The mouse as a model of fundamental concepts related to Turner syndrome

Abstract

Although XO mice do not show many of the overt phenotypic features of Turner syndrome (TS; 45,X or XO), mice and humans share different classes of genes on the X chromosome that are more or less likely to cause TS phenotypes. Based on the evolutionary history of the sex chromosomes, and the pattern of dosage balancing among sex chromosomal and autosomal genes in functional gene networks, it is possible to prioritize types of X genes for study as potential causes of features of TS. For example, X-Y gene pairs are among the most interesting because of the convergent effects of X and Y genes that both are likely to prevent the effects of TS in XX and XY individuals. Many of the high-priority genes are shared by mouse and human X chromosomes, but are easier to study in genetically tractable mouse models. Several mouse models, used primarily for the study of sex differences in physiology and disease, also produce XO mice that can be investigated to understand the effects of X monosomy. Using these models will lead to the identification of specific X genes that make a difference when present in one or two copies. These studies will help to achieve a better appreciation of the contribution of these specific X genes to the syndromic features of TS.

1 ∣. INTRODUCTION

Two ideas guide research on the causes of Turner syndrome (TS). One is that some X genes have significantly different functional effects when they are expressed from one versus two X chromosomes (Zinn & Ross, 1998). The second idea is that one versus two X chromosomes have different epigenetic effects on the rest of the genome, caused by differences in X gene expression and/or effects of the inactive X chromosome in 46,XX but not 45,X (TS) individuals (Alvarez-Nava & Lanes, 2018; Wijchers & Festenstein, 2011; Trolle et al., 2016). The two mechanisms likely operate simultaneously. To approach the first idea, it is important to develop a list of candidate genes that are most likely to have different effects because of their dose in the genome. The list of candidate genes is reduced based on our understanding of mechanisms of dosage compensation in the genome (Veitia, Veyrunes, Bottani, & Birchler, 2015; Disteche, 2016; Raznahan et al., 2018), and because of the sex-specific adjustments of sex chromosome gene dose that are required to keep both XX and XY cells functional. Here, I briefly discuss some issues involved in approaching this problem, and also a group of mouse models that will be useful in testing ideas about the types of X genes that make a difference depending on their genomic dose.

2 ∣. A QUESTION OF BALANCE

The biological problem underlying TS is the loss of one copy of a large chromosome that has evolved to be in balance with the rest of the genome. X genes are embedded in gene networks that are mostly autosomal, in which the X genes drive and are driven by many genes throughout the genome (Arnold, Itoh, & Melamed, 2008; Veitia et al., 2015; Zhang & Oliver, 2007). The expressed dose of X genes relative to their autosomal interacting partners can be critical, so that the dose of each gene is selected to be at a level that is not too high or too low in the network. The dosage relationship is altered, by removal of one copy of the X chromosome in TS, or addition of an extra X chromosome in Klinefelter Syndrome (XXY). Because some X genes can operate properly over a wide range of dosage, changing the copy number of those genes from two to one can have little physiological effect. However, for other X genes (called “dosage-sensitive” genes), the range of acceptable dose is more narrow, so that loss of one copy of those genes pushes those gene pathways outside the normal range of function (Naqvi, Bellott, Lin, & Page, 2018). A major goal of TS research is to find the X genes that underlie the syndromic features of TS, as a prelude for developing therapies to prevent or mitigate the undesirable consequences of TS.

To be sure, the dosage problem is severe. Reduction in copy number, from two to one, of only a small percentage of the genome is sufficient to be lethal (Birchler, Bhadra, Bhadra, & Auger, 2001; Birchler, Fernandez, & Kavi, 2006). Only some genes are dosage sensitive, and larger chromosomal deletions are more likely to alter the dose of enough dosage-sensitive genes to interfere with proper cell function. In humans, monosomy for an entire chromosome is lethal for all chromosomes except the X chromosome, because the rest of the genome cannot adjust for the reduction in dose of some highly dosage-sensitive genes. And X monosomy is reported to be lethal in perhaps 99% of conceptuses (Cockwell, MacKenzie, Youings, & Jacobs, 1991). The special nature of the X chromosome, which allows survival of some TS individuals, is probably related to the fact that the X chromosome is normally present in nature in healthy individuals with one (XY) or two (XX) copies. Unlike autosomes, the X chromosome has evolved some resistance to the deleterious effects of monosomy. This resistance stems from the evolution of mechanisms that compensate for the inequality of dose of the X chromosome in XY males versus XX females. Thus, the conceptualization of TS is strongly related to understanding of the evolution of the sex chromosomes and inherent sex bias in the genome.

