Introductory Paragraph
Intronic G4C2 hexanucleotide repeat expansions of C9orf72 are the most common cause of familial variants of frontotemporal dementia / amyotrophic lateral sclerosis (FTD/ALS)1,2. G4C2hexanucleotide repeat expansions (HREs) in C9orf72 undergo non-canonical repeat associated translation, producing dipeptide repeat (DPR) proteins, with various deleterious impacts on cellular homeostasis3. While five different DPRs are produced, poly(glycine-arginine) (GR) is amongst the most toxic, and is the only DPR to accumulate in the associated clinically relevant anatomical locations of the brain4,5. Previous work has demonstrated the profound effects of a poly(GR) model of c9FTD/ALS, including motor impairment, memory deficits, neurodegeneration, and neuroinflammation6. Neuroinflammation is hypothesized to be a driving factor in the disease course; microglia activation is present prior to symptom onset and persists throughout the disease7. Here, we establish the contributions of the NLRP3 inflammasome in the pathogenesis of FTD/ALS, resulting from a stress-induced neuronal-microglial crosstalk feedforward loop. In a mouse model of c9FTD/ALS, inflammasome-mediated inflammation was increased with microglial activation, cleavage of caspase-1, upregulation of Cxcl10, and production of IL-1β. We find that genetic ablation of Nlrp3 protected behavioral deficits and prevented neurodegeneration as seen in C57BL6J Wild Type mouse model of c9FTD/ALS. Ultimately, survival was improved by the genetic ablation of Nlrp3. Moreover, we identified the process by which neuronal stress signals induced by hexanucleotide expansions initiate an inflammatory cascade in microglia. These findings provide evidence of the integral role of inflammasome-mediated innate immunity in c9FTD/ALS pathogenesis, and suggest the NLRP3 inflammasome as a therapeutic target.
Main Text
To directly investigate the effects of GR induced neuronal stress in the brain in c9FTD/ALS, neonatal intracerebroventricular injections of adeno-associated virus serotype 9 (AAV9) vectors containing either GFP (control) or 100 repeats of the GR dipeptide (GFP-(GR)100), was conducted, as previously reported6. AAV9 have a strong expression pattern in neurons of the brain, but not microglia8, allowing for the targeted investigation of neuronal induced effects. Body weight of GFP and GFP-(GR)100 mice did not differ throughout the study (Supplementary Figure 1a).
Examination of c9FTD/ALS behavioral features in GFP-(GR)100 revealed significant mortality over the course of the paradigm, with over 25% of animals dying before completion (Fig 1a). To assess behavioral responses, mice were tested for contextual and cued memory, motor function, and anxiety. GFP-(GR)100 animals exhibited a significant reduction in freezing in both the contextual and cued memory paradigms (Figs 1b-c). GFP-(GR)100 animals exhibited a significant increase in falls in the hanging wire test (Fig 1d). GFP-(GR)100 mice exhibited anxiety with an increase in thigmotaxis in the Open Field Test (Fig 1e). The observed impairments in motor function, memory, anxiety, as well as mortality largely replicate those seen in c9FTD/ALS.
Next, as we hypothesized the inflammasome contributes to the neurodegeneration in behaviorally relevant brain regions in the central nervous system in c9FTD/ALS, we quantified GFP-expressing neurons, indicating virally infected cells, in brain regions with major contributions to the observed behavioral phenotype. These include the prefrontal cortex (PFC), the parietotemporal lobes of the cortex (CTX), and the hippocampus (HP). This GFP-(GR)100 model of c9FTD/ALS has previously been shown to not effect spinal cord pathology, allowing for specific evaluation of associated brain pathologies6. GFP-(GR)100 mice exhibit significant reduction in GFP-expressing neurons in the CTX, PFC, and HP (Fig 1g-i). Representative images are depicted in Figure 1f.
Microglia mediated neuroinflammation has been described in cases of ALS9. Microgliosis has also been previously described to occur in this model, beginning at 1.5 months of age6. We hypothesized that microglia, the innate immune cells of the brain, may be causally related to the pathogenesis of c9FTD/ALS. To determine microglial response to the disease process, we first stereologically quantified the density of microglia in the cortex, a relevant region of the brain that contributes to the behavioral impairments we observed in GFP-(GR)100 and found a significant increase in GFP-(GR)100, compared to GFP controls (Fig 1k). We next performed quantitative morphological analyses to characterize activation states microglia. In CTX of GFP-(GR)100 mice, there was a trending increase in the level of ramifications closer to the soma, indicated by the increase in intersecting segments within 10 µm of the soma (Fig 1i) suggesting microglia hyperramification, compared to GFP controls. This is accompanied by a statistically significant decrease in the number of intersecting segments further away from the soma (20 – 30 µm away from the soma) (Fig 1i), indicative of an activated state described by microglia branch contraction accompanied with increased branch complexity, as depicted by a representative microglia in situ in (Figure 1j.II). This increased density and morphological ramification of microglia in the CTX is indicative of a neuroinflammatory response occurring in GFP-(GR)100 mice.
