Nanoligomers targeting NF-κB and NLRP3 reduce neuroinflammation and improve cognitive function with aging and tauopathy

NF-κB and NLRP3 are central mediators of age-related inflammation (i.e., “inflammaging”), neuroinflammation, innate immune responses and cytokine release, including in the brain [6, 7, 21, 22]. Our published data show that NF-κB/NLRP3-targeting Nanoligomers downregulate these targets in multiple brain regions [18, 20]. However, to confirm this in the present study, we first treated older (aged 19 months) male and female C57Bl/6J mice with the same Nanoligomer cocktail or saline control (1 × intraperitoneal injection; 150 mg/kg body weight). Younger (aged 8 months) male and female mice, as a reference group, also received a saline injection. After 16 h, mice were euthanized and immunoblots were performed on isolated whole hippocampus. We observed significant reductions in both NF-κB and NLRP3 in old treated vs. untreated mice (Fig. 1B, D), and a modest but non-significant increase in phosphorylated (activated) NF-κB in older vs. young control animals that was absent in older treated animals (Fig. 1C). These effects of acute Nanoligomer administration on central mediators of neuroinflammation and innate immune responses are consistent with our reports showing that the drug crosses the blood–brain barrier and engages its targets [13, 14, 16, 18, 20], suggesting the potential for protective effects on mediators of neuroinflammation downstream of NF-κB/NLRP3 (e.g., reactive glia, pro-inflammatory gene induction and cytokine release) [23, 24]. Consistent with this idea, even with the short duration of this treatment, we also found a nearly significant trend for reduced expression of Cluster of Differentiation 68 (CD68), a marker of pro-inflammatory microglia in Nanoligomer-treated old mice (Fig. 1E).

Although NF-κB and NLRP3 play complex roles in brain health [8], most data suggest that inhibiting these proteins individually improves cognitive function [25, 26]. Therefore, given our finding that acute Nanoligomer treatment reduced NF-κB and NLRP3 in the hippocampus (a key area of the brain involved in memory and cognitive function) and the central role of neuroinflammation in cognitive dysfunction with brain aging and AD [27], we next tested the hypothesis that long-term Nanoligomer treatment (150 mg/kg body weight; 3 × week for one month) would improve cognitive function in old C57Bl/6J male and female mice. In addition, because tauopathy (a key hallmark of AD) is associated with NF-κB/NLRP3 activation [28] and cognitive dysfunction [29], we also treated 2 month-old male and female rTg4510 mice, a common model of tauopathy based on expression of human P301L tau in the forebrain, using the same approach. Importantly, we did not note any adverse effects with this treatment, and all body weights remained stable. After the treatment, we measured short-term recognition memory via Novel Object Recognition testing and anxiety-like behavior via Elevated Plus Maze testing. In Novel Object Recognition testing, we found that treated older and tauopathy mice performed significantly better (as measured by recognition indices) than untreated controls, and similarly to younger mice and littermate controls, respectively (Fig. 2A). In the Elevated Plus Maze test, we observed a non-significant trend for treated older and tauopathy mice to spend more time in open maze arms than untreated mice, reflecting reduced anxiety (Fig. 2B). Finally, in tests of grip strength, which we measured because of its relationship to cognitive function and overall health [30], we found a slight but non-significant trend for improvements in old treated mice, although these effects were not noted in treated tauopathy mice, and among all animals we observed no differences in frailty, which is closely associated with cognitive decline [31] (Supplementary Figure S1). Future, more comprehensive studies are needed to assess how NF-κB/NLRP3-targeting Nanoligomer treatment affects other cognitive and peripheral functional outcomes (e.g., spatial memory, treadmill endurance, wire hang, rotarod performance). However, these results show that long-term Nanoligomer treatment improves several domains of cognitive-behavioral function in old, non-transgenic mice, and in tauopathy model mice, without overt adverse effects on neuromuscular function/health.

