document

letter

nature genetics • volume 23 • october 1999 189

Truncating mutations in CCM1, encoding KRIT1,cause hereditary cavernous angiomas

Sophie Laberge-le Couteulx1, Hans H. Jung1, Pierre Labauge1, Jean-Pierre Houtteville2,Christelle Lescoat1, Michaelle Cecillon, Emmanuelle Marechal1, Anne Joutel1,3, Jean-François Bach1 & Elisabeth Tournier-Lasserve1,3

1INSERM U25, Faculté de Médecine Necker, 156 Rue de Vaugirard, 75730 Paris Cedex 15, France. 2Service de Neurochirurgie, CHR Côte de Nacre, Caen,France. 3Laboratoire de Cytogénétique, Hôpital Lariboisière, 2 rue A. Paré, 75010, Paris, France. Correspondence should be addressed to E.T.L. (e-mail: [email protected]).

Cavernous angiomas are vascular malformations mostlylocated in the central nervous system and characterized byenlarged capillary cavities without intervening brainparenchyma1. Clinical symptoms include seizures, haemor-rhage and focal neurological deficits. Cavernous angiomasprevalence is close to 0.5% in the general population2. Theymay be inherited as an autosomal dominant condition in asmuch as 50% of cases3. Cerebral cavernous malformations(CCM) loci were previously identified on 7q, 7p and 3q (refs4,5). A strong founder effect was observed in the Hispano-American population, all families being linked to CCM1 on 7q(refs 4,6,7). CCM1 locus assignment was refined to a 4-cMinterval bracketed by D7S2410 and D7S689 (ref. 8). Here wereport a physical and transcriptional map of this interval andthat CCM1, a gene whose protein product, KRIT1, interactswith RAP1A (also known as KREV1; ref. 9), a member of theRAS family of GTPases, is mutated in CCM1 families. Our datasuggest the involvement of the RAP1A signal transductionpathway in vasculogenesis or angiogenesis10.We recently established the clinical and genetic features ofhereditary cavernous angiomas in a series of 57 French fami-lies11. Neuroimaging investigation confirmed the high fre-quency of multiple lesions in hereditary cavernous angiomas(Fig. 1a). It also showed a correlation between the number oflesions and the age of the patient, suggesting a dynamic naturefor these lesions. A genetic linkage analysis conducted in 36 ofthese families showed that 65% were linked to CCM1, with nofounder effect12. Using a published YAC contig and genomicsequence information publicly available in GenBank(sequences from The Washington University Chromosome 7Project), we constructed BAC/PAC contigs covering 90% ofthe estimated 1,600-kb CCM1 interval8,13 (Fig. 2). We used 20families (179 potentially informative meioses) having a poste-rior probability of more than 85% to be linked to CCM1 forfine-mapping of CCM1 with polymorphic markers identifiedin BAC/PAC sequences (Fig. 2). A recombination eventobserved in an affected individual (family 27 from ref. 11)allowed us to reduce this interval, now bracketed by MS2456(centromeric boundary) and D7S689 (telomeric boundary).Multiple-database screening allowed us to map 574 ESTs inthis interval. We clustered them in 53 potential transcriptionalunits including 6 characterized human genes, CDK6, CYP51,CCM1, PEX1, MTERF and YOTIAO.

CCM1 was identified through a two-hybrid screeningdesigned to identify proteins interacting with RAP1A (ref. 9).It encodes a protein (of 529 aa) containing four ankyrin

domains and interacting with RAP1A through its carboxy-terminal region. RAP1A was identified on the basis of itshomology with DRAS3, encoding a Drosophila melanogasterRAS homologue14,15. Contribution of Ras signalling inangiogenesis is suggested by vascular developmental abnor-malities observed in mouse models16,17. In addition, RAP1Ahas been shown to be involved in morphogenesis and celldifferentiation10,18–20. RAP1A is also known to interact withB-Raf, whose deficiency leads to massive endothelial apopto-sis in mice10,17.

We identified 12 exons through alignment of the publishedCCM1 cDNA with BAC HSAC000120 (ref. 9). We designedintronic oligonucleotide primers to amplify all exonic andsplice-site sequences. A panel of 20 unrelated CCM patientsbelonging to families showing a posterior probability to belinked to CCM1 of more than 85% was screened for mutationsusing single-strand conformation polymorphism (SSCP)analysis, direct genomic DNA sequencing and screening ofCCM1 cDNA for deletions. Mutations were identified in 12patients (Fig. 3).

