1
ANALYSIS OF NF1 MUTATION MECHANISMS
By
REBECCA L. LODA-HUTCHINSON
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
2009
2
© 2009 Rebecca L. Loda-Hutchinson
3
To Mrs. Michelle Doyle and Mr. Stephen Sans, whose inspiration and enthusiasm started me on
this path many years ago
4
ACKNOWLEDGMENTS
First, I need to thank all the families and individuals who participated in these studies;
without them, this work would not have been possible. I want to thank Dr. Peggy Wallace for
taking me on as a student and letting me work in her lab for the past five years. I would not have
made it through this experience without her guidance, understanding, and support. I also want to
thank the members of my supervisory committee, Dr. Daniel Driscoll, Dr. Keith Robertson, and
Dr. Peter Sayeski, for their time, support, and suggestions throughout this process. My thanks
also goes to all the past and present members of the Wallace lab, especially Beth Fisher, for her
patience and smiles as I was learning the ropes; and Michelle Burch, for always having an
answer for me. I also want to thank Dr. Michele Tennant, Dr. Pauline Ng (Fred Hutchinson
Cancer Research Center, Seattle), and Maya Zuhl (University of Maryland Biotechnology
Institute) for their invaluable bioinformatics help.
Finally, I want to thank my family and friends for their support and encouragement no
matter what has come my way these past five years. I especially want to thank Randi Marie,
Deborah, Nicole, and Lauren, who have shared in the graduate school experience with me, for
helping me stay sane, giving me someone to talk to, and reminding me to relax and have fun.
Last, but not least, a huge thanks goes to my husband, Lance, who has seen this process through
with me from applications to a dissertation, and never stopped believing that I could do it.
5
TABLE OF CONTENTS
page
ACKNOWLEDGMENTS ...............................................................................................................4
LIST OF TABLES...........................................................................................................................7
LIST OF FIGURES .........................................................................................................................8
ABSTRACT.....................................................................................................................................9
CHAPTERS
1 INTRODUCTION ..................................................................................................................11
Neurofibromin and the NF1 Gene ..........................................................................................11
Neurofibromatosis 1 (NF1).....................................................................................................12
Mutations in NF1 ....................................................................................................................15
2 SOMATIC CpG C TO T TRANSITIONS AT NF1 GERMLINE HOTSPOTS....................19
Introduction.............................................................................................................................19
Materials and Methods ...........................................................................................................22
Mutation Detection by PCR and Restriction Digest........................................................22
Methylation Status Analysis............................................................................................23
Results.....................................................................................................................................24
Methylation Status Analysis............................................................................................24
Mutation Detection by PCR and Restriction Digest........................................................26
Discussion...............................................................................................................................26
3 ALTERNATIVE SPLICING OF EXON 23a AND mRNA EDITING .................................32
Introduction.............................................................................................................................32
Material and Methods .............................................................................................................36
Reverse-Transcription and PCR......................................................................................36
Cloning and Sequence Analysis ......................................................................................37
Results.....................................................................................................................................37
Alternative Splicing Patterns ...........................................................................................37
Analysis of RNA Editing.................................................................................................39
Discussion...............................................................................................................................40
4 MISSENSE MUTATION COMPUTATIONAL ANALYSIS OF PATHOGENICITY .......47
Introduction.............................................................................................................................47
Materials and Methods ...........................................................................................................49
Missense Computational Methods ..................................................................................49
Other Databases and Programs........................................................................................49
6
Data Sets ..........................................................................................................................50
Sequence Input Requirements .........................................................................................50
Splice Analysis ................................................................................................................53
Results.....................................................................................................................................53
Discussion...............................................................................................................................57
5 CONCLUSIONS AND FUTURE DIRECTIONS .................................................................64
Somatic CpG C to T Mutation................................................................................................64
Alternative Splicing of exon23a .............................................................................................66
Computational Analysis Comparison .....................................................................................69
APPENDIX
A CpG C to T Mutation Analysis Data ......................................................................................70
B Exon 23a Alternative Splicing Data .......................................................................................74
C Missense Mutation Computational Analysis Data .................................................................77
LIST OF REFERENCES...............................................................................................................85
BIOGRAPHICAL SKETCH .......................................................................................................100
7
LIST OF TABLES
Table page
2-1 Primers used in CpG C to T mutation screening. ..............................................................29
3-1 Summary of alternative splicing of exon23a seen in various sample types examined......45
4-1 Summary of computational missense prediction results for 5 data sets (4 controls and
1 unknown). .......................................................................................................................63
A-1 Results of CpG C to T mutation screen using TaqαI restriction enzyme digest................70
B-1 Relative concentrations of Type I v Type II mRNA in blood, tumor and culture
samples...............................................................................................................................74
C-1 Results for each mutation as returned by the various computational methods used. ........78
8
LIST OF FIGURES
Figure page
1-1. Diagram of neurofibromin’s putative domains..................................................................18
2-1. Methylation-sensitive restriction digest protocol...............................................................29
2-2. Visualization of methylation-specific restriction digest analysis. .....................................30
2-3. Results of bisulfite sequencing. .........................................................................................30
2-4. TaqαI digests of E23.2 PCR products visualized on 8% PAGE........................................31
3-1. Representative gels showing relative concentrations of Type I v Type II mRNA in
various tissue types studied................................................................................................46
3-2. Comparison of alternative splicing of exon23a in primary plexiform tumors v
cultured Schwann cells from the same tumors. .................................................................46
9
Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy
ANALYSIS OF NF1 MUTATION MECHANISMS
By
Rebecca L. Loda-Hutchinson
May 2009
Chair: Margaret R. Wallace
Major: Medical Sciences—Genetics
Neurofibromin is a large protein encoded by the NF1 gene, whose best understood
function is as a down-regulator of the RAS signaling pathway, leading to NF1’s classification as
a tumor suppressor gene. NF1 gene mutations, which occur at a rate 10x higher then the genome
average, lead to the autosomal dominant disorder neurofibromatosis 1 (NF1). NF1 has variable
expressivity and is clinically diagnosed using seven diagnostic criteria, of which the key features
are café-au-lait spots, neurofibromas, Lisch nodules, and skin fold freckling. Genetic diagnosis
is difficult, as the gene is very large and very few mutations are recurrent. Additionally, many
NF1 phenotypes are believed to originate with a second mutation in the wild type allele in
specific cell types. To gain insight into mutation mechanisms in NF1, I pursued three projects.
First, I analyzed the rate of somatic C to T mutations at four hotspots for germline mutations.
The methylation status of these sites in somatic cells makes them susceptible to C to T
transitions; however no such mutations were identified in 123 neurofibromas. Next, the
alternative splicing of exon23a, the inclusion of which reduces neurofibromin’s RAS-GAP
function, was examined in various tissue types and tumors. Transcripts containing exon23a
(Type II mRNA) are predominant in most tumors; mRNA lacking exon23a (Type I) is
predominant in blood leukocytes. While previously reported in tumors containing increased
10
Type II mRNA, no RNA editing was observed in the tumors in this study. Finally, I tested the
accuracy of computational methods at predicting the effects of NF1 missense mutations
(pathogenic versus neutral). These programs are needed clinically since mutations can not be
tested functionally. No program was 100% accurate, but each had advantages in different
situations. This work contributes to the knowledge in NF1, toward a goal of targeted therapies
and improved diagnosis.
11
CHAPTER 1
INTRODUCTION
Neurofibromin and the NF1 Gene
Neurofibromin is a 220 kD, 2818 amino acid protein expressed in vertebrate and
non-vertebrate animal species, with two homologs in yeast. While it is ubiquitously expressed, it
is found at highest levels in Schwann cells, neurons, and other neural crest-derived cell types
(Daston et al., 1992). Neurofibromin is known to be localized to the cytoplasm and interacts
with microtubules (Gregory et al., 1993), but recent work suggests that it may also travel to the
nucleus (Vandenbroucke et al., 2004) and may localize to different cell compartments based on
stage of development and cell type (Roudebush et al., 1997; Kaufmann et al., 2002; Malminen et
al., 2002; De Schepper et al., 2006). Additionally, four isoforms caused by alternative splicing
have been shown to occur at various developmental stages as well as in specific cell types, which
will be discussed in more detail in Chapter 3 (Nishi et al., 1991; Suzuki et al., 1992; Gutmann et
al., 1993a; reviewed by Skuse and Cappione, 1997). Figure 1-1 shows a diagram of the putative
domains of neurofibromin. While several functional domains have been proposed, including a
microtubule binding domain (Gregory et al., 1993) and a SEC14-like domain (D’Angelo et al.,
2006), the most well-characterized functional domain of neurofibromin is a 360 amino acid
RAS-GTPase activating protein (GAP)-related domain (GRD) (Ballester et al., 1990, Xu et al.,
1990; reviewed by Cichowski and Jacks, 2001). This domain binds activated p21-RAS-GTP and
leads to the hydrolysis of GTP to GDP, inactivating the RAS protein. This inactivation
decreases downstream RAS signaling, and thereby down-regulates cell proliferation and the
inhibition of apoptosis. Since increased RAS signaling is associated with tumorigenesis,
neurofibromin’s inhibitory activity, as well as its gene’s adherence to Knudson’s two-hit
hypothesis (Knudson, 1971), led to its classification as a tumor suppressor (Colman et al., 1995).
12
It is not surprising then, that mutations in the gene encoding neurofibromin predispose
individuals to developing multiple tumor types. Neurofibromin is encoded by the NF1 gene,
which contains 60 exons that span 280 kb of DNA on chromosome 17q11.2 (Cawthon et al.
1990; Viskochil et al. 1990; Wallace et al. 1990; Li et al. 1995). The GRD is encoded by exons
21 to 27a, and mutations in this region have been studied functionally in yeast. Expressing this
domain alone in yeast lacking the yeast NF1 homologs (Ira1, Ira2) can restore a normal
phenotype (Ballester et al., 1990). While the GRD is the best understood domain, mutations
have been found in all exons/domains of the NF1 gene, which lead to neurofibromatosis 1 (NF1).
Neurofibromatosis 1 (NF1)
Although the reasons are unclear, the mutation rate at the NF1 locus is ten times higher
than the average rate in the human genome, making NF1 one of the most common genetic
disorders (Riccardi, 1992; Hughes, 1994). It is inherited in an autosomal dominant manner, and
occurs in approximately 1 in 3,500 births worldwide (Riccardi, 1992). Occurrence rates do not
vary based on sex or ethnicity. Approximately half of all cases brought to medical attention have
no affected parent and represent a de novo NF1 mutation. Of these de novo mutations, 80% of
those that are not deletions are paternal in origin (Stephens et al., 1992). The physical
manifestations of NF1 are numerous and vary greatly in severity among individuals (variable
expressivity) (Carey and Viskochil, 1999). In addition, even within a family, phenotypic
features and severity may vary, although there is less intra-familial variation than inter-familial
(Easton et al., 1993; Szudek et al., 2002). Because of this, seven diagnostic criteria were
established in 1988 by the National Institutes of Health to aid in clinical diagnosis (reviewed by
Gutmann et al., 1997). These seven diagnostic criteria are: six or more café-au-lait macules
(must be 0.5 cm or larger before puberty, 1.5 cm or larger after puberty), neurofibromas (benign
Schwann cell tumors), optic pathway tumors (benign pilocytic astrocytomas), two or more Lisch
13
nodules (benign hyperpigmented lesions of the iris that look like freckles), skeletal dysplasia
(typically sphenoid bone dysplasia or tibial dysplasia), skin fold freckling, and having a first
degree relative with NF1. A patient is clinically diagnosed with NF1 if they meet any two or
more of these criteria. Although molecular genetic testing is available, these criteria remain
important in diagnosis as many patients do not have the expensive genetic analysis done
(McClatchey, 2007). NF1 is a progressive disorder, and these phenotypic features may arise or
increase unpredictably over time. Café-au-lait macules are often the earliest phenotype to
develop, usually present by 2 years of age. Tibial dysplasia and skin fold freckling develop in
the first few years of life as well, with the majority of other phenotypes arising by late
childhood/early adolescence (Riccardi, 1992; Williams et al., 2009). In addition to these criteria,
patients with NF1 are also at increased risk for additional complications, including learning
disabilities, short stature, scoliosis, renal artery stenosis, hypertension, macrocephaly, and
increased risk of certain malignancies, such as rhabdomyosarcoma, pheochromocytoma, and
myeloid leukemia (reviewed by Gutmann et al., 1997).
Of the main phenotypic characteristics, neurofibromas generally cause the most trouble
and discomfort to the patients. Discomfort may include itching, pain, and tenderness around
areas of tumor growth, and furthermore patients can suffer socially due to disfigurement
(Riccardi, 1992; Gottfried et al., 2006). Neurofibromas originate from the peripheral nerve
Schwann cells and are clonal in origin (Skuse et al., 1991; Colman et al., 1995). Schwann cells
are one of several cell types that make up the peripheral nerve sheath, which surrounds the nerve
and axons and helps maintain proper nerve function. The healthy peripheral nerve sheath is well
organized, with Schwann cells closely associated with the axon of the nerve, producing the
myelin used to insulate the axon. Disorganization and an increase in all peripheral nerve sheath
14
cell types are seen in neurofibromas (reviewed by Cichowski and Jacks, 2001). Thus, these
tumors contain primarily clonal, expanded (and somatically mutated, as described below)
Schwann cells, but also contain other cell types normally associated with peripheral nerves, such
as non-clonal Schwann cells, fibroblasts, mast cells, and axons (Serra et al., 2000). The
cutaneous, or dermal type neurofibromas, are derived from Schwann cells associated with nerve
twigs, close to or at the surface of the skin. Some patients develop thousands of these tumors
while others have few or none. They rarely grow beyond one centimeter in diameter and are not
known to become malignant. In contrast, plexiform neurofibromas develop from Schwann cells
associated with larger nerves, are generally deeper in the body, are larger then dermal
neurofibromas, and may involve large areas of the body. It is estimated that 30-40% of NF1
patients develop plexiform tumors, some of which may be asymptomatic nodular masses in the
thorax (Tonsgard et al., 1998). In addition to the risk that a plexiform tumor could grow quite
large and disrupt bone and organs, 10-15% of plexiform neurofibromas will transform into
malignant peripheral nerve sheath tumors (MPNSTs), presumably by the accumulation of genetic
and epigenetic changes (Ferner and Gutmann, 2002). Unfortunately, MPNSTs often become
metastatic and have a poor prognosis unless removed completely by surgery.
How the transformation from benign plexiform neurofibroma to MPNST occurs is unclear.
The development of benign neurofibromas is somewhat better understood and the initiating event
appears to follow Knudson’s two-hit hypothesis of tumor suppressors (1971). Because of the
dominant nature of NF1, the majority of patients are constitutionally heterozygous for an
inactivating mutation in the NF1 gene. In a minority of patients the disease-causing mutation
arose at a point in development shortly after the single cell stage, leading to mosaic distribution
of the mutation (Colman et al. 1996). Regardless, neurofibromas arise when Schwann cells
15
carrying this mutated NF1 allele develop a second mutation that reduces the function of the
previously wild-type NF1 allele (Serra et al., 1997, 2000; reviewed in MacCollin and Wallace,
2002). When a second, somatic mutation or other inactivating event interferes with the function
of the wild-type NF1 allele in a Schwann cell, neurofibroma development is thought to start or at
least be poised to begin upon additional signals. This somatically-mutated Schwann cell then
undergoes clonal expansion, dividing inappropriately, even without axonal contact. Similarly,
optic pathway tumors initiate from two NF1 mutations in an astrocyte (Bajenaru et al., 2003),
tibial dysplasia is associated with two mutations in osteoblasts (Stevenson et al., 2006), and
café-au-lait macules originate from a melanocyte with two NF1 mutations (DeSchepper et al.,
2007).
Mutations in NF1
The NF1 germline mutation rate is estimated between 1/7,800 and 1/23,000, approximately
10-fold higher than average (reviewed in Gottfried et al. 2006). It is unclear why the NF1 locus
is so susceptible to mutation; the size of the gene alone does not explain this phenomenon
(Friedman 1999; Fahsold et al; 2000). The NF1 germline mutation spectrum is broad, with over
1000 germline mutations identified. Of these, 70-80% are clearly inactivating (frameshift and
nonsense mutations), and none recur in more than 2% of cases (reviewed in Thomson et al.
2002). Approximately 20% are aberrant splicing mutations (Messiaen et al., 2000; Serra et al.,
2001; Wimmer et al., 2007; Pros et al., 2008). The exception to this is a 1.2-1.4 Mb
microdeletion of the region of chromosome 17 that contains the NF1 gene and 14 flanking genes.
This microdeletion was first identified in patients with an early age of tumor onset, high tumor
loads, specific facial features, and mental retardation (Kayes et al., 1992, 1994; Leppig et al.,
1996, 1997; Wu et al., 1995, 1997). Later studies, however, revealed patients with no
outstanding phenotype who also carry microdeletions in the NF1 region (Rasmussen et al., 1998;
16
Dorschner et al., 2000). This microdeletion is thought to account for 4-7% of NF1 cases, and
approximately 80% of de novo microdeletions are found on the maternally inherited
chromosome 17 (Lázaro et al., 1996; Valero et al., 1997; Rasmussen et al., 1998; Upadhyaya et
al., 1998; Lopez-Correa et al., 2001; Kluwe et al., 2004). These deletions are mediated by
unequal crossover meiotic events in the germline at mis-aligned repetitive regions. Both the 1.2
and 1.4 Mb versions of the microdeletion share a 5’ repeat sequence containing a pseudogene of
JJAZ1, called JAZFP (Forbes et al., 2004). The 1.2 Mb deletion’s 3’ end is in the JJAZ1 gene,
whereas the 1.4 Mb deletion’s breakpoint is distal to JJAZ1 (Raedt et al., 2006). In addition to
the NF1 gene being deleted, 14 other flanking genes are deleted as well, leaving patients
hemizygous for these genes. These deletions can also occur somatically, early in embryogenesis
(mitotic recombination), resulting in a patient who is mosaic for an NF1 microdeletion
(Rasmussen et al., 1998; Kehrer-Sawatzki et al., 2004) and who may show fewer NF1 features.
A much smaller body of knowledge exists for somatic NF1 mutations. Loss of
heterozygosity was first seen in neurofibromas by Colman et al. (1995). Since then, it has
become clear that in most cases the allelic imbalance is due to mitotic recombination, where a
region of the mutant chromosome replaces that of the wild-type chromosome (Serra et al.,
2001a). Loss of heterozygosity has been found in 10-40% of tumors analyzed (Rasmussen et al.,
2000; Serra et al., 2001; Upadhyaya et al., 2004). Sawada et al. (1996) were the first to identify
a specific somatic mutation in the NF1 gene in a tumor from a patient whose germline mutation
was already known. Since then reports of additional somatic mutations have been limited, and
there is still little known about the somatic mutation spectrum (Rasmussen and Wallace, 1998;
Upadhyaya and Cooper, 1998; Eisenbarth et al., 2000; Wiest et al., 2003; Upadhyaya et al.,
2004). The difficulty of identifying these mutations is due not only to the size and complexity of
17
the NF1 gene, but also to the fact that each tumor is thought to arise from an independent
inactivating event, so that somatic mutations differ not only between individuals, but also
between tumors from the same individual (Colman et al., 1995; Serra et al., 2001b).
Furthermore, the tumors have an admixture of cells and thus only a portion of the cells (the
clonally-expanded Schwann cells) carry the somatic mutation. Further understanding of the
nature and frequency of somatic mutations will provide important information about risk factors,
disease progression and tumorigenesis, as well as help elucidate the pathways that lead to
neurofibroma formation and the transformation to MPNSTs.
This work adds to this body of knowledge by addressing three specific areas of
investigation directly related to the somatic mutation spectrum of NF1. Chapter 2 discusses the
somatic rate of C to T transitions at four sites known to be hotspots for this mutation type in the
germline. Chapter 3 examines alternative splicing of exon23a in NF1-related tumors, its
expression level in various cell types, and its relationship to RNA editing. Chapter 4 addresses
the need for reliable ways to predict the effects of missense mutations (both somatic and
germline) on neurofibromin function, and evaluates several computational methods that make
such predictions.
18
Figure 1-1. Diagram of neurofibromin’s putative domains. Alternatively spliced exons are
shown, along with the number of amino acids they insert. Well-characterized
domains, with known crystal structure, are shown in red. GRD = GAP-related
domain (exons21-26); Sec14 = SEC14 protein-like domain (exons27b-28). Domains
shown in blue are not as well-characterized. TM = Transmembrane domain
(exon10a2); CSRD = Cystein/serine-rich domain (amino acids 543-909);
TBD = Tubulin-binding domain (amino acids 1095-1194); SB = Syndecan-binding
domains (amino acids 1356-1469 (with in the GRD), 2619-2715);
LRD = Leucine-rich domain (Sec14 domain is within the LRD) (amino acids
1530-1950); NLS = Nuclear localization sequence (exon43).
19
CHAPTER 2
SOMATIC CpG C TO T TRANSITIONS AT NF1 GERMLINE HOTSPOTS
Introduction
In mammalian DNA, a cytosine immediately 5’ to a guanine, designated CpG, can be
methylated, having a methyl group attached to carbon 5, producing 5-methylcytosine (5mC).
