Common variation in the CYP17A1 and IFIT1 genes on chromosome 10 does not contribute to the risk of endometriosis

Endometriosis is a complex disease involving multiple susceptibility genes and environmental factors. Our pre- vious studies on endometriosis identified a region of significant linkage on chromosome 10q. Two biological candidate genes (CYP17A1 and IFIT1) located on chromosome 10q, have previously been implicated in endometriosis and/or uter- ine function. We hypothesized that variation in CYP17A1 and/or IFIT1 could contribute to the risk of endometriosis and may account for some of the linkage signal on chromosome 10q. We genotyped 17 single nucleotide polymorphisms (SNPs) in the CYP17A1 and IFIT1 genes including SNP rs743572 previously associated with endometriosis in 768 endo- metriosis cases and 768 unrelated controls. We found no evidence for association between endometriosis and individual SNPs or SNP haplotypes in CYP17A1 and IFIT1. Common variation in these genes does not appear to be a major con- tributor to endometriosis susceptibility in our Australian sample.

Endometriosis is a complex disease involving multiple susceptibility genes and environmental factors [1-4]. Esti- mates of the population prevalence indicate that endometri- osis affects 8–10% of women of reproductive age [2, 5], but the reasons for establishment and progression of endometri- osis remain uncertain. Our previous studies on endometriosis identified a region of significant linkage on chromosome 10q26 [6]. The peak linkage signal is located at 148.75 cM between markers D10S587 and D10S1656 and the 95% confidence interval (CI) spans a region of 8.5 megabase pairs (Mbp). Gene map- ping of plausible candidates may help to define the mecha- nisms contributing to the genetic susceptibility of endome- triosis. Two candidate genes on chromosome 10q, which have previously been implicated in endometriosis and uter- ine function, are cytochrome P450, family 17, subfamily A, polypeptide 1 ( CYP17A1, MIM #609300) and interferon- induced protein with tetratricopeptide repeats 1 ( IFIT1, MIM #147690) [7-9]. Human endometrium is highly responsive to hormonal stimuli during the menstrual cycle with estrogen and proges- terone influencing maturational and functional changes in the endometrium. The CYP17A1 gene lies at 104.5 Mbp, on the shoulder of our linkage peak on chromosome 10 and en- codes the cytochrome P450c17
enzyme that is involved in estrogen biosynthesis and metabolism [10]. The gene is ex- pressed in human follicles, corpora lutea and endometrial carcinoma cells [11-13]. A number of studies on CYP17A1 variation suggest it could be a genetic biomarker for hor- mone related diseases [7, 14, 15]. One single nucleotide polymorphism (SNP, rs743572) at the -34 bp position rela- tive to the start codon in the 5’UTR promoter region of *Address correspondence to this author at the Queensland Institute of Medi- cal, Research, Brisbane, Queensland 4029, Australia; E-mail: [email protected] CYP17A1 has been commonly studied. The SNP was thought to be associated with a Sp-1 binding site that would lead to higher gene expression in male baldness [16]. In pa- tients with endometrial cancer, a marked decrease in the A2 allele of CYP17A1 was seen when compared with normal controls [7]. A subsequent study on 119 endometriosis cases and 108 normal controls demonstrated that the CYP17A1 allele was associated with an increased risk of endometriosis in a Chinese population [8], but lack of association between endometriosis and the CYP17A1 polymorphism was ob- served in UK and Japanese populations [17]. The IFIT1 gene is located at 91.1 Mbp, close to the CYP17A1 locus just outside the 95% confidence region for our linkage peak. However the linkage peak is broad and there is evidence for linkage and association with endometri- osis in Puerto Rican families at marker D10S677 [18], which is located at 113.34 cM (95.95 Mb) close to the IFIT1 locus. During early pregnancy, IFIT1 is highly expressed upon stimulation with interferon tau ( IFNT), a type I interferon produced by the conceptus trophectoderm [9, 19]. As an en- dometrial gene, IFIT1 is believed to respond to hormonal stimulation and may have a functional role for uterine sup- port of peri-implantation conceptus survival, growth, and implantation [9]. Although IFIT1 has been studied in the sheep model, it is unknown if variants in IFIT1 contribute to risk of human endometriosis. We previously excluded association between endometri- osis and candidate genes under the linkage peak including empty spiracles homeobox 2 ( EMX2), phosphatase and tensin homolog (PTEN) and fibroblast growth factor receptor 2 (FGFR2) in our Australian sample [20, 21]. To continue to search for the gene or genes contributing to the linkage peak, we examined variation in IFIT1 and/or CYP17A1 to deter- mine whether those genes contribute to the risk of endome- triosis and may account for some of the linkage signal on chromosome 10q. 36 The Open Reproductive Science Journal, 2008, Volume 1 Zhao et al.

