ADP- ribose transferase PARP16 mediated- unfolded protein response contributes to neuronal cell damage in cerebral ischemia/reperfusion

Ischemic stroke is known to cause the accumulation of misfolded proteins and loss of calcium homeostasis, leading to impairment of endoplasmic reticulum (ER) function and activating the unfolded protein response (UPR). PARP16 is an active (ADP- ribosyl)transferase known tail- anchored ER transmembrane protein with a cytosolic catalytic domain. Here, we find PARP16 is highly expressed in ischemic cerebral hemisphere and oxygen– glucose deprivation/reoxygena-tion (OGD/R)- treated immortalized hippocampal neuronal cell HT22. Using an adeno- associated virus- mediated PARP16 knockdown approach in mice, we find PARP16 knockdown decreases infarct demarcations and has a better neurological outcome after ischemic stroke. Our data indicate PARP16 knockdown decreases ER stress and neuronal death caused by OGD/R, whereas PARP16 overexpression promotes ER stress- mediated cell damage in primary cortical neurons. Furthermore, PARP16 functions mechanistically as ADP- ribosyltransferase to modulate the level of ADP- ribosylation of the corresponding PERK and IRE1 α arm of the UPR, and such modifications mediate activation of PERK and IRE1 α . Indeed, pharmacological stimulation of the UPR using Brefeldin A partly counteracts PARP16 knockdown- mediated neuronal protection upon OGD/R treatment. In conclusion, PARP16 plays a crucial role in post- ischemic UPR and PARP16 knockdown alleviates brain injury after ischemic stroke. This study demonstrates the potential of the PARP16- PERK/IRE1 α axis as a target for neuronal survival in ischemic stroke.


