A germline variant of human IgG1+ B cell receptor represses colorectal tumorigenesis and progression by shaping the tumor microenvironment

The presence of B lymphocytes in tumor tertiary lymphoid structures (TLSs) is an important prognostic indicator for different types of cancers. However, whether B cell responses in the tumor microenvironment (TME) can be harnessed for immunotherapy is unclear. Here we report that a protective germline variant of human immunoglobulin heavy constant gamma 1 gene (IGHG1) containing a Gly396 to Arg396 substitution (hIgG1-G396R) confers improved survival of colorectal cancer (CRC) patients. These hIgG1-G396R homozygous CRC patients displayed elevated tumor-associated antigen (TAA)-specic IgG1 antibody production and plasma cell inltration into tumors. In murine colon carcinoma models, mice expressing the murine functional homolog IgG2c-Gly400Arg variant (mIgG2c-G400R) also produce higher levels of tumor-specic IgG2c antibodies via enhanced plasma cell differentiation, together with alleviated tumorigenesis and progression. Mechanistically, this variant potentiates TAA-specic antibody-dependent cellular phagocytosis and antigen presentation. Comprehensive immune proling of the TME of CRC patients revealed that hIgG1-G396R prominently promotes broad mobilization of immune cells (IgG1 + plasma cells, CD8 + T cells, CD103 + DCs) and ecient TLS formation, both key components of an anti-tumor microenvironment. Notably, adoptive transfer of tumor-primed B cells with this variant exhibited therapeutic ecacy in murine tumor models, demonstrating clinical potential. These results prompt a prospective investigation of hIgG1-G396R in CRC patients as a biomarker for clinical prognosis and demonstrate that manipulating the functionality of IgG1 + B cells in tumors could improve immunotherapy outcomes. and γ-chain constant region (Cγ1), Cγ2, Cγ3 and Cγ4 in the tumor tissues of CRC patients, measured by RT-qPCR analyses. One of three representative experiments is shown (A-C, G-K). Statistical signicance was determined using an unpaired two-tailed t-test (A-C, F, G, J, K), two-way ANOVA (H), Log-rank test (H) and unpaired one-tailed t-test (L, O). Mean± SEM. NS, not signicant.


Abstract
The presence of B lymphocytes in tumor tertiary lymphoid structures (TLSs) is an important prognostic indicator for different types of cancers. However, whether B cell responses in the tumor microenvironment (TME) can be harnessed for immunotherapy is unclear. Here we report that a protective germline variant of human immunoglobulin heavy constant gamma 1 gene (IGHG1) containing a Gly396 to Arg396 substitution (hIgG1-G396R) confers improved survival of colorectal cancer (CRC) patients. These hIgG1-G396R homozygous CRC patients displayed elevated tumor-associated antigen (TAA)-speci c IgG1 antibody production and plasma cell in ltration into tumors. In murine colon carcinoma models, mice expressing the murine functional homolog IgG2c-Gly400Arg variant (mIgG2c-G400R) also produce higher levels of tumor-speci c IgG2c antibodies via enhanced plasma cell differentiation, together with alleviated tumorigenesis and progression. Mechanistically, this variant potentiates TAA-speci c antibodydependent cellular phagocytosis and antigen presentation. Comprehensive immune pro ling of the TME of CRC patients revealed that hIgG1-G396R prominently promotes broad mobilization of immune cells (IgG1 + plasma cells, CD8 + T cells, CD103 + DCs) and e cient TLS formation, both key components of an anti-tumor microenvironment. Notably, adoptive transfer of tumor-primed B cells with this variant exhibited therapeutic e cacy in murine tumor models, demonstrating clinical potential. These results prompt a prospective investigation of hIgG1-G396R in CRC patients as a biomarker for clinical prognosis and demonstrate that manipulating the functionality of IgG1 + B cells in tumors could improve immunotherapy outcomes.

