IgA- Tertiary Lymphoid Structures Correlates With a Better Prognosis in GADC Patients

Background: The roles of tumor inltrating B lymphocytes (TIBs) and tertiary lymphoid structures (TLSs) in solid tumor genesis and tumor therapy have been recognized by researchers, but the specic formation and effect of the TLSs have not been fully understood. In this study, we used single-cell RNA sequencing, multiple immunouorescence assays, and quantitative digital image analysis to study the formation and structure of TLSs in gastric adenocarcinoma (GADC). Furthermore, the study collected 165 cases of GADC with TLSs and analysis the relationships between TLS formation and clinicopathological characteristics and prognosis of GADC patients were analyzed. Results:The result identied the type of IgA-TLSs which contained higher level of IgA + -B cells in GADC, and the major structures of the IgA-TLSs were determined. We found that immune cells in IgA-TLSs had higher levels of cellular interactions and migration ability. The expression of signal sequence receptor subunit 4 (SSR4) was characterized and found to higher expressed in the IgA-TLSs. Furthermore, IgA-TLSs correlated with age, differentiation, distant metastasis, TNM stage, chemotherapy effect, expression of programmed death-ligand 1, J-chain, and SSR4, and better overall survival. Conclusions: Our research provided the information about CD79A/J-chain B cells in GADC and indicated that IgA-TLSs was associated with better prognosis for GADC patients. in present study we explored the information of CD79A/J-chain B cells in GADC tissue samples, and conducted Gene Ontology (GO) and Cellular Spatial Organization mapper (CSOmap) analysis on the CD79A/J-chain cell cluster. Next, we collected 24 cases of GADC with TLSs to compare structures and densities of tumor-associated TLSs using multiple immunouorescence assays. We also analyzed the relationships between IgA-TLSs formation and clinicopathological characteristics and prognosis in patients with GADC through 165 cases of TMA analysis. Our research provided the information about CD79A/J-chain B cells in GADC and indicated that IgA-TLSs was associated with better prognosis for GADC patients.


Introduction
Gastric cancer (GC) is one of the ve most common tumors worldwide and one of the most frequent malignant tumors of the digestive system in China. [1] More than 80% of patients with GC in China present with middle and advanced stages at the time of diagnosis because of the absence of obvious symptoms. Gastric adenocarcinoma (GADC) is a kind of gastric cancer, which is caused by malignant changes of gastric gland cells. Its incidence rate accounts for 90% of gastric malignant tumors. [1] The growth, invasion and metastasis of GADC is a complex and dynamic process, which involves the genetic abnormalities inherent in the tumor tissue and tumor-immune cell interactions in the local microenvironment. [3,4] Moreover, B cell mediated mucosal immunity and humoral immunity also plays an important role in gastrointestinal tumors. [3,5,6] IgA is the most abundant immunoglobulin in the body and the main force of mucosal immunity against infection. IgA induces B and T cell responses and recruits them to various effector sites, differentiating into plasma cells that can continue to secrete IgA and play a mucosal protective role. [7] The tumor microenvironment (TME) of GADC is a dynamic network and is a key factor affecting the metastasis and promotion or inhibition of immune response against the tumor cells. [8] In ltrating immune cells in the TME, also known as tumor-in ltrating lymphocytes (TILs), mediate the anti-tumor response. [9] In the gastric TME, there is bidirectional regulation between tumor cells and immune cells.
The interaction between the two types of cells can disrupt homeostasis in the body, mobilize internal and external resources of cells, produce TMEs suitable for tumor cell growth, and affect the response of GC cells to GC immunotherapy. [3] Immunotherapy is a major treatment method following surgery, radiotherapy, chemotherapy, and targeted therapy. [10] Currently, effective immunotherapy for GC mainly comprises immune checkpoint inhibitor (ICI) treatment with monoclonal antibodies against programmed cell death-1 or programmed cell death ligand-1. [11,12] Additionally, adoptive immunotherapy with chimeric antigen receptor T cells has shown good e cacy in the treatment of hematological tumors and some solid tumors. [13] To date, strategies for predicting biomarkers and enhancing clinical immunotherapeutic responses have focused on T cells, with less attention has been paid to the role of other immune cell subsets that may contribute to anti-tumor immune responses.
The TME contains not only TILs but also tertiary lymphoid structures (TLSs). [14,15] TLSs, also known as ectopic lymphoid tissue, is a new type of lymphoid tissue found in recent years in areas that are stimulated by certain autoimmune diseases, graft rejection, and chronic in ammatory responses. Its structure and function are very similar to secondary lymphatic organs such as lymph nodes and spleen.
TLSs can be seen in both primary and metastatic tumors, [16] and mainly composed of T and B lymphocytes, follicular dendritic cells (FDC), and mature dendritic cells, [17] which can penetrate into the "hinterland of the enemy" much like a "Trojan horse" and effectively inhibit tumor growth and in ammation. [18] Furthermore, the density of TLSs and the ratio of TLSs to tumor area, as well as the number of tumor-in ltrating B lymphocytes (TIBs) were reported to be higher in patients who were sensitive to ICI treatment than in ICI-insensitive patients. [19] In our previous study, we collected 4 cases of fresh GADC tissue samples and found the potential importance of IgA-mediated humoral immunity in GADC patients with TLSs through single-cell RNA sequencing. [19] To further explore the formation of IgA-TLSs in GADC, in present study we explored the information of CD79A/J-chain B cells in GADC tissue samples, and conducted Gene Ontology (GO) and Cellular Spatial Organization mapper (CSOmap) analysis on the CD79A/J-chain cell cluster. Next, we collected 24 cases of GADC with TLSs to compare structures and densities of tumor-associated TLSs using multiple immuno uorescence assays. We also analyzed the relationships between IgA-TLSs formation and clinicopathological characteristics and prognosis in patients with GADC through 165 cases of TMA analysis. Our research provided the information about CD79A/J-chain B cells in GADC and indicated that IgA-TLSs was associated with better prognosis for GADC patients.

