Nat. Biotechnol., published online 19 February 2012, doi:10.1038/nbt.2121

Human poly-ADP-ribose polymerases (PARPs) share a common catalytic domain that facilitates the transfer of ADP-ribose. Perhaps the best-studied PARP family member is PARP1, which participates in the DNA damage response. On the basis of observations suggesting that blocking PARP1 sensitizes certain cancer cells to drugs or radiation, PARP1 inhibitors are currently being tested in clinical trials. The biological function of many of the remaining human PARP enzymes has not been well characterized, in part because of a lack of selective chemical probes. Thus, Wahlberg et al. set out to determine the specificity of known and potential PARP inhibitors. The team looked at the binding of 185 compounds, including clinical PARP inhibitors, to purified catalytic domains from 13 of the 17 human PARPs. One subset of compounds was relatively nonselective, binding up to 10 of the 13 PARPs tested. A second subset bound preferentially to PARP1, PARP2, PARP3 and PARP4. A third subset was selective for the tankyrases, PARPs initially described as telomerase regulators. Very few of the compounds tested, and none of the clinical compounds, were highly selective for a single PARP. Cocrystal structures identified distinct binding modes for nonselective, PARP1–PARP4-selective and tankyrase-selective compounds. These biochemical and structural insights may enable the development of highly selective tool compounds, including those for relatively uncharacterized family members such as PARP10 and PARP14 that catalyze mono-ADP-ribosylation rather than the prototypical poly-ADP-ribosylation.