Background
Ovarian cancer is the most lethal gynecological cancer and ranks as the fourth most common cause of cancer-related death in women in the Western world [
1]. The standard treatment for advanced ovarian cancer is debulking surgery, followed by platinum-based chemotherapy. This standard treatment results in a complete response rate of 40-60%, however, more than 90% of these patients relapse after 2 years [
2]. Recurrent ovarian cancer in most cases becomes incurable due to the development of chemoresistance.
Chemoresistance is multifactorial in nature involving tumor-related and drug-related factors, as well as interactions with the tumor microenvironment itself. A well established cause involves increased expression of members of the membrane efflux ATP binding cassette (ABC) transporter family, which decrease the intracellular accumulation and retention of chemotherapy drugs [
3]. ABC transporters are a family of membrane proteins that transport a wide range of substrates, including metabolic products, nutrients, lipids, and drugs, across extracellular and intracellular membranes, and have been shown to play an important role in many human diseases [
4,
5]. Phylogenetic analysis places the 48 known human ABC transporters into 7 distinct subfamilies (ABCA-ABCG) [
4]. The first ABC transporter identified was MDR1 (ABCB1), also known as p-glycoprotein [
6]. ABCB1 plays a critical role in drug fluxes and chemoresistance in many malignancies, including ovarian cancer [
7‐
10]. Other ABC transporters, including ABCB3 (TAP2), ABCC1 (MRP1), ABCC2 (MRP2, cMOAT), and ABCC3 (MRP3), have been shown to be involved in ovarian cancer chemoresistance [
11‐
15].
A component of the tumor microenvironment linked to chemoresistance is the extracellular matrix (ECM) molecule, hyaluronan (HA) [
16]. HA is a large polysaccharide that is assembled into pericellular and ECM in many tissues [
17]. HA has a role in various cell functions such as adhesion, motility, and differentiation. It has also been implicated to be a key player in cancer metastasis [
17,
18]. Many human tumors, including ovarian cancer, are surrounded by a connective tissue matrix enriched with HA [
18‐
20]. Increased HA has been shown to be an independent predictor of ovarian cancer survival [
19]. HA levels significantly correlate with the degree of invasiveness and metastatic potential in malignant ovarian tumors [
19,
21], and it promotes the attachment of cancer cells to peritoneal cells via interactions with its major surface receptor, CD44 [
22‐
25].
HA has been shown to reduce the ability of chemotherapy drugs to cause cancer cell death in a variety of malignancies [
26‐
29]. Furthermore, several studies have demonstrated that HA interactions with CD44 can increase resistance to numerous chemotherapy drugs and increase the expression of the ABC drug transporters [
28,
30‐
34]. The co-localization of CD44 with ABCB1 and ABCC2 in ovarian cancer tissues suggests a functional link between CD44 expression and chemoresistance [
35]. However, knowledge regarding the importance of HA-CD44 interactions in mediating chemoresistance and regulating ABC transporter expression in ovarian cancer is limited. In this study we therefore assessed the ability of HA to block the cytotoxic effects of the chemotherapy drug carboplatin (CBP) on ovarian cancer cells and investigated a potential link between the HA-CD44 pathway and ABC transporter expression. To relate findings with the clinical situation, we measured HA serum levels in ovarian cancer patients at diagnosis, during chemotherapy and at recurrence and determined the relationship with patient outcome.
Methods
Cell lines
The human ovarian cancer cell lines OVCAR-3, SKOV-3, and OV-90 were obtained from American Type Culture Collection (ATCC, Manassas, VA, USA). OVCAR-5 cells were obtained from Dr. Thomas Hamilton (Fox Chase Cancer Center, Philadelphia, PA) [
36]. All ovarian cancer cell lines were maintained in RPMI 1640 medium supplemented with 4 mM L-glutamine and antibiotics (100 U penicillin G, 100 μg/ml streptomycin sulfate and 0.25 μg/ml amphotericin B, Sigma Aldrich, St Louis, MO, USA) at 37°C in an environment of 5% CO
2 as described previously [
37].
