Design of 2-cyclopentenone derivatives with enhanced NF-κB: DNA binding inhibitory properties

Design of 2-cyclopentenone derivatives with enhanced NF-κB: DNA binding inhibitory properties

Please download to get full document.

View again

of 10
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.


Publish on:

Views: 0 | Pages: 10

Extension: PDF | Download: 0

  Design of 2-cyclopentenone derivatives with enhanced NF- k B:DNA binding inhibitory properties Pedro A. Fernandes a , Ana I.S. Cruz b , Ana R.R. Maia b , Andre´ A.S. Almeida b ,Andre´ M.N. da Silva b , Bruno F.B. Silva b , Carla M.S. Ribeiro b , Ce´sar F.B. Ribeiro b ,Eva M.S. Cunha b , Filipe R.N.C. Maia b , Joa˜o A.C. Tedim b , Joaquim A.A.D. Ferreira b ,Lı´gia C. Gomes b , Liliana R.C. Matos b , Luı´s M.N.F.S. Cruz b , Manuel A.B.P. Pinto b ,Marisa A.R. da Encarnac¸a˜o b , Pedro F.R.D. Teixeira b , Raquel S.G.R. Seixas b ,Rui J.A.L. da Quinta b , Sandro S. Gomes b , So´nia G. Patrı´cio b , Susana D.S. Martins b ,Tiago F. Barros b , Tiago S.J.T. Sela˜o b , Vineet Pande a , Maria J. Ramos a, * a  REQUIMTE, Departamento de Quı´ mica, Faculdade de Cieˆ ncias, Universidade do Porto, Rua do Campo Alegre, 687, 4169-007 Porto, Portugal b  Departamento de Quı´ mica, Faculdade de Cieˆ ncias, Universidade do Porto, Rua do Campo Alegre, 687, 4169-007 Porto, Portugal Received 21 June 2004; accepted 22 June 2004Available online 1 October 2004 Abstract In this work we derived a series of new inhibitors for the association between nuclear factor kappa B (NF- k B) and the corresponding  k Bsite in DNA. They were derived through optimization of the lead compound 2-cyclopentenone (CP), which corresponds tothe reactive unit of natural product 15-Deoxy- D 12,14 -prostaglandin J 2  (PGJ2). Both CP and PGJ2 possess demonstrated inhibitory efficiency for this and otherbiological important systems.We began by studying the docking of CP to NF- k B. Subsequently, a set of rational strategies were derived to insert substituents into CPwhich increase its association to NF- k B, in terms of both the affinity and the specificity. Molecular mechanics calculations have beenperformed to decide on the suitability of the substitutions, and to evaluate the energies of association with NF- k B. One of the importantchemical properties of CP is that it is a weak electrophile, hence it selects attacking nucleophilic sites in proteins, rather than the nucleicacids. To assure that the designed compounds were notsubstantially more reactive than CP we performed high level density functional theorycalculations. Important conclusions have been obtained concerning the optimization of this inhibitor; namely, a set of methodologies forrational drug design have been derived to enhance the affinity of the CP derivatives to NF- k B. The efficacy of these methodologies has beendemonstrated by generating a set of substituted CPs, exhibiting increased affinity for NF- k B, and opening new ways to broaden thetherapeutic applications of this class of drugs. q 2004 Elsevier B.V. All rights reserved. Keywords:  2-Cyclopentenone; Nuclear factor kappa B; Molecular mechanics; Density functional theory; HIV-1; Cancer 1. Introduction Rel/NF-kappaB (NF- k B) transcription factors are afamily of structurally-related eukaryotic transcriptionfactors that are involved in the control of a large numberof cellular and organismal processes, such as immuneand inflammatory responses, developmental processes,cellular growth, and apoptosis. In addition, these factorsare active in a number of disease states, including cancer,AIDS, arthritis, chronic inflammation, asthma, neurodegen-erative diseases, and heart disease [1]. NF- k B is sequesteredin the cytoplasm by the inhibitory proteins-I k B. With arange of biochemical stimuli, the IkB protein is degradedvia a proteasome dependent-post-ubiquitination pathway.Following this the free transcription factor translocates to 0166-1280/$ - see front matter q 2004 Elsevier B.V. All rights reserved.doi:10.1016/j.theochem.2004.06.027Journal of Molecular Structure (Theochem) 685 (2004) 73–* Corresponding author. E-mail address: (M.J. Ramos).  the nucleus where it binds to consensus kB DNA sites [2,3].Having a multistep activation pathway, NF- k B has attracteda pleothora of inhibition strategies in the recent past [4,5].One such important inhibitors of NF- k B-DNA binding is the15-deoxy- D 12,14 -prostaglandin J 2  (15d-PGJ 2 ), a dehydrationproduct of the prostaglandin D2 [6]. This molecule has beenfound to react with the redox-regulated cysteine residues inNF- k B p50 (Cys62) and NF- k B p65 (Cys38) subunits, bythe virtue of its electrophilic carbon in the cyclopentenonering [7].The redox-regulatedcysteine residues in NF- k B areimportant determinant of the DNA recognition and bindingof the protein, as a reduced state has been described torender the protein functional for  k B site-DNA binding [8].It has been shown that the 