PTP1B inhibition promotes microglial phagocytosis in Alzheimer’s disease models by enhancing SYK signaling

Abstract

Amyloid-β (Aβ) accumulation is a hallmark of Alzheimer’s disease (AD). Emerging evidence suggests that impaired microglial Aβ phagocytosis is a key feature in AD, highlighting the therapeutic potential of enhancing this innate immune function. Here, we demonstrate that genetic deletion or pharmacological inhibition of protein tyrosine phosphatase 1B (PTP1B) ameliorated memory deficits and reduced Aβ burden in APP/PS1 mice. Moreover, we show that PTP1B was highly expressed in microglia, and its deficiency promoted a transcriptional shift toward immune activation and phagocytosis. Consistently, PTP1B deletion in microglia enhanced phagocytosis and energy metabolism, supported by increased AKT-mTOR signaling, a pathway essential for meeting the energy demands of activation. Mechanistically, we identified spleen tyrosine kinase (SYK), a key regulator of microglial phagocytosis, as a direct substrate of PTP1B. Inhibition of SYK showed that PTP1B modulates microglial activation in a SYK-dependent manner. These findings established PTP1B as a critical modulator of microglial activation and a potential therapeutic target for AD.

Introduction

Alzheimer’s disease (AD) is the most common form of dementia and an urgent global issue. As its prevalence rises, the demand for healthcare services grows, placing an increasing burden on society (1). AD has a complex pathogenesis, with the accumulation of amyloid β (Aβ) plaques in the brain recognized as a key initiating event that ultimately contributes to neurodegeneration and cognitive decline, as proposed by the amyloid cascade hypothesis (2, 3). Aβ exists in multiple aggregation states, ranging from soluble oligomers to insoluble fibrillar aggregates. Insoluble deposits form plaques that represent advanced stages of pathology, whereas the soluble species are now considered as the most synaptotoxic and proinflammatory species (4). Therapeutic anti-amyloid antibodies, such as aducanumab and lecanemab, target aggregated Aβ species at different assembly stages but, despite effectively reducing Aβ burden, have shown limited clinical benefits with concerns regarding potential side effects (5, 6). This suggests a need for novel therapeutic approaches that target multiple aspects of the disease.

In recent years, growing evidence suggests that brain metabolic dysfunction plays a critical role in AD pathology (7). Impaired cerebral glucose uptake and utilization may precede cognitive dysfunction by years, or even decades (8-10), suggesting this brain hypometabolic state contributes to AD progression. Additionally, systemic metabolic disorders, such as obesity and type 2 diabetes, are well-established risk factors for AD (11, 12). These conditions are associated with chronic inflammation, insulin resistance, and mitochondrial dysfunction, all of which may exacerbate AD pathology (13). Together, these findings suggest that targeting metabolic dysregulation and brain energy metabolism could open new therapeutic opportunities for AD (14). Consistent with this, repurposing antidiabetic drugs, such as insulin and GLP-1 receptor agonists, has shown promising preclinical and early-phase clinical results for AD (15-17).

Protein tyrosine phosphatase 1B (PTP1B) plays a central role in maintaining glucose homeostasis and energy balance (18-21). It attenuates insulin signaling and has been implicated in the development of insulin resistance (18). Moreover, PTP1B directly regulates leptin signaling in hypothalamic neurons (19-21). Inhibition or genetic deletion of PTP1B enhances insulin and leptin sensitivity and improves glucose metabolism in models of obesity and type 2 diabetes (22-24). Furthermore, PTP1B inhibition restored impaired insulin signaling and improved the behavioral deficits in a Rett syndrome model (25), suggesting its potential to alleviate metabolic dysfunction in neurological diseases. However, PTP1B has long been considered a challenging drug target due to its highly polar and conserved catalytic site (26). Notably, recent progress in development of allosteric inhibitors has made it possible to target PTP1B selectively, including in the brain. One such compound, MSI-1436, binds the C-terminal regulatory segment of PTP1B and has showed efficacy in preclinical models (27), progressing to clinical trials for obesity and type 2 diabetes, which were discontinued due to commercial considerations rather than scientific limitations (28). In this study, we utilized a novel MSI-1436 derivative, DPM-1003, characterized in our lab (24, 29), to explore the therapeutic potential of PTP1B inhibition in an AD animal model.

Beyond its role in metabolic regulation, PTP1B has also been recognized as an important modulator of immune cell signaling. It is well established that PTP1B can dephosphorylate the JAK and TYK2 kinases and inactivate STAT signaling (30). In addition, it has been shown to suppress T cell-mediated antitumor activity through effects on JAK/STAT5 signaling (31). In macrophages, PTP1B regulates myeloid cell differentiation and activation through CSF1 and IFN-γ signaling (32, 33). Considering that microglia are brain-resident macrophages, these findings prompt a broader immune-regulatory role for PTP1B that may extend to microglia. Interestingly, recent findings suggest that PTP1B is a positive regulator of microglia-mediated neuroinflammation (34); however, these findings were observed outside of the context of AD, highlighting the need for further investigation.

Microglia play a key role in AD by mediating chemotaxis and phagocytosis to clear toxic aggregates. Notably, anti-amyloid immunotherapies often rely on microglial activation to clear Aβ plaques (35, 36), highlighting the potential importance of targeting microglial function in AD therapies. Interestingly, microglial activation has been associated with enhanced cerebral glucose uptake in AD mouse models and patients (35, 36), suggesting that metabolic regulation is closely linked to microglial function. Indeed, modulating microglial metabolism directly impacts their functions, such as phagocytosis(37-41), suggesting that targeting metabolic pathways could enhance Aβ clearance and mitigate disease progression. Among the signaling pathways, SYK plays a central role downstream of multiple immunoreceptors in microglia, coordinating both metabolic fitness and phagocytosis (42-44). Interestingly, a previous study suggested that loss of PTP1B enhanced B cell activation signaling, including the phosphorylation of SYK (45). Based on this, we hypothesized a possible connection between PTP1B and SYK signaling in microglia. Considering its role in both metabolic and immune signaling, PTP1B has the potential to regulate microglial function at the intersection of these pathways.

