Synergistic effect of targeting the epidermal growth factor receptor and hyaluronan synthesis in oesophageal squamous cell carcinoma cells

Worldwide, oesophageal cancer is the eighth most common cancer and has a very poor survival rate. In order to identify new tolerable treatment options for oesophageal squamous cell carcinoma (ESCC), erlotinib was tested with moderate efficacy in phase I and II studies. As 4‐methylumbelliferone (4‐MU), an hyaluronan (HA) synthesis inhibitor showed anti‐cancer effects in vitro, and in ESCC xenograft tumours, we investigated whether the anti‐cancer effects of erlotinib could be augmented by combining it with 4‐MU.


Introduction
Oesophageal cancer accounted for 3.2% of new cancer cases in 2012. It is therefore the eighth most frequently diagnosed type of cancer and the sixth most common cause of cancer deaths (approx. 400 000 per year). The very poor survival prognosis for patients suffering from this cancer entity is defined by an overall ratio of mortality to incidence of 0.88 (Ferlay et al., 2015). The two major types of oesophageal cancer are squamous cell carcinoma (ESCC) and adenocarcinoma (EAC). Treatment of oesophageal cancer depends upon type, stage, location of the tumour and on the medical condition of the patient. The treatment of ESCC comprises surgery and/or chemoradiation with platinum derivatives and 5-fluorouracil or taxanes (Stahl et al., 2013). In the past years, new targeted therapies for epithelial tumours such as receptor TK inhibitors have been investigated in anti-cancer studies.
To date, there are few clinical phase I and II trials on the efficacy of treatments including the EGFR TK inhibitor erlotinib in oesophageal cancer. In the studies on the treatment of ESCC, progression-free survival ranged from 3.3 to 12 months (Ilson et al., 2011;Zhai et al., 2013). The most prominent adverse effects of erlotinib were diarrhoea and rash (Dragovich et al., 2006;Ilson et al., 2011). Most authors did not report a significant correlation of EGFR expression level and treatment outcome due to small sample sizes (Ilson et al., 2011;Iyer et al., 2013;Zhai et al., 2013). However, the efficacy of erlotinib might be better in the treatment of ESCC compared to EAC. In ESCC, EGFR overexpression was more frequently detected (Ilson et al., 2011;Fichter et al., 2014). In ESCC cell lines, the combination of erlotinib with cetuximab, a monoclonal antibody directed against the EGFR, or the combination of erlotinib with the tyrphostin AG 1024, an insulin-like growth factor receptor TK inhibitor, or fluvastatin, was shown to have additive or even synergistic antiproliferative effects (Sutter et al., 2006).
Hyaluronan (HA), a major component of the extracellular matrix, is found in the tumour stroma and parenchyma of ESCC, depending upon the degree of tumour differentiation (Wang et al., 1996). We previously showed that HA synthase (HAS) isoform 3 mRNA expression was up-regulated in human ESCC tissue samples compared with normal mucosa (Twarock et al., 2011). Tumour cell-associated HA is associated with poor prognosis in breast cancer (Auvinen et al., 2000;, colorectal cancer (Ropponen et al., 1998), pancreatic cancer (Cheng et al., 2013) and malignant peripheral nerve sheath tumours (Ikuta et al., 2014). Inhibition of HA synthesis by 4-methylumbelliferone (4-MU) and HAS3 knockdown leads to decreased tumour volume in ESCC xenograft tumours in nude mice, and similarly, 4-MU treatment reduces tumour volume in a prostate cancer xenograft model (Lokeshwar et al., 2010). Additionally, 4-MU inhibited metastases in several animal studies (Yoshihara et al., 2005;Arai et al., 2011;Okuda et al., 2012;Hiraga et al., 2013). First examples for combining inhibition of HA synthesis with other anti-cancer drugs are the combinations with gemcitabine (Nakazawa et al., 2006), with the multi-kinase inhibitor sorafenib (Benitez et al., 2013) and with trastuzumab, a recombinant humanized anti-ErbB2 antibody (Palyi-Krekk et al., 2007). However, no information is available about increasing the effect of 4-MU in ESCC by additional chemotherapeutic drugs.
It has been shown previously that EGFR mRNA expression correlated with HAS3 mRNA expression in human ESCC tissue samples. In OSC1 cells, an ESCC cell line, EGF stimulation induced HAS3 mRNA expression pointing towards a role of HA in EGFR-regulated processes (Twarock et al., 2011). In squamous cell carcinoma of the head and neck, HA and EGFRs were expressed in a similar way (Jonsson et al., 2012). It is noteworthy that HA binding to its receptor CD44 promoted association with and activation of the EGFR (Toole, 2009). Activation of both EGFR and CD44 by their ligands induces downstream signalling pathways such as MAPK/ERK and PI3K/Akt and promotes proliferation, migration and survival (Citri and Yarden, 2006;Toole, 2009). The present experiments show that combining erlotinib and 4-MU resulted in significant inhibition of proliferation and migration of the ESCC cell line KYSE-410.

