De novo mutations in zebrafish mitochondrial DNA

(1)

1

De novo mutations in zebrafish

mitochondrial DNA

(2)

2

De novo mutations in zebrafish

mitochondrial DNA

Performance data

:

Version 1

Date: 10-06-15 Author:

Adriana Jenneke (Rianne) Timmer

aj.timmer@student.avans.nl

Student number: 2048574

Internship supervisors: Graduation Internship Jo Vanoevelen

Start date: 19-01-2015 j.vanoevelen@maastrichtuniversity.nl

End date: 12-06-2015 Auke Otten

Maastricht, The Netherlands auke.otten@maastrichtuniversity.nl

Teacher-supervisor: Internship institution:

Pascal Hommelberg Maastricht UMC+

pph.hommelberg@avans.nl Department: Clinical Genomics Universiteitssingel 50

6229 ER Maastricht

University: Phone: 043-3875843

Avans University of Applied Science Course: Forensic Laboratory Research

Lovensdijkstraat 61-63 In collaboration with:

4818 AJ Breda Université de Liège, Belgium, Unité

(3)

3

Preface

A little more than a year ago I had to find an internship for the last year of my study. I knew that I liked to work for a university medical center so I tried to find an internship in Nijmegen or Maastricht, since I liked these cities. Unfortunately, in Nijmegen there were no opportunities for an internship. However, in Maastricht I could choose between two projects. One project focused on Chlamydia. The other project focused on mitochondrial DNA and zebrafish. At first I was not sure if I was a suitable person to work with zebrafish, since I did not had any experience in working with animals and also I do not feel comfortable to hurt animals. However, the zebrafish project attracted me more than the Chlamydia project and so I chose this project. Fortunately, nine months after I started this project, I can say that I do not regret this choice. I had an amazing time during my internship and for this I want to thank everyone of the Clinical Genomics department. However, I want to mention a few people in particular. Firstly, I would like to thank my supervisor, Auke Otten for his commitment and guidance, both theoretical and practical, during this project. I could not have had a better supervisor! I also want to thank my other supervisor Dr. Jo Vanoevelen for carefully reading this manuscript and all the other reports that I have written during this internship and for infecting me with the ‘zebrafish virus’. I thought it was quite fascinating to work with zebrafish! I also want to thank a few other colleagues, namely: Ivo Eijkenboom, Erika Timmer, Ellen Lambrichs and Bieke Vanherle, for their guidance and assistance at the lab and also their opinions about other matters. Finally, I would like to thank Dr. Pascal Hommelberg for his guidance from school.

Abstract

Mitochondrial DNA differs from nuclear DNA in many ways. Mitochondrial DNA is present in higher and variable amounts per cell. The amount of mitochondrial DNA is called the mitochondrial DNA copy number. Mitochondrial DNA is only inherited via the maternal lineage. In both mitochondrial and nuclear DNA de novo mutations can occur. De novo mutations are mutations which are only present in the germ cells or the progeny of the parents and not in the genomes of the parents. The occurrence of de novo mutations is known, however the rate at which they occur is not known. It is expected that a low copy number will result in a higher susceptibility for de novo mutations than an average or high copy number. The effect of the copy number on the mutation rate is also not known. To investigate the de novo mutation rate and the effect of the copy number on the mutation rate, the mitochondrial genome of four tissues and oocytes of seven female zebrafish was sequenced (using Next-Generation Sequencing) and analyzed. Mitochondrial DNA copy number of the oocytes was also determined using qPCR. The average copy number was 34 million copies per oocyte. Total amount of de novo mutations in all the oocytes varied from 23 to 29 mutations. Consequently, the de novo mutation rate varied from 1,36 x 10-5 to 1,71 x 10-5 mutations per site. No significant difference was found between copy numbers of the oocytes with a de novo mutation and without a de novo mutation.

Keywords

(4)

4

Samenvatting

Mitochondriaal DNA en nucleair DNA verschillen in een aantal opzichten. Mitochondriaal DNA is onder andere in veel grotere en variabele getalen aanwezig in de cel. De hoeveelheid mitochondriaal DNA wordt kopie nummer genoemd. Mitochondriaal DNA erft alleen via de moederlijke lijn over en in dit DNA kunnen de novo mutaties plaatsvinden. De novo mutaties zijn mutaties die aanwezig zijn in de genomen van geslachtscellen of nakomelingen van individuen. Waarbij deze mutaties niet in het genoom van de individuen aanwezig zijn. Het is niet bekend met welke frequentie de novo mutaties plaatsvinden in mitochondriaal DNA. Om de de novo mutatie frequentie en het effect van het kopie nummer op de frequentie te bepalen, werd het mitochondriale genoom van vier weefsels en verschillende oocyten van zeven vrouwelijke vissen gesequenced (door middel van Next-Generation Sequencing) en geanalyseerd. Er werd verwacht dat een laag kopie nummer een effect heeft op de de novo mutatie frequentie in vergelijking tot een gemiddeld of hoog kopie nummer. Echter was het nog niet bekend of het kopie nummer een effect heeft. Hierdoor werd het kopie nummer in de eicellen ook bepaald door middel van qPCR. Het gemiddelde kopie nummer in de eicellen was 34 miljoen kopieën per cel. Het totale aantal de novo mutaties in alle oocyten varieerden van 23 tot 29 mutaties. De de novo mutatie frequenties varieerden van 1,36 x 10-5 tot 1,71 x 10-5 mutaties per site. Er werd geen significant verschil gevonden tussen de kopie nummers van eicellen zonder de novo mutaties en eicellen met de novo mutaties.

(5)

5

1. Introduction

Mitochondria are essential for eukaryotes as their primary function is supporting the aerobic respiration and providing ATP. Mitochondria contain DNA, which is called mitochondrial DNA (mtDNA). Mutations in both mtDNA and nuclear DNA can cause mitochondrial diseases. Mitochondrial diseases can cause a great burden on the lives of relatives of individuals, carrying these diseases, and the individuals themselves. There is no cure available yet for these diseases and so it is important to discover how these diseases are inherited. Consequently, preventing mitochondrial diseases will be easier if the inheritance patterns are known. MtDNA is inherited only via the maternal lineage and is present in much higher amounts then nuclear DNA. The amount of mtDNA molecules (mtDNA copy number) can vary between individuals and also the sequence of mtDNA molecules is variable. The presence of mtDNA molecules with varying sequences is called heteroplasmy and is expressed in percentages. The percentages heteroplasmy can vary between a mother and her off-spring, since a mitochondrial bottleneck occurs after fertilization of the oocyte. The heteroplasmy percentage in a subsequent generation can also be altered by de novo mutations. [1] De novo mutations are mutations which are solely present in germ cells or progeny of the parents and not in the genomes of the parents. Occurrence of de novo mutations is known, however the rate at which they occur is not known. It is expected that a low copy number will result in a higher susceptibility for de novo mutations than an average or high copy number. However, the effect of the copy number on this mutation rate is also not known.

The aim of this study was, determining the de novo mutation rate in oocytes of zebrafish and also whether the mitochondrial copy number has an effect on the de novo mutation rate. This is important to study, since better understanding the mitochondrial bottleneck would improve genetic diagnostics. Consequently, genetic screening will be able to advise parents with affected children better regarding getting more children and the likelihood that these children will also be affected with a mitochondrial disease. The determination of the de novo mutation rate was performed by extracting mitochondrial DNA of four different tissues and fourteen to fifteen oocytes of several female zebrafish. The mitochondrial genome of the tissues and oocytes was amplified in three overlapping fragments by a Long-Range PCR. Next, the amplified genomes were sequenced using Next-Generation Sequencing. The mitochondrial copy number in the oocytes was determined by performing a qPCR with extracted mtDNA of these oocytes. It was expected that both the de novo mutation rate and the effect of the mtDNA copy number on the de novo mutation rate could be determined.