Participants and Sample Collection The project was approved by the Human Research Ethics Committee of the Queensland Institute of Medical Research and the Australian Twin Registry. Women with surgically confirmed endometriosis were selected from each of 768 Aus- tralian affected sister pair families as previously described [20]. The sister with the most severe stage of disease was cho- sen for genotyping. Disease severity was assessed retro- spectively from medical records using the revised American Fertility Society (rAFS) classification system [22]. Sixty one percent of cases were classified with minimal to mild endome- triosis (rAFS stages I/II). The remaining 39% of cases with moderate to severe (rAFS stages III/IV) endometriosis were more likely to have ovarian endometriosis. A total of 645 cases (84%) were diagnosed at laparoscopy; the remaining cases were mostly diagnosed at hysterectomy, or in a small number of cases at laparotomy or during another procedure. The controls were 768 unrelated women who had volun- teered for a twin study of gyn aecological health [2]. Controls were selected after consideration of the competing issues of ascertainment bias from clinic controls and presence of undi- agnosed cases. They were selected from women who self- reported they had never been diagnosed with endometriosis and were therefore considered to be at low risk of having endometriosis. Twins had been asked simply ‘have you had endometriosis?’[2]. Additional information from medical records was used where available. Women were also asked whether they had ever had a laparoscopy and/or a hysterec- tomy and the reasons for each. About 14% of control women reported having a hysterectomy and/or laparoscopy. No evi- dence of endometriosis was reported at any of these proce- dures in our control sample [20]. The mean ages (± SD) of the cases and controls at the time of data collection were 35.6 ± 9.1 years (range = 17-65) and 45.7 ± 12.2 (range = 29-90) years respectively. Genomic DNAs were extracted [23], and diluted to a working concentration of 2.5 ng/ μl. The case and control DNAs were randomly placed in 384- well PCR plates. SNP Selection We selected 7 SNPs in the CYP17A1 gene based upon the allelic association results in endometriosis [8] and allele frequency information from the National Center for Biotech- nology Information (NCBI, http://www.ncbi.nlm.nih.gov/). Ten SNPs were selected in the IFIT1 gene on the basis of the allelic frequency information and SNP distribution across the gene. The chosen CYP17A1 SNP list comprised two promoter, two intronic, two coding exonic and one 3’UTR SNPs. One 5’UTR promoter, seven intronic and two coding exonic SNPs were chosen in the IFIT1 gene. All SNP se- quences were downloaded from the Chip Bioinformatics database (http://snpper.chip.org/) and the sequences were cross checked in NCBI and Sequenom RealSNP databases (https://www.realsnp.com/) before assay design. Genotyping Multiplex assays were designed using the Sequenom MassARRAY Assay Design software (version 3.0). SNPs were typed using Sequenom iPLEX™ chemistry on a MALDI-TOF Mass Spectrometer. The 2.5 μL PCR reactions were performed in standard 384-well plates using 10 ng ge- nomic DNA, 0.5 unit of Taq polymerase (Qiagen, Valencia, CA), 500 μmol of each dNTP, and 100 nmol of each PCR primers. Standard PCR thermal cycling conditions and post- PCR extension reactions were carried out as described previ- ously [24]. The iPLEX reaction products were desalted by diluting samples with 15
l of water and adding 3
l of resin. The products were spotted on a SpectroChip (Sequenom), and data were processed and analysed in a Compact Mass Spectrometer by MassARRAY Workstation (version 3.4) software (Sequenom). Statistical Analysis The genotypes were inspected and results were tested for departures from Hardy-Weinberg equilibrium (HWE) sepa- rately for cases and controls using Haploview version 3.32 (Whitehead Institute for Biomedical Research, USA). The PLINK program (http://pngu.mgh.harvard.edu/purcell/ plink/) was used to test association between endometriosis and individual SNPs. Global p-values were obtained for each marker or each haplotype by performing 10,000 permutation tests. Haplotype frequencies, linkage disequilibrium (LD) estimates and analysis were determined by Haploview [25] using the default method of Gabriel [26]. A global p-value <0.05 was considered to be statistically significant. We performed power calculations for our case-control study assuming a disease (endometriosis) prevalence of 10% using the Genetic Power Calculator [27]. Power calculations were based on 768 unrelated cases and 768 unrelated con- trols using a significance threshold ( ) of P = 0.01.

