Galectin 3 expression in gastrointestinal tumors identified through comprehensive bioinformatic mapping

Abstract

Background

Gastrointestinal (GI) cancers are a major global health burden with high incidences and mortality. Despite advances in detection and treatment, prognosis for advanced disease remains poor, underscoring the need for new biomarkers and therapies. Galectin-3 (GAL-3), encoded by LGALS3, is a multifunctional β-galactoside-binding protein involved in adhesion, migration, apoptosis, angiogenesis, and immune regulation, making it an important modulator of tumor biology.

Methods

This study used in silico bioinformatic analyses to examine LGALS3 expression and its relationship with immune infiltration across seven GI cancers: cholangiocarcinoma (CHOL), colon adenocarcinoma (COAD), esophageal adenocarcinoma (ESCA), liver hepatocellular carcinoma (LIHC), pancreatic adenocarcinoma (PAAD), rectal adenocarcinoma (READ), and stomach adenocarcinoma (STAD).

Results

LGALS3 was significantly upregulated in CHOL, ESCA, and LIHC, with higher but non-significant expression in PAAD and STAD, while downregulated in COAD and READ. Elevated GAL-3 expression correlated with poorer survival, particularly in LIHC and PAAD. Immune infiltration analysis indicated a role in fostering an immunosuppressive tumor microenvironment. Protein–protein interaction analysis identified MAPK3 and PTEN as key partners, while Gene Ontology enrichment highlighted functions in T-cell activation and motility regulation.

Conclusions

These findings suggest LGALS3 as a promising diagnostic and prognostic biomarker, and a potential therapeutic target to enhance immune responses in GI cancers.

Introduction

Gastrointestinal (GI) cancers comprise more than a quarter of all cancer cases worldwide, with their incidences consistently increasing [1–3]. According to 2020 estimates from the Global Cancer Observatory (GLOBOCAN), there were 19.3 million new cancer diagnoses and 10 million cancer-related deaths globally. GI cancers accounted for 27% of these cases and 36% of the deaths [2, 4]. Despite advances in early detection, surgery, chemotherapy, and targeted therapies [5–7], the overall prognosis for patients with advanced GI cancers remains poor [8]. This situation underscores the urgent need for novel diagnostic markers and therapeutic strategies to enhance patient outcomes. One of the primary challenges in managing GI cancers is their tendency for early metastasis, therapeutic resistance, and the ability to evade the host immune system [8, 9]. In recent years, Galectin-3 (GAL-3), a multifunctional member of the beta-galactoside-binding protein family, has emerged as a key player in the pathogenesis and progression of various gastrointestinal tumors. GAL-3 is ubiquitously expressed in many tissues and is involved in a wide range of cellular functions, including cell adhesion, migration, apoptosis, angiogenesis, and immune regulation. It exhibits a unique ability to interact with diverse ligands, including cell surface glycoproteins and extracellular matrix (ECM) components, influencing critical signaling pathways in tumor cells. In the context of GI cancers, GAL-3 has been increasingly recognized for its role in promoting tumor growth and progression [10–13]. GAL-3 expression has been found upregulated in several GI malignancies, including liver hepatocellular carcinoma (LIHC), stomach adenocarcinoma (STAD), pancreatic cancer (PAAD), and colon cancer (COAD), and is associated with increased tumor cell proliferation, invasion, apoptosis and metastatic potential [13–21]. The expression of GAL-3 has been also linked to the biological characteristics of a small population of tumor-initiating cancer stem cells (CSCs), which are thought to partially drive the recurrence of GI adenocarcinomas following surgery and chemotherapy [20]. GAL-3 contributes to an immunosuppressive tumor microenvironment that promotes immune evasion and cancer progression [22]. It also interacts with signaling pathways like Wnt/β-catenin and Mitogen-Activated Protein Kinase (MAPK), facilitating epithelial-mesenchymal transition (EMT) and metastasis [23–25]. Beyond tumor cell regulation, GAL-3 modulates immune cell function by binding to glycosylated receptors on T cells and macrophages, altering their activity within the tumor microenvironment [26]. GAL-3 can also induce T-cell apoptosis and inhibit T-cell activation, fostering an immunosuppressive environment that enables tumor immune evasion [22]. This highlights GAL-3 as a potential therapeutic target, particularly for enhancing anti-tumor immunity in GI cancers. Additionally, GAL-3 has been implicated in drug resistance, notably in colorectal cancer, where its expression is linked to resistance to chemotherapeutic agents like 5-fluorouracil (5-FU) [27, 28], likely through modulation of apoptotic and survival pathways. Given its roles in tumor progression, metastasis, immune regulation, and therapy resistance, GAL-3 is increasingly recognized as both a biomarker and a therapeutic target. This study employs an extensive bioinformatic analysis to investigate LGALS3 gene expression and function in GI cancers, with the goal of uncovering insights that may aid in the development of targeted immunotherapies to enhance patient outcomes.

Materials and methods

UALCAN analysis

UALCAN (http://ualcan.path.uab.edu) is a public resource to explore TCGA gene expression data. UALCAN allows analysis of relative expression across cancer and normal samples, as well as different cancer subgroups based on individual cancer clinicopathological information [29]. We used UALCAN to analyze LGALS3 expression across seven GI cancers: cholangiocarcinoma (CHOL), colon adenocarcinoma (COAD), esophageal adenocarcinoma (ESCA), liver hepatocellular carcinoma (LIHC), pancreatic adenocarcinoma (PAAD), rectal adenocarcinoma (READ), and stomach adenocarcinoma (STAD). UALCAN was used for exploratory gene expression analysis across gastrointestinal cancer types. The significance of differences was assessed using Student’s t-test, and p < 0.05 was considered statistically significant. Multiple testing correction was not applied, as adjusted p-values are not available within the platform.

c-BioPortal analysis

The cBio Cancer Genomics Portal (http://cbioportal.org) serves as an online resource for analyzing multidimensional cancer genomics data. LGALS3 mutation and expression data for the seven investigated GI cancers were obtained in accordance with the portal’s standard procedures [30–32].

Survival and prognostic analysis

Survival analysis was performed using Kaplan-Meier (KM) curves, generated through the KM-plotter tool (https://kmplot.com/analysis/) [33, 34]. The correlations between LGALS3 probe set (208949_s_at) expression and the overall survival (OS) and relapse-free survival (RFS) for the seven investigated GI cancers were analyzed with associated patient samples separated into two groups by median expression. The hazard ratio (HR) with 95% confidence intervals and log-rank p-value were also determined.

GeneMANIA analysis

GeneMANIA (http://www.genemania.org) is a public resource for building protein–protein interaction (PPI) network, producing hypotheses about gene function, displaying gene lists, and prioritizing genes for functional assays [35]. We constructed a protein–protein interaction (PPI) network for LGALS3 and its associated proteins using GeneMANIA, with search parameters set to a maximum of 20 resulting genes and 10 resulting attributes. These parameter values are commonly used in exploratory GeneMANIA analyses to avoid excessive network density. Sensitivity testing using higher gene limits yielded comparable core interaction patterns, indicating robustness of the network structure.

