Medicinal Chemistry Research

, Volume 21, Issue 12, pp 4430–4436

Scouting new molecular targets for CFTR therapy: the HSC70/BAG-1 complex. A computational study

Authors

  • Elena Cichero
    • Dipartimento di Scienze FarmaceuticheUniversità di Genova
  • Anna Basile
    • Dipartimento di Scienze Farmaceutiche (FARMABIOMED)Università di Salerno—Via Ponte Don Melillo
    • Biouniversa Srl
  • Maria Caterina Turco
    • Dipartimento di Scienze Farmaceutiche (FARMABIOMED)Università di Salerno—Via Ponte Don Melillo
    • Biouniversa Srl
  • Mauro Mazzei
    • Dipartimento di Scienze FarmaceuticheUniversità di Genova
    • Dipartimento di Scienze FarmaceuticheUniversità di Genova
Original Research

DOI: 10.1007/s00044-012-9985-1

Cite this article as:
Cichero, E., Basile, A., Turco, M.C. et al. Med Chem Res (2012) 21: 4430. doi:10.1007/s00044-012-9985-1
  • 146 Views

Abstract

HSC70 has been identified as an important molecular target involved in the ΔF508-CFTR cystic fibrosis. HSC70 associates ΔF508-CFTR to a much greater extent than WT-CFTR and after this step, it recruits other co-chaperones (BAG1, CHIP) and performs the ubiquitination and proteosomal degradation of the protein. Up to now, several X-ray data concerning the HSC70:BAG1 complexes are available. Thus, we performed an “in silico” investigation focused to explore which different amino acid residues are involved in the binding of ATP, the natural substrate, and the co-crystallized ligands at the HSC70/BAG-1 interface. The study allowed us to evaluate sildenafil and KM11060, which proved to be also CFTR correctors, as potential HSC70:BAG1 inhibitors, and also let us derive interesting perspectives for the development of new CFTR correctors.

Keywords

CFTRHSC70BAG-1Computational study

Introduction

Cystic fibrosis (CF) is a multi-organ genetic disorder caused by the loss of function of the CF transmembrane conductance regulator (CFTR) (Riordan, 2008). CF is the most common autosomal recessive disease in people of European ancestry, with more than 1,700 different mutations identified in the CFTR gene. The CFTR protein is expressed in epithelial cells and is a member of the ATP-binding cassette (ABC) transporter superfamily, which also includes the multidrug-resistance protein. CFTR consists of two membrane-spanning domains, two nucleotide-binding domains (NBDs), and a regulatory domain, which controls channel activity. CFTR is a chloride channel and regulator of other transporters (Quinton, 1983; Quinton and Bijman, 1983).

Loss of function of CFTR results in viscous secretions of the exocrine glands in multiple organ systems and elevated sweat chloride levels. The most prevalent symptoms develop in the gastrointestinal and respiratory tracts. While administration of pancreatic enzyme supplements has significantly reduced the gastrointestinal complications and extended the lifespan of patients to more than 35 years, the chronic bacterial airway infections lead to respiratory failure. At present, the respiratory complications are responsible for the majority of mortality associated with this disease. In spite of the fact that symptomatic combination therapies have been employed, including antibiotic, anti-inflammatory, and mucolytic medications, the lung function often declines to less than 30%, and transplantation remains the last treatment option. Since the discovery of the CFTR gene in 1989 (Riordan et al., 1989), a major effort has been directed to develop causative therapies for CF. These include gene-replacement strategies, based on both viral and non-viral vectors (Griesenbach and Alton, 2009), and a number of alternative approaches (Amaral and Kunzelmann, 2007) among which potentiators, compounds that stimulate pre-activated CFTR channel activity and correctors, agents that rescue cell-surface expression of mutated/defective CFTR, in particular ΔF508-CFTR, the most abundant protein mutation. However, despite these efforts, the current arsenal of CFTR “drugs” is limited, with only three compounds reaching the clinical trials and none on the market.

