Identification of the Plant Compound Geraniin as a Novel Hsp90 Inhibitor

Besides its function in normal cellular growth, the molecular chaperone heat shock protein 90 (Hsp90) binds to a large number of client proteins required for promoting cancer cell growth and/or survival. In an effort to discover new small molecules able to inhibit the Hsp90 ATPase and chaperoning activities, we screened, by a surface plasmon resonance assay, a small library including different plant polyphenols. The ellagitannin geraniin, was identified as the most promising molecule, showing a binding affinity to Hsp90α similar to that of 17-(allylamino)-17-demethoxygeldanamycin (17AGG). Geraniin was able to inhibit in vitro the Hsp90α ATPase activity in a dose−dependent manner, with an inhibitory efficiency comparable to that measured for 17-AAG. In addition, this compound compromised the chaperone activity of Hsp90α, monitored by the citrate synthase thermal induced aggregation assay. Geraniin decreased the viability of HeLa and Jurkat cell lines and caused an arrest in G2/M phase. We also proved that following exposure to different concentrations of geraniin, the level of expression of the client proteins c-Raf, pAkt, and EGFR was strongly down−regulated in both the cell lines. These results, along with the finding that geraniin did not exert any appreciable cytotoxicity on normal cells, encourage further studies on this compound as a promising chemical scaffold for the design of new Hsp90 inhibitors.


Introduction
Heat shock protein 90 (Hsp90) is a highly conserved molecular chaperone that modulates cellular homeostasis and environmental stress responses by interacting with more than 200 different proteins, known as Hsp90 client proteins, to facilitate their correct folding and biological activity. Besides assisting proper protein folding and assembly, Hsp90 is also critical to target misfolded proteins for proteolytic degradation [123]. Most of the Hsp90 client proteins are involved in cell growth, differentiation and survival, and include kinases, nuclear hormone receptors, transcription factors and other proteins associated with almost all the hallmarks of cancer [4,5]. Consistent with these diverse activities, genetic and biochemical studies have demonstrated the implication of Hsp90 in a range of diseases, also including cancer and allograft rejection [6]. Although Hsp90 is required in all cells, tumor cells are especially sensitive to Hsp90 inhibitors due to the critical role played by this chaperone in stabilizing several oncoproteins [7]. Inhibition of Hsp90 activity incapacitates simultaneously multiple client proteins, resulting in a blockade of multiple signaling pathways and, ultimately, providing a combinatorial attack to cellular oncogenic processes [8]. Because of the potential therapeutic use in multiple cancer indications, Hsp90 has emerged as an interesting target for the development of antitumor agents: thirteen new Hsp90 inhibitors are currently under evaluation at various stages of clinical trials [9].
With the view to identifying new potential Hsp90 inhibitors, we have used a surface plasmon resonance (SPR) assay to screen a small library including different phenolic compounds, such as flavonoids, tannins and coumarins ( Figure 1). Among the different flavonoids and tannins that were able to bind Hsp90, we focused on the ellagitannin geraniin (compound 54 in Figure 1), the main polyphenolic compound in Geranium thunbergii, a medicinal plant used to treat diarrhea in Japan.
Ellagitannins belong to the hydrolyzable tannin group and occur in several food plants, such as raspberries, strawberries, blackberries, pomegranate, almonds, and walnuts [17]. In vitro and in vivo studies have demonstrated various biological activities of this class of compounds, including antioxidant [18], antiviral, antimutagenic, antimicrobial, and antitumor effects [19], suggesting that the consumption of ellagitannins might be beneficial to human health. Geraniin is a typical ellagitannin because it is composed entirely of common acyl units, such as galloyl, hexahydroxydiphenoyl (HHDP), and dehydrohexahydroxydiphenoyl (DHHDP) groups. Although variousstudies of geraniin have proved its antioxidant, antitumor, and antivirus properties [20,21], its mechanism of action is still poorly characterized. Herein, we report the results of a panel of chemical and biological approaches that demonstrate that geraniin binds to Hsp90a and inhibits its ATPase activity, thus compromising the stability of some oncogenic client proteins.
Our results indicated that geraniin could represent an innovative scaffold for the design of new Hsp90 inhibitors interacting with its ATPase domain.

