Genetic and Epigenetic Determinants of Low Dysferlin Expression in Monocytes

RESEARCH ARTICLEOFFICIAL JOURNAL

www.hgvs.org

Genetic and Epigenetic Determinants of Low DysferlinExpression in Monocytes

Eduard Gallardo,1,2∗ Arunkanth Ankala,3 Yaiza Nunez-Alvarez,4 Madhuri Hegde,3 Jordi Diaz-Manera,1,5,2 Noemı De Luna,6,7

Ana Pastoret,1,2 Monica Suelves,4 and Isabel Illa1,5,2

1Laboratori de Malalties Neuromusculars, Institut de Recerca de HSCSP, Universitat Autonoma de Barcelona (UAB), Barcelona, Spain; 2Centrode Investigacion Biomedica en Red de Enfermedades Neurodegenerativas (CIBERNED), Madrid, Spain; 3Department of Human Genetics, EmoryUniversity School of Medicine, Atlanta, Georgia; 4Institut de Medicina Predictiva i Personalitzada del Cancer (IMPPC) i Institut Germans Trias iPujol (IGTP), Badalona, Spain; 5Servei de Neurologia, Hospital de Sant Pau, Universitat Autonoma de Barcelona (UAB), Barcelona, Spain;6Laboratori de Patologia Mitocondrial i Neuromuscular, Hospital Universitari Vall d’Hebron, Institut de Recerca (VHIR), Universitat Autonoma deBarcelona; 7Centre for Biomedical Network Research on Rare Diseases (CIBERER), Instituto de Salud Carlos III, Valencia, Spain

Communicated by Johan T. den DunnenReceived 24 February 2014; accepted revised manuscript 2 May 2014.Published online 17 May 2014 in Wiley Online Library (www.wiley.com/humanmutation). DOI: 10.1002/humu.22591

ABSTRACT: Dysferlinopathies are autosomal recessive in-herited muscular dystrophies caused by mutations in thegene DYSF. Dysferlin is primarily expressed in skeletalmuscle, cardiac muscle, and peripheral blood monocytes.Expression in skeletal muscle and monocytes strongly cor-relates in healthy and disease states. We evaluated the ef-ficiency of the monocyte assay to detect carriers and to de-termine the carrier frequency of dysferlinopathies in thegeneral population. We enrolled 149 healthy volunteersand collected peripheral blood samples for protein analy-sis. While 18 of these individuals with protein levels in therange of 40%–64% were predicted to be carriers by themonocyte assay, subsequent DYSF sequencing analysis in14 of 18 detected missense variants in only four. Analysisof DNA methylation patterns at the DYSF locus showedno changes in methylation levels at CpG islands and shoresbetween samples. Our results suggest that: (1) dysferlinexpression can also be regulated by factors outside of thedysferlin gene, but not related to DNA methylation; (2)carrier frequency and therefore the number of affected in-dividuals could be higher than previously estimated; and(3) although reliable for evaluating dysferlinopathies, themonocyte assay cannot be used to determine the carrierstatus; for this, a molecular analysis of DYSF must beperformed.Hum Mutat 35:990–997, 2014. C© 2014 Wiley Periodicals, Inc.

KEY WORDS: DYSF; dysferlin; monocyte; methylation

Additional Supporting Information may be found in the online version of this article.∗Correspondence to: Eduard Gallardo, Neuromuscular Disorders Laboratory, De-

partment of Neurology, Institut de Recerca de l’Hospital de Sant Pau, Universitat

Autonoma de Barcelona (UAB), C/Pare Claret 167, 08025, Barcelona, Spain. E-mail:

[email protected]

Contract grant sponsors: Jain Foundation; (FIS 12/2291) to E.G.; Ministerio de Ciencia

e Innovacion (SAF2012-37427) to M.S.; Generalitat de Catalunya (2009 SGR1356) to M.S.;

Postdoctoral fellowship Sara Borrell (CD11/00060) from the Spanish Ministry of Health

to N.D.L; FPU fellowship from the Ministerio de Ciencia e Innovacion to Y.N.

IntroductionLimb-girdle muscular dystrophy type 2B (MIM# 253601),

Miyoshi myopathy (MIM# 254130), distal myopathy with ante-rior tibial onset (MIM# 606768), and other rare forms of musculardystrophy associated with mutations in the DYSF (MIM# 603009)gene are collectively referred to as dysferlinopathies. These inher-ited muscle disorders involve progressive muscle wasting, typicallybeginning in the late second decade. The DYSF gene maps to chro-mosome 2p13, has 55 exons and encodes a protein of 237 kDa [Liuet al., 1998]. Dysferlin is located in the sarcolemma where it has beenreported to be involved in sarcolemma repair and muscle differen-tiation [Bansal et al., 2003; de Luna et al., 2006]. Dysferlin is alsoexpressed in other cell types, including monocytes, heart, kidney,and placenta [Anderson et al., 1999; Ho et al., 2002]. Dysferlin-deficient monocytes reportedly display excessive phagocytosis andimpaired cell adhesion [Nagaraju et al., 2008; de Morree et al., 2013].

A complete diagnostic protocol for dysferlinopathies involves ini-tial clinical evaluation, preliminary protein studies, and molecularconfirmation by DYSF genetic analysis. The high genetic and phe-notypic heterogeneity of the different LGMDs, of which dysfer-linopathies are a subtype, make disease diagnosis very challenging.Therefore, minimally invasive and comparatively inexpensive assayssuch as the blood monocyte assay [Ho et al., 2002] that is availablefor dysferlinopathy diagnosis can be extremely useful, especiallyin countries where genetic testing is still not readily available andaffordable.

