Remaining brain immune cells after short PLX5622 treatment in Alzheimer's mice show lower inflammation and cell growth signals, but higher self-cleaning activity

The study employed three‐month‐old female 5xFAD mice with five human familial AD mutations driven by the mouse Thy1 promoter and age‐matched wild‐type mice (B6SJL) (Oakley et al., 2006). The AD mice were bred in‐house and maintained in the vivarium of Texas A&M University. The 5xFAD transgenic male mice and B6SJLF1/J female mice employed for breeding were procured from Jackson Laboratories (Cat No: 34840‐JAX and 100012‐JAX, Bar Harbor, ME, USA). Animals were housed in an environmentally controlled room with a 12:12‐h light–dark cycle and were given food and water ad libitum. The Animal Care and Use Committee of Texas A&M University approved all experimental procedures performed in the study.

Three‐month‐old 5xFAD female mice were randomly assigned to two groups: one group received AIN‐76A standard rodent chow containing PLX5622 (n = 12, AD+PLX group), and another group received AIN‐76A chow without PLX (n = 12, AD group) for ten days. We selected 3‐month‐old 5xFAD mice for this study because the development of AD pathogenesis begins at 1.5 months of age in these mice, which is typified by increased levels of Aβ, including Aβ deposits inside neurons. By 2 months of age, there is neuroinflammation due to activated microglia and reactive astrocytes, along with the appearance of extracellular β‐amyloid deposits, which continue to increase over time (Pádua et al., 2024). Our goal was to examine whether reducing the number of activated microglia at an early stage would decrease pro‐inflammatory markers in the remaining microglia. Furthermore, we decided to investigate female mice for an initial study, as a previous study has suggested that the progression of AD pathogenesis is more severe in female 5xFAD mice than in male 5xFAD mice (Poon et al., 2023). Hence, we first wanted to investigate the impact of partial microglia depletion on a severely affected gender.

The drug PLX5622 was purchased from Medkoo Biosciences (Morrisville, NC, USA), which was formulated into AIN‐76A standard chow at the dose of 1200 ppm by Research Diets Inc. (New Brunswick, NJ, USA). Age‐matched wild‐type mice that received standard lab chow in parallel served as naïve controls (n = 12, Naïve group). At the end of the 10‐day diet, as described above, 50% of mice from each group (n = 6/group) were deeply anesthetized, perfused with 4% paraformaldehyde, and the brain tissues were postfixed in 4% paraformaldehyde and processed for cryostat sectioning. Fresh brain tissues were harvested following deep anesthesia and decapitation from the remaining 50% of mice in each group (n = 6/group). The tissues were snap‐frozen and stored at −80°C until processed for biochemical assays.

The brain tissues were processed and sectioned using a cryostat as described elsewhere (Hattiangady & Shetty, 2011; Kodali et al., 2018; Rao et al., 2008). Thirty‐micrometer‐thick coronal sections through the entire brain were collected in 24‐well plates containing phosphate buffer (PB) and stored in a cryobuffer at −20°C until processed for immunohistochemistry. Serial sections (every 20th through the hippocampus and the cerebral cortex) were processed for immunohistochemical detection of the ionized calcium‐binding adaptor molecule 1 (IBA‐1, a marker of microglia), glial fibrillary acidic protein (GFAP, a marker of astrocytes), and Aβ42 (a marker of Aβ plaques). Furthermore, every 15th section through the hippocampus was employed to visualize doublecortin (DCX) positive newly born neurons. The methods employed for immunohistochemistry are described in our previous reports (Kodali et al., 2018; Madhu et al., 2019, 2021). The peroxidase reaction was developed using Vector SG (Vector Labs) as the chromogen. After a thorough wash, the sections were mounted on subbed slides, counterstained with nuclear fast red (Vector Labs), coverslipped, and observed under a Nikon E600 microscope. Table S1 provides a complete list of primary and secondary antibodies with dilutions employed and the source from which they were procured.

Hippocampal and cerebral cortex tissues from naïve control, AD, and AD‐PLX groups were micro‐dissected. The tissues were lysed through sonication in a tissue extraction reagent (Invitrogen) containing protease‐phosphatase inhibitor (1:100 dilution, ThermoFischer Scientific, Waltham, MA) for 15–20 s at 4°C. The resulting solution was centrifuged at 4°C for 10 min (15,000 g), and the supernatant was aliquoted and stored at −80°C until further use. The lysates were used to measure the concentration of multiple markers.

