CD22 modulation alleviates amyloid β-induced neuroinflammation

Abstract

Neuroinflammation is a crucial driver of multiple neurodegenerative diseases, including Alzheimer’s disease (AD), amyotrophic lateral sclerosis (ALS) and Parkinson’s disease (PD). Yet, therapeutic targets for neurodegenerative diseases based on neuroinflammation still warrant investigation. CD22 has been implicated in neuroinflammatory diseases, namely AD. Specifically, plasma soluble CD22 (sCD22) level is upregulated in patients with AD. Direct experimental evidence for the role of CD22 in neuroinflammation is needed, as is a better understanding of its impact on microglia activation and therapeutic potential. Here we reported that sCD22 promotes neuroinflammation both in vivo and in vitro. sCD22 activated microglia via both p38 and ERK1/2 signaling pathway for the secretion of TNFα, IL-6 and CCL3. Moreover, sCD22 activated microglia via sialic acid binding domain and 2,6 linked sialic acid glycan on sCD22. The pivotal therapeutic potential of targeting CD22 was demonstrated in Amyloid β (Aβ) induced-neuroinflammation in hCD22 transgenic mice. Suciraslimab improved working memory and resolved neuroinflammation in vivo. Further, membrane CD22 inhibited Amyloid β (Aβ) induced-NFκB signaling pathway and mechanistic study delineated that suciraslimab suppressed Aβ-induced IL-1β secretion in human microglia and PBMC. Suciraslimab also suppressed IL-12 and IL-23 secretion in human PBMC. Moreover, suciraslimab reduced the surface expression of α4 integrin on B cells. Intriguingly, we discovered that CD22 interact with Aβ and suciraslimab enhanced internalization of CD22-Aβ complex in microglia. Our data highlights the importance of sCD22 in driving neuroinflammation and the dual mechanism of targeting CD22 to resolve Aβ-induced inflammation and promote Aβ phagocytosis.

Introduction

Alzheimer’s disease is the most common form of dementia and is characterized by accumulation of amyloid plaque and excessive neuroinflammation [1, 2]. Mounting evidence suggests the role of immune cells in neurodegenerative diseases. Specifically, whole genome sequencing identifies mutation in innate immunity in Alzheimer’s disease [3,4,5]. Microglia is the pivotal immune cells that regulate neuroinflammation and is considered a key driver of neurodegenerative diseases [6,7,8]. However, despite the importance of microglial biology in neurodegenerative diseases, effective therapeutic option by targeting microglial-mediated neuroinflammation is lacking [9].

CD22, an inhibitory receptor on B cells, modulates peripheral immune response towards self-antigen [10]. CD22 regulates B cell activation via recognition of NeuAcα2–6Galβ1–4GlcNAc (α2,6-linked sialic acid) with its sialic acid binding domain (Domain 1 of CD22 extracellular domain) [11]. Dysregulated α2,6-linked sialic acid binding by CD22 has been suggested to cause autoimmune diseases [10] and drive Niemann Pick Type C (NPC) syndrome, a rare neurodegenerative disease, via lysosomal dysfunction of human microglia [12]. Recent findings in multiple human and mouse RNA-seq datasets demonstrated increased CD22 expression in aging population and patients with neurological disorders, including frontotemporal dementia (FTD) [13], amyotrophic lateral sclerosis (ALS) [14], NPC [15] and Alzheimer’s disease (AD) [13]. Particularly in AD, plasma levels of soluble CD22 (sCD22) are positively correlated with the severity of AD patients and are higher in levels in APOE ε4 carriers [16]. In mice, brain-infused anti-murine CD22 antibody (Clone: Cy34.1) was shown to improve cognitive function in aging mice and promoted clearance of toxic protein aggregates, namely Aβ and α-synuclein [17]. However, the pathological role of brain CD22 in driving AD and the therapeutic potential of CD22 modulation in AD remains largely unknown.

Given the relationship of CD22 with AD, this study aims to investigate the role of CD22 in promoting neuroinflammation and whether targeting CD22 provide therapeutic value in Aβ-induced neuroinflammation. Here, we find that sCD22 drives neuroinflammation via activating microglia. Mechanistically, sCD22 promotes neuroinflammation via its salic acid binding domain and MAPK signaling pathway. Pharmacological modulation of CD22 by suciraslimab alleviates Aβ-induced neuroinflammation in human CD22 transgenic mice. We further delineate the dual mechanism of action of suciraslimab which simultaneously suppresses Aβ-induced inflammation in microglia and peripheral mononuclear cells (PBMC) and promotes Aβ phagocytosis in microglia and oligodendrocyte, implicating a potential therapeutic target for AD.

Materials and methods

Materials

PBMC was purchased from iXCells (iXCells Biotech, CA, USA). All other materials included RPMI-1640 (Gibco, A10491-01), Dulbecco’s modified Eagle’s medium (DMEM) (Gibco, 11995-040), FBS (Gibco, A5256701), Pen/strep (Gibco, 15140-122), IL-34 (Sino Biological, 10948-H08S), GM-CSF (Sino Biological, 10015-HNAH), Geltrex (Gibco, A14133-02), human CD22 ECD (CD22 FL) (Sino Biological, 11958-H08H), human CD22 a.a. 176–687 (CD22 δ1) (Sino Biological, 11958-H08H1), mouse CD22 ECD (Sino Biological, 51177-M08H), Tanzisertib (Selleckchem, S8490), Perifosine (Selleckchem, S1037), Ravoxertinib (Selleckchem, S7554), SB856553 (Selleckchem, S7215). HMC-3 (ATCC^®CRL-3304) was purchased from ATCC and BV-2 (ABC-TC212S) was purchased from AcceGen.

Antibodies

Primary antibodies were anti phospho-ERK1/2 (Cell Signaling, 9101), anti ERK1/2 (Cell Signaling, 9102), anti phospho-p38 (Cell Signaling, 9211), anti p38 (Cell Signaling, 9212), anti-GFAP antibody (Abcam, Ab7260), anti Iba1/AIF-1 (Cell Signaling, 17198), anti-IbaI (FujiFilm Wako, 019-19741), anti Aβ antibody, clone 6E10 (BioLegend, 803014), anti-CD22 antibody (Abcam, ab181771), anti-CD22 antibody (Abcam, ab207727), anti-MAG antibody (Millipore, MAB1567) and anti-GAPDH antibody (Santa Cruz, sc-47724).

Mice

Wildtype C57BL/6Gpt and human CD22 transgenic mice C57BL/6JGpt-Cd22em1Cin(hCD22)/Gpt were provided by GemPharmatech Co., Ltd. Experiments were approved by the Institutional Animal Care and Use Committee (IACUC) at GemPharmatech. The IACUC approval code is GPTAP20230510-2.

Aβ (1–42) preparation

Beta-Amyloid (1–42), HFIP (rPeptide, A-1163-1) was purchased from rPeptide (GA, USA). Biotinylated monomeric Aβ was prepared by dissolving 0.5 mg Biotin Beta-Amyloid, with 200 μL hexafluoroisopropanol (HFIP). The mixture was then air-dried in fume hood overnight to remove HFIP. Dried Biotin Beta-Amyloid (1–42) was then dissolved with 30 μL DMSO. Further, 470 μL phosphate buffer saline (PBS) was added and monomeric Aβ was aliquoted and snap-freeze with liquid nitrogen. Monomeric biotinylated Beta-Amyloid (1–42) was used for BioLayer Interferometry Assay.

For the preparation of oligomeric Aβ, Beta-Amyloid (1–42), HFIP (rPeptide, A-1163-1) was dissolved with 30 μL DMSO. Further, 470 μL phosphate buffer saline (PBS) was added, and the solution was vortexed for 15 s and placed on ice and incubated at 4^oc for 26 h. After incubation, the mixture was aliquoted and snap-freezed with liquid nitrogen. The oligomeric Aβ was used for in vitro activation assay and in vivo generation of Aβ-induced neuroinflammation model in human CD22 transgenic mice. To prepare oligomeric FITC-conjugated Aβ (1–42), 0.5 mg Fluorescein Beta-Amyloid (1–42) (rPeptide, A-1119-1) was dissolved with 200 μL HFIP. The mixture was air-dried in fume hood overnight to remove HFIP. Dried fluorescein Beta-Amyloid (1–42) was then dissolved with DMSO and oligomeric form was prepared as described above.

