A Journey Through Genetics to Biology.

Although my engagement with human genetics emerged gradually, and sometimes serendipitously, it has held me spellbound for decades. Without my teachers, students, postdocs, colleagues, and collaborators, I would not be writing this review of my scientific adventures. Early gene and disease mapping was a satisfying puzzle-solving exercise, but building biological insight was my main goal. The project trajectory was hugely influenced by the evolutionarily conserved nature of the implicated genes and by the pace of progress in genetic technologies. The rich detail of clinical observations, particularly in eye disease, makes humans an excellent model, especially when complemented by the use of multiple other animal species for experimental validation. The contributions of collaborators and rivals also influenced our approach. We are very fortunate to work in this era of unprecedented progress in genetics and genomics.

Short-read whole genome sequencing identifies causative variants in most individuals with previously unexplained aniridia.

BACKGROUND: Classic aniridia is a highly penetrant autosomal dominant disorder characterised by congenital absence of the iris, foveal hypoplasia, optic disc anomalies and progressive opacification of the cornea. >90% of cases of classic aniridia are caused by heterozygous, loss-of-function variants affecting the PAX6 locus. METHODS: Short-read whole genome sequencing was performed on 51 (39 affected) individuals from 37 different families who had screened negative for mutations in the PAX6 coding region. RESULTS: Likely causative mutations were identified in 22 out of 37 (59%) families. In 19 out of 22 families, the causative genomic changes have an interpretable deleterious impact on the PAX6 locus. Of these 19 families, 1 has a novel heterozygous PAX6 frameshift variant missed on previous screens, 4 have single nucleotide variants (SNVs) (one novel) affecting essential splice sites of PAX6 5' non-coding exons and 2 have deep intronic SNV (one novel) resulting in gain of a donor splice site. In 12 out of 19, the causative variants are large-scale structural variants; 5 have partial or whole gene deletions of PAX6, 3 have deletions encompassing critical PAX6 cis-regulatory elements, 2 have balanced inversions with disruptive breakpoints within the PAX6 locus and 2 have complex rearrangements disrupting PAX6. The remaining 3 of 22 families have deletions encompassing FOXC1 (a known cause of atypical aniridia). Seven of the causative variants occurred de novo and one cosegregated with familial aniridia. We were unable to establish inheritance status in the remaining probands. No plausibly causative SNVs were identified in PAX6 cis-regulatory elements. CONCLUSION: Whole genome sequencing proves to be an effective diagnostic test in most individuals with previously unexplained aniridia.

Recurrent heterozygous PAX6 missense variants cause severe bilateral microphthalmia via predictable effects on DNA-protein interaction.

PURPOSE: Most classical aniridia is caused by PAX6 haploinsufficiency. PAX6 missense variants can be hypomorphic or mimic haploinsufficiency. We hypothesized that missense variants also cause previously undescribed disease by altering the affinity and/or specificity of PAX6 genomic interactions. METHODS: We screened PAX6 in 372 individuals with bilateral microphthalmia, anophthalmia, or coloboma (MAC) from the Medical Research Council Human Genetics Unit eye malformation cohort (HGUeye) and reviewed data from the Deciphering Developmental Disorders study. We performed cluster analysis on PAX6-associated ocular phenotypes by variant type and molecular modeling of the structural impact of 86 different PAX6 causative missense variants. RESULTS: Eight different PAX6 missense variants were identified in 17 individuals (15 families) with MAC, accounting for 4% (15/372) of our cohort. Seven altered the paired domain (p.[Arg26Gln]x1, p.[Gly36Val]x1, p.[Arg38Trp]x2, p.[Arg38Gln]x1, p.[Gly51Arg]x2, p.[Ser54Arg]x2, p.[Asn124Lys]x5) and one the homeodomain (p.[Asn260Tyr]x1). p.Ser54Arg and p.Asn124Lys were exclusively associated with severe bilateral microphthalmia. MAC-associated variants were predicted to alter but not ablate DNA interaction, consistent with the electrophoretic mobility shifts observed using mutant paired domains with well-characterized PAX6-binding sites. We found no strong evidence for novel PAX6-associated extraocular disease. CONCLUSION: Altering the affinity and specificity of PAX6-binding genome-wide provides a plausible mechanism for the worse-than-null effects of MAC-associated missense variants.

