Introduction
Colorectal cancer ranges among the three most frequent malignancies in Western countries [
1,
2]. Survival is determined by local recurrence, lymphatic, and hematogenous dissemination [
3]. Due to improved therapeutic strategies, the overall survival in stage IV colorectal cancer has increased from 8 months to more than 2 years during the last decade [
4‐
6].
Besides new chemotherapeutic drugs, such as platinum derivates (oxaliplatin) and topoisomerase II inhibitors (irinotecan), the introduction of biologicals targeting tumor neovascularization or growth signaling significantly has improved response and prognosis [
4‐
6].
Specific mutations in tumor-suppressor genes (APC, DCC, p53) and oncogenes (K-ras) occur during the adenoma–carcinoma sequence of colorectal cancer [
7‐
9]. The K-ras mutation status was the first key to personalized therapy in colorectal cancer, as anti-EGFR strategies were shown to be effective in K-ras wild types only [
10].
Receptor tyrosine kinases (RTKs) are transmembrane receptors containing extracellular ligand-binding domains connected to intracellular catalytic domains [
11]. The growth factors VEGF/PDGF/EGF and their receptors VEGFR1-3, PDGFRα/β, and EGFR are critical in the process of (lymphatic) neo-angiogenesis and dissemination in human cancer [
12‐
16]. Inhibition of RTKs with sorafenib has been successful in renal and hepatocellular cancers [
17,
18]. Two phase I studies revealed a disease stabilization in pretreated colorectal cancer patients receiving sorafenib in combination with either irinotecan or oxaliplatin [
19,
20]. However, recent phase II/III studies testing other multi-tyrosine kinase inhibitors in human colorectal cancer have failed to show any benefit [
21]. So far only one randomized Phase III study with Regorafenib improved survival times after failure of all approved standard therapies [
22]. Therefore, the impact of combinational therapies (sorafenib + chemotherapy) remains controversial. Preclinical data as well as experimental data explaining interaction mechanisms are widely missing. Thus, we initiated this study to examine sorafenib targeted RTK expression and to analyze the in vivo effect of sorafenib alone or in combination with the classical chemotherapeutic backbone 5-fluorouracil (5-FU).
Material and methods
Cell culture
The human colorectal cancer cell lines SW480 [K-ras mt, B-Raf wt, PI3K wt, p53 mt], SW620 [K-ras mt, B-raf wt, PI3K wt, p53 mt], and HT29 [K-ras wt, B-raf mt, PI3K wt, p53 mt] were cultured in RPMI 1640 (Invitrogen, Germany) supplemented with 10 % FCS, 100 U/ml penicillin, 100 μg/ml streptomycin (Cambrex, Germany), and 1 mM l-glutamine (Invitrogen, Germany). The human colorectal cancer cell line Caco2 [K-ras wt, B-Raf wt, PI3K wt, p53 mt] was cultured in DMEM (Invitrogen, Germany) supplemented with 10 % FCS, 100 U/ml penicillin, 100 μg/ml streptomycin (Cambrex, Germany), and 1 mM l-glutamine (Invitrogen, Germany).
Proliferation assays and chemosensitvity
For proliferation assays, 5 × 103 SW480, SW620, Caco2, or HT-29 cells were plated in 96-well plates and cultured as described above. Twelve hours after plating sorafenib (0, 5, and 10 μg/ml), 5-FU (0.5 mg/ml) ± sorafenib (5 μg/ml), irinotecan (1 mg/ml) ± sorafenib (5 μg/ml), or oxaliplatin (0.5 mg/ml) ± sorafenib (5 μg/ml) were added to the medium. The amount of cells per well was determined by luminescence assay (CellTiter-Glo Cell Viability assay, Promega, USA). Each condition was performed in quadruplicates.
For apoptosis analyses, 2 × 105 cells were seeded per 6 wells, respectively. Twelve hours after plating, cells were treated for 24 h as mentioned above. Suspended cells were collected, and adherent cells were trypsinized prior to fixation with 100 % ethanol, staining with propidium iodide and analyzation by FACS without gating. Each condition was performed in quadruplicates.
Migration assay
SW480, SW620, Caco2, or HT29 cells (2 × 106 ) were seeded per 6 wells, cultured for 24 h, serum-starved (2 % FCS only) for 12 h, and exposed to sorafenib at different concentrations (0, 5, or 10 μg/ml) for 6 h. Migration was assayed with 24-well HTS FluoroBlock Inserts in triplet approaches (8 μM pore size; Becton Dickinson, USA).
In brief, 4×104 cells were resuspended in RPMI1640/DMEM medium containing 2 % FCS and 10 ng/ml CXCL12 and added to the upper chamber. Subsequently, RPMI1640/DMEM medium with 20 % FCS and 100 ng/ml CXCL12 (Sigma, Germany) was added to the lower chamber. Chambers were incubated for 24 h at 37 °C in a humid atmosphere of 5 % CO2. After incubation, the amount of migrated cells in the lower chamber was determined by luminescence assay (CellTiter-Glo, Cell Viability assay, Promega, USA). Each condition was performed in triplicate.
