Amendment history:Erratum (November 2018)

Mosaic RAS/MAPK variants cause sporadicvascular malformations which respond totargeted therapy

Lara Al-Olabi, … , E. Elizabeth Patton, Veronica A. Kinsler

J Clin Invest. 2018. https://doi.org/10.1172/JCI98589.

In-Press Preview

BACKGROUND. Sporadic vascular malformations (VMs) are complex congenitalanomalies of blood vessels that lead to stroke, life-threatening bleeds, disfigurement,overgrowth, and/or pain. Therapeutic options are severely limited and multi-disciplinarymanagement remains challenging, particularly for high-flow arteriovenous malformations(AVM).

METHODS. To investigate the pathogenesis of sporadic intracranial and extracranial VMsin 160 children in which known genetic causes had been excluded, we sequenced DNAfrom affected tissue and optimised analysis for detection of low mutant allele frequency.

RESULTS. We discovered multiple mosaic activating variants in four genes of the RAS-MAPK pathway, KRAS, NRAS, BRAF, and MAP2K1, a pathway commonly activated incancer and responsible for the germ-line RAS-opathies. These variants were more frequentin high-flow than low-flow VMs. In vitro characterisation and two transgenic zebrafish AVMmodels which recapitulated the human phenotype validated the pathogenesis of the mutantalleles. Importantly, treatment of AVM-BRAF mutant zebrafish with the BRAF inihibitor,Vemurafinib, restored blood flow in AVM.

CONCLUSIONS. Our findings uncover a major […]

Clinical Medicine Therapeutics Vascular biology

Find the latest version:

http://jci.me/98589/pdf

Title

Mosaic RAS/MAPK variants cause sporadic vascular malformations which respond to targeted

therapy

Authors

Lara Al-Olabi1*, Satyamaanasa Polubothu1,2*, Katherine Dowsett3*, Katrina A Andrews4,5*,

Paulina Stadnik1, Agnel P Joseph6, Rachel Knox4,5, Alan Pittman7, Graeme Clark8, William

Baird1, Neil Bulstrode9, Mary Glover2, Kristiana Gordon10, Darren Hargrave11, Susan M

Huson12, Thomas S Jacques13, Gregory James14, Hannah Kondolf15, Loshan Kangesu9, Kim M

Keppler-Noreuil15, Amjad Khan2, Marjorie J Lindhurst15, Mark Lipson16, Sahar Mansour17,

Justine O’Hara9, Caroline Mahon2, Anda Mosica2, Celia Moss18, Aditi Murthy2, Juling Ong9,

Victoria E Parker4,5, Jean-Baptiste Rivière19, Julie C Sapp15, Neil J Sebire20, Rahul Shah9, Bran

Sivakumar9, Anna Thomas1, Alex Virasami13, Regula Waelchli2, Zhiqiang Zeng3, Leslie G

Biesecker15, Alex Barnacle21, Maja Topf6, Robert K Semple4,5,22+, E. Elizabeth Patton3+,

Veronica A Kinsler1,2+

*These first authors contributed equally to this work

+These senior authors contributed equally to this work

Affiliations

1. Genetics and Genomic Medicine, University College London GOS Institute of Child

Health, London WC1N 1EH, UK

2. Paediatric Dermatology, Great Ormond St Hospital for Children NHS Foundation Trust,

London WC1N 3JH, UK

3. MRC Human Genetics Unit and CRUK Edinburgh Centre, MRC Institute of Genetics and

Molecular Medicine, University of Edinburgh, Western General Hospital, Edinburgh

EH4 2XU, UK

4. Metabolic Research Laboratories, Wellcome Trust-MRC Institute of Metabolic

Science, University of Cambridge, Cambridge CB2 0QQ, UK

5. The National Institute for Health Research Cambridge Biomedical Research Centre,

Cambridge CB2 0QQ, UK.

6. Biological Sciences, Birkbeck, University of London, London WC1E 7HX, UK

7. Molecular Neuroscience, UCL Institute of Neurology, London WC1N 3BG, UK

8. Department of Medical Genetics, University of Cambridge, Cambridge Biomedical

Campus, Cambridge CB2 0QQ, UK

9. Plastic Surgery, Great Ormond St Hospital for Children NHS Foundation Trust, London

WC1N 3JH, UK

10. Dermatology and Lymphovascular Medicine, St. George’s Hospital NHS Trust, London

SW17 0QT, UK

11. Paediatric Oncology, Great Ormond St Hospital for Children NHS Foundation Trust,

London WC1N 3JH, UK

12. Manchester centre for Genomic Medicine, St Mary’s Hospital, Manchester M13 9WL,

UK

13. Developmental Biology and Cancer Programme, UCL GOS Institute of Child Health

and Department of Histopathology, Great Ormond Street Hospital for Children NHS

Foundation Trust, London

14. Paediatric Neurosurgery, Great Ormond St Hospital for Children NHS Foundation

Trust, London, WC1N 3JH, UK

15. National Human Genome Research Institute, National Institute of Health, Bethesda,

MD 20892, USA

16. Paediatrics and Clinical Genetics, Kaiser Permanente Medical Center, Sacramento, CA

95825, USA

17. Clinical Genetics, St. George’s Hospital NHS Trust, London SW17 0QT

18. Paediatric Dermatology, Birmingham Women’s and Children’s NHS Foundation Trust

Birmingham B4 6NH and University of Birmingham, UK

19. McGill University Health Centre and Research Institute, 1001 Boulevard Décarie,

Montréal (QC) H4A 3J1

20. Paediatric Pathology, Great Ormond St Hospital for Children NHS Foundation Trust,

London, WC1N 3JH, UK

21. Interventional Radiology, Great Ormond St Hospital for Children NHS Foundation

Trust, London, WC1N 3JH, UK

22. University of Edinburgh Centre for Cardiovascular Science, Queen’s Medical

Research Institute, Edinburgh EH16 4TJ

Licensing

This work was funded or supported by grants from the Wellcome Trust (UK), the Medical

Research Council (UK) and the UK National Institute for Health Research, and therefore

requires a Creative Commons CC-BY license in order to support publication fees for this

manuscript.

Key words Vascular malformation, Gene discovery, Targeted therapy, Mosaicism, MAPK, RAS

25 word summary

Activating mosaic mutations in four MAPK pathway genes are discovered to cause sporadic

intracranial and extracranial malformations, and respond to targeted inhibitors in a zebrafish

model.

Conflict of interest

LGB is an uncompensated advisor to the Illumina Corp, Receives royalties from Genentech,

Inc., and honoraria from Wiley-Blackwell. All other authors declare no conflict of interest.

Abstract

Background

Sporadic vascular malformations (VMs) are complex congenital anomalies of blood vessels

that lead to stroke, life-threatening bleeds, disfigurement, overgrowth, and/or pain.

Therapeutic options are severely limited and multi-disciplinary management remains

challenging, particularly for high-flow arteriovenous malformations (AVM).

Methods

To investigate the pathogenesis of sporadic intracranial and extracranial VMs in 160 children

in which known genetic causes had been excluded, we sequenced DNA from affected tissue

and optimised analysis for detection of low mutant allele frequency.

Results

We discovered multiple mosaic activating variants in four genes of the RAS-MAPK pathway,

KRAS, NRAS, BRAF and MAP2K1, a pathway commonly activated in cancer and responsible for

the germ-line RAS-opathies. These variants were more frequent in high-flow than low-flow

VMs. In vitro characterisation and two transgenic zebrafish AVM models which recapitulated

the human phenotype validated the pathogenesis of the mutant alleles. Importantly,

treatment of AVM-BRAF mutant zebrafish with the BRAF inihibitor, Vemurafinib, restored

blood flow in AVM.

Conclusions

Our findings uncover a major cause of sporadic vascular malformations of different clinical

types, and thereby offer the potential of personalised medical treatment by repurposing

existing licensed cancer therapies.

Funding

This work was funded or supported by grants from AVM Butterfly Charity, the Wellcome Trust

(UK), the Medical Research Council (UK), the UK National Institute for Health Research,

L’Oreal-Melanoma Research Alliance, the European Research Council and the National

Human Genome Research (US).

Sporadic vascular malformations (VMs) are congenital malformations of blood vessels, with

high associated morbidity and limited treatment options (Figure 1, Figure 2). VMs have

traditionally been divided into diagnostic groups according to anatomical site (intracranial

versus extracranial), and flow characteristics (high versus low), with detailed subclassification

incorporating multiple historical diagnostic labels(1). In high-flow, or arteriovenous,

malformations (AVM), direct complex interconnections between arteries and veins without

the normal interposed small-bore capillary network permits dangerous flow of high pressure

arterial blood into thin-walled veins leading to bleeds or stroke. In low-flow, capillary and/or

venous malformations, interconnections between anatomically abnormal capillaries and/or

veins can lead to sludging of blood, perilesional tissue anoxia, thrombosis, and overgrowth.

Application of next generation sequencing to a range of sporadic syndromes featuring VMs

has rapidly established the paradigm that such disorders are commonly caused by postzygotic

variants activating key cellular growth pathways(2-11). Variants overlap with those

documented in cancer, but the allelic spectra may differ. Accurate genotype-based

stratification of affected patients has begun to improve prognostication and in conditions

affecting the PI3K-AKT-MTOR signalling pathway has led to early clinical trials of targeted

therapies(12, 13). Despite these discoveries, however, a substantial proportion of patients

have no mutations in known associated genes(9, 14, 15). We sought to address this issue

using ultra-deep next generation sequencing and bio-informatic analyses aimed at detection

of low mutant allele frequency in affected tissue. Here, we discover multiple mosaic activating

variants in four genes of the RAS-MAPK pathway cause VM, and model therapy in an animal

model.

Results

25 patients with high-flow and 135 patients with low-flow VMs, in whom known VM-related

pathogenic variants had previously been excluded (Methods), were investigated to identify

the cause of the clinical phenotype. Using deep next generation sequencing of affected tissue

known or predicted pathogenic variants in RAS/MAPK pathway genes were identified in 9 of

the 25 patients with high-flow VMs (Table 1, Figure 3), and confirmed by a second method.

Identified variants were in KRAS (n=4) and BRAF (n=1), encoding key proximal components of

the RAS/MAPK signaling pathway, known to be pathogenic, but previously undescribed as

causal in any type of VM. In addition, a cluster of variants was identified in exon 2 of MAP2K1

(n=4), confirming a hotspot very recently reported in extracranial AVM but not known prior

to our study(9), and adding one novel small intragenic deletion (c.159_173del,

p.(F53_Q58delinsL)) (Table 1, Figure 3) to the allelic spectrum. Variant allele frequencies

ranged from 3-26% in affected tissue, but with no detectable variant in paired blood samples

where these were available (n=4/9 Table 1). One KRAS-variant patient had a sporadic AVM

restricted to the intra-cranial cavity, which presented with an intracranial haemorrhage at the

age of 13. The other eight RAS/MAPK variant patients had high-flow VMs centred in the skin

or soft tissues at various anatomical sites, with intra-cranial extension documented in two

(Table 1).