To confirm the formation of oligomeric Aβ, 2 separate batches of oligomer prepared were analyzed with western blot. In brief, 9 μg of oligomeric Aβ was loaded into pre-cast 4–20% Tris-Glycine gradient gel (Bio-Rad, 4561096) and transferred to nitrocellulose membrane (Cytiva, 10600004). Membrane was blocked with 5% BSA in TBST and incubated with anti-Aβ antibody (Clone: 6E10, dilution 1:1000) overnight at 4^oc. Next, the membrane was washed with TBST and incubated with anti-mouse HRP-conjugated secondary antibody for 1 h at room temperature. Chemiluminescence signal was developed with ECL Western Blotting Substrate (Pierce, 32109) and signal was detected with ChemiDoc system (Bio-Rad, ChemiDoc MP). Western blot analysis revealed that majority of Aβ prepared is of oligomeric form (Fig. S6A).

Intracerebroventricular (i.c.v.) injection of mouse sCD22 in C57BL6/J wildtype mice

C57BL6/J wildtype male mice at 7–8 week of age were anesthetized and placed in a stereotaxic frame. A skin incision was made, and holes were drilled at ML 1 mm, AP, -0.4 mm. A total of 9 μg of mCD22-His protein (Sino Biological, 51177-M08H) in 3μL of PBS were delivered at z-depths of 2.5 mm. Injection of PBS into the same region of wildtype mice was served as control. Three days or 7 days after injection, mice were anesthetized and perfused with ice-cold PBS. For RNA-seq analysis, cortex and hippocampus from the left brain were dissected out, snap frozen in liquid nitrogen, and stored at − 80 °C before extraction of RNA. For histological analysis, right brain hemispheres were fixed in 4% PFA overnight at 4 °C and dehydrated and embedded in paraffin.

Mouse brain microglia analysis

Brain sections of 3 μm thickness were cut and antigen retrieved with sodium citrate buffer. Brain sections were washed with PBS. Anti-IbaI antibody (Cell Signaling, 17198S) was used in 1:800 dilution and anti-GFAP antibody (Abcam, Ab7260) was used in 1:1000 dilution. Slides were incubated overnight at 4°C with primary antibodies followed by incubation with biotinylated secondary antibody for 2 hours. Immunohistochemical staining was performed according to manufacturer’s instruction (Maxim Biotechologies, KIT-9701) and 3,3’-diaminobenzidine tetrahydrochloride (Maxim Biotechnologies, DAB-0031) was used as chromogen. Sections were counterstained with hematoxylin, mounted in Permount medium (Fisher Chemical, SP15-500), and observed under the light microscope (Olympus, BX51).

Monocyte-derived microglia-like cell (MDMi) differentiation

MDMi cells were generated following previously reported instructions with modifications [18]. Briefly, 2.5 × 10⁶ cells/mL fresh PBMCs were first resuspended in complete media (RPMI-1640 containing 10% FBS and 1% P/S) and seeded onto plates pre-coated with 2% Geltrex (Day 0) and incubated overnight at 37 °C with 5% CO2. Then non-adherent cells were carefully removed. The adherent monocytes were cultured with RPMI-1640 supplemented with 1% P/S, 0.1 μg/ml IL-34 and 0.01 μg/ml GM-CSF (differentiation medium). During differentiation, culture was replaced with differentiation medium every 2 days. To validate the identity of MDMi generated, MDMi were immunostained with resident microglia surface marker P2RY12. Briefly, MDMi were fixed with 4% PFA and immunostained with anti-P2RY12 antibody (Alomone Labs, #APR-020). Next, cells were incubated with Alexafluor 488-conjugated anti-Rabbit secondary antibody (Jackson ImmunoResearch, 111-545-003) and nuclei were counterstained with DAPI. Images were acquired with Carl Zeiss LSM 880 Confocal Microscope (Zeiss, USA).

MDMi activation with sCD22 in vitro

MDMi was prepared as described above. Cells were stimulated with sCD22 FL (Sino Biological, 11958-H08H) for 24 h. Cells were then fixed with 4% paraformaldehyde for 10 min at room temperature and permeabilized with PBS, 0.1% Triton-X 100 for 5 min at room temperature. Activation of MDMi was examined with anti-IbaI antibody (FujiFilm Wako, 019-19741). Confocal image acquisition was performed using a Zeiss LSM 880 laser scanning microscope (Carl Zeiss) and images were analyzed with ImageJ software.

Semi-quantitation of signaling molecules in MDMi

MDMi was treated with designated concentration of soluble CD22 as stated in the Figure for 30 min. MDMi lysate was collected with 1xRIPA lysis buffer (Millipore, 20–188) supplemented with protease and phosphatase inhibitors (Thermo Scientific™, 1861281). Thirty micrograms of protein samples were loaded for western blot analysis. Protein were detected with antibodies for phospho-ERK1/2 (Cell Signaling, 9101), ERK1/2 (Cell Signaling, 9102), phospho-p38 (Cell Signaling, 9211) and p38 (Cell Signaling, 9212). Chemiluminescence signal was developed with ECL Western Blotting Substrate (Pierce, 32109) and signal detected with ChemiDoc system (Bio-Rad, ChemiDoc MP).

Enzyme-linked immunosorbent assay

To examine the effect of sCD22-indcued TNFα, IL-6 and CCL3 release, MDMi or THP-1 cells were treated with either HEK293-derived CD22 FL (Sino Biological, 11958-H08H), CD22δ1(Sino Biological, 11958-H08H1) or CHO-derived sCD22 (Peprotech, 100-01) for 24 h before the collection of culture medium for cytokines/chemokines ELISA assay. To examine the signaling pathway involved in sCD22-induced cytokines secretion, pathway inhibitors were co-treated with CD22 FL (Sino Biological, 11958-H08H) for 24 h and supernatants were collected and examined with ELISA.

To examine the effect of SM03 treatment in Aβ-injected mice on CX3CL1, CCL24 and CCL4 release in mouse cortex, brain tissues were collected on Day 21 after Y-maze test.

To examine the release of IL-1β, IL-6 and TNF-α in MDMi and PBMC after Aβ treatment, MDMi and PBMC were incubated with designated concentration of Aβ and 0.1 μg/mL SM03 or IgG1 isotype overnight before the collection of culture medium for ELISA assay.

All cytokines and chemokines were measured with ELISA according to manufacturer’s instructions (R&D Systems, MN, USA).

Microglia viability assay

MDMi was cultured in RPMI-1640 containing 10% FBS and 1% P/S in 24 well plate. HMC-3 was cultured in EMEM (ATCC, 30-2003) supplemented with 10% FBS (Gibco, A5256701) and BV-2 was cultured in DMEM (Gibco, 11965092) supplemented with 10% FBS. Both HMC-3 and BV-2 were seeded onto 96 well plate and MDMi was prepared as described above. All the Cells were treated with 10 μg/mL CD22 FL (Sino Biological, 11958-H08H) for 3 days. Cell viability was analyzed with Cell Counting Kit 8 (Abcam, ab228554).

Acute neuroinflammation model by Aβ i.c.v. injection in hCD22 transgenic mice

The acute Alzheimer’s disease model in male human CD22 transgenic mice were generated by i.c.v. injection of oligomeric Aβ as described previously with modification [19,20,21]. Briefly, C57BL6/J wildtype mice at 6–9 month of age were anesthetized and placed in a stereotaxic frame. Skin incision was made, and holes were drilled at ML 1 mm, AP -0.4 mm. A total of 9 μg of oligomeric Aβ in 4.5 μl PBS were delivered at z-depths of 2.5 mm. Seven days after injection, Y-maze test was performed.

Y-maze

The subject mice were placed in a Y-shaped maze with three opaque arms at 120° from each other. They were allowed to freely explore the Y-maze for 8 min. The number of three consecutive alternations (ABC, ACB, BCA, BAC, CAB, CBA) was counted. The percentage alternation was calculated as (Number of 3 consecutive alternations)/((Total number of arm entries)) x100%.

NFκB reporter assay in HEK293 and HEK293-human CD22 stable cell line

pEGFP-N1-hCD22 plasmid was constructed by Genscript Co. Ltd. HEK293-hCD22 stable cell line was generated by transfection of pEGFP-N1-hCD22 plasmid with Lipofectamine 300 according to manufacturer’s instructions. Transfected cells were selected with 800 μg/mL G418 (Invivogen, ant-gn-1) in culture medium (DMEM + 10% FBS) to generate stable clones.

