Modeling human gastric cancers in immunocompetent mice

Gastric cancer (GC) is a major cause of cancer-related mortality worldwide. GC is determined by multiple (epi)genetic and environmental factors; can occur at distinct anatomic positions of the stomach; and displays high heterogeneity, with different cellular origins and diverse histological and molecular features. This heterogeneity has hindered efforts to fully understand the pathology of GC and develop efficient therapeutics. In the past decade, great progress has been made in the study of GC, particularly in molecular subtyping, investigation of the immune microenvironment, and defining the evolutionary path and dynamics. Preclinical mouse models, particularly immunocompetent models that mimic the cellular and molecular features of human GC, in combination with organoid culture and clinical studies, have provided powerful tools for elucidating the molecular and cellular mechanisms underlying GC pathology and immune evasion, and the development of novel therapeutic strategies. Herein, we first briefly introduce current progress and challenges in GC study and subsequently summarize immunocompetent GC mouse models, emphasizing the potential application of genetically engineered mouse models in antitumor immunity and immunotherapy studies.


Introduction
Gastric cancer (GC) is a major cause of cancer-related mortality worldwide 1,2 .To date, surgery and radio-chemotherapy remain the major clinical treatments for GC.However, these treatments are frequently challenged by patients presenting in advanced or metastatic disease stages, because of a lack of early diagnosis markers 3 .Recently, immunotherapies such as anti-PD1/PD-L1 4,5 and targeted therapy (anti-Her2/ Claudin18.2) 6,7 have entered clinical trials and have been used as tumor treatments.However, many or even most patients with GC do not respond well to these treatments, thus emphasizing the highly heterogeneous nature of GC and the urgent need for in-depth understanding of GC pathology.Heterogeneity remains a major barrier to GC management 8 .
The first aspect of GC heterogeneity is the ever-refined subtyping.Traditional Lauren classification of GC includes intestinal-type, diffuse-type, and mixed-type GC 9 .Recent studies have comprehensively characterized the molecular features of gastric adenocarcinoma.In a milestone in this regard, The Cancer Genome Atlas project, at the molecular level, has revealed 295 primary GCs and defined 4 GC subtypes including Epstein-Barr virus-positive, microsatellite instability, chromosomal instability and genomic stability 10 .Single cell RNA sequencing (scRNA-seq) has been applied to decipher the cellular heterogeneity of the tumor microenvironment in primary and metastatic lesions of patients with GC [11][12][13][14][15][16][17][18][19][20][21][22][23] .ScRNA-seq of primary and peritoneal carcinomatosis cells from patients with GC has demonstrated that the diversity in tumor cell lineage/ state compositions is a key contributor to intratumoral heterogeneity 11,12 .These studies have identified a group of genes associated with differentiation and prognosis, and showing high diversity within and between tumors.Some subgroups show different degrees of differentiation, consistently with the histopathological features of the Lauren subtype.Two newly identified subgroups show unique transcriptional profiles: one expressing master cell markers and Wnt/β-catenin pathway signature genes, and the other expressing immune-related signature genes associated with Epstein-Barr virus infection.Despite progress in understanding of the molecular and cellular mechanisms of GC pathology, the underlying clonal evolution dynamics and cellular malignant transformation of human GC remain to be fully elucidated.
The second aspect of GC heterogeneity includes cell origin and genetic mutations.The homeostasis of the gastric epithelia, which is directly exposed to food intake and gastric acid, can be easily disrupted.As a countermeasure, gastric epithelial cells are continually renewed to maintain gastric gland structure and function.The gastric gland is composed primarily of pit cells, neck cells, parietal cells, isthmus stem cells, reserve stem cells, chief cells, and a small number of endocrine cells and tuft cells 24 .Gastric adenocarcinomas can originate from both stem cells and terminal differentiated cells located at the cardia, corpus, and antrum of the stomach.Furthermore, genetic mutations such as inactivation of tumor suppressors, including RNF43 25 , TP53 26,27 , and ROHA 10,28 , as well as activation of oncogenic Kras 29 and YAP 30 , synergistically initiate and drive the tumor evolution of GC.Among them, the Hippo signaling pathway has been extensively investigated as a major driving force of both gastric tumorigenesis and acquired drug resistance 31,32 .Targeting recovery of Hippo activity, such as with SHAP 33 and SAIP-1/2 34 peptide agonists or chemical agonists 35 , is a promising strategy to curb GC.However, the specific cellular origins and genetic drivers of human GC remain to be clarified, thus hindering the development of precision medicine and targeted therapies.
The third aspect of GC heterogeneity involves the dynamic remodeling of the immune microenvironment associated with disease progression and treatment.A comprehensive single-cell atlas characterizing the microenvironment across various stages of GC progression, from precancerous lesions to metastatic tumors, has identified 6 ecotypes associated with the phenotypic progression and outcomes of GC 14 .In particular, IgA + plasma cells accumulate in precancerous lesions, whereas immunosuppressive myeloid and SDC2 + cancer-associated fibroblasts dominate late-stage GC 14 .In addition, scRNA-seq of GC biopsy samples has revealed elevated plasma cell proportions in diffuse-type GC, in agreement with the upregulation of KLF2 expression in epithelial cells mediating plasma cell recruitment 15 .Using paired pre-and on-treatment samples during standard frontline chemotherapy, Kim et al. 16 have identified chemotherapy-induced NK cell infiltration, macrophage repolarization, and increased antigen presentation among responders.In contrast, the nonresponders showed increased LAG3 expression and decreased dendritic cell abundance, thus suggesting remodeling of the tumor micro environment during chemotherapy response and resistance.However, how the immune microenvironment regulates GC remains largely unknown.For example, how do the unique regional immune properties of the stomach regulate human GC initiation?How does the heterogeneity of the immune microenvironment determine the sensitivity or response to immuno-therapy?
A new dimension of GC heterogeneity, nerve-cancer crosstalk, is increasingly being implicated in gastric tumorigenesis.Although the stomach is innervated predominantly by the autonomic, non-autonomic, and enteric nervous systems, to maintain epithelial homeostasis and hormone secretion 36 , elevated infiltration of other neuronal fibers derived from vagal nerves, sympathetic nerves, and choline-acetyltransferase positive stromal neurons is frequently observed in GC tissues, and neural density positively correlates with GC progression and poor survival prognosis [37][38][39] .Indeed, sole surgical vagotomy or myenteric denervation has been found to efficiently decrease the incidence of gastric tumors, enhance chemotherapy effects, and prolong overall survival in both mouse and rat models, as well as in human patients with GC 37,40 , thus adding a new layer of complexity underlying GC carcinogenosis [36][37][38]41 . Exporing how nerves, or even emotions, regulate the initiation, progression, and response to targeted and/or immunotherapy of human GC should prove interesting.
Beyond intrinsic factors, environmental factors, such as microorganisms, have been well established to participate in GC initiation and development 2 .For example, gastric tissue injury and chronic inflammation triggered by Helicobacter pylori infection initiate sequential histopathologic progression of gastritis to gastric atrophy, intestinal metaplasia, dysplasia, and finally gastric adenocarcinoma 42 .Recently, several oral pathogens, including Streptococcus anginosus 43 , Candida albicans 44 , and intracellular bacteria 45 , have been identified to colonize the stomach and promote gastric tumorigenesis.Nevertheless, how these novel microbe-host interactions and genetic mutations synergistically drive GC tumor evolution awaits further investigation.In addition, how microbiota determine the sensitivity or response to targeted treatment and immunotherapy for GC remains poorly understood.
To address GC heterogeneity and uncover its pathological nature, multiple model systems can be applied, including cell lines, organoids, animal models, and clinical samples.In the past decade, organoids have been developed as powerful tools for both mechanistic study of tumorigenesis and drug screening.Comparisons between patient-derived organoids and single cells from primary tumors have highlighted interand intralineage similarities and differences 15 , thus suggesting that heterogeneity may decrease or disappear during the continuous passage of organoids.Moreover, current organoid models do not adequately mimic the complex microenvironment in vivo, where immune cells interact with tumor cells.Furthermore, tumorigenesis is increasingly understood to be regulated by crosstalk between the stomach and other organs, e.g., the liver and brain.Therefore, various GC animal models must be developed to meet research needs, including mechanistic study and target intervention, genealogical tracing of different GC subtypes, and assessment of chemotherapy and immunotherapy.
Multiple immunocompetent GC mouse models have been developed, including GC cell line transplantation, chemical carcinogen administration, and genetic engineered models (Figure 1).These models enable study of the cellular origins, clonal evolution, relapse, host-microbe interaction, tumor immunity, and neuronal/emotional control of tumor immunity during various stages of GC, including initiation, progression, invasion, and metastasis, with or without therapeutic treatment.Herein, we review the roles of well-established and emerging GC mouse models, particularly genetically engineered mouse models (GEMMs), in deciphering the heterogeneity in human GC; we further compare their pathological features, applications, and limitations.