The diagnostic utility of the blood-based monocyte assay wasestablished by showing a high correlation between dysferlin levelsin skeletal muscle and monocytes in healthy individuals, diseasemanifesting patients, and unaffected family members who carrya single mutation [Gallardo et al., 2011]. In a recent report fromour laboratory, we established the specificity and sensitivity of themonocyte assay and determined that dysferlin levels lower than23% of a healthy individual are characteristic of the presence ofbiallelic mutations in DYSF, while levels between 23% and 75%are indicative of the presence of one mutated allele [Luna et al.,2012]. However, for some of these individuals, a mutated allelehas not been identified, suggesting other mechanisms regulatinggene expression. In addition, although dysferlinopathies primarilyare present as autosomal recessive muscular dystrophies, there havebeen reports of symptomatic carriers with only one mutation inDYSF [Illa et al., 2007].

C© 2014 WILEY PERIODICALS, INC.

DNA methylation at CpG dinucleotides is the major epigeneticmechanism associated with long-term gene silencing [Jones, 2012],and although much of the human genome is CpG depleted, thisdinucleotide can be found at close to its expected frequency insmall genomic regions known as CpG island [Gardiner-Gardenand Frommer, 1987; Takai and Jones, 2002]. These areas are usu-ally protected from methylation and are mainly located in proxi-mal regulatory regions [Bird, 2002; Fazzari and Greally, 2004; Golland Bestor, 2005]. However, methylation of CpG islands, both atphysiological and pathological conditions, results in gene silenc-ing [Portela and Esteller, 2010; Sharma et al., 2010; Baylin andJones, 2011]. In the present study, we addressed for the first timewhether DNA methylation levels could modulate dysferlin expres-sion, comparing DNA methylation patterns of dysferlin CpG islandsbetween individuals, covering the two annotated transcription startsites (TSSs) according to the UCSC genome bioinformatics data(http://genome.ucsc.edu/).

In addition, we took advantage of the minimally invasive and rel-atively inexpensive monocyte assay to perform a carrier screening.Dysferlinopathies are rare and their exact incidence and worldwidecarrier frequencies are not known. Anecdotal evidence predicts thatthe disease frequency is around 1–5 per million [Norwood et al.,2009]. However, given the difficulty and high cost of genetic analy-sis, it is likely that this disease is underdiagnosed and may be moreprevalent. We undertook this study to determine the frequency ofdysferlin carriers in a sample population and estimate the true dis-ease incidence. We hypothesized that if the predicted 1–5 per millionincidence rate is accurate, we would find no more than one carrierin 100 individuals in the general population.

Materials and Methods

Volunteers

The clinical research ethics committee from Hospital de la SantaCreu i Sant Pau approved this study (reference number of our EthicsCommittee IIBSP-MON-2011-157). We enrolled 149 volunteers do-nating blood at the Blood and Tissue Bank of the Hospital de la SantaCreu i Sant Pau, Spain. All volunteers received information aboutthe study. Blood samples were collected from each participatingindividual after receiving informed consent.

Isolation of Peripheral Blood Monocytes CD14+ and CD14−

Cell Population

Blood was collected in a Cell Preparation Tube (CPT), and periph-eral blood mononuclear cells (PBMCs) were isolated from wholeblood of volunteers by CPT gradient centrifugation at 1650g for20 min (Becton, Dickinson and Company, NJ). Peripheral bloodmonocytes (PBMs) CD14 positive isolation was carried out in allvolunteers as previously described [Gallardo et al., 2011]. Briefly,PBMCs were washed with autoMACS running buffer (MiltenyBiotec, Bergisch Gladbach, Germany) at 300g for 10 min. Afterthat, 10 million PBMCs were mixed with 20 μl of anti-CD14-coatedmicrobeads (Milteny Biotec) and incubated at 4°C for 15 min. Un-bound microbeads were removed by washing cells in excess au-toMACS running buffer followed by centrifugation at 300g for10 min. Cell pellet was resuspended in 500 μl of the buffer andit was transferred to MACS separation columns 25 MS, accordingto manufacturer’s instructions (Milteny Biotec). CD14+ and CD14–

fractions were stored at –20°C.

Western Blot of PBM CD14+

Western blots using lysates of the CD14+ fraction of PBMcells were performed as described in detail elsewhere [Gallardoet al., 2011]. Nitrocellulose transfer membrane (Whatman, Dassel,Germany) was incubated for 1 h at room temperature with pri-mary antibody, human dysferlin NCL-Hamlet (Leica Byosystems,Wetzlar, Germany) (1:300 dilution), and mouse monoclonal anti-body to β-actin (Sigma-Aldrich, St Louis, MO) (1:2,000 dilution),used as loading control. IRDye R© goat anti-mouse (Li-cor, Lincoln,NE) (1:750 dilution) was used as a secondary antibody. Nitrocellu-lar membrane was incubated for 1 h at room temperature with thesecondary antibody. Dysferlin and β-actin fluorescent bands wereacquired and analyzed using the Odyssey Infrared Imaging System(Li-cor). The dysferlin double band was quantified and normalizedto the β-actin band.