The structure of this report is as follows. Chapter 2 provides the theoretical background and the research model. Chapter 3 describes the materials and methods and in chapter 4, the results are shown. Finally, chapter 5 describes the discussion and the end conclusions. Following chapter 5, the references and the supplemental sections are shown.

(7)

3

2. Theoretical background

2.1 Mitochondrial DNA

Mitochondria are cell organelles found in most eukaryotic cells and they provide cells the capability to generate energy. [2] Like all other organelles mitochondria can also malfunction and in the early 1960s the link between malfunctioning mitochondria and certain diseases was discovered. [3] In 1963 it was shown that besides the well-known nuclear DNA (nDNA), also mitochondrial DNA (mtDNA) is present in cells. [4] In contrast to nDNA, mtDNA is circular and the mtDNA molecule is much smaller than nDNA. The length of mtDNA molecules is approximately 16,6 kb and encodes for 22 transfer RNAs (tRNAs), two ribosomal RNAs (rRNAs) and thirteen proteins. The locations of these genes and the D-loop in the mtDNA molecule are shown in figure 1. [1] The D-loop region is a small non-coding region. However, it does contain regulatory elements for replication and expression of the mitochondrial genome. [5] MtDNA molecules contain virtually no introns in comparison to nDNA. Consequently, if a mutation occurs in mtDNA, it is more likely that the mutation will occur in an exon than in an intron. This causes the mtDNA to be more susceptible to mutations with functional consequences, in comparison to nDNA. Although mtDNA molecules encode for the basic protein synthesis machinery, it is still dependent on nDNA for providing enzymes for replication, repair, transcription and translation. Consequently, both mutations in mt- and nDNA can result in mitochondrial dysfunction. [1]

Figure 1: A schematic overview of human mtDNA. The tRNAs, rRNAs and proteins for which mtDNA encodes for are shown. Also the D-loop region and OH and OL, are shown, which are important for replication of mtDNA. [6]

(8)

4 2.1.1 Oxidative phosphorylation

Mitochondria are involved in cellular homeostasis, signaling and apoptosis. [3] However, the fundamental role, of mitochondria, is in the respiratory chain for adenosine triphosphate (ATP)-production. The ATP production is also called oxidative phosphorylation (OXPHOS). The OXPHOS pathway takes place after glycolysis and the citric acid cycle. During these processes reduced nicotinamide adenine dinucleotide (NADH) and reduced flavin adenine dinucleotide (FADH2) are

produced. NADH and FADH2 will be used as reduced cofactors in the OXPHOS pathway. The OXPHOS

pathway consists of five different enzyme complexes, which are imbedded in the inner mitochondrial membrane. The complexes consist of over 85 subunits and 13 of those subunits are encoded by mtDNA. Complex I is the largest complex, and is called NADH dehydrogenase. Complex II, III and IV are called respectively: succinate-Q oxidoreductase, Q-cytochrome c oxidoreductase and cytochrome c oxidase. Complex V is the last complex and is called ATP synthase. During OXPHOS NADH and FADH2 will donate

electrons to complex I and III, respectively. These electrons will flow down the respiratory chain via ubiquinone (ubiq) and cytochrome c (cyt c) to the complexes III and IV. The liberated energy is used by complex I, III and IV to actively pump protons out of the mitochondrial matrix and into the intermembrane space. The electrochemical gradient, which is created by pumping protons out of the matrix, is utilized by complex V for the synthesis of ATP. ATP is synthesized from adenosine diphosphate and inorganic phosphate. In figure 2 the order of the OXPHOS pathway is shown. [1, 7]

Figure 2: OXPHOS pathway in the inner membrane (IM) of mitochondria. Complexes 1 to 5, ubiquinone, cytochrome c, UCP and ANT are shown in this membrane. Also the outer membrane (OM), used products and created byproducts are shown. The order of the pathway is indicated by arrows. [7]

Following the synthase, matrix ATP is transferred to the intermembrane by adenine nucleotide translocase (ANT) in exchange for ADP. The uncoupling proteins (UCP) in the intermembrane allows protons to re-enter the matrix and this will reduce the formation of reactive oxygen species (ROS) in the matrix. This is important, since ROS can damage DNA and especially mtDNA is very susceptible to damage of ROS. The susceptibility for ROS is caused by the proximal location of mtDNA to ROS and the lack of histones and mitochondrial repair mechanisms. [7]

(9)

5 2.1.2 MtDNA replication

Replication of mtDNA occurs in both dividing and non-dividing cells and is continuous and independent of the nuclear replication. [3] Replication starts at the origin (OH) of the heavy strand (H-strand), at

which the H-strand separates of the light strand (L-strand). The single-stranded H-strand is called the displacement loop (D-loop). After separation a new H-strand will be synthesized. Elongation of this strand is unidirectionally, displacing the parental H-strand and enlarging the D-loop. Enlargement of the D-loop proceeds until the origin of the light strand (OL) is exposed at two-third of the molecule. At the

exposure of the OL, the new L-strand will be synthesized in opposite direction of the expanding loop. An

intermediate, which is called, Exp-D (l) will arise, in which two partially double stranded mtDNA molecules exist in the form of one molecule. Upon completion of the molecule with the new H-strand, the molecules will separate. Consequently, the molecule with the new L-strand will be incomplete after the separation and this molecule is called a gapped circle. In figure 3 the replication is shown. [1, 8-10]

Figure 3: Replication of the mtDNA molecule. A new H-strand is synthesized at the origin of the H-strand, creating a D-loop. The H-strand is elongated, expanding the D-loop. At the exposure of the origin of the L-strand, a new L-strand will start to form in the opposite direction of the single-stranded H-strand. When the daughter mtDNA molecules will separate one of the molecules will be incomplete. [10]

All cells contain two copies of nDNA, except for erythrocytes and germ cells. Germ cells contain only one copy of nDNA and erythrocytes contain no nDNA. During the differentiation of erythrocytes organelles, such as the nucleus and mitochondria, are ejected. Consequently, erythrocytes also do not contain mtDNA. [11] Cells which contain mitochondria can however contain varying amounts of mitochondria. This depends on the energy requirements of the tissue, of which the cell is part of. Brain and muscle tissue require large amounts of ATP, so the amount of mitochondria in these cells is high. The amount of mtDNA copies per mitochondrium is equal for different cell types. Consequently, the amount of mtDNA copies also varies between different cell types, high energy demanding tissues contain more mtDNA copies than low energy demanding tissues. Per cell there are 500 to 1.000 mtDNA copies present and their sequences are not identical. [1, 2, 12-14] However, in a healthy individual most copies will be the same. The state in which all copies have the same sequence is called homoplasmy and this is also the ideal situation. However, mutations do occur in mtDNA and this creates a mixture of different mtDNA molecules. This state is called heteroplasmy and is expressed in percentages. The percentage indicates the proportion mutant mtDNA to wild-type mtDNA. However, the percentages 0% and 100% are called homoplasmy, since only one sort of mtDNA copy is present. [1, 15]

Mutations in mtDNA can be inherited, but they can also arise spontaneously during life or occur spontaneously in a germ cell (de novo). All these mutations can give rise to a very similar amino acid or the same amino acid, which will not affect the function of the protein. MtDNA mutations can also give rise to an altered amino acid and be pathogenic. However, the presence of a pathogenic mutation will not always result in symptoms of a mitochondrial disease. This is caused by the variable heteroplasmy percentage. Many studies have been performed to determine the effect of the percentage heteroplasmy on mitochondrial metabolism and cell phenotype. During these studies cells lacking mtDNA were used and mtDNA of patients was transferred to these cells, creating transmitochondrial cell lines (cybrids) with known percentages of heteroplasmy, ranging from 0 to 100%. Activity of the

