-->
- Sequences, structures and databases
- Experimental methods (X-rays, electrons and NMR)
- Finding and visualising structures from the Protein Data Bank
- Comparing structures
- Modelling mutations
- Creating models
!?Proteins are polypeptides of aminoacids
- Folded proteins provide a well-defined 3D arrangement of functional groups, creating microenvironments and active sites.
- Structural changes are often involved in functional mechanisms (motor proteins, ...)
{{0-1}}
Uniprot approx. 570,000 curated sequences (SwissProt)
-
mainly determined by large-scale DNA sequencing of individual genes or whole genomes
-
increasingly automated annotation
{{1-2}}
{{2}}
Protein Data Bank approx. 208,000 experimentally determined structures
- Mainly determined by X-ray crystallography, electron microscopy and recently lesser high-resolution NMR spectroscopy
- Automation is increasing, but there are still significant limitations in the rate of solving new structures
{{3}}
- contains structures of proteins, nucleic acids and complexes
- No purely theoretical or ab initio models (since 2006)
- Also stores supporting experimental data
- Full deposition now required by all peer-reviewed journals
- Use the UniProt site to search for “dnak”.
- Use the PDB site to search for “dnak”.
- Compare the UniProt and PDB result lists.
- Use the sequence search function to see if there are structures with sequences similar to that of the DnaK C-terminal domain.
- Look at the summary pages of a number of structures and note some interesting properties.
- Select a number of structures and create a report page.
{{0}}
Via UniProt
The way of searching for a specific protein structure depends on the data you already have. You might already have the PDB ID (a unique identifier), that's an easy one. But mostly you have the protein name or you just have a sequence. In the last cases I recommend to start from the UniProt website at http://www.uniprot.org, which is the best annotated protein database in the world. Our first model protein will be the molecular chaperone DnaK from E. coli.
Explore a PDB structure on the Uniprot web site
Go to the UniProt website and search for the DnaK protein
- The UniProt search engine returns a list of DnaK protein sequences from a variety of organisms. An entry with accession code P0A6Y8 and entry name DNAK_ECOLI should be near the top of this list.
Click on the accession code (column Entry) to view the protein page of this DnaK from the model organism Escherichia coli.
Click on Structure in the left-side menu and then look at the 3D structure databases table.
Guidelines which PDB structures to select
Which structures (give the 4-character PDB ID) of the C-terminal domain of DnaK should preferentially be use for analysis and why?
Solution
Usually, the recommended selection criteria are using an X-ray structure with low resolution and low
$R_{free}$ factor. Furthermore, the PDB database has pre-calculated a validation report for all of the structures.As an example, have a look at https://www.ebi.ac.uk/pdbe/entry/pdb/4EZX under the section 'Experiments and Validation'. For many PDB structures, there is also a re-done structure available with a vast amount of information on the quality of the X-ray structure and suggested 'better' models e.g. (https://pdb-redo.eu/db/4ezx). In our case, we could opt for the structures 1DKX and 4EZX.
This is a difficult example since there are so many high resolution structures available. So, it is recommended to study the articles and compare the available structures to find your favorite structure for further analysis.
{{1}}
Via the Protein Data Bank by PDB ID
You can find structural information directly at the PDB database. The web site of the PDB consortium is located at http://www.wwpdb.org. This web site provides links to all members of the PDB (left side). It is a question of taste which resource you start off with. For X-ray structures, it is currently PDBe, RCSB PDB, PDBj. For NMR structres, you find the BMRB. In today's course, we focus on the PDB resources only.
If we visit RCSB at http://www.rcsb.org you can start your search for structures.
The PDB entry with ID 1DKX contains the atomic coordinates of the molecular chaperone (DnaK) from E. coli.
Explore a PDB structure on the Uniprot web site
- Go to the PDB website and type 1DKX in the search box
This will lead you to a page with similar information than on UniProt.
{{2}}
Via the Protein Data Bank by sequence
In lots of cases we only have a sequence of which we would like to find out if there is structural information. The PDB can be searched using a sequence as input. Here is the sequence of the C-terminal substrate binding domain of DnaK:
DVKDVLLLDVTPLSLGIETMGGVMTTLIAKNTTIPTKHSQVFSTAEDNQSAVTIHVLQGERKRAADNKSLGQFNLDGINPAPRGMPQIEVTFDIDADGILHVSAKDKNSGKEQKITIKASSGLNEDEIQKMVRDAEANAEADRKFEELVQTRNQGDHLLHSTRKQVEEAGDKLPADDKTAIESALTALETALKGEDKAAIEAKMQELAQVSQKLMEIAQQQHAQQQTAGADASANNAKDDDVVDAEFEEVKDKK
The PDB allows sequence searches through the same search box we used before.
There is also an Advanced Search section, with a Blast/Fasta option in the Sequence Features section.
Hands-on: BLAST search for PDB structure
- Go to the Advanced Search section
- Please select 'Sequence BLAST/PSI-BLAST' in the Query type drop down. This method allows you to change some parameters for the search.
- Copy and paste the sequence in the ''Sequence'' field
- Press ''Submit query''.
- You should see the same structures popping up as you saw in the UniProt page of DnaK.
The primary data format for PDB data is the PDBx/mmCIF format. There is a legacy PDB format which is still very popular.
A PDB (Protein Data Bank) file is a plain text file that contains the atom coordinates of a solved 3D structure of a protein or even DNA. Such coordinate files can be obtained at the Protein Data Bank at http://www.rcsb.org/. Each PDB entry has a unique identifier (ID) consisting of 4 characters, the first one is always a number. Note: It has been announced that the 4 character code will change in the future.
The PDB entry with ID 1DKX contains the atomic coordinates of the molecular chaperone (DnaK) from E coli.
