Genetic and Clinical Investigation of Noonan Spectrum Disorders

ACTA

Digital Comprehensive Summaries of Uppsala Dissertations from the Faculty of Medicine 830

Genetic and Clinical

Investigation of Noonan Spectrum Disorders

SARA EKVALL

Dissertation presented at Uppsala University to be publicly examined in Rudbecksalen, Rudbecklaboratoriet, Dag Hammarskjölds väg 20, Uppsala, Friday, December 7, 2012 at 09:15 for the degree of Doctor of Philosophy (Faculty of Medicine). The examination will be conducted in English.

Abstract

Ekvall, S. 2012. Genetic and Clinical Investigation of Noonan Spectrum Disorders. Acta Universitatis Upsaliensis. Digital Comprehensive Summaries of Uppsala Dissertations from the Faculty of Medicine 830. 73 pp. Uppsala. ISBN 978-91-554-8511-5.

Noonan spectrum disorders belong to the RASopathies, a group of clinically related developmental disorders caused by dysregulation of the RAS-MAPK pathway. This thesis describes genetic and clinical investigations of six families with Noonan spectrum disorders.

In the first family, the index patient presented with severe Noonan syndrome (NS) and multiple café-au-lait (CAL) spots, while four additional family members displayed multiple CAL spots only. Genetic analysis of four RAS-MAPK genes revealed a de novo PTPN11 mutation and a paternally inherited NF1 mutation, which could explain the atypically severe NS, but not the CAL spots trait in the family. The co-occurrence of two mutations was also present in another patient with a severe/complex NS-like phenotype. Genetic analysis of nine RASopathy-associated genes identified a de novo SHOC2 mutation and a maternally inherited PTPN11 mutation. The latter was also identified in her brother. Both the mother and the brother displayed mild phenotypes of NS. The results from these studies suggest that an additive effect of co-occurring mutations contributes to severe/complex NS phenotypes.

The inherent difficulty in diagnosing Noonan spectrum disorders is evident in families with neurofibromatosis-Noonan syndrome (NFNS). An analysis of nine RASopathy-associated genes in a five-generation family with NFNS revealed a novel NF1 mutation in all affected family members. Notably, this family was initially diagnosed with NS and CAL spots. The clinical overlap between NS and NFNS was further demonstrated in three additional NFNS families. An analysis of twelve RASopathy-associated genes revealed three different NF1 mutations, all segregating with the disorder in each family. These mutations have been reported in patients with NF1, but have, to our knowledge, not been associated with NFNS previously.

Together, these findings support the notion that NFNS is a variant of NF1. Due to the clinical overlap between NS and NFNS, we propose screening for NF1 mutations in NS patients negative for mutations in NS-associated genes, preferentially when CAL spots are present.

In conclusion, this thesis suggests that co-occurrence of mutations or modifying loci in the RAS-MAPK pathway contributes to the clinical variability observed within Noonan spectrum disorders and further demonstrates the importance of accurate genetic diagnosis.

Keywords: RASopathies, Noonan syndrome, neurofibromatosis type 1, neurofibromatosis- Noonan syndrome, RAS-MAPK pathway, mutation

Sara Ekvall, Uppsala University, Department of Immunology, Genetics and Pathology, Rudbecklaboratoriet, SE-751 85 Uppsala, Sweden.

© Sara Ekvall 2012 ISSN 1651-6206 ISBN 978-91-554-8511-5

urn:nbn:se:uu:diva-183325 (http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-183325)

Till min underbara familj

Supervisors Marie-Louise Bondeson, Associate Professor Göran Annerén, Professor, M.D.

Dept. of Immunology, Genetics and Pathology Uppsala University

Uppsala, Sweden

Faculty opponent Göran Andersson, Professor

Dept. of Animal Breeding and Genetics Swedish University of Agricultural Sciences Uppsala, Sweden

Review board Tobias Sjöblom, Associate Professor

Dept. of Immunology, Genetics and Pathology Uppsala University

Uppsala, Sweden

Jovanna Dahlgren, Associate Professor, M.D.

Dept. of Paediatrics

The Queen Silvia Children’s Hospital University of Gothenburg

Gothenburg, Sweden

Margareta Dahl, Associate Professor, M.D.

Dept. of Women’s and Children’s Health Uppsala University Children’s Hospital Uppsala University

Uppsala, Sweden

Chairman Berivan Baskin, Ph.D., FACMG, FCCMG Dept. of Immunology, Genetics and Pathology Uppsala University

Uppsala, Sweden

List of Papers

This thesis is based on the following papers, which are referred to in the text by their Roman numerals.

I Nyström A.M., Ekvall S., Strömberg B., Holmström G., Thuresson A.C., Annerén G., Bondeson M.L. (2009) A severe form of Noonan syndrome and autosomal dominant café-au-lait spots – evidence for different genetic origins. Acta Paediatr, 98(4):693–8

II Ekvall S., Hagenäs L., Allanson J., Annerén G., Bondeson M.L. (2011) Co-occurring SHOC2 and PTPN11 mutations in a patient with severe/complex Noonan syndrome-like phenotype.

Am J Med Genet A, 155A(6):1217-24

III Nyström A.M.*, Ekvall S.*, Allanson J., Edeby C., Elinder M., Holmström G., Bondeson M.L., Annerén G. (2009) Noonan syndrome and neurofibromatosis type I in a family with a novel mutation in NF1. Clin Genet, 76(6):524-34

IV Ekvall S., Sjörs K., Jonzon A., Annerén G., Bondeson M.L.

Mutations in NF1 in families with neurofibromatosis type I and neurofibromatosis-Noonan syndrome. Manuscript

*Equal first authors

Reprints were made with permission from the respective publishers.

Additional publications by the author

1. Nyström A.M., Ekvall S., Berglund E., Björkqvist M., Braathen G., Duchen K., Enell H., Holmberg E., Holmlund U., Olsson-Engman M., Annerén G., Bondeson M.L. (2008) Noonan and cardio-facio- cutaneous syndromes: two clinically and genetically overlapping disorders. J Med Genet, 45(8):500-6

2. Nyström A.M., Ekvall S., Thuresson A.C., Denayer E., Legius E., Kamali-Moghaddam M., Westermark B., Annerén G., Bondeson M.L. (2010) Investigation of gene dosage imbalances in patients with Noonan syndrome using multiplex ligation-dependent probe amplifi- cation analysis. Eur J Med Genet, 53(3):117-21

3. Wittström E., Ekvall S., Schatz P., Bondeson M.L., Ponjavic V., Andréasson S. (2011) Morphological and functional changes in mul- tifocal vitelliform retinopathy and biallelic mutations in BEST1.