(10)

6 respiratory chain and cell growth were measured in these cybrids. The results of this study showed that a low percentage of heteroplasmy does not result in an altered phenotype. However, if the percentage heteroplasmy was sufficiently high, the phenotype was altered. This showed that a certain threshold value exists for mitochondrial mutations. When the threshold value is exceeded, symptoms will arise. However, when the percentage heteroplasmy remains below the threshold value, a normal phenotype will occur. It was also determined that for mtDNA deletions the threshold value is 60% and for other mtDNA mutations the threshold value is approximately 90%. Thus normal proteins, mRNAs and tRNAs can complement the mutants to a certain extent. However, in some organs the threshold value can be lower or higher than in other organs before symptoms will arise. [15, 16]

2.2 Mitochondrial bottleneck

MtDNA is maternally inherited, although the percentages heteroplasmy between a mother and her off-spring can vary. During a study in a female Holstein cow with neutral mtDNA variants and her off-off-spring it was shown that allele frequencies could rapidly shift and become fixed in a few generations. It was proposed that this was caused by a mitochondrial genetic bottleneck. During this bottleneck a reduction of mtDNA copies occurs after which these copies would become the founder molecules for the next generation. [17, 18] In the study of Jenuth et al. [17] two mechanisms for the bottleneck, which occur at different stages, were hypothesized. The first mechanism occurs during maturation of primary oocytes and in this mechanism the mtDNA copy number would increase a 100-fold from approximately 103 to 105. During the copy number increase, a selected subpopulation would be replicated. This could cause a rapid shift in heteroplasmy percentages between generations. The second mechanism that was hypothesized was a reduction of mtDNA copy number during the first numerous cell divisions after fertilization of an oocyte. During the cell divisions no mtDNA is replicated and so the mtDNA is continuously divided among the daughter cells. These cells can contain differing mitochondrial genotypes, since mtDNA partitioning is stochastic. This can also cause a rapid genetic shift between generations. However, it was shown that neither mechanisms that were hypothesized were correct. [17] Human mature oocytes contain over 100,000 mtDNA molecules and upon fertilization they start dividing. [19] After the seventh cell division the cells are called a blastocyst. The blastocyst is spherical and in the blastocyst a fluid-filled cavity forms. Outer cells of the blastocyst form the trophoblast which will form the fetal part of the placenta. In the blastocyst, the inner cell mass will form which will develop into the embryo. [20] The inner cell mass develops into the epiblast and hypoblast layers and in the epiblast layer, precursors of primordial germ cells (PGCs) will develop. After the development of the precursors of PGCs, they are moved to an extra-embryonic region where PGCs are determined. During gastrulation, which occurs after the blastocyst stage, the gonadal ridges arise and these will develop into the gonads. PGCs will migrate during the gastrulation to the gonadal ridges and during migration they proliferate. [21] During this early embryogenesis the amount of mtDNA remains constant although cell divisions do occur. Consequently, the copy number in the cells and in the PGCs will be reduced to ~200 before replication is initiated again. [15, 22] This is the pre-implantation bottleneck. Next, the mtDNA copy number is increased again to 100.000 in PGCs and this is the post-implantation bottleneck. The two bottlenecks ensure rapid genetic drift and because of this, deleterious mutations can be removed and homoplasmic mtDNA can be maintained. [23, 24] The two mitochondrial bottlenecks and their effect on mtDNA copy number is shown in figure 4. In a study of Fan et al. it was showed that the bottlenecks have the best purifying effect on deleterious mutations. However, no selection for milder mutations was observed. [25]

(11)

7

Figure 4: The two mitochondrial bottlenecks and their effect on mtDNA copy number. After fertilization numerous cell divisions occur and replication has stalled and so the copy number is reduced from ~100.000 to ~200. This is the pre-implantation bottleneck. As soon as the low copy number is reached, replication will be initiated so the copy number will increase again to 100.000 in mature oocytes. This is called the post-implantation bottleneck. [23]

To assess the effect of the two bottlenecks a model was established, for the segregation and replication of mtDNA between dividing cells in the germ line. This showed that the pre-implantation bottleneck accounts for 70% of the variability and the post-implantation bottleneck accounts for 30% of the variability. However, the bottlenecks are not infallible, since some deleterious mutations are still transmitted to the progeny. [18, 24] In addition to this, there is also a risk of de novo mutations during the bottleneck. Especially during the pre-implantation bottleneck de novo mutations pose a major risk. The copy number is low during this bottleneck and a mutation would result in a greater increase of heteroplasmy during this bottleneck than it would in a somatic cell.

2.3 Research model

To gain more information about the effect of the bottleneck and the risk for de novo mutations it would be most preferable to use human oocytes. However, human oocytes are difficult to obtain and also solely studies, in which discarded oocytes of IVF procedures are used, are allowed. [26] Thus, an animal research model was required and in this study Danio rerio will be used. Danio rerio is the scientific name for zebrafish. Zebrafish live in tropical freshwater and originate from the area of Sri Lanka and India. Spawning in natural habit is induced by light, temperature and happens at the start of the monsoon season. However, in laboratory settings zebrafish can breed all year round, producing hundreds of off-spring per mating per week. Also unfertilized oocytes can easily be collected by squeezing the abdomen of mature females. [27] Production of mature oocytes does differ between humans and zebrafish. Oogonia in zebrafish keep on proliferating and oogonial mitosis is only activated after ovulation in the ovarian lumen. Consequently, the oocyte stock will remain young. Human oogonia already undergo mitosis during week 9-22 of the embryonic development and so they are subjected to aging more than zebrafish oocytes (A. B. Otten, unpublished review). Also, the development of a zebrafish embryo is different in comparison to the development of a human embryo. MtDNA of zebrafish and humans however, is very comparable. The length of the mtDNA genome in zebrafish is 16596 bp and it also encodes for thirteen proteins, twenty-two tRNAs and two rRNAs. [28] Preliminary data of our group also showed that the mitochondrial bottleneck occurs in zebrafish. This ensures that zebrafish are a suitable model for studying mitochondrial DNA and the mitochondrial bottleneck.

(12)

8

2.4 Next-Generation Sequencing

To determine if de novo mutations are present, a technique called massively parallel sequencing was used. This name is due to the ability of sequencing multiple DNA fragments at the same time. A more common name for this technique is next generation sequencing (NGS). DNA samples must undergo extensive sample preparation before NGS can be performed. Figure 5 shows an overview of this preparation. During preparation the DNA will be fragmented into parts of approximately 100 bp. This will be followed by ligating adapters to both sides of the fragmented DNA. The fragmented DNA with the adapters is then transferred to a solid carrier. The sequencing reaction takes place on a solid carrier, in this case a glass plate. In figure 6, the sequencing reaction is shown. The glass plate contains oligonucleotides, which have complementary sequences, to the adapters which are ligated to the DNA. Consequently, the DNA will attach to the glass plate. When the DNA is attached bridge amplification will be performed. During bridge amplification the free adapter of the DNA will bend over and attach to an oligo with a complementary sequence. The DNA will be amplified creating a forward and reverse strand. These strands will be separated and in the next cycle of the bridge amplification the DNA strands will bend over to other oligos nearby creating more strands. The amplicons will also be attached to the cell so DNA strands with the same sequence will form clusters. When the clusters are created the reverse strand will be denatured and washed away. During the first sequencing round the read number one primer will bind to a part of the adapter which is complementary. Fluorescent labeled nucleotides with inactive ‘3 OH will be added. The ‘3 OH is inactivated so only one nucleotide will bind to the DNA per cycle. A picture will be taken of the cell and in this image it can be seen which fluorescent color is emitted by which cluster. Subsequently, the fluorescent label will be washed away and the ‘3 OH will be activated and a new cycle will start. During the second read the two indexes of the adapters will be sequenced, for identification of the samples. After this, the number two primer will bind to the complementary part of the adapter and the reverse strand will be sequenced just like the forward strand. [29]

