Nonsense-mediated decay and the molecular pathogenesis of mutations in SALL1 and GLI3

� 2007 Wiley-Liss, Inc. American Journal of Medical Genetics Part A 143A:3150–3160 (2007)

Nonsense-Mediated Decay and the MolecularPathogenesis of Mutations in SALL1 and GLI3

Dominic Furniss,1,2 Paul Critchley,2 Henk Giele,2 and Andrew O.M. Wilkie1*1Weatherall Institute of Molecular Medicine, University of Oxford, John Radcliffe Hospital, Oxford, United Kingdom

2Department of Plastic and Reconstructive Surgery, Oxford Radcliffe Hospitals NHS Trust, The West Wing,John Radcliffe Hospital, Oxford, United Kingdom

Received 12 June 2007; Accepted 27 August 2007

Mutations in SALL1 and GLI3 are responsible for human limbmalformation syndromes. The molecular pathophysiology ofthese mutations is incompletely understood, and manyconclusions have been drawn from studies performed inthe mouse. We identified truncating mutations in SALL1 andGLI3 in patients with limb malformation and studied thecontribution of nonsense-mediated decay (NMD) to theexpression of mutant mRNA in patient-derived fibroblasts.Quantification of the relative proportions of mutant andwild-type alleles was performed by pyrosequencing. InSALL1, a mutant allele causing Townes–Brocks syndromewas unexpectedly resistant to NMD, whereas a different

mutation causing a much milder phenotype was susceptibleto NMD. In GLI3, all three mutant alleles tested weresusceptible to NMD. This work provides novel insights intothe molecular pathophysiology of SALL1 and GLI3 muta-tions, extends the phenotypic spectrum of SALL1 mutations,and provides an example of a human mutation which doesnot follow the usual accepted positional rules governingmammalian NMD. � 2007 Wiley-Liss, Inc.

Key words: SALL1; GLI3; nonsense-mediated decay; muta-tions; pathophysiology; pyrosequencing

How to cite this article: Furniss D, Critchley P, Giele H, Wilkie AOM. 2007. Nonsense-mediated decayand the molecular pathogenesis of mutations in SALL1 and GLI3. Am J Med Genet Part A 143A:3150–3160.

INTRODUCTION

SALL1 and GLI3 are genes encoding zinc-fingertranscription factors that have been implicated in thepathogenesis of dominantly inherited human limbmalformations. Heterozygous intragenic truncatingmutations in SALL1 causeTownes–Brocks syndrome(TBS), which is characterized by abnormalities of theanus, radial ray, ear, heart, kidney and genitourinarysystem, and rarely the brain [Kohlhase et al., 1998;Botzenhart et al., 2007]. Three complete SALL1deletions have also been described; however, thephenotype was mild compared to classical TBS[Borozdin et al., 2006].

Heterozygous mutations in GLI3 have beenreported in Greig cephalopolysyndactyly (GCPS),characterized by postaxial polydactyly of the hands,preaxial polydactyly of the feet, syndactyly, macro-cephaly, and hypertelorism [Vortkamp et al., 1991;Wild et al., 1997]; Pallister–Hall syndrome (PHS),characterized by central polydactyly of the hands,hypothalamic hamartoma, and other features includ-ing postaxial polydactyly of the hands, imperforateanus, and cleftingof the larynx [Kang et al., 1997]; and(rarely) acrocallosal syndrome, characterized bypostaxial polydactyly of the hands, preaxial poly-dactyly of the feet, and agenesis of the corpus

callosum [Elson et al., 2002]. Patients with GLI3mutations and limb malformations who lack othersyndromic manifestations have been classified withpostaxial polydactyly types A and B (PAP-A, PAP-B),characterized by the presence of a well formed extradigit on the ulnar border of the hand, with or withoutan extra metacarpal (type A), or a rudimentary extradigit (typeB) [Radhakrishna et al., 1997]; andpreaxialpolydactyly type IV, characterized by preaxialpolydactyly of the hands and feet, with simplesyndactyly of the 3rd web space of the hands andthe 2nd web space of the feet [Radhakrishna et al.,1999].

For both genes, genotype–phenotype correlationsin mouse and human give clues to the molecularpathogenesis of the underlying mutations. For exam-ple, the original Sall1 heterozygous knockout mousedid not phenocopy human TBS [Nishinakamura

Grant sponsor: Wellcome Trust.*Correspondence to: Andrew O.M. Wilkie, Weatherall Institute of

Molecular Medicine, University of Oxford, John Radcliffe Hospital,Headington, Oxford OX3 9DS, UK.E-mail: [email protected]

DOI 10.1002/ajmg.a.32097

et al., 2001], but a mouse heterozygous for a mutantallele, that produces a truncated protein, recapitu-lated the abnormalities found in human TBS [Kieferet al., 2003]. This suggests that an abnormal truncatedSALL1 protein (acting in a dominant negative or gain-of function fashion) is responsible for the fullspectrum of developmental defects seen in TBS,whereas haploinsufficiency for SALL1 causes amilder TBS-like phenotype. A more distinct patternof genotype–phenotype correlation for truncatingmutations of GLI3 was reported by Johnston et al.[2005]. Those truncations which occurred eitherwithin the first 1997 nucleotides of the GLI3 codingsequence, or after nucleotide 3481, caused GCPS,whilst those which occurred between these twonucleotide positions caused PHS, with the notableexception of a recurrent mutation 2374C>T(R792X), which was associated with GCPS. Trans-fection of mutant cDNA constructs into cell linessuggested that GCPS is caused by functional hap-loinsufficiency of GLI3, whereas in PHS and PAP-A/B, abnormal truncated GLI3 protein acts in adominant negative manner [Shin et al., 1999].

