FEBS Letters 580 (2006) 4449–4456

NF1 mRNA biogenesis: Effect of the genomic milieu in splicingregulation of the NF1 exon 37 region

Marco Barallea, Natasa Skokoa, Anna Knezevicha, Laura De Contia, Dario Mottia,Madhuri Bhuvanagiria, Diana Baralleb, Emanuele Burattia, Francisco E. Barallea,*

a International Centre for Genetic Engineering and Biotechnology (ICGEB), Padriciano 99, 34012 Trieste, Italyb Department of Medical Genetics, Addenbrooke’s Hospital, Box 134, Hills Road, Cambridge CB2 2QQ, UK

Received 18 May 2006; revised 3 July 2006; accepted 5 July 2006

Available online 14 July 2006

Edited by Takashi Gojobori

Abstract We have studied the splicing regulation of NF1 exons36 and 37. We show that they not only require an intact exonSplicing Enhancer (ESE) within exon 37, but also need the geno-mic region stretching from exons 31 to 38. Any nucleotidechange in two exon 37 third codon positions disrupts the ESE.The extent of exons 36 and 37 skipping due to a mutated ESEdepends on the genomic context. This is a unique example ofwhat may be a more general phenomena involved in the tuningof pre-mRNA processing and gene expression modulation inthe chromosomal setting.� 2006 Federation of European Biochemical Societies. Publishedby Elsevier B.V. All rights reserved.

Keywords: NF1; Splicing; Genomic context; Exonic splicingenhancers

1. Introduction

Over the recent years it has become well established that pre-

mRNA splicing is accomplished by the combinatorial recogni-

tion of multiple degenerate signals, resulting in a network of

RNA–protein interactions across an exon and/or intron [1–

3]. Moreover, recent evidence indicates that alternative splicing

and possibly splicing itself is not only regulated by the relative

abundance of antagonistic factors but by a more complex pro-

cess involving the transcription machinery [4]. However, the

fact that in many cases splicing systems are not easy to model

[5] indicates that the determinants of its regulation are poorly

understood. It is obvious that a greater knowledge of the inter-

play between all the cis and trans-acting elements that modu-

late splicing is needed.

The NF1 gene has been the subject of extensive studies since

its identification as the defective locus causing Neurofibroma-

tosis type 1. In particular, comprehensive sequencing has been

carried out for diagnostic and research purposes [6–9].

Premature termination codon (PTC) induced skipping of

NF1 exon 37, that preserves an open reading frame, has been

previously reported in the case of the nucleotide substitutions

c.6792C > G (p.Tyr2264X) and c.6792C > A (p.Tyr2264X)

[10] (Fig. 1A). The phenomenon by which PTCs cause exon

skipping during pre-mRNA splicing in the nucleus is known

*Corresponding author. Fax: +39 040 3757361.E-mail address: baralle@icgeb.org (F.E. Baralle).

0014-5793/$32.00 � 2006 Federation of European Biochemical Societies. Pu

doi:10.1016/j.febslet.2006.07.018

as nonsense altered splicing (NAS) [11,12]. However, in many

of these cases, it was found that these mutations disrupted an

exonic splicing enhancer (ESE) critical for exon inclusion [13].

In fact, mutations from other patients in the same area, a

c.6790_6791TTins and a c.6789_6792TTACdel (Fig. 1A), both

of which create PTC’s a few base pairs further down stream do

not result in the skipping of this exon [14]. These observations

suggested that the genomic variations were not subject to NAS

but were affecting a exonic Splicing Enhancer (ESE). Indeed,

in a recent report the splicing enhancer activities of the wild-

type and mutant putative ESE were tested in an heterologous

system by measuring the rescue of splicing of the 34 nucleotide

exon 2 of the pSXN minigene reporter construct [15]. In this

system, 15 nucleotides of the NF1 exon 37 WT sequence

(nucleotides c.6783–c.6796) were able to rescue splicing of

the alternative exon approximately 30% more than the mu-

tated sequences. Obviously, this is a limited indication on an

artificial system and a more detail molecular analysis was

desirable.

In this study we show that even a synonymous change at po-

sition c.6792 produces exon skipping conclusively demonstrat-

ing that altered splicing is not due to nonsense mutations at

this position but rather due to interference with a regulatory

region present in the NF1 exon 37, necessary for its definition.