In this study, we investigated whether PTP1B could serve as a therapeutic target for AD. Using the APP/PS1 mouse model of AD, we demonstrated that deletion or pharmacological inhibition of PTP1B significantly improved learning and memory and reduced Aβ burden in the brain. In addition, PTP1B deletion triggered a more reactive and phagocytic transcriptional profile in microglia, enhancing their phagocytic activity in vitro and in vivo. Mechanistically, we demonstrated that SYK is a direct substrate of PTP1B, and that this interaction is critical for regulating microglial activation in response to Aβ stimulation. Together, these findings highlight a critical role of the phosphatase in microglial function and suggest that targeting PTP1B may represent a promising strategy to enhance microglial-mediated Aβ clearance and mitigate AD progression.

Results

Deletion and inhibition of PTP1B in APP/PS1 mice resulted in beneficial cognitive effects

To test whether PTP1B can serve as a potential therapeutic target for AD, we cross-bred PTP1B knockout mice with APP/PS1 mice (APP/PS1-PTP1B^−/−). We chose APP/PS1 mice as our AD mouse model for two reasons. First, this model is well-established to study amyloid pathology, as it exhibits age-dependent Aβ accumulation starting at 6-months of age and progressing to cognitive deficits around 12-months (46, 47). Second, this mouse model demonstrates impaired glucose and insulin tolerance as early as 6-months (48, 49).

By 12-13 months of age, these mice were examined in three different behavior tests (fig. S1A). We performed an Open Field Test (OFT) to assess locomotor activity and anxiety-like behavior (Fig. S1B). PTP1B deletion did not significantly affect total distance moved (Fig. S1C) or time spent in the center zone (Fig. S1D) in either WT or APP/PS1 mice; as anxious mice typically avoid the central zone, this indicates that anxiety levels, as well as general motor ability, were unchanged. Recognition memory was assessed by the Novel Object Recognition (NOR) behavior test, which leverages the natural tendency of rodents to explore novel objects over familiar ones (Fig. 1A). Interestingly, PTP1B deletion improved recognition memory in APP/PS1 mice, as shown by increased exploration of the novel object compared to APP/PS1 controls (Fig. 1B). Furthermore, spatial learning and memory were evaluated using the Morris Water Maze (MWM) behavior test (Fig. 1C). During the acquisition trials, PTP1B deletion in APP/PS1 mice shortened the time taken to find the platform (escape latency) compared to APP/PS1 mice (Fig. 1D), suggesting improved spatial learning. Twenty-four hours after the last training session, with the platform removed, PTP1B deletion in APP/PS1 mice increased the number of crossings over the previous location (Fig. 1E) and the time spent in the target quadrant (Fig. 1F), indicating improved spatial memory. No differences in swimming speed were observed among the groups (fig. S1B), confirming that the observed effects were not influenced by variations in motor function. Importantly, when analyzed separately by sex, both male and female APP/PS1-PTP1B−/− mice showed a similar improvement in both NOR and MWM (Fig. S1F-G), indicating that the effect of PTP1B deletion was not gender dependent.

Figure 1. Deletion or inhibition of PTP1B in APP/PS1 model improved learning and memory.

(A) Schematic of the NOR test. (B) Percentage of exploration of old or novel object in WT- PTP1B^+/+ (n=16, 7 males, 9 females), WT-PTP1B^−/−(n=18, 8 males, 10 females), APP/PS1-PTP1B^+/+ (n=20, 10 males, 11 females), APP/PS1-PTP1B^−/− (n=18, 9 males, 9 females); unpaired two-tailed t-tests. (C) Schematic of the MWM tests. (D-F) MWM test in WT-PTP1B^+/+ (n=16, 7 males, 9 females), WT-PTP1B^−/− (n=18, 8 males, 10 females), APP/PS1-PTP1B^+/+ (n=20, 10 males, 10 females), APP/PS1-PTP1B^−/− (n=20, 10 males, 11 females); (D) Escape latency to submerged platform in acquisition trials (repeated-measures two-way ANOVA with Tukey’s post hoc test; *: WT-PTP1B^+/+ vs APP/PS1-PTP1B^+/+; #: APP/PS1-PTP1B^+/+ vs APP/PS1-PTP1B^−/−); (E) Crossing frequency in the platform area in probe trial and (F) Time in target quadrant in probe trial (One-way ANOVA with Tukey’s post hoc test); the symbol legend in Fig. 1B also applies to Fig. 1D-F. (G) Percentage of exploration on old or novel object on WT-veh (n= 16, 6 males and 10 females), WT-DPM-1003 (n= 19, 9 males and 10 females), APP/PS1-veh (n= 27, 13 males and 14 females), APP/PS1-DPM-1003 (n= 24, 14 males and 10 females) unpaired two-tailed t-tests. (H-J) Morris water maze in WT-veh (n= 17, 7 males and 10 females), WT-DPM-1003 (n= 19, 9 males and 10 females), APP/PS1-veh (n= 27, 15 males and 12 females), APP/PS1-DPM-1003 (n= 27, 14 males and 13 females); (H) Escape latency to submerged platform in acquisition trials (repeated-measures two-way ANOVA, Tukey’s post hoc test was applied; *: WT-veh vs APP/PS1-veh; #: APP/PS1-veh vs APP/PS1-DPM-1003); (I) Crossing frequence in the platform area in probe trial and (J) Time in target quadrant in probe (One-way ANOVA with Tukey’s post hoc test); the symbol legend in Fig. 1G also applies to Fig. 1H-J.

