Cytosolic receptor MDA5 mediates Poly-ICLC preconditioning induced neuroprotection against cerebral ischemic injury

Abstract

Background and Purpose

Preconditioning with poly-ICLC provides robust neuroprotection from cerebral ischemia in a mouse stroke model. However, the receptor that mediates neuroprotection is unknown. As a synthetic double stranded RNA (dsRNA) poly-ICLC may bind endosomal Toll-like receptor 3 (TLR3) or one of the cytosolic retinoic acid-inducible gene I (RIG-I)-like receptor family members, RIG-I or melanoma differentiation associated protein 5 (MDA5). Activation of these receptors culminates in type I interferons (IFNα/β) induction—a response required for poly-ICLC induced neuroprotection. In this study we investigate the receptor required for poly-ICLC induced neuroprotection.

Methods

TLR3-, MDA5- and IPS1-deficient mice were treated with poly-ICLC 24hr prior to middle cerebral artery occlusion. Infarct volume was measured 24hr after stroke to identify the receptor signaling pathways involved in protection. IFNα/β induction was measured in plasma samples collected 6hr after poly-ICLC treatment. IFNβ-deficient mice were used to test the requirement of IFNβ for poly-ICLC-induced neuroprotection. Mice were treated with recombinant IFNα-A to test the role of IFNα as a potential mediator of neuroprotection.

Results

Poly-ICLC induction of both neuroprotection and systemic IFNα/β requires the cytosolic receptor MDA5 and the adapter molecule IPS1, while it is independent of TLR3. IFNβ is not required for poly-ICLC induced neuroprotection. IFNα treatment protects against stroke.

Conclusions

Poly-ICLC preconditioning is mediated by MDA5 and its adaptor molecule IPS1. This is the first evidence that a cytosolic receptor can mediate neuroprotection, providing a new target for the development of therapeutic agents to protect the brain from ischemic injury.

Introduction

More than one million patients annually are at risk of brain ischemia and reperfusion injury that occurs secondary to life-saving endovascular or cardiac procedures.^{1, 2} Thus far, no treatment is available to confer brain protection in this ‘at risk’ patient population. Prophylactic brain protection can be achieved through preconditioning, a phenomenon whereby brief exposure to a potential harmful stimulus induces protection against a subsequent injury. We have previously published that preconditioning with polyinosinic polycytidylic acid stabilized with poly-L-lysine and carboxymethylcellulose (poly-ICLC), protects against cerebral ischemic injury in a mouse model of stroke resulting in reduced ischemic injury and attenuation of stroke induced neurological deficits.³ In addition we have shown that protection is dependent on type I interferon (IFN) signaling.⁴ As a synthetic dsRNA, Poly-ICLC may induce type I IFN signaling through the binding and activation of multiple different pattern recognition receptor signaling cascades. TLR3, RIG-I and MDA5, have all been shown to bind dsRNA, making them potential targets for poly-ICLC. While TLR3 localizes to the endosome, RIG-I and MDA5 are found in the cytoplasm.⁵ Identifying the specific receptor and intracellular signaling pathway engaged by poly-ICLC is important for defining drug delivery strategies for future translational development.

In this study we investigate the receptor mediated signaling cascade required for poly-ICLC induced neuroprotection against ischemic brain injury using TLR3^−/−, MDA5^−/− and IPS1^−/− mice. In addition, using IFNβ deficient mice, we show that IFN β is not involved in neuroprotection, and that exogenous administration of IFNα protects against stroke, supporting a role for IFNα in mediating poly-ICLC induced protection.

Materials and Methods

Mice

C57Bl/6 (WT), TLR3^tm1Flv (TLR3^−/−) and B6.Cg-^{Ifih1tm1.1Cln}/J (MDA5^−/−) mice were from Jackson Laboratories (West Sacramento, CA, USA). IPS1^−/− animals provided by Dr. Edith Janssen (University of Cincinnati, Cincinnati, Ohio). IFNβ^−/− mice provided by Dr. Tomas Leanderson (Lund University). Studies were performed with male mice between 8–14 weeks of age. Mice were given free access to food and water and housed in a facility approved by the Association for Assessment and Accreditation of Laboratory Animal Care International. Animal protocols were approved by the Oregon Health & Science University Institutional Animal Care and Use Committee (OWLAW# A3304-01) and met the guidelines set forth by the National Institutes of Health.