Statistical Analysis

To test the relation between normally distributed quantitativevariable such as dysferlin expression and sex and then comparedysferlin expression and age (younger and older than 35 years old),a nonpaired Student’s test was used. Differences were consideredsignificant if P-values were lower than 0.05.

Sequence Analysis of the Dysferlin Gene

DNA was extracted from CD14– cells using the DNeasy Bloodand Tissue Kit (Qiagen, Hilden, Germany) according to the man-ufacturer’s instructions. Sequencing analysis of the entire codingregion of the 55 exons of the DYSF gene and flanking intron/exonboundaries up to 50 bp was performed at the CLIA-approved DNAdiagnostic laboratory of Emory Genetics Laboratory (EGL) (At-lanta, GA).

Sequence Analysis and Variant Detection

Sequences were aligned with DYSF reference sequence(NM 003494.3) downloaded from NCBI (http://www.ncbi.nlm.nih.gov/) using the Mutation Surveyor v3.30 software (SoftGeneticsLLC, PA) and sequence variants were identified. Nucleotide num-ber 1 corresponds to the “A” of the translation initiation codonATG of the reference sequence. Observed sequence variants wereannotated using the Human Genome Variation Society guidelinesand the clinical significance of the variants detected in each samplewas interpreted using EGL’s in-house variant database EmVClass(http://genetics.emory.edu/egl/emvclass/emvclass.php) as well asonline databases such as the 1000 Genome Database (http://www.1000genomes.org/), exome variant server (EVS) (http://evs.gs.washington.edu/EVS/), and dbSNP database (http://www.ncbi.nlm.nih.gov/projects/SNP/). Additional mutation databases includingHuman Genome Mutation Database (www.hgmd.org), the Leidendysferlin-specific database (http://www.LOVD.nl/DYSF), and theUMD-DYSF database (http://www.umd.be/DYSF/) were also usedto determine whether the observed variants have been previouslyreported as benign variants or disease-causing mutations. The al-gorithm used in the UMD-DYSF was described by Frederic et al.(2009) and it is freely available at http://www.umd.be. Predictionof the effect of changes in DNA on protein should be taken withcaution since mRNA was not studied in any of the subjects of thisstudy. Variants that were observed for the first time, or those withno specific or conflicting information in the literature as to their

HUMAN MUTATION, Vol. 35, No. 8, 990–997, 2014 991

clinical consequence, were classified as variants of unknown oruncertain significance (VOUS). Additionally, all variants detectedin this study were submitted to NCBI’s ClinVar public database(https://www.ncbi.nlm.nih.gov/clinvar/).

Bisulfite Conversion

Bisulfite coversion was performed using 500 ng of DNA with EZDNA MethylationTM Kit (ZymoResearch, Orange, CA,) accordingto manufacturer’s instruction. Converted DNA was eluted in 30 μlof MiliQ water.

Direct Bisulfite Sequencing

Bisulfite sequencing was performed following Dr. Clark’s proce-dure [Clark et al., 2006]. For each region of interest, PCR ampli-fication was first performed on 1 μl of bisulfite-treated DNA withconventional PCR method in a final volume of 12.5 μl, in duplicate.These PCR products were diluted fivefold or directly used as tem-plate for a nested PCR with conventional PCR method, according toeach specific primer set. The nested PCR was performed one reac-tion per PCR product of the previous reaction in a final volume of25 μl. Primers were designed using MethPrimer [Li and Dahiya,2002] and Bisearch (Supp. Table S1). At the end, the duplicatesamples were pooled, purified with JETQUICK PCR Spin KIT(Genomed, St.Louis, MO) and sequenced with specific primers atGATC Biotech service (Constance, Germany).

Results

Dysferlin Expression Levels by Age and Gender

Dysferlin expression levels in monocytes were measured in all149 volunteers (47% female) by Western blotting (Supp. Table S2).The average age of female participants was 39.3 years (range 20–65)and of male participants was 45.3 years (range 21–68). To deter-mine whether dysferlin expression varied by age, we divided thevolunteers into two groups: those that were younger than 35 yearsold (n = 50) and those older than 35 years (n = 97). No significantdifferences in dysferlin expression were found between these twogroups (95.64% vs. 95.26%; Student’s t-test P = 0.91), suggestingthat dysferlin expression in monocytes does not vary significantlybetween these age groups (range 20–68 years old). Comparison ofdysferlin levels by gender in the 68 female and 79 male participantsshowed a significantly higher expression of dysferlin in females overmales (99.01% vs. 92.40%, Student’s t-test P = 0.04).

Identification of Carriers by Protein Level

Regardless of age and gender, most individuals had monocytedysferlin levels in the expected healthy noncarrier range (>75% ofnormal controls, we assign a value of 100% to a healthy individualthat has no mutations in DYSF) (Fig. 1). Eighteen (12%) of the149 participants studied had dysferlin levels between 40% and 64%(Fig. 2) placing them within the established carrier range of 23%–75%, and suggesting that they carry a pathogenic mutation in theDYSF gene.

Figure 1. Dysferlin expression analysis in monocytes by Western blot.Example of the membranes analyzed in this study. Lane 1: volunteernumber 21 showed a dysferlin expression of 75%.; Lane 2: volunteernumber 23 showing 100% expression; Lane 3: volunteer number 140showing 81% expression; Lane 4: volunteer number 45 showing 77%expression; and Lane 5: volunteer number 25 displayed 47% expressionof dysferlin. β-actin was used as a loading control.