Main Text
Tumor cells evade clearance by the immune system via several mechanisms, including inhibiting neoantigen presentation and inducing immune silencing in the tumor microenvironment (TME). Extensive studies have focused on effective ways to reinvigorate immunosurveillance by modulating T lymphocyte activation through T cell receptor (TCR) engineering or immune checkpoint blockade 1 . However, the response rate of immunotherapy is still low based on current paradigms [2][3][4] . Although growing numbers of studies suggest credible roles also for B lymphocytes in anti-tumor immunity [5][6][7][8] , it is unclear whether manipulation of B cell antigen receptor (BCR) activation could facilitate immunosurveillance against tumors.
Within the TME, tumor-associated antigen (TAA) can drive tumor-speci c B cells to undergo immune activation, class-switch to IgG1 + memory B cells and differentiation to IgG1 antibody-secreting plasma cells 9 . A single nucleotide polymorphism (SNP) rs117518546 (C > T in allele, Glycine > Arginine in amino acid, hIgG1-G396R) located in human immunoglobulin heavy constant gamma 1 gene (IGHG1) was recently reported to enhance the activation of IgG1-BCR signaling and promote IgG1 + plasma cell polarization 10 . Here, we investigated whether this variant could also modulate cancer progression and survival by focusing on a potential role in colorectal cancer (CRC).
To determine whether the presence of the hIgG1-G396R variant is correlated with clinical colorectal tumorigenesis, we established a cohort containing 1006 CRC patients with up to 100-month survival clinical records from Peking University People's Hospital in Beijing China (Extended Data Fig. 1, Table 1).
To determine the mechanism of action of hIgG1-G396R in inhibiting CRC tumorigenesis, we generated knock-in mice harboring the homologous mutation at the cytoplasmic tail of endogenous murine IgG2c (denoted as mIgG2c-G400R) in consideration of murine IgG2c sharing a conserved tail sequence and functional equivalence to human IgG1 11 (Extended Data Fig. 4A, B). In a transplantation tumor model using syngeneic MC38 murine colon carcinoma cells, mIgG2c-G400R mice showed signi cantly slower progression of inoculated MC38 tumor cells compared to the WT control mice (Fig. 1D, E). In order to more closely simulate colon tumorigenesis in humans, we introduced the azoxymethane (AOM) / dextran sodium sulfate (DSS) colitis-associated carcinoma (CAC) model to our study, which has been widely used for the induction of colitis-dependent neoplasia and primary cancer in mice (Extended Data Fig. 4C) 12,13 . In mIgG2c-G400R mice, the AOM-DSS treatment reduced weight loss (Fig. 1F, Extended Data Fig. 4D), decreased the disease activity index (Fig. 1G), alleviated phenotypes of tumor-associated in ammation Given that the hIgG1-G396R variant enhances IgG1 + B cells activation and IgG1 producing plasma cell differentiation 10 , and that human IgG1 antibodies are important modulators in immunosurveillance 14,15 , we investigated whether the hIgG1-G396R variant differentially regulates the production of tumor-speci c IgG1 antibodies. In the MC38 transplantation model, which presents a broad TAA pool 16 , mIgG2c-G400R mice generated elevated amounts of MC38 cell-speci c IgG2c antibodies compared to WT mice ( Fig. 2A). Tumor-primed B cells isolated from the tumor-draining lymph nodes (TDLNs) of mice could secret large amounts of IgG under the in vitro re-stimulation with tumor cells, comparing with unprimed B cells (Extended Data Fig. 5A) 17 . B cells from the TDLNs of mIgG2c-G400R mice secreted signi cantly higher amounts of IgG2c upon re-stimulation with irradiated MC38 cells than B cells from TDLNs of WT mice (Fig. 2B). Consistently, the levels of intestinal IgG2c antibodies were signi cantly increased in mIgG2c-G400R mice upon AOM/DSS induction, while the levels of intestinal IgA, IgG1, IgG2b and IgG3 antibodies in mIgG2c-G400R mice were comparable with WT mice (Fig. 2C).