Caspase assay
Cells were treated with placebo or sorafenib (5 and 10 μg/ml, respectively). After incubation for 16 h, cells were lysed in buffer containing 20 mM Tris/HCl pH 8.0, 5 mM EDTA, 0.5 % Triton X-100, and onefold complete protease inhibitor cocktail (Roche, Germany). Protein concentration was determined by Bradford assay (Sigma, Germany). Sixty micrograms of protein was incubated in reaction buffer (25 mM HEPES pH 7.5, 50 mM NaCl, 10 % glycerol, 0.05 % CHAPS, and 5 mM DTT) in the presence of 50 μM fluorogenic substrate (Biomol, Germany), which was specific for caspase 3 (DEVD-AMC); caspase 6, 8, and 10 (Ac-IETD-AFC); or caspase 9 (Ac-LEHD-AFC). Analyses were performed in triplicates.
Assays were performed in black micro-titer plates (Nunc, Germany), and after 1 h incubation at 37 °C, the generation of free AMC or AFC was measured using a fluorometer plate reader (Appliscan, Thermo Fisher, Germany) at an excitation wavelength of 380 nm (AMC and AFC) and an emission wavelength of 460 nm (AMC) or 505 nm (AFC).
Western blot analysis
SW480, SW620, Caco2, or HT29 cells (2 × 10
6) were harvested after a 12-h-long exposition to placebo or sorafenib (5 and 10 μg/ml, respectively). Cells were washed twice with phosphate-buffered saline (PBS; 1×) and lysed in 2× RIPA solution. For Western blot analysis, 100 μg of protein was loaded on 8–12 % SDS-PAGE gels, respectively. After separation, the gel was transferred to a PVDF membrane (Roth, Karlsruhe, Germany). Proteins (AKT/pAKT, MEK/pMEK, PI3K/pPI3K, mTOR/pmTOR, P53/pp53, FoxO3a/pFoxO3a, GADD45β, and alpha tubulin) were detected with specific primary antibodies (Table
1; 4 °C, overnight). The specific secondary antibodies were exposed for 1 h at room temperature (Table
1). For visualisation, the Roti Lumin systems 1 and 2 were applied (P79 and P80, Roth, Karlsruhe, Germany). Each condition was performed in duplicates.
Table 1
Antibodies used for Western blotting
Rabbit-anti-human pPI3K | Cell Signaling | 4228 | Goat-anti-rabbit IgG | 85/60 | 1:1,000 |
Rabbit-anti-human PI3K | Cell Signaling | 4257 | Goat-anti-rabbit IgG | 85 | 1:1,000 |
Rabbit-anti-human pAKT | Cell Signaling | 9267 | Goat-anti-rabbit IgG | 60 | 1:1,000 |
Rabbit-anti-human AKT | Cell Signaling | 4685 | Goat-anti-rabbit IgG | 60 | 1:1,000 |
Rabbit-anti-human pmTOR | Cell Signaling | 2971 | Goat-anti-rabbit IgG | 289 | 1:1,000 |
Rabbit-anti-human mTOR | Cell Signaling | 2983 | Goat-anti-rabbit IgG | 289 | 1:1,000 |
Rabbit-anti-human pMEK | Cell Signaling | 9121 | Goat-anti-rabbit IgG | 45 | 1:1,000 |
Rabbit-anti-human MEK | Cell Signaling | 9122 | Goat-anti-rabbit IgG | 45 | 1:1,000 |
Goat-anti-human GADD45β | Santa Cruz Biotechnology | sc-8776 | Donkey-anti-goat IgG | 18 | 1:500 |
Mouse-anti-human α-Tubulin | Sigma Aldrich | t5168 | Goat-anti-mouse IgG | 48,5 | 1:2,000 |
Goat-anti-mouse IgG | Santa Cruz Biotechnology | sc-2031 | – | – | 1:10,000 |
Goat-anti-rabbit IgG | Santa Cruz Biotechnology | sc-2030 | – | – | 1:10,000 |
Donkey-anti-goat IgG | Santa Cruz Biotechnology | sc-2033 | – | – | 1:10,000 |
Subcutaneous xenograft tumor system
HT29 tumor cells (1 × 107) were suspended in 0.2 ml pure RPMI1640 medium and 1× PBS (1:1) and applied by subcutaneous injection into the left flank of 7–8-week-old female nod-SCID mice. Nod-SCID mice were irradiated with 1.8 Gy 1 day prior to s.c. injection of tumor cells. As soon as the tumors reached a size of 10 mm, animals received i.p. injections of placebo (group 1; 200 μl, 5 days/week; 25 % cremophor in NaCl 0,9 %), sorafenib (group 2; 200 μl; 5 days per week; 0.12 mg/dose solved in 25 % cremophor; 30 mg/kg/week), 5-FU (group 3; 200 μl; three times a week; 0.18 mg/dose solved in 25 % cremophor; 25 mg/kg/week) or sorafenib + 5-FU (group 4; 200 μl; combination of group 2 and 3). The size of tumors was measured manually twice weekly. Tumors grew for 4 weeks. Thereafter, tumor nodules were excised and measured manually with a vernier micrometer.