Unique noncoding variants upstream of PRDM13 are associated with a spectrum of developmental retinal dystrophies including progressive bifocal chorioretinal atrophy.

The autosomal dominant progressive bifocal chorioretinal atrophy (PBCRA) disease locus has been mapped to chromosome 6q14-16.2 that overlaps the North Carolina macular dystrophy (NCMD) locus MCDR1. NCMD is a nonprogressive developmental macular dystrophy, in which variants upstream of PRDM13 have been implicated. Whole genome sequencing was performed to interrogate structural variants (SVs) and single nucleotide variants (SNVs) in eight individuals, six affected individuals from two families with PBCRA, and two individuals from an additional family with a related developmental macular dystrophy. A SNV (chr6:100,046,804T>C), located 7.8 kb upstream of the PRDM13 gene, was shared by all PBCRA-affected individuals in the disease locus. Haplotype analysis suggested that the variant arose independently in the two families. The two affected individuals from Family 3 were screened for rare variants in the PBCRA and NCMD loci. This revealed a de novo variant in the proband, 21 bp from the first SNV (chr6:100,046,783A>C). This study expands the noncoding variant spectrum upstream of PRDM13 and suggests altered spatio-temporal expression of PRDM13 as a candidate disease mechanism in the phenotypically distinct but related conditions, NCMD and PBCRA.

A Restricted Repertoire of De Novo Mutations in ITPR1 Cause Gillespie Syndrome with Evidence for Dominant-Negative Effect.

Gillespie syndrome (GS) is characterized by bilateral iris hypoplasia, congenital hypotonia, non-progressive ataxia, and progressive cerebellar atrophy. Trio-based exome sequencing identified de novo mutations in ITPR1 in three unrelated individuals with GS recruited to the Deciphering Developmental Disorders study. Whole-exome or targeted sequence analysis identified plausible disease-causing ITPR1 mutations in 10/10 additional GS-affected individuals. These ultra-rare protein-altering variants affected only three residues in ITPR1: Glu2094 missense (one de novo, one co-segregating), Gly2539 missense (five de novo, one inheritance uncertain), and Lys2596 in-frame deletion (four de novo). No clinical or radiological differences were evident between individuals with different mutations. ITPR1 encodes an inositol 1,4,5-triphosphate-responsive calcium channel. The homo-tetrameric structure has been solved by cryoelectron microscopy. Using estimations of the degree of structural change induced by known recessive- and dominant-negative mutations in other disease-associated multimeric channels, we developed a generalizable computational approach to indicate the likely mutational mechanism. This analysis supports a dominant-negative mechanism for GS variants in ITPR1. In GS-derived lymphoblastoid cell lines (LCLs), the proportion of ITPR1-positive cells using immunofluorescence was significantly higher in mutant than control LCLs, consistent with an abnormality of nuclear calcium signaling feedback control. Super-resolution imaging supports the existence of an ITPR1-lined nucleoplasmic reticulum. Mice with Itpr1 heterozygous null mutations showed no major iris defects. Purkinje cells of the cerebellum appear to be the most sensitive to impaired ITPR1 function in humans. Iris hypoplasia is likely to result from either complete loss of ITPR1 activity or structure-specific disruption of multimeric interactions.

Genome sequencing-the dawn of a game-changing era.

The development of genome sequencing technologies has revolutionized the biological sciences in ways which could not have been imagined at the time. This article sets out to document the dawning of the age of genomics and to consider the impact of this revolution on biological investigation, our understanding of life, and the relationship between science and society.

Functional assessment of disease-associated regulatory variants in vivo using a versatile dual colour transgenesis strategy in zebrafish.

