Huang et al . reply:

In response to the points raised by Leslie, Warne and Tate1, we reexamined the structural model that we deposited in the Protein Data Bank (PDB 4GPO) and obtained similar results to those that we reported previously2; i.e., we could observe some small extra density but no ligands. In addition, we asked another protein crystallography group to check the deposited data set, and they were also unable to find ligands in our structure (Jean Jakoncic (Brookhaven National Laboratory), personal communication).

We then checked the electron density maps of our structure with other web servers. On the PDB_REDO web server, the maps from both the 'conservatively optimized structure' and the 'fully optimized structure' did not show extra density for a ligand in either chain A or chain B (http://www.cmbi.ru.nl/pdb_redo/gp/4gpo/index.html). However, on the Electron Density Server (EDS), the map showed the presence of extra electron density in both chains. We do not know why the map on the EDS web server shows the presence of this extra density.

The presence or absence of cyanopindolol in our structural model was carefully discussed in the paper. From our radioligand binding studies, the β1-adrenergic receptor (β1-AR) protein sample used for the crystallization study yielded the same Kd value for dihydroalprenolol as that in cells expressing wild-type β1-ARs. This indicated that most β1-ARs in the protein sample were not bound with cyanopindolol; otherwise, we would have observed a rightward shift in the competition curve. In addition, at pH 4, the ligand binding was decreased by approximately ten-fold compared to that at pH 7.7. From our mass spectrometry analysis, we estimated that, if cyanopindolol were present, <1% of receptors would be bound with cyanopindolol. One could argue that only cyanopindolol-bound β1-ARs were stable; however, in our hands, purified β1-AR(m23) proteins were stable at pH 4 for up to 6 months.