3 ∣. SEX CHROMOSOMES EVOLVED TO TOLERATE X MONOSOMY

The sex chromosomes of mammals evolved from an ancient autosomal pair, because of the emergence of a dominant testis-determining gene (Sry) on one of the autosomes (Charlesworth, 1991; Charlesworth, Charlesworth, & Marais, 2005; Graves, 2016; Rice, 1984). The presence of Sry on one chromosome transformed that chromosome into the proto-Y chromosome, handed down asexually in the male lineage from fathers to sons. In early stages of sex chromosome evolution, the proto-X and proto-Y contained many of the same genes, stemming from their previous status as an autosomal pair. But then the proto-Y began to diverge in its sequence from its partner, the proto-X. Accordingly the Y chromosome lost its ability to recombine with the X chromosome, leading to the loss of most of its genes because of asexual inheritance of the Y chromosome (Vallender & Lahn, 2004). As XY cells lost Y genes, more and more X genes were present in a monosomic state in XY cells but in a disomic state in XX cells. Thus, compensatory mechanisms were favored that allowed X genes to operate in an acceptable stoichiometric balance with autosomal genes, both when the X genes were present in two doses in XX cells, and in one dose in XY cells. The compensatory mechanisms include: (a) In nongermline cells, one X chromosome is transcriptionally silenced in XX cells (X inactivation; Disteche, 2016). Thus, XX cells have one active X chromosome, similar to XY cells. The balance of many X and autosomal genes therefore is similar in XX and XY cells. In addition, X genes may be upregulated in their expression so that the X genes are in balance with the autosomal genes that are expressed from two chromosomes (Disteche, 2016). (b) Although the present-day Y chromosome is therefore partly viewed as a massively degraded X chromosome, it participates in dosage compensation because it has retained a small number of highly dosage-sensitive genes that are very similar to genes on the X chromosome (Bellott et al., 2014; Cortez et al., 2014; Hughes & Page, 2015; Graves, 2016; Naqvi et al., 2018; Raznahan et al., 2018). These X-Y gene pairs have unusual features. The X partner gene typically escapes X inactivation, so is expressed higher in XX than XY cells. The Y partner gene, however, has similar functions, and therefore offsets the out-of-balance expression of the X gene. One copy each of the X and Y genes have similar effects to two copies of the X gene. This arithmetic likely explains why XX and XY conceptuses are both not subject to the high mortality of XO conceptuses in humans. The presence of the Y chromosome compensates for the lack of a second X. From this view, studying the role of the Y chromosome should give insight into which X genes are responsible for lethality of X monosomy in humans (see below). (c) Importantly, the autosomes themselves have likely evolved mechanisms that compensate for the imbalance in effects of one versus two X chromosomes, because the autosomes are selected to function properly in both XY and XX cells (Veitia et al., 2015). Autosomal alleles that have some plasticity of function may operate better over a range of X doses in their gene pathways, relative to autosomal alleles that require a narrow range of X dose. In this manner, natural selection may have favored autosomal balance with X genes that works in both XX and XY cells. Relatively little is known about which autosomal genes might be affected. Nevertheless, some autosomal backgrounds in humans support the viability of X-monosomic conceptuses, perhaps because of specific dosage-compensating autosomal alleles, or network mechanisms that are more tolerant of X monosomy.

A central concept of dosage compensation is the idea that sexual equality of cells is advantageous for many cellular processes common to both XY males and XX females. Thus, one set of processes (e.g., X inactivation) is favored in XX cells that make them more like XY cells, and another set of processes (Y chromosome effects) is favored in XY cells that make them more like XX cells. Because XO cells lack both of these XX- and XY-specific compensation mechanisms, they are particularly vulnerable in ways that XX and XY cells are not.

4 ∣. WHERE TO LOOK FOR X GENES CAUSING TURNER SYNDROM

The ideas just discussed point to a subset of X genes that are top candidates for genes causing the syndromic features of TS, because the effects of these X genes in the XO condition are different from XX and XY cells, which both avoid the features of TS. Thus, the problem for XO cells is not just that they are female XX cells that have lost one chromosome, but they are cells that lack a second sex chromosome of either type. This throws the cell into new territory in which gene pathways are no longer sufficiently functional to maintain cell viability. The discussion below recapitulates ideas that have been extensively discussed by other authors (e.g., (Hughes & Page, 2015; Raznahan et al., 2018; Zinn & Ross, 1998).