Materials
All the tested compounds belong to the plant-derived chemical library of the Department of Pharmacy, University of Salerno. Solvents (HPLC grade) were purchased from Romil (ROMIL Ltd, Cambridge, UK). All buffers were prepared with a Milli-Q water apparatus (Millipore, Bedford, MA, USA). Recombinant human Hsp90a was purchased at Assay Designs (Ann Arbor, MI, USA). Proteomic grade trypsin was purchased from Sigma-Aldrich

Surface Plasmon Resonance Analyses
SPR analyses were carried out according to our previously published procedures [22,23]. Briefly, SPR analyses were performed using a Biacore 3000 optical biosensor equipped with research2grade CM5 sensor chips (GE Healthcare). Using this platform, two separate recombinant Hsp90a surfaces, a BSA surface and one unmodified reference surface were prepared for simultaneous analyses. Proteins (100 mg mL 21 in 10 mM CH 3 COONa, pH 5.0) were immobilized on individual sensor chip surfaces at a flow rate of 5 mL min 21 using standard amine2coupling protocols to obtain densities of 8-12 kRU.  Polyphenols, as well as 17-AAG and hardwickiic acid (HA), were dissolved in 100% DMSO to obtain 4 mM solutions, and diluted 1:200 (v/v) in PBS (10 mM NaH 2 PO 4 , 150 mM NaCl, pH 7.4) to a final DMSO concentration of 0.5%. Compounds concentration series were prepared as twofold dilutions into running buffer: for each sample, the complete binding study was performed using a six2point concentration series, typically spanning 0.025-1 mM, and triplicate aliquots of each compound concentration were dispensed into disposable vials. Binding experiments were performed at 25uC, using a flow rate of 50 mL min 21 , with 60 s monitoring of association and 300 s monitoring of dissociation. Simple interactions were suitably fitted to a single2site bimolecular interaction model (A+B = AB), yielding a single K D . Sensorgram elaborations were performed using the BIAevaluation software provided by GE Healthcare.

ATP Hydrolysis Inhibition
The assay was performed using the Discover RX ADP HunterTM Plus Assay kit, following the manufacturer's instructions. ATPase reactions were carried out for 60 min at 40uC temperature in 100 mM Tris pH 7.4, 100 mM ATP and 40 nM Hsp90a in the presence of different concentrations of geraniin or 17-AAG. ADP generation was measured using a Perkin Elmer LS 55 fluorimeter (540 nm excitation and 620 nm emission). Fluorescence intensity values measured for Hsp90a without any compound were set at 100% of enzyme activity. The background reaction rate was measured in a reaction lacking enzyme or substrate and subtracted from the experimental rates.

Citrate Synthase Aggregation Assay
Chaperone activity was evaluated by monitoring the therma-l2induced aggregation of citrate synthase (CS) (0.075 mM), as reported elsewhere [24], in the absence or presence of a stoichiometric amount of Hsp90a and 0.3 mM ATP, and with or without a four2fold molar excess of each testing compound. Aggregation was initiated by unfolding CS incubating the protein in 40 mM HEPES-KOH, pH 7.5 at 43uC. Aggregation was monitored by measuring light scattering at right angles with the Perkin Elmer LS 55 fluorimeter in stirred and thermostatted quartz cells. Both the emission and excitation wavelengths were set at 500 nm, while the band-pass was 2 nm. Kinetics traces reported are the averages of two separate measurements.

