Thesis
Reference
Genetic variability and epigenetic alterations in Down syndrome with congenital heart defects
SAILANI, Mohammad Reza
Abstract
This study aims to identify the genetic and epigenetic determinants of congenital heart defect (CHD) in Down syndrome (DS). We hypothesized that trisomy 21 in concert with additional genetic and epigenetic variations contribute to the risk of CHD in DS. Association studies were performed in samples from DS with and without CHD to identify genome-wide SNPs and chromosome 21 specific CNVs associated with CHD. The results revealed two SNPs and two CNVs on chromosome 21 that may modify the CHD risk in DS. Furthermore, a monozygotic twin approach was used to elucidate the role of trisomy 21 in DNA methylation modifications.
Reduced representation bisulphite sequencing (RRBS) revealed differentially methylated regions in promoters of genes involved in embryonic organ morphogenesis relevant to DS phenotypes including CHD. In summary, the results of this work support a multifactorial model for development of CHD in DS that includes trisomy 21, SNPs, CNVs, and DNA methylation modifications.
SAILANI, Mohammad Reza. Genetic variability and epigenetic alterations in Down syndrome with congenital heart defects. Thèse de doctorat : Univ. Genève, 2013, no. Sc.
4629
URN : urn:nbn:ch:unige-332506
DOI : 10.13097/archive-ouverte/unige:33250
Available at:
http://archive-ouverte.unige.ch/unige:33250
Disclaimer: layout of this document may differ from the published version.
1 / 1
UNIVERSITÉ DE GENÈVE
Département de Génétique et Evolution FACULTÉ DES SCIENCES
Professeur Alicia Sanchez- Mazas Département de FACULTÉ DE MÉDECINE
Médecine Génétique et Développement Professeur Stylianos E. Antonarakis
Genetic variability and epigenetic alterations in Down syndrome with congenital heart defects
THÈSE
Présentée à la Faculté des sciences de l’Université de Genève pour obtenir le grade de Docteur ès sciences, mention biologie
par
Mohammad Reza Sailani
de Kerman, Iran Thèse n° 4629
Université de Genève 2013
NOTE
Part of the results presented in this thesis have been published or are under review in peer-reviewed journals and referenced in the text. Copies of these papers are in the appendix section of this thesis.
1-Letourneau A, Santoni F, Bonilla X, Sailani M.R, Gonzalez D, Kind J, Chevalier C, Thurman R, Sandstromn R.S, Hibaoui Y, Falconnet E, Gagnebin M, Gehrig C, Vannier A, Guipponi M.L, Farinelli L, Robyr D, Garieri M, Migliavacca E, Borel C, Deutsch S, Feki A, Stamatoyannopoulos J, Herault Y, Steensel B, Guigo R, Antonarakis S.E. Discordant monozygotic twins reveal chromosomal domains of gene expression dysregulation in Down syndrome. 2013 (Under review).
2-Hibaoui Y, Grad I, Letourneau A, Sailani M.R, Dahoun S, Santoni F, Gimelli S, Guipponi M, Pelte F, Bena F, Antonarakis S.E, Feki A. Modelling and rescuing neurodevelopmental defect of Down syndrome using induced pluripotent stem cells from monozygotic twins discordant for trisomy 21. EMBO Molecular Medicine 2013 (In press).
3-Sailani MR, Makrythanasis P, Valsesia A, Santoni FA, Deutsch S, Popadin K, Borel C, Migliavacca E, Sharp AJ, Duriaux Sail G, Falconnet E, Rabionet K, Serra-Juhé C, Vicari S, Laux D, Grattau Y, Dembour G, Megarbane A, Touraine R, Stora S, Kitsiou S, Fryssira H, Chatzisevastou-Loukidou C, Kanavakis E, Merla G, Bonnet D, Pérez-Jurado LA, Estivill X, Delabar JM, Antonarakis SE. The complex SNP and CNV genetic architecture of the increased risk of congenital heart defects in Down syndrome. Genome Research. 2013 Sep; 23(9):1410-21.
I TABLE OF CONTENTS:
LIST OF FIGURES ... VI
LIST OF TABLES ... VIII
LIST OF ABBREVIATIONS ... IX
1 ACKNOWLEDGMENTS ... 1
2 ABSTRACT ... 4
3 RÉSUMÉ EN FRANCAIS ... 8
4 INTRODUCTION ... 12
4.1 Down syndrome historical background ... 13
4.2 Down syndrome associated phenotypes ... 14
4.3 Down syndrome associated congenital heart defect (CHD) ... 16
4.3.1 Atrioventricular septal defect (AVSD)... 17
4.3.2 Ventricular septal defect (VSD) ... 17
4.3.3 Atrial septal defect (ASD)... 18
4.4 Mouse models of Down syndrome with CHD ... 19
4.5 Molecular mechanisms of the heart development ... 22
4.6 Definition of chromosome 21 minimal critical region for CHD in Down syndrome ... 27
4.7 Molecular basis of CHD in Down syndrome and non-Down syndrome cases ... 33
II
4.8 Copy number variation and CHD ... 36
4.9 DNA methylation and Down syndrome ... 40
5 HYPOTHESIS ... 46
6 OBJECTIVES ... 49
7 MATERIAL AND METHODS ... 51
7.1 Down syndrome CHD and SNP association study ... 54
7.1.1 Discovery cohort collection and genotyping ... 54
7.1.1.1 Chromosome 21 SNPs calling ... 57
7.1.1.2 Association and interaction analyses ... 59
7.1.1.3 Population stratification ... 60
7.1.2 Replication cohort collection and genotyping ... 62
7.2 Down syndrome chromosome 21 CNVs analyses ... 65
7.2.1 Discovery cohort collection and chromosome 21 CNV calling ... 65
7.2.1.1 Chromosome 21 array CGH design ... 65
7.2.1.2 Chromosome 21 CNV analyses ... 67
7.2.2 Replication cohort and chromosome 21 CNV analyses by NanoString ... 69
7.2.2.1 NanoString codeset design ... 69
7.2.2.2 NonoString data analyses ... 71
7.3 Euploid population and chromosome 21 CNVs ... 73
7.4 DNA methylation analyses of Down syndrome cases... 75
7.4.1 Monozygotic twin’s samples description ... 75
7.4.2 Reduced Representation Bisulfite Sequencing (RRBS) ... 79
7.4.2.1 RRBS methodology... 79
III
7.4.2.2 RRBS benefits and limitations ... 83
7.4.2.3 RRBS library preparation protocol ... 85
7.4.3 DNA methylation data processing ... 88
8 RESULTS ... 92
8.1 Genome-wide SNP association study of CHD in Down syndrome ... 93
8.1.1 Goal: ... 94
8.1.2 Rational: ... 94
8.1.3 Data: ... 94
8.1.3.1 Population stratification ... 95
8.1.3.2 Non-chromosome 21 (disomic) SNPs association study ... 97
8.1.3.3 Chromosome 21 (trisomic) SNPs association study ... 101
8.1.3.4 Candidate SNP risk allele validation in replication cohort ... 106
8.1.4 Interpretation of the results ... 107
8.2 Two-locus (SNP-SNP) interaction study ... 108
8.2.1 Goal: ... 108
8.2.2 Rational: ... 108
8.2.3 Data: ... 109
8.2.3.1 Interactions among SNPs on diploid fraction of the genome ... 109
8.2.3.2 Interaction study and replication cohort ... 113
8.2.3.3 Interactions amongst SNPs on the trisomic chromosome 21 ... 113
8.2.3.4 Interactions between SNPs on chr21 and SNPs on the diploid fraction of the genome ... 114
8.2.4 Interpretation of the results: ... 114
8.3 Chromosome 21 CNV association study of AVSD in Down syndrome ... 116
8.3.1 Goal: ... 116
8.3.2 Rational: ... 116
8.3.3 Data: ... 116
IV
