- Open Access
Control of post-translational modifications in antithrombin during murine post-natal development by miR-200a
© Teruel et al.; licensee BioMed Central Ltd. 2013
Received: 21 February 2013
Accepted: 8 May 2013
Published: 16 May 2013
Developmental haemostatic studies may help identifying new elements involved in the control of key haemostatic proteins like antithrombin, the most relevant endogenous anticoagulant.
In this study, we showed a significant reduction of sialic acid content in neonatal antithrombin compared with adult antithrombin in mice. mRNA levels of St3gal3 and St3gal4, two sialyltransferases potentially involved in antithrombin sialylation, were 85% lower in neonates in comparison with adults. In silico analysis of miRNAs overexpressed in neonates revealed that mir-200a might target these sialyltransferases. Moreover, in vitro studies in murine primary hepatocytes sustain this potential control.
These data suggest that in addition to the direct protein regulation, microRNAs may also modulate qualitative traits of selected proteins by an indirect control of post-translational processes.
MicroRNAs (miRNAs) are small non-coding RNAs implicated in the modulation of a large number of physiological and pathological processes[1, 2] through a mechanism based on the repression of protein translation or degradation of messenger RNAs. MiRNAs have been recently involved in the modulation of several haemostatic factors such as fibrinogen, tissue factor, and proteins implicated in platelet function[4–6]. Actually, miRNAs can also be involved in the quantitative variations of elements of the haemostatic system observed during development. In humans, levels of haemostatic factors go in constant increase after birth until reaching levels similar to those found in adults within the first year[8–10]. In particular, antithrombin, an anticoagulant serpin crucial in the control of the hemostatic system, is significantly reduced (50%) in plasma of neonates in comparison with adults. Despite these differences, neonates maintain a perfect haemostatic equilibrium. Accordingly, a developmental model between neonatal and adult period is ideal to study the mechanisms that regulate haemostatic protein levels and the adaptation of this system to particular conditions. In addition to these quantitative changes, few works have shown that neonatal antithrombin has lower levels of sialic acid than its adult counterpart but the molecular mechanism of this regulation is unknown[12, 13]. Thus, the developmental model may also allow to investigating the regulation of N-glycosylation by miRNAs and thus to further enlarge the effects of miRNAs in gene regulation.
Non-inbred Swiss CD1 mice from different litters were sacrificed by cervical dislocation or decapitation at different ages, from day one after birth to adult age. Livers finely dissected were immediately snap-frozen in liquid nitrogen. Blood was anticoagulated with trisodium citrate, centrifuged at 1,500 × g for 5 minutes to obtain platelet poor plasma and immediately stored in aliquots at −80°C. All experimental procedures strictly followed the University of Murcia approved Institutional Animal Care guidelines and were approved by the local ethical committee (#C131002043; 15/02/2010).
Antithrombin levels and activity
Antithrombin activity was determined by chromogenic methods, as previously described. Anti-factor Xa (anti-FXa) assay was performed with pentasaccharide, bovine FXa, and S-2765 chromogenic substrate (Chromogenix, IZASA, Spain). Antithrombin levels were determined by enzyme-linked immunosorbent assay and electro-immunodiffusion (Laurell), as previously reported. Values were expressed as a percentage relative to a pool of citrated plasma from 10 adult control mice (100%).
Electrophoretic analysis of antithrombin
Mouse plasma samples were run in polyacrylamide gel electrophoresis under denaturing and non-denaturing conditions, blotted onto PVDF membranes, and immunodetected with goat anti-human antithrombin polyclonal antibody (Sigma-Aldrich, Madrid, Spain) and rabbit anti-goat IgG-horseradish peroxidase conjugate (Sigma-Aldrich, Madrid, Spain), with detection via an ECL kit (Amersham Biosciences, Little Chalfont, UK), essentially as described elsewhere.
Plasma samples from adult and neonate mice were subjected to isoelectrofocusing (IEF) analysis and electroelution using an OFFGEL fractionator with strips of 12 cm with a pH gradient of 4–7 (Agilent 3100, Agilent Technologies, Madrid, Spain). Each fraction collected was run in SDS-PAGE gel and immunodetected as described above.