Data represent mean ± SEM; p > 0.05(ns), p < 0.05 (*), p < 0.05(* or #), p < 0.01 (** or ##), p < 0.001 (***or ###), p < 0.0001 (**** or ####).

In addition, we examined the effects of an allosteric PTP1B inhibitor, DPM-1003, in APP/PS1 mice. Starting from 11 months of age, a stage at which the mice exhibit significant amyloid plaque deposition and cognitive deficits, the mice received the compound for 5 weeks (fig. S2A). Similar to gene ablation, PTP1B inhibitor treatment significantly improved the recognition memory, as illustrated by an increased novel object exploration percentage in DPM-1003-treated APP/PS1 mice compared to the saline-treated controls (Fig. 1G). In the MWM, DPM-1003-treated APP/PS1 mice showed shorter escape latency (Fig. 1H) and enhanced spatial memory, indicated by increased crossings of the former position of the platform and longer time spent in the target quadrant (Fig. 1I-J). Swimming speed was comparable across all groups (fig. S2B). These results demonstrate that both PTP1B deletion and inhibition alleviate cognitive impairments in APP/PS1 mice.

Deletion and inhibition of PTP1B reduced Aβ levels in APP/PS1 mice.

Since Aβ accumulation is associated with cognitive deficits in APP/PS1 mouse model, we investigated whether PTP1B deletion or inhibition could impact Aβ burden in this model. Thioflavin S (ThioS) selectively labels fibrillar Aβ, whereas 6E10 detects both the fibrillar and non-fibrillar forms (50); this dual labeling enables a comprehensive assessment of Aβ burden. Notably, WT animals did not show any plaques (fig. S3A), whereas APP/PS1 mice displayed plaque formation (Fig. 2A). Quantitative analysis in the hippocampus of APP/PS1 mice indicated that PTP1B deletion reduced amyloid deposition. Specifically, the area covered by ThioS-positive plaques was reduced by 33% in APP/PS1 lacking PTP1B mice compared to APP/PS1 controls (1.7 ± 0.11 vs 1.1 ±0.08) (Fig. 2B), and 6E10-positive area was reduced 27% (1.8 ± 0.1 vs 1.3 ± 0.08) (Fig. 2C). Similarly, treatment with DPM-1003 in APP/PS1 mice also led to reductions in Aβ burden, with ThioS- positive areas reduced by 28% (1.5 ± 0.17 vs 1.1 ±0.08) and 6E10-positive area decreased by 26% (1.5 ± 0.12 vs 1.1 ±0.08) (Fig. 2D-F).

Figure 2. Deletion or inhibition of PTP1B in APP/PS1 reduced Aβ levels in APP/PS1 mice.

(A-C) Immunofluorescence and quantitation of Aβ levels in APP/PS1 mice with or without PTP1B in hippocampal region; (A) Representative images showing Thioflavin S and 6E10 staining, Scale bar =500μm; Quantitation of percent area occupied by ThioS (B) and 6E10 (C); n=8 mice/group (4 males, 4 females), with 3-4 brain slices analyzed per mouse. (D-F) Immunofluorescence and quantitation of Aβ levels in APP/PS1 mice with or without DPM-1003 treatment in hippocampal region; (D) Representative images showing Thioflavin S and 6E10 staining, Scale bar =500μm; Quantitation of percent area occupied by ThioS (E) and 6E10 (F); n=8 mice/group (4 males, 4 females), with 3-4 brain slices analyzed per mouse. (G) ELISA quantitation of Aβ_1–42 levels in the brain diethylamine (soluble) and formic acid (insoluble) fractions of cortical tissue in APP/PS1-PTP1B^+/+ (n= 15, 8 males and 7 females) and APP/PS1-PTP1B^−/− (n= 16, 7 males and 9 females); (H) ELISA quantitation of Aβ_1–42 levels in the brain diethylamine (soluble) and formic acid (insoluble) of cortical tissue in APP/PS1-veh (n= 11, 5 males and 6 females) APP/PS1-DPM-1003 (n= 9, 5 males and 4 females)

Each dot represents an individual animal; data represent mean ± SEM; unpaired two-tailed t-tests; p < 0.05 (*), p < 0.01 (**), p < 0.001 (***).

Considering both soluble Aβ and insoluble Aβ aggregates contribute to AD pathology (2, 3), we examined whether PTP1B deletion altered levels of both forms in APP/PS1 mice. We performed sequential extraction of brain tissues using diethylamine for soluble Aβ and formic acid for insoluble Aβ (51). Notably, ELISA quantitation revealed that PTP1B deletion in APP/PS1 mice reduced soluble Aβ42 levels by 48% and insoluble Aβ42 levels by 35% (Fig. 2G). Similarly, treatment with DPM-1003 in APP/PS1 mice resulted in reduction in soluble Aβ42 by 28% and insoluble Aβ42 by 30% (Fig. 2H). These findings suggest that both genetic deletion and pharmacological inhibition of PTP1B attenuate amyloid pathology in APP/PS1 mice.

The levels of Aβ in the brain are dependent on the dynamic equilibrium between Aβ production and clearance. Aβ is produced through the proteolytic processing of APP, by both β- and γ-secretases. Interestingly, immunoblot analysis of brain lysates of APP/PS1 mice revealed that whereas PTP1B deletion reduced Aβ levels, it did not alter the expression levels of full-length APP, APP C-terminal fragments (APP-CTFs), or key APP-processing enzymes, including Beta-site APP cleaving enzyme-1 (Bace-1) and Presenilin 1 (PS1), a component of the γ-secretase complex (fig. S3B-S3D). Notably, PS1 expression was reduced in APP/PS1 mice due to the presence of a PS1 deletion mutation. These findings suggest that the reduction in Aβ levels following PTP1B deletion is likely driven by enhanced Aβ clearance rather than altered APP expression or processing in APP/PS1 mice.

