Transvection is a phenomenon wherein gene expression is effected by the interaction of alleles in trans and often results in partial complementation between mutant alleles. Transvection is dependent upon somatic pairing between homologous chromosome regions and is a form of interallelic complementation that does not occur at the polypeptide level. In this study we demonstrated that transvection could occur at the vestigial (vg) locus by revealing that partial complementation between two vg mutant alleles could be disrupted by changing the genomic location of the alleles through chromosome rearrangement. If chromosome rearrangements affect transvection by disrupting somatic pairing, then combining chromosome rearrangements that restore somatic pairing should restore transvection. We were able to restore partial complementation in numerous rearrangement trans-heterozygotes, thus providing substantial evidence that the observed complementation at vg results from a transvection effect. Cytological analyses revealed this transvection effect to have a large proximal critical region, a feature common to other transvection effects. In the Drosophila interphase nucleus, paired chromosome arms are separated into distinct, nonoverlapping domains. We propose that if the relative position of each arm in the nucleus is determined by the centromere as a relic of chromosome positions after the last mitotic division, then a locus will be displaced to a different territory of the interphase nucleus relative to its nonrearranged homolog by any rearrangement that links that locus to a different centromere. This physical displacement in the nucleus hinders transvection by disrupting the somatic pairing of homologous chromosomes and gives rise to proximal critical regions.
THE term transvection was first introduced by Lewis (1954) to describe his observations of the Ultrabithorax (Ubx) gene in Drosophila melanogaster, wherein complementation between certain mutant Ubx alleles could be disrupted by chromosome rearrangement. Since Lewis' time, transvection has been shown to occur at more than a dozen loci in D. melanogaster. There are a wide variety of transvection effects each with their own distinct properties and mechanisms of action, but for the purpose of this study the term transvection refers to transvection effects involving interallelic complementation between mutant alleles, such as those observed at Abdominal-B (Abd-B), decapentaplegic (dpp), eyes absent (eya), Salivary glue secretion-4 (Sgs-4), Ubx, yellow (y), whitespeckled (wsp), and wings-up A (wup A) (Korge 1981; Gelbart 1982; Davison et al. 1985; Martinez-Laborda et al. 1992; Leiserson et al. 1994; Morris et al. 1998, 1999a; Sipos et al. 1998; Zimmerman et al. 2000; Marin et al. 2004).
For every transvection effect, complementation occurs only with specific pairs of mutant alleles. In many cases one mutant allele contains a structural mutation, while the other allele has a mutation in a cis-regulatory element. The cis-regulatory element is often an enhancer (for examples, see Davison et al. 1985; Morris et al. 1998; Zimmerman et al. 2000). One transvection model proposes that the functional enhancer element on the chromosome with the structural mutation acts in trans on its target promoter located on the homologous chromosome containing the cis-regulatory element (Geyer et al. 1990).
In this study we examined a potential transvection effect at the vestigial (vg) locus. Mutations in the vg gene give rise to a range of phenotypes that variably include improperly developed wings and halteres, low viability, female sterility, developmental delay, and erect postscutellar bristles (see Williams and Bell 1988 and references therein). In this study we focused on vg83b27, a mutant with a deletion limited to the second intron of vg (Williams et al. 1990). During development, vestigial protein levels in the wing imaginal discs of vg83b27 homozygotes are normal until the third larval instar at which time they become very low relative to those in wild-type controls (Williams et al. 1991). These results suggested the presence of a stage-specific enhancer in the second intron of vg and, indeed, Williams et al. (1991) discovered a 750-base EcoRI fragment within the second intron of vg that is significantly conserved between D. melanogaster and D. virilis. This sequence contains a 360-bp minimal element that can drive β-gal expression in the wing and haltere imaginal discs during the third larval instar (Williams et al. 1994). This minimal element acted like a true enhancer in that it drove expression in an orientation-independent manner whether it was upstream or downstream of the target promoter.
The vg83b27 allele genetically complements many vg mutants possessing alterations in the transcription unit (Williams et al. 1990) but classical interallelic complementation at the polypeptide level is not sufficient in explaining the complementation behavior because of insufficient protein production from the vg83b27 allele during a critical stage of wing and haltere development (Williams et al. 1991). If transvection is responsible for the observed complementation behavior then complementation should be disrupted by chromosome rearrangements that affect the ability of complementing alleles to somatically pair and interact. Transvection should be restored in rearrangement trans-heterozygotes if the rearrangements are similar enough to sufficiently restore somatic pairing and interaction between vg alleles.