Cell line-derived GC graft model
Cell lines derived from patients with GC and mice provide powerful tools to explore the nature of tumor progression, and responsiveness to targeted therapy and immunotherapy.In addition to the multiple human GC cell lines available for xenograft study in immunocompromised mice, several mouse GC cell lines (e.g., MFC, MGCC3I, NCC-S1/3, YTN16, and M12), which can be transplanted into immunocompetent mice, have been generated to investigate the molecular and cellular mechanisms governing gastric tumorigenesis and related immune response.
The widely used mouse GC cell line MFC was generated from forestomach carcinoma developed in N-methyl-N'nitro-N-nitrosoguanidine (MNNG) treated mice on a 615 inbred background.MFC cells form tumors after orthotopic or subcutaneous implantation in mice, and are prone to spontaneous metastasis to the lungs 46  which tumor cells promote immune evasion 47,48 and impede responses to anti-PD-1 immunotherapy [48][49][50] .For example, through SLC6A6-mediated competitive uptake of taurine, gastric cancer cells induce CD8 + T cell exhaustion by increasing ER stress and ATF4 mediated upregulation of immune checkpoint genes, thus resulting in immune evasion and tumor progression 47 .MGCC3I, another forestomach carcinomaderived mouse GC cell line, forms poorly differentiated gastric carcinoma after orthotopic transplantation into the stomach serosa, and liver metastasis after intrasplenic injection into the syngeneic ICR mice 51 .The NCC-S1 and NCC-S3 cell lines were derived from primary GC developed in Smad4 fl/fl ; Trp53 fl/fl ; Cdh1 fl/+ ; Villin cre and Trp53 fl/fl ; Cdh1 fl/fl ; Pdx1 cre mice, respectively 52 .Metastatic NCC-S1M and NCC-S3M subclones were then isolated from lung metastatic foci.Activation of the Wnt/β-catenin signaling pathway is required for the metastatic phenotype 52 .Notably, 2 mouse GC cell lines (YTN16 and M12) transplantable into mice with a C57BL/6 background have been developed 53,54 .YTN16 cells were subcloned and established from N-methyl-N-nitrosourea (MNU) treated p53 heterozygous knockout mice.YTN16 cells form orthotopic tumors and metastasis foci in lymph nodes, the peritoneum, and lungs 53 .YTN16 cells have been used to develop novel mouse models of lymphatic and peritoneal metastasis [55][56][57] , gastric tumor immunogenicity 58 , neoantigen identification 59 , and response to immune checkpoint inhibitors in GC [60][61][62] .M12 cells have also been derived from gastric carcinoma in p53 knockout mice on a C57BL/6 background treated with a zinc-deficient diet and MNU administered in the drinking water.M12 cells show tumorigenic and metastatic properties in C57BL/6 syngeneic mice 54 .Serine/threonine-protein kinase 24 (STK24) is essential for immune regulation during the tumor progression of M12 cells in vivo.Knockdown of STK24 promotes myeloid derived suppressor cell (MDSC) expansion and tumor growth in C57BL/6 mice 54 .