Figure 5: Sample preparation by fragmenting the DNA to Figure 6: The adapters are indicated with purple and blue parts of 300 bp and ligating adapters to both sides of the and the DNA fragment with green. Following the DNA DNA fragments. The adapters contain two parts where the preparation which is shown in figure 4, the DNA is fragment can attach to the flow cell (indicated in red and amplified in the flow cell by bridge amplification. When the orange). The adapters also contain two index parts (purple clusters are created the DNA will be sequenced by and pink), which are used for identifying the samples and adding one nucleotide per cycle to the DNA. After each as last a read one (blue) and two (green) primer part. Both cycle the cell will be imaged and the added nucleotides can figure 4 and 5 were adapted from the website of Illumina. be identified.

(13)

9

3. Materials and methods

3.1 Sample collection and mtDNA extraction

Oocytes and tissues of the female zebrafish were collected by experienced personnel of the Unité de Biologie Moléculaire et Génie Génétique, Université de Liège, Belgium. Zebrafish were sedated with Tricaine (0,8 mg/ml) and oocytes were collected by squeezing the ovaries of the fish. Oocytes were collected in 4,33 μl 50 mM NaOH and frozen on dry-ice. Zebrafish could recover for a period before they were sacrificed in ice-cold water. Brain, gonads, liver and a piece of muscle were removed according to standard protocols and collected in 450 μl Nuclei Lysis Solution of Wizard Genomic Purification kit (A1120, 1125 and 1620, Promega Benelux b.v., Leiden, The Netherlands). Samples were stored at -80°C. MtDNA of brain, gonad, liver and muscle of the female fish was already extracted during a previous study with the use of the Wizard Genomic Purification kit according to the manufacturer’s specifications with minor adaptions (Supplemental section I). In short, after thawing the organs with Nuclei Lysis Solution 37,5 μl Proteinase K (20 mg/ml) was added. Samples were spun down shortly and incubated overnight at 56°C (PHMT-PSC18, Fisher Scientific Netherlands, Landsmeer, The Netherlands, used for all incubation steps above room temperature). Following the incubation, 2,5 μl RNase A solution (4 mg/ml) was added and the solution was incubated at 37°C for 30 minutes. Next, the samples were allowed to cool down to room temperature and 200 μl Protein Precipitation Solution was added. Samples were vortexed at full speed for 20 seconds and spun down for 10 minutes at 2.000 x g (Eppendorf, 5415D and R, Eppendorf Nederland B.V. Nijmegen, The Netherlands, used for all centrifuge steps). Subsequently, the supernatant was transferred to a new tube and 1 volume of isopropanol was added. Samples were mixed by inversion and centrifuged for 20 minutes at 24.100 x g at 4°C. Next, the supernatant was removed and 1 ml 70% ethanol was added. After this, the samples were centrifuged again at 24.100 x g and 4°C, for 5 minutes. The supernatant was removed and the DNA pellet was shortly air-dried. Finally, the DNA was dissolved in 30 μl Tris-EDTA (TE) buffer (containing 10 mM Tris-Cl and 1 mM EDTA).

MtDNA of the oocytes was extracted by adding 500 μl Lysis buffer (containing 75 mM NaCl, 50 mM EDTA and 20 mM HEPES) to the samples and then the samples were mixed. 10 μl 20% SDS and 20 μl Proteinase K (10 mg/ml) was added and the samples were mixed again. This was followed by an incubation for 4 hours at 50°C and after 2 hours the samples were mixed by inverting 5 times. Next, 420 μl isopropanol was added to the samples. Samples were mixed by inverting 15 to 20 times and incubated overnight at -20°C. Following the incubation, the samples were centrifuged for 30 minutes at 16.000 x g at 4°C. The supernatant was removed and 500 μl of 70% ethanol was added. Samples were centrifuged again for 10 minutes at 16.000 x g at 4°C and all the supernatant was removed. Finally, the DNA was dissolved in 10 μl TE buffer.

3.2 Long-Range PCR

MtDNA was amplified with the use of three different overlapping primers, amplifying fragments A, B and C with a length of 6000, 5651 and 5722 bp, respectively. These fragments cover the entire mitochondrial genome. A schematic overview of the overlapping fragments is shown in figure 7. Primers for fragment A were designed with the use of NCBI/Primer-BLAST.[30] Primers for fragments B and C were designed in a previous performed study using Primer 3.0 Software (v. 0.4.0) (Whitehead Institute for Biomedical Research, A. Untergasser et al. [31]). Primers were ordered from Invitrogen (Life Technologies Europe B.V., Bleiswijk, The Netherlands) and the sequence of the primers is displayed in table 1. The mitochondrial

Figure 7: The mitochondrial genome is shown in black. In red, blue and green the three fragments are shown. All fragments overlap with the two other fragments.

(14)

10 genome of the four tissues of the female zebrafish was already amplified during a previous study. The PCR mixture of 50 μl contained 1 x GC Buffer, 1 Unit Phusion Hot Start II High-Fidelity DNA Polymerase (F-549S, F-549L, Fisher Scientific, Landsmeer, The Netherlands), 0,2 mM deoxyribonucleotide triphosphate (BIO-39025, BIO-39026, BIO-39027, BIO-39049, Bioline, GC Biotech B.V., Alphen aan den Rijn, The Netherlands), and 0,5 μM forward and reverse primer. Per PCR 2 µl of extracted DNA of the oocytes was used and MilliQ was added to create a volume of 50 μl. The Long-Range PCR program consists of the following steps: first 30 seconds at 98°C, then 40 cycles consisting of 10 seconds at 98°C, 20 seconds at 58°C and 8 minutes at 72°C and as last 10 minutes at 72°C and a hold on 4°C. PCR was performed in a TProfessional Thermocycler of Biometra (Westburg B.V., Leusden, The Netherlands). The PCR products of the Long-Range PCR were visualized on 1% agarose (Ultrapure Agarose, 16500-100, 16500-500, Invitrogen, Life Technologies Europe B.V., Bleiswijk, The Netherlands) gels, containing 1 μl of 0,625 mg/ml Ethidium Bromide (E406-5ML, E406-15ML, Amresco, VWR International B.V., Amsterdam, The Netherlands) per 10 ml gel with O’Generuler mix (DNA ladder, ready-to-use 100 – 10.000 bp, Fisher Scientific, Landsmeer, The Netherlands), as the DNA molecular weight marker.

Table 1: Sequences of the different primer sets for the Long-Range PCR and the qPCR.