8.3.3.1 Chromosome 21 CNV association results ... 117
8.3.3.2 Chromosome 21 AVSD risk CNVs in the general population ... 126
8.3.3.3 Chromosome 21 AVSD risk CNVs in replication cohort ... 129
8.3.4 Interpretation of the results: ... 131
8.4 DNA methylome of Down syndrome ... 133
8.4.1 Goal: ... 133
8.4.2 Rational: ... 133
8.4.3 Data: ... 133
8.4.3.1 Samples and sequencing data ... 137
8.4.3.2 Data analysis ... 144
8.4.3.2.1 DNA methylation level at single CpG sites and genomic regions: ... 144
8.4.3.2.2 Identifying differentially methylated regions (DMRs) ... 159
8.4.4 Interpretation of the results: ... 174
8.4.4.1 Monozygotic twins discordant for trisomy 21 ... 174
8.4.4.2 Trisomy 21 monozygotic twins discordant for VSD ... 175
9 DISCUSSION: ... 177
9.1 Genome wide SNP association study of CHD in Down syndrome ... 179
9.2 Genome wide SNP-SNP interaction study ... 182
9.3 CNV association study of AVSD in Down syndrome ... 184
9.4 Down syndrome and DNA methylation ... 186
9.5 Challenges and considerations... 190
9.6 Future directions ... 193
10 CONCLUSIONS ... 195
V
11 REFERENCES ... 198
12 APPENDIX ... 228
VI LIST OF FIGURES
Figure 1. Down syndrome mouse models and heart defects. ... 20
Figure 2. Representation of human heart development. ... 26
Figure 3. Definition of the CHD-critical region on chromosome 21. ... 27
Figure 4. Down syndrome CHD candidate region defined by Barlow et al (2001). ... 30
Figure 5. Down syndrome genotype–phenotype mapping by Lyle et al (2009). ... 31
Figure 6. Down syndrome CHD candidate region defined by Korbel et al (2009). ... 32
Figure 7. Models for CNVs in modification of the phenotypic expression of trisomies. ... 39
Figure 8. Representation of our hypothesis regarding CHD risk in Down syndrome. ... 47
Figure 9. Representation of methods and samples used in this study. ... 52
Figure 10. Example of the chromosome 21 SNP calling for a trisomy 21 individual. ... 58
Figure 11. Distribution map of designed CNV probes over chromosome 21. ... 66
Figure 12. Box-plots of all 135,679 CNV probes log2 ratios per sample. ... 68
Figure 13. Reduced representation bisulfite sequencing library preparation steps. ... 82
Figure 14. Workflow of DNA methylation data processing. ... 89
Figure 15. Principle component analyses (PCA) for evaluating population stratification. ... 96
Figure 16. Genome-wide Manhattan plots for CHD in DS and its different sub-phenotypes. ... 98
Figure 17. Q-Q plots of the SNP association test statistic. ... 100
Figure 18. Chr21-wide Manhattan plot of SNP genotypic association test P-values. ... 104
Figure 19. Chr21-wide Q-Q plot of SNP genotypic association test P-values. ... 105
Figure 20. Q-Q plot of whole-genome cis-eQTLs interaction study P-values. ... 111
Figure 21. Chromosome 21-wide Manhattan plot of CNV association test P-values. ... 119
Figure 22. Overview of CNV1 region (Chr21: 42,066,443-442,071,313). ... 121
Figure 23. CNV1 and illumine SNP probes in normal and Down syndrome individuals. ... 122
Figure 24. Overview of CNV2 region (Chr21:42,284,480-42,286,300). ... 123
Figure 25. Overview of CNV3 region (Chr21:45,541,600-45,555,054). ... 125
Figure 26. Frequency of detected DS-AVSD risk CNVs in the general population. ... 127
VII
Figure 27. Gene expression and copy number states for CNV1, CNV2, and CNV3. ... 128
Figure 28. Box-plot of base quality alongside the reads. ... 138
Figure 29. Representation of the genomic features for the DNA methylation analyses. ... 144
Figure 30. An example of cytosine base methylation calling based on read coverage. ... 146
Figure 31. Histogram of DNA methylation level in MZ-twins discordant for trisomy 21. ... 147
Figure 32. Box-plots of overall DNA methylation level. ... 148
Figure 33. Box-plots of overall DNA methylation level over genomic features. ... 150
Figure 34. Average distribution of DNA methylation profile mapped onto a gene model. ... 152
Figure 35. DNA methylation fold change in trisomy 21 twin versus normal twin... 153
Figure 36. DNA methylation fold change based on gene model. ... 154
Figure 37. Average distribution of DNA methylation profile mapped onto a gene model. ... 155
Figure 38. Distribution of DNA methylation profile mapped onto LADs and iLADs. ... 158
Figure 39. Strategy used to define DMRs induced by the extra copy of chr21. ... 161
Figure 40. RPKM values for protocadherin gamma gene cluster (PCDHG). ... 173
VIII LIST OF TABLES
Table 1. Frequency of CHD phenotypes (modified from Freeman et al. 2008). ... 16
Table 2. Genes, pathways, and loci that have been associated with CHD. ... 36
Table 3. Distribution of phenotypic attributes of the discovery cohort. ... 55
Table 4. Distribution of phenotypic attributes of the replication cohort. ... 63
Table 5. List of primers used in PCR amplification and Pyrosequencing... 64
Table 6. NanoString CNV prob coordinates. ... 72
Table 7. Primers used for qPCR experiment. ... 74
Table 8. Description of samples used for DNA methylation analyses. ... 77
Table 9. Top non-chromosome 21 SNP association results for DS-CHD and its sub-phenotypes. ... 99
Table 10. Significant chromosome 21 trisomic SNPs association test results for DS-CHD and DS -ASD. ... 102
Table 11. Chromosome 21 candidate CHD risk SNPs in the replication cohort. ... 106
Table 12. P-values of top three pairs of interacting genome-wide 8,126 cis-eQTLs for DS-CHD. ... 111
Table 13. P-values of top pairs of potentially interacting SNPs from the 431,962 whole genome SNPs. ... 112
Table 14. Candidate chromosome 21 CNV regions for increased risk of AVSD in Down syndrome. ... 120
Table 15. NanoString CNV probe association test results in replication study ... 130
Table 16. Samples used for DNA methylation analyses and sequencing information. ... 139
Table 17. Number of CpGs covered by different read coverage cut-offs. ... 141
Table 18. Re-analyzed data from Jin et al, 2013... 143
Table 19. Number of DMRs identified in each pair of samples. ... 162
Table 20. GO analyses of final 49 candidate DMRs in MZ-twins discordant for T21. ... 165
Table 21. Absolute DNA methylation differences and gene expression fold changes. ... 166
Table 22. Gene expression fold change of HOXB and HOXD gene clusters . ... 166
Table 23. Gene expression fold change of DNA methylation and de-methylation enzymes ... 167
Table 24. GO analysis of DMRs associated with trisomy 21 MZ-twins discordant for VSD. ... 170
Table 25. DMRs in the genes associated with cell adhesion processes. ... 171 Table 26. Gene expression of the PCDHG cluster in cardiac cells with and without functional heart defect. 172
IX
LIST OF ABBREVIATIONS