In studying the pathogenesis of truncating muta-tions, an important factor that may contribute tophenotype variability is the extent to which themutant transcripts may be unstable, thus reducingthe production of mutant truncated proteins. Severalmechanisms exist within cells to ensure that onlycorrectly processed and error-free mRNA is trans-lated. These methods include a nuclear qualitycontrol mechanism which degrades improperlyprocessed mRNA, and nucleo-cytoplasmic mecha-nisms which target transcripts containing no termi-nation codon (nonstop-mediated decay), or apremature termination codon (PTC; nonsense-medi-ated decay [NMD]) for degradation. These processesprotect the cell from the accumulation of potentiallytoxic aberrant proteins [Fasken and Corbett, 2005;Khajavi et al., 2006].

The effects of NMD are likely to be significant in thepathogenesis of genetic disease, because approxi-mately one-third of all point mutations that causegenetic disorders are frameshift or nonsensemutations[Mendell and Dietz, 2001]. For example, unusual casesof dominantly inherited b-thalassemia were shown tobe related to the lack of mRNA instability when anonsense mutation lay in the terminal 30 exon of theHBB gene [Hall and Thein, 1994]. Similarly in SOX10,distinct neurological phenotypes are conveyed bytruncating mutations in the 50 or 30 portions of thetranscript, and the phenotypic variability is mediatedby differences in NMD: nonsense mutations in the 50

part of the gene are susceptible to NMD and cause amilderphenotype (Waardenburg–Shah syndrome)byhaploinsufficiency, whereas nonsense mutations inthe 30 part of the gene, which are resistant to NMD,cause a more complex neurocristopathy by a domi-nant negative mechanism [Inoue et al., 2004].

However, the contribution of NMD to molecularpathogenesis of truncating mutations either in SALL1or inGLI3 has not, to our knowledge, previously beenformally published.

In this report we describe a patient with a frame-shift mutation in SALL1who has an isolated unilaterallimb malformation without other organ systeminvolvement, thereby extending the phenotypicspectrum associated with SALL1 mutations. Further-more, we describe two novel mutations in GLI3 andthe results of studies to determine the relative levelsof expression of wild-type and PTC-containingmutant alleles of SALL1 and GLI3. These resultsprovide insight into the molecular pathophysiologyof SALL1 and GLI3 mutations, and raise furtherquestions with regard to the fundamental mecha-nism of NMD.

MATERIALS AND METHODS

Ascertainment of Patients

Approval for this work was obtained from theOxfordshire Research Ethics Committee C (C99.181).Patients were recruited from the pediatric handsurgery clinic at Oxford Radcliffe Hospitals NHSTrust between 1999 and 2006, as part of a larger studyinto the genetic basis of human limb malformations.All patients or families who presented to the clinicwith a congenital limb malformation requiringreconstructive surgery, and who gave informedconsent, entered the study. At operation, a maximumof 5 ml of venous blood was collected, from whichgenomicDNAwas isolatedusing phenol/chloroformextraction. Additionally, if any spare skin wasavailable from the surgery, this was collected inHank’s Balanced Salt Solution (HBSS, Cambrex BioScience, Verviers, Belgium) and used to set up afibroblast culture using complete medium.

Mutation Detection

Genomic DNA was amplified by PCR as previouslydescribed [Johnson et al., 2003]. All primer sequencesand reaction conditions are available on request.Genomic DNA was screened for heterozygous muta-tions using WAVE denaturing high performance liquidchromatography (dHPLC) (Transgenomic, Elancourt,France), and any abnormally eluting fragments weresubjected to DNA sequencing as previously described[Johnson et al., 2003]. Mutations were confirmed usingrestriction endonuclease digestion. Genbank acces-sion numbers for the reference sequences used wereNM_002968 (SALL1) and NM_000168 (GLI3); cDNAsare numbered starting at the A of the initiation codon.

Cycloheximide Treatment of Fibroblasts andRNA Extraction

For quantification of NMD, two 150 cm2 flasks offibroblast cells grown to confluence were obtained.

NONSENSE-MEDIATED DECAY IN SALL1 AND GLI3 3151

American Journal of Medical Genetics Part A: DOI 10.1002/ajmg.a

Mediumwaspoured from thefirst flask, and replacedby 100mgof cycloheximide dissolved in 10ml of a 1%dimethylsulfoxide solution in complete medium(final concentration of cycloheximide 10 ng/ml).The second flask was treated with complete mediumonly. Bothflaskswere then incubated at 378C for 4 hr,afterwhich themediumwaspoured frombothflasks.Cells were rinsed in sterile phosphate buffered saline(Cambrex Biowhitaker) and stored at �708C in 1 mlof TRIzol (Invitrogen, Paisley, UK). When required,samples were defrosted at room temperature anddivided into two aliquots for RNA extraction; 400 mlof chloroform was added to each aliquot andsamples were vortexed vigorously prior to centrifu-gation at 10,400g for 3 min at room temperature. Theupper aqueous phase was collected and RNAprecipitation performed using the RNeasy kit (QIA-GEN, Crawley, UK), following the manufacturer’sinstructions, from step 4 onwards. Both aliquots ofeach sample were processed through the samecolumn to allow repooling, and RNA was eluted ina final volume of 30 ml. The RNA concentration wasdetermined by spectrophotometry. Reverse tran-scription (RT)with randomdecamerswas performedon 1–2 mg of RNA using the RETROscript kit(Ambion/Applied Biosystems, Warrington, UK)according to the manufacturer’s instructions. Twomicroliters of the RT product was used in subsequentPCR amplification.

