The clinical characteristics are shown in Table 1 and S1. All AIG cases showed severe mucosal atrophy predominantly from the gastric body to the fundus and showed high anti-PCA titer (> 10 RU/mL) [13]. No significant differences were noted in age and sex between the AIG, HPG, and normal cases analyzed for comprehensive gene expression analysis (Table S3).

AIG displayed a unique gene expression profile from HPG in gastric mucosa

In comprehensive gene expression analyses, several gene transcripts were significantly (P < 0.05) upregulated (more than twofold change: 4219 of 48,858) and downregulated (3,536) in AIG samples, whereas 3,892 gene transcripts were upregulated and 3281 were downregulated in HPG samples (Fig. 2a). Unsupervised cluster analysis using 2500 genes with the greatest SD showed that the AIG and HPG samples were clearly separated from the normal mucosa samples and were further separated into the AIG-enriched cluster and the HPG-enriched cluster (Fig. 2b), indicating a unique expression profile of AIG. In addition, ATP4A was markedly downregulated (Fig. 2a), indicative of parietal cell damage. GAST (gastrin) and PAPPA2 (pappalysin 2), which have been reported to contribute to the development of NET [14], were highly expressed in the AIG samples (Fig. S1). The top 50 differentially expressed gene transcripts in AIG and HPG samples are listed in Tables S4, S5, S6, and S7.

Fig. 2

Results of the comprehensive analysis of gene expression among the AIG (n = 14), HPG (n = 13), and normal samples (n = 9). a Volcano plot analysis using the fold changes of gene expression levels between the normal and AIG samples, and the normal and HPG samples. The number of gene transcripts with twofold change and a small P-value (-log10 (P-values) > 1.301) was higher in the AIG samples than in the HPG samples. b Unsupervised hierarchical cluster analysis using gene expression levels of the total 36 samples. Using the 2500 gene transcripts with the highest standard deviation (TOP SD), the AIG and HPG samples were clearly separated from the normal samples and were further separated into the AIG-enriched cluster and the HPG-enriched cluster. c Pathway enrichment analyses conducted via GSEA using the upregulated genes in the AIG and HPG samples. The top eight activated gene sets in the AIG and HPG samples are shown. NES—normalized enrichment score

To explore the pathways upregulated in the gastric mucosa with AIG and HPG, pathway enrichment analyses were conducted using the upregulated genes. With a cutoff P-value of 0.01, GSEA identified 14 terms in both AIG and HPG samples (Fig. 2c). The AIG samples were enriched with gene sets associated with small intestinal absorption, such as “porphyrin and chlorophyll metabolism” and “starch and sucrose metabolism,” indicating abnormal intestinal differentiation in gastric mucosa. In contrast, the HPG samples were enriched with many gene sets associated with inflammatory response, such as “Fc-gamma receptor-mediated phagocytosis,” indicating aggressive macrophage- and neutrophil-related inflammation in gastric mucosa with HPG.

Upregulation of small intestine-specific genes and downregulation of stomach-specific genes induced in gastric mucosa with AIG

Abnormal intestinal differentiation is known to occur in gastric mucosa with HPG [15]. To investigate whether abnormal intestinal differentiation is induced in gastric mucosa with AIG as prominently as in that with HPG, tissue enrichment analysis (TissueEnrich) was performed using the top 50 upregulated genes in gastric mucosa with AIG and HPG. AIG showed higher enrichment of genes specific to the duodenum and small intestine (− log10 (P value) = 68.5 and 65.2, respectively) than HPG (25.8 and 22.2, respectively) (Fig. 3a). Additionally, mucosal samples were compared by analyzing the expression levels of tissue-specific genes, which were identified as the top 25 highly expressed genes in a tissue with an SD greater than 2SD among all tissue types using the GTEx database. Notably, almost all the AIG samples were clearly separated from the normal samples, in case of small intestine-specific and stomach-specific genes (Tables S8 and S9), whereas a few HPG samples grouped with the normal samples (Fig. 3b), indicating frequent intestinal differentiation in the gastric mucosa with AIG. In the histological analysis using the USS scores, the AIG samples demonstrated a high incidence of intestinal metaplasia (10/11, 90.9%), whereas the HPG samples had a lower incidence (3/7, 42.9%) (Fig. S2a). Moreover, similar to HPG-associated intestinal metaplasia [16], AIG-associated intestinal metaplasia exhibited MUC2 expression but lacked MUC5AC expression (Figs. 3c and S2b).