Plasma from adult and neonate (+1 day) mice (10 μL) were treated with 2 U α2-3,6,8,9 neuraminidase (sialidase) (N 3786, Sigma-Aldrich, Saint Louis, USA) at 37°C for 18 hours in 50 mM sodium phosphate buffer, pH 6.0. Samples were resolved by SDS-PAGE and detected as previously described.
Total RNA was isolated from frozen liver using Trizol® Reagent (Invitrogen, Carlsbad, CA) following manufacturer’s instructions. The RNA concentration and 260/280 ratio were determined by using NanoDrop spectrophotometer (Thermo Scientific, Wilmington, DE) and RNA integrity was verified by lab-on-chip technology using the Experion automated electrophoresis system (Bio-Rad Laboratories, Madrid, Spain).
MicroRNAs microarray profiling was performed using total RNA extracted from the liver from one adult mouse (day +50) and one neonate mouse (day +1) using the LC Sciences technology (LC Sciences, Houston, TX). The arrays were designed to detect and quantify miRNA transcripts corresponding to 558 mature miRNAs contained in the Sanger mirBase Release 10.0 (miRMouse 10.0:http://www.mirbase.org/pub/mirbase/10.0/). We used two chips (1 and 2) in which RNAs from each sample were labeled either with cy3 or with cy5. The signal values were derived by background subtraction and normalization. Additional details on the array are available elsewhere.
In silico studies
Several web databases and algorithms of miRNA target prediction were used for the search of miRNA targeting sialyltransferases. We essentially used TargetScan (release 5.1:http://www.targetscan.org), which provides the prediction results computed by the TargetScanS algorithm, PicTar (http://www.pictar.mdc-berlin.de), and miRanda (http://www.microrna.org/microrna/home.do).
Murine hepatocyte primary culture
Hepatocytes were isolated from livers of Swiss CD1 mice using a modified version protocol from Wu et al.. Mice were anesthetized with an intraperitoneal injection of a ketamine/xylazine mixture. A 24G clear cannula was inserted into the posterior vena cava and secured with a ligature. A second ligature was placed around the anterior vena cava, between the liver and the heart, and the portal vein was severed, allowing outflow of solution. The liver was then perfused at 37°C with oxygenized HBSS (in mM: 137 NaCl, 5.4 KCl, 0.8 MgSO4.7H2O, 0.3 NaHPO4.2H2O, 0.44 KH2PO4, 26 NaHCO3, pH 7.4) 3 min at 5 mL/min and 5 min at 7 mL/min. The perfusion solution was then changed to HBSS supplemented with 4 mM CaCl2 and containing 0.12% collagenase (Sigma-Aldrich, Madrid, Spain) for 8 min at 5 mL/min. The liver was additionally incubated with HBSS with 0.12% collagenase for 15 min, filtered through a cell strainer (100 μm from Becton Dickinson, Madrid, Spain) and hepatocytes were isolated by repeated 50 × g centrifugations. Viability was assessed using trypan blue to be >90% in all the cases. Six-well plates were pre-coated with 50 μg/mL collagen from Stemcell (Grenoble, France) for 12 h at 4°C and cells were seeded at 250,000/well.
Primary hepatocytes were maintained in DMEM/F12 supplemented with 10% fetal bovine serum at 37°C in a humidified incubator with 5% CO2. Cells were pre-cultured for 24 h in complete medium without antibiotics and transfected at 40-60% confluence with 100 nmol/L of precursor molecules for miR-17-3p, miR-200a, and negative scrambled control (Applied Biosystems, Madrid, Spain) by using siPORT™ NeoFX™ transfection agent (Applied Biosystems, Madrid, Spain). The cells were collected 48 hours after transfection and total RNA was extracted.
qRT-PCR and validation assays
Total RNA from mouse livers and from transfected hepatocytes was isolated using Trizol® Reagent (Invitrogen, Madrid, Spain). RNA integrity was verified using bioanalyzer (Bio-Rad, Madrid, Spain). RNA samples were stored at −80°C until used in the experiments. The miRNA and mRNA quantification were carried out as previously described. For St3gal3, St3gal4, and St6gal1, as well as for serpinc1 transcripts relative quantification, retrotranscription reactions were performed using 100 ng of total RNA for each sample according to the manufacturer instructions (SuperScript First Strand, Invitrogen, Madrid, Spain). One set of primers and a probe were chosen from the Applied Biosystems list of TaqMan® Gene Expression Assays for these sialyltransferases (Hs00544033_m1, Hs00920871_m1, and Hs00949382_m1, respectively). For serpinc1 expression was measured using assay Hs00166654_m1 (Applied Biosystems). Sialyltransferase mRNA expression analysis was performed in triplicate for each sample. Expression of β-actin (Hs99999903_m1) was used as endogenous reference control. The PCR reactions were performed using an LC480 Real Time PCR system (Roche Applied Science, Barcelona, Spain). We employed the 2-ΔCt method to calculate the relative abundance of miRNA and mRNA compared with endogenous control expression. Ct is the Threshold Cycle and ΔCt = Ct sample gene - Ct endogenous control.