To comprehensively characterize the TAA-speci c antibodies, we used microarrays to pro le tumor speci c antibody levels against a panel of 51 TAAs. The results further validated the burst of sera IgG2c antibody production with broad TAA speci city, but not sera IgM, IgG1 or IgG2b antibodies in mIgG2c-G400R CAC mice (Fig. 2D, Extended Data Fig. 5B). Similarly, in CRC patients, the hIgG1-G396R homozygotes produced elevated levels of TAA-speci c IgG1 antibodies (Fig. 2E), such as anti-AFP, anti-CEA and anti-CA125 antibodies (Fig. 2F), while this stimulatory effect was not signi cant for IgM antibodies. Previous studies have indicated human IgG3 as a negative prognostic marker for cancer patients, which is the opposite of human IgG1 18,19 . Notably, we identi ed an increase in the ratio of IgG1 to IgG3 TAA-speci c antibodies in hIgG1-G396R homozygous patients (Extended Data Fig. 5C, D). The boosted production of TAA-speci c IgG2c antibody was also detected in the colon explants of CACinduced mice by TAA microarrays (Extended Data Fig. 5E). Together, these results demonstrate that the hIgG1-G396R variant boosts the production of TAA-speci c IgG1 antibodies in human CRC patients, similar to the mIgG2c-G400R variant in tumor-bearing mice.
To investigate the role of this variant in immunosurveillance mediated by TAA-speci c IgG1 antibody upregulation, we utilized membrane-bound ovalbumin (OVA)-expressing MC38 tumor cells (denoted as MC38-mOVA) and B16F10-mOVA (denoted as B16-mOVA) (Extended Data Fig. 5F) to examine the effect of OVA priming on anti-tumor immunity in the mIgG2c-G400R mice. To establish a loss-of-function control in our experimental system, we constructed a new mIgG2c-tailless mouse with a truncated cytoplasmic tail of the membrane-bound mIgG2c heavy chain, which blocks the mIgG-tail signaling transduction for enhanced IgG + B cell activation and differentiation 20,21 . After immunization of OVA antigen combined with Th1-prone adjuvant 22 , which favors IgG2c antibody production, OVA-speci c IgG2c antibody production signi cantly increased in mIgG2c-G400R mice while decreasing in mIgG2ctailless mice compared to WT mice (Fig. 2G). Notably, tumor growth was clearly inhibited in mIgG2c-G400R mice during the subsequent MC38-mOVA challenge, whereas it was exacerbated in mIgG2ctailless mice compared with WT mice (Fig. 2H, Extended Data Fig. 5G). Consistently, in the B16-mOVA tumor cell inoculation model, mIgG2c-G400R mice exhibited slower tumor progression and prolonged survival compared to WT mice, while the bene ts were diminished in mIgG2c-tailless mice (Fig. 2H). Further linear regression tting analyses indicated that OVA-speci c IgG2c antibody levels negatively correlated with MC38-mOVA tumor size and positively correlated with survival time in the B16-mOVA transplantation tumor model (Fig. 2I).
Mechanistically, we detected higher percentages of OVA-speci c IgG2c + germinal center (GC) B cells and memory B cells, as well as an increase of plasma cells in the spleen of mIgG2c-G400R mice compared to WT mice after OVA priming ( from hIgG1-G396R homozygous CRC patients were consistently detected by IHC staining, while there were comparable numbers of total B cells. Speci cally, we detected higher levels of immunoglobulin heavy constant gamma 1 (Cγ1) transcripts in the tumor tissues from hIgG1-G396R homozygous CRC patients, which is a hallmark of elevated IgG1 expression in the TME 23 , while the levels of Cγ2, Cγ3, Cγ4 and immunoglobulin heavy constant alpha (Cα) transcripts were comparative (Fig. 2O). Thus, mIgG2c-G400R alleviates mice from colorectal tumor progression by promoting TAAspeci c IgG2c producing plasma cell differentiation and subsequent IgG2c antibody production within the TME.
