Induction of cancer-specific cytotoxicity towards human prostate and skin cells using quercetin and ultrasound

Bioflavonoids, such as quercetin, have recently emerged as a new class of chemotherapeutic drugs for the treatment of various cancer types, but are marred by their low potency and poor selectivity. We report that a short application of low-frequency ultrasound selectively sensitises prostate and skin cancer cells against quercetin. Pretreatment of cells with ultrasound (20 kHz, 2 W cm−2, 60 s) selectively induced cytotoxicity in skin and prostate cancer cells, while having minimal effect on corresponding normal cell lines. About 90% of the viable skin cancer cell population was lost within 48 h after ultrasound-quercetin (50 μM) treatment. Ultrasound reduced the LC50 of quercetin for skin cancer cells by almost 80-fold, while showing no effect on LC50 for nonmalignant skin cells.

Therapeutic selectivity plays a crucial role in determining the success of chemotherapy. Some of the current targeted therapies attempt to localise drugs to cancer cells based on overexpression of epidermal growth factor receptors (EGFR) (Mendelsohn and Baselga, 2000) or angiogenesis (Carter, 2001). Antibodies, inhibitors, antisense therapy and gene therapy are also among a few strategies that have gained momentum (Guillemard and Saragovi, 2004). Many of these strategies have now reached clinical trials; however, these methods are still limited by issues including low potency, delivery complications, multi-drug resistance, side effects, collateral damage (Tattersall and Clarke, 2003) or incomplete success (Lynch et al, 2004).
In an attempt to develop a targeted chemotherapeutic strategy, we propose the use of bioflavonoids, which are common dietary supplements, in conjunction with low-frequency ultrasound. Quercetin, a major bioflavonoid in human diet, has been identified as a chemotherapeutic agent for the treatment of breast cancer (Singhal et al, 1995;Choi et al, 2001), colon cancer (Salucci et al, 2002), ovarian cancer (Chan et al, 2003) and prostate cancer (Knowles et al, 2000;Nakanoma et al, 2001;Kobayashi et al, 2002). Antiproliferative action of quercetin is hypothesised to be mediated by attenuating phosphorylation of activated hsp transcription factor (hsf), shortly after its trimerisation (Nagai et al, 1995;Lee et al, 1998), thereby resulting in increased susceptibility of hsf to proteolytic degradation and as a consequence inhibiting all downstream events, including hsp expression (Li et al, 1999). Since hsps are constitutively overexpressed in many tumours (Jaattela, 1999), inhibition of hsps is an attractive chemotherapeutic strategy. hsps form a complex with mutant p53 protein (mp53), thereby prolonging the half-life of malignant mp53 and allowing tumour cells to bypass the normal mechanism of cell cycle arrest (Selkirk et al, 1996).
In spite of its therapeutic benefits, utilisation of quercetin in clinical applications has been limited by low potency and poor specificity. Additionally, it is difficult to sustain therapeutic quercetin concentrations in blood by oral ingestion (Lamson and Bringall, 2000).
Here, through in vitro studies, we demonstrate for the first time, using two pairs of normal and cancer cells (human skin fibroblast and human prostate epithelial cells), that ultrasound selectively sensitises cancer cells against quercetin. LC 50 of quercetin for skin cancer cells is selectively decreased by almost 80-fold by a short pretreatment with ultrasound.

