miR-424(322)^~503 impairs colon cancer progression driven by PTEN deficiency

Abstract

Colorectal cancer (CRC) is a leading cause of cancer-related morbidity and mortality worldwide, with molecular subtypes and signaling pathways playing critical roles in its progression. The miR-424(322)^~503 cluster, comprising miR-424 and miR-503, has been implicated in various malignancies, exhibiting dual roles as tumor suppressors or oncogenes depending on the context. However, its function in CRC remains poorly understood. This study investigates the role of the miR-424(322)^~503 cluster in CRC driven by PTEN deficiency using genetically modified mouse models. Our findings reveal that the loss of miR-424(322)^~503 significantly exacerbates CRC progression in PTEN-deficient mice. Double knockout (dKO) mice lacking both PTEN and miR-424(322)^~503 exhibited a higher number and larger size of colorectal lesions compared to PTEN-deficient counterparts. Histological analysis demonstrated increased severity of dysplasia and adenocarcinoma development in dKO mice. Mechanistically, while Wnt/β-catenin signaling remained unaltered, transcriptomic analyses highlighted dysregulation of MAPK and TGFβ pathways, alongside epithelial-to-mesenchymal transition (EMT)-related gene signatures. Protein-level validation confirmed hyperactivation of MAPK (ERK1/2 and p38) and TGFβ signaling, as well as elevated cyclin D1 expression in dKO colonic tissues. These results underscore the tumor-suppressive role of the miR-424(322)^~503 cluster in CRC by modulating key oncogenic pathways such as MAPK and TGFβ. Our study provides novel insights into the interplay between PTEN loss and miRNA regulation in CRC pathogenesis.

Introducton

Worldwide, colorectal cancer (CRC) is the fourth most common type of cancer and the third most deadly, considering men and women. In fact, in 2022, 1,926,425 new cases of colorectal cancer were diagnosed, and 904,019 deaths were recorded [1, 2]. Histologically, most colorectal cancers are adenocarcinomas originating in the epithelial cells of the mucosa of the colon or rectum. However, in the histological classification established in the fifth edition of the World Health Organization’s tumor classification in 2019 (WHO, 2019), there are nine more variants of this type of tumor [3, 4]. Due to the need to improve the prognostic classification provided by tumor histology, in 2015, a consensus of molecular subtypes (MSCs) was reached based on the transcriptomic analysis of many primary tumors worldwide. The results obtained made it possible to study aspects related to the tumor microenvironment and metabolic, genomic and epigenetic signatures, which resulted in the establishment of four molecular subgroups [5, 6]. The molecular classification of CRC is based on genetic and epigenetic characteristics, such as mutations, microsatellite or MSI instability, methylation of CpG islands (CpG Island Methylator Phenotype, CIMP), chromosomal instability (Chromosomal Instability, CIN), alterations in the number of copies (Copy Number Alterations, CNA) or overactivated signaling pathways [7]. The majority of colorectal cancers, between 70% and 90%, arise from precancerous polyps or adenomas that, over a long period of time, undergo a series of transformations until neoplasia develops [8, 9]. At the base of the colon crypts, there is the proliferation and differentiation of intestinal stem cells, which allows the cell turnover necessary to replace the cells that exfoliate in the intestinal lumen [10]. In the adenoma-adenocarcinoma sequence, cells undergo alterations in DNA repair mechanisms, which gives them a certain genomic instability that encourages the formation of a dysplastic epithelium [9]. The onset of adenoma formation coincides with the inactivating mutation of APC, which provides cells with proliferative self-sufficiency through sustained activation of WNT signaling. Later, mutations in KRAS, SMAD4 and TP53 provide telomere dysfunction and chromosomal instability and even invasive and metastatic capacity, resulting in the formation of invasive adenocarcinoma [8]. Apart from traditional adenomas, there are other precursor lesions that give rise to colorectal cancer. This is the case of serrated polyps, which represent the precursor lesion of 15% of the total CRCs, after having been transformed by the sawn pathway. This neoplastic pathway is characterized by presenting the activating mutation of BRAF V600E, which causes constitutive activation of MAPK signaling resulting in uncontrolled cell proliferation. After the mutation in BRAF, serrated tumors can evolve into two different pathways: the MSI pathway, which is based on alterations in DNA repair genes and results in a phenotype with a high MSI; and the way of TP53, which promotes the activation of signaling pathways such as TGF-β or EMT. It is important to note that in tumors originating from the sawn pathway, an overactivation of the WNT pathway can be found, which, unlike the traditional route, is not due to the inactivation of APC [9]. The inactivating mutations of PTEN, which are found in 10% of CRCs, also induce constitutive activation of AKT [11]. Identified as a mutation initiating the process of colorectal carcinogenesis, alterations in PTEN participate in the progression of this type of tumor. In this sense, it has been seen that the loss of expression of PTEN is highly related to genetically unstable tumors [12], as well as with tumors of greater stage and with the presence of nodular metastases [13]. Mutations in PTEN, moreover, often coexist with alterations in BRAF and, therefore, with the serrated pathway of colorectal pathogenesis that is characterized by having a high state of MSI and CIMP [14].

MicroRNAs (miRNAs) are small non-coding RNA molecules that regulate gene expression and play critical roles in cancer biology. The miR-424(322)^~503 cluster expresses miR-424(322 in mouse) and miR-503, two miRNAs belonging to the miR-16 family [15]. Among the growing number of miRNAs implicated in cancer, miR-424(322)^~503 has garnered attention due to its dual role in tumorigenesis, both as a tumor suppressor and as an oncogene, depending on the cellular context [16–18]. This miRNA-cluster is involved in the regulation of cancer-related cellular processes such as proliferation, differentiation, cellular plasticity or apoptosis [19] and has been found either up- or down-regulated in many types of cancers [16–18].

The role of miR-424(322)^~503 in CRC has not been deeply explored, and as in other types of malignancies, a dual role has been reported. As oncogenic functions, it can promote proliferation and metastasis [20, 21]. As tumor suppressive miRNA-424 has been shown to reduce tumor progression [22], angiogenesis [23] and Epithelial-to-Mesenchymal Transition (EMT) [24]. To date, there are no studies investigating the in vivo function of the miR-424(322)^~503 cluster in CRC using genetically modified mouse models. Here, we aimed to investigate the role of the miR-424(322)^~503 cluster in PTEN-loss-driven CRC.