Seven SNPs in the CYP17A1 gene and ten SNPs in the IFIT1 gene were typed in 768 endometriosis cases and 768 unrelated controls. All SNPs were in Hardy–Weinberg equi- librium. The minor allele frequencies of the CYP17A1 SNPs ranged from 0.230 to 0.414 in cases and from 0.200 to 0.443 in controls. The minor allele frequencies of the IFIT1 SNPs ranged from 0.058 to 0.484 in cases and from 0.072 to 0.498 in controls (Table 1). The minor allele frequency for the key SNP (rs743572) in the CYP17A1 gene was 0.386 and 0.397 in cases and controls, respectively. There was no significant difference in allele frequency between cases and controls for this key SNP (Table 1). SNP rs2486758, at the -362 bp posi- tion relative to the start codon in the 5’UTR promoter region of CYP17A1 gene showed nominal evidence of association (P < 0.05). However, the difference in allele frequency be- tween cases and controls was small and the effects were not significant after correcting for multiple testing of all SNPs. The positions of the SNPs genotyped in the IFIT1 gene and the CYP17A1 gene are shown in Fig. ( 1a). A linkage disequilibrium plot of SNPs and common haplotype blocks for the both genes are also shown in Fig. ( 1b,c). We found no evidence for association between endometriosis and indi- vidual SNPs in either IFIT1 or CYP17A1 for either the allelic or the genotypic association tests (Table 1). Stratification of cases according to stage of disease (469 Stage A cases and 768 controls) gave a best point-wise P- value of 0.02 for SNP rs619824 in the CYP17A1 gene, but the global result correcting for multiple tests was non- significant (P = 0.58). Analysis of 296 cases diagnosed with Variation in CYP17A1 and IFIT1 and Risk of Endometriosis The Open Reproductive Science Journal, 2008, Volume 1 37 stage B and 768 controls, showed no significant differences between cases and controls for any SNPs typed in the study.

Our results do not support an association between endo- metriosis and common variation in either CYP17A1 or IFIT1. Although both genes are biological candidates for endometriosis and are located near the peak of our linkage signal on chromosome 10, there was no evidence that vari- ants in either gene were associated with endometriosis or contribute to the linkage signal on chromosome 10q. Genetic studies of CYP17A1 variants to date have fol- lowed a defined biological hypothesis suggesting the 5’UTR promoter region SNP (rs743572) is associated with gene expression [16]. While the T allele (rs743572) was found to be associated with increased risk of endometriosis in a study of Chinese women [14], there was no association between CYP17A1 variants and endometriosis in studies in Brazilian, UK or Japanese populations [17, 28, 29]. Endometriosis is a sex steroid-dependent disease [30]. CYP17A1 is involved in estrogen biosynthesis and metabolism so that certain genetic polymorphisms in the gene could be associated with in- creased risk of developing endometriosis. The differences in the results may relate to study power (sample size) or popu- lation differences. We estimated power for our case-control study based on total sample of 768 cases and 768 controls. There is over 80% power to detect allele frequency of 0.05, 0.25 and 0.5 contributing a dominant genotype relative risk (GRR) of 1.7, 1.5 and 1.8, respectively. In contrast, when the sample size changed to total sample of 100 cases and 100 controls, there is only 10% power to detect allele frequency of 0.05, 0.25 and 0.5 contributing the similar genotype rela- tive risk as stated above. These calculations demonstrate our sample has high power to detect novel gene associations of moderate effect. However, because our cases are highly se- lected in terms of family history, compared to a standard case-control association study, our sample will