Cell survival assays
Cells were plated at 5,000 cells per well, grown for 24 hr, and then treated with CBP (0–300 μM, Mayne Pharma, Victoria, Australia) for 72 hr in normal growth media. Cell survival was calculated by MTT assay, as per the manufacturer’s instructions (Sigma Aldrich). The inhibitory concentration (LD50) of CBP was calculated from three independent experiments performed in triplicate using exponential regression curve fitting. To assess the effect of HA on cell survival, cells were plated at 5,000 cells per well, grown for 24 hr, and then treated with IC50 dose of CBP and HA (0–100 μg/ml, H1504, Sigma Aldrich) ± CD44 neutralizing antibody (Clone A020, 20 μg/ml Calbiochem, NSW, Australia), control IgG (20 μg/ml, BD Biosciences, North Ryde, NSW, Australia) or HA oligomers (HYA-OLIGO 6–10, 10–250 μg/ml, North Star Bioproducts, Associates of Cape Cod Inc, East Falmouth, MA, USA or 10-250 μg/ml HYA-OLIGO 8, Hyalose LLC, Oklahoma City, OK, USA) for 72 hr before cell survival assessment by MTT assay. Chemosensitivity of SKOV-3 cells to CBP in presence of HA oligomers (250 μg/ml) was determined using the MTT assay as described above. Exponential curve fitting was used to the calculate equation of the line and the CBP LD50 in the presence of CBP alone and combined with HA oligomers.
Quantitative real-time PCR
Cells were plated at 5,000 cells per well for 24 hr, and then treated with control medium, CBP LD50 dose or HA (5 μg/ml) for 72 hr. Cells were also treated CBP or HA in the presence or absence of HA oligomers (10-250 μg/ml HYA-OLIGO 6-8, Hyalose LLC). Total RNA was isolated from ovarian cancer cell lines and reverse transcribed using the TaqMan® Gene expression Cells-to-CT™ kit (Applied Biosystems, Mulgrave, Victoria, Australia), as per the manufacturer’s instructions. Briefly, lysis solution with DNAse was added to each well and incubated for 5 min at room temperature. Stop solution was then added to each well and mixed. The lysate (10 μl) was added to a 40 μl reverse transcription master mix and reverse transcribed for 1 hr. Resultant cDNA was stored as 50 μl aliquots at −20°C for qRT-PCR analysis.
qRT-PCR reactions were performed on triplicate samples using TaqMan® primer sets for genes of interest, as detailed in Additional file
1: Table S1 using the 7900HT Fast Real-Time PCR System (Applied Biosystems, Mulgrave, Victoria, Australia). Briefly, PCR reactions were made up to 10 μl and contained TaqMan® Gene Expression Master Mix (2×), primers for the gene of interest, nuclease free water, and the sample cDNA. PCR cycling conditions were as follows: 50°C for 2 min, 95°C for 10 min (with 40 cycles following of 95°C for 15 sec), and 60°C for 1 min. CT values were normalised to the house keeping gene β-actin and calibrated to no treatment control using the 2
-∆∆CT method.
HA ELISA assay
An HA ELISA kit (DY3614, R&D Systems, Minneapolis, MN) was used to determine the concentration of HA in serum samples or conditioned media (CM), as per manufacturer’s instructions. Serum was assessed from ovarian cancer patients at diagnosis (n = 101), after chemotherapy (n = 52), at 1st recurrence (n = 17) and 2nd recurrence (n = 5). Matched serum samples at diagnosis and following at least two cycles of chemotherapy were available from 32 patients. Serum samples were also obtained from patients with benign ovarian tumors (n = 22) and healthy controls (n = 27). All samples were collected with approval from the Royal Adelaide Hospital Human Ethics Committee and informed written consent was obtained from all participants involved in this study. CM was collected from cells (OVCAR-5, OVCAR-3, OV-90, and SKOV-3) plated at 5,000 cells per well and grown for 24 hr before 72 hr treatment with their IC50 concentration of CBP or control media. The detection limit of the assay was 100 pg/ml, and the coefficient of variation between assays was 9.21%.