We extensively mapped this element and show that it is not

only required for the definition of exon 37, but also that it con-

tributes towards the identification of the upstream exon 36, in

that the aberrant splicing of exon 37 affects the correct recog-

nition of exon 36. Intriguingly, we show that both these exons

need an extremely wide genomic context to obtain the in vivo

splicing pattern indicating that their correct definition relies, in

addition to the intrinsic properties of the exons themselves, on

the wider genomic environment within which the exons are

located.

2. Materials and methods

2.1. Plasmid constructionThe hybrid minigene constructs were prepared using the PTB plas-

mid [16] by inserting wild-type and mutated DNA fragments amplifiedwith PTBEX37F and PTBEX37R primers. The same sense mutation(6792C > T) was created by site directed mutagenesis. Minigenes carry-ing NF1 genomic sequences1 from exons 36 to 38 and from 34 to 38were made by amplifying wild-type and mutated genomic DNA with

1 GenBank Accession Number for NF1 Sequence NM_000267.

blished by Elsevier B.V. All rights reserved.

Fig. 1. Nonsense or same sense mutations at position 6792 of the NF1 gene behave in an identical manner. (A) Previously described mutations (grey)in the exon 37 of patients suffering from NF1 that give rise to PTC (denoted by X) and their effect on the splicing of this exon. The sequence shownhere only represents part of exon 37. (B) Schematic representation of the PTB hybrid minigene carrying the NF1 exon 37 along with intronic flankingsequence. Black, shaded and white boxes represent alpha globin, fibronectin and NF1 exons respectively. Below are the RNA products generated bythe splicing assay for the wildtype, 6792C > G nonsense mutation and the 6792C > T same sense mutation hybrid minigenes. Two products are seenon the agarose gel electrophoresis, the 239 bp band represents the RNA transcript lacking NF1 exon 37 (�Ex37) and the 342 bp band represents thetranscript that includes NF1 exon 37 (+Ex37).

4450 M. Baralle et al. / FEBS Letters 580 (2006) 4449–4456

the following forward primers: EX36BAMHI EX34BAMHI and thereverse primer EX38XHO. The amplicon was subsequently cloned inthe pcDNA3 vector (Invitrogen). PCR site directed mutagenesis wasperformed to create the nucleotide substitutions reported, makinguse of the internal cassette inside the minigene defined by the uniquerestriction enzymes sites EcoRI–AfeI.

The minigenes extending from NF1 exons 31–38 were constructedby cloning the genomic segment NF1 34–38 (amplified with EX34-BAMHI and EX38XHO) into SmaI site of PUC19 (Biolabs) vectorwhose BamHI site had being destroyed. The vector was then openedwith BamH1 and SphI, into which the genomic fragment ampli-fied with the oligos EX31BAMH1/SPH1R and subsequently digestedwith BamH1 and SphI, was cloned. The genomic fragment thus con-taining exons 31–38 was then sub-cloned into pcDNA3 vector Bam-HI/XhoI. The nucleotide substitutions of interest in the ESE regionwere introduced by excising the DNA fragment flanked by therestriction enzymes EcoRI–AfeI substituting it with the identicalfragment, bar the nucleotide transition of interest. All constructswere designed to maintain an open reading frame in the wild-typecontext.

The correct sequence of all constructs was checked by a CEQ2000sequencer (Beckman Coulter, Fullerton CA) according to the manu-facturers instructions.