have consid- erably more power to detect gene associations. We examined variation in CYP17A1 in cases from fami- lies contributing to the linkage peak in this region of chro- mosome 10q and genotyped 7 CYP17A1 SNPs including rs743572 in 768 endometriosis and 768 controls. We found no evidence for association with either SNP rs743572 in the promoter of CYP17A1 or the other 6 SNPs in the gene in our Australian sample. Strong linkage disequilibrium (LD) was detected between SNP rs743572 and the other three SNPs in the gene (rs3740397, r 2=0.948; rs6163, r2=0.958; rs6162, r2=0.919) in our sample. It is unlikely that any asymptomatic cases present in the control samples would affect the conclu- sion from this study for a disease with a prevalence of 8-10% [31]. IFIT1 is a hormonally responsive gene expressed in sheep endometrium following stimulation with IFNT [9, 19]. To test for association between IFIT1 variants and human endometriosis we typed 10 common IFIT1 SNPs in our 768 endometriosis cases and 768 unrelated controls. There was no evidence for association between the individual SNPs and endometriosis. Analysis of the SNPs across the IFIT1 locus Table 1. Association Analysis of 17 SNPs Across the IFIT1 and CYP17A1 Gene Locus Genotyped in 768 Endometriosis Cases and 768 Controls dbSNP ID SNP Position Gene(s) Role Alleles MAF-Cases MAF-Controls
2 P Value OR rs304478 chr10:91140902 IFIT1 Promoter G>T 0.456 0.463 0.139 0.709 0.972 rs303218 chr10:91142573 IFIT1 Intron G>A 0.175 0.186 0.620 0.431 0.929 rs303217 chr10:91143679 IFIT1 Intron T>C 0.470 0.469 0.002 0.968 1.003 rs303216 chr10:91145166 IFIT1 Intron C>T 0.169 0.182 0.858 0.354 0.915 rs303215 chr10:91146811 IFIT1 Intron T>C 0.170 0.182 0.754 0.385 0.921 rs304484 chr10:91149303 IFIT1 Intron G>T 0.231 0.244 0.770 0.380 0.928 rs304485 chr10:91149890 IFIT1 Intron T>A 0.484 0.498 0.578 0.447 0.946 rs303212 chr10:91151335 IFIT1 Intron T>C 0.232 0.241 0.357 0.550 0.950 rs303211 chr10:91152477 IFIT1 Coding exon G>A 0.058 0.072 2.581 0.108 0.790 rs303210 chr10:91152657 IFIT1 Coding exon C>T 0.169 0.182 0.822 0.365 0.916 rs619824 chr10:104571278 CYP17A1 3' UTR G>T 0.414 0.443 2.633 0.105 0.888 rs3740397 chr10:104582665 CYP17A1 Intron (boundary) C>G 0.375 0.396 1.404 0.236 0.916 rs4919687 chr10:104585238 CYP17A1 Intron (boundary) G>A 0.347 0.359 0.539 0.463 0.946 rs6163 chr10:104586914 CYP17A1 Coding exon C>A 0.376 0.395 1.058 0.304 0.926 rs6162 chr10:104586971 CYP17A1 Coding exon G>A 0.389 0.403 0.573 0.449 0.945 rs743572 chr10:104587142 CYP17A1 5' UTR T>C 0.386 0.397 0.376 0.540 0.954 rs2486758 chr10:104587470 CYP17A1 5' UTR T>C 0.230 0.200 3.870 0.049 1.189 MAF: minor allele frequency. dbSNP ID: database SNP identification. UTR: untranslated region. 38 The Open Reproductive Science Journal, 2008, Volume 1 Zhao et al. Fig. (1). Variants typed in the human IFIT1 and CYP17A1 genes (a) the genomic structure of the IFIT1 and CYP17A1 genes showing the location of the 17 SNPs genotyped ( b) the linkage disequilibrium plot of single nucleotide polymorphism estimated as
2 using Haploview (c) common haplotypes and association analysis with endometriosis. Shading key: white
2=0; shades of grey 0>
2<1; black
2=1 a) b) c) 91.140M 91.144M 91.148M 91.152M 104.580M 104.582M 104.584M 104.586M 104.588M rs304478 rs303218 rs303217 rs303216 rs303215 rs304484 rs304485 2Exons 1 IFIT1 rs303212 rs303211 rs303210 8 76 54 2 3 1 rs619824 rs3740397 rs4919687 rs6163 rs743572 rs6162 rs2486758 CYP17A1 Variation in CYP17A1 and IFIT1 and Risk of Endometriosis The Open Reproductive Science Journal, 2008, Volume 1 39 showed strong linkage disequilibrium, with two haplotype blocks and three major haplotypes accounting for 87% of the chromosomes in our case samples. Our data also show strong linkage disequilibrium (LD) in two blocks covering most of the gene in our samples (Fig. 1b). We did not find any evi- dence for association between SNP haplotype frequencies and endometriosis. IFIT1 may be important in uterine devel- opment and function, but common variation is not associated with risk of endometriosis. In this study, we examined the association between en- dometriosis and individual common SNPs and haplotypes in the IFIT1 and CYP17A1 genes on chromosome 10q in an Australian population including a functional SNP in the CYP17A1 promoter region. Our data does not provide evi- dence supporting an association between common variation in the IFIT1 and CYP17A1 genes and endometriosis suscep- tibility. We have previously demonstrated significant linkage to endometriosis on chromosome 10q [6]. Results of the pre- sent study demonstrate that variation in the CYP17A1 and IFIT1 genes does not explain linkage to endometriosis in this region of chromosome 10. Both genes are good candidates for endometriosis but the linkage region spans approximately 8.5 million DNA base pairs and contains over 50 known genes within the 95% confidence interval for the linkage peak. Many of these other genes can also be considered can- didates for endometriosis and variation in one or more of these genes may explain the linkage signal in this region. We conclude that common variants in the IFIT1 and CYP17A1 genes do not play a key role in the pathogenesis of endome- triosis. ACKNOWLEDGMENTS We thank Dr. Daniel T. O’Connor for confirmation of diagnosis and staging of disease from clinical records of 295 cases; Barbara Haddon for co-ordination of family recruit- ment, blood and phenotype collection; Anjali Henders, Me- gan Campbell and staff of the Molecular Epidemiology Laboratory for sample processing and DNA preparation. This study was supported by the Australian Government’s Cooperative Research Centre’s Program and National Health and Medical Research Council of Australia (339430, 339446).

[1] Kennedy S, Mardon H, Barlow D. Familial e ndometriosis. J Assist Reprod Genet 1995; 12: 32-4. [2] Treloar SA, O'Connor DT, O'Connor VM, Martin NG. Genetic influences on endometriosis in an Australian twin sample. Fertil Steril 1999; 71: 701-10. [3] Hadfield RM, Mardon HJ, Barlow DH, Kennedy SH. Endometrio- sis in monozygotic twins. Fertil Steril 1997; 68: 941-2. [4] Stefansson H, Geirsson RT, Steinthorsdottir V, et al . Genetic fac- tors contribute to the risk of developing endometriosis. Hum Re- prod 2002; 17: 555-9. [5] Eskenazi B, Warner ML. Epidemiology of endometriosis. Obstet Gynecol Clin North Am 1997; 24: 235-58. [6] Treloar SA, Wicks J, Nyholt DR, et al. Genomewide linkage study in 1,176 affected sister pair families identifies a significant sus- ceptibility locus for e ndometriosis on chromosome 10q26. Am J Hum Genet 2005; 77: 365-76. [7] Haiman CA, Hankinson SE, Colditz GA, Hunter DJ, De Vivo I. A polymorphism in CYP17 and endometrial cancer risk. Cancer Res 2001; 61: 3955-60. [8] Hsieh YY, Chang CC, Tsai FJ, Lin CC, Tsai CH. Estrogen receptor alpha dinucleotide repeat and cytochrome P450c17alpha gene polymorphisms are associated with susceptibility to endometriosis. Fertil Steril 2005; 83: 567-72. [9] Gray CA, Abbey CA, Beremand PD, et al . Identification of endo- metrial genes regulated by early pregnancy, progesterone, and interferon tau in the ovine uterus. Biol Reprod 2006; 74: 383-94. [10] Picado-Leonard J, Miller WL. Cloning and sequence of the human gene for P450c17 (steroid 17 alpha-hydroxylase/17,20 lyase): simi- larity with the gene for P450c21. DNA 1987; 6: 439-48. [11] Doody KJ, Lorence MC, Mason JI, Simpson ER. Expression of messenger ribonucleic acid species encoding steroidogenic en- zymes in human follicles and corpora lutea throughout the men- strual cycle. J Clin Endocrinol Metab 1990; 70: 1041-5. [12] Suzuki T, Sasano H, Tamura M, et al. Temporal and