Primer set Forward primer 5’-3’ Reverse primer 5’-3’

A CACACCCCTGACTCCCAAAG GGTCGTTTGTACCCGTCAGT

B AAATTAACACCCTAACAACGACCTG GGGGATCAGTACTTTTAGCATTGTAGT

C AAGTTTATCCACAGCCCCTATACTACT TATTGGTGGTCTCTCACTTGATATG

qPCR TCAGGAAGACACATGACTTCTACTTC CAAATGGTCCTGCTGCATAC

3.3 PCR product purification

The PCR products were purified with the use of bead purification. Agencourt AMPure XP Beads (A63880, A63881, A63882, Beckman Coulter Nederland B.V., Woerden, The Netherlands) were used according to the manufacturer’s specifications with minor adaptions (Supplemental section I). In short, 40 μl PCR product was transferred to a new tube and 72 μl AMPure XP Beads were added. Samples were mixed by pipetting up and down and this was followed by incubating 5 minutes at room temperature. Next, the samples were placed into a magnetic stand for 5 minutes before proceeding washing the samples. The supernatant was removed and the tubes were removed from the magnetic stand. 500 μl 70% ethanol was added to the tubes and the samples were mixed by vortexing shortly. Samples were placed again into the magnetic stand for 5 minutes before proceeding with the same washing steps as described above. After washing, the samples were placed again in the magnetic stand for 5 minutes. Following these 5 minutes, the ethanol was removed and the beads were air-dried until they had an even color and a dry appearance. The tubes were removed from the magnetic stand and 50 μl MilliQ was added to the samples. Samples were incubated for 5 to 10 minutes at room temperate and then the tubes were placed again into the magnetic stand for 5 minutes. Finally, the supernatant with the purified PCR product was transferred to a new tube.

3.4 Sample preparation for NGS

Purified PCR products of fragment A, B and C of each sample had to be pooled and so the DNA concentration of the products were measured with the use of a Qubit fluorometer (Life Technologies Europe B.V., Bleiswijk, The Netherlands) with the Qubit dsDNA HS assay (Q32851, Q32854, Life Technologies Europe B.V., Bleiswijk, The Netherlands). The Qubit assay was performed according to the manufacturer’s specifications (Supplemental section I). Pooling of the samples was performed with the use of the following formula: y = 0.0363 x, in which y is the amount of μl that has to be added to the pool and x is the length of the PCR product in bp. For all samples the amount of μl was calculated and

(15)

11 the pools were prepared. Next, 100, 50 or 0 μl MilliQ was added to the samples. The amount of μl that was added was determined with the use of the DNA concentrations that the separate PCR products had before pooling. Following this dilution, the DNA concentrations were measured again with the Qubit fluorometer and the Qubit dsDNA HS assay, according to the manufacturer’s specifications (Supplemental section I). Prior to the use of the Nextera XT DNA Library preparation kit (FC-131-1024, FC-131-1096, Illumina, Eindhoven, The Netherlands) samples were diluted to 0,2 ng/μl. The Nextera-based preparation kit was performed according to the manufacturer’s specifications (Supplemental section I). Following this preparation, a quantification and size estimation of the library of 22 samples of the 145 was performed with a Bioanalyzer 2100 High Sensitivity DNA chip (5067-4626, Agilent Technologies Netherlands B.V., Amstelveen, The Netherlands) by experienced personnel according to the manufacturer’s specifications (Supplemental section I). Next, the samples were pooled per lane. Libraries were normalized to 2 nM and sequenced on the HiSeq 2000 System (2 x 100 paired-end reads) (Illumina, Eindhoven, The Netherlands) at our department by experienced personnel.

3.5 Mitochondrial genome assembly, annotation and data analysis

Sequence output data was analyzed by in-house bioinformatics. In short, multiplexed files were generated using Casava version 1.8.2. (Illumina, Eindhoven, The Netherlands). The files were aligned to the reference genome (Reference Sequence mtDNA Danio rerio: NC_002333.2 [32], NCBI) with the use of Burrows-Wheeler Aligner 0.5.9. [33]. A Python 2.6.6., Python package pysam 0.7.8 and SAMTools 0.1.19 software was used to read the BAM files. The number of each nucleotide per position was determined and the coverage, presence of a variant and percentage heteroplasmy was determined. To determine if de novo mutations were present in the mtDNA of the oocytes, a few filters were established. Cut-off values were used for heteroplasmy percentages and the coverage, Heteroplasmy percentages had to be >5% and the coverage had to be >2000. To determine if variants in the oocytes were de novo mutations, the variants were not allowed to be present in the brain, gonad, liver and/or muscle tissues of the female fish, they originated from, with percentages higher than 5%. To filter even further, only the variants of which the tissues (except the gonads) who did contain <1 percentage of heteroplasmy were selected for being called a de novo mutation. For these de novo mutations, in the other oocytes of the same female fish the percentages heteroplasmy for the same location as the de novo mutation were also determined. Presence of the same mutation nucleotide in all the other oocytes was used as criteria for the optional filter. In table 2 the filter steps and descriptions are shown.

Table 2: The filter steps that were performed with the NGS data are shown. A short description of the filter steps is described in the last column. The * indicates that this is an optional filter.

Filter steps Description

1. Initial set without filters 102 samples x 16,596 sites

2. Variants > 5% heteroplasmy 5% heteroplasmy is used as the diagnostics cut-off value

3. Variants > 2000 coverage Filtering for low coverage

4. Variants not present in four tissues of the female fish the oocyte originated from

Filtering for variants with heteroplasmy percentages > 5% present in the brain, gonad, liver and/or muscle of female fish

5. Variants not present in three tissues of the female fish the oocyte originated from

Filtering for variants with heteroplasmy percentages > 1% present in the brain, liver and/or muscle of female fish

6. De novo mutation not present

in the other oocytes of the same female fish*

Filtering for the presence of the mutation nucleotide as the second most prevalent nucleotide in all the other oocytes

(16)

12

3.6 Quantitative PCR

Mitochondrial copy number of the oocytes was determined using qPCR. During the qPCR a part of the ND1 gene was amplified. Extracted mtDNA was diluted 5 times prior to the qPCR. The qPCR mixture of 10 µl contained 1 µl extracted DNA, 5 µl qPCR SyBr Green Fluorescein Mix (QT615-20, Bioline, GC Biotech B.V., Alphen aan den Rijn, The Netherlands), 1.25 µM forward and reverse primer and MilliQ was added to create a volume of 10 µl. Primers were designed and ordered similar to the Long-Range PCR primers for fragment B and C. The qPCR program consists of the following steps: first 2 minutes at 50°C and 10 minutes at 95°C, then 40 cycles consisting of 15 seconds at 95°C and 1 minute at 60°C, next 15 seconds at 95°C and to obtain a melting curve an increase from 60°C to 95°C. The PCR was performed in a 7900HT Fast Real-Time PCR System (Life Technologies Europe B.V., Bleiswijk, The Netherlands). Per qPCR all samples were measured in triplicates and for all samples two separate qPCRs were performed. Standard curves were used to determine the mtDNA copy number and per qPCR two standard curves were included. Standard curves were generated using serial dilutions of the ND1 region, which was ligated into the pGEM-T Easy Vector (Made by I. Eijkenboom) (A1360, Addgene, Teddington, United Kingdom). Calculations of the dilutions are shown in Supplemental section V.

(17)

13

4. Results

4.1 Amplification of the mitochondrial genome with Long-Range PCR

Figure 8 shows the gel electrophorese results of the first of the four different PCRs. PCR results of the other PCRs are shown in Supplemental section II. It is shown that for all the samples fragments A, B and C could be amplified. Samples in which fragment A was amplified, contained fragments with a size of 6000 bp. Samples in which fragment B or C were amplified, contained fragments with a size of 5651 or 5722 bp, respectively. Negative controls (indicated with red bars) did not result in fragments. The amplification of fragment A performed well for all 98 samples. While, for 8 samples fragment B was not amplified and for 2 samples fragment C was not amplified. A second attempt to amplify fragment B of 2 samples could not be performed. In conclusion all fragments could be amplified for 96 samples.

Figure 8: Long-Range PCR results visualized by a gel electrophorese. Fragments A, B and C were amplified for the samples 2.1 to 2.14 and 3.1 to 3.14. Samples were loaded in an ascending order and the red bars indicate the negative controls. All samples contained a fragment of approximately 6000 bp for fragment A, 5651 bp for fragment B and 5722 bp for fragment C. All negative controls did not contain fragments of these sizes.

4.2 DNA concentration determination with Qubit assay

All amplified fragments of the samples could be purified and the presence of DNA was confirmed. However, the results of the assay showed that a few samples contain low concentrations of DNA. Especially, the samples with DNA fragment B had lower DNA concentration compared to the samples with fragment A and C. The average DNA concentration of the samples of fragments A and C were 67,8 ng/μl and 61,4 ng/μl, respectively. The average DNA concentration of the samples of fragment B was 40,4 ng/μl, with a lowest concentration of 1,75 ng/μl. However, the lowest concentration of the samples with fragment C was approximately the same, namely 2,7 ng/μl. The lowest concentration of the samples with fragment A was 24,5 ng/μl. The highest concentration of the samples with fragments A, B and C were almost the same, 110,0, 99,7 and 110,0 ng/μl, respectively. The individual DNA concentrations of the samples are shown in table 3.

(18)

14

Table 3: Individual and average concentrations of samples with fragments A, B and C measured with the use of a Qubit assay are shown. Average concentrations of samples with fragments A and C were comparable and the average concentration of samples with fragments B was lower in comparison to fragments A and C. Lowest concentrations of samples with fragments B and C were comparable, however the lowest concentration of samples with fragments A were higher than B and C. Highest concentrations of samples with fragments A, B and C were comparable.

A B C A B C A B C

Conc. Conc. Conc. Conc. Conc. Conc. Conc. Conc. Conc.

1.1 65.8 1.75 21.2 4.1 63.2 24.6 88.3 6.8 77.9 45.8 50.7 1.2 63.3 18.0 27.0 4.2 65.6 44.8 90.5 6.9 82.9 17.9 46.6 1.3 87.4 23.0 67.4 4.3 71.3 14.8 84.9 6.10 60.1 0 2.74 3.15 74.6 4.65 36.5 4.4 64.0 56.8 69.7 6.11 73.6 10.1 49.1 5.15 80.5 51.8 67.4 4.5 62.2 62.3 79.4 6.12 73.7 48.7 47.7 7.15 92.5 5.01 49.9 4.6 75.6 46.7 80.0 6.13 68.2 11.0 6.25 8.15 94.5 4.99 38.2 4.7 70.2 60.2 88.9 6.14 74.4 18.1 36.7 2.1 81.0 81.6 100.0 4.8 78.1 57.6 76.7 7.1 78.0 22.3 59.5 2.2 79.0 65.7 76.7 4.9 61.5 27.5 86.3 7.2 77.0 23.0 47.1 2.3 87.2 24.5 83.1 4.10 66.6 12.9 79.3 7.3 72.8 14.9 64.1 2.4 86.0 68.8 52.5 4.11 65.4 23.2 89.7 7.4 84.8 32.7 70.4 2.5 98.1 63.0 80.0 4.12 60.1 7.29 83.1 7.5 34.7 28.8 67.3 2.6 91.8 77.8 91.6 4.13 66.1 51.7 91.9 7.6 82.6 5.49 51.7 2.7 84.8 78.7 97.9 4.14 64.0 60.3 80.7 7.7 72.3 18.6 54.2 2.8 89.0 74.6 64.6 5.1 62.7 34.2 96.6 7.8 73.4 9.31 20.7 2.9 85.0 31.9 43.1 5.2 70.3 33.8 83.9 7.9 70.9 19.8 52.8 2.10 87.2 75.9 79.4 5.3 63.5 33.2 96.3 7.10 80.0 20.3 57.5 2.11 90.6 75.6 72.2 5.4 63.0 39.0 88.8 7.11 78.6 38.1 58.4 2.12 86.9 67.6 80.0 5.5 65.1 50.5 89.2 7.12 79.3 45.3 61.9 2.13 74.9 55.7 89.8 5.6 66.8 17.3 53.7 7.13 75.6 69.0 61.2 2.14 72.5 52.6 110.0 5.7 54.9 22.4 84.2 7.14 51.6 46.1 60.8 3.1 93.0 84.9 78.0 5.8 64.3 26.1 89.6 8.1 28.4 8.9 23.2 3.2 110.0 91.8 66.2 5.9 60.2 32.3 81.4 8.2 28.6 23.6 24.4 3.3 99.3 51.8 74.2 5.10 58.2 20.2 89.8 8.3 28.1 20.9 21.9 3.4 88.2 99.7 73.2 5.11 70.5 61.5 95.3 8.4 29.6 23.9 15.0 3.5 82.7 99.2 68.6 5.12 63.4 58.1 98.5 8.5 27.6 13.6 24.9 3.6 80.5 78.0 72.9 5.13 62.7 67.5 91.4 8.6 30.6 23.8 19.4 3.7 74.4 86.0 69.0 5.14 62.7 43.9 75.6 8.7 29.0 22.9 20.4 3.8 76.9 78.9 63.5 6.1 68.8 12.6 45.2 8.8 29.0 13.8 22.2 3.9 76.9 72.6 67.4 6.2 81.0 23.1 50.3 8.9 26.3 16.0 22.7 3.10 79.8 77.7 75.1 6.3 70.7 54.8 65.3 8.10 27.8 19.0 27.7 3.11 77.8 76.8 70.8 6.4 75.9 1.81 13.3 8.11 29.2 21.6 23.7 3.12 64.3 65.1 71.1 6.5 73.2 54.4 58.2 8.12 27.2 12.4 22.1 3.13 62.7 75.0 63.1 6.6 77.5 62.6 65.3 8.13 24.5 28.6 25.3 3.14 53.2 81.9 71.1 6.7 61.1 0 11.2 8.14 27.8 23.8 24.5

(19)

15

4.3 NGS library preparation check with Bioanalyzer assay

A total of 145 samples were included for preparation and a list of all these samples is shown in Supplemental section II. To confirm that the first part of the NGS preparation was successful a Bioanalyzer assay was performed with 22 of the 145 samples. The Bioanalyzer system is a system which is used for sizing, quantitation and quality control of DNA, RNA, proteins and cells on a single platform. [34] Figure 9 shows the results of one Bioanalyzer assay: the graph shows the measured fluorescence units (FU) at different DNA fragment lengths. If the preparation performed well then a graph will arise in which the measured FU will rise slowly from 150 bp to 500 bp. At 500 bp the increase will be steeper and at 1000 bp there will be a peak which is followed by a sharp decrease of the FU to 0. Sample 80/4.5 displays a perfect graph. If the preparation did not work perfectly no clear peak will be observed at 1000 bp and this can be seen in the graphs of sample 93/2.3 and 140/7.1. If a sample is too concentrated then the increase of the measured FU at 150 bp will be steeper and no clear peak will be observed. Sample 114/3.10 was too concentrated and this can be seen in both the electrophoresis result and the graph of this sample. All results of the two Bioanalyzer assays are shown in Supplemental section IV.

Figure 9: Library preparation check with Bioanalyzer assay. Left part of this figure shows a sort of electrophoresis result. Right part of this figure shows the graphs of 5 samples and also the ladder is shown in the graph in the bottom right. Sample 80/4.5 contains a perfect graph indicating that the preparation performed well. If the preparation did not perform well then no clear peak will be observed at 1000 bp and this can be seen in the graphs of sample 93/2.3 and 140/7.1. Sample 114/3.10 was too concentrated and this can be seen in both the electrophoresis result and the graph of this sample. In the graph it can be seen that the increase of the measured FU at 150 bp will be steeper in comparison to the increase at 150 bp in sample 80/4.5 and no clear peak at 1000 bp can be observed.

4.4 De novo mutation load determination with NGS data analysis

To obtain reliable results, more than 80% of the reads of the performed sequence run should contain a Q30 quality score. 81,6% of the reads of the performed sequence run contained a Q30 quality score. A quality score of Q30 means that 1 in 1000 times a base is called incorrectly. Consequently, base call accuracy will be 99,9% and also virtually all of the reads will be identified correctly. [35] Also, total error rate of the sequence run was 0,32%. To determine the de novo mutation load different filter steps were performed and these filter steps are shown in table 4. Initially, there were 1.692.792 sites and after filtering for variants with a heteroplasmy percentage of more than 5% and a coverage of at least 2000, respectively 7559 and 4484 sites were retained. An example of the difference between a sample with a high coverage and a low coverage is shown in figure 10. Following the first two filters, more filters were applied. However, this time variants were removed which were also present in the tissues of the female

(20)

16 fish of which the oocyte originated from. This filtering was performed in two steps; the first step was removing the variants which were also present in the female fish at the same location with a higher percentage than 5% heteroplasmy, retaining 894 variants. The second filtering step was selecting the variants of which the same locations in the tissues of the female fish contained heteroplasmy percentages of maximum 1%. Following this filter step only 29 variants remained which were called de novo mutations. For some de novo mutations the mutation nucleotide was also the second most prevalent nucleotide in all the other oocytes, however causing a heteroplasmy percentages of less than 5%. Lastly, an optional filter can also be used. With this optional filter the variants, which are not solely present in one oocyte but also in all the other oocytes of the same female fish, will be removed, retaining only 23 de novo mutations.

Table 4: The filter steps that were performed with the NGS data and the effect of these filter steps on the retained sites are shown. The * indicates that this is an optional filter. As can be seen in the row of the number of retained sites, with each filter step the number of sites decreases. The de novo mutation load varies between 29 and 23 mutations, depending on whether the optional filter is applied.

Filter step 1. 2. 3. 4. 5. 6.*

Retained sites 1692792 7559 4484 894 29 23

Figure 10: Figure 10: In the left graph the coverage of a sample with a high coverage is shown and the right graph shows a low coverage sample. In both graphs there are area’s with a low or a high coverage. In the sample with the low coverage, however, the coverage varies much more throughout the genome. The red color indicates the coverage that is obtained by the forward reads and the blue color indicates the coverage that is obtained with the reverse reads.

To determine the de novo mutation rate the following formula was used: amount of de novo mutations / (total amount of samples x length of the studied genome in bp). Total amount of de novo mutations in all the oocytes varied from 23 to 29 mutations. Thus, the de novo mutation rate range was determined by performing two calculations: 23 / (102 x 16596) = 1,36 x 10-5 mutations per site and 29 / (102 x 16596) = 1,71 x 10-5 mutations per site.

To determine if the distribution of the de novo mutations is equally divided among the fish, the amount of de novo mutations per female fish were determined for both the mutations determined with and without the optional filter step. Oocytes of female fish 4, 5 and 7 did contain maximally one de novo mutation with and without the optional filter. Oocytes of female fish 2, 3, 6 and 8 contained four to ten de novo mutations without applying the optional filter and three to seven de novo mutations with applying the optional filter. In table 5 this distribution is shown and also whether the de novo mutations were in a protein-coding region or in a non-protein coding region. Most of the de novo mutations are located in a non-protein-coding region.

(21)

17

Table 5: Total number of de novo mutations and the division of those mutations over the protein-coding and non-protein-coding regions are shown. Numbers between parentheses show the amount of mutations including the last optional filter. Amount of de novo mutations varies between the oocytes of the different female fish. Most of the de novo mutations are located in the non-protein-coding region.

Total number of de

novo mutations

Number of mutations in protein-coding regions

Number of mutations in non-protein- coding region

Oocytes fish 2 6 (5) 1 (1) 5 (4) Oocytes fish 3 4 (3) 1 (0) 3 (3) Oocytes fish 4 0 (0) 0 (0) 0 (0) Oocytes fish 5 1 (1) 1 (1) 0 (0) Oocytes fish 6 10 (7) 4 (1) 6 (6) Oocytes fish 7 1 (0) 1 (0) 0 (0) Oocytes fish 8 7 (7) 1 (1) 6 (6)

For all female fish maximal one oocyte contained a de novo mutation in a protein-coding region except for female 6. Oocytes of female fish 6 contained four de novo mutations in protein-coding regions. One female fish did not contain more than three oocytes with one or more de novo mutations. This can be seen in table 6 for female fish 6 but also in Supplemental section V for the other female fish. Oocyte 6.4 and 6.13 contain three de novo mutations and oocyte 6.10 contains four de novo mutations. However, two de novo mutations in oocyte 6.4 and one de novo mutations in oocyte 6.10 would be excluded by applying the optional filter step.

Table 6: Distribution of the de novo mutations in oocytes with de novo mutations of female fish 6. Reference positions with **, indicate that the de novo mutations would be excluded if the optional filter step was used. Also heteroplasmy percentages for the mutation locations of the four tissues of the female fish are shown. Oocyte 6.4 and 6.13 contain 3 de novo mutations and oocyte 6.10 contains 4 de novo mutations. However, 2 de novo mutations in oocyte 6.4 and 1 de novo mutations in oocyte 6.10 would be excluded with the optional filter step.

Reference position Brain Gonad Liver Muscle Oocyte 6.4 Oocyte 6.10 Oocyte 6.13 30 0,108 0,262 0,000 0,000 8,336 9,235 34 0,252 0,590 0,446 0,000 8,055 8,398 36 0,149 0,460 0,000 0,000 7,905 8,571 5941** 0,124 0,128 0,165 0,000 7,568 11183** 0,909 1,460 0,386 0,760 14,948 11184 0,846 0,470 0,496 0,865 6,753 11186** 0,561 1,281 0,311 0,557 5,013

During this study in sample 2.1, 1 de novo mutations was found, by applying a 5% cut-off value. For this sample it was determined that 529 de novo mutations would be found when a heteroplasmy cut-off value of 1% was applied. If heteroplasmy cut-off was increased to 2, 3 or 4% then 72, 4 or 1 de novo mutations would be found, respectively.

4.5 Copy number determination with qPCR analysis

Most reliable mtDNA copy numbers were determined by performing two qPCRs per sample. Average copy numbers of two qPCR runs were determined and figure 11 shows the distribution of mtDNA copy numbers of the oocytes. Precise mtDNA copy numbers of the oocytes are listed in Supplemental section VI also, four standard curves and the corresponding formulas are shown. Average copy number of the 98 oocytes is 3,43 x 107 copies per cell with a standard deviation of 7,32 x 106 copies per cell. The oocyte

(22)

18 with the highest mtDNA copy number contained 5,96 x 107 copies per cell and the oocyte with the lowest copy number contained 1,05 x 107. In the graph the distribution of the copy numbers seems to be Gaussian distributed. To determine if the mtDNA copy numbers were Gaussian distributed a d'Agostino and Pearson omnibus normality test was performed and a P value of 0,0914 was found.

MtDNA copy number in zebrafish oocytes

0 2.0_107

4.0107

6.0107

8.0_107

3,43 x 107 copies per cell

Av e rage : Standard de v iation Oocyte s m tD N A c o p y n u m b e r

Figure 11: Scatter plot of the mtDNA copy number of 98 zebrafish oocytes. Each oocyte is represented by one circle. Average copy number of those 98 oocytes is 3,43 x 107 copies per cell and the average is indicated by the dark grey horizontal line. Standard deviation is 7,32 x 106 and indicated by the light gray error bars. Copy number distribution is Gaussian distributed.