Immunohistochemistry
Excised tumors obtained from the experimental animals were paraffin-embedded. After obtaining adequate slides, the tissue samples were screened for Ki-67, PDGFA, VEGFA, VEGFR1, VEGFR2, PDFGRα, and PDGFRβ protein expression by immunohistochemistry. To that purpose, the tissues were deparaffinized, rehydrated, and subsequently incubated with the respective primary antibodies [anti-PDGFRα (sc-338); 1:200, 2 h, Santa Cruz Biotechnology, CA, USA; anti-PDGFRβ (3169), 1:40, 2 h, Cell Signaling Technology, MA, USA; PDGFA (NBP1-19781), 1:100, 2 h, Novus Biologicals, Cambridge, UK; VEGFA (ab46154), 1:200, 2 h, Abcam plc, Cambridge UK; VEGFR1 (RB-9049-R7), 1:50, 2 h, Thermo Fisher Scientific GmbH Neomarkers, Germany; VEGFR2 (RB-9239-R7), 1:50, 2 h, Thermo Fisher Scientific GmbH Neomarkers, Germany; VEGFR3 (sc-321), 1:200, 2 h, Santa Cruz Biotechnology, Germany; Ki-67 (mib1), 1:100, 2 h, Dako, Germany; Envision flex plusTM, Autostainer, Dako, Germany]. The secondary antibody (anti-rabbit-mouse-goat antibody) was incubated for 15 min at room temperature, followed by incubation with streptavidin-POD (Dako, Germany) for 15 min. Antibody binding was visualized using AEC solution (Dako, Germany). Afterwards, the tissues were counterstained by haemalaun solution (Dako, Germany). The expression of the respective tyrosine kinase was evaluated using a scoring system. Expression strength of PDGFA, VEGFA, VEGFR1, VEGFR2, VEGFR3, PDFGRα, and PDGFRβ was classified as negative (0), low (1), medium, (2) and high (3). All slides were independently evaluated by three investigators. The Ki-67 expression was measured as percentage of Ki-67 expressing cells.
Statistics
In order to assess dependence of growth factor and Ki-67 expression on treatment with 5-FU and sorafenib, the minimum, the maximum, the median, and the quartiles in subgroups were calculated. For Ki-67 analyses, the mean and standard deviations were calculated and displayed in box plots. Ki-67 was measured three times for each specimen; averages were analyzed using two-way analysis of variance. To compare growth factor expression between treatment groups the Kruskal–Wallis tests was used, followed by pairwise Wilcoxon test if the Kruskal–Wallis test gave a p value ≤0.05.
All tests were performed with exploratory intention, associations with p values ≤0.05 might warrant further consideration. Statistical analysis was performed using SAS 9.3 2002–2010 by SAS Institute Inc., Cary, NC, USA.
Discussion
The approach of inhibiting RTKs with sorafenib has been successful in renal and hepatocellular cancers [
17,
18]. A phase I study revealed disease stabilization in pretreated colorectal cancer patients [
20]. Except of one recent study with Regorafenib, recent phase II/III studies testing other multi-tyrosine kinase inhibitors in colorectal cancer failed to show any benefit [
21,
22]. So far, no molecular markers have been identified which are helpful in stratifying the patients.
We performed defined functional in vitro analyses in order to identify sorafenib-sensitive and sorafenib-resistant cell lines. While HT29 and SW480 were found to be sorafenib sensible, Caco2 was resistant and SW620 showed features of resistance. However, the mutation status of K-ras, B-Raf, PI3K, or p53 did not correlate with resistance.
Combining sorafenib with chemotherapeutic drugs used in colorectal cancer revealed an additive effect in growth inhibition and apoptosis induction in SW480 (except for oxaliplatin) and HT29 cells, whereas in Caco2 cells, apoptosis was not increased and proliferation even stimulated (5-FU or oxaliplatin). These data are in line with previous reports describing a reduced cellular uptake of oxaliplatin and generation of DNA adducts in specific colorectal cancer cells through sorafenib [
23]. Thus, combination with oxaliplatin seems disadvantageous in specific settings. The effect of sorafenib on migration was marginal and of no significant importance.