aCGH array Comparative Genomic Hybridization
ALL Acute Lymphoblastic Leukemia
AMKL Acute Megakaryoblastic Leukemia
ASD Atrial Septal Defects
AVCD Atrio Ventricular Canal Defect
AVSD Atrio Ventricular Septal Defect
BPS Bronchopulmonary Sequestration
BMP Bone Morphogenetic Protein
CEU Utah residents from northern and western Europe
CHB Han Chinese in Beijing
CHD Congenital Heart Defect
eQTL expression Quantitative Trait Loci
CCAM Congenital Cystic Adenomatoid Malformation
CNV Copy Number Variant CGIs CpG Islands
DGV Database of Genomic Variant
X
DMR Differentially Methylated Region
DNMT DNA methyltransferase
DS Down syndrome
ECD Endocardial Cushion Defect
FDR False Discovery Rate
FGF Fibroblast Growth Factor
GSEA Gene Set Enrichment Analysis
GFF General Feature Format
GADA Genome Alteration Detection Analysis
GAII Genome Analyzer II
GWAS Genome Wide Association Studies GWIS Genome Wide Interaction Studies
GC Genomic Control
JPT Japanese in Tokyo
HWE Hardy–Weinberg equilibrium
HET Heterozygousity
HSA21 Homo sapiens chromosome21
XI IBS Identity By State
LAD Lamina Associated Domain
LD Linkage Disequilibrium
MAF Minor Allele Frequency
Mb Megabase
MDS Multidimensional Scaling
MEF2 Myocyte Enhancer Factor-2
MZ Mono Zygotic
OR Odd Ratio
PCA Principal Components Analysis
PCDHG Protocadherin Gamma Gene
qPCR quantitative Polymerase Chain Reaction
Q-Q Plot Quantile-Quantile plot
RRBS Reduced Representation Bisulphite Sequencing
RPKM Reads Per Kilobase Per Million Mapped Reads
SLIM Sliding Linear Model
SNP Single Nucleotide Polymorphism
XII T21 Trisomy 21
TOF Tetralogy Of Fallot
TSS Transcription Start Site
TTS Transcription Termination Site
TGFβ Transforming Growth Factor Beta
VSD Ventricular Septal Defect
YRI Yoruba in Ibadan
1 Acknowledgments
Foremost, I would like to sincerely thank my supervisor Prof. Stylianos E. Antonarakis, the director of the Department of Genetic Medicine and Development at the University of Geneva, for the excellent support and supervision during my four-year Ph.D study. I am grateful for the freedom he offered me in the managing of my thesis project. I would also like to thank him for the excellent opportunities he offered me to present my project in the international meetings. I appreciate his patience, generosity, enthusiasm, motivation, original ideas, and immense knowledge. It has been a privilege to be one of his students, and I could not have imagined having a better advisor for my Ph.D studies.
Moreover, I sincerely appreciate the time he spent reviewing my career goals and providing me with the letter of support for my postdoc applications.
Furthermore, I would like to thank the NCCR-Frontiers in Genetics Ph.D program for providing me with the great opportunity to pursue my Ph.D studies. This graduate program gave me the chance to do three laboratory rotations in Switzerland. I would like to thank Prof. Antonarakis, Prof. Gasser, and Prof. Trono for offering me the opportunity to perform my lab rotations in their laboratories. I wish also to thank Prof.
Duboule, the director of the NCCR Frontiers in Genetics program, and Dr. Suarez for administrative coordination.
Moreover, I wish to thank several people for their significant contribution in shaping my scientific interests and knowledge. First, I would like to sincerely acknowledge my mentor Dr. Andy Sharp during my lab rotation for the excellent mentoring and supervision. I very much enjoyed working with him. I am also thankful to the former
2
Postdoc of the lab, Dr. Samuel Deutsch, for his contribution and comments on my Ph.D project at the beginning. I am immensely grateful to Dr. Periklis Makrythanasis for the continued support of my project all the times. I wish also to thank Dr. Federico Santoni for his excellent support, encouragement and advice for my PhD thesis. I am grateful also to Dr. Armand Valsesia who generously contributed to this project with the CNV analysis.
My sincere thanks also go to Dr. Christelle Borel, Dr. Michel Guipponi, Dr. Kostya Popadin, Dr. Sergey Nikolaev, Dr. Daniel Robyr, Dr. Jean-Louis Blouin, and Dr. Eugenia Migliavacca for their support during my PhD studies. It was a privilege to work with them closely. I am also grateful to my fellows Audrey Louterneau, Ximena Bonilla, George Stamoulis, Marco Garieri, Marco Peligri and Muhammad Ansar who always motivated me. They offered me the best out of our friendships.
I am grateful also to the past and present technicians in the lab, Corinne Gehrig, Catherine Ucla, Maryline Gagnegin, Genevieve Duriaux Sail, Emilie Falconnet, Yann Dupre, Anne Vannier and Pascale Ribaux for their excellent technical skills that helped me during the performance of my experiments.
I am also thankful to Dr. Frédérique Béna, Dr. Stefania Gimelli, Dr. Sophie Dahoun, and Ms. Elisavet Stathaki from the cytogenetics lab for thier participation in this project.
I also thank Prof. Manolis Dermitzakis, from whom I learnt a lot about population genetics and evolutionary biology. I would like to point out Prof. Marguerite Neerman- Arbez and her lab with whom we shared our weekly lab meetings and from whom I acquired a lot of knowledge regarding fibrinogen gene regulation mechanisms. I am also thankful to Dr. Richard Fish for his advice and comments on my project.
3
Also, I would like to thank the administrative and secretarial team of the department. I do appreciate the kindness of Francine Chopard, Sarita Goutorbe, Dominique Sutter and Kim Nguyen for their enormous help and support at any time that I needed it.
I wish to thank the whole lab and the entire department for teaching me almost everything that I know today about human genetics and genomics.
I would also like to thank the genomics platform, primarily Dr. Patrick Descombes, Dr.
Mylene Docquier, Dr. Didier Chollet, Christelle Barraclough, Isabelle Durussel, and Brice Petit for their constant support, and always providing a welcoming atmosphere.
Of course, this project would have not been possible without the participation of our collaborators. I greatly appreciate their contribution. I also gratefully acknowledge the individuals and families for their participation in this project.
I wish to thank the Vital-IT of the SIB Swiss Institute of Bioinformatics for the excellent service for high-performance computing.
I would like to thank Prof. Dean Nizetic and Prof. Alicia Sanchez-Mazas for taking the time out of their busy schedule to be the referees of this thesis.
Finally, my special thanks go to my parents in Iran for all the support, motivation and love they have given me. I do appreciate their kindness.