Cell culture, drugs and siRNA transfection
The human ESCC cell lines KYSE-270, KYSE-410 and KYSE-520 were purchased from the Leibnitz Institute DSMZ -German Collection of Microorganisms and Cell Cultures (Braunschweig, Germany) (Shimada et al., 1992). Cells were grown in RPMI 1640 GlutaMAX TM medium (Gibco®, Life These Tables list key protein targets and ligands in this article which are hyperlinked to corresponding entries in http:// www.guidetopharmacology.org, the common portal for data from the IUPHAR/BPS Guide to PHARMACOLOGY (Pawson et al., 2014) and are permanently archived in the Concise Guide to PHARMACOLOGY 2013/14 ( a,b Alexander et al., 2013a,b).  (Rogojina et al., 2003;Twarock et al., 2009;Röck et al., 2012). Data were analysed by the ΔΔCq method using GADPH as a reference gene.

Cell count and determination of synergism
Cells were seeded at 5000 cm −2 in 12-well plates. After 24 h, the medium was changed and erlotinib or gefitinib at final concentrations of 0.25, 0.5, 1, 2 and 4 μmol·L −1 as well as 4-MU at final concentrations of 50, 75, 150, 300 and 600 μmol·L −1 were added alone or in combination at a ratio of 1:300 in five dilutions beginning with 0.167 μmol·L −1 erlo-tinib plus 50 μmol·L −1 4-MU to 2 μmol·L −1 erlotinib plus 600 μmol·L −1 4-MU. After an additional 72 h, cells were trypsinized and counted in a Neubauer chamber. The effect of each dose was calculated by counting the number of cells compared with the control-treated wells. In order to obtain a linear regression coefficient of r ≥ 0.95 appropriately, the average effects of at least five independent experiments were used to simulate the median-effect plots of the drugs as single agents or in combination and to subsequently determine the combination index (CI) using the CompuSyn Software (Chou and Martin, 2005) as described by Chou (2006).

Cell cycle analysis
Cells were seeded and treated as described earlier. After trypsinization, cells were washed with PBS and permeabilized using 75 μL of 0.1% sodium citrate solution (Carl Roth GmbH) containing 0.1% Triton X-100 (Sigma, St. Louis, MO, USA) as described by Nicoletti et al. (1991). Subsequently, 25 μL of Guava® Cell Cycle Reagent (EMD Millipore Corporation, Hayward, CA, USA) was added and DNA content was measured on the Guava easyCyte TM Flow Cytometer (EMD Millipore Corporation). Histogram deconvolution was performed by ModFit LT TM Software (Verity Software House, Topsham, ME, USA).

[ 3 H]-thymidine proliferation assay
One day after seeding, cells were incubated with 4-MU, erlotinib or vehicle for 24 h, and [ 3 H]-thymidine (Perkin Elmer, Waltham, MA, USA) was added for the last 6 h at a final concentration of 0.5 μCi·mL −1 and specific activity of 2 Ci·mmol −1 . After the cell layer had been washed with cold PBS, it was harvested by 0.3 mol·L −1 perchloric acid and 0.1 mol·L −1 sodium hydroxide. After addition of Rotiszint® ecoplus scintillation mix (Carl Roth GmbH), radioactivity was counted in the Beckman LS 6000 IC scintillation counter for 3 min. Counts were normalized to total protein in the lysates, which was quantified by the Bradford method-based Bio-Rad protein assay (Bio-Rad Laboratories, Inc., München, Germany).

