Dietary flavonoid fisetin increases abundance of high-molecular-mass hyaluronan conferring resistance to prostate oncogenesis
Dietary flavonoid fisetin increases abundance of high-molecular-mass hyaluronan conferring resistance to prostate oncogenesis
Rahul K.Lal 1
Deeba N.Syed 1
Mohammad Imran Khan 1
Vaqar M.Adhami 1
Yuansheng Gong 0
John A.Lucey 0
Hasan Mukhta r 0 1
0 Department of Food Science, University of Wisconsin , Madison, WI 53706-1510 , USA
1 Department of Dermatology
We and others have shown previously that fisetin, a plant flavonoid, has therapeutic potential against many cancer types. Here, we examined the probable mechanism of its action in prostate cancer (PCa) using a global metabolomics approach. HPLC-ESI-MS analysis of tumor xenografts from fisetin-treated animals identified several metabolic targets with hyaluronan (HA) as the most affected. Efficacy of fisetin on HA was then evaluatinedvitro and alsoin vivo in the transgenic TRAMP mouse model of PCa. Size exclusion chromatography-multiangle laser light scattering (SEC-MALS) was performed to analyze the molar mass (Mw) distribution of HA. Fisetin treatment downregulated intracellular and secreted HA levels both in vitro and in vivo. Fisetin inhibited HA synthesis and degradation enzymes, which led to cessation of HA synthesis and also repressed the degradation of the available high-molecular-mass (HMM)-HA. SEC-MALS analysis of intact HA fragment size revealed that cells and animals have more abundance of HMM-HA and less of low-molecular-mass (LMM)-HA upon fisetin treatment. Elevated HA levels have been shown to be associated with disease progression in certain cancer types. Biological responses triggered by HA mainly depend on the HA polymer length where HMM-HA represses mitogenic signaling and has anti-inflammatory properties whereas LMM-HA promotes proliferation and inflammation. SimilarMlyw, analysis of secreted HA fragment size revealed less HMM-HA is secreted that allowed more HMM-HA to be retained within the cells and tissues. Our findings establish that fisetin is an effective, non-toxic, potent HA synthesis inhibitor, which increases abundance of antiangiogenic HMM-HA and could be used for the management of PCa.
Tumor microenvironment plays a major role during prostate Increased HA levels in various cancers including PCa cor-re
cancer (PCa) development. Interactions between the microen-vi late with malignant progression and poor survival. These high
ronment and cancer cells are important for migration, met-as HA levels are associated with high expression of HAS and low
tasis and survival 1(). A major component of the extracellular expression of HYALs (
matrix in the tumor microenvironment is glycosaminoglycan, HA regulates several cellular function7s,8)(. The native,
hyaluronan (HA). HA is a non-sulfated, linear polymer co-m high-molecular-mass (HMM)-HA (~107 Da) can be broken down
posed of repeating disaccharides of glucuronic acid (GlcUA) and into smaller fragments in response to glycosidase activity and
N-acetyl glucosamine units (GlcNAc)2)(. HA is synthesized at environmental parameters such as pH and reactive oxygen
the cell surface by the membrane-bound enzyme HA synthase species (ROS) (
). High- and low-molecular-weight forms of
(HAS). Three HAS enzymes HAS1, HAS2 and HAS3 are involved HA provoke distinct anti-inflammatory and proinflammatory
in the production of HA, whereas HA degradation is controlled effects upon binding to CD44 and can deliver either proliferative
by hyaluronidases (HYALs), six of which have been identified3(). or antiproliferative signals in various cell type1s3,(14). Studies
have shown that HMM-HA exhibited antiangiogenic effects both
in vitro and in vivo (
). In contrary, the lower-molecular-mass
(LMM)-HA stimulated cell proliferation and exhibited proinfl-am
matory and proangiogenic effects in various studies14(
In recent years, considerable progress has been made in
identifying a specific inhibitor for HA synthesis and such inhi-bi
VA) in 2013, whereas PC3 (CRL-1435) and DU145 (HTB-81) cell lines were
obtained in 2012. ATCC ensures cell lines authenticity using morphology,
karyotyping and PCR-based approaches, which include assays to detect
cytochrome C oxidase I gene (COI analysis) to rule out interspecies c-on
tamination and short tandem repeat DNA profiling to rule out intra-spe
cies contamination as tabulated inSupplementary Table 7, available at
Carcinogenesis Online. Cell lines were immediately resuscitated upon
receipt and frozen in aliquots in liquid nitrogen. Once thawed, early-pas
sage cells were cultured within 3 months from a frozen vial of the same
batch of cells. Cells were routinely tested to ensure there was no m-yco
plasma contamination (MycoAlert Mycoplasma Detection Kit, Lonza).
RWPE1, NB11 and NB26 cells were cultured in human keratinocyte growth
supplement from Thermo-Fisher Scientific (Grand Island, NE),
supplemented with HKGS and 1% penicillin–streptomycin. PC3 and DU145 cells
were cultured in RPMI 1640 from Gibco (Carlsbad, CA), with 10% fetal
bovine serum and 1% penicillin–streptomycin. The cells were then inc-u
bated at 37°C with 5% C O2in a humid environment. For time-dependent
studies, cells (70% confluent) were treated with fisetin dissolved in d-ime
thyl sulfoxide (DMSO) (0–40μ M) for specified time points at 37°C in media
and harvested for further studies.
BrdU cell proliferation assay
The BrdU assay from Cell Signaling (Danvers, MA) was used as per the ma-n
ufacturer’s instructions to measure cell proliferation. Absorbance was read
at 450nm using a Synergy 2 multidetection microplate reader (BioTek, VT).
