2.1 Human Brain Tissue Samples

All paraffin-embedded human postmortem brain tissue samples (three stroke and three non-stroke patients) were obtained from the Anhui Provincial Key Laboratory for Brain Bank Construction and Resource Utilization. Informed consent for use of the human tissues for research was obtained in writing from all donors or the next of kin. All subjects were defined as normal controls or IS samples by pathologists. No subjects with prolonged agonal state were used; cause of death and related information are shown in Table S1. All the experiments related to human tissues were conducted in accordance with the Helsinki criteria and were also approved by the Ethics Committees of Anhui Medical University with the approval number 20200021.

2.2 Animals

Male Sprague–Dawley rats (6–8 weeks old, weight 150–200 g) were obtained from the Zhejiang Vital River Laboratory Animal Technology Company and kept under SPF conditions in the Experimental Animal Center of Anhui Medical University for 1 week prior to the experiment. Rats were housed in a temperature- and humidity-controlled condition with a 12 h light/dark cycle. Standard laboratory food pellets and water were available ad libitum. All the experimental procedures for animal surgery were approved by the Animal Ethics Committee of Anhui Medical University with the approval number SCXK2019-0001.

2.3 Transient Middle Cerebral Artery Occlusion Model

The focal cerebral ischemia rat model was performed as described previously [35]. Rats were anesthetized with 2% isoflurane mixed with air, and the body temperature was maintained using a homoeothermic blanket. Focal cerebral ischemia was induced by using MCAO with a suture [36]. Briefly, a nylon monofilament was inserted to the origin of the right MCA of rats. The filament was withdrawn at 2 h after occlusion to allow for reperfusion for 24 h. Sham-operated rats underwent similar surgical procedures except for filament insertion. After the operation, mice were kept in warm cages for 2 h for recovery. The other details of materials and methods are presented in the Supporting Information.

3.1 RNF2 Is Upregulated in the Brain After Focal Cerebral Ischemia

Given that PcG proteins are increased in the ischemic-tolerant rodent brain and overexpressing the PcG family proteins SCMH1 or BMI1 induces tolerance to ischemia, we asked whether RNF2 was involved in IS injury. We first analyzed the expression of RNF2 in human postmortem brain tissue samples with IS and rat brain under MCAO, respectively. RNF2 was widely expressed in the human ischemic brain compared to the control brain (Figure 1A,B). Consistently, the expression of RNF2 was also increased in the peri-infarct area of I/R rats subjected to MCAO for 2 h followed by reperfusion for 24 h, which exhibits massive necrosis (Figure 1C,F,G) and infarction volume (Figure 1D,E). Furthermore, both protein and mRNA levels of RNF2 were upregulated in the ischemic brain, as revealed by quantitative polymerase chain reaction (qPCR) (Figure 1J), western blot analysis (Figure 1H,I), and relative qPCR (Figure S1A,B), implying a potential involvement of RNF2 in IS. To further verify if RNF2 was regulated by ischemic insult, we examined RNF2 expression in cultured neurons after OGD/R treatment and found that the RNF2 level was significantly increased, which was consistent with that in the MACO brain (Figure 2M,N).

Next, we examined the distribution of RNF2 in the human and rat brain. RNF2 was widely expressed in the brain with high levels under the ischemic condition but weakly expressed in the control brain. To investigate the cellular localization of RNF2 in the brain, we examined the expression of RNF2 in human and rat brains. Pentuple-fluorescence immunofluorescent staining showed that most RNF2-positive cells were NeuN-positive neurons, while fewer RNF2-positive cells were GFAP-positive astrocytes or CD68-positive microglia, but there were no CNP-positive oligodendrocytes in I/R brain tissues (Figure 3A–D). Consistently, double-labeled immunofluorescent staining showed that RNF2 can be differentially expressed in glial cells, including astrocytes and microglia, and that neurons are the major source of RNF2 in the rat MACO brain (Figure 3E). These results, together with our previous findings, indicate that I/R upregulates the expression of RNF2, especially in neurons.

