Modifiers of Terminal Deficiency-Associated Position Effect Variegation in Drosophila

Corresponding author: Gary H. Karpen, The Salk Institute for Biological Studies, 10010 N. Torrey Pines Rd., La Jolla, CA 92037., karpen{at}salk.edu (E-mail)

Communicating editor: R. S. HAWLEY

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Terminal deletions of a Drosophila minichromosome (Dp(1;f)1187) dramatically increase the position effect variegation (PEV) of a yellow⁺ body-color gene located in cis. Such terminal deficiency-associated PEV (TDA-PEV) can be suppressed by the presence of a second minichromosome, a phenomenon termed "trans-suppression." We performed a screen for mutations that modify TDA-PEV and trans-suppression. Seventy suppressors and enhancers of TDA-PEV were identified, but no modifiers of trans-suppression were recovered. Secondary analyses of the effects of these mutations on different PEV types identified 10 mutations that modify only TDA-PEV and 6 mutations that modify TDA-PEV and only one other type of PEV. One mutation, a new allele of Su(var)3-9, affects all forms of PEV, including silencing associated with the insertion of a transgene into telomeric regions (TPE). This Su(var)3-9 allele is the first modifier of PEV to affect TPE and provides a unique link between different types of gene silencing in Drosophila. The remaining mutations affected multiple PEV types, indicating that general PEV modifiers impact TDA-PEV. Modifiers of TDA-PEV may identify proteins that play important roles in general heterochromatin biology, including proteins involved in telomere structure and function and the organization of chromosomes in the interphase nucleus.

THE organization of DNA in eukaryotic nuclei goes far beyond packaging into nucleosomes and chromosomes. Chromosomes in the metazoan interphase nucleus display two types of cytologically and functionally distinct chromatin: heterochromatin and euchromatin. Heterochromatin is late replicating, composed of highly and moderately repetitive sequences (satellite DNA and transposons), and is relatively gene poor. Conversely, euchromatin predominantly replicates in early to mid-S-phase, is composed mostly of unique sequences, and contains the vast majority of mutable genes (GATTI and PIMPINELLI 1992 ; WEILER and WAKIMOTO 1995 ; WALLRATH 1998 ; HENNIG 1999 ).

Cytological and genetic data suggest that the eukaryotic nucleus maintains a reproducible organization during interphase (BRIDGER and BICKMORE 1998 ; LAMOND and EARNSHAW 1998 ; LEITCH 2000 ). For example, interphase chromosomes can be organized into a "Rabl configuration," in which telomeres and centromeres (which are constitutively heterochromatic) are clustered at opposite sides of the nuclear periphery, while the euchromatic portion of the genome is located predominantly in the nuclear lumen (RABL 1885 ; COMINGS 1980 ; MATHOG et al. 1984 ; HOCHSTRASSER et al. 1986 ; FUNABIKI et al. 1993 ). Even in nuclei in which the Rabl configuration is not observed, cytological studies show that telomeres associate with the nuclear lamina in a wide variety of organisms. Furthermore, in some tissues each chromosome, or specific parts of a chromosome, inhabits restricted, unique domains within interphase nuclei (BRIDGER and BICKMORE 1998 ; DIETZEL et al. 1998 ; VISSER et al. 1998 ; ZINK et al. 1998 ).

Genetic and cytological studies provide evidence that the structure of chromosomes and the location of genes along the chromosome, and perhaps within the interphase nucleus, can impact gene expression. One particularly well-studied example of the relationship between nuclear organization, chromosome structure, and gene expression is position effect variegation (PEV). PEV describes the clonal inactivation of a euchromatic gene that has been positioned close to or within heterochromatin (via chromosome aberration or transgene insertion) or silencing of heterochromatic genes that have been positioned in distal parts of chromosome arms (WEILER and WAKIMOTO 1995 ; WALLRATH 1998 ). PEV is sensitive to a large number of genetic modifiers [Mod(var)s], loci known as suppressors or enhancers of variegation [Su(var)s and E(var)s]. Genetic screens for Drosophila Mod(var)s suggest that >100 genes affect PEV associated with chromosome aberrations (GRIGLIATTI 1991 ; REUTER and SPIERER 1992 ).

In addition to being directly affected by chromosome structure, gene expression is also impacted by chromosomal organization in the interphase nucleus. For example, translocation of the normally heterochromatic light (lt) gene to distal euchromatic regions results in lt variegation (WAKIMOTO and HEARN 1990 ). In addition, variegation of the euchromatic brown (bw) gene is enhanced by chromosome rearrangements that move the locus to more proximal positions within the euchromatin and is suppressed by rearrangements that move bw to more distal positions (TALBERT et al. 1994 ). Cytological studies indicate that bw variegation is correlated with increased associations of the gene with centric heterochromatin (CSINK and HENIKOFF 1996 ; DERNBURG et al. 1996 ). Such genetic and cytological observations have led to models suggesting that heterochromatic and euchromatic "domains" exist within the nucleus and that appropriate positioning of a gene within the nucleus and relative to other chromosome regions is required for normal function (WAKIMOTO and HEARN 1990 ; KARPEN 1994 ; HENIKOFF 1997 ; BRIDGER and BICKMORE 1998 ; LAMOND and EARNSHAW 1998 ).

Genes located within subtelomeric regions of yeast (GOTTSCHLING et al. 1990 ; LUSTIG 1998 ; NIMMO et al. 1998 ) and Drosophila (LEVIS et al. 1985 ; KARPEN and SPRADLING 1992 ; WALLRATH and ELGIN 1995 ) frequently undergo silencing. In Saccharomyces cerevisiae, telomeres are clustered at the edge of the nucleus, and mutations in some genes (e.g., Hdf1 or Hdf2) interfere with telomere-induced silencing and telomere length, as well as telomere clustering and association with the nuclear periphery (BOULTON and JACKSON 1998 ; LAROCHE et al. 1998 ). Drosophila telomeres do not contain the tandem, simple repeats observed in most eukaryotes. Instead, Drosophila telomeres are composed of more complex repetitive DNAs, including the retrotransposons TART and Het-A (LEVIS et al. 1993 ; BIESSMANN and MASON 1997 ), and subtelomeric repeats known as telomere-associated sequences (TAS; PARDUE and DEBARYSHE 1999 ). Despite the unusual nature of telomeric DNA in Drosophila, genes inserted into subtelomeric regions are silenced (LEVIS et al. 1985 ; KARPEN and SPRADLING 1992 ; WALLRATH and ELGIN 1995 ). Strikingly, general Mod(var)s in Drosophila fail to alter the silencing associated with transgene insertion into telomeres (known as telomeric position effect, or TPE). Thus, despite their phenotypic similarity, different types of silencing in Drosophila can involve distinct components and mechanisms.

Despite the growing amount of evidence that telomere function, chromosome position, and nuclear organization play key roles in gene expression, surprisingly little is known about the components and mechanisms responsible for nuclear organization in multicellular organisms. In addition, while studies in mammals and yeasts have identified a number of telomeric proteins (COOPER 2000 ; MCEACHERN et al. 2000 ), very few proteins are known to be involved in Drosophila telomere function. Previous studies showed that the yellow (y) PEV associated with minichromosome Dp(1;f)1187 (Dp1187) is enhanced by deletions removing the distal portion of the minichromosome, even when the break is >100 kb distal to the y locus (TOWER et al. 1993 ; DONALDSON and KARPEN 1997 ). This terminal deficiency-associated PEV (TDA-PEV) is suppressed by the presence of a second minichromosome, a phenomenon termed trans-suppression. Trans-suppression of TDA-PEV is altered by structural changes in the trans-suppressing minichromosome, suggesting that TDA-PEV and trans-suppression involve chromosome nuclear organization and/or telomere structure and function (TOWER et al. 1993 ; DONALDSON and KARPEN 1997 ). Here, we report the results of a genetic screen for modifiers of TDA-PEV and trans-suppression. We have identified and characterized specific modifiers of TDA-PEV, as well as general Mod(var)s. This collection should help elucidate the components and mechanisms involved in heterochromatin biology, telomere function, and nuclear organization in Drosophila and their impact on the long-distance regulation of gene expression.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Fly stocks and basic husbandry:
Our standard genetic background is y¹; ry⁵⁰⁶. X^X refers to C(1)RM, y v, which is composed of two X chromosomes attached to a single centromere (LINDSLEY and ZIMM 1992 ). Minichromosomes 878 and 158 have been previously described (LE et al. 1995 ; DONALDSON and KARPEN 1997 ). The bw^D; st stock was supplied by Dr. Steve Henikoff (TALBERT et al. 1994 ). 39C-3, 39C-4, 39C5, and 39C-27 were supplied by Dr. Lori Wallrath (WALLRATH and ELGIN 1995 ). The In(1)w^m4h chromosome was isolated by outcrossing an In(1)w^m4h; Su(var)2-10/SM1 stock (provided by Dr. Gunter Reuter) to our y¹; ry⁵⁰⁶ strain, and the In(1)w^m4; dp¹ bw¹ stock was from the Bloomington Drosophila Stock Center (stock no. 2880, in January 1997). All other mutations are as described in LINDSLEY and ZIMM 1992 or FlyBase (flybase.bio.indiana.edu). Flies were grown on standard corn flour, corn syrup media. All crosses were done at 21°, and all females used in crosses were virgins.

Mutagenesis and screen:
A diagram of the screen is shown in Fig 2. Adult y; ry; 158, ry⁺ males were collected 1–3 days posteclosion, aged for 2 days, and then treated with ethyl methanesulfonate (EMS). Mutagenesis was performed via overnight feeding of EMS in sugar water (12.5 mM EMS in 5% sucrose; ASHBURNER 1990 ). Green food coloring was added to the EMS solution to allow visual tracking of the EMS. EMS-mutagenized males were subsequently crossed en masse to 878, y⁺-carrying females (25 males x 50 females per bottle). Males were transferred to a new set of females after 2 days, while females were transferred to new bottles after 3 days to continue laying eggs. This procedure was repeated twice, such that each group of 25 EMS-treated males produced nine bottles of F₁ progeny (i.e., males were crossed to three sets of females, and each set of females deposited eggs on three sets of bottles). 878, y⁺-carrying females of two different types were used. Standard X/X; 878, y⁺ females were used to isolate dominant autosomal modifier mutations, and X^X/Y; 878, y⁺ virgins were used to isolate either dominant autosomal modifiers or dominant or recessive X-linked modifiers.

View larger version (25K): In this window In a new window Download PPT slide   Figure 1. Terminal deficiency-associated yellow PEV and trans-suppression phenotypes associated with Dp 8-23 minichromosome derivatives. 878, y+ and 158, ry+ are -irradiation derivatives of Dp 8-23 (LE et al. 1995 ). Structures include centric heterochromatin (shaded box), euchromatin (horizontal solid line), yellow locus (open circle), rosy+ P elements (solid circles), and subtelomeric heterochromatin (shaded box). yellow variegation phenotype is shown for adult male abdomens.

View larger version (15K): In this window In a new window Download PPT slide   Figure 2. Schematic describing the screen for modifiers of terminal deficiency-associated position effect variegation and trans-suppression. See text for details. 158, ry+ males were EMS mutagenized and mated en masse to 878, y+-carrying virgin females. Phenotypically yellow+ (and therefore 878, y+-carrying) male progeny were scored for modification of terminal deficiency-associated position effect variegation (TDA-PEV; phenotypically ry- males) or trans-suppression (phenotypically ry+ males).

Male progeny that carried 878, y⁺ (phenotypically y+ ry-) or both 878, y⁺ and 158, ry⁺ (phenotypically y+ ry+) were visually screened for increased or decreased levels of 878 y⁺ expression. A total of 49,554 F₁ males were screened from X/X mothers (from which we recovered dominant autosomal modifiers), and 25,420 F₁ males were screened from X^X/Y mothers (from which X-linked and autosomal mutations were recovered). Male progeny with increased or decreased y⁺ expression were crossed to females (X^X/Y females if the male had come from an X^X/Y mother, X/X females if the male had come from an X/X mother) to determine if the phenotypes were heritable.

Mapping mutations to a chromosome and balancing mutations:
Mutant male progeny from X^X/Y mothers were crossed to X^X/Y females. Mutations transmitted to all male progeny, and to no female progeny, were determined to be on the X chromosome; stocks of these mutations were maintained by crossing X^X/Y females to their mutant/Y; 878, y⁺ brothers. Autosomal mutations were mapped to a chromosome and balanced by standard methods (ASHBURNER 1990 ), using Sp/SM1, Cy, or ry/TM3, Sb stocks (DONALDSON 2000 ). Five additional mutations were identified that were not on the X, second, or third chromosomes and are presumed to have been on the fourth chromosome.

Determining lethality level:
Level of lethality was determined by standard lethality tests (ASHBURNER 1990 ). Five to seven females of genotype Mutant/Balancer were crossed to three to five males of genotype Mutant/Balancer. In the case of a completely viable mutation the number of expected nonbalancer animals should equal one-half the number of balancer animals . Mutations were classified on the basis of the number of nonbalancer animals obtained vs. expected: 0–5%, lethal; 6–50%, semilethal; and 51–75%, subviable (ASHBURNER 1990 ).

Complementation analysis:
All 24-sec chromosome mutations were tested for lethal complementation with each other, as were most of the 21 third chromosome mutations (DONALDSON 2000 ). Five to seven females of genotype Mutant A/Balancer were crossed to three to five males of genotype Mutant B/Balancer. Mutations were classified as noncomplementing if the number of nonbalancer animals was 0–25% of expected. All complementation crosses were done reciprocally (i.e., female Mutant A/Balancer crossed to male Mutant B/Balancer and male Mutant A/Balancer crossed to female Mutant B/Balancer). Crosses were classified as noncomplementing only if results from both crosses were consistent. At least 100 animals were counted for each cross, unless the outcome was obvious from 50 animals (e.g., all the animals were Cy, indicating complete lack of complementation). Note that noncomplementation does not necessarily indicate mutations are in the same gene, since lethality may be due to second-site mutations rather than to the Mod(var) in question. In addition, mutations that complement each other may not necessarily be in different genes, since many of the mutations are not homozygous lethal and may not be expected to induce lethality in combination with each other. Thus, the estimate of the number of genes mutated is approximate.

Tests for modifier effects on w variegation:
Tests for effects on w^m4 PEV utilized two different lines, In(1) w^m4h (w^m4h) and In(1)w^m4 y¹ w^m4; dp¹ bw¹ (w^m4-2880, previously known as line number 2880 from the Bloomington Stock Center). Each mutation was tested for its effect on w^m4 variegation at least twice; most mutations were tested using both alleles w^m4h and w^m4-2880. The 25 X-linked mutant chromosomes originally carried a wild-type copy of the white (w) gene, to be able to score for the presence of 158 (ry+). To test these mutations for effects on w PEV, we recombined the w¹¹¹⁸ allele onto the mutant X chromosomes (DONALDSON 2000 ).

To determine if mutations affect the PEV of a P element inserted into centric heterochromatin, all modifiers were tested for their effect on y¹ w¹¹¹⁸; P[w⁺] 39C-3 (39C-3), a stock in which a w⁺ P-element marker variegates due to its position within the 2L centric heterochromatin (WALLRATH and ELGIN 1995 ). A limited number of modifiers were also tested for their effect on y¹ w¹¹¹⁸; P[w⁺] 39C-4 (39C-4), which contains a w⁺-marked P element inserted into another site in the 2L centric heterochromatin (WALLRATH and ELGIN 1995 ). We observed that the 39C-4 insertion is significantly less sensitive to genetic modification than is 39C-3; therefore, most secondary tests were performed only on 39C-3.

Modifiers were also tested to determine if they affect TPE. These tests utilized lines y¹ w¹¹¹⁸; P[w⁺] 39C-5 (referred to as 39C-5) and y¹ w¹¹¹⁸; P[w⁺] 39C-27 (referred to as 39C-27). The w⁺ genes of lines 39C-5 and 39C-27 variegate due to their insertion near the 2L and 2R telomeres, respectively (WALLRATH and ELGIN 1995 ). For the sake of simplicity, the w^m4h, w^m42880, 39C-3, 39C-4, 39C-5, and 39C-27 chromosomes are described collectively as "w^var."

Two different crossing schemes were used to test the effects of the mutations on w^var, depending on whether the mutation was autosomal or X-linked. For autosomal mutations, five to seven w^var females were crossed to three to five Mutant/Balancer males. w^var, Mutant male progeny were visually scored for suppressed, enhanced, or unchanged variegation relative to control animals. Control crosses in which w^var females were crossed to +/Balancer males (where "+" represents the original, unmutagenized second or third chromosomes) were performed at the same time and on the same food, using virgins from the same collections. Phenotypes of +/Balancer progeny from test and control crosses were directly compared to ensure vial to vial consistency, and +/+ males (from the control vials) were compared to +/Mutant males (from the test vials) to ascertain the effect of the mutation on w^var. Each mutation was tested at least twice for its effect on all types of w^var PEV.

For X-linked mutations three to five y¹ w¹¹¹⁸ mutant/Y males were crossed to five to seven w^var females. y¹ w¹¹¹⁸/Y males were used for control crosses. Female y¹ w¹¹¹⁸ mutant/w^var progeny were scored for dominant suppressed, enhanced, or unchanged w⁺ variegation relative to y¹ w¹¹¹⁸/w^var progeny from control crosses.

Tests for modifier effects on bw^D position effect variegation:
The bw locus is required to produce the pteridine (bright red) eye pigments (LINDSLEY and ZIMM 1992 ). The expression of pteridine eye pigments is much more easily seen in the absence of ommochrome (brown) pigments, which require the function of the cinnabar (cn) and scarlet (st) genes. We assessed the level of bw expression in a cn or st background so small changes in bw expression could be easily observed. As was described for w^var tests (above), control crosses were undertaken in parallel using the original, unmutagenized chromosome.

X-linked modifier effects on bw^D: y¹ Mutant/FM7 females were crossed to FM7/Y; bw^D/+; st/ry males. Mutant/FM7; bw^D/+; st/ry progeny were crossed to bw^D; st males. Male Mutant/Y; bw^D/+; st/st progeny were scored for suppressed, enhanced, or unchanged bw^D variegation relative to control animals.

Second chromosome modifier effects on bw^D: Mutant/SM1, Cy; ry males were crossed to bw^D; st females. Male Mutant/bw^D; ry/st progeny were backcrossed to bw^D; st females. Mutant/bw^D; st progeny were scored for bw^D PEV phenotype relative to control animals.

Third chromosome modifier effects on bw^D: Mutant/TM3, Stubble (Sb) females were crossed to bw^D cn/bw; +/TM6B males. Male bw^D cn/+; Mutant/TM6B, Tubby (Tb) progeny (selected as Sb+, Tb) were crossed to cn; + females. Modification of bw^D PEV was scored in cn/cn bw^D; +/Mutant progeny (selected as Tb+ Hu+).

Tests for modifier effects on a variegating yellow⁺ P-element insertion:
To distinguish true modifiers of TDA-PEV from general modifiers of y PEV or y expression, we examined the effect of TDA-specific modifiers on a y⁺ P element inserted in the centric heterochromatin of the third chromosome (line B79; YAN et al. 2002 ). Experiments to test the effects of autosomal modifiers on B79 PEV were identical to those used to test mutant effects of w^var, except that the B79 chromosome was used instead of the w^var chromosomes. For X-linked modifiers, X^X y v/Y virgins carrying the B79 chromosome were crossed to males that carried the mutation. Male progeny were scored for changes in B79 y⁺ expression relative to progeny from control crosses.

Sequence analysis of Su(var)3-9 and Su(var)2-5 genes:
Genetic data suggested that mutation 1699 could be a new allele of Su(var)3-9 and that mutations 1009, 1097, 1207, and 1545 represented new alleles of Su(var)2-5 (see RESULTS). To confirm the allelism and identify the molecular lesions, PCR products from the mutant and the original, unmutagenized (y¹; ry⁵⁰⁶) chromosomes were sequenced directly on an ABI 3700 automated sequencer in The Salk Institute DNA Sequencing Facility. Base pair and corresponding amino acid changes were identified by comparing control and mutant sequences, using Sequencher 3.0 (see RESULTS).

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Isolation of modifiers of TDA-PEV and trans-suppression:
Previous studies determined that terminal deficiencies of minichromosome Dp 8-23 significantly enhanced the variegation of the yellow (y) locus located on the minichromosome (i.e., 878, y⁺ ; Fig 1; TOWER et al. 1993 ; DONALDSON and KARPEN 1997 ). This TDA-PEV is suppressed by the presence of a second, y-, minichromosome (158, ry⁺) in trans, a phenomenon termed trans-suppression (Fig 1; DONALDSON and KARPEN 1997 ).

The consistency of the 878, y⁺ yellow PEV phenotype allows for sensitive identification of suppressors and enhancers of TDA-PEV (Fig 1). We mutagenized 158, ry⁺-carrying adult males with the chemical mutagen EMS (see MATERIALS AND METHODS) and crossed them to females carrying 878, y⁺ (Fig 2). Potential modifiers of TDA-PEV were identified by screening male progeny that were phenotypically y+ ry- (therefore carrying only 878, y⁺). Potential modifiers of trans-suppression were identified by screening male progeny that were phenotypically y+ ry+ (therefore carrying both 878, y⁺ and 158, ry⁺). A pilot screen was performed to examine the feasibility of the system for identifying modifiers of TDA-PEV and trans-suppression. Ten mutations that affect TDA-PEV were identified (nos. 26, E69, E113, 224, E226, 177, 234, 242, 600, and E646) and were analyzed further (see below).

The success of the pilot screen encouraged us to undertake a larger screen, using an isogenic y¹; ry⁵⁰⁶ strain. Approximately 27,225 F₁ 878, y⁺-carrying males (62,017 chromosomes) were screened to identify 282 suppressors and 159 enhancers, of which 50 suppressors and 1 enhancer of TDA-PEV proved to be stable and heritable. Approximately 25,492 F₁ 878, y⁺/158, ry⁺ males (58,168 chromosomes) were screened to identify 360 mutations that reduced the y⁺ expression of 878, y⁺/158, ry⁺ animals. From these original 360 F₁ mutants, nine lines were identified that carried heritable, mappable presumptive modifiers of trans-suppression.

In total, we obtained 60 Mod(var) mutations that affected TDA-PEV or trans-suppression from the 120,185 chromosomes (52,717 males) examined in the large-scale screen (0.05% per chromosome, Table 1). Twenty-five of the 60 modifier mutations came from 14,571 males produced by X^X/Y-bearing mothers, allowing X-linked Mod(var) mutations to be recovered (see MATERIALS AND METHODS and Fig 2); 20 of these mutations are X-linked. Further analysis was performed on the 10 mutations identified in the pilot screen and on the 60 identified from the large-scale screen (see below).

Testing modifiers of trans-suppression for effects on TDA-PEV:
Mutations specifically altering trans-suppression (878, y⁺/158, ry⁺ animals) would not be expected to affect TDA-PEV (878, y⁺ animals). However, all nine mutations that reduced y⁺ expression in 878, y⁺/158, ry⁺ animals also reduced y expression in 878, y⁺-only siblings. Therefore, no modifiers specifically affecting trans-suppression were recovered from the screen of 25,492 F₁ 878, y⁺/158, ry⁺ progeny, and these mutations are classified as TDA-PEV modifiers in subsequent studies.

Characterization of mutations that modify TDA-PEV:
The 70 modifier mutations were mapped to chromosomes and balanced, using standard mapping and balancing procedures (see MATERIALS AND METHODS). Twenty-four mutations mapped to the second chromosome, 21 to the third chromosome, and 25 to the X chromosome (Table 2). Mutations on the second and third chromosomes were tested to determine if they were homozygous viable and were categorized as lethal, semilethal, subviable, or viable (see MATERIALS AND METHODS). Of 45 autosomal mutations, 18 are homozygous lethal, 15 are semilethal, 2 are subviable, and 10 are viable (Table 2).

All autosomal mutations in the collection were complementation tested to provide an estimate of how many different loci were identified and to identify putative alleles at the same locus. Lethal complementation was examined (see MATERIALS AND METHODS), since the dominant PEV phenotype cannot be used to identify alleles at the same locus; unlinked, nonallelic PEV modifiers often interact additively or synergistically, which in most cases cannot be distinguished from noncomplementation of true alleles. The 24-sec chromosome mutations comprise 17 complementation groups: 1 group of four mutations (1009/1097/1207/1545), 1 group of three mutations (E1047/E1060/E1178), 2 groups of two mutations (1683/1227 and E1672/E646), and 13 single-mutation groups. The 21 third chromosome mutations comprise 20 complementation groups: 1 group of two mutations (234/1551) and 19 single-mutation groups. Note that it is possible that noncomplementation is due to second-site mutations on the chromosomes. Furthermore, complementation for lethality may not mean two mutations are nonallelic, since most of the mutations are not homozygous lethal, and hypomorphs may complement due to the presence of sufficient protein to ensure viability. Thus, the number of complementation groups is likely to be an overestimate of the number of different modifier loci isolated in the screen.

All 25 X-linked modifiers were identified in males; thus they are hemizygous viable; all 25 are also homozygous viable in females. All six X-linked enhancers of TDA-PEV are recessive, whereas 17 of the X-linked suppressors of TDA-PEV are dominant and 2 are recessive.

Secondary screening for effects on other types of PEV:
Previous genetic screens suggest that >100 different loci modify PEV (GRIGLIATTI 1991 ; REUTER and SPIERER 1992 ). Mutations in at least two general Mod(var) genes suppress TDA-PEV: Su(var)2-5 [which encodes Heterochromatin Protein 1 (HP1); JAMES et al. 1989 ; EISSENBERG et al. 1990 ] and Su(var)2-10/PIAS (WUSTMANN et al. 1989 ; HARI et al. 2001 ). Therefore, we examined whether TDA-PEV modifiers affected TDA-PEV only or also impacted other types of PEV.