Of the 135 patients with low-flow VMs, five had mosaic RAS gene variants, four in KRAS and

one in NRAS, known to be pathogenic. Variants were present in each affected tissue tested

and were consistent within an individual (Table 1), with a variant allele load of 3-29%, but

undetectable in blood (n=4/5 Table 1). These patients exhibited a spectrum of phenotypic

manifestations as is typical of mosaic disorders (Table 1).

Structural modelling was undertaken of the four mosaic variants detected in exon 2 of the

MAP2K1 gene, two identical missense variants (p.(K57N)), and two small intra-exonic

deletions removing respectively codons 53-58 (novel c.159_173del, p.(F53_Q58delinsL)) and

58-62 (c.173_187del, p.(Q58_E62del)) (Uniprot Q02750.2 (MP2K1_HUMAN). All were found

to affect helix A in the 3D protein structure (residues 44-58). The deletions of residues 53-58

and 58-62 are predicted to affect the integrity of helix A, and K57 is a critical amino acid

involved in a hydrogen bond interaction with the backbone of a beta strand at the active site

(Figure 3c), a critical 3D-structure stabilizing interaction. Numerous interactions involving

Q46, R49, L50, F53, K57, V60 and the rest of the kinase domain, are found in the inhibited

form of MAP2K1 (Figure 3c), consistent with critical involvement of helix A in protein function.

Helix A forms an integral part of the Negative Regulatory Region of kinase function(16), and

these variants are therefore predicted to destabilise the conformation of the inactive state.

Transient overexpression of MAP2K1 and BRAF mutants in HEK293T cells led to increased

phosphorylation of ERK compared either to overexpression of corresponding wildtype genes

or to mock-transfected controls at 24 hours (Figure 4a,b). Overexpression of two of the same

mutants in human umbilical vein endothelial cells (HUVEC) led to disruption and disordering

of spontaneous vascular tube formation between 6-24 hours compared to mock-transfected

controls (Figure 4c,d), with significant reductions in mean number of master junctions, total

vascular tube length, and total mesh area (Figure 4e,f,g).

Two zebrafish models of VM were then developed to validate our findings in vivo and to serve

as a platform for screening of potential drug treatments. Artery and vein development in

zebrafish is regulated by similar mechanisms to those in humans(17), and blood flow through

the vasculature is clearly visible by two days of development (Supplementary Movie). We

injected separately BRAFV600E and MAPK2K1Q58del expressed from a pan-endothelial promoter,

flia, into single-cell zebrafish embryos to generate post-zygotic expression of BRAFV600E or

MAPK2K1Q58del in vessels, that like in the patients, is expressed in a mosaic fashion (Figure

5a). For some animals, both BRAF and BRAFV600E expression led to shortening along the

anterior-posterior axis, as reported previously in zebrafish with high MAPK expression, and

these were removed from the analysis (Supplementary Figure S1; Anastasaki et al., 2009;

Anastasaki et al., 2012). Notably, BRAFV600E, but not expression of wild type BRAF, led to

disordered vessel formation leading to severely impeded blood flow, and recapitulated the

clinical features of patient VMs (Figure 5c, d, e; Supplementary Movie). Interestingly, these

VM formed preferentially at the junction of the caudal artery and vein(18). Mosaic expression

of MAP2K1Q58del, but not wild type MAP2K1, also generated VM (n=5/37) (Figure 5f).

Zebrafish with established VM were then treated for two days with a low, continuous dose of

Vemurafenib (0.1uM), the approved anti-cancer BRAF inhibitor, and assessed by blind scoring

for improved blood flow (Figure 5g). Strikingly, almost all Vemurafenib-treated zebrafish

exhibited improved blood flow (Figure 5g). In contrast, drug treatment had no effect on

control zebrafish, and only a minority of BRAFV600E VMs demonstrated improved blood flow

spontaneously. These important results of improvement in phenotype from RAS-RAF-MAPK

targeted therapy in an animal model echo recent murine models of VM ameliorated by

targeted pathway inhibition of the PI3K-AKT-MTOR pathway(19, 20).

Discussion

We discover mosaic activating variants in oncogenes at several levels of the RAS/MAPK

signaling pathway as a major cause of sporadic VMs, and particularly of the clinically high-risk

group of AVMs. This discovery identifies the RAS-RAF-MAPK pathway as a major cause of

sporadic VMs, and opens the door to repurposing of targeted medical therapies on a

personalized medicine basis. The pathogenicity of these variants is supported by several lines

of evidence including the well-established pathogenicity of the BRAF, KRAS and NRAS variants

in other disorders and tissues, the mosaic occurrence of these variants in the lesions and their

absence in peripheral blood, and by the existence of a hotspot of MAP2K1 variants in the

COSMIC database, comprising 65 tumor variants from F53 to E62.

Several genetic causes of human inherited and sporadic vascular malformation are already

known(21). Inherited susceptibility to cerebral cavernous (low-flow) malformations is

conferred by germline mutations in KRIT1(22), CCM2(23) and PDCD10(24), however

malformations themselves require a somatic ‘second hit’(25). The same paradigm of an

inherited genetic diathesis interacting with a somatic mutation applies to heritable extra-

cranial low-flow malformations caused by mutations in TEK(10) and GLMN(26, 27), and

heritable (high-flow) arteriovenous malformations associated with mutations in PTEN,

ACVR1, ENG, RASA1(28, 29). The same mechanism is also likely in the recently described gene

for CM-AVM syndrome EPHB4(30). Known causes of sporadic vascular malformations are

post-zygotic mutations in TEK(10, 11), AKT1(2), PIK3CA(7), GNAQ(3, 4), GNA11(4) and

MAP3K3(8) in low-flow VMs, and recently described MAP2K1 mutations in high flow extra-

cranial malformations(9) (Figure 6). Thus, there is a pattern of milder predisposing mutations

in the germline requiring a second hit to lead to phenotype, and more severe mutations

leading directly to a phenotype only seen as post-zygotic hits. Our discovery that RAS-MAPK

VM mutations are strong activators of the MAPK signaling pathway and overlap with the

cancer allele spectrum is in line with the hypothesis that lethal genes survive by

mosaicism(31).

In silico protein-interaction analysis(32) of the full spectrum of genes associated with VMs,

including our new findings, demonstrates a strong network of functional interaction among

the gene products (Supplementary Figure S2). This aligns with recent mechanistic work on

vascular malformations secondary to germline variants in CCM2, PCD10 and KRIT1, which

drive vascular malformations by increasing MAP3K3 (MEKK3) signaling in endothelial

cells(33). MAP3K3 is essential for the development of embryonic cardiovascular system in

mice(34) and zebrafish(35), and is considered a critical nexus between the Congenital

Cavernous Malformation complex and downstream signaling pathways(33, 35, 36) such as

those mediated by Rho signaling(36), p38 MAPK(37) and MEK5/ERK5(35, 37). We hypothesise

that congenital vascular malformations ultimately result from dysregulation of vascular cell

MAPK and/or PI3K signaling during human embryonic development.

Analysis of multiple different affected tissues in two of the patients revealed consistency of

mutation within an individual, confirming a post-zygotic hit to a multipotent precursor in

those cases. Although several mosaic disorders have been described in association with

postzygotic activating variants in RAS oncogenes, vascular malformations have not been

described in those phenotypes(38-44). Arterial abnormalities have been described in

association with epidermal nevi, but without genotype data (45-47). Interestingly, post-natal

somatic second-hit RAS and RAF gene family variants have been described in the small

acquired vascular tumours pyogenic granulomas, arising on a background of a post-zygotic

GNAQ-variant congenital capillary malformation(48). In these cases, RAS variants were not

the cause of the underlying VM, but consistent with somatic RAS variants leading to tumours

of many types.

These findings are particularly important in the context of the lack of effective therapy for

complex VMs, particularly for AVMs. Embolization and surgical intervention are not always

possible for safety reasons, and where possible can have limited short-term results due to

very frequent localised recanalization and recurrence, or be associated with disfiguring

results. Identification of mosaic mutations in a third of our AVM cohort, including intracranial

AVMs, will radically alter the concept of management of these VMs by introducing the

possibility of targeted medical therapy.

The cause of the difference in frequency of these variants between the high- and low-flow

cohorts is not clear. As a minority of the low-flow cohort skin biopsies were cultured for

fibroblasts before DNA extraction it is possible that the mutations was either lost in culture

or not present in fibroblasts. We would not however expect this to explain the size of the

difference between the groups, and further work will be needed to explore the increased

frequency in high-flow vascular malformations.

In summary, these findings demonstrate a novel cause of sporadic VMs, and suggest a

common pathogenesis for sporadic VMs of different vessel types and at different anatomical

locations, including intra-cranial. Genotyping of affected tissue, where accessible, should

therefore be a key element of management across the diverse medical subspecialties to

which affected patients present. Resulting genetic stratification may not only be of value

prognostically, but may also now serve to guide therapy. Although formal clinical trials will be

required before routine repurposing of such agents is sanctioned in clinical care, licensed

targeted medical therapy on a compassionate basis may already be considered for severe or

life-threatening cases.

Methods

Patient cohorts

Twenty-three patients with high-flow vascular malformation (AVM), confirmed clinically and

radiologically, who were seen in the Paediatric Dermatology Department at Great Ormond

Street Hospital for Children, London (GOSH), were recruited and gave written informed

consent. At the time of recruitment, no genes were known to be causative in sporadic AVM.

After initial results of our study were available two patients with intracranial-only AVM were

also recruited for testing via the Paediatric Neurosurgery department.

135 patients with low-flow vascular malformations and/or overgrowth, and one patient with

a high-flow vascular malformation were recruited with written informed consent to the

Investigation of Segmental Overgrowth Disorders study (REC 12-EE-0405) and found to be

wildtype for genotype of known sporadic vascular and overgrowth-related genes PIK3CA,

AKT1, GNAQ, GNA11 and TEK. One further patient with a low-flow vascular malformation

and overgrowth was identified after genotyping results from the Phase 1 dose-finding trial of

ARQ 092 in children and adults with Proteus syndrome study at the NIH, Bethesda, USA,

bringing the totals of high-flow to 25, and low-flow to 135. Detailed clinical phenotyping was

undertaken in all patients before genotyping (Table 1).

DNA extraction

DNA was extracted directly from samples by DNeasy Blood and Tissue Kit (Qiagen), and from

paraffin embedded tissue using the RecoverAll total nucleic acid extraction kit for FFPE (Life

Technologies).

Next generation sequencing and variant calling

High-flow vascular malformation patients

21 FFPE and 4 fresh samples were sequenced using the SureSeq™ Solid Tumour Panel (Oxford

Gene Technology). This was chosen due to the difficulty in biopsying AVMs due to risk of

bleeding, and because this panel is optimized for FFPE tissue. DNA library preparation was

performed for each sample using the Agilent SureSelect™ XT Reagent Kit, as per the

manufacturers protocols

(http://www.ogt.co.uk/assets/0000/4457/990162_HB_SureSeqSolidTumour_110215.pdf).

Samples were then pooled and sequenced on a MiSeq instrument (Illumina) according to the

manufacturer’s recommendations for paired-end 150-bp reads. In-depth sequencing was

performed to achieve a mean sequencing depth of 500 reads for all targeted coding bases.