Limited Proteolysis
Limited proteolysis experiments were performed on recombinant Hsp90a at 37uC, PBS 0.1% DMSO, using trypsin or chymotrypsin as proteolytic agents; 30 mL of a 3 mM Hsp90a solution were used for each experiment. Binary complex Hsp90a/ geraniin was formed by incubating the protein with a 5:1 molar excess of geraniin at 37uC for 15 min prior to proteolytic enzyme addition. Both Hsp90 and Hsp90/geraniin complex were digested using a 1:100 (w/w) enzyme to substrate ratio. The extent of the reactions was monitored on a time2course basis by sampling the incubation mixture after 5, 15, and 30 min of digestion. Samples were analyzed by MALDITOF/MS using a MALDI micro MX (Waters). Mass data were elaborated using the Masslynx software (Waters). Preferential hydrolysis sites on Hsp90a under different conditions were identified on the basis of the fragments released during enzymatic digestion.

Cell Culture and Treatment
Jurkat and HeLa (epithelial carcinoma) cells, obtained from Cell Bank in GMP-IST (Genova, Italy) were maintained in RPMI 1640  medium or Dulbecco's modified Eagle medium (DMEM), respectively, supplemented with 10% (v/v) FBS, 2 mM L2glutamine and antibiotics (100 U mL 21 penicillin, 100 mg mL 21 streptomycin) purchased from Invitrogen (Carslbad, CA, USA), at 37uC in humidified atmosphere with 5% CO 2 . To ensure logarithmic growth, cells were subcultured every two days. As control cells, human peripheral blood mononuclear cells (PBMC) were isolated from buffy coats of healthy donors (kindly provided by the Blood Center of the Hospital of Battipaglia, Salerno, Italy) by using standard Ficoll2Hypaque gradients. Freshly isolated PBMC contained 92.863.1% live cells. Proliferation of PBMC was induced by phytohemagglutinin (PHA) (10 mg mL 21 ). Tumor and PBMC cells were treated with different concentrations of geraniin or 17-AAG to evaluate their cytotoxic effects. Stock solutions of geraniin (50 mM in DMSO) and 17-AAG (10 mM in DMSO) were stored a 220uC in the dark and diluted just before addition to the sterile culture medium. In all the experiments, the final concentration of DMSO was 0.10% (v/v).

Cell Proliferation and Viability
Jurkat (2610 4/ well) and HeLa (5000/well) cells were seeded in triplicate in 96 well2plates and incubated for 24 h in the absence or presence of different concentrations of geraniin (between 0,5 mM to 50 mM) or 17-AAG (between 0,5 nM to 10 mM).
The number of viable cells was determined by using a [3-4,5dimethyldiazol-2-yl]-2,5-diphenyl tetrazolium bromide (MTT, Sigma-Aldrich) conversion assay, according to the method described by Mosmann [25]. Briefly, following the treatment, 25 mL of MTT (5 mg/mL in PBS) was added and the cells were incubated for additional 3 h at 37uC. Thereafter, cells were lysed and suspended with 100 mL of buffer containing 50% (v/v) N,Ndimethylformamide, 20% SDS (pH 4.5). The absorbance was  Cell population growth inhibition was tested by cytometric counting (trypan blu exclusion). IC 50 values were calculated from cell viability dose2response curves and defined as the concentration resulting in 50% inhibition of cell survival, compared to control cells treated with 0.10% DMSO.

Analysis of Cell Cycle and Hypodiploidy by Flow Cytometry
Cell DNA content was measured by propidium iodide (PI) incorporation into permeabilized cells, as described by Nicoletti et al. [26]. Briefly, the cells were harvested after geraniin or 17-AAG treatments were washed with cold PBS and incubated with a PI solution (0.1% sodium citrate, 0.1% Triton X-100 and 50 mg mL 21 of prodium iodide, Sigma-Aldrich, 10 mg/ml Rnase A) for 30 min at room temperature. Data from 10.000 events per sample were collected by a FACScalibur flow cytometer (Becton Dickinson, San José, CA) and cellular debris was excluded from analysis by raising the forward scatter threshold. The percentage of cells in the sub G 0 /G 1 phase, apoptotic fraction, was quantified using CellQuest software (Becton Dickinson). The distribution of cells in G 0 /G 1 , S, G 2 /M phases was determined using ModFit LT cell cycle analysis software (Becton Dickinson). Results were expressed as a mean 6 SD of three experiments performed in triplicate.

