Journal list menu

Volume 284, Issue 20 p. 3362-3373
Review Article
Free Access

The background puzzle: how identical mutations in the same gene lead to different disease symptoms

Jan E. Kammenga

Corresponding Author

Jan E. Kammenga

Laboratory of Nematology, Wageningen University, The Netherlands

Correspondence

J. E. Kammenga, Wageningen University, Droevendaalsesteeg 1, 6708 PB, Wageningen, The Netherlands

Fax: +31 317 484254

Tel: +31 317 482998

E-mail: [email protected]

Search for more papers by this author
First published: 08 April 2017
Citations: 60

Abstract

Identical disease-causing mutations can lead to different symptoms in different people. The reason for this has been a puzzling problem for geneticists. Differential penetrance and expressivity of mutations has been observed within individuals with different and similar genetic backgrounds. Attempts have been made to uncover the underlying mechanisms that determine differential phenotypic effects of identical mutations through studies of model organisms. From these studies evidence is accumulating that to understand disease mechanism or predict disease prevalence, an understanding of the influence of genetic background is as important as the putative disease-causing mutations of relatively large effect. This review highlights current insights into phenotypic variation due to gene interactions, epigenetics and stochasticity in model organisms, and discusses their importance for understanding the mutational effect on disease symptoms.

Abbreviations

  • CGV
  • cryptic genetic variants
  • Hsp 90
  • heat shock protein 90
  • MAPK
  • mitogen-activated protein kinase
  • POLG
  • polymerase-γ
  • Rh1
  • rhodopsin-1
  • RNAi
  • RNA interference
  • sd
  • scalloped mutation
  • TGFβ
  • transforming growth factor β
  • Introduction

    A long-standing view is that ‘monogenic’ or Mendelian (named after the founding father of genetics Gregor Mendel) disorders or diseases are caused by a single defective gene (see Glossary). Extensive research in induced single gene mutants of model organisms like mouse, flies and worms has been conducted to understand the underlying molecular mechanism of ‘monogenic’ disease characteristics (Box 1). Next to these induced mutations in model organisms, mutations can occur spontaneously in wild populations. In wild populations single gene mutations within the coding region of genes might have a phenotypic effect, and only in a very few cases do these mutations cause death or disease. The effects of both induced and natural mutations depend on the contribution of allelic heterogeneity influencing variation in disease penetrance and expressivity, i.e. that different alleles at the ‘causative’ gene can result in different disease states. The penetrance of a mutation is the proportion of a population of individuals that express a phenotype caused by the mutant allele. Expressivity measures the extent to which a mutation displays its phenotypic expression.

    Glossary

    Mendelian disease or disorder: a disease or disorder assumed to me controlled by a single locus in an inheritance pattern.

    Transposable element: insertion DNA sequences that can be used to knock out genes.

    Genome editing: genetic engineering in which DNA is inserted, deleted or replaced in the genome of an organism.

    Complementation assay: a relatively simple test for assigning a mutation to a genetic locus.

    Modifier: gene or locus in the genetic background that affects the penetrance off a mutation.

    Penetrance of a mutation: the proportion of a population of individuals that express a phenotype caused by the mutant allele.

    Expressivity: measures the extent to which a mutation displays its associated phenotype.

    Modifier: allelic variant in the genetic background that interacts with a mutation affecting the disease phenotype.

    GWAS: Genome Wide Association Study.

    Box 1. Mutant screens

    Specific mutations made in model organisms have been very valuable for identification of mutations that affect complex phenotypes. These include mutations caused by exposing model organisms like nematodes, flies and mice to mutagenic agents. This is followed by further breeding for the phenotypes of interest, gene mapping and identification. More recently genome editing tools have become available to very precisely study the penetrance and expressivity effects of mutations.

    Different methods are used to generate gene knock-outs (mutations) in different model organisms [ultraviolet radiation (UV) and ethyl methane-sulfonate (EMS)]. Clustered regularly interspaced short palindromic repeats (CRISPR) and CRISPR-associated (Cas) are part of the gene editing tools. Mouse figure taken from Wikimedia, Knockout mouse breeding scheme.