4.1 ∣. Pseudoautosomal genes

The pseudoautosomal region (PAR) of the X chromosome is identical to the PAR of the Y chromosome. Pairing of the two PAR regions is required for segregation of the two sex chromosomes during meiosis. XO cells have only one PAR, versus two in XX and XY cells. Thus, as has long been recognized, PAR genes are primary candidates for causal genes in TS (Urbach & Benvenisty, 2009). Historically, the equal number of PARs in XX and XY cells suggested equality of expression in XX and XY cells. However, recently PAR genes are found to be expressed lower on average in XX females than XY males (Tukiainen et al., 2017). It appears that X-inactivation spills over onto the PAR, and reduces PAR gene expression in the inactive X chromosome in XX cells. In the absence of X inactivation in XY cells, expression is somewhat higher. The only X gene strongly implicated as a causal gene in TS is SHOX, a PAR gene in humans. Two copies leads to greater stature than one copy (Marchini, Ogata, & Rappold, 2016). The human PAR contains at least 25 genes, compared to two in the mouse (Helena & Morris, 2007; Raudsepp & Chowdhary, 2015). The nearabsence of genes on the mouse PAR genes could contribute to the better viability of mouse XO embryos compared to XO humans.

4.2 ∣. X-Y gene pairs

Outside of the PAR, X genes with a closely related Y partner gene are the most likely to be TS causal genes. These X-Y pairs represent former alleles present on the autosomes that evolved into the sex chromosomes. A number of X-Y gene pairs have been present on the mammalian sex chromosomes either since the emergence of Sry, or since these genes were subsequently added to both X and Y chromosomes at later stages of evolution (Bellott et al., 2014; Cortez et al., 2014; Graves, 2016). The survival of these Y genes on the Y chromosome, in the face of massive degradation of the Y chromosome over evolutionary time, is explained by their importance as balancers of the lower genomic dose of X chromosome genes in XY cells. The X-Y gene pairs are highly dosage sensitive (Raznahan et al., 2018) because of their crucial involvement in fundamental cell processes, including as epigenetic regulators of the rest of the genome (Bellott et al., 2014; Cortez et al., 2014; Naqvi et al., 2018). Evolutionary theory predicts that many of these genes must be present in two doses for cell viability, either as two doses of the X gene, or one dose each of the X and Y gene. Thus, XO cells lacking a second sex chromosome are more vulnerable than either XX or XY cells.

Comparison of the sex chromosomes of humans and mice indicate that although more ancestral X-Y gene pairs survive in humans than mice, some X-Y gene pairs are common in the two species (Bellott et al., 2014). In mice, for example, four of the top X-Y gene pairs are Utx (a.k.a. Kdm6a, X-linked) and Uty (Y-linked), Kdm5c (X) and Kdm5d (Y), Ddx3x and Ddx3y, and Eif2s3x and Eif2s3y (Chen et al., 2012; Lopes et al., 2010; Wolstenholme, Rissman, & Bekiranov, 2012). All are also X-Y gene pairs in humans. Study of these genes in mice offers the chance to discover where in the body dosage of these genes leads to abnormal function, which will be important for hypothesizing about the role of the genes in human tissues related to TS. However, one might also argue that because mice lack some of the syndromic features of TS, it is precisely the X-Y genes pairs found in humans but not in mice that are the highest priority targets for study related to TS. If several X-Y pairs account for the TS phenotypes, however, those found in both species could be important. In any case, comparison of the function of both X and Y partner genes, in mice will help illuminate the degree to which their functions overlap, and if the Y gene is likely to help prevent the TS phenotype. In mice, complete knockout of the X partner gene Utx (Kdm6a) is lethal in embryos at mid-gestation, but the presence of one copy of either Utx or its Y partner Uty prevents or partially prevents lethality (Shpargel, Sengoku, Yokoyama, & Magnuson, 2012). Although Utx and Uty have divergent functions, they overlap in their prevention of lethality, and are among top candidates for contributors to the TS phenotype. Utx is a tumor suppressor in bladder cancer (Kaneko & Li, 2018), so a lower dose in TS individuals may be problematic.

