Whole exome sequencing of patients with diffuse idiopathic

ÓRgÃO OfICIAL dA SOCIEdAdE PORTUgUESA dE REUMATOLOgIA

116

ORIgInAL PAPERS

1. Serviço Especializado de Epidemiologia e Biologia Molecular(SEEBMO) / Hospital de Santo Espírito da Ilha Terceira (HSEIT)2. Comprehensive Health Research Center, CHRC, Lisbon, Portugal3. Centre of Marine Science (CCMAR) / University of Algarve

could be involved in this phenotype in an as yet un-known way.

Keywords: Rheumatic and musculoskeletal diseases ;Genetic association; Rheumatology.

INTRODUCTION

Previous studies undertaken by our group, identifiedand characterized twelve families affected with DiffuseIdiopathic Skeletal Hyperostosis (DISH, MIM 106400)and/or Calcium Pyrophosphate Dihydrate (CPPD)Chondrocalcinosis (CC), hereafter designated,DISH/CC. DISH/CC is a poorly understood phenotypecharacterised by peripheral and axial enthesopathic cal-cifications, frequently fulfilling the radiological criteriafor DISH, and in some cases associated with CPPDChondrocalcinosis. A common pathogenic mechanism,shared by the two conditions, has been suggested1.DISH is a common skeletal disorder characterized byprogressive calcification and ossification of ligamentsand entheses2-3. The exact prevalence and incidence ofDISH is unknown, however it is well known that it ismore frequent in males and its prevalence rapidly in-creases with age, mainly affecting subjects over the ageof 404. The prevalence of DISH in patients over 50 yearsof age is 25% in males and 15% in females,5 and thisdisease is now becoming a serious problem in aging so-cieties. Several lines of evidence suggest that geneticfactors might play a part in the etiology of DISH, suchas the existence of familial cases with early onset (in thethird decade of life)6 and the higher frequency of DISHin the boxer dogs relative to other dog breeds7-8. So far,however, no single gene has been conclusively associa -ted with the disease. Chondrocalcinosis is characteri -zed by deposition of crystals of calcium pyrophosphatedihydrate (CPPD) in articular hyaline and fibro-carti-

Whole exome sequencing of patients with diffuse idiopathic skeletal hyperostosis and calcium

pyrophosphate crystal chondrocalcinosis

Parreira B1,2, Couto AR1,2, Rocha F1,2, Sousa M1,2, Faustino V1, Power DM3, Bruges-Armas J1,2

ACTA REUMATOL PORT. 2020;45:116-126

ABSTRACT

Objectives: DISH/CC is a poorly understood pheno-type characterised by peripheral and axial entheso-pathic calcifications, frequently fulfilling the radiolog-ical criteria for Diffuse Idiopathic Skeletal Hyperostosis(DISH, MIM 106400), and in some cases associatedwith Calcium Pyrophosphate Dihydrate (CPPD) Chon-drocalcinosis (CC). The concurrence of DISH and CCsuggests a shared pathogenic mechanism. In order toidentify genetic variants for susceptibility we performedwhole exome sequencing in four patients showing thisphenotype.Materials and methods: Exome data were filtered inorder to find a variant or a group of variants that couldbe associated with the DISH/CC phenotype. Variantsof interest were subsequently confirmed by Sanger se-quencing. Selected variants were screened in a cohortof 65 DISH/CC patients vs 118 controls from Azores.The statistical analysis was performed using PLINKV1.07.Results:We identified 21 genetic variants in 17 genesthat were directly or indirectly related to mineraliza-tion, several are predicted to have a strong effect at aprotein level. Phylogenetic analysis of altered aminoacids indicates that these are either highly conservedin vertebrates or conserved in mammals. In case-con-trol analyses, variant rs34473884 in PPP2R2D was sig-nificantly associated with the DISH/CC phenotype(p=0.028; OR=1.789, 95% CI= 1.060 - 3.021)).Conclusion: The results of the present and precedingstudies with the DISH/CC families suggests that thephenotype has a polygenic basis. The PPP2R2D gene


117

PARREIRA B ET AL

lage9. For the moment two genes, ANKH (CCAL2) andTNFRSF11B, are known to be involved in the devel-opment of Chondrocalcinosis10-11. Previously, we per-formed genetic studies of DISH/CC using a “Wholegenome wide linkage study” and an “Identity byState/Identity by Descent” association study and twogenes were identified as good candidate genes forDISH/CC; RSPO4 on chromosome 20, and LEMD3 onchromosome 12. Several RSPO4 gene variants wereidentified and nucleotide modifications located in theregulatory region were more frequent in control indi-viduals than in the DISH/CC group12. Even though thegenetic basis of DISH/CC is unknown, it is consideredto be a bone forming disease and genes related to cal-cification and ossification are considered good candi-dates. In the present study, we took advantage of wholeexome sequencing (WES) so that even in the absenceof sufficient pedigrees and samples for traditional link-age approaches we could identify rare protein-codingvariants that are the cause of the majority of mono-genic diseases13-14. We performed targeted exome se-quencing on four DISH/CC patients, with an appar-ently autosomal dominant DISH/CC phenotype, inorder to capture rare and pathogenic variants expect-ed to have potentially damaging effects on proteinfunction that lead to modified calcification and/or os-sification. To our knowledge this is the first report ofWES analysis in patients affected with DISH.