GEPIA analysis

GEPIA (Gene Expression Profiling Interactive Analysis, http://gepia.cancer-pku.cn/index.html) is an online platform containing RNA sequencing expression data from tumor and normal samples from the TCGA and GTEx projects [19]. We used GEPIA to search for the first 100 LGALS3-related genes derived from all TCGA tumor tissues and corresponding normal counterparts. The thresholds for similar genes of LGALS3 were defined as Pearson correlation coefficient (PCC) r ≥ 0.3.

Gene ontology analysis

Gene Ontology (GO) enrichment analysis was performed using GO annotations from the Bioconductor package org.Hs.eg.db (version 3.19), with GO term mappings obtained from GO.db (version 3.19). The analysis focused on biological processes, cellular components, and molecular functions. Genes co-expressed with LGALS3 were identified through GEPIA using Pearson correlation analysis, and those with a correlation coefficient r ≥ 0.3 were considered biologically relevant. The top 100 positively correlated genes were selected for enrichment analysis to reduce background noise and enhance interpretability. P-values were adjusted for multiple testing using the Benjamini–Hochberg procedure, and q-values were calculated to estimate the false discovery rate. All statistical analyses were conducted using R version 4.4.1.

LGALS3 expression and infiltrating immune cells

We utilized the Tumor Immune Estimation Resource 3 (TIMER 3, https://compbio.cn/timer3/) web tool [36] to assess the relative abundance of six major tumor-infiltrating immune cell types, B cells, CD4⁺ T cells, CD8⁺ T cells, macrophages, neutrophils, and dendritic cells, based on TCGA RNA-Seq data. TIMER applies a constrained least squares deconvolution algorithm to estimate immune cell proportions for each tumor sample. Spearman’s rank correlation was calculated between LGALS3 expression and the infiltration levels of each immune cell type. Statistical significance was defined as p < 0.05. Results were visualized as scatter plots displaying the p-value and partial correlation coefficient (cor).

LGALS expression, cancer-associated cells and endothelial cells

We utilized the TIMER 3.0 web tool (TIMER 3, https://compbio.cn/timer3/) [36] to assess the association between LGALS3 expression and the infiltration of cancer-associated fibroblasts and endothelial cells across the seven GI cancers examined.

Results

LGALS3 expression in GI cancers

Through UALCAN we examined LGALS3 expression levels across various GI cancer samples using data from TCGA, revealing that LGALS3 expression was significantly higher in CHOL (Fig. 1A, p = 3.7 × 10^− 7), LIHC (Fig. 1B, p = 3.7 × 10^− 6) and ESCA (Fig. 1C, p = 1.4 × 10^− 2) all showed elevated LGALS3 levels in tumor samples. In PAAD (Fig. 1D) and STAD (Fig. 1E), expression levels were higher than in normal tissues but did not reach statistical significance (PAAD: p = 5.7 × 10^− 2; STAD: p = 1.4 × 10^− 1). Expression was significantly lower in COAD vs. normal mucosa (Fig. 1F, p < 1 × 10⁻¹²), and similarly reduced in READ (Fig. 1G, p = 1.9 × 10⁻³).

Fig. 1.

Differential Expression of LGALS3 in GI Cancers Based on TCGA Data. Box plots represent LGALS3 gene expression levels in tumor versus normal tissues across seven GI cancer types, analyzed using UALCAN and TCGA datasets. Expression was significantly elevated in A CHOL (p < 0.001), B ESCA (p < 0.05), and C LIHC (p < 0.001). No statistically significant differences were observed in D PAAD or E STAD or F COAD or (G) READ. Statistical significance is denoted as follows: *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001

LGALS3 expression in GI according to demographic and histopathological characteristics