2.2. Primers

PTBEX37F TTCATATGTAGCTACCAAGATCA

PTBEX37R TTCATATGTAACTTATTCTCAAT

EX36BAMH1

ACGTGGATCCATGGATTTGCATTCCAATATAATC-

CAT

EX34BAMH1 ACGTGGATCCATGGAAGAGACCAAG-

CAAGTTTTGA

EX38XHO TGCACTCGAGTCACTTGTCATTGAATATA-

CGGAGA

EX31BAMH1 ACGTGGATCCATGAGTATTGAATTGA-

AACACCTTTG

Fig. 2. The importance of the genomic context on NF1 exon 37 definition. (A) In scale schematic representation of the genomic region encoding NF1exons 30–39. Agarose gel electrophoresis showing RT-PCR products obtained with a minigene. (B) Composed of exons 36–38. The upper bandcorresponds to the transcript that includes all three exons as depicted on the right. The lower band was a mixture of transcripts, where although all ofthem skipped exon 37 several showed an incorrect use of splice sites (depicted by the black box) in the graphical representation. (C) Composed ofexons 34–38. The wildtype construct results in a transcript that carries all five exons. The nonsense and same sense mutations result in a three bandprofile where the upper band represents correct assembly of the five exons, the intermediate band lacks exon 37, and the lower band lacks both exons36 and 37. (D) Agarose gel electrophoresis showing RT-PCR products obtained from RNA prepared from EBV lymphoblast derived from a normalindividual (WT) or from a NF1 patient carrying the c.6792C > G substitution in one allele. The splicing pattern of the latter shows three bands, theupper one corresponds to the correct mature mRNA derived from the normal and to some extent from the mutant allele. The intermediate band lacksexon 37 and the lower band lacks both exons 36 and 37. (E) Comparison of RT-PCR products obtained with the use of minigenes composed of exon34–38 with those composed of exons 31–38. From the gel we can see that the more genomic sequences are present the more correctly spliced mRNA ispresent with the wider context resembling the chromosomal situation (compare gels D with E). (F) Graphical representation of the position of thespurious splice sites used from the transcripts lacking exon 37 (Fig. 2B). **indicates correct splice site, *splice site recognized as such by the software‘‘Splice Site Prediction by Neural Network Site’’ (http://www.fruitfly.org/seq_tools/splice.html). Bold uppercase indicates exonic sequence, lowercaseintronic/denotes the splice junction.

M. Baralle et al. / FEBS Letters 580 (2006) 4449–4456 4451

SPH1R GTTGGAATATCTCTCATGCATGCCT

Sp34F ACAGAAGCTTTGTTGGAGATC

Sp38R CTCATCAAGCTGCACAGCC

2.3. Cell culture, transfection, and RT-PCR analysisTransfections of the human cervical carcinoma (HeLa) cells were

carried out as previously described [17]. Total RNA was then preparedfrom the transfected cells using RNAwiz (Ambion). Analysis of thealternatively spliced products was performed on each cDNA by PCRamplification with the primers previously described, in the case of hy-brid minigenes [16], T7 and SP6 in the case of the 36–38 minigene,Sp34F and Sp38R in the case of the lymphoblasts and Sp34F withSP6 in the case of the minigenes 31 or 34–38.

2.4. EBV patient transformed lymphoblastsEBV transformed lymphoblasts from a patient suffering from NF1

and previously identified to carry the c.6792C > G mutation [7] wereobtained from the European Collection of Cell Cultures ECACC.

3. Results

3.1. c.6792C > A and c.6792C > G mutations result in exon

skipping due to the disruption of an exonic regulatory

element

To test the hypothesis that mutations c.6792C > A and

c.6792C > G in NF1 exon 37 were affecting a splicing regula-

tory region in NF1 exon 37 and not causing NAS, wild-type

and mutated NF1 exon 37 sequences along with flanking intro-

nic sequence were inserted in the previously reported PTB

hybrid minigene (Fig. 1B). Surprisingly, we observed that the

wild-type sequence is not efficiently processed (Fig. 1B) in this

context as only approximately 50% of exon inclusion could be

observed. This was unexpected, as in many other cases the

minigene reflected accurately the splicing pattern of the endog-

enous gene [16,18,19]. In any case, it was still possible to see

4452 M. Baralle et al. / FEBS Letters 580 (2006) 4449–4456

the deleterious effect on splicing of the disease causing muta-

tion 6792C > G as this nucleotide transition in the minigene re-

sulted in complete skipping of NF1 exon 37. The fact that also

the same sense (translationally silent) substitution 6792C > T

(Fig. 1B) introduced at this position caused total exon skipping

indicated that the aberrant processing of mRNA carrying

c.6792C > A or c.6792C > G observed in the patients is indeed

due to the disruption of a ESE and not to NAS.

