See related article on pg 992
Next-Generation Sequencing for Mutation Detection in Heritable Skin Diseases: The Paradigm of Pseudoxanthoma Elasticum Andrew P. South1, Qiaoli Li1 and Jouni Uitto1 Next-generation sequencing applied either to the entire genome or to a subset, such as a whole exome, has revolutionized the search for pathogenic mutations in heritable diseases, including genodermatoses. In this issue, Hosen et al. applied whole-exome sequencing to identify potential pathogenic mutations in four candidate genes associated with pseudoxanthoma elasticum, the prototype of ectopic mineralization disorders. The study highlights the advantages of this approach over traditional Sanger sequencing, including expedience and cost, but it also illustrates some of the challenges encountered in implementing this rapidly evolving technology. Journal of Investigative Dermatology (2015) 135, 937–940; doi:10.1038/jid.2014.521
One of the emerging paradigms of contemporary medicine is the concept of personalized medicine, an approach that tailors the diagnosis and its management to match unique findings in individual patients. Of course, we have always practiced personalized medicine based on an individual’s clinical presentation and laboratory findings. However, what makes the concept of personalized medicine topical at this juncture is that we now have an unprecedented capability to decipher the genetic makeup of each individual by sequencing an entire genome or a subset of it, such as a whole exome. The primary purpose is to find pathogenic mutations that would explain phenotypic manifestations in a patient, but this approach also allows us to identify sequence variants that may predispose individuals to late-onset diseases or that may modify the phenotypic presentation of a primary pathogenic mutation.
Mutation detection in candidate genes over the last several decades has provided new opportunities to manage genodermatoses (Uitto, 2009). Identification of specific mutations in Mendelian disorders has provided means to confirm the diagnosis with subclassification, and the mutation databases have allowed assessment of genotype/phenotype correlations with long-term prognostication of disease outcome. Identification of mutations in families at risk for severe skin diseases has provided a basis for genetic counseling, and it has also allowed carrier detection and presymptomatic testing in diseases with late onset. Finally, identification of specific mutations has formed the basis for prenatal testing and preimplantation genetic diagnosis, and, importantly, emerging allele-specific treatment approaches often require precise knowledge about a mutation. Until recently, mutation detection in candidate genes has relied primarily on Sanger sequencing, which examines
1
Department of Dermatology and Cutaneous Biology, Sidney Kimmel Medical College at Thomas Jefferson University, Philadelphia, Pennsylvania, USA Correspondence: Jouni Uitto, Department of Dermatology and Cutaneous Biology, Sidney Kimmel Medical College at Thomas Jefferson University, 233S. 10th Street, Suite 450 BLSB, Philadelphia, Pennsylvania 19107, USA. E-mail: [email protected] Abbreviations: NGS, next generation sequencing; WES, whole exome sequencing; PXE, pseudoxanthoma elasticum
& 2015 The Society for Investigative Dermatology
COMMENTARY
potential candidate genes one at a time. It has become evident, however, that several heritable skin diseases have underlying mutations in different genes, with resultant phenocopies, and in many cases attempts to identify candidate genes by conventional approaches, such as positional cloning and homozygosity mapping, have not been successful. The landscape of mutation detection is changing rapidly with genome-wide, next-generation sequencing (NGS) approaches that allow global evaluation of sequence variants in an individual case. NGS technologies have revolutionized our ability to survey nucleic acid base-pair composition in the entire genome. Just over a decade ago, Sanger sequencing was the mainstay methodology, with a cost of approximately $1,000 per million base-pairs. Now, the prospect of an entire human genome (some 3.2 billion base-pairs) being sequenced for less than $1,000 is a reality (Hayden, 2014; Wetterstrand, 2014). Although Sanger sequencing assays the predominant nucleotide or nucleotides at a given position from a pool of DNA, NGS yields sequences from individual DNA molecules in massive numbers: Sanger sequencing can give us the average sequence of 1,000 DNA molecules representing a single genomic position, whereas NGS can give us the individual sequence from each of those 1,000 DNA molecules or 1,000 molecules each from different genomic positions. Technologies of NGS, in various formats and different capabilities and capacities, are available, ranging from bench-top instruments taking just a few hours to run to large machines requiring dedicated laboratory space and weeks to generate and analyze data (for review see Metzker (2010). Increasing application of NGS technologies to numerous biological and clinical questions is driving the cost down, and progress toward commercial application of nanopore technology, which has the potential to revolutionize cost and output speed even further, is gathering momentum (Laszlo et al., 2014). www.jidonline.org
937
COMMENTARY
Clinical Implications
Next-generation sequencing provides an expedient and cost-effective platform to identify pathogenic mutations in patients with genodermatoses.
This technology will also assist in identification of modifier genes contributing to the phenotypic variability of such diseases.