We found point mutations leading to nonsense stopcodons in three probands (pedigree 10, a C→T transition, nt1,283; pedigree 41, a G→A transition, nt 615; pedigree 19, a

Fig. 1 Radiological and histological features of familial cavernous angiomas.a, Cerebral magnetic resonance of a patient: multiple lesions are observed.b, Histological section of a cerebral cavernous angioma showing juxtapositionof thin vascular cavities lined by endothelium and collagen without interven-ing brain parenchyma (Modified Gomori Trichrome Stain, ×40 magnification,courtesy of F. Chapon).

a b

© 1999 Nature America Inc. • http://genetics.nature.com©

199

9 N

atu

re A

mer

ica

Inc.

• h

ttp

://g

enet

ics.

nat

ure

.co

m

letter

190 nature genetics • volume 23 • october 1999

G→A transition, nt 261), and deletions causing frameshiftsresulting in premature stop codons in five probands (1 bp, nt1,342 in pedigree 6; 1 bp, nt 436, in pedigree 25; 1 bp, nt 633,in pedigree 27; 4 bp, nt 681–684, in pedigree 42; 26 bp, nt1,012–1,037, in pedigree 35). Two probands harboured aninsertion leading to a frameshift and a premature stopcodon: insertion of a cytosine after nt 1,271 in pedigree 58and insertion of a cytosine after nt 247 in pedigree 18.

Amplification of cDNA prepared with total RNA from patientsEBV cell lines revealed in two probands (pedigrees 11 and 5) anadditional shorter band of equal intensity with the normal sizedband, which was absent in healthy controls and other patients(data not shown). Sequence analysis of cloned RT-PCR productsof the pedigree 11 proband showed deletion of exon 11 (nt1,430–1,546), leading to a putative truncated protein lacking 39amino acids in the putative RAP1A interacting region. Directgenomic sequencing in this patient with an intronic primerlocated upstream of exon 11 revealed an A→G substitution inintron 10 (12 bp upstream from the acceptor splice site), suggest-ing that the truncated transcript observed in this patient mayresult from abnormal splicing. Sequence analysis of cloned RT-PCR products of the pedigree 5 proband revealed in 30% of theclones an in-frame 84-bp cDNA deletion (nt 1,348–1,431), lead-ing to a putative truncated protein lacking 28 amino acids in theputative RAP1A-interacting region. Genomic DNA amplifica-tion and sequencing suggests that this cDNA deletion is due to agenomic deletion encompassing at least the 3´ end of exon 10 andthe intronic sequence containing the reverse primer used forsequencing of exon 10.

SSCP or sequence analysis of affected and unaffected first-degree relatives showed co-segregation of the mutations with

the affected phenotype in all 12 pedigrees (Fig. 4, and data notshown). The truncating nature of these mutations, theirabsence in healthy controls and their co-segregation with theaffected phenotype suggest that they are disease-causing. Wewere not able to detect mutations in 8 of 20 families tested. Itmay be that some pedigrees, despite a high probability to belinked to CCM1, may be linked to another CCM locus. It isalso possible that large deletions may be missed by the meth-ods used here, or unidentified exons may be missing.

To estimate CCM1 transcript sizes, we hybridized a humanadult multiple-tissue northern blot with a cDNA probeencompassing nt 553–1,367. We detected a 5–6-kb transcriptin all tissues, although it was less abundant in brain, kidneyand lung (Fig. 5). A 3.5-kb band was also observed in heart,skeletal muscle and pancreas. An additional faint 2-kb bandwas identified in pancreas. The size difference between the5–6-kb band and the 1,986-bp cloned and published cDNAsuggests that larger transcripts containing additional, uniden-tified exons may exist9.

Mutations identified here might result in a dominant-nega-tive effect or a loss of function. We favour the second hypothe-sis. Sporadic forms of cavernous angiomas manifest as uniquelesions and familial forms as multiple lesions, which evokes a‘Knudson double hit mechanism’ and would be consistentwith the need for a complete loss of CCM1 function for theappearance of cavernous angiomas. All mutations reportedhere predict truncated CCM1 proteins completely or partiallydevoid of the putative RAP1A-interacting region. Two of thesemutations would lead to limited deletions of this region,enforcing the potential role of an alteration of CCM1/RAP1Ainteraction in CCM pathophysiology and the putative role of

a

b

Fig. 2 Genetic, physical and transcriptional map of the CCM1 interval. a, Genetic map of CCM1. Previous D7S2410–D7S689 and reduced MS2456–D7S689 geneticintervals are indicated by horizontal brackets. Previously published microsatellites are boxed. New microsatellites are identified by bold characters. Some STSs arealso shown. STS sWSS1703 corresponds to nt 393–658 of CCM1. Markers between vertical brackets are less than 1 kb apart. b, Physical and transcriptional maps ofCCM1. BAC contigs spanning the CCM1 interval. BAC AC000120 is indicated by a thicker line. Hits with either STS or microsatellite markers are indicated by smallvertical lines. Black polygon, CCM1; grey polygons, well-characterized human genes; open polygons, genes showing strong homology to genes of other species(not characterized in humans).

