- Open Access
Butyrate suppresses expression of neuropilin I in colorectal cell lines through inhibition of Sp1 transactivation
- Danny CW Yu†1,
- Jennifer S Waby†1, 2,
- Haridasan Chirakkal1,
- Carolyn A Staton1 and
- Bernard M Corfe1Email author
© Yu et al; licensee BioMed Central Ltd. 2010
Received: 21 December 2009
Accepted: 15 October 2010
Published: 15 October 2010
Neuropilin is a transmembrane receptor for vascular endothelial growth factor (VEGF) and is expressed in normal endothelial cells and upregulated in cancer cells. Neuropilin-1 (NRP-1) has been shown to promote tumour cell migration and survival in colon cancer in response to VEGF binding. The expression profiles of neuropilins, associated co-receptors and known ligands have been mapped in three colorectal cell lines: Caco-2, HCT116 & HT29. We have previously shown that butyrate, a naturally occurring histone deacetylase inhibitor (HDACi) produced by fermentation of fibre in the colon, causes apoptosis of colon cancer cell lines.
Here we demonstrate that butyrate down-regulates NRP-1 and VEGF at the mRNA and protein level in colorectal cancer cell lines. NRP-1 is a known transcriptional target of Sp1, whose activity is regulated by acetylation. NRP-1 down-regulation by butyrate was associated with decreased binding affinity of Sp1 for canonical Sp-binding sites in the NRP-1 promoter. siRNA-mediated knock-down of Sp1 implied that Sp1 may have strong DNA binding activity but weak transactivation potential.
The downregulation of the key apoptotic and angiogenesis regulator NRP-1 by butyrate suggests a novel contributory mechanism to the chemopreventive effect of dietary fibre.
Fermentation of fibre in the colon leads to production of short-chain fatty acids (SCFA) including butyrate. Butyrate has been implicated in cellular homeostasis of the normal colonic mucosa, and this is thought to underwrite the chemoprotective effect of fibre . In vitro studies indicate that butyrate causes cell cycle arrest, differentiation or apoptosis in a number of transformed cell lines. These outcomes are mediated by butyrate's inhibition of histone deacetylases (HDACs). Transcription factors Sp1 and Sp3 share canonical GC boxes and are thought to bind with equivalent affinity. Moreover, both Sp1 and Sp3 have been reported as acetylated and are targets for HDAC1 and HDAC2 [2, 3]. Butyrate has been shown to inhibit HDAC activity thereby down-regulating Sp1 binding and up-regulating Sp3 binding. This leads to an increase in p21 expression, which ultimately causes cell cycle arrest ; and an increase in Bak expression which ultimately causes apoptosis . Both events may contribute to the chemopreventive action of butyrate. In an accompanying paper, we use a novel anti-acetyl-Sp1 antibody to show that upregulation of p21, Bak and acetylation of Sp1 respond to the same subset of HDACi with highly similar EC50, implying a simple and causal relationship .
The vascular endothelial growth factor (VEGF) family comprises members that share structural homology and have been linked to cancer angiogenesis, metastasis and survival . VEGF signalling on endothelial cells is mediated by three tyrosine kinase receptors VEGFR1-3 . However, the Neuropilin (NRP) family of 130-140 kDa transmembrane glycoprotein receptors has recently been implicated in both VEGF-mediated angiogenesis  and colon cancer cell survival . NRP-1 and NRP2 are non-tyrosine kinase receptors that bind with specific members of the VEGF family: NRP-1 binds VEGF165, VEGF-B, VEGF-E and placenta growth factor-2; whereas NRP-2 binds VEGF145, VEGF165, VEGF-C and VEGF-D . NRP-1 is expressed in morphologically normal colonic epithelium  and is commonly over-expressed in human colon cancer where it correlates with advanced grade, metastatic potential , and decreased patient survival . Cloning of the promoter region of NRP-1 demonstrated the presence of an AP-1, a CCAAT box and two Sp1 elements all of which contribute to induced promoter activity . However, as yet it is not clear which of these elements is involved in the up-regulation of NRP-1 expression in colon cancer.