Identification of specific pathogenic mutations will assist in implementing personalized medicine, consisting of improved genetic counseling, prognostication of long-term outcome, prenatal testing, and preimplantation genetic diagnosis, as well as the development of allele-specific treatment modalities.
NGS is not without drawbacks, however, and the field is continuing to evolve to address a number of issues. Costs per base-pair are usually quoted on the basis of sequencing alone, whereas DNA isolation, storage, and preparation can incur significant cost and capital outlay. In addition, the cost of labor and expertise associated with bioinformatics analyses of data should certainly be considered as part of any NGS project. Current technologies of NGS are also inherently error prone (ranging from 0.1 to 15% compared with Sanger sequencing at 0.1–1%), and therefore there is a need to sequence more than a handful of DNA molecules from the same genomic position to be certain that any variation at that position is correct. This is called coverage: 10 DNA molecules corresponding to a single genomic locus is 10x coverage, and when preparing libraries of DNA molecules simultaneously from different loci, coverage is rarely even. As an example, a 100 average coverage of a 39-exon gene can give differences between exons that are orders of magnitude (South et al., 2014). Careful design and optimization is required to be confident that all areas of interest are sequenced with sufficient confidence to identify all potential variants (Chilamakuri et al., 2014). The other problem with such large amounts of sequence information, in particular from multiple loci generated by wholeexome sequencing (WES), is mapping the individual sequences back to a reference, which itself is still evolving (http://ncbiinsights.ncbi.nlm.nih.gov/ 2013/12/24/introducing-the-new-humangenome-assembly-grch38/ Accessed 938
10/20/2014). Many algorithms are designed to map short, 100 bp sequences back to a large, single sequence reference genome, and significant deviations from this (in particular, insertions or deletions) are often missed by current workflows because of repetitive elements within the genome and within the sequence itself (Li and Durbin, 2009). Nevertheless, decreasing costs and increasing amounts of data, and an array of publicly available analysis tools, such as the Galaxy platform (Giardine et al., 2005), means that we will soon see even more applications of NGS to diagnostic services (Takeichi et al., 2013). In this issue, Hosen et al. (Hosen et al., 2015) have approached the genetic basis of pseudoxanthoma elasticum (PXE), a heritable multisystem disorder with skin manifestations, with an expanded, NGS-based mutation detection strategy. PXE, an autosomal recessive disease, is the paradigm of ectopic mineralization disorders affecting the skin, as well as the eyes and arterial blood vessels, with considerable morbidity and occasional mortality (Uitto et al., 2010). The phenotypic presentation of PXE is quite variable, but the classic form demonstrates late onset with slowly progressing tissue involvement and clinical manifestations that lead to loss of visual acuity and cardiovascular complications. PXE was initially shown to be caused by mutations in the ABCC6 gene, which encodes a putative efflux transporter, ABCC6, and it is expressed primarily in the liver and kidneys. Although the pathomechanistic details leading from mutations in ABCC6 to ectopic minera-
Journal of Investigative Dermatology (2015), Volume 135
lization of peripheral connective tissues remain to be identified, it is clear that in the majority of families (over 80%) PXE is caused by mutations in this gene (Uitto et al., 2013). The spectrum of ABCC6 mutations consists of missense or nonsense mutations, or small insertions and deletions, which result in frame-shift and premature termination of translation, as well as of large gene deletions. However, no definitive genotype/phenotype correlations have been discovered, possibly implying the presence of modifier genes and a role for environmental and life-style factors (Pfendner et al., 2007). In addition to ABCC6, phenotypic presentations similar to classic PXE have been encountered in patients with mutations in ENPP1, a gene that underlies generalized arterial calcification of infancy. This disease is often diagnosed by prenatal ultrasound demonstrating extensive mineralization of arterial blood vessels and leading in most cases to demise during the first year of life (Nitschke and Rutsch, 2012). In addition, PXE-like cutaneous findings have been reported in patients with mutations in GGCX, a gene encoding an enzyme involved in the vitamin K cycle in the liver (Vanakker et al., 2007; Li et al., 2009). Thus, there is evidence of genetic heterogeneity resulting in a spectrum of PXE-like phenotypes. The investigators in the study by Hosen et al. (2015) streamlined the mutation detection strategy in PXE by exploring NGS in a manner that allows identification of mutations in many potential candidate genes, including ABCC6, ENPP1, GGCX, and VKORC1, in a single workflow. Their approach utilized WES, which allows the assessment of B1.5% of the human genome (some 50 million base-pairs) containing the protein-coding sequences of about 20,000 genes (Figure 1). By doing so, they were able to generate a list of variants specific to each patient, which can potentially total up to 25,000 singlenucleotide polymorphisms when compared with the reference genome (Bamshad et al., 2011). The prior knowledge of candidate genes known to cause PXE, when mutated, and the types of mutations likely to lead to pathology no doubt greatly assisted
COMMENTARY
Genomic DNA containing exons (~1.5% sequence)
Exon
Exon
Exon
Genomic DNA fragments
Denature and mix with “baits” (immobilized exon sequences) Exome capture-hybridization and purification of exon sequences Discard non-exome fragments
Sequence exome fragments 2.2
q12.3
q13.1
q13.2
q13.31 q13.32
18.470 kb
Coverage of exon sequences
Figure 1. Schematic presentation of basic principles of whole-exome sequencing. Genomic DNA is first fragmented and denatured, followed by hybridization to baits containing sequences representing protein-coding exons of B20,000 genes (depicted in red). These exon-containing DNA fragments are eluted from the baits, processed by techniques depending on the sequencing platform, and sequenced to yield individual DNA fragment sequences or ‘‘reads’’. The sequence information can be aligned to the reference genome, and alignment variations can be scored to uncover pathogenic mutations.