Figure 2. Representation of dysferlin expression of volunteers withreduced levels of dysferlin compared to volunteers with normal ex-pression. Unaffected volunteers have a dysferlin expression averageof 101%, SD: 13.3. Volunteer number 5 has a dysferlin expression aver-age of 53.2 %, SD: 3.9. Volunteer 6 has a dysferlin expression average of43.9%, SD: 12.4. Volunteer 9 has a dysferlin expression average of 51.6%,SD: 12.2. Volunteer 14 has a dysferlin expression average of 46.7%, SD:9.5. Volunteer 16 has a dysferlin expression average of 51.4%, SD: 17.14.Volunteer 18 has a dysferlin expression average of 56.2%, SD: 11. Volun-teer 23 has a dysferlin expression average of 58.9%, SD: 11. Volunteer27 has a dysferlin expression average of 40%, SD: 17.2. Volunteer 30has a dysferlin expression average of 61.2%, SD: 13.7. Volunteer 37has a dysferlin expression average of 53.3%, SD: 15.7. Volunteer 38 hasa dysferlin expression average of 47.7%, SD: 15.7. Volunteer 50 has adysferlin expression average of 64.1%, SD: 5.1. Volunteer 119 has a dys-ferlin expression average of 59.4%, SD: 8.4. Volunteer 125 has a dysferlinexpression average of 55%, SD: 2.6. Volunteer 132 has a dysferlin expres-sion average of 50.6%, SD: 2.1. Volunteer 143 has a dysferlin expressionaverage of 54.8%, SD: 13.7. Volunteer 145 has a dysferlin expression av-erage of 62.6%, SD: 3.3. Volunteer 146 has a dysferlin expression averageof 54%, SD: 11.1.

Mutation Analysis of Predicted Carriers

Genetic analysis of the dysferlin gene was performed in 14 of the18 volunteers who had dysferlin protein levels in the carrier range.Genetic analysis of participants 14 and 132 could not be completedbecause sufficient DNA was not available. Molecular analysis ofDYSF was not performed on participants 145 and 146 because theyare first-degree relatives of participants 6 and 38, and likely share anallele. Since no pathogenic DYSF variants were detected in partici-pants 6 or 38, we reasoned that the same result would be found inparticipants 145 and 146.

992 HUMAN MUTATION, Vol. 35, No. 8, 990–997, 2014

Table 1. Molecular Analysis of DYSF in Volunteers with Reduced Levels of Dysferlin

Volunteer Variant UMD pathogenicity scorea Seen in other patient/s Additional information

5 c.463G>A (p.G155R) 82 (Pathogenic) 16 None9 None14 N/D Not enough DNA to sequence.16 None18 None23 None27 c.509C>A (p.A170E) 71 (Probably pathogenic) 4/1

c.5626G>A (p.D1876N) 53 (Probably polymorphism)30 c.509C>A (p.A170E) 71 (Probably pathogenic)37 c.3191 3196dupCGGAGG N/D 1 Found in 1 other patient who has a second pathogenic mutation.38 None50 None119 c.3191 3196dupCGGAGG N/D 1 Found in 1 other patient who has a second pathogenic mutation.125 None132 N/D Not enough DNA to sequence.143 c.509C>A (p.A170E) 71 (Probably pathogenic) 4145 N/D Not sequenced child of 6 who had no detectable pathogenic mutations.146 N/D Not sequenced sibling of 38 who had no detectable pathogenic mutations.

aUsing the pathogenicity prediction algorithm, UMD-DYSF predictor score [Blandin et al., 2012] any missense variant with a score of 65 or higher is considered pathogenic. Thereference sequence used for sequence analysis was NM_003494.3. Nucleotide number 1 corresponds to the “A” of the translation initiation codon ATG of the reference sequence.All variants have been submitted to the NCBI ClinVar database (see Supp. Table S2 for ClinVar Ids for each variant).N/D; no data.

Several polymorphisms in DYSF were found in the 14 volunteersanalyzed (Supp. Table S3), but the vast majority did not have de-tectable changes that were likely to be pathogenic or explain whythey have reduced dysferlin protein levels by the blood monocytetest. In 4/14 samples, possible pathogenic missense mutations inDYSF were detected (Table 1).

To predict the pathogenicity of the missense mutations, we usedan algorithm specifically designed for the dysferlin gene (UMD-DYSF predictor score) [Blandin et al., 2012]. This algorithm takesinto account the position of the variant in functional domains,and the conservation of the affected amino acid. Using this tool,any missense variant with a score of 65 or higher is consideredpathogenic. In volunteer 5, we identified a heterozygous missensevariant c.463G>A (p.G155R) in exon 6 of the DYSF gene. The UMD-pathogenicity predictor score for this variant is 82, indicating that itis likely pathogenic [Blandin et al., 2012]. In addition, this varianthas been reported in a dysferlinopathy patient in the Leiden database(www.dmd.nl), supporting the notion that the p.Gly155Arg variantis likely to be pathogenic [Krahn et al., 2008].

In volunteers 27, 30, and 143, we identified a c.509C>A (p.A170E)variant in heterozygous state with or without a second variant (Ta-ble 1). The p.Ala170Glu variant has been reported in eight differentindividuals, all with dysferlinopathy and the UMD-predictor scoreis 71, suggesting that the variant is likely to be pathogenic (UMD-DYSF database). However, this variant has a minor allele frequency(MAF) of >1% in the dbSNP database and EVS. If the variant werereally pathogenic, given the high frequency in the general popula-tion (>1%), we would expect to see the variant at a much higherfrequency in dysferlinopathy patients than we currently do. Variantsreported in dbSNP, EVS, locus-specific databases, or EGL databaseat a population frequency higher than expected, given the preva-lence of the disease and mode of inheritance, are generally classifiedas benign variants. Taken together, the p.Ala170Glu variant appearsto be a benign variant. The conflicting information on this variantmakes it difficult to determine whether the p.Ala170Glu is trulypathogenic.

