A novel and ancient group of type I keratins with members in bichir, sturgeon and gar
© Schaffeld et al; licensee BioMed Central Ltd. 2007
Received: 22 December 2006
Accepted: 06 June 2007
Published: 06 June 2007
Vertebrate epithelial cells typically express a specific set of keratins. In teleosts, keratins are also present in a variety of mesenchymal cells, which usually express vimentin. Significantly, our previous studies revealed that virtually all known teleost keratins evolved independently from those present in terrestrial vertebrates. To further elucidate the evolutionary scenario that led to the large variety of keratins and their complex expression patterns in present day teleosts, we have investigated their presence in bichir, sturgeon and gar.
We have discovered a novel group of type I keratins with members in all three of these ancient ray-finned fish, but apparently no counterparts are present in any other vertebrate class so far investigated, including the modern teleost fish. From sturgeon and gar we sequenced one and from bichir two members of this novel keratin group. By complementary keratin blot-binding assays and peptide mass fingerprinting using MALDI-TOF mass spectrometry, in sturgeon we were able to assign the sequence to a prominent protein spot, present exclusively in a two-dimensionally separated cytoskeletal preparation of skin, thus identifying it as an epidermally expressed type I keratin. In contrast to the other keratins we have so far sequenced from bichir, sturgeon and gar, these new sequences occupy a rather basal position within the phylogenetic tree of type I keratins, in a close vicinity to the keratins we previously cloned from river lamprey.
Thus, this new K14 group seem to belong to a very ancient keratin branch, whose functional role has still to be further elucidated. Furthermore, the exclusive presence of this keratin group in bichir, sturgeon and gar points to the close phylogenetic relationship of these ray- finned fish, an issue still under debate among taxonomists.
In vertebrates the cytoskeleton of epithelial cell types is typically reinforced by a specific set of type I and type II keratins that assemble into 10 nm thick filaments formed from typeI/II heterodimers. The keratins are members of the large multigene family of intermediate filament proteins (IFproteins) of which they form by far the most complex group. In human, 53 of the hitherto nearly 70 identified IF protein genes code for keratins [1–4] that are expressed in tissue and developmental specific patterns. Without including human hair and nail forming keratins, the number of keratin genes found in teleost fish is comparably high, but in contrast to human and other tetrapods, teleost fish possess a large excess of type I keratin genes [5–7]. By analysing the molecular evolution of keratins, as well as the evolution of their expression patterns in lower vertebrates, we want to further elucidate the scenario and probable evolutionary forces that led to this extraordinary variety of keratins in vertebrates.
Our investigations of the keratin systems in lamprey, shark, trout, zebrafish, carp, goldfish and lungfish have so far revealed that type I and type II keratins are apparently present in all classes of vertebrates and that the various keratins can generally be subdivided into the "E" keratins, expressed in epidermal keratinocytes and other stratified epithelia, and those appearing in cells forming simple epithelia, thus named "S" keratins [8–19]. Nevertheless, our data based on cDNA sequence analysis, followed by thorough phylogenetic analyses [13–18] as well as the studies based on the recently available genome data from man and teleost fish [1, 2, 5–7], strongly support the view of largely independent origins of the keratin genes found in fish and man. According to our present data, solely the typical "S" keratin pair K8 and K18 can at least be found in all gnathostomian vertebrate groups, indicating the unique and general importance of these "ancient" keratins. In contrast to other vertebrates investigated so far, in modern teleost fish keratins, including K8 and K18, in addition to their typical IF epithelial appearance show a widespread IF occurrence in mesenchymally derived cells and tissues, such as fibroblasts, chondrocytes and blood vessel endothelia (for review see ). The latter in the non-teleost vertebrates usually do not express keratins but the type III IF protein vimentin [20–26]. To further trace the origin of the different "E" and "S" keratins as well as the evolution of the mesenchymal keratin expression in teleosts, we have investigated the keratin systems in a bichir, a sturgeon and a gar that are believed to represent the most ancient groups of the extant ray-finned fish. In the course of these studies, from all three species we obtained sequences that apparently belong to a novel branch of type I keratins, without counterparts in any other vertebrate group investigated so far, including the teleost fish. Here we present and discuss their sequences as well as their phylogenetically relationships to the other members of the type I keratin subfamily, which may also provide clues to the early evolution of ray-finned fish. The latter is still strongly debated among taxonomists, whether on the basis of molecular or morphological data (for an overview see ).
