Variants in genes encoding pyrophosphate metabolizing enzymes are associated with Pseudoxanthoma elasticum

Clinical Biochemistry 47 (2014) 60–67

Contents lists available at ScienceDirect

Clinical Biochemistry journal homepage: www.elsevier.com/locate/clinbiochem

Variants in genes encoding pyrophosphate metabolizing enzymes are associated with Pseudoxanthoma elasticum Mareike Dabisch-Ruthe a, Alexander Brock a, Patricia Kuzaj a, Peter Charbel Issa b, Christiane Szliska c, Cornelius Knabbe a, Doris Hendig a,⁎ a b c

Institut für Laboratoriums-und Transfusionsmedizin, Herz-und Diabeteszentrum Nordrhein-Westfalen, Universitätsklinik der Ruhr-Universität Bochum, Bad Oeynhausen, Germany Department of Ophthalmology, University of Bonn, Germany Krankenhaus Bethesda, Dermatologie, Lehrkrankenhaus der Ruhr-Universität Bochum, Freudenberg, Germany

a r t i c l e

i n f o

Article history: Received 22 April 2014 Received in revised form 30 June 2014 Accepted 3 July 2014 Available online 12 July 2014 Keywords: Pseudoxanthoma elasticum Calcification Tissue nonspecific alkaline phosphatase Ectonucleotide pyrophosphatase 1 Ankylosis

a b s t r a c t Objectives: Pseudoxanthoma elasticum (PXE) is a rare hereditary disorder characterized by progressive calcification and fragmentation of elastic fibers. Because of the great clinical variability between PXE patients the involvement of modifier genes was recently suggested. Therefore, we investigated the association of single nucleotide variants (SNVs) in selected candidate genes known to regulate cellular pyrophosphate metabolism. Design and methods: We used RLFP analyses to evaluate the distribution of SNVs in alkaline phosphatase (ALP), ectonucleotide pyrophosphatase 1 (ENPP1) and ankylosis (ANKH) in DNA samples from 190 German PXE patients and 190 age- and sex-matched healthy controls. Statistical analyses were performed using Fisher exact test and Bonferroni correction. Results: The screening revealed three different SNVs in three genes, which were associated with PXE. The SNV c.1190-65C N A (rs1780329, minor allele frequency (MAF) patients: 0.17; controls: 0.11; P = 0.04) in the ALP gene was significantly more frequent in PXE patients. Furthermore, PXE was highly associated with ANKH p.A98A genotype TT (P = 0.0012), although the MAF was not different between patients and controls. After correction for multiple testing according to the Bonferroni method, one SNV in the ENPP1 gene (c.313 + 9G N T, rs7773477) remained significantly associated with PXE with significantly higher MAF values in the patient cohort (MAF: 0.04 vs. 0.00; P = 0.0024) and a high association with PXE susceptibility (OR 27.96). Conclusion: Polymorphisms in ALP, ENPP1 and ANKH are important genetic risk factors contributing to PXE. © 2014 The Canadian Society of Clinical Chemists. Published by Elsevier Inc. All rights reserved.

Introduction Progressive calcification and fragmentation of elastic fibers, which affect skin, eyes and the cardiovascular system, are major characteristics of Pseudoxanthoma elasticum (PXE, OMIM 264800). PXE, a rare hereditary disorder, is caused by mutations in ABCC6 (ATP-binding cassette subfamily C, member 6), a gene encoding for an ABC-transporter protein also known as multidrug resistance-associated protein 6 (MRP6) [1]. The clinical course of PXE is highly variable, with age at disease onset and the number and magnitude of its symptoms differing considerably among patients [2]. It has been speculated that other

⁎ Corresponding author at: Herz- und Diabeteszentrum Nordrhein-Westfalen, Universitätsklinik der Ruhr-Universität Bochum, Institut für Laboratoriums-und Transfusionsmedizin, Georgstr. 11, 32545 Bad Oeynhausen, Germany. Fax: +49 5731 97 2013. E-mail addresses: [email protected] (M. Dabisch-Ruthe), [email protected] (A. Brock), [email protected] (P. Kuzaj), [email protected] (P. Charbel Issa), [email protected] (C. Szliska), [email protected] (C. Knabbe), [email protected] (D. Hendig).

genes, so-called modifier genes, as well as environmental factors might contribute to the expression and severity of PXE. In the last years the relevance of this hypothesis could be confirmed by identification of modifier genes for PXE [3–7]. The encoded proteins are involved in the biosynthesis of glycosaminoglycans, response to oxidative stress and regulation of biological calcification, which were discussed as playing key roles in the pathomechanisms underlying PXE. Biological calcification was previously reported to be a passive process, but recent evidence suggests a complex regulation involving multiple steps and mediators. Due to progressive calcification of elastic fibers as a hallmark of PXE and the fact that the calcification process is actively regulated, we suggest further modifier genes for PXE. Candidate genes for PXE susceptibility are tissue-nonspecific alkaline phosphatase (ALP), inorganic pyrophosphate (PPi) channel progressive ankylosis protein homolog (ANKH) and transmembrane ectonucleotide pyrophosphatase 1 (ENPP1). We have just recently shown the essential role of PPi as the central regulatory metabolites preventing matrix calcification in PXE [8]. PPi acts as an inhibitor of basic calcium phosphate crystal growth in the extracellular matrix [9]. The concentration of PPi is controlled by ENPP1, which generates AMP and PPi from ATP, by the PPi

http://dx.doi.org/10.1016/j.clinbiochem.2014.07.003 0009-9120/© 2014 The Canadian Society of Clinical Chemists. Published by Elsevier Inc. All rights reserved.