    Only recently our insight of ‘monogenic’ disorders and diseases has shifted. Evidence has emerged showing that variation of phenotypes due to ‘causative’ genes is not the result of allelic variation in the mutated gene. In a recent study, human adults were found to harbour mutations associated with severe Mendelian conditions, although disease symptoms were not shown [1]. Here, the authors studied healthy carriers of disease-causing mutations; 874 defined disease genes in 589 306 genomes were screened and 13 healthy adults with mutations associated with eight severe Mendelian conditions were identified. These findings suggest that incomplete penetrance of Mendelian disease genes can depend on the so-called genetic background, which is defined as the other genes that may interact with the gene of interest and therefore potentially influence the specific phenotype [2]. It should be noted, though, that these studies have not been followed through time to assess their validity.

    The effect of the genetic background has a close analogy with the performance of soccer players in a football match (Fig. 1A). The yellow circles are considered as ‘key’ players. The other team members have a different role in the background, but determine the score of the team. Changing these background players will change the characteristics of the team. Analogously, Fig. 1B depicts a cell with different genes. The yellow genes are key members, and the other genes play a different background role but might affect gene functions. Changing these background genes will change the characteristics of the cell and individual.

    Details are in the caption following the image
    Understanding genetic background effects. The analogy between a soccer team and background genes in a cell. (A) Cartoon of a soccer field with ‘key’ members (in yellow). Grey circles are other players. Yellow and grey players function as a team but the grey players can substantially affect the key players. (B) Cell with key player genes (yellow). Grey circles are background genes determining complex traits. The grey genes can substantially affect the key players.

    A strong background effect was found for mutation in polymerase-γ (POLG) in humans (affecting almost 30% of diseases of mitochondria). More than 200 pathogenic mutations have been reported in POLG, some of which are dominant or recessive. A pathogenic POLG mutation involves an Ala467Thr missense mutation. Many patients carry a homozygous mutation but the disease phenotypes are highly variable, ranging from childhood-onset Alpers–Huttenlocher syndrome (uncommon mitochondrial disease) to neuropathy and deterioration of eye muscles [3]. Another striking case of background effects was presented by Dorman et al. [4] on the development of intestinal polyp formation in mice. The authors reported a huge impact of the genetic background that totally overshadowed the effect of an inactivating mutation in the adenomatous polyposis coli gene (Fig. 2).

    Details are in the caption following the image
    Intestinal adenoma modifiers. Genetic background effects on intestinal polyps in mice. The intestines were fixed in neutral buffered formalin and stained by methylene blue. Mice carrying a non-sense mutation in adenomatous polyposis coli gene at site R850, designated ApcR850X/+ (Min) can develop intestinal adenomas. The two images show the poly counts in the small intestine (parts 1–3) and colon of two different mouse lines each carrying the mutation. Line IL-1300 harbours many polyps while hardly any polyp is seen in line IL-2462. Courtesy of F. Iraqi. Taken with permission from J. E. Kammenga, Inaugural Lecture, Wageningen University, 12 December 2016.

    Classic cases of background effects on mutations of so-called ‘monogenic’ diseases have been reported for cystic fibrosis. Approximately 50% of the European cystic fibrosis patients carry the same disease-causing allele [5, 6]. Nevertheless the disease symptoms are variable due to variation in ‘modifier background genes’ [6]. Likewise, pedigrees carrying a deletion in the TBCE exon (encoding tubulin-specific chaperone E) display either hypoparathyroidism–retardation–dysmorphism syndrome or Kenny–Caffey syndrome. Kenny–Caffey patients have the same symptoms as hypoparathyroidism–retardation–dysmorphism syndrome but in combination with osteosclerosis and recurrent bacterial infections [7]. This pedigree-specific variable phenotypic expression of mutations is likely an effect of a difference in genetic background. In addition to these examples, there are many other inherited diseases in which the same mutation is not, or to a low extent, expressed in the individuals who carry it [8]. Table 1 provides an overview of a number of recent studies of background modifier effects of mutations in ‘causal’ genes in humans and model organisms.