Induction of apoptosis might explain the different observations made upon sorafenib exposure: While activity of caspases 6, 8, and 10 was induced in sensitive SW480 cells, it was decreased in the resistant cell lines. Furthermore, SW480 reacted with an increased activity of caspase 9. In contrast, activity of caspase 3 was decreased in Caco2 cells upon exposure to sorafenib. An induction of caspase 3 activity, as seen in prostate cancer cells, was not observed in colorectal cancer cells [
24]. Our data reveal that resistance to sorafenib is associated with inhibition of specific pro-apoptotic pathways. However, sorafenib is also known to induce caspase-independent apoptosis, mediated through nuclear translocation of
AIF [
25].
We observed an inhibition of the Ras–Raf pathway (
pMEK) in SW620 cell lines only, matching sorafenib’s function as a Raf inhibitor [
25]. While sensitive cell lines revealed only a weak–absent
pAKT expression,
AKT expression was clearly suppressed upon exposure with increasing sorafenib doses. In contrast, the resistant cell line Caco2 did not show such
AKT suppressive behavior. These observations match a previous report that a constitutively active
AKT protects cells against sorafenib/bortezomib-induced apoptosis [
26,
27].
Sorafenib-sensitive cells lines were defined by almost absent pAKT, medium–strong FoxO3a, and hint GADD45β levels. The tumor suppressor FoxO3 belongs to a subclass of the forkhead transcription factors, being inhibited by activation of the PI3K pathway. Downregulation of FoxO3 is thus considered a consequence of pAKT activity.
In contrast, resistant cell lines showed medium
pAKT, weak
FoxO3a, and very intense
GADD45β levels.
GADD45β expression levels discriminated best between sensitive and resistant cell lines.
GADD45 is a stress sensor modulating the response of cells to genotoxic or oxidative stress [
28‐
30]. In specific colon cancer cells,
GADD45β over-expression was linked to protection from platin induced death, matching our observations [
31]. Being an apoptosis modulator, activation of
GADD45β prevents the propagation of damaged cells, causing an arrest in cell growth and apoptosis after exposure to toxins [
32]. This regulation seems intact in SW480 cells but reversed in resistant cells;
GADD45ß was downregulated in Caco2 upon sorafenib treatment, going along with a sorafenib-mediated inhibition of caspases 6, 8, and 10. As a downstream effector of
p53,
GADD45β was confirmed to be specifically downregulated in HCC, which was associated to the extent of
p53 mutation [
33]. We observed a
pp53 and a
GADD45β upregulation in some sensitive cell line (SW480) upon exposure to sorafenib. In contrast, resistant cell lines showed primarily high (Caco2, SW620) and, upon sorafenib exposure, decreasing (Caco2)
GADD45β and
pp53 levels. These data are in line with observations in HCC, in which
GADD45β induction by sorafenib occurred only in sensitive hepatocellular carcinoma cell lines, independent of the
Raf/MEK/ERK signaling pathway [
34].These findings confirm our definition of sensitive cell lines, in which sorafenib induces apoptosis and inhibits proliferation.
In vivo, Wilhelm and colleagues described a potent growth inhibition of HT29 xenografts at sorafenib doses of 7.5 mg/kg. We studied four different groups in vivo: placebo, 5-FU, sorafenib, and 5-FU + sorafenib. 5-FU was chosen, being the backbone of most chemotherapeutic protocols in colorectal cancer. Sorafenib was applied at 5 mg/kg, matching 400 mg/day as used in combination therapies [
19,
20].
Interestingly, we observed that a sorafenib monotherapy was at least equally effective as the 5-FU monotherapy or as the combination therapy and even tended to inhibit in vivo tumor growth somewhat better than the combination therapy.
The proliferation index was significantly reduced in all treatment groups as compared to the control group but displayed similar results for mono-agent therapy and the combination therapy. Since only small numbers were analyzed, a possibility exists that larger treatment groups might demonstrate even more distinct differences. However, we clearly demonstrate that combination of sorafenib and chemotherapy did not result in any additive effects. In contrast, it seems that treatment effects are partially cancelled when 5-FU and sorafenib are applied simultaneously.
Expression rates of receptor tyrosine kinases VEGFR1 and PDGFRβ as well as of the ligand PDGFA were decreased by all treatment regimens used. However, no significant differences were detected between treatment groups.
Inhibition of receptor tyrosine kinases through sorafenib could potentially lead to a selection of low target expressing tumor cells. Combination regimens of sorafenib and 5-FU might reduce sorafenib target expression leading to a similar proliferation effect as under 5-FU monotherapy. However, the adverse events in humans might rather be additive. Our results indicate that there is no additive effect in combination of these two treatment mechanisms and that combination might only add adverse events. Therefore, in future studies preferentially sorafenib monotherapy versus sequential treatment regimens (inductiontherapy via chemotherapy–maintenance via sorafenib) should be explored.