CCM1


199

9 N

atu

re A

mer

ica

Inc.

• h

ttp

://g

enet

ics.

nat

ure

.co

m

letter


RAP1A in angiogenesis or vasculogenesis. We can not exclude,however, that CCM1 interacts with other unknown proteins.

MethodsPatients. We included 20 unrelated patients from reported CCM fami-lies with their informed consent11. HOMOG admixture analysisshowed that these families had a posterior probability to be linked toCCM1 higher than 85% (ref. 12). Genomic DNA was extracted fromperipheral leukocytes as described12.

Transcriptional and physical map. We selected 16 YACs spanning theD7S2410–D7S689 region from a previous published contig, and ampli-fied with 23 STS markers to assess their DNA integrity8. We thenamplified 77 ESTs identified in Unigene and mapped 13 in the CCM1interval. We also used a new in silico strategy to assemble genomic con-tigs of the region and constructed a database containing the sequencesof the 23 STS and 13 ESTs previously identified. All sequences, filtered

for repeats (average of 234 bp/STS, ranging from 48 to 504 bp), wereused for a BLASTN search on GenBank non-redundant (nr) databaseto identify BACs and PACs (refs 13,21). Position assignment of BACsand PACs in the contig were based on GenBank (BACs or PACs) neigh-bouring sequence information and hits with either BAC/PAC endsequences, STSs (including microsatellites), ESTs (mapped by PCR onYAC contig) or known genes. Using the identified BAC/PAC genomicsequences, we then launched AAT searches to identify more genes andEST sequences22.

Reduction of the genetic interval. We selcted 12 polymorphicmicrosatellite markers spanning the D7S2410–D7S689 interval for link-age analysis. D7S2410, D7S2409, D7S1813, D7S1789, D7S646, D7S689and D7S558 had been used previously4–8,12. We identified MS65,MS2453A, MS2456A, MS119 and MS120 based on sequencing data ofBACs mapped in the interval. New markers, primer sequences, type ofrepeats and allele sizes and frequencies will be available on dbSNPs. Allmicrosatellites were mapped by PCR on the DNA of the selected YACs.

Fig. 3 CCM1 mutations. Top, CCM1 andpredicted protein product. del, deletion;ins, insertion; intron, mutation in intron.For insertions, nucleotide number corre-sponds to the nucleotide preceding theinsertion. ‘RAP1A interacting region’corresponds to the region (aa 483–529)whose deletion abolishes the interactionwith RAP1A by double-hybrid assay9.Bottom, sequencing data. Arrows, sitesof mutations; WT, wild-type sequence;MT, mutant sequence. Larger bold char-acters in amino acid sequences representthe amino acid corresponding to the lastnormal codon. *, reverse-strand chro-matograms. Note that for pedigrees 35and 42, genomic PCR products werecloned to precisely define deletionboundaries. For pedigree 5, amplifiedcDNA products were cloned and only theabnormal sequence is shown at the bot-tom of the chromatogram frame.

RAP1Ainteracting region


199

9 N

atu

re A

mer

ica

Inc.

• h

ttp

://g

enet

ics.

nat

ure

.co

m

letter

192 nature genetics • volume 23 • october 1999

Mutation screening SSCP, sequence and cDNA analysis. We designed15 sets of primers to amplify all 12 exons and flanking intronic splicesites of CCM1 using genomic DNA (available on request) based on BACHSAC000120 sequence. PCR reactions were conducted as follows: afteran initial denaturation step at 94 °C (4 min), 30 cycles of amplificationconsisting of steps at 94 °C (15 s), optimized annealing temperature of45–55 °C (15 s) and 72 °C (15 s), followed by a final extension step at72 °C (10 min).