Results

Lack of miRNA-322~503 enhances CRC tumorigenesis in Pten-deficient mice

We have recently demonstrated that miR-322^~503 is required for endometrial cancer (EC) progression initiated by Pten deficiency [25]. In this previous study, we analyzed the effects of miR-322^~503 in EC initiated by Pten deficiency by crossing a tamoxifen-inducible conditional Pten Knock-out mice (Cre: ER^(T)+/–Pten^F/F) with a miR-322^~503 knock-out mice miRNA-322 ~ 503^–/– to obtain four genotype combinations: Cre: ER^(T–/–Pten^F/FmiRNA-322 ~ 503^+/+ (Wt), Cre: ER^(T/–/–Pten^F/FmiRNA-322 ~ 503^–/– (miR KO), Cre: ER^(T)+/–Pten^F/FmiRNA-322 ~ 503^+/+ (Pten KO) mice and the Cre: ER^(T)+/–Pten^F/FmiRNA-322 ~ 503^–/– (dKO) mice (Supplementary Figure S1). In this previous study, we demonstrated that dKO mice had impaired progression of EC compared to Pten KO mice. While performing necropsy of mice to study the role of miR-322^~503 in the endometrium (Fig. 1A), we noticed that dKO animals had markedly enlarged intestines. To further investigate this finding, new independent cohorts of male and female mice were generated and treated with tamoxifen to induce Pten deletion. At the human endpoint, mice were euthanized and intestines were dissected, opened longitudinally and examined macroscopically. Neither Wt nor miR KO intestines showed detectable lesions. In contrast, numerous superficial lesions were detected in Pten KO mice, and their number was further increased in dKO mice compared to all other genotypes (Fig. 1B). Quantification of lesion number and volume confirmed a significant elevation in both parameters in dKO colons relative to Pten KO colons (Fig. 1C, D). To ensure that differences in lesion number were not attributable to intestinal length, colon lengths were measured and found comparable across genotypes (Fig. 1E). Expression of Pten, miR-322 and miR-503 was determined in colorectal tissues from Wt, miR KO, Pten KO and dKO mice to confirm the expected genotype-specific alterations (Fig. 1F, G).

Fig. 1. Double deficiency of Pten and miR-322 ~ 503 increases multiplicity, volume and complexity of colorectal polyps.

A Timeline for the in vivo necropsies performed in Cre:ER^(T)–/– Pten^F/F miR-322 ~ 503^+/+ (Wt), Cre:ER^(T)–/– Pten^F/F miR-322 ~ 503^–/– (miR KO), Cre:ER^(T)+/– Pten^F/F miR-322 ~ 503^+/+ (Pten KO) and Cre:ER^(T)+/– Pten^F/F miR-322 ~ 503^–/– (dKO) mice. B Schematic representation for miR-322, miR-503 and Pten detection in Wt, miR KO, Pten KO and dKO mice. C Relative miR-322, miR-503 and Pten expression in colorectal epithelial tissue. Statistical differences were assessed by t-test analysis. Significant differences are indicated in the figure. D Representative macroscopic images of colons from Wt, miR KO, Pten KO and dKO mice. Scale bar: 1 cm. E Violin plot showing multiplicity of colorectal lesions in Pten KO and dKO mice. Statistical significance was determined using t-test analysis. Significant differences are indicated in the figure. F Volume of colorectal lesions observed in Pten KO and dKO mice. Statistical differences were assessed by t-test analysis. Significant differences are indicated in the figure. G Colon length of Pten KO and dKO mice. Data is presented as mean ± S.E.M, according to t-test analysis. H Representative hematoxylin and eosin (H&E) staining of colorectal sections from Wt, miR KO, Pten KO and dKO mice. Scale bar: 100 μm. I Histopathological analysis of colorectal sections from Wt, miR KO, Pten KO and dKO mice showed in (F). Statistical analysis was performed with Chi-squared analysis. Significant differences are indicated in the figure. J Representative immunohistochemistry images of E-cadherin, cytokeratin-8 (CK8) and Vimentin in CRC lesions od dKO mice. Scale bar: 100 μm.

After conducting the macroscopic analysis, the colons were embedded in paraffin for subsequent histological and immunohistochemical analysis. The histological analysis was performed on colons from Wt, Pten KO, miR KO or dKO mice. For this purpose, we carried routine hematoxylin-eosin staining to observe histopathological alterations (Fig. 1H). Consistent with the lack of macroscopic intestinal lesions, all Wt and miR KO colons exhibited normal histology. On the other hand, Pten KO colons, 30% showed normal histology, while 70% presented adenomas with low-grade dysplasia. Finally, in dKO mice, 10% exhibited normal colorectal histology, while the remaining cases were distributed as follows: 50% with adenomas with low-grade dysplasia, 20% with adenomas with high-grade dysplasia, and 20% with intraepithelial adenocarcinomas (Fig. 1I). IHC staining of Cytokeratin-8 (CK8), E-cadherin and Vimentin, evidenced the epithelial origin of CRC developed in dKO mice (Fig. 1J).

Lack of Pten enhances miR-322^~503 expression in colorectal epithelium

In our previous study, we demonstrated that Pten loss caused an increase of miR-322^~503 expression in endometrial cells [25]. Having found a completely antagonistic effect of miRNA deletion in the colon, we questioned whether the expression of miR-322 and miR-503 in the colorectal epithelium of Pten KO mice was also affected. For this purpose, the miRNA fraction was isolated from total epithelial tissue, polyp lesions, or colon organoids derived from Wt and Pten KO colons, and miRNA expression was analyzed by quantitative PCR (Fig. 2A). Expression analyses revealed an upward trend in both miR-322 and miR-503 expression in Pten KO mice compared to Wt controls (Fig. 2B). It is important to mention that PTEN deficiency triggered upon tamoxifen injection in epithelial cells occurs in a mosaic pattern resulting in ablation only in a fraction of total epithelial cells, thereby diluting the possible effects of PTEN deficiency in miRNA expression (Supplementary Fig. S2A). To sort this out, miRNA was extracted from colorectal polyps dissected from Pten KO mice, in which most cells have lost Pten expression (Supplementary Fig. S2,B). The results demonstrated a significant increase in the expression of both miR-322 and miR-503 in PTEN-deficient colorectal polyps (Fig. 2C). This increase in the miRNA expression was further amplified in Pten KO colon organoids compared to Wt organoids, in which Pten deletion was confirmed by quantitative PCR (Fig. 2D, E and Supplementary Fig. S2,C).

Fig. 2. Pten deficiency increases miRNA-424(322)/503 cluster expression in colorectal epithelial cells.

A Timeline for the analysis of miR-322 and miR-503 expression in colorectal epithelium from Cre:ER^(T)–/– Pten^F/F (Wt) and Cre:ER^(T)+/– Pten^F/F (Pten KO) mice. B Relative miR-322 and miR-503 expression in total colorectal epithelium or C polyps from Wt and Pten KO mice. Data is presented as mean ± S.E.M, according to t-test analysis. Significant differences are indicated in the figure. D Representative images of colon organoids from Wt and Pten KO mice. Scale bar: 500 μm. E Relative expression of miR-322 and miR-503 in Wt and Pten KO mouse colon organoids. Data is presented as mean ± S.E.M, according to t-test analysis. Significant differences are indicated in the figure. F Relative expression of miR-322 and miR-503 in Wt and Pten KO mouse colon organoids treated with 10 ng/μl TGF-β for 16 h. Data is presented as mean ± S.E.M, according to t-test analysis. Significant differences are indicated in the figure.