    Table 1. An overview of recent studies illustrating the effect of modifier genes on ‘causal’ disease genes
    ‘Causal’ gene Disease (pathway) Modifier(s) Source
    Human
    POLG Mitochondrial dysfunction mtDNA variants [48]
    KCNQ1 Long QT syndrome 3′UTR SNPs [49] (but see [50])
    HbE/β-thal Thalassemia XmnI locus, rs11886868, and rs766432 [51]
    C12orf65 Leigh syndrome ? [52]
    RPGR X-linked retinitis pigmentosa Minor allele (N) of I393N in IQCB1 and the common allele (R) of R744Q in RPGRIP1L [53]
    RPGR Retinal degeneration CEP290 [54]
    BRCA1 Breast cancer 2 SNPs [55]
    BRCA1 Ovarian cancer 5 SNPs [56]
    BRCA2 Ovarian cancer 6 SNPs [56]
    Model organisms
    SCN1A Dravet syndrome Gabra2 [57]
    HNF1A Maturity onset diabetes of the young type 3 Moda1 [58]
    Dmd Muscular dystrophy LTBP4 [59]
    RAS/MAPK Cancer pathway Amx-2 [18]
    RPGR Retinal degeneration CEP290 [54]
    Atm Cancer Bid [60]

    These examples clearly show that the genetic background is capable of altering the effects of highly penetrant mutations. The penetrance level of mutations determines the severity of genetically inherited diseases and it is of paramount importance to identify and characterize the background modifiers that affect disease outcomes and understand the mechanisms. Here I do not attempt to provide an exhaustive review of mutations affecting human disease phenotypes; rather the review outlines the various modes of phenotypic variation and discusses their functional importance for understanding the mutational effect on complex traits, including disease phenotypes. Cooper et al. [8] comprehensively discussed reduced mutation penetrance resulting from allele dosage, differential allelic expression, copy number variation, and sex and age dependence. The present review provides recent insights regarding variable penetrance resulting from gene interactions, epigenetic changes, environmental factors and stochasticity in model organisms [9] but also humans.

    Single gene mutations depend on multiple background loci: insights from Drosophila and mice

    In a constant environment both penetrance and expressivity of a mutation, M, is the result of M and the interaction between M and the modifier genes in background B, i.e. mutation-by-background (M × B) effect. Attempts have been made to understand the molecular interaction of background effects using model organisms like the fly Drosophila melanogaster. Dworkin et al. [10] found that the genetic background interacted with the allelic effect of a scalloped mutation (sdE3), a mutation known for its reduced wing size. The background effects were partly associated with differentially expressed genes between mutant and wild-type where quantitative transcription level differences correlated with variation for the sd phenotype. Different genes were involved compared with the ones that mediate the effects of the mutant sd allele. They also showed a background dependency of the epistatic interaction between sdE3 and a mutation in the optomotor blind gene. In a follow up study Chari and Dworkin [11] showed that the background effect on such mutational interactions is a general entity of genetic systems. It was found that 74% of all interactions between modifiers and the sdE3 phenotype are background dependent. Interestingly, background loci that affect the mutational phenotype also affected these interactions. Chandler et al. [12] showed that the majority of these interactions were not caused by the variation at the locus interacting with scalloped, thereby excluding the possibility of quantitative non-complementation. The complex nature of M × B was further supported by Lachance et al. [13] who studied background effects on vesiculated, an X-linked gene in D. melanogaster that results in wing defects. They placed a natural variant into a range of backgrounds and quantified penetrance and expressivity of wing defects. They found significant complex interactions that were affected by the genetic background. Focusing on a different phenotype, Chow et al. [14] studied mutations causing retinitis pigmentosa, an inherited, degenerative eye disease that leads to severe vision impairment. The authors used a panel of a few hundred genotypic strains as different genetic backgrounds to find potential genetic modifiers in the background interacting with the mutation. They crossed the rhodopsin-1 Rh1G69D mutation into the strains and investigated the effect of natural variation on the Rh1G69D retinal phenotype. The mutant phenotype was strongly affected by the genetic background. Subsequent mapping identified 10 genes that were supported by gene knock-down testing to be each interacting with the expressivity of Rh1G69D. Interestingly, these candidate genes had human orthologues that previously had not been implicated as retinitis pigmentosa modifiers.

    Together these findings show that background dependency is based on complex interactions between several genomic regions each containing multiple candidate genes. The results for the fly were supported by recent findings in mice. Sittig et al. [15] studied background effects of mutations in genes affecting type-2 diabetes. The vast majority of mutant screens are conducted in a strain called Black-6. Sittig et al. performed a cross of the mutation with multiple different backgrounds after which they analysed disease phenotypes. It was found that, across all backgrounds, phenotypic results were very different and even opposite to the observed effects in Black-6.

