Myeloid gasdermin D drives early-stage T cell immunity and peripheral inflammation in a mouse model of Alzheimer’s disease

Abstract

Background

It is now realized that peripheral inflammation and abnormal immune responses, especially T cells, contribute to the development of Alzheimer’s disease (AD). Gasdermin D (GSDMD) -mediated pyroptosis has been associated with several neuroinflammatory diseases, but whether GSDMD is involved in the peripheral inflammation and T cell immunity during AD remains unclear.

Methods

We dynamically investigated GSDMD activation in the peripheral and central nervous system of 5×FAD mouse model and dissected the role of myeloid GSDMD using genetic knockout mice, especially its influence on peripheral T cell responses and AD inflammation. RNA sequencing and in vitro coculture were used to elucidate the underlying immune mechanisms involved. Targeted inhibitor experiments and clinical correlation analysis were used to further verify the function of GSDMD in AD.

Results

In the present study, caspase activated GSDMD in the spleen of 5×FAD mice earlier than in the brain during disease progression. Loss of myeloid cell GSDMD was shown to impair early-stage effector T cell activation in the periphery and prevent T cell infiltration into the brain, with an overall reduction in neuroinflammation. Furthermore, myeloid cell GSDMD induced T cell PD-1 expression through the IL-1β/NF-κB pathway, restricting regulatory T cells. The administration of a GSDMD inhibitor combined with an anti-PD-1 antibody was found to mitigate the development of AD-associated inflammation. In some AD patients, plasma sPD-1 is positively correlated with IL-Iβ and clinical features.

Conclusions

Our study systematically identified a role for GSDMD in the AD-related peripheral inflammation and early-stage T cell immunity. These findings also suggest the therapeutic potential of targeting GSDMD for the early intervention in AD.

Background

Alzheimer’s disease (AD) is a neurodegenerative disease typically characterized by progressive dementia and a neuropathology that includes deposits of amyloid-beta (Aβ) plaques and neurofibrillary tangles [1]. Dysfunction in innate immunity and adaptive immunity contributes to the development of AD [2].

Pyroptosis is an important innate immune inflammatory response that mainly activated by inflammasomes which consist of pattern recognition receptors, apoptosis-associated speck-like protein, and inflammatory caspases (caspase-1/-11) [3]. Upon stimulation by pathogen invasion or tissue damage, oligomerized inflammasomes induce caspase auto-cleavage, which activates gasdermin D (GSDMD). The N-terminus of GSDMD forms transmembrane pores, causing cell rupture and the release of cell contents, especially IL-1β and IL-18 [4]. Inflammasome independent pathways (e.g., caspase-8) are also reported to activate GSDMD; thus, GSDMD is a key executive molecule that controls pyroptosis [5, 6]. GSDMD plays a critical role in several neurological disorders including Parkinson’s disease, ischemic stroke, and depression [7,8,9]. Although recent studies have described the activation of brain GSDMD during AD, the immune function of GSDMD in AD is unclear [10, 11].

In addition, increasing evidence suggests that neuroinflammation in AD is affected by T cells. Infiltrating T cells are found in the cerebrospinal fluid (CSF), leptomeninges, and parenchyma of AD patients, as well as in the brain of AD model mice [12,13,14]. CD8⁺ T cells have been shown to expand clonally in the peripheral blood mononuclear cells (PBMCs) and CSF of AD patients [13, 15]. These cells can be cytotoxic for neurons once activated and interact with microglia, promoting the production of various inflammatory signals [13, 16]. Mice lacking T cells exhibit less inflammation and are resistant to the development of AD [17]. Central nervous system (CNS) antigens, including Aβ, presented by antigen presenting cells (APCs) trigger AD T cell activation [18, 19]. GSDMD in innate immune myeloid cells has been reported to regulate T cell adaptive immune responses in several diseases, including multiple sclerosis, orchitis, and cancer [20,21,22]. Our previous study demonstrated GSDMD activation in the PBMCs of AD patients [23]. However, little is known about the process and mechanism by which GSDMD regulates T cell immunity in AD.

Herein, we found that GSDMD activation occurs much earlier in the periphery than in the CNS. We systematically investigated the role of myeloid cell GSDMD in AD inflammatory responses using transgenic knockout mice. GSDMD was found to promote effector T cell activation by enhancing antigen presentation and suppressing regulatory T (T-reg) cells by inducing PD-1 expression during the early stage of AD. In addition, combinatorial treatment with a GSDMD inhibitor and PD-1 blockade reduced the excessive activation of T cells and efficiently restricted inflammation, thereby mitigating the development of AD.

Materials and methods

Animal experiments

Female mice with the C57BL/6 background were used in this study. 5×FAD (Tg6799) mice were kindly provided by Dr. Ming Xiao (Nanjing Medical University, Jiangsu, China). Gsdmd^−/− mice and Cx3cr1-Cre mice were gifts from Dr. Shuo Yang (Nanjing Medical University, Jiangsu, China). We crossed 5×FAD mice with Gsdmd^−/− mice to generate Gsdmd^−/−5×FAD mice. To obtain myeloid cell conditional knockout mice (Gsdmd^fl/flCx3cr1-Cre), Gsdmd^fl/fl mice were generated via conditional gene targeting methods as described previously [22]. These mice were crossed with Cx3cr1-Cre mice. Then, Gsdmd^fl/flCx3cr1-Cre mice were crossed with 5×FAD mice to generate myeloid cell-conditional GSDMD-depleted AD mice (Gsdmd^fl/flCx3cr1-Cre;5×FAD). All animal experiments were conducted according to the National Institutes of Health Guidelines for the Care and Use of Laboratory Animals, and all animal procedures were approved by the Ethical Review Committee for Laboratory Animal Welfare of Tongji University School of Medicine. Mice were bred in a barrier facility with a 12-h light/dark cycle at a temperature of 18–23 °C and a humidity of 40–60%.

Behavior and short-term memory analysis

Morris water maze (MWM)

A circular pool with a diameter of 150 cm and a height of 60 cm was filled with water to a depth of 40 cm. Food grade titanium dioxide powder (nontoxic) was added to the water for opacity. The pool was divided into four quadrants. The white escape platform was located in the center of one quadrant and submerged 1 cm below the water surface. Animals were subjected to four training trials each day for a total of 5 days as the learning phase. Each animal was given 60 s to locate the hidden platform; if the animal could not find the platform within 60 s, they were placed on the platform for 10 s. After 5 days of training, the hidden platform in the target quadrant was removed. On the sixth day, the mice were allowed to swim in the pool for 60 s, and the number of crossings over the previous position of the platform and the time spent in the target quadrant were recorded.

Y maze test (Y-maze)

The Y-maze test was used to assess short-term working memory. The Y-maze consists of three arms (marked as A, B, and C). The task was divided into two stages (training and testing). In the training stage, food was placed in one of the three arms. Then, the mice were placed at the end of the starting arm facing the wall and allowed to explore the maze for 10 min. Twenty-four hours later, the food in the target arm was removed, and the test stage was performed in which the mice were permitted to explore the three arms for 5 min. The number of target arm crossings was recorded, and the ratio of time spent in the target arm compared to the other arms was recorded.

Novel object recognition (NOR) and novel location recognition (NLR) tests

The NOR and NLR test apparatus was a white opaque cube. The experiment was divided into three stages. The first stage was the adaptation period in which the mice were allowed to move freely within the apparatus (without objects) for 10 min. The second stage was the familiarization period with two objects (same color, shape, and size) at the bottom of the apparatus. Mice were placed in the apparatus facing one sidewall and allowed to explore freely for 10 min. Mouse exploration time and the number of times on each object were recorded. The third stage was the testing period, during which memory was tested and evaluated as a timed interval. It took place 1 h after the second stage. For the NOR test, one of the two identical objects was replaced by a different object. For the NLR test, the location of one object was changed. Then, the mice were placed in the apparatus with their backs facing the object from an equal distance and allowed them to explore freely for 5 min. The preference for novel and familiar objects or location was recorded to calculate the novel object/location discrimination index.

