Pre‐ and post‐conditioning with poly I:C exerts neuroprotective effect against cerebral ischemia injury in animal models: A systematic review and meta‐analysis

Abstract Background Toll‐like receptor (TLR) agonist polyinosinic–polycytidylic acid (poly I:C) exerts neuroprotective effects against cerebral ischemia (CI), but concrete evidence supporting its exact mechanism of action is unclear. Methods We evaluated the neuroprotective role of poly I:C by assessing CI indicators such as brain infarct volume (BIV), neurological deficit score (N.S.), and signaling pathway proteins. Moreover, we performed a narrative review to illustrate the mechanism of action of TLRs and their role in CI. Our search identified 164 articles and 10 met the inclusion criterion. Results Poly I:C reduces BIV and N.S. (p = 0.00 and p = 0.03). Interestingly, both pre‐ and post‐conditioning decrease BIV (preC p = 0.04 and postC p = 0.00) and N.S. (preC p = 0.03 and postC p = 0.00). Furthermore, poly I:C upregulates TLR3 [SMD = 0.64; CIs (0.56, 0.72); p = 0.00], downregulates nuclear factor‐κB (NF‐κB) [SMD = −1.78; CIs (−2.67, −0.88); p = 0.0)], and tumor necrosis factor alpha (TNF‐α) [SMD = −16.83; CIs (−22.63, −11.02); p = 0.00]. Conclusion We showed that poly I:C is neuroprotective and acts via the TLR3/NF‐κB/TNF‐α pathway. Our review indicated that suppressing TLR 2/4 may illicit neuroprotection against CI. Further research on simultaneous activation of TLR3 with poly I:C and suppression of TLR 2/4 might open new vistas for the development of therapeutics against CI.

cognitive decline. 14,15 Patients that survive severe CI bear an exorbitant financial burden, where total medical care cost associated with stroke for Americans will soon be around $3 trillion from 2005 to 2050 (inflation-adjusted to 2021). 16 If the high mortality and financial impact of CI are to be mitigated, novel therapy development targeted against CI is necessary.
Toll-like receptors (TLRs) are expressed in astrocytes, microglia, and endothelial cells and are upregulated in the brain following ischemia. 17 Although activation of TLRs signaling worsens stroke injury, some reports suggest that the TLR3 activation by polyinosinicpolycytidylic acid (poly I:C) mediates innate immune responses and thereby provides neuroprotection. 18,19 On the other hand, no significant differences in indicators of neuroprotection such as brain infarct volume (BIV) and neurological deficit score (N.S.) between TLR3 knockout (KO) and wild-type (WT) mice after CI was observed, 20 indicating that the TLR3 signaling pathway might not be directly linked to ischemic brain injury. 21 Thus, identifying the receptor and transduction pathways involved with TLR signaling is critical for outlining drug delivery methods.
A safe and effective form of the neuroprotective drug against CI remains elusive. Poly I:C, a synthetic analog of double-stranded RNA (dsRNA), has shown promise as a neuroprotective agent in rodents with induced ischemic stroke in both adults 19,22,23 and neonates. 24,25 However, Stridh et al. 26 demonstrated that preconditioning (preC) of poly I:C might increase brain BIV and augment the loss of myelin basic protein by a factor of 5, which indicates an increased vulnerability of the neonatal brain to CI compared to control. Current literature on this topic is divided and the neuroprotective effect of poly I:C is currently inconclusive. Therefore, a meta-analysis is warranted to estimate the therapeutic potential of poly I:C against CI.
Primarily, we set out to clarify the therapeutic potential and signal transduction pathways of poly I:C following CI in rodent models.
Brain infarct volume, N.S., cell death, and level of poly I:C-associated pathway protein were chosen as outcomes for our meta-analysis to measure the therapeutic potential of poly I:C. Secondly, as other TLRs can be a potential therapeutic target against CI, we narratively reviewed various signaling pathway of TLR and their role in stroke.