M. Dabisch-Ruthe et al. / Clinical Biochemistry 47 (2014) 60–67

channel ANKH and ALP [9–11]. ALP cleaves PPi and phosphorylated osteopontin (OPN), whereby these two calcification inhibitors become inactive and Pi is released. Pi is a component of hydroxyapatite crystal deposition and supports extracellular matrix mineralization. Investigations into ENPP1 metabolism led to a correlation between PXE and another rare disorder known as generalized arterial calcification of infancy (GACI, OMIM 208000). GACI is characterized by a progressive calcification of the internal elastic lamina, fibrotic myointimal proliferation of muscular arteries, and resultant arterial stenosis in the neonatal period [12]. It was observed that PXE and GACI can be caused by mutations in either ABCC6 or ENPP1 [13]. The two other candidate genes, ALP and ANKH, also induce calcification associated diseases, but none of them were already linked to PXE. Mutations in the ALP gene cause ankylosing spondylitis (AS), a form of chronic arthritis characterized by inflammatory response and pathological mineralization that primarily affects young adults. Hypophosphatasia, a second disease distinguished by defective bone mineralization and deficiency of serum ALP activity, is a rare autosomal recessive inborn error of metabolism. ANKH mutations are linked to autosomal-dominant chondrocalcinosis (CC), which manifests in joint pain and arthritis caused by the deposition of calcium-containing crystals within articular cartilage [14]. These candidate genes play important roles in the regulation of matrix calcification. The aim of our current study was to investigate selected single nucleotide variants (SNVs) in ALP, ANKH and ENPP1 as modifiers for PXE manifestation and disease severity, because PPi seems to be a central regulatory metabolite preventing matrix calcification in PXE.

61

PCR was used to detect the frequent 16-kb deletion c.EX23_EX29del as reported by Hu et al. [21]. To detect ABCC6 deletions/insertions, which can be missed by direct sequencing, multiplex ligationdependent probe amplification (MLPA) was applied in PXE patients with an incomplete genotype [22]. Polymerase chain reaction PCR primers were designed using the published sequence (GenBank accession no. ALP NG_008940, ANKH NG_008273, ENPP1 NG_008206). PCR was performed in 25 μl reaction volume, containing 25 pmol of each primer (Biomers, Ulm, Germany), 1 U HotStar Taq DNA polymerase (Qiagen, Hilden, Germany) in 1 × reaction buffer including 2.5 mM MgCl2 (Qiagen, Hilden, Germany), 0.25 mM of each dNTP (Solis Biodyne, Tartu, Estonia) and 5 μl DNA template. The PCR conditions were as follows: initial denaturation at 95 °C for 2 min, 35 cycles of denaturation at 95 °C for 1 min, annealing for 1 min, extension at 72 °C for 1 min and final extension at 72 °C for 10 min. The primer sequences, annealing temperatures and sizes of the PCR products are summarized in Supplement 1. Allele-specific PCR The polymorphism c.294C N T (rs17251667) was analyzed using allele-specific PCR. PCR was performed twice as described above using primers with the specific sequence for each allele. Agarose gel electrophoresis was applied to detect the amplification products.

Materials and methods Restriction fragment length polymorphism analysis Patients and controls The study was approved by the ethics commission of the Ruhr University of Bochum Faculty of Medicine, located in Bad Oeynhausen. All patients provided their written informed consent to participate in the study. EDTA-anticoagulated whole blood samples were obtained from 190 German PXE patients and from 190 age- and sex-matched healthy controls. The control cohort consists of Westphalian blood donors, who were healthy and had no symptoms suggestive for PXE. Patients were recruited between 2001 and 2008 from the multidisciplinary referral center for patients with PXE, Bethesda Hospital, Freudenberg, Germany. Clinical data were collected retrospectively by reviewing patient charts from the Departments of Dermatology, Ophthalmology and Medicine. The diagnosis of PXE in all patients was consistent with the reported consensus criteria [15]. We analyzed 56 male and 134 female patients with a mean age of 46.9 years (± 15.1 years) and an age of diagnosis of 31.8 years (± 16.4 years). Clinical characteristics were as follows: 169 (89%) patients with skin involvement,157 (83%) patients with involvement of the eyes, 79 (42%) patients with involvement of the vascular system, 58 (31%) patients with hypertension, and 29 (15%) patients with heart involvement (e.g. myocardial infarction, cardiomyopathy). Of note, these numbers do not exclude subclinical or faint disease manifestations. For instance, ocular alterations sometimes may be detected best using advanced imaging technologies [16–18] which, however, have not been used for phenotyping the cohort described herein.

The obtained DNA fragments were digested overnight with 1 U of either restriction endonuclease (Supplement 1) at the temperature recommended by the supplier and subsequently separated on a 1.5% agarose gel. All restriction endonucleases were purchased from New England Biolabs (Frankfurt, Germany). Statistical analysis Allele and haplotype frequencies were compared between cases and controls using Fisher's exact test. Correction for multiple testing was performed using the Bonferroni method. The chi-square test was used to examine whether the genotype distributions were within the Hardy–Weinberg equilibrium by comparing observed and expected genotypes. P-values of less than 0.05 were considered significant after Bonferroni correction. The association of each SNV with PXE was measured by the odds ratio (OR) and 95% CI. All tests were performed using GraphPad Prism 5. Determination of linkage disequilibrium (LD) and haplotype blocks and frequencies were performed by Haploview 3.2 [23]. Haplotype blocks were defined according to the “spine of LD” setting in Haploview software, on the basis of each end marker of a block having a D′ value of more than 0.8. Multifactor dimensionality reduction (MDR) software 3.0.2 was used for detecting gene–gene and gene–environment interactions [24]. Power calculations were performed using the program developed by Faul et al. [25].