Disruption of gene regulation by sequence variation in non-coding regions of the genome is now recognised as a significant cause of human disease and disease susceptibility. Sequence variants in cis-regulatory elements (CREs), the primary determinants of spatio-temporal gene regulation, can alter transcription factor binding sites. While technological advances have led to easy identification of disease-associated CRE variants, robust methods for discerning functional CRE variants from background variation are lacking. Here we describe an efficient dual-colour reporter transgenesis approach in zebrafish, simultaneously allowing detailed in vivo comparison of spatio-temporal differences in regulatory activity between putative CRE variants and assessment of altered transcription factor binding potential of the variant. We validate the method on known disease-associated elements regulating SHH, PAX6 and IRF6 and subsequently characterise novel, ultra-long-range SOX9 enhancers implicated in the craniofacial abnormality Pierre Robin Sequence. The method provides a highly cost-effective, fast and robust approach for simultaneously unravelling in a single assay whether, where and when in embryonic development a disease-associated CRE-variant is affecting its regulatory function.

A recurrent de novo mutation in ACTG1 causes isolated ocular coloboma.

Ocular coloboma (OC) is a defect in optic fissure closure and is a common cause of severe congenital visual impairment. Bilateral OC is primarily genetically determined and shows marked locus heterogeneity. Whole-exome sequencing (WES) was used to analyze 12 trios (child affected with OC and both unaffected parents). This identified de novo mutations in 10 different genes in eight probands. Three of these genes encoded proteins associated with actin cytoskeleton dynamics: ACTG1, TWF1, and LCP1. Proband-only WES identified a second unrelated individual with isolated OC carrying the same ACTG1 allele, encoding p.(Pro70Leu). Both individuals have normal neurodevelopment with no extra-ocular signs of Baraitser-Winter syndrome. We found this mutant protein to be incapable of incorporation into F-actin. The LCP1 and TWF1 variants each resulted in only minor disturbance of actin interactions, and no further plausibly causative variants were identified in these genes on resequencing 380 unrelated individuals with OC.

Heterozygous loss-of-function mutations in YAP1 cause both isolated and syndromic optic fissure closure defects.

Exome sequence analysis of affected individuals from two families with autosomal-dominant inheritance of coloboma identified two different cosegregating heterozygous nonsense mutations (c.370C>T [p.Arg124*] and c. 1066G>T [p.Glu356*]) in YAP1. The phenotypes of the affected families differed in that one included no extraocular features and the other manifested with highly variable multisystem involvement, including hearing loss, intellectual disability, hematuria, and orofacial clefting. A combined LOD score of 4.2 was obtained for the association between YAP1 loss-of-function mutations and the phenotype in these families. YAP1 encodes an effector of the HIPPO-pathway-induced growth response, and whole-mount in situ hybridization in mouse embryos has shown that Yap1 is strongly expressed in the eye, brain, and fusing facial processes. RT-PCR showed that an alternative transcription start site (TSS) in intron 1 of YAP1 and Yap1 is widely used in human and mouse development, respectively. Transcripts from the alternative TSS are predicted to initiate at codon Met179 relative to the canonical transcript (RefSeq NM_001130145). In these alternative transcripts, the c.370C>T mutation in family 1305 is within the 5' UTR and cannot result in nonsense-mediated decay (NMD). The c. 1066G>T mutation in family 132 should result in NMD in transcripts from either TSS. Amelioration of the phenotype by the alternative transcripts provides a plausible explanation for the phenotypic differences between the families.

Monoallelic and biallelic mutations in MAB21L2 cause a spectrum of major eye malformations.

