|UFDC Home||myUFDC Home | Help|
This item has the following downloads:
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 st arted me on this path many years ago
4 ACKNOWLEDGMENTS First, I need to thank all the families and individuals who participated in these studie s; 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 wa nt to thank the members of my supervisory committee, Dr. Daniel Driscoll, Dr. Keit h 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 a lways having an answer for me. I also want to thank Dr. Michele Tennant, Dr. Pauline Ng (Fred Hut chinson 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 encourage ment 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 t hrough 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 PATHOGENI CITY.......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 ex amined......45 4-1 Summary of computational missense prediction results for 5 data sets (4 co ntrols and 1 unknown)........................................................................................................................63 A-1 Results of CpG C to T mutation screen using TaqI 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 us ed.........78
8 LIST OF FIGURES Figure page 1-1. Diagram of neurofibromins putative domains..................................................................18 2-1. Methylation-sensitive restriction digest protocol...............................................................29 2-2. Visualization of methylation-specific restriction digest anal ysis......................................30 2-3. Results of bisulfite sequencing..........................................................................................30 2-4. TaqI 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 SciencesGenetics 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 vari able expressivity and is clinically diagnosed using seven diagnostic criteri a, of which the key features are caf-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 alle le in specific cell types. To gain insight into mutation mechanisms in NF1, I purs ued three projects. First, I analyzed the rate of somatic C to T mutations at four hotspots for ger mline mutations. The methylation status of these sites in somatic cells makes them suscept ible to C to T transitions; however no such mutations were identified in 123 neurofibromas. Next, t he alternative splicing of exon23a, the inclusion of which reduces neurofibromins RASGAP function, was examined in various tissue types and tumors. Transcripts containi ng 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 sinc e mutations can not be tested functionally. No program was 100% accurate, but each had advantages in differ ent situations. This work contributes to the knowledge in NF1, toward a goal of targeted t herapies 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 ce ll 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 tra vel to the nucleus (Vandenbroucke et al., 2004) and may localize to different cell compartme nts based on stage of development and cell type (Roudebush et al., 1997; Kaufmann et al., 2002; Malminen e t al., 2002; De Schepper et al., 2006). Additionally, four isoforms caused by alternative spl icing 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; G utmann 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 (DAngelo et al., 2006), the most well-characterized functional domain of neurofibromin is a 360 amino acid RAS-GTPase activating protein (GAP)-related domain (GRD) (Balle ster 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 inac tivation decreases downstream RAS signaling, and thereby down-regulates cel l proliferation and the inhibition of apoptosis. Since increased RAS signaling is associated with tumorig enesis, neurofibromins inhibitory activity, as well as its genes adherence to Knuds ons 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 t his 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, mutat ions 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, a nd 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 medica l 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 physica l manifestations of NF1 are numerous and vary greatly in severity among indivi duals (variable expressivity) (Carey and Viskochil, 1999). In addition, even within a family, phenoty pic 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 diagnosi s (reviewed by Gutmann et al., 1997). These seven diagnostic criteria are: six or more caf-au -lait macules (must be 0.5 cm or larger before puberty, 1.5 cm or larger after puberty), neurofibroma s (benign Schwann cell tumors), optic pathway tumors (benign pilocytic astrocytomas), t wo or more Lisch
13 nodules (benign hyperpigmented lesions of the iris that look like freckles), skeleta l 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 the y meet any two or more of these criteria. Although molecular genetic testing is availabl e, these criteria remain important in diagnosis as many patients do not have the expensive genetic analysi s done (McClatchey, 2007). NF1 is a progressive disorder, and these phenotypic features may arise or increase unpredictably over time. Caf-au-lait macules are often the ea rliest phenotype to develop, usually present by 2 years of age. Tibial dysplasia and skin fold fre ckling 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 the se criteria, patients with NF1 are also at increased risk for additional complications, in cluding learning disabilities, short stature, scoliosis, renal artery stenosis, hypertensi on, macrocephaly, and increased risk of certain malignancies, such as rhabdomyosarcoma, pheochrom ocytoma, and myeloid leukemia (reviewed by Gutmann et al., 1997). Of the main phenotypic characteristics, neurofibromas generally cause t he most trouble and discomfort to the patients. Discomfort may include itching, pain, and tenderness a round areas of tumor growth, and furthermore patients can suffer socially due to disf igurement (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). Schw ann cells are one of several cell types that make up the peripheral nerve sheath, which surrounds t he 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, producin g the myelin used to insulate the axon. Disorganization and an increase in all peripher al nerve sheath
14 cell types are seen in neurofibromas (reviewed by Cichowski and Jacks, 2001). Thus, t hese tumors contain primarily clonal, expanded (and somatically mutated, as descri bed below) Schwann cells, but also contain other cell types normally associated with peri pheral 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 associa ted with nerve twigs, close to or at the surface of the skin. Some patients develop thousands of thes e tumors while others have few or none. They rarely grow beyond one centimeter in diam eter and are not known to become malignant. In contrast, plexiform neurofibromas develop from Schw ann cells associated with larger nerves, are generally deeper in the body, are large r 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 masse s in the thorax (Tonsgard et al., 1998). In addition to the risk that a plexiform tumor could grow quit e large and disrupt bone and organs, 10-15% of plexiform neurofibromas will transform int o malignant peripheral nerve sheath tumors (MPNSTs), presumably by the accumulat ion of genetic and epigenetic changes (Ferner and Gutmann, 2002). Unfortunately, MPNSTs ofte n become metastatic and have a poor prognosis unless removed completely by surgery. How the transformation from benign plexiform neurofibroma to MPNST occurs is unclea r. The development of benign neurofibromas is somewhat better understood and the initia ting event appears to follow Knudsons two-hit hypothesis of tumor suppressors (1971). Because of the dominant nature of NF1, the majority of patients are constitutionally heteroz ygous 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 func tion 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 Schwa nn cell then undergoes clonal expansion, dividing inappropriately, even without axonal contact. Simil arly, 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 e t al., 2006), and caf-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 onse t, high tumor loads, specific facial features, and mental retardation (Kayes et al., 1992, 1 994; 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 (Lzaro et al., 1996; Valero et al., 1997; Rasmussen et al., 1998; Upadh yaya et al., 1998; Lopez-Correa et al., 2001; Kluwe et al., 2004). These deletions are media ted by unequal crossover meiotic events in the germline at mis-aligned repetit ive regions. Both the 1.2 and 1.4 Mb versions of the microdeletion share a 5 repeat sequence containing a pseudogene o f JJAZ1 called JAZFP (Forbes et al., 2004). The 1.2 Mb deletions 3 end is in the JJAZ1 gene, whereas the 1.4 Mb deletions 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 pati ents hemizygous for these genes. These deletions can also occur somatically, ea rly 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 feat ures. 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 t hen, it has become clear that in most cases the allelic imbalance is due to mitotic reco mbination, where a region of the mutant chromosome replaces that of the wild-type chromosome (Se rra et al., 2001a). Loss of heterozygosity has been found in 10-40% of tumors analyzed (Rasmussen et a l., 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 li mited, and there is still little known about the somatic mutation spectrum (Rasmussen a nd 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 complexi ty 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 als o 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 understa nding of the nature and frequency of somatic mutations will provide important information about ri sk factors, disease progression and tumorigenesis, as well as help elucidate the pathwa ys 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. Chapte r 2 discusses the somatic rate of C to T transitions at four sites known to be hotspots for this mutation typ e in the germline. Chapter 3 examines alternative splicing of exon23a in NF1-rela ted tumors, its expression level in various cell types, and its relationship to RNA editing. Chapt er 4 addresses the need for reliable ways to predict the effects of missense mutations (bot h somatic and germline) on neurofibromin function, and evaluates several computational methods that make such predictions.