Hands-on: BLAST search for PDB structure
- Go to the PDB website at http://www.rcsb.org/
- Type 1DKX in the search and try to answer the questions in the following step.
Questions
- How many molecules were solved in this PDB file? What kind of molecules are these (proteins, peptides, DNA, ...)?
- Does the structure represent the full protein? If not, how many residues are missing? Hint: Click on the UniProt KB link in the Sequence tab to see the full sequence.
- Was this structure solved by X-Ray or NMR?
- What is the atomic resolution and R-factor?
Solution
- Two, called polymers or chains: they are polypeptides
- To answer this question you can go to the sequence tab at the top:
- Summary: a large chunk of the N-terminus is missing from the structure, the C-terminus is virtually complete.
- X-RAY diffraction, as shown by Experimental Details
- Atomic resolution: 2.00 Ångstrom and R-factor of 0.206
Hands-on: BLAST search for PDB structure
- Go to the PDB website at http://www.rcsb.org/
- Type 1DKX in the search box.
- The file that holds the 3D coordinates can be downloaded by clicking on Download files in the top right corner and then choosing PDBx/mmCIF Format. For convenience, save this file on your desktop. The filename is the 4-character unique PDB ID.
Hands-on: Open downloaded PDB file in an editor
- Open this file with a text editor, e.g. WordPad is an excellent tool for that.
- Do you see the different sections in the PDB file? Analyse some ATOM lines and try to explain what kind of data is in each column.
Additional exercises on searching PDB can be found on the basic bioinformatics exercises page.
{{0-1}}
--{{0}}--
The Protein Data Bank (PDB) format provides a standard representation for macromolecular structure data derived from X-ray diffraction and NMR studies. This representation was created in the 1970's and a large amount of software using it has been written.
{{1-2}}
Atomic descriptions in classical PDB format
ATOM 21 N ALEU A 392 64.898 118.959 40.827 0.60 24.19 N
ATOM 22 N BLEU A 392 64.981 118.958 40.769 0.45 24.73 N
ATOM 23 CA ALEU A 392 64.818 117.251 41.235 0.60 22.53 C
ATOM 24 CA BLEU A 392 64.908 117.527 41.044 0.45 23.55 C
ATOM 25 C ALEU A 392 63.999 117.260 40.013 0.60 21.73 C
ATOM 26 C BLEU A 392 64.972 116.794 39.706 0.45 22.95 C
ATOM 27 O ALEU A 392 64.789 117.304 38.508 0.60 19.91 O
ATOM 28 O BLEU A 392 64.270 117.164 38.760 0.45 21.80 O
ATOM 29 CB ALEU A 392 62.994 117.229 41.239 0.60 22.01 C
ATOM 30 CB BLEU A 392 63.588 117.170 41.739 0.45 23.26 C
ATOM 31 CG ALEU A 392 62.020 116.866 42.907 0.60 22.43 C
ATOM 32 CG BLEU A 392 63.411 117.414 43.240 0.45 24.38 C
ATOM 33 CD1ALEU A 392 63.906 117.128 42.987 0.60 22.36 C
ATOM 34 CD1BLEU A 392 62.004 117.019 43.655 0.45 24.55 C
ATOM 35 CD2ALEU A 392 62.660 117.298 44.405 0.60 22.73 C
ATOM 36 CD2BLEU A 392 64.432 116.611 44.023 0.45 24.44 C
ATOM 37 N AASP A 393 65.422 116.659 39.809 0.60 21.96 N
ATOM 38 N BASP A 393 65.827 115.776 39.625 0.45 22.93 N
ATOM 39 CA AASP A 393 65.972 115.093 38.433 0.60 22.44 C
ATOM 40 CA BASP A 393 65.967 114.977 38.409 0.45 23.84 C
ATOM 41 C AASP A 393 64.830 113.969 38.617 0.60 21.62 C
ATOM 42 C BASP A 393 64.900 113.889 38.543 0.45 22.45 C
ATOM 43 O AASP A 393 64.810 113.091 39.785 0.60 21.11 O
ATOM 44 O BASP A 393 64.962 113.054 39.456 0.45 21.89 O
ATOM 45 CB AASP A 393 66.589 114.675 39.315 0.60 24.09 C
ATOM 46 CB BASP A 393 67.374 114.360 38.335 0.45 26.75 C
ATOM 47 CG AASP A 393 66.965 114.032 38.106 0.60 25.18 C
ATOM 48 CG BASP A 393 67.888 114.196 36.899 0.45 29.40 C
ATOM 49 OD1AASP A 393 66.303 115.375 36.780 0.60 25.41 O
ATOM 50 OD1BASP A 393 67.190 114.588 35.938 0.45 31.28 O
ATOM 51 OD2AASP A 393 68.508 115.272 37.824 0.60 25.36 O
ATOM 52 OD2BASP A 393 69.014 113.671 36.724 0.45 32.22 O
--{{1}}--
The original short term goal of the working group was to fulfill the mandate set by the IUCr: to define mmCIF data names that needed to be included in the CIF dictionary in order to adequately describe the macromolecular crystallographic experiment and its results. In January 1997, the mmCIF dictionary was completed and submitted to COMCIFS for review, and version 1.0 was released in June 1997.
