The TargetScanHuman 7.2 database (http://www.targetscan.org/vert_72/) (14) was used to predict the same seed sequence and consequential pairs of target regions of miR-221/222-3p. The miRbase database (http://www.mirbase.org/) (15) was used to acquire stem-loop sequences, and the RNA structure (http://rna.urmc.rochester.edu/RNAstructureWeb/) was used to construct stem-loop structures. The human miRNA tissue atlas (https://ccb-web.cs.uni-saarland.de/tissueatlas/) (16) was used to determine the expression of miR-221/222-3p in normal tissues.

As an integrated platform linking miRNAs and target genes, the miRNET database (https://www.mirnet.ca/) (17) integrates data from the following miRNA databases: miRTarBase, miRecords, miRanda, miR2Disease, HMDD, PhenomiR, SM2miR, DIANA-TarBase, PharmacomiR, EpimiR, and StarBase but not from TargetScan. Therefore, we combined miRNET and TargetScanHuman to search for the downstream target genes of miR-221/222-3p. In addition, the intersecting genes of both miR-221-3p and miR-222-3p screened by miRNET were further analysed via the DAVID 6.8 database (https://david.ncifcrf.gov/) to determine tissue locations (18).

The Kyoto Encyclopedia of Genes and Genomes (KEGG, https://www.kegg.jp/) is a database used to understand the high-level functions and utilities of biological systems. The enrichment database (https://amp.pharm.mssm.edu/Enrichr/) (19) was used to perform the KEGG analysis. The Catalog of Somatic Mutations in Cancer (COSMIC, https://cancer.sanger.ac.uk/cosmic) is a comprehensive database that contains information on somatic mutations in multiple human cancer types. Funrich software (http://www.funrich.org/) (20) is a functional enrichment analysis tool used to perform ‘COSMIC’ analysis.

STRING (https://string-db.org/) (21), the most widely used database for exploring proteins and their functional interactions, was used to construct a complete PPI network. Cytoscape (https://cytoscape.org/) (22) is an open-source software platform for visualizing complex networks and hub genes among networks selected using the ‘CytoHubba’ plug-in.

A literature search for potential regulatory oles of miR-221-3p and miR-222-3p in cancers was based on PubMed (https://www.ncbi.nlm.nih.gov/pubmed) and the miRCancer database (http://mircancer.ecu.edu/) (23). The Cancer Genome Atlas (TCGA, https://portal.gdc.cancer.gov/) is the most extensive cancer genomics programme that describes the molecular characteristics of more than 20,000 primary cancers and matched normal samples spanning 33 cancer types. OncoLnc (http://www.oncolnc.org/) is a web tool that provides easy access to TCGA data and is widely used to analyse the survival rate of patients with different cancers stratified by miRNA expression. The MiRTargetLink database (https://ccb-compute.cs.uni-saarland.de/mirtargetlink2) provides miRNAs with target genes verified by experiments (24). Gene Expression Profiling Interactive Analysis (GEPIA, http://gepia.cancer‐pku.cn/) is another database used to analyse the survival rate of mRNAs in TCGA data (25). cBioportal (http://www.cbioportal.org/) was used to research applications, mutations, and deletions of target genes in the TCGA database (26).

The mRNA array dataset of colorectal cancer (GSE126092) was obtained from the GEO database (https://www.ncbi.nlm.nih.gov/geo/), and mRNA expression levels were analysed using the GEO2R online tool to identify differentially expressed genes (DEGs) (https://www.ncbi.nlm.nih.gov/geo/geo2r/). The screening criteria for DEGs were as follows: fold change (FC) > 1 and adjusted P < 0.05. The PPI network was constructed using STRING and the hub genes were selected out by the cytoHubba plug-in of Cytoscape according to ‘Degree’.

