2.1 Cell Culture

BV2 cells, 293F cells, and 293 T cells (sourced from the cell bank of the Chinese Academy of Sciences) were cultured in DMEM (Wisent, Montreal, Canada). The culture medium was supplemented with 10% fetal bovine serum (FBS) (PAN-Biotech, Germany) and 1% penicillin/streptomycin (Wisent, Montreal, Canada). Cells were cultured in a humidified incubator maintained at 37°C with 5% CO₂.

Cys-D (Cat# A53299, China) was purchased from OKA, Tg2-IN1 (Cat# HY117678, USA) was purchased from MCE, and etoposide (Cat# SC0173, China) was purchased from Beyotime.

2.2 Animal Experiments

Mice experiments were performed in the animal facility of Tsinghua University (Beijing, China) with approval of the Institutional Animal Care and Use Committee of Tsinghua University (approval number: 22-DHT3). C57BL/6J mice were purchased from Jackson Laboratory through the Laboratory Animal Research Center and housed in the animal facility of Tsinghua University with ad libitum access to diet and water. Animal rooms were maintained at 20°C–22°C with 30%–70% relative humidity and a 12-h light/dark cycle.

Using a dose of 200 mg/kg, Cys-D was administered by gastric gavage to 16-month-old C57 mice (n = 8) every other day for 2 months. A control group of mice (n = 8) received sterile water via gavage during the same period. Within 2 weeks of completing the Cys-D treatment, the mice were individually subjected to a series of behavioral tests, including the rotarod test, grip strength test, open-field test, and Y-maze test.

2.3 Isolation of Primary Senescent Microglia

The Single Cell Suspension Preparation Instrument (DCS-400, RWD Life Science Co Ltd., China) was used to prepare single-cell suspensions from the brains of 22-month-old mice (n = 3), following the program “M-NeoBrain Heater-1.” The CD11b + Cell Sorting Kit (K1306-10, RWD Life Science Co Ltd) was employed to isolate microglial cells from single-cell suspensions. The isolated cells were stained with fluorescent antibodies: FITC anti-mouse/human CD11b (M1/70, Biolegend) and IBA1/AIF-1 (E4O4W, CST). Subsequently, flow cytometry was performed to sort and obtain relatively pure primary microglial cells. These primary microglial cells were further cultured for 3 weeks, and senescent microglia were identified using the SPiDER-βGal assay (SG02, Dojindo). All cell sorting procedures were performed on the Bigfoot Spectral Cell Sorter (Thermo Fisher Life Science Co Ltd., USA).

2.4 Western Blotting

Cells were lysed in RIPA lysis buffer (Cat# P0013K, Beyotime, China) added with protease inhibitors cocktail. Equal volumes of protein were separated by 12% sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) and electrotransferred onto a polyvinylidene difluoride (PVDF) membrane. Primary rabbit anti-Tgm2 (Cell Signaling Technology, Danvers, USA), rabbit anti-P105, rabbit anti-P65/Rel, anti-phospho-P65 (Ser 536), rabbit anti-H3 (Abcam, Cambridge, UK), rabbit anti-IκBα, rabbit anti-phospho-IκBα (Ser 32) (HuaBio, Hangzhou, China), rabbit anti-P50, rabbit anti-phospho-P50 (Ser 337) (Proteintech, Chicago, USA), rabbit anti-β-actin, secondary anti-rabbit HRP-IgG antibodies (Cell Signaling Technology, Danvers, USA), and M5 HiClear Prestained Protein Ladder (Mei5bio, Beijing, China) were used for immunoblotting.

BV2 cells were subjected to nuclear and cytosolic fractionation using the Nuclear/Cytosol Fractionation Kit (invent, SC003), following the protocol recommended by the manufacturer.

2.5 Tandem Mass Tag (TMT)-Based Quantitative Proteomic Analysis

2.6 Cross-Linking Identification by MS

A PVDF membrane of separated proteins was excised for in-gel digestion, and proteins were identified by mass spectrometry. Briefly, proteins were reduced with 25 mM dithiothreitol (DTT) and alkylated with 55 mM iodoacetamide. In-gel digestion was performed using sequencing grade-modified trypsin in 50 mM ammonium bicarbonate at 37°C overnight. The peptides were extracted twice with 1% trifluoroacetic acid in 50% acetonitrile aqueous solution for 30 min. The peptide extracts were then centrifuged in a SpeedVac to reduce the volume.

