Knapp et al. 10.1073/pnas.0509411103. |
Supporting Movie 1
Supporting Materials and Methods
Supporting Movie 2
Supporting Movie 3
Supporting Figure 4
Supporting Table 1
Supporting Table 2
Supporting Figure 5
Fig. 4. Stereo diagram showing structures from difference refinement against 5-ns (yellow) and 9-ms (gray) data along with static R (red) and T (blue) structures (light lines). Included are the heme groups, the proximal His (F8), Phe F4, and Asn F7, and main-chain atoms from residues F3 to F8. Labels "A" and "D" refer to the A and D pyrrole rings, respectively. The 5-ns structure shows characteristics of the early intermediate including the buckled heme and displacement of F-helix residues. The 9-ms structure shows attributes of tertiary T-state including the displacement of the hemes toward the subunits interface, movement of F4 into the heme pocket, and displacement of the F helix. {Figs. 4 and 5 were produced with PYMOL [DeLano, W. L. (2002) http://pymol.sourceforge.net].}
Fig. 5. Difference Fourier map HbI* (photoproduct) minus HbI-CO at time delays of 5 ns and 60 ms is shown for CD, E, F, and heme regions of subunit B (A); and the Phe F4 of subunit B (B). These regions are equivalent to those shown for subunit A in Fig. 1. (A) An a-carbon trace (gray) for the CD region, E and F helices, the heme group (salmon), and side chains for CD1, CD3, E7, F7, and F8 (cyan) and F4 (yellow) are shown, with maps contoured at ± 3.5s (blue and red, respectively). Note the extensive structural rearrangement involving the heme group at 5 ns, along with that of the CD region and F helix. (B) Stereo views of the difference electron density are shown for the region around F4 Phe at ± 2.5s in blue and red, respectively, along with the atomic model for the liganded (salmon) and unliganded (cyan) structures. Phe F4 undergoes the largest ligand-linked side-chain rearrangement during the R-to-T transition. As the density maps show, this movement has not occurred at 5 ns after the ligand release but is completed by 60 ms. Note that the elongated positive density near the F4 side chain at 5 ns does not appear to be associated with the F4 T-state; rather it is located near (below and to the left) of the R- state F4 side chain. This density as well as counterpart negative density (located behind the elongated density at a similar location to that evident in Fig. 1C for subunit A) are both close to the R-F4 Cb and most likely result from the early shift of R-F4 Cb, as the F4 side chain responds to the motion of the backbone in this region. An additional isolated peak of positive density can be seen to the left of the F4 T-state position. Given that the integrated density around the entire F4 T-state ring sums to zero in subunit B, and that subunit A and both subunits at subsequent time points in the nanosecond range do not show this peak, we attribute it to noise. Strong density indicating the movement of F4 to its T-state location occurs only later, as is evident in the 60-ms image below.
Movie 1. Movie showing key structural differences between liganded and unliganded structures of M37V, which are nearly identical to those observed in wild-type HbI structures. The a-carbon traces from residues F2 to F9 are shown in gray, the side chains of residue F4 Phe are shown in yellow, the side chains of residue F8 His are shown in cyan, and the heme groups are shown in magenta for the unliganded structure and red for the CO-liganded structure, with the CO ligand in yellow and water molecules shown as spheres. Water molecules are colored by group, with the core of five water molecules present in both R and T states shown in blue, the pair of water molecules observed to be rapidly disordered after deligation shown in green, and the rest of the interface water molecules shown in cyan. [Note that three R-state water molecules used in the integration of Dr(t) values in Fig. 2D are not included in this figure because they would obscure the view of other water molecules.] Note the movement of the heme closer to the interface upon deligation, the ligand-linked transition of residue F4 between the interface and heme pocket, and the ligand-linked reorganization of well ordered water molecules. (Movies 1-3 were produced with PYMOL.)
Movie 2. Difference electron density features at ± 3.5s (blue and red, respectively) in the difference Fourier map HbI* minus HbI-CO at time points between 5 ns and 60 ms. Included is a ribbon diagram of the HbI-CO dimer (gray) with side chains for His F8 (cyan), Phe F4 (yellow), and two key interface water molecules (small cyan spheres). Note the concentration of difference density near the binding site for CO during the early (nanosecond) time points, which distributes to the interface as the protein responds allosterically to the loss of ligand in the microsecond time range. The time delays shown here correspond to the first and second time series of data, as described in Materials and Methods.