Chemical carcinogen-induced GC mouse models
MNU, one of the best-characterized chemical carcinogens, can be supplied in the drinking water to induce GC in mice 63 .By introducing alkyl radicals into DNA, MNU causes DNA mutation and dysfunction, thereby promoting gastric tumorigenesis 64 .MNU-induced primary GCs are usually localized in the antrum, and involve well to poorly differentiated adenocarcinoma 63,65 .The tumorigenic efficacy of MNU varies in mice with different genetic backgrounds; male mice on a BALB/c background are relatively susceptible to MNU-induced tumors 65 .In addition, MNU-induced GC is significantly enhanced in combination with other GC risk factors, such as a high-salt diet 66 , H. pylori infection 67 , and Streptococcus anginosus infection 43 .Notably, genetic alterations also significantly influence MNU-dependent tumorigenesis.For example, p53 knockout mice are relatively sensitive to MNU-induced carcinogenesis 68,69 .
MNNG is another chemical carcinogen particularly widely used in combination with Helicobacter infection to induce GC in mice 70 .MNNG is supplied in the drinking water in 3 cycles at 2 week intervals to induce GC in mice.MNNGinduced primary GC varies across model organisms, including squamous cell carcinoma in the forestomach in mice 46,70 and adenocarcinomas in the glandular stomach in Mongolian gerbils 71 .Similarly, environmental GC risk factors, including a high-salt diet 72 , calcium-deficient diet 71 , or catechol 73 , promote the incidence and progression of GC induced by MNNG administration.Moreover, this model has been extensively used to investigate gastric tumorigenesis and targeted therapy against GC 32,33,[74][75][76] .

IL-1β β transgenic mice
Gastrointestinal cancers are frequently associated with chronic inflammation.For example, chronic inflammation triggered by H. pylori infection or tissue injury in the stomach can initiate sequential histopathologic progression of gastritis to gastric atrophy, intestinal metaplasia, dysplasia, and finally gastric adenocarcinoma.Interleukin-1 polymorphisms have been associated with increased risks of both hypochlorhydria induced by H. pylori infection and gastric carcinogenesis 77 .H/K-ATPase:hIL-1β transgenic mice expressing secretory human IL-1β specifically in parietal cells have been generated to explore the pathogenic role of hIL-1β during gastric tumorigenesis.These mice spontaneously develop chronic gastritis, hyperplasia, and high-grade dysplasia/adenocarcinoma without invasion into the submucosa or metastasis to distant organs 78 .In a setting of H. felis Infection, IL-1β has been found to accelerate the development of gastric inflammation and carcinoma, thereby indicating a causative effect of IL-1β in inflammation-associated GC 79 .This model has been used to explore mechanisms of tumor resistance to immune checkpoint blockade of GC.Overexpression of IL-1β in the stomach results in recruitment of MDSCs through the IL-1RI/ NF-κB signaling pathway.MDSCs exert an immunosuppressive function through upregulation of PDL1, and anti-PD-1 treatment does not block GC progression in these IL1β transgenic mice 78 .

