Author response: Variable paralog expression underlies phenotype variation
Year: 2022
Type: peer-review
Abstract: Article Figures and data Abstract Editor's evaluation Introduction Results Discussion Materials and methods Data availability References Decision letter Author response Article and author information Metrics Abstract Human faces are variable; we look different from one another. Craniofacial disorders further increase facial variation. To understand craniofacial variation and how it can be buffered, we analyzed the zebrafish mef2ca mutant. When this transcription factor encoding gene is mutated, zebrafish develop dramatically variable craniofacial phenotypes. Years of selective breeding for low and high penetrance of mutant phenotypes produced strains that are either resilient or sensitive to the mef2ca mutation. Here, we compared gene expression between these strains, which revealed that selective breeding enriched for high and low mef2ca paralog expression in the low- and high-penetrance strains, respectively. We found that mef2ca paralog expression is variable in unselected wild-type zebrafish, motivating the hypothesis that heritable variation in paralog expression underlies mutant phenotype severity and variation. In support, mutagenizing the mef2ca paralogs, mef2aa, mef2b, mef2cb, and mef2d demonstrated modular buffering by paralogs. Specifically, some paralogs buffer severity while others buffer variability. We present a novel, mechanistic model for phenotypic variation where variable, vestigial paralog expression buffers development. These studies are a major step forward in understanding the mechanisms of facial variation, including how some genetically resilient individuals can overcome a deleterious mutation. Editor's evaluation In this elegant genetic study, Bailon-Zambrano et al. draw on classical genetic concepts to address the clinically pertinent question of how genetic variants in the same gene can yield wildly different phenotypes in different individuals. Specifically, this work makes significant contributions to our understanding of how a mutant phenotype may be modified by the expression levels of paralogues of the mutant gene. https://doi.org/10.7554/eLife.79247.sa0 Decision letter Reviews on Sciety eLife's review process Introduction Human faces are variable Human craniofacial variation allows us – and even computers – to identify specific individuals. A recent study indicated that human facial structures are more variable than other human anatomical features and that evolution selected for increased variation in craniofacial-associated genomic regions (Sheehan and Nachman, 2014). While human genomic variation contributes to craniofacial phenotypic variation (Liu et al., 2012; Paternoster et al., 2012), facial variation persists among more genetically homogeneous populations such as Finnish people (Sheehan and Nachman, 2014) and even monozygotic twins (Kohn, 1991). Therefore, even among individuals with similar genomes, craniofacial development may remain noisy or sensitive to minor fluctuations, compared with other developmental processes. Phenotype severity and variation can be measured in different ways. Penetrance is the frequency of a phenotype associated with a genotype. Expressivity is different degrees of the same phenotype associated with a genotype. Here, we use both penetrance and average expressivity to measure phenotype severity. We also use the distribution in expressivity to measure phenotypic variation. Furthermore, there are two types of variation. Among-individual variation can be quantified by measuring differences between individuals. Within-individual variation is measured by quantifying departures from symmetry on left versus right sides of an individual known as fluctuating asymmetry (Valen, 1962). It is unknown if among- and within-individual variation are products of the same biological mechanisms (Hallgrimsson et al., 2019). There is evidence to support that these two types of variation are associated (Breuker et al., 2006; Santos et al., 2005); however, other studies indicate that they are independent (Debat et al., 2009; Breno et al., 2011). Further work is needed to resolve this question. Variability in human craniofacial disease In the 1940s, Waddington observed that mutant organisms are often associated with increased phenotypic variation compared to wild types (Waddington, 1942), and more recent studies support this finding (Fish, 2016; Hallgrímsson et al., 2006; Scharloo, 1991; Parsons et al., 2008). Human genetic craniofacial disease phenotypes also appear more variable than normal human facial phenotypes (Ward et al., 