MATERIALS AND METHODS
Drosophila stocks were raised in an incubator on standard yeast and sugar medium. The incubator was kept at 25° and set on a 12-hr light/dark cycle. The vg alleles used in this study include vg83b27 and vg1 (also known as vgBG or vgBG1). The vg83b27 allele is a mutant that deletes a stage-specific enhancer within the second intron of vg (Williams et al. 1991). The deletion is limited to the second intron and does not alter exons in the vg transcription unit (Williams et al. 1990). The homozygous vg83b27 stock used in this study also contains the X-linked white (w) eye color marker. The vg1 allele contains a viable insertion in intron 3 (Williams et al. 1990). The vg1 strain also contains the cinnibar (cn) and w eye color markers, and the black (b) body color marker. A derivative of vg1 (vg1-7b) was also used in this study and differs from vg1 as described in results. The vg1-7b chromosome has the cn marker but does not contain the w marker. Homozygous vg1-7b stocks are difficult to maintain so the stock has a Curly Oster (CyO) second chromosome balancer segregating in the population to increase the overall health of the stock. The CyO balancer contains the following genetic markers: Curly (Cy), Star (S), cn, and brown (bw). Full descriptions of all stocks and genetic markers can be found in Lindsley and Zimm (1992) and online at http://flybase.org (FlyBase Consortium 2002).
Scoring the wings and halteres of vestigial mutants:
The wings and halteres were examined for each fly and these individual results were combined to produce a phenotype score for the whole group. An ordinal scoring system for individual flies was developed that would allow a wing score to be calculated. A fly was given an overall wing rank of 1–6 on the basis of the phenotype of its wings (Table 1) with 1 representing the most extreme mutant phenotype and 6 being indistinguishable from wild type. For flies whose wings differed in phenotype, such that the individual wings fit the descriptions of different ordinal rankings, an average of the two individual wing scores was made to create a score for the fly. For consistency, in cases where the average produced a decimal of 0.5, scores were always rounded down to the nearest whole number. A mean wing score (Wm) for a group of flies was calculated by taking the mean of all individual wing pair ranks. A standardized Wilcoxon rank sum statistic with continuity correction was used to statistically compare wing scores from different crosses. This statistical test is equivalent to a Mann-Whitney test and is used as a nonparametric alternative to the two-sample t-test to compare two independent populations with ranked data (Zar 1999). Halteres were scored as being present or absent in a given individual, so for a group of flies the haltere distribution score (H) was simply the proportion of individuals that had at least one haltere. Haltere scores from different crosses were compared using chi-square analysis.
Creating noncomplementing derivatives of vg mutants:
Figure 1 shows the mutagenesis screens used to create noncomplementing derivatives of vg mutants. Three- to 4-day-old males, homozygous for vg83b27, vg1, or vg1-7b, were exposed to 2000 Rads of X-ray radiation and mated en masse to virgin females homozygous for the complementing allele (cross 1). Male progeny with no wings and no halteres were considered potential carriers of noncomplementing derivative chromosomes. These males were individually mated to virgin females with the vg1-7b chromosome balanced over the CyO second chromosome balancer (cross 2). It was possible to recover irradiated chromosomes because the vg1 and vg1-7b chromosomes contained the recessive cn eye color marker while the complementing vg83b27 chromosome did not contain the cn marker. Male and female progeny containing the CyO balancer chromosome and an irradiated second chromosome were selected, on the basis of the presence or absence of the cn phenotype, and mated to create independent derivative lines (cross 3).
Every noncomplementing derivative line was examined cytologically to determine if it contained a visible chromosome aberration. Third instar larval polytene chromosomes were examined when heterozygous with standard wild-type chromosomes. For derivative lines containing a second chromosome heterochromatic break, mitotic chromosomes were examined. The location of the prominent secondary constriction at the heterochromatic/euchromatic junction in 2L (Kaufmann 1934) was used as a marker in mitotic preparations to determine if heterochromatic breaks were in 2Lh or 2Rh. Mitotic chromosomes were prepared as in Moore (1971), with modification from Hilliker (1975), and are briefly described here. The brain of an active third-instar larva was dissected into Drosophila Ringer's and then placed in 1% sodium citrate for 30–90 sec. The brain was then transferred to a drop of 2% aceto-lacto-orcein on a siliconized microscope slide and allowed to fix for 5 min. The tissue was then squashed and observed 24–48 hr later.
