Background
Cancer is the leading cause of death worldwide, accounting for around 13 % of all fatalities, with mortality numbers continuing to rise. Cancer therapy is one of the fastest growing segments in the pharmaceutical industry, which itself is undergoing a major shift from conventional chemotherapy and radiotherapy to specific agents aimed at targeting abnormalities that are specific to cancer cells. Amongst these are dysregulations of crucial growth factor cascades.
Fibroblast growth factors (FGFs) are pivotal regulators to key cellular processes, and their deregulation can lead to excessive proliferation, evasion of apoptosis, invasive behavior as well as aberrant angiogenic responses, all hallmarks of cancer. FGFs and their receptors (FGFRs) have been shown to be oncogenes in mouse tumor models [
1] and their over-expression or over-activation have been observed in many cancers [
2‐
4]. Of the over 500 pathways that have been implicated in oncogenic cascades, the kinases in FGF signaling are the most commonly mutated [
5]. FGFs and FGFRs have thus become important targets for cancer intervention.
Research laboratories and pharmaceutical companies have been developing FGFR tyrosine kinase inhibitors (TKIs), antibodies against FGFRs and inactive receptor decoys as targeted cancer therapeutics. Several of the TKIs are undergoing clinical trials. However, most of these compounds are nucleotide analogs that have multiple targets due to similarity in structure of the intracellular kinase domains of FGFRs with vascular endothelial growth factor receptors (VEGFRs) and platelet-derived growth factor receptors (PDGFRs), as well as off-target inhibition of other proteins. Thus, recent efforts have focused on developing more specific FGFR TKIs, as well as neutralizing antibodies that are more specific than chemical inhibitors, particularly antibodies that target individual FGFR isoforms, so avoiding pan-FGFR inhibition [
6]. Antibodies against FGFRs have been reported to inhibit tumor growth in xenograft models [
7‐
9].
The family of FGFRs, consisting of four members (FGFRs1-4), are highly conserved, sharing 55–72 % amino acid sequence with each other [
10]. FGFRs1 - 3 are widely expressed in adult human tissues, with FGFR4 exhibiting more limited tissue distribution [
11]. The extracellular region of most FGFRs contains three immunoglobulin-like motifs (Ig domains). Alternative splicing of FGFRs1-3 mRNA transcripts within the IgIII domain results in two isoforms, IIIb and IIIc, which have altered affinity for FGFs and exhibit tissue-specific expression. The IIIb isoform is expressed mainly in epithelium and the IIIc in mesenchyme [
12]. A heparin-binding domain, essential for interactions with heparin/heparan sulfate (HS), is located between IgII and IgIII [
13]. FGFs are known to bind to the IgII and IgIII domains on FGFRs [
14], triggering dimerization that subsequently activates intracellular tyrosine kinase activity and downstream signaling.
Ubiquitously distributed on cell surfaces and within the pericellular and extracellular matrix (ECM), HS is a highly complex molecule consisting of varying lengths of sulfated polysaccharides, some of which regulate growth factor signaling in a relatively specific manner through association with heparin-binding domains located on the ligand and/or receptor [
15,
16]. HS is required to stabilize the FGF-FGFR interaction [
17,
18] through the formation of an FGF:HS:FGFR ternary complex [
14,
19].
Active mutations of FGFR2 or FGFR3 are common in various cancers [
20], whereas the oncogenic impetus of FGFR1 results mainly from either gene amplification or overexpression of wild-type receptor [
21‐
23]. FGFR1 overactivation occurs frequently in breast [
24] and prostate cancer [
25], with changes also reported in oral squamous carcinoma, ovarian cancer and bladder cancer [
15,
26,
27]. In this work we hypothesized that FGFR1 signaling might be most effectively inhibited by blocking its interaction with HS. An antibody (IMB-R1) was thus raised to mask the heparin-binding domain on FGFR1. It proved to be particularly effective in suppressing FGF signaling, inducing strong apoptosis in cancer cells with tissue-specific potency. Our study supports the idea that FGFs/FGFRs are crucial for the survival of many cancer cells, and that targeting FGFR1 is a rational and effective strategy for therapy. More importantly, a novel strategy has been opened up for the targeted therapy of many other growth factor/receptors, that of preventing heparan sulfate/protein interactions.
Discussion
The growth of many aggressive tumors involves FGF signaling which in turn is dependent on HS as an essential ‘co-receptor’. Here a highly specific FGFR1 antibody, IMB-R1, was engineered in an attempt to prevent the engagement of endogenous HS with FGFR1, so preventing formation of an active FGF-HS-FGFR1 signaling complex. The antibody recognizes specific epitopes that are adjacent to the unique heparin-binding domain within FGFR1, so disrupting the sugar-receptor association. Thus, our approach differs markedly from other strategies that have sought to block ligand-receptor interactions by concentrating on the ligand-binding site. Our approach not only yields a novel agent for preclinical testing, but a novel way of blocking all HS-dependent interactions involved in carcinogenesis.