2000). For example, facial clefting among monozygotic twins can be incompletely penetrant or unilateral (Trotman et al., 1993; Young et al., 2021), demonstrating both among- and within-individual variation. How the same genetic disease allele can have devastating consequences in some individuals, while other resilient individuals can overcome the deleterious mutation is not well understood (Chen et al., 2016). One human genetic disease that presents variable craniofacial phenotypes is MEF2C haploinsufficiency syndrome. Patients heterozygous for mutations affecting the transcription factor encoding gene MEF2C show variable facial dysmorphologies (Le Meur et al., 2010; Zweier et al., 2010; Gordon et al., 2018; Tonk et al., 2011). To our knowledge, no MEF2C homozygous mutant patients have been identified, likely due to embryonic lethality. Additionally, craniofacial asymmetry is documented in some patients with this disorder (Nowakowska et al., 2010; Novara et al., 2013). Thus, both among- and within-individual variation are present in MEF2C haploinsufficiency syndrome patients. Numerous MEF2C mutant alleles cause this disorder. Therefore, a model system in which the same allele can be studied in many different individuals is needed. Variable buffering by gene family members might underlie among-individual variation During evolution, whole genome duplications produced multiple mef2 genes, or paralogs, in vertebrate genomes. Thus far, no mef2 paralogs are associated with craniofacial development besides mef2c. When whole genome duplications occur, the most likely outcome is the loss of one of the duplicate genes through the accumulation of deleterious mutations and eventual nonfunctionalization (Takahata and Maruyama, 1979; Watterson, 1983). However, sometimes mutations occur in gene regulatory elements partitioning gene expression domains among the new duplicates. This preserves both copies and is called subfunctionalization (Force et al., 1999; Ohno, 2013). Subfunctionalized duplicates may experience only partial loss of expression subdomains (Force et al., 1999), retaining relics of their ancestral expression pattern even if the gene is no longer required for the original function. Compensation, or the ability of paralogs or gene family members to make up for the loss of a gene due to an overlap in function, has long been recognized as a source of genetic robustness (Wagner, 2000; Kirschner and Gerhart, 1998; Gu et al., 2003; Conant and Wagner, 2004; Kamath et al., 2003). Robustness is defined as the ability of a biological system to overcome genetic or environmental perturbation (Diss et al., 2014; de Visser et al., 2003). Buffering lessens the impact of perturbation on a system. Although buffering by paralogs has long been proposed, whether paralogs retain buffering capacity after subfunctionalization has not been sufficiently addressed. We hypothesize that vestigial expression can buffer against loss of another paralog and that variation in vestigial expression underlies phenotypic variation. This hypothesis is supported by work in systems ranging from yeast to human cells demonstrating that paralogs contribute to genetic robustness (Diss et al., 2014; De Kegel and Ryan, 2019; Dandage and Landry, 2019; Li et al., 2010; Hsiao and Vitkup, 2008; Wagner, 2008; Dean et al., 2008). However, whether paralog expression variation underlies phenotypic variation has not been directly tested. Finally, whether variation in paralogous compensation is subject to selection is unknown. We previously found evidence to support this hypothesis in zebrafish. mef2ca single mutants produce dramatically variable craniofacial phenotypes (Nichols et al., 2016). mef2cb is the most closely related mef2ca paralog. However, mef2cb is not required for craniofacial development, and homozygous mutants are viable and indistinguishable from wild types (Hinits et al., 2012). However, we reported that ventral cartilage defects associated with mef2ca mutation become more severe when we removed a single functional copy of mef2cb from mef2ca homozygous mutants (De Laurier et al., 2014). Our vestigial buffering hypothesis predicts that although mef2cb is no longer overtly required for craniofacial development, remaining mef2cb expression in neural crest cells might partially substitute for mef2ca loss. There is further evidence from mouse mutants that Mef2 paralogs can functionally substitute for one another (Majidi et al., 2019). Because mutations in none of the other four zebrafish paralogous mef2 genes have been reported, their function is unknown. These paralogs have human orthologs, MEF2A, MEF2B, and MEF2D. None of these