Quantification of NMD by Pyrosequencing

Pyrosequencing, performed on the PSQ-HS96Asystem (Biotage, Uppsala, Sweden), was used tomeasure cDNA expression levels of alleles contain-ing PTCs in SALL1 and GLI3 relative to the wild-typeallele, in the presence and absence of prior in-cubation with cycloheximide. PCR primer sequen-ces, annealing temperatures, sequencing primersequences, and nucleotide dispensation orders forthe mutations studied are shown in Table I. Forpyrosequencing performed in the forward direction,the reverse primer was biotinylated, and vice versa.Single stranded DNA was obtained from 10 ml of PCRproduct by immobilization on streptavidin-coatedsepharose beads (Streptavidin Sepharose high per-formance, GE Healthcare, Chalfont St. Giles, UK) andNaOH denaturation.

The peak heights were recorded using PSQ96 SQAsoftware and the values exported into an MS Exceldata sheet. Each pyrogram had to pass three qualitycontrol criteria to be included in the final calcu-lations. These were the absence of unexpectedmutant allele specific peaks in the control samples(mutant/wild-type ratio <0.1); no decrease in peakheight as the dispensation progressed, which wouldmake comparisons between early and late peaksinvalid; and no unexpected peaks from deliberatelyblank dispensations. For each pair of peaks chosenfor quantification, the mutant/wild-type peak ratio

was obtained. At least three independent experi-ments were performed for each mutation.

RESULTS

Identification of Mutations in SALL1 and GLI3

As part of a study of the prevalence of mutationsassociated with limb malformations requiring surgi-cal correction, we screened patients for mutations inSALL1 (N¼ 197) and GLI3 (N¼ 198) using WAVEdHPLC and DNA sequencing. In this report, we haveaddressed the pathogenesis of those patients whohad truncating mutations in either one of thesegenes, and in whom fibroblasts were available forfunctional analysis. Two patients with SALL1 muta-tions and three with GLI3 mutations matched thesecriteria.

A summary of the clinical features and mutationsidentified in the probands is shown in Table II. Theclinical features of patient OX3335 (TBS) [independ-ently reported by Botzenhart et al., 2007], andpatients OX2879 and OX3536 (GCPS), were typicalfor their respective disorders. Patient OX2948 pre-sented with isolated right hand preaxial polydactyly,classified as Wassel type 7 [Wassel, 1969] (Fig. 1A).There was no family history of congenital malforma-tion. DNA sequencing of SALL1 revealed themutation 3414_3415delAT, leading to a frameshift,and predicting the addition of 13 abnormal aminoacids before termination (Fig. 1B). The deletioncauses the addition of only 13 abnormal amino acidsbecause the patient is also homozygous for apreviously described single nucleotide [C/T] poly-morphism (SNP) at position 3456 (rs11645288)(Fig. 1C). The mutation was confirmed by BsaAIrestriction digest of blood and fibroblast derivedDNA,whichdid not showany evidenceofmosaicismfor the mutation (Fig. 1D).

Patient OX2877 presented with isolated bilateralPAP-B of the hands (Fig. 2A). A family history of PAP-B, stretching back at least three generations waselicited (Fig. 2B). The affected father (II-2) hadunilateral PAP-B, and individual II-3, who is pre-dicted to carry the mutation, was reported to beunaffected, demonstrating both incomplete pene-trance and variable expressivity of this phenotype.Radiographs of the hands of the proband and heraffected father did not reveal any occult centralpolydactyly. There was no family history of othercongenital malformations, in particular there was nohistory of central nervous system lesions or ofunexplained death or miscarriage. DNA sequencingof GLI3 revealed the mutation 2372delC, predictingthe addition of two abnormal amino acids beforetranslation prematurely terminates (Table II).

Reverse transcriptase-PCR was performed for eachmutation, and provided no evidence of abnormalsplice variants produced by any of the mutations

3152 FURNISS ET AL.


TABLE

I.Pyro

sequenci

ng

Prim

ers

and

Conditio

ns

Gene

Muta

tion

Direct

ion

Am

plifica

tion

prim

ers

(50 !