We screened genomic DNA of all 20 patients by SSCP. PCR productswere electrophoresed in four different conditions (10% acrylamide-bisacrylamide 37.5:1 with or without 10% glycerol at both 4 °C and20 °C) on a Mighty Small II apparatus (Pharmacia-Biogen) at a con-stant current of 35 mA. Conformers were revealed by silver staining(Silver Stain Plus kit, Biorad). Amplimers showing an atypical SSCPbanding pattern were sequenced using a big dye-terminator cycle-sequencing kit according to manufacturer’s instructions (ABI377,

Perkin Elmer). We tested all confirmed mutations for co-segregation bySSCP using informative relatives; 50 unrelated healthy spouses wereused as negative controls. Genomic amplicons of pedigrees 25 and 42probands were cloned (pGEMT easy cloning kit) and sequenced toallow us to delineate the exact boundaries of the deletions.

We screened all index patients in whom no abnormal SSCP conformerwas observed (12 patients) using both direct sequencing of genomic DNA(all 12 exons) and cDNA screening for deletions. cDNA was preparedusing patients and healthy control EBV cell lines total RNA, oligo-dTprimers and Moloney Murine Virus reverse transcriptase, according tothe manufacturer’s instructions (Amersham). Amplification of probandsand healthy controls was conducted with 4 sets of primers (available onrequest): set 1 (nt 68–658), set 2 (nt 553–943), set 3 (nt 845–1,642) andset 4 (nt 1,334–1,642). Unique and intense bands were obtained for eachof these sets with control cDNA, suggesting that CCM1 is expressed inEBV cell lines. We cloned set 3 RT-PCR products from one healthy con-trol and probands of pedigrees 5 and 11 with pGEMT easy vector cloningkit according to the manufacturer’s instructions (Promega). For pedi-grees 5 and 11 probands, 10 colonies were amplified to assess the size ofcloned RT-PCR products and 4 clones were sequenced on both strands,including 2 clones of normal size and 2 shorter clones. For control, weamplified 10 colonies and sequenced 2 clones.

Northern-blot analysis. We generated a cDNA PCR probe spanning exons5–10 (nt 553–1,367), which excludes ankyrin repeats and 3´ Alu sequenceto avoid non-specific cross-hybridization. QIAquick PCR (Qiagen) puri-fied probe was labelled with [32P] adCTP and used first on a genomicDNA HindIII-digested Southern blot to assess its specificity. Probe wasthen hybridized in a Dextran-containing medium (Dextran 10%, NaCl 1M, SDS 1% Tris HCl 50 mM, EDTA 5 mM) on a human adult multiple-tissue northern blot (MTN1, Clontech) at 65 °C overnight. It was washedfor 20 min in SSPE 0.1×SDS (0.1%) at RT, 20 min in SSPE 0.1×SDS (0.1%)at 65 °C. A Kodak film was then exposed for 5 d at –80 °C for 120 h withthe blot.

GenBank accession numbers. Human CCM1, U90268, U90269; humangenomic sequences including CCM1, HSAC000120; human STSs corre-sponding to CCM1, G31685 (EST M62210).

Fig. 5 CCM1 northern-blot analysis. Human adult multiple-tissue northern blot(MTN1, Clontech) was hybridized with a PCR cDNA probe (nt 553–1,367). Bot-tom, hybridization of the same blot with a GAPD probe. He, heart; Br, brain; Pl,placenta; Lu, lung; Li, liver; SkM, skeletal muscle; Ki, kidney; Pa, pancreas. Sizemarkers are on the left.

kb

GAPD

CCM1

Fig. 4 Co-segregation of abnormal conformers with the affected phenotype in the eight pedigrees showing abnormal SSCP patterns. Open, normal MRI; filled,symptomatic patients harbouring cavernoma on MRI; half-filled, asymptomatic individuals harbouring cavernoma on MRI; ?, unknown status; \, deceased.Patients deceased or with unknown status were not tested for mutations and are shown for a better understanding of the familial structures. Abnormal bandsare indicated by arrows.


199

9 N

atu

re A

mer

ica

Inc.

• h

ttp

://g

enet

ics.

nat

ure

.co

m

letter


1. Russell, D.S. & Rubenstein, L.J. Pathology of Tumours of the Nervous System730–736 (Williams and Wilkins, Baltimore, 1989).

2. Otten, P., Pizzolato, G.P., Rilliet, B. & Berney, J. A propos de 131 cas d’angiomescaverneux (cavernomes) du SNC, repérés par l’analyse rétrospective de 24 535autopsies. Neurochirurgie 35, 82–83 (1989).

3. Rigamonti D. et al. Cerebral cavernous malformations. Incidence and familialoccurence. N. Engl. J. Med. 319, 343–347 (1988).