Sample Collection
All sample collection was approved by the Human Research Ethic Committee of Bayannur Hospital and Nanjing First Hospital. 4 cases of GADC tissue samples (named as T1, T2, T3 and T4) with TLSs were immediately collected after resection and dissociated into a single-cell suspension. Single-cell sequencing data came from the previous research data of our team. Please refer to the information of the 4 patients [19] . All para n tissue blocks were received from the department of Pathology at the Bayannur Hospital and Nanjing First Hospital between Jan. 2008 to Dec. 2010, and preparation onto tissue microarray (all samples from 120 GADC patients). The formation of germinal center structures is determined by two experienced pathologist. An effective TLS is de ned as an area of ectopic lymphocyte aggregation greater than a 200 × visual eld (1mm in diameter). Two distinct regions of TLSs are observed under the microscope: the bright region and dark region. Clinical pathology information includes gender, age, differentiation, Tumor Node Metastasis (TNM) stage, chemotherapy effect and TLSs location. TNM stage standard according to American Joint Committen on Cancer (AJCC) Cancer Staging Manual Eighth Edition. [20] We divided the locations of TLSs into tumor center, para-cancerous tissues and the junction between tumor (T) and normal (N) tissues (T/N junction). Overall survival (OS) period was de ned from initial diagnosis to death. Matched normal tissue was obtained from sites displaced at least 5 centimeters from the tumor and was con rmed to lack tumor cells on histopathology review. All patients did not receive any other therapy before surgery, more details about samples please see Table 2. For clustering and dimension reduction analysis, t-distributed stochastic neighbor embedding (tSNE)was used in the algorithm, resolution was set at 0.5, and harmony was used to remove the batch effect. A total of 14,660 cells were retained after screening. Bimod algorithm is adopted in the difference analysis between different clusters, and p value is set as < 0.01, the lter condition log2FC ≥ 0.26. [20] Gene Ontology(GO) Analysis The GO is an internationally standardized functional classi cation system for genes. It provides a dynamically updated standard vocabulary to comprehensively describe the properties of genes and gene products in an organism. There are three ontologies (ontologies) in GO. The molecular function, cellular component and biological process of the gene were described respectively. In this study, only the biological process was analyzed. The basic unit of GO is "term", and each term corresponds to an attribute. Firstly, all the signi cantly differentially expressed genes were mapped to the terms of the GO database, and the number of genes in each term was calculated. Then, hypergeometric test was applied to nd out the GO items that were signi cantly enriched in the signi cantly differentially expressed genes compared with the whole genome background. [22] GGPLOT2 was used to carry out enrichment analysis on the GO results and scatter plot display: Rich factor represents the number of differential genes located in the GO/the total number of genes located in the GO. The greater the Rich factor, the higher the GO enrichment degree. [23] Cellular Spatial Organization Mapper (CSOmap) To reconstruct the cell spatial organization, we analyzed 2,557 ligand-receptor paired in single cell a nity matrix. The ligand-receptor interaction database in CSOmap is based on FANTOM5 (https: // fantom. gsc. riken. Jp / 5 / suppl / Ramilowski_et_al_2015/) for estimating the cell-cell a nity matrix. [24,25] Function Annotation Of The Mammalian Genome (FANTOM) 5 compiled putative ligands from known interacting ligands, orphan Human Plasma Membrane Receptome (HPMR) ligands and from a set of secreted proteins that were not found in the set of known receptors. FANTOM5 compiled putative receptors from known interacting receptors, orphan HPMR receptors and from a set of PM proteins that were not found in the set of known ligands. And then FANTOM5 obtained predicted ligand-receptor pairs by searching for validated protein-protein interactions (Human Protein Reference Database and STRING32 databases) between putative ligands and putative receptors. [25] Immunohistochemistry, Multiple Immuno uorescence Assay and HALO Spatial Analysis Tissue para n sections were incubated in the oven at 65°C for 1h, dewaxed by xylenefor 3 times, gradient alcohol hydration was performed, and antigens were retrieved by microwave. Sections were pretreated and stained with monoclonal antibodies directed against CD79A (SP18, Abcam), J-chain (OTI2B1, Invitrogen), CD4 (EPR6855, Abcam), SSR4 (11655-2-AP, Proreintech), CD8 (SP16, Abcam), CD14 (SP192, Abcam) and PD-L1 (22C3, Dako), DAPI (D3571, Thermo Fisher Scienti c) on a Leica-Bond III/max autostainer platform (Leica Biosystems). Incubated overnight at room temperature (RT), added biotin-labeled secondary antibody at RT for 1h, and adding horseradish peroxidase labeled streptomycin opaletin working solution. Incubated at 37℃ for 30min, and using anti -immunoglobulin -coupled horseradish peroxidase with 3,3'-diaminobenzidine (DAB, OptiView Kit, Roche Diagnostics) as substrate.
Lymphocyte regions on tissue images were manually circled, based on Halo Highplex FL V4.0.4 (ICA Labs, Albuquerque) algorithm, the algorithm was adjusted according to debug the algorithm according to the localization and staining characteristics of different positive cells. The number and proportion of positive cells labeled by antibodies in the whole image and lymphocyte cells region were quantitatively analyzed, and the morphology of each cell was quantitatively analyzed. The area to be analyzed on the tissue image was observed manually, and the in ltration degree of immune cells 300µm inside and outside the tissue boundary was calculated and analyzed. Each interval of 50um was divided into 1 interval. The number of immune cells in each interval, the tissue area of each interval, and the cell density in each interval were calculated, and the histogram was automatically generated.