| INTRODUCTION
Ischemia stroke is a severe and life-threatening disease, remains a major cause of disability worldwide. 1,2 The primary treatments for ischemic stroke are thrombosis or recanalization, which are believed to a standard therapeutic option. 3 However, reperfusion itself may lead to more severe tissue injury and inflammatory response. 4 The development of new treatments requires a comprehensive understanding of the diverse mechanisms that are responsible for neuronal death during ischemic brain damage. The mechanisms underlying cerebral ischemia and reperfusion (I/R) injury are complex, involving apoptosis, necrosis, and inflammatory responses. [5][6][7] Consequently, an improved understanding of the mechanisms of I/R damage has helped to find promising and novel approaches to ischemic stroke therapy.
Endoplasmic reticulum (ER) is the site of synthesis and folding of secretory proteins. This organelle is essential for multiple other cellular functions, in order to carry out their cellular functions, proteins must be folded into proper conformations. 8 However, perturbations of ER homeostasis can lead to the accumulation of unfolded or misfolded proteins in the ER lumen, known as ER stress, which activates the unfolded protein response (UPR). 9 In mammalian cells, the UPR is mediated by three main signaling components: protein kinase RNA-like ER kinase (PERK), inositolrequiring enzyme 1 (IRE1), and activation transcription factor 6 (ATF6). UPR activation can promote cell survival, however, when the ER stress is chronic or severe, the organelle elicits apoptotic signals. In ischemia stroke, pathological changes cause the accumulation of misfolded proteins and alterations of calcium homeostasis, leading to ER stress and activating the UPR. 10,11 Then, ER stress contributes to cell death or to neurologic impairment in the penumbra of focal ischemia, which was encouraged to further study the potential of UPR-based therapies.
PARP16 (also known as ARTD15) is the only known ADP-ribosyltransferase tail-anchored ER transmembrane protein with a cytosolic catalytic domain. 12 Of note, PARP16 is required for activation of the functionally related ER stress sensors PERK and IRE1α during the UPR. During ER stress, PARP16 enzymatic activity is upregulated and modulates the level of ADP-ribosylation of PERK and IRE1α, but ATF6 is not regulated by PARP16, 13 ultimately helping to regulate the unfolded protein response. Therefore, PARP16 plays an important role in cellular protein folding regulation and homeostasis.
In cerebral ischemic injuries, pathological injury can lead to ER stress and activating the UPR. Given an important role of PARP16 in cellular protein folding regulation and homeostasis, we examined the role of PARP16 and ADP-ribosylation of PERK and IRE1α in a pathological state associated with ischemic stroke. We postulated that PARP16 may be a useful target for therapeutic intervention to prevent secondary progression of ischemic brain injury.
Male C57BL/6J mice (22-25 g) were randomly divided into three groups: sham-operated group (AAV-Control+Sham); middle cerebral artery occlusion (MCAO) group (AAV-Control+MCAO); and a shPARP16-treated group (AAV-Parp16 shRNA+MCAO). 5 μl AAV-Control or AAV-Parp16 shRNA was injected into the lateral ventricle unilaterally. The coordinates were as follows 14 : anteroposterior, −2.0 mm; mediolateral, ±1.5 mm; dorsoventral, −2.0 mm. The injection was performed using a 10 μl Hamilton syringe driven by a minipump (KDS Model 310 Plus, KD Scientific, Holliston, MA). The needle was kept in the injection site for another 3 min and then withdrawn slowly to prevent solution leakage. Mice were allowed to recover on a heating pad and returned to the animal facility. Similarly, the sham-operated and MCAO groups received equal volumes of AAV-Control. At 21 days post-injection, MCAO model was set up with middle cerebral artery (MCA) occlusion by intraluminal block. In brief, mice were kept anesthetized during surgery with 1% pentobarbital sodium (50 mg/kg, i.p, Sigma-Aldrich, USA). Transient ischemia was induced by using the suture occlusion technique, as previously described, 15 with minor modifications. The common carotid artery area was exposed, a silicon rubber-coated filament of size 6-0 was inserted through the left internal carotid artery until the approximate branch of the left MCA, and kept in place for 90 min to block blood flow. Subsequently, the suture and the filament were removed to allow reperfusion for 24 h. For sham animals, the filament was advanced to the MCA and withdrawn immediately. At 24 h after reperfusion, the neurological deficits scores of all mice were evaluated in line with the criteria described by Bederson. 16 Briefly, a grading scale of 0-3 was used to assess the extent of neurologic deficits. Mice extended both forelimbs toward the floor and were assigned grade 0; mice with any amount of consistent forelimb flexion were graded 1; mice had consistently reduced resistance to lateral push toward the paretic side, and were graded 2; mice circled toward the paretic side consistently were graded 3. After the end of scoring, mice were anesthetized and samples were collected immediately.

| Assessment of cerebral infarction volume
TTC staining was used to evaluate infarct size. 17 Mice were killed after anesthesia, and their brains were quickly and completely removed. The olfactory bulbs and cerebellums were excised, and the brains were continuously and equidistantly cut into five coronal sections. The brain slices were completely immersed in 1% TTC solution and incubated at 37°C for 30 min. Then, sections were photographed and the infarcted regions in each section were evaluated by using ImageJ software. The percentage of the infarct volume was calculated by the following formula: the infarcted area of the ipsilateral hemisphere/total area.

| Histological staining and TUNEL assay
Paraffin-embedded tissue was subjected to stain with hematoxylin-eosin (H&E) staining for morphological changes in brain, and a TUNEL (TdT-mediated dUTP Nick-End Labeling) assay for detecting cell death was performed with an in situ apoptosis detection kit (ThermoFisher Scientific, USA) or TUNEL Apoptosis Assay Kit (Meilunbio, Dalian, China) according to the manufacturer's instructions.