MATERIAlS AND METHODS

This study was approved by the “Comissão de Éticado Hospital de Santo Espírito da Ilha Terceira”. Allmethods were performed in accordance with the ap-proved guidelines, obtaining informed consent fromeach subject before conducting the experiments.

ExOME CApTURE

The selection of patients was made after ruling out mu-tations in ANKH and secondary causes for Chondro-calcinosis. DNA was extracted from peripheral bloodsamples using a standard salting out procedure. Sam-ples were resequenced, using an ABI-SOLiD platformand Agilent’s SureSelect Target Enrichment System for38 Mb, by “Sistemas Genómicos, S.L.” in Valencia,Spain. After standard DNA quality control, SOLiDFragment libraries were prepared and enriched withSureSelect All Human Enrichment Target Exon. Thequality and quantity of the libraries were assessed by

analysis with an Agilent 2100 Bioanalyzer and Qubit.Each library went through a process of emulsion PCRfor clonal amplification of the fragments, followed byan enrichment process and chemical modification toallow loading into the reaction chamber. The qualityand quantity of the beads obtained for each librarywere estimated taking into account the parameters giv-en by Work Flow Analysis. Then, ligation sequencingwas done to obtain sequences of 50nt +35nt (Paired-end) reads using a SOLiD4 sequencer. The data qual-ity was estimated using the parameters provided bythe software SETS (SOLiD Experimental Tracking Sys-tem). Single Nucleotide Variants (SNVs) were classifiedusing Ensembl’s nomenclature and grouped using thefollowing scheme: Known and Novel (Coding, Splic-ing, Others). The coding variants were divided intononsynonymous and synonymous. The “Others” in-cluded intronic, UTR, regulatory region, intergenic,downstream and upstream variants.

WES fIlTERINg

The segregation analysis of these families indicated thatthe most likely model for this disease is an autosomaldominant model1. However, given that a previouslinkage study did not identify a major dominant lo-cus, the data were analyzed assuming both a recessiveand a dominant model.Genetic variants were identified in 815, 917, 872

and 593 genes under a dominant model, for indivi -duals AZ1 to AZ4. Assuming that the genetic cause isthe same for the individuals AZ1 to AZ4, identifiedgenes were then filtered to select common, sharedgenes (supplementary Table I, available in supple-mentary file). A group of 52 genes were common to thefour patients; candidate genes for testing were select-ed taking into account their involvement in calcifica-tion and/or ossification or related conditions that couldbe associated with the DISH/CC disease. Gene vari-ants shared between the genes common to the four pa-tients were identified and nonsynonymous, splice site,stop lost/gain and frameshift variants were targeted, inanticipation that synonymous and intronic variantswould be far less likely to be relevant in thepathogenicity.

CANDIDATE gENES – SElECTION By fUNCTION

A group of 20 candidate genes was selected and in-cluded genes that are reported to be involved in bonemetabolism and/or related conditions. These geneswere Alkaline Phosphatase (ALPL/TNAP), the Calcium


118

EcToPIc cALcIfIcATIon

TA

BlE

I. R

AD

IOlO

gy

RE

SU

lTS

fR

OM

fO

UR

SE

lE

CT

ED

pA

TIE

NT

S f

OR

WE

S.

RADIOLOGY

Age at

Patient

Sex

C-Spine

T-Spine

L-Spine

Knees

Elbow

onset

Other diseases

AZ1

MNor

mal

Nor

mal

Synde

smop

hytos

is D

ISH

N/A

Periarticular

<40

Obe

sity

calcifications

AZ2

FDISH

DISH

DISH

Arthrosis,

Periarticular

30Lithiasis,

enthesop

athy

calcifications

Diabe

tes mellitus

AZ3

FN/A

DISH

DISH

Enthesop

athy,

Periarticular

N/A

Obe

sity

osteop

hytos

is,

calcifications

arthrosis

AZ4

MNor

mal

Synde

smop

hytos

isSy

nde

smop

hytos

is,

Osteo

phytos

is,

Periarticular

<40

Lithiasis,

anterior

osteo

phytos

iscalcifications in

calcifications

cardiac

caps

ule, a

rthrosis

arrythmia

Definition of Rad

iological terminolog

y 1)

DISH: C

ontinuou

s os

sification

along the an

terolateral a

spect of three or

fou

r co

ntigu

ous ve

rteb

ral b

odies; 2) Sy

nde

smop

hytos

is: V

ertical a

nd

symmetrical c

alcification

of the lateral m

argins of the intervertebr

al disc sp

ace; 3) Osteo

phytos

is: p

resence of sp

urs, w

hich are outgrows of bon

e tissue; 4) Enthesop

athy:

calcification/ossification process at the site of the insertion of liga

men

ts, ten

dons, fascia or

articular caps

ule in

to bon

e; 5) Arthrosis: presence of joint sp

ace narrowing, scleros

is and

osteop

hytos

is.

Abb

reviations: N

/A: n

ot ava

ilab

le, C

-Spine: C

ervical S

pine, T-Spine: Thor

acic spine an

d L-Spine: Lumba

r sp

ine.