More recently, increasing attention has been devoted to those molecular chaperones which assist the protein folding processes within the cell and, among these, HSC70 has been identified as an important molecular target. HSC70, in fact, associates ΔF508-CFTR to a much greater extent than WT-CFTR and after this step, it recruits other co-chaperones (BAG1, CHIP) and performs the ubiquitination and proteosomal degradation of the protein (Alberti et al., 2004). Interestingly, the X-rays of HSC70 in complex with BAG-1 and ATP, or small inhibitors are available on PDB (PDB code: 3FZF and 3LDQ, respectively) (Williamson et al., 2009; Macias et al., 2011).

On these basis, we found it could be interesting to perform an “in silico” investigation focused to explore which different amino acid residues are involved in the binding of ATP, the natural substrate, and the co-crystallized ligands at the HSC70/BAG-1 interface, so as to define which interactions are specific for switching the activation and which ones are instead specific for switching the inhibition of the chaperone-co-chaperone assembly. This information could in fact represent the first feature able to discriminate between the two groups of substances and could subsequently be useful to select from large databases new potential CFTR agents to be preliminary tested in vitro.

Recently, NSC71948 (Fig. 1), a mixture of the I and II components, has been discovered as acting on the HSC70:BAG1 assembly, displaying an IC50 of 0.9–0.1 μM in the in vitro HSC70:BAG-1 interaction assay (Sharp et al., 2009).
https://static-content.springer.com/image/art%3A10.1007%2Fs00044-012-9985-1/MediaObjects/44_2012_9985_Fig1_HTML.gif
Fig. 1

Chemical structure of the mixture NSC71948

Thus, both the two components have been submitted to docking simulations, and then compared with the X-ray data concerning the HSC70:BAG1 complexes. The derived key interactions, which resulted to be effective on the HSC70:BAG1 inhibition, have been successively employed to evaluate, through docking procedures, a number of compounds recently described as CFTR correctors. In particular, we focused our attention on sildenafil analogues (Fig. 2), which proved to show CFTR corrector behavior, as reported by Kalid et al. (2010). Among them, KM11060 resulted to have greater corrector activity than sildenafil.
https://static-content.springer.com/image/art%3A10.1007%2Fs00044-012-9985-1/MediaObjects/44_2012_9985_Fig2_HTML.gif
Fig. 2

Chemical structure of sildenafil and its analogues. The common structure is depicted in blue (color figure online)

Notably, these compounds are well-known type 5 phosphodiesterase inhibitors, wherein the pyrazolo pyrimidine scaffold is able to resemble the cAMP behavior, by properly fitting the enzyme catalytic site. Similarly, it could be probable that the same compounds could act as HSC70:BAG1 ligands, occupying the protein ATP domain.

The obtained results, being in agreement with the experimental data, allowed us to reasonably hypothesize (for the examined compounds) a binding mode able to suggest interesting perspectives for expedite the development of new CFTR correctors.

Materials and methods

Ligand preparation

For our studies, compounds I and II (Fig. 1) were selected as well-known HSC70:BAG1 inhibitors, while sildenafil and KM11060 (Fig. 2) were chosen since they have proved to be the most promising CFTR correctors, among the PDE5 inhibitors.

All the compounds were built, parameterised (Gasteiger–Huckel method) and energy minimized within MOE using MMFF94 force field (MOE: Chemical Computing Group Inc. Montreal. H3A 2R7 Canada. http://www.chemcomp.com).

The I and II sulphonic groups were considered in the undissociated (Iu and IIu) and in the dissociated (Id and IId) acid forms.

Successively docking studies were performed, using the HSC70:BAG1 complexes. The X-ray coordinates of HSC70 in complex with BAG-1 and ATP (PDB code: 3FZF) and those of HSC70 in complex with BAG-1 and a small inhibitor (PDB code: 2LDQ) were downloaded by the PDB protein data bank.