Aqueous and organic metabolites were isolated from control and
fisetintor would not only help elucidate the function of HA, but also treated tumor xenograft tissues using a methanol/water and dichlorom-eth
can be used for treatment of diseases with elevated HA levels ane/water extraction as described previously36(). Metabolite extracts
). 4-Methylumbelliferone (4-MU) is one of the well-known (10 µl) were then diluted to 5µ0l with 0.1% formic acid in water. Untargeted
and potent HA synthesis inhibitor, and its role has been explored metabolomics using HPLC–ESI–MS was performed with a 10 µl injection on
extensively in various cancers1(
), but the IC50 of 4-MU to an Agilent 6210 ESI-TOF mass spectrometer with Agilent 1200 series HPLC
inhibit HA synthesis in prostate and other cancers is as high as (Santa Clara, CA). Solvents were 0.1% formic acid in water (A) and 0.1%-for
400 µM (~70 µg/ml) (
). Therefore, identifying non-toxic inhib-i mic acid in acetonitrile (B). The HPLC column was an Agilent Zorbax SB C18
tors that effectively suppress HAS levels and its production will 1.8 µm, 2.1 mm i.d. × 50 mm length (Santa Clara, CA). Metabolites were g-ra
be important. Generally considered as non-toxic, dietary fl-avo dient-eluted at a flow rate of 0.m25l/min starting at 2% B with a m1in hold,
noids act as key modulators of signaling pathways and are the-re ramping to 50% B at 35min, then to 95% B at 40min, returning to 2% B at
42 min and re-equilibrating at 2% B for 1m8in. Electrospray was performed
fore considered desirable chemopreventive agents27(,28). Fisetin, in positive-ion (+) mode at 3.6kV and fragmentor at 130V. Metabolite dete-c
found in many fruits and vegetables, belongs to the flavonol -sub tion was performed over the m/z range 50–1700 by summing 10013
trangroup of flavonoids and has shown potential against PC2a9(–34). sients/scan (0.89 scans/s). Blank injections (10µl ) of 20% MeOH and 0.08%
In this study, we identified HA as a unique target of fisetin in formic acid in water were made between sample injections.
PCa cells, tumor xenografts and TRAMP mouse model. We utilized Raw data acquired using HPLC–ESI–MS system were processed by
SEC-MALS for a quick, accurate and quantitative determination Xcalibur software that provides an appropriate format for further data
of different molecular masses of HA bothin vitro and in vivo. analysis. The structured data files were then uploaded on XCMS Online
(https://xcmsonline.scripps.edu), a high-quality cloud-based platform
linked with METLIN h(ttps://metlin.scripps.edu), to facilitate metabolite
Materials and methods identification using a repository of metabolite and tandem mass spe-c
trometry information of known compounds.
TRAMP animals were obtained as described previously 3(
and care of the animals was approved by the University of Wisconsin’s
Research Animal Resource Committee in accordance with the NIH
Guidelines for the Care and Use of Laboratory Animals.
Fisetin was purchased from Sigma Chemical Co. (St Louis, MO). Antibodies In vivo tumor xenograft model
were obtained from Cell Signaling Technology (Danvers, MA), Abcam Seven- to eight-week-old athymicn(u/nu) male nude mice (Harlan) were
(Cambridge, MA) and Santa Cruz Biotechnology (Dallas, TX), and a list xenografted as described previously3(0). We selected unique PCa cells
is provided in Supplementary Table 1, available atCarcinogenesis Online. NB11 and NB26 for determining thein vivo effects of fisetin due to their
Quantitative PCR (qPCR) primers were synthesized and obtained from DNA ability to form rapid and reproducible tumors. After injected, 12 animals
synthesis laboratory at the University of Wisconsin Biotechnology Center. were then randomly divided into 2 groups with 6 animals each. The
The sequences of oligonucleotides used are listed iSnupplementary Table 2, first group of animals received 30μ l i.p. injection of phosphate-buffered
available atCarcinogenesis Online. Commercially available HA oligos (HA60K-1 saline (PBS)/DMSO (1:1) and served as control. The animals of second
and HA200K-1) were purchased from Lifecore Biomedical (Chaska, MN). group received i.p. injection of fisetin (40mg/kg ~ 1 mg/animal) in 30 μl of
PBS/DMSO (1:1) twice weekly. Tumor sizes were measured twice weekly
Animals as described previously 3(0). All animals were sacrificed when tumors
reached a volume of 1200mm 3 in the control group. Samples were co-l
lected and stored at −20°C until further analysis.
Spontaneous PCa progression (TRAMP) model
Eight-week-old transgenic TRAMP micen( = 27) were randomly divided
Cell culture into two groups for five different time points from 8–12–16–20–24 weeks.
Non-tumorogenic RWPE1 (CRL-11609), tumorogenic NB11 (CRL-2851) and Each time point/group contained three animals, respectively. The first
NB26 (CRL-2852) cell lines were directly obtained from ATCC (Manassas, group of TRAMP (n = 15) animals received 50 µl PBS/DMSO (1:1) and served
as the control. The second groupn( = 12) received fisetin (1mg/animal) series with a Superose 12 and 10/300 GL column (GE Healthcare, Pittsburgh,
dissolved in 50 µl PBS/DMSO (1:1) thrice weekly, beginning at 8 weeks of PA) attached to a Waters HPLC system (Waters Corporation, Milford, MA).
age and continued until the animals were 12, 16, 20 and 24 weeks old, A solution containing 2m0M imidazole and 50 mM NaCl at pH 7.0 was used
respectively. Throughout the experiment, the animals had access to chow as an eluting buffer. All the samples were filtered through a 0μ.2m2 filter.
dietad libitum. Sample injection volume was 100 μl, and nominal flow rate was 0.m5l/min.
Animals in all groups were observed weekly for body weight, tumor The chromatography system consisted of columns, a ultraviolet det-ec
progression by abdominal palpation and survival. At the termination of tor (model 2998; Waters Corporation, Milford, MA) operating at 2n8m0, a
the experiment at their respective time points, blood samples were c-ol DAWN-DSP MALS photometer Wyatt Technology, Santa Barbara, CA) fitted
lected by the ‘mandibular bleed’ and serum was separated and stored at with a helium-neon laserλ( = 632.8 nm) and a K-5 flow cell, and a differe-n
−20°C until further analysis. The prostate was excised under a dissecting tial refractometer (DRI) detector (model 2414; Waters Corporation, Milford,
microscope and snap frozen in liquid nitrogen for further analysis. MA). The electronic outputs of the ultraviolet, DRI and MALS were sent to a
Dell computer. The data were processed with ASTRA (version 4.0) software.