Next, we investigated the effects of microglial activation seen in GFP-(GR)100 animals. We hypothesized that NLRP3 inflammasome responses may be related to the neuronal stress induced by GFP-(GR)100. NLRP3 inflammasome complexes contain pro-caspase-1, which is cleaved to active-caspase-1, which in turns processes the zymogenic forms of IL-1β and IL-18 into active forms; as such, we measured levels of pro-caspase-1 and active-caspase-1 via Western Blot in the cortex, and IL-1β and IL-18 via ELISA. GFP-(GR)100 mice exhibit an increase in active caspase 1 (Fig 1m), pro-caspase-1 (Fig 1n) and the ratio of active caspase-1 to pro-caspase-1 (Fig 1o), in the CTX. Additionally, IL-1β was significantly increased in the CTX of GFP-(GR)100, compared to GFP controls, while there was no change in IL-18 (Fig 1q-r). Collectively, the elevation of active caspase-1, pro-caspase-1, IL-1β, microgliosis and activation of microglia suggests that there is a significant innate immune driven response in vivo in GFP-(GR)100 mice as a result of inflammasome assembly and activation in microglia.
Based upon our results identifying significant innate immune related inflammation in a GFP-(GR)100 model of c9FTD/ALS, we targeted the NLRP3 inflammasome complex via genetic ablation. To test the effects of genetic ablation of Nlrp3, GFP-(GR)100 or GFP expression in the mouse brain was achieved by neonatal intracerebroventricular injections of adeno-associated virus serotype 9 (AAV9) vectors in mice lacking the Nlrp3 gene (Nlrp3-/-). Weights of Nlrp3-/--GFP and Nlrp3-/--GFP-(GR)100 did not differ over the course of the study (Supplementary Figure 1b).
Examination of the behavioral effects of inflammasome targeted interventions revealed no change in mortality in Nlrp3-/- GFP-(GR)100 compared to Nlrp3-/- GFP and WT-GFP over the course of the experimental paradigm (Fig 2a). Assessment of memory function demonstrated unaffected contextual memory in Nlrp3-/--GFP-(GR)100 animals compared to Nlrp3-/--GFP; however, significant cued memory deficits are still evident (Fig 2b-c). Motor function, which was significantly impaired in WT-GFP-(GR)100, remained unaffected in Nlrp3-/- GFP-(GR)100, with animals performing at a statistically similar level to Nlrp3-/--GFP and WT-GFP animals. (Fig 2d). Anxiety, as assessed by the OFT, revealed no changes in Nlrp3-/--GFP-(GR)100 animals compared to Nlrp3-/--GFP (Fig 2e). Interestingly, both Nlrp3-/--GFP and Nlrp3-/- GFP-(GR)100 exhibited less freezing time in both the contextual memory task, while exhibiting a greater preference for the center zone than WT-GFP.
We next investigated the effects of targeting the NLRP3 inflammasome on neurodegeneration. GFP-(GR)100 mice. In the CTX, no change in the count of GFP-expressing neurons was noted in Nlrp3-/--GFP-(GR)100 compared to either WT-GFP or Nlrp3-/--GFP (Fig 2g). Lack of neurodegeneration in the CTX and protection in the associated behavioral tasks indicate a role of Nlrp3 in c9FTD/ALS pathogenesis.
As we above demonstrated the presence of inflammasome mediated neuroinflammation in the cortex in GFP-(GR)100 mice, we further investigated these findings in the cortex of Nlrp3-/--GFP-(GR)100 mice. To assess levels of proinflammatory cytokines produced as a result of activation of inflammasome activation in response to GFP-(GR)100, we quantified levels of IL-1β and IL-18 in the CTX of WT-GFP, Nlrp3-/--GFP, and Nlrp3-/--GFP-(GR)100 mice. Nlrp3-/--GFP-(GR)100 mice exhibited a significant increase in IL-1β, compared to WT-GFP controls (Fig 2h). Interestingly, IL-18 production was significantly lower in both Nlrp3-/--GFP and Nlrp3-/--GFP-(GR)100 compared to WT-GFP mice (Fig 2i). We next assessed the levels of the pro-IL-1β and pro-IL-18 processing caspase-1. No change in active caspase-1 (Fig 2j), pro-caspase-1 (Fig 2k), and the ratio of active caspase-1 to pro-caspase-1 (Fig 2l) was seen in Nlrp3-/--GFP-(GR)100 animals compared to Nlrp3-/--GFP or WT-GFP.