The activation of NF-κB and NLRP3 via aging or tauopathy induces the production of many pro-inflammatory cytokines [32]. Therefore, we performed multiplex ELISAs on brains from the same mice (long-term Nanoligomer-treated) described above, to assess levels of key cytokines involved in neuroinflammation and innate immune responses. Principal component analyses (PCA) revealed that aging and tauopathy were associated with robust and broad increases in many cytokines, whereas cytokine levels in treated old and tauopathy animals were similar to those in young mice and littermate controls, respectively (Fig. 3A, B, left panels). At the individual cytokine level, several important pro-inflammatory and immune-activating cytokines that increased with aging and tauopathy included Interleukin 5 (IL-5) and Macrophage-Inflammatory Protein 2 (MIP-2α), but these effects were reduced with Nanoligomer treatment. Similar trends were noted for most Interleukins (ILs), Monocyte Chemoattractant Proteins (MCPs), and Interferons (IFNs) (Supplementary Figure S2). Additionally, we found that cytokine levels (global/PC-based and individual) were significantly inversely correlated with short-term recognition memory (Fig. 3A, B center and right panels), which is consistent with existing evidence that increased cytokine levels in the brain are associated with cognitive decline [33]. Together, these data show that long-term NF-κB/NLRP3-targeting Nanoligomer treatment broadly reduces neuroinflammation, including many cytokines that are directly modulated by NF-κB and NLRP3 [22], and that these changes may contribute to improvements in cognitive function in old wildtype and tauopathy model mice.

Importantly, tau pathology can both activate and be exacerbated by NF-κB and NLRP3 [34–36], driving cycles of inflammation-mediated tau propagation and toxicity [37], and phosphorylated (activated) tau is closely associated with cognitive dysfunction [38]. To characterize the effects of Nanoligomer treatment on tau, we performed immunohistochemistry for two tau residues (phosphorylated at threonine-181 and threonine-231) that have been associated with AD pathology and progression [39, 40]. We found that these tau species increased in hippocampus and cortex of rTg4510 mice, similar to what others have reported in tauopathy models [41, 42], and that this effect was attenuated with Nanoligomer treatment (Fig. 4A). We also performed immunoblots for tau, including for phosphorylated species more typical of later pathology, and we found that long-term Nanoligomer treatment was associated with modest reductions (p = 0.05) in both total and phosphorylated (at serine 396) tau using this approach (Fig. 4B).

Finally, in these same mice, additional immunohistochemistry analyses showed that Nanoligomer treatment and the reduced tau burden we observed were associated with: (1) neuroprotection, as reflected by greater numbers of Fox-3 (NeuN)-positive cells in the CA1 and CA2 regions of the hippocampus (Fig. 5A); and (2) fewer potentially reactive glia, as reflected by reductions in Glial Fibrillary Acidic Protein (GFAP)-positive astrocytes and a non-significant trend for reductions in Ionized Calcium-Binding Adapter molecule 1 (IBA1)-positive microglia (Fig. 5B). Although GFAP and IBA1 counts do not reflect neuroinflammation per se, these results suggest that NF-κB/NLRP3-targeting Nanoligomer treatment may suppress tau pathology and confer neuroprotection, and that this effect may involve changes in glial cell-associated processes.

Although aged wildtype mice do not develop tau pathology, they do develop gliosis like humans. Indeed, in human aging/AD, glial cells (astrocytes and microglia) become immune-activated, secreting cytokines that impair neuronal and cognitive function [43]. This neuroinflammation is also associated with morphological changes [44, 45], including alterations in process length [46] and branching [47, 48], which have detrimental effects on neurons (e.g., synaptic dysfunction). To evaluate the effects of NF-κB/NLRP3-targeting Nanoligomer treatment on astrocyte and microglia activation/morphology with aging, we performed immunofluorescence staining on cortex tissue from old, old treated, and young mice. Consistent with previous findings showing that astrocyte activation increases with aging [44, 49], we found evidence of age-related increases in GFAP [50], but this effect was less pronounced in old treated mice (Fig. 6A). Similar to previous studies [51, 52], we also found that aging was associated with reduced astrocyte process length and branch number (Fig. 6B). However, these effects were reversed with Nanoligomer treatment. We also found age-related increases in the activation of microglia, as reflected by increases in IBA1 (Fig. 6C). This effect of aging on microglia was not altered by Nanoligomer treatment, although we did observe changes in microglial process length and branch number in old animals that were attenuated with treatment (Fig. 6D). Finally, similar to our observations in rTg4510 mice, using immunohistochemistry we found that all of these glial cell changes were associated with a reduction in NeuN + cells in old mice that was absent in treated animals, although only in the CA2 region of the hippocampus (Fig. 6E). Collectively, these results suggest that the beneficial effects of Nanoligomers targeting NF-κB/NLRP3 in old mice may be due in part to reducing age-related changes in glial cells and the preservation of neurons in the hippocampus.