    Genome-wide and molecular insights into background interactions: lessons from C. elegans

    Genetic background effects in the model nematode Caenorhabditis elegans have increasingly been explored over the past few years. Vu et al. [16] took a genome-wide approach to investigate background effects on gene perturbation. They used RNA interference (RNAi) knock-downs and mutations to compare phenotypic effects for 1400 genes in two backgrounds (wild-types Bristol N2 and CB4856) of C. elegans. One-fifth of the genes differed in the phenotypic penetrance between these two backgrounds. The phenotypic effect of the background on the severity of both knock-down as well as mutant phenotypes could be predicted from expression level variation of the affected gene. In addition to C. elegans, this effect was also reported in mammalian cells, suggesting it is a common characteristic of genetic networks. They extended their search to four different wild-types, and at the same time limited their phenotypic search to a subset of 35 genes involved in the energy transport chain rather than screening all 1400 genes. It was found that genes with a lower expression had more severe phenotypes when knocked down than genes with a higher expression. A comparable study was conducted by Paaby et al. [17] who reported on RNAi approaches for knocking down genes in different genetic backgrounds of C. elegans affecting worm embryogenesis. They conducted quantitative genetic mapping to study how different background alleles modified the penetrance of embryonic lethality. Each knock-down revealed numerous genetic modifiers of small effect but together having a strong phenotypic effect or penetrance. They identified two different types of background modifiers: (a) non-gene-specific or informational modifiers, which suppress null-mutations, and (b) gene-specific modifiers. They showed that heritable modifier variation among worms accounts for 50% of non-gene-specific modifiers and 50% of gene-specific modifier effects. Whether these findings can be generalized to other species remains to be seen. Paaby et al. [17] based their study on maternally expressed genes by RNAi to induce embryonic lethality in 29 genes. One of the limitations of both the Vu et al. and Paaby et al. studies is the fact that the variation in efficacy of RNAi across strains was not evaluated or experimentally confirmed. Interestingly, Paaby et al. [17] did not find any relation between a lower gene expression explaining more severe phenotypes in mutations as reported by Vu et al. [16]. This might suggest that the effect of the background modifiers found by Paaby et al. [17] depends on variation beyond the target gene. Together these studies show that the effect of mutations in a genetic background likely depends on the expression levels of the perturbed gene and that complex phenotypes are mediated by many interacting background modifiers with low levels of pleiotropy.

    Molecular insight into background effects on a pathway was reported by Schmid et al. [18]. They investigated M × B by searching for modifiers of the oncogenic RAS/mitogen-activated protein kinase (MAPK) signalling pathway in C. elegans. They developed ‘mutation included recombinant inbred lines’ derived from a cross between strain MT2124 that carries the activating let-60 RAS (n1046 gain-of-function) mutation in the Bristol N2 background and the wild-type Hawaiian CB4856 background. let-60(n1046) is a point mutation that changes glycine to glutamine at residue 13, a substitution that is equivalent to some oncogenic RAS mutations in mammalian systems. The effect of let-60(n1046) can be measured by vulval induction. It was shown that the vulval induction effect caused by let-60(n1046) depends strongly on the genetic background and ranges with a factor of 2. Subsequent quantitative trait locus analysis combined with transgenic experiments identified the polymorphic monoamine oxidase A (MAOA) gene amx-2 to be negatively regulating RAS/MAPK signalling. They also showed that a monoamine oxidase A mutation that affects gene activity controls for MAPK activation.