4
2 Abstract
5
Congenital heart defects (CHDs) are common developmental phenotypes of Down syndrome (DS), occurring in 40-60 % of cases. While carrying three copies of genes or other functional genomic elements on chromosome 21 increase the risk for CHD, trisomy 21 itself is not sufficient to cause CHD. Thus, additional genetic/epigenetic variations on the chromosome 21 and/or elsewhere in the genome could contribute to the risk of CHD. In this thesis, genome wide association studies (GWAS) were used to identify genomic variations that, in concert with the trisomy 21, determine the risk of CHD in Down syndrome. This case-control GWAS includes 187 Down syndrome individuals with CHD as cases, and 151 Down syndrome individuals without CHD as controls. A chromosome 21-specific association study revealed rs2832616 and rs1943950 (both cis-eQTLs for KRTAP7-1) as CHD risk alleles (adjusted genotypic P- values < 0.05). These SNPs were confirmed in a replication cohort of 92 Down syndrome individuals with CHD and 80 Down syndrome individuals without CHD (nominal P-value
= 0.0022). Since Down syndrome is likely to be a disorder of gene expression, a two- locus interaction study was applied for whole genome cis-eQTLs that are functionally linked to gene expression variations. A pair of interacting cis-eQTLs involving CNOT11 on the chromosome 2 and NRGN on the chromosome 11 were identified (adjusted P- value < 0.05). Furthermore, a search for chromosome 21 risk CNVs for AVSD (Atrioventricular septal defects) using a customized chromosome 21 aCGH of 135K probes in 55 Down syndrome AVSD cases and 53 Down syndrome without CHD controls revealed two risk CNV regions (FDR ≤ 0.05). Both CNVs are located in the previously identified CHD region on chromosome 21. The first CNV is located upstream of RIPK4, and the second CNV is located within the ZNF295. These risk CNVs were further confirmed in a replication study of 49 Down syndrome AVSD cases and 45 Down
6
syndrome without CHD controls (FDR ≤ 0.05). Furthermore, to elucidate the role of the extra copy of chromosome 21 in genome wide DNA methylation modifications in Down syndrome pathogenesis including CHD, a monozygotic twin approach was applied.
Reduced representation bisulphite sequencing (RRBS) revealed 183 differentially methylated regions (DMRs) (FDR < 0.001) between a rare pair of monozygotic twins discordant for trisomy 21, of which 49 DMRs were also confirmed in unrelated normal and trisomy 21 individuals. The identified DMRs are enriched for genes involved in embryonic organ morphogenesis (FDR = 0.0051) and include genes such as HOXB5, HOXB6, HOXD3, and HOXD11. The DMRs in HOXD and HOXB clusters are maintained in iPS cells generated from this pair of monozygotic twins discordant for trisomy 21 and are strongly correlated with the gene expression changes of HOXD and HOXB clusters (R
= -0.71). The results also showed an increase in DNA methylation level in the Down syndrome methylome compared to the normal euploid methylome. This increase in the level of DNA methylation is more pronounced in promoters and CpG islands of Down syndrome cases compared to normal euploid controls. This observation is concordant with the up-regulation of DNA methyltransferase enzymes (DNMT3B and DNMT3L) and down-regulation of DNA de-methylation enzymes (TET2 and TET3) in trisomy 21 twin versus normal twin cells. Moreover, a pair of rare trisomy 21 monozygotic twins discordant for VSD (Ventricular septal defect) revealed 111 DMRs in the genes that are enriched for cell-cell adhesion process (FDR = 8.92e-08) including the protocadherin gamma gene cluster. In summary, the results of this work support a multifactorial model for development of CHD in Down syndrome that includes trisomy 21, SNPs, CNVs, and DNA methylation modifications.
7
8
3 RÉSUMÉ EN FRANCAIS
9
La cardiopathie congénitale est un phénotype développemental commun du syndrome de Down qui touche 40-60% des cas. Même si être porteur de trois copies des gènes et autres éléments fonctionnels du chromosome 21 augmente le risque de cardiopathie congénitale, la trisomie seule n’est pas suffisante pour causer la cardiopathie. D’autres variations génétiques ou épigénétiques sur le chromosome 21 et/ou sur le reste du génome pourraient donc contribuer au risque d’être atteint de cardiopathie congénitale.
Dans ce travail de thèse, des études d’association pangénomiques ont été utilisées dans le but d’identifier des variations génomiques qui, de concert avec la trisomie 21, déterminent un risque pour la cardiopathie congénitale chez les patients atteints du syndrome de Down. Cette étude cas-témoins inclut 187 individus atteints du syndrome de Down ainsi que de la cardiopathie congénitale, comme cas et 151 personnes atteintes du syndrome de Down sans cardiopathie comme contrôles. Une étude d’association spécifique du chromosome 21 a révélé deux allèles, rs2832616 et rs1943950 (deux cis- eQTLS de KRTAP7-1) comme allèles à risque (valeurs P génotypiques ajustées < 0.05).
Ces polymorphismes nucléotidiques, ou plus communément SNPs (pour Single- Nucleotide Polymorphisms) ont été confirmés dans une cohorte de réplication de 92 individues atteints du de syndrome de Down atteints de cardiopathie congénitale ainsi que 80 patients contrôles atteints du syndrome de Down mais sans cardiopathie (valeur P nominale = 0.0022). Comme le syndrome de Down est probablement un trouble lié à l’expression génique, une étude d’interaction de deux loci a été utilisée pour identifier des cis-eQTLs qui sont liés fonctionnellement dans le génome entier. Une paire de cis- eQTLs interagissant l’un avec l’autre a été identifiée. Elle implique CNOT11 sur le chromosome 2 et NRGN sur le chromosome 11 (valeur P ajustée <0.05). D’autre part, l’étude de la variabilité du nombre de copies (CNV pour Copy Number Variant en
10
anglais) de 21 segments à risque pour le canal atrio-ventriculaire (CAV) a été réalisée à l’aide d’une puce d’hybridation génomique comparative personnalisée du chromosome 21 de 135'000 de sondes. Cette étude, portant sur 55 cas de trisomiques 21 atteints de CAV et 53 trisomiques ne présentant pas cette malformation, a révélé deux régions CNV à risque (FDR ≤ 0.05). Les deux CNVs sont situés dans la région du chromosome 21, précédemment identifiée comme étant à risque pour la cardiopathie congénitale. Le premier CNV est situé en amont de RIPK4 et le second est situé à l’intérieur de ZNF295.
Ces CNVs à risque ont été ensuite confirmés dans une étude de réplication de 49 patients atteints du syndrome de Down et du CAV et 45 individus trisomiques sans malformation cardiaque. (FDR ≤ 0.05). Enfin, dans le but d’étudier le rôle de la copie supplémentaire du chromosome 21 sur les modifications de méthylation de l’ADN du génome entier dans la pathogenèse du syndrome de Down incluant la cardiopathie congénitale, une approche faisant recours à des jumeaux monozygotiques a été utilisée.