18 Figure 1-1. Diagram of neurofibromins putative domains. Alternatively s pliced 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-methylcytos ine (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 dinucleoti des not in CpG islands are methylated (Cooper and Krawczak, 1993). 5-methylcytosine c an be associated with transcriptionally inactive regions, especially if such m ethylation 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 r esidues 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 t hymine (a C to T transition). There are mismatch repair enzymes to recognize and repai r both G:U and G:T mispairs, but G:U mispairing is more efficiently repaired, in part becaus e 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 repair ed 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 transit ions to become fixed in the cells DNA. These base changes can be neutral (e.g. in non-coding DN A, or a silent coding-region substitution), or they can cause errors in splicing or coding regi ons 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 muta tion has been repeatedly identified as a germline mutation in unrelated neurofibroma tosis 1 (NF1) patients, accounting for 1-2% of mutant alleles. This CpG C to T mutation, R1947X (CGA t o 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 (Kr kljus et al., 1998; Fashold et al., 2000; Messiaen et al., 2000). Of these, the majority that we re 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 an d Migeon, 1990). In contrast, point mutations at CpG sites account for approximately 50% of the reporte d 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 s trong 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 m utation 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 s amples. Only 142 mutations of any type were identified out of 372 tumors. These studies scree ned 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 si te (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 sinc e 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 te st 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 fre quencies of somatic mutations such as C to T transitions will be useful for understanding genetic chan ges
22 contributing to NF1 tumors. In this work, I will test for somatic presence of eac h 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 4C. 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 detail s are given in Table 2-1. Diluted DNA samples were used as template (50-100 ng) in poly merase chain reactions (PCR) under standard conditions. These exon PCR products were then dige sted using the TaqI restriction enzyme in 25 l reactions incubated at 65C for 2 hours, with more enzyme added after 1 hr. The sequence recognized and cut by TaqI spans the CpG site of interest in each exon, and any changes in the sequence will result in failure of t he enzyme to cut. These digest reactions were visualized using ethidium bromide staining afte r electrophoretic separation on 8% native polyacrylamide gels. With this assay, germline he terozygotes 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 TaqI and methylsensitive digestion for each exon. Numbers of mutation-positive and mutation-ne gative 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 c ytosines were methylated in normal somatic Schwann cells, to ensure that they could be a ta rget for 5meC deamination. This was first done using methyl-sensitive restriction digest and then bisulfite genomic sequencing. A schematic of the methyl-sensitive restriction dige st 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 65C; AvaI f or exon22, incubated at 37C). These enzymes recognize and cut the sequence spanning the CpG sites of interest, but cut only if the cytosine residues are unmethylated. T he sequence surrounding the CpG site of interest in exon10a is 5GGTTGAACTT CG AAATATGTTT 3; in exon41 the sequence is 5TGAAGAAGTT CG AAGTCGCTGC 3; in exon23.2 the sequence is 5CCCTCAACTT CG AAGTGTGTGC 3. These three sites are cut by BstBI, which recognize s and cuts the sequence 5TT/CGAA 3. The sequence surrounding the CpG site of interest in exon22 is 5TGAACTAGCT CG AGTTCTGGTT 3. This site contains the recognition sequence of AvaI, which is 5C/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 a s untreated DNA. The primers listed in Table 2-1 were used in these reactions. This sam e methylation analysis was carried out on control samples as well as samples from a plexiform neurof ibroma. As a control, PCR samples were set up using the same samples, and these reactions wer e digested with TaqI restriction enzyme as described above. The TaqI site (TC/GA) is within the recognition sites of the methyl-sensitive enzymes, and is cut by TaqI whether the C of interest is methylated or not. The presence of digested product in these reactions will confi rm that the recognition sites are not mutated in the samples tested, and are therefore ab le to be digested by
24 the methyl-sensitive enzymes. All digestion products were visualized usi ng 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 a s an independent measure of somatic methylation. Genomic DNA from cultured normal hum an 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 (5GTTAGAATTATTAGAGAGTTTTGAG 3) and 23-2MER (5ATAATCTCTAACTATAAACATACCTAATA 3) with an annealing tempera ture of 54C. The sequence of these primers was determined assuming the conversion of unmethyl ated cytosine to thymine after sodium bisulfite treatment. These primers ampl ify 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 manufacturers protocol. The cells were plated on L B-ampicillin agar plates with IPTG and X-gal, and incubated at 37C overnight. Bacterial c olonies (clones) positive for the insert were identified by blue/white selection, followed by P CR 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 a nd 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 analy sis. 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 unmet hylated, these enzymes will cut the genomic DNA, and there will be no PCR product. If methylation is not c omplete at a site, there will be uncut band visible after separation by electorphoresis on a PAG E gel, but the concentration should appear less. As a control, TaqI digest reactions were also set up, to ensure that the sites of interest in these samples did not contain any mutations that would pre vent either enzyme from cutting, regardless of methylation status. Representative gel s from these experiments can be seen in Figure 2-2. All four CpG sites are free of muta tion in these samples, as they can be digested by TaqI (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 rea ction (middle lane for each sample), it appears that the cytosine at each of the four CpG si tes 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 t he 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 sodi um bisulfite, unmethylated cytosines are converted to uracils by deamination; uracil is not nor mally present in DNA (rather it is seen in place of thymine in RNA) and during PCR amplifica tion, thymine takes the place of the uracil, leading to a C to T change at the unmethylated sites. T he cloning of a sodium bisulfite treated PCR product into a vector allowed for the isolation of a si ngle allele per bacterial colony. Sequencing was done to reveal the relative number of cytosi nes versus thymines at the CpG site of interest, allowing an estimate of the percent m ethylation 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 unde rlined in the untreated sequence for comparison. All 5 clones analyzed contained a cytosine a t the CpG dinucleotide of interest (indicating that it is methylated), while all other c ytosines in the fragment had been replaced by thymine. The agreement between the methylation specific r estriction digest results and the bisulfite sequencing results for the exon23.2 site of intere st suggests that the four germline mutation hotspots are all heavily methylated in somatic t issue. 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 am plified with each of the four primer sets listed in Table 2-1 and then subjected to restrict ion digest by TaqI. All 4 CpG sites were analyzed in 117 samples, and 1 to 3 of the CpG sites was analy zed in the remaining 29 (due to lack of PCR amplification by on or more primer pair). Acr oss all the samples, 531 CpG dinucleotides were analyzed for C to T transitions. Figure 2-4 s hows 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 pr oducts). While this analysis was used to screen for somatic CpG C to T mutations, it a lso revealed two previously-identified constitutional mutations, one at the exon10a CpG site and the ot her at the exon23.2 (Figure 2-2) CpG site. These were the only mutations, germline or somat ic, 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 5methyl cytosine (5mC) (Razin and Riggs, 1980). The percent of isolated CpG sites that are methylate d is likely higher,