their approach. In this context, it should be noted that de novo gene hunting through WES or whole-genome approaches has disclosed a surprising number of loss-of-function mutations in apparently healthy individuals (Tennessen et al., 2012). This is where basic genetics has a role and parent–child trios or sibpairs will facilitate comparisons to narrow down variants to a manageable number. This could possibly be a reason for the difficulties encountered by Hosen et al. (2015) to find potential modifiers of the disease phenotypes that NGS may ultimately provide for PXE. The study by Hosen et al. (2015) also highlights several of the challenges and pitfalls of WES. For example, although the overall coverage on the targeted bases was 37x, the overall sequencing depth of Z5 reads was highly variable. Specifically, 93% of the exons in ABCC6 met these criteria, but only 33% of exons in VKORC1 had the
sequence depth of more than 5 reads. This required the investigators to resort to traditional Sanger sequencing for all exons with a depth of less than 20x. Also, the importance of preparation of quality DNA is emphasized by their finding that, in two patients, coverage was poor for all four genes tested and this correlated with the poor overall sequencing depth of the exome. Another challenge of NGS relates to the bioinformatics and particularly to the ability to call sequence variants pathogenic. For example, as a rule of thumb, missense mutations identified in more than 1% of a population are not considered to be disease-causing. However, this does not imply that all nucleotide polymorphisms present in the SNP database at a frequency of o1% are disease-causing. In fact, these rare polymorphisms constitute the majority of variation between individuals in larger populations (Tennessen et al., 2012).
Similarly, some of the SNPs with low allelic frequency affecting the splice sites could be considered pathogenic, yet splicing prediction programs do not reveal evidence for pathogenicity. Another shortfall of WES is that it may not detect large insertions or deletions that can be searched separately by multiplex ligation–dependent probe amplification. However, multiplex ligation–dependent probe amplification is not available for many genes, including three of the candidate genes in the study by Hosen et al. Finally, a critical element in looking for mutations in cohorts, such as PXE, is making an accurate clinical diagnosis. This problem emphasizes the importance of clinical acumen and adherence to diagnostic criteria for patients who may fall into a spectrum of genotypically and phenotypically overlapping diseases, as in the case of PXE (Uitto et al., 2014). Specifically, in the ectopic www.jidonline.org
939
COMMENTARY
mineralization disorders affecting the skin and arterial vasculature, there are now as many as half a dozen genes that can result in presentations often with overlapping clinical features (Li and Uitto, 2013). Nevertheless, NGS clearly provides novel tools and strategies to identify previously unrecognized candidate genes in patients with similar phenotypic presentations, and refinement of the bioinformatics analyses in the context of clinical phenotypes may allow identification of modifier genes that contribute to the clinical heterogeneity of these diseases, including PXE. CONFLICT OF INTEREST
The authors state no conflict of interest.
