Background: Canonical hydrogen-bonding multi-nucleotide matches of microRNAs (miRs)
with mRNAs are considered as important in mRNA regulation. MiR "seed" positions 2-8 are frequently
viewed as mRNA partners, but there is ample evidence for use of other (and even non-contiguous) miR
parts. No detailed information is available about canonical matching, and the GC content of the matches
is rarely considered, although it should have a major regulatory potential.
Methods: Sequences of 2586 human miRs and of 5'utr, cds and 3'utr in 18810 human mRNAs were examined
for number and GC content of contiguous Watson-Crick antisense matches of six or more nucleotides
(nt) in successive windows shifted by 1 nt.
Results: Frequency of the antisense matches is within all sectors similar for segments of up to 10 nt
starting at positions 1-10 of miR sequences, with decrease of 3.5 to 4-fold for each 1-nt increment. Adenine
and uracil rich elements (ARE-like) are very frequent in cds and 3'utr. All mRNAs have matches of
up to 10 nt, and most also those of 11-15 nt. The match density is largest in 5'utr, and the match number
in cds. The 5'utr and cds matches average much higher GC content than those of 3'utr. The GC content
of matches is above that for the whole sector in 5'utr and cds, but lower in 3'utr.
Conclusion: Human mRNA matches across miR sequences constitute a positionally similar matrix of
canonical hydrogen-bonding reactivity. This presents ample opportunities for contiguous binding independent
of miR position. The ubiquitous 10 to 15-nt matches could serve as binding foci. Interaction of
miRs with the abundant GC-rich 5'utr and cds counterparts could be important in the regulation of
mRNA-ribosome interaction as well as in mRNA disposal. The lower density and GC content of a majority
of 3'utr matches could mainly support a dynamic regulation by miRs.