Detecting a T(2;3) containing only heterochromatic breakpoints:
To detect any translocations with two heterochromatic breakpoints (not detectable in polytene chromosomes) the crossing scheme in Ashburner (1989) was used. Transvection-disrupting derivatives without visible rearrangements were tested for the presence of a T(2;3) with heterochromatic breaks using a bw ; scarlet (st) marker strain. The bw and st markers produce brown eyes and bright red eyes, respectively. In combination the two markers produce white eyes.
The zeste (z)a null allele was tested to see if it effected complementation between vg83b27 and vg1-7b. The allele was tested by crossing za; vg1-7b virgin females to z+; vg83b27 males and scoring the wings of za; vg1-7b/vg83b27 male progeny.
Complementation between vg83b27 and vg1:
Investigating the possibility of transvection at the vestigial locus began with confirming the complementation ability of vg83b27 with vg1 as reported by Williams et al. (1990). Table 2 provides progeny counts, haltere scores, and wing scores for vg83b27/vg1 and the two homozygous controls. The trans-heterozygotes were shown to have higher wing and haltere scores than both of the homozygous controls. This increase in scores was both visually obvious (Figure 2) and statistically significant [vg83b27/vg1 flies were compared to vg1 homozygotes using a standardized Wilcoxon rank sum test; the calculated standardized Wilcoxon statistic, with continuity correction, was 12.25 (P < 0.001)].
Transvection at vg is not effected by the za null allele:
Table 3 provides wing scores comparing za; vg1-7b/vg83b27 males with z+; vg1-7b/vg83b27 males. The za allele had no effect on complementation [the calculated standardized Wilcoxon statistic, with continuity correction, was 1.026 (P = 0.305)].
Rearrangements of the vg83b27 chromosome disrupt complementation:
An X-ray mutagenesis screen was performed as described in materials and methods, in an attempt to obtain complementation-disrupting chromosome rearrangements of the vg83b27 chromosome. A total of 19,019 males were scored in five separate mutagenesis screens, and 60 individual males carrying potential complementation-disrupting mutations were recovered. Thirty-nine of these males were successfully mated, leading to the creation of independent putative mutant lines. The thirty-nine independent derivative lines were crossed to vg1 virgin females to determine if they carried complementation-disrupting mutations. Eighteen lines showed a significant reduction in wing score compared to that in the control cross (Table 4). One line, PIIB, showed no significant reduction in wing score but showed a significant reduction in haltere score (Table 4). Twenty of the lines complemented to a degree similar to the vg83b27 × vg1 control cross and were considered false positives.
Polytene chromosomes were examined from heterozygotes for each complementation-disrupting derivative (see materials and methods). Of the 19 noncomplementing derivative lines of vg83b27, 14 were found to have visible chromosome rearrangements (Table 5). Twelve of these derivatives had one rearrangement breakpoint proximal to vg. Analysis of the two remaining derivatives showed line H6a to be a deletion that included the vestigial locus (thereby attributing the disruption of complementation to the deletion), while line 327c, a three-break rearrangement, had one breakpoint very near vg83b27, making it unclear whether the rearrangement was disrupting complementation or if the breakpoint was within the vg83b27 coding region.
The five lines lacking visible rearrangements were tested for the presence of a translocation between the second and third chromosomes that contained only heterochromatic breakpoints. All five lines lacked such a translocation. Several possibilities remained as to why complementation was disrupted in these lines, including a mutation in a vg modifier, a mutation in the coding region of vg83b27, or a T(X:2) or T(2;4) with heterochromatic breaks. Further analysis of these lines was not performed since 14 independent rearrangement lines were already obtained and were considered sufficient for our analysis of transvection.