NFκB signaling reporter plasmid was constructed to examine NFκB signaling pathway in HEK293 or HEK293-hCD22. NFκB responsive element (5’-CTAGCGGGAATTTCCGGGGACTTTCCGGGAATTTCCGGGGACTTTCCGGGAATTTCCTAGAGGGTATATAATGGAAGCTCGACTTCCAGC-3’ and 5’- TCGAGCTGGAAGTCGAGCTTCCATTATATACCCTCTAGGAAATTCCCGGAAAGTCCCCGGAAATTCCCGGAAAGTCCCCGGAAATTCCCG-3’) were purchased from Sangon (China, Shanghai). The 2 oligos (50 μM each) were mixed and annealed with 1 x NEB buffer 2. Mixture was boiled for 5 min and allowed to cool down to room temperature. Annealed oligos were ligated with NheI and NotI-cut pNL3.3 plasmid (Promega, N1051) with T4 ligase (Invitrogen, 15224017). Ligation mix was transformed into TOP10 competent cells (Invitrogen, C404010). Positive clones were verified by sequencing.

To examine NFκB signaling, reporter plasmid was transfected into HEK293 or HEK293-hCD22 stable cell line with lipofectamine 3000 (Invitrogen, L3000008). Designated concentration of oligomeric Aβ were incubated with transfected cells overnight. Supernatant were collected and luciferase signaling was detected with Nano-Glo^® Luciferase Assay System (Promega, N1120) with Varioskan™ LUX multimode microplate reader (Thermo Scientific™, VL0000D0).

PBMC cytokines assay

Human PBMC (iXCells Biotech) was cultured in RPMI-1640 supplemented with 10% FBS and 1% P/S. PBMC was seeded onto 96 well plate at a density of 2.5 M/mL in 200 μL medium. On Day 1, PBMC was first stimulated with 10 ng/mL IFNγ (Sino Biological, 11725-HNAE). On Day 2, PBMC was stimulated with 100 ng/mL LPS with the presence of 1 μg/mL SM03 (SinoMab BioScience Limited) or IgG1 isotype (Sino Biological, HG1K). On Day 3, supernatant was collected and examined with IL-12 and IL-23 with ELISA kit according to manufacturer’s instruction (R&D Systems, USA, MN).

Examination of TLR4 expression on CD14 + cells in human PBMC

Human PBMC (iXCells Biotech) was cultured in RPMI-1640 supplemented with 10% FBS and 1% P/S. Cells were activated with IFNγ and LPS and incubated with IgG1 or suciraslimab as described above. Cells were collected and stained with anti-CD3-Alexa Flour F700 (Biolegend, 300424), anti-CD19-Pacific Blue (Biolegend, 302224), anti-CD14-APC (Biolegend, 325608) and anti-TLR4-PE (Biolegend, 312806). Cells were analyzed with Cytoflex S Flow Cytometer (Beckman Coulter, USA). Cells which were CD3-CD19-CD14 + were gated and the expression of TLR4 were examined.

Examination of α4 integrin on human B and T cells

Human PBMC (iXCells Biotech) was cultured in RPMI-1640 supplemented with 10% FBS and 1% P/S. PBMC was incubated with 1 μg/mL of antibodies overnight. Cells were then collected and stained with anti-CD3-Alexa Flour F700 (Biolegend, 300424), anti-CD19-Pacific Blue (Biolegend, 302224) and anti-α4 integrin-FITC (Biolegend, 304316). CD19 + CD3- B cells and CD19-CD3 + T cells were then gated and the effect of suciraslimab on the expression of α4 integrin were analyzed with Cytoflex S Flow Cytometer (Beckman Coulter, USA).

To examine the effect of T cell depletion and monocyte depletion on suciraslimab-modulation of α4 integrin, PBMC were depleted with T cells using CD3 microbeads (Miltenyi, 130-050-101) or depleted with monocyte using CD14 microbeads (Miltenyi, 130-050-201) according to manufacturer’s instruction. Depleted PBMC were treated with suciraslimab and the expression of α4 integrin were analyzed as described above.

Immunofluorescent staining

To examine the expression of CD22 on MDMi, MDMi were fixed with 4% PFA and permeabilized with 0.1% triton-X 100. Cells were immunostained with anti-IbaI antibody (FujiFilm Wako, 019-19741) and suciraslimab (SinoMab BioScience Limited). Next, cells were incubated with respective secondary antibodies and nuclei were counterstained with DAPI.

To examine the expression of NLRP3 and ASC after Aβ treatment in MDMi, cells were activated with 5 μg/mL oligomeric Aβ in the presence of suciraslimab or IgG1 isotype overnight. MDMi were then fixed with 4% PFA and permeabilized with 0.1% triton-X 100. Cells were immunostained with anti-NLRP3 antibody (AdipoGen, AG-20B-0014-C100) and anti-ASC antibody (AdipoGen, AG-25B-0006-C100). Respective secondary antibodies were incubated, and nuclei were counterstained with DAPI.

To examine the interaction of Aβ with human CD22, HEK293 cells and HEK293-hCD22 cells were cultured in DMEM supplemented with 10% FBS on coverslip. Cells were fixed with 4% PFA and immunostained with 10 μg/mL FITC-conjugated Aβ overnight. Cells were then counterstained with DAPI visualized nuclei. All the confocal images were acquired with Carl Zeiss LSM 880 Confocal Microscope (Zeiss, USA).

MO3.13 differentiation and examination of CD22 expression

MO3.13 was culture in Dulbecco’s modified Eagle’s medium (DMEM) (Gibco, 11995-040) supplemented with 10% FBS. To differentiate MO3.13 into oligodendrocyte phenotype, cells were cultured in DMEM supplemented with 100 nM PMA (Biogems, 1652981) and differentiated for designated days as stated in the figure. Cells lysates were collected and expression level of CD22, MAG and GAPDH were immunoblotted with anti-CD22 antibody (Abcam, ab207727), anti-MAG antibody (Millipore, MAB1567) and anti-GAPDH antibody (Santa Cruz, sc-47724).

To examine surface CD22 expression on differentiated MO3.13 with flow cytometry, cells were detached with TrypLE™ Express Enzyme (Gibco, 12605010) and incubated with anti-CD22 antibody conjugated with PE (Clone: RFB4) (Life Technologies, MHCD2204). Cells were analyzed with Cytoflex S Flow Cytometer (Beckman Coulter, USA).

Examination of CD22 expression in peripheral human monocyte and macrophage

Human PBMC (iXCells Biotech, CA, USA) was culture in RPMI1640 supplemented with 10% FBS. Human monocytes were isolated with Pan Monocyte Isolation Kit (Miltenyi, 130-096-537). To differentiate monocytes into peripheral macrophages, monocytes were cultured in RPMI1640 medium supplemented with 50 ng/mL M-CSF for 7 days. Differentiation medium was changed every 3 days. Both monocytes and peripheral macrophages were stained with 2 different clones of anti-CD22 antibody, namely anti-CD22 antibody (Clone: S-HCl-1)-APC (Biolegend, 363506) and anti-CD22 antibody (Clone: RFB4)-PE (Invitrogen, MHCD2204). Cells were analyzed with Cytoflex S Flow Cytometer (Beckman Coulter, USA).

BioLayer interferometry

Streptavidin sensors (Fortebio) were used to capture biotinylated monomeric Aβ proteins. In the Aβ loading step, 10 μg/mL biotinylated Aβ proteins were incubated with the sensors for 300s. In the association steps, designated concentration of soluble Human CD22 FL (Sino Biological, 11958-H08H) were incubated for 600s and dissociation steps in kinetic buffer was done in 600s. Background wavelength shifts were measured from reference biosensors that were loaded only with biotin monomeric Aβ. All experiments were performed at 30 °C with shaking at 1,000 rpm. ForteBio’s data analysis software 12.0.1.55 was used to fit the data to a 2:1 binding model.

Proximity ligation assay

HMC-3 cells were seeded on coverslips. Cells were incubated with 10 μg/mL oligomeric Aβ for 1 h and then fixed with 4% paraformaldehyde. The cells were incubated overnight at 4 °C with anti Aβ antibody, clone 6E10 (BioLegend, 803014) and anti-CD22 antibody (SM03) or IgG1 isotype (Sino Biological, HG1K) for the control group. The cells were washed and allowed to react to a pair of Duolink proximity probes (Millipore). Nuclei were counterstained with DAPI. Confocal image acquisition was performed using a Zeiss LSM 880 laser scanning microscope (Carl Zeiss).

