Med Review article Molecular genetics of …Review article Moleculargenetics ofneurofibromatosis type 1 (NFl) MingHongShen, Peter S Harper, MeenaUpadhyaya Institute ofMedical Genetics,

Review article

Molecular genetics of neurofibromatosis type 1(NFl)

Ming Hong Shen, Peter S Harper, Meena Upadhyaya

Institute of MedicalGenetics, University ofWales College ofMedicine, Heath Park,Cardiff CF4 4XN, UKM H Shen*P S HarperM Upadhyaya* Present address:CRC ChromosomeMolecular Biology Group,Department of Biochemistry,University of Oxford, SouthParks Road, OxfordOXI 3QU, UK.

Correspondence to: Dr Shenor Dr Upadhyaya.

AbstractNeurofibromatosis type 1 (NF1), alsocalled von Recklinghausen disease or peri-pheral neurofibromatosis, is a commonautosomal dominant disorder char-acterised by multiple neurofibromas, cafeau lait spots, and Lisch nodules ofthe iris,with a variable clinical expression. Thegene responsible for this condition, NF1,has been isolated by positional cloning.It spans over 350kb of genomic DNA inchromosomal region 17qll.2 and encodesan mRNA of 11-13 kb containing at least59 exons. NFI is widely expressed in avariety of human and rat tissues. Fouralternatively spliced NFI transcripts havebeen identified. Three of these transcriptisoforms (each with an extra exon: 9br,23a, and 48a, respectively) show differ-ential expression to some extent in varioustissues, while the fourth isoform (2-9kbin length) remains to be examined. Theprotein encoded by NFI, neurofibromin,has a domain homologous to the GTPaseactivating protein (GAP) family, anddownregulates ras activity. The iden-tification of somatic mutations in NFIfrom tumour tissues strongly supports thespeculation that NFl is a member of thetumour suppressor gene family. Althoughthe search for mutations in the gene hasproved difficult, germline mutation ana-lysis has shown that around 82% of all thefully characterised NF1 specific mutationsso far predict severe truncation of neuro-fibromin. Further extensive studies arerequired to elucidate the gene function andthe mutation spectrum. This should thenfacilitate the molecular diagnosis and thedevelopment of new therapy for the dis-ease.(J7Med Genet 1996;33:2-17)

Key words: neurofibromatosis 1; molecular genetics.

The diseaseThe neurofibromatoses (NF) are a group ofneurocutaneous syndromes primarily affectingtissues derived from the neural crest. The ex-treme clinical heterogeneity of NF has evokedseveral attempts to classify the disease into

several discrete entities.'` At least two dis-tinctive forms, neurofibromatosis type 1 (NF1)and neurofibromatosis type 2 (NF2), were re-cognised and later confirmed by the cloning oftwo separate genes, the NF1 gene on chro-mosome 1746 and the NF2 gene on chro-mosome 22.78 Other NF related syndromeswere also reported, such as segmental NF,9Watson syndrome,'0 Noonan syndrome,"spinal NF,'2 familial cafe au lait spots(CLS)," and Schwannomatosis.'4 It remainsto be determined whether such disorders aregenetically discrete.NFl, also called von Recklinghausen disease

or peripheral neurofibromatosis, is a commonautosomal dominant disorder affecting about1 in 3000 to 5000 people. It exhibits full pene-trance and a high mutation rate with 30 to50% ofNFl patients representing a new muta-tion. '-'7 The NF mutation events show biastowards paternal origin.'819 The condition ischaracterised by multiple CLS, neurofibromas,and Lisch nodules of the iris.'5-'7 Its clinicalexpression in a wide spectrum involving mul-tiple body systems varies greatly from onepatient to another, between families, and evenwithin a given family carrying the same muta-tion. Although the three characteristic features(CLS, neurofibromas, and Lisch nodules) eachoccur in over 90% of all NF1 patients bypuberty, the number of lesions is extremelyvariable. Other features are only present ina minority of NF cases, such as learningdisabilities, seizures, macrocephaly, shortstature, scoliosis, pseudoarthrosis, andmalignancies.20-22

The cloning of the NF1 geneThe NFl gene was isolated by positional clon-ing. Initial exclusion mapping suggested thatthe NFI gene was most likely to be on eitherchromosome 5 or 17.23 Seizinger et al'4 firstreported linkage of the NF gene to the nervegrowth factor receptor (NGFR) gene locatedat 17q22 with the marker pE5 1. Barker et al'5established significant linkage with the markersderived from chromosome 17, especially thecentromeric markers p3.6 (D17Z1) andpA10.41 (D17S71). White et al"6 observedthat the centromeric marker p3.6, and a newmarker pHHH202 (D17S33), which maps to

2717q11.2, were very close to the NEl gene.

2 J7 Med Genet 1996;33:2-17

on March 4, 2020 by guest. P

rotected by copyright.http://jm

g.bmj.com

/J M

ed Genet: first published as 10.1136/jm

g.33.1.2 on 1 January 1996. Dow

nloaded from

http://jmg.bmj.com/

3Molecular genetics of neurofibromatosis type 1 (NFI)

A

17 qcen

BStart codon (ATG)

5''-V----

30 bp insertion at base1260 in 5'ALT2 (exon 9br)

ATG5' f

1641 bp 5'codingsequence of 5'ALT1

-350 kb NF1 genomic DNA

t(;17) 7 tt(l7;22)OMGP EVl2B EV12A

11-13 kb NF1 mRNA8457 bp coding sequence

NF1-GRD region(exons 21-27)

63 bp insertion at base4111 in GRD II (exon 23a)

17 qter

Stop codon (TGA)

3,

TGA3'

54 bp insertion at base 8315 in3'ALT (exon 48a)

The NFI gene structure. (A) The size of the NFI gene and the embedded genes. (B) The size of the NFl transcript.(C) The location ofNFl-GRD and the alternatively spliced NFl transcripts.

These results coupled with other reports sum-marised by Skolnick et al28 indicated that theNFl gene was located on the long arm ofchromosome 17. Further genetic analyses byeight teams with 31 markers clearly located theNFl gene to the proximal long arm ofchromosome 17. The combination of thesedata resulted in an estimation of the mostlikely order of 13 loci on chromosome 17 as:

pter-pA10.41-EW30 1-p 1 7H8(centromere)-pHHH202-NF 1 -EW206-EW207-EW204-EW203-CRI.L58 1-CRI.L946-HOX2-NGFR-qter.29 In this multipoint linkage map, the twomarkers pHHH202 and EW206 narrowed thelocation of the NF1 locus to about 3 cM of17ql 1.2.Cytogenetic abnormalities from two un-

related NFl patients provided the first physicalevidence that the NF1 gene was on the proximallong arm of chromosome 17. One of thepatients had a balanced translocation betweenchromosomes 1 and 17, t(1;17)(p34.3;ql 1.2).30 The other carried a balanced trans-location between chromosomes 17 and 22,t(17;22)(qll.2;qll.2).3' Somatic hybrid map-ping analysis showed that the derived order ofDNA markers and the translocation break-points agreed with the genetic map.31 32 Two ofthe DNA markers derived from chromosome17 detected the breakpoints on PFGE gels.One (clI.lF1 0) identified abnormal PFGEfragments in both translocation patients andplaced the breakpoints within a 600 kb regionexpected to contain the NFl gene.32 The other(17L1) detected abnormal PFGE fragments inthe t(1; 17) patient and her affected offspring.33These results led to the construction of a longrange restriction map of the NF1 region en-

compassing the two breakpoints which were

separated by a minimum of60 kb.34 Both PFGEanalysis with a jump clone pEH135 and chro-mosomal walking with cosmids cEVI20 andcEVI3636 further indicated that the breakpointswere approximately 60 kb apart.

Subsequent search for coding sequences inthe region between the translocation break-points identified three candidate genes

EVI2A,37 EVI2B,38 and OMGP,39 and apseudogene AK3.4° However, none of thesegenes spanned either of the two translocationbreakpoints and no NFI specific mutationscould be found in these genes from NF 1patients tested.3739The NF1 transcript was finally identified by

two American groups led by Drs Collins andWhite. Using PFGE and Southern blottinganalyses, Viskochil et al5 identified three largedeletions (190 kb, 40 kb, and 11 kb) in the NFlcandidate region. The 11 kb deletion did notcontain any of the previously characterisedgenes located between the translocation break-points. However, it partially deleted the 3-8 kbEcoRI fragment (probe EE3.8) which wasshown to span the NFl t(17;22) breakpointand strongly hybridised to a 3-35 kb EcoRImouse fragment on Southern blots from so-

matic cell hybrids.3637 By screening severalcDNA libraries with the conserved sequenceEE3.8, a number of cDNA clones were iden-tified from the translocation breakpoint region(TBR), generating an overlapping cDNA se-

quence of approximately 5 kb. One of thesecDNA clones hybridised to an mRNA of about11 kb. These cDNA clones represented partof a new gene, TBR. By sequencing thesecDNA clones, a 4kb sequence was obtainedwhich contained a single open reading frameof about 3-3 kb.4 The TBR gene was found tobe interrupted by the translocation t(17;22),the three large deletions, and six small NF1specific mutations in the coding sequence.45

Simultaneously, Wallace et alt screenedcDNA libraries with the jump clone pEHI andthe YAC clone Al 13D7,4' spanning the twotranslocation breakpoints. Two cDNA clones(p5 and B3A) were isolated covering a sequenceof 2012 bp with a single open reading framecontaining six exons, and spanning at least33 kb ofgenomic DNA. Northern blotting ana-

lysis with the clone p5 led to the identification ofa transcript of about 13 kb. This gene, initiallytermed NF1LT (NFl large transcript) gene,was found to be disrupted by the translocationt(I7;22). An NFl specific de novo insertion of

C



g.bmj.com

/J M



nloaded from

http://jmg.bmj.com/

Shen, Harper, Upadhyaya

0 5 kb was also detected in the gene. NF1LTand TBR represent the same gene, NF1, butwere isolated from different cDNA clones.

The NFI gene structureNF1 spans over 350 kb of genomic DNA inchromosomal region 17ql1.2 and encodes anmRNA of 11-1 3 kb containing at least 59 exons(figure). An 8457 bp open reading frame ofthe NFI transcript predicts a protein of 2818amino acids with an estimated molecular massof 327 kDa.5 642-44There is a 360 amino acid region of the

predicted gene product which shows homologyto the catalytic domain of the mammalianGTPase activating protein (GAP) and theproducts of the yeast IRAl and IRA2 genes.4445This region is referred to as NFI GAP relateddomain (NF1-GRD) and is located in the cent-ral portion of the NF1 gene, encompassingexons 21-27 (figure).The three genes, EVI2A, EVI2B, and

OMGP, are embedded within intron 27 (fig-ure).37-3 These genes are transcribed in theopposite orientation to NFl. The OMGP genemaps within 4 kb of the t(1; 17) breakpoint.The EVI2A and EVI2B genes lie approximately20 and 5 kb telomeric to the OMGP gene,respectively. The intronless AK3 pseudogenelies in intron 37 and is transcribed in the sameorientation as NF1 (figure) 40

Four alternatively spliced NF1 transcriptshave been identified. The first is called GRD II,with a 63 bp insertion (exon 23a) in the GAPrelated domain, and the second 3'ALT, witha 54 bp insertion (exon 48a) at the 3' end(figure).4434647 The third isoform of the NF1transcript (termed here 5'ALT1), about 2-9 kbin length, has an open reading frame predictinga protein of 551 amino acid residues and shar-ing the same 547 amino-terminal residues withthe original NF1 product (figure).48 A recentlyidentified isoform of the NF1 transcript(termed here 5'ALT2) contains a 30 bp in-sertion (exon 9br) between exons 9b and 10(figure).4

NFI gene expressionTHE NF1 mRNANF1 has been shown to be widely expressedin many tissues. The 13 kb transcript was de-tected in human brain, neuroblastoma, kidney,and melanoma.6 An 11-13kb transcript wasdetected in murine kidney, brain, B 16 melan-oma cells, but not in mouse skin, spleen,thymus, or liver.45 Expression of the NF1 genewas also found by RT-PCR analysis in manyhuman tissues, including WBC, skin fibroblast,brain, spleen, lung, muscle, neuroblastoma,thymoma, neurofibroma, an NF 1 neuro-fibrosarcoma cell line, a colon carcinoma cellline, and breast cancer.65051 Ribonucleajseprotection assays and in situ hybridisation ana-lysis showed that the NF1 transcript was alsoexpressed in a variety of tissues in the chickembryo.5The alternatively spliced isoforms ofthe NF 1

transcript have been investigated to determine

their expression pattern in different tissues.GRD I and GRD II are equally represented inthe RT-PCR products amplified from EBVtransformed lymphocytes either from NF1patients or from normal controls.43 These twoisoforms have been found to have equal ratiosin human kidney, placenta, and lung.5' UsingRT-PCR, Andersen et al46 were able to detectGRD I and GRD II transcripts in all tissuestested, but with a variation in the relativeamounts of the two NF1 transcripts. Thesetissues included kidney, spleen, stomach, lung,colon, muscle, brain, peripheral nerve, skinfibroblasts, an EBV lymphoblastoid cell line,HeLa cells, HepG2, HEL, HT29, and NF88-2 cell lines, neuroblastoma, thymoma, breastcarcinoma, and neurofibroma. Interestingly,the expression of these two isoforms has beenobserved to be associated with the differ-entiation status of a particular tissue.50 GRD Ipredominates in fetal brain and un-differentiated primitive neuroectodermal tu-mours, whereas GRD II was predominantlyexpressed in adult brain and differentiated celllines. GRD II expression was higher in 22 weekfetal brain cells than in 20 week fetal braincells. Treatment of the SH-SY5Y neuro-blastoma cells with retinoic acid resulted inan immediate switch from GRD I to GRD IIexpression during the course of the induceddifferentiation. In contrast, Suzuki et al5' founda higher GRD I/GRD II ratio in human adultbrain while GRD II was preferentially ex-pressed in most (13/16) primary brain tumoursanalysed. Gutmann et aP3 used high doses offorskolin or 8-bromo-cAMP to induce bothmyelin PO protein and neurofibromin. Theyobserved that differentiation induced by eithercAMP stimulation or high density culture con-ditions was associated with predominant ex-pression of GRD II and that GRD I expressionwas observed in untreated Schwann cells orthose stimulated with mitogenic doses of for-skolin or 8-bromo-cAMP. Using RNase pro-tection and RT-PCR, Baizer et al4 found thatmost chicken tissues including brain expressedpredominantly GRD II throughout embryonicdevelopment. However, GRD I increased dra-matically in brain during later development.Gutmann et al'7 analysed 22 different human

tissue samples to investigate the expression ofthe transcript isoform 3'ALT by RT-PCR. Thehighest expression of 3'ALT was found in adultcardiac muscle, skeletal muscle, and bladder.This isoform was also expressed in fetal cardiacmuscle, adult left ventricle, and adult cardiacPurkinje cells. Trace levels of the 3'ALT ex-pression was detected in brain and nerve.Northern blot analysis enabled the sim-

ultaneous identification of at least three tran-scripts ofabout 2-9 kb, 1 1 kb, and 13 kb in bothkidney and placenta.48 The 11-13 kb transcriptsare consistent with three transcripts reportedpreviously.56 S1 mapping confirmed that the2-9 kb mRNA was expressed as the NFl tran-script isoform 5'ALT1 lacking the GAP relateddomain.48