PTP1B is highly expressed in brain immune cells and limits microglial reactivity in APP/PS1 mice

Considering the reduced Aβ burden in APP/PS1-PTP1B^−/− mice, and since microglia are known to play a crucial role in Aβ uptake and degradation, we hypothesized that these effects may be driven by enhanced microglial activity. To investigate this, we conducted an unbiased assessment of gene expression changes in the brains of APP/PS1 mice using single-cell RNA sequencing. Specifically, we used 8 age-matched (13–month-old) female mice, equally divided between APP/PS1 and APP/PS1-PTP1B^−/−. Cells from the hippocampal region from 2 mice of the same genotype were pooled, yielding 4 samples for single-cell RNA sequencing. (Fig. 3A). Only female mice were used to minimize sex-related variability and to reflect the higher prevalence of AD patients in females.

Figure 3. PTP1B is highly expressed in brain immune cells and limits microglial reactivity in APP/PS1 mice.

(A) Schematic of the scRNA-seq experimental design on 13-month-old female mice. (B) Ptpn1 expression across all annotated cell types in APP/PS1 mice. (C) UMAP plots of microglia subclusters split by genotype and colored according to subclusters. (D) Proportion of microglia subclusters from each genotype. (E) Volcano plot showing significant differential expressed genes in red dots (adjusted P < 0.05, log₂FC> 0.3 or log₂FC< −0.3) in microglia of APP/PS1-PTP1B^−/− versus APP/PS1; pseudobulk expression was generated by summing all counts per gene. (F) Gene ontology bar graph generated from genes significantly upregulated (adjusted P < 0.05, basemean>200, log₂FC>0.3) in APP/PS1-PTP1B^−/− mice.

After quality control and filtering, a total of 31740 cells across two genotypes were arranged by uniform manifold approximation and projection (UMAP) for visualization (fig. S4A). Clusters were then manually annotated as astrocytes, choroid plexus cells, endothelial cells, macrophages, microglia, neurons, oligodendrocytes, OPCs and T cells, based on the expression of cell type-specific signature genes (fig. S4B). Then, we investigated the expression level of PTP1B across all cell types in the brain. Interestingly, Ptpn1, the gene that encodes PTP1B, was found to be highly expressed in immune cells, including microglia, macrophages, and T cells (Fig. 3B), suggesting a potential role in brain immune regulation.

As microglia are the primary phagocytes in the brain, we subclustered them for further analysis. This revealed three distinct subclusters: homeostatic microglia (Homeostatic; expressing markers: Fcrls, Tmem119 and P2ry12), disease-associated microglia (DAM; expressing markers: Axl, Ctsl and Trem2) and interferon-responsive microglia (IFN; expressing markers: Oasl2, Stat1 and Irf7) (Fig. 3C and fig. S4C). The proportion of DAM microglia showed a modest increase in APP/PS1-PTP1B^−/− mice (51.7%) compared to the controls APP/PS1 mice (44.9%) (Fig.3D), suggesting a possible shift towards a more enhanced microglial activation in the absence of PTP1B.

To examine further how PTP1B affects the microglial response to Aβ pathology, we analyzed the differentially expressed genes in the microglia. Genes upregulated in the absence of PTP1B include those encoding major histocompatibility complex (MHC) proteins (e.g. H2-D1, Cd74 and B2m), interferon-response proteins (e.g. Stat1, Ifitm3 and Oasl2), members of the cathepsin family of cysteine proteases (e.g. Ctsh, Ctsl and Ctsc) and lysosomal proteins (e.g. Hexa, Hexb and Arsb). Additionally, several genes associated with DAM signatures were upregulated, including Axl, Cd9, Trem2, Csf1 and Spp1. Some microglial homeostatic genes, such as Serincs and Cx3cr1, were downregulated in PTP1B-deficient microglia (Fig. 3E). These findings revealed that PTP1B deletion in APP/PS1 mice drives a transcriptional shift towards a more active and phagocytic state. Indeed, gene ontology (GO) pathway analysis showed enrichment for processes involving the “proteosome”, “lysosome” and “microglia pathogen phagocytosis pathway” (Fig. 3F). Interestingly, PTP1B-deficient microglia exhibited an enrichment of genes related to the oxidative phosphorylation (OXPHOS) pathway. This metabolic enhancement likely supports the increased activation and phagocytosis. Together, these data suggest that microglia could be more activated and phagocytic in APP/PS1 mice upon the deletion of PTP1B.

PTP1B deficiency enhances microglial phagocytic activity in response to Aβ

Considering the reduction of Aβ levels (Fig. 2) and microglial transcriptional changes (Fig. 3) observed in APP/PS1 mice lacking PTP1B, we hypothesized that PTP1B deletion would increase microglial phagocytic activity. To test this, an assay measuring uptake of fluorescent beads was performed on primary microglia isolated from WT and PTP1B-knockout pups. Phagocytic activity was quantitated by flow cytometry as the percentage of cells that internalized the fluorescent beads. In WT microglia, treatment with Aβ oligomers (AβOs) increased phagocytic activity compared to vehicle-treated control. Interestingly, PTP1B-deficient microglia exhibited a higher baseline phagocytic activity, and this effect was further enhanced upon stimulation with AβOs (Fig. 4A and fig. S5A). These data suggest that PTP1B deletion enhances microglial phagocytic capacity in vitro, supporting a role of PTP1B as a negative regulator of microglial phagocytosis.

Figure 4. PTP1B deletion enhanced Aβ engulfment in APP/PS1 mice and promoted phagocytosis in vitro.