Migration assay
Cells were seeded at 20 000 cells per chamber in ibidi® cell culture inserts composed of two chambers separated by a 500 μm wall. After 24 h, the inserts were removed, resulting in two confluent cell monolayers separated by a defined gap. Medium containing either 1 μmol·L −1 erlotinib, 300 μmol·L −1 4-MU, a combination of both or vehicle DMSO was added followed by phase contrast time-lapse microscopy using the 5× objective of the Zeiss AxioObserver Z.1 (Carl Zeiss Micro-Imaging GmbH, Göttingen, Germany). The addition of 5 mmol·L −1 hydroxyurea prevented proliferation of the cells.
In the pictures taken every 120 min, the distance between the cell layers was measured using Zen2012 Software (Carl Zeiss MicroImaging GmbH) at the two sites of minimum and maximum distance after 24 h. The difference in the mean distances was divided by 120 min to calculate the gap closing speed within a period of 24 h or until the gap was closed.

ELISA-like HA assay
For quantification of HA in the cell culture supernatants, KYSE-410 were seeded and treated as described earlier. After 24 h of treatment, supernatants were harvested and the amount of HA was determined with the hyaluronic acid test kit (Corgenix, Westminster, CO, USA) and normalized to total protein quantified by the Bradford method, as described previously.

Statistical analysis
GraphPad Prism 6 Software Version 6.04 (GraphPad Software, Inc., La Jolla, CA, USA) was used for statistical analysis. The relative expression values obtained by qPCR were logarithmically transformed and then compared with the given expression level in the control group by one-sample t-test in case of knockdown control and ordinary one-way ANOVA with Sidak's multiple comparison test. MCTS growth curves were analysed by two-way ANOVA ( Figure 7B). Western blot data of pERK/ERK and pAkt/Akt were normalized to control samples and Kruskal-Wallis test was performed. Migration assay data did not pass the Kolmogorov-Smirnov normality test and were therefore tested using the Kruskal-Wallis test and Dunn's multiple comparisons test. Likewise, percentage values obtained in the cell cycle analysis experiments were analysed using the Kruskal-Wallis test and Dunn's multiple comparisons test. In other cases, ordinary one-way ANOVA and Sidak's multiple comparisons test were used. All groups were compared with the control group and the combined treatment group. P values < 0.05 were defined as statistically significant.

Combined inhibition of EGFR TK and HA signalling reduced the cell number
In KYSE-410 cells, treatments using either 4-MU or erlotinib or the combination resulted in decreased cell counts ( Figure 1A) compared with the control. The combination of both drugs lowered the cell count most effectively and significantly fewer cells were counted compared to the samples treated with single agents. In order to investigate if interfering with HA signalling by knockdown of the two major HA receptor CD44 and receptor for HA-mediated motility (RHAMM) would show the same effect, cells were transfected with siRNA and subsequently treated with erlotinib or vehicle. Samples were transfected in parallel and randomly checked for knockdown efficiency ( Figure 1B and D). Erlotinib treatment combined with a knockdown of CD44 significantly reduced the cell number as compared to control and single erlotinib treatment as well as compared to vehicletreated siCD44 transfected cells ( Figure 1C). In contrast, comparing erlotinib treatment of cells with and without knockdown of RHAMM, only a non-significant trend towards a further reduction of the cell number in erlotinib and siRHAMM-treated cells ( Figure 1E) was observed. Hence, inhibition of HA synthesis by 4-MU or blocking HA signalling by CD44 knockdown augmented the effect of erlotinib. In contrast to 4-MU treatment, the knockdown of CD44 and RHAMM alone did not significantly reduce the number of cells. However, the combination of 4-MU with knockdown of CD44 caused a further, albeit non-significant, decrease in the cell number ( Figure 1F).