Using the UALCAN platform, we observed statistically significant differences in LGAL3 expression in CHOL between non-tumoral tissues and both stage I (p = 3.01 × 10⁻⁷) and stage II (p = 1.24 × 10⁻²⁴) tumors. Additionally, LGAL3 expression differed significantly between non-tumoral tissues and tumors with nodal metastasis status N0 (p = 5.62 × 10⁻⁹). No significant differences in LGAL3 expression were found in relation to age, race, gender, BMI, or tumor grade. In COAD, LGAL3 expression was significantly different between non-tumoral tissues and all tumor stages—stage I (p = 1.63 × 10⁻¹²), stage II (p = 2.224 × 10⁻¹⁶), stage III (p = 5.79 × 10⁻¹⁴), and stage IV (p = 1.63 × 10⁻¹²). Additionally, LGAL3 expression differed significantly between subjects aged 21–40 and those aged 41–60 (p = 3.69 × 10⁻²), as well as between 21 and 40 and 61–80 years (p = 3.54 × 10⁻²), with higher expression observed in the younger group. A significant difference in LGAL3 expression was also found between non-tumoral tissues and both adenocarcinoma (p < 1 × 10⁻¹²) and mucinous adenocarcinoma (p = 2.66 × 10⁻¹⁵); however, no significant difference was observed between the two histological subtypes. Regarding nodal metastasis, LGAL3 expression was significantly elevated compared to non-tumoral tissues for N0 (p = 1.62 × 10⁻¹²), N1 (p = 2.13 × 10⁻¹²), and N2 (p < 1 × 10⁻¹²) status. Using TP53 mutation status derived from TCGA whole-exome sequencing data, LGAL3 expression was significantly different between non-tumoral tissues and both TP53 mutant (p < 1 × 10⁻¹²) and non-mutant (p = 2.25 × 10⁻¹³) groups. Moreover, expression was significantly higher in non-mutant tumors compared to mutant ones (p = 7.95 × 10⁻³). No significant differences in LGAL3 expression were found based on race, gender, or BMI. In ESCA, LGAL3 expression showed statistically significant differences between non-tumoral tissues and both stage I (p = 1.06 × 10⁻²) and stage III (p = 3.084 × 10⁻²) tumors. Additionally, significant differences were observed between stage I and stage III (p = 2.08 × 10⁻²), as well as between stage I and stage IV (p = 3.00 × 10⁻²). LGAL3 expression also varied significantly by race, with differences noted between Caucasians and African Americans (p = 9.54 × 10⁻⁷), and between Caucasians and Asians (p = 1.58 × 10⁻¹¹). Statistically significant differences in LGAL3 expression were observed across various BMI categories (p < 0.01). Specifically, differences were noted between individuals with normal weight (BMI ≥ 18.5 and < 25) and those classified as overweight (BMI ≥ 25 and < 30), obese (BMI ≥ 30 and < 40), and extremely obese (BMI > 40). Additional significant differences were found between overweight and extremely obese individuals, as well as between obese and extremely obese groups. Age-related differences were also significant, with higher LGAL3 expression in individuals aged 61–80 compared to those aged 41–60 (p = 3.07 × 10⁻³). LGAL3 expression differed significantly between non-tumoral tissues and grade II tumors (p = 2.07 × 10⁻²). Smoking habits showed a significant association (p < 0.01), with higher expression in individuals who had quit smoking within the past 15 years. Likewise, drinking habits were significant (p < 0.01), with higher expression observed in non-drinkers. Regarding histological type, significant differences were found between non-tumoral tissues and adenocarcinoma (p = 2.27 × 10⁻⁶), as well as between adenocarcinoma and squamous cell carcinoma (p = 1.62 × 10⁻¹²). For nodal metastasis status, LGAL3 expression differed significantly between non-tumoral tissues and both N0 and N2 (p < 0.04), and between N0 and both N2 and N3 (p < 0.04). Using TP53 mutation status obtained from TCGA whole-exome sequencing data, LGAL3 expression was significantly different between non-tumoral tissues and both TP53 mutant and non-mutant groups (p < 0.04). No significant differences in expression were observed based on gender. In LIHC, LGAL3 expression was significantly different between non-tumoral tissues and tumor stages I, II, and III (p < 0.01), as well as between stage I and stage III (p = 2.56 × 10⁻²). Significant differences were also observed across tumor grades: between non-tumoral tissues and grades G1, G2, and G3 (p < 0.03), and between grades G2/G3 and G4 (p < 0.01). Regarding nodal metastasis status, LGAL3 expression differed significantly between non-tumoral tissues and both N0 and N1 (p < 0.01). Analysis of TP53 mutation status using TCGA whole-exome sequencing data revealed a statistically significant difference in LGAL3 expression between non-tumoral tissues and both TP53 mutant and non-mutant groups (p < 0.001). No significant differences in LGAL3 expression were observed based on gender, race, age, weight, or tumor histology. In PAAD, LGAL3 expression was significantly different between non-tumoral tissues and stage II tumors (p = 2.85 × 10⁻²). Significant differences were also observed across tumor grades: between non-tumoral tissues and grades G2 and G3 (p < 0.02); between G1 and G2/G3 (p < 0.001); and between G2/G3 and G4 (p < 0.02). For nodal metastasis, LGAL3 expression differed significantly between non-tumoral tissues and N1 status (p = 3.28 × 10⁻²). Using TP53 mutation data from TCGA whole-exome sequencing, LGAL3 expression was significantly different between non-tumoral tissues and TP53 mutant tumors (p = 5.57 × 10⁻³), as well as between TP53 mutant and non-mutant tumors (p = 1.51 × 10⁻⁸). No significant differences in LGAL3 expression were observed with respect to race, gender, drinking habits, diabetes status, or chronic pancreatitis status. In READ, LGAL3 expression was significantly different between non-tumoral tissues and tumor stages I, II, III, and IV (p < 0.01). Significant differences were also observed by histological subtype, with higher expression in both adenocarcinoma and mucinous adenocarcinoma compared to non-tumoral tissues (p < 0.001). Nodal metastasis status showed significant differences between non-tumoral tissues and N0, N1, and N2 (p < 0.01). Using TP53 mutation status from TCGA whole-exome sequencing data, LGAL3 expression was significantly different between TP53 mutant and non-mutant tumors (p < 0.01). No significant differences were found with respect to race, gender, BMI, or age. In STAD, a significant difference in LGAL3 expression was observed between patients aged 41–60 and those aged 61–80 (p = 4.86 × 10⁻³). Tumor grade comparisons showed significant differences between non-tumoral tissues and grade G2 (p = 1.77 × 10⁻²), and between grades G1/G2 and G3 (p < 0.04). Considering Helicobacter pylori infection, tumors from infected patients showed significantly different LGAL3 expression compared to those without infection (p = 1.64 × 10⁻²). It has been recently reported that Helicobacter pylori infection promotes the accumulation of cytosolic GAL-3 around damaged lysosomes in gastric epithelial cells, where GAL-3 facilitates apoptosis independently of the VacA toxin and its glycan-binding activity [37]. These findings identify GAL-3 as a pro-apoptotic host factor that links lysosomal damage, autophagy initiation, and epithelial cell death during Helicobacter pylori infection [37]. In terms of histological subtype, significant differences were found between non-tumoral tissues and intestinal (tubular) adenocarcinoma (p = 2.81 × 10⁻²), and between diffuse and intestinal (tubular) adenocarcinoma (p = 2.5 × 10⁻²). No significant differences were observed for race, gender, nodal metastasis status, or TP53 mutation status.

Genomic alterations of LGALS3 in GI cancers

Using sequencing data from the TCGA database, cBioPortal was utilized to analyze the types and frequencies of LGALS3 alterations across the seven investigated GI cancers (Fig. 2A). Our analysis revealed that LGALS3 alterations varied across GI cancers, although they occurred at low frequencies. Specifically, alterations were identified in 1 out of 36 CHOL patients (2.78%), 9 out of 478 (1.88%) STAD patients, 2 out of 185 (1.08%) ESCA patients, 3 out of 377 (0.8%) LIHC patients, 1 out of 185 (0.54%) PAAD patients, 1 out of 619 (0.16%) COAD patients. No alterations were found in READ patients. The types of alterations varied among the cancer types. In CHOL, there was one case of amplification (2.78%). In STAD, there were 3 cases of mutations (0.63%) and 6 cases of amplification (1.26%). ESCA presented with one case of mutation (0.54%) and one case of amplification (0.54%). For LIHC, we observed one case of mutation (0.26%), one case of amplification (0.27%), and one case of deep deletion (0.27%). PAAD showed a single case of mutation (0.54%) (Fig. 2B). Lastly, COAD showed 1 amplification (0.16%). We also assessed LGALS3 mutation counts across the different alteration states, providing insight into how these genetic changes may impact its expression (Fig. 2C).

Fig. 2.

Genomic alterations of LGALS3 in GI cancers and their impact on expression. A Summary plot of LGALS3 genomic alterations, including mutations, amplifications, and deletions, in GI cancers extracted from TCGA data using cBioPortal. B Bar chart showing the alteration frequencies across cancer types. C LGALS3 mutation counts are compared among samples with and without alterations, illustrating how specific genomic events correlate with transcriptional changes in LGALS3

Survival analysis and LGALS3 in GI cancers

Kaplan–Meier survival analysis demonstrated that high LGALS3 expression was significantly associated with poorer OS in patients with pancreatic adenocarcinoma (PAAD; n = 390; p = 0.0017; Fig. 3A) and liver hepatocellular carcinoma (LIHC; n = 178; p = 0.0011; Fig. 3B). In contrast, no significant association was observed between LGALS3 expression and RFS in either cancer (Fig. 3C and D). For the remaining GI cancer types included in the study, LGALS3 expression levels were not significantly associated with OS or RFS.

Fig. 3.

Kaplan–Meier analysis revealed a statistically significant improvement in overall survival for PAAD (A) and LIHC (B), whereas relapse-free survival was not significant for either neoplasm (C and D). No association was observed between LGALS3 expression levels (low or high) and overall or relapse-free survival in the other GI cancers included in the study

PPI network analysis of interacting genes with LGALS3

The protein-protein interaction (PPI) network of genes interacting with LGALS3 was generated using GeneMANIA. This network comprised 21 nodes and 119 edges (Fig. 4). The top five proteins with the highest degree of interaction were MAPK3 (degree = 6), PTEN (degree = 6), PDCD6IP (degree = 5), CAPN2 (degree = 4), and FCGR2A (degree = 4) (Table 1).

Fig. 4.

