This document describes a computational method for redesigning the HIV-1 protease enzyme to cleave a desired target sequence. The method uses a docking protocol to predict substrate binding and generate protease mutants with lower binding energies. The top mutant structures are refined with quantum chemistry calculations to determine qualitative binding energies. The goal is to iteratively mutate the best cleaving sequence into the target sequence one amino acid at a time to obtain a protease variant that cleaves the target.

1. A Computational Enzyme Activity Design of HIV-1 Protease Luca De Vico WATOC 2011 Santiago de Compostela, Spain

2. What the project is about Modify a protease in order to cleave another desired sequence New enzymatic activity towards a specific sequence Rational redesign of the enzyme Valuable tool for biological applications Extended to any sequence Experimental and computational collaboration

3. Enzyme redesign: what is needed - A target sequence to cleave - A template protease: HIV-1 protease ✂ Pro His Leu Ser Phe Met Ala Ile Pro Pro

5. Enzyme redesign: what is needed X-ray structure of wild type protease docked with one of its natural substrates: 1KJ7 The protocol has to produce: - reasonable structures with many different substrate peptides bound - generate mutants with lower binding energy

6. Our PyRosetta optimization protocol Substrate peptide sequence Starting structure x6 f, ψ, χ perturbations Energy minimization (David-Fletcher-Powell) x8 x4 Out of 500 decoy structures the lower in energy is chosen. x6 MC criterion Side-chain packing MC criterion ca. 24 hours on 5 cpus Output decoy

7. The perturbation and minimization are on: Backbone Side chains Substrate peptide Protease cavity Residues inside 5 Å radius from the peptide Any residue reported as active in Chaudhury et al. Their ±1 sequence neighbors Both chains have the same residues as “movable”

9. FMO inputs from FRAGIT

10. ca. 8 hours on 64 cpus per structure

11. Used to compute qualitative binding energiesEbinding = Ecomplex – Eapo – Epeptide

12. Qualitative binding energies FMO MP2/PCM/6-31G* (kcal/mol), Wild Type protease Average binding energy Natural Target Cleavable Peptides -60.9 Min value -79.2 Max value -41.1 Standard deviation 11.0 Average binding energy Known NON-CleavablePeptides -13.7 Min value -68.3 Max value 61.0 Standard deviation 28.1 Target peptide sequence binding energy: -24.1

13. Enzyme redesign: mutation protocol Among the cleavable peptides sequences, the closest to the target is mutated into the target sequence, one amino acid at the time Cleavable start ✂ ✂ Val Ser Phe Asn Phe Met Ala Ile Leu Thr Pro His Leu Ser Phe Pro Gln Ile Pro Pro Target

15. Only the protease 6 specificity determining residues are allowed to mutate (ca. 2000 total rotamers per perturbation)

18. Evaluation of energy barrier for qualitative differences between wild type and mutated protease

19. Comparison with on-going experimental data

20. Dep. Chemistry: Jan Jensen, Casper Steinmann

21. Dep. Biology: JakobWinther, Martin Willemöes, Helen Webb

22. Funding provided by the Danish Research Council for Technology and Production Sciences (FTP)

24. Extra

25. Cleavable sequences WT HIV-1 protease recognized sequences

26. Example of optimization convergence MA-CA Each optimization cycle requires ca. 24 hours on 5 cpus

27. FMO MP2/PCM/6-31G* binding energies of cleavable peptides Kcal/mol Differences between the binding energies of WT HIV-1 protease and its natural substrates are expected and will be experimentally checked

28. Enzyme redesign Among the experimentally verified cleavable peptides, the closest to the target sequence is chosen Target sequence TF-PR cleavage sequence, candidate starting sequence Specificity most involved peptide residues The candidate substrate sequence is mutated into the target sequence one amino acid at the time

29. Enzyme redesign The candidate substrate sequence is mutated into the target sequence one amino acid at the time

30. Specificity determining residues P3’ V82 G48’ L76 P4’ high P1’ I47’ medium P2 P2’ D30’ low P1 P3 P4 I84’