30 )

PCR

Sequen

cing

prim

er

Forw

ard

Revers

e

Anneal

ing

tem

pera

ture

(8C)

Additiv

e

SALL1

995del

CForw

ard

ACAG

AG

CCTCG

CCAG

CCAAT-

(Bio

tin)-

TTTCAG

AG

GACG

GG

GTG

GTA-

64

CAACACCATCATTC

CTG

CCAG

CAT

ACTG

CCG

CTG

995del

CRevers

e(B

iotin)-

Fa

Ra

64

AAG

AG

CCG

CTG

TTG

GATG

3414_

3415delA

TForw

ard

CTG

TCTTCCTCTG

CCACATC-

(Bio

tin)-

TAG

AAATG

TCATG

GG

GCCAT-

64

CCCAAG

CAG

CACTACTG

CAA

CCCAG

TTCTG

CTCCC

CCACAG

AG

AG

C3414_

3415delA

TRevers

e(B

iotin)-

GTCTG

GG

CCTCTG

TCTTCCT-

Ra

64

CG

ATG

AG

GAG

AAG

GTTTTG

CCTG

CCACATC

GLI3

1320-2

1insT

Forw

ard

GCAG

CACTG

GCG

ACCCTG

CG

-(B

iotin)-

TG

TTTG

CTTCG

GTCTTTG

TC-

56.2

CCAAG

ATCAAACCCG

AT

CACAACAAG

AG

CCCTTCCTCCTTG

AC

1320-2

1insT

Revers

e(B

iotin)-

Fa

Ra

56.2

GG

GCTG

GG

GAG

GTCTTCA

2372d

elC

Forw

ard

CCCG

GCAG

GG

ACCAAATG

GA-

(Bio

tin)-

GCAG

GTG

TTG

TTG

GACTG

TG

-66.8

DM

SOb

ACAAG

TG

AATG

GAATG

TTTC

TG

GAG

CACG

TTG

CCATTTCCTATG

2372d

elC

Revers

e(B

iotin)-

GG

ACTCAACCATTTCCACTG

-G

GTG

TTG

TTG

GACTG

TG

TG

C-

61.6

DM

SOb

AG

AATG

GG

GTTCAG

TCG

CG

CAACCACAG

CCATTTCCTATG

2374C>

TForw

ard

CCCG

GCAG

GG

ACCAAATG

GA-

(Bio

tin)-

GCAG

GTG

TTG

TTG

GACTG

TG

-66.8

DM

SOb

AACAAG

TG

AATG

GAATG

TT

TG

GAG

CACG

TTG

CCATTTCCTATG

2374C>

TRevers

e(B

iotin)-

GG

ACTCAACCATTTCCAG

AG

-G

GTG

TTG

TTG

GACTG

TG

TG

C-

61.6

DM

SOb

GG

GTAG

AATG

GG

GTTCAG

CAACCACAG

CCATTTCCTATG

aSa

me

asForw

ard

(F)orre

vers

e(R

)se

quence

show

nin

colu

mn

imm

edia

tely

above.

bD

imeth

ylsu

lfoxid

e(1

0%

final

conce

ntrat

ion).


(data not shown). Furthermore, the creation ofneither donor nor acceptor splice sites was predictedby the neural network splice site prediction program(http://www.fruitfly.org/seq_tools/splice) for any ofthe mutations. Figure 3 shows a schematic of thegene structure of SALL1 and GLI3, together with theposition at which each mutation is predicted totruncate the protein, in relation to key functionaldomains and theposition equivalent to the final exonjunction.

Quantification of NMD by Pyrosequencing

We used pyrosequencing to provide a highlyquantitative measure of the relative amounts ofwild-type and mutant alleles in cDNA samples[Langaee and Ronaghi, 2005]. Figure 4 shows typicalpyrograms (forward orientation) obtained fromSALL1 cDNA prepared from fibroblasts of patientsOX3335 (Fig. 4A) andOX2948 (Fig. 4B), respectively,with and without cycloheximide treatment, com-pared with control fibroblasts. There were nosignificant mutant-specific peaks in pyrograms fromthe control samples, in addition these pyrogramspassed all other quality control criteria (see MaterialsandMethods Section). In Figure 4A, theheights of themutant-specific and wild-type-specific peaks incDNA from OX3335 are similar, and are unchangedafter treatment with cycloheximide. This shows,surprisingly, that the SALL1 mutation 995delC is notsubject to NMD. In contrast, inspection of Figure 4Bshows that the initial proportion of mutant SALL1cDNA from patient OX2948 is only�40% of the wild-type allele, but treatment with cycloheximideincreases this proportion to almost 100%. These dataimply that the SALL1 transcript containing the3414_3415delAT mutation is subject to NMD. Wecorroborated these findings by pyrosequencing thesamples in the reverse orientation; Figure 5 summa-rizes these data.

A similar analysis was performed on fibroblastsderived from patients OX2879, OX2877 and OX3536,all of whom have PTC mutations in GLI3 (Fig. 6). Bycontrast with the variable findings for SALL1, eachGLI3 mutation was shown to undergo NMD to asimilar extent.

DISCUSSION

In this work, we have analyzed NMD in fibroblastsfrom two patients with truncating SALL1 mutationsand three patients with truncatingGLI3mutations. Infour of the five patients (one of two with SALL1mutations and all three with GLI3 mutations) wedemonstrated that the mutant allele was subject toNMD of similar magnitude (cDNA from mutant allelepresent at 40–50% of wild-type allele, restored tonear normal levels after cycloheximide treatment).Unexpectedly, however, in one patient with a SALL1mutation and typical clinical features of TBS,

TABLE

II.

