Discovery Of Hsp90 Inhibitors For Disrupting The Hsp90/Cdc37 Complex

Inhibition of molecular chaperone Hsp90(90 k Da heat shock protein) has become an attractive therapeutic approach for the treatment of cancer. Hsp90 facilates the folding, assembly and maturation of its client proteins in cancer cells, thus promoting cancer cell survival and invasion. Various Hsp90 inhibitors have been developed in the past two decades, such as the purine scaffold Hsp90 inhibitor geldanamycin and its derivative 17-AAG, the natural product radicicol and some small peptides. Most of these Hsp90 inhibitors block Hsp90 N-terminal ATP binding site. To date, 17 agents have been tested in clinical trials. However, none of them has received approval from the US Food and Drug Administration(US FDA).Hsp90 binds and releases different cochaperones at different stages of its ATP cycle to achieve its chaperoning function. Thus, targeting Hsp90 cochaperones rather than focusing on Hsp90 ATPase activity may represent a new concept for developing Hsp90 inhibitors. Our preliminary data and recent literature demonstrate that Cdc37(cell division cycle protein 37) plays a central role in loading kinase client proteins into Hsp90 superchaperone complex. Many of these kinase clients are oncogenic proteins in cancer cells. The crucial role of the Hsp90/Cdc37 complex in kinase maturation has made it an exciting target for cancer treatment. Therefore, screening small molecule inhibitors targeting Hsp90/Cdc37 interaction might be a promising strategy for developing novel cancer therapeutics.However, the complexity of the interaction patch between Hsp90 and Cdc37 makes it difficult to identify Hsp90/Cdc37 inhibitors. Further, investigation of the interaction between Hsp90 and Cdc37 has been limited to the use N-terminal yeast Hsp90 and middle terminal human Cdc37 fragments due to the large molecular weights of these two proteins, which may not truly reflect the nature of interactions between the two proteins. Furthermore, the methods commonly used for studying Hsp90/Cdc37 interactions, such as co-immunoprecipitation, NMR, and crystallography, can only qualitatively measure the interacting status of Hsp90 and Cdc37, making these methods not suitable for screening of Hsp90/Cdc37 inhibitors. To date, only a few small molecule compounds have been identified to disrupt Hsp90/Cdc37 thus leading to tumor cell apoptosis. Therefore, there is an urgent need for the establishment of an effective approach to study the Hsp90/Cdc37 complex and screen small molecule inhibitors specifically targeting Hsp90/Cdc37 interactions.In our previous study, we utilized a newly developed bioluminescence imaging method split renilla luciferase protein fragmentassisted complementation(SRL-PFAC), to study Hsp90/Cdc37 interactions in living cells. We also proved that SRL-PFAC is sensitive and specific to quantitatively monitor Hsp90/Cdc37 interactions in HEK293 cells, thus providing a rationale to screen Hsp90/Cdc37 inhibitors using SRL-PFAC method. However, this bioluminescence system needs the cells to be transiently transfected with NRL-Hsp90α, Cdc37-CRL and RL eukaryotic expression plasmids in 24-well plates to produce exogenous proteins prior to carrying out the complement renilla luciferase(RL) assay, which is costly and time consuming. This makes the eukaryotic system unlikely to be suitable for the screening of Hsp90/Cdc37 inhibitors.In this study, we first successfully established an in vitro Hsp90α/Cdc37 inhibitors screening system based on the SRL-PFAC method. We first fused human full-length Hsp90α and Cdc37 to the N- and C-terminus RL(split from amino acid 229 and 230), respectively, and then cloned them into a prokaryotic expression vector p ET-30 a. Then, we expressed and purified the recombinant NRL-Hsp90α and Cdc37-CRL proteins in Escherichia coli strain BL21(DE3). Hsp90α tends to express as inclusion bodies in E. coli strains due to its large molecular weight and quaternary structure. The study of purified Hsp90α has been limited to the use of N-terminal Hsp90α fragment, which cannot truly reflect the nature of protein-protein interactions. By optimizing the induction and purification conditions, we successfully obtained soluble NRL-Hsp90α and Cdc37-CRL proteins with biological functions. Our SDS-PAGE results showed that the purities of the two fusion proteins were both over 90%. The yield of NRL-Hsp90α and Cdc37-CRL protein was 1.0 mg and 3.0 mg per liter, respectively, which was suitable for our screening system. We also performed western blot to further confirm the SDS-PAGE results, and the results were consistent with our SDS-PAGE data.After establishment of the optimal culture and purification conditions of NRL-Hsp90α and Cdc37-CRL, we