DNA preparation Results Genomic DNA was extracted from 200 μl EDTA blood using the QIAamp blood kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions. DNA was stored at −20 °C. ABCC6 genotyping Mutational analysis of the ABCC6 gene and the common c.3421C N T mutation were performed as described previously [19,20]. A multiplex

Association of single markers with PXE Twelve single nucleotide variants (SNV) in the genes ALP, ANKH and ENPP1 were genotyped in DNA samples from 190 German PXE patients and 190 age- and sex-matched healthy controls. The SNVs were selected according to the following criteria: 1) functional relevance, 2) involvement in pathological calcification, and 3) at least 5% reported frequency.

62

M. Dabisch-Ruthe et al. / Clinical Biochemistry 47 (2014) 60–67

Comparison of the allelic frequencies of the investigated SNVs between PXE patients and healthy controls revealed three variants c.1190-65C N A (rs1780329, ALP, frequency in PXE patients vs. controls: 0.17 vs. 0.11, P = 0.010), c.313 + 9G N T (rs7773477, ENPP1, 0.04 vs. 0, P = 0.0002) and c.2101-14delT (rs1799774, ENPP1, 0.20 vs. 0.14, P = 0.042) to be significantly more frequent in the PXE group (P b 0.05 each) (Table 1). After correction for multiple testing according to the Bonferroni method, only the one SNV c.313 + 9G N T (ENPP1) remained significantly associated with PXE (Pcorrected = 0.0024) (Table 1). The presence of the disease-associated allele conferred an odds ratio (OR) of 27.96 (1.66–472.30) for allele c.313 + 9T. The c.294C N T genotype TT in the ANKH gene was associated with PXE due to differences in comparing genotype frequencies (0.11 vs. 0.02, Pcorrected = 0.0036, Table 2), although the minor allele frequency (MAF) was not different between patients and controls (0.53 vs. 0.51, P = 0.235). All genotype distributions were within the Hardy– Weinberg equilibrium (HWE) except for c.876A N G (ALP) and c.294C N T (ANKH). To calculate the power of our sample size and genetic models, we used the algorithm of Faul et al. [25]. The results indicate that our sample has a sufficient size to detect associations with a power of 75%–93% for a relative risk of ≥2.0 for the most frequent polymorphisms (minor allele frequency 25%–45%, significance α = 5%), assuming an additive or multiplicative model for PXE. Assuming a dominant or recessive model, our study had a power of 33%–64% or 19%–43%, respectively. For the less abundant sequence variants (minor allele frequency b 8%), our study had a power ranging from 29% to 47%, assuming a multiplicative, additive, or dominant model if the relative risk exceeds 2.0; assuming a recessive model for PXE, the power of our study was 5% for these rare polymorphisms.

Association of ENPP1 variant c.313 + 9G N T with ABCC6 genotype and PXE phenotype The association of ENPP1 variant c.313 + 9G N T (rs7773477) with ABCC6 genotype and PXE phenotype is presented in Table 3. Two PXE patients were found to carrying this ENPP1 variant in a homozygous state whereas eight PXE patients were heterozygous carriers. The variant was not found in healthy age-and sex-matched controls. No correlation was detected between ABCC6 genotype and PXE phenotype.

Determination of LD structure and haplotype blocks LD and haplotype blocks were evaluated in PXE patients and controls using Haploview 3.2 [23]. Haplotype analysis yielded one block of LD in each analyzed gene in the pooled population, healthy controls and PXE patients (Fig. 1). Highest pairwise D′ (0.84 and 0.80) was observed between c.978delC and c.-4G N A (ANKH), as well as between ENPP1 variant c.511A N C (rs1805101) and c.517A N C (rs1044498). Two distinct blocks of LD were detected in the PXE group in the ALP gene by haplotype analysis. The second block was not observed in the group of healthy controls and the pooled population. There was no significant association between these haplotypes. Haplotype analysis comparing all three candidate genes yielded four blocks of LD (Fig. 2). The haplotype CCG in block 2 was significantly more frequent among the control group (P = 0.019, Supplement 2), but after correction for multiple testing the haplotype did not remain significant (Pcorrected = 0.228). However, the association of haplotype AT in block 3 with PXE remained significant after correction for multiple testing (Pcorrected = 0.048).

Table 1 Allele frequencies of ALP, ANKH and ENPP1 polymorphisms in PXE patients and controls. SNVa

Ref. SNVb

Allele

PXEc

Controlsc

OR (95% Cl)

Pd

Pcorrectede

C T T C A G C A

360 (95%) 20 (5%) 338 (89%) 42 (11%) 370 (97%) 10 (3%) 313 (83%) 67 (17%)

357 (94%) 23 (6%) 347 (91%) 33 (9%) 377 (99%) 3 (1%) 342 (89%) 38 (11%)

0.93 (0.50–1.72)

0.876

1.000

1.31 (0.81–2.11)

0.331

1.000

3.42 (0.93–12.52)

0.055

0.660

1.78 (1.16–2.75)

0.010

0.120

A C G A C T A G

230 (61%) 150 (39%) 355 (93%) 25 (7%) 177 (47%) 203 (53%) 367 (97%) 13 (3%)

217 (57%) 163 (43%) 361 (95%) 19 (5%) 188 (49%) 192 (51%) 368 (97%) 12 (3%)