Movie 3. Models based on difference refinement at 5 ns, 200 ns, 700 ns, 2 ms, 9 ms, and 80 ms after photolysis. Included are main-chain atoms for residues F3–F8, side-chain atoms for residues Phe F4 (green) and His F8, heme groups (red), and CO ligand (yellow). Main-chain atoms and His F8 are colored by atom type, with carbon in yellow, nitrogen in blue, and oxygen in red. Light lines show the structures of liganded-R (red) and unliganded-T (blue) structures. The early intermediate structure has formed by 5 ns and remains largely unchanged between 5 and 700 ns but then is followed by a clear progression from the R to the T state during the microsecond time domain. It is important to emphasize that each refined structure does not necessarily represent a single intermediate state but more likely represents an average between a mixture of states. These structures are presented to help assess the direction of motion of structural moieties and populations of the underlying intermediates. The very early and late time points are most likely to result predominately from a single species.
Table 1. Data reduction statistics
Series |
Crystal |
Time delay |
X-ray exposure time |
R(F2), % |
R(|F|), % |
Observations |
Unique reflections; (R > 1/total)* |
Overall completeness, % (100-1.6 Å) |
Last shell completeness, % (1.67-1.6 Å) |
1 | Two (of four) | Dark | 100 ps ´ 120 | 7.2 | 4.1 | 37,266 | 10,674/20,574 | 54.5 | 22.1 |
1 | Two (of four) | 5 ns | 100 ps ´ 75 | 8.2 | 4.8 | 34,366 | 9,982/18,629 | 49.3 | 15.8 |
3 | One (of three) | Dark | 2 μs ´ 24 | 6.6 | 3.7 | 36,935 | 10,896/19,823 | 52.5 | 18.4 |
3 | One (of three) | 2 μs | 2 μs ´ 15 | 8.4 | 4.7 | 31,471 | 8,995/17,226 | 45.6 | 10.4 |
Representative statistics for individual crystals (21 images per data set; 9º angular increment).
*R, redundancy.
Table 2. Representative completeness of ΔF(t) data sets
Resolution range, Å | Time series 1 (5-ns time delay) | Time series 2 (3-μs time delay) | Time series 3 (2-μs time delay) |
100–1.60 | 74.2 | 69.5 | 66.8 |
100–3.20 | 88.0 | 85.7 | 84.0 |
3.20–2.54 | 91.8 | 89.0 | 84.4 |
2.54–2.22 | 89.1 | 86.6 | 79.2 |
2.22–2.02 | 87.4 | 83.3 | 75.2 |
2.02–1.87 | 76.9 | 74.5 | 61.9 |
1.87–1.76 | 68.8 | 63.5 | 49.2 |
1.76–1.67 | 58.0 | 48.3 | 34.6 |
1.67–1.60 | 33.7 | 25.0 | 17.5 |
Data from multiple crystals are merged.
Supporting Materials and Methods
Time-Resolved Data Collection.
For each recorded Laue image, the pump-probe sequence was repeated multiple times with data accumulated on a MAR345 image plate detector before the detector readout. The pump-probe sequence was repeated 75 times for the first time series (single 100-ps x-ray probe pulse) and 15 times for the second and third time series (500-ns or 2-ms probe pulse). The wait time between consecutive pump- probe sequences during an image accumulation was 2 s.Difference Electron Density Integration.
Before integration, each map was multiplied by a correction factor s(5 ns)/s(t) to account for differences in S/N ratio [s is the rms value of Dr within the asymmetric unit; it was shown that s is dominated by noise (1)].Within the integration mask, a grid point was included in the integration if a Dr(t) value was above a specified s level (0 level was used in all cases, except for helices where 2s level was used). In the case of helices, absolute values |Dr(t)| were summed to include both positive and negative signal (for improved signal-to-noise ratio) within a radius of 2 Å around the R-state atomic coordinates. Similar helix integration was performed for a difference map free of signal (such a map was calculated by using two HbI-CO data sets) to estimate the noise contribution and subtract it from the integrated |Dr(t)| values.
To compare the magnitude of the difference signal for various helices (Fig. 2B), the integrated |Dr(t)| values for each helix were divided by the number of residues included in the integrated region to obtain the integrated |Dr(t)| values per residue. Similarly, for clusters of water molecules (Fig. 2D) integrated Dr(t) values per water molecules were obtained.
Estimate of Photolysis.
