Abstract
We investigated the fate of dicentric chromosomes in the mitotic divisions of Drosophila melanogaster. We constructed chromosomes that were not required for viability and that carried P elements with inverted repeats of the target sites (FRTs) for the FLP site-specific recombinase. FLP-mediated unequal sister-chromatid exchange between inverted FRTs produced dicentric chromosomes at a high rate. The fate of the dicentric chromosome was evaluated in the mitotic cells of the male germline. We found that dicentric chromosomes break in mitosis, and the broken fragments can be transmitted. Some of these chromosome fragments exhibit dominant semilethality. Nonlethal fragments were broken at many sites along the chromosome, but the semilethal fragments were all broken near the original site of sister-chromatid fusion, and retained P element sequences near their termini. We discuss the implications of the recovery and behavior of broken chromosomes for checkpoints that detect double-strand break damage and the functions of telomeres in Drosophila.
IN most eukaryotes a chromosome must consist of a single molecule of double-stranded DNA, with a single functional centromere and two telomeres, in order to be stably transmitted. The behavior of dicentric chromosomes during cell division has been extensively examined in maize, Saccharomyces, and Drosophila. In maize, a dicentric chromosome—produced by meiotic recombination between a paracentric inversion and its normal homolog—breaks during anaphase I and a chromosome fragment is delivered to each daughter cell (McClintock 1939). Broken chromosomes are subject to repeated cycles of dicentric formation and breakage because replicated broken chromatids fuse, again forming a dicentric chromosome. McClintock called this the breakage-fusion-bridge cycle (BFB cycle; McClintock 1939). In contrast, dicentric chromosomes produced in the same way in Drosophila females do not break. Sturtevant and Beadle (1936) deduced that dicentric X chromosomes remain stretched between the two poles of the MI division, and are excluded from the functional oocyte nucleus. Novitski (1952) reported that the fate of a dicentric chromosome in meiosis could be altered by modifying the structure of the chromosome. He produced dicentric chromosomes from Xs that carried parts of the Y chromosome appended as an additional arm and concluded that these dicentric chromosomes do break. While the behavior of dicentric chromosomes in mitotic divisions in Drosophila has received limited attention, there is evidence that breakage occurs (Merriamet al. 1972; Golic 1994). However, chromosome loss in mitosis after dicentric formation may also occur, and has been proposed to underlie the mitotic instability of ring chromosomes (Hinton 1955, 1959; Leigh 1976).
Whether chromosome breakage or loss occurs after dicentric formation is a key issue, because these may have different cellular consequences. Muller (1941; Muller and Herskowitz 1954) extensively characterized chromosome rearrangements in Drosophila and deduced that the natural end of a chromosome, which he termed the telomere, was an essential structure. In addition, Muller performed screens for broken chromosomes after X-irradiation. Muller found no case in which a chromosome missing a telomere (i.e., a terminal deficiency) was recovered, and argued that there was no evidence that broken chromosomes could be healed by de novo addition of a telomere in Drosophila.
Recently, a number of groups have recovered the broken chromosomes that Muller looked for. Mason et al. (1984) used a mutation, mu2, that permits the recovery of broken X chromosomes after irradiation of females. The requirement for a mu2 mutation to recover the broken chromosomes suggests that such a chromosome may normally be lethal. Even so, the mu2 mutation is not required to propagate the broken chromosome. In addition, Levis (1989) produced a chromosome 3 missing its tip distal to a P element after destabilization by P transposase. The mu2 mutation was not required to recover this broken chromosome. It is not clear why these screens have recovered broken chromosomes while those Muller performed did not.
The mechanisms of broken chromosome healing have been extensively characterized in Saccharomyces cerevisiae (McCusker and Haber 1981; Dunnet al. 1984; Haber and Thorburn 1984; Wang and Zakian 1990). Telomere acquisition on the end of a broken chromosome primarily occurs by recombination, but may also occur through de novo addition. A chromosome without a telomere cannot be propagated in yeast, but the functions of the telomere have not been fully defined; telomeric sequences promote replication of the chromosome end by telomerase, but this function is probably only necessary for long-term survival of mitotically active cells and not for cell viability itself (Zakian 1989). One essential role of the telomere must be to distinguish the chromosome end from a double-strand break (DSB) within the chromosome, so that only the latter activate DNA-damage responses and checkpoints. However, broken chromosomes in Saccharomyces show continuing instability even in strains where DNA-damage responses are defective, suggesting that the telomere has additional roles (Sandell and Zakian 1993). The detection of physical associations between telomeres and between telomeres and the nuclear envelope has suggested that telomeres play a role in the organization of the nucleus (reviewed in Blackburn and Szostak 1984; Dernberget al. 1995).