Fig. 3

Marked intestinal differentiation in gastric mucosa with AIG. a Tissue enrichment analysis using top 50 upregulated genes in gastric mucosa with AIG and HPG. AIG showed higher enrichment of genes specific to the duodenum and the small intestine compared with HPG. b Unsupervised hierarchical cluster analysis using gene expression levels of small intestine-specific and stomach-specific genes among the AIG (n = 14), HPG (n = 13), and normal samples (n = 9). The AIG samples were clearly separated from the normal samples using small intestine and stomach-specific genes, while a fraction of the HPG samples were grouped with the normal samples. c Immunostaining of MUC2 and MUC5AC using gastric mucosa with AIG. Scale bar: 100 μm. d Unsupervised hierarchical cluster analysis using gene expression levels of CDX signature genes, along with CDX2/1 expression (cutoff signal intensity = 25). The AIG and HPG samples were clearly separated from the normal samples and were further separated into the AIG-enriched cluster and the HPG-enriched cluster. Moreover, all the AIG samples expressed CDX2/1, while the CDX-negative HPG samples showed a unique cluster

Caudal-type homeobox (CDX) 2 and CDX1 are master regulator genes for intestinal differentiation [17, 18] and are thought to complement each other [19]. All AIG samples and a half of HPG samples showed high expression levels of CDX2/1 (Fig. 3d). To examine whether CDX2/1 plays a potential role in AIG-induced transcriptional changes, the gene expression data of AIG, HPG, and normal mucosa were evaluated according to the expression levels of CDX signature genes, which were characterized as the genes affected in gastric cancer cells stably transfected with CDX [20]. The AIG and HPG samples were clearly separated from the normal samples and were further separated into the AIG-enriched cluster and the HPG-enriched cluster (Fig. 3d). Moreover, all AIG samples expressed CDX2/1, whereas the CDX-negative (signal intensity < 25) HPG samples showed a unique cluster. These data indicate that intestinal trans-differentiation may occur in a CDX-dependent manner.

Gastric mucosa with AIG showed trans-differentiation into pancreas and lung

Next, we identified the top 25 upregulated genes specific to the mucosa with AIG (Fig. 4a, left), which showed low expression levels in the normal and HPG samples (average signal intensity < 25) and minimal expression changes between the normal and HPG samples (< twofold change) (Table S10). Notably, AIG-specific genes include pancreatic digestion-related genes, pancreatic triacylglycerol lipase (PNLIP) [21], carboxyl ester lipase (CEL) [22], and chymotrypsin B1/C (CTRB1 and CTRC) [23]. In addition, a master regulator gene of the lung, NK2 homeobox 1/thyroid transcription factor 1 (NKX2-1/TTF1) [24], and alveolar fluid secretion-related genes, surfactant proteins B and C (SFTPB and SFTPC) [25], were included. Furthermore, the protein–protein interaction (PPI) network analysis using AIG-specific genes showed enrichment of gene sets related to digestion and alveolar lamellar bodies (Fig. 4a, right). A fraction of AIG samples actually expressed these genes (Figs. S3 and S4) and were clearly separated from the other samples (Fig. 4b) using the pancreas- and lung-specific genes (Tables S11 and S12). We also confirmed these ectopic gene expressions (Figs. S5 and S6) by RT-PCR on an additional sample set (Table S1).