Study Design and Blinding

For each genotype (WT, TLR3^−/−, MDA5^−/−, IPS1^−/− and IFNβ^−/−), mice were randomly given a subcutaneous (s.c.) injection of either vehicle (carboxymethyl cellulose; Sigma Aldrich) or poly-ICLC (Hiltonol®; 60µg per mouse, gifted from Oncovir, Washington DC). For stroke studies, mice were treated 24hr before ischemia. To measure systemic type I interferon induction, mice were treated 6hr before blood collection. For studies involving IFNα, mice were injected intraperitoneally (i.p.) with vehicle (saline) or recombinant IFNα-A (1 × 10⁴ units/mouse, PBL Biomedical Laboratories) 18 hours prior to surgery. Surgery and analysis were performed by investigators blinded to treatment and genotype. A total of 73 WT, 22 TLR3^−/−, 25 MDA5^−/−, 25 IPS1^−/− and 8 IFNβ^−/− mice have been used in these studies.

Mouse ischemia-reperfusion model

Focal cerebral ischemia was induced by 60 or 45 minute middle cerebral artery occlusion (MCAO) as described previously.⁴ Mice that did not show >80% reduction in cerebral blood flow, monitored with laser Doppler flowmetry (Transonic System Inc.) were excluded. Twenty-four hours following MCAO infarct volume was measured as previously described.⁴

Cytokine Quantification

Plasma concentrations of IFNα and β were determined using a Mouse IFNα/β ProcartaPlex™ Multiplex Immunoassay (Affymetrix eBioscience) according to manufacturer instructions. Cytokine concentrations were extrapolated from the standard curve with samples below the standard curve assigned a value half that of the lowest standard.

Statistical analysis

Data are represented as group means ± standard error of the mean (SEM) and statistical analysis performed using GraphPad Prism6 software. Two-way ANOVA with Bonferroni post hoc test was used for infarct analysis and IFNα/β measurements.

Results

Poly-ICLC requires MDA5, but not TLR3, to induce neuroprotection

Potential cognate receptors for poly-ICLC include TLR3, and the RIG-I-like receptor (RLR) family members, RIG-I and MDA5 (Fig 1A). RIG-I and MDA5 recognize dsRNA of different size,⁵ thus we postulated that poly-ICLC (~500KDa) would bind to MDA5, which recognizes larger dsRNAs.⁶ To determine whether TLR3 or MDA5 mediated poly-ICLC neuroprotection, TLR3^−/− and MDA5^−/− mice were preconditioned with poly-ICLC 24hr prior to MCAO. As expected, poly-ICLC preconditioning significantly reduced ischemic injury in WT animals (15.18±2.5% versus vehicle 32.57±1.8%; Fig 1B). In addition, poly-ICLC preconditioning significantly reduced ischemic injury in TLR3^−/− mice (4.81±2.1 versus vehicle 29.14±4.1%; Fig. 1B). No significant difference between poly-ICLC treated WT and TLR3^−/− mice was evident. In contrast, poly-ICLC treated MDA5^−/− mice failed to show a reduction in ischemic volume (29.63±3.1% versus vehicle 32.1±1.8%; Fig.1B). These data indicate that poly-ICLC preconditioning induced neuroprotection is MDA5 dependent, demonstrating for the first time that a cytosolic receptor can mediate neuroprotection.

Figure 1. (A) Schematic representation of the potential intracellular signaling pathways activated by poly-ICLC. (B) Poly-ICLC is dependent on MDA5 and IPS1 for induction of neuroprotection.