Perturbations in the transcriptome affect downstream biological processes that mediate brain aging/AD [53]. To provide broad insight into how Nanoligomer treatment affected genes/pathways most related to aging and tauopathy, we performed RNA-seq on the hippocampus. We found that aging and tauopathy were associated with gene expression changes (Fig. 7A, top rows of heatmaps) that were largely reversed in terms of relative expression with Nanoligomer treatment (Fig. 7A, bottom rows of heatmaps). For example, one of the top transcripts that increased with aging but decreased with Nanoligomer treatment was S100 calcium-binding protein A8 (S100A8), which activates NF-κB and induces microglial activation [54]. One of the top transcripts that increased with tauopathy but decreased with Nanoligomer treatment was Interleukin 17 Receptor E (IL17RE), which is also associated with the activation of NF-κB [55] and cognitive dysfunction [56].

Given the above observations and because aging and tau pathophysiology intersect [57], to understand how Nanoligomer treatment affected genes/pathways common to both aging and tauopathy, we compared the top 1000 most increased/decreased transcripts common to both aging and tauopathy, and those same transcripts in old and tauopathy-treated mice (Fig. 7A, top Venn diagrams). Gene ontology (GO) analyses revealed that increased genes common to both aging and tauopathy reflected pathways associated with inflammation, immune responses and cognitive dysfunction, including immune receptor activity, major histocompatibility complex (MHC) binding, and synapse pruning (Fig. 7B, left). Decreased genes/transcripts common to both aging and tauopathy comprised pathways associated with neurotransmitter receptor activity and signal transduction (Fig. 7B, right). However, these patterns were generally reversed with treatment. That is, among the transcripts that changed in both treated old and treated tauopathy mice (Fig. 7A, bottom Venn diagrams), those that were increased reflected pathways involved in brain aging and neuronal/cognitive function including ion channel activity, neuronal signaling, and neurogenesis (Fig. 7C, left), whereas common decreased transcripts with treatment in aged and tauopathy mice reflected pathways associated with inflammation and immune responses, including cell adhesion and stress response (Fig. 7C, right).

To determine which cell types might have contributed most to the observed transcriptome patterns, we performed a deconvolution analysis (intersecting bulk RNA-seq data with single-cell RNA-seq data to identify cell types most associated with gene expression differences in the bulk data [58]). This revealed a common pattern in both mouse/treatment models (Fig. 7D). With aging and tauopathy, increased gene expression signatures (i.e., of inflammation) were largely associated with glia, whereas decreased gene expression signatures were primarily associated with neurons—and in all cases these cell type-specific contributions to gene expression were at least somewhat reversed with Nanoligomer treatment. To confirm our RNA-seq data, we performed immunoblotting for several proteins whose gene expression levels changed (Figure S4), and we found similar patterns of protein expression for these targets. Also, given our findings above suggesting changes in glial cells and neuroprotection, we immunoblotted for CD68 (a marker of activated/reactive microglia [59]) and Brain-Derived Neurotrophic Factor (BDNF), a protein responsible for maintaining neuronal survival and growth [60] (Fig. 7E). We observed reduced CD68 in Nanoligomer-treated animals, although this effect was not significant in old mice, but no treatment-associated changes in BDNF in either model. Although not definitive, these results suggest that Nanoligomer targeting of NF-κB/NLRP3 favorably influences several important biological processes associated with brain aging and tauopathy, particularly those involved in neuroinflammation/immune activation and neuronal health, and that these effects may be mostly driven by changes in glial cells, upstream of neuroprotection (i.e., because the inflammation-related gene expression patterns that changed with treatment mapped to these cell types in deconvolution analyses, and neuron-intrinsic protective proteins like BDNF were unchanged).