For two further patients fresh 4mm skin biopsies were taken from the AVM, and DNA

extracted directly from the whole sample, and from paired blood samples. These were

sequenced by whole exome sequencing, library preparation was with SureSelect™ Agilent

QXT v6 following the manufacturers protocols, and sequencing was on the HiSeq 3000

(Illumina), with a mean read depth of 500X.

Sequence alignment to the human reference genome (UCSC hg19), and variant calling and

annotation was performed with our in-house pipeline. Briefly this involves alignment with

NovoAlign, removal of PCR-duplicates with Picard Tools followed by local realignment around

indels and germline variant calling with HaplotypeCaller according to the Genome Analysis

Toolkit (GATK) best practices. We identified potentially mosaic variants with GATK muTECT2

in tumor-only somatic variant calling mode. The raw list of single nucleotide variants (SNVs)

and indels were then filtered using ANNOVAR. Only exonic and donor/acceptor splicing

variants were considered. Priority was given to rare variants (<1% in public databases,

including 1000 Genomes project, NHLBI Exome Variant Server, Complete Genomics 69, and

Exome Aggregation Consortium). Furthermore, we have an in-house set of approximately six

thousand exomes encompassing controls, rare diseases for cross-checking any shortlisted

candidate variants, and for sequencing artefact removal. Identification of candidate variants

where paired samples were available was also performed using Ingenuity Variant Analysis by

selecting variants present in skin but not in blood (or in mosaic levels in both). For all

candidate variants BAM files were viewed using the Integrative Genomics Viewer (Broad

Institute), and mosaicism percentage taken from the mutant allele reads divided by the total

directly from the BAM. Candidate post-zygotic variants were confirmed by Sanger sequencing

in all DNA samples available from each patient. To maximize detection of mutant alleles at

low percentage mosaicism, restriction enzyme digests of the normal allele were designed

where necessary using validated methods(49) and Sanger sequencing performed. See

Supplementary Table S1 for primer sequences. Touchdown PCR programmes were used

throughout, with 40 cycles for the first PCR (annealing and extension times of 1 minute), and

15 for the second hemi-nested PCR where required.

Low-flow vascular malformation patients

A 4mm punch biopsy of affected tissue was taken from an area of overgrowth and/or vascular

malformations. DNA was extracted either directly from the biopsy (51/134 patients), or from

dermal fibroblasts grown from the biopsy (59/134 patients), using the QiaAMP DNA Micro Kit

(Qiagen). In 9/134 patients, DNA was extracted from FFPE tissue samples using the QIAamp

DNA FFPE Tissue Kit. In cases of facial involvement (15/134 patients), a buccal swab served as

affected tissue, and DNA was extracted via standard methods of phenol-chloroform

extraction followed by ethanol precipitation. Blood samples were collected where possible,

and lymphocyte DNA was extracted via the Illustra BACC3 DNA extraction kit (GE Healthcare).

Targeted next generation sequencing was performed on affected tissue DNA using a custom

panel of overgrowth-related genes on an Illumina MiSeq platform. This panel was designed

for the low flow study as a selection of possible candidate genes for overgrowth, on the basis

of the genes/pathways already known to be involved in this phenotype. 10 ng of DNA was

amplified for 18 cycles of PCR with the Ion Ampliseq custom DNA panel, enriching the 195

target amplicons. This included full coverage of all coding regions of PIK3CA, PTEN and CCND2,

and “hotspot” regions in 57 other genes (regions of coverage listed in Supplementary Table

S2). The panel was split into two primer pools and amplified with 5x Ion Ampliseq HiFi master

mix, followed by FuPa treatment (Ion Ampliseq DNA Library Kit 2.0). The two primer pools for

each sample were then pooled, and purified with 1.8x Agencourt AMPure XP magnetic beads

(Beckman Coulter). Amplicons were 3’ adenylated using the NebNext Ultra II end repair/dA

tailing module (NEB), followed by NextFlex DNA Barcode adapter (Bio Scientific) ligation using

the NebNext Ultra II ligation module (NEB). Ligation products were purified and size-selected

using 0.8x Agencourt AMPure XP beads. Library concentration was determined with Kapa

Biosystems Library qPCR quantification kit on the Lightcycler 480 real-time PCR system

(Roche). Libraries were subsequently diluted to 2 nM and pooled in equimolar amounts.

Pooled libraries were spiked with 1% PhiX DNA (Illumina), and sequenced on the MiSeq

desktop sequencer using version 2 chemistry at 250 bp read length paired end.

VCF files tailored for mosaic variant calling were created in MiSeq Reporter (Illumina) and

annotated in Illumina Variant Studio version 2.2, resulting in a list of 300-1,000 variants per

sample. The programming language R was used to apply hard filters according to the following

parameters: read depth >5; quality score >10; absence of strand bias (as determined by MiSeq

Reporter), cross-sample subtraction of artefactual variants called in >8 samples per batch of

24; exonic non-synonymous variants only. This resulted in a list of 1-4 candidate variants, the

clinical relevance and sequencing quality of which were then assessed.

Mosaic variants considered to be causative were confirmed and tested in other tissue samples

from the same patient, alongside DNA from healthy controls, either by Sanger sequencing

(primers listed in Supplementary Table S1) or by custom RFLP. For RFLP, genomic DNA was

amplified with GoTaq Green (Promega) using the primers listed in Supplementary Table S3.

The PCR products were designed to include a restriction enzyme recognition site allowing

specific digestion of the mutant allele, while leaving the wildtype allele intact. The digested

PCR fragments were then mixed with GeneScan™ 500 LIZ® Size Standard (Applied Biosystems)

and loaded on an ABI3730 capillary sequencer. The area under the curve of

undigested:digested DNA was used to calculate the mutation burden using Genemapper v5.0

software (Applied Biosystems).

Mutant plasmid construction, HEK293T and HUVEC cell transfection

pCMV6-MAP2K1:

The pCMV6-MAP2K1 plasmid for in vivo expression in mammalian cells was ordered from

Origene (ID: SC118424). To generate mutant MAP2K1 expression vector, an improved

QuickChangeTM site-directed mutagenesis protocol was used(50). p.(K57N) is a predicted

missense alteration caused by a single nucleotide change (g>c); p.(Q58_E62del) (hereafter

termed Q58del) is a deletion mutation caused by a deletion of 15 bp. The sequences of

mutagenesis PCR primers are: K57N-Fw: 5'-gaggcctttcttacccagaaccagaaggtggg-3’ K57N-Rev:

5'-cccaccttctggttctgggtaagaaaggcctc-3' Q58del-Fw: 5’-

cttacccagaagctgaaggatgacgactttgagaagatcag-3' Q58del-Rev: 5’-

gtcgtcatccttcagcttctgggtaagaaaggcctcaagg-3' The mutagenesis PCR runs for 15 cycles and

each cycle consists of denaturation at 95 qC for 30 sec, annealing at 58 qC for 1 min and

extension at 72 qC for 5 min.

pDEST26-BRAF:

The wildtype and mutant (p.V600E) human BRAF cDNA are in the same Gateway� middle

entry clones used in a previous study(51). These two BRAF cDNAs were cloned into the

destination vector pDEST26 for expression in mammalian cells using the Gateway� cloning

system (Invitrogen) according to the manufacturers instructions.

HEK293T cell line transfection:

HEK293T cells (ATCC, Catalogue Number CRL-11268) were maintained as per established

protocols and were transfected with mutant and wildtype cDNA expression plasmids, and

an empty vector control, using Lipofectamine® 2000. One day before transfection, 6 x

105 cells were plated per well of a 6-well culture vessel in 500 μl of growth medium without

antibiotics so that cells would be 70-90% confluent at the time of transfection. For each

transfection sample, 4.0 μg of plasmid DNA was diluted in 250 μl of Opti-MEM® I Reduced

Serum Medium and mixed gently. Lipofectamine® 2000 was gently mixed before use, then

10 μl was diluted in 250 μl of Opti-MEM® I Medium, then incubated for 5 minutes at room

temperature before combining with the diluted DNA (total volume = 500 μl), mixed gently

and incubated for 20 minutes at room temperature. 500 μl of complexes were added to

each well containing cells and medium and mixed gently by rocking the plate back and forth.

Finally cells were incubated cells at 37 °C in a CO2 incubator for 30 hours prior to testing for

transgene expression.

HUVEC cells transfection :

Based on transfection optimisation Human Umbilical Vein Endothelial Cells (HUVEC,

(ThermoFisher, catalogue number C0035C) were incubated in transfection reagents with a

1:3 DNA:Lipofectamine®LTX ratio at 37 °C, 5% (v/v) CO2 for 48 h. Volumes described in this

section are representative of a single well in a 6-well plate. 2.5 µg of either plasmids

containing the wild-type (WT) or mutated gene of interest (BRAFWT, BRAFV600E, MAP2K1WT,

MAP2K1K57N and MAP2K1Q58_E62del as detailed above), or pDest26 empty vector or a mock

transfection that represented an additional control was diluted in 500 µl in Opti-MEM®I

Reduced Serum Medium without serum (#31985062). Next, 2.5 µl of PLUS™Reagent

(#11514015) was added to the DNA dilution and incubated for 10 minutes at room

temperature. Transfection complexes were formed after adding 7.5 µl Lipofectamine®LTX to

DNA:PLUS™Reagent solution prior to 30 minutes incubation at room temperature and then

added to a well containing 2 ml fresh growth supplemented medium.

Quantitative real time PCR

To assess the efficiency of transfection of plasmid DNA in the cells, quantitative expression

analysis of the genes of interest MAP2K1 and BRAF as well as the endogenous control GAPDH

was determined by real-time qPCR with TaqMan gene expression assays using the StepOne

plus instrument. Standard protocol as per manufacturer’s guidelines for Applied Biosystems

TaqMan Gene Expression Assay on the StepOne plus instrument. Probes used for quantitative

analysis were: MAP2K1 (Assay ID: Hs00983247_g1), BRAF (Assay ID: Hs00269944_m1) and

GAPDH (Assay ID: Hs02786624_g1) and were all from Life Technologies). Results of the

quantitative gene expression levels were obtained after the amplification reaction using the

StepOne software (version 2.3). Results are based on analysis of two biological replicates,

with triplicate technical replicates in each experiment, and quantitative values of the cycle

threshold (Ct) averaged. The relative expression of the genes of interest was determined by

calculating the ratio of their expression compared to that of the GAPDH endogenous control

in the same sample.

Western Blotting

Cell lysates were prepared using standard protocols. Primary and secondary antibodies are

shown in Supplementary Table S4. Western Blot basic protocol: 40ug of protein was used

after heating at 95 °C with 2x Laemmli sample buffer in 5% Beta-mercaptoethanol was run on

4–20% Mini-PROTEAN® TGX™ Precast Protein Gels (#4561096 BIORAD). The transfer was

blocked at room temperature for 1hr in 5% dry milk or 5% BSA dissolved in TBS-0.05%Tween.

The transfer was incubated with primary antibody (1:1000 dilution in 3% BSA (BSA in 1xTBST)

at room temperature overnight. The membrane was then washed 3 times for 15 minutes in

TBST. Incubation with the secondary antibody was conducted at room temperature for 1hr.