(A) Quantitation of WT and PTP1B^−/− primary microglia, treated with vehicle or AβOs, and measurement of phagocytic activity by fluorescence-activated cell sorting (FACS)-based microparticle-uptake assay (n = 9/group); repeated-measures one-way ANOVA with Tukey’s post hoc test; representative micrographs are shown in Supplementary Figure S5A. (B) Representative images of microglia (IBA1, pseudo green), Aβ plaques (6E10, pseudo grey) and lysosomal marker (CD68, pseudo red) co-stained in APP/PS1 mice with or without PTP1B; nuclei are stained with DAPI (blue); Aβ engulfed by IBA1⁺ microglia and by CD68⁺ microglia were shown in grey, scale bar=10μm. (C) Quantitation of the percent volume of Aβ engulfed by IBA1⁺ microglia, averaged across multiple images per mouse; each dot represents one mouse (n=7 mice/group); unpaired two-tailed t-tests. (D) Quantitation of the percent volume of Aβ engulfed by CD68⁺ microglia, averaged across multiple images per mouse; each dot represents one mouse (n=7 mice/group); unpaired two-tailed t-tests.

Data represent mean ± SEM; p > 0.05 (ns), p < 0.05 (*), p < 0.01 (**), p < 0.001 (***), p < 0.0001 (****).

To assess whether this effect was recapitulated in vivo, we performed immunohistochemistry and quantitative confocal imaging on brain slices from APP/PS1 mice, in the presence or absence of PTP1B (Fig. 4B and fig. S5B). We quantitated Aβ engulfment by calculating the volume of Aβ signal colocalized with IBA1-positive microglia, normalized to the total Aβ volume. Notably, PTP1B-deficient microglia exhibited a significant increase in the percentage of intracellular Aβ compared to controls (Fig. 4B and 4C). Additionally, immunofluorescence staining of CD68, a lysosomal marker of phagocytic microglia, revealed that Aβ internalization within CD68+ phagolysosomes was more than doubled in the absence of PTP1B (Fig. 4B and 4D). These results suggest that enhanced Aβ plaque engulfment by PTP1B-deficient microglia may contribute to the reduced Aβ burden observed in APP/PS1 mice lacking PTP1B.

PTP1B deletion enhanced AβO-induced PI3K-AKT-mTOR signaling and energy metabolism in microglia

The activation of myeloid cells, such as microglia, is regulated by the PI3K-AKT-mTOR signaling pathway, which plays a critical role in cellular metabolism (38, 52). Upon activation, this pathway promotes the phosphorylation of AKT and mTOR, leading to the upregulation of HIF1α, a transcriptional regulator of glycolysis. Enhanced glycolysis enables microglia to generate ATP rapidly to meet the immediate energy demands associated with immune functions such as phagocytosis. Although glycolysis provides a fast and flexible energy source, OXPHOS remains a more efficient pathway of sustaining longer-term energy production. Both pathways can contribute to the metabolic demands of activated microglial function (37-39, 41).

To investigate the role of PTP1B in this metabolic regulation, we examined signaling changes in primary microglia with or without PTP1B following exposure to AβOs. In WT microglia, stimulation with AβOs increased phosphorylation of AKT and mTOR and elevated HIF1α expression. Interestingly, PTP1B-knockout microglia displayed further enhancement of the PI3K-AKT-mTOR pathway, as demonstrated by increased AKT and mTOR phosphorylation, and increased HIF1α (Fig. 5A and fig. S6A), suggesting that PTP1B deletion amplifies microglial signaling responses to AβOs.

Figure 5. PTP1B deletion enhanced AβOs-induced PI3K-AKT-mTOR signaling and energy metabolism in microglia.

(A) Representative immunoblot analysis of AKT, p-AKT, mTOR, p-mTOR, HIF1α, and Actin in WT or PTP1B^−/− primary microglia treated with vehicle or AβOs; quantitation from independent biological replicates (n = 5-6) are shown in Supplementary Figure S6A. (B) Glycolysis measured by lactate secretion (n = 4 per group); repeated-measures one-way ANOVA with Tukey’s post hoc test. (C-D) WT and PTP1B^−/− primary microglia were treated with and without AβOs for 24h. After treatment, real time oxygen consumption rate (OCR) was measured following sequential addition of oligomycin, FCCP, rotenone and antimycin A as illustrated (n = 3 independent experiments); (C) OCR curve; (D) quantitation of basal respiration (repeated-measures one-way ANOVA with Bonferroni's post hoc test).

Data represent mean ± SEM; p < 0.05 (*), p < 0.01 (**).

To assess downstream metabolic effects, we measured both glycolytic output and mitochondrial respiration of WT and PTP1B-deficient microglia. Glycolytic activity was first evaluated by measuring lactate secretion into the culture medium, as lactate is a by-product of glycolysis. Stimulation with AβOs increased lactate production in WT microglia, whereas in PTP1B-deficient microglia lactate production was further increased, suggesting elevated glycolytic metabolism in the absence of PTP1B (Fig. 5B). Consistently, Seahorse analysis revealed a higher extracellular acidification rate (ECAR) in PTP1B-deficient microglia in response to AβOs (fig. S6B), indicative of increased glycolytic flux. In the same assay, measurement of the oxygen consumption rate (OCR) illustrated that PTP1B deletion also led to an increase in mitochondrial respiration, as indicated by elevated basal respiration under both vehicle and AβO-treatment conditions (Fig. 5C and 5D). Additionally, PTP1B-deficient microglia exhibited enhanced maximal respiration upon stimulation with AβOs (Fig. 5C and fig. S6C), indicating a greater respiration capacity. Similarly, mitochondrial-linked ATP production was higher in PTP1B-deficient microglia under both vehicle and AβO-treatment conditions (Fig. 5C and fig. S6C). Together, these findings suggest that PTP1B deletion in microglia enhances both mitochondrial and glycolytic metabolism, promoting a metabolically active state that may support enhanced microglial function such as phagocytosis.