genes have been associated with human craniofacial development or disease. We developed a system for understanding craniofacial development and variability One prominent phenotype in zebrafish mef2ca homozygous mutants is expansion of the opercle, a bone supporting the gill flap. This phenotype is remarkably variable (De Laurier et al., 2014; Kimmel et al., 2003); mef2ca homozygous mutants have many opercle shapes, and some even develop wild-type-looking opercle bones; the phenotype is incompletely penetrant (Nichols et al., 2016). With some genes, mutants encoding a premature termination codon (PTC) upregulate compensating genes through transcriptional adaptation, explaining why some mutant alleles do not produce a phenotype (Rossi et al., 2015; El-Brolosy et al., 2019). Transcriptional adaptation does not contribute to incomplete penetrance in our system (Sucharov et al., 2019). Specifically, we pharmacologically inhibited transcriptional adaptation (by inhibiting RNA decay) and did not observe changes in penetrance. We demonstrated that the factors underlying opercle variation are heritable through selective breeding, which shifted the penetrance of the expanded bone phenotype to generate strains with consistently low or high penetrance of this phenotype (Nichols et al., 2016). Unlike traditional mutagenesis modifier screens (St Johnston, 2002; Kile and Hilton, 2005), our selective breeding paradigm likely enriched for standing genetic variation without the need for further mutagenesis. Thus, we likely bred naturally occurring genetic modifiers in our background to fixation. Similar approaches have proven successful in other systems (Vu et al., 2015; Gasch et al., 2016). mef2ca is pleiotropic, and although we only selected on opercle bone penetrance, nearly all mef2ca-associated phenotypes increased or decreased penetrance in the high- and low-penetrance strains, respectively (Sucharov et al., 2019). The only phenotype remaining fully (100%) penetrant in the low- and high-penetrance strains is a shortened symplectic cartilage, a linear, rod-shaped cartilage that functions as a jaw support structure. In unselected lines, mef2ca-associated phenotypes are only found in mef2ca homozygous mutants (Miller et al., 2007). However, in high-penetrance heterozygotes, we observed the shortened symplectic cartilage, suggesting this phenotype is highly sensitive to mef2ca loss (Sucharov et al., 2019). Penetrance is a binary measurement (a shortened symplectic is present or not), and we have not yet examined expressivity (to what extent a mutant symplectic is shortened). We do not know if the average length of the shortened symplectic or the distribution of symplectic length is different between low- and high-penetrance strains. Here, we capitalize on the strengths of our zebrafish system to address the mechanisms underlying craniofacial variation. We quantified expressivity by taking linear measurements of the symplectic cartilage, allowing us to measure severity, among-individual variation, and within-individual variation in the zebrafish craniofacial skeleton. Combining these measurements with penetrance scoring, we compared phenotype severity and variation between selectively bred strains. We found that increased severity is associated with increased variation. In the high-penetrance strain, we found a more severe symplectic phenotype, and more among-individual variation but not within-individual variation compared to the low-penetrance strain. What factors were selected upon from the original genetic background to increase or decrease penetrance? Work in nematodes demonstrates that expression variation in otherwise redundant genes contributes to variable penetrance in mutants (Raj et al., 2010). Therefore, we hypothesized that one or more of the five zebrafish mef2ca paralogs might be redundant with mef2ca and ameliorate the phenotype in the low-penetrance line or reciprocally intensify the phenotype in the high-penetrance line. We compared the expression of the mef2 paralogs between selectively bred strains and found that many of the paralogs are more highly expressed in the low-penetrance strain compared to the high-penetrance strain. Furthermore, we uncovered standing variation in paralog expression in unselected strains. To determine if the increased paralog expression we discovered in the low-penetrance strain buffers the mef2ca mutant phenotype, we mutagenized the mef2 paralogs. Double mutant analyses indicate that the different paralogs modularly buffer different aspects of the pleiotropic mef2ca mutant phenotype: some affect severity, while others affect variation and some affect