A Systems-Genetics Analyses of Complex Phenotypes

Home , CENPL, FBXO4, IL1RAPL2, INTS8, KCTD8, MGST3

A systems-genetics analyses of complex phenotypes

A thesis submitted to the University of Manchester for the degree of

Doctor of Philosophy in the Faculty of Life Sciences

2015

David Ashbrook

Table of contents

Table of contents ...... 1 Tables and figures ...... 10 General abstract ...... 14 Declaration ...... 15 Copyright statement ...... 15 Acknowledgements...... 16 Chapter 1: General introduction ...... 17 1.1 Overview...... 18 1.2 Linkage, association and gene annotations ...... 20 1.3 ‘Big data’ and ‘omics’ ...... 22 1.4 Systems-genetics ...... 24 1.5 Recombinant inbred (RI) lines and the BXD ...... 25 Figure 1.1: Derivation of the BXD set...... 27 1.6 Indirect genetic effects (IGEs) ...... 28 Figure 1.2: Indirect genetic effects influence phenotypes in interacting individuals...... 31 1.7 Parent-of-origin effects (POEs), epigenetics and genomic imprinting ...... 32 Table 1.1: Diseases linked to abnormal genomic imprinting, with their associated features and causative mutations...... 36 1.8 Parental care ...... 37 1.9 Aims of the PhD ...... 39 1.10 Alternative format ...... 40 1.11 References ...... 41 Chapter 2: Joint genetic analysis of hippocampal size in mouse and human identifies a novel gene linked to neurodegenerative disease ...... 56 Chapter 2: Preface...... 57 Abstract ...... 59 2.1 Introduction ...... 60 2.2 Materials and methods ...... 63 2.2.1 Data ...... 63 2.2.2 Identification of significant genes for hippocampus size in mouse and human ...... 64 2.2.3 Expression quantitative trait loci ...... 66 2.2.4 Functional analysis ...... 67 2.2.5 Co-expresion and ‘Guilt-by-association’ ...... 67 2.3 Results ...... 70

Table of contents

2.3.1 Identification of genes significant in both species ...... 70 Figure 2.1: Quantile-quantile plot of human homologues of significant mouse genes for hippocampus size...... 71 2.3.2 Regulation of gene expression ...... 72 Table 2.1: Pearson’s correlations between probes for Mgst3 in adult mouse hippocampus...... 73 2.3.3 Functional analysis of significant genes ...... 73 2.3.4 ‘Guilt-by-association’ ...... 74 2.4 Discussion ...... 76 2.4.1 Conclusion ...... 77 2.5 References ...... 78 Chapter 3: A cross-species genetic analysis identifies candidate genes for mouse anxiety and human bipolar disorder...... 84 Chapter 3: Preface...... 85 Abstract ...... 87 3.1 Introduction ...... 88 Figure 3.1: Bipolar disorder traits and analogous mouse phenotypes...... 90 3.2 Materials and methods ...... 92 3.2.1 Mouse and human data ...... 92 3.2.2 Evolutionary conservation ...... 94 3.2.3 Identification of areas of expression ...... 94 3.2.4 ‘Guilt-by-association’ ...... 94 3.2.5 Principal component analysis ...... 96 3.2.6 Biological networks ...... 97 3.3 Results ...... 98 Figure 3.2: Graphical representation of the research method used, providing a summary of the main findings...... 98 3.3.1 QTL for bipolar related phenotypes in mice ...... 99 Table 3.1: Activity traits in the elevated zero maze which have a significant QTL ...... 101 Figure 3.3: QTL maps showing two QTL within each of the regions we identified ...... 102 Table 3.2: Vertical activity traits in the open field which have a significant QTL...... 104 3.3.2 Gene expression in the brain ...... 106 3.3.3 ‘Guilt-by-association’ ...... 106 3.3.4 Co-expression analysis ...... 107

Table of contents

3.3.5 Correlation analysis ...... 109 3.4 Discussion ...... 111 3.4.1 Biological function of candidate genes ...... 113 3.4.2 RXRG ...... 113 3.4.3 MCTP1 ...... 114 3.4.4 TNR...... 114 3.4.5 Role of striatum ...... 115 3.4.6 Conclusion ...... 116 3.5 References ...... 115 Chapter 4: Genetic variation in offspring indirectly influences the quality of maternal behaviour in mice ...... 129 Chapter 4: Preface...... 130 Abstract ...... 132 4.1 Introduction ...... 133 Figure 4.1: Experimental cross-foster design ...... 134 4.2. Materials and methods ...... 136 4.2.1 Experimental animals ...... 136 4.2.2 Husbandry and mating protocol ...... 136 4.2.3 Phenotyping protocols ...... 137 Table 4.1: Maternal and offspring behaviours recorded...... 138 4.2.4. Genetic analysis ...... 138 4.3. Results ...... 140 Figure 4.2: Offspring indirect genetic effect modifying maternal nestbuidling behaviour (OspIge7.1)...... 140 Figure 4.3 Offspring indirect genetic effect modifying maternal care (OspIge5.1) ...... 141 Figure 4.4: Maternal indirect genetic effect modifying offspring growth (MatIge17)...... 142 Table 4.2: A summary of the QTL identified ...... 143 Table 4.3: Tests of between-subjects effects of litter size, B6 foster mothers’ maternal care and day on BXD offspring solicitation, without phenotypes being divided by the litter size ...... 144 Table 4.4: Tests of between-subjects effects of litter size, B6 foster mothers’ maternal care and day on BXD offspring solicitation...... 145 Figure 4.5: Correlation between BXD mothers’ nestbuilding on day 6 and expression of Hsd17b7 in adrenal gland ...... 147 4.3.1 Co-adaptation of parental and offspring traits ...... 148

Table of contents

Figure 4.6: Correlation between offspring and maternal traits in biological BXD families ...... 150 Figure 4.7: Correlation between offspring solicitation and corresponding bodyweight in BXD lines on day 10 and day 14, respectively ...... 150 4.4 Discussion ...... 151 4.4.1 Caveats and future work ...... 151 4.4.2 Conclusion ...... 152 4.5 References ...... 154 Chapter 5: Indirect genetic effects influence sibling and maternal behaviour in mice...... 159 Chapter 5: Preface...... 160 Abstract ...... 162 5.1 Introduction ...... 163 5.2 Materials and methods ...... 167 5.2.1 Experimental animals ...... 167 5.2.2 Husbandry and mating protocol ...... 167 5.2.3 Cross fostering and data recording ...... 168 5.2.3 Behavioural observations ...... 169 Table 5.1: Maternal and offspring behaviours recorded...... 170 5.2.4 QTL mapping ...... 170 5.2.5 Identification of candidate genes ...... 171 5.3 Results ...... 173 Figure 5.1 Experimental half-litter cross-fostering design ...... 173 5.3.1 Day 14 activity – SocInt2.1 ...... 174 Table 5.2: Summary of correlations between B6 and BXD pup phenotypes which show a QTL on chromosome 2 ...... 175 Figure 5.2: QTL map for traits which have a QTL on chromosome 2 ...... 176 5.3.2 Day 6 maternal care and offspring feeding – SocInt4.1 ...... 176 Table 5.3: Summary of correlations between B6 mother, and B6 and BXD offspring phenotypes which show a QTL on chromosome 4, and traits which correlate significantly with them ...... 177 Figure 5.3: QTL map for traits which have a suggestive QTL on chromosome 4, and traits which correlate significantly with them ...... 178 5.3.3 Day 14 maternal care and offspring feeding – SocInt15.1 ...... 178 Table 5.4: Summary of correlations between B6 and BXD offspring phenotypes which show a QTL on chromosome 15, and those which correlate significantly with them ...... 179 Figure 5.4: QTL maps for correlated traits as shown in table 5.4 ...... 180

Table of contents

Table 5.5: A summary of the traits within each of the three loci identified .. 181 5.4 Discussion ...... 184 5.4.1 SocInt2.1 ...... 185 5.4.2 SocInt4.1 ...... 186 5.4.3 SocInt15.1 ...... 186 5.4.4 Implications and future directions ...... 186 5.4.5 Conclusion ...... 189 5.5 References ...... 191 Chapter 6: The BXD lines as a tool to identify parent-of-origin effects with an indirect genetic effect ...... 197 Chapter 6: Preface...... 198 Abstract ...... 200 6.1 Introduction ...... 201 Figure 6.1: Production of reciprocal heterozygotes ...... 203 6.2 Materials and methods ...... 206 6.2.1 Experimental animals ...... 206 Table 6.1: Heterozygote litters produced from breeding two inbred BXD lines, identified by their line numbers ...... 206 6.2.2 Developmental measurements ...... 207 6.2.3. Behavioural observations ...... 207 Table 6.2: Maternal and offspring behaviours recorded ...... 208 6.2.4 Statistical analysis ...... 209 6.3 Results ...... 210 6.3.1 Weight ...... 210 Table 6.3: Tests of between-subjects effects on day 21 weight of RH offspring on 21 ...... 210 Figure 6.2: Weight of RH offspring on post-natal day 21...... 212 6.3.2 Behaviour ...... 213 Table 6.4: Phenotypes with a significant difference between the 55x92A and 92Ax55 reciprocal heterozygotes...... 213 6.4 Discussion ...... 215 6.4.1 Conclusion ...... 217 6.5 References ...... 218 Chapter 7: General discussion ...... 222 7.1 Main findings ...... 223 7.2 Knockout, normal variation and disease ...... 225 7.3 Fostering and maternal care ...... 227

Table of contents

7.4 Missing heritability ...... 229 7.5 Psychiatric disorders ...... 232 7.6 Collaborative cross ...... 234 7.7 Future work ...... 236 7.8 Conclusion ...... 238 7.9 References ...... 239 Appendix 1: Supplementary materials for chapter 2 ...... 249 A1.1 Supplementary tables ...... 250 Supplementary table 2.1: A table showing the human homologue gene symbols for all the mouse genes with a genome-wide p-value ≤ 0.05 for hippocampus weight in BXD, showing their unadjusted p-value in human. 250 Supplementary table 2.2: Genes which commonly co-express with MGST3 as determined by GeneFriends ...... 251 Supplementary table 2.3: KEGG pathway annotations significantly enriched (calculated by DAVID) in genes that are significantly co-expressed with MGST3 (calculated by GeneFriends) ...... 252 Supplementary table 2.4: Probes which correlate with all six of the correlating probes for Mgst3 in the adult mouse hippocampus as determined by Pearson correlations in GeneNetwork ...... 253 Supplementary figure 2.5: KEGG pathway annotations significantly enriched (calculated by DAVID) in genes significantly co-expressed with Mgst3 (calculated by Pearson correlation in GeneFriends) ...... 254 Supplementary table 2.6: Genes which commonly co-express with MGST3 as determined by GeneFriends and which co-express with Mgst3 in the adult mouse ...... 255 Supplementary table 2.7: KEGG pathway annotations significantly enriched (calculated by DAVID) for homologous genes which commonly co-express with MGST3, independent of tissue or treatment (as identified by GeneFriends) and which co-express with Mgst3 in the adult mouse hippocampus, as determined by Pearson correlation in GeneNetwork ...... 256 Appendix 2: Supplementary materials for chapter 3 ...... 257 A2.1 Supplementary figures ...... 258 Supplementary figure 3.1: Whole genome QTL maps for all elevated zero maze phenotypes ...... 258 Supplementary figure 3.2: Whole genome QTL maps for all open field test phenotypes ...... 258 A2.2 Supplementary tables ...... 259 Supplementary table 3.1: Genes within the mouse QTL we identified with human homologues ...... 259

Table of contents

Supplementary table 3.2: Genes commonly co-expressed (co-expression value ≥ 0.5, p-value ≤ 0.05) with MCTP1 and RXRG in humans, independent of tissue or treatment, as calculated by GeneFriends ...... 260 Supplementary table 3.3: Enrichment of disease annotations from WebGestalt for genes commonly co-expressed with MCTP1 and RXRG .. 261 Supplementary table 3.4: Gene Ontology (GO) annotation enrichment from WebGestalt for genes commonly co-expressed with MCTP1 and RXRG .. 262 Supplementary table 3.5: Genes commonly co-expressed (co-expression value ≥ 0.5, p-value ≤ 0.05) with RXRG and TNR, in humans, independent of tissue or treatment, as calculated by GeneFriends ...... 262 Supplementary table 3.6: Enrichment of disease annotations from WebGestalt for genes commonly co-expressed of RXRG and TNR ...... 262 Supplementary table 3.7: Gene Ontology (GO) annotation enrichment from WebGestalt for genes commonly co-expressed with RXRG and TNR ...... 263 Supplementary table 3.8: KEGG pathway annotation enrichment from WebGestalt for genes commonly co-expressed with RXRG and TNR ...... 263 Supplementary table 3.9: Pathway Commons annotation enrichment from WebGestalt for genes commonly co-expressed with RXRG and TNR ...... 264 Supplementary table 3.10: Wikipathways annotation enrichment from WebGestalt for genes commonly co-expressed with RXRG and TNR ...... 264 Supplementary table 3.11: Pearson correlation matrix (calculated by GeneNetwork) of striatal expression of mental disorder related gene probes and probes for Cmya5, Mctp1, Rxrg and Tnr ...... 264 Supplementary table 3.12: Pearson correlation matrix (calculated by GeneNetwork) of hippocampal expression of mental disorder related gene probes and probes for Cmya5, Mctp1, Rxrg and Tnr ...... 265 Supplementary table 3.13: Genes which commonly co-express with MCTP1, RXRG and TNR (from GeneFriends; Supplementary Tables 3.2 and 3.5) and co-express with Mctp1, Rxrg and Tnr in the mouse striatum (from GeneNetwork; Supplementary table 3.11) ...... 266 Supplementary table 3.14: Phenotypes which correlate with expression of all correlated probes for Cmya5, Mctp1, Rxrg and Tnr ...... 267 Supplementary table 3.15: Pearson correlations between our target phenotypes (Tables 3.1 and 3.2) and phenotypes found to correlate with expression of our candidate genes (Supplementary table 3.14) ...... 267 Appendix 3: Supplementary materials for chapter 4 ...... 268 A3.1 Supplementary tables ...... 269 Supplementary table 4.1: Functional and other details about the genes within the OspIge5.1 QTL for B6 maternal care on day 14, obtained from GeneNetwork, Entrez genes, and Mouse Genome Informatics ...... 269

Table of contents

Supplementary table 4.2: Functional and other details about the genes within the MatDge1.1 QTL for BXD nestbuilding on day 6, obtained from GeneNetwork, Entrez genes, and Mouse Genome Informatics ...... 269 Supplementary table 4.3: Functional and other details about the genes within the MatDge10.1 QTL for BXD maternal care on day 6, obtained from GeneNetwork, Entrez genes, and Mouse Genome Informatics ...... 269 Supplementary table 4.4: Functional and other details about the genes within the OspDge5.1 QTL BXD offspring solicitation on day 6, obtained from GeneNetwork, Entrez genes, and Mouse Genome Informatics ...... 270 Appendix 4: Supplementary materials for chapter 5 ...... 271 A4.1 Supplementary tables ...... 272 Supplementary table 5.1: Genes within the SocInt2.1 QTL ...... 272 Supplementary table 5.2: Genes within the SocInt4.1 QTL ...... 273 Supplementary table 5.3: Genes within the SocInt15.1 QTL ...... 274

Final word count (excluding references and appendixes): 45,881

Tables and figures

Chapter 1: General introduction ...... 17 Figure 1.1: Derivation of the BXD set...... 27 Figure 1.2: Indirect genetic effects influence phenotypes in interacting individuals...... 30 Table 1.1: Diseases linked to abnormal genomic imprinting, with their associated features and causative mutations...... 36 Chapter 2: Joint genetic analysis of hippocampal size in mouse and human identifies a novel gene linked to neurodegenerative disease ...... 56 Figure 2.1: Quantile-quantile plot of human homologues of significant mouse genes for hippocampus size...... 71 Table 2.1: Pearson’s correlations between probes for Mgst3 in adult mouse hippocampus...... 72 Chapter 3: A cross-species genetic analysis identifies candidate genes for mouse anxiety and human bipolar disorder...... 84 Figure 3.1: Bipolar disorder traits and analogous mouse phenotypes...... 90 Figure 3.2: Graphical representation of the research method used, providing a summary of the main findings...... 98 Table 3.1: Activity traits in the elevated zero maze which have a significant QTL ...... 101 Figure 3.3: QTL maps showing two QTL within each of the regions we identified ...... 100 Table 3.2: Vertical activity traits in the open field which have a significant QTL...... 102 Chapter 4: Genetic variation in offspring indirectly influences the quality of maternal behaviour in mice ...... 129 Figure 4.1: Experimental cross-foster design...... 134 Table 4.1: Maternal and offspring behaviours recorded...... 138 Figure 4.2: Offspring indirect genetic effect modifying maternal nestbuilding behaviour (OspIge7.1)...... 140 Figure 4.3 Offspring indirect genetic effect modifying maternal care (OspIge5.1) ...... 141 Figure 4.4: Maternal indirect genetic effect modifying offspring growth (MatIge17)...... 142 Table 4.2: A summary of the QTL identified ...... 143 Table 4.3: Tests of between-subjects effects of litter size, B6 foster mothers’ maternal care and day on BXD offspring solicitation, without phenotypes being divided by the litter size ...... 144 Table 4.4: Tests of between-subjects effects of litter size, B6 foster mothers’ maternal care and day on BXD offspring solicitation...... 145

Tables and figures

Figure 4.5: Correlation between BXD mothers’ nestbuilding on day 6 and expression of Hsd17b7 in adrenal gland ...... 147 Figure 4.6: Correlation between offspring and maternal traits in biological BXD families ...... 150 Figure 4.7: Correlation between offspring solicitation and corresponding bodyweight in BXD lines on day 10 and day 14, respectively ...... 150 Chapter 5: Indirect genetic effects influence sibling and maternal behaviour in mice ...... 159 Table 5.1: Maternal and offspring behaviours recorded...... 170 Figure 5.1 Experimental half-litter cross-fostering design ...... 173 Table 5.2: Summary of correlations between B6 and BXD pup phenotypes which show a QTL on chromosome 2 ...... 175 Figure 5.2: QTL map for traits which have a QTL on chromosome 2 ...... 176 Table 5.3: Summary of correlations between B6 mother, and B6 and BXD offspring phenotypes which show a QTL on chromosome 4, and traits which correlate significantly with them ...... 177 Figure 5.3: QTL map for traits which have a suggestive QTL on chromosome 4, and traits which correlate significantly with them ...... 178 Table 5.4: Summary of correlations between B6 and BXD offspring phenotypes which show a QTL on chromosome 15, and those which correlate significantly with them ...... 179 Figure 5.4: QTL maps for correlated traits as shown in table 5.4 ...... 180 Table 5.5: A summary of the traits within each of the three loci identified .. 181 Chapter 6: The BXD lines as a tool to identify parent-of-origin effects with an indirect genetic effect ...... 197 Figure 6.1: Production of reciprocal heterozygotes ...... 203 Table 6.1: Heterozygote litters produced from breeding two inbred BXD lines, identified by their line numbers ...... 201 Table 6.2: Maternal and offspring behaviours recorded ...... 208 Table 6.3: Tests of between-subjects effects on day 21 weight of RH offspring on 21 ...... 210 Figure 6.2: Weight of RH offspring on post-natal day 21...... 212 Table 6.4: Phenotypes with a significant difference between the 55x92A and 92Ax55 reciprocal heterozygotes...... 213 Chapter 7: General discussion ...... 222 Appendix 1: Supplementary materials for chapter 2 ...... 249 Supplementary table 2.1: The human homologue gene symbols for all the mouse genes with a genome-wide p-value ≤ 0.05 for hippocampus weight in BXD, showing their unadjusted p-value in human ...... 250

Tables and figures

Supplementary table 2.2: Genes which commonly co-express with MGST3 as determined by GeneFriends ...... 251 Supplementary table 2.3: KEGG pathway annotations significantly enriched (calculated by DAVID) in genes that are significantly co-expressed with MGST3 (calculated by GeneFriends) ...... 252 Supplementary table 2.4: Probes which correlate with all six of the correlating probes for Mgst3 in the adult mouse hippocampus as determined by Pearson correlations in GeneNetwork ...... 253 Supplementary figure 2.5: KEGG pathway annotations significantly enriched (calculated by DAVID) in genes significantly co-expressed with Mgst3 (calculated by Pearson correlation in GeneFriends) ...... 254 Supplementary table 2.6: Genes which commonly co-express with MGST3 as determined by GeneFriends and which co-express with Mgst3 in the adult mouse ...... 255 Supplementary table 2.7: KEGG pathway annotations significantly enriched (calculated by DAVID) for homologous genes which commonly co-express with MGST3, independent of tissue or treatment (as identified by GeneFriends) and which co-express with Mgst3 in the adult mouse hippocampus, as determined by Pearson correlation in GeneNetwork ...... 256 Appendix 2: Supplementary materials for chapter 3 ...... 257 Supplementary figure 3.1: Whole genome QTL maps for all elevated zero maze phenotypes ...... 258 Supplementary figure 3.2: Whole genome QTL maps for all open field test phenotypes ...... 258 Supplementary table 3.1: Genes within the mouse QTL we identified with human homologues ...... 259 Supplementary table 3.2: Genes commonly co-expressed (co-expression value ≥ 0.5, p-value ≤ 0.05) with MCTP1 and RXRG in humans, independent of tissue or treatment, as calculated by GeneFriends ...... 260 Supplementary table 3.3: Enrichment of disease annotations from WebGestalt for genes commonly co-expressed with MCTP1 and RXRG .. 261 Supplementary table 3.4: Gene Ontology (GO) annotation enrichment from WebGestalt for genes commonly co-expressed with MCTP1 and RXRG .. 262 Supplementary table 3.5: Genes commonly co-expressed (co-expression value ≥ 0.5, p-value ≤ 0.05) with RXRG and TNR, in humans, independent of tissue or treatment, as calculated by GeneFriends ...... 262 Supplementary table 3.6: Enrichment of disease annotations from WebGestalt for genes commonly co-expressed of RXRG and TNR ...... 262 Supplementary table 3.7: Gene Ontology (GO) annotation enrichment from WebGestalt for genes commonly co-expressed with RXRG and TNR ...... 263 Supplementary table 3.8: KEGG pathway annotation enrichment from WebGestalt for genes commonly co-expressed with RXRG and TNR ...... 263

Tables and figures

Supplementary table 3.9: Pathway Commons annotation enrichment from WebGestalt for genes commonly co-expressed with RXRG and TNR ...... 264 Supplementary table 3.10: Wikipathways annotation enrichment from WebGestalt for genes commonly co-expressed with RXRG and TNR ...... 264 Supplementary table 3.11: Pearson correlation matrix (calculated by GeneNetwork) of striatal expression of mental disorder related gene probes and probes for Cmya5, Mctp1, Rxrg and Tnr ...... 264 Supplementary table 3.12: Pearson correlation matrix (calculated by GeneNetwork) of hippocampal expression of mental disorder related gene probes and probes for Cmya5, Mctp1, Rxrg and Tnr ...... 265 Supplementary table 3.13: Genes which commonly co-express with MCTP1, RXRG and TNR (from GeneFriends; Supplementary tables 3.2 and 3.5) and co-express with Mctp1, Rxrg and Tnr in the mouse striatum (from GeneNetwork; Supplementary table 3.11) ...... 266 Supplementary table 3.14: Phenotypes which correlate with expression of all correlated probes for Cmya5, Mctp1, Rxrg and Tnr ...... 267 Supplementary table 3.15: Pearson correlations between our target phenotypes (Tables 3.1 and 3.2) and phenotypes found to correlate with expression of our candidate genes (Supplementary table 3.14) ...... 267 Appendix 3: Supplementary materials for chapter 4 ...... 268 Supplementary table 4.1: Functional and other details about the genes within the OspIge5.1 QTL for B6 maternal care on day 14, obtained from GeneNetwork, Entrez genes, and Mouse Genome Informatics ...... 269 Supplementary table 4.2: Functional and other details about the genes within the MatDge1.1 QTL for BXD nestbuilding on day 6, obtained from GeneNetwork, Entrez genes, and Mouse Genome Informatics ...... 269 Supplementary table 4.3: Functional and other details about the genes within the MatDge10.1 QTL for BXD maternal care on day 6, obtained from GeneNetwork, Entrez genes, and Mouse Genome Informatics ...... 269 Supplementary table 4.4: Functional and other details about the genes within the OspDge5.1 QTL BXD offspring solicitation on day 6, obtained from GeneNetwork, Entrez genes, and Mouse Genome Informatics ...... 270 Appendix 4: Supplementary materials for chapter 5 ...... 271 Supplementary table 5.1: Genes within the SocInt2.1 QTL ...... 272 Supplementary table 5.2: Genes within the SocInt4.1 QTL ...... 273 Supplementary table 5.3: Genes within the SocInt15.1 QTL ...... 274

General abstract

Institution: The University of Manchester

Name: David Ashbrook

Degree Title: Doctor of Philosophy

Thesis Title: A systems-genetics analyses of complex phenotypes

Date: 2015

Complex phenotypes are traits which are influenced by many factors, and not just a single gene, as for classical Mendelian traits. The brain, and its resultant behaviour, gives us a large subset of complex phenotypes to examine. Variation in these traits is affected by a range of different influences, both genetic and environmental, including social interactions and the effects of parents.

Systems-genetics provides us with a framework in which to examine these complex traits, seeking to connect genetic variants to the phenotypes they cause, through intermediate phenotypes, such as gene expression and protein levels. This approach has been developed to exploit and analyse massive data sets generated for example in genomics and transcriptomics.

In the first half of this thesis, I combine genetic linkage data from the BXD recombinant inbred mouse panel with genome-wide association data from humans to identify novel candidate genes, and use online gene annotations and functional descriptions to support these candidates. Firstly, I discovered MGST3 as a novel regulator of hippocampus size, which may be linked to neurodegenerative disorders. Secondly, I identified that CMYA5, MCTP1, TNR and RXRG are associated with mouse anxiety-like phenotypes and human bipolar disorder, and provide evidence that MCTP1, TNR and RXRG may be acting via inter-cellular signalling in the striatum.

The second half of this thesis uses different cross-fostering designs between genetically variable BXD lines and the genetically uniform C57BL/6J strain to identify indirect genetic effects and the loci underlying them. With this, I have found novel loci expressed in mothers that alter offspring behaviour, novel loci expressed in offspring affecting the level of maternal care, and novel loci expressed in offspring, which alter the behaviour of their nestmates, as well as the level of maternal care they receive. Further I provide evidence of co- adaptation between maternal and offspring genotypes, and a positive indirect genetic effect of offspring on their nestmates, supportive of a role for kin selection. Finally, I demonstrate that the BXD lines can be used to investigate genes with parent-of-origin dependent expression, which have an indirect genetic effect on maternal care.

In conclusion, this thesis identifies a number of novel loci, and in some cases genes, associated with complex traits. Not only are these techniques applicable to other phenotypes and other questions, but the candidates I identify can now be examined further in vitro or in vivo.

Declaration

No portion of the work referred to in the thesis has been submitted in support of an application for another degree or qualification of this or any other university or other institute of learning.

Copyright statement i. The author of this thesis (including any appendices and/or schedules to this thesis) owns certain copyright or related rights in it (the “Copyright”) and s/he has given The University of Manchester certain rights to use such Copyright, including for administrative purposes. ii. Copies of this thesis, either in full or in extracts and whether in hard or electronic copy, may be made only in accordance with the Copyright, Designs and Patents Act 1988 (as amended) and regulations issued under it or, where appropriate, in accordance with licensing agreements which the University has from time to time. This page must form part of any such copies made. iii. The ownership of certain Copyright, patents, designs, trade marks and other intellectual property (the “Intellectual Property”) and any reproductions of copyright works in the thesis, for example graphs and tables (“Reproductions”), which may be described in this thesis, may not be owned by the author and may be owned by third parties. Such Intellectual Property and Reproductions cannot and must not be made available for use without the prior written permission of the owner(s) of the relevant Intellectual Property and/or Reproductions. iv. Further information on the conditions under which disclosure, publication and commercialisation of this thesis, the Copyright and any Intellectual Property and/or Reproductions described in it may take place is available in the University IP Policy (see http://documents.manchester.ac.uk/DocuInfo.aspx?DocID=487), in any relevant Thesis restriction declarations deposited in the University Library, The University Library’s regulations (see http://www.manchester.ac.uk/library/aboutus/regulations) and in The University’s policy on Presentation of Theses

Acknowledgements

I would firstly like to thank my supervisor, Reinmar Hager, for all his help, support, guidance and patience during my PhD. The time and opportunities he has given me throughout my PhD have been invaluable.

Thank you to Claire, for her help and support, and putting up with me during my

PhD. I would like to thank everyone in the office, past and present, for keeping me sane and social: Naorin, Chrissy, Emma, Xaali, Jade, Marco, Miguel,

Kayleigh, Amy, Bea, Barbara (sorry to everyone I’ve missed!). Thanks to the entire Systems Biology cohort for sharing the pain of the PhD.

I would also like to thank my collaborators, especially Rob Williams, Jason Stein,

Derrek Hibar and Sarah Medland. Their advice, data and insight helped to enrich my PhD.

I would like to acknowledge the support of the BBSRC for funding my PhD, and giving me the amazing opportunity which I have had.

But I am very poorly today & very stupid & hate everybody & everything.

One lives only to make blunders.— I am going to write a little Book for

Murray on orchids & today I hate them worse than everything so farewell

& in a sweet frame of mind, I am | Ever yours | C. Darwin (Darwin, C. R.

to Lyell, Charles, 1 Oct 1861)

Chapter 1: General introduction

1.1 Overview

Complex traits, or complex phenotypes, can be defined as any trait in which the variability is due to a large number of factors, which can be separated into genetic and non-genetic sources of variation and their interplay (Gratten et al.,

2014). This, therefore, means that most traits which are not Mendelian can be considered to be complex traits.

This project concentrates on the brain and behaviour, as a (large) subset of complex traits. It has been said that “The brain is the chief architect, orchestrator and driver of behavior; behavior, in turn, is the principal function of the brain” (Gomez-Marin et al., 2014, p. 1455), and therefore to understand one we need to understand the other. The brain and the behaviours that it causes are highly complex traits influenced by many factors including genes (Hager et al.,

2012; Hitzemann et al., 2013; McCarroll and Hyman, 2013), environment (Carola et al., 2006; Paus et al., 2012), social interactions (Caldji et al., 2004; Prakash et al., 2006), parent-of-origin effects (Keverne et al., 1996; Curley, 2011; Kopsida et al., 2011) and epigenetically inherited factors (Keverne and Curley, 2008;

Tarantino et al., 2011; Keverne, 2014), to give but a few examples.

Investigating this complexity requires not just novel experimental design, but also new methods of analyses. Systems approaches seek to connect the different layers of a biological system to give a more complete picture of the whole, rather than concentrating on one level or one aspect of the system (Raad et al., 2012). Indeed, systems-genetics has a definition almost perfectly designed to investigate complex traits: “A global analysis of the molecular factors that underlie variability in physiological or clinical phenotypes across individuals in a population. It considers not only the underlying genetic variation but also intermediate phenotypes such as gene expression, protein levels and metabolite

Chapter 1: General introduction levels, in addition to gene-by-gene and gene-by-environment interactions”

(Civelek and Lusis, 2014, p. 34).

In previous approaches phenotypic changes would be directly attributed to different genotypes, without being able to examine the mechanisms linking them or specific variants underlying them (e.g. Gelman et al., 1988). The huge amounts of data now available, and tools to analyse them, provide great opportunities to go beyond identifying regions of the genome which may be associated with a trait, allowing us to hypothesise about the biological pathways which lead from gene to phenotype, how these interact with the non-genetic factors outlined above and make specific predictions which can then be tested.

In this introduction I will seek to outline some of the techniques, tools and concepts which I will make use of throughout my thesis. Firstly, I will discuss genetic links, associations and gene annotations, before moving on to systems- genetics, the approach it takes and some previous uses of it. I will go on to talk about some of the tools and resources available and the advantages of publically available ‘big data’. Then I will move on to describe recombinant inbred (RI) lines, and in particular the BXD panel of which I have made extensive use. I will describe indirect genetic effects (IGEs) and parent-of-origin effects (POEs), and how they can influence phenotype. Finally, I will finish with a description of parental care, why it is important, and the behaviours we examine.

Chapter 1: General introduction

1.2 Linkage, association and gene annotations

Quantitative trait loci (QTL) are segments of the genome affecting a particular phenotype (Falconer and Mackay, 1996). QTL mapping, identifying QTL to explain the genetic basis of complex traits, relies on being able to make correlations between genetic markers and phenotypic traits in a population.

Individuals are scored for their phenotype for a particular trait, and their genotype at a marker. If there is a difference in mean phenotype between those individuals with one genotype compared with the other, than we can infer that there is a QTL linked to that marker. If there is no difference between the means, then we can conclude that the loci does not influence the phenotype in that population

(Falconer and Mackay, 1996; Miles and Wayne, 2008). Early studies relied on a very limited number of markers, however with technological advances such as whole genome sequencing and microarrays, large numbers of markers can be quickly and accurately genotyped throughout the genome. Every QTL study is a trade-off between cost, the number of markers used (more allowing a more precise localisation of the QTL) and the number of samples (increasing power to find a QTL). To efficiently map a QTL to a precise location requires a large segregating experimental population, such as an F2 cross population or genetic reference populations. In this thesis I have primarily used a specific type of genetic reference population, RI lines, and these will be discussed later. Although

QTL mapping has shown the genetic basis of some very complex behavioural traits (e.g. burrow construction; Weber et al., 2013), these QTL can often contain tens or hundreds of genes.

Genome-wide association studies (GWASs) are conceptually similar to

QTL studies, but rather than using hundreds or thousands of markers, they genotype millions of single nucleotide polymorphisms (SNPs) throughout the genome. In contrast to QTL, GWASs can, potentially, narrow down to single

Chapter 1: General introduction

SNPs, showing the causative polymorphism. Due to a number of complications, particularly linkage-disequilibrium (the fact that closely linked SNPs are often inherited together), it is in practice difficult to narrow to a single SNP, and instead the associated SNP represents one or more variants that have not been directly genotyped but that are associated with the genotyped variant. GWASs have been able to identify thousands of loci for many different diseases and phenotypes, but, alone, the list of candidate genes they provide does not give any mechanistic insight into physiology. Indeed, the vast majority of commonly detected SNPs only have a very small effect size, and therefore by looking at any of these genes individually we may gain little insight (Goldstein, 2009; Lander,

2011).

Various gene ontologies have been developed to describe genes in terms of their molecular action, biological function, subcellular localisation (Ashburner et al., 2000; Gene Ontology Consortium, 2015), biological pathways they are involved in, and diseases they have been associated with (Kanehisa and Goto,

2000; Kanehisa et al., 2012). This allows the huge amount of literature that may be available for a given gene, or its protein products, to be accessed and understood quickly. This is advantageous because when a candidate gene is identified (e.g. from QTL mapping or GWASs), its known functions can be examined to see if they relate to the phenotype. These annotations also allow enrichment analyses to be carried out. Enrichment analysis takes a list of genes

(e.g. generated from a co-expression analysis), and then examines if any annotation is enriched within the list (e.g. are more genes in the list annotated with a particular annotation than would be expected from a random selection of genes). This is useful, as most genes will play a role in many systems, but the shared annotations should be those which are relevant to the trait of interest

(Zhang et al., 2005; Wang et al., 2013).

Chapter 1: General introduction

1.3 ‘Big data’ and ‘omics’

We have reached a point where large amounts of data can be gathered on a whole range of biological scales (‘omics’ technologies), we have the computational power to analyse these data, and they can be quickly shared across the internet. These ‘omics’ technologies derive from leaps forwards in nucleotide sequencing (Smith and Hood, 1987), applying mass spectrometry to biological questions (Fenn et al., 1989; Wasinger et al., 1995), and microarrays

(Schena et al., 1995), all of which allow large amounts of data to be rapidly gathered.

Although some datasets are produced with the specific aim of being shared, e.g. the Allen Brain Atlases (Lein et al., 2007), more are produced to answer a specific question (Ferguson et al., 2014). Repositories have been established to contain data of different types, for example genomes on the

UCSC Genome Browser (Kent et al., 2002) and transcriptomes on the Gene

Expression Omnibus (GEO; Edgar et al., 2002; Barrett et al., 2011, 2013), as well as more species specific repositories (e.g. Chesler et al., 2004; Harris et al.,

2010; Eppig et al., 2015). Some of these repositories integrate their own tools, while other sites have been developed as resources to analyse and interpret large datasets (e.g. Wang et al., 2003; Zhang et al., 2005; Wang et al., 2013).

These repositories and tools allow us to bring together diverse datasets, each of which may contain only a small amount of data, and use them to answer new biological questions which would not have been possible with any of them individually.

For a gene to be able to influence a trait it needs to be transcribed to

RNA, from which it can then act directly or be translated into a protein.

Transcriptomics is the identification and quantification of all transcripts within a

Chapter 1: General introduction sample. For many years this has been done using microarrays. Early microarrays primarily hybridized the transcripts’ 3′ end, meaning that isoforms could not be easily identified (Ghazalpour et al., 2011). Later, exome arrays were produced, which had probes for a number of locations within each gene, and therefore were more able to deal with isoform differences (Mulligan et al., 2012).

RNA-seq is a newer technology for transcriptomics, which has the advantage over microarrays that, as well as providing expression levels, also gives information about RNA splicing, copy number variants (CNVs) and transcripts which may not be detected on commercial microarrays. Fortunately, measured expression levels appear consistent between the two approaches

(Marioni et al., 2008; Ghazalpour et al., 2011; Liu et al., 2011), meaning previously collected data can be used and compared to current experiments.

Transcript levels can be treated like any other phenotypic trait, in that QTL mapping or GWASs can be carried out to examine which loci are influencing the expression of the transcript (Brem et al., 2002; Brem and Kruglyak, 2005;

Bennett et al., 2010; Van Nas et al., 2010; Lappalainen et al., 2013). These loci which are associated with changes in transcript expression are often termed expression QTL (eQTL): a variant (or variants) within the locus alters the expression of the gene of interest. An eQTL found near to the location (~ ≤

1Mbp) of the transcript is described as a local eQTL, and are often called cis- eQTL. This is in contrast to trans-eQTL which are found more distally. Cis-eQTL are interesting when they are found for a gene within a QTL for another phenotype (e.g. mapping a QTL for a behavioural phenotype, and then looking at the eQTL for genes within this QTL), as it indicates that the gene is regulating its own expression, and therefore increases its likelihood of being the causal variant, as, in this case, it shows that a genetic variant within the locus causes a change in the expression of the mRNA, and this change in mRNA is likely to have

Chapter 1: General introduction functional consequences. These cis-eQTL variants are found within gene transcription elements, such as promoter elements or TATA boxes (Doss et al.,

2005). Also, cis-eQTL appear to be conserved between studies and tissues, and so variants may be effecting several systems (Van Nas et al., 2010). However, finding no eQTL or a trans-QTL does not rule it out as a causal variant as, for example, structural changes or alternative splicing could be changing the function of the protein product, without a change in expression. On the other hand, most SNPs found by GWASs map outside of the protein-coding regions of the genome, which suggests that changes in expression, rather than changes in function, might be more important in natural variation of complex traits (Civelek and Lusis, 2014).

1.4 Systems-genetics

Gene variants identified as associated with a trait, for example using QTL mapping or GWASs, can be examined in many different ways, from individual molecules, to cell culture, to whole animal models. However, these can be expensive in terms of both time and money, and many genetic variants revealed through GWASs have modest effects which may be difficult to examine in isolation. Indeed, in some cases it may be impossible, as the gene may only affect the phenotype of interest when it interacts with another particular allele

(gene-by-gene interactions, or epistasis) or in a particular environment (gene-by- environment effects; GxE) (e.g. Ayroles et al., 2009; Wu et al., 2012).

Systems-genetics provides a different, complementary, approach, taking a more global, holistic view, which shares many traits with systems biology. In systems-genetics approaches genetic variation (often natural genetic variation) is used to perturb the biological system, and a global analysis of biological molecules is carried out. Technological advances have now allowed us to look at

Chapter 1: General introduction millions of gene variants, the expression level of tens of thousands of transcripts and hundreds of proteins (Foss et al., 2007; Melzer et al., 2008; Ghazalpour et al., 2011), which can all be measured in the same individuals, alongside more traditional phenotypes. From this data, correlations, networks and associations can be made, as well as more complex statistical models (Civelek and Lusis,

2014). Further, by looking at inter-individual variation within a population, we can examine the action of different combinations of allelic variants (gene-by-gene interactions) and the affect a particular variant exerts on different genetic backgrounds (Civelek and Lusis, 2014).

1.5 Recombinant inbred (RI) lines and the BXD

RI lines are an incredibly useful resource for systems-genetics, and are used throughout this thesis, as they provide an opportunity to study genetics, epigenetics and environment. RI lines are produced by breeding founder strains together for several generations to produce recombination of the genome and these recombinant litters are then inbred by brother/sister mating for at least 20 generations to produce a series of fully inbred lines, which are homozygous at every locus but have a fixed pattern of possible alleles (Figure 1.1; Peirce et al.,

2004; Pollard, 2012; Gini and Hager, 2012).

RI lines allow the mapping of QTL for complex traits due to the fact that they have known genotypes that can be easily compared between the lines to identify regions of the genome associated with the trait (Peirce et al., 2004). Their advantage over other populations to map QTL (such as intercrosses, heterogeneous stock or human cohorts) is that they represent a stable genotype over time, and therefore give the ability to collect huge amounts of phenotype data which is coherent and consistent. Since they have a fixed genotype, RI lines can be used in different places and at different times, allowing many data points

Chapter 1: General introduction to be collected for the same genome, reducing confounding environmental noise and therefore allowing mapping of traits with low heritability (Peirce et al., 2004).

Conversely, genetics can be controlled by using only individuals of the same genotype, allowing non-genetic factors (environmental factors) to be investigated

(Aziz et al., 2007). Further, since multiple traits can be examined in the same genotype, this makes it perfect for a systems approach (Andreux et al., 2012).

Chapter 1: General introduction

Figure 1.1: Derivation of the BXD set. The BXD lines are derived from the C57BL/6J and DBA/2J strains. Following a cross between the two, the F1 generation consists of genetically identical individuals that inherited one chromosome from each parental strain.

Intercrosses were then carried out between F1 individuals, generating recombination in the F2. Patterns of recombination were frozen by >20 generations of sibling matings, which resulted in almost complete homozygosity in generation F23. Breeding was continued by full sibling matings, and individuals were monitored to ensure consistency in the genotype of each line over time. Each line represents a unique mosaic of C57BL/6J and DBA/2J alleles; there is extensive variation between lines, and virtually no genetic variation between individuals of one line. Modified from Gini and Hager, 2012.

Chapter 1: General introduction

The BXD strain are a group of RI lines, bred from the parental C57BL/6J

(B6) and DBA/2J (D2) lines, first established in the 1970s (Taylor et al., 1973), with progressively more lines being added (Taylor et al., 1999; Peirce et al.,

2004). There are approximately 80 extant, fully inbred, BXD lines (Gini and

Hager, 2012) with more being developed. The phenome of the BXD lines has now been extensively catalogued (http://www.genenetwork.org), and is currently the largest of any mammalian model system, both in terms of number of extant inbred lines available and the number of phenotypes analysed. The BXD have been used to study both normal biological processes (including pharmacology

(Rulten et al., 2006), behaviour (Philip et al., 2010), physiology (Seecharan et al.,

2003; Hager et al., 2012) and genetics (Mozhui et al., 2011)) as well as the interaction between genes and interventions, sex or age (e.g. Wu et al., 2014;

Hayes et al., 2014). Further, the BXD phenome is particularly enriched for neuronal and behavioural phenotypes, and transcriptome data is available for many brain regions (e.g. Overall et al., 2009). The number of lines, and the extensive study made of the lines, has allowed not just QTL to be identified in the

BXD, but the causal genes underlying them (e.g. Miyairi et al., 2007; Mozhui et al., 2008; Koutnikova et al., 2009; Boughter et al., 2012; Miyairi et al., 2012;

Williams et al., 2014). Combining them with other resources (Houtkooper et al.,

2013; Ashbrook et al., 2014a; Williams and Auwerx, 2015), including human resources (Koutnikova et al., 2009; Ashbrook et al., 2014b, 2015), allows specific candidate genes and mechanisms to be identified.

1.6 Indirect genetic effects (IGEs)

Phenotypes have traditionally been investigated for direct genetic effects (DGEs), i.e. the effect of an individual’s genotype on the same individual’s phenotype

(Geldermann, 1975). IGEs occur when the genotype of one individual alters the phenotype of a second individual, and can be produced in any group of

Chapter 1: General introduction interacting individuals (Moore et al., 1997). Phenotypes expressed during social interactions have long been recognised as evolutionarily important (Moore et al.,

1997). Indeed, some traits cannot exist without social interactions, for example courtship behaviour (Whitney et al., 1973; Doty, 1974; Neunuebel et al., 2015), play behaviour (Cox and Rissman, 2011; Lukas and Wöhr, 2015), aggression

(Anholt and Mackay, 2012) and, importantly for this thesis, parental care, which will be detailed in section 1.8. The phenotypes upon which IGEs act have been referred to as interacting phenotypes (Moore et al., 1997).

Understanding IGEs in evolutionary terms is complicated as they are

‘evolving environments’: that is they are environmental because they exist outside of the individual expressing the behaviours, but are evolvable, since they are dependent on the genes of the second individual (Figure 1.2). Therefore, this represents an environment which can respond to selective pressures (Bijma and

Wade, 2008; McGlothlin et al., 2010). Further, since interactions are two-way,

IGE related genes can evolve in both individuals, causing feedback between the individuals (Moore et al., 1997). The first formal treatment of the genetics of

IGEs, which incorporated phenotypic evolution and interactions between unrelated individuals, was by Moore et al. (1997).

It is important to understand IGEs for two reasons. Firstly, they can have a significant effect on evolution; for example they are to able alter the rate of evolution and allow evolution to occur independent of the DGE (Moore et al.,

1997). Secondly, they can have a significant effect on health, e.g. metabolic disease (Wells, 2007). For example, maternal genotype, at least in part, influenced offspring birthweight (Lunde et al., 2007), which in turn can have health implications later in life, predisposing to coronary heart disease, type 2 diabetes and hypertension (Barker et al., 2002).

Chapter 1: General introduction

Expression of a phenotype in a focal individual (P), is dependent upon the genotype of that focal individual (G1) and the environment (E) [P = G1*E]. Most genetic mapping methods however, treat the phenotype as being dependent on the genotype [P = G1] as environment is controlled in laboratory conditions. For phenotypes influenced by IGEs their expression is also dependent on the genotype of the interacting individual (G2; P = G1*G2*E; Figure 1.2). Using our experimental design, we are able to control for E (as all animals are kept under the same conditions), G1 (as all animals are of the same genotype i.e. B6) and again, we assume that changes to individual B’s phenotype are due to their own genotype. Therefore, the equation can be simplified to P = G2: differences in the phenotype of the observed individual are due to the phenotype of the interacting individual, which is determined by the interacting individual’s genotype. This means that normal mapping methods can be used, replacing the genotype of the focal individual with the genotype of the interacting individual.

Chapter 1: General introduction

Figure 1.2: Indirect genetic effects influence phenotypes in interacting individuals.

A) Environment, genotype and phenotype are shown for two individuals, showing how they interact to cause the phenotype of the focal individual. Both the indirect genetic effect (IGE; shown by the blue arrow), and the direct genetic effect (DGE) of focal genotype on focal phenotype are labelled. B) Using our experimental design we are able to control for variation in environment and genotype (represented by being crossed out), and therefore the focal individual’s phenotype (P) is dependent on individual B’s genotype (G2).

Chapter 1: General introduction

Although the above shows that the importance of IGEs has been recognised for many years, and it has been shown that behaviours expressed in mice are influenced by the genotype of interacting individuals, e.g. Hughes

(1989), there have been few studies to identify the genetic loci influencing these traits (Wolf et al., 2002; Wilson et al., 2004; Casellas et al., 2009; Wolf et al.,

2011; Wolf and Cheverud, 2012).

A specific sub-category, and the classic example of IGEs, is the maternal genetic effect (MGE) of a mother’s genotype on her offspring’s phenotype, independent of any shared genetic material (Kirkpatrick and Lande, 1989; Wolf and Wade, 2009; Wolf and Cheverud, 2012). In many species, and particularly mammals, the mother provides a large part of the early life environment, and therefore changes in the mothers’ phenotype, due to her genotype, will alter the developing young’s phenotype. These maternal effects can have significant effects on the phenotype of offspring, such as maternal nursing on offspring adult blood pressure (Gouldsborough et al., 1998) and behaviour (Caldji et al., 2000;

Cameron et al., 2008). However MGE can easily be confused with maternal inheritance (that is influenced from the mother due to genetic factors, including nuclear or mitochondrial DNA), and therefore experiments need to be carefully designed to separate these influences (Kirkpatrick and Lande, 1989).

1.7 Parent-of-origin effects (POEs), epigenetics and genomic imprinting

As well as maternal effects described above, parental genotype can affect offspring in other ways as well. An example of this are POEs, epigenetic effects whereby the expression of phenotypes or genes are dependent on which parent it was inherited from, and not on the sequence of the allele inherited (Keverne and Curley, 2008). Epigenetic effects such as this are due to changes in expression of genes within the individual, but are not a result of a change in the

Chapter 1: General introduction nucleotide sequence. There are several forms of POEs, reviewed by Guilmatre and Sharp (2012), amongst which is genomic imprinting.

Genomic imprinting is differential expression of paternal and maternal genes, and in the most extreme cases complete silencing of the paternal or maternal allele, resulting in mono-allelic expression (Khatib, 2007; Cheverud et al., 2008; Wolf et al., 2008a). There appear to be several mechanisms by which imprinting can occur, however, since they manifest in the same way phenotypically, they are subject to the same selective pressures and can therefore be analysed from an evolutionary perspective together (Wolf et al.,

2008b).

From a mechanistic point of view, methylation appears to be the most important control on genomic imprinting (Strogantsev and Ferguson-Smith,

2012). The majority of imprinted genes found so far occur in clusters including imprinted non-coding RNA, imprinted protein coding genes and non-imprinted genes. Differently methylated regions are found on the maternally and paternally derived chromosomes, and these appear to regulate imprinting across the whole cluster, acting as imprinting control regions (Edwards and Ferguson-Smith,

2007). Chromatin- , transcription- and non-coding RNA- mediated mechanisms have been demonstrated to play a role in controlling imprinting of some genes, building up a more complicated picture which needs further exploration (Kacem and Feil, 2009).

Genomic imprinting has been shown in both the B6 and D2 strains

(Downing et al., 2011) as well as the BXD lines (Shibata et al., 1995) and online data (http://www.genenetwork.org) from the parental reciprocal crosses also shows evidence of POEs. This demonstrates that the BXD are a valid system to explore the phenomenon.

Chapter 1: General introduction

Over 150 imprinted genes and ncRNAs have been reported, with more being suggested as imprinted, although there is debate around these studies and the potential for tissue specific imprinting (Barlow, 1995; DeVeale et al., 2012;

Babak et al., 2015; Bonthuis et al., 2015; Crowley et al., 2015). Imprinted QTL

(iQTL), that is QTL which are dependent upon which parent the allele was derived from, have been identified for a range of phenotypes (Cheverud et al.,

2008; Gao et al., 2013). It is important to understand genomic imprinting, as it may influence evolution, development and heath. Imprinting is known to have a strong effect on the placenta, with the majority of known imprinted genes being expressed there, and therefore this has been reviewed extensively (Frost and

Moore, 2010; Lim and Ferguson-Smith, 2010; Fowden et al., 2011; Renfree et al., 2013).

In terms of this thesis’s focus on the brain and behaviour, there is also strong evidence for imprinted genes playing a role in the brain (Dent and Isles,

2014; Perez et al., 2015) with early studies showing that paternal and maternal genes differentially contribute to the limbic system and cortex respectively (Allen et al., 1995; Keverne et al., 1996; Badcock and Crespi, 2006). Further work suggests a change from more maternally expressed imprinted genes in the embryonic brain to more paternally expressed imprinted genes in the adult brain which is supportive of a role of imprinted genes that changes through the lifetime of the animal (Gregg et al., 2010). Imprinted genes can be found influencing a number of neuronal systems, e.g. maternally expressed 5-hydroxytryptamine

(serotonin) receptor 2A (HTR2A) in serotonergic transmission (Kato et al., 1998), and behaviours, such as growth factor receptor bound protein 10 (Grb10) which influences social behaviour and is paternally expressed in specific brain regions

(for example the diencephalon, ventral midbrain and the medulla oblongata) but maternally or biallelically expressed elsewhere (Garfield et al., 2011). Very

Chapter 1: General introduction recently it has been found that imprinting may be specific to individual neurons, and that this can have direct behavioural consequences (Bonthuis et al., 2015).

A number of disorders are associated with abnormal imprinting in humans

(Lim and Ferguson-Smith, 2010). Table 1.1 shows that, although there are a diversity of different phenotypes, these diseases share a major detrimental effect on both physical and mental development, as would be predicted from the placenta and brain being strongly influenced by imprinting. These disorders are directly caused by alteration of normal imprinting, either by a genetic mutation

(causing altered or loss of expression of the imprinted gene) or by loss or gain of imprinting (resulting in both alleles being expressed or neither).

POEs have also been suggested to influence susceptibility to a number of common diseases, with genomic imprinting specifically implicated in some and this has been reviewed by Guilmatre and Sharp (2012). Examples include diabetes type one and two (Unoki et al., 2008; Mackay and Temple, 2010;

Wallace et al., 2010) and neuropsychiatric disorders (Crespi, 2008; Kopsida et al., 2011).

Chapter 1: General introduction

Table 1.1: Diseases linked to abnormal genomic imprinting, with their associated features and causative mutations. Adapted from Lim and Ferguson-Smith (2010).

UPD, Uniparental disomy. IUGR, Intrauterine growth restriction. ICR, Imprinting control region. DMR, Differentially methylated region.

Disease / Associated features Parent of origin effects phenotypes Beckwith- Pre-and post-natal overgrowth, pUPD11, maternal Wiedemann macroglossia, organomegaly, microdeletions in H19DMR or syndrome (BWS) neonatal hypoglycemia, maternal deletion hemihypertrophy, predisposition KCNQ1OT1DMR, CDKN1C to some embryonal tumours mutation, hypomethylation of including Wilms’ tumour H19 and/or KCNQ1OT1DMR Silver-Russell IUGR, facial dysmorphism, post- mUPD7, paternal syndrome (SRS) natal growth retardation, hypomethylation at H19 urogenital abnormalities, skeletal asymmetry Angelman Microcephaly, mental retardation, pUPD15, maternal 15q11 syndrome (AS) autistic behaviour, seizures, →q13 deletion, maternal absence of speech, sleep mutation within UBE3A, disorder hypomethylation of maternal ICR Prader-Willi IUGR, pre- and post-natal mUPD15, paternal syndrome (PWS) hypotonia, mild mental 15q11 → q13 deletion, retardation, facial dysmorphology, hypermethylation of paternal hypogondism, short stature and ICR small extremities mUPD14 or IUGR, hypotonia, obesity, short mUPD14, decreased DLK1 mUPD(14)-like stature, developmental retardation and RTL1 expression, phenotype and early puberty onset paternal deletion, paternal hypomethylation of IG-DMR and MEG3-DMR on HSA14q32.2 pUPD14 or Mental retardation, growth delay, pUPD14, increased RTL1 pUPD(14)-like facial abnormalities, small and expression, maternal phenotype bell-shaped thorax, abdominal microdeletion of IG-DMR wall defects and/or MEG3-DMR on HSA14q32.2. Maternal hypermethylation of IG-DMR Pseudohypoparath Short stature, obesity, mUPD 20, Maternal or yroidism/ Albrights brachydactyly, ectopic paternal inherited Gsα hereditary ossifications, mental retardation mutations, loss-of-imprinting osteodystrophy at GNAS exon A/B DMR Transient neonatal Severe IUGR, dehydration, Overexpression of diabetes mellitus persistent hyperglycemia in first 6 ZAC1/PLAGL1, paternal (TNDM) weeks after birth, elevated later duplication of 6q24, risk of diabetes recurring hypomethylation of PLAG1 ICR

Chapter 1: General introduction

1.8 Parental care

Several of the chapters of this thesis deal with parental care, specifically maternal care, and therefore the evolution and importance of this behaviour should be outlined. Parental care can be defined as “Any parental trait that enhances the fitness of a parent’s offspring, and that is likely to have originated and/or to be currently maintained for this function” (Smiseth et al., 2012, p. 7) . At first look it would seem that parental care is a poor deal for the parents, as it results in a loss of time and energy (e.g. from nestbuilding; Bucher et al., 1982), supressed immunity (Medina, 1993), increased parasite load (Nordling et al.,

1998; Descamps et al., 2009) and potentially reduced future reproductive success or mortality. However, that parental care has evolved numerous times, within different taxa, and in a variety of different forms, shows that there must be an evolutionary benefit for it (Klug and Bonsall, 2014). To understand this, we need to understand kin selection (Hamilton, 1964): that a behaviour will evolve if the benefit to the recipient (in this case offspring) is larger than the cost to the actor (in this case the parent) dependent on their relatedness coefficient. Each parent has a relatedness co-efficient of 0.5 with their offspring (they share approximately 50% of their DNA). Therefore, parental care will evolve when the benefit to offspring is twice the cost to the parent. This is because the parents gain greater fitness (e.g. more grand-offspring) through increasing their offspring’s fitness than they would from increasing their own fitness directly

(Smiseth et al., 2012).