optimized complement conditions of these two proteins in 96-well plates, including the protein concentrations, assay buffer, and incubation time and temperature. We then evaluated the specificity of our in vitro SRL-PFAC assay to monitor Hsp90/Cdc37 interactions. The specificity of the complement assay was further confirmed by two site-direct mutations, NRL-Hsp90α(Q133A) and Cdc37(R167A)-CRL, which only retained 20% of the complement RL activities compared with their wild type counterparts. In addition, we expressed and purified the full-length RL protein as a control to rule out non-specific inhibition of RL catalytic activities by the compounds to be screened. We then evaluated the sensitivity of the in vitro SRL-PFAC system for screening of small molecule compounds. After testing over 60 compounds on both RL and complemented NRL-Hsp90α/Cdc37-CRL proteins, we found that sulforaphane, withaferin A, celastrol and EGCG all decreased the complemented NRL-Hsp90α/Cdc37-CRL activities in a concentration dependent manner. However, celastrol and EGCG treatments also caused decrease of the full-length RL activity, whereas sulforaphane and withaferin A treatments did not. In contrast, LBH-589 and 17-AAG showed no effect on the leuciferase activities of either NRL-Hsp90α/Cdc37-CRL or RL. These results demonstrate that our in vitro SRL-PFAC system is sensitive and accurate to reflect the effects of small molecule compounds on Hsp90/Cdc37 interactions. Therefore, this system is ideal for screening of Hsp90/Cdc37 inhibitors.By using this in vitro SRL-PFAC bioluminescence method, we screened numerous small molecule compounds and obtained some candidates which had the potential to disrupt the Hsp90/Cdc37 complex. And then, we further utilized SRL-PFAC method in HEK293 cells to further verify if these candidates are indeed specific inhibitors of Hsp90/Cdc37.Luckily, we found that the natural plant flavonoid apigenin(4′,5,7,-trihydroxyflavone), which widely exists in different fruits and vegetables, reduced the complemented NRL-Hsp90α/Cdc37-CRL activity in a concentration-dependent manner, whereas its analogues, galangin, baicalein, genistein and naringenin, did not show similar disrupting function. In addition, the false positive assay data showed that apigenin did not show any inhibitory effect on the full-length RL catalytic activity compared to the non-treated group, indicating that apigenin directly disrupted Hsp90/Cdc37 complex with structural specificity. Apigenin also inhibited NRL-Hsp90α and Cdc37(Ser13Ala)-CRL complementation. Moreover, casein kinase II(CK2) specific inhibitor 4, 5, 6, 7-tetrabromobenzotriazole(TBB) did not affect NRL-Hsp90α/Cdc37-CRL complementation, indicating that the inhibitory function of apigenin on Hsp90/Cdc37 was not reliant on the activity of CK2, which regulates the phosphorylation of Cdc37 N-terminal Ser13 and mediates Cdc37 binding to client kinases and recruiting Hsp90 to the kinase-Cdc37 complex. Furthermore, the results of our co-immunoprecipitation and western blotting showed that apigenin blocked Hsp90/Cdc37 complex formation and induced degradation of Hsp90/Cdc37 downstream client proteins Akt, Cdk4 and MMP-9 in pancreatic cancer cell line MIAPa Ca-2, in a concentration- and time-dependent fashion. Apigenin also inhibited the growth of pancreatic cell lines, MIAPa Ca-2, PANC-1, As Pc-1 and Bx PC-3, with IC50 values of 47.90 ± 3.12 μM, 51.17 ± 3.08 μM, 30.62 ± 4.31 μM and 13.57 ± 2.06 μM, respectively. Scratch wound healing assay also showed that apigenin inhibited MIAPa Ca-2 and PANC-1 cell migration in a concentration-dependent manner. Finally, results of our 2′,7′-dichlorofluorescein diacetate(DCFH-DA) probe assay showed that apigenin induced accumulation of intracellular reactive oxygen species(ROS) in MIAPa Ca-2 and PANC-1 cells. All these findings indicate that apigenin works as a structural specific inhibitor of Hsp90/Cdc37. It directly blocks Hsp90/Cdc37 interaction and leads to downstream client protein degradation in pancreatic cancer cells, and as a consequence, induces intracellular ROS accumulation and inhibits the growth and migration of pancreatic cancer cells.Taken together, our data suggest that purified NRL-Hsp90α and Cdc37-CRL appeares to be pure, stable and active conformation for small molecule inhibitors binding, thus this in vitro Hsp90/Cdc37 SRL-PFAC system we established in this work is ideal for Hsp90/Cdc37 inhibitors screening. The constructed SRL-PFAC method in living cells can be used to further verify small molecule compounds which target Hsp90/Cdc37 interactions. Therefore, both of the SRL-PFAC systems are powerful platforms facilatiing the development of novel inhibitors for disruption of the Hsp90/Cdc37 interactions.