MiRNA assay kits for miR-200a (Applied Biosystems, Madrid, Spain) were used to validate expression levels in mouse hepatocytes during post-natal development (neonates day+1, n=14; adults day+50, n=5). Expression of U6 snRNA (Applied Biosystems, Madrid, Spain) was used as endogenous reference control.
Quantitative differences of antithrombin between neonate and adult mice
Qualitative differences of antithrombin in neonate and adult mice
To further evaluate the differences of neonatal and adult antithrombins, we performed IEF of plasma from neonate and adult mice. Our results showed that neonates expressed more antithrombin isoforms with higher pI (5.19) and lacked of isoforms with lower pI (4.94) (Figure 2D). These results were in accord with those obtained in native electrophoresis.
Expression of sialyltransferases potentially involved in glycosylation of antithrombin
Regulation of St3gal3 and St3gal4 by miR-200a
miR-200a putative target site in St3gal3 and St3gal4 mRNA using different target prediction software
Antithrombin is the main endogenous anticoagulant and, thus, its role in regulating haemostasis is absolutely essential. Indeed, complete antithrombin deficiency is incompatible with life and partial deficiency is an important risk factor for developing venous thrombosis. Besides its role in haemostasis, antithrombin may also regulate other important physiological processes such as inflammation, angiogenesis or apoptosis[22–24]. Intriguingly, the levels of antithrombin in neonates are severely reduced in comparison with adults without relevant physiological consequences[8, 9, 25]. This study confirms that the difference not only relies on protein expression levels but also in post-translational modifications. Here, we studied the expression, features, and functionality of antithrombin in a mouse model to deepen into the impact of post-translational modifications of this protein in developmental haemostasis.
Our previous results suggested that the lower levels of antithrombin in neonate mice are mainly explained by a concomitant reduction of mRNA in hepatocytes. In addition, electrophoretic data in the present study suggest that the lower molecular weight of antithrombin from neonates is due to a post-translational modification: an abnormal N-glycosylation (Figures 2 and3). Our results show that the sialic acid content of antithrombin is smaller in neonates than adults. Unfortunately, we were unable to perform fine glycomic studies to calculate the exact sialic acid content of neonate antithrombin due to the large amount of purified protein that is required for this procedure. Interestingly, a reduced sialylation of antithrombin has also been described for antithrombin in chicken and sheep neonates[12, 13]. These data strongly suggest that this has to be a process highly regulated in different species. Aiming to identify the mechanisms involved in such control, we evaluated the mRNA levels of three sialyltransferases potentially involved: St3gal3, St3gal4, and St6gal1. Indeed, St6gal-I performs α2-6 sialic acid linkage as that present in antithrombin. Accordingly, this enzyme seems to be the main responsible for the sialylation of antithrombin. However, in St6gal-I KO mice, St3gal-IV, that performs α2-3 linkages, may also achieve α2-6 linkages in von Willebrand factor. In addition, a study by Fan et al. revealed that recombinant human antithrombin expressed in baby hamster kidney cells is fully sialylated containing α2-3 linkage. Thus, it is worth suggesting that the lower levels of sialic acid in neonate’s antithrombin might be explained by the reduced expression of St3gal3 and St3gal4. Further experiments are necessary to clarify these issues.