27 as this statistic includes all CpG dinucleotides, both isolated and those found in CpG isl ands, 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 woul d 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 nor mal 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 cell s, which are typically quiescent unless stimulated to divide by injury, or occasional divis ions to keep up with nerve growth. Interestingly, no such mutations were identified in 146 tumors analy zed at the CpG germline mutation hotspots. All four of these CpG sites have been previously identified as germline mutati on hot spots with a 1-2% recurrence rate (arginine-to-stop), and the estimated combined ger mline mutation rate at these four sites is approximately 7% (they account for 7% of all NF1 germline mutations) based on our labs data and published comprehensive studies from other labs. Thus, if the somatic mutation rate was equivalent, I would have expected to see approximate ly 37 total mutations (7% of 531 sites analyzed) among the four sites in the 146 tumors. Since no som atic mutations were identified, it appears that the rate of somatic CpG C to T trans itions is very low in neurofibromas, (at least at these sites), despite the presence of a methy lated 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 muta tion
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 m utations in neurofibromas compared to other tumor types may be due to differences in thei r ability to repair this type of mutation. As many cancers exhibit mutations in DNA repa ir pathways, it may also be that more malignant tumors are more susceptible to C to T transition muta tions, (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 cytogene tic 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 (Wall ace 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 fre quent. This may be pointing to basic differences in these tumor types for frequency of 5meC deaminati on and/or robustness of excision repair. Understanding the frequencies and mechanisms of CpG somatic mutation may help pre dict whether certain individuals or tumors are more at risk for these, or lead to a speci fic 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 e t 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 Taq I 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) TaqI 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 di gest analysis. a: normal human Schwann cell culture; b and c: non-NF1 patient blood; d: plexiform neurofibroma DNA. GTTAGAA CC AT C AGAGAG CC TTGAGGAAAA CC AG C GGAA CC T CC TT C AGATG A C TGAAAAGTT C TT CC ATG CC AT C AT C AGTT CC T CC T C AGAATT CCCCCC T C A A C TT CG AAGTGTGTG CC A C TGTTTATA CC AGTTTATA CC AGGTATG C TTA C AG TTAGAGATTA C Figure 2-3. Results of bisulfite sequencing. The sequence given is the untreat ed 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 sit e 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. TaqI 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 p atient 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 t ranscripts from a single gene. This usually affects the coding region in between invari ant translation start and stop sites, but this phenomenon can also produce transcripts with different transc ription or translation start and/or end sites. Alternative splicing is often regulat ed 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, eliminati on of the resulting proteins function, changes in the resulting proteins function, creation of a new function of the protein, and effects at the RNA level, such as transcript stability or efficiency o f 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 cance rs (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 is oform 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 splice d 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 inserti on of 21 amino acids in neurofibromin, and is known as the Type II isoform. These additional amino acids ar e
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 is oform 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 cell s expressing mostly the Type I isoform (lacking exon23a). This was shown to be due to ten-f old 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). I t has been suggested that in addition to altering or eliminating known functions of neurofi bromin, 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 a nd 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 inci dence of cognitive deficits, relative to the exon31 knockout mouse (which has no cognitive impairment). It also appears that this isoforms is developmentally signi ficant, 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 a nd chick, as well as differentiating human cell types (Nishi et al., 1991; Baizer et al., 1993; G utmann 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 transcr ipt, present at levels greater than or equal to Type I in most adult tissues, with the exception of the adul t central nervous system (Suzuki et al., 1992; Uchida et al., 1992; Teinturier et al., 1992; Baiz er et al.,
34 1993). Two of these studies also indicated that in various non-neurofibromatosis 1 (NF 1) associated cancers the expression of Type II mRNA is preferentially inc reased (Suzuki et al., 1992; Teinturier et al., 1992). Ogata et al. (2001) found that a particularly malig nant 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). T he other includes exon48a, which is a muscle-specific 18 amino acid insertion near the Cterminal 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 N F1 tumorigenesis. In addition to these well-studied isoforms, there have been several reports of mul tiple 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 norma l 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, im pacts 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 the se novel isoforms have been shown to occur in a tissue specific manner, implying that they may be funct ionally 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 transcripts cytosine at bp 3916 of the mRNA to form a uracil. Thi s nucleotide is within the region encoding the neurofibromin GRD. It is unclear wh at effect this premature stop codon may have on the cells that express the edited mRNA, but if trans lated 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, pote ntially 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; Mukhopadhy ay et al., 2002). Cappione et al. (1997) observed a correlation between increased invasivenes s of tumors with increased levels of mRNA editing, with malignant tumors having highe r levels of editing than plexiform neurofibromas, which in turn had higher levels of editing than der mal 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). Mukhopadh yay et al. (2002) evaluated RNA editing in malignant peripheral nerve sheath tumors (MPNS Ts) 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%). Tumor s that exhibited this higher, reproducible level of mRNA editing at position 3916 also exhibi ted
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 neurofibr omin function by leading to higher levels of activated RAS. These data were suggested to su pport a connection between RNA editing, alternative exon23a splicing, and tumorigenesis. The ex tent of these post-transcriptional modifications in NF1 tumorigenesis has yet to be validat ed by multiple independent groups. I examined alternative splicing of exon23a and mRNA editing in our s et 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 furthe r clarify the roles of these modifications in NF1 tumorigenesis and provide potential new avenues of trea tment, 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 sample s (3 normal Schwann cell, 4 dermal tumor cultures, 9 plexiform tumor cultures, 2 immortalized ple xiform neurofibroma cell lines) using the Trizol reagent and manufacturers protocol (Invitrogen), and stored at -80C. Reverse transcription reactions were carried out using Superscr ipt II reverse transcription kit (Invitrogen) and random hexamer primers. The resulting cDNA s 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 a mplified, one each from Type I and Type II mRNA (303 and 366 bp, respectively). The two isoforms wer e separated by electrophoresis on 8% native polyacrylamide gels and visualize d by ethidium bromide staining. Based on visual observation of band intensities, the relative ra tio of Type I to
