Data collection, data processing, and structure determination
Data collection was carried out at beamline 21ID-D of LS-CAT at the Advanced Photon Source (Argonne National Laboratory). Data were processed, integrated, and scaled with the programs Mosflm and Scala in the CCP4 package [52]. Phasing was achieved by using the molecular replacement method, which was carried out by using the program Phaser [53] that is implemented in the PHENIX package [54]. The monomeric structure of the E. coli PPase (PDB code 4um4) was used as a search model. This structure is chosen as a search model because All plausible space groups of the P222 family were tried. A single solution was found in the space group P212121. Interactive model building was carried out with Coot [55]. Structure refinement was carried out by PHENIX. Structure determination statistics are shown in Table 1. The program UCSF Chimera [56] was used to generate all of the molecular graphic figures. Figures for electrostatic surface rendering were generated by Chimera using the Coulombic Surface Coloring tool with the default settings.
Sequence alignments of the eukaryotic soluble PPases with known structures were carried out by Clustal Omega [57]. Molecular interactions at the PPA1 homodimerization interface were depicted by using the program LIGPLOT [58]. Atomic coordinates and diffrac­tion data have been deposited in the Protein Data Bank with accession code 6C45.

Structural modelling of PPA1 with various putative substrates

The determined crystal structure of human PPA1 does not contain substrate at the putative active site. To gain insights into the possible catalytic mechanism of human PPA1, modelling studies were carried out to place various putative substrates at the active site. At first, a pyrophosphate molecule was docked into the active site by using AutoDock Vina [59]. Chimera was used to prepare the PPA1 receptor structure for docking and to visualize the docking results. To explore whether the PPA1 active site is capable of accommodating peptide substrates containing phosphorylated tyrosine and/or phosphorylated threonine residues, a series of phosphotyrosine and/or phosphothreonine containing peptides were modelled into the PPA1 active site by using Chimera (including energy minimization of the bound peptide at the active site). The phosphotyrosine and/or phosphothreonine containing peptides were initially derived from the phosphorylated MAP kinase Erk2 (PDB code 2erk) [60]. Two Erk2-derived tetrapeptides were used: 184-EYpVA (superscript p indicates phosphotyrosine) and 181-FLTpE (superscript p indicates phosphothreonine), to model a Yp and a Tp at the active site respectively. To prepare peptides corresponding to the JNK1 sequence for modelling studies, the 181-FLTpEYpV sequence in the Erk2 structure were mutated to JNK1 sequence 181-MMTpPYpV by Chimera. A pentapeptide 182-MTpPYpV was used to model the Yp at the active site. A hexapeptide 181-MMTpPYpV was used to model the Tp at the active site.