In our previous study, we also demonstrated that treatment of organoids with TGF-β caused an increase of miR-322^~503 expression in endometrial cells [25]. Therefore, we questioned whether this regulation was also important in colon epithelial cells. To address this question, we treated Wt and Pten KO colon organoid cultures with TGF-β, and we measured expression of miRNA-322 and miRNA-503. As previously demonstrated in the endometrium, TGF-β treatment, also led to a pronounced increase in the expression of both miRNAs in colon organoids (Fig. 2F). These results underscore the role of PI3K/Akt and TGF-β signaling in the regulation of miR-424(322)^~503 cluster expression.

CRC from double Pten and miR-322^~503-deficient mice do not display alterations in β-catenin signaling

It is well documented that miR-322^~503 plays a role in regulating the Wnt/β-catenin signaling pathway in mammary tissues through its binding to the messenger RNA of LRP6, one of the receptors of this pathway [26]. This, combined with the critical role of this signaling pathway in colorectal cancer [27], leads to the hypothesis that dKO mice may suffer an hyperactivation of the Wnt/β-catenin pathway in colorectal tissue, resulting in the development of malignant lesions in this organ. To address this question, we analyzed Lrp6 and Ctnnb1 expressions, together with their immunohistochemical profiles, in colorectal tissues from Wt, miR KO, Pten KO and dKO animals (Fig. 3A). Expression analyses and immunostaining for the LRP6 receptor and β-catenin revealed no detectable differences between genotypes. Particularly, no increase in either LRP6 or β-catenin was observed in dKO colons compared to Pten KO colons (Fig. 3B, C). Based on these results, the loss of miR-322 ~ 503 expression in PTEN-deficient colons does not affect the Wnt/β-catenin pathway, ruling it out as the cause of the increased colorectal lesions.

Fig. 3. Carcinogenesis induced by lack of Pten and miRNA-322 ~ 503 is not mediated by Wnt/β-catenin signaling.

A Schematic representation for protein analyses in Cre:ER^(T)–/– Pten^F/F miR-322 ~ 503^+/+ (Wt), Cre:ER^(T)–/– Pten^F/F miR-322 ~ 503^–/– (miR KO), Cre:ER^(T)+/– Pten^F/F miR-322 ~ 503^+/+ (Pten KO) and Cre:ER^(T)+/– Pten^F/F miR-322 ~ 503^–/– (dKO) mice. B Relative expression of Ctnnb1 and Lrp6 in Wt, miR KO, Pten KO and dKO colorectal tissue. Statistical differences were determined with ANOVA followed by Tukey’s multiple comparisons test. Significant differences are indicated in the figure. C Representative immunohistochemistry images of β-catenin and Lrp6 in colon sections from Wt, miR KO, Pten KO and dKO mice. Scale bar: 50 μm.

Double Pten and miR-322^~503 deficiency regulates the expression of genes associated with MAPK signaling, TGF-β signaling and epithelial-to-mesenchymal transition

Since β-catenin did not seem to be altered in dKO mice, we pursued alternative molecular alterations that could explain colorectal carcinogenesis induced by the dual loss of miR-322^~503 and PTEN. To have a global view of this dual loss effects on gene expression, we performed a comprehensive mRNA of colon epithelium from Wt, Pten KO, miRNA KO or dKO mice. To investigate the transcriptional impact of miR-322^~503 deletion, we performed a differential gene expression analysis (DEG, log₂ FC > abs(1.5) plus adj. FDR p < 0.05) to compare gene expression profiles across the following groups: miR KO vs. Wt, Pten KO vs. Wt, dKO vs. Wt, and Pten KO vs. dKO (Fig. 4A, B and Supplementary Table 1). As expected by the dramatic impact of double Pten and miRNA ablation in CRC development, the comparison between samples dKO and Wt exhibited the highest number of genes with significantly altered expression. However, the comparison between Pten KO and Wt samples also exhibited many genes with significantly altered expression, indicating the profound transcriptional changes associated with Pten loss. Then, to find differentially regulated biological processes or molecular functions (i.e., pathway-level insights) underlying CRC development, we conducted an exploratory analysis by GSEA⁴⁵ using predefined gene set annotations.

Fig. 4. Transcriptome analysis of colon epithelium deficient in Pten and/or miR-322^~503.

A Volcano plots of differentially expressed genes (DEGs) comparing Cre:ER^(T)–/– Pten^F/F miR-322 ~ 503^+/+ (Wt), Cre:ER^(T)–/– Pten^F/F miR-322 ~ 503^–/– (miR KO), Cre:ER^(T)+/– Pten^F/F miR-322 ~ 503^+/+ (Pten KO) and Cre:ER^(T)+/– Pten^F/F miR-322 ~ 503^–/– (dKO) replicates, as indicated. Green dots represent significantly upregulated genes, while red dots indicate significantly downregulated genes. B Heatmap of hierarchical clustering analysis showing DEGs across Wt, Pten KO and dKO genotypes. C Dot plot representing gene set enrichment analysis (GSEA) of enriched gene signatures from the molecular signatures database (MSigDB) gene ontology (GO) biological processes (BP), molecular function (MF), cellular components (CC) and hallmarks comparing Pten KO and dKO genotypes. The plot highlights transcriptomic signatures deregulated in dKO colorectal tissue. Each red node represents a distinct positive enrichment, while blue nodes represent negative enrichment annotations; node size corresponds to the number of genes included within the annotation. D Enriched gene annotation from GSEA for “Protein Serin Threonin Kinase Activity”, “ERK1 and ERK2 Cascade”, “p38MAPK Cascade”, “Response to Transforming Growth Factor β” and “Epithelial Mesenchymal Transition”.

Given that the CRC development was dramatically increased in dKO over Pten KO colonic epithelium, we focused our analysis on comparing these two genotypes. Our GSEA primarily retrieved gene signatures associated with tissue remodeling processes such as Cell-cell Adhesion, Cell Adhesion Molecule Binding, Response to Wounding/Wound healing or Epithelial-to-Mesenchymal Transition (EMT) (Fig. 4C, D). Regarding signaling pathway modifications, we found significative changes in the signatures associated with TGF-β signaling and the MAPK signaling pathways ERK1/2 and p38 (Fig. 4C, D). These pathways are also known to play critical roles in cellular growth and survival and are commonly dysregulated in CRC. These findings suggest that the loss of miR-322^~503 and Pten triggers a transcriptional program that actively promotes cell proliferation and EMT, likely through its impact on TGF-β and MAPK signaling.