Docking simulations

Each compound was docked into the HSC70:BAG1 assembly, by means of the Surflex docking module implemented in Sybyl-X 1.0 (Sybyl-X 1.0., Tripos Inc 1699 South Hanley Road, St Louis, Missouri, 63144, USA). Then, for all the compounds, the best docking geometries (selected on the basis of the SurFlex scoring functions) were refined by ligand–receptor complex energy minimisation (CHARMM27), by means of the MOE software. In order to verify the reliability of the derived docking poses, the obtained protein/ligand complexes were further investigated by docking calculations (10 run), using MOE-Dock (Genetic algorithm; applied on the poses already located into the protein-binding site). The conformers showing lower energy scoring functions and RMSD values (respect to the starting poses) were selected as the most stable.

Results and discussion

The ATP-binding mode into the HSC70:BAG1 assembly

On the basis of the experimental data (PDB: code: 3FZF), the ATP behavior seems to be determined by interactions with two narrow region (P1 and P2). P1 includes Thr13, Thr14, Tyr15, Gly202, Gly203, Thr204, Gly338, Glu339, and Asp366 while P2 consists of Gly230, Glu231, Glu268, Lys271, Ser275, Arg272, Ser340, and Arg342 (Fig. 3a).
https://static-content.springer.com/image/art%3A10.1007%2Fs00044-012-9985-1/MediaObjects/44_2012_9985_Fig3_HTML.jpg
Fig. 3

X-ray pose of ATP (a) and of the small inhbitor (b) into the 3FZF (C atom: yellow) and 3LDQ (C atom: green) binding site, respectively (color figure online)

In details, as shown in the MOE LigPlot depicted in Fig. 4, ATP is engaged in several H-bond interactions between: (i) one of the ribose core oxydryl groups and Glu268 and Lys271, (ii) one nitrogen atom of the purine core and Ser275 and (iii) between the phosphate groups and Thr14, Tyr15, and Gly339.
https://static-content.springer.com/image/art%3A10.1007%2Fs00044-012-9985-1/MediaObjects/44_2012_9985_Fig4_HTML.gif
Fig. 4

MOE LigPlot of ATP into the 3FZF binding site

The binding mode of a small inhibitor into the HSC70:BAG1 assembly

According to the 3LDQ X-ray coordinates, the small inhibitor exclusively interacts with the P2 region, being also projected outside the ATP-binding pocket (Fig. 3b). In particular, the ligand ribose group is engaged H-bond interactions with Glu268 and Lys271 while the purine ring displays one H-bond with Ser275 (Fig. 5, MOE LigPlot). Furthermore, the purine ring and the quinoline one are engaged in cation–π contacts with Arg272.
https://static-content.springer.com/image/art%3A10.1007%2Fs00044-012-9985-1/MediaObjects/44_2012_9985_Fig5_HTML.gif
Fig. 5

MOE LigPlot of the small inhibitor into the 3LDQ binding site

A comparison between HSC70:BAG1 in complex with ATP and in complex with the small inhibitor

Taking into account all the experimental data previously discussed, the agonist behavior seems to be determined by several hydrophilic contacts with the P1 region, which is unoccupied by the compound showing inhibitory activity. Notably, the agonist and the inhibitor share H-bonds with Ser275, Glu268, and Lys271 residues, suggesting that the three amino acids could become an interesting starting point to design new HSC70:BAG1 ligands.

To better explore those differences in the 3LDQ and 3FZF complexes which could be related to the binding of compounds showing agonist or inhibitory activity, the two assemblies were superimposed using MOE software (RMSD = 0.534 Å). Regarding the HSC70:BAG1 area located 5 Å far from the two ligands, the RMSD value increases to 5.492 Å. As shown in Fig. 6, the 3LDQ (C atom in green) and the 3FZF (C atom in yellow) binding domains are mainly different in the orientation of residue Arg272 side-chain. In particular, the HSC70:BAG1 complex with the inhibitor appears to be accompanied by a HSC70 protein-rearrangement, by switching the residue Arg272 outside the ATP-binding site. Accordingly, the inhibitor could display more interaction with the already mentioned Arg272, reducing its contacts with the P1 region.
https://static-content.springer.com/image/art%3A10.1007%2Fs00044-012-9985-1/MediaObjects/44_2012_9985_Fig6_HTML.jpg
Fig. 6

Superimposition of the 3FZF (C atom: yellow) and 3LDQ (C atom: green) X-ray coordinates. The two ligands are depicted in stick (color figure online)