Western blotting The DRI response factor was measured by injecting a series of known NaCl
After treatment with fisetin (4μ0M , 48 h), whole cell lysates were prepared concentrations into the detector with the syringe pump. This response -fac
and western blot analysis was performed as described previously29(). tor was obtained from the slope of the linear plot between NaCl con-cen
Densitometric measurements of the bands were done with image analysis tration and DRI response. The factor to correct the Rayleigh ratio to 90°
software using the Biorad ChemiDoc MP imaging system. for instrument geometry was obtained by measuring the LS intensity of
filtered (0.025μ m) HPLC quality toluene at 90°. The responses to LS inte-n
HA ELISA sity of the photodiodes arrayed around the scattering cell were normalized
to the diode at 90° with a bovine serum albumin (BSA) sample (monomeric
BSA with a nominal molecular weight of 6k6Da).
Human and Mouse HA ELISA kit were obtained from TSZ ELISA (Waltham,
MA) and R&D systems (Minneapolis, MN). These immunoassays have
been shown to accurately quantitate HA. Both kits are solid-phase ELISA Statistical analysis
designed to measure at least 3k5Da HA in biological samples. HA levels in
mouse serum and human PCa cells treated with or without fisetin (μ40M ,
48 h) were measured according to the manufacturer’s protocol.
Microsoft Excel software was used to calculate the mean and standard
error of the mean (SEM). For the HPLC–ESI–MS metabolomics study, a l-ib
eralP value of 0.1 was considered significant. Two-tailed, Studentt-’stest
RNA isolation and qPCR analysis was used to assess statistical significance. Data points in all the rest
of studies represent mean ± SEM, andP values < 0.05 were considered
Total RNA was extracted from cells and animal tissues using RNeasy kit significant.
(Qiagen, Germantown, MD), and reverse transcribed with iScript Reverse
transcription supermix kit (Biorad, Hercules, CA). cDNA was amplified
in triplicate using gene-specific primers S(upplementary Table 2, avail- Results
able atCarcinogenesis Online). Threshold cycleC(T) values obtained from
the instrument’s software were used to calculate the fold change of the Identification of HA as a unique target of fisetin
respective mRNAs. ΔCT was calculated by subtracting thCeT value of the in PCa
housekeeping gene from that of the mRNA of interestΔ.ΔCT for each
mRNA was then calculated by subtracting thCeTvalue of the control from
the experimental value. Fold change was calculated by the formul−aΔΔC2T
To identify new biomarkers in PCa, we designed a comprehen
sive metabolic profiling of fisetin-treated NB11/NB26 tumor
xenografts. NB11 and NB26 cells are unique tumorogenic PCa
cells derived from non-tumorgenic RWPE1 cells upon exposure
Intracellular ROS assay to N-methyl-N-nitrosourea 3(
). We selected these unique PCa
The OxiSelect™ Intracellular ROS Assay Kit obtained from Cell Biolabs Inc. cells, NB11 and NB26, for determining thien vivo effects of
fise(San Diego, CA) provides a cell-based assay for measuring primarily hydr-o tin due to their ability to form rapid and reproducible tumors.
gen peroxide, along with hydroxyl, peroxyl and other ROS levels within a Aqueous and organic extracts were isolated from the tissue
cell. Cells treated with or without fisetinμ(M40) for specified times were samples and analyzed using HPLC–ESI–MS in both positive and
processed as per the manufacturer’s instructions. Hydrogen peroxide was negative scan modes. The total ion chromatogram resulted in
used as a positive control in the assay. Fluorescence was evaluated on a excellent separation between the two tissue groupns =( 6). The
Synergy H1 (BioTek) multimode microplate reader at 480/53n0m (excita- XCMS online analysis of the total ion chromatogram revealed
tion/emission) using Gen5 2.0 software (BioTek). a list of newly identified metabolites. The quasimolecular ions
Immunofluorescence analysis were confirmed and the exact masses of the monoisotopic
molecular weights were used to search the METLIN database.
gNrBa1f1t, aNnBd26T,RPACM3PanadniDmUa1l4s5licdeelslswterreeaitnecduwbaitthedfiwseittihnH(4μA0M a,n2t4ibho),dxyenaso-per Total metabolites identified were 2086, out of which 1203 were
protocol as described previously3(0). After rinsing in PBS tween 20 (PBST), downregulated and 883 were upregulated between vehicle c-on
slides were incubated with fluorescent-conjugated secondary antibody trol and fisetin-treated NB11 xenografts. Similarly, a total of 2203
at 1:500 dilution in blocking buffer. Slides were then rinsed in PBST, and metabolites were identified in control versus fisetin-treated
sections were mounted with ProLong Gold Antifade reagent containing NB26 xenografts, out of which 1654 were downregulated and 549
4′,6-diamidino-2-phenylindole (Invitrogen) and left in dark overnight. were upregulated S(upplementary Figure 1 and Supplementary
Slides were imaged with the Andor Revolution XD spinning-disk confocal Table 8, available atCarcinogenesis Online).