and genotype-phenotype analysis in a large international case series affected by pseudoxanthoma elasticum. J Med Genet 44:621–8 South AP, Purdie KJ, Watt SA et al. (2014) NOTCH1 mutations occur early during cutaneous squamous cell carcinogenesis. J Invest Dermatol 134:2630–8 Takeichi T, Nanda A, Liu L et al. (2013) Impact of next generation sequencing on diagnostics in a genetic skin disease clinic. Exp Dermatol 22:825–31 Tennessen JA, Bigham AW, O’Connor TD et al. (2012) Evolution and functional impact of rare coding variation from deep sequencing of human exomes. Science 337:64–9 Uitto J (2009) Progress in heritable skin diseases: translational implications of mutation analysis and prospects of molecular therapies. Acta Derm Venereol 89:228–35 Uitto J, Jiang Q, Varadi A et al. (2014) Pseudoxanthoma elasticum: diagnostic features,
classification and treatment options. Expert Opin Orphan Drugs 2:567–77 Uitto J, Li Q, Jiang Q (2010) Pseudoxanthoma elasticum: molecular genetics and putative pathomechanisms. J Invest Dermatol 130: 661–70 Uitto J, Varadi A, Bercovitch L et al. (2013) Pseudoxanthoma elasticum: progress in research toward treatment: summary of the 2012 PXE International Research Meeting. J Invest Dermatol 133:1444–9 Vanakker OM, Martin L, Gheduzzi D et al. (2007) Pseudoxanthoma elasticum-like phenotype with cutis laxa and multiple coagulation factor deficiency represents a separate genetic entity. J Invest Dermatol 127: 581–587 Wetterstrand KA (2014) DNA sequencing costs: Data from the NHGRI genome sequencing program (GSP) Available at www. genome.gov/sequencingcosts. Accessed on 20 October 2014
ACKNOWLEDGMENTS We thank Carol Kelly for the assistance with manuscript preparation. QL is recipient of an NIH/NIAMS Grant K01 AR064766.
REFERENCES Bamshad MJ, Ng SB, Bigham AW et al. (2011) Exome sequencing as a tool for Mendelian disease gene discovery. Nat Rev Genet 12:745–55 Chilamakuri CS, Lorenz S, Madoui MA et al. (2014) Performance comparison of four exome capture systems for deep sequencing. BMC Genomics 15:449 Giardine B, Riemer C, Hardison RC et al. (2005) Galaxy: a platform for interactive large-scale genome analysis. Genome Res 15:1451–5 Hayden EC (2014) Technology: The $1,000 genome. Nature 507:294–5 Hosen MJ, Van Nieuwerburgh F, Steyaert W et al. (2015) Efficiency of exome sequencing for the molecular diagnosis of pseudoxanthoma elasticum. J Invest Dermatol 135:992–8 Laszlo AH, Derrington IM, Ross BC et al. (2014) Decoding long nanopore sequencing reads of natural DNA. Nat Biotechnol 32:829–33 Li H, Durbin R (2009) Fast and accurate short read alignment with Burrows-Wheeler transform. Bioinformatics 25:1754–60 Li Q, Schurgers LJ, Smith AC et al. (2009) Co-existent pseudoxanthoma elasticum and vitamin K-dependent coagulation factor deficiency: compound heterozygosity for mutations in the GGCX gene. Am J Pathol 174:534–40 Li Q, Uitto J (2013) Mineralization/anti-mineralization networks in the skin and vascular connective tissues. Am J Pathol 183:10–8 Metzker ML (2010) Sequencing technologies—the next generation. Nat Rev Genet 11:31–46
940
See related article on pg 1033
The Double Life of Connexin Channels: Single Is a Treat Roberto Bruzzone1,2 Although several genetic diseases are caused by mutations in channels made by connexin family members, there has been little progress in the development and validation of therapeutic options. An in vitro study in this issue of JID suggests that an anti-malarial drug may be beneficial in keratitis-ichthyosis deafness, a severe conexin channel disease associated with potentially fatal recurrent infections. Journal of Investigative Dermatology (2015) 135, 940–943; doi:10.1038/jid.2014.524
Ever since the report of a human genetic disease linked to connexin mutations more than 20 years ago, pharmacological modulation of this class of channels has acquired clinical relevance. Connexins are special channels, as they consist of oligomeric structures, termed connexons or hemichannels, which can either dock with cognate hemichannels to form complete intercellular (gap junction) channels that provide ionic and metabolic coupling between adjacent cells, or function as unpaired connexons, whose physiological role, if any,
remains to be firmly established (Saez et al., 2003). Of note, dysregulated hemichannel activity by mutated connexins has been suspected to have a pathogenic role in some diseases, perhaps as a consequence of the resulting loss of cell integrity (Xu and Nicholson, 2013; Kelly et al., 2014). In the current issue of JID by Levit et al. (2015, this issue) describe the effect of a molecule approved for anti-malarial treatment, on some of the most common Cx26 mutations that cause keratitis-ichthyosis deafness (KID), a
Nitschke Y, Rutsch F (2012) Generalized arterial calcification of infancy and pseudoxanthoma elasticum: two sides of the same coin. Front Genet 3:302
1
Pfendner EG, Vanakker OM, Terry SF et al. (2007) Mutation detection in the ABCC6 gene
Correspondence: Roberto Bruzzone, HKU-Pasteur Research Pole, School of Public Health, The University of Hong Kong, Hong Kong, People’s Republic of China. E-mail: [email protected]
HKU-Pasteur Research Pole, School of Public Health The University of Hong Kong, Hong Kong, People’s Republic of China and 2Department of Cell Biology and Infection, Institut Pasteur, Paris, France
Journal of Investigative Dermatology (2015), Volume 135