The clinical study in this project was approved by the IEC for Clinical Research of Zhongda Hospital, affiliated with Southeast University (2021ZDSYLL167-P01). Research involving human research participants was performed in accordance with the Declaration of Helsinki. Twenty patients with coronary artery disease (CAD) from Zhongda Hospital, Southeast University were enrolled in this study. The inclusion criteria for patients were as follows: (1) aged older than 18 years and younger than 75 years and (2) diagnosed with HF-reduced ejection fraction (HFrEF) because of ischaemic heart disease. The exclusion criteria for patients were as follows: (1) had severe hepatic and kidney dysfunction; (2) had autoimmune diseases, tumor diseases or severe pulmonary diseases; (3) were pregnant; and (4) were deemed inappropriate by researchers. The patients were divided into HF and non-HF groups (Supplementary Table 1). Written informed consent for participation in the study was obtained.

HT-29 cells (human colon cancer cells) were purchased from Procell Company (CL-0118, Wuhan, China). Serum from HF and non-HF groups was pooled equally due to limited individual serum volume, serum (5% in complete medium) from patients with or without HF was added to the HT-29 cell culture for 2 h (at 37 °C and 5% CO2). miR-221-3p and miR-222-3p inhibitors (Antisense oligonucleotides, ASOs) were transfected into cells with RNATransMate (Sangon Biotech, Shanghai, E607402). The experiment set non-inhibitor HF serum-treated group as control to verify the specific role of miRNAs. When the cell confluence reached 60-90%, RNATransMate and RNA were added to tubes A and B, which contained serum-free culture medium and diluted. Then, tubes A and B were mixed together. The RNA/RNATransMate complex mixture was allowed to stand at room temperature for 5–10 minutes. After that, the RNA/RNATransMate mixture was added to the wells.

A CCK-8 (CK04, Dojindo, Japan) and BeyoClick™ EdU Cell Proliferation Kit with Alexa Fluor 594 (C0078S, Beyotime, China) were used to detect cell proliferation. The 96-well plate culture plates were preincubated with serum in an incubator for 48 h (at 37 °C and 5% CO2). 10μl CCK-8 solution was added to each well. The plates were incubated for 1–4 hours. The absorbance was measured at 450 nm using a microplate reader. For EdU staining (C0078S, Beyotime, China), EdU working solution preheated at 37 °C was added to the confocal dish (10 μM) and incubated for 2 h. The culture medium was then removed, and 4% paraformaldehyde was added for 15 min at room temperature. After fixation and washing, the penetrating solution was added at room temperature for 10 min. Then, the Click reaction solution was added to each well and the samples were incubated in the dark at room temperature for 30 min. Finally, the nuclei were stained using a Hoechst kit. A confocal microscope (A1RHD25, Nikon, Japan) was used to observe the fluorescence.

RNA was extracted using the TRIzol reagent (15596018, Invitrogen, USA). A miRNA 1st Strand cDNA Synthesis Kit (by stem-loop) (MR101, Vazyme, China) was used for cDNA synthesis. A ChamQ Universal SYBR qPCR Master Mix Kit (Q711; Vazyme, China) was used for qRT-PCR. All procedures were performed according to the manufacturer’s instructions. The sequences of hsa-miR-221-3p were as follows: RT-primer, GTCGTATCCAGTGCAGGGTCCGAGGTATTCGCACTGGATACGACGAAACC; forward primer, AACGGCAGCTACATTGTCTGCT; reverse primer ATCCAGTGCAGGGTCCGAGG. The sequences of hsa-miR-222-3p were as follows: RT-primer, GTCGTATCCAGTGCAGGGTCCGAGGTATTCGCACTGGATACGACACCCAG; forward primer, AGCGCCTAGCTACATCTGGCT; and reverse primer, ATCCAGTGCAGGGTCCGAGG.