We identified four different missense mutations in the single-exon gene MAB21L2 in eight individuals with bilateral eye malformations from five unrelated families via three independent exome sequencing projects. Three mutational events altered the same amino acid (Arg51), and two were identical de novo mutations (c.151C>T [p.Arg51Cys]) in unrelated children with bilateral anophthalmia, intellectual disability, and rhizomelic skeletal dysplasia. c.152G>A (p.Arg51His) segregated with autosomal-dominant bilateral colobomatous microphthalmia in a large multiplex family. The fourth heterozygous mutation (c.145G>A [p.Glu49Lys]) affected an amino acid within two residues of Arg51 in an adult male with bilateral colobomata. In a fifth family, a homozygous mutation (c.740G>A [p.Arg247Gln]) altering a different region of the protein was identified in two male siblings with bilateral retinal colobomata. In mouse embryos, Mab21l2 showed strong expression in the developing eye, pharyngeal arches, and limb bud. As predicted by structural homology, wild-type MAB21L2 bound single-stranded RNA, whereas this activity was lost in all altered forms of the protein. MAB21L2 had no detectable nucleotidyltransferase activity in vitro, and its function remains unknown. Induced expression of wild-type MAB21L2 in human embryonic kidney 293 cells increased phospho-ERK (pERK1/2) signaling. Compared to the wild-type and p.Arg247Gln proteins, the proteins with the Glu49 and Arg51 variants had increased stability. Abnormal persistence of pERK1/2 signaling in MAB21L2-expressing cells during development is a plausible pathogenic mechanism for the heterozygous mutations. The phenotype associated with the homozygous mutation might be a consequence of complete loss of MAB21L2 RNA binding, although the cellular function of this interaction remains unknown.

Disruption of autoregulatory feedback by a mutation in a remote, ultraconserved PAX6 enhancer causes aniridia.

The strictly regulated expression of most pleiotropic developmental control genes is critically dependent on the activity of long-range cis-regulatory elements. This was revealed by the identification of individuals with a genetic condition lacking coding-region mutations in the gene commonly associated with the disease but having a variety of nearby chromosomal abnormalities, collectively described as cis-ruption disease cases. The congenital eye malformation aniridia is caused by haploinsufficiency of the developmental regulator PAX6. We discovered a de novo point mutation in an ultraconserved cis-element located 150 kb downstream from PAX6 in an affected individual with intact coding region and chromosomal locus. The element SIMO acts as a strong enhancer in developing ocular structures. The mutation disrupts an autoregulatory PAX6 binding site, causing loss of enhancer activity, resulting in defective maintenance of PAX6 expression. These findings reveal a distinct regulatory mechanism for genetic disease by disruption of an autoregulatory feedback loop critical for maintenance of gene expression through development.

Sequencing of Pax6 loci from the elephant shark reveals a family of Pax6 genes in vertebrate genomes, forged by ancient duplications and divergences.

Pax6 is a developmental control gene essential for eye development throughout the animal kingdom. In addition, Pax6 plays key roles in other parts of the CNS, olfactory system, and pancreas. In mammals a single Pax6 gene encoding multiple isoforms delivers these pleiotropic functions. Here we provide evidence that the genomes of many other vertebrate species contain multiple Pax6 loci. We sequenced Pax6-containing BACs from the cartilaginous elephant shark (Callorhinchus milii) and found two distinct Pax6 loci. Pax6.1 is highly similar to mammalian Pax6, while Pax6.2 encodes a paired-less Pax6. Using synteny relationships, we identify homologs of this novel paired-less Pax6.2 gene in lizard and in frog, as well as in zebrafish and in other teleosts. In zebrafish two full-length Pax6 duplicates were known previously, originating from the fish-specific genome duplication (FSGD) and expressed in divergent patterns due to paralog-specific loss of cis-elements. We show that teleosts other than zebrafish also maintain duplicate full-length Pax6 loci, but differences in gene and regulatory domain structure suggest that these Pax6 paralogs originate from a more ancient duplication event and are hence renamed as Pax6.3. Sequence comparisons between mammalian and elephant shark Pax6.1 loci highlight the presence of short- and long-range conserved noncoding elements (CNEs). Functional analysis demonstrates the ancient role of long-range enhancers for Pax6 transcription. We show that the paired-less Pax6.2 ortholog in zebrafish is expressed specifically in the developing retina. Transgenic analysis of elephant shark and zebrafish Pax6.2 CNEs with homology to the mouse NRE/Pα internal promoter revealed highly specific retinal expression. Finally, morpholino depletion of zebrafish Pax6.2 resulted in a "small eye" phenotype, supporting a role in retinal development. In summary, our study reveals that the pleiotropic functions of Pax6 in vertebrates are served by a divergent family of Pax6 genes, forged by ancient duplication events and by independent, lineage-specific gene losses.