Histology and immunohistochemistry

After completing the behavioral tests, mice were anesthetized and sacrificed, brain and spleen tissues were fixed with 4% paraformaldehyde and embedded in paraffin. For immunohistochemistry, and after heat-induced antigen retrieval, the sections were blocked with 3% H₂O₂ for 15 min and 5% goat serum for 30 min at room temperature. The sections were then incubated with primary antibody overnight at 4 °C and treated with horseradish peroxidase (HRP)-labeled secondary antibody followed by diaminobenzidine (DAB) staining. For immunofluorescence, the sections were permeabilized in PBS supplemented with 0.25% Triton X-100 (PBST) for 15 min, blocked, and then incubated with primary antibodies diluted in 0.1% PBST overnight at 4 °C. The next day, the sections were incubated with the corresponding fluorescence-labeled secondary antibodies and treated with 4′,6-diamidino-2-phenylindole (DAPI) or thioflavin S (Sigma-Aldrich, St. Louis, MO, USA). Primary antibodies were specific for: β-Amyloid (4G8; 1:500; SIG-39200; Biolegend, San Diego, CA, USA), GSDMD (1:500; ab219800; Abcam, Cambridge, MA, USA), caspase-1 (1:200; AG-20B-0042; AdipoGen, San Diego, CA, USA), caspase-11 (1:100; NB120-10454; Novus, Littleton, CO, USA), caspase-8 (1:100; sc-5263; Santa Cruz, CA, USA), IBA1 (1:500; 019-19741; Wako, Osaka, Japan), IBA1 (1:500; GB12105; Servicebio, Wuhan, China), GFAP (1:500; G3893; Sigma-Aldrich), NeuN (1:300; MAB377; Merck Millipore, Billerica, MA, USA), CD11b (1:300; MA1-80091; Invitrogen, Carlsbad, CA, USA), CD3 (1:200; GB12014; Servicebio), CD8 (1:200; GB114196; Servicebio), TMEM119 (1:300; ab209064; Abcam), P2RY12 (1:500; 55043; AnaSpec, Fremont, CA, USA), PD-1 (1:1000; 66220-1-Ig; Proteintech, Rosemont, IL, USA), and MHC-II (1:200; 107650; Biolegend). The secondary antibodies used were: Alexa Fluor 488-conjugated anti-rabbit (1:500; 111-545-144; Jackson ImmunoResearch, West Grove, PA, USA), Alexa Fluor 555-conjugated anti-mouse (1:500; GB21301; Servicebio), and Alexa Fluor 555-conjugated anti-rat (1:500; A-21434; Invitrogen) antibodies. Images were acquired with a Leica DM3000 microscope and processed using Image-Pro Plus software.

Western blot analysis

Mouse brains and spleens were quickly collected on ice and homogenized at 1:10 (w/v) in NP-40 lysis buffer (50 mM Tris–HCl, pH 7.4, containing 150 mM NaCl, 1% [vol/vol] Igepal, 10% [wt/vol] glycerol, 50 mM NaF, 1 mM Na₃VO₄, 1 mM dithiothreitol, and 1 mM phenylmethylsulphonyl fluoride, supplemented with protease inhibitor cocktail), and then incubated at 4 °C for 1 h. Equal amounts of protein were mixed, dissolved in 5× SDS/PAGE sample buffer, heated for 10 min at 100 °C, separated on SDS–polyacrylamide gels and blotted onto polyvinylidene difluoride (PVDF) membranes. For clinical samples, approximately 1.5 × 10⁶ PBMCs were combined with 70 µl of sample buffer, processed as above, and PVDF membranes incubated overnight at 4 °C with the following primary antibodies specific for: caspase-1 (1:1000; ab179515; Abcam), GSDMD (1:1000; ab209845; Abcam); Cleaved-caspase-8 (1:1000; 8592; Cell Signaling Technology, Danvers, MA, USA), caspase-11 (1:1000; NB120-10454; Novus), IκB-α (1:1000; 9242; Cell signaling Technology), p-IκB-α (Ser32/Ser36) (1:1000; 9246; Cell signaling Technology), NF-κB (p65) (1:1000; 8242; Cell Signaling Technology), p-NF-κB (p-p65) (1:1000; 3033; Cell Signaling Technology), PD-1 (1:1000; 66220-1-Ig; Proteintech), and β-Actin (1:1000; 4967; Cell Signaling Technology). The next day, after being washed three times in TBST, the membranes were incubated with the following secondary antibodies: HRP-conjugated anti-rabbit (1:1000; 7074; Cell Signaling Technology), HRP-conjugated anti-rat (1:1000; 7077; Cell Signaling Technology), and HRP-conjugated anti-mouse (1:1000; 7076; Cell Signaling Technology). The intensity of target protein bands was analyzed using ImageJ software.

Enzyme‑linked immunosorbent assay

Brain tissue samples and serum were collected from the mice to measure inflammatory factors. The levels of IL-1β, IL-6, and TNF-α were measured with ELISA kits (Abcam). Plasma was collected from AD patients and age-matched subjects to measure the levels of PD-1 (USCN, Wuhan, China) and IL-1β (Jingmei, Yancheng, China). CSF specimens were used to measure the protein levels of AD biomarkers, including Aβ1–40, Aβ1–42, p-tau-181, and t-tau, using commercially available ELISA kits (INNOTEST, Fujirebio, Ghent, Belgium). A microplate reader (Antobio, Zhengzhou, China) was used to obtain the absorbance.

Flow cytometry

Brain and spleen tissues were prepared as single-cell suspensions as previously described [22]. In brief, chopped brain tissues were digested with DNase I (10 U/ml; Roche) and collagenase IV (0.5 mg/ml; Sigma-Aldrich) in RPMI 1640 with agitation (200 rpm) at 37 °C for 60 min. Cell suspensions were filtered through a 70 μm cell strainer and centrifuged through a Percoll density gradient (GE Healthcare, Waukesha, WI, USA). Immune cells were collected from the interface fractions between 37% and 70% Percoll. For splenic tissue, splenocytes were released by mechanical force and red blood cells were lysed with ammonium–chloride–potassium (ACK) lysing buffer. After intensive washing, the cells were labeled with the following antibodies specific for: CD45 (553079), CD8 (553032), CD4 (552051), CD44 (561859), CD19 (551001), PD1 (562671), MHC-II (562363), IFN-γ (560660), and Foxp3 (563101), which were purchased from BD Biosciences. Specific antibodies against CD11b (101212), CD62L (104412), CD11c (117320), F4/80 (123107), MHC-II (107650), and Ki67 (151209) were purchased from Biolegend. An FVD specific antibody (65-0863-14) was purchased from Invitrogen. For intracellular Foxp3 and Ki67 staining, cell samples were fixed and permeabilized using a Transcription Factor Buffer Set (BD Pharmingen, San Jose, CA, USA) according to the manufacturer’s instructions and then stained with specific antibodies. For intracellular cytokine staining, cell samples were stimulated for 4 h with PMA, ionomycin, and brefeldin A, and then subjected to IFN-γ staining. Flow cytometry was performed using a FACS Calibur (BD Biosciences, San Jose, CA, USA), and the data were analyzed with FlowJo 7.6.1 software.

RNA-sequencing analysis

For RNA-sequencing (RNA-seq) analysis, splenocytes were collected. RNA isolation and cDNA library construction were performed with the MGISEQ-2000RS system (Beijing Genomic Institution). Clean reads were mapped to the mouse genome (GRCm38.p5) by HISAT2, and matched reads were normalized by FPKM. Fold changes were calculated for all possible comparisons, and a 1.5-fold cutoff was used to select genes with expression changes. Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis and Gene ontology biological process (GO-BP) analysis were performed using the R package, and target genes were filtered for significant differentially expressed genes (P < 0.05). The gene network was analyzed by the Beijing Genomics Institute in-house customized data mining system. Gene set enrichment analysis (GSEA) was performed as previously described [24] using the Gene Ontology Biological Process Database. The sequencing data files have been deposited in the NCBI Sequence Read Archive (SRA) database under the accession code PRJNA1096339.

Quantitative RT-PCR

Total RNA was extracted with TRIzol reagent (Takara, Tokyo, Japan) and then subjected to cDNA synthesis. Real-time quantitative PCR (RT-qPCR) was performed using a TB Green Premix Ex Taq Kit (Takara). The expression of target genes was calculated by the standard curve method and standardized to the expression of Hprt. The primers used were as follows:

Il1b 5′-TCATTGTGGCTGTGGAGAAG-3′, forward.

5′-AGGCCACAGGTATTT TGTCG-3′, reverse.

Il-6 5′-CTTGGGACTGATGCTGGTGAC-3′, forward.