Total cell protein was extracted with a protein extraction kit (KGP2100, KeyGEN BioTECH). The protein concentration was detected using a BCA Protein Assay Kit (KGP902, KeyGEN BioTECH). Proteins were separated using SDS-PAGE and the target proteins were transferred to the PVDF membranes. The membrane was blocked with 5% skimmed milk at 4°C for 90 minutes. The membrane was washed with TBST 3 times for 5 minutes each and then incubated with primary antibodies overnight at 4°C in a shaker. Next, the membrane was washed 3 times for 5 minutes each, and incubated with the corresponding secondary antibodies at room temperature for 1 hour. After being washed with TBST 3 times for another 5 minutes each, the membrane was exposed by an automated chemiluminescence imaging analysis system (Tanon 5200, Shanghai).

Data were analyzed using SPSS 25.0 (USA) and GraphPad Prism 8 (USA). Continuous variables are presented as mean ± SD. Comparisons between two groups were performed using independent samples t-tests or rank-sum tests. Categorical variables were compared using χ² tests or Fisher’s exact tests. ANOVA was used for inhibitor intervention experiments. Multiple comparisons were corrected using the Benjamini-Hochberg method (FDR < 0.05). Effect sizes (Cohen’s d, OR) and 95% confidence intervals (CI) were reported for key outcomes.

PubMed miRNA search revealed that “miR‐221” and “miR‐222” are the most commonly used terms names in a variety of studies. The term miR-221/222 serves as the family designation, and their precursors can be processed into two mature isoforms: 3p and 5p. Among these, the 3p isoforms are more frequently documented in the literature and were more prominent in our preliminary analysis, with their functions being better characterized. Therefore, the subsequent experiments will focus specifically on miR-221-3p and miR-222-3p. Similarly, both miR-221-3p and miR-222-3p have mature sequences and are suitable for follow-up research. The mature sequences of miR-221-3p and miR-222-3p in humans are shown in Supplementary Figure 1A, and one target gene, CDKN1B, was used as an example of all the common target genes of both miR-221 and 222-3p. In addition, considering that the stem‐loop structure of precursors is critical for processing into miRNA‐3p or 5p, we further searched for pre-miR-221/222 sequences and subsequently constructed an RNA secondary structure by the ‘RNA structure’ presented in Supplementary Figure 1B. The expression levels and distribution ranges of miRNAs in different tissues and/or organs usually indicate their different roles in tissue development and disease. Next, we analyzed the tissue location of miR-221/222-3p with the ‘human miRNA tissue atlas,’ and the results showed that miR-221/222-3p was widely distributed in the bladder, brain, epididymis, colon, lung, prostate, and heart (Supplementary Figures 2, 3), indicating a potentially wide range of biological functions for miR-221/222-3p. We confirmed miR-221-3p/222-3p as the core research objects, and all subsequent experiments focused on their expression and function.

It is well known that miRNAs usually target mRNAs to inhibit their expression, and thus exert multiple biological functions. Therefore, identifying potential target genes of miRNAs is important. The miRNET webtool includes 11 miRNA-related databases and provides reliable results. We found that 368 genes were targeted by miR-221-3p while 394 were targeted by miR-222-3p. Intersecting these two groups of genes revealed that almost one-third of the genes (124) were targeted by the two miRs (Figure 1A). The biological function of miRNAs is mediated by the mRNAs of target genes. As shown in Figure 1B, these 124 cotargeted genes were widely distributed in various tissues, including the kidney, bone marrow, and testis, indicating that miR-221/222-3p has multiple regulatory functions in the human body.

To reveal the specific biological function of miR-221/222-3p, we performed KEGG analysis of 124 target genes to comprehensively evaluate the role of these miRNAs. These results suggest that miR-221/222-3p and its downstream genes are closely related to cancer. The COSMIC database is a comprehensive database of mutations in cancer genes. Therefore, we further analysed the 124 targeted genes in the COSMIC database, and the results showed that genes were only enriched in the genital tract and were enriched in the cancer gene (Figure 1C). The results showed that miR-221/222-3p was significantly correlated with chronic myeloid leukemia, central carbon metabolism in cancer, non-small cell lung cancer, apoptosis, and the prolactin signalling pathway (Figure 1D).