Dual or triple immunofluorescence procedures were employed to visualize the following markers, cells, and complexes using methods described in our earlier reports (Madhu et al., 2021; Kodali, Attaluri, et al., 2021; Kodali, Mishra, et al., 2021). (1) Proinflammatory microglia positive for IBA‐1 and CD68. (2) NLRP3 inflammasome complex positive for IBA‐1, NLRP3, and apoptosis‐associated speck‐like protein containing a CARD (ASC). (3) the protein sequestosome 1 (p62+ structures) in the IBA‐1+ microglia and NeuN+ neurons as an autophagy reporter. (4) phosphorylated S6 ribosomal protein (pS6) expression in the NeuN+ neurons and IBA‐1+ microglia to determine the activation of mTOR signaling. Briefly, the sections were washed in phosphate‐buffered saline (PBS), blocked with 10% normal donkey serum, and incubated overnight at 4°C in individual primary antibodies (for single immunofluorescence studies) or a cocktail of two or three primary antibodies (for dual and triple immunofluorescence studies). Next, following thorough washing in PBS, the sections were treated with the matching secondary antibodies tagged with fluorescent markers for 60 min. Then, the sections were rinsed in PBS and coverslipped with a slow fade/antifade mounting medium (Invitrogen). Table S1 provides a complete list of primary and secondary antibodies with dilutions employed and the source from which they were procured.

The IBA‐1+ microglia surrounding Aβ‐42+ plaques in the hippocampus and cerebral cortex were visualized and quantified using 2‐μm‐thick Z‐sections from brain tissue sections processed for IBA‐1 and Aβ‐42 dual immunofluorescence. In both AD and AD‐PLX groups, the number of microglia/unit area of the Aβ‐42 plaque was computed using three sections per animal (n = 4/group). The number of microglia was expressed per 100 μm² of plaque area.

The IBA‐1+ microglia expressing Clec7a (dectin‐1, a marker of neurodegenerative microglia [MGnD]) in and around Aβ‐42+ plaques were visualized using 1‐μm thick Z‐sections from brain tissue sections processed for IBA‐1, Clec7a, and Aβ‐42 triple immunofluorescence. In both AD and AD‐PLX groups, the extent of Clec7a expression within IBA‐1+ microglia in and around Aβ‐42 plaques was evaluated (n = 6/group).

The number of IBA‐1+ microglia in the hippocampus was quantified by stereological counting of IBA1+ cells using every 20th section through the entire hippocampus of naïve, AD, and AD+PLX groups (n = 6/group). Microglia in the cerebral cortex were also quantified stereologically using every 20th section, but counts were taken as numbers per unit volume (0.1 mm³). In addition, we quantified the number of clusters of microglia per unit volume (0.1 mm³) of the hippocampus and cerebral cortex in AD and AD+PLX groups (n = 6/group) using every 20th section. Stereological counting using StereoInvestigator was performed as described in our previous reports (Hattiangady & Shetty, 2011; Mishra et al., 2015). Percentages of microglia expressing CD68 were quantified from sections processed for IBA‐1 and CD68 dual immunofluorescence using Z‐section analysis in a confocal microscope as described in our previous study (Kodali, Attaluri, et al., 2021; Kodali, Mishra, et al., 2021). The percentages of IBA‐1+ microglia expressing CD68 were computed and compared between naïve, AD, and AD+PLX groups for both hippocampus and cerebral cortex (n = 6/group). The percentages of activated microglia were collected from all subfields of the hippocampus (DG, CA1, and CA3) to compile data for the entire hippocampus (3 sections/animal, n = 6/group).

The area fractions (AFs) of GFAP+ structures in the hippocampus and cerebral cortex were measured through Image J, using representative sections from all groups (Kodali et al., 2015, 2018). Three sections separated by 600 μm distance were employed in each animal (n = 6/group).

Neurolucida (Microbrightfield Inc., Williston, VT) was employed to trace the soma and processes of microglia from the dentate gyrus (DG) of the hippocampus and cerebral cortex, as described in our previous reports (Kodali et al., 2015; Kodali, Attaluri, et al., 2021; Kodali, Mishra, et al., 2021). Twenty microglia per DG or cortex (i.e., five microglia/animal, n = 6 animals/group) were individually traced in each group using a 100X oil immersion lens. The data, such as total process length, number of nodes, and endings, were computed and compared between naïve, AD, and AD+PLX groups. Sholl's concentric circle analysis was also employed in the Neurolucida program's NeuroExplorer component to determine the pattern and extent of processes in microglia from each subregion at various distances from the soma.

Triple immunofluorescence for NLRP3, ASC, and IBA‐1 detected the NLRP3 inflammasome complexes in hippocampal and cerebral cortical microglia of naïve, AD, and AD+PLX groups. Using Z‐section analysis in a Leica THUNDER 3D Imager, area fractions (AFs) of NLRP3 inflammasome complexes (i.e., structures positive for NLRP3 and ASC) within individual microglia were measured.

