Article Text

Significant association of a M129V independent polymorphism in the 5′ UTR of the PRNP gene with sporadic Creutzfeldt-Jakob disease in a large German case-control study
  1. C Vollmert1,
  2. O Windl2,3,
  3. W Xiang2,
  4. A Rosenberger4,
  5. I Zerr5,
  6. H-E Wichmann6,
  7. H Bickeböller4,
  8. T Illig1,
  9. the KORA group,
  10. H A Kretzschmar2
  1. 1Institute of Epidemiology, GSF-National Research Center for Environment and Health, Neuherberg, Germany
  2. 2Center for Neuropathology and Prion Research (ZNP), Ludwig-Maximilians-University, Munich, Germany
  3. 3Veterinary Laboratories Agency, Weybridge, UK
  4. 4Department of Genetic Epidemiology (GEM), University of Göttingen, Göttingen, Germany
  5. 5Department of Neurology, Georg-August University, Göttingen, Germany
  6. 6Institute of Medical Informatics, Biometry and Epidemiology, Chair of Epidemiology, Ludwig Maximilians University, Munich, Germany
  1. Correspondence to:
 Professir Dr med H A Kretzschmar
 Center for Neuropathology and Prion Research, Ludwig Maximilians University Munich, Feodor-Lynen-Straße 23, D-81377 Munich, Germany; hans.kretzschmar{at}


Background: A single nucleotide polymorphism (SNP) in the coding region of the prion protein gene (PRNP) at codon 129 has been repeatedly shown to be an associated factor to sporadic Creutzfeldt-Jakob disease (sCJD), but additional major predisposing DNA variants for sCJD are still unknown. Several previous studies focused on the characterisation of polymorphisms in PRNP and the prion-like doppel gene (PRND), generating contradictory results on relatively small sample sets. Thus, extensive studies are required for validation of the polymorphisms in PRNP and PRND.

Methods: We evaluated a set of nine SNPs of PRNP and one SNP of PRND in 593 German sCJD patients and 748 German healthy controls. Genotyping was performed using MALDI-TOF mass spectrometry.

Results: In addition to PRNP 129, we detected a significant association between sCJD and allele frequencies of six further PRNP SNPs. No significant association of PRND T174M with sCJD was shown. We observed strong linkage disequilibrium within eight adjacent PRNP SNPs, including PRNP 129. However, the association of sCJD with PRNP 1368 and PRNP 34296 appeared to be independent on the genotype of PRNP 129. We additionally identified the most common haplotypes of PRNP to be over-represented or under-represented in our cohort of patients with sCJD.

Conclusion: Our study evaluated previous findings of the association of SNPs in the PRNP and PRND genes in the largest cohorts for association study in sCJD to date, and extends previous findings by defining for the first time the haplotypes associated with sCJD in a large population of the German CJD surveillance study.

  • BSE, bovine spongiform encephalopathy
  • KORA S4, Kooperative Gesundheitsforschung im Raum Augsburg, survey 4
  • LD, linkage disequilibrium
  • MALDI-TOF, matrix assisted laser desorption ionisation-time of flight
  • MS, mass spectrometry
  • PRND, prion-like doppel gene
  • PRNP, prion protein gene
  • SAP, shrimp alkaline phosphatase
  • sCJD, sporadic Creutzfeldt-Jakob disease
  • SNP, single nucleotide polymorphism
  • SSCP, single strand conformational polymorphism
  • prion protein gene
  • prion-like doppel gene
  • Creutzfeldt-Jakob disease
  • single-nucleotide polymorphism
  • genetic association study

Statistics from

Creutzfeldt-Jakob disease (CJD) is a fatal transmissible neurodegenerative disease in humans, which can present in acquired, familial, or sporadic forms. A minority of CJD cases is caused by transmission from material contaminated with infectious human CJD or bovine spongiform encephalopathy (BSE). Around 10% of cases are inherited and are associated with mutations in the prion protein gene (PRNP), located on chromosome 20p12.1–3 The majority of CJD cases (more than 90% of the CJD cases) occur sporadically, representing the most common form of CJD worldwide. Prion diseases such as CJD in humans, BSE in cattle, and scrapie in goats and sheep are characterised by spongiform degeneration, neuronal cell loss, astrogliosis, and deposition of the scrapie isoform of the prion protein, PrPSc, in the brain. The latter is derived through a posttranslational conformational conversion from the cellular prion protein, PrPC.