Surface CD22 expression

HMC-3 were seeded onto coverslip and treated with SM03 for designated time as stated. Treated cells were fixed with 4% paraformaldehyde for 5 min at room temperature without permeabilization to stain for surface CD22. The cells were washed and probed with anti-CD22 antibody (Abcam, ab181771) overnight at 4^oc and then probed with anti-mouse antibody conjugated with AlexaFluor 488 for 1 h at room temperature. Nuclei were counterstained with DAPI and confocal images were acquired with a Zeiss LSM 880 laser scanning microscope (Carl Zeiss).

Surface suciraslimab binding on HMC-3

HMC-3 were seeded and cultured in EMEM + 10% FBS. Cells were incubated on ice with 10 μg/mL suciraslimab (α-CD22 Ab) for 1 h. Cells were washed with ice-cold PBS and then incubated at 37^oc for 5 min to allow internalization of suciraslimab. Next, cells were fixed with 4% PFA in PBS. Subsequently, fixed cells were washed and incubated with Goat anti-Human IgG (H + L) AlexaFluor488 conjugated secondary antibody (Invitrogen, A-11013). Cells were then analyzed with BD FACSLyric flow cytometry (BD biosciences, NJ, USA).

FITC-Aβ phagocytosis in vitro

HMC-3 were seeded and cultured in EMEM + 10% FBS. Cells were starved with EMEM only for 1 h and then incubated with 1 μg/mL FITC-Aβ and 0.1 μg/mL SM03 (α-CD22 Ab IgG), IgG1 isotype, or 0.016 μg/mL SM03 scFv (α-CD22 Ab scFv) for 4 h. Cells were washed with PBS and trypsinized. Cells were resuspended with 1% in BSA in PBS and analyzed with BD FACSLyric flow cytometry (BD biosciences, NJ, USA).

MO3.13 cells were differentiated as described above for 3 days. Cells were starved with DMEM for 1 h and then incubated with 1 μg/mL FITC-Aβ and 0.1 μg/mL SM03 or IgG1 isotype for 4 h. Cells were washed with PBS and trypsinized. Cells were resuspended with 1% in BSA in PBS and analyzed with BD FACSLyric flow cytometry (BD biosciences, NJ, USA).

RNA sequencing analysis

Total RNA from cultured MDMi and mouse cortex samples were extracted with TRIzol Reagent (Invitrogen, 15596026) according to the instruction. mRNA was purified with Oligo(dT)-attached magnetic beads, followed by fragmentation. The first strand and second cDNA were generated by random hexamer-primed reverse transcription. The cDNA was amplified by PCR and purified by AMPure XP Beads. The quality control of the products was verified by Agilent Technologies 2100 Bioanalyzer. After generating cluster on flow cell, sequencing by synthesis technology was used to obtain the base-by-base genome information. For data analysis, we first used fastp (version 0.23.2) to remove low-quality reads from raw data with default parameters [22]. Then, high-quality reads were aligned to the Homo sapiens (human) genome assembly GRCh38(hg38) or Mus musculus (house mouse) genome assembly GRCm39 (mm39) with the gene annotation file by HISAT2 (version 7.5.0) [23] using the parameters --no-mixed --no-discordant --qc-filter. Gene abundance was quantified by featureCounts (version 2.0.1) [24] with the default parameters. The gene-level read count matrix was then imported into the R (version 4.2.1) [25] for differential gene expression analysis. In this process, DEseq2 (version1.38.3) [26], edgeR (version 3.40.2) [27], and limma (version 3.48.3) [28] packages were used. Differentially expressed genes from each package were further filtered to retain that with adjusted P value < 0.05 and fold change contrast ≥ 1.5. The union of filtered differentially expressed genes from each packaged were used for functional enrichment with clusterProfiler (version4.6.2) [29].

GSEA

GSEA (Gene Set Enrichment Analysis) was conducted using the hallmark gene set from the Molecular Signatures Database [30]. The analysis was performed with the default settings, which included 1,000 permutations for the gene sets and the Signal2Noise metric for ranking the genes.

Statistical analysis

Two-tailed unpaired Student t-test, one-way ANOVA with Tukey’s post hoc test, and two-way ANOVA with Tukey’s post hoc test were conducted using GraphPad Prism software. All data are represented as mean ± SEM.

Data availability

RNA-seq data was deposited to GEO database under accession number GSE269121.

Result

Soluble CD22 promotes neuroinflammation

sCD22 was first reported as 100 kDa and 80 kDa protein in hairy cell leukemia patients [31]. sCD22 was detected with ELISA using 2 different clones of anti-CD22 antibodies, S-HCL-1 and RFB4, which target domain 1 and domain 2 of CD22 extracellular structure. Using the same methodology, plasma sCD22 was reported to be associated with brain amyloid burden and cognitive decline in Alzheimer’s disease [16]. It is postulated that the sCD22 protein associated with AD should contain both domain 1 and domain 2. Therefore, we will use sCD22 of complete extracellular domain to study its effect on neuroinflammation. We first investigated the effect of sCD22 in vivo by intracerebroventricular (i.c.v.) injection of His-tagged mouse sCD22 in wildtype C57BL/6 mice. Mice injected with the same volume of PBS were served as control (Fig. 1A). Cortices and hippocampi of sCD22-treated mice were collected 3 days or 7 days after i.c.v. injection. RNA seq analysis of cortices revealed that 3-day sCD22 treatment elicited robust neuroinflammatory response as manifested by the upregulation of pro-inflammatory genes including C3, Spp1, Cybb, Ccr5, and Dpp4 (Fig. 1B). Gene ontology analysis revealed that sCD22 promoted leukocyte migration (Lgals3, Itgb2, Icam1, Cxcl16 and Ccl9) and mediated B and T cell immune responses (Irf7, Fcgr1, Ptprc, Il4ra, Il21r, Ccr2, and Tnfrsf1b) as well as antigen presentation (Cd74, H2-Eb1, H2-K1, H2-Aa and B2m) (Fig. 1C-D). These data recapitulated with previous findings that CD22 functioned as an adhesion molecule and its ligand binding domain regulated the expression of major histocompatibility complex class II on B lymphocytes [32]. Intriguingly, RNA seq analysis of hippocampus after 3 days of sCD22 treatment revealed a less obvious inflammatory status (Fig. S1A-B). The set of upregulated inflammatory genes observed in cortices (Fig. 1B) was not as obvious in hippocampi (Fig. S1C). These data highlight that the neuroinflammatory effect of sCD22 is mainly contributed by mouse cortex. After 7-day treatment of sCD22, RNA seq and gene ontology analysis of cortex revealed upregulated genes in complement activation (C4b, Serpinb1, A2m and Mfap4) and neutrophil/granulocyte chemotaxis (Edn1, Thbs4 and Lbp) (Fig. 1E-F). Moreover, gene set enrichment analysis (GSEA) also revealed that sCD22-treated mouse cortices were positively enriched for inflammatory pathways, including TNFα signaling via NFκB, IL6/JAK/STAT3 signaling, IFN response, complement activation and cholesterol homeostasis (Fig. 1G, Fig. S1D), while PI3K/AKT/MTOR signaling was negatively enriched (Fig. S1E). The enrichment in cholesterol homeostasis resonated with dysregulated cholesterol function induced by sCD22 in previous report [12]. The enhanced neuroinflammation observed might be elicited by over-activation of microglia, but not astrocyte, as suggested by significantly enhanced IbaI + cells, but not GFAP + cells, after sCD22 treatment for 7 days (Fig. 1H). Intriguingly, we found that sCD22 treatment upregulated key AD-related Disease Associated Microglia (DAM) genes, including C3, Lgals3, Spp1, B2m, Ctss, Cybb, and H2-Ab1 (Fig. 1B & E). Many of which have been proven to be potential therapeutic targets for AD [33,34,35]. Collectively, sCD22-mediated neuroinflammation could possibly aggravate neurodegenerative diseases such as AD in vivo. Next, to explore the effect of sCD22 on microglia, human monocyte-derived microglia-like cells (MDMi) were generated by stimulating human monocytes with IL-34 and GM-CSF for 10 days. To better characterize the MDMi generated, we immunostained MDMi cells with resident microglia surface marker P2RY12 and found that MDMi cells were positive for P2RY12 (Fig. S1F). We treated MDMi with 10 μg/mL sCD22 for 24 h and differential gene expression was analyzed (Fig. 1I). Principal component analysis revealed that sCD22-treated MDMi samples had distinct transcriptional clustering profiles when compared to MDMi treated with PBS (Control) (Fig. 1J) sCD22 treatment steered microglia from homeostatic state to pro-inflammatory state by downregulating homeostatic genes, namely P2RY12 and P2RY13, while upregulating a spectrum of pro-inflammatory genes, including IL1B, TNF, CCL5, CXCL1 and CXCL10 (Fig. 1K). GSEA also identified sCD22-treated MDMi was positively enriched with IL6-JAK-STAT3 signaling, TNFα Signaling, Interferon γ response, inflammatory signaling and complement activation (Fig. S2), similar to what we previously observed in vivo (Fig. S1D-E & Fig. 1G). Gene ontology analysis revealed that sCD22 treatment upregulated leukocyte migration genes (CCL8, CXCL5, CXCL2, CXCL1 and ICAM1) and possibly via upregulation in MAPK signaling genes (MAP3K20, MAP2K6, MAP3K10, MAP3K7 and MAP3K9) (Fig. 1L & M).