Depending on whether the p.Ala170Glu variant is pathogenic,the genetic analysis of the volunteers in this study suggests a carrierfrequency between 0.7% (1/149) and 2.7% (4/149) (Fig. 3).

Figure 3. Graphic representation of dysferlin expression in the pop-ulation studied percentage of volunteers with normal levels of dysferlin(131/149; 87.9%), reduced levels without mutation (10/149; 6.7%), reducedlevels with mutation (4/149; 2.7%), and reduced levels with no geneticstudy (4/149; 2.7%). Overall, 12.1% (18/149) of volunteers showed reducedlevels of dysferlin.

DNA Methylation Analysis

Next, we assessed the DNA methylation status of the DYSF locusin order to determine whether the reduced DYSF levels in carrierand disease range had an underlying epigenetic mechanism. Weanalyzed the DNA methylation profiles of the three CpG islands(defined following the Takai and Jones (2002) criteria) by bisulfitesequencing. Bisulfite conversion of DNA relies on the deaminationof unmethylated cytosines into uracils, while methylated cytosinesare protected from deamination. Bisfulfite sequencing analysis inDYSF locus revealed that the first two CpG islands (CpGi64 andCpGi93) are completely unmethylated in all samples, meanwhilethe CpGi33 (the one in an intronic location) is highly methylatedalso in all analyzed samples (Figs. 4a and 5a). Unmethylated CpGislands are frequently associated with chromatin transcriptionallyactive, and CpGi64 and CpGi93 cover the two annotated TSSs,suggesting an important role for these two CpG islands regulatingDYSF expression. However, no DNA hypermethylation has been


Figure 4. DNA methylation analysis of dysferlin locus in carriers with reduced levels of dysferlin compared to volunteers with normal expression.Genomic structure for dysferlin locus including annotated CpG islands (CpGi) and transcription start sites (TSSs). CpGi are named according tothe number of CpG sites they contain and methylation levels were analyzed by bisulfite sequencing. Each CpG is represented by a dot that canbe unmethylated (white), partially methylated (light, intermediate, or dark gray depending of methylation degree), and heavily methylated (black);X corresponds to not determined. Each square represents an analyzed region and on the top is indicated the relative position to the first TSS. A:DNA methylation analysis in regions I1, I2, and I3, all inside the CpG islands. B: DNA methylation analysis in regions S1, S2, S3, and S4, all inside toCpG island shores. C: DNA methylation analysis in R1 and R2 inside the proximal regulatory region. Samples correspond as follow: N1–N7 showednormal levels of dysferlin. V125 and V23 showed 50% of dysferlin levels without pathogenic mutations. P1–P4 showed dysferlin levels lower than10% with only a pathogenic mutation.

found in individuals with lower expression levels of DYSF proteinaccording to its genetic profiles (Figs. 4a and 5a).

Next, we took advantage of the ENCODE data inCD14+/monocyte cells to look at histone modification profiles inDYSF promoter regions (ENCODE Histone Modification Tracksby ChIP-seq from ENCODE/Broad Institute; http://genome-euro.ucsc.edu/cgi-bin/hgGateway). The levels of positive marks(H3K4me3 and H3K9Ac) in both promoters were very high and nosignal for the heterocromatic histone mark H3K9me3 was detected(data not shown). However and very interestingly, there is a differ-ent pattern for the H3K27me3 mark depending on the promoter,showing this inactive mark only in the second promoter/TSS. Thisresult could suggest that this could be an alternative promoterof the DYSF gene being poised to be activated under certainconditions.

To extend the analysis of DNA methylation, we examined themethylation state of the named CpG island shores (CpG dinu-cleotides located in the sequence next to the CpG island), sincechanges in CpG island shore methylation have been correlated withchanges in gene expression [Doi et al., 2009; Irizarry et al., 2009]. Asshown in regions S1, S2, S3, and S4 in Figures 4b and 5b, there werealso no differences in the methylation levels of CpG island shorebetween samples, suggesting that DNA hypermethylation was notinvolved in DYSF downregulation in these individuals.

Finally, we examined DNA methylation levels of CpGs located atthe proximal regulatory region of DYSF gene [Foxton et al., 2004].The methylation profile of CpG poor regions can be very variable,following the idea that its methylation status can be closely con-nected with the regulatory function. The results showed that themost distal CpGs were completely methylated (R1 region), mean-while the two closer to the TSS1 were totally unmethylated (R2region in Figs. 4c and 5c), although no differences were found be-tween samples.

All together, these results provide evidences that DNA methy-lation seems not to play a role regulating DYSF expression andstrongly suggest a methylation-independent mechanism involvedin DYSF pathogenic silencing.