Statistical Analysis
All data were analyzed by SPSS 23.0 software. Survival curves were generated by the Kaplan-Meier method. The results of Kaplan-Meier method is displayed with HR and P or Cox P-values from a log-rank test. Pearson x 2 test was used to compare the relationship between the expression levels of gene and clinicopathological parameters.s. The independent paired t test was used to compare the mean values of the two groups. P < 0.05 is considered a signi cant difference.

The information of CD79A/J-chain B cells in GADC Tissue Samples
In our previous study, we performed single-cell RNA sequencing to analyze of 49,765 single cells from four GADC patients, which contained a larger number of B cells. In the part of B cell, we found CD79A and J-chain higher expressed. After clustering and dimension reduction analysis, 14,660 CD79A/J-chain B cells were re-clustered into 27 separate subsets through t-SNE algorithm ( Figure 1A). The distribution of CD79A/J-chain B cells in tumor tissue samples is shown in the Figure 1B, and we found that they mainly existed in T1 and T2 ( Figure 1C). Therefore, we choosed CD79A/J-chain B cells cluster from tissue samples to analysis and found that the cluster exhibited related biological process, including classical complement activation, leukocyte migration, and immune responsiveness through gene ontology (GO) analysis ( Figure 1D). After performing enrichment analysis, we determined that the main functions of these cells were related to production of secretory and monomeric IgA immunoglobulin complexes, and the monomeric IgA immunoglobulin complex, and antibacterial humaral response ( Figure 1E).