However, there is also conflict between parents and offspring about the amount of care which should be given, due to the fact that the relatedness of each to themselves is 1, while each to the other is 0.5. This means that there is a disagreement between parents and offspring about the amount of parental

Chapter 1: General introduction investment which should be made (Trivers, 1974). Therefore, offspring may develop mechanisms to manipulate parental investment (Trivers, 1974).

In most mammals, the post-natal period is characterized by offspring being dependent on their mother for warmth, nutrition and protection. Indeed, differences in maternal behaviour between different strains of mice has a significant correlation with offspring survival (Carlier et al., 1982). I will concentrate on two of these post-natal maternal behaviours, food provisioning via lactation, and warmth, via nestbuilding (Klug and Bonsall, 2014), as these are relevant in a laboratory situation.

There is clear evidence of differences in maternal care between different strains of rodents (McIver and Jeffrey, 1967; Carlier et al., 1982; Myers et al.,

1989; Champagne et al., 2007; Chourbaji et al., 2011), which strongly suggests a genetic basis to maternal care, and indeed loci have been mapped for maternal care in mice (Peripato and Cheverud, 2002; Peripato et al., 2002). Further, it has been shown in both humans and rodents that maternal behaviour can alter offspring phenotypes when they become adults, particularly social behaviour, and some of the molecular changes in offspring have been explored (Liu et al.,

1997; Francis et al., 1999, 2000; Boccia and Pedersen, 2001; Champagne et al.,

2003b, 2003a; Weaver et al., 2004b, 2004a; Priebe et al., 2005; Veenema et al.,

2006, 2007; Murgatroyd et al., 2010; Curley et al., 2010; Gudsnuk and

Champagne, 2011; Curley et al., 2011; Wolf and Cheverud, 2012; Veenema,

2012; Franks et al., 2015). Cross-fostering can be used to sever the connection between maternal genotype and post-natal maternal care and therefore to explore how natural variation in maternal care (i.e. by different genotypes) can alter offspring phenotype (Ressler, 1962; Bester-Meredith and Marler, 2001;

Champagne et al., 2006; Hager et al., 2009; Cox et al., 2013; Peña et al., 2013).

Chapter 1: General introduction

These changes in adult behaviour can potentially have an impact on fitness, and therefore on selection.

However, whilst differences in maternal care between strains have been examined for many years (Ressler, 1962), what has not been well studied are the offspring genes which underlie altered maternal behaviour (offspring to mother

IGE), and how these maternal behaviours then influence offspring phenotype.

Chapters 4-6 will begin to examine this.

1.9 Aims of the PhD

This thesis aims to develop novel analyses in systems-genetics, uncovering more about the genetics of the brain and resultant behavioural phenotypes, and the pathways leading from genes to observed phenotypes.

The first approach to this aim is to use diverse datasets in two different species to identify genes underlying brain morphology (chapter 2) and behavioural phenotypes (chapter 3). I aim to identify novel candidate genes, and plausibly link these to other genes known to influence the phenotype, and therefore, link our candidate to the phenotype itself. This will not only provide understanding of the genetics and physiology of these traits, but also give insight into pathways which may be involved in disease. I aim to present predictions, made in silico, which can then be tested in vitro or in vivo.

The second part aims to use a series of experimental designs to provide novel information about the genetics of maternal effects, IGEs and POEs. All three of these effects are important to understand, for two reasons. Firstly, they provide mechanisms upon which selection can act, beyond selection on the individual due to their own genes. These effects can mean that evolution can act in ways which would not be predicted using simple direct genetic effects.

Secondly, the early life environment, upon which they have a substantial impact,

Chapter 1: General introduction is an important period of development, influencing health and fitness throughout life. I aim to show that maternal effects, IGEs and POEs can all influence parental care, and identify loci and candidate genes, which can then be followed up in further study.

Both sections make use of systems-genetics tools and datasets available online to elucidate the pathway from genotype to phenotype. The final aim of the project is to demonstrate how these can be combined to deliver greater understanding than would be possible with any one tool or one dataset individually.

1.10 Alternative format

The thesis is being presented in the alternative format in accordance with the rules and regulations of the University of Manchester. The results chapters, chapters 2-6, are presented in the form of manuscripts for publication. Each results chapter is preceded by a preface, where details about its publication (if published) and the contributions of the authors are outlined. All chapters have been formatted to the same style, and therefore published manuscripts have had alterations made to ensure the thesis form a cohesive body of work.

Chapter 1: General introduction

1.11 References

Allen, N. D., Logan, K., Lally, G., Drage, D. J., Norris, M. L., and Keverne, E. B. (1995). Distribution of parthenogenetic cells in the mouse brain and their influence on brain development and behavior. Proc. Natl. Acad. Sci. U. S. A. 92, 10782–10786.

Andreux, P. A., Williams, E. G., Koutnikova, H., Houtkooper, R. H., Champy, M.- F., Henry, H., Schoonjans, K., Williams, R. W., and Auwerx, J. (2012). Systems genetics of metabolism: the use of the BXD murine reference panel for multiscalar integration of traits. Cell 150, 1287–1299. doi:10.1016/j.cell.2012.08.012.

Anholt, R. R. H., and Mackay, T. F. C. (2012). Genetics of aggression. Annu. Rev. Genet. 46, 145–164. doi:10.1146/annurev-genet-110711-155514.

Ashbrook, D. G., Delprato, A., Grellmann, C., Klein, M., Wetzel, R., Overall, R. W., and Badea, A. (2014a). Transcript co-variance with Nestin in two mouse genetic reference populations identifies Lef1 as a novel candidate regulator of neural precursor cell proliferation in the adult hippocampus. Front. Neurosci. 8, 418. doi:10.3389/fnins.2014.00418.

Ashbrook, D. G., Williams, R. W., Lu, L., and Hager, R. (2015). A cross-species genetic analysis identifies candidate genes for mouse anxiety and human bipolar disorder. Front. Behav. Neurosci. 9, 171. doi:10.3389/fnbeh.2015.00171.

Ashbrook, D. G., Williams, R. W., Lu, L., Stein, J. L., Hibar, D. P., Nichols, T. E., Medland, S. E., Thompson, P. M., and Hager, R. (2014b). Joint genetic analysis of hippocampal size in mouse and human identifies a novel gene linked to neurodegenerative disease. BMC Genomics 15, 850. doi:10.1186/1471-2164-15-850.

Ashburner, M., Ball, C. A., Blake, J. A., Botstein, D., Butler, H., Cherry, J. M., Davis, A. P., Dolinski, K., Dwight, S. S., Eppig, J. T., et al. (2000). Gene ontology: tool for the unification of biology. The Gene Ontology Consortium. Nat. Genet. 25, 25–29. doi:10.1038/75556.

Ayroles, J. F., Carbone, M. A., Stone, E. A., Jordan, K. W., Lyman, R. F., Magwire, M. M., Rollmann, S. M., Duncan, L. H., Lawrence, F., Anholt, R. R. H., et al. (2009). Systems genetics of complex traits in Drosophila melanogaster. Nat. Genet. 41, 299–307. doi:10.1038/ng.332.

Aziz, R. K., Kansal, R., Abdeltawab, N. F., Rowe, S. L., Su, Y., Carrigan, D., Nooh, M. M., Attia, R. R., Brannen, C., Gardner, L. A., et al. (2007). Susceptibility to severe Streptococcal sepsis: use of a large set of isogenic mouse lines to study genetic and environmental factors. Genes Immun. 8, 404–415. doi:10.1038/sj.gene.6364402.

Babak, T., DeVeale, B., Tsang, E. K., Zhou, Y., Li, X., Smith, K. S., Kukurba, K. R., Zhang, R., Li, J. B., van der Kooy, D., et al. (2015). Genetic conflict reflected in tissue-specific maps of genomic imprinting in human and mouse. Nat. Genet. 47, 544–549. doi:10.1038/ng.3274.

Chapter 1: General introduction

Badcock, C., and Crespi, B. (2006). Imbalanced genomic imprinting in brain development: an evolutionary basis for the aetiology of autism. J. Evol. Biol. 19, 1007–32. doi:10.1111/j.1420-9101.2006.01091.x.

Barker, D. J. P., Eriksson, J. G., Forsén, T., and Osmond, C. (2002). Fetal origins of adult disease: strength of effects and biological basis. Int. J. Epidemiol. 31, 1235–1239. doi:10.1093/ije/31.6.1235.

Barlow, D. P. (1995). Gametic imprinting in mammals. Science 270, 1610–1613. doi:10.1126/science.270.5242.1610.

Barrett, T., Troup, D. B., Wilhite, S. E., Ledoux, P., Evangelista, C., Kim, I. F., Tomashevsky, M., Marshall, K. A., Phillippy, K. H., Sherman, P. M., et al. (2011). NCBI GEO: archive for functional genomics data sets--10 years on. Nucleic Acids Res. 39, D1005–10. doi:10.1093/nar/gkq1184.

Barrett, T., Wilhite, S. E., Ledoux, P., Evangelista, C., Kim, I. F., Tomashevsky, M., Marshall, K. A., Phillippy, K. H., Sherman, P. M., Holko, M., et al. (2013). NCBI GEO: archive for functional genomics data sets--update. Nucleic Acids Res. 41, D991–995. doi:10.1093/nar/gks1193.

Bennett, B. J., Farber, C. R., Orozco, L., Min Kang, H., Ghazalpour, A., Siemers, N., Neubauer, M., Neuhaus, I., Yordanova, R., Guan, B., et al. (2010). A high-resolution association mapping panel for the dissection of complex traits in mice. Genome Res. 20, 281–290. doi:10.1101/gr.099234.109.

Bester-Meredith, J. K., and Marler, C. A. (2001). Vasopressin and Aggression in Cross-Fostered California Mice (Peromyscus californicus) and White-Footed Mice (Peromyscus leucopus). Horm. Behav. 40, 51–64. doi:10.1006/hbeh.2001.1666.

Bijma, P., and Wade, M. J. (2008). The joint effects of kin, multilevel selection and indirect genetic effects on response to genetic selection. J. Evol. Biol. 21, 1175–1188. doi:10.1111/j.1420-9101.2008.01550.x.

Boccia, M. L., and Pedersen, C. A. (2001). Brief vs. long maternal separations in infancy: contrasting relationships with adult maternal behavior and lactation levels of aggression and anxiety. Psychoneuroendocrinology 26, 657–672. doi:10.1016/S0306-4530(01)00019-1.

Bonthuis, P. J., Huang, W.-C., Stacher Hörndli, C. N., Ferris, E., Cheng, T., and Gregg, C. (2015). Noncanonical genomic imprinting effects in offspring. Cell Rep. 12, 979–991. doi:10.1016/j.celrep.2015.07.017.

Boughter, J. D., Mulligan, M. K., St. John, S. J., Tokita, K., Lu, L., Heck, D. H., and Williams, R. W. (2012). Genetic control of a central pattern generator: Rhythmic oromotor movement in mice is controlled by a major locus near atp1a2. PLoS One 7. doi:10.1371/journal.pone.0038169.

Brem, R. B., and Kruglyak, L. (2005). The landscape of genetic complexity across 5,700 gene expression traits in yeast. Proc. Natl. Acad. Sci. U. S. A. 102, 1572–1577. doi:10.1073/pnas.0408709102.

Chapter 1: General introduction

Brem, R. B., Yvert, G., Clinton, R., and Kruglyak, L. (2002). Genetic dissection of transcriptional regulation in budding yeast. Science 296, 752–755. doi:10.1126/science.1069516.

Bucher, T. L., Ryan, M. J., and Bartholomew, G. A. (1982). Oxygen consumption during resting, calling, and nest building in the frog Physalaemus Pustulosus. Physiol. Zool. 55, 10–22. doi:10.2307/30158439.

Caldji, C., Diorio, J., Anisman, H., and Meaney, M. J. (2004). Maternal behavior regulates benzodiazepine/GABAA receptor subunit expression in brain regions associated with fear in BALB/c and C57BL/6 mice. Neuropsychopharmacology 29, 1344–1352. doi:10.1038/sj.npp.1300436.

Caldji, C., Diorio, J., and Meaney, M. J. (2000). Variations in maternal care in infancy regulate the development of stress reactivity. Biol. Psychiatry 48, 1164–1174. doi:10.1016/S0006-3223(00)01084-2.

Cameron, N. M., Fish, E. W., and Meaney, M. J. (2008). Maternal influences on the sexual behavior and reproductive success of the female rat. Horm. Behav. 54, 178–184. doi:10.1016/j.yhbeh.2008.02.013.

Carlier, M., Roubertoux, P., and Cohen-Salmon, C. (1982). Differences in patterns of pup care in Mus musculus domesticus I—Comparisons between eleven inbred strains. Behav. Neural Biol. 35, 205–210. doi:10.1016/S0163- 1047(82)91213-4.

Carola, V., Frazzetto, G., and Gross, C. (2006). Identifying interactions between genes and early environment in the mouse. Genes. Brain. Behav. 5, 189– 199. doi:10.1111/j.1601-183X.2005.00152.x.

Casellas, J., Farber, C. R., Gularte, R. J., Haus, K. A., Warden, C. H., and Medrano, J. F. (2009). Evidence of maternal QTL affecting growth and obesity in adult mice. Mamm. Genome 20, 269–280. doi:10.1007/s00335- 009-9182-9.

Champagne, F. A., Curley, J. P., Keverne, E. B., and Bateson, P. P. G. (2007). Natural variations in postpartum maternal care in inbred and outbred mice. Physiol. Behav. 91, 325–334. doi:10.1016/j.physbeh.2007.03.014.

Champagne, F. A., Francis, D. D., Mar, A., and Meaney, M. J. (2003a). Variations in maternal care in the rat as a mediating influence for the effects of environment on development. Physiol. Behav. 79, 359–371. doi:10.1016/S0031-9384(03)00149-5.

Champagne, F. A., Weaver, I. C. G., Diorio, J., Dymov, S., Szyf, M., and Meaney, M. J. (2006). Maternal care associated with methylation of the estrogen receptor-α1b promoter and estrogen receptor-α expression in the medial preoptic area of female offspring. Endocrinology 147, 2909–2915. doi:10.1210/en.2005-1119.

Champagne, F. A., Weaver, I. C. G., Diorio, J., Sharma, S., and Meaney, M. J. (2003b). Natural variations in maternal care are associated with estrogen

Chapter 1: General introduction

receptor alpha expression and estrogen sensitivity in the medial preoptic area. Endocrinology 144, 4720–4724. doi:10.1210/en.2003-0564.

Chesler, E. J., Lu, L., Wang, J., Williams, R. W., and Manly, K. F. (2004). WebQTL: rapid exploratory analysis of gene expression and genetic networks for brain and behavior. Nat. Neurosci. 7, 485–6. doi:10.1038/nn0504-485.

Cheverud, J. M., Hager, R., Roseman, C., Fawcett, G., Wang, B., and Wolf, J. B. (2008). Genomic imprinting effects on adult body composition in mice. Proc. Natl. Acad. Sci. U.S.A. 105, 4253–4258. doi:10.1073/pnas.0706562105.

Chourbaji, S., Hoyer, C., Richter, S. H., Brandwein, C., Pfeiffer, N., Vogt, M. A., Vollmayr, B., and Gass, P. (2011). Differences in mouse maternal care behavior - is there a genetic impact of the glucocorticoid receptor? PLoS One 6, e19218. doi:10.1371/journal.pone.0019218.

Civelek, M., and Lusis, A. J. (2014). Systems genetics approaches to understand complex traits. Nat. Rev. Genet. 15, 34–48. doi:10.1038/nrg3575.

Cox, K. H., and Rissman, E. F. (2011). Sex differences in juvenile mouse social behavior are influenced by sex chromosomes and social context. Genes, Brain Behav. 10, 465–472. doi:10.1111/j.1601-183X.2011.00688.x.

Cox, K. H., So, N. L. T., and Rissman, E. F. (2013). Foster dams rear fighters: Strain-specific effects of within-strain fostering on aggressive behavior in male mice. PLoS One 8, e75037. doi:10.1371/journal.pone.0075037.

Crespi, B. (2008). Genomic imprinting in the development and evolution of psychotic spectrum conditions. Biol. Rev. 83, 441–493. doi:10.1111/j.1469- 185X.2008.00050.x.

Crowley, J. J., Zhabotynsky, V., Sun, W., Huang, S., Pakatci, I. K., Kim, Y., Wang, J. R., Morgan, A. P., Calaway, J. D., Aylor, D. L., et al. (2015). Analyses of allele-specific gene expression in highly divergent mouse crosses identifies pervasive allelic imbalance. Nat. Genet. doi:10.1038/ng.3222.

Curley, J. P. (2011). Is there a genomically imprinted social brain? Bioessays 33, 662–8. doi:10.1002/bies.201100060.

Curley, J. P., Jensen, C. L., Mashoodh, R., and Champagne, F. A. (2011). Social influences on neurobiology and behavior: Epigenetic effects during development. Psychoneuroendocrinology 36, 352–371. doi:10.1016/j.psyneuen.2010.06.005.

Curley, J. P., Rock, V., Moynihan, A. M., Bateson, P., Keverne, E. B., and Champagne, F. A. (2010). Developmental shifts in the behavioral phenotypes of inbred mice: The role of postnatal and juvenile social experiences. Behav. Genet. 40, 220–232. doi:10.1007/s10519-010-9334-4.

Chapter 1: General introduction

Dent, C. L., and Isles, A. R. (2014). Brain-expressed imprinted genes and adult behaviour: the example of Nesp and Grb10. Mamm. Genome 25, 87–93. doi:10.1007/s00335-013-9472-0.

Descamps, S., Gilchrist, H. G., Bety, J., Buttler, E. I., and Forbes, M. R. (2009). Costs of reproduction in a long-lived bird: large clutch size is associated with low survival in the presence of a highly virulent disease. Biol. Lett. 5, 278– 281. doi:10.1098/rsbl.2008.0704.

DeVeale, B., van der Kooy, D., and Babak, T. (2012). Critical evaluation of imprinted gene expression by RNA-Seq: a new perspective. PLoS Genet. 8, e1002600. doi:10.1371/journal.pgen.1002600.

Doss, S., Schadt, E. E., Drake, T. A., and Lusis, A. J. (2005). Cis-acting expression quantitative trait loci in mice. Genome Res. 15, 681–691. doi:10.1101/gr.3216905.

Doty, R. L. (1974). A cry for the liberation of the female rodent: Courtship and copulation in rodentia. Psychol. Bull. 81, 159–172. doi:10.1037/h0035971.

Downing, C., Johnson, T. E., Larson, C., Leakey, T. I., Siegfried, R. N., Rafferty, T. M., and Cooney, C. A. (2011). Subtle decreases in DNA methylation and gene expression at the mouse Igf2 locus following prenatal alcohol exposure: effects of a methyl-supplemented diet. Alcohol 45, 65–71. doi:10.1016/j.alcohol.2010.07.006.

Edgar, R., Domrachev, M., and Lash, A. E. (2002). Gene Expression Omnibus: NCBI gene expression and hybridization array data repository. Nucleic Acids Res. 30, 207–210. doi:10.1093/nar/30.1.207.

Edwards, C. A., and Ferguson-Smith, A. C. (2007). Mechanisms regulating imprinted genes in clusters. Curr. Opin. Cell Biol. 19, 281–9. doi:10.1016/j.ceb.2007.04.013.

Eppig, J. T., Blake, J. A., Bult, C. J., Kadin, J. A., and Richardson, J. E. (2015). The Mouse Genome Database (MGD): facilitating mouse as a model for human biology and disease. Nucleic Acids Res. 43, D726–D736. doi:10.1093/nar/gku967.

Falconer, D. S., and Mackay, T. F. C. (1996). Introduction to Quantitative Genetics. 4th ed. Burnt Mill, England: Longman.

Fenn, J., Mann, M., Meng, C., Wong, S., and Whitehouse, C. (1989). Electrospray ionization for mass spectrometry of large biomolecules. Science 246, 64–71. doi:10.1126/science.2675315.

Ferguson, A. R., Nielson, J. L., Cragin, M. H., Bandrowski, A. E., and Martone, M. E. (2014). Big data from small data: data-sharing in the “long tail” of neuroscience. Nat. Neurosci. 17, 1442–1447. doi:10.1038/nn.3838.

Foss, E. J., Radulovic, D., Shaffer, S. A., Ruderfer, D. M., Bedalov, A., Goodlett, D. R., and Kruglyak, L. (2007). Genetic basis of proteome variation in yeast. Nat. Genet. 39, 1369–1375. doi:10.1038/ng.2007.22.

Chapter 1: General introduction

Fowden, A. L., Coan, P. M., Angiolini, E., Burton, G. J., and Constancia, M. (2011). Imprinted genes and the epigenetic regulation of placental phenotype. Prog. Biophys. Mol. Biol. 106, 281–288. doi:10.1016/j.pbiomolbio.2010.11.005.

Francis, D. D., Champagne, F. A., Liu, D., and Meaney, M. J. (1999). Maternal care, gene expression, and the development of individual differences in stress reactivity. Ann. N. Y. Acad. Sci. 896, 66–84. doi:10.1111/j.1749- 6632.1999.tb08106.x.

Francis, D. D., Champagne, F. C., and Meaney, M. J. (2000). Variations in maternal behaviour are associated with differences in oxytocin receptor levels in the rat. J. Neuroendocrinol. 12, 1145–1148. doi:10.1046/j.1365- 2826.2000.00599.x.

Franks, B., Champagne, F. A., and Curley, J. P. (2015). Postnatal maternal care predicts divergent weaning strategies and the development of social behavior. Dev. Psychobiol., [Epub ahead of print]. doi:10.1002/dev.21326.

Frost, J. M., and Moore, G. E. (2010). The importance of imprinting in the human placenta. PLoS Genet. 6, e1001015. doi:10.1371/journal.pgen.1001015.

Gao, H., Liu, Y., Zhang, T., Yang, R., and Prows, D. R. (2013). Parametric proportional hazards model for mapping genomic imprinting of survival traits. J. Appl. Genet. 54, 79–88. doi:10.1007/s13353-012-0120-2.

Garfield, A. S., Cowley, M., Smith, F. M., Moorwood, K., Stewart-Cox, J. E., Gilroy, K., Baker, S., Xia, J., Dalley, J. W., Hurst, L. D., et al. (2011). Distinct physiological and behavioural functions for parental alleles of imprinted Grb10. Nature 469, 534–538. doi:10.1038/nature09651.

Geldermann, H. (1975). Investigations on inheritance of quantitative characters in animals by gene markers I. Methods. Theor. Appl. Genet. 46, 319–330. doi:10.1007/BF00281673.

Gelman, R., Watson, A., Bronson, R., and Yunis, E. (1988). Murine chromosomal regions correlated with longevity. Genetics 118, 693–704.

Gene Ontology Consortium (2015). Gene Ontology Consortium: going forward. Nucleic Acids Res. 43, D1049–D1056. doi:10.1093/nar/gku1179.

Ghazalpour, A., Bennett, B., Petyuk, V. A., Orozco, L., Hagopian, R., Mungrue, I. N., Farber, C. R., Sinsheimer, J., Kang, H. M., Furlotte, N., et al. (2011). Comparative analysis of proteome and transcriptome variation in mouse. PLoS Genet. 7, e1001393. doi:10.1371/journal.pgen.1001393.

Gini, B., and Hager, R. (2012). Recombinant inbred systems can advance research in behavioral ecology. Front. Genet. 3, 198. doi:10.3389/fgene.2012.00198.

Goldstein, D. B. (2009). Common genetic variation and human traits. N. Engl. J. Med. 360, 1696–1698. doi:10.1056/NEJMp0806284.

Chapter 1: General introduction

Gomez-Marin, A., Paton, J. J., Kampff, A. R., Costa, R. M., and Mainen, Z. F. (2014). Big behavioral data: psychology, ethology and the foundations of neuroscience. Nat. Neurosci. 17, 1455–1462. doi:10.1038/nn.3812.

Gouldsborough, I., Black, V., Johnson, I. T., and Ashton, N. (1998). Maternal nursing behaviour and the delivery of milk to the neonatal spontaneously hypertensive rat. Acta Physiol. Scand. 162, 107–114. doi:10.1046/j.1365- 201X.1998.0273f.x.

Gratten, J., Wray, N. R., Keller, M. C., and Visscher, P. M. (2014). Large-scale genomics unveils the genetic architecture of psychiatric disorders. Nat. Neurosci. 17, 782–790. doi:10.1038/nn.3708.

Gregg, C., Zhang, J., Weissbourd, B., Luo, S., Schroth, G. P., Haig, D., and Dulac, C. (2010). High-resolution analysis of parent-of-origin allelic expression in the mouse brain. Science 329, 643–648. doi:10.1126/science.1190830.

Gudsnuk, K. M. A., and Champagne, F. A. (2011). Epigenetic effects of early developmental experiences. Clin. Perinatol. 38, 703–717. doi:10.1016/j.clp.2011.08.005.

Guilmatre, A., and Sharp, A. J. (2012). Parent of origin effects. Clin. Genet. 81, 201–9. doi:10.1111/j.1399-0004.2011.01790.x.

Hager, R., Cheverud, J. M., and Wolf, J. B. (2009). Change in maternal environment induced by cross-fostering alters genetic and epigenetic effects on complex traits in mice. Proc. Biol. Sci. 276, 2949–2954. doi:10.1098/rspb.2009.0515.

Hager, R., Lu, L., Rosen, G. D., and Williams, R. W. (2012). Genetic architecture supports mosaic brain evolution and independent brain-body size regulation. Nat. Commun. 3, 1079. doi:10.1038/ncomms2086.

Hamilton, W. D. (1964). The genetical evolution of social behaviour. I. J. Theor. Biol. 7, 1–16. doi:10.1016/0022-5193(64)90038-4.

Harris, T. W., Antoshechkin, I., Bieri, T., Blasiar, D., Chan, J., Chen, W. J., De La Cruz, N., Davis, P., Duesbury, M., Fang, R., et al. (2010). WormBase: a comprehensive resource for nematode research. Nucleic Acids Res. 38, D463–D467. doi:10.1093/nar/gkp952.

Hayes, K. S., Hager, R., and Grencis, R. K. (2014). Sex-dependent genetic effects on immune responses to a parasitic nematode. BMC Genomics 15, 193. doi:10.1186/1471-2164-15-193.

Hitzemann, R., Bottomly, D., Darakjian, P., Walter, N., Iancu, O., Searles, R., Wilmot, B., and McWeeney, S. (2013). Genes, behavior and next- generation RNA sequencing. Genes Brain Behav. 12, 1–12. doi:10.1111/gbb.12007.

Houtkooper, R. H., Mouchiroud, L., Ryu, D., Moullan, N., Katsyuba, E., Knott, G., Williams, R. W., and Auwerx, J. (2013). Mitonuclear protein imbalance as a

Chapter 1: General introduction

conserved longevity mechanism. Nature 497, 451–457. doi:10.1038/nature12188.

Hughes, A. L. (1989). Interaction between strains in the social relations of inbred mice. Behav. Genet. 19, 685–700. doi:10.1007/BF01066031.

Kacem, S., and Feil, R. (2009). Chromatin mechanisms in genomic imprinting. Mamm. Genome 20, 544–556. doi:10.1007/s00335-009-9223-4.

Kanehisa, M., and Goto, S. (2000). KEGG: kyoto encyclopedia of genes and genomes. Nucleic Acids Res. 28, 27–30.

Kanehisa, M., Goto, S., Sato, Y., Furumichi, M., and Tanabe, M. (2012). KEGG for integration and interpretation of large-scale molecular data sets. Nucleic Acids Res. 40, D109–14. doi:10.1093/nar/gkr988.

Kato, M. V, Ikawa, Y., Hayashizaki, Y., and Shibata, H. (1998). Paternal imprinting of mouse serotonin receptor 2A gene Htr2 in embryonic eye: a conserved imprinting regulation on the RB/Rb locus. Genomics 47, 146–8. doi:10.1006/geno.1997.5089.

Kent, J. W., Sugnet, C. W., Furey, T. S., Roskin, K. M., Pringle, T. H., Zahler, A. M., and Haussler, D. (2002). The human genome browser at UCSC. Genome Res. 12, 996–1006. doi:10.1101/gr.229102.

Keverne, E. B. (2014). Significance of epigenetics for understanding brain development, brain evolution and behaviour. Neuroscience 264, 207–217. doi:10.1016/j.neuroscience.2012.11.030.

Keverne, E. B., and Curley, J. P. (2008). Epigenetics, brain evolution and behaviour. Front. Neuroendocr. 29, 398–412. doi:10.1016/j.yfrne.2008.03.001.

Keverne, E. B., Fundele, R., Narasimha, M., Barton, S. C., and Surani, M. A. (1996). Genomic imprinting and the differential roles of parental genomes in brain development. Brain Res. Dev. Brain Res. 92, 91–100.

Khatib, H. (2007). Is it genomic imprinting or preferential expression? BioEssays 29, 1022–1028. doi:10.1002/bies.20637.

Kirkpatrick, M., and Lande, R. (1989). The evolution of maternal characters. Evolution 43, 485. doi:10.2307/2409054.

Klug, H., and Bonsall, M. B. (2014). What are the benefits of parental care? The importance of parental effects on developmental rate. Ecol. Evol. 4, 2330– 2351. doi:10.1002/ece3.1083.

Kopsida, E., Mikaelsson, M. A., and Davies, W. (2011). The role of imprinted genes in mediating susceptibility to neuropsychiatric disorders. Horm. Behav. 59, 375–82. doi:10.1016/j.yhbeh.2010.04.005.

Koutnikova, H., Laakso, M., Lu, L., Combe, R., Paananen, J., Kuulasmaa, T., Kuusisto, J., Häring, H.-U., Hansen, T., Pedersen, O., et al. (2009).

Chapter 1: General introduction

Identification of the UBP1 locus as a critical blood pressure determinant using a combination of mouse and human genetics. PLoS Genet. 5, e1000591. doi:10.1371/journal.pgen.1000591.

Lander, E. S. (2011). Initial impact of the sequencing of the human genome. Nature 470, 187–197. doi:10.1038/nature09792.

Lappalainen, T., Sammeth, M., Friedländer, M. R., ‘t Hoen, P. A. C., Monlong, J., Rivas, M. A., Gonzàlez-Porta, M., Kurbatova, N., Griebel, T., Ferreira, P. G., et al. (2013). Transcriptome and genome sequencing uncovers functional variation in humans. Nature 501, 506–511. doi:10.1038/nature12531.

Lein, E. S., Hawrylycz, M. J., Ao, N., Ayres, M., Bensinger, A., Bernard, A., Boe, A. F., Boguski, M. S., Brockway, K. S., Byrnes, E. J., et al. (2007). Genome- wide atlas of gene expression in the adult mouse brain. Nature 445, 168– 176. doi:10.1038/nature05453.

Lim, A. L., and Ferguson-Smith, A. C. (2010). Genomic imprinting effects in a compromised in utero environment: implications for a healthy pregnancy. Semin. Cell Dev. Biol. 21, 201–8. doi:10.1016/j.semcdb.2009.10.008.

Liu, D., Diorio, J., Tannenbaum, B., Caldji, C., Francis, D., Freedman, A., Sharma, S., Pearson, D., Plotsky, P. M., and Meaney, M. J. (1997). Maternal care, hippocampal glucocorticoid receptors, and hypothalamic- pituitary-adrenal responses to stress. Science 277, 1659–1662. doi:10.1126/science.277.5332.1659.

Liu, S., Lin, L., Jiang, P., Wang, D., and Xing, Y. (2011). A comparison of RNA- Seq and high-density exon array for detecting differential gene expression between closely related species. Nucleic Acids Res. 39, 578–588. doi:10.1093/nar/gkq817.

Lukas, M., and Wöhr, M. (2015). Endogenous vasopressin, innate anxiety, and the emission of pro-social 50-kHz ultrasonic vocalizations during social play behavior in juvenile rats. Psychoneuroendocrinology 56, 35–44. doi:10.1016/j.psyneuen.2015.03.005.

Lunde, A., Melve, K. K., Gjessing, H. K., Skjaerven, R., and Irgens, L. M. (2007). Genetic and environmental influences on birth weight, birth length, head circumference, and gestational age by use of population-based parent- offspring data. Am. J. Epidemiol. 165, 734–741. doi:10.1093/aje/kwk107.

Mackay, D. J. G., and Temple, I. K. (2010). Transient neonatal diabetes mellitus type 1. Am. J. Med. Genet. C. Semin. Med. Genet. 154C, 335–42. doi:10.1002/ajmg.c.30272.

Marioni, J. C., Mason, C. E., Mane, S. M., Stephens, M., and Gilad, Y. (2008). RNA-seq: An assessment of technical reproducibility and comparison with gene expression arrays. Genome Res. 18, 1509–1517. doi:10.1101/gr.079558.108.

Chapter 1: General introduction

McCarroll, S. a, and Hyman, S. E. (2013). Progress in the genetics of polygenic brain disorders: significant new challenges for neurobiology. Neuron 80, 578–587. doi:10.1016/j.neuron.2013.10.046.

McGlothlin, J. W., Moore, A. J., Wolf, J. B., and Brodie III, E. D. (2010). Interacting phenotypes and the evolutionary process. III. Social evolution. Evolution 64, 2558–2574. doi:10.1111/j.1558-5646.2010.01012.x.

McIver, A. H., and Jeffrey, W. E. (1967). Strain differences in maternal behavior in rats. Behaviour 28, 210–216. doi:10.1163/156853967X00244.

Medina, K. L. (1993). Suppression of B lymphopoiesis during normal pregnancy. J. Exp. Med. 178, 1507–1515. doi:10.1084/jem.178.5.1507.

Melzer, D., Perry, J. R. B., Hernandez, D., Corsi, A.-M., Stevens, K., Rafferty, I., Lauretani, F., Murray, A., Gibbs, J. R., Paolisso, G., et al. (2008). A genome-wide association study identifies protein quantitative trait loci (pQTLs). PLoS Genet. 4, e1000072. doi:10.1371/journal.pgen.1000072.

Miles, C. M., and Wayne, M. (2008). Quantitative trait locus (QTL) analysis. Nat. Educ. 1, 208.

Miyairi, I., Tatireddigari, V. R. R. A., Mahdi, O. S., Rose, L. A., Belland, R. J., Lu, L., Williams, R. W., and Byrne, G. I. (2007). The p47 GTPases Iigp2 and Irgb10 regulate innate immunity and inflammation to murine Chlamydia psittaci infection. J. Immunol. 179, 1814–1824. doi:179/3/1814 [pii].

Miyairi, I., Ziebarth, J., Laxton, J. D., Wang, X., van Rooijen, N., Williams, R. W., Lu, L., Byrne, G. I., and Cui, Y. (2012). Host genetics and Chlamydia disease: prediction and validation of disease severity mechanisms. PLoS One 7, e33781. doi:10.1371/journal.pone.0033781.

Moore, A. J., Brodie III, E. D., and Wolf, J. B. (1997). Interacting phenotypes and the evolutionary process: I. Direct and indirect genetic effects of social interactions. Evolution 51, 1352–1362. doi:10.2307/2411187.

Mozhui, K., Ciobanu, D. C., Schikorski, T., Wang, X., Lu, L., and Williams, R. W. (2008). Dissection of a QTL hotspot on mouse distal chromosome 1 that modulates neurobehavioral phenotypes and gene expression. PLoS Genet. 4, e1000260. doi:10.1371/journal.pgen.1000260.

Mozhui, K., Wang, X., Chen, J., Mulligan, M. K., Li, Z., Ingles, J., Chen, X., Lu, L., and Williams, R. W. (2011). Genetic regulation of Nrxn1 expression: an integrative cross-species analysis of schizophrenia candidate genes. Transl. Psychiatry 1, e25. doi:10.1038/tp.2011.24.

Mulligan, M. K., Wang, X., Adler, A. L., Mozhui, K., Lu, L., and Williams, R. W. (2012). Complex control of GABA(A) receptor subunit mRNA expression: variation, covariation, and genetic regulation. PLoS One 7, e34586. doi:10.1371/journal.pone.0034586.

Murgatroyd, C., Patchev, A. V, Wu, Y., Micale, V., Bockmühl, Y., Fischer, D., Holsboer, F., Wotjak, C. T., Almeida, O. F. X., and Spengler, D. (2010).

Chapter 1: General introduction

Dynamic DNA methylation programs persistent adverse effects of early-life stress. Nat. Neurosci. 13, 649–649. doi:10.1038/nn0510-649e.

Myers, M. M., Brunelli, S. A., Shair, H. N., Squire, J. M., and Hofer, M. A. (1989). Relationships between maternal behavior of SHR and WKY dams and adult blood pressures of cross-fostered F1 pups. Dev. Psychobiol. 22, 55–67. doi:10.1002/dev.420220105.

Van Nas, A., Ingram-Drake, L., Sinsheimer, J. S., Wang, S. S., Schadt, E. E., Drake, T., and Lusis, A. J. (2010). Expression quantitative trait loci: Replication, tissue- and sex-specificity in mice. Genetics 185, 1059–1068. doi:10.1534/genetics.110.116087.

Neunuebel, J. P., Taylor, A. L., Arthur, B. J., and Egnor, S. R. (2015). Female mice ultrasonically interact with males during courtship displays. Elife 4. doi:10.7554/eLife.06203.

Nordling, D., Andersson, M., Zohari, S., and Lars, G. (1998). Reproductive effort reduces specific immune response and parasite resistance. Proc. R. Soc. B Biol. Sci. 265, 1291–1298. doi:10.1098/rspb.1998.0432.

Overall, R. W., Kempermann, G., Peirce, J., Lu, L., Goldowitz, D., Gage, F. H., Goodwin, S., Smit, A. B., Airey, D. C., Rosen, G. D., et al. (2009). Genetics of the hippocampal transcriptome in mouse: a systematic survey and online neurogenomics resource. Front. Neurosci. 3, 55. doi:10.3389/neuro.15.003.2009.

Paus, T., Bernard, M., Chakravarty, M. M., Davey Smith, G., Gillis, J., Lourdusamy, A., Melka, M. G., Leonard, G., Pavlidis, P., Perron, M., et al. (2012). KCTD8 gene and brain growth in adverse intrauterine environment: a genome-wide association study. Cereb. Cortex 22, 2634–42. doi:10.1093/cercor/bhr350.

Peirce, J. L., Lu, L., Gu, J., Silver, L. M., and Williams, R. W. (2004). A new set of BXD recombinant inbred lines from advanced intercross populations in mice. BMC Genet. 5, 7. doi:10.1186/1471-2156-5-7.

Peña, C. J., Neugut, Y. D., and Champagne, F. A. (2013). Developmental timing of the effects of maternal care on gene expression and epigenetic regulation of hormone receptor levels in female rats. Endocrinology 154, 4340–4351. doi:10.1210/en.2013-1595.

Perez, J. D., Rubinstein, N. D., Fernandez, D. E., Santoro, S. W., Needleman, L. A., Ho-Shing, O., Choi, J. J., Zirlinger, M., Chen, S.-K., Liu, J. S., et al. (2015). Quantitative and functional interrogation of parent-of-origin allelic expression biases in the brain. Elife 4. doi:10.7554/eLife.07860.

Peripato, A. C., De Brito, R. A., Vaughn, T. T., Pletscher, L. S., Matioli, S. R., and Cheverud, J. M. (2002). Quantitative trait loci for maternal performance for offspring survival in mice. Genetics 162, 1341–1353.

Peripato, A. C., and Cheverud, J. M. (2002). Genetic influences on maternal care. Am. Nat. 160 Suppl, S173–S185. doi:10.1086/342900.

Chapter 1: General introduction

Philip, V. M., Duvvuru, S., Gomero, B., Ansah, T. A., Blaha, C. D., Cook, M. N., Hamre, K. M., Lariviere, W. R., Matthews, D. B., Mittleman, G., et al. (2010). High-throughput behavioral phenotyping in the expanded panel of BXD recombinant inbred strains. Genes. Brain. Behav. 9, 129–159. doi:10.1111/j.1601-183X.2009.00540.x.

Pollard, D. A. (2012). Design and construction of recombinant inbred lines. Methods Mol. Biol. 871, 31–39. doi:10.1007/978-1-61779-785-9_3.

Prakash, P., Merali, Z., Kolajova, M., Tannenbaum, B. M., and Anisman, H. (2006). Maternal factors and monoamine changes in stress-resilient and susceptible mice: cross-fostering effects. Brain Res. 1111, 122–133. doi:10.1016/j.brainres.2006.06.089.

Priebe, K., Brake, W. G., Romeo, R. D., Sisti, H. M., Mueller, A., McEwen, B. S., and Francis, D. D. (2005). Maternal influences on adult stress and anxiety- like behavior in C57BL/6J and BALB/cJ mice: A cross-fostering study. Dev. Psychobiol. 47, 398–407. doi:10.1002/dev.20098.

Raad, M., El Tal, T., Gul, R., Mondello, S., Zhang, Z., Boustany, R.-M., Guingab, J., Wang, K. K., and Kobeissy, F. (2012). Neuroproteomics approach and neurosystems biology analysis: ROCK inhibitors as promising therapeutic targets in neurodegeneration and neurotrauma. Electrophoresis 33, 3659– 68. doi:10.1002/elps.201200470.

Renfree, M. B., Suzuki, S., and Kaneko-Ishino, T. (2013). The origin and evolution of genomic imprinting and viviparity in mammals. Philos. Trans. R. Soc. Lond. B. Biol. Sci. 368, 20120151. doi:10.1098/rstb.2012.0151.

Ressler, R. H. (1962). Parental handling in two strains of mice reared by foster parents. Science 137, 129–130. doi:10.1126/science.137.3524.129.

Rulten, S. L., Ripley, T. L., Hunt, C. L., Stephens, D. N., and Mayne, L. V (2006). Sp1 and NFkappaB pathways are regulated in brain in response to acute and chronic ethanol. Genes Brain Behav. 5, 257–73. doi:10.1111/j.1601- 183X.2005.00157.x.

Schena, M., Shalon, D., Davis, R. W., and Brown, P. O. (1995). Quantitative monitoring of gene expression patterns with a complementary DNA microarray. Science 270, 467–470. doi:10.1126/science.270.5235.467.

Seecharan, D. J., Kulkarni, A. L., Lu, L., Rosen, G. D., and Williams, R. W. (2003). Genetic control of interconnected neuronal populations in the mouse primary visual system. J. Neurosci. 23, 11178–11188.

Shibata, H., Yoshino, K., Muramatsu, M., Plass, C., Chapman, V. M., and Hayashizaki, Y. (1995). The use of restriction landmark genomic scanning to scan the mouse genome for endogenous loci with imprinted patterns of methylation. Electrophoresis 16, 210–7.

Smiseth, P. T., Kölliker, M., and Royle, N. J. (2012). “What is parental care?,” in The evolution of parental care, eds. N. J. Royle, P. T. Smiseth, and M.

Chapter 1: General introduction

Kölliker (Oxford, UK: Oxford University Press), 1–18. doi:10.1093/acprof:oso/9780199692576.001.0001.

Smith, L., and Hood, L. (1987). Mapping and sequencing the human genome: How to proceed. Bio/Technology 5, 933–939. doi:10.1038/nbt0987-933.

Strogantsev, R., and Ferguson-Smith, A. C. (2012). Proteins involved in establishment and maintenance of imprinted methylation marks. Brief. Funct. Genomics 11, 227–239. doi:10.1093/bfgp/els018.

Tarantino, L. M., Sullivan, P. F., and Meltzer-Brody, S. (2011). Using animal models to disentangle the role of genetic, epigenetic, and environmental influences on behavioral outcomes associated with maternal anxiety and depression. Front. Psychiatry 2, 44. doi:10.3389/fpsyt.2011.00044.

Taylor, B. A., Heiniger, H. J., and Meier, H. (1973). Genetic analysis of resistance to cadmium-induced testicular damage in mice. Proc. Soc. Exp. Biol. Med. 143, 629–633.

Taylor, B. A., Wnek, C., Kotlus, B. S., Roemer, N., MacTaggart, T., and Phillips, S. J. (1999). Genotyping new BXD recombinant inbred mouse strains and comparison of BXD and consensus maps. Mamm. Genome 10, 335–348. doi:10.1007/s003359900998.

Trivers, R. L. (1974). Parent-offspring conflict. Am. Zool. 14, 249–264. doi:10.1093/icb/14.1.249.

Unoki, H., Takahashi, A., Kawaguchi, T., Hara, K., Horikoshi, M., Andersen, G., Ng, D. P. K., Holmkvist, J., Borch-Johnsen, K., Jørgensen, T., et al. (2008). SNPs in KCNQ1 are associated with susceptibility to type 2 diabetes in East Asian and European populations. Nat. Genet. 40, 1098–102. doi:10.1038/ng.208.

Veenema, A. H. (2012). Toward understanding how early-life social experiences alter oxytocin- and vasopressin-regulated social behaviors. Horm. Behav. 61, 304–312. doi:10.1016/j.yhbeh.2011.12.002.

Veenema, A. H., Blume, A., Niederle, D., Buwalda, B., and Neumann, I. D. (2006). Effects of early life stress on adult male aggression and hypothalamic vasopressin and serotonin. Eur. J. Neurosci. 24, 1711–1720. doi:10.1111/j.1460-9568.2006.05045.x.

Veenema, A. H., Bredewold, R., and Neumann, I. D. (2007). Opposite effects of maternal separation on intermale and maternal aggression in C57BL/6 mice: Link to hypothalamic vasopressin and oxytocin immunoreactivity. Psychoneuroendocrinology 32, 437–450. doi:10.1016/j.psyneuen.2007.02.008.

Wallace, C., Smyth, D. J., Maisuria-Armer, M., Walker, N. M., Todd, J. a, and Clayton, D. G. (2010). The imprinted DLK1-MEG3 gene region on chromosome 14q32.2 alters susceptibility to type 1 diabetes. Nat. Genet. 42, 68–71. doi:10.1038/ng.493.

Chapter 1: General introduction

Wang, J., Duncan, D., Shi, Z., and Zhang, B. (2013). WEB-based GEne SeT AnaLysis Toolkit (WebGestalt): update 2013. Nucleic Acids Res. 41, W77– 83. doi:10.1093/nar/gkt439.

Wang, J., Williams, R. W., and Manly, K. F. (2003). WebQTL: web-based complex trait analysis. Neuroinformatics 1, 299–308. doi:10.1385/NI:1:4:299.

Wasinger, V. C., Cordwell, S. J., Cerpa-Poljak, A., Yan, J. X., Gooley, A. A., Wilkins, M. R., Duncan, M. W., Harris, R., Williams, K. L., and Humphery- Smith, I. (1995). Progress with gene-product mapping of the Mollicutes:Mycoplasma genitalium. Electrophoresis 16, 1090–1094. doi:10.1002/elps.11501601185.

Weaver, I. C. G., Cervoni, N., Champagne, F. A., D’Alessio, A. C., Sharma, S., Seckl, J. R., Dymov, S., Szyf, M., and Meaney, M. J. (2004a). Epigenetic programming by maternal behavior. Nat. Neurosci. 7, 847–854. doi:10.1038/nn1276.

Weaver, I. C. G., Diorio, J., Seckl, J. R., Szyf, M., and Meaney, M. J. (2004b). Early environmental regulation of hippocampal glucocorticoid receptor gene expression: Characterization of intracellular mediators and potential genomic target sites. Ann. N. Y. Acad. Sci. 1024, 182–212. doi:10.1196/annals.1321.099.

Weber, J. N., Peterson, B. K., and Hoekstra, H. E. (2013). Discrete genetic modules are responsible for complex burrow evolution in Peromyscus mice. Nature 493, 402–405. doi:10.1038/nature11816.

Wells, J. C. K. (2007). The thrifty phenotype as an adaptive maternal effect. Biol. Rev. 82, 143–172. doi:10.1111/j.1469-185X.2006.00007.x.

Whitney, G., Coble, J. R., Stockton, M. D., and Tilson, E. F. (1973). Ultrasonic emissions: Do they facilitate courtship of mice? J. Comp. Physiol. Psychol. 84, 445–452. doi:10.1037/h0034899.

Williams, E. G., and Auwerx, J. (2015). The convergence of systems and reductionist approaches in complex trait analysis. Cell 162, 23–32. doi:10.1016/j.cell.2015.06.024.

Williams, E. G., Mouchiroud, L., Frochaux, M., Pandey, A., Andreux, P. a, Deplancke, B., and Auwerx, J. (2014). An evolutionarily conserved role for the aryl hydrocarbon receptor in the regulation of movement. PLoS Genet. 10, e1004673. doi:10.1371/journal.pgen.1004673.

Wilson, A. J., Coltman, D. W., Pemberton, J. M., Overall, A. D. J., Byrne, K. A., and Kruuk, L. E. B. (2004). Maternal genetic effects set the potential for evolution in a free-living vertebrate population. J. Evol. Biol. 18, 405–414. doi:10.1111/j.1420-9101.2004.00824.x.

Wolf, J. B., Cheverud, J. M., Roseman, C., and Hager, R. (2008a). Genome-wide analysis reveals a complex pattern of genomic imprinting in mice. PLoS Genet. 4, e1000091. doi:10.1371/journal.pgen.1000091.

Chapter 1: General introduction

Wolf, J. B., Hager, R., and Cheverud, J. M. (2008b). Genomic imprinting effects on complex traits: a phenotype-based perspective. Epigenetics 3, 295–299.

Wolf, J. B., Leamy, L. J., Roseman, C. C., and Cheverud, J. M. (2011). Disentangling prenatal and postnatal maternal genetic effects reveals persistent prenatal effects on offspring growth in mice. Genetics 189, 1069– 1082. doi:10.1534/genetics.111.130591.

Wolf, J. B., Vaughn, T. T., Pletscher, L. S., and Cheverud, J. M. (2002). Contribution of maternal effect QTL to genetic architecture of early growth in mice. Heredity 89, 300–310.

Wolf, J. B., and Wade, M. J. (2009). What are maternal effects (and what are they not)? Philos. Trans. R. Soc. B Biol. Sci. 364, 1107–1115. doi:10.1098/rstb.2008.0238.

Wolf, J., and Cheverud, J. M. (2012). Detecting maternal-effect loci by statistical cross-fostering. Genetics 191, 261–277. doi:10.1534/genetics.111.136440.

Wu, C., Kraft, P., Zhai, K., Chang, J., Wang, Z., Li, Y., Hu, Z., He, Z., Jia, W., Abnet, C. C., et al. (2012). Genome-wide association analyses of esophageal squamous cell carcinoma in Chinese identify multiple susceptibility loci and gene-environment interactions. Nat. Genet. 44, 1090– 1097. doi:10.1038/ng.2411.

Wu, Y., Williams, E. G., Dubuis, S., Mottis, A., Jovaisaite, V., Houten, S. M., Argmann, C. A., Faridi, P., Wolski, W., Kutalik, Z., et al. (2014). Multilayered genetic and omics dissection of mitochondrial activity in a mouse reference population. Cell 158, 1415–1430. doi:10.1016/j.cell.2014.07.039.

Zhang, B., Kirov, S., and Snoddy, J. (2005). WebGestalt: an integrated system for exploring gene sets in various biological contexts. Nucleic Acids Res. 33, W741–8. doi:10.1093/nar/gki475.

Chapter 2: Joint genetic analysis of hippocampal size in mouse and human identifies a novel gene linked to neurodegenerative disease

Chapter 2: Preface

This chapter is based on a paper I have published in BMC genomics, but with changes to fit the format of this thesis.

The idea for this work was developed in collaboration between all authors.

Data analysis was carried out by David Ashbrook, with all authors providing advice.

The paper was written by David Ashbrook, with editing and comments from all other authors.

This chapter (and the paper it is based on) builds upon a paper published by

Hager et al. (2012) investigating the genetic architecture of brain region size in mice and another by Stein et al. (2012) investigating the genetics of brain region size in humans. By collaborating with both groups, and combining their data, we were able to identify a novel gene which may be influencing hippocampus size.

Chapter 2: Joint genetic analysis of hippocampal size in mouse and human identifies a novel gene linked to neurodegenerative disease

Joint genetic analysis of hippocampal size in mouse and human identifies a novel gene linked to neurodegenerative disease

David G. Ashbrook1, Robert W. Williams2, Lu Lu2, 3, Jason L. Stein4, Derrek P.

Hibar5, Thomas E. Nichols6, Sarah E. Medland7, Paul M. Thompson4, Reinmar

Hager1

1Computational and Evolutionary Biology, Faculty of Life Sciences, University of

Manchester, Manchester, M13 9PT, UK.

2University of Tennessee Health Science Center, Memphis, TN 38163, USA.

3Jiangsu Key Laboratory of Neuroregeneration, Nantong University, China

4Laboratory of Neuro Imaging, Department of Neurology, UCLA School of

Medicine, Los Angeles, CA 90095-1769, USA.

5Imaging Genetics Center, Institute for Neuroimaging and Informatics, Keck

School of Medicine, University of Southern California, 2001 N. Soto Street, Los

Angeles, CA 90033, USA

6Department of Statistics & Warwick Manufacturing Group, The University of

Warwick, Coventry CV4 7AL, UK

7Genetic Epidemiology Laboratory, Queensland Institute of Medical Research

Berghofer, Brisbane, Australia

Chapter 2: Joint genetic analysis of hippocampal size in mouse and human identifies a novel gene linked to neurodegenerative disease

Abstract

Variation in hippocampal volume has been linked to significant differences in memory, behaviour, and cognition among individuals. To identify genetic variants underlying such differences and associated disease phenotypes, multinational consortia such as ENIGMA have used large magnetic resonance imaging (MRI) data sets in human GWASs. In addition, mapping studies in mouse model systems have identified genetic variants for brain structure variation with great power. A key challenge is to understand how genetically based differences in brain structure lead to the propensity to develop specific neurological disorders.

We combine the largest human GWAS of brain structure with the largest mammalian model system, the BXD recombinant inbred mouse population, to identify novel genetic targets influencing brain structure variation that are linked to increased risk for neurological disorders. We first use a novel cross-species, comparative analysis using mouse and human genetic data to identify a candidate gene, MGST3, associated with adult hippocampus size in both systems. We then establish the co-regulation and function of this gene in a comprehensive systems-analysis.

We find that MGST3 is associated with hippocampus size and is linked to a group of neurodegenerative disorders, such as Alzheimer’s.

Chapter 2: Joint genetic analysis of hippocampal size in mouse and human identifies a novel gene linked to neurodegenerative disease

2.1 Introduction

The hippocampus is a key forebrain region involved in declarative memory, cognition, and spatial navigation. Hippocampal volume is highly variable with unilateral values ranging from ~2500 to 5000 mm3 among healthy young humans

(mean 3,917 mm3, s.d. = 441 mm3) and from 15.2 to 23.0 mm3 among young adult mice (Peirce et al., 2003; Stein et al., 2012). Heritability ranges from 40% to

70% in both species (Lu et al., 2001; Peper et al., 2007), and a small fraction of the difference in volume is also attributable to sex (Lu et al., 2001; Fjell et al.,

2009). This wide range of natural variation raises the possibility that susceptibility to a subset of neurodegenerative and psychiatric disorders linked to defects in the hippocampus may depend, in part, on its initial healthy volume. Individuals who develop and retain a large hippocampus into adulthood may be comparatively resistant to some forms of disease, particularly Alzheimer's. Such a "reserve" hypothesis of neurological disease (Barnett et al., 2006;

Nithianantharajah and Hannan, 2009) has been proposed for Parkinson's

(Biundo et al., 2013), Huntington's (Bonner-Jackson et al., 2013) and Alzheimer’s

(Braskie and Thompson, 2013) diseases. Lower than average volume has been linked to a number of disorders (Geuze et al., 2005) including depression (Frodl et al., 2002; Videbech et al., 2004; Sheline, 2011; Sexton et al., 2012; Sawyer et al., 2012), Alzheimer’s disease (den Heijer et al., 2006) and schizophrenia

(Adriano et al., 2012). Understanding the genetic factors that contribute to individual differences in hippocampal volume is thus crucial in providing insight into vulnerability and severity of disease.

Prior efforts to identify genetic variants underlying differences in brain structure have used large data sets in human genome-wide association studies

(GWASs) or extensive mapping populations in mouse model systems. GWASs in humans typically have modest statistical power due to high corrections needed to

Chapter 2: Joint genetic analysis of hippocampal size in mouse and human identifies a novel gene linked to neurodegenerative disease compensate for multiple testing. However, loci are defined with very high precision, often down to the level of single nucleotide polymorphisms (SNPs). In contrast, mouse linkage studies often have high statistical power to detect genetic effects but lower genetic resolution, producing loci that include hundreds of genes (Mackay et al., 2009; Ackert-Bicknell et al., 2010). Combining data from mice and humans overcomes some of these problems, gaining power from mouse crosses and precision from human GWAS. This method also ensures the translational relevance, giving confidence to the human results, as the same gene controlling the same phenotype is found in a related species. Further, this approach illustrates that the homologous mouse gene is relevant to the human phenotype, as well as the significance of experimental research in model systems that would not be possible in humans. Homologous genes are genes that share a common evolutionary ancestor. In this study we are specifically looking at a subset of homologous genes, orthologues, which derive from a speciation event, rather than paralogues, which arise because of a gene duplication event: that is we are looking at genes which have not duplicated in either genome since the common ancestor of mice and humans.