Genetic ablation of the NLRP3 inflammasome prevented the significant mortality, motor impairment, contextual memory deficits, and anxiety exhibited in our WT GFP-(GR)100 mice. Furthermore, significant protection to GFP-expressing neurons in CTX as well as the ratio of active-caspase-1 to pro-caspase-1 is conferred by Nlrp3 ablation. Thus, these data further support the idea of microglia driven innate immune responses in the brain to the pathogenesis of c9FTD/ALS, and identify a novel treatment strategy of targeting microglia mediated innate immune responses.
The above data demonstrate the link of innate immune driven inflammation to disease pathogenesis in GFP-(GR)100 animals. We next investigated the transcriptional regulation of genes in the cortex, prefrontal cortex, and hippocampus of GFP-(GR)100 mice. Principal component analysis indicated samples tend to cluster amongst treatment groups, with both cortex and pre-frontal cortex tissues clustering within treatment groups, showing similar transcriptional profiles within GFP and GFP-(GR)100 groups (Fig 3a).
Using DESeq2, a total of 165 genes are differentially regulated in the cortex of GFP-(GR)100 animals (padj < 0.1). Overall transcriptional patterns depicting the significantly differentially expressed genes are visualized in Figure 3b, with the top three mapped upregulated and downregulated genes labelled (as identified by the magnitude of Log2Fold Change) (Figure 3b). The top upregulated mapped gene in GFP-(GR)100 is Cxcl10, which has been previously shown to result in the proliferation and activation of microglia10.
Based on our above-described in vivo data characterizing a causal link between inflammasome activation and neurodegeneration resulting in behavioral impairments, we hypothesized that GFP-(GR)100 expression would invoke transcriptional changes in innate immune responses. Examination of the pathways and regulators enriched in the differentially-expressed genes using Ingenuity Pathway Analysis confirms enrichment in neuroinflammatory signaling pathways including “Complement System”, “TREM1 Signaling”, “Chemokine Signaling”, and “NF-κB Signaling”, in the cortex (Fig 3c). These pathways are core regulators of inflammatory responses and microglial activation. Amongst validated genes (Supplemental Figure 2), we found that Cxcl10 is significantly upregulated, with a 7-fold increase in GFP-(GR)100 animals, compared to GFP controls (Fig 3d).
The transcriptomic changes exhibited in GFP-(GR)100 animals heavily point to significant innate immune, microglia mediated responses. Our findings of significant Cxcl10 upregulation suggests there is a neuronal signal released which can activate microglia, and induce innate immune cascades in GFP-(GR)100 mice.
As Cxcl10 is an important signal for the immune responses of microglia10, we investigated the production and effects of Cxcl10 in primary neuronal and microglial cultures, in vitro. We studied, in vitro, the neuronal effect of GFP-(GR)100 infection on CXCL10 production, and whether CXCL10 induces microglia activation. First, we investigated whether the expression of GFP-(GR)100 induced neuronal death in vitro. GFP-(GR)100 neurons exhibited a significant increase in the release of LDH compared to GFP neurons (Fig 4a), confirming the GFP-(GR)100-induced neuronal injury and death observed in vivo. As Cxcl10 was shown to be highly upregulated in vivo in GFP-(GR)100 mice, we hypothesized that neurons are able to produce and secrete CXCL10, in response to GFP-(GR)100. To test this, primary cortical neurons were transfected with either AAV9-GFP or AAV9-GR100 and cell supernatants were analyzed for CXCL10 concentration. GFP-(GR)100 neurons demonstrated significantly increased production of CXCL10 compared to GFP neurons (Fig 4b).
