Monoallelic Expression

Gene expression is termed "monoallelic" when only one of the two copies of a gene is active, while the other is silent. Monoallelic expression is frequently initiated early in the development of an organism and stably maintained thereafter. The mechanisms involved in monoallelic expression thus touch on central questions about the effects of active and silent chromatin states during development, and how these chromatin states can be inherited. The focus of this article, however, is another potentially important aspect of monoallelic expression — namely, its possible role in certain genetic disorders that do not show standard Mendelian patterns of inheritance. With state-of-the-art methods, researchers are beginning to screen for genes that display monoallelic expression. The hope is that someday they can couple information about these particular genes with genetic mapping, thereby tracking down the genes involved in various non-Mendelian inherited genetic disorders.

Monoallelic expression can be understood in comparison to bi-allelic expression, the more prevalent form of gene expression (Figure 1). In most cases, both alleles of a gene are transcribed; this is known as bi-allelic expression. However, a minority of genes show monoallelic expression. In these cases, only one allele of a gene is expressed. Which of the two alleles is expressed may be determined by the parental origin of the allele (such as in imprinting), or the choice may be random.

The best-known case of monoallelic expression is the widely studied phenomenon of X chromosome inactivation, which occurs early in the development of most female mammals. Recall that females typically have two copies of the X chromosome, whereas males have only one copy, along with a Y chromosome. In females, most of the genes on one X chromosome are inactivated in every cell. As a result, the overall dosage (i.e., level of transcription) of most X-linked genes is equal in males and females. This phenomenon is known as dosage compensation (Erwin & Lee, 2008).

Researchers know that X inactivation takes place early in development, at roughly the 64-to-128-cell (blastocyst) stage in mouse zygotes. Interestingly, the silencing of one chromosome at this stage is random; that is, each cell makes an independent choice of which chromosome to inactivate. However, once this decision is made, all descendants of that cell keep the same pattern. The resulting mosaic-like pattern of expression can be seen in tortoiseshell cats (Figure 2), which are females that are heterozygous for an X-linked coat-color gene. In these cats, each patch of differently colored fur represents the progeny of one embryonic cell. Thus, the cells in any one patch have the same inactive X chromosome, and these cells have remained in the same region throughout multiple cell divisions. Intriguingly, the "identical twin" of such a cat would also have a mosaic-like fur pattern, but the exact pattern would be unique to that cat, dependent upon the random inactivation decision of the founder embryonic cells.

X chromosome inactivation affects nearly all the genes on the X chromosome, and for many years, it was thought to be a unique process. However, about twenty years ago, a similar effect was observed on some autosomes (non-sex chromosomes) as well. Unlike random X inactivation, the silencing of single alleles on autosomes appeared to be dependent upon the parent of origin of a given chromosome. In fact, because there seemed to be a mark or imprint distinguishing one parental allele from the other, the silencing of one allele was termed imprinting. Imprinted genes give rise to a pattern of inheritance in which the parental origin of a mutation is important. For example, some mutations at the imprinted Prader-Willi/Angelman syndrome locus are inherited only from the father, and they may skip several generations of female offspring before reappearing in the progeny of a carrier father.

Aside from imprinting, an increasing number of autosomal genes are now known to show random monoallelic expression as well. These include some immune response genes, such as those for various immunoglobulin receptors and cytokines (Cedar & Bergman, 2008; Kelly & Locksley, 2000). Similarly, in 1994, Chess and colleagues reported something called "allelic exclusion" of olfactory receptor genes in the central nervous system (CNS). This meant that olfactory receptor genes would show expression of only one allele of one gene among many. This finding was exciting because it raised the possibility that other genes in the CNS might also show allele-specific expression. One might argue that olfactory receptors are unusual in that only one allele of one gene is expressed from an entire gene family. However, recent studies indicate that at least 1% of all autosomal genes show monoallelic expression, including genes expressed in the CNS (Gimelbrant et al., 2007; Wang et al., 2007).