37 Type II NF1 mRNA was noted (I<
38 and how that compared to the literature. A summary of the results of this analysis i s 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 grea ter levels then Type II in 7/7 samples, with the majority showing Type I as the pre dominant 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 e qual levels of the two isoforms, and lanes 5-7 again show more Type I than Type II trans cript. In leukocytes from seven NF1 patients, there was more variability between sam ples. Two out of seven showed Type II as the predominant transcript, three out of seven had approxi mately equal amounts of the two isoforms, and the remaining two had more Type I transcript tha n Type II. Three cultures from normal human Schwann cells were analyzed. All three of thes e cultures contained Type II mRNA as the main transcript, with one culture exhi biting 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 der mal tumors were analyzed for their exon23a splicing patterns. Of the 21 primary de rmal neurofibroma samples, there were two tumors from which corresponding cultures w ere 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 ha ving 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 transcri pt, and lane 6 is UF80T2 (which has no detectable Type I transcript). Twenty-five plexiform neurofibromas and nine cultures derived from plexiform t umors were also analyzed for exon23a splicing. Of the nine cultures analyzed, the cor responding primary tumor was also analyzed for four. The majority of primary plexiform t umors (23/25) contained predominantly Type II transcript, three of which had very little Typ e I in comparison, and two having only trace amounts of Type I. Of the remaining two tumors, one had about equ al 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 pre dominant transcript, with 3 having barely detectable levels of Type I transcript. T he remaining culture, UF469Tc, contained only Type II transcript and was also selected for RNA editi ng analysis. Figure 3-1B shows the experimental results from representative prima ry plexiform tumors, and one plexiform-derived culture. Lanes 8-10, and lane 12 show samples with predominantl y 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 show s the experimental results for representative MPNSTs. Lanes 1 and 2 show samples wi th much more Type II then Type I transcript, while lanes 3 and 4 show samples with predominantl y Type II mRNA. One MPNST, SNF94.3, had previously been analyzed for exon23a splicing level s, and was found to contain only Type II transcript. This sample was also chosen for RNA edi ting analysis. Analysis of RNA Editing Three samples were selected for RNA editing analysis based on their ex on23a splicing patterns. All three samples contained only Type II transcript, based on polyacryl amide 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 cl oned 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 w ere undergoing RNA editing at C3916. Thirty-five (UF469Tc) to Forty (SNF94.3 and UF80T2) clones were anal yzed for each sample of interest. This number was chosen based on the levels of RNA editi ng previously detected in MPNSTs lacking Type I transcript (3-12%) (Mukhopadhya y et al., 2002). Even at the lowest levels previously seen, at least one in forty clones would be expe cted to contain an edited cDNA fragment. No RNA editing was seen in any of these tum or samples, despite previous reports of increased levels of RNA editing in tumors containing predom inantly 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 i nterest. 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 syste m, with all the samples (7/7) showing Type I transcript at equal or greater levels com pared 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 deviati on 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 a l., 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 aff ects 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 pres ence of an NF1 gene mutation shouldnt affect relative ratios of Type I to Type II transcript. Y et 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 ma jority 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 pa tient 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 neurofi broma samples. There was no significant difference in these relative ratios bet ween 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, a ll contained more Type II mRNA then Type I. Observation of a relative (but not dramatic) predomi nance of Type II transcript suggests a relatively reduced RAS-GAP activity in those cell s, which would be inferred for most normal adult tissues, including, as I have shown here, Schwann ce lls, based on their Type I to Type II ratio. However, many of the NF1 tumors had a profile of T ype 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 exami ned 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 der mal neurofibromas and their corresponding cultured Schwann cells, and four were plexiform neurofibromas a nd their corresponding Schwann cell cultures. All 6 primary tumors analyzed containe d predominantly Type II mRNA, and there was no obvious change in the relative rat io of the two isoforms compared to the corresponding cultures. The result of one of these compa risons is shown in Figure 3-2 (plexiform neurofibromas). These results suggest that t here is an inherent Type I:Type II control mechanism in these tumor Schwann cells that is not susc eptible to influences of tissue culture, and it appears that the relative ratio of Type I to T ype 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 analyze d here showed Type II as the predominant transcript and one (SNF94.3) that had been previous ly analyzed showed complete loss of the Type I transcript by ethidium bromide visua lization. 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 wer e 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 thei r MPNSTs of interest at 3-12%. If RNA editing occurred to a similar extent in all tum ors 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 simpl y 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 m inimal. 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 t ranscript 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 on e of the germline NF1 microdeletions, but no somatic mutation has been found in the remaining allele despi te sequencing the entire mRNA open reading frame and the 3 UTR. This raises the inte resting possibility that the alternative splicing in this tumor substitutes as a mutat ion, producing a hypomorphic allele. The cells thus would lack sufficient NF1 RAS-GAP activi ty to prevent tumorigenesis. This suggests that there may be a threshold of neurofibromin ac tivity 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 chan ges (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 i n downstream regulatory effects (Schmegner et al., 2005; Ling et al., 2006; Martinez and Sc hackert, 2007; Shelton et al., 2008; Bartels and Tsongalis, 2009). The latter is evidenced by studie s 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 effe ct on NF1 expression levels (Zhu et al., 2004). The observation of different Type I to Type II ratios in two dermal neurofibromas f rom 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 type s 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 Typ e I. A) leukocytes from 7 non-NF1 patients. B) Primary tissue samples f rom dermal (lanes 1-7) and plexiform (lanes 8-12) tumors. C) Cultured (left two l anes) and primary tissue (right two lanes) from MPNSts. Figure 3-2. Comparison of alternative splicing of exon23a in primary plexifor m tumors v cultured Schwann cells from the same tumors. Tumor culture sample is loade d 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 PATHOGENI CITY Introduction Missense mutations are single base substitutions that result in an altered mR NA 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 (r eviewed by Thomson and Wallace, 2002). Missense mutations a priori may be pathogenic or represent neutral polymorphisms. For this reason, determining the pathogenicity of such mutat ions 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: i f the mutation is in an important structural of functional region of the gene; if the mutation alters a n 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 t he 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 i n 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 lit erature regarding a novel mutations effects, other methods must be used to predict the mutation's pathogeni city.