Synergistic reduction of cell number by combined erlotinib and 4-MU treatment
To

Erlotinib and 4-MU reduced the number of cells in the S-phase
Flow cytometry analysis of propidium iodide-stained cells was subsequently used to elucidate the underlying mechanism of the decreased number of KYSE-410 by 4-MU and erlotinib treatment. No significant increase of cells in the sub-G1 phase was observed after any of the treatments, excluding an effect on apoptosis ( Figure 3A). Furthermore, the fraction of cells in the G0/G1 phase increased in the double treatment group ( Figure 3B) as compared to erlotinib or vehicle-treated cells alone. Accordingly, after 24 h of treatment, the proportion of KYSE-410 in the S-phase was significantly lower in dual erlotinib and 4-MU-treated cells compared with erlotinib or vehicle-treated cells ( Figure 3C   with combined, 4-MU and erlotinib ( Figure 3F), supporting the anti-proliferative effect of the combination.

Phosphorylation of ERK but not Akt was reduced by erlotinib and 4-MU
Western blot analysis of cells harvested after 24 h of treatment showed significantly reduced ERK phosphorylation when treated with erlotinib combined with 4-MU. Additionally, there was a strong trend towards a reduced ERK phosphorylation in KYSE-410 treated with erlotinib alone ( Figure  4A and B).  Information Fig. S1C and D). In these cases, the receptor TK inhibitors alone were able to significantly reduce ERK phosphorylation. Furthermore, treating KYSE-410 with erlotinib and erlotinib combined with 4-MU resulted in a significantly reduced ERK phosphorylation even after 20 min. At this time point, 4-MU treatment resulted in a trend towards decreased ERK phosphorylation as well (Supporting Information Fig. S1E and F). As both CD44 and the EGFR could addition- ally signal via the PI3K/Akt pathway, Akt phosphorylation was analysed. However, after 24 h of treatment, no significant reduction in Akt phosphorylation in any of the treatment groups was observed in KYSE-410 (Supporting Information Fig. S4C and D).

Erlotinib and 4-MU-treated KYSE-410 cells showed impaired migration
The impact of the treatment on cell migration was studied in addition to the anti-proliferative effects of 4-MU and erlotinib. The mean speed of gap closure was significantly lower in erlotinib plus 4-MU-treated KYSE-410 as compared to vehicle or erlotinib-treated cells ( Figure 5).

Combined inhibition of EGFR and HAS activity was effective in the 3D cell culture
The efficacy of combining 4-MU and erlotinib was studied in MCTS 3D cell culture of KYSE-410 since a 3D system potentially reflects more realistically the in vivo situation regarding metabolic gradients, drug gradients and proliferative gradients and allows for 3D cell-cell and cell-matrix interaction. Starting 4 days after seeding, MCTS were treated with 1 μmol·L −1 erlotinib, 300 μmol·L −1 4-MU or a combination of both for 10 days. As depicted in Figure 6A and B, MCTS growth was most effectively suppressed in the double treatment group. Interestingly, erlotinib and 4-MU showed varying efficacy in reducing ERK phosphorylation when comparing two-dimensional (2D) and three-dimensional (3D) cell culture formats. 4-MU led to a significant reduction in ERK phosphorylation after 24 h of treatment in the 3D cell culture, whereas in the 2D cell culture, no significant difference could be observed compared with control ( Figures 6C and 4A and B, respectively). However, the combination of erlotinib and 4-MU decreased ERK phosphorylation significantly in either setting. The significantly suppressed ERK phosphorylation by combined erlotinib and 4-MU treatment was detectable after 24 h and lasted for 10 days of treatment ( Figure 6C and D).