Fig. 1

CD22 promotes neuroinflammation via microglia. A Schematic diagram of sCD22 i.c.v. injection into wildtype C57BL/6 mice. B Volcano plot of sCD22-treated mouse cortex after 3 days treatment. N = 3. C-D Gene ontology analysis of sCD22-treated mice (3 days treatment) in KEGG pathway (C) and Biological function (D). E Volcano plot of sCD22-treated mouse cortex after 7 days treatment. N = 4. F Gene ontology analysis of sCD22-treated mice (7 days treatment) in Biological function. G GSEA showing enrichment of IL6/JAK/STAT3, TNFα Signaling via NFκB, and Cholesterol Homeostasis of mouse cortex after sCD22 7 days treatment relative to PBS group. H Representative image and quantitation showing the effect of sCD22 on IbaI and GFAP expression in mouse cortex. IbaI: Student t-test, ** P < 0.01; GFAP: ns: not significant. N = 6. I Schematic diagram of MDMi differentiation and sCD22 treatment. J Principal component analysis showing sCD22 treated MDMi versus control MDMi. N = 4. K Volcano plot of sCD22-treated MDMi. L Gene ontology analysis of sCD22-treated MDMi in Biological function. M Gene ontology analysis of sCD22-treated MDMi in KEGG

sCD22 triggers pro-inflammatory cytokines release in a MAPK signaling- and sialic acid-dependent manner

As we have shown that sCD22 activated microglia both in vivo and in vitro (Fig. 1), next we investigated how sCD22 activates microglia. To examine its underlying mechanism, MDMi was treated with 10 μg/mL human sCD22 ECD recombinant protein for 24 h and immunostained with anti-IbaI antibody to examine its activation status. Indeed, sCD22 significantly enhanced immuno-signal of IbaI without affecting the total area of the cells, indicating that MDMi was activated by sCD22 in vitro (Fig. 2A). sCD22 treatment did not affect the viability of microglia as assessed with CCK8 assay in MDMi (Fig. 2B), human microglia cell line (HMC-3) (Fig. 2C) and mouse microglia cell line (BV-2) (Fig. 2D). To further comprehend the pro-inflammatory mechanism of sCD22, biochemical analysis revealed that sCD22 increased both ERK1/2 and p38 phosphorylation in MDMi (Fig. 2E). To confirm if sCD22 promotes neuroinflammation via both ERK1/2 and p38 signaling pathway, MDMi was stimulated with sCD22 in the presence or absence of Ravoxertinib (ERK1/2 inhibitor) and SB856553 (p38 inhibitor). Indeed, inhibition of either ERK1/2 or p38 partially inhibited sCD22-mediated release of TNFα, IL-6 and CCL3. Co-inhibition of ERK1/2 and p38 exerted stronger inhibition on sCD22-mediated cytokines release (Fig. 2F). The signaling cascade of sCD22-induced cytokine release was independent on JNK and AKT pathway as inhibitors of both pathways did not interfere with sCD22-induced CCL3 release in MDMi (Fig. 2G). Previous studies have shown that CD22 ectodomain adopted a tilted rod-like structure which forms CD22 nanocluster via sialic acid binding domain (D1) [36]. Also, the D1 structure of CD22 is essential for eliciting lysosomal dysfunction of microglia [12]. Therefore, we further examined if D1 structure has an essential function in triggering neuroinflammation in microglia (Fig. 2H). We activated MDMi withD1-truncated sCD22 (CD22 δ1). Intriguingly, D1-truncated sCD22 (CD22 δ1) induces a significant reduction in TNFα secretion and a trend of reduced secretion for IL-6 and CCL3 from MDMi (Fig. 2I). As CD22 is a glycoprotein which also presents 2,6 sialic acid on the ectodomain [35], we further addressed the possible effect of CD22 glycan on neuroinflammation. We utilized CHO (Chinese hamster ovary)-derived 2,6 sialic acid-null sCD22 by leveraging the fact that CHO cells lack β-galactoside α2,6-sialyltransferases activity [37]. Though HEK293-derived sCD22 significantly activated neuroinflammation in MDMi (Fig. 2F), CHO-derived sCD22 failed to induce CCL3 release in MDMi (Fig. 2J). This finding was validated in THP-1 cell line, a human monocytic cell line which has previously been used to model human microglia [38, 39]. HEK293-derived sCD22 significantly enhanced CCL3 secretion while CHO-derived sCD22 failed to do so (Fig. 2K). Collectively, these data suggested that sCD22 induced microglia-mediated neuroinflammation in a 2,6 sialic acid- and sialic acid binding-dependent manner.

Fig. 2

sCD22 promotes microglial neuroinflammation via MAPK-signaling pathway and in a sialic acid-dependent manner. A Immunostaining with anti-IbaI antibody to examine microglia activation after sCD22 treatment in MDMi. Student t-test, **** P < 0.0001, ns: not significant. N = 26–27, from 3 independent experiments. B-D Effect of sCD22 on viability of MDMi (B), HMC-3 (C) and BV-2 cells (D). E Representative and quantitation of western blot examining ERK1/2 and p38 phosphorylation in sCD22-treated MDMi. p38: One-way ANOVA, F = 16.01, P = 0.001, Tukey post hoc test ** P < 0.01; ERK1/2: One-way ANOVA, F = 7.38, P = 0.0108, Tukey post hoc test * P < 0.05. F Effect of ERK1/2 inhibitor (Ravoxertinib) and p38 inhibitor (SB856553) on sCD22-mediated TNFα, IL-6 & CCL3 release. TNFα: One-way ANOVA, F = 16.09, Tukey post hoc test **** P < 0.0001; IL-6: One-way ANOVA, F = 4.917, Tukey post hoc test * P < 0.05, ** P < 0.01; CCL3: One-way ANOVA, F = 6.672, Tukey post hoc test ** P < 0.01. N = 4–5. G Effect of pan JNK inhibitor (Tanzisertib) and Akt inhibitor (Perifosine) on sCD22-mediated CCL3 release. H Schematic diagram of sCD22 with complete extracellular domain (CD22-FL) and with D1-truncated (CD22-δ1). I Full length and D1-truncated sCD22 effect on TNFα, IL-6 & CCL3 release in MDMi. TNFα: One-way ANOVA, F = 7.847, Tukey post hoc test * P < 0.05, ** P < 0.01; IL-6: One-way ANOVA, F = 4.375, Tukey post hoc test * P < 0.05; CCL3: One-way ANOVA, F = 3.669, Tukey post hoc test * P < 0.05. ns: not significant. N = 5–6. J Effect of CHO-derived sCD22 on CCL3 release in MDMi. Student t-test, P = 0.83. N = 2. K Effect of HEK293-derived sCD22 and CHO-derived sCD22 on CCL3 release in THP-1. Two-way ANOVA, source of sCD22: F_(1,4) = 124, P = 0.0007; CCL3 release: F_(1,4) = 87.57, P = 0.0007. Tukey post hoc test, i < 0.001. ns: not significant. N = 2. All data are presented as mean ± SEM

CD22 is expressed in MDMi, PMA-differentiated MO3.13 cell line and human monocyte

Human RNA-seq data revealed that CD22 is highly expressed in oligodendrocyte and moderately expressed in microglia [13]. Particularly, CD22 expression is upregulated upon oligodendrocyte differentiation as the CD22 expression in oligodendrocyte precursor cells is negligible [13]. To validate the expression of CD22 in human microglia, we examined CD22 expression in MDMi with immunofluorescence staining. We found that CD22 was localized both in the surface membrane and in cytoplasmic compartment (Fig. S3A). This is in line with its expression in B cells that CD22 was present in the surface membrane and internalized constitutively into intracellular compartment [40]. Further, using MO3.13 human oligodendrocyte cell line, we examined the expression of CD22 upon oligodendrocyte differentiation. MO3.13 was differentiated with PMA and CD22 expression was examined 3 days and 7 days after. Immunoblot revealed that upon PMA treatment, MAG, a marker for oligodendrocyte maturation, was upregulated. Intriguingly, the expression of CD22 was also enhanced (Fig. S3B). To assess if the upregulated CD22 was expressed on the surface or resided in the cytoplasmic compartment, PMA-differentiated MO3.13 was detached and examined with anti-CD22 antibody (Clone: RFB4) in flow cytometry. The surface expression of CD22 was evident in differentiated MO3.13 cell line, indicating that upregulated CD22 was localized on the surface membrane and was amendable to therapeutic modulation (Fig. S3C).