48 It has been estimated that over 50% of gene lesions known to cause hereditary disorder s in humans are missense mutations (Cooper et al., 1998), and predicting the effects of the se mutations on their corresponding proteins as well as their contribution to disease ca n be difficult as well. Previously computational-based methods were designed to aid in the unders tanding of the importance of specific amino acids in protein structure and function. These orig inal 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 protei n (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 pathogenicit y, 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 m utation, aided partially by the increasing amount of genomic sequence information avail able, 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 w as 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 ly sozyme, and in all cases was more accurate overall (Ng and Henikoff, 2001). As new programs ha ve been developed, their predictions have been compared to those from SIFT as a test of each pr ograms efficiency and accuracy. These programs include MAPP (Multivariat e 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 sma ll, 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 alt ernative exons) (DeClue et al., 1991; Marchuk et al., 1991). With its large size and complex na ture, neurofibromin was examined here as a robust test of the efficiency and accura cy 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 perfo rmance, as well as applicability to the analysis of neurofibromin (lack of structural data, e tc.). 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 Ana lysis 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 (OBrien 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 a l. 1997)) Data Sets Compiled from a literature search, the Human Gene Mutation Database (www.hgmd.o rg), and unpublished laboratory findings, the three data sets included missense mutations of known effect, mutations created by site-directed mutagenesis in neurofibromins isolated GAP Related Domain (GRD) and tested for RAS-GTP activity, and missense mutations of unknown e ffect. For mutations known to create cryptic splice sites, the amino acid residue that w ould normally result from the new codon was used for analysis. Sample size: known germline neut ral 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 repo rt. I reported these references to our best ability from the literature and the Human Gene Mutati on 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 assemble d a
51 multiple sequence alignment (MSA) from similar sequences found in the Swiss PROT/TrEMBLE database. In total, 43 sequences were aligned by the SIFT program, and saved for fu rther use. All - in the returned alignment were converted to X to allow proper analysi s by SIFT when the sequences were re-entered (Ng, personal communication). Not all sequenc es were considered at all mutation sites (due to high sequence similarity); SIFT re turned 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 thes e 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 cortecDNA, RIKEN full-length enric hed 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, individua l mutations were then entered in the Your SNP box for analysis, and data was retur ned regarding the sequences chosen for use in the prediction for that specific mutat ion site. The number and species of the chosen sequences varied between mutation sites, ranging from 5 to 8, and links were provided to each sequences NCBI entry. Finally, MAPP requires a strong ortholog set for best results, and I teste d 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 align ments (MSAs) by Clustal. In addition, a phylogenetic tree is required for MAPP analysis. Usi ng an MSA from Clustal, SEMPHY was employed to produce a phylogenetic tree, and both the MSA a nd 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 ret urned. 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 spl icing 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. P rediction parameters were: Organism: Human or other, search for both 5 and 3 splice sites, no r everse 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 (MA PP) are used for those missense mutations predicted to substantially alter protein structure a nd thus likely to be associated with disease. For SNPs3D and MAPP, Failed indicates inabili ty of the program to make a prediction because there was insufficient data about that residue in other s pecies. SIFT also returned some predictions as pathogenic with low confidence (low conf i n Table 4-1 and (!) in Table C-1). These were residues with prediction scores very cl ose to the tolerated/pathogenic threshold. The first two data sets, germline mutations with known effect and mutations fr om site-directed mutagenesis studies (control columns 1-5 of Table 4-1), were us ed to gauge the accuracy of the various programs in predicting the functional effects of misse nse 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, l acking 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 vari ed by program, as well as by the sequences used in making the predictions. When using t he 8 mammalian ortholog sequences, SIFT correctly predicted only 1/8 (12.5%) of the known neutr al 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 increas ed to 50%. The results were similar for MAPP: 1/8 (12.5%) of the known neutral germline mutati ons were correctly predicted when the 8 mammalian ortholog sequences were used. Howe ver, adding the additional sequences to the ortholog set (amphibian, chicken, fish, and insects) only increa sed MAPPs correct neutral predictions to 25%. The SNPs3D program itself chooses w hich 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 neutr al 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. W hen only using the 8 mammalian ortholog set, SIFT predicted all 18 mutations to be pathogenic; howev er, 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 onl y two
55 given a rating of low confidence. The differences seen between the various M APP analyses are not as great as those seen for the known neutral germline mutations. When MAPP used t he 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 accurate ly predicted 12/18 (66.7%) of the known pathogenic germline mutations. For the remaining two data subsets (site-directed mutagenesis data, contr ol columns 4-5, Table 4-1), the number of ortholog sequences used did not affect the accuracy of the pr edictions by SIFT and MAPP. SNPs3D called all of these mutations pathogenic. In the c ase of the 7 mutations found to be neutral in the site-directed mutagenesis RAS-GAP studie s, all three programs failed to predict any of them correctly. In contrast, all three prog rams correctly predicted a pathogenic effect for the 5 mutations found to alter RAS-GAP acti vity 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 les s likely to occur since some have 2 or more bases altered in the codons. Overall, among the 46 control mutations, SIFT (8 mammals) accurately pre dicted 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 cont rol 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 inaccur ate 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 sl ightly 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 predi cting the known effects of the control mutations. All three programs (all predictions under all conditions, not including failures) w ere 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 thre e 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 predicte d 37 or fewer to be pathogenic (inversely related to number of homolog sequences). In total, 14 differe nt 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 inac curate if the point change actually induced cryptic splicing. NNSPLICE predicted 2 of the 7 repor ted cryptic splice sites (29% accuracy). As missense mutations, SIFT predicted 2 of these as pat hogenic, 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 r un (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. Interesting ly, NNSPLICE found two mutations in the Unknown mutations set that it predicted to create cryptic s plice
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 program s. To express the accuracy of these tests in terms of specificity (probabi lity that a test calls a mutation neutral when it really is neutral) and sensitivity (probability tha t a test calls a mutation pathogenic when it really is pathogenic), data from control columns 1 and 2 (Table 4-1) we re 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 mamm als), 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 progra m. The user must assemble multiple sequence alignments and generate phylogenetic tre es from these MSAs prior to using MAPP to make predictions. The program returns results regarding subs titution 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 mutat ions of interest. However, if some of the orthologs used in the MSA have different numbe rs of encoded amino acids (or insertions/deletions relative to other orthologs), finding the am ino acid of interest
