlin 3 (Derl3), a key component of the ERAD machinery.²⁴, ²⁵ Using quantitative reverse transcription–polymerase chain reaction, we confirmed significant upregulation of 4 ERAD core genes (Derl3, Hspa5, Herpud1, and Hrd1) as well as 7 other genes from the list of DEGs (Figure 3E).

To further verify an increase in ERAD components in sATF6‐KI versus control mouse brains, we performed Western blotting analysis in CA samples. Control and sATF6‐KI mice were subjected to sham or CA/CPR surgery, and at 3 hours reperfusion, hippocampus samples were collected and analyzed. Compared with control mice, sATF6‐KI mice showed significantly higher levels of Sel1l (sel‐1 suppressor of lin‐12‐like), OS9 (osteosarcoma amplified 9), VCP (valosin‐containing protein), and HERP (or Herpud1; Homocysteine‐inducible endoplasmic reticulum stress‐inducible ubiquitin‐like domain member 1). However, at this time point of reperfusion, CA/CPR itself appeared to have no significant effect on the protein levels of any of these ERAD components (control groups; Figure 4). Notably, previous studies have suggested that improving ERAD capacity by upregulating related genes markedly protects the heart from ischemic insult.²⁴, ²⁶

In addition, from the RNA‐Seq data, we also noted that mesencephalic astrocyte‐derived neurotrophic factor (MANF) was significantly upregulated in sATF6‐KI mouse brains, which was validated by qRT‐PCR (Figure 3E). Because MANF is a potent neuroprotective molecule, we assessed MANF protein levels, and confirmed its increase in sATF6‐KI mice (Figure 4G). Taken together, data indicate that the ERAD machinery as a whole is boosted by activation of the ATF6 UPR branch. It is reasonable to suggest that augmented ERAD capacity to clear misfolded proteins and thus resolve perturbed ER proteostasis in neurons contributes to the improved CA outcome observed in sATF6‐KI mice.

Protein Ubiquitination Was Differentially Increased in Control and sATF6‐KI Mouse Brains After CA/CPR

Based on the data above, the next step would be to confirm an increase in ERAD activity in sATF6‐KI mouse brains after CA/CPR. However, because no universal or brain‐specific ERAD substrate has been identified, an established way to directly measure ERAD activity in vivo in the brain is not yet available. Alternatively, we analyzed ubiquitination levels in the post‐CA brain because ubiquitination is a key step in the ERAD pathway. We speculated that if sATF6‐expressing neurons can more efficiently label unwanted proteins for degradation via the ERAD pathway, the level of proteins that are ubiquitinated for degradation would be higher in sATF6‐KI mice than in control mice. To test this, we measured ubiquitination levels in the post‐CA brain (Figure 5). Consistent with our previous reports,¹³, ²⁷, ²⁸ levels of global ubiquitination were robustly increased after CA/CPR, although to a similar extent in control and sATF6‐KI mice. Of note, ubiquitination regulates different biological processes according to the linkage types of which K48‐ and K63‐linkages have been commonly examined.²⁹ Especially, K48‐linked polyubiquitin chains serve as the principal linkage signal for protein degradation. Thus, we further examined the changes of these 2 polyubiquitin linkages. Interestingly, the increase in K48‐linked polyubiquitin chains was significantly higher in sATF6‐KI versus control mice, whereas the K63‐linkage signal was significantly lower in sATF6‐KI mice (Figure 5).

CA Outcome Was Improved in Mice Treated With Compound 147

Finally, we set out to determine whether pharmacologic activation of the ATF6 branch improves CA outcome. To this end, we tested the compound 147. Compound 147 was among the first small molecules identified as specific ATF6 activators through phenotypic screening.³⁰ A previous time‐course experiment indicates that induction of ATF6‐regulated genes reaches a maximum at 24 hours after compound 147 administration.¹⁶ Thus, as the first attempt of testing compound 147 in CA, we chose a dosing regimen similar to one used in the recent study¹⁶ (ie, pretreatment at 24 hours before CA/CPR surgery plus post‐treatment at 30 minutes after return of spontaneous circulation). We had 2 experimental groups: CA+vehicle and CA+compound 147. There was no significant difference in CPR time between the 2 groups (Figure 6A). On day 3 after CA/CPR, however, we found that mice treated with compound 147 performed significantly better in both neurologic scoring and the open field test (Figure 6B and 6C). Moreover, the survival rate over 3 days following CA was significantly improved after compound 147 treatment (Figure 6D).