An anomalous effect of N-isopropyl substitution in determining beta-adrenergic activity