HA staining in ovarian cancer tissues
Formalin fixed tissue was obtained from ovarian cancer patients at surgery (n = 10), postchemotherapy (n = 9) and at recurrence (n = 4). Tissues samples were collected with approval from the Royal Adelaide Hospital Human Ethics Committee. Informed written consent was obtained from all participants involved in this study. Tissue sections (5 μm) were mounted onto positively charged slides (SuperFrost VR Plus, Menzel-Glaser, Braunschweig, Germany) and heated at 60°C for 1.5 hr. HA in ovarian cancer sections was detected using HABP (2 μg/ml, Seikagaku Corp, Japan) [
38]. The intensity of HA in the epithelial and stromal compartments was assessed using a manual scoring method: strong (3+), moderate (2+), weak (1+), or negative (0). A score of 0 or 1+ was defined as low HA staining and a score or 2+ or 3+ was defined as high HA staining. The % of HA positive cancer cells were independently assessed in the three tissue groups.
HA staining and ABCC2 immunocytochemistry
OVCAR-5 cells (2×104 cells/well) were plated in 8 well tissue culture chamber slides (Nunclon™ Lab-Tek II Chamber slide, RS Glass Slide, Naperville, IL) in 500 μl 10% FBS RPMI for 24 hr and treated for 72 hr with IC50 concentration of CBP or control media. OVCAR-5 cells were fixed with cold 100% methanol (5 min) and cold 100% acetone (3 min), washed with PBS and blocked with 5% donkey serum and incubated overnight with biotinylated HABP (2 μg/ml, Seikagaku Corp, Japan) and/or mouse ABCC2 monoclonal antibody (1/25, Clone M2I-4, Abcam, Cambridge, United Kingdom). HA and ABCC2 were visualized with Cy3-conjugated streptavidin (1/100, 1 hr at RT, catalogue no: 016-160-084, Jackson Immunoresearch Laboratories, West Grove, PA, USA) or FITC-conjugated AffiniPure donkey anti-mouse (1/100, 1 hr at RT, catalogue no: 715-096-151, Jackson Immunoresearch Laboratories), respectively. Nuclei were stained with 4′,6′-diamidino-2-phenylindole dihydrochloride solution (DAPI, Molecular Probes, Eugene, OR, USA) and slides were mounted with fluorescent mounting medium (Dako Labs, Glostrup, Denmark). Cells were viewed with an epifluorescence microscope (BX50, Olympus Australia) and imaged using a 20× objective and a Spot RT digital camera (Diagnostic Instruments, Sterling Heights, MI). Negative controls included only Cy3-conjugated streptavidin and FITC-conjugated AffiniPure donkey anti-mouse and also Streptomyces hyaluronidase (30 min RT, 10 U/ml, Sigma-Aldrich) treatment for the HA staining.
Statistical analyses
All analyses were performed using the PASW 17.0 Statistical software (SPSS Inc., Chicago, IL). The Mann–Whitney U, Kruskal-Wallis, Wilcoxon signed rank or Chi Square tests were used to determine statistical significance between patient groups. For cell line studies, the Student’s t-test and one way ANOVA test with the Tukey or Dunnett C post-hoc tests were used to determine statistical significance between control and treatment groups. Clinical follow-up data for disease progression and survival was available for 77 and 83 ovarian cancer patients respectively. All patients received either CBP alone or a combination of CBP and paclitaxel. Forty five percent (35/77) of the patient’s progressed and 27.7% (23/83) died from ovarian cancer at the time of census (1st December 2012). Kaplan-Meier and Cox regression analyses with progression or death due to ovarian cancer was used as the endpoint to determine whether HA levels prior to chemotherapy treatment were related to progression-free survival (PFS) or overall survival (OS). Statistical significance was accepted at P < 0.05.
Discussion
Chemotherapy resistance is one of the most challenging problems in cancer treatment. The molecular mechanisms mediating chemoresistance are widely studied but still poorly understood. Multidrug transporter proteins, such as ABC transporters, are well known for their contribution to chemoresistance through the efflux of cytotoxic drugs from cancer cells [
40,
41]. ABC transporter expression and chemoresistance has been reported to be modulated by HA-CD44 interactions [
30,
32‐
34,
42]. Although the importance of HA and CD44 in ovarian cancer progression has been well established [
18,
25,
43], the knowledge about their significance in mediating ovarian cancer chemoresistance is limited.