3.2 RNF2 Overexpression Ameliorates Brain Injury Induced by MCAO

To explore the functional role of RNF2 in I/R injury, mCherry-fluorescent-labeled AAV2/9 virus carrying the RNF2 was stereotaxically injected into the unilateral lateral ventricle 3 weeks before MACO (Figure 2A and Figure S2A). Whole brain fluorescence images were collected using an in vivo imaging system at 3 weeks after AAV-RNF2 or AAV-Con injection, and fluorescent signals can be detected both in the left and in right hemispheres of the brain (Figure 2B and Figure S2A). Compared with that of AAV-Con-infected rats, RNF2 expression was acutely upregulated both in sham and I/R brains, indicating that RNF2 was successsfully transfected into brains (Figures S3A,B and S2B). In addition, most of mCherry-positive cells are NeuN-positive neurons, but also existed in some NeuN-negative cells, suggesting that AAV-RNF2 leads to a wide expression of RNF2 in the ischemia brain (Figure S2C). RNF2 overexpression markedly reduced the infarct volume after reperfusion compared with AAV-Con-injected rats (Figure 2C,E). Compared to the AAV-Con-injected I/R brain, the number of NeuN-positive cells is remarkably increased in AAV-RNF2-injected rats following I/R induction, although there are no difference between AAV-Con- and AAV-RNF2-transfected sham rat (Figure 2D,F), suggesting that RNF2 alleviates cerebral injury induced by IS.

3.3 RNF2 Inhibits I/R-Induced Neuron Apoptosis and Protects Against TM and OGD/R-Induced Nerve Cells Apoptosis

Given that RNF2 overexpression obviously increased NeuN-positive cells after I/R injury, we wondered whether RNF2 facilitates neuroprotection against cerebral ischemia via inhibiting neuronal apoptosis. As we expected, RNF2 overexpression alleviated neuronal apoptosis in the cortex after MCAO, as shown by TUNEL staining (Figure 2G,H) and cleaved caspase-3 (c-casp-3) (Figure 2I,J and Figure S4A–C), indicating a protective role of RNF2 in IS injury. Next, we used OGD/R and TM to mimic I/R in vitro and observed the effect of RNF2 overexpression or knockdown on neuronal apoptosis. TM treatment (2.5 μg/mL, 16 h) increased the expression of RNF2 in N2a cells, accompanied by upregulated levels of the ER stress marker BIP and pro-apoptotic markers CHOP and c-casp-3 (Figure 2K,L,O,P). Consistently, glucose reoxygenation following OGD for 1, 2, 3, and 4 h gradually upregulated the levels of RNF2, BIP, CHOP, and c-casp-3 in SH-SY5Y cells, especially at 3 h hypoxia and 24 h reoxygenation (Figure 2M,N,Q,R). RNF2 overexpression inhibited the activation of caspase-3 and CHOP expression both in TM and OGD/R cells (Figure 2O–R), suggesting that RNF2 playing a protective role again I/R induced neuronal apoptosis. Then N2a cells were transiently transfected with RNF2-siRNA and treated with TM for 16 h (Figure 2S,T). SH-SY5Y cells were transiently transfected with RNF2-siRNA and treated with OGD for 3 h and reoxygenated for 24 h (Figure 2U,V). The upregulated levels of c-casp-3 and CHOP in the RNF2 knockdown group suggested that RNF2 knockdown aggravates TM and OGD/R-induced neuronal apoptosis. Additionally, RNF2 knockdown significantly increased the number of apoptotic cells detected by flow cytometry (Figure S5A,B). Together, these data show that RNF2 alleviates ischemia-induced neuronal apoptosis.

3.4 RNF2 Interacts With MANF in the Nuclei

To explore the mechanism underlying the neuroprotective effect of RNF2 in focal cerebral ischemia, we screened the RNF2-interacting proteins using a yeast two-hybrid method and found that MANF may be the candidate for RNF2. The interaction between MANF and RNF2 was verified by yeast back hybridization in which pGBKT7-MANF and PGADT7-RNF2 were co-transformed into yeast gold strain. We found that the co-transformed yeast grew normally and appeared blue on defective medium QDO/X, suggesting that RNF2 could interact with MANF in yeast (Figure 4A). Co-IP also confirmed the interaction between MANF and RNF2 in vivo (Figure 4B). As we mentioned before, ER stress inducer TM obviously upregulated MANF and RNF2 expression (Figure 4C,D). MANF is mainly located in the cytoplasm in normal cells, while TM elicits MANF translocation to the nucleus, which is colocalized with RNF2 in the N2a cell (Figure 4E,F). Consistently, RNF2 levels in the nucleus were increased in the brain tissue from stroke patients and I/R rats and co-existed with MANF under ischemic injury (Figure 4G,H). GST pull-down assay further confirm the direct interaction between MANF and RNF2 (Figure 4I). To further clarify the functional interacting domain between RNF2 and MANF, the plasmids carrying different RNF2 or MANF truncation mutants were transfected into N2a cells, following co-IP experiments. As shown in Figure 4I,K, the RNF2 mutant lacking the RING-finger domain failed to interact with MANF, while the MANF mutant lacking the C-terminal including SAP-like domain and RTDL sequence failed to interact with RNF2, indicating that the C-terminal of MANF was able to bind to the N-terminal of RNF2 (Figure 4J–M).