Sequen

ceChan

ges

and

Phenoty

pes

ofPat

ients

With

Tru

nca

tingSA

LL1

andGLI3

Muta

tions

Gene

Pat

ient

IDSe

xa

Muta

tion

(exon)b

Pre

dic

ted

amin

oac

idch

ange

Phenoty

pe

Bilat

era

lFam

ily

his

tory

Num

ber

of

lim

bs

affe

cted

Han

ds

Feet

Oth

er

SALL1

OX3335

M995del

C(2

)P332H

fsX10

Pre

axia

lpoly

dac

tyly

—Im

perf

ora

tean

us,

rect

alat

resi

a,hyposp

adia

s,overf

old

ed

helice

s

Yes

No

2

SALL1

OX2948

F3414_

3415

delA

T(2

)C11

39W

fsX14

Pre

axia

lpoly

dac

tyly

,trip

hal

angeal

thum

b—

—N

oN

o1

GLI3

OX2879

M1320_

1321

insT

(9)

E441X

Pre

axia

lpoly

dac

tyly

Pre

axia

lpoly

dac

tyly

,sy

ndac

tyly

—Y

es

No

4

GLI3

OX2877

F2372delC

(14)

P791Rfs

X3

Post

axia

lpoly

dac

tyly

B—

—Y

es

Yes

2GLI3

OX3536

M2374C

>T

(14)

R792X

Post

axia

lpoly

dac

tyly

BPre

axia

lpoly

dac

tyly

Hypertelo

rism

Yes

Yes

4

aM

,m

ale;F,fe

mal

e.

bSA

LL1

com

prise

s3

exons

andGLI3

com

prise

s15

exons.

3154 FURNISS ET AL.


the mutant allele did not undergo NMD. Thisidentifies an additional mechanism that may influ-ence the phenotype associated with truncatingSALL1 mutations.

The two mutations that we studied in SALL1,namely 995delC (P332HfsX8) and 3414_3415delAT(C1139WfsX14), have both been recently describedin TBS [Botzenhart et al., 2007]. The 995delCmutation presented with a typical phenotype andis, in fact, the same case as reported by Botzenhartet al. [2007]. However, the phenotype of patientOX2948, who presented with an isolated unilateralpreaxial polydactyly and has not been reportedpreviously, is milder either than patients with thesame mutation reported by Botzenhart et al. [2007],or than the mild phenotype associated with hetero-

zygous SALL1 deletions [Borozdin et al., 2006]. Thiscase extends the phenotypic spectrum of SALL1mutations, and suggests that the 3414_3415delATmutation is associatedwith variable expressivity. Theparents of OX2948 were unaffected, but sampleswere not available for molecular analysis.

Interestingly, the 3414_3415delAT mutation inSALL1 has also been described in a 17-year-oldfemale with bilateral renal hypodysplasia withoutany extra-renal manifestations [Weber et al., 2006].The annotation of the mutation described by Weberet al. [2006] (T1138fs1152X) suggests that the stopcodon in this patient occurs at the same unusualposition as in patient OX2948, and thus that thedeletion is inherited in cis with the rare 3456T alleleof the rs11645288 SNP. The sequence context of the

FIG. 1. Phenotype and SALL1gene analysis in patientOX2948. A: Preaxial polydactyly of the right hand, note that the radial thumb is triphalangeal. B: Forward (uppertrace) and reverse sequencing of exon 2 reveals a heterozygous dinucleotide deletion, 3414_3415delAT. C: Reverse sequencing of exon 2 from patient OX2948 (uppertrace) and control (lower trace) shows a homozygous 3456C>T transition in the patient (arrow). D: BsaAI restriction digest of genomic DNA derived from blood andfibroblasts from patient OX2948 shows no evidence of mosaicism (similar relative intensities of normal and mutant fragments in the two different tissues). [Color figurecan be viewed in the online issue, which is available at www.interscience.wiley.com.]



3415_3415delAT mutation does not suggest anypredisposition to recurrent mutation, making it likelythat patient OX2948 and the patient described byWeber et al. [2006] share a common ancestor inwhom the deletion occurred in cis with the 3456Tallele. It will be important to determine whether theapparently identical mutation described by Botzen-hart et al. [2007] occurs in cis to this same unusualallele.

Of the three mutations inGLI3 that we studied, two(1320_1321insT and 2372delC) are newly described,whereas the 2374C>T is a recurrent mutationpreviously observed in at least six unrelated casesof GCPS [Johnston et al., 2005]. Both the 2372delCand 2374C>T mutations lie in the region of the GLI3protein in which truncating mutations are typicallyassociated with PHS [Johnston et al., 2005]. Interest-ingly the 2372delC mutation was associated with aparticularly mild phenotype of PAP-B that exhibitedboth incomplete penetrance and variable expressiv-ity (Fig. 2B). We are only aware of a single previousreport documenting incomplete penetrance in afamily with a GLI3 mutation [Debeer et al., 2003].None of the additional features of GCPS or PHS wereevident in any of the affected family members.

Our studies ofNMDshed light on thepathogenesis,and genotype–phenotype correlations, of bothSALL1 and GLI3 truncating mutations. In the case ofSALL1, we obtained strikingly different results for thetwo mutations. The 3414_3415delAT mutation waspresent at only �40% of the level of the wild-typeallele, but this was fully restored after cycloheximidetreatment (Fig. 5, right). These findings are typical ofNMD, and the observed under-representation of themutant allele is in the range reported in previousstudies of naturally occurring PTC mutations [Perrin-Vidoz et al., 2002; Harries et al., 2005]. NMD of themutant transcript may mitigate the adverse effects ofthe mutation, consistent with the associated mild andvariable phenotype.