Mutations were characterized using five different PEV types, which we refer to as w^m4, P(cen), bw^D, P(telo), and P(y⁺) (MATERIALS AND METHODS). The structures of these chromosomes are shown in Fig 3, while the phenotypes for these lines are displayed in Fig 4 (top row, wild type). w^m4 is a large inversion derivative of the X chromosome (Fig 3; TARTOF et al. 1984 ) that causes the w locus to be positioned very close to centric heterochromatin, inducing variegated expression (see MATERIALS AND METHODS). w^m4 and w^m4 alleles have been used in previous screens for modifiers of PEV (REUTER and WOLFF 1981 ; SINCLAIR et al. 1983 ; LOCKE et al. 1988 ; WUSTMANN et al. 1989 ; DORN et al. 1993 ).

View larger version (14K): In this window In a new window Download PPT slide   Figure 3. Diagram of chromosome structures of PEV types used for 2° screening. Secondary screening of modifiers of TDA-PEV was undertaken using five different PEV types. Boxes, heterochromatin; thick solid lines, euchromatin; thin vertical lines, gene locations; inverted triangles, insertions. wm4 is a chromosomal inversion of the X chromosome that moved the w+ locus close to heterochromatin, causing w+ PEV. Arrows represent inversion breakpoints. P(cen) is an insertion of a w+ P element into centric heterochromatin, causing PEV of the w+ transgene. The bwD allele contains a 1.5-Mb insertion of heterochromatin into the bw locus, which causes PEV of the bw+ locus on the homologous chromosome. P(telo) is an insertion of a w+ P element into subtelomeric heterochromatin, which induces PEV of the w+ transgene. P(y+) is an insertion of a y+ SUPor-P element into centric heterochromatin of the third chromosome (B79), causing PEV of the y+ transgene.

View larger version (65K): In this window In a new window Download PPT slide   Figure 4. Representative phenotypes of PEV types used for 2° screening. Phenotypes are shown for wild-type, suppressed, and enhanced PEV. Images shown are typical of observed phenotypes.

Previous studies indicated that PEV can occur when a transgene is positioned within centric heterochromatin (REUTER and WOLFF 1981 ; WUSTMANN et al. 1989 ; DORN et al. 1993 ; WALLRATH and ELGIN 1995 ). While PEV induced by chromosome rearrangements (i.e., w^m4) and P-element insertion are phenotypically similar, experiments to determine whether they are mechanistically similar have produced varied results (WALLRATH and ELGIN 1995 ; SASS and HENIKOFF 1998 ). To address this issue, and to identify mutations that have differential effects on PEV due to inversion vs. P-element centric insertion, secondary testing was also applied to a white⁺ P element inserted in centric heterochromatin [P(cen); lines 39C-3 and 39C-4; WALLRATH and ELGIN 1995; see MATERIALS AND METHODS and Fig 3).

The third PEV type utilized for secondary screening was bw^D. The bw^D allele contains a large (1.5 Mb) insertion of centric heterochromatin within the coding region of the bw locus (PLATERO et al. 1998 ). The insertion eliminates all bw expression from the bw^D chromosome. In addition, the bw^D chromosome causes PEV of the bw⁺ allele on the homologous chromosome. Genetic and cytological data indicate that bw^D PEV is linked to an increased frequency of association of the bw^D allele and its paired bw⁺ homolog with the centric heterochromatin (TALBERT et al. 1994 ; CSINK and HENIKOFF 1996 ; DERNBURG et al. 1996 ). bw^D was used here as a secondary screen to identify mutations potentially involved in the nuclear organization or pairing of chromosomes.

Transgenes located in the subtelomeric regions of the chromosome [termed P(telo) in this study] are subject to position effect (LEVIS et al. 1985 ; WALLRATH and ELGIN 1995 ). Such TPE is phenotypically similar to PEV induced by other means. However, previous analyses failed to identify any Su(var)s or E(var)s that alter TPE (TALBERT et al. 1994 ; WALLRATH and ELGIN 1995 ; CRYDERMAN et al. 1999 ), indicating that the two phenomena are functionally distinct and are likely associated with different sets of proteins. Since TDA-PEV and TPE are both thought to be associated with telomere behavior, a mutation altering both TDA-PEV and TPE may identify a gene involved in telomere function. We used the 39C-5 and 39C-27 telomeric insertions of a w⁺-marked P element to analyze effects of the modifier mutations on TPE (see MATERIALS AND METHODS; WALLRATH and ELGIN 1995 ).

To identify mutations specifically affecting TDA-PEV (and not just y PEV or y expression), 26 mutations were tested for their effect on a strain containing a variegating y⁺ P element [referred to here as P(y⁺)] that is inserted in the centric heterochromatin of chromosome 3 (DOBIE et al. 2001 ; YAN et al. 2002 ).

Secondary screens identify four major groups of mutations:
Sixty-seven of 70 TDA-PEV Mod(var) mutations were determined to enhance, suppress, or have no effect on each type of PEV, by direct comparisons with control animals that lacked the mutations (see MATERIALS AND METHODS). The three remaining mutations are all X-linked and for technical reasons could not be recombined onto the y w chromosome [a necessity for testing the effect of the mutations on w^m4, P(cen), and P(telo) variegation; see MATERIALS AND METHODS].

The results of the secondary screening allowed us to place the mutations into one of four major groups, distinguished by their effects on the different PEV types (Table 2). Group I consists of mutations that affect only TDA-PEV. Group II contains a single mutation that has an effect on all types of PEV tested, including telomeric PEV. Group III includes mutations that affect TDA-PEV and only one other type of PEV. Group IV contains mutations that behave like classic modifiers of PEV; these mutations affect multiple variegating types, but not P(telo). Images of characteristic phenotypes are displayed in Fig 4. Our data indicate that all members of a given complementation group behave similarly with respect to lethality and effect on different PEV types. Thus, subsequent details regarding the different mutation groups include the number of complementation groups in addition to, or instead of, the number of mutations showing a given phenotype.

Ten mutations affect only TDA-PEV:
Of the 67 mutations (comprising at most 59 complementation groups) that have been fully characterized, 10 (8 complementation groups) affect only TDA-PEV and are placed in group I (Table 2). Mutations specific to TDA-PEV were found on the X, second, and third chromosomes and consist of both suppressors and enhancers of TDA-PEV. Three of the second chromosome mutations, all enhancers of TDA-PEV (E1047, E1060, and E1178), are homozygous lethal and comprise a single complementation group. Of the 4 remaining second chromosome group I mutations, 1 is homozygous viable (1038), 1 is semilethal (1310), and 2 are homozygous lethal (1429 and 1657). Of the mutations on the third chromosome, 1, 1025, is homozygous viable, while the other, 1650, is homozygous semilethal. The X-linked mutation E1350 is homozygous viable; E1350 has not been tested for its effect on P(y⁺) and thus could eventually be characterized as a group III mutation.