La méthode de « reduced representation bisulfite sequencing (RRBS) » a révélé 183 régions méthylées de manière différentielle (FDR < 0.001) entre un cas rare de jumeaux monozygotes discordants pour la trisomie 21. 49 de ces régions différentiellement méthylées ont par la suite aussi été confirmées chez deux individus sans lien de parenté, l’un présentant un caryotype normal, l’autre trisomique 21. Ces régions identifiées de par leur méthylation, sont enrichies en gènes impliqués dans la morphogenèse d’organes embryonnaires (FDR = 0.0051) et incluant des gènes tels que HOXB5, HOXB6, HOXD3 et HOXD11. Les régions différentiellement méthylées dans les clusters HOXD et HOXB sont maintenues dans les cellules souches embryonnaires pluripotentes induites générées à partir de ces jumeaux discordants pour la trisomie 21 et sont fortement corrélées avec le changement d’expression génique des clusters HOXD et HOXB (R = -
11
0.72). Les résultats montrent aussi une augmentation de méthylation dans le méthylome des cas de trisomie 21 comparé à un méthylome normal euploïde. Cette augmentation de niveau de méthylation est plus prononcée aux promoteurs et aux îles de CpG. Cette observation concorde avec la régulation positive de l’enzyme ADN methyltransférase (DNMT3B et DNMT3L) et la régulation négative d’enzymes de déméthylation (TET2 et TET3) dans les cellules du jumeau trisomique 21 comparativement aux cellules du jumeau normal. Finalement, l’étude de jumeaux discordants pour la communication inter-ventriculaire (CIV) a révélé 111 régions méthylées différentiellement comprenant des gènes enrichis dans des processus d’adhésion entre cellules (FDR = 8.92e-08), incluant le cluster de gènes de la protocadherine gamma. En résumé, les résultats de ce travail soutiennent un modèle multifactoriel pour le développement de la cardiopathie congénitale chez les patients souffrant du syndrome de Down. Ce modèle inclut la trisomie 21, des SNPs, des CNVs et des modifications de la méthylation de l’ADN.
12
4 Introduction
13
4.1 Down syndrome historical background
Genomic aneuploidy, the deletions or duplications of chromosomal or sub-chromosomal regions, is a common cause of human genetic disorders leading often to dysregulation of gene expression patterns. A classic example of genomic aneuploidy is trisomy 21, resulting from the presence of supernumerary copies of a chromosome 21. Trisomy 21 gives rise to a collection of phenotypes defined as Down syndrome which initially was described by Èdouard Sèguin in 1846 (Neri and Opitz 2009) and later on in 1866 these phenotypes were further characterized by John Langdon Down (Down 1995). However, the genetic etiology of Down syndrome was not clear for about 100 years till Lejeune and Gautier in 1959 discovered that Down syndrome was indeed a chromosomal anomaly resulted from the presence of three copies of the chromosome 21 (Lejeune et al. 1959). Forty years later, the significant advances in exploring the human genome leaded to the determination of the complete DNA sequence of human chromosome 21 (HSA21) (Hattori et al. 2000). This achievement provided the opportunity to find the molecular mechanisms underlying the phenotypic manifestation of trisomy 21, and the functional links between the genomic and phenotypic variability. Nevertheless, there are still massive gaps in our understanding of the mechanisms underlying the complex spectrum of phenotypes in Down syndrome. Recent advances in genomic technologies and next-generation sequencing platforms in conjunction with advances in transgenic and trans-chromosomal mouse models will further help to have a better understanding of the molecular pathophysiology of Down syndrome.
14
4.2 Down syndrome associated phenotypes
Down syndrome, which refers to a collection of phenotypes resulted from an extra copy of chromosome 21, occurs generally in 1 out of 750 live births. Although survival in trisomy 21 can range from death in utero to late adulthood, the rate is higher at the conception stage. It is shown that about 75% of trisomy 21 fetuses during the first and 50% during the second trimester are lost before term (Spencer 2001; Roper and Reeves 2006). One of the characteristics of Down syndrome phenotypes, which make the discovery of the underlying gene(s) difficult, is the variable expressivity and reduced penetrance of the causative genes. There are over 30 phenotypic features in Down syndrome that vary in frequency amongst the cases (Nizetic 2001). Some of the phenotypes (e.g. cognitive impairment) are consistently present in all of the cases of Down syndrome, albeit in variable severity, while others show incomplete penetrance and are highly variable in incidence and severity from patient to patient (Antonarakis et al. 2002; Antonarakis et al. 2004; Antonarakis and Epstein 2006). Some of the common phenotypes of trisomy 21 are characteristic facial dysmorphology, a small and hypocellular brain, immune system defects, and Alzheimer-like dementia, which appear by the fourth decade of life (Korenberg et al. 1994; Spencer 2001; Roper and Reeves 2006). Interestingly, the phenotypes with incomplete penetrance are present in a higher rate in trisomy 21 population than in the euploid population. This observation highlights the point that perhaps chromosome 21 harbors the genes that directly or indirectly are involved in the pathogenesis of theses phenotypes, and that the dysregulation of these genes is the underlying cause of the symptoms associated with Down syndrome. For instance, Down syndrome patients are frequently affected by leukemia (Kivivuori et al.
1996; Nikolaev et al. 2013; Yoshida et al. 2013). Down syndrome newborns diagnosed
15
for transient preleukemic state are about 10% (Massey 2005), 20-30% of which progress to acute megakaryoblastic leukemia (AMKL). This event is very rare in the general population. In addition to AMLK, Down syndrome cases are also at higher risk for acute lymphoblastic leukemia (ALL). Collectively, infants with Down syndrome are 500 times at higher risk of developing AMKL and 20 times at higher risk of developing ALL than euploid infants (Seewald et al. 2012). However, Down syndrome cases have a significant lower incidence of most solid tumours (Nizetic and Groet 2012) and a markedly decreased risk of atherosclerosis (Lott and Head 2005). Another example of Down syndrome phenotype with the reduced penetrance is congenital defects of the gastrointestinal tract which is present in 7% of Down syndrome cases (Freeman et al.
2009). Similar variability occurs for many other phenotypes of Down syndrome, however, the most notable phenotype with reduced penetrance and expressivity in Down syndrome are congenital heart defects (CHD). CHD is one of the most common birth defect, affecting 0.8% of live births in the general population (Harris 2011).
Interestingly, CHDs occur in approximately 40-60% of Down syndrome individuals (Ferencz et al. 1989; Torfs and Christianson 1998; Freeman et al. 2008). This represents almost a 50-fold higher risk for the Down syndrome cases to be affected with CHD compared to the general population. The most frequent forms of CHD in Down syndrome cases are atrioventricular septal defects (AVSD) also known as atrioventricular canal defect (AVCD) and endocardial cushion defect (ECD) (MIM 606215), comprising 43% of the CHD cases in Down syndrome individuals. While ventricular septal defects (VSD), atrial septal defects (ASD), and tetralogy of fallot (TOF) and others comprise 32%, 19%, and 6% of the CHD cases in Down syndrome respectively. The AVSD cases are very rare in the general population (3-5 per 10,000 live
16
births) which makes it of high interest to investigate in the cases of Down syndrome, as the trisomy 21 individuals account for approximately 70% of the AVSD cases. This represents approximately 2000-fold increase in the risk of AVSD in the context of Down syndrome compared to the general population (Freeman et al. 1998; Freeman et al.
2008).