This is the only nucleotide that can be methylated in mammalian DNA (Bird, 2002). Cytosines
not in the context of CpG are not subject to methylation, with the exception of those found in a
CpNpG context (Clark et al., 1995), and it is believed that the majority of CpG dinucleotides not
in CpG islands are methylated (Cooper and Krawczak, 1993). 5-methylcytosine can be
associated with transcriptionally inactive regions, especially if such methylation occurs in a gene
promoter region where CpGs can be clustered (termed a CpG island) (Bird, 1986) or in other
transcriptional regulatory elements (Cooper and Krawczak, 1993). Cytosine residues in CpG
dinucleotides are susceptible to the spontaneous loss of the amine group at carbon 4 by
hydrophilic attack or by chemical deamination. Once the amine group is lost, a tautomeric shift
can occur. In the unmethylated state this shift produces uracil, while 5mC becomes thymine (a C
to T transition). There are mismatch repair enzymes to recognize and repair both G:U and G:T
mispairs, but G:U mispairing is more efficiently repaired, in part because uracil is not used in the
production of DNA (Cooper and Krawczak, 1993; Lari et al., 2002, 2004). Brown and Jiricny
(1987) found that G:T mispairs (resulting from C to T transitions) were repaired in favor of
guanine 90% of the time. These mispairs were not repaired 2% of the time, and in the remaining
8% of cases, they were corrected in favor of thymine, causing the C to T transitions to become
fixed in the cell’s DNA. These base changes can be neutral (e.g. in non-coding DNA, or a silent
coding-region substitution), or they can cause errors in splicing or coding regions leading to a
premature stop codon, amino acid deletion, or amino acid substitution. No neutral CpG C to T
20
transitions have been reported in the NF1 gene despite studies that would have detected such
changes in the coding regions and UTRs; no such polymorphisms have been found in the 11 Kb
mRNA. When considering 880 base changes previously reported to be involved in genetic
disorders, Cooper and Krawczak (1993) found that 32.8% were C to T or G to A (resulting from
a C to T transition on the antisense strand) mutations at CpG dinucleotides. One such mutation
has been repeatedly identified as a germline mutation in unrelated neurofibromatosis 1 (NF1)
patients, accounting for 1-2% of mutant alleles. This CpG C to T mutation, R1947X (CGA to
TGA), results in a stop codon in exon31 of the NF1 gene (Horiuchi et al., 1994; Lazaro et al.,
1995; Dublin et al., 1995). There are several other such nonsense germline mutations due to
CpG C to T transitions, each of which has a frequency of 0.5-2%: R416X, R440X, R816X,
R1241X, R1276X, R1362X, R1748X, and R2429X. C to T transitions account for 18-30% of
NF1 germline mutations, most causing nonsense codons, the rest missense mutations (Krkljus et
al., 1998; Fashold et al., 2000; Messiaen et al., 2000). Of these, the majority that were de novo
arose in the paternal genome (Jadayel et al. 1990; Wallace et al. 1991; Krkljus et al. 1998),
which may be related to methylation during spermatogenesis (Driscoll and Migeon, 1990). In
contrast, point mutations at CpG sites account for approximately 50% of the reported germline
mutations in TP53 (Greenblatt et al., 1994), ~30% reported for RB1 (Lohmann et al., 1996), and
54% of germline mutations in NF2 (Baser, 2006). These mutations may be under-represented in
NF1 relative to some other genetic disorders.
In neurofibromatosis 2 (NF2), caused by mutations in the NF2 gene, CpG C to T mutations
are reported to account for 38-52% of somatic mutations (Baser 2006). While there is a strong
body of knowledge on the germline frequency of these mutations in NF1, there are no studies
focused on the rate at which they occur somatically. There are multiple mutation screens that
21
have identified somatic NF1 CpG mutations (Sawada et al., 1996; Eisenbarth et al., 2000; Serra
et al., 2001b; Wiest et al., 2003; Upadhyaya et al., 2004, 2008; Maertens et al., 2006; Bottillo et
al., 2009). In many of these studies, mutations were not identified in more then half the samples.
Only 142 mutations of any type were identified out of 372 tumors. These studies screened for
mutations in germline C to T transition hotspots, such as the previously mentioned R1947X
mutation. Of the NF1 somatic mutations identified, 7/142 (4.9%) were CpG mutations. Of the 7
CpG somatic mutations found, 5 were at a previously identified germline mutation site
(Eisenbarth et al., 2000; Wiest et al., 2003; Upadhyaya et al., 2004, 2008; Bottillo et al., 2009).
While NF1 contains more CpG containing codons, and specifically CGA codons (which create a
premature stop codon with a C to T transition) than NF2, it does not appear to be as susceptible
to CpG C to T mutations.
Further identification of somatic mutation mechanisms is important since somatic
inactivation of the wild-type NF1 allele in NF1 patients is the initiating step in neurofibroma
tumorigenesis in Schwann cells. It is also possible that sporadic neurofibromas are the result of
two somatic NF1 mutations in a single cell; one such case has been reported, involving
chromosome translocation (Storlazzi et al., 2005). The goal of my work was to test whether
somatic C to T transitions at CpG dinucleotides in the NF1 gene may be a common mechanism
of generating a second hit in Schwann cells. Furthermore, tumors that have any DNA
hypermethylation (which may be present at low levels in some plexiform neurofibromas, based
on promoter studies by Fishbein et al. (2005)) could be at risk for an increase in CpG transitions
since more cytosines are methylated. Further knowledge of the types and frequencies of somatic
mutations such as C to T transitions will be useful for understanding genetic changes
22
contributing to NF1 tumors. In this work, I will test for somatic presence of each of four
recurrent germline NF1 C to T transition stop mutations in tumor DNA.
Materials and Methods
Mutation Detection by PCR and Restriction Digest
One hundred ninety-seven DNAs were previously prepared by phenol/chloroform
extraction from NF1 patient blood and tumor samples as described in Colman et al. (1993) and
kept in concentrated stocks at 4°C. PCR primers were designed previously for exon10a, 22,
23.2, and 41 of the NF1 gene. These exons contain the mutations R440X, R1241X, R1362X,
and R2429X, respectively. The primers are in the introns flanking each exon, and details are
given in Table 2-1. Diluted DNA samples were used as template (50-100 ng) in polymerase
chain reactions (PCR) under standard conditions. These exon PCR products were then digested
using the TaqαI restriction enzyme in 25 µl reactions incubated at 65°C for 2 hours, with more
enzyme added after 1 hr. The sequence recognized and cut by TaqαI spans the CpG site of
interest in each exon, and any changes in the sequence will result in failure of the enzyme to cut.
These digest reactions were visualized using ethidium bromide staining after electrophoretic
separation on 8% native polyacrylamide gels. With this assay, germline heterozygotes for one of
these mutations show an uncut band plus two smaller bands in the gel. The presence of an uncut
fragment was suggestive of the presence of a mutation and a second digest was used to confirm
the presence of uncut DNA, and controls were used to ensure complete digestion. Table 2-1 lists
the size of both the uncut fragments and the cut fragments produced by TaqαI and methyl-
sensitive digestion for each exon. Numbers of mutation-positive and mutation-negative samples
at each site were determined, as well as total mutations per tumor type.
23
Methylation Status Analysis
In addition to screening for the four mutations, I also investigated whether those cytosines
were methylated in normal somatic Schwann cells, to ensure that they could be a target for 5meC
deamination. This was first done using methyl-sensitive restriction digest, and then bisulfite
genomic sequencing. A schematic of the methyl-sensitive restriction digest method is given in
Figure 2-1. Genomic DNA was used in restriction digest reactions using methylation-sensitive
restriction enzymes (BstBI for sites in exon10a, 23.2 and 41, incubated at 65°C; AvaI for
exon22, incubated at 37°C). These enzymes recognize and cut the sequence spanning the CpG
sites of interest, but cut only if the cytosine residues are unmethylated. The sequence
surrounding the CpG site of interest in exon10a is 5’GGTTGAACTTCGAAATATGTTT 3’; in
exon41 the sequence is 5’TGAAGAAGTTCGAAGTCGCTGC 3’; in exon23.2 the sequence is
5’CCCTCAACTTCGAAGTGTGTGC 3’. These three sites are cut by BstBI, which recognizes
and cuts the sequence 5’TT/CGAA 3’. The sequence surrounding the CpG site of interest in
exon22 is 5’TGAACTAGCTCGAGTTCTGGTT 3. This site contains the recognition sequence
of AvaI, which is 5’C/YCGRG 3’, where Y is either T or C, and R is either A or G. PCR
reactions were then set up using DNA digested with these enzymes as well as untreated DNA.
The primers listed in Table 2-1 were used in these reactions. This same methylation analysis
was carried out on control samples as well as samples from a plexiform neurofibroma. As a
control, PCR samples were set up using the same samples, and these reactions were digested
with TaqαI restriction enzyme as described above. The TaqαI site (TC/GA) is within the
recognition sites of the methyl-sensitive enzymes, and is cut by TaqαI whether the C of interest
is methylated or not. The presence of digested product in these reactions will confirm that the
recognition sites are not mutated in the samples tested, and are therefore able to be digested by
24
the methyl-sensitive enzymes. All digestion products were visualized using ethidium bromide
staining after electrophoretic separation on 8% native polyacrylamide gels.
The CpG site in exon23.2 (R1362X) was chosen for analysis by bisulfite sequencing as an
independent measure of somatic methylation. Genomic DNA from cultured normal human
Schwann cells was subjected to sodium bisulfite treatment per the method given in Fishbein et al.
(2005). The treated DNA was then PCR amplified using primers 23-2MEF
(5’GTTAGAATTATTAGAGAGTTTTGAG 3’) and 23-2MER
(5’ATAATCTCTAACTATAAACATACCTAATA 3’) with an annealing temperature of 54°C.
The sequence of these primers was determined assuming the conversion of unmethylated
cytosine to thymine after sodium bisulfite treatment. These primers amplify a 158 bp fragment
spanning exon23-2 and 23 bp of intron23-2. The PCR products were ligated into a plasmid
vector (Topo TA, Invitrogen) and transformed into chemically competent E. coli cells (One Shot
TOP10, Invitrogen) using the manufacturer’s protocol. The cells were plated on LB-ampicillin
agar plates with IPTG and X-gal, and incubated at 37°C overnight. Bacterial colonies (clones)
positive for the insert were identified by blue/white selection, followed by PCR amplification of
picked colonies using the original primers. The PCR products from five of these clones were
sequenced, using cycle sequencing with the ABI Prism Big Dye 3 system and the UF CEG
sequencing core, with the 23-2MEF primer as the sequencing primer.
Results
Methylation Status Analysis
In one set of experiments, the methylation status of all four CpG C to T germline mutation
hotspots was analyzed using methylation-sensitive restriction digest analysis. I examined these
sites in a normal Schwann cell culture, leukocytes from two non-NF1 patients, and one plexiform
neurofibroma. As mentioned above, this analysis used restriction enzymes that will not cut in
25
the presence of a methylated cytosine. If a cytosine of interest is unmethylated, these enzymes
will cut the genomic DNA, and there will be no PCR product. If methylation is not complete at a
site, there will be uncut band visible after separation by electorphoresis on a PAGE gel, but the
concentration should appear less. As a control, TaqαI digest reactions were also set up, to ensure
that the sites of interest in these samples did not contain any mutations that would prevent either
enzyme from cutting, regardless of methylation status. Representative gels from these
experiments can be seen in Figure 2-2. All four CpG sites are free of mutation in these samples,
as they can be digested by TaqαI (left-most lane of three for each sample). Based on the
presence of roughly equivalent amounts of uncut product in the methylation-sensitive reactions
(right-most lane of three for each sample) when compared to the uncut control reaction (middle
lane for each sample), it appears that the cytosine at each of the four CpG sites is methylated in
all samples. The equivalent amount of product in the uncut control and methyl-sensitive
reactions suggests that these sites are completely methylated, to the level of detection possible
with this method.
To further validate the extent of methylation at these four sites of interest, I chose the CpG
site in exon23.2 for analysis by bisulfite sequencing. In DNA treated with sodium bisulfite,
unmethylated cytosines are converted to uracils by deamination; uracil is not normally present in
DNA (rather it is seen in place of thymine in RNA) and during PCR amplification, thymine takes
the place of the uracil, leading to a C to T change at the unmethylated sites. The cloning of a
sodium bisulfite treated PCR product into a vector allowed for the isolation of a single allele per
bacterial colony. Sequencing was done to reveal the relative number of cytosines versus
thymines at the CpG site of interest, allowing an estimate of the percent methylation at that site.
Figure 2-3 shows the untreated sequence of exon23.2, as well as the chromatogram from the
26
sequencing of one of the clones. The sequence represented in the chromatogram is underlined in
the untreated sequence for comparison. All 5 clones analyzed contained a cytosine at the CpG
dinucleotide of interest (indicating that it is methylated), while all other cytosines in the fragment
had been replaced by thymine. The agreement between the methylation specific restriction
digest results and the bisulfite sequencing results for the exon23.2 site of interest suggests that
the four germline mutation hotspots are all heavily methylated in somatic tissue. These sites are
thus susceptible to CpG C to T transitions.
Mutation Detection by PCR and Restriction Digest
Somatic mutation detection analysis was carried out on DNA from 63 dermal
neurofibromas and 83 plexiform neurofibromas (Table A-1). Each sample was PCR amplified
with each of the four primer sets listed in Table 2-1 and then subjected to restriction digest by
TaqαI. All 4 CpG sites were analyzed in 117 samples, and 1 to 3 of the CpG sites was analyzed
in the remaining 29 (due to lack of PCR amplification by on or more primer pair). Across all the
samples, 531 CpG dinucleotides were analyzed for C to T transitions. Figure 2-4 shows an
example of the visualization of the digest products on an 8% PAGE, with uncut PCR product
indicating presence of a mutation (normal sequence is digested into two smaller products).
While this analysis was used to screen for somatic CpG C to T mutations, it also revealed two
previously-identified constitutional mutations, one at the exon10a CpG site and the other at the
exon23.2 (Figure 2-2) CpG site. These were the only mutations, germline or somatic, seen at
any of the four CpG sites in the 146 neurofibromas, from 82 independent patients.
Discussion
The methylation of cytosine in a CpG dinucleotide is a common occurrence in the human
genome. It is estimated that 70-80% of all CpG dinucleotides contain 5- methylcytosine (5mC)
(Razin and Riggs, 1980). The percent of isolated CpG sites that are methylated is likely higher,
27
as this statistic includes all CpG dinucleotides, both isolated and those found in CpG islands, the
latter of which account for the majority of non-methylated CpGs (reviewed by Bird, 2002).
Based on these observations, it was expected that the four CpG sites studied here would be
methylated in normal tissue. The results of my methylation status analysis confirm that all four
of these sites are indeed predominantly, if not completely, methylated in normal Schwann cells
and neurofibromas.
The positive methylation status of these CpG sites makes them susceptible to C to T
transitions due to the spontaneous deamination of 5meC. Since C to T transitions can occur
without need for replication, this mutation mechanism is feasible in Schwann cells, which are
typically quiescent unless stimulated to divide by injury, or occasional divisions to keep up with
nerve growth. Interestingly, no such mutations were identified in 146 tumors analyzed at the
CpG germline mutation hotspots.
All four of these CpG sites have been previously identified as germline mutation hot spots
with a 1-2% recurrence rate (arginine-to-stop), and the estimated combined germline mutation
rate at these four sites is approximately 7% (they account for 7% of all NF1 germline mutations)
based on our lab’s data and published comprehensive studies from other labs. Thus, if the
somatic mutation rate was equivalent, I would have expected to see approximately 37 total
mutations (7% of 531 sites analyzed) among the four sites in the 146 tumors. Since no somatic
mutations were identified, it appears that the rate of somatic CpG C to T transitions is very low
in neurofibromas, (at least at these sites), despite the presence of a methylated cytosine. The
detection of the two known germline mutations by my methods indicated that this apparent
dearth of somatic CpG C to T mutations is not due to a faulty detection method. The level of
mutation seen in my study is consistent with the low level reported in the NF1 mutation
28
literature. This also suggests that C to T transitions do not play a major role in somatic events in
NF1 tumorigenesis. This is in contrast to the rate of CpG C to T mutations in other, non-NF1
tumors. The rate of somatic CpG C to T mutations in TP53 ranges from 25% in bladder cancer
to nearly 50% in colon cancer (Jones et al., 1991; Greenblatt et al., 1994; Olivier et al., 2002). In
a study of hereditary non-polyposis colorectal cancer, 30.7% (4/13) of somatic APC mutations
were CpG C to T mutations (Huang et al., 2004). The differences in the rate of these mutations
in neurofibromas compared to other tumor types may be due to differences in their ability to
repair this type of mutation. As many cancers exhibit mutations in DNA repair pathways, it may
also be that more malignant tumors are more susceptible to C to T transition mutations, (whereas
neurofibromas are benign) due to reduced ability to repair the mismatches. The status of base
excision repair in neurofibromas has not been analyzed, although it is known that cytogenetic
rearrangements virtually never occur in dermal neurofibromas and less then half the time in
plexiforms, so these tumors tend to have fairly good chromosomal stability (Wallace et al.,
2000). However, NF2 schwannomas are also benign Schwann cell tumors that do not show
chromosomal rearrangements, yet somatic CpG C to T mutations are very frequent. This may be
pointing to basic differences in these tumor types for frequency of 5meC deamination and/or
robustness of excision repair.
Understanding the frequencies and mechanisms of CpG somatic mutation may help predict
whether certain individuals or tumors are more at risk for these, or lead to a specific therapy for
tumors containing such mutations. For example, there are several therapies that have shown
potential in allowing translations through premature stop codons, including gentamicin and
related compounds, and antisense oligonucleotides (reviewed by Ainsworth, 2005; Kulyte et al.,
2005; Pinotti et al., 2006).
29
Genomic DNA
Digest with methylation-sensitive enzymes
(BstBI for e10a, e41; AvaI for e22, e23.2)
PCR with flanking primers PCR with flanking primers
Digest with TaqαI
Polyacrylamide gel Polyacrylamide gel
Unmethylated C Methylated C not Me-sensitive
No band Uncut band Cut band
Table 2-1. Primers used in CpG C to T mutation screening.
Exon Primer sequence (5’-3’)
Product
size (bp)
Annealing
temp (°C)
TaqαI digest
band sizes (bp)
5’ ACGTAATTTTGTACTTTTTCTTCC 10a
3’ CAATAGAAAGGAGGTGAGATTC 222 60
105
117
5’ TGCTACTCTTTAGCTTCCTAC 22
3’ CCTTAAAAGAAGACAATCAGCC 331 62
87
244
5’ TTTTAAGGAGTGATTTTTGTTATTTG 23.2
3’ CCTTCTTTGATAAAGCATTCTTC 276 55
179
97
5’ TTCATCCTGTTTTAAGTCACACTTG 41
3’ TTGCCTCCATTAGTTGGAAAATTG 273 60
94
179
Figure 2-1. Methylation-sensitive restriction digest protocol.
30
Figure 2-2. Visualization of methylation-specific restriction digest analysis. a: normal human
Schwann cell culture; b and c: non-NF1 patient blood; d: plexiform neurofibroma
DNA.
GTTAGAACCATCAGAGAGCCTTGAGGAAAACCAGCGGAACCTCCTTCAGATG
ACTGAAAAGTTCTTCCATGCCATCATCAGTTCCTCCTCAGAATTCCCCCCTCA
ACTTCGAAGTGTGTGCCACTGTTTATACCAGTTTATACCAGGTATGCTTACAG
TTAGAGATTAC
Figure 2-3. Results of bisulfite sequencing. The sequence given is the untreated sequence of
exon23.2. Upon treatment with bisulfite followed by sequencing, unmethylated Cs
are converted to Ts, methylated Cs are not converted. All Cs are in blue, the CpG site
of interest is in red. The chromatogram shows the underlined region of sequence
after bisulfite treatment. The only C not converted to T is in the CpG dinuleotide,
indicating it is methylated.
E10a E22
m a b c d m m a b c d m
TaqI + - - + - - + - - + - - + - - + - - + - - + - -
BstBI - - + - - + - - + - - + AvaI - - + - - + - - + - - +
m a b c d m m a b c d m
TaqI + - - + - - + - - + - - + - - + - - + - - + - -
BstBI - - + - - + - - + - - + BstBI - - + - - + - - + - - +
E23.2 E41
31
U M
Figure 2-4. TaqαI digests of E23.2 PCR products visualized on 8% PAGE. All samples are from
plexiform neurofibromas. The sample indicated by the asterisk is from a patient with
a known constitutional CpG mutation. C= control with known mutation at cut site;
U= uncut sample; M=1 kb marker.
*
*
32
CHAPTER 3
ALTERNATIVE SPLICING OF EXON 23a AND mRNA EDITING
Introduction
The alternative splicing of exons allows for the production of multiple different transcripts
from a single gene. This usually affects the coding region in between invariant translation start
and stop sites, but this phenomenon can also produce transcripts with different transcription or
translation start and/or end sites. Alternative splicing is often regulated in a spatial,
tissue-specific, and/or temporal manner, and its effects can generally be divided into five
categories: effects on localization of the resulting protein, elimination of the resulting protein’s
function, changes in the resulting protein’s function, creation of a new function of the protein,
and effects at the RNA level, such as transcript stability or efficiency of translation (reviewed by
Smith et al., 1989). While there are many examples of alternative splicing events, there is also
evidence that disregulated alternative splicing plays a role in human cancers (Early et al., 1980;
Nagoshi et al., 1988; Lee and Feinberg, 1997; Stimpfl et al., 2002; Adams et al., 2002). One
array study identified 845 alternative splicing isoforms from throughout the genome that appear
to be tumor associated (Wang et al., 2003). While no NF1 isoforms were included in this data
set (which also did not include Schwann cell tumors), it is known that several common NF1
isoforms normally exist at at least a 10% level relative to the major isoform in the pertinent
tissue. There are three alternatively spliced exons in NF1, as well as a short isoform that lacks
exon11 through most of exon49 (Nishi et al., 1991; Suzuki et al., 1992; Gutmann et al., 1993a;
reviewed by Skuse and Cappione, 1997). The first of these alternatively spliced exons to be
identified was exon23a (Nishi et al., 1991). The inclusion of this exon produces an mRNA
containing 63 additional base pairs from intron23-2, leading to an in-frame insertion of 21 amino
acids in neurofibromin, and is known as the Type II isoform. These additional amino acids are
33
within the NF1 GAP-related domain (GRD), and have been shown to change the hydrophobicity
and secondary structure of this region of the protein (Nishi and Saya, 1991). While this isoform
is still able to bind RAS and has GAP activity, Andersen et al. (1993) found that cells expressing
the Type II isoform contained much more RAS-GTP (activated RAS) than cells expressing
mostly the Type I isoform (lacking exon23a). This was shown to be due to ten-fold decreased
GAP activity of the protein encoded by NF1 Type II mRNA. It has also been shown that Type II
neurofibromin does not associate with microtubules as Type I does (Gutmann et al., 1995). It
has been suggested that in addition to altering or eliminating known functions of neurofibromin,
the insertion of exon23a may introduce novel functions to the protein. Teinturier et al. (1992)
found that exon23a showed sequence homology with vaccinia virus nucleoside triphosphotase I,
and Andersen et al. (1993) point out similarities between the 21 amino acid insert and common
nuclear localization signals.