| Cell culture
Hippocampal neuronal cell HT22 was incubated in DMEM-F12 supplemented with 10% fetal bovine serum (FBS; Gibco, USA) at 37°C in a humidified atmosphere with 5% CO 2 . Primary cortical neurons were obtained from rats after birth day 1 as previously described 18 and cultured at 37°C in Dulbecco's modified Eagles's medium (DMEM) medium containing 10% FBS. After 8 h, the medium was changed into FBS-free DMEM-F12 (Hyclone, USA) medium containing 2% B27. Cells were cultured for 7 days and were identified using specific neuron marker Map-2 by immunofluorescent staining.

| Lentiviral preparations for PARP16 overexpression and transfection
The lentiviral construct co-expresses PARP16 was obtained as previously described. 19 The rat PARP16 was amplified from rat genomic DNA and then cloned into the vector pCDH-CMV-MCS-EF1-copGFP. To obtain the lentivirus, the recombinant plasmid and packaging vectors pMD2.0G and psPAX2 were co-transfected into HEK293T cells using the transfection reagent lipofectamine™ 2000 (ThermoFisher Scientific, USA). After 48 and 72 h, the lentivirus in the culture medium was collected and infiltrated with 0.45 μm filters. The lentivirus was added to the culture medium with 8 μg/ml polybrene (Sigma-Aldrich, USA) in rat primary cortical neurons.

| Oxygen-glucose deprivation/ reoxygenation (OGD/R) model in vitro
Oxygen-glucose deprivation/reoxygenation (OGD/R) was established as follows: cells were cultured in DMEM glucose-free medium and were incubated for 4 h in <0.1% oxygen in the hypoxia box. Then cultures were switched to completely normal conditions for 24 h.

| Immunohistochemical staining
Tissue sections were treated with autoclaving at 121°C for 15 min in 0.01 mol/L citrate-buffered saline (pH 6.0) for antigen retrieval. The sections were then immersed in 3% H 2 O 2 for 30 min at room temperature to block the endogenous peroxidase activity. After deactivation, 10% normal goat serum was used to block non-specific binding of the immunological reagents. After incubation of the antibody against iNOS at 4°C overnight, each slide was rinsed three times in PBS and incubated with biotinylated anti-rabbit IgG and HRP-streptavidin at room temperature according to the immunohistochemical staining kit (ThermoFisher Scientific, USA), stained with DAB substrate. Finally, nuclear counterstaining was done using hematoxylin.

| Immunofluorescence staining
Cells slides or tissue sections were fixed with 4% paraformaldehyde and then permeabilized for 10 min using 1% Triton X-100 in dissolved in PBS followed by incubation with the primary antibodies at 4°C overnight. Slides were washed three times with PBS and incubated with Fluor-conjugated secondary antibody (ThermoFisher Scientific, USA). The nucleus was stained with 4′,6-diamidino-2-phenylindole (DAPI) to locate the cell. Fluorescent pictures were obtained with a Zeiss fluorescence microscope (Zeiss, Germany).

| Western blot analysis
Samples from the culture cells and tissues were pooled in ristocetin-induced platelet aggregation (RIPA) buffer containing protease and phosphatase inhibitors (Complete Protease Inhibitor Cocktail and PhosSTOP Phosphatase Inhibitor Cocktail; both from Roche), and then centrifuged at 12 000g for 10 min at 4°C. The supernatant was collected. The total protein concentration in each sample was determined with a bicinchoninic acid (BCA) assay (ThermoFisher Scientific, USA) according to the manufacturer's instructions. An equal amount of protein samples was separated by SDS-polyacrylamide gel electrophoresis and transferred to nitrocellulose membranes (Millipore). The membranes were blocked with 5% skimmed milk (wt/ vol) in Tris-buffered saline supplemented with 0.1% Tween 20 (TBST) for 1 h at room temperature and then incubated with the corresponding primary antibodies at 4°C overnight. VCAM-1 antibody was purchased from GeneTex (GeneTex, USA); COX-2, IL-18, IL-6, and IL-1β antibodies were purchased from ABclonal (ABclonal Technology Co., Ltd. China); iNOS, BIP, Calnexin, zona occludens 1 (ZO-1), βtubulin, and GAPDH antibodies were obtained from Proteintech (Proteintech Group, Inc, USA); Bax, p53, Bcl-2, Caspase 3, Caspase 9, p-elf2α, and p-PERK antibodies were purchased from CST (Cell Signaling Technology, Inc, USA); XBP-1 and Claudin-1 antibodies were obtained from Abcam (Abcam, UK); Caspase-1 (p20) was obtained from AdipoGen (AdipoGen Life Sciences, Switzerland); p-IRElα antibody was obtained from (ThermoFisher Scientific, USA); PARP16 antibody was obtained from Aviva (Aviva Systems Biology, USA). The membranes were then washed and incubated for 1 h at room temperature with the species-appropriate horseradish peroxidase (HRP)labeled secondary antibody (Jackson ImmunoResearch Inc., USA). Protein-specific signals were detected using a Bio-Rad Imager (Bio-Rad, Hercules, CA, USA), and the bands were quantified by densitometric analysis (ImageJ software, NIH). The relative amounts of proteins were normalized against βtubulin.