0.87 (0.65–1.16)

0.371

1.000

1.34 (0.72–2.48)

0.436

1.000

1.12 (0.84–1.50)

0.467

1.000

1.06 (0.48–2.36)

1.000

1.000

A C G T A C T -

330 (86%) 50 (14%) 368 (97%) 12 (3%) 333 (87%) 47 (13%) 305 (80%) 75 (20%)

331 (87%) 49 (13%) 380 (100%) 0 326 (86%) 54 (14%) 327 (86%) 53 (14%)

1.03 (0.67–1.56)

0.914

1.000

27.96 (1.66–472.30)

0.0002

0.0024

0.87 (0.57–1.32)

0.523

1.000

1.52 (1.03–2.23)

0.042

0.504

ALP c.330C N T

p.S110S

rs1780316

c.787T N C

p.Y263H

rs3200254

c.876A N G

p.P292P

rs3200255

c.1190-65C N A

Intron

rs1780329

ANKH c.-978delC

Promotor

Reference [32]

c.-4G N A

5′-UTR

Reference [32]

c.294C N T

p.A98A

rs17251667

c.963A N G

p.A321A

rs2288474

ENPP1 c.511A N C

p.K171Q

rs1805101

c.313 + 9G N T

Intron

rs7773477

c.517A N C

p.K121Q

rs1044498

c.2101-14delT

Intron

rs1799774

a Nucleotide numbering refers to the genomic DNA sequence (GenBank accession no. ALP NG_008940, ANKH NG_008273, ENPP1 NG_008206), with the first nucleotide of exon 1 as the transcription initiation start site referred to as nucleotide +1. b Reference SNV in National Center for Biotechnology Information database, June 2013. c Allele counts (allele frequencies). d Allelic frequencies were compared between PXE patients and controls by Fisher 2-sided exact test. e P values b0.05 were considered significant after correction for multiple testing according to the Bonferroni method, Pcorrected = p(α) × k, k = 12.

M. Dabisch-Ruthe et al. / Clinical Biochemistry 47 (2014) 60–67

63

Table 2 Genotype frequencies of ALP, ANKH and ENPP1 polymorphisms in PXE patients and controls. SNVa

Ref. SNVb

Genotype

PXEc

Controlsc

Pd

Pcorrectede

CC CT TT TT TC CC AA AG GG CC CA AA

171 (90%) 18 (9%) 1 (1%) 151 (79%) 36 (19%) 3 (2%) 182 (96%) 6 (3%) 2 (1%) 131 (69%) 51 (27%) 8 (4%)

167 (88%) 23 (22%) 0 (0%) 158 (83%) 31 (16%) 1 (1%) 188 (98%) 1 (1%) 1 (1%) 156 (82%) 30 (16%) 4 (2%)

0.526

1.000

0.465

1.000

0.133

1.000

0.036

0.432

AA AC CC GG AG AA CC CT TT AA AG GG

61 (33%) 108 (57%) 21 (11%) 165 (87%) 25 (13%) 0 (0%) 8 (4%) 161 (85%) 21 (11%) 178 (94%) 11 (5%) 1 (1%)

53 (29%) 111 (58%) 26 (14%) 171 (90%) 19 (10%) 0 (0%) 2 (1%) 184 (97%) 4 (2%) 179 (94%) 10 (5%) 1 (1%)

0.567

1.000

0.332

1.000

0.0003

0.0036

0.985

1.000

AA AC CC GG GT TT AA AC CC TT T–

142 (74%) 46 (25%) 2 (1%) 180 (95%) 8 (4%) 2 (1%) 145 (76%) 43 (23%) 2 (1%) 121 (64%) 63 (33%) 6 (3%)

144 (76%) 43 (22%) 3 (2%) 188 (99%) 2 (1%) 0 (0%) 139 (73%) 48 (25%) 3 (2%) 143 (75%) 41 (22%) 6 (3%)

0.577

1.000

0.036

0.432

0.785

1.000

0.039

0.468

ALP c.330C N T

p.S110S

rs1780316

c.787T N C

p.Y263H

rs3200254

c.876A N G

p.P292P

rs3200255

c.1190-65C N A

Intron

rs1780329

ANKH c.-978delC

Promotor

Reference [32]

c.-4G N A

5′-UTR

Reference [32]

c.294C N T

p.A98A

rs17251667

c.963A N G

p.A321A

rs2288474

ENPP1 c.511A N C

p.K171Q

rs1805101

c.313 + 9G N T

Intron

rs7773477

c.517A N C

p.K121Q

rs1044498

c.2101-14delT

Intron

rs1799774

a Nucleotide numbering refers to the genomic DNA sequence (GenBank accession no. ALP NG_008940, ANKH NG_008273, ENPP1 NG_008206), with the first nucleotide of exon 1 as the transcription initiation start site referred to as nucleotide +1. b Reference SNV in National Center for Biotechnology Information database, June 2013. c Genotype counts (genotype frequencies). d Genotype frequencies were compared between PXE patients and controls by chi-square test. e P values b0.05 were considered significant after correction for multiple testing according to the Bonferroni method, Pcorrected = p(α) × k, k = 12.

Association of SNVs with clinical features

Detecting gene–gene and gene–environment interactions

Allelic frequencies of the investigated variants in ALP, ANKH and ENPP1 genes were analyzed in subgroups of the PXE patient group to evaluate an association with the clinical features summarized in part 2.1. We found no significant association between genotype/allelic frequencies and age, age at PXE diagnosis, and number or kind of organs involved.