The mass spectrometer was run under DDA mode. The survey of full-scan MS spectra (200–1500 m/z) was acquired in the Orbitrap with 60,000 resolutions. The AGC target of 3e⁶ and the maximum injection time of 25 ms. Then, the precursor ions were selected into collision cell for fragmentation by higher-energy collision dissociation, the normalized collection energy was 30. The MS/MS resolution was set at 15,000, the AGC target of 5e⁴, the maximum injection time of 22 ms, and dynamic exclusion was 30 s.

Raw data were searched using the pLink2 software against IκBα sequences with trypsin specificity. The search parameters were as follows: peptides mass tolerance of 20 ppm; MS/MS tolerance of 0.02 Da; Peptide length: 4–100 aa; three missed cleavages allowed; oxidation on Met as the dynamic modification; and carbamidomethylation on Cys as static modification. A crosslinker monoisotopic mass of −17.026 Da (Q-K = −17.026 Da) was manually added.

2.7 Cloning, Expression, and Characterization of Tgm2 Protein

Tgm2 plasmid was purchased from Wuhan Miaoling Biotechnology Co. Ltd. Oligonucleotide primers:

(5′-GCTAGCGTTTAAACGGGCCCGCCACCATGGCAGAGGAGCTG-3′) and (5′-GCGGTTTAAACTTAAGCTTCTAATGATGATGATGATGATGGGCCGGGCCGATGATAAC-3′) were designed for PCR amplification of Tgm2 fragments. Primer pairs (5′-AAGCTTAAGTTTAAACCGC-3′) and (5′-GGGCCCGTTTAAACGCTAGC-3′) were used to amplify the pCDNA3.1 vector fragment. The two fragments were ligated by using Seamless Cloning Kit (Beyotime, D7010s). After sequence validation, the resulting plasmid constructs were used to produce the recombined Tgm2 proteins in HEK-293F cells. Before transfection, the fresh culture medium was replaced, and the final concentration of the expression vector plasmid DNA in the transfection volume was 3 μg/mL, with a final concentration of PEI at 9 μg/mL. After 24 h of transfection, fresh culture medium was added at a 1:1 ratio. After 3 days of culture, cells were collected by centrifugation at 4000 rpm for 15 min. The media were discarded and cells were resuspended in PBS buffer. Initially, cells were disrupted using a low-temperature ultrahigh-pressure cell disruptor at 800 MPa for 30 min and centrifuge at 17,000 rpm for 1 h. Subsequently, the target protein was purified using a Ni-NTA column on an AKTA purification platform. After concentration and buffer exchange on Centricon-10 (Millipore), the proteins in PBS buffer were aliquoted, flash frozen in liquid nitrogen, and then stored at −80°C. The purified proteins were separated by 12% SDS-PAGE under reducing conditions and stained with Coomassie brilliant blue. Finally, the target protein was identified using WB (6 × His antisera as the primary antibody and rabbit antisera as secondary antibody) quantified using BCA (Beyotime).

2.8 Immunofluorescence

BV2 cells cultured on glass coverslips were fixed in 4% paraformaldehyde for 20 min. Cells permeabilized in PBS with 0.5% Triton X-100 for 15 min at room temperature. Blocked as above, and then incubated with anti-Tgm2 antibody and anti-p65 antibody at 4°C overnight. Then, it was incubated with a secondary antibody of Alexa Fluor 488-conjugated anti-rabbit IgG secondary antibody (1:200) (Abcam, ab150077) or Alexa Fluor 647-conjugated anti-mouse IgG secondary antibody (1:200) (Abcam, ab150115). Finally, the nuclei were stained with DAPI. Images were captured with LSM980 (ZEISS, Germany).

2.9 Cell Viability Assay

Cells were seeded and cultured in 96-well plates. After treatment, the cell survival rate was measured using CCK8 reagent (Cat# K1018, ApexBio Technology, USA) according to the manufacturer's instructions. Briefly, CCK8 reagent was added into the wells, then the plates were incubated in a cell incubator for 1.5 h. The absorbance at 450 nm was measured to calculate cell viability.

2.10 Real-Time Quantitative PCR Analysis

Total RNA was isolated from cells by using Trizol reagent (TIANGEN, Beijing, China). Liver tissues were added into Trizol reagent and homogenized with a grinding rod. Then, cDNA was synthesized using a commercially available kit following the manufacturer's instructions (CWBIO, Beijing, China). qPCR analysis was performed using SYBR green reaction mixture (CWBIO, Beijing, China) and Roche LightCycler 96 System (Roche, Basel, Switzerland). β-Actin was used as the internal control for relative quantification. Primers used in qPCR are listed in Data S1.