4.3 ∣. Genes escaping X inactivation but lacking a Y partner gene

Recent extensive analysis of the human X chromosome transcriptome indicates that approximately 25% of X genes escape inactivation and are expressed from both X chromosomes in highly diverse tissues of XX females (Balaton, Dixon-McDougall, Peeters, & Brown, 2018; Carrel & Willard, 2005; Tukiainen et al., 2017), many more than the 25 or so X genes that have a Y partner. Expression of escapees is higher in XX than XY cells, although often not twofold higher. The constitutively higher level of expression of several hundred X genes in XX than XO individuals could account for some features of TS. Nevertheless, escape from X inactivation does not prove that the higher expression in XX than XO causes any difference in emergent phenotypes such as lethality or other TS syndromic features. In part, this is because it is unclear if the dose difference exerts any differential effect on other genes in its network, and network compensation mechanisms are thought to dampen the effect of different levels of expression in one part of the gene network (Birchler et al., 2001; Birchler, Riddle, Auger, & Veitia, 2005; Zhang & Oliver, 2007). Indeed, the failure of X inactivation to compensate dosage of the X escapees is explained by the lack of strong selection pressure to compensate dosage of these genes between XX and XY cells. If the X escapees are among the least dosage-sensitive genes, then they should be among the genes least likely to be problematic for TS. In the mouse, a much smaller percentage of X genes, approximately 3%, appears to escape X inactivation (Berletch et al., 2015; Berletch, Yang, & Disteche, 2010; Lopes et al., 2010). Nevertheless, the functional effects of this category of genes can be modeled in mice. If X escapee genes are contributing to TS, then the lower incidence of such genes in mice might help account for the less lethal nature of X monosomy in this species. However, other reasons can also be invoked, for example the low number of genes in the PAR, and smaller number of X-Y genes. To date, at least one X escapee gene lacking a Y partner, O-GlcNAc transferase (Ogt), has been implicated in causing a sex difference phenotype (Howerton & Bale, 2014). The higher expression of Ogt in XX than XY placenta leads to differences in methylation of histones, and creates better protection of females to prenatal stress (Nugent, O'Donnell, Epperson, & Bale, 2018). Although XO mice were not studied, the X monosomy of males would predict that X monosomy in females would be equally problematic.

4.4 ∣. Imprinted X genes

XX and XY cells differ in the parental imprint of X genes, because XY cells inherit an imprint from the mother (XmY), whereas XX cells inherit an imprint from both parents (XmXp). Because one of the X chromosomes is inactivated in each XX cell, XX females are a mosaic of cells with a paternal and maternal X imprint. The effect of the parental imprint is therefore thought to be reduced in XX cells because only half of the cells experience any one pattern of imprint. Because of the singular parental imprint on X genes in TS (XpO or XmO), the difference caused by X gene imprinting between XpO and XmY cells is likely to be more pronounced than the difference of XmXp versus XmY. Different parental origin of the X chromosome is reported to affect neurocognitive phenotypes in TS patients (Skuse et al., 1997), and also in XO mice (Davies et al., 2005; Davies, Isles, Burgoyne, & Wilkinson, 2004, 2006; Davies, Humby, Isles, Burgoyne, & Wilkinson, 2007). The mouse offers the chance to study the imprinting process mechanistically, in a whole animal model. Some authors question the role of imprinting in causing TS phenotypes (Urbach & Benvenisty, 2009; Zinn & Ross, 1998).

4.5 ∣. Other mechanisms

It is conceivable that cells with a large inactive X chromosome differ in their epigenetic status compared to cells lacking the inactive chromosome. In Drosophila, the large heterochromatic Y chromosome alters autosomal gene expression via epigenetic interactions that do not require any expression of Y genes (Lemos, Araripe, & Hartl, 2008; Lemos, Branco, & Hartl, 2010). In mice, the number of X chromosomes appears to regulate the heterochromatic status of autosomal loci, possibly because of an “heterochromatin sink” in which the inactive X chromosome sequesters epigenetic mechanisms favoring heterochromatin (Wijchers & Festenstein, 2011; Wijchers et al., 2010). However, little is known about these kinds of epigenetic effects in mammals. A related idea is that Xist itself, which is expressed in XX but not XY or XO cells, might have direct effects on autosomal genes only in cells with more than one X chromosome. Indeed, reduced expression of Xist in adult XX cells leads to hematologic cancer, although this might be the result of reactivation of the inactive X chromosome, rather than to a trans effect of Xist on the autosomes (Yildirim, Kirby, Brown, Mercier, Sadreyev, Scadden, & Lee, 2013).

5 ∣. FRAMEWORK FOR STUDYING MICE AND OTHER NON-HUMAN MODEL SYSTEMS IN TURNER SYNDROME RESEARCH

When it comes to studies of TS, the mouse has a bad reputation. Some investigators consider the mouse a poor or even useless model of TS, because XO mice often do not share many overt syndromic features typical of TS, including sterility and embryonic lethality (Urbach & Benvenisty, 2009). Is there any reason to study mice to clarify the mechanisms of TS?