As we identified Cxcl10 is highly upregulated in vivo, and was determined experimentally to be significantly released by primary cortical GFP-(GR)100 neurons in vitro (Fig 4b), we next investigated whether CXCL10 activates microglia. Tumor necrosis factor-alpha (TNF-α) constitutes a hallmark for microglia activation and has previously been shown to play an important role in the transcriptional regulation of components of the NLRP3 inflammasome11. To test the ability of CXCL10 to induce TNF-α expression, primary cortical microglia cultures were treated with escalating doses of CXCL10 and the TNF-α released to cell supernatant was measured. CXCL10 significantly increased the microglial production of TNF-α in a dose-dependent manner (Fig 4c), while the co-treatment with a CXCL10 neutralizing antibody results in a significant reduction in the TNF-α secreted by microglia (Fig 4d). Finally, to analyze if CXCL10 mediates the activation of IL-1β processing inflammasome complexes, we measured the release of IL-1β from primary cortical microglia in response to escalating doses of CXCL10. There is a significant, dose-dependent escalation in the secretion of IL-1β (Fig 4e). Co-treatment with a CXCL10 neutralizing antibody results in a significant reduction in IL-1β released by microglia (Fig 4f). These data demonstrate a neuronal-microglial crosstalk whereby GFP-(GR)100 induces secretion of neuronal CXCL10, activating a microglial innate immune response, resulting in inflammasome activation and subsequent neuroinflammation.
In conjunction with the rest of our data, these experiments suggest that neurons in c9FTD/ALS produce DAMPs such as CXCL10 as a result of GFP-(GR)100, which significantly stimulates pro-inflammatory reactions from microglia, leading to inflammasome activation (Fig 4g). Targeting the resultant NLRP3 inflammasome dependent responses serves as a potent substrate for therapeutic intervention in c9FTD/ALS.
Our present study provides evidence linking inflammasome activation to neurodegeneration, and provides a basis for the investigation of innate immune inflammasome inhibitors as a treatment of disease in c9FTD/ALS. Here, we identify a key mechanism by which neuronal stress initiated by G4C2 HREs in c9FTD/ALS can activate microglia, producing an inflammasome-dependent innate immune response, which can be therapeutically targeted. As neuronal stress occurs throughout various forms of FTD/ALS12–14, the self-perpetuating cycle of neuronal stress inducing microglial inflammasome activation and resultant neuroinflammation may represent a conserved therapeutic substrate throughout the various disease subtypes of FTD/ALS. Ultimately, we demonstrate that by targeting microglial reactivity through inflammasome inhibition, FTD/ALS pathogenesis was significantly attenuated through the genetic ablation of Nlrp3. The results of these studies identify the contributions of inflammasome activation in microglia to the pathogenesis of c9FTD/ALS.
Microglial activation and the activation of the NLRP3 inflammasome have been demonstrated as crucial mediators in other neurodegenerative conditions including Alzheimer’s disease15, Parkinson’s disease16, and primary progressive multiple sclerosis 17. Inflammation has also been noted pre-symptomatically in models of ALS18,19 and incidence of ALS has been demonstrated to be higher in individuals who have been diagnosed with an autoimmune disease20, indicating a putative role in the disease pathogenesis, and a target for therapeutic intervention. Neuroimaging studies conducted on individuals with ALS have demonstrated microglial activation throughout the brain9,21. Upon examination of the microglia in the cortex, we observed microgliosis consistent with previously identified findings6. Morphologically, microglia in GFP-(GR)100 were in an activated state; these findings are congruent with those identified in other neurodegenerative diseases, including Alzheimer’s Disease, in which morphologically activated microglia are present22.
Inflammasome activation has been noted in other forms of ALS23–25. We noted an increase in active caspase-1, pro-caspase-1, as well as the ratio of active-caspase-1 to pro-caspase-1 in the cortex of GFP-(GR)100 mice; these increases were attenuated via genetic ablation of Nlrp3. Interestingly, while IL-1β production was not attenuated by the genetic ablation of Nlrp3, Nlrp3-/- mice had a significantly lower level of IL-18 in the cortex compared to WT animals. This Nlrp3 independent production of IL-1β has been demonstrated previously, whereby Nlrp3 deficient mice still produce IL-1β in response to a stimulus; however, IL-18 levels were significantly decreased26. Studies employing caspase-1 deficient mice and caspase-1 targeting treatments should be employed in the future to further evaluate the therapeutic value of targeting inflammasome-produced cytokines in FTD/ALS.
In addition to improvements in neurodegeneration and neuroinflammation, full genetic ablation of the Nlrp3-/- inflammasome conferred significant protection behaviorally in Nlrp3-/--GFP-(GR)100 mice compared to Nlrp3-/--GFP controls; mortality was almost completely eliminated, contextual memory was preserved, motor function was protected, and there was no increase in anxiety. To determine if the NLRP3 inflammasome can be therapeutically targeted, further studies may employ experimental models whereby NLRP3 ablation by Cre dependent-designer receptor exclusively activated by designer drugs (DREADD) mice allow for temporal control of NLRP3 expression, thereby evaluating both its role in presymptomatic disease, and evaluating whether the NLRP3 inflammasome may serve as a therapeutic target after symptom onset.