Results and discussion
Novel type I keratin sequences from bichir, sturgeon and gar
Properties of the isolated cDNA clones encoding K14 from bichir, sturgeon and gar.
EMBL accession number
Size of cDNA (bp)
Number of encoded amino acids
Subsequent mining of the available genome and EST databases for K14 counterparts in other vertebrates such as teleosts, amphibians, birds and mammals, so far has not yet yielded any matches, suggesting that this keratin group may only be present in the ancient groups of ray- finned fish. We only found two matches encoding K14 of another sturgeon, notably Acipenser transmontanus (white sturgeon; [EMBL: DR975435, DR975694]), which both stem from a skin-derived cDNA library.
Biochemical identification of K14 in sturgeon
Molecular evolution of keratins – the ancient origin of K14
Furthermore, the phylogenetic tree illustrated in Fig. 4 suggests that most of the type I keratins from ray-finned fish evolved independently from those present in lungfish, frog or man and that very early in actinopterygian evolution a gene duplication gave rise to at least two different type I keratin branches, each with members in both ancient and modern ray- finned fish (see the boxed ray-fins twig in Fig. 4). Compared to our recent analyses [14–18, 28], we have now included additional type I keratin sequences from zebrafish and Xenopus; therefore, in this study the K18 branch shows a more complex branching pattern, not allowing a clear identification of the authentic K18 counterparts in the different vertebrates solely on the basis of their position in the tree. However, our previous identification of K18 in shark, bichir, sturgeon, trout and zebrafish was additionally based on its typical occurrence in simple epithelia, such as liver hepatocytes or intestinal mucosal epithelium, corresponding to the situation in man and other tetrapods [10, 12–16, 19, 29]. Nevertheless, these "functional K18 counterparts" identified in the different vertebrate groups have to be considered as paralogous protein sequences. The branching pattern of the K18 twig supports the suggestion that early as well as more recent duplication events led to the various K18 related genes. Future analysis of their expression patterns may provide further clues for possible functions of these K18 relatives.
Compared to the evolution of type II keratins, the phylogenetic tree of type I keratins appears rather complex and several nodes cannot be resolved. Moreover, in teleosts the number of detected type I keratin genes virtually triples the number of those coding for type II keratins [5–7], in contrast to the almost equal number of type I and type II keratin genes detected in the tetrapod genomes. However, when including the hair- and nail-forming keratins the total number of keratin genes in tetrapods is clearly higher than in teleost fish [1–4]. A group of two sequences, one from zebrafish, the other from Xenopus, also occupy a rather basal position in the type I keratin tree, directly branching off between the K14 and K18 twig. But it has to be taken into consideration that both sequences stem from genomic DNA sequencing and may only represent non-active "relics" of ancient keratin genes. From the current Bayesian phylogenetic analysis (but not from our NJ and ML analysis) another phenomenon emerged for the first time: There are now four highly supported keratin subgroups in the tetrapod twig, each with members in both, frog and man (in Fig. 4 encircled by dotted lines). Moreover, the tree topology indicates that the human hair keratins may have emerged prior to these radiation events.
The type I and type II keratins represent the two most complex and abundant groups of intermediate filament (IF) proteins among vertebrates and their structure, function and extraordinary variety cannot be completely understood without phylogenetic considerations. Therefore, we investigated the keratin systems in more ancient representatives of the vertebrate lineage. In particular, the keratins present in skin may play an important role in the transition of vertebrates from water- to land- living animals. Here we present a novel group of epidermal type I keratins, which we termed keratin 14 and so far have only found them in the basal groups of living ray-finned fish, notably bichir, sturgeon and gar. Our phylogenetic analysis revealed a rather basal position of this keratin group in the tree of type I keratin evolution. The keratin 14 group even emerged prior to the gnathostomian K18 sequences, which together with its binding partner K8 and with the exception of the recently discovered thread keratins in teleosts and amphibians, was hitherto considered as most ancient group of gnathostomian keratins. Future analyses will hopefully shed more light on the expression and functional role of K14 in the fish epidermis and clarify the identity of its type II keratin binding partner.