This study takes a cross-species approach to identify genes with an evolutionarily conserved role in influencing hippocampus size; i.e. because a given gene is playing the same role in two different species we hypothesise that it was playing the same role in the ancestral species. Previous studies have begun to show the utility of using a cross-species approach to identify genes underlying a phenotype of interest (Koutnikova et al., 2009; de Mooij-van Malsen et al., 2009; Poot et al., 2011; Leduc et al., 2011; Schofield et al., 2012). This approach has the advantage that it allows the investigation of disease phenotypes without requiring data from experimental perturbations. Instead we

Chapter 2: Joint genetic analysis of hippocampal size in mouse and human identifies a novel gene linked to neurodegenerative disease utilize data obtained from populations that segregated for large numbers of common sequence variants and associated differences in phenotype.

Here, we use data from the most extensively phenotyped mouse model system, BXD, to identify a set of genes associated with hippocampus size in a joint analysis with human hippocampus MRI data obtained by the ENIGMA consortium for GWASs (Thompson et al., 2014). We identify, MGST3 and use a systems-genetics approach that links this gene to neurodegenerative disorders such as Alzheimer’s disease and Parkinson’s disease.

Chapter 2: Joint genetic analysis of hippocampal size in mouse and human identifies a novel gene linked to neurodegenerative disease

2.2 Materials and methods

2.2.1 Data

We used mouse hippocampus weight data from 35 BXD lines plus the parental

C57BL/6J and DBA/2J strains, adjusted for age, sex, body weight, and brain weight minus hippocampus weight (GN13031; Lu et al., 2001). Over 3800 SNP markers are used for QTL interval mapping, thus for each marker significance values are available. Using the gene’s base pair distance from the nearest two markers we developed a Python script to produce an estimated p-value for each gene. For example a gene positioned halfway between two markers would have an estimated p-value half way between the two marker values. Therefore an estimated p-value could be produced for any gene in the mouse genome

(NCBI37/mm9) by using the gene’s known start position and any set of mouse markers.

Human MRI-generated hippocampus volume from healthy subjects and patients was generated for GWAS meta-analyses by the Enhancing Neuro

Imaging Genetics Through Meta-Analysis (ENIGMA) network (Stein et al., 2012;

Novak et al., 2012; Thompson et al., 2014) and can be visualised with

ENIGMAvis (Novak et al., 2012). Association analyses used multiple linear regression with hippocampus volume as a dependent variable and the additive dosage of each SNP as an independent variable, controlling for covariates of population stratification (four MDS components), intracranial volume, age, age2, sex and the interactions between age and sex and age2 and sex. Dummy covariates were used to control for different scanner sequences or equipment within a site. We converted the p-values available for each SNP marker to gene level significance values using the Versatile Gene-based Association Study (Liu et al., 2010) website (VEGAS). This tool tests for association between the

Chapter 2: Joint genetic analysis of hippocampal size in mouse and human identifies a novel gene linked to neurodegenerative disease phenotype and a gene by summarizing the full set of SNPs in the gene. Linkage disequilibrium between SNPs in a gene is taken into account by using simulation based on the pre-calculated linkage disequilibrium structure of a set of reference individuals from the HapMap phase 2 CEU population. SNPs are assigned to each of 17,787 autosomal genes on the UCSC Genome Browser hg18 assembly, with boundaries defined as ±50 kb of 5’ and 3’ UTRs. Association p- values for any given gene with n SNPs are converted to uppertail chi-squared statistics with one degree of freedom (df). The gene-based test statistic is then the sum of all of the chi-squared 1 df statistics within that gene. If the SNPs are in perfect linkage equilibrium, the test statistic will have a chi-squared distribution with n degrees of freedom under the null hypothesis. However, this is unlikely to be the case, therefore the true null distribution given the LD structure (and hence p-values that correlate accordingly) will need to be taken into account. This is done by simulating a large number of multivariate normal vectors, and the empirical gene based p-value is the proportion of simulated test statistics that exceed the observed gene-based test statistic (Liu et al., 2010). Thus, we are able to identify genes associated with hippocampus size that may be significant, independent of whether individual SNPs are significant.

All data used for the above is from existing, previously published, anonymised data and therefore no further ethical approval was needed.

2.2.2 Identification of significant genes for hippocampus size in mouse and human

To be able to compare the data between species, mouse homologues for the human genes need to be identified. The marker method above can produce a p- value for any mouse gene; therefore it is the human genes produced by VEGAS that limit the total number of genes in our analysis. Using the human SNP p-

Chapter 2: Joint genetic analysis of hippocampal size in mouse and human identifies a novel gene linked to neurodegenerative disease values, VEGAS produced p-values for 17,787 human genes (the number of autosomal genes on the UCSC Genome Browser hg18 assembly). We used several tools to identify mouse homologues for these human genes:

MammalHom (http://depts.washington.edu/l2l/mammalhom.html), Mouse

Genome Informatics (www.informatics.jax.org) and HomoloGene

(www.ncbi.nlm.nih.gov/homologene). Thus, 15,705 mouse genes with a corresponding human homologue were identified, representing 88.3% of human genes.

To determine if genes affecting hippocampus weight in mice also influence hippocampus volume in humans we used a protocol developed in R (R

Core Team, 2013). Firstly, we produced a quantile-quantile plot using the human p-values of those of the 42 genes which had a mouse p-value of ≤ 0.05.

Secondly, the genomic control λ-value (Bacanu et al., 2000) was calculated. This value is generally a measure of inflation of statistics due to population stratification, i.e. if significance is increased due to the populations being related.

In our case a high lambda would show that overall those genes with a significant mouse p-value have a higher human p-value than would be expected by chance.

In other words, by using genes which are significant in mouse, the p-values of the homologous human genes would be inflated. In our study we tested this value by permutations, with the same number of random genes sampled from the genome and the λ-value calculated (random λ). The number of times that the random λ was greater than the calculated λ was divided by the number of permutations (100,000) to give the p-value of the calculated lambda values. The permutations determine if a high λ-value is simply due to an overall high λ between the two datasets, i.e. that all the p-values in human are higher than would be expected by chance. We validated results thus obtained using an additional approach, the Rank Rank Hypergeometric Overlap test (Plaisier et al.,

Chapter 2: Joint genetic analysis of hippocampal size in mouse and human identifies a novel gene linked to neurodegenerative disease

2010).This was carried out using the RRHO R package (Rosenblatt and Stein,

2013), which computes the number of overlapping elements, and return the observed signiﬁcance of this overlap using a hypergeometric test.

Thirdly, to assess if any particular gene is associated with brain region size in both mouse and human the significance of the homologues for the 42 genes found to be significant in BXD mice were examined in the human GWAS data. This was corrected for multiple comparisons using the number of genes compared (de Mooij-van Malsen et al., 2009) (42 significant mouse genes), therefore 0.05/42 = p < 0.0012.

2.2.3 Expression quantitative trait loci

Expression quantitative trait loci (eQTL) show regions of the genome that influence the expression of a gene of interest. A cis-eQTL, i.e. an eQTL in the same position of the candidate gene, suggests that the candidate gene regulates its own expression, whereas a trans-eQTL, i.e. a QTL elsewhere in the genome, indicates that a gene at this position is influencing the expression of the candidate gene. Data for exon mRNA expression in the hippocampus of mouse lines (mainly BXD but with data from other inbred mouse lines) available at

GeneNetwork were used and WebQTL (Wang et al., 2003) produced eQTL for genes identified above. The database of microarray results used from

GeneNetwork was UMUTAffy Hippocampus Exon (Feb09) RMA (GN206)

(Mulligan et al., 2012). This allows the examination of the cis- or trans- regulation of our identified gene (Mgst3) in the mouse hippocampus. Using this exon gene expression data, all probes for Mgst3 were correlated using Pearson’s product- moment correlation as implemented in GeneNetwork, and those probes which showed a significant correlation (r ≥ 0.5, p ≤ 0.05) were said to represent the expression of the gene.

Chapter 2: Joint genetic analysis of hippocampal size in mouse and human identifies a novel gene linked to neurodegenerative disease

2.2.4 Functional analysis

Functional analysis allows us to investigate enrichment; for example if the molecular function of a gene product is over-represented in a submitted list of genes. Enrichment therefore suggests whether a particular gene or a set of genes is associated with a particular function or disorder.

The Database for Annotation, Visualization and Integrated Discovery

(DAVID; Huang et al., 2008, 2009) identifies if a given list of genes is significantly enriched in an annotated gene term. DAVID uses a range of databases, including

Gene Ontologies (GO) terms (Ashburner et al., 2000), Kyoto Encyclopedia of

Genes and Genomes (KEGG) pathways (Kanehisa and Goto, 2000), Online

Mendelian Inheritance in Man diseases and InterPro protein domains (Hunter et al., 2012). Separate lists of all significant mouse genes (p ≤ 0.05) and all nominally significant human genes (p ≤ 0.05) were analysed, and the results examined for any annotations that appeared in both datasets. The latter would suggest that the same pathways or networks were involved in the phenotype in both species, even if the same individual genes are not significant.

2.2.5 Co-expression and ‘Guilt-by-association’

Shared regulation and function of genes can also be established using co- expression analysis (Allocco et al., 2004). However, co-expression can differ between species or between tissues within an organism. To examine if genes are commonly co-expressed in humans, GeneFriends can be used (van Dam et al.,

2012). GeneFriends takes submitted list of genes and uses a large database of microarray data (4164 Micro array datasets containing 26,113 experimental conditions and 19,080 genes)(van Dam et al., 2012), from the Gene Expression

Omnibus (Edgar et al., 2002; Barrett et al., 2013) to find genes that are commonly co-expressed with the entered gene list. However it is not specific for

Chapter 2: Joint genetic analysis of hippocampal size in mouse and human identifies a novel gene linked to neurodegenerative disease tissue type or treatment, and therefore can only inform us which genes tend to co-express together, and not which genes specifically co-express in the hippocampus or at what time points.

GeneFriends produces a list of genes that are co-expressed with the submitted genes in a significant number of datasets, to identify commonly co- expressed genes (co-expressed independent of treatment or tissue). Common co-expression suggests that the genes are under the same regulation in particular since co-expression is analysed across conditions and tissues. This list of commonly co-expressed genes was analysed using DAVID as above, producing annotations for these genes. This allows a ‘guilt-by-association’ approach, where the roles played by genes that are commonly co-expressed with our genes are used to suggest the networks that the genes are part of (van Dam et al., 2012).We next used Pearson product-moment correlations, as implemented in GeneNetwork, to examine co-expression in mice by producing correlation matrices of hippocampal gene expression (Chesler et al., 2003). In contrast to GeneFriends this is specific to the hippocampus. Hippocampus mRNA expression was found for Mgst3 in the UMUTAffy Hippocampus Exon

(Feb09) RMA (GN206) microarray database. The probes for Mgst3 were correlated with each other, and six showed a significant correlation (r ≥ 0.5, p ≤

0.05) and were used to determine gene expression. For each of these six correlating probes for Mgst3, the top 20,000 correlations were then found within the whole hippocampus exon array dataset (1,236,087 probes). 5906 probes correlated with all six of the probes for Mgst3, representing 2971 genes. This list of 2971 was submitted to DAVID to determine KEGG pathway enrichment.

Significance testing using permutations was then carried out to determine the overlap between six random samples of 20,000 values (the number of

Chapter 2: Joint genetic analysis of hippocampal size in mouse and human identifies a novel gene linked to neurodegenerative disease correlations) from a total of 1,236,087 values (the total number of probes). With

1,000,000 permutations this produced a value of p < 1x10-6.

The overlap between the genes identified by GeneFriends and those identified by GeneNetwork was also examined. The resulting list of genes was then submitted to DAVID for KEGG pathway enrichment analysis. The significance of the number of overlapping genes was again determined by permutation. Samples of sizes 8135 (the number of co-expressing genes found by GeneFriends) and of 2971 (the number of co-expressing genes found by

GeneNetwork) were taken from a list of all protein coding human genes

(downloaded from the HUGO gene nomenclature committee website) (Gray et al., 2013), and the overlap between these two samples recorded. This was repeated 1,000,000 times and the significance value was calculated by dividing the number of times the overlap between the two samples was greater than 1579

(the overlap we see) by the number of permutations (1,000,000).

Chapter 2: Joint genetic analysis of hippocampal size in mouse and human identifies a novel gene linked to neurodegenerative disease

2.3 Results

2.3.1 Identification of genes significant in both species

Associations between genes and hippocampus size in BXD mice were estimated using p-values for over 3800 markers obtained for QTL interval mapping. QTL mapping identifies a region of the genome significantly linked to variation in the phenotype. Having identified QTL, we then estimated a particular gene’s significance based on its base pair distance from the nearest two markers and the significance of these two markers. Therefore, any particular gene will have a p-value somewhere between the p-values of its two closest markers. The next step in our analysis was to obtain SNP level p-values for association with human hippocampus volume, which were converted to gene p-values to allow comparison with data for the mouse hippocampus. Using the SNP p-values from the human GWAS, the Versatile Gene-based Association Study (VEGAS) website (Liu et al., 2010) produced gene p-values for a total of 17,787 human genes. Secondly, the mouse homologues of these human genes were identified and yielded a total of 15,705 genes (88.3% of the human genes).

Using a relaxed (i.e. uncorrected) p-value of ≤ 0.05, 1015 human genes

(916 with mouse homologues; 90.2%) were then nominally identified as having an effect on hippocampus size. Overall, there is no indication that the significance of any given gene with the entire region identified in the QTL analysis of BXD mouse hippocampus weight is indicative of the homologous gene’s significance on human hippocampus volume, as judged by the quantile- quantile plot and lambda (λ = 0.912, p = 0.82; Figure 2.1). This is corroborated by a separate Rank Rank Hypergeometric Overlap test (Plaisier et al., 2010) used to compare the two datasets, which yielded a non-significant result (p = 0.38, corrected by the familywise error rate). This is unsurprising as a QTL analysis

Chapter 2: Joint genetic analysis of hippocampal size in mouse and human identifies a novel gene linked to neurodegenerative disease identifies a region of the genome associated with a trait, and therefore in our analysis all genes within the mouse QTL were significant. However, not all the genes within a QTL contribute to the phenotype, but only a subset or even a single gene.

Figure 2.1: Quantile-quantile plot of human homologues of mouse genes within the significant QTL on chromosome 1 for hippocampus size. For genes with a significant influence on hippocampus weight in mice (≤ 0.05) the significance of their influence on human hippocampus volume was plotted against a normal distribution of p-values.

Although there are outliers, most of the points lie close to the y=x line, indicating there is no difference between what is seen in the data and what would be expected by chance.

This is reinforced by the non-significant lambda value close to 1, which indicates no inflation of significance values. The point with the highest observed p-value is MGST3.

Therefore, we sought to identify which genes are associated with both

BXD mouse hippocampus weight and human hippocampus volume. The 42 genes which were significant in mouse are all within a QTL on chromosome 1 (Lu

Chapter 2: Joint genetic analysis of hippocampal size in mouse and human identifies a novel gene linked to neurodegenerative disease et al., 2001; Hager et al., 2012).Only one gene had a significant human p-value when corrected for multiple testing (0.05/42 = p ≤ 0.0012; Supplementary table

2.1), MGST3.

2.3.2 Regulation of gene expression

To establish if our candidate gene was regulating its own expression, mouse hippocampus microarray data from GeneNetwork was used to find probes corresponding to expression of the gene, and WebQTL was used to produce eQTL. Of the 17 probes for Mgst3 within the exon array data, six have a significant Pearson’s correlation (r ≥ 0.5, p ≤ 0.05), and these probes were used to represent expression of Mgst3 in the mouse hippocampus. The six probes represent four probes for exons and one each from the 5’ and 3’ UTR (Table

2.1). The remaining 11 probes were for introns and UTRs. This shows that the correlating probes represent the protein coding parts of the gene.

Mgst3 has a cis-eQTL, suggesting it regulates its own expression. No trans-QTL was found which was consistent between probes. The QTL and eQTL analysis also showed that the C57BL/6J (B6) allele increased hippocampus weight, whereas the DBA/2J (D2) allele increased the expression of Mgst3.

Chapter 2: Joint genetic analysis of hippocampal size in mouse and human identifies a novel gene linked to neurodegenerative disease

Table 2.1: Pearson’s correlations between probes for Mgst3 in adult mouse hippocampus. Pearson product-moment correlations produced by GeneNetwork.

The dataset used was UMUTAffy Hippocampus Exon (Feb09) RMA (GN206). All probes are for Mgst3 located on chromosome 1.

Probe ID Target 1 2 3 4 5 6 1 4654447 3'UTR 1 r = 0.921 r = 0.873 r = 0.787 r = 0.8 r = 0.62 p < 1E-16 p < 1E-16 p < 1E-16 p < 1E-16 p = 9.87E-10 2 5358488 exon5 r = 0.921 1 r = 0.903 r = 0.84 r = 0.828 r = 0.632 p < 1E-16 p < 1E-16 p < 1E-16 p < 1E-16 p = 3.71E-10 3 5399827 exon4 r = 0.873 r = 0.903 1 r = 0.937 r = 0.884 r = 0.534 p < 1E-16 p < 1E-16 p < 1E-16 p < 1E-16 p = 5.33E-7 4 5566068 exon2 r = 0.787 r = 0.84 r = 0.937 1 r = 0.92 r = 0.512 p < 1E-16 p < 1E-16 p < 1E-16 p < 1E-16 p = 1.84E-6 5 5025657 exon1 r = 0.8 r = 0.828 r = 0.884 r = 0.92 1 r = 0.519 p < 1E-16 p < 1E-16 p < 1E-16 p < 1E-16 p = 1.24E-6 6 5280988 5'UTR r = 0.62 r = 0.632 r = 0.534 r = 0.512 r = 0.519 1 p = 9.87E-10 p = 3.71E-10 p = 5.33E-7 p = 1.84E-6 p = 1.24E-6

2.3.3 Functional analysis of significant genes

To investigate the function of our candidate genes, we used the Database for

Annotation, Visualization and Integrated Discovery (DAVID) as it allows us to analyse a number of different annotation databases. Significance was determined by the false discovery rate (FDR), which corrects the significance value for the large number of multiple comparisons.

To determine if any annotations were enriched in both mouse and human, even though individual genes were not shared, separate lists of genes nominally significant in human (915 genes p ≤ 0.05) and human homologues of the 42 mouse genes with p ≤ 0.05 were entered into DAVID. No overlapping significant annotations were found, i.e. no annotations were significantly enriched in both the genes significant in human and the genes significant in mouse. Again, this suggests that not all 42 genes within the mouse QTL influence the phenotype, but only a subset, although it could be that several genes, each of which

Chapter 2: Joint genetic analysis of hippocampal size in mouse and human identifies a novel gene linked to neurodegenerative disease influence hippocampus size by a different mechanism, are found within the QTL, as these would not cluster together functionally, and therefore would not be detected as enriched.

2.3.4 ‘Guilt-by-association’

Co-expression of genes implies that they share the same regulatory mechanisms

(Allocco et al., 2004) and/or are involved in the same biological processes (van

Dam et al., 2012). A ‘guilt-by-association’ approach asserts that the function of a gene, or list of genes, can be indicated by the genes that it commonly co- expresses with, as common co-expression indicates they are part of the same biological process (van Dam et al., 2012). The large datasets of gene expression provided by GeneNetwork and GeneFriends allows this ‘guilt-by-association’ approach to be used. This is especially useful for genes such as MGST3/Mgst3, which previously have not been investigated in detail.

GeneFriends shows human genes which co-express in a large number of datasets from the Gene Expression omnibus. However, it is not specific for tissue or treatment. This identified 8135 genes that were co-expressed with MGST3 in over half of the datasets (co-expression value ≥ 0.5; Supplementary table 2.2).

These were analysed using DAVID, to find what KEGG pathway annotations were significant (FDR ≤ 0.05; Supplementary table 2.3). One of the six KEGG pathways is particularly interesting; Alzheimer's disease (FDR = 0.0029).

To support a specific link between genes that are co-expressed with

Mgst3 and Alzheimer’s disease we used the exon array data from GeneNetwork, as this is specific to the hippocampus. Each of the six above identified probes for

Mgst3 was correlated against the entire exon array dataset (1,236,087 probes) to find the top 20,000 probes with which it correlates. These six lists of probes were then combined to find which probes correlated with all six probes for Mgst3. This

Chapter 2: Joint genetic analysis of hippocampal size in mouse and human identifies a novel gene linked to neurodegenerative disease produced a list of 5906 probes which correlated with all six of the probes for

Mgst3, representing 2971 genes (Supplementary table 2.4). Permutation testing was carried out to determine an empirical p-value for how often six lists of 20,000 values, from a choice of 1,236,087, values would overlap and produced a p-value of < 1x10-6. This shows that the overlap between our probes is highly significant, and that these 2971 genes really do co-express with Mgst3 in the mouse hippocampus. Enrichment of this list of genes was then examined in DAVID, and again we see genes involved in neurodegenerative diseases significantly enriched (Supplementary table 2.5): Huntington's disease (95 genes, FDR =

3.29E-27, Parkinson's disease (77 genes, FDR = 1.56E-25) and Alzheimer's disease (83 genes, FDR = 1.29E-18).

Finally, the overlap between the genes that are commonly co-expressed with MGST3 and human homologues of the genes that are co-expressed with

Mgst3 in the mouse hippocampus was examined. This showed that 1579 genes which commonly co-express with MGST3 also co-express with its mouse homologue in the mouse hippocampus (Supplementary table 2.6). We tested this by permutation taking samples of 8135 genes and 2971 genes from a list of all known human protein coding genes and determining how often an overlap larger than 1579 was seen. This produced an empirical p-value ≤ 1x10-6. Again with

KEGG enrichment analysis, the three neurodegenerative diseases are highly significant (Supplementary figure 2.7): Huntington's disease (78 genes; FDR =

3.08E-22), Parkinson's disease (63 genes; FDR = 5.68E-21) and Alzheimer's disease (69 genes; FDR = 1.34E-18).

Chapter 2: Joint genetic analysis of hippocampal size in mouse and human identifies a novel gene linked to neurodegenerative disease

2.4 Discussion

We found strong evidence that MGST3, on chromosome 1 in both mouse and human, is significantly associated with hippocampus size. MGST3 has previously been found to be down-regulated in Alzheimer’s disease (Xu et al., 2006). The fact that the gene appears to have an evolutionarily conserved role in both species suggests a role in hippocampus morphology. MGST3 has been found to be particularly highly expressed in the rat hippocampus (Fetissov et al., 2002). A

‘guilt-by-association’ approach shows that these genes co-express with genes linked to neurodegenerative disorders associated with reduced hippocampus volume: Huntington's disease (Spargo et al., 1993; Rosas et al., 2003; Bohanna et al., 2008), Alzheimer's disease (den Heijer et al., 2006; Frisoni et al., 2010) and Parkinson's disease (Camicioli et al., 2003; Pereira et al., 2013; Zhang et al.,

2014).

The potential mechanism for this link is more speculative. Genes that co- express with MGST3 are also associated with cellular energy production, as the oxidative phosphorylation KEGG pathway appears in our results (Supplementary table 2.3, Supplementary table 2.5, and Supplementary table 2.7). Mitochondrial dysfunction has been implicated in both neuropsychiatric and neurodegenerative disorders (Deheshi et al., 2013; Chaturvedi and Flint Beal, 2013), linking the mitochondrial and neurodegenerative annotations. Recently it has been reported that dysfunction of mtDNA genes, which have been implicated in Alzheimer’s disease, directly influence left hippocampal atrophy (Ridge et al., 2013). Further, links have also been found between oxidative stress and regulation of Mgst3 in mice (Higgins and Hayes, 2011).

MGST3 has also been linked to inflammation, as it and other family members show leukotriene C4 (LTC4) synthase activity (Jakobsson et al., 1997).

Chapter 2: Joint genetic analysis of hippocampal size in mouse and human identifies a novel gene linked to neurodegenerative disease

Leukotrienes are physiological important mediators of various inflammatory and immediate hypersensitivity processes (Jakobsson et al., 1997). When porcine kidney cells were exposed a nephrotoxin, aristolochic acid I, MGST3 and FLAP

(another family member) were upregulated before an increase in LTs synthesis.

This is relevant as Alzheimer’s disease, as well as other neurodegenerative disorders, have been linked to inflammation (reviewed by Amor et al. (2014)).

However, other research has found that rat MGST3 does not have LTC4 synthase activity (Schröder et al., 2003), is not upregulated in response to lipopolysaccharide (Schröder et al., 2005) and does not appear to be directly involved in the inflammation response (Fetissov et al., 2002). In this last paper, the authors speculate that it may have a neuroprotective role against oxidative stress (Fetissov et al., 2002).

2.4.1 Conclusion

In summary, the combination of human GWAS and mouse QTL data from some of the largest study systems available has enabled us to identify a novel gene,

MGST3, which is associated with hippocampus size across species and, when dysregulated, is linked to neurodegenerative disorders.

Chapter 2: Joint genetic analysis of hippocampal size in mouse and human identifies a novel gene linked to neurodegenerative disease

2.5 References

Ackert-Bicknell, C. L., Karasik, D., Li, Q., Smith, R. V, Hsu, Y.-H., Churchill, G. A., Paigen, B. J., and Tsaih, S.-W. (2010). Mouse BMD quantitative trait loci show improved concordance with human genome-wide association loci when recalculated on a new, common mouse genetic map. J. Bone Miner. Res. 25, 1808–1820. doi:10.1002/jbmr.72.

Adriano, F., Caltagirone, C., and Spalletta, G. (2012). Hippocampal volume reduction in first-episode and chronic schizophrenia: a review and meta- analysis. Neuroscientist 18, 180–200. doi:10.1177/1073858410395147.

Allocco, D. J., Kohane, I. S., and Butte, A. J. (2004). Quantifying the relationship between co-expression, co-regulation and gene function. BMC Bioinformatics 5, 18. doi:10.1186/1471-2105-5-18.

Amor, S., Peferoen, L. A. N., Vogel, D. Y. S., Breur, M., van der Valk, P., Baker, D., and van Noort, J. M. (2014). Inflammation in neurodegenerative diseases--an update. Immunology 142, 151–166. doi:10.1111/imm.12233.

Bacanu, S. A., Devlin, B., and Roeder, K. (2000). The power of genomic control. Am. J. Hum. Genet. 66, 1933–1944. doi:10.1086/302929.

Barnett, J. H., Salmond, C. H., Jones, P. B., and Sahakian, B. J. (2006). Cognitive reserve in neuropsychiatry. Psychol. Med. 36, 1053–1064. doi:10.1017/S0033291706007501.

Biundo, R., Calabrese, M., Weis, L., Facchini, S., Ricchieri, G., Gallo, P., and Antonini, A. (2013). Anatomical correlates of cognitive functions in early Parkinson’s disease patients. PLoS One 8, e64222. doi:10.1371/journal.pone.0064222.

Bohanna, I., Georgiou-Karistianis, N., Hannan, A. J., and Egan, G. F. (2008). Magnetic resonance imaging as an approach towards identifying neuropathological biomarkers for Huntington’s disease. Brain Res. Rev. 58, 209–225. doi:10.1016/j.brainresrev.2008.04.001.

Bonner-Jackson, A., Long, J. D., Westervelt, H., Tremont, G., Aylward, E., and Paulsen, J. S. (2013). Cognitive reserve and brain reserve in prodromal Huntington’s Disease. J. Int. Neuropsychol. Soc. 19, 739–750. doi:10.1017/S1355617713000507.

Chapter 2: Joint genetic analysis of hippocampal size in mouse and human identifies a novel gene linked to neurodegenerative disease

Braskie, M. N., and Thompson, P. M. (2013). Understanding cognitive deficits in Alzheimer’s disease based on neuroimaging findings. Trends Cogn. Sci. 17, 510–516. doi:10.1016/j.tics.2013.08.007.

Camicioli, R., Moore, M. M., Kinney, A., Corbridge, E., Glassberg, K., and Kaye, J. A. (2003). Parkinson’s disease is associated with hippocampal atrophy. Mov. Disord. 18, 784–790. doi:10.1002/mds.10444.

Chaturvedi, R. K., and Flint Beal, M. (2013). Mitochondrial diseases of the brain. Free Radic. Biol. Med. 63, 1–29. doi:10.1016/j.freeradbiomed.2013.03.018.

Chesler, E. J., Wang, J., Lu, L., Qu, Y., Manly, K. F., and Williams, R. W. (2003). Genetic correlates of gene expression in recombinant inbred strains: a relational model system to explore neurobehavioral phenotypes. Neuroinformatics 1, 343–357. doi:10.1385/NI:1:4:343.

Van Dam, S., Cordeiro, R., Craig, T., van Dam, J., Wood, S. H., and de Magalhães, J. P. (2012). GeneFriends: an online co-expression analysis tool to identify novel gene targets for aging and complex diseases. BMC Genomics 13, 535. doi:10.1186/1471-2164-13-535.

Deheshi, S., Pasqualotto, B. A., and Rintoul, G. L. (2013). Mitochondrial trafficking in neuropsychiatric diseases. Neurobiol. Dis. 51, 66–71. doi:10.1016/j.nbd.2012.06.015.

Edgar, R., Domrachev, M., and Lash, A. E. (2002). Gene Expression Omnibus: NCBI gene expression and hybridization array data repository. Nucleic Acids Res. 30, 207–210. doi:10.1093/nar/30.1.207.

Fetissov, S. O., Schröder, O., Jakobsson, P.-J. P.-J. P.-J., Samuelsson, B., Haeggström, J. Z., and Hökfelt, T. (2002). Expression of microsomal glutathione S-transferase type 3 mRNA in the rat nervous system. Neuroscience 115, 891–897.

Fjell, A. M., Westlye, L. T., Amlien, I., Espeseth, T., Reinvang, I., Raz, N., Agartz, I., Salat, D. H., Greve, D. N., Fischl, B., et al. (2009). Minute effects of sex on the aging brain: a multisample magnetic resonance imaging study of healthy aging and Alzheimer’s disease. J. Neurosci. 29, 8774–8783. doi:10.1523/JNEUROSCI.0115-09.2009.

Frisoni, G. B., Fox, N. C., Jack, C. R., Scheltens, P., and Thompson, P. M. (2010). The clinical use of structural MRI in Alzheimer disease. Nat. Rev. Neurol. 6, 67–77. doi:10.1038/nrneurol.2009.215.

Frodl, T., Meisenzahl, E. M., Zetzsche, T., Born, C., Groll, C., Jäger, M., Leinsinger, G., Bottlender, R., Hahn, K., and Möller, H.-J. (2002). Hippocampal changes in patients with a first episode of major depression. Am. J. Psychiatry 159, 1112–1118. doi: 10.1176/appi.ajp.159.7.1112

Geuze, E., Vermetten, E., and Bremner, J. D. (2005). MR-based in vivo hippocampal volumetrics: 2. Findings in neuropsychiatric disorders. Mol. Psychiatry 10, 160–184. doi:10.1038/sj.mp.4001579.

Chapter 2: Joint genetic analysis of hippocampal size in mouse and human identifies a novel gene linked to neurodegenerative disease

Gray, K. A., Daugherty, L. C., Gordon, S. M., Seal, R. L., Wright, M. W., and Bruford, E. A. (2013). Genenames.org: the HGNC resources in 2013. Nucleic Acids Res. 41, D545–552. doi:10.1093/nar/gks1066.

Den Heijer, T., Geerlings, M. I., Hoebeek, F. E., Hofman, A., Koudstaal, P. J., and Breteler, M. M. B. (2006). Use of hippocampal and amygdalar volumes on magnetic resonance imaging to predict dementia in cognitively intact elderly people. Arch. Gen. Psychiatry 63, 57–62. doi:10.1001/archpsyc.63.1.57.

Higgins, L. G., and Hayes, J. D. (2011). Mechanisms of induction of cytosolic and microsomal glutathione transferase (GST) genes by xenobiotics and pro- inflammatory agents. Drug Metab. Rev. 43, 92–137. doi:10.3109/03602532.2011.567391.

Huang, D. W., Sherman, B. T., and Lempicki, R. A. (2009). Bioinformatics enrichment tools: paths toward the comprehensive functional analysis of large gene lists. Nucleic Acids Res. 37, 1–13. doi:10.1093/nar/gkn923.

Huang, D. W., Sherman, B. T., and Lempicki, R. A. (2008). Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources. Nat. Protoc. 4, 44–57. doi:10.1038/nprot.2008.211.

Hunter, S., Jones, P., Mitchell, A., Apweiler, R., Attwood, T. K., Bateman, A., Bernard, T., Binns, D., Bork, P., Burge, S., et al. (2012). InterPro in 2011: new developments in the family and domain prediction database. Nucleic Acids Res. 40, D306–D312. doi:10.1093/nar/gkr948.

Jakobsson, P. J., Mancini, J. A., Riendeau, D., and Ford-Hutchinson, A. W. (1997). Identification and characterization of a novel microsomal enzyme with glutathione-dependent transferase and peroxidase activities. J. Biol. Chem. 272, 22934–22939.

Kanehisa, M., and Goto, S. (2000). KEGG: kyoto encyclopedia of genes and genomes. Nucleic Acids Res. 28, 27–30.

Koutnikova, H., Laakso, M., Lu, L., Combe, R., Paananen, J., Kuulasmaa, T., Kuusisto, J., Häring, H.-U., Hansen, T., Pedersen, O., et al. (2009). Identification of the UBP1 locus as a critical blood pressure determinant using a combination of mouse and human genetics. PLoS Genet. 5, e1000591. doi:10.1371/journal.pgen.1000591.

Leduc, M. S., Lyons, M., Darvishi, K., Walsh, K., Sheehan, S., Amend, S., Cox, A., Orho-Melander, M., Kathiresan, S., Paigen, B., et al. (2011). The mouse QTL map helps interpret human genome-wide association studies for HDL cholesterol. J. Lipid Res. 52, 1139–1149. doi:10.1194/jlr.M009175.

Liu, J. Z., McRae, A. F., Nyholt, D. R., Medland, S. E., Wray, N. R., Brown, K. M., Hayward, N. K., Montgomery, G. W., Visscher, P. M., Martin, N. G., et al.

Chapter 2: Joint genetic analysis of hippocampal size in mouse and human identifies a novel gene linked to neurodegenerative disease

(2010). A versatile gene-based test for genome-wide association studies. Am. J. Hum. Genet. 87, 139–145. doi:10.1016/j.ajhg.2010.06.009.

Lu, L., Airey, D. C., and Williams, R. W. (2001). Complex trait analysis of the hippocampus: mapping and biometric analysis of two novel gene loci with specific effects on hippocampal structure in mice. J. Neurosci. 21, 3503– 3514.

Mackay, T. F. C., Stone, E. A., and Ayroles, J. F. (2009). The genetics of quantitative traits: challenges and prospects. Nat. Rev. Genet. 10, 565–577. doi:10.1038/nrg2612.

De Mooij-van Malsen, A. J. G., van Lith, H. A., Oppelaar, H., Hendriks, J., de Wit, M., Kostrzewa, E., Breen, G., Collier, D. a, Olivier, B., and Kas, M. J. (2009). Interspecies trait genetics reveals association of Adcy8 with mouse avoidance behavior and a human mood disorder. Biol. Psychiatry 66, 1123– 1130. doi:10.1016/j.biopsych.2009.06.016.

Nithianantharajah, J., and Hannan, A. J. (2009). The neurobiology of brain and cognitive reserve: mental and physical activity as modulators of brain disorders. Prog. Neurobiol. 89, 369–382. doi:10.1016/j.pneurobio.2009.10.001.

Novak, N. M., Stein, J. L., Medland, S. E., Hibar, D. P., Thompson, P. M., and Toga, A. W. (2012). EnigmaVis: online interactive visualization of genome- wide association studies of the Enhancing NeuroImaging Genetics through Meta-Analysis (ENIGMA) consortium. Twin Res. Hum. Genet. 15, 414–418. doi:10.1017/thg.2012.17.

Peirce, J. L., Chesler, E. J., Williams, R. W., and Lu, L. (2003). Genetic architecture of the mouse hippocampus: identification of gene loci with selective regional effects. Genes Brain Behav. 2, 238–252.

Peper, J. S., Brouwer, R. M., Boomsma, D. I., Kahn, R. S., and Hulshoff Pol, H. E. (2007). Genetic influences on human brain structure: a review of brain imaging studies in twins. Hum. Brain Mapp. 28, 464–473. doi:10.1002/hbm.20398.

Pereira, J. B., Junqué, C., Bartrés-Faz, D., Ramírez-Ruiz, B., Marti, M.-J., and Tolosa, E. (2013). Regional vulnerability of hippocampal subfields and memory deficits in Parkinson’s disease. Hippocampus 23, 720–728. doi:10.1002/hipo.22131.

Plaisier, S. B., Taschereau, R., Wong, J. A., and Graeber, T. G. (2010). Rank- rank hypergeometric overlap: identification of statistically significant overlap between gene-expression signatures. Nucleic Acids Res. 38, e169. doi:10.1093/nar/gkq636.

Chapter 2: Joint genetic analysis of hippocampal size in mouse and human identifies a novel gene linked to neurodegenerative disease

Poot, M., Badea, A., Williams, R. W., and Kas, M. J. (2011). Identifying human disease genes through cross-species gene mapping of evolutionary conserved processes. PLoS One 6, e18612. doi:10.1371/journal.pone.0018612.

R Core Team (2013). R: A Language and Environment for Statistical Computing. R Found. Stat. Comput. Available at: http://www.r-project.org/.

Ridge, P. G., Koop, A., Maxwell, T. J., Bailey, M. H., Swerdlow, R. H., Kauwe, J. S. K., and Honea, R. A. (2013). Mitochondrial haplotypes associated with biomarkers for Alzheimer’s disease. PLoS One 8, e74158. doi:10.1371/journal.pone.0074158.

Rosas, H. D., Koroshetz, W. J., Chen, Y. I., Skeuse, C., Vangel, M., Cudkowicz, M. E., Caplan, K., Marek, K., Seidman, L. J., Makris, N., et al. (2003). Evidence for more widespread cerebral pathology in early HD: an MRI- based morphometric analysis. Neurology 60, 1615–1620. doi:10.1212/01.WNL.0000065888.88988.6E.

Rosenblatt, J. D., and Stein, J. L. (2013). RRHO: Test overlap using the Rank- Rank Hypergeometric test. R package version 1.0.0.

Sawyer, K., Corsentino, E., Sachs-Ericsson, N., and Steffens, D. C. (2012). Depression, hippocampal volume changes, and cognitive decline in a clinical sample of older depressed outpatients and non-depressed controls. Aging Ment. Health 16, 753–762. doi:10.1080/13607863.2012.678478.

Schofield, P. N., Hoehndorf, R., and Gkoutos, G. V (2012). Mouse genetic and phenotypic resources for human genetics. Hum. Mutat. 33, 826–836. doi:10.1002/humu.22077.

Schröder, O., Sjöström, M., Qiu, H., Jakobsson, P.-J., and Haeggström, J. Z. (2005). Microsomal glutathione S-transferases: selective up-regulation of leukotriene C4 synthase during lipopolysaccharide-induced pyresis. Cell. Mol. Life Sci. 62, 87–94. doi:10.1007/s00018-004-4366-7.

Schröder, O., Sjöström, M., Qiu, H., Stein, J., Jakobsson, P. J., and Haeggström, J. Z. (2003). Molecular and catalytic properties of three rat leukotriene C(4) synthase homologs. Biochem. Biophys. Res. Commun. 312, 271–276.

Sexton, C. E., Le Masurier, M., Allan, C. L., Jenkinson, M., McDermott, L., Kalu, U. G., Herrmann, L. L., Bradley, K. M., Mackay, C. E., and Ebmeier, K. P. (2012). Magnetic resonance imaging in late-life depression: vascular and glucocorticoid cascade hypotheses. Br. J. Psychiatry 201, 46–51. doi:10.1192/bjp.bp.111.105361.

Sheline, Y. I. (2011). Depression and the hippocampus: cause or effect? Biol. Psychiatry 70, 308–309. doi:10.1016/j.biopsych.2011.06.006.

Spargo, E., Everall, I. P., and Lantos, P. L. (1993). Neuronal loss in the hippocampus in Huntington’s disease: a comparison with HIV infection. J. Neurol. Neurosurg. Psychiatry 56, 487–491.

Chapter 2: Joint genetic analysis of hippocampal size in mouse and human identifies a novel gene linked to neurodegenerative disease

Stein, J. L., Medland, S. E., Vasquez, A. A., Hibar, D. P., Senstad, R. E., Winkler, A. M., Toro, R., Appel, K., Bartecek, R., Bergmann, Ø., et al. (2012). Identification of common variants associated with human hippocampal and intracranial volumes. Nat. Genet. 44, 552–561. doi:10.1038/ng.2250.

Thompson, P. M., Stein, J. L., Medland, S. E., Hibar, D. P., Vasquez, A. A., Renteria, M. E., Toro, R., Jahanshad, N., Schumann, G., Franke, B., et al. (2014). The ENIGMA Consortium: large-scale collaborative analyses of neuroimaging and genetic data. Brain Imaging Behav. 8, 153–182. doi:10.1007/s11682-013-9269-5.

Videbech, P., Ravnkilde, B., and Ph, D. (2004). Hippocampal volume and depression: a meta-analysis of MRI studies. Am. J. Psychiatry 161, 1957– 1966. doi:10.1176/appi.ajp.161.11.1957.

Wang, J., Williams, R. W., and Manly, K. F. (2003). WebQTL: web-based complex trait analysis. Neuroinformatics 1, 299–308. doi:10.1385/NI:1:4:299.

Xu, P.-T., Li, Y.-J., Qin, X.-J., Scherzer, C. R., Xu, H., Schmechel, D. E., Hulette, C. M., Ervin, J., Gullans, S. R., Haines, J., et al. (2006). Differences in apolipoprotein E3/3 and E4/4 allele-specific gene expression in hippocampus in Alzheimer disease. Neurobiol. Dis. 21, 256–575. doi:10.1016/j.nbd.2005.07.004.

Zhang, Y., Zhang, J., Xu, J., Wu, X., Zhang, Y., Feng, H., Wang, J., and Jiang, T. (2014). Cortical gyrification reductions and subcortical atrophy in Parkinson’s disease. Mov. Disord. 29, 122–126. doi:10.1002/mds.25680.

Chapter 3: A cross-species genetic analysis identifies candidate genes for mouse anxiety and human bipolar disorder

Chapter 3: A cross-species genetic analysis identifies candidate

genes for mouse anxiety and human bipolar disorder

Chapter 3: A cross-species genetic analysis identifies candidate genes for mouse anxiety and human bipolar disorder

Chapter 3: Preface

This chapter is based on a paper I have published as first author in Frontiers in

Behavioural Neuroscience, but minor styling changes have been made to fit into the format of the thesis.

Data analysis was carried out by David Ashbrook

The paper was written by David Ashbrook, with advice, edits and comments from the co-authors.

This chapter builds upon the previous chapter, applying a similar approach to behavioural, rather than morphological, phenotypes. As this project was completed almost two years after the first, many of the methods have been updated. Our results show that by using analogous behaviours between species, we may be able to identify novel genes influencing mouse and human behaviour, and shed light on the biological mechanisms linking them.

Chapter 3: A cross-species genetic analysis identifies candidate genes for mouse anxiety and human bipolar disorder

A cross-species genetic analysis identifies candidate genes for mouse anxiety and human bipolar disorder

David G. Ashbrooka*, Robert W. Williamsb, Lu Lub, c, Reinmar Hagera aComputational and Evolutionary Biology, Faculty of Life Sciences, University of

Manchester, Manchester, UK bDepartment of Genetics, Genomics and Informatics, University of Tennessee

Health Science Center, University of Tennessee, Memphis, TN, USA cJiangsu Key Laboratory of Neuroregeneration, Nantong University, China

Chapter 3: A cross-species genetic analysis identifies candidate genes for mouse anxiety and human bipolar disorder

Abstract

Bipolar disorder (BD) is a significant neuropsychiatric disorder with a lifetime prevalence of ~1%. To identify variants underlying BD genome-wide association studies (GWASs) have been carried out. While many variants of small effect have been identified few have yet been confirmed, partly because of the low power of GWASs due to multiple comparisons. Complementary mapping studies using murine models have identified loci for behavioural traits linked to BD, often with high power, but these regions often contain too many genes for clear identification of candidate genes. In this study we have aligned human BD

GWAS results and mouse linkage studies to help define and evaluate candidate genes linked to BD, seeking to use the power of the mouse mapping with the precision of GWASs.

We use quantitative trait mapping for open field test and elevated zero maze data in the largest mammalian model system, the BXD recombinant inbred mouse population, to identify genomic regions associated with these BD-like phenotypes. We then investigate these regions in whole genome data from the

Psychiatric Genomics Consortium’s bipolar disorder GWAS to identify candidate genes associated with BD. Finally we establish the biological relevance and pathways of these genes in a comprehensive systems-genetics analysis.

We identify four genes associated with both mouse anxiety and human

BD. While TNR is a novel candidate for BD, we can confirm previously suggested associations with CMYA5, MCTP1 and RXRG. A cross-species, systems- genetics analysis shows that MCTP1, RXRG and TNR co-express with genes linked to psychiatric disorders and identify the striatum as a potential site of action. We hypothesise that MCTP1, RXRG and TNR influence intercellular signalling in the striatum.

Chapter 3: A cross-species genetic analysis identifies candidate genes for mouse anxiety and human bipolar disorder

3.1 Introduction

Bipolar disorder (BD) is a neuropsychiatric disorder characterized by recurrent periods of mania and depression. Bipolar I disorder has a lifetime prevalence (i.e. at least one case of both depression and mania) of ~1%, although the bipolar spectrum (at least one episode of sub-threshold mania or hypomania) has a lifetime prevalence of up to 6.4% (Judd and Akiskal, 2003; Merikangas et al.,

2007). Twin studies estimate heritability of ~60-80%, indicating a substantial genetic component (McGuffin et al., 2003; Edvardsen et al., 2008; Lichtenstein et al., 2009; Wray and Gottesman, 2012).

To identify genetic variants underlying the disorder many genome-wide association studies (GWASs) have been conducted showing that BD is highly polygenic, with many single nucleotide polymorphisms (SNPs) each of small effect (Ferreira et al., 2008; Scott et al., 2009; Purcell et al., 2009; Psychiatric

GWAS Consortium Bipolar Disorder Working Group, 2011). Due to this polygenic nature large sample sizes are required to detect SNPs with genome-wide significance. Indeed, despite large cohorts of patients used, GWASs have found only ten SNPs which are strongly and consistently associated with the disorder

(Szczepankiewicz, 2013; Mühleisen et al., 2014), although many genes have been identified with lower confidence or using additional analyses. For example, a recent pathway analysis study has linked 226 genes to BD (Nurnberger et al.,

2014), however, approximately 5% of these would be expected to be false positives and the method is biased against genes of small size (Nurnberger et al., 2014). In order to understand the aetiology and biology of bipolar disorder it is critical to know the underlying causal variants. This allows us to link the disorder to specific proteins and pathways, potentially leading to novel treatments and a better ability to predict genetic predisposition.

Chapter 3: A cross-species genetic analysis identifies candidate genes for mouse anxiety and human bipolar disorder

GWASs in humans typically have modest statistical power due to high multiple testing corrections. However, loci are defined with high precision, often to the individual SNP level. By contrast, mouse linkage studies often have high statistical power to detect genetic effects but lower resolution, producing loci that include tens or hundreds of genes (Mackay et al., 2009; Ackert-Bicknell et al.,

2010; Hager et al., 2012; Wu et al., 2014). Combining data from mice and humans overcomes some of these problems, gaining power from mouse crosses and precision from human GWASs. This method also ensures translational relevance, as the same gene controlling similar phenotypes is found in a related species (Ashbrook et al., 2014). Moreover, the approach illustrates that the mouse homologue is relevant to the human phenotype, allowing research to be carried out on the gene to phenotype pathway which would not be possible in humans (Kas et al., 2007).

Mammalian model systems have been extensively used to investigate the genetic basis of disease traits through the experimental study of analogous behavioural or developmental traits (e.g. Hayes et al., 2014). We chose to investigate BD, as several of the symptoms are related to behaviours which can be measured in mice. For example, in manic patients typical behaviours include motor hyperactivity, increased risk taking and impulsive behaviour (Goodwin and

Jamison, 1990). Indeed, anxiety and activity measures in mice have previously been used to evaluate animal models of BD (Gould et al., 2001; Kirshenbaum et al., 2011). Furthermore, up to 93% of bipolar I disorder patients have a comorbid anxiety disorder at some stage of their life, and comorbidity between BD and anxiety results in significantly worse patient outcomes (Freeman et al., 2002;

MacKinnon, 2002; Boylan et al., 2004; Simon et al., 2004; Merikangas et al.,

2007; Goldstein and Levitt, 2008; Goodwin and Sachs, 2010; Vázquez et al.,

2014). Additionally, measures of global anxiety correlate well with time spent in

Chapter 3: A cross-species genetic analysis identifies candidate genes for mouse anxiety and human bipolar disorder depressive episodes in bipolar patients (Coryell et al., 2009).This suggests a shared underlying aetiology of anxiety and BD. Therefore, there is a clear relationship between symptoms of bipolar disorder (Goodwin and Sachs, 2010), behaviours disrupted in bipolar patients (Young et al., 2007; Perry et al., 2009;

Minassian et al., 2011) and measureable behaviours in mice (Figure 3.1).

Figure 3.1: Bipolar disorder traits and analogous mouse phenotypes. On the left, symptoms of the manic and depressive phases of bipolar disorder are shown, as well as behaviours which have been demonstrated to be disrupted in bipolar disorder patients.

On the right are mouse phenotypes which are measured by the zero maze and open field test, showing how the human traits link to the mouse traits.

Most organisms will balance risk against potential reward; whether to put greater emphasis on minimizing exposure to danger (being more anxious), or greater emphasis on reward (being more risk taking) (Marks and Nesse, 1994;

Bateson et al., 2011). This trade-off is reflected in a number of behavioural patterns, and individuals in a population differ in the degree to which anxiety-like behaviour is displayed (Erhardt and Spoormaker, 2013). Since normal and

Chapter 3: A cross-species genetic analysis identifies candidate genes for mouse anxiety and human bipolar disorder clinical anxiety exist upon a continuum (Bateson et al., 2011) and aspects of pathological anxiety and BD overlap in terms of aetiology and phenotype, it is reasonable to assume that some aspects of the underlying anxiety system should also overlap with BD, although true anxiety is more easily segregated from BD in humans.

Previous studies demonstrated the utility of using a cross-species approach to identify genes underlying specific traits (Koutnikova et al., 2009; de

Mooij-van Malsen et al., 2009; Poot et al., 2011; Leduc et al., 2011; Schofield et al., 2012; Ashbrook et al., 2014). This approach has the advantage that it allows the investigation of phenotypes without requiring experimental perturbations. We utilize data obtained from populations that segregated for large numbers of common sequence variants and associated differences in phenotype (Ashbrook et al., 2014). Here, we use data for two common measures of anxiety and activity in mouse, the elevated zero maze and the open field test, to identify QTL in the largest mammalian model system, both in terms of number of extant inbred lines available and the number of phenotypes analysed, the recombinant inbred mouse panel BXD (Wu et al., 2014). This identifies areas of the genome containing genetic variants which influence BD-like phenotypes in mice, and therefore may be influencing aspects of BD in humans. These genomic regions were then investigated in a large human GWAS of BD, the SNP summary of which has been made available online (Psychiatric GWAS Consortium Bipolar

Disorder Working Group, 2011).

We identify four genes (TNR, RXRG, MCTP1 and CMYA5) associated with anxiety in mice and risk of BD in humans. A systems-genetics approach suggests that TNR, RXRG and MCTP1 co-express with several other genes related to mental disorders in the striatum, providing a potential mechanism of action.

Chapter 3: A cross-species genetic analysis identifies candidate genes for mouse anxiety and human bipolar disorder

3.2 Materials and methods

3.2.1 Mouse and human data

The BXD recombinant inbred population consists of experimentally tractable and genetically defined mouse lines capturing a large amount of naturally occurring genetic variation, which underlies variation at the phenotypic level (e.g. Chesler et al., 2005; Mozhui et al., 2011; Hager et al., 2012). Over 80 BXD lines incorporate ~5 million segregating SNPs, 500000 insertions and deletions, and

55000 copy-number variants (Mozhui et al., 2011). These lines are used for complex systems-genetics analyses integrating massive phenotype and gene expression data sets obtained in different studies (Andreux et al., 2012; Hayes et al., 2014). For our analysis we used data for two murine phenotypes, the elevated zero maze and open field test (Philip et al., 2010) available in

GeneNetwork (genenetwork.org) that measure a combination of anxiety, exploration and activity (Henry et al., 2010), all of which are altered in BD

(Goodwin and Jamison, 1990; Young et al., 2007). We selected traits with at least one significant QTL (p ≤ 0.05) at the genome-wide level. Significance was calculated within GeneNetwork, using 5000 permutations of trait values and genotypes. A p-value of 0.05 is defined as an LRS score greater than 95% of the permuted datasets (Wang et al., 2003).

Human GWAS data for BD were obtained from the Psychiatric Genomics

Consortium (PGC; https://pgc.unc.edu), containing 11974 BD cases and 51792 controls (Psychiatric GWAS Consortium Bipolar Disorder Working Group, 2011).

The majority of cases were of BD type 1 although also included BD type 2, schizoaffective disorder bipolar type and individuals with other bipolar diagnoses

(Psychiatric GWAS Consortium Bipolar Disorder Working Group, 2011). The

Knowledge-Based Mining System for Genome-wide Genetic Studies (KGG;

Chapter 3: A cross-species genetic analysis identifies candidate genes for mouse anxiety and human bipolar disorder http://statgenpro.psychiatry.hku.hk/limx/kgg; version 3.5) was used to convert these SNP p-values to gene p-values, using both the GATES and the hybrid set- based test (HYST) methods (Li et al., 2010, 2011, 2012). The GATES test is a

Simes test extension that is valid for correlated SNPs, and, of the two methods, is more powerful for genes with one or few independent causal variants (Li et al.,

2011). HYST combines the GATES test and the scaled chi-square test

(Moskvina et al., 2011) to examine the overall association significance in a set of

SNPs. This test is more powerful for genes with a number of independent causal variants. Both tests are advantageous as they require only summary GWAS and ancestral population linkage disequilibrium data, rather than raw data. Gene locations were from the Hg18 genome build. We used the complete set of data from the Psychiatric Genomics Consortium GWAS for BD

(http://www.med.unc.edu/pgc/downloads; pgc.bip.full.2012-04.txt), and linkage disequilibrium data from the CEPH (Utah residents with ancestry from northern and western Europe; CEU) Hapmap population dataset

(http://hapmap.ncbi.nlm.nih.gov/downloads/ld_data/?N=D). For each gene in the

Hg18 genome build a GATES and HYST p-value was calculated for associated with BD.

For the joint mouse – human analysis, the human homologues of genes within the mouse QTL were identified using Homologene

(http://www.ncbi.nlm.nih.gov/homologene). To assess if any particular gene is associated with both anxiety in mouse and BD, we examined both the GATES derived and HYST derived p-values of human homologues of genes within the significant BXD QTL. The human GWAS significance values were Bonferroni corrected for multiple comparisons using the total number of homologous genes compared and the number of tests used (two: HYST and GATES) (de Mooij-van

Malsen et al., 2009; Ashbrook et al., 2014).

Chapter 3: A cross-species genetic analysis identifies candidate genes for mouse anxiety and human bipolar disorder

3.2.2 Evolutionary conservation

A highly conserved nucleotide sequence between mouse and human would support that the protein plays the same role in both species. For each gene the protein it codes for was identified using the Entrez gene database

(http://www.ncbi.nlm.nih.gov/gene) and the NCBI reference sequence for the protein identified. For each pair of homologous proteins a Protein BLAST was carried out with default settings

(http://blast.ncbi.nlm.nih.gov/Blast.cgi?PROGRAM=blastp&PAGE_TYPE=BlastS earch&LINK_LOC=blasthome). The reference sequences used were: RXRG human NP_001243499.1, mouse NP_001153203.1; TNR human NP_003276.3, mouse NP_071707.2; MCTP1 human NP_001002796.1, mouse NP_084450.2;

CMYA5 human NP_705838.3, mouse NP_076310.2.

3.2.3 Identification of areas of expression

To identify the regions in the mouse brain in which our candidate genes are expressed we used images provided by the Allen Brain Atlas showing patterns of gene expression (http://mouse.brain-map.org) throughout the adult mouse brain for > 20000 genes, using in situ hybridization (Lein et al., 2007). For the human data, the BrainSpan Atlas of the Developing Human Brain (Miller et al., 2014) contains RNA-seq gene expression data at different life stages. There are a total of 578 samples, covering an age range from prenatal to 40 years and taken from

26 different brain regions. To identify locations in the human brain where our candidate genes are expressed we used gene expression heatmaps that show levels of gene expression in different brain regions (http://www.brainspan.org).