While the functional effects caused by the inclusion of exon23a have yet to be fully
characterized, Costa et al (2001) found that exon23a knockout mice had an increased incidence
of cognitive deficits, relative to the exon31 knockout mouse (which has no cognitive
impairment). It also appears that this isoforms is developmentally significant, as there is a
switch from the predominant expression of Type I to Type II neurofibromin through
embryogenesis into post-natal life in the majority of tissues in rat, mouse and chick, as well as
differentiating human cell types (Nishi et al., 1991; Baizer et al., 1993; Gutmann et al., 1994,
1995; Huynh et al., 1994; Mantani et al., 1994). Although there are some conflicting reports, the
consensus is that the Type II transcript is typically the predominant transcript, present at levels
greater than or equal to Type I in most adult tissues, with the exception of the adult central
nervous system (Suzuki et al., 1992; Uchida et al., 1992; Teinturier et al., 1992; Baizer et al.,
34
1993). Two of these studies also indicated that in various non-neurofibromatosis 1 (NF1)
associated cancers the expression of Type II mRNA is preferentially increased (Suzuki et al.,
1992; Teinturier et al., 1992). Ogata et al. (2001) found that a particularly malignant breast
cancer cell line (MDA-MB-231) does not express any type I NF1 mRNA, further supporting a
role of aberrant alternative splicing of exon23a in tumors.
There are two other well-documented NF1 isoforms. One is expressed exclusively in the
central nervous system. This isoform includes exon9br, which encodes 10 additional amino
acids inserted at residue 420 of the protein (Danglot et al., 1995; Geist and Gutmann, 1996). The
other includes exon48a, which is a muscle-specific 18 amino acid insertion near the C-terminal
end of the neurofibromin molecule (Gutmann et al., 1993a). These were not included in this
tumor study since they do not affect RAS-GAP activity, and do not appear to be involved in NF1
tumorigenesis.
In addition to these well-studied isoforms, there have been several reports of multiple rare
novel splice variants of NF1 mRNA (Thomson and Wallace, 2002; Vandenbroucke et al., 2002a,
2002b). While some of these alternative splice events insert or delete intronic or exonic
sequence, many of them involve exon skipping due to splicing at the same sites used in normal
NF1 RNA (Thomson and Wallace, 2002; Vandenbroucke et al., 2002b). Some of these produce
out-of-frame transcripts. Thomson and Wallace (2002) found that the conditions under which
blood samples were drawn, or length of time stored before RNA was isolated, impacts this rare
variant splicing profile of NF1, with the frequency of these increasing over time. However,
relative ratios of the Type I to Type II mRNA are not affected. Some of these novel isoforms
have been shown to occur in a tissue specific manner, implying that they may be functionally
significant despite relatively low levels (Vandenbroucke et al., 2002a, 2002b). It is clear that
35
these isoforms do exist in vivo, but further study is needed to determine the role of these rare
NF1 isoforms in normal tissues and in NF1 phenotype development.
Another form of post-transcriptional modification that has been shown to occur in NF1
mRNA is RNA editing. Skuse et al. (1996) found that some NF1 mRNA undergoes base
modification editing at position 3916. An in-frame stop codon (R1306X) is created by the
deamination of the transcript’s cytosine at bp 3916 of the mRNA to form a uracil. This
nucleotide is within the region encoding the neurofibromin GRD. It is unclear what effect this
premature stop codon may have on the cells that express the edited mRNA, but if translated it
would produce a truncated protein that would likely be degraded or lack full function.
Additionally, editing can occur in mRNA transcribed from either allele, potentially inactivating
that allele regardless of NF1 gene mutation state. While this editing has been found at low levels
(1.5-2%) in all cell types studied, there has been evidence that NF1 mRNA editing occurs at
somewhat higher levels in tumor cells (Skuse et al., 1996; Cappione et al., 1997; Mukhopadhyay
et al., 2002). Cappione et al. (1997) observed a correlation between increased invasiveness of
tumors with increased levels of mRNA editing, with malignant tumors having higher levels of
editing than plexiform neurofibromas, which in turn had higher levels of editing than dermal
neurofibromas. Importantly, this trend at the NF1 gene does not appear to be due to an overall
increase in general mRNA editing in malignant cells (Cappione et al., 1997). Mukhopadhyay et
al. (2002) evaluated RNA editing in malignant peripheral nerve sheath tumors (MPNSTs) and
found that 76.5% of tumors examined exhibited low levels of mRNA editing (0-2.5%), near the
reliable detection limits of their assay. However, they also identified a distinct subpopulation of
these tumors (23.5%) that exhibited mRNA editing at higher levels (3-12%). Tumors that
exhibited this higher, reproducible level of mRNA editing at position 3916 also exhibited
36
increased levels of Type II mRNA relative to Type I. The creation of this stop codon in some
transcripts in cells expressing Type II mRNA would further reduce neurofibromin function by
leading to higher levels of activated RAS. These data were suggested to support a connection
between RNA editing, alternative exon23a splicing, and tumorigenesis. The extent of these
post-transcriptional modifications in NF1 tumorigenesis has yet to be validated by multiple
independent groups. I examined alternative splicing of exon23a and mRNA editing in our set of
NF1 tumors and normal Schwann cells to test for a connection between these two
post-transcriptional modifications. It is hoped that these studies will further clarify the roles of
these modifications in NF1 tumorigenesis and provide potential new avenues of treatment, aimed
at altering mRNA expression and post-transcriptional modifications.
Material and Methods
Reverse-Transcription and PCR
RNA was previously isolated from blood (8 non-NF1 patients, 7 NF1 patients), tumor (22
dermal neurofibromas, 21 plexiform neurofibromas, 6 MPNSTs), and culture samples (3 normal
Schwann cell, 4 dermal tumor cultures, 9 plexiform tumor cultures, 2 immortalized plexiform
neurofibroma cell lines) using the Trizol reagent and manufacturer’s protocol (Invitrogen), and
stored at -80°C. Reverse transcription reactions were carried out using Superscript II reverse
transcription kit (Invitrogen) and random hexamer primers. The resulting cDNAs were used in
PCR reactions under standard conditions using primers in exon23.1 and exon24, flanking
exon23a (CAT-H: 5’ ATTGTGATCACATCCTCTGATTGG 3’;
CAT-I: 5’ ATCTAAAATCCCTGCTTCATACGG 3’). Two fragments were amplified, one each
from Type I and Type II mRNA (303 and 366 bp, respectively). The two isoforms were
separated by electrophoresis on 8% native polyacrylamide gels and visualized by ethidium
bromide staining. Based on visual observation of band intensities, the relative ratio of Type I to
37
Type II NF1 mRNA was noted (I<<II, I<II, I≈II, I>II, etc.) and compared between tumor
samples and controls, and between tumor types. Samples having only (or mostly) type II mRNA
were selected for cloning and sequencing to determine if RNA editing at C3916 was occurring in
these tumors.
Cloning and Sequence Analysis
New RT-PCR products were generated from the cDNA samples identified above, using
primers E21c5’ (5’CACAATGATGGGTGATCAAGG 3’) and
E23.1c3’ (5’CATGTTGCCAATCAGAGGATG 3’), spanning the 3916 edit site (377 bp
fragment from Type II mRNA). The products were ligated into a plasmid vector (Topo TA,
Invitrogen) and transformed into chemically competent E. coli cells (One Shot TOP10,
Invitrogen) using the manufacturer’s protocol. The cells were plated on LB-ampicillin agar
plates with IPTG and X-gal, and incubated at 37°C overnight. Bacterial colonies (clones)
positive for the insert were identified by blue/white selection, followed by PCR amplification
using the original primers. The PCR products from 30 positive clones were sequenced. Cycle
sequencing was done using the ABI Big Dye 3 system and automated sequencers in the UF CEG
sequencing core, with one of the PCR primers as a sequencing primer. Controls were used at all
steps.
Results
Alternative Splicing Patterns
I investigated the relative concentrations of Type I versus Type II mRNA in fresh blood
leukocytes, dermal neurofibromas, plexiform neurofibromas, MPNSTs, and multiple cell
cultures including normal Schwann cells. The use of a single primer pair to amplify both
isoforms helps preserve the ratio present in the original sample. Multiple samples of each tissue
type were analyzed, to determine if there was a common splicing pattern for the given cell type,
38
and how that compared to the literature. A summary of the results of this analysis is given in
Table 3-1, and a detailed list of samples analyzed can be found in Appendix B (Table B-1).
In leukocytes from non-NF1 patients, Type I mRNA was present at equal to or greater
levels then Type II in 7/7 samples, with the majority showing Type I as the predominant
transcript (5/7). Results from these samples are shown in Figure 3-1A. From left to right in the
figure, lane 1 and 2 show more Type I than Type II, lanes 3 and 4 show approximately equal
levels of the two isoforms, and lanes 5-7 again show more Type I than Type II transcript. In
leukocytes from seven NF1 patients, there was more variability between samples. Two out of
seven showed Type II as the predominant transcript, three out of seven had approximately equal
amounts of the two isoforms, and the remaining two had more Type I transcript than Type II.
Three cultures from normal human Schwann cells were analyzed. All three of these
cultures contained Type II mRNA as the main transcript, with one culture exhibiting a much
higher level of Type II compared to Type I than the others.
Both primary dermal neurofibroma tissue and Schwann cell cultures derived from dermal
tumors were analyzed for their exon23a splicing patterns. Of the 21 primary dermal
neurofibroma samples, there were two tumors from which corresponding cultures were also
analyzed. An additional two dermal tumor-derived cultures were analyzed. The majority of
primary dermal tumors contained predominantly Type II transcript (18/21), with 7 having very
low levels of Type I in comparison. Two dermal tumor samples, UF80T32 and UF505T4, had
Type I as the main transcript. One primary dermal sample, UF80T2, contained no Type I
transcript, and was selected for cloning and sequencing to test for RNA editing. All 4 dermal
tumor-derived cultures contained predominantly Type II transcript. Figure 3-1B shows
experimental results from some representative dermal samples. Lane 1 is a sample with more
39
Type I mRNA, lanes 2-5 and 7 are samples with predominantly Type II transcript, and lane 6 is
UF80T2 (which has no detectable Type I transcript).
Twenty-five plexiform neurofibromas and nine cultures derived from plexiform tumors
were also analyzed for exon23a splicing. Of the nine cultures analyzed, the corresponding
primary tumor was also analyzed for four. The majority of primary plexiform tumors (23/25)
contained predominantly Type II transcript, three of which had very little Type I in comparison,
and two having only trace amounts of Type I. Of the remaining two tumors, one had about equal
levels of Type I and Type II mRNA, and the other had Type I as the predominant transcript. The
majority of plexiform tumor-derived cultures (8/9) also had Type II as the predominant
transcript, with 3 having barely detectable levels of Type I transcript. The remaining culture,
UF469Tc, contained only Type II transcript and was also selected for RNA editing analysis.
Figure 3-1B shows the experimental results from representative primary plexiform tumors, and
one plexiform-derived culture. Lanes 8-10, and lane 12 show samples with predominantly Type
II transcript, and lane 11 is UF469Tc, with only Type II mRNA.
Finally, 6 MPNSTs were analyzed. All 6 had Type II mRNA as the predominant
transcript, with 3 showing much lower levels of Type I in comparison. Figure 3-1C shows the
experimental results for representative MPNSTs. Lanes 1 and 2 show samples with much more
Type II then Type I transcript, while lanes 3 and 4 show samples with predominantly Type II
mRNA. One MPNST, SNF94.3, had previously been analyzed for exon23a splicing levels, and
was found to contain only Type II transcript. This sample was also chosen for RNA editing
analysis.
Analysis of RNA Editing
Three samples were selected for RNA editing analysis based on their exon23a splicing
patterns. All three samples contained only Type II transcript, based on polyacrylamide gel
40
electrophoresis, but they were all from different tissue types. UF80T2 is a primary dermal
neurofibroma, UF469Tc is a plexiform neurofibroma-derived Schwann cell culture, and
SNF94.3 is an MPNST. Plasmid vectors containing individual cDNA fragments were cloned in
E. coli and then PCR amplified for sequencing. This allowed for the analysis of the sequence of
a single cDNA molecule at a time, to detect the number of transcripts that were undergoing RNA
editing at C3916. Thirty-five (UF469Tc) to Forty (SNF94.3 and UF80T2) clones were analyzed
for each sample of interest. This number was chosen based on the levels of RNA editing
previously detected in MPNSTs lacking Type I transcript (3-12%) (Mukhopadhyay et al., 2002).
Even at the lowest levels previously seen, at least one in forty clones would be expected to
contain an edited cDNA fragment. No RNA editing was seen in any of these tumor samples,
despite previous reports of increased levels of RNA editing in tumors containing predominantly
Type II transcript.
Discussion
The relative ratios of Type I to Type II NF1 mRNA in a given tissue sample can be
determined using a single set of PCR primers to amplify the fragment of interest. The alternative
splicing pattern of exon23a has previously been studied for many embryonic and adult solid
tissues in mouse and human, but the levels of alternative splicing seen in blood leukocytes were
not well established. I observed that the ratio of Type I to Type II NF1 transcript in non-NF1
blood leukocytes is similar to that reported in the adult human central nervous system, with all
the samples (7/7) showing Type I transcript at equal or greater levels compared to Type II, with 5
of those 7 having Type I as the predominant transcript. This is in contrast to most other human
and mouse postembryonic tissues studied, where Type II predominates. The deviation from this
pattern in leukocytes is interesting. It is known that some alleles with NF1 mutations have
decreased mRNA levels, presumably due to nonsense-mediated decay (Colman et al., 1993;
41
Hoffmeyer et al., 1995; Pros et al., 2006). However, there should be no such mechanism
operating in normal leukocytes, and it is known that this alternative splicing affects both alleles
(Thomson and Wallace, 2002). Thus, there may be functional reasons in leukocytes and the
central nervous system for a relative lack of Type II transcript, and the presence of an NF1 gene
mutation shouldn’t affect relative ratios of Type I to Type II transcript. Yet in blood leukocytes
from NF1 patients, there is more variation in the relative ratio of Type I to Type II transcript
observed. Only 5/7 samples showed relative ratios similar to those seen in the majority of
non-NF1 bloods, with the majority of those 5 having equal levels of the two transcripts. It
appears that there may be a trend toward inclusion of exon23a in the mRNA from NF1 patient
leukocytes, compared to leukocytes from non-NF1 patients.
To determine if the level of exon23a inclusion varied based on tumor type, I analyzed the
relative ratio of Type I versus Type II mRNA in both dermal and plexiform neurofibroma
samples. There was no significant difference in these relative ratios between the two types of
neurofibromas. The majority of both dermal (18/21) and plexiform (23/25) neurofibromas had
Type II as the predominant transcript. This is not unexpected as cultures of normal human
Schwann cells, the cell type from which neurofibromas are clonally-derived, all contained more
Type II mRNA then Type I. Observation of a relative (but not dramatic) predominance of Type
II transcript suggests a relatively reduced RAS-GAP activity in those cells, which would be
inferred for most normal adult tissues, including, as I have shown here, Schwann cells, based on
their Type I to Type II ratio. However, many of the NF1 tumors had a profile of Type II
transcript at much higher levels relative to Type I, with several tumors, of each variety of
neurofibroma, showing near or complete loss of the Type I transcript. This could result in
sufficiently less RAS-GAP activity relative to the native Schwann cell, which could contribute to
42
tumorigenesis. If true, then a mechanism to shift the ratio back toward more equal levels of the
two transcripts could potentially decrease tumorigenic potential.
Previous reports had indicated that sample collection conditions and other environmental
factors may influence the level of alternative splicing in NF1 mRNA (Thomson and Wallace,
2002). To determine if culturing condition altered the level of exon23a inclusion, I examined the
relative ratios of Type I to Type II NF1 mRNA in several primary tumor and corresponding
tumor Schwann cell culture pairs. Six such pairs were analyzed: two were dermal neurofibromas
and their corresponding cultured Schwann cells, and four were plexiform neurofibromas and
their corresponding Schwann cell cultures. All 6 primary tumors analyzed contained
predominantly Type II mRNA, and there was no obvious change in the relative ratio of the two
isoforms compared to the corresponding cultures. The result of one of these comparisons is
shown in Figure 3-2 (plexiform neurofibromas). These results suggest that there is an inherent
Type I:Type II control mechanism in these tumor Schwann cells that is not susceptible to
influences of tissue culture, and it appears that the relative ratio of Type I to Type II mRNA seen
in a tumor cell culture can be a good estimate of the ratio seen in the primary tumor.
It was previously reported that MPNSTs containing little or no Type I transcript may
exhibit RNA editing (Mukhopadhyay et al., 2002). All six of the MPNSTs analyzed here
showed Type II as the predominant transcript and one (SNF94.3) that had been previously
analyzed showed complete loss of the Type I transcript by ethidium bromide visualization. I
analyzed this MPNST, as well as one dermal neurofibroma and one plexiform
neurofibroma-derived culture, both of which also showed loss of the Type I transcript, for the
presence of RNA editing. Thirty-five to forty clones from each sample were sequenced to check
for RNA editing at C3916, but this edit was not seen in any of the three samples with little to no
43
Type I transcript. Mukhopadhyay et al. (2002) reported levels of RNA editing in their MPNSTs
of interest at 3-12%. If RNA editing occurred to a similar extent in all tumors showing a near or
complete loss of the Type I transcript, I would have expected to see 1-4 clones out of 40
undergoing RNA editing. The absence of editing in my samples could have been due simply to
chance. However, it could imply that any connection seen between increased levels of Type II
mRNA and RNA editing is not universal, even in MPNSTs of this type, or that the amount of
such editing is quite low, in which case its functional significance would seem minimal.
While the creation of a premature stop codon in NF1 mRNA by RNA editing has a clear
negative effect on neurofibromin function, the effect of having predominantly Type II transcript
may also have a negative impact, as the protein encoded by this transcript has a reduced
RAS-GAP activity. The first MPNST that I studied for RNA editing has one of the germline
NF1 microdeletions, but no somatic mutation has been found in the remaining allele despite
sequencing the entire mRNA open reading frame and the 3’ UTR. This raises the interesting
possibility that the alternative splicing in this tumor substitutes as a mutation, producing a
hypomorphic allele. The cells thus would lack sufficient NF1 RAS-GAP activity to prevent
tumorigenesis. This suggests that there may be a threshold of neurofibromin activity that keeps a
cell from becoming tumorigenic. This is an important notion that could have relevance for future
therapies. Alternatively, there could be one of several epigenetic changes (other than RNA
editing) that could constitute the second hit in this MPNST, and possibly other NF1 tumors.
These include histone modifications, changes in microRNA effects, and changes in downstream
regulatory effects (Schmegner et al., 2005; Ling et al., 2006; Martinez and Schackert, 2007;
Shelton et al., 2008; Bartels and Tsongalis, 2009). The latter is evidenced by studies by Hawes
et al. (2007) that found that different mouse strains have different levels of Nf1 expression. This
44
implies that different background levels of transcription factors can have an effect on NF1
expression levels (Zhu et al., 2004).
The observation of different Type I to Type II ratios in two dermal neurofibromas from the
same individual (UF80T2 and UF80T32; UF505T4 and UF505T7) suggests that this ratio is
specific to each tumor and is not heavily controlled by systemic factors. This is consistent with
the fact that each neurofibroma has a different NF1 somatic mutation (and possibly other genetic
or epigenetic alterations) and is therefore independent.
45
Table 3-1. Summary of alternative splicing of exon23a seen in various sample types examined.
mRNA levels
Cell Type
Total
sample # No Type I Type I < II Type I ≈ II Type I > II
Normal Schwann cell culture 3 3 (1+)
Non-NF1 patient blood 7 2 5
Non-NF1 patient fibroblasts 1 1
NF1 patient blood 7 2 3 2
Primary dermal tumor 21 1 18 (5+, 2*) 2
Dermal tumor cell culture 4 4 (2+)
Primary plexiform tumor 25 23 (3+, 2*) 1 1
Plexiform tumor cell culture 9 1 8 (1+, 3*)
Immortalized plexiform cell
lines 2 2
MPNSTs 7 1 6 (3+)
+: Much more Type II; *: Barely detectable Type I in comparison to Type II.
46
A
B
C
Figure 3-1. Representative gels showing relative concentrations of Type I v Type II mRNA in
various tissue types studied. Upper band represents Type II mRNA, lower band Type
I. A) leukocytes from 7 non-NF1 patients. B) Primary tissue samples from dermal
(lanes 1-7) and plexiform (lanes 8-12) tumors. C) Cultured (left two lanes) and
primary tissue (right two lanes) from MPNSts.
Figure 3-2. Comparison of alternative splicing of exon23a in primary plexiform tumors v
cultured Schwann cells from the same tumors. Tumor culture sample is loaded first,
corresponding primary tumor sample in the lane to the right. T= primary tumor,
C= tumor culture.