microscope using a ×20/×40/1.4 NA oil objective with identical exposures To obtain a comprehensive global metabolic signature
and gains for each antibody stained. for this discovery-based approach, we considered a liberal
SEC-MALS analysis P value of 0.1 as significant for this data set. Among identified
metabolites, an m/z of 776.25 stood out as a single metabolite
Cells with or without fisetin treatment were collected over time and pooled HA (Figure 1A). The relative intensity of the identified HA peak
in order to isolate enough HA to be detected in Size exclusion chroma-tog (retention time = 2546.73min) exhibited a significant decrease
raphy-multiangle laser light scattering (SEC-MALS) analysis. HA was- iso
lated using the PEGNAC HA size protocol (available online). This method in fisetin-treated NB26 xenograftsFi(gure 1B) and plotted as
sequentially removes proteins, nucleic acids and isolate HA from cells, a function of relative abundance (fold change = 76.09) in both
media, tissues or any biological samples. Separation of HA molecular fr-ag fisetin-treated xenograft groupFsi(gure 1C). We next performed
ments was carried out by SEC-MALS on a Superose 6, 10/300 GL column in immunofluorescence staining for HA in NB11 and NB26 tumor
tissues and observed that fisetin-treated animals exhibited retention of more HA inside the cell with increasing doses
significantly decreased HA expression as compared with the (Supplementary Figure 4A, available atCarcinogenesis Online).
untreated controlsFi(gure 1D and Supplementary 2, available at Also, fisetin treatment (40µM , 48 h) significantly decreased cell
Carcinogenesis Online). proliferation in NB11 and NB26 cellsSu(pplementary Figure 4B,
available atCarcinogenesis Online). Fisetin was able to decrease
Fisetin decreases intracellular and extracellular HA both intracellular and extracellular HA levels signifying HA as a
abundance in PCa cells potential target of fisetin in PCa.
We performed immunofluorescence staining for HA in NB11,
NB26, PC3 and DU145 PCa cells treated with or without fisetin Fisetin decreases HA levels that are associated
(40 µM, 48 h). Fluorescence intensity in all four cell types suggested with inhibition of disease progression in TRAMP
that fisetin-treated cells showed significantly decreased HA mouse model
expression as compared with the untreated controlFsig(ure 2A The in vitro results of NB11/NB26 cells prompted us to look for si-m
and Supplementary Figure 3, available atCarcinogenesis Online). ilar effects of fisetin in another mouse model of PCa. The TRAMP
Next, we evaluated intracellular and extracellular HA upo n- fise model is a spontaneous PCa progression model that closely
tin treatment (40µ M, 48 h) using a solid-phase sandwich human mirrors the pathogenesis of human PCa 3(
). Compared with
HA ELISA assay. Our data showed that intracellular HA levels vehicle-treated animals, fisetin-treated animals of 24 weeks
significantly decreased in fisetin-treated PCa celFlisg u(re 2B). exhibited decreased hyperplasia S(upplementary Figure 5A,
Similarly, secreted HA levels in the fisetin-treated group were available atCarcinogenesis Online) in the prostate. We found no
significantly lower than the untreated contFrioglu(re 2C). significant change in body weight of the animals between the
It was further noted that PCa cells secrete more HA in the vehicle- and fisetin-treated groups as shownSu( pplementary
media and retain less HA within the cellFi(gure 2B, Figure 2C). Figure 5B, available atCarcinogenesis Online), indicating that fis-e
PC3 and NB26 cells treated with or without fisetin (5–µ4M0 ) tin is not associated with any significant adverse side effects. We
showed a similar decrease in secreted HA levels in a dose- also observed less palpable tumors in fisetin-treated animals
dependent manner. We found that fisetin decreases HA levels when compared with the vehicle-treated controls. Many of the
secreted in the media in a dose-dependent manner and enables fisetin-treated animals had well-differentiated adenocarcinoma
with no evidence of poorly differentiated carcinoma with cell types F(igure 4B and Supplementary Figure 4C, available at
disease progression (Supplementary Figure 5C, available at Carcinogenesis Online).
Carcinogenesis Online). qPCR analysis of PCa tumors of TRAMP tissues revealed a
Next, we performed immunofluorescence staining for HA in similar significant decrease in HAS2 and HAS3 mRNA levels in
TRAMP prostate tissues and compared HA expression intensity fisetin-treated mice as compared with the vehicle-treated c-on
among control and fisetin-treated groups. Fluorescence in-ten trols with increasing weeks of disease progressionFi(gure 4C).
sity showed that the vehicle-treated control animals dem- on Previous studies have shown that native HA can be broken
strated increased expression with disease progression that down into smaller fragments either via enzymatic (via HYALs)
was significantly reduced in fisetin-treated animalFsig(ure 3A, or non-enzymatic catabolism (via ROS). Since fisetin decreases
Figure 3B). Next, we evaluated the effect of fisetin on secreted enzymatic catabolism via reduction in HYAL levelsFi(gure 4A,
HA levels in TRAMP mouse serum using a mouse HA ELISA. Our Figure 4B) both in vitro/in vivo, we investigated the role of
nonresults revealed that serum HA levels increased with disease enzymatic catabolism in fisetin-treated cells by measuring ROS
progression and fisetin-treated animals showed significantly levels in both PC3 and NB26 cells. Increase in ROS levels can
lower serum HA levels as compared with the respective controls breakdown the native HMM-HA into smaller-molecular-weight
(Figure 3C). fragments. Our data show that fisetin increased ROS levels-sig
nificantly in both PC3 and NB26 cellFsi(gure 4D).
