search

jensengroup / propka

PROPKA predicts the pKa values of ionizable groups in proteins and protein-ligand complexes based in the 3D structure.
http://propka.org
GNU Lesser General Public License v2.1
256 stars 58 forks source link

Handling of metal ions #142

Closed guidsari closed 2 years ago

guidsari commented 2 years ago

Good morning, I'm trying to use PROPKA3.4 to obtain an estimation of the pKa of an aspartate residue inside a mutant of a small bacterial protein. My mutant protein includes one Zn2+ ion with a complexed water molecule, not too far from the aspartate. 1) Is it possible for the PROPKA3 software to handle the charged zinc without ignoring it, so that I can estimate the pKa of the water molecule bound to the zinc? 2) In the case this were not possible, can the software handle the Zn2+ charge, since it is relatively close to the aspartate I want to investigate and it could electrostatically influence its dissociation properties? 3) I suppose the software will be able to handle Na+ and Cl- solute ions, am I correct?

Thanks for your kind attention. Yours sincerely, Guido Frisari PhD Student EPFL - 1015 Lausanne (CH) SB - ISIC - LCBC (Computational Chemistry and Biochemistry Laboratory - Ursula Roethlisberger's group) guido.frisari@epfl.ch

sobolevnrm commented 2 years ago

I don't think PROPKA is the best choice for this use case. I'd recommend MCCE from the Gunner lab instead.

guidsari commented 2 years ago

In principle, I can neglect the Na+ and Cl- solute ions. So the program would not be able to compute the pKa of an aspartate which is close to a Zn2+?

sobolevnrm commented 2 years ago

Have you tried this? What error did you receive?

guidsari commented 2 years ago

It complains about histidines which had abnormal protonation states, like in the following case: Expected 2 interaction atoms for bases, found: 261- ND1 16-HIS (X) [ 34.138 24.262 35.169] N 265- NE2 16-HIS (X) [ 33.107 26.174 34.604] N

The histidines have a weird protonation state because they interact with the Zn2+. I'm using the -k option to keep the original protonation state of each residue. Anyway, I'm not interested in the histidine residues.

The program seems to be producing a normal output. Lines are similar to the following ones:

                        DESOLVATION  EFFECTS       SIDECHAIN          BACKBONE        COULOMBIC

RESIDUE pKa BURIED REGULAR RE HYDROGEN BOND HYDROGEN BOND INTERACTION

ASP 22 X 5.50 91 % 2.17 536 0.00 0 0.00 XXX 0 X -0.23 ALA 23 X -0.02 N+ 1 X ASP 22 X 0.00 XXX 0 X 0.00 XXX 0 X -0.17 LYS 28 X ASP 22 X 0.00 XXX 0 X 0.00 XXX 0 X -0.05 LYS 50 X

ASP 33 X 3.16 100 % 2.86 596 0.45 0 -0.68 LYS 37 X 0.00 XXX 0 X -0.60 HIS 5 X ASP 33 X 0.00 XXX 0 X 0.00 XXX 0 X -0.64 HIS 16 X ASP 33 X 0.00 XXX 0 X 0.00 XXX 0 X -2.03 LYS 37 X (many other lines)

(another section with pKa and model pKa for each protonable residue)

The output file ends this way: Free energy of folding (kcal/mol) as a function of pH (using neutral reference) 0.00 39.40 1.00 39.30 2.00 38.63 3.00 36.95 4.00 36.80 5.00 40.50 6.00 43.86 7.00 44.82 8.00 44.23 9.00 41.99 10.00 39.03 11.00 39.29 12.00 42.62 13.00 46.15 14.00 48.75

The pH of optimum stability is 3.6 for which the free energy is 36.4 kcal/mol at 298K Could not determine pH values where the free energy is within 80 % of minimum Could not determine the pH-range where the free energy is negative

Protein charge of folded and unfolded state as a function of pH pH unfolded folded 0.00 11.00 10.98 1.00 10.98 10.81 2.00 10.83 9.89 3.00 9.64 8.45 4.00 5.25 6.69 5.00 0.48 3.73 6.00 -1.54 -0.06 7.00 -3.35 -3.27 8.00 -4.45 -5.43 9.00 -5.30 -7.56 10.00 -7.67 -9.14 11.00 -12.14 -10.36 12.00 -13.77 -11.03 13.00 -13.98 -11.66 14.00 -14.00 -12.53 The pI is 6.00 (folded) and 5.20 (unfolded)

Now, as you can see, I get a pKa value for ASP33, which is relatively close to a Zn (the aspartate does NOT coordinate the zinc, but it will feel its electromagnetic field; the name of the Zn residue in the PDB is ZN6). What I don't know is if I can trust the algorithm to work correctly with a Zn2+ ion just a few angstrom away from the aspartate side chain. Moreover, I don't know whether Zn2+ is parametrized in the program.

I can easily neglect Na+ ions because they are far from the protein. I'm only interested in that aspartate 33. I would like to know if I can trust the predicted value. Thank you for your attention.

sobolevnrm commented 2 years ago

I'm sorry but I don't know how well PROPKA performs with divalent atoms near titratable groups.