HAS2 mRNA expression was reduced by 3D cell culture or by erlotinib and 4-MU treatment in 2D cell culture
As 4-MU seemed to act differently in the 2D and 3D cell cultures regarding the phosphorylation of ERK, the expression of HA-related genes and of the EGFR was analysed. 4-MUtreated MCTS had significantly lower HAS2, RHAMM and EGFR mRNA expression and higher HAS3 mRNA expression compared with the 2D cell culture ( Figure 7A). In the 2D cell culture, HAS2 mRNA expression was significantly reduced in all treatment groups, and combining erlotinib and 4-MU led to the lowest HAS2 mRNA expression ( Figure 7B). Treatment with erlotinib and erlotinib in combination with 4-MU reduced the mRNA expression of HAS3 ( Figure 7C). As a result of reduced HAS mRNA expression by erlotinib treatment, decreased amounts of HA in cell culture supernatants were detected ( Figure 7E). Additionally, CD44 mRNA expression was reduced by 4-MU and erlotinib alone or in combination.
In contrast to the 2D cell culture, HAS2 and CD44 mRNA expressions were not significantly reduced in the 3D cell culture by any treatment (Figure 7F and H). HAS3 mRNA expression was even increased by the combined erlotinib and 4-MU treatment ( Figure 7G). The lack of HAS2 mRNA reduction in the 3D cell culture may be explained by the already low HAS2 mRNA expression in MCTS compared with the 2D cell culture ( Figure 7A). The mean relative HAS2 mRNA expression of control-treated MCTS was found to be 0.69 (±0.085 SEM)-fold of that in the combined erlotinib and 4-MU-treated cells in the 2D cell culture (data not shown).