3.2. The importance of the genomic context and the peculiar

genomic organization in the NF1 exon 34–38 region

The poor performance of the hybrid minigene exon 37 wild-

type construct indicated that efficient definition of NF1 exon

37 could need additional flanking intronic and/or exonic NF1

sequences localized within the exon-rich genomic cluster in

which it is found (Fig. 2A). We then constructed a minigene

with additional genomic sequences spanning the region from

NF1 exons 36–38 (Fig. 2B). In this case the wild-type construct

showed efficient processing of exon 37 whilst the 6792C > G

(nonsense mutation) and 6792C > T (translationally silent)

mutant constructs showed significant exon skipping. This re-

sult confirms that, as observed with the PTB minigenes, the

translationally silent mutation acts in the same way as the non-

sense mutation. However, this expanded context was not com-

pletely satisfactory in mimicking the natural effects of these

mutations. In fact, an intriguing and albeit unexplained obser-

vation was made when the transcripts lacking exon 37 (lower

bands Fig. 2B) were analysed. Cloning and sequencing these

bands showed that the junction between exons 36 and 38

was not always derived from clean skipping of exon 37 but

in addition to the normal 36 and 38 splice sites customarily

used there was an heterogeneous collection of unusual splice

sites usage (Fig. 2F). This was true for mRNA lacking exon

37 obtained from wild-type and 6792C > G constructs.

To better assess the influence of the flanking genomic se-

quences on NF1 exon 37 splicing, a further set of minigenes

was then assembled comprising NF1 exons 34 through to exon

38 (Fig. 2C). The wild-type construct when transfected in Hela

cells mimicked what has being previously reported to occur in

human tissues [8,20], with proper processing of the exons and

an extremely small amount of transcript lacking NF1 exon 37.

Again, the results show that as with the other minigenes the

translationally silent mutation 6792C > T acts in the same

way as the nonsense mutation 6792C > G, causing exon 37

skipping. In addition, a third type of mRNA processing is also

observed, namely one in which both the exons 36 and 37 are

skipped. This, to a smaller extent, was also observed in EBV

transformed lymphocytes obtained from a patient suffering

from NF1 and carrying the c.6792C > G substitution in hetero-

zygosis with a normal allele (Fig. 2D). The bands resulting

from the processing of the minigene and the lymphoblasts were

sequenced and, in contrast with those derived from the minig-

enes made from exons 36 to 38, this time all of them resulted

from correct splice site usage.

Although exons 34–38 are sufficient for correct wild-type

processing of the minigene we observed that the mutated con-

struct did not imitate exactly the splicing pattern obtained with

the patient lymphoblast. Extending the minigene to exons 31

(31–38 constructs) caused a decrease of the exon 36–37 double

skipping (Fig. 2E). This is a shift towards the splicing pattern

observed in lymphoblasts derived from a NF1 patient carrying

the c.6792C > G mutation (Fig. 2D), further underlining the

long-range interplay of splicing regulatory elements within this

exon cluster.

3.3. Definition of NF1 exon 37 regulatory element of splicing

We have then carried out a study to define the ESE and the

interplay with other exon 37 sequences. Extensive site directed

mutagenesis was undertaken of the bases between positions

c.6790 and c.6797 with all three possible nucleotide substitu-

tions (Fig. 3). We observed that any nucleotide transition at

positions c.6790, 6791, 6793, 6794, 6796, and 6797 had no ef-

fect on splicing. However, every nucleotide change at positions

c.6792 and c.6795 led to aberrant splicing, resulting in a three

band profile where the upper band represents correct assembly

of the five exons, the intermediate band lacks exon 37 and the

lower band lacks both exons 36 and 37. This observation is

also valuable for future diagnostic procedures as any nucleo-

tide variation at position c.6795 will very likely be a disease

causing mutation.

3.4. Exon 37 definition influences exon 36 recognition

In previous reports of upstream exon skipping the defect was

always located in the downstream exon splice sites [21–23].

This is the first report of decreased downstream exon definition

due to an ESE disruption affecting the upstream exon defini-

tion. In order to distinguish if this was a direct effect of disrup-

tion of the ESE or an indirect effect in as much as the exon 36

recognition is dependent on exon 37 definition we analyzed the

splicing pattern resulting from a minigene spanning exon 34–

38 with the donor site of exon 37 mutated by a

IVS37 + 1G > T mutation. From Fig. 4 we can see that this re-

sults in two transcripts, one in which only exon 37 is skipped

and a second where both exons 36 and 37 are skipped. These

results indicate that the definition of exon 36 is dependent on

that of 37. The severity of which the definition of exon 37 is

affected reflects on exon 36 definition as shown by comparing

the splicing pattern of transcripts resulting from disruption of

ESE (6792C > G) with the splicing pattern caused by the more

drastic donor site disruption (IVS37 + 1G > T).

3.5. The exon regulatory element in NF1 exon 37 offsets the

presence of a non-consensus 3 0 splice site

Exonic or intronic cis-acting regulatory regions are often de-

scribed on exons with weakly defined 5 0 and 3 0 splice site [1,24].

The ESE that we have characterized may thus be necessary be-

cause of the extremely weak definition of NF1 exon 37 3 0 splice

site that does not compare favorably to consensus (Fig. 5).

To evaluate the effect of ESE disruption in relation to the

strength of the 3 0 splice site we tested a minigene construct with

an improved 3 0 splice site, created by the nucleotide transition

IVS36-A > C, carrying the mutation c.6792C > G that caused

aberrant splicing. Fig. 5 shows that the double mutant

(c.6792C > G and IVS36-A > C) has a splicing pattern identi-

cal to the wild-type, with full inclusion of exons 36 and 37.

4. Discussion

We have studied the effect on splicing of naturally occurring

and site directed mutations in NF1 exon 37. Our work, in

accordance with the greater enhancer activity of the wild-type

Fig. 3. Definition of the ESE. Schematic representation of the minigene used in this assay with nucleotides c.6790–c.6797 in NF1 exon 37 highlightedin their codon triplets. The lower panel shows the effect on RNA processing of changing each nucleotide present between position c.6790 and c.6797of the NF1 gene, into the three alternative nucleotides for that position. It can be observed that every nucleotide change at positions c.6792 and c.6795led to aberrant splicing, resulting in a three band profile where the upper band represents correct assembly of the five exons, the intermediate bandlacks exon 37 and the lower band lacks both exons 36 and 37.

Fig. 4. Effect of decrease of NF1 exon 37 definition on exon 36 recognition. To test this phenomena we have decreased exon 37 definition by ESEdisruption (6792C > G) and the more drastic 5 0ss disruption (IVS37 + 1G > T). Upper panel shows a schematic representation of the minigenescarrying the ESE 6792G > C mutation or the IVS37 + 1G > T mutation that disrupts the donor site of NF1 exon 37. Lower panel shows the effect onRNA processing of disrupting NF1 exon 37 definition. We can observe that the greater the skipping of exon 37 the more exon 36 is also skipped.

M. Baralle et al. / FEBS Letters 580 (2006) 4449–4456 4453

sequence over the mutated sequence [15], shows that the dis-

ruption of a ESE as the cause of the inefficient inclusion in

NF1 mRNA of exon 37 carrying the c.6792C > G and

c.6792C > A mutations found in patients [10,14]. In fact, even

the same sense variation, 6792C > T results in the skipping of

NF1 exon 37.

Exon 37 is localized in an exon-rich genomic cluster and is

badly recognized by the splicing machinery if it is placed with

limited flanking intronic sequences in a minigene with heterol-

ogous genomic context. It needs to be embedded in the region

spanning NF1 exons 34–38 to become efficiently recognized

(Fig. 6A–C, wildtype constructs). Indeed, a blanket mutagen-

esis analysis carried out in the 34–38 context shows that

changes in 2 out of the 8 nucleotide positions analyzed (specif-

ically the C’s in the ACAAC motif) produced aberrant tran-

scripts: exon 37 skipping and the double skipping of exons

36 and 37, indicating unequivocally that this region contains

a regulatory element important for correct exon definition. It

is interesting to note that both the c.6790_6791TTins and a

c.6789_6792TTACdel observed in patients that did not cause

exon skipping (Fig. 1A) preserve the sequence ACAACA. sim-

ilar to the A/C splicing regulatory element previously reported

[25]. Widening the genomic context further from NF1 exons 31

to 38 partially offsets the importance of the exon 37 ESE

reducing the skipping of both exons 36 and 37 in the minigene

carrying the c.6792C > G mutation. Indeed, if we consider that

in lymphoblasts obtained from a patient carrying the

c.6792C > G mutations the splicing pattern observed derives

equally from both the normal and mutant alleles, as has been

shown for c.6792C > A [14] (Fig. 2D) and hence the contribu-

tion to the top band is predominantly from the non-mutated

allele it follows that the pattern resulting from the 31 to 38

minigene carrying the c.6792C > G mutation is practically

identical to that of the chromosomal setting (Fig. 6, compare

lines C, D and E in the c.6792C > G minigene constructs/lym-

phoblast allele column).