| RNA extraction and quantitative real-time polymerase chain reaction (qRT-PCR)
Total RNA was isolated from the cells or tissues using TRIzol reagent (Takara Bio Inc., China). Reverse transcription was performed according to the instruction of the Prime RT Master Mix kit (Takara Bio Inc., China). Quantitative real-time PCR was performed in triplicate using SYBR Premix EX Taq II (Yeasen Biotech Co., Ltd. China). The sequences of primers are shown in Table 1.

| Small interfering RNA (siRNA) transfection in vitro
Cells were plated in six-well culture dishes and were transfected with Parp16 and control small interfering RNA (siRNA) (GenePharma Co. Ltd. Shanghai, China) using Lipofectamine™ RNAiMAX (ThermoFisher Scientific) according to the manufacturer's instructions. The sequences were as follows: mouse Parp16 siRNA: 5′-CGGAU UUC CUA UUU GAA AUTT-3′, rat Parp16 siRNA: 5′-CCUAC CUC ACA AGU GAC UUTT-3′; and a control siRNA was applied as a negative control. The experiments were performed 24 h after transfection. NaCl, 10% glycerol, 1% Nonidet P-40 (NP-40), 2 mmol/L EDTA, and protease inhibitor cocktail. Precleaning was achieved with 20 μl of protein A/G agarose beads (Santa Cruz, Dallas, TX, USA) for 1 h, and lysates were incubated overnight at 4°C in presence of protein A/G agarose and 1-2 μg PARP16 or ADP-ribose antibody (MABE1075, Sigma-Aldrich, USA) before analysis by western blotting. The same amount of irrelevant rabbit IgG (Santa Cruz) was used to control for non-specific IgG binding to PARP16 or ADP.

| Statistical analysis
Data were expressed as mean ± SD. Differences of means were analyzed by using one-way ANOVA with the Tukey-Kramer post hoc test for multiple groups, and when comparing between two groups using unpaired Student's t-test or a non-parametric test. All analyses were made using the GraphPad Prism 9.0 statistical software package and p value of <.05 was accepted as statistically significant.

| PARP16 expression is upregulated in HT22 cells after OGD/R and in ischemic cerebral hemisphere of mice after MCAO
HT22 cells were cultured and injured by anaerobic bags to make OGD/R model. We assessed PARP16 expression by western blot and immunofluorescence on HT22 neuron cells after OGD treatment for 4 h and reoxygenation for 24 h. After OGD/R, PARP16 expression was significantly increased compared to untreated cells ( Figure 1A). In addition, we tested the ischemic hemisphere of mice injured by MCAO and found that after MCAO injury, PARP16 expression was also significantly increased ( Figure 1B). To evaluate this further, PARP16 expression was detected by immunofluorescence staining in OGD/R-treated HT22 cells ( Figure 1C), and in vivo cerebral tissues were processed for immunofluorescent doubly staining using PARP16 and the antibody against neuron-specific marker Map-2 ( Figure 1D). Consistently, PARP16 expression was upregulated in HT22 cells after OGD/R and in brain of mice after MCAO, suggesting that PARP16 may contribute to cerebral I/R injury.

