Heterochromatin and gene expression in Drosophila.

INTRODUCTION

The study of heterochromatin and its relationship to gene expression has a long and rich history that began in the early years of Drosophila genetics. In the 1920s and 1930s, the yearly reports from TH Morgan's laboratory described the discovery of new genetic markers and noted that only a few genes mapped to the Y chromosome, the proximal portion of the X chromosome, or areas surrounding the spindle attachment sites on the autosomes. So few functions mapped to these regions, which were later shown to correspond to the deeply staining, heterochromatic portions of the genome, that heterochromatin became known as the "inert" region (107). However, subsequent observations showed that heterochromatin could affect gene expression. A striking demonstration of the repressive effect of heterochromatin on a euchromatic gene was described by Muller in 1930 (85). He recovered several exceptional mutations of the white+ (w+) gene that resulted in variegated eye color phenotypes. In each case, the mutation was associated with a chromosome rearrangement that placed the gene adjacent to heterochromatin (106). Studies of the white-mottled ([w.sup.m]) alleles and other rearrangements that induced mosaic phenotypes outlined the basic properties of the phenomenon that is now known as position effect variegation (PEV) (72, 114). Early investigators showed that breakpoints that created novel junctions between euchromatin and heterochromatin could induce the mosaic inactivation of euchromatic and heterochromatic genes, genes expressed at diverse developmental stages, and those expressed in different tissues. Position effects induced on a euchromatic gene could be relieved by recombining the gene away from the heterochromatin-euchromatin breakpoint or by inducing a second chromosome rearrangement that changed the location of the gene. Especially intriguing was the observation that a single breakpoint could affect more than one gene, with the strength of the effect diminishing with increasing distance from the breakpoint. The severity and extent of variegation was sensitive to the dose of the Y chromosome and other regions of heterochromatin. Thus, simply changing the balance between heterochromatin and euchromatin in the nucleus could influence gene expression in trans.

PEV captured the attention of early geneticists because of the striking phenotypes and because of its potential for understanding chromosome organization and gene expression during cell differentiation (4, 107). The fascination with PEV continues today with the recognition that it provides a genetic approach to investigate how large chromosomal domains are specified and maintained and how chromatin structure influences gene expression (39, 113). This review concentrates on selected aspects of Drosophila heterochromatin and PEV. Certain properties are commonly considered to be defining features of heterochromatin and PEV. We examine the strength of the data upon which these definitions rest and discuss the implications for understanding the regulation of heterochromatic and euchromatic genes. At the time PEV was first discovered and defined, variegation was a novel phenotype. Since then, a number of phenomena that result in mosaic inactivation of genes have been discovered in a variety of organisms. Variegated phenotypes in Drosophila are associated with repeated transposon arrays, dominant trans-inactivation, and position effects induced on transgenes inserted into pericentric heterochromatin, chromosome 4, or telomeric sites. We examine how these phenomena compare to variegated position effects induced in cis by chromosome rearrangements. We refer the reader to other references for reviews of genetic functions and molecular organization of heterochromatin (34, 60), or discussions of the relationships of PEV to homeotic gene regulation (25), transvection effects (50), and gene silencing in yeast (1, 61, 89).

CYTOGENETIC AND MOLECULAR ORGANIZATION OF HETEROCHROMATIN

The definition of heterochromatin is based strictly on morphological criteria. Heitz (45) introduced the term to refer to chromosomal regions that appear as deeply staining, compact bodies throughout the cell cycle, including interphase. These regions are distinguishable at the level of light microscopy from euchromatin, which condenses at metaphase, but appears diffuse at interphase. Thus, the definition of heterochromatin was originally, and remains today, limited by the spatial and temporal resolution of a cytological analysis. Even regions generally considered to be "constitutively" heterochromatic may change over time or vary with tissue type, and the boundaries between heterochromatin and euchromatin may not be precisely defined (60). Despite these limitations, heterochromatin is a useful term because it has important implications for chromosome structure and function. The distinction between heterochromatin and euchromatin reflects differences visible at the level of whole chromosomes or large chromosomal domains in the timing or extent of condensation. Heterochromatic regions are generally located at pericentric or telomeric locations (45). They can aggregate together in certain cell types, and thereby promote the association of nonhomologous chromosomes and influence the three-dimensional organization of chromosomes in the nucleus.

The heterochromatin of D. melanogaster has been characterized more extensively than that of any other organism (34). In a typical diploid cell, 30-35% of the karyotype is heterochromatic. This includes the entire Y chromosome, 40% of the X chromosome, 25% of chromosomes 2 and 3, and over half of chromosome 4. Heterochromatin is highly enriched in repetitive DNAs (77). The distribution of the major types of repetitive DNAs on the cytogenetic map of the heterochromatin has been determined by combining chromosome banding techniques and in situ hybridization with repetitive DNA probes (14, 79, 90). The general picture that emerges from these studies is that heterochromatin is molecularly heterogeneous, with specific types of satellite and middle repetitive sequences preferentially clustered at certain chromosomal locations. Lohe and coworkers (79) have estimated that 70-80% of the heterochromatin in diploid cells can be accounted for by 11 different satellite sequences. The molecular dimensions of the simple sequence arrays are estimated to be within the 100- to 900-kb size range, based on long-range molecular mapping (69). The arrays of satellite DNA are interspersed with (69, 90), and may be interrupted by (78), more complex sequences that include different types of middle repetitive DNAs, some of which have homologies to known transposable elements (TEs). Unlike the single TEs found in the euchromatin, heterochromatic clusters of TE-like DNAs show surprisingly little variation in chromosomal location among natural populations of D. melanogaster. Hence, middle repetitive DNAs may constitute rather stable structural components of the heterochromatin (90). Single copy DNA sequences from the heterochromatin have also been isolated and in several cases, these are known to correspond to functional genes (8, 11, 19, 20, 35, 88, 131).