3.5 RNF2 Regulates the Stability of MANF by Monoubiquitylation

Since RNF2 is a E3 ubiquitin ligase, we suspected that MANF might be a substrate of RNF2. First, we investigated the mRNA and protein levels of MANF in N2a cells transfected with RNF2-overexpressing plasmids or siRNA, respectively, along with TM treatment. Obviously, the protein levels of MANF increased in RNF2 overexpressing cells and decreased in RNF2 knockdown cells (Figure 5A,B). Further study showed that RNF2 could dose-dependently upregulate MANF protein levels with or without TM treatment (Figure 5C,D). However, we found that the altered expression of RNF2 did not affect the mRNA levels of MANF (Figure 5E,F), suggesting that RNF2 regulation of the MANF level merely occurs at the protein level. To further assess the effect of RNF2 on MANF protein stability, we inhibited de novo protein synthesis by cyclohexamide (CHX) and measured MANF protein levels in the RNF2-overexpressed cells. Consistently, RNF2 significantly enlonged the half-life of MANF in N2a cells treated with CHX (Figure 5G,H), which indicated that RNF2 stabilized MANF protein expression.

Next, we assessed whether E3 ubiquitin ligase RNF2 could mediate the ubiquitination of MANF, which makes it more stabilized. Co-IP results showed that monoubiquitinated MANF was markedly increased in the presence of RNF2 overexpression (Figure 5I). The in vitro ubiquitination experiment further verifies that RNF2 facilitates MANF monobiquitination (Figure 5J). Together, these data show that RNF2 mediated MANF monoubiquitination to stabilize MANF.

3.6 RNF2 Enhances MANF Nuclear Translocation Under ER the Stress Condition

Given the functions of RNF2 in promoting the stabilization of MANF via monoubiquitination, we inferred that RNF2 might facilitate MANF nuclear translocation under stroke stress. We found that MANF mostly expressed in the cytoplasm in control but translocated into nuclear and colocalized with RNF2 upon TM stimulation. Moreover, RNF2 overexpression further increased the expression of MANF in the nucleus and facilitated the colocalization between RNF2 and MANF in the presence of TM (Figure 6A,C). To further verify the effect of RNF2 on MANF nuclear translocation, total cellular and nuclear proteins were extracted with nuclear plasma separation kit and western blot result showed that RNF2 enhanced nuclear MANF levels under TM treatment (Figure 6E,F). Consistent with this result, RNF2 knockdown attenuated TM-induced MANF nuclear translocation (Figure 6B,D,G,H).

Mechanistically, RNF2 stabilized MANF and facilitated MANF nuclear translocation under the ER stress condition.

3.7 RNF2 Inhibits Neuron Apoptosis Dependently on MANF

To investigate whether RNF2 functions dependently on MANF in the ischemic brain, c-casp-3 expression was detected in MANF knockdown cells. We found that MANF siRNA obviously enhanced c-casp-3 levels, but the effect was weakened to some extent in RNF2 overexpressing cells (Figure 6I,J, Lane 5 vs. Lane 2). rhMANF was proved to play a neuroprotective role in a series of brain injury, although the underlying mechanism is not very clear. Here, we observed that rhMANF treatment decreased caspase-3 activation whether RNF2 overexpressing or not, which is consistent with our previous studies (Figure 6I,J, Lane 3 vs. Lane 2, Lane 6 vs. Lane 5). However, in RNF2 overexpressing cells, c-casp-3 levels were more significantly relieved accompanied by rhMANF treatment (Figure 6I,J, Lane 6 vs. Lane 3). Flow cytometry analysis (Figure 6K,L) and TUNEL staining (Figure S6A,B) further proved that the inhibitory effect of RNF2 on neuron apoptosis might be abolished partially by MANF knockdown and reversed by rhMANF. These findings suggest that RNF2 inhibits neuronal cells apoptosis partly dependently on MANF.