The lysates from the hippocampus and cerebral cortex were utilized for quantifying the mediators and end products of NLRP3 inflammasome activation and additional proinflammatory cytokines using individual enzyme‐linked immunosorbent assay (ELISA) kits (n = 6 animals/group: Madhu et al., 2019, 2021). The specific ELISA kits employed in the study include the nuclear subunit of nuclear factor kappa B (NF‐kB p65; LSBio, Lynnwood, WA, USA); NLRP3 (Aviva Systems Biology, San Diego, CA, USA); ASC (MyBioSource, San Diego, CA, USA); cleaved caspase‐1 (BioVision Inc., Milpitas, CA, USA); interleukin‐18 (IL‐18), IL‐1β (R&D Systems, Minneapolis, MN, USA); tumor necrosis factor‐alpha (TNFα; R&D Biosystems); macrophage inflammatory protein‐1 alpha (MIP1α; LSBIO); and IL‐6 (R&D Biosystems).

The extent of pS6 expression was quantified within NeuN+ neurons and IBA‐1+ microglia in the hippocampus and cerebral cortex using Z‐section analysis in a Leica THUNDER 3D Imager (3 images/section, three sections/animal, n = 6 animals/group). The pS6 expression in the cytoplasm of neurons and microglia varied within and across groups. Therefore, the AF of pS6+ staining within the soma of neurons and microglia was measured using Image J. The fractions of pS6+ structures in neurons and microglia were measured from all hippocampus subfields (DG, CA1, and CA3) to compile data for the entire hippocampus (3 sections/animal, n = 6/group). We also measured percentages of neurons and microglia presenting pS6.

Tissue lysates from the hippocampus and cerebral cortex were also utilized to quantify the phospho‐mTOR and pan‐mTOR using an ELISA kit (n = 6 animals/group, Ray Biotech, GA, USA). Next, we determined the ratio of phospho‐mTOR and pan‐mTOR and compared them across the groups.

The p62+ structures within microglia and neurons in different hippocampal subfields (the DG, CA1, and CA3 subfields) were quantified. First, the percentages of IBA‐1+ microglia expressing p62+ structures in the entire hippocampus were computed and compared between naïve, AD, and AD+PLX groups (3 sections/animal, n = 6 animals/group). Next, the area fraction (AF) of p62+ structures within the soma of IBA‐1+ microglia and NeuN+ neurons were quantified using Image J. Fourteen microglia from 2 to 3 images in each animal were measured for each brain region (n = 6/group). Neurons were evaluated using ten randomly selected NeuN+ neurons in each cell layer from 2 to 3 images per animal (n = 6/group).

The lysates from the hippocampus were utilized to quantify the various autophagy markers using ELISA. The kits employed in the study include beclin‐1 (Aviva Systems Biology, San Diego, CA, USA), autophagy‐related gene‐5 (ATG‐5, MyBioSource), and microtubule‐associated protein‐1 light chain 3B (MAP1LC3B, MyBioSource).

Aβ plaques in the hippocampus and cerebral cortex were quantified using serial sections (every 15th) and Image J (n = 6 animals/group). Serial sections through the posterior half of the hippocampus were employed since Aβ plaques were minimal in the anterior half of the hippocampus in both AD and AD+PLX groups. Furthermore, we determined the concentrations of Aβ‐42 in the hippocampus and cerebral cortex tissue lysates using an ELISA kit from Invitrogen (n = 6 animals/group).

The status of hippocampal neurogenesis was measured through the stereological quantification of doublecortin‐positive (DCX+) newly born neurons in the subgranular zone‐granule cell layer (SGZ‐GCL) of the DG using serial sections (every 15th, n = 6/group) through the entire hippocampus as detailed in our previous reports (Attaluri et al., 2022; Hattiangady et al., 2007; Rao et al., 2007; Shetty et al., 2020).

The animal numbers for immunohistochemical studies per group were determined through a power analysis using G*Power software, using the effect size of 1.2 (based on our previous data) and alpha of 0.05, which suggested a requirement of final data from a minimum of 4 mice/group to obtain a power of 0.8 and above. However, to obtain robust datasets, we employed n = 6/group in all studies. We used one‐way ANOVA with Tukey's post hoc tests to compare data across three groups. We performed the Kruskal–Wallis test with Dunn's post hoc tests when individual groups did not pass the normality test. In comparisons involving two groups, we employed a two‐tailed, unpaired Student's t‐test (or a Mann–Whitney U‐test when standard deviations differed significantly). A statistically significant value of p < 0.05 was employed in all comparisons.