    Background effects: defining cryptic genetic variants

    Background effects have not been widely studied while focusing on individual single-nucleotide variants [19]. A genome-wide analysis of yeast revealed that 75% of yeast gene pairs displayed 170 000 interactions underlying growth of cells [20]. Taylor and Ehrenreich [21] investigated the background effect of mutations by defining cryptic genetic variants (CGV). CGV are ‘standing polymorphisms (single-nucleotide variants) that only show phenotypic effects under atypical conditions, such as when specific genes are mutated, rare combinations of segregating alleles are generated, or the environment markedly changes’ [22]. Taylor and Ehrenreich [21] investigated the background effects of a single mutation on a colony trait in yeast. The colony trait was expressed when the mutation interacted with CGV in six genes. In this case, the mutation unveiled the phenotypes of the cryptic variants by perturbation of the transcriptional silencing of some of the genes. These findings imply that CGV may affect the phenotype by mutual interacting or with the mutation, and that gene regulatory networks play a role in genetic background effects. These results are in line with the complex genetic interactions found between modifiers and mutations in Drosophila and C. elegans although the exact nature of the interaction differs between species because of different methodology and approaches.

    Taylor et al. [22] investigated the role of CGV in yielding similar phenotypes by studying the genetics of 17 different architectures leading to the same colony phenotype (a so-called rough morphology phenotype) in yeast. The rough phenotype resulted from interactions between cryptic variants and de novo mutations in 50% of all cases. The other half did not depend on de novo mutations but solely on interactions between CGV. Interestingly, the authors showed that a large number of the CGV are involved in Ras signalling, which is mainly conserved across species, implying that disease-associated signalling pathways are important in these interactions. The role of CGV in modifying the effect of mutations depends strongly on the chaperone heat shock protein 90 (Hsp90). It was shown that, across different species, Hsp90 activity can lead to buffering and potentiation of the effect of CGV on mutations [23].

    The majority of the research in model organisms aims to minimize background effects rather than understanding

    Although the M × B effect can influence the penetrance and expressivity of mutations, in most studies on genetic background effects in model organisms the emphasis is on minimizing the confounding effects of genetic background rather than understanding it. A major reason for this is the aim to generate reproducible results in a standardized way. Standard test assays using model organisms require the use of a single genetic background in order to increase reproducibility across different labs. Although understandable, hardly any attention has been paid to genetic background effects, thereby overlooking the effect of background modifiers that could enhance or decrease the effect of mutations [10, 12, 24-29]. For mice, The Jackson Laboratory, an institute with a long standing record in mouse genetics, briefly mentions that the genetic background can confound the phenotypic effect of a mutation (https://www.jax.org/research-in-action/why-mouse-genetics), and suggests how to take this into account (correct for it) in experiments involving mice. Another example is in C. elegans research in which almost all (thousands) mutations identified are on the canonical background Bristol N2. The results emanating from these single genotype studies are used for elucidating function of genes. This implicitly ignores the M × B effect because in populations each individual has its own unique genetic background. Improved understanding of how identical mutations in the same gene lead to different phenotypes is important for understanding phenotypic variability in human diseases, including so called ‘monogenic disorders’.

    An experimental set-up required for gaining a mechanistic understanding in human disease pathways associated with monogenic diseases using model organisms of background-dependent modifiers is shown in Fig. 3. Two individuals with different backgrounds, B1 and B2, do not differ in a disease symptom. If these two individuals carry the same ‘disease-causing’ mutation M, B1 might not show a symptom (M × B1) in contrast to B2 (M × B2). Both backgrounds carry the same disease-causing mutation in the same gene but differ in their disease symptoms. The individuals are then crossed and the offspring are genetic recombinants each carrying the mutation. In the case where linkage mapping reveals a single locus associated with variation of the disease symptom, then fine mapping in combination with genetic complementation assays and transgenic analysis or genome editing technologies could lead to the identification and characterization of the causal interacting variant. Transgenic analysis of gene homologues in, for instance, human cell lines can then reveal the causal relation of the identified gene and disease symptoms in humans. In addition to background effects, the phenotypic effects of identical mutations in the same gene in the same genetic background can be subjected to environmental and stochastic influences.

    Details are in the caption following the image
    Schematic representation of identification of background modifiers in model organisms. (A) Backgrounds B1 and B2 (purple) do not differ in their disease symptoms. (B) In a situation that both backgrounds carry the same mutation M (M × B1; M × B2; blue), the disease symptoms differ. After crossing the two different backgrounds, the offspring (recombinant inbred strains) do not display large variation for the B1 × B2 cross. The offspring of M × B1 × B2 display large variation due to background effects (recombination and transgressive segregation). (C) Subsequent genetic mapping of quantitative trait loci (QTL) identifies a significant QTL associated with the disease symptom (large peak in purple) whereas the blue mapping did not detect any significant QTL. Further fine mapping and causal gene identification underlying the QTL can be established using target gene identification by means of CRISPR/CAS-9 homologous recombination.