Some biomedical scientists may have unreasonably high expectations that the physiological mechanisms at play in an inbred mouse line are likely to be the same as mechanisms in humans. Generally we do not expect mice to be very similar to humans, for virtually any phenotype. Moreover, virtually all mechanisms and phenotypes are quite variable across laboratory mouse strains, so that no one mouse strain can be seen to be representative even of Mus musculus. For example, if one were to study sex differences in insulin resistance in mice, one could choose from mouse strains that show either no sex difference at all, or those showing moderate or pronounced sex differences (Parks et al., 2015). In mice, the viability of XO embryos depends on the parental imprint (Hunt, 1991), and varies with background strain. For example, XO mice are not produced by the XY* cross discussed below when the strain background is inbred C57BL/6, but are produced when the strain background is outbred MF1. The differences in production of XO offspring could be related to viability of fetuses or production of XO zygotes, but understanding the mechanisms in mice helps to approach the study of TS both when mouse strains are similar to or different from humans.

The mouse is a useful model of TS, because it is the only genetically tractable mammalian species in which one can easily produce viable female individuals with one or two X chromosomes. Moreover, it is possible to produce XX, XY, or XO mice with either ovaries or testes, to separate the hormonal and sex chromosome effects that are confounded in all human aneuploidy syndromes (Burgoyne & Arnold, 2016). The functional roles of specific candidate genes can be tested one by one. But, what are the questions that the model can answer? Rather than attempting to replicate all features of TS in a mouse, we can ask about the functional effects of different classes of genes that could cause TS phenotypes, such as X-Y gene pairs, or X genes escaping inactivation, or X genes with different parental imprints, and so on. The answers to these questions in mice strongly condition how we think about the operation of genes in the same classes in humans. The sex chromosomes of mice and humans share important genes, and those may be the genes that can best be tested in mice because of the power of mouse molecular genetics.

6 ∣. RELEVANT MOUSE MODELS

Recent reviews have summarized mouse models in which the numbers of X and Y chromosomes are manipulated (Arnold, 2014, 2017; Burgoyne & Arnold, 2016; Cox, Bonthuis, & Rissman, 2014; Grgurevic & Majdic, 2016). Here, we mention three models, in which the number of X and Y chromosomes is manipulated, either independently of each other or in tandem. In some models, the gonadal sex and type of gonadal hormones are unrelated to sex chromosome complement, allowing better separation of the effects of these two classes of sex-biasing variables. Most of these models have been applied to the study of factors causing sex differences in physiology or disease phenotypes. Two of the models produce XO mice, compared to XX and XY mice (for another XO model, see Probst, Cooper, Cheung, & Justice, 2008; Raznahan et al., 2015; Raznahan, Probst, Palmert, Giedd, & Lerch, 2013).

Four core genotypes.

In this model, the Y chromosome is deleted for the testis-determining gene Sry, making the Y⁻ chromosome (Table 1, Figure 1a; De Vries et al., 2002; Burgoyne & Arnold, 2016). XY⁻ mice have ovaries. In addition, some mice have an Sry transgene on chromosome 3, which ensures the development of testes. Breeding XY⁻[Sry+] gonadal males with XX gonadal females produces four types of progeny; XX and XY⁻ gonadal females, and XX(Sry+) and XY⁻(Sry+) gonadal males. The model is a 2 × 2 comparison to detect effects of gonadal sex (male vs. female) and sex chromosome complement (XX vs. XY), and the interaction of these two factors (Arnold, 2014; Burgoyne & Arnold, 2016). The model can be used to measure virtually any phenotype for which there is a sex difference, to determine if the sex difference is caused by sex hormones or sex chromosomes or both. The model is the first step in establishing that there is an effect of sex chromosome dose, but does not separate effects of X dose (one vs. two, XY vs. XX) and Y dose (zero vs. one, XX vs. XY). The latter question can be answered in conjunction with other models, especially the XY* model (see below). To date, the FCG model has been used to discover sex chromosome effects (i.e., XX not equal to XY) in mouse models of autoimmune disease, metabolism and adiposity, cardiovascular ischemia/reperfusion, aneurysms, pulmonary hypertension, stroke, Alzheimer's disease and longevity, neural tube closure effects, pain, and various behavioral and neural traits (Smith-Bouvier et al., 2008; Chen et al., 2008; Arnold et al., 2016; Arnold, Cassis, Eghbali, Reue, & Sandberg, 2017; Li et al., 2014; Umar et al., 2018; Alsiraj et al., 2017; Broestl et al., 2015; Chen, Watkins, Delot, Reliene, Schiestl, Burgoyne, & Arnold, 2008; Gioiosa, Chen, Watkins, Klanfer, Bryant, Evans, & Arnold, 2008; Corre et al., 2016; Vousden et al., 2018; Cox et al., 2014; Cox, Quinnies, Eschendroeder, Didrick, Eugster, & Rissman, 2015).