Preparation of tissues and cytoskeletal proteins
Sturgeons (Acipenser baeri) were purchased from a local hatching farm (Fischzucht Rhönforelle, Gersfeld, Germany), and gars (Lepisosteus oculatus) from a local pet shop (Fauna Exotica, Mainz-Kostheim, Germany). The bichirs (Polypterus senegalus) were a gift from Dr. Latz (Institute of Zoology, University of Mainz). Within the scope of this study we employed one specimen for each species. Animals were killed by cutting the neck or the tail artery after MS 222 narcosis (0.5 g/l). For further procedures, the tissues were excised and used immediately or snap frozen according to . Cytoskeletal proteins were extracted as described in .
Electrophoresis, immunoblotting, CKBB, PMF and immunofluorescence microscopy
Two-dimensional polyacrylamide gel electrophoresis (2D-PAGE), complementary keratin blot-binding (CKBB) assays, Western blotting and indirect immunofluorescence microscopy were essentially performed as described in . For immunoblotting we used 10% milk powder as the blocking reagent and the antibody incubations were performed overnight at 8°C. Peptide mass fingerprinting (PMF) using matrix-assisted laserdesorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) was performed as described in [16, 18]. The mass tolerance for matching fragments was set to ≤ 100 ppm.
Preparation of RNA
Total RNA from bichir was extracted according to a protocol modified from , including a GTC extraction and subsequent sedimentation of RNA by ultracentrifugation through a dense cushion of caesium chloride (detailed protocol given in ). For purification of sturgeon RNA we applied a protocol that included consecutive steps of guanidinium thiocyanate (GTC) homogenizations, acidic phenol/chloroform extractions and precipitations. The final RNA precipitation was accomplished with 8 M LiCl. Gar RNA was prepared using the GeneMATRIX Universal RNA Purification Kit purchased from Roboklon (Germany) according to the suppliers' instructions. Isolation of mRNA was generally performed with the "PolyATract®-mRNA isolation system" from Promega.
Applying 0.5 to 1.0 μg of total RNA, RT-PCR was performed either using the Superscript II reverse transcriptase from Invitrogen (RT for 1 h at 42°C), followed by standard PCR (Taq Polymerase, Invitrogen) or the QIAGEN®OneStep RT-PCR Kit (RT for 0.5 h at 50°C). PCR was done for 30–35 cycles and primer specific annealing temperatures. We used different combinations of degenerate, IF-specific primers, as previously described  (notably primer numbers P4, P5, P8, P9, P10). 3' RACE-PCR was accomplished using an oligodT-19mer as downstream and sequence-specific oligos as upstream primers. For gel extraction of PCR products we employed the QIAquick Gel Extraction Kit (Qiagen) or the GeneMATRIX DNA Purification Kit AGAROSE-OUT (Roboklon). Cloning of the isolated DNA fragments was accomplished either with the TOPO-TA Cloning Kit (Invitrogen), the StrataClone™ PCR Cloning Kit (Stratagene) or the pGEM-T Easy vector kit (Promega). All kits were used according to the instruction manual. Primers were synthesized by Roth and Sigma-Ark, respectively. Nucleotide sequencing was performed on both strands using the Taq Dye Deoxy Terminator system. The subsequent gel run was done by commercial services (Genterprise, Germany).
cDNA library construction and screening
According to the supplier's instructions we constructed λ-phage cDNA libraries (ZAP-Express®, Stratagene) from sturgeon and bichir, respectively, in each case using 5 μg of mRNA purified from a mixture of tissues, including skin, eyes, brain and internal organs. Clones were isolated using the fish keratin-specific antiserum GPpoly  and digoxygenin-labeled cDNA probes derived from obtained RT-PCR fragments. Digoxygenin labeling was either performed using the DIG-High Prime Kit from Roche, according to the instruction manual or by standard PCR, using digoxygenin labeled nucleotides from Roche.