58 can be difficult due to altered numbering systems. SIFT can also provide a simil ar analysis for later use, and this output is simpler to search through then the MAPP results. SIFT a lso allows for several different options in the amount of preparation needed before analysis. T he 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 ana lysis. If reusing an MSA previously produced by SIFT, every "-" used as a place holder in partial sequence s 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 s imultaneously. If more then one mutation occurs at the same amino acid, they must be analyzed separate ly, 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 s equences to use in analysis. These will vary by position somewhat, but the user cannot choose which t o include or exclude. Though analysis of large numbers of mutations can be time cons uming since each mutation is entered separately, the SNPs3D results are quickly returne d since the program only aligned a small number of sequences (5-8) over a short sequence surrounding the s ite of your mutation. The results are also well explained through links, and easy to inter pret. 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 reliabi lity of predictions (additional known neutral mutation predicted correctly compared to mamm als-only
59 results). This is also the case when comparing the results for the control set s returned by SIFT when 8 mammalian sequences are used for analysis (where nearly all res ults were returned with low confidence) versus the 43 sequences gathered by SIFT from the SwissPRO T/TrEMBLE database (very few low confidence). The program that used the fewest s equences (5-8) was SNPs3D. Interestingly, for those mutations where SNPs3D was able to return a re sult, it appears to be the most accurate. However, a major weakness was that SNPs3D analysis f ailed 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 est ablished. 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 a ll 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 pa thogenic/deleterious, while only 13.1% of the predictions made for a known pathogenic mutation (germline and thos e from site-directed mutagenesis) were neutral. In the control sets (not inc luding splice errors), known neutral mutations were incorrectly predicted to be pathogenic 4 times as ofte n 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 the se programs have an inherently higher false-positive rate than false-negative rate (highe r 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 neutra l. It has been estimated that the majority of missense mutations in the human genome are slig htly 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 ph enotypes, the biochemical effect is enough to be picked up by these prediction programs. Such mutat ions 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 m utation alters RAS-GAP activity but isnt seen in NF1 patients, or is found in individuals w ho only meet one NF1 diagnostic criteria). Hypomorphic alleles would also confound the type of anal ysis 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 resul ts for my unknown data set. In my analyses of the control sets (across all programs and ana lyses), 88/262 pathogenic predictions were incorrect (33.6%). This number varies based on the spec ific 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 dat a 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 expect ation that novel NF1 mutations in NF1 patients are likely to be pathogenic. In the analyses o f the control sets, 66.7% of the neutral predictions made were incorrect. If this is approximate ly the rate of incorrect neutral predictions within the unknown set as well, it can be estimate d 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 analysi s. 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 missens e mutation. As these programs were not designed to replace functional studies, it is understandable that they seem to consider mutations pathogenic until proven neutral. SNP s3D had the highest control accuracy but also the highest failed rate. SIFT predic ted pathogenic mutations at the expense of a high false-positive rate. MAPP called some known pathoge nic mutations neutral, but also had false-positives. No single program stood out as super ior 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 res ults are consistent with SIFT data for other monogenic situations (sensitivity 80-90%), although SI FT 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 pr ograms, with greater faith in the results agreed upon by more than one analysis (Mat he et al., 2006; Chan et al., 2007; Valdmanis et al., 2008). These programs are often updated and so the accuracy m ay improve with newer versions. In addition, there are other programs becoming availabl e and refined (some of which are based on amino acid biochemistry rather than MSA), s uch as PolyPhen. When choosing a program (or set of programs) one should consider the nature o f 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 mutat ions, 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 als o found that a missense analysis should be complimented by a splice site analysis (computat ional or in the lab with RT-PCR) to attempt to find mutations that cause exon skipping or cryptic splici ng rather than an amino acid substitution. The data reported here will be useful for individuals cons idering computational methods for testing pathogenicity of missense mutations, particul arly in large genes such as NF1
63 Table 4-1. Summary of computational missense prediction results for 6 data se ts (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 he terogeneous phenotypes of NF1. Additional knowledge of the mutation mechanisms occurring both in the germline and somatic cells, as well as elucidating which mutation mechanis ms do not play a major role in NF1 progression, may contribute to developing targeted therapie s and better diagnosis. In this chapter, I reiterate the major findings of my work and discus s possible future directions. Somatic CpG C to T Mutations From my work described here, C to T transition mutations at CpG dinucleotides are cl early 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 sam e 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. Compa red 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 mor e malignant tumors, leading to a decreased ability of cells to recognize and correctly re pair 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 mutat ion 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 plexifor m 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 pathw ays. These data suggest that MPNSTs, with a higher occurrence of MSI (and likely repa ir 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 l arge cohort of MPNSTs. If future studies were able to identify C to T transitions at CpG sites in neurofibr omas, 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 indivi duals were found to be generally susceptible to such mutations, (e.g. less-than-ade quate DNA excision repair mechanisms, or exposed to mutagens), perhaps they could benefit from therapi es being designed to compensate for mutations causing premature stop codons, which are often ca used by C to T transitions at CpG sites. An interesting report described two siblings w ith some features of NF1 (caf-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 repa ir (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 thes e 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 r eplication) 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 star ts clonal cell expansion. There is no evidence for outright mutation at other genes in neurofibrom as. Conversely, perhaps second hits occur randomly throughout life and are followed shortly b y clonal expansion. Dermal neurofibromas rarely appear before adolescence, but pl exiform 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 unders tand. However, I saw no CpG mutations in large sets of both tumor types, so the occurrence of this mutat ion type does not appear different. CpG C to T transitions can occur in the germline (pater nal), or any time after the egg is fertilized throughout life. Knowing the timing of the s econd hit could also useful for future therapies/preventions. Alternative Splicing of exon23a While it appears that the alternative splicing of exon23a is developmentally si gnificant, the role that inclusion of exon23a might play in tumorigenesis is still unclear. Analy sis of RNA transcripts from multiple cell types provided an estimate of the level of inc lusion of exon23a occurring in these cells (the relative ratio of Type I to Type II trans cript). Leukocytes produce predominantly Type I transcript, with a slight trend towards increased leve ls of Type II compared to Type I mRNA in leukocytes from NF1 patients versus non-NF1 patients, a lthough 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 bet ween primary tumor samples and their corresponding cultures. There also does not appear to be a dif ference 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 wo uld be of interest to analyze samples from more male patients to rule out gender-specific tre nds in the alternative splicing of exon23a. Additionally, I found several cases where two tumors from the sa me individual had very different Type I to Type II ratios, indicating that these ra tios are tumor-, not individual-specific. Development of a real-time PCR protocol to distinguish leve ls 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 T ype