PTP1B plays a critical role in regulating microglia activation upon AβOs stimulation via SYK-dependent pathway

Since we demonstrated that PTP1B deletion enhanced phagocytic activity (Fig. 4) and energy metabolism (Fig. 5) in microglia in response to AβOs, we examined upstream signaling pathways to identify mediators of these responses. We focused on SYK, a protein tyrosine kinase previously reported to be a central node in the microglial responses to Aβ pathology (42-44). In addition, SYK was reported to act upstream of the PI3K-AKT-mTOR signaling pathway, playing a key role maintaining microglial energy metabolism (42-44).

Interestingly, following treatment with AβOs, PTP1B^−/− microglia exhibited increased SYK phosphorylation compared to WT (Fig. 6A, fig. S7A and S7B), suggesting PTP1B acts as a negative regulator of SYK. To determine whether SYK signaling contributes to the role of PTP1B in microglial activation, we pretreated both WT and PTP1B^−/− microglia with the SYK inhibitor BAY61-3606 at varying concentrations. Treatment with BAY61-3606 effectively reduced SYK tyrosine phosphorylation in both WT and PTP1B^−/− microglia, indicating successful inhibition of SYK activity (Fig. 6A). Functionally, the microparticle uptake assay showed that treatment with BAY61-3606 diminished the enhanced phagocytic activity of PTP1B^−/− microglia, reducing it to the levels seen in treated WT microglia (Fig. 6B and fig. S7C). Immunoblot analysis further demonstrated that BAY61-3606-treated microglia displayed reduced AKT-mTOR signaling in both groups (Fig. 6A and fig. S7B). Consistently, treatment with SYK inhibitor led to a reduction in lactate production (Fig. 6C) and basal oxygen consumption rate (Fig. 6D) to levels similar to both WT and PTP1B-deficient microglia. Taken together, these data demonstrate that PTP1B can regulate SYK function, which in turn plays an important regulatory role in controlling microglial responses to AβOs, including phagocytosis, signaling and energy metabolism.

Figure 6. PTP1B plays a critical role in regulating microglia activation upon AβOs stimulation via SYK-dependent pathway.

(A) Representative immunoblot analysis of SYK, p-SYK, AKT, p-AKT, mTOR, p-mTOR, HIF1α, and Actin in WT or PTP1B^−/− microglia treated with AβOs or varying concentrations of SYK inhibitor BAY-61-3606, quantitations from independent biological replicates (n = 3-6/group) are shown in Supplementary Figure S7B. (B) Quantitation of microglia and measurement of phagocytic activity by fluorescence-activated cell sorting (FACS)-based microparticle-uptake assay (n = 7/group); repeated-measures one-way ANOVA with Tukey’s post hoc test; representative micrographs are shown in Supplementary Figure S7C. (C) Glycolysis measured by lactate secretion (n = 4/group); repeated-measures one-way ANOVA with Tukey’s post hoc test. (D) OCR measurements of primary cultured microglia with or without the presence of BAY61-3606; repeated-measures one-way ANOVA with Bonferroni's post hoc test.

Data represent mean ± SEM; p > 0.05 (ns), p < 0.05 (*), p < 0.01 (**), p < 0.001 (***).

SYK is a direct PTP1B substrate

A previous study in B cells suggested that SYK signaling may be regulated by PTP1B, as phosphorylation at Y525/Y526 was increased in the absence of PTP1B (45). However, it remains unclear whether PTP1B regulates SYK activation directly by targeting phosphotyrosine residues on SYK or indirectly through other signaling molecules. Our earlier structural studies identified a sequence motif —D/E–pY–pY–R/K— that is critical for optimal recognition of substrates by PTP1B (53). Notably, SYK’s activation loop features a similar motif: D-E-N-pY-pY-K. With this in mind, we hypothesized that SYK may be a direct substrate of PTP1B.

Upon overexpression in HEK293T cells (HEK293T-SYK), we observed SYK autophosphorylation (fig. S8B and S8C). Consistent with our findings in microglia, SYK phosphorylation was increased in PTP1B-KO 293T cells compared to WT controls (fig. S8B). In examining whether PTP1B directly affected kinase activation, we observed a dose-response relationship with increasing levels of WT PTP1B ectopically expressed in PTP1B-deficient HEK 293T cells (fig. S8C and S8D), coinciding with decreased phosphorylation of Y525/526 from the activation loop of SYK. Notably, expression of the catalytically inactive form (C215S) of PTP1B did not alter the phosphorylation of SYK, indicating that the phosphatase activity of PTP1B is required for this regulatory function (Fig. S8C and S8E).

Additionally, we took advantage of the strategy involving substrate-trapping mutant forms of the phosphatase (54) to investigate whether SYK is a direct substrate of PTP1B. Unlike WT PTP1B, which dephosphorylates and releases its substrate, the PTP1B-D181A mutant forms a stable complex with its substrate, allowing for substrate identification and characterization (54, 55). The substrate-trapping mutant, but not the WT form of the phosphatase, formed a stable complex with SYK (Fig.7A and Fig. S8A, lane 1 and 2), supporting a direct enzyme-substrate interaction. To identify the critical tyrosine residues mediating the PTP1B-SYK interaction, we expressed WT and mutant (Y525F, Y526F, YY525/Y526FF, Y352F) forms of SYK together with the PTP1B-D181A mutant in HEK293T cells. Mutation of the activation loop residues Y525 and Y526 in SYK to phenylalanine disrupted this interaction (Fig.7A and Fig. S8A, lanes 3-5), whereas mutation at Y352 had no observable effect (Fig.7A and Fig. S8A, lane 6). Taken together, these findings demonstrate a direct interaction between PTP1B and SYK at the critical autophosphorylation sites, Y525/Y526.