Western Blot Analyses
Treated cells were harvested and disrupted by freeze2thawing in RIPA buffer (50 mM Hepes, 10 mM EDTA, 150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS, pH 7.4), supplemented with protease inhibitors cocktail (Sigma-Aldrich). Cell debris was removed by centrifugation at 4uC and the supernatant protein concentration was determined, according to the Bio-Rad Protein assay (Biorad Laboratories, CA, USA). Total proteins (30 mg) were separated by SDS-PAGE under denatured reducing conditions. Separated proteins were then transferred to nitrocellulose membranes and blocked for 3 h with a solution containing 3% BSA in 50 mM Tris, 200 mM NaCl, 0.1% Tween 20, before incubation at 4uC overnight with the different primary antibodies (diluted 1:1000) against specific proteins. After washing, the membranes were incubated with an appropriate peroxidaseconjugate secondary antibodies at room temperature (1 h). Immunoreactive protein bands were detected by enhanced chemiluminescence reagent (ECL, Rockford, USA), according to the manufacturer's instructions. Quantitative densitometry analyses were performed using a Gel Doc 2000 system (Biorad Laboratories, CA, USA).

Statistical Analysis
All the reported data represent the mean 6 standard deviation (SD) of at least two independent experiments, performed in triplicate. Data were statistically compared by t-test; in that aim we checked for normal distribution of data and comparable variance among the groups compared. The statistical significance of DNA content of hypodiploid nuclei was examined in the two-way analysis of variance (ANOVA) with Bonferroni post-test analysis.

Screening for Ligand-HSP90 Complex Formation by Surface Plasmon Resonances
Putative interactions of different plant-derived phenolic compounds with Hsp90a were evaluated by a SPR-based approach [27]. Natural compounds belonging to different polyphenol classes (38 flavonoids, 12 coumarins and 4 tannins) were assayed for their binding affinity for the chaperone (Figure 1). The geldanamycin derivative 17-AAG, one of the best characterized Hsp90 inhibitors, was used as a positive control, whereas the clerodane diterpene hardwickiic acid (HA) (Figure S1), was selected as a negative control on the basis of our previous observations [28]. In Figure 2 some of the obtained sensorgrams are reported. Eight out of the 54 tested compounds interacted with immobilized Hsp90a, as inferred by the concentration-dependent responses, and by the clearly discernible exponential curves, during the association and dissociation phases. None of the tested coumarins was found to interact with the protein, whereas several of the tested flavonoids and tannins were able to bind to Hsp90a.
To measure the kinetic and thermodynamic parameters for each complex formation, the sensorgrams were fitted to a single2site bimolecular interaction model (A+B = AB); each constant was calculated fitting at least 12 curves, obtained by injecting the different compounds three times at four different concentrations, ranging from 0.025 mM to 1 mM (Table 1). Among the tested compounds, geraniin showed the highest affinity for the immobilized Hsp90a (Figure 2A). A K D of 415627 nM was measured for the Hsp90a/geraniin complex, comparable to that measured for the Hsp90a/17-AAG complex (388689 nM).

Inhibition of Hsp90a by Geraniin
The elevated affinity towards Hsp90a measured for geraniin, prompted us to investigate further the ability of this ellagitannin to affect different biological activities of the chaperone. Firstly, we evaluated the effects of different concentration of geraniin on the ATPase activity of Hsp90a compared to those produced by 17-AAG or HA, selected as a positive and a negative control, respectively. Geraniin was found to inhibit Hsp90a ATPase activity in a dose2dependent manner, with an inhibitory efficiency similar to that measured for 17-AAG (Figure 3 A).
The next question we addressed was to establish whether geraniin2dependent ATPase inhibition could compromise the chaperone activity of Hsp90a, which was evaluated by monitoring the citrate synthase (CS) thermal induced aggregation in the presence of Hsp90a, with or without geraniin. Upon incubation at elevated temperatures, CS underwent partial unfolding resulting in a quantitative protein aggregation; the presence of stoichiometric amounts of Hsp90a changed the aggregation kinetics significantly. When a four2fold molar excess of geraniin was added to the reaction, the curve slope clearly increased, indicating that geraniin, interacting with Hsp90a, affects its affinity towards denatured CS (Figure 3 B). Anti2chaperone activity of the tannin resembled that observed performing the same experiment using a four2fold molar excess of 17-AAG, whereas the addition of the same amount of HA did not perturb the CS+Hsp90a curve.