A survey of ancient conserved non-coding elements in the PAX6 locus reveals a landscape of interdigitated cis-regulatory archipelagos.

Biological differences between cell types and developmental processes are characterised by differences in gene expression profiles. Gene-distal enhancers are key components of the regulatory networks that specify the tissue-specific expression patterns driving embryonic development and cell fate decisions, and variations in their sequences are a major contributor to genetic disease and disease susceptibility. Despite advances in the methods for discovery of putative cis-regulatory sequences, characterisation of their spatio-temporal enhancer activities in a mammalian model system remains a major bottle-neck. We employed a strategy that combines gnathostome sequence conservation with transgenic mouse and zebrafish reporter assays to survey the genomic locus of the developmental control gene PAX6 for the presence of novel cis-regulatory elements. Sequence comparison between human and the cartilaginous elephant shark (Callorhinchus milii) revealed several ancient gnathostome conserved non-coding elements (agCNEs) dispersed widely throughout the PAX6 locus, extending the range of the known PAX6 cis-regulatory landscape to contain the full upstream PAX6-RCN1 intergenic region. Our data indicates that ancient conserved regulatory sequences can be tested effectively in transgenic zebrafish even when not conserved in zebrafish themselves. The strategy also allows efficient dissection of compound regulatory regions previously assessed in transgenic mice. Remarkable overlap in expression patterns driven by sets of agCNEs indicates that PAX6 resides in a landscape of multiple tissue-specific regulatory archipelagos.

Identification of novel craniofacial regulatory domains located far upstream of SOX9 and disrupted in Pierre Robin sequence.

Mutations in the coding sequence of SOX9 cause campomelic dysplasia (CD), a disorder of skeletal development associated with 46,XY disorders of sex development (DSDs). Translocations, deletions, and duplications within a ∼2 Mb region upstream of SOX9 can recapitulate the CD-DSD phenotype fully or partially, suggesting the existence of an unusually large cis-regulatory control region. Pierre Robin sequence (PRS) is a craniofacial disorder that is frequently an endophenotype of CD and a locus for isolated PRS at ∼1.2-1.5 Mb upstream of SOX9 has been previously reported. The craniofacial regulatory potential within this locus, and within the greater genomic domain surrounding SOX9, remains poorly defined. We report two novel deletions upstream of SOX9 in families with PRS, allowing refinement of the regions harboring candidate craniofacial regulatory elements. In parallel, ChIP-Seq for p300 binding sites in mouse craniofacial tissue led to the identification of several novel craniofacial enhancers at the SOX9 locus, which were validated in transgenic reporter mice and zebrafish. Notably, some of the functionally validated elements fall within the PRS deletions. These studies suggest that multiple noncoding elements contribute to the craniofacial regulation of SOX9 expression, and that their disruption results in PRS.

Discovery and assessment of conserved Pax6 target genes and enhancers.

The characterization of transcriptional networks (TNs) is essential for understanding complex biological phenomena such as development, disease, and evolution. In this study, we have designed and implemented a procedure that combines in silico target screens with zebrafish and mouse validation, in order to identify cis-elements and genes directly regulated by Pax6. We chose Pax6 as the paradigm because of its crucial roles in organogenesis and human disease. We identified over 600 putative Pax6 binding sites and more than 200 predicted direct target genes, conserved in evolution from zebrafish to human and to mouse. This was accomplished using hidden Markov models (HMMs) generated from experimentally validated Pax6 binding sites. A small sample of genes, expressed in the neural lineage, was chosen from the predictions for RNA in situ validation using zebrafish and mouse models. Validation of DNA binding to some predicted cis-elements was also carried out using chromatin immunoprecipitation (ChIP) and zebrafish reporter transgenic studies. The results show that this combined procedure is a highly efficient tool to investigate the architecture of TNs and constitutes a useful complementary resource to ChIP and expression data sets because of its inherent spatiotemporal independence. We have identified several novel direct targets, including some putative disease genes, among them Foxp2; these will allow further dissection of Pax6 function in development and disease.