Sensing Receptor (CASR), Bone MorphogeneticProtein Receptor Type 1B (BMPR1B), Osteopon-tin (OPN/SPP1), Integrin Binding Sialoprotein(IBSP), Fibroblast Growth Factor 2 (FGF2), Inor-ganic Pyrophosphate Transport Regulator(ANKH), Collagen Type XI Alpha 2 (COL11A2),Nucleotide Pyrophosphatase 1 (ENPP1), Runt-related Transcription Factor (RUNX2), DickkopfWNT Signaling Pathway (DKK-1), Insulin LikeGrowth Factor 1 (IGF1), Matrix Gla Protein(MGP); Vitamin D (VDR), Bone MorphogeneticProtein 4 (BMP-4), Collagen Type 1 Alpha 1(COL1A1), Transforming Growth Factor Beta 1(TGF�1), Solute Carrier Family 29 Member 1(SLC29A1), Bone Morphogenetic Protein 2 (BMP--2) and Collagen Type VI Alpha 1 (COL6A1).

gENETIC vARIANTS – SElECTION By

pATHOgENICITy

In the selection by variant pathogenicity filter-ing strategy we used two levels; a “variant level”and a “knowledge level” (Figure 1). In the “vari-ant level” we first filtered according to varianttype and focused on functional significance(splice sites, frameshift coding, nonsynonymouscoding and loss or a gain of a stop). Only theSIFT prediction with Deleterious and Unknownand GERP values equal or higher than 3 were in-cluded. We excluded variants with MAF valueshigher than 0.05 by filtering the SNP_IDs (rsnumber or chromosome position) against theEnsemble Variant Effect Predictor tool(http://www.ensembl.org/ info/docs/tools/vep/index.html). Then in the selection by “knowl-edge” we focused only on genes associated withcalcification and/or ossification or related con-ditions. Lastly, we evaluated the functional sig-nificance and pathogenic potential of each vari-ant found in the selected genes.

EvAlUATION Of gENES AND vAlIDATION Of

vARIANTS

Information about candidate genes was obtainedfrom several databases, including Ensembl(http://www.ensembl.org/index), National Centerfor Biotechnology Information (NCBI) (http://www.ncbi.nih.gov), Pubmed (https:// www.ncbi.nlm.nih.gov/pubmedwe), GeneCards (http://www.genecards.org/), Online Mendelian Inheri-tance in Man (OMIM) (http://www. omim.org/)


119

PARREIRA B ET AL

and MalaCards (http://www.malacards.org/).PCR primers were designed using the software

Primer3 (http://www.bioinformatics.nl/cgi-bin/primer3plus/primer3plus.cgi) to amplify and validatevariants detected by exome sequencing. Sanger se-quencing using standard protocols was performed us-ing an automated DNA sequencer ABI 3130xl (AppliedBiosystemsÒ). Genetic variants were screened using Se-quencing Analysis and SeqScape (Applied Biosystem-sÒ) and using a reference NCBI sequence.The functional significance and the potential deleteri-

ous effect of each variant was explored using the follow-ing databases: Ensembl (http://www.ensembl. org/index),Human Gene Mutation Database (HGMD) (http://www.hgmd.cf.ac.uk/ac/index.php) and dbSNP (https://www.ncbi.nlm.nih.gov/projects/SNP/), ma king use ofPolyPhen-2 (Polymorphism Phenotyping v2) (http://ge-netics.bwh.harvard.edu/pph2/), SIFT and GERP.PolyPhen is classified as, benign [0-0.2], possibly dam-aging [0.2-0.85], and probably damaging [0.85-1], SIFTas deleterious if less than 0.05 and GERP, ranges from -12.3 to 6.17, with 6.17 being the most conserved15. The MAF values of each variant were also analyzed.

Protein conservation analysis was performed usingClustalW (http://www.genome.jp/tools/clustalw/) tocompare homologous amino acid sequences amongmultiple vertebrates at the sites where the variations oc-cur (accession numbers of transcripts available in sup-plementary Table II, available in supplementary file).

fIgURE 1. The two-level filter approach used to analyze the WES results from 4 patients with DISH/CC disease. SIFT: Sorting Intolerant From Tolerant, GERP: Genomic Evolutionary Rate Profiling score and MAF: Minor Allele Frequency

ASSOCIATION STUDIES AND STATISTICAl

ANAlySIS

In order to verify a possible association between iden-tified genes from WES and the DISH/CC phenotype,selected variants were screened in family members (af-fected with DISH/CC and unaffected), and the mostconserved variants and those that were present in threeor four WES patients, were selected for screening in agroup of 65 DISH/CC patients and a group of 118 con-trols. Association testing was performed by chi-squaredanalysis using PLINK software (V1.07)16: p-values<0.05 were considered significant. ORs with a 95% CIwere calculated for the minor alleles of each variant.OR >1 indicates a susceptibility allele and OR<1 indi-cates a protective allele for the disease.

RESUlTS

The WES was performed using DNA from four patientsfrom unrelated DISH/CC families (AZ1-AZ4) fromAzores. A detailed description of these families is pro-vided in our previous study1. The patients who gaveinformed consent, a blood sample was obtained andstandard X-rays were taken from: the knees, axial skele-ton, wrists, hands, elbows, and pelvis. The 4 patientswere selected because they shared a very evident DISHphenotype. The radiological characterization of pa-tients selected for WES are indicated in Table I.


120


TA

BlE

II. l

IST

Of

fIl

TE

RE

D C

AN

DID

AT

E g

EN

ET

IC v

AR

IAN

TS

CO

Nf

IRM

ED

By

SA

Ng

ER

SE

qU

EN

CIN

g.

