Error during protein/DNA docking with mutated nucleotide

HADDOCK
1

To the HADDOCK support team,

I would like to express my compliments to your team for your work.

I am attempting to dock a modified DNA strand containing a deoxyuracil in place of a standard thymine using HADDOCK 2.4. However, in the final output, this nucleotide appears detached from the rest of the strand and lacks its phosphate groups. I named the nucleotide as U, so it changed the structure to match a standard uracil.

I have performed several runs with different parameters. My best configuration involved keeping the terminal phosphates on and defining unambiguous restraints to prevent the stretching of the uracil’s phosphodiester bonds. While this kept the mutated uracil closer to the strand, the final model still showed stretched bonds.

The restraint file I used was:

text

! Unambiguous distance restraints for DNA backbone - Covalent phosphate bonds
! Based on experimental distance of ~1.6A

! Restraint 1: Phosphate of residue 15 to O3’ of residue 14 (covalent bond)
assign ( resid 15 and name P   and segid B ) (
resid 14 and name O3’ and segid B ) 1.4 1.6 1.8

! Restraint 2: Phosphate of residue 16 to O3’ of residue 15 (covalent bond)
assign ( resid 16 and name P   and segid B ) (
resid 15 and name O3’ and segid B ) 1.4 1.6 1.8

Despite this, the final bond distance was approximately 3.2 Å.

Following the tutorial, I also selected the strand as a semi-flexible segment, but I did not modify any configurations in the Advanced Sampling Parameters tab. The DNA strand kept its helicity as expected.

I am aware that HADDOCK 2.4 does not officially support mutated nucleotides, but I have read several literature reports of successful docking experiments with modified strands using the online server. Is there a recommended way to proceed with this type of experiment in HADDOCK 2.4?

From my research, the issue seems to be related to the lack of specific parameters for non-standard nucleotides like uracil or deoxyuracil in DNA strands. Is it possible to submit custom topology files to the web server to address this? Or, to edit the files that are already there to properly build this system?

Finally, it is important to note that we plan to use these docked structures for subsequent Molecular Dynamics simulations. I am uncertain if a 3.2 Å bond distance is too large for energy minimization to correct effectively.

I appreciate your patience and any guidance you can provide.

example of best run: HADDOCK results page

2

Hi there

We don’t have support for modified bases…

If you define those at HETATM parameters/topologies are automatically defined, but those won’t be connected to the remaining of the strand.
Because of van der Waals interaction, even defining distance restraints as you did will not be able to maintain the covalent bond distance.

In the context of an EU-India project (ganana.eu) we are working on adding support for modified bases, but this will take some time.

Note also that you definition of distances is rather loose. The three number are: target distance, lower limit correction, upper limit correction.
So better to define those as:

! Unambiguous distance restraints for DNA backbone - Covalent phosphate bonds
! Based on experimental distance of ~1.6A ! Restraint 1: Phosphate of residue 15 to O3’ of residue 14 (covalent bond) assign ( resid 15 and name P and segid B ) ( resid 14 and name O3’ and segid B ) 1.6 0.2 0.2 ! Restraint 2: Phosphate of residue 16 to O3’ of residue 15 (covalent bond) assign ( resid 16 and name P and segid B ) ( resid 15 and name O3’ and segid B ) 1.6 0.2 0.2

And you could add a few more (other measurements, e.g. to have the angle also defined).

3

We appreciate the answer

I submitted a test job to check if those restraints will generate a better result

We’re aware of the long queue situation, so I reduced the amount of rigid body structures generated to just 10, and only 2 for subsequent steps. I just want to know how will the software handle the nucleotide

Now, I am wondering. Even if we could force the nucleotide to stay in a good position, since we are neglecting a clear limitation of the tool, would this be honest? Would it generate a publishable result?

So, I am working with 2 more researchers who have some experience with molecular modelling. We are quite invested on this subject and are considering parametrizing the mutated nucleotide to run the job locally.

We believe we can use most of thimine parameters, since it is a very similar structure.

However, what is the best approach to fill the gaps of the missing info?

Would you give us some guidance on the matter?

4

Now, I am wondering. Even if we could force the nucleotide to stay in a good position, since we are neglecting a clear limitation of the tool, would this be honest? Would it generate a publishable result?

You might be able to refine your model with some other software to restore the proper covalent bonds.

So, I am working with 2 more researchers who have some experience with molecular modelling. We are quite invested on this subject and are considering parametrizing the mutated nucleotide to run the job locally.

We believe we can use most of thimine parameters, since it is a very similar structure.

You can give it a try indeed. The files to modify will be the topology and parameter files for nucleic acids.

In haddock3 these are found in haddock3/src/haddock/cns/toppar/

topology: dna-rna-allatom-hj-opls-1.3.top
parameter: dna-rna-allatom-hj-opls-1.3.param

1 Like
5

Thank you for the prompt response and for your suggestion to refine the model using other software; we plan to try that approach shortly.

Following up on that, another idea occurred to me, though I’m uncertain about its scientific merit. Given that the RMSD between the original DNA strand and the docked DNA is less than 1 Ångstrom, would it be valid to superimpose the original DNA onto the docked structure, remove the DNA strand from the HADDOCK output, and then proceed with molecular dynamics?

image
image700×779 107 KB

This is the superimposition between both strands. In yellow, the one from haddock. In green, the input strand. The mutated nucleotides are almost on the same position relative to the protein.

6

Assuming that the binding mode of the modified DNA remains the same this seems like a possible strategy.

7

Would, for example, listing the interactions between the docked and superimposed strand with the protein through PLIP and the RMSD be enough arguments to stand our experiment to a reviser? Provided the interactions are kept the same between both structures.

Thank you so much for your help so far!

8

This will be of course no proof that the complex is formed… only that it is plausible.

1 Like