Age-related deregulation of TDP-43 after stroke enhances NF-κB-mediated inflammation and neuronal damage

The wild-type (C57Bl/6) mice of 3 and 12 month old (representing a middle aged mouse group) were selected for study. The TLR2-luc-GFP transgenic reporter mice were developed, validated, and genotyped as described previously [23]. These animals do not develop any overt phenotype and were used for in vivo bioluminescence imaging analysis of microglia activation/innate immune response. The TDP-43 A315T transgenic mice were generated, described, and genotyped as described in [24]. The TDP-43 A315T mice develop age-related cognitive deficits resembling frontotemporal dementia phenotypes. All our transgenic colonies are kept in C57Bl/6 genetic background. All experimental animals used in this study were provided with water and healthy diet and were monitored during the entire experimental protocol. The animals were held in the pathogen-free animal facility of the CERVO Brain Research Institute, 3–5 mice per cage in the controlled environment having the 12 h day and night cycles. To avoid the biological effects of sex on ischemic injury, the experiments were performed on male mice. All the experimental procedures were approved by the Laval University Animal care Ethics Committee and are in accordance with the Guide to the Care and Use of Experimental Animals of the Canadian Council on Animal Care.

Transient focal cerebral ischemia was induced by unilateral left middle cerebral artery occlusion (MCAO) as described [23]. Wild-type mice of around 3–12 months were selected and unilateral transient focal cerebral ischemia was induced by intraluminal filament occlusion of the left middle cerebral artery (MCA) with a 6–0 silicone-coated monofilament suture for 1 h followed by reperfusion times of 24 h, 48 h, 72 h, 5 days, and 10 days after surgery. The body temperature was maintained at 37 °C with a heating pad.

The images were obtained by using IVIS 200 imaging system (Caliper LS-Xenogen, Alameda, CA, USA). Twenty minutes prior to imaging session, the mice were administered with d-luciferin (150 mg/kg bw), a substrate for luciferase dissolved in 0.9% saline. The mice were then anesthetized in 2% isoflurane in 100% oxygen at a flow rate of 2 L/min, placed in a heated light-tight imaging chamber. All the animals were imaged before for baseline expression and then continued at different time points post MCAO. Images were captured using a high sensitivity CCD camera with wavelengths ranging from 300 to 600 nm and exposure time for imaging of brain was set for 1 min. The bioluminescence emission was quantified by determining the total number of photons emitted per second (p/s) using live image 2.5 acquisitions and imaging software. Region of interest measurements were used to convert surface radiance (p/s/cm²/sr) to source flux or the total flux of photons expressed in photons/second. The data are represented as pseudo-color images indicating light intensity (red and yellow, most intense), which were superimposed over gray-scale reference photographs [25].

Cytoplasmic and nuclear fractions were obtained as per the protocol described earlier [26, 27]. Five hundred milligrams of fresh brain tissue samples were transferred to 1 ml of cell lysis buffer (10 mM HEPES, 10 mM NaCl, 1 mM KH₂PO₄, 5 mM MgCl_2, phosphatase, and protease inhibitors), and homogenized by applying two strokes in a glass homogenizer. The suspension was incubated for 10 min on ice and then homogenized by applying six strokes in a motorized homogenizer at 250 rpm. Differential centrifugation was performed after restoration with 100 μl of 2.5 M sucrose. The first round of centrifugation was performed at 6300×g for 10 min at 4 °C. The pellet was collected and suspended in TSE buffer (10 mM Tris, 300 mM sucrose, 1 mM EDTA, 0.1% Non idet P-40, 10× v/w, phosphatase, and protease inhibitors pH 7.5) and homogenized with 30 strokes using a motorized Teflon potter at 250 rpm. The obtained suspension was centrifuged at 4000×g for 5 min. The resulting supernatant was discarded, and the pellet was washed with TSE buffer until the supernatant was clear. Pellet was then suspended in RIPA buffer (50 mM Tris, 1 mM EDTA, 150 mM Nacl, 0.1% SDS, 1% Non-idet P-40, 0.5% sodium deoxycholate, phosphatase, and protease inhibitors) with 2% SDS as nuclear fraction. The supernatant collected from the first round of differential centrifugation was subjected for centrifugation at 10,700 g for around 30 min and the supernatant collected was used as cytoplasmic fraction. The protein lysates from different fractions were quantified by Bradford and subjected for Western blot and co-immunoprecipitation as described previously [24].