Eucaryotes with linear chromosomes normally have special DNA sequences that confer the properties of a telomere. In most organisms these are short G- and T-rich sequences that are repeated many times (GT repeats; Zakian 1989), but in Drosophila the DNA sequences at the ends of chromosomes are not GT repeats and instead belong to two families of middle-repetitive sequences: the HeT family and the TART family (Younget al. 1983; Levis 1989; Leviset al. 1993; Biessmannet al. 1992; Karpen and Spradling 1992; Walteret al. 1995). Both elements are arranged as tandem repeats, frequently with deletions at their distal ends, and this has suggested that the sequences are acquired by retrotransposition to the chromosome end, alleviating the need for telomerase-mediated extension. Other higher dipterans probably have this type of telomere structure and elongation as well (Biessmann and Mason 1994).
Most broken chromosomes that have been isolated in Drosophila do not have HeT or TART sequences on their ends (Levis 1989; Biessmannet al. 1990). Because Drosophila carrying these chromosomes are viable and the chromosomes are transmitted normally, it is not apparent whether HeT and TART sequences contribute to telomere functions. Despite this, occasional acquisitions of these sequences by broken chromosome ends have been observed. Traverse and Pardue (1988) examined a spontaneous linear derivative of a ring chromosome. HeT sequences were appended onto both new ends of the chromosome. Some chromosomes that were broken by P transposase or in mu2 females have later acquired HeT or TART sequences (Biessmannet al. 1990; Leviset al. 1993). Such observations support the model where occasional retrotranspositions of HeT or TART to the chromosome ends maintain the lengths of Drosophila chromosomes.
FLP-mediated unequal sister-chromatid exchange and dicentric bridge formation. FLP-mediated recombination between FRTs in opposite orientation on sister chromatids fuses those chromatids together. The dicentric is stretched between the poles of division at the following mitosis. The solid ovals and circles represent centromeres, and the solid arrows represent FRTs.
In this article we describe our studies of the behavior of dicentric chromosomes and broken chromosomes in Drosophila. Falco et al. (1982) observed that the FLP site-specific recombinase can mediate unequal sister-chromatid exchange (USCE) between inverted repeats of chromosomal FLP Recombination Targets (FRTs) to generate dicentric chromosomes in Saccharomyces. Golic (1994) showed that FLP-mediated USCE also produces dicentric chromosomes in Drosophila (Figure 1). Sister chromatids become fused at the site of recombination to form a dicentric chromosome and an acentric chromosome. The dicentric chromosome forms a bridge between the two poles of division when the cell divides, while the acentric portion of the chromosome is probably lost. In Drosophila, the loss of the acentric chromosome resulted in substantial aneuploidy in the daughter cells (Golic 1994). Aneuploidy of such magnitude severely reduces the viability of cells (Ripoll and Garcia-Bellido 1979; Ripoll 1980), making it difficult to assess the fate of the dicentric bridge. In this study, we generated dicentric chromosomes in the male germline using chromosomes that are not essential for viability. We show that dicentric chromosomes can break in mitosis and that the chromosome fragments can be transmitted to progeny. Although the broken chromosomes were propagated in strains where the genes that were lost have no effect on viability, we observed that some broken chromosomes had persisting lethal effects.
MATERIALS AND METHODS
Mutations and chromosomes not described here are described by Lindsley and Zimm (1992). Abbreviated names for certain chromosomes are listed in Table 1. All flies were raised at 25° on standard cornmeal medium.
P element lines: Transformation of flies with each P element construct was carried out by standard procedures (Rubin and Spradling 1982), and additional insertion lines of these constructs were collected by mobilization with P transposase from the insertion P[Δ2-3, ry+](99B) (Robertsonet al. 1988). New insertions were mapped by segregation from dominant markers. P element lines are designated by numbers or letters; these do not indicate insertion sites. The insertions used here are listed in Table 1.
Abbreviations used in this work
The FLP construct P[ry+, hsFLP] has been described by Golic and Lindquist (1989). The construct P[ry+, 70FLP] is a similar heat-inducible FLP gene but is inducible to higher levels than P[ry+, hsFLP] with the same heat-shock (Golicet al. 1997).
pP[RS5] carries two FRTs in direct relative orientation and a whs selectable marker (K. G. Golic and M. M. Golic 1996). P[RS5]1B is a single recessive lethal insertion of this element on chromosome 4 at 102C2-5. All other insertion lines of P[RS5] described here were transposase-induced mobilizations from P[RS5]1B.