{{2-3}}
Atomic descriptions in mmCIF format ^1
ATOM 21 N N A LEU A 1 4 ? 64.898 118.959 40.827 0.60 24.19 ? 392 LEU A N 1
ATOM 22 N N B LEU A 1 4 ? 64.981 118.958 40.769 0.45 24.73 ? 392 LEU A N 1
ATOM 23 C CA A LEU A 1 4 ? 64.818 117.251 41.235 0.60 22.53 ? 392 LEU A CA 1
ATOM 24 C CA B LEU A 1 4 ? 64.908 117.527 41.044 0.45 23.55 ? 392 LEU A CA 1
ATOM 25 C C A LEU A 1 4 ? 63.999 117.260 40.013 0.60 21.73 ? 392 LEU A C 1
ATOM 26 C C B LEU A 1 4 ? 64.972 116.794 39.706 0.45 22.95 ? 392 LEU A C 1
ATOM 27 O O A LEU A 1 4 ? 64.789 117.304 38.508 0.60 19.91 ? 392 LEU A O 1
ATOM 28 O O B LEU A 1 4 ? 64.270 117.164 38.760 0.45 21.80 ? 392 LEU A O 1
ATOM 29 C CB A LEU A 1 4 ? 62.994 117.229 41.239 0.60 22.01 ? 392 LEU A CB 1
ATOM 30 C CB B LEU A 1 4 ? 63.588 117.170 41.739 0.45 23.26 ? 392 LEU A CB 1
ATOM 31 C CG A LEU A 1 4 ? 62.020 116.866 42.907 0.60 22.43 ? 392 LEU A CG 1
ATOM 32 C CG B LEU A 1 4 ? 63.411 117.414 43.240 0.45 24.38 ? 392 LEU A CG 1
ATOM 33 C CD1 A LEU A 1 4 ? 63.906 117.128 42.987 0.60 22.36 ? 392 LEU A CD1 1
ATOM 34 C CD1 B LEU A 1 4 ? 62.004 117.019 43.655 0.45 24.55 ? 392 LEU A CD1 1
ATOM 35 C CD2 A LEU A 1 4 ? 62.660 117.298 44.405 0.60 22.73 ? 392 LEU A CD2 1
ATOM 36 C CD2 B LEU A 1 4 ? 64.432 116.611 44.023 0.45 24.44 ? 392 LEU A CD2 1
ATOM 37 N N A ASP A 1 5 ? 65.422 116.659 39.809 0.60 21.96 ? 393 ASP A N 1
ATOM 38 N N B ASP A 1 5 ? 65.827 115.776 39.625 0.45 22.93 ? 393 ASP A N 1
ATOM 39 C CA A ASP A 1 5 ? 65.972 115.093 38.433 0.60 22.44 ? 393 ASP A CA 1
ATOM 40 C CA B ASP A 1 5 ? 65.967 114.977 38.409 0.45 23.84 ? 393 ASP A CA 1
ATOM 41 C C A ASP A 1 5 ? 64.830 113.969 38.617 0.60 21.62 ? 393 ASP A C 1
ATOM 42 C C B ASP A 1 5 ? 64.900 113.889 38.543 0.45 22.45 ? 393 ASP A C 1
ATOM 43 O O A ASP A 1 5 ? 64.810 113.091 39.785 0.60 21.11 ? 393 ASP A O 1
ATOM 44 O O B ASP A 1 5 ? 64.962 113.054 39.456 0.45 21.89 ? 393 ASP A O 1
ATOM 45 C CB A ASP A 1 5 ? 66.589 114.675 39.315 0.60 24.09 ? 393 ASP A CB 1
ATOM 46 C CB B ASP A 1 5 ? 67.374 114.360 38.335 0.45 26.75 ? 393 ASP A CB 1
ATOM 47 C CG A ASP A 1 5 ? 66.965 114.032 38.106 0.60 25.18 ? 393 ASP A CG 1
ATOM 48 C CG B ASP A 1 5 ? 67.888 114.196 36.899 0.45 29.40 ? 393 ASP A CG 1
ATOM 49 O OD1 A ASP A 1 5 ? 66.303 115.375 36.780 0.60 25.41 ? 393 ASP A OD1 1
ATOM 50 O OD1 B ASP A 1 5 ? 67.190 114.588 35.938 0.45 31.28 ? 393 ASP A OD1 1
ATOM 51 O OD2 A ASP A 1 5 ? 68.508 115.272 37.824 0.60 25.36 ? 393 ASP A OD2 1
{{3}}
mmCIF file looks like a paired collection of data item names and values.
data_1DKX
#
_entry.id 1DKX
#
_audit_conform.dict_name mmcif_pdbx.dic
_audit_conform.dict_version 5.279
_audit_conform.dict_location http://mmcif.pdb.org/dictionaries/ascii/mmcif_pdbx.dic
#
loop_
_database_2.database_id
_database_2.database_code
PDB 1DKX
WWPDB D_1000172826
#
_pdbx_database_status.status_code REL
_pdbx_database_status.entry_id 1DKX
_pdbx_database_status.recvd_initial_deposition_date 1996-06-03
_pdbx_database_status.deposit_site ?
_pdbx_database_status.process_site ?
_pdbx_database_status.SG_entry .
_pdbx_database_status.pdb_format_compatible Y
_pdbx_database_status.status_code_mr ?
_pdbx_database_status.status_code_sf ?
_pdbx_database_status.status_code_cs ?
#
loop_
_audit_author.name
_audit_author.pdbx_ordinal
'Zhu, X.' 1
'Zhao, X.' 2
'Burkholder, W.F.' 3
'Gragerov, A.' 4
'Ogata, C.M.' 5
'Gottesman, M.E.' 6
'Hendrickson, W.A.' 7
#
_citation.id primary
_citation.title 'Structural analysis of substrate binding by the molecular chaperone DnaK.'
_citation.journal_abbrev Science
_citation.journal_volume 272
_citation.page_first 1606
_citation.page_last 1614
_citation.year 1996
_citation.journal_id_ASTM SCIEAS
_citation.country US
_citation.journal_id_ISSN 0036-8075
_citation.journal_id_CSD 0038
_citation.book_publisher ?
_citation.pdbx_database_id_PubMed 8658133
_citation.pdbx_database_id_DOI ?