Ophthalmic Genet, 32(2):83-96

Contents

Introduction ... 15

The human genome ... 15

Human genetic variation ... 15

Disease-causing variants ... 17

Human genetic disorders ... 17

Monogenic disorders ... 17

Methods in disease-gene identification ... 19

Linkage analysis ... 19

Sanger sequencing ... 19

Restriction fragment length polymorphism (RFLP) ... 20

Multiplex ligation-dependent probe amplification (MLPA) ... 20

SNP arrays ... 21

Next-generation sequencing ... 21

The RAS-MAPK pathway ... 22

Activation of the RAS-MAPK pathway ... 22

Regulation of the RAS-MAPK pathway ... 24

Phosphorylation and dephosphorylation ... 24

Scaffolding proteins, phosphatases and inhibitors ... 25

Internalization and degradation of receptors ... 25

Histone modifications ... 25

Post-transcriptional regulation ... 25

Determination of signal specificity of the RAS-MAPK pathway ... 26

Signal strength and duration ... 26

Cross-talk with other pathways ... 27

Subcellular localization of components of the pathway ... 27

Cancer and the RAS-MAPK pathway ... 27

Drug development ... 28

RASopathies ... 29

Noonan and Noonan-like syndromes ... 30

Clinical description ... 30

Genetic description ... 31

Genotype-phenotype correlations ... 35

Neurofibromatosis type 1 ... 37

Clinical description ... 37

Genetic description ... 38

Genotype-phenotype correlations ... 40

Neurofibromatosis-Noonan syndrome ... 41

Clinical description ... 41

Genetic description ... 41

Genotype-phenotype correlations ... 42

Animal models and future treatments ... 42

Present investigations ... 44

Background ... 44

Aims ... 45

Paper I ... 45

Paper II ... 47

Paper III ... 50

Paper IV ... 52

Concluding remarks and future perspectives ... 54

Populärvetenskaplig svensk sammanfattning ... 58

Acknowledgements ... 60

References ... 62

Abbreviations

A Adenine or alanine

AKT v-akt murine thymoma viral oncogene

ARAF v-raf murine sarcoma 3611 viral oncogene homolog BRAF v-raf murine sarcoma viral oncogene homolog B1

C Cytosine or cysteine

CAL Café-au-lait

cAMP Cyclic adenosine monophosphate

CBL Casitas B-lineage Lymphoma protein/gene Cdc25 Cell division cycle 25 protein

cDNA Complementary deoxyribonucleic acid CFCS Cardio-facio-cutaneous syndrome

c-Fos v-fos FBJ murine osteosarcoma viral oncogene CNV Copy number variant

CR1-3 Conserved region 1-3

CS Costello syndrome

CSRD Cysteine/serine-rich domain

DECIPHER DatabasE of Chromosomal Imbalance and Phenotype in Humans using Ensembl Resources

DH Dbl homology

DNA Deoxyribonucleic acid DUSP Dual-specificity phosphatase EGF Epidermal growth factor

EGFR Epidermal growth factor receptor Elk1 E twenty-six-like transcription factor 1 EMH Extramedullary hematopoiesis

ENCODE Encyclopaedia of DNA Elements ERK Extracellular signal regulated kinase EVI2A Ecotropic viral integration site 2A EVI2B Ecotropic viral integration site 2B

F Phenylalanine

FPPS Farnesyl diphosphate synthetase

G Guanine or glycine

GAP GTPase-activating protein GDP Guanosine diphosphate

GEF Guanine-nucleotide-exchange factor

Grb2 Growth factor receptor-bound protein 2

GRD GAP-related domain

GTP Guanosine triphosphate

HD Histone domain

HGMD Human Gene Mutation Database HRAS v-Ha-ras Harvey RAS homolog

HuR Human antigen R

I Isoleucine

JMML Juvenile myelomonocytic leukaemia JNK c-Jun N-terminal kinases

KRAS v-Ki-ras2 Kirsten RAS homolog KSR Kinase suppressor of RAS

L Leucine

LCRs Low copy repeats

let-7 Lethal-7

LOD Logarithm of the odds

LOVD Leiden Open Variation Database

LS LEOPARD syndrome

MAP2K1/2 Mitogen-activated protein kinase kinase 1/2 MAPK Mitogen-activated protein kinase

Mb Mega bases

MEK1/2 Mitogen-activated protein kinase kinase 1/2 miRNA Micro ribonucleic acid

MKP-1 MAPK phosphatase 1

MLPA Multiplex ligation-dependent probe amplification MPNSTs Malignant peripheral nerve sheath tumours mRNA Messenger ribonucleic acid

mTOR Mammalian target of rapamycin MYST4/KAT6B K(lysine) acetyltransferase 6B

NCBI National Center for Biotechnology Information NCFCs Neuro-cardio-facio-cutaneous syndromes NF1 Neurofibromatosis type 1 or neurofibromin gene NFNS Neurofibromatosis-Noonan syndrome

NF-κB Nuclear factor kappa-light-chain-enhancer of activated B cells

NGF Nerve growth factor

NIH National Institutes of Health

NRAS Neuroblastoma RAS viral (v-ras) oncogene homolog

NS Noonan syndrome

NS/LAH Noonan syndrome with loose anagen hair OMIM Online Mendelian Inheritance in Man OMGP Oligodendrocyte myelin glycoprotein OPG Optic pathway glioma

ORF Open reading frame

PC12 Pheochromocytoma cell line 12 PCR Polymerase chain reaction PDGF Platelet-derived growth factor

PH Pleckstrin homology

PI3K-AKT Phosphatidylinositol 3-kinase-v-akt murine thymoma viral oncogene

PKA Protein kinase A

PLA2 Phospholipase A2

PP1C Catalytic protein phosphatase 1 subunit PTP Protein tyrosine phosphatase

PTPN11 Protein tyrosine phosphatase, non-receptor type 11 gene PUM2 Pumilio homolog 2

R Arginine

RAC1 RAS-related C3 botulinum toxin substrate 1

RAF1 v-raf-1 murine leukaemia viral oncogene homolog 1 RAS Rat sarcoma viral oncogene; HRAS, KRAS and NRAS RAS-MAPK RAS-induced mitogen-activated protein kinase

Rem RAS exchanger motif

RFLP Restriction fragment length polymorphism RT-PCR Reverse transcription-polymerase chain reaction

S Serine

SAPK Stress-activated protein kinase

Ser Serine

SH2 Src-homology 2 domain

SHOC2 Soc-2 suppressor of clear homolog

SHP2 Protein tyrosine phosphatase, non-receptor type 11 SNP Single nucleotide polymorphism

SOLiD Supported oligonucleotide ligation and detection SOS1/2 Son of sevenless homolog 1/2

SPRED1/2 Sprouty-related, EVH1 domain-containing protein 1/2 SPRY Sprouty homolog, antagonist of FGF signaling

T Thymine or Threonine

Tyr Tyrosine

UTR Untranslated region

Introduction

The human genome

Almost 60 years have passed since Watson and Crick discovered the struc- ture of deoxyribonucleic acid, DNA, in 1953. [1] They put forward the prin- ciple that the four nucleotide bases, adenine (A), cytosine (C), thymine (T) and guanine (G), pair up with each other in a specific manner, A to T and C to G, and form a complementary double helix; thereby, explaining how an organism is able to copy its DNA.

In 2004, another important step in understanding the human genome was taken, when a near complete sequence of the human genome was published.

[2] From this, we learned that the human genome is approximately three billion base pairs long and contains 20,000-25,000 protein-coding genes.

Recently, the Encyclopaedia of DNA Elements (ENCODE) project re- leased a number of publications, where regions of transcription, transcription factor association, chromatin structure and histone modification were sys- tematically mapped. Together, these researchers could assign biochemical functions for 80% of the genome, providing new insights into the organiza- tion of our genome and the mechanisms of gene regulation. This demon- strates that much more than just the protein-coding genes are of great im- portance for us. [3]

Human genetic variation

Although the genomes between two randomly selected humans resemble each other, a considerable number of differences exist between them. [4, 5]

These differences are called genetic variations and can vary in type and size, ranging from differences in single nucleotides to duplications of large seg- ments. A genetic variant present in more than 1% of the population is con- sidered to be a polymorphism. [6]

Two large projects, the International HapMap project and the 1000 Ge-

nomes project, have been initiated to identify genetic similarities and differ-

ences in humans. The International HapMap project started in 2002, with the

aim of genotyping and characterizing single nucleotide polymorphisms

(SNPs) and structural variations in large groups of individuals from different

geographical origins. [7] In 2007, the 1000 Genomes project was launched,

with the purpose of sequencing 1000 individuals from different populations using high-throughput sequencing technologies. [8]

One of the most common types of variation is SNPs. As the name suggests, a SNP is a difference, e.g. deletion, insertion or substitution, of one single nucleotide. They occur on average once every 100 to 300 bases, although the density can vary throughout the genome. [9] In June 2012, the NCBI’s SNP database made a new release containing approximately 38 million validated reference SNPs. (ncbi.nlm.nih.gov/projects/SNP/snp_summary.cgi)

Another type of variant is repeat sequences, either tandemly repeated or interspersed, and they account for more than 50% of the genome. [10] The tandem repeats can be subdivided into satellites, minisatellites, microsatel- lites and mononucleic tracts, depending on the length of the total repeat tract (more than 10

base pairs for satellites).