Fisetin reduces HA synthesis and degradation
enzyme levels in vitro and in vivo Fisetin increases abundance of HMM-HA in PCa cells
Previous studies indicate that elevated HAS and HYAL levels In non-diseased conditions, native HA is present as HMM-HA
are responsible for the high turnover (both synthesis and d-eg (Mw ~107 Da), whereas in diseased condition, it gets broken
radation) of HA in advanced disease state and may cooperate down into LMM-HA. HMM-HA is known to have antiangiogenic
together in cancerous growth. To determine whether fisetin properties; however, LMM-HA has shown to be proangiogenic.
has any effect on HA synthesis and degradation, we evaluated Since the biological functions of HA is size dependent, we first
mRNA levels of HAS and HYAL enzymes in NB11 and NB26 evaluated the molar mass distribution profile of HA fragments
cells with or without fisetin treatment (µ4M0 ) for 6h. Fisetin- in normal versus cancerous cells. Intracellular HA was- col
treated cells exhibited a significant decrease in HAS and HYAL lected from normal RWPE1 and PCa PC3 cells by employing the
mRNA levels (Figure 4A). The change in transcriptional levels of PEGNAC HA size protocol. The isolated HA was then used for
HAS and HYALs were reflected in protein expression that were SEC-MALS analysis to identify and compare the molar mass of
markedly reduced upon fisetin treatment between the two HA fragment in both cell types. The molar mass versus elution
volume chromatograms of SEC-MALS analysis and theMw of respectively. We measured the HA fragment sizes in
vehicleHA fragments showed that normal RWPE1 cells predominantly versus fisetin-treated animals. The light-scattering MALS data
exhibited higher levels of HMM-HA fragment cluster and no in vehicle-treated TRAMP animals showed a decreased abu-n
detectable LMM-HA cluster. In PC3 cells, we observed lower-lev dance of HMM-HA fragment cluster with increased abundance
els of HMM-HA fragment cluster and higher levels of LMM-HA of proangiogenic LMM-HA fragment clusters. Interestingly, in
(Figure 5A, Supplementary Table 3, available atCarcinogenesis fisetin-treated animals, we observed increased abundance of
Online) that could contribute to the pro-oncogenic nature of antiangiogenic HMM-HA fragment clusters and lower ab-un
cancer cells. dance of LMM-HA fragment clustersF(igure 6A, Supplementary
Next, we determined the effect of fisetin on molecular mass Table 5, available atCarcinogenesis Online). Studies have shown
profile of HA fragments in PCa cells. Intracellular and extr-acel that increased HMM-HA levels are directly associated with re-sist
lular HA with or without fisetin treatment were isolated from ance to oncogenesis 1(
). These results suggest a novel role
PC3 cells/culture media and analyzed using SEC-MALS. We of fisetin in increasing the abundance of intracellular HMM-HA,
observed an increased abundance of intracellular HMM-HA which is associated with resistance to PCa progression.
fragment cluster in fisetin-treated cells when compared with Serum analysis of vehicle-treated TRAMP animals revealed
the untreated control cells. Even the intracellular LMM-HAhigher secretion of antiangiogenic HMM-HA and lower sec-re
fragments in fisetin-treated cells were found to have comp-ar tion of LMM-HA fragment clusters. Notably, serum analysis of
atively higherMw than the untreated control celFlisgu(re 5B, fisetin-treated animals showed lower secretion of antian-gio
Supplementary Table 4, available at Carcinogenesis Online). genic HMM-HA and higher secretion of proangiogenic LMM-HA
Interestingly, we observed that PC3 cells release more HMM-HA fragment clustersF(igure 6B, Supplementary Table 6, available
in the media, whereas no detectable HMM-HA was secreted in at Carcinogenesis Online). Taken together, these findings clearly
media of fisetin-treated cells suggesting retention of antian-gio show that fisetin inhibits the development of PCa in TRAMP
genic HMM-HA fragments within the cells. mouse model.
Fisetin increases abundance of HMM-HA conferring resistance to prostate oncogenesis in TRAMP mouse model
Our study demonstrates that fisetin, which is found in many
TRAMP (n = 15) and TRAMP + fisetin (n = 12) animals were fruits and vegetables, is a potent HA synthesis inhibitor and
selected randomly, and vehicle or fisetin was administered can be developed for management of PCa. We and others have
beginning at 8 weeks of age and continued until 24 weeks old, previously reported that fisetin is an effective proapoptotic
agent with anti-invasive and antiangiogenic propertie2s9(–34). advantage when compared with 4-MU bothin vitro and in vivo
The inhibition of HA synthesis by fisetin is significant cons-id PCa models.
ering the fact that targeting the HA pathway with other agents Accumulation of HA is associated with progression of va-ri
has met with limited success. Only 4-MU has been found to ous cancers including PCa. Nevertheless, there are inconsistent
be the most promising till date 1(
). d-Mannose has shown views regarding the significance and relevance of HA secreted in
promise in vitro although at very high concentrations ofm20M serum. Studies in breast cancer have shown that the total serum
(42). Similarly, a curcumin analogue was shown to inhibit HA HA levels reflect disease progression44(), whereas other studies
export from fibroblasts at 5µM (43); however, antitumor acti-v have shown that the serum HA level does not have any progn-os
ity of both these agents has not been evaluated at these doses. tic significance with disease4(
). Our results suggest that fisetin
Sulfated HA has been reported to display antitumor activity significantly decreases both cellular and secreted HA levelins in
by blocking HA signaling, but its oral bioavailability and- evi vitro/in vivo and shows promise in prevention of other cancers
dence in experimental in vivo models has not yet been evalu- that are associated with increased HA levels.
). 4-MU, which is a known HA synthesis inhibitor, is The concept of HA present inside the cells has met with a
effective at high concentrations. The 5I0Cof 4-MU is 400 µM, lot of skepticism. Several studies have addressed this question,
whereas fisetin inhibits HA synthesis at 4µ0M , a 10-fold lower but the actual mechanism of intracellular HA uptake remains
dose. The in vitro IC50 dose of fisetin translates to 4m0g/kg body elusive. HA is a large molecule synthesized in the plasma me
mweight (~1mg/day) in vivo, which is less than the dose required brane; hence, the existence of intracellular HA seems less pl-au
in humans (1.5–2 g/day) as dietary supplement. At these low sible. However, evidence for intracellular HA has continued to
doses, fisetin exhibits no detectable toxicity as revealed by grow over the years, and it is now clear that HA can get into the
xenograft and TRAMP studies. We observed that fisetinm(1g/ cells to be catabolized and perhaps perform a number of impo-r
animal/day) significantly reduced tumor growth and delayed tant cellular functions. HA internalization is tightly regulated
progression of PCa in the TRAMP mouse model. Fisetin acts as and appears to occur simultaneously with increased HA synt-he
a potent HA synthesis inhibitor and exhibits significant dose sis during cell migration and proliferatio4n6(–52). It seems that
normal removal of HA from the matrix occurs via CD44 along replaced by newly formed HA every 24h (
). Importantly, the
with HYAL2 activation. HA binds to CD44 and other specific cell presence of ROS is known to enhance HA turnover5(
surface receptors and gets internalized. Increased HYAL2 results results show that fisetin treatment significantly inhibits both
in a cleavage of HMM-HA into smaller fragments, and CD44 is HAS and HYAL enzymes and halts the synthesis of new HA
then released and recycled back to the cell surface. Most of the and also prevents the degradation of previously synthesized
intracellular HA translocated into the cytoplasm or nucleus isHA. Fisetin treatment increased ROS levelins vitro but did not
primarily low molecular weight (<3k0Da). Our results showed enhance HA turnover in PCa cells.