DiscussionAccording to some rough estimates, the prevalence of dysfer-

linopathies is around 1–5 per million, which would predict a carrierfrequency of around 1 in 100 [Norwood et al., 2009]. However,based on the number of healthy individuals with reduced dysferlinlevels that we previously identified by the monocyte assay fromamong those volunteering as test controls, we suspected that thecarrier frequency could be higher than estimated. To investigate this


Figure 5. DNA methylation analysis of dysferlin locus in carriers with monoallelic mutation and patients with biallelic mutations. Genomicstructure for dysferlin locus including annotated CpG islands (CpGi) and transcription start sites (TSSs). CpGi are named according to the numberof CpG sites they contain and methylation levels were analyzed by bisulfite sequencing. Each CpG is represented by a dot that can be unmethylated(white), partially methylated (light, intermediate, or dark gray depending of methylation degree), and heavily methylated (black); X corresponds tonot determined. Each square represents an analyzed region and on the top is indicated the relative position to the first TSS. A: DNA methylationanalysis in regions I1, I2, and I3, all inside the CpG islands. B: DNA methylation analysis in regions S1, S2, S3, and S4, all inside to CpG island shores.C: DNA methylation analysis in R1 and R2 inside the proximal regulatory region. Samples correspond as follow: C1–C7 and V119 are carriers withmonoallelic mutation showing 50% of dysferlin levels. P5–P7 are patients (0%–10% of dysferlin levels) with biallelic mutations.

further, we conducted a clinical study to determine dysferlin mu-tation carrier frequency in a sample population using the dysferlinmonocyte assay for initial screening, followed by sequencing forconfirmation.

In one of our previous studies, we reported that expression levelsof dysferlin in PBM cells of genetically confirmed carriers (onemutation in the DYSF gene identified) range from 24.5% to 78.2%with an average of 51.2 ± 15.4% and that dysferlin expression levelslower than 75% of normal are indicative of carrier status with asensitivity of 97.8% and a specificity of 97.1% [Luna et al., 2012].A different study looking at dysferlin protein expression in wholePBM cell lysates also reported a similar range for dysferlinopathycarriers [Ankala et al., 2014]. However, in our current study, eventhough 12.1% (18/149) of the study participants (clinically healthyindividuals) showed dysferlin levels in the carrier range, subsequentmolecular analysis by sequencing detected missense variants in theDYSF gene in only 2.7% (4/149) of them. This could be explainedby the presence of other types of mutations such as exonic deletionsor duplications, as well as intronic regulatory and promoter specificmutations, which were not evaluated due to their low prevalencein the patient population. These types of dysferlin mutations arespeculated to be the source of the second pathogenic mutation inthe �20% of cases of clinically diagnosed dysferlinopathy in whichonly one bona fide mutation have been found [De Luna et al.,2007; Krahn et al., 2008]. However, the only reported study thatinterrogated potential mutations in the promoter region of DYSFdid not detect any mutations [Foxton et al., 2004].

DNA methylation is an epigenetic mechanism involved in tran-scriptional regulation, and CpG methylation is inversely correlated

with gene expression. Nowadays, there is no doubt about the se-vere effects of abnormal CpG island methylation in diseases, forexample, in cancer, where it affects repression of tumor suppres-sor genes [Jones and Baylin, 2007; Irizarry et al., 2009; Fernandezet al., 2012]. To address the possibility that DYSF expression couldbe repressed/regulated by DNA methylation, we analyzed for thefirst time the methylation levels of CpG islands and shores in theDYSF gene by bisulfite sequencing. Our results, however, showed nodifferences in DNA methylation profiles in none of the three CpGislands, CpG island shores, and neither in the proximal regulatoryregion between healthy people, mutated, and nonmutated carriersand monoallelic and biallelic mutated patients. These data provideclear evidences that DNA methylation seems not to play a pivotalrole regulating DYSF expression and would suggest a methylation-independent mechanism involved in dysferlin downregulation.

An alternative explanation could be the existence of pathogenicchanges in other genes and factors that contribute to the reduc-tion in dysferlin expression. For instance, secondary reductions inmuscle dysferlin levels have been observed in patients with muta-tions in calpain 3 (LGMD2A) and caveolin-3 (LGMD1C) [Matsudaet al., 2001; Hermanova et al., 2006]. However, since calpain-3 andcaveolin-3 are not expressed in monocytes, a secondary reductionof dysferlin in these cells cannot be explained by a molecular defectin these two genes. Because dysferlin levels can be susceptible tosecondary reductions, it might be that variants in genes expressedin monocytes could cause the reduction of dysferlin levels seen inthis study. In addition, we have recently shown that vitamin D levelsregulate dysferlin expression, both in skeletal muscle primary cul-tures and in PBMs [Luna et al., 2012]. However and unfortunately,


vitamin D analysis could not be performed on these individualsbecause serum samples were not collected.

Finally, in the last few years, the role of miRNAs as regulatorsof gene expression is an emerging field of research. In fact, thereare several papers demonstrating modulation of gene expressionin monocytes by miRNAs [Wang et al., 2009; Xie et al., 2013]. Itwould be interesting to study a possible effect of specific miRNAson dysferlin expression in monocytes and skeletal muscle because itcould also have therapeutic implications.

Another important observation from this study is the high in-cidence (12.1%) of reduced PBM dysferlin levels in an other-wise healthy general population. Although the absence of clearlypathogenic variants (i.e., frameshift or nonsense mutations) andthe limited number of volunteers makes it difficult to do a defini-tive estimation about the carrier frequency of dysferlinopathy, ourresults would indicate that the carrier frequency may be higher thanpreviously estimated (between 1% and 2.7%). If this is true, thenthe disease incidence may also be higher than previously estimated.This would not be surprising since dysferlinopathy patients havefrequently been misdiagnosed with polymyositis [Vinit et al., 2010;Angelini et al., 2011] and may be confused with other degenerativemuscle diseases. In addition, the high cost and limited access to dys-ferlin molecular analysis leave a large majority of dysferlinopathywith a diagnosis of undefined limb-girdle muscular dystrophy.