By contrast, the level of mutant SALL1 cDNA fromfibroblasts derived from patient OX3335 (995delC)with typical TBS was equal to wild-type, and did notchange when the cells were treated with cyclo-heximide. Although it had previously been specu-lated that truncating TBS mutations might escapeNMD [Borozdin et al., 2006], this finding wassurprising because the lack of NMD in this casecontravenes the rule that it is predicted tooccurwhenthe PTC is �50–55 nucleotides 50 of the final exon

FIG. 2. Phenotype and pedigree of patient OX2877. A: The right (left panel) and left (right panel) hands demonstrate bilateral PAP-B. B: Family pedigree; theproband is indicated with an arrow.

3156 FURNISS ET AL.


junction [Zhang and Maquat, 1997]. The 995delCmutation resides in the very large exon 2 of the threeexon SALL1 gene, well upstream of this boundary(and indeed upstream of the 3414_3415delATmutation, which does undergo NMD; Fig. 3). Thisprovides the first direct evidence in humans that TBSmay be caused by a truncated SALL1 protein acting ina dominant negative manner. It is consistent withmouse models of TBS [Kiefer et al., 2003], and alsovalidates the in vitro experimental approach usingmutant cDNA constructs taken by Sakaki-Yumotoet al. [2006].

Turning to the analysis ofGLI3, weobtained similarresults with all three PTC mutants, which exhibitedmutant cDNA levels of 40–60% of the wild-typeallele, that were restored to normal after cyclo-heximide treatment. Thus all three alleles appearsubject to NMD, which is likely to contribute to themilder phenotypes (GCPS and PAP-B) observed inassociation with these mutations, and may accountfor why the two truncations in the ‘‘PHS region’’ thatwe studied (2372delC and 2374C>T) are notassociated with any definite PHS features. Ourfinding of NMD in the case of the 2374C>T mutationdiffers from unpublished observations on lympho-

blast cells from a different patient, in which the levelof mutant message appeared similar to wild-type[Johnston et al., 2005], but concords with the GCPSphenotype consistently associated with this recur-rent mutation. The demonstration that GLI3 muta-tions undergo NMD is important, as the in vitro workreported by Shin et al. [1999] assumed that themutations causing GCPS, PHS, and PAP all lead to theproduction of truncated proteins, without taking thepossibility of NMD into account. Ideally, the con-tribution of NMD should be assessed in parallel withsuch functional analyses. We speculate that PHSmutations may be characterized by partial orcomplete escape from NMD, in a manner analogousto the situation that we have documented for the995delC mutation in SALL1. Unfortunately no fibro-blasts from PHS patients were available to us to testthis hypothesis.

Why certain mutations more than 55 nucleotides 50

of thefinal exon junctionmayescapeNMD isnotwellunderstood, although other examples have beenreported [Zhang andMaquat, 1997; Inoue et al., 2004;Harries et al., 2005]. Most of these mutations arelocated close to the initiation codon, and it ispostulated that NMD avoidance is mediated by

FIG. 3. Schematic representation of the gene and protein structure of (A) SALL1 (3 exons) and (B)GLI3 (15 exons; not drawn to scale). In each part, the upper panelshows the genomic and transcript structure and the lower panel shows key functional domains in the protein (filled rectangles in the SALL1 protein represent zincfinger domains). The filled arrows show the positions of mutations described in this report, and the unfilled arrows depict the equivalent positions of the final exonjunctions. Zn finger, zinc finger; PKA, multiple phosphokinase A phosphorylation sites.



reinitiation of translation [Zhang and Maquat, 1997;Harries et al., 2005], or by physical proximity to theinitiation codon [Inacio et al., 2004]. The 995delCmutation, in common with other mutations reportedto cause TBS, is not close to the initiation codon.Indeed, a PTC too close to the initiation codon wouldyield a truncated protein of insufficient length topossess the dominant negative functions required togive rise to the phenotype. This suggests that a novel

mechanism of NMD avoidance is operating in TBS,and further study may provide insights into the basicmechanism of NMD.

Finally, it is important to note that whilst our resultsare consistent with a role for NMD in the molecularpathology of mutations in SALL1 and GLI3, strictlyspeaking the experimental method that weusedonlyprovides evidence that a process dependent onprotein synthesis is involved. The protein synthesis

FIG. 4. Pyrosequencing analysis of SALL1 cDNA in the forward direction from patients OX3335 (995delC) (A) and OX2948 (3414_3415delAT) (B). In each figure partthe cDNA sequence analyzed is shown at the top, with the PCR primer sequences in bold, sequencing primer in italics, and the deleted base(s) underlined. Below, thedispensation order and mutant-(M)/wild-type- (WT) specific bases used for comparison are indicated, with a table listing the allele-specific incorporation of bases ateach dispensation (denoted by superscripted numbers). The lower part shows the pyrograms obtained after amplification of untreated cDNA from a control without aSALL1 mutation (upper left), and of untreated (upper right) and cycloheximide-treated (lower left) cDNA from the patient. The mutant allele-specific peaks used forquantification are marked with dashed arrows, wild-type allele-specific peaks are marked with solid arrows.