2-cyclopentenone itself alsoinhibits the NF- k B-DNA binding, although a concentration100-fold higher was needed to achieve the same level of inhibition of 15d-PGJ 2  (probably due to the lower affinityand/or specificity of 2-cyclopentenone for NF- k B).In contrast, the compounds cyclopentanone andcyclopentene, which do not contain a reactive center fornucleophilic addition, did not inhibit NF- k B activity [7].Other therapeutic properties have also been assigned to2-cyclopentenone, as the induction of Heat Shock Protein70 synthesis [9] or the induction of apoptosis in endothelialcells [10]. Based on these observations, we have designed aseries of 2-cyclopentenone (CP, Scheme 1) ring analogues,which should mimick the reactivity and therapeuticproperties of 15-deoxy- D 12,14 -prostaglandin J 2 , alongwithimproved protein binding. 2. Methodology 2.1. Drug design The model system used in this work has been derivedfrom the structure for the NF- k B p50/p50 homodimer(pdb code: 1NFK) [11], from where one of the p50monomers and the DNA sequence were deleted.The remaining p50 monomer was kept for further modeling.The sequence numbering used throughout the textcorresponds to the one of the 1NFK entry.The structure of the rigid 2-cyclopentenone was drawnby us (Scheme 1). One should note that this small ring hasonly one conformer. We began by docking the CP intoNF- k B. Assuming that a productive complexation mustobey to the constraint that the electrophilic carbon (C3) hasto be close to the thiol of Cys59, few docking alternativesremained. Therefore, there was no need for a systematicdocking search to realize which could be the best dockingpossibilities. Simple visual inspection was enough toconclude that there are two possible binding pockets forCP: one between Cys59 and Tyr57 and the other betweenCys59 and Glu60. Which one is the most favoured dependson the size and nature of the substituents. Once these firsttwo structures were modeled, the geometry of the wholesystem was optimized using molecular mechanics.It should be noticed that it is not necessary to include thewhole p50 monomer for the calculations. The inclusion of the whole p50 monomer would imply to use a lower-leveltheoretical approach, due to the long calculation timeneeded to treat the whole monomer. Instead, we includedthe residues which had any of their atoms within a radius of circa 10 A˚ , centered on the sulfur thiol of Cys59.The obtained model included the amino acids 54–68,140–146 and 207. To confirm if this model was representa-tive of the p50 monomer (in terms of the CP:NF- k B bindingproperties), and to assure that no meaningful long rangeinteracting residues were deleted, two additional modelswere built, in which the included residues were within aradius of 15 and 20 A˚ from the sulfur thiol of Cys59,respectively. The first of these included all the amino acidsof the previous model (10 A˚ cutoff) plus amino acids 53,107–110, 138–139, 147–150, 208, 239–243, and thesecond included all the amino acids of the previous model(15 A˚ cutoff), plus amino acids 52, 97–99, 106, 111–120,136–137, 151–154, 198, 201–206, 209–212, 238, 244–247.In Fig. 1 we can see the three NF- k B models.Subsequently, the CP:NF- k B binding energy wascalculated in each model, to confirm that all the modelsresulted in similar binding energies.All the calculations were performed in vacuum, using thesoftware package  GAUSSIAN 98 [12]. To perform molecular mechanics geometry optimizations and energy calculations Scheme 1. The CP molecule. An asterisk marks the electrophilic carbonwhich suffers nucleophilic addition by the thiolate of Cys59.Fig. 1. The models used in this study. The residues included in the 10 A˚ cutare shown in purple. The gray and yellow residues were also included in the15 and 20 A˚ cuts, respectively. The reactive cysteine (Cys59) is shownin cyan. (For interpretation of the references to colour in this figure legend,the reader is referred to the web version of this article.) P.A. Fernandes et al. / Journal of Molecular Structure (Theochem) 685 (2004) 73–82 74  we have used the universal force field (UFF) force field [13].This corresponds to an adequate theoretical level for asystem of this size. UFF is the only force field which hasbeen parameterized for all elements of the Periodic Table.Several benchmarks of this force field have been publishedbefore. Average errors have been determined for moleculeswhere good quality, reliable experimental data wereavailable. In terms of energy, typical deviations fromexperiment found in several organic and inorganicmolecules are just a few (1–8) kcal/mol [14,15].These errors are essentially systematic, and in the presentcase most of the errors cancel when comparing thecomplexation energy for the srcinal CP and a very similarsubstituted drug, and therefore this error amplitude isperfectly acceptable. Accurate geometries can also beobtained with this force field. In organic