Characterization of the mechanism of action of IMB-R1 demonstrated that it selectively affects cancer cell survival by preventing the formation of FGF/FGFR1 complexes, thereby inhibiting the FGF/FGFR1 signaling axis. This is particularly important because heparin is known to facilitate FGF and FGFR1 dimerization [
40], which has resulted in attempts to modulate heparin’s action through competition with other carbohydrate compounds [
41‐
44]. Rather than compete against heparin activity, here we sought to block heparin binding and subsequent ligand/receptor dimerization by using an antibody approach.
FGFR1 is one of the most widely expressed FGFRs [
11], and here we confirmed that FGFR1 is strongly expressed in both normal and cancer tissues. We note that treatment with IMB-R1 may result in altered levels of FGF-R1 expressed on the cell surface by interfering with FGF-2/FGFR-1 binding and subsequent receptor endocytosis (internalisation) [
45]. FGFR1 mRNA and/or protein levels are both significantly elevated in distinct cancers (e.g., breast and bone cancer), suggesting that the FGFR1 receptor is a rational target for therapeutic intervention. We found that normal mammary epithelial cells are less sensitive to IMB-R1 than breast cancer cells, which in turn suggests an antibody-based regime for preferential elimination of breast cancer cells with minimal harm towards normal mammary cells at the appropriate dosage is possible. However, this needs to be balanced against the finding that normal osteoblastic cells are more sensitive to IMB-R1 than osteosarcoma cells. Thus, the antibody acts in a cell and tissue type -specific manner, presumably due to differences in the relative expression of FGFR1. The latter result, obtained with cultured cells, provides a counter-indication for its utility
in vivo. Systemic administration of the IMB-R1 antibody might exacerbate the osteolysis induced by metastatic breast cancers. On the other hand, local delivery of the antibody into mammary gland tumors might prove effective in constraining growth and thus offer a less invasive therapy for inhibition of the primary breast tumor mass.
We have used a variety of methods to determine the mechanism by which the IMB-R1 antibody affects cell survival. The cDNA microarray expression analysis indicates that IMB-R1 consistently down-regulates the expression of proteins that contain selenocysteine (GPX1, SEPP and SEPW), which are an important component of antioxidant defense in human cells [
46]. In view of their protective role for tumor cells against oxidative insult, and their overexpression in many cancer cells, selenoproteins have recently become targets for anti-cancer therapies [
47]. Suppressing selenoprotein expression leads to increased oxidative stress in tumor cells and subsequent apoptosis, which may also contribute to the observed cancer cell growth arrest and death induced by IMB-R1. Collectively, our data suggest that suppression of antioxidants to induce apoptosis and inhibition of FGFR1-dependent mitogenic activity may together represent two principal mechanisms for the anti-oncogenic effects of the IMB-R1 antibody.
As well as inactivation of selenoproteins, the data also reveal that IMB-R1 increases the expression of the HMOX1, NQO1 and GCLM. These and other genes are downstream targets of the transcription factor NRF2, which is a master transcriptional regulator of the oxidative stress response. In normal cells, NRF2-mediated mitigation of oxidative stress prevents cancer initiation and progression by removing excessive ROS, so preventing DNA damage and avoiding spontaneous mutations [
48,
49]. However, one caveat for these beneficial cytoprotective effects is that overactivation of NRF2 can be oncogenic, through the survival of cancer cells that have accumulated oxidative damage and genetic mutations [
50,
51]. Importantly, FGF signaling is known to activate NRF2 to provide cytoprotection against oxidative stress [
52]. Therefore, if our hypothesis is correct, IMB-R1 should decrease the expression of NRF2-dependent genes.
To explain the paradoxical effects of IMB-R1 on selenoproteins and NRF2 target proteins, we suggest that the opposing expression profiles for the two classes of anti-oxidant proteins may reflect regulatory feedback. The IMB-R1 antibody might initially mediate suppression of selenoproteins, triggering oxidative stress, which in turn might induce NRF2 and activate the program of NRF2 - responsive genes (HMOX1, NQO1 and GCLM). It appears that the reduction in selenoprotein expression prevails, so inducing apoptosis upon IMB-R1 treatment. Hence our data indicates that a key survival function of FGFR1 is to support synthesis of selenoproteins. The universal alteration of different groups of antioxidant genes in opposing directions further indicate that IMB-R1- induced cell death is a net outcome of unbalanced oxidative insult vs antioxidative defense.