Discussion and conclusions
Erlotinib is used in advanced or metastatic non-small-cell lung cancer for treatment after chemotherapy and it has recently been approved for first-line treatment of metastatic non-small-cell lung cancers that are characterized by activating EGFR exon 19 deletions or exon 21 substitution mutations. In this patient subgroup, erlotinib treatment was superior to standard chemotherapy. This was not observed in other large trials without stratification for these mutations (Khozin et al., 2014). In ESCC, EGFR mutations are rare (Liu et al., 2011;Gonzaga et al., 2012;Kato et al., 2013) and erlotinib has shown only modest efficacy in clinical trials. Our experiments showed that 4-MU may be an eligible candidate for combination with erlotinib. As an approved choleretic drug, it is well tolerated (Gonzalo-Garijo et al., 1996). We investigated the efficacy of the combination of erlotinib and decreased HA signalling on proliferation, cell migration, MAPK signalling and its efficacy in the 3D cell culture.
Most markedly reduced cell numbers and migration were observed after treatment with the combination of 4-MU and erlotinib. It has previously been demonstrated that both RHAMM and CD44 are involved in proliferative signalling (Twarock et al., 2010). Additionally, both CD44 and RHAMM may interact with growth factor receptors such as EGF and PDGF receptors (Turley et al., 2002;Toole, 2009). This may contribute to an at least by trend augmented effect of erlotinib observed in siCD44 and siRHAMM-transfected cells ( Figure 1C and E), although siRNA targeting CD44 or RHAMM alone had no significant effects on cell counts. Alternatively, 4-MU might affect cell growth by additional mechanisms independent of HA synthesis (Nakamura et al., 2007;Edward et al., 2010).
We further investigated the underlying mechanism of the reduced cell number after combined treatment. For this purpose, the cell cycle was analysed and [ 3 H]-thymidine incorporation was quantified after 24 h of treatment. As the proportion of cells in the sub-G1 phase did not change but the proportion of cells in the S-phase was reduced, we concluded that the proliferative activity of cells was decreased and that apoptosis was not affected in KYSE-410, KYSE-270 and KYSE-520. In line with our findings, Sutter et al. (2006) did not observe apoptotic effects in ESCC cells treated with erlotinib. However, other investigations report a proapoptotic effect of 4-MU (Lokeshwar et al., 2010;Urakawa et al., 2012) or erlotinib (Fichter et al., 2014).
The Akt and ERK pathways are prominent downstream pathways of EGFR and HA signalling and 4-MU alone has been shown to reduce ERK phosphorylation in OSC1 cells (Twarock et al., 2010). In our setting, 4-MU alone was only able to consistently reduce ERK phosphorylation in the 3D cell culture (Figure 6). Also, the strong additional effect of the combination, compared to EGFR TK inhibition alone, on ERK phosphorylation was not seen in the 2D cultures. This may point to an important role of extracellular matrix molecules such as HA in the 3D cell structures. Moreover, 4-MU or erlotinib was reported to additionally decrease Akt phosphorylation (Lokeshwar et al., 2010;Arai et al., 2011;Urakawa et al., 2012;Fichter et al., 2014), which could not be detected in the present experimental set-up.
Furthermore, it is important to test the efficacy of combination of erlotinib and 4-MU in a 3D model of intermediate complexity that is characterized by cellular heterogeneity, nutrient and oxygen gradients and which allows for 3D cell-cell interactions and a 3D extracellular matrix arrangement. Therefore, we established an MCTS growth assay as a model of, that is, tumour micro-regions or micro-metastases (Vinci et al., 2012). Additionally, there may be different efficacies of drugs in monolayer cell culture as compared to the 3D cell culture (Friedrich et al., 2009). In the experiments reported here, not only ERK phosphorylation but also gene expression was differentially regulated in the 3D cell culture. For example, the reduction in HAS2 and CD44 mRNA expression that was observed after 4-MU treatment in other settings (Kultti et al., 2009;Lokeshwar et al., 2010) was only seen in the 2D cell culture. Further research is required to reveal the underlying mechanisms of the differences in ERK activation after 4-MU treatment. Possibly, ERK activation is more sensitive to changes in HA concentration in an environment of low HAS2, RHAMM and EGFR expression that was observed in the 3D cell culture. Importantly, a comparison of the anti-cancer activities of the drugs in the 2D and 3D cell culture, as measured by the cell number or MCTS volume, showed that single agents had moderate but the combination strong effects in both settings.
In order to determine if 4-MU and erlotinib acted in a synergistic or in an additive way, we calculated the CI as described by Chou (2006). We chose the initial cell count experiment as a key finding to evaluate the mutual action. In order to achieve acceptable linear regression coefficients, we used the average effect values of independent experiments to calculate median-effect plots and the CI. In cancer studies, the CI values corresponding to high effects such as ED 95 are of special interest. In our setting, the value for ED95 was calculated to be 0.58 in KYSE-410, 0.59 in KYSE-270 and 0.43 in KYSE-520, thus being in the range of 0.3-0.7, which is defined to describe synergism (Chou, 2006).
The mechanism of the synergistic action of 4-MU and erlotinib remains to be elucidated. In the 2D cell culture, treatment with erlotinib alone effectively decreased ERK phosphorylation even though it was not decreased significantly in all settings investigated. At an earlier time point, 4-MU also tended to inhibit this pathway, which may contribute to the synergism between 4-MU and EGFR inhibition. Furthermore, in some cell lines, the HA-receptor CD44 could interact and activate the EGFR (Toole, 2009) and it may additionally interact with the PDGFR, c-Met, ErbB2 and TGFR (TGFBR1) or cytoskeletal proteins such as ankyrin or adaptor proteins and GTPases, such as, RhoA (Toole, 2009;Misra et al., 2011). Moreover, other receptors such as RHAMM can contribute to HA signalling. A synergistic effect due to a more pronounced inhibition of HA synthesis by combining 4-MU and erlotinib seemed to be unlikely as the amount of HA in cell culture supernatants was not significantly lower in the double treatment group compared to single treatments. Additionally, an HA-independent mechanism of 4-MU action cannot be excluded. Hence, there are several possible mechanisms that could be responsible for the observed synergistic inhibitory effects of erlotinib plus 4-MU on cell number.
The results of the present study show that treatment of ESCC cells with the combination of the EGFR inhibitor erlotinib and the HA synthesis inhibitor 4-MU profoundly inhibits the proliferation and migration of three different ESCC cell lines as well as 3D MCTS growth. Therefore, 4-MU may be a promising agent to increase erlotinib efficacy in ESCC.

Supporting information
Additional Supporting Information may be found in the online version of this article at the publisher's web-site: http://dx.doi.org/10.1111/bph.13240 Data are presented as mean ± SEM; *P < 0.05 compared to control.