Gene–gene and gene–environment interactions were detected with multifactor dimensionality reduction (MDR) software for reducing the dimensionality of multilocus genotype information to improve the identification of polymorphism combinations associated with disease risk [24]. Fig. 3 shows a tree diagram of the interaction system for the genotype (12 SNVs) and phenotype (cardiovascular tissue involvement)

Table 3 Association of ENPP1 variant c.313 + 9T with ABCC6 genotype and PXE phenotype. PXE

Sex

Genotype

ABCC6 genotype allele 1

ABCC6 genotype allele 2

Status

Organ involvement

143 19 34 66 204 252 341 140 467 294

f f f m f m m m f f

TT TT GT GT GT GT GT GT GT GT

E24: c.3421C N T (p.R1141X) E24: c.3421C N T (p.R1141X) E16: c.1995delG i-26: c.3736-1G N A E24: c.3421C N T (p.R1141X) E24: c.3421C N T (p.R1141X) E12: c. 1552C N T (p.R518X) E24: c.3421C N T (p.R1141X) E24: c.3421C N T (p.R1141X) E24: c.3490C N T (p.R1164X)

None E24: c.3421C N T (p.R1141X) i-21: c.2787 + 1G N T E30: c.4209C N A (p.S1403R) E24: c.3412C N T (p.R1138W) E12: c.1574_1575insG None None E24: c.3421C N T (p.R1141X) None

ht hm cht cht cht cht ht ht hm ht

Skin, eyes, heart, cardiovascular tissue, hypertension, gastrointestinal tract Skin, eyes, hypertension Skin Skin, eyes, cardiovascular tissue Skin, eyes, cardiovascular tissue Eyes, cardiovascular tissue, hypertension Skin, eyes, heart, hypertension Skin, eyes, cardiovascular tissue, hypertension, gastrointestinal tract Skin, eyes, cardiovascular tissue, hypertension Skin, eyes

f: female, m: male, ht: heterozygote, hm: homozygote, cht: compound heterozygote, ht: heterozygote, wt: wild type.

64

M. Dabisch-Ruthe et al. / Clinical Biochemistry 47 (2014) 60–67

Fig. 1. Linkage disequilibrium structure of polymorphisms within the ALP, ANKH and ENPP1 genes. LD structure of four polymorphisms within the ALP gene in the pooled population (A), PXE patients (B), and healthy controls (C). LD structure of four polymorphisms within the ANKH (D) and the ENPP1 (E) genes in the pooled population. LD patterns and haplotype blocks in each group were defined according to the “spine of LD” setting in Haploview 3.2 software, which is on the basis of each end marker of a block having a D′ value of more than 0.8 (23). A standard color scheme is used to display LD pattern, with black for perfect LD (LOD ≥ 2, D′ = 1), white for no LD (LOD b 2, D′ b 1), shades of gray (LOD ≥ 2, D′ ≤ 1) and red for intermediate LD (LOD b 2, D′ = 1).

associations. Both blue and green lines represent a synergetic interaction, while the blue interaction is stronger than the green one. Red and orange describe an antagonistic interaction effect, where the red interaction is stronger than the orange one. From the color in the diagram, we know that the interaction of the polymorphism c.294C N T (ANKH) and cardiovascular tissue involvement shows a strong synergistic effect. Weaker synergistic effects were detected between c.1190-65C N A (ALP), c.2101-14delT (ENPP1), c.294C N T (ANKH) and the cardiovascular tissue involvement, between c.-978delC, c.-4G N A (both ANKH) and c.313 + 9G N T (ENPP1), as well as between c.787T N C and c.876A N G (both ALP).

Discussion Characteristic for PXE is a great variability in the clinical course and phenotype of PXE, even among patients with the same or functionally similar ABCC6 mutations. A number of genes (e.g. osteopontin, vascular endothelial growth factor gene) are discussed as contributing to PXE manifestation and disease severity. Due to progressive calcification of elastic fibers as the main characteristic of PXE and evidence that the calcification process is actively regulated, we suggest further modifier genes, which are involved in the regulation of biological calcification. Our results show a significant association of the SNV c.313 + 9G N T

(intron) in ENPP1 and c.294C N T (p.A98A) in ANKH, and also one haplotype AT being associated with PXE. ALP is an important promoter of calcification because it catalyzes the hydrolysis of PPi, thereby decreasing the concentration of this calcification inhibitor and concomitantly increasing the levels of Pi. The SNV c.1190-65C N A in the ALP gene is localized in intron 10. The functional effect of the C to A transition at this position is unknown. But we found a haplotype between the variants c.876A N G (p.P292P) and c.1190-65C N A (intron) in the PXE patients. We suggest that these SNVs, alone or in combination, might modify ALP activity. Higher ALP activities were previously reported in PXE fibroblasts [26], which catalyzed dephosphorylation of pyrophosphates and phosphorylated proteins such as osteopontin. A previous family-based association study by Tsui et al. in a Canadian population documented that an ALP haplotype marker consisting of the variants (c.62-1924C/c.787C/ c.1190-65A) in man is significantly associated with AS in multiple families affected [27]. Previous studies showed that deficiency of ENPP1 or ANKH results in pathological deposition of hydroxyapatite crystals [28,29]. Deficiency in ENPP1 is associated with vascular calcification in GACI and results in a decrease of extracellular PPi, whereas decreased ANKH potentiates endochondral vascular calcification [10,30,31]. We found two SNVs in ENPP1: c.313 + 9G N T and c.2101-14delT, which were significantly more frequent in the PXE patient group. After correction for multiple