3.2.4 ‘Guilt-by-association’

To establish if specific genes are co-expressed in humans we used GeneFriends

(van Dam et al., 2012; http://genefriends.org/microArray) which contains

Chapter 3: A cross-species genetic analysis identifies candidate genes for mouse anxiety and human bipolar disorder microarray data (4164 microarray datasets containing 26113 experimental conditions and 19080 genes) from the Gene Expression Omnibus (van Dam et al., 2012; Barrett et al., 2013). This enabled us to find genes that are commonly co-expressed (i.e. genes with a co-expression value ≥ 0.50 and a p-value ≤ 0.05) with a submitted gene list. A co-expression value of 0.50 indicates that a particular gene is increased in expression at least 2-fold in 50% of the cases that a target gene is increased in expression at least 2-fold.

However, GeneFriends is not tissue or treatment specific, and therefore can only show the genes that are co-expressed together, not when or where.

This common co-expression suggests that the genes are under the same regulation. The list of commonly co-expressed genes was analysed using

WebGestalt, producing annotations for these genes. This ‘guilt-by-association’ approach enables us to identify the biological networks of our candidates (van

Dam et al., 2012). WebGestalt (http://bioinfo.vanderbilt.edu/webgestalt) is a web- based enrichment analysis tool that incorporates information from online sources including Gene Ontology (GO), KEGG pathways, Wikipathways, Pathways

Commons and disease association analysis (Zhang et al., 2005; Wang et al.,

2013). The lists of genes generated by our ‘guilt-by-association’ analysis were submitted to WebGestalt to identify pathways or diseases that our candidate genes may be involved in. Significance of enrichment was determined by the

Benjamini & Hochberg method of multiple test adjustment (Benjamini and

Hochberg, 1995) as implemented in WebGestalt, and the whole human genome was used as the background set of genes.

Shared function of genes can be established using co-expression analysis (Allocco et al., 2004). We analysed co-expression in adult mouse brain by producing Pearson product-moment correlation matrices of striatal and hippocampal gene expression (Chesler et al., 2003) as implemented in

Chapter 3: A cross-species genetic analysis identifies candidate genes for mouse anxiety and human bipolar disorder

GeneNetwork. The striatum was chosen as it is an area where several of our genes are expressed, whereas the hippocampus was chosen to identify if any co-expression seen in the striatum was seen in other brain regions. Expression data was obtained from GeneNetwork HQF Striatum Exon (Feb09) RMA data

(GN163) and UMUTAffy Hippocampus Exon (Feb09) RMA (GN206). Since both used the same microarray we could directly compare the data. Probes for exons were used unless no exon probes were available, in which case introns were investigated for consistent co-expression within a given gene. Co-expression between our candidate genes and genes previously associated with BD (as identified in Szczepankiewicz, 2013; Mühleisen et al., 2014), and other mental disorders was calculated in GeneNetwork using Pearson correlations. Multiple comparisons were corrected for by dividing the p-value obtained in the correlation analysis by the number of probes for our genes of interest (Cmya5,

Mctp1, Rxrg, Tnr). Probes were said to co-express if they have an r ≥ 0.5 or ≤ -

0.5 and an adjusted p ≤ 0.05. This allowed us to establish if our candidate genes co-express with genes known to be associated with neuropsychiatric disorders, and whether this is specific to the striatum.

3.2.5 Principal component analysis

Principal component analysis (PCA) was used to jointly analyse multiple phenotypes including gene expression. PCA reduces the dimensionality of data and captures the shared variability between traits. If the first principal component

(PC1) explains a high proportion of the variability it can be used as a synthetic trait, capturing the main common source of variation within the traits (Mozhui et al., 2011). PCA was carried out in GeneNetwork to find the PC1 of striatal expression of our candidate gene MCTP1 and the PC1 of the open field test phenotypes.

Chapter 3: A cross-species genetic analysis identifies candidate genes for mouse anxiety and human bipolar disorder

3.2.6 Biological networks

In order to establish the biological networks of our candidate genes we utilized the massive phenotype and expression data sets available for the BXD panel

(Gini and Hager, 2012; Ashbrook and Hager, 2013). Generally, microarray analysis uses a number of probes, targeted at different parts of a gene. Exon level microarrays, such as those used above in the co-expression analysis, can be used to calculate expression of exons, introns or untranscribed regions

(UTRs). Consistent correlation between different probes for a gene and a phenotype would suggest that the gene is associated with the phenotype.

First, using the same exon level data as used for the co-expression analysis, probes for each of our four candidate gene were examined for high

Pearson correlation (r ≥ 0.5, p ≤ 0.05). For each gene, this produced a group of highly correlated probes. Each of these highly correlated probes was individually correlated against BXD phenotypes, which yielded a list of phenotypes that correlated with that probe (r ≥ 0.5 or ≤ -0.5 and p ≤ 0.05). Next, the phenotypes which correlated with all the probes for a particular candidate gene were identified, to produce a smaller list of phenotypes which consistently correlated with all the probes for that gene.

Chapter 3: A cross-species genetic analysis identifies candidate genes for mouse anxiety and human bipolar disorder

3.3 Results

Figure 3.2: Graphical representation of the research method used, providing a summary of the main findings.

Chapter 3: A cross-species genetic analysis identifies candidate genes for mouse anxiety and human bipolar disorder

In Figure 3.2 we give a graphical summary of our results. We begin by identifying

QTL for the open field test and elevated zero maze in the BXD recombinant inbred lines. We then investigate the homologous genomic regions in human bipolar disorder GWAS data and thus identify four candidate genes (TNR,

RXRG, MCTP1 and CMYA5). The protein products of three of these genes

(TNR, RXRG and MCTP1) are highly conserved (≥ 90% identical), and are expressed in the adult striatum of both mouse and human. In non-tissue specific human gene expression data these three genes co-express with genes that are associated with mental disorders. In the BXD mouse striatum the three genes co- express together, as well as with known mental disorder related genes. Finally, when we correlate striatal expression of all four of our candidate genes against the large BXD phenotype dataset, we find that Tnr, Rxrg and Mctp1 expression correlates with dopamine related traits, whereas Cmya5 expression correlates with anxiety- and depression-like traits.

3.3.1 QTL for bipolar related phenotypes in mice

Bipolar disorder is characterized by disrupted anxiety, activity and exploration

(Goodwin and Jamison, 1990; Goodwin and Sachs, 2010; Figure 3.1) and we thus investigated genetic variation in those traits, focusing on the elevated zero maze and the open field test. Both of these are used to measure anxiety, activity and exploration in mice (Shepherd et al., 1994; Chauhan et al., 2005; Kalueff et al., 2006; Ariyannur et al., 2013). We limited the datasets to only those with no or saline only treatment, and with at least one genome-wide significant QTL (p ≤

0.05).

For the seven available elevated zero maze traits, QTL were consistently found on distal chromosome 1 (156.2-175.3 Mbp), containing 185 genes with human homologues (Table 3.1, Supplementary figure 3.1). Therefore the

Chapter 3: A cross-species genetic analysis identifies candidate genes for mouse anxiety and human bipolar disorder significance threshold for human genes is p = 1.4E-4 (185 genes, each compared against both the GATES and HYST significance values; 0.05/185*2).

However the size of the above region differed between the datasets, and appears to separate into two smaller QTL, based on phenotypes showing either two peaks, with a non-significant region between them, or a significant peak at only one of the two QTL (Figure 3.3A; Supplementary table 3.1; Supplementary figure 3.1).

100

Chapter 3: A cross-species genetic analysis identifies candidate genes for mouse anxiety and human bipolar disorder

Table 3.1: Activity traits in the elevated zero maze which have a significant QTL.

GeneNetwork ID, phenotype and number of lines phenotyped are from GeneNetwork, number of genes refers to the number of genes within the QTL which have human homologues. Position identifies the chromosome and megabase pair (Mbp) location of the gene on mm9.

ID Phenotype Position of genes

(mm9)

Numberof lines phenotyped Numberof genes 12359 Central nervous system, behavior: Anxiety 74 3 Chr1:160.6- assay, baseline untreated control (BASE 161.5 group), activity in closed quadrants using an elevated zero maze in 60 to 120-day-old males and females during first 5 min [n beam breaks] 12361 Central nervous system, behavior: Anxiety 74 8 Chr1:160.6- assay, baseline untreated control (BASE 162.2 group), activity in closed quadrants using an elevated zero maze in 60 to 120-day-old males and females during 10 min [n beam breaks] 12409 Central nervous system, behavior: Anxiety 73 30 Chr1:156.2- assay, saline treated [0.18 ml/kg i.p.] (NOS 161.5 group), activity in closed quadrants using an elevated zero maze in 60 to 120-day-old males 44 Chr1:165.3- only during first 5 min [n beam breaks] 171.0 12411 Central nervous system, behavior: Anxiety 73 30 Chr1:156.2- assay, saline treated [0.18 ml/kg i.p.] (NOS 161.5 group), activity in closed quadrants using an 7 Chr1:169.1- elevated zero maze in 60 to 120-day-old males 170.0 only during 10 min [n beam breaks] 12419 Central nervous system, behavior: Anxiety 75 60 Chr1:156.2- assay, saline treated [0.18 ml/kg i.p.] (NOS 164.9 group), activity in closed quadrants using an elevated zero maze in 60 to 120-day-old males 9 Chr1:169.1- and females during first 5 min [n beam breaks] 171.5 12420 Central nervous system, behavior: Anxiety 75 31 Chr1:156.2- assay, saline treated [0.18 ml/kg i.p.] (NOS 162.0 group), activity in closed quadrants using an elevated zero maze in 60 to 120-day-old males and females during last 5 min [n beam breaks] 12421 Central nervous system, behavior: Anxiety 75 61 Chr1:157.8- assay, saline treated [0.18 ml/kg i.p.] (NOS 165.0 group), activity in closed quadrants using an elevated zero maze in 60 to 120-day-old males 9 Chr1:169.1- and females during 10 min [n beam breaks] 171.5

101

Figure 3.3: QTL maps showing two QTL within each of the regions we identified. A)

Chromosome 1, 155-175 Mbp, elevated zero maze traits (GN ID 12419, 12361 and 12409; full details in Table 3.1). B) Chromosome 13, 70-100 Mbp, open field test traits 11606, 11759, 11607 (Full details in Table 3.2). This shows that some traits have two peaks, with a non-significant region between, while other traits have significant peaks only at one end or the other of the regions. The higher red line indicates genome-wide significance, (genome-wide p-value of ≤ 0.05) with the blue line showing the significance of the trait at each position. The lower red line indicates the additive coefficient, i.e. that C57BL/6J alleles increase trait values, with the scale in green on the left. This increase caused by the C57BL/6J allele is consistent across all the phenotypes. The coloured blocks at the top of the figure show the positions of genes. The orange track at the bottom of each map is the SNP

Seismograph track, showing the positions of SNPs within the BXD lines. The location of our four candidate genes are shown, with the width of the line representing the size of the gene.

102

The first QTL is located between 156.2 - 161.5 Mbp containing 30 homologous genes (significance threshold for human genes p-value = 8.3E-4).

Only one gene, TNR, has a p-value lower than this threshold (GATES p-value

7.48E-4; 161.4 Mbp). The second QTL is between 169.1 - 171.5 Mbp, containing

9 homologous genes, with a p-value threshold of ≤ 2.7E-3. Only one gene has a lower p-value, RXRG (GATES p-value 9.63E-4; 169.5 Mbp). These two genes are the most significant genes within this chromosome 1 region, and their positions line up well with the LRS peaks in the QTL map (Figure 3.3A).

Consistent QTL for the 34 open field test phenotypes are found on chromosome 13 (73 - 97 Mbp), containing 63 homologous genes. The threshold for human genes is thus p ≤ 3.96E-4. One gene has a p-values lower than this threshold: CMYA5 (HYST p = 1.57E-4; 94.2 Mbp). Interestingly, this larger region again appears to segregate into two smaller QTL (Table 3.2; Figure 3.3B;

Supplementary figure 3.2), the first from 73.3 - 84 Mbp (41 homologous genes, p- value threshold = 6.1E-4) and the second from 93.9 - 95.2 Mbp (12 homologous genes, p-value threshold = 2.1E-3). This reveals another significant gene,

MCTP1 (HYST p = 4.59E-4; 76.5 Mbp). Again, these are the two most significant genes within the region, and their positions match those of the LRS peaks on the

QTL maps (Figure 3.3B).

Further, we investigated the conservation of the amino acid sequence in the proteins which our candidate genes code for using Protein BLAST. This revealed that the proteins are conserved between mouse and human (RXRG

99% identical; TNR 95% identical; MCTP1 90% identical; CMYA5 68% identical).

The strong conservation of RXRG, TNR and MCTP1 supports that they may have an evolutionarily conserved function.

103

Chapter 3: A cross-species genetic analysis identifies candidate genes for mouse anxiety and human bipolar disorder

Table 3.2: Vertical activity traits in the open field which have a significant

QTL. GeneNetwork ID, phenotype and number of lines phenotyped are from

GeneNetwork, number of genes refers to the number of genes within the QTL which have human homologues. Position identifies the chromosome and megabase pair (Mbp) location of the gene on mm9.

ID Phenotype Position of

genes (mm9)

Number of Number lines phenotyped of Number genes 11350 Central nervous system, behavior: Novel open field 63 25 Chr13:74.7- behavior, vertical activity (rears) from 15-30 min for males [n 82.0 beam breaks] 11351 Central nervous system, behavior: Novel open field 63 38 Chr13:74.0- behavior, vertical activity (rears) from 30-45 min for males [n 82.0 beam breaks] 11352 Central nervous system, behavior: Novel open field 63 38 Chr13:74.0- behavior, vertical activity (rears) from 45-60 min for males [n 82.0 beam breaks] 11355 Central nervous system, behavior: Novel open field 63 41 Chr13:73.9- behavior, vertical activity (rears) in the center from 0-60 min 84.0 for males [n beam breaks] 11606 Central nervous system, behavior: Novel open field 64 10 Chr1:97.0- behavior, vertical activity (rears) from 0-15 for females [n 107.2 beam breaks] 10 Chr13:93.9- 94.9 11607 Central nervous system, behavior: Novel open field 64 37 Chr13:75.2- behavior, vertical activity (rears) from 15-30 min for females 95.2 [n beam breaks] 11608 Central nervous system, behavior: Novel open field 64 24 Chr13:75.2- behavior, vertical activity (rears) from 30-45 min for females 83.7 [n beam breaks] 11609 Central nervous system, behavior: Novel open field 64 28 Chr9:73.3- behavior, vertical activity (rears) from 45-60 min for females 78.0 [n beam breaks] 26 Chr13:74.7- 84.0 11612 Central nervous system, behavior: Novel open field 64 38 Chr13:74.7- behavior, vertical activity (rears) in the center from 0-60 min 94.9 for females [n beam breaks]

11759 Central nervous system, behavior: Novel open field 64 17 Chr13:75.8- behavior, vertical activity (rears) from 0-60 min in the center 78.7 for females [n beam breaks]

11772 Central nervous system, behavior: Novel open field 64 17 Chr13:75.8- behavior, vertical activity (rears) from15-30 min in the center 78.7 for females [n beam breaks]

11773 Central nervous system, behavior: Novel open field 64 15 Chr13:75.8- behavior, vertical activity (rears) from 30-45 min in the center 78.4 for females [n beam breaks]

11784 Central nervous system, behavior: Novel open field 64 22 Chr13:75.8- behavior, vertical activity (rears) in the periphery from 15-30 82.0 min for females [n beam breaks]

11785 Novel open field behavior, vertical activity (rears) in the 64 24 Chr13:75.2- periphery from 30-45 min for females [n beam breaks] 83.7 11786 Central nervous system, behavior: Novel open field 64 26 Chr13:74.7- behavior, vertical activity (rears) in the periphery from 45-60 83.6 min for females [n beam breaks]

104

Chapter 3: A cross-species genetic analysis identifies candidate genes for mouse anxiety and human bipolar disorder

11790 Central nervous system, behavior: Novel open field 64 26 Chr13:74.7- behavior, vertical activity (rears) in the periphery from 0-60 83.6 min for females [n beam breaks]

11793 Central nervous system, behavior: Saline control response 64 25 Chr13:74.7- (10 ml/kg), open field behavior, vertical activity (rears) 0-60 82.0 min after injection for females [n beam breaks/60 min] 11806 Central nervous system, behavior: Saline control response 64 48 Chr13:74.7- (10 mg/kg ip), open field behavior, vertical activity (rears) 0- 96.3 15 min after injection for females [n beam breaks/15 min]

11807 Central nervous system, behavior: Saline control response 64 17 Chr13:75.2- (10 mg/kg ip), open field behavior, vertical activity (rears) 15- 78.7 30 min after injection for females [n beam breaks/15 min] 11808 Central nervous system, behavior: Saline control response 64 17 Chr13:75.2- (10 mg/kg ip), open field behavior, vertical activity (rears) 30- 78.7 45 min after injection for females [n beam breaks/15 min] 11809 Central nervous system, behavior: Saline control response 64 17 Chr13:75.2- (10 mg/kg ip), open field behavior, vertical activity (rears) 45- 78.7 60 min after injection for females [n beam breaks/15 min] 11863 Central nervous system, behavior: Novel open field 64 8 Chr1:97.0- behavior, vertical activity (rears) from 0-15 min for males and 107.2 females [n beam breaks] 9 Chr13:93.9- 94.6 11864 Central nervous system, behavior: Novel open field 64 39 Chr13:74.4- behavior, vertical activity (rears) from 15-30 min for males 94.6 and females [n beam breaks]

11865 Central nervous system, behavior: Novel open field 64 34 Chr13:74.2- behavior, vertical activity (rears) from min 30-45 for males 84.0 and females [n beam breaks]

11866 Central nervous system, behavior: Novel open field 64 2 Chr9:73.3- behavior, vertical activity (rears) from 45-60 min for males 74.0 and females [n beam breaks] 4 Chr9:77.2- 78.0 40 Chr13:73.9- 84.0 11869 Central nervous system, behavior: Novel open field 64 39 Chr13:74.0- behavior, vertical activity (rears) in the center from 0-60 min 84.0 for males and females [n beam breaks]

12042 Central nervous system, behavior: Novel open field 64 22 Chr13:75.8- behavior, vertical activity (rears) in the periphery from 30-45 82.0 min for males and females [n beam breaks]

12043 Central nervous system, behavior: Novel open field 64 22 Chr13:75.8- behavior, vertical activity (rears) in the periphery from 45-60 82.0 min for males and females [n beam breaks]

12047 Central nervous system, behavior: Novel open field 64 23 Chr13:75.8- behavior, vertical activity (rears) in the periphery from 0-60 84.0 min for males and females [n beam breaks]

12050 Central nervous system, behavior: Saline control response 64 23 Chr13:75.2- (10 ml/kg), open field behavior, vertical activity (rears) 0-60 82.0 min after injection for males and females [n beam breaks/60 min] 12063 Central nervous system, behavior: Saline control response 64 23 Chr13:75.8- (10 mg/kg ip), open field behavior, vertical activity (rears) 0- 84.0 15 min after injection for males and females [n beam breaks/15 min] 12064 Central nervous system, behavior: Saline control response 64 16 Chr13:75.8- (10 mg/kg ip), open field behavior, vertical activity (rears) 15- 79.0 30 min after injection for males and females [n beam breaks/15 min] 12065 Central nervous system, behavior: Saline control response 64 16 Chr13:75.8- (10 mg/kg ip), open field behavior, vertical activity (rears) 30- 79.0 45 min after injection for males and females [n beam breaks/15 min] 12066 Central nervous system, behavior: Saline control response 64 14 Chr13:75.8- (10 mg/kg ip), open field behavior, vertical activity (rears) 45- 78.2 60 min after injection for males and females [n beam breaks/15 min]

105

Chapter 3: A cross-species genetic analysis identifies candidate genes for mouse anxiety and human bipolar disorder

3.3.2 Gene expression in the brain

The Allen Brain Atlas indicates that Rxrg and Tnr are expressed throughout the mouse brain (Lein et al., 2007) but that Cmya5 has high expression only in the cerebellum. For Mctp1 no area shows strong expression in the mouse brain atlas. Human RNA-seq data from BrainSpan (Miller et al., 2014), however, suggests an increase in expression of MCTP1 in the striatum. TNR appears to have a similar level of expression throughout the brain. Interestingly, RXRG also shows high expression in the striatum while CMYA5 shows low expression levels throughout the brain.

3.3.3 ‘Guilt-by-association’

Next we used GeneFriends to identify genes which commonly co-express with our candidates, irrespective of tissue or treatment. GeneFriends uses the gene expression omnibus database and calculates a co-expression value and a p- value for this co-expression. Co-expression of a gene with two of our candidates suggests it is involved in their shared function (as a single gene may have several functions, only some of which are shared). We can then use WebGestalt to look for enriched annotations in these co-expressed genes, which may indicate how our candidate genes function to influence BD. CMYA5 had no commonly co-expressed genes. However, MCTP1 and RXRG have 13 commonly co-expressed genes (Supplementary table 3.2), which are significantly enriched for relevant disease annotations (e.g. ‘Brain diseases’ and ‘Mental disorders’; Supplementary table 3.3) and several relevant GO annotations (e.g.

‘Synaptic transmission’; Supplementary table 3.4).

RXRG and TNR share 797 commonly co-expressed genes

(Supplementary table 3.5). This shows enrichment for 10 diseases, all of which are highly relevant to neuropsychiatric disorders (e.g. ‘Substance-related

106

Chapter 3: A cross-species genetic analysis identifies candidate genes for mouse anxiety and human bipolar disorder disorders’ and ‘Mental disorders’; Supplementary table 3.6). GO and pathway annotations (Supplementary table 3.7, Supplementary table 3.8, Supplementary table 3.9 and Supplementary table 3.10) show a clear relationship to intercellular signalling (e.g. ‘Synaptic transmission’, ‘Neuroactive ligand-receptor interaction’ and ‘Signaling by G protein-coupled receptors’). Interestingly, MCTP1 and TNR share no commonly co-expressed genes. This suggests that although genes related to, for example, mental disorders are enriched in the overlap between

TNR and RXRG, and in the overlap between MCTP1 and RXRG, they are not the same genes, i.e. both sets of genes are independently related to mental disorders.

3.3.4 Co-expression analysis

The GeneFriends analysis suggests that MCTP1 and RXRG co-express with mental disorder related genes. BrainSpan shows that both MCTP1 and RXRG are expressed in the striatum, and it has been found that striatal expression of

Rxrg effects depression-like behaviours in mice (Krzyzosiak et al., 2010).

Therefore we analysed striatal expression of our four candidate genes in

GeneNetwork, using the HQF Striatum Exon RMA data (GN163). This data for mouse striatal exon probes shows clear co-expression between Mctp1, Rxrg and

Tnr (Supplementary table 3.11). We could not investigate Cmya5 exons, as no exon probes were available. However, there are seven intronic probes for Cmya5 which show a significant cis-eQTL and correlate strongly with each other (r ≥

0.5). These do not correlate with Mctp1, Rxrg or Tnr in the striatum.

We further investigated if known mental disorder genes, which commonly co-express with our candidates in the GeneFriends data, also co-express specifically in the striatum. Therefore we built a Pearson correlation matrix including our candidate genes (Cmya5, Mctp1, Rxrg and Tnr), genes commonly

107

Chapter 3: A cross-species genetic analysis identifies candidate genes for mouse anxiety and human bipolar disorder co-expressed with MCTP1 and RXRG or TNR and RXRG within the mental disorders and CNS Diseases annotations (Supplementary table 3.3 and

Supplementary table 3.6), and ten genes containing SNPs which have been strongly associated with BD (SLC6A4, BDNF, DAOA, DTNBP1, NRG1, DISC1,

CACNA1C, ANK3, ODZ4, COMT; Szczepankiewicz, 2013; Mühleisen et al.,

2014). Since there is strong evidence for shared genetics between mental disorders it is likely that we will see an overlap (Doherty and Owen, 2014). As above, we only used probes specific to exons, with the exception of Cmya5. An identical table was built for the hippocampus, as the same mRNA microarray has been used and therefore probes were directly comparable. Multiple comparisons were corrected for by dividing the p-value of the correlation by the number of probes (n = 63) targeting our four candidates (Cmya5, Mctp1, Rxrg, Tnr). Probes were said to co-express if they have an r ≥ 0.5 or ≤ -0.5 and an adjusted p ≤ 0.05

(p = 7.94E-4; r = 0.554 or r = -0.554 in striatum).

The tables for striatum (Supplementary table 3.11) and hippocampus

(Supplementary table 3.12) show that there are many more co-expressing probes in the striatum than in the hippocampus. Genes with probes which consistently co-express with several probes for the candidate genes are shown in

Supplementary table 3.13.

Interestingly, Cmya5 shows no correlation with any of these genes, either in the adult hippocampus or striatum, indicating that it may be acting in another brain region or during development. The other three genes show strong co- expression with each other, and with other genes related to mental disorders in the striatum, but not in the hippocampus.

108

Chapter 3: A cross-species genetic analysis identifies candidate genes for mouse anxiety and human bipolar disorder

3.3.5 Correlation analysis

Next, we correlated the expression of our four candidates with the behavioural phenotypes shown in Tables 3.1 and 3.2. This showed a correlation between several Mctp1 probes (Supplementary table 3.11) and the open field test phenotypes (Table 3.2). We can sum up each of these groups of related traits into two principal components, i.e. the first principal component for the Mctp1 probes and the first principal component for the open field traits and look at the correlation between these two. The open field PC1 explains ~75% of the variance between traits and the Mctp1 PC1 explains ~60% of the variance.

These two principal components show a Pearson’s correlation of 0.562, p =

1.86E-3. This correlation suggests that there may be a link between expression level of Mctp1 in the striatum and vertical activity in the open field test.

We next investigated Pearson correlations between expression of our candidates and the GeneNetwork phenotype database. Each of the four candidates has multiple probes for striatal expression, which were correlated against each other. We then selected those probes that showed a significant correlation (r ≥ 0.5) with at least one other probe (Supplementary table 3.11).

These selected probes were then correlated against the GeneNetwork database of phenotypes. Supplementary table 3.14 shows that the expression of Mctp1,

Rxrg and Tnr correlates with dopamine related gene expression in the striatum.

Further, we found that Cmya5 expression correlates with anxiety-like and depression-like behaviour (Supplementary table 3.14). Interestingly, when these phenotypes were correlated against our target phenotypes (i.e. open field test and elevated zero maze traits; Supplementary table 3.15), there was no significant correlation with dopamine related gene expression. However, this may be due to a much lower number of overlapping samples (n ≤ 27 rather than n ≥

63).

109

Chapter 3: A cross-species genetic analysis identifies candidate genes for mouse anxiety and human bipolar disorder

These results suggest that Mctp1, Rxrg and Tnr may be acting within the same striatal dopamine network, whereas Cmya5 is acting elsewhere, but still has an influence on anxiety and depression related phenotypes.

110

Chapter 3: A cross-species genetic analysis identifies candidate genes for mouse anxiety and human bipolar disorder

3.4 Discussion

This study identified four candidate genes associated with anxiety in mice and

BD in humans, CMYA5, MCTP1, RXRG and TNR. The association between TNR and BD is a novel finding. In mice, three of our candidates (Mctp1, Rxrg and Tnr) co-express in the striatum with other genes related to mental disorders. This suggests that these three genes are part of a pathway which is shared between neuropsychiatric disorders, and which involves the striatum.

Interval mapping in the BXD mouse set produces two regions, within which the QTL for the open field traits (34 measures) and the elevated zero maze traits (7 measures) are located (open field at Chr1:156.2 - 171.5 Mbp and elevated maze at Chr13:73 - 97 Mbp). Our results suggest that the QTL defining these two regions are actually made up of two loci rather than one. Firstly, the

QTL for some traits are found only in a small part of each region, or have two significant loci separated by a non-significant region (Tables 3.1 and 3.2; Figure

3.3). Secondly, we find one gene associated with BD at each end of these two regions.

Distal chromosome 1 has been linked to a wide range of phenotypes, many of them neuropsychiatric related (Flint et al., 1995; Gershenfeld et al.,

1997; Wehner et al., 1997; Turri et al., 2001a, 2001b; Talbot et al., 2003; Yalcin et al., 2004; Singer et al., 2005; Valdar et al., 2006; Ponder et al., 2007; Mozhui et al., 2008; Eisener-Dorman et al., 2010; Vogel et al., 2013). Genes influencing related phenotypes can often collocate in the genome, e.g. (Legare et al., 2000).

Often, when a trait is mapped to a single QTL, the locus actually contains several

QTL, each with a small effect size (Flint et al., 2005; Valdar et al., 2006).

Therefore our finding of an elevated zero maze QTL on chromosome 1 fits well with the literature.

111

Chapter 3: A cross-species genetic analysis identifies candidate genes for mouse anxiety and human bipolar disorder

We find that our genes have previously been associated with schizophrenia, which agrees with an identified overlap in the phenotypes and genetics underlying the two disorders, as well as other psychiatric disorders

(Craddock et al., 2006; Owen et al., 2007; Purcell et al., 2009; Lichtenstein et al.,

2009; Lee et al., 2013; Cardno and Owen, 2014; Pearlson and Ford, 2014;

Gratten et al., 2014; Ruderfer et al., 2014). All four candidates have been associated generally with neuropsychiatric disorders, or psychiatric-like behaviour, before. Mouse Tnr knock-outs have decreased motivation and increased anxiety (Freitag et al., 2003), a depression-like phenotype. Further

TNR appears in a GWAS of efficacy of an antipsychotic, iloperidone, for treating schizophrenia (Lavedan et al., 2009). There is some previous association between RXRG and BD, as well as other disorders, such as schizophrenia (Le-

Niculescu et al., 2008, 2009). RXRG has been associated with sensation seeking in humans (Alliey-Rodriguez et al., 2011) and ablation of Rxrg in mice leads to depression-like behaviour (Krzyzosiak et al., 2010). CMYA5 has been associated with schizophrenia (Watanabe et al., 2014; Wang et al., 2014), depression

(Wang et al., 2014) and BD (Nurnberger et al., 2014).CMYA5 interacts with dysbindin (Benson et al., 2004), which has been linked to both schizophrenia and

BD (Breen et al., 2006; Pae et al., 2007; Joo et al., 2007). MCTP1 has already been suggested to be associated with BD, but this result was non-significant in their data (Scott et al., 2009).

Finally, we show that three of our four genes (Mctp1, Rxrg and Tnr) co- express with other genes known to be involved in mental disorders in the striatum, but not in the hippocampus, of adult BXD mice.

112

Chapter 3: A cross-species genetic analysis identifies candidate genes for mouse anxiety and human bipolar disorder

3.4.1 Biological function of candidate genes

Our evidence identifies the striatum as a point of convergence between three of our candidate genes, MCTP1, RXRG and TNR. The striatum is a relevant brain region in BD, since the diagnostic symptoms of mania include increased risk- taking and deficits in goal regulation (Johnson, 2005; Goodwin and Sachs, 2010), and the striatum is involved in both of these (Balleine and O’Doherty, 2010;

Bartra et al., 2013; Mason et al., 2014). Indeed, when making choices about risky decisions, BD patients have increased activity in the nucleus accumbens, part of the ventral striatum (Mason et al., 2014). This may relate back to the mouse phenotypes, as they have to balance the rewards of exploring an exposed area

(e.g. mating opportunities) against the risks (e.g. predation), a system that is necessary for all animals and therefore likely to be conserved.

3.4.2 RXRG

Rxrg knockout mice show a reduction in ambulatory activity (Krezel et al., 1998), accompanied by reduction in striatal dopamine receptor expression (Krezel et al., 1998; Krzyzosiak et al., 2010). This alteration is due to direct transcriptional regulation of Drd2 by retinoid receptors, such as RXRG (Samad et al., 1997;

Krezel et al., 1998; Krzyzosiak et al., 2010). In another Rxrg knockout, expression of choline acetyltransferase in striatal cholinergic interneurons was reduced and response to antipsychotic dopamine antagonists was altered (Saga et al., 1999).

These studies show that loss of RXRG signalling leads to depression-like behaviour in mice, and indicates that decreased dopamine signalling in the striatum plays a critical role in this (Krzyzosiak et al., 2010). Additionally, striatal dopamine receptor expression is clearly involved in inhibition and behavioural control (Lawrence et al., 1998; Cropley et al., 2006; Dalley et al., 2007; Pattij et

113

Chapter 3: A cross-species genetic analysis identifies candidate genes for mouse anxiety and human bipolar disorder al., 2007; Hamidovic et al., 2009; Eagle et al., 2011), a trait disrupted in the disorder. This supports that mutations in RXRG potentially play a role in BD.

Since both Mctp1 and Tnr expression correlated with dopamine traits and with

Rxrg, it follows that they may be part of this same network.

3.4.3 MCTP1

MCTP1 is an understudied gene, however one of the studies has demonstrated that MCTP1 is a calcium binding protein (Shin et al., 2005). It has been suggested that MCTP1’s calcium binding properties may be involved in BD (Scott et al., 2009) as other calcium-related genes are associated with BD, such as

CACNA1C (Ferreira et al., 2008). We find that the calcium signalling pathway is enriched in genes commonly co-expressed with RXRG and TNR (Supplementary table 3.8), and disruptions in calcium signalling are part of the pathophysiology of bipolar disorder (Berridge, 2013, 2014). In relation to the above, MCTP1 could be part of an intracellular signalling pathway activated by RXRG or dopamine, or part of a calcium dependent system of regulating dopamine receptor expression, however this hypothesis remains to be tested.

3.4.4 TNR

Tnr knockout mice show decreased motivation to explore and increased anxiety

(Freitag et al., 2003). Further, knockout mice spent more time resting and less time eating/drinking (Freitag et al., 2003) and again, activity and appetite are altered in BD (Goodwin and Jamison, 1990). In contrast to this, another Tnr knockout experiment showed no increase in anxiety, but increased exploration, although with a different pattern than the wildtype, and an impaired ability to construct a goal-independent representation of space (Montag-Sallaz and

Montag, 2003), which may be linked to disruption of goal-orientated behaviours in bipolar patients. These differences may be due to different genetic

114

Chapter 3: A cross-species genetic analysis identifies candidate genes for mouse anxiety and human bipolar disorder backgrounds. Freitag et al. backcrossed their mice to C57BL/6J twice, while

Montag-Sallaz and Montag backcrossed to C57BL/6J ten times, and therefore this difference may be indicative of epistatic interactions

TNR is a major component of the perineuronal nets (PNNs) of inhibitory interneurons, including those in the striatum (Fuss et al., 1993; Hargus et al.,

2008). PNNs are part of the extracellular matrix and ensheath many CNS neurons and their axons and dendritic processes, but not the sites of synaptic contact (Ojima et al., 1995; Alpár et al., 2006; Brückner et al., 2006; Bitanihirwe and Woo, 2014). Specific removal of striatal PNNs in adult mice has direct behavioural consequences (Lee et al., 2012). Further, dysfunction of PNNs has been implicated in schizophrenia (Bitanihirwe and Woo, 2014). In Tnr knockout mice the composition and formation of PNNs is significantly altered (Brückner et al., 2000; Haunsø et al., 2000). Therefore dysregulation of TNR may affect the

PNNs of striatal cells and consequently connectivity between cells.

3.4.5 Role of striatum

Medium spiny neurons are the principal neurons of the striatum. Supplementary table 3.14 shows that Mctp1, Rxrg and Tnr expression correlates with gene expression signatures for medium spiny neurons. These GABAergic neurons are regulated by glutamatergic and dopaminergic neurons, and presynaptic receptors

(including kappa opioid and muscarinic receptors) regulate glutamatergic and dopaminergic transmission (McGinty, 1999). The postsynaptic glutamatergic, dopamine D1 and D2, and muscarinic receptor signals in the medium spiny neurons trigger a complex intracellular network, resulting in changes in gene and protein expression (McGinty, 1999; Surmeier et al., 2007). In our list of genes which co-express with Mctp1, Rxrg and Tnr in the striatum (Supplementary table

115

Chapter 3: A cross-species genetic analysis identifies candidate genes for mouse anxiety and human bipolar disorder

3.13) we find four GABA receptors, five glutamate receptors and the kappa 1 opioid receptor, further demonstrating a potential link to the above network.

Although the full details of the molecular mechanisms remain to be established, especially as different subtypes of striatal neurons express different genes (Surmeier et al., 2007), it is interesting to note that many of these intracellular cascades are Ca2+-dependent (McGinty, 1999). The latter may link back to MCTP1 and the role of calcium in bipolar disorder (Berridge, 2013,

2014). Further, medium spiny neurons are a major downstream target of parvalbumin-positive cells, which PNNs particularly co-localise with (Lee et al.,

2012).

3.4.6 Conclusion

The large number of commonly co-expressed genes between RXRG and TNR, and the enrichment of these genes in mental disorder related annotations, strongly suggests that they are part of the same mental disorder related network.

Synaptic transmission related genes are enriched in the commonly co-expressed genes shared between RXRG and MCTP1 and between RXRG and TNR.

Therefore, we hypothesise that disruption of MCTP1, RXRG or TNR alters the complex intercellular signalling within the striatum, leading to changes in intracellular signalling and gene expression. Overall, our study provides evidence for the association of CMYA5, MCTP1, RXRG and TNR with bipolar disorder.

116

Chapter 3: A cross-species genetic analysis identifies candidate genes for mouse anxiety and human bipolar disorder

3.5 References

Alliey-Rodriguez, N., Zhang, D., Badner, J. A., Lahey, B. B., Zhang, X., Dinwiddie, S., Romanos, B., Plenys, N., Liu, C., and Gershon, E. S. (2011). Genome-wide association study of personality traits in bipolar patients. Psychiatr. Genet. 21, 190–194. doi:10.1097/YPG.0b013e3283457a31.

Allocco, D. J., Kohane, I. S., and Butte, A. J. (2004). Quantifying the relationship between co-expression, co-regulation and gene function. BMC Bioinformatics 5, 18. doi:10.1186/1471-2105-5-18.

Alpár, A., Gärtner, U., Härtig, W., and Brückner, G. (2006). Distribution of pyramidal cells associated with perineuronal nets in the neocortex of rat. Brain Res. 1120, 13–22. doi:10.1016/j.brainres.2006.08.069.

Ariyannur, P. S., Arun, P., Barry, E. S., Andrews-Shigaki, B., Bosomtwi, A., Tang, H., Selwyn, R., Grunberg, N. E., Moffett, J. R., and Namboodiri, A. M. A. (2013). Do reductions in brain N-acetylaspartate levels contribute to the etiology of some neuropsychiatric disorders? J. Neurosci. Res. 91, 934– 942. doi:10.1002/jnr.23234.

Ashbrook, D. G., and Hager, R. (2013). Empirical testing of hypotheses about the evolution of genomic imprinting in mammals. Front. Neuroanat. 7, 6. doi:10.3389/fnana.2013.00006.

Ashbrook, D. G., Williams, R. W., Lu, L., Stein, J. L., Hibar, D. P., Nichols, T. E., Medland, S. E., Thompson, P. M., and Hager, R. (2014). Joint genetic analysis of hippocampal size in mouse and human identifies a novel gene linked to neurodegenerative disease. BMC Genomics 15, 850. doi:10.1186/1471-2164-15-850.

Balleine, B. W., and O’Doherty, J. P. (2010). Human and rodent homologies in action control: corticostriatal determinants of goal-directed and habitual action. Neuropsychopharmacology 35, 48–69. doi:10.1038/npp.2009.131.

117

Chapter 3: A cross-species genetic analysis identifies candidate genes for mouse anxiety and human bipolar disorder

Bartra, O., McGuire, J. T., and Kable, J. W. (2013). The valuation system: a coordinate-based meta-analysis of BOLD fMRI experiments examining neural correlates of subjective value. Neuroimage 76, 412–427. doi:10.1016/j.neuroimage.2013.02.063.

Bateson, M., Brilot, B., and Nettle, D. (2011). Anxiety: An evolutionary approach. Can. J. Psychiatry 56, 707–715.

Benjamini, Y., and Hochberg, Y. (1995). Controlling the false discovery rate: A practical and powerful approach to multiple testing. J. R. Stat. Soc. Ser. B 57, 289–300. doi:10.2307/2346101.

Benson, M. A., Tinsley, C. L., and Blake, D. J. (2004). Myospryn is a novel binding partner for dysbindin in muscle. J. Biol. Chem. 279, 10450–10458. doi:10.1074/jbc.M312664200.

Berridge, M. J. (2014). Calcium signalling and psychiatric disease: Bipolar disorder and schizophrenia. Cell Tissue Res. 357, 477–492. doi:10.1007/s00441-014-1806-z.

Berridge, M. J. (2013). Dysregulation of neural calcium signaling in Alzheimer disease, bipolar disorder and schizophrenia. Prion 7, 2–13. doi:10.4161/pri.21767.

Bitanihirwe, B. K. Y., and Woo, T.-U. W. (2014). Perineuronal nets and schizophrenia: The importance of neuronal coatings. Neurosci. Biobehav. Rev. 45C, 85–99. doi:10.1016/j.neubiorev.2014.03.018.

Boylan, K. R., Bieling, P. J., Marriott, M., Begin, H., Young, L. T., and MacQueen, G. M. (2004). Impact of comorbid anxiety disorders on outcome in a cohort of patients with bipolar disorder. J. Clin. Psychiatry 65, 1106–1113. doi:10.4088/JCP.v65n0813.

Breen, G., Prata, D., Osborne, S., Munro, J., Sinclair, M., Li, T., Staddon, S., Dempster, D., Sainz, R., Arroyo, B., et al. (2006). Association of the dysbindin gene with bipolar affective disorder. Am. J. Psychiatry 163, 1636– 1638. doi:10.1176/appi.ajp.163.9.1636.

Brückner, G., Grosche, J., Schmidt, S., Härtig, W., Margolis, R. U., Delpech, B., Seidenbecher, C. I., Czaniera, R., and Schachner, M. (2000). Postnatal development of perineuronal nets in wild-type mice and in a mutant deficient in tenascin-R. J. Comp. Neurol. 428, 616–629. doi:10.1002/1096- 9861(20001225)428:4<616::AID-CNE3>3.0.CO;2-K.

Brückner, G., Szeöke, S., Pavlica, S., Grosche, J., and Kacza, J. (2006). Axon initial segment ensheathed by extracellular matrix in perineuronal nets. Neuroscience 138, 365–375. doi:10.1016/j.neuroscience.2005.11.068.

Cardno, A. G., and Owen, M. J. (2014). Genetic relationships between schizophrenia, bipolar disorder, and schizoaffective disorder. Schizophr. Bull. 40, 504–515. doi:10.1093/schbul/sbu016.

118

Chapter 3: A cross-species genetic analysis identifies candidate genes for mouse anxiety and human bipolar disorder

Chauhan, A. K., Moretti, F. A., Iaconcig, A., Baralle, F. E., and Muro, A. F. (2005). Impaired motor coordination in mice lacking the EDA exon of the fibronectin gene. Behav. Brain Res. 161, 31–38. doi:10.1016/j.bbr.2005.02.020.

Chesler, E. J., Lu, L., Shou, S., Qu, Y., Gu, J., Wang, J., Hsu, H. C., Mountz, J. D., Baldwin, N. E., Langston, M. A., et al. (2005). Complex trait analysis of gene expression uncovers polygenic and pleiotropic networks that modulate nervous system function. Nat. Genet. 37, 233–242. doi:10.1038/ng1518.

Coryell, W., Solomon, D. A., Fiedorowicz, J. G., Endicott, J., Schettler, P. J., and Judd, L. L. (2009). Anxiety and outcome in bipolar disorder. Am. J. Psychiatry 166, 1238–1243. doi:10.1176/appi.ajp.2009.09020218.

Craddock, N., O’Donovan, M. C., and Owen, M. J. (2006). Genes for schizophrenia and bipolar disorder? Implications for psychiatric nosology. Schizophr. Bull. 32, 9–16. doi:10.1093/schbul/sbj033.

Cropley, V. L., Fujita, M., Innis, R. B., and Nathan, P. J. (2006). Molecular imaging of the dopaminergic system and its association with human cognitive function. Biol. Psychiatry 59, 898–907. doi:10.1016/j.biopsych.2006.03.004.

Dalley, J. W., Fryer, T. D., Brichard, L., Robinson, E. S. J., Theobald, D. E. H., Lääne, K., Peña, Y., Murphy, E. R., Shah, Y., Probst, K., et al. (2007). Nucleus accumbens D2/3 receptors predict trait impulsivity and cocaine reinforcement. Science 315, 1267–1270. doi:10.1126/science.1137073.

Doherty, J. L., and Owen, M. J. (2014). Genomic insights into the overlap between psychiatric disorders: implications for research and clinical practice. Genome Med. 6, 29. doi:10.1186/gm546.

Eagle, D. M., Wong, J. C. K., Allan, M. E., Mar, A. C., Theobald, D. E., and Robbins, T. W. (2011). Contrasting roles for dopamine D1 and D2 receptor subtypes in the dorsomedial striatum but not the nucleus accumbens core during behavioral inhibition in the stop-signal task in rats. J. Neurosci. 31, 7349–7356. doi:10.1523/JNEUROSCI.6182-10.2011.

Edvardsen, J., Torgersen, S., Røysamb, E., Lygren, S., Skre, I., Onstad, S., and Øien, P. A. (2008). Heritability of bipolar spectrum disorders. Unity or heterogeneity? J. Affect. Disord. 106, 229–240. doi:10.1016/j.jad.2007.07.001.

119

Chapter 3: A cross-species genetic analysis identifies candidate genes for mouse anxiety and human bipolar disorder

Eisener-Dorman, A. F., Grabowski-Boase, L., Steffy, B. M., Wiltshire, T., and Tarantino, L. M. (2010). Quantitative trait locus and haplotype mapping in closely related inbred strains identifies a locus for open field behavior. Mamm. Genome 21, 231–246. doi:10.1007/s00335-010-9260-z.

Erhardt, A., and Spoormaker, V. I. (2013). Translational approaches to anxiety: Focus on genetics, fear extinction and brain imaging. Curr. Psychiatry Rep. 15, 417. doi:10.1007/s11920-013-0417-9.

Ferreira, M. A. R., O’Donovan, M. C., Meng, Y. A., Jones, I. R., Ruderfer, D. M., Jones, L., Fan, J., Kirov, G., Perlis, R. H., Green, E. K., et al. (2008). Collaborative genome-wide association analysis supports a role for ANK3 and CACNA1C in bipolar disorder. Nat. Genet. 40, 1056–1058. doi:10.1038/ng.209.

Flint, J., Corley, R., DeFries, J., Fulker, D., Gray, J., Miller, S., and Collins, A. (1995). A simple genetic basis for a complex psychological trait in laboratory mice. Science 269, 1432–1435. doi:10.1126/science.7660127.

Flint, J., Valdar, W., Shifman, S., and Mott, R. (2005). Strategies for mapping and cloning quantitative trait genes in rodents. Nat. Rev. Genet. 6, 271–286. doi:10.1038/nrg1576.

Freeman, M. P., Freeman, S. A., and McElroy, S. L. (2002). The comorbidity of bipolar and anxiety disorders: prevalence, psychobiology, and treatment issues. J. Affect. Disord. 68, 1–23. doi:10.1016/S0165-0327(00)00299-8.

Freitag, S., Schachner, M., and Morellini, F. (2003). Behavioral alterations in mice deficient for the extracellular matrix glycoprotein tenascin-R. Behav. Brain Res. 145, 189–207. doi:10.1016/S0166-4328(03)00109-8.

Fuss, B., Wintergerst, E. S., Bartsch, U., and Schachner, M. (1993). Molecular characterization and in situ mRNA localization of the neural recognition molecule J1-160/180: a modular structure similar to tenascin. J. Cell Biol. 120, 1237–1249. doi:10.1083/jcb.120.5.1237.

Gershenfeld, H. K., Neumann, P. E., Mathis, C., Crawley, J. N., Li, X., and Paul, S. M. (1997). Mapping quantitative trait loci for open-field behavior in mice. Behav. Genet. 27, 201–210. doi:10.1023/A:1025653812535.

Gini, B., and Hager, R. (2012). Recombinant inbred systems can advance research in behavioral ecology. Front. Genet. 3, 198. doi:10.3389/fgene.2012.00198.

Goldstein, B. I., and Levitt, A. J. (2008). The specific burden of comorbid anxiety disorders and of substance use disorders in bipolar I disorder. Bipolar Disord. 10, 67–78. doi:10.1111/j.1399-5618.2008.00461.x.

Goodwin, F. K., and Jamison, K. R. (1990). Manic-depressive illness. First ed. New York: Oxford University Press.

Goodwin, G., and Sachs, G. (2010). Fast Facts: Bipolar Disorder. Second ed. Abingdon: Health Press Limited.

120

Chapter 3: A cross-species genetic analysis identifies candidate genes for mouse anxiety and human bipolar disorder

Gould, T. J., Keith, R. A., and Bhat, R. V (2001). Differential sensitivity to lithium’s reversal of amphetamine-induced open-field activity in two inbred strains of mice. Behav. Brain Res. 118, 95–105. doi:10.1016/S0166-4328(00)00318- 1.

Gratten, J., Wray, N. R., Keller, M. C., and Visscher, P. M. (2014). Large-scale genomics unveils the genetic architecture of psychiatric disorders. Nat. Neurosci. 17, 782–790. doi:10.1038/nn.3708.

Hamidovic, A., Dlugos, A., Skol, A., Palmer, A. A., and de Wit, H. (2009). Evaluation of genetic variability in the dopamine receptor D2 in relation to behavioral inhibition and impulsivity/sensation seeking: an exploratory study with d-amphetamine in healthy participants. Exp. Clin. Psychopharmacol. 17, 374–383. doi:10.1037/a0017840.

Hargus, G., Cui, Y., Schmid, J.-S., Xu, J., Glatzel, M., Schachner, M., and Bernreuther, C. (2008). Tenascin-R promotes neuronal differentiation of embryonic stem cells and recruitment of host-derived neural precursor cells after excitotoxic lesion of the mouse striatum. Stem Cells 26, 1973–1984. doi:10.1634/stemcells.2007-0929.

Haunsø, A., Ibrahim, M., Bartsch, U., Letiembre, M., Celio, M. R., and Menoud, P. (2000). Morphology of perineuronal nets in tenascin-R and parvalbumin single and double knockout mice. Brain Res. 864, 142–145. doi:10.1016/S0006-8993(00)02173-9.

Hayes, K. S., Hager, R., and Grencis, R. K. (2014). Sex-dependent genetic effects on immune responses to a parasitic nematode. BMC Genomics 15, 193. doi:10.1186/1471-2164-15-193.

Henry, B. L., Minassian, A., Young, J. W., Paulus, M. P., Geyer, M. A., and Perry, W. (2010). Cross-species assessments of motor and exploratory behavior related to bipolar disorder. Neurosci. Biobehav. Rev. 34, 1296–1306. doi:10.1016/j.neubiorev.2010.04.002.

Johnson, S. L. (2005). Mania and dysregulation in goal pursuit: a review. Clin. Psychol. Rev. 25, 241–262. doi:10.1016/j.cpr.2004.11.002.

Joo, E. J., Lee, K. Y., Jeong, S. H., Chang, J. S., Ahn, Y. M., Koo, Y. J., and Kim, Y. S. (2007). Dysbindin gene variants are associated with bipolar I disorder in a Korean population. Neurosci. Lett. 418, 272–275. doi:10.1016/j.neulet.2007.03.037.

Judd, L. L., and Akiskal, H. S. (2003). The prevalence and disability of bipolar spectrum disorders in the US population: re-analysis of the ECA database taking into account subthreshold cases. J. Affect. Disord. 73, 123–131. doi:10.1016/S0165-0327(02)00332-4.

121

Chapter 3: A cross-species genetic analysis identifies candidate genes for mouse anxiety and human bipolar disorder

Kalueff, A. V, Keisala, T., Minasyan, A., Kuuslahti, M., and Tuohimaa, P. (2006). Temporal stability of novelty exploration in mice exposed to different open field tests. Behav. Processes 72, 104–112. doi:10.1016/j.beproc.2005.12.011.

Kas, M. J. H., Fernandes, C., Schalkwyk, L. C., and Collier, D. A. (2007). Genetics of behavioural domains across the neuropsychiatric spectrum; of mice and men. Mol. Psychiatry 12, 324–330. doi:10.1038/sj.mp.4001979.

Kirshenbaum, G. S., Clapcote, S. J., Duffy, S., Burgess, C. R., Petersen, J., Jarowek, K. J., Yucel, Y. H., Cortez, M. A., Snead, O. C., Vilsen, B., et al. (2011). Mania-like behavior induced by genetic dysfunction of the neuron- specific Na+,K+-ATPase 3 sodium pump. Proc. Natl. Acad. Sci. U. S. A. 108, 18144–18149. doi:10.1073/pnas.1108416108.

Krezel, W., Ghyselinck, N., Samad, T. A., Dupé, V., Kastner, P., Borrelli, E., and Chambon, P. (1998). Impaired locomotion and dopamine signaling in retinoid receptor mutant mice. Science 279, 863–867. doi:10.1126/science.279.5352.863.

Krzyzosiak, A., Szyszka-Niagolov, M., Wietrzych, M., Gobaille, S., Muramatsu, S., and Krezel, W. (2010). Retinoid x receptor gamma control of affective behaviors involves dopaminergic signaling in mice. Neuron 66, 908–920. doi:10.1016/j.neuron.2010.05.004.

Lavedan, C., Licamele, L., Volpi, S., Hamilton, J., Heaton, C., Mack, K., Lannan, R., Thompson, A., Wolfgang, C. D., and Polymeropoulos, M. H. (2009). Association of the NPAS3 gene and five other loci with response to the antipsychotic iloperidone identified in a whole genome association study. Mol. Psychiatry 14, 804–819. doi:10.1038/mp.2008.56.

Lawrence, A. D., Weeks, R. A., Brooks, D. J., Andrews, T. C., Watkins, L. H., Harding, A. E., Robbins, T. W., and Sahakian, B. J. (1998). The relationship between striatal dopamine receptor binding and cognitive performance in Huntington’s disease. Brain 121 (Pt 7), 1343–1355. doi:10.1093/brain/121.7.1343.

Lee, H., Leamey, C. A., and Sawatari, A. (2012). Perineuronal nets play a role in regulating striatal function in the mouse. PLoS One 7, e32747. doi:10.1371/journal.pone.0032747.

Lee, S. H., Ripke, S., Neale, B. M., Faraone, S. V, Purcell, S. M., Perlis, R. H., Mowry, B. J., Thapar, A., Goddard, M. E., Witte, J. S., et al. (2013). Genetic

122

Chapter 3: A cross-species genetic analysis identifies candidate genes for mouse anxiety and human bipolar disorder

relationship between five psychiatric disorders estimated from genome-wide SNPs. Nat. Genet. 45, 984–994. doi:10.1038/ng.2711.

Legare, M. E., Bartlett, F. S., and Frankel, W. N. (2000). A major effect QTL determined by multiple genes in epileptic EL mice. Genome Res. 10, 42–48. doi:10.1101/gr.10.1.42.

Le-Niculescu, H., McFarland, M. J., Ogden, C. A., Balaraman, Y., Patel, S., Tan, J., Rodd, Z. A., Paulus, M., Geyer, M. A., Edenberg, H. J., et al. (2008). Phenomic, convergent functional genomic, and biomarker studies in a stress-reactive genetic animal model of bipolar disorder and co-morbid alcoholism. Am. J. Med. Genet. B. Neuropsychiatr. Genet. 147B, 134–66. doi:10.1002/ajmg.b.30707.

Le-Niculescu, H., Patel, S. D., Bhat, M., Kuczenski, R., Faraone, S. V, Tsuang, M. T., McMahon, F. J., Schork, N. J., Nurnberger, J. I., and Niculescu, A. B. (2009). Convergent functional genomics of genome-wide association data for bipolar disorder: comprehensive identification of candidate genes, pathways and mechanisms. Am. J. Med. Genet. B. Neuropsychiatr. Genet. 150B, 155–181. doi:10.1002/ajmg.b.30887.

Li, M.-X., Gui, H.-S., Kwan, J. S. H., and Sham, P. C. (2011). GATES: a rapid and powerful gene-based association test using extended Simes procedure. Am. J. Hum. Genet. 88, 283–293. doi:10.1016/j.ajhg.2011.01.019.

Li, M.-X., Kwan, J. S. H., and Sham, P. C. (2012). HYST: a hybrid set-based test for genome-wide association studies, with application to protein-protein interaction-based association analysis. Am. J. Hum. Genet. 91, 478–488. doi:10.1016/j.ajhg.2012.08.004.

Li, M.-X., Sham, P. C., Cherny, S. S., and Song, Y.-Q. (2010). A knowledge- based weighting framework to boost the power of genome-wide association studies. PLoS One 5, e14480. doi:10.1371/journal.pone.0014480.

Lichtenstein, P., Yip, B. H., Björk, C., Pawitan, Y., Cannon, T. D., Sullivan, P. F., and Hultman, C. M. (2009). Common genetic determinants of schizophrenia and bipolar disorder in Swedish families: a population-based study. Lancet 373, 234–239. doi:10.1016/S0140-6736(09)60072-6.

Mackay, T. F. C., Stone, E. A., and Ayroles, J. F. (2009). The genetics of quantitative traits: challenges and prospects. Nat. Rev. Genet. 10, 565–577. doi:10.1038/nrg2612.