5′ -GCCATTGCACAACTCTTTTCTC-3′, reverse.

Tnfα 5′-TACTGAACTTCGGGGTGATCG-3′, forward.

5′-TCCTCCACTTGGTGGTTTGC-3′, reverse.

Inos 5′-GGGAATCTTGGAGCGAGTTG-3′, forward.

5′-TGTCCAGGAAGTAGGTGAGGG-3′, reverse.

H2-ob 5′-GACAACAGTAATGCTGGAAATGA-3′, forward.

5′-TGA GCCTTGAGATGGATAACAAC-3′, reverse.

H2-Ab1 5′-AGCCCCATCACTGTGGAGT-3′, forward.

5′ -GATGCCGCTCAACATCTTGC-3′, reverse.

Ccl8 5′-TATCCAGAGGCTGGAGAGCTAC-3′, forward.

5′-TGGAATCCCTGACCCATCTCTC-3′, reverse.

Cxcr4 5′-CTCCTCTTTGTCATCACGCTTCC-3′, forward.

5′-GGATGAGGACACTGCTGTAGAG-3′, reverse.

Ccr1 5′-GTTGGGACCTTGAACCTTGA-3′, forward.

5′-CCCAAAGGCTCTTACAGCAG-3′, reverse.

Pdcd1 5′-ACCCTGGTCATTCACTTGGG-3′, forward.

5′-CATTTGCTCCCTCTGACACTG-3′, reverse.

Il18 5′-CCGCCTCAAACCTTCCAAAT-3′, forward.

5′-TGTGTTTCTTTTCTGGGTGCC-3′, reverse.

Cd74 5′ -ATGGCGACGAGAACGGTAAC-3′, forward.

5′-CGTTGGGGAACACACACCA-3′, reverse.

Nfkb1 5′-ATGGCAGACGATGATCCCTAC-3′, forward.

5′-TGTTGACAGTGGTATTTCTGGTG′, reverse.

Csk 5′-CGACGAGACTCCTGACGTG-3′, forward.

5′-GCACCTCGGGACCCAAATC-3′, reverse.

Ccr6 5′-TCACGACTCGGATTGCTC-3′, forward.

5′-CTGCTGGGTATGGGACTG-3′, reverse.

Hprt 5′-GTCCCAGCGTCGTGATTAGC-3′, forward.

5′-TGGCCTCCCATCTCCTTCA-3′, reverse.

In vitro dendritic cell and microglia antigen presentation assay

As previously described [19], Aβ1–42 peptide (2 mg per mouse) was emulsified with complete Freund’s adjuvant (including 2 mg/ml M. tuberculosis H37Ra, 100 µl per mouse) and used to subcutaneously immunize 5×FAD mice. Fourteen days later, the mice were boosted with Aβ peptide. Seven days later, CD8⁺ T cells were sorted from the splenocytes of immunized mice by flow cytometry. A total of 2 × 10⁵ CD8⁺ T cells were labeled with 5 µM carboxyfluorescein succinimidyl ester (CFSE) and cocultured with WT or Gsdmd^−/− dendritic cells (DCs) (CD11c⁺, 4 × 10⁴) or microglia (CD45^lowCD11b⁺, 1 × 10⁵) in the presence of 10 µM Aβ peptide and IL-2 (10 ng/ml). Anti-PD-1 antibody (10 µg/ml, BE0273, BioXCell, West Lebanon, NH, USA) was added as indicated. CFSE dilution and CD44⁺ percentages were determined by flow cytometry 5 days later. For IL-1β stimulation, 2 × 10⁵ CD8⁺ T cells were cultured with anti-CD3 (1 µg/ml) and anti-CD28 (1 µg/ml) antibodies, and recombinant IL-1β (rIL-1β, 5 µg/ml) was added as indicated.

GSDMD inhibitor treatment and PD-1 blockade in 5×FAD mice

To chronically suppress the cleavage and activation of GSDMD, 5×FAD mice were intraperitoneally injected with dimethyl fumarate (DMF, 5 mg/kg) every 3 days from 2 to 3.5 months of age. To chronically inhibit PD-1 signaling, PD-1 specific antibodies (500 µg per mouse) were intraperitoneally injected into 5×FAD mice every 5 days from 2 to 3.5 months of age [25]. Behavioral feature, histologic, and flow cytometry analyses were performed as described above at 6 months of age.

Cerebrospinal fluid and plasma were collected from AD patients

The criteria of the National Institute on Aging and the Alzheimer’s Association were used for AD diagnosis [26]. We recruited 84 patients with AD from the Department of Neurology, Affiliated Brain Hospital of Nanjing Medical University, from January 2021 to April 2022. For controls, we recruited 33 age-matched subjects. Blood was collected and centrifuged at 3500 rpm for 10 min at 4 °C, and plasma was collected and stored at − 80 °C for subsequent testing. To obtain PBMCs, PBS diluted whole blood (1:1) was transferred to 15 ml centrifuge tubes containing 3 ml of Ficoll Paque (GE Healthcare, Uppsala, Sweden) and centrifuged at 400 × g for 20 min at room temperature. PBMCs were collected and washed twice in 10 ml of PBS. The cells were resuspended in 1× protein loading buffer for subsequent analysis. CSF was collected by lumbar puncture at the L3/L4 or L4/L5 level in the morning. Approximately 2 ml of CSF was collected in a polypropylene tube, centrifuged at 2000 × g for 10 min at room temperature to eliminate cells and other insoluble materials, aliquoted, and stored at − 80 °C for further processing. All participants or their legal guardians provided informed written consent. The Institutional Review Board of the Affiliated Brain Hospital of Nanjing Medical University approved this study.

Statistical analysis

The data were analyzed by GraphPad Prism 7.0 software (Graph Pad Software Inc., San Diego, CA) and are presented as the mean ± standard error of the mean (SEM). The statistics were analyzed by using an unpaired t test for two groups and one-way ANOVA with Dunnett’s multiple comparisons test or two-way ANOVA with Sidak’s multiple comparisons test for multiple groups. P values are indicated as *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001.

Results

At the peak stage, GSDMD is activated in the brains of 5×FAD mice

To dynamically dissect changes in GSDMD during AD progression, we first evaluated CNS pathological features of 5×FAD mice at 2, 4, 6, and 8 months. The deposition of Aβ around the hippocampus and cortex area was mildly increased until 6 months, when the fibrous plaque started to explode (Additional file 1: Fig. S1a, e and Additional file 2: Fig. S2a, c). The activation of microglia and astrocytes occurred earlier than Aβ deposition (Additional file 1: Fig. S1b, c,f, g and Additional file 2: Fig. S2b, d), with no effect on neuron survival (Additional file 1: Fig. S1d, h). More importantly, brain GSDMD expression gradually increased as the disease progressed, as detected by immunohistochemistry at 6 months of age (Fig. 1a, b and Additional file 3: Fig. S3a). Furthermore, with the activation of GSDMD, the cleavage of caspase-1/-11/-8 (which have been reported to induce pyroptosis as upstream signals of GSDMD) were all markedly increased in the brain from 6 months of age (Fig. 1b and Additional file 3: Fig. S3b-i). These results are consistent with public RNA-seq (GEO accession no. GSE1297) data that showing a gradual increase in the transcription level of Gsdmd in the brain tissues of AD patients at different disease stages (Additional file 3: Fig. S3j). To identify GSDMD-positive cells, immunofluorescence colocalization analysis was performed. The results indicated that GSDMD was mainly expressed in activated microglia (amoeboid-like, IBA1⁺TEME119⁺ or IBA1⁺P2RY12⁺) and infiltrated macrophages (ellipse, IBA1⁺TMEM119^- or IBA1⁺P2RY12^-) (Fig. 1c and Additional file 4: Fig. S4a, b) but not in resting microglia (ramified-like, IBA1⁺TMEM119⁺ or IBA1⁺P2RY12⁺), astrocytes, neurons, or T cells (Fig. 1c and Additional file 4: Fig. S4a-c). Analysis of a public database (https://brainrnaseq.org) also revealed that GSDMD is expressed mainly in microglia and macrophages (Additional file 4: Fig. S4d). The levels of proinflammatory cytokines (IL-1β, IL-6, and TNF-α), especially IL-1β, in the brain were also significantly elevated at 6 months, indicating the development of neuroinflammation with GSDMD activation (Fig. 1d-f). These results suggest that GSDMD is activated in the brain at the peak of disease and is also derived from infiltrating macrophages in addition to microglia.