Finally, previous studies from our laboratory clearly show thatthe monocyte assay can successfully aid the diagnosis of dysfer-linopathies. Yet, the current study strongly demonstrates that theblood monocyte test is not able to reliably identify carriers of dys-ferlin mutations in the general population. Understanding the limitsof the blood monocyte assay has important clinical implications forcarrier status interpretation, as well as for genetic counseling infamilies with dysferlinopathy. Given that there is some variability ofdysferlin expression in healthy population, addition of more thanone control in every test may improve its reliability. Hence, evalu-ation of dysferlin protein levels in monocytes is a meaningful firststep to identify carriers of mutations in families with dysferlinopa-thy, though molecular confirmation is always essential.

Acknowledgments

We would like to thank healthy blood donors and Dr. A. Bosch and techni-cians (Jenny and Isabel) at the blood bank (BST) in Hospital de Sant Pau fortheir kindness and technical support.

Disclosure statement: The authors declare no conflict of interest.

References

Anderson LV, Davison K, Moss JA, Young C, Cullen MJ, Walsh J, Johnson MA, Bashir R,Britton S, Keers S, Argov Z, Mahjneh I, et al. 1999. Dysferlin is a plasma membraneprotein and is expressed early in human development. Hum Mol Genet 8:855–861.

Angelini C, Grisold W, Nigro V. 2011. Diagnosis by protein analysis of dysferlinopathyin two patients mistaken as polymyositis. Acta Myol 30:185–187.

Ankala A, Nallamilli BR, Rufibach LE, Hwang E, Hegde MR. 2014. Diagnosticoverview of blood-based dysferlin protein assay for dysferlinopathies. MuscleNerve. 10.1002/mus.24195.

Bansal D, Miyake K, Vogel SS, Groh S, Chen CC, Williamson R, McNeil PL, CampbellKP. 2003. Defective membrane repair in dysferlin-deficient muscular dystrophy.Nature 423:168–172.

Baylin SB, Jones PA. 2011. A decade of exploring the cancer epigenome–biological andtranslational implications. Nat Rev Cancer 11:726–734.

Bird A. 2002. DNA methylation patterns and epigenetic memory. Genes Dev 16:6–21.Blandin G, Beroud C, Labelle V, Nguyen K, Wein N, Hamroun D, Williams B, Monnier

N, Rufibach LE, Urtizberea JA, Cau P, Bartoli M, et al. 2012. UMD-DYSF, a novellocus specific database for the compilation and interactive analysis of mutationsin the dysferlin gene. Hum Mutat 33:E2317–E2331.

Clark SJ, Statham A, Stirzaker C, Molloy PL, Frommer M. 2006. DNA methylation:bisulphite modification and analysis. Nat Protoc 1:2353–2364.

De Luna N, Freixas A, Gallano P, Caselles L, Rojas-Garcia R, Paradas C, Nogales G,Dominguez-Perles R, Gonzalez-Quereda L, Vilchez JJ, Marquez C, Bautista J, et al.2007. Dysferlin expression in monocytes: a source of mRNA for mutation analysis.Neuromuscul Disord 17:69–76.

de Luna N, Gallardo E, Soriano M, Dominguez-Perles R, de la Torre C, Rojas-Garcia R,Garcia-Verdugo JM, Illa I. 2006. Absence of dysferlin alters myogenin expressionand delays human muscle differentiation "in vitro". J Biol Chem 281:17092–17098.

de Morree A, Flix B, Bagaric I, Wang J, van den Boogaard M, Grand Moursel L, FrantsRR, Illa I, Gallardo E, Toes R, van der Maarel SM. 2013. Dysferlin regulates celladhesion in human monocytes. J Biol Chem 288:14147–14157.

Doi A, Park IH, Wen B, Murakami P, Aryee MJ, Irizarry R, Herb B, Ladd-Acosta C, RhoJ, Loewer S, Miller J, Schlaeger T, et al. 2009. Differential methylation of tissue-and cancer-specific CpG island shores distinguishes human induced pluripotentstem cells, embryonic stem cells and fibroblasts. Nat Genet 41:1350–1353.

Fazzari MJ, Greally JM. 2004. Epigenomics: beyond CpG islands. Nat Rev Genet 5:446–455.

Fernandez AF, Assenov Y, Martin-Subero JI, Balint B, Siebert R, Taniguchi H, Ya-mamoto H, Hidalgo M, Tan AC, Galm O, Ferrer I, Sanchez-Cespedes M, et al. 2012.A DNA methylation fingerprint of 1628 human samples. Genome Res 22:407–419.

Foxton RM, Laval SH, Bushby KM. 2004. Characterisation of the dysferlin skeletalmuscle promoter. Eur J Hum Genet 12:127–131.

Frederic MY, Lalande M, Boileau C, Hamroun D, Claustres M, Beroud C, Collod-Beroud G. 2009. UMD-predictor, a new prediction tool for nucleotide substitutionpathogenicity – application to four genes: FBN1, FBN2, TGFBR1, and TGFBR2.Hum Mutat 30:952–959.