Immunostaining for IBA‐1 visualized the density and distribution of microglia. The examples from the hippocampal and cerebral cortex regions of naïve, AD, and AD+PLX groups are illustrated (Figure 1a–l; Figure S1A–D). Microglia in AD mice were seen as clusters as well as individual cells (Figure 1b,e,h,k) in contrast to naïve control mice displaying a homeostatic phenotype, with smaller soma and highly ramified processes (Figure 1a,d,g,j). Furthermore, microglia within clusters in AD mice displayed the activated phenotype, typified by hypertrophied soma and short processes with no or reduced ramifications (Figure 1e,k). On the other hand, the residual microglia in AD+PLX mice were mainly scattered and displayed the phenotype of homeostatic microglia, including smaller soma and highly ramified processes (Figure 1c,f,i,l). Stereological quantification revealed that the total number of microglia significantly increased in the hippocampus and cerebral cortex of the AD group compared to the naïve control group (p < 0.01–0.0001; Figure 1m,n). Moreover, the microglial number was significantly reduced within both the hippocampus and cerebral cortex in the AD+PLX group compared to the AD group (65% reduction, p < 0.0001) and the naïve control group (p < 0.0001, Figure 1m,n). The reduced number of microglia in the AD+PLX group was also associated with a significant reduction in microglial clusters (40% reduction, p < 0.05–0.01, Figure 1o,p).

We next investigated plaque‐associated microglia using Aβ‐42 and IBA‐1 dual immunofluorescence in AD and AD+PLX groups (Figure 1 [q‐ad]). In both groups, microglia were observed around plaques. In the AD group, there was a higher density of cells immediately around the plaques (Figure 1q–w) compared to the AD+PLX group (Figure 1 [x‐ad]). Notably, microglia in the AD + PLX group were dispersed around plaques (Figure 1 [x‐ad]). Quantification of microglia per unit area (100 μm²) of Aβ‐42 plaques revealed a significant reduction in microglial number around plaques in the AD‐PLX group vis‐à‐vis the AD group in the hippocampus (p < 0.05, Figure 1 [ae]) and cerebral cortex (p = 0.05, Figure 1 [af]). Further evaluation through Aβ‐42, IBA‐1, and Clec7a triple immunofluorescence revealed varying Clec7a expression within plaque‐associated microglia between AD and AD‐PLX groups. While all plaque‐associated microglia in the AD group displayed robust Clec7a expression (Figure 1 [ag‐aj]), implying the MGnD phenotype (Krasemann et al., 2017), plaque‐associated microglia in the AD‐PLX group exhibited greatly reduced Clec7a expression (Figure 1 [ak‐an]), likely suggesting a non‐MGnD phenotype.

The expression of CD68 within microglia (i.e., activated microglia) was detected through IBA‐1 and CD68 dual immunofluorescence and Z‐section analysis in a confocal microscope (Figure S2A–I,K–S). Most hypertrophied microglia in AD mice contained large amounts of CD68+ structures, whereas the residual microglia in AD+PLX mice contained reduced CD68+ structures. In the AD group, percentages of IBA‐1+ microglia displaying CD68 were higher in the entire hippocampus and the cerebral cortex compared to the naïve control group (p < 0.0001, Figure S2J,T). The percentages of CD68+ microglia reduced in the AD+PLX group for the hippocampus and the cerebral cortex (p < 0.01–0.0001, Figure S2J,T). Moreover, the percentages of CD68+ microglia in the AD+PLX group were normalized to naïve control levels of the cerebral cortex (p > 0.05, Figure S2T). Thus, 10 days of PLX treatment in AD mice reduced the extent of activated microglia in both the hippocampus and cerebral cortex.

Tracing of microglia and morphometric analysis using Neurolucida and NeuroExplorer revealed that 10 days of PLX treatment in AD mice altered microglial morphology in the hippocampus and cortex (Figure 2 and Figure S3). Examples of microglia from naïve control, AD, and AD+PLX groups are illustrated for the hippocampal DG (Figure 2) and cerebral cortex (Figure S3). The naive control group exhibited homeostatic and non‐inflammatory phenotypic features (Figure 2, Figure S3A,D). In contrast, in AD mice, microglia exhibited proinflammatory phenotype (Figure 2, Figure S3B,E) with reduced process length and a reduced number of nodes and endings (p < 0.05–0.0001, Figure 2, Figure S3G–I). The residual microglia in the AD+PLX group displayed a non‐inflammatory phenotype (Figure 2, Figure S3C,F), akin to that seen in naive control mice, which is evident from increased total process length and a higher number of nodes and endings compared to the AD group (p < 0.05–0.0001, Figure 2, Figure S3G–I). Furthermore, Sholl's analysis of microglial processes revealed that microglia in the AD group displayed a reduced number of intersections, reduced length of processes, and reduced number of nodes and endings at multiple distances from the soma than the naïve group (Figure 2, Figure S3J–M). The AD + PLX group exhibited a higher number of intersections, increased length of the processes, and a higher number of nodes and endings at multiple distances from the soma compared to the AD group (p < 0.05–0.001, Figure 2, Figure S3J–M).