    Genetic background effects in different environments

    M × B can be caused by gene-by-gene interactions where the effect of a mutant allele depends on the genetic background and the environment. This aspect was further investigated by Marigorta and Gibson [30]. They argued that current modern lifestyle changes in humans unlocked the effect of mutations which has led to the spread of obesity and other diseases, a case of gene-by-environment (G × E) interactions. By performing simulations of a quantitative trait resembling diabetes and myocardial infarction-related disease that are controlled by 2500 polymorphic gene variants they reported G × E to be having a strong prevalent effect. Their results suggest that the genetic control of complex disease is environment dependent and that detection of causal gene variants such as in genome-wide association studies implies ample hidden effects. Kulminski et al. [31] investigated whether environmental change affected genes associated with cholesterol and cardiovascular disease. They reported that the APOE e4 allele and APOB CC genotype affected cardiovascular disease in different environments. They found differential effects of the same APOE and APOB alleles on cardiovascular disease and cholesterol throughout multiple generations. These and the other examples show that changing environments throughout life are likely to play crucial roles in the genetics of health span. Background effects can also depend on the diet. Friedline et al. [32] investigated the metabolic effect of lymphocyte mutations in the non-obese diabetic (NOD) mouse vs wild-type C57BL/6 mouse. NOD is an inbred mouse strain, widely used as a model for studying the pathogenesis of autoimmune, T-cell-mediated type 1 diabetes in humans. Homozygous mice carrying the immunodeficiency (scid) mutation in C57BL/6 are deficient in functional T and B lymphocytes. NOD mice carrying the same mutation were also more insulin sensitive but showed tolerance to diet-related obesity. They also had increased muscle glucose metabolism. These findings highlight the important role of diet and genetic background on lymphocyte signalling in obesity and insulin resistance.

    Stochastic and epigenetic effects

    Differential phenotypes of identical mutations in the same gene have been reported within the same genetic background, for instance in monozygotic twins [33]. It was shown that mutations can have variable penetrance in isogenic model organisms. Burga et al. [34] introduced a model for the incompleteness of mutation penetrance using C. elegans based on two compensatory mechanisms that differ among individuals with the same genetic background and influence the mutant effect. A feedback induction of a duplicate ancestral gene was different across individuals, where an increased level of transcription overshadowed the mutation effect. Second, they reported large variation in the induction of chaperone proteins like Hsp90. Chaperones buffer genetic variation [35], and inherited mutation were less likely to have phenotypic effects. They hypothesized that redundant function is continuously present in genomes to buffer stochastic developmental mistakes [36] and present a model for understanding the causes of altered mutation penetrance. Their results show that individual variation in buffering systems are causal to the phenotypic effect of inherited mutations in each individual. The penetration of similar mutations in the same gene in isogenic individuals was studied by Casanueva et al. [37]. They reported that induction of stress response was able to decrease the mutation penetrance in C. elegans. The stress-induced mutation buffering was found to vary in genetically identical individuals affecting stress signalling. This had ramifications for wild-types, leading to individuals with increased stress tolerance and diminished reproductive output and individuals with decreased stress tolerance and increased reproduction. This would favour survival in unpredictable environments, where ‘bet-hedging’ would diversify risk. These results show how the environment can induce protection against mutations by mutation buffering. It still needs to be investigated whether these mechanisms play a role in complex phenotypes, including disease, in humans or other species.

    Next to the aforementioned mechanisms of differential mutation penetrance in isogenic backgrounds in model organisms, epigenetic changes are important in incomplete penetrance of the same mutations in the same gene in humans. For instance monozygotic twins differ in childhood leukaemia due to divergent breast cancer 1 (BRCA1) methylation [38]. A differential methylation status of SLC6A4, a serotonin transporter, was found in a member of a monozygotic twin pair expressing a bipolar disorder [39]. Grundberg et al. [40] investigated the variation of DNA methylation in adipose tissue from twins, including monozygotic twins, and reported associations to disease-related variants in regulatory parts. In conclusion, epigenetic mechanisms, stochastic effects and a case study of chaperone-mediated buffering illustrate how identical mutations in the same gene in isogenic backgrounds can lead to different phenotypic effects, including disease symptoms. These phenomena are not only specific for monogenic diseases but also for complex polygenic diseases.