4.3 Down syndrome associated congenital heart defect (CHD)
Congenital heart defect (CHD) refers to abnormalities in the heart’s structure or function that arise before birth (Bruneau 2008). Heart abnormalities in Down syndrome include mainly atrioventricular septal defect (AVSD), ventricular septal defect (VSD), isolated secundum atrial septal defect (ASD), and few percentages of isolated tetralogy of fallot (TOF) and others. Table 1 shows the frequencies of CHD phenotypes in Down syndrome reported in different studies (Freeman et al. 1998, Stoll et al. 1998, Freeman et al. 2008, and Sailani et al. 2013).
Table 1. Frequency of CHD phenotypes (modified from Freeman et al. 2008).
Freeman et al. 1998 Stoll et al. 1998 Freeman et al. 2008 Sailani et al. 2013
Sample's origin USA France USA Europeb
Collection period 1989-1999 1979-1996 2000-2004 Since 2008
Number of DS samples 423 398 1469 510
% with CHD 41% 46% 44% 54%
AVSDa 47% 43% 39% 48%
ASDa 37% NA 42% 21%
VSDa 44% 32% 43% 30%
Othersa 7% 3% 6% 1%
NA, not available.
aAmong those with any CHD. Some individuals may have more than one type of CHD.
b Europe includes samples from France, Italy, Greece and Spain.
In the following sections each of these heart developmental phenotypes are discussed.
17 4.3.1 Atrioventricular septal defect (AVSD)
The term AVSD describes a spectrum of congenital heart defects defined by a common atrioventricular junction co-existing with abnormal or insufficient atrioventricular septation (Craig 2006). AVSD is the most severe form of CHD that necessitates surgical operation for its correction. There are three main structural defects seen in a typical AVSD heart. First is the absence of the atrial septum that separates the left and right atria. Second is the absence of the ventricular septum that divides the lower chamber of the heart into the left and right ventricles. Third is the complete AVSD which is due to malformation of the tricuspid and mitral valves (Craig 2006). It is noteworthy that AVSD cases in general are genetically associated with three different genetic patterns. The first is association with Down syndrome, the second is association with an autosomal dominant trait, and the third is isolated AVSD cases. AVSD cases with autosomal dominant inheritance are not linked to chromosome 21 (Craig 2006). However, the most cases of AVSDs (approximately 76%) are in the context of Down syndrome (Loffredo et al. 2001).
4.3.2 Ventricular septal defect (VSD)
VSD is the second common cardiac defect in Down syndrome (Marino et al. 1990) after AVSD. VSDs are defined as openings in the ventricular septum, the wall that separates the left and right ventricles of the heart. The VSD classification is based on the location of these openings (Minette and Sahn 2006). The ventricular septum can be further sub- divided into two morphological compartments, the membranous septum and the muscular septum. The magnitude of pathogenesis of VSDs is determined by the size of the opening and the pressure in the right and left ventricular chambers (Minette and
18
Sahn 2006). VSDs are found in 30 to 60% of all infants with a congenital heart defect, or about 2 to 6 cases per 1000 births (Meberg et al. 1994). VSDs in the context of Down syndrome also constitute about 35% of the CHDs in Down syndrome (Freeman et al.
1998)
4.3.3 Atrial septal defect (ASD)
ASD is a type of CHD that enables blood flow between two compartments of the heart called the left and right atria. In a healthy heart, the right and the left atria are well divided by the inter-atrial septum. In the absence or defect of this septum, the oxygen- rich blood and the oxygen-poor blood in the two atria can be mixed (Webb and Gatzoulis 2006; John et al. 2011). There are three major types of ASDs or inter-atrial communications as discussed by Webb and Gatzoulis (Webb and Gatzoulis 2006). These ASDs types are ostium secundum, ostium primum, and sinus venosus defects. The ostium secundum is a true defect of the atrial septum and involves the region of the fossa ovalis and is the main type of ASD observed in Down syndrome cases. In fact, ASDs are the most common cardiac abnormalities of Holt-Oram syndrome, which is caused by mutations in TBX5 (Basson et al. 1999).
19
4.4 Mouse models of Down syndrome with CHD
In order to understand the impact of increased gene expression of trisomic genes on the developmental phenotypes of Down syndrome cases, animal models serve as a critical tool. In particular, genetically engineered mice have been shown to be useful in exploring the developmental phenotypes (genotype-phenotype correlation) of Down syndrome. There are several mouse models for partial or complete trisomy syntenic to human chromosome 21 (HAS21) that demonstrate some of the characteristics of Down syndrome (Sago et al. 1998; Shinohara et al. 2001; Dunlevy et al. 2010; Yu et al. 2010;
Zhang et al. 2012) (Figure 1). HSA21 is syntenically conserved with three different regions in mouse chromosome 10 (Mmu10), chromosome 16 (Mmu16) and chromosome (Mmu17) (Figure 1). The majority of the conserved genes present in Mmu16. The centromere-proximal region of HSA21 through ZNF295 contains approximately 240 genes and is orthologous with the telomeric region of Mmu16 (Gardiner et al. 2003). The next segment (about 1.5 Mb) of HSA21 contains approximately 20 genes and is orthologous with the centromere-proximal region of Mmu17. The 1.5 Mb of telomeric region of HSA21 contains approximately 70 genes, making it an ortholog of an internal segment of Mmu10. Initially, the first link between HSA21 and Mmu16 was pointed out in 1979 and subsequently the mice that carried an extra copy of Mmu 16 (Ts16) were identified as a potential model for research on Down syndrome (Patterson and Costa 2005). Ts16 embryos demonstrate some of the anatomical features of Down syndrome, but they die before birth. Therefore their behavior cannot be studied (Davisson et al. 1993; Patterson and Costa 2005). It is of note that Ts16 embryos have a 100% incidence of AVSDs (Webb et al. 1996). A closer genetic match of the Down syndrome is generated by mouse models that are trisomic for
20
only a part of Mmu 16. Such a mouse model is Ts65Dn developed by Dr. Davisson's research group at the Jackson Laboratory.
Figure 1. Down syndrome mouse models and heart defects.
Mouse models: 1, Tc1; 2, Dp(10)1Yey/+;Dp(16)1Yey/+;Dp(17)1Yey/+; 3, p(16)1Yey/+;
4, Dp(17)1Yey/+; 5, Dp(10)1Yey/+; 6, Ts1Yah; 7, Ts65Dn; 8, Ts1Cje; 9, Dp(16)2Yey/+;
10, Ts1Rhr. –, data not available. Source: “Zhang L, et al. Bioeng Bugs. 2012; 3(1):8-12.”
21
Ts65Dn has been extensively used to study the molecular mechanisms responsible for the Down syndrome features (Davisson et al. 1993). This mouse model is trisomic for 13.4 Mb of the 22.9 Mb chromosome 21 syntenic regions (Moore 2006; Williams et al.
2008) and have many of the behavioral characteristics common to individuals with Down syndrome. Also Williams and colleagues characterized the cardiac phenotype in neonatal Ts65Dn mice (Williams et al. 2008). The histological analyses of Ts65Dn showed the presence of inter-ventricular septal defects. In addition, the immune- histochemistry of trisomic neonates confirmed abnormal muscle composition in the cardiac valves. These results indicated that the genetic imbalance present in Ts65Dn disrupts vital pathways during cardiac development (Williams et al. 2008). However, the frequency of CHD in Ts65Dn mice is less than that in humans. O'Doherty and colleagues in 2005 reported a trans-chromosomal Down syndrome mouse model carrying almost the complete HSA21 (O'Doherty et al. 2005). This trans-chromosomic mouse line manifests a number of developmental problems analogous to those in Down syndrome individuals, including defects in heart development (O'Doherty et al. 2005; Dunlevy et al.