As a secondary antibody we used anti rabbit RB96 in 1:7000 dilution and diluted in 3% BSA

(BSA in 1xTBST) or 3% milk.

Endothelial cell tube formation assay and microscopy

Cells and all reagents in this subsection were obtained from Invitrogen (Paisley, UK) unless

otherwise stated. The formation of endothelial tubes by Human Umbilical Vein Endothelial

Cells (HUVEC, (ThermoFisher, catalogue number C0035C) on growth factor-reduced Geltrex®

(#A1413202) was conducted according to the manufacturer’s protocol (#MAN0001687) using

Medium 200PRF supplemented with Low Serum Growth Supplement. Briefly, 24 well culture

plates were coated with 100 µl/well (50 µl/ cm2) Geltrex® and incubated for 30 minutes at 37

°C. After harvesting transfected and untreated HUVECs (P2) cultured on 6 well plates, they

were seeded on coated plates at a density of 4.5x104/cm2 (9x104 cells/well) in 200 µl/cm2

(400µl/well) supplemented 200PRF medium and cultured in a CO2 incubator (37 °C, 5% (v/v)

CO2, 95% humidity). 30 minutes before the end of the incubation period cells were treated

with 2 µg/ml (0.8μl/well) calcein AM (#C3099) and incubated at 37 °C, 5% (v/v) CO2. Tube

formation observed at 14 h time point was imaged with a 5x objective lens of Olympus IX71

inverted fluorescence and bright field microscope using HCImage software. The degree of

tube formation was assessed by measuring all aspects of tubule, node and mesh growth in

triplicate, using randomly chosen fields from each well using the angiogenesis analyzer for

ImageJ (http://image.bio.methods.free.fr/ImageJ/?Angiogenesis-Analyzer-for-ImageJ).

Transgenic zebrafish

The zebrafish fli1a promoter was generated by gateway® PCR with primers (forward: 5’-

GGGGACAACTTTGTATAGAAAAGTTGCCTGGCTGTCAAGCTCCAGC-3’, reverse:

GGGGACTGCTTTTTTGTACAAACTTGATATGTGGCGGAGAGACAGAG-3’, promoter specific

sequences are underlined). The promoter comprises 2.2 kb immediately upstream of the

first ATG of the fli1a gene. The PCR product was then clone into the gateway® 5’ donor

vector pDONRP4-P1R to obtain the 5’ entry clone p5Efli1a2.2k. Middle entry clones

containing human BRAFwt, BRAFV600E, MAP2K1wt or MAP2K1Q58del cDNA were recombined

with the fl1a promoter in p5Efli1a2.2k and the pDestTol2CG2 expression vector using the

Tol2kit gateway® cloning method(52), resulting fli1a-BRAFwt, fli1a-BRAFV600E , fli1a-

MAP2K1wt, fli1a- MAP2K1Q58del constructs. One nl of mixed fli1a-BRAF or fli1a-MAP2K1

plasmid DNA and Tol2 mRNA (37 ng/µl and 35 ng/µl respectively) was injected into 1-cell

stage of fli1a:GFP zebrafish embryos (Species Danio rerio, AB line)(53). Embryos were

raised at 28.5 °C and screened for phenotypes. Embryos were then fixed in 4% PFA.

Drug treatments

Zebrafish embryos were treated with the BRAF inhibitor Vemurafenib (PLX4032). Embryos

were incubated with 0.1 µM of the drug from 2.5 dpf after initial imaging, refreshed daily.

Control embryos were incubated in E3 with DMSO at the same concentration as drug used.

During live imaging, embryos were kept in 1:5,000 MS222 and 1.5% LMP agar. Leica stereo

brightfield microscopy was used for live colour imaging. Fluorescence images were taken

using Leica Sp5 confocal microscopy. Images were processed using FIJI(ImageJ). Graph pad

prism was used to analyse data.

Statistics

Western blot (WB) data (Figure 4a,b) was analysed using one-way ANOVA comparing mock

transfection to wild-type and mutant allele groups, per mutant. Densitometry data was

pooled from biological replicates and used to calculate means and SDs. A p-value of <0.05

was used for 95% confidence. Angiogenesis data (Figure 4e,f,g) was analysed using one-way

ANOVA comparing mock transfection controls to wild-type and mutant allele groups, per

mutant. An initial one-way ANOVA was undertaken to demonstrate that biological replicates

were significantly different, and therefore data between replicates was first standardized to

the within-experiment control before being pooled for analysis of standardized means and

SD. Correction for multiple testing was applied after the ANOVA, reducing the 95%

confidence p-value to <0.0167.

Zebrafish data – significance was assessed for data in Figure 5b using unpaired parametric t-

test with Welch’s correction, and in Figure 5c using paired t-test.

Study approval

These studies were conducted according to Declaration of Helsinki principles, and were

approved by the local Research Ethics Committees of each centre involved (London

Bloomsbury, University of Cambridge and University of Edinburgh). All participants were

recruited with written informed consent. Separate written informed consent was obtained

for publication of all clinical photographs.

All zebrafish work was done in accordance with United Kingdom Home Office Animals

(Scientific Procedures) Act (1986) and approved by the University of Edinburgh Ethical Review

Committee.

Acknowledgements

We gratefully acknowledge the participation of all patients and families in this study. We

gratefully acknowledge Craig Nicol’s help with figure preparation. VAK and RKS are funded by

the Wellcome Trust (Grants WT104076MA and WT098498 respectively). Support was also

provided by the Medical Research Council [MRC_MC_UU_12012/5] and by the UK National

Institute for Health Research (through Biomedical Research Centres at Great Ormond Street

Hospital for Children NHS Foundation Trust, University College London and in Cambridge, and

through the Rare Disease Translational Research Collaboration). KA is funded by Health

Education East of England and the Wellcome Trust Translational Medicine and Therapeutics

programme. EEP is funded by the Medical Research Council, L’Oreal-Melanoma Research

Alliance Team Award for Women in Science, and the European Research Council. LGB is

funded by the Intramural Research Program of the National Human Genome Research

Institute (Grants HG200328 11 and HG200388 03).

Author contributions

LA-O and SP were responsible for patient sample DNA extraction, sequencing and

confirmation, patient recruitment and phenotyping of the low and high flow cohorts,

HEK293T cell work, and figure preparation. PS was responsible for the HUVEC cell work and

figure preparation. KAA and VEP were responsible for patient recruitment and phenotyping

of the low-flow cohort. RK, GC and WB contributed to sequencing and cell culture

maintenance. KD, ZZ and EEP were responsible for the zebrafish work and figure preparation.

AJ and MT were responsible for the protein modelling work and figure preparation. AP and

JBR contributed to the design of the sequencing data analysis. Authors listed in the

alphabetical section of the authorship contributed patient samples and phenotypic data to

the cohort, and for HJ, KMKN, MJL and JS also to the mutation discovery of their patient. LGB

contributed patient genotype/phenotype and critical review of the manuscript. AB reviewed

all the imaging and contributed radiological images for figures. RKS conceived, designed and

directed the research for the low-flow cohort, as well as contributing a large number of

patients and the genotypic and phenotypic data to that cohort, with contribution to and

critical review of the manuscript. EEP conceived, designed and directed the zebrafish

research, with contribution to and critical review of the manuscript. VAK conceived,

designed, directed the research and contributed patients to the high-flow cohort, contribute

patients to the low-flow cohort, analysed the sequencing and angiogenesis data, wrote the

manuscript and prepared figures.

Table 1 - Phenotypic and genotypic characterisation of 15 individuals with M

APK pathway variants and sporadic vascular m

alformations.

VM – vascular m

alformation; AVM

– arteriovenous malform

ation; R – right; L = left; FFPE – formalin-fixed paraffin-em

bedded. Patient

# Age

Phenotype gDN

A (hg19) cDN

A, aa change M

utant allele count

Total allele count

Percentage m

osaicism by

sample origin

1 19

years High flow

AVM

L face, recurrent bleeding,

progressive enlargem

ent

chr7:g.140453136T>A N

M_004333.4: c.1799T>A;

BRAF p.(V600E) 27

103 FFPE tissue

from vascular

malform

ation tissue from

surgical

resection 26%

Blood 0%

2

33 years

High flow

AVM R face,

recurrent bleeding, progressive

enlargement

chr12:g.25380275A>C

NM

_004985.4; NM

_033360.3 :c.183A>C;

KRAS p.(Q61H)

114 2241

FFPE tissue from

vascular m

alformation

tissue from

surgical resection of VM

5%

Blood 0%

3

12 years

High flow

AVM R face,

recurrent bleeding, loss of vision R eye,

progressive enlargem

ent

chr12:g.25398284G>T

NM

_004985.4; NM

_033360.3: c.35G>T;

KRAS p.(G12V)

5 174

Fresh skin biopsy from

VM

3%

Blood 0%

4 24

years High flow

AVM

L buttock, recurrent bleeding,

progressive enlargem

ent

chr15: g.66727443

NM

_002755.3: c.159_173delTCTTACCCAGAAGCA;

MAP2K1 p.(F53_Q

58delinsL)

44 752

FFPE tissue from

vascular m

alformation

tissue from

surgical resection of

vascular m

alformation

tissue 6%

5 20

years High flow

AVM

R temple area,

recurrent bleeding, progressive

enlargement

chr15:g.66727455G>C

NM

_002755.3 : c.171G>C; M

AP2K1 p.(K57N)

10 136

FFPE tissue from

vascular m

alformation

tissue from

surgical resection 7%

6

9 years

High flow

AVM R external ear,

and posterior auricular soft tissues

chr15:g.66727456

NM

_002755.3: c.173_187delAGAAGGTG

GGAGAAC; M

AP2K1 p.(Q58_E62del)

31 782

FFPE tissue from

vascular m

alformation

tissue from

surgical resection 4%

Blood 0%

7A

24 years

High flow

Deep vessel mixed

arterial and venous m

alformation of the

left posterior chest w

all and overlying skin. Extensive

chr12:g.25398284G>A N

M_004985.4; N

M_033360.3:

KRAS c.35G>A; KRAS, p.(G12D)

Multiple

samples

Cultured

fibroblasts from

skin biopsy of epiderm

al nevus 39%

asymm

etric linear verrucous

keratinocytic epiderm

al nevus of the back w

ith a few

areas suggestive of sebaceous nevus.

Hypertension caused by bilateral renal

artery stenosis, with

long strictures of the descending aorta,

coeliac axis and superior m

esenteric artery also

abnormal. Surgery

for removal of

symptom

atic spinal root plexiform

neurofibrom

as on tw

o occasions and an intra-spinal lipom

a on one occasion. Post-

surgical kyphoscoliosis.

Mutations in NF1

and PTEN genes excluded.

Fresh skin biopsy of epiderm

al nevus 30%

Paraspinal ‘N

eurofibroma-

like’ overgrow

th 30%

8 13

years High flow

AVM

L external ear, posterior auricular

soft tissues, intracranial extension,

progressive enlargem

ent

chr15:g.66727455G>C

NM

_002755.3 : c.171G>C; M

AP2K1 p.(K57N)

18 992

Fresh skin biopsy 2%

9 14 years

High flow

AVM in the right

frontal lobe, supplied by a branch of the right anterior

cerebral artery. Presented aged 13

with intracranial

haemorrhage and

intraventricular extension,

previously well.