4. Dubovsky, J. et al. A gene responsible for cavernous malformations of the brainmaps to chromosome 7q. Hum. Mol. Genet. 4, 453–458 (1995).

5. Craig, H.D. et al. Multilocus linkage identifies two new loci for a Mendelian formof stroke, cerebral cavernous malformation, at 7p15–13 and 3q25.2–27. Hum.Mol. Genet. 7, 1851–1858 (1998).

6. Günel, M. et al. A founder mutation as a cause of cerebral cavernousmalformation in Hispanic Americans. N. Engl. J. Med. 334, 946–951 (1996).

7. Günel, M. et al. Genetic heterogeneity of inherited cerebral cavernousmalformation. Neurosurgery 38, 1265–1271 (1996).

8. Johnson, E.W. et al. Refined localization of the cerebral cavernous malformationgene (CCM1) to a 4 cM interval of chromosome 7q contained in a well-definedYAC contig. Genome Res. 5, 368–380 (1995).

9. Serebriiskii, I. et al. Association of Krev-1/rap1a with Krit1, a novel ankyrin repeat-containing protein encoded by a gene mapping to 7q21–22. Oncogene 15,1043–1049 (1997).

10. Bos, J.L. All in the family? New insights and questions regarding interconnectivityof ras, rap1 and raI. EMBO J. 17, 6776–6782 (1998).

11. Labauge, P. et al. Hereditary cerebral cavernous angiomas, clinical and genetic

features in 57 French families. Lancet 352, 1892–258 (1998).12. Laberge, S., Labauge, P., Maréchal, E., Maciazek, J. & Tournier-Lasserve, E. Genetic

heterogeneity and absence of founder effect in a series of 36 non Hispano-American cerebral cavernomas families. Eur. J. Hum. Genet. 7, 499–504 (1999).

13. Benson, D.A. et al. GenBank. Nucleic Acids Res. 27, 12–17 (1999).14. Pizon, V., Chardin, P., Lerosey, I., Olofsson, B. & Tavitian, A. Human cDNAs rap 1

and rap2 homologous to the Drosophilia gene Dras3 encode proteins closelyrelated to ras in the effector region. Oncogene 3, 201–204 (1988).

15. Kitayama, H., Sugimoto, Y., Matsuzaki, T., Ikawa, Y. & Noda, M. A ras-related genewith transformation suppressor activity. Cell 56, 77–84 (1989).

16. Henkemeyer, M. et al. Vascular system defects and neuronal apoptosis in micelacking ras GTPase-activating protein. Nature 377, 695–701 (1995).

17. Wojnowski, L. et al. Endothelial apoptosis in Braf-deficient mice. Nature Genet.16, 293–297 (1997).

18. Asha, H., de Ruiter, N.D., Wang, M.G. & Hariharan, I.K. The Rap1 GTPase functionsas a regulator of morphogenesis in vivo. EMBO J. 18, 605–615 (1999).

19. Quarck, R. et al. Differential up-regulation of Rap1a and Rap1b proteins duringsmooth muscle cell cycle. Eur. J. Cell Biol. 70, 269–277 (1996).

20. Pizon, V, Cifuentes-Diaz, C., Mege, R.M., Baldacci, G. & Rieger, F. Expression andlocalization of rap1 proteins during myogenic differentiation. Eur. J. Cell Biol. 69,224–235 (1996).

21. Altschul, S.F. et al. Gapped BLAST and PSI-BLAST, a new generation of proteindatabase search programs. Nucleic Acids Res. 25, 3389–3402 (1997).

22. Huang, X., Adams, M.D., Zhou, H. & Kerlavage, A.R. A tool for analysing andannotating genomic sequences. Genomics 46, 37–45 (1997).

AcknowledgmentsWe thank all members of the Societe Francaise de Neurochirurgie, inparticular, L. Capelle, J.P. Castel, D. Fohano, B. George, J. Philippon,A. Rey and F. Roux, this study would not have been possible without theirhelp; F. Chapon for pathological advice and the families for theirparticipation. S.L.L.C. had a studentship from the Fonds de la Rechercheen Santé du Québec (FRSQ, Canada). P.L. had a poste d’accueil INSERMand H.H.J. had a fellowship from the Schweizerishe Stiftung fûr

Medizinisch-Biologische Stipendien. This work was supported byINSERM, Ministère de l’Enseignement Supérieur et de la Recherche(MESR, ACCSV 1995).

Received 5 May; accepted 30 August 1999.


199

9 N

atu

re A

mer

ica

Inc.

• h

ttp

://g

enet

ics.

nat

ure

.co

m

Documents

document