that fisetin treatment significantly decreased hyaluronidase HA is a ‘dynamic’ molecule with a constant turnover in
protein levels, especially HYAL1 and HYAL2. Interestingly, we many tissues via rapid metabolism leading to HA fragments of
still found HA internalization within the cells even in absence of various sizes: HMM-HA, LMM-HA and o-HA. The innumerable
HYAL2. As observed by others also5(0), treatment of cells with oncogenic functions of HA mainly depend on its molecular size
the monoclonal antibody to the HA receptor CD44 increased (
). In this study, we used SEC-MALS analysis to
endosomal HA that was significantly decreased upon fisetin accurately measure the molecular distribution of HA size f-rag
treatment S( upplementary Figure 6, available atCarcinogenesis ments in all biological samples. Although large molecules of
Online). We suggest a novel concept that HA internalization in HA promote tissue integrity and quiescence, the generation of
fisetin-treated cells occurs via CD44 alone and is independent of breakdown products enhances signal transduction, contributing
polymer digestion by HYAL2 activation. This significant finding to the pro-oncogenic behavior of cancer cells. Previous studies
represents the mechanistic aspect of HA internalization upon (15,40,41) have revealed that cancer resistance in naked mole rat
fisetin treatment. derives from the abundant production of HMM-HA. In contrary,
HAS enzymes synthesize HA, whereas the HYALs degrade it LMM-HA has been proven to play a crucial role in breast cancer
into smaller fragments. This constant HA turnover is important and seems to be a more promising molecular biomarker than
for the maintenance of tissue homeostasis, and ~30% of HA is total HA for detecting breast cancer metastas5i7s).( Our in vivo
results showed that during PCa progression in TRAMP animals, states, which undergoes breakdown with progression of disease
there is increased pool of HMM-HA that further breaks down into and larger pool of proangiogenic LMM-HA accumulates. Fisetin
LMM-HA. This suggests higher activity of HYAL enzymes leading reduces the larger pool of LMM-HA and increases the abundance
to higher degradation with disease progression. Increased ROS of HMM-HA, which mimics the HA fragment profile of the
nonlevels in diseased state may also contribute to the breakdown diseased state. The precise mechanism underlying the actual
of HMM-HA into smaller fragments. Upon fisetin treatment, we role of fisetin and its ability to directly or indirectly restructure
observed an increase in accumulation of HMM-HA botihn vitro/in the smaller HA fragments into larger fragments though very
vivo. This increase of HMM-HA fragments in presence of fisetin promising requires further exploration.
mimics the HA fragment profile of the non-diseased state that In summary, our study demonstrates fisetin, as a non-toxic,
unveils a novel effect of fisetin. This effect is because fisetin -sig orally bioavailable, potent HA synthesis inhibitor in PCa. Fisetin
nificantly decreases both HAS and HYAL levels. In this condition, treatment enables accumulation of a larger pool of ant-ian
the HA synthesis process is halted and no new HMM-HA is pr-o giogenic HMM-HA and lowers level of proangiogenic LMM-HA,
duced. At the same time with reduced activity of HYALs upon which leads to reduction of HA signaling. The dose of fisetin
fisetin treatment, there seems to be very little or no degradationused in our in vivo studies was comparable with those in human
of the preexisting HMM-HA. Fisetin treatment induces ROS l-ev use as a dietary supplement. Hence, by inhibiting HA synthesis,
els, but surprisingly, increased ROS levels do not seem to cont-rib fisetin displays significant potential for prevention and tr-eat
ute toward HA degradation as evidenced by larger accumulation ment of PCa in preclinical models.
of HMM-HA upon fisetin treatmentin vitro. Hence, the HMM-HA
stays intact within the cells and tissues upon fisetin treatment. Supplementary material
Another possible reason for this phenomenon could be fis-e
tin’s ability to directly or indirectly catalyze and reassembleSupplementary Tables 1–9 and Figures 1–6 can be found at
the LMW-HA fragments into HMW-HA fragments. Interestingly, http://carcin.oxfordjournals.org/
SEC-MALS analysis of commercially available HA oligos (HA-60K
and HA-200K) incubated with fisetin revealed an increase in the Funding
molecular weight of the starting HA oligos in a cell-free- sys United States Public Health Service Grants (RO1 CA160867 to H.M.).
tem (Supplementary Table 9, available atCarcinogenesis Online).
These preliminary data open up newer promising avenues and
suggest fisetin probably possesses the potential to reassemble Acknowledgements
the smaller HA fragments to larger ones. On the basis of our We gratefully acknowledge Cleveland Clinic for using their
findings, antiangiogenic HMM-HA is increased in non-diseased HA isolation protocol (available online NHLBI # PO1HL107147).
We thank the CORE resource facility at UW-Madison for helping in 24 . Piccioni , F. et al. ( 2012 ) Antitumor effects of hyaluronic acid inhibitor our confocal imaging studies (National Institutes of Health grant 4-methylumbelliferone in an orthotopic hepatocellular carcinoma P30 EY016665) . We thank Dr G.Barett-Wilt at the UW-Madison model in mice . Glycobiology , 22 , 400 - 410 .