1 2 3 4 5 6 7
1 2 3 4 5 6 7 8 9 10 11 12
1 2 3 4
C T C T
47
CHAPTER 4
MISSENSE MUTATION COMPUTATIONAL ANALYSIS OF PATHOGENICITY
Introduction
Missense mutations are single base substitutions that result in an altered mRNA codon,
leading to an amino acid substitution at the protein level. In neurofibromatosis 1 (NF1),
missense mutations account for 10-20% of disease-causing germline lesions (reviewed by
Thomson and Wallace, 2002). Missense mutations a priori may be pathogenic or represent
neutral polymorphisms. For this reason, determining the pathogenicity of such mutations is
critical and is a major challenge in molecular diagnosis. Cooper and Krawczak (1993) outline
eight points of evidence that may indicate a missense mutation is pathogenic: if the mutation is in
an important structural of functional region of the gene; if the mutation alters an highly
conserved codon; if there are multiple unrelated reports of the mutation in patients; if there is no
observation of the mutation in healthy individuals; if the mutation segregates with the disease
phenotype within a family; if the mutant protein produced in vitro has the same properties and
characteristics as protein produced in vivo; and if introduction of wild-type protein can rescue
the disease phenotype in patients or culture. The first 6 of these points can be useful in
determining the pathogenicity of NF1 missense mutations, but the general difficulty of this
process is exacerbated by the fact that most NF1 mutations cannot be tested functionally in the
lab, eliminating the final 2 points of evidence. Thus, the finding of an NF1 missense mutation in
a person lacking sufficient diagnostic criteria can be a clinical dilemma. Missense mutations
should also be tested for a cryptic splicing effect, to best understand pathogenesis. When there
are no splicing errors, and if there is no useful information from the family or literature regarding
a novel mutation’s effects, other methods must be used to predict the mutation's pathogenicity.
48
It has been estimated that over 50% of gene lesions known to cause hereditary disorders in
humans are missense mutations (Cooper et al., 1998), and predicting the effects of these
mutations on their corresponding proteins as well as their contribution to disease can be difficult
as well. Previously computational-based methods were designed to aid in the understanding of
the importance of specific amino acids in protein structure and function. These original methods
did not take into account all information specific to the protein of interest, and were not generally
designed to predict the effects that a missense mutation will have on its protein (Henikoff and
Henikoff, 1992; Ng and Henikoff, 2001). Rather, they provided information about the likelihood
of finding a particular amino acid at a particular position based on ortholog sequence alignments.
Additionally, while some studies have used these likelihoods to extrapolate pathogenicity, this
use has not been experimental validated. In recent years, however, several new computational
methods have been developed for the purpose of predicting the effect of a missense mutation,
aided partially by the increasing amount of genomic sequence information available, and taking
into account more factors specific to the protein, such as biochemical properties of the amino
acids in specific regions of the protein. The first of these new program was SIFT (Sorting
Intolerant From Tolerant amino acid substitutions (Ng and Henikoff, 2001, 2002)), which uses
protein sequence orthologs to predict the tolerance of a particular amino acid at a particular site.
SIFT has been compared to traditional substitution matrices in the accuracy of predicting the
effect of amino acid substitutions in LacI, HIV-I protease, and bacteriophage T4 lysozyme, and
in all cases was more accurate overall (Ng and Henikoff, 2001). As new programs have been
developed, their predictions have been compared to those from SIFT as a test of each program’s
efficiency and accuracy. These programs include MAPP (Multivariate Analysis of Protein
Polymorphism (Stone and Sidow, 2005)) and SNPs3D (Yue and Moult, 2006). These two latter
49
programs have been shown to be more accurate in their predictions than SIFT, using small, well
characterized proteins for analysis, with functional analyses available to confirm the results. In
comparison to these previously analyzed proteins, the NF1 protein product, neurofibromin, is a
much larger, 2818-amino acid ubiquitously-expressed peptide (plus or minus a few alternative
exons) (DeClue et al., 1991; Marchuk et al., 1991). With its large size and complex nature,
neurofibromin was examined here as a robust test of the efficiency and accuracy of these new
prediction programs, not only in comparison to SIFT, but to each other. This work has been
submitted for publication (Loda-Hutchinson et al., 2009).
Materials and Methods
Missense Computational Methods
Three freely available programs were chosen based on their reported performance, as well
as applicability to the analysis of neurofibromin (lack of structural data, etc.). The three
programs chosen were SIFT: Sorts Intolerant From Tolerant
(http://blocks.fhcrc.org/sift/SIFT.html (Ng and Henikoff, 2001; 2002)),
SNPs3D: (http://www.snps3d.org/ (Yue and Moult, 2006)), and MAPP: Multivariate Analysis of
Protein Polymorphism (http://mendel.stanford.edu/SidowLab/downloads/MAPP/MAPP.html
(Stone and Sidow, 2005)).
Other Databases and Programs
The following databases and programs were also used.
For sequence acquisition and analysis:
• NCBI
• http://snpper.chip.org
• SwissPROT/TrEMBL (http://www.expasy.ch/sprot/)
50
For sequence alignment:
• CLUSTAL (www.clustal.org)
For ortholog data:
• Inparanoid (http://inparanoid.sbc.su.se/cgi-bin/index.cgi (O’Brien et al. 2005))
To generate phylogenetic trees:
• SEMPHY-Structural EM Phylogenetic Reconstruction
(http://compbio.cs.huji.ac.il/semphy/ (Ninio et al. 2006))
For prediction of novel splice sites created by mutations:
• NNSPLICE version 0.9 (www.fruitfly.org/seq_tools/splice.html (Reese et al. 1997))
Data Sets
Compiled from a literature search, the Human Gene Mutation Database (www.hgmd.org),
and unpublished laboratory findings, the three data sets included missense mutations of known
effect, mutations created by site-directed mutagenesis in neurofibromin’s isolated GAP Related
Domain (GRD) and tested for RAS-GTP activity, and missense mutations of unknown effect.
For mutations known to create cryptic splice sites, the amino acid residue that would normally
result from the new codon was used for analysis. Sample size: known germline neutral n=8,
known germline pathogenic n=18, neutral in site-directed mutagenesis n=7, pathogenic in
site-directed mutagenesis n=5, Known pathogenic due to splicing error n=8, and unknown n=39.
References listed for these mutations do not necessarily include every report. I reported these
references to our best ability from the literature and the Human Gene Mutation Database
(www.hgmd.org).
Sequence Input Requirements
For SIFT, the human neurofibromin amino acid sequence (gi|4557793|ref|NP_000258.1|
neurofibromin [Homo sapiens]) was provided as a reference, and the program assembled a
51
multiple sequence alignment (MSA) from similar sequences found in the SwissPROT/TrEMBLE
database. In total, 43 sequences were aligned by the SIFT program, and saved for further use.
All “-” in the returned alignment were converted to “X” to allow proper analysis by SIFT when
the sequences were re-entered (Ng, personal communication). Not all sequences were
considered at all mutation sites (due to high sequence similarity); SIFT returned data as to the
number of sequences used in the analysis of each mutation, but not on which sequences these
were. Listed below is the identification information provided by SIFT for these sequences.
• NF1_RAT Neurofibromin (Neurofibromatosis-related protein NF-1)
• NF1_HUMAN Neurofibromin (Neurofibromatosis-related protein)
• Q5SYI1_MOUSE (Q5SYI1) Neurofibromatosis 1
• NF1_MOUSE Neurofibromin (Neurofibromatosis-related protein)
• Q5SYI2_MOUSE (Q5SYI2) Neurofibromatosis 1
• Q9YGV2_FUGRU (Q9YGV2) Neurofibromatosis type 1
• Q59DT9_DROME (Q59DT9) CG8318-PD, isoform D
• Q9VBJ2_DROME (Q9VBJ2) CG8318-PB, isoform B
• O01399_DROME (O01399) Neurofibromin
• O01398_DROME (O01398) Neurofibromin
• O01397_DROME (O01397) Neurofibromin
• Q7QBJ9_ANOGA (Q7QBJ9) ENSANGP00000003216 (Fragment)
• Q7PGW6_ANOGA (Q7PGW6) ENSANGP00000025084 (Fragment)
• Q8IMS2_DROME (Q8IMS2) CG8318-PC, isoform C
• Q4T1K3_TETNG (Q4T1K3) Chromosome 16 SCAF10562, whole genome shotgun
• Q4T1K5_TETNG (Q4T1K5) Chromosome 16 SCAF10562, whole genome shotgun
• Q8WZ-6_NEUCR (Q8WZ-6) Related to NEUROFIBROMIN
• Q6CMT2_KLULA (Q6CMT2) Kluyveromyces lactis strain NRRL Y-1140
• IRA2_YEAST Inhibitory regulator protein IRA2
• Q6FJ13_CANGA (Q6FJ13) Candida glabrata strain CBS138 chromosome M
• Q757I8_ASHGO (Q757I8) AER025Cp
• Q3TYD2_MOUSE (Q3TYD2) Visual corte- cDNA, RIKEN full-length enriched
• Q59F-3_HUMAN (Q59F-3) Neurofibromin variant (Fragment)
• Q5SYH9_MOUSE (Q5SYH9) Neurofibromatosis 1 (Fragment)
• Q8CCE8_MOUSE (Q8CCE8) 15 days embryo male testis cDNA, RIKEN
• Q4HTV9_GIBZE (Q4HTV9) Hypothetical protein
• Q14931_HUMAN (Q14931) NF1 N-isoform-e-on11
• Q4R3N5_MACFA (Q4R3N5) Testis cDNA clone: QtsA-15713, similar to human
• Q7RWZ8_NEUCR (Q7RWZ8) Hypothetical protein
• Q8BQG3_MOUSE (Q8BQG3) 9 days embryo whole body cDNA, RIKEN
• Q95U43_DROME (Q95U43) GH08833p
52
• Q5C2A5_SCHJA (Q5C2A5) SJCHGC09051 protein (Fragment)
• Q62595_RATLE (Q62595) Neurofibromatosis protein type 1 (Fragment)
• Q62596_RATLE (Q62596) Neurofibromatosis protein type 1 (Fragment)
• Q8UVE4_9FALC (Q8UVE4) Neurofibromatosis type 1 (Fragment)
• NF1_CHICK Neurofibromin (Neurofibromatosis-related protein
• Q55CR5_DICDI (Q55CR5) Hypothetical protein
• Q62597_RATLE (Q62597) Neurofibromatosis protein type 1 (Fragment)
• Q8SSU4_DICDI (Q8SSU4) Similar to Dictyostelium discoideum (Slime
• Q7Z3J5_HUMAN (Q7Z3J5) Hypothetical protein DKFZp686J1293 (Fragment)
• Q5C284_SCHJA (Q5C284) SJCHGC08175 protein (Fragment)
• P79186_9PRIM (P79186) Neurofibromin (Fragment)
• P79796_HYLCO (P79796) Neurofibromin (Fragment)
SNPs3D required “NF1” to be entered as a “gene ID” for SNP analysis, individual
mutations were then entered in the “Your SNP” box for analysis, and data was returned
regarding the sequences chosen for use in the prediction for that specific mutation site. The
number and species of the chosen sequences varied between mutation sites, ranging from 5 to 8,
and links were provided to each sequence’s NCBI entry.
Finally, MAPP requires a strong ortholog set for best results, and I tested several different
sets (mammals only: Homo Sapien, Bos taurus (cow), Canis familiaris (dog), Macaca mulatta
(rhesus macaque), Monodelphis domestica (opossum), Pan troglodytes (chimp), Rattus
norvegicus (rat), Mus musculus (mouse); mammals, amphibians, and birds: mammalian
sequences, Gallus gallus (chicken), Xenopus tropicalis (frog); mammals, amphibians, birds, and
fish: previously listed sequences, Takifugu rubripes (puffer fish), Gasterosteus aculeatus (3
spine stickleback), Tetraodon nigroviridis; mammals, amphibians, birds, fish, and insects:
Previously listed sequences, Anopheles gambiae (mosquito), Apis mellifera (honeybee),
Drosophila melanogaster (fruit fly)). Inparanoid was used to determine the best sequences to
use, and these were arranged into several different multiple sequences alignments (MSAs) by
Clustal. In addition, a phylogenetic tree is required for MAPP analysis. Using an MSA from
Clustal, SEMPHY was employed to produce a phylogenetic tree, and both the MSA and the
53
corresponding tree were used in MAPP analysis. No specific mutations are analyzed by MAPP,
but rather the results for all possible substitutions at all sites are returned. Differences in
terminology used in reporting the results reflect those used by each program.
Splice Analysis
To analyze the possibility that mutations in our data sets could produce abnormal splicing
as their pathogenic mechanism rather than an amino acid substitution, the sequence of the entire
exon in which the mutation occurred, as well as the last 10 nucleotides of the 5’ and first 10
nucleotides of the 3’ intron were used for analysis by the NNSPLICE algorithm. Prediction
parameters were: Organism: Human or other, search for both 5’ and 3’ splice sites, no reverse
strand included, and a minimum score of 0.4 for both site types (out of a maximum 1.0).
Results
Table 4-1 shows the summary of the results, while Table C-1 (Appendix C) gives detailed
results for each individual mutation analyzed. In both tables, “tolerated” and “no significant
impairment (NSI)” are the terms used to infer a non-pathogenic (neutral) prediction by the
respective programs. “Pathogenic” (SIFT, SNPs3D) and “deleterious” (MAPP) are used for
those missense mutations predicted to substantially alter protein structure and thus likely to be
associated with disease. For SNPs3D and MAPP, “Failed” indicates inability of the program to
make a prediction because there was insufficient data about that residue in other species. SIFT
also returned some predictions as pathogenic with “low confidence” (“low conf” in Table 4-1
and “(!)” in Table C-1). These were residues with prediction scores very close to the
tolerated/pathogenic threshold.
The first two data sets, germline mutations with known effect and mutations from
site-directed mutagenesis studies (control columns 1-5 of Table 4-1), were used to gauge the
accuracy of the various programs in predicting the functional effects of missense mutations on
54
neurofibromin. Each of these two data sets can be further broken into two subsets, neutral
mutations (present in an unaffected relative, or did not alter RAS-GAP activity) and pathogenic
mutations (proven de novo mutations, or found in two or more independent patients but not
controls, or shown to have affected RAS-GAP activity by in vitro methods). The “Unknown”
mutations set included mutations identified in our lab or reported in the literature, lacking
conclusive data about pathogenicity beyond initial identification in an NF1 patient.
For the eight mutations known to be neutral (control column 2, Table 4-1), accuracy varied
by program, as well as by the sequences used in making the predictions. When using the 8
mammalian ortholog sequences, SIFT correctly predicted only 1/8 (12.5%) of the known neutral
germline mutations. When SIFT was allowed to compile a set of sequences from the available
databases (43 total), the number of correct neutral germline predictions increased to 50%. The
results were similar for MAPP: 1/8 (12.5%) of the known neutral germline mutations were
correctly predicted when the 8 mammalian ortholog sequences were used. However, adding the
additional sequences to the ortholog set (amphibian, chicken, fish, and insects) only increased
MAPP’s correct neutral predictions to 25%. The SNPs3D program itself chooses which
sequences to consider at each mutation site, and correctly predicted 4/8 (50%) of the known
neutral germline mutations. SNPs3D analysis of the remaining four known neutral mutations
failed due to inadequate data at those sites.
For the 18 known pathogenic germline mutations (control column 1, Table 4-1), the
sequences included in the analysis also affected the accuracy of the predictions. When only
using the 8 mammalian ortholog set, SIFT predicted all 18 mutations to be pathogenic; however,
the program reported low confidence in all these predictions. When using the 43 sequences,
SIFT accurately predicted 13/18 known pathogenic germline mutations (72.2%), with only two
55
given a rating of low confidence. The differences seen between the various MAPP analyses are
not as great as those seen for the known neutral germline mutations. When MAPP used the 8
mammalian ortholog sequences, 17/18 mutations were predicted correctly (94.0%), and this only
decreased to 77.8% (14/18) when the entire ortholog set was used. SNPs3D accurately predicted
12/18 (66.7%) of the known pathogenic germline mutations.
For the remaining two data subsets (site-directed mutagenesis data, control columns 4-5,
Table 4-1), the number of ortholog sequences used did not affect the accuracy of the predictions
by SIFT and MAPP. SNPs3D called all of these mutations “pathogenic”. In the case of the 7
mutations found to be neutral in the site-directed mutagenesis RAS-GAP studies, all three
programs failed to predict any of them correctly. In contrast, all three programs correctly
predicted a pathogenic effect for the 5 mutations found to alter RAS-GAP activity in the
site-directed mutagenesis studies (100% accuracy)(control column 4, Table 4-1). To the best of
my knowledge, these mutations have not been reported in NF1 patients yet, and are less likely to
occur since some have 2 or more bases altered in the codons.
Overall, among the 46 control mutations, SIFT (8 mammals) accurately predicted the
effects of 69.5% of the mutations, and SIFT (43 sequences) was accurate for 54.3% of the
mutations. While SNPs3D accurately predicted the effects of 58.7% of the control mutations,
analysis did not return a result (failed) for 9 of the mutations due to insufficient sequence data at
those sites. Because in these cases the program needs more information to ensure an accurate
prediction, it is slightly misleading to include these samples in the “inaccurate” category. If only
taking into account those control mutations for which a prediction was returned (either correct or
incorrect) the accuracy of SNPs3D is 73% (27/37). The accuracy of MAPP varied slightly based
on the ortholog sequences used in analysis. Using the MAPP predictions made when the
56
mammalian, amphibian, chicken, and fish neurofibromin ortholog sequences were used in
analysis (the most accurate), MAPP had a 60.9% accuracy rate (28/46) in predicting the known
effects of the control mutations.
All three programs (all predictions under all conditions, not including failures) were in
agreement, correctly calling 15 germline pathogenic control mutations (out of 18). However,
only one known germline neutral mutation was correctly called neutral by all three programs.
Among the 39 “Unknown” missense mutations, SNPs3D predicted that 16 would be
pathogenic, SIFT (43 sequences) predicted 31, SIFT (8 sequences) predicted 37, MAPP (8
sequences) predicted 36, and the remaining three MAPP analyses predicted 37 or fewer to be
pathogenic (inversely related to number of homolog sequences). In total, 14 different
“Unknown” mutations were predicted to be neutral by one or more programs. Of those 14, 6
were predicted to be neutral in only one analysis, and 8 were predicted to be neutral in two or
more of the analyses.
Seven of the mutations were already known to create cryptic splice sites. This is a pitfall
for predicting missense pathogenesis, since a “neutral” result could be inaccurate if the point
change actually induced cryptic splicing. NNSPLICE predicted 2 of the 7 reported cryptic splice
sites (29% accuracy). As missense mutations, SIFT predicted 2 of these as pathogenic, 5 as
tolerated. SNPs3D predicted 2 to be pathogenic, 1 to be tolerated, and the analysis of the
remaining 4 failed due to insufficient sequence data. The most accurate MAPP run (mammal,
amphibian, chicken, and fish orthologs) predicted 4 to be deleterious, 2 to have no significant
impairment, and the analysis of 1 failed. Thus, the 3 computational methods would have missed
the true pathogenicity of these mutations half or more of the time. Interestingly, NNSPLICE
found two mutations in the “Unknown” mutations set that it predicted to create cryptic splice
57
sites, one a splice acceptor (score of 0.95) and one a splice donor (score of 0.56). Both of these
mutations happened to be predicted to be pathogenic/deleterious by all three programs.
To express the accuracy of these tests in terms of specificity (probability that a test calls a
mutation neutral when it really is neutral) and sensitivity (probability that a test calls a mutation
pathogenic when it really is pathogenic), data from control columns 1 and 2 (Table 4-1) were
used (real NF1 patient data, no splice errors). For SNPs3D, sensitivity was 92.3% (12/13),
specificity was 100% (4/4) and failure rate was 34.6% (9/26). For SIFT (8 mammals), sensitivity
was 100% (18/18), and specificity was 12.5% (1/8). For SIFT (43 orthologs), sensitivity was
72.2% (13/18), and specificity was 50% (4/8). For MAPP (8 mammals), sensitivity was 94.4%
(17/18), and specificity was 12.5% (1/8). For MAPP (mammals + amphibians + birds),
sensitivity was 94.4% (17/18), and specificity was 12.5% (1/8). For MAPP (mammals +
amphibians + birds + fish), sensitivity was 94.1% (16/17), and specificity was 25% (2/8), with
one failed analysis. For MAPP (mammals + amphibians + birds + fish + insects), sensitivity was
82.4% (14/17), and specificity was 25% (2/8), with one failed analysis.
Discussion
I compared the ability of three freely-available programs with various parameters to predict
NF1 missense mutation pathogenicity, and each had pros and cons in ease-of-use and in
accuracy. MAPP requires the most preparation and the ability to use a Java program. The user
must assemble multiple sequence alignments and generate phylogenetic trees from these MSAs
prior to using MAPP to make predictions. The program returns results regarding substitution of
all 22 amino acids for every amino acid position in your sequence (provided there is enough
information in the MSA) which prevents the need to re-run analysis for future mutations of
interest. However, if some of the orthologs used in the MSA have different numbers of encoded
amino acids (or insertions/deletions relative to other orthologs), finding the amino acid of interest
58
can be difficult due to altered numbering systems. SIFT can also provide a similar analysis for
later use, and this output is simpler to search through then the MAPP results. SIFT also allows
for several different options in the amount of preparation needed before analysis. The program
will except a gene ID number or a protein sequence and produce a MSA, or align a group of
sequences already collected. Finally, an MSA can also be submitted for analysis. If reusing an
MSA previously produced by SIFT, every "-" used as a place holder in partial sequences must be
replaced with "X" for analysis to proceed correctly. The length of time an analysis takes varies
based on amount of input given, with analysis of an MSA taking the least amount of time. In all
cases, a list of mutations for analysis is also entered, and all are analyzed simultaneously. If
more then one mutation occurs at the same amino acid, they must be analyzed separately, as the
program only returns specific results (number of sequences considered, scores, etc.) for one. In
the case of SNPs3D, all mutations are analyzed individually. Once the correct protein sequence
is found, individual mutations are entered one at a time and the program selects the sequences to
use in analysis. These will vary by position somewhat, but the user cannot choose which to
include or exclude. Though analysis of large numbers of mutations can be time consuming since
each mutation is entered separately, the SNPs3D results are quickly returned since the program
only aligned a small number of sequences (5-8) over a short sequence surrounding the site of
your mutation. The results are also well explained through links, and easy to interpret. These, in
addition to accuracy data (discussed below), are important considerations for future users.