#
--{{0}}--
Macromolecular crystals are composed of many individual molecules packed into a symmetrical arrangement. In some crystals, there are slight differences between each of these molecules. For instance, a sidechain on the surface may wag back and forth between several conformations, or a substrate may bind in two orientations in an active site, or a metal ion may be bound to only a few of the molecules. When researchers build the atomic model of these portions, they can use the occupancy to estimate the amount of each conformation that is observed in the crystal. For most atoms, the occupancy is given a value of 1, indicating that the atom is found in all of the molecules in the same place in the crystal. However, if a metal ion binds to only half of the molecules in the crystal, the researcher will see a weak image of the ion in the electron density map, and can assign an occupancy of 0.5 in the PDB structure file for this atom. Occupancies are also commonly used to identify side chains or ligands that are observed in multiple conformations. The occupancy value is used to indicate the fraction of molecules that have each of the conformations. Two (or more) atom records are included for each atom, with occupancies like 0.5 and 0.5, or 0.4 and 0.6, or other fractional occupancies that sum to a total of 1.
-
Nucleic Acid Database: DNA and RNA structures
-
PDB-REDO: automatically re-refined deposited structures, using the latest methods
-
EBI: many online tools for structure analysis
-
Replaced Electron Density Server: convenient overview of quality parameters for crystal structures
-
BioMagResBank: experimental data for NMR- derived structures (lists of distance restraints and other experimentally derived properties)
-
BioIsis: database of SAXS-derived structures
-
EMBL database of SAXS-derived structures
-
EM Databank for cryo-EM structures
General geometric properties (bond lengths and angles, Ramachandran distribution, …): MolProbity
Crystal Structures
- Diffraction data resolution and completeness (PDB)
- Final $ R_{cryst} $ and $ R_{free} $ factors (PDB)
- Local correspondence to electron density EDS
NMR Structures
- Number of distance restraints and additional experimental data sources BMRB
- Local restraint density on-line NMR constraint analyser
Other techniques
- Difficult to generalise: carefully read and consider the published protocol
- PyMOL: high-quality output, good examples on Wiki
- ChimeraX: good documentation on website
- VMD: excellent for the analysis of MD trajectories
- Yasara
- SwissPDBViewer
-
ChimeraX is the next-generation molecular visualization program from the UCSF RBVI.
-
Provides high-performance rendering of large structures and data
-
Calulate fast and robust solvent-excluded surfaces
-
Integration with AlphaFold and ESMFold AI-predicted protein structures (fetch existing or calculate new)
-
Load PDB entry 1TRZ using the File menu.
-
Compare the default representations and use the various controls to change the view of the protein.
-
Explore the Edit and View menus to change various aspects of the graphical representation of the structure.
-
Examine the hydrogen bonding patterns. Display a molecular surface.
-
Create an interesting view of the molecule and save it as a high-resolution image.
There are great videos on working with ChimeraX that you can access via a playlist from Brown Lab.
We suggest to start with the first and the second video of this playlist.
Open the PDB with code 1TRZ in ChimeraX via pasting open 1TRZ into the command field at the bottom. This will fetch the structure from the RCSB Protein Data Bank and load it into ChimeraX. Please have a look at the log window where more information about the structure is shown next to the full command open 1TRZ format mmcif fromDatabase pdb.
The default view is a quite nice one highlighting the interactions of the side chains of three residues with zinc and sodium atoms as well as selected water molecules.
Changing the molecule display can be easily achieved by toggling the icons Show and Hide from the sections Atoms, Cartoons and Styles in the Home menu.
If you click the Hide icon followed by the Show icon of the Atoms section, all atoms and residues are shown by default. Toggling the Cartoons to state Show doen not explicitly show atoms and residues but are merely a impressionistic representation of the structure. The menu command Actions --> Atoms/Bonds --> Show Sidechain/Base, to a certain extent, shows atoms, but only of side chains, not main chain atoms. Mostly to do structure analysis, we want to show only the most interesting residues, the ones we want to analyze, and hide all the others.
The structure of insulin was crystallized together with some water molecules. In many cases, it is no problem to permanently delete those waters via Select > Residues > HOH followed by Actions > Delete. To visualize the water molecules, got to Select > Residues > HOH followed by Atoms/Bonds > Show. Do you see the red water (oxygen) atoms floating around the surface?
There are several ways to show the residues of interest:
From the sequence selector. Via Tools > Sequence > Sequence Viewer, you will see the sequence selector opening. Search for Cys7 from Molecule B, select the residue (will get green) and click on Atoms > Show..
Now show the atoms of His5 in Molecule B using a method of choice.
And now that we’re on it, what is special about the two cysteines we just visualized?
Hiding individual atoms or residues works in the same way as showing them, only now you should go to Hide in the menu.
Most published molecular images show a detailed active site and all the rest is hidden for clarity. From the previous exercise we show the atoms of 3 residues (let’s assume this is our active site). Now secondary structure of the rest of the molecule is also still visible. To hide all that, we do not have to hide atoms, but hide the secondary structure from the rest of the structure. Atoms and residues are balls and sticks, ‘secondary structure’ is an artistic impression of the structure (beta sheet arrows, helix ribbons, …).
So let’s hide many of the secondary structure, but keep just a few stretches around our active site. Our active site is Cys7 (A), Cys7 (B) and His 5 (B). This can be done in several ways. Since we would have to hide almost everything, I propose to hide first everything and then show again those stretches that we want. But if you have a better idea, I would like to hear it.
Hide all secondary structure via Cartoons > Hide
Open the Sequence viewer and select the stretches via the mouse.
Then show stretches of residues 2-10 in Mol B and residues 4-10 in Mol A in Cartoons > Show.