The next step to understand the mechanism responsible for these differences was the identification of the element(s) controlling the levels of these sialyltransferases. In this framework, the recent report suggesting that some conserved genes implicated in glycosylation pathway may be regulated by miRNAs during animal development, reveals miRNAs as potential candidates. n fact, in silico searching identified miR-200a as an excellent regulator of St3gal3 and St3gal4. Interestingly, the levels of this miRNA during development show a fully compatible change (overexpressed in neonate mice, but reduced expression in adults). The final proof indicating the control of these sialyltransferases by miR-200a was obtained by transfecting this miRNA in adult primary hepatocytes. These experiments suggest that miR-200a may be in part implicated in the regulation of St3gal3 and, in a lesser degree, in the regulation of St3gal4, as predicted by in silico studies (Table 1). Specificity of this regulation is further suggested by the lack of effect of another miRNA overexpressed in neonate’s liver, miR-17-3p. However, other mechanisms and additional miRNAs still to characterize may be involved in the reduced expression of these two sialyltransferases in neonate mice.
Finally, it would be of great interest to evaluate whether or not these qualitative modifications regulated indirectly by miRNAs could have functional significance apart of contributing to an increased clearance. In our case, it is necessary to investigate the functional relevance of the lower sialylation in antithrombin, not only on the anticoagulant function, which might contribute to explain the dramatic change of the haemostatic system after birth, but also on other functions of this molecule.
Our results supported by those of Kahai et al. showing that UDP-N-acetyl-alpha-D-galactosamine:polypeptide N-acetylgalactosaminyltransferase 7 (GalNAc-T7) is inhibited by miR-378 with consequences in the rate of osteoblast differentiation, open new and interesting perspectives, as the regulation of proteins involved in N-glycosylation (and potentially any other post-translational modification) of antithrombin (and extensively other proteins) may be done by miRNAs. The role of miRNAs in diseases and physiological processes is therefore not restricted to the direct control of proteins of one system (in this case, the haemostatic system), but could be extended to an indirect effect by affecting elements involved in transcriptional[32, 33], translational or post-translational processes.
MEM-B is a holder of a predoctoral research grant from ISCIII. CM and IM-M are investigators from Fundación para la Formación e Investigación Sanitarias de la Región de Murcia (FFIS). This work was supported by research grants from ISCIII and FEDER (PI11/00566, PI12/00657, Red RECAVA RD12/0042/0050).
- Fabbri M, Croce CM, Calin GA: MicroRNAs. Cancer J. 2008, 14: 1-6. 10.1097/PPO.0b013e318164145e.View ArticlePubMedGoogle Scholar
- Bartel DP: MicroRNAs: target recognition and regulatory functions. Cell. 2009, 136: 215-233. 10.1016/j.cell.2009.01.002.PubMed CentralView ArticlePubMedGoogle Scholar
- Guo H, Ingolia NT, Weissman JS, Bartel DP: Mammalian microRNAs predominantly act to decrease target mRNA levels. Nature. 2010, 466: 835-840. 10.1038/nature09267.PubMed CentralView ArticlePubMedGoogle Scholar
- Fort A, Borel C, Migliavacca E, Antonarakis SE, Fish RJ, Neerman-Arbez M: Regulation of fibrinogen production by microRNAs. Blood. 2010, 116: 2608-2615. 10.1182/blood-2010-02-268011.View ArticlePubMedGoogle Scholar