molecules, averagedeviations from experiment are 0.021 A˚ for C–C bondlength and 0.024 A˚ for C–N, and 5–10 8  in angle bend [16].The visualization of all geometries as well as allsubstitutions in the CP molecule, was carried out with thehelp of programs Gaussview [17] and Molden [18]. The use of partial charges, calculated by the method of charge equilibration, is possible with  GAUSSIAN 98 [19], and isthe default method for charge generation with the UFF forcefield. This method generates point charges which aregeometry-dependent, representing a considerable improve-ment in relation to traditional, non-polarizable force fieldsas CHARMM or AMBER [20,21]. However, it is notpossibleneitherefficienttocalculatethepointchargesateachoptimization step, as this would make the calculations tooslow to be feasible. To overcome this problem the pointcharges were calculated only at the beginning and at the endof the optimization process. Subsequently, it was verifiedwhether the obtained geometry could be consideredoptimized using the new calculated charges and the sameconvergence criteria. If not, a new optimization was carriedout, with the point charges calculated at the end of thepreceding optimization. The overall procedure was repeateduntil self-consistency was achieved, i.e. the optimizedgeometry should obey the same convergence criteria withboththeinitialandfinalpointcharges.Thisimpliedtypically5–10 full consecutive geometry optimizations. This meth-odology was evaluated in a previous work, by running thesamecalculationbothdeterminingthe pointchargesateveryoptimization step, and by using the self-consistent methodcalculating the point charges only at the beginningand at theend of each optimization. The final geometry was the samewithbothprocedures,demonstratingthatthesameminimumwas obtained with both methods [22].To calculate the complexation energy we used thefollowing procedure: we began by introducing the tentativesubstitutions on the CP molecule (within the CP:NF- k Bcomplex); next the geometry of the complex was optimized.Starting from that geometry the substituted CP was obtainedby deleting the NF- k B and re-optimising the geometry.The same type of procedure was applied to obtainthe optimized NF- k B fragment. This same method wasapplied also to obtain the complexation energy of theunsubstituted CP. The energy of complexation wascalculated as DD E  complex Z D E  subscomplex K D E  CPcomplex Z ð E  subs : NF- k B K E  CP : NF- k B Þ K ð E  subs K E  CP Þ K ð E  subsNF- k B K E  CPNF- k B Þ ð 1 Þ with obvious notations. Assuming that the entropiccontribution for  D G complex  is similar for the srcinal andsubstituted complexes, the free energy variation induced bythe substitution can be approximated by the correspondingenergy difference: DD G complex Z D G subscomplex K D G CPcomplex z D E  subscomplex K D E  CPcomplex (2) 2.2. Reactivity To study the reactivity of the new drugs we calculated theactivation and reaction energies for a set of representativemolecules. Density functional theory was used in allcalculations, with the  GAUSSIAN 98 suite of programs [12],at the unrestricted Becke3LYP level of theory [23–25].The 6-31G(d) basis set was used. A scan along theCys-S-Carbon 3 of CP was performed. The scan startedfrom a productive structure of the CP:methylthiolatenon-bonded complex. Afterwards the distance wassuccessively shortened by  K 0.2 A˚ increments, and thegeometry was re-optimized. This procedure continued untilthe energy has raised through S-C3 steric clash. This lastpoint was freely optimized to generate the product’sgeometry. The transition state energy was taken from thescan, and was identified as the highest energy structure.It was not freely optimized because such structures are quitedifficult to obtain for all the reactions studied, and to takethem from the scan was thought to be a reasonablecompromise between accuracy and computing time.We should emphasize that no vibrational frequencies wereevaluated, and the energies reported here only include theelectronic energy plus the interaction with the continuum.To calculate the Hessian in all stationary points and toobtain the reactants and products from internal reactioncoordinate calculations would be preferable. However, forthe sake of making a semiquantitative comparison of activation and reaction energies between molecular analogs,and considering the intrinsic error associated with thetheoretical method (estimated in 3–5 kcal/mol), the usedprotocol seems to be accurate enough. A proton was addedto the products to obtain the final protonated enol.The solvent was accounted for by the introduction of adielectric continuum ( 3 Z 78.4), with which the energy of the optimized points along the scan was recalculated.For this purpose we have used a polarized continuum model,called C-PCM, as implemented in  GAUSSIAN 98 [26]. P.A. Fernandes et al. / Journal of Molecular Structure (Theochem) 685 (2004) 73–82  75  This method considers the solute as a set of interlockingspheres, centered on each atom, with apparent surfacecharges, that interact with the wave function. Thecontinuum is modeled as a conductor, instead of a dielectric.This simplifies the electrostatic computations, and correc-tions are made a posteriori for dielectric behavior. It isusually assumed that geometry optimizations can be carriedout in vacuum, and transferred to the continuum to calculatefinal energies, without introducing significant error [27].However, in the case of the reactants, it was noted that theseparated solvated species have a lower energy than thereactants complex. Therefore, the energy of the reactantswas calculated as the sum of the energies of the solvatedmeyhylthiolate and the 2-cyclopentenone derivatives atinfinite separation. 3. Results and discussion 3.1. Molecular design To begin with, we have docked the CP molecule on thep50 monomer of NF- k B. It should be noted that no bindingpocket is expected to exist for CP. Instead, this region hasevolved to bind a specific DNA sequence, which size andshape are completely different from our lead compound.Considering the constraint that the reactive electrophiliccarbon (C3) should lie close to the nucleophilic sulfur atomof Cys59, the small size of CP, the lack of conformationalfreedom of the CP ring, and the position of Cys59 in the p50monomer (on the ‘top of a hill’, with few neighbor residuesin a radius accessible to interact with the small CP, seeFig. 1), very few options remained for the docking. Therewere mainly two possible binding positions for CP: onebetween Cys59 and Tyr57 and the other between Cys59 andGlu60. The latter was found to be the most favorable for CP.The obtained geometry was further optimized by minimiz-ation using the UFF force field. The result is shown in Fig. 2.CP binds to NF- k B in a very small crevice formed bythe side chains of Cys59 and Glu60. The most relevantinteractions consist in two pseudo hydrogen bonds betweeneach of the carboxylate oxygens of Glu60 and eachhydrogen atom bonded to C5, with distances of 2.88 and3.33 A˚, one hydrogen bond between the backbone carbonyloxygen of Cys59 and a hydrogen atom at C4 (with a lengthof 2.52 A˚), and one interaction between the hydrogen at C3and the thiol sulfur, at 3.35 A˚; the reactive carbon liesclose to the thiol sulfur, at a distance of only 3.85 A˚.The complexation energy was calculated in K 24 kcal/mol.The two larger models (with 15 and 20 A˚ cutoff) did notshow meaningful differences in the binding energy ( K 19and K 22 kcal/mol, respectively), but made the simulationsmuch more time consuming. Therefore, we have used the10 A˚ cut throughout the rest of the study.OncethebeststructurefortheCP:NF- k Bcomplexhasbeendetermined, the strategies for rational improvement of theaffinityandspecificityofthecomplexcanbederived.Wehavefollowed several strategies and will discuss them one by one. 3.2. Strategy 1: increasing the interaction between the sidechain of Cys59 and the hydrogen atom at C3. As mentioned previously, there is a strong interactionbetween the thiol of Cys59 and one of the hydrogens boundto the C3 carbon atom. To increase the interaction betweenthe thiol of Cys59 and carbon C3 has the advantage of placing the latter close to the sulfur atom, and thus in aproductive geometry for the subsequent nucleophilic attack.This interaction can be further increased by replacing thehydrogen on carbon C3 by a more polar atom or group.Moreover, this substituent can be either electronegative orelectropositive, as the thiol group can establish strongdipole–dipole interactions with both, through either theproton or the sulfur atom. Although halides, thiolates andother electron rich substituents at atom C3 should be thesource of four electron repulsions with the Cys59 sulfuratom, the attraction involves in these cases the Cys59 thiolhydrogen atom. Fig. 3 shows an example of the binding of  Fig. 2. A detail of the NF- k B:CP complex. Relevant interactions aredepicted in Angstrom.Fig. 3. A detail of the complex between compound  3  and NF- k B, as anexample of the binding of CP analogs with electron rich substituents.Relevant interactions are depicted in Angstrom. P.A. Fernandes et al. / Journal of Molecular Structure (Theochem) 685 (2004) 73–82 76  Table 1Results (in kcal/mol) for single substitutions in carbons C3–C5No. Molecule  DD G complex  No. Molecule  DD G complex 1 0 16  K 132  K 16 17  K 83  K 11 18  K 84  K 13 19  K 95  K 20 20  K 56  K 5 21  K 137  K 3 22  K 268  K 7 23  K 299  K 6 24  K 1610  K 15 25  K 1811  K 13 26  K 2012  K 8 27  K 3213  K 4 28  K 4714  K 11 29  K 8115  K 13 30  K 93The complexation free energy differences are expressed in kcal/mol. P.A. Fernandes et al. / Journal of Molecular Structure (Theochem) 685 (2004) 73–82  77
Related Search
Similar documents
View more...
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks