STUDIES OF GENETIC MOSAICISM IN RARE DISEASES

(1)

Department of Molecular Medicine and Surgery Karolinska Institutet, Stockholm, Sweden

STUDIES OF

GENETIC MOSAICISM IN RARE DISEASES

Sofia Frisk

Stockholm 2022

(2)

Cover art and all illustrations produced by Sofia Frisk.

All previously published papers were reproduced with permission from the publisher.

Published by Karolinska Institutet.

Printed by Universitetsservice US-AB, 2022

ISBN 978-91-8016-524-2

(3)

Studies of Genetic Mosaicism in Rare Diseases THESIS FOR DOCTORAL DEDGREE (PH.D.)

By

Sofia Frisk

The thesis will be defended in public at Eva & Georg Klein, BioMedicum, Karolinska Institutet, Solnavägen 9 on Friday April 29, at 9 a.m.

Principal Supervisor:

Professor Ann Nordgren Karolinska Institutet

Department of Molecular Medicine and Surgery

Co-supervisors:

Professor Anna Lindstrand Karolinska Institutet

Fulya Taylan, PhD Karolinska Institutet

Tobias Laurell, MD, PhD Karolinska Institutet

Bianca Tesi, MD, PhD Karolinska Institutet

Opponent:

James T Bennett, MD, PhD University of Washington Department of Pediatrics

Examination Board:

Associate Professor Cecilia Gunnarsson Linköping University

Department of Clinical Genetics, and Department of Biomedical and Clinical Sciences

Associate Professor Ingrid Øra Lund University

Department of Clinical Sciences

Professor Filip Farnebo Karolinska Institutet

(4)

(5)

To my parents

(6)

(7)

PREFACE

I was about 18 years old when genetics came into my mind. I attended a summer research school and did a laboratory project about the expression of SBE1, SBE2A and SBE2B; genes encoding starch branching enzymes in barley. Perhaps not the coolest thing for a teenager to do during the summer vacation, but I was amazed by the world of genetics. Several years later, here I am writing my thesis, this time about human genetics. I maintain the same fasciantion for this field as I had back then. As trained physician, I now also find strength and motivation to learn more and, at times, struggle as I try to understand mechanisms of rare diseases. Understanding underlying mechanisms in rare diseases and deliver

explanations to patients, and in the long run even treatment, are some of reasons that motivate me to continue learning.

This thesis unfolds as follows: In chapter 1, I start by introducing some key terms and background knowledge of human genetics in general. Then, I introduce genetic mosaicism and the disease groups that have been object of my studies. I continue by stating the aims and methodology of our studies in Chapter 2 and 3 respectively, followed by results and discussion in Chapter 4. Chapter 5 and 6 describe conclusions and some future perspectives respectively. Chapter 7 acknowledges those who have contributed to my development as a researcher and Chapter 8 lists the references of this thesis. In the printed version, the publications are also attached. I hope you will enjoy reading this thesis and will share some of my fascination for this fantastic field.

Sofia Frisk, Stockholm, March 16, 2022

(8)

(9)

POPULÄRVETENSKAPLIG SAMMANFATTNING

Likt mosaikplattor på ett badrumsgolv kan kroppens celler se olika ut och bära på olika egenskaper. Det här kan bero på att cellerna har olika genetiska uppsättningar och kallas mosaicism. Mosaicism kan ge upphov till mycket sällsynta sjukdomar som ofta inte har standardiserade vårdförlopp inom sjukvården. Mosaicism riskerar att missas som förklaring till sjukdom eftersom det är så ovanligt eller okänt och då inte tillräckligt känsliga metoder används för detektion. I den här avhandlingen har vi med hjälp av finkänsliga metoder kunnat upptäcka mosaicism som förklaring till sjukdom.

I delstudie I och II beskriver vi två ovanliga symptombilder vid överväxtsyndrom. I delstudie I beskriver vi PIK3CA-orsakad överväxt som omfattar muskelöverväxt. Märkligt nog har dessa patienter muskler på ställen där det vanligtvis finns bindväv, till exempel i handflatan. Hos en av patienterna finns muskelöverväxten på båda sidor vilket tyder på att mutationen hänt mycket tidigt under den embryologiska utvecklingen. I delstudie II beskrivs ytterligare en ovanlig kombination av symptom, denna gång orsakad av en DICER1-mutation. I delstudie III beskriver vi en tidigare okänd mutation hos en pojke med fokal dermal hypoplasi och rapporterar en sammanställning över alla beskrivna manliga patienter med denna diagnos.

I delstudie IV har vi flyttat fokus till föräldrarna till barn med genetiska syndrom. Enligt dagens rutin lämnar föräldrarna blodprov och hittar man inte samma genetiska avvikelse i deras blod som påvisats hos deras sjuka barn får de därför informationen att de inte är bärare och att risken för att de ska få fler barn med samma genetiska syndrom därför är mycket låg. Men tänk om förändringen finns där i så liten grad att vi inte kan upptäcka den med vanliga metoder? Eller att förändringen bara finns i könscellerna och därför inte upptäcks med blodprov? Vi har letat efter mosaicism i blod och sperma och hittade det hos två pappor där vi också såg högre nivåer i sperma än i blod.

Att förstå mekanismerna bakom en sällsynt sjukdom möjliggör bättre uppföljning med kontroller och eventuell behandling. Inom till exempel överväxtsyndrom har

behandlingsalternativ tagits fram efter att man förstått att PIK3CA-mosaicism är orsaken.

Det är också viktigt för att kunna ge så korrekt information som möjligt till patienter och föräldrar. Fynd av mosaicism kan till exempel precisera varje familjs specifika

upprepningsrisk att få ett ytterligare sjukt barn.

(10)

ABSTRACT

Mosaicism in human genetics refers to an individual harboring two or more genetic compositions, all derived from the same fertilized egg. Common signs of genetic mosaicism are asymmetric growth, skin aberrations or vascular malformations. Each clinical picture is in itself rare, but together mosaic disorders form a growing group of identifiable characteristic abnormalities. Interestingly, several pharmacological treatment possibilities for these conditions have evolved in the last couple of years.

In study I, we found mosaic hotspot PIK3CA variants in two patients with ectopic muscles and muscular overgrowth, by performing whole genome sequencing and digital PCR. This adds information about timing of PIK3CA mutagenesis during embryogenesis in correlation to phenotype and confirms the diagnostic entity PIK3CA-related muscular overgrowth with ectopic muscles.

In study II, we describe a genetic mechanism in DICER1-related overgrowth. We show that a constitutional DICER1 variant encoding the RNase IIIa domain causes a severe subtype of DICER1 syndrome with intellectual disability, macrocephaly, extensive bilateral lung cysts, early onset of Wilms tumor, and well-differentiated fetal lung adenocarcinoma.

This phenotype is similar to, but distinct from, the phenotype reported in two patients with GLOW syndrome caused by mosaic hotspot variants encoding the RNase IIIb domain.

In study III, we add knowledge of genotype-phenotype correlations in male focal dermal hypoplasia patients by describing a previously unknown disease-causing variant in a male patient, and by highlighting that focal dermal hypoplasia can be suspected in patients with characteristic limb malformations, such as ectrodactyly, or ocular manifestations, even in the absence of typical skin findings.

In study IV, we used droplet digital PCR to analyze blood- and sperm-derived DNA from 87 parents to children with intellectual disability syndromes caused by de novo variants.

We found germline mosaicism in two fathers and showed that analysis of blood alone may underestimate germline mosaicism.

Taken together, these studies have improved our understanding of methodological approaches in mosaicism diagnostics. In addition, these studies contribute to our

understanding of the phenotypic and/or genetic spectrum of PIK3CA-related overgrowth, DICER1-related overgrowth, focal dermal hypoplasia and germline mosaicism in rare diseases.

(11)

LIST OF SCIENTIFIC PAPERS

I. Frisk S, Taylan F, Blaszczyk I, Nennesmo I, Annerén G, Herm B, Stattin EL, Zachariadis V, Lindstrand A, Tesi B, Laurell T, Nordgren A. Early activating somatic PIK3CA mutations promote ectopic muscle development and upper limb overgrowth. Clinical Genetics. 2019 Aug;96(2):118-125. doi:

10.1111/cge.13543. PMID: 30919936.

II. Pontén E, Frisk S, Taylan F, Vaz R, Wessman S, de Kock L, Pal N, Foulkes WD, Lagerstedt-Robinson K, Nordgren A. A complex DICER1 syndrome phenotype associated with a germline pathogenic variant affecting the RNase IIIa domain of DICER1. Journal of Medical Genetics. 2020 Nov 18:2020-107385. doi: 10.1136/jmedgenet-2020-107385. PMID: 33208384.

III. Frisk S, Grandpeix-Guyodo C, Popovic Silwerfeldt K, Hjartarson HT,

Chatzianastassiou D, Magnusson I, Laurell T, Nordgren A. Goltz syndrome in males: A clinical report of a male patient carrying a novel PORCN variant and a review of the literature. Clinical Case Reports. 2018 Sep 21;6(11):2103- 2110. doi: 10.1002/ccr3.1783. PMID: 30455901.

IV. FriskS, WachtmeisterA, Laurell T, Lindstrand A, Jäntti N, Malmgren H, Lagerstedt-Robinson K, Tesi B, Fulya Taylan F, Nordgren A. ^.Detection of germline mosaicism in fathers of children with intellectual disability syndromes caused by de novo variants. Molecular Genetics & Genomic Medicine. 2022 Feb 4:e1880. doi: 10.1002/mgg3.1880. Epub ahead of print.

PMID: 35118825.

(12)

(13)

LIST OF ABBREVIATIONS

bp: Base pair

BWS: Beckwith-Wiedemann syndrome CGH: Comparative genomic hybridization CNV: Copy number variant

DCMO: Diffuse capillary malformation with overgrowth ddPCR: Droplet digital polymerase chain reaction

DMEG: Dysplastic megalencephaly DNA: Deoxyribonucleic acid

dPCR: Digital polymerase chain reaction ERN: European reference network FAVA: Fibroadipose vascular anomaly FDH: Focal dermal hypoplasia

FFPE: Formalin fixed paraffin embedded FISH: Fluorescence in situ hybridization GOF: Gain-of-function

ID: Intellectual disability

KOGS: Kosaki overgrowth syndrome KTS: Klippel-Trenaunay syndrome LM: Lymphatic malformation LOD: Limit-of-detection LOF: Loss-of-function LOH: Loss-of-heterozygosity

(16)

MCAP: Megalencephaly capillary malformation M-CM: Macrocephaly capillary malformation

MLPA: Multiplex ligation-dependent probe amplification MPS: Massively parallel sequencing

NGS: Next generation sequencing

PIK3CA: Phosphatidylinositol-4,5-bisphospate 3-kinase, catalytic subunit alpha

PIP3: Phosphatidylinositol 3,4,5-trisphosphate PLWRD: Persons living with a rare disease PROS: PIK3CA-related overgrowth spectrum RNA: Ribonucleic acid

SGBS: Simpson-Golabi-Behmel syndrome SNP: Single nucleotide polymorphism SNV: Single nucleotide variant

SV: Structural variant

VAF: Variant allelic fraction WES: Whole exome sequencing WGS: Whole genome sequencing

(17)

1 INTRODUCTION

1.1 RARE DISEASES

1.1.1 Definition of rare disease

In this thesis, the term rare disease is used according to the European definition. In Europe, a disease is defined as rare when it affects fewer than 1 in 2,000 people (Czech et al., 2019).

In the United States, the definition is slightly different, saying a disease is defined as rare when it affects fewer than 200,000 people in the US at any given time. Considering a population of 330 million people, this definition can be translated to a prevalence of approximately 1 in 1,650 people (Whicher, Philbin, & Aronson, 2018).

1.1.2 Rare diseases in the world

Despite the rarity of each disease, when considered as a group, rare diseases are common.

Between 6,000 and 8,000 rare diseases have been identified and they collectively affect approximately 6-8% of the global population, half of them being children (Auvin, Irwin, Abi-Aad, & Battersby, 2018; Czech et al., 2019; Richter et al., 2015). In 2017, European Reference Networks (ERN) were launched. They consist of European expert teams on complex and rare diseases that require highly specialized health care. As of January 2022, the total number of ERN members were almost 1,500 healthcare units. In addition, the United Nations’ General Assembly recently adopted the first-ever resolution recognizing the over 300 million persons living with a rare disease (PWLRD) worldwide and their families (United Nations, 2021). The resolution focuses on the importance of non-

discrimination and emphasizes key pillars of the Sustainable Development Goals, including inclusion and meaningful participation in society, tackling gender inequality, and ensuring access to quality health services without financial hardship. All of the 193 signing member states have agreed to strengthen their health systems and empower PLWRD. Some

measures are to design and implement national policies and strategies, to strengthen international collaborations and coordination of research efforts, and to share data (Eurordis, 2021).

Rare diseases are often life-long and progressive. The aetiology behind rare diseases varies, the cause can for example be infections, auto-immunity or environmental factors. For a proportion of these conditions, the aetiology is still unknown. However, the great majority,

(18)

80% of rare diseases, are of genetic origin. (Auvin et al., 2018; Czech et al., 2019; Liu et al., 2019)

1.2 HUMAN GENETICS

1.2.1 The human genome

The human genome consists of approximately 20,000 genes packed into 46 chromosomes in a normal body cell. All the chromosomes are in pairs (23 pairs in a normal body cell).

The germ cells are different. They are haploid cells, meaning they contain just a single set of chromosomes (n = 23). The genome is made of deoxyribonucleic acid (DNA). Basically, DNA is a coding sequence combined of four different nucleotides, also called base pairs (bp): Adenine (A), Guanine (G), Thymine (T) and Cytosine (C). Genes are composed of introns and exons, where only the exons are used as templates to make complementary ribonucleic acid (RNA) sequences (transcription), which are later translated to polypeptides (proteins) via messenger RNA (translation) or becomes mature non-coding RNA. Three nucleotides form a codon encoding one amino acid. These are formed to polypeptides and proteins. Introns are non-coding regions accounting the majority of our genome. Only 1.1%

of the 3.2 billion nucleotides in our genome are protein-coding regions.

1.2.2 Genetic variation

The human genome is polymorphic, meaning that there are differences in the DNA

sequence at specific positions between individuals. The variation can be divided into small or large-scale variants.

1.2.2.1 Small-scale variants

A genetic small-scale variant is a variant affecting one or a small number of base pairs (<50 bp). Some small-scale variants relevant for this thesis are:

1. Synonymous variants: Not altering the encoded amino acid.

2. Missense variants: Substitution of a specific amino acid into another.

3. Nonsense (also known as stop-gain) variants: Introduction of an immediate stop codon. This can be caused by deletions, duplications, insertions or substitions of base pairs.

(19)

4. Splice site variants: Leading to abnormal splicing of the exons.

5. Indels: Duplications, deletions, insertions or a combination of them that lead to a size change at a specific site: In-frame indels do not affect the reading frame while out-of-frame indels, also called frameshift variants, changes the reading frame and may introduce a premature stop codon.

1.2.2.2 Large-scale variants

Copy number variants (CNVs), refer to deletions or duplications > 50 bp that change the total gene dosage. They can also be referred to as unbalanced structural variants (SVs). A balanced SV is due to rearrangements of the chromosomal material, for example a

translocation or an inversion, but does not lead to any changes in the total gene dosage.

1.2.2.3 Interpretation of genetic variants

Since the human genome is polymorphic, there are a lot of variants not leading to disease.

Making the distinction between a normal variant and a disease-causing variant can be very challenging. To predict the functional effect, there are several bioinformatic tools. In addition, there is a system to classify variants. Variants in this thesis were classified into benign (1), likely benign (2), variants of unknown significance (3), likely pathogenic (4) and pathogenic variants (5) according to a well-established classification system (Richards et al., 2015). Likely pathogenic and pathogenic variants are considered as potential disease- causing variants and are traditionally called mutations. In this thesis, disease-causing variant is preferably used. However, if the word mutation is used, it refers to disease- causing variant (pathogenic or likely pathogenic variant).

It is also important to have reduced penetrance in mind when interpreting genetic findings.

Penetrance refers to the proportion of people with a particular genetic variant who exhibit signs and symptoms of the disorder. Reduced penetrance means there is a chance the genetic variant does not lead to any symptoms, even when classified as likely pathogenic or pathogenic. Variable expressivity may also complicate the interpretation, meaning that there is a spectrum of signs and symptoms that can occur in different people with the same genetic condition. The same genetic variant can lead to mild or very severe symptoms. The reason for reduced penetrance and variable expressivity is unknown but may be due to epigenetic modifications, the impact of additional genes and biological factors.

(20)

1.2.3 De novo variants in rare diseases

A fertilized egg becomes a zygote and inherits half of its genome from the mother through the oocyte, and the other half from its father via the sperm. In addition, each one of us is born with novel, de novo, genetic changes from the inherited information. Compared to the human reference genome, the human genome varies, on average, at 4-5 million sites (Pasmant & Pacot, 2020). Studies have estimated the de novo mutation rate of single nucleotide variants (SNVs) to be approximately 10^-8 mutations per generation (Kong et al., 2012; Michaelson et al., 2012; Pasmant & Pacot, 2020; Rahbari et al., 2016; Roach et al., 2010). Thus, a newborn child is estimated to harbor 40-80 de novo SNVs, with the potential to cause disease (Breuss et al., 2020; Jónsson et al., 2018; Kong et al., 2012; Pasmant &

Pacot, 2020). Such variants are either pre-zygotic, originating from parental germ cells, or post-zygotic, arising from an early mitotic division in the developing embryo (fig. 1). De novo variants arise 3-4 times more often in paternal germ cells compared to maternal germ cells, and the amount of paternal de novo variants has a positive correlation to increasing age of the father (Kong et al., 2012; Michaelson et al., 2012; Rahbari et al., 2016). Lately, the increased usage of trio genome and exome sequencing has revealed that de novo variants are a significant cause behind genetic disease, underlying about 30% of all diagnosed cases (Acuna-Hidalgo, Veltman, & Hoischen, 2016; Rauch et al., 2012;

Stefanski et al., 2021; Stranneheim et al., 2021).

1.3 EARLY HUMAN DEVELOPMENT

After the fertilization, a series of cell divisions takes place to form an embryo. Between 15- 21 days postfertilization, three different germ layers are formed; the endo-, meso-, and ectodermal layer (Madsen, Vanhaesebroeck, & Semple, 2018). This process is called gastrulation. These three layers give rise to different parts of the body. The outer layer of the skin (epidermis), and the nervous system (neuroectoderm) origin from the ectoderm.

The endoderm gives rise to the gastrointestinal and respiratory systems, the epithelial lining of the bladder and urethra, as well as many endocrine glands. The mesoderm gives rise to a variety of tissues in between these two layers, including bone, cartilage, connective tissue, fat, muscles, the vascular system, gut and lungs.

The cell divisions continue, and step by step creates the embryo that the human originates from. During each cell division, the human genome needs to replicate along the way. To

(21)

make this happen, the cell has to divide its genetic material into two. This is done by a series of events. Microtubules attach to a site called kinetochore at the center of the

chromosome, the centrosome. The chromosome starts to divide when microtubules begin to depolymerize, pulling the chromatids to two opposite poles. Cytokinesis begins and the cell divides, creating two identical cells. To generate an adult human, as many as 10¹⁴ cells are required (Frank, 2014). If an error occurs during any cell division, a new cell line might be created, different from the previous. Depending on the timing and what kind of protein change the error gives rise to, and in what kind of tissue, the error can cause a change in the function of that body part. It may look or behave differently. This is an example of genetic mosaicism (fig. 1).

Figure 1. Development of somatic mosaicism. Dark purple is illustrating affected (mutated) cells. The sperm fertilizes the egg and becomes a zygote. Cell divisions occur. A post-zygotic mutational event happens during embryonic development, leading to somatic mosaicism.

1.4 GENETIC MOSAICISM

1.4.1 Definition of genetic mosaicism

Mosaicism in human genetics refers to an individual harboring two or more genetic compositions, which are all derived from the same fertilized egg. There is a genetic variation of mosaicism involving entire chromosomes, small-scale variants, large-scale variants and epigenetic variants (Biesecker & Spinner, 2013; Martínez-Glez et al., 2020).

The mutational event may occur at any time after fertilization, in any type of cell, making a huge variation in phenotypic effect.

(22)

1.4.2 Genetic mosaicism in rare diseases

In theory, any monogenic disease may appear in mosaic form. In that case, it is usually associated with a milder or atypical course compared to a mutational event affecting all cells. Another scenario is if a genetic variant, that would be lethal if it occurred in every cell, occurs in mosaic form. Then, it may be tolerated and compatible with life. There are many potential mechanisms of a mosaic disorder, such as DNA replication errors, break- induced repair, mitotic recombination, anaphase lagging, endoreplication, non-disjunction, and environmental factors such as ultraviolet radiation, environmental genotoxic agents, and viral integration. A majority of mosaic events do not lead to any functional

consequence and therefore do not cause disease (Moog, Felbor, Has, & Zirn, 2020). If a mutational event occurs at an important site during early human development, and confer a growth or selective advantage, a mutant clone will expand and reach clinical relevance (Breuss, Yang, & Gleeson, 2021).

Common signs of genetic mosaicism are asymmetric growth, focal brain malformations, different skin patterns or vascular malformations. Each disorder is in itself rare, but together these diseases form a growing group of identifiable clinical pictures displaying

characteristic abnormalities. Interestingly, several pharmacological treatment possibilities have evolved in the last couple of years. In chapter 1.5.2.3, some pharmacological

treatment possibilities in overgrowth syndromes are described.

1.4.3 Different types of mosaicism

There are several ways to classify mosaicism (Martínez-Glez et al., 2020). Depending on the distribution of affected cell populations, mosaicism may be divided into general,

somatic or confined. General mosaicism is present throughout the entire organism, whereas confined mosaicism refers to genetic mosaicism present in only a particular area, for

example the brain, placenta or gonads (Taylor et al., 2014). Germline mosaicism, formerly gonadal mosaicism, is the appearance of different genetic cell lines confined to the gonads (fig. 2b) (Schwab, Tuohy, Condie, Neklason, & Burt, 2008). Somatic mosaicism, on the other hand, occurs when somatic cells, the body cells, have more than one genotype (fig.

2a) (Biesecker & Spinner, 2013). This is the most reported type of mosaicism in disease and occurs for example in skin disorders, overgrowth and vascular anomalies (Moog et al., 2020). The main mechanism in cancer is also somatic mosaicism (Martínez-Glez et al.,

(23)

2020). Gonadosomatic, or gonosomal, mosaicism can be used as a term to describe mosaicism present in both germline and somatic cells (Martínez-Glez et al., 2020).

Regardless of whether mosaicism is present in the entire body or is confined to a specific organ or body part, the underlying mutational mechanisms are similar (Taylor et al., 2014).

Figure 2. Examples of different types of mosaicism. Dark purple is illustrating affected (mutated) cells. A proportion of cells affecting the body cell is called somatic mosaicism (a). Affected cells confined to the gonads is called germline mosaicism (b). Affected cells following the lines of Blaschko (c). This figure is also illustrating lateralization, meaning only one half of the body is affected. Revertant mosaicism is when a proportion of cells has reverted to unaffected while the rest remains affected (c).

Other classification types are segmental or non-segmental mosaicism. Segmental means that the mosaicism involves several body parts, which reflects clonal expansion in the developing embryo. The mosaic body pattern is usually respecting the midline but can appear asymmetric. One example of pattern is Blaschko linear distribution (fig. 2c).

Blaschko lines are streaks representing pathways of cell migration during embryonic development. Other types of segmental mosaicism are phylloid, lateralization and checkerboard/block/flag-like mosaicism. The opposite, non-segmental mosaicism, is restricted to one location of the body. This is more common than segmental and can be patchy, disseminated or single point mosaicism. So far, the mechanisms behind the patterns are poorly understood (Martínez-Glez et al., 2020).

There are more dimensions to mosaicism, for example the direction of the mutational event.

The great majority of individuals with mosaicism have a direction of benign to pathogenic.

However, there are forms of mosaicism where the direction of change is from pathogenic to benign (fig. 2d). This means that affected individuals with generalized disease, have one or

(24)

more areas where the affected tissue has reverted to unaffected. This is called revertant mosaicism and has been described mostly in skin cells but also in liver, muscle and

hematopoietic cells (Biesecker & Spinner, 2013; Martínez-Glez et al., 2020). The reversion may occur through mitotic recombination, presumably with positive selection of cells with the unaffected allele (Biesecker & Spinner, 2013; Martínez-Glez et al., 2020).

1.5 MOSAICISM IN OVERGROWTH SYNDROMES

1.5.1 Overgrowth syndromes – an overview

Overgrowth syndromes are characterized by excessive growth and can appear anytime in life. Growth regulation is a complex process involving genetic, epigenetic, endocrine, and metabolic factors. The nutrition and/or oxygen exchange between the placenta and the fetus during pregnancy can also be of significance, as well as fetal exposures to toxins,

infections, and pollutants (Brioude et al., 2019). The overgrowth can be generalized, involving the entire body, or segmental, affecting one or several body parts (Manor &

Lalani, 2020). In cases of segmental overgrowth, mosaic mechanisms can be suspected.

Still, approximately 50% of syndromic overgrowth patients have no identified molecular findings (Brioude et al., 2019). There are many diagnoses within the overgrowth

syndromes, here are a short description of some of them:

Beckwith-Wiedemann Syndrome (BWS) is the most common overgrowth syndrome with an incidence of approximately 1:10,500. Clinical findings in BWS include lateralized overgrowth (hemihyperplasia), macrosomia, macroglossia, exomphalos, midface hypoplasia, prominent mandible, ear pits, and facial naevus flammeus. There is an increased risk of developing childhood cancer, particularly Wilms tumor and

hepatoblastoma. Genetic and epigenetic findings within the 11p15 region are found in approximately 80% of patients with a clinical diagnosis of BWS. (Duffy et al., 2019) Simpson-Golabi-Behmel syndrome (SGBS) is a rare X-linked disorder due to GPC3 loss- of-function (LOF) variants. Macrosomia, macroglossia, and visceromegaly are the most frequent findings. Cognitive impairment, supernumerary nipples, skeletal anomalies, and different congenital malformations are also common. There is an increased risk of Wilms tumor and other embryonal tumors. (Brioude et al., 2019; Cottereau et al., 2013)

Sotos syndrome is mainly caused by mutations in NSD1, which encodes a histonemethyltransferase and therefore affecting epigenetic regulation and gene

(25)

transcription. Patients with Sotos syndrome usually present with excessive birth length but normal weight. Typical findings include postnatal overgrowth, hyperglossia, intellectual disability and facial dysmorphic features such as frontal bossing, sparse frontotemporal hair, malar flushing and a prominent chin. Malformations can also be seen, for example in the heart, brain, kidney or skeletal. Behavioural problems and symptoms of autism

spectrum disorder have also been reported. The syndrome is associated with various malignancies, including lymphomas and acute lymphoblastic leukemia. Weaver and Malan syndrome are two syndromes with phenotypic similarities to Sotos. (Brioude et al., 2019)

Kosaki overgrowth syndrome (KOGS) is associated with overgrowth, scolios, large hands and large feet. The skin is usually hyperelastic, thin and fragile. White matter lesions can be seen in the brain, as well as facial features, such as frontal bossing and ptosis. Variants in PDGFRB are associated with this syndrome. Tyrosine kinase inhibitor Imatinib has been trialed in KOGS with no further disease progression or serious side effects noted after 18 months of treatment (Rustad et al., 2021) and in other PDGFRB-related disorders with positive results (Mudry et al., 2017; Pond et al., 2018; Wenger et al., 2020).

Patients carrying constitutional mutations of PTEN present with various phenotypes, which are grouped as PTEN-related syndromes. Mosaic variants in PTEN have been reported (Nathan, Keppler-Noreuil, Biesecker, Moss, & Darling, 2017). Macrocephaly,

gastrointestinal polyposis, breast cancer, papules, and penile freckling associated with vascular or lymphatic malformations are very frequently seen in these patients (Macken, Tischkowitz, & Lachlan, 2019). Intellectual disability and/or autism spectrum disorders can also be observed (Brioude et al., 2019). Cowden syndrome and Bannayan-Riley-

Ruvalcaba syndrome are included in the PTEN-related syndromes.

Patients with segmental overgrowth and lipomatous cerebriform hyperplastic tissue can be suspected of having Proteus syndrome. The overgrowth is progressive and is often not evident at birth (Martinez-Lopez et al., 2019). Other findings include cutaneous naevi, abnormal adipose tissue and vascular malformations. The phenotype has been linked to one somatic AKT1 variant (Lindhurst et al., 2011). AKT1 and PTEN are both involved in the PI3K-AKT-mTOR-signaling pathway, just like PIK3CA (fig. 3).

(26)

1.5.2 PIK3CA-related overgrowth spectrum

In 2012, it was discovered that segmental overgrowth and/or vascular malformation could be due to mosaic hotspot variants in the phosphatidylinositol-4,5-bisphosphate 3-kinase, catalytic subunit alpha gene (PIK3CA). Nevertheless, PIK3CA was known before that; as an oncogene in several common cancer forms (Samuels et al., 2004). More and more overgrowth patients were reported and the phenotype expanded to include several separate diagnoses. In 2015, the umbrella term PIK3CA-related overgrowth spectrum (PROS) was proposed (Keppler-Noreuil et al., 2015).

Figure 3. Illustration of the PI3K-AKT-mTOR signaling pathway.

1.5.2.1 Molecular mechanisms in PROS

PROS is due to gain-of-function (GOF) variants in PIK3CA. This gene encodes the p110 α catalytic subunit of phosphoinositide 3-kinase (PI3K), that phosphorylates

phosphatidylinositol to generate phosphatidylinositol 3,4,5-trisphosphate (PIP3). PIP3 plays a key role in activating signaling cascades including inhibition of apoptosis, activation of protein synthesis, and enhanced cell survival (fig. 3) (Tatton-Brown & Weksberg, 2013).

The PI3K-AKT-mTOR signaling pathway is central during embryonic development and seems to play an important role regarding cell growth, differentiation and pluripotency (Van Keymeulen et al., 2015; Ying, Sandoval, & Beronja, 2018). Mosaic events in the PI3K-AKT-mTOR pathway have been frequently observed. The reason for enrichment of mosaicism in this pathway is unclear (Cao et al., 2019). The variable expression of

symptoms within PROS is mainly explained by the timing and location of the initiating PIK3CA mutation, but the reason behind the high degree of interindividual phenotypic heterogeneity is unknown (Madsen et al., 2018). Three sites in the PIK3CA gene account for 80% of the found mutations: glutamates (E) 542 and 545, and histidine (H) 1047. Other PIK3CA sites identified in PROS patients seem to correlate with a less severe phenotype

(27)

(Madsen et al., 2018). Approximately 10% of molecular confirmed cases are due to constitutional PIK3CA variants (non-mosaicism) (Moog et al., 2020). In such cases, the growth disorder is not asymmetric (Moog et al., 2020).

1.5.2.2 Phenotypic spectrum in PROS

CLOVES is the acronym for a combination of symptoms; Congenital Lipomatous

Overgrowth, Vascular malformations, Epidermal nevi and Skeletal anomalies/Scoliosis. A characteristic feature of CLOVES syndrome is the lipomatous overgrowth involving the trunk, sometimes infiltrated by vascular malformations. Epidermal nevi can be seen,

usually following the Blaschko lines. Skeletal anomalies, often scoliosis, can be progressive and deforming. Skeletal anomalies usually manifest distally and become more proximal with time. Malformations of the hands are common. Characteristic for CLOVES are wide hands with ulnar deviation of the fingers and furrowing of the palms and soles, due to fatty deposition. (Kurek et al., 2012; Martinez-Lopez et al., 2017)

Klippel-Trenaunay Syndrome (KTS) has similarities to CLOVES, particularly the combination of vascular malformation and overgrowth. Typically, the findings in KTS are isolated to a lower extremity. The capillary malformation is often a complex combined venous-lymphatic anomaly extending to muscular fascia, located to the lateral side of the affected extremity. (Martinez-Lopez et al., 2019)

Brain overgrowth, or megalencephaly, with associated neurologic manifestations has been seen in PROS patients (Jansen et al., 2015). Dysplastic megalencephaly (DMEG) is

characterized by bilateral brain overgrowth together with intractable epilepsy, global developmental delay, and complex neurologic deficits. Megalencephaly capillary malformation (MCAP), also named macrocephaly capillary malformation (M-CM), is characterized by congenital or early postnatal segmental overgrowth, megalencephaly, and reticulated or confluent capillary malformations. In addition, neurologic manifestations have been reported, including ventriculomegaly and cerebellar tonsillar ectopia (Mirzaa, Rivière, & Dobyns, 2013).

Diffuse capillary malformation with overgrowth (DCMO) is characterized by overgrowth and capillary malformations, not necessarily on the same location. The overgrowth is usually located to one extremity and may involve soft tissue or bone.

(Hughes, Hao, & Luu, 2020).

(28)

The clinical spectrum of PIK3CA-related disorders expands to not only include overgrowth.

Vascular manifestations may also be due to mosaic variants in PIK3CA (Luks, Kamitaki, Vivero, Uller, Rab, Bovée, et al., 2015). Fibroadipose vascular anomaly (FAVA) and some common (cystic) lympatic malformations (LM) are classified as PROS when they are associated with overgrowth (Canaud, Hammill, Adams, Vikkula, & Keppler-Noreuil, 2021). Fibroadipose hyperplasia, fibroadipose or facial infiltrating lipomatosis, lipomatosis of the nerve, macrodactyly and muscular hemihyperplasia are also classfied within the PROS group (Canaud et al., 2021).

1.5.2.3 Treatment possibilities in the PI3K-AKT-mTOR pathway

Traditionally, the treatment of overgrowth syndromes has been based on supportive

surgical debulking and blocking of overgrowth vessels. In muscular overgrowth, correction of angular bone deformities of hands and removing tight ectopic tissues may improve function. However, these approaches are difficult to perform and have limited effect or even lead to recrudescence of overgrowth (Pagliazzi et al., 2021). Regarding

pharmacological treatment, there are now several studies targeting the PI3K-AKT-mTOR signaling pathway and several drugs are developing. Proteus syndrome has been treated with AKT1-inhibitor (Agarwal et al., 2015). Sirolimus has been reported to stabilize disease severity and diminish symptoms in PROS through direct inhibition of mTORC1, which is found further downstream in the PI3K-AKT-mTOR signaling pathway. However, a majority of patients reported adverse effects and 18% reported withdrawing from the treatment due to adverse effects in a study (Parker et al., 2019). Low-dose Sirolimus

treatment affects progressive overgrowth of adipose tissue rather than significant regression of existing overgrowth (Pagliazzi et al., 2021). A switch from Sirolimus to Alpelisib has been reported as beneficial regarding decrease in overgrowth and tolerability (Pagliazzi et al., 2021). Alpelisib, or BYL719, is a p110 alpha-specific PI3K inhibitor approved for breast cancer (Venot et al., 2018). Since GOF PIK3CA variants are over-activating the p110-alpha subunit of PI3K, a specific inhibition of this site seems promising for efficient outcome and reduction of off-target effects. A clinical trial based on 19 PROS patients reported a significant reduction of overgrowth on radiological follow-up at 3 and 6 months and improved life quality with Alpelisib treatment (Venot et al., 2018). Reported adverse effects were oral ulcerations and mild hyperglycemia. Recently, two infants and two children with PROS were treated with Alpelisib. Clinical improvements and good

tolerability were reported at a 12-months follow-up (Kolitz, Fernandes, Agim, & Ludwigl, 2022; Morin et al., 2022).

(29)

1.5.3 DICER1-related overgrowth

1.5.3.1 Molecular mechanisms in DICER1-related disease

DICER1 encodes for an RNase endonuclease important for proper messenger RNA expression in early embryonic development. DICER1 variants are identified in several neoplastic transformations and more than 40 heterozygous constitutional variants are reported to cause cancers in early life (de Kock, Wu, & Foulkes, 2019). The most

frequently observed mechanism underlying tumor formation in DICER1-related disease is germline LOF variants, combined with somatic variants in the sequence encoding the hotspot residues of the RNase IIIb domain (de Kock et al., 2019; Foulkes, Priest, &

Duchaine, 2014). Coupling analyses have showed that another site, RNase IIIa-S1344, located in proximity to the active cleft of RNase IIIb, impacts tumorigenesis through functional interaction with the RNase IIIb activity, resulting in a similar phenotype (Vedanayagam et al., 2019).

1.5.3.2 Phenotypic spectrum in DICER1-related disease

DICER1 syndrome is an autosomal dominant pleiotropic tumor predisposition syndrome with variable expression and reduced penetrance of benign and malignant tumors,

commonly pleuropulmonary blastoma, cystic nephroma, Sertoli-Leydig cell tumor and hyperplastic proliferations such as multinodular goiter (Foulkes et al., 2014). In addition, Wilms tumor has been described. Extending the phenotypic spectrum associated

with DICER1 variants, a novel syndrome has been proposed, related to mosaic variants in hotspot residues of the RNase IIIb domain. GLOW syndrome presents with a constellation of clinical features beyond cancer predisposition in affected patients, and the acronym compiles the core findings; Global developmental delay, Lung cysts, Overgrowth and Wilms tumor. A highly penetrant and severe phenotype has been shown in GLOW patients with mosaic (de Kock et al., 2016; S. Klein et al., 2014) or germline (de Kock et al., 2014) RNase IIIb missense variants, compared to patients with classic DICER1 syndrome caused by germline LOF variants.

(30)

1.6 MOSAICISM IN SKIN DISORDERS

Somatic mosaicism has been described in several skin disorders, for example McCune Albright syndrome, incontinentia pigmenti, vascular malformation syndromes, CHILD syndrome, neurofibromatosis and trisomy 13. Several patterns have been described in mosaic skin disorders, for example Blaschko linear distribution, checkerboard and phylloid pattern. Focal dermal hypoplasia (FDH) is a rare skin disorder due to somatic segmental mosaicism with a Blaschko linear distribution.

1.6.1 Focal dermal hypoplasia

1.6.1.1 Molecular mechanisms in focal dermal hypoplasia

Focal dermal hypoplasia, also known as Goltz syndrome, is a rare X-linked syndrome with variable meso-ectodermal abnormalities (Peters, Perrier, & Haber, 2014). The molecular basis of FDH was first reported in 2007, when it was discovered that FDH is caused by LOF variants in the PORCN gene, located on the X chromosome (Xp11.23). The PORCN gene encodes an endoplasmic reticulum protein, the porcupine protein, thought to be essential for modification and excretion of Wnt proteins (Alkindi, Battin, Aftimos, &

Purvis, 2013; Takada et al., 2006). Wnt proteins are important for interactions between ectoderm and mesoderm during embryogenesis (Durack et al., 2017; X. Wang et al., 2007).

Currently, PORCN is the only gene associated with FDH. A total of 219 predicted pathogenic variants are reported in the online PORCN mutation database and these are scattered throughout the entire coding sequence of the gene except exon 7. Nonsense variants are the most common variant type (n = 101), followed by missense (n = 95).

Frameshift (n = 38) and splicing variants (n = 28) are also reported. Large deletions have been reported in twelve patients. (Arlt et al., 2022)

1.6.1.2 Phenotypic spectrum in focal dermal hypoplasia

There are at least 400 individuals diagnosed with FDH (Bree, Grange, Hicks, & Goltz, 2016; L. Wang et al., 2014). FDH was long believed to be incompatible with life in males (Vreeburg, van Geel, van den Heuij, Steijlen, & van Steensel, 2011; Young, Sawyer, &

Hartnett, 2014). Today, reports show that approximately 10% of all FDH patients are males (Bornholdt et al., 2009; Bostwick, Fang, Patel, & Sutton, 2016). The majority of surviving males can be explained by mosaic PORCN variants that arise post-zygotically during

(31)

embryogenesis (Yoshihashi, Ohki, Torii, Ishiko, & Kosaki, 2011) or by Klinefelter syndrome, since these males carry an additional X chromosome (Alkindi et al., 2013).

There is no genotype-phenotype correlation, which can be explained by random X-

inactivation in females or mosaicism (Clements, Mellerio, Holden, McCauley, & McGrath, 2009; Maas et al., 2009; Rao, Shenoy, Salian, & Girisha, 2016). Common clinical findings are skin findings, syndactyly, ocular defects, and ectrodactyly. A study from 2016,

including 18 patients with FDH, demonstrated a higher incidence of ophthalmologic manifestations in patients with FDH than previously reported (77% vs. 40%) (Gisseman &

Herce, 2016). This has also been shown in later reports (Mary et al., 2017). Interestingly, four male patients with non-mosaic PORCN mutations lacked skin manifestations but showed other severe symptoms compatible with FDH (Brady et al., 2015; Madan et al., 2017). In addition, diaphragm abnormalities were detected in two of these patients, resembling previously reported patients with pentalogy of Cantrell (Maas et al., 2009;

Smigiel et al., 2011).

1.7 PARENTAL MOSAICISM IN RARE DISEASES

Parental mosaicism means that a parent of a child with a genetic rare disease, carries a proportion of cells with the same disease-causing variant as found in the child. It can be confined to the gonads of the parent, by definition germline mosaicism, or discovered in other tissues as well, for example blood or skin, making it gonadosomatic. In case of transmission, the offspring will harbor the variant in all of his/her cells (fig. 4).

During genetic counselling of parents to children with disease-causing de novo variants, clinical geneticists rely on empirical data concerning the recurrence risk in future pregnancies. Currently, a recurrence risk of 1% is commonly used during counselling (Campbell, Stewart, et al., 2014; de Lange et al., 2019; Myers et al., 2018; Röthlisberger &

Kotzot, 2007). However, the recurrence risk may be higher or lower depending on if one of the parents is a germline mosaic or not. Parental genetic testing is most often based on blood-derived parental DNA. Unless multiple children are born with the same de novo variant, germline mosaicism is generally not investigated, leaving the recurrence risk unknown for most families.

Several studies have reported on mosaicism in parents of children diagnosed with de novo variants. For example, one study analyzed parental blood with long range-PCR from parents to 100 children with CNVs (Campbell, Yuan, et al., 2014). In four parents, low-

(32)

level mosaicism was detected and probabilistic modelling was done. The conclusion was that parental mosaicism is underrecognized and influence recurrence risk. Parental

mosaicism has also been observed in epileptic disorders (Møller et al., 2019; Myers et al., 2018; Nakayama et al., 2018; Yang et al., 2017). One example is Dravet syndrome, which is a severe fever-sensitive, refractory epileptic disorder, often occurring in the infancy period. In a study from 2017, 56 paternal sperm samples were analyzed with droplet digital polymerase chain reaction (ddPCR) and ten of 56 fathers exhibited mosaicism of the proband's mutation (Yang et al., 2017).

Figure 4. Germline mosaicism. In this example, mutated cells (dark purple) in the gonads of the father leading to mutated cells in the entire child.

Some larger cohorts on parental mosaicism have been presented (Cao et al., 2019; Jónsson et al., 2018; Krupp et al., 2017; Rahbari et al., 2016; C. F. Wright et al., 2019). In 2019, Cao et al. investigated approximately 12,000 samples of SNVs in disease-causing genes (Cao et al., 2019). 2,373 were based on trio exome sequencing and 9,619 on proband-only exome sequencing (mean read depth 130x). Peripheral blood was analyzed. Parental mosaicism was identified in 0.3% of analyzed families. Two parents with mosaic variants exhibited phenotypes related to the mosaic change. Wright et al. reported in 2019 on 4,293 families of probands with severe developmental disorders (C. F. Wright et al., 2019). They re-analyzed exome data and used ultra-deep sequencing to validate candidate mosaic variants. Parental mutations were found in 21 trios (0.5%) based on blood-derived DNA.

Krupp et al. re-evaluated de novo SNVs found in children with autism spectrum disorders and found that parental mosaic variants accounted for 6.8% of the de novo variants in the children (Krupp et al., 2017). One Icelandic study launched an online calculator to estimate de novo mutation recurrence probability, based on trio-analysis from 251 couples (Jónsson et al., 2018). They concluded that depending on the properties of the de novo mutation, the recurrence probability ranges from 0.011% to 28.5%.

(33)

1.8 APPROACHES TO INVESTIGATE GENETIC MOSAICISM

Detection of mosaicism is challenging since it may be limited to certain tissue(s). It is sometimes not easy to predict which tissue might be affected and even if it is, that tissue may not be available for analysis. Moreover, even if DNA from affected tissue is analyzed, there is no guarantee to detect mosaicism since the affected tissue may be highly

heterogenous. On top of that, a method sensitive enough to detect the mosaic genetic change is needed.

There are several methods to consider for detection of mosaicism (table 1). Sanger sequencing has traditionally been used for many applications but has several limitations.

Since both alleles are sequenced simultaneously, low-level mosaicism cannot be accurately determined (Biesecker & Spinner, 2013).

Cytogenetic studies require actively dividing cells which limits the detection of mosaicism to certain tissue types. Chromosomal analysis for detection of cytogenetic mosaicism usually includes evaluation of 20 cells from peripheral blood. This has been calculated to rule out mosaicism down to 14%. If lower levels are suspected, more cells are counted (30 cells detect 10%, 50 cells detect 6% and 100 cells detect 3%) (Spinner & Conlin, 2014). In 2005, microarray-based techniques started to replace some of the traditionally cytogenetic analyzes. Single nucleotide polymorphism (SNP) array is more sensitive than array comparative genomic hybridization (CGH) for detection of mosaicism and mosaicism involving 5% of cells has been detected using these arrays (Biesecker & Spinner, 2013).

A relatively new and highly sensitive amplification technology that utilizes a water-oil emulsion droplet system is ddPCR (Hindson et al., 2011). Basically, input DNA is partitioned, along with PCR reagents, into approximately 20,000 droplets within a single thermocycled reaction well (Pinheiro et al., 2012). The principles are the same for digital PCR (dPCR). However, instead of droplets, the DNA mix is distributed over a nanofluidic chip that contains 20,000 wells. The methods serve essentially the same function as in a regular PCR assay, but each droplet/well is analyzed separately, and the fraction of PCR- positive droplets/wells in the original sample can be determined. With the standard setting, dPCR allows detection levels down to 0.1%. Therefore, it is a sensitive method and

amplification platform as a choice to identify and quantify low-level mosaicism.

Thanks to continuous improvements in massively parallel sequencing (MPS) technologies, read depth in excess of 1000x (ultra-deep sequencing) is possible (Mirebrahim, Close, & Lonardi, 2015). Read depth of a given region > 10⁸x has been

(34)

described (Jee et al., 2016). Mosaicism detection by MPS is suitable when screening for unknown mosaic variants, compared to ddPCR/dPCR, for which you need to design probes and primers for each specific target. The lower limit-of-detection (LOD) differs with various MPS techniques and even if many MPS platforms are able to identify variants <1%

variant allelic fraction or allele frequency (VAF), most are not designed to distinguish such findings from sequencing artefacts (Singh, 2020). However, studies have shown that joint analyzes of library-level replicates can reduce false positive signals (Kim et al., 2019).

Still, ddPCR/dPCR is considered to be a more robust method for confirmation and

quantification of mosaic findings (Gambin et al., 2020). With modern ddPCR technique, it is possible to merge droplets from several wells to a combined interpretation, resulting in LOD close to 0.001%.

Method LOD Detection of variant type

Sanger sequencing 20% Sequence variants within the exons

Standard chromosome analysis

14% Aneuploidy, large deletions, duplications, inversions, translocations

(balanced/unbalanced)

Standard WGS 10% All variants within the genome

Arrays 5% Large unbalanced deletions and duplications

Quantitative PCR 1% Targeted variants

Digital PCR 0.1% Targeted variants

Ultra-deep MPS 0.1% If WES, all variants within the exome. If a targeted gene panel; all variants within the coding regions of targeted genes

Droplet digital PCR 0.001% Targeted variants

Table 1. Summary of approaches to mosaicism detection. LOD=Limit-of-detection, indicating the lowest level of mosaicism possible to detect; WGS=Whole genome sequencing; PCR= polymerase chain reaction

MPS=Massively parallel sequencing.

(35)

2 RESEARCH AIMS

The overall aims of our research were to contribute to improved understanding of underlying mechanisms in some rare mosaic disorders, in order to improve genetic counselling and care of the patients and their families.

Specific aims of the thesis were to:

• Improve understanding of underlying mechanisms in PIK3CA- and DICER1-related overgrowth by identification, quantification and analysis of tissue distribution of the disease-causing variants (study I-II).

• Improve understanding of underlying mechanisms in focal dermal hypoplasia by identification, quantification and analysis of tissue distribution of the disease- causing variant (study III).

• Delienate the phenotypes associated with the investigated disorders (study I-III).

• Investigate whether mosaicism is present in parents to children diagnosed with rare intellectual disability syndromes caused by de novo small-scale variants (study IV).

(36)

(37)

3 PARTICIPANTS AND METHODS

3.1 PARTICIPANTS

All participants in our research have either been patients at the Department of Clinical Genetics, Karolinska University Hospital, or parents to patients investigated at the Department of Clinical Genetics, Karolinska University Hospital.

3.1.1 Ethical approval and informed consents

The studies were performed in accordance with the Declaration of Helsinki and approved by the local ethical board. Prior to inclusion to our studies, a written informed consent was signed by each participating individual or their legal guardians.

3.2 METHODS

3.2.1 DNA isolation (study I-IV)

DNA extraction from peripheral blood was performed at the Department of Clinical Genetics at Karolinska University Hospital following standard procedures.

Tissue biopsies were taken during surgery and sent to the Department of Clinical Genetics for DNA-extraction according to standard procedures.

Sperm samples were collected in 15 ml Falcon tubes and were stored at -18 °C until DNA extraction. DNA was isolated using the Qiagen Mini Amp Kit (Qiagen, Hilden, Germany) according to the manufacturer’s instructions for tissue isolation.

The concentrations were determined by Qubit^TM dsDNA Broad Range Assay Kit in Qubit^TM 3.0 fluorometer (Life Technologies, Carlsbad, CA, USA).

3.2.2 DNA sequencing (study I-IV)

As part of routine clinical diagnostics, whole exome sequencing (WES) or whole genome sequencing (WGS) were performed at Clinical Genomics at Science for Life Laboratory, using either HiSeq X (Illumina Inc, San Diego, CA, USA) or NovaSeq 6000 (Illumina Inc, San Diego, CA, USA) aiming at 100x (WES) or 30x (WGS) median read depth

(38)

(Stranneheim et al., 2021). Probands included in this cohort were diagnosed with genetic variants classified as “pathogenic” or “likely pathogenic” according to the American College of Medical Genetics and Genomics (ACMG) guidelines for interpretation of genetic variants (Richards et al., 2015).

In study I, DNA from patient 1 was extracted from paraffin-embedded tissue from affected muscle and analyzed with the Ion Torrent Oncomine Solid Tumor DNA Panel kit on the Ion Torrent S5 (ThermoFisher Scientific, Waltham, MA, USA) according to the

manufacturer's instructions. The analyzed genes were KRAS, BRAF, EGFR, NRAS and PIK3CA (NM_006218.1). In study III, DNA was sent to a commercial laboratory according to of that time current clinical routines for analysis of the PORCN gene. The PORCN gene was amplified by PCR and screened for variants by genomic sequencing of both DNA strands of the coding region and the highly conserved exon-intron splice junctions.

3.2.3 Digital PCR (study I, II, III)

Digital PCR was used in study I-III (fig. 5). Fifty nanograms of genomic DNA were amplified with 1X QuantStudio 3D Digital PCR Master mix (Applied Biosystems, CA, USA) and commercially available TaqMan^® assays (Applied Biosystems, CA, USA) were designed for our targets in the studies. Fifteen microliters of PCR reaction mixes were loaded into QS3D Digital 20K V2 chips (Applied Biosystems, CA, USA) which were then placed into GeneAmpÔ PCR System 9700 thermocycler (ThermoFisher Scientific,

Waltham, MA, USA). The chips were incubated at 96°C for 10 minutes, followed by 39 cycles at 54°C for 2 minutes and 98°C for 30 seconds. PCR was completed with a final incubation at 60°C for 2 minutes. After completion of PCR, each chip surface was cleaned with isopropanol to remove any dirt on the surface and scanned by using QuantStudio 3D reader (Applied Biosystems, CA, USA). Digital PCR data were analyzed by QuantStudio^â 3D AnalysisSuite^ä(version 3.1.2-PRC-build-03) using PoissonPlus algorithm (version 4.4.10). Using the 2D scatter plots, partitions for positive and negative amplification signals were classified for target and reference sequences using FAM and VIC channels. Each sample was run at least in duplicates.

(39)

3.2.4 Droplet digital PCR (study IV)

Droplet digital PCR was used in study IV (fig. 5). For each analyzed variant in this cohort, a TaqMan^® assay was designed by and ordered from ThermoFisher Scientific (Waltham, MA, USA). Probes detecting the mutant allele were labelled with FAM fluorophore and wild type allele with VIC fluorophore. Sixty-six nanograms of genomic DNA was mixed with ddPCR Supermix for Probes (No dUTP) (Bio-Rad, Hercules, CA, USA) according to the manufacturer’s protocol. The droplets were generated using QX200 Droplet Generator (Bio-Rad, Hercules, CA, USA) according to the manufacturer’s instructions. After the amplification was performed in CFX96 Real-Time Thermal Cycler (Bio-Rad, Hercules, CA, USA) according to the manufacturer's protocol with adjustments of annealing temperature for each probe (58.8-60.5°C), the droplets were scanned using QX200 QuantaSoft Droplet Reader (Bio-Rad, Hercules, CA, USA). QuantaSoft Analysis Pro Software (Bio-Rad, Hercules, CA, USA) was used to analyze ddPCR data through a Poisson distribution. After analysis, each data was manually inspected and when needed, the droplets were further grouped using the manual selection tools in the software. Each sample was run in triplicates at minimum.