The accuracy of these programs is in part based on the gene sequences used in making the
predictions. As can be seen in the results from the multiple MAPP analyses, the inclusion of
more diverse sequences (in this case amphibians, birds, and fish) can improve the reliability of
predictions (additional known neutral mutation predicted correctly compared to mammals-only
59
results). This is also the case when comparing the results for the control sets returned by SIFT
when 8 mammalian sequences are used for analysis (where nearly all results were returned with
“low confidence”) versus the 43 sequences gathered by SIFT from the SwissPROT/TrEMBLE
database (very few “low confidence”). The program that used the fewest sequences (5-8) was
SNPs3D. Interestingly, for those mutations where SNPs3D was able to return a result, it appears
to be the most accurate. However, a major weakness was that SNPs3D analysis failed to return a
prediction for 34.6% of the controls (50% of the known neutral mutations, 27.7% of known
pathogenic mutations), preventing the accuracy of the program from being fully established.
While including sequences from more divergent species provides insight into what
mutations and amino acid substitutions are tolerated, this increased diversity may represent
divergence of function rather than tolerance for mutations. If that is the case, some mutations
may erroneously be predicted to be tolerated. In contrast to this idea, however, all three
programs showed error in favor of pathogenicity in the NF1 analysis. Across all analyses of the
control sets, (not including splicing errors), 68.9% of the predictions made for a known neutral
mutation (germline and those from site-directed mutagenesis) were “pathogenic/deleterious”,
while only 13.1% of the predictions made for a known pathogenic mutation (germline and those
from site-directed mutagenesis) were “neutral”. In the control sets (not including splice errors),
known neutral mutations were incorrectly predicted to be pathogenic 4 times as often as known
pathogenic mutations were predicted to be neutral. This suggests that the diversity included in
my sequences may not represent divergence of function. It also indicates that these programs
have an inherently higher false-positive rate than false-negative rate (higher sensitivity, lower
specificity)
60
The poor performance of these programs in correctly predicting the effects of the known
neutral mutations could partially be a result of how a mutation is defined as neutral. It has been
estimated that the majority of missense mutations in the human genome are slightly deleterious
(Kryukov et al, 2007). It is possible that, while my known neutral mutations do not alter the
protein structure/function sufficiently to lead to a clinical-diagnosed NF1 phenotypes, the
biochemical effect is enough to be picked up by these prediction programs. Such mutations
could be considered NF1 hypomorphs, not completely neutral but not absolutely pathogenic.
Such mutations have not yet been proven in NF1, but it is theoretically possible (e.g. a mutation
alters RAS-GAP activity but isn’t seen in NF1 patients, or is found in individuals who only meet
one NF1 diagnostic criteria). Hypomorphic alleles would also confound the type of analysis
done here, and the interpretation of the outcome.
The inaccuracy of these programs in predicting neutral NF1 mutations does call into
question the number of false-pathogenic predictions that might be contained in the results for my
“unknown” data set. In my analyses of the control sets (across all programs and analyses),
88/262 pathogenic predictions were incorrect (33.6%). This number varies based on the specific
program, but it is possible that there are some mutations in my “unknown” set that may actually
be neutral despite being predicted to be pathogenic. None of the substitutions in this data set are
reported in dbSNP (http://www.ncbi.nlm.nih.gov/SNP/), which decreases the likelihood that my
unknowns are non-disease-related polymorphisms. This is consistent with the expectation that
novel NF1 mutations in NF1 patients are likely to be pathogenic. In the analyses of the control
sets, 66.7% of the neutral predictions made were incorrect. If this is approximately the rate of
incorrect neutral predictions within the “unknown” set as well, it can be estimated that
61
approximately 9/16 of the unknowns predicted to be neutral are actually pathogenic. More
confidence can be placed in the “unknowns” predicted to be neutral by more than one analysis.
My results indicate that a prediction by any of these programs, whether it is a prediction of
pathogenic or neutral, cannot be taken alone in determining the effects of a given missense
mutation. As these programs were not designed to replace functional studies, it is
understandable that they seem to consider mutations pathogenic until proven neutral. SNPs3D
had the highest control accuracy but also the highest failed rate. SIFT predicted pathogenic
mutations at the expense of a high false-positive rate. MAPP called some known pathogenic
mutations “neutral”, but also had false-positives. No single program stood out as superior
overall. All three had high sensitivity (with at least one version of analysis). Only SNPs3D had
high specificity (100% of the germline neutral mutations, the others were ≤50%), but that
analysis also had a 50% failure rate. My SIFT accuracy and sensitivity results are consistent
with SIFT data for other monogenic situations (sensitivity 80-90%), although SIFT had a worse
specificity score in NF1 analysis (compared to 67-74% in other studies)(Mathe et al., 2006; Chan
et al., 2007). As indicated in other studies, it is considered beneficial to run multiple programs,
with greater faith in the results agreed upon by more than one analysis (Mathe et al., 2006; Chan
et al., 2007; Valdmanis et al., 2008). These programs are often updated and so the accuracy may
improve with newer versions. In addition, there are other programs becoming available and
refined (some of which are based on amino acid biochemistry rather than MSA), such as
PolyPhen. When choosing a program (or set of programs) one should consider the nature of the
question to be answered, the sequence data available, failure rate, as well as which type of errors
are more important to avoid. If one is looking to catch all possible pathogenic mutations, for
example to test further in functional studies, the program used will likely differ from one chosen
62
to correctly identify as many possible neutral mutations as possible. We also found that a
missense analysis should be complimented by a splice site analysis (computational or in the lab
with RT-PCR) to attempt to find mutations that cause exon skipping or cryptic splicing rather
than an amino acid substitution. The data reported here will be useful for individuals considering
computational methods for testing pathogenicity of missense mutations, particularly in large
genes such as NF1.
63
Table 4-1. Summary of computational missense prediction results for 6 data sets (5 controls and
1 unknown).
Controls (n=46)
Data set:
Program
Known
(germline)
Pathogenic
(n=18)
Known
(germline)
Neutral
(n=8)
Known
pathogenic
due to
altering
splicing
(n=8)
Altered GAP
function in
Site-directed
Mutagenesis
(n=5)
Neutral in
Site-directed
Mutagenesis
(n=7)
Unknown
(n=39)
SNPs3D 12 pathogenic
1 tolerated
5 failed
4 tolerated
4 failed
2 pathogenic
2 tolerated
4 failed
5 pathogenic 7 pathogenic 16 pathogenic
5 tolerated
18 failed
SIFT (43
sequences)
13 pathogenic
-2 low conf
5 tolerated
4 pathogenic
-1 low conf
4 tolerated
3 pathogenic
-1 low conf
5 tolerated
5 pathogenic 7 pathogenic 31 pathogenic
-16 low conf
8 tolerated
SIFT (8
mammalian
sequences
18 pathogenic
-18 low conf
7 pathogenic
-7 low conf
1 tolerated
8 pathogenic
-8 low conf
5 pathogenic
-5 low conf
7 pathogenic
-7 low conf
37 pathogenic
-37 low conf
2 tolerated
MAPP (8
mammalian
sequences)
17 deleterious
1 NSI
7 deleterious
1 NSI
6 deleterious
2 NSI
5 deleterious 7 deleterious 36 deleterious
3 NSI
MAPP
(mammals,
amphibians,
and birds)
17 deleterious
1 NSI
7 deleterious
1 NSI
5 deleterious
3 NSI
5 deleterious 7 deleterious 37 deleterious
2 NSI
MAPP *
(Mammals,
amphibians,
birds, and
fish)
16 deleterious
1 NSI
1 failed
6 deleterious
2 NSI
5 deleterious
2 NSI
1 failed
5 deleterious 7 deleterious 35 deleterious
4 NSI
MAPP
(Mammals,
amphibians,
birds, fish,
and insects)
14 deleterious
3 NSI
1 failed
6 deleterious
2 NSI
3 deleterious
4 NSI
1 failed
5 deleterious 7 deleterious 33 deleterious
6 NSI
*the most accurate MAPP analysis; NSI= no significant impairment.
64
CHAPTER 5
CONCLUSIONS AND FUTURE DIRECTIONS
More then 20 years after the identification of the gene and protein responsible for
neurofibromatosis 1(NF1), there is still much that is unclear about NF1 mutations mechanisms,
how these mutations affect neurofibromin function, and how this relates to the heterogeneous
phenotypes of NF1. Additional knowledge of the mutation mechanisms occurring both in the
germline and somatic cells, as well as elucidating which mutation mechanisms do not play a
major role in NF1 progression, may contribute to developing targeted therapies and better
diagnosis. In this chapter, I reiterate the major findings of my work and discuss possible future
directions.
Somatic CpG C to T Mutations
From my work described here, C to T transition mutations at CpG dinucleotides are clearly
not a common somatic mutation mechanism in the NF1 gene. While only four such sites were
analyzed, they were chosen based on the fact that they are hotspots for these same mutations
occurring in the germline, are scattered across the gene, and are not known to be involved in
exon skipping. These same sites are also examined in screens for somatic NF1 mutations. It is
however still possible that other NF1 CpG sites are more susceptible to C to T transitions in
somatic cells, and that these mutations could play a role in NF1 tumorigenesis. Compared to
some genes, such as TP53, RB1, and NF2, germline NF1 C to T mutations are somewhat less
common, and somatic C to T mutations are much rarer. The difference in somatic C to T
mutation rates between NF1 (in neurofibromas), TP53 (in bladder and colon cancers), and APC
(in colon cancer) could be due to defects of the base excision repair pathway in more malignant
tumors, leading to a decreased ability of cells to recognize and correctly repair a G:T mispair.
Several reports have shown that neurofibromas, which are not malignant, show no evidence of
65
microsatellite instability (MSI), which is often an indication of mutation in the DNA repair
pathways (Luijten et al., 2000; Upadhyaya et al., 2004; in contrast: Ottini et al., 1995). However,
malignant peripheral nerve sheath tumors (MPNSTs), which can develop from a plexiform
neurofibroma, have been shown to exhibit MSI in 30-45% of cases (Upadhyaya et al., 2004;
Kobayashi et al., 2006). Additionally, Wang et al. (2003) found that half of 10 non-NF1 human
cancer cells lines containing MSI had point mutations in NF1, whereas no mutations in NF1
were found in non-NF1 human cancer cell lines with functional mismatch repair pathways.
These data suggest that MPNSTs, with a higher occurrence of MSI (and likely repair pathway
mutations), may be more susceptible to CpG C to T transitions in NF1 than neurofibromas. This
could be determined by a similar mutation screen as the one I performed, with a large cohort of
MPNSTs.
If future studies were able to identify C to T transitions at CpG sites in neurofibromas, it
would be of interest to see if this is a common mutation mechanism within an individual.
Multiple neurofibromas from the same patient could help answer this question. If individuals
were found to be generally susceptible to such mutations, (e.g. less-than-adequate DNA excision
repair mechanisms, or exposed to mutagens), perhaps they could benefit from therapies being
designed to compensate for mutations causing premature stop codons, which are often caused by
C to T transitions at CpG sites. An interesting report described two siblings with some features
of NF1 (café-au-lait spots and skin-fold freckling) but also non-NF1-related cancer at an early
age; these children were found to be homozygous for a MSH2 mutation, suggesting that
pigmentary NF1 features may be mimicked by deficiency in mismatch repair (Toledano et al.,
2008). Full characterization of the NF1 somatic mutation repertoire may indicate whether
66
certain types of mutations besides C to T transitions predominate, and whether these are related
to mutagen exposure.
Of interest, we do not yet know when somatic mutations occur in the Schwann cells that
initiate neurofibromas. The peripheral nervous system growth (and Schwann cell replication) is
predominantly done by adulthood, with Schwann cells going quiescent. These only become
active and enter mitosis when there is a nerve injury. It is possible that two-hit Schwann cells
could lie dormant until an environmental trigger such as injury or endocrine change starts clonal
cell expansion. There is no evidence for outright mutation at other genes in neurofibromas.
Conversely, perhaps second hits occur randomly throughout life and are followed shortly by
clonal expansion. Dermal neurofibromas rarely appear before adolescence, but plexiform tumors
can appear anytime in life (including infancy). So these two tumor types may have some basic
differences relative to somatic mutation occurrence that we do not yet understand. However, I
saw no CpG mutations in large sets of both tumor types, so the occurrence of this mutation type
does not appear different. CpG C to T transitions can occur in the germline (paternal), or any
time after the egg is fertilized throughout life. Knowing the timing of the second hit could also
useful for future therapies/preventions.
Alternative Splicing of exon23a
While it appears that the alternative splicing of exon23a is developmentally significant, the
role that inclusion of exon23a might play in tumorigenesis is still unclear. Analysis of RNA
transcripts from multiple cell types provided an estimate of the level of inclusion of exon23a
occurring in these cells (the relative ratio of Type I to Type II transcript). Leukocytes produce
predominantly Type I transcript, with a slight trend towards increased levels of Type II
compared to Type I mRNA in leukocytes from NF1 patients versus non-NF1 patients, although
the numbers were low. There were no strong differences seen in the relative ratios of Type I to
67
Type II transcript seen in dermal versus plexiform neurofibroma samples, or between primary
tumor samples and their corresponding cultures. There also does not appear to be a difference in
the ratio of Type I to Type II mRNA when samples are analyzed by gender. However, there was
nearly twice the number of female individuals in this study as males. It would be of interest to
analyze samples from more male patients to rule out gender-specific trends in the alternative
splicing of exon23a. Additionally, I found several cases where two tumors from the same
individual had very different Type I to Type II ratios, indicating that these ratios are tumor-, not
individual-specific. Development of a real-time PCR protocol to distinguish levels of Type I and
Type II mRNA, as has been used to quantify skipping of exon37 in NF1 mRNA (Vandenbroucke
et al., 2001), would be useful in quantifying more subtle differences in the ratios of Type I to
Type II mRNA between primary tumors and their corresponding cultures, between tumor types,
or between genders. However, this would require a large sample set since the ratios would fall
into a greater number of bins.
Despite some differences in the amount of alternative splicing of exon23a between tumors,
there is a trend toward increased levels of exon23a inclusion in NF1 tumors compared to normal
Schwann cells, making this mechanism a potential target for therapy. Before such therapies can
be developed it must be shown that having (virtually) only Type II transcript can substitute as a
somatic mutation. To help determine if the reduced GAP activity of Type II neurofibromin is
“pathogenic” one of two experiments could prove informative. A construct containing the Type
II GAP domain driven by a low-level promoter could be introduced to an NF1 null cell with an
abnormal phenotype (e.g. increased invasiveness on Matrigel, increased passage number prior to
senescence, growth-factor independent) to see if the Type II GAP has sufficient activity to rescue
the phenotype. Alternatively, Schwann cells hemizygous for NF1 (e.g. from a patient with a
68
germline large deletion) could be forced, through recently developed splicing therapies, to
express only Type II transcript to determine if this causes any changes in phenotype
(development of tumorigenic characteristics). Additionally, both the trans- and cis-acting
elements involved in controlling the splicing of exon23a must be identified. These elements
would be the targets of therapies designed to decrease the amount of exon23a inclusion, most
likely in Schwann cells if targeted correctly. Zhu et al. (2008) have identified some of these
factors that appear to regulate the inclusion of exon23a in neurons. Hu proteins are mRNA
binding proteins that are proposed to bind to AU-rich regions on either side of exon23a and
inhibit inclusion of the exon. As this interaction is in neurons, further studies need to be carried
out to identify splicing elements in Schwann cells. Interestingly, mice lacking exon23a have a
high rate of cognitive deficits (e.g. water maze memory test) where as exon31 knockout mice (an
out of frame deletions) do not show these deficits (Silva et al., 2001). Perhaps increased
inclusion of exon23a in more NF1 transcripts could be used to address this problem, which
affects approximately half of children with NF1. Such therapies are currently being developed to
alter the alternative splicing of exon10 of tau, a gene indicated in several progressive dementia
disorders (reviewed by Zhou et al., 2008). Interestingly, alternative splicing of exon10 in tau is
developmentally regulated in the central nervous system, as is exon23 alternative splicing in
NF1, although the pattern is not the same (Gao et al., 2000).
RNA editing at C3916 did not appear to be associated with loss of the Type I transcript in
tumors in this study, in contrast to published results. I saw no evidence of any RNA editing in
our samples, including MPNSTs. A larger study, repeating the analysis of the published samples
as well as new samples, may shed light on this inconsistency. It may also be of interest to
determine which of these changes is occurring first, in the tumors reported to exhibit both loss of
69
the Type I transcript and RNA editing. My study used a qualification of complete or nearly
complete loss of the Type I transcript to identify samples to be evaluated for RNA editing, and
since none was found, it appears that this shift in alternative splicing alone does not lead to
increased RNA editing.
Computational Analysis Comparison
Recent advances in bioinformatics, as well as the vast amount of data being generated by
high-throughput techniques has greatly improved the methods available to predict the effects of
missense mutations on a protein of interest. In the study of NF1 and neurofibromin, this is
especially important as there is no way to study these mutations functionally. Using the limited
set of NF1 missense mutations with known effect to test the accuracy of these new prediction
programs, it appears that each may have its advantages in a given situation. It would be
advantageous to use multiple programs when possible. These programs are already being used
to supplement molecular diagnostic data, and so it is important to determine the best tools for the
job. This is a rapidly growing and evolving field, and it will be key to continue to increase the
control data set as more NF1 mutations with known effect are identified, to test the abilities of
these programs and ensure that best sequence data is being used. A clinical diagnosis may hinge
on data provided by programs such as these.
70
APPENDIX A
CPG C TO T MUTATION ANALYSIS DATA
Table A-1. Results of CpG C to T mutation screen using TaqαI restriction enzyme digest.
Mutation at CPG site in:
Sample ID Exon 10a Exon 22 Exon 23.2 Exon 41
Dermal tumors
UF80 T1 N N
UF80 T2 N N
UF80 T4 N N
UF80 T6 N N N N
UF80 T8 N N
UF80 T11 N N
UF80 T12 N N
UF80 T12c N
UF80 T23 N N N N
UF80 T32 N N N N
UF113 T3 N N N
UF113 T4 N N N
UF113 T5 N N N N
UF113 T6 N N N
UF113 T7 N N N N
UF113 T9 N N N N
UF113 T10 N N N
UF113 T11 N N N
UF113 T12 N
UF113 T13 N
UF113 T14 N N
UF113 T15 N N
UF113 T16 N N N
UF113 T17 N N
UF113 T18 N N
UF113 T19 N N
UF113 T20 N N
UF113 T21 N N
UF113 T22 N N N
UF113 T23 N N
UF113 T24 N N
UF233 T1 N N N N
UF233 T2 N N N N
UF287 T1 N N N N
UF287 T2 N N N N
UF327 T1 N N N N
N= no, Y=yes, C=constitutional
71
Tabel A-1. Continued
Mutation at CPG site in:
Sample ID Exon 10a Exon 22 Exon 23.2 Exon 41
Dermal tumors
UF328 T4 N N N N
UF328 T6 N N N N
UF328 T7 N N N N
UF328 T14 N N N N
UF417A N N N N
UF417D N N N N
UF431 T2 N N N N
UF474 T3 N N N N
UF486 T N N N N
UF486 T2 N N N N
UF505 T3 N N N N
UF505 T4 N N N N
UF509 T2 N N N N
UF510 T1 N N N N
UF512 T3 N N N N
UF532 T1 N N N N
UF552 T3 N N N N
UF705 T1 N N N N
UF743 T1 N N N N
UF831 T1 N N N N
UF831 T2 N N N N
UF831 T2c N N N N
UF835 T1 N N N N
UF1150 T N N N N
UF1345 T1 N N N N
UF1346 T1 N N N N
AW T2 N N N N
Plexiform tumors
UF158 T3 N N N N
UF158 T4 N N N N
UF181 T1 N N N N
UF303 T N N N N
UF310 T N N N N
UF327 T2 N N N N
UF340 T1 N N N N
UF344 T2 N N N N
UF346 T1 N N N N
UF356 T1 N N N N
UF362 T N N N N
N= no, Y=yes, C=constitutional
72
Table A-1. Continued
Mutation at CPG site in:
Sample ID Exon 10a Exon 22 Exon 23.2 Exon 41
Plexiform tumors
UF375 T1 N N N N
UF375 T2 N N N N
UF378 T1 N N N N
UF378 T2a N N N N
UF378 T2b N N N N
UF386 T1 N N N N
UF387 T2 N N N N
UF389 T1 N N N N
UF420 T1 Y/C N N N
UF428 T1 N N N N
UF429 T N N N N
UF440 T1 N N N ?
UF440 Tc N N N N
UF450 T1 N N N N
UF452 T1 N N Y/C N
UF454 T1 N N N N
UF454 T2 N N N N
UF454 T3 N N N N
UF454 T4 N N N N
UF454 T5 N N N N
UF454 T6 N N N N
UF456 T1 N N N N
UF456 T3 N N N N
UF468 T N N
UF469 T1 N N N N
UF475 T1 N N N N
UF495 T N N N N
UF499 T N N N N
UF504 T N N N N
UF511 T1 N N N N
UF511 T2 N N N N
UF526 T1 N N N N
UF526 T2 N N N N
UF537 T1 N N N N
UF549 T1 N N N N
UF550 T1 N N N N
UF554 T1 N N N
UF554 Tc N N N N
UF555 T N N N N
UF562 T N N N N
N= no, Y=yes, C=constitutional
73
Table A-1. Continued
Mutation at CPG site in:
Sample ID Exon 10a Exon 22 Exon 23.2 Exon 41
Plexiform tumors
UF572 T1 N N N N
UF572 Tc N N N
UF573 T1 N N N N
UF573 T2 N N N N
UF573 T3 N N N N
UF573 T4 N
UF593 T1 N N N N
UF609 T N N N N
UF622 T N N N N
UF632 T1 N N N N
UF746 T N N N N
UF787 T N N N N
UF836 T N N N N
UF836 Tc N N N N
UF860 T N N N N
UF1072 T2 N N N N
UF1093 T N N N N
UF1151 T N N N N
UF1160 T N N N N
UF1169 T N N N N
UF1207 T N N N N
UF1243 T1 N N N N
UF1243 Tc N N N N
UF1258 T N N N N
UF1296 T N N N N
UF1296 Tc N N N N
UF1308 T N N N N
UF1308 Tc N N N N
UF1371 Tdr ? N N N
UF1371 Tsw N N N N
UF1371 Tc dr N N N N
UF1371 Tc sw N N N N
N= no, Y=yes, C=constitutional
74
APPENDIX B
EXON 23a ALTERNATIVE SPLICING DATA
Table B-1. Relative concentrations of Type I v Type II mRNA in blood, tumor and culture
samples.