Structural variations are the largest variation with regards to the length of the involved DNA segment. They are defined as genomic alterations involv- ing segments of DNA larger than 1kb and can be divided into five subcate- gories: copy number variants (CNVs), segmental duplications, inversions, translocations and segmental uniparental disomy. [11]

CNVs are segments of DNA present at different copy numbers compared to a reference sequence and represent insertions, deletions or duplications.

The total number of CNVs collected within the Database of Genomic Vari- ants has now reached approximately 67,000 (projects.tcag.ca/variation/). A recent study revealed that less than 5% of the human genome is affected by large CNVs [12], rather than the 12% first estimated [13].

The second category of structural variations, segmental duplications, is DNA segments occurring in two or more copies and with a sequence identity larger than 90%. They can vary in copy number; hence, segmental duplica- tions can also be CNVs. About 5% of the human genome is constituted of segmental duplications and they can be both inter- and intra-chromosomal.

[2]

A DNA segment with a reversed orientation in reference to the rest of the chromosome is called an inversion, whereas a change in position of a DNA segment within the genome without a change in the total DNA content is called a translocation. [11] Translocations can also be either intra- or inter- chromosomal.

The final type of structural variation, segmental uniparental disomy, is a

phenomenon where a pair of homologous chromosomes in one individual is

derived from a single parent.

Disease-causing variants

Changes in the nucleotide sequence are often referred to as variants; some of them are responsible for causing human genetic disorders and are commonly denoted as mutations. There are many types of mutations and they can be subdivided based on either the type and size of the mutation, or the effect of the aberration on a molecular level, i.e. a loss-of-function or a gain-of- function mutation.

A point mutation is a substitution, a deletion or an insertion of a single nu- cleotide and it can be located in either coding regions or non-coding regions.

In the coding region, point mutations can be classified as missense, non- sense, frameshift, splice site or silent mutations, all depending on the out- come of the protein. Point mutations in non-coding regions can be positioned in a promoter, a splice site or another regulatory sequence and still affect the protein outcome and cause disease. [14]

Larger aberrations, such as duplications, deletions, inversions, insertions and translocations, can range from only a few nucleotides up to several Mb and the effects are similar to point mutations, but here more than one gene can be affected.

Another type of mutation is trinucleotide repeat expansions, particularly associated with neurodegenerative disorders, e.g. Huntington’s disease. [15]

There are several databases collecting mutations associated with human dis- orders, such as the DatabasE of Chromosomal Imbalance and Phenotype in Humans using Ensembl Resources (DECIPHER; decipher.sanger.ac.uk), the Human Gene Mutation Database (HGMD; www.hgmd.org), the Online Mendelian Inheritance in Man (OMIM; ncbi.nlm.nih.gov/omim) and the Leiden Open Variation Database (LOVD; www.lovd.nl).

Human genetic disorders

Genetic disorders can be classified into four different categories, depending on the type of mutation associated with the disorder and the possible in- volvement of the environment. The categories are monogenic (also called Mendelian) disorders, complex (also called multifactorial) disorders, chro- mosomal disorders and mitochondrial disorders. The present thesis will fo- cus on disorders belonging to the group of monogenic disorders.

Monogenic disorders

Monogenic disorders are rare disorders, caused by mutations in one single

gene or its regulatory sequences. However, allelic heterogeneity is usually

present, i.e. when different mutations in the same gene cause the same disor- der. In some monogenic disorders, mutations at different loci cause the same phenotype, called locus heterogeneity. A third type of heterogeneity is clini- cal heterogeneity, where mutations in the same gene are associated with different disorders, often denoted as allelic disorders. [16]

Five different inheritance patterns exist for monogenic disorders, namely:

 Autosomal dominant, in which affected individuals are heterozygous for the mutated allele, located on one of the autosomes. A disorder with this inheritance pattern can affect both males and females, and can be trans- mitted by either sex.

 Autosomal recessive, where affected individuals are homozygous or compound heterozygous for the mutated allele, located on one of the au- tosomes. An autosomal recessive disorder can also affect both males and females and parents of affected individuals are usually heterozygous for the mutated allele.

 X-linked dominant. Here, the mutated allele is located on the X- chromosome; thus, the only X-chromosome of affected males harbours the mutated allele and affected females are heterozygous. Both sexes can transmit this type of disorder, but females are more often affected, since an affected male will always transmit the disorder to his daughters.

 X-linked recessive, in which affected females are homozygous for the mutated allele and the only X-chromosome of affected males harbours the mutated allele. Mainly males are affected by X-linked recessive dis- orders, since they carry only one X-chromosome. The mother of an af- fected male is usually heterozygous for the mutated allele, whereas the status of the father is of no importance. However, an affected male will always transmit the mutated allele to his daughters, but they will only be carriers of the disorder, not affected, unless they inherit a mutated allele from their mother as well.

 Y-linked, where the mutated allele is located on the Y-chromosome;

thus, only affecting males and also only transmitted by males. This type of disorder is extremely rare.

Reduced penetrance is a complicating factor when studying the inheritance

of a disorder, meaning that not all individuals harbouring a mutation will

express the disorder. Another complicating factor is the clinical variability,

where individuals with the same disorder and even the same mutation ex-

press different features or different severity of features. Furthermore, the age

of onset of features and the changing of features with age are additional

complicating factors.

Both reduced penetrance and clinical variability could in some cases be explained by different genetic components that modify the phenotype, such as modifier genes, allelic or locus heterogeneity, or environmental factors.

Methods in disease-gene identification

Many different strategies exist, which aim to identify the genetic defects causing a disorder. A few examples, most of them used in the present thesis, will be discussed here.

Linkage analysis

Linkage analysis identifies the disease gene by its position in the genome, with no assumptions about its function. This type of approach is most suita- ble for studies of large families, where the clinical diagnosis is well-defined.

The basic principle is that genetic markers, e.g. SNPs or microsatellites, close to the mutation will be inherited together with the mutation as a block (haplotype) more often than is expected by random segregation and are, thus, said to be linked to the disease locus. This linkage is due to the fact that re- combination is less likely to occur between closely positioned loci. By using linkage analysis, genomic regions associated with the disease can be identi- fied by studying the inheritance of these genetic markers in affected and unaffected family members, and statistically calculating the so-called LOD (Logarithm of the ODds) score, which is a measure of the likelihood of ge- netic linkage between two loci and a function of the recombination fraction.

For a monogenic trait, a LOD score >3 (>2 for X-linked) is required for sig- nificant evidence of linkage, which means that the odds of the two loci being linked is 1000 times greater than the odds of them being unlinked. [17]

Linkage analysis can be performed in either a candidate-gene manner or a genome-wide approach.

Sanger sequencing

Once linkage analysis has revealed a genomic region that is linked to a cer- tain disease, candidate genes can be selected by literature searches in public databases or by the use of interaction network programs, e.g. Ingenuity pathway analysis. A number of candidate genes can then be sequenced by Sanger sequencing to possibly find the disease-causing variant.