Mutational

Patient

Gene

Chr

SNP_ID

spectrum

Variant

AA

MAF

SIFT

PolyPhen

AZ1

AZ2

AZ3

AZ4

PPP2R

2D10

rs34

4738

84HC

c.10

72G>A

G35

8S0.18

DB

hm

ht

ht

ht

COL11

A2

6rs92

7793

4RR

c.82

6G>A

E27

6K0.32

TPs D

hm

ht

hm

ht

VDR

12rs22

2857

0CM

c.2T

>CM1T

0.33

DB

hm

hm

ht

ht

CASR

3rs18

0172

6RR

c.30

31G>C

E10

21Q

0.08

TB

hm

hm

hm

FGF2

4rs10

4820

1NA

c.*7

57C>T

R20

9Q0.23

NA

NA

ht

ht

ht

BMP2

20rs23

5768

HC

c.57

0A>T

R19

0S0.24

DPb D

ht

ht

ht

ENPP1

6rs10

4449

8RR

c.51

7A>C

K17

3Q0.34

TB

ht

ht

ht

BMP4

14rs17

563

CM

c.45

5T>C

V15

2A0.37

TB

hm

hm

ht

MGP

12rs42

36RR

c.30

4A>G

T10

2A0.39

TB

ht

ht

ht

PLCG2

16rs13

8158

454

RR

c.20

54+7

G>A

NA

0.01

NA

NA

ht

ht

ALPL

1rs32

0025

4RR

c.78

7T>C

Y26

3H0.27

TB

ht

ht

CASR

3rs18

0172

5HC

c.29

56G>T

A98

6S0.09

TB

ht

ht

COL1A

117

rs37

2029

024

CM

c.32

47G>T

A10

83T

<0.01*

TB

hm

PLCG2

16rs37

4430

619

RR

c.37

86G>C

K12

62N

<0.01*

DB

ht

ALPL

1rs14

9344

982

RR

c.45

5G>A

R15

2H0.01

TB

ht

ABCC6

16rs41

2781

74CM

c.31

90C>T

R10

64W

0.01

DPb D

ht

COL11

A2

6rs22

2979

2RR

c.51

65C>T

P17

22L

0.01

DB

ht

TGFβ1

19rs55

6590

02RR

c.71

3-8d

elC

NA

0.01

NA

NA

ht

AMER3

2rs72

8549

96RR

c.13

00C>G

L43

4V0.05

TB

ht

FLNC

7rs22

9156

9HC

c.47

00G>A

R15

67Q

0.06

DPb D

ht

COL6A

121

rs10

5331

2RR

c.25

49G>A

R85

0H0.27

TB

hm

The va

rian

ts are listed

in the table acco

rding to occurren

ce in

two or

more of the pa

tien

ts A

Z1 to A

Z4 in the study.

Abb

reviations: C

hr: chromos

ome, H

C: h

ighly con

served

, CM: c

onserved

in m

ammals, RR: relax

ed reg

ion, N

A: N

ot A

pplicable, A

A: A

minoa

cid, M

AF: M

inor

Allele Frequ

ency, h

m:

hom

ozyg

ous, ht: heteroz

ygou

s, Blank cells represen

t wild type

, T: T

olerated

, D: D

eleterious, Pb D for

“Proba

bly Dam

aging”

, Ps D for

“Pos

sibly Dam

aging”

and B for

Ben

ign. *

Allele

freq

uen

cies assumed

from N

HLBI GO Exo

me Se

quen

cing Project.


121

PARREIRA B ET AL

fIlTERINg RESUlTS

Approximately 38 Mb of sequence per patient was gen-erated and the capture specificity and sensitivity in allsamples was about 55% and 94%, respectively. The re-sults for each sample after SNV calling and indel identi-fication steps are summarized in supplementary Table III.After filtering, 21 missense, deletion and splice site

variants in 17 genes were obtained: PLCG2, ALPL,

CASR, FGF2, COL11A2, ENPP1, MGP, VDR, BMP-4,COL1A1, TGF�1, BMP-2, COL6A1, FLNC, AMER3,PPP2R2D and ABCC6. Ten of the identified variantswere present in the HGMD mutation database, linkedto phenotypes, other than DISH/CC. The identifiedvariants and available functional information are pre-sented in Table II. Sequence conservation of the missense variants were

fIgURE 2. Sequence conservation in six vertebrates of the variants identified in candidate genes possibly associated with the DISH/CCphenotype in humans (some genes are not available in all species). The variant for the FGF2 gene is not presented since it is located inthe 3’ untranslated region.Abbreviations: PLCG2- Phospholipase C Gamma 2, ALPL- Alkaline Phosphatase, Liver/Bone/Kidney, CASR- calcium-sensing receptor,COL11A2- Collagen Type XI Alpha 2 Chain, ENPP1– ectonucleotide pyrophosphatase/phosphodiesterase 1, MGP- Matrix Gla Protein,VDR- Vitamin D Receptor, BMP4- Bone morphogenetic protein 4, COL1A1- Collagen Type I Alpha 1 Chain, BMP2- Bone morphogeneticprotein 2, COL6A1- Collagen Type VI Alpha 1 Chain, FLNC- Filamin C, AMER3- APC Membrane Recruitment Protein 3, PPP2R2D-Protein Phosphatase 2 Regulatory Subunit B delta, ABCC6 – ATP-binding cassete subfamily C, member 6.


122


then evaluated in 6 different vertebrates. It is evidentin Figure 2 that the variants in the genes PPP2R2D,BMP2, FLNC, and CASR are highly conserved in all thevertebrates analyzed. Variants of the genes VDR, BMP4,COL1A1 and ABCC6 are only conserved in mammals.The degree of conservation is not directly linked to low-er allele frequency, since several of these variants havehigh Minor Allele Frequencies (MAF).