M. Dabisch-Ruthe et al. / Clinical Biochemistry 47 (2014) 60–67

65

Fig. 2. Linkage disequilibrium structure of 12 polymorphisms within the ALP, ANKH and ENPP1 genes in one analysis in the pooled population. LD patterns and haplotype blocks in each group were defined according to the “spine of LD” setting in Haploview 3.2 software, which is on the basis of each end marker of a block having a D′ value of more than 0.8 (23). A standard color scheme is used to display LD pattern, with black for perfect LD (LOD ≥ 2, D′ = 1), white for no LD (LOD b 2, D′ b 1), shades of gray (LOD ≥ 2, D′ ≤ 1) and red for intermediate LD (LOD b 2, D′ = 1).

testing according to the Bonferroni method, only the SNV c.313 + 9G N T remained significantly associated with PXE (Pcorrected =0.0024). The variants c.2101-14delT and c.313 + 9G N T are localized at the intron/exon

boundaries and are speculated to lead to an impairment of mRNA splicing, which results in a faulty processing of mRNA. The c.313 + 9T allele was only found in PXE patients and not in the age- and sex-matched

Fig. 3. A tree diagram of the interactions among 12 polymorphisms and cardiovascular tissue involvement for PXE using MDR. Blue and green represent a synergetic interaction, where the blue intensity is greater than the green intensity. Red and orange describe an antagonistic interaction effect, where the red intensity is stronger than the orange intensity.

66

M. Dabisch-Ruthe et al. / Clinical Biochemistry 47 (2014) 60–67

healthy controls (OR 27.96). This highly significant association suggests single nucleotide alteration c.313 + 9G N T as a strong risk factor for PXE. Another study discovered that PXE and GACI can be caused by mutations in either ABCC6 or ENPP1 [13], which supported our finding that sequence variants in ENPP1 might be an important genetic cofactor for PXE. We did not detect any association between ABCC6 genotype and the ENPP1 variant c.313 + 9C N T. Homozygous as well as heterozygous carriers of the ENPP1 variant were also found to carry ABCC6 mutations in a compound-heterozygous, heterozygous or homozygous state. Therefore, we suggest that the ENPP1 variant c.313 + 9C N T is not a disease causing mutation, but rather functions as a genetic co-factor for PXE susceptibility. However, we did not identify a correlation to PXE severity (e.g. organ involvement, age of disease onset). In-depth clinical phenotyping and follow-up studies of these PXE patients would be highly interesting, in order to discover differences in disease expression of patients carrying the ENPP1 variant c.313 + 9C N T. The genotype TT of the ANKH SNV c.294C N T (p.A98A) was found to be significantly associated with the PXE group, although an association could not be observed when comparing allelic frequencies. The exchange from C to T on the third nucleotide position did not change the type of amino acid. Therefore we did not suggest a potential functional relevance. Sequence variations in the ANKH gene are found to be associated with CC and up-regulation of ANKH expression, consequently resulting in a decrease of intracellular PPi and elevation of extracellular PPi concentrations to prevent matrix mineralization [32,33]. Deviations from the HWE were found for the two sequence variations c.876A N G (ALP) and c.294C N T (ANKH). All other genotype distributions were within the HWE. Discrepancies from HWE can be due to inbreeding, population stratification or selection or disease association. If the population prevalence of the trait is low and in the presence of an association, cases are expected to deviate from HWE and also do not need to be in the HWE [34,35]. The c.294C N T genotype TT in the ANKH gene was associated with PXE and the G allele of the variant c.876A N G in ALP gene was roughly associated with PXE. Haploview analyses identified one block (block 3) consisting of the variants c.963A N G (ANKH) and c.313 + 9G N T (ENPP1), which revealed one haplotype AT to be significantly associated with PXE. This combination could have an additional function for the occurrence of symptoms in PXE. MDR has been applied to identify gene–gene interactions conferring susceptibility to common diseases, including hypertension [36], bladder cancer [37], Type 2 diabetes [38] and rheumatoid arthritis [39]. The main advantage of MDR is its ability to simultaneously detect and characterize the impact of the combined effects of multiple disease factors [40]. In our study MDR analyses revealed strong interactions between c.294C N A (ANKH), c.2101-14delT (ENPP1), c.1190-65C N A (ALP) and cardiovascular tissue involvement. The three variants were also found to be significantly associated with PXE. Variants in ANKH, ALP and ENPP1 are known to be involved in cardiovascular diseases [10,30,31]. There are some limitations to our study. Since PXE is a rare disease, cohort size is relatively small and, therefore, the statistical power is limited. We have reduced false-positive results due to multiple testing by using strict Bonferroni correction criteria. This conservative correction assumes independence among markers and tends to underestimate association, so that the associations of single SNV markers and haplotypes we have found in our study are unlikely to be alone by chance. The power of the present study was adequate to detect an association of the investigated SNV and susceptibility to PXE reliably. It cannot be totally excluded that relationships of smaller magnitude may have been missed in our analysis. However, for the less frequent variants with a minor allele frequency below 8%, the power did not exceed 80%, indicating a possibility that the results might be falsely negative. In conclusion, we identified further variants that are likely to contribute to PXE. These findings pave the way for further studies

investigating the effects of altered ALP, ANKH and ENPP1 expression in PXE pathogenesis. Abbreviations ABCC6 ATP-binding cassette subfamily C, member 6 ALP non-tissue specific alkaline phosphate ANKH progressive ankylosis protein homolog AS ankylosing spondylitis CC chondrocalcinosis CI confidence interval ENPP1 ectonucleotide pyrophospatase 1 GACI generalized arterial calcification of infancy HWE Hardy–Weinberg equilibrium MDR multifactor dimensionality reduction LD linkage disequilibrium OR odds ratio PCR polymerase chain reaction inorganic pyrophosphate PPi PXE Pseudoxanthoma elasticum SNV single nucleotide variant