MacKinnon, D. F. (2002). Comorbid bipolar disorder and panic disorder in families with a high prevalence of bipolar disorder. Am. J. Psychiatry 159, 30–35. doi:10.1176/appi.ajp.159.1.30.

123

Chapter 3: A cross-species genetic analysis identifies candidate genes for mouse anxiety and human bipolar disorder

Marks, I. fM., and Nesse, R. M. (1994). Fear and fitness: An evolutionary analysis of anxiety disorders. Ethol. Sociobiol. 15, 247–261. doi:10.1016/0162-3095(94)90002-7.

Mason, L., O’Sullivan, N., Montaldi, D., Bentall, R. P., and El-Deredy, W. (2014). Decision-making and trait impulsivity in bipolar disorder are associated with reduced prefrontal regulation of striatal reward valuation. Brain 137, 2346– 2355. doi:10.1093/brain/awu152.

McGinty, J. F. (1999). Regulation of neurotransmitter interactions in the ventral striatum. Ann. N. Y. Acad. Sci. 877, 129–139. doi:10.1111/j.1749- 6632.1999.tb09265.x.

McGuffin, P., Rijsdijk, F., Andrew, M., Sham, P., Katz, R., and Cardno, A. (2003). The heritability of bipolar affective disorder and the genetic relationship to unipolar depression. Arch. Gen. Psychiatry 60, 497–502. doi:10.1001/archpsyc.60.5.497.

Merikangas, K. R., Akiskal, H. S., Angst, J., Greenberg, P. E., Hirschfeld, R. M. A., Petukhova, M., and Kessler, R. C. (2007). Lifetime and 12-month prevalence of bipolar spectrum disorder in the National Comorbidity Survey replication. Arch. Gen. Psychiatry 64, 543–552. doi:10.1001/archpsyc.64.5.543.

Miller, J. A., Ding, S.-L., Sunkin, S. M., Smith, K. A., Ng, L., Szafer, A., Ebbert, A., Riley, Z. L., Royall, J. J., Aiona, K., et al. (2014). Transcriptional landscape of the prenatal human brain. Nature 508, 199–206. doi:10.1038/nature13185.

Minassian, A., Henry, B. L., Young, J. W., Masten, V., Geyer, M. A., and Perry, W. (2011). Repeated assessment of exploration and novelty seeking in the human behavioral pattern monitor in bipolar disorder patients and healthy individuals. PLoS One 6, e24185. doi:10.1371/journal.pone.0024185.

Montag-Sallaz, M., and Montag, D. (2003). Severe cognitive and motor coordination deficits in tenascin-R-deficient mice. Genes, Brain Behav. 2, 20–31. doi:10.1034/j.1601-183X.2003.00003.x.

Moskvina, V., O’Dushlaine, C., Purcell, S., Craddock, N., Holmans, P., and O’Donovan, M. C. (2011). Evaluation of an approximation method for assessment of overall significance of multiple-dependent tests in a genomewide association study. Genet. Epidemiol. 35, 861–866. doi:10.1002/gepi.20636.

124

Chapter 3: A cross-species genetic analysis identifies candidate genes for mouse anxiety and human bipolar disorder

Mühleisen, T. W., Leber, M., Schulze, T. G., Strohmaier, J., Degenhardt, F., Treutlein, J., Mattheisen, M., Forstner, A. J., Schumacher, J., Breuer, R., et al. (2014). Genome-wide association study reveals two new risk loci for bipolar disorder. Nat. Commun. 5, 3339. doi:10.1038/ncomms4339.

Nurnberger, J. I., Koller, D. L., Jung, J., Edenberg, H. J., Foroud, T., Guella, I., Vawter, M. P., and Kelsoe, J. R. (2014). Identification of pathways for bipolar disorder: a meta-analysis. JAMA psychiatry 71, 657–664. doi:10.1001/jamapsychiatry.2014.176.

Ojima, H., Kuroda, M., Ohyama, J., and Kishi, K. (1995). Two classes of cortical neurones labelled with Vicia villosa lectin in the guinea-pig. Neuroreport 6, 617–620. doi:10.1097/00001756-199503000-00008.

Owen, M. J., Craddock, N., and Jablensky, A. (2007). The genetic deconstruction of psychosis. Schizophr. Bull. 33, 905–911. doi:10.1093/schbul/sbm053.

Pae, C.-U., Serretti, A., Mandelli, L., Yu, H.-S., Patkar, A. A., Lee, C.-U., Lee, S.- J., Jun, T.-Y., Lee, C., Paik, I.-H., et al. (2007). Effect of 5-haplotype of dysbindin gene (DTNBP1) polymorphisms for the susceptibility to bipolar I disorder. Am. J. Med. Genet. B. Neuropsychiatr. Genet. 144B, 701–703. doi:10.1002/ajmg.b.30439.

Pattij, T., Janssen, M. C. W., Vanderschuren, L. J. M. J., Schoffelmeer, A. N. M., and Van Gaalen, M. M. (2007). Involvement of dopamine D1 and D2 receptors in the nucleus accumbens core and shell in inhibitory response control. Psychopharmacology (Berl). 191, 587–598. doi:10.1007/s00213- 006-0533-x.

Pearlson, G. D., and Ford, J. M. (2014). Distinguishing between schizophrenia and other psychotic disorders. Schizophr. Bull. 40, 501–503. doi:10.1093/schbul/sbu055.

Perry, W., Minassian, A., Paulus, M. P., Young, J. W., Kincaid, M. J., Ferguson, E. J., Henry, B. L., Zhuang, X., Masten, V. L., Sharp, R. F., et al. (2009). A reverse-translational study of dysfunctional exploration in psychiatric disorders: from mice to men. Arch. Gen. Psychiatry 66, 1072–1080. doi:10.1001/archgenpsychiatry.2009.58.

Ponder, C. A., Kliethermes, C. L., Drew, M. R., Muller, J., Das, K., Risbrough, V. B., Crabbe, J. C., Gilliam, T. C., and Palmer, A. A. (2007). Selection for contextual fear conditioning affects anxiety-like behaviors and gene

125

Chapter 3: A cross-species genetic analysis identifies candidate genes for mouse anxiety and human bipolar disorder

expression. Genes. Brain Behav. 6, 736–749. doi:10.1111/j.1601- 183X.2007.00306.x.

Psychiatric GWAS Consortium Bipolar Disorder Working Group (2011). Large- scale genome-wide association analysis of bipolar disorder identifies a new susceptibility locus near ODZ4. Nat. Genet. 43, 977–983. doi:10.1038/ng.943.

Purcell, S. M., Wray, N. R., Stone, J. L., Visscher, P. M., O’Donovan, M. C., Sullivan, P. F., and Sklar, P. (2009). Common polygenic variation contributes to risk of schizophrenia and bipolar disorder. Nature 460, 748– 752. doi:10.1038/nature08185.

Ruderfer, D. M., Fanous, A. H., Ripke, S., McQuillin, A., Amdur, R. L., Gejman, P. V, O’Donovan, M. C., Andreassen, O. A., Djurovic, S., Hultman, C. M., et al. (2014). Polygenic dissection of diagnosis and clinical dimensions of bipolar disorder and schizophrenia. Mol. Psychiatry 19, 1017–1024. doi:10.1038/mp.2013.138.

Saga, Y., Kobayashi, M., Ohta, H., Murai, N., Nakai, N., Oshima, M., and Taketo, M. M. (1999). Impaired extrapyramidal function caused by the targeted disruption of retinoid X receptor RXRgamma1 isoform. Genes Cells 4, 219– 228. doi:10.1046/j.1365-2443.1999.00253.x.

Samad, T. A., Krezel, W., Chambon, P., and Borrelli, E. (1997). Regulation of dopaminergic pathways by retinoids: activation of the D2 receptor promoter by members of the retinoic acid receptor-retinoid X receptor family. Proc. Natl. Acad. Sci. U. S. A. 94, 14349–14354. doi:10.1073/pnas.94.26.14349.

Schofield, P. N., Hoehndorf, R., and Gkoutos, G. V (2012). Mouse genetic and phenotypic resources for human genetics. Hum. Mutat. 33, 826–836. doi:10.1002/humu.22077.

Scott, L. J., Muglia, P., Kong, X. Q., Guan, W., Flickinger, M., Upmanyu, R., Tozzi, F., Li, J. Z., Burmeister, M., Absher, D., et al. (2009). Genome-wide association and meta-analysis of bipolar disorder in individuals of European ancestry. Proc. Natl. Acad. Sci. U. S. A. 106, 7501–7506. doi:10.1073/pnas.0813386106.

Shepherd, J. K., Grewal, S. S., Fletcher, A., Bill, D. J., and Dourish, C. T. (1994). Behavioural and pharmacological characterisation of the elevated “zero- maze” as an animal model of anxiety. Psychopharmacology (Berl). 116, 56– 64. doi:10.1007/BF02244871.

Shin, O.-H., Han, W., Wang, Y., and Südhof, T. C. (2005). Evolutionarily conserved multiple C2 domain proteins with two transmembrane regions (MCTPs) and unusual Ca2+ binding properties. J. Biol. Chem. 280, 1641– 1651. doi:10.1074/jbc.M407305200.

126

Chapter 3: A cross-species genetic analysis identifies candidate genes for mouse anxiety and human bipolar disorder

Simon, N. M., Otto, M. W., Wisniewski, S. R., Fossey, M., Sagduyu, K., Frank, E., Sachs, G. S., Nierenberg, A. A., Thase, M. E., and Pollack, M. H. (2004). Anxiety disorder comorbidity in bipolar disorder patients: data from the first 500 participants in the Systematic Treatment Enhancement Program for Bipolar Disorder (STEP-BD). Am. J. Psychiatry 161, 2222–2229. doi:10.1176/appi.ajp.161.12.2222.

Singer, J. B., Hill, A. E., Nadeau, J. H., and Lander, E. S. (2005). Mapping quantitative trait loci for anxiety in chromosome substitution strains of mice. Genetics 169, 855–62. doi:10.1534/genetics.104.031492.

Surmeier, D. J., Ding, J., Day, M., Wang, Z., and Shen, W. (2007). D1 and D2 dopamine-receptor modulation of striatal glutamatergic signaling in striatal medium spiny neurons. Trends Neurosci. 30, 228–235. doi:10.1016/j.tins.2007.03.008.

Szczepankiewicz, A. (2013). Evidence for single nucleotide polymorphisms and their association with bipolar disorder. Neuropsychiatr. Dis. Treat. 9, 1573– 1582. doi:10.2147/NDT.S28117.

Talbot, C. J., Radcliffe, R. A., Fullerton, J., Hitzemann, R., Wehner, J. M., and Flint, J. (2003). Fine scale mapping of a genetic locus for conditioned fear. Mamm. Genome 14, 223–230. doi:10.1007/s00335-002-3059-5.

Turri, M. G., Datta, S. R., DeFries, J., Henderson, N. D., and Flint, J. (2001a). QTL analysis identifies multiple behavioral dimensions in ethological tests of anxiety in laboratory mice. Curr. Biol. 11, 725–734. doi:10.1016/S0960- 9822(01)00206-8.

Turri, M. G., Henderson, N. D., DeFries, J. C., and Flint, J. (2001b). Quantitative trait locus mapping in laboratory mice derived from a replicated selection experiment for open-field activity. Genetics 158, 1217–1226.

Valdar, W., Solberg, L. C., Gauguier, D., Burnett, S., Klenerman, P., Cookson, W. O., Taylor, M. S., Rawlins, J. N. P., Mott, R., and Flint, J. (2006). Genome-wide genetic association of complex traits in heterogeneous stock mice. Nat. Genet. 38, 879–887. doi:10.1038/ng1840.

Vázquez, G. H., Baldessarini, R. J., and Tondo, L. (2014). Co-occurrence of anxiety and bipolar disorders: Clinical and therapeutic overview. Depress. Anxiety 31, 196–206. doi:10.1002/da.22248.

Vogel, H., Montag, D., Kanzleiter, T., Jonas, W., Matzke, D., Scherneck, S., Chadt, A., Töle, J., Kluge, R., Joost, H.-G., et al. (2013). An interval of the obesity QTL Nob3.38 within a QTL hotspot on chromosome 1 modulates behavioral phenotypes. PLoS One 8, e53025. doi:10.1371/journal.pone.0053025.

Wang, J., Duncan, D., Shi, Z., and Zhang, B. (2013). WEB-based GEne SeT AnaLysis Toolkit (WebGestalt): update 2013. Nucleic Acids Res. 41, W77– 83. doi:10.1093/nar/gkt439.

127

Chapter 3: A cross-species genetic analysis identifies candidate genes for mouse anxiety and human bipolar disorder

Wang, J., Williams, R. W., and Manly, K. F. (2003). WebQTL: web-based complex trait analysis. Neuroinformatics 1, 299–308. doi:10.1385/NI:1:4:299.

Wang, Q., He, K., Li, Z., Chen, J., Li, W., Wen, Z., Shen, J., Qiang, Y., Ji, J., Wang, Y., et al. (2014). The CMYA5 gene confers risk for both schizophrenia and major depressive disorder in the Han Chinese population. World J. Biol. Psychiatry 15, 553–560. doi:10.3109/15622975.2014.915057.

Watanabe, Y., Shibuya, M., and Someya, T. (2014). The cardiomyopathy- associated 5 (CMYA5) gene and risk of schizophrenia: Meta-analysis of rs3828611 and rs4704591 in East Asian populations. Asian J. Psychiatr. 7, 95–96. doi:10.1016/j.ajp.2013.10.013.

Wehner, J. M., Radcliffe, R. A., Rosmann, S. T., Christensen, S. C., Rasmussen, D. L., Fulker, D. W., and Wiles, M. (1997). Quantitative trait locus analysis of contextual fear conditioning in mice. Nat. Genet. 17, 331–334. doi:10.1038/ng1197-331.

Wray, N. R., and Gottesman, I. I. (2012). Using summary data from the danish national registers to estimate heritabilities for schizophrenia, bipolar disorder, and major depressive disorder. Front. Genet. 3, 118. doi:10.3389/fgene.2012.00118.

Yalcin, B., Willis-Owen, S. A. G., Fullerton, J., Meesaq, A., Deacon, R. M., Rawlins, J. N. P., Copley, R. R., Morris, A. P., Flint, J., and Mott, R. (2004). Genetic dissection of a behavioral quantitative trait locus shows that Rgs2 modulates anxiety in mice. Nat. Genet. 36, 1197–1202. doi:10.1038/ng1450.

Young, J. W., Minassian, A., Paulus, M. P., Geyer, M. A., and Perry, W. (2007). A reverse-translational approach to bipolar disorder: rodent and human studies in the Behavioral Pattern Monitor. Neurosci. Biobehav. Rev. 31, 882–896. doi:10.1016/j.neubiorev.2007.05.009.

Zhang, B., Kirov, S., and Snoddy, J. (2005). WebGestalt: an integrated system for exploring gene sets in various biological contexts. Nucleic Acids Res. 33, W741–8. doi:10.1093/nar/gki475.

128

Chapter 4: Genetic variation in offspring indirectly influences the quality of maternal behaviour in mice

Chapter 4: Genetic variation in offspring indirectly influences the

quality of maternal behaviour in mice

129

Chapter 4: Genetic variation in offspring indirectly influences the quality of maternal behaviour in mice

Chapter 4: Preface

This chapter is based on an unpublished manuscript which is currently in submission. It has been reformatted to help the flow of the thesis.

Empirical data was collected by Beatrice Gini and Reinmar Hager.

Analysis was carried out by David Ashbrook.

The paper was jointly written by David Ashbrook and Reinmar Hager.

This chapter is the first of three which begin to explore indirect genetic effects, using variations on a cross-fostering design in the BXD mouse lines. Here we investigate the maternal genetic effects of mothers on foster-litters.

130

Chapter 4: Genetic variation in offspring indirectly influences the quality of maternal behaviour in mice

Genetic variation in offspring indirectly influences the quality of maternal behaviour in mice

Authors: David G. Ashbrooka, Beatrice Ginia, Reinmar Hagera aComputational and Evolutionary Biology, Faculty of Life Sciences, University of

Manchester, Manchester M13 9PT, UK

131

Chapter 4: Genetic variation in offspring indirectly influences the quality of maternal behaviour in mice

Abstract

The conflict over the level of parental investment between parent and offspring is predicted to lead to selection on genes expressed in offspring for traits influencing maternal investment, and on genes expressed in parents affecting offspring behaviour. However, the specific genetic variants that indirectly modify maternal or offspring behaviour remain largely unknown. Using a genetics experiment in a cross-fostered population of mice, we map maternal behaviour in genetically uniform mothers as a function of genetic variation in offspring and identify two loci on offspring chromosomes 5 and 7 that modify maternal behaviour. Conversely, we found that genetic variation among mothers influences offspring development, when offspring genotype was kept constant.

Offspring solicitation and maternal care show signs of co-adaptation as they are negatively correlated between mothers and their biological offspring, which may be linked to the costs of increased solicitation on growth found in our study (e.g. needing to expend more energy to increase solicitation). Our results show overall levels of parental provisioning and offspring solicitation are unique to specific genotypes.

132

Chapter 4: Genetic variation in offspring indirectly influences the quality of maternal behaviour in mice

4.1 Introduction

The close interaction between mother and offspring in mammals is fundamental to offspring development and fitness. However, parent and offspring are in conflict over how much parents should invest in their young where offspring typically demand more than is optimal for the parent (Trivers, 1974; Godfray,

1995). The resulting selection pressures are predicted to lead to the evolution of traits in offspring that influence parental behaviour (and thus investment).

Conversely, parental traits should be selected for their effects on offspring traits that influence parental behaviour indirectly (Kilner and Hinde, 2012). This would lead to indirect genetic effects (IGE): loci expressed in one individual which influence the phenotype of a second individual.

The correlation between parental and offspring traits has been the focus of co-adaptation models where specific combinations of demand and provisioning are selectively favoured (Wolf and Brodie III, 1998; Kölliker et al.,

2005). The fundamental assumption underlying predictions about the evolution of traits involved in parent-offspring interactions is that genetic variation in offspring exists for traits that indirectly influence maternal investment and vice versa.

However, it remains to be shown whether specific genetic variants in offspring indirectly influence maternal behaviour in mammals. In an experimental mouse population, we demonstrate that genes expressed in offspring modify the quality of maternal behaviour and thus affect, indirectly, offspring fitness.

To investigate the genetics of parent-offspring interactions we conducted a cross-fostering experiment between genetically variable and genetically uniform mice, using the largest genetic reference panel in mammals, both in terms of number of extant inbred lines available and the number of phenotypes analysed, the BXD mouse population. We set up an experimental population of

133

Chapter 4: Genetic variation in offspring indirectly influences the quality of maternal behaviour in mice families consisting of mothers and adoptive young, in which, across the population, either mothers or offspring are genetically variable while the corresponding mothers or offspring show no genetic variation (Figure 4.1). Thus, we can analyse the effects of genetic variation in mothers or offspring while controlling for genetic variation in the other. This cross-fostering design has been successfully utilized in previous studies on family interactions because it breaks the correlation between maternal and offspring traits. Here, different families, or naturally occurring variation of maternal and offspring trait combinations across different broods, are assumed to represent distinct evolved strategies (Agrawal et al., 2001; Hager and Johnstone, 2003; Meunier and Kölliker, 2012).

Figure 4.1: Experimental cross-foster design. Females of different lines of the

BXD strain (large light to grey mice) adopt B6 offspring (small dark mice), and B6 females (large dark mice) adopt offspring born to females of different BXD lines

134

Chapter 4: Genetic variation in offspring indirectly influences the quality of maternal behaviour in mice

(small light to grey mice). A total of 42 BXD lines with three within-line repeats plus the corresponding B6 families were set up for the experiment.

In our experiment, we used mice of the BXD recombinant inbred population (Wu et al., 2014), derived from two divergent mouse strains

(C57BL/6 J and DBA/2 J, hence BXD). The BXD population consist of experimentally tractable and genetically defined mouse lines capturing a large amount of naturally occurring genetic variation, which underlies variation at the phenotypic level (e.g. Chesler et al., 2005; Hayes et al., 2014). The BXD panel incorporates ~5 million segregating SNPs, 500,000 insertions and deletions, and

55,000 copy-number variants. These lines are used for complex systems- genetics analyses integrating massive phenotype and gene expression data sets obtained across years and studies (e.g. Andreux et al., 2012; Ashbrook et al.,

2014). When generating families of genetically variable mothers and genetically uniform offspring we cross-fostered C57BL/6J (B6) litters, in which no genetic variation occurs between animals of this strain, to mothers of a given BXD strain.

Conversely, a BXD female’s litter was cross-fostered to B6 mothers. From birth until weaning at 3 weeks of age we recorded offspring and maternal bodyweights and measured maternal behaviour and offspring solicitation once during each of the three weeks, following Hager & Johnstone(Hager and Johnstone, 2003).

We were able to identify IGE loci expressed in offspring which influence the phenotype of foster mothers, and loci expressed in mothers which influence the phenotype of foster offspring. Further, we find evidence that offspring solicitation and maternal care show signs of co-adaptation as they are negatively correlated between mothers and their biological offspring.

135

Chapter 4: Genetic variation in offspring indirectly influences the quality of maternal behaviour in mice

4.2 Materials and methods

4.2.1 Experimental animals

The 42 BXD strains used in this study (1, 11, 12, 14, 24, 32, 34, 38-40,

43-45, 48a, 49-51, 55, 56, 60-64, 66-71, 73a, 73b, 73-75, 83, 84, 87, 89, 90, 98,

102) were obtained from Professor Robert W. Williams at the University of

Tennessee Health Science Centre, Memphis, TN. C57BL/6J (B6) mice were obtained from Charles River, UK. Three within-line repeats plus the corresponding B6 families were set up for the experiment (Figure 4.1).

4.2.2 Husbandry and mating protocol

Mice were maintained under standard laboratory conditions in individually ventilated cages, maintained at 20°C (± 2°) with a relative humidity of 55% (± 10), reverse dark:light cycle with red light (active phase) between 10:00 and 22:00 hours and food and water ad libitum. Cages were cleaned once a week but never within the first six days after birth to minimise disturbance. The parental mice were all sexually mature and females were unmated. Groups of up to five siblings were housed together in single-sex cages until mating, which occurred between

6-10 weeks of age, when females were ≥ 18g.

Before breeding, bedding from male cages was added to the female cages to encourage synchronised oestrus (Hau and Hau, 2007) and individual males were moved to new cages, to allow them to scent mark. Two days later, two sisters were added to the male’s cage. From 16 days after, females were checked daily for evidence of pregnancy. If they were visibly pregnant (weight gain ≥ 8g or distended abdomen), females were separated into an individual cage. This ensured that neither father nor aunt had a social interaction with the offspring.

136

Chapter 4: Genetic variation in offspring indirectly influences the quality of maternal behaviour in mice

4.2.3 Phenotyping protocols

Cross-fostering of entire litters took place within 24h of birth of corresponding B6 and BXD females and analyses used trait values per adopted pup to adjust for differences in litter size. Both mothers and litters were weighed at cross-fostering and for three weeks until weaning, i.e. on post-natal days 6, 10 and 14, to enable the calculation of growth during these periods. In addition, we recorded maternal and offspring behaviour on post-natal days 6, 10 and 14 when we simulated maternal departure to standardize observation conditions (Hager and Johnstone,

2003, 2005). After a 4h separation, both mothers and litters were weighed then re-joined in the original cage. Maternal and offspring behaviours were recorded simultaneously over 15 minutes, using scan sampling every 20 seconds (Table

4.1) (Martin and Bateson, 2007). Because rodents are nocturnal, all observations occurred under red light, i.e. the active phase. Maternal care was recorded as the sum of nursing, suckling and nestbuilding. Nursing is defined as attending the litter, sitting on the nest and suckling up to half the litter while suckling refers to half or more of the litter being suckled at the same time. This distinction was used as sometimes it cannot be ascertained whether pups are suckling or not because of the position of the mothers in the nest. Mothers and pups were then reweighed after two hours of having been together. Pups’ short-term weight change during this time was recorded and the change in maternal bodyweight was used as a measure of maternal provisioning (as it is assumed that bodyweight lost indicates milk given to foster offspring). At day 21, mothers and litters were reweighed, and the pups were sexed. All procedures were approved by the University of Manchester Ethics Committee.

137

Chapter 4: Genetic variation in offspring indirectly influences the quality of maternal behaviour in mice

Table 4.1: Maternal and offspring behaviours recorded. Behaviours in bold (states) were recorded every 20 seconds for 15 minutes and behaviours in italics (events) were recorded every time they occurred.

Maternal care behaviour Self Nestbuilding (gathering nesting material and Autogroom (mother grooming constructing a nest) herself) Nursing (attending the litter, sitting on the nest and Feeding or drinking suckling up to half the litter) Suckling (greater than half the litter being suckled at Resting the same time) Sniffing Other active (moving around the cage) Pup retrieve Licking

Offspring behaviour Feeding in the nest Feeding outside the nest Resting in the nest Resting outside the nest Other active (moving around the cage) Suck attempt/solicitation (move towards mother or attempt to feed) Playfight/sibling competition (attempt to burrow under a sibling or pushing a sibling from a teat) Autogroom (pup grooming itself) Allogroom (pup grooming another pup

4.2.4 Genetic analysis

Interval mapping (Haley and Knott, 1992) relies on 3795 informative SNP markers across all chromosomes, except Y, as implemented in GeneNetwork

(Wang et al., 2003; Hager et al., 2012). The BXD strains were genotyped using the MUGA array in 2011, along with genotypes generated earlier using Affymetrix and Illumina platforms (Shifman et al., 2006), and mm9 is used. Loci are identified in GN by the computation of a likelihood ratio statistic (LRS) score and significance was determined using 5000 permutations of the phenotype data.

Thresholds were determined as suggestive (p < 0.63; a threshold which is defined as that which yields, on average, one false positive per genome scan), or

138

Chapter 4: Genetic variation in offspring indirectly influences the quality of maternal behaviour in mice significant (p < 0.05) genome-wide p-value. Confidence intervals were calculated using the 1.5 LOD drop, as is common practice (Dupuis and Siegmund, 1999;

Manichaikul et al., 2006). LRS scores were converted to LOD scored by dividing by 4.615, and the first markers with a LOD score ≤ (peak LOD – 1.5) were used as the ends of the confidence interval. All genetic locations are given using the mm9 mouse genome build.

Potential candidate genes were identified using GeneNetwork to find all the genes within the confidence interval calculated above. Information for these candidates was collected from QTLminer (Alberts and Schughart, 2010), Entrez genes (http://www.ncbi.nlm.nih.gov/gene) and Mouse Genome Informatics

(Eppig et al., 2015).

139

Chapter 4: Genetic variation in offspring indirectly influences the quality of maternal behaviour in mice

4.3 Results

We first investigated whether there is evidence for indirect genetic effects in offspring influencing maternal behaviour, while keeping maternal genotype constant. In other words, is maternal behaviour modified by genes expressed in offspring? To find out, we mapped variation in B6 maternal behaviour as a function of their adoptive BXD offspring genotype. We found that variation in offspring genotype affects maternal behaviour, which in turn influences offspring development and fitness. Throughout, we denote loci as either maternal or offspring, Mat or Osp, followed by whether it is an indirect genetic effect (Ige) or a direct genetic effect locus (Dge).

Figure 4.2: Offspring indirect genetic effect modifying maternal nestbuilding behaviour (OspIge7.1). An offspring genome scan of maternal nestbuilding on day 6.

The blue line represents the genome scan, showing the likelihood ratio statistic (LRS) associated with each marker across the 19 autosomal and the X chromosome. The top, pink, line marks genome-wide significance (p = 0.05), the lower, grey, line marks the suggestive significance threshold (p = 0.63). The green or red line shows the additive coefficient, with green showing that the DBA/2J alleles increase trait values and red that the C57BL/6J alleles increase trait values. The green axis on the right shows by how much the respective alleles increase trait values.

140

Chapter 4: Genetic variation in offspring indirectly influences the quality of maternal behaviour in mice

During the first post-natal week we mapped a locus on offspring chromosome 7, OspIge7.1, modifying maternal nestbuilding on day 6 (Figure

4.2). At this locus, the D2 allele increases the trait value such that B6 mothers showed more nestbuilding activity when fostering BXD pups carrying the D2 allele at the locus. Nestbuilding is particularly important for offspring fitness during early development as thermoregulation is underdeveloped and hypothermia the primary cause of early death, even if milk is supplied (Lynch and

Possidente, 1978).

We detected a further locus on offspring distal chromosome 5

(OspIge5.1) that affects maternal care on day 14; around the time we expect the weaning conflict to be highest (Figure 4.3). Here, BXD offspring with the D2 allele receive increased levels of maternal care.

Figure 4.3 Offspring indirect genetic effect modifying maternal care (OspIge5.1). An offspring genome scan of maternal care on day 14. The blue line represents the genome scan, showing the likelihood ratio statistic (LRS) associated with each marker across the

19 autosomal and the X chromosome. The top, pink, line marks genome-wide significance (p = 0.05), the lower, grey, line marks the suggestive significance threshold

(p = 0.63). The green or red line shows the additive coefficient, with green showing that the DBA/2J alleles increase trait values and red that the C57BL/6J alleles increase trait values. The green axis on the right shows by how much the respective alleles increase trait values.

141

Chapter 4: Genetic variation in offspring indirectly influences the quality of maternal behaviour in mice

Conversely, we can look at how variation in maternal genotype influences offspring traits, when offspring genotype was kept constant, mapping variation in offspring traits as a function of BXD genotype. Here, we found that offspring growth between days 6 and 10 is affected by a locus on maternal chromosome

17, MatIge17.1, where the B6 allele increases the trait value (Figure 4.4).

Figure 4.4: Maternal indirect genetic effect modifying offspring growth (MatIge17).

A maternal genome scan of offspring growth between post-natal days 6 and 10. The blue line represents the genome scan, showing the likelihood ratio statistic (LRS) associated with each marker across the 19 autosomal and the X chromosome. The top, pink, line marks genome-wide significance (p = 0.05), the lower, grey, line marks the suggestive significance threshold (p = 0.63). The green or red line shows the additive coefficient, with green showing that the DBA/2J alleles increase trait values and red that the

C57BL/6J alleles increase trait values. The green axis on the right shows by how much the respective alleles increase trait values.

In addition to looking at indirect genetic effects we can analyse direct genetic effects, i.e. how an individual’s genotype influences its own traits. In mothers we found a direct genetic effect locus for maternal care on day 6 on proximal chromosome 10, and for nestbuilding specifically on chromosome 1

(MatDge10.1 and MatDge1.1, respectively), where the B6 allele increases the trait value in both cases. We also detected a locus for offspring solicitation behaviour on day 6 on chromosome 5 (OspDge5.1) with the D2 allele increasing

142

Chapter 4: Genetic variation in offspring indirectly influences the quality of maternal behaviour in mice the level of solicitation shown. Its location is at the opposite end on chromosome

5 from where OspIge5.1 is located. All loci are summarized in table 4.2.

Table 4.2: A summary of the QTL identified. The QTL position is the coordinates of the most significant marker(s). The confidence interval is defined by a 1.5 LOD drop. LRS, p- value and additive effect size are for the most significant marker(s) and are calculated by

GeneNetwork. A positive additive effect indicates that the D2 allele increases the trait value, whereas a negative additive effect indicates that the B6 allele increases the trait value. Number of genes is from QTLminer.

Confidence p- Additive Number Loci QTL position LRS interval value effect of genes

MatDge1.1 168.323 - 169.018 165.243 - 172.067 17.033 0.069 -1.019 68

OspIge7.1 53.683 47.762 - 56.652 17.852 0.039 0.864 232

81.492 - 89.629 76.126 - 90.928 16.06 0.87 0.816 133

MatIge17.1 23.323 - 26.351 11.482 - 31.179 19.026 0.022 -0.28 422

33.026 31.327 - 40.657 18.571 0.028 -0.279 321

MatDge10.1 19.098 18.617 - 21.834 22.376 0.008 -1.75 30

OspDge5.1 23.827 17.827 - 24.622 17.855 0.046 0.889 73

OspIge5.1 146.689 - 147.394 145.297 - 147.659 18.651 0.038 2.362 40

Overall, our results show that IGEs, as well as DGEs, can be linked to specific loci that affect parent-offspring interaction, and it is thus possible that selection may occur on genes with indirect and direct effects on parental behaviour. Importantly, our result that variation in maternal behaviour is affected by genes expressed in offspring is also clearly borne out at the phenotypic level: offspring solicitation behaviour in genetically variable BXD pups is positively correlated with the level of maternal care in their genetically uniform B6 adoptive mothers on all three days we measured behaviour (day 6: Pearsons’ r = 0.56, p =

0.003, df = 36; day 10: r = 0.63, p = 0.001, df = 36; and day 14: r = 0.55, p =

0.008, df = 33). This appears to be a result of genotype, not environment, since

143

Chapter 4: Genetic variation in offspring indirectly influences the quality of maternal behaviour in mice the mean was taken per genotype, meaning these are genotype values, not litter values. Further, a general linear model (GLM) shows that litter size may affect the amount of offspring solicitation: if solicitation is not divided by litter size then litter size significantly affects solicitation (Table 4.3) but is no longer significant once the phenotypes have been divided by litter size (Table 4.4). Therefore these correlations do represent the effect of genotype, and not environment or litter size.

Table 4.3: Tests of between-subjects effects of litter size, B6 foster mothers’ maternal care and day on BXD offspring solicitation, without phenotypes being divided by the litter size. This shows that litter size does have an effect and therefore needs to be corrected for. Adapted from SPSS (version 21, IBM Corporation, Armonk,

NY, USA).

Dependent Variable: BXD offspring solicitation

Source Type III Sum of df Mean Square F Sig.

Squares

Hypothesis 110.179 1 110.179 1.449 .267 Intercept

Error 537.632 7.071 76.038a

Hypothesis 448.081 1 448.081 11.835 .001 Litter size

Error 3899.650 103 37.861b

Hypothesis 396.164 1 396.164 10.464 .002 B6 mothers’

maternal care Error 3899.650 103 37.861b

Hypothesis 1556.120 2 778.060 20.551 .000 Day

Error 3899.650 103 37.861b

a. .052 MS(day) + .948 MS(Error)

b. MS(Error)

144

Chapter 4: Genetic variation in offspring indirectly influences the quality of maternal behaviour in mice

Table 4.4: Tests of between-subjects effects of litter size, B6 foster mothers’ maternal care and day on BXD offspring solicitation. Both phenotypes have been divided by the number of pups in the litter. Adapted from SPSS (version 21, IBM

Corporation, Armonk, NY, USA).

Dependent Variable: BXD offspring solicitation

Source Type III Sum df Mean Square F Sig. of Squares

Hypothesis 17.367 1 17.367 8.388 .007 Intercept

Error 55.527 26.819 2.070a

Hypothesis 3.122 1 3.122 1.971 .163 Litter size

Error 163.206 103 1.585b

Hypothesis 27.915 1 27.915 17.617 .000 B6 mothers’ maternal care

Error 163.206 103 1.585b

Hypothesis 45.074 2 22.537 14.223 .000 Day

Error 163.206 103 1.585b a. .023 MS(day) + .977 MS(Error) b. MS(Error)

Similarly, we can investigate how traits in genetically uniform B6 offspring correlate with maternal traits in their genetically variable BXD mothers (i.e. within adoptive families). We found that offspring solicitation behaviour is correlated with maternal care on days 6 and 14 (day 6: r = 0.52, p = 0.004, df = 40; and day

14: r = 0.30, p = 0.06, df = 39), although the day 14 correlation is just non- significant at the 0.05 level. While one might generally assume that mothers behave in response to offspring solicitation behaviour, these results show, perhaps surprisingly, that variation in maternal behaviour influences the level of solicitation: here, we need to remember that there is no variation in offspring

145

Chapter 4: Genetic variation in offspring indirectly influences the quality of maternal behaviour in mice genotype so we assume that across families differences in offspring behaviour are due to differences in the genotype of their adoptive mothers. As with the correlation between BXD offspring and B6 mother, this represents an effect of genotype and not environment.

We next investigated potential candidate genes in the regions defined by the loci using GeneNetwork, Entrez gene information and gene functional annotations, focusing on the four loci that define regions containing less than 100 genes: OspIge5.1, MatDge1.1, MatDge10.1 and OspDge5.1 (Table 4.2;

Supplementary tables 4.1-4) .

Beginning with OspIge5.1, which is an IGE of BXD offspring genotype on

B6 maternal care on day 14, the locus contains 40 genes (Supplementary table

4.1), among them several involved in steroid hormone biosynthesis (Cyp3a16,

Cyp3a44, Cyp3a11, Cyp3a25, Cyp3a41a), which therefore may be involved in the regulation of offspring behaviour, which, in turn, influences maternal behaviour. Next, MatDge1.1 for nestbuilding on day 6 contains 68 genes

(Supplementary table 4.2) with Hsd17b7 a potential candidate, as it is involved in steroid hormone synthesis, and hormonal regulation is needed to initiate nestbuilding behaviour (Gammie et al., 2007; Keisala et al., 2007; Bester-

Meredith and Marler, 2012).

Data contained within the GeneNetwork database (Figure 4.5) shows that there is a significant positive correlation between the trait and Hsd17b7 expression in the adrenal gland of adult BXD mice (r = 0.476, p = 0.003, df = 34).

However, no non-synonymous SNPs (nsSNPs) or insertions or deletions (indels) have been identified within Hsd17b7 (Supplementary table 4.2). Further relevant genes at this locus have been linked to activity (Pou2f1, Lmx1a, Rgs4), which may affect nestbuilding behaviour, the first two of which contain both nsSNPs

146

Chapter 4: Genetic variation in offspring indirectly influences the quality of maternal behaviour in mice and indels. Finally, two of the genes within the locus are linked to changes in craniofacial morphology (Pbx1, Ddr2), the former of which has both nsSNPs and indels. It is well established that craniofacial abnormalities affect nestbuilding and therefore it maybe females are physically less able to build nests (Schneider et al., 2012). It is interesting to note that several candidate genes within the locus affect very different traits that overall influence nestbuilding, which is a complex and derived trait to which several intermediate traits contribute.

Figure 4.5: Correlation between BXD mothers’ nestbuilding on day 6 and expression of Hsd17b7 in adrenal gland. Hsd17b7 expression data from GeneNetwork

INIA Adrenal Affy MoGene 1.0ST dataset (GN Accession GN388), record ID 10359917.

147

Chapter 4: Genetic variation in offspring indirectly influences the quality of maternal behaviour in mice

MatDge10.1 affects maternal care on day 6 and contains 30 genes

(Supplementary table 4.3), of which the key candidate is Ifngr1 as it is related to depression-like behaviour, which in turn has been linked to reduced maternal care (Smith et al., 2004; Malkesman et al., 2008). It has an indel, and non- synonymous SNPs, suggesting structural changes in the protein.

Finally, OspDge5.1 directly affects offspring solicitation behaviour and contains 73 genes (Supplementary table 4.4),among which Cdk5 is a good candidate as it is involved in several neuronal annotations (e.g. axonogenesis and synaptic transmission), and knockout models show either reduced food intake (Okamura et al., 2012) or do not have a sucking reflex (Ohshima et al.,

1996). Again, it shows an indel and nsSNPs.

4.3.1 Co-adaptation of parental and offspring traits

While in the previous section we have focused on analysing traits within foster families, we now turn to the correlation between traits of biological families, i.e. mothers and their biological offspring. This correlation has been analysed in co- adaptation models, which make specific predictions about how parental and offspring traits are correlated, and in empirical work (Kölliker et al., 2000; Agrawal et al., 2001; Curley et al., 2004; Lock et al., 2004; Kölliker et al., 2005; Hinde et al., 2010; Meunier and Kölliker, 2012; Kölliker et al., 2015). Because in our foster families one of the two parts (either mother or offspring) is genetically variable while the other is not we can determine both offspring and maternal traits controlling for genetic variation in the other. Prior experimental studies have found both a positive (Parus major (Kölliker et al., 2000), Nicrophorus vespilloides (Lock et al., 2004)) and negative correlation between offspring solicitation and parental traits (Sehirus cinctus (Agrawal et al., 2001)). In our study, we found a negative correlation. When we measured provisioning (i.e. the

148

Chapter 4: Genetic variation in offspring indirectly influences the quality of maternal behaviour in mice mothers’ weight change after two hours with the foster offspring), we found a negative correlation between BXD offspring short-term weight change and the corresponding provisioning by their biological (BXD) mothers on day 10, as well as a negative correlation between BXD offspring solicitation and the corresponding provisioning by their biological (BXD) mothers on day 14 (r = -

0.34, p = 0.04, df = 36 and r = -0.48, p = 0.003, df = 31; Figure 4.6). Thus, our results suggest that mothers who are generous providers produce young that solicit less maternal resources than offspring born to less generous mothers.

Such a negative correlation is predicted to occur when maternal traits are predominantly under selection as long as parents respond to offspring demand

(which we have shown above) (Kölliker et al., 2005). One scenario to explain this negative correlation might be that each BXD line, i.e. genotype, is characterized by a unique (to this line, everything else being equal) combination of offspring and maternal behaviours where higher maternal provisioning is correlated with lower offspring solicitation. This may be due to the cost of increased solicitation, for which we found evidence in our study: the level of offspring solicitation is negatively correlated with bodyweight (e.g. day 10 and 14 solicitation is negatively correlated with bodyweight at day 10 and day 14, respectively: r = -

0.39, p = 0.02, df = 35; and r = -0.44, p = 0.01, df = 32; Figure 4.7).

149

Chapter 4: Genetic variation in offspring indirectly influences the quality of maternal behaviour in mice

Figure 4.6: Correlation between offspring and maternal traits in biological BXD families. The first panel shows the correlation between BXD offspring short-term weight change and provisioning by their corresponding biological BXD mother on day 10 (r = -

0.34, p = 0.04, df = 36, slope = -0.01371). The second panel shows the correlation between the level of BXD offspring solicitation on day 14 and their mothers’ provisioning on day 14 (r = -0.48, p = 0.003, df = 31, slope = -0.6283).

Figure 4.7: Correlation between offspring solicitation and corresponding bodyweight in BXD lines on day 10 and day 14, respectively. Day 10, r = -0.39, p =

0.02, df = 35, slope = -0.9645; day 14, r = -0.44, p = 0.01, df = 32, slope = -1.046.

150

Chapter 4: Genetic variation in offspring indirectly influences the quality of maternal behaviour in mice

4.4 Discussion

4.4.1 Caveats and future work

One important caveat of our results is that we cannot control for pre-natal effects of the biological mother, and since the mother and offspring genotypes are identical then any QTL we detect could actually be due to the maternal genotype.

However, methods to avoid this problem, such as artificial insemination and embryo transfer, may cause disruption of epigenetic markers, such as methylation and imprinting (Lim and Ferguson-Smith, 2010; Fowden et al., 2011;

Calle et al., 2012; Kohda and Ishino, 2013). Therefore, these techniques may have their own confounding effect. Further, the IGE loci which we find would still be interesting: if the QTL are due to the biological mothers’ genotype, then we have shown an IGE QTL affecting the foster mother. This would raise an interesting question of why the maternal genotype should alter the maternal phenotype via changing the offspring phenotype, rather than the maternal genotype directly altering maternal phenotype. A possible explanation comes from the fact that mice often show communal nesting and nursing (Sayler and

Salmon, 1969; König, 1997; Weidt et al., 2014; Heiderstadt et al., 2014).

Therefore, IGEs could have evolved to increase offspring solicitation to non- biological mothers within a communal nest. This could be investigated using a communal nursing experiment with the BXD mouse lines. For instance, having a genetically variable BXD mother and her offspring sharing a nest with a genetically uniform B6 mother and her B6 offspring, measuring maternal behaviour from the BXD and B6 mothers, and observing loci which alter the maternal care given by the B6 mother but not the BXD mother.

We, of course, have to acknowledge the fact that we are using inbred lines, themselves derived from ‘fancy’ mice bred in the 19th century, (Wade et al.,

151

Chapter 4: Genetic variation in offspring indirectly influences the quality of maternal behaviour in mice

2002) to make inferences about evolutionary processes. These lines therefore have been outside the reach of natural selection for an extended period of time, and instead under the artificial selection of laboratory conditions. That said, maternal care is a fundamental process, which is vital for offspring survival, and therefore should be maintained in the absence of natural selection under standard laboratory conditions. A similar study could be carried out using individuals taken from a wild population, e.g. fostering B6 offspring to wild derived mothers and looking for MGEs, but this would require genotyping all the mothers, however the opposite direction of fostering would be more difficult, as, for example, there will be a number of different genotypes within each litter, which would prevent mapping of offspring QTL to maternal traits.

We do not explore potentially causative genes for two of our loci,

OspIge7.1 and MatIge17.1, due to the large number of genes within the confidence interval, and therefore the difficulty in narrowing the list of potential candidate genes. However, it should be noted that the MatIge17.1 covers the location of the mouse major histocompatibility complex, histocompatibility 2, and that mothers alter their maternal behaviour depending on a shared genotype

(Yamazaki et al., 2000). Therefore, the MatIge17.1 loci could be due to the shared histocompatibility 2 genotype between mothers with the B6 MatIge17.1 loci and their foster offspring of the B6 strain.

4.4.2 Conclusion

Our study of the genetics underlying family interactions has revealed that genes expressed in offspring can indirectly influence the quality of maternal behaviour and thus offspring fitness. At the same time, we detected specific loci in maternal genotype that indirectly modify offspring traits, which shows that IGEs can be an important component of the genetic architecture of complex traits (Bijma and

152

Chapter 4: Genetic variation in offspring indirectly influences the quality of maternal behaviour in mice

Wade, 2008). We now need to investigate the candidate genes identified here and how their effects on parental and offspring traits are integrated into the gene networks determining individual development. By controlling for genetic variation in either mothers or offspring we have been able to show that levels of maternal provisioning and offspring solicitation are unique to specific genotypes (here each BXD line) and that solicitation is costly. The ability to conduct complex systems-genetics analyses in experimental systems of parent offspring interactions will enable us to concentrate now on understanding the underlying pathways involved, and how they are modified by social environmental conditions that determine adult phenotypes and associated reproductive success.

153

Chapter 4: Genetic variation in offspring indirectly influences the quality of maternal behaviour in mice

4.5 References

Agrawal, A. F., Brodie III, E. D., and Brown, J. (2001). Parent-offspring coadaptation and the dual genetic control of maternal care. Science 292, 1710–1712. doi:10.1126/science.1059910.

Alberts, R., and Schughart, K. (2010). QTLminer: identifying genes regulating quantitative traits. BMC Bioinformatics 11, 516. doi:10.1186/1471-2105-11- 516.

Bester-Meredith, J. K., and Marler, C. A. (2012). Naturally occurring variation in vasopressin immunoreactivity is associated with maternal behavior in female peromyscus mice. Brain Behav. Evol. 80, 244–253. doi:10.1159/000341899.

Calle, A., Fernandez-Gonzalez, R., Ramos-Ibeas, P., Laguna-Barraza, R., Perez-Cerezales, S., Bermejo-Alvarez, P., Ramirez, M. A., and Gutierrez- Adan, A. (2012). Long-term and transgenerational effects of in vitro culture on mouse embryos. Theriogenology 77, 785–93. doi:10.1016/j.theriogenology.2011.07.016.

Curley, J. P., Barton, S., Surani, A., and Keverne, E. B. (2004). Coadaptation in mother and infant regulated by a paternally expressed imprinted gene. Proc. R. Soc. B Biol. Sci. 271, 1303–1309. doi:10.1098/rspb.2004.2725.

Dupuis, J., and Siegmund, D. (1999). Statistical methods for mapping quantitative trait loci from a dense set of markers. Genetics 151, 373–386.

Eppig, J. T., Blake, J. A., Bult, C. J., Kadin, J. A., and Richardson, J. E. (2015). The Mouse Genome Database (MGD): facilitating mouse as a model for

154

Chapter 4: Genetic variation in offspring indirectly influences the quality of maternal behaviour in mice

human biology and disease. Nucleic Acids Res. 43, D726–D736. doi:10.1093/nar/gku967.

Gammie, S. C., Bethea, E. D., and Stevenson, S. A. (2007). Altered maternal profiles in corticotropin-releasing factor receptor 1 deficient mice. BMC Neurosci. 8, 17. doi:10.1186/1471-2202-8-17.

Godfray, H. C. (1995). Evolutionary theory of parent-offspring conflict. Nature 376, 133–138. doi:10.1038/376133a0.

Hager, R., and Johnstone, R. A. (2005). Differential growth of own and alien young in mixed litters of mice: a role for genomic imprinting? Ethology 111, 705–714. doi:10.1111/j.1439-0310.2005.01097.x.

Hager, R., and Johnstone, R. A. (2003). The genetic basis of family conflict resolution in mice. Nature 421, 533–535. doi:10.1038/nature01239.

Haley, C. S., and Knott, S. A. (1992). A simple regression method for mapping quantitative trait loci in line crosses using flanking markers. Heredity 69, 315–324. doi:10.1038/hdy.1992.131.

Hau, A. R., and Hau, J. (2007). Correlation between duration of courtship and litter size in outbred NMRI mice. In Vivo 21, 33–34.

Hayes, K. S., Hager, R., and Grencis, R. K. (2014). Sex-dependent genetic effects on immune responses to a parasitic nematode. BMC Genomics 15, 193. doi:10.1186/1471-2164-15-193.

Heiderstadt, K. M., Vandenbergh, D. J., Gyekis, J. P., and Blizard, D. A. (2014). Communal nesting increases pup growth but has limited effects on adult behavior and neurophysiology in inbred mice. J. Am. Assoc. Lab. Anim. Sci. 53, 152–60.

Hinde, C. A., Johnstone, R. A., and Kilner, R. M. (2010). Parent-offspring conflict and coadaptation. Science 327, 1373–1376. doi:10.1126/science.1186056.

Keisala, T., Minasyan, A., Järvelin, U., Wang, J., Hämäläinen, T., Kalueff, A. V, and Tuohimaa, P. (2007). Aberrant nest building and prolactin secretion in vitamin D receptor mutant mice. J. Steroid Biochem. Mol. Biol. 104, 269– 273. doi:10.1016/j.jsbmb.2007.03.031.

Kilner, R. M., and Hinde, C. A. (2012). “Parent-offspring conflict,” in The Evolution of Parental Care, eds. N. Royle, P. T. Smiseth, and M. Kölliker (Oxford, UK: Oxford University Press), 119–132.

155

Chapter 4: Genetic variation in offspring indirectly influences the quality of maternal behaviour in mice

Kohda, T., and Ishino, F. (2013). Embryo manipulation via assisted reproductive technology and epigenetic asymmetry in mammalian early development. Philos. Trans. R. Soc. Lond. B. Biol. Sci. 368, 20120353. doi:10.1098/rstb.2012.0353.

Kölliker, M., Boos, S., Wong, J. W. Y., Röllin, L., Stucki, D., Raveh, S., Wu, M., and Meunier, J. (2015). Parent–offspring conflict and the genetic trade-offs shaping parental investment. Nat. Commun. 6, 6850. doi:10.1038/ncomms7850.

Kölliker, M., Brinkhof, M. W. G., Heeb, P., Fitze, P. S., and Richner, H. (2000). The quantitative genetic basis of offspring solicitation and parental response in a passerine bird with biparental care. Proc. R. Soc. B Biol. Sci. 267, 2127–2132. doi:10.1098/rspb.2000.1259.

Kölliker, M., Brodie III, E. D., and Moore, A. J. (2005). The coadaptation of parental supply and offspring demand. Am. Nat. 166, 506–516. doi:10.1086/491687.

König, B. (1997). Cooperative care of young in mammals. Naturwissenschaften 84, 95–104. doi:10.1007/s001140050356.

Lock, J. E., Smiseth, P. T., and Moore, A. J. (2004). Selection, inheritance, and the evolution of parent-offspring interactions. Am. Nat. 164, 13–24. doi:10.1086/421444.

Lynch, C. B., and Possidente, B. P. (1978). Relationships of maternal nesting to thermoregulatory nesting in house mice (Mus musculus) at warm and cold temperatures. Anim. Behav. 26, 1136–1143. doi:10.1016/0003- 3472(78)90103-3.

Malkesman, O., Lavi-Avnon, Y., Maayan, R., and Weizman, A. (2008). A cross- fostering study in a genetic animal model of depression: Maternal behavior and depression-like symptoms. Pharmacol. Biochem. Behav. 91, 1–8. doi:10.1016/j.pbb.2008.06.004.

Manichaikul, A., Dupuis, J., Sen, S., and Broman, K. W. (2006). Poor Performance of Bootstrap Confidence Intervals for the Location of a Quantitative Trait Locus. Genetics 174, 481–489. doi:10.1534/genetics.106.061549.

Martin, P., and Bateson, P. P. G. (2007). Measuring Behaviour: An Introductory Guide. 3rd ed. Cambridge: Cambridge University Press.

Meunier, J., and Kölliker, M. (2012). Parental antagonism and parent-offspring co-adaptation interact to shape family life. Proc. R. Soc. B Biol. Sci. 279, 3981–3988. doi:10.1098/rspb.2012.1416.

156

Chapter 4: Genetic variation in offspring indirectly influences the quality of maternal behaviour in mice

Ohshima, T., Ward, J. M., Huh, C. G., Longenecker, G., Veeranna, Pant, H. C., Brady, R. O., Martin, L. J., and Kulkarni, A. B. (1996). Targeted disruption of the cyclin-dependent kinase 5 gene results in abnormal corticogenesis, neuronal pathology and perinatal death. Proc. Natl. Acad. Sci. U. S. A. 93, 11173–11178. doi:10.1073/pnas.93.20.11173.

Okamura, T., Yanobu-Takanashi, R., Takeuchi, F., Isono, M., Akiyama, K., Shimizu, Y., Goto, M., Liang, Y. Q., Yamamoto, K., Katsuya, T., et al. (2012). Deletion of CDKAL1 affects high-fat diet-induced fat accumulation and glucose-stimulated insulin secretion in mice, indicating relevance to diabetes. PLoS One 7, e49055. doi:10.1371/journal.pone.0049055.

Sayler, A., and Salmon, M. (1969). Communal nursing in mice: influence of multiple mothers on the growth of the young. Science 164, 1309–1310. doi:10.1126/science.164.3885.1309.

Schneider, T., Skitt, Z., Liu, Y., Deacon, R. M. J., Flint, J., Karmiloff-Smith, A., Rawlins, J. N. P., and Tassabehji, M. (2012). Anxious, hypoactive phenotype combined with motor deficits in Gtf2ird1 null mouse model relevant to Williams syndrome. Behav. Brain Res. 233, 458–473. doi:10.1016/j.bbr.2012.05.014.

Shifman, S., Bell, J. T., Copley, R. R., Taylor, M. S., Williams, R. W., Mott, R., and Flint, J. (2006). A high-resolution single nucleotide polymorphism genetic map of the mouse genome. PLoS Biol. 4, 2227–2237. doi:10.1371/journal.pbio.0040395.

Smith, J. W., Seckl, J. R., Evans, A. T., Costall, B., and Smythe, J. W. (2004). Gestational stress induces post-partum depression-like behaviour and alters maternal care in rats. Psychoneuroendocrinology 29, 227–244. doi:10.1016/S0306-4530(03)00025-8.

Trivers, R. L. (1974). Parent-offspring conflict. Am. Zool. 14, 249–264. doi:10.1093/icb/14.1.249.

Wade, C. M., Kulbokas, E. J., Kirby, A. W., Zody, M. C., Mullikin, J. C., Lander, E. S., Lindblad-Toh, K., and Daly, M. J. (2002). The mosaic structure of variation in the laboratory mouse genome. Nature 420, 574–578. doi:10.1038/nature01252.

Wang, J., Williams, R. W., and Manly, K. F. (2003). WebQTL: web-based complex trait analysis. Neuroinformatics 1, 299–308. doi:10.1385/NI:1:4:299.

Weidt, A., Lindholm, A. K., and König, B. (2014). Communal nursing in wild house mice is not a by-product of group living: Females choose. Naturwissenschaften 101, 73–76. doi:10.1007/s00114-013-1130-6.

Wolf, J. B., and Brodie III, E. D. (1998). The coadaptation of parental and offspring characters. Evolution 52, 299–308. doi:10.2307/2411068.

Wu, Y., Williams, E. G., Dubuis, S., Mottis, A., Jovaisaite, V., Houten, S. M., Argmann, C. A., Faridi, P., Wolski, W., Kutalik, Z., et al. (2014). Multilayered

157

Chapter 4: Genetic variation in offspring indirectly influences the quality of maternal behaviour in mice

genetic and omics dissection of mitochondrial activity in a mouse reference population. Cell 158, 1415–1430. doi:10.1016/j.cell.2014.07.039.

Yamazaki, K., Beauchamp, G. K., Curran, M., Bard, J., and Boyse, E. A. (2000). Parent-progeny recognition as a function of MHC odortype identity. Proc. Natl. Acad. Sci. U. S. A. 97, 10500–10502. doi:10.1073/pnas.180320997.

158

Chapter 5: Indirect genetic effects influence sibling and maternal behaviour in mice.

Chapter 5: Indirect genetic effects influence sibling and maternal

behaviour in mice.

159

Chapter 5: Indirect genetic effects influence sibling and maternal behaviour in mice.

Chapter 5: Preface

This chapter is unpublished and so has been formatted to the style of the previous chapters.