ASSOCIATION BETWEEN vARIANTS AND THE

DISH/CC pHENOTypE

Two cohorts, one of 65 unrelated DISH/CC patients(45 males, 20 females; age of onset around 40 years)and another of 118 unrelated controls without anysigns of DISH/CC, with a similar ethnic background,were selected after radiological characterization (47males, 71 females; mean current age, 68 years; range,60-104) and used for association studies. In cases weselected younger patients around the age of 40, wheregenetic factors may be associated, however in the con-trols we selected older individuals to ensure that theyhave not developed the disease.All patients had at least 2 highly conserved genetic

variants in combination with other variants that arenormally conserved in mammals (Table II). Variantswith a high degree of conservation, present in 4 or 3WES patients, irrespective of the MAF, were selectedfor the association study. The variants rs34473884,rs9277934 and rs2228570 in PPP2R2D, COL11A2 andVDR genes, respectively, were present in all four WESpatients. The variants rs1048201, rs235768,rs1044498 and rs17563 in FGF2, BMP2, ENPP1 and

BMP4 genes, respectively, were present in three of thefour patients selected for WES. Seven variants werescreened in a group of 65 DISH/CC patients and 118controls and the results are indicated in Table III. Thevariant rs34473884 in the PPP2R2D gene was the onlythat gave significant results. It was found to be muchmore frequent in the DISH/CC group than in the con-trols (p=0.028; OR=1.789, 95% CI= 1.060 - 3.021).

DISCUSSION

In this study, we used WES as a method to identify can-didate genes for DISH/CC aetiology and associationstudies to investigate some of the identified variants.As expected, thousands of protein coding variants perpatient were identified across each exome. Conse-quently, a number of filtering strategies were used to se-lect potential high risk variants of genes potentially as-sociated with DISH. The filtering strategies employedwere based on previous knowledge about gene func-tion. The major limitations of the strategy used to iden-tify candidate genes is that unannotated genes andthose with unknown functions were not investigated inthe study.Very few genetic studies have been published about

DISH so far. DISH susceptibility genes have previous-ly been investigated, including Human Leukocyte Anti-gens (HLA),17-24 Collagen 6A1 gene (COL6A1),25 Fi-broblast Growth factor 2 (FGF2),26 Vitamin D (1,25-Dihydroxyvitamin D3) Receptor (VDR) and CollagenType I�1 (COL1A1),27 but only two genes are known to

TABlE III. ASSOCIATION STUDy BETWEEN SEvEN vARIANTS fROM SEvEN gENES AND THE DISH/CC pHENOTypE

MAFSNP Allele DISH/CC Controls

Chr Gene M/m MAF (N=65) (N=118) OR (95% CI) p-value10 PPP2R2D rs34473884 G/A 0.262 0.165 1.789 (1.060 - 3.021) 0.02814 BMP4 rs17563 T/C 0.454 0.390 1.301 (0.843 - 2.006) 0.2344 FGF2 rs1048201 C/T 0.139 0.131 1.063 (0.569 - 1.985) 0.8496 ENPP1 rs1044498 A/C 0.200 0.203 0.979 (0.574 - 1.670) 0.9386 COL11A2 rs9277934 G/A 0.385 0.360 1.110 (0.713 - 1.728) 0.64320 BMP2 rs235768 A/T 0.292 0.284 1.042 (0.650 - 1.671) 0.86512 VDR rs2228570 T/C 0.400 0.373 1.121 (0.723 - 1.739) 0.609

The minor alleles are indicated in bold.Abbreviations: Chr: Chromossome, SNP: Single nucleotide polymorphism, M/m: major allele/minor allele, MAF: Minor allele frequency andOR (95% CI): Odds Ratio (95% Confidence Interval).


123

PARREIRA B ET AL

have a positive association with DISH susceptibility;COL6A128-29 and FGF2.26 However, the COL6A1 andFGF2 variants associated with the disease, are locatedin non-coding regions and are common variants with-in the general population, suggesting they have a smallgenetic effect. In our study we found 3 genetic variants that were

present in all 4 DISH/CC patients and 6 that were pre-sent in 3 DISH/CC patients. Minor allele frequencies ofthese variants were high (mean 0.25), meaning thatthese were common variants in the general population.Rare nonsynonymous SNPs are two times more likelyto be predicted as protein-affecting, when compared tocommon SNPs. However, in our study three commonSNPs, shared by the 4 patients (G358S, E276K andM1T in PPP2R2D, COL11A2 and VDR genes, respecti -vely), according to SIFT and/or PolyPhen algorithmswere predicted to have a strong effect on the protein.Eight of these genetic variants, in heterozygous and/orhomozygous states, were in conserved positions of pro-teins associated with mineralization; four of which werein regions that were highly conserved across the verte-brates. The four WES selected patients had at least 2highly conserved genetic variants in combination withother variants that were conserved in mammals. The as-sociation study indicated that the SNP rs34473884 inthe PPP2R2D gene was significantly associated with theDISH/CC phenotype. In humans, the full PPP2R2D (Protein Phosphatase