Gallardo E, de Luna N, Diaz-Manera J, Rojas-Garcia R, Gonzalez-Quereda L, Flix B,de Morree A, van der Maarel S, Illa I. 2011. Comparison of dysferlin expressionin human skeletal muscle with that in monocytes for the diagnosis of dysferlinmyopathy. PLoS One 6:e29061.

Gardiner-Garden M, Frommer M. 1987. CpG islands in vertebrate genomes. J Mol Biol196:261–282.

Goll MG, Bestor TH. 2005. Eukaryotic cytosine methyltransferases. Annu Rev Biochem74:481–514.

Hermanova M, Zapletalova E, Sedlackova J, Chrobakova T, Letocha O, Kroupova I,Zamecnik J, Vondracek P, Mazanec R, Marikova T, Vohanka S, Fajkusova L. 2006.Analysis of histopathologic and molecular pathologic findings in Czech LGMD2Apatients. Muscle Nerve 33:424–432.

Ho M, Gallardo E, McKenna-Yasek D, De Luna N, Illa I, Brown Jr RH. 2002. A novel,blood-based diagnostic assay for limb girdle muscular dystrophy 2B and Miyoshimyopathy. Ann Neurol 51:129–133.

Illa I, De Luna N, Dominguez-Perles R, Rojas-Garcia R, Paradas C, Palmer J, MarquezC, Gallano P, Gallardo E. 2007. Symptomatic dysferlin gene mutation carriers:characterization of two cases. Neurology 68:1284–1289.

Irizarry RA, Ladd-Acosta C, Wen B, Wu Z, Montano C, Onyango P, Cui H, Gabo K,Rongione M, Webster M, Ji H, Potash JB, et al. 2009. The human colon cancermethylome shows similar hypo- and hypermethylation at conserved tissue-specificCpG island shores. Nat Genet 41:178–186.

Jones PA. 2012. Functions of DNA methylation: islands, start sites, gene bodies andbeyond. Nat Rev Genet 13:484-492.

Jones PA, Baylin SB. 2007. The epigenomics of cancer. Cell 128:683–692.Krahn M, Beroud C, Labelle V, Nguyen K, Bernard R, Bassez G, Figarella-Branger D,

Fernandez C, Bouvenot J, Richard I, Ollagnon-Roman E, Bevilacqua JA, et al.2008. Analysis of the DYSF mutational spectrum in a large cohort of patients.Hum Mutat 30:E345–E375.

Li LC, Dahiya R. 2002. MethPrimer: designing primers for methylation PCRs. Bioin-formatics 18:1427–1431.

Liu J, Aoki M, Illa I, Wu C, Fardeau M, Angelini C, Serrano C, Urtizberea JA, HentatiF, Hamida MB, Bohlega S, Culper EJ, et al. 1998. Dysferlin, a novel skeletal musclegene, is mutated in Miyoshi myopathy and limb girdle muscular dystrophy. NatGenet 20:31–36.

Luna ND, Diaz-Manera J, Paradas C, Iturriaga C, Rojas-Garcia R, Araque J, Genebri-era M, Gich I, Illa I, Gallardo E. 2012. 1alpha,25(OH)(2)-Vitamin D3 increasesdysferlin expression in vitro and in a human clinical trial. Mol Ther 20:1988–1997.

Matsuda C, Hayashi YK, Ogawa M, Aoki M, Murayama K, Nishino I, Nonaka I, ArahataK, Brown RH, Jr. 2001. The sarcolemmal proteins dysferlin and caveolin-3 interactin skeletal muscle. Hum Mol Genet 10:1761–1766.

Nagaraju K, Rawat R, Veszelovszky E, Thapliyal R, Kesari A, Sparks S, Raben N, PlotzP, Hoffman EP. 2008. Dysferlin deficiency enhances monocyte phagocytosis: amodel for the inflammatory onset of limb-girdle muscular dystrophy 2B. Am JPathol 172:774–785.

Norwood FL, Harling C, Chinnery PF, Eagle M, Bushby K, Straub V. 2009. Prevalenceof genetic muscle disease in Northern England: in-depth analysis of a muscle clinicpopulation. Brain 132:3175–3186.

Portela A, Esteller M. 2010. Epigenetic modifications and human disease. Nat Biotechnol28:1057–1068.


Sharma S, Kelly TK, Jones PA. 2010. Epigenetics in cancer. Carcinogenesis 31:27–36.Takai D, Jones PA. 2002. Comprehensive analysis of CpG islands in human chromo-

somes 21 and 22. Proc Natl Acad Sci USA 99:3740–3745.Vinit J, Samson M, Jr., Gaultier JB, Laquerriere A, Ollagnon E, Petiot P, Marie

I, Levesque H, Rousset H. 2010. Dysferlin deficiency treated like refractorypolymyositis. Clin Rheumatol 29:103–106.

Wang B, Yang Z, Brisson BK, Feng H, Zhang Z, Welch EM, Peltz SW, Barton ER, BrownRH, Jr., Sweeney HL. 2009. Membrane blebbing as an assessment of functionalrescue of dysferlin-deficient human myotubes via nonsense suppression. J ApplPhysiol 109:901–905.

Xie W, Li M, Xu N, Lv Q, Huang N, He J, Zhang Y. 2013. MiR-181a regulates inflam-mation responses in monocytes and macrophages. PLoS One 8:e58639.


Genetic and Epigenetic Determinants of Low Dysferlin Expression in Monocytes

Documents