The interactions between immune system and tumor are governed by a complex network of cell-cell interactions. Thus, we considered that the hIgG1-G396R variant may exert a pleiotropic immunomodulatory role by also affecting interrelated immune cells for effective immunosurveillance. To test this, we analyzed the effect of the variant on in ltration of both innate and adaptive immune cells in the TME, including T cells, natural killer (NK) cells, macrophages and DCs. In the MC38-mOVA tumor model, mIgG2c-G400R mice exhibited increased amounts of CD8 + T cells within subcutaneous tumor tissues (Fig. 3A), and especially increased CD44 high CD62L low CD8 + effector T cells, with elevated levels of cytolytic granzyme B and interferon-γ (IFN-γ) production ( Fig. 3B, C). This indicates that tumorin ltrating T cells from MC38-inoculated mIgG2C-G400R mice maintain an optimal functionality, which may help to limit tumor progression. Notably, DCs, especially CD103 + cDC1, which are the key antigen presenting cells transporting antigens to lymphoid structure and priming tumor-speci c CD8 + T cells 24 , were also elevated in mIgG2c-G400R mice compared to WT mice (Fig. 3D). The MHC-I presentation of OVA epitopes by CD103 + cDC1 had a moderate increase in the tumor tissues from mIgG2c-G400R mice (Fig. 3E). The amounts of tumor-in ltrating macrophages, NK cells and NKT cells displayed no signi cant differences (Extended Data Fig. 6C). Remarkably, IHC staining revealed that the colon tumor specimens from CAC-induced mIgG2c-G400R mice exhibited dramatically increased in ltration speci cally of CD8 + T cells compared with those from WT mice (Fig. 3F, Extended Data Fig. 7A, B).
Tertiary lymphoid structures (TLSs) were recently identi ed as important indicators of favorable outcomes of immunotherapies [5][6][7] . B cells and DCs have been shown to be the major initiators of TLS formation, and together with T cells are key constituents of TLS 25 . In addition, plasma cells and antibodies are positively associated with TLS formation. IHC staining revealed a trend of more and larger TLSs formation within the colon tumors of mIgG2c-G400R mice (Fig. 3G). The elevated immune cell in ltration and TLS formation were largely recapitulated in the tumor tissues of CRC patients: hIgG1-G396R homozygotes had prominently increased tumor in ltrating CD8 + T cells accompanied with an accumulation of S100 + DCs compared with other genotypes (Fig. 3H, Extended Data Fig. 7C, D). There was also a mild but consistent increase of TLSs in the tumor specimens from hIgG1-G396R homozygotes in comparison with WT (Fig. 3I). The transcriptional levels of ve chemokines (CXCR5, CXCL12, CXCL13, CCL19, CCL21) that have been reported to be involved in the formation of TLS 26 , were prominently elevated in both the tumor specimens from mIgG2c-G400R mice and hIgG1-G396R homozygous CRC patients (Fig. 3J). Collectively, these ndings suggest that the hIgG1-G396R variant potentiates anti-tumor responses also by invigorating CD8 + effector T cell priming and activation by DCs within TLSs.
Single cell RNA sequencing (scRNA-seq) is widely used to pro le immune cells in organs and tissues.
Thus, we re-analyzed the proportions of immune cell subsets in the TME of 18 CRC patients using the scRNA-seq results from our recent studies 27,28 . Both patient genome sequencing and single B cell RNAseq data identi ed 11 heterozygotes, 6 WT and only one hIgG1-G396R homozygote within these 18 CRC patients, so we could only compare immune cell pro les within the TME from 12 hIgG1-G396R carriers (CT + TT) to those from 6 non-carriers (CC) (Extended Data Table 5). Nevertheless, in hIgG1-G396R carriers, there were increased proportions of IgG + plasma cells, CXCR5 + T follicular helper cells, and CD6 + tumor-resident memory T cells, and decreased proportions of CTLA4 + regulatory T cells (Fig. 3K, L, Extended Data Fig. 7E). Moreover, the proportions of LAYN + exhausted T cells were also decreased in hIgG1-G396R carriers, implying that the variant may be bene cial to ameliorate T cell exhaustion (Fig. 3L). All these results indicate that the hIgG1-G396R variant favors the formation of an anti-tumor microenvironment.