The aetiology of sporadic CJD (sCJD) is not known. A common polymorphism in the coding region of the PRNP gene at codon 129 (M129V) is a recognised genetic risk factor. Homozygotes for methionine and to a lesser extent for valine are over-represented in every group of sCJD examined to date.1,4–6 In addition, this polymorphism has been shown to modify the clinical and neuropathological phenotype of the disease.5 In white populations, according to studies in several countries, 37.5–45% of individuals are homozygous for methionine and 40–51% heterozygous, and 10–15% are homozygous for valine.6 However, PRNP M129V may not be the sole genetic factor predisposing to the disease. In mouse models of prion disease, inbred mouse strains with the same genotype of the prion protein gene show notable differences in the incubation time following experimental infection. Detailed studies have detected several quantitative trait loci for susceptibility to prion disease in mice, suggesting the influence of genetic factors other than PRNP.7–9

There is growing interest in polymorphisms outside the coding region of the PRNP gene and in other candidate genes. Mead et al identified 56 polymorphic sites within 25 kb of the PRNP locus, including sites within the PRNP promoter and the PRNP 3′ untranslated region. In an association study comprising 93 sCJD patients from the UK and 652 healthy controls from families in the UK and those registered with CEPH (the Centre d’Etude du Polymorphisme Humain), a significant association between an SNP upstream of PRNP exon 1 (designated SNP 1368) and sCJD was demonstrated, in addition to the strong susceptibility conferred by codon 129.10 However, this finding could not be confirmed by Croes et al in a study based on a Dutch cohort.11 McCormack et al evaluated three SNPs at position 101 bp upstream of exon 1 and at 310 bp and 385 bp downstream of exon 1 of human PRNP, which are within or adjacent to the regulatory regions of PRNP.12 This group suggested an association of CJD with the polymorphisms in the regulatory regions of PRNP. However, their data were based on a case-control study with limited sample size (sCJD, n = 25; controls, n = 100). Thus, extensive studies are required for validation of the polymorphisms of PRNP in larger cohorts.

Another gene of interest is the prion-like doppel gene (PRND), located 16 kb downstream of PRNP.13PRND shares 24% coding sequence identity with PRNP, and its product, the doppel protein (Dpl), was suggested to have biological properties antagonistic to PrP.14 Although there have been several published studies on the association of SNPs in PRND with CJD, particularly PRND T174M, other studies could not confirm this association.10,11,15–17

To evaluate the published data, which in most studies were derived from small cohorts and in part produced conflicting data, we investigated the association of PRNP and PRND polymorphisms with sporadic CJD. The association study was performed in the largest case and control cohorts to date. We analysed the genomic DNA of 593 sCJD patients and 748 healthy controls matched for age and sex from a population based German study (Kooperative Gesundheitsforschung im Raum Augsburg, survey 4, (KORA S4), 2000). Analysing nine polymorphic positions in the PRNP locus and one SNP of the PRND, we identified two SNPs that act as possible risk factors for sCJD in addition to the PRNP 129 polymorphism. Additionally, we found the most common haplotypes to be differentially distributed in healthy German controls and German patients with sCJD.



CJD suspects were referred to the German CJD surveillance unit ( and and were clinically classified as “probable” or “possible” CJD or “other”.18,19 The diagnosis of “definite” CJD was made on post-mortem examination using neuropathological criteria.20 DNA of patients with “probable” and “definite” CJD were collected and screened for pathogenic mutations in the PRNP gene. To screen for pathogenic mutations, the coding region of PRNP was amplified by PCR and subjected to either SSCP analysis or direct sequence analysis (n = 316 of all analysed cases).1 Familial cases with pathogenic mutations in PRNP1 were excluded. The remaining 593 cases, including 387 “definite” sCJD and 206 “probable” sCJD cases were enrolled in our association study.

In total, 748 age and sex matched healthy controls were taken from a population based German study performed in the city and region of Augsburg (KORA S4),21,22 which is a representative sample of the adult general population of German nationality.

All study participants gave informed written consent according to the Bavarian ethics committee and the ethics committee of the Ludwig Maximilians University of Munich, and every attempt was made to ensure anonymity of the participants.