2 Regulatory Subunit � delta) gene structure is com-posed of 9 exons and spans 79.19 kb in chromosome10. The gene encodes a crucial serine/threonine proteinphosphatase that regulates basal cellular activities bydephosphorylating substrates. Protein Ser/Thr phos-phatases are a group of enzymes that catalyze the re-moval of phosphate groups from serine and/or threo-nine residues by hydrolysis of phosphoric acidmonoesters. They oppose the action of kinases andphosphorylases and are involved in signal transduc-tion. Protein phosphatases have long been postulatedto influence TGF-� superfamily signaling, which regu-lates numerous cellular responses30. Despite the long-standing suspected influence of protein phosphatasesin TGF-� signaling, concrete data confirming the inter-action only started to emerge recently. In human celllines, PPP2R2D negatively modulates TGF-�/Activin//Nodal signaling by inhibiting the type I receptorsALK4 and ALK531. The variant c.1072G>A (G327S) found in PPP2R2D

gene is a missense variant which causes substitution of

glycine for serine at amino acid 327 in the protein.These two amino acids are hydrophilic but glycine isnon-polar and serine is polar. The glycine is normallytotally conserved in all vertebrates studied, and thevariant has a low, deleterious, SIFT score (0.03) but thePolyphen score (0.33) does not corroborate its harm-ful effect. The frequency of this variant is high in Eu-ropean populations with a MAF of 0.18 and this vari-ant was identified in all four DISH/CC patients usedfor WES; AZ1 was homozygous and AZ2, 3 and 4 wereheterozygous. As far as we known no phenotype hasbeen associated with this gene. Rare variants, with a deleterious effect on proteins in

the human genome, are the basis for the developmentof Mendelian diseases. However, common polymor-phisms that individually exert small effects might, as agroup, also play a substantial genetic role. Despite thelack of association of most of the studied gene variantsin the present study, this may be in part linked to theheterogeneous features of the phenotype, and so im-proved characterization of the disorder under study isessential for the planning of future studies. The resultsof our study lead us to suggest that DISH/CC is poly-genic, and is influenced by the interaction of multiple,small effect gene variants and possibly by unknown en-vironmental factors.

CONClUSION

Our results underline the polygenic nature of DISH anda number of conserved and sometimes deleterious vari-ants were identified in genes with a role in mineraliza-tion. The variant rs34473884 in the PPP2R2D gene wassignificantly associated with the DISH/CC phenotypeand we propose it may contribute to the developmentof this disorder. Further studies will be needed to con-firm the association of PPP2R2D with the phenotypeunder study.

ACkNOWlEDgEMENTS

The authors thank all the patients who participated in this researchand made it possible. We also thank Isa Dutra (MSc), João PauloPinheiro (BsC) and Raquel Meneses (BsC) from Hospital Santo Es-pírito da Ilha Terceira, EPE, Epidemiology and Molecular BiologyService (SEEBMO), Angra do Heroísmo, Portugal, who performedthe DNA extractions and Vânia Machado (MSc), who participatedin the elaboration of the pedigrees. BP was supported by “Fundo Re-gional para a Ciência e Tecnologia (FRCT)” (M3.1.2/F/023/2011).

COMpETINg fINANCIAl INTERESTS

The authors declare no competing interests.


124


CORRESpONDENCE TO

Jácome Bruges ArmasServiço Especializado de Epidemiologia e Biologia Molecular (SEEBMO) Hospital de Santo Espírito da Ilha Terceira (HSEIT)Canada do Briado9700-049, Angra do HeroísmoE-mail: [email protected]

REfERENCES

1. Bruges-Armas J, Couto AR, Timms A, Santos MR, BettencourtBF, Peixoto MJ, et al. Ectopic calcification among families in theAzores: clinical and radiologic manifestations in families withdiffuse idiopathic skeletal hyperostosis and chondrocalcinosis.Arthritis Rheum 2006; 54: 1340-1349.

2. Mader R. Clinical manifestations of diffuse idiopathic skeletalhyperostosis of the cervical spine. Sem Arthritis Rheum 2002;32: 130-135.

3. Mader R, Verlaan JJ, Buskila D. Diffuse idiopathic skeletal hy-perostosis: clinical features and pathogenic mechanisms. NatRev Rheumatol 2013; 9: 741-750.

4. Utsinger PD. Diffuse Idiopathic Skeletal Hyperostosis. ClinRheum Dis 1985; 11: 325-351.

5. Weinfeld RM, Olson PN, Maki DD, Griffiths HJ. The prevalenceof diffuse idiopathic skeletal hyperostosis (DISH) in two largeAmerican Midwest metropolitan hospital populations. SkeletalRadiol 1997; 26: 222-225.

6. Beardwell A. Familial Ankylosing Vertebral Hyperostosis WithTylosis. Ann Rheum Dis 1969; 28: 518-523.

7. Woodard JC, Poulos PW, Jr., Parker RB, Jackson RI, Jr., EurellJC. Canine diffuse idiopathic skeletal hyperostosis. Vet Pathol1985; 22: 317-326.

8. Morgan J, Stavenborn M. Disseminated idiopathic skeletal hy-perostosis (DISH) in a dog. Vet Radiol Ultrasound 1991; 32:65-70.

9. Gutierrez M, Silveri F, Bertolazzi C, Giacchetti G, Tardella M, DiGeso L, et al. [Gitelman syndrome associated with chondrocal-cinosis: description of two cases]. Reumatismo 2010; 62: 60-64.

10. Netter P, Bardin T, Bianchi A, Richette P, Loeuille D. The ANKHgene and familial calcium pyrophosphate dihydrate depositiondisease. Joint Bone Spine 2004; 71: 365-368.