Biotechnology Center for his LCMS expertise and assistance . 25. Uchakina , O.N. et al. ( 2013 ) Targeting hyaluronic acid production for the Conflict of Interest Statement: None declared. treatment of leukemia: treatment with 4-methylumbelliferone leads to induction of MAPK-mediated apoptosis in K562 leukemia . Leuk. Res. , 37 , 1294 - 1301 . 26 . Wang , R. et al. ( 2015 ) Role of hyaluronan and glucose on 4-methylu-m References belliferone-inhibited cell proliferation in breast carcinoma cells .-Anti
1. Sleeman , J.P. et al. ( 2012 ) Concepts of metastasis in flux: the stromal cancer Res ., 35 , 4799 - 4805 . progression model. Semin. Cancer Biol ., 22 , 174 - 186 . 27 . Lall , R.K. et al. ( 2015 ) Dietary polyphenols in prevention and treatment
2. Yang , C. et al. ( 2012 ) The high and low molecular weight forms of h-ya of prostate cancer . Int. J. Mol. Sci ., 16 , 3350 - 3376 . luronan have distinct effects on CD44 clustering . J. Biol. Chem ., 287 , 28 . Sak , K. ( 2014 ) Cytotoxicity of dietary flavonoids on different human 43094-43107 . cancer types. Pharmacogn. Rev. , 8 , 122 - 146 .
3. Slevin , M. et al. ( 2007 ) Hyaluronan-mediated angiogenesis in vascular 29 . Mukhtar , E. et al. ( 2015 ) Dietary flavonoid fisetin bindsβt-otubulin and disease: uncovering RHAMM and CD44 receptor signaling pathways. disrupts microtubule dynamics in prostate cancer cells . Cancer Lett ., Matrix Biol ., 26 , 58 - 68 . 367, 173 - 183 .
4. Anttila , M. A . et al. ( 2000 ) High levels of stromal hyaluronan predict 30 . Khan , M.I. et al. ( 2014 ) YB-1 expression promotes epithelial-to-mese-n poor disease outcome in epithelial ovarian cancer . Cancer Res. , 60 , chymal transition in prostate cancer that is inhibited by a small -mol 150-155 . ecule fisetin. Oncotarget , 5 , 2462 - 2474 .
5. Fuchs , K. et al. ( 2013 ) Opposing effects of high- and low-molecular 31 . Adhami , V.M. et al. ( 2012 ) Dietary flavonoid fisetin: a novel dual inh-ibi weight hyaluronan on CXCL12-induced CXCR4 signaling depend on tor of PI3K/Akt and mTOR for prostate cancer management . Biochem. CD44. Cell Death Dis. , 4 , e819. Pharmacol., 84 , 1277 - 1281 .
6. Lokeshwar , V.B. et al. ( 2001 ) Stromal and epithelial expression of tumor 32 . Suh , Y. et al. ( 2010 ) Fisetin induces autophagic cell death through s-up markers hyaluronic acid and HYAL1 hyaluronidase in prostate cancer. pression of mTOR signaling pathway in prostate cancer cells . Carci-no J. Biol. Chem ., 276 , 11922 - 11932 . genesis, 31 , 1424 - 1433 .
7. Simpson , M. A . et al. ( 2008 ) Hyaluronan and hyaluronidase in genitou-ri 33 . Khan , N. et al. ( 2008 ) A novel dietary flavonoid fisetin inhibits androgen nary tumors . Front. Biosci. , 13 , 5664 - 5680 . receptor signaling and tumor growth in athymic nude mice . Cancer
8. Itano , N. et al. ( 2002 ) Abnormal accumulation of hyaluronan matrix Res ., 68 , 8555 - 8563 . diminishes contact inhibition of cell growth and promotes cell mig-ra 34 . Khan , N. et al. ( 2008 ) Fisetin, a novel dietary flavonoid, causes apoptosis tion . Proc. Natl Acad. Sci. USA , 99 , 3609 - 3614 . and cell cycle arrest in human prostate cancer LNCaP cells . Car-cino
9. Beasley , K.L. et al. ( 2009 ) Hyaluronic acid fillers: a comprehensive genesis , 29 , 1049 - 1056 . review. Facial Plast. Surg. , 25 , 86 - 94 . 35 . Adhami , V.M. et al. ( 2012 ) Oral infusion of pomegranate fruit extract
10. Collins, M.N. et al. ( 2013 ) Hyaluronic acid based scaffolds for tissue inhibits prostate carcinogenesis in the TRAMP model. Carcinogenesis, engineering-a review . Carbohydr . Polym., 92 , 1262 - 1279 . 33, 644 - 651 .
11. Rayahin , J.E. et al. ( 2015 ) High and low molecular weight hyaluronic 36 . Want , E.J. et al. ( 2013 ) Global metabolic profiling of animal and human acid differentially influence macrophage activation . ACS Biomater. Sci. tissues via UPLC-MS. Nat. Protoc ., 8 , 17 - 32 . Eng., 1 , 481 - 493 . 37 . Webber , M.M. et al. ( 2001 ) Human cell lines as anin vitro/in vivo model
12. Stern , R. et al. ( 2007 ) The many ways to cleave hyaluronan. Biotechnol. for prostate carcinogenesis and progression . Prostate , 47 , 1 - 13 . Adv., 25 , 537 - 557 . 38 . Gingrich , J.R. et al. ( 1997 ) Androgen-independent prostate cancer p-ro
13. Gao , F. et al. ( 2010 ) Hyaluronan oligosaccharides promote excisional gression in the TRAMP model . Cancer Res. , 57 , 4687 - 4691 . wound healing through enhanced angiogenesis . Matrix Biol ., 29 , 107 - 39 . Gingrich, J.R. et al. ( 1996 ) Metastatic prostate cancer in a transgenic 116. mouse . Cancer Res. , 56 , 4096 - 4102 .
