Cation-p and Salt Bridge Visualization in Protein Explorer
Contents
Cation-p Interactions       Snapshots
Salt Bridges

Background.

Gallivan & Dougherty (1999) reported results from a quantitative survey of cation-p (cation-pi) interactions in high-resolution structures in the Protein Data Bank. Using an energy-based criterion for identifying significant sidechain interactions, they studied 593 sequence dissimilar proteins taken from the "PDB Select" list of Hobohm and Sander. They found an average of one such interaction per 77 residues, with no significant effect of chain length, or multiple-chain vs. single chain structures. Arg was more likely than Lys to participate in a cation-pi interaction, and the liklihood of aromatic sidechain participation was Trp > Tyr > Phe. Over one quarter of all Trp's were involved in cation-pi interactions, with the cation typically positioned over the 6-atom ring of Trp. Because of the frequencies of amino acids in the database, Arg participates in nearly twice as many cation-pi interactions as does Lys, and the numbers of cation-pi interactions involving Trp, Tyr and Phe are roughly similar. Their study did not include His because, depending on its protonation state, it could participate either as a cation or as a pi-system. Lys and Arg were assumed always to be protonated and hence cationic.

Gallivan and Dougherty conclude "When a cationic sidechain is near an aromatic sidechain, the geometry is biased toward one that would experience a favorable cation-pi interaction", and "cation-pi interactions should be considered alongside the more conventional hydrogen bonds, salt bridges, and hydrophobic effects in any analysis of protein structure". They provide a Gallery of energetically significant cation-pi interactions and a server that lists text results from their program CaPTURE.

Zacharias and Dougherty (2002) reviewed cation-pi interactions in the binding of ligands to proteins. Cation-pi interactions are usually energetically important when the ligand has either positive charge or an aromatic ring, and are involved in control of ion channels, G-protein-coupled receptors, transporters, and enzymatic catalysis. An example is 1L8B, a portion of a eukaryotic translation initiation factor that recognizes N7-methylated guanosine. The ligand's heterocyclic base (cationic) is sandwiched between Trp56 and Trp102. The easiest way to see this in PE is in QuickViews: SELECT Ligand, DISPLAY Contacts. (Alternatively, you can designate the cationic ligand nitrogen in the Cation-Pi form in Advanced Explorer. Note that QuickViews DISPLAY Cation-Pi does not display cation-pi interactions involving ligands.)

An introductory gallery of examples of cation-pi interactions and other noncovalent interactions is provided by Ricky Cox.

Contents   Close

Pros & Cons of Protein Explorer's Cation-Pi Display

As detailed below, Protein Explorer's present mechanism to display cation-pi interactions tends to show more cation-pi pairs than the number deemed energetically significant by Gallivan & Dougherty's CaPTURE (up to twice as many pairs). It also sometimes misses one or more pairs deemed energetically significant by CaPTURE. Therefore, Protein Explorer's cation-pi display is useful for a first look, but important cases should then be compared to the output of CaPTURE. (It is planned eventually to enable CaPTURE to display its output directly in Protein Explorer.)

Protein Explorer allows you to include ligands in the cation-pi display, a capability that is not implemented in the CaPTURE web server.

In order to include ligand cations or aromatic rings, you must specify the appropriate atoms in the cation-pi form slots. Several examples of how to do this are explained in the tutorials below.

Contents   Close

Interesting Examples and Gallery

Examples given by Gallavan and Dougherty include:

Interesting examples noted using Protein Explorer:

Contents   Close

Tutorials: How to See Cation-Pi Interactions in Protein Explorer

    Contents   Close

  1. Tutorial: Interchain and intrachain amino acid pairs in 1b07.
    1. Click here to start a new PE session with 1B07, a small Src-Homology 2 (SH2) domain bound to a regulatory peptide.
    2. Check "Cation-Pi" but not "Salt Bridges".
    3. Click the [Show] button. After a short pause for processing, three cation-pi pairs will be displayed. The default color scheme is to color by chain. This makes it easy to see that two of the pairs are within the longer chain, while one is between the peptide and the longer chain. Note that you can distinguish Arg sidechains from Lys because the cationic atoms of Arg are shown larger.
    4. Once you have displayed interacting pairs, you can change the color scheme of these pairs immediately. Under Color interacting pairs, change the radio buttons from "as below" to "CPK" (color by element).
    5. Try another combination of the Display Options if you wish. Click the [Show] button again. (Click OK on the message that pops up.)
        Advanced feature: The message will suggest that you show the cation-pi pairs for the originally selected 615 atoms -- the entire molecule. When the cation-pi pairs are shown, Protein Explorer leaves them selected. This makes it easy to change their rendering or coloring using either QuickViews or typed commands. However, if you re-run the cation-pi routine, you usually want to show the pairs for the entire molecule.
    6. Comparison of the 3 pairs shown by Protein Explorer with the results of CaPTURE reveals that one of these three is deemed as energetically insignificant.
    7. Reduce the maximum distance in Angstroms used in the cation-pi routine from 6.0 to 5.5, and click [Show] again. Only two pairs meet the new distance cutoff, and these happen to be the two deemed energetically significant by CaPTURE. However, it is unusual for Protein Explorer's simple distance criterion to agree so well with the results of CaPTURE.
    8. Increasing the cutoff distance to 6.5 Angstroms reveals a 4th cation-pi pair (also deemed energetically insignificant by CaPTURE).
    9. Way down at the bottom of the form, under the Advanced section, check the Autostep feature. In this mode, the detection distance will be incremented by 0.5 Angstroms (or whatever you specify in the Autostep form slot) on each click of the [Show] button. Click the [Show] button several times, until you get to 8.5 Angstroms, noting the newly found moieties in each cycle. This makes it easy to see that there are no more candidates for energetically significant cation-pi pairs.

    Contents   Close

  2. Tutorial: Cationic substrate interaction with enzyme in 2ACE.
    1. For this portion of the tutorial, you will need to load 2ace.pdb (acetylcholinesterase with acetylcholine placed into the catalytic gorge by modeling). If you are connected to the Internet, click here to start a new PE session with 2ACE. (If not, you will need to get a copy of 2ace.pdb on your hard disk and load it with the Browse button on the Load Molecule page.)
    2. Check "Cation-Pi" but not "Salt Bridges", and make sure "Autostep" (near the bottom of the page) is not checked.
    3. Click the link Restore Default Parameters in the Cation-p section of the form, and accept the confirmation by clicking OK.
    4. Click the [Show] button. Nine pairs are shown. In this instance, Protein Explorer's distance-based criteria do particularly badly: CaPTURE reports 12 energetically significant cation-pi interactions; Protein Explorer misses 4 of these, and shows one deemed energetically insignificant by CaPTURE. (By increasing the distance criterion to 6.5 Angstroms, three of the missing pairs are shown, at the expense of four new but energetically insignificant pairs also being shown.)
    5. Replace the contents of the cations slot with "ach.n1" (the cationic nitrogen in the substrate). Click the [Show] button. Trp84 and Phe380 are shown as candidates for cation-pi interactions with the substrate. Increasing the cutoff distance to 8.0 Angstroms shows Phe331, but it is oriented with the edge of the ring facing the cation.

    Contents   Close

  3. Tutorial: Cationic ligand interaction with antibody in PROBLEMATIC PDB file 2mcp.
    1. 2mcp is the Fab fragment of an antibody bound to the hapten phosphocholine. This case is problematic because of a bug in Chime 2 that prevents it from correctly handling residues with names shorter than 3 characters In this instance, the phosphocholine is named "PC" in the PDB file. (The bug is that after selecting "PC", Chime fails to report its residue name with the "show residue" command.) Until a new version of Chime is released in which this bug is fixed, the only solution to this problem is to save the PDB file to your hard disk and edit it with a text editor, being careful to save it as plain/ASCII/DOS text. You must give any residues of interest three-character names. In this case, renaming PC to PC1 works fine. Be careful to place the 3-character name in columns 18-20, so it lines up with other 3-character residue names in the PDB file. For your convenience, the PDB file so-modified has been made placed on our server, and you can load it into Protein Explorer by clicking here: http://www.umass.edu/microbio/chime/beta/pdb/2mcp-pc1.pdb
    2. Check "Cation-Pi" but not "Salt Bridges", and make sure "Autostep" (near the bottom of the page) is not checked.
    3. Click the link Restore Default Parameters in the Cation-p section of the form, and accept the confirmation by clicking OK.
    4. Change the contents of the cations slot to "[pc1].n1". (Chime requires square brackets around residue names that contain numerals.)
    5. Click the [Show] button. Trp107 and Tyr100 are shown in favorable positions for cation-pi interactions with the hapten.