In specifically blocking signaling of FGF2/HS complexes through FGFR1, IMB-R1 selectively affects cancer cell survival and exhibits reduced non-specific toxicity compared to chemical pathway inhibitors. This set of attributes compares favorably with those of other FGFR inhibitors, including SU5402 [
53] and PD173074 [
54], both of which tend to be indiscriminately toxic to both normal and cancer cells. The efficacy of IMB-R1 also compares favorably to the commercial neutralizing FGFR1 antibody, MAB765 that failed to reduce the basal growth of cancer cells. One limitation of this particular antibody is that it is directed against the FGFR1 IIIb isoform, which is preferentially expressed in epithelial cells. However, MAB765 does not antagonize the activity of the IIIc isoform, the form which is expressed prominently in mesenchymal cells. In contrast, IMB-R1 recognizes both isoforms, so offering inhibition of FGFR1 signaling in cancers of either epithelial or mesenchymal origin. IMB-R1 differs from other existing FGFR1-neutralizing antibodies in that it expressly disrupts HS-FGFR1 interactions, highlighting the importance of targeting heparin-binding sites as a potential anti-cancer strategy.
Methods
Chemicals and inhibitors
SU5402, Staurosporine and U0126 were obtained from Merck. PD173074, protease inhibitor cocktails and other chemicals were purchased from Sigma-Aldrich.
Cell culture
Cells were purchased from ATCC and maintained in the corresponding recommended medium, except human osteosarcoma cells (OS1) [
55] that were cultured in DMEM (1000 mg/L glucose) supplemented with 10 % FCS, 2 mM L-glutamine, 25 mM HEPES (Biopolis Shared Facility, A*STAR, Singapore) and antibiotics. Media changes were performed every 2–3 days.
Taqman real-time quantitative PCR analysis
Cells were grown in triplicates and treated as indicated. The mRNA expression of target genes were analysed using the Taqman® real-time PCR method as described previously [
56]. Primers and probes were all pre-designed by Applied Biosystems.
Western blot analysis
Cells were treated as indicated and lysed in Laemmli buffer at 95 °C for 5 min. The denatured protein lysates (~20 μl) were separated by sodium dodecyl sulfate polyacrylamide gel electrophoresis and proteins transferred to nitrocellulose membranes. The blots were divided into three to five horizontal strips guided by protein standards stained by Ponceau Red to permit analysis of multiple proteins from the same sample without antibody stripping. Thereafter membranes were immunoblotted, protein targets visualized and their levels quantified as described previously [
56]. The p21 antibody was obtained from BD Biosciences. The antibodies against FGFRs or p53 were purchased from Santa Cruz. FGFR1 antibody (#MAB765) was from R&D Systems. All other antibodies were supplied by Cell Signaling Technology.
Antibody engineering
The peptide SSSEEKETDNTKPNR, located immediately upstream of the heparin-binding domain of FGFR1, was chosen as the antigen for the production of rabbit polyclonal FGFR1-neutralising antibodies as described previously [
56]. The rabbit antiserum was designated as IMB-R1, and was further affinity-purified using Reacti-Gel beads (Thermo Scientific) coupled with the above peptide. With this method we obtained two purified polyclonal antibodies, IMB-R1A and IMB-R1B, from two rabbit sera.
Sandwich Enzyme-linked immunosorbent assay (ELISA)
Maxisorp™ EIA plates (Thermo Scientific) were coated with 0.5 μg/ml goat anti-human IgG-Fc (Jackson ImmunoResearch Laboratories) in PBS at 4 °C overnight. Thereafter the plate was blocked with 2 % bovine serum albumin (BSA) for 1 h. Recombinant human FGFRs-Fc (R&D systems) (500 ng/ml) or the control human IgG-Fc (Abcam) was then added for 2 h, followed by incubation with IMB-R1 or the control rabbit IgG (Invitrogen) at the indicated doses for 1 h. The bound antibodies were detected with 0.5 μg/ml HRP-conjugated goat anti-rabbit IgG (Jackson ImmunoResearch Laboratories) and visualized with TMB substrate (Thermo Scientific). The color was read at 450 nm using the Victor3 multilabel plate reader (PerkinElmer). All reactions were performed at room temperature, unless indicated otherwise, and protected from light, with each step followed by extensive washing in blocking buffer. The readings were normalized against the controls.
Protein-GAG binding assay
To investigate whether IMB-R1 prevented heparin/HS binding to FGFR1, 50 μg/ml heparin was coated onto GAG-binding plates in the standard assay buffer provided (Iduron). After blocked with 0.5 % BSA, the plate was incubated with 2 μg/ml FGFR-Fc in PBS for 2 h at 37 °C. FGFR-Fc bound to heparin was then detected using 0.5–1 μg/ml HRP-conjugated goat anti-human IgG-Fc (Jackson ImmunoResearch Laboratories) for 1 h and visualized as described above.
