On the specificity of ribonuclease U2