3158 FURNISS ET AL.


inhibitor used in these experiments, cycloheximide,is indeed widely employed to prevent NMD bypreventing the ‘‘pioneer round’’ of translation whichdetects a PTC [Fasken and Corbett, 2005]. However,strategies to directly inhibit the NMD pathway, forexample by RNAi knock down of UPF1, would berequired to establish more rigorously the role ofNMD in these patients.

ACKNOWLEDGMENTS

We are grateful to the families for their cooperationwith this study, to D. Johnson for assistance withpatient recruitment, S. Butler for fibroblast culture,and K. Clarke for DNA sequencing. Pyrosequencingwas conducted using the London IDEAS KnowledgePark facility (Institute of Child Health, London) withthe help of K. Pearce, P. Scambler, A. Goriely and S.Twigg. This work was supported by grants from theWellcome Trust to D.F. (Clinical Training Fellow-ship) and A.O.M.W. (Programme Grant).

REFERENCES

Borozdin W, Steinmann K, Albrecht B, Bottani A, Devriendt K,Leipoldt M, Kohlhase J. 2006. Detection of heterozygousSALL1 deletions by quantitative real time PCR proves thecontribution of a SALL1 dosage effect in the pathogenesis ofTownes-Brocks syndrome. Hum Mutat 27:211–212.

Botzenhart EM, Bartalini G, Blair E, Brady AF, Elmslie F, ChongKL, Christy K, Torres-Martinez W, Danesino C, Deardorff MA,Fryns JP, Marlin S, Garcia-Minaur S, Hellenbroich Y, Hay BN,Penttinen M, Shashi V, Terhal P, Van Maldergem L, WhitefordML, Zackai E, Kohlhase J. 2007. Townes-Brocks syndrome:Twenty novel SALL1 mutations in sporadic and familial casesand refinement of the SALL1 hot spot region. Hum Mutat 28:204–205.

Debeer P, Peeters H, Driess S, De Smet L, Freese K, Matthijs G,Bornholdt D, Devriendt K, Grzeschik KH, Fryns JP, Kalff-Suske M. 2003. Variable phenotype in Greig cephalopoly-syndactyly syndrome: Clinical and radiological findings in 4independent families and 3 sporadic cases with identifiedGLI3 mutations. Am J Med Genet Part A 120A:49–58.

Elson E, Perveen R, Donnai D, Wall S, Black GC. 2002. De novoGLI3 mutation in acrocallosal syndrome: Broadening thephenotypic spectrum ofGLI3 defects andoverlapwith murinemodels. J Med Genet 39:804–806.

Fasken MB, Corbett AH. 2005. Process or perish: Quality controlin mRNA biogenesis. Nat Struct Mol Biol 12:482–488.

FIG. 6. Relative quantification of mutant and wild-type alleles inGLI3 cDNA. The mean proportion (�2 SD) of mutant allele compared to wild-type, before (unfilledsymbols) and after (black symbols) cycloheximide treatment, is shown for the 1320_1321insT (left), 2372delC (center) and 2374C>T (right) mutations. Squares andtriangles indicate the results of pyrosequencing in the forward and reverse orientations, respectively.

FIG. 5. Relative quantification of mutant and wild-type alleles in SALL1 cDNA. The mean proportion (�2 SD) of mutant allele compared to wild-type, before (unfilledsymbols) and after (black symbols) cycloheximide treatment, is shown for the 995delC (left) and 3414_3415delAT (right) mutations. Squares and triangles indicate theresults of pyrosequencing in the forward and reverse orientations, respectively.



Hall GW, Thein S. 1994. Nonsense codon mutations in theterminal exon of the b-globin gene are not associated with areduction in b-mRNA accumulation—A mechanism for thephenotype of dominant b-thalassemia. Blood 83:2031–2037.

Harries LW, Bingham C, Bellanne-Chantelot C, Hattersley AT,Ellard S. 2005. The position of premature termination codons inthe hepatocyte nuclear factor-1 beta gene determines suscept-ibility to nonsense-mediated decay. Hum Genet 118:214–224.

Inacio A, Silva AL, Pinto J, Ji X, Morgado A, Almeida F, Faustino P,Lavinha J, Liebhaber SA, Romao L. 2004. Nonsense mutationsin close proximity to the initiation codon fail to trigger fullnonsense-mediated mRNA decay. J Biol Chem 279:32170–32180.

Inoue K, Khajavi M, Ohyama T, Hirabayashi S, Wilson J, RegginJD, Mancias P, Butler IJ, Wilkinson MF, Wegner M, Lupski JR.2004. Molecular mechanism for distinct neurological pheno-types conveyed by allelic truncating mutations. Nat Genet36:361–369.

Johnson D, Kan SH, Oldridge M, Trembath RC, Roche P, EsnoufRM, Giele H, Wilkie AOM. 2003. Missense mutations in thehomeodomain of HOXD13 are associated with brachydactylytypes D and E. Am J Hum Genet 72:984–997.