| PARP16 knockdown reduces neuronal inflammatory response after OGD/R
Inflammation plays significant role in the pathogenesis of ischemic stroke, and it can be regulated by intracellular complex inflammasomes. 20 During ischemic brain injury, inflammasome activates Caspase-1, further leading to production of proinflammatory IL-1β and IL-18. 21 The expressions of inflammation-related proteins were assessed by western blot. The results showed that after OGD/R, the expressions of NLRP3 inflammatory body-associated protein (Caspase-1(p20), IL-1β, and IL-18) and inflammationassociated protein (COX-2, iNOS, and VCAM-1) were increased, but PARP16 knockdown could significantly inhibit their expressions ( Figure 2A). In addition, immunofluorescence analysis showed that IL-1β expression was significantly reduced upon PARP16 knockdown in OGD/R-treated HT22 cells ( Figure 2B). Subsequently, the mRNA expressions of inflammation-related genes were tested, and the results showed that PARP16 knockdown could significantly inhibit the increase in the expression of inflammation-related genes Nos2 and Vcam1 induced by OGD/R ( Figure 2C). Taken together, these data suggest that PARP16 contributes to inflammatory response after OGD/R in HT22 cells.

| PARP16 knockdown reduces neuronal ER stress through inhibiting ADP-ribosylation of PERK and IRE1α
Given the above data, we investigated whether two main signaling components of endoplasmic reticulum (such as p-IRElα, p-PERK, XBP-1, Calnexin, p-elf2α) were involved in PARP16-mediated neuronal apoptosis. The results showed that after OGD/R, the expressions of XBP-1, Calnexin, p-elf2α, p-IRElα, and p-PERK were increased, but PARP16 knockdown could significantly reverse those changes ( Figure 3A). Immunofluorescence analysis showed also that after OGD/R, BIP expression was upregulated, while PARP16 knockdown could significantly downregulate BIP expression ( Figure 3B). These results suggest that PARP16 knockdown reduces neuronal ERstress induced by OGD/R in HT22 hippocampal neurons. Next, we further identified how the ER stress was activated by PARP16. Because PARP16 is an effective regulator of ER stress sensor PERK and IRE1α, to clarify the interaction between the PARP16 and ER stress sensor, cell lysates were immunoprecipitated with antibody against PARP16. Co-IP results showed that PARP16 could interact with PERK and IRE1α in OGD/R-induced HT22 cells ( Figure 3C). It is generally known that ADP-ribosylation by PARP16 directly activates PERK and IRE1α. To determine whether the ADP-ribosylation of PERK and IRE1α was upregulated under OGD/R, we used Co-IP to examine the level of ADP-ribosylation of PERK and IRE1α. Our results showed that OGD/R resulted in an increased ADP-ribosylation of PERK and IRE1α, whereas PARP16 knockdown decreased ADP-ribosylation of PERK and IRE1α ( Figure 3D), suggesting that ADP-ribosylation by PARP16 was sufficient to activate PERK and IRE1α. These results suggest that PARP16 modulates the level of ADPribosylation of UPR activator PERK and IRE1α involved in a pathological state associated with OGD.