Competing interests The authors declare that they have no competing interests. Authors' contributions M.D.-R. conducted the experiments, analyzed the data and wrote the paper, A.B. performed some experiments, P.K. and P.C.I. discussed and revised the manuscript critically, P.C.I. and C.S. collected the patients' samples and clinical data, and C.K. and H.D. initiated the study, planned the experimental design and contributed to write the paper. Acknowledgments We are very grateful to all the PXE patients and their relatives, whose cooperation made our studies possible. Furthermore, we would like to thank Peter Hof, chairman of the Selbsthilfe für PXE Erkrankte Deutschlands e.V., the members of the clinical ambulance for PXE at the Bethesda hospital in Freudenberg, Germany and the University Eye Hospital in Bonn, Germany. This work was funded by the German Research Foundation (DFG He 5900/2-1). We thank Christoph Lichtenberg for his excellent technical assistance, Melanie Weinstock for helpful discussions and Sarah L. Kirkby for her linguistic advice. Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.clinbiochem.2014.07.003. References [1] Bergen AAB, Plomp AS, Schuurman EJ, Terry S, Breuning M, Dauwerse H, et al. Mutations in ABCC6 cause pseudoxanthoma elasticum. Nat Genet 2000;25(2):228–31. [2] Plomp AS, Toonstra J, Bergen AAB, van Dijk MR, de Jong PTVM. Proposal for updating the pseudoxanthoma elasticum classification system and a review of the clinical findings. Am J Med Genet A 2010;152(4):1049–58. [3] Hendig D, Arndt M, Szliska C, Kleesiek K, Götting C. SPP1 promoter polymorphisms: identification of the first modifier gene for pseudoxanthoma elasticum. Clin Chem 2007;53(5):829–36. [4] Zarbock R, Hendig D, Szliska C, Kleesiek K, Götting C. Pseudoxanthoma elasticum: genetic variations in antioxidant genes are risk factors for early disease onset. Clin Chem 2007;53(10):1734–40. [5] Zarbock R, Hendig D, Szliska C, Kleesiek K, Götting C. Analysis of MMP2 promoter polymorphisms in patients with pseudoxanthoma elasticum. Clin Chim Acta 2010;411(19):1487–90. [6] Schön S, Schulz V, Prante C, Hendig D, Szliska C, Kuhn J, et al. Polymorphisms in the xylosyltransferase genes cause higher serum XT-I activity in patients with

M. Dabisch-Ruthe et al. / Clinical Biochemistry 47 (2014) 60–67

[7]

[8]

[9] [10]

[11] [12] [13]

[14] [15]

[16] [17] [18]

[19]

[20]

[21]

[22]

[23]

pseudoxanthoma elasticum (PXE) and are involved in a severe disease course. J Med Genet 2006;43(9):745–9. Zarbock R, Hendig D, Szliska C, Kleesiek K, Götting C. Vascular endothelial growth factor gene polymorphisms as prognostic markers for ocular manifestations in pseudoxanthoma elasticum. Hum Mol Genet 2009;18(17):3344–51. Dabisch-Ruthe M, Kuzaj P, Götting C, Knabbe C, Hendig D. Pyrophosphates as a major inhibitor of matrix calcification in Pseudoxanthoma elasticum. J Dermatol Sci 2014;75(2):109–20. Terkeltaub RA. Inorganic pyrophosphate generation and disposition in pathophysiology. Am J Physiol Cell Physiol 2001;281(1):C1–C11. Harmey D, Hessle L, Narisawa S, Johnson KA, Terkeltaub R, Millán JL. Concerted regulation of inorganic pyrophosphate and osteopontin by Akp2, Enpp1, and Ank: an integrated model of the pathogenesis of mineralization disorders. Am J Pathol 2004;164(4):1199–209. Rutsch F, Nitschke Y, Terkeltaub R. Genetics in arterial calcification pieces of a puzzle and cogs in a wheel. Circ Res 2011;109(5):578–92. Iravathy Goud Kalal DS, Panda A, Nitschke Y, Rutsch F. Molecular diagnosis of generalized arterial calcification of infancy (GACI). J Cardiovasc Dis Res 2012;3(2):150. Nitschke Y, Baujat G, Botschen U, Wittkampf T, du Moulin M, Stella J, et al. Generalized arterial calcification of infancy and pseudoxanthoma elasticum can be caused by mutations in either ENPP1 or ABCC6. Am J Hum Genet 2012;90(1):25. Pendleton A, Johnson MD, Hughes A, Gurley KA, Ho AM, Doherty M, et al. Mutations in ANKH cause chondrocalcinosis. Am J Hum Genet 2002;71(4):933–40. Lebwohl M, Neldner K, Pope FM, De Paepe A, Christiano AM, Boyd CD, et al. Classification of pseudoxanthoma elasticum: report of a consensus conference. J Am Acad Dermatol 1994;30(1):103. Charbel Issa P, Finger RP, Götting C, Hendig D, Holz FG, Scholl HP. Centrifugal fundus abnormalities in pseudoxanthoma elasticum. Ophthalmology 2010;117(7):1406–14. Gliem M, De Zaeytijd J, Finger RP, Holz FG, Leroy BP, Charbel Issa P. An update on the ocular phenotype in patients with pseudoxanthoma elasticum. Front Genet 2013;4. Charbel Issa P, Finger RP, Holz FG, Scholl HP. Multimodal imaging including spectral domain OCT and confocal near infrared reflectance for characterization of outer retinal pathology in pseudoxanthoma elasticum. Invest Ophthalmol Vis Sci 2009;50(12):5913–8. Schulz V, Hendig D, Henjakovic M, Szliska C, Kleesiek K, Götting C. Mutational analysis of the ABCC6 gene and the proximal ABCC6 gene promoter in German patients with pseudoxanthoma elasticum (PXE). Hum Mutat 2006;27(8):831. Götting C, Schulz V, Hendig D, Grundt A, Dreier J, Szliska C, et al. Assessment of a rapidcycle PCR assay for the identification of the recurrent c. 3421C N T mutation in the ABCC6 gene in pseudoxanthoma elasticum patients. Lab Invest 2003;84(1):122–30. Hu X, Plomp A, Gorgels T, Ten Brink J, Loves W, Mannens M, et al. Efficient molecular diagnostic strategy for ABCC6 in pseudoxanthoma elasticum. Genet Test 2004;8(3):292–300. Costrop LM, Vanakker OO, Van Laer L, Le Saux O, Martin L, Chassaing N, et al. Novel deletions causing pseudoxanthoma elasticum underscore the genomic instability of the ABCC6 region. J Hum Genet 2010;55(2):112–7. Barrett J, Fry B, Maller J, Daly M. Haploview: analysis and visualization of LD and haplotype maps. Bioinformatics 2005;21(2):263–5.