Z‐section analysis of brain tissue sections processed for IBA‐1, NLRP3, and ASC visualized NLRP3 inflammasome complexes (i.e., structures co‐expressing NLRP3 and ASC) within IBA‐1+ microglia in the hippocampus (Figure 3a–f) and cerebral cortex (Figure 4a–f). The presence of NLRP3 inflammasome complex was more frequent in the AD group than in AD+PLX groups (Figures 3 and 4a–f). Notably, microglia in the AD‐PLX group only occasionally displayed the NLRP3 inflammasome complex (Figures 3 and 4g). Quantification revealed that the area fraction of inflammasome complexes in microglia was significantly reduced in the AD‐PLX group compared to the AD group in the hippocampus and cerebral cortex (p < 0.05–0.01, Figures 3 and 4g).

Next, we quantified concentrations of proteins implicated in NLRP3 inflammasome activation and proinflammatory cytokines in the hippocampus and the cerebral cortex. AD mice displayed higher levels of proteins that mediate NLRP3 inflammasome activation (NF‐kB‐p65, NLRP3, ASC, cleaved caspase‐1) compared to naïve mice. The differences were significant for all proteins in the hippocampus (p < 0.05–0.0001, Figure 3h–k) and for NLRP3, ASC, and cleaved caspase‐1 in the cerebral cortex (p < 0.05–0.001, Figure 4i–k). In contrast, in AD+PLX mice, the concentrations of all these proteins were comparable to naïve control mice in both the hippocampus and cerebral cortex (p > 0.05, Figures 3 and 4h–k). The concentrations of end products of NLRP3 inflammasome activation (IL‐1β and IL‐18) and additional proinflammatory cytokines (TNFα, MIP1α, and IL‐6) also showed a similar trend. AD mice displayed higher concentrations of these cytokines than naïve mice (p < 0.05–0.001, Figures 3 and 4l–p), whereas AD+PLX mice exhibited similar concentrations as naïve control mice for all these cytokines in the hippocampus and IL‐1β, IL‐18, MIP1α, and IL‐6 in the cerebral cortex (p > 0.05, Figures 3 and 4l,m,o,p). Ten days of PLX treatment in AD mice diminished the concentration of various proteins involved in NLRP3 inflammasome activation and multiple downstream proinflammatory cytokines.

Examples of GFAP immunoreactive astrocytes in DG, CA1, and CA3 subfields and the cerebral cortex are illustrated (Figure S4A–I,N–P). Quantification using Image J revealed that AFs of astrocyte elements in the hippocampal CA3 region and cerebral cortex of the AD group were elevated compared to the naïve group (p < 0.05–0.001; Figure S4L,Q). In the AD+PLX group, the AF of astrocyte elements was normalized to levels in the naïve group in the CA3 subfield (p > 0.05; Figure S4L) but not in the cerebral cortex (p < 0.001; Figure S4Q), where AF of astrocyte elements remained comparable to AF in the AD group (p > 0.05). However, AFs of astrocyte elements in the DG, CA1 subfield, and the entire hippocampus did not show significant differences between naive, AD, and AD‐PLX groups (p > 0.05; Figure S4J,K,M). Thus, 10 days of PLX treatment in AD mice did not alter astrocyte hypertrophy in the hippocampus and cerebral cortex.

mTOR signaling within hippocampal and cerebral cortical neurons and microglia were visualized and quantified through NeuN/pS6 and IBA‐1/pS6 dual immunofluorescence, Z‐section, and Image J analyses (Figures 5 and 6a–f,h–m). pS6 is one of the primary downstream targets and the effector of the mTOR pathway and hence serves as a marker for mTOR activity (Leontieva et al., 2012; Morgan‐Warren et al., 2013). AF analysis of pS6 expression within individual neurons and the percentages of neurons expressing pS6 suggested diminished pS6 expression within hippocampal and cerebral cortical neurons in the AD+PLX group compared to the AD group (Figures 5 and 6g,h). However, such reductions were statistically significant only for the cerebral cortex (p < 0.01, Figure 6g,h). AFs of average pS6 expression within individual microglia and the percentages of microglia expressing pS6 were decreased in the AD+PLX group compared to the AD group in the hippocampus and cerebral cortex (Figures 5 and 6o,p). However, statistically, only the AFs of average pS6 expression within individual microglia were significant between AD and AD‐PLX groups (p < 0.05–0.01, Figures 5 and 6o). Thus, microglia depletion significantly reduced the extent of downstream mTOR signaling within individual neurons in the cerebral cortex and within individual microglia in both hippocampus and the cerebral cortex.

To further probe the extent of changes in mTOR signaling, we measured the concentrations of pan‐mTOR, phospho‐mTOR, and the ratio of phospho‐ and pan‐mTOR in the hippocampus and the cerebral cortex (Figures 5 and 6q–s). The ratio of phospho‐ and pan‐mTOR levels was upregulated in the AD group compared to the naive group in both the hippocampus and the cerebral cortex (p < 0.01, Figures 5 and 6s), but the AD+PLX group displayed a similar ratio of phospho‐ and pan‐mTOR as the naive group (p > 0.05, Figures 5 and 6s), implying reduced mTOR signaling in both brain regions. Thus, 10 days of PLX treatment in AD mice did significantly dampen mTOR signaling in both the hippocampus and cerebral cortex.