Protein–Protein Interaction Network (PPI) of LGALS3 and Associated Genes. The PPI network was generated using GeneMANIA and includes LGALS3 and 20 functionally related genes. Each node represents a gene, and each edge represents an interaction supported by co-expression, shared pathways, or co-localization data. The most connected nodes include MAPK3 and PTEN, suggesting LGALS3’s involvement in critical signaling networks relevant to cancer progression and immune modulation

Table 1.

Key genes identified in the LGALS3 protein–protein interaction (PPI) network and their connectivity

Gene	Degree	Interacting genes
MAPK3	6	AHSG, FCGR2B, FCGR2A, CUBN, ABL1, LGALS3
PTEN	6	AGPS, ABL1, AHSG, LGALS3, LGALS1, CAPN2
PDCD6IP	5	ATP5PB, PTPRT, LGALS1, UACA, LGALS3
CAPN2	4	LGALS1, LGALS3, TMSB10, UACA
FCGR2A	4	AHSG, FCGR2B, LGALS3, ATP5PB
GEMIN4	4	UACA, PTPRT, LGALS3, LGALS1
ABL1	3	AHSG, UACA, LGALS3
IMPA2	3	LGALS3, AHSG, UACA
PTPRT	3	SUFU, AHSG, LGALS3
TMSB10	3	ATP5PB, LGALS1, LGALS3
UACA	3	FCGR2B, PTPRT, LGALS3
ATP5PB	2	SUFU, LGALS3
CUBN	2	PTPRT, LGALS3
FCGR2B	2	LGALS3, CLEC7A
LGALS1	2	LGALS3, FCGR2B
AGPS	1	LGALS3
AHSG	1	LGALS3
CLEC7A	1	LGALS3
CYHR1	1	LGALS3
SUFU	1	LGALS3

This table lists genes identified through GeneMANIA-based PPI network analysis as directly interacting with LGALS3

The “Degree” column indicates the number of direct interactions (edges) each gene has within the network

The “Interacting genes” column provides the names of the individual genes connected to each node

High-degree genes such as MAPK3 and PTEN suggest central roles in LGALS3-mediated regulatory networks relevant to cancer progression, immune modulation, and signal transduction

Genes correlation and functional annotation of similar genes of LGALS3

Gene correlation analysis performed using GEPIA identifying the top 10 genes most strongly correlated with LGALS3 expression (Table 2), measured using the Pearson correlation coefficient (PCC). The analysis reveals that ETHE1 and MISP have the highest correlation with LGALS3, both with a PCC of 0.57 and 0.53 respectively. The top 10 similar genes identified in Table 2 with LGALS3 were selected for GO pathway enrichment analysis.

Table 2.

Top 10 genes most strongly correlated with LGALS3 expression in TCGA tumor datasets

Gene Symbol	Ensemble ID	PCC
ETHE1	ENSG00000105755.7	0.57
MISP	ENSG00000099812.8	0.53
GPA33	ENSG00000143167.11	0.53
MYO1D	ENSG00000176658.16	0.52
CDX1	ENSG00000113722.16	0.51
LGALS4	ENSG00000171747.8	0.51
CDH17	ENSG00000079112.9	0.49
LINC00483	ENSG00000167117.8	0.49
PLS1	ENSG00000120756.12	0.49
PPP1R14D	ENSG00000166143.9	0.47

The table reports the ten genes with the highest Pearson correlation coefficients (PCC) with LGALS3, as identified via GEPIA analysis of transcriptomic data across TCGA tumor types

Each entry includes the official gene symbol, corresponding Ensembl gene identifier, and the PCC value indicating the strength of association

These genes were used as input for subsequent GO enrichment analysis to explore the functional context of LGALS3-associated gene regulation in GI cancers

Enrichment analysis in the Biological Process category revealed a concerted emphasis on cytoskeletal organization, nutrient absorption and protein localization. The eight most significantly enriched pathways were actin filament organization (p = 8.00 × 10^− 6), intestinal absorption (p = 2.80 × 10^− 5), actin filament–based movement (p = 3.70 × 10^− 5), digestive system process (p = 1.06 × 10^− 4), microvillus organization (p = 1.66 × 10^− 4), digestion (p = 3.72 × 10^− 4), regulation of protein localization to cell periphery (p = 3.84 × 10^− 4) and regulation of wound healing (p = 3.97 × 10^− 4). Together, these results point to an integrated program in which dynamic remodeling of actin structures facilitates both absorptive functions, via microvillus assembly in epithelial surfaces, and directed intracellular trafficking of proteins. The concurrent enrichment of digestive and absorptive processes further suggests a functional coordination between cytoskeletal rearrangements and metabolic activity, while wound-healing regulation indicates an overlay of repair mechanisms that may be engaged under the experimental conditions (Fig. 5A). Within the Molecular Function domain (Fig. 5B), activities associated with cytoskeletal motors, filament binding and enzymatic catalysis predominated. The top eight terms comprised microfilament motor activity (p = 1.86 × 10^− 8), actin filament binding (p = 4.84 × 10^− 8), actin binding (p = 8.53 × 10^− 7), cytoskeletal motor activity (p = 1.45 × 10^− 5), catalytic activity acting on glycoproteins (p = 1.66 × 10^− 4), calmodulin binding (p = 4.10 × 10^− 4), acetylglucosaminyltransferase activity (p = 1.23 × 10^− 3) and structural constituent of the cytoskeleton (p = 1.92 × 10^− 3). These enrichments underscore a dual focus on mechanical force generation and substrate modification: motor proteins and filament‐binding factors drive dynamic cellular movements, while glycoprotein‐processing enzymes and calcium‐dependent interactors modulate extracellular matrix assembly and signaling cascades.

Fig. 5.

Gene ontology (GO) enrichment analysis of genes correlated with LGALS3 expression. Plots representing the top significantly enriched GO terms among the 100 genes most strongly correlated with LGALS3, based on GEPIA analysis. A Biological processes include actin filament organization, intestinal absorption and microvillus organization. B Molecular functions include microfilament motor activity. C Cellular component terms highlight associations with cluster actin-based cell projection. Enrichment is expressed as − log₁₀(p-value)

Analysis of the Cellular Component (Fig. 5C) ontology highlighted membrane projections and specialized compartments as principal locales for the differentially expressed gene products. The eight most overrepresented components were microvillus (p = 3.60 × 10^− 10), brush border (p = 8.72 × 10^− 10), cluster of actin-based cell projections (p = 3.97 × 10^− 8), basal plasma membrane (p = 4.21 × 10^− 8), actin-based cell projection (p = 4.89 × 10^− 8), basal part of cell (p = 8.10 × 10^− 8), myosin complex (p = 1.66 × 10^− 7) and basolateral plasma membrane (p = 1.58 × 10^− 6). This pattern indicates that the molecular machinery underpinning absorptive and motile functions is predominantly assembled at apical and basal cell surfaces, with actin-rich projections and myosin assemblies facilitating both nutrient uptake and cellular adhesion. Such spatial specialization reinforces the link between cytoskeletal dynamics and polarized cell function revealed in the other ontologies.

LGALS3 expression and immune cells

We investigated the association between LGALS3 expression levels and immune cell infiltration across various cancer types using data from TIMER3. In CHOL, LGALS3 expression was positively correlated with neutrophil infiltration (r = 0.399, p = 1.95 × 10^− 2) as well as with CD8⁺ T cell infiltration (r = 0.404, p = 1.78 × 10^− 2) (Fig. 6A). In COAD, LGALS3 expression exhibited negative correlations with neutrophil infiltration (r = −0.124, p = 4.15 × 10^− 2) and macrophage infiltration (r = −0.214, p = 4.15 × 10^− 2), while showing a positive correlation with B cell infiltration (r = 0.185, p = 2.24 × 10^− 3) (Fig. 6B). In ESCA, LGALS3 expression was negatively correlated with neutrophil infiltration (r = −0.197, p = 8.08 × 10^− 3) and positively correlated with CD4⁺ T cell infiltration (r = 0.208, p = 5.30 × 10^− 3) (Fig. 6C). In LIHC, LGALS3 expression demonstrated positive correlations with neutrophil infiltration (r = 0.387, p = 9.04 × 10^− 14), CD8⁺ T cell infiltration (r = 0.270, p = 4.83 × 10^− 5), CD4 + T cell infiltration (r = 0.200, p = 1.88 × 10^− 4), myeloid dendritic cell infiltration (r = 0.465, p = 6.51 × 10^− 20), macrophage infiltration (r = 0.325, p = 6.32 × 10^− 10), and B cell infiltration (r = 0.221, p = 3.57 × 10^− 5) (Fig. 6D). In PAAD, LGALS3 expression correlated positively with B cell infiltration (r = 0.243, p = 1.38 × 10^− 3) and negatively with macrophage infiltration (r = −0.154, p = 4.53 × 10^− 2) (Fig. 6E). In STAD, LGALS3 expression was negatively correlated with CD4 + T cell infiltration (r = −0.154, p = 2.74 × 10^− 3), CD8 + T cell infiltration (r = −0.143, p = 6.35 × 10^− 3), myeloid dendritic cell infiltration (r = −0.119, p = 2.08 × 10^− 2), and macrophage infiltration (r = −0.162, p = 1.61 × 10^− 3) (Fig. 6F). No significant correlation between LGALS3 expression and immune cell infiltration were observed in READ.

Fig. 6.

Association of LGALS3 expression with immune cell infiltration in GI cancers. A In CHOL, LGALS3 expression was positively correlated with neutrophil and CD8⁺ T cell infiltration. B In COAD, LGALS3 showed a negative correlation with neutrophil and macrophage infiltration, but a positive correlation with B cell infiltration. C In ESCA, LGALS3 was negatively correlated with neutrophils and positively correlated with CD4⁺ T cells. D In LIHC, LGALS3 expression positively correlated with multiple immune cell types, including neutrophils, CD8⁺ and CD4⁺ T cells, myeloid dendritic cells, macrophages, and B cells. E In PAAD, LGALS3 was positively correlated with B cells and negatively with macrophages. F In STAD, LGALS3 expression showed negative correlations with CD4⁺ and CD8⁺ T cells, myeloid dendritic cells, and macrophages. No significant correlations between LGALS3 expression and immune cell infiltration were observed in READ

LGALS expression and cancer-associated cells and endothelial cells

We utilized the Tumor Immune Estimation Resource 3 (TIMER 3) web tool to assess the association of LGALS3 expression with cancer-associated fibroblasts (CAFs) and endothelial cells in GI cancers. A statistically significant correlation between CAFs and LGALS3 expression was observed across various gastrointestinal tumors (Fig. 7), though this association is highly dependent on the analytical algorithm employed [36]. Except for ESCA and LIHC, where endothelial cell abundance was found to be significantly positively correlated with LGALS3 levels (p < 0.05), the other GI cancers analyzed exhibited an inverse correlation. No significant correlation was observed in CHOL and READ (Fig. 7).

Fig. 7.

Association of LGALS3 expression with cancer associated fibroblasts and endothelial cell in GI cancers. Scatter plots illustrate correlations between LGALS3 expression and infiltration levels of cancer-associated fibroblasts and endothelial cells across five GI cancer types, using the TIMER platform. Immune score (i.e. a quantitative measure of the infiltration of immune cells in the tumor microenvironment, often derived from gene expression data. It reflects the immune activity within the tumor tissue.) and stromal score (i.e. A metric estimating the presence of stromal cells in the tumor microenvironment, also based on gene expression profiles. It indicates the extent of the non-tumor, supportive tissue within the tumor) were also explored

Conclusions

GI cancers remain among the most aggressive malignancies, contributing significantly to cancer-related morbidity and mortality worldwide [1]. Cancers of the digestive tract present unique challenges in diagnosis and treatment due to their tendency for early metastasis and resistance to conventional therapies. Despite advances in surgical techniques, chemotherapy, and targeted therapies, patient prognosis, particularly in advanced stages, remains poor [38]. The critical role of the tumor microenvironment (TME), especially the interplay between tumor cells and the immune system, is increasingly recognized as a key factor in tumor progression and therapy resistance in GI cancers. The TME plays a critical role in tumor progression and response to therapy [39]. It is composed not only of cancer cells but also various noncancerous cells, such as fibroblasts, endothelial cells, neurons, adipocytes, and both innate and adaptive immune cells. The TME also includes noncellular elements, such as the extracellular matrix (ECM), soluble factors like chemokines, cytokines, growth factors, and extracellular vesicles. This understanding has driven a growing interest in uncovering molecular targets that could enhance immune surveillance and improve the efficacy of current treatments [40, 41].

Galectins, a family of carbohydrate-binding proteins that recognize galactose-containing glycans and are expressed by various cell types, have been shown to regulate signaling pathways through multivalent interactions, thereby influencing cell fate and function [42–45]. In cancer, they drive transformation, angiogenesis, immune evasion, and metastasis, making them key therapeutic targets [44, 46–49]. A key regulatory mechanism in the TME involves the glycoprotein GAL-3. GAL-3 binds to the TCR at the immunological synapse on the cell surface, thereby limiting TCR mobility, promoting TCR downregulation, and inhibiting early T-cell activation via the TCR signaling pathway [50]. GAL-3 is a structurally distinct glycoprotein, extensively studied in various pathological states, including fibrosis, inflammation, and cancer [51–54]. As a member of the lectin family, GAL-3 belongs to one of the 14 identified mammalian galectins, which bind specifically to ß-glycoside structures. These galectins are categorized into three groups based on their conserved carbohydrate-recognition-binding domain (CRD) structures: prototypes, tandem repeats, and chimera groups. Research shows that GAL-3 expression increases during cancer progression, contributing to negative outcomes such as enhanced tumor growth, invasiveness, and metastasis [55–57]. Notably, GAL-3 impacts more cancer types than other galectins. Given its wide array of functions in both tumor cells and the immune microenvironment, GAL-3 represents a potential biomarker and therapeutic target in GI cancers, particularly in the development of immune-based therapies. Our in silico analysis, based on TCGA data, offers valuable insights into the relationship between LGALS3 expression and seven gastrointestinal cancers: CHOL, COAD, ESCA, LIHC, PAAD, READ, and STAD.