Fig. 1

The dynamic changes of GSDMD in the brains from 5×FAD mice. a Immunohistochemical detection of GSDMD in the hippocampus and cortex of 5×FAD mice at indicated months of age. Scale bar = 50 μm, 20 μm. b Western blot analysis of caspase-1, caspase-11, cleaved caspase-8, and GSDMD in the brain tissues of 5×FAD mice at indicated months of age. F: full length; C: cleavage. c Immunofluorescence colabeling of GSDMD (green) and IBA1 (red) shows that GSDMD was mainly expressed in activated microglia and infiltrated macrophages (white arrowheads) in the brains of 5×FAD mice at 6 months of age. Scale bar = 100 μm, 50 μm. R-MG: resting microglia; A-MG: activated microglia; M∅: macrophage. d-f ELISA analysis of IL-1β (d), IL-6 (e), and TNF-α (f) in the brain tissue of 5×FAD mice at indicated months of age (n = 5/group). Data are pooled from three independent experiments. *P < 0.05, ***P < 0.001, ****P < 0.0001. Error bars show means ± SEM. One-way ANOVA with Dunnett’s multiple comparisons test for d-f

At the early stage, GSDMD is activated in the spleens of 5×FAD mice

Our previous study showed that canonical inflammasome signaling and GSDMD-induced pyroptosis were activated in the PBMCs of amnestic mild cognitive impairment (aMCI) and AD patients [23]. Herein, we assessed pyroptosis-associated protein activation in the spleens of 5×FAD mice during disease progression. The results showed that GSDMD and upstream caspase-1/-11/-8 were significantly activated beginning at 2 months of age (Fig. 2a and Additional file 5: Fig. S5a-g), which was much earlier than their activation in the brain. The splenic size and weight gradually increased with age (Fig. 2b, c). Immunofluorescence analysis demonstrated that splenic GSDMD was localized to myeloid cells (CD11b⁺ cells within red pulp) but not to T cells (CD3⁺ cells within white pulp) in 5×FAD mice (Fig. 2d). The serum levels of IL-1β, IL-6, and TNF-α were also increased earlier than those in the brain (Fig. 2e-g). These results indicate that GSDMD is activated in peripheral myeloid cells during the early stage of AD.

Fig. 2

The dynamic changes of GSDMD in the spleens from 5×FAD mice. a Western blot analysis of caspase-1, caspase-11, cleaved caspase-8, and GSDMD in the spleen tissues of 5×FAD mice at indicated months of age. F: full length; C: cleavage. b-c The size (b) and quantified weight (c) of spleens from 5×FAD mice at indicated months of age (n = 3/group). d Immunofluorescence colocalization of GSDMD (green) with CD11b (red), or with CD3 (red) demonstrates that GSDMD was expressed on myeloid cells in the spleens from 5×FAD mice at 2 months of age. Scale bar = 50 μm, 20 μm. WP: white pulp; RP: red pulp. e-g ELISA analysis of IL-1β (e), IL-6 (f), and TNF-α (g) in the serum of 5×FAD mice at indicated months of age (n = 5/group). Data are pooled from three independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Error bars show means ± SEM. One-way ANOVA with Dunnett’s multiple comparisons test for c, e-g

GSDMD deficiency alleviates behavioral impairment and pathology in 5×FAD mice

To directly assess the role of GSDMD in AD progression, we generated Gsdmd^−/−5×FAD mice (Additional file 5: Fig. S5h) and compared their disease severity with that of 5×FAD mice at 6 months of age. The MWM test was used to evaluate spatial learning memory. Escape latency was significantly lower in Gsdmd^−/−5×FAD mice than in 5×FAD mice during training phase days 4 and 5 (Fig. 3a, b). In the probe phase, Gsdmd^−/−5×FAD mice exhibited significantly more platform crossings and more time spent in the target quadrant than did 5×FAD mice (Fig. 3c, d). The Y maze and NOR/NLR tests (Additional file 5: Fig. S5i, j) were used to assess cognitive memory function. Gsdmd^−/−5×FAD mice had an improved ability to distinguish maze arms (Fig. 3e, f) and discriminate between familiar and novel objects (Fig. 3g), although those animals failed to distinguish a novel location (Fig. 3h), indicating improvement in cognitive memory function. Immunofluorescence revealed a reduction in Aβ deposition (Fig. 3i, k and Additional file 5: Fig. S5k, l), along with decreased microgliosis (Fig. 3j, l and Additional file 5: Fig. S5k, l) and astrocytes activation (Additional file 5: Fig. S5m, n) in Gsdmd^−/−5×FAD mice compared to those in 5×FAD mice. Consistent with these pathological changes, the levels of IL-1β, IL-6, and TNF-α were also decreased in the brains of Gsdmd^−/−5×FAD mice (Fig. 3m-o), indicating less severe neuroinflammation. Furthermore, flow cytometric analysis revealed a marked reduction in infiltrating CD8⁺ T cells (CD45^highCD8⁺) and myeloid cells (CD45^highCD11b⁺) in the brains of Gsdmd^−/−5×FAD mice compared to those in the brain of 5×FAD mice (Fig. 3p-r). The number of brain-infiltrated CD4⁺ T cells (CD45^highCD4⁺) was much lower than that of CD8⁺ T cells; hence, there was no noticeable difference between Gsdmd^−/−5×FAD and 5×FAD mice (Fig. 3p-r). Collectively, these results suggest that GSDMD is essential for the development of AD and the associated neuroinflammation.

Fig. 3

The effects of GSDMD deletion on CNS of 5×FAD mice at the peak stage. a-d MWM test for 5×FAD and Gsdmd^−/−5×FAD mice at 6 months of age. Data are presented as representative swimming traces (a), escape latency in the training phase (b), number of crossing platform (c) and time spend in the target quadrant during the probe test (d) (n = 7/group). e-f Y-maze test for 5×FAD and Gsdmd^−/−5×FAD mice at 6 months of age. Data are presented as the bouts of entering target arm (e) and the percentage of duration in target arm (f) (n = 7/group). g-h The discrimination index of NOR (g) and NLR (h) tests for 5×FAD and Gsdmd^−/−5×FAD mice at 6 months of age (n = 7/group). i-l Immunofluorescence staining and quantification of Aβ plaques (Thioflavin S) (i,k) and IBA1 (j,l) in the hippocampus and cortex of 5×FAD and Gsdmd^−/−5×FAD mice at 6 months of age (n = 5/group). Scale bar = 200 μm. m-o ELISA analysis of IL-1β (m), IL-6 (n), and TNF-α (o) in the brain tissue of 5×FAD and Gsdmd^−/−5×FAD mice at 6 months of age (n = 5/group). p-r Flow cytometry analysis of immune cells (CD45^high), including infiltrating T cells (CD45^highCD8⁺ and CD45^highCD4⁺), myeloid cells and activated microglia (CD45^highCD11b⁺), in the brain tissue of 5×FAD and Gsdmd^−/−5×FAD mice at 6 months of age (n = 5/group). Data are presented as a representative plot (p), quantified percentages (q), and absolute cell numbers (r). Data are pooled from three independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Error bars show means ± SEM. One-way ANOVA for B. Unpaired t test for c-h, k-o. Two-way ANOVA with Sidak’s multiple comparisons test for q, r

GSDMD deficiency reduces early-stage peripheral inflammation and T cell responses in 5×FAD mice

Substantial evidence suggests that T cells, especially CD8 T cells, are involved in AD pathogenesis and neuroinflammation [14,15,16]. Given the above results, we hypothesized that GSDMD is essential for peripheral T cell responses during the early stage of AD. Therefore, we compared peripheral inflammation and immune cell populations between Gsdmd^−/−5×FAD and 5×FAD mice at 2 months of age. The levels of serum IL-1β, IL-6, and TNF-α were reduced in Gsdmd^−/−5×FAD mice (Fig. 4a-c). Spleen swelling was reduced in Gsdmd^−/−5×FAD mice (Fig. 4d, e). Flow cytometric analysis revealed that the percentages and absolute numbers of CD8⁺ and CD4⁺ T cells were both lower in the spleen of Gsdmd^−/−5×FAD mice than in those of 5×FAD mice (Fig. 4f, i,j). The proportion and number of antigen-activated effector T cells (CD44⁺CD62L⁻) among both CD8⁺ and CD4⁺ T cells were also reduced. In contrast, the proportion of naive T cells (CD44⁻CD62L⁺) was increased, although the absolute number of naive T cells did not change due to a decrease in the total number of CD8⁺ and CD4⁺ T cells in Gsdmd^−/−5×FAD mice (Fig. 4g, h,k-n). GSDMD deficiency resulted in the downregulation of IFN-γ⁺ CD8⁺ T cells (Fig. 4o-q). In addition, the number of myeloid cells (CD11b⁺) in Gsdmd^−/−5×FAD mice were reduced, but there was no significant difference in the number of B cells (CD19⁺) (Fig. 4r-w). These results indicate that GSDMD deficiency impairs the activation and differentiation of T cells in peripheral lymphoid tissue and restricts peripheral inflammation during the early stage of disease model.

