Poly(ADP-ribosyl) transferase (pADPRT) catalyzes the transfer of the ADP-ribose moiety from NAD+ onto proteins as well as onto protein-bound ADP-ribose. As a result, protein-bound polymers of ADP-ribose are formed. pADPRT itself contains several acceptor sites for ADP-ribose polymers and may attach polymers to itself (automodification). In this study the influence of substitutions in the purine base of NAD+ on the polymerization reaction was investigated. The adenine moiety of NAD+ was replaced by either guanine, hypoxanthine or 1,N6-ethenoadenine. These analogs served as substrates for polymer synthesis as judged from the extent of automodification of the enzyme and the sizes of the polymers formed. Time course experiments revealed that 1,N6-etheno NAD+ (epsilon-NAD+) and nicotinamide hypoxanthine dinucleotide (NHD+) were rather poor substrates as compared to NAD+. Synthesis of GDP-ribose polymers from nicotinamide guanine dinucleotide (NGD+) was more efficient, but still significantly slower than poly(ADP-ribosyl)ation of the enzyme using NAD+. The size of the different polymers appeared to correlate with these observations. After 30 min of incubation in the presence of 1 mM substrate, polymers formed from epsilon-NAD+ or NHD+ contained up to 30 epsilon-ADP-ribose or IDP-ribose units, respectively. Using NGD+ as substrate polymers consisted of more than 60 GDP-ribose units, an amount similar to that achieved by poly(ADP-ribosyl)ation in the presence of only 0.1 mM NAD+ as substrate. These results suggest that the presence of an amino group in the purine base of NAD+ may facilitate catalysis. Substitution of the nicotinamide moiety of NAD+ with 3-acetylpyridine had no detectable effect on polymer formation. Oligomers of GDP-ribose and epsilon-ADP-ribose exhibited a slower mobility in polyacrylamide gels as compared to ADP-ribose or IDP-ribose oligomers. This feature of the two former analogs as well as their markedly attenuated polymerization by pADPRT provide valuable tools for the investigation of the enzymatic mechanism of this protein. Moreover, polymers of epsilon-ADP-ribose may be useful for studying enzymes degrading poly(ADP-ribose) owing to the fluorescence of this analog. Digestion of epsilon-ADPR polymers with snake venom phosphodiesterase was accompanied by a significant fluorescence enhancement.