Next-Generation Sequencing for Mutation Detection in Heritable Skin Diseases: The Paradigm of Pseudoxanthoma Elasticum Andrew P. South1, Qiaoli Li1 and Jouni Uitto1 Next-generation sequencing applied either to the entire genome or to a subset, such as a whole exome, has revolutionized the search for pathogenic mutations in heritable diseases, including genodermatoses. In this issue, Hosen et al. applied whole-exome sequencing to identify potential pathogenic mutations in four candidate genes associated with pseudoxanthoma elasticum, the prototype of ectopic mineralization disorders. The study highlights the advantages of this approach over traditional Sanger sequencing, including expedience and cost, but it also illustrates some of the challenges encountered in implementing this rapidly evolving technology. Journal of Investigative Dermatology (2015) 135, 937–940; doi:10.1038/jid.2014.521
One of the emerging paradigms of contemporary medicine is the concept of personalized medicine, an approach that tailors the diagnosis and its management to match unique findings in individual patients. Of course, we have always practiced personalized medicine based on an individual’s clinical presentation and laboratory findings. However, what makes the concept of personalized medicine topical at this juncture is that we now have an unprecedented capability to decipher the genetic makeup of each individual by sequencing an entire genome or a subset of it, such as a whole exome. The primary purpose is to find pathogenic mutations that would explain phenotypic manifestations in a patient, but this approach also allows us to identify sequence variants that may predispose individuals to late-onset diseases or that may modify the phenotypic presentation of a primary pathogenic mutation.
Mutation detection in candidate genes over the last several decades has provided new opportunities to manage genodermatoses (Uitto, 2009). Identification of specific mutations in Mendelian disorders has provided means to confirm the diagnosis with subclassification, and the mutation databases have allowed assessment of genotype/phenotype correlations with long-term prognostication of disease outcome. Identification of mutations in families at risk for severe skin diseases has provided a basis for genetic counseling, and it has also allowed carrier detection and presymptomatic testing in diseases with late onset. Finally, identification of specific mutations has formed the basis for prenatal testing and preimplantation genetic diagnosis, and, importantly, emerging allele-specific treatment approaches often require precise knowledge about a mutation. Until recently, mutation detection in candidate genes has relied primarily on Sanger sequencing, which examines
1
Department of Dermatology and Cutaneous Biology, Sidney Kimmel Medical College at Thomas Jefferson University, Philadelphia, Pennsylvania, USA Correspondence: Jouni Uitto, Department of Dermatology and Cutaneous Biology, Sidney Kimmel Medical College at Thomas Jefferson University, 233S. 10th Street, Suite 450 BLSB, Philadelphia, Pennsylvania 19107, USA. E-mail: [email protected] Abbreviations: NGS, next generation sequencing; WES, whole exome sequencing; PXE, pseudoxanthoma elasticum
& 2015 The Society for Investigative Dermatology
COMMENTARY
potential candidate genes one at a time. It has become evident, however, that several heritable skin diseases have underlying mutations in different genes, with resultant phenocopies, and in many cases attempts to identify candidate genes by conventional approaches, such as positional cloning and homozygosity mapping, have not been successful. The landscape of mutation detection is changing rapidly with genome-wide, next-generation sequencing (NGS) approaches that allow global evaluation of sequence variants in an individual case. NGS technologies have revolutionized our ability to survey nucleic acid base-pair composition in the entire genome. Just over a decade ago, Sanger sequencing was the mainstay methodology, with a cost of approximately $1,000 per million base-pairs. Now, the prospect of an entire human genome (some 3.2 billion base-pairs) being sequenced for less than $1,000 is a reality (Hayden, 2014; Wetterstrand, 2014). Although Sanger sequencing assays the predominant nucleotide or nucleotides at a given position from a pool of DNA, NGS yields sequences from individual DNA molecules in massive numbers: Sanger sequencing can give us the average sequence of 1,000 DNA molecules representing a single genomic position, whereas NGS can give us the individual sequence from each of those 1,000 DNA molecules or 1,000 molecules each from different genomic positions. Technologies of NGS, in various formats and different capabilities and capacities, are available, ranging from bench-top instruments taking just a few hours to run to large machines requiring dedicated laboratory space and weeks to generate and analyze data (for review see Metzker (2010). Increasing application of NGS technologies to numerous biological and clinical questions is driving the cost down, and progress toward commercial application of nanopore technology, which has the potential to revolutionize cost and output speed even further, is gathering momentum (Laszlo et al., 2014). www.jidonline.org
937
COMMENTARY
Clinical Implications
Next-generation sequencing provides an expedient and cost-effective platform to identify pathogenic mutations in patients with genodermatoses.
This technology will also assist in identification of modifier genes contributing to the phenotypic variability of such diseases.
