For validated (or predicted) interactions, I would recommend miRWalk2.0 for this. This allows you to search for predicted or validated (via CLASH) mir-to-gene interactions. It has data for gene promotors, 5'UTR, CDS, and 3'UTR.
Alternatively, if you have a list of mirs and want to find in silico predicted gene targets, then miRNAtap is useful and is on Bioconductor. It uses information from 4 databases in order to increase confidence in predictions:
Finally, if you want to perform functional enrichment via GO and KEGG for each mir's predicted targets (and, thus, each mir), you can again use miRNAtap for GO (which ultimately uses topGO), and KEGGprofile for KEGG pathways.