2.11 Hematoxylin–Eosin Staining

Tissues were fixed with 4% paraformaldehyde overnight. Then, tissues were embedded in paraffin. Five-micrometer sections were prepared, deparaffinized, and hydrated sequentially, stained with HE. The sections were dehydrated by gradient alcohol, transparentized with xylene, and mounted with neutral resin for light microscopy.

2.12 Immunohistochemistry Staining

Tissues were fixed with 4% paraformaldehyde overnight. Then, tissues were embedded in paraffin according to the standard protocol. Five-micrometer sections were prepared, deparaffinized, and hydrated sequentially. Antigen retrieval was performed using EDTA buffer (Servicebio), and endogenous peroxidase was blocked. The slides were blocked with 3% BSA at room temperature for 30 min, before incubation with Tgm2 primary antibody at 4°C overnight. After being washed with PBS twice, slides were incubated with secondary antibody (Servicebio). The signal was detected by DAB staining. Nuclear was stained by hematoxylin. The sections were dehydrated by gradient alcohol, transparentized with xylene, and mounted with neutral resin for light microscopy.

2.13 SA-β-Galactosidase Staining

Cellular senescence was assessed by examining the activity of β-galactosidase using SA-β-Gal staining kit (Cat# C0602, Beyotime, China) according to the manufacturer's instructions. Briefly, the cells in 12-well plates were fixed with fixation solution for 10 min, then washed with PBS twice and stained with SA-β-Gal solution at 37°C overnight. Then, the samples were imaged by light microscopy (Nikon-90i, Japan), and the blue-stained cells were identified as senescent cells.

2.14 Rotarod Test

Mice were trained on the Rotarod (Ugo Basile, Italy) at a speed of 10 rpm for 3 days. During the test, the speed of the Rotarod was 10 rpm, and the time on the rod before falling was recorded.

2.15 Grip Strength Test

The maximal grip strength was measured using a grip strength meter (Ugo Basile, Italy). The animals were put on a metal grid, and their tail was caught and pulled back. The force exerted on the grid by the animals' legs was measured in grams. Three tests were performed, and the mean of the three measurements was calculated.

2.16 Y-Maze Test

The Y-maze test serves as a tool to evaluate the short-term spatial working memory capabilities of rodents. The Y-maze test apparatus consists of three arms that intersect at 120° angles, forming a shape reminiscent of the letter Y. Each arm has walls that are 11 cm high, 10 cm wide, and 31 cm long. Herein, the Y-maze test was conducted according to a previously reported protocol (Choi et al. 2024). The cumulative duration in the food arm and frequency of leaving the food arm was recorded for 5 min. These data are statistically analyzed by automated behavioral video software (EthoVision XT 11.5, Noldus Information Technology b.v., Netherlands).

2.17 Open Field Test

The open field was utilized to assess the effects of Cys-D treatment on aged mice. The box (60 × 60 × 60 cm) was placed directly under the camera, and the mice were allowed to freely move within the box for 5 min. After the test of every mouse, the apparatus was cleaned and sterilized, and the experimental chamber was wiped with 75% alcohol. The movement of the mice was recorded on video. The distance walked and the average speed were analyzed using the EthoVision XT 11.5.

2.18 Construction of Tgm2 Knockdown BV2 Cell Line

Tgm2 shDNA of the mouse was subcloned into lentivirus vector TRC2-pLKD-puro to construct a lentiviral vector TRC2-pLKD-Tgm2-puro. Then, lentiviral vectors were packaged and the titer was determined. BV2 cells were transfected with the constructed TRC2-pLKD-Tgm2-puro (Tgm2-KD or TKD group) and TRC2-pLKD-puro (NC group) as controls. Tgm2-KD and NC groups were identified by RT-qPCR and WB.

Tgm2 gene shRNA sequence:

CCGGCCTGGTCTTTATCCTAAGATACTCGAGTATCTTAGGATAAAGACCAGGTTTTT.

2.19 Statistical Analysis

Student's t-test was used to compare the two groups. One-way ANOVA was used for multiple comparisons. p < 0.05 was considered statistically significant. Data were presented as the means ± SD. GraphPad Prism software (version 9.0) was used for statistical analysis.