Empirical data was collected by Sophie Lyst.

Analysis was carried out by David Ashbrook, with contributions from Naorin

Sharmin.

The paper was written by David Ashbrook, with editing and comments from

Reinmar Hager.

This chapter builds upon the previous chapter, investigating indirect genetic effects within families, but the change in experimental design allows the examination of sibling effects on behaviour and development.

160

Chapter 5: Indirect genetic effects influence sibling and maternal behaviour in mice.

Indirect genetic effects influence sibling and maternal behaviour in mice.

David G. Ashbrooka,*, Naorin Sharmina, Reinmar Hagera aComputational and Evolutionary Biology, Faculty of Life Sciences, University of

Manchester, Manchester M13 9PT, UK

161

Chapter 5: Indirect genetic effects influence sibling and maternal behaviour in mice.

Abstract

Indirect genetic effects (IGEs) occur when genes expressed in one individual influence traits in a second individual. Significant IGEs are predicted to occur during interactions between family members because there is conflict over resource allocation between parents and offspring, and between siblings.

However, the complex effects of genes expressed in family members on phenotypes in social partners has not been investigated to date. In a large systems-genetics study using the recombinant inbred mouse panel BXD we have set up a population of experimental mouse families using a split litter, cross- fostering design which enables us to determine the effect of genetic variation in offspring, on mothers and siblings. We measured life history traits, maternal behaviour and offspring behaviour throughout lactation and mapped trait variation in focal individuals as a function of the social partner genotype.

Our analysis revealed three loci with significant and complex IGE patterns on both maternal and offspring traits. For example, a locus on chromosome 4 showed a significant direct effect on suckling behaviour but also an indirect effect on nestmate feeding and the level of maternal care. Subsequently, we identified potential candidate genes underlying our loci. Our results show that offspring genotype influences the amount of care they receive from their mother, yet, surprisingly, they also suggest that pups may gain an indirect benefit from their nestmates through increased maternal care.

162

Chapter 5: Indirect genetic effects influence sibling and maternal behaviour in mice.

5.1 Introduction

Early life environment has significant effects on both the physiology and behaviour of animals throughout their life (Anisman et al., 1998; Akers et al.,

2006). Traits influenced by this period, such as weight (Hager et al., 2012a), can alter the fitness of the offspring, and early life environment may result in epigenetically heritable changes, further increasing evolutionary importance

(Francis et al., 1999; Caldji et al., 2000). Two of the most important influencers of early life environment are parents and nestmates. In this study we will concentrate on mothers, as in most mammals they are the primary care givers and provide nutrition in the form of milk (Lee et al., 1991; Champagne et al.,

2003a). Nestmates spend the entire pre-weaning period together, providing time for many interactions to take place (Bergsma et al., 2008). The influence of mothers and nestmates may have a significant impact on development during this early life period.

The parent-offspring conflict hypothesis suggests that offspring will endeavour to secure a larger share of resources from their mother than is optimal for the mother to give which, therefore, will have a detrimental effect on their mother and siblings (Trivers, 1974; Godfray, 1995). This is due to an asymmetry of relatedness, where offspring are twice as related to themselves as they are to their mothers or full-siblings (Hamilton, 1964; Trivers, 1974). Of course, this is applicable to fathers as well, but we are only examining mothers in our experimental design. There are two aspects of parent-offspring conflict which are important for this study: ‘inter-brood conflict’, as the mother has to divide finite resources between the current litter (or brood) and future litters she may have, and ‘intra-brood’ conflict, in which the offspring enter into conflict with the mother and with each other over the division of resources among the current litter

(Macnair and Parker, 1979; Kilner and Hinde, 2012). In our study, ‘inter-brood

163

Chapter 5: Indirect genetic effects influence sibling and maternal behaviour in mice. conflict’ will be detected as offspring conflict with the mother, whereas ‘intra- brood conflict’ will cause conflict with both the mother and nestmates.

An important way that offspring can influence provisioning is by begging, or, more generally speaking, solicitation behaviour. Offspring know their own condition, and how much provisioning they require, whereas parents need this information to known optimal allocation levels, but do not have it (Grodzinski and

Lotem, 2007). Therefore, offspring can exaggerate their need in an attempt to receive more resources, and consequently parents may become less responsive

(Trivers, 1974; Macnair and Parker, 1979).

In contrast to the above conflict hypothesis, parent-offspring co- adaptation has been proposed, stating that it is necessary for mothers and offspring to evolve in compatible ways, leading to maximum inclusive fitness for both individuals (Bateson, 1994; Wolf et al., 1998; Kölliker et al., 2005). A correlation and co-adaptation may develop between offspring genes influencing solicitation behaviour, and maternal genes influencing provisioning (Wolf and

Brodie III, 1998; Kölliker et al., 2005). There can also be co-operation between nestmates, for example grooming each other (allogrooming) or huddling together to improve thermal regulation (Roulin and Dreiss, 2012).

These two concepts, conflict and co-adaptation, are not exclusive, and both may affect the evolution of traits (Meunier and Kölliker, 2012). Co- adaptation allows for the resolution of parent-offspring conflict, selecting for combinations of traits which maximise the fitness of family members, e.g. in the above example high begging offspring with low response mothers, or low begging offspring with high response mothers (Wolf et al., 1998; Kölliker et al.,

2005). This means that in some cases loci will be selected for conflict, while in other cases loci are selected for co-adaptation. Therefore it is of interest to see

164

Chapter 5: Indirect genetic effects influence sibling and maternal behaviour in mice. how offspring influence the care they receive from their mothers and the impact this has on their nestmates.

An important method by which an individual may influence others is by indirect genetic effects (IGEs); that is the genotype of the target individual alters the phenotype of a second individual (Moore et al., 1997; Wolf et al., 1998). IGEs occur because of social interactions between individuals, and we are able to map the loci by only allowing one genotype to change between experiments. It is expected that IGEs are important for parental care for several reasons. Firstly, in relation to the conflict hypothesis, offspring need to alter maternal phenotype e.g. to receive more provisioning. Secondly, the nest environment is stable over a prolonged period of time, giving IGEs the opportunity to take effect (Kölliker et al.,

2012). Previous studies have shown the substantial effect that IGEs can have on growth (Bergsma et al., 2008) and on interactions between mothers and offspring

(chapter 4).

The effect of maternal IGEs on her offspring (also called maternal genetic effects) has been well known for some time (Willham, 1963; Cheverud, 1984), and loci have been mapped for maternal care behaviour (Peripato et al., 2002;

Peripato and Cheverud, 2002). There has been increasing interest and exploration of interactions in the other direction, IGEs of offspring on their mother

(Cierpial et al., 1990; Hager and Johnstone, 2003; van der Veen et al., 2008;

Curley et al., 2010), where it has been shown that the genotype of offspring alters the level of maternal care given. For example, maternal behaviours in mice depends upon both the genotype of the mother and the genotype of the offspring

(Hennessy et al., 1980; Curley et al., 2010). However, finding QTL underlying these offspring IGEs has been slower (Champagne and Curley, 2012). Further, the loci causing IGEs of one sibling upon another have had very limited study. It is known, however, that the genotype of conspecifics can alter behaviour in both

165

Chapter 5: Indirect genetic effects influence sibling and maternal behaviour in mice. rats and mice (Randall and Lester, 1975; Holmes et al., 2005; Curley et al., 2010;

Branchi and Cirulli, 2014), such as exploratory and antagonistic behaviour being dependent upon the genotype of interacting individuals (Hughes, 1989).

In this study we set out to examine how the genotype of a pup influences the phenotype of its mother and nestmates. Consequently, we have investigated

IGEs on post-natal behaviour in the BXD recombinant inbred mouse lines during early life, pre-weaning. Using half-litter cross-fostering of BXD offspring to B6 mothers we show that BXD pups are able to exert IGEs on their foster mothers, resulting in changes in care to both themselves and their B6 nestmates.

166

Chapter 5: Indirect genetic effects influence sibling and maternal behaviour in mice.

5.2 Materials and methods

5.2.1 Experimental animals

To explore IGEs we used the largest mammalian model system, both in terms of number of extant inbred lines available and the number of phenotypes analysed, the recombinant inbred mouse panel BXD (Wu et al., 2014), derived from two divergent mouse strains (C57BL/6J and DBA/2J, hence BXD). The BXD recombinant inbred population consists of experimentally tractable and genetically defined mouse lines capturing a large amount of naturally occurring genetic variation, which underlies variation at the phenotypic level (e.g. Chesler et al., 2005; Mozhui et al., 2011; Hager et al., 2012b; Ashbrook et al., 2014).

Each of the lines has a unique recombination pattern of exactly two possible alleles, incorporate ~5 million segregating SNPs, 500000 insertions and deletions

(indels), and 55000 copy-number variants (Mozhui et al., 2011). In this experiment we used data from ~32 lines (from BXD lines 1, 11, 12, 14, 24, 32,

34, 38-40, 43-45, 48a, 49-51, 55, 56, 60-64, 66-71, 73a, 73b, 73-75, 83, 84, 87,

89, 90, 98 and 102, dependent on the phenotype measured). For each line there were between 1-4 within-line replicates.

C57BL/6J (B6) inbred mice were also used as a genetically uniform strain such that all mothers and half of each litter had the same genotype in all cases.

BXD mice were obtained from Professor Robert W. Williams at the University of

Tennessee Health Science Centre, Memphis, TN., USA, and C57BL/6J mice were obtained from Charles River, UK. All procedures were approved by the

University of Manchester Ethics Committee.

5.2.2 Husbandry and mating protocol

Mice were housed in individually ventilated cages, maintained at 20°C (± 2°) with a relative humidity of 55% (± 10), reverse dark:light cycle with red light between

167

Chapter 5: Indirect genetic effects influence sibling and maternal behaviour in mice.

10:00 and 22:00 hours and food and water ad libitum. Cages were cleaned once a week but never within the first six days after birth to minimise disturbance. The parental mice were all sexually mature and females were unmated. Groups of up to five sibs were housed together in single-sex cages until mating, which occurred between 6-10 weeks of age, when females were ≥ 18g.

Before breeding bedding from males’ cages was added to the females’ cages to encourage synchronised oestrus (Hau and Hau, 2007) and individual males were moved to new cages, to allow them to scent mark. Two days later two sisters were added to the male’s cage. From 16 days after, females were checked daily for evidence of pregnancy. If they were visibly pregnant (weight gain ≥ 8g or distended abdomen), females were separated into an individual cage. This ensured that neither father nor aunt had a social interaction with the offspring.

5.2.3 Cross fostering and data recording

Females in individual cages were checked daily for new born litters. Litters were weighed and cross-fostered, such that each B6 mother had a litter composing of half B6 and half BXD offspring. On a few occasions (< 10% of litters) no corresponding litters were available for cross-fostering, in which case the procedure was delayed by a day. If no corresponding litter was available on the following day, the individuals were removed from the study. Cross-fostered soon after birth meant that all litters were genetically, and as far as possible environmentally, identical with the exception of the genotype of the BXD pups.

For six days following cross-fostering, the litter was left undisturbed, apart from visual checks from outside the cage twice daily, to minimise maternal stress and the risk of infanticide.

168

Chapter 5: Indirect genetic effects influence sibling and maternal behaviour in mice.

5.2.3 Behavioural observations

On the 6th, 10th and 14th day after the litters were cross-fostered observations were made following the example of Hager and Johnstone (2003, 2007). Litters and mothers were weighed then separated for four hours. This stimulates the mothers’ natural foraging behaviour and encourages maternal care when the pups are reunited. Behaviour of mother and pups was recorded during the 15 minutes after the pups were reunited. Behaviour was recorded in an ethogram

(table 5.1), and separated into states (long-lasting commonly occurring activities) and events (short, less common activates). States were recorded every 20 seconds, and events were recorded whenever they occurred. Pups behaviour was recorded as the number of pups (of an individual genotype) engaged in the behaviour at any given time, and an average for the litter was used for statistical analysis. In addition an aggregate phenotype was calculated: maternal care

(nursing + suckling + nestbuilding).

169

Chapter 5: Indirect genetic effects influence sibling and maternal behaviour in mice.

Table 5.1: Maternal and offspring behaviours recorded. Behaviours in bold (states) were recorded every 20 seconds for 15 minutes and behaviours in italics (events) were recorded every time they occurred.

5.2.4 QTL mapping

To compensate for differences between litters, residuals were calculated from a general linear model (GLM) with covariates: maternal bodyweight, average bodyweight of the B6 offspring (weight of the B6 litter divided by the B6 litter size), B6 litter size, average bodyweight of BXD offspring (weight of the BXD litter divided by the BXD litter size), BXD litter size and batch. The least significant of these was then removed from the model, and the model was re-run. This was repeated until the only covariates left were significant (in which case the residuals of the GLM were kept) or until no covariates remained significant (in which case the raw data was kept, as none of the covariates are having a

170

Chapter 5: Indirect genetic effects influence sibling and maternal behaviour in mice. significant effect on the phenotype). All GLM were carried out using SPSS

(version 21, IBM Corporation, Armonk, NY, USA).

For QTL analysis the average per line was then taken and QTL were plotted using the “interval mapping” tool on GeneNetwork. Interval mapping

(Haley and Knott, 1992) relies on 3795 informative SNP markers across all chromosomes, except Y, as implemented in GeneNetwork (GN) (Wang et al.,

2003; Hager et al., 2012b). Loci are identified in GN by the computation of a likelihood ratio statistic score and significance was determined using 5000 permutations of the phenotype data. The number of lines varied between phenotypes, for example due to infanticide leading to a loss of a litter on later days. For all behavioural phenotypes QTL were plotted, and the position of any suggestive (p < 0.63; a threshold which is defined as that which yields, on average, one false positive per genome scan), or significant (p < 0.05) genome- wide p-value was recorded. Confidence intervals were given by a LOD drop of

1.5 (Manichaikul et al., 2006).

All significant QTL were recorded, as well as traits which had a suggestive QTL at the same location as a significant QTL. These traits were then correlated within GN, using Spearman’s correlation (due to < 30 lines being used in some cases and the presence of outliers in some traits).

5.2.5 Identification of candidate genes

To identify candidates, we firstly used the ‘Phenotypes, Alleles & Disease Models

Search’ (http://www.informatics.jax.org/allele) on Mouse genome informatics

(Eppig et al., 2015; Bello et al., 2015) to find phenotypes associated with each of the genes within the loci we identified. We used this, as it is a database of phenotypes found in knockout and other mutant mouse models. As we do not know what tissue(s) any genetic variants will be acting upon then this provides an

171

Chapter 5: Indirect genetic effects influence sibling and maternal behaviour in mice. unbiased examination of genes within the loci. Since the BXD genotype was influencing the phenotype of the mother and nestmates we assumed that the causative gene must be altering the behaviour of BXD pups. For each of the loci identified we searched the ‘Phenotypes, Alleles & Disease Models Search’ database. This list was then manually examined for QTL and behaviours which may be relevant to the phenotypes measured, i.e. development or behaviour related.

QTLminer (Alberts and Schughart, 2010) was used to summarise information about candidate genes, including if they have non-synonymous SNPs

(nsSNPs) or insertions or deletions (indels) in the BXD lines.

172

Chapter 5: Indirect genetic effects influence sibling and maternal behaviour in mice.

5.3 Results

IGEs can confound the mapping of direct genetic effects (DGEs), as the phenotype is being influenced by both the focal individual and interacting individuals (Moore et al., 1997; Wolf, 2003). We control for this by keeping the B6 nestmates and B6 foster mother genotype constant, so that the only difference between their phenotypes is the IGEs of the BXD offspring (Figure 5.1). The mapping of DGEs in BXD offspring traits may be influenced by IGEs from their foster mother (i.e. maternal genetic effects) or nestmates, but again, the consistent B6 genotype should mean consistent IGEs on all BXD litters.

Figure 5.1 Experimental half-litter cross-fostering design. B6 females (large dark mice) adopt half-litters of different lines of the BXD panel (small white to grey mice), and half-litters of B6 offspring (small dark mice).

173

Chapter 5: Indirect genetic effects influence sibling and maternal behaviour in mice.

Our results showed that although QTL with genome-wide significance only showed up in a small number of the traits measured, these same QTL locations appeared in other traits at a suggestive level. Further, those traits which shared a QTL are often correlated.

We can identify three loci which have a suggestive or significant QTL

(genome-wide p-value < 0.05) on at least one BXD behavioural phenotype, and correlate with B6 offspring and/or B6 maternal phenotypes which themselves show a suggestive or significant QTL at the same position, demonstrating IGEs.

Because IGEs are mediated by social interactions (Moore et al., 1997; Kölliker et al., 2012), we refer to these QTL as social interaction QTL, as the DGEs are causing a social interaction, and the IGEs are caused by the social interaction.

Therefore we have named these traits SocInt2.1, SocInt4.1 and SocInt15.1.

5.3.1 Day 14 activity – SocInt2.1

The first QTL is on chromosome 2, from ~ 67.516 - 80.16 megabase pairs (Mbp)

(Figure 5.2; Supplementary table 5.1), where the C57BL/6J allele increases the time spent resting in the nest by both the BXD and B6 offspring and decreases solicitation by BXD pups on day 14. The recording of these two traits were not dependent upon each other: pups could have been resting in the nest and performing solicitation behaviour, or moving around the nest and not performing solicitation behaviour. This QTL only contains phenotypes for the BXD and B6 pups, which indicates a DGE of BXD offspring’s genotype on their own phenotype and a consequent IGE on the phenotype of their B6 nestmates.

The correlation analysis (Table 5.2) shows that on day 14 the time spent by B6 offspring resting in the nest increases with the time spent by BXD offspring resting in the nest. Secondly, both of these traits correlate negatively with BXD offspring solicitation, indicating that pups which are less active also solicit less

174

Chapter 5: Indirect genetic effects influence sibling and maternal behaviour in mice. feeding. We can hypothesise that the BXD pups are not soliciting maternal care

(resulting in reduced solicitation and more time spent resting in the nest, rather than pursuing the mother), causing the mother to show reduced maternal care behaviour (resulting in a negative correlation with maternal care; rho = -0.63, p =

1.49E-4, n = 29), which in turn results in the B6 offspring also remaining in the nest, resting (resulting in a correlation and an IGE QTL; Figure 5.2, Table 5.2).

Although these phenotypes do correlate with some other traits, such as maternal care shown above, these other traits do not show a suggestive QTL on chromosome 2, indicating that the IGE QTL is only having a strong effect on the pup traits, and that the other traits are being more strongly influenced by other

QTL.

Table 5.2: Summary of correlations between B6 and BXD pup phenotypes which show a QTL on chromosome 2. Spearman’s rho, its p-value and the number of BXD lines used are shown. Pink shading represents a positive correlation, rho ≥ 0.5, whereas green shading represents a negative correlation, rho ≤ -0.5.

B6 offspring resting BXD offspring resting BXD offspring in the nest day 14 in the nest day 14 solicitation day 14

rho = 0.915 rho = -0.502 B6 offspring resting p = 2.00E-15 n = 29 in the nest day 14 n = 29 p = 4.75E-3 BXD offspring rho = 0.915 rho = -0.505 resting in the nest p = 2.00E-15 n = 29 day 14 n = 29 p = 4.53E-3 rho = -0.502 rho = -0.505 BXD offspring n = 29 n = 29 solicitation day 14 p = 4.75E-3 p = 4.53E-3

175

Chapter 5: Indirect genetic effects influence sibling and maternal behaviour in mice.

Figure 5.2: QTL map for traits which have a QTL on chromosome 2 (Table 5.2). The likelihood ratio statistic (LRS) is shown on the left, and the position along chromosome 2 in megabase pairs at the bottom. Each trait is represented by a colour, shown at the top of the map, with the lines showing the traits LRS at each position. The coloured blocks at the top of the figure show the positions of genes. The orange track at the bottom of each map is the SNP Seismograph track, showing the positions of SNPs within the BXD lines.

Adapted from GeneNetwork. The significance of each trait is calculated individually for each trait by permutation, so is not shown. However, the suggestive threshold (p = 0.63) is around LRS 10 and the genome-wide significant threshold (p = 0.05) is around LRS >

17.

5.3.2 Day 6 maternal care and offspring feeding – SocInt4.1

The second QTL was found on chromosome 4, between 3.327 and 13.563 Mbp

(Figure 5.3; Supplementary table 5.2), where the C57BL/6J allele in BXD offspring increases maternal care from their B6 foster mothers. The traits within this group are measures of maternal care behaviour and offspring feeding. The correlations demonstrate that pup feeding is highly correlated with maternal behaviour, which is a predictable result, as for the offspring to feed the mother

176

Chapter 5: Indirect genetic effects influence sibling and maternal behaviour in mice. needs to nurse or suckle them (Table 5.3), and the pups cannot be resting if they are feeding. However, because the only difference between the litters is the BXD offspring genotype, this shows that the level of maternal behaviour is dependent on the BXD offspring genotype, i.e. an IGE. Further, this IGE which increases maternal care to the BXD offspring had a further indirect effect, causing the B6 offspring to also receive more maternal care (demonstrated by positively correlation with B6 offspring feeding; Table 5.3).

Table 5.3: Summary of correlations between B6 mother, and B6 and BXD offspring phenotypes which show a QTL on chromosome 4, and traits which correlate significantly with them. Spearman’s rho, its p-value and the number of lines used are shown. Pink shading represents a positive correlation rho ≥ 0.5, whereas green shading represents a negative correlation, rho ≤ 0.5. Trait names are shaded to represent the

QTL mapped for that trait: orange = significant (p ≤ 0.05) QTL on chromosome 4, yellow

= suggestive (p ≤ 0.63) QTL on chromosome 4.

B6 B6 B6 BXD BXD B6 B6 mothers’ offspring offspring offspring offspring mothers’ mothers’ other feeding in resting in resting in feeding in suckling maternal active day the nest the nest the nest the nest day 6 care day 6 6 day 6 day 6 day 6 day 6 B6 mothers’ rho = -0.853 rho = 0.943 rho = 0.996 rho = -0.987 rho = -0.96 rho = 0.994 p p = 1.71E-10 p < 1.00E-16 p < 1.00E-16 p < 1.00E-16 p < 1.00E-16 < 1.00E-16 suckling day 6 n = 28 n = 28 n = 28 n = 28 n = 28 n = 28 B6 mothers’ rho = -0.853 rho = -0.826 rho = -0.866 rho = 0.862 rho = 0.861 rho = -0.861 other active p = 1.71E-10 p = 4.17E-09 p = 3.24E-11 p = 8.16E-11 p = 1.17E-10 p = 5.88E-11 day 6 n = 28 n = 28 n = 28 n = 28 n = 28 n = 28 B6 mothers’ rho = 0.943 rho = -0.826 rho = 0.928 rho = -0.912 rho = -0.894 rho = 0.925 maternal care p < 1.00E-16 p = 4.17E-09 p = 2.22E-16 p = 1.27E-14 p = 1.11E-15 p = 4.44E-16 day 6 n = 28 n = 28 n = 28 n = 28 n = 28 n = 28 B6 offspring rho = 0.996 rho = -0.866 rho = 0.928 rho = -0.992 rho = -0.965 rho = 0.998 feeding in the p < 1.00E-16 p = 3.24E-11 p = 2.22E-16 p < 1.00E-16 p < 1.00E-16 p < 1.00E-16 nest day 6 n = 28 n = 28 n = 28 n = 28 n = 28 n = 28 B6 offspring rho = -0.987 rho = 0.862 rho = -0.912 rho = -0.992 rho = 0.994 rho = -0.989 resting in the p < 1.00E-16 p = 8.16E-11 p = 1.27E-14 p < 1.00E-16 p < 1.00E-16 p < 1.00E-16 nest day 6 n = 28 n = 28 n = 28 n = 28 n = 28 n = 28 BXD offspring rho = -0.96 rho = 0.861 rho = -0.894 rho = -0.965 rho = 0.994 rho = -0.802 resting in the p < 1.00E-16 p = 1.17E-10 p = 1.11E-15 p < 1.00E-16 p < 1.00E-16 p = 1.44E-9 nest day 6 n = 28 n = 28 n = 28 n = 28 n = 28 n = 28 BXD offspring rho = 0.994 p rho = -0.861 rho = 0.925 rho = 0.998 rho = -0.989 rho = -0.802 feeding in the < 1.00E-16 p = 5.88E-11 p = 4.44E-16 p < 1.00E-16 p < 1.00E-16 p = 1.44E-9 nest day 6 n = 28 n = 28 n = 28 n = 28 n = 28 n = 28

177

Chapter 5: Indirect genetic effects influence sibling and maternal behaviour in mice.

Figure 5.3: QTL map for traits which have a suggestive QTL on chromosome 4, and traits which correlate significantly with them (Table 5.3). The likelihood ratio statistic

(LRS) is shown on the left, and the position along chromosome 4 in megabase pairs at the bottom. Each trait is represented by a colour, shown at the top of the map, with the lines showing the traits LRS at each position. The coloured blocks at the top of the figure show the positions of genes. The orange track at the bottom of each map is the SNP

Seismograph track, showing the positions of SNPs within the BXD lines. Adapted from

GeneNetwork. The significance of each trait is calculated individually for each trait by permutation, so is not shown. However, the suggestive threshold (p = 0.63) is around

LRS 10 and the genome-wide significant threshold (p = 0.05) is around LRS > 17.

5.3.3 Day 14 maternal care and offspring feeding – SocInt15.1

The final cluster of phenotypes are traits for maternal care and offspring feeding on day 14. These traits show a QTL on chromosome 15 (between 3.229 - 6.298

Mbp; Supplementary table 5.3; Figure 5.4). However, some of these traits influenced by the SocInt15.1 QTL correlate with the traits which are influenced by the SocInt2.1 QTL, identified above (Figure 5.4; Table 5.4). Again, as with the

SocInt4.1 QTL, these traits are intuitively related, as to be feeding the pups must be receiving care from their mothers, but again, the differences between litters is due to the BXD offspring genotype, i.e. an IGE. The finding that two QTL,

178

Chapter 5: Indirect genetic effects influence sibling and maternal behaviour in mice.

SocInt15.1 and SocInt2.1 influence traits which are correlated suggests that each of the QTL is influencing a different aspect of the BXD offspring’s behaviour, for instance that SocInt15.1 is influencing solicitation behaviour and SocInt2.1 is influencing activity.

Table 5.4: Summary of correlations between B6 and BXD offspring phenotypes which show a QTL on chromosome 15, and those which correlate significantly with them. Spearman’s rho, its p-value and the number of lines used are shown. Pink shading represents a positive correlation rho ≥ 0.5, whereas green shading represents a negative correlation, rho ≤ 0.5. Trait names are shaded to represent the QTL mapped for that trait: dark green = significant (p ≤ 0.05) QTL on chromosome 15, light green = suggestive (p ≤

0.63) QTL on chromosome 15, blue = suggestive (p ≤ 0.63) QTL on chromosome 2.

B6 BXD B6 offspring B6 B6 offspring BXD mothers’ B6 mothers’ offspring feeding in mothers’ resting in offspring other maternal feeding in the nest day suckling the nest day resting in the active day care day 14 the nest day 14 day 14 14 nest day 14 14 14

B6 mothers’ rho = -0.578 rho = -0.674 rho = -0.578 rho = -0.735 rho = 0.525 rho = 0.580 other active p = 7.23E-4 p = 1.26E-5 p = 6.42E-4 p = 6.44E-7 p = 2.91E-2 p = 7.33E-4 day 14 n = 29 n = 29 n = 29 n = 29 n = 29 n = 29

B6 mothers’ rho = -0.578 rho = 0.759 rho = 0.866 rho = 0.657 rho = -0.586 rho = -0.630 maternal p = 7.23E-4 p = 4.84E-7 p = 2.11E-11 p = 6.67E-5 p = 5.87E-4 p = 1.49E-4 care day 14 n = 29 n = 29 n = 29 n = 29 n = 29 n = 29 B6 offspring feeding in rho = -0.674 rho = 0.759 rho = 0.760 rho = 0.893 rho = -0.577 rho = -0.622 p = 1.26E-5 p = 4.84E-7 p = 4.70E-7 p = 4.05E-13 p = 4.54E-4 p = 1.05E-4 the nest day n = 29 n = 29 n = 29 n = 29 n = 29 n = 29 14 B6 mothers’ rho = -0.578 rho = 0.866 rho = 0.760 rho = 0.634 rho = -0.543 rho = -0.581 suckling day p = 6.42E-4 p = 2.11E-11 p = 4.70E-7 p = 1.58E-4 p = 1.66E-3 p = 5.91E-4 14 n = 29 n = 29 n = 29 n = 29 n = 29 n = 29 BXD offspring rho = -0.735 rho = 0.657 rho = 0.893 rho = 0.634 rho = -0.468 rho = -0.556 feeding in p = 6.44E-7 p = 6.67E-5 p = 4.05E-13 p = 1.58E-4 p = 7.53E-3 p = 9.52E-4 the nest day n = 29 n = 29 n = 29 n = 29 n = 29 n = 29 14 rho = 0.525 B6 offspring rho = -0.586 rho = -0.577 rho = -0.543 rho = -0.468 rho = 0.915 p = 2.91E-2 p = 5.87E-4 p = 4.54E-4 p = 1.66E-3 p = 7.53E-3 p = 2.00E-15 resting in the n = 29 n = 29 n = 29 n = 29 n = 29 n = 29 nest day 14 BXD offspring rho = 0.580 rho = -0.630 rho = -0.622 rho = -0.581 rho = -0.556 rho = 0.915 p = 7.33E-4 p = 1.49E-4 p = 1.05E-4 p = 5.91E-4 p = 9.52E-4 p = 2.00E-15 resting in the n = 29 n = 29 n = 29 n = 29 n = 29 n = 29 nest day 14

179

Chapter 5: Indirect genetic effects influence sibling and maternal behaviour in mice.

Figure 5.4: QTL maps for correlated traits as shown in table 5.4. For each the likelihood ratio statistic (LRS) is shown on the left, and the position along the chromosome in megabase pairs. Each trait is represented by a colour, shown at the top of the map, with the lines showing the traits LRS at each position. The coloured blocks at the top of the lower two panels show the positions of genes. The orange track at the bottom of each map is the SNP Seismograph track, showing the positions of SNPs within the BXD lines. Adapted from GeneNetwork. The significance of each trait is calculated individually for each trait by permutation, so is not shown. However, the suggestive threshold (p = 0.63) is around LRS 10 and the genome-wide significant threshold (p =

0.05) is around LRS > 17

180

Chapter 5: Indirect genetic effects influence sibling and maternal behaviour in mice.

5.3.4 Identification of candidate genes

The three loci identified above (summarised in table 5.5) were examined to determine which genes within the confidence intervals were candidate quantitative trait genes.

Table 5.5: A summary of the traits within each of the three loci identified. The QTL position is the coordinates of the most significant marker(s). The confidence interval is defined by a 1.5 LOD drop, but was not calculated for traits with a p-value > 0.9. LRS, p- value and additive effect size are for the most significant marker(s) and are calculated by

QTL Confidence p- Additive Number Loci Trait LRS position interval value effect of genes 67.516- B6 resting in nest 14 77.355 17.099 0.085 -0.586 132 80.160 BXD resting in nest 67.516- SocInt2.1 77.355 17.507 0.079 -0.618 132 14 80.160 BXD offspring 74.928- 73.315- 16.230 0.149 0.527 50 solicitation 14 76.309 77.878 B6 mothers’ suckling 11.148- 10.826- 18.848 0.058 -9.149 20 day 6 11.507 13.031 B6 mothers’ other 11.148- 10.826- 18.371 0.054 7.969 20 active day 6 11.507 13.031 B6 mothers’ 11.148- 10.826- 19.313 0.037 -9.036 20 maternal care day 6 11.507 13.031 B6 offspring feeding 11.148- 10.826- 18.201 0.058 -0.720 20 in nest day 6 11.507 13.031 SocInt4.1 B6 offspring resting 11.148- 10.826- 18.437 0.06 0.721 20 in the nest day 6 11.507 13.031 BXD offspring 11.148- 3.327- resting in the nest 10.640 0.683 8.203 63 11.507 13.563 day 6 BXD offspring 11.148- 10.826- feeding in the nest 19.205 0.041 -9.297 20 11.507 13.031 day 6 B6 mothers’ other 3.410 na 8.56 0.939 -0.442 na active day 14 B6 mothers’ maternal care day 3.410 3.229-5.324 20.330 0.022 0.622 17 14 B6 offspring feeding 3.619 3.229-6.298 14.566 0.131 0.590 19 in nest day 14 B6 mothers’ suckling 3.619 3.229-6.298 15.276 0.099 0.598 19 SocInt15.1 day 14 BXD offspring feeding in the nest 3.619 3.229-6.298 13.913 0.144 6.028 19 day 14 B6 offspring resting 3.410 na 4.295 1 -0.335 na in the nest day 14 BXD offspring resting in the nest 3.619 na 4.928 1 -0.380 na day 14

181

Chapter 5: Indirect genetic effects influence sibling and maternal behaviour in mice.

Using the ‘Phenotypes, Alleles & Disease Models Search’ we find that the SocInt2.1 locus has been linked for a number of development related phenotypes, such as early growth, body length, and weight at weeks 3, 6 and 10.

Further, it shows that three of the genes have been associated with abnormal suckling behaviour: Dlx2 (Qiu et al., 1995; Nery et al., 2003), Gad1 (Condie et al., 1997) and Zfp650 (also known as Ubr3) (Tasaki et al., 2007). QTLminer shows that Gad1 and Zfp650 both have nsSNPs and indels in the BXD lines.

Therefore, we can suggest these are good candidates, as the alleles are different between the B6 and D2, and disruption of the gene causes changes in relevant phenotypes.

At the SocInt4.1 locus the ‘Phenotypes, Alleles & Disease Models Search’ did not contain any QTL at the same locus for relevant phenotypes, but does contain two genes which have been linked to early life behaviour previously:

Chd7, which is associated with abnormal suckling behaviour (Hurd et al., 2010), and abnormal maternal nurturing (Hurd et al., 2007) and Runx1t1, deletion of which causes the absence of gastric milk in neonates (Calabi et al., 2001). Chd7 has nsSNPs and Runx1t1 has both nsSNPs and indels, and are therefore strong candidates. Further, only 32 genes within this QTL have nsSNPs or indels, which is a small list of potential candidate genes.

Finally, the SocInt15.1 locus overlaps with previously identified QTL for tail length and weight at 3 weeks old, both of which are indicators of early life growth. Again, we find two genes altering behaviours, Ghr and Sepp1. Ghr influences feeding behaviour and growth (Egecioglu et al., 2006), while Sepp1 influences parental behaviour (Hill et al., 2003) and grooming behaviour (Raman et al., 2012). Sepp1 has nsSNPs and Ghr has both nsSNPs and indels. Within this locus there are only eight genes with nsSNPs or indels within the BXD lines, so again, this gives a very small number of candidate genes. That we find two

182

Chapter 5: Indirect genetic effects influence sibling and maternal behaviour in mice. genes out of these eight which are plausibly linked to the observed phenotypes is highly encouraging.

183

Chapter 5: Indirect genetic effects influence sibling and maternal behaviour in mice.

5.4 Discussion

An obvious caveat of our method of finding candidate genes within QTL is that it requires that the gene has already been linked to phenotypes which may be involved in maternal and offspring interactions. Therefore, we are biased against understudied genes. The interactions between mothers and offspring are highly complex, and therefore many genes and many systems will be involved. For example, an increase in feeding may be due to increased vocalisation, changes in odour cues, increased metabolism, increased activity etc., all of which would be under the control of different genes in different tissues. Further experiments would be required to clarify the causative gene(s) within each QTL, e.g. targeted gene mutations or mice with greater genetic diversity at the loci. In particular, both SocInt4.1 and SocInt15.1 are relatively narrow regions, containing few genes, and even fewer genes with nsSNPs or indels (32/63 and 8/19 respectively). This makes it feasible to carry out more targeted studies, especially if genes are ranked, e.g. by using GeneNetwork to prioritise those genes which are known to be expressed in the brain of BXD mice.

We have to acknowledge the fact that we are using inbred lines, many of which were derived from ‘fancy’ mice bred in the 19th century (Wade et al., 2002), to make inferences about evolutionary processes. These lines therefore have been outside the reach of natural selection for many decades and instead under the artificial selection of laboratory conditions. Despite this, maternal care is fundamental for offspring survival, and therefore should be maintained in the absence of natural selection under standard laboratory conditions. It would not seem to be possible to do this analysis in wild derived mice, as there will be a number of different genotypes within each litter, which would prevent mapping of offspring QTL to maternal traits or to their inbred littermates. Even if a single wild derived pup was used within the litter, we would still have to acknowledge that

184

Chapter 5: Indirect genetic effects influence sibling and maternal behaviour in mice. the response of the B6 inbred mice might be influenced by being laboratory derived.

A difficulty in finding candidate genes for the complex traits examined here is that we do not know where or when gene expression is influencing the phenotype. As a result, we could not look specifically at any tissue or time point.

Further, very little transcriptome data has been collected in BXD pups, although this is being rectified (e.g. Gaglani et al., 2009).Despite this there are several good candidate genes within each of our loci, and we can hypothesise about how they are causing changes in phenotype.

5.4.1 SocInt2.1

Gad1 and Ubr3 are our top candidates within SocInt2.1. Gad1 is key to the synthesis of γ-aminobutyric acid (GABA) and is involved in the formation of the palate, and knockout can cause difficulty suckling (Condie et al., 1997). Loss of function is known to influence social behaviour (Zhang et al., 2014; Sandhu et al.,

2014). Therefore, it could be that the reduced solicitation observed in the BXD offspring could be a reflection of dysfunctional social behaviour, or physical difficulty suckling. Ubr3 knockout pups show background dependent prenatal mortality, or post-natal mortality due to a reduction in suckling, and therefore starvation (Tasaki et al., 2007). However, with a reduced litter-size mortality was also reduced (Tasaki et al., 2007), and the authors hypothesise that the phenotype is due to disruption of the sensory systems. Hence, it is possible that our BXD pups are not getting olfactory cues, and are consequently not attempting to suckle, resulting in more time spent resting in the nest. That the phenotype in knockout mice was shown to be dependent on genetic background suggests an epistatic effect, which would need to be considered when studying this locus further.

185

Chapter 5: Indirect genetic effects influence sibling and maternal behaviour in mice.

5.4.2 SocInt4.1

Chd7 is a plausible candidate genes, as it is important for development, and mutations within the human homologue can cause CHARGE syndrome, a developmental disorder, in humans (Janssen et al., 2012). In mice a full knockout is lethal, and heterozygous knockouts show hearing deficits and increased mortality at weaning (Hurd et al., 2007, 2010). In both mouse and human, Chd7 mutations are associated with defects within the ear (Hurd et al., 2010; Janssen et al., 2012), and therefore it could be that the changes in the maternal care associated with the SocInt4.1 locus are due to changes in the pups’ ability to communicate effectively. However, as Chd7 is expressed in many organs, it could be associated with a number of phenotypes which could lead to changes in maternal care, for example abnormalities of the pituitary gland were observed

(Hurd et al., 2007), so it could be hormonal changes which are changing levels of offspring solicitation.

5.4.3 SocInt15.1

Sepp1 appears to be a good candidate at this locus. Sepp1-/- knockout offspring of Sepp1-/- knockout mothers show an increased rate of mortality before weaning, compared with Sepp1+/- heterozygous, or Sepp1-/- knockout offspring raised by

Sepp1+/- heterozygous mothers (Hill et al., 2003). This suggests that there is an effect of both mother and offspring genotype at this locus influencing post-natal survival before weaning.

5.4.4 Implications and future directions

Our finding that IGEs may have a positive effect on siblings would seem at odds with common perceptions of ‘intra-brood’ competition, however other studies have reported similar results (Mutic and Wolf, 2007; Bergsma et al., 2008). This is perhaps not surprising if we consider the framework of kin selection: offspring

186

Chapter 5: Indirect genetic effects influence sibling and maternal behaviour in mice. will have a relatedness coefficient of 0.5 with both their mother and (in normal circumstances) their nestmates (Hamilton, 1964). Therefore, as long as cost is equal or less than half of the gain received by the nestmates, this altruistic behaviour which increases provisioning to nestmates will be selected for (Roulin and Dreiss, 2012). It would be interesting to examine if this positive relationship holds when competition increases, for example by increasing litter size and therefore decreasing the resources available per pup, and, as a result, to be able to map ‘genotype by family environment’ interactions (Kölliker et al., 2012; Hager et al., 2012a). Further, it would also be interesting to look at ‘inter-brood competition’, i.e. if the genotype of previous litters influences the phenotype of later litters. This may show a different pattern, as in polygamous species such as mice, offspring will only expect to have a relatedness of 0.25 to their mother’s future offspring, rather than the relatedness of 0.5 between members of the same litter. To test this, BXD offspring could be fostered to B6 mothers, allowed to be weaned, and then the B6 mothers bred again, with the phenotype of her second biological litter being mapped to the genotype of her first foster litter.

We have assumed that the benefit to nestmates is coincidental, i.e. that increased solicitation brings the mother to the nest and the nestmates can then be fed at the same time, and that the pups do not identify if nestmates are full siblings or not. Using the BXD lines we can test this assumption, by using different combinations of BXD litters, which have different percentages of B6 derived alleles, and are therefore more or less genetically similar to the B6 genotype. Different experimental setups could then tell us if: a) mothers alter their provisioning dependent on how closely related they are to the foster offspring and b) if the indirect effect of nestmate genotype is dependent upon their genetic similarity. It has been shown previously that mothers and pups are able to determine differences in the major histocompatibility complex (MHC)

187

Chapter 5: Indirect genetic effects influence sibling and maternal behaviour in mice. between each other and that this can lead to differences in maternal behaviour

(Yamazaki et al., 2000), therefore the possibility nestmates may be able identify each other’s MHC genotype seems likely.

Further, although we have been considering how maternal genotype influences offspring phenotype (chapter 4), how offspring genotype influences maternal phenotype (chapters 4 and 5), and how one offspring’s genotype alters another’s phenotype (chapter 5), these do not happen in isolation, with all these

IGEs occurring at the same time, in the same family, along with DGEs. This allows for the possibility of more complex interactions taking place, such as epistasis between alleles expressed in different individuals. Such ‘social epistasis’ could be investigated using an alteration to the experimental design demonstrated here, perhaps varying pairs of genotypes simultaneously, although new mapping strategies would be required (e.g. as suggested by Cui et al.

(2004).

We did not find any significant QTL affecting offspring growth in either the

BXD or B6 pups. This is intriguing, as it would be expected that differences in maternal care would result in differences in growth. However, there are a number of potential explanations for this. Firstly, it could be that some genotypes get increased maternal care due to increased solicitation, but that the solicitation is costly enough that the increased care is countered by the cost. A second possibility is that we used relatively small litter sizes, with mothers having unlimited access to food, and therefore all the offspring were receiving enough care to grow at a high rate, even though there was variation within the amount of care. A third explanation is that maternal care was only measured after separation, and therefore the QTL we detect are for maternal care after separation, and not for maternal care under uninterrupted conditions, which could balance the short term gains. A final, and highly probable, possibility is that

188

Chapter 5: Indirect genetic effects influence sibling and maternal behaviour in mice. growth is under the control of many loci, all of which have a lower level of effect size, and therefore could not be detected with the number of lines used. Although offspring growth is often used as a proxy for offspring fitness, it could be that other phenotypes, not examined here, could be altered, and therefore change the pups’ fitness. For example, the amount of parental care that offspring give as adults is influenced by the amount of parental care they themselves received

(Champagne et al., 2001, 2003b), and this could increase the fitness of the next generation.

It should also be noted that the QTL found here for IGEs are not those found previously (chapter 4). It could be that the presence of the B6 half litter is having an effect, as although the effect they have on each litter is consistent, the

B6 offspring will still be having their own IGEs on their B6 mother and BXD nestmates. Another potential reason is that there could be a number of loci showing IGEs on maternal care, all of which have an effect size close to our detection threshold, and therefore it is chance which of these QTL we detect. In this case we would expect that both sets of IGEs (those from this chapter and the previous one) would be found if a larger study with more BXD lines was carried out. Finally, it has been shown that cross-fostering itself can alter maternal care, even if both the offspring and mother are the same genotype (Curley et al.,

2010). As the B6 offspring used in this study were the B6 mothers’ biological offspring, using B6 offspring fostered from another mother could have caused different IGE loci to be found.

5.4.5 Conclusion

In conclusion, we are able to show that the BXD pups’ genotype can have an

IGE on maternal care, resulting in changes to nestmate phenotype. Further, we

189

Chapter 5: Indirect genetic effects influence sibling and maternal behaviour in mice. observe that the effect is positive, i.e. the B6 nestmates of BXD pups which receive increased maternal care also receive increased maternal care.

190

Chapter 5: Indirect genetic effects influence sibling and maternal behaviour in mice.

5.5 References

Akers, K. G., Nakazawa, M., Romeo, R. D., Connor, J. A., McEwen, B. S., and Tang, A. C. (2006). Early life modulators and predictors of adult synaptic plasticity. Eur. J. Neurosci. 24, 547–554. doi:10.1111/j.1460- 9568.2006.04921.x.

Alberts, R., and Schughart, K. (2010). QTLminer: identifying genes regulating quantitative traits. BMC Bioinformatics 11, 516. doi:10.1186/1471-2105-11- 516.

Anisman, H., Zaharia, M. D., Meaney, M. J., and Merali, Z. (1998). Do early-life events permanently alter behavioral and hormonal responses to stressors? Int. J. Dev. Neurosci. 16, 149–164. doi:10.1016/S0736-5748(98)00025-2.

Ashbrook, D. G., Delprato, A., Grellmann, C., Klein, M., Wetzel, R., Overall, R. W., and Badea, A. (2014). Transcript co-variance with Nestin in two mouse genetic reference populations identifies Lef1 as a novel candidate regulator of neural precursor cell proliferation in the adult hippocampus. Front. Neurosci. 8, 418. doi:10.3389/fnins.2014.00418.

Bateson, P. (1994). The dynamics of parent-offspring relationships in mammals. Trends Ecol. Evol. 9, 399–403. doi:10.1016/0169-5347(94)90066-3.

Bello, S. M., Smith, C. L., and Eppig, J. T. (2015). Allele, phenotype and disease data at Mouse Genome Informatics: improving access and analysis. Mamm. Genome 26, 285–294. doi:10.1007/s00335-015-9582-y.

Bergsma, R., Kanis, E., Knol, E. F., and Bijma, P. (2008). The contribution of social effects to heritable variation in finishing traits of domestic pigs (Sus scrofa). Genetics 178, 1559–1570. doi:10.1534/genetics.107.084236.

Branchi, I., and Cirulli, F. (2014). Early experiences: Building up the tools to face the challenges of adult life. Dev. Psychobiol. 56, 1661–1674. doi:10.1002/dev.21235.

Calabi, F., Pannell, R., and Pavloska, G. (2001). Gene targeting reveals a crucial role for MTG8 in the gut. Mol. Cell. Biol. 21, 5658–5666. doi:10.1128/MCB.21.16.5658-5666.2001.

Champagne, F. A., and Curley, J. P. (2012). “Genetics and epigenetics of parental care,” in The evolution of parental care, eds. N. Royle, P. T. Smiseth, and M. Kölliker (Oxford, UK: Oxford University Press), 304–324. doi:10.1093/acprof:oso/9780199692576.003.0017.

Champagne, F. A., Diorio, J., Sharma, S., and Meaney, M. J. (2001). Naturally occurring variations in maternal behavior in the rat are associated with

191

Chapter 5: Indirect genetic effects influence sibling and maternal behaviour in mice.

differences in estrogen-inducible central oxytocin receptors. Proc. Natl. Acad. Sci. U. S. A. 98, 12736–12741. doi:10.1073/pnas.221224598.

Champagne, F. A., Weaver, I. C. G., Diorio, J., Sharma, S., and Meaney, M. J. (2003b). Natural variations in maternal care are associated with estrogen receptor alpha expression and estrogen sensitivity in the medial preoptic area. Endocrinology 144, 4720–4724. doi:10.1210/en.2003-0564.

Cheverud, J. M. (1984). Evolution by kin selection: A quantitative genetic model illustrated by maternal performance in mice. Evolution 38, 766–777. doi:10.2307/2408388.

Cierpial, M. A., Murphy, C. A., and McCarty, R. (1990). Maternal behavior of spontaneously hypertensive and Wistar—Kyoto normotensive rats: Effects of reciprocal cross-fostering of litters. Behav. Neural Biol. 54, 90–96. doi:10.1016/0163-1047(90)91271-C.

Condie, B. G., Bain, G., Gottlieb, D. I., and Capecchi, M. R. (1997). Cleft palate in mice with a targeted mutation in the gamma-aminobutyric acid-producing enzyme glutamic acid decarboxylase 67. Proc. Natl. Acad. Sci. U. S. A. 94, 11451–11455. doi:10.1073/pnas.94.21.11451.

Cui, Y., Casella, G., and Wu, R. (2004). Mapping quantitative trait loci interactions from the maternal and offspring genomes. Genetics 167, 1017– 1026. doi:10.1534/genetics.103.024398.

Egecioglu, E., Bjursell, M., Ljungberg, A., Dickson, S. L., Kopchick, J. J., Bergström, G., Svensson, L., Oscarsson, J., Törnell, J., and Bohlooly-Y, M. (2006). Growth hormone receptor deficiency results in blunted ghrelin feeding response, obesity, and hypolipidemia in mice. Am. J. Physiol. Endocrinol. Metab. 290, E317–E325. doi:10.1152/ajpendo.00181.2005.

192

Chapter 5: Indirect genetic effects influence sibling and maternal behaviour in mice.

Francis, D., Diorio, J., Liu, D., and Meaney, M. J. (1999). Nongenomic transmission across generations of maternal behavior and stress responses in the rat. Science 286, 1155–1158. doi:10.1126/science.286.5442.1155.

Gaglani, S. M., Lu, L., Williams, R. W., and Rosen, G. D. (2009). The genetic control of neocortex volume and covariation with neocortical gene expression in mice. BMC Neurosci. 10, 44. doi:10.1186/1471-2202-10-44.

Godfray, H. C. (1995). Evolutionary theory of parent-offspring conflict. Nature 376, 133–138. doi:10.1038/376133a0.

Grodzinski, U., and Lotem, A. (2007). The adaptive value of parental responsiveness to nestling begging. Proc. R. Soc. B Biol. Sci. 274, 2449– 2456. doi:10.1098/rspb.2007.0658.

Hager, R., Cheverud, J. M., and Wolf, J. B. (2012a). Genotype-dependent responses to levels of sibling competition over maternal resources in mice. Heredity 108, 515–520. doi:10.1038/hdy.2011.115.

Hager, R., and Johnstone, R. A. (2007). Maternal and offspring effects influence provisioning to mixed litters of own and alien young in mice. Anim. Behav. 74, 1039–1045. doi:10.1016/j.anbehav.2007.01.021.

Hager, R., and Johnstone, R. A. (2003). The genetic basis of family conflict resolution in mice. Nature 421, 533–535. doi:10.1038/nature01239.

Hager, R., Lu, L., Rosen, G. D., and Williams, R. W. (2012b). Genetic architecture supports mosaic brain evolution and independent brain-body size regulation. Nat. Commun. 3, 1079. doi:10.1038/ncomms2086.

Haley, C. S., and Knott, S. A. (1992). A simple regression method for mapping quantitative trait loci in line crosses using flanking markers. Heredity 69, 315–324. doi:10.1038/hdy.1992.131.

Hamilton, W. D. (1964). The genetical evolution of social behaviour. I. J. Theor. Biol. 7, 1–16. doi:10.1016/0022-5193(64)90038-4.

Hau, A. R., and Hau, J. (2007). Correlation between duration of courtship and litter size in outbred NMRI mice. In Vivo 21, 33–34.

Hennessy, M. B., Li, J., Lowe, E. L., and Levine, S. (1980). Maternal behavior, pup vocalizations, and pup temperature changes following handling in mice of 2 inbred strains. Dev. Psychobiol. 13, 573–584. doi:10.1002/dev.420130603.

Hill, K. E., Zhou, J., McMahan, W. J., Motley, A. K., Atkins, J. F., Gesteland, R. F., and Burk, R. F. (2003). Deletion of selenoprotein P alters distribution of selenium in the mouse. J. Biol. Chem. 278, 13640–13646. doi:10.1074/jbc.M300755200.

Holmes, A., le Guisquet, A. M., Vogel, E., Millstein, R. A., Leman, S., and Belzung, C. (2005). Early life genetic, epigenetic and environmental factors

193

Chapter 5: Indirect genetic effects influence sibling and maternal behaviour in mice.

shaping emotionality in rodents. Neurosci. Biobehav. Rev. 29, 1335–1346. doi:10.1016/j.neubiorev.2005.04.012.

Hughes, A. L. (1989). Interaction between strains in the social relations of inbred mice. Behav. Genet. 19, 685–700. doi:10.1007/BF01066031.

Hurd, E. A., Capers, P. L., Blauwkamp, M. N., Adams, M. E., Raphael, Y., Poucher, H. K., and Martin, D. M. (2007). Loss of Chd7 function in gene- trapped reporter mice is embryonic lethal and associated with severe defects in multiple developing tissues. Mamm. Genome 18, 94–104. doi:10.1007/s00335-006-0107-6.

Hurd, E. A., Poucher, H. K., Cheng, K., Raphael, Y., and Martin, D. M. (2010). The ATP-dependent chromatin remodeling enzyme CHD7 regulates pro- neural gene expression and neurogenesis in the inner ear. Development 137, 3139–3150. doi:10.1242/dev.047894.

Janssen, N., Bergman, J. E. H., Swertz, M. A., Tranebjaerg, L., Lodahl, M., Schoots, J., Hofstra, R. M. W., van Ravenswaaij-Arts, C. M. A., and Hoefsloot, L. H. (2012). Mutation update on the CHD7 gene involved in CHARGE syndrome. Hum. Mutat. 33, 1149–1160. doi:10.1002/humu.22086.

Kölliker, M., Brodie III, E. D., and Moore, A. J. (2005). The coadaptation of parental supply and offspring demand. Am. Nat. 166, 506–516. doi:10.1086/491687.

Kölliker, M., Royle, N. J., and Smiseth, P. T. (2012). “Parent–offspring co‐ adaptation,” in The evolution of parental care, eds. N. J. Royle, P. T. Smiseth, and M. Kölliker (Oxford, UK: Oxford University Press), 285–303. doi:10.1093/acprof:oso/9780199692576.003.0016.

Lee, P. C., Majluf, P., and Gordon, I. J. (1991). Growth, weaning and maternal investment from a comparative perspective. J. Zool. 225, 99–114. doi:10.1111/j.1469-7998.1991.tb03804.x.

Macnair, M. R., and Parker, G. A. (1979). Models of parent-offspring conflict. III. Intra-brood conflict. Anim. Behav. 27, 1202–1209. doi:10.1016/0003- 3472(79)90067-8.

Manichaikul, A., Dupuis, J., Sen, S., and Broman, K. W. (2006). Poor performance of bootstrap confidence intervals for the location of a quantitative trait locus. Genetics 174, 481–489. doi:10.1534/genetics.106.061549.

Meunier, J., and Kölliker, M. (2012). Parental antagonism and parent-offspring co-adaptation interact to shape family life. Proc. R. Soc. B Biol. Sci. 279, 3981–3988. doi:10.1098/rspb.2012.1416.

194

Chapter 5: Indirect genetic effects influence sibling and maternal behaviour in mice.

Mutic, J. J., and Wolf, J. B. (2007). Indirect genetic effects from ecological interactions in Arabidopsis thaliana. Mol. Ecol. 16, 2371–2381. doi:10.1111/j.1365-294X.2007.03259.x.

Nery, S., Corbin, J. G., and Fishell, G. (2003). Dlx2 progenitor migration in wild type and Nkx2.1 Mutant telencephalon. Cereb. Cortex 13, 895–903. doi:10.1093/cercor/13.9.895.

Peripato, A. C., and Cheverud, J. M. (2002). Genetic influences on maternal care. Am. Nat. 160 Suppl, S173–S185. doi:10.1086/342900.

Qiu, M., Bulfone, A., Martinez, S., Meneses, J. J., Shimamura, K., Pedersen, R. A., and Rubenstein, J. L. R. (1995). Null mutation of Dlx-2 results in abnormal morphogenesis of proximal first and second branchial arch derivatives and abnormal differentiation in the forebrain. Genes Dev. 9, 2523–2538. doi:10.1101/gad.9.20.2523.

Raman, A. V., Pitts, M. W., Seyedali, A., Hashimoto, A. C., Seale, L. A., Bellinger, F. P., and Berry, M. J. (2012). Absence of selenoprotein P but not selenocysteine lyase results in severe neurological dysfunction. Genes, Brain Behav. 11, 601–613. doi:10.1111/j.1601-183X.2012.00794.x.

Randall, C., and Lester, D. (1975). Social modification of alcohol consumption in inbred mice. Science 189, 149–151. doi:10.1126/science.1138373.

Roulin, A., and Dreiss, A. N. (2012). “Sibling competition and cooperation over parental care,” in The evolution of parental care, eds. N. Royle, P. T. Smiseth, and M. Kölliker (Oxford, UK: Oxford University Press), 133–149. doi:10.1093/acprof:oso/9780199692576.003.0008.