WT, TLR3^−/− MDA5^−/− and IPS1^−/− mice were administered poly-ICLC (60µg) or vehicle 1 day prior to MCAO (60min) and ischemic injury was measured 24hr after MCAO. The WT mice used as controls for TLR3^−/−, MDA5^−/− and IPS1^−/− stroke studies were combined because no significant difference between the groups was evident. Data reported as group means ± SEM, n=7–27, two-way ANOVA and Bonferroni’s post hoc test, *** P<0.0001 vs. corresponding vehicle.

Poly-ICLC induced neuroprotection is mediated through IPS1

MDA5 signals through the adaptor molecule IFNβ-promoter stimulator (IPS1) leading to the induction of type I IFNs⁵ (Fig 1A). Therefore, we tested the involvement of IPS1 in poly-ICLC preconditioning using IPS1^−/− mice. IPS1^−/− mice preconditioned with poly-ICLC failed to show a significant reduction in ischemic injury (40.8±2.5% versus vehicle: 37.11±2.0%; Fig. 1B), demonstrating that poly-ICLC preconditioning depends on both MDA5 and IPS1 for the induction of neuroprotection.

Poly-ICLC induction of type I interferons requires MDA5 and IPS1, not TLR3

We have previously shown that poly-ICLC requires type I IFN signaling for protection.⁴ Therefore, we postulate that MDA5 and IPS1 mediate poly-ICLC induction of IFNα and/or IFNβ. We measured plasma IFNα and IFNβ levels 6hr after poly-ICLC treatment in WT, TLR3^−/−, MDA5^−/− and IPS1^−/− mice. Poly-ICLC significantly increased plasma levels of IFNα in both WT and TLR3^−/− mice (WT vehicle: 1.72 pg/ml versus poly-ICLC: 596.37±181.8 pg/ml; TLR3^−/− vehicle: 1.72 pg/ml versus poly-ICLC: 720.88±371.8 pg/ml; Fig. 2A). However, MDA5^−/− and IPS1^−/− mice showed no significant induction of IFNα in response to poly-ICLC (Fig. 2A). Similar results were obtained for IFNβ (Fig. 2B) with TLR3^−/− and WT mice inducing equivalent levels of IFNβ (WT 33.17±14.2 pg/ml versus TLR3^−/− 60.53±40.3 pg/ml; Fig. 2B), while no increase in IFNβ was detected in MDA5^−/− and IPS1^−/− mice. These results are consistent with poly-ICLC mediating protection through binding of the MDA5 receptor, signaling through IPS1 and subsequent induction of type I IFNs, independent of TLR3 signaling.

Figure 2. Poly-ICLC induction of IFNα/β is dependent on MDA5 and IPS1.

IFNα (A) and IFNβ (B) were measured in plasma samples collected from WT, TLR3^−/−, MDA5^−/− and IPS1^−/− mice 6hr following vehicle or poly-ICLC (60µg) treatment. Data reported as group means ± SEM, n=3–4, two-way ANOVA and Bonferroni’s post hoc test, ** P<0.001 and * P<0.05 vs corresponding vehicle.

IFNβ is not required for poly-ICLC neuroprotection

IFNβ has been shown to provide protection against cerebral ischemic injury in experimental models of stroke^7–9, thus we hypothesized that IFNβ may be the key mediator of neuroprotection. To test this we preconditioned IFNβ^−/− mice with poly-ICLC 24hr before ischemia. We found that IFNβ^−/− mice were significantly protected against ischemic brain injury (poly-ICLC 21.54±5.3% versus vehicle 41.91±5.8%, Fig. 3A), indicating that poly-ICLC induced neuroprotection is independent of IFNβ.

Figure 3. Poly-ICLC neuroprotection is independent of IFNβ while IFNα treatment confers protection against stroke.

A) IFNβ^−/− mice were administered poly-ICLC (60µg) or vehicle 1 day prior to MCAO (60min) and ischemic injury was measured 24hr after MCAO. B) Mice were administered recombinant IFNα-A (1 × 10⁴ units/mouse) or vehicle 18 h prior to MCAO (45min) and ischemic injury was measured 24hr after MCAO. Data reported as group means ± SEM, n=3–6, unpaired t-test, * P<0.05 vs vehicle.