Study of Hsp90a/geraniin Interaction
In an effort to identify the Hsp90a region involved in geraniin binding, a limited proteolysis2mass spectrometry2based strategy was used for the structural analysis of the Hsp90a/geraniin complex. This approach is based on the evidence that exposed, weakly structured, and flexible regions of a protein can be recognized by a proteolytic enzyme [29,30]. The proteolytic patterns obtained in the experiments performed on Hsp90a and on the Hsp90a/geraniin complex, using trypsin or chymotrypsin as proteolytic agents, are summarized in Figure 4. Comparison of the differential proteolytic pattern obtained from the digestion of native Hsp90a or of the Hsp90a/geraniin complex confirmed a direct interaction between geraniin and the chaperone. In addition, we observed that peptide bonds following Arg86, Lys237, Lys293, and Tyr380, preferential cleavage sites of the native chaperone in the absence of geraniin, were protected in the complex. Conversely, following interaction with geraniin, peptide bonds located at the level of Lys488 and Lys512 became susceptible to enzymatic hydrolysis. These data suggest that binding of geraniin to Hsp90a induced significant conformational changes of its three2dimensional structure. Specifically, the observed overall protection of the N2terminal region from proteolysis indicated that this is the protein portion g004mainly involved in interaction with geraniin.

Effect of Geraniin on Cancer Cell Viability
On the basis of the ability of geraniin to bind and inhibit HSP90a activity, we evaluated its potential anti2proliferative or cytotoxic activity in HeLa (epithelial carcinoma) and  [31] and Bisht et al. [32]. Minimal cytotoxic effects, evaluated by tryptan blue dying, were observed in control PHA-stimulated proliferating PBMC only at high doses of geraniin or 17-AAG ( Figure 5).

Induction of Cell Cycle Arrest and Cell Death in Cancer Cell Lines by Geraniin
To establish the mechanism of action underlying the inhibition of cancer cell viability caused by geraniin treatment, the cell cycle progression of cancer cells and PHA-stimulated PBMC were analyzed by flow cytometry.
The Jurkat and HeLa cells were incubated for 24 h with concentrations close to the IC 50 values of geraniin (0.7 and 5 mM, respectively) and 17-AAG (10 and 0.2 mM, respectively), The effects of geraniin and 17-AAG-treatments on the distribution of non-apoptotic cells differed significantly. In both tumor cell lines, treatment with geraniin caused a G 2 /M arrest (Figure 6 A, B), whereas 17-AAG induced a G 1 and G 2 /M arrest (Figure 6 A, B), consistent with previous published results [31,33,34]. Moreover, following geraniin treatment the percentage of Jurkat and Hela cells with hypodiploid nuclei was 9.7561.02% and 16.260.7%, respectively, whereas following 17-AGG treatment these values were 6.0560.4% and 11.660.5%, respectively. These results indicate a similar pro-apoptotic effect of the two compounds on both the cell lines. Neither geraniin nor 17-AAG caused any prodeath or cytostatic effects in PHA-stimulated proliferating PBMC, since the levels of hypodiploidy/necrosis in geraniin and 17-AAG treated PBMC were 2.661.3% and 1.860.6%, respectively, very similar to those observed in untreated cells (1.560.8%).