Colorectal tissue deficient in both Pten and miR-322^~503 exhibits hyperactivation and TGF-β signaling pathways and enhanced Cyclin D1 expression

To analyze whether transcriptomic dysregulation of ERK1/2 and p38 MAPK signaling were translated to alterations in the activation of such pathways, we carried-out Western Blot analysis of protein lysates and immunohistochemistry staining from colorectal epithelium from Wt, miR KO, Pten KO or dKO mice (Fig. 5A). The effect of Pten deletion on the PI3K/AKT pathway was investigated by analyzing the phosphorylated forms of AKT, p70S6K, and S6 as a read-out of pathway activation. As we mentioned before (Supplementary Fig. S2,A), it is important to highlight that Pten deletion in epithelial cells occurs in a mosaic pattern, which explains the partial expression of PTEN observed in Pten KO and dKO conditions. Despite such mosaicism Pten deletion, a marked increase in the phosphorylated forms of AKT, p70S6K, and S6 was observed in all Pten KO and dKO conditions. Interestingly, dKO did not show further significant increase of AKT, p70S6K, and S6 phosphorylation over Pten KO conditions, suggesting that lack of miR-322^~503 does not affect PI3K/AKT activation directly (Fig. 5B, C). This result is further supported by the miR KO conditions, in which PI3K/Akt signaling pathway is not significantly increased over Wt controls. Regarding MAPK signaling activation, Western blot analysis revealed an elevated phosphorylation levels of JNK, p38, MKK4, and ERK1/2, in dKO conditions compared to the rest on genotypes, suggesting an activation of the three main MAPK signaling pathways (Fig. 5B, C). Among all three MAPKs, the ERK1/2 signaling pathway plays a pivotal role in development of CRC. Therefore, we further analyzed the increase in ERK1/2 phosphorylation by IHC staining of Wt, miR KO, Pten KO or dKO colon epithelium. Notably, a marked increase of p-ERK1/2 staining was observed in lesions from dKO mice. Moreover, analysis of Cyclin D1 expression in consecutive tissue sections revealed a pronounced increase in lesions displaying increased ERK phosphorylation (Fig. 5D, E).

Fig. 5. Double loss of Pten and miR-322^~503 enhance MAPK and TGF-β signaling.

A Schematic representation for protein analyses in Cre:ER^(T)–/– Pten^F/F miR-322 ~ 503^+/+ (Wt), Cre:ER^(T)–/– Pten^F/F miR-322 ~ 503^–/– (miR KO), Cre:ER^(T)+/– Pten^F/F miR-322 ~ 503^+/+ (Pten KO) and Cre:ER^(T)+/– Pten^F/F miR-322 ~ 503^–/– (dKO) mice. B Western blot images of phosphorylated signaling proteins: p-p70(S6K)^Thr389, p-Akt^Ser473, p-S6^Ser235/236, p-Sapk/Jnk^{Thr183/Tyr185}, p-Sek1/Mkk4^Thr261, p-Erk1/2^{Thr202/Tyr204} and p-p38^{Thr180/Tyr182}. The analysis was performed on Wt, miR KO, Pten KO and dKO epithelial colorectal tissue. Membranes were re-probed with Gapdh as loading control. C Densitometric analysis of Western blot bands in (B). Phosphorylated protein levels were normalized to Gapdh as loading control. Data represent mean ± S.E.M from three independent mice for each of the four genotypes. Statistical differences were performed with ANOVA followed by Tukey’s multiple comparisons test. Significant differences are indicated in the figure. D Representative IHC images of p-Erk1/2^{Thr202/Tyr204} and Cyclin D1 in colorectal sections from Wt, miR KO, Pten KO and dKO transgenic mice. Scale bar: 100 μm. E Quantitative analysis of IHC staining across genotypes using QuPath to quantify staining intensity (H-score, p-Erk1/2) and positive nuclei (Cyclin D1). Data represent mean ± S.E.M of >5 biological replicates per genotype. Statistical differences were performed with ANOVA followed by Tukey’s multiple comparisons test. Significant differences are indicated in the figure. F Western blot images showing the expression of SMAD3, SMAD4 and SMAD7. The analysis was performed on Wt, miR KO, Pten KO and dKO epithelial colorectal tissue. Membranes were re-probed with Gapdh as loading control. G Densitometric analysis of Western blot bands in (F). Phosphorylated protein levels were normalized to Gapdh as a loading control. Data represent mean ±S.E.M from three independent mice for each of the four genotypes. Statistical differences were performed with ANOVA followed by Tukey’s multiple comparisons test. Significant differences are indicated in the figure. H Representative immunohistochemistry images of Smad2/3 in Wt, miR KO, Pten KO and dKO colorectal sections. Scale bar: 100 μm. I Quantitative analysis of IHC staining across genotypes using QuPath to quantify staining intensity (H-score) of Smad2/3. Data represent mean ± S.E.M of >5 biological replicates per genotype. Statistical differences were performed with ANOVA followed by Tukey’s multiple comparisons test. Significant differences are indicated in the figure.

Finally, we investigated alterations in signaling pathways activated in response to TGF-β. The SMAD family of transcription factors is one of the main transducers of cellular responses after TGF-β receptor engagement [28]. Among them, the SMAD3 and SMAD4 are the main transcription factors activated by TGF-β signaling, while SMAD7 is an inhibitory SMAD. Western blot analysis of Wt, miR KO, Pten KO or dKO derived lysates revealed a global increase of SMAD3 and SMAD4 expression and a decrease in SMAD7 in dKO mice (Fig. 5F, G). Moreover, immunohistochemistry showed an increased intensity staining of SMAD2/3 in dKO tumors (Fig. 5H, I). This result suggests that TGF-β/SMAD signaling is enhanced by the lack of Pten and the miR-322^~503.

IGF1 is a direct conserved target of the miR-424(322)^~503 cluster in colorectal cancer

To identify conserved targets of the miR-322^~503 cluster potentially contributing to the intestinal phenotype, we intersected transcripts significantly upregulated in dKO compared with Wt cells (log2FC > 3) with conserved predicted miR-322^~503 targets. Among the 318 upregulated genes, 24 overlapped with conserved miR-322^~503 target transcripts, including Igf1, suggesting that loss of miR-322^~503 leads to derepression of canonical target mRNAs (Fig. 6A). We then examined IGF1 expression in human tumor datasets and found that IGF1 mRNA levels were significantly higher in tumors with low miR-424 or miR-503 expression, indicating an inverse correlation consistent with miR-424(322)^~503-mediated repression (Fig. 6B). To gain structural insight into this regulatory interaction, we analyzed the IGF1 mRNA sequence and identified conserved miR-424 (7merA1) and miR-503 (8mer) binding sites within its 3’UTR, showing canonical seed-pairing patterns (Fig. 6C).

Fig. 6. IGF1 is a conserved direct target of the miR-424^~503 cluster in colon cancer.