11. Ramos YF, Bos SD, van der Breggen R, Kloppenburg M, Ye K,Lameijer EW, et al. A gain of function mutation in TNFRSF11Bencoding osteoprotegerin causes osteoarthritis with chondro-calcinosis. Ann Rheum Dis 2015; 74: 1756-62.

12. Couto AR, Parreira B, Thomson R, Soares M, Power DM,Stankovich J, et al. Combined approach for finding suscepti-bility genes in DISH/chondrocalcinosis families: whole-genome-wide linkage and IBS/IBD studies. Hum Genome Var 2017; 4:17041.

13. Ng SB, Buckingham KJ, Lee C, Bigham AW, Tabor HK, DentKM, et al. Exome sequencing identifies the cause of a mendeliandisorder. Nat Genet 2010; 42: 30-35.

14. Rabbani B, Tekin M, Mahdieh N. The promise of whole-exomesequencing in medical genetics. J Hum Genet 2014; 59: 5-15.

15. Cooper GM, Stone EA, Asimenos G, Green ED, Batzoglou S,Sidow A. Distribution and intensity of constraint in mammaliangenomic sequence. Genome Res 2005; 15: 901-913.

16. Purcell S, Neale B, Todd-Brown K, Thomas L, Ferreira MA, Ben-der D, et al. PLINK: a tool set for whole-genome association andpopulation-based linkage analyses. Am J Hum Genet 2007; 81:559-575.

17. Rosenthal M, Bahous I, Müller W. Letter: HL-A B27 and modi-fied bone formation. Lancet 1976 1: 543.

18. Rotés Querol J, Ercilla González G. Letter: HLA-B27 and mod-ified bone formation. Lancet 1976 1: 482-3.

19. Brigode M, Francois RJ. Histocompatibility antigens in vertebralankylosing hyperostosis. J Rheumatol 1977 4: 429-434.

20. Spagnola AM, Bennett PH, Terasaki PI. Vertebral ankylosing hy-perostosis (Forestier's disease) and HLA antigens in Pima Indi-ans. Arthritis Rheum 1978 21: 467-72.

21. Perry JD, Wolf H, Festenstein H, Storey GO. Ankylosing hy-perostosis: a study of HLA A, B, and C antigens. Ann Rheum Dis1979 38: 72-73.

22. Dostál C, Ivasková E, Hána I, Sváb V. HLA antigens in vertebralankylosing hyperostosis (Forestier's disease). J Hyg EpidemiolMicrobiol Immunol 1983; 27: 98-102.

23. Rosenthal M, Bahous I, Muller W. Increased frequency of HLAB8 in hyperostotic spondylosis. J Rheumatol Suppl 1977; 3: 94--96.

24. Ercilla González G, Brancós MA, Breisse Y. Histocompatabilityantigens in Forestier's disease, polyarthrosis and ankylosingspondylitis. HLA and Disease. Paris: INSERM; 1976. p. 28.

25. Tsukahara S, Miyazawa N, Akagawa H, Forejtova S, Pavelka K,Tanaka T, et al. COL6A1, the candidate gene for ossification ofthe posterior longitudinal ligament, is associated with diffuse id-iopathic skeletal hyperostosis in Japanese. Spine 2005 30: 2321--2324.

26. Jun JK, Kim SM. Association study of fibroblast growth factor 2and fibroblast growth factor receptors gene polymorphism inkorean ossification of the posterior longitudinal ligament pa-tients. J Korean Neurosurg Soc 2012; 52: 7-13.

27. Havelka S, Uitterlinden AG, Fang Y, Arp PP, Pavelková A, VeseláM, et al. Collagen type I(alpha 1) and vitamin D receptor poly-morphisms in diffuse idiopathic skeletal hyperostosis. ClinRheumatol 2002 21: 347-348.

28. Tanaka T, Ikari K, Furushima K, Okada A, Tanaka H, FurukawaK, et al. Genomewide linkage and linkage disequilibrium anal-yses identify COL6A1, on chromosome 21, as the locus for os-sification of the posterior longitudinal ligament of the spine.Am J Hum Genet 2003; 73: 812-822.

29. Kong Q, Ma X, Li F, Guo Z, Qi Q, Li W, et al. COL6A1 poly-morphisms associated with ossification of the ligamentumflavum and ossification of the posterior longitudinal ligament.Spine (Phila Pa 1976) 2007; 32: 2834-2838.

30. Liu T, Feng XH. Regulation of TGF-beta signalling by proteinphosphatases. Biochem J 2010; 430: 191-198.

31. Batut J, Schmierer B, Cao J, Raftery LA, Hill CS, Howell M. Twohighly related regulatory subunits of PP2A exert opposite ef-fects on TGF-beta/Activin/Nodal signalling. Development 2008;135: 2927-2937.


125

PARREIRA B ET AL

SUpplEMENTARy fIlES

SUpplEMENTARy TABlE I. NUMBER Of

CANDIDATE gENES pER SAMplE AND NUMBER

Of CANDIDATE gENES SHARED By DISH/CC

pATIENTS

Number ofcandidate genes

Dominant Recessive Samples model ModelAZ1 815 48AZ2 917 58AZ3 872 47AZ4 593 25AZ1+AZ2 220 6AZ1+AZ3 212 4AZ1+AZ4 139 4AZ2+AZ3 249 6AZ2+AZ4 149 4AZ3+AZ4 162 3AZ1+AZ2+AZ3 118 2AZ1+AZ2+AZ4 65 1AZ2+AZ3+AZ4 78 2AZ1+AZ3+AZ4 77 1AZ1+AZ2+AZ3+AZ4 52 1

SU

pp

lE

ME

NTA

Ry

TA

BlE

II. l

IST

Of

gE

NE

S A

ND

TH

EIR

AC

CE

SS

ION

TR

AN

SC

RIp

T N

UM

BE

RS

US

ED

fO

R C

ON

SE

Rv

AT

ION

AN

Aly

SIS

.