Fig. 4

Trans-differentiation into the pancreas and lung in gastric mucosa with AIG. a Protein–protein interaction (PPI) network analysis using AIG-specific genes. The observed networks (Network A and Network B) showed the enrichment of gene sets related to digestion and the alveolar lamellar body. FDR—false discovery rate. b Unsupervised hierarchical cluster analysis using gene expression levels of pancreas-specific and lung-specific genes among the AIG (n = 14), HPG (n = 13), and normal samples (n = 9). A fraction of the AIG samples was clearly separated from the other samples. c Immunostaining of AIG-specific genes related to abnormal differentiation of the pancreas (PNLIP and BCL10) and lung (NKX2-1/TTF1, SFTPB, and SFTPC) using gastric mucosa with AIG. Scale bar: 100 μm

To confirm the level and distribution of protein expression, we performed immunohistochemical analysis of the newly identified gene products in the AIG samples (Figs.4c and S7). PNLIP and BCL10, which are markers of pancreatic acinar cells [26], were co-expressed in a restricted area of the gastric mucosa with AIG, indicating that pancreatic metaplasia [27] causes ectopic pancreas-specific gene expression. In contrast, diffused expression of NKX2-1/TTF1 was observed in gastric mucosa with AIG, along with two types of surfactant proteins, SFTPB and SFTPC, suggesting that pulmonary trans-differentiation universally occurs in the stomach with AIG. Taken together, gastric mucosa with AIG undergoes more diverse trans-differentiation than that with HPG.

In addition, the upregulated genes specific to the mucosa with HPG (Table S13) included cytokines, IL6, CXCL8, and CXCR2 and showed enrichment of multiple gene sets related to the inflammatory response (Fig. S8). CXCL8 is mainly secreted by macrophages and induces neutrophil migration, which is in accordance with the histological findings [28] and the results of the pathway enrichment analysis described above (Fig. 2c).

Environmental acidic condition possibly affects expression levels of intestine-related genes

To explore the impact of AIG on gene expression, we first focused on DNA methylation, as gastric mucosa with AIG showed aberrant DNA methylation at promoter CpG islands (CGI) [6]. Notably, most of the upregulated and downregulated genes had no promoter CGI (Tables S4 and S5), indicating their regulation by TATA box elements. Typically, CGI-regulated genes are crucial for fundamental cellular processes, including housekeeping and tumor suppressor genes [29,30,31]. In contrast, TATA-regulated genes often exhibit regulated or inducible gene expression [32], meaning their expression levels are modulated in response to specific environmental conditions and stimuli.

Therefore, we focused on the environmental changes predominantly observed in the gastric mucosa with AIG. Reflecting the destruction of parietal cells by AIG, the expression levels of ATP4A and ATP4B were drastically decreased in AIG samples compared with those in HPG samples (Fig. S9 and Tables S14 and S15), and a markedly high pH has been reported in the mucosa with AIG [33, 34]. Therefore, we hypothesized that an increased environmental pH might induce abnormal trans-differentiation in the gastric mucosa with AIG. To test this hypothesis, gastric cancer cells (AGS, MKN74, MKN1, and GCIY) were cultured under acidic or control conditions to analyze changes in gene expression. Remarkably, acidic conditions downregulated the expression levels of CDX2, a master regulator gene of the small intestine [17], and stem marker genes LGR5 [35], ASCL2 [36], and OLFM4 [37] (Fig. 5a), and the average fold changes of small intestine-specific genes among the cell lines tended to be downregulated in acidic conditions (Fig. S10). In contrast, there was minimal change in the expression of pancreatic and pulmonary genes identified in the AIG.

Fig. 5

Abnormal intestinal differentiation by environmental acidic conditions. a Gene expression changes in gastric cancer cell lines (AGS, MKN74, MKN1, and GCIY). Acidic conditions downregulated the expression levels of a master regulator gene of the small intestine, CDX2, and stem marker genes, LGR5, ASCL2, and OLFM4. b Unsupervised hierarchical cluster analysis using gene expression levels of pancreas-specific and lung-specific genes among the AIG (n = 14), HPG (n = 13), and normal samples (n = 9). The AIG and HPG samples were clearly separated from the normal samples and were further separated into the AIG-enriched cluster and the HPG-enriched cluster

Next, we comprehensively analyzed the genes that were upregulated and downregulated (> twofold change) under acidic conditions in each cell line and identified the commonly upregulated (n = 7) and downregulated genes (n = 33) (Fig. S11 and Table. S16). Based on these possible pH-dependent genes, the AIG and HPG samples were clearly separated from the normal samples and were further separated into the AIG-enriched cluster and the HPG-enriched cluster (Fig. 5b), suggesting that increased environmental pH may alter gene expression in the gastric mucosa, consequently resulting in abnormal differentiation in the intestine.