Stochasticity in genetics and gene regulation.

Development from fertilized egg to functioning multi-cellular organism requires precision. There is no precision, and often no survival, without plasticity. Plasticity is conferred partly by stochastic variation, present inherently in all biological systems. Gene expression levels fluctuate ubiquitously through transcription, alternative splicing, translation and turnover. Small differences in gene expression are exploited to trigger early differentiation, conferring distinct function on selected individual cells and setting in motion regulatory interactions. Non-selected cells then acquire new functions along the spatio-temporal developmental trajectory. The differentiation process has many stochastic components. Meiotic segregation, mitochondrial partitioning, X-inactivation and the dynamic DNA binding of transcription factor assemblies-all exhibit randomness. Non-random X-inactivation generally signals deleterious X-linked mutations. Correct neural wiring, such as retina to brain, arises through repeated confirmatory activity of connections made randomly. In immune system development, both B-cell antibody generation and the emergence of balanced T-cell categories begin through stochastic trial and error followed by functional selection. Aberrant selection processes lead to immune dysfunction. DNA sequence variants also arise through stochastic events: some involving environmental fluctuation (radiation or presence of pollutants), or genetic repair system malfunction. The phenotypic outcome of mutations is also fluid. Mutations may be advantageous in some circumstances, deleterious in others. This article is part of a discussion meeting issue 'Causes and consequences of stochastic processes in development and disease'.

Long-range gene regulation links genomic type 2 diabetes and obesity risk regions to HHEX, SOX4, and IRX3.

Genome-wide association studies identified noncoding SNPs associated with type 2 diabetes and obesity in linkage disequilibrium (LD) blocks encompassing HHEX-IDE and introns of CDKAL1 and FTO [Sladek R, et al. (2007) Nature 445:881-885; Steinthorsdottir V, et al. (2007) Nat. Genet 39:770-775; Frayling TM, et al. (2007) Science 316:889-894]. We show that these LD blocks contain highly conserved noncoding elements and overlap with the genomic regulatory blocks of the transcription factor genes HHEX, SOX4, and IRX3. We report that human highly conserved noncoding elements in LD with the risk SNPs drive expression in endoderm or pancreas in transgenic mice and zebrafish. Both HHEX and SOX4 have recently been implicated in pancreas development and the regulation of insulin secretion, but IRX3 had no prior association with pancreatic function or development. Knockdown of its orthologue in zebrafish, irx3a, increased the number of pancreatic ghrelin-producing epsilon cells and decreased the number of insulin-producing beta-cells and glucagon-producing alpha-cells, thereby suggesting a direct link of pancreatic IRX3 function to both obesity and type 2 diabetes.

Mutations in SOX2 cause anophthalmia.

A submicroscopic deletion containing SOX2 was identified at the 3q breakpoint in a child with t(3;11)(q26.3;p11.2) associated with bilateral anophthalmia. Subsequent SOX2 mutation analysis identified de novo truncating mutations of SOX2 in 4 of 35 (11%) individuals with anophthalmia. Both eyes were affected in all cases with an identified mutation.

Incapacitating the evolutionary capacitor: Hsp90 modulation of disease.

The nature-nurture argument surrounding the mechanisms of disease causation cannot be resolved, as the roles of genes and environment are inextricably entwined. Environmental fluctuation is clearly a major modifier of phenotype, as well as a promoter of evolutionary change. Both types of variability can be mediated by the stress response pathway, with the Hsp90 chaperone family as key components. Hsp90 has been hailed as a capacitor for evolutionary change, because partial inhibition of its functions can uncover cryptic mutations, leading to unexpected phenotypes that, although generally deleterious, will under rare new environmental conditions provide improved survival to the carrier of that variant. There is, therefore, a strong environmentally elicited link between the capacity to reveal hidden variation as human disease phenotype and as novel morphological forms for evolutionary selection.