neCesiDADes De LAs HortALizAs

Clustering in imprinted regions appears to be an important organisational component of the imprinting process. As described in Section 1.5, clusters of imprinted genes are associated with both the BWS and PWS/AS regions. These clusters have several properties in common including interspersion of both maternally and paternally expressed genes and distribution of imprinted genes over a large distance. It is clear from studies of the control of allele-specfic expression that regional, as well as local, mechanisms have a part to play within these clusters.

1.6.3.1 Local control mechanisms

Evidence for local, cw-acting control elements comes from the observation that even relatively small genomic fragments injected artificially into mouse zygotes can become differentially methylated. As described in Section 1.6.3, when Igf2r was introduced on a YAC into transgenic mice, parent-of-origin specific expression was observed suggesting that local signals are sufficient for imprinting (Wutz et al., 1997). Mutation of the critical ‘region 2’ in IgfZr resulted in loss of differential méthylation and allele-specific expression of the transgene. Thus imprinting signals can be mapped by reducing the size of the transgene injected. Ainscough et a/.(1997) used a similar YAC approach to study Igf2-H19

imprinting. They found that although imprinting was conserved in transgenic mice, Jgf2

expression was affected by both position and copy number of the YACs. It is therefore likely that additional, regional control mechanisms are needed to achieve perfect imprinting.

One of the most difficult aspects of imprinting to explain is the presence of closely linked but reciprocally imprinted genes such as H I 9 and Igf2. Bartolomei and Tilghman (1992) have proposed an enhancer competition model in which the two genes compete with each other for access to shared enhancers. Although this model cannot account for some

recent experimental findings (Tilghman, 1999), it is still widely accepted as one possible mechanism for gene silencing.

Recent studies have provided evidence for an alternative mechanism involving reciprocally imprinted antisense transcripts. Antisense RNA transcripts have been detected for both Igf2r (Wutz et a l, 1997) and UBE3A (Rougeulle et a l, 1998) genes. The antisense transcript of Igf2r is thought to be responsible for monoallelic expression of the gene by repressing the paternal allele in cis (Wutz et a l, 1997).

1.6.3.2 Regional control mechanisms

‘Imprinting mutations’ which disrupt imprinting in both the PWS/AS and BWS clusters provide evidence for regional control of imprinted genes. As mentioned in Section 1.5.1, small deletions of an IC within the PWS/AS region can alter in cis the appropriate imprinting of a number of imprinted genes over a 2 Mb domain (Sutcliffe et a l, 1994; Buiting e ta l, 1995). The IC has been mapped to the SN RPN locus (Buiting et a l, 1995; Dittrich et a l, 1996). Immediately upstream of the SN RPN promoter lie a series of alternative 5’ exons that are spliced to exons 2-10 (Dittrich et a l, 1996). In AS, IC mutations map within these upstream BD exons; in PWS, the deletions map further 3’ to

SNRPN exon 1. Based on these findings, Dittrich et a i (1996) have proposed a model for imprint switching during gametogenesis. They suggest that the alternative SN R PN

transcript (the imprintor) acts in cis on the exon 1 region (the imprint-switch initiation site) to bring about the parental switch. In the female germline a trans-acûng matemal-specific factor allows the paternal > maternal switch. In the male germline the maternal > paternal switch occurs, probably by default (Buiting et a l, 1998). The model does not make any predictions about the fate of the paternal chromosome in the male germline and the maternal chromosome in the female germline. These imprints may be retained or erased and reestablished (Ferguson-Smith, 1996).

Imprinting mutations have also been implicated in BWS. Reik et a l (1995) identified patients with normal biparental inheritance of the l i p 15.5 region but altered allelic méthylation of H I9 and IGF2. Translocation breakpoints in this disorder have been mapped to two clusters that do not disrupt the coding sequences of known imprinted genes. One of these results in activation of the repressed IGF2 allele while expression from H19

Chapter L Introduction

remains unaffected (Brown et al., 1996). Distant rearrangements are therefore capable of affecting the imprinting process within this region.

Many features of the PWS/AS and BWS imprinting domains are reminiscent of X chromosome inactivation. The process of X inactivation has recently been reviewed by Goto and Monk (1998). Although in most cells X inactivation is random, in some species, including mice, the paternal allele is preferentially inactivated in extraembryonic cells during early embryogenesis (Takagi and Sasaki, 1975). X inactivation spreads throughout the entire chromosome from a single centre (XIC) in a fashion analogous to the proposed action of ICs in PWS/AS. The process involves the action of XIST (X-inactive specific transcript), a gene within the XIC which is uniquely transcribed from the inactive chromosome alone. X IS T codes for an RNA molecule with several conserved tandem repeats which coats the inactive X chromosome. It is likely that X inactivation and imprinting of autosomal genes share at least some mechanistic components.

1.6.3.3 Implications for other imprinted regions

Control of imprinting is undoubtedly complex and our understanding of the process is still rapidly evolving. It is clear from studies so far that both méthylation and chromatin structure have central roles to play. Imprinted genes tend to be localised in clusters and regulated by ICs. Multiple control mechanisms seem to be involved, some local and others regional. Experimental findings in PWS, AS and BWS have led to several mechanistic models. However, we are still far from being able to incorporate all the existing evidence into one coherent scheme. It is generally assumed that the mechanisms underlying other imprinting disorders such as SRS will eventually prove similar to those described for PWS, AS and BWS. Characterisation of the imprinted regions responsible in these disorders should help to formulate a unified model for the control of imprinting.