Required craniotomy

and resection of AVM

.

chr12:25398285G>T N

M_004985.4; N

M_033360.3:

c.34G>T; KRAS, p.(G12C)

23 841

FFPE tissue 3%

10 18

years Low

flow

Mild overgrow

th of right arm

and shoulder girdle from

age 10, w

ith prom

inent superficial veins and

chr12:g.25398284G>A N

M_004985.4; N

M_033360.3:

c.35G>A; KRAS p.(G12D)

275 7628

Fresh skin biopsy of capillary

malform

ation 4%

Blood – 0%

reticulate capillary m

alformation on the

right upper arm. N

o deep vessel

malform

ation detectable, chronic

pain in palpably w

arm right hand,

with brow

nish-pink uniform

capillary m

alformation and

scattered telangiectasia.

11 39

years Low

flow

Subtle congenital right foot

overgrowth w

hich gradually progressed

and extended to buttock by second decade. Prom

inent superficial veins and dependent edem

a from

20 years old, still progressing. N

o capillary

malform

ation.

chr1:g.114713908A>G

N

M_002524.4:c.182A>G;

NRAS p.(Q61R)

524 7715

Fresh skin biopsy of VM

7%

Blood – 0%

Saliva – 0%

12 14

years Low

flow

Non-progressive

congenital mild

chr12:g.25398284G>T

NM

_004985.4; NM

_033360.3: c.35G>T;

KRAS p.(G12V)

193 7105

Fresh skin biopsy of VM

3%

overgrowth w

ith co-localised capillary

malform

ation of the left arm

, hand, shoulder girdle, chest w

all and breast w

ith clear m

idline dem

arcation. The capillary

malform

ation was

uniform brow

nish-red and punctuated

by telangiectasia w

ith a pale halo. The overgrow

n area was

painful and palpably w

arm, but

hematological

indices and deep vessels w

ere normal.

Blood – 0%

13 7 years

Low flow

O

vergrowth of the

third and fourth fingers on the right hand (length and breadth), forearm

sm

aller in girth on R than L. Venous

prominence on the

chr12:g.25398284G>A N

M_004985.4; N

M_033360.3:

c.35G>A: KRAS p.(G12D)

Cultured fibroblasts from

skin biopsy of

capillary m

alformation

29%

Blood – 0%

right forearm and

hand, and confluence of

hyperpigmentation

and increased vascularity of the

right chest, shoulder and back. Pain on

palpation of his right third finger. O

ne fracture of his left

distal radius. Skeletal X-rays

showed

enlargement of his

phalanges and m

etacarpals of the right third and

fourth fingers, and m

ultifocal osteal lesions described as

lucent, expansile, and lytic lesions

involving the fingers, foot, tibia, hum

erus, fem

ur and clavicle on the right. N

eedle biopsy and surgical

pathology of the right tibia, including

small fragm

ents of bone associated

with scant fibrous

tissue and skeletal m

uscle, which

showed spindle and giant cell

proliferation diagnosed as nonossifying

fibroma. Clinical

genetic testing of the germ

line with a

neurofibromatosis

panel and whole

exome sequencing

were negative.

14

10 m

onths

Low flow

W

idespread linear keratinocytic

epidermal nevus in a

Blaschko-linear distribution,

localised to the right side except on the

mid back w

here there is a sm

all streak on the left.

Affects trunk and leg

chr12:g.25398284G>T

KRAS c.35G>T: KRAS p.(G12V)

177 3328

Skin biopsy from

capillary m

alformation

5%

Skin biopsy epiderm

al nevus 13%

but not head. M

oderate right-sided

hemihypertrophy

mainly affecting the

leg. Capillary vascular

malform

ation apparent on right foot and buttock. N

evus simplex on

glabella and nape of neck.

15

36 years

Low flow

Left low

er limb

hypertrophy requiring epiphyseal

fusion, lym

phoedema,

varicose veins with

previous recurrent DVT and cutaneous

vascular m

alformation.

Complicated by

recurrent cellulitis and left hip pain.

chr12:g.25398284G>A N

M_004985.4; N

M_033360.3:

c.35G>A: KRAS p.(G12D)

200 9510

Affected skin biopsy from

left low

er limb 2%

References

1. Wassef M, Blei F, Adams D, Alomari A, Baselga E, Berenstein A, Burrows P, Frieden IJ, Garzon MC, Lopez-Gutierrez JC, et al. Vascular Anomalies Classification: Recommendations From the International Society for the Study of Vascular Anomalies. Pediatrics. 2015;136(1):e203-14.

2. Lindhurst MJ, Sapp JC, Teer JK, Johnston JJ, Finn EM, Peters K, Turner J, Cannons JL, Bick D, Blakemore L, et al. A mosaic activating mutation in AKT1 associated with the Proteus syndrome. The New England journal of medicine. 2011;365(7):611-9.

3. Shirley MD, Tang H, Gallione CJ, Baugher JD, Frelin LP, Cohen B, North PE, Marchuk DA, Comi AM, and Pevsner J. Sturge-Weber syndrome and port-wine stains caused by somatic mutation in GNAQ. The New England journal of medicine. 2013;368(21):1971-9.

4. Thomas AC, Zeng Z, Riviere JB, O'Shaughnessy R, Al-Olabi L, St-Onge J, Atherton DJ, Aubert H, Bagazgoitia L, Barbarot S, et al. Mosaic Activating Mutations in GNA11 and GNAQ Are Associated with Phakomatosis Pigmentovascularis and Extensive Dermal Melanocytosis. The Journal of investigative dermatology. 2016;136(4)(770-8.

5. Riviere JB, Mirzaa GM, O'Roak BJ, Beddaoui M, Alcantara D, Conway RL, St-Onge J, Schwartzentruber JA, Gripp KW, Nikkel SM, et al. De novo germline and postzygotic mutations in AKT3, PIK3R2 and PIK3CA cause a spectrum of related megalencephaly syndromes. Nature genetics. 2012;44(8):934-40.

6. Lindhurst MJ, Parker VE, Payne F, Sapp JC, Rudge S, Harris J, Witkowski AM, Zhang Q, Groeneveld MP, Scott CE, et al. Mosaic overgrowth with fibroadipose hyperplasia is caused by somatic activating mutations in PIK3CA. Nature genetics. 2012;44(8):928-33.

7. Kurek KC, Luks VL, Ayturk UM, Alomari AI, Fishman SJ, Spencer SA, Mulliken JB, Bowen ME, Yamamoto GL, Kozakewich HP, et al. Somatic mosaic activating mutations in PIK3CA cause CLOVES syndrome. American journal of human genetics. 2012;90(6):1108-15.

8. Couto JA, Vivero MP, Kozakewich HP, Taghinia AH, Mulliken JB, Warman ML, and Greene AK. A somatic MAP3K3 mutation is associated with verrucous venous malformation. American journal of human genetics. 2015;96(3):480-6.

9. Couto JA, Huang AY, Konczyk DJ, Goss JA, Fishman SJ, Mulliken JB, Warman ML, and Greene AK. Somatic MAP2K1 Mutations Are Associated with Extracranial Arteriovenous Malformation. American journal of human genetics. 2017;100(3):546-54.

10. Limaye N, Wouters V, Uebelhoer M, Tuominen M, Wirkkala R, Mulliken JB, Eklund L, Boon LM, and Vikkula M. Somatic mutations in angiopoietin receptor gene TEK cause solitary and multiple sporadic venous malformations. Nature genetics. 2009;41(1):118-24.

11. Soblet J, Kangas J, Natynki M, Mendola A, Helaers R, Uebelhoer M, Kaakinen M, Cordisco M, Dompmartin A, Enjolras O, et al. Blue Rubber Bleb Nevus (BRBN) Syndrome Is Caused by Somatic TEK (TIE2) Mutations. The Journal of investigative dermatology. 2017;137(1):207-16.

12. Gripp KW, Baker L, Kandula V, Conard K, Scavina M, Napoli JA, Griffin GC, Thacker M, Knox RG, Clark GR, et al. Nephroblastomatosis or Wilms tumor in a fourth patient with

a somatic PIK3CA mutation. American journal of medical genetics Part A. 2016;170(10):2559-69.

13. Nathan N, Keppler-Noreuil KM, Biesecker LG, Moss J, and Darling TN. Mosaic Disorders of the PI3K/PTEN/AKT/TSC/mTORC1 Signaling Pathway. Dermatol Clin. 2017;35(1):51-60.

14. Mirzaa G, Timms AE, Conti V, Boyle EA, Girisha KM, Martin B, Kircher M, Olds C, Juusola J, Collins S, et al. PIK3CA-associated developmental disorders exhibit distinct classes of mutations with variable expression and tissue distribution. JCI insight. 2016;1(9).

15. Thomas AC, Zeng Z, Riviere JB, O'Shaughnessy R, Al-Olabi L, St-Onge J, Atherton DJ, Aubert H, Bagazgoitia L, Barbarot S, et al. Mosaic Activating Mutations in GNA11 and GNAQ Are Associated with Phakomatosis Pigmentovascularis and Extensive Dermal Melanocytosis. The Journal of investigative dermatology. 2016;136(4):770-8.

16. Fischmann TO, Smith CK, Mayhood TW, Myers JE, Reichert P, Mannarino A, Carr D, Zhu H, Wong J, Yang RS, et al. Crystal structures of MEK1 binary and ternary complexes with nucleotides and inhibitors. Biochemistry. 2009;48(12):2661-74.

17. Gore AV, Monzo K, Cha YR, Pan W, and Weinstein BM. Vascular development in the zebrafish. Cold Spring Harb Perspect Med. 2012;2(5):a006684.

18. Lawson ND, and Weinstein BM. Arteries and veins: making a difference with zebrafish. Nat Rev Genet. 2002;3(9):674-82.

19. Castillo SD, Tzouanacou E, Zaw-Thin M, Berenjeno IM, Parker VE, Chivite I, Mila-Guasch M, Pearce W, Solomon I, Angulo-Urarte A, et al. Somatic activating mutations in Pik3ca cause sporadic venous malformations in mice and humans. Sci Transl Med. 2016;8(332):332ra43.

20. Shenkar R, Shi C, Austin C, Moore T, Lightle R, Cao Y, Zhang L, Wu M, Zeineddine HA, Girard R, et al. RhoA Kinase Inhibition With Fasudil Versus Simvastatin in Murine Models of Cerebral Cavernous Malformations. Stroke. 2017;48(1):187-94.

21. Wetzel-Strong SE, Detter MR, and Marchuk DA. The pathobiology of vascular malformations: insights from human and model organism genetics. J Pathol. 2017;241(2):281-93.

22. Sahoo T, Johnson EW, Thomas JW, Kuehl PM, Jones TL, Dokken CG, Touchman JW, Gallione CJ, Lee-Lin SQ, Kosofsky B, et al. Mutations in the gene encoding KRIT1, a Krev-1/rap1a binding protein, cause cerebral cavernous malformations (CCM1). Human molecular genetics. 1999;8(12):2325-33.

