Opened 3 years ago

Last modified 17 months ago

#7140 assigned enhancement

Problems with the display of cyclic peptides

Reported by: chen_mangkun0101@… Owned by: Tom Goddard
Priority: normal Milestone:
Component: Depiction Version:
Keywords: Cc: Elaine Meng, risa.shibuya@…, Eric Pettersen, srettie@…
Blocked By: Blocking:
Notify when closed: Platform: all
Project: ChimeraX

Description (last modified by Eric Pettersen)

Hello Developer

Chimera 1.3 and 1.4 had some problems with the display of cyclic peptides. Taking cyclosporine (PDB: 1CYA) as an example, in Chimera 1.4, the ring cannot be closed in Cartoon mode. But it is closed on the PDB website. Other cyclic peptides have the same problem. Is this a setting issue or a software issue?

Version:1.4
OS: Windows 10
Protein: Cyclosporine (PDB: 1CYA) / Voclosporin (PDB: 3ODL)

[cid:image001.png@01D8855F.744BED70]

________________________________

Save paper. Protect the environment. Print only when necessary.


This email message is intended only for use by the intended recipient and may contain confidential and privileged information. Any unauthorized review, use, disclosure or distribution is prohibited. If you are not the intended recipient, then please contact the sender by reply email and destroy all copies of the original message.

Added by email2trac

Attachments (4)

image001.png (195.7 KB ) - added by chen_mangkun0101@… 3 years ago.
Added by email2trac
1cya_pseudobond.png (57.3 KB ) - added by goddard@… 3 years ago.
Added by email2trac
1cya2.pdb (15.5 KB ) - added by Elaine Meng 3 years ago.
frankenloop.png (156.3 KB ) - added by Elaine Meng 3 years ago.
image made from splicing non-broken segment of ribbon from circularly permuted structure

Download all attachments as: .zip

Change History (20)

by chen_mangkun0101@…, 3 years ago

Attachment: image001.png added

Added by email2trac

comment:1 by Eric Pettersen, 3 years ago

Cc: Elaine Meng added
Component: UnassignedDepiction
Owner: set to Tom Goddard
Platform: all
Project: ChimeraX
Status: newassigned

Chimera does show a closed loop whereas ChimeraX does not.

comment:2 by Tom Goddard, 3 years ago

Type: defectenhancement

The ChimeraX ribbon code does not know about cyclic peptides. I am not sure how difficult it would be to add the code needed to close the loop.

in reply to:  4 comment:3 by goddard@…, 3 years ago

An alternative to depict the closure is to show a pseudobond.

open 1cya
distance :11@C:1@N
label delete

Added by email2trac

Last edited 3 years ago by Tom Goddard (previous) (diff)

by goddard@…, 3 years ago

Attachment: 1cya_pseudobond.png added

Added by email2trac

in reply to:  6 ; comment:4 by Elaine Meng, 3 years ago

At least in the PDB file, all of the peptide bonds are represented by LINK records. There is probably equivalent information in the mmCIF file.  All of the residues are already correctly recognized by ChimeraX as polymer chain and amino acid (or else they wouldn't be shown as ribbon segments), so that leaves the problem of accounting for cyclic peptides when computing the ribbon path.  There are many cyclic peptide structures, and all of the peptide bonds are equally peptidic so it is misleading to have to show one of them as a pseudobond, implying some kind of discontinuity.

I recognize, however, that this may be rather complicated/difficult to fix and since I'm not the programmer, I would not be the one to make the ultimate decision of whether it is feasible in a reasonable amount of time.  Although Chimera already handles this case, the ChimeraX ribbon code is somewhat different than in Chimera.