Sandhu, K. V., Lang, D., Müller, B., Nullmeier, S., Yanagawa, Y., Schwegler, H., and Stork, O. (2014). Glutamic acid decarboxylase 67 haplodeficiency impairs social behavior in mice. Genes, Brain Behav. 13, 439–450. doi:10.1111/gbb.12131.

Tasaki, T., Sohr, R., Xia, Z., Hellweg, R., Hörtnagl, H., Varshavsky, A., and Kwon, Y. T. (2007). Biochemical and genetic studies of UBR3, a ubiquitin ligase with a function in olfactory and other sensory systems. J. Biol. Chem. 282, 18510–18520. doi:10.1074/jbc.M701894200.

195

Chapter 5: Indirect genetic effects influence sibling and maternal behaviour in mice.

Trivers, R. L. (1974). Parent-offspring conflict. Am. Zool. 14, 249–264. doi:10.1093/icb/14.1.249.

Van der Veen, R., Abrous, D. N., de Kloet, E. R., Piazza, P. V, and Koehl, M. (2008). Impact of intra- and interstrain cross-fostering on mouse maternal care. Genes, Brain Behav. 7, 184–192. doi:10.1111/j.1601- 183X.2007.00337.x.

Wang, J., Williams, R. W., and Manly, K. F. (2003). WebQTL: web-based complex trait analysis. Neuroinformatics 1, 299–308. doi:10.1385/NI:1:4:299.

Willham, R. L. (1963). The covariance between relatives for characters composed of components contributed by related individuals. Biometrics 19, 18–27. doi:10.2307/2527570.

Wolf, J. B. (2003). Genetic architecture and evolutionary constraint when the environment contains genes. Proc. Natl. Acad. Sci. U. S. A. 100, 4655– 4660. doi:10.1073/pnas.0635741100.

Wolf, J. B., and Brodie III, E. D. (1998). The coadaptation of parental and offspring characters. Evolution 52, 299–308. doi:10.2307/2411068.

Wolf, J. B., Brodie III, E. D., Cheverud, J. M., Moore, A. J., and Wade, M. J. (1998). Evolutionary consequences of indirect genetic effects. Trends Ecol. Evol. 13, 64–69. doi:10.1016/S0169-5347(97)01233-0.

Zhang, K., Hill, K., Labak, S., Blatt, G. J., and Soghomonian, J.-J. (2014). Loss of glutamic acid decarboxylase (Gad67) in Gpr88-expressing neurons induces learning and social behavior deficits in mice. Neuroscience 275, 238–247. doi:10.1016/j.neuroscience.2014.06.020.

196

Chapter 6: The BXD lines as a tool to identify parent-of-origin effects with an indirect genetic effect

Chapter 6: The BXD lines as a tool to identify parent-of-origin effects

with an indirect genetic effect

197

Chapter 6: The BXD lines as a tool to identify parent-of-origin effects with an indirect genetic effect

Chapter 6: Preface

This chapter combines elements of my first year continuation report with elements of a perspectives paper published in Frontiers in Neuroanatomy

(Ashbrook and Hager, 2013).

Data was collected and analysed by David Ashbrook.

The chapter was written by David Ashbrook, with edits, comments and intellectual input from Reinmar Hager.

The chapter represents initial empirical work and proof of principal for further analysis. As such, it would not be publishable, but it shown here in a paper format. We use a similar fostering design to chapter 4, but use reciprocal heterozygote litters to investigate parent-of-origin effects which have indirect genetic effects on maternal care.

198

Chapter 6: The BXD lines as a tool to identify parent-of-origin effects with an indirect genetic effect

The BXD lines as a tool to identify parent-of-origin effects with an indirect genetic effect

David Ashbrooka* and Reinmar Hagera aComputational and Evolutionary Biology, Faculty of Life Sciences, University of

Manchester, Manchester M13 9PT, UK

199

Chapter 6: The BXD lines as a tool to identify parent-of-origin effects with an indirect genetic effect

Abstract

Parent-of-origin effects (POEs) can be defined as the phenotypic effect of an allele which depends upon whether the allele was inherited from the mother or the father, and not on the sequence of the allele inherited. POEs have been shown to have a significant impact during early life on both development and behaviour.

In this study we use reciprocal heterozygous crosses between lines of the

BXD recombinant inbred panel to identify phenotypes which show POEs.

Reciprocal heterozygotes are genetically identical, only differing between which chromosome from each pair was inherited from each parent. A fostering design allows us to separate POEs from post-natal maternal effects. We measured life history traits, maternal behaviour and offspring behaviour throughout lactation, and then looked for direct genetic effects of POEs (i.e. a significant difference in phenotype between reciprocal heterozygous pups) and indirect genetic effects of

POEs (i.e. a significant difference in maternal phenotypes of mothers, dependent upon the reciprocal heterozygous pups they fostered).

This initial analysis reveals POEs on offspring bodyweight at day 21, and

POEs on phenotypes in the foster mother, resulting in differences in maternal care on post-natal day 14 dependent upon the genotype of the foster offspring.

This shows that the BXD lines are an appropriate model for examining these effects, and leads the way for future studies using a greater number of BXD lines to map POE loci.

200

Chapter 6: The BXD lines as a tool to identify parent-of-origin effects with an indirect genetic effect

6.1 Introduction

2008). These epigenetic effects are due to changes in gene expression, but are not a result of a change in the nucleotide sequence. There are several mechanism by which POEs can take place or which show a similar pattern to

POEs, including the effect of mitochondria DNA, sex chromosomes and in utero effects, reviewed by Guilmatre and Sharp (2012), but the most well-known POE is genomic imprinting.

Genomic imprinting is a specific form of POE, defined by differential expression of paternally and maternally inherited genes, and in the most extreme cases complete silencing of the paternal or maternal allele, resulting in monoallelic expression (Khatib, 2007; Cheverud et al., 2008; Wolf et al., 2008a). A paternally expressed (maternally imprinted) gene is passed from the mother to both sons and daughters, but is not expressed in either; instead the allele inherited from the father is active. The maternally expressed gene can only become expressed again in the offspring of her sons, but not the offspring of her daughters. This contrasts with classical Mendelian genetics, where both the maternally inherited and paternally inherited genes are expressed in offspring

(baring nucleotide changes which result in loss of expression). Molecularly, genomic imprinting results from chromosome remodelling and epigenetic modifications, such as methylation, during gametogenesis, which are then maintained through fertilisation and development (Ideraabdullah et al., 2008). It should be noted that genomic imprinting can be both time and tissue dependent, i.e. monoallelic expression of a given gene may only occur in a certain tissue at a certain developmental stage (Garfield et al., 2011).

201

Chapter 6: The BXD lines as a tool to identify parent-of-origin effects with an indirect genetic effect

To investigate POEs we used the largest genetic reference panel in mammals, both in terms of number of extant inbred lines available and the number of phenotypes analysed, the BXD mouse population. These are a panel of inbred lines, all of which are derived from crosses between the C57BL/6J (B6) and the DBA/2J (D2) inbred strains. Genomic imprinting has been shown in both the B6 and D2 lines (Downing et al., 2011) as well as the BXD lines (Shibata et al., 1995) and online data (http://www.genenetwork.org) from the parental reciprocal crosses also shows evidence of POEs. This demonstrates that they are a valid system to explore the phenomenon, but the range of phenotypes which show POEs in BXD has not been explored.

Genomic imprinting has a strong effect on the placenta, with the majority of known imprinted genes being expressed there (Lim and Ferguson-Smith,

2010; Frost and Moore, 2010; Fowden et al., 2011; Renfree et al., 2013), and on the brain, where they can influence behaviour (Allen et al., 1995; Keverne et al.,

1996; Kato et al., 1998; Badcock and Crespi, 2006; Gregg et al., 2010; Garfield et al., 2011; DeVeale et al., 2012; Dent and Isles, 2014; Bonthuis et al., 2015;

Perez et al., 2015). Disruption of normal genomic imprinting can result in severe developmental disorders, which alter both early development and feeding behaviour (Lim and Ferguson-Smith, 2010). Therefore, measuring the development and behaviour of mouse pups provides a set of highly relevant traits to investigate POEs.

To explore genomic imprinting reciprocal heterozygotes (RHs), bred by reciprocal mating of two RI lines, can be used (Figure 6.1). The RH offspring should be genetically identical and therefore differences in phenotype or gene expression will be due to POEs (Hager et al., 2008; Ashbrook and Hager, 2013).

All the BXD lines used are derived from a cross between a DBA/2J male and a

C57BL/6J female, and therefore all lines carry the DBA/2J Y chromosome and

202

Chapter 6: The BXD lines as a tool to identify parent-of-origin effects with an indirect genetic effect the C57BL/6J X chromosome and mitochondria. As the genetics of the F1 RH can be reliably predicted this means that no further genotyping is needed and that the mice can be replicated by other groups to build up a body of data that is easily compared, similar to that which has been produced for the ‘pure’ RI lines, for example at GeneNetwork (genenetwork.org).

Figure 6.1: Production of reciprocal heterozygotes. Reciprocal heterozygotes are bred from two fully inbred parental strains (Strain 1 and Strain 2) to produce offspring with identical genotypes but different phenotypes (in this hypothetical example coat colour showing a maternal expression pattern). Modified from Ashbrook and Hager

(2013).

It has been shown that quantitative trait locus (QTL) analysis can be used to identify imprinted loci (Mantey et al., 2005; Wolf et al., 2008a; Cheverud et al.,

2008). By breeding RHs and then fostering to foster mothers of the B6 inbred mouse strain we can separate the effects of imprinted genes expressed in the offspring from post-natal maternal effects due to the foster mother’s genotype, as

203

Chapter 6: The BXD lines as a tool to identify parent-of-origin effects with an indirect genetic effect has been identified as a complicating factor in previous imprinted QTL studies

(Hager et al., 2008; Leamy et al., 2008; Hager et al., 2009). The B6 strain was chosen as foster mothers as they are one of the original strains, have been shown to demonstrate good maternal care and to accept foster offspring (Priebe et al., 2005; Chourbaji et al., 2011). We examine offspring behaviour and development during the post-natal, pre-weaning period, as imprinted genes have been shown to be important during early life. For instance, the paternally expressed gene Rasgrf1 is expressed in the neonatal mouse brain and becomes biallelically expressed at weaning. The level of expression of Rasgrf1 correlates with size: pups which over-express Rasgrf1 are larger than wildtype, and knockouts are smaller (Drake et al., 2009). Similarly, infant mice lacking expression of the paternally expressed Gnasxl transcript show reduced post- natal growth and suckling impairment (Plagge et al., 2004).

The advantage of RHs is that the effect of variation in imprinted genes can be seen, i.e. both alleles are potentially functional but only one is expressed.

This contrasts with traditional methods of finding imprinting, such as knockouts or uniparental disomy studies which, although they have provided a wealth of information (Schulz et al., 2008), lead to gross abnormalities which are unlikely to survive in a wild population and therefore unlikely to contribute to selection (Wolf et al., 2008b). The use of the more physiologically relevant RI lines might show more subtle changes in phenotype which are more representative of natural variation.

This project will examine early life history traits, including weight gain and maternal provisioning over the first three weeks of life (measured on post-natal days 6, 10 and 14), using an adaptation of the method of Hager and Johnstone

(2003, 2007). Since all pups will be fostered to the same maternal genotype changes in weight should be due to the offspring genotype and their ability to

204

Chapter 6: The BXD lines as a tool to identify parent-of-origin effects with an indirect genetic effect solicit suckling from the foster mother. Further, changes in maternal behaviour must be due to pup genotype and phenotype, as all the B6 foster mothers have identical genotypes, and therefore should have similar phenotypes. This means that alterations in maternal behaviour are due to indirect genetic effects (IGE): the influence of a gene expressed in one individual on the phenotype of a second individual. We would expect to see IGE showing a POE, as the conflicts over the level of maternal care given are one of the main hypothesis as to why POEs evolved (Ashbrook and Hager, 2013). This could result in, for example, paternal expression of genes in offspring which result in increased maternal care, or maternal expression of genes which reduce maternal care.

In this study we set out to examine if POEs on early life development and behaviour can be found in the BXD recombinant inbred lines and if these result in IGEs on maternal phenotype. Using a fostering design of reciprocal heterozygous pups to B6 mothers, we show that POEs can influence pup bodyweight on post-natal day 21, and can result in IGE on maternal behaviour at day 14.

205

Chapter 6: The BXD lines as a tool to identify parent-of-origin effects with an indirect genetic effect

6.2 Materials and methods

6.2.1 Experimental animals

To explore POEs, eight BXD mouse lines (table 6.1) plus B6 mice were used. All mice were housed in individually ventilated cages, maintained at 20°C (±2°) with a relative humidity of 55% (±10), reverse dark:light cycle with red light between

10:00 and 22:00 hours and food and water ad libitum. Cages were cleaned once a week but never within the first 6 days after birth to minimise disturbance. The parental mice were all sexually mature (5–7 weeks of age) and females were unmated and maintained together. Bedding from the male’s cage was placed in the females’ cage to synchronise oestrus (Hau and Hau, 2007). Females were placed into a male’s cage either individually or in pairs. Female weight was checked 18 days later, and pregnancy determined by weight gain (> 8g) and/or distended abdomen. Pregnant females were separated into new cages and visually checked daily until a litter was observed. Heterozygous offspring were fostered to B6 mothers which had litters < 2 days old on the day that the RH litters were born (day 1).

Table 6.1: Heterozygote litters produced from breeding two inbred BXD lines, identified by their line numbers. RH litters could not be produced for 97x1 and

92Ax32, as the mothers did not get pregnant or infanticide occurred.

Mother Father Heterozygote litter 1 97 1x97 24 34 24x34 34 24 34x24 55 92A 55x92A 92A 55 92Ax55 55 74 55x74 74 55 74x55 32 92A 32x92A

206

Chapter 6: The BXD lines as a tool to identify parent-of-origin effects with an indirect genetic effect

6.2.2 Developmental measurements

Immediately before fostering on day 1, BXD biological mother, B6 foster mother,

BXD litter and B6 litter weights were measured to the nearest hundredth of a gram. B6 mothers' bodyweight and BXD litter weight was again recorded on post-natal days 6, 10, 14 and 21. Short term weight measurements were also taken as part of the behavioural observations outlined below.

6.2.3 Behavioural observations

Behavioural observations were carried out on days 6, 10 and 14 following the example of Hager and Johnstone (2003, 2007). The weight of foster mothers and offspring was measured to the nearest hundredth of a gram, and then the pups were returned to the home cage and kept warm using a heat mat placed under the cage. The foster mother was kept separate in a new cage for 4 hours with ad libitum food and water. This separation replicates the mother’s normal foraging expeditions and is not believed to influence animal welfare (Hager and

Johnstone, 2007). After 4 hours the mother, pups and food were measured again, and the mother was returned to the home cage. Behavioural observations were taken over the next 15 minutes (Table 6.2). Long lasting behaviours were recorded every 20 seconds, whereas less common behaviours were recorded whenever they occurred (bold and italicised respectively in table 6.2). This set of observations was called time point a.

The animals were then left for a further hour and 45 minutes before being re-weighed and a further 15 minutes of behavioural observations made, and this was called time point b. This allowed us to look at maternal behaviour both after separation (time point a) and without separation (time point b). The change in weight from time point a to time point b represents a measure of maternal

207

Chapter 6: The BXD lines as a tool to identify parent-of-origin effects with an indirect genetic effect provisioning, as we assume that bodyweight lost by the mothers during this period is due to milk being given to the offspring. There is a caveat however that level of urination and defecation has long been recognised as a sign of

‘emotionality’ in rodents (Hall, 1934), with more being produced in novel or stressful environments (Gentsch et al., 1981; Castanon and Mormède, 1994;

Carola et al., 2004; Henderson et al., 2004). Therefore weight changes could, potentially, be due to differences in anxiety levels between the inbred mothers.

Due to the fostering method used, whereby all offspring were reared by foster mothers of a uniform genotype, differences in maternal phenotypes can be attributed to the genotype of the pups.

Table 6.2: Maternal and offspring behaviours recorded. Behaviours in bold (states) were recorded every 20 seconds for 15 minutes and behaviours in italics (events) were recorded every time they occurred.

Maternal care behaviour Self Nestbuilding (gathering nesting material Autogroom (mother grooming herself) and constructing a nest) Nursing (attending the litter, sitting on the Feeding or drinking nest and suckling up to half the litter) Suckling (greater than half the litter being Resting suckled at the same time) Sniffing Other active (moving around the cage) Pup retrieve Licking

208

Chapter 6: The BXD lines as a tool to identify parent-of-origin effects with an indirect genetic effect

6.2.4 Statistical analysis

A univariate general linear model (GLM) in SPSS (version 21, IBM Corporation,

Armonk, NY, USA) was used to calculate the unstandardized residuals, taking into account relevant covariates (outlined with each result).

For each pair of reciprocal heterozygotes, an F-test was used to determine if there was equal variance, and then a T-test was used to investigate if unstandardized residuals were significantly different between RH litters.

209

Chapter 6: The BXD lines as a tool to identify parent-of-origin effects with an indirect genetic effect

6.3 Results

6.3.1 Weight

A GLM was calculated for offspring bodyweight on day 21, using B6 mothers' bodyweight, litter size and proportion of males at day 21 as covariates (Table

6.3). The covariates were used to account for potential confounding effects: larger mothers may be able to provide more for their foster pups, and therefore the pups may be bigger; if there are more pups in a litter, the mother may be able to provide less per pup and therefore pups may be smaller; if there are more males in a litter then males may grow more rapidly than females, and therefore would be larger. With the relatively small sample size available all covariates were kept as some which were close to significance (e.g. B6 mothers' bodyweight in table 6.3) may be having an effect but we may not have the power to detect it at a statistically significant level.

Table 6.3: Tests of between-subjects effects on day 21 weight of RH offspring on

21. The residuals of the model were then used for comparing between genotypes.

Dependent Variable: RH offspring bodyweight day 21 Source Type III Sum df Mean Square F Sig. of Squares Corrected Model 8.477a 3 2.826 2.354 .076 Intercept 25.495 1 25.495 21.238 .000 B6 mothers’ 4.501 1 4.501 3.749 .055 bodyweight day 21 Proportion of males 4.106 1 4.106 3.420 .067 Litter size day 21 .135 1 .135 .112 .738 Error 138.056 115 1.200 Total 8266.988 119 Corrected Total 146.533 118 a. R Squared = .058 (Adjusted R Squared = .033)

The unstandardized residuals of the model were plotted (Figure 6.2). The residuals were then used for T-tests between the reciprocal genotypes (two

210

Chapter 6: The BXD lines as a tool to identify parent-of-origin effects with an indirect genetic effect tailed, unequal variance). For the three genotypes where reciprocals were available we see that there is a significant difference between the 24x34 and

34x24 genotypes (p = 8.45E-10) and the 92Ax55 and 55x92A genotypes (p =

0.016), but not between the 55x74 and 74x55 reciprocal heterozygotes (p =

0.52).

For the other three time points (day 6, 10 and 14) we only have litter averages. Because of the low sample size, the only pair of reciprocal genotypes for which we have three litters in both directions are the 55x92A and 92Ax55 genotypes. Therefore for bodyweight before separation, weight after 4 hours

(time point a) and weight after 6 hours (time point b) a GLM was calculated, with litter size, mothers’ bodyweight and proportion of males as covariates. As above, this was to attempt to correct for any confounding factors which may produce spurious results. An F-test was then carried out on the residuals of the 55x92A and 92Ax55 weights to determine if the variance was equal, and the appropriate

T-test calculated (two tailed, either equal or unequal variance). None of the measures reached significance.

211

Figure 6.2: Weight of RH offspring on post-natal day 21. Weight is the unstandardised residuals from a GLM with B6 mothers' bodyweight, proportion of males and litter size as covariates. For each box the upper and lower hinge corresponds to the first and third quartile. A band represents the median and a diamond the mean. The upper whisker extends from the hinge to the highest value which is within 1.5 times the interquartile range. The lower whisker extends from the hinge to the lowest value within 1.5 times the interquartile range. Outliers are shown as filled circles. Produced using ggplot2 (Wickham, 2009).

212

6.3.2 Behaviour

Again, for behavioural phenotypes we only have values per litter, and therefore a low number of replicates. Similar to our weight measurements, a GLM was carried out for each behaviour, using litter weight, litter size and B6 mothers' bodyweight at the specific time point, proportion of males at day 21 and litter weight at fostering as covariates, and a T-test was carried out between the

55x92A and 92Ax55 reciprocal heterozygotes.

Table 6.4: Phenotypes with a significant difference between the 55x92A and

92Ax55 reciprocal heterozygotes. For each phenotype, a GLM was run with all genotypes, using litter size, litter weight and B6 mothers' bodyweight at that time point, offspring weight at fostering, and proportion of males at day 21 as covariates. An F-test was used to determine the two tailed probability that the variance was not significantly different. Dependent upon this the appropriate T-tests were then carried out on the residuals of this GLM between the reciprocal heterozygotes. The uncorrected mean is the mean number of times what the behaviour was recorded in the three litters of each

RH genotype. Time point a represents behaviour after four hours separation, whereas time point b represents behaviour after two hours undisturbed.

Mothers’ Offspring Mothers’ Mothers’ ‘Self’ rest outside nestbuilding maternal care behaviour the nest day 14 b day 14 a day 14 a day 6 b Mean uncorrected 3.666 38.333 6.666 0.612 55x92A Mean uncorrected 1 27.666 17.333 0.759 92Ax55 Mean residual 2.897 12.737 -12.737 -0.087 55x92A Mean residual 0.04 -0.24 0.24 0.23 92Ax55 F-test 0.713 0.984 0.984 0.863 T-test 0.015 0.031 0.031 0.026

213

Chapter 6: The BXD lines as a tool to identify parent-of-origin effects with an indirect genetic effect

Table 6.4 shows that there is a significant difference in maternal care on day 14 time point a, between the 55x92A and 92Ax55 reciprocal heterozygotes.

As maternal care is a composite measure produced by combining suckling, nursing and nestbuilding, we then examined these behaviours, and found a just sub-significant increase in suckling and nursing in the foster mothers of the

55x92A pups on day 14 time point a (p = 0.056 and 0.058 respectively). There is a negative correlation in the maternal non-pup directed behaviours (‘self’).

Therefore, this shows that there is an indirect genetic effect on maternal care, due to a POE of the pup genotype.

We also see other indicators that the 55x92A pups are able to solicit greater levels of care more generally, as their foster mothers also have increased nestbuilding on day 14 time point b, and the pups spend less time resting outside of the nest (when covariates are taken into account) on day 6, time point b.

214

Chapter 6: The BXD lines as a tool to identify parent-of-origin effects with an indirect genetic effect

6.4 Discussion

The results of this work were constrained by the complex format of the experiment, whereby both directions of the reciprocal cross are needed for meaningful data to be produced, and must be timed with the birth of B6 litters for fostering to be possible. Several of the BXD lines showed high levels of infant death, with pups being either stillborn or culled soon after birth. Therefore, it would be necessary to have larger numbers of within line repeats, to ensure that sufficient mice are available for a complete statistical analysis. However, all litters fostered to B6 females were successfully reared.

We are able to show a POE on offspring bodyweight on day 21.

Therefore, this demonstrates that, as expected, POEs influence development and that these can be found in RHs of BXD lines. Future experiments, using more RHs, will be able to map the loci underlying these changes (Cheverud et al., 2008). These loci may map to the position of previously identified imprinted genes, which would suggest them as candidates, and therefore that the BXD are an appropriate model for examining the effect of that gene. Alternatively, the mapped loci may not contain any known imprinted genes, at which point it would be necessary to identify which gene is showing a parent-of-origin effect, in what tissue and at what developmental stage. This could be done by performing transcriptomic analyses (e.g. microarray or RNAseq based) on tissues from the

RHs (Garfield et al., 2011).

Additionally, we find an IGE on maternal behaviour which shows a POE, and this is exactly what we expect to see, as it shows the pup genotype increasing the amount of care they receive. We might suspect that it is due to a paternally expressed gene, as it causes an increase in maternal care, which presumably causes an increase in maternal investment and therefore an

215

Chapter 6: The BXD lines as a tool to identify parent-of-origin effects with an indirect genetic effect increase in pup fitness. Again, future studies would be able to identify the causative loci for the phenotype, and potentially the specific imprinted gene.

These initial results suggest POEs for some the maternal behaviours, and therefore that the maternal behaviour is being influenced by pup genotype.

However, the results are confounded by maternal genetic effects, for example it can be seen that the offspring bodyweight at fostering of the 55x92A litter is higher than that of the reciprocal 92Ax55 litter, and, although we corrected for this, it is unclear if this is a result of the pups’ genotype (allowing them to grow faster) or maternal genotype (providing them with more pre-natal resources).

Indeed, there appears to be a POE on day 14 weight between the 55x92A and

92Ax55 reciprocal heterozygotes, but this becomes non-significant when offspring weight at fostering is included as a covariate. Therefore, in future work we need to be careful to take into consideration these effects.

This demonstrates the down side of this experimental design: it cannot separate true parent-of-origin effects (i.e. the effect of a maternally or paternally derived gene within the target individual) from pre-natal maternal effects (i.e. the effect of the in utero maternal environment on the target individual). This is due to the fact that for any particular RH genotype, the mother’s genotype will always be the same. However, methods to avoid this problem, such as artificial insemination and embryo transfer (Cowley et al., 1989), may cause disruption of epigenetic markers, such as methylation and imprinting, in both mice and humans (Lim and Ferguson-Smith, 2010; Fowden et al., 2011; Calle et al., 2012;

Kohda and Ishino, 2013). Therefore, these techniques may have their own confounding effect, would be far from trivial to accomplish, and would further increase the time taken and cost of what would already be a large and expensive experiment.

216

Chapter 6: The BXD lines as a tool to identify parent-of-origin effects with an indirect genetic effect

Even if it was demonstrated that the IGE found here were a result of in utero maternal effects, and not due to POEs, it would still be an interesting finding, as it would represent an action of maternal genotype on maternal phenotype, which is mediated by changing the pups’ phenotype.

6.4.1 Conclusion

This work represents a demonstration that the BXD line can be used to show IGE which also show a POE effect on early life development and behaviour.

Therefore, increasing the number of lines used would allow QTL to be mapped for POEs (Cheverud et al., 2008). Once QTL are identified, the genes within the locus can be examined for parent-of-origin specific expression.

217

Chapter 6: The BXD lines as a tool to identify parent-of-origin effects with an indirect genetic effect

6.5 References

Ashbrook, D. G., and Hager, R. (2013). Empirical testing of hypotheses about the evolution of genomic imprinting in mammals. Front. Neuroanat. 7, 6. doi:10.3389/fnana.2013.00006.

Carola, V., D’Olimpio, F., Brunamonti, E., Bevilacqua, A., Renzi, P., and Mangia, F. (2004). Anxiety-related behaviour in C57BL/6 <--> BALB/c chimeric mice. Behav. Brain Res. 150, 25–32. doi:10.1016/S0166-4328(03)00217-1.

Castanon, N., and Mormède, P. (1994). Psychobiogenetics: adapted tools for the study of the coupling between behavioral and neuroendocrine traits of emotional reactivity. Psychoneuroendocrinology 19, 257–282. doi:10.1016/0306-4530(94)90065-5.

Cowley, D. E., Pomp, D., Atchley, W. R., Eisen, E. J., and Hawkins-Brown, D. (1989). The impact of maternal uterine genotype on postnatal growth and adult body size in mice. Genetics 122, 193–203.

Dent, C. L., and Isles, A. R. (2014). Brain-expressed imprinted genes and adult behaviour: the example of Nesp and Grb10. Mamm. Genome 25, 87–93. doi:10.1007/s00335-013-9472-0.

DeVeale, B., van der Kooy, D., and Babak, T. (2012). Critical evaluation of imprinted gene expression by RNA-Seq: a new perspective. PLoS Genet. 8,

218

Chapter 6: The BXD lines as a tool to identify parent-of-origin effects with an indirect genetic effect

e1002600. doi:10.1371/journal.pgen.1002600.

Drake, N. M., Park, Y. J., Shirali, A. S., Cleland, T. A., and Soloway, P. D. (2009). Imprint switch mutations at Rasgrf1 support conflict hypothesis of imprinting and define a growth control mechanism upstream of IGF1. Mamm. Genome 20, 654–663. doi:10.1007/s00335-009-9192-7.

Frost, J. M., and Moore, G. E. (2010). The importance of imprinting in the human placenta. PLoS Genet. 6, e1001015. doi:10.1371/journal.pgen.1001015.

Gentsch, C., Lichtsteiner, M., and Feer, H. (1981). Locomotor activity, defecation score and corticosterone levels during an openfield exposure: a comparison among individually and group-housed rats, and genetically selected rat lines. Physiol. Behav. 27, 183–186. doi:10.1016/0031-9384(81)90320-6.

Gregg, C., Zhang, J., Butler, J. E., Haig, D., and Dulac, C. (2010). Sex-specific parent-of-origin allelic expression in the mouse brain. Science 329, 682–5. doi:10.1126/science.1190831.

Guilmatre, A., and Sharp, A. J. (2012). Parent of origin effects. Clin. Genet. 81, 201–9. doi:10.1111/j.1399-0004.2011.01790.x.

Hager, R., Cheverud, J. M., and Wolf, J. B. (2008). Maternal effects as the cause of parent-of-origin effects that mimic genomic imprinting. Genetics 178, 1755–1762. doi:10.1534/genetics.107.080697.

Hager, R., and Johnstone, R. A. (2003). The genetic basis of family conflict resolution in mice. Nature 421, 533–535. doi:10.1038/nature01239.

219

Chapter 6: The BXD lines as a tool to identify parent-of-origin effects with an indirect genetic effect

Hall, C. S. (1934). Emotional behavior in the rat. I. Defecation and urination as measures of individual differences in emotionality. J. Comp. Psychol. 18, 385–403. doi:10.1037/h0071444.

Hau, A. R., and Hau, J. (2007). Correlation between duration of courtship and litter size in outbred NMRI mice. In Vivo 21, 33–34.

Henderson, N. D., Turri, M. G., DeFries, J. C., and Flint, J. (2004). QTL analysis of multiple behavioral measures of anxiety in mice. Behav. Genet. 34, 267– 293. doi:10.1023/B:BEGE.0000017872.25069.44.

Ideraabdullah, F. Y., Vigneau, S., and Bartolomei, M. S. (2008). Genomic imprinting mechanisms in mammals. Mutat. Res. 647, 77–85. doi:10.1016/j.mrfmmm.2008.08.008.

Keverne, E. B., and Curley, J. P. (2008). Epigenetics, brain evolution and behaviour. Front. Neuroendocr. 29, 398–412. doi:10.1016/j.yfrne.2008.03.001.

Khatib, H. (2007). Is it genomic imprinting or preferential expression? BioEssays 29, 1022–1028. doi:10.1002/bies.20637.

Leamy, L. J., Klingenberg, C. P., Sherratt, E., Wolf, J. B., and Cheverud, J. M. (2008). A search for quantitative trait loci exhibiting imprinting effects on mouse mandible size and shape. Heredity 101, 518–526. doi:10.1038/hdy.2008.79.

Mantey, C., Brockmann, G. a, Kalm, E., and Reinsch, N. (2005). Mapping and exclusion mapping of genomic imprinting effects in mouse F 2 families. J. Hered. 96, 329–338. doi:10.1093/jhered/esi044.

220

Chapter 6: The BXD lines as a tool to identify parent-of-origin effects with an indirect genetic effect

Plagge, A., Gordon, E., Dean, W., Boiani, R., Cinti, S., Peters, J., and Kelsey, G. (2004). The imprinted signaling protein XL alpha s is required for postnatal adaptation to feeding. Nat. Genet. 36, 818–826. doi:10.1038/ng1397.

Schulz, R., Woodfine, K., Menheniott, T. R., Bourc’his, D., Bestor, T., and Oakey, R. J. (2008). WAMIDEX: a web atlas of murine genomic imprinting and differential expression. Epigenetics 3, 89–96.

Wickham, H. (2009). ggplot2: elegant graphics for data analysis. Springer New York.

Wolf, J. B., Hager, R., and Cheverud, J. M. (2008b). Genomic imprinting effects on complex traits: a phenotype-based perspective. Epigenetics 3, 295–299.

221

Chapter 7: General discussion

222

Chapter 7: General discussion

In this thesis I have performed systems-genetics analyses for a number of developmental and behavioural traits, seeking to identify causative loci and potential candidate genes. For this two main approaches were taken: combining data collected in different species to identify candidate genes for phenotypes

(chapters 2 and 3), and using different experimental designs of between strain fostering in mice to investigate indirect genetic effects (IGEs) and the loci causing them (chapters 4-6). In this chapter I will discuss the main findings, some of the limitations of these approaches, and potential areas of future work.

7.1 Main findings

In chapter 2, I use quantitative trait loci (QTL) mapping data for hippocampus weight from the BXD recombinant inbred (RI) mouse panel, to calculate the significance of linkage between all mouse genes and hippocampus weight.

Human genome-wide association study (GWAS) data for hippocampus volume was used to produce a p-value for each gene within the human genome.

Combining these datasets revealed one gene, MGST3, which appeared to influence hippocampus size in both species. A co-expression analysis revealed that MGST3 co-expresses with genes involved in mitochondrial function and oxidative phosphorylation, suggesting an involvement in oxidative stress as a potential mechanism of action. Further, genes which co-express with MGST3 are significantly enriched for genes associated with neurodegenerative disorders, suggesting that MGST3 may also be of interest when studying these diseases.

In chapter 3, I used a similar technique, but instead identified two regions of the genome linked to traits which measure a combination of anxiety, activity and exploration. Human GWAS data for bipolar disorder was then used to explore these regions of the genome, as anxiety, activity and exploration behaviour are all disrupted in bipolar patients. This produced four candidate genes, TNR, CMYA5, MCTP1 and RXRG, which are associated with the traits in

223

Chapter 7: General discussion both species. This allows me to confirm previous associations between bipolar disorder and CMYA5, MCTP1 and RXRG, as well as present TNR as a novel candidate. MCTP1, RXRG and TNR co-express with genes previously linked to psychiatric disorders, and from this I hypothesise that they act via influencing signalling in the striatum. This hypothesis could be directly tested in future work, potentially giving insight into the biology of psychiatric disorders.

Chapter 4 examines how the maternal genome can influence post-natal offspring phenotype, and how the offspring genome can influence post-natal maternal behaviour, using a cross-fostering design. I was able to identify two maternal direct genetic effect (DGE) loci, where maternal genotype influences maternal care and nestbuilding behaviour, and an offspring DGE loci which influences offspring solicitation. I was also able to identify a maternal indirect genetic effect (IGE) loci, where maternal genotype influences offspring growth, and two offspring IGE loci, where offspring genotype influences maternal care and nestbuilding. This shows that IGEs can be an important component of the genetic architecture of complex traits (Bijma and Wade, 2008). Further, I was able to show that offspring solicitation and maternal care show signs of co- adaptation, which may be linked to the costs of increased solicitation on growth found in the study (e.g. needing to expend more energy to increase solicitation).

Chapter 5 uses a similar design to chapter 4, but using half-litter fostering, such that each B6 mother had half a litter of the B6 genotype, and half a litter of a

BXD genotype. I only examined one direction of cross, as in the other direction

(BXD mothers, with half a litter of B6 and half a litter of BXD), the genotype of the genetically variable mother and offspring are correlated, and therefore conclusions would be more ambiguous. In this I was able to not just examine the effect of offspring genotype on maternal phenotype, but also the effect of offspring genotype on their nestmates’ phenotype. I was able to identify three loci

224

Chapter 7: General discussion showing IGEs, one of which appears to have an effect on nestmate phenotype, whereas the other two alter maternal phenotype, which then alters nestmate phenotype. Again, this shows the importance of IGEs in influencing complex traits. Also, the knowledge that offspring are able to influence the phenotype of their mothers and nestmates is important for understanding kin selection and parent-offspring conflict (Hamilton, 1964; Trivers, 1974). As mentioned in chapter

5, the loci I find here are not the same that were found in chapter 4. I speculate on some of the reasons, including the difference in experimental design and the possibility of many loci which are near the detection threshold.

Finally, chapter 6 uses reciprocal heterozygous (RH) crosses to examine parent-of-origin effects (POEs), using a fostering design to break the correlation between maternal genotype and post-natal maternal care. Although this is only an initial study, I am able to show a POE on day 21 post-natal weight and an IGE

POE (i.e. the offspring genotype altering the maternal phenotype in parent-of- origin dependent manner) on maternal care. This, therefore, shows that RH crosses of the BXD lines can be used to examine POEs, and that POE loci can influence the level of maternal care. The next step would be to expand this approach so we are able to map loci for these POEs.

7.2 Knockout, normal variation and disease

Knockout animals are often used to investigate the phenotypes which a gene influences (White et al., 2013; de Angelis et al., 2015), and can provide huge insight, showing what novel genes do in a whole animal system. However a single gene may influence many traits (pleiotropy) (de Angelis et al., 2015), the genetic background of the mouse can exacerbate or mask the gene’s effects

(epistasis) (Sisay et al., 2013) and subtle variations may not be noticed (i.e. a serious physical deficit may mask a behavioural change). This is analogous to

225

Chapter 7: General discussion human disease, where protein coding mutation can have much larger effects than changes in regulatory elements. For example, the gene CACNA1C, in which rare, gain-of-function mutations causes Timothy Syndrome, a multi-organ disorder whose symptoms include potentially lethal cardiac arrhythmias, immune deficiency, cognitive disability, and autism (Splawski et al., 2004), whereas common variation in the regulatory regions are associated with bipolar disorder and schizophrenia, without evidence of associated with, for example, cardiac or immune phenotypes (Smoller et al., 2013; Ripke et al., 2013; McCarroll and

Hyman, 2013; Erk et al., 2014).

In light of the above, studying sequence variation within a population (e.g. the variation found within RI populations) is a necessary complement to the major loss- or gain-of-function mutations found in knockin or knockout animal models, especially when investigating common, polygenetic disorders or variations in normal phenotypes. This is not to discount the huge amount of information that knockin and knockout models have given us, especially in the area of rare, monogenic disorders, e.g. Shahbazian et al. (2002) and Peça et al.

(2011).

In relation to this thesis, we can observe the usefulness of both approaches. Genetic variation influencing a phenotype of interest was used to identify QTL in chapters 2-5. In chapters 2 and 3, I then used natural variation in another population for a similar phenotype to identify candidates underlying the disease in both populations (in this case, populations from two species). In chapters 4 and 5, I identified candidates within my QTL using data from a mutant mouse model database. Both these approaches rely on others having investigated relevant phenotypes, which is less common for behavioural traits

(e.g. maternal care) than for morphological traits (e.g. hippocampus size) or disease traits (e.g. bipolar disorder). This demonstrates the overlap between all

226

Chapter 7: General discussion the chapters, as literature from knockout models were used to support the candidate genes in chapters 2 and 3, whereas identifying IGE underlying maternal behaviour in a different population (either the same or a different species) may help us to identify candidates within the loci detected in chapters 4 and 5.

7.3 Fostering and maternal care

An important caveat to remember in relation to chapters 4-6 is that that fostering itself can alter behaviour in both mothers and offspring. Mothers are able to differentiate between biological or non-biological young (Ostermeyer and Elwood,

1983), and show altered maternal behaviour when rearing foster young (Ressler,

1962; Hager and Johnstone, 2003; van der Veen et al., 2008; Curley et al.,

2010), even if they are of the same strain (van der Veen et al., 2008; Curley et al., 2010; Cox et al., 2013; Lerch et al., 2014). In offspring, fostering can alter behaviour, including both anxiety- and depression-like behaviours (Maxson and

Trattner, 1981; Priebe et al., 2005; Prakash et al., 2006; George et al., 2010; Cox et al., 2010), autoimmune disease (Case et al., 2010) and bodyweight

(Bartolomucci et al., 2004; Ďureje et al., 2011; Matthews et al., 2011).

Therefore some of the QTL I detect may be due to gene-by-environment interactions (i.e. different genotypes responding differently to fostering), rather than ‘pure’ genotype effects. Embryo-transfer (Cowley et al., 1989) could be used to examine this (e.g. a B6 mothers with BXD embryos compared to B6 mothers with fostered BXD pups), but again, embryo-transfer may have its own effects, potentially resulting in their own genotype-by-environment effects (i.e. different offspring genotypes responding differently to the B6 intra uterine environment).

Another potential caveat which needs to be mentioned is that the four hours of separation before behavioural observations could affect behaviour of

227

Chapter 7: General discussion both the mother and offspring (Nylander and Roman, 2013; Nishi et al., 2014).

More frequent separations (i.e. ≥ 3 hours separation, daily) are used to simulate a deprived maternal environment and produces poor outcomes later in life

(Plotsky and Meaney, 1993; Ogawa et al., 1994; Hall et al., 1999), whereas shorter separations (≥ 15 minutes, often called handling) have been shown to increase maternal care and has a positive impact on later life (Weininger, 1954;

Levine, 1957, 1967; Hennessy et al., 1980).

Therefore, again, there is the potential that I am detecting gene-by- environment effects (how different genotypes respond to long, infrequent maternal separation) rather than loci which alter maternal care independent of environment. On the other hand, in a natural environment the mother would leave the nest to go and find food, and, in fact, an environment where the mother is within a few centimetres of the nest during the entire post-natal period, such as normal laboratory condition, may provide its own environmental effects. This brings us to the observation that, for the majority of behaviours, it is not genes alone which produce behaviours, but instead their interactions with the environment (Gomez-Marin et al., 2014). Therefore, the best we can do in an experimental setting is to keep environment constant and remember that different genes may influence the same phenotype in different environments.

228

Chapter 7: General discussion

7.4 Missing heritability

Heritability is defined as the fraction of phenotypic variation explained by genetic variation. Narrow sense heritability (h2), the heritability typically measured in twin studies, is the additive genetic variance as a proportion of the phenotypic variance. Broad sense heritability (H2) also includes other variables, such as dominance and epistasis (Visscher et al., 2008). Gene-by-environment interactions (GxE) and gene-environment correlations can also influence heritability estimates, but are often ignored due to the difficulty in measuring them. Narrow sense heritability is the normally reported value, as individuals only transmit one copy of each gene, and therefore most relatives share either a single or no copies of the allele, and so dominance and other genetic effects which depend on sharing two copies of the allele do not contribute to the phenotypic resemblance (Visscher et al., 2008)

There is clear heritability of many traits in humans, including both normal variation and disease traits. For example the heritability of height is over 0.8

(Silventoinen et al., 2003; Macgregor et al., 2006) and it has been estimated that at least 45% of variance can be accounted for just by SNPs (Yang et al., 2010).

For disease phenotypes, comparisons between monozygotic and dizygotic twins has shown heritability of 0.81 for schizophrenia (Sullivan et al., 2003), 0.75 for bipolar disorder (Smoller and Finn, 2003) and 0.80 for autism spectrum disorders

(Ronald and Hoekstra, 2011).

Since their first, small scale use a decade ago (Klein et al., 2005) to recent efforts using tens of thousands of subjects (Schizophrenia Working Group of the Psychiatric Genomics Consortium, 2014), GWASs have found SNPs associated with many phenotypes (Hindorff et al., 2009; Welter et al., 2014).

However, there has become a disconnect between the amount of heritability

229

Chapter 7: General discussion explained by the SNPs found to be significant, and the level of heritability predicted for traits, and this has been termed ‘missing heritability’ (Maher, 2008;

Manolio et al., 2009). One major cause of missing heritability appears to be that there are many genes influencing any given phenotype, each of which has a very small effect size, and therefore increasingly large cohorts, or new analyses, are needed to identify specific variants (Yang et al., 2010). An assumption of GWASs is that common diseases are caused by common variants. However, it could be that some of the heritability is due to less common, but still not rare, variants, which are not detected by the SNP arrays currently used for GWASs. In this sense, the heritability may not really be missing, just difficult to find.

All of the approaches within this thesis can be used to address the problem of missing heritability. One cause of ‘missing’ heritability is just that there are many loci with very small effect sizes, which are therefore very difficult to detect with traditional association methods, requiring tens, and potentially hundreds, of thousands of individuals to detect these genes of small effect. This is because GWASs typically have modest statistical power due to high corrections needed to compensate for multiple testing. However, loci are defined with very high precision, theoretically down to the level of SNPs. In contrast, mouse linkage studies often have high statistical power to detect genetic effects but lower genetic resolution, producing loci that include tens or hundreds of genes (Mackay et al., 2009; Ackert-Bicknell et al., 2010). Combining data from mice and humans overcomes some of these problems, gaining power from mouse crosses and precision from human GWAS. As I have shown, RI lines provide an opportunity to identify loci, even with small effect sizes, and therefore chapters 2 and 3 may help to address the missing heritably problem, by finding novel genes for phenotypes.

230

Chapter 7: General discussion

It is unsurprising that there are many alleles of small effect, and few alleles of large effect, for common, complex diseases (e.g. psychiatric disorders), as these disorders cause a marked reduction in reproductive fitness (Haverkamp et al., 1982; Bassett et al., 1996; Power et al., 2013), and therefore will be selected against. On the other hand, alleles which have a small effect on phenotypic variation will have a far lower selective pressure because their effect may be compensated for by other alleles, and consequently they can survive in the general population. Therefore, these disorders result from the interactions between large numbers of genes (polygenicity), and interactions with the environment. None of these alleles of small effect can be said to causal, but rather each is a risk factor, increasing an individual’s probability of developing the disorder, but none of these alleles are necessary nor sufficient to cause the disorder alone (McCarroll and Hyman, 2013). For autism is has been estimated that there are 350 and 400 genes which increase risk (Iossifov et al., 2012) and for major depression thousands of genes (Flint and Kendler, 2014). Indeed Flint and Kendler (2014) suggest that as many as one in five genes expressed in the brain may contribute to major depression.

In a more broad sense heritability can be confounded by, for example,

IGEs (particularly maternal effects) and POEs. Both of these can cause an individual to be phenotypically similar to a sibling or parent (two of the traditional methods of estimating heritability) without them sharing the same genotype (e.g. an IGE effecting offspring phenotype due to parental genotype). POEs may confound GWASs, as two individuals could have the same genotype, but a different phenotype due to an identical imprinted allele being inherited maternally in one individual and paternally in the other. Understanding and identifying IGEs and POEs could help to differentiate between ‘real’ heritability (due to genotype), and non-genetic effects.

231

Chapter 7: General discussion

7.5 Psychiatric disorders

Taking psychiatric disorders, mentioned above, as an example, I can show the utility of the approaches demonstrated in this thesis. As in chapters 2 and 3, I can identify genes which are associated with both the disease phenotype in humans and normal variation in linked phenotypes in animals. This implies that the genes

I find are related to specific symptoms of the disorder and that the animal model is appropriate for further investigation. This approach can help us elucidate the link between the gene and the disorder, since it suggests what specific symptoms of the disorder the gene may be influencing. We can then build a

‘parts list’, dividing the genes associated with a particular disorder into groups dependent upon what underlying phenotypes they influence. It is possible, and indeed likely, that individual genes may influence several different phenotypes, so would appear in several groups.

Further, we could specifically look at how the candidate genes interact within specific tissues in the BXD mice and what effects the alleles have on the expression or activity of the particular proteins. This kind of transcriptome and proteome data is not possible to get from human cohorts (as, for example, large amounts of brain tissue cannot be removed from patients), but is possible to produce in animal models.

IGEs are also important to understand in relation to psychiatric disorders.

It is well known that the environment influences the inheritance of many psychiatric disorders, particularly early life events (Heim and Nemeroff, 2001;

Carter et al., 2002; Charney and Manji, 2004; Nemeroff, 2004; Bebbington et al.,

2011; Varese et al., 2012; Van Winkel et al., 2013; De Bellis and Zisk, 2014;

Laurens et al., 2015), including family interactions (Burman et al., 1987;

Goldstein, 1987; Schiffman et al., 2002; Asselmann et al., 2014; Laurens et al.,

232

Chapter 7: General discussion

2015). As I explored in chapters 4-6, this early life environment can be significantly influenced by genes expressed in other individuals, and the direction of action is not always simple to understand. For example, it could be that a gene expressed in offspring decreases the amount of maternal care they receive, and that the reduced maternal care increases the offspring’s susceptibility to psychiatric-like behaviours. This gene, therefore, may appear to be associated with the disorder, but could act by a very different pathway than genes with a direct effect. Identifying IGEs is even more complicated in humans, where many environmental factors will also be contributing to offspring early life and maternal behaviour, so identifying candidate genes in mice may allow these genes to then be examined in human (Lucion and Bortolini, 2014).

POEs and imprinting have been implicated in several common, complex, psychiatric disorders (Crespi, 2008; Kopsida et al., 2011; Wilkins and Úbeda,

2011; Lin et al., 2012), including attention-deficit/hyperactivity disorder (Hawi et al., 2005; Wang, 2012) and schizophrenia (Palmer et al., 2006; Pun et al., 2011).

However, conclusive evidence has been lacking for these complex disorders, unlike for disorders where a single gene or loci have been shown to be causative, such as Prader-Willi or Angelman syndromes (Kalsner and

Chamberlain, 2015). One possible reason for this lack of definitive linkage between POEs and common psychiatric disorders is small sample sizes, due to the necessity of having data from at least duos of parent and offspring, and preferably the full trio of father, mother and offspring (Connolly and Heron, 2015).

As we have seen earlier when discussing missing heritability, tens of thousands of subjects are needed to detect the small effect sizes for many genes associated with psychiatric disorders, whereas parent-of-origin studies have often used just hundreds of patients. Therefore if parent-of-origin related genes only have a small effect size they may simply have been missed in the past

233

Chapter 7: General discussion analyses. This brings us back to my first point: perhaps reciprocal heterozygous crosses of RI lines, as used in chapter 6, could be used to identify loci showing parent-of-origin effects and then these loci could be investigated in human

GWASs, as was done in chapters 1 and 2. As this would narrow the region of interest, the genome-wide multiple-comparisons correction would not be needed, and therefore genes could be identified using smaller sample sizes.

It should be acknowledged, of course, that animals are only a model for human behaviour, especially when dealing with psychiatric-like behaviours. Since their common ancestor, approximately 65-100 million years ago (Emes et al.,

2003), both species have diversified into very different niches and have evolved different repertoires of behaviour. That being said, some behaviours, such as maternal care and anxiety, are found throughout the mammalian order, as all mammals care for their young and must balance risk and reward. Here, we make the assumption that genes which regulated these behaviours in the common ancestor are still influencing the behaviours now, even if the details of the phenotypic manifestations (i.e. the behaviours) have been modified in the interim.

This can give us insight about the pathways and mechanisms of action, potentially suggesting new treatment methods, but must always interpreted in the light of them being models and not exact replicates (van der Staay et al., 2009).

7.6 Collaborative cross

A resource which could be used in the future for the kind of analyses carried out in this thesis are the Collaborative Cross (CC) mouse lines (Threadgill et al.,

2002; Churchill et al., 2004; Morahan et al., 2008; Iraqi et al., 2008; Chesler et al., 2008; Threadgill and Churchill, 2012; Collaborative Cross Consortium.,

2012). The CC are a recombinant inbred mouse panel which are derived from eight strains, five inbred (C57BL/6J, A/J, 129S1/SvImJ, NOD/ShiLtJ, and

234

Chapter 7: General discussion

NZO/H1LuJ) and three wild-type-derived strains (CAST/EiJ, PWK/PhJ, and

WSB/EiJ). This contrasts with the two inbred lines used in classical RI panels, such as the BXD (Threadgill et al., 2011). The aim is to produce ~1000 CC lines, and to use a two-step mapping strategy to produce very small candidate loci

(Churchill et al., 2004).

Using the CC would almost certainly increase the number of loci found for any of the phenotypes examined, as at any loci there are, potentially, eight possible alleles, rather than the two found in the BXD. Indeed, it has been estimated that the BXD capture 16% of the genetic variation in Mus musculus, whereas the CC capture 89% (Roberts et al., 2007). This means that many areas of the BXD genome are identical by descent, making them effectively invisible to

QTL mapping (Threadgill et al., 2011). Further, with the large number of lines, and therefore recombination events in the CC, the loci identified are predicted to be much smaller, and the phenotypic diversity between the lines much greater

(Churchill et al., 2004). Finally, because of how many lines will be available, the number of reciprocal heterozygous crosses (also called recombinant inbred intercrosses; Zou et al., 2005) which could be made is, for all practical purposes, limitless. This means that they would be an invaluable resource for the kinds of analyses of POEs begun in chapter 6. Since the CC panel has been specifically designed to model the complexity of the human genome, it will be invaluable for future studies (Threadgill et al., 2011).

The current, major disadvantage of the CC panel is that it does not yet have the huge phenome which has been developed over the last few decades for the BXD panel, and therefore some of the systems-genetics approaches used in this thesis would not be possible without additional experimental work (e.g. using previously collected transcriptome data, or comparing between phenotypes).

However, many studies are now being carried out with the CC, so this gap is

235

Chapter 7: General discussion rapidly being filled (Ram et al., 2014; Weiser et al., 2014; Zheng et al., 2015;

Chow et al., 2015; Dickson et al., 2015; Morgan and Welsh, 2015; Lorè et al.,

2015).

One specific example of how the CC panel could be used, related to the work presented here, would be to map both DGEs and IGEs for maternal behaviour phenotypes and resultant offspring phenotypes. The advantage of the

CC over the BXD is that more loci would be detected (as there is a far greater range of allelic variation), and that the loci would be narrower (due to more lines, and therefore smaller recombination regions). These loci could then be examined in human GWAS for related phenotypes, for example the phenotypes collected by the Avon Longitudinal Study of Parents and Children (Boyd et al., 2013;

Fraser et al., 2013).

7.7 Future work

In addition to the future experimental work suggested in the results chapters of this thesis and in this chapter, there are several areas which have a broader application to several of the chapters, which act as ‘next steps’ to the work presented here.

One area not explored, but for which RI lines are very useful, is regulation of gene expression and transcriptomics. This is because identical genotypes can be used in different environmental conditions, and the links between gene expression, genotype and environment can be explored. As mentioned in section 7.2, many loci found to be associated with complex traits are in regulatory regions, not protein coding regions. Using RI mouse lines we can measure gene expression in many individuals, in different brain regions and at different times, which can help to show which variants alter expression of which genes, and when and where these are active. Indeed this information was

236

Chapter 7: General discussion used in chapters 2 and 3. However, a next step for chapters 4-6 would be to perform transcriptome analysis (e.g. RNA-seq) of brain regions of interest. In the case of IGEs this would allow us to map specific genes, the expression of which are altered by IGEs, and further to be able to map IGE eQTL, showing exactly what loci in the interacting individual is altering the expression of a gene in the focal individual.

Similarly, the epigenome of the mice could be mapped, e.g. DNA methylation (Pelizzola and Ecker, 2011), to determine the proximal biological mechanisms causing the IGE eQTL identified by the transcriptomics methods, and again these epigenetic signals could be mapped onto the genotype of the interacting individual. Using this combination of approaches we could begin to build a full systems-genetics analysis, showing a gene expressed in one individual causing a change in that individual’s phenotype, changing DNA methylation in the second individual, and therefore gene expression, and finally altering the second individual’s phenotype.

The sample sizes for the phenotypic analyses in this thesis are relatively small (around 40 genotypes), which may have reduced the power to detect loci.

However, the advantage of RI lines is that the genotypes are stable and therefore more data can be added. A small overlap in lines could be used so that a model could be produced, taking into account the different batches that the data were collected in. Increasing sample size would firstly, increase the power for detecting loci, and secondly, would allow the exploration of epistasis. Epistasis has been strongly suggested in regulating crucial maternal behaviours such as milk ejection (Góes et al., 2012) and nestbuilding (Sauce et al., 2012).

237

Chapter 7: General discussion

7.8 Conclusion

This thesis has made several novel discoveries. Firstly, I have discovered a novel association between MGST3 and hippocampus size in mouse and human, finding evidence that this is due to changes in oxidative stress and that it may be linked to neurodegenerative disorders. Secondly, I have shown associations between TNR, CMYA5, MCTP1 and RXRG, and anxiety in mice and bipolar disorder in humans. The association between TNR and bipolar disorder is novel, and I suggest that the action of TNR, MCTP1 and RXRG is at least partially mediated by disruption of signalling in the striatum. In chapters 4 and 5, I find novel loci underlying maternal and offspring behaviours, as well as IGEs of mothers on offspring, offspring on mothers and nestmates on nestmates. These loci can now be investigated further, giving insight into the biology of behaviour, and the evolution of family interactions. Finally, in chapter 6 I demonstrate that reciprocal heterozygous offspring of BXD mice can be used to find POEs, and that, again, the BXD lines provide a system in which these effects can be investigated.

In conclusion, my results show the necessity of integrating data from different sources, even across species, and how this can provide novel insight which would not be available from an experiment in isolation.

238

Chapter 7: General discussion

7.9 References

De Angelis, M. H., Nicholson, G., Selloum, M., White, J. K., Morgan, H., Ramirez- Solis, R., Sorg, T., Wells, S., Fuchs, H., Fray, M., et al. (2015). Analysis of mammalian gene function through broad-based phenotypic screens across a consortium of mouse clinics. Nat. Genet. 47, 969–978. doi:10.1038/ng.3360.