Our results show that HA induces chemoresistance against CBP, and increases the expression of the ABC transporters, ABCB3, ABCC1, ABCC2, and ABCC3 in ovarian cancer cell lines expressing the HA receptor, CD44. By measuring serum HA levels in ovarian cancer patients, we demonstrate for the first time that HA levels are elevated in patients following chemotherapy treatment and at recurrence compared with HA levels at diagnosis. Importantly, higher serum HA levels (> 50 μg/ml) were associated with reduced progression-free and overall survival. Our findings confirm that, in addition to their important role in promoting malignant ovarian cancer cell behaviour, CD44-HA interactions also play a significant role in mediating chemoresistance.
HA production in ovarian cancer cells was increased in ovarian cancer tissues from patients that received neoadjuvent chemotherapy and at recurrence compared to tissues collected at surgery prior to any treatment. Other potential mechanisms that may also contribute to the increased HA serum levels following chemotherapy were not investigated further in this study. Pro-inflammatory cytokines including tumour necrosis factor and interleukin-1β shown to be increased by chemotherapy treatment [
44,
45] and stimulate HA production in a variety of cell types [
46‐
49] may also contribute to increasing serum HA levels.
Our data supports the model that chemoresistance is acquired by chemotherapy-induced HA production and increases ovarian cancer cell survival which contributes to chemoresistance by increasing the expression of ABC transporter proteins via a HA-CD44 mediated pathway. CBP treatment increased the secretion of HA in ovarian cancer cell lines by increasing HA synthesis, as we found corresponding increased
Has2 and
Has3 expression in OVCAR-5 and increased
Has3 expression in OV-90 and OVCAR-3 cells. In accord with our results, increased chemotherapy resistance was observed in MCF7 breast cancer cells overexpressing
Has2[
50]. Furthermore, chemoresistant lymphoma cell lines were also found to have greater expression of
Has1,
Has2, and
Has3 transcripts, and to secrete higher levels of HA [
51].
Our finding that HA promotes chemoresistance of ovarian cancer cells agrees with previous studies using other cancer cell lines showing that HA can increase the LD
50 of various chemotherapy drugs [
26‐
29,
31]. We have found that the addition of either neutralizing CD44 antibody or HA oligomers (6–10 sugar residues), which interact monovalently with CD44 and competitively block polyvalent interactions between CD44 and endogenous HA, blocked the HA induced chemoresistance in CD44 positive ovarian cancer cell lines but not in the CD44 negative, OVCAR-3 cells. These findings demonstrate the significance of HA-CD44 interactions in this mechanism. Furthermore, HA oligomers were able to sensitize chemoresistant SKOV-3 cells to CBP. These findings agree with previous studies demonstrating that HA oligomers can sensitize various carcinoma cell lines to chemotherapy drugs, including doxorubicin, taxol, and vincristine, both
in vitro and
in vivo[
30,
32,
34]. More recently, Slomiany
et al. found that HA oligomers decreased doxorubicin resistance of malignant peripheral nerve sheath tumours and suppressed HA secretion in these tumors [
52]. Importantly, the HA oligomers and doxorubicin acted synergistically at suboptimal doses and induced tumour regression to a greater extent than either agent alone [
52].
Several studies have also shown that HA-CD44 interactions regulate ABC transporter expression and activity [
28,
30‐
32]. HA has been shown to stimulate
ABCB1 expression and
ABCB1 activity in various cancer cell lines [
28,
30‐
32]. HA-CD44 interactions also regulate expression of
ABCG2 (BCRP) in glioma cells [
34].