Construction and identification of chromosomes bearing inverted-orientation FRTs: Duplications of a whs-bearing P element can be recovered by screening for darker eye colors after mobilization with P transposase (Golic 1994). An insertion of the FRT-bearing P element P[RS5] on chromsome 4 was mobilized and the two-copy insertion line P[RS5]23 was identified as carrying inverted P elements. To provide another marker for crosses, a small duplication of the X chromosome (Dp(1;4)193, y+) on the left arm of chromosome 4 was crossed onto the P[RS5]23 chromosome. Females heterozygous for P[RS5]23 and Dp(1;4)193, y+ · spapol were heat-treated to induce recombination on chromosome 4 (Grell 1971), and a chromosome 4 carrying y+ on 4L with P[RS5]23 and spa+ on 4R was recovered (Figure 2). This chromosome is listed as y+·DcIV in Table 1. The designation DcIV is intended to indicate that the chromosome carries the inverted FRT-bearing elements that are required to form a dicentric chromosome upon FLP induction. The y+ gene of the duplication variegates.
P[RS5] and the Dc chromosomes. (A) Structure of the inverted P-element duplication inserted in a chromosome. The arrangement of insertions on DcY and X·DcYL is shown; their orientation on y+·DcIV is unknown. The line beneath the construct indicates the transcript of the whs gene (open boxes). Only the first intron of whs, which contains an FRT sequence (half-arrows), is indicated. The second FRT is downstream of the gene. Triangles indicate the P-element inverted terminal repeats. (B) Structures of the Dc chromosomes. The sites of the P[RS5] insertions are indicated as inverted FRTs; the circle represents the centromere. Heterochromatic segments are indicated by thick lines; euchromatic segments by thin lines.
To construct a Y chromosome that carried inverted FRT-bearing P-element insertions, we used Dp(4;Y)E spaCat as a Y chromosome with euchromatic sequences attached, into which P elements may insert. [This chromosome is probably the same as Dp(4;Y) in Muller and Edmundson (1957) and Lindsley and Zimm (1992). It carries most of chromosome 4 appended at the tip of YL.] The ci+ marker on this chromosome does not complement recessive ci alleles on regular fourth chromosomes and is hereafter ignored. An insertion of P[RS5] was recovered on the chromosome Dp(4;Y)E spaCat by crossing y w/Dp(4;Y)E spaCat; Sb P[Δ2-3 ry+](99B)/+; P[RS5]1B/+ males to w1118 virgins and recovering Sb+ male progeny with eye colors distinct from that of P[RS5]. These males were crossed to w1118 virgins to test for sex linkage. One line (P[RS5]W1) out of 98 tested was identified as an insertion on Dp(4;Y)E, spaCat. This insertion exhibits variegated expression of whs.
The insertion P[RS5]W1 was mobilized by P[Δ2-3, ry+] (99B) and flies with darker eyes were recovered. Genomic DNA from each line was screened by PCR with the Pfoot primer (Table 2) to identify lines with two close insertions, and rescreened with either the primer Pout5′ or the primer Pout3′ to identify insertions in inverted relative orientation. The line P[RS5]ZY2 had two P elements 0.7 kb apart with adjacent 3′ ends, and is referred to as DcY (Figure 2).
DcY was used to generate an X-Y rearrangement that placed the P[RS5]ZY2 double insertion on the right arm of the X chromosome. w1118/DcY males were X-irradiated with 800 rads in a Torrex 120D machine and brooded with w1118 virgins daily. Progeny from the broods of days six through nine were screened for w+ females and seven such females were retested for linkage of whs to the X chromosome. The line d7 was identified as an X-Y exchange that placed YL (and the chromosome 4 duplication) as a right arm of the X. The structure of this chromosome was confirmed by mitotic cytology. We refer to this chromosome as X·DcYL. The y+·DcIV, DcY and X·DcYL chromosomes are diagrammed in Figure 2.
Broken chromosome lines: We established lines from numerous male progeny carrying derivatives of the DcY, X·DcYL, and y+·DcIV chromosomes. Only those derivatives of DcY that carried all the Y fertility factors were propagated, because they were maintained in males as the only Y chromosome. Chromosomes derived from X·DcYL were recovered in males that also had a Y chromosome (alleviating the requirement for the Y fertility factors of the fragment chromosome) and were balanced with C(1)DX.