Patient data indicate that colorectal tumours with increased NRP-1 expression have a greater incidence of metastases, increased proliferation index and reduced numbers of apoptotic cancer cells than tumours with low NRP-1 staining  suggesting that NRP-1 may protect colon cancer cells from apoptosis. Interestingly siRNA-induced down-regulation of NRP-1 has been shown to increase sensitivity to chemotherapy by induction of apoptosis  suggesting that down-regulation of NRP-1 may have therapeutic potential. Therefore, modification of Sp family activity by butyrate and the potential of NRP-1 as an Sp1 target led us to investigate the ability of butyrate to modulate NRP-1 expression, with a view to providing an alternative therapy or chemopreventive strategy for colon cancer.
Profile of angiogenic factors and their receptors mRNA expression in human colon cancer cell lines
Butyrate downregulates NRP-1 at the mRNA and protein levels
To validate further, we undertook a high-content analysis (HCA) of NRP-1 in HCT116 cells (Figure 2C &2D). The distribution of NRP-1 was very heterogenous (see also Additional file 1: Fig S2 for larger image) and was variously observed as periplasmic, perinuclear and pan-cytosolic. The levels of NRP-1 in cells were quantified by HCA as total NRP-1 (Figure 2D). Following butyrate treatment a decrease in level of NRP-1 cross-reactivity was seen in the cell as a whole consistent with the findings from immunoblotting and qRT-PCR. The functionality of HCA was used to assess whether subcellular distribution was altered however no profound alterations were observed over and above the downgregulation (see Additional file 1: Fig S3).
Regulation of NRP-1 by transcription factors Sp1 and Sp3
In certain cellular contexts and at several genes dysregulated by butyrate, Sp1 appears to be displaced by Sp3 following treatment. We therefore examined the effect of butyrate on binding of both short (S) and long (L) forms of Sp3 at the SpA and SpB binding sites. Both Sp3L and Sp3S bound to SpA and SpB. Following butyrate treatment binding to both sequences by Sp3S was reduced significantly, whereas Sp3L binding remained essentially constant (Figure 3Ci). Quantification of three independent repeats showed that the reduction in Sp3S binding to both SpA and SpB was significant at 5 mM and above (Figure 3Ciii). Nuclear extracts were also immunoprobed for Sp3 to verify that the observed mobility shift changes were due to altered binding affinity and not attributable to altered expression of Sp3 (Figure 3Cii).
Taken together these data imply that down-regulation of NRP-1 transcription may primarily be through reduced Sp1 and Sp3 binding.
Conservation of NRP-1 response to HDAC inhibition
Role of Sp1/3 and HDACs in NRP-1 regulation
Butyrate downregulates VEGF at the mRNA and protein levels
Taken together these data suggest a reduction in VEGF production and signalling by cells following butyrate treatment. We hypothesize that the apparent intracellular increase in VEGF may be due to accumulation owing to cessation of export and synthesis preceding degradation of the protein.
Neuropilin-1 was initially characterised as a receptor for semaphorins which mediate neuronal cell guidance, however, it has more recently been identified as an isoform-specific receptor for VEGF in endothelial cells. Furthermore it is expressed in some tumour cells including colon cancer. While it is known to augment angiogenesis through enhancing the binding of VEGF to VEGF-R2 on endothelial cells, it has also been shown to play an essential role in autocrine anti-apoptotic signalling by VEGF in NRP-1 positive breast cancer cells lacking VEGF-R2 [17, 18]. More recently NRP-1 has been shown to be expressed in colon cancer  and inhibition of NRP-1 in colon cancer cell lines using siRNA significantly increased cancer cell apoptosis . These data suggest that inhibition of NRP-1 may result in improved prognosis for colon cancer patients, especially as high levels of NRP-1 expression correlated with poor patient survival .
Using a panel of tumour-derived lines with epithelial phenotype, we established a profile of gene expression for the families of VEGF ligands and receptors. There was a general consistency in these lines of expression of VEGF A and B, but not C, and little or no expression of any of the classical VEGFR. HGF was not expressed although its receptor was. Contrastingly, the PDGF ligands, but not the receptors were expressed. These data may suggest HGFR as a likely co-receptor with NRP-1, especially as others have shown that this occurs in pancreatic cancer cell lines . However, it is possible that plexins may also be binding partners and these have yet to be studied in non-endothelial cell lines. These data imply expression patterns and roles in epithelia not previously anticipated for NRP-1.