Human IgG1, or the functional equivalent murine IgG2c, can enhance the anti-tumor activity via antibodydependent cellular phagocytosis (ADCP) and antibody-dependent cellular cytotoxicity (ADCC) 29,30 . To determine whether this mechanism is also triggered by the upregulation of TAA antibodies in the carriers of this variant, we co-incubated necrotic exogenous antigen-expressing tumor cells of either LLC-M2e (In uenza virus M2 ectodomain, in three tandem repeats) or MC38-mOVA, with in vitro differentiated phagocytes in the presence of either anti-M2e IgG2c antibodies or OVA antiserum. We then calculated the e ciency of phagocytosis by ow cytometry and confocal uorescence imaging. Indeed, anti-M2e IgG2c antibodies induced e cient phagocytosis of LLC-M2e cells by bone marrow derived macrophages (BMDMs) in a dose-dependent manner (Fig. 4A, Extended Data Fig. 7F). More importantly, OVA antiserum or further puri ed IgG from OVA-immunized mIgG2c-G400R mice with MC38-mOVA induction showed more potent effects on BMDM activation and phagocytosis towards antibody targeted tumor cells (Fig. 4B, Extended Data Fig. 7G). The enhanced ADCP activity of BMDMs largely relied on dominant TAA speci c IgG2c antibodies in the serum on account of similar ADCP effects with puri ed IgG antibodies ( Fig. 4C-E). Considering tumor antigen presentation by DCs is crucial to initiate T-cell-mediated protective immunity, we investigated the impact of mIgG2c-G400R on antibody-mediated tumor antigen uptake by FLT3L-DCs 31,32 . Similarly, FLT3L-DCs showed dose-dependent phagocytosis of antibody-coated tumor antigens examined by ow cytometry (Fig. 4F, Extended Data Fig. 7F, G) and confocal uorescence microscopy (Fig. 4G). We found that puri ed IgG antibodies from OVA-immunized mIgG2c-G400R mice enhanced the antigen uptake activities of FLT3L-DCs (Fig. 4H, I), which in turn promoted OT-I CD8 + T cell activation and proliferation (Fig. 4J, K). Collectively, this variant strengthens anti-tumor effects in part by enhancing ADCP and subsequent antigen presentation.
The enhanced anti-tumor immunity in mIgG2c-G400R mice prompted us to assess whether adoptive transfer of tumor-speci c B cells is a potential and feasible immunotherapy. Thus, we adoptively transferred MC38-primed B cells to B-cell-de cient mice (µMT) followed by MC38 tumor inoculation. Clearly, tumor primed B cells from mIgG2c-G400R mice conferred improved protection from tumor progression (Fig. 4L). We next assessed the roles of tumor-speci c IgG on anti-tumor response by IgG reinfusion to µMT mice followed by MC38 tumor inoculation. µMT mice with IgG reinfusion showed signi cantly decreased MC38 tumor progression compared with untreated µMT mice (Fig. 4M). Furthermore, µMT mice showed much slower tumor growth after administration of IgG puri ed from MC38-inoculated mIgG2c-G400R mice (Fig. 4M). These results demonstrate the potential therapeutic effect of tumor-primed highly reactive B cells and the feasibility of BCR manipulation in future immunotherapy.
Although the impacts of B cells on tumor progression are still somewhat controversial based on tumor types and tumor stages in both human patients and murine disease model studies 33,34 , our study here uncovers the roles of IgG1 + B cells in antitumor immunity. A protective germline variant, hIgG1-G396R, which enhances IgG1 + B cell activation and differentiation to plasma cells by potentiating the phosphorylation of IgG1 immunoglobulin tail tyrosine motif, fuels immune responses against tumors in at least four ways: (i) elevated and dominant tumor-speci c IgG1 versus IgG3 production, which promotes avidity of macrophages, DCs and NK cells towards IgG1 bound on tumors 35,36 ; (ii) improved ADCP by macrophages and DCs; (iii) enhanced presentation of neoantigens to potentiate effector CD8 + T cell activation; and (iv) supports functional TLS formation, in which switched memory B cells are enriched in responders and synergize with killer T cells for effective tumor cell targeting. Concurrently, we expect that B cells might also play additional roles. For example, activated tumor-in ltrating B cells might secrete cytokines such as IFN-γ and tumor necrosis factor-alpha (TNF-α) to ne-tune the immune context in tumors 37 . In addition, hIgG1-G396R might alter the a nity of antibodies targeting neoantigens, and in turn block immune evasion of tumor cells. Future studies will be important to explore these phenomena and other potential mechanisms, and determine whether hIgG1-G396R has similar effects on survival and progression in other cancer types. Most importantly, the favorable antitumor outcomes after adoptive transfer of highly reactive B cells in mice highlight the possibility of combined adoptive T cell therapies with B cells. Thus, our work suggests that IgG1 + B cells might serve as tuners for orchestrating broad and robust immunosurveillance and provide a notion that harnessing human IgG1 + B cells in the TME could be a powerful strategy for cancer immunotherapy.