Recently, the NF 1 transcript isoform5'ALT2, containing exon 9br, has been ana-lysed by RT-PCR in several normal tissues and

4



g.bmj.com

/J M



nloaded from

http://jmg.bmj.com/


brain tumours.49 A high level expression of thisisoform was observed in the central nervoussystem but no expression was detected in allthe other normal tissues tested. Furthermore,the expression of this transcript decreased inmedulloblastomas and oligodendrogliomas.

THE NFI GENE PRODUCT: NEUROFIBROMIN

Using antisera raised against synthetic peptidesor fusion proteins containing the GRD regionof neurofibromin, a 220-320 kDa protein was

identified in a variety of human and murinetissues such as HeLa cells, NIH3T3 cells, spinalcord, brain, neuroblastoma, and PC 12 phaeo-chromocytoma cell lines.55-60 Immuno-precipitation followed by western blottingallowed the detection ofneurofibromin at high-est levels in adult rat brain and spinal cord.Neurofibromin was also identified in the sciaticnerve and adrenal gland and, to a lesser extent,in liver, spleen, and pancreas. Only minuteamounts of neurofibromin was detected in car-

diac tissue. No neurofibromin was detectablein rat skeletal muscle, lung, kidney, or skin.Similarly, less neurofibromin was detected inhuman peripheral nerve than in human spinalcord. Immunostaining of tissue sections in thesame study detected neurofibromin in neur-

ones, oligodendrocytes, dorsal root ganglia, andnon-myelinating Schwann cells but not in as-

trocytes or myelinating Schwann cells.57The observed size of neurofibromin

(220-320 kDa) is smaller than that (327 kDa)predicted based on the NF1 open readingframe. Recently, the complete coding sequence

has been subcloned into a baculovirus transfervector and expressed in the infected Sf9 insectcells. The purified full length neurofibrominproved to be approximately 220 kDa.6" Thisdiscrepancy is most likely to be caused byprotein folding during electrophoresis. Thereis no evidence of glycosylation or processing ofthe full length neurofibromin.62

In a recent study, neurofibromin was iden-tified in all parts of the brain, especially in largeprojection neurones such as pyramidal andPurkinjee cells.63 Neurofibromin was also loc-alised to keratinocytes and melanocytes in de-veloping rat and human skin.64 Gutmann et

al65 examined the expression of neurofibrominduring in vitro myoblast differentiation.Differentiating C2C 12 cells were shown to up-

regulate the expression ofNFl mRNA and thisresulted in an increase in neurofibromin levels.This was paralleled by a decreased level ofactivated p2 lras. Nakamura et al166 observedstrong expression of neurofibromin specific tothe Schwann cells of normal peripheral nerves

in an N-nitroso-N-ethylurea induced Syrianhamster NF1 model. Although neoplasticSchwann cells exhibited neurofibromin specificsignals, the proportion of positive cells was

diminished.

The availability of antibodies raised againstportions of the NF1 product has also enabledthe subcellular localisation of neurofibromin.DeClue et al55 lysed NIH 3T3 cells in theabsence of detergent and separated them intoparticulate and non-particulate fractions. By

immunoprecipitation analysis, neurofibrominwas found solely in the particulate fractionwhile GAP was primarily in the soluble fraction.Hattori et al967 also located neurofibromin tothe particulate fraction using similar strategies.In another study with 1% NP-40 in variouscell lysates, NFl-GAP activity was found dis-tributed between post 100 000 x g spin super-natant and the particulate fraction.68Immunohistochemical staining showed thelocation of neurofibromin in the cytoplasm ofcells from various tissues.57 Further experi-ments showed the co-localisation of neuro-fibromin with cytoplasmic microtubules.69

NF1 gene functionTUMOUR SUPPRESSIONThe identification of oncogenes and tumoursuppressor genes has directly confirmed theinvolvement of genetic components in tumori-genesis.70 Several genetic models have beenproposed for tumorigenesis. Examples includethe "two hit" model for retinoblastoma (RB),7'the dominant negative model for the mutantp53,72 and the multistep model for the familialadenomatous polyposis (FAP)/colorectal can-cer syndrome (CRC).7 In all these models, theinactivation of one or both alleles of a tumoursuppressor gene is supposed to be involvedin the development of a particular type ofmalignancy.NF1 patients have multiple benign tumours

which predispose to malignancies.'574 Se-quence analysis of NF 1 showed a markedhomology of a 360 residue region of predictedneurofibromin to the catalytic domain ofmam-malian GAP and yeast IRA1 and IRA2 geneproteins which can downregulate p21"activity.44 75 These observations suggest thatNFl is a tumour suppressor gene.A second hit or a somatic mutation in a

tumour suppressor gene may occur in tumourcells such as LOH or other types of geneticalterations which inactivate the gene. The firstsomatic mutations were detected in the NFl-GRD region by SSCP analysis and DNA se-quencing.76 These mutations involve an A toG transition leading to a Lys to Glu substitutionat codon 1423 in exon 24 in colon adeno-carcinoma, myelodysplastic syndrome, andanaplastic astrocytoma, and an A to C trans-version resulting in a Lys to Gln substitutionat the same codon in an anaplastic astrocytoma.This Lys is one of the 14 amino acids in thecatalytic domain that are absolutely conservedacross all members of the GAP family.447577 Invitro site directed mutagenesis analysis of theA to G transition showed that the GAP activityof the mutant NF1-GRD is 200 to 400-foldlower than that of the wild type. Sawada et al8reported a mutational SSCP band in an NF1exon in DNA extracted from a malignantSchwannoma of an NF1 patient. Ludwig et al9screened 57 patients with myelodysplasticsyndrome (MDS) and 27 patients with acutemyelocytic leukaemia (AML) for mutationsin exoti 24 and its flanking sequences usingrestriction enzyme digestion and PCR-SSCPanalysis. None ofthese patients showed a codon



g.bmj.com

/J M



nloaded from

http://jmg.bmj.com/


1423 mutation. However, a patient withchronic myelomonocytic leukaemia (CMML)exhibited a 3 bp deletion within the splice ac-ceptor region in front of exon 24.Some large somatic mutations identified in

NFI by Southern blotting include a loss of thewild type allele in three phaeochromocytomasfrom NF1 patients,80 a large homozygous de-letion extending from the 5' end of the gene tointron 28 in a sporadic malignant melanoma,8'a 200 kb deletion in a neurofibrosarcomafrom a patient with NFl,82 and variants infour out of 10 human neuroblastoma linesexpressing little or no neurofibromin, one ofwhich is a large homozygous deletion.60

However, the actual mechanisms of tumori-genesis, in which NFl is involved, remain un-known. No evidence ofLOH has been observedin neurofibromas.83 There may be several ex-planations for this. Firstly, neurofibromas maybe multicellular in origin, as proposed byFialkow et al.84 However, Skuse et al83 showedthat neurofibromas from all eight NF1 patientstested were ofmonoclonal origin by performingX chromosome specific RFLP analysis andmethylation sensitive digestion with HpaII. Thesecond explanation is that a second mutationin another gene may be required for the genesisof neurofibromas, or they may arise because ofthe loss of function of one allele. The latter istrue in FAP, in which only one mutation isrequired for tumorigenesis.73 The NFI mutantallele may function in a dominant negativefashion causing partial functional inactivationof the wild type allele as does the mutant p53.72The third explanation is that a second mutationin the NFI gene may exist in neurofibromas,which is not identifiable by LOH. Such muta-tions may represent a point mutation or asmall deletion or insertion.

In some malignant tumours from NFIpatients, however, LOH is confined to chro-mosomal region 17p only and one of the twop53 alleles has been deleted.8586 Point muta-tions in the retained p53 allele have also beenobserved in neurofibrosarcomas.85 Severalother tumour suppressor genes may be locatedon chromosome 17.87-90 These observationssuggest that the NF1 mutations may representsmall alterations and may be involved in themultistep process of tumorigenesis in certainmalignant tumours.

It should be noted that although no muta-tions in the NF 1 gene have so far been detectedin some types of tumours such as MDS,AML,79 and familial gliomas,9' further muta-tion analysis in these tumours is required be-cause only a small portion of the NF1 gene hasbeen screened in such tumour studies.

NEUROFIBROMIN AND GAP ACTIVITYNeurofibromin shows homology to variousmembers of the GAP superfamily.444592 Theseproteins include the mammalian GAP, theIRA1 and IRA2 gene products ofSaccharomycescerevisiae, the sarl gene product of Schizo-saccharomyces pombe, and Drosophila GAP1.The homology of neurofibromin to the mam-malian GAP and the Drosophila GAPI is re-

stricted to the catalytic domain of 360 aminoacids. There is more extensive homology ofneurofibromin to the yeast IRAI, IRA2, andsarl gene proteins. The first mammalian GAP,a cytosolic protein with a molecular mass of120 kDa (p120-GAP), was identified by itsproperty of stimulating the conversion of theactive GTP bound form of the ras proteins tothe inactive GDP bound form.93 The IRAl andIRA2 genes encode large proteins of 2938amino acids and 3079 amino acids respectively,which negatively regulate the yeast ras-cyclicAMP pathway.9495 The sarl gene product(80 kDa) regulates rasl but does not appearto participate in the modulation of adenylatecyclase.77 The Drosophila homologue GAPIwas also shown to be a negative regulator ofrasl.96 All these GAP-like proteins can down-regulate the activity of normal ras proteins.They may also function as effectors for rasproteins.93The ras genes were first discovered to be

oncogenes in the acutely transforming Harveyand Kirsten rat sarcoma viruses.97 These viraloncogenes were later found to be derived fromcellular proto-oncogenes in the host's genome.Among more than 30 ras genes, three closelyrelated ones are known as H-ras, K-ras, andN-ras, encoding a 21 kDa protein.879899 Thep21 ras protein (p21ras) binds guanine nuc-leotides with high affinity and possesses anintrinsic GTPase activity, functioning as asingle switch molecule: in the active GTPbound state, it is switched on; in the GDPbound state, it is switched off. This molecularswitch plays a critical role in the control ofcellular growth and differentiation. Activatingmutations in ras lead to aberrant signalling forcell proliferation and are involved in the genesisof many human malignant tumours.'°°

Genetic experiments suggest that the NFl-GRD, as other GAP-like proteins, can interactwith yeast and mammalian ras proteins. Xu etatl5 cloned DNA fragments covering the NF1GAP related domain (NF1-GRD) into theyeast expression plasmid pKT10 and trans-formed the yeast IRAl- and IRA2- mutantswith the constructs. They found that the IRA-mutations were complemented by expressingthe NF1-GRD peptide. Similar results wereobtained in other studies by cloning and ex-pressing cDNA fragments containing the NF 1-GRD in a yeast expression system.'0' 02 Previouswork has shown that either the complete pro-tein or the catalytic domain of the mammalianGAP expressed in yeast can suppress the IRA 1 -

phenotype.94'03 Ballester et al'0l also showedthat the expressed NF1-GRD peptide couldinhibit both wild type and mutant activatedhuman H-ras genes that were expressed inyeast. However, the human GAP can inhibitthe wild type human H-ras protein but not themutant H-ras protein.'03 Nur-E-Kamal et al'04found that both NF1-GRD and a fragment of91 amino acids derived from NF1-GRD wereable to reverse v-Ha-ras induced malignantphenotype. Nakafuku et all5 examined a poolof mutagenised NFl expression plasmids andobtained two mutant NFl cDNA clones thatcould suppress oncogenic ras including RAS-



g.bmj.com

/J M



nloaded from

http://jmg.bmj.com/

Molecular genetics of neurofibromatosis type 1 (NFl)

2Val 19 cells. When expressed in mammaliancells, these mutant NFl-GRDs had the abilityto reverse the morphological phenotype of v-ras transformed NIH3T3 cells. Wood et al'06analysed two ras mutants in yeast. One of thesemutants in the at3 region of ras was found tobe less sensitive to neurofibromin activity andto bind less well to neurofibromin.

Biochemical analysis has shown that the GAPrelated domain ofneurofibromin has the abilityto accelerate the conversion of the active GTPbound p2l's to the inactive form of GDPbound p21ras and thus downregulates the rasactivity. Xu et al5 used the purified glutathionS-transferase/NFI-GRD fusion protein ex-pressed in E coli and observed that this fusionprotein strongly stimulated the GTPase activityof the yeast ras2 and human H-ras proteins.Ballester et al'0' found that a peptide of 412amino acids containing the NFl-GRD, whenexpressed in yeast, increased the H-ras GTPaseactivity. Martin et all'02 used an epitope taggedpeptide of 474 amino acids, which flanks theNF1-GRD and is expressed in the Sf9 insectcells, and showed that the GTPase activity ofwild type ras protein was stimulated by theNF1-GRD. All these studies showed that thepeptides containing the NFl-GRD did notstimulate the oncogenic ras mutants H-rasVal'2,ras2`19 rasAa42, N-rasA,pl2 or N-ras`l2although the GTPase activity of the N-rasAla38effector mutant was slightly increased.7510' 02

Bollag and McCormick68 reported that al-though both human GAP and NF 1-GRD couldbind oncogenic mutants ofp2 Iras, neither couldstimulate their GTPase activity. Andersen etal46 showed that both forms of the NF1-GRD,GRD I and GRD II, could not only com-plement the loss of IRA function but also in-crease the conversion of GTP bound ras to itsGDP bound form, although the GRD II hada weaker effect. In addition, lower levels ofneurofibromin were detected in malignant tu-mours from NF1 patients while normal levelsof both p120-GAP and p21raS were obtained inthese tumours."'107 In these studies, the totalGTPase activating activity was found to bereduced. Furthermore, site directed mu-

tagenesis analysis confirmed the decreasedGTPase activating activity of the mutation atresidue 1423 in the NF1-GRD region.76108 Re-cently, Poullet et al'09 extensively examined therole of lysine at codon 1423 by mutating it toall possible different amino acids and found thatlysine was the only amino acid that produced a

functional NF1 product having normal GAPactivity. They also found that the mutation atcodon 1434 from phenylalanine to serine couldpartially restore GAP activity in the lysine mut-ant at codon 1423.

Additionally or alternatively, neurofibrominmay serve as a downstream effector for p21ras.This model has also been proposed for theinteraction between p120-GAP and p21 ras.93Mutations in the p21ras effector domain in-activated the transforming ability of ras andblocked GTPase activation by p120-GAP. 11'

Similarly, the NF1-GRD peptides did notstimulate the GTPase activity of mutant p21rasin the effector domain.687502 Both p120 GAP

and NF1-GRD could bind well to con-stitutively signalling oncogenic p2 Iras but in-teracted poorly with the effector mutant inwhich signalling is impaired.68The NF1-GRD and p120-GAP showed

different affinities for p2 Iras. The affinity of theNF1-GRD was estimated to be 20-fold higherthan that of p120-GAP, while the specific ac-tivity ofNFI-GRD was approximately 30-foldlower than that ofp 120-GAP. 102 Further studiesindicated that the NF1-GRD binds to p21Ia,proteins up to 300 times more efficiently thanp120-GAP.68 However, at low N-ras con-centrations, the p120-GAP and NF1-GRD ac-tivities were comparable. This suggests thatneurofibromin may be a significant regulatorof p2 Iras activity, particularly at low p2 iras con-centrations. 10The NF1-GRD and human GAP also

showed differences in sensitivity to inhibitionby various lipids. Mitogenesis inducedby lipid related mitogenic agents was found todepend on cellular ras activity."2 It was laterobserved that several lipids inhibited the humanGAP activity."'3 Golubic et all4 reported thatarachidonic acid inhibited the stimulation ac-tivity of the catalytic fragments of the GAP,NF1-GRD, and the yeast IRA2 protein whilephosphatidic acid (containing arachidonic andstearic acid) inhibited NF1-GRD but not thecatalytic fragments ofthe GAP or IRA2 protein.Han et all" obtained similar inhibition of theGAP activity by arachidonic acid and alsofound that GAP stimulation of H-ras was in-creased by prostaglandins PGF2cc or PGA2and decreased by PGI2, whereas NF1-GRDactivity was unaffected by these prostaglandins.In another study, micromolar concentrationsof arachidonate, phosphatidate, and phos-phatidylinositol-4,5-bisphosphate affected onlyNF1-GRD. The p120-GAP activity wasonly weakly affected by these lipids.68 Fur-thermore, the detergent dodecyl P-D-maltosidewas found to be a more stable selective inhibitorspecifically of the NF1-GRD activity. The se-lective inhibition of these lipids can be used todifferentiate the two types ofGTPase activatingactivity in human tissues and also indicatespossible subtle distinctions in the biologicalfunctions of the NF1-GRD and p120-GAP.

Nearly all of the above experiments wereperformed with the NF1-GRD, not the fulllength neurofibromin. Bollag et al6' have puri-fied full length neurofibromin obtained from abaculovirus/Sf9 expression system containingthe complete coding sequence. To examinethe GAP activity of full length neurofibromin,activation of the ras GTPase activity was as-sayed by monitoring phosphate release. Thestimulation activity offull length neurofibrominwas found to be similar to that of the truncatedprotein, NF1-GRD. Full length neurofibrominalso showed similar sensitivity to the detergentdodecyl maltoside. Such observations allow acautious extrapolation from results of ex-periments with the NF1-GRD peptide to in-terpretations of the full length neurofibrominactivity.

Recently, Johnson et al"6 introduced a fulllength NFI cDNA into melanoma cell lines



g.bmj.com

/J M



nloaded from

http://jmg.bmj.com/


deficient in neurofibromin expression. This in-troduction resulted in growth inhibition andinduced differentiation in these cell lines. Over-expression of neurofibromin in NIH3T3 cellswas growth inhibitory, but did not alter thelevel ofGTP bound Ras in the cells. In addition,transformation by v-ras resistant to neuro-fibromin GAP activity was also inhibitedby overexpressed neurofibromin. These resultssuggest that neurofibromin can inhibit Ras de-pendent growth by a mechanism independentof its GTPase activating function.

NEUROFIBROMIN AND TUBULINTubulin has been found to be another inhibitorof the GTPase activating activity of both theNF1-GRD peptide and the full length neuro-fibromin.6' The interaction between neuro-fibromin, either truncated (NF1-GRD) orfull length, and tubulin was shown by theircopurification. This interaction was sensitiveto the microtubule depolymerising agent col-chicine. Bovine brain tubulin could cause sig-nificant inhibition of neurofibromin activitywhile the activity ofp 120-GAP was not affectedby tubulin. Some tubulin binding determinantswere localised to an 80 residue segment im-mediately N-terminal to the GAP related do-main of neurofibromin by the identification ofthe reduced sensitivity of a truncated mutantneurofibromin to tubulin. In the same study,full length neurofibromin was found to be con-siderably less sensitive to phosphatidic acidthan the truncated fragments (NF1-GRD andN-NF1-GRD) in the presence or absence oftubulin. This suggests that tubulin and phos-pholipids indeed bind to distinct sites and thatother domains of full length neurofibrominimpede the interaction of phospholipids withthe GAP related domain.6' The relationshipobserved between neurofibromin and tubulinis consistent with the colocalisation of neuro-fibromin and microtubules.69 Tubulin is theprimary component of microtubules. Micro-tubules play a role in many aspects of cellularorganisation."17 118 The interaction betweenneurofibromin and tubulin may be implicatedin any cellular functions and it has been spec-ulated that neurofibromin is a link betweentubulin and the growth regulator ras.6'

OTHER FUNCTIONS AND MODIFYING GENESNeurofibromin is involved in the control ofcellular growth and differentiation by at leastthree possible mechanisms: as an upstreamdownregulator of p21ras, a downstream effectorofp2 1ras, and a link between tubulin and p2 ras.These mechanisms are not mutually exclusiveand they may function in concert in varioustissues. However, the NF1-GRD is the onlydomain that has been extensively examinedand accounts for only about 13% of neuro-fibromin. To date, the vast majority of muta-tions identified in the NF1 gene do not disruptthe NF1-GRD. Furthermore, an isoform ofneurofibromin lacking the NF1-GRD has beenidentified.48 A recent study showed that whena mutation at codon 1434 from phenylalanine

to serine was introduced into wild type neuro-fibromin, the resulting product acquired theability to suppress the activated phenotypes ofRAS2Val19 cells. However, this suppressiondid not involve ras interaction since the mutantdid not stimulate the intrinsic GTPase activityof RAS2Val19 protein and did not have anincreased affinity for ras proteins.'09 All thesedata suggest that neurofibromin may par-ticipate in other biological activities.A family based analysis of the variable ex-

pression of NF1 suggested a strong geneticcomponent in the variation of NFI expressionbut a minor role that the NFI mutation typemay play. "' Modifying genes may be involvedin the variability of NF1 expression. This issupported by the observation that identicalNF1 specific mutations do not cause the samephenotypes in unrelated NF1 patients.4' 20-122Five NFl patients with large deletions re-moving the whole NFl gene and its flankingsequences showed large numbers of neuro-fibromas.'23 This suggests that deletion of anunknown gene in the NF1 region may affecttumour initiation or development. Neuro-fibromin has several potential sites for phos-phorylation, suggesting that neurofibromincould be regulated by kinases.42 The activity ofp120-GAP was found to be modified byphosphorylation of tyrosine residues when ac-tivated by various growth factors.'24'25 How-ever, neurofibromin lacks similar domains toSH2 and SH3 of GAP which are thought todirect interactions with phosphotyrosine pro-teins involved in signal transduction.'26The function ofneurofibromin may be modi-

fied by different mechanisms. Neurofibromasare known to contain a large number of mastcells, the development ofwhich can be inducedby stem cell factor (SCF). The receptor ofSCFis encoded by the c-kit gene. Hirota et all27observed that the c-kit mRNA was stronglyexpressed in mast cells in the neurofibromasand that the amount ofSCF cDNA was greaterin neurofibroma tissues than in normal skintissues. Ryan et all28 found that SCF was se-creted by both normal Schwann cells and amalignant Schwannoma. The c-kit gene prod-uct was not expressed in normal Schwann cellsbut expressed in the malignant Schwannomas.Another observation suggested that certaingrowth factors such as platelet derived growthfactor and transforming growth factor I1 maybe implicated in the development of neuro-fibromas in NFl.129 These factors may thusfunction by direct or indirect interaction withneurofibromin. In addition, the functions ofthe three embedded genes (EVI2A, EVI2B, andOMGP) at theNF 1 locus and their relationshipwith NF1 remain to be examined.37-39 Thesegenes may participate in the modification ofNFl function. The search for homologous locihas shown NF1 related loci on chromosomes2,12, 14, 15, 20, 21, and 22.130-132 The locus onchromosome 12 contains open reading frameshomologous to NF1 in at least two exons andis expressed in a number of tissues.'3' Suchexpressed sequences may also play a role in theregulation of NFl.

8



g.bmj.com

/J M



nloaded from

http://jmg.bmj.com/


Gerniline mutation analysisMUTATION SPECTRUMMutation analysis in the NFl gene has proveddifficult.'33 The techniques commonly used formutation detection in the NFl gene includetwo relatively simple ones: heteroduplex ana-

lysis (HA) and single strand conformation poly-morphism analysis (SSCP). Reportedly, neitherHA nor SSCP can identify 100% of pointmutations.'34 Recently, several more sensitivetechniques have been adapted to mutation ana-

lysis at the NFl locus. Using denaturing gra-dient gel electrophoresis (DGGE), Valero et

all35 identified NFl specific mutations in fourof 70 unrelated NFl patients by screeningexons 29 and 31. Hoffmeyer et all36 used a

polymorphism in exon 5 of the NFl geneto evaluate allele specific dosage of the NFlmRNA in fibroblast-like cells and observedreduced amounts of mRNA from one of theNFl alleles in six out of eight NFl patients.Using chemical cleavage of mismatch (CCM),Purandare et al'37 screened 70% of the NFlcoding sequence amplified by RT-PCR frommRNA or by PCR from genomic DNA. Theyidentified seven new NFl specific mutationsfrom 25 unrelated NFl patients. In anothermutation study, the protein truncation test(PTT) based on the in vitro transcription-translation of the whole NF1 coding sequenceallowed the identification of truncatedneurofibromin in 14 of 21 NFl patients.'38To date, over 100 NFl specific germline

mutations have been reported to the In-ternational NF1 Genetic Analysis Con-sortium.'39 Seventy-eight NF1 specificmutations have been described (table 1). Themutations include three cytogenetic re-

arrangements, 34 deletions, 12 insertions, twoduplications, and 27 substitutions.

Interestingly, 82% (49/60) of all the fullycharacterised NF 1 specific mutations are eithernonsense or frameshift mutations. These muta-tions presumably lead to severe truncation ofneurofibromin. The large deletions are alsoexpected to cause severe truncation or completeremoval of the gene. Nonsense and frameshiftmutations have been reported to be associatedwith decreased or absent mRNA levels in a

variety of both human159'-62 and rodentgenes. 163-165 Mutations introducing an in-appropriate stop codon have also been reportedto result in undetectable protein.'66 Reducedamounts ofmRNA from one of the NF 1 alleleswere observed in six of eight NFl patients.136However, no mRNA reduction was observedfrom an NF1 patient with an inappropriatestop codon in the centre of NFl in the same

study. In contrast, a decrease in mutant mRNAlevel has been observed in a 10 bp duplicationin exon 38155 and a splice junction mutation inintron 31,157 both of which cause frameshift.

Seven germline substitutions identified in theNFl gene do not generate premature stop

codons (table 1). These mutations replace am-

ino acids that may be important for maintainingthe tertiary structure ofneurofibromin as foundin the case of rhodopsin.'67 Analysis by sitedirected mutagenesis of the substitutionA4267G in exon 24 showed a decreased GAP-

like activity of the mutant fusion protein.76 108 109

The tandem duplication 5095dup42bp inexon 28 does not cause a frameshift and thepredicted neurofibromin has an additional se-quence of 14 amino acids, 13 of which containa motif normally found in neurofibromin fromcodon 1701 to 1713.154 It is not clear howthis mutation affects gene function. Twoexons of the hypoxanthine-guanine phospho-ribosyltransferase (HPRT) gene were observedto be duplicated in a patient with Lesch-Nyhansyndrome. 168 This duplication generated a tran-script of increased length consistent with thesize of the duplicated exons and the resultingenzyme had no detectable activity.'69 Anotherduplication in the collagen type 2A gene hasbeen identified in spondyloepiphyseal dys-plasia, resulting in excessive post-translationalmodification of the gene product.'70 It wassuggested that the tandem duplication in NFlmight result in changes in protein folding, gly-cosylation, or excessive post-translational mo-dification. '54 Alternatively, the duplication mayalter the tertiary structure of mRNA and thusaffect its stability or translation efficiency orboth. 151The 3 bp deletion 2970delAAT results in loss

of the codon ATG (methionine) with a silentchange of codon 990 from ACA (threonine)to ACG (threonine).'22 RT-PCR heteroduplexanalysis has confirmed the expression of thismutation at the RNA level in this study. Theneurofibromin has been found to be structurallyabnormal al Downward, personal com-munication).The 320 bp insertion in intron 32 of NFl

results in a splicing error leading to loss of exon33.6146 Other examples of splicing errors in theNF 1 gene resulted from four intronic sub-stitutions in the splice consensus region.'37 157 158All these splicing abnormalities are expected toresult in truncated neurofibromin.The majority of the NFl specific mutations

are unique and have been found in only onefamily. However, six unrelated NFl patientsfrom different countries carry an identicaltransition (C5839T) in exon 31 (table 1).Three other examples of recurrent causativemutations include the mutations 2027insC inexon 13, 2970delAAT in exon 17, and5449insC in exon 29, each of which was foundin two unrelated NFl patients (table 1).Neurofibromin has been shown to function

75 101 102as a GAP-like protein. However, only sixgermline mutations have been described inthe NFl-GRD region and all the other fullycharacterised small or point mutations do notinterrupt this domain (table 1).

MUTATION MECHANISMSMutagenesis in mammalian genomes is shownto be influenced by genomic position. 71 171Mutations have been found to occur non-ran-domly with respect to the surrounding DNAsequence in human genes.'73-'75 NFl has amutation rate about 100-fold higher than theaverage mutation rate per generation per locus.The mutations observed so far in the NFl geneare not randomly distributed. This suggests



g.bmj.com

/J M



nloaded from

http://jmg.bmj.com/


Table I Germline mutations identified in the NFI gene

Mutation Location Neurofibromin change predicted No of Referencepatients

Chromosomal aberrationst(1;17) (p34.3;qll.2)t(17;22) (qlI.2;qlI.2)del(17) (pter-cen::qll.2

or 12-.qter), +?r(17)(cen-*qll.2 or q12)

Large deletionsdel(731-835) Exon 6del(I 528-1845) Exons lOc-12a517bpdel Introns 32-33Large deletion Introns 2-31 kbdel 3' end40kbdel Undefined40kbdel Containing EVI280kbdel Undefined90kbdel 5' end1 90kbdel UndefinedHemizygosity Intron 27-exon 37>700kbdel NFI and adjacent lociLarge insertions320bpins Intron 32(splicing error: exon skipping)1 Okbins Undefined1 Okbins UndefinedSmall deletions2970delAAT Exon 173050delAATT Exon 18del(4088-4 110) Exon 234190delT Exon 244868delAC Exon 28501 OdelG Exon 285077del(13bp) Exon 285108delAG Exon 285123delCCACC Exon 285679delACTG Exon 305843delAA Exon 315949delA Exon 32?delGAGA Exon 34?delT Exon 347260delT Exon 417745del(1Obp) Exon 44Small insertions2027insC Exon 135449insC Exon 295466insT Exon 295816insG Exon 315852insTT Exon 316519insG Exon 347485insGG Exon 42Duplications5095dup(42bp) Exon 286922dup(10bp) Exon 38Substitutions in coding sequencesC1318T Exon lOaC3049T Exon 18G3497A Exon 21C3826T Exon 22A4256G Exon 24A4267G Exon 24C5242T Exon 29C5260T Exon 29C5380T Exon 29T5795C Exon 31C5839T Exon 31

T6339A Exon 33T?A Exon 33C6427A Exon 34T6511G Exon34G6624A Exon 35C7486T Exon 42G7522T Exon 42Substitutions in introns and splicing errorsA1721+3G Intron II(exon skipping)T5749 + 2G Intron 30(exon skipping)A47768G Intron 31(4 bp insertion in transcript)A6364 + 4G Intron 33(exon skipping)Total

1 301 31

TruncatedTruncatedTruncatedLoss of 28 amino acids

TruncatedNo neurofibromin

Truncated

Loss of MetTruncatedTruncatedTruncatedTruncatedTruncatedTruncatedTruncatedTruncatedTruncatedTruncatedTruncatedTruncatedTruncatedTruncatedTruncated

TruncatedTruncatedTruncatedTruncatedTruncatedTruncatedTr-uncated

14 additional amino acidsTruncated

TruncatedTruncatedGlylIl66AspTruncatedLysl14l9ArgLys 1 423GlutTruncatedTruncatedTruncatedLeu 1932PmoTruncated

TruncatedAsn?LysLeu2l43MetTyr2 17 lAspTruncatedTruncatedTruncated

Truncated

Truncated

Truncated

Truncated

1 140

1 1381 1381 1411 1421 51 1431 51 1431 1441 52 1455 123

1 146

1 1431 143

2 1221 1381 1381 1471 1411 1481 1491 1501 1511 1521 1351 1521 1531 1531 1381 122

2 1382 1431 1431 1501 1201 1371 137

1 1541 155

1 1381 1381 1371 1381 1371 761 1351 1351 1381 46 4, 120, 121,

135, 1561 1381 1411 1431 1431 1381 1371 138

1 137

1 137

1 157

1 158

78

that the sequence composition of NFI alsoinfluences the mutation formation.CpG dinucleotides show a high mutation rate

in the human genome owing to spontaneousdeamination of 5-methylcytosinel76 and about32% of point mutations involve this di-nucleotide. 173 Nine substitutions detected inNFl occur within CpG dinucleotides. Six of

these point mutations involve a single CpG sitein exon 31 representing an identical mutationC5839T (table 1). Other CpG involved sub-stitutions include three C to T transitions inexons 1Oa,138 29,135 and 42,'13 respectively. TheNF1 coding sequence contains 107 CpG di-nucleotides. Several positions are methylatedboth within and around the NFl locus.'77 Fur-

10

??



g.bmj.com

/J M



nloaded from

http://jmg.bmj.com/

Molecular genetics of neurofibromatosis type 1 (NFI)

ther mutational analysis may allow the iden-tification of additional hot spots at such CpGsites in NFl.

Particular sequence patterns in the humangenome are associated with the preferentialformation of insertions and deletions. 174 175

These patterns include direct repeats, pal-indromes, quasi-palindromes, symmetrical ele-ments, and runs of identical bases. Preliminaryexamination of the local sequence environmentflanking some deletions and insertions detectedin NFl has provided similar examples for suchsequences.'22 138 151 152 164

Several NFl related loci have been identifiedon chromosomes 2, 12, 14, 15, 20, 21, and22. 131132 These sequences have homologygreater than 90% with some NFI exons andintronic sequences.'3' The majority of theseloci appear to represent pseudogenes, whichmay be involved in mutagenesis ofNF1 by geneconversion.'1' This mechanism might possiblyexplain the high mutation rate of NFI.

NFI MUTATIONS AND PHENOTYPIC EXPRESSIONWhether particular NF1 mutations are cor-related with the variable clinical expressionclearly needs to be analysed. However, thecurrent paucity of the disease causing muta-tions so far documented in NFI causes diffi-culties in correlating the positions and types ofmutations with the variable clinical features.The variability ofphenotypic expression ofNFleven within families makes this more com-plicated.No clear relationship between mutation size

and clinical features has been found. However,large deletions of more than 700 kb have beendetected in five NFl patients. These mutationsdelete the whole NF1 gene and the three em-bedded genes (EVI2A, EVI2B, and OMGP)as well as the sequences flanking NFI. Thesepatients have mild facial dysmorphology, men-tal retardation, numerous cutaneous neuro-fibromas, and plexiform neurofibromas. 123They show a variable number of physical an-omalies that were not apparently correlatedwith the extent of their deletion but a largenumber of neurofibromas for their age wereobserved in all these patients. This suggests thatthe variable NFl phenotypes may be related toother genes at or near the NFl locus, and thatdeletion of such unknown genes may lead totumour initiation or development.123Two identical mutations (2970delAAT) in

exon 17 (table 1) did not cause similar clinicalfeatures in two unrelated NF1 patients. Oneofthem had developmental abnormalities, mildpulmonary stenosis, multiple CLS, axillaryfreckling, and freckles over the bridge of thenose, but had no neurofibromas, while theother had developmental delay, multiple CLS,and several neurofibromas. The C5839T trans-ition within exon 31 has been identified in six

unrelated NFl patients (table 1). Clinical datawere available from four of these patients. Twoof them show scoliosisl3' 156 and the two othersshow "classical" NF1.132 In addition, affectedmembers in the same family may have differentmanifestations.'78 These observations suggest

that disruption of the NFl gene itself may notbe enough to account for the clinical variability.They also appear to cast doubts on the validityof genotype-phenotype correlation in NFl.

It is not clear whether the genes (EVI2A,EVI2B, and OMGP) embedded in intron 27of the NFl gene play a role in the variable NFlexpression. EVI2A and EVI2B are the humanhomologues of murine Evi-2A and Evi-2Brespectively.3738 Evi-2A is a putative oncogeneand is involved in retrovirus induced murinemyeloid leukaemogenesis.179 Evi-2B lies in themidst of a cluster of viral integration sites iden-tified in retrovirus induced myeloid tumoursand may thus also function as an oncogene inthese tumours.38 EVI2A is highly expressed inbrain, bone marrow, and peripheral blood whileEVI2B is expressed in bone marrow and peri-pheral blood but not in brain. Juvenile chronicmyelogenous leukaemias occur more oftenin NF1 patients than in the generalpopulation.7418018' Recently, LOH was ob-served in bone marrow samples from five outof 11 children with NFl and malignant myeloiddisorders.'82 In these cases, the retained NF1allele was the one inherited from the affectedparent. It is possible that mutations involvingboth NF1 and EVI2A (or EVI2B) may causeNFl/leukaemia syndromes. A 40 kb deletionhas been reported to remove a number ofNF1exons, all ofEVI2A, and the 5' exon ofEVI2B.5Several other large deletions have been reportedto remove the whole NFI gene and the threeembedded genes.'23 However, none of thepatients with these deletions shows features ofleukaemia.OMGP encodes the oligodendrocyte-myelin

glycoprotein (OMGP),39 which is expressed onlyin oligodendrocytes of the central nervous sys-tem. OMGP may function as a cell adhesionmolecule in the myelin of the central nervoussystem.'83 No mutations involving only OMGPhave been reported. As proposed by Viskochilet al,39 transcriptional regulation of these em-bedded genes might play a role in the NFlphenotype. In the past, there have been isolatedreports of patients with NFI and multiplesclerosis (MS). MS is a disorder in whichmultiple plaques of demyelination occur in thecentral nervous system.'84 Perhaps the patientswith NFI and MS have a mutation disruptingboth NF1 and OMGP, which may result inthis particular phenotype.

Animal models for NFIAnimal models of human genetic disease havebeen very helpful in tackling a range of scientificand clinical problems.'85 Several NFI modelshave been reported in different species. A Dro-sophila model has been generated by a lethalinteraction of the eye colour mutant prune(pn) with Killer-of-prune (k-pn) allele of theabnormal wing disc (awd) locus.'86 The awdgene is the Drosophila homologue of the mam-malian tumour metastasis gene nm23 and thepn gene is postulated to encode a GAP-likeprotein. In this model, hypertrophy ofthe neuro-glia and the perineurium of the larval brainwas observed. The lymph glands of the larvae

1 1



g.bmj.com

/J M



nloaded from

http://jmg.bmj.com/


were found to be highly abnormal and mel-anotic pseudotumours were formed upon agingof the larvae. These manifestations involve celltypes in Drosophila equivalent to those affectedby human NF1. Spontaneous neurofibromasand hyperpigmented spots have been observedin the bicolour damselfish.'87 188 The disease inthe damselfish appears to be transmissible andthe tumours tend to be more malignant thanhuman neurofibromas. The Syrian hamsterNFl model has been achieved by N-nitroso-N-ethylurea treatment of pregnant Syrian goldenhamsters.'89190 The hamster offspring developsimilar neurofibromas and pigmented lesionsto those observed in NFI patients, and alsoshow Wilms' tumours and other malignanciesnot typically seen in NFl patients. Point muta-tions in the neo proto-oncogene have beenidentified in these hamster tumours but nomutations have been found in NFl.90 A trans-genic mouse model for neurofibromatosis hasbeen described expressing the HTLV-I tatgene."9' 192 Neurofibromas developed in thetransgenic mice but no other features ofhumanNF1 were observed.The mouse NF 1 gene was identified and

characterised.'9' It encodes an approximately12 kb mRNA with an open reading frame pre-dicting a protein of 2841 amino acids. Se-quence comparison ofNF1 showed more than98% amino acid sequence identity betweenmouse and man. Both the 5' and 3' end un-translated regions of the mouse NF1 transcriptare highly conserved. Hajra et all94 comparedthe promoter regions of human and mouseNF1 genes and found that the transcriptionsites were the same. Both contain a cAMPresponse element, AP2 consensus binding sites,and a serum response element. Recently, genetargeting has been used to disrupt the mouseNFl gene by the insertion of a neomycin cas-sette into the murine DNA sequence equivalentto exon 31 of the human NFl.'95 Mice het-erozygous for the targeted germline mutationshow none of the classical symptoms observedin human NFI, but are highly predisposed tothe formation of various tumour types, espe-cially phaeochromocytoma and myeloid leuk-aemia. Homozygosity for the mutation causescardiac developmental abnormality and mid-gestational embryonic lethality. Branna etall96 used a similar strategy to disrupt the mouseNF1 by generating a null mutation. The het-erozygous mutant mice, aged up to 10 months,do not show any obvious abnormalities butthe homozygous mutant embryos die in utero,exhibiting developmental abnormalities in theheart and various neural crest derived tissues.Another animal homologue ofthe NF1 gene,

the chicken NF1 gene, has been cloned52 54 andagain the predicted amino acid sequence washighly conserved between chick and man. Thechicken NF1 cDNA hybridised to a 12 5 kbtranscript of RNA blots. The potential ex-istence ofspontaneous mutations in the chickenNF1 gene might provide another form ofNF1model. Alternatively, similar models might begenerated by genetically mutating the chickenNF1 gene.

Molecular diagnosis of NFIDNA testing provides a powerful tool in pre-natal diagnosis and in presymptomatic or car-rier screening for genetic diseases. At present,NF 1 allele specific diagnosis with the mutationsdetected in the NF1 gene are only available toa few families. The molecular diagnosis ofNFlhas been mainly based on linkage studies withavailable DNA markers."` Although a bat-tery of closely linked markers has been de-veloped for prenatal and predictive testing,marker informativeness can not be achieved inall NFI families. A number of documentedintragenic polymorphisms appear to be in link-age disequilibrium with each other but no dis-equilibrium was found between the diseaseallele and any of these polymorphisms.20020'

Several PCR based polymorphisms havebeen identified."02-206 The highly informativemicrosatellite in intron 38 has a heterozygosityof up to 0.82.203 Some polymorphisms haverecently been found in the NF1 coding se-quence including a polymorphic RsaI re-striction site in exon 5207 and three single basesubstitutions in exons 23, 24, and 29 re-spectively. 137 These PCR based poly-morphisms, in conjunction with the previouslydescribed markers identified by Southern blot-ting, as well as an increasing number of docu-mented NF1-specific mutations, should makemolecular diagnosis possible in more NFlfamilies (table 2).The current demand for molecular diagnosis

of NF1 is low. Many couples would probablyrequest a prenatal diagnosis if it could predictdisease severity. This is not possible at present.Presymptomatic DNA diagnosis is probablynot going to be in huge demand because theclinical diagnosis of NFI is usually straight-forward, even in early childhood. Both theseverity and prognosis of the disease are ex-tremely variable. The observed variation maynot only be related to different mutations dis-rupting the NF1 gene but also to the in-volvement of unknown modifying genes withinor outside the NFl locus."9 This implies thatmolecular prediction of disease severity andprognosis may either be very complicated oreven impossible.

Therapy for NF1No medical treatment is available to preventor reverse the characteristic lesions of NF1.Instead, medical management is focused ongenetic counselling and the early detection oftreatable complications. Drug therapies dir-ected at upregulating neurofibromin GAP ac-tivity or downregulating p2 Iras activity may beuseful for controlling the growth of neuro-fibromas. Several lipids have been found toinhibit the neurofibromin GAP activity.68 114Tubulin has also been reported to inhibit theneurofibromin GAP activity.6' These findingsmay be helpful to the development of a drugcapable of upregulating or replacing neuro-fibromin. Alternatively, pharmaceutical agentsthat block farnesylation and thus interfere withp21 ras activity may prove to have therapeuticpotential in NFlI13.25 These agents have been

12



g.bmj.com

/J M



nloaded from

http://jmg.bmj.com/


Table 2 Polymorphisms in the NFI gene

Probe sequence Restniction enzyme Heterozygosity Reference

Southern blottingAE25 BglII, PvuII, TaqI 0 5 208NF1C2 EcoRI 0-5 209pRC9.4 BamHI 0 45 210fHB5E3 EcoRI 0 37 200pHa39.3 EcoRI 0-42 200pEV136.5 TaqI 0 43 200fPL37.B2 BamHI 0-42 200pDV1.9 HindIII 0 41 200ff315.1 EcoRI 0 49 200f7G4.GL.2 BglII 0 41 200GE2 EcoRI ? 211PCR basedRsaI polymorphism in exon 5 0 40 207T to C transition in intron 16 0 09 205C to T transition in exon 23 ? 137G to A transition in exon 24 ? 137Compound nucleotide repeat in intron 26 0-72 212Alu repeats in intron 27 High 202CA/GT repeats in intron 27 0-46 204CA/GT repeats in intron 27 0-72 204C to T transition in exon 29 ? 137CA/GT repeats in intron 38 0-82 203RsaI polymorphism in intron 39 0-14 206G to A transition in intron 41 0 43 205

found to inhibit the mitogenic effects of growthfactors and the tumorigenic properties ofneurofibromas. More useful therapies mightinvolve drugs capable of inhibiting famesyltransferase that adds famesyl groups to thep2lras protein. 21

The ultimate cure of an inherited diseasemay be gene therapy. Gene therapy for NF1may not be easy for several reasons. First,because NF1 is an autosomal dominant dis-ease, successful gene therapy may require cor-rection of the mutant NF1 allele. Mutationanalysis in NF1, however, showed that themajority of the causative mutations identifiedare expected to cause severely truncated neuro-fibromin. This suggests that NF1 may resultfrom dosage insufficiency of normal neuro-fibromin. If this is true, introduction of a nor-mal NFl allele may be able to achieve therapywithout correction ofthe mutant allele. Second,NFl often involves multiple tissues with a highvariation. It would be very difficult to deliverthe normal NFl copy to target tissues. Thismight be achieved only by germline gene ther-apy, about which there is still much ethicaldebate. Third, it has been difficult to generate asatisfactoryNFl animal model although severalmodels in different species have been reported.The recently generated mouse models by genetargeting did not mimic human NFl.95l196 Itmay be possible to target the mouse NFl geneusing different mutations identified from NFlpatients and so to create a suitable NFl mousemodel. However, the involvement ofmodifyinggenes in the NFl expression may make it moredifficult to achieve such a model for gene ther-apy.

Possibly, a normal copy of NF1 may betransferred into target tumour tissues to achievetreatment. These tumours include NF1 as-sociated malignant tumours and some sporadicmalignancies not related to NF1. Nur-E-Kamalet all04 observed that when the NF1-GRD wasoverexpressed in v-Ha-Ras transformedNIH3T3 cells, it greatly reduced their abilityto form colonies in a soft agar. Further ex-amination showed that a fragment of 91 aminoacids (NF9 1) derived from the NF1-GRD was

able to reverse the v-Ha-Ras induced malignanttransformation as the NF 1-GRD. These resultssuggest that the NFl -GRD or even the smallerfragment NF9 1 may be used as a potential curefor human cancers caused by mutations in theras gene.

Conclusions and speculationsThe isolation of the NFI gene has advancedthe understanding of the molecular basis ofthe disease and the gene function. Mutationanalysis has allowed the identification of over100 disease causing germline mutations in theNFI gene. However, these mutations are onlyresponsible for a small proportion of all NF1patients. Recent applications of some tech-niques with high sensitivity, such as CCM,DGGE, and PTT, have proved to be highlyefficient in mutation detection in the NFl gene.It is essential to use such techniques for furthermutation search in the gene, especially theregions not yet examined which include thecoding sequence, splice site consensus se-quences, and other essential elements for geneexpression. The functions of the three em-bedded genes (EVI2A, EVI2B, and OMGP)and their relationship to NF 1 remain unknown,and thus mutation analysis in these genes mayalso be interesting. On the other hand, ex-tensive analysis of somatic mutations in a vari-ety of human tumours should help usunderstand the possible mechanisms by whichthe NFl gene is involved in tumorigenesis.Further mutation analysis should lead to mo-lecular diagnosis for more NF1 families andmay also provide clues for tumour diagnosis inthe future.

Evidence is accumulating that NF 1 as atumour suppressor gene plays an important rolein the control of cell growth and differentiation.Neurofibromin genetically and biochemicallyshows GTPase activating activity, down-regulating the activity of normal ras proteins.However, this may not be the only wayneurofibromin functions. Many questions needto be answered. For example, what would bethe effect of the interaction between neuro-

13



g.bmj.com

/J M



nloaded from

http://jmg.bmj.com/


fibromin and tubulin? How does the differentialexpression of the alternatively spliced NFItranscripts contribute to the gene function indifferent tissues? Do these transcript isoformsplay different roles? Are there any modifyinggenes to NF1? If so, where do these genes liein the genome and are they associated with thevariable clinical expression ofNFl? To addresssuch questions more precisely, further studiesare required by using a variety of strategies.Examples include in vitro site directed mu-tagenesis in different domains of the NFI geneand in vivo interruption of the mouse NFIgene by generating different types of mutationsthrough gene targeting. Hopefully, this wouldlead to the generation of a proper mouse modelfor NFI. Breeding heterozygous NF1 miceagainst different strains should help to identifymodifying genes. Further detailed studies onthe interaction of neurofibromin with p21 ras,tubulin, growth factors, or other proteins arealso important for the elucidation of the genefunction. Further knowledge of the gene func-tion may also lead to the development of newtherapy for NFI.

We thank Drs N Thomas and A Clarke for their invaluablecomments on the manuscript. M H Shen is also grateful to DrW Brown for his encouragement during the preparation of thismanuscript. This work was supported by Action Research.

1 Riccardi VM. Neurofibromatosis: clinical heterogeneity.Curr Probl Cancer 1982;7:1-34.

2 Carey JC, Baty BJ, Johnson JP, Morrison T, Skolnick M,Kivlin J. The genetic aspects of neurofibromatosis. AnnNYAcad Sci 1986;486:45-56.

3 Stumpf DA, Alksne JF, Annegers JF, et al. Neuro-fibromatosis. NIH consensus development conferencestatement. Arch Neurol 1988;45:575-8.

4 Cawthon RM, Weiss R, Xu G, et al. A major segmentof the neurofibromatosis type 1 gene: cDNA sequence,genomic structure, and point mutations. Cell 1990;62:193-201.

5 Viskochil D, Buchberg AM, Xu G, et al. Deletions and atranslocation interrupt a cloned gene at the neuro-fibromatosis type 1 locus. Cell 1990;62:187-92.

6 Wallace MR, Marchuk DA, Anderson LB, et al. Type 1neurofibromatosis gene: identification of a large transcriptdisrupted in three NFI patients. Science 1990;249:181-6.

7 Rouleau GA, Merel P, Lutchman M, et al. Alteration ina new gene encoding a putative membrane-organizingprotein causes neurofibromatosis type 2. Nature 1993;363:515-21.

8 Trofatter JA, MacCollin MM, Rutter JL, et al. A novelmoesin-like, ezrin-like, radixin-like gene is a candidatefor neurofibromatosis-2 tumor suppressor. Cell 1993;72:79 1-800.

9 Trattner A, David M, Hoda KE, Bendavid E, SandbankM. Segment neurofibromatosis. J7 Am Acad Dermatol1 990;23:866-9.

10 Watson GH. Pulmonary stenosis, cafe-au-lait spots, anddull intelligence. Arch Dis Child 1967;42:303-7.

11 Allanson JE. Noonan syndrome. JT Med Genet 1987;24:9-13.

12 Pulst SM, Riccardi VM, Fain P, Korenberg JR. Familialspinal neurofibromatosis: clinical and DNA linkage ana-lysis. Neurology 1991;41:1923-7.

13 Brunner HG, Hulsebos T, Steijlen PM, der Kinderen DG,Steen AVD, Hamel BCJ. Exclusion of the neuro-fibromatosis locus in a family with inherited cafe-au-laitspots. AmJMed Genet 1993;46:472-4.

14 Purcell S, Dixon SL. Schwannomatosis. An unusual variantof neurofibromatosis or a distinct clinical entity. ArchDermatol 1989;125:390-3.

15 Riccardi VM, Eichner JE. Neurofibromatosis: phenotype, nat-ural history and pathogenesis. Baltimore: Johns HopkinsUniversity Press, 1986.

16 Huson SM, Harper PS, Compton DAS. Von Reckling-hausen neurofibromatosis: a clinical and population studyin South-East Wales. Brain 1988;111:1355-81.

17 Huson SM, Hughes RAC. The neumfibromatosis: a patho-genetic and clinical overview. London: Chapman & Hall,1994.

18 Jadayel D, Fain P, Upadhyaya M, et al. Paternal origin ofnew mutations in Von Recklinghausen neurofibromatosis.Nature 1990;343:558-9.

19 Stephens K, Kayes L, Riccardi VM, Rising M, Sybert VP,Pagon RA. Preferential mutation ofthe neurofibromatosistype 1 gene in paternally derived chromosomes. HumGenet 1992;88:279-82.