Butyrate is thought to have chemopreventative properties in the colon. We have previously shown that at physiologically relevant concentrations in vitro it is a potent inducer of apoptosis in all the cell lines used . We and others have shown that up-regulation of the pro-apoptotic protein Bak occurs in multiple colon cell lines preceding apoptosis and have proposed this as a major contributory mechanism of cell death [5, 20, 21]. The down-regulation of both VEGF and its receptor NRP-1, shown in this study, suggest an alternative or contributory mechanism to colon cell death following butyrate treatment. The downregulation of NRP-1 is conserved amongst all three cell lines (see fig 2) following butyrate treatment. Critically, there is little or no detectable expression of the principle VEGF receptors in these cell lines, implicating NRP-1, and possibly NRP-2, as the primary VEGF receptor and signal transducer. The down-regulation of NRP-1 is paralleled by the down-regulation of the ligand VEGF, at least at the mRNA level in two of the three cell lines. The regulation of VEGF by butyrate seems more complex as our data suggest that transcription and secretion cease faster than the protein is degraded, leading to an observed increase in intracellular VEGF. This parallels the multi-tier level of regulation previously reported in the study of HIF-regulated genes by butyrate . This may imply a coordinate mechanism of gene regulation, which may have a mechanistic basis . We noted that although there was a clear concentration-responsiveness of downregulation of VEGF by butyrate at the mRNA and secreted protein levels, there was a trend to increased expression at the protein level. We hypothesise that this reflects a cessation of secretion in parallel to downregulation of transcription, but in advance of cessation of translation and protein degradation. The application of a high-content analysis approach was used to study the response of NRP-1 to multiple HDACi. The conservation of NRP-1 downregulation following treatment with multiple HDACi with distinct activities suggests that the downregulation of NRP-1 is mediated through an acetylation-dependent pathway and is not specific to the enterocytic response to butyrate. These data indicate the potential of HDACi as a chemotherapeutic route to suppression of NRP-1 dependent chemoresistance and angiogenesis in the colon.
As NRP-1 is involved in both angiogenesis and the prevention of apoptosis in colorectal cancer these data suggest two potential mechanisms for chemoprevention - through the apoptotic regulatory function of NRP-1 and through its pro-angiogenic role. The recent findings that NRP-1 may be expressed in normal epithelia [9, 12, 23], indicate a role for it in epithelial biology and work must be directed at establishing the extent or otherwise to which endothelial signalling mechanisms are replicated in epithelia.
Our data therefore suggest routes to improve therapeutic outcome in the colon: through dietary- (or enema-) mediated alteration in luminal butyrate, or through development of specific HDACi to the NRP-1/VEGF pathway. Both of these areas are the subject of our ongoing study.
Materials and methods
Colon cancer cell line, cell culture and reagent
Caco-2, HCT116 and HT29 human colon cancer cell lines were kindly donated to Prof C Dive's group (PICR, Manchester, UK). The cells were grown in 1 g/L glucose DMEM (GIBCO) with Penicillin/Streptomycin antibiotics and 10% fetal calf serum (Biosera) at 37°C in 5% CO2, 95% air. Dose response effects of sodium butyrate (Calbiochem) were detected by treating cells with culture media supplemented 0 to 20 mM sodium butyrate for 24 hours. 1M stocks of sodium butyrate were made up in PBS and diluted in culture media to give the correct final concentration.
Protein extraction and immunoblotting
Whole cell lysates were collected by resuspending cell pellets in cell kinase buffer (1M Tris pH5.5, 1M NaF, 1M β-glycerophosphate, 0.2M EDTA, 0.2M EGTA, 10% Triton X-100, 0.1M PMSF, 0.1M NaVO4 and 1× proteases inhibitor cocktail, Sigma). Protein concentrations were measured in triplicate by Bradford reagent (Bio-Rad), and 20 μg of protein samples were loaded per well. Transfer membranes were incubated in primary anti-NRP-1 (1:1000 dilution; Santa Cruz) or anti-VEGF (1:1000 dilution; Santa Cruz) for 2 hours at room temperature with gentle agitation, and then in peroxidase-conjugated goat anti-rabbit IgG (1:2000 dilution; Dakocytomation). The immunoblotting results were visualized by CHEMI GENIUS Bio imaging system. The membranes were washed and re-probed with anti-human actin antibody (1:10000 dilution; Abcam) as an equal loading control.