Down2regulation of Hsp90 Client Proteins by Geraniin Treatment
Geraniin was tested for its effects on the intracellular levels of Hsp70, Hsp90 and Hsp90 client proteins. HeLa and Jurkat cell lines were treated with a geraniin concentration corresponding to their IC 50 value and at additional lower and higher concentrations ( Figure 7). These treatments caused a dose-dependent decrease in the level of the client proteins c-Raf, pAkt and EGFR, whereas the level of Akt was unaffected. Interestingly, the level of Hsp70 expression was enhanced in both tumor cell lines, in a dosedependent fashion, with 3.560.12-fold increase in Jurkat cells upon 1 mM treatment and 2.760.09 fold increase in HeLa cells upon 10 mM geraniin treatment.

Discussion
Inhibition of HSP90 has received significant attention in cancer research due to its ability to retard or block tumor growth [6]. In this respect, Hsp90 plays a critical role in the maintenance of multiple oncogenic pathways and is required to maintain the folding, stability and functionally active conformation of many aberrant oncoproteins [7]. In healthy cells, Hsp90 is involved in dynamic, low-affinity interactions with a plethora of proteins during folding and maturation; however, in tumor cells, it assists folding of dysregulated oncoproteins and sustains their aberrant activity [4,5]. Given the diversity of the Hsp90 client proteins involved in critical cellular pathways and processes, inhibition of Hsp90 was predicted to have efficacy in a broad variety of human tumors. However, although several Hsp90 inhibitors have thus far entered into clinical trials, the development of Hsp90 inhibitors has encountered difficulties, including drug solubility and hepatic toxicity [35].
Based on the notion that natural products are compounds pre2optimized by evolution to act against specific biological targets, we performed a structure2based screening of different plant2derived polyphenols to identify new potential Hsp90 inhibitors. By SPR analysis, the tannin geraniin was identified as an efficient ligand of Hsp90a, showing a high affinity for this chaperone, similar to that found for 17-AAG and derivates. The comparison of the HSP90a proteolytic patterns in the presence or absence of geraniin indicated that this compound binds at the Nterminus of the chaperone, as reported for several Hsp90 inhibitors [10]. Through targeting the ATP-binding site of the N-terminal domain, the inhibitors probably prevent Hsp90 from forming a closed N-terminal dimeric state and, consequently, alter the chaperone activity of the molecule [36]. This mechanism was proved for geraniin, which was able to reduce Hsp90 chaperone activity more efficiently than 17-AAG.
The impaired ATPase and chaperone activities of Hsp90, caused by geraniin binding, induce cytotoxic effects in the tumor cell lines tested, with a large percentage of cells containing hypodiploid DNA. These results are in agreement with those reported in human melanoma cells, where geraniin treatment caused apoptosis, through up-regulation of the Fas ligand expression, the activation of caspase-8, the cleavage of Bid, and the induction of cytochrome c release from mitochondria to the cytosol [21]. Our results also revealed that geraniin induces cell cycle arrest in the G 2 /M phase in both cancer cell lines, whereas 17-AAG-treated cells accumulated in G 1 phase of cell cycle.
Moreover, the geraniin-dependent inhibition of Hsp90a chaperone activity caused a dose-dependent decrease in the level of the oncogenic proteins c-Raf, pAkt and EGFR, further supporting the potential of this compound to interfere with tumor progression. As already reported in other cellular systems, Hsp90 inhibition by geraniin was associated to the up-regulation of Hsp70, as a compensatory mechanism aiding the proper folding and assem-bling of oncogenic client proteins. Although it is still unclear if the induction of Hsp70 following Hsp90 inhibition attenuates the cytotoxic effects of Hsp90 inhibitors, Hsp70 might also be considered as a pharmacodynamic marker for drug response of HSP90 inhibition [37].
In conclusion, the results presented here, along with the finding that geraniin did not exert any appreciable cytotoxicity in normal cells, encourages further studies on this compound as a promising chemical scaffold for the design of new Hsp90 inhibitors. Figure S1 Structure of hardwickiic acid. (DOCX)