Twin discordance and disease: not just an environmental cause

Monozygotic (MZ) twins have been used for decades in genetic-epidemiological studies to tease out the relative contributions of genes and the environment to a trait. Phenotypic discordance in MZ twins has traditionally been ascribed to non-shared environmental factors acting after birth, however recent data shows this explanation as being far too simple. We review here other reasons for discordance, including differences in the in utero environment, mosaicism and epigenetics. Epigenetic differences are gaining increasing recognition. While it is clear that in specific cases epigenetic alterations provide a causal factor in disease aetiology, the overall significance of epigenetics in twin discordance as well as its dependence on environmental and genetic factors remains unclear. Epigenomic profiling studies have recently shed more light on the dynamics of temporal methylation change and methylome heritability, yet have not given a definite answer on the relevance to disease due to limitations in establishing causality. This review explores the subject of epigenetics as another component in human phenotypic variability focusing on evidence from MZ twin studies. Phenotypic Variability and Discordance The extent to which phenotypic traits are heritable has been a subject of scientific interest at least since Galton’s classic twin study design[1]. Twins offer a unique means to study inheritance. Monozygotic (MZ) twins arise from a single zygote and have always been thought to inherit identical genomic sequences[1]. Dizygotic (DZ) twins arise from two different zygotes and, just like siblings, inherit on average 50% of identical genomic sequence. To assess the relative contribution of genes to a trait, comparisons are made between MZ and DZ twin concordance, with a greater MZ than DZ concordance rate implicating a role for genetics in determining the trait[1]. Phenotypic discordance between MZ twins has traditionally been ascribed to non-shared environmental exposures. However, recent research highlights this as being a too simplified explanation. In this review, we focus on potential sources of MZ discordance. Environment and Discordance There are a range of early environmental factors that need to be considered as potentially explaining MZ twin discordance. Twinning itself is thought to be a rare malformation and a stochastic event, although there exists evidence for familiality [1-3]. MZ twins occur in about 3.5 in 1000 pregnancies or 4 in 1000 live births[4, 5]. Depending on the time of zygote splitting, MZ twins can be divided into four groups[1]. If the zygote splits within 3 days, the twins are dichorionic and diamniotic (DC DA) (18-36% of all MZ births). If splitting occurs after that time but before the 7 th day, the twins are monochorionic but diamniotic (60-80% of cases)[1, 6]. If division occurs between days 7 and 14, twins are monochorionic and monoamniotic (MC MA). This type of twining accounts for 2-4% of all MZ twins[6]. Conjoined twins arise when splitting happens after the 13 th or 14 th day[1, 6]. All multifoetal pregnancies are more prone to complications (like foetal malnutrition and growth restriction, premature birth), with six times higher mortality rate than in singletons and a shorter average duration of twin pregnancy (35 weeks)[2, 7-15]. Intrauterine growth restriction (IUGR) is a common issue in twin pregnancies affecting 12-47% of all twin pairs[16]. It often leads to discordance in birth weight[16-18] and has been linked with discordance for a range of phenotypes like height, head circumference, intelligence, language comprehension and expression, fine motor performance, balance, coordination, and visual-motor perception [16, 18, 19]. The potential causes of IUGR include genetic predisposition, in-utero crowding, uneven allocation of blastomere and uneven blood supply, as well as placental dysfunction (complications like abruptio placentae, infarcts, stem vessel thrombosis, velamentous insertion of the cord, and single umbilical artery)[16, 17, 19-21]. IUGR is even more pronounced in MC twins where differences in placental sharing and vascularisation lead to occasional unequal blood and nutrient sharing, and, in about 15% MC diamniotic pregnancies, result in twin-to-twin transfusion syndrome (TTTS)[4, 16, 19]. MC twins have a higher incidence of congenital heart disorders and TTTS increases this risk even further[4]. However, even in the absence of twin to twin transfusion, MC twins are seven times more likely to develop congenital heart disease, which usually manifests in one twin only[3-5, 11, 19, 22]. The higher risk nature of multiple pregnancies, their proclivity towards