Fig. 4

The effects of GSDMD deletion on peripheral immune of 5×FAD mice at the early stage. a-c ELISA analysis of IL-1β (a), IL-6 (b), and TNF-α (c) in the serum of 5×FAD and Gsdmd^−/−5×FAD mice at 2 months of age (n = 5/group). d-e The spleen size (d) and weight (e) of 5×FAD and Gsdmd^−/−5×FAD mice at 6 months of age (n = 3 /group). f-n Flow cytometry analysis of CD8⁺ T cells (CD8⁺CD4⁻) and CD4⁺ T cells (CD4⁺CD8⁻) in the spleen and effector T cells (CD44⁺CD62L⁻) and naive T cells (CD44⁻CD62L⁺) from CD8⁺ or CD4⁺ T cells in the spleen of 5×FAD and Gsdmd^−/−5×FAD mice at 2 months of age (n = 5/group). Data are presented as a representative plot (f-h), quantified percentages (i,k,m), and absolute cell numbers (j,l,n). o-q Flow cytometry analysis of IFN-γ⁺ cells from CD8⁺ T cells in the spleens of 5×FAD and Gsdmd^−/−5×FAD mice at 2 months of age (n = 5/group). Data are presented as a representative plot (o), quantified percentages (p), and absolute cell numbers (q). r-w Flow cytometry analysis of myeloid cells (CD11b⁺) and B cells (CD19⁺) in the spleens of 5×FAD and Gsdmd^−/−5×FAD mice at 2 months of age (n = 5/group). Data are presented as a representative plot (r,u), quantified percentages (s,v), and absolute cell numbers (t,w). Data are pooled from three independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Error bars show means ± SEM. Unpaired t test for a-c, e, p-q, s-t, v-w. Two-way ANOVA with Sidak’s multiple comparisons test for i-n

Myeloid GSDMD deficiency alleviates AD pathogenesis and early-stage peripheral T cell responses in 5×FAD mice

The results from the ImmGen database (https://www.immgen.org/) and our previous study [22] revealed that GSDMD was more highly expressed in myeloid cells than in other types of immune cells (Additional file 6: Fig. S6a). Based on the immunofluorescence results described above, we hypothesized that GSDMD derived from myeloid cells contributes to AD inflammation. We generated myeloid cell-conditional knockout mice (Gsdmd^fl/flCx3cr1-Cre;5×FAD) and assessed whether these mice exhibited the same phenotype as Gsdmd^-/-5×FAD mice (Additional file 6: Fig. S6b). The MWM test demonstrated improvement in learning memory function in Gsdmd^fl/flCx3cr1-Cre;5×FAD mice (Fig. 5a-d). Furthermore, the brain deposition of Aβ and activation of microglia were suppressed in Gsdmd^fl/flCx3cr1-Cre;5×FAD mice compared to their counterparts (Cx3cr1-Cre;5×FAD) (Fig. 5e-g). CNS-infiltrated CD8⁺ T cells and myeloid cells were also reduced compared to those in littermate controls at 6 months of age (Fig. 5h-j), with decreased numbers of peripheral CD8⁺ and CD4⁺ T cells at 2 months of age (Fig. 5k and Additional file 5: Fig. 1c, d), which is consistent with fewer differentiated effector cells and more naive cells (Fig. S6l-q). RT-qPCR results confirmed inflammation changes in both the CNS and periphery, compared with those in Cx3cr1-Cre;5×FAD mice, the relative expression of Il1b, Il6, Il18, Tnfα, and Inos decreased (Additional file 6: Fig. S6e, f). These data suggest that myeloid intrinsic GSDMD regulates early-stage T cells activation and inflammation in the periphery, promoting the development of disease model.

Fig. 5

The effects of myeloid GSDMD deletion on the disease progression and T cells response. a-d MWM test for Cx3cr1-Cre;5×FAD and Gsdmd^fl/flCx3cr1-Cre;5×FAD mice at 6 months of age. Data are presented as representative swimming traces (a), escape latency in the training phase (b), number of crossing platform (c) and time spend in the target quadrant during probe test (d) (n = 6/group). e-g Immunofluorescence analysis of Aβ plaques (Thioflavin S) and IBA1 in the hippocampal area of brains from Cx3cr1-Cre;5×FAD and Gsdmd^fl/flCx3cr1-Cre;5×FAD mice at 6 months of age (n = 5 /group). Scale bar = 200 μm. Data are presented as representative images (e) and quantified statistics (f-g). h-j Flow cytometry analysis of immune cells (CD45^high), including infiltrating T cells (CD45^highCD8⁺ and CD45^highCD4⁺), myeloid cells and activated microglia (CD45^highCD11b⁺), in the brain tissue of Cx3cr1-Cre;5×FAD and Gsdmd^fl/flCx3cr1-Cre;5×FAD mice at 6 months of age (n = 5/group). Data are presented as a representative plot (h), quantified percentages (i), and absolute cell numbers (j). k-q Flow cytometry analysis of CD8⁺ T cells (CD8⁺CD4⁻) and CD4⁺ T cells (CD4⁺CD8⁻) in the spleen and effector T cells (CD44⁺CD62L⁻) and naive T cells (CD44⁻CD62L⁺) from CD8⁺ or CD4⁺ T cells in the spleens of Cx3cr1-Cre;5×FAD and Gsdmd^fl/flCx3cr1-Cre;5×FAD mice at 2 months of age (n = 5 /group). Data are presented as a representative plot (k-m), quantified percentages (n,p), and absolute cell numbers (o,q). Data are pooled from three independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Error bars show means ± SEM. One-way ANOVA for b. Unpaired t test for c-d, f-g. Two-way ANOVA with Sidak’s multiple comparisons test for i-j, n-q

Myeloid GSDMD deficiency attenuates T cell response-related genes expression in 5×FAD mice

To assess the mechanistic basis for myeloid cell GSDMD regulation of T cells, we performed bulk RNA-sequencing analysis using splenocytes from 2-month-old Gsdmd^fl/flCx3cr1-Cre;5×FAD and Cx3cr1-Cre;5×FAD mice. GO-BP analysis identified the top pathways that were downregulated in Gsdmd^fl/flCx3cr1-Cre;5×FAD mice, including Adaptive immune response, Antigen processing and presentation of exogenous peptide, Peptide antigen assembly with MHCII complex, and Positive regulation of T cell activation (Fig. 6a). And based on KEGG analysis, downregulated pathways included Antigen processing and presentation, and Th1/Th2 cell differentiation (Fig. 6a). Heatmap and RT-qPCR analyses demonstrated significant reductions in the expression of many genes associated with T cell response, chemokines, and antigen presentation, including Pdcd1, Nfkb1, Csk, Cd74, Cxcr4, Ccr6, Ccr1, Ccl8, H2-Ob, and H2-Ab1, in Gsdmd^fl/flCx3cr1-Cre;5×FAD mice (Fig. 6b, f). GSEA highlighted the enrichment of downregulated genes involved in Adaptive immune response and Positive regulation of immune effector process in Gsdmd^fl/flCx3cr1-Cre;5×FAD mice compared to Cx3cr1-Cre;5×FAD mice (Fig. 6c). Protein-protein interaction (PPI) network analysis identified communications between genes related to the adaptive immune response, antigen processing and presentation, and chemokine signaling pathway (Fig. 6d). Notably, a volcano plot showed several genes with the greatest decreases in Gsdmd^fl/flCx3cr1-Cre;5×FAD mice, including Pdcd1, Cxcl13, Cxcr4, Ccl8, Syk, and Ccr6 (Fig. 6e), with Pdcd1 being the most obvious hub gene for the PPI network (Fig. 6d). Therefore, during the early stage of disease model, impaired expression of T cell response-related genes in peripheral lymphoid tissue is due to myeloid cell GSDMD deficiency. Moreover, for the progression of AD, the relationship between myeloid cell GSDMD and T cell PD-1 is noteworthy.