Sequence analyses and phylogenetic inference
For database searches, we employed the Basic Local Alignment Search Tool (BLAST ) of the National Center for Biotechnology Information (NCBI) and of the Ensembl Genome Browser. The multiple sequence alignments were performed with ClustalX version 1.8  using default gap penalties. When necessary, the alignment was edited by hand. The final alignment was analysed by Neighbor Joining (NJ), Maximum Likelihood (ML) and Bayesian methods (B). All analyses were conducted under the Jones-Taylor-Thornton substitution model of amino acid evolution (JTT ). Furthermore, the model applied for the ML and Bayesian analyses included observed amino acid frequencies (F), estimated proportion of invariant sites (I), and estimation of among-site rate variation for the remaining sites according to a gamma distribution (G) that was set to four rate categories. For the NJ analysis we used the programs PROTDIST and NEIGHBOR of the Phylogeny Inference Package (PHYLIP, version 3.6 b ). The reliability of the tree topology was then tested by bootstrap analysis  with 100 replications, using the PHYLIP programs seqboot, protdist, neighbor and consensus. The ML analysis was conducted with the PHYLIP-like interface PHYML . For bootstrapping we generated 100 pseudo data sets. The Bayesian inference analysis was performed with MRBAYES 3.1  with the frequencies fixed to the Jones frequencies. In each of the two parallel runs four Markov chains (one hot and three cold chains) were run simultaneously for 3,000,000 generations, starting with random trees. Sampling from the trees was set to every 10th generation. Under these conditions the average value for the deviation of split frequencies reached a value < 0.01. The burn-in was set to 115000, based on the stationary phase. All consensus trees were drawn using TREEVIEW version 1.6.6 .
- CKBB :
Complementary keratin blot-binding
- MALDI :
Matrix-assisted laser desorption/ionisation
- TOF :
- MS :
- PAGE :
Polyacrylamide gel electrophoresis
- PMF :
Peptide mass fingerprint
- 2D :
We thank Prof. Dr. Harald Herrmann for providing the recombinant human keratins 8 and 18, Simon Höffling for additional sequencing work, Nicole Tappe for additional biochemical experiments and Prof. Dr. J. Robin Harris for proof reading of the manuscript. We especially thank Prof. Dr. Jürgen Markl for initiating and continuously supporting this work and critical reading of the manuscript. Financially aided by a grant to J.M. from the Deutsche Forschungsgemeinschaft (MA 843/5-1).
- Hesse M, Magin TM, Weber K: Genes for intermediate filament proteins and the draft sequence of the human genome: novel keratin genes and a surprisingly high number of pseudogenes related to keratin genes 8 and 18. J Cell Sci. 2001, 114: 2569-2575.PubMedGoogle Scholar
- Hesse M, Zimek A, Weber K, Magin TM: Comprehensive analysis of keratin gene clusters in humans and rodents. Eur J Cell Biol. 2004, 83: 19-26. 10.1078/0171-9335-00354.View ArticlePubMedGoogle Scholar
- Rogers MA, Winter H, Langbein L, Bleiler R, Schweizer J: The human type I keratin gene family: characterization of new hair follicle specific members and evaluation of the chromosome 17q21.2 gene domain. Differentiation. 2004, 72: 527-540. 10.1111/j.1432-0436.2004.07209006.x.View ArticlePubMedGoogle Scholar
- Rogers MA, Edler L, Winter H, Langbein L, Beckmann I, Schweizer J: Characterization of new members of the human type II keratin gene family and a general evaluation of the keratin gene domain on chromosome 12q13.13. J Invest Dermatol. 2005, 124: 536-544. 10.1111/j.0022-202X.2004.23530.x.View ArticlePubMedGoogle Scholar
- Zimek A, Stick R, Weber K: Genes coding for intermediate filament proteins: common features and unexpected differences in the genomes of humans and the teleost fish Fugu rubripes. J Cell Sci. 2003, 116: 2295-2302. 10.1242/jcs.00444.View ArticlePubMedGoogle Scholar