There are still some metal-bound histidines flying around that weren’t hidden because they are metal bound. Hide those histidines by clicking on one of the sidechain atoms, go to Select > Broaden and click Atoms > Hide.
The metal coordination bonds can be hidden simply by selecting the model 1.1 and hide it by un-checking the checkbox next to the colored square.
Et voilà, a publication ready image!
You can put labels on the residues you want to highlight by selecting an atom from a residue (right-click in default mode). Subsequently, you select Actions > Label > Residues and choose the formatting of the label.
Via the same menu, you can change the height to 0.7 A or e.g. 20 pixels.
You can color on all levels: atoms, residues, molecules and objects. So be careful, if you color a residue, all of its atoms will get that color. If you color a molecule, all atoms in that molecule will get that color.
Let’s color the secondary structure (the backbone in our case) of our active site in orange. But the sidechains should keep their Element colors. So we shouldn’t color entire residues, but only a selected atom set. Therefore our selection will be at the atom level, not the residue level. Go to Actions > Color > All Options and unselect the objects to which coloring applies.
Then select the orange color and select Apply. Hopefully, it is a satisfying result.
It would be a pitty that you spent hours creating fancy molecular graphics for that next paper while you can’t continue on the work the next day. That’s why ChimeraX can save the entire session including orientations, colors, views, everything. To save the current session, go to File > Save and select ChimeraX session as File type.
Choose a filename such as MyInsulin.cxs. To load the work again in ChimeraX go to File > Open.
To save the current view to a high quality publication ready image file, click on the Snapshot icon which will save a PNG file at the default location.
Distances between atoms are calculated as follows:
- Select the first atom
- Keep SHIFT pressed and select the second atom.
- Go to
Tools > Structure Analysis > Distancesand click on 'Create'.
To show hydrogen bonds, you can go to Analysis section in the Molecule Display tab. Before you click on the H-bonds icon, it is recommended to define a search space by selecting residues or molecules between which you would like to calculate hydrogen bonds.
To hide or remove the hydrogen bonds, you have multiple choices: use Molecule Display > Hide H-bonds or select the hydrogen bonds object in the Models overview on the right side of the ChimeraX window and click on Close.
It can be very useful and informative to show the molecular surface of a protein. you can visualize cavities, ligand binding sites, etc … To show the molecular surface of one monomer of dimeric insulin, go to Molecule Display > Surfaces > Show.
In our case, 4 surfaces will be calculated which display status and color can be individually changed using the toggles in the Models window.
Try to recreate the rendering of the insulin structure – pdb code: 1TRZ as shown below.
Hints:
- Choose the proper secondary structure scene style;
- Find the correct orientation first;
- Color all backbone atoms in gray;
- Find the residue numbers of the 2 colored helices;
- Color those residues magenta;
- Show the sidechain atoms and the CA of the two histidines and the glutamate;
- Color the sidechain atoms of all residues in the Element color;
- Label the histidines and the glutamate;
- If you need some help how to change the parameters for the label, please have a look at Help > User guide and search for the label command.
Download GroEL via PDB code 1WE3 in ChimeraX.
Try to reproduce (approximately) the following image (hints below):
Hints:
- Load the PDB via
open 1we3 - Zoom out and find the correct orientation
- Delete the ADP, DMS and Mg molecules.
- Color by molecule (every molecule will get another color) and color by gradient (now you need to specify 2 colors, the begin and end color).
- Choose a first color (eg. color with code 0)
- Choose a second color (eg. color with code 300, so you go over the entire color wheel spectrum)
Now, you have explored the ChimeraX interface and acquainted with basic visualisations. You have identified how you can visualise secondary structure elements, surfaces, and hydrogen bonds. And most importantly, you can create publication-ready figures using ChimeraX.
Structures can be structurally aligned to achieve the best possible match between corresponding atoms. It is possible to consider all atoms, or only a subset (such as the Cα atoms).
When the structures are very similar, determining which atoms should be matched to each other is trivial. When there are larger differences, it takes more preparation.
Different algorithms use different combinations of sequence- and secondary of tertiary structure-based information.
- DALI [^1](Holm and Sander, 1993 and Holm 2019 https://doi.org/10.1093/bioinformatics/btz536)
- TM-Align
- DeepAlign
- LGA
- CE
- FatCat
- MUSTANG
- MMLigner
-
Download the five provided PDB files and open them in ChimeraX.
-
Use matchmaker function to align the four last structures with the first one.
-
Check the reported root mean square deviations as well as the number of residues matched.
- Color all structures by B factor and compare the distribution to the local variability of the structures.
We compare structures by structurally aligning them on top of each other. That is, we align structurally equivalent atoms. For now, we will only use CA atoms as a representation of the backbones. You always need to specify:
- source object(s): the structure(s) that needs to be rotated and translated to superpose on anoth er structure
- target object: the structure to superpose on
An optimal alignment is found when the root-mean-square deviation (RMSD) is at a minimum. The RMSD is given as:
where R is the distance between two structurally equivalent atom pairs (CA in our case) and n is the total number of atom pairs.
Hands-on: Data download
Download the following adapted PDB files from Zenodo
1DKX_1.pdb 1DKY_1.pdb 1DKZ_1.pdb 3DPO_1.pdb 3DPP_1.pdb
Hands-on: Loading PDF files and structurally align them
Now load all of them in :
File > OpenType the following command in the command interface:
matchmaker #2-5 to #1Notice that ChimeraX prints the RMSD of every structural alignment in the console.
Color the atoms by their B-factor:
Molecule Display > Coloring > BfactorHigh BFactors are red, low BFactors are blue.
Questions
Do you see a correlation between the BFactors and the variability in the structure?