Fig. 6

T cell immunity related genes were suppressed in myeloid GSDMD knockout 5×FAD mice. a GO-BP and KEGG analyses showing downregulated genes related to the TOP10 signaling pathway in the splenocytes from Cx3cr1-Cre;5×FAD (n = 3) and Gsdmd^fl/flCx3cr1-Cre;5×FAD mice (n = 4) at 2 months of age. b Heatmap of genes with an adjusted P value < 0.05, a false discovery rate < 0.05, and a log2-fold change > 1.5 according to RNA-seq analysis. c GSEA of the genes associated with “Adaptive immune response” and “Positive regulation of immune effector process” in the splenocytes from the mice in a based on the Gene Ontology Biological Process Database. d PPI network analysis showing the connections between hub genes associated with the indicated pathways. Node colors from blue to red indicate the changes in mRNA levels in Gsdmd^fl/flCx3cr1-Cre;5×FAD mice relative to those in Cx3cr1-Cre;5×FAD mice. e Volcano plot showing the results of differential gene expression analysis of splenocytes from the mice in a. f RT-qPCR analysis of the indicated genes in the spleens of 2-month-old Cx3cr1-Cre;5×FAD and Gsdmd^fl/flCx3cr1-Cre;5×FAD mice (n = 3/group). Data were normalized to a reference gene, Hprt. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Error bars show means ± SEM. Two-way ANOVA with Sidak’s multiple comparisons test for f

Myeloid GSDMD deficiency reduces the expression of PD-1 on T cells and MHC-II on myeloid cells

The expression of PD-1 by splenic T cells from 2-month-old Gsdmd^fl/flCx3cr1-Cre;5×FAD and Cx3cr1-Cre;5×FAD mice was assessed by flow cytometry (Additional file 7: Fig. S7a). Compared to those in Cx3cr1-Cre;5×FAD mice, the proportion and number of PD-1⁺ cells among both CD8⁺ and CD4⁺ T cells were decreased in Gsdmd^fl/flCx3cr1-Cre;5×FAD mice (Fig. 7a-c). The percentages of PD-1⁺ cells among CD8⁺ effector T cells (CD8⁺CD44⁺) and CD4⁺ T-reg cells (CD4⁺Foxp3⁺) were also decreased, although the decrease in absolute numbers was slight (Fig. 7a-c). We also evaluated the levels of PD-1 on CNS-infiltrated CD8⁺ T cells at 6 months of age, the proportion and number of PD-1⁺ CD8⁺ T cells were also much lower in Gsdmd^fl/flCx3cr1-Cre;5×FAD mice, with decreased numbers of effector CD8⁺ T cells (Additional file 7: Fig. S7b-f). Immunofluorescence analysis confirmed the presence of PD-1⁺ CD8⁺ T cells in the brains of 5×FAD mice at 6 months of age (Additional file 7: Fig. S7g). There were evidence showed that PD-1 restricts T-reg cell suppressive activity [27, 28], with PD-1 inhibition increasing the number of T-reg cells in AD tauopathy mice [25]. Consistent with the decrease in PD-1⁺ T-reg cells, the percentage and number of Foxp3⁺ T-reg CD4⁺ T cells were increased in the spleens of Gsdmd^fl/flCx3cr1-Cre;5×FAD mice (Fig. 7d-f). These results indicate that the decrease in T cell PD-1 expression is closely related to myeloid cell GSDMD deficiency during disease.

Myeloid cells regulate T cell responses by secreting cytokines and by antigen processing and presentation. The percentage and number of MHC-II positive monocyte-macrophages (CD11b⁺MHC-II⁺) and dendritic cells (CD11c⁺MHC-II⁺) were all reduced, as were the numbers of CD11b⁺F4/80⁺ cells and CD11c⁺ cells in the spleens of Gsdmd^fl/flCx3cr1-Cre;5×FAD mice (Fig. 7g-l). Microglia, as myeloid cells in CNS, also process and present antigens [29]. Immunofluorescence demonstrated the colocalization of MHC-II and IBA1 in 5×FAD mouse brains (Additional file 7: Fig. S7h). Flow cytometric analysis showed a decreased percentage and number of CD11b⁺MHC-II⁺ cells, gated as microglia (CD11b⁺CD45^low), in Gsdmd^fl/flCx3cr1-Cre;5×FAD mice (Fig. 7m-o). These results imply that GSDMD knockout influences antigen processing and presentation by myeloid cells during AD.

Fig. 7

T cell PD-1 and myeloid cell MHC-II were suppressed in myeloid GSDMD knockout 5×FAD mice. a-c Flow cytometry analysis of PD-1⁺ cells from CD8⁺ T cells (CD8⁺CD4⁻), CD4⁺ T cells (CD4⁺CD8⁻), CD8⁺ effector T cells (CD8⁺CD44⁻), and CD4⁺ T-reg cells (CD4⁺Foxp3⁺) in the spleens of Cx3cr1-Cre;5×FAD and Gsdmd^fl/flCx3cr1-Cre;5×FAD mice at 2 months of age (n = 5/group). Data are presented as a representative plot (a), quantified percentages (b), and absolute cell numbers (c). d-f Flow cytometry analysis of T-reg cells (Foxp3⁺) from CD4⁺ T cells in the spleens of Cx3cr1-Cre;5×FAD and Gsdmd^fl/flCx3cr1-Cre;5×FAD mice at 2 months of age (n = 5 /group). Data are presented as a representative plot (d), quantified percentages (e), and absolute cell numbers (f). g-l Flow cytometry analysis of macrophages (CD11b⁺F4/80⁺), DCs (CD11c⁺), and MHCII-expressing cells (CD11b⁺MHCII⁺ and CD11c⁺MHCII⁺) in the spleens of Cx3cr1-Cre;5×FAD and Gsdmd^fl/flCx3cr1-Cre;5×FAD mice at 2 months of age (n = 5/group). Data are presented as a representative plot (g-h), quantified percentages (i,k), and absolute cell numbers (j,l). m-o Flow cytometry analysis of MHCII-expressing cells (MHCII⁺) from infiltrated myeloid cells and microglia (CD45⁺CD11b⁺) in the brains of Cx3cr1-Cre;5×FAD and Gsdmd^fl/flCx3cr1-Cre;5×FAD mice at 6 months of age (n = 5/group). Data are presented as a representative plot (m), quantified percentages (n), and absolute cell numbers (o). Data are pooled from three independent experiments. **P < 0.01, ***P < 0.001, ****P < 0.0001. Error bars show means ± SEM. Two-way ANOVA with Sidak’s multiple comparisons test for b-c, i-l. Unpaired t test for e-f, n-o

In vitro, myeloid GSDMD deficiency influences the proliferation and differentiation of T cells, reducing the effect of PD-1 blockade

To directly address the influence of myeloid GSDMD on the priming and differentiation of antigen-activated T cells, we cocultured CD8⁺ T cells and Gsdmd^-/- or WT DCs in vitro (Fig. 8a). After prepriming in vivo, CD8⁺ T cells proliferation (CFSE⁺) and differentiation (CD44⁺) induced by Aβ peptide were impaired by Gsdmd^-/- DCs compared to WT DCs in vitro (Fig. 8b-d). Moreover, although anti-PD-1 treatment improved the Aβ-induced T cell response when cocultured with WT DCs, no effect was observed when cocultured with Gsdmd^-/- DCs (Fig. 8b-d). Similar results were obtained with cocultured Gsdmd^-/- microglia, indicating that antigen processing and presentation by microglia were also impaired by GSDMD deficiency (Fig. 8e-g). Reduced proliferation of CD8⁺ T cells was confirmed by Ki67 staining of the spleens and brains from Gsdmd^fl/flCx3cr1-Cre;5×FAD mice (Fig. 8h-j). These results emphasize the critical role of APCs GSDMD in promoting CD8⁺ T cell responses in AD, with GSDMD deficiency blocking the effect of ani-PD-1 antibody. IL-1 has been shown to regulate PD-1 expression on tumor-associated macrophages by promoting the recruitment of NF-κBp65 to the Pdcd1 promoter [30]. To determine whether IL-1β directly affects PD-1 expression on T cells, a rIL-1β was added to T cells in vitro. Western-blot analysis showed that the phosphorylation of IκB-α and p65 was enhanced by stimulation with rIL-1β and was accompanied by increased PD-1 expression (Fig. 8k and Additional file 8: Fig. S8a-e).