23. Liquori CL, Berg MJ, Siegel AM, Huang E, Zawistowski JS, Stoffer T, Verlaan D, Balogun F, Hughes L, Leedom TP, et al. Mutations in a gene encoding a novel protein containing a phosphotyrosine-binding domain cause type 2 cerebral cavernous malformations. American journal of human genetics. 2003;73(6):1459-64.

24. Bergametti F, Denier C, Labauge P, Arnoult M, Boetto S, Clanet M, Coubes P, Echenne B, Ibrahim R, Irthum B, et al. Mutations within the programmed cell death 10 gene cause cerebral cavernous malformations. American journal of human genetics. 2005;76(1):42-51.

25. Akers AL, Johnson E, Steinberg GK, Zabramski JM, and Marchuk DA. Biallelic somatic and germline mutations in cerebral cavernous malformations (CCMs): evidence for a two-hit mechanism of CCM pathogenesis. Human molecular genetics. 2009;18(5):919-30.

26. Vikkula M, Boon LM, Carraway KL, 3rd, Calvert JT, Diamonti AJ, Goumnerov B, Pasyk KA, Marchuk DA, Warman ML, Cantley LC, et al. Vascular dysmorphogenesis caused

by an activating mutation in the receptor tyrosine kinase TIE2. Cell. 1996;87(7):1181-90.

27. Amyere M, Aerts V, Brouillard P, McIntyre BA, Duhoux FP, Wassef M, Enjolras O, Mulliken JB, Devuyst O, Antoine-Poirel H, et al. Somatic uniparental isodisomy explains multifocality of glomuvenous malformations. American journal of human genetics. 2013;92(2):188-96.

28. Zhou XP, Marsh DJ, Hampel H, Mulliken JB, Gimm O, and Eng C. Germline and germline mosaic PTEN mutations associated with a Proteus-like syndrome of hemihypertrophy, lower limb asymmetry, arteriovenous malformations and lipomatosis. Human molecular genetics. 2000;9(5):765-8.

29. Revencu N, Boon LM, Mendola A, Cordisco MR, Dubois J, Clapuyt P, Hammer F, Amor DJ, Irvine AD, Baselga E, et al. RASA1 mutations and associated phenotypes in 68 families with capillary malformation-arteriovenous malformation. Human mutation. 2013;34(12):1632-41.

30. Amyere M, Revencu N, Helaers R, Pairet E, Baselga E, Cordisco M, Chung W, Dubois J, Lacour JP, Martorell L, et al. Germline Loss-of-Function Mutations in EPHB4 Cause a Second Form of Capillary Malformation-Arteriovenous Malformation (CM-AVM2) Deregulating RAS-MAPK Signaling. Circulation. 2017;136(11):1037-48.

31. Happle R. Lethal genes surviving by mosaicism: a possible explanation for sporadic birth defects involving the skin. Journal of the American Academy of Dermatology. 1987;16(4):899-906.

32. Szklarczyk D, Franceschini A, Wyder S, Forslund K, Heller D, Huerta-Cepas J, Simonovic M, Roth A, Santos A, Tsafou KP, et al. STRING v10: protein-protein interaction networks, integrated over the tree of life. Nucleic acids research. 2015;43(Database issue):D447-52.

33. Zhou Z, Tang AT, Wong WY, Bamezai S, Goddard LM, Shenkar R, Zhou S, Yang J, Wright AC, Foley M, et al. Cerebral cavernous malformations arise from endothelial gain of MEKK3-KLF2/4 signalling. Nature. 2016;532(7597):122-6.

34. Yang J, Boerm M, McCarty M, Bucana C, Fidler IJ, Zhuang Y, and Su B. Mekk3 is essential for early embryonic cardiovascular development. Nature genetics. 2000;24(3):309-13.

35. Cullere X, Plovie E, Bennett PM, MacRae CA, and Mayadas TN. The cerebral cavernous malformation proteins CCM2L and CCM2 prevent the activation of the MAP kinase MEKK3. Proceedings of the National Academy of Sciences of the United States of America. 2015;112(46):14284-9.

36. Fisher OS, Deng H, Liu D, Zhang Y, Wei R, Deng Y, Zhang F, Louvi A, Turk BE, Boggon TJ, et al. Structure and vascular function of MEKK3-cerebral cavernous malformations 2 complex. Nature communications. 2015;6(7937.

37. Deng Y, Yang J, McCarty M, and Su B. MEKK3 is required for endothelium function but is not essential for tumor growth and angiogenesis. Am J Physiol Cell Physiol. 2007;293(4):C1404-11.

38. Groesser L, Herschberger E, Ruetten A, Ruivenkamp C, Lopriore E, Zutt M, Langmann T, Singer S, Klingseisen L, Schneider-Brachert W, et al. Postzygotic HRAS and KRAS mutations cause nevus sebaceous and Schimmelpenning syndrome. Nature genetics. 2012;44(7):783-7.

39. Wang H, Qian Y, Wu B, Zhang P, and Zhou W. KRAS G12D mosaic mutation in a Chinese linear nevus sebaceous syndrome infant. BMC medical genetics. 2015;16(101.

40. Sun BK, Saggini A, Sarin KY, Kim J, Benjamin L, LeBoit PE, and Khavari PA. Mosaic activating RAS mutations in nevus sebaceus and nevus sebaceus syndrome. The Journal of investigative dermatology. 2013;133(3):824-7.

41. Igawa S, Honma M, Minami-Hori M, Tsuchida E, Iizuka H, and Ishida-Yamamoto A. Novel postzygotic KRAS mutation in a Japanese case of epidermal nevus syndrome presenting with two distinct clinical features, keratinocytic epidermal nevi and sebaceous nevi. The Journal of dermatology. 2016;43(1):103-4.

42. Farschtschi S, Mautner VF, Hollants S, Hagel C, Spaepen M, Schulte C, Legius E, and Brems H. Keratinocytic epidermal nevus syndrome with Schwann cell proliferation, lipomatous tumour and mosaic KRAS mutation. BMC medical genetics. 2015;16(6.

43. Peacock JD, Dykema KJ, Toriello HV, Mooney MR, Scholten DJ, 2nd, Winn ME, Borgman A, Duesbery NS, Hiemenga JA, Liu C, et al. Oculoectodermal syndrome is a mosaic RASopathy associated with KRAS alterations. American journal of medical genetics Part A. 2015.

44. Boppudi S, Bogershausen N, Hove HB, Percin EF, Aslan D, Dvorsky R, Kayhan G, Li Y, Cursiefen C, Tantcheva-Poor I, et al. Specific mosaic KRAS mutations affecting codon 146 cause oculoectodermal syndrome and encephalocraniocutaneous lipomatosis. Clinical genetics. 2016;90(4):334-42.

45. Aizawa K, Nakamura T, Ohyama Y, Saito Y, Hoshino J, Kanda T, Sumino H, and Nagai R. Renal artery stenosis associated with epidermal nevus syndrome. Nephron. 2000;84(1):67-70.

46. Alsohim F, Abou-Jaoude P, Ninet J, Pracros JP, Phan A, and Cochat P. Bilateral renal artery stenosis and epidermal nevus syndrome in a child. Pediatr Nephrol. 2011;26(11):2081-4.

47. Parent JJ, Bendaly EA, and Hurwitz RA. Abdominal coarctation and associated comorbidities in children. Congenit Heart Dis. 2014;9(1):69-74.

48. Groesser L, Peterhof E, Evert M, Landthaler M, Berneburg M, and Hafner C. BRAF and RAS Mutations in Sporadic and Secondary Pyogenic Granuloma. The Journal of investigative dermatology. 2015.

49. Kinsler VA, Thomas AC, Ishida M, Bulstrode NW, Loughlin S, Hing S, Chalker J, McKenzie K, Abu-Amero S, Slater O, et al. Multiple congenital melanocytic nevi and neurocutaneous melanosis are caused by postzygotic mutations in codon 61 of NRAS. The Journal of investigative dermatology. 2013;133(9):2229-36.

50. Zheng L, Baumann U, and Reymond JL. An efficient one-step site-directed and site-saturation mutagenesis protocol. Nucleic acids research. 2004;32(14):e115.

51. Anastasaki C, Estep AL, Marais R, Rauen KA, and Patton EE. Kinase-activating and kinase-impaired cardio-facio-cutaneous syndrome alleles have activity during zebrafish development and are sensitive to small molecule inhibitors. Human molecular genetics. 2009;18(14):2543-54.

52. Kwan KM, Fujimoto E, Grabher C, Mangum BD, Hardy ME, Campbell DS, Parant JM, Yost HJ, Kanki JP, and Chien CB. The Tol2kit: a multisite gateway-based construction kit for Tol2 transposon transgenesis constructs. Developmental dynamics : an official publication of the American Association of Anatomists. 2007;236(11):3088-99.

53. Lawson ND, and Weinstein BM. In vivo imaging of embryonic vascular development using transgenic zebrafish. Dev Biol. 2002;248(2):307-18.

Figure 1 – A broad clinical spectrum of vascular malformations in somatic RAS/MAPK mutations Inexorable enlargement of high-flow arteriovenous malformations with age, affecting the face and leading to loss of vision in the right eye (a-d), and the right ear helix and posterior auricular soft tissues leading to eventual resection of the helix (e-g). Varied clinical examples of the spectrum of high-flow VMs of the temple, left leg/buttock and left face (h,i). Segmental overgrowth of the left arm, chest wall and breast with a co-localised low-flow VM, detectable by a uniform brownish-pink macular capillary malformation and superimposed scattered telangiectasia, with clear midline demarcation (j-l). Segmental overgrowth of the right arm and hand, with a co-localised uniform brownish- pink capillary malformation with superimposed scattered telangiectasia (m,n).

a b c d

e f g h i

j k

l

m n

Fig. 1

Figure 2 – Imaging of sporadic VMs secondary to mutations in MAPK pathway genes demonstrating involvement of all blood vessel sizes Lateral image from a digitally subtracted angiogram showing a leash of small, abnormal vessels shunting through a dense capillary bed to early-filling veins (a); axial contrast-enhanced fat-saturated T1 weighted MRI image showing a grossly enlarged right pinna and thickened posterior auricular soft tissues. The abnormal tissue is filled with multiple signal voids, representing enlarged abnormal vessels. The pinna enhances avidly. (b). Lateral image from a digitally subtracted angiogram, with the catheter tip in the grossly enlarged left internal maxillary artery which supplies a leash of abnormal high flow vessels in the face (c). Thermography of low-flow VMs with overgrowth of the left chest wall and arm demonstrates increased temperature (shown in magenta) compared to the right-sided structures (d), and in the right forearm and thumb compared to the left (f). 3D reconstruction of an abdominal CT angiogram demonstrating multifocal vascular disease, with severe stenoses of the descending aorta, coeliac axis, superior mesenteric artery origin, and right renal artery (e).