SEQRES   1 A   11  DAL MLE MLE MVA BMT ABA SAR MLE VAL MLE ALA
HET    DAL  A   1      10
HET    MLE  A   2      22
HET    MLE  A   3      22
HET    MVA  A   4      19
HET    BMT  A   5      30
HET    ABA  A   6      13
HET    SAR  A   7      10
HET    MLE  A   8      22
HET    MLE  A  10      22
HETNAM     DAL D-ALANINE
HETNAM     MLE N-METHYLLEUCINE
HETNAM     MVA N-METHYLVALINE
HETNAM     BMT 4-METHYL-4-[(E)-2-BUTENYL]-4,N-METHYL-THREONINE
HETNAM     ABA ALPHA-AMINOBUTYRIC ACID
HETNAM     SAR SARCOSINE
FORMUL   1  DAL    C3 H7 N O2
FORMUL   1  MLE    4(C7 H15 N O2)
FORMUL   1  MVA    C6 H13 N O2
FORMUL   1  BMT    C10 H19 N O3
FORMUL   1  ABA    C4 H9 N O2
FORMUL   1  SAR    C3 H7 N O2
LINK         N   DAL A   1                 C   ALA A  11     1555   1555  1.31
LINK         C   DAL A   1                 N   MLE A   2     1555   1555  1.32
LINK         C   MLE A   2                 N   MLE A   3     1555   1555  1.32
LINK         C   MLE A   3                 N   MVA A   4     1555   1555  1.32
LINK         C   MVA A   4                 N   BMT A   5     1555   1555  1.34
LINK         C   BMT A   5                 N   ABA A   6     1555   1555  1.31
LINK         C   ABA A   6                 N   SAR A   7     1555   1555  1.32
LINK         C   SAR A   7                 N   MLE A   8     1555   1555  1.32
LINK         C   MLE A   8                 N   VAL A   9     1555   1555  1.31
LINK         C   VAL A   9                 N   MLE A  10     1555   1555  1.32
LINK         C   MLE A  10                 N   ALA A  11     1555   1555  1.31

comment:5 by Eric Pettersen, 3 years ago

If you switch to stick representation, you can see that all the peptide bonds are present.

comment:6 by Tom Goddard, 3 years ago

How do you find RCSB cyclic peptide structures? I just looked at 10 that had cyclic peptide in the title and none of them had cyclic peptide backbones. Most were made cyclic by a disulfide bond. Some were made cyclic by some other non-peptide bond.

I am curious because the Chimera implementation only works if all residues are observed. For instance 1cya has residues 1-11 and if you delete residue 6 suddenly two breaks in the ribbon occur. I can understand why, because the spline code to handle a broken loop would have to be different from a closed loop, so both were not handled in Chimera. So I am curious whether there are lots of cyclic peptides using all peptide bonds and where all residues are observed in the RCSB.

comment:7 by Eric Pettersen, 3 years ago

Well, 6AXI, 6AWM, 6AZF, 6AZG, and 6AWK as per: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6223261/ . I had to google "Homodetic cyclic peptide pdb" to find hits like these. Note that for 6AXI at least (didn't check the others) the cyclic peptide bond is missing because the needed LINK/CONECT record is missing, though that bond should clearly be there.

Last edited 3 years ago by Eric Pettersen (previous) (diff)

comment:8 by Eric Pettersen, 3 years ago

"cyclic peptide *bond*..."

comment:9 by Tom Goddard, 3 years ago

Yeah, none of those (6AXI, 6AWM, 6AZF, 6AZG, and 6AWK) have the closing peptide bond in PDB or mmCIF format.

in reply to:  12 comment:10 by Elaine Meng, 3 years ago

Although you are right that cyclization can often be through sidechains (not always peptide backbone), my understanding is backbone cyclization not only occurs in important natural products but is also a way of designing manmade therapeutic peptides to be less susceptible to proteolytic inactivation.  There may be a few manmade ones in the PDB, but many people may be modeling ones that aren't in the public database.  I don't know how to search for them in the PDB -- maybe it would work to find an example and search by its common name or by same uniprot sequence to entity, but I haven't tried it.