Regarding exon 36, as far as we know this is the first time

that upstream exon definition is shown to be at least in part

dependant on an ESE in the downstream exon, in as much

as this element is necessary for definition of exon 37 whose

Fig. 5. The ESE is necessary due to the weak definition of exon 37 splice sites. Schematic representation of the minigenes carrying the wild-type 3 0ssor improved 3 0 splice site (IVS36-3A > C). Splice site strength was evaluated using the ‘‘Splice Site Prediction by Neural Network Site’’ (http://www.fruitfly.org/seq_tools/splice.html)./ denotes intron exon junction. The nucleotides 6790–6798 in NF1 exon 37 are highlighted in their codontriplets. The lower panel shows the effect on RNA processing of improving the 3 0 splice site. The aberrant splicing previously observed with themutations 6792C > G that disrupted the ESE is no longer present (lanes 1 and 2).

4454 M. Baralle et al. / FEBS Letters 580 (2006) 4449–4456

definition is in turn necessary for the correct recognition of

exon 36 by the splicing machinery. In previous reports of up-

stream exon skipping the defect was always located in the

downstream exon splice sites [21–23]. These results highlight

the dynamic setting of the splicing process, where early-and-

late transcribed sequences condition the definition of specific

exons located at considerable distance. The experimental data

therefore point to a particular long-range combinatorial con-

trol of exon definition in this cluster. The recognition of splice

sites involves crosstalk between multiple, at times relatively

weak, interactions that contribute to establish complexes defin-

itively committed to splicing [1–3]. Our work shows that there

is a much larger long distance network across exons and in-

trons than previously thought, which contributes to correct

pre mRNA processing. In fact, the more genomic sequences

present in the minigenes, the greater amount of correctly pro-

cessed transcript is produced, even in the versions of the mini-

gene where the ESE is disrupted (Figs. 2 and 5). This type of

control may not be restricted to NF1 exon 37. Although we

do not yet fully understand the intricate mechanisms of the

interactions involved in this cluster it is possible that up and

down stream sequences may modify the transcription kinetic

and/or the spliceosome composition. For example, it has been

shown that in the human thrombopoietin gene processing of a

small intron present within the last exon of the gene, is depen-

dent on processing of previous introns [26]. Furthermore, the

effect of this type of long distance interactions was highlighted

by Fededa et al. in a recent study [27] on splicing in the Fibro-

nectin gene where the processing undergone by an upstream

alternatively spliced exon has an influence on the pattern of

alternative splicing of downstream exons.

Another interesting observation related to our finding re-

gards the overlapping of codon usage and splicing regulatory

elements. The latter represents an additional constraint during

the evolution of an exon sequence. Synonymous variations are

assumed to be functionally neutral both in clinical diagnosis

and when measuring evolutionary distances between species

because they do not change the peptide sequence. However,

a recent study [28] has shown that synonymous variations in

CFTR exon 12 cannot evolve neutrally, but are significantly

constrained by splicing requirements. In fact, roughly one

quarter of the synonymous changes studied in CFTR exon 9

[29] and CFTR exon 12 [28] have an effect on splicing. A sim-

ilar situation is observed, albeit in a limited setting, with the

synonymous changes introduced during the mutational analy-

sis on the NF1 exon 37 where the only two nucleotide transi-

tions that result in a same sense mutation (c.6792C > T and

c.6795C > T) result in aberrant splicing. Our study shows that

a weak 3 0 splice site, like the one flanking exon 37, conditions

the evolution of the exon that must maintain exonic splicing

regulatory sequences to ensure its inclusion in mature mRNA.

We have tested this hypothesis experimentally by improving

the NF1 exon 37 3 0 splice site. The introduction of a single

nucleotide transition IVS36-3A > C overrides the need for a

functional ESE in NF1 exon 37. From an evolutionary bearing

this means that any nucleotide variation, in an improved con-

text, and hence amino acid changes may occur without altering

the pre-mRNA processing.