Cell culture
Normal and cancer cells derived from prostrate and skin tissues were investigated in this study. DU145 prostate cancer cells were provided by Dr L Wilson at UC Santa Barbara, CA, USA. Nonmalignant prostrate normal cells (Catalog No. CRL-11609), nonmalignant skin cells (Catalog No. CRL-7761) and skin cancer cells  were obtained from the American Type Culture Collection (ATCC, Rockville, MD, USA). All cells were grown as monolayers and were kept in a 5% CO 2 environment at 371C. Cell cultures were maintained in Dulbecco's modified Eagle's medium (DMEM) with glucose (1 g l À1 ), NaHCO 3 (3.7 g l À1 ), L-glutamine (2 mM), nonessential amino acids (0.0815 g l À1 ) and 10% FBS. Antibiotic -antimycotic cocktail (Catalog No. 15240-062, Gibco, Invitrogen Corporation, Carlsbad, CA, USA), at a final concentration of 100 U ml À1 of penicillin, 100 mg ml À1 of streptomycin and 0.25 mg ml À1 of amphotericin B, was added to all cultures. Cells were harvested at a concentration of about 3 Â 10 5 cells ml À1 , by washing with versene (NaCl -8 g l À1 , KCl -0.2 g l À1 , NaH 2 PO 4 -1.15 g l À1 , K 2 HPO 4 -0.2 g l À1 , Na 2 -EDTA -0.2 g l À1 in distilled water with pH adjusted to 7.2) followed by 2 -3-min digestion with trypsin/EDTA (0.25%/0.02%).

Ultrasound application and quercetin treatment
Aliquots of 2 ml cell suspension (3 Â 10 5 cells ml À1 ) were plated in 12-well plates. Ultrasound was applied to cells prior to quercetin exposure. Other sequences of ultrasound and quercetin application were not studied and may yield different results. Ultrasound was applied at a frequency of 20 kHz and an intensity of 2 W cm À2 (Sonics and Materials, Danbury, CT, model VCX 400). Ultrasound intensity was determined using a hydrophone (Tezel et al, 2002). Ultrasound was applied at room temperature for 60 s by directly immersing the transducer half-way down the meniscus in the well. The temperature of the cell suspension was recorded to ensure that no significant elevation of temperature (o51C) was observed. 20 ml of quercetin solution prepared in dimethyl sulphoxide (DMSO) was immediately added after each sonication to the wells to achieve a final quercetin concentration of 0 -50 mM. Cells were analysed for viability at the end of 48 h. In experiments involving multiple exposures of ultrasound, the adhering monolayer of cells in the wells was washed with the procedure described above. The washed cell suspension from each well was made up to 2 ml by adding DMEM and subsequently sonicated by plating it in a new 12-well plate. Additional exposures were performed in some experiments at the end of 48 and 72 h and cell viability assessed at the end of 96 h. Cell viability prior and during experimentation was determined using Trypan blue exclusion under a light microscope.

Gel electrophoresis and Western blots
Malignant and nonmalignant skin cells were treated with ultrasound and quercetin (50 mM) and their hsp content was assessed using Western blots after 48 h. Specifically, the culture medium was removed after the treatment and the wells were washed three times with PBS to remove the serum and dead cells. The removed culture medium was mixed with PBS and centrifuged for 10 min to recover the dead cell pellet. 200 ml of lysis buffer containing 20 mM Tris (pH ¼ 7.4, Sigma), 150 mM NaCl (Sigma, St Louis, MO, USA), 1% Triton X-100 nonionic detergent buffer (ICN Biomedicals, Aurora, OH, USA) with 1 mM pepstatin, leupeptin and PMSF (Sigma Chemicals, St Louis, MO, USA) was added to the well and the dead cell pellet separately to obtain respective cell lysates. The cell lysates were then centrifuged for 10 min and the supernatant protein extracts were used for electrophoresis measurements. Electrophoresis samples were prepared on a cell number basis by mixing the two protein extracts and using the previously obtained cell density data, such that all the samples contained proteins from an equal number of starting cell population (roughly 6 Â 10 5 cells per well). Heat shock protein 72 (hsp72) mouse monoclonal IgG antibody (Catalog No: SPA-810, Stressgen, Victoria, BC, Canada) was used to measure the induction of inducible form of heat shock protein 70 family (hsp70), viz., hsp72. Anti-mouse IgG horseradish peroxidaseconjugated antibody (Amersham Pharmacia, UK) was used as the secondary antibody. Dilutions of 1 : 1000 for the primary antibody and 1 : 5000 for the secondary antibody were typically used. Images were captured using X-ray films by the ECL Western blotting detection kit (Amersham Pharmacia, UK) and quantified by densitometry by using the software ImageQuantt TL (Amersham Biosciences, UK).