A Venn diagram showing the overlap between mRNAs significantly upregulated in double knockout (dKO) versus wild-type (Wt) cells (log₂FC > 3) and conserved miR-424^~503 predicted targets. Among the 318 upregulated transcripts, 24 overlapped with conserved miR-424^~503 target genes, including IGF1, highlighting potential derepression of canonical miRNA targets following loss of miR-424^~503. B Boxplots showing IGF1 mRNA abundance (log₂RSEM counts +1]) in tumors stratified by the top (20% high) and bottom (20% low) expression levels of miR-424 (left) or miR-503 (right) across human colorectal cancer samples. IGF1 expression was significantly higher in tumors with low miR-424 (p = 6.92 × 10^–4, Wilcoxon test) and low miR-503 (p = 4.55 × 10^–2, Wilcoxon test), indicating an inverse regulatory relationship consistent with miR-424^~503 mediated suppression of IGF1. C Schematic representation of the human IGF1 mRNA showing the coding region and 3′UTR with the predicted miR-424 and miR-503 binding sites (BS). The seed pairing alignments are shown for each interaction: a 7mer-A1 site for miR-424 and an 8mer site for miR-503, located at positions 6053-6060 of the IGF1 3′UTR. Both sites display canonical base-pairing patterns consistent with direct miRNA-mRNA targeting. D Luciferase reporter assay showing the effect of miR-424^~503 overexpression on constructs containing the 3′UTRs of IGF1 and IGF1R downstream of a CMV-driven luciferase reporter. IGF1R served as positive control. Co-transfection with the miR-424^~503 expression vector significantly reduced normalized luciferase activity for both IGF1 (p = 0.004) and IGF1R (p = 0.001), confirming that these transcripts are direct miR^~424/503 targets. E Relative expression of Igf1 in Wt and dKO mouse colon epithelium. Data is presented as mean ± S.E.M, according to t-test analysis (p = 0.03).

To confirm that IGF1 is a conserved target of miR-424(322)^~503 cluster, we assessed direct targeting using a luciferase reporter assay. IGF1R, a well-established miR-424(322)^~503 target [29], was included as a positive control. Co-transfection with the miR-424(322)^~503 expression construct significantly reduced luciferase activity from reporters containing the IGF1 or IGF1R 3’UTRs, confirming direct targeting of IGF1 by the miR-322^~503 cluster (Fig. 6D). Finally, we also examined Igf1 expression directly in colorectal tissues from Wt and dKO mice. Quantitative analysis revealed that Igf1 levels were significantly increased in dKO tissues compared to Wt controls (Fig. 6E). This supports the notion that loss of miR322^~503 leads to derepression and upregulation of Igf1 in vivo, further corroborating the regulatory relationship established in our molecular and bioinformatic analyses.

MIR503HG-miR-424-miR-503 axis is dysregulated in human colorectal cancer

Consistent with our in vivo findings in the Cre: ER^(T) Pten^F/F miR-322^~503 mouse model, we investigated the clinical relevance of the miR-424^~503 cluster in human colorectal cancer. First, we analyzed human colorectal samples, which revealed a significant positive correlation between miR-424 and miR-503 expression levels, indicating co-expression of these miRNAs (Fig. 7A). Moreover, the distribution of their expression demonstrated that miR-424 exhibits a wider dynamic range and generally higher abundance compared to miR-503. In addition, miR-424 expression was found to be significantly higher than that of miR-503 (Fig. 7B, C), supporting the notion of distinct absolute abundance levels despite their coordinated expression.

Fig. 7. Genomic organization and coordinated dysregulation of the MIR503HG–miR-424–miR-503 axis in human colon cancer.

A Scatter plot depicting normalized expression levels (log₂[RPM + 1]) of miR-424 and miR-503 across human colorectal tumor samples. A significant positive correlation was observed between miR-424 and miR-503 abundance (R = 0.526, p = 2.82 × 10^–19). B Histograms showing the distribution of miR-424 (left) and miR-503 (right) expression levels (counts per million, CPM) across human colorectal cancer samples. miR-424 displayed a wider dynamic range (average=128.03 CPM; median=91.67; standard deviation=116.05; range=19-838) compared with miR-503 (average=8.4 CPM; median=5.23; standard deviation=10.93; range=1-120), suggesting higher and more variable expression of miR-424 across patient tumors. C Boxplot illustrating normalized expression levels (log₂[RPM + 1]) of miR-424 and miR-503 across 251 colorectal cancer samples. miR-424 expression was significantly higher than that of miR-503 (Wilcoxon test, p < 2.2 × 10^–16), indicating that although these microRNAs are co-expressed, their absolute abundance markedly differs in tumor tissues. D Left, schematic representation of the X chromosome region q26.3, illustrating the genomic arrangement of the long non-coding RNA MIR503HG (H19X) and its embedded microRNAs miR-424 and miR-503. The figure highlights the co-localization of these transcripts within a shared genomic cluster, supporting their potential transcriptional co-regulation and coordinated expression in colorectal tissue. Right, boxplots showing the normalized expression levels (log₂[TPM + 1]) of MIR503HG across normal colorectal tissues and tumors of increasing pathological stage (I-IV). MiR503HG exhibited significant stage-dependent upregulation (Wilcoxon test p < 2.22 × 10^–16). E Boxplots showing normalized expression levels (log₂[RPM + 1]) of miR-424-5p (left) and miR-503-5p (right) across normal colon tissues and tumors of increasing stage (I-IV). MiR-424 displayed a marked increase in expression in tumor samples relative to normal tissue (p < 5 × 10^–6, Wilcoxon test).

Next, we examined their genomic localization, revealing that both miRNAs are situated within the long non-coding RNA MIR503HG cluster on chromosome Xq26.3, which implies potential transcriptional co-regulation (Fig. 7D). Notably, MIR503HG itself was significantly upregulated in colorectal tumors in a stage-dependent manner (Fig. 7E). Similarly, miR-424 and miR-503 showed a marked increased expression in tumor tissues relative to normal tissues, with levels rising with higher tumor stages (Fig. 7F). Collectively, these findings demonstrate that MIR503HG-miR-424-miR-503 axis is coordinately dysregulated in colon cancer, with miR-424 particularly exhibiting higher and more variable expression, which could influence tumor progression.

Discussion

MicroRNAs serve as key regulators of post-transcriptional gene expression and have been increasingly recognized as pivotal players in oncogenic processes. While numerous miRNAs have been associated with CRC pathogenesis, the functional characterization of individual miRNAs—particularly their in vivo roles— remains an underexplored area of research. This study addresses a critical knowledge gap by investigating the miR-424(322)^~503 cluster through a genetically engineered mouse model, revealing its essential role in counteracting PTEN deficiency-driven CRC progression. Our findings demonstrate that genetic ablation of miR-424(322)^~503 markedly accelerates colorectal tumorigenesis in Pten-deficient mice, establishing this miRNA cluster as a crucial tumor suppressor in the Pten-loss context. The tumor-suppressive mechanism appears to involve coordinated regulation of two fundamental signaling axes: the ERK-MAPK signaling, a pathway frequently hyperactivated in human CRCs and the TGF-β/Smad signaling, a critical pathway maintaining epithelial differentiation and apoptotic responses. Notably, the combined loss of Pten and miR-424(322)^~503 creates a permissive environment for malignant transformation by simultaneously disrupting these interconnected networks. This dual pathway dysregulation likely explains the observed acceleration of adenoma-to-adenocarcinoma progression in our model, providing mechanistic insights into how miRNA-mediated regulation protects intestinal epithelium integrity under oncogenic stress conditions.