| PARP16 promotes ER stressmediated primary cortical neurons damage
To ascertain whether overexpression of PARP16 induces ER stress-mediated cell injury, lentiviral expression vector of PARP16 was used in primary cortical neuronal culture ( Figure 5A). In addition, the levels of two main signaling components of endoplasmic reticulum (such as p-PERK, p-IRElα, p-elf2α, XBP-1, Calnexin, BIP) protein expressions were increased in PARP16-overexpressed cortical neurons (Parp16 OE) ( Figure 5A). The results implied that the ER stress was induced by PARP16 overexpression. Overexpression of PARP16 resulted in increased expression of pro-apoptotic proteins (such as p53, Bax, cleaved-Caspase 9), and decreased anti-apoptotic protein Bcl-2 ( Figure 5B). Consistently, positive cells of TUNEL staining increased in PARP16-overexpressed cortical neurons ( Figure 5C) and F I G U R E 3 PAPR16 knockdown reduces neuronal ER stress at least partly through inhibiting ADP-ribosylation of PERK and IRE1α after OGD/R. (A and B) PARP16 knockdown reduced neuronal ER stress after OGD/R. HT22 cells transfected with either control siRNA (siNC) or Parp16 siRNA (siParp16) were cultured for 48 h and subsequently treated with OGD 4 h and 24 h of reoxygenation. Immunoblot analysis of PARP16 and ER stress-related proteins (p-PERK, p-elf2α, p-IRE1α, XBP-1, and Calnexin) in HT22 cells. Data are presented as mean ± SD of three independent experiments, ***p < .001 compared with Control+siNC, ## p < .01, ### p < .001 compared with OGD/R + siNC (A); Immunofluorescence staining of BIP in HT22 cells after OGD/R. Scale bars: 50 μm. Data are presented as mean ± SD, n = 3, *p < .05 compared with Control+siNC, ### p < .001 compared with OGD/R + siNC (B). (C and D) PARP16 knockdown inhibited ADP-ribosylation of PERK and IRE1α after OGD/R. HT22 cells transfected with either siNC or siParp16 were cultured for 48 h and subsequently treated with OGD 4 h and 24 h of reoxygenation. Co-immunoprecipitation assay using antibody against PARP16 in lysates of HT22 cells induced by OGD/R (C); Co-immunoprecipitation assay using anti-ADP-ribose binding reagents in lysates of OGD/R-induced HT22 cells (D). knockdown of PARP16 also decreased the apoptosis of primary cortical neuron cells exposed to OGD ( Figure 5D). Furthermore, the expressions of NLRP3 inflammatory body-associated protein (Caspase-1(p20), IL-1β, IL-18) and inflammation-associated protein (COX-2, iNOS, and VCAM-1) were increased in PARP16-overexpressed cortical neurons ( Figure 5E). These results suggest that PARP16 promotes ER stress-mediated primary cortical neuron damage.