20 Riccardi VM. Cutaneous manifestations of neuro-fibromatosis: cellular interaction, pigmentation, and mastcells. Birth Defects 1981;17:129-45.

21 Huson SM, Compton DAS, Clarke P, Harper PS. A geneticstudy of von Recklinghausen neurofibromatosis in southeast Wales. I. Prevalence, fitness, mutation rate, and theeffect of parental transmission on severity. 7 Med Genet1989;26:704-1 1.

22 Huson SM, Compton DAS, Harper PS. A genetic studyof von Recklinghausen neurofibromatosis in south eastWales. II. Guidelines for genetic counselling. JMed Genet1989;26:712-21.

23 Safarazi M, Huson S, Edwards JH. An exclusion mapfor von Recklinghausen neurofibromatosis. J Med Genet1987;24:515-20.

24 Seizinger BR, Rouleau GA, Ozelius LJ, et al. Geneticlinkage of von Recklinghausen neurofibromatosis to thenerve growth factor receptor gene. Cell 1987;49:589-94.

25 Barker D, Wright E, Nguyen K, et al. Gene for vonRecklinghausen neurofibromatosis is in the peri-centromeric region of chromosome 17. Science 1987;236:1100-2.

26 White R, Nakamura Y, O'Connell P, et al. Tightly linkedmarkers for the neurofibromatosis type 1 gene. Genomics1987;1:364-7.

27 VanTuinen P, Rich DC, Summers KM, Ledbetter DH.Regional mapping panel for human chromosome 17:application to neurofibromatosis type 1. Genomics 1987;1:374-81.

28 Skolnick MH, Ponder BAJ, Seizinger B. Linkage of NF1to 12 chromosome 17 markers: a summary of eightconcurrent reports. Genomics 1987;1:382-3.

29 Goldgar DE, Green P, Parry DM, Mulvihill JJ. Multipointlinkage analysis in neurofibromatosis type 1: an in-ternational collaboration. AmJ7Hum Genet 1989;44:6-12.

30 Schmidt MA, Michels W, Deward W. Cases of neuro-fibromatosis with rearrangement of chromosome 17 in-volving band 17ql1.2. Am JMed Genet 1987;28:771-7.

31 Ledbetter DH, Rich DC, O'Connell P, Leppert M, CareyJC. Precise localisation of NFI to 17q11.2 by balancedtranslocation. Am J Hum Genet 1989;44:20-4.

32 O'Connell P, Leach R, Cawthon RM, et al. Two NF1translocations map within a 600-kilobase segment of17qll.2. Science 1989;244:1087-8.

33 Fountain JW, Wallace MR, Bruce MA, et al. Physicalmapping of a translocation breakpoint in neuro-fibromatosis. Science 1989;244: 1085-7.

34 Collins FS, O'Connell P, Ponder BAJ. Seizinger BR. Pro-gress towards identifying the neurofibromatosis (NFl)gene. Trends Genet 1989;5:217-21.

35 Wallace MR, Andersen LB, Fountain JW, et al. A chro-mosome jump crosses a translocation breakpoint in thevon Recklinghausen neurofibromatosis region. GenesChrom Cancer 1990;2:271-7.

36 O'Connell P, Viskochil D, Buchberg AM, et al. The humanhomolog of murine Evi-2 lies between two vonRecklinghausen neurofibromatosis translocations.Genomics 1990;7:547-54.

37 Cawthon RM, O'Connell P, Buchberg AM, et al. Iden-tification and characterisation of transcripts from theneurofibromatosis 1 region: the sequence and genomicstructure of EVI2 and mapping of other transcripts. Geno-mics 1990;7:555-65.

38 Cawthon RM, Andersen LB, Buchberg AM, et al. cDNAsequence and genomic structure of Evi2B, a gene lyingwithin an intron of the neurofibromatosis type 1 gene.Genomics 1991;9:446-60.

39 Viskochil D, Cawthon R, O'Connell P, et al. The geneencoding the oligodendrocyte-myelin glycoprotein is em-bedded within the neurofibromatosis type 1 gene. MolCell Biol 1991;11:906-12.

40 Xu G, O'Connell P, Stevens J, White R. Characterisationof human adenylate kinase 3 (AK3) cDNA and mappingof the AK3 pseudogene to an intron of the NF1 gene.Genomics 1992;3:537-42.

41 Marchuk DA, Tavakkol R, Wallace MR, et al. A yeastartificial chromosome contig encompassing the type 1neurofibromatosis gene. Genomics 1992;13:672-80.

42 Marchuk DA, Saulino AM, Tavakkol R, et al. cDNAcloning of the type 1 neurofibromatosis gene: completesequence of NFl gene product. Genomics 1991;11:931-40.

43 Viskochil D, White R, Cawthon R. The neurofibromatosistype 1 gene. Annu Rev Neurosci 1993;16:183-205.

44 Xu G, O'Connell P, Viskochil D, et al. The neuro-fibromatosis type 1 gene encodes a protein related toGAP. Cell 1990;62:599-608.

45 Buchberg AM, Cleveland LS, Jenkins NA, Copeland NG.Sequence homology shared by neurofibromatosis type 1gene and IRA-1 and IRA-2 negative regulators of theRAS cyclic AMP pathway. Nature 1990;347:291-4.

46 Andersen LB, Ballester R, Marchuk DA, et al. A conservedalternative splice in the von Recklinghausen neuro-fibromatosis (NFl) gene produces two neurofibrominisoforms, both with GAP activity. Mol Cell Biol 1993;13:487-95.

47 Gutmann DH, Andersen LB, Cole JL, Swaroop M, CollinsFS. An alternatively-spliced mRNA in the carboxy ter-minus of the neurofibromatosis type 1 (NFl) gene isexpressed in muscle. Hum Mol Genet 1993;2:989-92.

48 Suzuki H, Takahashi K, Kubota Y, Shibahara S. Molecularcloning of a cDNA coding for neurofibromatosis type 1protein isoform lacking the domain related to ras GTPase-activating protein. Biochem Biophys Res Commun 1992;187:984-90.

14



g.bmj.com

/J M



nloaded from

http://jmg.bmj.com/


49 Danglot G, Regnier V, Fauvet D, Vassal G, Kujas M,Bemheim A. Neurofibromatosis 1 (NFl) mRNAs ex-pressed in the central nervous system are differentiallyspliced in the 5 part of the gene. Hum Mol Genet 1995;4:915-20.

50 Nishi T, Lee PSY, Oka K, et al. Differential expression oftwo types of the neurofibromatosis type 1 (NFl) genetranscripts related to neuronal differentiation. Oncogene1991;6:1555-9.

51 Suzuki Y, Suzuki H, Kayama T, Yoshimoto T, ShibaharaS. Brain tumors predominantly express the neuro-fibromatosis type 1 gene transcripts containing the 63-base insert in the region coding for GTPase activatingprotein-related domain. Biochem Biophys Res Commun1991;181:955-61.

52 Schafer GL, Ciment G, Stocker KM, Baizer L. Analysisof the sequence and embryonic expression of chickenneurofibromin mRNA. Mol Chem Neuropathol 1993;18:267-78.

53 Gutmann DH, Tennekoon GI, Cole JL, Collins FS,Rutkowski JL. Modulation of the neurofibromatosis type1 gene product, neurofibromin, during Schwann celldifferentiation. 7 Neurosci Res 1993;36:216-23.

54 Baizer L, Ciment G, Hendrickson SK, Schafer GL. Regu-lated expression of neurofibromin type 1 transcript in thedeveloping chicken brain. J Neurochem 1993;61 :2054-60.

55 DeClue JE, Cohen BD, Lowry DL. Identification andcharacterisation of the neurofibromatosis type 1 proteinproduct. Proc Natl Acad Sci USA 199 1;88:9914-18.

56 Gutmann DH, Wood DL, Collins FS. Identification ofthe neurofibromatosis type 1 gene product. Proc NatlAcad Sci USA 199 1;88:9658-62.

57 Daston MM, Scrable H, Nordlund M, Sturbaum AK,Nissen LM, Ratner N. The protein product of the neuro-fibromatosis type 1 gene is expressed at the highest abund-ance in neurons, Schwann cells, and oligodendrocytes.Neuron 1992;8:415-28.

58 Basu TN, Gutmann DH, Fletcher JA, Glover TW, CollinsFS, Downward J. Aberrant regulation of ras proteins inmalignant tumor cells from type 1 neurofibromatosispatients. Nature 1992;356:713-15.

59 Hattori S, Maekawa M, Nakamura S. Identification ofneurofibromatosis type 1 gene product as an insolubleGTPase-activating protein toward ras p21. Oncogene1992;7:481-5.

60 The I, Murthy AE, Hannigan GE, et al. Neurofibromatosistype 1 gene mutations in neuroblastoma. Nature Genet1993;3:62-6.

61 Bollag G, McCormick F, Clark R. Characterization of full-length neurofibromin: tubulin inhibits Ras GAP activity.EMBO J 1993;12:1923-7.

62 Gutmann DH, Collins FS. The neurofibromatosis type 1gene and its protein product, neurofibromin. Neuron1993;10:335-43.

63 Nordlund M, Gu X, Shipley MT, Ratner N. Neuro-fibromin is enriched in the endoplasmic reticulum ofCNA neurons. J Neurosci 1994;13: 1588-600.

64 Malhotra R, Ratner N. Localisation of neurofibromin tokeratinocytes in developing rat and human skin. _7 InvestDermatol 1994;102:812-18.

65 Gutmann DH, Cole JL, Collins FS. Modulation of neuro-fibromatosis type 1 gene expression during in vitro myo-blast differentiation. _7 Neurosci Res 1994;37:398-405.

66 Nakamura T, Nemoto T, Arai M, et al. Specific expressionofthe neurofibromatosis type 1 gene (NF 1) in the hamsterSchwann cell. Am _7 Pathol 1994;144:549-55.

67 Hattori S, Ohmi N, Maekawa M, Hoshino M, KawakitaM, Nakamura S. Antibody against neurofibromatosis type1 gene product reacts with a triton-insoluble GTPaseactivating protein toward ras p21. Biochem Biophys ResCommun 1991;177:83-9.

68 Bollag G, McCormick F. Differential regulation of rasGAPand the neurofibromatosis gene product activities. Nature199 1;351:576-9.

69 Gregory PE, Gutmann DH, Mitchell A, et al. The neuro-fibromatosis type 1 gene product (neurofibromin) as-sociated with microtubules. Som Cell Mol Genet 1993;19:265-74.

70 Diamandis EP. Oncogenes and tumor suppressor genes:new biochemical tests. Crit Rev Clin Lab Sci 1992;29:269-305.

71 Knudson AG Jr. Mutation and cancer: statistical study ofretinoblastomas. Proc NatlAcad Sci USA 197 1;68:820-3.

72 Oren M. p53: the ultimate tumor suppressor gene? FASEB.7 1992;6:3169-76.

73 Fearon ER, Vogelstein B. A genetic model for colorectaltumorigenesis. Cell 1990;61:759-67.

74 Bader JL, Miller RW. Neurofibromatosis and childhoodleukemia. _7 Pediatr 1978;92:925-9.

75 Xu G, Lin B, Tanaka K, et al. The catalytic domain of theneurofibromatosis type 1 gene product stimulates rasGTPase and complements ira mutants of S. cerevisiae.Cell 1990;63:835-41.

76 Li Y, Bollag G, Clark R, et al. Somatic mutations in theneurofibromatosis 1 gene in human tumors. Cell 1992;69:275-81.

77 Wang Y, Boguski M, Riggs M, Rodgers L, Wigler M.sarl, a gene from the yeast Schizosaccharomyces pombeencoding a protein that regulates rasl. Cell Reg 1991,2:453-65.

78 Sawada S, Monda M, Niimura M. Molecular geneticanalysis of the von Recklinghausen neurofibromatosis(NF 1) gene using polymerase chain reaction-single strandconformation polymorphism (PCR-SSCP) method. _7Dermatol 1994;21:294-300.

79 Ludwig L, Jassen JW, Schulz AS, Bartram CR. Mutationswithin the FLR exon of NF1 are rare in myelodysplasticsyndromes and acute myelocytic leukemias. Leukenmia1993;7: 1058-60.

80 Xu W, Mulligan M, Ponder BAJ. Loss of NF1 alleles inphaeochromocytomas from patients with type 1 neuro-fibromatosis. Genes Chrom Cancer 1992;4:337-42.

81 Andersen LB, Fountain JW, Gutmann DH, et al. Mutationsin the neurofibromatosis 1 gene in sporadic malignantmelonoma cell lines. Nature Genet 1993;3:118-21.

82 Legius E, Marchuk DA, Collins FS, Glover TW. Somaticdeletion of the neurofibromatosis type 1 gene in a neuro-fibrosarcoma supports a tumour suppressor gene hypo-thesis. Nature Genet 1993;3:122-6.

83 Skuse GR, Kosciolek BA, Rowley DT. The neurofibromain von Recklinghausen neurofibromatosis has a unicellularorigin. Ani 7 Hum Genet 199 1;49:600-7.

84 Fialkow PJ, Sagebiel RW, Gartler SM, Rimoin DL. Mul-tiple cell origin of hereditary neurofibromas. N Engl.Med 1971;284:298-300.

85 Menon AG, Anderson KM, Riccardi VM, et al. Chro-mosome 17p deletions and p53 gene mutations associatedwith the formation of malignant neurofibrosarcomas invon Recklinghausen neurofibromatosis. Proc Natl AcadSci USA 1990;87:5435-9.

86 Glover TW, Stein CK, Legius E, Anderson LB, BreretonA, Johnson S. Molecular and cytogenetic analysis oftumours in von Recklinghausen neurofibromatosis. GenesChrom Cancer 1991;3:62-70.