Fig. 6. Quantification of transcripts from Figs. 1 and 2 by Image J (http://rsb.info.nih.gov/ij/) shows the influence of the genomic context induplicating the in vivo effects (E). Wild-type exon 37 alone with a small portion of flanking intronic regions is not sufficient to yield the observed>99% inclusion in normal lymphoblasts (compare lines A and E). However, this result can be readily achieved when larger portions of this genomicregion are included in the genomic construct (compare lines C and D with E). In addition, an increased genomic context also provides a closercorrespondence with regards to the exon skipping effects of the 6792C > G substitution In fact, only the construct carrying NF1 exons 31–38 togetherwith a mutated exon 37 (D) yields a splicing pattern that reflects the observed percentage in a patient’s lymphoblast that carry the C6792Gsubstitution in one allele (compare D and E). Quantification in the minigene 36–38 of the bands lacking exon 37 was performed on the totalaberrantly spliced bands.

M. Baralle et al. / FEBS Letters 580 (2006) 4449–4456 4455

Acknowledgements: This work was supported by grants from TelethonOnlus Foundation and by a European community grant (EURAS-NET) to FEB.

References

[1] Graveley, B.R. (2000) Sorting out the complexity of SR proteinfunctions. Rna 6, 1197–1211.

[2] Reed, R. (2000) Mechanisms of fidelity in pre-mRNA splicing.Curr. Opin. Cell Biol. 12, 340–345.

[3] Smith, C.W. and Valcarcel, J. (2000) Alternative pre-mRNAsplicing: the logic of combinatorial control. Trends Biochem. Sci.25, 381–388.

[4] Proudfoot, N.J., Furger, A. and Dye, M.J. (2002) IntegratingmRNA processing with transcription. Cell 108, 501–512.

[5] Buratti, E. and Baralle, F.E. (2005) Another step forward forSELEXive splicing. Trends Mol. Med. 11, 5–9.

[6] Ars, E., Serra, E., Garcia, J., Kruyer, H., Gaona, A., Lazaro, C.and Estivill, X. (2000) Mutations affecting mRNA splicing are themost common molecular defects in patients with neurofibroma-tosis type 1. Hum. Mol. Genet. 9, 237–247.

[7] Mattocks, C., Baralle, D., Tarpey, P., Ffrench-Constant, C.,Bobrow, M. and Whittaker, J. (2004) Automated comparativesequence analysis identifies mutations in 89% of NF1 patients andconfirms a mutation cluster in exons 11–17 distinct from the GAPrelated domain. J. Med. Genet. 41, e48.

[8] Messiaen, L.M., Callens, T., Mortier, G., Beysen, D., Van-denbroucke, I., Van Roy, N., Speleman, F. and Paepe, A.D.(2000) Exhaustive mutation analysis of the NF1 gene allowsidentification of 95% of mutations and reveals a high frequency ofunusual splicing defects. Hum. Mutat. 15, 541–555.

[9] Whittaker, J.L., Mattocks, C., Baralle, D., Tarpey, P., Ffrench-Constant, C. and Bobrow, M. (2005) Re: Pitfalls of automatedcomparative sequence analysis as a single platform for routineclinical testing for NF1 (Messiaen and Wimmer). J. Med. Genet.42, e41.

[10] Messiaen, L., Callens, T., De Paepe, A., Craen, M. and Mortier,G. (1997) Characterisation of two different nonsense mutations,C6792A and C6792G, causing skipping of exon 37 in the NF1gene. Hum. Genet. 101, 75–80.

[11] Wang, J., Hamilton, J.I., Carter, M.S., Li, S. and Wilkinson,M.F. (2002) Alternatively spliced TCR RNA induced by disrup-tion of reading frame. Science 297, 108–110.

[12] Wang, J., Chang, Y.F., Hamilton, J.I. and Wilkinson, M.F.(2002) Nonsense-associated altered splicing: a frame-dependentresponse distinct from nonsense-mediated decay. Mol. Cell 10,951–957.

[13] Buhler, M. and Muhlemann, O. (2005) Alternative splicinginduced by nonsense mutations in the immunoglobulin mu VDJexon is independent of truncation of the open reading frame. Rna11, 139–146.

[14] Hoffmeyer, S., Nurnberg, P., Ritter, H., Fahsold, R., Leistner,W., Kaufmann, D. and Krone, W. (1998) Nearby stop codons inexons of the neurofibromatosis type 1 gene are disparate spliceeffectors. Am. J. Hum. Genet. 62, 269–277.