Identification of specific pathogenic mutations will assist in implementing personalized medicine, consisting of improved genetic counseling, prognostication of long-term outcome, prenatal testing, and preimplantation genetic diagnosis, as well as the development of allele-specific treatment modalities.
NGS is not without drawbacks, however, and the field is continuing to evolve to address a number of issues. Costs per base-pair are usually quoted on the basis of sequencing alone, whereas DNA isolation, storage, and preparation can incur significant cost and capital outlay. In addition, the cost of labor and expertise associated with bioinformatics analyses of data should certainly be considered as part of any NGS project. Current technologies of NGS are also inherently error prone (ranging from 0.1 to 15% compared with Sanger sequencing at 0.1–1%), and therefore there is a need to sequence more than a handful of DNA molecules from the same genomic position to be certain that any variation at that position is correct. This is called coverage: 10 DNA molecules corresponding to a single genomic locus is 10x coverage, and when preparing libraries of DNA molecules simultaneously from different loci, coverage is rarely even. As an example, a 100 average coverage of a 39-exon gene can give differences between exons that are orders of magnitude (South et al., 2014). Careful design and optimization is required to be confident that all areas of interest are sequenced with sufficient confidence to identify all potential variants (Chilamakuri et al., 2014). The other problem with such large amounts of sequence information, in particular from multiple loci generated by wholeexome sequencing (WES), is mapping the individual sequences back to a reference, which itself is still evolving (http://ncbiinsights.ncbi.nlm.nih.gov/ 2013/12/24/introducing-the-new-humangenome-assembly-grch38/ Accessed 938
10/20/2014). Many algorithms are designed to map short, 100 bp sequences back to a large, single sequence reference genome, and significant deviations from this (in particular, insertions or deletions) are often missed by current workflows because of repetitive elements within the genome and within the sequence itself (Li and Durbin, 2009). Nevertheless, decreasing costs and increasing amounts of data, and an array of publicly available analysis tools, such as the Galaxy platform (Giardine et al., 2005), means that we will soon see even more applications of NGS to diagnostic services (Takeichi et al., 2013). In this issue, Hosen et al. (Hosen et al., 2015) have approached the genetic basis of pseudoxanthoma elasticum (PXE), a heritable multisystem disorder with skin manifestations, with an expanded, NGS-based mutation detection strategy. PXE, an autosomal recessive disease, is the paradigm of ectopic mineralization disorders affecting the skin, as well as the eyes and arterial blood vessels, with considerable morbidity and occasional mortality (Uitto et al., 2010). The phenotypic presentation of PXE is quite variable, but the classic form demonstrates late onset with slowly progressing tissue involvement and clinical manifestations that lead to loss of visual acuity and cardiovascular complications. PXE was initially shown to be caused by mutations in the ABCC6 gene, which encodes a putative efflux transporter, ABCC6, and it is expressed primarily in the liver and kidneys. Although the pathomechanistic details leading from mutations in ABCC6 to ectopic minera-
Journal of Investigative Dermatology (2015), Volume 135
lization of peripheral connective tissues remain to be identified, it is clear that in the majority of families (over 80%) PXE is caused by mutations in this gene (Uitto et al., 2013). The spectrum of ABCC6 mutations consists of missense or nonsense mutations, or small insertions and deletions, which result in frame-shift and premature termination of translation, as well as of large gene deletions. However, no definitive genotype/phenotype correlations have been discovered, possibly implying the presence of modifier genes and a role for environmental and life-style factors (Pfendner et al., 2007). In addition to ABCC6, phenotypic presentations similar to classic PXE have been encountered in patients with mutations in ENPP1, a gene that underlies generalized arterial calcification of infancy. This disease is often diagnosed by prenatal ultrasound demonstrating extensive mineralization of arterial blood vessels and leading in most cases to demise during the first year of life (Nitschke and Rutsch, 2012). In addition, PXE-like cutaneous findings have been reported in patients with mutations in GGCX, a gene encoding an enzyme involved in the vitamin K cycle in the liver (Vanakker et al., 2007; Li et al., 2009). Thus, there is evidence of genetic heterogeneity resulting in a spectrum of PXE-like phenotypes. The investigators in the study by Hosen et al. (2015) streamlined the mutation detection strategy in PXE by exploring NGS in a manner that allows identification of mutations in many potential candidate genes, including ABCC6, ENPP1, GGCX, and VKORC1, in a single workflow. Their approach utilized WES, which allows the assessment of B1.5% of the human genome (some 50 million base-pairs) containing the protein-coding sequences of about 20,000 genes (Figure 1). By doing so, they were able to generate a list of variants specific to each patient, which can potentially total up to 25,000 singlenucleotide polymorphisms when compared with the reference genome (Bamshad et al., 2011). The prior knowledge of candidate genes known to cause PXE, when mutated, and the types of mutations likely to lead to pathology no doubt greatly assisted
COMMENTARY
Genomic DNA containing exons (~1.5% sequence)
Exon
Exon
Exon
Genomic DNA fragments
Denature and mix with “baits” (immobilized exon sequences) Exome capture-hybridization and purification of exon sequences Discard non-exome fragments
Sequence exome fragments 2.2
q12.3
q13.1
q13.2
q13.31 q13.32
18.470 kb
Coverage of exon sequences
Figure 1. Schematic presentation of basic principles of whole-exome sequencing. Genomic DNA is first fragmented and denatured, followed by hybridization to baits containing sequences representing protein-coding exons of B20,000 genes (depicted in red). These exon-containing DNA fragments are eluted from the baits, processed by techniques depending on the sequencing platform, and sequenced to yield individual DNA fragment sequences or ‘‘reads’’. The sequence information can be aligned to the reference genome, and alignment variations can be scored to uncover pathogenic mutations.