a b c

d e f

Fig. 2

Figure 3 – Somatic variants in MAP2K1 cluster in exon 2 and are predicted to destabilise the 3D structure of the inactive form of the kinase (a) Schematic representation of clustered somatic mutations in exon 2 of MAP2K1; (b) Low allele frequency mutations on IGV visualization of deep next generation sequencing data from VM tissue samples of three patients; (c) 3D structural modelling of an inhibitor(4BM)-bound form (PDB ID: 3EQG) of MAP2K1 demonstrating the mutated residues. Deletions are highlighted in magenta (53-58) and orange (residues 58-62, with residue 58 common to both in pink). Critical residue K57 which is substituted as a result of the missense mutation is shown using ball and stick representation, with the dashed line indicating a hydrogen-bond interaction with the beta-sheet of the kinase. Residues involved in interaction between helix A and the core kinase domain are also shown.

a

b

c

Fig. 3

Na

Helix-CL63

V60

K57

L92

F53E144 MG ADP

4BMF129G128Helix-A

Activation loop

393 aa residues

39336144331 67

p.F5

3_Q

58de

linsL

K57N

p.Q

58_E

62de

lD Kinase DomainNES

Figure 4 - Mutations MAPK pathway-encoding genes lead to activation of downstream signaling, and disruption of vascular endothelial tube formation in vitro (a,b) Expression of mutant BRAFV600E, MAP2K1K57N and MAP2K1Q58_E62del in HEK293T cells leads to significantly increased phosphorylation of ERK detected by immunoblotting (representative blot shown from duplicate biological replicates), compared to wild-type gene overexpression and controls. Error bars represent mean +/- SD. Significance was determined using one-way ANOVA, and p values <0.05 are indicated by an asterisk. (c) Expression of mutant BRAFV600E and MAP2K1K57N in HUVEC cells seeded onto Geltrex® Matrix leads to visible disruption of endothelial vascular tube formation compared to controls, with (d,e,f) significant reductions in mean number of master junctions, total length of tubes, and total mesh area indicated by asterisks. Error bars represent mean +/- SD. Means were taken from triple technical replicates, for each of duplicate biological replicates, standardised to within-replicate controls, and analysed by one way ANOVA with correction for multiple testing (p value for significance reduced to <0.0167).

ERKGAPDH

WT

MAP

2K1

K57N

MAP

2K1

Q58

_E62

del

MAP

2K1

WT

BRAF

BRAF

V60

0E

Empt

y ve

ctor

P-ERKGAPDH

WT

MAP

2K1

K57N

MAP

2K1

Empt

y ve

ctor

Empt

y ve

ctor

Q58

_E62

del

MAP

2K1

GAPDHMAP2K1

a b

c d

e

Fig. 4

0 0.5

1 1.5

2 2.5

3

Mean ratio phospho-ERK to total ERK

0 0.2 0.4 0.6 0.8

1 1.2 1.4 1.6 1.8

Empty vector

WT BRAF

BRAF V600E

Mean ratio phospho-ERK to total ERK

* * *

Empty vector

WTMAP2K1

K57N MAP2K1

Q58del MAP2K1

0 0.2 0.4 0.6 0.8

1 1.2 1.4

Mock BRAF WT BRAF V600E

0 0.2 0.4 0.6 0.8

1 1.2 1.4 1.6

Standardised mean number master junctions

Standardised mean totallength channels

Standardised mean totalarea meshes

Standardised mean number master junctions

Standardised mean totallength channels

Standardised mean totalarea meshes

MAPK2K1 WT MAPK2K1 K57N MAP2K1 Q58del Mock

* *

* * *

*

* * * *

Figure 5 – BRAF and MAP2K mutations induce VM phenotypes in zebrafish that respond to targeted therapy. (a) Schematic of zebrafish embryos injected with Tg(fli1a:GFP) at the one-cell stage generate zebrafish larvae that are mosaic for the transgene integration and expression. (b) Image of mosaic expression of Tg(fli1a:GFP) in a zebrafish vessel. (c) Images of zebrafish expressing wild type or mutant BRAFV600E in a mosaic fashion from a fli1a promoter in endothelial cells. Accumulation of blood is visible at the junction of the caudal vein and artery in the BRAFV600E

–expressing zebrafish and outlined in dashed red line. (d) Quantification of VM phenotype in BRAFWT (n=511) and BRAFV600E (n=779) -expressing zebrafish. (e) BRAFWT and BRAFV600E mosaic expression in stable Tg(fli1a:GFP) larvae to visualize all vessels in the zebrafish in the VM lesion. Increased numbers of vascular channels and disorganized architecture of VM lesions are clearly detectable. (f) Schematic of VM treatment protocol, and quantification of the percentage of VM BRAFV600E zebrafish with improved blood flow following treatment with vemurafenib. VM BRAFV600E zebrafish were randomized prior to DMSO (n=31) or vemurafenib (n=19) treatment, and blind scored. (g) Images of zebrafish expressing MAPK2K1Q58del in a mosaic fashion from a fli1a promoter in endothelial cells (n=5/37). Zebrafish larvae in 5b were analyzed by an unpaired parametric t-test with Welch’s correction, and in 5d by a paired t-test comparing matched pairs (before and after treatment). Error bars are SEM.

a

gf

Fig. 5

1 Day 2 3 4 5

Treatment Select VM phenotypes Image Injection

Perc

ent

0

20

40

60

80

100

no phenotype VM phenotype

BRAF n=511BRAFV600E n=779

***

% b

lood

flow

rest

ored

0

20

40

60

80

100

DMSO BRAFi

*** Tg(fli1a:GFP)

DMSO n=31BRAFi n=19

b

mosaic Tg(fli1:gfp) expression

BRAF V600E c d

e

MAP2K1Q58del

VM phenotypeTg(fli1a:GFP)

VM phenotypeno phenotype

Tg(fli1:gfp) DNA

Mos

aic

BRAF

V6

00E

Control Mosaic

Con

trol Mosaic

Figure 6 - Schematic summary of the known and newly-identified signalling proteins affected by mosaic mutations which lead to a vascular malformation phenotype. Key signalling pathways PI3K-AKT-MTOR and RAS-RAF-MEK-ERK control cellular growth, apoptosis, and differentiation through complex transcriptional regulation. Multiple receptor types feed into one or both pathways. In addition, there is cross-talk between the two pathways at multiple levels (not shown). Proteins affected by the genetic mutations presented in this paper are shown in red, with previously identified sites shown in blue. Key classes of potential targeted therapeutics are shown in green boxes. RTK – receptor tyrosine kinase; GPCR – G-protein coupled receptor.

PTEN

PIP2

PIP3

AKT1

TSC1/2

PI3K

AKTInhibitors

MTORInhibitors

RASA1

RAS GTP

KRAS/NRAS

GPCR

GNAQ/GNA11

BRAF

RAS GDP

BRAFInhibitors

MEK 1/2(MAP2K1)

MEKInhibitors

REGULATION OF TRANSCRIPTION

mTORC1ERK1/2MAPK1

TIE2 (TEK) RTK

1

SUPPLEMENTARY MATERIAL

Figure S1 – the effects of BRAF overexpression on tail development in zebrafish larvae. (a) Image of an example of a zebrafish larvae expressing high levels of MAPK signaling, leading to a "stunting" phenotype, as previously described(1, 2) (b) Graph of zebrafish embryos expressing BRAF WT or V600E alleles. Both WT and mutant alleles lead to a stunted phenotype that confounds analysis of the vasculature in those animals, and were excluded from further analysis.

2

Figure S2 - In silico modelling using STRING of networks of genes implicated in human

vascular malformations gene products.

3

Table S1: Primers used for Sanger sequencing

Mosaic variant Forward primer sequence (5’ to 3’) Reverse primer sequence (5’ to 3’)

KRAS c.35G>A, p.Gly12Asp

AGCGTCGATGGAGGAGTTTG

ACAGAGAGTGAACATCATGGACC

MAP2K1 • c.171G>C,

p.(K57N) • c.159_173delTCTT

ACCCAGAAGCA, p.(F53_Q58delinsL)

• c.173_187delAGAAGGTGGGAGAAC, p.(Q58_E62del)

TGGGTTGACTTCTCTGGTGA

GAGACCTTGAACACCACACC

4

Table S2: Gene targets covered by the custom-designed 60 gene overgrowth sequencing

panel (AmpliSeq, Life Technologies).

195 amplicons ranging from 125-275 bp in size were designed to cover the codons of interest listed below, except in the case of PIK3CA, PTEN and CCND2, where full coverage of all coding regions is provided. Annotations are according to reference genome hg19. Where multiple transcript variants exist, the longest transcript variant is stated for the purpose of amino acid numbering.