    Contents   Close

  4. Tutorial: Aromatic ligand complex with enzyme in 3pcb.
    1. File 3pcb contains the enzyme dioxygenase complexed with 3-hydroxybenzoate. Unfortunately it contains six copies of a heterodimer. This produces images of unnecessary complexity. For your convenience, a PDB file containing a single heterodimer (chains A, M) has been made available, and can be loaded (if you are connected to the Internet) by clicking here: http://www.umass.edu/microbio/chime/beta/pdb/3pcb-am.pdb
    2. Check "Cation-Pi" but not "Salt Bridges", and make sure "Autostep" (near the bottom of the page) is not checked.
    3. Click the link Restore Default Parameters in the Cation-p section of the form, and accept the confirmation by clicking OK.
    4. Protein Explorer needs to be told the names of three alternate carbons in each type of aromatic ring for which it is to show proximal cations. Clicking on the carbons in the 3HB ligand shows their names in the message box. Change the contents of the 3 slots for aromatic ring carbons to these:
      • [3hb].c1
      • [3hb].c3
      • [3hb].c5
      (Chime requires square brackets around residue names that contain numerals. You could equally well use carbons 2, 4, and 6.)
    5. Click the [Show] button. Arg333 and Arg457 are shown in favorable positions for cation-pi interactions with the two copies of hydroxybenzoate.
Contents   Close

Protein Explorer's Distance-Based Search Mechanism for Cation-Pi Interactions

Chime has built-in at its disposal only distance, and the ability to identify specific atoms within specific residues, to use in searching for cation-pi interactions. Gallivan and Dougherty considered potential interactions out to a distance of 10 Angstroms, yet found that 99% of significant cation-pi interactions occurred at distances not exceeding 6 Angstroms. In distances, both NZ and CE of Lys, and both CZ and CD of Arg were considered.

The method built into Protein Explorer locates all cationic atoms (by default, lys.nz, lys.ce, arg.cz, arg.cd) within (by default) 6.0 Angstroms of three alternate carbons in a six-carbon ring of Trp, Tyr or Phe. Advanced Explorer allows customization of the list of cationic atoms considered, the list of aromatic atoms considered, or the distance. This enables, for example, a bound substrate to be included in the display. (The entire Chime script employed can be displayed.)

There are two reasons for overestimation by the present distance-based method employed in Protein Explorer. In some cases, a cation is within the requisite 6 Angstroms of an aromatic sidechain, but the interaction would in fact be energetically insignificant. In other cases, the requirement for three alternate carbon atoms in aromatic rings are met by carbons from different residues. The latter type of incorrect results are usually obvious because a ring will be shown with no proximal cation, or vice versa. (The effort that would be needed to correct this flaw in the method seemed not worthwhile, considering that interactions would still be overestimated due to the lack of an energetics calculation). Detailed comparisons of results are given below.

Contents   Close

Comparison of Results from Protein Explorer vs. CaPTURE.

A goal for the future is to connect CaPTURE with Protein Explorer so that CaPTURE's results can be visualized directly.

For the present, the simplistic distance-based method used in Protein Explorer (when used with default parameters) in a few cases agrees perfectly with the results of CaPTURE (1bl8, 2wea), but in most cases shows more cation-pi pairs than the number of energetically significant cation-pi interactions reported by CaPTURE. Occasionally the number of excess pairs can exceed the total number of pairs reported by CaPTURE (5 added to 4 in 1bfg). Detailed comparisons for a few cases are listed in a table below.

Occasionally, Protein Explorer fails to show an interaction reported by CaPTURE (1 out of 10 in 1axi; 1 out of 14 in 1gai; 1 out of 6 in 2vab). CaPTURE usually assigns barely significant energies to the pairs not shown by Protein Explorer. They can be shown by Protein Explorer if the default distance of 6.0 Angstroms is increased in Advanced Explorer (at the expense of showing more energetically insignificant pairs).

Reducing the distance (e.g. to 5.5 Angstroms) in some cases bring Protein Explorer's results into agreement with those of CaPTURE (1v39), but in others increases the number of energetically significant pairs missed, while still finding some energetically insignificant pairs (1bfg). Therefore, the default distance was left at 6.0 Angstroms.