3.1 Tgm2 is Highly Expressed in Senescent Microglia Both In Vivo and In Vitro

As previously reported, microglia are the primary source of Tgm2 in the brain (Giera et al. 2018; Liu et al. 2023). To explore the role and mechanism of Tgm2 in the brain aging process, we performed IHC and WB experiments on mouse brain tissue. Immunohistochemical staining demonstrated increased expression of Tgm2 in the CA1 and CA3 regions of the hippocampus in aged C57/B6 mice (Figure 1A). Furthermore, WB analysis confirmed that Tgm2 was highly expressed in the hippocampus of aged mice (p < 0.05, Figure 1B,C). These results suggested that Tgm2 was upregulated in aged mice. Additionally, we utilized flow cytometry in combination with cell sorting (Figure 1D and Data S1) to determine that Tgm2 was highly expressed in primary senescent microglia (p < 0.01, Figure 1E–G).

In addition, we established an in vitro model of senescent microglia by treating the BV2 cell line with etoposide. WB (p < 0.0001, Figure 2A,B) was used to confirm the high expression of Tgm2 in senescent BV2 cells. IF results also confirmed the high expression of Tgm2 in senescent BV2 cells, primarily localized in the cytoplasm (Figure 2C). Moreover, we performed a six-plex TMT quantitative proteomics analysis on senescent BV2 cells (Figure 2D), the heat map (fold change > 2.2, Figure 2E), and volcano map (Figure 2F) revealed that Tgm2 ranked seventh among the 187 upregulated proteins (p < 0.001, Figure 2G). These results suggest that Tgm2 is highly expressed in senescent microglia both in vivo and in vitro.

3.2 The NF-κB Pathway is Activated in Senescent Microglia and the Brain Tissues of Aged Mice

The NF-κB pathway was enriched in the results of TMT quantitative proteomics for senescent microglia by KEGG pathway analysis (Data S1), and the NF-κB pathway plays a critical role in the brain aging process (Lee et al. 2004; Liu and Mouradian 2024). The WB results of senescent cells revealed an increase in the expression of Tgm2, as well as an upregulation of P105 and p-P65(Ser 536) in the NF-κB pathway. However, the expression of IκBα was decreased (Figure 3A). In addition, it was found that compared to 3-month-old C57/B6 mice, the protein content of p-P65 (Ser 536) in the olfactory bulb, cortex, and hippocampal tissues of 20-month-old mice also significantly increased (Figure 3B). These results indicated that the NF-κB pathway was activated in senescent microglia as well as in the brain tissue of aged mice.

3.3 Tgm2 Inhibitors Can Reduce the Nuclear Translocation of NF-κB in Senescent Microglia

To investigate the role of Tgm2 in senescent BV2 cells, we used a Tgm2 inhibitor Cys-D to treat senescent cells. We extracted cytosolic and nuclear proteins from normal BV2 cells (labeled as Ctrl), senescent cells (labeled as Sene), and Cys-D-treated senescent cells (labeled as Sene+CD) for WB analysis. The results showed that the protein levels of p-P65 (Ser 536) and p-IκBα (Ser 32) were increased in the cytosol of senescent BV2 cells compared to the Ctrl group. The cytosolic protein of P65 and IκBα were increased in the Sene+CD group compared to the Sene group (Figure 4A). Additionally, it was observed that the protein levels of P65, p-P65 (Ser 536), and P50 were elevated in the nuclei of senescent BV2 cells from the Sene group. However, in the nuclei of the Sene+CD group, the protein levels of P65, P50, and p-P50 (Ser 337) were reduced, while the protein levels of Tgm2 and p-P65 (Ser 536) remained unchanged (Figure 4B).

Airyscan IF microscopy was used to confirm further the enhanced nuclear translocation of P65 in senescent microglia, which was attenuated by the administration of the Tgm2 inhibitors TIN1 and CD (Figure 4C). In conclusion, it was observed that the nuclear translocation of NF-κB increased in senescent microglial cells, and the Tgm2 inhibitor reduced the nuclear translocation of NF-κB.