As CD22 was expressed in microglia, a phagocytic cell type of myeloid lineage, we hypothesized that CD22 might also be expressed in monocyte or peripheral macrophage, which is the main phagocytic myeloid cells present in the peripheral blood. We tested the surface expression of CD22 on monocyte and peripheral macrophage with flow cytometry. As differential reactivity of anti-CD22 antibodies to basophils and dendritic cells have been reported [41, 42], we examined CD22 expression on isolated monocyte with 2 different clones of anti-CD22 antibodies. Clone S-HCl-1 targeted domain 1 of CD22 extracellular domain (ECD) while clone RFB4 targeted domain 2 of CD22 ECD [41]. Surprisingly, a modest CD22 expression was detected in isolated human monocyte using clone RFB4 while there is no signal while using clone S-HCl-1 (Fig. S3D). Further, we differentiated human monocyte with M-CSF to peripheral macrophage for 7 days. CD22 expression was detected in peripheral macrophage with both antibody clones. More importantly, we observed upregulation of CD22 expression in macrophage when compared to monocyte (Fig. S3E). This broadens our knowledge that CD22 expression is not restricted to B cells. CD22 expressed in monocyte and peripheral macrophage provides opportunities for therapeutic modulation by antibody in disease conditions.

Suciraslimab alleviates Aβ-induced neuroinflammation in human CD22 transgenic mice

Knowing that sCD22 could promote neuroinflammation and CD22 is expressed in microglia and monocyte that is essential for neuroinflammation, we further to examine if targeting CD22 by antibody could provide therapeutic value in mouse model of neuroinflammation. Previous study showed that administration of anti-murine CD22 via CNS-targeted osmotic infusion improved cognitive functions in aging mice [17]. However, CNS osmotic infusion is less feasible when applied clinically while intravenous or subcutaneous injections are more suitable in clinic. Therefore, we test suciraslimab (SM03), an anti-human CD22 antibody that has already been tested clinically for rheumatoid arthritis in China, in mouse model of neuroinflammation. As suciraslimab only reacts with CD22 of human and non-human primate, but not of murine origin, the therapeutic effect of suciraslimab could not be directly tested in traditional AD mouse model, namely, APP/PS1, 3xTG and 5xFAD mice. Therefore, we generated knockin mice expressing human CD22, but not mouse CD22 with CRISPR design (Fig. S4A). The expression of human CD22 was validated in both blood and spleen tissue (Fig. S4B & C). Next, we evaluated the potential of suciraslimab in resolving neuroinflammation in hCD22 transgenic mice i.c.v. injected with exogenous Aβ (model mice) (Fig. 3A). While the model mice displayed reduced spatial working memory, intravenous administration of suciraslimab, but not IgG1 isotype, significantly restored the spatial working memory as assessed in Y-maze. The improved memory was not a function of locomotion, as evidenced by a similar number of arm entry among groups (Fig. 3B). We further performed RNA seq with the cortices of model mice administered with suciraslimab or IgG1 isotype. Suciraslimab downregulated leukocyte migration genes (Icam1, Vcam1, Cxcl1 and Adam8) and sialic acid binding genes (Sele and Selp) (Fig. 3C-E). RNA seq data from hippocampus also displayed a similar downregulation of leukocyte migration genes and sialic acid binding genes (Fig. S5A-C). These findings recapitulated previous report showing reduced leukocytes infiltration after brain-infusion of anti-murine CD22 antibody in a mouse model of intracerebral hemorrhage [43]. Moreover, GSEA revealed suciraslimab-treatment was negatively enriched with inflammatory signaling, including TNFα signaling via NFκB and interferon α response (Fig. S5D). ELISA analysis on the cortices of suciraslimab-treated model mice further confirmed the anti-inflammatory role of targeting CD22 in reducing the levels of CX3CL1, CCL24 and CCL4 (Fig. 3F). Collectively, these data suggested that suciraslimab reduced Aβ-induced neuroinflammation in mouse models.

Fig. 3

CD22 modulation by suciraslimab alleviates Aβ-induced neuroinflammation in human CD22 transgenic mice. A Schematic diagram of Aβ-induced neuroinflammation model in human CD22 transgenic mice. B Effect of suciraslimab on Aβ-injected model mice in Y-maze test. Alternation: One-way ANOVA, F = 4.724, P = 0.0196. Tukey post hoc test, * P < 0.05. Number of arm entry: One-way ANOVA, F = 0.07, P = 0.93. N = 8–9. C Volcano plot of suciraslimab-treated mouse cortex. N = 3. D Gene ontology analysis of suciraslimab-treated mouse cortex in Biological function. E Gene ontology analysis of suciraslimab-treated mouse cortex in molecular function. F Effect of suciraslimab on chemokine release in mouse brain of model mice. Student t-test, P value as stated in the figure. N = 3. All data are presented as mean ± SEM

Suciraslimab alleviates Aβ-induced inflammation in microglia and human PBMC

Albeit the pro-inflammatory nature of sCD22, membrane CD22 is an inhibitory receptor for B cell receptor to prevent excessive activation in B cells [10]. We tested whether membrane CD22 could exert inhibitory effect on Aβ-induced signaling. HEK293 cell stably-expressing CD22 was constructed and Aβ-induced NFκB inflammatory signaling was examined using a nano luciferase reporter system. Overexpression of CD22 significantly abolished Aβ-induced NFκB inflammatory signaling in HEK293 cells (Fig. 4A). Further, we delineated the mechanism of action of suciraslimab in the context of Aβ-induced neuroinflammation, we tested if suciraslimab suppressed Aβ-induced IL-1β in multiple cellular systems, as IL-1β is upregulated in AD patients [44] and promote Aβ deposition [45]. Indeed, suciraslimab significantly suppressed the level of Aβ-induced IL-1β release in MDMi (Fig. 4B). As NLRP3 activation is the major signaling platform for the secretion of IL-1β in microglia [46], we examined if suciraslimab suppressed Aβ-induced NLRP3-ASC activation. We found that Aβ promoted NLRP3 and ASC aggregation in MDMi as represented with enhanced punta signal with immunostaining. Suciraslimab significantly reduced NLRP3 and ASC punta signal in Aβ-treated MDMi (Fig. 4C). However, Suciraslimab did not suppress Aβ-induced IL-6 and TNFα secretion in MDMi system (Fig. S6B). Further, we examined if suciraslimab also suppressed inflammation in the peripheral system. We treated human PBMC with Aβ to induce IL-1β secretion and we found that suciraslimab significantly suppressed IL-1β release while IgG1 isotype did not (Fig. 4D). Single nucleotide polymorphism in IL-12/IL-23 axis has been associated with AD [47, 48] and anti-IL-12/IL-23 was shown to have therapeutic potential in mouse model of AD [49]. We further tested if suciraslimab could suppress IL-12 and IL-23 in PBMC stimulated with IFNγ and LPS. Indeed, suciraslimab significantly suppressed their secretion upon stimulation in PBMC (Fig. 4E). To further examine the underlying mechanism of IL-12 and IL-23 downregulation by suciraslimab, we examine the expression of TLR4 on monocyte, which is the major cell type for the release of IL-12 and IL-23 upon IFNγ and LPS stimulation. Upon IFNγ and LPS stimulation, we gated CD3-CD19-CD14 + cells and examined the expression level of TLR4 with flow cytometry. We found that suciraslimab significantly reduced the surface expression of TLR4 on CD3-CD19-CD14 + monocyte when compared to IgG1 isotype (Fig. 4F). These data suggested that suciraslimab effectively suppressed multiple pro-inflammatory cytokines, not only in microglia, but also in human PBMC.