both. These findings demonstrate that heritable, variable paralog expression is a major factor affecting phenotype severity and among-individual phenotypic variation but does not contribute to within-individual variation. Results Craniofacial phenotypes are more variable in mef2ca mutants compared with wild types The phenotypic variation often associated with human genetic diseases might be partially due to individuals inheriting different mutant alleles retaining different levels of functional activity, expression domains, and/or transcriptional levels. A strength of the zebrafish system is that variation can be studied in many different individuals with the same mutant allele. Wild-type mef2ca encodes an N-terminal MADS box (MCM1, agamous, deficiens, and SRF) and an adjacent MEF2 domain (Chen et al., 2017; Figure 1A). These highly conserved domains mediate dimerization, DNA binding, and co-factor interactions (Potthoff and Olson, 2007). The mef2cab1086 mutant allele arose from a forward genetic screen and produces a PTC just downstream of the MADS box. This mutant allele variably displays phenotypes including ectopic bone near the opercle, interhyal joint fusion, dysmorphic ceratohyal, reduced Meckel's cartilage, jaw joint fusion, and shortened symplectic (Figure 1B and C). Skeletal preparations from two full-sibling individuals illustrate both among- and within-individual variation easily observed in mef2ca mutants (Figure 1B). Comparing the two mutant individuals (B3 and B4) demonstrates among-individual variation associated with the opercle bone phenotype; one individual (B3) has phenotypically wild-type opercles, while the other individual has bilateral mutant phenotypes (B4). Meanwhile, within-individual (left-right) symplectic cartilage phenotype variation is present in one of the mef2cab1086 animals (B4); in this animal, the symplectic cartilage is longer on one side than the other. Figure 1 with 1 supplement see all Download asset Open asset mef2ca mutant craniofacial phenotypes are more variable than wild types. (A) Schematic of mef2ca exonic structure. The mef2cab1086 mutant allele used in this study and regions encoding known functional domains are annotated. (B1–B4) Zebrafish heterozygous for mef2ca were pairwise intercrossed, and 6 days post fertilization (dpf), larvae were stained with Alcian blue and Alizarin red to label cartilage and bone. The individuals were then genotyped, flat mounted, and imaged. Two examples (upper and lower) are provided for each genotype. The following craniofacial skeletal elements are indicated in a wild-type individual: opercle bone (op), branchiostegal ray (br), Meckel's (mc), ceratohyal (ch), symplectic (sy) cartilages, interhyal (ih), and jaw (jj) joints. Indicated phenotypes associated with mef2ca mutants include: ectopic bone (arrowheads), interhyal and jaw-joint fusions (^), reduced mc (double arrowhead), and a shortened sy (red arrows). Dashed outline indicates symplectic cartilage. Scale bar: 50 μm. (C) The penetrance of mef2ca mutant-associated phenotypes observed in 6 dpf homozygous mutant larvae is indicated. (D) Schematic indicating how the symplectic cartilage length was measured in this study. (E) Symplectic cartilage length was measured from 6 dpf wild type or homozygous mutant larvae. The p-value from a Welch's t-test is indicated (****≤0.0001). F-test value testing for significant differences in variation between genotypes is indicated. (F) Symplectic cartilage length on left and right sides of 6 dpf zebrafish was measured to determine fluctuating asymmetry, or the absolute difference between left and right, for wild type or mutant larvae. The p-value from a Welch's t-test is indicated (****≤0.0001). For box and whisker plots, the box extends from the 25th to 75th percentiles. The line in the middle of the box is plotted at the median, and the bars are minimum and maximum values. For E and F, n=22 for wild types and 44 for mutants. Figure 1—source data 1 Penetrance data and raw symplectic cartilage length measurements. https://cdn.elifesciences.org/articles/79247/elife-79247-fig1-data1-v2.xlsx Download elife-79247-fig1-data1-v2.xlsx To quantify the variation associated with mef2ca mutants, we first scored penetrance of the various mef2ca phenotypes in an unselected AB background (Figure 1C). Based on penetrance, the symplectic cartilage is the craniofacial structure most sensitive to mef2ca loss (Sucharov et al., 2019). Moreover, we found that symplectic length correlates with opercle bone area (Figure 1—figure supplement 1), suggesting the shortened symplectic phenotype is a good proxy for the entire craniofacial complex. We