- Teruel R, Perez-Sanchez C, Corral J, Herranz MT, Perez-Andreu V, Saiz E, Garcia-Barbera N, Martinez-Martinez I, Roldan V, Vicente V, López-Pedrera C, Martinez C: Identification of miRNAs as Potential Modulators of Tissue Factor Expression in Patients with Systemic Lupus Erythematosus and Antiphospholipid Syndrome. J Thromb Haemost. 2011, 9: 1985-1992. 10.1111/j.1538-7836.2011.04451.x.View ArticlePubMedGoogle Scholar
- Nagalla S, Shaw C, Kong X, Kondkar AA, Edelstein LC, Ma L, Chen J, McKnight GS, Lopez JA, Yang L, Jin Y, Bray MS, Leal SM, Dong JF, Bray PF: Platelet microRNA-mRNA coexpression profiles correlate with platelet reactivity. Blood. 2011, 117: 5189-5197. 10.1182/blood-2010-09-299719.PubMed CentralView ArticlePubMedGoogle Scholar
- Teruel R, Corral J, Perez-Andreu V, Martinez-Martinez I, Vicente V, Martinez C: Potential role of miRNAs in developmental haemostasis. PLoS One. 2011, 6: e17648-10.1371/journal.pone.0017648.PubMed CentralView ArticlePubMedGoogle Scholar
- Andrew M, Paes B, Milner R, Johnston M, Mitchell L, Tollefsen DM, Castle V, Powers P: Development of the human coagulation system in the healthy premature infant. Blood. 1988, 72: 1651-1657.PubMedGoogle Scholar
- Andrew M, Paes B, Milner R, Johnston M, Mitchell L, Tollefsen DM, Powers P: Development of the human coagulation system in the full-term infant. Blood. 1987, 70: 165-172.PubMedGoogle Scholar
- Andrew M, Vegh P, Johnston M, Bowker J, Ofosu F, Mitchell L: Maturation of the hemostatic system during childhood. Blood. 1992, 80: 1998-2005.PubMedGoogle Scholar
- Rau JC, Beaulieu LM, Huntington JA, Church FC: Serpins in thrombosis, hemostasis and fibrinolysis. J Thromb Haemost. 2007, 5 (Suppl 1): 102-115.PubMed CentralView ArticlePubMedGoogle Scholar
- Amrani DL, Mosesson MW, Koide T: Evidence that chicken antithrombin III is a developmentally regulated glycoprotein synthesized by hepatocytes. Biochim Biophys Acta. 1985, 847: 324-334. 10.1016/0167-4889(85)90038-2.View ArticlePubMedGoogle Scholar
- Niessen RW, Lamping RJ, Peters M, Lamers WH, Sturk A: Fetal and neonatal development of antithrombin III plasma activity and liver messenger RNA levels in sheep. Pediatr Res. 1996, 39: 685-691. 10.1203/00006450-199604000-00021.View ArticlePubMedGoogle Scholar
- Ishiguro K, Kojima T, Kadomatsu K, Nakayama Y, Takagi A, Suzuki M, Takeda N, Ito M, Yamamoto K, Matsushita T, Kusugami K, Muramatsu T, Saito H: Complete antithrombin deficiency in mice results in embryonic lethality. J Clin Invest. 2000, 106: 873-878. 10.1172/JCI10489.PubMed CentralView ArticlePubMedGoogle Scholar
- Hagiwara S, Iwasaka H, Shingu C, Matsumoto S, Uchida T, Noguchi T: High-dose antithrombin III prevents heat stroke by attenuating systemic inflammation in rats. Inflamm Res. 2010, 59: 511-518. 10.1007/s00011-009-0155-y.View ArticlePubMedGoogle Scholar
- Larsson H, Akerud P, Nordling K, Raub-Segall E, Claesson-Welsh L, Bjork I: A novel anti-angiogenic form of antithrombin with retained proteinase binding ability and heparin affinity. J Biol Chem. 2001, 276: 11996-12002. 10.1074/jbc.M010170200.View ArticlePubMedGoogle Scholar
- Guerrero JA, Teruel R, Martinez C, Arcas I, Martinez-Martinez I, de la Morena-Barrio ME, Vicente V, Corral J: Protective role of antithrombin in mouse models of liver injury. J Hepatol. 2012, 57: 980-986. 10.1016/j.jhep.2012.06.023.View ArticlePubMedGoogle Scholar
- Manco-Johnson MJ, Jacobson LJ, Hacker MR, Townsend SF, Murphy J, Hay W: Development of coagulation regulatory proteins in the fetal and neonatal lamb. Pediatr Res. 2002, 52: 580-588. 10.1203/00006450-200210000-00019.View ArticlePubMedGoogle Scholar
- McCoy AJ, Pei XY, Skinner R, Abrahams JP, Carrell RW: Structure of beta-antithrombin and the effect of glycosylation on antithrombin's heparin affinity and activity. J Mol Biol. 2003, 326: 823-833. 10.1016/S0022-2836(02)01382-7.View ArticlePubMedGoogle Scholar