TABLE 1.

Genetic composition of mice in three mouse models

Model	Father	Mother	Genotypes Produced	Shorthand	Gonad	NPX	NPY	PAR	Sry
Four core genotypes	XY−(Sry+)	XX	XX	XXF	F	2	0	2	0
			XY−	XYF	F	1	1	2	0
			XX(Sry+)	XXM	M	2	0	2	1
			XY−(Sry+)	XYM	M	1	1	2	1
Comparisons: XX vs. XY (both groups M or F); M vs. F (both groups XX or XY)
XY*	XY*	XX	XY*X	XO + PAR	F	1	0	2	0
			XX	XX	F	2	0	2	0
			XY*	XY	M	1	1	2	1
			XXY*	XXY	M	2	1	2	1
Comparisons: XO vs. XX, XY vs. XXY, XO vs. XY, XX vs. XXY
Cross B	XY−Y*X(Sry+)	XX	XY*X	XOF + PAR	F	1	0	2	0
Burgoyne and Arnold (2016)			XY*X(Sry+)	XOM + PAR	M	1	0	2	1
			XX	XXF	F	2	0	2	0
			XX(Sry+)	XXM	M	2	0	2	1
			XY−	XYF	F	1	1	2	0
			XY−(Sry+)	XYM	M	1	1	2	1
Comparisons: XO vs. XX (both groups either M or F); XO vs. XY (both M and F); XX vs. XY (both groups M or F); M vs. F (for any genotype)

The table shows three different mouse crosses discussed. PAR = pseudoautosomal region; NPX = non-PAR region of the X chromosome; NPY = non-PAR region of the Y chromosome; Sry = testis-determining gene; (Sry+) = Sry transgene. See Burgoyne and Arnold (2016) for further discussion.

FIGURE 1.

Changes in body weight in XX and XY mice. In the four core genotypes model (a), gonadal males (XXM, XYM) weigh more than gonadal females (XXF, XYF) at 75 days of age, week 0. When mice are gonadectomized (GDX) on day 75, the gonadal difference is lost by 4 weeks later, but XX mice gain more weight than XY mice, irrespective of the type of gonad that they previously possessed. In the XY* model (b), the same experiment shows that the greater increase in body weight is caused by the number of X chromosomes, not the Y chromosome. Mice with one X chromosome (XO + PAR group and XY group) have lower body weight than those with two X chromosomes (XX and XXY groups). All mice have a C57BL/6J background. From Chen et al. (2012)

XY* model.

The Y* chromosome has an aberrant PAR region of the Y chromosome that recombines abnormally with the X chromosome to produce four types of progeny (Table 1, Figure 1b; Eicher et al., 1991; Burgoyne, Mahadevaiah, Perry, Palmer, & Ashworth, 1998; Burgoyne & Arnold, 2016). The offspring are nearly equivalent to XO and XX gonadal females, and XY and XXY gonadal males. Thus, when a sex chromosome effect on a phenotype is discovered with the FCG model, the phenotype can be further tested in the XY* model to determine if the effect is caused by the number of X chromosomes (a phenotypic difference between XO and XX, or between XY and XXY), or a difference in the presence/absence of a Y chromosome (XO vs. XY, and XX vs. XXY; for extensive discussion, see Burgoyne & Arnold, 2016). Most of the time, the sex chromosome effects found in the FCG model have been found to be caused by the number of X chromosomes, attesting to the dominant role of X gene dose in sex differences in disease models (Arnold et al., 2016; Chen et al., 2012; Li et al., 2014).

The third cross utilizes XY⁻Y*^X(Sry+) males as fathers (equivalent to “Cross B” of Burgoyne and Arnold (2016). These males have the Y⁻ from the FCG model, and the minute Y*^X chromosome derived from the XY* model (Table 1, Figure 2a), which is similar to an X chromosome with a massive deletion of over 99% of genes, leaving mostly the PAR. When bred with XX females, 14 genotypes are produced (Chen, McClusky, Itoh, Reue, & Arnold, 2013). For present purposes, the most interesting groups are XX, XY⁻, and XY*^X, each genotype with or without the autosomal Sry genes, thus with testes or ovaries (Table 1). The XY*^X genotype is the near-equivalent of XO, so the model allows comparison of XO mice to XX and XY mice, all with the same type of gonads in the same litters. That comparison measures the effects of “adding in” either a second X or Y chromosome to an XO genotype.