Solution
-
Similarities in assembly of secondary structure elements
-
So not based on sequence like motifs but on 3D structure
-
Folds represent the shapes of protein domains
!?https://youtu.be/nJVkwdNm_HY !?https://youtu.be/1ljeUscKE9Y !?https://youtu.be/wWHtcbAnOmQ
-
SCOP (http://scop.mrc-lmb.cam.ac.uk/scop/)
Class, Fold, Superfamily, Family The human hemoglobin protein (1HHO-A), is a domain classfi ed by SCOP as the following. Class: all helical proteins; fold: globin-like; superfamily: globin-like; and family: globins.
!?https://youtu.be/Wz6Ugn7BZjg
- CATH (http://www.cathdb.info/) class (equivalent to SCOP class), architecture, topology (equivalent to SCOP fold), and homologous superfamily (equivalent to SCOP superfamily). 1HHO-A is a domain classi ed by CATH as the following. Class: mainly helical; architecture: orthogonal bundle; topology: globin-like; and superfamily: globins.
!?https://youtu.be/mb898MT3eLc
Similarity on structural level: aligning 3D structures
Structure of query protein is known and aligned to PDB structures
Further reading:
!?Presentation by Martin Steinegger about FoldSeek
Compare proteins with low sequence similarity: similar structure implies homology -> same function
Imagine that we would like to find out whether there are similar structures of the human MEAK7 protein in the proteome of S. pombe.
Usually, we would start with doing a BLAST using Uniprot and limit the search space to the proteome from S. pombe.
How many hits do you find and is there any useful hit?
Alternatively, we can use FoldSeek to find similar 3D structures.
How would you find out what structural information is available about MEAK7?
Execute a search in FoldSeek using restriction on the search space for the PDB and AlphaFoldDB.
Did you find anything interesting from this search?
- Missing atoms, residues and loops
- Multiple molecules in the asymmetric unit
- Incomplete oligomers due to coincident crystal and oligomer symmetry
- Cα-only models
- alternate conformations
- multiple models, especially for NMR ensembles
- use of B factors to represent other properties
- other non-standard extensions (PDBQT, ...)
-
Energy terms representing physical interactions
- Covalent bond lengths and angles
- Dihedral angles and van der Waals forces (steric effects)
- Electrostatic interactions and hydrogen bonds
- …
-
Energy can be minimized, and forces can literally be derived from the potential function.
-
Consistency and careful consideration of the properties to be simulated are essential.
Each energy term has a functional form, which includes one or more parameters:
-
Covalent bond energy term
$ E_{bond} = k(r-r_{0})^{2} $ with
$r_0$ being the equilibrium length -
Van der Waals contact energy term
$ E_{VdW} = C_6r^{-6} + C_{12}r^{-12} $ with
$C_6$ attractive,$C_{12}$ repulsive term
The parameters are collectively optimized to reproduce a chosen set of experimentally observed parameters.
A given force field should be used as a consistent system, and can only be used to predict properties that are covered by the training set.
Is designed for quantitative modelling of the contribution of structural interactions to the stability of proteins and protein complexes. It also supports protein/DNA complexes.
The force field describes the different interactions in a protein structure or complex in analytical terms. It has been calibrated using a set of experimentally determined stabilities.
Applications include the optimisation of structures, the calculation of the stability of complexes, and predicting the effect of amino-acid or base-pair mutations on these properties.
-
Load the PDB entry 2AC0.
-
Set an appropriate structure representation.
-
Locate residue Ala159 using the sequence view.
-
Use the swapaa command to change it to a Trp residue.
-
Look at the effect of the substitution on the structure.
-
Mutate Arg273 to an alanine side chain. Discuss the effects of this substitution.
{{1}}
swapaa #6.2:159 trp log true
Additional exercise: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1304567/ Further reading: https://dasher.wustl.edu/chem430/software/chimera/users-guide.pdf
--{{0}}
Computational methods of predicting protein stability changes upon missense mutations are invaluable tools in high-throughput studies involving a large number of protein variants. Site saturation mutagenesis (SSM), or simply site saturation, is a random mutagenesis technique used in protein engineering, in which a single codon or set of codons is substituted with all possible amino acids at the position. There are many variants of the site saturation technique, from paired site saturation (saturating two positions in every mutant in the library) to scanning site saturation (performing a site saturation at every site in the protein, resulting in a library of size [20^(number of residues in the protein)] that contains every possible point mutant of the protein).
We are exploring the tool MutateX as an example for computational site saturation mutagenesis experiments.
For the exercise, we will use the HPC Tier-2 of the UGent. Documentation on how to connect via the Web portal.
- Login to the VSC HPC Tier-2 of the UGent at https://login.hpc.ugent.be
- Open an interactive session on the login server via
Cluster-->login Shell Access - Once you get the prompt, switch to your scratch folder by typing
cd $VSC_SCRATCH - Copy two files from the
/tmpfolder to your scratch folder by typingcp /tmp/mutatex-v1.sif .andcp /tmp/foldx_20241231 . - With the subsequent command, we copy the necessary files from a remote repository to your scratch folder
git clone https://github.com/vibbits/protein-structure-analysis.git - Change the directory by
cd protein-structure-analysis - Switch to the donphan debug cluster by executing
module swap cluster/donphan - Run the deep screen by executing
sbatch mutatex-run.pbs - Wait for max 40 min
- Create a heatmap of the results in the analysis folder.
The goal of this exercise is appreciate how protein interactions can be studied through visual inspection and other software tools. Protein interactions can be classified into different groups regarding the molecular properties and functions of the interacting partners. (These groups are intertwined in several cases.) Some examples include:
- the interactions of proteins with other proteins, small molecules, carbohydrates, lipids or nucleic acids;
- Receptor-ligand interactions;
- Peptide-protein interactions;
- Antigen-antibody interactions;
- Enzymatic interactions, enzyme-inhibitor interactions.