Figure 7. SYK is a direct substrate of PTP1B.

(A) Representative immunoblot of HA-SYK co-immunoprecipitated with FLAG-PTP1B in CRISPR-generated PTP1B-KO HEK293T cells transiently transfected with FLAG-PTP1B (WT or D181A mutant) and HA-SYK (WT, Y525F, Y526F, YY525/526FF, or Y352F mutant) constructs, alone or in combination as indicated. Whole-cell lysates were probed for HA and FLAG; quantitation from independent biological replicates (n = 4) is shown in Supplementary Figure S8A. (B) Representative Immunoblot and quantitation of SYK and PTP1B substrate trapping assay in primary microglia stimulated with AβOs; Lysates were incubated with purified His-tagged PTP1B (WT or D181A mutant, residue 1-321 truncated form comprising the catalytic domain) in the presence or absence of pervanadate prior to immunoprecipitation; repeated-measures one-way ANOVA followed by Dunnet post hoc test, comparing each group to the control (lane 2); n=3. (C) Proposed model: In microglia, PTP1B directly dephosphorylate SYK, suppressing the downstream AKT-mTOR signaling; deletion of PTP1B prevents SYK dephosphorylation, leading to enhanced AKT-mTOR signaling, which promotes increased activation, phagocytosis, mitochondrial and glycolytic metabolism in response to Aβ. In APP/PS1 mice, PTP1B deletion facilitates Aβ clearance and improve cognitive function.

Data represent mean ± SEM; p < 0.05 (*).

Furthermore, we tested whether PTP1B directly interacts with phosphorylated SYK in primary microglia following treatment with AβOs. Lysates of primary microglia, treated with or without AβOs, were collected and incubated with either WT or D181A mutant PTP1B protein. In the presence of AβOs, a low level of co-precipitation of SYK with WT PTP1B was detected (Fig. 7B, lane 1), whereas this association was more pronounced with the PTP1B-D181A substrate–trapping mutant (Fig.7B, lane 2). Furthermore, pretreatment of the PTP1B substrate trapping mutant with pervanadate, which disrupts the enzyme’s active site, prevented its binding to SYK (Fig.7B, lane 3), indicating that the interaction between PTP1B and SYK requires the PTP1B catalytic site. Therefore, these results suggest that SYK is a direct substrate of PTP1B, dephosphorylation of which represents a potential mechanism underlying the effects of PTP1B deletion or inhibition in AD mouse model (Fig 7C). Together, these findings established that PTP1B as a critical regulator in microglial function and provides new insights into its potential as a therapeutic strategy for AD.

Discussion

Accumulating evidence suggests that metabolic alterations in the brain, such as insulin resistance and impaired glucose metabolism, are early features of AD (11, 56-61). PTP1B is a widely expressed phosphatase well known for its role in metabolic regulation, particularly through its regulation of insulin and leptin signaling, both of which have been associated with neuroprotective functions (62). These connections led us to hypothesize that targeting PTP1B may restore the disrupted brain energy metabolism and improve cognitive outcome in AD.

PTP1B modulates diverse signaling pathways across several tissues – including insulin signaling in metabolic organs, HER2 signaling in breast cancer, and JAK/STAT cytokine signaling in immune cells. As a result, PTP1B has been explored as a therapeutic target in a range of diseases and contexts, including diabetes (18-20), breast cancer (63) and immunotherapy involving T cell checkpoint blockade (31). Although one study reported that lifetime deletion of PTP1B in myeloid lineage cells predisposed aged mice to leukemia (64), the heterozygous animals appeared normal, suggesting that partial inhibition with a PTP1B inhibitor could provide therapeutic efficacy. In the central nervous system (CNS), most studies of PTP1B have focused its role on neurons, where it negatively regulates key signaling pathways involved in neuronal survival, synaptic function, and metabolism (25, 65-67). For example, PTP1B dephosphorylates the TRKB receptor in neurons, attenuating BDNF signaling (25). In the hAPP-J20 AD model, neuron-specific PTP1B deletion enhanced the phosphorylation of synaptic proteins such as NSF and NMDAR GluN2B, leading to improved synaptic plasticity (66). Thus, neuronal PTP1B has been proposed as a potential therapeutic target for neurological diseases.

Although prior studies have primarily focused on the role of PTP1B in neurons, our RNA sequencing data showed higher mRNA levels of PTP1B in microglia compared to neurons (Fig. 3B). However, its function in microglia, especially in the context of AD, remains largely unknown. Our findings reveal a novel role for PTP1B in microglia, highlighting a broad involvement in the inflammatory response within CNS. Specifically, PTP1B deletion in APP/PS1 mice reduced brain amyloid levels. This was accompanied by enhanced Aβ engulfment by microglia, suggesting that increased phagocytic activity may contribute to the observed reduction of Aβ burden. Consistent with this, single-cell RNA sequencing revealed that PTP1B is highly expressed in microglia, and its deletion in APP/PS1 mice induces a transcriptional shift toward a more metabolically active and phagocytic microglial state. These findings extend the role of PTP1B beyond neurons, highlighting its influence on microglial signaling pathways.