Cxcl10, which we found to be highly upregulated in the cortex of GFP-(GR)100 mice, encodes for a small cytokine belonging to the CXC chemokines family. Upon neuronal death, CXCL10 is produced and exerts a chemotactic function by attracting microglia and CD8+ T cells, and can induce the activation of microglia, thereby causing the release of pro-inflammatory cytokines27. Chemotactic effects are demonstrated in our FTD/ALS model by the microgliosis in the cortex of GFP-(GR)100 mice where we see a significant increase in the density of microglia. Neuronal stress has been demonstrated in other forms of ALS, including in cases of sporadic ALS28. Dysregulation of CXCL10 chemotaxis in peripheral blood cells from ALS patients has been directly observed, which was associated to increased inflammatory responses29. These facts further support our in vitro findings, where we demonstrated that neuronal cultures expressing (GR)100 dipeptides released higher concentrations of CXCL10, and the treatment of microglia cultures with increasing concentrations of CXCL10 promoted a dose-dependent increase in TNF-α and in IL-1β production, indicating microglial activation.
In summary, our studies identify a novel neuron-microglia crosstalk mechanism in c9FTD/ALS whereby neuronal stress induced secretion of CXCL10 triggers inflammasome activation in microglia, thereby creating a self-propagating cycle of neurodegeneration and neuroinflammation. Our findings show targeting the inflammasome responses represents a putative therapeutic strategy, as evidenced by the genetic ablation of Nlrp3 and resultant protection from behavioral impairment and neuropathologies.
Author Contributions
KJT and CS conducted in vivo experiments of the GFP-(GR)100 model. Immunofluorescence staining, image acquisition, and analysis was performed by KJT, CS, UHI, UR, and TO. In vitro experiments were conducted by MSV, MAR, and RIA. Molecular analyses, including ELISA, Western Blots, and RT-qPCR were performed by KJT, CS, MSV, HW, MAR, and EZ. Bioninformatic analyses were conducted by HW and ME. Generation and purification of the GFP-(GR)100 and GFP viruses were completed by YJZ and LP. KJT, CS, FJH, and GMP designed the project. KJT, CS, and GMP prepared the manuscript. All authors discussed and commented on the article.
Declaration of Interests
The authors declare that they have no conflicts of interest with the contents of this article.
Supplemental Figures
Materials and Methods
Materials
The AAV9-GFP and AAV9-GFP-(GR)100 viruses were a kind gift from Dr. Leonard Petrucelli. Unless specified otherwise, cell culture reagents were obtained from ThermoFisher Scientific (Waltham, MA) and antibodies were obtained from Abcam (Cambridge, MA).
Subjects
Wild-type C57BL/6J mice (#000664) and Nlrp3-/- mice (#021302) were obtained from The Jackson Laboratory (Bar Harbor, ME) and socially housed on a 12:12-h light/dark cycle with lights on at 07:00 h in a temperature-controlled (20 ± 2 °C) vivarium. Mice were given food and water ad libitum and were bred to obtain mouse pups used in our studies. All procedures were approved by the Institutional Animal Care and Use Committee of the Icahn School of Medicine at Mount Sinai.
C9orf72 dipeptide model of frontotemporal dementia / amyotrophic lateral sclerosis
Upon confirmation of pregnancy, wild-type and Nlrp3-/- dames were separated and monitored daily for litters. After identification of a milk spot to confirm that newborn (p0) pups were being nursed, the pups were injected with AAV9-GFP or AAV9-GFP-(GR)100 by the ICV method. Pups were cryoanesthetized by placing in a padded 15mL tube partially submerged in a slurry of water and crushed ice. Upon confirmation of anesthesia by toe-pinch, we identified the point of injection, approximately two-fifths of the distance between bregma and each eye. A Hamilton syringe (#7634-01; Reno, NV) with a 32 ga needle (10 mm long, Hamilton #7803-04) was loaded with 4 µL of the AAV vector and placed 2 mm deep, orthogonal to the mouse’s head at the point of injection. 2 µL of the AAV vector was injected at a rate of 1 µL min-1 into each lateral ventricle. Upon completion of injections, each pup was rapidly returned to physiological temperatures by placing on a warming pad. The litters were returned to dames upon completion of injection for the entire litter and monitored for acceptance. Mice were weaned at p30 by sex. For confirmation of diffusion prior to studies, a small number of mice were injected with Trypan Blue and sacrificed two hours later for dissection of the brain and examination by stereoscopic microscope.