I to Type II mRNA between primary tumors and their corresponding cultures, betw een tumor types, or between genders. However, this would require a large sample set since the rat ios would fall into a greater number of bins. Despite some differences in the amount of alternative splicing of exon23a between t umors, there is a trend toward increased levels of exon23a inclusion in NF1 tumors compared to nor mal Schwann cells, making this mechanism a potential target for therapy. Befor e such therapies can be developed it must be shown that having (virtually) only Type II transcript can subs titute as a somatic mutation. To help determine if the reduced GAP activity of Type II neur ofibromin 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 s ufficient 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 t herapies, to express only Type II transcript to determine if this causes any changes i n phenotype (development of tumorigenic characteristics). Additionally, both the trans and cisacting elements involved in controlling the splicing of exon23a must be identified. These el ements would be the targets of therapies designed to decrease the amount of exon23a inclusi on, most likely in Schwann cells if targeted correctly. Zhu et al. (2008) have identif ied 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 ca rried out to identify splicing elements in Schwann cells. Interestingly, mice l acking 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 incre ased 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 curre ntly 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 s plicing 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 transcr ipt in tumors in this study, in contrast to published results. I saw no evidence of any RNA edi ting in our samples, including MPNSTs. A larger study, repeating the analysis of the publi shed 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 e xhibit both loss of
69 the Type I transcript and RNA editing. My study used a qualification of comple te or nearly complete loss of the Type I transcript to identify samples to be evaluated for R NA editing, and since none was found, it appears that this shift in alternative splicing alone doe s not lead to increased RNA editing. Computational Analysis Comparison Recent advances in bioinformatics, as well as the vast amount of data being g enerated 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. Us ing the limited set of NF1 missense mutations with known effect to test the accuracy of these new predic tion programs, it appears that each may have its advantages in a given situation. It w ould be advantageous to use multiple programs when possible. These programs are already be ing 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 inc rease 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 TaqI 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 cultur e 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 met hods 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 De l no R1825W exon29 CGG-TGG Ars et al 2003 N failed Patho Patho (!) Del Del Del De l no A2058D exon33 GCT-GAT Mattocks 2004 N failed Patho Patho (!) Del Del Del D el 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 N SI no V1093M exon19b GTG-ATG Ars et al. 2003 C.S. Tol Tol Patho (!) Del Del Del NS I 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 D el 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 De l 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 AATATT 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, affe cts Microtubule binding, N= Neutral, and ?= Unknown. For results returned by programs Patho=Pathogenic, Tol= Tolerated, Del= Dele terious, NSI= No Significant Impairment. For results reported in the SIFT columns, (!)= a warning of low confidence in the predic tion 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 t he 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., Estivi ll, 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., Garca, J., Kruyer, H., Gaona, A., Lzaro, C., Es tivill, X., 2000. Mutations affecting mRNA splicing are the most common molecular defects in patients wi th 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 N f1 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 distribut ion 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 mammalia n 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., R obertson, 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 mutat ions. 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 disea se 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-pai r 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., Rgnier, 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 characte rization of the neurofibromatosis type 1 protein product. Proc. Natl. Acad. Sci. U S A. 88, 99148. De Luca, A., Bottillo, I., Sarkozy, A., Carta, C., Neri, C., Bellac chio, E., Schirinzi, A., Conti, E., Zampino, G., Battaglia, A., Majore, S., Rinaldi, M.M., Carella, M., M arino, B., Pizzuti, A., Digilio, M.C., Tartaglia, M., Dallapiccola, B., 2005. NF1 gene mut ations 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 analy sis 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 caf 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 mela nosomes. 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 gene s in human meiotic germ cells: implications for X chromosome inactivation, parental im printing, 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 processin g 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 gen es. 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. A m. 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., Nurnber g, 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 periphe ral 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 Pathoge nesis. 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-specifi c 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 g ene 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. Neurosurg ery 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. Cance r 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) associat es 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 Growt h Dif. 6, 315-323. Gutmann, D.H., Cole, J.L., Collins, F.S., 1994. Modulation of neurofibromatosis type 1 g ene 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 alternativelyspliced mRNA in the carboxy terminus of the neurofibromatosis type 1 (NF1) gen e is expressed in muscle. Hum. Mol. Genet. 2, 989-92. Gutmann, D.H., Boguski, M., Marchuk, D., Wigler, M., Collins, F.S., Ballest er, R., 1993b. Analysis of the neurofibromatosis type 1 (NF1) GAP-related domain by sitedirected mutagenesis. Oncogene 8, 761-9. Gutmann, D.H., Wood, D.L., Collins, F.S., 1991. Identification of the neurofibrom atosis 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 per formance liquid chromatography (DHPLC) for the mutational analysis of the neurofibromat osis 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. Neuroge netics 8, 121-30. Henikiff, S., Henikoff, J.G., 1992. Amino acid substitution matrices from protei n 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 wi th 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 p atients 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., Mller, 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 f ive 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 som atic 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., Maut ner, 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 tumor s. Hum. Mutat. 22, 420. Kluwe, L., Friedrich, R.E., Korf, B., Fahsold, R., Mautner, V.F., 2002. NF 1 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. Na tl. 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 peri pheral 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., Wallac e, 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 allel es are deleterious in humans: implications for complex disease and association studies. Am. J. Hum. G enet. 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 ba se 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 pr eference of G:T mismatch repair: incision at G:T, O6-methylguanine:T, and G:U mispai rs 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 i n 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. Gene t. 3, 361363. Leppig, K.A., Kaplan, P., Viskochil, D., Weaver, M., Orterberg, J., Stephens, K., 1997. Familial neurofibromatosis 1 gene deletions: cosegregation with distinctive facial fea tures 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 ty pe 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., OConnell, 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., War d, K., Friedman, E., Samowitz, W., Robertson, M., Bradley, P., McCormick, F., White, R., Caw thon, 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., Hpping, 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., Dooije s, 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 d uring 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 mur ine cell differentiation and proliferation. Gene 148, 245-51. Martinez, R., Schackert, G., 2007. Epigenetic aberrations in malignant gliomas: an ope n 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 c omparison of three sequence analysis based tools. Nucleic Acids Res 34, 1317-1325.
