Missing residues and incomplete sidechains

From Proteopedia

(Redirected from Missing residues)
Jump to: navigation, search

Most empirical models are missing residues that were present in the experimental material, and/or sidechain atoms, and therefore are often missing charges present in the intact molecule. These missing parts make errors in the shape of the molecule as well as the distribution of charges, hydrophobic and polar regions, and often there are missing salt bridges and cation-pi interactions. FirstGlance in Jmol makes it less likely, compared to other popular viewers, to overlook missing parts. AlphaFold models, which lack no atoms (other than hydrogen) should be analyzed alongside empirical models in order to see complete structures.

Contents

Missing Residues

In about 80%[1] of the Empirical models in the PDB, some residues (amino acids or nucleotides) that were present in the experimental material are absent (have no coordinates) in the empirical model. X-ray crystallography gives a clear electron density map only where every molecule in the protein crystal has the same conformation. Usually, some parts of the molecule vary in conformation between copies in the crystal, that is, some regions are disordered. The same may occur with protein molecules on a cryo-electron microscopy grid. These disordered portions of the molecule are not clearly resolved in the density map used to construct the structure model. Without density to guide where to place these residues, the experimenter omits them from the model. These are called missing residues. It is very common for a few residues at the ends of protein chains to be missing in the atomic model. (Example: 5 residues are missing from the carboxy terminus of the protein in 1ijw.) In addition to disorder rendering residues "invisible", residues can actually be lost by proteolysis during purification, or radiation damage during X-ray crystallography can sometimes remove atoms[2], and conformations can be affected by crystallization conditions[3].

To emphasize, the missing residues were believed to be present in the experimental material, but are absent in the resulting atomic model. Missing residues are distinct from intentionally deleted residues that were not present in the experimental material due to engineering of the cloned and expressed protein.

Get The Best Model Available

Before you spend time working with an empirical model, make sure you have chosen the highest resolution and most complete model available. Search for sequence-identical (or nearly identical) models. Things will be much easier if the sequence numbers in the empirical model are the same as the UniProt numbering. If not, consider renumbering the PDB file.

Missing Ends of Chains

Unlike other viewers, FirstGlance in Jmol ensures that you are aware of missing ends by marking them with spherical "empty baskets". In the example shown below, 2ace, the 3 missing N-terminal residues DDH have net negative charge.

2ace amino-terminus missing 3 amino acids

P04058 AlphaFold:
None Missing

Image:2ace-fg-missing-nterm.png

FirstGlance "empty basket".

PyMOL

ChimeraX

Molstar

iCn3D

FirstGlance

Missing Loops

FirstGlance ensures that you are aware of missing loops with an ellipsoidal "empty basket". Other viewers use a dotted line, which is easier to overlook when viewing the entire structure. Empty baskets are easily hidden[4].

2ace missing 5 amino acid loop

P04058 AlphaFold:
None Missing

FirstGlance "empty basket" (see #below)[5]

PyMOL

ChimeraX

Molstar

iCn3D

FirstGlance

See For Yourself

AlphaFold models have no missing atoms

When all empirical models of the protein of interest have missing atoms, the best way to get a model without missing atoms is to download the AlphaFold model. The downloaded PDB file can then be uploaded to FirstGlance. iCn3D will accept the UniProt sequence ID to retrieve the AlphaFold model directly.

Models from the AlphaFold Database are limited to single-chain structures without ligands, but starting in 2024, see How to predict structures with AlphaFold for servers that can predict multiple-chain structures (protein, DNA, RNA) with ligands or modified residues.

The AlphaFold model will usually be nearly identical to the empirical model (you should verify this by superposition), but will include the missing residues/atoms. Structures of missing loops will be predicted by AlphaFold, but typically with lower reliability (higher pLDDT) than the remainder of the structure. Many long loops that are missing in empirical models are actually intrinsically disordered, in which case the AlphaFold prediction will have very low reliability and be meaningless.

Why Missing Residues Matter

Where residues are completely missing in empirical models, the shape of the molecule will be incorrect, and when charged residues are missing, the distribution of charges will be incorrect, as well as the distribution of hydrophobic and polar patches. Salt bridges and cation-pi interactions may be missing.

Example: 5nyp

5nyp is a bacterial protein believed to be an ancestor of the 20S proteasome[6]. All sequence numbers given below are UniProt numbering. Subtract one to match numbering in 5nyp.

  • 5nyp has 3 missing loops, clustered on one face. Their lengths are 8, 10, and 12 residues (20-27, 102-111, 183-194).
  • None of the 3 missing loops are predicted to be intrinsically disordered by RCSB[7]. The first missing loop is predicted to be disordered by flDPnn2[8]. The other two are not. But flDPnn2a predicts disorder for 71-83, which is not missing.
  • The missing loops include 7 charged amino acids (4–, 3+).
  • Although the authors did not detect proteolytic activity with the substrates tested, the putative catalytic triad residues Thr2, Asp18, and Lys33 (UniProt numbers) are present.
  • When present in the AlphaFold model, two of the missing loops partially obscure the putative catalytic triad.
  • Two salt bridges are missing in 5nyp that are present in the AlphaFold model. UniProt Glu107 (present) forms a salt bridge with missing Arg99. UniProt Asp183 forms a salt bridge with arg187 -- both are missing in 5nyp.