One novel modifier of TDA-PEV also affects telomeric position effect:
Group II consists of a single mutation, 1699, which suppresses TPE of both P(telo) lines tested. This homozygous viable third chromosome mutation also suppresses all the other types of PEV examined in the secondary tests and is the first example of a general modifier of PEV that can affect TPE (TALBERT et al. 1994 ; WALLRATH and ELGIN 1995 ; CRYDERMAN et al. 1999 ).

Analysis of 1699 provided clues regarding its identity. For example, 1699 has a particularly strong effect on w^m4, increasing w expression to nearly wild-type levels (data not shown). Recombination mapping of 1699 localized it to the genetic map region 3-56, very close to the location of a previously identified, extremely strong modifier of PEV, Su(var)3-9 (TSCHIERSCH et al. 1994 ). Sequence analysis of 1699 (see MATERIALS AND METHODS) identified a mutation in the Su(var)3-9 gene, a T to A change in the first position of the codon encoding amino acid 456. This cysteine to serine change is in a residue that is completely conserved among Su(var)3-9 homologs, including clr4 in Schizosaccharomyces pombe (IVANOVA et al. 1998 ), suv39h1 and suv39h2 in the mouse (AAGAARD et al. 1999 ; O'CARROLL et al. 2000 ), and SUV39H1 in humans (AAGAARD et al. 1999 ). These proteins have been demonstrated to encode a histone methyltransferase (see DISCUSSION).

Five mutations affect TDA-PEV and only one additional type of PEV:
The six group III mutations have effects on TDA-PEV and only one other type of PEV. Lines E69 and E113 affect only TDA-PEV and bw^D PEV. These mutations are the first identified modifiers of bw^D PEV that are not also general modifiers of PEV. E69 has a particularly unusual phenotype: although it enhances TDA-PEV, it suppresses bw^D PEV. Such differential modification of PEV types is a unique behavior for a PEV modifier; previously described mutations act either as suppressors or as enhancers of PEV in general. Line 1309 affects only TDA-PEV and w^m4, while E1451 affects only TDA-PEV and P(cen). Finally, 26 and E226 affect only TDA-PEV and the y⁺ centric P insertion; these mutations may affect y⁺ expression, rather than PEV per se. Strikingly, all six mutations in group III are on the X chromosome. These mutations comprise a unique group that may provide insight into the mechanisms of and overlap between different types of PEV.

Mutations affecting position effect variegation in general:
The remaining 50 mutations affect at least two PEV types in addition to TDA-PEV, indicating that they play a general role in position effect variegation (see Table 2). These 50 mutations comprise 44 complementation groups; 41 are suppressors and 3 are enhancers. In every case TDA-PEV suppressors also suppressed the other types of PEV, and enhancers of TDA-PEV behaved only as enhancers of other types of PEV. Group IV mutations are subdivided into subgroups A, B, and C on the basis of their effects on bw^D, w^m4, and P(cen) PEV. Mutations in subgroup A affect bw^D, w^m4, and P(cen), mutations in subgroup B fail to affect bw^D but do modify w^m4 and P(cen), and mutations in subgroup C affect bw^D and w^m4 but fail to affect P(cen). No mutations that modified bw^D and P(cen) and did not affect w^m4 were identified.

Of the 44 group IV complementation groups, 21 are in subgroup A (Table 2), including 20 suppressors and 1 enhancer. Ten of these mutations are located on the X chromosome, 4 are on the second, and 7 are on the third. None of the autosomal subgroup A mutations are homozygous viable. Of particular interest is the single enhancer in subgroup A, E1377. Previous experiments have suggested that enhancers of w^m4 variegation do not affect bw^D PEV (SASS and HENIKOFF 1998 ). Indeed, no genic enhancers of bw^D PEV have been identified until now; E1377 and another enhancer, E1261 (in subgroup C), provide intriguing exceptions to this observation, as do mutations E69 and E113 (in group III, discussed previously).

Subgroup A also includes one complementation group that has four mutations (1009, 1097, 1207, and 1545). Analysis of mitotic chromosomes from 1207/1207 and 1009/1009 homozygous larvae (H. LE, K. DONALDSON and G. KARPEN, unpublished results) demonstrated that these mutations exhibit telomere fusions, which is similar to the behavior of Su(var)2-5 (HP1) alleles (FANTI et al. 1998 ). Complementation analysis indicated that 1009, 1097, 1207, and 1545 are all new alleles of Su(var)2-5, and sequence analyses confirmed that all four mutations represent lesions in the Su(var)2-5 locus. 1009 contains an A to T change at the first position of the codon for amino acid 46, resulting in a premature stop codon and presumed truncation of the protein; 1207 contains a C to T change at the first position of the codon for amino acid 69, leading to a premature stop codon and presumed truncation of the protein; 1097 contains an A to T change at the first position of the codon encoding amino acid 53, causing an asparagine to tyrosine change; 1545 has two mutations within the Su(var)2-5 locus, one a G to A change in the first position of the codon for amino acid 27, causing a glutamic acid to lysine change, and the other an A to G change in the first position of the codon for amino acid 183, causing an asparagine to aspartic acid change.

Sixteen complementation groups comprise subgroup B. These mutations affect TDA-PEV, w^m4, and P(cen) but have no effect on bw^D. Fifteen of these mutations are suppressors of PEV; 1 is on the X, 4 are on the second, and the remaining 10 are on the third. The single enhancer in subgroup B is on the second chromosome; this complementation group is composed of two mutations, E646 and E1672. Whereas none of the subgroup A autosomal mutations were viable as homozygotes, nearly one-half (7 out of 15) of the subgroup B autosomal mutations are homozygous viable.

The remaining seven mutations comprise subgroup C. These six suppressors and one enhancer affect TDA-PEV, bw^D, and w^m4 but do not affect P(cen). Four of these suppressors are on the X chromosome, and the two second chromosome suppressors are both homozygous lethal, as is the single enhancer in this subgroup, E1261. Mutation 1535 has not been tested for its effect on bw^D and thus could be a member of group III.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Very few proteins involved in chromosome nuclear organization and telomere structure and function in Drosophila have been identified. Previous studies have described and characterized a novel form of position effect variegation associated with terminal deficiencies that position new chromosome ends distal to the yellow gene (TOWER et al. 1993 ; DONALDSON and KARPEN 1997 ). This TDA-PEV is suppressed by the presence of a second minichromosome in trans, a phenomenon termed trans-suppression (DONALDSON and KARPEN 1997 ). Previous data suggested that TDA-PEV and trans-suppression involve chromosome pairing, nuclear organization, and telomere structure and function (DONALDSON and KARPEN 1997 ).