NF-κB1-deficient mice
Deficiency in NF-κB1, even loss of a single allele, can lead to spontaneous intestinal-type gastric adenocarcinoma in mice.Interestingly, such gastric adenocarcinoma is not accelerated by H. pylori infection and a high salt diet in these NF-κB1deficient mice 80 , thus indicating a GC pathology independent of commensal microorganisms.This model has been used to study inflammation associated malignancy in GC.Deficiency in NF-κB1 results in increased expression of a variety of inflammatory cytokines, including tumor necrosis factor (TNF), interleukin-6 (IL-6), IL-22, and IL-11, thereby driving aberrant activation of signal transducer and activator of transcription 1 (STAT1).Further genetic depletion of TNF or STAT1 in NF-κB1-deficient mice has been found to prevent invasive GC development 81 .In agreement with these observations, genetic analysis has identified a significant association between the Nfκb1 locus and gastric tumor susceptibility in a collaborative cross-mouse population 82 .

Gp130 F/F transgenic mice
Excessive secretion of IL-6 cytokine family members, including IL-6, IL-11, IL-27, IL-31, oncostatin M, leukemia inhibitory factor, ciliary neurotrophic factor, ardiotrophin-like cytokine factor 1, and cardiotrophin 1, promote GC through persistent activation of the JAK-STAT1/3 and/or SHP2-Ras-ERK signaling pathways through the IL-6 cytokine family of receptors 83 .To investigate the role of dysregulated activation of STAT3 in regulating gastrointestinal epithelial cell homeostasis, Tebbutt et al. 84 have generated gp130 F/F mice by using a phenylalanine knock-in substitution of the IL-6 receptor β-chain Gp130 at the cytoplasmic tyrosine 757 residue, thus preventing its binding to the suppressor of cytokine signaling 3 and enhancing activation of STAT3.Gp130 F/F mice spontaneously develop gastric adenoma at the antrum by 4-6 weeks of age, accompanied by splenomegaly and extra-gastric pathologies in the liver and lung 84,85 .Further knockout of STAT3 in Gp130 F/F mice alleviates gastric adenoma progression, thereby highlighting an essential role of STAT3 hyperactivation in GC pathology 85 .
Among IL-6 cytokine family members, IL-11 is a major cytokine promoting gastrointestinal tumorigenesis.Pharmacological inhibition of IL-11 signaling through mIL-11 Mutein administration or genetic depletion of IL-11 ligand-binding receptor subunit in Gp130 F/F mice has been found to inhibit GC development 86,87 .Beyond the IL-6 cytokine family, STAT3 activation directly increases Tolllike receptor (TLR) 2 expression, thereby promoting gastric tumor cell survival and proliferation.Accordingly, genetic or therapeutic targeting of TLR2 has been found to alleviate gastric tumorigenesis 88 .Moreover, IL-6/IL-11-gp130-dependent mTORC1 activation has been implicated in promoting inflammation-associated gastrointestinal tumorigenesis, which is druggable through treatment with the mTORC1specific inhibitor RAD001 89 .Finally, Gp130 F/F transgenic mice have also been used to study the interaction of tumor cells with immune cells involved in GC progression.For example, mast cells have been found to be activated by GC cell-derived alarmin IL-33 and to recruit macrophages via secreting attracting chemokines, such as CSF2, CCL3, and IL-6; moreover, deletion of macrophages has been found to suppress gastric tumorigenesis 90 .

Transgenic mice with aberrant inflammation induced by T cells
Deregulated T cell activation mediates gastritis and promotes gastric hyperplasia and adenocarcinomas.For example, T cell-specific deletion of the tumor suppressor liver kinase B1 (LKB1) results in excessive production of proinflammatory cytokines and chemokines such as IL-6, IL-11 and CXCL2, which is accompanied by increased STAT3 activation and infiltration of inflammatory monocytes and neutrophils.The related inflammation promotes development of gastrointestinal polyposis, a cancer predisposition syndrome 91 .In addition, autoimmune gastritis mediated by self-reactive CD4 + T cells has been found to promote GC development.In a T cell receptor transgenic mouse model of autoimmune gastritis, the T cell receptor targets a peptide from the H + / K + ATPase proton pump, which is highly expressed on parietal cells in the stomach.Transgenic mice display chronic gastritis with intensive CD4 + T cell infiltration, and elevated IFNγ and IL1-17 production, which is followed by initiation and progression of GC from oxyntic atrophy, mucinous hyperplasia to spasmolytic polypeptide-expressing metaplasia, and intraepithelial neoplasia 92 .Collectively, these mouse models illustrate a causal link between gastric inflammation and GC development.