Behavioral testing
After reaching 3 months of age, mice were tested through a battery of behavioral tests. All behavioral testing took place during the light phase of the day (9:00AM). On all days of behavioral testing, mice were acclimated to an anteroom directly adjacent to the behavioral testing room for 30 min. On Day One, mice were tested in the Open Field Test for basal anxiety and general locomotor impairments (Seibenhener et al., 2015). The Open Field apparatus consisted of a 40cm x 40cm x 40cm Plexiglass box with opaque white walls, situated within a dimly-lit room (200 lux). Mice were placed in the center of the apparatus and were allowed to freely explore for 10 minutes before being returned to their home cage. On Day Three through Five, mice were tested in a Hanging Wire Test for muscular impairments (Aartsma-Rus and van Putten, 2014). The Hanging Wire apparatus consisted of a 2mm-thick wire suspended 35cm over a layer of corn cob bedding, situated in a brightly lit room (500 lux). Mice were lifted from their home cage by the base of their tail and were placed near the wire until they grasped it with their forelimbs. The number of falls over a 2 min period were recorded. Falls in which the mice hanged from the wire from their hindlimbs were excluded from the number of falls. At the end of the behavioral trial, mice were returned to their home cages. On Days Seven through Nine, mice were tested in a Contextual and Cued Memory Test in two Contexts. Context A was a 30cm x 24cm x 21cm conditioning chamber (Med Associates, Fairfax VT) within a room with white walls and bright lighting. Context A had a bare metal grid floor, bare grey walls, bright lighting, and background fan noise. Context A was cleaned with a 0.5% hydrogen peroxide solution (Virox, Oakville Canada) between each trial. Context B was a chamber of the same dimensions within a room with dim lighting. The chamber had a white plastic floor, curved white plastic walls, dim lighting, no background fan noise, and scented with 0.25% benzaldehyde in 70% ethanol. Context B was cleaned with a 70% ethanol solution between each trial. On Day Seven, mice were allowed to explore Context A for 4 min. At 180s, white-noise (85 dB) played for 30 s and was co-terminated with a footshock (2 s, 0.75 mA). After the 4 min trial, the mouse was returned to its home cage. On Day Eight, mice were allowed to explore Context A for 4 min in the absence of white-noise and footshock (Context Recall). On Day Nine, mice were allowed to explore Context B in the constant presence of white-noise (85 dB, Cue Recall). For all tasks, behavior was analyzed and recorded with Any-Maze v6.0 (Stoelting, Wood Dale IL).
Immunofluorescence
After completion of behavioral studies, a subset of mice were deeply anesthetized with ketamine/xylazine (100 mg/kg + 10 mg/kg, IP), then were perfused transcardially with cold sterile PBS followed by 4% PFA in PBS. Brains were removed, drop-fixed in 4% PFA overnight, then washed once with cold PBS and stored in PBS. Tissue sections (50 μm thick) were taken with a vibratome (Leica; Wetzlar, Germany) and were stored in PBS with 0.02% sodium azide (w/v). Sections were washed in PBS followed by 10 min permeabilization in 0.1% Triton X-100 in PBS (PBST). Sections were then incubated in blocking solution (5% goat serum in PBST) for 1.5 hr. The sections were washed three times with PBST and incubated with primary antibodies diluted in blocking solution: 1:500 Rabbit anti-Iba1 (ab178846, Abcam) and 1:250 chicken anti-GFP (A10262, ThermoFisher) overnight at 4°C. Post incubation, the sections were washed three times in PBST and incubated with secondary antibodies diluted in blocking solution for two hours at room temperature: Goat anti-Rabbit conj AlexaFluor 568 (Abcam #175471, 1:500) and 1:500 Goat anti-Chicken AlexaFluor 488 (A11039, Thermofisher). The sections were washed three times in PBST and incubated in 1μM DAPI solution (ab228549, Abcam) for 5 min. The sections were washed twice in PBS and mounted on slides using ProLong Diamond Antifade Mountant (P36970, ThermoFisher).