- Zimek A, Weber K: Terrestrial vertebrates have two keratin gene clusters; striking differences in teleost fish. Eur J Cell Biol. 2005, 84: 623-635. 10.1016/j.ejcb.2005.01.007.View ArticlePubMedGoogle Scholar
- Krushna Padhi B, Akimenko MA, Ekker M: Independent expansion of the keratin gene family in teleostean fish and mammals: an insight from phylogenetic analysis and radiation hybrid mapping of keratin genes in zebrafish. Gene. 2006, 368: 37-45. 10.1016/j.gene.2005.09.016.View ArticlePubMedGoogle Scholar
- Schaffeld M, Markl J: Fish keratins. Methods Cell Biol. 2004, 78: 627-671.View ArticlePubMedGoogle Scholar
- Markl J, Franke WW: Localization of cytokeratins in tissues of the rainbow trout: fundamental differences in expression pattern between fish and higher vertebrates. Differentiation. 1988, 39: 97-122. 10.1111/j.1432-0436.1988.tb00086.x.View ArticlePubMedGoogle Scholar
- Markl J, Winter S, Franke WW: The catalog and the expression complexity of cytokeratins in a teleost fish, the rainbow trout. Eur J Cell Biol. 1989, 50: 1-16.Google Scholar
- Markl J, Schechter N: Fish intermediate filament proteins in structure, function and evolution. Subcell Biochem. 1998, 31: 1-33.PubMedGoogle Scholar
- Conrad M, Lemb K, Schubert T, Markl J: Biochemical identification and tissue-specific expression patterns of keratins in the zebrafish Danio rerio. Cell Tissue Res. 1998, 293: 195-205. 10.1007/s004410051112.View ArticlePubMedGoogle Scholar
- Schaffeld M, Löbbecke A, Lieb B, Markl J: Tracing keratin evolution: Catalog, expression patterns and primary structure of shark (Scyliorhinus stellaris) keratins. Eur J Cell Biol. 1998, 77: 69-80.View ArticlePubMedGoogle Scholar
- Schaffeld M, Haberkamp M, Braziulis E, Lieb B, Markl J: Type II keratin cDNAs from the rainbow trout: Implications for keratin evolution. Differentiation. 2002, 70 (6): 292-299. 10.1046/j.1432-0436.2002.700607.x.View ArticlePubMedGoogle Scholar
- Schaffeld M, Höffling S, Haberkamp M, Conrad M, Markl J: Type I keratin cDNAs from the rainbow trout: independent radiation of keratins in fish. Differentiation. 2002, 70 (6): 282-291. 10.1046/j.1432-0436.2002.700606.x.View ArticlePubMedGoogle Scholar
- Schaffeld M, Knappe M, Hunzinger C, Markl J: cDNA sequences of the authentic keratins 8 and 18 in zebrafish. Differentiation. 2003, 71: 73-82. 10.1046/j.1432-0436.2003.700607.x.View ArticlePubMedGoogle Scholar
- Schaffeld M, Höffling S, Markl J: Sequence, evolution and tissue expression patterns of an epidermal type I keratin from the shark Scyliorhinus stellaris. Eur J Cell Biol. 2004, 83: 359-368. 10.1078/0171-9335-00395.View ArticlePubMedGoogle Scholar
- Schaffeld M, Bremer M, Hunzinger C, Markl J: Evolution of tissue-specific keratins as deduced from novel cDNA sequences of the lungfish Protopterus aethiopicus. Eur J Cell Biol. 2005, 84: 363-377. 10.1016/j.ejcb.2004.12.006.View ArticlePubMedGoogle Scholar
- García DM, Bauer H, Dietz T, Schubert T, Markl J, Schaffeld M: Identification of keratins and analysis of their expression in carp and goldfish: comparison with the zebrafish and trout keratin catalog. Cell Tissue Res. 2005, 322 (2): 245-56. 10.1007/s00441-005-0031-1.View ArticlePubMedGoogle Scholar
- Herrmann H, Fouquet B, Franke WW: Expression of intermediate filament proteins during development of Xenopus laevis . I. cDNA clones encoding different forms of vimentin. Development. 1989, 105: 279-298.PubMedGoogle Scholar
- Lazarides E: Intermediate filaments: a chemically heterogeneous, developmentally regulated class of proteins. Annu Rev Biochem. 1982, 51: 219-250. 10.1146/annurev.bi.51.070182.001251.View ArticlePubMedGoogle Scholar