If the disorder of a patient has been previously associated with certain

genes, the genomic region of interest is already defined; thus, Sanger se-

quencing can be performed directly on that particular gene or those particu-

lar genes of interest.

Sanger sequencing is similar to the well-known technique polymerase chain reaction (PCR), but only one primer is used in each reaction instead of two.

Furthermore, some of the nucleotides are fluorescently labelled terminators, which will stop the synthesis once incorporated into the PCR fragment.

Thus, different sizes of fragments ending with a fluorescently labelled nu- cleotide terminator will exist in one reaction. These fragments are then sepa- rated according to size by capillary gel electrophoresis and the fluorophores are detected by a laser. Since each of the four types of nucleotide terminators has a different colour depending on the fluorophore, the sequence of the fragment can then be read and analysed by certain computer software. [18]

Restriction fragment length polymorphism (RFLP)

When a variant is found by Sanger sequencing, there is usually a need for screening of additional family members or unrelated controls, in order to be able to determine whether it is the causative variant or not. This can be per- formed by Sanger sequencing, but another method of choice is RFLP.

This method is based on the fact that restriction enzymes recognize spe- cific sequence motifs and cleave a DNA fragment at these recognition sites.

A mutation can either introduce or delete such a site through its change in DNA sequence, which will then give a different cleavage pattern when com- pared to the cleavage pattern of a sequence without that specific mutation.

After the cleavage has been performed, the different cleavage patterns can be analysed by gel electrophoresis. [19]

Multiplex ligation-dependent probe amplification (MLPA)

Several methods exist for detecting gains and losses of regions in the ge-

nome. One such method is MLPA, which allows for simultaneous detection

of several different targets. Each target has two probes, designed to bind

adjacent to each other. The probes contain one target-specific hybridization

sequence and one universal PCR primer recognition sequence. The length of

the two probes together is unique for each target. After hybridization of the

probes to the targets, the probes are ligated and then denatured. The next step

is amplification of the ligated probes with a fluorescently labelled primer

pair. Since all probes contain universal PCR primer recognition sequences,

the amplification can be performed simultaneously for all probes. The ampli-

fication products are then separated according to size by capillary gel elec-

trophoresis and the unique length of each ligated probe pair makes it possi-

ble to directly relate the amount of amplification product to the amount of

initial target. By calculating the ratio of the amplification product in patients

and controls, possible gains or losses for each target can be determined. A

ratio of 1.5 indicates a gain and 0.5 a loss, whereas a ratio of 1 is considered

as normal. [20]

SNP arrays

Another method to identify gains and losses in the genome is the use of SNP arrays, which are high-density synthetic oligonucleotide microarrays, requir- ing only a small amount of DNA to genotype hundreds of thousands or even millions of SNPs simultaneously. [12] This technique can also be used for linkage analysis (described earlier) and autozygosity mapping, which is the search for chromosomal regions where affected individuals are homozygous for an allele identical by descent. Autosomal recessive disease-causing genes are usually identified by autozygosity mapping of consanguineous families or individuals originating from the same geographical area.

The procedure starts with digestion of genomic DNA by a restriction en- zyme and ligation of adaptors to the digested fragments. The fragments are then amplified simultaneously by PCR, using primers recognizing the adap- tor sequences. This is followed by fragmentation of the amplified fragments, labelling and finally hybridization to the SNP array. The resulting hybridiza- tion pattern can then be interpreted by computer analysis, to identify the genotype of each SNP and evaluate possible copy number variations or linked regions by linkage analysis. (www.affymetrix.com)

Next-generation sequencing

New technologies for sequencing have quite recently been developed, with the possibility of sequencing whole genomes in a significantly shorter period of time than traditional Sanger sequencing. These technologies, such as the Solexa sequencing-by-synthesis (Illumina), the 454 pyrosequencing (Roche 454) and the Supported oligonucleotide ligation and detection platform tech- nology (SOLiD; Applied Biosystems, now Life Technologies), will probably make sequencing of the whole genome or at least the exome, i.e. the 1-2% of the genome consisting of exons, a standard component of biomedical re- search and patient care in the future. [21] Already, they have been used suc- cessfully for screening of patients with unknown genetic causes and where there is no history of the disorder in the family or the family size is small, making linkage analysis very difficult. [22] Targeted re-sequencing of larger regions, for example those identified by linkage analysis, is also possible with these new techniques, and is much more time- and cost-efficient com- pared to using regular Sanger sequencing. [23] Sequencing of the exons of the X-chromosome in patients with mental retardation and control individu- als has also been performed with these types of technologies. [24]

The basic principle of these technologies is fragmentation of genomic

DNA into short fragments, which are then amplified, either by emulsion or

solid phase PCR, and sequenced by different techniques depending on the

platform used. [25]

The RAS-MAPK pathway

In order for an organism to function, the cells building up the organism must be able to communicate with each other and with the extracellular surround- ings. That is, the cells need to be able to respond to external signals either from other cells or from the environment, such as drugs, light or different kinds of antigens. This cellular communication is mediated by signalling pathways, in which receptors on the cell surface sense different molecules, such as growth factors, cytokines or hormones, and initiate a signalling cas- cade into the nucleus of the cell. In the nucleus, the expression of different genes can be regulated in a specific manner, depending on the external stim- ulus and the desired outcome in different cellular processes, such as differen- tiation, proliferation, apoptosis, cell survival or stress response.

Numerous signalling pathways exist in humans, but one central group is the MAPK signalling pathways. At least six different MAPK pathways exist, each named after their terminal kinases: ERK1/2, JNK1/2/3 or SAPKs, p38 MAPK, ERK3/4, ERK5 and ERK7/8, where the RAS-ERK1/2 (also denoted RAS-MAPK) pathway is one of the best characterized signalling pathways.

This pathway (Figure 1) is involved in many cellular processes, such as pro- liferation, differentiation, motility and survival, and is activated by almost all growth factors and cytokines.

Activation of the RAS-MAPK pathway

A number of different receptors, such as receptor Tyr kinases, G protein-

coupled receptors and ion channels, can initiate the activation of the RAS-

MAPK pathway. Upon stimulation of the extracellular domain of these re-

ceptors, kinase activity in the cytoplasmic domain of the receptors is in-

duced. This kinase activation phosphorylates C-terminal tyrosine residues of

the receptors, providing docking sites for a complex of molecules, including

enzymes, adaptors and docking proteins. The adaptor proteins, e.g. Grb2,

further interact with guanine-nucleotide exchange factor SOS (SOS1 and

SOS2) and recruit SOS to the plasma membrane, where small GTP-binding

proteins, the RAS proteins (KRAS, NRAS and HRAS), are localized. SOS

then catalyses the conversion of inactive guanosine-diphosphate-bound RAS

(RAS-GDP) to active guanosine-triphosphate-bound RAS (RAS-GTP). Once

RAS is activated, it activates the RAF family of kinases (ARAF, BRAF and

Figure 1. A simplified overview of the RAS-MAPK pathway. (Updated and adapted from Ekvall et al. [26])

RAF1) by phosphorylation, causing them to further phosphorylate two serine residues of the MAP2 kinases (also known as MEKs; MEK1 and MEK2).

The MEKs are dual-specificity kinases that when activated can phosphory- late two conserved threonine and tyrosine residues of ERK (ERK1 and ERK2), resulting in a conformational change in ERK and increased catalytic activity. [27-33] See Figure 1 for an overview.

Both RAF and MEK have restricted substrate specificity, whereas ERK

has a wide range of different cytosolic and nuclear substrates. To date, ap-

proximately 200 distinct substrates of ERK1/2 have been identified, where

cytoplasmic PLA2, different cytoskeletal elements and intracellular domains

substrates include the nuclear transcription factors Elk1, c-Fos and c-Jun.