Figure 5. Workflow for dPCR/ddPCR. DNA was extracted from a selected sample. The sample was partitioned into many separate PCR reactions and read out as positive signals for the designed target or reference sequence. An absolute quantification of signals was done and graphical plotted through a software program.

3.2.5 Array comparative genomic hybridization (study III)

For array CGH, a custom 4x180K array CGH platform was used (Oxford Gene Technologies, Oxfordshire, UK). This platform has a genome wide average base pair spacing of about 20 Kb. Experiments were performed according to the manufacturer’s protocol.

(40)

3.2.6 Pathology (study I, III)

Formalin fixed paraffin embedded (FFPE) sections of affected tissue stained with

Hematoxylin-eosin and van Gieson as well as with antibodies against p62 were analyzed for microscopic evaluation. In study I, biopsies from ectopic dorsal muscle of the right hand (patient 1), ectopic dorsal interosseous muscle (patient 2) and ectopic extensor digitorum muscle (patient 2) were analyzed. In study III, skin biopsies were performed on atrophic patches.

(41)

4 RESULTS AND DISCUSSION

4.1 STUDY I

In a male patient with ectopic muscles and unilateral isolated muscular overgrowth, we identified a mosaic PIK3CA hotspot variant in affected muscle tissue; c.3140A>G,

p.(His1047Arg) in 1154 out of 6524 reads (18%). The variant was confirmed and quantified using dPCR, which showed VAF 18% (fig. 6a). In a female patient with bilateral isolated muscular overgrowth of both upper limbs, more than 13 ectopic muscles were found during surgery. A mosaic PIK3CA hotspot variant; c.1624G>A, p.(Glu542Lys) was found in 10 out of 42 WGS reads (24%) in the affected tissue. The variant was confirmed and

quantified using dPCR, which showed VAF 26% in muscle biopsy I and 17% in muscle biopsy II (fig. 6b). The variants were not detected in blood samples from any of the patients.

Figure 6. Genetic and pathologic findings in patients in study I. A-B: Representative results from dPCR of DNA from ectopic muscle biopsies. Digital PCR quantifies the load of mosaic mutations, c.3140A>G, p.(His1047Arg) (A), and c.1624G>A (p.Glu542Lys) (B), in DNA extracted from ectopic muscles from patient 1 (A) and 2 (B). Blue cluster (FAM) shows signals from mutant allele, red signals (VIC) from reference allele and green signals from both mutant and reference alleles. Yellow cluster represents the wells where no amplification signal was detected. C: Histopathological findings from affected muscle of patient 1. Severe changes with increase in connective tissue (van Gieson). D: Histopathological findings from affected muscle of patient 2. Scattered rounded eosinophilic muscle fibers in otherwise normal looking tissue (Hematoxylin- eosin). Bars: 100 μm

(42)

Here, we establish that isolated upper limb muscle overgrowth is a phenotype within

PROS. In our described patients, PIK3CA mutated cells seem to have obtained the ability to become muscles when they normally should differentiate to tendons or fascia. Our findings indicate that PIK3CA plays an important role during early human development. It is also interesting whether PIK3CA mosaic variants are associated with an increased malignancy risk or not. Wilms tumor in PROS might be explained by early mosaic mutations since muscles and kidneys origin from the same migrating cells that later develop the mesoderm.

Possibly, the PROS patients who are at risk of Wilms tumor are those with a significant number of PIK3CA-mutated renal cells (Biderman Waberski et al., 2018). Therefore, ddPCR analysis of cell-free DNA in urine could possibly identify PROS patients at risk of Wilms tumor. Still, more studies are needed.

4.2 STUDY II

In study II, we showed that an individual with a heterozygous germline DICER1 variant (c.4031C>T, p.Ser1344Leu), encoding the RNase IIIa domain, has a severe form of DICER1 syndrome (’DICER1 syndrome plus’). The patient present with intellectual disability, macrocephaly, physical abnormalities, Wilms tumor, and a well-differentiated fetal adenocarcinoma of the lung. This phenotype resembles the GLOW subphenotype of DICER1 syndrome which has been described in two children with mosaic findings in another domain, RNase IIIb, of the DICER1 gene. By performing dPCR, we could confirm that the variant was in a non-mosaic state (table 2, fig. 7). DNA prepared from kidney and lung tumor tissue revealed that the variant was more abundant in the tumor tissue, 87% and 90% respectively, suggesting loss-of-heterozygosity (LOH) (table 2). To verify LOH in the tumor, DNA from the kidney tumor was subjected to WGS. Whole genome sequencing data confirmed a 0.61 Mb deletion covering the whole DICER1 gene (fig.7).

Table 2. Summary of genetic findings of the DICER1 variant c.4031C>T in different tissues in study II.

(43)

There are close similarities to the phenotype of our patient and two children reported with GLOW syndrome due to mosaic hotspot variants in the RNase IIIb domain, although the postnatal overgrowth was more pronounced in these individuals. Activation of the

PI3K/AKT/mTOR pathway has been described in genetically modified cells with hotspot variants in the RNase IIIb domain (S. D. Klein & Martinez-Agosto, 2019). It is noteworthy that some patients with variants in the PI3K/AKT/mTOR pathway have overlapping symptoms with our patient, such as polydactyly in patients with variants in AKT3 or PIK3CA, and multiple nevi, macrocephaly and ID seen in PTEN- and PIK3CA-related disorders. Posterior helical pits are rare and have previously been described in BWS, SGBS and GLOW syndrome, all of which are cancer predisposition syndromes.

Figure 7. Digital PCR results of DNA from skin biopsy (a) and WGS SV filtration of DNA from kidney tumor from the DICER1 patient (b). A: Blue cluster represents amplification of the target region. Red signals represent the reference sequence and green signals both mutant and reference alleles. Yellow cluster represents the wells where no amplification signal was detected. B: A 0.61 Mb deletion covering the whole DICER1 gene was detected with WGS.

In addition to our patient, we identified RNase IIIa or RNase IIIb domain hotspot variants in 27 sequenced Wilms tumor. Among them, 39% had alterations on both alleles, while 61% had only one single allele affected. It would be interesting to find out if these tumors have copy neutral LOH or a deletion as in our patient, or other genetic alterations that escaped detection. The severe phenotypes seen in patients with germline or mosaic RNase III hotspot variants might be explained by a second somatic mutational event stochastically occuring in DICER1, in combination with tissue-specific neomorphic outcomes of that specific mutational event.

(a) (b)

STUDIES OF GENETIC MOSAICISM IN RARE DISEASES

Department of Molecular Medicine and Surgery Karolinska Institutet, Stockholm, Sweden

STUDIES OF

GENETIC MOSAICISM IN RARE DISEASES

Studies of Genetic Mosaicism in Rare Diseases THESIS FOR DOCTORAL DEDGREE (PH.D.)

Sofia Frisk

PREFACE

POPULÄRVETENSKAPLIG SAMMANFATTNING

ABSTRACT

LIST OF SCIENTIFIC PAPERS

TABLE OF CONTENTS

LIST OF ABBREVIATIONS

1 INTRODUCTION

2 RESEARCH AIMS

3 PARTICIPANTS AND METHODS

4 RESULTS AND DISCUSSION