The mice were anesthetized and perfused transcardially with phosphate buffer solution followed by 4% paraformaldehyde at pH 7.4. The brain tissue was post fixed overnight in 4% paraformaldehyde and then cryo-preserved in 30% sucrose. The following procedure was adopted for the study as described earlier [24]. The fixed brains were sliced into 25 μm sections, washed with phosphate buffer saline thrice, and blocked with 5% goat serum for 1 h. The sections were incubated over night with respective primary antibodies—rabbit polyclonal TDP-43 (Protein tech, IL, USA, 1:1000), mouse monoclonal NeuN (Millipore; 1:1000), rabbit polyclonal Iba1 (WAKO; 1:500), rat monoclonal CD11b (serotech 1:500), and caspase-3 (Cell signaling; 1:100), mouse monoclonal ubiquitin (Millipore; 1:500) followed by incubation with respective fluorescent goat Alexa Fluor 488 and 594 (1:500) secondary antibodies (Invitrogen) for 2 h at room temperature. Finally, the microscopic images were captured using confocal microscope (Zeiss) and Apotome (Zeiss Axio vision).

The mice were anesthetized and perfused transcardially using phosphate buffer saline followed by 4% paraformaldehyde solution (pH 7.4). The brains were then sectioned to 25-μm-thick slices and stained with cresyl violet histological stain. The mean stroke area of approximately ten sections from 3 to 12 months old were calculated after 72 h post MCAO by using ImageJ software and expressed as % stroke area. (%Stroke area = (infarct size / total contra lateral side of the section) × 100) [28].

The mouse cytokine array (Ray bio Mouse cytokine Antibody Array, C1 series, Ray bio, Inc.) was used to detect the levels of different cytokines in sham, acute, chronic stroke-operated mice. The array was performed according to the manufacturer’s instructions. The protein lysates were obtained by homogenization of brains of from respective groups using 1x cell lysis buffer (provided in the kit). The protein concentration was determined for each sample and diluted to 300 μg in 1x blocking buffer. Samples for each group (three mice/group) were pooled and incubated with array membrane overnight at 4 °C. After washes, the membranes were incubated with biotin conjugated primary antibody provided in the kit overnight at 4 °C and next day following successive washes, membranes were incubated with secondary antibody provided for 2 h at room temperature. As per the ray biotech protocol, protein levels were visualized by chemiluminescence and quantified using ImageJ software. The protein levels on each array were standardized against an internal positive control on the array (Lalancette et al. 2007; Bohacek et al. 2012).

As described previously [29], paraffin-embedded human brain sections of 5 μm from the frontal cortical lobe were examined to evaluate reactivity towards TDP-43 protein. The paraffin-embedded sections were deparaffinized in xylene and rehydrated in a descending series of ethanol. Endogenous peroxidases were blocked with 5% hydrogen peroxide in methanol, and antigen retrieval was carried out using sodium citrate buffer for 30 min. Sections were then blocked by using goat serum and incubated overnight with anti-TDP-43 (abnova: E2 clone 1:1000). Next, labeling was detected by using species-specific biotinylated the EnVision™ + System, Peroxidase (Dako, Agilent). Slides were cover slipped using DPX mounting medium (#06522, Sigma). For the used tissues, we obtained written consent from the families for tissue removal after death for diagnostic and research purposes at the Neurological Tissue bank of the Biobank-Hospital Clínic-Institut d’Investigacions Biomèdiques August Pi i Sunyer (IDIBAPS). We also obtained tissue from control subjects stored at this Biobank. The study had the approval of the Ethics Committee of Hospital Clínic de Barcelona (CEIm). Features of the cases are shown in Additional file 1: Table S1.

The data quantified and represented as mean ± SEM. Statistical analysis was carried using Prism 7 (Graph Pad Software, La Jolla, CA, USA). Comparison between two groups was achieved by unpaired t test and comparisons between multiple groups were done using one-way ANOVA followed by Tukey’s post-hoc multiple comparison test. Statistical significance was defined when ***p < 0.001, **p < 0.01, *p < 0.05. In all experimental procedures, n (per group) = 5–10 animal.

