THE PURIFICATION OF A CYSTEINE-DEPENDENT NAD(+) GLYCOHYDROLASE ACTIVITY FROM BOVINE ERYTHROCYTES AND EVIDENCE THAT IT EXHIBITS A NOVEL ADP-RIBOSYLTRANSFERASE ACTIVITY

An NAD(+):cysteine ADP-ribosyltransferase activity was purified from bovine erythrocytes on the assumption that, like pertussis toxin, the enzyme would exhibit a cysteine-dependent NAD(+) glycohydrolase activity. A three-step purification procedure was developed involving (1) precipitation with 40 % (NH4)(2)SO4, (2) binding to a cysteine-Sepharose affinity column, and (3) binding to an NAD(+) affinity column. PAGE showed a single band of M(r) 45000. The enzyme had been purified 47000-fold and had a specific activity of 1900 nmol nicotinamide released/min per mg. A study of the kinetic properties of this enzyme showed saturation kinetics for cysteine (K-m = 4.0 mM). The ability of this enzyme to ADP-ribosylate protein was investigated using re-sealed inverted bovine erythrocyte ghosts. Incubation of the purified enzyme with erythrocyte ghosts and [adenylate-P-32]NAD(+) led to the enhanced dose-dependent labelling of several proteins, a doublet of high M(r) and proteins of M(r) 60000, 55000 and 29000, identified by autoradiography of separated proteins on SDS/PAGE. The enzyme-catalysed labelling of the major component at M(r) 55000 was blocked by pre-treatment of the erythrocyte ghosts with N-ethylmaleimide, a sulphydryl alkylating agent, and the label was released by mercuric ion, but not by hydroxylamine. These experiments suggested that a cysteine residue on the target protein had been mono-ADP-ribosylated. This supposition was further supported by identification of the mercuric-ion-released radiolabelled product as ADP-ribose by HPLC, and the observation that free ADP-ribose was unable to modify the membrane target protein directly.