We will start with exploring one crystal structure of the β2 adrenoceptor. Together with the Steyaert lab from VIB, Kobilka published several crystal structures of the β2 adrenoceptor in its various activation states (Rasmussen et al. Nature 2011, 477)
{{1}}
Get the structure
Download the crystal structure 3P0G from the PDB into ChimeraX.
open 3P0GAs you can immediately appreciate, it is a bigger crystal structure with more than one molecule.
Questions
- How many molecules are present in the crystallized structures?
- And how many chain identifiers are used?
Solution
- There are three molecules, chain A Beta-2 adrenergic receptor; Endolysin, chain B Camelid Antibody Fragment, and a small molecule ligand. Also have a look at PDBe 3P0G which gives a very nice overview of the structure and its composition.
- Only two chain identifiers A and B. Sometimes, this leads to issues depending on the software you might want to use for downstream processing.
{{2}}
Some software routines need seperate chain identifiers for molecular entities to work correctly, so we suggest to rename the small molecule to chain L.
Rename the chain of a residue
Select the residue
Tools - Structure Editing - Change Chain IDsIn the subsequent dialogue enter 'L' while keeping 'To the one' highlighted.
Confirm with 'OK'
We first have a look whether we can find out if there are specific interactions of the small molecule ligand with the adrenoreceptor.
In order to do so, we could add hydrogens to all present molecules first.
Add hydrogens to the molecules
Tools - Structure Editing - Add hydrogens
Note: ChimeraX does not need hydrogens to be present to calculate hydrogen bonds.
Select the ligand
Select - Residues - P0GNote: in the Log panel on the right side, you identify that there is a non-standard residue present in the structure.
Calculate hydrogen bonds
Tools – Structure Analysis – H-bondscheck Limit by selection and check the Log checkbox at the bottom of the dialogue and press OK in the subsequent window
{{3}}
Given that hydrogen bonding is dependent on the definition of a hydrogen bond in the program, it is not a bad idea to use other tools to compare the analysis. There are many options to do this online if you look at published crystal structures. Next to the tools which are directly linked out from the web site of the crystal structure at the PDB database you can use the ProteinPlus server
Go to the web site of ProteinPlus and enter the PDB code 3P0G into the search box. After clicking on Go, you should be presented with on overview of tools the ProteinPlus server provides.
We do not go into great detail on all the tools but only mention PoseView. With this tool, you can prepare an automatic sketch of the small molecule-protein interactions.
Questions
- Between which amino acids and the ligand do you see hydrogen bonds using ChimeraX?
- According to PoseView, between which amino acids and the ligand do you see hydrogen bonds?
- What other interactions are presented in the sketch?
- Inspect the visualisation in ChimeraX: Do you see the interactions in ChimeraX as well?
Solution
- In ChimeraX, you observe hydrogen bonds between Asp113A as well as the carbonyl function of Asn312A and the charged amine function.
- PoseView indicates hydrogen bonds between Asp113A as well as the carbonyl function of Asn312A and the charged amine function. Furthermore, hydrogen bonds are indicated between the phenolic OH and Ser207A and Ser203A as well as the amine function and Ser203A.
- Furthermore, hydrophobic interactions are indicated for the methylbenzyl moiety and pi-pi interactions of Phe290A and the phenolic moiety.
- With ChimeraX, those hydrophobic interactions can also be visualised.
- Explore the tool in
Tools > Structure Analysis > Contactsand the help of ChimeraX.
In order to estimate the binding energy between the nanobody and the β2 adrenoceptor, we can use the FoldX tool AnalyseComplex. It is recommended to calculate these binding energies on energy-minimized structures. To illustrate the effect of the energy minimization, we compare the interaction energy of the current crystal structure and its minimized structure.
In case, there is still time, I would recommend to try to use some of your capabilities you learned today and create a superposition of the inactive and active conformation of the β2 adrenoceptor. We take one of the crystal structures which are available: 3SN6
open 3SN6
You might be overwhelmed once the structure is loaded into ChimeraX. In order to get a first quick overview, click on the 'View' buttom in the right menu of the Models window. Then, it is time to look at the PDB entry of 3SN6 in the PDB database to have a first idea on what molecules are in the PDB file.
As you see on the website 3SN6, the chain R consists of 2 molecules, the β2 adrenoceptor and lysozyme. In the corresponding article, it is stated that 'the unstructured amino terminus of the β2AR is replaced with T4 lysozyme (T4L)'.
Since this is an extra molecule in the crystal structure which disturbes our view, we will delete it.
After the manipulation, the overall picture should look roughly like this.
In the following step, we structurally align only the receptors. The rest of the structures will move along. It is suggested to use the first chain A from 3P0G as target and structurally align the receptor from 3SN6.
Investigate the differences in TM helices and the binding of the nanobody compared to the subunit of the G protein.
Tip: Color the secondary structures to better identify the individual chains/units of G protein.
Given that energy-minimization takes a while for this rather large complex, please download the Yasara scene here
Calculate the interaction energies between the chain A and B of the object 3P0G and the RepairObj1, respectively.
Analyze - FoldX - Interaction energy of molecules
Questions
- What is the dG in the two cases?
- Any idea why the difference is rather hugh?
Solution
- first case (X-ray structure): Interaction energy between molecule(s) A and B in object 1 = -9.86 (kcal/mol) second case: Interaction energy between molecule(s) A and B in object 2 = -20.19 (kcal/mol)
- Through the energy minimisation of the Repair Object function, the interactions of the amino acids are optimised.
This command also creates a list of residues forming the interface of the two proteins. Hit space to see the list of residues in the interface.
Tip: This list can also be useful if you want to make visualisations of the interaction site.
