IntroductionGastric cancer (GC) ranks as the fifth most prevalent cancer globally and the fourth leading cause of cancer-related mortality among solid tumors, though its prognosis remains poor [1,2,3]. For diagnosis, several classification criteria are used, with Lauren’s criteria serving as the standard for histologic classification, categorizing GC into intestinal- and diffuse-types [4,5,6,7]. The diffuse-type, typically found in women and younger patients, is characterized by single cells or loosely connected cells infiltrating the gastric wall [2, 8, 9]. In contrast, the intestinal-type, which is more commonly found in older patients and men, shows glandular, solid, or intestinal structures, including tubular formations, and is frequently associated with environmental factors like Helicobacter pylori infection [10,11,12]. Recent research indicates that while the intestinal-type is considered genomically more unstable, the diffuse-type is more invasive and displays more aggressive traits in cancer progression [13, 14].Intratumoral heterogeneity (ITH) describes the presence of genetically diverse subpopulations of tumor cells within a single tumor, either intermingled or spatially separated. This diversity results in variations in tumor cell growth, immune response, metabolism, and metastatic potential [15,16,17]. ITH is a well-documented phenomenon in various cancers, including GC [18, 19]. Although significant research has confirmed its existence, more in-depth studies are still required to explore its impact on treatment strategies [20].One-carbon (1C) metabolism, comprised of folate metabolism and methionine metabolism, is known to serve as an important mechanism in regulating cancer cells [21]. 1C metabolism is generally upregulated in cancer, primarily because cells require one-carbon units for nucleotide synthesis, methylation reactions, and for the generation of reducing cofactors [22, 23]. Cancer cells utilize the outputs obtained from these folate and methionine cycles to supply energy to cancer cells, facilitate cancer cell proliferation through nucleic acid synthesis, regulate cancer cell fate through redox control, and maintain homeostasis [24,25,26]. A significant number of 1C metabolism genes play crucial roles in cancer proliferation and survival through purine production. Additionally, drugs such as Methotrexate and 5-fluorouracil (5-FU), which have been used for cancer patients for decades and remain actively utilized today, function by inhibiting key genes within 1C metabolism. The clinical efficacy of these therapies against various cancers clearly demonstrates that cancer cells are highly dependent on 1C metabolism [27, 28].Although various metabolic pathways contribute to cancer pathogenesis, 1C metabolism has attracted attention as a potential therapeutic target due to its observable upregulation in cancer patients. Recent studies have demonstrated a close association between GC and 1C metabolism, and while numerous investigations are currently underway regarding 1C metabolism in GC, further in-depth research remains necessary [29, 30].Patient-derived organoids (PDOs) are three-dimensional models created from tumor cells taken from patients’ primary tumors [31]. They effectively replicate the tissue architecture and cellular composition of the original tumor [32,33,34]. PDOs allow researchers to perform individualized tumor response testing, which plays a key role in discovering potential predictive biomarkers [35,36,37,38]. This study utilized single-nucleus transcriptomics to examine ITH in PDOs from two GC subtypes: intestinal- and diffuse-types [39]. While sequencing analyses of PDOs may not completely capture the spectrum of tumor cell subpopulations found in the primary tumor, we addressed this limitation by validating our results with single-cell RNA sequencing data from human tissue.Our objective is to leverage the findings from this study to confirm consistent features of ITH in both PDOs and human samples. By examining both histological classifications and genetic expression differences in histologically pure GC types, we aim to develop effective treatment strategies.ResultsGenetic features of organoids derived from clonal samples of gastric cancer tissueTissue samples were obtained from patients diagnosed with pure intestinal- and pure diffuse-type GC, and four distinct regions were collected from each sample for organoid culture. Organoids were successfully cultured from all eight samples (Fig. 1A). To assess their genetic characteristics, we performed staining on the cultured organoids. The staining results showed that all eight samples were positive for MLH1, PMS2, MLH2, and MLH6, classifying them as microsatellite stable. Additionally, C-erb2 was found to be negative in all eight samples (Fig. 1B, C).Fig. 1: Clonal selection sites and genetic profiling of organoids derived from gastric cancer tissue.A Intestinal- and diffuse-type gastric cancer specimens marked with regions selected for clonal sampling. B, C Immunohistochemical experiments performed on organoids derived from each clonal sample of gastric cancer specimens, GC141 and GC143, including H&E staining. Scale bars: B 200 μm, C 100 μm.Full size imageDiffuse- and intestinal-type gastric cancers display unique molecular characteristicsAfter collecting tissue samples from two subtypes of GC, we extracted four different regions from each subtype. We were able to successfully culture organoids from all eight samples, and nuclei were isolated from the resulting organoid samples. Single-nucleus RNA sequencing was performed on the nuclei isolated from PDOs derived from four regions of intestinal-type (GC141_C1-4) and four regions of diffuse-type (GC143_C1-4) GC tissues (Fig. 2A). Nine transcriptomic clusters were identified through analysis and projected into two-dimensional space using UMAP. Visualization by histological subtype and clonal origin revealed that cells were uniformly distributed across all conditions (Fig. 2B–D). When examining cell proportions across clusters within the two subtypes, we found no significant differences between the two samples. Similarly, the proportions of clonal samples within each subtype showed minimal variation, although the number of cells in cluster C2 was notably lower in GC141 (Fig. 2E). Through pathway analysis, we observed differences in enriched pathways between the pure intestinal- and pure diffuse-type, as identified using both KEGG and Gene Set Enrichment Analysis (GSEA) HALLMARK gene sets (Fig. 2F, G). Diffuse-type GC is well-known for its high malignancy and elevated stemness. Our data confirmed this, revealing a significant upregulation of cancer stem cell-related pathways in the diffuse-type GC sample GC143, aligning with established findings (Fig. 2H). These results highlight that GC141 and GC143 not only differ histologically as pure intestinal- and pure diffuse-type but also exhibit distinct molecular characteristics.Fig. 2: Distinct molecular characteristics of intestinal and diffuse subtypes in gastric cancer.A Schematic illustration of patient-derived organoid generation and snRNA-seq analysis. B Uniform Manifold Approximation and Projection (UMAP) plots displayed nine individual clusters from 24,340 single cells. C, D The distribution of annotated cell clusters shown on the UMAP, divided into two subtypes, with each cluster distinguished by clonal samples from the respective subtypes. E The cell proportion of clonal samples and nine clusters of each subtype. F Comparison of KEGG pathway enrichment results for DEGs in each subtype. G Comparison of Hallmark pathway enrichment between the two subtypes using Gene Set Enrichment Analysis (GSEA). H Dotplot of cancer stem cell-related pathway expression between the two subtypes.Full size imageIntratumoral heterogeneity in intestinal-type gastric cancer correlates with high CD44 expression and 1C metabolism enrichmentDiffuse-type GC, a major histological subtype, displays a highly invasive phenotype with pronounced malignant and stemness-associated characteristics [40, 41]. Additionally, as previously mentioned cancer stem cell-related pathways were significantly enriched in the diffuse-type GC sample GC143. To investigate this further, we focused on the expression of CD44, a well-established marker of cancer stem cells. CD44 is widely recognized as a key marker for cancer stem cells in various solid malignancies, including GC, and has been extensively studied as both a tumor biomarker and a therapeutic target [42]. As anticipated, CD44 expression was significantly higher in GC143 compared to GC141, a trend that was consistent across all clonal samples (Fig. 3A). Similarly, in all clonal samples of GC143, CD44 expression was elevated. Interestingly, one clonal sample from GC141, specifically GC141_C2, exhibited CD44 levels comparable to those in GC143 (Fig. 3B). When comparing CD44 expression between clusters of GC141 and GC143, CD44 was elevated in most clusters of GC143. However, some clusters of GC141 also exhibited clear expression of CD44 (Supplementary Fig. 1A). Given the well-known ITH of GC, we performed IHC staining for CD44 on all eight organoid clonal samples to validate these findings from single-nuclei transcriptomic analysis [43]. The staining revealed strong signals in all clonal samples of GC143, with GC141_C2 also showing a relatively strong signal, further confirming the presence of ITH (Fig. 3C). Interestingly, consistent results were also observed in CD44 staining of GC tissues. While CD44 expression was generally high across all areas in GC143, GC141 exhibited regions with high CD44 expression alongside areas with lower expression, reflecting the ITH characteristic of GC141 (Supplementary Fig. 1B). Additionally, immunocytochemistry in GC141 organoids confirmed the high CD44 expression in GC141_C2 (Fig. 3D). Since both transcriptomic analysis and experimental validation indicated elevated expression of stemness-related genes in GC141_C2, we used ‘CellChat’ to further analyze the increased communication signals between this CD44-high clonal sample and other clonal samples of intestinal-type GC, GC141 [44]. Among the four clonal samples, GC141_C2 showed significant enrichment in cancer stem cell-related pathway networks, including the ‘NOTCH signaling pathway,’ ‘BMP signaling pathway,’ and ‘NRG signaling pathway,’ compared to the other clonal samples (Fig. 3E). To further explore the enriched pathways in GC141_C2, we conducted a deeper analysis by subsetting GC141 and comparing the differentially expressed genes (DEGs) of GC141_C2 with those of the other clonal samples. Cancer stem cell-related pathways were significantly enriched in KEGG analysis and GC141_C2 also displayed enriched pathways similar to those upregulated in GC143, indicating its diffuse-like characteristics (Fig. 3F, Supplementary Fig. 1C, D). To gain deeper insights into the stemness-related molecular characteristics of GC141_C2, we isolated clusters corresponding to GC141_C2 and divided them into high and low groups based on CD44 expression. The highly CD44-expressing clusters in GC141_C2 exhibited increased activity in 1C metabolism-related pathways (Fig. 3G). Subsequent DEG analysis between ‘GC141_C2 high’ and ‘GC141_C2 low’ revealed that, among all pathways exceeding the significance threshold for p-values, key 1C metabolism-related pathways—such as ‘DNA replication,’ ‘One carbon pool by folate,’ ‘Alanine, aspartate, and glutamate metabolism,’ and ‘Pyrimidine metabolism’—were prominently enriched (Fig. 3H, Supplementary Fig. 1E). DNA replication, essential for cell division and transmission of genetic information, is closely linked to the folate and methionine cycles—core components of 1C metabolism [27, 45]. Both our KEGG and HALLMARK analyses consistently revealed enrichment of proliferation and cell cycle pathways, which are critically dependent on 1C metabolism. This finding further supports that the ‘GC141_C2 high’ population is characterized by enhanced 1C metabolic activity (Supplementary Fig. 1F). Additionally, a comparison of 1C metabolism-related gene expression showed that most genes, including TYMS and SHMT1, were highly expressed in ‘GC141_C2 high’ (Fig. 3I). These findings underscore the clear presence of ITH within intestinal-type GC and highlight the enrichment of 1C metabolism pathways in samples with elevated cancer stem cell-related pathways.Fig. 3: Intratumoral heterogeneity of intestinal-type gastric cancer and 1C metabolism enrichment in CD44-high clusters of intestinal-type gastric cancer.A Vlnplot of CD44 expression between the two subtypes and B each clonal samples. C Immunohistochemical experiments for CD44 staining were performed on each clonal samples from the two subtypes. Scale bars: 100 μm. D Immunocytochemistry of GC141 clonal sample organoids to show the expression of CD44 (Alexa488, green). DAPI (blue) was used for nuclear staining. E ‘CellChat’ chord diagram showing ligand-receptor pairs and their weights contributing to signalings among clonal samples of intestinal-type GC, GC141. F KEGG pathway enrichment analysis of identified DEGs in GC141_C2 compared to other clonal samples of GC141. G Dotplot showing pathway enrichment of ‘CD44-high’ and ‘CD44-low’ expressing clusters of GC141_C2. H KEGG pathway enrichment analysis of identified DEGs in ‘CD44-high’ expressing clusters compared to ‘CD44-low’ expressing clusters of GC141_C2. I Vlnplot displaying the expression of 1C metabolism-related genes between ‘CD44-high’ and ‘CD44-low’ clusters of GC141_C2. Statistical comparisons were performed using two-tailed Student’s t-test (*P