INS-GAS mice
Gastrin, produced by antrum G cells, is crucial for gastric acid secretion and parietal cell differentiation.A transgenic mouse model termed INS-GAS expressing human gastrin specifically in β islet cells under control of the insulin promoter was originally generated to investigate the potential role of gastrin in regulating islet differentiation 93,94 .INS-GAS mice show a twofold elevation of serum amidated gastrin and gastrointestinal mucosal hyperplasia 94 .These mice have been further used to examine the role of hypergastrinemia in GC pathology and have shown elevated maximal gastric acid secretion and parietal cell number within 4 months old, but progressive sustained loss of parietal and hypochlorhydria 95 .Eventually, INS-GAS mice develop metaplasia, dysplasia, and invasive GC at 20 months of age (Figure 2).
Notably, INS-GAS mice exhibit accelerated progression to gastric carcinoma in the presence of H. felis infection 95 .Moreover, this tumor progression is influenced by sex, genetic background, and commensal flora.Female INS-GAS mice are more resistant than male INS-GAS mice to H. felis infection-induced GC [96][97][98] .Ovary derived estradiol may contribute to the protective role of this sexual dimorphism, because ovariectomized female mice develop more severe gastritis and gastrointestinal neoplasia than intact female mice 97 .In addition, INS-GAS mice on an FVB/N background have been found to be susceptible to H. felis infection-induced GC, whereas those on a C57BL/6 background develop only metaplasia and dysplasia 99 .In addition, H. felis-infected INS-GAS mice raised in germ-free conditions develop mitigatory gastritis and delayed intraepithelial neoplasia, in contrast to those raised in specific-pathogen-free conditions 100 .Moreover, gastric colonization with restrict altered Schaedler's flora in male germ-free INS-GAS mice is sufficient to promote gastric inflammation and dysplasia to a similar extent as diverse intestinal microbiota in the presence of H. pylori infection 101 .In the setting H. felis infection, INS-GAS mouse models have served as an important tool for validation of anti-inflammatory strategies for GC treatment.The combination of the nonsteroidal anti-inflammatory drug sulindac and antibiotic mediated H. pylori eradication has been found to alleviate the production of pro-inflammatory cytokines in the stomach, as well as the progression from H. pylori-associated severe dysplasia to GC 102 .In addition, the combination of the gastrin receptor antagonist YF476 and the histamine H2-receptor antagonist loxtidin has been found to completely suppress gastric acid secretion and progression to neoplasia 103 .

Gastrin −/− mice
Gastrin-deficient mice (gastrin −/− ) have been generated to investigate the role of gastrin in regulating the development and function of the gastrointestinal tract 104,105 .Gastrin −/− mice show impaired gastric acid secretion, accompanied by marked abnormalities in gastric gland architecture, with diminished numbers of parietal and enterochromaffin-like cells, and enhanced numbers of mucous neck cells 104,105 .The loss of parietal cells in gastrin −/− mice has been attributed to bacterial overgrowth and chronic gastritis, and the parietal cell number has been found to normalize after antibiotic treatment 106 .The chronic inflammation resulting from gastric acid secretion disorder in gastrin −/− mice promotes intestinal metaplasia of the gastric epithelium, which eventually develops into polyps by the age of 12-month 107-109 (Figure 2).

GCs induced by genetic mutations in gastric epithelial cells
Genetically engineered mice with genetic mutations in pan-epithelial cells Claudin18.2 is a tight junction membrane protein specifically expressed in the gastric epithelium.Claudin18.2knockout mice (CLDN18KM) exhibit preneoplastic lesions at 7 weeks and eventually develop highgrade intraepithelial neoplasia at 2 years of age 110 .However, CLDN18KM mice are resistant to H. pylori colonization and are not suitable to investigate pro-tumor of H. pylori infection.Of note, the yes-associated protein 1 (YAP1) signaling pathway has been found to be up-regulated and to contribute to the proliferation of metaplastic cells in CLDN18KM mice 110 .
Given that Wnt/β-catenin signaling, receptor tyrosine kinase, and Trp53 pathways are commonly perturbed pathways in GC, Fatehullah et al. 111 have developed a claudin18.2-IRES-CreERT2allele to selectively drive conditional knock in of Kras G12D and deletion of Apc and Trp53 (Cldn18-ATK) in the gastric epithelium.Cldn18-ATK mice develop high grade CIN GC, which metastasizes to the liver, lymph nodes, and diaphragm 111 .This model has been used to evaluate the roles of Lgr5 + stem cells in GC initiation and distant metastasis 111 .Expression of Anxa10, a member of the annexin family of calcium-dependent phospholipid-binding proteins, is restricted to the gastric epithelium.Mutations in oncogenes and/or tumor suppressor genes specifically in Anxa10 + cells result in mouse GCs with various subtypes possibly mimicking human GC.Knock in of Kras G12D and Trp53 R172H , and deletion of Smad4 in Anxa10 + cells leads to intestinal-type GC, which is prone to metastasis to the liver and the lungs.Knock in of Kras G12D and deletion of Cdh1 and Smad4 in Anxa10 + cells result in poorly differentiated signet ring cell carcinoma, and metastasis to the lung and peritoneum; moreover, knock in of Kras G12D and deletion of Cdh1 and Apc in Anxa10 + cells lead to serrated adenomatous GC 112 .Tumor organoids have been derived from these models to test responses to conventional chemotherapeutics and targeted therapeutics.Intestinal-type CIN organoids are relatively sensitive to docetaxel treatment but resistant to trametinib treatment targeting the EGF receptor (EGFR).In addition, overexpression of peroxisome proliferator-activated receptor delta (Ppard1/2) in Villin + cells promotes gastric inflammation and tumorigenesis 113 .