87 Hall A. ras and GAP - who's controlling whom? Cell 1990;61:921-3.

88 Biegel JA, Burk CD, Barr FG, Emanuel BS. Evidence fora 17p tumor related locus distinct from p53 in pediatricprimitive neuroectodermal tumors. Cancer Res 1992;52:3391-5.

89 Cogen PH, Daneshvar L, Metzger AK, Duyk G, EdwardsMS, Sheffield VC. Involvement of multiple chromosome17p loci in medulloblastoma tumorigenesis. An _7 HumGenet 1992;50:584-9.

90 Chen P, Ellmore N, Weissman BE. Functional evidencefor a second tumor suppressor gene on human chro-mosome 17. Mol Cell Biol 1994;14:534-42.

91 van Meyel DJ, Ramsay DA, Chambers AF, MacdonaldDR, Cairmcross JG. Absence of hereditary mutations inexons 5 through 9 of the p53 gene and exon 24 of theneurofibromin gene in families with glioma. Ann Neurol1994;35: 120-2.

92 Hall A. Signal transduction through small GTPases - atale of two GAPs. Cell 1992;69:389-91.

93 McCormick F. Ras GTPase activating protein: signal trans-mitter and signal terminator. Cell 1989;56:5-8.

94 Tanaka K, Nakafuku M, Satoh T, et al. S. cerevisiae genesIRA1 and IRA2 encode proteins that may be functionallyequivalent to mammalian ras GTPase activating protein.Cell 1990;60:803-7.

95 Tanaka K, Nakafuku M, Tamanoi F, Kaziro Y, MatsumotoK, Toh-e A. IRA2, a second gene of Sacchromycescerevisiae that encodes a protein with a domain ho-mologous to mammalian ras GTPase activating protein.Mol Cell Biol 1990;1O:4303-13.

96 Gaul U, Mardon G, Rubin G. A putative ras GTPaseactivating protein acts as a negative regulator of signallingby the sevenless receptor tyrosine kinase. Cell 1992;68:1001-19.

97 Ellis RW, Defeo D, Shih TY, et al. The p21 src genesof Harvey and Kirsten sarcoma viruses originate fromdivergent members of a family of normal vertebrae genes.Nature 1981;292:506-11.

98 Barbacid M. Ras genes. Annu Rev Biochenm 1987;56:779-828.

99 Bourne HR, Sanders DA, McCormick F. The GTPasesuperfamily: a conserved switch for diverse cell functions.Nature 1990;384: 125-32.

100 Bos JL. Ras oncogenes in human cancer. A review. CancerRes 1989;49:4682-9.

101 Ballester R, Marchuk D, Boguski M, et al. The NFl locusencodes a protein functionally related to mammalian GAPand yeast IRA proteins. Cell 1990;63:851-9.

102 Martin GA, Viskochil D, Bollag G, et al. The GAP-relateddomain of the neurofibromatosis type 1 gene productinteracts with ras p21 regulation. Cell 1990;63:843-9.

103 Ballester R, Michaeli T, Ferguson K, Xu HP, McCorickF, Wigler M. Genetic analysis of mammalian GAP ex-pressed in yeast. Cell 1989;59:681-6.

104 Nur-E-Kamal MS, Varga M, Maruta H. The GTPase-activating NFl fragment of 91 amino acids reverses v-Ha-Ras-induced malignant phenotype. J BiolChem 199 3;268:22331-7.

105 Nakafuku M, Nagamine M, Ohtoshi A, Tanaka K,Toh-e A, Kaziro Y. Suppression of oncogenic Ras bymutant neurofibromatosis type 1 genes with single aminoacid substitutions. Proc Natl Acad Sci USA 1993;90:6706-10.

106 Wood DR, Poullet P, Wilson BA, et al. Biochemical char-acterisation of yeast ras2 mutants reveals a new region ofras protein involved in the interaction with GTPase-activating proteins. _7 Biol Chemi 1994;269:5322-7.

107 DeClue JE, Papageorge AG, Flether JA, et al. Abnormalregulation ofmammalian p2 1ras contributes to malignanttumour growth in von Recklinghausen (type 1) neuro-fibromatosis. Cell 1992;69:265-73.

108 Gutmann DH, Boguski M, Marchuk D, Wigler M, CollinsFS, Ballester R. Analysis of the neurofibromatosis type 1(NFl1) GAP-related domain by site-directed mutagenesis.Oncogene 1993;8:761-9.

15



g.bmj.com

/J M



nloaded from

http://jmg.bmj.com/


109 Poullet P, Lin B, Esson K, Tamanoi F. Functional sig-nificance of lysine 1423 of neurofibromin and char-acterisation of a second site suppressor which rescues

mutations at this residue and suppresses RAS2Val-19-activated phenotypes. Mol Cell Biol 1994,14:815-21.

110 Adari H, Lowry DR, Willumsen BM, Der CJ, McCormickF. Guanosine triphosphatase activating protein (GAP)interacts with the p21 effector binding domain. Science1988;240:518-2 1.

111 Cales C, HancockJF, Marshall CJ, Hall A. The cytoplasmicprotein GAP is implicated as the target for regulation bythe ras gene product. Nature 1988;332:548-51.

112 Yu CL, Tsai MH, Stacey DW. Cellular ras activity andphospholipid metabolism. Cell 1988;52:63-71.

113 Tsai MH, Yu CL, Wei FS, Stacey DW. The effect ofGTPase activating protein upon ras is inhibited by mi-togenically responsive lipids. Science 1989;243:522-6.

114 Golubic M, Tanaka K, DobrowolskiS, et al. The GTPasestimulatory activities of the neurofibromatosis type 1 andthe yeast IRA2 proteins are inhibited by arachidonic acid.EMBO]_ 1991;1O:2897-903.

115 HanJW, McCormick F, Masara IG. Regulation of Ras-GAP and neurofibromatosis type 1 gene product byeicosanoids. Science 1991;252:576-9.

116 Johnson MR, DeClue JE, FelzmannS, et al. Neurofibromincan inhibit Ras-dependent growth by a mechanism in-dependent of its GTPase-accelerating function. Mol CellBiol 1994;14:641-5.

117 Carney DH, Crossin KL, Ball R, Fuller GM, Albrecht T,Thompson WC. Changes in the extent of microtubuleassembly can regulate initiation of DNA synthesis. AnnNYAcad Sci 1986;466:919-32.

118 Gelfand VI, Bershadsky AD. Microtubule dynamics -

mechanism, regulation and function. Annu Rev Cell Biol1991;7:93-116.

119 Easton DF, Ponder MA, Huson SM, Ponder BAJ. Ananalysis of variation in expression of neurofibromatosis(NF) type1 (NFl): evidence for modifying genes. An, _7Hun,1 Genet 1993;53:305-13.

120 Ainsworth PJ, Rodenhiser DI, Costa MT. Identificationand characterisation of sporadic and inherited mutationsin exon 31 of the neurofibromatosis (NFI) gene. HnwnGeniet 1993;91:151-6.

121 Horiuchi T, Hatta N, Matsumoto M, et al. Nonsensemutations at Arg-1947 in two cases of familial neuro-

fibromatosis type1 in Japanese. Hun,iiGeniet 1994;93:81-3.

122 Shen MH, Harper PS, Upadhyaya M. Neurofibromatosistype 1 (NF1): the search for mutations by PCR-het-eroduplex analysis on Hydrolink gels. Hun, MolGenet1993;2: 1861-4.

123 Kayes LM, Burke W, Riccardi VM, et al. Deletions span-

ning the neurofibromatosis1 gene: identification andphenotype of five patients. An,7 HuniGeniet 1994;54:424-36.

124 Ellis C, Maron M, McCormick F, Pawson T. Phos-phorylation of GAP and GAP associated proteins bytransforming and mitogenic tyrosine kinases. Nature 1990;343:377-81.

125 MolloyCJ, Bottaro DP, Fleming TP, Marshall MS, GibbsJB, Aaronson SA. PDGF induction of tyrosine phos-phorylation of GTPase activating protein. Natlure 1989;342:711-14.

126 Koch CA, Andersen D, Moran MF, Ellis C, Dawson T.SH2 and SH3 domains: elements that control interactionsof cytoplasmic signaling proteins. Science 1991;252:669-74.

127 HirotaS, NomuraS, Asada H, Ito A, MoriiE, KitamuraY. Possible involvement of c-kit receptor and its ligandin increase of mast cells in neurofibroma tissues. ArchPathol Lab Med 1993;117:996-9.

128 RyanJJ, Klein KA, NeubergerTJ, et al. Role for the stemcell factor/KIT complex in Schwann cell neoplasia andmast cell proliferation associated with neurofibromatosis..7 Neurosci Res 1994;37:415-32.

129 Kadono T, Soma Y, Takkehara K, Nakagawa H, IshibashiY, Kikuchi K. The growth regulation of neurofibromacells in neurofibromatosis type-1: increased responses toPDGF-BB and TGF-j31. Biochen,i Biophys Res Coniniu,,1994;198:827-34.

130 Legius E, Marchuk DA, Hall BK, et al. NFl-related locuson chromosome 15. Genomiics 1992;13:1316-18.

131 Cummings LM, Glatfelter A, Marchuk DA. NFl-relatedloci on chromosomes 2, 12, 14, 15, 20, 21, and 22: apotential role for gene conversion in the high spontaneousmutation rate of NFl? Ani Hun, Genet (suippl) 1993;53:672.

132 Gasparini P, Grifa A, Origone P, Coviello D, AntonacciR, Rocchi M. Detection of a neurofibromatosis type 1(NF1) homologous sequence by PCR: implications forthe diagnosis and screening of genetic diseases. Mol CellProbes 1993;7:415-18.

133 Upadhyaya M, Shaw DJ, Harper PS. Molecular basis ofneurofibromatosis type 1 (NFl): mutation analysis andpolynorphisms l Hu7 n IMtat83-101.

134 Cotton RGH. Current methods of mutation detection.Muitat Res 1993;285:125-44.

135 Valero MC, Velasco E, Moreno F, Hernnandez-Chico C.Characterisation of four mutations in the neuro-fibromatosis type 1 gene by denaturing gradient gelelectrophoresis (DGGE). Hun,i Mol Geniet 1994;3:639-4 1.

136 Hoffmeyer 5, Assum G, Kaufmann D, KroneW. Unequalexpression of NFl alleles. Nature Geniet 1994;6:331.

137 Purandare SM, Lanyon WG, ConnorJM. Characterisationof inherited and sporadic mutations in neurofibromatosistype-1. Hun, Mol Ge,,et 1994;3:1109-15.

138 Heim RA, Kam-Morgan LNW, Binnie CG, et al. Dis-tribution of 13 truncating mutations in the neuro-fibromatosis 1 gene. Hun, Mol Genet 1995,4:975-81.

139 International NFl Genetic Analysis Consortium. News-letter. Communicated by Dr Bruce Korf of Harvard Med-ical School, USA, June 1995.

140 Andersen LB, Tommerup N, Koch J. Formation of aminichromosome by excision of the proximal region of1 7q in a patient with von Recklinghausen neuro-fibromatosis. Cytogenet Cell Genet 1990;53:206-10.

141 Xu W, Qi Y, Liu L, et al. Molecular analysis of neuro-fibromatosis type 1 mutations. Hun, Mutat 1992;1:474-7.

142 HoffmeyerS, Assum G, Kaufmann D, Schwenk K, KroneW. A deletion in the 5'-region of the neurofibromatosistype 1 (NF1) gene. Hun, Genet 1994,94:97-100.

143 Upadhyaya M, Shen MH, Cherryson A, et al. Analysis ofmutations at the neurofibromatosis 1 (NFl) locus. H,on,Mol Gen,et 1992;1:735-40.

144 Upadhyaya M, Cherryson A, Broadhead W, et al. A 90 kbdeletion associated with neurofibromatosis type 1.Y MedGetnet 1990;27:738-41.

145 Lazaro C, Gaona A, Ravella A, et al. Novel alleles, hemi-zygosity and deletions at an Alu-repeat within the neuro-fibromatosis type 1 (NFl) gene. H,i,,n Mol Ge,,et 1993,2:725-30.

146 Wallace MR, Andersen LB, Saulino AM, Gregory PE,Glover TW, Collins FS. A de novo Alu insertion resultsin neurofibromatosis. Natzure 1991;353:864-6.

147 Anglani F, Murgia A, BedinS, et al. A new disease-causingmutation in the GAP-related domain of the NFI gene.Hun, Mol Genet 1993;2:1057-9.

148 Colman SD, Collins FS, Wallace MR. Characterisation ofa single base-pair deletion in neurofibromatosis type 1.Hun, Mol Geniet 1993;2:1709-11.

149 Shen MH, Upadhyaya M. A de novo nonsense mutationin exon 28 of the neurofibromatosis type1 (NFI) gene.Hun, Genet 1993;92:410-12.

150 Zhong J, Spiegel R, BoltshauserE, Schmid W. Two novelmutations: 5108delAG and 5816insG in the NFl genedetected by SSCP analysis.Hu.n, Mol Ge,iet 1993;2:149 1-2.

151 Stark M, Assum G, Krone W. A small deletion and anadjacent base exchange in a potential stem-loop regionof the neurofibromatosis1 gene. Hun,i Ge,,et 1991;87:685-7.

152 Hatta N, Horiuchi T, Fujita S. Analysis of NFl genemutations in neurofibromatosis type1 patients in Japan.Biochen, Biophys Res Connunin 1994;199:207-12.

153 Stephens K, Riccardi VM, Ehrlich P. Deletions in exon34 of the neurofibromatosis1 gene. An,Y H,o,, GenetSuippl 1992;51:114.