The presence of pathological TDP-43-positive inclusions in cytoplasm and nucleus of both the neurons and glia in ALS, FTLD, and Alzheimer’s has been widely established [17, 24]. Moreover, growing evidence suggests that TDP-43 may have a role in the acute neurodegeneration/neuroinflammation triggered by different types of brain injuries including TBI and stroke. To better understand the role of TDP-43 in the brain response to ischemic injury, we first characterized the expression pattern of TDP-43 by immunofluorescence staining employing anti-TDP-43 antibody in the wild-type (WT) mice that were subjected to transient MCAO followed by different reperfusion periods. While within the first 24 h after stroke TDP-43 expression was mostly limited to the nuclei, at 48 h after MCAO, few cells start to show cytoplasmic TDP-43, while at 72 h post-stroke till the latest time point analyzed (30 days), most of the cells displayed strong cytoplasmic TDP-43 immunoreactivity (Fig. 1a). Because incidence of stroke increases with aging (peaking at middle age time), we next performed MCAO on 3- and 12-month-old WT mice and asked whether TDP-43 subcellular distribution and expression patterns in control conditions and after stroke are affected by processes of aging. Normally, TDP-43 is a nuclear protein (Fig. 1b, c upper panels) and as shown in Fig. 1d, e, the expression of whole-length TDP-43 from the nuclear lysates isolated from control brains and 72 h post MCAO mice did not show significant difference between 3 and 12 month mice. Because brain injury may trigger formation of the pathological TDP-43 species including TDP-35 and TDP-25 C-terminal fragments, we next asked whether aging may contribute to a formation of pathological TDP-43 species. To address this issue, we analyzed the expression patterns of TDP-43, TDP-35, and TDP-25 fragments in the cytoplasmic lysate collected 72 h post MCAO. While ischemic injury caused mislocalization of TDP-43 into cytoplasm in both age groups (Fig. 1b, c, f), Western blot analysis revealed a significant increase in the expression of whole-length cytoplasmic TDP-43 in the ischemic brains of 12 months compared to 3-month-old mice and corresponding controls (Fig. 1f, g). The same expression pattern was observed for pathological TDP-35 and TDP-25 fragments. As further revealed in Fig. 1f, ischemic injury in older, 12-month-old animals was associated with significantly higher increase in the expression of pathological cytoplasmic TDP-35 and TDP-25 fragments when compared to 3 month old (Fig. 1f, g). Importantly, the pathological TDP-35 and TDP-25 fragments were not detected in the cytoplasmic lysates from the brains of 3- and 12-month-old controls, non-stroked mice (Fig. 1f, g). Of note, a low amount of full-length cytoplasmic TDP-43 was detected in the brains of non-stroked 12-month-old mice. Taken together, our results suggest that cytoplasmic mislocalization of TDP-43 after stroke may represent a common mechanism of the neuronal response to ischemic injury while formation and expression of pathogenic, truncated TDP-35 and TDP-25 species suggests an age-related process. Importantly, as shown in Fig. 1h, unlike the chronic neurodegeneration, acute brain ischemia did not trigger formation of phosphorylated TDP-43 aggregates.

The main histological feature of neurodegenerative disorders such as ALS/FTLD and to a lesser extent AD is a presence of ubiquitinated TDP-43 immunoreactive cytoplasmic inclusions in neuronal and glial cells. In order to determine if cerebral ischemia increases the level of ubiquitination of TDP-43 in an age-dependent manner, we performed a double immunofluorescence analysis using anti-TDP-43 and anti-ubiquitin antibodies on 3- and 12-month-old control mice and 72 h after MCAO. As shown in Fig. 2a, a double immunofluorescence analysis revealed a colocalization of ubiquitin staining with the cytoplasmic TDP-43 forming a ring-shaped TDP-43/ubiquitin aggregates. The TDP-43/ubiquitin aggregates were present in the mice of both age groups (Fig. 2a); however, the number of TDP-43/ubiquitin positive inclusions seems to be higher in the ischemic brains of the 12 months mice. These results obtained by immunofluorescence were further confirmed by co-immunoprecipitation experiments of TDP-43 protein with ubiquitin (Fig. 2b). The immunoprecipitation experiments clearly demonstrated that TDP-43 is highly ubiquitinated 72 h after MCAO. The intensity of ubiquitination of TDP-43 was more pronounced in the brains of 12-month-old mice compared to the 3-month-old mice 72 h after MCAO. Importantly, the control samples of both the age groups were devoid of ubiquitinated TDP-43 protein. Next, we asked whether ischemic injury triggers cytoplasmic mislocalization of TDP-43 in glial cells. As further revealed in Fig. 2c, we observed few microglial cells showing cytoplasmic TDP-43 72 h post MCAO in both age groups suggesting that deregulation of TDP-43 is not only confine to neurons after acute stroke.