Dual immunofluorescence for IBA‐1/p62 and NeuN/p62 and Z‐section analysis and quantification via Image J revealed the extent of autophagy within neurons (Figure 7a–f; Figure S5A–F) and microglia (Figure 7h–m; Figure S5H–M) in the hippocampus and cerebral cortex. The p62 protein serves as a classical marker of autophagic flux because it accumulates with autophagic cargo but declines when autophagy is active (Liu et al., 2019; Pankiv et al., 2007). Compared to the AD group, the AD‐PLX group displayed reduced AFs of p62 within hippocampal and cerebral cortex neurons (Figure 7g; Figure S5G). However, such reductions were statistically significant only in the cerebral cortex (p < 0.01, Figure S5G). Analysis of the p62 expression in microglia also revealed a similar trend. The AD+PLX group exhibited reduced p62 expression compared to the AD group in the hippocampus and cerebral cortex (p > 0.001–0.0001, Figure 7n; Figure S5N). Overall, p62 expression in neurons and microglia implied improved autophagy in the AD+PLX group, although the improvement was not statistically significant in hippocampal neurons. Consistent with these results, we found decreased concentrations of beclin‐1 and ATG‐5 in the hippocampus and cerebral cortex of the AD group compared to the naive group (Figure 7o; Figure S5O), and the decrease was significant for beclin‐1 in the hippocampus (Figure 7o). Moreover, both beclin‐1 and ATG‐5 concentrations in the AD+PLX group were comparable to the naive group in the hippocampus (p > 0.05, Figure 7o,p) and significantly higher than the AD group (p < 0.01, Figure 7o,p). Measurements from the cerebral cortex showed a similar trend, but the differences were not statistically significant. Thus, partial microglia depletion improved autophagy within the leftover microglia in both the hippocampus and the cerebral cortex of 5xFAD mice. Furthermore, microglia depletion in 5xFAD mice also improved autophagy in cortical neurons and showed a similar trend in hippocampal neurons.

Examples of Aβ plaques from the hippocampus and cerebral cortex are illustrated (Figure S6A–D). In these regions, the density of Aβ plaques appeared similar between AD and AD+PLX groups. Quantification of AFs of Aβ plaques in these brain regions using Image J confirmed that the extent of Aβ plaques in the AD group matched the AD+PLX group (p > 0.05, Figure S6E,F). Furthermore, the concentrations of soluble Aβ‐42 were also not statistically significant between the AD and AD+PLX groups in both the hippocampus and cerebral cortex (p > 0.05, Figure S6G,H). Thus, 10 days of PLX treatment in AD mice did not significantly reduce the extent of Aβ plaques and the soluble Aβ‐42 in the hippocampus and cerebral cortex.

The extent of hippocampal neurogenesis was ascertained through DCX immunostaining (Figure S6I–N). Stereological quantification of DCX+ newly born neurons in the SGZ‐GCL revealed no differences between naïve, AD, and AD+PLX groups (p > 0.05, Figure S6O). Thus, 10 days of PLX treatment in AD mice did not alter hippocampal neurogenesis.

A series of assessments has verified that the residual microglia in the hippocampus and cerebral cortex of 5xFAD mice following 10 days of CSF1R inhibition predominantly displayed a noninflammatory phenotype. These assessments encompassed morphometric analyses quantifying the intricacies of microglial processes, such as branches, nodes, and process lengths at various distances from the soma, illustrating the resemblance of residual microglial processes to those of the naive control group. Additional observations supporting the predominant elimination of activated microglia in AD mice receiving 10 days of CSF1R inhibition include the AD group receiving a standard diet displaying large numbers of microglial clusters comprising microglia with hypertrophied soma, thicker and shorter processes, and increased incidence of NLRP3 inflammasome complexes. In contrast, in the AD+PLX group, such clusters were reduced, and the dispersed residual microglia displayed diminished NLRP3 inflammasome complexes. Moreover, while microglia were found around plaques in both AD and AD+PLX groups, there were significantly fewer plaque‐associated microglia in the AD+PLX group compared to the AD group. Additional evaluation through Aβ‐42, IBA‐1, and Clec7a triple immunofluorescence demonstrated that all plaque‐associated microglia in the AD group exhibited pronounced expression of Clec7a, indicative of an MGnD phenotype (Krasemann et al., 2017). Conversely, microglia associated with plaques in the AD‐PLX group displayed diminished Clec7a expression, suggesting a non‐MGnD phenotype. There was also an overall reduction in proinflammatory milieu in the hippocampus and cerebral cortex, which could be gleaned from reduced concentrations of NLRP3 inflammasome activation mediators (NF‐kB‐p65, NLRP3, ASC, and cleaved caspase‐1), end products (IL‐1β and IL‐18), and other proinflammatory cytokines such as TNFα, MIP1α, and IL‐6 in the hippocampus and cerebral cortex of the AD‐PLX group compared to the AD group. These results are consistent with the diminished proinflammatory cytokine concentrations seen after CSF1R inhibition in another model of AD (Mancuso et al., 2019).