Genetically engineered mice with genetic mutations in stem cells
With their capabilities of self-renewal, proliferation, and differentiation into various types of functional cells, stem cells play important roles in tissue homeostasis and injury repair 24,114 .Moreover, stem cells have been found to be the cellular origins of gastric tumorigenesis and metastasis 24,114 .Lgr5 marks homeostatic stem cells in multiple tissues including the gastrointestinal tract.In the human and mouse stomach, Lgr5 is expressed in a subpopulation of chief cells located at the base of the corpus gland.Lgr5-expressing chief cells drive epithelial renewal after injury and are the cells of origin of GC 115 .Knock in of Kras G12D or Trp53 deletion in Lgr5-expressing chief cells promotes metaplastic lesions in the corpus 69,115 .In Cldn18-ATK mice, Lgr5 + cells function as cancer stem cells in gastric tumorigenesis and distal metastasis 111 .
Recently, Aqp5 has been identified as a new pyloric-specific marker of Lgr5-expressing stem cells.Hyperactivation of the WNT/β-catenin, PI3K, and KRAS signaling pathways by deletion of Apc and Pten and knock in of Kras G12D in Aqp5 + stem cells cooperatively drives invasive gastric tumorigenesis 116 .In addition, pepsinogen C (PGC) is a predominant marker secreted by gastric chief cells, and successive activation of Kras G12D and depletion of Apc and Trp53 in Pgc + chief cells have been found to result in progressive development of metaplasia, dysplasia, and invasive and metastatic gastric carcinoma 117 .Recently, knock in of Kras G12D alone in zymogensecreting chief cells has been reported to lead to the development of precancerous metaplasia and high-grade dysplasia.Metabolic rewiring from glycolysis to fatty acid metabolism occurs during the progression from metaplasia to dysplasia 118 .Stearoyl-coenzyme A desaturase dependent production of monounsaturated fatty acids fuels dysplastic cells 118 .
Mist1 expression marks the stem cells located in the isthmus of the gastric corpus 24 .Knock in of Kras G12D and Apc deletion in Mist1-expressing stem cells give rise to intestinaltype metaplasia and cancer.In addition, depletion of Cdh1 in Mist1 + stem cells may cause diffuse-type GC.Importantly, Cxcl12 + endothelial cells recruit Cxcr4 + innate lymphoid cells (ILCs), which form a peri-vascular inflammatory niche supporting diffuse-type GC development from Mist1 + cells through Wnt5a produced by ILCs 119 .ILC-derived Wnt5a mediates RhoA activation and promotes tumor cell survival.Concordantly, RHOA gain of function through knock in RHOA Y42C combined with Cdh1 deletion in Mist1 + cells induces metastatic diffuse-type GC 120 .Mechanistically, Cdh1 loss and RHOA Y42C mutation induce cytoskeletal rearrangements and focal adhesion kinase activation, which in turn further promote the activation of YAP/TAZ, PI3K/AKT, and WNT/β-catenin signaling 120 .
The transcription factor Sox2 marks adult stem cells in multiple epithelial tissues, including the glandular stomach, anus, cervix, testes, and lens 121 .Sox2 is highly expressed in the basal progenitor cells of the stratified epithelium in the esophagus and forestomach 121 , and drives gastric specification and regionalization by maintaining chromatin accessibility of forestomach lineage-specific genes 122 .Overexpression of Sox2 in basal progenitor cells results in the development of invasive squamous cancer in the forestomach and is involved in inflammation-mediated Stat3 activation 123 .However, deletion of Apc in Sox2 + cells leads to gastric adenoma formation in the corpus, and loss of Sox2 enhances gastric tumorigenesis, thus suggesting that Sox2 may also act as a tumor suppressor by restraining Wnt/β-catenin signaling and intestinal genes 124 .Moreover, activation of KRAS in Sox2 + cells also leads to precancerous lesions in gastric tissues, accompanied by accumulation of Sox9 + cells in the stomach.A combined Cre-loxp and Flipase-Frt system to specifically activate Kras and deplete Trp53 in Sox2 + Sox9 + cells has been found to result in the development of aggressive GC, in which SOX9 promotes the transformation of SOX2 + stem cells through biased symmetric cell division 125 .In addition, SOX9 in epithelial tumor cells promotes M2 macrophage polarization and CD8 + T cell functional inhibition through paracrine secretion of LIF, thus driving the progression and metastasis of gastric adenocarcinoma.

Genetically engineered mice with genetic mutations in terminally differentiated cells
Tff1 −/− mice Trefoil factor 1 (TFF1) is a tumor suppressor gene that belongs to the trefoil factor family and is expressed predominantly in gastric pit cells.TFF1 transcription is positively regulated by the gastrin hormone 126 , and decreased abundance of TFF1 resulting from epigenetic silencing is involved in gastric carcinogenesis 127,128 .TFF1 knockout mice (Tff1 −/− ) have been generated by Lefebvre et al. 129 to investigate the roles of this factor in gastrointestinal homeostasis and tumorigenesis.Tff1 −/− mice develop severe hyperplasia and dysplasia, marked by elongated pits and enlarged nuclei.Moreover, 30% of Tff1 −/− mice develop invasive pyloric adenoma, but no metastatic dissemination to the lung or liver, at the age of 5 months 129 .In line with the role of dysregulated inflammation in promoting GC, knockout of TFF1in mice and progressive loss of TFF1in human gastric tissues have been associated with activation of NF-κB-mediated inflammation and progression to gastric tumorigenesis 130 .This model has been used to investigate the role of prostanoid metabolism in GC progression.Expression of cyclooxygenase-2 (Cox-2) is elevated in pyloric adenoma in Tff1 −/− mice, and is involved in the conversion of arachidonic acid to prostanoid precursors.Moreover, inhibition of Cox-2 through genetic deletion or treatment with the selective inhibitor celecoxib decreases adenoma size and ulceration in Tff1 −/− mice 131,132 .