Rearrangements of the vg1 and the vg1-7b chromosomes disrupt complementation:
An attempt was also made to create noncomplementing derivatives of the vg1 chromosome, but due to viability effects there was difficulty in obtaining these derivatives. Only one noncomplementing derivative of vg1 (line B9b) was successfully recovered out of 8400 males screened. Therefore, an attempt was made to create additional derivatives starting from the vg1-7b chromosome. The vg1-7b line contains a small inversion that includes vg (breakpoints 49EF–50B) but this inversion has no effect on its ability to complement vg83b27 for wing phenotype (Table 6) and would not interfere with our ability to recover noncomplementing derivatives.
A total of 2844 males were scored in a mutagenesis screen for noncomplementing vg1-7b derivatives. Seventeen individual males carrying potential complementation-disrupting mutations were recovered. Ten of these males were successfully mated, leading to the creation of independent derivative lines. The 10 independent vg1-7b derivative lines were crossed to vg83b27 females to determine if they carried complementation-disrupting mutations. Six of the lines showed a significant reduction in wing score compared to the control cross of vg1-7b × vg83b27 (Table 6) while 4 of the lines were false positives.
Polytene chromosomes from the complementation-disrupting derivatives of vg1 and vg1-7b were examined over standard wild-type chromosomes for visible chromosome rearrangements. Five noncomplementing derivative lines were found to have visible chromosome rearrangements with at least one rearrangement breakpoint proximal to vg (Table 6). The sixth line was lost before polytene analysis.
The seemingly large number of false positives in all screens (see Table 7 for a summary of all screens) was an unavoidable attribute of the mutagenesis screens and the alleles used in this study. On rare occasions the progeny of vg83b27 × vg1 control crosses have been observed to contain flies with no wings or halteres; these flies would be recovered as false positives. On the basis of all screens in this study, a false positive was recovered in ∼1 out of 1000 flies (0.1%), approximately the same rate as the recovery of transvection-disrupting mutations.
Complementation can be rescued in rearrangement trans-heterozygotes:
If chromosome rearrangements disrupt partial complementation between vg1-7b and vg83b27, and if the complementation is due to a transvection effect, then complementation should be rescued if somatic pairing between the two alleles can be restored. In an attempt to rescue complementation, five vg83b27 derivatives (550, 835, 805, 327c, and 789) and five vg1-7b derivatives (75, 81, 84, 88, and 89) were tested for their ability to complement one another. These lines were chosen in hopes of obtaining a wide variety of complementation results ranging from full restoration between lines with similar rearrangements to a lack of restoration between dissimilar rearrangements. Complementation was considered to be restored for any trans-heterozygous combination that rescued wing and/or haltere scores at, or above, the level of complementation observed in the control cross (vg1-7b × vg83b27). Complementation should be restored in rearrangement trans-heterozygotes if the rearrangements are similar enough to allow sufficient restoration of pairing and interaction between the specific vg1-7b and vg83b27 derivatives.
Out of the 25 possible combinations, 11 trans-heterozygotes, involving eight independent rearrangement lines, showed significant restoration of complementation (Table 8). Lines 805 and 84 were lethal in combination and the remaining combinations did not show significant restoration (data not shown).
Figure 3 provides a diagrammatic representation of the eight complementation-disrupting rearrangement chromosomes that made up the 11 trans-heterozygous combinations in Table 8. Figure 4 provides a visual representation of the proposed somatic pairing in rearrangement trans-heterozygotes in which complementation was rescued. The trans-heterozygotes in Figure 4 can be divided into four groups. In the first group, wing and haltere scores were fully restored to a level higher than that in the vg83b27/vg1-7b control and approached levels expected for a completely nonrearranged combination of vg83b27/vg1. The second group produced full restoration for either the wing or the haltere score and partial restoration for the other score. The third group showed partial restoration for both wing and haltere scores to a level expected for vg83b27/vg1-7b trans-heterozygotes. In the last group, complementation was the weakest and showed only partial restoration for either the wing or the haltere score and no rescue for the other score.
The basic trend in Figure 4 is that complementation is best restored in trans-heterozygote combinations where the breakpoints are on the same chromosome arm, they produce arms of similar length, and little pairing competition is expected between rearrangement products. The next best pairings are similar to the first but involve opposite arms of the third chromosome (in Figure 4 these trans-heterozygote combinations are drawn in the Rabl orientation to show how in three dimensions, close proximity between vg alleles is possible in the interphase nucleus even when the alleles reside on opposite arms of the same chromosome).