Cell type Sample # No Type I Type I < II Type I ≈ II TypeI > II
Normal pm 97.3 X
Schwann pm 97.4 X+
cells pm 02.3 X
Dermal UF80T2 X
tumors UF80T4 X
UF80T5 X
UF80T6 X
UF80T18 X+
UF80T31 X
UF80T32 X
UF328T4 X
UF328T6 X
UF328T11 X
UF389T1 X
UF470T1 X
UF505T4 X
UF505T7 X*
UF526T1 X*
UF526T2 X+
UF1312T1 X+
UF1312T2 X
UF1313T1 X+
UF1313T1 sc-/- X+
UF1313T2 X
Dermal UF328T8c X
cultures UF470T1c X
UF470T2c X+
UF1313T2c X+
Immortal PNF95.11b P23 X
cell lines PNF95.11b P24 X
+: Much more Type II; *: Barely detectable Type I in comparison to Type II
75
Table B-1. Continued
Cell type Sample # No Type I Type I < II Type I ≈ II TypeI > II
Plexiform PNF95.11b2 X
tumors PNF95.11a X
PNF95.6 X*
UF428T1 X+
UF429T(GD) X
UF429T X
UF450T1 X
UF450T1(P) X
UF532T1 X*
UF548T4 X
UF548T5 X
UF548T6 X
UF609T X
UF622T1 X
UF746T1A X
UF746T1B X
UF746T1C X
UF836TA X+
UF836TB X
UF860T X
UF1151T X
UF1160T X
UF1201T1 X+
UF1258T X
UF1371 DR X
Plexiform UF440Tc X*
cultures UF469Tc X
UF554T1c X*
UF609Tc X*
UF746TcA X
UF746TcB X
UF1243Tc X
UF1258Tc X+
UF1371 DR -/- X
+: Much more Type II; *: Barely detectable Type I in comparison to Type II
76
Table B-1. Continued
Cell type Sample # No Type I Type I < II Type I ≈ II TypeI > II
NF blood UF80 X
UF389 X
UF450 X
UF470 X
UF746 X
UF836 X
UF1160 X
mPNSTs SNF 02.2 X+
SNF94.3 X
SNF 96.2 X+
UF158T X
UF344T1 X+
UF459T1 X
UF860T X
Control UF86 X
blood UF91 X
UF563 X
UF733 X
T80G GG X
PW Fresh X
RET Fresh X
fibroblasts UF1104 X+
Timed UF328fresh X
blood UF328 1 day X
UF328 3 day X
+: Much more Type II; *: Barely detectable Type I in comparison to Type II
77
APPENDIX C
MISSENSE MUTATION COMPUTATIONAL ANALYSIS DATA
78
Table C-1. Results for each mutation as returned by the various computational methods used.
Mutation
and
location
Codon
change Reference Type SNPs3D SIFT
SIFT
8
mammal
MAPP
mammals
MAPP
mam/am/
ck
MAPP
mam/am/
ck/fsh
MAPP
mam/am/
ck/fs/ins
Splice
Site
L549P
exon11 CTG-CCG
Fahsold et al
2000; Wallace
lab,
unpublished P failed Tol Patho (!) Del Del Del Del no
K505E
exon10b AAG-GAG Park et al 1998 P failed Tol Patho (!) Del Del Del NSI no
L508P
exon10b CTT-CCT
Wallace lab,
unpublished P failed Patho (!) Patho (!) Del Del Del Del no
S665F
exon12b TCC-TTC
Mattocks et al
2004; Fahsold
et al 2000 P failed Tol Patho (!) Del Del N/A N/A no
T780K
exon15 ACA-AAA
Han et al 2001;
Fahsold et al
2000 P Patho Patho Patho (!) Del Del Del Del no
W784R
exon15 TGG-CGG
Upadhyaya et
al 2008 P Patho Patho Patho (!) Del Del Del Del no
L844F
exon16 CTT-TTT
Mattocks et al
2004; Girodon
& Bouduret
2000 P Patho Patho Patho (!) Del Del Del Del no
L847P
exon16 CTT-CCT
Fahsold et al
2000; Wallace
lab,
unpublished P Patho Patho Patho (!) Del Del Del Del no
L898P
exon15 CTG-CCG
Fahsold et al.
2000; Maynard
et al 1997 P Patho Patho Patho (!) Del Del Del Del no
79
Table C-1. Continued
Mutation
and
location
Codon
change Reference Type SNPs3D SIFT
SIFT
8
mammal
MAPP
mammals
MAPP
mam/am/
ck
MAPP
mam/am/
ck/fsh
MAPP
mam/am/
ck/fs/ins
Splice
Site
M968R
exon17 ATG-AGG
DeLuca et al
2003 P Patho Patho Patho (!) Del Del Del Del no
R1204W
exon21 CGG-TGG
Pros et al 2008;
Ars et al 2000 P Patho Patho Patho (!) Del Del Del Del no
R1276Q
exon22 CGA-CAA
Jeong et al
2006; Fahsold
et al 2000 P Tol Patho Patho (!) Del Del Del Del no
R1391S
exon24 AGA-AGT
Gutmann et al
1993b D/M Patho Patho Patho (!) Del Del Del Del no
K1419Q
exon24 AAG-CAG
Mattocks et al
2004;
Upadhyaha et
al 1997 P Patho Patho Patho (!) Del Del Del NSI no
K1423E
exon24 AAG-GAG
Li et al 1992;
Xu and
Gutmann 1997 D/M Patho Patho Patho (!) Del Del Del Del no
K1423Q
exon24 AAG-CAG
Li et al 1992;
Xu and
Gutmann 1997 D/M Patho Patho Patho (!) Del Del Del Del no
K1423S
exon24 AAG-TCG
Gutmann et al
1993b D/M Patho Patho Patho (!) Del Del Del Del no
L1425P
exon25 CTT-CCT
Peters et al
1999 P Patho Patho Patho (!) Del Del Del Del no
Q1426R
exon25 CAG-CGT
Gutmann et al
1993b D Patho Patho Patho (!) Del Del Del Del no
N1430R
exon25
AAT-AGA
or AGG
Wallace lab,
unpublished P Patho Patho Patho (!) Del Del Del Del no
S1468G
exon26 AGT-GGT
Mattocks et al
2004;
Upadhyaha et
al 1997 P Patho Tol Patho (!) NSI NSI NSI NSI no
80
Table C-1. Continued
Mutation
and
location
Codon
change Reference Type SNPs3D SIFT
SIFT
8
mammal
MAPP
mammals
MAPP
mam/am/
ck
MAPP
mam/am/
ck/fsh
MAPP
mam/am/
ck/fs/ins
Splice
Site
G1498E
exon26 GGG-GAG
Pros et al
2008; Ars et
al. 2003 P Patho Tol Patho (!) Del Del Del Del no
L2317P
exon38 CTT-CCT
Wu et al 1999;
Wallace lab,
unpublished P failed Patho Patho (!) Del Del Del Del no
D176E
exon4b GAT-GAA
Wallace lab,
unpublished N Tol Tol Patho (!) Del Del NSI Del no
R765H
exon14 CGC-CAC
Wallace lab,
unpublished N failed Tol Patho (!) Del Del Del Del no
S858C
exon16 TCT-TGT
Wallace lab,
unpublished N Tol Tol Tol NSI NSI NSI NSI no
R873C
exon16 CGT-TGT Mattocks 2004 N Tol Patho (!) Patho (!) Del Del Del Del no
N1229S
exon21 AAT-AGT Ars et al 2003 N Tol Tol Patho (!) Del Del Del NSI no
E1264Y
exon22 GAA-TAC
Gutmann et al
1993b N Patho Patho Patho (!) Del Del Del Del no
A1281R
exon22 GCC-CGC
Gutmann et al
1993b N Patho Patho Patho (!) Del Del Del Del no
F1389H
exon24 TTC-CAC
Poullet et al
1994 N Patho Patho Patho (!) Del Del Del Del no
P1395I
exon24 CCT-ATT
Gutmann et al
1993b N Patho Patho Patho (!) Del Del Del Del no
A1396G
exon24 GCC-GGC
Poullet et al
1994 N Patho Patho Patho (!) Del Del Del Del no
P1400R
exon24 CCG-CGG
Gutmann et al
1993b N Patho Patho Patho (!) Del Del Del Del no
N1430M
exon25 AAT-ATG
Gutmann et al
1993b N Patho Patho Patho (!) Del Del Del Del no
81
Table C-1. Continued
Mutation
and
location
Codon
change Reference Type SNPs3D SIFT
SIFT
8
mammal
MAPP
mammals
MAPP
mam/am/
ck
MAPP
mam/am/
ck/fsh
MAPP
mam/am/
ck/fs/ins
Splice
Site
R1809C
exon29 CGC-TGC Ars et al 2003 N failed Patho Patho (!) Del Del Del Del no
R1825W
exon29 CGG-TGG Ars et al 2003 N failed Patho Patho (!) Del Del Del Del no
A2058D
exon33 GCT-GAT Mattocks 2004 N failed Patho Patho (!) Del Del Del Del no
D186V
exon4b GAT-GTT
Zatkova et al.
2004 C.S. Patho Patho (!) Patho (!) Del Del Del Del no
Y489C
exon10b TAT-TGT
Messiaen et al
1999 C.S. Tol Tol Patho (!) Del Del Del NSI 0.97 A
G629R
exon12b GGG-AGG Ars et al 2000 CS failed Tol Patho (!) Del NSI N/A N/A 0.98 A
G922S
exon16 GGT-AGT Ars et al 2000 CS failed Tol Patho (!) NSI NSI NSI NSI no
V1093M
exon19b GTG-ATG Ars et al. 2003 C.S. Tol Tol Patho (!) Del Del Del NSI no
S1479G
exon26 AGT-GGT
Wallace lab,
unpublishes;
Upadhyaya et
al. 1997
(missense) C.S. Patho Tol Patho (!) NSI NSI NSI Del no
R1849Q
border
exon 29
+ 30 CGG-CAG Ars et al. 2000 C.S. failed Patho Patho (!) Del Del Del Del no
K2286N
exon37
AAG-AAC
or AAT
Messiaen et al
2000 CS failed Patho Patho (!) Del Del Del NSI no
82
Table C-1. Continued
Mutation
and
location
Codon
change Reference Type SNPs3D SIFT
SIFT
8
mammal
MAPP
mammals
MAPP
mam/am/
ck
MAPP
mam/am/
ck/fsh
MAPP
mam/am/
ck/fs/ins
Splice
Site
H31R
exon2 CAT-CGT
Mattocks et al
2004 ? failed Tol Patho (!) Del Del Del Del no
S82F
exon3 TCT-TTT
Kluwe et al
2002 ? failed Patho (!) Patho (!) Del Del Del Del no
C93Y
exon3 TGT-TAT
Messiaen et al
2000 ? failed Patho (!) Patho (!) NSI Del Del Del no
L145P
exon4a CTC-CCC
Mattocks et al
2004 ? Patho Patho (!) Patho (!) Del Del Del Del no
C187Y
exon 4b TGT-TAT
Messiaen et al
2000 ? Tol Patho (!) Patho (!) Del Del Del NSI no
L194R
exon4b CTG-CCG
De Luca et al.
2005 ? Patho Patho (!) Patho (!) Del Del Del Del no
C324R
exon7 TGT-CGT
Mattocks et al
2004 ? Patho Patho (!) Patho (!) Del Del Del Del no
E337V
exon7 GAA-GTA
Mattocks et al
2004 ? Tol Patho (!) Patho (!) Del Del Del NSI no
D338G
exon7 GAT-CGT
Upadhyaya
1997 ? Patho Patho (!) Patho (!) Del Del Del Del 0.56 D
R440P
exon10a CGA-TGA
Wallace lab,
unpublished ? Patho Patho (!) Patho (!) Del Del Del Del no
Q519P
exon10c CAA-CCA
Upadhyaya et
al 2004 ? failed Patho (!) Patho (!) Del Del Del Del no
L532P
exon10c CTG-CCG
Mattocks et al
2004 ? Patho Patho (!) Patho (!) Del Del Del Del no
S574R
border
exon 11
& 12a AGC-CGC
Mattocks et al
2004 ? Tol Patho (!) Patho (!) Del Del NSI NSI no
L578P
exon12a CTT-CCT
Jeong et al
2004 ? Patho Patho (!) Patho (!) Del Del Del Del no
83
Table C-1. Continued
Mutation
and
location
Codon
change Reference Type SNPs3D SIFT
SIFT
8
mammal
MAPP
mammals
MAPP
mam/am/
ck
MAPP
mam/am/
ck/fsh
MAPP
mam/am/
ck/fs/ins
Splice
Site
I581S
exon12a ATC-AGC Lee et al. 2006 ? Failed Patho (!) Patho (!) Del Del Del Del 0.95 A
R815M
exon16 AGG-ATG
Wallace lab,
unpublished ? Patho Patho (!) Patho (!) Del Del NSI NSI no
L1015R
exon18 CTG-CGG
Kluwe et al
2003 ? Patho Patho Patho (!) Del Del Del Del no
M1073V
exon19b ATG-GTG
Mattocks et al
2004 ? Tol Tol Patho (!) Del Del Del Del no
G1166D
border
exon 20
& 21 GGC-GAC
Purandare et al
1994 ? failed Tol Patho (!) Del Del Del NSI no
L1196R
exon21 CTT-CGT
Mattocks et al
2004 ? failed Patho Patho (!) Del Del Del Del no
R1204G
exon21 CGG-GGG
Krkljus et al
1998 ? Patho Patho Patho (!) Del Del Del Del no
R1276G
exon22 CGA-GGA
Mattocks et al
2004 ? Patho Patho Patho (!) Del Del Del Del no
R1325G
exon23-1 AGG-GGG Lee et al. 2006 ? Patho Patho Patho (!) Del Del Del Del no
P1400S
exon24 CCG-TCG
Xu and
Gutmann 1997 ?/M Patho Patho Patho (!) Del Del Del Del no
K1419R
exon24 AAG-AGG
Purandare et al
1994 ? Tol Patho Patho (!) NSI NSI NSI NSI no
N1430I
exon25 AAT-ATT Wallace lab,
unpublished ? Patho Patho Patho (!) Del Del Del Del no
N1430T
exon25 AAT-ACT
De Luca et al.
2005 ? Patho Patho Patho (!) Del Del Del Del no
V1432L
exon25 GTT-CTT
De Luca et al.
2005 ? Patho Patho Patho (!) Del Del Del Del no
84
Table C-1. Continued
Mutation
and
location
Codon
change Reference Type SNPs3D SIFT
SIFT
8
mammal
MAPP
mammals
MAPP
mam/am/
ck
MAPP
mam/am/
ck/fsh
MAPP
mam/am/
ck/fs/ins
Splice
Site
R1590W
exon27b CGG-TGG
Upadhyaha et
al 1997 ? failed Patho Patho (!) Del Del Del Del no
V1621R
exon28 GTA-ATA
Jeong et al
2006 ? failed Patho Patho (!) Del Del Del Del no
D1623V
exon28 GAC-GTC
Wallace lab,
unpublished ? failed Patho Patho (!) Del Del Del Del no
A1764S
exon29 GCT-TCT Han et al 2001 ? failed Patho Tol Del Del Del Del no
T1787M
exon29 ACG-ATG Lee et al. 2006 ? failed Tol Tol NSI NSI Del Del no
L1812P
exon29 CTG-CCG
Wallace lab,
unpublished ? failed Tol Patho (!) Del Del Del Del no
C1909R
exon30 TGT-CGT Lee et al. 2006 ? failed Tol Patho (!) Del Del Del Del no
L1932P
exon31 CTG-CCG
Cawthon et al
1990 ? failed Patho Patho (!) Del Del Del Del no
R2129S
exon34 AGA-AGC
Upadhyaya et
al 2004 ? failed Tol Patho (!) Del Del Del Del no
Y2171D
exon34 TAT-GAT
Upadhyaya et
al 1992 ? failed Tol Patho (!) Del Del NSI Del no
S2739Y
exon48 TCT-TAT
Wallace lab,
unpublished ? failed Patho (!) Patho (!) Del Del Del Del no
Under “Type” P= Pathogenic, CS= Cryptic Splice, D/M= Detrimental, affects Microtubule binding, N= Neutral, and ?= Unknown.
For results returned by programs Patho=Pathogenic, Tol= Tolerated, Del= Deleterious, NSI= No Significant Impairment. For
results reported in the SIFT columns, (!)= a warning of low confidence in the prediction was returned by the program. Under
“Splice Site” A= Acceptor, D= Donor.
85
LIST OF REFERENCES
Adams, M., Jones, J.L., Walker, R.A., Pringle, J.H., Bell, S.C., 2002. Changes in tenascin-C
isoform expression in invasive and preinvasive breast disease. Cancer Res. 62, 3289-97.
Ainsworth, C., 2005. Nonsense mutations: running the red light. Nature 438, 726-8.
Andersen, L.B., Ballester, R., Marchuk, D.A., Chang, E., Gutmann, D.H., Saulino, A.M.,
Camonis, J., Wigler, M., Collins, F.S., 1993. A conserved alternative splice in the von
Recklinghausen neurofibromatosis (NF1) gene produces two neurofibromin isoforms, both
of which have GTPase-activating protein activity. Mol. Cell Biol. 13, 487-495.
Ars, E., Kruyer, H., Morell, M., Pros, E., Serra, E., Ravella, A., Estivill, X., Lazaro, C., 2003.
Recurrent mutations in the NF1 gene are common among neurofibromatosis type 1
patients. J. Med. Genet. 40, e82.
Ars, E., Serra, E., García, J., Kruyer, H., Gaona, A., Lázaro, C., Estivill, X., 2000. Mutations
affecting mRNA splicing are the most common molecular defects in patients with
neurofibromatosis type 1. Hum. Mol. Genet. 9, 237-47. Erratum in: Hum Mol Genet 9,
659.
Baizer, L., Ciment, G., Hendrickson, S.K., Schafer, G.L., 1993. Regulated expression of the
neurofibromin type I transcript in the developing chicken brain. J. Neurochem. 61, 2054-
60.
Bajenaru, M.L., Hernandez, M.R., Perry, A., Zhu, Y., Parada, L.F., Garbow, J.R., Gutmann,
D.H., 2003. Optic nerve glioma in mice requires astrocyte Nf1 gene inactivation and Nf1
brain heterozygosity. Cancer Res. 63, 8573-7.
Ballester, R., Marchuk, D., Boguski, M., Saulino, A., Letcher, R., Wigler, M., Collins, F., 1990.
The NF1 locus encodes a protein functionally related to mammalian GAP and yeast IRA
proteins. Cell 63, 851-9.
Bartels, C.L., Tsongalis, G.J., 2009. MicroRNAs: novel biomarkers for human cancer. Clin.
Chem. 55, 623-31.
Baser, M.E., Contributors to the International NF2 Mutation Database, 2006. The distribution of
constitutional and somatic mutations in the neurofibromatosis 2 gene. Hum. Mutat. 27,
297-306.
Bird, A., 2002. DNA methylation patterns and epigenetic memory. Genes Dev. 16, 6–21.
Bird, A.P., 1986. CpG-rich islands and the function of DNA methylation. Nature 321, 209-13.
Bottillo, I., Ahlquist, T., Brekke, H., Danielsen, S.A., van den Berg, E., Mertens, F., Lothe, R.A.,
Dallapiccola, B., 2009. Germline and somatic NF1 mutations in sporadic and NF1-
associated malignant peripheral nerve sheath tumours. J. Pathol. 217, 693-701.
86
Brown, T.C., Jiricny, J., 1987. A specific mismatch repair event protects mammalian cells from
loss of 5-methylcytosine. Cell 50, 945-50.
Cappione, A.J., French, B.L., Skuse, G.R., 1997. A potential role for NF1 mRNA editing in the
pathogenesis of NF1 tumors. Am. J. Hum. Genet. 60, 305-12.
Carey, J.C., Viskochil, D.H., 1999. Neurofibromatosis type 1: A model condition for the study of
the molecular basis of variable expressivity in human disorders. Am. J. Med. Genet. 89, 7-
13.
Cawthon, R.M., Weiss, R., Xu, G.F., Viskochil, D., Culver, M., Stevens, J., Robertson, M.,
Dunn, D., Gesteland, R., O'Connell, P., et al., 1990. A major segment of the
neurofibromatosis type 1 gene: cDNA sequence, genomic structure, and point mutations.
Cell 62, 193-201. Erratum in: Cell 1990 62, following 608.
Cichowski, K., Jacks, T., 2001. NF1 tumor suppressor gene function: narrowing the GAP. Cell
104, 593-604.
Chan, P.A., Duraisamy, S., Miller, P.J., Newell, J.A., McBride, C., Bond, J.P., Raevaara, T.,
Ollila, S., Nystrom, M., Grimm, A.J., Christodoulou, J., Oetting, W.S., Greenblatt, M.S.,
2007. Interpreting missense variants: comparing computational methods in human disease
genes CDKN2A, MLH1, MSH2, MECP2, and tyrosinase (TYR). Hum. Mutat. 28, 683-
693.
Clark, S.J., Harrison, J., Frommer, M., 1995. CpNpG methylation in mammalian cells. Nat.
Genet. 10, 20-7.
Colman, S.D., Rasmussen, S.A., Vu, T.H., Abernathy, C.R., Wallace, M.R., 1996. Somatin
mosaicism in a patient with neurofibromatosis type 1. Am. J. Hum. Genet. 58, 484-490.
Colman S.D., Williams C.A., Wallace M.R., 1995. Benign neurofibromas in type 1
neurofibromatosis (NF1) show somatic deletions of the NF1 gene. Nat. Genet. 11, 90-2.