Mutations in parietal cells Parietal cells, marked by Atp4b
expression, account for one-third of all gastric epithelial cells.Parietal cells secrete gastric acid in response to gastrin stimulation, thereby maintaining the acidic environment of the stomach and inhibiting the invasion of pathogenic microorganisms.Manipulation of oncogenes and/or tumor suppressor genes in parietal cells also contributes to gastric tumorigenesis.For example, activation of Notch signaling by knock in of Notch1 intracellular domain (NICD1) in Atp4b + parietal cells induces dedifferentiation into multipotential progenitors that populate the gastric epithelium.Sustained Notch activation within parietal cells eventually induces adenomas characterized by focal Wnt/βcatenin signaling activation 133 .In addition, loss of AT-rich interaction domain 1A (Arid1a), a key subunit of the chromatin remodeling BAF complex in Notch-signaling-activated parietal cells, further accelerates GC progression in a dose-dependent manner 134 .Mechanistically, homozygous depletion of Arid1a leads to a competitive disadvantage through activation of the p53 pathway and thus promotes gastric tumorigenesis 134 .
Multiple studies have shown that the Cdh1 gene, encoding E-cadherin, displays hypermutation in diffuse-type GC 10,135 .Deletion of Cdh1 alone in Atp4b + parietal cells leads to mucosal hyperplasia and spasmolytic polypeptide-expressing metaplasia 136 .Synergistic depletion of Cdh1 and Trp53 in parietal cells results in the development of invasive diffuse-type GC, thus leading to a high frequency of lymphatic metastases and tumorigenic activity in immunodeficient mice 137 .In addition to specific deficiency in Cdh1 and Trp53, knock in of oncogenic Kras G12D in parietal cells accelerates intestinal and diffuse-type gastric tumorigenesis, and lymphatic and hematogenous metastasis in the lymph nodes, liver, and lungs 138 .In this regard, KRAS activation promotes epithelialto-mesenchymal transition and the generation of cancer stem cells, and consequently metastasis to the lungs 139 .

Mutations in neuroendocrine cells
Neuroendocrine cells in Neurogenin 3-expressing progenitor cells in the gastric epithelium play an essential role in maintaining gastrointestinal homeostasis and have been proposed as a potential cellular origin of gastric neuroendocrine neoplasms [140][141][142] .Neuroendocrine neoplasms are characterized by the expression of neuroendocrine markers and are divided into subclasses of well-differentiated neuroendocrine tumors, aggressive poorly differentiated neuroendocrine carcinoma, and mixed neuroendocrine/non-neuroendocrine neoplasia 143 .A missense mutation (p.R703C) in the human ATP4a gene has been identified in aggressive familial gastric neuroendocrine tumors.Mice with knock in of human ATP4a R703C develop severe metaplasia and dysplasia in the stomach 144 .Recently, by characterizing the genomic landscapes and transcriptional subtyping of human gastric neuroendocrine carcinoma (G-NEC), Griger et al. 145 have identified MYC as a critical driver of G-NEC.The Cγ1-cre allele was used to drive overexpression of MYC in the gastric neuroendocrine compartment.MYC-driven mouse G-NECs develop aggressive malignancies and distal metastatic foci in the lungs and liver 145 .The G-NEC cell line and organoid resources derived from this GC model were generated to perform genome-scale CRISPR and pharmacologic screens.

Gastrointestinal stromal tumors (GISTs)
GISTs are among the most common human sarcomas in human gastrointestinal tracts.GIST originates from the interstitial cells of Cajal (ICC) which depends on high expression of KIT for lineage commitment.GIST is characterized primarily by activating mutations in KIT or PDGFRA receptor tyrosine kinase 146,147 .Multiple mouse models of GIST have been established through knock in of KIT mutations.For example, knock in of Kit V558∆ or Kit K641E results in the development of human GIST-like tumors marked by ICC hyperplasia within the myenteric plexus of the GI tract 148,149 .Imatinib (Gleevec), a multitargeted tyrosine kinase inhibitor targeting KIT/PDGFR, is the standard first-line therapy for advanced GIST.However, patients with GIST frequently develop imatinib resistance resulting from second-site mutations of KIT.Further knock in of Kit T669I or Kit K653A in Kit V558∆ mice promotes GIST development and induces resistance to imatinib 150,151 .
The ETS family member ETV1 is another lineage survival factor of ICC.Activating mutation of KIT stabilizes the ETV1 protein through constitutive activation of the KIT-MAPK signaling pathway, and augments ETV1 transcriptional output, thus promoting GIST 152 .In the Kit V558∆ mouse model, ETV1 ablation inhibits GIST initiation and progression 153 .Moreover, the Forkhead family member FOXF1 directly controls the transcription of KIT and ETV1, and is required for tumor growth and maintenance via regulating the GIST lineage-specific transcriptome 154 .Combining knock in of Braf V600E with Trp53 deletion in ETV1 + ICC or smooth muscle cells drives ICC hyperplasia and multifocal GIST-like tumor formation in the mouse gastrointestinal tract 155,156 .