RESULTS
Cytotoxic effects of quercetin and ultrasound were assessed using two pairs of normal and cancer cell lines (human skin fibroblast and human prostate epithelial cells). The pair of skin cells was obtained from the same donor and differed from each other only in terms of malignancy. Cells were incubated with quercetin (0 -50 mM) with or without prior exposure to ultrasound (20 kHz, 2 W cm À2 , 60 s). A strong concentration-dependent cytoxicity was observed in skin cancer cells for the combined ultrasound and quercetin treatment ( Figure 1A, closed squares), but not in nonmalignant skin cells ( Figure 1A, open squares, Po0.001 for quercetin ¼ 50 mM). About 90% of viable population of skin cancer cells was lost in 48 h after ultrasound and quercetin (50 mM) treatment ( Figure 1A, closed squares). In the absence of ultrasound, quercetin showed no significant effect on either malignant or nonmalignant skin cells after 48 h incubation ( Figure 1A, closed circles and open circles, respectively; P40.90 for 50 mM quercetin concentration). Similar results were obtained for prostate cancer and normal cells (data not shown, Po0.05 for ultrasound, followed by quercetin (50 mM) treatment).
Enhancement in quercetin cytotoxicity towards skin cancer cells due to ultrasound exposure (defined as the fraction of cells killed with ultrasound exposure divided by the fraction of cells killed without the use of ultrasound at the same quercetin concentration) increased with increasing quercetin concentrations ( Figure 1B, closed circles; Po0.02 for 50 mM quercetin concentration). Ultrasound had no effect on quercetin toxicity towards nonmalignant skin cells ( Figure 1B, open circles). Tumour selectivity (defined as the number of dead cancer cells divided by number of total dead cells; for equal number of normal and cancer cells treated) as high as 82% was observed. Ultrasound alone had no effect on cell viability of either type of skin cells (viability of 9675% for both types of skin cells). The effect of ultrasound on quercetin-induced cytotoxicity is clearly due to the synergistic activity between the two and not due to the direct effect of ultrasound on cell viability.
The LC 50 (quercetin concentration necessary to reduce cell viability by 50%) for skin cancer cells was also significantly reduced by ultrasound pre-exposure (Figure 2A: filled bar -skin cancer cells, open bar -nonmalignant skin cells). In the absence of ultrasound, LC 50 of skin cancer cells was 98 mM. However, a single exposure to ultrasound for 60 s reduced LC 50 to about 9 mM and two further applications of ultrasound 24 h apart reduced LC 50 by 80-fold to about 1.2 mM. LC 50 of nonmalignant skin cells was not significantly altered (450 mM in all cases). To assess the specificity of synergy between quercetin and ultrasound, similar experiments were performed using another drug geldanamycin (a drug known to interfere with hsp90 cycle) and ultrasound. Geldanamycin alone exhibited cytotoxicity consistent with prior reports (Gan et al, 1998); however, no synergistic effect with ultrasound was found.
Selective effect of quercetin and ultrasound on skin cancer cells was accompanied by an effect on the inducible form of hsp70 (hsp72), which has long been known to confer protection to cells under severe stress (Kiang and Tsokos, 1998) and has been identified as a target of quercetin (Hansen et al, 1997). Skin cancer cells exhibited higher concentrations of hsp72 (1.8-fold, Po0.05) compared to corresponding nonmalignant cells ( Figure 2B, lane 4 vs lane 1). This observation is consistent with the generally accepted notion that cancer cells overexpress heat shock proteins (Jaattela, 1999;Jolly and Morimoto, 2000;Nylandsted et al, 2000). Ultrasound alone or ultrasound þ quercetin had minimal effect on cellular hsp72 in nonmalignant skin fibroblasts ( Figure 2B, 12% decrease for ultrasound alone, P40.90, and 22% decrease for ultrasound þ 50 mM quercetin, P40.76). A combination of ultrasound and quercetin (50 mM) induced a significant decrease in hsp72 concentration in skin cancer cells (72% decrease, Po0.01, Figure 2B). In the same cells, quercetin alone decreased hsp72 concentration by 31.4% ( Figure 2B, lane 7) and ultrasound alone decreased hsp72 concentration by 31.7% ( Figure 2B, lane 5).