Three Missing Loops in 5nyp. Snapshots from FirstGlance.

5nyp: missing loops (in front) as ellipsoidal "empty baskets"[9].
N               C

5nyp: yellow halos on putative catalytic triad, showing proximity to missing loops (in front).

AlphaFold-predicted structure for F8JB59. Loops missing in 5nyp are at top. Confidence (>70 is "confident"):

AlphaFold Prediction Superposed on 5nyp

The three missing loops are present in the AlphaFold prediction for UniProt F8JB59. FATCAT superposed the AlphaFold prediction onto all 213 alpha carbons of 5nyp with RMSD 2.4 Å. The FATCAT-generated morph between the two models shows their similarity. Their similarity shows that the AlphaFold model is largely correct.

AlphaFold Prediction Superposed on 5nyp

Morph Between Models

Image:Morph-5nyp-af-cut2.gif

Missing loops at top.

Captured from FATCAT.

Missing Loops Obstruct Catalytic Site

Missing loops affect the shape of the molecule. The putative catalytic site is exposed on 5nyp, but when the 3 loops are added in AlphaFold's prediction for UniProt F8JB59, they partially obstruct the putative catalytic site.

5nyp

AlphaFold-Predicted Structure

Image:5nyp-charge-catres.gif

Image:AF-F8JB59-charge-catres-loops.gif

Charged atoms colored by FirstGlance: Positive +, Negative –. Putative catalytic triad: yellow halos. Loops predicted by AlphaFold in dark gray.

Missing Charges in 5nyp

The three missing loops include 4 negatively-charged amino acids, and 3 positively charged ones. Their absence affects the distribution of charge on the surface, as shown in these electrostatic potential maps generated by iCn3D (from a link in the Views tab in FirstGlance).

5nyp: 3 loops with 7 charges are missing.

AlphaFold-Predicted Structure for F8JB59: No missing charges.

Electrostatic potential maps by iCn3D (from a link in the Views tab of FirstGlance):
Positive +, Negative –.

Missing Salt Bridges in 5nyp

Two salt bridges are missing in 5nyp as a result of two of the missing loops. At left: Missing UniProt Glu107 forms a salt bridge with Arg99 (present). At right: UniProt Asp183 forms a salt bridge with Arg187 -- both are missing in 5nyp. A salt bridge also forms between two of the putative catalytic triad residues, Lys31 and Asp169. (All 3 triad residues are present in 5nyp.)

AlphaFold-Predicted Structure for F8JB59.

Salt bridges rendered by FirstGlance: Positive +, Negative –. Yellow halos: Charges missing in 5nyp. Green halos: Putative catalytic triad, present in 5nyp. Black backbone: loops missing in 5nyp.

Incomplete Sidechains

In about half[1] of the Empirical models in the PDB, some residues have coordinates missing for some of their sidechain atoms, due to local disorder. But their main chain atoms are present in the model; thus, they are residues with incomplete sidechains. For example, the long sidechain of a lysine on the surface of a protein may have too blurry an electron density to indicate its position. In some cases, the model builder may give that sidechain coordinates with high temperatures, or low or zero occupancy (example: Arg321 in 2ade[10]). In other cases, the model builder simply omits the coordinates for the sidechain, so the aforementioned surface lysine may have the sidechain of an alanine (example: Lys498 in 2ace).

Among the five viewers shown below, only FirstGlance alerts you to incomplete sidechains in its initial view. It marks them S—. FirstGlance and iCn3D are the only ones that show disulfide bonds in their initial views. The S— labels are easily hidden when desired[11].

Initial views of incomplete sidechains in 2ace

FirstGlance marks incomplete sidechains with S—, and shows disulfide bonds.

PyMOL

ChimeraX

Molstar

iCn3D shows disulfide bonds.

Do incomplete sidechains matter?

Obviously, incomplete sidechains matter when they occur near a region of interest in an empirical model. The easiest way to get a model without incomplete sidechains is to use the AlphaFold model, bearing in mind that residue positions and sidechain rotamer orientations may be inexact.

Atoms colored by charge

When the charged atoms are missing in incomplete sidechains, there is little effect when spacefilling atoms are colored by charge in FirstGlance or iCn3D, because they color the entire residue by charge. FirstGlance initially colors only charged atoms. Where the charged sidechain atoms are missing, FirstGlance colors the remainder of the residue in a lighter color, with a checkbox option to color all atoms in charged residues, as does iCn3D.

Atoms Colored By Charge: 2ace With Incomplete Sidechains, Positive +, Negative –

FirstGlance: orientation for all snapshots.