their approach. In this context, it should be noted that de novo gene hunting through WES or whole-genome approaches has disclosed a surprising number of loss-of-function mutations in apparently healthy individuals (Tennessen et al., 2012). This is where basic genetics has a role and parent–child trios or sibpairs will facilitate comparisons to narrow down variants to a manageable number. This could possibly be a reason for the difficulties encountered by Hosen et al. (2015) to find potential modifiers of the disease phenotypes that NGS may ultimately provide for PXE. The study by Hosen et al. (2015) also highlights several of the challenges and pitfalls of WES. For example, although the overall coverage on the targeted bases was 37x, the overall sequencing depth of Z5 reads was highly variable. Specifically, 93% of the exons in ABCC6 met these criteria, but only 33% of exons in VKORC1 had the
sequence depth of more than 5 reads. This required the investigators to resort to traditional Sanger sequencing for all exons with a depth of less than 20x. Also, the importance of preparation of quality DNA is emphasized by their finding that, in two patients, coverage was poor for all four genes tested and this correlated with the poor overall sequencing depth of the exome. Another challenge of NGS relates to the bioinformatics and particularly to the ability to call sequence variants pathogenic. For example, as a rule of thumb, missense mutations identified in more than 1% of a population are not considered to be disease-causing. However, this does not imply that all nucleotide polymorphisms present in the SNP database at a frequency of o1% are disease-causing. In fact, these rare polymorphisms constitute the majority of variation between individuals in larger populations (Tennessen et al., 2012).
Similarly, some of the SNPs with low allelic frequency affecting the splice sites could be considered pathogenic, yet splicing prediction programs do not reveal evidence for pathogenicity. Another shortfall of WES is that it may not detect large insertions or deletions that can be searched separately by multiplex ligation–dependent probe amplification. However, multiplex ligation–dependent probe amplification is not available for many genes, including three of the candidate genes in the study by Hosen et al. Finally, a critical element in looking for mutations in cohorts, such as PXE, is making an accurate clinical diagnosis. This problem emphasizes the importance of clinical acumen and adherence to diagnostic criteria for patients who may fall into a spectrum of genotypically and phenotypically overlapping diseases, as in the case of PXE (Uitto et al., 2014). Specifically, in the ectopic www.jidonline.org
939
COMMENTARY
mineralization disorders affecting the skin and arterial vasculature, there are now as many as half a dozen genes that can result in presentations often with overlapping clinical features (Li and Uitto, 2013). Nevertheless, NGS clearly provides novel tools and strategies to identify previously unrecognized candidate genes in patients with similar phenotypic presentations, and refinement of the bioinformatics analyses in the context of clinical phenotypes may allow identification of modifier genes that contribute to the clinical heterogeneity of these diseases, including PXE. CONFLICT OF INTEREST
The authors state no conflict of interest.
