Docking HSP90 in Negative-Stain Electron Microscopy

HSP90 EM mapHSP90 PDB 4ipe

Tom Goddard
December 11, 2017
UCSF Biophysics 204A tutorial

How to dock an HSP90 atomic model (x-ray) into a low resolution (~20 Angstrom) negative stain electron microscopy single-particle map using UCSF Chimera version 1.12.

Color chains of dimer

OperationChimera menu entry
Open HSP90 protein, PDB 4ipe File / Fetch by ID.
Select chain A Select / Chain / A.
Color chain A orange Actions / Color / Orange.
Clear selection Ctrl-click on background.

Hand align atomic model with map and optimize

OperationChimera menu entry
Open HSP90 map File / Open....
Adjust map surface threshold Volume Viewer panel, move vertical bar on histogram.
Freeze map motion Favorites / Model Panel. Click active button (column headed by "A") off for map.
Hand dock model Drag atomic model into map with mouse.
Unfreeze map Click active button in Model Panel for map.
Optimize fit Tools / Volume Data / Fit in Map. Press Fit button on Fit in Map panel.
Make surface transparent Press Volume Viewer panel color button, click opacity, change A (opacity) value.

This fit optimization rotated and shifted atomic model to maximize the average map value at the atom positions. Reports number of atoms outside displayed map contour, for example

1754 of 9940 atoms outside contour

Measure correlation coefficient

OperationChimera menu entry
Fit with simulated map Press Options button on Fit in Map,
click Use map simulated from atoms...,
set resolution 15 Angstroms
Set optimization mode click Optimize correlation
Optimize fit click Fit.
Show simulated map. In Model Panel check button in S (show) column for molmap 4ipe res 15.

Fit in Map panel reports correlation coefficient and average map value, for example,

Correlation 0.8675 Average map value 1.034

Correlation values compare two maps and range from -1 to 1, with 0 indicating no correspondence, and 1 identical maps.

Is the Atomic Model Dimer Symmetric?

Notice that atomic model pokes out of map surface in an asymmetric way, the helices and turns of the two chains that poke out do not match. Why?

Is the 4ipe atomic model symmetric?

OperationChimera menu entry
Open second copy of 4ipe Press lightning bolt icon in lower right corner, click 4ipe button.
Hide maps unclick shown column buttons for maps in Model Panel
Align chain B of one 4ipe
with chain A of other 4ipe
Tools / Structure Comparison / Matchmaker.
Chain pairing: specific chains with specific chains.
Reference chain #0 A, match #2 B.
Press Apply.

Note the N-terminal domains align well but the C-terminal domains misalign by more than 10 Angstroms.

Is the map C2 symmetric?

Map single-particle reconstruction imposed C2 symmetry. But let's check that it really has C2 symmetry.

OperationChimera menu entry
Open second copy of map Press lightning bolt icon in lower right corner, click Hsp90_ATP_Cryosparc_C2.mrc button.
Hand align map copy to original map Click line in Model Panel for map copy, press Activate Only button. Move map copy with mouse to align with original copy. Press Activate All button.
Optimize fit of map copy in original map
Tools / Volume Data / Fit in Map.
Choose map copy and original map in Fit panel top line. Press Fit
Rotate map copy and fit again
Use Model Panel to activate only map copy, approximately hand rotate 180 degrees. Press Fit button.

Note correlation coefficient is 1 for map fit into map even after 180 degree rotation. Has exact 2-fold (C2) symmetry.

Are other HSP90 atomic models symmetric?

Do other HSP90 atomic dimer models have symmetry? Align another PDB HSP90 model 5tth to 4ipe to see how similar they are.

OperationChimera menu entry
Open 5tth File / Fetch by ID
Align 5tth to 4ipe Tools / Structure Comparison / Matchmaker.
Reference 4ipe chain A, match 5tth chain B.
Press Apply.
Check RMSD difference. Favorites / Reply Log

Visually 4ipe is nearly identical to 5tth except the latter has an extra domain. The reply log reports 0.464 Angstroms root mean square deviation (C-alpha atoms) for 587 paired residues.

RMSD between 587 pruned atom pairs is 0.464 angstroms; (across all 588 pairs: 0.471)

Can find 348 other PDB structures with high sequence similarity (BLAST E-value < 1e-38) using menu Tools / Sequence / Blast Protein. Can select some or all and press "Load" to load them and automaticallly align them.