One of the most intriguing questions about miR-424(322)^~503 is its paradoxical role as oncogene or tumor suppressor gene depending on the cellular context [16–18]. However, tumor suppressive and oncogenic functions have been demonstrated in independent models, leaving the possibility that different experimental scenarios may affect the function of miR-424(322)^~503. We have recently demonstrated that miR-424(322)^~503 acts as an oncogene in EC initiated by Pten deletion by regulating cell proliferation and apoptosis. Here, using the same mouse model, we have found that miR-424(322)^~503 functions as tumor suppressor in CRC initiated by Pten deficiency. The induction of the miR-424(322) ~ 503 cluster following PTEN loss and TGF-β stimulation in the colon likely represents a compensatory tumor-suppressive response, rather than a driver event. In this context, upregulation of the cluster appears to function as a protective mechanism that helps restrain excessive MAPK and TGF-β pathway activation triggered by PTEN deficiency. Accordingly, the loss of miR-322/503 removes this compensatory brake, leading to unchecked signaling and a marked exacerbation of tumorigenesis in the PTEN-deficient colonic epithelium. These completely opposite in vivo effects in the endometrium and the colonic epithelium highlight the duality of miR-424(322)^~503 depending on the tissue, even when carcinogenesis is initiated by the same driver. Furthermore, the function of this miRNA specifically in CRC is still controversial, and both oncogenic [20, 21] and tumor suppressive functions have been proposed [22–24]. Another result that deserves discussion is the effect of Pten ablation on miR-424(322)^~503 expression in colorectal epithelial cells. Our previous results demonstrated that loss of Pten or TGF-β caused an up-regulation of miR-424(322)^~503 expression, which acts as an oncogene by promoting EC progression [25]. Similarly, we found that Pten or TGF-β increases the expression of miR-424(322)^~503, which ultimately has the contrary function to that found in the endometrium. This result strengthens the importance of PI3K/Akt and TGF-β signaling pathways as critical regulators of miR-424(322)^~503 expression, independently of its function in the target tissue.

At the heart of CRC pathogenesis remain disrupted signaling pathways that promote tumor formation, support cancer cell growth, and facilitate metastatic spread. Among them, the Wnt/β-catenin, the RAS/RAF/ERK, the PI3K/AKT, and the TGF-β are frequently altered by a combination of genetic mutations [30]. It has been recently demonstrated that miR-424(322)^~503 cluster can modulate β-catenin signnalling by targeting LRP6 receptor in mammary epithelium [26]. Based these findings and the crucial role of β-catenin signaling in CRC [31], it was easy to speculate that CRC caused by double Pten and miR-424(322)^~50 ablation was caused by dysregulation of classic β-catenin signaling. However, no changes in β-catenin expression or cellular localization, nor in LRP6 expression, have been found in double Pten and miR-424(322)^~503 KO mice, suggesting this pathway is not participating in our model. Instead, our findings demonstrate that miR-424(322)^~503 knockout enhances MAPK signaling in Pten-deficient colorectal cells, resulting in an epithelium with exacerbated activation of both PI3K/Akt and MAPK pathways. Deregulation of the Ras/Raf/MEK/ERK MAPK signaling pathway is a critical factor that drives the progression of CRC [32]. This pathway acts as a critical signaling hub that governs cell proliferation and cell cycle progression. Among human cancers, somatic RAS mutations occur in roughly 30% of cases, triggering constitutive activation of the downstream kinase cascade. The process begins with mutant RAS activating RAF, which phosphorylates MEK, ultimately leading to sustained MAPK/ERK signaling. This stepwise amplification drives uncontrolled cellular growth and survival. The cooperation between the PI3K/Akt and MAPK signaling pathways is well-documented in CRC, as both pathways play critical roles in tumor initiation, progression, metastasis and resistance to therapy [30]. It is worth mentioning that besides RAS/RAF/ERK MAPK signaling, we have also found increased phosphorylation of p38 [33] and JNK [34, 35], two other MAPK that can participate CRC development and progression. Our identification of IGF1 as a direct target of the miR-424(322) ~ 503 cluster provides a mechanistic explanation for the MAPK/ERK hyperactivation seen in dKO colons. IGF1 is a strong upstream activator of the RAS/ERK cascade, and its derepression in the absence of miR-322/503 aligns with the increased phospho-ERK and phospho-p38 levels observed in vivo. These findings support a model in which the miRNA cluster normally restrains IGF1-driven MAPK signaling in the PTEN-deficient epithelium. When both PTEN and miR-322/503 are lost, this dual inhibitory control is removed, creating a signaling environment that favors epithelial proliferation and tumor progression. These findings collectively underscore the critical role of PI3K/Akt and MAPK pathway cooperation in CRC pathogenesis and highlight their potential as therapeutic targets.

Finally, we have identified signatures associated with TGF-β signaling EMT, two closely interconnected processes involved in carcinogenesis [36]. Regarding TGF-β signaling, the increased expression of SMAD3 and SMAD4, along with the decreased expression of SMAD7 and the enhanced nuclear accumulation of SMAD2/3 in colorectal cancer (CRC) derived from miRNA and Pten double knockout mice, strongly suggests an activation of the pathway. The upregulation of TGF-β signaling in tumors is closely linked to EMT, as these two processes are highly interconnected and play critical roles in cancer progression. Therefore, it is reasonable to speculate that the increased EMT signatures are associated with the activation of TGF-β/SMAD signaling.

Consistent with our observations in the Pten-deficient mouse colon, several independent studies have demonstrated that the miR-424(322) ~ 503 cluster is downregulated in human colorectal cancer, supporting a conserved tumor-suppressive role in the intestinal epithelium. Li et al. reported that miR-424 is significantly reduced in both primary CRC tissues and multiple human CRC cell lines compared with adjacent normal mucosa, and that loss of miR-424 enhances SMAD7 expression to promote proliferation, migration, and invasion of CRC cells [20]. More recently, Ghonbalani et al. showed that miR-424 is frequently silenced by promoter hypermethylation in human CRC, leading to marked downregulation of miR-424 and concomitant upregulation of oncogenic targets such as VEGF, thereby promoting tumor angiogenesis and progression [23]. Together, these studies indicate that reduced miR-424 expression is a recurrent feature of human colorectal tumors and often arises through epigenetic repression. These findings align with our results showing that genetic deletion of miR-322/503 exacerbates PTEN-loss–driven tumorigenesis and support a model in which the miR-424(322) ~ 503 cluster normally restrains key oncogenic pathways, including IGF1–MAPK and TGF-β signaling, in the colonic epithelium. Thus, the decreased expression of this miRNA cluster observed in human CRC further underscores the clinical relevance of the tumor-suppressive functions uncovered in our mouse model.