Asselmann, E., Wittchen, H.-U., Lieb, R., and Beesdo-Baum, K. (2014). The role of the mother-child relationship for anxiety disorders and depression: results from a prospective-longitudinal study in adolescents and their mothers. Eur. Child Adolesc. Psychiatry. doi:10.1007/s00787-014-0596-x.

Bartolomucci, A., Gioiosa, L., Chirieleison, A., Ceresini, G., Parmigiani, S., and Palanza, P. (2004). Cross fostering in mice: behavioral and physiological carry-over effects in adulthood. Genes, Brain Behav. 3, 115–122. doi:10.1111/j.1601-183X.2003.00059.x.

Bassett, A. S., Bury, A., Hodgkinson, K. A., and Honer, W. G. (1996). Reproductive fitness in familial schizophrenia. Schizophr. Res. 21, 151–160. doi:10.1016/0920-9964(96)00018-7.

Bebbington, P., Jonas, S., Kuipers, E., King, M., Cooper, C., Brugha, T., Meltzer, H., McManus, S., and Jenkins, R. (2011). Childhood sexual abuse and psychosis: Data from a cross-sectional national psychiatric survey in England. Br. J. Psychiatry 199, 29–37. doi:10.1192/bjp.bp.110.083642.

De Bellis, M. D., and Zisk, A. (2014). The biological effects of childhood trauma. Child Adolesc. Psychiatr. Clin. N. Am. 23, 185–222. doi:10.1016/j.chc.2014.01.002.

Boyd, A., Golding, J., Macleod, J., Lawlor, D. A., Fraser, A., Henderson, J., Molloy, L., Ness, A., Ring, S., and Davey Smith, G. (2013). Cohort Profile: the “children of the 90s”--the index offspring of the Avon Longitudinal Study of Parents and Children. Int. J. Epidemiol. 42, 111–127. doi:10.1093/ije/dys064.

Burman, B., Mednick, S. A., Machón, R. A., Parnas, J., and et al (1987). Children at high risk for schizophrenia: Parent and offspring perceptions of family relationships. J. Abnorm. Psychol. 96, 364–366. doi:10.1037/0021- 843X.96.4.364.

239

Chapter 7: General discussion

Carter, J. W., Schulsinger, F., Parnas, J., Cannon, T., and Mednick, S. A. (2002). A multivariate prediction model of schizophrenia. Schizophr. Bull. 28, 649– 682.

Case, L. K., del Rio, R., Bonney, E. A., Zachary, J. F., Blankenhorn, E. P., Tung, K. S. K., and Teuscher, C. (2010). The postnatal maternal environment affects autoimmune disease susceptibility in A/J mice. Cell. Immunol. 260, 119–127. doi:10.1016/j.cellimm.2009.10.002.

Charney, D. S., and Manji, H. K. (2004). Life stress, genes, and depression: multiple pathways lead to increased risk and new opportunities for intervention. Sci. STKE 2004, re5. doi:10.1126/stke.2252004re5.

Chesler, E. J., Miller, D. R., Branstetter, L. R., Galloway, L. D., Jackson, B. L., Philip, V. M., Voy, B. H., Culiat, C. T., Threadgill, D. W., Williams, R. W., et al. (2008). The Collaborative Cross at Oak Ridge National Laboratory: developing a powerful resource for systems genetics. Mamm. Genome 19, 382–389. doi:10.1007/s00335-008-9135-8.

Chow, C. Y., Wang, X., Riccardi, D., Wolfner, M. F., and Clark, A. G. (2015). The genetic architecture of the genome-wide transcriptional response to ER stress in the mouse. PLoS Genet. 11, e1004924. doi:10.1371/journal.pgen.1004924.

Churchill, G. a, Airey, D. C., Allayee, H., Angel, J. M., Attie, A. D., Beatty, J., Beavis, W. D., Belknap, J. K., Bennett, B., Berrettini, W., et al. (2004). The Collaborative Cross, a community resource for the genetic analysis of complex traits. Nat. Genet. 36, 1133–1137. doi:10.1038/ng1104-1133.

Collaborative Cross Consortium. (2012). The genome architecture of the Collaborative Cross mouse genetic reference population. Genetics 190, 389–401. doi:10.1534/genetics.111.132639.

Connolly, S., and Heron, E. A. (2015). Review of statistical methodologies for the detection of parent-of-origin effects in family trio genome-wide association data with binary disease traits. Brief. Bioinform. 16, 429–448. doi:10.1093/bib/bbu017.

Cowley, D. E., Pomp, D., Atchley, W. R., Eisen, E. J., and Hawkins-Brown, D. (1989). The impact of maternal uterine genotype on postnatal growth and adult body size in mice. Genetics 122, 193–203.

Cox, K. H., Gatewood, J. D., Howeth, C., and Rissman, E. F. (2010). Gestational exposure to bisphenol A and cross-fostering affect behaviors in juvenile mice. Horm. Behav. 58, 754–761. doi:10.1016/j.yhbeh.2010.07.008.

Crespi, B. (2008). Genomic imprinting in the development and evolution of psychotic spectrum conditions. Biol. Rev. 83, 441–493. doi:10.1111/j.1469- 185X.2008.00050.x.

240

Chapter 7: General discussion

Dickson, P. E., McNaughton, K. A., Hou, L., Anderson, L. C., Long, K. H., and Chesler, E. J. (2015). Sex and strain influence attribution of incentive salience to reward cues in mice. Behav. Brain Res. 292, 305–315. doi:10.1016/j.bbr.2015.05.039.

Ďureje, L., Bímová, B. V., and Piálek, J. (2011). No postnatal maternal effect on male aggressiveness in wild-derived strains of house mice. Aggress. Behav. 37, 48–55. doi:10.1002/ab.20371.

Emes, R. D., Goodstadt, L., Winter, E. E., and Ponting, C. P. (2003). Comparison of the genomes of human and mouse lays the foundation of genome zoology. Hum. Mol. Genet. 12, 701–709. doi:10.1093/hmg/ddg078.

Erk, S., Meyer-Lindenberg, A., Linden, D. E. J., Lancaster, T., Mohnke, S., Grimm, O., Degenhardt, F., Holmans, P., Pocklington, A., Schmierer, P., et al. (2014). Replication of brain function effects of a genome-wide supported psychiatric risk variant in the CACNA1C gene and new multi-locus effects. Neuroimage 94, 147–154. doi:10.1016/j.neuroimage.2014.03.007.

Flint, J., and Kendler, K. S. (2014). The genetics of major depression. Neuron 81, 484–503. doi:10.1016/j.neuron.2014.01.027.

Fraser, A., Macdonald-Wallis, C., Tilling, K., Boyd, A., Golding, J., Davey Smith, G., Henderson, J., Macleod, J., Molloy, L., Ness, A., et al. (2013). Cohort Profile: the Avon Longitudinal Study of Parents and Children: ALSPAC mothers cohort. Int. J. Epidemiol. 42, 97–110. doi:10.1093/ije/dys066.

George, E. D., Bordner, K. A., Elwafi, H. M., and Simen, A. A. (2010). Maternal separation with early weaning: a novel mouse model of early life neglect. BMC Neurosci. 11, 123. doi:10.1186/1471-2202-11-123.

Góes, C. P., Sauce, B., and Peripato, A. C. (2012). Milk ejection in mice LG/J x SM/J. Mamm. Genome 23, 770–779. doi:10.1007/s00335-012-9431-1.

Goldstein, M. J. (1987). The UCLA High-risk Project. Schizophr. Bull. 13, 505– 514. doi:10.1093/schbul/13.3.505.

Hager, R., and Johnstone, R. A. (2003). The genetic basis of family conflict resolution in mice. Nature 421, 533–535. doi:10.1038/nature01239.

Hall, F. S., Wilkinson, L. S., Humby, T., and Robbins, T. W. (1999). Maternal deprivation of neonatal rats produces enduring changes in dopamine function. Synapse 32, 37–43. doi:10.1002/(SICI)1098- 2396(199904)32:1<37::AID-SYN5>3.0.CO;2-4.

241

Chapter 7: General discussion

Hamilton, W. D. (1964). The genetical evolution of social behaviour. I. J. Theor. Biol. 7, 1–16. doi:10.1016/0022-5193(64)90038-4.

Haverkamp, F., Propping, P., and Hilger, T. (1982). Is there an increase of reproductive rates in schizophrenics? I. Critical review of the literature. Arch. Psychiatr. Nervenkr. 232, 439–450. doi:10.1007/BF00345599.

Hawi, Z., Segurado, R., Conroy, J., Sheehan, K., Lowe, N., Kirley, A., Shields, D., Fitzgerald, M., Gallagher, L., and Gill, M. (2005). Preferential transmission of paternal alleles at risk genes in attention-deficit/hyperactivity disorder. Am. J. Hum. Genet. 77, 958–965. doi:10.1086/498174.

Heim, C., and Nemeroff, C. B. (2001). The role of childhood trauma in the neurobiology of mood and anxiety disorders: preclinical and clinical studies. Biol. Psychiatry 49, 1023–1039. doi:10.1016/S0006-3223(01)01157-X.

Hindorff, L. A., Sethupathy, P., Junkins, H. A., Ramos, E. M., Mehta, J. P., Collins, F. S., and Manolio, T. A. (2009). Potential etiologic and functional implications of genome-wide association loci for human diseases and traits. Proc. Natl. Acad. Sci. U. S. A. 106, 9362–9367. doi:10.1073/pnas.0903103106.

Iossifov, I., Ronemus, M., Levy, D., Wang, Z., Hakker, I., Rosenbaum, J., Yamrom, B., Lee, Y. ha, Narzisi, G., Leotta, A., et al. (2012). De novo gene disruptions in children on the autistic spectrum. Neuron 74, 285–299. doi:10.1016/j.neuron.2012.04.009.

Iraqi, F. A., Churchill, G., and Mott, R. (2008). The Collaborative Cross, developing a resource for mammalian systems genetics: A status report of the Wellcome Trust cohort. Mamm. Genome 19, 379–381. doi:10.1007/s00335-008-9113-1.

Kalsner, L., and Chamberlain, S. J. (2015). Prader-Willi, Angelman, and 15q11- q13 Duplication syndromes. Pediatr. Clin. North Am. 62, 587–606. doi:10.1016/j.pcl.2015.03.004.

Klein, R. J., Zeiss, C., Chew, E. Y., Tsai, J.-Y., Sackler, R. S., Haynes, C., Henning, A. K., SanGiovanni, J. P., Mane, S. M., Mayne, S. T., et al. (2005). Complement factor H polymorphism in age-related macular degeneration. Science 308, 385–389. doi:10.1126/science.1109557.

Laurens, K. R., Luo, L., Matheson, S. L., Carr, V. J., Raudino, A., Harris, F., and Green, M. J. (2015). Common or distinct pathways to psychosis? A systematic review of evidence from prospective studies for developmental

242

Chapter 7: General discussion

risk factors and antecedents of the schizophrenia spectrum disorders and affective psychoses. BMC Psychiatry 15, 205. doi:10.1186/s12888-015- 0562-2.

Lerch, S., Brandwein, C., Dormann, C., Gass, P., and Chourbaji, S. (2014). What makes a good mother? Implication of inter-, and intrastrain strain “cross fostering” for emotional changes in mouse offspring. Behav. Brain Res. 274, 270–281. doi:10.1016/j.bbr.2014.08.021.

Levine, S. (1957). Infantile experience and resistance to physiological stress. Science 126, 405. doi:10.1126/science.126.3270.405.

Levine, S. (1967). Maternal and environmental influences on the adrenocortical response to stress in weanling rats. Science 156, 258–260. doi:10.1126/science.156.3772.258.

Lin, M., Hrabovsky, A., Pedrosa, E., Wang, T., Zheng, D., and Lachman, H. M. (2012). Allele-biased expression in differentiating human neurons: implications for neuropsychiatric disorders. PLoS One 7, e44017. doi:10.1371/journal.pone.0044017.

Lorè, N. I., Iraqi, F. A., and Bragonzi, A. (2015). Host genetic diversity influences the severity of Pseudomonas aeruginosa pneumonia in the Collaborative Cross mice. BMC Genet. 16, 106. doi:10.1186/s12863-015-0260-6.

Lucion, A. B., and Bortolini, M. C. (2014). Mother-pup interactions: Rodents and humans. Front. Endocrinol. 5. doi:10.3389/fendo.2014.00017.

Macgregor, S., Cornes, B. K., Martin, N. G., and Visscher, P. M. (2006). Bias, precision and heritability of self-reported and clinically measured height in Australian twins. Hum. Genet. 120, 571–580. doi:10.1007/s00439-006- 0240-z.

Mackay, T. F. C., Stone, E. A., and Ayroles, J. F. (2009). The genetics of quantitative traits: challenges and prospects. Nat. Rev. Genet. 10, 565–577. doi:10.1038/nrg2612.

Maher, B. (2008). Personal genomes: The case of the missing heritability. Nature 456, 18–21. doi:10.1038/456018a.

Manolio, T. A., Collins, F. S., Cox, N. J., Goldstein, D. B., Hindorff, L. A., Hunter, D. J., McCarthy, M. I., Ramos, E. M., Cardon, L. R., Chakravarti, A., et al. (2009). Finding the missing heritability of complex diseases. Nature 461, 747–753. doi:10.1038/nature08494.

Matthews, P. A., Samuelsson, A.-M., Seed, P., Pombo, J., Oben, J. A., Poston, L., and Taylor, P. D. (2011). Fostering in mice induces cardiovascular and metabolic dysfunction in adulthood. J. Physiol. 589, 3969–3981. doi:10.1113/jphysiol.2011.212324.

Maxson, S. C., and Trattner, A. (1981). Interaction of genotype and fostering in the development of behavior of DBA and C57 mice. Behav. Genet. 11, 153– 165. doi:10.1007/BF01065626.

243

Chapter 7: General discussion

McCarroll, S. a, and Hyman, S. E. (2013). Progress in the genetics of polygenic brain disorders: significant new challenges for neurobiology. Neuron 80, 578–587. doi:10.1016/j.neuron.2013.10.046.

Morahan, G., Balmer, L., and Monley, D. (2008). Establishment of “The Gene Mine”: A resource for rapid identification of complex trait genes. Mamm. Genome 19, 390–393. doi:10.1007/s00335-008-9134-9.

Morgan, A. P., and Welsh, C. E. (2015). Informatics resources for the Collaborative Cross and related mouse populations. Mamm. Genome. doi:10.1007/s00335-015-9581-z.

Nemeroff, C. B. (2004). Neurobiological consequences of childhood trauma. in Journal of Clinical Psychiatry, 18–28.

Nishi, M., Horii-Hayashi, N., and Sasagawa, T. (2014). Effects of early life adverse experiences on the brain: implications from maternal separation models in rodents. Front. Neurosci. 8. doi:10.3389/fnins.2014.00166.

Nylander, I., and Roman, E. (2013). Is the rodent maternal separation model a valid and effective model for studies on the early-life impact on ethanol consumption? Psychopharmacology 229, 555–569. doi:10.1007/s00213- 013-3217-3.

Ogawa, T., Mikuni, M., Kuroda, Y., Muneoka, K., Mori, K. J., and Takahashi, K. (1994). Periodic maternal deprivation alters stress response in adult offspring: Potentiates the negative feedback regulation of restraint stress- induced adrenocortical response and reduces the frequencies of open field- induced behaviors. Pharmacol. Biochem. Behav. 49, 961–967. doi:10.1016/0091-3057(94)90250-X.

Ostermeyer, M. C., and Elwood, R. W. (1983). Pup recognition in Mus musculus: Parental discrimination between their own and alien young. Dev. Psychobiol. 16, 75–82. doi:10.1002/dev.420160202.

Palmer, C. G. S., Hsieh, H.-J., Reed, E. F., Lonnqvist, J., Peltonen, L., Woodward, J. A., and Sinsheimer, J. S. (2006). HLA-B maternal-fetal genotype matching increases risk of schizophrenia. Am. J. Hum. Genet. 79, 710–715. doi:10.1086/507829.

Peça, J., Feliciano, C., Ting, J. T., Wang, W., Wells, M. F., Venkatraman, T. N., Lascola, C. D., Fu, Z., and Feng, G. (2011). Shank3 mutant mice display autistic-like behaviours and striatal dysfunction. Nature 472, 437–442. doi:10.1038/nature09965.

Pelizzola, M., and Ecker, J. R. (2011). The DNA methylome. FEBS Lett. 585, 1994–2000. doi:10.1016/j.febslet.2010.10.061.

Plotsky, P. M., and Meaney, M. J. (1993). Early, postnatal experience alters hypothalamic corticotropin-releasing factor (CRF) mRNA, median eminence CRF content and stress-induced release in adult rats. Mol. Brain Res. 18, 195–200. doi:10.1016/0169-328X(93)90189-V.

244

Chapter 7: General discussion

Power, R. A., Kyaga, S., Uher, R., MacCabe, J. H., Långström, N., Landen, M., McGuffin, P., Lewis, C. M., Lichtenstein, P., and Svensson, A. C. (2013). Fecundity of patients with schizophrenia, autism, bipolar disorder, depression, anorexia nervosa, or substance abuse vs their unaffected siblings. JAMA psychiatry 70, 22–30. doi:10.1001/jamapsychiatry.2013.268.

Pun, F. W., Zhao, C., Lo, W.-S., Ng, S.-K., Tsang, S.-Y., Nimgaonkar, V., Chung, W. S., Ungvari, G. S., and Xue, H. (2011). Imprinting in the schizophrenia candidate gene GABRB2 encoding GABA(A) receptor β(2) subunit. Mol. Psychiatry 16, 557–68. doi:10.1038/mp.2010.47.

Ram, R., Mehta, M., Balmer, L., Gatti, D. M., and Morahan, G. (2014). Rapid identification of major-effect genes using the collaborative cross. Genetics 198, 75–86. doi:10.1534/genetics.114.163014.

Ressler, R. H. (1962). Parental handling in two strains of mice reared by foster parents. Science 137, 129–130. doi:10.1126/science.137.3524.129.

Ripke, S., O’Dushlaine, C., Chambert, K., Moran, J. L., Kähler, A. K., Akterin, S., Bergen, S. E., Collins, A. L., Crowley, J. J., Fromer, M., et al. (2013). Genome-wide association analysis identifies 13 new risk loci for schizophrenia. Nat. Genet. 45, 1150–9. doi:10.1038/ng.2742.

Roberts, A., Pardo-Manuel de Villena, F., Wang, W., McMillan, L., and Threadgill, D. W. (2007). The polymorphism architecture of mouse genetic resources elucidated using genome-wide resequencing data: implications for QTL discovery and systems genetics. Mamm. Genome 18, 473–481. doi:10.1007/s00335-007-9045-1.

Ronald, A., and Hoekstra, R. A. (2011). Autism spectrum disorders and autistic traits: A decade of new twin studies. Am. J. Med. Genet. Part B Neuropsychiatr. Genet. 156, 255–274. doi:10.1002/ajmg.b.31159.

Sauce, B., de Brito, R. A., and Peripato, A. C. (2012). Genetic architecture of nest building in mice LG/J x SM/J. Front. Genet. 3. doi:10.3389/fgene.2012.00090.

Schiffman, J., LaBrie, J., Carter, J., Cannon, T., Schulsinger, F., Parnas, J., and Mednick, S. (2002). Perception of parent–child relationships in high-risk families, and adult schizophrenia outcome of offspring. J. Psychiatr. Res. 36, 41–47. doi:10.1016/S0022-3956(01)00046-2.

245

Chapter 7: General discussion

Schizophrenia Working Group of the Psychiatric Genomics Consortium (2014). Biological insights from 108 schizophrenia-associated genetic loci. Nature 511, 421–427. doi:10.1038/nature13595.

Shahbazian, M. D., Young, J. I., Yuva-Paylor, L. A., Spencer, C. M., Antalffy, B. A., Noebels, J. L., Armstrong, D. L., Paylor, R., and Zoghbi, H. Y. (2002). Mice with truncated MeCP2 recapitulate many Rett syndrome features and display hyperacetylation of histone H3. Neuron 35, 243–254. doi:10.1016/S0896-6273(02)00768-7.

Silventoinen, K., Sammalisto, S., Perola, M., Boomsma, D. I., Cornes, B. K., Davis, C., Dunkel, L., De Lange, M., Harris, J. R., Hjelmborg, J. V. B., et al. (2003). Heritability of adult body height: a comparative study of twin cohorts in eight countries. Twin Res. 6, 399–408. doi:10.1375/twin.6.5.399.

Sisay, S., Pryce, G., Jackson, S. J., Tanner, C., Ross, R. A., Michael, G. J., Selwood, D. L., Giovannoni, G., and Baker, D. (2013). Genetic background can result in a marked or minimal effect of gene knockout (GPR55 and CB2 receptor) in experimental autoimmune encephalomyelitis models of multiple sclerosis. PLoS One 8, e76907. doi:10.1371/journal.pone.0076907.

Smoller, J. W., Craddock, N., Kendler, K., Lee, P. H., Neale, B. M., Nurnberger, J. I., Ripke, S., Santangelo, S., and Sullivan, P. F. (2013). Identification of risk loci with shared effects on five major psychiatric disorders: a genome- wide analysis. Lancet 381, 1371–1379. doi:10.1016/S0140-6736(12)62129- 1.

Smoller, J. W., and Finn, C. T. (2003). Family, twin, and adoption studies of bipolar disorder. Am. J. Med. Genet. C. Semin. Med. Genet. 123C, 48–58. doi:10.1007/s11920-002-0046-1.

Splawski, I., Timothy, K. W., Sharpe, L. M., Decher, N., Kumar, P., Bloise, R., Napolitano, C., Schwartz, P. J., Joseph, R. M., Condouris, K., et al. (2004). CaV1.2 calcium channel dysfunction causes a multisystem disorder including arrhythmia and autism. Cell 119, 19–31. doi:10.1016/j.cell.2004.09.011. van der Staay, F. J., Arndt, S. S., and Nordquist, R. E. (2009). Evaluation of animal models of neurobehavioral disorders. Behav. Brain Funct. 5, 11. doi:10.1186/1744-9081-5-11.

Sullivan, P. F., Kendler, K. S., and Neale, M. C. (2003). Schizophrenia as a complex trait: evidence from a meta-analysis of twin studies. Arch. Gen. Psychiatry 60, 1187–1192. doi:10.1001/archpsyc.60.12.1187.

Threadgill, D. W., and Churchill, G. A. (2012). Ten years of the collaborative cross. Genetics 190, 291–294. doi:10.1534/genetics.111.138032.

Threadgill, D. W., Hunter, K. W., and Williams, R. W. (2002). Genetic dissection of complex and quantitative traits: from fantasy to reality via a community effort. Mamm. Genome 13, 175–178. doi:10.1007/s00335-001-4001-y.

246

Chapter 7: General discussion

Threadgill, D. W., Miller, D. R., Churchill, G. A., and de Villena, F. P.-M. (2011). The collaborative cross: a recombinant inbred mouse population for the systems genetic era. ILAR J. 52, 24–31. doi:10.1093/ilar.52.1.24.

Trivers, R. L. (1974). Parent-offspring conflict. Am. Zool. 14, 249–264. doi:10.1093/icb/14.1.249.

Varese, F., Smeets, F., Drukker, M., Lieverse, R., Lataster, T., Viechtbauer, W., Read, J., van Os, J., and Bentall, R. P. (2012). Childhood adversities increase the risk of psychosis: A meta-analysis of patient-control, prospective- and cross-sectional cohort studies. Schizophr. Bull. 38, 661– 671. doi:10.1093/schbul/sbs050.

Visscher, P. M., Hill, W. G., and Wray, N. R. (2008). Heritability in the genomics era — concepts and misconceptions. Nat. Rev. Genet. 9, 255–266. doi:10.1038/nrg2322.

Wang, K.-S. (2012). Parent-of-origin effects of FAS and PDLIM1 in attention- deficit/hyperactivity disorder. J. Psychiatry Neurosci. 37, 46–52. doi:10.1503/jpn.100173.

Weininger, O. (1954). Physiological damage under emotional stress as a function of early experience. Science 119, 285–286. doi:10.1126/science.119.3087.285.

Weiser, M., Mukherjee, S., and Furey, T. S. (2014). Novel distal eQTL analysis demonstrates effect of population genetic architecture on detecting and interpreting associations. Genetics 198, 879–893. doi:10.1534/genetics.114.167791.

Welter, D., MacArthur, J., Morales, J., Burdett, T., Hall, P., Junkins, H., Klemm, A., Flicek, P., Manolio, T., Hindorff, L., et al. (2014). The NHGRI GWAS Catalog, a curated resource of SNP-trait associations. Nucleic Acids Res. 42, D1001–D1006. doi:10.1093/nar/gkt1229.

White, J. K., Gerdin, A., Karp, N. A., Ryder, E., Buljan, M., Bussell, J. N., Salisbury, J., Clare, S., Ingham, N. J., Podrini, C., et al. (2013). Genome- wide generation and systematic phenotyping of knockout mice reveals new roles for many genes. Cell 154, 452–464. doi:10.1016/j.cell.2013.06.022.

Wilkins, J. F., and Úbeda, F. (2011). Diseases associated with genomic imprinting. Prog. Mol. Biol. Transl. Sci. 101, 401–45. doi:10.1016/B978-0- 12-387685-0.00013-5.

Van Winkel, R., Van Nierop, M., Myin-Germeys, I., and Van Os, J. (2013). Childhood trauma as a cause of psychosis: Linking genes, psychology, and biology. Can. J. Psychiatry 58, 44–51.

247

Chapter 7: General discussion

Yang, J., Benyamin, B., McEvoy, B. P., Gordon, S., Henders, A. K., and others (2010). Common SNPs explain a large proportion of the heritability for human height. Nat. Genet. 42, 565–569. doi:10.1038/ng.608.Common.

Zheng, C. L., Wilmot, B., Walter, N., Oberbeck, D., Kawane, S., Searles, R. P., McWeeney, S. K., and Hitzemann, R. (2015). Splicing landscape of the eight collaborative cross founder strains. BMC Genomics 16, 52. doi:10.1186/s12864-015-1267-0.

Zou, F., Gelfond, J. A. L., Airey, D. C., Lu, L., Manly, K. F., Williams, R. W., and Threadgill, D. W. (2005). Quantitative trait locus analysis using recombinant inbred intercrosses: Theoretical and empirical considerations. Genetics 170, 1299–1311. doi:10.1534/genetics.104.035709.

248

Appendix 1: Supplementary materials for chapter 2

249

Appendix 1: Supplementary materials for chapter 2

Some of the supplementary tables are too large to fit onto an A4 page format while still being readable. Such tables are contained within the attached DVD and online. Descriptions of the tables are contained here.

A1.1 Supplementary tables

Supplementary table 2.1: The human homologue gene symbols for all the mouse genes with a genome-wide p-value ≤ 0.05 for hippocampus weight in BXD, showing their unadjusted p-value in human. Also available on the attached DVD or online at: https://www.dropbox.com/s/kl8lryiz3hccuct/Supplementary_table_2.1.docx?dl=0

Gene Mouse p-value Human p-value MGST3 0.013 0.00095 ALDH9A1 0.013 0.00299 LRRC52 0.013 0.018 TMCO1 0.013 0.029 OLFML2B 0.045 0.058 RXRG 0.013 0.146 ANKRD45 0.024 0.196 RFWD2 0.024 0.201 LMX1A 0.013 0.253 TNFSF18 0.024 0.306 PAPPA2 0.024 0.334 FASLG 0.024 0.369 TNN 0.024 0.421 TNFSF4 0.024 0.44 ASTN1 0.03 0.453 PRDX6 0.024 0.469 UHMK1 0.016 0.469 PIGC 0.024 0.481 RGS4 0.006 0.484 KLHL20 0.024 0.504 UAP1 0.013 0.504 TNR 0.024 0.535 SH2D1B 0.02 0.582 ZBTB37 0.024 0.626 CENPL 0.024 0.643 DARS2 0.024 0.643 RABGAP1L 0.024 0.643 C1ORF111 0.021 0.647 SERPINC1 0.024 0.697 C1ORF9 0.024 0.698 RC3H1 0.024 0.731 UCK2 0.013 0.737 MRPS14 0.024 0.749 RGS5 0.005 0.793 FAM78B 0.013 0.804 CACYBP 0.024 0.815 PBX1 0.009 0.816 NUF2 0.005 0.835 DDR2 0.009 0.913 HSD17B7 0.009 0.934 NOS1AP 0.021 0.955 C1ORF110 0.008 0.979

250

Appendix 1: Supplementary materials for chapter 2

Supplementary table 2.2: Genes which commonly co-express with MGST3 as determined by GeneFriends. A table showing the 8134 genes which are commonly co-expressed with MGST3, independent of tissue or treatment, as identified by GeneFriends. For each gene its Entrez gene ID, gene symbol and co-expression value with MGST3 are shown. Available on the attached DVD or online at: https://www.dropbox.com/s/7k66bxb3gggzjm7/Supplementary_table_2.2.xlsx?dl

251

Supplementary table 2.3: KEGG pathway annotations significantly enriched

(calculated by DAVID) in genes that are significantly co-expressed with

MGST3 (calculated by GeneFriends). The table shows the annotation, the number of submitted genes in the annotation / total number of submitted genes in the category, the fold enrichment and the false discovery rate (FDR). Also available on the attached DVD or online at: https://www.dropbox.com/s/pz771gwplpho1ng/Supplementary_table_2.3.docx?dl

Number of submitted genes in the Fold Term annotation / total FDR Enrichment number of submitted genes in the category hsa04142:Lysosome 95/117 1.427 6.86E-05 hsa04510:Focal adhesion 151/201 1.320 9.45E-05 hsa05010:Alzheimer's disease 122/163 1.316 0.00286 hsa00071:Fatty acid metabolism 36/40 1.582 0.0371 hsa00190:Oxidative phosphorylation 97/130 1.312 0.0462 hsa05200:Pathways in cancer 222/328 1.190 0.0466

252

Supplementary table 2.4: Probes which correlate with all six of the correlating probes for Mgst3 in the adult mouse hippocampus as determined by Pearson correlations in GeneNetwork. For each probe, the probe ID, Entrez gene ID of the mouse gene, Entrez gene ID of the homologous gene, the mouse gene symbol and the gene’s location is given. Available on the attached DVD or online at: https://www.dropbox.com/s/xzbhlwk996plabi/Supplementary_table_2.4.xlsx?dl=0

253

Supplementary table 2.5: KEGG pathway annotations significantly enriched

(calculated by DAVID) in genes significantly co-expressed with Mgst3

(calculated by Pearson correlation in GeneFriends). The table shows the category of the enrichment and the specific annotation, the number of submitted genes in the annotation / total number of submitted genes in the category, the fold enrichment and the false discovery rate FDR. Also available on the attached

DVD or online at: https://www.dropbox.com/s/vjsz69mvxqt5p70/Supplementary_figure_2.5.docx?dl

Number of submitted genes in the annotation / Fold Term FDR Total number of Enrichment submitted genes in the category mmu03010:Ribosome 79/89 5.640 9.23E-49 mmu00190:Oxidative phosphorylation 78/130 3.813 2.32E-27 mmu05016:Huntington's disease 95/183 3.299 3.29E-27 mmu05012:Parkinson's disease 77/133 3.679 1.56E-25 mmu05010:Alzheimer's disease 83/182 2.898 1.29E-18 mmu03050:Proteasome 30/47 4.056 1.08E-09 mmu03040:Spliceosome 52/124 2.665 7.40E-09 mmu00240:Pyrimidine metabolism 34/96 2.251 0.00556

254

Supplementary table 2.6: Genes which commonly co-express with MGST3 as determined by GeneFriends and which co-express with Mgst3 in the adult mouse. The table shows the homologous genes which commonly co- express with MGST3, independent of tissue or treatment (as identified by

GeneFriends) and those which are co-expressed with Mgst3 in the adult mouse hippocampus, as determined by Pearson correlation in GeneNetwork. For each gene, its human gene symbol, human Entrez gene ID, human chromosome, human gene location (megabase pairs), mouse gene symbol, mouse Entrez

Gene ID, mouse chromosome and mouse gene location (megabase pairs) is shown. Available on the attached DVD or online at: https://www.dropbox.com/s/o8hihvublwdqmjx/Supplementary_table_2.6.xlsx?dl=

255

Supplementary table 2.7: KEGG pathway annotations significantly enriched

(calculated by DAVID) for homologous genes which commonly co-express with MGST3, independent of tissue or treatment (as identified by

GeneFriends) and which co-express with Mgst3 in the adult mouse hippocampus, as determined by Pearson correlation in GeneNetwork. The table shows the annotation, the number of submitted genes in the annotation / total number of submitted genes in the category, the fold enrichment and the false discovery rate (FDR). Also available on the attached DVD or online at: https://www.dropbox.com/s/nrslvt69v711efq/Supplementary_figure_2.7.docx?dl=

Number of submitted genes in the annotation / Fold Term Total number of FDR Enrichment submitted genes in the category hsa00190:Oxidative phosphorylation 66/130 4.059 4.17E-23 hsa05016:Huntington's disease 78/180 3.465 3.08E-22 hsa05012:Parkinson's disease 63/128 3.935 5.68E-21 hsa05010:Alzheimer's disease 69/163 3.385 1.34E-18 hsa03050:Proteasome 29/47 4.933 2.45E-11 hsa00020:Citrate cycle (TCA cycle) 14/31 3.611 0.0470

256

Appendix 2: Supplementary materials for chapter 3

257

Appendix 2: Supplementary materials for chapter 3

Some of the supplementary figures and tables are too large to fit onto an A4 page format while still being readable. Such figures are contained within the attached DVD and online. Descriptions of the figures and tables are contained here.

A2.1 Supplementary figures

Supplementary figure 3.1: Whole genome QTL maps for all elevated zero maze phenotypes (full details in Table 3.1). Chromosome numbers are shown at the top of each map and megabase pair positions within each chromosome are shown below each map. The higher red line indicates the level of genome- wide significance, i.e. a genome-wide corrected p-value of ≤ 0.05, with the blue line showing the significance of the trait at each position. Significance thresholds are shown in the upper left corner, the significant LRS being equivalent to a genome-wide corrected p-value of ≤ 0.05. The upper right corner shows the trait

ID, as shown in Table 3.1. The lower red or green line shows the additive coefficient, where a positive additive coefficient (green line) indicates that DBA/2J alleles increase trait values while a negative additive coefficient (red line) indicates that C57BL/6J alleles increase trait values. The scale for the additive coefficient is shown in green on the left of each map. Available on the attached

DVD or online at: https://www.dropbox.com/s/oyqsomsamhbbmj9/Supplementary%20Figure%203.

1.jpeg?dl=0

Supplementary figure 3.2: Whole genome QTL maps for all open field test phenotypes (full details in Table 3.2). Chromosome numbers are shown at the top of each map and megabase pair positions within each chromosome are shown below each map. The higher red line indicates the level of genome-wide significance, i.e. a genome-wide corrected p-value of ≤ 0.05, with the blue line

258

Appendix 2: Supplementary materials for chapter 3 showing the significance of the trait at each position. Significance thresholds are shown in the upper left corner, the significant LRS being equivalent to a genome- wide corrected p-value of ≤ 0.05. The upper right corner shows the trait ID, as shown in Table 3.2. The lower red or green line shows the additive coefficient, where a positive additive coefficient (green line) indicates that DBA/2J alleles increase trait values while a negative additive coefficient (red line) indicates that

C57BL/6J alleles increase trait values. The scale for the additive coefficient is shown in green on the left of each map. Available on the attached DVD or online at: https://www.dropbox.com/s/o99oeqghz240esw/Supplementary%20Figure%203.2

.jpg?dl=0

A2.2 Supplementary tables

Supplementary table 3.1: Genes within the mouse QTL we identified with human homologues (Tables 3.1 and 3.2). This shows the mouse gene symbol, chromosome (chr), megabase pair (Mbp) start, gene length in kilobase pairs

(Kbp), SNP count and SNP Density (SNP/Kbp) along with the human homologue, chromosome, Mbp start, gene length in Kbp, GATES p-value and hybrid set-based test (HYST) p-value. The two genes with the highest p-values in each region are highlighted red. Available on the attached DVD or online at: https://www.dropbox.com/s/37fppx92ud3qiqc/Supplementary%20Table%203.1.xl sx?dl=0

259

Supplementary table 3.2: Genes commonly co-expressed (co-expression value ≥ 0.5, p-value ≤ 0.05) with MCTP1 and RXRG in humans, independent of tissue or treatment, as calculated by GeneFriends. This shows the human Entrez gene ID, the gene symbol, the p-value of the co- expression, the human chromosome (chr) on which the gene is located, the megabase pair (Mbp) location of the start and end of the gene, and their correlation with MCTP1 and RXRG. Also available on the attached DVD or online at: https://www.dropbox.com/s/c82xsqtt6ir3o59/Supplementary%20Table%203.2.xlsx?dl=0

Entrez gene Gene symbol p-value Geneset friends Total Chr Start End MCTP1 RXRG ID friends 6622 SNCA 8.63E-04 3 1878 4 90645250 90759466 0.504 0.534 3759 KCNJ2 2.49E-03 3 2674 17 68164814 68176189 0.502 0.525 55790 CSGALNACT1 3.58E-03 3 3019 8 19261672 19615540 0.502 0.525 5142 PDE4B 4.99E-03 3 3370 1 66258197 66840259 0.51 0.53 4208 MEF2C 5.12E-03 3 3400 5 88013975 88199922 0.506 0.534 7130 TNFAIP6 6.87E-03 3 3751 2 152214106 152236560 0.51 0.526 23462 HEY1 7.92E-03 3 3933 8 80676245 80680098 0.505 0.554 2162 F13A1 1.22E-02 3 4543 6 6144318 6321246 0.501 0.534 51299 NRN1 2.29E-02 3 5602 6 5998232 6007200 0.502 0.567 5569 PKIA 2.42E-02 3 5704 8 79428374 79517502 0.508 0.533 8900 CCNA1 3.02E-02 3 6141 13 37005967 37017019 0.5 0.535 5376 PMP22 3.69E-02 3 6568 17 15133095 15168643 0.517 0.531 590 BCHE 4.03E-02 3 6762 3 165490692 165555260 0.501 0.578

260

Supplementary table 3.3: Enrichment of disease annotations from WebGestalt for genes commonly co-expressed with MCTP1 and RXRG

(Supplementary table 3.2). Showing the disease annotations, annotation ID, number of reference genes in the category, number of genes in the gene set and also in the category, expected number of genes in the category, ratio of enrichment, p-value from hypergeometric test and p-value adjusted for multiple testing. The human gene symbols, Entrez IDs and Ensemble IDs are shown for each gene within an annotation. Also available on the attached DVD or online at: https://www.dropbox.com/s/okjpfkdpcxxqtwa/Supplementary%20Table%203.3.xlsx?dl=0

Number of Number of p-value Expected Symbols of Entrez ID of reference genes in the p-value from adjusted Ensemble ID of Disease number of Ratio of genes genes Annotation ID genes in gene set and hypergeomet for genes within the annotation genes in the enrichment within the within the the also in the ric test multiple annotation category annotation annotation category category testing PA165108563 78 2 0.02 85.06 2.00E-04 9.00E-04 KCNJ2 3759 ENSG00000123700 Tooth malformation PMP22 5376 ENSG00000109099 PA445233 83 2 0.03 79.94 3.00E-04 9.00E-04 KCNJ2 3759 ENSG00000123700 Paralysis PMP22 5376 ENSG00000109099 PA443459 67 2 0.02 99.03 2.00E-04 9.00E-04 F13A1 2162 ENSG00000124491 Atrial fibrillation KCNJ2 3759 ENSG00000123700 PA443553 411 3 0.12 24.22 2.00E-04 9.00E-04 BCHE 590 ENSG00000114200 Brain diseases F13A1 2162 ENSG00000124491 SNCA 6622 ENSG00000145335 PA443657 438 3 0.13 22.72 3.00E-04 9.00E-04 BCHE 590 ENSG00000114200 Central nervous F13A1 2162 ENSG00000124491 system diseases SNCA 6622 ENSG00000145335 PA445619 81 2 0.02 81.91 3.00E-04 9.00E-04 KCNJ2 3759 ENSG00000123700 Scoliosis PMP22 5376 ENSG00000109099 PA447127 38 2 0.01 174.6 5.86E-05 9.00E-04 BCHE 590 ENSG00000114200 Lewy Body disease SNCA 6622 ENSG00000145335 PA446858 404 3 0.12 24.63 2.00E-04 9.00E-04 BCHE 590 ENSG00000114200 Neurodegenerative PMP22 5376 ENSG00000109099 diseases SNCA 6622 ENSG00000145335 PA447208 564 3 0.17 17.65 6.00E-04 1.50E-03 BCHE 590 ENSG00000114200 Mental disorders PDE4B 5142 ENSG00000184588 SNCA 6622 ENSG00000145335 PA152241270 147 2 0.04 45.14 9.00E-04 2.10E-03 BCHE 590 ENSG00000114200 Alexander disease PMP22 5376 ENSG00000109099

261

Supplementary table 3.4: Gene Ontology (GO) annotation enrichment from

WebGestalt for genes commonly co-expressed with MCTP1 and RXRG

(Supplementary table 3.2). Showing the GO category, GO annotation, GO ID, number of reference genes in the category, number of genes in the gene set and also in the category, expected number of genes in the category, ratio of enrichment, p-value from hypergeometric test and p-value adjusted for multiple testing. The human gene symbols, Entrez IDs and Ensemble IDs are shown for each gene within an annotation. Available on the attached DVD or online at: https://www.dropbox.com/s/crkng1ewdtw1dca/Supplementary%20Table%203.4.x lsx?dl=0

Supplementary table 3.5: Genes commonly co-expressed (co-expression value ≥ 0.5, p-value ≤ 0.05) with RXRG and TNR, in humans, independent of tissue or treatment, as calculated by GeneFriends. This shows the human

Entrez gene ID, the gene symbol, the p-value of the co-expression, the human chromosome on which the gene is located, the megabase pair (Mbp) location of the start and end of the gene, and their correlation with RXRG and TNR.

Available on the attached DVD or online at: https://www.dropbox.com/s/2bzwlqazdkya2sw/Supplementary%20Table%203.5. xlsx?dl=0

Supplementary table 3.6: Enrichment of disease annotations from

WebGestalt for genes commonly co-expressed of RXRG and TNR

(Supplementary table 3.5). Showing the disease annotation, the annotation ID, number of reference genes in the category, number of genes in the gene set and also in the category, expected number of genes in the category, ratio of enrichment, p-value from hypergeometric test and p-value adjusted for multiple testing. The human gene symbols, Entrez IDs and Ensemble IDs are shown for

262

Appendix 2: Supplementary materials for chapter 3 each gene within an annotation. Available on the attached DVD or online at: https://www.dropbox.com/s/2ugqjwgt3krmfty/Supplementary%20Table%203.6.xls x?dl=0

Supplementary table 3.7: Gene Ontology (GO) annotation enrichment from

WebGestalt for genes commonly co-expressed with RXRG and TNR

(Supplementary table 3.5). Showing the GO category, GO annotation, GO ID, number of reference genes in the category, number of genes in the gene set and also in the category, expected number of genes in the category, ratio of enrichment, p-value from hypergeometric test and p-value adjusted for multiple testing. The human gene symbols, Entrez IDs and Ensemble IDs are shown for each gene within an annotation. Available on the attached DVD or online at: https://www.dropbox.com/s/4mmsw6267l5cgbj/Supplementary%20Table%203.7. xlsx?dl=0

Supplementary table 3.8: KEGG pathway annotation enrichment from

WebGestalt for genes commonly co-expressed with RXRG and TNR

(Supplementary table 3.5). Showing the KEGG pathway annotation, the annotation ID, number of reference genes in the category, number of genes in the gene set and also in the category, expected number of genes in the category, ratio of enrichment, p-value from hypergeometric test and p-value adjusted for multiple testing. The human gene symbols, Entrez IDs and Ensemble IDs are shown for each gene within an annotation. Available on the attached DVD or online at: https://www.dropbox.com/s/n5ygv9ofpmq7kcr/Supplementary%20Table%203.8.x lsx?dl=0

263

Appendix 2: Supplementary materials for chapter 3

Supplementary table 3.9: Pathway Commons annotation enrichment from

WebGestalt for genes commonly co-expressed with RXRG and TNR

(Supplementary table 3.5). Showing the Pathway Commons annotation, the annotation ID, number of reference genes in the category, number of genes in the gene set and also in the category, expected number of genes in the category, ratio of enrichment, p-value from hypergeometric test and p-value adjusted for multiple testing. The human gene symbols, Entrez IDs and Ensemble IDs are shown for each gene within an annotation. Available on the attached DVD or online at: https://www.dropbox.com/s/u7fels0uwx9zqmk/Supplementary%20Table%203.9.x lsx?dl=0

Supplementary table 3.10: Wikipathways annotation enrichment from

WebGestalt for genes commonly co-expressed with RXRG and TNR

(Supplementary table 3.5). Showing the Wikipathways annotation, the annotation ID, number of reference genes in the category, number of genes in the gene set and also in the category, expected number of genes in the category, ratio of enrichment, p-value from hypergeometric test and p-value adjusted for multiple testing. The human gene symbols, Entrez IDs and Ensemble IDs are shown for each gene within an annotation. Available on the attached DVD or online at: https://www.dropbox.com/s/w8mgazta2hngim1/Supplementary%20Table%203.1

0.xlsx?dl=0

Supplementary table 3.11: Pearson correlation matrix (calculated by

GeneNetwork) of striatal expression of mental disorder related gene probes and probes for Cmya5, Mctp1, Rxrg and Tnr. Significant correlations are highlighted in red for positive and purple for negative correlations. For each

264

Appendix 2: Supplementary materials for chapter 3 probe its ID, the gene symbol of the gene it targets, the chromosome and megabase pair (Mbp) position it targets and where this is within the gene are shown. Available on the attached DVD or online at: https://www.dropbox.com/s/82rx4if1wyugape/Supplementary%20Table%203.11. xlsx?dl=0

Supplementary table 3.12: Pearson correlation matrix (calculated by

GeneNetwork) of hippocampal expression of mental disorder related gene probes and probes for Cmya5, Mctp1, Rxrg and Tnr. Significant correlations are highlighted in red for positive and purple for negative correlations. For each probe its ID, the gene symbol of the gene it targets, the chromosome and position it targets and where this is within the gene are shown. Available on the attached DVD or online at: https://www.dropbox.com/s/yigjf10yw1ff317/Supplementary%20Table%203.12.xl sx?dl=0

265

Appendix 2: Supplementary materials for chapter 3

Supplementary table 3.13: Genes which commonly co-express with MCTP1, RXRG and TNR (from GeneFriends; Supplementary tables 3.2 and 3.5) and co-express with Mctp1, Rxrg and Tnr in the mouse striatum (from GeneNetwork;

Supplementary table 3.11). This shows the mouse gene symbol, the Entrez ID of the human homologue and the human gene name. Also available on the attached DVD or online at: https://www.dropbox.com/s/c360lvtk9e47t50/Supplementary%20Table%203.13.xlsx?dl=0

Mouse gene Human homologue symbols Entrez ID Human gene name A2bp1 54715 RNA binding protein, fox-1 homolog 1 Acsl6 23305 acyl-CoA synthetase long-chain family member 6 Ank3 288 ankyrin 3, node of Ranvier (ankyrin G) Cacna1c 775 calcium channel, voltage-dependent, L type, alpha 1C subunit Cartpt 9607 CART prepropeptide Dcx 1641 doublecortin v-erb-b2 avian erythroblastic leukemia viral oncogene Erbb4 2066 homolog 4 Gabra4 2557 gamma-aminobutyric acid (GABA) A receptor, alpha 4 Gabrb1 2560 gamma-aminobutyric acid (GABA) A receptor, beta 1 Gabrb2 2561 gamma-aminobutyric acid (GABA) A receptor, beta 2 Gabrg2 2566 gamma-aminobutyric acid (GABA) A receptor, gamma 2 glutamate decarboxylase 2 (pancreatic islets and brain, Gad2 2572 65kDa) Gria4 2893 glutamate receptor, ionotropic, AMPA 4 Grik1 2897 glutamate receptor, ionotropic, kainate 1 Grin2a 2903 glutamate receptor, ionotropic, N-methyl D-aspartate 2A Grin2b 2904 glutamate receptor, ionotropic, N-methyl D-aspartate 2B Grm3 2913 glutamate receptor, metabotropic 3 Hcrt 3060 hypocretin (orexin) neuropeptide precursor 5-hydroxytryptamine (serotonin) receptor 2C, G protein- Htr2c 3358 coupled Il1rapl2 26280 interleukin 1 receptor accessory protein-like 2 Mctp1 79772 multiple C2 domains, transmembrane 1 Mog 4340 myelin oligodendrocyte glycoprotein Nos1 4842 nitric oxide synthase 1 (neuronal) Ntrk2 4915 neurotrophic tyrosine kinase, receptor, type 2 Ntrk3 4916 neurotrophic tyrosine kinase, receptor, type 3 Odz4 26011 teneurin transmembrane protein 4 Oprk1 4986 opioid receptor, kappa 1 Pde4b 5142 phosphodiesterase 4B, cAMP-specific Phkg1 5260 phosphorylase kinase, gamma 1 (muscle) Reln 5649 reelin Rxrg 6258 retinoid X receptor, gamma solute carrier family 1 (glial high affinity glutamate Slc1a2 6506 transporter), member 2 Syp 6855 synaptophysin Tnr 7143 tenascin R Vgf 7425 VGF nerve growth factor inducible

266

Appendix 2: Supplementary materials for chapter 3

Supplementary table 3.14: Phenotypes which correlate with expression of all correlated probes for Cmya5, Mctp1, Rxrg and Tnr. For each gene, the phenotypes which correlated with all the probes is shown, including the

GeneNetwork record ID, the phenotype name and description, the authors of the study, the year it was published on GeneNetwork, the Pubmed ID if it is published data and the number of lines in which the phenotype and gene expression has been measured. For each probe the ID is shown with a

Pearson’s correlation value (r) and a significance value (p). Available on the attached DVD or online at: https://www.dropbox.com/s/pd3l114ybrhw94t/Supplementary%20Table%203.14. xlsx?dl=0

Supplementary table 3.15: Pearson correlations between our target phenotypes (Tables 3.1 and 3.2) and phenotypes found to correlate with expression of our candidate genes (Supplementary table 3.14). Phenotype

IDs are shown for our target phenotypes, while IDs and descriptions are shown for the phenotypes which correlated with expression of our candidate genes.

Pearson’s correlation values (r) are shown in the table, as well as the number of lines in which both phenotypes have been measured (n). Correlations of r ≥ 0.5 are shown in red and of r ≤ -0.5 are shown in purple. Available on the attached

DVD or online at: https://www.dropbox.com/s/0mou025rf98fgqa/Supplementary%20Table%203.15. xlsx?dl=0

267

Appendix 3: Supplementary materials for chapter 4

268

Appendix 3: Supplementary materials for chapter 4

The supplementary tables are too large to fit onto an A4 page format while still being readable. The tables are contained within the attached DVD and online.

Descriptions of the tables are contained here.

A3.1 Supplementary tables

Supplementary table 4.1: Functional and other details about the genes within the OspIge5.1 QTL for B6 maternal care on day 14, obtained from

GeneNetwork, Entrez genes, and Mouse Genome Informatics. Available on the attached DVD or online at: https://www.dropbox.com/s/bfth6f8jroebb60/Supplementary%20table%204.1.xlsx

?dl=0

Supplementary table 4.2: Functional and other details about the genes within the MatDge1.1 QTL for BXD nestbuilding on day 6, obtained from

GeneNetwork, Entrez genes, and Mouse Genome Informatics. Available on the attached DVD or online at: https://www.dropbox.com/s/hrr0fiygcq4ybqh/Supplementary%20table%204.2.xlsx

?dl=0

Supplementary table 4.3: Functional and other details about the genes within the MatDge10.1 QTL for BXD maternal care on day 6, obtained from

GeneNetwork, Entrez genes, and Mouse Genome Informatics. Available on the attached DVD or online at: https://www.dropbox.com/s/a7atiw0n2u7jgnn/Supplementary%20table%204.3.xls x?dl=0

269

Appendix 3: Supplementary materials for chapter 4

Supplementary table 4.4: Functional and other details about the genes within the OspDge5.1 QTL BXD offspring solicitation on day 6, obtained from GeneNetwork, Entrez genes, and Mouse Genome Informatics. Available on the attached DVD or online at: https://www.dropbox.com/s/kvul6r7mhplj9ws/Supplementary%20table%204.4.xls x?dl=0

270

Appendix 4: Supplementary materials for chapter 5

271

Appendix 4: Supplementary materials for chapter 5

A4.1 Supplementary tables

Supplementary table 5.1: Genes within the SocInt2.1 QTL. The table shows the position of the start of the genes (megabase pairs), and if they contain non- synonymous SNPs (nsSNPs), or insertions or deletions (indels). All data is adapted from GeneNetwork and is from the mm9 genome build. Available on the attached DVD or online at: https://www.dropbox.com/s/8i78ev23rou5x7s/Supplementary%20table%205.1.xls x?dl=0

272

Appendix 4: Supplementary materials for chapter 5

Supplementary table 5.2: Genes within the SocInt4.1 QTL. The table shows the position of the start of the genes (megabase pairs), and if they contain non-synonymous SNPs (nsSNPs), or insertions or deletions (indels). All data is adapted from GeneNetwork and is from the mm9 genome build. Also available on the attached DVD or online at: https://www.dropbox.com/s/qy1z3wmv5vhnp54/Supplementary%20table%205.2.xlsx?dl=0 Gene Start nsSNPs C57BL/6J vs DBA/2J Indels in BXD mice Tmem68 3.47619 34 1 B230117O15Rik 3.48223 0 0 Tgs1 3.50203 23 0 2210414B05Rik 3.58366 0 0 Lyn 3.60527 42 10 6330407A03Rik 3.64211 0 0 9430025C20Rik 3.73378 9 0 Rps20 3.76162 3 1 Mos 3.7978 5 0 Plag1 3.8283 1 0 Chchd7 3.86606 0 0 RP23-24J10.7 3.92309 0 0 4833413O15Rik 3.98381 0 0 Penk1 4.06068 0 0 A830012C17Rik 4.06547 0 0 Impad1 4.6915 0 0 4930423M02Rik 5.55947 0 0 1700012H17Rik 5.57133 0 0 3110003A22Rik 6.11825 0 0 Cyp7a1 6.19276 0 0 4930430E12Rik 6.27925 0 0 Sdcbp 6.29283 0 0 Nsmaf 6.32335 0 0 Tox 6.61453 0 5 2610024J18Rik 6.86904 0 0 8430436N08Rik 7.48783 0 0 Car8 8.06864 184 9 Rab2 8.46279 5 12 D130047N11Rik 8.50992 0 0 C530036F05Rik 8.52419 0 0 Chd7 8.61807 6 0 Rlbp1l1 9.19651 25 7 D130060J02Rik 9.32573 0 0 Asph 9.37623 47 35 4930412C18Rik 9.69773 80 5 5730591J02Rik 9.69783 72 5 Gdf6 9.77152 1 0 4930448K20Rik 9.84322 0 0 LOC242317 10.3662 0 1 1700123O12Rik 10.4352 6 14 2610301B20Rik 10.8016 17 3 Plekhf2 10.9158 10 1 2310030N02Rik 10.9782 7 1 Trp53inp1 11.0836 18 1 A630034I12Rik 11.1031 0 2 Ccne2 11.1185 4 2 Ints8 11.1263 35 4 Narg3 11.1888 0 0 Dpy19l4 11.1922 18 3 Rbm35a 11.2592 22 2 1110037F02Rik 11.4131 26 9 E130016E03Rik 11.4861 28 3 Gem 11.6315 0 0 Cdh17 11.6853 0 0 Ppm2c 11.8853 0 0 1700123M08Rik 11.8937 0 1 4932430A15Rik 11.9418 0 0 Tmem67 11.9668 0 0 C430048L16Rik 12.0165 0 0 Rbm12b 12.0673 5 0 6720467C03Rik 12.0809 0 0 0610009K14Rik 13.537 0 0 Runx1t1 13.6704 8 8

273

Appendix 4: Supplementary materials for chapter 5

Supplementary table 5.3: Genes within the SocInt15.1 QTL. The table shows the position of the start of the genes (megabase pairs), and if they contain non-synonymous

SNPs (nsSNPs), or insertions or deletions (indels). All data is adapted from

GeneNetwork and is from the mm9 genome build. Also available on the attached DVD or online at: https://www.dropbox.com/s/grl8yerjgrlvtsb/Supplementary%20table%205.3.xlsx?dl=0

nsSNPs C57BL/6J vs Indels in BXD Gene Start DBA/2J mice Sepp1 3.220976 3 0 Ghr 3.267759 2 4 Fbxo4 3.915462 7 1 C7 3.929744 0 0 AW549877 3.932034 37 5 A630020A06 3.946038 34 4 BC037032 3.972972 1 0 Oxct1 3.976427 0 3 Plcxd3 4.32549 0 2 C6 4.677209 0 0 4930455B06Rik 4.848736 0 0 A530026G17 4.997354 0 0 Card6 5.047451 0 0 Rpl37 5.066644 0 0 Prkaa1 5.09386 0 0 Ttc33 5.135559 0 0 Ptger4 5.183405 0 0 4930505O19Rik 5.438619 0 0 5430437J10Rik 5.446317 0 0

274