Stereology and cell morphology
Sections were stained for immunofluorescence as described above. Microglia density for the cortex, prefrontal cortex, and hippocampus of the brain was determined through the use of MBF Stereo Investigator (Williston VT). The number of sections required for each region were determined by an initial pilot study which found that six sections were needed for both cortex and prefrontal cortex regions and seven sections for the hippocampal region, each spaced equidistantly. A widefield microscope (AxioImager M2/Z2, Carl Zeiss, Oberkochen Germany) along with MBF Stereo Investigator were used to visualize each section and to determine the number of microglia in each region. Using the Optical Fractionator workflow, the region of interest was traced onto each section based on the mouse brain atlas by Paxinos and Franklin (Paxinos and Franklin). This was followed by a systematically random grid overlay consisting of squares measuring 300 µm by 300 µm, covering the region of interest. From each of these squares an unbiased sampling region (100 µm by 100 µm) was used to manually count microglia somata in the area. Once each section was completed, the software ran an algorithm to estimate the number of microglia in the region. In parallel, the Cavalieri estimator in the software was used to determine the volume of each region. The microglia count and volume were then used to determine the microglia density in the CTX, PFC, and hippocampal regions. For assessment of cell morphology, images of coronal sections were taken on a Zeiss LSM880 Airyscan confocal microscope (Oberkochen, Germany) using an X20/0.8 NA air immersion objective controlled by Zeiss Zen Black software. For 3D analysis, z-stack images were obtained by capturing an image every 0.7μm covering the entire 50μm-thick section. Images were deconvoluted using AutoQuant X3.1 (Media Cybernetics, Rockville MD) and 3D analysis was performed using Imaris 9.1.2 (Bit Plane Inc, Concord MA) using the surface tool to reconstruct the soma and the filaments tool to reconstruct the branches.
Microglial cultures
Cortices from 1-3 day-old C57BL/6J mouse pups were isolated, digested, and seeded at a density of 8 cortices per 10 mL culture dish. Every three days, medium (DMEM + 10% FBS + 1% penicillin-streptomycin) was replenished. After 3 weeks, mixed glial cultures reached confluence and were isolated by mild trypsinization as previously described (Saura et al., 2003). Briefly, cells were washed with culture medium without FBS and treated with a mixture of trypsin (0.25% without EDTA) and DMEM-F12 medium in a 1:3 ratio. After 40 min incubation, mixed glial cells detached and left a layer of microglia attached to the bottom of the culture dish. Pure microglia were isolated by 15 min incubation with trypsin (0.05% with EDTA) at 37°C followed by gentle shaking. Cells were counted and seeded in 24-well plates at a density of 7.5 × 104 cells/wells.
Cortical neuron cultures
Cortices from 3-day old C57BL/6J mouse pups were isolated and finely diced in ice-cold HBSS. Cortices were then incubated in 10X Trypsin Solution (Sigma #59427C) with DNase I (Sigma #D4513) for 15 min at 37°C with period inversion. Cells were then spun at 200g for 5 min, and the pellet was triturated with a serological pipette and strained in a 40 μm cell strainer (BD Falcon #352340). Centrifugation followed by trituration was repeated once, then the pellet was spun down and resuspended in Neurobasal medium supplemented with 0.25% GlutaMAX, 2% B-27, 10% Fetal Bovine Serum, and 1% Penicillin/Streptomycin at a density of 5 × 105 cell/mL. 2 hours later, medium was changed to Neurobasal medium supplemented with 0.25% GlutaMax, 2% B-27, and 1% Penicillin/Streptomycin, and half the medium was replenished every 3-4 days afterwards. 3 days later, cells were infected with 2 × 1010 vg/mL AAV. Expression was confirmed by fluorescence microscopy. Cell supernatant and lysates were collected for assessment of LDH release and CXCL10 production (R&D Systems #DY466).