Materials And Methods
CRC Patients, healthy controls and ethics 1006 colorectal cancer (CRC) patients with histological veri cation by Peking University People's Hospital and 583 healthy controls (HCs) were informed with for-in vitro study-purpose consents. After surgery, the CRC samples and tumor-adjacent tissues were immediately collected and separated as aliquots in cryogenic vials. All tissues were preserved at -80°C before further treatments and assays. Basic features
Cell culture and transfection MC38, LLC, B16F10 and HEK 293T cells were purchased from national Biomedical Cell Resource (BMCR, China). Generally, cells were cultured in complete DMEM medium containing 10% heat-inactivated fetal bovine serum, 2 mM L-Glutamine and 100 U/ml penicillin/streptomycin. Retrovirus was constructed on the PHAGE backbone. MC38 and B16F10 cells were infected by retrovirus to stably express both uorescent protein mCherry and exogenous membrane-bound OVA (MC38-mOVA, B16-mOVA). LLC cells were infected by retrovirus to stably express both uorescent protein mCherry and exogenous membranebound m2e (in uenza virus M2 extracellular domain) in three tandem repeats (LLC-m2e). Retrovirus infected tumor cells were sorted through Aria III basing on the expression levels of mCherry.

Mice
Rag1 -/-( B6.129S7-Rag1 tm1Mom /J, 002216) and μMT (B6.129S2-Igh-6 tm1Cgn /J, 002288) mice were obtained from professor Dr. Hai Qi (Tsinghua University). mIgG2c-tailless and mIgG2c-G400R genetically modi ed mice were generated by CRISPR/Cas9 based gene manipulation in wild type C57BL/6 background, as described in detail below. Mice were maintained under speci c-pathogen free conditions in the animal facility of Tsinghua University. Mice experiments were performed according to the governmental and institutional guidelines to guarantee animal welfare. All experimental studies were approved with an assurance identi cation number: 15-LWL2 and 19-LWL1 by the Institutional Animal Care and Use Committee (IACUC) of Tsinghua University.
Mice immunization and ELISA 6-week-old gender-matched mIgG2c-tailless, WT and mIgG2c-G400R mice were injected with 100 μg OVA (Sigma-Aldrich) in combination with 100μg TH-Z93 adjuvant through foot pad 2 . Serum samples were collected from tail vein on day 0 and weekly after immunization to quantify OVA-speci c antibody titers.
Sera samples were stored at -80℃ before further processing. To detect antibody levels, 2 μg/ml biotinconjugated goat anti-mouse IgG or 2 μg/ml goat anti-mouse IgA was coated on the ELISA plate overnight at 4ºC to capture antibodies. ELISA plates were blocked with 5% skim milk in PBS at 37℃ for 1 hour and then incubated with serum, colon explants culture medium or control culture medium for 2 h. Peroxidaseconjugated immunoglobulin subclass (IgG, IgG1, IgG2b, IgG2c, IgG3, IgA)-speci c antibodies were further used for nal detection. To detect OVA-speci c antibodies, 2 μg/ml OVA in PBS was coated on the ELISA plate.