[15] Zatkova, A., Messiaen, L., Vandenbroucke, I., Wieser, R.,Fonatsch, C., Krainer, A.R. and Wimmer, K. (2004) Disruptionof exonic splicing enhancer elements is the principal cause of exonskipping associated with seven nonsense or missense alleles ofNF1. Hum. Mutat. 24, 491–501.

[16] Pagani, F. et al. (2000) Splicing factors induce Cystic FibrosisTransmembrane Regulator exon 9 skipping through a non-evolutionary conserved intronic element. J. Biol. Chem. 275,21041–21047.

4456 M. Baralle et al. / FEBS Letters 580 (2006) 4449–4456

[17] Baralle, M., Baralle, D., De Conti, L., Mattocks, C., Whittaker,J., Knezevich, A., Ffrench-Constant, C. and Baralle, F.E. (2003)Identification of a mutation that perturbs NF1, a gene splicingusing genomic DNA samples and a minigene assay. J. Med.Genet. 40, 220–222.

[18] Pagani, F., Buratti, E., Stuani, C., Bendix, R., Dork, T. andBaralle, F.E. (2002) A new type of mutation causes a splicingdefect in ATM. Nat. Genet. 30, 426–429.

[19] Pagani, F., Stuani, C., Tzetis, M., Kanavakis, E., Efthymiadou,A., Doudounakis, S., Casals, T. and Baralle, F.E. (2003) Newtype of disease causing mutations: the example of the compositeexonic regulatory elements of splicing in CFTR exon 12. Hum.Mol. Genet. 12, 1111–1120.

[20] Vandenbroucke, I., Callens, T., De Paepe, A. and Messiaen, L.(2002) Complex splicing pattern generates great diversity inhuman NF1 transcripts. BMC Genomics 3, 13.

[21] Fang, L.J., Simard, M.J., Vidaud, D., Assouline, B., Lemieux, B.,Vidaud, M., Chabot, B. and Thirion, J.P. (2001) A novelmutation in the neurofibromatosis type 1 (NF1) gene promotesskipping of two exons by preventing exon definition. J. Mol. Biol.307, 1261–1270.

[22] Haire, R.N., Ohta, Y., Strong, S.J., Litman, R.T., Liu, Y., Prchal,J.T., Cooper, M.D. and Litman, G.W. (1997) Unusual patterns ofexon skipping in Bruton tyrosine kinase are associated withmutations involving the intron 17 30 splice site. Am. J. Hum.Genet. 60, 798–807.

[23] Schneider, S., Wildhardt, G., Ludwig, R. and Royer-Pokora, B.(1993) Exon skipping due to a mutation in a donor splice site inthe WT-1 gene is associated with Wilms’ tumor and severe genitalmalformations. Hum. Genet. 91, 599–604.

[24] Fairbrother, W.G., Yeh, R.F., Sharp, P.A. and Burge, C.B.(2002) Predictive identification of exonic splicing enhancers inhuman genes. Science 297, 1007–1013.

[25] Stickeler, E., Fraser, S.D., Honig, A., Chen, A.L., Berget, S.M.and Cooper, T.A. (2001) The RNA binding protein YB-1 bindsA/C-rich exon enhancers and stimulates splicing of the CD44alternative exon v4. Embo. J. 20, 3821–3830.

[26] Romano, M., Marcucci, R. and Baralle, F.E. (2001) Splicing ofconstitutive upstream introns is essential for the recognition ofintra-exonic suboptimal splice sites in the thrombopoietin gene.Nucleic Acids Res. 29, 886–894.

[27] Fededa, J.P. et al. (2005) A polar mechanism coordinatesdifferent regions of alternative splicing within a single gene.Mol. Cell 19, 393–404.

[28] Pagani, F., Raponi, M. and Baralle, F.E. (2005) Synonymousmutations in CFTR exon 12 affect splicing and are notneutral in evolution. Proc. Natl. Acad. Sci. USA 102, 6368–6372.

[29] Pagani, F., Buratti, E., Stuani, C. and Baralle, F.E. (2003)Missense, nonsense, and neutral mutations define juxtaposedregulatory elements of splicing in cystic fibrosis transmembraneregulator exon 9. J. Biol. Chem. 278, 26580–26588.