[30, 34]

Regulation of the RAS-MAPK pathway

Regulation of the RAS-MAPK pathway occurs by a large variety of different mechanisms, some of which will be discussed in further detail below.

Phosphorylation and dephosphorylation

ERK1/2 can phosphorylate RAF, which inhibits its phosphorylation of MEK, or SOS1/2 can be phosphorylated by ERK1/2, causing SOS1/2 to dissociate from the adaptor protein, Grb2, and preventing its activation of RAS. Specific phosphorylation of some receptors, e.g. EGFR, by ERK1/2 is also possible, which then inhibits the signal output. Another example of sim- ilar feedback controls is ERK1/2-dependent expression of dual-specificity phosphatases (DUSPs), which can dephosphorylate ERK1/2, making them inactive. See Figure 2. [35]

Figure 2. Regulation of the pathway by ERK. Phosphorylation (P) at specific protein

residues by ERK inhibits the signals of the pathway. Furthermore, dephosphoryla-

tion of ERK by ERK-dependent DUSPs also inhibits the signalling.

Scaffolding proteins, phosphatases and inhibitors

The regulation of the RAS-MAPK pathway is also modulated by a number of different scaffolding proteins, phosphatases and inhibitors. An example of a scaffolding protein is KSR, which coordinates assembly of the RAF-MEK complex, and catalyses the phosphorylation of MEK. [36] SHOC2 is another scaffolding protein, which link RAS to downstream signal transducers. [36]

Negative regulation is the effect of the inhibitors SPRED1/2, which have their targets located between RAS and RAF and prevent phosphorylation and activation of RAF. [37] The GTPase activating protein, neurofibromin, is another example of a negative regulator, which accelerates the hydrolysis of active RAS-GTP to inactive RAS-GDP. [38] Furthermore, the phospha- tase SHP2 has been shown to regulate the RAS-MAPK pathway in several different aspects; first, it can act as a scaffolding protein and recruit the Grb2/SOS complex to the membrane. Second, it has been demonstrated to dephosphorylate and inactivate SPRY, an inhibitor binding to Grb2, and third, SHP2 can also dephosphorylate several other targets, which in turn promote activation of RAS. [39] See Figure 1 for an overview.

Internalization and degradation of receptors

Another mechanism regulating this pathway is internalization and degrada- tion of active receptor tyrosine kinases. This is possible through recruitment of ubiquitin ligases, e.g. CBL (Figure 1), which connect ubiquitin to the receptors, making them prone to degradation. [40]

Histone modifications

Recently, histone acetyltransferase MYST4/KAT6B (Figure 1) was found to primarily regulate MAPK signalling pathways, including the RAS-MAPK pathway, via H3 acetylation. MYST4/KAT6B does not interact directly with genes in the RAS-MAPK pathway, but affects the expression of genes, which interact with members of the RAS-MAPK pathway. [41]

Post-transcriptional regulation

Besides post-translational regulations, the RAS-MAPK pathway is also sub-

jected to post-transcriptional regulation. One example is binding of PUM2 to

3’UTR regulatory elements in ERK mRNA, which represses translation and

stability of the mRNA. PUM2 can also regulate the pathway indirectly by

targeting DUSP6, an inhibitor of ERK1/2. (Figure 3A) Another RNA-

binding protein is HuR, which binds to the 3’UTR of MEK1 mRNA and

makes it more stable, thereby promoting translation. Like PUM2, HuR can

affect the regulation in a negative manner as well, by binding to MKP-1

mRNA, which then can dephosphorylate and inactivate members of the MAPK family. (Figure 3B) Moreover, the 3’UTR of RAS possesses several conserved and presumed binding sites for miRNA let-7, which has been shown to repress expression upon binding to its targets (Figure 3C). [42]

Figure 3. Post-transcriptional regulation of different compo- nents of the RAS-MAPK path- way. A) PUM2 inhibits transla- tion of ERK directly, but can also indirectly function as a positive regulator of ERK sig- nalling. B) HuR promotes MEK1 translation directly, but can also negatively regulate the pathway indirectly by promot- ing MKP-1 translation. C) miRNA let-7 can directly re- press expression of RAS by binding to the 3’UTR of RAS.

Determination of signal specificity of the RAS-MAPK pathway

As mentioned, the RAS-MAPK pathway is involved in a number of different cellular processes, which raises the question as to what determines the speci- ficity of the signals within this pathway.

Signal strength and duration

Differences in strength and duration of the signal have previously been

found to have an impact on the biological outcome in response to extracellu-

lar stimulation. PC12 cells were stimulated with either EGF or NGF, two

growth factors that strongly induce ERK1/2 activation. In EGF-stimulated

cells, a transient activation of ERK was detected, which promoted prolifera-

tion of the cells, whereas in NGF-stimulated cells, ERK activation was sus-

tained and cells were differentiating. An explanation for this difference was

the effect of immediate early genes, which induce different cellular process-

es depending on the duration of the signal. [34] However, sustained ERK

signalling does not always lead to differentiation. In fibroblasts, sustained

ERK signalling by PDGF has been shown to induce proliferation, whereas

transient ERK signalling by EGF could not. [43] Despite these differences,

one can conclude that the duration of the signal is of importance for the bio-

logical output, but different types of cellular systems can result in different outcomes.

Cross-talk with other pathways

Furthermore, the RAS-MAPK pathway is not an independent pathway oper- ating alone, but part of a multi-dimensional signalling network, and can cross-talk with several other signalling pathways within this network. This, in turn, can influence and modulate the biological outcome. Members of this multi-dimensional network include other MAPK pathways, but also the PI3K-AKT pathway and the NF-κB pathway, among others. [34] In fact, the RAS-MAPK pathway and the PI3K-AKT pathway interact at multiple points with different outcomes, but in general, it seems as though members of the PI3K-AKT pathway have a positive impact on the RAS-MAPK pathway, which is most effective at low doses of growth factors, whereas RAS-MAPK negatively regulates the PI3K-AKT pathway, but at high doses of growth factors. [32]

Subcellular localization of components of the pathway

A final mechanism in determining the specificity of signals in the RAS- MAPK pathway is localization of its components to specific subcellular compartments. In most resting cells, all components of the pathway are pri- marily localized in the cytosol, due to interaction with specific scaffolding proteins. Upon stimulation, RAF is recruited to the plasma membrane to interact with active RAS and start a phosphorylation cascade. Once MEK and ERK are activated, they are released from their anchors within the cyto- sol and can translocate to the nucleus or other organelles in the cell to per- form further interactions. However, a portion of the components stay at- tached to their anchors upon stimulation, to be directed to other specific tar- gets in the cytoplasm. Together, these specific localizations in the cell influ- ence the biological outcome in a distinct manner. [30, 34]

Cancer and the RAS-MAPK pathway

The hallmarks of a cancer cell include increased or inappropriate prolifera-

tion, motility and survival; all processes where the RAS-MAPK pathway is

of great importance, making this pathway a hot target for many human can-

cers. [35] In fact, several components within the pathway have been associ-

ated with different types of cancers. Mutations in RAS genes have been iden-

tified in ~30% of human cancers, where mutations in KRAS are by far the

most common type (~85%). [31] Furthermore, BRAF and different types of

ferent cancer-types, whereas the other two RAF genes as well as the MEK genes are rarely mutated in cancer. [33] In addition, mutations in PTPN11, the gene encoding SHP2, and CBL have been found to cause juvenile mye- lomonocytic leukaemia (JMML) and mutations in NF1, encoding neurofi- bromin, contribute mainly to solid tumours and myeloid leukaemias. [44-46]

Drug development

Being a hot target for many cancers also makes the RAS-MAPK pathway an attractive and important target for development of new cancer therapeutics.