RNA extraction, RT-PCR and quantitative-PCR (q-PCR)
Total RNA was extracted using Trizol reagent (Invitrogen) following the manufacturer's instructions. Reverse transcription was performed with random primers (Promega) in 1 μg of total RNA using the Superscript III Reverse Transcriptase system (Invitrogen). PCR or qPCR were performed in ReadyMix PCR master mix (Thermo) or 1× SYBR green reagent (QIAGEN) with 1 μl cDNA. Forward and reverse primers for RT-PCR (Sigma-genosys) were designed for amplifying VEGF isoforms and their receptors, NRP-1, NRP-2, PDGFA, PDGFB, PDGFRα, PDGFRβ, HGF, HGFR and actin (Additional file 1: Fig S4). QRT-PCR quantitect assays were purchased from Qiagen for quantitative PCR. QPCR data were collected on the ABI StepOnePlus platform. Relative fold changes were calculated using the ΔΔCT method as previously described .
Westerns of electromobility shift assays (WEMSAs)
An adapted version of the EMSA protocol, a western of a mobility shift gel (WeMSA) was carried out as previously described . Briefly: unlabelled oligonucleotides were incubated with nuclear extracts as per Lightshift EMSA kit instructions; complexes were separated by molecular weight using 5% TBE acrylamide mini-gels in 0.5 × TBE; gels were incubated in SDS buffer (25 mM Tris; 192 mM glycine; 0.2% (w/v) SDS) for 10 min prior to being transferred to PVDF at 100V for 1 hr in 0.5 × TBE; the membranes were blocked in 5% milk TBST for 1 hr prior to immunoprobing and ECL detection of HRP conjugated secondary antibodies. Oligonucleotides for binding assays were commissioned from Sigma Genosys. Oligonucleotides used for for EMSA were 3' biotin-labelled.
siRNA transfections were carried out using Ambion NeoFX as per manufacturer's instructions. Briefly siRNAs and transfection reagent were diluted in OptiMEM (Invitrogen) in two separate tubes. Diluted siRNA and diluted transfection reagent were then mixed together and incubated at room temp for ten min. Complexes were transferred to 24 well plates and 0.5 ml HCT116 cell suspension in antibiotic free media (1 × 104 cells) were added to each well. A final concentration of 30 nM of siRNA was used. HCT116 cells were transfected with siRNA to Sp1/3 or validated siRNA to HDACs 1, 2 and 3. Controls used were siRNA to GAPDH, a proprietary negative control siRNA, and mock transfected cells. 24 and 48 hours transfection, the cells were collected for RNA and protein extraction respectively, followed by qRT-PCR and immunoblotting analyses.
Enzyme-linked immunosorbent assay (ELISA)
Conditioned media from cultured cells were collected after each experiment and spun down to remove dead cells and debris. Total VEGF concentration in the culture media was determined by ELISA according to the manufacturer's instruction (R&D system). The degree of intensity was measured using BioTek microplate reader at 450 nm optical density with correction at 560 nm.
High Content Analysis (HCA)
Cells were grown for HCA analysis as described above, in 96 well plates. Cells were fixed by 3.5% formaldehyde in PBS for 15 minutes at room temperature. NRP-1 (R&D system), acetyl-SP1 (in-house) and VEGF (Abcam) antibodies were diluted in digitonin (500 μl/ml; Sigma) treating cells for 30 minutes at room temperature. The cells were then incubated with anti-sheep Alexa Fluor-blue (Invitrogen) or anti-mouse Alexa Fluor-555 (Invitrogen) antibodies. DNA was stained with Hoechst at 2.5 μg/ml. Plates were analysed on a Cellomics Arrayscan. The Arrayscan compartmental analysis algorithm was used to generate a mask to measure cell surface and cytoplasmic staining independently.
The results were expressed as mean ± standard error of the mean (SEM). Densitometric analysis of western blots was performed by Gene Tool software from SynGene to quantify the results of western blotting and westerns of electromobility shift assays. All statisical analyses were conducted using SPSS v18 software (Chicago, IL, USA) and Prism 5. Statistical significance was determined using one-way ANOVA in ELISA and HCA results, and Student's t test was used in densitometric and qRT-PCR results. Each experiment was repeated at least 3 times. P < 0.05 was considered statistically significant.