Colman, S.D., Collins, F.S., Wallace, M.R., 1993. Characterization of a single base-pair deletion
in neurofibromatosis type 1. Hum. Mol. Genet. 2, 1709-11.
Cooper, D.N., Ball, E.V., Krawczak, M., 1998. The human gene mutation database. Nucleic
Acids Res. 26, 285-7.
Cooper, D.N., Krawczak, M., 1993. Human Gene Mutation. Oxford: BIOS Scientific Publishers.
Costa, R.M., Yang, T., Huynh, D.P., Pulst, S.M., Viskochil, D.H., Silva, A.J., Brannan, C.I.,
2001. Learning deficits, but normal development and tumor predisposition, in mice lacking
exon 23a of Nf1. Nat. Genet. 27, 399-405.
D'Angelo, I., Welti, S., Bonneau, F., Scheffzek, K., 2006. A novel bipartite phospholipid-binding
module in the neurofibromatosis type 1 protein. EMBO. Rep. 7, 174-9.
87
Danglot, G., Régnier, V., Fauvet, D., Vassal, G., Kujas, M., Bernheim, A., 1995.
Neurofibromatosis 1 (NF1) mRNAs expressed in the central nervous system are
differentially spliced in the 5' part of the gene. Hum. Mol. Genet. 5, 915-20.
Daston, M.M., Scrable, H., Nordlund, M., Sturbaum, A.K., Nissen, L.M., Ratner, N., 1992. The
protein product of the neurofibromatosis type 1 gene is expressed at highest abundance in
neurons, Schwann cells, and oligodendrocytes. Neuron 8, 415-28.
DeClue, J.E., Cohen, B.D., Lowy, D.R., 1991. Identification and characterization of the
neurofibromatosis type 1 protein product. Proc. Natl. Acad. Sci. U S A. 88, 9914-8.
De Luca, A., Bottillo, I., Sarkozy, A., Carta, C., Neri, C., Bellacchio, E., Schirinzi, A., Conti, E.,
Zampino, G., Battaglia, A., Majore, S., Rinaldi, M.M., Carella, M., Marino, B., Pizzuti, A.,
Digilio, M.C., Tartaglia, M., Dallapiccola, B., 2005. NF1 gene mutations represent the
major molecular event underlying neurofibromatosis-Noonan syndrome. Am. J. Hum.
Genet. 77, 1092-101. Epub 2005 Oct 26.
De Luca, A., Buccino, A., Gianni, D., Mangino, M., Giustini, S., Richetta, A., Divona, L.,
Calvieri, S., Mingarelli, R., Dallapiccola, B., 2003. NF1 gene analysis based on DHPLC.
Hum. Mutat. 21, 171-2.
De Schepper, S., Maertens, O., Callens, T., Naeyaert, J.M., Lambert, J., Messiaen, L., 2008.
Somatic mutation analysis in NF1 café au lait spots reveals two NF1 hits in the
melanocytes. J. Invest. Dermatol. 128, 1050-3.
De Schepper, S., Boucneau, J.M., Westbroek, W., Mommaas, M., Onderwater, J., Messiaen, L.,
Naeyaert, J.M., Lambert, J.L., 2006. Neurofibromatosis type 1 protein and amyloid
precursor protein interact in normal human melanocytes and colocalize with melanosomes.
J Invest. Dermatol. 3, 653-9.
Dorschner, M.O., Sybert, V.P., Weaver, M., Pletcher, B.A., Stephens, K., 2000. NF1
microdeletion breakpoints are clustered at flanking repetitive sequences. Hum. Mol. Genet.
9, 35-46.
Driscoll, D.J., Migeon, B.R., 1990. Sex difference in methylation of single-copy genes in human
meiotic germ cells: implications for X chromosome inactivation, parental imprinting, and
origin of CpG mutations. Somat. Cell Mol. Genet. 16, 267-282.
Dublin, S., Riccardi, V.M., Stephens, K., 1995. Methods for rapid detection of a recurrent
nonsense mutation and documentation of phenotypic features in neurofibromatosis type 1
patients. Hum. Mutat. 5, 81-85.
Early, P., Rogers, J., Davis, M., Calame, K., Bond, M., Wall, R., Hood, L., 1980. Two mRNAs
can be produced from a single immunoglobulin mu gene by alternative RNA processing
pathways. Cell 20, 313-9.
88
Easton, D.F., Ponder, M.A., Huson, S.M., Ponder, B.A., 1993. An analysis of variation in
expression of neurofibromatosis (NF) type 1 (NF1): evidence for modifying genes. Am. J.
Hum. Genet. 53, 305-13.
Eisenbarth, I., Beyer, K., Krone, W., Assum, G., 2000. Toward a survey of somatic mutation of
the NF1 gene in benign neurofibromas of patients with neurofibromatosis type 1. Am. J.
Hum. Genet. 66, 393-401.
Fahsold, R., Hoffmeyer, S., Mischung, C., Gille, C., Ehlers, C., Kucukceylan, N., Abdel-Nour,
M., Gewies, A., Peters, H., Kaufmann, D., Buske, A., Tinschert, S., Nurnberg, P., 2000.
Minor lesion mutational spectrum of the entire NF1 gene does not explain its high
mutability but points to a functional domain upstream of the GAP-related domain. Am. J.
Hum. Genet. 66, 790-818.
Ferner, R.E., Gutmann, D.H., 2002. International consensus statement on malignant peripheral
nerve sheath tumors in neurofibromatosis 1. Cancer Res. 62, 1573-1577.
Fishbein, L., Eady, B., Sanek, N., Muir, D., Wallace, M.R., 2005. Analysis of somatic NF1
promoter methylation in plexiform neurofibromas and Schwann cells. Cancer Genet.
Cytogenet. 157, 181-186.
Forbes, S.H., Dorschner, M.O., Le, R., Stephens, K., 2004. Genomic context of paralogous
recombination hotspots mediating recurrent NF1 region microdeletion. Genes
Chromosomes Cancer 41, 12-25.
Friedman, J.M., 1999. Clinical Genetics, in: Friedman, J.M., Gutmann, D.H., MacCollin, M.,
Riccardi, V.M. (eds.), Neurofibromatosis: Phenotype, Natural History, and Pathogenesis.
Baltimore, Johns Hopkins University Press.
Gao, Q.S., Memmott, J., Lafyatis, R., Stamm, S., Screaton, G., Andreadis, A., 2000. Complex
regulation of tau exon 10, whose missplicing causes frontotemporal dementia. J.
Neurochem. 74, 490-500.
Geist, R.T., Gutmann, D.H., 1996. Expression of a developmentally-regulated neuron-specific
isoform of the neurofibromatosis 1 (NF1) gene. Neurosci. Lett. 211, 85-8.
Girodon-Boulandet, E., Pantel, J., Cazeneuve, C., Gijn, M.V., Vidaud, D., Lemay, S., Martin, J.,
Zeller, J., Revuz, J., Goossens, M., Amselem, S., Wolkenstein, P., 2000. NF1 gene analysis
focused on CpG-rich exons in a cohort of 93 patients with neurofibromatosis type 1. Hum.
Mutat. 16, 274-5.
Gottfried, O.N., Viskochil, D.H., Fults, D.W., Couldwell, W.T., 2006. Molecular, genetic, and
cellular pathogenesis of neurofibromas and surgical implications. Neurosurgery 58, 1-16.
Greenblatt, M.S., Bennett, W.P., Hollstein, M., Harris, C.C., 1994. Mutations in the p53 tumor
suppressor gene: clues to cancer etiology and molecular pathogenesis. Cancer Res. 54,
4855-78.
89
Gregory, P.E., Gutmann, D.H., Mitchell, A., Park, S., Boguski, M., Jacks, T., Wood, D.L., Jove,
R., Collins, F.S., 1993. Neurofibromatosis type 1 gene product (neurofibromin) associates
with microtubules. Somat. Cell Mol. Genet. 19, 265-74.
Gutmann, D.H., Aylsworth, A., Carey, J.C., Korf, B., Marks, J., Pyeritz, R.E., Rubenstein, A.,
Viskochil, D., 1997. The diagnostic evaluation and multidisciplinary management of
neurofibromatosis 1 and neurofibromatosis 2. JAMA 278, 51-7.
Gutmann, D.H., Geist, R.T., Wright, D.E., Snider, W.D., 1995. Expression of the
neurofibromatosis 1 (NF1) isoforms in developing and adult rat tissues. Cell Growth Dif.
6, 315-323.
Gutmann, D.H., Cole, J.L., Collins, F.S., 1994. Modulation of neurofibromatosis type 1 gene
expression during in vitro myoblast differentiation. J. Neuro. Res. 37, 398-405.
Gutman, D.H., Andersen, L.B., Cole, J.L., Swaroop, M., Collins, F.S., 1993a. An alternatively-
spliced mRNA in the carboxy terminus of the neurofibromatosis type 1 (NF1) gene is
expressed in muscle. Hum. Mol. Genet. 2, 989-92.
Gutmann, D.H., Boguski, M., Marchuk, D., Wigler, M., Collins, F.S., Ballester, R., 1993b.
Analysis of the neurofibromatosis type 1 (NF1) GAP-related domain by site-directed
mutagenesis. Oncogene 8, 761-9.
Gutmann, D.H., Wood, D.L., Collins, F.S., 1991. Identification of the neurofibromatosis type 1
gene product. Proc. Natl. Acad. Sci. U S A 88, 9658-62.
Han, S.S., Cooper, D.N., Upadhyaya, M.N., 2001. Evaluation of denaturing high performance
liquid chromatography (DHPLC) for the mutational analysis of the neurofibromatosis type
1 (NF1) gene. Hum. Genet. 109, 487-97. Epub 2001 Oct 11.
Hawes, J.J., Tuskan, R.G., Reilly, K.M., 2007. Nf1 expression is dependent on strain
background: implications for tumor suppressor haploinsufficiency studies. Neurogenetics
8, 121-30.
Henikiff, S., Henikoff, J.G., 1992. Amino acid substitution matrices from protein blocks. Proc.
Natl. Acad. Sci. U S A 89, 10915-10919.
Horiuchi, T., Hatta, N., Matsumoto, M., Ohtsuka, H., Collins, F.S., Kobayashi, Y., Fujita, S.,
1994. Nonsense mutations at Arg-1947 in two cases of familial neurofibromatosis type 1 in
Japanese. Hum. Genet. 93, 81-83.
Huang, J., Zheng, S., Jin, S.H., Zhang, S.Z., 2004. Somatic mutations of APC gene in
carcinomas from hereditary non-polyposis colorectal cancer patients. World J.
Gastroenterol. 10, 834-6.
Huynh, D.P., Nechiporuk, T., Pulst, S.M., 1994. Alternative transcripts in the mouse
neurofibromatosis type 2 (NF2) gene are conserved and code for schwannomins with
distinct C-terminal domains. Hum. Mol. Genet. 3, 1075-9.
90
Jadayel, D., Fain, P., Upadhyaya, M., Ponder, M.A., Huson, S.M., Carey, J., Fryer, A., Mathew,
C.G., Barker, D.F., Ponder, B.A., 1990. Paternal origin of new mutations in von
Recklinghausen neurofibromatosis. Nature 343, 558-559.
Jeong, S.Y., Park, S.J., Kim, H.J., 2006. The spectrum of NF1 mutations in Korean patients with
neurofibromatosis type 1. J. Korean Med. Sci. 21, 107-12.
Jones, P.A., Buckley, J.D., Henderson, B.E., Ross, R.K., Pike, M.C., 1991. From gene to
carcinogen: a rapidly evolving field in molecular epidemiology. Cancer Res. 51, 3617-
3620.
Kaufmann, D., Müller, R., Kenner, O., Leistner, W., Hein, C., Vogel, W., Bartelt, B., 2002. The
N-terminal splice product NF1-10a-2 of the NF1 gene codes for a transmembrane segment.
Biochem. Biophys. Res. Commun. 294, 496-503.
Kayes, L.M., Burke, W., Riccardi, V.M., Ehrlich, P., Rubenstein, A., Stephens, K., 1994.
Deletions spanning the neurofibromatosis 1 gene: identification and phenotype of five
patients. Am. J. Hum. Genet. 54, 424-436.
Kayes, L.M., Riccardi, V.M., Burke, W., Bennett, R.L., Stephens, K., 1992. Large de novo DNA
deletion in a patient with sporadic neurofibromatosis 1, mental retardation, and
dysmorphism. J Med. Genet. 29, 686-690.
Kehrer-Sawatzki, H., Kluwe, L., Sandig, C., Kohn, M., Wimmer, K., Krammer, U., Peyrl, A.,
Jenne, D.E., Hansmann, I., Mautner, V.F., 2004. High frequency of mosaicism among
patients with neurofibromatosis type 1 (NF1) with microdeletions caused by somatic
recombination of the JJAZ1 gene. Am. J. Hum. Genet. 75, 410-23. Epub 2004 Jul 15.
Kluwe, L., Siebert, R., Gesk, S., Friedrich, R.E., Tinschert, S., Kehrer-Sawatzki, H., Mautner,
V.F., 2004. Screening 500 unselected neurofibromatosis 1 patients for deletions of the NF1
gene. Hum. Mutat. 23, 111-116.
Kluwe, L., Friedrich, R.E., Peiper, M., Friedman, J., Mautner, V.F., 2003. Constitutional NF1
mutations in neurofibromatosis 1 patients with malignant peripheral nerve sheath tumors.
Hum. Mutat. 22, 420.
Kluwe, L., Friedrich, R.E., Korf, B., Fahsold, R., Mautner, V.F., 2002. NF1 mutations in
neurofibromatosis 1 patients with plexiform neurofibromas. Hum. Mutat. 19, 309.
Knudson, A.G. Jr., 1971. Mutation and cancer: statistical study of retinoblastoma. Proc. Natl.
Acad. Sci. U.S.A. 68, 820-823.
Kobayashi, C., Oda, Y., Takahira, T., Izumi, T., Kawaguchi, K., Yamamoto, H., Tamiya, S.,
Yamada, T., Oda, S., Tanaka, K., Matsuda, S., Iwamoto, Y., Tsuneyoshi, M., 2006.
Chromosomal aberrations and microsatellite instability of malignant peripheral nerve
sheath tumors: a study of 10 tumors from nine patients. Cancer Genet. Cytogenet. 165, 98-
105.
91
Krkljus, S., Abernathy, C.R., Johnson, J.S., Williams, C.A., Driscoll, D.J., Zori, R., Stalker, H.J.,
Rasmussen, S.A., Collins, F.S., Kousseff, B.G., Baumbach, L., Wallace, M.R., 1998.
Analysis of CpG C-to-T mutations in neurofibromatosis type 1. Mutations in brief no. 129.
Online. Hum. Mutat. 11, 411.
Kryukov, G., Pennacchio, L., Sunyaev, S., 2007. Most rare missense alleles are deleterious in
humans: implications for complex disease and association studies. Am. J. Hum. Genet. 80,
727-39. Epub 2007 Mar 8.
Kulyte, A., Dryselius, R., Karlsson, J., Good, L., 2005. Gene selective suppression of nonsense
termination using antisense agents. Biochem. Biophys. Acta. 1730, 165-72.
Lari, S.U., Famulski, K., Al-Khodairy, F., 2004. Multiplicity of strand incision at G:T base
mismatches in DNA by human cell extracts. Biochemistry 43, 6691-7.
Lari, S.U., Al-Khodairy, F., Paterson, M.C., 2002. Substrate specificity and sequence preference
of G:T mismatch repair: incision at G:T, O6-methylguanine:T, and G:U mispairs in DNA
by human cell extracts. Biochemistry 29, 9248-55.
Lazaro, C., Gaona, A., Ainsworth, P., Tenconi, R., Vidaud, D., Kruyer, H., Ars, E., Volpini, V.,
Estivill, X., 1996. Sex differences in mutational rate and mutational mechanism in the NF1
gene in neurofibromatosis type 1 patients. Hum. Genet. 98, 696-699.
Lazaro, C., Kruyer, H., Gaona, A., Estivill, X., 1995. Two further cases of mutation R1947X in
the NF1 gene: screening for a relatively common recurrent mutation. Hum. Genet. 3, 361-
363.
Leppig, K.A., Kaplan, P., Viskochil, D., Weaver, M., Orterberg, J., Stephens, K., 1997. Familial
neurofibromatosis 1 gene deletions: cosegregation with distinctive facial features and early
onset of cutaneous neurofibromas. Am. J. Med. Genet. 73, 197-204.
Leppig, K.A., Viskochil, D., Neil, S., Rubenstein, A., Johnson, V.P., Zhu, X.L., Brothman, A.R.,
Stephens, K., 1996. The detection of contiguous gene deletions at the neurofibromatosis 1
locus with fluorescence in situ hybridization. Cytogenet. Cell Genet. 72, 95-98.
Lee, M.J., Su, Y.N., You, H.L., Chiou, S.C., Lin, L.C., Yang, C.C., Lee, W.C., Hwu, W.L.,
Hsieh, F.J., Stephenson, D.A., Yu, C.L., 2006. Identification of forty-five novel and
twenty-three known NF1 mutations in Chinese patients with neurofibromatosis type 1.
Hum. Mutat. 27, 832.
Lee, M.P., Feinberg, A.P., 1997. Aberrant splicing but not mutations of TSG101 in human breast
cancer. Cancer Res. 57, 3131-4.
Li, Y., O’Connell, P., Breidenbach, H.H., Cawthon, R., Stevens, J., Xu, G., Neil, S., Robertson,
M., White, R., Viskochil, D., 1995. Genomic organization of the neurofibromatosis 1 gene
(NF1). Genomics 25, 9-18.
92
Li, Y., Bollag, G., Clark, R., Stevens, J., Conroy, L., Fults, D., Ward, K., Friedman, E.,
Samowitz, W., Robertson, M., Bradley, P., McCormick, F., White, R., Cawthon, R., 1992.
Somatic mutations in the neurofibromatosis 1 gene in human tumors. Cell 69, 275-81.
Ling, J.Q., Li, T., Hu, J.F., Vu, T.H., Chen, H.L., Qiu, X.W., Cherry, A.M., Hoffman, A.R.,
2006. CTCF mediates interchromosomal colocalization between Igf2/H19 and Wsb1/Nf1.
Science 312, 269-72.
Loda-Hutchinson, R.L., Driscoll, D.J., Wallace, M.R., 2009. Comparison of computational
methods in predicting pathogenicity of NF1 missense mutations.
Lohmann, D.R., Brandt, B., Höpping, W., Passarge, E., Horsthemke, B., 1996. The spectrum of
RB1 germ-line mutations in hereditary retinoblastoma. Am. J. Hum. Genet. 58, 940-9.
Lopez-Correa, C., Dorschner, M., Brems, H., Lazaro, C., Clementi, M., Upadhyaya, M., Dooijes,
D., Moog, U., Kehrer-Sawatzki, H., Rutkowski, J.L., Fryns, J.P., Marynen, P., Stephens,
K., Legius, E., 2001. Recombinations hotspot in NF1 microdeletion patients. Hum. Mol.
Genet. 10, 1387-1392.
Luijten, M., Redeker, S., van Noesel, M.M., Troost, D., Westerveld, A., Hulsebos, T.J., 2000.
Microsatellite instability and promoter methylation as possible causes of NF1 gene
inactivation in neurofibromas. Eur. J. Hum. Genet. 8, 939-45.
MacCollin, M., Wallace, M.R., 2002. Molecular genetics of NF1 and NF2, in: Cowell, J.K. (ed.),
Molecular Genetics of Cancer, Second ed. Oxford BIOS Scientific Publishers, 95-113.
Maertens, O., Brems, H., Vandesompele, J., De Raedt, T., Heyns, I., Rosenbaum, T., De
Schepper, S., De Paepe, A., Mortier, G., Janssens, S., Speleman, F., Legius, E., Messiaen,
L., 2006. Comprehensive NF1 screening on cultured Schwann cells from neurofibromas.
Hum. Mutat. 27, 1030-40.
Malminen, M., Peltonen, S., Koivunen, J., Peltonen, J., 2002. Functional expression of NF1
tumor suppressor protein: association with keratin intermediate filaments during the early
development of human epidermis. BMC Dermatol. 2,10.
Mantani, A., Wakasugi, S., Yokota, Y., Abe, K., Ushio, Y., Yamamura, K., 1994. A novel
isoform of the neurofibromatosis type-1 mRNA and a switch of isoforms during murine
cell differentiation and proliferation. Gene 148, 245-51.
Martinez, R., Schackert, G., 2007. Epigenetic aberrations in malignant gliomas: an open door
leading to better understanding and treatment. Epigenetics 2,147-50.
Mathe, E., Olivier, M., Kato, S., Ishioka, C., Hainaut, P., Tavtigian, S.V., 2006. Computational
approaches for predicting the biological effect of p53 missense mutations: a comparison of
three sequence analysis based tools. Nucleic Acids Res. 34, 1317-1325.
93
Mattocks, C., Baralle, D., Tarpey, P., ffrench-Constant, C., Bobrow, M., Whittaker, J., 2004.
Automated comparative sequence analysis identifies mutations in 89% of NF1 patients and
confirms a mutation cluster in exons 11-17 distinct from the GAP related domain. J. Med.
Genet. 41, e48.
Maynard, J., Krawczak, M., Upadhyaya, M., 1997. Characterization and significance of nine
novel mutations in exon 16 of the neurofibromatosis type 1 (NF1) gene. Hum. Genet. 99,
674-6.
McClatchey, A,I., 2007. Neurofibromatosis. Annu. Rev. Pathol. 2, 191-216.
Messiaen, L.M., Callens, T., Mortier, G., Beysen, D., Vandenbroucke, I., Van Roy, N.,
Speleman, F., Paepe, A.D., 2000. Exhaustive mutation analysis of the NF1 gene allows
identification of 95% of mutations and reveals a high frequency of unusual splicing
defects. Hum. Mutat. 15, 541-55.
Messiaen, L.M., Callens, T., Roux, K.J., Mortier, G.R., De Paepe, A., Abramowicz, M., Pericak-
Vance, M.A., Vance, J.M., Wallace, M.R., 1999. Exon 10b of the NF1 gene represents a
mutational hotspot and harbors a recurrent missense mutation Y489C associated with
aberrant splicing. Genet. Med. 1, 248-53.