The critical region for this transvection effect can be deduced from the breakpoints in the derivative lines that make up complementing trans-heterozygote combinations in Table 5. Those breakpoints range from 2Rh to 48AB, demonstrating that the critical region is any point proximal to vg.
Williams et al. (1990) investigated the possibility of transvection at vg when they were first studying the vg83b27 allele. At that time several lines of evidence argued against transvection: two deficiencies, Df(2R)vg56 and Su(z)25, showed partial complementation with vg83b27 and these deficiencies were expected to disrupt pairing in the region; all transvection effects known at that time were dependent on the allelic state of the z gene and Williams et al. (1990) found that zv77h3, a strong zeste allele, did not disrupt the complementation ability of vg83b27. These facts no longer argue against transvection at vg. First, the two deficiencies both have one breakpoint in the last exon of vg (at the distal end of the coding region). Since these two deletions do not compromise the second intron enhancer (and the breakpoints are not in the proximal critical region), these coding region mutations are expected to partially complement with vg83b27. Second, zeste-independent transvection effects have been discovered at Scr, Abd-B, and Ubx (Pattatucci and Kaufman 1991; Martinez-Laborda et al. 1992; Hopmann et al. 1995; Southworth and Kennison 2002).
In this study we tested the za null allele and found that it did not affect transvection at vg. The za allele does affect transvection effects at numerous other loci, all of which involve enhancers. These loci include dpp, eya, Ubx, and y (Lewis 1954; Wu and Gelbart 1982; Geyer et al. 1990; Leiserson et al. 1994; Chen et al. 2002). Our results with za, in combination with the zv77h3 results from Williams et al. (1990), do suggest that the transvection effect involving vg83b27 is zeste independent or at least a unique zeste dependence. However, a more exhaustive analysis of the gain-of-function alleles of zeste needs to be conducted to establish the true relationship between the allelic state of zeste and transvection at vg.
Lewis (1954) established a set of rules to predict when, and how severely, a simple two-break rearrangement would disrupt transvection. The term “critical region” was coined to describe the chromosomal region in which a rearrangement breakpoint would need to occur to disrupt transvection. In the transvection effect studied here, the critical region encompasses the region proximal to vg. This is consistent with the proximal critical regions characterized for Ubx, dpp, and eya (Lewis 1954; Gelbart 1982; Leiserson et al. 1994). A determining factor for the size of critical regions is linked to the presence or absence of genetic elements termed cis-factors that suppress the ability of control elements to act in trans. Cis-factors include promoter integrity, the strength of the control element acting in trans, pairing sensitive DNA sequences, and other gene-specific factors. For example, studies have shown that cis-promoter integrity supports cis-preference for enhancer-promoter interactions whereas mutations in the cis-promoter can enhance transvection (Martinez-Laborda et al. 1992; Casares et al. 1997; Sipos et al. 1998; Gibson et al. 1999; Morris et al. 1999b, 2004).
There is a relationship between the strength of a trans-acting control element and the size of critical regions. In terms of enhancers, strong, long-range enhancers are less sensitive to rearrangements and give rise to smaller critical regions than do transvection effects involving weak, short-range enhancers. There are examples of both rearrangement-insensitive and -sensitive transvection effects at Abd-B (see Duncan 2002), revealing how the specific combinations of cis-factors and trans-acting regulatory elements contribute to the specific properties of a given transvection effect. Complementation involving vg83b27 and vg1 involves a short-range intronic enhancer that we show to be very sensitive to chromosome rearrangement. It also has a large critical region. Critical regions for transvection effects that do not involve the trans-action of a cis-regulatory element will likely reflect the specific mechanisms behind the effect. For example, the critical region for transvection involving the brownvariegated alleles will be determined by the properties of heterochromatin and position-effect variegation (Henikoff and Dressen 1989; Henikoff et al. 1995).