Systemic IFNα protects against stroke

The previous data indicates that IFNβ is not required, indicating that another type I IFN receptor ligand, such as IFNα may mediate the protective response. To test this IFNα was administered 18h prior to MCAO. This timing corresponds with the increase in serum levels of IFNα induced by preconditioning with poly-ICLC. We found that IFNα treated mice showed a significant reduction in ischemic volume (IFNα 18.11±4% versus vehicle 33.19±5.1%, Fig. 3B) indicating that poly-ICLC induction of systemic IFNα may mediate protection against stroke.

Discussion

We demonstrate that poly-ICLC preconditioning mediates neuroprotection through the cytoplasmic receptor MDA5 and its adapter IPS1. We have previously published that the type I IFN receptor is required for protection.⁴ Here we show that, while systemic levels of both IFNα and IFNβ are increased in response to poly-ICLC preconditioning, the protective response is independent of IFNβ while direct treatment with IFNα protects the brain from ischemic injury, supporting a key role for IFNα in poly-ICLC neuroprotection. These findings broaden our knowledge of endogenous targets for the induction of neuroprotection against cerebral ischemic injury. In particular, the localization of these mediators of protection (both the receptor and its adapter molecule) in the cytoplasm may offer alternative drug delivery strategies for the translational development of therapeutic agents for stroke.

We found that neuroprotection induced by poly-ICLC preconditioning is mediated exclusively through MDA5 and independent of TLR3. This is in contrast to a recent study using native poly-IC in which acute neuroprotection against brain ischemic injury was mediated through TLR3.¹⁰ The apparent difference may relate to the chemical modification of poly-ICLC, which is a version of poly-IC that has been stabilized with poly-L-lysine and carboxymethylcellulose to improve pharmacokinetics. These alterations may influence the structure and trafficking of the compound resulting in different responses between poly-ICLC and native poly-IC. Consistent with this postulate, we found that although poly-ICLC was able to induce IFNα and β in TLR3^−/− mice (Fig. 2), poly-IC was unable to induce either IFN in the TLR3^−/− mice (data not shown). This difference is likely the result of altered trafficking of the two poly-IC compounds into the cell. This is supported by recent data by Dann et. al demonstrating that native (non-complexed) poly-IC is a ligand for TLR3,¹¹ while poly-IC in complex with polyethylenimine (PEI) targets the cytoplasmic receptor MDA5.¹² Poly-IC complexation with PEI changes its cell delivery making it available for the cytosolic receptors. We propose that the -LC component of poly-ICLC modifies the cell delivery as well, making poly-ICLC a specific ligand for cytosolic MDA5.

The association between stimulation of RLRs and reduction of central nervous system inflammation was previously reported by Dann et al., who showed that activation of RIG-I and MDA5 reduced macrophage and lymphocyte infiltration into the spinal cord in an experimental model of multiple sclerosis.¹¹ Similar to our results, this process was dependent on type I interferon receptor signaling.^{4, 11}

We and others, have shown that prophylactic administration of TLR ligands induces neuroprotection against MCAO in mice.¹³ Here we demonstrate for the first time that the pattern recognition receptor (PRR), MDA5, is involved in neuroprotection. This may indicate the existence of a conserved feature shared among PRRs for conferring protection against brain ischemic injury. This discovery has real significance in neuroprotection because it broadens substantially the possibilities of new approaches to achieve neuroprotection, particularly in the realm of preconditioning modalities.

Conclusions

Our results demonstrate that cytoplasmic receptors play an important role in preconditioning and that MDA5 and its adapter IPS1 are required for poly-ICLC induced neuroprotection. These findings increase our understanding of the neuroprotective mechanisms associated with poly-ICLC, furthering its development as a prophylactic treatment against ischemic injury. Furthermore, the results reported here expand our knowledge of potential endogenous targets for the development of new therapeutic agents for stroke.