14. Gao , F. et al. ( 2008 ) Hyaluronan oligosaccharides are potential stim-ula 40 . Fisher, G.J. ( 2015 ) Cancer resistance, high molecular weight hyaluronic tors to angiogenesis via RHAMM mediated signal pathway in wound acid, and longevity . J. Cell Commun . Signal., 9 , 91 - 92 . healing. Clin. Invest. Med ., 31 , E106 - E116 . 41 . Seluanov , A. et al. ( 2009 ) Hypersensitivity to contact inhibition provides
15. Tian , X. et al. ( 2013 ) High-molecular-mass hyaluronan mediates the a clue to cancer resistance of naked mole-rat . Proc. Natl Acad. Sci. USA , cancer resistance of the naked mole rat . Nature , 499 , 346 - 349 . 106, 19352 - 19357 .
16. Lesley , J. et al. ( 2000 ) Hyaluronan binding by cell surface CD44 . J. Biol . 42 . Jokela , T.A. et al. ( 2013 ) Mannose reduces hyaluronan and leukocytes Chem ., 275 , 26967 - 26975 . in wound granulation tissue and inhibits migration and hyaluronan-
17. Wang , Y.Z. et al. ( 2011 ) CD44 mediates oligosaccharides of hyalu-ro dependent monocyte binding . Wound Repair Regen. , 21 , 247 - 255 . nan-induced proliferation, tube formation and signal transduction in 43 . Prehm, P. ( 2013 ) Curcumin analogue identified as hyaluronan export endothelial cells . Exp. Biol. Med . (Maywood), 236 , 84 - 90 . inhibitor by virtual docking to the ABC transporter MRP5 . Food Chem.
18. Benitez , A. et al. ( 2011 ) Targeting hyaluronidase for cancer therapy: Toxicol ., 62 , 76 - 81 . antitumor activity of sulfated hyaluronic acid in prostate cancer cells . 44 . Delpech , B. et al. ( 1990 ) Serum hyaluronan (hyaluronic acid) in breast Cancer Res ., 71 , 4085 - 4095 . cancer patients. Int. J. Cancer , 46 , 388 - 390 .
19. Lokeshwar , V.B. et al. ( 2010 ) Antitumor activity of hyaluronic acid s-yn 45 . Ponting , J. et al. ( 1992 ) Prognostic relevance of serum hyaluronan levels thesis inhibitor 4-methylumbelliferone in prostate cancer cells. Cancer in patients with breast cancer . Int. J. Cancer , 52 , 873 - 876 . Res., 70 , 2613 - 2623 . 46 . Brecht , M. et al. ( 1986 ) Increased hyaluronate synthesis is required for
20. Nagy , N. et al. ( 2015 ) 4-Methylumbelliferone treatment and hyaluronan fibroblast detachment and mitosis . Biochem. J. , 239 , 445 - 450 . inhibition as a therapeutic strategy in inflammation , autoimmunity, 47. Evanko , S.P. et al. ( 1999 ) Intracellular localization of hyaluronan in- pro and cancer . Front. Immunol. , 6 , 123. liferating cells. J. Histochem. Cytochem. , 47 , 1331 - 1342 .
21. Tamura , R. et al. ( 2014 ) 4-Methylumbelliferone inhibits ovarian cancer 48 . Hascall , V.C. et al. ( 2004 ) Intracellular hyaluronan: a new frontier for growth by suppressing thymidine phosphorylase expression . J. Ovar- inflammation? Biochim. Biophys. Acta , 1673 , 3 - 12 . ian Res., 7 , 94 . 49. Ripellino , J.A. et al. ( 1988 ) Light and electron microscopic studies on
22. Yates , T.J. et al. ( 2015 ) Dietary supplement 4-methylumbelliferone: an the localization of hyaluronic acid in developing rat cerebellum . J. Cell effective chemopreventive and therapeutic agent for prostate cancer . J. Biol ., 106 , 845 - 855 . Natl. Cancer Inst. , 107 , 85 - 94 . 50 . Tammi , R. et al. ( 2001 ) Hyaluronan enters keratinocytes by a novel
23. Benitez , A. et al. ( 2013 ) Dietary supplement hymecromone and endocytic route for catabolism . J. Biol. Chem ., 276 , 35111 - 35122 . sorafenib: a novel combination for the control of renal cell carcinoma . 51. Tammi , R. et al. ( 1991 ) Correlations between hyaluronan and epidermal J. Urol., 190 , 285 - 290 . proliferation as studied by [3H]glucosamine and [3H]thymidine inco-r porations and staining of hyaluronan on mitotic keratinocytes. Exp. oxidative effect of cysteine-derived compounds . Chem . Biodivers., 9 , Cell Res ., 195 , 524 - 527 . 309 - 317 .
52. Vigetti , D. et al. ( 2009 ) Modulation of hyaluronan synthase activity in 55 . Hrabarova , E. et al. ( 2011 ) Pro-oxidative effect of peroxynitrite reg-ard cellular membrane fractions . J. Biol. Chem ., 284 , 30684 - 30694 . ing biological systems: a special focus on high-molar-mass hyaluronan
53. Fox , S.B. et al. ( 1994 ) Normal human tissues, in addition to some degradation . Gen. Physiol. Biophys. , 30 , 223 - 238 . tumors, express multiple different CD44 isoforms . Cancer Res. , 54 , 56 . Soltés , L. et al. ( 2006 ) Degradative action of reactive oxygen species on 4539-4546 . hyaluronan. Biomacromolecules , 7 , 659 - 668 .
54. Hrabárová , E. et al. ( 2012 ) Free-radical degradation of high-molar-mass 57 . Wu , M. et al. ( 2015 ) A novel role of low molecular weight hyaluronan in hyaluronan induced by ascorbate plus cupric ions: evaluation of a-nti breast cancer metastasis . FASEB J ., 29 , 1290 - 1298 .