154 Tassabehji M, Strachan T, Sharland M, et al. Tandemduplication with a neurofibromatosis type1 (NFl) geneexon in a family with features of Watson syndrome andNoonan syndrome. An,i7 Huni, Genet 1993;53:90-5.

155 Legius E, Hall BK, Wallace MR, Collins FS, Glover TW.Ten base pair duplication in exon 38 of the NFl gene.Hurn, Mol Geniet 1994;3:829-30.

156 Estivill X,Lazaro C, Casals T, Ravella A. Recurrence ofa nonsense mutation in the NFl gene causing neuro-fibromatosis type 1. Hun,Genet 1991;88:185-8.

157 Ainsworth P, Rodenhiser D, Stuart A, Jung J. Char-acterization of an intron 31 splicejunction mutation inthe neurofibromatosis type1 (NF 1) gene. Hu,, M,sl Genet1994;3:1179-81.

158 Hutter P, Antonarakis SE, Delozier-Blanchet CD, MorrisMA. Exon skipping associated with A-G transition at+4 of theIVS33 splice donor site of the neurofibromatosistype 1 (NFl) gene. Hun, Mol Geniet 1994;3:663-5.

159 Atweh GF, Brickner HE, Zhy XX, Kazazian HH Jr, ForgetBG. New amber mutation in a3-thalassemic gene withnonmeasurable levels of mutant messenger RNA in vivo.YCli,, Invest 1988;82:557-61.

160 BasergaSJ, Benz EJ Jr. ~-globin nonsense mutation: de-ficient accumulation of mRNA occurs despite normalcytoplasmic stability. Proc Natl Acad Sci USA 1992;89:2935-9.

161 Hamosh A, Trapnell BC, Zeitlin PL, et al. Severe deficiencyof cystic-fibrosis transmembrane conductance regulatormessenger RNA carrying nonsense mutationsr553x andww163x in respiratory epithelial cells of patients withcystic-fibrosis. _I Clin, Invest 1991;88:1880-5.

162 Kadowaki T, Kadowaki H, Taylor SI. A nonsense mutationcausing decreased levels of insulin receptor messengerRNA-detection by a simplified technique for direct se-quencing of genomic DNA amplified by the polymerasechain reaction. ProcNatl Acad Sci USA 1990;87:658-62.

163 Baumann B, Potash MJ, Kohler G. Consequences offrameshift mutations at the immunoglobulin heavy-chainlocus of the mouse. EMBO Y 1985;4:351-9.

164 Lim 5, Mullins JJ, Chen CM, Gross KW, Maquat LE.Novel metabolism of several beta-thalassemic beta-globinmessenger RNAs in the erythroid tissues of transgenicmice. EMBOJ_ 1989;8:2613-19.

165 Urlaub G, Mitchell PJ, Ciudad CJ, Chasmn LA. Nonsensemutations in the dihydrofolate reductase gene affect RNAprocessing.Mol Cell Biol 1989;9:2868-80.

166 Pattinson JK, Millar DS, McVey JH, et al. The moleculargenetic analysis of haemophilia A: a directed strategv forthe detection of point mutations in the human factor VIIIgene. Blood 1990;76:2242-8.

16



g.bmj.com

/J M



nloaded from

http://jmg.bmj.com/


167 Dryja TP. Rhodopsin and autosomal dominant retinitispigmentosa. Eye 1992;6:1-10.

168 Yang TP, Stout JT, Konecki DS, Patel PI, Alford RL,Caskey CT. Spontaneous reversion of novel Lesch-Nyhanmutation by HPRT gene rearrangement. Som Cell MolGenet 1988;14:293-303.

169 Wilson JM, Stout JT, Palella TD, Davidson BL, KelleyWN, Caskey CT. A molecular survey of hypoxanthine-guanine phosphoribosyltransferase deficiency in man. ]fClin Invest 1986;77:188-95.

170 Tiller GE, Rimoin DL, Murray LW, Cohn DH. Tandemduplication within a type II collagen gene (COL2A1)exon in an individual with spondyloepiphyseal dysplasia.Proc Natl Acad Sci USA 1990;87:3889-93.

171 Wolfe KH, Sharp PM, Li WH. Mutation rates differ amongregions of the mammalian genome. Nature 1989;337:283-5.

172 Lichtenauer-Kaligis EGR, van der Velde-van Dijke I, denDulk H, van de Putte P, Giphart-Gassler M, de Jong JGT.Genomic position influences spontaneous mutagenesis ofan integrated retroviral vector containing the hprt cDNAas target for mutagenesis. Hum Mol Genet 1993;2: 173-82.

173 Cooper DN, Krawczak M. The mutational spectrum ofsingle base-pair substitutions causing human genetic dis-ease: patterns and predictions. Hum Genet 1990;85:55-74.

174 Cooper DN, Krawczak M. Mechanisms of insertionalmutagenesis in human genes causing genetic disease.Hum Genet 1991;87:409-15.

175 Krawczak M, Cooper DN. Gene deletions causing humangenetic disease: mechanisms of mutagenesis and the roleof the local DNA sequence environment. Hum Genet1991 ;86:425-44.

176 Coulondre C, Miller JH, Farabaugh PJ, Gilbert W. Mo-lecular basis of base substitution hotspots in Escherichiacoli. Nature 1978;274:775-80.

177 Rodenhiser DI, Coulter-Mackie MB, Singh SM. Evidenceof DNA methylation in the neurofibromatosis type 1

(NFl) gene region of 17q11.2. Hum Mol Genet 1993;2:439-44.

178 Riccardi VM. Genotype, malleotype, phenotype, and ran-domness: lessons from neurofibromatosis-1 (NFl). AmJtHum Genet 1993;53:301-4.

179 Buchberg AM, Bedigan HG, Jenkins NA, Copeland NG.Evi 2, a common integration site involved in murinemyeloid leukemogenesis. Mol Cell Biol 1990;10:4658-66.

180 Mays JA, Neerhout RC, Bagby GC, Koler RD. Juvenilechronic granulocytic leukemia: emphasis of cutaneousmanifestations and underlying neurofibromatosis. Am ]fDis Child 1980;134:654-8.

181 Clark RD, Hutter JJ Jr. Familial neurofibromatosis andjuvenile chronic myelogenous leukemia. Hum Genet 1982;60:230-2.

182 Shannon MK, O'Connell P, Martin GA, et al. Loss of thenormal NF1 allele from the bone marrow of childrenwith type 1 neurofibromatosis and malignant myeloiddisorders. N Engl _r Med 1994;330:597-601.

183 Mikol D, Gulcher J, Stefansson K. The oligodendrocyte-myelin glycoprotein belongs to a distinct family of proteinsand contains the HNK-1 carbohydrate. Cell Biol 1990;110:471-80.

184 Dick G. Disorders of myelin - natural history of multiplesclerosis. _J R Soc Med 1992;85:424-5.

185 Smithies 0. Animal models of human genetic diseases.Trends Genet 1993;9:112-16.

186 Hackstein JH. The lethal prune/killer-of-prune interactionof Drosophila causes a syndrome resembling humanneurofibromatosis (NFl). Eur] Cell Biol 1992;58:429-44.

187 Schmale MC, Hensley GT, Udey LR. Neurofibro-matosis in the bicolor damselfish as a model of vonRecklinghausen neurofibromatosis. Ann NY Acad Sci1986;486:386-402.

188 Schmale MC, Hensley GT. Transmissibility of a neuro-fibromatosis-like disease in bicolor damselfish. Cancer Res1988;48:3828-33.

189 Nakamura T, Hara M, Kasuga T. Transplacental inductionofperipheral nervous tumour in the Syrian golden hamsterby N-nitroso-N-ethylura. Am _7 Pathol 1989;135:251-9.

190 Nakamura T, Ushijima T, Ishizaka Y, et al. neu proto-oncogene mutation is specific for the neurofibromasin a N-nitroso-N-ethylura-induced hamster neuro-fibromatosis model but not for hamster melanomas andhuman Schwann cell tumors. Cancer Res 1994;54:976-80.

191 Hinrichs SH, Nerenberg M, Reynolds RK, Khoury G,Jay G. A transgenic mouse model for human neuro-fibromatosis. Science 1987;237:1340-3.

192 Green JE, Baird AM, Hinrichs SH, Klintworth GK, JayG. Adrenal medullary tumours and iris proliferation in a

transgenic mouse model of neurofibromatosis. Amii _rPa-thol 1992;140:1401-10.

193 Bemards A, Snijders AJ, Hannigan GE, Murthy AE,Gusella JF. Mouse neurofibromatosis type 1 cDNAsequence reveals high degree of conservation of bothcoding and non-coding mRNA segments. Hum Mol Genet1993;2:645-50.

194 Hajra A, Martin-Gallardo A, Tarle SA, et al. DNA se-

quences in the promoter region of the NFI gene are

highly conserved between human and mouse. Genomics1994;1 :49-52.

195 Jacks T, Shih TS, Schmitt EM, Bronson RT, Bemards A,Weinberg RA. Tumour predisposition in . mice hetero-zygous for a targeted mutation in NF 1. Nature Genet1994;7:353-61.

196 Brannan CI, Perkins AS, Vogel KS, et al. Targeted dis-ruption of the neurofibromatosis type-I gene leads to

developmental abnormalities in heart and various neuralcrest-derived tissues. Genes Dev 1994;8: 1019-29.

197 Ward K, O'Connell P, Carey JC, et al. Diagnosis of neuro-fibromatosis 1 by using tightly linked flanking DNAmarkers. Am J Hum Genet 1990;46:943-49.

198 Hofman KJ, Boehm CD. Familial neurofibromatosis type1: clinical experience with DNA testing. _7 Pediatr 1992;120:394-8.

199 Upadhyaya M, Fryer A, MacMillan J, Broadhead W,Huson SM, Harper PS. Prenatal diagnosis and pre-symptomatic detection of neurofibromatosis type 1. _7Med Genet 1992;29:180-3.

200 Jorde LB, Watkins WS, Viskochil D, O'Connell P, WardK. Linkage disequilibrium in the neurofibromatosis 1(NF1) region: implication for gene mapping. Am ] HunmGenet 1993;53:1038-50.

201 Messiaen L, De Bie S, Moens T, Van den Enden A, LeroyJ. Lack of independence between five DNA poly-morphisms in the NF1 gene. Hum Mol Genet 1993;2:485.

202 Xu G, Nelson L, O'Connell P, White R. An Alu poly-morphism intragenic to the neurofibromatosis type 1 gene(NF1). Nucleic Acids Res 199 1;19:3764.

203 Lazaro C, Gaona A, Xu G, Weiss R, Estivill X. A highlyinformative CA/GT repeat polymorphism in intron 38 ofhuman neurofibromatosis type 1 (NF1) gene. Hunm Genet1993;92:429-30.

204 Lazaro C, Gaona A, Estivill X. Two repeat CA/GT repeatpolymorphisms in intron 27 of the human neuro-fibromatosis type 1 (NFI) gene. Humn Genet 1994;3:51-2.

205 Shen MH, Upadhyaya M. Two single base polymorphismsin introns 41 and 16 of the NF1 gene. Anm Med Genet1995;60: 147-9.

206 Rodenhiser D, Hovland K. A novel RsaI polymorphismwithin intron 39 of the neurofibromatosis type 1 (NFI)gene. Hum Genet 1995;95:241-2.

207 Hoffmeyer S, Assum G. An RsaI polymorphism in theneurofibromatosis (NF1) gene. Hum Genet 1994;93:481-2.

208 Andersen LB, Wallace MR, Marchuk DA, et al. A poly-morphic cDNA probe in the NFl gene. Nucleic Acids Res1991 ;19:3754.

209 Andersen LB, Wallace MR, Marchuk DA, et al. A poly-morphic cDNA probe on chromosome 17qII.2 locatedwithin the NFl gene (D17S376). Nucleic Acids Res 1991;19:197.

210 Xu W, Liu L, Ponder BAJ. A BamHI polymorphism inthe human EVI2A gene (human homology of the murinegene Evi-2). Nucleic Acids Res 199 1;19:2520.

211 Reyniers E, De Boulle K, Marchuk DA, Andersen LB,Collins FS, Willems PJ. An EcoRI RFLP in the 5' regionof the human NFl gene. Hum Genet 1993;92:631.

212 Andersen LB, Tarle SA, Marchuk DA, Legius E, CollinsFS. A compound nucleotide repeat in the neuro-fibromatosis (NFl) gene. Hum Mol Genet 1993;2:1083.

213 Chen WJ, Andres DA, Goldstein JL, Brown MS. Cloningand expression of a cDNA encoding the a-subunit of ratp21... protein famesyltransferase. Proc latl Acad Sci USA1991;88:11368-72.

214 Kinsella BT, Erdman RA, Maltese WA. Posttranslationalmodification of Ha-ras p21 by famesyl versus ger-anylgeranyl isoprenoids is determined by the COOH-terminal amino acid. Proc Natl Acad Sci USA 1991;88:8934-8.

215 Qiu MS, Pitts AF, Winters TR, Green SH. Ras iso-prenylation is required for ras-induced neuronal differ-entiation of PC12 cells. _7 Cell Biol 1991;115:795-808.

216 Gibbs JB. Ras C-terminal processing enzymes - new drugtargets? Cell 1991;65:1-4.

17



g.bmj.com

/J M



nloaded from

http://jmg.bmj.com/

Documents

Med Review article Molecular genetics of …Review article Moleculargenetics ofneurofibromatosis type 1 (NFl) MingHongShen, Peter S Harper, MeenaUpadhyaya Institute ofMedical Genetics,