Correlation between SSR4 and Immune Cells in GADC
Through cluster analysis, the study discovered that the 27 types of sub-clusters had higher gene expression of immunoglobin heavy chain alpha (IGH) A1, IGHA2 and signal receptor subunit 4 (SSR4) genes ( Figure 2A). Therefore, we speculated that the expression of SSR4 gene is related to the function of CD79A/J-chain B cells.
The expression levels of SSR4 were evaluated in immune cells through Gene Expression Pro ling Interactive Analysis (GEPIA) database (http://gepia.cancer-pku.cn/), and the results indicated that SSR4 gene was higher expressed in immune cells, especially in B cells and T cells. Furthermore, the expression levels of SSR4 were signi cantly different between distinct types of B cells in gastric tissue ( Figure 2B).
The study also found signi cant correlations between SSR4 and marker genes of B cell, T cells, tumorassociated macrophage (TAM), such as CD79A, CD3D, IL10, HLA-DPB1, and IFN-g in Tumor IMmune Estimation Resource (TIMER) database (https://cistrome.shinyapps.io/timer/) ( Table 1). These results further con rm that SSR4 is signi cantly related with immune in ltrating cells in GADC, which suggests that SSR4 plays a important role in the immune microenvironment.

Different Structures and Densities of Tumor Associated TLSs
The study collected 24 cases of para n tissue blocks, and each samples with TLSs. Twelve of these tissues contained TLSs that formed germinal centers (bright region and dark region). Multiple immuno uorescence assays detected the morphology and structure of the TLSs in these tissues. We

The information of IgA-TLSs (CD79A+/J-chain+) and the Correlation with Clinicopathological Characteristics in Patients with GADC
To analyze the function of IgA-TLSs, we determined the relationships between IgA-TLS in GADC and clinicopathological parameters. We collected 165 samples from 120 GADC patients that contained TLSs and analyzed protein expression using tissue microarray. Each sample point detected the formation of TLSs by determining the location of CD79A protein and the expression of J-chain. We de ned IgA-TLSs as CD79A expressed in the core of the TLSs and simultaneously detected the expression of J-chain. We also detected the expression levels of PD-L1, and SSR4 proteins and combined analyses with clinicopathological information was performed (Figure 4).

IgA-TLS was Associated with Better Prognosis for GADC Patients
To determine the relationship between IgA-TLSs and overall survival (OS) of GADC patients, Kaplan-Meier survival curve analyses were performed and it was suggested that IgA-TLSs formation was signi cantly associated with a better prognosis for GADC patients (log rank = 13.604, P < 0.001) ( Figure 5).  (Table  3). These results indicated that IgA-TLSs may predict a better treatment outcome and a longer survival prognosis for GADC patients.