Docking studies on compounds I and II

As reported by Sharp et al. (2009), the mixture NSC71948 shows HSC70:BAG1 inhibitor activity (IC50 of 0.9–0.1 μM in the in vitro HSC70:BAG-1 interaction assay). Thus, the two components, I and II, (taking into account both the un-dissociated and the dissociated form) were docked into the HSC70:BAG1 complex, using the 3LDQ X-ray coordinates. According to our calculations, the Id form is able to interact with the residue Glu268, also involved in the ATP and in the inhibitor binding modes, as previously discussed. The sulphonic group of Id, bearing a negative charge, is projected towards the interface region located between the HSC70 and BAG1 proteins (Fig. 7a). On the other hand, Iu form results to be involved in H-bonds with Glu231, Asp366, and Ser275 (Fig. 7b). Furthermore, cation–π contacts are also detected with Tyr15 and Arg272. Notably, both ATP and the small inhibitor interact with Ser275, while only the inhibitor could interact with Arg272. As far as concern compound II, the dissociated form IId is able to display H-bond contacts between the sulphonic group and Arg272 side-chain, while the two quaternary nitrogen atoms are engaged in hydrophilic interactions with Glu268 and Asp366, respectively, (Fig. 8a). The undissociated form IIu is involved in H-bonds with its sulphonic group and Lys271 and Ser340, and also shows cation–π contacts with Arg272 side-chain (Fig. 8b).
https://static-content.springer.com/image/art%3A10.1007%2Fs00044-012-9985-1/MediaObjects/44_2012_9985_Fig7_HTML.gif
Fig. 7

MOE LigPlot of the Id (a) and Iu (b) forms into the 3LDQ binding site

https://static-content.springer.com/image/art%3A10.1007%2Fs00044-012-9985-1/MediaObjects/44_2012_9985_Fig8_HTML.gif
Fig. 8

MOE LigPlot of the IId (a) and IIu (b) forms into the 3LDQ binding site

Taking into account all the results previously discussed, and comparing the derived docking poses with the 3LDQ experimental data, the IIu form results to be the one displaying the most similar behavior with respect to that of the small inhibitor (compare Fig. 8b with Fig. 4). On the contrary, the component I, bearing a more bulky scaffold, is unable to share the configuration of the small inhibitor into the HSC70:BAG1 complex. Thus, it is probable that the component II, and in particular the IIu form, more efficiently interacts with the HSC70:BAG1 assembly, being in hydrophilic contact with Arg272, and by properly occupying the complex interface (Fig. 9). Accordingly, for the mixture NSC71948, IC50 value falls in the micromolar range.
https://static-content.springer.com/image/art%3A10.1007%2Fs00044-012-9985-1/MediaObjects/44_2012_9985_Fig9_HTML.jpg
Fig. 9

The selected docking pose of the IIu form (in spacefilled mode, C atom in green) into the 3LDQ binding site is depicted. The HSC70 and the BAG1 backbones are colored in blue and yellow, respectively. The 3LDQ complex small inhibitor is reported in space filled mode, C atoms in gray (color figure online)

With the aim of optimizing the compound II scaffold, we here propose some structural modifications which could be made.

Interestingly, the introduction of a flexible linker between the ring bearing the sulphonic group and the subsequent one, could be beneficial for the inhibitory activity, by enhancing the formation of cation–π contacts with Arg342. Furthermore, the substitution of the two quaternary nitrogen atoms with an H-bond acceptor function could allow the establishment of H-bonds with the key residue Arg272.

Docking studies on sildenafil and KM11060

The docking studies performed by us highlight that sildenafil and KM11060 are involved in cation–π interactions with Arg272 and Arg342 residues, by means of the pyrazolopyrimidone ring and of the quinoline one, respectively, (Fig. 10a). Furthermore, KM11060, is also engaged in an additional H-bond with Lys271 (Fig. 10b). Notably, Arg272 is a key residue for the binding of the HSC70:BAG1 inhibitor, while Lys271 results to be involved in the interaction between the HSC70:BAG1 assembly and both ATP and the small inhibitor compound.
https://static-content.springer.com/image/art%3A10.1007%2Fs00044-012-9985-1/MediaObjects/44_2012_9985_Fig10_HTML.gif
Fig. 10

MOE LigPlot of Sildenafil (a) and KM11060 (b) into the 3LDQ binding site

Accordingly, KM11060 proves to have greater corrector activity than sildenafil.