Through UALCAN, LGALS3 expression levels reveal that LGALS3 expression was significantly higher in CHOL, LIHC, and ESCA all showed elevated LGALS3 levels in tumor samples. Although not statistically significant, LGALS3 expression appeared elevated in PAAD and STAD. In contrast, LGALS3 expression was significantly lower in COAD compared to normal mucosa and similarly reduced in READ. As a preliminary assessment, we examined GAL-3 expression patterns in human tissues using the Human Protein Atlas (HPA; https://www.proteinatlas.org/), an open-access resource that integrates protein expression data from normal and cancer tissues [58]. This analysis was conducted to illustrate histological differences in GAL-3 expression across gastrointestinal cancer specimens. Notably, GAL-3 displayed heterogeneous expression patterns, both in staining intensity and spatial distribution, across different GI tumor types. These observations support the notion that GAL-3 expression is context-dependent and underscore the need for prospective validation in well-characterized clinical cohorts. Future studies integrating real-world data, including epidemiological and clinical variables, as well as radiological and histopathological findings, will be essential to substantiate these preliminary observations and to better define the potential biological and clinical relevance of GAL-3 in GI malignancies. While overall GAL-3 levels may not differ markedly between neoplastic and cancer cells, its distribution varies, with metastatic COAD cells exhibiting higher galectin-3 expression. This suggests a role for GAL-3 in promoting colon cancer metastasis [12, 17]. The statistically significant expression of LGALS3 in CHOL, ESCA and LIHC represent an interesting finding suggesting the potential application of immunotherapy targeted for these malignancies. Especially LIHC survival analysis showed a statistically significant worse prognosis for patients with high expression of LGALS3, making it even more interesting for targeted therapy. In LIHC, LGALS3 expression showed positive correlations with the infiltration of neutrophils, CD8⁺ T cells, CD4⁺ T cells, myeloid dendritic cells, macrophages, and B cells. In CHOL, it was positively associated with neutrophil and CD8⁺ T cell infiltration. In ESCA, LGALS3 expression was negatively correlated with neutrophil infiltration but positively correlated with CD4⁺ T cell infiltration. Additionally, Kaplan–Meier analysis revealed a statistically significant improvement in overall survival (OS) for PAAD. In this cancer type, LGALS3 expression was positively associated with B cell infiltration and negatively associated with macrophage infiltration.

In addition to exploring LGALS3 impact on immune infiltration, we also conducted a PPI network analysis to identify key molecular partners of LGALS3. The PPI network revealed that LGALS3 interacts with several proteins known to be involved in critical biological processes related to cancer progression, including MAPK3, PTEN, PDCD6IP, CAPN2, and FCGR2A. These interactions suggest that LGALS3 participates in signaling pathways that regulate tumor cell motility, immune evasion, and cell survival. For instance, MAPK3 is a well-established mediator of cell proliferation and survival, while PTEN is a well-known tumor suppressor involved in cell cycle regulation and apoptosis. The interaction of LGALS3 with PDCD6IP (which plays a role in apoptosis) and FCGR2A (a receptor involved in immune responses) further underscores LGALS3 role in both modulating the tumor microenvironment and promoting tumor cell survival. These findings highlight the potential for targeting LGALS3’s interactions with these key proteins as part of therapeutic strategies aimed at inhibiting pathways that support tumor growth and immune suppression. By disrupting these interactions, it may be possible to enhance the efficacy of current treatments or develop novel approaches to combat resistance in GI cancers. Our GO analysis provided additional insights into the biological processes, cellular components, and molecular functions associated with LGALS3 and its top correlated genes. The GO enrichment analysis revealed significant associations across several biological processes that are critical to cancer development and immune regulation.

Among the biological processes, notable pathways included regulation of protein localization to the cell periphery, T cell activation, lymphocyte activation involved in immune response, and antimicrobial peptide production. These findings suggest that LGALS3 and its correlated genes are involved in regulating immune responses, particularly through the activation and regulation of T cells, which are key players in anti-tumor immunity. The involvement of LGALS3 in antimicrobial peptide production also points to its role in modulating the immune defense mechanisms within the tumor microenvironment. For cellular components, the analysis highlighted significant enrichment in the cortical actin cytoskeleton, brush border, and actin filament categories. These components are essential for maintaining cell shape, motility, and intracellular signaling, all of which are critical for tumor progression and metastasis. The connection between LGALS3 and actin filament binding supports its involvement in promoting cell motility and metastasis, particularly through the EMT process, which is commonly observed in advanced GI cancers. In terms of molecular functions, LGALS3 was significantly associated with actin filament binding, actin binding, and oligopeptide transmembrane transporter activity. These functions further suggest that LGALS3 plays a crucial role in regulating cell motility, cytoskeletal organization, and possibly nutrient transport mechanisms, all of which are key elements in supporting tumor growth and invasion. The GO enrichment analysis provides a comprehensive view of LGALS3’s functional role within the cell and its potential contribution to cancer progression. The involvement of LGALS3 in immune-related processes, coupled with its interactions with proteins that regulate tumor cell behavior, underscores its multifaceted role in modulating both tumor growth and immune evasion. These findings support the idea that LGALS3 is not only a critical modulator of the tumor microenvironment but also a promising therapeutic target in GI cancers. Genetic alterations in LGALS3 have been found relatively rarely across GI cancers, but their type and frequency vary among cancer subtypes. The highest alteration frequencies were observed in CHOL and STAD, suggesting potential subtype-specific relevance. In contrast, no LGALS3 alterations were detected in READ, indicating possible tumor-specific genetic stability or alternative regulatory mechanisms. These findings highlight the heterogeneous nature of LGALS3 involvement in GI malignancies.

In several cancers, including CHOL, COAD, ESCA, LIHC, and READ, LGALS3 expression increases with disease progression or nodal metastasis, as evidenced by its association with demographic and histopathological characteristics in GI cancers. These findings are consistent with previous reports [12, 59].

Given the TME’s significant role in promoting cancer progression and metastasis, efforts to modulate the TME by reducing immunosuppression and enhancing immune activation have gained momentum in cancer immunotherapy. The substantial evidence of GAL-3 involvement in promoting tumor growth, metastasis, and immune suppression makes it an attractive target for therapeutic intervention [60, 61]. Inhibiting GAL-3 in solid tumors, especially when combined with T-cell checkpoint blockade or T-cell agonists, holds promise for enhancing anti-tumor immunity and improving tumor regression.