Microglial activation coupled with a marked induction of Toll-like receptor 2 (TLR2) are characteristic features of the brain response to ischemic injury [23]. Our previous reports have demonstrated that overexpression of TDP-43 in transgenic mice increases expression of inflammatory markers and TLR2 in glial cells [17, 24]; thus, we hypothesized that increase in TDP-43 levels and the formation of ubiquitin/TDP-43-positive inclusions observed predominately in 12-month-old mice would favor shift of microglial profiles toward pro-inflammatory phenotype resulting in an increase of the TLR2 response after stroke. To visualize microglial activation patterns and TLR2 response after MCAO, we took advantage of the TLR2-luc/GFP reporter mouse model previously generated and validated in our laboratory (Fig. 3a) [23]. In this in vivo model system, the reporter genes luciferase and GFP are co-expressed under transcriptional control of the murine TLR2 gene promoter thus allowing visualization of the luciferase signals from the brains of living mice [23]. Importantly, our previous studies have demonstrated that the TLR2 signal induction represents a valid readout measure of microglia activation after stroke [30‐32]. As shown in Fig. 3b, we analyzed and compared the TLR2 signals normalized to a baseline value after stroke in young and old mice. The quantitative analysis of the TLR2 signals after MCAO revealed a significant increase in the TLR2 signal in 12-month-old mice when compared to 3-month-old mice. The differences were more pronounced in the acute phase of the response, i.e., 48–72 h after stroke (Fig. 3a, b). That microglial cells were more activated in the ischemic brains of 12-month-old mice was further confirmed by the quantitative analysis of the standard microglial marker Iba1. As shown in Fig. 3c, Western blot analysis of brain homogenates collected from 3- and 12-month-old mice in control conditions and 72 h after MCAO showed a marked increase in Iba 1 expression levels after stroke when compared to control conditions. The expression of Iba1 was strongly induced by MCAO in both age groups. However, a significantly higher induction of Iba1 was observed in the ischemic brains of the 12-month-old mice when compared to younger mice (Fig. 3c). Importantly when comparing the non-stroke controls, a baseline expression level of Iba1 was also significantly higher in 12-month-old controls animals when compare to younger animals. Together, these findings suggest the existence of age-dependent processes that are associated with an increase in inflammatory signaling in the brain.

We previously demonstrated that deregulation of TDP-43 observed in a mouse model of ALS/FTDL potentiates NF-κB-mediated pathogenic pathways [22]. We showed that TDP-43 may act as a co-activator of the P65 subunit of NF-κB and helps in the transcription of pro-inflammatory genes leading to release of inflammatory cytokines and causing inflammation-induced neurodegeneration [22]. The active form of P65, the phospho P65 (P-P65), can be used as an indicator of NF- κB-associated inflammation. We next examined whether NF-κB-mediated pathway is activated after stroke and whether the observed age-related increase in cytoplasmic TDP-43 is associated with an increase in P-P65. As shown in Fig. 3d, we quantified the P-P65 level in nuclear lysate from the brains of 3- and 12-month-old controls and acutely stroked mice (72 h post MCAO). The P-P65 levels were found to be significantly up-regulated in 12-month-old mice 72 h post MCAO when compared to 3-month-old mice in the same conditions (Fig. 3d). Although we observed a tendency, there was no significant difference in P-P65 level between 3- and 12-month-old control mice (Fig. 3d). To further asses the age-dependent immune profile of the brain after stroke, as previously described [33], we performed a multiplex cytokine array analysis. We measured expression levels of several pro- and anti-inflammatory cytokines in the brain. As shown in Fig. 3e, analysis of the inflammatory response in the brains of stroked mice revealed an increase in general immune response in the 12-month-old mice 72 h after MCAO. In fact, cytokine array data showed a significant increase in the levels of both the pro- and anti-inflammatory cytokines and chemokines like IL-1β, TNF-α, CCL5, GM-CSF, IL-6, IL-4, IL-10, and IL-17 in this specific group (Fig. 3e).