The combination of cis-factors with cis-regulatory elements acting in trans provides insight as to why a given critical region is small or large, but it does not explain why the large critical regions for Ubx, dpp, eya, and vg encompass only the region proximal to, and not distal to, each locus. Lewis (1954) attempted to explain the proximal nature of critical regions at Ubx by proposing a zippering model of somatic pairing. He proposed that homologous chromosomes pair in the interphase nucleus in a directional manner, beginning at the centromere and proceeding toward the telomere. Hence, only the regions proximal to a rearrangement breakpoint would be paired. However, advances in the study of homologous pairing suggest that proximal loci are not the first loci to pair when compared to distal euchromatic loci (Csink and Henikoff 1998; Fung et al. 1998; Gemkow et al. 1998; Vazquez et al. 2001). In fact, these studies suggest that somatic pairing in the Drosophila interphase nuclei proceeds through a nondirectional random-walk model. We propose that if homologous pairing does not proceed in a directional manner, then the organization of chromosomes in the interphase nucleus must play an important role in proximal critical regions.
There is a nonrandom architecture to the Drosophila interphase nucleus. In both diploid and polytene nuclei, chromosome arms occupy distinct, nonoverlapping domains, or territories, in the nucleus (Hilliker 1985; Hochstrasser et al. 1986; Hilliker and Appels 1989). Additionally, Drosophila interphase chromosomes are initially found in the Rabl orientation (Rabl 1885) with centromeres clustered at one end of the nucleus and the chromosome arms extending out toward the other end (Hilliker and Appels 1989). Transvection-disrupting rearrangements that move one allele of a complementing pair to a different chromosome arm might disrupt transvection by isolating that allele in an improper chromosome territory, making it less likely that it will “find” and pair with its complementing allele (Figure 5).
The notion that transvection-disrupting rearrangements can isolate an allele in an improper chromosome territory is supported by two observations: there is a correlation between the relative positions of chromosomes before mitosis and their relative positions after mitosis, and the arrangement of chromosome arms in eukaryotic interphase nuclei tends to be defined by their positions at the end of the last mitotic division (Hilliker and Appels 1989). Since homologous chromosomes in Drosophila are still associated during metaphase, homologous chromosomes begin anaphase in close proximity to one another relative to other chromosomes and likely end anaphase in relative proximity. This would make somatic pairing through random walking an efficient mechanism, except in cases where chromosomal regions are physically separated from one another by chromosome rearrangement. The position of each chromosome in the nucleus after mitosis would be determined by the final resting place of the centromere. A chromosome rearrangement with breakpoints proximal to an allele would link that allele to a new centromere, or to the other arm of the same chromosome, and would change the relative position of that allele in the nucleus as it relates to chromosome territories established by nonrearranged chromosomes. This would lead to a large proximal critical region in cases of transvection effects involving weak cis-regulatory elements acting in trans.
Several lines of evidence suggest that the somatic pairing in the Drosophila interphase nucleus is not essential, implying that transvection is not essential. First, Hilliker and Trusis-Coulter (1987) created numerous translocations between the second and third chromosomes and tested trans-heterozygotes for lethality. Breakpoints were randomly distributed throughout the genome, and out of ∼1000 trans-heterozygous combinations tested, only 2 were lethal. If somatic pairing is essential then more trans-heterozygotes should have been lethal or at least have shown some visible defects. Second, the polytene nuclei of the hindgut and midgut of third instar larvae lack the same level of nuclear organization found in salivary gland nuclei. These nuclei frequently have overlapping chromosome territories with telomeres and centromeres found throughout the nuclei, sometimes without a clear Rabl orientation (Hochstrasser and Sedat 1987). Despite the disorderly organization observed in hindgut and midgut nuclei, the gut cells are capable of functioning without the level of nuclear organization found in other nuclei. Last, somatic pairing does not appear to be essential on the basis of the large number of deletion lines and transposon insertion lines that have been successfully created and maintained in Drosophila. Most deletion heterozygotes can survive, and, with the exception of position effects (such as those occurring when heterochromatic/euchromatic boundaries are crossed), single, unpaired transgenic insertions tend to function adequately (as do the phenotypic markers embedded in the transposon).
Transvection might not be an essential process in the interphase nucleus, but its study has provided valuable insights into various areas including enhancer and promoter function, somatic pairing, nuclear architecture, the structure of genes, and others. Here we have shown that transvection is likely the mechanism underlying the interallelic complementation observed with the vg83b27 allele. This transvection effect it is not altered in za mutant background.
The authors acknowledge research support from discovery grants from the Natural Science and Engineering Research Council of Canada to A.J.H. and J.B.B.
Communicating editor: J. Birchler
- Received February 1, 2005.
- Accepted April 25, 2005.
- Genetics Society of America