| Searchstrategyandinclusioncriteria
Research articles reporting on the intervention of poly I:C in ischemic animal models were included in this systematic review and metaanalysis. The literature search was executed using keywords such as "poly I:C" in combination with "ischemia, reperfusion, hypoxia," in PubMed, and Embase, for studies published up until March 8, 2021. References (reverse citation tracking) and citations (forward citation tracking) of included studies were examined to identify any remaining studies or unpublished studies such as preprints that might be missing from our literature search results. The search strategies are given in Table S1 and S2. No limits on language or publication date were used. The full inclusion criteria used in our study are available in Appendix 1.

| Studyselectionanddataextraction
Two authors (Z.A.K. and D.M.S.) individually screened relevant articles by title and abstract, and full texts to identify articles that fulfilled our eligibility criteria. Disagreements regarding study selection were settled by a discussion with a third and fourth author (J.C. and G.C.K.).
Two authors (Z.A.K. and D.M.S.) individually extracted the data from selected studies. Information related to the authors, country, year of publication, sex, age, sample size, and outcome measures was extracted. The full details of data extraction used in our study are available in Appendix 1.

| Qualityassessmentoftheselectedarticle
To assess various aspects of bias related to animal interventionbased studies, the risk of bias (RoB) was estimated by two reviewers independently using the SYRCLE RoB tool (Cochrane RoB tool). 27,28 The RoB tool covers 10 items related to six different types of bias including selection, performance, detection, attrition, reporting, and other bias. Responses of "yes," "no," and "unsure" indicated low, high, and unclear RoB, respectively.

| Analysisofextracteddata
The extracted data from the selected studies were then entered into Stata SE 16 software. For effect size analysis, the standardized mean difference (SMD; Hedges'g) was used when studies assessed the same outcome by different measurements. 29,30 The mean difference (MD) was used whenever the outcome measures of associated studies utilized the same scale without other significant differences. 31 The detailed data analysis methods used in our study are available in Appendix 1.

| Studysearchandselection
In total, 164 studies were found through our PubMed and Embase electronic database search. Duplicate studies were removed, 137 potentially relevant articles were screened based on title and abstract screening, and 120 articles were excluded. The full-text screening was performed with the remaining 17 studies and 7 studies were excluded due to: unrelated outcomes (n = 2), conference abstract (n = 2) inappropriate study design (n = 1), and unrelated population (n = 2) (Table S3). Finally, 10 studies 19,21-26,32-34 fulfilled our eligibility criteria and were selected for systematic review and meta-analysis ( Figure 1).

| Studycharacteristicsandriskof biasassessment
The key characteristics of the studies included for systematic review and meta-analysis are presented in Table 1. The included studies were published between 2011 and 2020. MCAo was used to induce brain ischemia in 8 of the 10 included studies, 19,[21][22][23]25,[32][33][34] while two studies used carotid artery ligation. 24,26 All the included studies used male animals except Stridh et al., 26 who used both male and female mice.
Eight studies used mice, 19,21,[24][25][26][32][33][34] only one study used rats 23 and both mice and rats were used in one study. 22 The four studies were from the USA 21,25,32,33 and China, 19,22,23,34 while one study was from Canada 24 and Sweden, 26 respectively. Eight studies used an intraperitoneal injection of poly I:C, [21][22][23][24]26,[32][33][34] one used intramuscular 19 and subcutaneous 25 each. Various concentrations of the poly I:C ranging from 0.3 to 4 mg/kg were used in the included studies. The RoB analysis summary and individual RoB scores have shown in Figure 2A,B, and a detailed description of RoB is available in Appendix 1.
Using a random-effects model, we found that poly I:C significantly poly I:C is neuroprotective against CI ( Figure 4). We also performed meta-regression of the BIV with various doses of poly I:C and found no correlation between them (R 2 = 0.54 and p = 0.19) ( Figure S1).
The positive correlation between BIV and the N.S. is well established. 35 Our N.S. meta-analysis with poly I:C and CI consist of five studies. 19,22,23,25,34 Using a random-effects model, we found that poly I:C reduces the N.S., which is an indicator of better functional recovery  Figure 5A). The effects of poly I:C on cell death were assessed from three studies. 22,34 The cell death was measured by TUNEL assay. Using a random-effects model, we found that poly I:C intervention did not affect cell death 3.3.2 | Effect of poly I:C on the level of TLR3, NF-κB, and TNF-α The change in TLR3 and NF-κB level in the brain after CI injury was determined based on two 22,33 and three studies, 22,24,32 respectively.
All the studies quantified TLR3 by Western blot, 22,33 while NF-κB level was measured by the western blot 22,24 and electrophoretic mobility shift assay. 32 Using a fixed-effects inverse variance model, we found that poly I:C level significantly upregulates the level of TLR3 treatment on the level of IRF3 and GFAP, respectively. Using a random-effect model, we found that poly I:C has no effect on the IRF3 level in the brain after CI [I 2 = 94.36%; SMD = 3.62; CIs (−1.24, 8.47); p = 0.14] ( Figure 8A). The reason behind the high heterogeneity in IRF3 results was explored by leaving one publication/year out (Appendix 1). On the other hand, we applied fixed-effects inverse variance model and found that poly I:C significantly downregulated the level of GFAP in the brain after CI [I 2 = 0.00%; SMD = −3.38; CIs (−4.75, −2.01); p = 0.00] ( Figure 8B).