Fig. 8

CD8⁺ T cell responses were affected by APCs GSDMD deletion and PD-1 inhibition. a Schematic representation of the experiments in b-g. b-d Flow cytometry analysis of CFSE labeled cells, and CD44⁺ cells among CD8⁺ T cells on day 5 after cocultured with DCs (n = 3/group). Data are presented as a representative plot (b) and quantified percentages (c-d). e-g Flow cytometry analysis of CFSE-labeled cells, and CD44⁺ cells among CD8⁺ T cells on day 5 after cocultured with microglia (n = 3/group). Data are presented as a representative plot (e) and quantified percentages (f-g). h-j Flow cytometry analysis of Ki67⁺ cells from CD8⁺ T cells in the spleens and brains of Cx3cr1-Cre;5×FAD and Gsdmd^fl/flCx3cr1-Cre;5×FAD mice (n = 5/group). Data are presented as a representative plot (h) and quantified percentages (i-j). k Western-blot analysis of phospho-IκB-α, IκB-α, phospho-p65, p65, and PD-1 in CD8⁺ T cells after treatment with rIL-1β. Data are pooled from three independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Error bars show means ± SEM. One-way ANOVA with Dunnett’s multiple comparisons test for c-d, f-g. Unpaired t test for i-j

Attenuation of 5×FAD mice pathogenesis by pharmacological inhibition of GSDMD and/or anti-PD-1 antibody treatment at the early-stage of AD model

Since Gsdmd^fl/flCx3cr1-Cre;5×FAD mice with low levels of T cell PD-1 have milder symptoms, we assessed whether pharmacological inhibition of GSDMD and anti-PD-1 affected the development of disease model. 5×FAD mice were treated with the GSDMD inhibitor DMF [31] or an anti-PD-1 antibody at the early disease stage (Fig. 9a). Behavioral and pathological assessments demonstrated that AD model progression was partially restricted by DMF treatment (Fig. 9b-g). Flow cytometric results showed that brain CD8⁺ T cells and the peripheral immune response were both suppressed, while CD4⁺ T-reg cells were modestly increased with DMF treatment (Fig. 9h-o). In the presence of GSDMD, treatment with an anti-PD-1 antibody had a limited effect on AD development (Fig. 9b-g). This may be due to the positive effect of PD-1 blockade on the activation of effector CD8⁺ T cells. According to the flow cytometry results, despite the number of splenic T-reg cells increased, the activation of CD8⁺ T cells in spleen and its number in brain were not reduced (Fig. 9h-o). This result is consistent with previous findings that anti-PD-1 treatment in AD tauopathy mice increased the number of CD4⁺ T-reg cells but no obvious effect on effector CD8⁺ T cells in the brain [25]. Therefore, an approach combining GSDMD inhibition and anti-PD-1 blockade was employed. Mice treated with both DMF and the anti-PD-1 antibody had a significantly weakened AD phenotype (Fig. 9b-g), accompanied by greatly increased numbers of splenic T-reg cells and decreased numbers of brain CD8⁺ T cells (Fig. 9h-o). Taken together, these results demonstrate that GSDMD inhibition combined with anti-PD-1 treatment at the early stage reduces excessive T cell activation and restricts inflammation, thereby reducing the development of AD.

Fig. 9

The inhibitory effects of DMF and/or the anti-PD-1 antibody on 5×FAD mice. a Schematic representation of the experiments in b-l. b-d MWM test for the indicated mice at 6 months of age. Data are presented as representative swimming trace (b), number of crossing platform (c) and time spend in the target quadrant during probe test (d) (n = 5/group). e-g Immunofluorescence analysis of Aβ plaques (Thioflavin S) and IBA1 in the hippocampus of brains from the indicated mice (n = 5/group). Scale bar = 200 μm. Data are presented as representative images (e) and quantified statistics (f-g). h-i The quantified percentages (h) and absolute numbers (i) of CD8⁺ T cells that infiltrated into the brains of indicated mice were quantified by flow cytometry (n = 5/group). j-l Flow cytometry analysis of effector T cells (CD44⁺CD62L⁻) and naive T cells (CD44⁻CD62L⁺) among CD8⁺ T cells in the spleens of indicated mice (n = 5/group). Data are presented as a representative plot (j), quantified percentages (k), and absolute cell numbers (l). m-o Flow cytometry analysis of T-reg cells (Foxp3⁺) from CD4⁺ T cells in the spleens of indicated mice (n = 5/group). Data are presented as a representative plot (m), quantified percentages (n), and absolute cell numbers (o). Data are pooled from three independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Error bars show means ± SEM. One-way ANOVA with Dunnett’s multiple comparisons test for c-d, f-i, n-o. Two-way ANOVA with Sidak’s multiple comparisons test for k-l

Plasma sPD-1 levels in AD patients are correlated with IL-1β levels and disease progression

To assess the associations between PD-1, IL-1β, and AD clinical characteristics, a total of 84 patients and 33 controls were enrolled in this study. No statistical differences were found in age or gender between the AD patients and control subjects (Table 1). The Mini-Mental State Examination (MMSE) and Montreal Cognitive Assessment (MoCA) scores are commonly used in clinical settings as indicators of cognitive function, with lower scores suggesting more severe cognitive impairment. In this study, the scores for AD patients were significantly different from those for control subjects (Table 1). AD biomarkers including Aβ1–42, Aβ1–40, t-tau, and p-tau-181, in the CSF of AD patients were evaluated by ELISA (Table 1). The plasma IL-1β and soluble PD-1 (sPD-1) levels of the AD patients were both higher than those of the controls (Fig. 10a, b). However, no correlation was found between plasma sPD-1 and IL-1β levels in AD patients (Fig. 10c). We divided AD patients into two groups based on the average value (20 pg/ml) of plasma IL-1β in controls (Additional file 8: Fig. S8f). Plasma sPD-1 levels in IL-1β^high AD patients were significantly higher than those in controls (Additional file 8: Fig. S8g), with a positive correlation with plasma IL-1β (Fig. 10d). Moreover, in IL-1β^high AD patients, plasma sPD-1 levels were negatively correlated with CSF Aβ1–42 levels (Fig. 10e) and positively correlated with CSF p-tau-181 levels (Fig. 10f) but not CSF t-tau levels (Additional file 8: Fig. S8h). Plasma sPD-1 levels were negatively correlated with MMSE and MoCA scores for IL-1β^high AD patients (Fig. 10g, h). Similar correlations were observed between plasma IL-1β and the above clinical indices (Additional file 8: Fig. S8i-m). In addition, the activation of NF-κB in the PBMCs of IL-1β^high AD patient was enhanced compared to controls and was accompanied by increased levels of PD-1 (Fig. 10i and Additional file 8: Fig. S8n-r). Collectively, these results indicate that in AD patients with evidence of peripheral inflammation, plasma sPD-1 levels are correlated with IL-Iβ and disease progression. Furthermore, these results suggest that PD-1 may be regulated via the NF-κB pathway.

Table 1 Clinical characteristics of the study participants

Fig. 10

Plasma sPD-1 levels in AD patients correlate with IL-1β levels and AD progression. a Plasma IL-1β levels in controls and AD patients measured by ELISA. b Plasma sPD-1 levels in controls and AD patients measured by ELISA. c Correlation between sPD-1 and IL-1β levels in plasma from AD patients. d Correlation between sPD-1 and IL-1β levels in plasma from IL-1β^high AD patients. e-f Correlations between plasma sPD-1 levels and CSF Aβ1–42 (e), and p-tau-181 (f) levels in IL-1β^high AD patients. g-h Correlations between plasma sPD-1 levels and MMSE (g) and MoCA (h) scores in IL-1β^high AD patients. i Western-blot analysis of phospho-IκB-α, IκB-α, phospho-p65, p65, and PD-1 in PBMCs from controls and IL-1β^high AD patients (n = 3/group). n = 33 for controls, n = 40 for IL-1β^low AD patients, and n = 44 for IL-1β^high AD patients. ***P < 0.001. Error bars show means ± SEM. Unpaired t test for a-b. Correlations were established by calculating correlation coefficients

Discussion

Innate and adaptive immune responses contribute to the development of chronic AD neuroinflammation [2]. Recent studies have demonstrated the roles of peripheral immune cells, particularly T cells, in AD inflammation [25, 32, 33]. However, the process and mechanism by which the innate immune regulates T cell immune response in AD remain largely unknown. In this study, we found that peripheral inflammation is an often overlooked early event in AD pathophysiology, and that myeloid GSDMD is central to the peripheral inflammatory response in AD. GSDMD deficiency was found to impair T cell activation and the expression of PD-1, which prevents T cells infiltration into the brain and reduces inflammation in both the CNS and periphery. Furthermore, the combination of a GSDMD inhibitor and anti-PD-1 antibody treatment synergistically reduced T cell-mediated inflammation and the pathological burden in AD mice.