FIGURE 2.

Changes in body weight in mice with an MF1 outbred background. (a) In mice of “cross B” (Table 1), after gonadectomy (GDX) at 75 days of age, XO + PAR mice (that had either testes, M, or ovaries, F) gain less weigh than XX or XY mice that have two sex chromosomes. Thus, the second sex chromosome of either type prevent the lower body weight of XO mice. All mice have two PARs, so the difference is caused by non-PAR genes. In the XY* model (b), XO mice and XO + PAR mice have lower body weight after GDX, relative to XX and XY and XXY groups. From Chen, McClusky, Itoh, Reue and Arnold (2013)

7 ∣. EXAMPLES OF MOUSE DISEASE PHENOTYPES IN WHICH X DOSE IS IMPORTANT

These mouse models have been used extensively in the study of X chromosome effects on body weight, metabolism, and adiposity. In most mouse strains, male mice weigh more than female mice. In FCG mice (strain C57BL/6J), this difference is associated with the type of gonad: mice with testes are about 25% larger than mice with ovaries. That finding suggests that the primary cause of the sex difference is the type of gonadal hormones (Chen et al., 2012). To test this idea, the gonadal hormones were removed by gonadectomy (GDX) of adult mice (Figure 1a). Indeed, the sex difference in body weight disappears within about a month after GDX. After that, however, the two groups of mice with two X chromosomes (one group born with testes, the other with ovaries) gained much more weight than XY mice of either gonadal type. By 10 months after GDX, the XX-XY difference in body weight was as large as the difference between gonadal males and females prior to GDX. The large increase in weight of the XX group is attributable to an accumulation of fat in XX mice, more than XY. To test whether the effect of sex chromosomes is caused by X or Y dose, the same experiment was performed in the XY* model (Figure 1b; Chen et al., 2012). Again, before GDX the gonadal males weighed more than the gonadal females. After GDX, mice with two X chromosomes gained weight and fat more than those with one X chromosomes (Figure 1b), and the presence or absence of the Y chromosome had little effect. Among the four genotypes produced by XY* fathers (Figure 1b) are XY*^X daughters, which have one X chromosome and the minute Y*^X chromosome that, despite the name, is not a Y chromosome, and comprises mostly the PAR (Burgoyne & Arnold, 2016). Thus, comparison of XY*^X to XX is of mice with one versus two copies of the non-PAR region of the X chromosome, with each genotype having two PARs (Table 1). These XY*^X females have lower body weight and adiposity than XX mice, indicating that X genes cause the group difference and that metabolism of mice depends on the number of X chromosomes. Importantly, in this phenotype on this strain background (C57BL/6J), the Y chromosome does not prevent the XO phenotype of lower body weight and fat. This model will be useful for dissecting the X genes that cause XO-XX difference.

One advantage of mouse models is that one can control strain background, unlike in humans. If we repeat the above experiments in mice with a different strain background, the results are similar except in one important respect (Chen et al., 2013). XO mice in Cross B (with XY⁻Y*^X(Sry+) fathers) have lower body weight than XX mice (Figure 2a), replicating the result found in C57BL6/J mice (Figure 1b). The same difference is found in the XY* model (Figure 2b). But an important difference in the MF1 strain is that XX and XY mice have similar body weight, and both are heavier than XO mice. In other words, in the MF1 strain background, adding a second sex chromosome of either type prevents the phenotype of XO mice. Similarly, a second sex chromosome of either type influences postnatal growth (Burgoyne, Ojarikre, & Turner, 2002). These results suggest that the second X and Y chromosomes have convergent functions, as we expect based on the lack of TS features in both XX women and XY men. The results point to X-Y genes as the class of factors potentially causing the group differences. Even if these studies do not model the metabolic changes found in XO versus XX versus XY humans (where sex chromosome complement is confounded by major endocrine differences between groups), they offer greater experimental control and the ability to separate the causal effects of sex chromosome and hormonal variables. These studies are leading to the identification of X- and Y-linked genes that affect metabolism and adiposity in mice. Once the genes are identified, it will be important to measure how variation in those specific genes influences metabolism in humans, including TS patients. The translation of findings in mice to humans might best be accomplished when specific genes in mice are found to control relevant disease phenotypes, leading to studies in humans of the effects of the same genes.