We next investigated to what extent the observed age-related deregulation of cytokine response after stroke affects evolution of the post-stroke ischemic injury. As previously described [33] and shown in Fig. 4a, we first measured and compared the size of the ischemic lesion in 3- and 12-month-old mice. The cresyl-violet-stained sections were analyzed 72 h after MCAO. In agreement with previous results, there was a significant, 17.4% increase in the size of ischemic lesion in the 12-month-old mice when compared to 3-month-old mice (Fig. 4a, b). As further shown in Fig. 4c, d, the observed increase in brain damage was accompanied by an increase in cellular death/apoptosis. Namely, a Western blot analysis of brain tissue homogenates from the ipsilateral/ischemic region of the brain showed a significant, twofold increase in the expression levels of the cleaved caspase-3 in the ischemic brains of 12-month-old mice, 72 h after MCAO (Fig. 4d). Next, a double immunofluorescence analysis revealed that the vast majority of the cleaved caspase-3 expressing cells were positive for neuronal marker NeuN (Fig. 4e). In accordance with our previous work [31, 34], this suggests that neurons are the principal cell type undergoing cellular stress and apoptosis after MCAO.

Previous evidence suggests that cytoplasmic mislocalization of TDP-43 and formation of ubiquitinated aggregates is toxic to neurons and may enhance neuroinflammation [22, 24, 35]. Here, we hypothesized that age-related increase in the cytoplasmic TDP-43 observed in the brains of 12-month-old mice in control conditions and after stroke alters the immune microenvironment and may increase the susceptibility of neurons to ischemic damage. To test our hypothesis, we took advantage or the transgenic TDP-43 mice with moderate and ubiquitous overexpression of human TDP-43 (Tg line A315T). Importantly, at young age, starting at 2–3 months, these mice show an increase in cytoplasmic/ubiquitinated TDP-43 in neurons and glial cells, corresponding (resembling) to TDP-43 expression levels and/or expression patterns in aged, 12-month-old WT mice. In keeping with our hypothesis and as shown in Fig. 5a, b, the size of ischemic lesion was significantly increased in TDP-43 transgenic mice when compared to WT aged-matched littermates. Here, it is noteworthy that the observed increase in the size of ischemic lesion was corresponding to a size of ischemic lesion observed in 12-month-old mice (Figs. 3a and 4a). As shown by Western blot analysis, the cleaved caspase-3 expression levels were significantly increased in the ischemic brains of TDP-43 transgenic mice (Fig. 5c, d). A double immunofluorescence labeling for neuronal markers NeuN and cleaved caspase-3 revealed a marked co-localization of these two markers indicating an increase in the number of apoptotic neurons in the ischemic brains of TDP-43 transgenic mice (Fig. 5e). Next, we asked whether observed exacerbation of the ischemic injury and injury-induced neuronal apoptosis is caused by an increase in NF-κB-mediated inflammation. As described above (see Fig. 5f, g), we measured the levels of nuclear P65, an indicator of NF-κB-associated inflammation, in nuclear lysate from the brains of 3- and 12-month-old controls and TDP-43 transgenic mice in control conditions and following stroke (72 h post MCAO). As further shown in Fig. 5f, g, the expression of nuclear P65 is significantly increased in 12-month-old WT mice when compared to young WT controls. Interestingly, as shown in Fig. 5f, g, the expression of nuclear P65 is at comparable levels between old controls and 3-month-old TDP-43 transgenic mice thus suggesting a correlation of cytoplasmic TDP-43 levels and NF-κB activation.

At present, it remains unclear to what extent and/or whether TDP-43 pathology is associated with ischemic injury in human stroke. To address this issue, we performed immunohistochemistry analyses of the post-mortem post-stroke brain tissue autopsied at 1 to 5 days after stroke. The analysis was performed by focusing on two distinct region of the ischemic lesion, the peri-infarct and core region (cortical sections) and compared with corresponding controls. The analysis was performed using a human anti-TDP-43 antibody (Fig. 6a, b). As shown in Fig. 6a, control sections stained with an anti-human TDP-43 antibody displayed well-circumscribed and positively stained nuclei, while the cytoplasmic compartment was almost completely devoid of TDP-43 immunoreactivity. Interestingly, resembling the TDP-43 expression patterns following single traumatic brain injury in humans [20], the TDP-43 staining following acute ischemic stroke revealed increased immunoreactivity within the cytoplasm and in some cases extending into processes (Fig. 6b). The increase in the cytoplasmic TDP-43 immunoreactivity was more prominent at day 5 after stroke. Hence, ischemic stroke in humans is associated with an increase in TDP-43 immunoreactivity in the cytoplasmic compartment. However, unlike the TDP-43 expression patterns observed in the chronic TDP-43 proteinopathies, in stroke, the nuclei remain positively stained for TDP-43.