67

[24] Hahn LW, Ritchie MD, Moore JH. Multifactor dimensionality reduction software for detecting gene–gene and gene–environment interactions. Bioinformatics 2003;19(3):376–82. [25] Faul F, Erdfelder E, Lang A-G, Buchner A. G* Power 3: a flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behav Res Methods 2007;39(2):175–91. [26] Boraldi F, Annovi G, Vermeer C, Schurgers LJ, Trenti T, Tiozzo R, et al. Matrix Gla protein and alkaline phosphatase are differently modulated in human dermal fibroblasts from PXE patients and controls. J Investig Dermatol 2013;133(4):946–54. [27] Tsui HW, Inman RD, Reveille JD, Tsui FW. Association of a TNAP haplotype with ankylosing spondylitis. Arthritis Rheum 2007;56(1):234–43. [28] Okawa A, Nakamura I, Goto S, Moriya H, Nakamura Y, Ikegawa S. Mutation in Npps in a mouse model of ossification of the posterior longitudinal ligament of the spine. Nat Genet 1998;19(3):271–3. [29] Nitschke Y, Hartmann S, Torsello G, Horstmann R, Seifarth H, Weissen-Plenz G, et al. Expression of NPP1 is regulated during atheromatous plaque calcification. J Cell Mol Med 2011;15(2):220–31. [30] Rutsch F, Ruf N, Vaingankar S, Toliat MR, Suk A, Höhne W, et al. Mutations in ENPP1 are associated with ‘idiopathic’ infantile arterial calcification. Nat Genet 2003;34(4):379–81. [31] Johnson K, Polewski M, van Etten D, Terkeltaub R. Chondrogenesis mediated by PPi depletion promotes spontaneous aortic calcification in NPP1−/− mice. Arterioscler Thromb Vasc Biol 2005;25(4):686–91. [32] Zhang Y, Johnson K, Russell RGG, Wordsworth BP, Carr AJ, Terkeltaub RA, et al. Association of sporadic chondrocalcinosis with a -4-basepair G-to-A transition in the 5′untranslated region of ANKH that promotes enhanced expression of ANKH protein and excess generation of extracellular inorganic pyrophosphate. Arthritis Rheum 2005;52(4):1110–7. [33] Ho AM, Johnson MD, Kingsley DM. Role of the mouse ank gene in control of tissue calcification and arthritis. Science 2000;289(5477):265–70. [34] Lunetta KL. Genetic association studies. Circulation 2008;118(1):96–101. [35] Salanti G, Amountza G, Ntzani EE, Ioannidis JP. Hardy–Weinberg equilibrium in genetic association studies: an empirical evaluation of reporting, deviations, and power. Eur J Hum Genet 2005;13(7):840–8. [36] Moore JH, Williams SM. New strategies for identifying gene–gene interactions in hypertension. Ann Med 2002;34(2):88–95. [37] Andrew AS, Karagas MR, Nelson HH, Guarrera S, Polidoro S, Gamberini S, et al. DNA repair polymorphisms modify bladder cancer risk: a multi-factor analytic strategy. Hum Hered 2007;65(2):105–18. [38] Cho Y, Ritchie M, Moore J, Park J, Lee K-U, Shin H, et al. Multifactor-dimensionality reduction shows a two-locus interaction associated with Type 2 diabetes mellitus. Diabetologia 2004;47(3):549–54. [39] Becker ML, Gaedigk R, van Haandel L, Thomas B, Lasky A, Hoeltzel M, et al. The effect of genotype on methotrexate polyglutamate variability in juvenile idiopathic arthritis and association with drug response. Arthritis Rheum 2011;63(1):276–85. [40] Wu Y, Zhang L, Liu L, Zhang Y, Zhao Z, Liu X, et al. A multifactor dimensionality reduction-logistic regression model of gene polymorphisms and an environmental interaction analysis in cancer research. Asian Pac J Cancer Prev 2011;12:2887–92.