Preclinical studies have shown that treatment with the GAL-3 inhibitor GR-MD-02 enhances antigen-specific T-cell expansion in vivo [61, 62]. Combining GR-MD-02 with the agonist anti-OX40 antibody improved survival in MCA-205 sarcoma, 4T1 mammary carcinoma, and TRAMP-C1 prostate cancer models, and reduced lung metastases in the 4T1 model. GR-MD-02 also showed strong anti-tumor effects when paired with CTLA-4 or PD-1 inhibitors in various murine models. These findings supported its evaluation in two phase 1 trials, where GR-MD-02 was combined with ipilimumab or pembrolizumab in patients with metastatic melanoma, head and neck squamous cell carcinoma, and NSCLC. Overexpression of GAL-3 has been linked to poorer prognosis in several non-GI cancers, including ovarian carcinoma, nasopharyngeal carcinoma, malignant melanoma, nervous system and osteosarcoma [17, 63–68]. However, in laryngeal squamous-cell carcinoma, clear cell renal carcinoma, and breast carcinoma, GAL-3 overexpression has been associated with improved prognosis [11, 69–71]. Our in silico analysis provides compelling evidence that LGALS3 plays a key role in modulating the tumor-immune interface in GI cancers. The differential effects of LGALS3 on immune cell infiltration across various GI cancers underscore the complexity of its involvement in tumor progression and immune evasion. These findings lay a foundation for future studies aimed at targeting LGALS3 to enhance immune responses, particularly in cancers where it exerts an immunosuppressive effect, potentially improving outcomes for patients with GI malignancies. We identified GAL-3 as playing a pivotal role in CHOL, ESCA, and LIHC (Fig. 7). Despite this, the association between GAL-3 and CHOL, particularly intrahepatic CHOL, remains still underexplored. Shimura et al. [72] first reported a connection between GAL-3 expression and prognosis in extrahepatic CHOL, showing that patients with GAL-3-positive tumors had poorer prognoses than those who were GAL-3-negative. In distal CHOL, GAL-3 was identified as the only independent prognostic factor in their study. However, while GAL-3 levels were higher in tumor cells compared to adjacent normal bile duct epithelia, no clear association with prognosis was found. Although little attention has been paid to the subcellular distribution of GAL-3 in relation to prognosis, its overexpression is known to promote key tumor cell functions, including resistance to apoptosis, therapeutic resistance, proliferation, and migration. Junking et al. [73] immunohistochemically examined 53 patients with intrahepatic CHOL who underwent surgery without preoperative therapy. They found that bile duct epithelium expressed GAL-3 at varying intensities depending on histological subtype, with poorly differentiated types expressing GAL-3 less intensely than well- to moderately-differentiated types. Additionally, they noted an association between low GAL-3 expression and lymphatic invasion. Suppressing GAL-3 expression in two CHOL cell lines via GAL-3-targeted siRNA significantly increased cell migration and invasion without affecting proliferation. Thus, regulating GAL-3 expression may be a promising therapeutic approach to control metastasis in CHOL. The role of galectins in ESCA remains poorly understood. An overview of current knowledge highlights the need for further research to clarify the function of galectins in ESCA progression and to identify potential therapeutic opportunities. In particular, the functional role of LGALS3 in esophageal squamous cell carcinoma (ESCC) has been studied by Liang and colleagues [74], who investigated the effects of LGALS3 overexpression in ESCC cells. Using viral transduction to increase GAL-3 expression in the Eca-109 cell line, they compared the parental and GAL-3-overexpressing cells in terms of proliferation, apoptosis, migration, and invasion. Although GAL-3 levels only increased by 1.5-fold, the overexpressing cells exhibited reduced apoptosis and increased proliferative, migratory, and invasive capabilities. This suggests that GAL-3 may contribute to ESCC progression and could serve as a therapeutic target, even though its diagnostic or prognostic value remains limited. In a follow-up study [75], the same group used siRNA to knock down GAL-3 expression in Eca-109 cells, resulting in a 65–90% reduction in GAL-3 levels. This knockdown led to inhibited cell proliferation after 72 h, reduced migration and invasion, and increased apoptosis. These findings suggest that targeting GAL-3 could be a promising therapeutic approach for ESCC. Supporting this, Cui et al. [76] demonstrated that knocking down GAL-3 sensitized ESCC cells to EGFR-targeted therapy with gefitinib by reducing EGFR internalization. This effect was confirmed in vivo, where GAL-3 knockdown enhanced tumor sensitivity to gefitinib, similar to the inhibition of dynamin-dependent endocytosis. Additionally, Xu et al. [77] found that synephrine, a compound isolated from citrus tree leaves, inhibited the proliferation, migration, invasion, and colony formation of two ESCC cell lines (KYSE30 and KYSE270). In a xenograft mouse model, synephrine also reduced tumor growth and sensitized cells to 5-FU treatment. Proteomic analysis revealed that synephrine reduced AKT and ERK signaling, with GAL-3 identified as an upstream regulatory protein. These findings suggest that targeting GAL-3 in ESCC may provide dual benefits: direct effects on tumor cells and increased sensitivity to combination treatments. Nangia-Makker et al. [78] showed that orally administered modified citrus pectin (MCP) inhibits tumor growth, angiogenesis, and metastasis in vivo, likely by targeting GAL-3. These findings highlight the potential of dietary carbohydrates in cancer prevention and therapy. GAL-3 is an emerging biomarker of active inflammation and disease severity, with potential clinical utility, either alone or alongside existing scoring systems, for distinguishing advanced cirrhosis and predicting post-transplant infectious complications [79]. GAL-3 expression in LIHC appears independent of hepatitis B virus (HBV) status, with both HBV-positive and HBV-negative patients demonstrating high levels of GAL-3. However, co-transfection studies indicate that HBV-X protein can transactivate the GAL-3 promoter, suggesting that HBV infection may influence GAL-3 expression. GAL-3 overexpression may contribute to tumor transformation, invasiveness, and cell survival, and its presence in cirrhotic liver, particularly in regenerating nodules, may signal early neoplastic events due to its association with high mitotic activity. In contrast to other GI cancers, GAL-3 appears to have a less significant impact on PAAD. However, a recent study showed that GAL-3 promotes PAAD progression and immunosuppression through both intrinsic cancer cell mechanisms and immune-related pathways [11]. This study proposed that combined therapies targeting GAL-3, the CXCL12-CXCR4 axis, and PD-1 could offer a novel treatment strategy for PAAD [11]. It has been shown that GAL-3 loss induces broad transcriptomic alterations affecting cell cycle progression, stress responses, and morphogenetic pathways across cancer cell lines, while high LGALS3 expression is associated with poorer survival in multiple tumor types, including PAAD [80]. Together, these findings position GAL-3 as a key regulator of oncogenic programs and a potential prognostic biomarker and therapeutic target in selected cancers. Interestingly, two novel non-carbohydrate small molecules, K2 and L2, were identified as potent GAL-3 inhibitors that bind the canonical carbohydrate-recognition domain, alter GAL-3 conformation, and suppress its pro-tumorigenic and pro-inflammatory activities. These compounds effectively inhibit cancer cell adhesion, invasion, angiogenesis, and macrophage cytokine secretion in vitro, reduce tumor growth and metastasis in vivo, and exhibit no detectable cytotoxicity or genotoxicity, representing a promising new class of GAL-3–targeted therapeutics for cancer and fibrosis- or inflammation-associated diseases [81].