Discussion
In recent years, it has been suggested that the TLSs and the TIBs may regulate the e cacy of immunotherapy in several types of tumors, and may closely related to the prognosis of immunotherapy.
Therefore, these possibilities open a new immunotherapy research direction. The roles of TIBs and TLSs in tumorigenesis and immunotherapy have been recognized by researchers, but the speci c mechanisms of action are not fully understood. [26,27] In our study, the results showed that a large number of B cells in GADC tumor tissues expressed CD79A/J-chain gene, and most of these cells were IgA+ plasma cells (high expression of IGHA1 and IGA2 genes). Through GO enrichment analysis, the study also found CD79A/J-chain B cells have complement activation, B cell actication and antibacterial humoral response. These results con rmed that CD79A/Jchain B cells are involved in mucosal and humoral immunity and were mainly IgA-B cells cluster.
Next, we determined the spatial structure of single cells through CSOmap, and found that T1 and T2 samples contained more cells and different types of cellular interactions and migration ability compared with cells in T3 and T4 tissues. Moreover, we observed increased expression of receptors and ligands associated with chemokines, and these data suggests that the immune cells in T1 and T2 samples may have a stronger ability to migrate. Denton et al., con rmed that CXCL13 can recruit immune cells to generate an environment permissive for germinal center formation in the lung. [28] It supports our results that the TLSs contained in T1-2 tissues form germinal centers.
The germinal center is located in the follicular region of secondary lymphatic organs. It is a special structure formed by the aggregation of B cells in the humoral immune response, and it is also a transient and dynamic micro-anatomical structure. [29] Antigen-speci c B cells gather in this particular structure, and then they expand to produce somatic high frequency mutations, antibody type changes and eventually become high a nity B cells for speci c antigens. [30] Through cluster analysis, CD79A/J-chain B cells expressed high levels of the IgA1, IgA2, and SSR4 gene.
We continue to nd that SSR4 plays a important role in the immune microenvironment through GEPIA and TIMER database analysis. IgA is the most abundant antibody produced in mammals. IgA produced in the gastrointestinal mucosa is secretory IgA (SIgA), which is mainly composed of dimers. [31,32] The two monomers are linked by a J-chain protein, forming an important immune protective layer on the mucosal surface. The J-chain is a glycoprotein with a molecular weight of approximately 15 kD and is mainly synthesized by IgA or IgM plasma cells and connects two monomeric units of IgA or IgM. [33] Although J-chain can polymerize IgA and IgM, IgM mainly exists in blood, and SIgA plays an important role in mucosal immunity of the body. Based on this, it is reasonable to speculate that CD79A/J-chain B cells play a role in mucosal immunity.
The SSR4 gene encodes the translocon-associated protein δ (TRAPδ) subunit. The TRAP complex comprises four transmembrane subunits (α, β, γ and δ), which exist in the ER and are involved in protein transport across the ER membrane. Studies suggest that the TRAP complex is involved in the regulation of humoral immunity by guiding the secretion and transport of immunoglobulins. [34] It is reasonable to speculate that the immune cells with high expression of SSR4 may be related to the performance of humoral immunity in human body.
In order to obtain more information about TLSs, we continued to collect 24 cases of GADCs tumor samples with or without germinal centers. We compared the structural differences by multiple immuno uorescence and digital quantitative analysis techniques. And found that CD79A+-B cells and SSR4-immune cells existed mainly in the core of germinal center in the TLSs. Moreover, The interaction between immune cells were involved in the TLSs with germinal centers. These results also support the analysis results of CSOmap prediction model.
B cells are a type of lymphocytes, which enters the peripheral lymphoid tissue after maturation of the fetal liver and bone marrow mature. [35] Subsequently, B cells proliferate and mutate in germinal centers, differentiate into memory B cells and plasma cells, and secrete antibodies to induce humoral immunity and mediate cellular immunity. The germinal center is the site of clonal expansion and a nity maturation of B cells. [30,36] TIBs exists in some tumor tissues and are also an important part of TLSs. At present, there is still some controversy about the role of TIBs in anti-tumor immunity. Some studies showed that some immunosuppressive TIBs subtypes promoted tumor progression; [37] Another study suggested that TIBs promotes tumor immunity and inhibits the growth of tumor cells by producing tumor-speci c antibodies and presenting tumor antigens. [38] The function of TIBs and the formation of TLSs will likely affect the response to immunotherapy in GADC patients.
To understand more about the role of TIBs and TLSs in GADC, we collected 165 tissue samples from 120 patients with GADC who developed TLSs, and each case included complete clinicopathological information, treatment and prognosis information. IHC analyzed results indicated that the formation of IgA-TLSs was correlated with age, differentiation, M stage, TNM stage, chemotherapy effect, PD-L1 expression, and SSR4 expression. Moreover, the presence of IgA-TLS in GADC patients was associated with better OS period.
In this study, we provided the newly insight about TLSs in GADCs and found SSR4 higher expressed in the IgA-TLSs, which may promote a better prognosis for GADC patients. Future studies will be required to con rm that the presence of SSR4 in the core of TLS is predictive of an IgA-immune response and promotes an improved prognosis in patients with GADC.

Declarations
Availability of data and material All data included in this study are available upon request by contacting with the corresponding author.
Ethics approval and consent to participate      Survival curves for GADC patients with the IgA-TLS or the nTLSs using the Kaplan-Meier method and the log-rank test Overall survival curves for patients with the IgA-TLS (green line) or the nTLSs (blue line); Log