SPECIES

GENES

Human

Chimpanzee

Dog

Mouse

Chicken

Zebrafish

PLCG2

ENST

0000

0564

138.5

ENSP

TRT00

0000

1547

9.3

ENSC

AFT00

0000

3181

5.3

ENSM

UST

0000

0081

232.8

ENSG

ALT

0000

0050

238.1

ENSD

ART00

0000

2139

9.7

ALPL

ENST

0000

0374

840.7

ENSP

TRT00

0000

0059

2.2

ENSC

AFT00

0000

2357

8.3

ENSM

UST

0000

0030

551.10

ENSG

ALT

0000

0068

960.1

ENSD

ART00

0001

3110

1.2

CASR

ENST

0000

0498

619.3

ENSP

TRT00

0000

4399

6.4

ENSC

AFT00

0000

1876

0.3

ENSM

UST

0000

0063

597.13

XM_4

1649

1.5.1

ENSD

ART00

0000

1093

4.8

COL11

A2

ENST

0000

0341

947.6

ENSP

TRT00

0000

4856

4.4

ENSC

AFT00

0000

0140

9.4

ENSM

UST

0000

0087

497.10

Ni

ENSD

ART00

0001

0575

4.4

ENPP1

ENST

0000

0360

971.6

ENSP

TRT00

0000

3435

6.3

ENSC

AFT00

0000

0000

1.3

ENSM

UST

0000

0105

520.7

ENSG

ALT

0000

0066

498.1

ENSD

ART00

0001

2735

0.3

MGP

ENST

0000

0539

261.5

ENSP

TRT00

0000

0873

2.4

ENSF

CAT

0000

0031

714.1 (C

at)

ENSM

UST

0000

0032

342.2

ENSG

ALT

0000

0019

173.4

ENSD

ART00

0001

4962

2.2

VDR

ENST

0000

0229

022.7

Ni

ENSC

AFT00

0000

4347

3.2

ENSM

UST

0000

0023

119.14

ENSG

ALT

0000

0071

682.1

ENSD

ART00

0001

6189

2.1

BMP4

ENST

0000

0245

451.8

ENSP

TRT00

0000

1164

3.4

ENSC

AFT00

0000

2362

4.3

ENSM

UST

0000

0074

077.11

ENSG

ALT

0000

0020

316.5

ENSD

ART00

0000

7515

0.4

COL1A

1ENST

0000

0225

964.9

ENSP

TRT00

0000

1723

1.4

ENSC

AFT00

0000

2695

3.3

ENSM

UST

0000

0001

547.7

XM_0

1527

3228

.1.1

ENSD

ART00

0000

0939

3.7

BMP2

ENST

0000

0378

827.4

ENSP

TRT00

0000

2460

6.3

ENSC

AFT00

0000

4969

0.1

ENSM

UST

0000

0028

836.6

ENSG

ALT

0000

0065

435.1

ENSD

ART00

0001

6665

7.1

COL6A

1ENST

0000

0361

866.7

ENSP

TRT00

0000

2615

6.3

ENSC

AFT00

0000

1891

8.3

ENSM

UST

0000

0001

147.4

ENSG

ALT

0000

0039

669.3

ENSD

ART00

0001

1060

8.3

FLNC

ENST

0000

0325

888.12

ENSP

TRT00

0000

3644

4.5

ENSC

AFT00

0000

0250

4.3

ENSM

UST

0000

0065

090.7

NM_2

0457

3.1.1

Ni

AMER3

ENST

0000

0423

981.1

ENSP

TRT00

0000

2312

4.3

ENSC

AFT00

0000

0674

7.2

ENSM

UST

0000

0052

670.10

ENSG

ALT

0000

0055

960.1

ENSD

ART00

0001

4999

2.2

PPP2R

2DENST

0000

0455

566.5

ENSP

TRT00

0000

0582

9.4

ENSC

AFT00

0000

4441

6.1

ENSM

UST

0000

0041

097.12

ENSG

ALT

0000

0081

063.1

ENSD

ART00

0001

7217

5.1

ABCC6

ENST

0000

0205

557.11

ENSP

TRT00

0000

1439

8.4

ENSC

AFT00

0000

2890

8.3

ENSM

UST

0000

0002

850.7

ENSG

ALT

0000

0048

627.1

ENSD

ART00

0001

7294

3.1

Data was retriev

ed from the Ensembl gen

ome da

taba

se; a

ccessed on

Nov

embe

r 20

16.

Ni: not id

entified

.


126


SUpplEMENTARy TABlE III. kNOWN, NOvEl AND TOTAl NUMBER Of SNvS AND INDElS fOR THE fOUR

SAMplES TESTED (AZ1-4)

SNVs Indels TOTAL/Sample Known Novel Total Known Novel Complex Total sampleAZ1 17281 1619 18900 457 405 51 913 19813AZ2 18086 1759 19845 440 496 66 1002 20847AZ3 18327 1650 19977 524 505 73 1102 21079AZ4 17575 1223 18798 475 424 72 971 19769

SNVs: Single nucleotide variants.

Documents

Whole exome sequencing of patients with diffuse idiopathic