2010). Also CHDs have been observed in the full “trisomy 21” homologous mice Dp(10)1Yey/+; Dp(16)1Yey/+; Dp(17)1Yey/+ mice (Yu et al. 2010). In addition, engineering the duplication of a 5.43 Mb region of Mmu16 from TIAM1 to KCNJ6, Dp(16)2Yey, was reported to cause CHD (Liu et al. 2011). In summary, studies of mouse models of trisomy 21 collectively support a complex genetic architecture model for development of CHD in Down syndrome, and shows that the dosage imbalance is one of the major contributors to CHD.
22
4.5 Molecular mechanisms of the heart development
Developmental heart defects are the most common of all human birth defects and affect about one percent of the population (Srivastava 2006a; Bruneau 2008). A crucial principle in heart development is that the regulation of different heart’s cell lineages should be precisely regulated and controlled, such that the correct heart’s cell lineage differentiates at the correct time and in the right location (Bruneau 2008). The biology of heart growth and development is a large field of research. The shaping of the heart tube, looping, septation and the resultant systemic and pulmonary circulations are complex, but precisely well regulated processes. Any perturbation in these developmental processes during heart morphogenesis leads to a wide spectrum of congenital heart defects (Srivastava and Olson 2000; Srivastava 2006b; Srivastava 2006a). Therefore, a comprehensive understanding of the molecular mechanisms of normal heart development is necessary for any attempt to unravel genetic factors underlying developmental heart defects. Rapid progress in molecular biology techniques has helped a great deal in learning about the molecular basis of the morphogenesis and development of the heart (Srivastava 2006b; Xu and Baldini 2007; Bruneau 2008; von Gise and Pu 2012). Briefly, the cells that will ultimately form the earliest cardiac progenitors originate in the lateral plate mesoderm shortly after gastrulation controlled by a cascade of interacting transcription factors (Srivastava 2006b). Additional factors are secreted molecules such as fibroblast growth factors (FGF), bone morphogenetic proteins (BMP), Wnt proteins, etc (Bruneau 2008). Epithelial to mesenchymal transition (EMT) converts epithelial cells to developmentally plastic and mobile mesenchymal cells. These cells will differentiate into myocardial, endocardial, and smooth muscle cells to form the developed heart (von Gise and Pu 2012). Two fundamental factors in the
23
EMT process are the down-regulation of intercellular adhesion junctions and the expression of EMT-inducing transcription factors (Thiery and Sleeman 2006). CER1, which encodes a cytokine from the anterior endoderm, signals the future migrating cells toward the midline via the homeodomain transcription factor NKX2-5 (Lints et al. 1993;
Harvey 1996). NKX2-5 expression subsequently activates GATA transcription factors that are a family of transcription factors characterized by their ability to bind to the myocyte enhancer factor-2 (MEF2) family transcription factors (Lints et al. 1993; Harvey 1996). It has been reported that MEF2 proteins are recruited to target the promoter of genes by the cell-specific GATA transcription factors, and thereby up-regulate the transcriptional activity of this family of heart specific genes (Morin et al. 2000). As cell migration happens, the expression of N-cadherin which is cell context-dependent (Radice 2013) contributes to the differentiation of the mesoderm into two distinct cell types, an N-cadherin expressing epithelial layer cells and a group of cells that do not express N-cadherin. The former will become the myocardium and the latter will become the endocardium (Linask et al. 1997). The endocardium subsequently covers the inside of the heart and creates the valves that separate the atria and ventricles. Precise positioning and function of the valves is vital for chamber constitution and normal blood flow. The cardiac tubes finally join into a single tube, and the two endocardial masses join into a single endocardial mass, called the endocardial cushion. Subsequently by the fusion of the dorsal and ventral endocardial cushion, atrioventricular septum will be generated. The atrioventricular septum in turn will divide the atrial canal into a right and left side. Subsequently, the tricuspid valve will be generated from the right side of the atrial canal and the bicuspid will be generated from the left side of the atrial canal.
The next developmental step is to convert the anterior (ventricular) and posterior
24
(atrial) structure of the linear heart tube into a primitive heart that contains two atrial and two ventricular chambers organized in a left to right configuration (Wagner and Siddiqui 2007). This happens at five weeks of gestation and is mediated by NODAL and LEFTY-2 factors (Harvey 2002; Wagner and Siddiqui 2007). The Nodal family of proteins is a subset of the transforming growth factor beta (TGFβ) super-family. Nodal is responsible for mesoendoderm induction and determination of dorsal- ventral axis in vertebrate embryos (Harvey 2002). Nodal signaling induces activation of genes that encode the Nodal inhibitor proteins Lefty1 and Lefty2 (Hamada et al. 2002). Nodal and Lefty factors mediate looping of the heart tube through HAND1 and HAND2 transcription factors. Interestingly, HAND1 is expressed in the future left ventricle, and HAND2 in the right ventricle. Nodal also regulates the left side specific expression of PITX-2 and XIN (Biben and Harvey 1997; Wagner and Siddiqui 2007). PITX-2 (Paired- like homeodomain transcription factor 2) is another transcription factor that has a crucial task in directing cardiac asymmetric morphogenesis. It has been shown that the expression of the myocardial Pitx2 regulates left atrial identity and ventricular asymmetric remodeling programs (Tessari et al. 2008). It is a target gene for the Nodal factors and is restricted from leaking to the right side of the midline barrier by the Lefty1 factor (Schlueter and Brand 2007). After the development of looping heart, at approximately seven weeks of gestation, another sequence of transcription factors signal the differentiation between the upper and lower chambers of the heart. For instance, the atrioventricular canal will separate the atrial and ventricular chambers (Moorman et al. 2003). The myocardium, the muscle layer of the heart responsible for the heart's pumping action, grows ventrally from the roof of the heart into the atrium, and dorsally from the base into the ventricular area, to generate the muscular portion of
25
the atrial and ventricular septa. This process is the start of generating four distinct chambers of the heart (Bao et al. 1999; Anderson et al. 2003). However, the muscularization of the septa is not enough to build up the walls separating the right and left chambers of the heart. Completion of the walls is one of the primary functions of the endocardium. Endocardium is the deepest layer of tissue that lines the chambers of the heart (Combs and Yutzey 2009). In addition to completing the structural atrial and ventricular septa, this mass of cells performs two other essential functions. First, it creates an endothelial lining on the inside of the heart to be connected with the blood vessels, and second it generates the valve leaflets that separate the upper and lower chambers, allowing directed transport of blood into and out of the heart (Srivastava and Olson 2000). Figure 2 demonstrates a graphical representation of heart development from the undifferentiated crescent on the left to the complete four-chambered heart on the right (from Srivastava, D 2006).
26
Figure 2. Representation of human heart development.