Here, we describe the results of a screen for modifiers of TDA-PEV and trans-suppression. Seventy modifiers of TDA-PEV were identified, corresponding to at most 62 complementation groups. Secondary analysis of the mutations identified four classes of mutations, including mutations that affect only TDA-PEV, or TDA-PEV and a limited number of PEV types, as well as general PEV modifiers. In addition, we isolated a suppressor mutation (1699), corresponding to a new allele of Su(var)3-9, which is the first identified modifier of PEV to also affect TPE. We predict that the loci identified in this screen will help elucidate the relationship between chromosome nuclear organization, telomere structure and function, and gene expression.

Modifiers of TDA-PEV may provide a unique system for the study of eukaryotic nuclear organization:
Genetic and cytological observations have led to models suggesting that appropriate positioning of genes and chromosomes within interphase nuclei is required for normal expression (WAKIMOTO and HEARN 1990 ; KARPEN 1994 ; HENIKOFF 1997 ; BRIDGER and BICKMORE 1998 ; LAMOND and EARNSHAW 1998 ). In S. cerevisiae, telomeres are clustered at the edge of the nucleus, and mutations that interfere with telomere-induced silencing and telomere length also affect telomere clustering and association with the nuclear periphery (BOULTON and JACKSON 1998 ; LAROCHE et al. 1998 ). Similarly, in S. pombe, silenced chromatin is formed near the mating-type loci, telomeres, and centromeres. Mutations in some proteins alleviate silencing at all three loci (e.g., Swi6; EKWALL et al. 1995 ), while other mutations affect only one type of silencing (e.g., Mis6; SAITOH et al. 1997 ). Thus, studies of unicellular eukaryotes suggest that different types of silencing are regulated by different proteins. Such silencing may involve associations between genes and specific parts of the nucleus, reflecting the possible presence of distinct heterochromatic structural "domains."

Similar conclusions can be gleaned from studies of gene silencing in higher eukaryotes. In many cell types, heterochromatic telomere and centromere regions are associated with the nuclear lamina and are clustered at opposite sides of the nucleus, while the euchromatin is located predominantly in the nuclear lumen (RABL 1885 ; COMINGS 1980 ; MATHOG et al. 1984 ; HOCHSTRASSER et al. 1986 ; FUNABIKI et al. 1993 ). Chromosomes in mammals have also been demonstrated to occupy distinct and reproducible domains in interphase nuclei (BRIDGER and BICKMORE 1998 ; DIETZEL et al. 1998 ; VISSER et al. 1998 ; ZINK et al. 1998 ). Gene repression in Drosophila can be associated with the location of a gene at telomeres, at pericentric heterochromatin, and in association with Polycomb-group (Pc-G) proteins. Several factors have been found to affect only one or two of these types of silencing, suggesting there may be functionally separable silencing regions in Drosophila, as in yeasts. For example, analyses of general PEV modifiers failed to identify any that affected telomeric position effect (WALLRATH and ELGIN 1995 ; CRYDERMAN et al. 1999 ).

Secondary analyses of the TDA-PEV modifiers described here provide additional insight into the relationships among different types of PEV. The classification of these mutations into different groups on the basis of their spectrum of modification effects suggests that distinct components and mechanisms are responsible for different forms of silencing in Drosophila. The majority of mutations impacted w^m4, bw^D, and the PEV associated with centric heterochromatin P-element insertions (group IV, subgroup A), suggesting that many components are common to different types of PEV. However, only one mutation (1699, group II) affected all types of PEV tested, including TPE, suggesting that the Su(var)3-9 protein may be a component of most or all types of silencing in Drosophila.

Most importantly, we have identified mutations that alter a limited number of PEV types (groups III and IV, subgroups B and C) or that affect only TDA-PEV (group I). These data are consistent with the existence of multiple silencing mechanisms and suggest that the different types of silencing are physically and/or functionally distinct from each other, due perhaps to the presence or absence of specific proteins, specific protein domains, or associations with distinct nuclear domains.

Mutations that modify only TDA-PEV may help dissect telomere structure and function in Drosophila:
Drosophila telomeres do not contain "canonical" telomerase-produced short tandem repeats (PARDUE and DEBARYSHE 1999 ; see Introduction). Nevertheless, telomeres in Drosophila perform the same basic functions as telomeres in other organisms (DERNBURG et al. 1995 ; COOPER 2000 ); they counter terminal sequence loss due to replication of a linear molecule, are associated with the nuclear envelope (HARI et al. 2001 ), prevent chromosome ends from fusing (FANTI et al. 1998 ), and identify the end of the chromosome as normal and not a double-stranded break that must be repaired (MCEACHERN et al. 2000 ). Terminal sequence loss by incomplete DNA replication is counterbalanced by rare TART and Het-A retroposon additions. However, it is obvious that the remaining functions of telomeres occur independently of TART and Het-A elements, since fully functional telomeres in Drosophila have been identified that have no TART or Het-A sequence near the ends (MASON et al. 1984 ; LEVIS 1989 ; BIESSMANN et al. 1990 ). These terminal deficiency chromosomes appear to have completely normal telomere function. They do not cause double-stranded break-dependent cell cycle arrest and show no evidence of chromosome end-to-end fusion (MASON et al. 1984 ; LEVIS 1989 ; BIESSMANN et al. 1990 ). Drosophila telomeres are also associated with at least one protein known to play a role in telomere function, HP1 (FANTI et al. 1998 ). These observations suggest that Drosophila telomeres are not defined by primary DNA sequence, but by some other factor, such as a specific chromatin structure, the presence of specific proteins, or some other epigenetic mark (BIESSMANN et al. 1990 ; AHMAD and GOLIC 1999 ). Indeed, in other organisms there is mounting evidence that although the telomerase enzyme is responsible for maintaining the length of telomeres, equally important proteins are playing critical roles in telomere functions, including telomere clustering, interaction with the nuclear envelope, and prevention of telomere fusion (VAN STEENSEL et al. 1998 ; COOPER 2000 ; THAM and ZAKIAN 2000 ). One particularly telling piece of data is that mice lacking telomerase are viable and appear normal for several generations (BLASCO et al. 1997 ; RUDOLPH et al. 1999 ), implying that other components can still act to promote telomere functions in the absence of telomerase-mediated maintenance of telomeric DNA sequence.

Since TDA-PEV involves placement of a chromosome end closer to the yellow gene, further study of the genes that modify TDA-PEV is likely to improve our understanding of telomere biology and nuclear organization in Drosophila. Indeed, one Mod(var) [Su(var)2-10] encodes a protein [protein inactivator of activated STAT (PIAS)] that is localized to the nuclear lamina and telomeres (HARI et al. 2001 ). Su(var)2-10 mutations suppress TDA-PEV and alter telomere-telomere and telomere-lamina associations. These results support the hypothesis that the SU(VAR)2-10 protein coordinates telomere functions and is required for normal nuclear organization in interphase. They also suggest that other modifiers of TDA-PEV are likely to play roles in telomere structure and function.