Finally, current therapies meant to target neuroinflammation in the brain often have off-target effects [12, 61], and drug-induced hepatotoxicity is common following exposure to many bioactive compounds [62]. The liver is a key metabolic organ that processes such compounds and therefore a relevant target for therapies meant to help the brain [63]. To understand how Nanoligomer treatment affects the liver, we performed multiplex ELISAs on liver tissue from young, old, and old treated mice (Fig. 8A). We found that several pro-inflammatory and immune-activating cytokines that increased with aging were reduced with Nanoligomer treatment, including IFN-Ɣ inducible protein (IP-10) and chemokine ligand 1 (CXCL1). Additionally, the anti-inflammatory cytokine interleukin 10 (IL-10) was reduced with aging, but this effect was reversed with Nanoligomer treatment. We also found that aging was associated with evidence of hepatocyte polyploidy and polymorphonuclear leukocytes in the liver, a key source of tissue damage due to inflammation [64], and these effects were attenuated with Nanoligomer treatment (Fig. 8B, C). Furthermore, in Nanoligomer-treated old mice, these favorable changes in the liver were associated with a preservation of circulating (plasma) levels of fibroblast growth factor 21 (FGF-21), a hormone synthesized by the liver and linked with longevity and neuroprotection [65, 66]. Collectively, these results suggest that, in addition to their protective effects in the brain, Nanoligomers targeting NF-κB/NLRP3 are not hepatotoxic and may even favorably modulate several markers of inflammation and liver/peripheral health in old mice.

Nanoligomers were synthesized by Sachi Bioworks as previously reported [13–18]. Briefly, Nanoligomers comprise an antisense peptide nucleic acid (PNA) conjugated to a gold nanoparticle for improved delivery and membrane transport. The PNA sequences, provided in [14, 15], were previously screened for solubility, self-complementarity, and off-targeting in the human genome. The PNA portions of the Nanoligomers (NLRP3 Sequence: CTTCTACTGCTCACAGG, NFKB Sequence: AGTGGTACCGTCTGCTA) were synthesized on a Vantage peptide synthesizer (AAPPTec, LLC) with solid-phase Fmoc chemistry. Fmoc-PNA monomers were obtained from PolyOrg Inc., with A, C, and G monomers protected with Bhoc groups. Following synthesis, the peptides were conjugated with gold nanoparticles and purified via size-exclusion filtration. Conjugation and concentration of the purified solution was monitored through measurement of absorbance at 260 nm (for detection of PNA) and 400 nm (for quantification of nanoparticles). The sequence of the synthesized PNA was confirmed using LC–MS [13].

Older and younger male and female C57Bl/6J mice (aged 19 months and 8 months, respectively) were purchased from the National Institute on Aging aged rodent colony. Male and female rTg4510 mice and appropriate littermate controls (2 months of age) were purchased from the Jackson Laboratory (strain #024854). For the first part of the study (proof of concept to show that the Nanoligomer cocktail reduced both NF-κB and NLRP3 in the brain), cohorts were treated with Nanoligomer acutely (1 × intraperitoneal injection; 150 mg/kg body weight) and then studied 16 h later. For long-term treatment, mice were treated for 1 month (150 mg/kg body weight; 3 × week). Mice were housed throughout all experiments at ~ 18–23 °C on a reverse 12 light/12 dark cycle. Fresh water and ad-libitum food (Tekland 2918; 18% protein) was routinely provided to all cages. Animals were consistently health checked by the veterinary staff at Colorado State University.