developed a quantitative phenotyping assay where the symplectic length can be rapidly measured in many animals (Figure 1D). We used the total linear symplectic length (left side plus right side) to examine expressivity (Figure 1E). The average symplectic length was significantly shorter in mutants compared to wild types. The among-individual variation was significantly greater in mutants compared to wild types. When we examined within-individual variation by determining the absolute value of the difference between left- and right-symplectic cartilage length, we found that mutants have more within-individual variation than the wild type (Figure 1F). Selective breeding reveals that severity and among-individual variation, but not within-individual variation, is heritable and segregates with penetrance Another strength of the zebrafish system is that penetrance can be altered by selective breeding to better understand the influence of genetic background on severity and variation. An ongoing, long-term selective-breeding experiment in our laboratory derived two strains of zebrafish with consistently low and high penetrance of mef2ca-associated phenotypes in mef2cab1086 homozygous mutants (Nichols et al., 2016; Brooks and Nichols, 2017; Figure 2A). Our previous work examined the phenotypes present in homozygous mutant fish from these strains (Sucharov et al., 2019). Here, we closely examined mef2ca wild types (mef2ca+/+) from these strains. We were surprised to observe that occasionally some mef2ca-associated phenotypes, like shortened symplectic cartilages (19%, n=16) and fused interhyal joints (13%, n=16), were observed in mef2ca+/+ individuals from the high-penetrance strain (Figure 2B). In contrast, we never observed these phenotypes in mef2ca+/+ individuals from the low-penetrance strain. Importantly, these phenotypes are not likely due to general developmental delay because other skeletal structures like pharyngeal teeth are unaffected. Thus, the phenotypes we discovered in high-penetrance mef2ca+/+ individuals are specific to developmental processes associated with mef2ca function. Quantifying the severity of the shortened symplectic cartilage phenotype, we found significant differences between low- and high-penetrance strains for both mef2ca+/+ as well as mef2ca+/-, but homozygous mutants were statistically the same by this measure in the two strains (Figure 2C). Figure 2 Download asset Open asset Selective breeding affects mef2ca-associated phenotype severity and variation in mef2ca wild types, heterozygotes, and homozygous mutants. (A) Selective breeding pedigree illustrating ectopic bone phenotype penetrance inheritance. Dashed boxes indicate families used in this study. Six generations of full-sibling inbreeding produced the animals used here, a more complete pedigree extending back >10 generations can be found in our previous publication (Sucharov et al., 2019). (B) Alcian blue- and Alizarin red-stained animals from the low- and high-penetrance strains were genotyped, and mef2ca homozygous wild types were flat mounted and imaged. The following craniofacial skeletal elements are indicated in a wild-type individual from the low-penetrance strain: opercle bone (op), Meckel's (mc), ceratohyal (ch), symplectic (sy) cartilages, interhyal (ih), and jaw (jj) joints. Phenotypes normally associated with mef2ca homozygous mutants are present in some wild types from the low-penetrance strain including: ih joint fusions (^) and shortened sy (red arrows). Bars indicating sy length are presented to illustrate the shortened symplectic phenotype present in some high-penetrance wild types but not low-penetrance wild types. A stage-appropriate complement of ankylosed of pharyngeal teeth (pt) are present, and normal sized op bones are present in the individual with shortened sy, indicating the phenotypes we discovered in high-penetrance mef2ca+/+ are not due to general delay. Dashed outline indicates symplectic cartilage. Scale bar: 50 μm (C) Symplectic cartilage length was measured from 6 days post fertilization (dpf) larvae from wild types, heterozygotes, and homozygous mutants from both the low- and high-penetrance strains. p-Values from a Dunnet's T3 test are indicated (**≤0.01). (D) The coefficient of variation for symplectic length in all three genotypes from both strains was plotted (E) Table listing F-test values testing for significant differences in variation between strains comparing the same genotype. (F) Symplectic cartilage length on left and right sides of 6 dpf zebrafish was measured to determine fluctuating asymmetry or the absolute difference between left and right for all three