3.4 Tgm2 Inhibition or Knockdown Reduces SASP of Senescent Microglia

To further explore the role of Tgm2 in senescent microglia, we used Tgm2 inhibitor and knocked down Tgm2 expression. The results showed that senescent BV2 cells exhibited enlarged morphology and positive staining for SA-β-gal, whereas the Cys-D could reduce senescent BV2 cells (Figure 5A). The increase in SASP levels is a crucial hallmark of senescent cells (Birch and Gil 2020; Tchkonia et al. 2013). Therefore, we analyzed the expression of SASP factors using RT-qPCR. The results revealed that SASP was upregulated in senescent BV2 cells, and Cys-D decreased the relative mRNA expression levels of SASP factors, including interleukin-6 (IL6), matrix metallopeptidase 3 (MMP3), matrix metallopeptidase 12 (MMP12), and chemokines such as CXCL1, CXCL2, and CXCL10 (Figure 4B). Additionally, Cys-D also reduced the relative mRNA expression levels of P21 and P53 in senescent BV2 cells (Figure 4C).

To further confirm the regulatory role of Tgm2 in SASP expression in senescent BV2 cells, we constructed a Tgm2 knockdown cell line. Consistent with the results obtained using Tgm2 inhibitors, we found that Tgm2 knockdown in senescent BV2 cells significantly reduced the expression of SASP factors, such as IL6, MMP3, and CXCL1 (Figure 4D). These results indicate that Tgm2 plays a regulatory role in SASP in senescent BV2 cells. Furthermore, Tgm2 inhibitors represent potential prophylactics or senomorphics to mitigate the SASP in senescent microglia.

3.5 Tgm2-Catalyzed Cross-Linking of IκBα Promotes NF-κB Nuclear Translocation

Next, we aim to understand the mechanism underlying the regulation of SASP factors by Tgm2 in senescent BV2 cells. Tgm2 inhibitors effectively reduced the SASP level in senescent microglia (Figure 4B). IκBα protein, as an inhibitory factor for NF-κB, was decreased in both total protein extracts (Figure 3A) and cytoplasmic protein fractions (Figure 4A) of senescent BV2 cells. IκBα is a key protein that enables Tgm2 to promote the nuclear translocation of NF-κB. We enriched the IκBα protein in the cytoplasm of senescent BV2 cells using IκBα monoclonal antibody and then performed nondenaturing WB. We found that the content of IκBα dimers was significantly higher in senescent BV2 cells compared to control cells (Figure 6A). To further verify Tgm2 cross-linking IκBα, we coincubated 10 μg purified Tgm2 protein (Data S1) and 50 μM CaCl₂ with enriched natural IκBα protein in the cytoplasm of senescent BV2 cells at 37°C for 1 h. The results showed that Tgm2 promoted the cross-linking of IκBα into dimers in the cytoplasm of senescent BV2 cells, reducing the non–cross-linked IκBα content (Figure 6B). The structure of IκBa dimer was predicted using AlphaFold 3.0 (https://alphafold.com), revealing that the two IκBa molecules are centrosymmetric (Figure 6C). Finally, mass spectrometry identified new covalent cross-linking sites K22 and Q248 of IκBα mediated by Tgm2 (Figure 6D). We suggest that the K22 and Q248 residues of each IκBa monomer form covalent bonds under the catalysis of Tgm2.

3.6 Tgm2 Inhibitor Alleviates the Aged-Associated Phenotype in Aged Mice

Based on the results showing the reduction of SASP in senescent microglia with Cys-D, we administered a dose of 200 mg/kg Cys-D every other day via oral gavage to 16-month-old mice (n = 8) for 2 months, while a control group of mice (n = 8) received sterile water via oral gavage. Within 2 weeks after the end of the gavage period, the mice underwent the rotarod test, grip strength test, open-field test, and Y-maze tests (Figure 7A). The Cys-D-treated group of mice initially showed a decrease in body weight, followed by stabilization, compared to the control group (Figure 7B). After treatment with Cys-D, the hair of the aged mice became shiny, and the hair loss condition improved compared to that of the Ctrl group (Figure 7C). Cys-D improved the performance of aged mice in the rotarod test, especially for female mice, whose duration on the rod increased by about 20% (Figure 7D). It did not affect the grip strength of male mice, but the grip strength of female mice increased by more than 20% (Figure 7F). Additionally, results from the open-field test (Figure 7E) and Y-maze test (Figure 7G) indicated that Cys-D improved the adaptability to new environments and spatial memory of old mice. In conclusion, Cys-D not only prevented hair loss in old mice but also partially reduced their body weight, and increased their agility without affecting their strength.

Treatment with Cys-D significantly reduced the levels of Tgm2, p-P65, and P21 in the hippocampal tissue of aged mice (Figure 7H). We speculate that the improvement in cognitive behavior in mice may be associated with reduced inflammation and delayed aging in hippocampal cells.