Secondary tests were performed on 67 TDA-PEV modifier mutations, using various PEV types (Fig 2). Of greatest interest are 10 mutations, corresponding to at most eight complementation groups, which affected only TDA-PEV and no other type of PEV (group I, Table 2). Three of the mutations (E1047, E1060, and E1178) fail to complement each other for lethality, suggesting that they are all mutations in the same locus. Group I mutations are not general PEV modifiers, suggesting that TDA-PEV is a specialized type of PEV that involves novel factors in addition to general modifiers of PEV.

If the group I mutations alter telomere function, then why do they have no effect on TPE? It is possible that TDA-PEV and TPE are associated with different telomere regions, functions, or proteins. Alternatively, TPE may occur independently of telomere function. In most cases of TPE, the variegation occurs after insertion of a transgene into the TAS elements, large tandemly repeated arrays that are found just proximal of the Het-A and TART arrays at the ends of Drosophila chromosomes (KARPEN and SPRADLING 1992 ; KURENOVA et al. 1998 ). Complementary results suggest that TPE is induced by the TAS elements rather than telomeric location. First, TAS elements can act as boundary elements when they are located in more proximal positions along the chromosome arm (KURENOVA et al. 1998 ). Thus, TAS elements affect gene expression independent of their telomeric position. Second, terminal deficiencies broken just distal to the inserted transgene, which eliminate the distal TAS repeats, result in reversion of TPE (LEVIS 1989 ; TOWER et al. 1993 ). These results suggest that flanking TAS elements are responsible for TPE, rather than the presence of a chromosome end. Finally, at least two proteins with demonstrated roles in telomere behavior [Su(var)2-10/PIAS and Su(var)2-5/HP1; FANTI et al. 1998; HARI et al. 2001] affect TDA-PEV and not TPE. Thus, it is likely that mutations that affect only TDA-PEV will identify proteins involved in the maintenance of chromosome ends or their nuclear positions and other chromosomal functions.

Mutation 1699, a general modifier of PEV, also affects telomeric position effect:
Only 1 of the 67 mutations that affect TDA-PEV, 1699, modified all types of PEV tested, including TPE. This result is surprising; only a few mutations are known to modify TPE, and 1699 represents the first example of a mutation that affects both PEV and TPE (TALBERT et al. 1994 ; WALLRATH and ELGIN 1995 ; KURENOVA et al. 1998 ). Recombination mapping, phenotypic analyses, and sequence analyses demonstrated that 1699 represents a new allele of Su(var)3-9 (TSCHIERSCH et al. 1994 ). SU(VAR)3-9 homologs are found in a wide range of organisms, including S. pombe (IVANOVA et al. 1998 ), mice, and humans (AAGAARD et al. 1999 ; O'CARROLL et al. 2000 ). SU(VAR)3-9 homologs encode methyltransferases that modify lysine 9 of histone H3 and have been demonstrated to play roles in the recruitment of HP1 to heterochromatin and in chromosome segregation and mitotic progression (EKWALL et al. 1996 , EKWALL et al. 1999 ; AAGAARD et al. 1999 ; REA et al. 2000 ; BANNISTER et al. 2001 ; LACHNER et al. 2001 ). 1699 contains a mutation in a cysteine residue that is conserved throughout the SU(VAR)3-9 homologs, part of a cysteine-rich region required for histone H3 lysine 9 methyltransferase activity (REA et al. 2000 ). We conclude that SU(VAR)3-9 may also play a role in telomere function and nuclear organization and that at least one protein functions in both TPE and PEV.

Some mutations affect TDA-PEV and only one other type of PEV:
Most modifiers of PEV act on multiple types of PEV, suggesting they represent mutations in genes involved in general heterochromatin biology and common aspects of gene silencing. However, secondary analysis of TDA-PEV modifiers also identified mutations that affect only a subset of PEV types. Two enhancer mutations (E69 and E113) affected only TDA-PEV and bw^D. Genetic and cytological studies indicate that bw^D PEV is associated with mitotic pairing of homologous chromosomes and alterations in the associations of the bw⁺ chromosome with centric heterochromatin and likely reflects the importance of nuclear organization to gene expression (TALBERT et al. 1994 ; CSINK and HENIKOFF 1996 ; DERNBURG et al. 1996 ). Thus, mutations that act specifically on TDA-PEV and bw^D may identify proteins that organize chromosomes within the nucleus or impact chromosome pairing in somatic cells. Cytological analyses of interphase chromosome organization and pairing in these mutants are being used to address these hypotheses.

Interestingly, a previous screen for modifiers of bw^D PEV failed to produce any mutations that acted specifically on the trans-inactivation of the bw⁺ allele seen with bw^D (TALBERT et al. 1994 ). However, the Talbert screen looked specifically for mutations that dominantly affect bw^D and did not affect the cis-inactivation of bw; this eliminated isolation of general modifiers of bw expression and general modifiers of PEV. No modifiers that behaved in such a fashion were identified. The two mutations we have identified that specifically affect bw^D and TDA-PEV (E69 and E113) are both X-linked recessive enhancers and would therefore have not been found in the Talbert screen. Since these bw^D modifiers are both recessive, it is possible that recessive autosomal mutations that affect bw^D could also be found. Further analyses of these mutations may provide insight into their roles in bw^D PEV specifically and in chromosome nuclear organization in general.

Two mutations, 26 and E226, alter only PEV associated with yellow expression [i.e., TDA-PEV and P(y⁺), Table 2]. These mutations are likely to identify genes associated with general yellow expression, rather than PEV. A more interesting behavior involves mutations 1309, which affects TDA-PEV and only w^m4, and E1451, which impacts TDA-PEV and only P(cen). Such differential behavior suggests that although w^m4 and centric P-insertion silencing must have many components in common [general Mod(var)s], they also may have unique components. Alternatively, these mutations may identify proteins that function generally in PEV, but with lesions in specific residues or domains that play distinct roles in different types of silencing. Studies of such unique mutations as 1309 and E1451 are likely to provide insight into the complex regulation of heterochromatin formation and function.

A new collection of general modifiers of PEV:
In the last 20 years, multiple screens have been undertaken to identify mutations that modify position effect variegation (REUTER and WOLFF 1981 ; SINCLAIR et al. 1983 ; LOCKE et al. 1988 ; WUSTMANN et al. 1989 ; DORN et al. 1993 ). The majority of these screens focused on mutations that alter w^m4 PEV. Most TDA-PEV modifier mutations also modify w^m4, suggesting that this collection of PEV modifiers may overlap significantly with those from previous screens. Some of these general Mod(var)s may be new alleles of previously i