    Implications for complex disease phenotypes

    It is becoming increasingly clear that even so-called ‘simple’ Mendelian diseases caused by the same mutation in the same gene can be affected by background modifiers. Figure 4 provides an overview of the different mechanisms that play a role in M × B effects. The fact that the genetic background strongly can affect Mendelian diseases makes the distinction with complex polygenic disease less clear. Indeed, due to background effects, Mendelian diseases might be regarded as complex polygenic diseases. Although this phenomenon is widely recognized, the effect of genetic background is still regarded as an unwanted variation [41]. Each individual person has his or her own unique genetic makeup. Establishing how identical mutations lead to different disease phenotypes is important to predict how severe any inherited genetic diseases will be in each affected individual person. Identification of modifiers of complex diseases comprises only very few cases. A recent and clear example is the identification of modifiers affecting lung disease [6] and the discovery of an interaction of the BBS1 mutation associated with non-Mendelian Bardet–Biedl syndrome with alleles at other BBS loci [42]. Detailed insights into background modifiers affecting mutations was reported for the down-regulation of transforming growth factor β (TGFβ) signalling, which is a therapy for specific types of cancer and fibrosis. Kawasaki et al. [43] asked if genetic variants are involved in reduced TGFβ signalling and what the molecular mechanism was. They detected a polymorphic variant of disintegrin and metalloprotease 17 in mice. The variants differentially regulated TGFβ signalling and affected the severity of Tgfβ1-dependent vascular pathology.

    Details are in the caption following the image
    Schematic overview of phenotypic variation due to gene interactions, epigenetics and stochasticity for the same mutation in a single gene. B1 and B2 represent different genetic backgrounds of healthy individuals with distinctive phenotypic differences (large and small smiley). Mutations in the same gene are depicted by the red lightning bolt. For B1 the mutation is in interaction with background modifiers and leads to a devastative and very harmful phenotype. Due to stochastic or epigenetic effects the impact of the mutation can be severe but not devastative. The same interactions can lead to a moderate effect in a different environment (depicted by the closed box). For B2 the same mutation is in interaction with background modifiers and leads to a moderately harmful phenotype. Due to stochastic or epigenetic effects the impact of the mutation can be devastative. The same interactions can lead to a severe effect in a different environment (depicted by the closed box).

    However, the exploration of modifiers of complex human disorder and disease are not so widespread. The sheer number of potential genetic modifiers, their heterogeneity, and environmental variables are likely to contribute to detection complexity, especially in humans where such complexity is difficult to investigate experimentally. This shows the importance of animal models for detection and confirmation of human modifier genes [44]. The mechanistic insights will facilitate further research into this important field of genetics. At the moment background effects have mainly been studied at the level of DNA sequence mutations [45]. An important translational step would be to assess how these mutations affect protein function. Deep mutational scanning involves high-throughput sequencing and investigate the functional consequences for many protein variants at the same time [46]. Zhu et al. [47] reported that human genetic studies show that mutations in the same gene are able to increase the severity of many complex neuropsychiatric disorders. Although mechanisms have not been elucidated, it was clear that the influence of genetic background can be very pervasive and complex.

    In conclusion, evidence is mounting that the genetic background is as important – if not more important – than the ‘disease-causing’ mutation itself due to gene interactions, epigenetics and stochasticity. Despite the fact that progress has been made in understanding the underlying mechanisms, steps need to be made to provide a more general insight (Box 2).

    Box 2. Outstanding questions

    Can we identify patterns of interactions between the genetic background and mutations? Are there certain ‘rules’ for how modifiers affect the outcome of disease mutations?

    Are these interaction patterns conserved across species, and can we translate the mechanisms discovered in model species to humans?

    Does there exist a classification of disease types that have similar interactions with the genetic background?

    Should we redefine the phenotypic effect of a mutation by stating that the effect of a mutation is not caused by the mutation itself but by the interaction of the mutation and the genetic background?

    Acknowledgement

    The author is grateful for discussions with all members of the C. elegans group.