DNA preparation and analysis

Genomic DNA of sCJD patients was extracted in most cases from blood or in some cases from frozen postmortem brain tissue using commercial kits (QIAamp Blood Kit or QIAamp Tissue Kit; Qiagen, Hilden, Germany) according to the manufacturer’s protocols. For screening for the pathogenic mutations, the coding region of PRNP was amplified by PCR and subjected to single strand conformational polymorphism (SSCP) analysis or direct sequence analysis.1 Genomic DNA from the KORA probands was extracted from blood leukocytes (Puregene DNA Isolation Kit; Gentra Systems, Minnesota 55441, USA), following the manufacturer’s instructions.

MALDI-TOF MS genotyping

Genotyping of single nucleotide polymorphisms was performed using matrix assisted laser desorption ionisation-time of flight mass spectrometry (MALDI-TOF MS) (Mass Array; Sequenom, San Diego, CA, USA). Genomic DNA (5 ng) was amplified by PCR using 0.1 U HotStar Taq DNA polymerase (Qiagen). PCR conditions were 95°C for 15 minutes, followed by 44 cycles of 95°C for 30 seconds, 56°C for 30 seconds and 72°C for 1 minute, with a final cycle of 72°C for 10 minutes. (Primer information is given in supplementary table 1, available online at PCR products were treated with shrimp alkaline phosphatase (SAP; Amersham, Freiburg, Germany) for 20 minutes at 37°C to remove excess dNTPs, followed by 10 minutes at 85°C to inactivate SAP. Base extension (homogenous MassEXTEND; Sequenom) reactions in a final volume of 10 µl contained extension primers (supplementary table 2, available online at at a final concentration of 0.54 µmol/l and 0.6 U ThermoSequenase (Amersham, Freiburg, Germany). Base extension reaction conditions were 94°C for 2 minutes, followed by 40 cycles of 94°C for 5 seconds, 52°C for 5 seconds, and 72°C for 5 seconds. The final base extension products were treated with SpectroCLEAN resin (Sequenom). Aliquots (10 nl) of the reaction solution were dispensed onto a 384 format microarray (SpectroCHIP; Sequenom) prespotted with a matrix of 3-hydroxypicolinic acid. A modified Bruker Biflex MALDI-TOF MS was used for data acquisitions from the SpectroCHIP. Genotype calling was performed in real time with MassARRAY RT software (version; Sequenom). For quality control, negative controls were included in all assays. To establish reproducibility of genotyping data, 10% of randomly selected samples were genotyped in duplicate. Genotype frequencies at all loci were subjected to Hardy-Weinberg equilibrium analysis.

Table 1

 SNP allele frequencies in sCJD cases and healthy controls

Table 2

 Odds ratios of examined SNPs in PRNP and PRND in the subgroup of PRNP 129 methionine homozygote patients

Statistical analysis

Hardy-Weinberg equilibrium of all SNPs was determined individually in controls using log likelihood tests. To evaluate the non-random association of SNPs, pairwise linkage disequilibrium (LD) statistics D′ and correlation coefficient r2 were calculated using Haploview software (version 3.2; Whitehead Institute for Biomedical Research; D′ explains the difference in frequency between the observed and expected number of SNP pairs. Being scaled to Dmax, it spans the range −1 to 1, and r2 is the squared correlation coefficient between the markers.23

Differences in genotype distribution between cases and controls were tested by the robust version of Cochran-Armitage trend test24, adjusting for multiple testing according to the Sidak’s method.25 Genetic association was expressed as odds ratios (OR) for heterozygotes and homozygotes of the rare allele versus homozygotes of the common allele, and estimated within logistic regression models. These estimations were repeated within subgroups defined by the genotype of marker PRNP M129V (that is, homozygous MM and VV, and heterozygous MV).