TargetScan is a web program that provides prediction results for miRNA target genes in humans. Considering that miRNET excludes the potential results of TargetScan, we analyzed miRNET and TargetScan. First, we predicted the genes likely to be targeted by both miR-221 and 222-3p and identified 504 potential genes. Second, we screened the intersection with the miRNET results, and 43 genes were selected by both miRNET and TargetScan, while 461 genes were only predicted by only TargetScan (Figure 2A). Moreover, we constructed a PPI network of 43 intersecting genes, and the results revealed a close relationship among these genes (Supplementary Figure 4). Finally, we performed KEGG analysis of the remaining 461 genes to fully understand the function of miR-221/222-3p. The ErbB signalling pathway and transcriptional dysregulation in cancer were important components of the KEGG pathway (Figure 2B). These results confirmed that miR-221/222-3p is closely related to cancer.

Several pathological processes, such as cardiac fibrosis and hypertrophy related to HF, are regulated by a series of miRNAs (27), especially miR-221-3p and miR-222-3p (12, 28). Malacards is an integrated database that includes genes related to various diseases. To further evaluate the role of miR-221/222-3p in HF, we screened the database with the keyword “HF” and obtained a total of 43 genes from the Congestive HF, Systolic HF, and Diastolic HF clauses. Next, we intersected these 43 genes with miR-221/222-3p target genes that were predicted by miRNET and TargetScan. However, the results (Figure 2C) showed that no genes coexisted in the three groups or even in the malacards according to the miRNET or Targetscan results. Therefore, we searched for 43 genes targeted by miRNET and TargetScan in PubMed to identify potential HF-related genes. Fifteen genes, including BMF, BCL2L11, and TIMP3, were found to have potential functions in HF (Figure 2D). We also constructed a PPI network of a combination of 15 (PubMed searched) and 43 (Malacards searched) HF genes. The complex network structure (Supplementary Figure 5) indicated strong interactions between these 58 genes and showed that miR-221/222-3p could directly or indirectly regulate HF development.

KEGG analysis of the target genes identified by miRNET and TargetScan revealed that these genes were significantly enriched in cancer-related pathways, indicating the critical role of miR-221/222-3p in cancer. To confirm the connection between miR-221/222-3p and cancer, we performed a literature search in PubMed and identified 62 articles related to ‘miR-221-3p and cancer’ and 126 articles related to ‘miR-221-3p’, while 62 and 110 articles related to ‘miR-222-3p and cancer’ and ‘miR-222-3p’, respectively, suggested that miR-221/222-3p is an important tumor factor. The miRcancer database is a literature-supported miRNA-cancer database. The expression of miR-221/222-3p was dysregulated in cervical cancer, hepatocellular cancer, medulloblastoma cancer, and endometrial cancer (Figures 3A, 4A). Therefore, we performed a survival analysis of patients with diverse cancers stratified according to the miR-221/222-3p signature using the OncoLnc database in the TCGA cohort. Kaplan-Meier (KM) curves (Figures 3B, 4B) showed that high levels of miR-221-3p and miR-222-3p usually resulted in poor overall survival (OS) in most cancers, except for STAD and CESC, which indicated the dual effects of miR-221/222-3p in different cancer types.