Johnston JJ, Olivos-Glander I, Killoran C, Elson E, Turner JT,Peters KF, Abbott MH, Aughton DJ, Aylsworth AS, BamshadMJ, Booth C, Curry CJ, David A, Dinulos MB, Flannery DB, FoxMA, Graham JM, Grange DK, Guttmacher AE, Hannibal MC,Henn W, Hennekam RC, Holmes LB, Hoyme HE, Leppig KA,Lin AE, Macleod P, Manchester DK, Marcelis C, Mazzanti L,McCann E, McDonald MT, Mendelsohn NJ, Moeschler JB,Moghaddam B, Neri G, Newbury-Ecob R, Pagon RA, PhillipsJA, Sadler LS, Stoler JM, Tilstra D, Walsh Vockley CM, ZackaiEH, Zadeh TM, Brueton L, Black GC, Biesecker LG. 2005.Molecular and clinical analyses of Greig cephalopolysyndac-tyly and Pallister-Hall syndromes: Robust phenotype predic-tion from the type and position of GLI3 mutations. Am J HumGenet 76:609–622.

Kang S, Graham JM Jr, Olney AH, Biesecker LG. 1997. GLI3frameshift mutations cause autosomal dominant Pallister-Hallsyndrome. Nat Genet 15:266–268.

Khajavi M, Inoue K, Lupski JR. 2006. Nonsense-mediated mRNAdecay modulates clinical outcome of genetic disease. Eur JHum Genet 14:1074–1081.

Kiefer SM, Ohlemiller KK, Yang J, McDill BW, Kohlhase J,Rauchman M. 2003. Expression of a truncated Sall1 transcrip-tional repressor is responsible for Townes-Brocks syndromebirth defects. Hum Mol Genet 12:2221–2227.

Kohlhase J, Wischermann A, Reichenbach H, Froster U, Engel W.1998.Mutations in the SALL1putative transcription factor genecause Townes-Brocks syndrome. Nat Genet 18:81–83.

Langaee T, Ronaghi M. 2005. Genetic variation analyses bypyrosequencing. Mutat Res 573:96–102.

Mendell JT, Dietz HC. 2001. When the message goes awry:Disease-producing mutations that influence mRNA contentand performance. Cell 107:411–414.

Nishinakamura R, Matsumoto Y, Nakao K, Nakamura K, Sato A,Copeland NG, Gilbert DJ, Jenkins NA, Scully S, Lacey DL,Katsuki M, Asashima M, Yokota T. 2001. Murine homolog ofSALL1 is essential for ureteric bud invasion in kidneydevelopment. Development 128:3105–3115.

Perrin-Vidoz L, Sinilnikova OM, Stoppa-Lyonnet D, Lenoir GM,Mazoyer S. 2002. The nonsense-mediated mRNA decaypathway triggers degradation of most BRCA1 mRNAs bearingpremature termination codons. Hum Mol Genet 11:2805–2814.

Radhakrishna U, Wild A, Grzeschik KH, Antonarakis SE. 1997.Mutation in GLI3 in postaxial polydactyly type A. Nat Genet17:269–271.

RadhakrishnaU,BornholdtD, ScottHS, PatelUC, Rossier C, EngelH, Bottani A, Chandal D, Blouin JL, Solanki JV, Grzeschik KH,Antonarakis SE. 1999. The phenotypic spectrum of GLI3morphopathies includes autosomal dominant preaxial poly-dactyly type-IV and postaxial polydactyly type-A/B; Nophenotype prediction from the position of GLI3 mutations.Am J Hum Genet 65:645–655.

Sakaki-Yumoto M, Kobayashi C, Sato A, Fujimura S, MatsumotoY, Takasato M, Kodama T, Aburatani H, Asashima M, YoshidaN, Nishinakamura R. 2006. The murine homolog of SALL4, acausative gene in Okihiro syndrome, is essential for embry-onic stem cell proliferation, and cooperates with Sall1 inanorectal, heart, brain and kidney development. Develop-ment 133:3005–3013.

Shin SH, Kogerman P, Lindstrom E, Toftgard R, Biesecker LG.1999. GLI3 mutations in human disorders mimic Drosophilacubitus interruptus protein functions and localization. ProcNatl Acad Sci USA 96:2880–2884.

Vortkamp A, Gessler M, Grzeschik KH. 1991. GLI3 zinc-fingergene interrupted by translocations in Greig syndromefamilies. Nature 352:539–540.

Wassel HD. 1969. The results of surgery for polydactyly of thethumb. Clin Orthop Relat Res 64:175–193.

Weber S, Moriniere V, Knuppel T, Charbit M, Dusek J, GhiggeriGM, Jankauskiene A, Mir S, Montini G, Peco-Antic A, Wuhl E,Zurowska AM, Mehls O, Antignac C, Schaefer F, Salomon R.2006. Prevalence of mutations in renal developmental genesin children with renal hypodysplasia: Results of the ESCAPEstudy. J Am Soc Nephrol 17:2864–2870.

Wild A, Kalff-Suske M, Vortkamp A, Bornholdt D, Konig R,Grzeschik KH. 1997. Point mutations in human GLI3 causeGreig syndrome. Hum Mol Genet 6:1979–1984.

Zhang J, Maquat LE. 1997. Evidence that translation reinitiationabrogates nonsense-mediated mRNA decay in mammaliancells. EMBO J 16:826–833.

3160 FURNISS ET AL.


Nonsense-mediated decay and the molecular pathogenesis of mutations in SALL1 and GLI3

Documents

Transcript of Nonsense-mediated decay and the molecular pathogenesis of mutations in SALL1 and GLI3