GSDMD is a pyroptosis executive molecule that has been associated with neurodegenerative diseases, including AD [34]. The brains of AD patients exhibit increased numbers of cleaved GSDMD-positive cells in close proximity to Aβ plaques [10]. In addition, the level of GSDMD is increased in the CSF of AD patients and can be used as a biomarker for AD diagnosis and identification of vascular dementia [35]. In the periphery, our previous study found that GSDMD is activated in the PBMCs of aMCI and AD patients [23]. Herein, we comprehensively evaluated the dynamic changes in GSDMD at different stages of AD and found that GSDMD activation in splenic myeloid cells occurs much earlier than that in the CNS. In our animal model, GSDMD activation is induced by different caspase signals, implying a broader effect than that of inflammasomes in AD. Notably, GSDMD activation in the CNS also includes infiltrated macrophages, which may have been overlooked in previous studies [10]. The activation of microglia and astrocytes before Aβ deposition could be due to initial responses to early Aβ oligomers or protofibrils, and perhaps could be priming early inflammation that leads to GSDMD activation and the subsequent downstream effects. To the best of our knowledge, this is the first study in which genetic knockout of GSDMD in mice was used to determine the role of GSDMD in AD peripheral inflammation. We found that both conventional knockout and myeloid cell conditional knockout mice were resistant to the development of AD, as assessed by behavioral and cognitive improvement, Aβ plaque reduction, and inflammation suppression. More importantly, peripheral T cell response impairment in mice lacking myeloid GSDMD implied that GSDMD plays a role in peripheral myeloid cells rather than microglia at the early stage of AD.

In this study, RNA-seq analysis identified T cell response-related pathways including the Adaptive immune response, Antigen processing and presentation of exogenous peptide, Peptide antigen assembly with MHCII complex, and Positive regulation of T cell activation, were significantly downregulated in Gsdmd^fl/flCx3cr1-Cre;5×FAD mice. The heatmap and volcano plot also found significant reductions in several genes critical for T cell immune response (Pdcd1, Nfkb1, Csk, Cd74, H2-Ob, H2-Ab1) and migration (Cxcr4, Ccr6, Ccr1, Ccl8), indicating that GSDMD knockout in myeloid cells not only reduced the production of inflammatory cytokines, but also restricted T cell immunity at an early stage, reducing the migration of T cells into the brain at a later stage when the blood–brain barrier was weakened. Recent studies have found a critical role for CD8⁺ T cells in AD patients as well as in murine disease models of pathogenesis [13, 15, 16]. Consistent with these findings, our results showed that CD8⁺ T cells were more enriched than CD4⁺ T cells in the brains of AD mice. Notably, the activation of splenic CD4⁺ T cells is also affected by GSDMD. Furthermore, recent studies have described a role for microglia in AD CD8⁺ T cell responses and associated with tauopathy in mice [25, 36]. In vitro, we also proved that microglia could serve as APCs and present Aβ antigens to T cells and that antigen presentation is weakened by GSDMD deficiency. However, GSDMD activation in microglia can be barely detected until 6 months of age in our mouse model, suggesting that the early-stage T cell response is mainly affected by GSDMD in peripheral APCs.

From the PPI network analysis, we found that Pdcd1 reduction was most obvious among the hub genes related to T cell responses. The PD-1 receptor is an immune checkpoint protein expressed on conventional CD4⁺ and CD8⁺ T cells, as well as T-reg cells. It delivers inhibitory signals that regulate the magnitude of adaptive immune responses and immune tolerance [37, 38]. Studies have shown that PD-1 blockade reduces pathology and improves memory function in mouse models of AD [39, 40]. PD-1 blockade can not only increase the activation of CD8 effector T cells, but also enhance CD4 T-reg cell suppressive activity through the PI3K/AKT pathway [27, 28]. Our in vitro experiments showed that treatment with an anti-PD-1 antibody promoted the proliferation and differentiation of Aβ-specific CD8⁺ effector T cells. However, no obvious CD8⁺ effector T cell changes were observed in the periphery or the CNS after treatment with the anti-PD-1 antibody in vivo. This may be explained by the simultaneous increase in immunosuppressive T-reg cells induced by PD-1 blockade, which eventually attenuates the 5×FAD mouse phenotype. Similar results have been reported recently in tauopathy mice [25].

Notably, we found that myeloid cell GSDMD deficiency reduced PD-1 expression on T cells and blocked the positive effect of the PD-1 inhibitor on CD8 effector T cells in vitro. However, GSDMD blockade may not entirely replace PD-1 inhibition in AD, especially for T-reg cells, as the expression and function of PD-1 can be influenced by signals other than GSDMD in vivo [41, 42]. This observation prompted us to consider combinatorial AD treatment using GSDMD inhibition and PD-1 blockade in 5×FAD mice. Combinatorial treatment at an early stage had marked effects on 5×FAD mice, with reduced CD8⁺ T cells activation and increased CD4⁺ T-reg cell numbers. IL-1β, which is released downstream of GSDMD, is widely known to promote T cells activation. The IL-1 receptor can regulate PD-1 expression on tumor-associated macrophages by promoting the recruitment of NF-kBp65 to the Pdcd1 promoter [30]. Our results showed that IL-1β increased the phosphorylation of IκB-α and p65 in T cells, suggesting that GSDMD regulates T cell PD-1 expression through the IL-1β/NF-κB pathway. Nevertheless, IL-1β may also counteract the suppressive effects of PD-1 ligation on T cells [43, 44]. Moreover, we found that AD patients with high plasma IL-1β levels also exhibited increased sPD-1 levels. The plasma levels of IL-1β and sPD-1 are correlated with the clinical characteristics of AD patients. In addition, the NF-κB/PD-1 pathway is activated in the PBMCs of IL-1β^high AD patients, emphasizing that GSDMD and PD-1 may be novel potential therapeutic targets for AD.

Several limitations should be noticed. First, the specific signals that trigger GSDMD-mediated pyroptosis in the periphery at early stage of AD model require further investigation. It remains unclear whether Aβ oligomers or protofibrils produced at early ages could be transported to periphery by CSF and act as damage signals inducing pyroptosis before they deposit and form plaques in the CNS. Additionally, alterations in gut microbiota and metabolites, observed as early as 2 months of age in 5×FAD mice [45], may also contribute to peripheral pyroptosis. Furthermore, the Thy1 promoter driving transgenes overexpression in 5×FAD mice is also expressed in T cells, natural killer cells, and fibroblasts. This renders extracerebral transgene expression a potentially independent factor in the activation of peripheral pyroptosis. While our study demonstrated that GSDMD activation in periphery occurs much earlier than in CNS, the Gsdmd^fl/flCx3cr1-Cre model simultaneously deletes GSDMD in both peripheral myeloid cells and microglia. Utilizing bone marrow chimeric mice could provide clearer insights into whether peripheral GSDMD influences GSDMD activation in the CNS. Lastly, caution is warranted in interpreting some statistical results due to the limited number of animals in each group.

Conclusions

This study demonstrated that myeloid GSDMD plays an important role in the development of AD. Myeloid GSDMD affects the immune response by directly regulating the activation of effector T cells and the level of PD-1, as well as promoting peripheral inflammation, which contributes to AD neuroinflammation (Additional file 9: Fig. S9). Developing therapeutic strategies that specifically target GSDMD-mediated pyroptosis and PD-1-mediated immunosuppression may be useful means by which to inhibit inflammation and enhance treatment outcomes of AD patients with signs of peripheral inflammation.