To examine whether IMB-R1 affected the interaction between FGFRs and FGF2, 5 μg/ml heparin was coated onto the plate (using the standard assay buffer) as a substrate to bind FGF2. The heparin-coated surface was blocked with 0.5 % fish gelatin before 200 ng/ml FGF2 (R&D systems) in PBS was added for 2 h at 37 °C. Next, FGFR-Fc (500 ng/ml) was pre-complexed in PBS with increasing amounts of IMB-R1 (on ice for 2 h). The complex was then applied to the FGF2 surface described above for 1.5 h. The amount of FGFR-Fc bound to FGF2 was determined as described above.
Receptor Tyrosine Kinase array
The phosphorylation status of the FGFRs was determined with the Human Phospho-Receptor Tyrosine Kinase Array (R&D Systems). MG63 cells were seeded at 20,000 cells/cm2 for 1 day and deprived of serum for 48 h before treated with 20 ng/ml FGF2 in the presence of IMB-R1 or rabbit IgG (at 1:250 dilution) for 5 min. The cells were then rinsed with PBS and lysed with the buffer provided in the kit. Cell lysates (250 μg) were assayed for of FGFR kinase activity following the manufacturer’s instruction.
Proliferation assay
The cells were plated in triplicate at 20,000 cells/cm
2 except for the MDAMB468 cells, that were at 100,000 cells/cm
2. They were treated with indicated doses of IMB-R1 or other reagents for 2 days, before viable cell numbers were assessed by the GUAVA Flow Cytometry Viacount Program as described previously [
56].
Annexin V- propidium iodide (PI) staining
Cells were plated at the densities nominated above and allowed to adhere overnight. Cells were treated with reagents for 24 h and then stained with Annexin V-FITC and/or PI [
46] protected from light for 15 min.. Fluorescent cells were detected by BD FACS Array (BD Biosciences), and viable and apoptotic cells analyzed using FlowJo software (Tree Star Inc). The unstained cells were used for gating purposes.
Caspase 3 activity assay
MG63 cells were seeded at 20,000 cells/cm2 and treated as indicated. Cells were then lysed and Caspase 3 activity measured using the Caspase-3 Colorimetric Assay Kit (BioVision) following the manufacturer’s instructions.
Microarray
Three consecutive passages of cells were treated as indicated and the total RNA extracted in TRIzol®-reagent and purified with the PureLink™ RNA mini kit (Invitrogen). The purified RNA was processed to cRNA using the Illumina® TotalPrep™-96 RNA Amplification kit (Ambion) as per the manufacturer’s instructions. cRNA was hybridized to the probes on the Human HT-12 v4 Expression BeadChip (Illumina). After washing and staining, the chip was scanned by the BeadArray™ Reader. The data was processed and heatmaps were generated using GenomeStudio software. The gene expression data were further analysed using GeneSpring GC 11.0 software and DAVID Bioinformatics Resources 6.7 (
http://david.abcc.ncifcrf.gov) [
57]. The function analysis was performed using Ingenuity Pathways Analysis software. The overlapping of gene targets in different cells was analysed using VENNY software (
http://bioinfogp.cnb.csic.es/tools/venny/index.html).
Cancer tissue array
The human multiple organ cancer tissue array (US Biomax, Inc) contains 19 types of cancer with 20 cases/type and 5 cases/type of normal controls. The sections were stained with IMB-R1 using a standard immunohistochemistry (IHC) paraffin staining method. The section was first deparaffinized and heat-induced antigen retrieval performed. After blocking in horse serum, the section was incubated with IMB-R1 or rabbit serum at a dilution of 1:800 for 1 h. The bound IMB-R1 was detected with ImmPRESS™ peroxidase anti-rabbit (Vector Laboratories) for 30 min followed by addition of peroxidase substrate DAB solution (DAKO Cytomation). Thereafter the section was counterstained with Hematoxylin QS (Vector Labs) and mounted in permanent mounting medium (Sigma). All steps were performed at room temperature, and between incubations sections were rinsed free of non-specific binding. The total positive cell numbers and intensity of the antibody staining were measured by ImageScope (Aperio Scanning System) and 20x object images captured. The intensity of staining was scored as negative (0), weak (1+), moderate (2+), or strong (3+).
Statistical analysis
Each experiment was repeated at least three times and numeric data were expressed as mean ± SD of triplicate samples. Differences among treatments were analyzed by Student’s t test. Significant differences were considered as those with a p value < 0.05.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made.
The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
SMC, AvW and LL designed research; LL, SKT and THG performed the experiments and data analysis; LL, SMC and AvW wrote the manuscript; EC and VN participated in its design and helped to draft the manuscript. All authors read and approved the manuscript.