Several inhibitors with direct or indirect effect on different components of the pathway have been developed with varying success. Promising develop- ment of drugs inhibiting farnesyltransferases, which localize RAS to the membrane, turned out to be unsuccessful, due to the ability of RAS to use an alternative transferase for this localization after inhibition.

Drugs targeting the receptors in patients with oncogenic receptor signal- ling have been more successful. However, these drugs have no effect in pa- tients with oncogenic mutations further downstream of the pathway. [47] To overcome this problem, several different inhibitors targeting RAF or MEK have been developed with successful results.

Then again, tumours develop resistance over time. For instance, cancer cells targeted with inhibitors of BRAF adapt and gain resistance by switch- ing from BRAF to RAF1, which then maintains ERK1/2 activation. [35]

Another way for tumours to develop escape mechanisms is by activation of

other signalling pathways that cross-talk with the RAS-MAPK pathway,

such as the PI3K-AKT pathway. Indeed, increased signalling of the PI3K-

AKT pathway has been detected in breast cancer cells, with almost complete

blockade of the RAS-MAPK pathway. Therefore, an efficient strategy of

inhibiting tumour growth has been shown to be the use of a combination of

inhibitors targeting both the RAS-MAPK and the PI3K-AKT pathway. [33]

RASopathies

The RASopathies are a group of clinically and genetically related develop- mental disorders, including Noonan syndrome (NS) and NS-like syndromes, cardio-facio-cutaneous syndrome (CFCS), LEOPARD syndrome (LS), Cos- tello syndrome (CS), Legius syndrome, neurofibromatosis type 1 (NF1) and neurofibromatosis-Noonan syndrome (NFNS). They can also be denoted as neuro-cardio-facio-cutaneous syndromes (NCFCs) or RAS-MAPK- syndromes. The two most common syndromes within the RASopathies are NS and NF1, with and incidence of 1/1000-2500 and 1/2500-3000 respec- tively, whereas the remaining syndromes are much less frequent. [48, 49]

Mutations associated with the RASopathies have been identified in 14 different genes, all regulating the RAS-MAPK signalling pathway; hence, the name RASopathies (Figure 4). [41, 50] This pathway is often affected in various types of cancers; however, most mutations identified in the RASopa- thies do not overlap with the cancer mutations. In general, it is believed that both types of mutations lead to dysregulation of the pathway, but somatic oncogenic mutations cause a stronger activation than germline mutations, which might explain the absence of overlapping mutations. This common pathogenic mechanism, dysregulation of the RAS-MAPK pathway, explains the clinical similarities within the RASopathies, where reduced growth, typi- cal facial features, cardiac defects, ectodermal abnormalities, variable cogni- tive deficits and susceptibilities to certain malignancies are all identified characteristics. Despite this clinical overlap between the different syn- dromes, an extensive clinical variability is seen within each syndrome. [31]

Both the clinical overlap and the clinical variability can cause a difficulty in diagnosing patients with RASopathies correctly. Since different syn- dromes have different prognoses, for example mental retardation is more common in CFCS than NS and the increased risk of developing malignan- cies differs with each syndrome, setting the correct diagnosis is of great im- portance for future follow-up. By combining clinical characteristics with genetic defects, diagnosing is greatly improved. Genetics can also be of help to better understand clinical variability between patients. [49, 51]

In the present thesis, the focus will be on Noonan spectrum disorders, in-

cluding NS, NS-like syndromes and NFNS, with detailed clinical and genet-

ic descriptions of each of them. In addition, NF1 will also be discussed in

Figure 4. The RAS-MAPK pathway and the different RASopathies together with their associated genes. (Updated and adapted from Ekvall et al. [26])

Noonan and Noonan-like syndromes

Clinical description

Jacqueline Noonan was one of the first to publish a comprehensive descrip- tion of this group of patients in 1963. In 1985, it was therefore suggested to change the name from male Turner syndrome to Noonan syndrome.

NS (OMIM 163950) is one of the most common monogenic disorders in humans with an incidence of one in 1000-2500 births. The inheritance pat- tern is autosomal dominant and both familial and sporadic cases exist. Clini- cally, NS is a very variable disorder, both within families and between unre- lated patients harbouring the same mutation. [52, 53]

The main characteristics of NS are congenital heart defects, short stature,

typical facial features and unusual pectus deformity. Pulmonic stenosis is the

most common heart defect (50-65%), followed by hypertrophic cardiomyo-

pathy (~20%). Other types of heart defects include atrioventricular canal

defects and atrial and ventricular septal defects. [54] The adult height of

females (without growth hormone treatment) is 148.4±5.6cm and of males

(without growth hormone treatment) 157.4±8.0cm, but around 30% of pa- tients with NS have an adult height in the normal range. [54, 55] Typical facial features include a broad forehead, hypertelorism, ptosis, downslanting palpebral fissures, low-set posteriorly rotated ears with thick helices, deep philtrum, high arched palate, low posterior hairline and broad webbed neck.

The facial features usually become less prominent with age. Besides the main characteristics, a number of associated features exist, such as neonatal feeding difficulties, developmental and motor delay, learning disabilities, bleeding abnormalities (e.g. coagulation deficits or thrombocytopenia), skin manifestations (e.g. café-au-lait spots, pigmented naevi, lentigines or kerato- sis), cryptorchidism, ocular problems (e.g. strabismus or refractive errors) and skeletal defects (e.g. scoliosis). [54, 56]

NS-like disorder with loose anagen hair (NS/LAH, OMIM 607721) is a syndrome that greatly resembles NS. This syndrome was first presented in 2003, and patients show characteristics such as more severe growth and cog- nitive deficits, distinctive hyperactive behaviour, diffuse skin pigmentation, hoarse/hypernasal voice, easily pluckable, sparse, thin and slowly growing hair and cardiac defects, with a significant overrepresentation of mitral valve dysplasia and septal defects compared to the general NS population. [57, 58]

Another NS-like condition is NS-like disorder with or without JMML (OMIM 613563), which was reported in 2010. These patients have a rela- tively variable phenotype, although clearly overlapping with NS, with short stature, developmental delay, cryptorchidism and predisposition to JMML.

[40, 44, 50, 59]

A predisposition to develop cancer exists in patients with NS, but the risk is relatively low considering the mutations in the RAS-MAPK pathway, which is often implicated in cancer pathogenesis. JMML, acute lympho- blastic leukaemia, rhabdosarcoma and neuroblastoma are types of cancer observed in NS patients [53, 60]. Furthermore, tumour-like lesions such as giant cell lesions affecting the jawbones or joints have also been observed in patients with NS. [50]

Genetic description

Of the 14 genes associated with RASopathies, patients with NS or NS-like conditions have been found to harbour mutations in ten of these genes. How- ever, these ten NS-associated genes are not only associated with NS, but the majority of them are associated with other RASopathies as well, such as LS or CFCS (Figure 4). This further explains the clinical similarities within RASopathies. Despite association to these ten genes, the genetic aetiology in

~25% of patients with NS is still unknown.

