Previously we showed that a biotinylated version of the HSP70 inhibitor PES (Biotin-PES, Supp. Fig. 1A
) can pull down HSP70 from cell extracts (7
). We next wanted to test whether PES binds directly to HSP70, and whether un-biotinylated compound could compete this interaction. Toward this end we purified recombinant human HSP70 from bacteria, and used this protein in pull down assays with Biotin-PES, as described (7
). Biotin-PES was consistently able to pull-down purified HSP70; moreover, we were able to efficiently compete this interaction using untagged compound (). These data support the premise that PES binds directly to HSP70, and that this interaction is specific.
Binding to PES requires residues within a C-terminal helical bundle of HSP70
We next sought to narrow down the region of HSP70 responsible for interaction with PES. We previously identified the substrate binding domain (SBD) of HSP70 (amino acids 386-641) as required for PES binding (7
). The substrate binding domain of HSP70 can be divided into two subdomains: a beta-sandwich region that maps to amino acids 393-507 (SBD-beta), and a C-terminal alpha-helical “lid” that maps to amino acids 507-616 (SBD-alpha) (20
). We generated deletion mutants of these sub-domains and used these in pull-down assays with biotinylated PES. To ensure that these deletion mutants were properly folded, we first checked them for their ability to interact with APAF1 in transfected cells. Several deletion mutants of SBD-alpha (encoding amino acids 386-616, 386-602 and 386-573) retained the ability to interact with APAF1, indicating that they were not grossly misfolded (data not shown). In contrast, deletion mutants in SBD-beta were unable to bind to APAF1 or CHIP, suggesting they were misfolded (data not shown); these were therefore not analyzed further. In biotinylated PES pull-down assays, the HSP70 construct encoding amino acids 386-616 showed strong binding to PES, while that encoding amino acids 386-602 demonstrated markedly impaired binding, and those encoding 386-573 and 386-543 consistently failed to bind (). The combined mapping data indicate that HSP70 requires SBD-alpha (the C-terminal helical lid, including amino acids 573-616) in order to interact with PES ().
We next chose to use in silico
docking in order to identify potential PES binding sites in the SBD of HSP70. Toward this end, the human sequences of HSP70 (Uniprot code P08017: residues 391 to 607) and Hsc70 (Uniprot code P11142: residues 361 to 613) were aligned to two different homologous proteins of known structure (the E. coli
DNA-K and Bovine Hsc70 substrate binding domains, PDB codes 1DKZ and 1YUW, respectively) using the program MolIDE (17
). Side chain conformations were predicted with the program SCWRL4 (19
), and the SCWRL4-generated models were subjected to a simple minimization using Chimera (UCSF). Following generation of this model, in silico
docking routines with PES were generated using the program AutoDock. This- analysis revealed three potential PES interaction sites within the SBD of HSP70. We identified predicted contact residues for all three potential docking sites, and generated HSP70 point mutants for each residue; these point mutants were then used in Biotin-PES pull down assays. We found that three different point mutants comprising one of the docking sites all showed markedly impaired PES binding; notably, this docking site was located in the C-terminal helical domain, so this site was consistent with our mapping data. In this docking model, residues N548, I607 and Y611 all make significant contacts with PES (). Mutagenesis of each of these residues produced HSP70 proteins that showed either normal, or somewhat decreased, interaction with Hsc70 and CHIP (), but which showed markedly impaired binding to biotinylated PES (). In contrast, the majority of the other point mutants showed normal ability to bind to PES (data not shown). Our combined data support this binding pocket in the C terminal helical domain of HSP70 as the potential binding site for PES.
A logical next step for these studies was to perform a preliminary structure-activity relationship for PES. Toward this goal we identified and obtained three analogues of PES from the National Cancer Institute chemical repository (Supp. Fig. 1B
). These compounds were compared to PES for cytotoxicity and the ability to inhibit autophagy. Two of these compounds contained a reduced acetylene group and had greatly decreased cytotoxicity, along with decreased ability to inhibit autophagy (Supp. Figs. 1C, D
). We next synthesized several derivatives of PES; three of these derivatives replaced the amide group of PES with a pyrrolidine ring (PES-P). One of these derivatives introduced a chloride at the meta position of the phenyl ring (PES-PCl), and another introduced a carboxymethyl group at the para position (PES-PCM) (Supp. Fig. 2A
). We found that the compounds with the pyrrolidine ring generally showed reduced cytotoxicity compared to PES; however, the compound containing a pyrrolidine ring plus a chloride on the phenyl ring (PES-PCl) had comparable cytotoxicity (Supp. Figs. 2B, C
). Additionally, PES-PCl was consistently able to induce programmed cell death, an attribute unique to this derivative (Supp. Fig. 2C
). These data prompted us to synthesize and test a PES derivative containing only a chloride on the phenyl ring, or 2-(3-chlorophenyl) ethynesulfonamide, hereafter referred to as PES-Cl.
