Evolution of novel enzymatic activities is believed to require a period of gene sharing in which a single enzyme must serve both its original function and a novel function that has become advantageous to the organism. The potential for evolution of the novel activity depends on many factors, including the effect of mutations that improve the novel activity on the efficiency of the original activity, and the possibility of changes in gene expression that compensate for poor enzymatic activity. We have investigated the evolution of a novel activity under circumstances in which both the original and the novel activity are critical for growth. ProA (glutamyl phosphate reductase) has a very low promiscuous activity with N-acetylglutamylphosphate, the normal substrate for ArgC (N-acetylglutamyl phosphate reductase). A mutation that changes Glu383 to Ala increases the promiscuous activity by 12-fold, but decreases the original activity by 2800-fold. The impairment in proline and arginine synthesis results in 20-fold overexpression of E383A ProA, providing sufficient N-acetylglutamylphosphate reductase activity to allow a strain lacking ArgC to grow on glucose. Thus, reaching the threshold level of N-acetylglutamylphosphate reductase activity required for survival required both a structural mutation and overexpression of the enzyme. Notably, overexpression does not require a mutation in the regulatory region of the protein; amino acid limitation due to the poor catalytic abilities of E383A ProA causes a physiological response that results in overexpression of the enzyme.