This work was funded by the BBSRC and Yorkshire Cancer Research.
- Bingham SA, Day NE, Luben R, Ferrari P, Slimani N, Norat T, Clavel-Chapelon F, Kesse E, Nieters A, Boeing H: Dietary fibre in food and protection against colorectal cancer in the European Prospective Investigation into Cancer and Nutrition (EPIC): an observational study. Lancet. 2003, 361: 1496-1501. 10.1016/S0140-6736(03)13174-1View ArticlePubMedGoogle Scholar
- Doetzlhofer A, Rotheneder H, Lagger G, Koranda M, Kurtev V, Brosch G, Wintersberger E, Seiser C: Histone deacetylase 1 can repress transcription by binding to Sp1. Mol Cell Biol. 1999, 19: 5504-5511.PubMed CentralView ArticlePubMedGoogle Scholar
- Sun JM, Chen HY, Moniwa M, Litchfield DW, Seto E, Davie JR: The transcriptional repressor Sp3 is associated with CK2-phosphorylated histone deacetylase 2. J Biol Chem. 2002, 277: 35783-35786. 10.1074/jbc.C200378200View ArticlePubMedGoogle Scholar
- Davie JR: Inhibition of histone deacetylase activity by butyrate. J Nutr. 2003, 133: 2485S-2493S.PubMedGoogle Scholar
- Chirakkal H, Leech SH, Brookes KE, Prais AL, Waby JS, Corfe BM: Upregulation of BAK by butyrate in the colon is associated with increased Sp3 binding. Oncogene. 2006, 25: 7192-7200. 10.1038/sj.onc.1209702View ArticlePubMedGoogle Scholar
- Waby JS, Chirakkal H, Yu DC, Griffiths GJ, Benson RSP, Bingle CD, Corfe BM: Sp1 acetylation results in loss of DNA binding at promoters associated with cell cycle arrest and cell death in a colon cell line. Mol Cancer. 2010, In revisionGoogle Scholar
- Folkman J: Angiogenesis in cancer, vascular, rheumatoid and other disease. Nat Med. 1995, 1: 27-31. 10.1038/nm0195-27View ArticlePubMedGoogle Scholar
- Kowanetz M, Ferrara N: Vascular endothelial growth factor signaling pathways: therapeutic perspective. Clin Cancer Res. 2006, 12: 5018-5022. 10.1158/1078-0432.CCR-06-1520View ArticlePubMedGoogle Scholar
- Staton CA, Kumar I, Reed MW, Brown NJ: Neuropilins in physiological and pathological angiogenesis. J Pathol. 2007, 212: 237-248. 10.1002/path.2182View ArticlePubMedGoogle Scholar
- Parikh AA, Fan F, Liu WB, Ahmad SA, Stoeltzing O, Reinmuth N, Bielenberg D, Bucana CD, Klagsbrun M, Ellis LM: Neuropilin-1 in human colon cancer - Expression, regulation, and role in induction of angiogenesis. American Journal of Pathology. 2004, 164: 2139-2151.PubMed CentralView ArticlePubMedGoogle Scholar
- Klagsbrun M, Takashima S, Mamluk R: The role of neuropilin in vascular and tumor biology. Adv Exp Med Biol. 2002, 515: 33-48.View ArticlePubMedGoogle Scholar
- Hansel DE, Wilentz RE, Yeo CJ, Schulick RD, Montgomery E, Maitra A: Expression of neuropilin-1 in high-grade dysplasia, invasive cancer, and metastases of the human gastrointestinal tract. Am J Surg Pathol. 2004, 28: 347-356. 10.1097/00000478-200403000-00007View ArticlePubMedGoogle Scholar
- Ochiumi T, Kitadai Y, Tanaka S, Akagi M, Yoshihara M, Chayama K: Neuropilin-1 is involved in regulation of apoptosis and migration of human colon cancer. Int J Oncol. 2006, 29: 105-116.PubMedGoogle Scholar
- Rossignol M, Pouyssegur J, Klagsbrun M: Characterization of the neuropilin-1 promoter; gene expression is mediated by the transcription factor Sp1. J Cell Biochem. 2003, 88: 744-757. 10.1002/jcb.10384View ArticlePubMedGoogle Scholar