Several limitations of this study should be acknowledged. First, due to its in silico design, the findings are inherently exploratory. P-values derived from UALCAN were not adjusted for multiple testing and should therefore be interpreted with caution and validated independently. In addition, the analyses relied on publicly available TCGA datasets, which are characterized by substantial biological and clinical heterogeneity, including variability in tumor purity, disease stage, treatment history, and molecular subtypes. Such heterogeneity may introduce bias into gene expression comparisons. Sample sizes also differed markedly among gastrointestinal cancer types, particularly with respect to normal tissue controls, potentially affecting statistical power and contributing to false-positive or false-negative results. Moreover, the use of bulk RNA-sequencing data does not capture TME complexity and observed expression patterns may partly reflect stromal or immune cell contributions rather than tumor cell–intrinsic expression. Finally, the absence of experimental validation precludes definitive conclusions regarding the functional or clinical relevance of LGALS3 and its associated molecular networks. Confirmation of these findings will require independent patient cohorts and mechanistic validation using in vitro and in vivo models. Despite these limitations, this study provides a systematic bioinformatic framework that may guide future experimental and translational investigations of LGALS3 in GI cancers. Although bioinformatic analyses are valuable for identifying molecular trends across large datasets, experimental validation is necessary to confirm these findings (Figs. 8 and 9). While in silico bioinformatic analyses provide valuable insights, they are limited by their reliance on publicly available datasets, which may vary in quality and completeness [82]. These studies cannot establish causality and may not fully capture the biological complexity of tumor microenvironments. Moreover, their findings require experimental validation to confirm clinical relevance. However, in silico bioinformatic studies are crucial for uncovering molecular patterns, predicting gene function, and identifying potential biomarkers or therapeutic targets. They enable large-scale data analysis quickly and cost-effectively, guiding experimental research and accelerating discoveries in cancer biology and immunology. In vitro and in vivo studies will be crucial to determine the functional roles of GAL-3 in modulating immune cell behavior within the tumor microenvironment. Additionally, exploring the mechanisms by which GAL-3 influences immune cell infiltration, whether through direct interaction with immune cells or by altering cytokine and chemokine networks, will be essential for developing effective therapeutic strategies. We acknowledge that the role of GAL-3 in gastrointestinal tumorigenesis has been investigated in prior studies. However, the present work is not intended to replicate existing mechanistic investigations. Instead, it provides an integrative, systems-level bioinformatic framework that synthesizes publicly available transcriptomic and proteomic data across multiple gastrointestinal cancer types. The novelty of this study lies in its methodological approach, which combines expression profiling, mutation analysis, survival assessment, protein–protein interaction networks, and immune infiltration analyses in a unified and reproducible workflow. By comparing multiple tumor entities within a single analytical framework, our study highlights inter-tumor heterogeneity, identifies co-expression and interaction patterns, and reveals subtype-specific prognostic and immune associations of LGALS3. These findings are intended to be hypothesis-generating and to inform the design of future experimental and translational studies rather than to propose new molecular mechanisms. In contrast to prior work by Ilmer et al. [20], which experimentally characterized GAL-3–positive cancer stem cell subpopulations using in vitro and in vivo models, our study does not address cellular mechanisms or stemness biology directly. Similarly, unlike the study by Fan et al. [83], which focused on a specific gastric cancer subtype and experimentally elucidated GAL-3–CD47–mediated immune evasion, our analysis adopts a broader perspective across gastrointestinal malignancies. Thus, while previous studies provide mechanistic insight in defined biological contexts, our work offers a complementary, clinically oriented, and hypothesis-generating bioinformatic framework that contextualizes LGALS3 across GI cancers and helps prioritize tumor types and immune-related pathways for future validation. In conclusion, this study provides an integrative bioinformatics analysis of LGALS3 expression and its potential clinical relevance across GI cancers. Our findings indicate that LGALS3 shows subtype-specific expression patterns and prognostic associations, with more consistent relevance observed in PAAD and LIHC. However, the heterogeneity of results across cancer types and the lack of experimental and clinical validation preclude definitive conclusions regarding the diagnostic or prognostic utility of LGALS3. Therefore, LGALS3 should be considered a candidate biomarker warranting further investigation rather than a validated clinical marker. Future studies using independent patient cohorts and in vitro and in vivo models are required to confirm its biological and clinical significance. It is essential to validate our findings through additional techniques, such as immunohistochemistry and multiplex fluorescence, alongside clinical data. This could deepen our understanding of the role of galectins in GI tumors and improve therapeutic approaches targeting this family of proteins.

Fig. 8.

Schematic Overview of the Bioinformatics Workflow Used to Evaluate LGALS3 in GI Cancers. This diagram outlines the bioinformatics analysis, including data exploration from TCGA, expression analysis via UALCAN and GEPIA, mutation profiling through cBioPortal, survival assessment using KM-plotter, PPI network construction via GeneMANIA, immune infiltration analysis via TIMER 3, and functional annotation using GO enrichment analysis. The flowchart underscores the integration of multiple web-based platforms for a comprehensive evaluation of LGALS3’s role in GI tumor biology, defining the most potential target (i.e. cancer type) for further experimental evaluations

Fig. 9.

Schema of a multi-layered bioinformatic approach and its principal findings, underscoring LGALS3’s potential as both a prognostic biomarker and a therapeutic target in selected malignancies. It presents the stepwise analysis applied to gastrointestinal cancers, encompassing expression profiling, mutation assessment, survival analysis, protein–protein interaction mapping, and immune infiltration evaluation

Author contributions

Maurizio Chiriva-Internati: Conceptualization, data collection, formal analysis, writing original draft, writing, reviewing and editing, and supervision. Fabio Grizzi: Conceptualization, data collection, formal analysis, writing original draft, writing, reviewing and editing, and supervision. Mohamed A.A.A. Hegazi: data collection, formal analysis, writing original draft, Writing, reviewing and editing. Jose A. Figueroa: Writing, reviewing and editing, and supervision. Robert S. Bresalier: Writing, reviewing and editing, and supervision. The authors have read and approved the final manuscript.

Funding

The author(s) report there is no funding associated with the work featured in this article.

Data availability

The datasets used in this study are publicly available and can be accessed from the respective repositories. All data supporting the findings of this study are provided within the manuscript and are also available through publicly accessible databases. Open access data for CHOL, COAD, ESCA, LIHC, PAAD, READ, and STAD were retrieved from the TCGA database through the UALCAN portal ([http://ualcan.path.uab.edu/analysis.html](http:/ualcan.path.uab.edu/analysis.html)) to evaluate gene expression differences between tumor and normal samples across various clinical characteristics. Survival analysis was performed using the Kmplotter tool ([https://kmplot.com/analysis/](https:/kmplot.com/analysis)). The associations of LGALS3 with immune cell infiltration, cancer-associated fibroblasts, and endothelial cells were analyzed using TIMER 3 ([https://compbio.cn/timer3/](https:/compbio.cn/timer3)). Furthermore, the public databases CBioPortal ([http://cbioportal.org](http:/cbioportal.org)), GeneMANIA ([http://www.genemania.org](http:/www.genemania.org)), and GEPIA ([http://gepia.cancer-pku.cn/index.html](http:/gepia.cancer-pku.cn/index.html)) were utilized to investigate genomic alterations, including mutations, amplifications, and deletions, in CHOL, COAD, ESCA, LIHC, PAAD, READ, and STAD cancers derived from TCGA data. These platforms were also used to construct the protein–protein interaction (PPI) network of LGALS3 and its associated genes, as well as to identify the most strongly correlated with LGALS3 expression in TCGA tumor datasets.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.