Panel A) shows frontal views of the cardiac precursors during human cardiac development. The primary or first heart field (FHF) cells form a crescent in the anterior embryo and second heart field (SHF) cells form medial and anterior to the FHF. In the panel B) cells lie dorsal to the straight heart tube and start to move (arrows) into the anterior and posterior extreme ends of the tube to generate the conotruncus (CT), the right ventricle (RV), and part of the atria (A). In the panel C) cardiac neural crest (CNC) cells move (arrow) into the outflow tract from the neural folds to septate the outflow tract and to pattern the bilaterally symmetric aortic arch arteries (III, IV, and VI). In the panel D) the four-chambered heart will be generated by the septation of the ventricles, atria, and atrioventricular valves (AVV). LV, left ventricle; V, ventricle; LA, left atrium;
RA, right atrium; Ao, aorta; PA, pulmonary artery; AS, aortic sac; RSCA, right subclavian artery; LSCA, left subclavian artery; RCA, right carotid artery; LCA, left carotid artery;
DA, ductus arteriosus. Source: "Srivastava D. Cell, 2006. 126(6): 1037-1048."
A) B) C) D)
27
4.6 Definition of chromosome 21 minimal critical region for CHD in Down syndrome
One potential approach to identify regions on chromosome 21 associated with CHD in Down syndrome is to use partial trisomy 21 cases who present CHD phenotype. This approach is based on identifying the minimal overlapping region across all partial trisomy 21 cases with CHD. Figure 3 illustrates a schematic representation of this approach.
Partial T21 cases with CHD
Chr 21
Minimum overlap region
Figure 3. Definition of the CHD-critical region on chromosome 21.
The minimal overlapping region
28
Recent studies have attempted to build up a genotype-phenotype correlation map of Down syndrome associated phenotypes by using partial trisomy 21 cases (Barlow GM et al 2001; Lyle et al. 2009; Korbel et al. 2009). In the first study that was published in 2001, 19 partial trisomy 21 cases were analyzed by using quantitative Southern blot dosage analysis in combination with fluorescence in situ hybridization (FISH) using subsets of 32 BAC probes, spanning the D21S16 (21q11.2) region through the telomeric region of chromosome 21 (Barlow GM et al 2001). Results of this study showed that the CHD candidate region was duplicated in 14 cases, of which 57% (eight cases) have some of the CHD characteristics (Barlow GM et al 2001). As illustrated in figure 4, the candidate region responsible for CHD is highlighted by D21S3 marker through PFKL marker. In the second study, Lyle et al. 2009 performed this approach on a panel of 19 partial trisomy 21 cases of European origin who had some of the characteristics of Down syndrome, including CHD in two of the cases (Lyle et al. 2009). In this study, by using a BAC array covering HSA21q and aCGH, a genotype-phenotype correlation map of different Down syndrome phenotypes was obtained. In this study, in 19 cases with partial trisomy 21, the size of the duplicated region range from 5.98Mb to 28.56Mb. The authors reported the identification and mapping of 30 pathogenic chromosomal aberrations of HSA21. They also identified the genimic map of the CHD region in a large segment between 31.5Mb and qter on chromosome 21. This region overlaps the region reported in the previous study (Figure 5). However, the identified region in these studies is considered too large to be informative as it contains many genes. In the third study, Korbel and colleagues used FISH, quantitative southern blot dosage analysis and high-density oligonucleotide tiling array to investigate a panel of 30 partial trisomy 21 cases, 14 of whom affected by CHD. They also took advantage of available data on
29
trisomy 21 in mouse, and combined results defined the minimal critical region for CHD as 1.77Mb region between DSCAM and ZNF295 (Korbel et al. 2009). This candidate region shown in figure 6 falls within the previously reported regions, though much smaller. In summary the results of these analyses suggest that only parts of the chromosome 21 are responsible for CHDs associated with Down syndrome. However, the main limitation of this approach is that partial trisomy 21 cases are very rare in the population. A panel of many partial trisomy 21 individuals with CHD is needed to achieve a precise critical region. Although this approach defined the minimal critical region on chromosome 21 for CHD phenotype, the defined region requires further analysis as it is large and contains many candidate genes.
30
Figure 4. Down syndrome CHD candidate region defined by Barlow et al (2001).
Minimal critical CHD candidate region as identified from studies of the cases with partial trisomy 21. Solid lines define regions of the known duplication from partial trisomy 21 cases. The candidate region is defined by a line with arrow heads at both ends. The extent of the region in kb indicated at the left of this line. The locations of the known genes mapping within the candidate CHD region are indicated by black bars to the left of the gene symbols. Source: " Barlow GM, et al. Genet Med, 2001: 3(2): 91-101."
31
Figure 5. Down syndrome genotype–phenotype mapping by Lyle et al (2009).
Each panel represents one aspect of the Down syndrome phenotype. The CHD panel (Cardiac anomaly) is highlighted in red color and shows duplication from 31.5Mb to qter. The X-axis represents the position along HSA21q in Mb; Y-axis is the phenotype score for each BAC array, with the maximal region shown in bold. Source: "Lyle R, et al.
Eur J Hum Genet, 2009: 17(4): 454-466."
32
Figure 6. Down syndrome CHD candidate region defined by Korbel et al (2009).
A) Purple boxes indicate presence of phenotype, yellow boxes indicate no phenotype, and solid boxes indicate increased copy-number. The red box shows Down syndrome CHD candidate region. 23 subjects have duplications encompassing the Down syndrome CHD region, 14 of whom show CHD associated phenotypes. In the left, it shows corresponding regions for Down syndrome in 6 mouse models. B) The proposed candidate region for CHD phenotype defined by including human and mouse data from panel A. The region which is 1.77 Mb is highlighted by red box. Source: " Korbel JO, et al.
Proc Natl Acad Sci, 2009: 106(29): 12031-12036."
33
4.7 Molecular basis of CHD in Down syndrome and non-Down syndrome cases
The variable expressivity and reduced penetrance of CHD associated phenotypes in general, and in particular in Down syndrome cases, have made the identification of CHD- associated genes very challenging. In addition, it has been shown that the genetic background has an impact on the frequency CHD among Down syndrome individuals (Freeman et al, 2008). Among the possible causes, the genetic and/or epigenetic background of each individual may largely contribute to this phenotypic variability.
Identification of the gene(s) responsible for AVSD and other heart defects associated with trisomy 21 remains a challenge. Studies of partial trisomy 21 cases as mentioned before identified the Down syndrome minimal CHD candidate region, but the region is too large to pinpoint the underlying genes. Recently a candidate-gene approach among individuals with Down syndrome AVSD and Down syndrome without CHD suggests a potential contribution of VEGF-A (Ackerman et al. 2012), ciliome, hedgehog (Ripoll et al.
2012) and folate (Locke et al. 2010) pathways to the pathogenesis of CHD in Down syndrome. In addition, altered expression of mitochondrial genes was reported in the heart of human fetuses with Down syndrome (Conti et al 2007). It is also suggested that up-regulation of HSA21 genes in combination with down regulation of nuclear encoded mitochondrial genes might contribute to the risk of CHD in Down syndrome (Piccoli et al 2013).
Candidate non-chromosome 21 genes have also been identified for susceptibility to several CHDs and AVSDs not associated with Down syndrome. For example, pathogenic mutations in the CREDL1 gene (on 3p25) have been found in ~6% of individuals with