genotypes from both strains. (G) Table listing the Dunnett's test for significant differences in fluctuating asymmetry between all three genotypes from both strains. For box and whisker plots, the box extends from the 25th to 75th percentiles. The line in the middle of the box is plotted at the median, and the bars are minimum and maximum values. Figure 2—source data 1 Symplectic cartilage length measurements from selectively bred strains. https://cdn.elifesciences.org/articles/79247/elife-79247-fig2-data1-v2.xlsx Download elife-79247-fig2-data1-v2.xlsx When we examined variation in symplectic cartilage length, we compared like genotypes between strains and observed that among-individual variation was greater in the high-penetrance strain for homozygous wild types and heterozygotes, and there was no significant difference in variation between strains for homozygous mutants (Figure 2D and E). Thus, the genetic background producing more severe phenotypes also exhibited more among-individual variation, even in wild types. Within-individual variation was not significantly different between backgrounds (Figure 2F and G). Six mef2 paralogs in the zebrafish genome share highly conserved amino acid sequences We sought to explore potential mechanisms underlying the differences in severity and variation between strains. Work from budding yeast demonstrates that gene duplication contributes to genetic robustness against null mutations and that the probability of compensation by gene duplicates is correlated with sequence similarity among the gene duplicates (Gu et al., 2003). Therefore, we hypothesized that mef2ca duplicates with high sequence similarity would severity and variation. There are five mef2ca paralogs in the zebrafish To determine which are the most similar to mef2ca and more likely to for we used to the amino acid sequences of mef2 paralogs. While the domain is among different mef2 genes et al., Figure supplement 1), the zebrafish mef2 paralogs each a MADS box and MEF2 domain which are remarkably similar in sequence all paralogs. We found that the paralog with the most sequence similarity to mef2ca is mef2cb (Figure and these genes are the of and of also share similar sequences to each other and the these mef2d and sequences are more There is only one zebrafish for and the other for each was These and are with previous analyses of the of the mef2 family (Chen et al., 2017; et al., et al., 2004; et al., 2019; et al., Thus, these duplicates are for underlying the differences we observe between strains. Figure with 1 supplement see all Download asset Open asset Zebrafish mef2 paralogs highly conserved MADS box (MCM1, agamous, deficiens, and SRF) and MEF2 (A) by the between the different zebrafish mef2 paralogs. The values are which the between the individual amino acid sequences and a (B) sequence reveals high of MADS box and MEF2 domains among all paralogs. These domains are for DNA binding, dimerization, and used for the are in the Materials and related mef2 paralogs share similar expression Our previous and that of demonstrates when post and where neural crest mef2ca functions craniofacial development (De Laurier et al., 2014; et al., 2019; et al., 2007). We hypothesized that mef2ca duplicates with expression similar to mef2ca would severity and variation. To examine paralog expression we on wild-type zebrafish from to (Figure We found that closely related paralogs share similar gene expression For example, mef2ca and mef2cb have both and expression and share an expression mef2d and have a single and expression respectively. Figure with 1 supplement see all Download asset Open asset Wild-type gene expression studies mef2 paralog expression expression and standing paralog expression variation. (A) Wild-type expression of all mef2 paralogs was quantified by at 1 from to of each paralog was to bars are (B) of each paralog in neural crest cells at was by single on the cells four populations as et al., and a cells (C) Wild-type expression of all mef2 paralogs was quantified by at to paralog expression levels between the low- and high-penetrance strains. of each paralog was to indicate significant difference and bars are (D) Wild-type expression of all mef2 paralogs was quantified by at to paralog expression levels between two families from the unselected AB strain. of each paralog was to indicate significant difference and bars are Figure data 1 raw values. Download Figure data 2 raw values. Download
Source: eLife
Institution: University of Colorado Anschutz
Language: English
Subfield: Genetics
Domain: Life Sciences
Open Access status: gold
Estimated APC paid: $2,500
OpenAlex ID: W4312259834