Fig. 4

Suciraslimab suppresses Aβ-induced inflammation in microglia and human PBMC. A Effect of CD22 overexpression on Aβ-induced NFκB signaling in HEK293. Two-way ANOVA: CD22 expression, F_(1,20) = 62.97, i < 0.0001; Aβ treatment, F_(4,20) = 16.83, P < 0.0001. Tukey post hoc test, **** P < 0.0001. N = 3. B Effect of suciraslimab on Aβ-induced IL-1β release in MDMi. One-way ANOVA, F = 7.767, P = 0.004. Tuley post hoc test, * P < 0.05, ** P < 0.01. N = 6–7. C Immunofluorescent staining and quantitation of NLRP3 and ASC after Aβ and suciraslimab treatment in MDMi. NLRP3: One-way ANOVA, F = 10.09, P < 0.0001, Tukey post hoc test * P < 0.05, *** P < 0.001; ASC, One-way ANOVA, F = 19.10, P < 0.0001, Tukey post hoc test **** P < 0.0001. N = 6–15. D Effect of suciraslimab on Aβ-induced IL-1β release in human PBMC. One-way ANOVA, F = 6.833, P = 0.0052. Tukey post hoc test, ** P < 0.01, ns = not significant. N = 8. E Effect of suciraslimab on IFNγ + LPS-induced IL-23 and IL-12 release in human PBMC. IL-23: One-way ANOVA, F = 26.93, P = 0.0002. Tukey post hoc test, ** P < 0.01, *** P < 0.001; IL-12: One-way ANOVA, F = 10.21, P = 0.0008. Tukey post hoc test, * P < 0.05, *** P < 0.001. N = 4–8. F Effect of suciraslimab on TLR4 surface expression on monocyte upon IFNγ and LPS activation. Two-tailed paired Student’s t test, P = 0.0293, t = 3.322, df = 4. N = 5 G Effect of suciraslimab on α4 integrin surface expression on T cell of human PBMC. One-way ANOVA, F = 0.7059, P = 0.5131. N = 5. H Effect of suciraslimab on α4 integrin surface expression on B cell of human PBMC. One-way ANOVA, F = 66.02, P < 0.0001. Tukey’s post hoc test, * P < 0.05, **** P < 0.0001. N = 4–5. I Effect of suciraslimab on α4 integrin surface expression on T cell-depleted human PBMC. One-way ANOVA, F = 16.91, P = 0.0009. Tukey’s post hoc test, ** P < 0.01. N = 4. J Effect of suciraslimab on α4 integrin surface expression on monocyte-depleted human PBMC. One-way ANOVA, F = 8.565, P = 0.0083. Tukey’s post hoc test, IgG1 vs. αCD22 Ab, P = 0.1748. N = 4. All data are presented as mean ± SEM

We have previously shown that suciraslimab-treatment in mice down-regulated genes related to leukocyte migration in the brain (Fig. 3). CD22 is well-known to be an adhesion molecule and Epratuzumab (Emab), a clinical stage anti-human CD22 antibody, was reported to downregulated β7 integrin [50], which is essential for lymphocyte migration to Peyer’s patches in the gut. We tested if suciraslimab will affect the expression of α4 integrin protein, which is essential for lymphocyte migration into the brain and is clinically proven as a therapeutic target for neurological disorders, namely multiple sclerosis. We found that suciraslimab treatment effectively downregulated the expression of α4 integrin in B cells, but not T cells (Fig. 5G & H). Moreover, we compared suciraslimab with Emab on α4 integrin reduction in B cells. We found that suciraslimab drives a greater reduction on α4 integrin surface expression in B cells (Fig. 5H). To further delineate the regulation by suciraslimab, we tested if the Fc region of suciraslimab is essential for the function as it was previously shown that anti-CD22 antibody regulated β7 integrin surface expression via trogocytosis, a process required interaction with FcγR on monocyte with the Fc portion of IgG. We introduced LALA or ADE mutation into the Fc region of suciraslimab separately. LALA mutation abolished the interaction between Fc region with FcγR while ADE mutation enhanced their interaction [51]. We found that LALA mutation abolished α4 integrin suppression induced by suciraslimab while ADE mutation enhanced its effect (Fig. 4H). Further, we examined the cell type involved in the downregulation of α4 integrin. With T cell-depleted PBMC, suciraslimab retained its effect on α4 integrin suppression (Fig. 4I). Surprisingly in monocyte-depleted PBMC, suciraslimab failed to reduce α4 integrin on B cells (Fig. 4J).

Fig. 5

Suciraslimab promotes Aβ phagocytosis. A BLI analysis of mouse CD22-Aβ interaction. Association: 600s; Dissociation: 600s. B BLI analysis of human CD22-Aβ interaction. Association: 600s; Dissociation: 600s. C Immunofluorescent staining and quantitation of FITC-Aβ on HEK293 and HEK293-hCD22 cells. Student t-test, ** P < 0.01. N = 20–21, from 3 independent experiments. D Representative image and quantitation of Proximity-ligation assay of CD22-Aβ complex in HMC-3. Student’s t-test, **** P < 0.0001. N = 41, from 3 independent experiments. E Structural alignment of mouse CD22 and human CD22. The structures of both mouse and human CD22 extracellular domain were generated with Alphafold2. Pairwise structural alignment score (TM-score) higher than 0.5 assumes generally proteins aligned of the same fold. F Surface CD22 expression in HMC-3 after suciraslimab treatment. One-way ANOVA, F = 2.892, P = 0.0139. Tukey post hoc test, * P < 0.05. N = 76–83. G Surface suciraslimab binding on HMC-3. One-way ANOVA, F = 125, P < 0.0001. Tukey post hoc test, ** P = 0.002. N = 3. H Effect of suciraslimab on FITC-Aβ phagocytosis in HMC-3. One-way ANOVA, F = 43.92, P < 0.0001. Tukey post hoc test, * P = 0.046, ** P = 0.0018, **** P < 0.0001. N = 3. I Effect of suciraslimab on FITC-Aβ phagocytosis in PMA-differentiated MO3.13. Two-tailed Student’s t test, P = 0.0477, t = 2.482, df = 6

CD22 interacts with amyloid β

Anti-murine CD22 antibody (clone: Cy34.1) enhanced clearance of exogenous Aβ i.c.v. injected into mouse brain [17]. However, the underlying mechanism of how CD22 targeting promote Aβ clearance remains unknown. It is known that Aβ binds to cells through an interaction with cell surface glycolipids or glycoproteins expressing sialic acid moiety [52, 53]. We hypothesized that interactions between CD22 and Aβ exist as CD22 is a well-known glycoprotein expressing sialic acid moiety. We further to examine the interaction between CD22 and Aβ with Bio-layer interferometry assay. Surprisingly, both mouse and human recombinant CD22 were found to interact with Aβ, forming a CD22-Aβ complex with appreciable affinity, as indicated in Bio-Layer Interferometry study (Fig. 5A & B). Next, we tested if CD22 expression on cell surface interacted with Aβ. We generated HEK293 cell over-expression human CD22. Cells were immunostained with FITC-conjugated Aβ. It is shown that HEK293 cells overexpressing human CD22 had significantly enhanced FITC signal when compared to HEK293 cell, indicating interaction between membrane CD22 and Aβ existed (Fig. 5C). Further, we tested if endogenous CD22 expression in human microglia is sufficient for Aβ interaction. Aβ was incubated with HMC-3 during culture and CD22-Aβ interaction was examined with proximity ligation assay. CD22-Aβ interaction was examined with anti-CD22 antibody/anti-Aβ antibody pair or IgG control/anti-Aβ antibody pair. We observed that there were significantly more PLA signal for the anti-CD22 antibody/anti-Aβ antibody pair group than control pair in HMC-3 cell (Fig. 5D). As we observed both mouse and human CD22 extracellular domain (ECD) interacted with Aβ. We performed structural alignment of both ECD and found that they had high TM-score (0.92), indicating that they had high structural similarity (Fig. 5E). Further to break down individual domain to examine the sequence similarity, domain 6 (76.2%) and 7 (82.1%) had higher sequence similarity, suggesting that these 2 domains might be essential for the interaction with Aβ.

We have shown that CD22 interacted with Aβ and CD22 functions as a constitutive internalizing receptor, which could be utilized to shuttle excessive Aβ into microglia for clearance. Suciraslimab, as proven to induce internalization of CD22 in B cells [54], is hypothesized to enhance the internalization of CD22-Aβ complex into microglia for clearance. We tested if suciraslimab could internalize CD22-Aβ complex in microglia. Indeed, suciraslimab-treated HMC-3 showed significantly reduced surface CD22 expression (Fig. 5F), suggesting that surface CD22 is internalized by suciraslimab. Next, we tested if suciraslimab binds onto the surface of HMC-3 could be internalized. Suciraslimab was incubated with HMC-3 on ice for 1 h to allow binding but not internalization. We detected significant signals of suciraslimab on HMC-3 with flow cytometry, indicating that suciraslimab binds onto the surface of HMC-3. With 37^oc incubation for 5 min, we observed significant reduction on the surface level of suciraslimab on HMC-3 (Fig. 5G). Next, we tested if suciraslimab enhanced phagocytosis of FITC-Aβ into HMC-3. With suciraslimab treatment, we observed a modest but significantly enhancement in FITC-Aβ phagocytosis when compared to IgG1 isotype. To test if Fc portion hinders suciraslimab ability to phagocytose Aβ, we treated HMC-3 with suciraslimab scFv to avoid Fc effect. Surprisingly, we observed a more obvious enhancement in phagocytosis when compared to IgG1 isotype and suciraslimab IgG (Fig. 5H). Moreover, suciraslimab also enhanced phagocytosis of FITC-Aβ in PMA-differentiated MO3.13 oligodendrocytic cell line (Fig. 5I). Altogether, these data suggested that suciraslimab not only reduced neuroinflammation, but also promoted Aβ removal by enhancing phagocytosis in microglia and oligodendrocyte in vitro.

Discussion

The importance of CD22 in the function of glia cells and hence its contributions in neurodegenerative diseases such as AD, ALS, and frontotemporal dementia (FTD) is increasingly recognized. Here we report a novel role of sCD22 in promoting inflammation in microglia. We provide the first evidence that sCD22 activates microglia to secrete a plethora of pro-inflammatory cytokines and chemokines in a sialic acid dependent manner. In addition, we present evidence that suciraslimab, a clinical-stage anti-human CD22 antibody, alleviates Aβ-induced neuroinflammation in a human CD22 transgenic mouse model (Fig. 6). Using human PBMC and MDMi to delineate the mechanism of action of suciraslimab, we also established a dual mechanism in which suciraslimab suppresses Aβ-induced inflammation and promotes Aβ clearance simultaneously.

Fig. 6

Working model presenting suciraslimab’s mechanism of action. Anti-inflammatory model: Suciraslimab targets microglia, suppresses Aβ-induced inflammation and reduces the release of CX3CL1, CCL24 and CCL4. Through reducing neuroinflammation, suciraslimab improves memory and cognitive functions in hCD22 transgenic mice

CD22 has recently been implicated and associated with neuro-inflammatory diseases, such as AD, ALS, FTD and NPC. The plasma levels of sCD22 are positively correlated with Aβ burden and worsening of cognitive function in AD patients [16]. Although sCD22 was recently reported to dysregulate lysosomal degradation of cholesterol in microglia in the context of NPC [12], the pathogenic role of sCD22 in the context of neuroinflammatory disease is largely unknown. CD22 is well-documented as a peripheral regulator of adaptive and innate immunity for the control of inflammation and autoimmunity [55]. We hypothesized that sCD22 also participated in neuroinflammation or even drive disease progression of neurodegenerative diseases. Here, we are first to report sCD22 as a driver of neuroinflammation. sCD22 binds to surface receptors of microglia in a sialic acid dependent manner and causes eventual release of pro-inflammatory cytokines and chemokines. Interestingly, sCD22 causes upregulation of a plethora of cytokines and chemokines (Fig. 1). sCD22 might serve as a key driver to change the transcriptomic signature of microglia from homeostatic to inflamed and disease-associated states, thereby causing diseases. AD is a heterogeneous disease where proper patient stratification by re-defining, not only amyloid and tau, but neuroinflammation biomarker might yield more accurate clinical outcomes. CD22 is a promising neuroinflammation biomarker that links peripheral adaptive immunity with central innate neuroinflammation. In particular, other than the reported upregulation in plasma sCD22 level in AD patients, CSF level of sCD22 might be a valuable index of neuroinflammation and could be incorporated into the current [¹⁸F] DPA-714 TSPO PET imaging method [56] for accessing brain neuroinflammation status.

Unlike soluble TREM2, of which the mechanism of its production is well-studied [57, 58], the source of sCD22 in the brain is still unknow. CD22 is highly expressed in oligodendrocyte and moderately expressed in microglia. Microglial expression of CD22 increases upon exposure to Aβ, as per our unpublished observations. The enhanced CD22 expression upon Aβ stimulation could be a physiological negative feedback mechanism to clear Aβ or other neurotoxic aggregates. This hypothesis is supported by upregulated CD22 expression in neurodegenerative diseases characterized with neurotoxic aggregates, such as TDP-43 in ALS and Aβ in AD. Upon chronic activation of Aβ, the upregulated CD22 might fail to clear Aβ, which eventually oligomerize to form fibrils. The elevated membrane CD22 might be cleaved off and act as the source of sCD22 observed in pathological conditions and further aggravate cognitive dysfunction via neuroinflammation observed in this study. In this study, we proposed that CD22 is a constitutive internalizing receptor that binds to Aβ and could be utilized by antibody to enhance Aβ phagocytosis by microglia (Fig. 5). Interestingly, we found that while suciraslimab has a modest but significant enhancement in Aβ in HMC-3, suciraslimab scFv has enhanced phagocytic ability when compared to suciraslimab whole IgG. We reasoned that phagocytosis of Aβ relies on binding of suciraslimab to CD22 to induce internalization. The presence of Fc portion of suciraslimab IgG might hinder binding to CD22 molecules. Suciraslimab IgG might instead bind to FcγR with its Fc portion. Suciraslimab scFv, having no Fc part, will not be affected by FcγR binding, displayed much obvious phagocytic ability in HMC-3 (Fig. 5H). The potential of utilizing CD22 to clear neurotoxic aggregate could be expanded as previous study has reported that anti murine CD22 antibody not only enhances the clearance of Aβ, but also α-synuclein and myelin debris [17].

Despite the encouraging efforts in recent approval of anti-amyloid monoclonal antibodies (mAb), the modest clinical benefits and underlying risk of increasing neuroinflammation, as the clearance of anti-amyloid mAb relies on activating microglia via ligation to FcγR [59,60,61], highlight the need for innovative immunotherapy to address multiple components (amyloid, tau, & neuroinflammation) of disease pathophysiology. In this study, we showed that suciraslimab suppressed both inflammation in the microglia and in PBMC in vitro. More surprisingly, not only modulating B cell activation as reported [54], suciraslimab also regulated T helper cell cytokines IL-12 and IL-23 from human PBMC (Fig. 4E). Co-incidentally, we also observed that suciraslimab downregulated the surface expression of TLR4 on CD14 + monocyte. Further experiments are required to validate if suciraslimab regulates IL-12 and IL-23 release via regulating the surface expression of TLR4. Monocyte-specific control of TLR4 expression might help to delineate its role in IL-12 and IL-23 release and how suciraslimab control IL-12 and IL-23 release. In this study, suciraslimab also controls chemokine levels in the mouse brain. (Fig. 3). We also found that suciraslimab downregulated surface expression of α4 integrin, a migration-related molecules that is clinically relevant to neurological diseases. Natalizumab, a humanized IgG4κ monoclonal antibody that binds to α4 integrin, is approved for the treatment of Crohn’s disease (2008) and multiple sclerosis (2004) [62]. In this study, by using a preliminary PBMC ex vivo model, we have shown that suciraslimab downregulate B cell α4 integrin in a Fc- and monocyte- dependent manner. It would be essential to further examine if suciraslimab-treated B cells have altered homing ability to different parts of the body, namely brain or gut. These will further provide insight onto the mechanism of action of suciraslimab and unleash its possible functions in other indications. To conclude, we present the existence of dual mechanisms where suciraslimab mediates anti-inflammatory response and promotes Aβ removal in a CD22-dependent manner. These findings may pave the way for novel, safer and more efficacious treatment options for AD and other neurodegenerative diseases. However, some interesting questions remain. Though we discovered that sCD22 elicits robust neuroinflammation, the receptor that sCD22 targets remains unknown. sCD22 is found to be upregulated in disease conditions, namely Alzheimer’s disease. However, the mechanism on how sCD22 is generated is not reported. Moreover, we have shown that suciraslimab promotes Aβ phagocytosis by enhancing internalization of CD22-Aβ complex. Other mechanisms for protein aggregate clearance might exist as it is reported that anti-murine CD22 antibody (Cy34.1), not only clear Aβ, but also clear α-synuclein and myelin debris.