So, what do we know about autosomal genes that are expressed from only one allele? Like X inactivation, monoallelic expression is frequently accompanied by differences in DNA methylation and modification of histones (Vu et al., 2006). There is also evidence that, just as for X inactivation, silenced alleles may replicate later in the cell cycle (Gimelbrant et al., 2005). Furthermore, in some cases, monoallelic expression may help control the level of transcription. For example, some cytokines are expressed from only one allele in latent T cells, but after stimulation, when the level of transcription is higher, both alleles may be expressed (Kelly & Locksley, 2000). In such cases, expression is said to be stochastic (i.e., variable), as opposed to the permanent silencing seen with X inactivation. However, autosomal allele-specific silencing is also different from silencing of the entire X chromosome, in that silencing of each inactive allele appears to occur independently of other monoallelically expressed genes on the same chromosome. Despite these discoveries, little more is known about the specific mechanisms that initiate this gene-specific inactivation (Krueger & Morison, 2008).

Twins are usually classified into two main categories: fraternal and identical. These categories also have different clinical names: dizygotic (DZ) and monozygotic (MZ), respectively. DZ twins result from the fertilization of two different eggs, each by different sperm. Thus, genetically speaking, DZ twins are as different as siblings, even though they developed in utero at the same time. In contrast, MZ twins arise from a single fertilized egg. Following fertilization, at some point early in development, the embryo splits in half. Each resulting fetus then develops separately within the mother's uterus. MZ twins, therefore, possess the same chromosomes, and for many physical traits, they show a near-identical phenotype.

There are some fascinating exceptions to this expectation among MZ twins. These include a number of major diseases with a genetic component, such as schizophrenia, bipolar disorder, epilepsy, and type I diabetes. For these disorders, if one twin becomes ill, there is a probability ranging from 40% to 70% that the other twin will not be affected. Because these twins share the same genes, developed in the same uterus, and were most likely raised together, what could be the cause of such a major difference between them? For one, conditions in utero, such as access to oxygen, might have differed subtly between the two developing twins. Furthermore, the twins may have experienced any number of different events that impacted their health following their initial "split," such as infection of only one twin by a particular virus. Yet another possibility is that some critical genes may have been expressed from only one of the two gene copies present in each cell, as described above. For such genes, some critical cells in one twin might express the mutant allele, whereas the same cells in the unaffected twin might express the normal allele.

Thus far, screening for monoallelic expression has been difficult. Most problematically, analysis of expression from a mixture of cells showing random monoallelic expression gives a composite result indistinguishable from bi-allelic expression. Recently, however, researchers have started using genome-wide and transcriptome-wide screening methods successfully. For example, Gimelbrant et al. (2007) utilized human SNP arrays (i.e., SNP "chips") to analyze clonal cell lines derived from B lymphocytes. Their surprising finding was that human monoallelic expression may be several times more common than the previous estimate of approximately 1%. In another study, a screen for differential DNA methylation led to the identification of several genes that are expressed from only one allele (Wang et al., 2007). More recently, a screen for imprinting based upon transcriptome-wide RNA sequencing (RNA-seq) has been reported (Wang et al., 2008). With this technique, after candidate genes are identified by large-scale screening, each gene may be assayed for allele-specific expression by methods such as RNA FISH and allele-specific sequencing of RT-PCR products. For example, a screen for differential DNA methylation has led to the identification of several genes that are expressed from only one allele in mouse clonal neural stem cell lines (Wang et al., 2007) (Figure 3).

Therefore, research has demonstrated that there is a class of autosomal genes that show monoallelic expression. The identified function of some of these genes, coupled with their chromosomal locus and tissue-specific patterns of expression, hints at their possible involvement in various inherited disorders. Examining instances of monoallelic expression and finding connections to the genetic and biochemical bases for these disorders is one of many challenges facing scientists in the years to come.