Haplotype frequencies for all observed markers were estimated applying the EM algorithm of Excoffier and Slatkin,26 using Arlequin software (version 2.000). Statistical analysis was performed using SAS (version 8.2; SAS Institute, Cary, NC, USA). The project was accomplished according the Guter Epidemiologischer Praxis (GEP) recommendations.27


Association study

To reveal the association of PRNP and PRND with sCJD, we genotyped nine SNPs in the PRNP locus and one SNP in the PRND locus in 593 sporadic CJD patients and 748 healthy controls (fig 1). In addition to the SNPs in the coding region of PRNP (PRNP A117A and PRNP M129V) and PRND (PRND T174M), other SNPs were identified by Mead etal.10 The SNP panel was chosen for genotyping because they are evenly distributed across the region of interest and have been shown to be highly polymorphic in the UK population and the families registered with CEPH.10

Figure 1

 Association of genotypes of PRNP SNPs with sCJD. Estimated odds ratios of heterozygous (SNP ID-1) and homozygous (SNP ID-2) minor alleles versus the homozygous common allele, and 95% confidence intervals are shown. No homozygous minor allele was observed in PRNP A117A. Detailed information is given in supplementary table 3 (available online at * p = 0.01, ** p = 0.001, *** p<0.0001.

The genotype frequencies of all SNPs studied achieved the criteria of Hardy-Weinberg equilibrium in the healthy control group (data not shown). As shown in table 1, we observed a significant association between sCJD and the allele frequency of the PRNP 129, confirming the predisposing role of the PRNP 129 polymorphism in sCJD. Additionally, we detected a highly significant association between sCJD and six other SNPs of the PRNP locus: apart from the rare SNPs PRNP 12533 and PRNP A117A, all eight PRNP SNPs with a minor allele frequency >15% showed significant differences in allele frequencies between sCJD and controls. The allele frequency of PRND T174M did not show any association with sCJD.

To define more precisely the role of the SNPs in risk of developing sCJD, we further investigated the association of various genotypes and sCJD (fig 1 and supplementary table 3). Apart from PRNP 12533 and PRND T174M, we observed a significant association with the genotypes of all other PRNP SNPs including the rare SNP PRNP A117A. While PRNP 1368 and PRNP 34296 showed increased ORs in a minor allele dose dependent manner, heterozygosity for all other PRNP SNPs significantly decreased the risk of sCJD. The homozygous common alleles of these SNPs (including PRNP 13436, 16987, 22976, M129V, and 28878) were associated with higher risk of developing sCJD compared with the homozygous minor alleles.

Table 3

 Major haplotypes of PRNP SNPs and their frequencies in cases and controls

Linkage disequilibrium analysis

We observed strong LD in controls within eight adjacent SNPs, from PRNP 1368 to PRNP 28878 (D′ = 0.78 to 1.00, r2 = 0.5 to 0.9), but markedly less LD to PRNP 34296 (D′ = 0.33 to 0.86, r2 = 0.12 to 0.53) (fig 2). Owing to the lower frequency of the minor alleles in PRNP 12533 (13%) and PRNP A117A (3%), the r2 values of these less frequent SNPs were markedly lower (r2<0.3). There was no LD between the PRNP SNPs and PRND T174M (D′ = 0.01 to 0.38, r2 = 0.00 to 0.15).

Figure 2

 SNP location and pairwise linkage disequilibrium between genotyped variants in PRNP and PRND. The top indicates the physical SNP locations relative to the exons of the PRNP and PRND locus on chromosome 20p12. SNPs in coding regions are indicated by the location of the codons and exchange of amino acids. All other SNPs are designated by their location on clone U29185. The base exchange in each SNP is given in parentheses. The first letter represents the major allele and the second the minor allele. Exons of PRNP and PRND genes are highlighted by arrows. Shading of the diamonds represents magnitude and significance of the pairwise r2 between the two SNPs, defined by the top left and the top right sides of the diamond, with black reflecting high r2 (r2>0.8) and white reflecting low r2 (r2<0.2).

Additional PRNP risk factors aside of the PRNP M129V polymorphism

Given the strong LD between SNPs in the PRNP gene, the significant association of PRNP SNPs with sCJD may be due to hitchhiking effects with the known predisposing PRNP M129V polymorphism. To identify possible risk factors additional to the PRNP M129V, we performed analysis in the PRNP M129V subgroups. This revealed a significantly higher risk for minor allele homozygotes of PRNP 1368 in the subgroup of PRNP codon 129 MM homozygotes (OR = 1.7, 95% confidence interval (CI) = 1.002 to 2.931) (table 2). In the same subgroup, only the PRNP 34296 minor allele showed a trend for association with sCJD (OR = 1.5, 95% CI = 0.94 to 2.45), independent of the PRNP M129V polymorphism. All other SNPs did not show any association with sCJD independent of the genotype of marker PRNP M129V.