DISCUSSION
The effects reported in Figures 1 and 2 are unlikely to originate from enhanced transport of quercetin by ultrasound. Quercetin is a small and slightly lipophilic molecule (molecular weight ¼ 302 Da, octanol-water partition coefficient, K o/w B1.270.13 (Brown et al, 1998 and is expected to diffuse across cell membranes at a high rate. Intracellular quercetin concentrations are expected to be in equilibrium with extracellular concentration even without ultrasound. Moreover, under the conditions used for the experiments in this study, a moderate degree of cavitation was observed (data not shown) and was not strong enough to induce significant membrane permeabilisation (as judged by lack of intracellular uptake of calcein under identical conditions), and hence incapable of pushing quercetin into cells. It is not clear at this stage as to how ultrasound selectively sensitises cancer cells against quercetin. It is possible that the selectivity originates from the effect of quercetin as well as  Quercetin alone has an LC 50 of about 98 mM for skin cancer as well as skin normal cells. A single application of ultrasound (20 kHz, 2 W cm À2 , 60 s) prior to incubation with quercetin substantially decreased LC 50 for skin cancer cells to 9 mM, but only moderately affected LC 50 for skin normal cells (86 mM). Application of three ultrasound pulses (prior to quercetin application, at 48 h and 72 h after the first application) further reduced LC 50 for skin cancer cells to 1.2 mM. Application of two pulses had no significant effect on LC 50 for skin normal cells. (B) Cellular concentrations of hsp72 in nonmalignant skin cells (first three lanes) and skin cancer cells (last four lanes). The first lane shows hsp72 concentration in nonmalignant skin cells prior to exposure to ultrasound or quercetin (control). The second lane shows hsp72 concentration in nonmalignant cells exposed to ultrasound alone. The third lane shows hsp72 concentration in nonmalignant skin cells exposed to ultrasound, followed by 50 mM quercetin for 48 h. The fourth lane shows control samples for skin cancer cells. hsp72 concentration in skin cancer cells is higher than that in skin normal cells. The fifth lane represents skin cancer cells exposed to ultrasound alone (20 kHz, 2 W cm À2 , 60 s). The sixth lane shows hsp72 concentration in skin cancer cells exposed to ultrasound and subsequently to 50 mM quercetin for 48 h. The seventh lane represents cells exposed to 50 mM quercetin alone for 48 h.
ultrasound on stress response. Quercetin has been shown to interfere with the stress response and inhibit hsp72 both at protein and mRNA levels in certain cells (Hosokawa et al, 1990;Elia and Santoro, 1994;Jakubowicz-Gil et al, 2002). Ultrasound, being a mild stress, may also induce a stress response in mammalian cells. It is possible that the interplay between the effect of ultrasound and quercetin on hsp cycle leads to selective sensitisation of cancer cells against ultrasound. Whether or not other mild stresses, for example, hypoxia, yield similar results remains to be seen. Since elevated levels of hsps are broadly associated with survival of cancer cells (Burdon, 1987;Lasunskaia et al, 1997), chemotherapeutic strategies that target hsps are attractive. With further studies focused on in vivo testing and mechanistic understanding, this technique may provide a potential treatment for the treatment of cancer, especially skin cancer.