FirstGlance default: charged atoms, light colors for incomplete sidechains.

FirstGlance charged residues checked.

iCn3D

(No way to color spacefilling atoms by charge was evident in ChimeraX or Molstar.)

Surface colored by electrostatic potential

When either ChimeraX or iCn3D color a surface by electrostatic potential, incomplete sidechains affect the map. Therefore, it is critical to use the AlphaFold-predicted structure that has neither missing residues nor incomplete sidechains.

Surface Colored By Electrostatic Potential vs. Incomplete Sidechains for 2ace, Positive +, Negative –

ChimeraX

iCn3D[12]

Structure Visualized

2ace with incomplete sidechains.

2ace with all sidechains complete.[13]

AlphaFold-predicted structure for P04058, with both ends trimmed.[14][15].

Orientations for all snapshots are the same as in the previous table.

How to avoid overlooking missing residues or incomplete sidechains

Regions with missing residues are clearly marked with "empty baskets" when a PDB ID model is displayed in FirstGlance in Jmol, as shown above. FirstGlance reports the total number missing, and the resulting number of missing charges. And it offers a detailed report (see snapshot below), which also lists all the missing residues (from PDB file REMARK 465).

FirstGlance also marks incomplete sidechains with S–, and (when you opt to Show More Details) provides a count of Incomplete Sidechains (see snapshot below). Clicking Find/List at that count enables you to list all the residues with incomplete sidechains.

Example in FirstGlance 4ifd: (When there are >100 incomplete sidechains, the S– labels are initially suppressed to reduce clutter.)

Notes & References

  1. 1.0 1.1 Percentages are based on searches for REMARK 465 and REMARK 470 at OCA, and searches for unobserved residues and unobserved atoms at RCSB.
  2. Weik M, Ravelli RB, Kryger G, McSweeney S, Raves ML, Harel M, Gros P, Silman I, Kroon J, Sussman JL. Specific chemical and structural damage to proteins produced by synchrotron radiation. Proc Natl Acad Sci U S A. 2000 Jan 18;97(2):623-8. PMID:10639129
  3. Dym O, Song W, Felder C, Roth E, Shnyrov V, Ashani Y, Xu Y, Joosten RP, Weiner L, Sussman JL, Silman I. The Impact of Crystallization Conditions on Structure-Based Drug Design: A Case Study on the Methylene Blue/Acetylcholinesterase Complex. Protein Sci. 2016 Mar 14. doi: 10.1002/pro.2923. PMID:26990888 doi:http://dx.doi.org/10.1002/pro.2923
  4. See Hiding empty baskets.
  5. The "S-" indicates that Lys491 has an incomplete sidechain, extending only to the gamma carbon.
  6. Vielberg MT, Bauer VC, Groll M. On the Trails of the Proteasome Fold: Structural and Functional Analysis of the Ancestral beta-Subunit Protein Anbu. J Mol Biol. 2018 Feb 2. pii: S0022-2836(18)30007-X. doi:, 10.1016/j.jmb.2018.01.004. PMID:29355501 doi:http://dx.doi.org/10.1016/j.jmb.2018.01.004
  7. Erdős G, Dosztányi Z. Analyzing Protein Disorder with IUPred2A. Curr Protoc Bioinformatics. 2020 Jun;70(1):e99. PMID:32237272 doi:10.1002/cpbi.99
  8. Wang K, Hu G, Basu S, Kurgan L. flDPnn2: Accurate and Fast Predictor of Intrinsic Disorder in Proteins. J Mol Biol. 2024 Sep 1;436(17):168605. PMID:39237195 doi:10.1016/j.jmb.2024.168605
  9. There is a spherical blue "empty basket" on the amino terminus because Met1 is missing. This spherical empty basket was removed in the image showing the putative catalytic triad.
  10. In 2ade, the temperature of the distal nitrogens in Arg321 is 79, while the average temperature is 31.
  11. See Labels on Atoms: S—, X, D, ?.
  12. Electrostatic potential maps were created in iCn3D using its Analysis, Delphi Potential, Surface with potential, with default settings except the max potential was set to 6 kT/e.
  13. To fill in incomplete sidechains in 2ace, the PDB file was loaded into Swiss-PDBViewer, then saved. It completes sidechains automatically. It does NOT fill in missing residues.
  14. The ends absent from the protein crystallized for 2ace were deleted using a Plain text editor. The signal sequence 1-21 was deleted from the N terminus. The C terminus starting with 559 was deleted; it is a tetramerization domain (see next reference).
  15. Dvir H, Harel M, Bon S, Liu WQ, Vidal M, Garbay C, Sussman JL, Massoulie J, Silman I. The synaptic acetylcholinesterase tetramer assembles around a polyproline II helix. EMBO J. 2004 Nov 10;23(22):4394-405. Epub 2004 Nov 4. PMID:15526038 doi:7600425

Proteopedia Page Contributors and Editors (what is this?)

Eric Martz

Personal tools