In conclusion, this study emphasizes the complex and context-dependent role of the miR-424(322) ~ 503 cluster in CRC, highlighting its critical involvement in regulating key signaling pathways such as MAPK and TGF-β/SMAD. By demonstrating how the loss of this miRNA cluster accelerates tumorigenesis in a Pten-deficient mouse model, we provide important mechanistic insights into its tumor-suppressive function in CRC. Furthermore, our findings underscore the intricate interplay between PI3K/Akt, MAPK, and TGF-β signaling pathways in driving cancer progression, revealing potential therapeutic targets for the treatment of CRC. The paradoxical dual role of miR-424(322) ~ 503, acting as both an oncogene and tumor suppressor depending on the tissue context, opens new avenues for future research aimed at understanding its precise molecular mechanisms.

Material and methods

3’UTR cloning and luciferase reporter assays

The 3′ untranslated regions (3′UTRs) of IGF1 and IGF1R were amplified by PCR from human genomic DNA using the following primers: IGF1R-F(MluI)/-R(PmeI) aaacgcgtCGACGAGAGACAGCCTTA/aagtttaaacTGACATCAGTAACTCGGACA, and IGF1-F(MluI)/-R(HindIII) aaacgcgtCTTTCAAGCCACCCATTG/aaaagcttGGGGCCTTTATGTAAACTG. PCR amplicons were visualized on agarose gels, excised, and purified (28706, Qiagen). In parallel, 3 mg of the pMIR-REPORT vector (AM5795M, Life Technologies) were digested with the appropriate NEB restriction enzymes (PmeI R0560, MluI R3198 and HindIII R3104). PCR products were digested with the same enzymes, purified (Qiagen, 28104) and subsequently ligated into the linearized vector using a 3:1 insert-to-vector ratio (T4 DNA Ligase, M0202, NEB). Ligation products were transformed into E. coli DH5a chemically competent cells (18265017, Invitrogen). Up to ten bacterial colonies per construct were screened, and plasmid DNA was isolated (27106, Qiagen) and sequence-verified by Sanger sequencing to confirm correct 3′UTR insertion. To assess the regulatory activity of miR-424 ~ 503 on these target 3′UTRs, Phoenix cells were seeded at 70% confluence in 96-well black plates (3603/153603, Corning/Cultek). After 24 h, cells were co-transfected with 50 hg of each validated pMIR-REPORT-3′UTR construct and a Renilla luciferase control plasmid, together with 100 hg of pLEMIR-miR-424 ~ 503 expression vector [29], using the jPEI transfection reagent. Twenty-four hours post-transfection, luciferase activity was measured using the Dual-Glo Luciferase Assay System (E2940, Promega), and relative luciferase units (RLUs) were recorded on a GloMax-Multi+ Microplate Multimode Reader (SA3030, Promega).

Experimental mouse models

Mice were housed in a barrier facility, and pathogen-free procedures were used in all mouse rooms. Animals were under 12 h of light/dark cycles at 22 °C, and they had ad libitum access to water and food. All procedures were performed according to the guidelines of the Ethical Committee of Universitat de Lleida and the National Institute of Health Guide for the Care and Use of Laboratory Animals. Conditional Pten knockout (C; 129S4-Ptentm1Hwu/J or Pten^F/F) and CAG-Cre: ER^(T) mice were obtained from the Jackson Laboratory (Bar Harbor, ME, USA). miR-322 ~ 503^–/– (FVB/NJ) mice were a gift from Prof. Jose Silva. Cre:ER^(T) Pten^F/F miR-322 ~ 503^–/– mice were generated by crossing Cre:ER^(T)+/– Pten^F/F and miR-322 ~ 503^–/– mice. Three weeks after birth, mice were weaned and genotyped as previously described. Genotyping primers and PCR conditions are available in Supplementary Materials (Table S1).

Tamoxifen administration

Tamoxifen (T5648, Sigma-Aldrich) was prepared and administered as previously described. Briefly, tamoxifen powder was dissolved in 100% ethanol at a concentration of 100 mg/ml. Then, tamoxifen was emulsified in corn oil (C8267, Sigma-Aldrich) at a final concentration of 10 mg/ml. Mice between 5 and 8 weeks were injected with a single 0.5 mg of tamoxifen intraperitoneally.

Isolation of epithelial colon cells and organoid culture

Mice were euthanized by cervical dislocation and the colorectal portions compressed between cecum and rectum were dissected. Colon portions were longitudinally opened and gently washed in ice-cold Phosphate Buffered Saline (PBS). Colons were incubated in digestion medium (50 mg/ml collagenase type II (17101-015, Gibco) and 10% FBS in DMEM for 2 h at 37 °C. To isolate colon crypts, digested colons were passed through a 70μm sterile strainer and washed in DMEM/F12 (11039-021, Gibco). When Pten ablation was required, colon crypts were incubated for 30 min at 37 °C with TAT-Cre peptide [37] and gently washed in DMEM/F12 before seeding. Colon crypts were resuspended in IntestiCult™ Organoid Growth Medium (Mouse) (06005, STEMCELL) containing 50% Matrigel™ (354234, Corning) to obtain 1500 crypts/ml and plated as domes in a multiwell plate. Domes containing 50% Matrigel™ were polymerized for 10 min at 37 °C and then, IntestiCult™ Organoid Growth Medium (Mouse) was added to cover the dome. Every 2–3 days medium was replaced until organoids were completely formed. Organoid treatments were performed two days after medium replacement. TGF-β (PHG9214, Gibco) treatments were performed at 10 ng/μl for 16 h.

miRNA extraction and real-time qPCR

miRNA extraction was performed in total epithelial colon tissue, isolated polyps, colon organoids, and HCT116 transfected cells using mirVana miRNA isolation kit (AM1561, Ambion) following manufacturer’s instructions. miRNA extracts were quantified with NanoPhotometer (N60 UV/Vis Spectrophotometer, IMPLEN) and stored frozen at –80 °C. Reverse transcription was performed using 50 ng of RNA using High-Capacity cDNA Reverse Transcription (4368815, Applied Biosystems), according to manufacturer’s protocol. Retrotrancription was performed at 16 °C for 30 min, followed by 60 cycles of 20 °C for 3 s, 42 °C for 3 s and 50°C for 2 s, reaction was inactivated at 85 °C for 5 min. Primers for retrotranscription are detailed in Supplementary Materials (Table S2). Quantitative real-time PCR detection of miRNAs was performed with the CFX96 (BioRad) using PowerUp SYBR Green MasterMix (A25742, Applied Biosystems). Real-time PCR was performed at 95 °C for 3 min, followed by 40 cycles of 95 °C for 10 s and 60 °C for 30 s, and finished with a descending gradient of temperature. Sequences of primers used in real-time PCR are shown in Supplementary Materials (Table S3). Relative expressions were determined from cycle threshold (Ct) values, which were normalized to sno-202 as a housekeeping miRNA. Experiments were performed at least three times, and every group was performed in triplicate.