To further understand the specific role of miR-221/222-3p in regulating cancer, the expression of five experimental validated target genes (CDKN1C, KIT, ICAM1, SSX2IP, and TNFSF10) from the miRTargetLink Human database (Figure 5) among several previous cancers was calculated via the ‘GEPIA’. Consistently, the expression of CDKN1C and KIT was lower in cancers, including KIRC, KIRP, and BRCA, while ICAM1, SSX2IP, and TNFSF10 were higher in cancers, including STAD and CESC, opposing the effects of miR-221/222-3p (Figure 5). KIT was selected as an example for survival analyses for patients in BRCA and KIRC. The results showed that low expression of KIT had an adverse effect on the survival of cancer patients, which was contrary to the effects of miR-221/222-3p in BRCA and KIRC (Figures 6A, B). Finally, somatic mutations in KIT itself are the main cause of cancer (Figure 6C), but this finding suggests that the function of KIT in different types of cancer is partly regulated by miR-221/222-3p.

Patients with HF have been demonstrated to have a higher risk of cancer. Neurohormonal activation or proteins secreted from the failing heart are important mechanisms linking these two diseases (29). In addition, a failing heart usually generates a large amount of abnormal noncoding RNAs, including long noncoding RNAs (lncRNAs) and microRNAs (miRNAs), compared to normal hearts (30). Previous studies have shown that miR-221 and miR-222 are highly expressed in the peripheral blood of patients with ischaemic HF (IHF) (31, 32). Meijers et al. (8) found that post-infarction HF could significantly stimulate intestinal tumor growth in vivo. Theoretically, high levels of miR-221/222-3p could mediate the tumor-promoting effects of HF, especially ischaemia.

Therefore, we selected colorectal cancer as a candidate disease to assess the core role of miR-221/222-3p in cancer. The GSE126092 dataset (10 human colorectal cancer tissues and 10 corresponding normal-appearing tissues) was screened from the GEO database and analysed using GEO2R. The dataset included lncRNA and mRNA data simultaneously and 1226 downregulated genes after deleting unqualified data (Figures 7A, B). Intersection with target genes provided by miRNET and TargetScan revealed that 48 downregulated genes (accounting for approximately 4% of the 1226 genes) were directly regulated by miR-221/222-3p, suggesting that these genes play an important role in colorectal cancer. The PPI network of all downregulated genes revealed an intricate network in which most proteins could interact with a variety of other proteins. In contrast, only a few proteins did not participate in this network (Supplementary Figure 6), indicating both direct and indirect regulation of miR-221/222-3p in colorectal cancer. To illustrate the vital role of miR-221/222-3p in colorectal cancer, we used the cytoHubba plug-in of Cytoscape software to perform a deep analysis of the PPI network and selected the top 10 genes according to degree score from high to low (Figure 7C). Intriguingly, 3 of the genes (IGF1, PIK3R1, and CDH2) of the 10 were targets of miR-221/222-3p, indicating that miR-221-3p and miR-222-3p play undeniable roles in colorectal cancer. Moreover, the expression of these three genes was all significantly lower in colon adenocarcinoma (COAD) patients in the TCGA database, consistent with the trend obeserved in the GSE dataset (Figure 7D). The above bioinformatics results further verify that miR-221/222-3p is involved in an active regulatory network in colorectal cancer, thus providing the justification for our use of HT-29 cells to investigate the ‘heart failure-induced cancer’ mechanism.

Twenty patients with CAD were enrolled, and divided into HF group and non-HF group (no tumor patients included, per exclusion criteria). qRT-PCR showed that miR-221/222-3p levels were elevated in the serum of patients with HF (Figures 8A, B). Compared to those in HT-29 cells treated with normal serum, miR-221/222-3p levels were elevated in cells treated with HF serum (Figures 8C, D).

HT-29 cells were treated with serum from patients with or without HF. CCK-8 and EdU staining revealed that, compared with those in HT-29 cells treated with normal serum, the proliferation rate was significantly greater in cells incubated with serum from HF patients (Figures 8E, F, H). Then, HT-29 cells were transfected with miR-221-3p and miR-222-3p inhibitors (Figure 8G). Compared to those in the HT-29 cells treated with HF serum, there were significant decreases in PCNA expression and proliferation in the miR-221-3p- and miR-222-3p-inhibiting groups (Figures 8H-J).