Mice were culled in a fed state in the late morning. After deep anesthetizing with isoflurane, ~ 1 ml of blood was removed via cardiac puncture followed by cervical dislocation. The left hippocampus and a piece of left cortex tissue were removed, flash-frozen on dry ice, and stored at − 80 °C until further processing. The right hippocampus and right cortex were processed (paraffin-embedded) for immunostaining (details below).

Frozen hippocampus samples were lysed in RIPA buffer containing 100 mM KCl, 40 mM Tris HCl, 10 mM Tris base, 5 mM MgCl₂, 1 mM EDTA, and 1 mM ATP (pH 7.5), phoSTOP phosphatase inhibitors and cOmplete protease inhibitors (ThermoFisher). Sample protein concentration was determined using a BCA kit (ThermoFisher/Pierce), and 20 μg of protein per sample was separated by electrophoresis (Bio-Rad Criterion system), transferred to nitrocellulose membranes (Trans-Blot Turbo, Bio-Rad), and blocked in TBS-T with 5% Bovine Serum Albumin (BSA). Primary antibodies were purchased from ABclonal and included phosphorylated nuclear factor kappa B (p-NF-κB, 1:1000 dilution); nuclear factor kappa B (NF-κB, 1:1000); NLR Family Pyrin Domain Containing 3 (NLRP3, 1:1000); total tau (1:1000); p-tau S396 (1:1000); p-tau T205 (1:1000); Cluster of Differentiation 68 (CD68, 1:1000); Indoleamine 2,3‐dioxygenase 1 (IDO1, 1:1000); Interleukin 17 Receptor E (IL17RE, 1:1000); and Brain-Derived Neurotrophic Factor (BDNF, 1:1000). GAPDH (1:2000) was purchased from Novus Biologicals. Membranes were incubated in primary antibody (TBS-T in 5% BSA) for 24 h at 4°C, followed by HRP-conjugated secondary antibodies (TST-T in 5% milk; 1:5000; Cell Signaling) for 1 h. Membranes were then incubated in ECL, imaged on a ProteinSimple Fluor Chem imaging system, and quantified with ImageJ.

Novel Object Recognition: For NOR testing, mice were assessed for short-term memory performance using no habituation phase [91]. Briefly, mice were placed in an opaque arena (50 cm × 50 cm) with two identical objects (tower of Legos) placed 8 cm from each side wall. After an inter-trial interval of two hours, mice were returned to the arena, but one of the old objects was replaced with a new object of similar shape and color (cell culture flask filled with multi-colored sand). Trials ended once 20 s total exploration time was reached, or 10 min had elapsed. If mice were unable to meet the exploration criteria of 20 s or 10 min, they were excluded from the analysis. Recognition index was calculated as novel object exploration time divided by 20 s [91].

Elevated Plus Maze: Elevated Plus Maze testing was performed as previously described [90, 92] and consisted of a cross with four arms (67 cm in length; 2 exposed arms, 2 enclosed arms). Mice were placed in the center of the maze facing an open arm and allowed to freely explore for 5 min. The amount of time in open arms (a measure of anxiety-like behavior) was recorded and analyzed with AnyMaze software.

Grip Strength: The grip strength test was performed as previously described [93]. Mice were tested using the DST-110 Digital Force Gauge (Maze Engineers). A total of 5 trials were performed per mouse and the highest and lowest values were discarded for analysis. The remaining 3 force values were averaged for each mouse. Data are presented as g force/g body weight.

Frailty Testing: Mouse frailty scores were assessed using an established scale [94]. Briefly, trained investigators blindly scored mice on a 31-point scale, which included measurements of health in the following areas: integumentary, physical/musculoskeletal, vestibular/auditory, ocular/nasal, digestive/urogenital, and respiratory systems. Higher scores indicate worse outcomes, while lower scores indicate better outcomes.