Total mRNA extraction and real-time qPCR

Total RNA from colon epithelial tissue and HCT116 transfected cells was extracted using SurePrep TrueTotal RNA Purification kit (BP2800-50, Fisher BioReagents) following manufacturer’s instructions. Total RNA extracts were quantified with NanoPhotometer and stored frozen at –80 °C. Quantitative real-time PCR was performed with 50 ng of total RNA using the one-step protocol qPCRBIO Probe 1-step Go (PB25.44-01, PCR Biosystems) according to manufacturer’s protocol. Primers used for gene expression assays were commercially obtained from Applied Biosystems and are listed in Supplementary Materials (Table S4). Relative expressions were assessed by cycle threshold (Ct) values, which were normalized to Gapdh expression as an internal control. Experiments were performed at least three times, and every group was performed in triplicate.

Whole-genome mRNA sequencing

2 × 150 bp mRNA sequencing was performed at GeneWiz (Azenta Life Sciences) on NovaSeq equipment at a depth of 20 M reads per sample. Quality control of all FASTQ files was performed using the FASTQC tool (v0.11.9) and MultiQC tool [38] implemented in R. For mRNAseq quantification the mapping-based mode of Salmon (v1.5.2) [39] was used to perform transcript-level quantification with default parameters, using the tximport library (v1.28.0) [40]. Low-expression genes, defined as those with fewer than 10 reads across all samples, were filtered out prior to analysis. Data normalization and differential expression analysis between miR KO, Pten KO, dKO and Wt conditions were performed using the DESeq2 R package (v1.30.1) [41]. To conduct Gene Set Enrichment Analysis (GSEA) [42], genes from the dKO vs Wt comparison were ranked by fold change and analyzed using the msigdbr R function. Selected pathways were visualized in a dot plot, displaying FDR values and gene set sizes, created using the ggplot2 package in R.

Western blot

Western blot analysis was performed as previously described with minor variations. To isolate the total colorectal epithelium, colons were dissected, opened longitudinally and washed twice with PBS. Epithelial cells were isolated with a scalpel and lysed with 2% SDS, 125 mM Tris-HCl, pH 6.8 (lysis buffer). Relative protein concentrations were determined with a colorimetric protein assay kit (5000112, BioRad). Equal amounts of protein were loaded onto an acrylamide gel and transferred to PDVF membranes (IPVH00010, Millipore). To avoid nonspecific antibody binding, membranes were blocked for 1 h with 5% non-fat milk in TBS-T (20 mM Tris-HCl, pH 7.4, 150 mM NaCl, 0.1% Tween-20). Membranes were incubated with primary antibodies for 16 h at 4 °C. Then, membranes were incubated for 1 h with secondary antibodies at room temperature. Finally, a signal was detected with Immobilon Forte Western HRP Substrate (WBLUF0100, Millipore). Primary and secondary antibodies used for Western blot and their concentrations are detailed in Supplementary Materials (Table S5). Intensity band quantification was performed in an image analyzer (ImageJ, version 1.46r; NIH, Bethesda, MD, USA). Signal values were normalized to the corresponding loading control and expressed relative to control samples.

Tissue processing and immunohistochemistry analysis on paraffin sections

Mice were euthanized and colons were collected, formalin-fixed overnight at 4 °C and paraffin-embedded. Then, paraffin sections of 3 μm were dried for an hour at 80 °C, dewaxed in xylene, gradually rehydrated in ethanol and washed in PBS. For antigen retrieval, slides were incubated in EnVision FLEX and high pH or low pH solutions (K8004 or K8005, DAKO) for 20 min at 95 °C, depending on each antibody. Then, slides were blocked with endogenous peroxidase through 3% H₂O₂ incubation and washed three times in PBS. Then, primary antibodies were incubated for 30 min at room temperature, washed in PBS and incubated with Horseradish Peroxidase (HRP)-conjugated secondary antibodies. Staining was visualized through reaction with EnVision Detection Kit (K4065, DAKO) using diaminobenzidine (DAB) substrate. Finally, slides were counterstained with Harry's hematoxylin. All primary and secondary antibodies used for immunohistochemistry and their concentrations are listed in Supplementary Materials (Table S6). When required, quantification of positive cells or positive nuclei in immunohistochemical stained sections was performed using QuPath (version 0.5.1) [43]. Automated cell segmentation and detection algorithms were employed to identify and classify cells or nuclei based on stain intensity. Intensity thresholds were established to segment nuclei into categories (negative, weak, moderate and strong) according to DAB chromogen signal, Histoscore (H-Score) quantification was subsequently calculated by combining the percentage of nuclei in each intensity category using the formula: H-score = (1 × % weakly positive) + (2 × % moderately positive) + (3 × % strongly positive), resulting in a value ranging from 0 to 300. Positive cell detection was optimized for nuclear or cellular localization and exported as the percentage of positive cells. All analyses were visually validated, and thresholding parameters were kept consistent across samples to allow quantitative comparison.

TCGA-COAD analyses

For TCGA-COAD analyses, normalized RPM values for miR-503 and miR-424 were obtained from XenaBrowser [44], and IGF1 and MIR503HG expression data were obtained from cBioPortal [45, 46]. Pearson correlation was performed across all samples for miR-503 and miR-424, and group differences were assessed using the Wilcoxon test [47].

Statistical analysis

Statistical analysis was performed according to each experiment. Shortly, comparisons between two groups were analyzed with Student’s t test and represented as mean ± standard error for the mean. Contingency tables were analyzed by χ²-test, followed by Fisher’s exact test. All experiments were performed at least three times, and all experimental groups were done in triplicate. For statistical analysis, GraphPad Prism (Version 8.0, Graphpad Software, Inc.) was used.

Author contributions

MV-S, DL-N, and XD designed the research and developed the project concept. MV-S, DL-N, and XD prepared the manuscript. JE, ME, RR-B, JMS, XM-G, DL-N, and XD provided essential materials and resources necessary for conducting the research. MV-S, NB, RN, JT and XD performed the experiments and collected the data. MV-S, NB, RN, DL-N, and XD analyzed the data and conducted statistical analyses. All authors contributed to data interpretation and critically revised the manuscript.

Data availability

The datasets generated and/or analyzed in this study can be obtained from the corresponding author upon reasonable request.

Competing interests

The authors have nothing to disclose. All authors of this manuscript have participated in the execution and analysis of the study, are aware of and agree to the content of the manuscript. They have approved the final version submitted, being listed as authors on the manuscript. The contents of this manuscript have not been copyrighted or published previously. There are no directly related manuscripts or abstracts, published or unpublished, by one or more authors of this manuscript. The contents of this manuscript are not under consideration for publication elsewhere. The submitted manuscript or any similar script, in whole or in part, will not be copyrighted, submitted, or published elsewhere while it is under consideration.

Supplementary information

The online version contains supplementary material available at 10.1038/s41419-026-08504-8.