In AD, microglia change from homeostatic to a disease-associated phenotype, often referred to disease-associated microglia (DAM) (68), which is characterized by the upregulation of genes involved in Aβ recognition and clearance. Accumulating evidence suggests that promoting expression of DAM genes can enhance microglial function and may represent a promising therapeutic strategy for halting AD progression. For example, transcriptomic studies in AD models and patients treated with anti-amyloid antibodies have shown increased expression of DAM genes, suggesting that DAM activation may be a key feature of effective Aβ-targeting therapies (35, 36). Similarly, activating the microglial receptor TREM2, a critical regulator of DAM progression (68), using agonist antibodies, has been reported to enhance microglial responses to Aβ, leading to increased microglial proliferation and metabolism (41-43, 69-74). Additionally, modulating signaling molecules downstream of TREM2, such as SYK (42-44) and mTOR (40), has been shown to attenuate Aβ accumulation and promote phagocytosis, suggesting that targeting these pathways holds promise for developing novel AD therapies. Our data are consistent with these findings, showing that PTP1B deletion induces a DAM-like phenotype, enhancing Aβ clearance.

Mechanistically, we show that PTP1B directly regulates the autophosphorylation sites of SYK within its kinase domain, acting as a brake on SYK function. The absence of PTP1B leads to enhanced SYK activation, which is critical for microglial activation. Consistent with this, inhibition SYK using BAY61-3606 led to a reduction in phagocytosis activity, regardless of PTP1B presence. Consistent with SYK signaling acting upstream of PI3K-AKT-mTOR signaling in microglia (42, 43), treatment with BAY61-3606 also attenuated AKT-mTOR phosphorylation, irrespective of PTP1B presence, underscoring SYK as a key regulatory node. Together, we demonstrated that regulation of SYK by PTP1B plays a critical role in regulating microglial activation upon Aβ stimulation.

The PI3K-AKT-mTOR pathway is a key regulator of energy balance and metabolism. Microglia rely on glycolysis and OXPHOS to fulfill their energy demands and dynamically adjust these metabolic pathways upon activation (38). Notably, microglia lacking critical activation molecules such as TREM2 (75) or SYK (43) have impaired metabolism, reducing both glycolysis and OXPHOS, together with reduced PI3K-AKT-mTOR signaling. Interestingly, we found that PTP1B-deficient microglia increased activation of the PI3K-AKT-mTOR pathway, exhibited enhanced glycolysis and OXPHOS, particularly upon Aβ stimulation. Transcriptomic analysis further supports an increase in oxidative metabolism, with upregulation of OXPHOS-related genes in PTP1B-deficient microglia. Collectively, these results position PTP1B as a key regulator of microglial metabolism, offering novel insight into how metabolic pathways intersect with immune activation in AD. Furthermore, previous studies demonstrated that the myeloid-specific deletion of PTP1B significantly improved glucose homeostasis in insulin resistance models (76, 77), suggesting that PTP1B may also modulate glucose uptake and metabolism in microglia. Targeting PTP1B could potentially enhance microglial metabolic capacity and improve glucose availability in the brain, thereby benefiting the AD patients that display brain glucose hypometabolism (78).

In our study, global PTP1B deletion and systemic administration of PTP1B inhibitor DPM-1003 improved cognitive performance and reduced amyloid burden. Although we have revealed a new perspective on the function of PTP1B in microglia during Aβ pathology, which contributes to the reduced amyloid levels, we cannot exclude the potential importance of neuronal PTP1B in this context. PTP1B inhibition enhances microglial phagocytic activity and potentially improves neuronal and synaptic function, which could synergistically contribute to the behavioral rescue observed here. This dual action may provide a distinct therapeutic advantage of targeting PTP1B in AD. Going forward, it will be important to generate animals in which PTP1B is specifically ablated in microglia and neurons, and assess directly the cell-type specific contributions to the overall disease phenotype.

In summary, we have now established a critical signaling function of PTP1B in microglia, in addition to its reported effects in neurons (65-67). We have identified PTP1B as a central regulator of microglial activation, metabolism, and Aβ clearance. By directly modulating SYK signaling and downstream PI3K-AKT-mTOR axis, PTP1B deletion enhances microglial phagocytosis and promotes a more metabolically active, DAM-like phenotype. These findings provide mechanistic insight into how targeting PTP1B could enhance innate immune responses and improve microglial energy metabolism in AD. Our work demonstrates that targeting PTP1B could represent a promising strategy for mitigating cognitive decline in AD patients.

Materials and Methods

All animal procedures were approved by the Institutional Animal Care and Use Committee of Cold Spring Harbor Laboratory. Full experimental details are provided in SI Appendix, Materials and Methods. Briefly, both genetic deletion of PTP1B and pharmacological inhibition using DPM-1003 on WT or APP/PS1 mice were employed to assess the impact of PTP1B loss on cognitive outcomes and Aβ pathology. Behavioral testing, brain tissue processing, histological staining, and biochemical quantification of Aβ species were performed accordingly. Single-cell RNA-seq was performed on hippocampal tissues of female APP/PS1 mice to define PTP1B expression profile in different cell types and transcriptional alterations associated with PTP1B loss. Primary microglia were isolated from pups for AβO stimulation, followed by phagocytosis assays, immunoblotting, Seahorse metabolic analysis, and substrate-trapping experiments using purified PTP1B (WT or DA mutant). List of antibodies used and detailed statistical analyses conducted in this study are as described in SI Appendix: Table 1 and Materials and Methods.

Significance Statement.

In Alzheimer’s disease, toxic amyloid-beta accumulation is believed to play a central role in the disease progression. We found that deleting or inhibiting a protein tyrosine phosphatase enzyme, PTP1B, improved cognitive behaviors and reduced Aβ burden in an Alzheimer’s disease mouse model. Mechanistically, we show that PTP1B deletion enhanced SYK-dependent signaling in microglia, the brain innate immune cell, boosting their phagocytic activity and mitochondrial and glycolytic metabolism. These findings highlight PTP1B inhibition as a potential therapeutic strategy to promote amyloid clearance and improve cognitive function in Alzheimer’s disease.

Data availability

All data supporting the findings of this study are available within the article and its supplementary materials. Single-cell RNA sequencing data were deposited in the Gene Expression Omnibus (GSE315736).