spatial locali- zation of steroidogenic enzymes in premenopausal human ovaries: in situ hybridization and immunohistochemical study. Mol Cell Endocrinol 1993; 97: 135-43. [13] Sugawara T, Nomura E, Fujimoto S. Expression of enzyme associ- ated with steroid hormone synthesis and local production of steroid hormone in endometrial carcinoma cells. J Endocrinol 2004; 180: 135-44. [14] Hsieh YY, Chang CC, Tsai FJ, Lin CC, Tsai CH. Cytochrome P450c17alpha 5'-untranslated region *T/C polymorphism in endo- metriosis. J Genet 2004; 83: 189-92. [15] Olson SH, Bandera EV, Orlow I. Variants in estrogen biosynthesis genes, sex steroid hormone levels, and endometrial cancer: a HuGE review. Am J Epidemiol 2007; 165: 235-45. [16] Carey AH, Waterworth D, Patel K, et al . Polycystic ovaries and premature male pattern baldness are associated with one allele of the steroid metabolism gene CYP17. Hum Mol Genet 1994; 3: 1873-6. [17] Asghar T, Yoshida S, Nakago S, et al. Lack of association between endometriosis and the CYP17 MspA1 polymorphism in UK and Japanese populations. Gynecol Endocrinol 2005; 20: 59-63. [18] Flores I, Mandal D, Bailey-Wilson J. Search for endometriosis susceptibility genes in Puerto Rico. Abstract 1905, American Soci- ety of Human Genetics meeting, Toronto. 2004. [19] Kim S, Choi Y, Bazer FW, Spencer TE. Identification of genes in the ovine endometrium regulated by interferon tau independent of signal transducer and activator of transcription 1. Endocrinology 2003; 144: 5203-14. [20] Treloar SA, Zhao ZZ, Le L, et al. Variants in EMX2 and PTEN do not contribute to risk of endometriosis. Mol Hum Reprod 2007; 13: 587-94. [21] Zhao ZZ, Pollock PM, Thomas S, et al . Common variation in the fibroblast growth factor receptor 2 gene is not associated with endometriosis risk. Hum Reprod 2008; 23: 1661-8. [22] Revised American Fertility Society classification of endometriosis: 1985. Fertil Steril 1985; 43: 351-2. [23] Miller SA, Dykes DD, Polesky HF. A simple salting out procedure for extracting DNA from human nucleated cells. Nucleic Acids Res 1988; 16: 1215. [24] Zhao ZZ, Nyholt DR, Le L, et al . KRAS variation and risk of endometriosis. Mol Hum Reprod 2006; 12: 671-6. [25] Barrett JC, Fry B, Maller J, Daly MJ Haploview: analysis and visualization of LD and haplotype maps. Bioinformatics 2005; 21: 263-5. [26] Gabriel SB, Schaffner SF, Nguyen H, et al. The structure of haplo- type blocks in the human genome. Science 2002; 296: 2225-9. 40 The Open Reproductive Science Journal, 2008, Volume 1 Zhao et al. [27] Purcell S, Cherny SS, Sham PC. Genetic Power Calculator: de- signof linkage and association ge netic mapping studies of complex traits. Bioinformatics 2003; 19: 149-50. [28] Kado N, Kitawaki J, Obayashi H, et al . Association of the CYP17 gene and CYP19 gene polymorphisms with risk of endometriosis in Japanese women. Hum Reprod 2002; 17: 897-902. [29] De Carvalho CV, Nogueira-De-Souza NC, Costa AM, et al . Ge- netic polymorphisms of cytochrome P450cl7alpha (CYP17) and progesterone receptor genes (PROGINS) in the assessment of endometriosis risk. Gynecol Endocrinol 2007; 23: 29-33. [30] Guo SW. Association of endometriosis risk and genetic poly- morphisms involving sex steroid biosynthesis and their receptors: a meta-analysis. Gynecol Obstet Invest 2006; 61: 90-105. [31] Moskvina V, Holmans P, Schmidt KM, Craddock N. Design of case-controls studies with unscreened controls. Ann Hum Genet 2005; 69: 566-76. Received: April 23, 2008 Revised: June 17, 2008 Accepted: June 17, 2008 © Zhao et al.; Licensee Bentham Open. This is an open access article distributed under the terms of the Creative Commons Attribution License (http ://creativecommons.org/licenses/by/2.5/), which permits unrestrictive use, distribution, and reproduction in any medium, provided the original work is properly cited.