Gene Regions covered Amino acids covered

AKT1 Chr14: 105246445 – 105246583 p.G16 – p.L52

Chr14: 105241352 – 105241535 p.E151 – p.K189

AKT2 Chr19: 40762852 – 40763002 p.G16 – p.L52

AKT3 Chr1: 243858913 – 243859106 p.G16 – p.L51

ALK Chr2: 29443490 – 29443714 p.K1173 – p.P1215

Chr2: 29436788 – 29437003 p.S1216 – p.H1247

Chr2: 29432582 – 29432804 p.D1249 – p.Y1278

BRAF Chr7: 140481298 – 140481511 p.K439 – p.H477

Chr7: 140453028 – 140453243 p.I582 – p.M620

CCND2 All coding regions

CDK2 Chr12: 56361535 – 56361762 p.E40 – p.V64

Chr12: 56362526 – 56362676 p.Y107 – p.L143

DEPTOR Chr8: 121018956 – 121019175 p.L310 – p.T332

EGFR Chr7: 55241601 – 55241801 p.L688 – p.K728

Chr7: 55248900 – 55249123 p.E762 – p.D807

Chr7: 55249121 – 55249200 p.D807 – p.K823

Chr7: 55259353 – 55259582 p.G824 – p.K875

ERBB2 Chr17: 37868167 – 37868373 p.N302 – p.R340

Chr17: 37880147 – 37880352 p.G737 – p.D769

ERBB3 Chr12: 56478786 – 56479009 p.R81 – p.A155

EZH2 Chr7: 148508619 – 148508841 p.H618 – p.E649

FGFR1 Chr8: 38285848 – 38286070 p.A121 – p.M149

Chr8: 38282094 – 38282318 p.R281 – p.L321

FGFR2 Chr10: 123279547 – 123279766 p.R251 – p.E295

Chr10: 123274672 – 123274890 p.P364 – p.V416

Chr10: 123257952 – 123258121 p.D523 – p.D558

FGFR3 Chr4: 1803433 – 1803647 p.G235 – p.C275

Chr4: 1806061 – 1806208 p.E362 – p.V413

FLT3 Chr13: 28610030 – 28610230 p.K438 – p.P472

Chr13: 28608204 – 28608422 p.Q569 – p.F612

Chr13: 28602195 – 28602418 p.K649 – p.S684

Chr13: 28592526 – 28592736 p.C807 – p.N847

GNA11 Chr19: 3118790 – 3118993 p.M203 – p.I226

5

GNAQ Chr9: 80409380 – 80409598 p.M203 – p.E245

GNAS Chr20: 57484401 – 57484626 p.D839 – p.F862

HRAS Chr11: 534102 – 534306 p.K5 – p.E37

Chr11: 533780 – 533936 p.R41 – p.I93

IDH1 Chr2: 209113051 – 209113255 p.E84 – p.Q138

IDH2 Chr15: 90631752 – 90631976 p.F126 – p.Q178

IGF1R Chr15: 99440011 – 99440237 p.E325 – p.G367

IGF2R Chr6: 160485824 – 160485949 p.P1340 – p.F1371

Chr6: 160496879 – 160497103 p.D1723 – p.M1772

Chr6: 160517434 – 160517663 p.G2220 – p.L2280

JAK2 Chr9: 5073668 – 5073862 p.S593 – p.E621

JAK3 Chr19: 17954041 – 17954267 p.E113 – p.Q140

Chr19: 17947870 – 17948080 p.S568 – p.D595

Chr19: 17945623 – 17945811 p.L684 – p.K733

KDR Chr4: 55980271 – 55980456 p.Y221 – p.K266

Chr4: 55979507 – 55979727 p.Q268 – p.M314

Chr4: 55972826 – 55973041 p.Q472 – p.K512

Chr4: 55962395 – 55962621 p.G873 – p.G909

Chr4: 55960908 – 55961127 p.T940 – p.E990

Chr4: 55955033 – 55955251 p.Y1136 – p.Q1170

Chr4: 55953691 – 55953917 p.G1172 – p.I1220

Chr4: 55946189 – 55946355 p.G1284 – p.I1330

Chr4: 55946011 – 55946232 p.I1330 – p.V1356

KIT Chr4: 55561667 – 55561888 p.S24 – p.G93

Chr4: 55592066 – 55592292 p.S464 – p.K513

Chr4: 55593338 – 55593546 p.Q515 – p.Q549

Chr4: 55593499 – 55593700 p.K550 – p.L589

Chr4: 55594136 – 55594358 p.S628 – p.G663

Chr4: 55595409 – 55595630 p.P665 – p.L707

Chr4: 55597389 – 55597603 p.S715 – p.I744

Chr4: 55599232 – 55599442 p.C788 – p.N828

Chr4: 55602591 – 55602803 p.A829 – p.L865

KRAS Chr12: 25398183 – 25398385 p.M1 – p.E37

Chr12: 25380260 – 25380337 p.R41 – p.A66

Chr12: 25378561 – 25378743 p.E98 – p.A146

LAMTOR1 Chr11: 71810184 – 71810406 p.D15 – p.L54

Chr11: 71809744 – 71809973 p.N64 – p.Y88

LAMTOR2 Chr1: 156025008 – 156025204 p.L24 – p.M74

MAP2K1 Chr15: 66729047 – 66729265 p.L98 – p.M146

Chr15: 66774032 – 66774239 p.V191 – p.S231

Chr15: 66727373 – 66727563 p.T28 – p.A95

6

MAPKAP1 Chr9: 128246741 – 128246970 p.S357 – p.F396

Chr9: 128201149 – 128201292 p.V482 – p.Q522

MET Chr7: 116339530 – 116339753 p.I131 – p.F206

Chr7: 116340170 – 116340389 p.G344 – p.R400

Chr7: 116403120 – 116403338 p.S812 – p.K879

Chr7: 116411806 – 116412001 p.L982 – p.V1014

Chr7: 116417424 – 116417614 p.H1106 – p.N1131

Chr7: 116423291 – 116423508 p.L1230 – p.T1280

MLST8 Chr16: 2256502 – 2256711 p.Q63 – p.L114

Chr16: 2258233 – 2258463 p.L199 – p.S232

MTOR Chr1: 11190760 – 11190931 p.A1789 – p.E1813

Chr1: 11190565 – 11190770 p.Q1807 – p.E1871

Chr1: 11187658 – 11187880 p.V2012 – p.Q2072

Chr1: 11174371 – 11174529 p.V2389 – p.D2433

Chr1: 11169263 – 11169482 p.G2484 – p.T2509

NRAS Chr1: 115258627 – 115258821 p.M1 – p.E37

Chr1: 115256452 – 115256669 p.D38 – p.S87

Chr1: 115252147 – 115252254 p.T127 – p.Q150

PDGFRA Chr4: 55140927 – 55141154 p.K552 – p.L595

Chr4: 55144059 – 55144277 p.T632 – p.S667

Chr4: 55151970 – 55152162 p.C814 – p.S854

PDK1 Chr2: 173427380 – 173427601 p.R138 – p.M156

Chr2: 173450928 – 173451139 p.M336 – p.L372

Chr2: 173457684 – 173457906 p.P380 – p.K410

PHLPP1 Chr18: 60642573 – 60642794 p.K1253 – p.S1307

PHLPP2 Chr16: 71683541 – 71683769 p.A999 – p.G1075

PIK3CA All coding regions

PIK3CB Chr3: 138374192 – 138374406 p.D1026 – p.S1070

PIK3R1 Chr5: 67593218 – 67593445 p.V663 – p.R724

PIK3R2 Chr19: 18273734 – 18273817 p.K371 – p.G385

PTEN All coding regions

PTPN11 Chr12: 112888117 – 112888297 p.R47 – p.C104

Chr12: 112926780 – 112926999 p.V484 – p.Q533

RHEB Chr7: 151187964 – 151188182 p.K19 – p.N41

Chr7: 151167603 – 151167827 p.V128 – p.Q154

RICTOR Chr5: 38959878 – 38959945 p.I663 – p.Q683

Chr5: 38952408 – 38952632 p.T967 – p.D1006

Chr5: 38950550 – 38950776 p.S1058 – p.L1134

RPS6KB1 Chr17: 58011763 – 58011967 p.A261 – p.A290

Chr17: 58013815 – 58014032 p.A348 – p.L373

RPS6KB2 Chr11: 67201650 – 67201787 p.R324 – p.L349

7

RPTOR Chr17: 78882538 – 78882765 p.V802 – p.K840

Chr17: 78899135 – 78899346 p.T937 – p.K973

Chr17: 78933883 – 78934090 p.D1160 – p.E1201

RRAGA Chr9: 19049675 – 19049856 p.T4 – p.Q66

RRAGB ChrX: 55748556 – 55748778 p.V43 – p.T75

ChrX: 55757816 – 55758046 p.E132 – p.L200

ChrX: 55779786 – 55780009 p.A233 – p.L273

RRAGC Chr1: 39322645 – 39322863 p.V80 – p.D116

Chr1: 39322526 – 39322746 p.D116 – p.Q147

Chr1: 39321410 – 39321639 p.D148 – p.A204

RRAGD Chr6: 90077747 – 90077976 p.L352 – p.L400

SMAD4 Chr18: 48575081 – 48575265 p.H92 – p.D142

Chr18: 48575548 – 48575763 p.D142– p.N151

Chr18: 48581005 – 48581225 p.A152 – p.H177

Chr18: 48584490 – 48584654 p.S223 – p.H262

Chr18: 48586137 – 48586349 p.P303 – p.P318

Chr18: 48591742 – 48591964 p.A319 – p.I376

Chr18: 48593403 – 48593514 p.G384 – p.P422

Chr18: 48603025 – 48603205 p.Q442 – p.I482

Chr18: 48604603 – 48604826 p.S483 – p.P550

SOS1 Chr2: 39281647 – 39281866 p.E198 – p.N240

Chr2: 39249808 – 39250023 p.E515 – p.I587

SRC Chr20: 36031585 – 36031773 p.R472 – p.E534

STAT3 Chr17: 40474350 – 40474557 p.K631 – p.G684

8

Table S3: Restriction fragment length polymorphism assay reagents.

PCR products were designed to cover the codon of interest. One primer was labelled with 6-carboxyfluorescein (6FAM) for detection via Genemapper software on an AB13730 capillary sequencer. The other primer was designed to include a restriction enzyme recognition site (altered bases depicted in capitals), resulting in specific digestion of the mutant but not the wildtype allele.

Mosaic

variant

Forward primer

sequence (5’ to 3’) Reverse primer

sequence (5’ to 3’) Restriction

enzyme

Size of

wildtype

fragment

(bp)

Size of

mutant

fragment

following

digestion

(bp)

KRAS, c.35G>A, p.Gly12Asp

aaacttgtggtagttggagcGg

[6FAM]aagaatggtcctgcaccagta

FokI 145 111

NRAS c.182A>G, p.Gln61Arg

gacatactggatacagcCgTac

[6FAM]tggggaaatgaggttaccaca

BsiWI 206 188

KRAS c.35G>T, Gly12Val

[6FAM]aaaaggtactggtggagtatttga

tcaaggcactcttgcctacgTTa

HpaI 158 136

9

Table S4: Antibodies used for immunoblotting

Antibody name Dilution Catalogue number

MAP2K1 (MEK1) Antibody 1:1000 Cell Signalling Technology #9124 p44/p42 MAPK (Erk1/2)(137F5) Rabbit mAb

1:500 Cell Signalling Technology #4695

Phospho-p44/42 MAPK (Erk1/2)(Thr202/Tyr204)(197G2) Rabbit mAb

1:1000 Cell Signalling Technology #4377

GAPDH (14C10) Rabbit mAb 1:3000 Cell Signalling Technology #2118 Anti-Rabbit IgG (γ-chain specific)–Peroxidase antibody, Mouse monoclonal RG-96 (secondary antibody used in all immunoblotting experiments)

1:7000 Sigma A1949

10

Table S5: Results of genotyping for PIK3CA and KRAS in multiple tissue samples in patient

7. MAF = minor allele frequency; FFPE = formalin-fixed, paraffin-embedded; RFLP = fluorescent Restriction Fragment Length Polymorphism assay; NGS = next generation sequencing

Sample Genotype MAF Technique Comment

Intradural, extra-medullary spinal plexiform tumour (FFPE)

PIK3CA H1047L KRAS G12D

7% 0% 0%

RFLP NGS NGS

Poor quality FFPE DNA. PIK3CA variant only present in 1 FFPE block out of 4. Low read depth in NGS (mean read depth < 50)

Neurofibroma (FFPE)

KRAS G12D PIK3CA H1047L

30% 0%

Sanger Sanger, RFLP

Lipoma (FFPE) PIK3CA H1047L 3% 0%

RFLP Sanger

Poor quality FFPE DNA.

Epidermal nevus fresh tissue

KRAS G12D PIK3CA H1047L

30% 0%

NGS, Sanger NGS, RFLP

Fibroblasts grown from epidermal nevus

KRAS G12D PIK3CA H1047L

39% 0%

NGS, Sanger NGS, RFLP

Blood PIK3CA H1047L KRAS WT

0% 0%

RFLP, Sanger Sanger

References

1. Anastasaki C, Estep AL, Marais R, Rauen KA, and Patton EE. Kinase-activating and

kinase-impaired cardio-facio-cutaneous syndrome alleles have activity during zebrafish development and are sensitive to small molecule inhibitors. Human molecular genetics. 2009;18(14):2543-54.

2. Anastasaki C, Rauen KA, and Patton EE. Continual low-level MEK inhibition ameliorates cardio-facio-cutaneous phenotypes in zebrafish. Dis Model Mech. 2012;5(4):546-52.