Plugin>interface residues between A and B
Plugin>TA66 TA68 IA72 IA127 RA131 AA134 IA135 TA136 SA137 PA138 FA139 KA140 QA142 YA219 VA222 EA225 AA226 LA266 KA267 EA268 AA271 LA272 TA274 LA275 IA278 IA325 YA326 RA328 SA329 PA330 SB27 IB28 FB29 SB30 IB31 TB33 AB50 IB51 eB52 SB56 TB57 NB58 YB100 AB102 VB103 LB104 YB105 EB106 YB107
- Search SCOP/CATH for protein with same fold and known 3D structure
- Align each amino acid of query sequence to a position in the template structure
- Evaluate how well the sequence fits the fold and select best-fit fold
- Build structural model of query based on alignment with selected fold
- HHpred (http://toolkit.lmb.uni-muenchen.de/hhpred)
- Phyre (http://www.sbg.bio.ic.ac.uk/phyre2/html/page.cgi?id=index)
- DescFold (http://202.112.170.199/DescFold/)
Works because:
- Number of different folds in nature is fairly small (approximately 2000-3000)
- 90% of new submissions in PDB have similar folds to those already in PDB
- Not always accurate ^1
- Run as many methods as you can
- Run each method on many sequences from your homologous protein family
- After all of these runs, build up a consensus picture of the likely fold
- Compare function of your protein to function of the proteins with the likely fold
- Compare secondary structure of your protein to that of the likely fold
- When a structure is available for a protein with a similar sequences, it is possible to predict the structure of a new sequence with varying degrees of confidence.
- Use PSI-BLAST/DELTA-BLAST to detect sequences with similar structures.
- All homology modelling procedures start from an alignment of the template and target sequences. The quality of this alignment will have a major impact on the resulting model.
- Available applications include stand-alone programs (Modeller, FoldX, …) and web-based services (such as SwissModel).
-
Energy terms representing physical interactions
- Covalent bond lengths and angles
- Dihedral angles and van der Waals forces (steric effects)
- Electrostatic interactions and hydrogen bonds
- …
-
Energy can be minimized, and forces can literally be derived from the potential function.
-
Consistency and careful consideration of the properties to be simulated are essential.
Each energy term has a functional form, which includes one or more parameters:
-
Covalent bond energy term
$ E_{bond} = k(r-r_{0})^{2} $ with
$r_0$ being the equilibrium length -
Van der Waals contact energy term
$ E_{VdW} = C_6r^{-6} + C_{12}r^{-12} $ with
$C_6$ attractive,$C_{12}$ repulsive term
The parameters are collectively optimized to reproduce a chosen set of experimentally observed parameters.
A given force field should be used as a consistent system, and can only be used to predict properties that are covered by the training set.
Is designed for quantitative modelling of the contribution of structural interactions to the stability of proteins and protein complexes. It also supports protein/DNA complexes.
The force field describes the different interactions in a protein structure or complex in analytical terms. It has been calibrated using a set of experimentally determined stabilities.
Applications include the optimisation of structures, the calculation of the stability of complexes, and predicting the effect of amino-acid or base-pair mutations on these properties.
In order to make FoldX more accessible and integrate its functions into Yasara, dr. Joost van Durme (SWITCH laboratory) made a Yasara plugin module that can apply FoldX functions to structures that are loaded as Yasara objects.
This greatly simplifies the use of FoldX, and allows for a quick visual analysis of the resulting changes in the structures.
More information can be found at wiki FoldX
- Load the 1CRN PDB file.
- Use the “Repair object” option in the
Analysis|FoldXmenu to activate the corresponding FoldX function. - Select the object to repair.
This exports the object as a temporary PDB file, starts FoldX with the appropriate options, and loads the repaired PDB file as a new object in Yasara.
- Compare the original and repaired objects.
- Describe the changes that were introduced.
{{0-1}}
--{{0}}--
In a crystal, a large number of macromolecules are packed together in a regular grid, with consistent orientations and relative distances. When exposed to an X-ray beam, this arrangement gives rise to diffracted rays in specific directions, resulting in discrete spots on the planar detector. By rotating the crystal, a series of images is obtained. From these, the intensities of all the diffracted rays of the crystal can be derived.
{{1}}
| Diffraction Spot Intensities and Phases | Electron density | |
|---|---|---|
|
|
{{0}}
--{{0}}--
Many atomic nuclei, including the ubiquitous hydrogen nuclei, resonate at specific radio frequencies when placed in a strong, uniform magnetic field. The chemical environment of each individual atom slightly modulates its exact resonance frequency. In macromolecules with thousands of atoms, many different effects combine to generate an extremely complicated pattern of chemical shifts, which therefore more or less uniquely identify each atom. Multidimensional spectra allow these frequencies to be assigned to specific atoms.
{{0}}
--{{0}}--
When two atoms are near each other in 3D space, they can exchange magnetization, giving rise to crosspeaks at the intersection of their respective frequencies. This nuclear Overhauser effect (NOE) is used to effectively measure the distances between pairs of atoms, at least qualitatively.
--{{0}}--
After identification of the atoms by means of their unique chemical shifts, distance restraints are derived from the Overhauser crosspeaks. An extended model of the protein is generated, and then condensed into a shape that is consistent with as many of distance restraints as possible.
- Electron microscopy (and especially Cryo-electron microscopy): Electron crystallography and single particle reconstruction
- Small-angle X-ray and neutron scattering (SAXS and SANS)
Docking runs
Assemblies
https://pubmed.ncbi.nlm.nih.gov/35001912/
Foldseek
https://academic.oup.com/bioinformaticsadvances/article/2/1/vbac072/6749558
Further reading
https://kpwulab.com/2021/03/02/chimerax-example-scripts-commands/
https://rbvi.github.io/chimerax-recipes/sidechains/sidechains.html