Brain homogenates from all mice were assessed with a 36-plex Procartaplex cytokine/chemokine panel (ThermoFisher) as previously reported [15, 16], which included the following cytokines and chemokines: ENA-78 (CXCL5), Eotaxin (CCL11), GRO-α (CXCL1), IP-10 (CXCL10), MCP-1 (CCL2), MIP-1α (CCL3), MIP-1β (CCL4), MIP-2α (CXCL2), RANTES (CCL5), G-CSF (CSF-3), GM-CSF, IFN-α, IFN-γ, IL-1α, IL-1β, IL-2, IL-3, IL-4, IL-5, IL-6, IL-9, IL-10, IL-12p70, IL-13, IL-15/IL-15R, IL-17A (CTLA-8), IL-18, IL-22, IL-23, IL-27, IL-28, IL-31, LIF, MCP-3 (CCL7), M-CSF, and TNF-α. For systemic effects (Fig. 7), separate individual ELISAs were used. Briefly, 25uL of 10 mg/ml hippocampus, cortex, liver homogenate, or serum was processed using standard ELISA techniques and analyzed on a Luminex MAGPIX xMAP instrument. Standards for each cytokine/chemokine were used with 1:4 dilutions (eightfold dilutions) and background, controls and sample concentrations were determined from a standard curve using Five Parameter Logistic curve fit/quantification.

Brain and liver tissues were fixed in 10% neutral buffered formalin at room temperature for at least 48 h. Brain samples were processed on a Leica TP1020 Automatic Benchtop Tissue Processor and embedded in paraffin wax (Cancer Diagnostics). For immunohistochemistry analyses of phosphorylated tau, paraffin-embedded tissue was sectioned at 4 µm, deparaffinized, rehydrated, sodium citate treated, and blocked in Tris A/2% donkey serum (Jackson ImmunoResearch), then incubated overnight in the following antibodies: phospho-tau (T181, 1:400 dilution; Invitrogen) and phospho-tau (T231, 1:400; ABClonal). Wash steps were performed using 2% bovine serum albumin and 2% Triton-X in 1 M TBS, and an ABC HRP peroxidase detection kit (Vector Laboratories) and ImmPACT DAB Substrate Peroxidase (HRP) Kit (Vector Laboratories) were used for a chromogen. Slides were counterstained with hematoxylin (ThermoFisher), secured with a coverslip in mounting medium and stored at room temperature until imaging. Whole tissue images were taken on an Olympus BX53 microscope with an Olympus DP70 camera using an Olympus UPlanSApo 20 × objective (N.A. = 0.75). Representative images were taken using an Olympus BX53 microscope with an Olympus DP70 camera using an Olympus UPlanFL N 40 × objective (N.A. = 0.75). To quantify and select the phosphorylation of tau-positive stained cells the manual threshold on the Count and Measure function of Olympus CellSens software (v1.18) was used. For immunofluorescence analyses of GFAP, IBA1 and glial cell process length and branching, sections were incubated overnight in primary antibodies: IBA1 (1:50 dilution; Abcam) and GFAP (1:100; Dako). Slides were washed with TBS and incubated for 1 h in the dark with AlexaFluor 555 or 647 (Invitrogen) secondary antibodies at 1:500 dilution and 2% normal donkey serum, then washed and incubated in Hoechst dye (1:2000; ThermoFisher) for 3 min. Slides were mounted with Prolong Gold Antifade mounting media (ThermoFisher) and stored in the dark at 4 °C prior to imaging. Images were captured using an Olympus BX63 fluorescence microscope equipped with a motorized stage and Hamamatsu ORCA-flash 4.0 LT CCD camera using a 40 × Olympus X-Apochromat air objective (N.A. = 0.80). For astrocyte/microglial process length and number analyses, four regions between the dentate gyrus and CA1-CA3 region of the hippocampus were imaged at 40×. To identify/quantify processes and branches, skeletonization of GFAP+ astrocytes and IBA1+ microglia was performed using IMARIS software as previously described [95] to trace and quantify stained filaments. For histology analyses of liver, tissues were sectioned on a ThermoFisher HM 325–2 microtome at 5 μm thickness and mounted on positively charged glass slides (Superfrost Plus, Cancer Diagnostics). Liver sections were deparaffinized and stained with hematoxylin and eosin (Cancer Diagnostics), and whole tissue images were taken for analysis using an Olympus DP70 camera and an Olympus UPlanSApo 20 × objective. Livers were scored for pathological infiltration of immune cells specifically around the portal triad, and for overall hepatocyte morphology by a trained animal pathologist.