and genotype-phenotype analysis in a large international case series affected by pseudoxanthoma elasticum. J Med Genet 44:621–8 South AP, Purdie KJ, Watt SA et al. (2014) NOTCH1 mutations occur early during cutaneous squamous cell carcinogenesis. J Invest Dermatol 134:2630–8 Takeichi T, Nanda A, Liu L et al. (2013) Impact of next generation sequencing on diagnostics in a genetic skin disease clinic. Exp Dermatol 22:825–31 Tennessen JA, Bigham AW, O’Connor TD et al. (2012) Evolution and functional impact of rare coding variation from deep sequencing of human exomes. Science 337:64–9 Uitto J (2009) Progress in heritable skin diseases: translational implications of mutation analysis and prospects of molecular therapies. Acta Derm Venereol 89:228–35 Uitto J, Jiang Q, Varadi A et al. (2014) Pseudoxanthoma elasticum: diagnostic features,
classification and treatment options. Expert Opin Orphan Drugs 2:567–77 Uitto J, Li Q, Jiang Q (2010) Pseudoxanthoma elasticum: molecular genetics and putative pathomechanisms. J Invest Dermatol 130: 661–70 Uitto J, Varadi A, Bercovitch L et al. (2013) Pseudoxanthoma elasticum: progress in research toward treatment: summary of the 2012 PXE International Research Meeting. J Invest Dermatol 133:1444–9 Vanakker OM, Martin L, Gheduzzi D et al. (2007) Pseudoxanthoma elasticum-like phenotype with cutis laxa and multiple coagulation factor deficiency represents a separate genetic entity. J Invest Dermatol 127: 581–587 Wetterstrand KA (2014) DNA sequencing costs: Data from the NHGRI genome sequencing program (GSP) Available at www. genome.gov/sequencingcosts. Accessed on 20 October 2014
ACKNOWLEDGMENTS We thank Carol Kelly for the assistance with manuscript preparation. QL is recipient of an NIH/NIAMS Grant K01 AR064766.
REFERENCES Bamshad MJ, Ng SB, Bigham AW et al. (2011) Exome sequencing as a tool for Mendelian disease gene discovery. Nat Rev Genet 12:745–55 Chilamakuri CS, Lorenz S, Madoui MA et al. (2014) Performance comparison of four exome capture systems for deep sequencing. BMC Genomics 15:449 Giardine B, Riemer C, Hardison RC et al. (2005) Galaxy: a platform for interactive large-scale genome analysis. Genome Res 15:1451–5 Hayden EC (2014) Technology: The $1,000 genome. Nature 507:294–5 Hosen MJ, Van Nieuwerburgh F, Steyaert W et al. (2015) Efficiency of exome sequencing for the molecular diagnosis of pseudoxanthoma elasticum. J Invest Dermatol 135:992–8 Laszlo AH, Derrington IM, Ross BC et al. (2014) Decoding long nanopore sequencing reads of natural DNA. Nat Biotechnol 32:829–33 Li H, Durbin R (2009) Fast and accurate short read alignment with Burrows-Wheeler transform. Bioinformatics 25:1754–60 Li Q, Schurgers LJ, Smith AC et al. (2009) Co-existent pseudoxanthoma elasticum and vitamin K-dependent coagulation factor deficiency: compound heterozygosity for mutations in the GGCX gene. Am J Pathol 174:534–40 Li Q, Uitto J (2013) Mineralization/anti-mineralization networks in the skin and vascular connective tissues. Am J Pathol 183:10–8 Metzker ML (2010) Sequencing technologies—the next generation. Nat Rev Genet 11:31–46
940
See related article on pg 1033
The Double Life of Connexin Channels: Single Is a Treat Roberto Bruzzone1,2 Although several genetic diseases are caused by mutations in channels made by connexin family members, there has been little progress in the development and validation of therapeutic options. An in vitro study in this issue of JID suggests that an anti-malarial drug may be beneficial in keratitis-ichthyosis deafness, a severe conexin channel disease associated with potentially fatal recurrent infections. Journal of Investigative Dermatology (2015) 135, 940–943; doi:10.1038/jid.2014.524
Ever since the report of a human genetic disease linked to connexin mutations more than 20 years ago, pharmacological modulation of this class of channels has acquired clinical relevance. Connexins are special channels, as they consist of oligomeric structures, termed connexons or hemichannels, which can either dock with cognate hemichannels to form complete intercellular (gap junction) channels that provide ionic and metabolic coupling between adjacent cells, or function as unpaired connexons, whose physiological role, if any,
remains to be firmly established (Saez et al., 2003). Of note, dysregulated hemichannel activity by mutated connexins has been suspected to have a pathogenic role in some diseases, perhaps as a consequence of the resulting loss of cell integrity (Xu and Nicholson, 2013; Kelly et al., 2014). In the current issue of JID by Levit et al. (2015, this issue) describe the effect of a molecule approved for anti-malarial treatment, on some of the most common Cx26 mutations that cause keratitis-ichthyosis deafness (KID), a
Nitschke Y, Rutsch F (2012) Generalized arterial calcification of infancy and pseudoxanthoma elasticum: two sides of the same coin. Front Genet 3:302
1
Pfendner EG, Vanakker OM, Terry SF et al. (2007) Mutation detection in the ABCC6 gene
Correspondence: Roberto Bruzzone, HKU-Pasteur Research Pole, School of Public Health, The University of Hong Kong, Hong Kong, People’s Republic of China. E-mail: [email protected]
HKU-Pasteur Research Pole, School of Public Health The University of Hong Kong, Hong Kong, People’s Republic of China and 2Department of Cell Biology and Infection, Institut Pasteur, Paris, France
Journal of Investigative Dermatology (2015), Volume 135