Analysis of Cytokines and Caspase-1 activity
After completion of behavioral trials, a subset of mice were sacrificed and brain regions were immediately frozen on dry ice and plasma was isolated from trunk blood using Lithium Heparin-treated tubes (Becton Dickinson #365985, Franklin Lakes NJ). Brain regions were lysed with 1X Cell Lysis Buffer supplemented with 1 mM PMSF (Cell Signaling #8553S, Danvers MA) and Protease Inhibitor Cocktail (Sigma-Aldrich, #11873580001). IL-1β in brain tissue, plasma, and microglia culture supernatant was measured with Mouse IL-1 beta/IL-1F2 DuoSet ELISA Kit (R&D Systems DY401, Minneapolis MN) according to the manufacturer’s instructions. CXCL10 was measured in neuronal culture supernatant with Mouse CXCL10 DuoSet ELISA (R&D Systems DY466-05, Minneapolis MN) according to manufacturer’s instructions. Primary microglia were stimulated with recombinant Mouse CXCL10 protein (R&D Systems 466-CR-050/CF, Minneapolis MN) and incubated with Mouse CXCL10 antibody (R&D Systems AF-466-NA, Minneapolis MN). TNF-α and IL-1β were measured in supernatant by ELISA with Mouse TNF-alpha Quantikine ELISA Kit (R&D Systems MTA00B, Minneapolis MN) and Mouse IL-1 beta/IL-1F2 DuoSet ELISA Kit (R&D Systems DY401, Minneapolis MN) according to manufacturer’s instructions. For analysis of caspase-1 activity, 35 µg of protein was loaded into a Western blot using a PVDF membrane (Bio-Rad #1620177; Hercules, CA) with Mouse IL-1 beta/IL-1F2 DuoSet ELISA Kit (R&D Systems DY401, Minneapolis MN) according to the manufacturer’s instructions. Membranes were blocked with 5% Bovine Serum Albumin (BSA) in 0.1% TBST and were probed with the following antibodies. Primary antibodies: Mouse anti-Caspase-1 1:1000 (AdipoGen #AG-20B-0042; San Diego, CA), Mouse anti-Tubulin 1:1000 (Sigma #T9026). Secondary antibodies: Goat anti Mouse conj HRP (1:10000, ThermoFisher #G-21040).
Gene expression and RNAseq
After completion of behavioral trials, a subset of mice were sacrificed and brain regions were immediately frozen on dry ice. Total RNA was isolated using the RNeasy Minikit (QIAGEN #74106; Hilden, Germany) and precipitated by the ethanol/sodium acetate method. RNA concentration and quality was initially measured using a Nanodrop 2000 (ThemoFisher). Secondary analysis of concentration and quality was conducted by the Genomics CoRE Facility at the Icahn School of Medicine at Mount Sinai using Qubit RNA BR Assay Kit (ThermoFisher #Q10211). Library construction and RNA sequencing was performed by Novogene (Durham, NC).
Tissue processing
Four mice from each group (GR100, HBA, and GFP) were processed for sequencing, providing three distinct brain regions (Cortex (CTX), Frontal Cortex (FC), and Hippocampus (HP)) per mouse. After assessing sample quality, one Frontal Cortex sample was excluded due to poor quality, and all tissue samples from a single mouse were removed due to sample misclassification.
Gene expression
Reads were processed using the NGS-Data-Charmer pipeline. Briefly, adaptors and low-quality bases were trimmed from reads, which were then aligned to the mm10 genome using HISAT2 (version 2.2.1). Read counts in the mm10 GENCODE annotation (version M22) were generated using FeatureCounts DESeq2 (version 1.24.0, R version 3.6.1) was then used to calculate differential expression from the read counts. The R package ‘biomaRt’ (version 2.40.5) was used to translate ensembl ids into common gene symbols.
Gene ontology
Mouse genes involved in inflammatory processes were extracted from the Mouse Genome Informatics group database (MGI). MGI mouse genes with the GO term ‘Inflammatory response’ were extracted. Ingenuity Pathway Analysis was applied to all differential genes identified in the previous analysis.
Quantitative reverse transcription PCR
For qPCR analysis, gene expression was measured in 4 replicates by PowerUP SYBR Green Master Mix (ThermoFisher # A25778) using an ABI PRISM 7900HT Sequence Detection System. Hypoxanthine phosphoribosyltransferase (Hprt) expression level was used as an internal control and data was normalized using the 2−ΔΔCt method55. Levels of target gene mRNA was expressed relative to those of GFP + Veh mice for in vivo studies. Primers used in this study were designed using Primer-BLAST software (Ye et al., 2012) and are listed in Supplementary Table S1.
Statistical Analysis
All figure values are presented as mean and standard error of the mean (s.e.m.). Statistical tests are indicated in the figure legends. A confidence interval of 95% was used for all analyses. In all studies, outliers (> 2 SD from the mean) were excluded. All statistical analysis was performed using GraphPad Prism 9 software (GraphPad Software, San Diego CA). *p < 0.05, **p < 0.01, ***p<0.001, ****p<0.0001, ns not significant. Statistically insignificant trends are indicated by p-value.
Acknowledgements
This research was supported by a grant (AT008661) from the NIG’s Office of Dietary Supplements (ODS) and the National Center for Complementary and Integrative Health, awarded to G.M.P. The study was supported by the generous support of the Altschul Foundation to G.M.P. G.M.P. holds a Senior VA Career Scientist Award.
We acknowledge that the contents of this study do not represent the views of the NCCIH, the ODS, the National Institutes of Health, the U.S. Department of Veterans Affairs, or the United States Government.
Footnotes
↵‡ Co-first authors