93 Mattocks, C., Baralle, D., Tarpey, P., ffrench-Constant, C., Bobrow, M., Whit taker, J., 2004. Automated comparative sequence analysis identifies mutations in 89% of NF1 patie nts 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 s ignificance 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 g ene 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., PericakVance, M.A., Vance, J.M., Wallace, M.R., 1999. Exon 10b of the NF1 gene represent s a mutational hotspot and harbors a recurrent missense mutation Y489C associated wit h 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 ty pe 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. m elanogaster. Cell 53, 229-36. Ng, P.C., Henikoff, S., 2001. Predicting deleterious amino acid substitutions. Ge nome 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 transcri pts related to neuronal differentiation. Oncogene. 6, 1555-1559. Nishi, T., Saya, H., 1991. Neurofibromatosis type 1 (NF1) gene: implication in neuroectode rmal differentiation and genesis of brain tumors. Cancer Metastasis Rev. 10, 301-10. O'Brien, K.P., Remm, M., Sonnhammer, E.L., 2005. Inparanoid: a comprehens ive 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 a nd 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., Ver, M.C., B attista, P., Frati, L., Caramia, F.G., Calvieri, S., et al., 1995. Alterations of microsatelli tes 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 trunc ation assay yielding identification of mutations in 73% of patients. J. Med. Genet. 35, 813-20. Peters, H., Hess, D., Fahsold, R., Schlke, M., 1999. A novel mutation L1425P in the G APregion of the NF1 gene detected by temperature gradient gel electrophoresi s (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 coa gulation 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 mutat ions at this residue and suppresses RAS2Val-19-activated phenotypes. Mol. Cell Biol. 14, 815 -21. Pros, E., Gmez, C., Martn, T., Fbregas, P., Serra, E., Lzaro, C., 2008. N ature and mRNA effect of 282 different NF1 point mutations: focus on splicing alterations. Hum. Muta t. 29, E173-E193. [Epub ahead of print] Purandare, S.M., Lanyon, W.G., Connor, J.M., 1994. Characterisation of inherite d 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., Lazar o, 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 t he 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, 604610. 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. Ge net. 14, 110-112. Serra, E., Rosenbaum, T., Nadal, M., Winner, U., Ars, E., Estivill, X., Lzaro, C., 2001a. Mitoti c 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 i n 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 e diting. 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., Zeillinge r, 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 missens e 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 sporadi c 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., B ernheim, A., 1992. The neurofibromatosis 1 gene transcripts expressed in peripheral nerve and neurofibr omas 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 tes ting 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., Pot ts, C., Griffiths, S., Ruggieri, M., von Deimling, A., Cooper, D.N., 2004. Character ization 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 T ype 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 mater nal origin and commonly associated with a learning disability, dysmorphic features a nd developmental delay. Hum. Genet. 102, 591-597. Upadhyaya, M., Maynard, J., Osborn, M., Harper, P.S., 1997. Six novel mutations in t he 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., Har per, P.S., 1995. Characterisation of germline mutations in the neurofibromatosis type 1 (N F1) 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 ori gin 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 spl ice 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, 87192. 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. G enes 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 i n 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 gen etic 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 defect s uncovers determinants for splicing outcome upon 5' splice-site disruption. Hum. Mutat. 28, 599-612. Wu, R., Lpez-Correa, C., Rutkowski, J.L., Baumbach, L.L., Glover, T.W. Legius, E., 1999. Germline mutations in NF1 patients with malignancies. Genes Chromosomes C ancer 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 seve re 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 abil ity 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., K rainer, A.R., Wimmer, K., 2004. Disruption of exonic splicing enhancer elements is the principal cause of ex on skipping associated with seven nonsense or missense alleles of NF1. Hum. Mutat 24, 491501. 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 neurofibrom in 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 genetic s began in her high school biology classes. After graduating valedictorian of the cl ass of 2000, she attended the University of Nevada, Reno, majoring in biology. During her time a t UNR, she participated in several summer research opportunities, the first of which wa s in the laboratory of Dr. John Postlethwait at the University of Oregon, studying armor formation i n threespine stickleback. The second was in the laboratory of Dr. Lee Weber and Dr. Eilee n Hickey at UNR. Dr. Weber was her honors thesis advisor, overseeing her investigation of using SNPs i n the HSP30 gene to detect hybrid populations of cutthroat trout. She graduated Summa cum Laude with a bachelors degree in biology in 2004. Rebeccas love of learning and desire to a dd 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 laborat ory of Dr. Margaret R. Wallace, where her doctoral research focused on Neurofibromatos is 1. Now that she has graduated with her PhD, Rebecca isnt sure which direction her career pa th will take, but is looking forward to having a little more time to spend with her husband, Lance, and their t wo rescue kitties.