This table displays best in a window at least 1024 pixels wide.
PDB ID
Resol.		 CaPTURE's Results* PE's Results* Total
Discrepancies**		 Comments
1axi
2.1A		 AA  3 (1 1 0 0 1 0)
BB  6 (2 1 2 0 0 1)
AB  1 (0 0 1 0 0 0)
   10 (3 2 3 0 1 1)		 AA  5 (2 2 0 0 1 0)
BB 11 (2 2 2 1 2 2)
AB  1 (0 0 0 0 0 0)
   16 (4 4 2 1 3 2)		 AA  2 (-0, +2)
BB  5 (-0, +5)
AB  1 (-1, +0)
    8 (-1, +7)		 The one miss is the least energetically significant pair (-2.72). From the K179 end: CPCPCPc (only one excess hit, lower case).
1bfg
1.6A		  4 (2 1 0 0 1 0)  9 (5 1 0 0 3 0) 5 (-0, +5) Least energetically significant pair -2.16.
1bl8
3.2A		  4 (0 0 4 0 0 0)  4 (0 0 4 0 0 0) 0 (-0, +0) Least energetically significant pair -7.80.
1gai
1.7A		 14 (1 3 5 0 3 2) 14 (1 2 6 0 3 2) 2 (-1, +1) The one miss is the least energetically significant pair (-2.10)
1v39
1.8A		  4 (2 2 0 0 0 0)  7 (2 3 0 2 0 0) 3 (-0, +3) Least energetically significant pair -4.34.
2ace
2.5A		  12 (4 0 2 2 1 3)  9 (4 0 0 3 1 1) 5 (-4, +1) The one miss is the least energetically significant pair (-3.36)
2vab
2.5A		  5 (0 2 3 1 0 0)  8 (1 3 3 1 0 0) 4 (-1, +3) The one miss is the least energetically significant pair (-3.36)
2wea
1.25A		  0 (0 0 0 0 0 0)  0 (0 0 0 0 0 0) 0 (-0, +0)  
* Total pairs (counts of pairs RF RY RW KF KY KW).
** (-) missed pairs; (+) excess pairs.

Contents   Close

Inaccurate results: Rings Displayed without Cations and Vice Versa

If a ring is highlighted with no cation nearby, or vice versa, this is consistent with the method employed. The explanation is as follows.

Inaccurate results: Misoriented rings.

Another inappropriate result occurs in 1fod, where Arg114 is shown proximal to Phe39, but the cation is in the plane of the ring instead of sitting on one face. Phe39 is shown because it is close to another cation, Arg179; it is not close enough to Arg114. Arg114 is shown because three alternate ring carbons are close enough, but not in the same residue. Some are in Phe39, and others are in Tyr11 (not shown).

Contents   Close

Acknowledgements

Thanks to Joel Sussman (Weizmann Institute of Science, Rehovot, Israel) for reminding me of the importance of cation-pi interactions, providing me with the relevant literature, and general support and encouragement. Thanks also to Ricky Cox (Georgia Southern U, Statesboro GA) for calling cation-pi interactions to my attention.

Contents   Close

Salt Bridge Detection

The default distance used for detection of salt bridges is 4.0 Angstroms (suggested by Ricky Cox, Murray State U, KY). This can be changed in the form slot.

At the bottom of the form is a checkbox for Autostep. This causes the detection distance to be incremented by 0.5 Angstroms on each click of the [Show] button. (The increment size can also be changed in a form slot.) Ions newly detected in each cycle are highlighted with a dot surface.

If you believe the default distance for salt bridge detection should be different than 4.0 Angstroms, or if you know of published summaries of the properties of protein salt bridges, please enlighten me!

Contents   Close

Tutorial: Detecting Ligand Salt Bridges to Protein

  1. You will need to load 1cbr.pdb (cytoplasmic retinoic acid-binding protein complexed to retinoic acid). If you are connected to the Internet, click here to start a new PE session with 1CBR. (If not, you will need to get a copy of 1cbr.pdb on your hard disk and load it with the Browse button on the Load Molecule page.)
  2. Check "Salt Bridges" but not "Cation-Pi", and make sure "Autostep" (near the bottom of the page) is not checked.
  3. Click the link Restore Default Parameters in the Cation-p section of the form, and accept the confirmation by clicking OK.
  4. Click on the carboxyl oxygens in the retinoic acid to show their atom and residue names in the message box.
  5. In the Salt Bridges form, replace the anion atoms in the form slot with "rea.o?" (letter o for oxygen, not numeral; the question mark matches either "rea.o1" or "rea.o2").
  6. Click the [Show] button. Arg 131 is identified as within salt bridge distance of the ligand anion.
  7. Feel free to try the "Autostep" mode to look for more distant cations, or restore the default anions and look for within-protein salt bridges.

Contents   Close