- Wey JS, Gray MJ, Fan F, Belcheva A, McCarty MF, Stoeltzing O, Somcio R, Liu W, Evans DB, Klagsbrun M: Overexpression of neuropilin-1 promotes constitutive MAPK signalling and chemoresistance in pancreatic cancer cells. Br J Cancer. 2005, 93: 233-241. 10.1038/sj.bjc.6602663PubMed CentralView ArticlePubMedGoogle Scholar
- Pellizzaro C, Coradini D, Daidone MG: Modulation of angiogenesis-related proteins synthesis by sodium butyrate in colon cancer cell line HT29. Carcinogenesis. 2002, 23: 735-740. 10.1093/carcin/23.5.735View ArticlePubMedGoogle Scholar
- Bachelder RE, Crago A, Chung J, Wendt MA, Shaw LM, Robinson G, Mercurio AM: Vascular endothelial growth factor is an autocrine survival factor for neuropilin-expressing breast carcinoma cells. Cancer Res. 2001, 61: 5736-5740.PubMedGoogle Scholar
- Barr MP, Byrne AM, Duffy AM, Condron CM, Devocelle M, Harriott P, Bouchier-Hayes DJ, Harmey JH: A peptide corresponding to the neuropilin-1-binding site on VEGF(165) induces apoptosis of neuropilin-1-expressing breast tumour cells. Br J Cancer. 2005, 92: 328-333.PubMed CentralPubMedGoogle Scholar
- Matsushita A, Gotze T, Korc M: Hepatocyte growth factor-mediated cell invasion in pancreatic cancer cells is dependent on neuropilin-1. Cancer Res. 2007, 67: 10309-10316. 10.1158/0008-5472.CAN-07-3256View ArticlePubMedGoogle Scholar
- Hague A, Manning AM, Hanlon KA, Huschtscha LI, Hart D, Paraskeva C: Sodium butyrate induces apoptosis in human colonic tumour cell lines in a p53-independent pathway: implications for the possible role of dietary fibre in the prevention of large-bowel cancer. Int J Cancer. 1993, 55: 498-505. 10.1002/ijc.2910550329View ArticlePubMedGoogle Scholar
- Ruemmele FM, Dionne S, Qureshi I, Sarma DS, Levy E, Seidman EG: Butyrate mediates Caco-2 cell apoptosis via up-regulation of pro-apoptotic BAK and inducing caspase-3 mediated cleavage of poly-(ADP-ribose) polymerase (PARP). Cell Death Differ. 1999, 6: 729-735. 10.1038/sj.cdd.4400545View ArticlePubMedGoogle Scholar
- Prasanna Kumar S, Thippeswamy G, Sheela ML, Prabhakar BT, Salimath BP: Butyrate-induced phosphatase regulates VEGF and angiogenesis via Sp1. Arch Biochem Biophys. 2008, 478: 85-95. 10.1016/j.abb.2008.07.004View ArticlePubMedGoogle Scholar
- Pellet-Many C, Frankel P, Jia H, Zachary I: Neuropilins: structure, function and role in disease. Biochem J. 2008, 411: 211-226. 10.1042/BJ20071639View ArticlePubMedGoogle Scholar
- Dennig J, Beato M, Suske G: An inhibitor domain in Sp3 regulates its glutamine-rich activation domains. EMBO J. 1996, 15: 5659-5667.PubMed CentralPubMedGoogle Scholar
- Noti JD: Sp3 mediates transcriptional activation of the leukocyte integrin genes CD11C and CD11B and cooperates with c-Jun to activate CD11C. J Biol Chem. 1997, 272: 24038-24045. 10.1074/jbc.272.38.24038View ArticlePubMedGoogle Scholar
- Kennett SB, Udvadia AJ, Horowitz JM: Sp3 encodes multiple proteins that differ in their capacity to stimulate or repress transcription. Nucleic Acids Res. 1997, 25: 3110-3117. 10.1093/nar/25.15.3110PubMed CentralView ArticlePubMedGoogle Scholar
- Spengler ML, Kennett SB, Moorefield KS, Simmons SO, Brattain MG, Horowitz JM: Sumoylation of internally initiated Sp3 isoforms regulates transcriptional repression via a Trichostatin A-insensitive mechanism. Cell Signal. 2005, 17: 153-166. 10.1016/j.cellsig.2004.06.007View ArticlePubMedGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.