The initial photolysis (average for subunits A and B at 5 ns) was estimated according to two methods. (i) Comparison of the integrated Dr of the negative CO feature with the expected 14 electrons for the fully photodissociated CO results in a photolysis estimate of ~36%. The integration mask consisted of two spheres centered at the bound CO atomic coordinates, with a radius of 1.2 Å. (ii) The integrated CO density from the experimental map was also compared with the similar density (same CO mask) in a reference difference map corresponding to 100% photolysis. This reference map was obtained by using SF amplitudes calculated for the HbI-CO model with and without the bound CO molecule (the model without the CO was subjected to simulated annealing). Based on this approach the initial photolysis was estimated to be ~45%. The scatter in initial photolysis for different crystals was estimated to ±10%.Assuming random photolysis and rebinding, initial photolysis of about 40% will result in a 3:1 ratio of singly liganded molecules compared to fully unliganded molecules; with the observed rebinding rate, this ratio increases to 18:1 with about 10% unliganded hemes in the microsecond time domain. The fraction of doubly unliganded hemes was therefore negligible in this time domain.
Error Estimates for Integrated Density.
Errors in integrated Dr(t) or |Dr(t)| values shown in Fig. 2 were estimated as follows. For the integrated Dr(t) values corresponding to the loss of bound CO molecule and to the photodissociated CO molecule (CO and CO* in Figs. 1 and 2A), errors correspond to the integrated Dr value for the same volumes in a difference map calculated by using two HbI-CO data sets. For the integrated |Dr(t)| values corresponding to the E and F helices and the CD turn (Fig. 2B), the error was calculated as the rms value of the integrated |Dr(t)| across all time points for the H helix (no significant signal was observed for this helix). Similarly, errors for the integrated Dr(t) values for Phe F4 (Fig. 2C) and allosteric water molecules (Fig. 2D) were the rms values across all time points of the integrated Dr(t) for Phe GH3 (R-state) and core water molecules, respectively, which show no signal.Difference Refinement.
In the difference refinement protocol the weighted difference SF amplitudes <DF(hkl, t)> were added to the structure factors calculated from the M37V HbI-CO model (Fcalc + w<DF(hkl, t)>), according to the weighting scheme described in the main text. A total of 5% of the reflections were used to calculate a free R-factor (the test set reflections were identical for all data sets within a time series). Structural models included two conformations with the M37V-CO structure serving both as the major conformation that remained fixed throughout refinement and the starting point of the minor conformation that was optimized. The residues Lys-96 and Phe-97 also included a third conformation that assumes a tertiary T-state position because none of the refinement protocols were capable of moving Phe-97 from its R-state to the T-state location. The fraction of each conformation (occupancy) present in a given subunit was set to the fractional occupancy of the CO ligand of that subunit as determined by the integration of the CO density (Fig. 2A) normalized by the electron content of CO. The fractions of Lys-96 and Phe-97 in their R- and T-state positions were determined in the same way, based on the integration of the Phe-97 ring in its T-state position (Fig. 2C).The minor (optimized) conformations of the model underwent a rigid-body refinement step followed by a simulated annealing step using the CNS software package (2). The models presented here are based on using a simulated annealing step limited to those atoms in the B helix, CD corner, E, F, and G helices, as well as the heme groups. Further refinement did not improve the model as judged by the free R-factor. Refinement using all atoms resulted in free R-factors that were no better than the refinement based on this more limited atom set, although the conventional R-factor was lower for the all-atom refinement. This indicates refinement of all of the atoms with simulated annealing only improved the R-factor by fitting noise. Use of the tertiary T-state model as the starting point for the minor conformation resulted in a refined structure that was within experimental error of that obtained when R-state model was used as a starting point for refinement, demonstrating the robustness of this procedure.
The positions of the heme iron atoms used in Fig. 3 are based on the refined iron atom locations from the difference refinement. The coordinates of each iron were transformed to a coordinate system in which the R-state iron occupies position 0, 0, 0 and the x, y, and z axes are aligned with the principal moments of inertia (from increasing to decreasing) of the R-state heme group. Thus, the z axis provides a vector perpendicular to the heme plane, and the square root of x2 + y2 describes a vector parallel to the heme plane. The alignment of the heme group was carried out with MOLEMAN2 (3), and transformation was applied with LSQMAN (4).
1. Srajer, V., Ren, Z., Teng, T. Y., Schmidt, M., Ursby, T., Bourgeois, D., Pradervand, C., Schildkamp, W., Wulff, M. & Moffat, K. (2001) Biochemistry 40, 13802-13815.
2. Brunger, A. T., Adams, P. D., Clore, G. M., DeLano, W. L., Gros, P., Grosse- Kunstleve, R. W., Jiang, J. S., Kuszewski, J., Nilges, M., Pannu, N. S., et al. (1998) Acta Crystallogr. D 54, 905-921.
3. Kleywegt, G. J. (1997) J. Mol. Biol. 273, 371-376.
4. Kleywegt, G. J. (1996) Acta Crystallogr. D 52, 842-857.