SNP haplotypes

In the region between PRNP 1368 and PRNP 34296, we revealed the existence of 188 complete and incomplete haplotypes. As shown in table 3, five major PRNP haplotypes showed a frequency >5% in all cases and controls, representing 62% of all estimated haplotypes. Two of these five haplotypes (haplotypes 4 and 5 in table 3), which only differed by the allele of the marker PRNP 34296 at the far end of the screen region, were significantly over-represented in sCJD patients. The remaining three haplotypes (haplotypes 1–3 in table 3) , which differed again at marker PRNP 34296 and additionally at PRNP 28878, were significantly under-represented in patients.


The identification of genetic risk factors in sporadic CJD is important for our understanding of possible pathogenic mechanisms and the susceptibility to the disease. Efforts have been made to identify genetic risk factors other than the well described PRNP M129V polymorphism. Several previous studies focused on the characterisation of SNPs in PRNP and PRND genes, generating contradictory results on relatively small sample sets. This led us to evaluate a panel of nine SNPs in the PRNP locus and one SNP in the PRND gene in a large German sCJD cohort and population based healthy controls.

We found a significant association of sCJD with seven of nine PRNP SNPs studied, including PRNP 1368, 13436, 16987, 22976, M129V, 28878, and 34296. In agreement with previous studies, individuals with homozygosity of both methionine and valine for PRNP M129V showed an increased risk in sCJD. We further found that the association of PRNP 1368, located ∼24 kb upstream of the coding region PRNP, was independent of the genotype of PRNP M129V. This result confirmed a previous study that compared sCJD cases in the UK with either the CEPH families from France or healthy UK population.10 However, this finding disagrees with the result from an earlier study based on a Dutch population.11 The difference between these studies and the Dutch study may be due to differences in the ethnic populations, or more likely, may be influenced by the small sample size in the Dutch study (sCJD n = 23; controls n = 83). In addition to PRNP 1368, we also found that the homozygous minor allele of PRNP 34296, an SNP located ∼8 kb downstream of the 3′ coding region of PRNP, tended to increase the susceptibility to sCJD of individuals homozygous for methionine at PRNP M129V. It should be noted that the power of this study is limited owing to lower sample size within the heterozygous PRNP M129V group (96 patients versus 324 control probands) and the homozygous valine group (78 patients versus 99 control probands). The power to detect genotype related risks in these two PRNP M129V subgroups for a marker with, for example, an allele frequency of 0.3, is only 44% and 28%, respectively. Therefore, larger studies are needed for further evaluation of the association of PRNP SNPs with sCJD, independent of PRNP M129V.

Attention has been paid to the genetic variances in the regulatory region of PRNP, as there is evidence in animal models that PrP gene expression levels influence the susceptibility to prion diseases and the incubation time of disease. Overexpression of the PrP gene in transgenic animals demonstrated shortening of incubation time of animals inoculated with prions,28 while mice heterozygous for a disrupted PrP gene and expressing only half of the normal PrP have prolonged incubation times following infection, whereas PRNP knockout mice are completely resistant to prion disease. Moreover, Hardy etal stated that those who express high levels of prion protein and are homozygous at codon 129 are most susceptible to disease, and that in diseases where protein deposition is part of the process, genetic variability in the promoter of the gene should be considered as a factor influencing risk of sporadic diseases.29 Therefore, the underlying mechanism of association of SNP 1368, located in the promoter region of PRNP, with sporadic prion diseases may possibly originate in high expression of the prion protein gene. However, this hypothesis has to be proven by functional analyses.

In addition, our data clearly show the independence of association of SNP1368 with sCJD from that of M129V with sCJD. This is in line with Hardy’s “law of mass action”, which states that the concentration of the pathogenic protein might only cause the initiation of the disease through the formation of a pathogenic template.30 After initiation, the proteins adopt the same conformation of the template. This process is probably facilitated by homozygosity at codon 129, due to symmetry considerations in prion-prion interactions. This process becomes self propagating. Thus, the latter pathogenesis may be largely independent of prion protein concentration.