In 2001, PTPN11 on chromosome 12q24.13 was the first gene to be associ-

phosphatase, termed SHP2, ubiquitously expressed in the cytoplasm. Two tandemly arranged N-terminal src-homology 2 domains (N-SH2 and C- SH2), a catalytic protein tyrosine phosphatase (PTP) domain and a C- terminal tail, containing two tyrosol phosphorylation sites and a proline-rich stretch, build up SHP2. The two SH2 domains bind to phosphotyrosol resi- dues on other target proteins, which promote localization of SHP2 to e.g. cell surface receptors or scaffolding proteins. SHP2 alternates between an active and inactive form by the release or binding of the N-SH2 domain to the PTP domain. [56]

Approximately 50% of patients with NS have mutations in PTPN11, mak- ing it the major gene associated with NS. The mutations identified are main- ly missense mutations, whereas deletions, insertions/duplications and indels are rare. ([62] and www.hgmd.org) The pathogenic mechanism is suggested to be destabilization of the inactive form of SHP2, i.e. a gain-of-function mechanism, and most mutations are located in residues involved in, or in close proximity to, the interaction between the N-SH2 and the PTP domain.

Mutations in PTPN11 have also been identified in patients with LS, who are sometimes diagnosed as NS in very young ages. [51, 63, 64].

The next gene to be associated with NS was KRAS on chromosome 12p12.1 [65]. KRAS consists of six exons and encodes a GTPase with two splice var- iants, KRASA and KRASB, where KRASB is predominant and often denot- ed as KRAS. The expression of KRAS is ubiquitous. Like all RAS proteins, KRAS consists of a conserved G domain, required for signalling, and a less conserved C-terminal tail, denoted as the hypervariable region, which medi- ates post-translational processing and plasma membrane anchoring. The protein cycles between inactive GDP-bound state and active GTP-bound state.

Less than 2% of NS patients harbour mutations in KRAS and, so far, only missense mutations have been identified. The outcome of all KRAS muta- tions is a gain-of-function, generated by different mechanisms, such as im- paired intrinsic GTPase hydrolysis or increased GDP/GTP dissociation rate.

[66]

Patients with CFCS have large clinical overlap with NS, and CFCS has previously even been suggested to be a variant of NS. In 2006, mutations in KRAS were also identified in patients with CFCS. [67]

Mutations in SOS1 on chromosome 2p22.1 and RAF1 on chromosome 3p25.2 have also been identified in patients with NS (~13% and 3-17% re- spectively). [68-71] SOS1 is comprised of 23 exons and encodes SOS1, which is a ubiquitously expressed guanine-nucleotide-exchange factor of RAS, catalysing conversion of inactive RAS-GDP to active RAS-GTP.

SOS1 is build up by five different domains: a histone domain (HD), a Dbl

homology (DH) domain, a pleckstrin homology (PH) domain, a RAS ex-

changer motif (Rem) domain and a Cdc25 domain. Furthermore, the C- terminal contains recognition sites, which together with the PH domain and the HD domain promote interaction with certain adaptor proteins, allowing localization to the plasma membrane upon stimulation. SOS1 is autoinhibit- ed by interaction between the DH and the Rem domain, which blocks bind- ing site for RAS. The majority of mutations in SOS1 are missense mutations, located in regions predicted to be involved in maintaining the catalytically inactive conformation. These mutations then destabilize this inactive con- formation, resulting in a gain-of-function. [56] Rarely, a mutation in SOS1 has been identified in patients with CFCS. [16, 72]

RAF1 is comprised of 17 exons and its protein RAF1, a serine threonine kinase, is also ubiquitously expressed. Three functional domains reside in RAF1, conserved regions 1 to 3 (CR1-3). CR1 is involved in RAS-GTP binding and promotes localization to the membrane, CR2 also regulates translocation to the membrane, but is also responsible for the catalytic activi- ty, and CR3 mediates phosphorylation. RAF1 has one inactive and one ac- tive conformation, where the N-terminal part of the protein interacts with the kinase domain in CR3 and inactivates it. Missense mutations are the main type of mutation in RAF1 as well, and they are clustered mainly in CR2, but also in CR3 or just C-terminal of CR3. The majority of these mutations cause a gain-of-function. [56] A few patients with LS have also been found to harbour mutations in RAF1. [68]

In 2009, two additional genes, SHOC2 and NRAS, were found to harbour mutations in patients with NS or NS-like conditions [57, 73]. SHOC2 on chromosome 10q25.2 is a nine-exon gene and encodes a widely expressed protein, SHOC2, mainly composed of leucine-rich repeats. SHOC2 has been found to have two functions, either it can act as a scaffold and guide RAS to downstream targets or it is part of PP1C and promotes PP1C’s translocation to the plasma membrane. Once at the plasma membrane, PP1C mediates RAF1 dephosphorylation at Ser259, which is a requirement for stable trans- location of RAF1 to the plasma membrane and catalytic activation. [56]

Mutations in SHOC2 are found in patients with the NS-like condition NS/LAH, which corresponds to less than 5% of the entire NS population.

Hitherto, only one single missense mutation has been identified in SHOC2-

positive patients. This mutation changes serine to glycine in residue two of

the protein, which introduces an N-myristolation site. N-myristolation is a

process where a 14-carbon saturated fatty acid is attached to an N-terminal

glycine residue with a satisfactory consensus sequence surrounding it, which

promotes anchoring to the plasma membrane. Mutated SHOC2 fulfils the

consensus requirements and becomes N-myristolated, resulting in constitu-

tive membrane translocation of SHOC2. In turn, this constitutive transloca-

tion promotes prolonged dephosphorylation of RAF1 at Ser259 mediated by

Hence, this single missense mutation of SHOC2 is a gain-of-function muta- tion. [56, 57]

The other gene, NRAS on chromosome 1p13.2, is constituted by six exons and as KRAS, it encodes a small GTPase, cycling between an active GTP- bound state and an inactive GDP-bound state. Only a few patients with NS have been identified with mutations in NRAS and the observed mutation type is only missense, where the mutation in each case results in enhanced phos- phorylation of MEK and ERK, i.e. a gain-of-function. [73-75]

Furthermore, a few reports have been published on NS patients with mis- sense mutations in BRAF and MEK1 [16, 69, 76, 77]. BRAF, located on chromosome 7q34, consists of 18 exons and as RAF1, it encodes a serine threonine kinase expressed in various tissues and contains the three con- served regions (CR1-3), but BRAF has been shown to have higher MEK kinase activity compared to RAF1 and ARAF. Mutations in BRAF are the major cause in CFCS; however, most mutations identified in NS patients do not overlap with mutations associated with CFCS. Two patients with LS have also been found to harbour mutations in BRAF. [77, 78]

MEK1 on chromosome 15q22.31 consists of eleven exons and encodes mitogen-activated protein kinase kinase 1, a kinase downstream of RAF.

MEK1 consists of one negative regulatory domain at the N-terminal and a single kinase domain. Mutations in MEK1, and also the functionally related MEK2, are mainly associated with CFCS. [56]

The last two genes to be associated with NS or NS-like conditions were CBL and MYST4/KAT6B. CBL is a 16-exon gene located on chromosome 11q23.3. It encodes a ubiquitously expressed RING finger E3 ubiquitin lig- ase, one of the enzymes required for targeting substrates for degradation by the proteasome. Four domains comprise CBL: one N-terminal tyrosine ki- nase-binding domain, involved in protein-protein interaction, one zinc-finger RING-finger domain, mediating E3 ubiquitin ligase activity, one proline-rich region and one ubiquitin-associated domain, promoting ubiquitin binding and overlapping with a leucine zipper motif, involved in protein dimeriza- tion. Mutations in CBL have been shown to impair ubiquitylation of recep- tors, causing increased pathway signalling, and patients with mutations in CBL have features more or less reminiscent of NS. [40, 44, 59] Somatic mu- tations in this gene are found in patients with different myeloid malignan- cies, and in fact, the spectrum of somatic mutations overlaps with the germline mutations. [40]

MYST4/KAT6B is located on chromosome 10q22.2 and consists of 16 ex-

ons. It encodes a histone acetyltransferase, which regulates the RAS-MAPK

pathway via H3 acetylation. This gene was found to be associated with an

NS-like phenotype in a patient with a balanced chromosomal translocation,

where one of the breakpoints was located within the MYST4/KAT6B gene.