Mukhopadhyay, D., Anant, S., Lee, R.M., Kennedy, S., Viskochil, D., Davidson, N.O., 2002. C--
>U editing of neurofibromatosis 1 mRNA occurs in tumors that express both the type II
transcript and apobec-1, the catalytic subunit of the apolipoprotein B mRNA-editing
enzyme. Am. J. Hum. Genet. 70, 38-50.
Nagoshi, R.N., McKeown, M., Burtis, K.C., Belote, J.M., Baker, B.S., 1988. The control of
alternative splicing at genes regulating sexual differentiation in D. melanogaster. Cell 53,
229-36.
Ng, P.C., Henikoff, S., 2001. Predicting deleterious amino acid substitutions. Genome Res. 11,
863-874.
Ninio, M., Privman, E., Pupko, T., Friedman, N., 2007. Phylogeny reconstruction: increasing the
accuracy of pairwise distance estimation using Bayesian inference of evolutionary rates.
Bioinformatics 23, e136-41.
Nishi, T., Lee, P.S., Oka, K., Levin, V.A., Tanase, S., Morino, Y., Saya, H., 1991. Differential
expression of two types of the neurofibromatosis type 1 (NF1) gene transcripts related to
neuronal differentiation. Oncogene. 6, 1555-1559.
Nishi, T., Saya, H., 1991. Neurofibromatosis type 1 (NF1) gene: implication in neuroectodermal
differentiation and genesis of brain tumors. Cancer Metastasis Rev. 10, 301-10.
O'Brien, K.P., Remm, M., Sonnhammer, E.L., 2005. Inparanoid: a comprehensive database of
eukaryotic orthologs. Nucleic Acids Res. 33(Database issue), D476-80.
94
Ogata, H., Sato, H., Takatsuka, J., De Luca, L.M., 2001. Human breast cancer MDA-MB-231
cells fail to express the neurofibromin protein, lack its type I mRNA isoform and show
accumulation of P-MAPK and activated Ras. Cancer Letters. 172, 159-164.
Olivier, M., Eeles, R., Hollstein, M., Khan, M.A., Harris, C.C., Hainaut, P., 2002. The IARC
TP53 database: new online mutation analysis and recommendations to users. Hum. Mutat.
19, 607-614.
Ottini, L., Esposito, D.L., Richetta, A., Carlesimo, M., Palmirotta, R., Verí, M.C., Battista, P.,
Frati, L., Caramia, F.G., Calvieri, S., et al., 1995. Alterations of microsatellites in
neurofibromas of von Recklinghausen's disease. Cancer Res. 55, 5677-80.
Park, V.M., Pivnick, E.K., 1998. Neurofibromatosis type 1 (NF1): a protein truncation assay
yielding identification of mutations in 73% of patients. J. Med. Genet. 35, 813-20.
Peters, H., Hess, D., Fahsold, R., Schülke, M., 1999. A novel mutation L1425P in the GAP-
region of the NF1 gene detected by temperature gradient gel electrophoresis (TGGE).
Mutation in brief no. 230. Online. Hum. Mutat. 13, 337.
Pinotti, M., Rizzotto, L., Chuansumrit, A., Mariani, G., Bernardi, F., International Factor VII
Deficiency Study Group, 2006. Gentamicin induces sub-therapeutic levels of coagulation
factor VII in patients with nonsense mutations. J. Thromb. Haemost. 4, 1828-30.
Poullet, P., Lin, B., Esson, K., Tamanoi, F., 1994. Functional significance of lysine 1423 of
neurofibromin and characterization of a second site suppressor which rescues mutations at
this residue and suppresses RAS2Val-19-activated phenotypes. Mol. Cell Biol. 14, 815-21.
Pros, E., Gómez, C., Martín, T., Fábregas, P., Serra, E., Lázaro, C., 2008. Nature and mRNA
effect of 282 different NF1 point mutations: focus on splicing alterations. Hum. Mutat. 29,
E173-E193. [Epub ahead of print]
Purandare, S.M., Lanyon, W.G., Connor, J.M., 1994. Characterisation of inherited and sporadic
mutations in neurofibromatosis type-1. Hum. Mol. Genet. 3, 1109-15.
Raedt, T.D., Stephens, M., Heyns, I., Brems, H., Thijs, D., Messiaen, L., Stephens, K., Lazaro,
C., Wimmer, K., Kehrer-Sawatzki, H., Vidaud, D., Kluwe, L., Marynen, P., Legius, E.,
2006. Conservation of hotspots for recombination in low-copy repeats associated with the
NF1 microdeletion. Nat. Genet. 38, 1419-23. Epub 2006 Nov 19.
Rasmussen, S.A., Colman, S.D., Ho, V.T., Abernathy, C.R., Arn, P.H., Weiss, L., Schwartz, C.,
Saul, R.A., Wallace, M.R., 1998. Constitutional and mosaic large NF1 gene deletions in
neurofibromatosis type 1. J. Med. Genet. 35, 468-471.
Rasmussen, S.A., Wallace, M.R., 1998. Somatic mutations of the NF1 gene in type 1
neurofibromatosis and cancer, in: Upadhyaya, M., Cooper, D.N. (eds.), Neurofibromatosis
Type 1: from Genotype to Phenotype, Oxford: BIOS Scientific Publishers, 153-165.
95
Rasmussen, S.A., Overman, J., Thomson, S.A., Colman, S.D., Abernathy, C.R., Trimpert, R.E.,
Moose, R., Virdi, G., Roux, K., Bauer, M., Rojiani, A.M., Maria, B.L., Muir, D., Wallace,
M.R., 2000. Chromosome 17 loss-of-heterozygosity studies in benign and malignant
tumors in neurofibromatosis type 1. Genes Chromosomes Cancer 28, 425-31.
Razin, A. and Riggs, A. D., 1980. DNA methylation and gene function. Science 210, 604–610.
Reese, M.G., Eeckman, F.H., Kulp, D., Haussler, D., 1997. Improved splice site detection in
Genie. J. Comput. Biol. 4, 311-23.
Riccardi, V.M., 1992. The prenatal diagnosis of NF-1 and NF-2. J. Dermatol. 19, 885-91.
Roudebush, M., Slabe, T., Sundaram, V., Hoppel, C.L., Golubic, M., Stacey, D.W., 1997.
Neurofibromin colocalizes with mitochondria in cultured cells. Exp. Cell Res. 236, 161-72.
Sawada, S., Florell, S., Purandare, S.M., Ota, M., Stephens, K., Viskochil, D., 1996.
Identification of NF1 mutations in both allales of a dermal neurofibroma. Nat. Genet. 14,
110-112.
Serra, E., Rosenbaum, T., Nadal, M., Winner, U., Ars, E., Estivill, X., Lázaro, C., 2001a. Mitotic
recombination effects homozygosity for NF1 germline mutations in neurofibromas. Nat.
Genet. 28, 294-6. Erratum in: Nat Genet 2001 29, 100.
Serra, E., Ars, E., Ravella, A., Sanchez, A., Puig, S., Rosenbaum, T., Estivill, X., Lazaro, C.,
2001b. Somatic NF1 mutational spectrum in benign neurofibromas: mRNA splice defects
are common among point mutations. Hum. Genet. 108, 416-429
Serra, E., Rosenbaum, T., Winner, U., Aledo, R., Ars, E., Estivill, X., Lenard, H., Lazaro, C.,
2000. Schwann cells harbor the somatic NF1 mutation in neurofibromas: evidence of two
different Schwann cell subpopulations. Hum. Mol. Genet. 9, 3055-3064.
Serra, E., Puig, S., Otero, D., Gaona, A., Kruyer, H., Ars, E., Estivill, X., Lazaro, C., 1997.
Confirmation of a double-hit model for the NF1 Gene in benign neurofibromas. Am. J.
Hum. Genet. 61, 512-519.
Shelton, B.P., Misso, N.L., Shaw, O.M., Arthaningtyas, E., Bhoola, K.D., 2008. Epigenetic
regulation of human epithelial cell cancers. Curr. Opin. Mol. Ther. 10, 568-78.
Skuse, G.R., Cappione, A.J., 1997. RNA processing and clinical variability in neurofibromatosis
type I (NF1). Hum. Mol. Genet. 6, 1707-1712.
Skuse, G.R., Cappione, A.J., Sowden, M., Metheny, L.J., Smith, H.C., 1996. The
neurofibromatosis type I messenger RNA undergoes base-modification RNA editing.
Nucleic Acids Res. 24, 478-85.
Skuse, G.R., Kosciolek, B.A., Rowley, P.T., 1991. The neurofibroma in von Recklinghausen
neurofibromatosis has a unicellular origin. Am. J. Hum. Genet. 49, 600-607.
96
Smith, C.W., Patton, J.G., Nadal-Ginard, B., 1989. Alternative splicing in the control of gene
expression. Annu. Rev. Genet. 23, 527-77.
Stephens, K., Kayes, L., Riccardi, V.M., Rising, M., Sybert, V.P., Pagon, R.A., 1992.
Preferential mutation of the neurofibromatosis type 1 gene in paternally derived
chromosomes. Hum. Genet. 88, 279-82.
Stevenson, D.A., Zhou, H., Ashrafi, S., Messiaen, L.M., Carey, J.C., D'Astous, J.L., Santora,
S.D., Viskochil, D.H., 2006. Double inactivation of NF1 in tibial pseudarthrosis. Am. J.
Hum. Genet. 79, 143-8. Epub 2006 May 10.
Stimpfl, M., Tong, D., Fasching, B., Schuster, E., Obermair, A., Leodolter, S., Zeillinger, R.,
2002. Vascular endothelial growth factor splice variants and their prognostic value in
breast and ovarian cancer. Clin. Cancer Res. 8, 2253-9.
Stone, E.A., Sidow, A., 2005. Physicochemical constraint violation by missense substitutions
mediates impairment of protein function and disease severity. Genome Res. 15, 978-86.
Epub 2005 Jun 17.
Storlazzi, C.T., Von Steyern, F.V., Domanski, H.A., Mandahl, N., Mertens, F., 2005. Biallelic
somatic inactivation of the NF1 gene through chromosomal translocations in a sporadic
neurofibroma. Int. J. Cancer 117,1055-7.
Suzuki, H., Takahashi, K., Kubota, Y., Shibahara, S., 1992. Molecular cloning of a cDNA
coding for neurofibromatosis type 1 protein isoform lacking the domain related to ras
GTPase-activating protein. Biochem. Biophys. Res. Commun. 187, 984-990.
Szudek, J., Joe, H., Friedman, J.M., 2002. Analysis of intrafamilial phenotypic variation in
neurofibromatosis 1 (NF1). Genet. Epidemiol. 23, 150-64.
Teinturier, C., Danglot, G., Slim, R., Pruliere, D., Launay, J.M., Bernheim, A., 1992. The
neurofibromatosis 1 gene transcripts expressed in peripheral nerve and neurofibromas bear
the additional exon located in the GAP domain. Biochem. Biophys. Res. Commun. 188,
851-857.
Thomson, S.A., Fishbein, L., Wallace, M.R., 2002. NF1 mutations and molecular testing. J.
Child. Neurol. 17, 555-561.
Toledano, H., Goldberg, Y., Kedar-Barnes, I., Baris, H., Porat, R.M., Shochat, C., Bercovich, D.,
Pikarsky, E., Lerer, I., Yaniv, I., Abeliovich, D., Peretz, T., 2008. Homozygosity of MSH2
c.1906G-->C germline mutation is associated with childhood colon cancer, astrocytoma
and signs of Neurofibromatosis type I. Fam. Cancer Dec 20.
Tonsgard, J.H., Kwak, S.M., Short, M.P., Dachman, A.H., 1998. CT imaging in adults with
neurofibromatosis-1: frequent asymptomatic plexiform lesions. Neurology 50, 1755-60.
97
Uchida, T., Matozaki, T., Suzuki, T., Matsuda, K., Wada, K., Nakano, O., Konda, Y., Nishisaki,
H., Nagao, M., Sakamoto, C., et al., 1992. Expression of two types of neurofibromatosis
type 1 gene transcripts in gastric cancers and comparison of GAP activities. Biochem.
Biophys. Res. Commun. 187, 332-9.
Upadhyaya, M., Spurlock, G., Monem, B., Thomas, N., Friedrich, R.E., Kluwe, L., Mautner, V.,
2008. Germline and somatic NF1 gene mutations in plexiform neurofibromas. Hum.
Mutat. 29, E103-E111
Upadhyaya, M., Han, S., Consoli, C., Majounie, E., Horan, M., Thomas, N.S., Potts, C.,
Griffiths, S., Ruggieri, M., von Deimling, A., Cooper, D.N., 2004. Characterization of the
somatic mutational spectrum of the neurofibromatosis type 1 (NF1) gene in
neurofibromatosis patients with benign and malignant tumors. Hum. Mutat. 23, 134-46.
Upadhyaya, M., Cooper, D.N., 1998. The mutational spectrum in neurofibromatosis 1 and its
underlying mechanisms, in: Upadhyaya, M., Cooper, D.N., (eds.), Neurofibromatosis Type
1: from Genotype to Phenotype, Oxford: BIOS Scientific Publishers, 65-88.
Upadhyaya, M., Ruggieri, M., Maynard, J., Osborn, M., Hartog, C., Mudd, S., Penttinen, M.,
Cordeiro, I., Ponder, M., Ponder, B.A., Krawczak, M., Cooper, D.N., 1998. Gross
deletions of the neurofibromatosis type 1 (NF1) gene are predominantly of maternal origin
and commonly associated with a learning disability, dysmorphic features and
developmental delay. Hum. Genet. 102, 591-597.
Upadhyaya, M., Maynard, J., Osborn, M., Harper, P.S., 1997. Six novel mutations in the
neurofibromatosis type 1 (NF1) gene. Hum. Mutat. 10, 248-50
Upadhyaya, M., Osborn, M.J., Maynard, J., Kim, M.R., Tamanoi, F., Cooper, D.N., 1997.
Mutational and functional analysis of the neurofibromatosis type 1 (NF1) gene. Hum.
Genet. 99, 88-92.
Upadhyaya, M., Maynard, J., Osborn, M., Huson, S.M., Ponder, M., Ponder, B.A., Harper, P.S.,
1995. Characterisation of germline mutations in the neurofibromatosis type 1 (NF1) gene.
J. Med. Genet. 32, 706-10.
Upadhyaya, M., Shen, M., Cherryson, A., Farnham, J., Maynard, J., Huson, S.M., Harper, P.S.,
1992. Analysis of mutations at the neurofibromatosis 1 (NF1) locus. Hum. Mol. Genet. 1,
735-40.
Valero, M.C., Pascual-Castroviejo, I., Velasco, E., Moreno, F., Hernandez-Chico, C., 1997.
Identification of de novo deletions at the NF1 gene: no preferential paternal origin and
phenotypic analysis of patients. Hum. Genet. 99, 720-726.
Vandenbroucke, I., Van Oostveldt, P., Coene, E., De Paepe, A., Messiaen, L., 2004.
Neurofibromin is actively transported to the nucleus. FEBS Lett. 560, 98-102.
Vandenbroucke, I.I., Vandesompele, J., Paepe, A.D., Messiaen, L., 2001. Quantification of splice
variants using real-time PCR. Nucleic Acids Res. 29, E68-8.
98
Viskochil, D., Buchberg, A.M., Xu, G., Cawthon, R.M., Stevens, J., Wolff, R.K., Culver, M.,
Carey, J.C., Copeland, N.G., Jenkins, N.A., White, R., O'Connell, P., 1990. Deletions and
a translocation interrupt a cloned gene at the neurofibromatosis type 1 locus. Cell 62, 87-
192.
Wallace, M.R., Rasmussen, S.A., Lim, I.T., Gray, B.A., Zori, R.T., Muir, D., 2000. Culture of
cytogenetically abnormal schwann cells from benign and malignant NF1 tumors. Genes
Chromosomes Cancer 27,117-23.
Wallace, M.R., Andersen, L.B., Saulino, A.M., Gregory, P.E., Glover, T.W., Collins, F.S., 1991.
A de novo Alu insertion results in neurofibromatosis type 1. Nature 353, 864-866.
Wallace, M.R., Marchuk, D.A., Andersen, L.B., Letcher, R., Odeh, H.M., Saulino, A.M.,
Fountain, J.W., Brereton, A., Nicholson, J., Mitchell, A.L., Brownstein, B.H., Collins, F.S.,
1990. Type 1 neurofibromatosis gene: identification of a large transcript disrupted in three
NF1 patients. Science 249, 181-186.
Wang, Q., Montmain, G., Ruano, E., Upadhyaya, M., Dudley, S., Liskay, R.M., Thibodeau, S.N.,
Puisieux, A., 2003. Neurofibromatosis type 1 gene as a mutational target in a mismatch
repair-deficient cell type. Hum. Genet. 112, 117-23.
Wang, Z., Lo, H.S., Yang, H., Gere, S., Hu, Y., Buetow, K.H., Lee, M.P., 2003. Computational
analysis and experimental validation of tumor-associated alternative RNA splicing in
human cancer. Cancer Res. 633, 655-7.
Wiest, V., Eisenbarth, I., Schmegner, C., Krone, W., Assum, G., 2003. Somatic NF1 mutation
spectra in a family with neurofibromatosis type 1: toward a theory of genetic modifiers.
Hum. Mut. 22, 423-427.
Williams, V.C., Lucas, J., Babcock, M.A., Gutmann, D.H., Korf, B., Maria, B.L., 2009.
Neurofibromatosis type 1 revisited. Pediatrics 123, 124-33.
Wimmer, K., Roca, X., Beiglbock, H., Callens, T., Etzler, J., Rao, A.R., Krainer, A.R., Fonatsch,
C., Messiaen, L., 2007. Extensive in silico analysis of NF1 splicing defects uncovers
determinants for splicing outcome upon 5' splice-site disruption. Hum. Mutat. 28, 599-612.
Wu, R., López-Correa, C., Rutkowski, J.L., Baumbach, L.L., Glover, T.W., Legius, E., 1999.
Germline mutations in NF1 patients with malignancies. Genes Chromosomes Cancer 26,
376-80.
Wu, B.L., Schneider, G.H., Korf, B.R., 1997. Deletion of the entire NF1 gene causing distinct
manifestations in a family. Am. J. Med. Genet. 69, 98-101.
Wu, B.L., Austin, M.A., Schneider, G.H., Boles, R.G., Korf, B.R., 1995. Deletion of the entire
NF1 gene detected by the FISH: four deletion patients associated with severe
manifestations. Am. J. Med. Genet. 59, 528-535.
99
Xu, G.F., O'Connell, P., Viskochil, D., Cawthon, R., Robertson, M., Culver, M., Dunn, D.,
Stevens, J., Gesteland, R., White, R., et al., 1990. The neurofibromatosis type 1 gene
encodes a protein related to GAP. Cell 62, 599-608.
Xu, H., Gutmann, D.H., 1997. Mutations in the GAP-related domain impair the ability of
neurofibromin to associate with microtubules. Brain Res. 759, 149-52.
Yue, P., Melamud, E., Moult, J., 2006. SNPs3D: Candidate Gene and SNP selection for
Association Studies. BMC Bioinformatics 7, 166.
Yue, P., Moult, J., 2006. Identification and analysis of deleterious human SNPs. J. Mol. Biol.
356,1263-74. Epub 2005 Dec 27.
Zatkova, A., Messiaen, L., Vandenbroucke, I., Wieser, R., Fonatsch, C., Krainer, A.R., Wimmer,
K., 2004. Disruption of exonic splicing enhancer elements is the principal cause of exon
skipping associated with seven nonsense or missense alleles of NF1. Hum. Mutat. 24, 491-
501.
Zhou, J., Yu, Q., Zou, T., 2008. Alternative splicing of exon 10 in the tau gene as a target for
treatment of tauopathies. BMC Neurosci. 9(Suppl 2), S10.
Zhu, C., Saberwal, G., Lu, Y., Platanias, L.C., Eklund, E.A., 2004. The interferon consensus
sequence-binding protein activates transcription of the gene encoding neurofibromin 1. J.
Biol. Chem. 279, 50874-85.
Zhu, H., Hinman, M.N., Hasman, R.A., Mehta, P., Lou, H., 2008. Regulation of neuron-specific
alternative splicing of neurofibromatosis type 1 pre-mRNA. Mol. Cell Biol. 28, 1240-51
BIOGRAPHICAL SKETCH
Rebecca Loda-Hutchinson grew up in northern Nevada, where her interest in genetics
began in her high school biology classes. After graduating valedictorian of the class of 2000, she
attended the University of Nevada, Reno, majoring in biology. During her time at UNR, she
participated in several summer research opportunities, the first of which was in the laboratory of
Dr. John Postlethwait at the University of Oregon, studying armor formation in threespine
stickleback. The second was in the laboratory of Dr. Lee Weber and Dr. Eileen Hickey at UNR.
Dr. Weber was her honor’s thesis advisor, overseeing her investigation of using SNPs in the
HSP30 gene to detect hybrid populations of cutthroat trout. She graduated Summa cum Laude
with a bachelor’s degree in biology in 2004. Rebecca’s love of learning and desire to add to the
knowledge of others had led her to apply to graduate school, and, in the fall of 2004, she began
her studies at the University of Florida. While all her previous research involved fish, Rebecca
was excited to make the transition to studying human genetics, and joined the laboratory of Dr.
Margaret R. Wallace, where her doctoral research focused on Neurofibromatosis 1. Now that
she has graduated with her PhD, Rebecca isn’t sure which direction her career path will take, but
is looking forward to having a little more time to spend with her husband, Lance, and their two
rescue kitties.
![Comorbid Behavioral Problems and Psychiatric …...predispositiontowards anxiety disorders in children with NF1 [17]. The prevalence of anxiety disorders in the adult NF1 population](https://static.documents.pub/doc/80x56/5f0a52c87e708231d42b176c/comorbid-behavioral-problems-and-psychiatric-predispositiontowards-anxiety-disorders.jpg)