The exact reasons for leftover microglia displaying a predominantly noninflammatory phenotype in PLX‐treated 5xFAD mice are unclear. Short‐term CSF1R inhibition in the early stage of AD predominantly ablates activated microglia, possibly due to a higher demand of CSF1 for survival by such microglia. Although direct evidence supporting enhanced CSF1R expression in activated microglia due to a higher demand for CSF1 is lacking, an overall CSF1R upregulation has been observed in mouse models of amyloidosis and neurotoxic models of Parkinson's disease (Murphy Jr. et al., 2000; Neal et al., 2020) and post‐mortem samples from AD and Parkinson's disease (PD) patients (Akiyama et al., 1994; Neal et al., 2020). Also, a study has shown that microglial proliferation increases progressively near Aβ plaques in the APP/PS1 AD model (Olmos‐Alonso et al., 2016), which may be mediated through increased CSF1R in plaque‐associated microglia. Repolarizing the activated microglia into a homeostatic phenotype after CSF1R inhibition has also been observed in models of multiple sclerosis (Nissen et al., 2018) and Parkinson's disease (Neal et al., 2020). Regardless of the reasons, the data support the increased vulnerability of activated microglia to CSF1R inhibition, at least in the early stage of AD. It remains to be investigated whether a similar elimination of activated microglia would occur in the later stages of AD. In this context, a study in ~5.5 months old tau‐seeded 5xFAD mice showing that 45 days of CSF1R inhibition preferentially eliminates non‐plaque‐associated microglia vis‐à‐vis plaque‐associated microglia is relevant (Lodder et al., 2021). The discrepancy between the latter study and the current study regarding the type of microglia eliminated following CSF1R inhibition likely reflects the age of the mice, disease stage, the extent of microglial activation, and the duration of CSF1R inhibition.

Reductions in NLRP3 inflammasome complexes in the residual microglia following 10 days of CSF1R inhibition in the early stage of AD are beneficial because NLRP3 inflammasome activation perpetuates chronic neuroinflammation and progression of AD pathogenesis (Bai & Zhang, 2021; Ising et al., 2019). The NLRP3 inflammasome complex comprises NLRP3, ASC, and caspase‐1 precursor proteins. Increases in NLRP3, a NOD‐like receptor protein with pattern recognition, recruit caspase‐1 precursor protein by binding to ASC through the pyrin domain at the N‐terminal, which activates caspase‐1. Activation of caspase‐1 facilitates the maturation of pro‐IL‐1β and pro‐IL‐18 into mature IL‐1β and IL‐18. The secreted mature IL‐1β and IL‐18 contribute to neurodegeneration through downstream inflammatory cascades (Franchi et al., 2009; Zhang et al., 2017). The events that upregulate NLRP3 in AD include increased levels of the danger‐associated molecular patterns (DAMPs) such as adenosine triphosphate, reactive oxygen species, and cathepsin B (Savage et al., 2012), which activate nuclear factor kappa B through toll‐like receptors (Yang et al., 2020). Mitochondrial DNA released from the damaged mitochondria can also directly upregulate NLRP3 in AD via potassium outflow or calcium influx (Murphy et al., 2012; Wyss‐Coray, 2006). In addition, activation of NLRP3 inflammasomes in AD can occur by fibrillar Aβ aggregates, Aβ oligomers, and protofibrils (Friker et al., 2020).

While activation of the NLRP3 inflammasome is beneficial to the organism in certain conditions as it can restrain microbial infection or endogenous cell damage, its overactivation is detrimental in AD (Hanslik & Ulland, 2020; Ising et al., 2019; Heneka et al., 2013). Overactivation of the NLRP3 inflammasomes in microglia transforms them into a proinflammatory phenotype, resulting in reduced phagocytosis of Aβ‐42 by microglia, which can lead to enhanced Aβ deposition and progression of AD pathogenesis (Lučiūnaitė et al., 2020). Such effects are also apparent in microglia with reduced NLRP3 inflammasome activation, displaying a noninflammatory phenotype and exhibiting increased phagocytosis of Aβ (Cherry et al., 2014). Moreover, NLRP3 inflammasome overactivation promotes the pathological formation of tau protein, which can accelerate AD‐related neurodegeneration (Ising et al., 2019). Therefore, many studies have focused on inhibiting NLRP3 inflammasome activation in AD to suppress chronic neuroinflammation and improve brain function (Bai & Zhang, 2021; Feng et al., 2020; Lonnemann et al., 2020; Milner et al., 2021; Van Zeller et al., 2021). For example, reduction of caspase‐1 and IL‐1β through knockdown of NLRP3 in AD models can considerably eliminate Aβ‐42 (Wu et al., 2017), and functional loss of the NLRP3 inflammasome could diminish hyperphosphorylation and aggregation of tau by regulating tau kinase and phosphorylase (Dempsey et al., 2017). Thus, short‐term CSF1R inhibition resulting in reduced NLRP3 inflammasome complexes in residual microglia in the early stage of AD, along with an overall reduction in concentrations of NLRP3 inflammasome activation mediators and end products observed in this study, promises to maintain better brain function for extended periods. However, it remains to be addressed whether short‐term CSF1R inhibition at regular intervals (e.g., 10 days of CSF1R inhibition every 6–8 weeks) would postpone cognitive and memory problems of AD for extended periods.