Heitz (46) provided early evidence that different regions of heterochromatin have distinctive behaviors and morphologies. He noted that the heterochromatin in salivary gland nuclei coalesces into a single chromocenter and forms two morphologically distinguishable types of heterochromatin. termed [Alpha] and [Beta]. The [Alpha] heterochromatin was defined as the small compact body in the middle of the chromocenter. Subsequent in situ hybridization studies have shown that [Alpha] heterochromatin is composed largely of satellite sequences that are greatly underrepresented in salivary gland nuclei relative to other sequences (32). The Y chromosome also appears to be underrepresented as judged by cytological (92) and molecular criteria (134). Heitz (46) defined [Beta] heterochromatin as the chromocentral material that is similar to euchromatin in being capable of "growth" and expansion, but that differs in forming a diffuse meshwork instead of well-structured chromomeres. As suggested by Traverse & Pardue (120), [Beta] heterochromatin is formed from sequences located throughout the heterochromatin, which replicate to high levels, loop out, then aggregate together to form the bulk of the chromocenter. These sequences include middle repetitive and single-copy sequences located in the heterochromatin of the X chromosome and autosomes (20, 131, 133). At least some [Beta] heterochromatic sequences are actively transcribed in salivary gland nuclei, as evidenced by 3H-uridine incorporation (68) and by the detection of RNA transcribed from specific single-copy sequences (8, 19). The functional significance of the differential sequence representation that distinguishes [Alpha] from [Beta] heterochromatin is unknown. Differential representation may be important for regulating the activity of heterochromatic genes (134) and as discussed below, may also be relevant to the mechanisms causing PEV in polytene tissues.

HETEROCHROMATIC GENES

The relationship between heterochromatin and gene expression is often oversimplified to the extent that "heterochromatinization" is frequently equated with gene repression. While it is true that some regions of Drosophila heterochromatin appear to be largely transcriptionally inactive and induce the inactivation of euchromatic genes in chromosome rearrangements, this is not true of all heterochromatic regions (113). As noted above, heterochromatin also contains transcriptionally active sequences, some of which correspond to genes with vital functions (55).

Approximately 40 heterochromatic loci have been identified thus far in D. melanogaster (34). A few of these, including the tandemly arrayed 18+28S rDNAs, and the ABO and DAL regions that interact with specific euchromatic genes, are repetitive in nature. Thirty genes have been identified by mutations that cause defects in fertility, viability, or morphology. The light+ (lt+) gene, an essential gene located in chromosome 2 heterochromatin, was the first protein-encoding heterochromatic gene to be cloned (19, 20). This gene is expressed in a wide variety of cell types. Its organization differs from that typical of a euchromatic gene. The exonic regions of the lt transcription unit are single copy, whereas the flanking regions and largest introns contain a high density of middle repetitive DNAs (19). The repetitive DNAs are heterogeneous and similar to those located in other regions of [Beta] heterochromatin (82). Molecular analyses of other heterochromatic loci, such as the concertina gene that encodes an [Alpha]-like subunit of a G protein (88), the rolled/ERK-A MAP kinase gene (11; SL Zipursky, unpublished information) and the kl-5 gene (35) have similar molecular organizations, with repetitive DNAs located near or within the coding regions. The kl-5 gene, which encodes a testis-specific dynein, is particularly interesting in that it maps to one of the giant Y chromosome loops visible in spermatocyte nuclei (15). The primary transcripts of this and other Y-linked fertility factors are larger than 1 Mb and contain predominantly satellite or middle repetitive sequences (13, 40).

POSITION EFFECT VARIEGATION OF HETEROCHROMATIC GENES

The existence of genetic functions in the heterochromatin raises the question of whether the regulation of some or all of these genes is affected by the surrounding environment. Germline transformation has been used to test this possibility for an isolated 18+28S rDNA coding unit (62). The results showed that a single rDNA unit inserted into euchromatin is actively transcribed and forms a mini-nucleolus in salivary gland nuclei. Therefore, a heterochromatic location is not necessary for high levels of rDNA transcription, at least in polytene nuclei.

Evidence to support the possibility that some heterochromatic genes require heterochromatin for normal expression was first obtained by Schultz for the lt+ gene (108). He demonstrated that inversions that remove the gene from the heterochromatin induce its variegated expression. Subsequent studies have shown that PEV of lt+ is not simply the reverse of PEV of euchromatic genes. Rearrangements that variegate for lt+ have one breakpoint located between the gene and the centromere, and another breakpoint in the distal euchromatin of the X or autosomes (54, 124). Breakpoints on the Y, chromosome 4, or in heterochromatin or proximal euchromatin of chromosomes 2 or 3 were not recovered in screens for lt+ variegating alleles.

The conventional view of PEV, based on studies of euchromatic genes, attributes variegation to changes that initiate at the heterochromatin-euchromatin breakpoint and extend to neighboring regions. The breakpoints of rearrangements that induce lt+ variegation might then be considered extraordinary in having extremely long-range effects that, in some cases, would extend over 5 Mb of DNA (57). A more reasonable explanation proposes that heterochromatin-euchromatin breakpoints cause effects on lt+ by reducing the amount of heterochromatin that surrounds the gene. This idea is consistent with the observation that smaller blocks of displaced heterochromatin result in lower levels of lt+…