| AAV-mediated PARP16 knockdown reduces I/R-induced apoptosis and neuroinflammation
Apoptosis and neuroinflammation are the common features of I/R injury diseases. Western blot analysis confirmed the increased level of apoptosis in the damaged hemisphere of MCAO mice. After MCAO, the expressions of pro-apoptotic proteins (such as Bax, cleaved-Caspase 3, cleaved-Caspase 9, p53) were upregulated, but the anti-apoptotic protein Bcl-2 expression was downregulated. After the treatment with AAV-Parp16 shRNA, the F I G U R E 6 PARP16 knockdown reduces I/R-induced-brain infarct volume and ER stress. (A) Experimental scheme was shown. After administration of AAV-Parp16 shRNA or AAV-Control for 3 weeks, mice were subjected to 90-min middle cerebral artery occlusion (MCAO) followed by 24 h reperfusion. (B) qPCR and immunoblot analysis of PARP16 in brain of mice subjected to 90-min MCAO followed by 24 h reperfusion. Data are presented as mean ± SD; n = 3-4, *p < .05, ***p < .001 compared with AAV-Control+Sham, # p < .05, ### p < .001 compared with AAV-Control+MCAO. (C) Immunoblot analysis of ER stress-related proteins (p-PERK, p-elf2α, p-IRE1α, BIP, XBP-1, and Calnexin) in brain of mice subjected to 90-min MCAO followed by 24 h reperfusion. Data are presented as mean ± SD; n = 4, ***p < .001 compared with AAV-Control+Sham, ### p < .001 compared with AAV-Control+MCAO. (D) Representative images of immunofluorescence double staining of BIP and Map-2 in brain of mice subjected to 90-min MCAO followed by 24 h reperfusion. Scale bars: 20 μm. (E) The neurological deficit scores of mice subjected to 90-min MCAO followed by 24 h reperfusion. Data are presented as box plot with the median. Significance was determined by Mann-Whitney test, n = 10, # p < .05 compared with AAV-Control+MCAO. (F) PARP16 knockdown reduces I/R-induced-brain infarct volume showed by TTC staining. Data are presented as mean ± SD; n = 6, ### p < .001 compared with AAV-Control+MCAO. (G) Representative images of H&E staining at 24 h of reperfusion after MCAO, black arrow showed shrunken neurons with pyknotic nuclei. Scale bars: 50 μm. changes in protein expressions were obviously reversed ( Figure 7A). Consistently, positive cells of TUNEL staining decreased in MCAO mice treated with AAV-Parp16 shRNA ( Figure 7B). The expressions of inflammationrelated proteins (COX-2, iNOS, VCAM-1, and IL-6) were significantly reduced upon PARP16 knockdown ( Figure 7C). Immunohistochemical staining further showed that PARP16 knockdown reduced the expression of iNOS in MCAO-treated mice ( Figure 7D). Caspase-1(p20), IL-1β, and IL-18 were increased in the damaged hemisphere of mice after MACO, but PARP16 knockdown could significantly reverse the expression change ( Figure 7E). Furthermore, breakdown of the blood-brain barrier (BBB) has been reported in stroke pathology. 22 We examined the effect of PARP16 knockdown on the expression of some proteins important to tight junction structure and function, including plasma protein ZO-1 and membrane protein Claudin-1. As shown in Figure 7F, the expressions of Claudin-1 and ZO-1 in the damaged hemisphere were significantly reversed upon PARP16 knockdown. These results suggest that PARP16 knockdown reduces I/R-induced apoptosis and neuroinflammation.