Conclusions and perspectives
Although gastric anatomy differs between mice and humans, many mouse models have been established to study GC pathology and related immune responses.In combination with gastric disorders, such as H. pylori infection, sophisticated GC mouse models-including cell line-derived graft tumors, treatment with chemical carcinogens, and genetic engineering-have laid an important foundation for exploring GC pathogenesis and antitumor immunity (Figure 1).Mutation of oncogenes and tumor suppressor genes in gastric epithelial and interstitial cells induces different types of GC, thus contributing to the high heterogeneity in GC (Table 1).Therefore, illustrating the pathogenesis and molecular features of GC induced by specific genetic mutations in specific types of gastric cells will not only uncover cell signaling networks crucial for GC development, but also promote accurate diagnosis and efficient treatment of GC.
Mouse GC cell lines provide powerful tools to explore the molecular mechanisms regulating immune responses.The most widely used mouse GC cell line is MFC, derived from squamous carcinoma in the forestomach from mice on a 615 background.However, mice with conditional knockout of certain immune system genes (e.g., Cd4 cre in T cells and Cd19 cre in B cells) are usually bred on a C57BL/6 background, which is not suitable for MFC cell tumorigenicity.In this regard, the development of mouse gastric adenocarcinoma cell lines, particularly those with a C57BL/6 background, has become an urgent need for tumor immunity research in GC.In addition, graft tumor models usually induce an unnaturally hyperinflammatory state after tumor cell transplantation 157 .Without high spatial and temporal specificity, chemical carcinogens such as MNU induce a high mutational burden and immunogenicity, but also dampen the immune system, including T cells.In contrast, GEMMs of GC provide multiple advantages for studying tumor cell initiation, progression, and antitumor immunity (Figure 2).The advantages of GEMMs include the following: (1) Modeling gastric tumorigenesis through genetic manipulation within distinct cell types aids in understanding the high heterogeneity arising from different cell origins and the genotype-phenotype relationships during human GC progression.
(2) In combination with the lineage tracing strategy, the gastric local spontaneity of GEMM helps reveal the clonal evolution of tumor cells within a complete immune system.(3) GEMMs of GC can replicate the effects of the gastric immune microenvironment and the crosstalk between the stomach and other organs, such as the brain and liver, in GC initiation and metastasis.However, the extent to which currently available genetically engineered GC mouse models reflect the pathological nature of human GC is questionable.Systemic comparative studies defining the similarity between human and mouse GC subtypes at single cell-resolution are lacking.Moreover, use of GEMMs is usually time-and resource-consuming, because of the need to intercross multiple germline strains.Recently, Leibold et al. 158 have developed somatic mouse models of GC by introducing various oncogenic lesions into the murine gastric epithelium through an electroporation-based approach.This strategy accelerates the development of GC mouse models, although the cellular origin of GC is in suspense and the injury is unavoidable during surgery 158 .
In addition, patient-derived xenografts (PDXs) and human gastric cell line-derived xenografts in immunocompromised mice are crucial tools for GC research 159,160 .Originating from human gastric tumor tissue, PDXs retain the genetic and phenotypic characteristics of tumors in the presence of stroma and immune cells, thus favorably modeling the natural tumor microenvironment.These models reflect the heterogeneity

Table 1 Continued
in patients' cancers as well as the biological characteristics and mutational landscape of cancer cells 161,162 .Additionally, humanized mouse models are an important GC research tool providing a humanized immune microenvironment for PDX and gastric cell line-derived xenograft growth, which may serve as a platform for the evaluation of drugs modulating the anti-tumor immune response 159 .
Given the heterogeneity in GC, no single GC model can answer all GC-associated scientific questions.Taking advantage of GEMM in GC research, further endeavors may focus on clarifying the cellular origins and clonal evolution of tumor cells during GC initiation and metastasis, through intercrossing with lineage tracing germline strains; the molecular and cellular mechanisms driving GC invasion, metastasis and metastatic organ tropism, by using cell lines and organoid resources derived from GEMMs; the heterogeneity in the immune microenvironment of primary GC located in the distinct anatomy of the stomach; the roles of stomach-brain and stomach-liver organ communication in regulating GC progression; the mechanisms mediating T cell exhaustion, formation of tertiary lymphoid structures, and resistance to immunotherapy; and the effects and mechanisms of emotions, biological rhythm, nerves, and microorganism infections in regulating GC tumorigenesis, immune evasion, and responses to targeted treatments and immunotherapy.

Figure 1
Figure 1Immunocompetent mouse models of GC. (A) Allograft GC models generated by subcutaneous and orthotopic transplantation of mouse gastric cell lines, such as MFC and YTN16, show favorable replicability and stability, but also induce an unnaturally hyperinflammatory response.(B) Chemical carcinogen-induced GC mouse models, such as those using MNU and MNNG, exhibit high mutational burden and immunogenicity.(C) Genetically engineered mouse models of GC generated through genetic manipulation within stomach specific cells have uncovered genotype-phenotype relationships during GC initiation and progression.

Table 1
Genetically engineered mouse models of gastric carcer GEMM