Cyclosporin is an important immunosuppressant drug (prevents organ rejection, among other things) and there are >347 structures from searching "cyclosporin" ... of which a goodly proportion do appear to contain a cyclic peptide. 1cya, 1cyb, 1csa, 1cwa, 4yv9 [...]

Some other structures are 
1tk2 (gramicidin S)
5eoc (cyclized version of viral epitope)
4z0f chain C
4oin chain I
5cs2 chain B
1bm2 chain L
1pqz chain M
3avb chains X,Y
1l5g chain C
3p8f, 1jbl
1sld chain P
1tmb chain T

These are just some examples from a review article I found.
Elaine

comment:11 by Tom Goddard, 3 years ago

Displaying closed loop ribbons for cyclic peptides would be cool. But it is also a lot of work and adds substantial complexity to the code for a rare case. I'd estimate that the code changes and future maintenance and possible debugging would be about 5 days work. If we wanted to handle the case where not all loop residues have coordinates (not handled by Chimera) I'd estimate maybe 10 days of work. Here is what would be needed.

1) A closed loop natural cubic spline code. Our current C++ cubic spline code can't handle this. SciPy has a version but it might be slow, maybe ok given it will rarely be used.
2) Detection of loop closing peptide bonds between first and last residue of a ribbon. Should be pretty easy.
3) Alternate code to convert spline to ribbon geometry. For a loop there is an extra segment instead of the non-loop which extrapolates the end segment.
4) Tether positions on ribbon need new special case code. The tethers on the loop closing segment need different handling than what is currently done.

If any residue of the loop is missing then it becomes a harder because the ChimeraX code handles the disconnected ribbon pieces as independent segments. It would need to be changed to possibly handle the closed loop as a single segment only with the residues not numbered in order. I think the Chimera code also handles the ribbons only as independent pieces and this is why Chimera does not show the loop closure if any residue in the loop is missing. The difficulty of handling this case is that the code is not organized to do it. It is not clear how much reorganization would be needed. We don't want to change the code organization to handle a rare case.

Overall the ribbon code is large and messy (~5000 lines?) and the original developer Conrad is retired. This makes it hard to modify. I don't think the effort to make cyclic peptides show as closed loops makes sense, too little value for too much (unpleasant) work.

comment:12 by Elaine Meng, 3 years ago

Another fairly poor hack -- not convenient and still doesn't make a great display -- is to edit a copy of the PDB file to put residues 6-11 before residues 1-5 creating a structure that has the break in the opposite part of the loop (1cya2.pdb, attached). Then open the original and the new one and show only the ribbon halves without the break. Not great since the "seams" don't exactly line up (as you can see in the image, also attached).

open 1cya
open 1cya2.pdb
~ribb #1:1-3,9-11 #2:4-8

Also in case we wish to refer to it later, the review article that included several of the examples I listed before is
https://pubs.acs.org/doi/10.1021/acs.chemrev.8b00807
Crystal Structures of Protein-Bound Cyclic Peptides
Alpeshkumar K. Malde, Timothy A. Hill, Abishek Iyer, and David P. Fairlie*
Chem. Rev. 2019, 119, 17, 9861–9914

by Elaine Meng, 3 years ago

Attachment: 1cya2.pdb added

by Elaine Meng, 3 years ago

Attachment: frankenloop.png added

image made from splicing non-broken segment of ribbon from circularly permuted structure

comment:13 by Eric Pettersen, 2 years ago

Cc: risa.shibuya@… added
Description: modified (diff)

comment:14 by Eric Pettersen, 2 years ago

Cc: Eric Pettersen added

comment:15 by Eric Pettersen, 17 months ago

Cc: srettie@… added

comment:16 by Tom Goddard, 17 months ago

A trick to make a cyclic peptide ribbon appear closed is described in ticket #15360. It takes 6 commands to double the length of the cyclic peptide then hide all but the middle part.

Note: See TracTickets for help on using tickets.