- Schaffeld M, Herrmann H, Schultess J, Markl J: Vimentin and desmin of a cartilaginous fish, the shark Scyliorhinus stellaris : Sequence, expression patterns and in vitro assembly. Eur J Cell Biol. 2001, 80: 692-702. 10.1078/0171-9335-00206.View ArticlePubMedGoogle Scholar
- Schultess J: Molekulare Evolution der Intermediärfilament-Proteine des Flussneunauges Lampetra fluviatilis. Phd thesis. 2001, University of Mainz, Faculty of BiologyGoogle Scholar
- Schultess J, Schaffeld M, Markl J: Intermediate filament proteins type I to III in the lamprey Lampetra fluviatilis : Searching for the origin of the IF protein evolution in vertebrates. Zoology. 2000, 103 (Suppl.III): 51-Google Scholar
- Schultess J, Schaffeld M, Markl J: Intermediate filament protein sequences of the cyclostome Lampetra fluviatilis. Biol Cell. 2001, 93: 235-(First Joint French-German Congress on Cell Biology)Google Scholar
- Schultess J, Schaffeld M, Scheinert I, Markl J: Intermediate filament protein sequences of the cyclostome Lampetra fluviatilis. Zoology. 2001, 104 (Suppl.IV): 36-Google Scholar
- Zardoya R, Meyer A: Recent advances in the (molecular) phylogeny of vertebrates. Annu Rev Ecol Syst. 2003, 34: 311-338. 10.1146/annurev.ecolsys.34.011802.132351.View ArticleGoogle Scholar
- Schaffeld M, Schultess J: Genes coding for intermediate filament proteins closely related to the hagfish "thread keratins (TK)" alpha and gamma also exist in lamprey, teleosts and amphibians. Exp Cell Res. 2006, 312: 1447-1462. 10.1016/j.yexcr.2006.01.005.View ArticlePubMedGoogle Scholar
- Schaffeld M, Schultess J, Haberkamp M, Bremer M, Markl J: Intermediate Filament Protein Evolution in Fish: Sequences from lamprey, shark, bichir, sturgeon and trout. Biol Cell. 2001, 93: 235-(First Joint French-German Congress on Cell Biology)Google Scholar
- Haberkamp M: Zur Evolution des Cytoskeletts beim Sibirischen Stör Acipenser baeri. Phd thesis. 2002, University of Mainz, Faculty of BiologyGoogle Scholar
- Delsuc F, Brinkmann H, Chourrout D, Philippe H: Tunicates and not cephalochordates are the closest living relatives of vertebrates. Nature. 2006, 439: 965-8. 10.1038/nature04336.View ArticlePubMedGoogle Scholar
- Chirgwin JM, Przybyla AE, MacDonald RJ, Rutter WJ: Isolation of biologically active ribonucleic acid from sources enriched in ribonuclease. Biochemistry. 1979, 18: 5294-5299. 10.1021/bi00591a005.View ArticlePubMedGoogle Scholar
- Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ: Basic local alignment search tool. J Mol Biol. 1990, 215: 403-410.View ArticlePubMedGoogle Scholar
- Thompson JD, Gibson TJ, Plewniak F, Jeanmougin F, Higgins DG: The CLUSTAL_X windows interface: flexible strategies for multiple sequence alignment aided by quality analysis tools. Nucleic Acids Res. 1997, 25: 4876-4882. 10.1093/nar/25.24.4876.PubMed CentralView ArticlePubMedGoogle Scholar
- Jones DT, Taylor WR, Thornton JM: The rapid generation of mutation data matrices from protein sequences. Comput Appl Biosci. 1992, 8: 275-282.PubMedGoogle Scholar
- Felsenstein J: PHYLIP (phylogeny inference package). Version 3.6b. Edited by: . 2004, Distributed by the author, Department of Genetics, University of Washington, SeattleGoogle Scholar
- Felsenstein J: Confidence limits on phylogenies: an approach using the bootstrap. Evolution. 1985, 39: 783-791. 10.2307/2408678.View ArticleGoogle Scholar
- Guindon S, Gascuel O: A simple, fast, and accurate algorithm to estimate large phylogenies by maximum likelihood. Syst Biol. 2003, 52: 696-704. 10.1080/10635150390235520.View ArticlePubMedGoogle Scholar
- Huelsenbeck JP, Ronquist F: MRBAYES: Bayesian inference of phylogenetic trees. Bioinformatics. 2001, 17: 754-755. 10.1093/bioinformatics/17.8.754.View ArticlePubMedGoogle Scholar
- Page RDM: TREEVIEW: an application to display phylogenetic trees on personal computers. Comput Appl Biosci. 1996, 12: 357-358.PubMedGoogle 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.