- Ellies LG, Ditto D, Levy GG, Wahrenbrock M, Ginsburg D, Varki A, Le DT, Marth JD: Sialyltransferase ST3Gal-IV operates as a dominant modifier of hemostasis by concealing asialoglycoprotein receptor ligands. Proc Natl Acad Sci USA. 2002, 99: 10042-10047. 10.1073/pnas.142005099.PubMed CentralView ArticlePubMedGoogle Scholar
- Fan B, Crews BC, Turko IV, Choay J, Zettlmeissl G, Gettins P: Heterogeneity of recombinant human antithrombin III expressed in baby hamster kidney cells. Effect of glycosylation differences on heparin binding and structure. J Biol Chem. 1993, 268: 17588-17596.PubMedGoogle Scholar
- Zhang L, Hammell M, Kudlow BA, Ambros V, Han M: Systematic analysis of dynamic miRNA-target interactions during C. elegans development. Development. 2009, 136: 3043-3055. 10.1242/dev.039008.PubMed CentralView ArticlePubMedGoogle Scholar
- Ashwell G, Harford J: Carbohydrate-specific receptors of the liver. Annu Rev Biochem. 1982, 51: 531-554. 10.1146/annurev.bi.51.070182.002531.View ArticlePubMedGoogle Scholar
- Kahai S, Lee SC, Lee DY, Yang J, Li M, Wang CH, Jiang Z, Zhang Y, Peng C, Yang BB: MicroRNA miR-378 regulates nephronectin expression modulating osteoblast differentiation by targeting GalNT-7. PLoS One. 2009, 4: e7535-10.1371/journal.pone.0007535.PubMed CentralView ArticlePubMedGoogle Scholar
- Guttilla IK, White BA: Coordinate regulation of FOXO1 by miR-27a, miR-96, and miR-182 in breast cancer cells. J Biol Chem. 2009, 284: 23204-23216. 10.1074/jbc.M109.031427.PubMed CentralView ArticlePubMedGoogle Scholar
- Wan HY, Guo LM, Liu T, Liu M, Li X, Tang H: Regulation of the transcription factor NF-kappaB1 by microRNA-9 in human gastric adenocarcinoma. Mol Cancer. 2010, 9: 16-10.1186/1476-4598-9-16.PubMed CentralView ArticlePubMedGoogle Scholar
- Raja SM, Chhablani N, Swanson R, Thompson E, Laffan M, Lane DA, Olson ST: Deletion of P1 arginine in a novel antithrombin variant (antithrombin London) abolishes inhibitory activity but enhances heparin affinity and is associated with early onset thrombosis. J Biol Chem. 2003, 278: 13688-13695. 10.1074/jbc.M300062200.View ArticlePubMedGoogle Scholar
- Hernandez-Espinosa D, Minano A, Martinez C, Perez-Ceballos E, Heras I, Fuster JL, Vicente V, Corral J: L-asparaginase-induced antithrombin type I deficiency: implications for conformational diseases. Am J Pathol. 2006, 169: 142-153. 10.2353/ajpath.2006.051238.PubMed CentralView ArticlePubMedGoogle Scholar
- Martinez-Martinez I, Ordonez A, Navarro-Fernandez J, Perez-Lara A, Gutierrez-Gallego R, Giraldo R, Martinez C, Llop E, Vicente V, Corral J: Antithrombin Murcia (K241E) causing antithrombin deficiency: a possible role for altered glycosylation. Haematologica. 2010, 95: 1358-1365. 10.3324/haematol.2009.015487.PubMed CentralView ArticlePubMedGoogle Scholar
- Lewis BP, Burge CB, Bartel DP: Conserved seed pairing, often flanked by adenosines, indicates that thousands of human genes are microRNA targets. Cell. 2005, 120: 15-20. 10.1016/j.cell.2004.12.035.View ArticlePubMedGoogle Scholar
- Krek A, Grun D, Poy MN, Wolf R, Rosenberg L, Epstein EJ, Macmenamin P, Da PI, Gunsalus KC, Stoffel M, Rajewsky N: Combinatorial microRNA target predictions. Nat Genet. 2005, 37: 495-500. 10.1038/ng1536.View ArticlePubMedGoogle Scholar
- John B, Enright AJ, Aravin A, Tuschl T, Sander C, Marks DS: Human MicroRNA targets. PLoS Biol. 2004, 2: e363-10.1371/journal.pbio.0020363.PubMed CentralView ArticlePubMedGoogle Scholar
- Wu W, Kocarek TA, Runge-Morris M: Sex-dependent regulation by dexamethasone of murine hydroxysteroid sulfotransferase gene expression. Toxicol Lett. 2001, 119: 235-246. 10.1016/S0378-4274(01)00263-6.View 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://www.creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.