In particular, by comparing the KM11060 selected docking pose and the 3LDQ X-ray coordinates, the sulfonyl moiety of the compound is partially able to resemble the ribose behavior in the inhibitor, detecting contacts with Lys271, while the quinoline ring is properly located in the area surrounding the corresponding ring of the inhibitor (Fig. 11). On the contrary, the piperazine ring does not occupy the region around the purine core, showing no contacts with Ser275.
https://static-content.springer.com/image/art%3A10.1007%2Fs00044-012-9985-1/MediaObjects/44_2012_9985_Fig11_HTML.jpg
Fig. 11

The selected docking pose of KM11060 (in stick mode, C atoms in white) into the 3LDQ binding site is depicted. The 3LDQ complex small inhibitor is reported in stick mode, C atoms in gray (color figure online)

On the basis of these data, the introduction of an indole ring between the quinoline one and the sulfonyl group (instead of a piperazine), could be useful to properly occupy the protein-binding pocket. On the other hand, the quinoline ring could be substituted by a flexible linker, with an H-bond acceptor function, which could be engaged in H-bond with the key residue Arg272.

Any detail concerning the docking results is reported in Table 1.
Table 1

Key interactions observed in the derived HSC70:BAG1/ligand complexes

Compound

H-bond interactions

Cation-π

3FZF ATP

Thr14 (3.02 Å), Tyr15 (2.95 Å), Gly339 (3.39 Å),

Glu268 (2.33 Å), Lys271 (2.93 Å), Ser275 (2.51 Å)

2LDQ small inhibitor

Glu268 (2.73 Å), Lys271 (3.37 Å), Ser275 (2.68 Å)

Arg272

Id

Glu268 (2.82 Å)

Iu

Asp366 (2.78 Å),

Tyr15

Glu231 (2.83 Å), Ser275 (2.75 Å)

Arg272

IId

Asp366 (2.85 Å)

Arg272 (2.75 Å), Glu268 (3.02 Å)

IIu

Lys271 (2.73 Å), Ser340 (2.84 Å)

Arg272

Sildenafil

Arg272, Arg342

KM11060

Lys271 (3.06 Å),

Arg272, Arg342

The residues which are involved both in the 2LDQ complex contacts and in the I, II, Sildenafil, and KM11060 binding mode, are reported in bold

The residues which belong to the P1 pocket are underlined. The H-bond distances between the ligand atoms and the protein amino acids are also reported

Conclusion

The docking studies here presented highlight the key structural features impacting the behavior of compounds which act as HSC70:BAG1 agonists or inhibitors. In particular, the agonist behavior seems to be determined by several hydrophilic contacts with the P1 region, which is unoccupied by the 2LDQ small inhibitor. On the other hand, the agonist and the inhibitor share H-bonds with the Ser275, Glu268, and Lys271 residues, suggesting that these three amino acids could become an interesting starting point to design new HSC70:BAG1 ligands. Accordingly, on the basis of our docking studies, the NSC71948 mixture (I and II) which acts as HSC70:BAG1 inhibitor, interacts with (at least) one of the three key residues previously described.

Furthermore, the derived key interactions have been employed to evaluate two PDE5 inhibitors, sildenafil and KM11060, which proved to be also CFTR correctors. The results obtained allowed us to reasonably hypothesize the binding mode of the two compounds onto the HSC70:BAG1 complex, and also to derive interesting perspectives for the development of new CFTR correctors.

Acknowledgments

This research was supported by Italian Cystic Fibrosis Foundation (Grant FFC#5/2010), with the contribution of Philip Watch–Morellato & Sector Group. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. E.C. was financially supported by a post-doc fellowship, Area Chimica, University of Genova.

Copyright information

© Springer Science+Business Media, LLC 2012