| AsystematicreviewofeffectpolyI:C interventiononapoptosis-relatedproteinsafterCI
Neuronal apoptosis plays a key role in CI injury. 36  other key markers of apoptosis. 37 Zhang et al. 32 showed that poly I:C prevented hypoxia ischemia-induced caspase-3 and −8 activity in microglial cells, indicating that poly I:C may attenuate microglial activation and apoptosis in response to ischemic stimulation. The results further support the hypothesis that the neuroprotective effect of poly I:C treatment is due to the reduction of apoptosis.
Cell death or survival pathways were conjointly affected with a rise in expression of the apoptosis-associated factor Fas, whereas pro-survival pathways including AKT phosphorylation were reduced. As opposed to other neuroprotective reports, Stridh et al. 26 showed that poly I:C treatment transiently downregulated the Akt phosphorylation and upregulated Fas ligand mRNA in the neonatal rodent ischemia model. This finding was in agreement with the study showing that poly I:C exacerbated neurodegeneration by the upregulation of Fas ligand. 38 These results indicate that poly I:C might negatively impact cell survival.

| AnarrativereviewofTLRsignalingpathways
Eleven types of TLRs are known to exist in humans, with 13 in animals. 39

| Toll-like receptors in CI
Toll-like receptors and their ligands play a significant role in the repercussion of brain ischemia. Toll-like receptors mediate inflammatory responses in immune cells, suggesting that these receptors aid in causing ischemia damage. After brain ischemia, astrocytes and mi- Each item in the SYRCLE tool was scored as "yes," "no," or "unclear" NE-induced neutrophil extracellular trap formation. TLR3 appears to be implicated in brain ischemia. The totality of TLR3 signaling is widely explored in the meta-analysis section; therefore, our focus will be on other TLRs which play a critical role in stroke.
The role of TLRs in CI is well studied and mostly intended to determine which TLR subpopulations are required for the development of ischemic damage in the brain. For example, TLR2 upregulation is linked with sterile inflammation, one type being ischemic brain injury. Ziegler et al. 49 showed that TLR2 acts on glial cells and proinflammatory factors that contribute to the spread of brain injury.
Toll-like receptor 2 has been shown to cause leukocyte infiltration into the injured region via a broken blood-brain barrier (BBB), as well as the subsequent activation of neuronal death. 50 Following brain ischemia in mice, TLR2 mRNA levels in resident microglia rises and TLR2 can binds to endogenous ligands. After ischemia injury, High mobility group box 1 (HMGB1), which is regarded as an important damage-associated molecular pattern (DAMP) in ischemic damage, is localized in the cell nucleus and translocated into the cytosol to activate TLR2. 51 High-mobility group box 1 neutralizing antibodies have been shown to decrease infarct volume following ischemic injury. 52 Another DAMP having a neuroprotective effect is the peroxiredoxin family protein, which is expressed in the damaged region.
Neurons of TLR2-deficient mice are also protected against cell death caused by an ischemia-like energy deprivation paradigm. In addition,

TLR2-KO animals showed reduced CNS injury after localized CI.
These findings imply that the TLR2-CD36 complex is important for inflammatory responses and may operate as a key marker of ischemia at the initiation of death signals. As a result, TLR2 inhibition might be explored in the future as potential therapy against ischemic stroke. Other TLR family members, such as TLR4 are thought to play a key role in the development of BIV in the ischemic brain by binding to endogenous ligands like HMGB1, which trigger immune cell infiltration into the infarct location and its surrounding areas via the BBB. Following cerebral ischemia, TLR4 gene expression is upregulated in neurons, along with an increase in inflammatory cytokines.
Knocking down the TLR4 gene in neurons can help them survive in glucose-depleted environments. This was also seen in TLR4deficient animals, which had smaller infarct volumes than WT mice.

F I G U R E 3
Forest plot comparing changes in BIV between poly I:C and vehicle-treated groups. Compared with vehicle treatment, BIV was significantly reduced in the poly I:C group. The prism represents the overall statistical results of the experimental data, squares represent the weight of each study, and horizontal lines represent the 95% CIs for each study. Normality of BIV was checked using the Shapiro-Wilk test (p = 0.76329). BIV, brain infarct volume; poly I:C, Polyinosinic:polycytidylic acid; CIs, Confidence intervals; SD, Standard deviation; IV, Independent variable

| DISCUSS ION
Our meta-analysis showed that both preC and postC with poly I:C offer substantial neuroprotection against CI. Overall, poly I:C lowers BIV, improves N.S., but does not affect cell death. Following CI, the generation of infarct in the brain is the final pathological step leading to neurological deficits. Injury severity is not directly proportional to the size of the brain infarct. 54 For example, slight injury in the medial temporal lobe may lead to severe disability such as speech impairment. However, a considerable injury to other parts of the brain may exert a mild functional deficit. Hence, one of the main aims of the CI treatment is functional recovery from neurological damage such as improvement in spasticity or limb impairment. 55,56 Thereby, N.S. is generally studied together with BIV. 57  showing that poly I:C treatment significantly reduces brain cell death compared to control, 24,32 suggesting that poly I:C treatment lowers the cell death after CI.
Our results also suggest that poly I:C significantly upregulates TLR3 levels and prevented ischemia-induced upregulation of NF-κB. Likewise, the level of TNFα was also downregulated. Also, the results of TLR3, NF-κB, and TNFα showed low heterogeneity, which is an indicator of robust results. These results showed that poly I:C acts through TLR3/NF-κB/TNF-α pathway. Our results reject the hypothesis that poly I:C acts independent of TLR3 and supports the previous findings that poly I:C stimulates TLR3, 58 which downregulates the production of TNFα by the NF-κB signaling pathway. 59,60 Even though our narrative review indicated that TLR3 can activate the IRF3, our meta-analysis showed that the level of IRF3 was Our narrative review suggests that TLR2 and TLR4 are known to have a larger role in the pathological progression of ischemic brain injury than other TLRs. As TLR4 suppression can downregulate both MyD88 and TRIF signaling, it can be a powerful neurotherapeutic target. Caso et al. 53 showed that TLR4-KO mice have minor infractions and less inflammatory response but no change in IL-1β and TNFα levels after an ischemic insult than wild-type animals. Subsequently, under ischemic stroke settings, Nalamolu et al. 62 reported that simultaneous TLR2/TLR4 suppression is more effective than individual suppression, which they conclude is achieved by reducing the production of pro-inflammatory cytokines TNF, IL-1, and IL-6. As a result, TLR2 and TLR4 might be regarded as potential stroke therapeutic targets.

F I G U R E 7
Forest plot comparing changes in TLR3, NF-κB, and TNFα in the brain between poly I:C and vehicle-treated groups following cerebral ischemia. Compared with vehicle treatment, (A) TLR3 was increased, (B) NF-κB and (C) TNFα, were significantly reduced in the poly I:C group. The prism represents the overall statistical results of the experimental data, squares represent the weight of each study, and horizontal lines represent the 95% CIs for each study. Normality of NF-κB was checked using the Shapiro-Wilk test (p = 0.70173). Poly I:C, polyinosinic:polycytidylic acid; TLR3, Toll-like receptor 3; NF-κB, nuclear factor-κB; TNFα, tumor necrosis factor alpha; CIs, confidence intervals; SD, standard deviation; IV, independent variable One of the key strengths of this study is combined as well as a separate meta-analysis of preC and postC of poly I:C. Interestingly, we observed that both preC and postC offer neuroprotection against CI.
Moreover, we applied a comprehensive search strategy, had access to the full texts of all identified studies, used the SYRCLE RoB tool to assess the methodological quality of the studies, and relevant extracted data. Furthermore, we performed a detailed subgroup and sensitivity analysis to validate our findings. On the other hand, the postC was performed only in the hyperacute phase of the ischemia, thereby inciting therapeutic uncertainty of poly I:C in later phases of ischemia.
Therefore, further research in acute, subacute, and chronic phases of CI is required to establish the therapeutic potential of poly I:C. Another limitation of this work is the high RoB in most of the included studies.
Methodological and reporting limitations in reporting/designing are common in animal studies and prevent us from reaching plausible conclusions. 63,64 Improvements in the preclinical data reporting should be paramount and guidelines regarding the reporting of animal studies should be followed to enhance the quality of research. 32,65,66 Although the potential source of heterogeneity has been investigated through various subgroup and sensitivity analyses, N.S. showed a moderate level of heterogeneity. Therefore, further research is warranted to establish the neuroprotective role of poly I:C.

| CON CLUS ION
Our meta-analysis showed that preC or postC with poly I:C offers neuroprotection against CI. The findings of this study suggest that poly I:C administration reduces BIV, N.S., and brain cell death. We conclude that poly I:C is a potential therapeutic agent for attenuating neuronal damage and promoting recovery after brain ischemia.
Our results reveal that TLR3, NF-κB, and TNFα could be utilized as predictive biomarkers for poly I:C treatment against cerebral ischemia injury. Furthermore, our narrative review highlights the importance of using multiple TLRs inhibitors against stroke. Also, it will be interesting to study the combined effect of a TLR activator such as poly I:C with TLR 2/4 inhibitors. TLRs' molecular structure, genetic differences, and regulation by a variety of reagents can all be used to assist manage stroke prevalence and therapies in the future.

F I G U R E 8
Forest plot comparing changes in IRF3 and GFAP protein level in brain between poly I:C and vehicle-treated groups following cerebral ischemia. No change was observed in the (A) IRF3 protein level, while (B) GFAP protein was downregulated in the poly I:C group. The prism represents the overall statistical results of the experimental data, squares represent the weight of each study, and horizontal lines represent the 95% CIs for each study. Normality of IRF was checked using the Shapiro-Wilk test (p = 0.57957). Poly I:C, polyinosinic:polycytidylic acid; IRF3, interferon regulatory factor 3; GFAP, glial fibrillary acidic protein; CIs, confidence intervals, CI; SD, standard deviation; IV, independent variable

ACK N OWLED G M ENT
The authors would like to acknowledge the invaluable support and critical comments of members in "Biological Clock & Aging Control" laboratory. The graphical abstract and figure 9 were created with BioRender.com, and a current subscription was valid at the time of submission and publication.

CO N FLI C TO FI NTE R E S T
The authors declare no conflict of interest.

DATAAVA I L A B I L I T YS TAT E M E N T
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.