Average copy numbers and standard deviations were determined for oocytes of separate female fish. A scatter plot of the average copy numbers and standard deviations per female fish is shown in figure 12. Average copy numbers of different female fish are not equal. An Analysis Of Variance was performed to determine if there were significant differences between average copy numbers between the female fish. Also a Bonferroni’s Multiple Comparison Test was performed. Means of female fish 2 and 5, 3 and 5, 5 and 8 and 7 and 8 were significantly different.

MtDNA copy number in zebrafish oocytes

Female 2 Female 3 Female 4 Female 5 Female 6 Female 7 Female 8 0 2.0107 4.0107 6.0107 8.0107 m tDNA c o p y n u m b e r

Figure 12: The distribution of the copy numbers in the separate female fish. The average copy number of each female fish is indicated by the dark grey horizontal lines. The light grey error bars indicate the standard deviations. The average copy numbers of the oocytes of female fish 2, 3, 4 and 6 are roughly the same. The average copy numbers of the oocytes of female fish 7 and 5 are respectively slightly and much higher than the average copy numbers of female fish 2, 3, 4 and 6. The average copy number of the oocytes of female fish 8 is lower than the average copy numbers of female fish 2, 3, 4 and 6. Most of the copy numbers are centered on the average, however there are also some outliers.

(23)

19 To determine if the copy number has an effect on the de novo mutation rate a scatter plot was generated in which the group of oocytes with a de novo mutation could be compared with the group of oocytes without a de novo mutation. Average copy number of the oocytes with a de novo mutation was 3,33 x 107 copies per cell and the standard deviation was 1,01 x 107 copies per cell. Average copy number and standard deviation of the oocytes without a de novo mutation are respectively 3,44 x 107 copies per cell and 6,95 x 106 copies per cell. A t-test indicates that the mean copy number of oocytes with a de novo mutation and the mean copy number of oocytes without a de novo mutation are not significantly different.

MtDNA copy number in zebrafish oocytes with and without a de novo mutation.

0 2.0107 4.0107 6.0107 8.0107 Oocytes with a de novo mutation Oocytes without a de novo mutation

Oocytes with a de novo mutation Average:

Standard deviation:

Oocytes without a de novo mutation Average:

Standard deviation:

m tDNA c o p y n u m b e r

Figure 13: Scatter plot of the copy numbers of the oocytes with and without a de novo mutation. The dark grey horizontal line indicates the average copy numbers and the light grey error bars represent the standard deviations. The average copy number of the oocytes with a de novo mutation was 3,33 x 107 copies per cell and the standard deviation was 1,01 x 107 copies per cell. The average copy number and standard deviation of the oocytes without a de novo mutation are respectively 3,44 x 107 copies per cell and 6,95 x 106 copies per cell.

(24)

20

5. Discussion

The aim of this study was to determine the de novo mutation rate in oocytes of zebrafish and also whether the mitochondrial copy number has an effect on the de novo mutation rate.

In this study the performance of the mtDNA extraction could not be verified, because the mtDNA was dissolved in 10 μl and at least 6 μl of these 10 μl were needed for the Long-Range PCR. Also it was determined that the theoretical DNA concentration would be 0,02 ng/μl. This was determined by using the formula: m = n (1,096 x 10-21 g/bp), in which m is the mass of one DNA fragment and n is the length of DNA fragment in bp. [36] The length of one mtDNA molecule is approximately 16.600 bp and so the mass of one mtDNA molecule would be 16600 bp x (1,096 x 10-21 g/bp) = 1,8177 x 10-17 g. During a recent study on the mtDNA copy number in zebrafish oocytes a copy number of 12 x 106 copies was found. [2] Consequently, the total mass of the mtDNA in oocytes would be 1,8177 x 10-17 g/mt DNA molecule x 12 x 106 mtDNA molecules = 2,1812 x 10-10 g or 0,21812 ng. MtDNA was dissolved in 10 μl and so the DNA concentration would be 0,021812 ng/μl. The most sensitive DNA concentration measurement technique available at the department is the Qubit High Sensitivity assay and the detection range of this assay is 0,02 ng/μl to 100 ng/μl. DNA concentrations of our samples would be 0,021812 ng/μl and this would mean that the concentration of the samples would be at the lowest point of the detection range. If the extraction method would not be able to extract all the mtDNA than the DNA concentration would not be measurable. Consequently, only the Long-Range PCR would show whether the mtDNA extraction performed correctly. The Long-Range PCR performed well for all the samples and all the fragments, expect for fragment B of sample 6.7 and 6.10. Probably, these two oocytes of female fish 6 had lower quality and therefore also a poorer DNA quality. Consequently, amplification of fragments could be inhibited during the PCR. Following the first Long-Range PCR attempt only fragment A was amplified for these samples. However, both fragment B and C could not be re-amplified, since there was insufficient DNA left. Only 4 μl DNA was left and per Long-Range PCR reaction 2 μl DNA would be used. Also a qPCR to determine the mtDNA copy number of the samples would have to be performed with the DNA of these samples. In a previous study it was attempted to perform the Long-Range PCR with just 1 μl of the sample. However during this PCR, fragments could be amplified for just a few samples. [37] Thus, only one fragment could be amplified and because fragment C seemed easier to amplify this fragment was chosen for the second Long-Range PCR attempt. In retrospect this was a good choice, since fragment C was amplified during the second Long-Range PCR. This was also confirmed by the Qubit results. Average DNA concentration of the purified fragments C was higher than the average DNA concentration of the purified fragments B. Indicating that amplification of fragment B was less efficient than the amplification of fragment A and C. The DNA that was used in the three Long-Range PCRs was identical and also the amount of DNA in the PCR reaction was the same. Poor performance of the PCR could not be caused by an excessive length of the fragment that was amplified, since fragment B was the smallest fragment. GC contents and melting temperatures (Tm) of the three fragments were equal. GC contents of the forward and reverse primer for fragments B and C were equal with percentages of 40%. For fragment A GC contents of the forward and reverse primer were 60 and 55%. However, the Tm of the primers for all three fragments were equal. Consequently, no explanation could be found for the poor amplification of fragment B.

After the first part of the sample preparation for the NGS run two Bioanalyzer assays were performed as a means of quality control. Sampling was performed, since conducting a Bioanalyzer assay for all the 145 samples would be too labour intensive and expensive. The 22 samples that were selected were chosen, since they were likely to be representative for all the samples. The sample preparation was performed in thirteen series and per series one sample was chosen. This was done, since the samples of one series

De novo mutations in zebrafish mitochondrial DNA

De novo mutations in zebrafish

mitochondrial DNA

De novo mutations in zebrafish

mitochondrial DNA

:

Table of Contents

1. Introduction

2. Theoretical background

2.1 Mitochondrial DNA

2.2 Mitochondrial bottleneck

2.3 Research model

2.4 Next-Generation Sequencing

3. Materials and methods

3.1 Sample collection and mtDNA extraction

3.2 Long-Range PCR

3.3 PCR product purification

3.4 Sample preparation for NGS

3.5 Mitochondrial genome assembly, annotation and data analysis

3.6 Quantitative PCR

4. Results

4.1 Amplification of the mitochondrial genome with Long-Range PCR

4.2 DNA concentration determination with Qubit assay

4.3 NGS library preparation check with Bioanalyzer assay

4.4 De novo mutation load determination with NGS data analysis

4.5 Copy number determination with qPCR analysis

5. Discussion