Transcriptome (RNA-seq) and gene expression analyses were performed on whole hippocampus via methods previously described [89, 90]. Frozen samples (n = 6/group; 3M and 3F) were transferred into Trizol (ThermoFisher) and homogenized on ice (SP Bel-Art ProCulture Micro-Tube Homogenizer System; VWR Scientific). RNA was isolated using an RNA-specific spin column kit (Direct-Zol, Zymo Research) and treated with DNase to remove genomic DNA. Poly(A)-selected libraries were generated using magnetic oligo dT beads (ThermoFisher) and Illumina TruSeq kits. Libraries were sequenced on an Illumina NovaSeq 6000 platform to yield > 40 M 151-bp paired-end FASTQ reads/sample. Reads were trimmed and filtered with the fastp program [96], then aligned to the mm10 Mus musculus genome using the STAR aligner [97]. Differential gene expression was analyzed with DESeq2 [98] and genes/transcripts with BaseMean read counts of < 10 were filtered and removed for analyses [99, 100]. Differentially expressed genes were analyzed for gene ontology (GO) enrichment using the Metascape program [101], and heatmaps were constructed with GraphPad Prism software. Deconvolution analyses of bulk RNA-seq data were performed as previously described [88] using a published protocol and R script [58]. Briefly, single-cell RNA-seq data on mouse hippocampus (SMART-seq, 2019; 10x-SMART-seq taxonomy, 2020) were downloaded from the Allen Brain Map portal [102], and cells from the hippocampus, including astrocytes, endothelial cells, microglia, oligodendrocytes and neurons, were sub-setted for analysis. Normalized DESeq2 gene counts for old vs. young mice, old Nanoligomer-treated vs. old, rTg4510 vs. LM Controls, and rTg4510 Nanoligomer-treated vs. rTg4510 comparisons in bulk RNA-seq data were filtered for the top 250 genes by p-value, and then sorted by Log2Fold change. Sorted gene lists were scaled to the single-cell RNA-seq dataset, normalized expression of the same genes in the single-cell dataset was obtained and used for dimensional reduction via principal component analysis and hierarchical clustering/heat mapping to obtain cell-type specific contributions to differences in the bulk RNA-seq comparisons, and cell types associated with each cluster were determined using CellMarker 2.0 [103].

One-way ANOVA with Tukey’s post-hoc testing was used to assess differences among groups in protein/cytokine levels and measures of cognitive function, as well as glial cell morphology, and simple linear regression (Pearson correlations) was used to relate cytokine levels to cognitive function data. A non-parametric ANOVA equivalent (Kruskal–Wallis test) was used for small sample sizes with non-normally distributed data. All data were analyzed and presented using GraphPad Prism software. Differentially expressed genes were detected using DESeq2, and genes were sorted by p-value and Log2Fold change to construct volcano plots and heatmaps, respectively. The Metascape program was used for GO and TRRUST analyses [101].

This protocol was approved by IACUC #1441 and Laboratory Animal Resources (RRID: SCR_022157) staff at Colorado State University.

Not applicable.

S.S, V.G, A.C., and P.N. work at Sachi Bio, a for-profit company, where the Nanoligomer technology was developed. A.C. and P.N. are the founders of Sachi Bio, and P.N. has filed a patent on this technology. All other authors declare no competing interest.

The RNA-seq data generated in this study are available on the Gene Expression Omnibus website under accession number GSE251888↗. All additional data are included in the data supplement, and raw data will be made available upon request to the corresponding author (T.J.L).