We next compared the properties of PES-Cl () with the parent compound PES using assays for cell viability, apoptosis, autophagy inhibition, and the ability to reduce the levels of HSP90 client proteins. Cell viability assays indicated that PES-Cl has an IC50
in tumor cells that is up to 8-fold reduced relative to the parent compound PES (). We also found that PES-Cl has superior ability to cause increased p62SQSTM1
and LC3 accumulation, indicators of autophagy inhibition (). Notably, unlike PES, treatment of cells with PES-Cl was accompanied by robust programmed cell death, as determined by the appearance of cleaved lamin A () and Annexin V positive cells (). Like PES, we found that PES-Cl showed comparable ability to bind to HSP70 (Supp. Fig. 3A
), limited cytotoxicity to non-transformed human fibroblasts (Supp. Fig. 3B
), and consistent ability to down-regulate the soluble levels of HSP90 client proteins (Supp. Fig. 3C
). In silico
docking revealed that PES-Cl has the potential to dock in the identical site on the C-terminal helical bundle in HSP70 (Supp. Fig. 3D
). These data indicated that PES-Cl might prove to be a superior anti-cancer compound.
PES-Cl shows increased cytotoxicity and superior ability to inhibit autophagy
In order to further test the potential of PES-Cl as a promising anti-cancer compound, we chose to compare the cytotoxicity of PES-Cl in primary melanocytes to that in five different melanoma cell lines. Two of these melanoma lines have acquired resistance to BRAF inhibitors (451Lu-R and 1617-R), and one is intrinsically resistant (WM1366), therefore giving us the opportunity to analyze the efficacy of PES-Cl in what are otherwise chemo-resistant cancer lines (20
). We found that the IC50
for PES-Cl in all five melanoma cell lines is between 2–5 uM; in contrast the IC50 for primary melanocytes is over 100 uM (). The increased cytotoxicity in melanomas compared to primary melanocytes is associated with increased inhibition of autophagy and increased apoptosis, as assessed by p62SQSTM1
accumulation and the appearance of caspase-cleaved lamin A and caspase-3 in melanomas (). Interestingly, the BRAF-inhibitor-resistant melanomas showed either indistinguishable or increased apoptosis in response to PES-Cl, compared to parental cells (). Electron microscopy analysis supported the premise that PES-Cl has superior ability to inhibit autophagy compared to PES, as evidenced by a significantly greater accumulation of autophagosomes in tumor cells (). The combined data indicate that PES-Cl retains all of the activities of PES (including the preferential death of tumor over normal cells), but in addition shows superior cytotoxicity and the unique ability to induce apoptosis.
Enhanced autophagosome accumulation in cells treated with PES-Cl
In the course of these analyses we noted apparent mitotic defects in cells exposed to PES-Cl. Therefore, we performed cell cycle analysis of tumor cells treated with both HSP70 inhibitors. This analysis revealed that treatment of tumor cells with these compounds led to a significant accumulation of cells in the G2/M phase of the cell cycle (). These data are consistent with the finding that silencing HSP70 leads to G2/M arrest (4
). In order to probe the possible mechanism underlying G2/M arrest by HSP70 inhibitors, we noted a previous study suggesting that an ATP-dependent chaperone controls the Anaphase Promoting Complex/Cyclosome (APC/C) (16
). To test the hypothesis that this ATP-dependent chaperone was HSP70, we monitored cyclin B1 degradation in cell-free extracts made from synchronized HeLa cells. In this assay, extracts were made from HeLa cells synchronized in G1/S using thymidine block; following addition of ATP, the activity of the APC/C is monitored by the degradation of cyclin B1. Notably, the ability of the extracts to activate the APC/C and degrade cyclin B was markedly inhibited by PES-Cl, to levels comparable to the positive control tautomycin (). These data suggested that these compounds were affecting a biochemical process required for mitotic exit and thus sustaining the mitotic checkpoint. To probe this finding further, HeLa cells stably transfected with GFP-tagged histone H2B were synchronized in G1/S following thymidine block, released, and then treated with PES or PES-Cl, followed by time-lapse video microscopy. This analysis revealed that the majority of PES and PES-Cl-treated cells arrested at G2/M with condensed chromosomes, but that in a percentage of cases these arrested cells appeared to transit through mitosis, resulting in defects in chromosome segregation (, grey arrows) and eventual cell death.
Treatment with PES or PES-Cl causes G2/M cell cycle arrest and inhibition of APC/C function
Previously we showed that PES was efficacious as an anti-cancer agent in Eμ-myc mice, which develop pre-B cell lymphoma (7
); this transgenic tumor model appears to be particularly sensitive to the effects of autophagy inhibition (22
). To compare the efficacy of PES and PES-Cl in lymphoma we treated Eμ-myc mice of 8 weeks of age and no evidence for cancer with dilution vehicle (n=19 mice), PES (20 mg/kg; n=20 mice) and PES-Cl (20 mg/kg; n=21 mice), administered by intra-peritoneal injection once per week for 20 weeks. Whereas all of the vehicle-treated mice succumbed to lymphoma by day 210, 35% of the PES-treated mice survived to this time point, and 71.4% of mice treated with PES-Cl survived (). The difference between PES and PES-Cl was significant (p=0.015) and the difference between vehicle and PES-Cl was highly significant (p<0.000006). To confirm that PES and PES-Cl functioned as predicted in these B cell lymphomas, we isolated tumor-containing lymph nodes of approximately equal-size from control, PES- and PES-Cl treated mice, and assessed the level of apoptosis in these tumors 24 hours after treatment. Notably, whereas there were low levels of cells positive for cleaved lamin A in the control and PES-treated mice, there was a marked increase in apoptotic cells in the PES-Cl treated mice (). These data indicate that PES-Cl functions as expected to induce programmed cell death in the tumors of treated mice.
PES-Cl significantly promotes survival in the Eμ-myc mouse model of lymphoma