Tumor growth and treatment
For MC38 cell inoculation-derived tumor model, 3×10 5 MC38 cells were re-suspended in 100 μl sterile PBS and injected subcutaneously into the right ank of mice. For MC38-mOVA cell inoculation-derived tumor model, 2×10 6 MC38-mOVA cells were also re-suspended in sterile PBS and injected subcutaneously into the right ank of mice at week 6 after recall immunization with OVA. For B16-mOVA cell inoculationderived tumor model, 2×10 6 B16-mOVA cells in sterile PBS were injected intravenously into mice at week 6 after recall immunization with OVA. Tumor growth was measured every two days, and tumor size was calculated by length (a) and width (b): tumor size = 0.52×ab 2 . Mice were sacri ced when tumor size reached to 2000 mm 3 .
AOM/DSS induced colitis-associated carcinoma (CAC) model The AOM/DSS induced CAC model has been described previously 3,4 . Brie y, 10-week-old male WT or mIgG2c-G400R mice were injected intraperitoneally with a single dose of organotropic carcinogen AOM (10 mg/kg of body weight) on day 0. Then mice were treated with 2.5% DSS (molecular weight 35-50 kDa) in drinking water for 7 days in each turn (day 1-7, day 22-28, day 43-49), followed by recovery for 14 days with normal drinking water. Mice were monitored for weight loss twice a week. The serum samples were isolated at indicated time points in each gure legends. Mice were euthanized at week 16 after induction and the colons were isolated to assess the tumor generation.
In vivo imaging of intestinal cancer-associated in ammation After iso urane supported anesthesia, mice were intraperitoneally injected with 25 mg/kg L-012 solution (Wako Chemicals). Bioluminescent images of each mouse were obtained under iso urane anesthesia using an IVIS Spectrum system (Perkin Elmer). For quantitative analyses, the Living Image software was utilized to calculate the intensity of bioluminescent signals at standardized regions of interest (ROIs) for each mouse.

In vitro culture of colon explants
The levels of IgG subclasses and IgA in colon explants were measured by ELISA. The method of colon explants in vitro culture was described previously 5 . Brie y, the fresh colon was collected from AOM-DSS induced mice and cut into pieces approximately 1 cm long. Then the colon tissue was washed vigorously in sterile PBS for three times, and in turn incubated in complete DMEM medium for 24 hours. To exclude unnecessary variables, colon pieces were also weighed before incubation, and colon pieces harboring equivalent wet weight were included into the in vitro culture procedure.

Hematoxylin and Eosin (H&E) staining and immunohistochemical (IHC) staining
After euthanasia, colon specimens were immediately isolated from CAC-induced mice and xed in 10% formalin solution overnight, and colon specimens were subsequently embedded in para n. Standard hematoxylin-eosin (H&E) staining was used to evaluate the pathological severity of colon tissues. For IHC analyses, slides were incubated with indicated primary antibodies in PBS containing 5% BSA overnight at 4°C. Then, streptavidin-HRP was added, and nally the slides were stained with both DAB staining kit and Hematoxylin staining solution (ZSGB-Bio). The slides were scanned by Axio Scan Z1 (Zeiss) and all the images were analyzed via Zen imaging software (Zeiss). TAA-speci c IgM, IgG1, IgG2b and IgG2c subclasses in the serum samples from CAC-induced mice and TAA-speci c IgM, IgG1 and IgG3 subclasses in the plasma samples from CAC patients were measured.
The serum samples and plasma samples were diluted 50 times and the secondary antibodies were diluted 1000 times.

RNA extraction and RT-qPCR
After iso urane supported anesthesia, fresh colon specimens were immediately isolated and ground with Tissue Grinder G50 (Coyote Bioscience) to get colon tissue homogenate. HiPure Total RNA Mini Kit was utilized to extract total RNA following manufacturer's protocols. cDNA was then synthesized through reverse transcription according to manufacturer's protocol and examined by quantitative PCR analyses, and each group was detected in triplicate. All the qPCR primers are listed in detail as an online table.