| DISCUSSION
Here, we report ER-associated tail-anchored protein PARP16 selectively (ADP-ribosyl)ates PERK and IRE1α during the UPR, and that such modification can lead to activation of PERK and IRE1α, which partly contributes to brain ischemia injury. In contrast, PARP16 knockdown has anti-apoptotic properties and prevents cell death in OGD insults and after cerebral I/R injury, therefore suggesting that it may be a potential treatment strategy for neuroprotection in ischemic stroke.
It is our current understanding that transient brain ischemia may trigger the accumulation of misfolded proteins in the ER lumen, thereby resulting in ER stress and activation of the UPR. 23,24 UPR activation may represent an inherent protective mechanism to avoid ER stress and to reestablish homeostasis in the cell. 25 On the other hand, under persisting or excessive ER stress, the adaptive ER responses fail and prolonged UPR may also ignite pro-apoptotic response. 26,27 It has been suggested that ER stress-induced neuronal cell death plays an important role in stroke pathophysiology, 28,29 inhibition of ER stress seems to be neuroprotective after cerebral ischemia/ reperfusion injury. 30,31 Our study reveals a progressive increase in the ER stress levels and the ER stress-induced apoptosis after MCAO, which could be responsible for the increase in the infarct volume. PARP16 is a novel ADPribosyltransferase associated with the endoplasmic reticulum. It is interesting that OGD treatment dramatically upregulates PARP16 expression in neurons and PARP16 might have a role in ischemia injury through its ADPribosylation of two components of the UPR. Indeed, ADP-ribosylation of PERK and IRE1α increased under ER stress, suggesting that ADP-ribosylation by PARP16 is sufficient to activate PERK and IRE1α. In our study, we found OGD stress caused ADP-ribosylation of PERK and IRE1α of PARP16 proximal signaling components, as well as resulted in robust phosphorylation of PERK and eIF2α. The requirement of PARP16 for PERK and IRE1α activation suggests that PARP16 functions upstream of PERK and IRE1α and mediates downstream signaling cascades.
In the present study, we found upregulation of proapoptotic signaling, which is involved in ER stress-induced cell death in OGD-induced cells and in the damaged hemisphere of mice. In addition, the inflammatory response and NLRP3 inflammasome participate in the ischemic reperfusion process, and inhibiting the inflammasome signaling pathway may be a possible strategy for treating this disease. Ischemic injury activates pro-Caspase-1 to generate active Caspase-1(p20), and activated Caspase-1 processes and matures IL-1β/18. 32,33 Previous studies have reported that ER stress can activate the NLRP3 inflammasome in various diseases and ischemia-reperfusion insult. [34][35][36] In our study, we found IL-1β/18 release in OGD-induced neuron and the damaged hemisphere of mice. PARP16 functions as upstream of PERK and IRE1α and plays a role in ER stress. Consistent with this notion, our observation found shRNA mediated PARP16 knockdown not only inhibited apoptosis, but also rescued neuronal inflammation via PARP16 signaling axis in cerebral ischemia injury. In contrast, PARP16 overexpression promoted ER stress-mediated apoptosis and inflammatory response in cortical neurons.
(ADP-ribosyl)ation of PERK and IRE1α was increased under OGD stress, suggesting that (ADP-ribosyl)ation by PARP16 is sufficient to activate PERK and IRE1α. PARP16 knockdown similarly failed to activate PERK or IRE1α and significantly inhibited the phosphorylation of PERK or IRE1α, tended to inhibit the ER stress, and also seemed to be neuroprotective after I/R injury, as revealed by reduced infarct volume. Therefore, it appeared that blocking PARP16 might act via PERK and IRE1α for its neuroprotective effect. Under these experimental conditions, blocking PARP16 or downstream signals appeared to be neuroprotective.
Our further investigation showed that PARP16 knockdown could reduce apoptotic signaling, but the ER activator Brefeldin A could partially counteract this effect on apoptotic signaling. Meanwhile, PARP16 knockdowninhibited IRE1a phosphorylation partially was reactivated upon an ER activator Brefeldin A. On other hand, in presence of an ER inhibitor TUDCA, the results showed neuronal protection with little difference compared to just PARP16 knockdown. This is consistent with results that TUDCA reduces apoptosis and protects against neurological injury. 35,37 In view of the present results, we further confirm that ER stress-induced neuronal cell death plays an important role in stroke pathophysiology, and that inhibition of ER stress provides neuroprotective from ischemic injury.
reperfusion, and that PARP16 may be a key molecule in ER stress-mediated neural damage. However, an important limitation worth noting is that only male mice were studied. We show that PARP16 selectively (ADPribosyl)ates PERK and IRE1α during the UPR, and that such modification is required for activation of PERK and IRE1α, at least in part contributing to the pathogenesis of cerebral I/R injury. Importantly, we show that the blocking PARP16 attenuates ischemic damage and partly restores the neurologic functions impaired by I/R, therefore suggesting that it may be a novel therapeutic target for ischemic stroke.

ACKNOWLEDGMENTS
We thank Dr. Jun Chang for technical assistance and discussion.

AUTHOR CONTRIBUTIONS
Jinghuan Wang and Jie Xu designed the research, performed experiments, and wrote original paper. Yejun Dong, Zhenghua Su, Haibi Su, and Qianwen Cheng performed some experiments and conducted analysis, and interpreted results. Xinhua Liu supervised the whole study process and revised the manuscript. All authors have read and approved the version of final manuscript.

FUNDING INFORMATION
This work was supported by grants from Shanghai Municipal Science and Technology Major Project (Grant No. 2017SHZDZX01).

DISCLOSURES
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available in the results and/or supplementary material of this article.