complications and twin-twin competition for maternal resources should increase the probability of a skewed environment affecting the twins in utero[23, 24]. After birth, any non-shared environmental exposure including factors such as diet, smoking, toxin exposure and infection could possibly contribute towards twin discordance [20, 25-31]. Moreover, early phenotypic differences arising in twins could potentially cause shared exposures to have different effects leading to dissimilarity between twins. De Novo Mutations and Genetic Mosaicism It is always assumed that MZ twins are genetically identical, but a wealth of data is accumulating to show that this is not necessarily the case. Mosaicism for de novo mutations, retro-transpositions, indels, duplications and chromosomal rearrangements may play a role in MZ twin discordance[32-44]. The rate for de novo base substitutions has been estimated at about 10 −8 per base pair per generation making some genetic differences between adult twins likely[45]. Postzygotic mutations have been linked with discordance for diseases such as oral-facial-dygital syndrome type 1 or Joubert syndrome[46-49]. Postzygotic point mutations have been found to be the source of MZ twin discordance in Van der Woude syndrome, Darier’s Disease and neurofibromatosis type 1 while mosaicism for chromosomal abnormalities has been implicated in MZ discordance for conditions like Turner syndrome, trisomy 21, trisomy 13, skin pigmentation and sex phenotypes[50, 51]. Postzygotic karyotypic mosaicism caused by faulty mitotic division has also been reported in cases of Ullrich-Turner syndrome[20]. Copy number variants (CNVs) accounting for a major portion of the genome, are highly polymorphic and relatively unstable, with mutation rates 100 to 10 000 times higher than single base substitutions[52]. Phenotypic discordance in MZ twins may in part be caused by de novo mutations of CNVs and CNV mosaicism[32, 34, 36, 53]. Indeed, it has been indicated that de novo CNVs may occur at a rate of 10% per twinning event, however so far studies failed to link CNV mosaicism to any specific case of phenotypic discordance in MZ twins[40, 54-56]. Additionally, unequal exchange of cells during gestation might potentially lead to discordant fetomaternal microchimerism[20]. Developmental Noise and Stochasticity Some variation is inevitable as a result of transcriptional/translational stochasticity entailed by the random movements of molecules and the complexity of their interactions [57-63]. It should be expected that such noise can lead to markedly different effects in identical environmental conditions[64]. The effect of developmental stochasticity might amass in a drift-like fashion and thus be more relevant to discordance in complex polygenic traits like height or weight, which develop over long periods of time[58]. Stochastic events like unequal division of the inner mass cells during twining, unequal allocation of the developmental markers or precursor cells to different somatic lineages have been reported as potential sources of discordance in MZ twins[3, 37]. Slight differences in eye or hair colour as well as fingerprint profiles, cases of mirror twinning (affecting up to 25% of MZ twins) ranging from occipital hair whorls and handedness to situs inversus as well as major malformations could constitute examples of stochastic developmental discordance. Alternatively, they could arise through gene-in utero environment interactions[20]. Stochasticity could lead to differential results from shared environmental exposures in twins. Certain cases of differential allelic expression (DAE), including random monoallelic gene expression, could result in random mosaic expression patterns[65, 66]. These are primarily associated with X-inactivation, however the phenomenon also affects autosomal genes, primarily those with olfactory, pheromone receptor as well as immune functions that undergo allelic exclusion, and can constitute a mechanism for stochastically-driven phenotypic discordance in MZ twins[20, 67-70]. A study by Cheung et al. estimated that about 50% of heterozygous loci are subject to DAE within MZ twin pairs[71]. These inter-allelic differences in expression were similar for at least 30% of heterozygous loci, indicating some genetic control but making stochastic gene expression differences likely [71]. In contrast, a comprehensive whole genome expression experiment conducted by Baranzini et al. (2010) produced different results[72]. Their findings, based on a single pair, indicated that only 1.9% of heterozygous coding loci showed significant evidence for DAE, but out of these, 57% were concordant within between the co-twins, still leaving room for stochastic effects.