Previous work has identified several regulatory regions of human PRNP, located about 273 bp from exon 1,12,31 and in the intronic region between 292 and 625 bp downstream of exon 1.12 McCormack et al evaluated three SNPs in these regions in an association study with small sample sizes (controls n = 100, CJDs n = 25), and suggested that these SNPs, including an SNP corresponding to PRNP 12533 in the present study may be risk factors for CJD. However, we could not confirm a significant association between PRNP 12533 and sCJD. In addition, PRNP 13436, an SNP in the intronic regulatory region, did not show a significant association with sCJD, independent of PRNP M129V. Our data suggest that the SNPs in these previously defined regulatory regions may not function as additional risk factors to sCJD.

In the present study, we demonstrated the existence of five major haplotypes of PRNP, determined by eight SNPs. It should be noted that the markers PRNP 12533 and PRNP A117A showed the same allele in all five major haplotypes, while markers PRNP 28878 and PRNP 34296 showed different alleles within the under-represented and over-represented haplotypes. The remaining five markers (PRNP 1368, 13436, 16987, 22976 and M129V) distinguished the under-represented and over-represented haplotypes. Therefore, a predisposing role of the region spanned by these five SNPs is highly probable. This region is 24 491 bp in length, and shows strong LD as described above. The identified major haplotypes showed strong consistency with the findings in the UK study,10 both in genotypes and in their distribution in sCJD and controls. For example, the two most common haplotypes in the UK study (haplotypes A and B; table 3) coincide with the two most common haplotypes in the present study (haplotypes 1 and 5), suggesting that German PRNP genealogy can also be characterised by these two proposed major haplotype branches.

Recently, attention has also been focused on PRND and the association of SNPs at the PRND locus with CJD. However, the results of earlier studies are contradictory. Studies on PRND T174M either showed no significant relationship,10,16,17 or a significantly increased risk for those heterozygous15 or homozygous for methionine.11 The present work showed neither significant association of different genotypes of PRND T174M with sCJD, nor strong linkage disequilibrium between the PRNP SNPs and PRND T174M, confirming the finding of a previous study based on UK and French populations.10 This result suggests a less important role of PRND T174M in prion disease pathogenesis. However, PRND cannot be excluded as a possible candidate gene. According to our study and a previous study,10PRNP 34296 demonstrated a significant association with sCJD and was located in a region of strong LD, also including SNPs upstream of PRND.10 Thus, associations of other genetic variants of PRND with sCJD, especially those upstream of PRND, independent of PRND T174M, may be possible.

In conclusion, we identified a significant association of seven PRNP SNPs with sCJD. Most of these associations are hitchhiking effects with the known predisposing effect of the PRNP M129V polymorphism. We confirmed the results of a previous study indicating that PRNP 1368 is associated with sCJD, independent of PRNP M129V. In addition, our results suggest that PRNP 34296 may also be an additional independent risk factor. Furthermore, we detected a strong LD region from PRNP 1368 to PRNP 28878, and identified the major haplotypes of PRNP SNPs that were over-represented or under-represented in sCJD cases. The cases and controls analysed in this study are the largest cohorts for association study in sCJD to date. These cohorts provide a basis for the association studies of further genetic variants in candidate genes involved in the susceptibility and pathogenesis of sCJD.


Accession numbers and URLs for data in this article are as follows:

NCBI Entrez Gene, for clones U29185 and AL133396;

Online Mendelian Inheritance in Man (OMIM) for PRNP (MIM 176640), CJD (MIM 123400), and PRND (MIM604263);


We thank all study participants. We further thank E Staniszewski, S Walter and M Wimmer for excellent technical assistance and G Fischer for perfect data management. Genotyping was performed in the GSF genotyping facility located in the Genome Analysis Center (GAC) chaired by J Adamski. This study was supported within the German National Genomic Research Network (NGFN) by the Federal Ministry of Education and Research (BMBF). The German CJD surveillance study was supported by a grant from the Bundesministerium für Gesundheit (BMG) (GZ: 325-4471-02/15). The KORA research platform (KORA, Cooperative Research in the Region of Augsburg) was initiated and financed by the GSF-National Research Centre for Environment and Health, which is funded by the German Federal Ministry of Education and Research and of the State of Bavaria.


Supplementary materials

  • Files in this Data Supplement:


  • The KORA group consists of H-E Wichmann (speaker), H Löwel, C Meisinger, T Illig, R Holle, J John, and their coworkers, who are responsible for the design and conduct of the KORA studies.

  • The first two authors contributed equally to this work.

  • Competing interests: there are no competing interests.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.