ABCC2 expression is upregulated in non small cell lung cancer cells (H322) overexpressing CD44, which are more resistant to cisplatin when cultured on a HA matrix [
33]. Human mesenchymal stem cells cultured on a layer of HA are more resistant to doxorubicin and produce increased levels of
ABCB1[
53]. Interestingly, we found that SKOV-3 cells, which produced the highest HA levels, were most resistant to CBP and expressed the highest levels of
ABCC1 and
ABCC3. CD44 co-localizes in the plasma membrane of cancer cells with
ABCB1 and
ABCG2, and HA antagonists rapidly induce internalization of these transporters and CD44 to make them ineffective [
34,
52,
54]. These findings have led to the suggestion that multivalent interactions between HA and CD44 may be necessary for stabilization of transporter interactions within the plasma membrane [
52].
Bourguignon
et al. have previously reported
ABCB1 expression in SKOV-3.ipl, a variant cell line established from mouse xenograft, to be increased by HA treatment [
28]. However, we did not observe
ABCB1 expression in any of the ovarian cancer cell lines examined, neither in the absence nor presence of HA. In contrast to Bourguignon
et al. 2009 [
31], we did not observe an effect on
Bcl-2l expression by HA treatment in the ovarian cancer cell lines. In our experiments, HA alone had no effect on cell survival and an anti-apoptotic mechanism was excluded. Our findings suggest that HA induces chemoresistance by inducing ABC transporter expression, which increases ovarian cancer cell survival by increasing the efflux of CBP from the cells.
Our data supports a role for
ABCB3,
ABCC1,
ABCC2, and
ABCC3 in ovarian cancer chemoresistance, which is in agreement with previous studies. Overexpression of
ABCC1 and
ABCC2 in human ovarian cancer cells conferred marked resistance to chemotherapeutics, such as methotrexate [
55]. Furthermore, increased expression of
ABCC1 was observed in ovarian cancers from chemotherapy non-responders [
56]. Higher gene expression of
ABCC1 and
ABCC3 was also found in ovarian cancer patients with unfavourable outcome following debulking surgery and platinum based chemotherapy [
12]. Nuclear membrane localization of
ABCC2 has been shown to predict resistance to cisplatin [
39]. A recent study by Auner
et al. identified gene expression of the four ABC transporters
ABCC1,
ABCC2,
ABCC3 and
ABCB3 to be significantly elevated in recurrent ovarian cancer compared to benign tumors and untreated primary cancer [
15]. Oxaliplatin resistant ovarian cancer cells have been shown to overexpress
ABCC1 and
ABCC4[
57]. The importance of
ABCC2 in mediating chemoresistance is supported by our findings showing increased expression of
ABCC2 in ovarian cancer cells following treatment with CBP and verified by studies which showed that
ABCC2 siRNA knockdown could reverse cisplatin and paclitaxel resistance in ovarian cancer cell lines [
39,
58].
HA is a promising candidate for increasing efficacy and reducing toxicity of cancer therapies. Administration of HA-chemotherapy conjugates, which allows chemotherapy drugs to enter cancer cells via a CD44 receptor-mediated mechanism, resulted in increased therapeutic activity compared with chemotherapy alone [
59,
60]. In the taxane resistant HeyA8-MDR ovarian cancer xenograft model, metronomic doses (more frequent lower therapeutic doses) of paclitaxel-HA conjugate had a more effective anti-tumor activity and exhibited reduced toxicity compared with mice that were administered the maximum tolerated doses of the paclitaxel-HA conjugate [
61]. More recent studies have demonstrated that HA is a barrier for chemotherapy drugs in pancreatic cancers [
62,
63]. Systemic administration of chemotherapy together with hyaluronidase (PEGPH20), which degrades HA, improved blood vessel perfusion and resulted in increased sensitivity and improved survival in pancreatic cancer mouse models [
62,
63]. Clinical trials are ongoing to examine the effect of depleting HA in patients with pancreatic cancer (NCT01959139, NCT01839487 & NCT01453153,
http://clinicaltrials.gov). Together these studies and our recent findings support the notion that targeting HA in the tumor microenvironment is an important future direction for cancer therapy.
Competing interests
The authors declares that they have no competing interests.
Authors’ contributions
CR, MPW and MKO conceived the idea and designed the experiments; CR, MPW, NAL, IAT and CEP performed the experiments. All authors read and approved the manuscript.