1. The dataset 1BCC is from a crystal with no inhibitors, and 3BCC is from one with antimycin and stigmatellin. STG, ANT, MYX, MOA and KRS are from crystals with stigmatellin (SIG), antimycin(AMY), myxothiazol, MOA-stilbene, and kresoxym-methyl; respectively. The asterisk on 1BCC and 3BCC indicates the original refinement of these models used for the coordinates submitted in Spring 1998 and released July 1998.

2. The isotropic overall B-factor was estimated by scaling each dataset against structure factors calculated from the model of the native bc1 complex (1bcc) in which all atomic B-factors were set to 20. To avoid low-resolution information from the solvent/protein contrast (solvent was absent in the model) The relative B-factor obtained was added to 20 to give the B-factor for the crystal. Anisotropic temperature factors B11, B22, and B33 were determined by scaling the raw dataset against Fcalc as described above but using anisotropic scaling.

3. In calculating R-sym, reflections with negative measured intensity were not rejected unless the absolute value was greater than 3 sigma. Therefore, the R-sym value is very high in the highest-resolution shells where the intensity of most reflections is below the noise level, and should not be compared with Rsym values calculated after excluding weak rejections. The French and Wilson method (), as implemented in the CCP4 program truncate, was used to estimate maximum liklihood values of F and s_F from all reflections including those with negative intensity. Only reflections with F>2s_F were used in refinement.

4. Average F/s_F and completeness were calculated in a narrow shell around 3.5 A (shell slightly different for each dataset). F was calculated from I by the truncate method (materials and methods). Completeness was calculated excluding those reflections for which F<s_F.

5. "Effective resolution" is a resolution such that a complete dataset with the same cell parameters and extending to that resolution would have the same number of reflections as the number of reflections in the dataset in question greater than two times the sigma level.

6. The data to parameters ratio is calculated as the ratio of reflections to 4 times the number of atoms in a monomer or (dimer) The factor of four is because x, y, and z coordinates and B-factor are refined for each atom. Because we used noncrystallographic constraints to link atoms in different monomers, the number of atoms in a monomer is more appropriate.

7. Coordinate error is the ESD from cross-validated SigmaA treatment over the same resolution range and using the same bulk solvent correction as used in refinement,

8. Root-Mean-Square deviations from geometry of small molecules as determined by Engh and Huber (), in Ångstroms (bond length) or degrees.

9. B-iso is the average of the isotropic atomic B-factors (refined for each atom to best fit the F-obs); i.e. the overall isotropic B-factor of the dataset. B11, B22, and B33 are the diagonal elements of the remaining anisotropic B-tensor needed to scale the structure factors calculated from the coordinate set and the solvent model to the Fobs. These three values indicate direction-specific disorder in the directions of the a, b, and c axes of the crystal, respectively. In each case lower numbers indicate greater order.

10. Average atomic B-factors for cytochrome b and the occupants of the Qi and Qo quinone binding sites.

11. In calculating mean B-factor for ubiquinone, only the 18 atoms of the head group and first isoprenoid unit were counted. The tail is disordered.