Natural Seed Informed and Structure Aware Computational Design of Artificial miRNA Seed Regions Targeting the KRAS 3′UTR
Amirhossein Amini Soror,1,*Mohammad Taghizadeh,2Arezoo Rezaei,3
1. Department of Biology , NT.C , Islamic Azad University , Tehran , Iran 2. Department of Biology , NT.C , Islamic Azad University , Tehran , Iran 3. Malek Ashtar University of Technology (MUT), Tehran, Iran
Introduction: KRAS is a major cancer-associated oncogene whose dysregulation contributes to the development and progression of several malignancies. MicroRNAs (miRNAs) are important post-transcriptional regulators of gene expression, with target recognition being strongly influenced by complementarity within the seed region. However, sequence complementarity alone may not fully represent the structural context of miRNA target sites within the 3′ untranslated region (3′UTR). This study aimed to characterize the sequence and secondary-structure features of natural miRNA seed sites targeting the KRAS 3′UTR and to use the identified patterns as a basis for the computational design of Artificial Seed candidates.
Methods: The human KRAS reference transcript and its 3′UTR sequence were retrieved and analyzed. Candidate miRNAs targeting the KRAS 3′UTR were identified using TargetScan and filtered based on Context++ Score Percentile. RNA secondary structure was predicted using RNAfold and visualized with FORNA. Natural seed sites were mapped onto the predicted structure and characterized according to seed length, nucleotide composition, structural position, and Loop:Stem distribution. Recurrent structural patterns identified among Natural Seeds were subsequently used as design criteria for generating Artificial Seed candidates at selected unoccupied positions within the KRAS 3′UTR. Natural and Artificial Seeds were further compared using a study-specific computational scoring approach and statistical analysis of their Loop/Stem distributions. Finally, known sequence variants within the KRAS 3′UTR were examined for overlap with the designed Artificial Seed sites.
Results: Sixty miRNAs met the selection criteria, corresponding to 61 Natural Seed sites within the KRAS 3′UTR, including 28 7-mer and 33 8-mer sites. Structural mapping identified 204 nucleotides in Loop regions (44.35%) and 256 in Stem regions (55.65%) among the Natural Seeds. Based on recurrent sequence and structural characteristics of these sites, 60 Artificial Seed candidates were designed, comprising 30 7-mers and 30 8-mers. The Artificial Seeds contained 210 Loop nucleotides (46.67%) and 240 Stem nucleotides (53.33%). Direct comparison of Loop/Stem distributions between Natural and Artificial Seeds showed no statistically significant difference (χ² = 0.493, df = 1, P = 0.482). In addition, seven unique known variants within the KRAS 3′UTR overlapped with at least one designed Artificial Seed site, highlighting sequence variation as an additional consideration in candidate prioritization.
Conclusion: This study presents a Natural-seed-informed and structure-aware computational framework for Artificial Seed design in the KRAS 3′UTR. Rather than relying solely on sequence-based design, the proposed approach derives design criteria from the sequence, structural, and positional characteristics of naturally occurring miRNA target sites and incorporates sequence variation into candidate evaluation. The resulting Artificial Seeds represent computational candidates for further prioritization and experimental validation. Future studies incorporating target accessibility, hybridization energetics, off-target assessment, and functional validation will be required to determine their biological activity and potential relevance to KRAS regulation.