Quantitative RT-PCR in this patient revealed a 50% decrease in mRNA ex- pression levels of MYST4/KAT6B. This haploinsufficiency resulted in in- creased phosphorylation of MEK1/2 and ERK1/2, and also enhanced AKT phosphorylation. However, no further mutations in this gene could be identi- fied in a cohort of 131 subjects with suggestive NS features, who were nega- tive for mutations in previously associated NS-genes. [41]

Genotype-phenotype correlations

Although operating within the same pathway, and in the majority of cases causing increased signalling, mutations in different genes or residues are associated with different clinical features (Table 1). However, NS patients with mutations in the same gene or even the same nucleotide position still can display clinical variability.

A significant relationship between NS patients harbouring a PTPN11 muta- tion and pulmonic stenosis has been revealed, whereas the frequency of hy- pertrophic cardiomyopathy was significantly lower in patients with a PTPN11 mutation. [46, 79] Familial cases also have a significantly higher occurrence of mutations in PTPN11 compared to sporadic cases. In addition, PTPN11-positive patients more often have easy bruising, thorax deformities, short stature and cryptorchidism compared to other genotypes. [71, 80] Fur- thermore, a specific mutation in PTPN11, p.T73I, has been identified as a JMML risk genotype. [50]

Patients with SOS1 mutations have a similar spectrum of heart defects as PTPN11-positive patients, but are less likely to have short stature and to need special education. However, they have significantly more often thorax deformities and ectodermal manifestations, such as curly hair, sparse eye- brows or keratosis pilaris. [71, 80, 81]

Hypertrophic cardiomyopathy and hyperpigmented cutaneous lesions occur at a significantly higher frequency (75-95% and 33% respectively) in pa- tients with a RAF1 mutation compared to NS patients in general. [48, 50, 68]

KRAS-positive patients more often have cognitive impairment and a more severe phenotype than the general NS population; otherwise the phenotype is quite variable. [48, 50]

Patients with mutations in SHOC2 represent a separate entity, NS/LAH;

hence, they have several features associated with the genotype. Loose

anagen hair is one feature, which is only observed in this condition. Other

associated features include hoarse/hypernasal voice, reduced growth caused

Table 1. Genotype-phenotype correlations in NS and NS-like syndromes Mutated gene Clinical feature

PTPN11

+ Pulmonic stenosis + Thorax deformities + Easy bruising + Short stature

+ Cryptorchidism (in males) + Familial cases

+ JMML (p.T73I)

- Hypertrophic cardiomyopathy

SOS1

+ Pulmonic stenosis + Thorax deformities + Ectodermal manifestations - Hypertrophic cardiomyopathy - Short stature

- Special education

RAF1 + Hypertrophic cardiomyopathy

+ Hyperpigmented cutaneous lesions

KRAS + Cognitive impairment

+ Severe phenotype in general

SHOC2

+ Loose anagen hair + Hoarse/hypernasal voice + Reduced growth (GH deficiency) + Cognitive impairment

+ Distinctive hyperactive behaviour

+ Darkly pigmented skin with eczema or ichtyosis + Mitral valve dysplasia

+ Septal defects

CBL

+ JMML + Short stature + Developmental delay + Café-au-lait spots + Cryptorchidism (in males)

BRAF

+ Neonatal growth failure (due to feeding difficulties) + Mild-to-moderate cognitive deficits

+ Hypotonia

+ Multiple naevi or dark colored lentigines

+ More severe phenotype in adulthood

MEK1, NRAS, MYST4/KAT6B No clear correlations

by GH deficiency, cognitive deficits, distinctive hyperactive behaviour and darkly pigmented skin with eczema or ichtyosis. These patients also have a significant overrepresentation of the heart defects mitral valve dysplasia and septal defects compared to the general NS population. [57]

Likewise, patients with mutations in CBL are denoted as a separate disorder.

The main associated feature is JMML, although CBL-positive patients with- out hematologic abnormalities exist. Short stature, developmental delay, café-au-lait spots and cryptorchidism are additional associated features. [40, 44, 59]

BRAF-positive NS patients are associated with neonatal growth failure due to feeding difficulties, mild-to-moderate cognitive deficits, hypotonia and multiple naevi or dark colored lentigines. They also present with a more severe phenotype in adulthood compared to patients with mutations in PTPN11 or SOS1. [48]

For MEK1, NRAS and MYST4/KAT6B, no clear genotype-phenotype correla- tions have been identified.

Neurofibromatosis type 1

Clinical description

The first description of neurofibromatosis type 1 (NF1, OMIM 162200), made by von Recklinghausen, dates back to 1882. [82] NF1 is one of the most common disorders with an autosomal dominant inheritance pattern, and the incidence is approximately one in 2500-3000 births. About half of the cases are inherited and the other half are caused by de novo mutations. As in NS, there is a clinical variability in NF1 both within families and between unrelated patients harbouring the same mutation. [49]

To establish a clinical diagnosis of NF1, the National Institutes of Health (NIH) consensus statement is used, which is a statement of diagnostic crite- ria for NF1 proposed by the NIH Consensus Development Conference in 1987 and later reviewed in 1997. [83] The statement requires at least two of the following seven clinical criteria to be present in order to set the diagnosis to NF1:

 Six or more café-au-lait spots with a greatest diameter

 >5mm in children

 >15mm in adults

 Two or more neurofibromas of any type or one plexiform neurofi- broma

 Axillary or inguinal freckling

 Two or more Lisch nodules

 Optic pathway glioma (OPG)

 Skeletal abnormalities, e.g. scoliosis, pseudoarthrosis, sphenoid dys- plasia or thinning of long bone cortex

 A first-degree relative with NF1

The presence of café-au-lait spots in NF1 patients is high, >99% of patients have these spots. The benign tumours neurofibromas are, as the name sug- gests, one of the main hallmarks of NF1. Cutaneous neurofibromas are the most common type, present in >99% of patients with NF1, whereas plexi- form neurofibromas are less common (30-50%). Both café-au-lait spots and neurofibromas are less common in infants, but develop later in childhood.

Axillary or inguinal freckling is also a quite common symptom, present in approximately 85% of NF1 patients. Like neurofibromas, Lisch nodules are also a type of benign tumour, affecting the iris. These tumours are very common (90-95%) in patients with NF1 and is one of the best markers for NF1 in older children and adults. A third type of benign tumour found in NF1 patients, but at a much lower frequency (15%) than neurofibromas or Lisch nodules, are OPGs, which affect the central nervous system. Skeletal abnormalities are found in ~60% of patients with NF1. [82, 84, 85]

Besides these seven clinical criteria, NF1 patients display several addi- tional features, for instance learning disabilities (>50%), epilepsy (6-7%), congenital heart defects (~3%), hypertension, gastrointestinal problems, delayed puberty, mildly short stature or macrocephaly [82, 85-88].

As noted, benign tumours occur in NF1 patients with high frequency. Never- theless, the presence of malignant tumours is relatively low (10-20%), alt- hough with an increased frequency compared to the general population. The most common reason for premature death in NF1 patients is presence of malignant peripheral nerve sheath tumours (MPNSTs), which are more prone to develop in plexiform neurofibromas. Other malignant tumours pre- sent in NF1 patients are central nervous system tumours (particularly associ- ated with OPGs), rhabdomyosarcoma and leukaemias (especially JMML).

[83, 85, 87]

Genetic description

As opposed to NS, NF1 has only been associated with mutations in one

gene, NF1 on chromosome 17q11.2, and the first mutations were identified