A predominant elimination of activated microglia in AD mice receiving 10 days of CSF1R inhibition led to decreased mTOR signaling within neurons in the cerebral cortex and microglia in both the hippocampus and the cerebral cortex. Moreover, the overall mTOR signaling, indicated by concentrations of p‐mTOR and the ratio of p‐mTOR and pan‐mTOR, decreased in AD mice receiving 10 days of CSF1R inhibition. mTOR hyperactivation has been seen in both animal models of AD and post‐mortem brain samples from AD patients (Caccamo et al., 2010, 2014; Li et al., 2005). Increased mTOR signaling in AD increases Aβ production and promotes aggregation by directly inhibiting autophagy (Caccamo et al., 2011; Zhao et al., 2015). The precise reason for decreased mTOR signaling with the predominant elimination of activated microglia in the early stage of AD observed in this study is unknown. It could be due to an overall decrease in NLRP3, as studies have shown that NLRP3 is a binding partner of mTOR, and NLRP3 silencing reduces the phosphorylation of mTOR (Cosin‐Roger et al., 2017). Thus, reduced NLRP3 levels, as observed in this study, diminished the physical interaction between NLRP3 and mTOR with a net effect of diminished mTOR signaling. One of the benefits of decreased mTOR signaling is the resulting increase in autophagy. Indeed, increased autophagy in the AD‐PLX group, evident from diminished p62 expression, was particularly apparent within microglia in the hippocampus and the cerebral cortex. Moreover, compared to the AD group, there was an overall increase in autophagy in the hippocampus of the AD‐PLX group, indicated by higher levels of autophagy enhancers like Beclin‐1 and ATG‐5. In this study, the increased autophagy in the AD‐PLX group is likely due not only to reduced mTOR signaling but also to lower levels of proinflammatory cytokines in the environment, as elevated proinflammatory cytokines like TNF‐α can impede autophagic flux in microglia (Jin et al., 2018). Thus, the reduced NLRP3 inflammasome activation, decreased mTOR signaling, and increased autophagy in residual microglia after 10‐day CSF1R inhibition are closely connected.

Enhanced autophagy in residual microglia is beneficial because it helps remove NLRP3 inflammasome activators, such as intracellular DAMPs, NLRP3 inflammasome components, and cytokines (Galluzzi et al., 2017; Ma et al., 2013). Conversely, dysfunctional autophagy can result in inflammatory diseases due to excessive inflammasome activation (Deretic et al., 2013; Levine et al., 2011). Therefore, maintaining increased autophagy in AD is crucial for reducing the hyperinflammatory response. Enhanced autophagy in microglia may also improve the phagocytosis of Aβ‐42 (Lucin et al., 2013; Plaza‐Zabala et al., 2017). However, the current study did not find reductions in Aβ plaques in the hippocampus or cerebral cortex after 10‐day CSF1R inhibition. The results are consistent with previous studies. A study has reported that 4 months of CSF1R inhibition commencing 1.5 months of age in 5xFAD mice did not alter Aβ levels or APP processing despite considerably depleting microglia (Spangenberg et al., 2019). Furthermore, a month‐long CSF1R inhibition in an advanced stage of the disease (i.e., in 10‐month‐old 5xFAD mice) did not affect either Aβ plaques or soluble or insoluble fractions of Aβ (Spangenberg et al., 2016). Also, two‐week CSF1R inhibition in aged 3xTg and APP/PS1 mice did not reduce Aβ plaques (Karaahmet et al., 2022). However, studies employing CSF1R inhibition for extended periods in 5xFAD mice reported diminished parenchymal plaque development (Son et al., 2020; Spangenberg et al., 2019). Thus, the overall impact of inhibiting CSF1R on Aβ plaques is minimal unless the inhibition is maintained for extended periods (Son et al., 2020; Spangenberg et al., 2019). However, inhibiting CSF1R for short or moderate durations can reduce synapse loss and neurodegeneration and increase circuitry connectivity in AD models (Liu et al., 2021; Spangenberg et al., 2016).