Flow cytometry
Murine primary cells were separated from the spleen, bone marrow, lymph node, lamina propria, subcutaneous tumor tissues and maintained as single cell suspensions in MACS buffer. Single cell suspensions were pre-incubated with anti-mouse CD16/CD32 (BD bioscience) on ice for 30 min to block Fcγ receptors, and then cells were stained with speci c uorescent dye-conjugated cell surface marker antibodies. Cells were also stained with either propidium iodide (PI) or DAPI to exclude the dead cells.
After washing, cells were re-suspended in MACS buffer and analyzed by either LSRII or Symphony A5 (BD). Cell sorting was performed with the Aria III (BD). All data was further processed with FlowJo (V10) software. For IgG2c + plasma cell staining, primary cells were rst blocked and stained with uorochromeconjugated rat anti-mouse B220 and CD138. After washing with MACS buffer, cells were xed and permeabilized. Subsequently, cells were stained with Alexa Fluor 647-conjugated Fab fragment antimouse IgG2c, Fc speci c (Jackson ImmunoResearch) on ice for 30 min before further FACS processing.
Re-stimulation of tumor-primed B cells 8 B cells puri ed from TDLNs were activated in RPMI-1640 medium with 10 μg/ml LPS for 2 days. Activated B cells were used for immune function analysis. Then the B cells were stimulated by irradiated MC38 cells. After 24 hours, the levels of IgG subclasses secretion in supernatants were detected by ELISA.

Antibody-dependent cellular phagocytosis (ADCP) of tumor cells by FLT3L-DCs and BMDMs
Primary cells were isolated from murine bone marrow and were then re-suspended into proper density. For FLT3L-DCs induction, primary cells were cultured in complete IMDM medium together with 100 ng/ml t3l cytokine for 9-10 days. For BMDMs induction, primary cells were cultured in complete RPMI-1640 medium together with 20 ng/ml M-CSF cytokine for 6 days 9 .
MC38-mOVA cells or LLC-m2e cells were prepared into necrotic cells (at least 3 frozen-thaw cycles) and co-cultured with either BMDMs or FLT3L-DCs as effector cells at a 1:1 ratio in the presence of speci c antibody or antiserum. After co-culture, both ow cytometry and confocal uorescence imaging methods were applied to distinguish tumor cells with phagocytic cells and calculated the e ciency of phagocytosis After co-culture, tumor cells pulsed FLT3L-DCs (following methods mentioned above) were evaluated for their ability to prime OVA-speci c OT-I CD8 + T cells. Puri ed OT-I CD8 + T Cells were stained by CTV following the instructions of user manual. DCs were sorted and co-cultured with OT-I CD8 + T cells at a 1:10 ratio. After either 48h or 72h, cells were collected and stained with CD8 and PI. The levels of proliferation were indicated by CTV dilution by LSRII. DCs were able to cross-prime OT-I T cells to a greater extent mediating by antibody, which was clari ed by increased percentages of proliferating OT-I CD8 + T cells. On day 3, Cytometric bead array (CBA) cytokine kit was applied to evaluate IFN-γ secretion by CD8 + T cells in supernatant following the manufacturer's instructions 10 .
Olympus FLUOVIEW FV1000 confocal laser scanning microscope equipped with 4 lasers (405 nm, 473 nm, 557 nm, and 635 nm) for uorescence excitation, 2 photomultiplier tubes (PMTs) for uorescence detection, and a 10× objective lens was utilized. Images were processed and analyzed by Image Pro Plus software (Media Cybernetics).

Statistical analysis
Actuarial analyses on OS and PFS were performed basing on the detailed follow-up information. Kaplan-Meier analyses were carried out to build the survival curves, and log-rank tests were employed to assess the signi cance of the difference. R software v4.0.2 was used to analyze the probability of cumulative 5-year and 100-month survival rates. Student's t-test or Wilcoxon signed-rank test was used to explore quantitative variables as appropriateness. Pearson's χ 2 test or Fisher's exact test were used to explore categorical variables as appropriateness. Univariable and multivariable COX regression analyses were performed to identify prognostic factors. The Pearson χ 2 test or binary logistic regression analysis were used for statistical analysis of allelic and genotypic differences between HCs and CRC patients. The