In addition to studies of sex chromosome effects on metabolism, the number of X chromosomes has been demonstrated to influence incidence of neural tube closure defects (Chen et al., 2008) and susceptibility to cardiac ischemia/reperfusion injury (Li et al., 2014). Various behaviors are also altered in mice with one versus two X chromosomes, including fear reactivity, (Cox et al., 2015; Isles, Davies, Burrmann, Burgoyne, & Wilkinson, 2004), sexual behavior (Bonthuis, Cox, & Rissman, 2012), and juvenile play behavior (Cox et al., 2015; reviewed by Arnold et al., 2016; Cox et al., 2014).

8 ∣. BRAIN IMAGING

Differences in the brain of TS patients, relative to XX and XY, have been studied with increasing frequency (Raznahan et al., 2010; Raznahan, Lee, Greenstein, Wallace, Blumenthal, Clasen, & Giedd, 2016; Fish et al., 2016; Mankiw et al., 2017; Reardon, Clasen, Giedd, Blumenthal, Lerch, Chakravarty, & Raznahan, 2016; Zhao & Gong, 2017). Because TS individuals have both endocrine and sex chromosomal differences relative to XX females, the effect of TS on the brain and neurocognitive phenotypes is likely to be caused by both factors. In mice, the hormonal and sex chromosome causal factors can be identified and manipulated in isolation of each other. Recent studies have used whole-brain MRI to compare brain structure in mice as a function of sex (male vs. female, with all sex-biasing factors operating), sex chromosome complement (XX vs. XY), and gonadal hormones (ovaries vs. testes; Spring, Lerch, & Henkelman, 2007; Corre et al., 2016; Vousden et al., 2018; Qiu, Germann, Spring, Alm, Vousden, Palmert, & Lerch, 2013; Qiu et al., 2018). These studies establish where in the brain sex chromosomes or gonadal hormones induce sex differences that are found in adults. Morphometric and histological studies of brain sexual dimorphisms have classically focused predominantly on brain regions with large volumetric differences predominantly in the limbic system, which are caused by prenatal and postnatal effects of gonadal hormones (Morris, Jordan, & Breedlove, 2004). In contrast to these studies of a few salient sexual dimorphisms in hormone-sensitive areas of the brain, the use of whole-brain MRI reveals a larger role for sex chromosome complement than previously suspected. For example, using MRI imaging of FCG mice, of 30 brain areas showing differences among the four groups, 14 were influenced by sex chromosomes, versus 20 varying by gonad type (hormonal effects; Corre et al., 2016). The sex chromosome effects are not associated with group differences in the levels of gonadal hormones, and generally are not modified by removal of the gonads before puberty, and thus are not likely caused by group differences in hormones that might confound sex chromosome effects. Moreover, measurement of the developmental emergence of sex differences in brain structure in wild-type XX brain regions demonstrate that brain regions larger in males developed earlier than those that were larger in females (Qiu et al., 2018). Structural MRI has also been applied to XO brains of gonad-intact mice (Raznahan et al., 2013, 2015). Perhaps most interesting is group differences in regional brain volumes that are found both when increasing X dosage in females (XO vs. XX) and males (XY vs. XXY), in the lateral septum, bed nucleus of the stria terminalis, ventral group thalamic nuclei, and periaqueductal gray (Raznahan et al., 2015). These are likely to represent true X dosage effects, and point to the need for future studies to determine which X genes are causing the effects.

9 ∣. SUMMARY

The study of factors causing TS is intimately related to the study of sex as a biological variable in disease models. The two research questions can be addressed using the same mouse models, and both seek to establish the functional effects of differences in copy number of X genes. We do not expect XO mice to have similar phenotypes to 45, X humans, because of differences in the evolution of the mouse and human X chromosomes, differences in gene content of the X chromosome, and differences in the autosomal genome. Nevertheless, there is considerable conservation of gene content on the X chromosome among mammals, so that some of the X gene classes (e.g., X-Y gene pairs) include orthologous genes in mice and humans. Moreover, both humans and mice have the same classes of X genes (dosage sensitive vs. not sensitive, X-Y pairs, X escapees, etc.). The study of mice has and will lead to generalizable conclusions about how each of these classes, and the phenotypes that they control, are influenced by X monosomy in the absence of a Y chromosome. In addition, both sex chromosomes participate in dosage compensation, making euploid humans (XX and XY) more similar to each other than they otherwise would be, and preventing the features of TS. Comparing the balancing roles of the second sex chromosome (X or Y) will help prioritize those genes that cause TS and protect from it, because of their dose. Studying animal models will continue to play a critical role, even when the overt features of TS are not found in the animal models.