Chromosomes derived from y+·DcIV could not be balanced by standard procedures. These chromosomes were recovered in males that also carried a C(4)RM, spapol chromosome and were crossed to females carrying this compound chromosome to expand lines. We attempted to balance the derivative chromosomes by crossing in a chromosome 4 marked with ciD. However, progeny that carried the derivative chromosome and ciD proved to also carry C(4), and a balanced stock could not be established. These crosses indicated that the lines were accumulating additional copies of chromosome 4, which is not uncommon in stocks that carry mutations on chromosome 4. Because of the multiple fourth chromosomes segregating in these lines, we used the pch+ marker on Dp(1;4)193 to select for the chromosome. pch2 males do not survive unless they carry Dp(1;4)193 (on the chromosome derived from y+·DcIV). The duplication does not rescue viability in pch2 females, presumably because the pch+ gene on the duplication variegates.
PCR amplification from Drosophila genomic DNA: Genomic DNA was prepared from adults as described by H. Steller (personal communication) or by W. R. Engels (personal communication) for PCR tests. The primers used in these tests are listed in Table 2 and were used in the following combinations: PE5 + BSF1; BSF2 + w7703D; w7926U + PE3; and Pout3′ + Pout3′. The four amplified products span the FLP-excised derivative of P[RS5] and the interval between them in DcY. The distance between P elements in the y+·DcIV chromosome was too great for PCR amplification; thus the test between the P elements could not be performed. All samples of genomic DNA were tested with the primer pair w11678U/w10683D, which amplifies a 1-kb product from the endogenous white gene, to confirm that they contained amplifiable DNA.
P transposase-induced terminal deficiencies: To generate a terminal deficiency on chromosome 4, we crossed Sb P[Δ2-3, ry+](99B)/+; P[RS5]1B/+ males to y w; Dp(1;4)193, y+·spapol virgins and screened the progeny for Sb+ spa flies. To generate a terminal deficiency on the Dp(4;Y)E, spaCat chromosome, males of genotype w1118/DcY; TMS, Sb P[Δ2-3,ry+](99B)/+; ci gvl eyR svn were individually crossed to w1118; ci gvl eyR svn virgins and Sb+ sv males were recovered.
Cytology: Larvae carrying DcY, X·DcYL or their derivative chromosomes were identified by their sex. Larvae carrying y +·DcIV or its derivative chromosomes were identified by scoring the yellow phenotype of the ventral cuticle belts. Salivary gland polytene chromosomes were prepared as described by Lefevre (1976). Polytene chromosome in situ hybridization (Pardue 1986) was performed using the GENIUS system (Boehringer Mannheim, Indianapolis, IN). We used the chromosome 4 maps of Sorsa (1988) to identify bands on the chromosome. The plasmid pP[RS5] was used as probe for the P-element insertions. Their locations are listed in Table 1. The plasmid pHeT-A/FR3 (Younget al. 1983) was used as a probe for HeT-A sequences and was a gift from M. L. Pardue. Chromosomes were examined with brightfield and phase contrast optics.
Primers used for PCR amplification
Metaphase chromosomes were prepared as described by Gatti and Pimpinelli (1983). Chromosomes were stained with 0.5 μg/ml 4,6-diamidine-2′-phenylindole dihydrochloride (Boehringer Mannheim) in PBS and observed by epifluorescence with UV excitation. A Ziess (Thornwood, NY) Axioplan microscope, 100X Plan-NEOFLUAR objective, and FT 510, LP 520 filter set were used. Video image capture and processing was performed as described in Golic (1994).
Heat-shock regimens: To induce FLP expression in most cells of the male germline, males must be heat-shocked early in development since the hsp70 promoter of the FLP gene is efficiently induced only in the mitotically dividing stem cells (Bonneret al. 1984; M. M. Golic and K. G. Golic 1996). Broods of eggs from crosses were collected by transferring the parents of the cross to fresh vials every 24 hr. Each vial was heat-shocked at 38° for 1 hr in a circulating water bath as described by Golic and Lindquist (1989), 2 hr after the parents were transferred [effectively 2–26 hr after egg-laying (AEL)].
For the visualization of dicentric chromosomes in larval neuroblast cells, larvae carrying hsFLP2B and a Dc chromosome were heat-shocked at 38° for 1 hr and allowed to recover for 2 to 6 hr before dissection.
Crosses: For most crosses where the progeny were counted, individual males were crossed to two or three virgin females and the progeny were counted 14 and 18 days afterward. Crosses that produced <10 progeny were considered sterile. At least five fertile individual flies of a genotype were tested. Selected lines were examined more carefully.
Statistical analyses: Descriptive and analytical statistics were performed using StatView (Abacus Concepts, Berkeley, CA) and the P-stat program (provided by W. R. Engels) running on a Macintosh computer.
Determination of lethal phase: To determine whether offspring with chromosome fragments died, we set up single-pair matings of males with a fragment chromosome and females from a standard tester strain. Three days after setting up the crosses, the parents from vials with progeny were collected and placed together in a bottle to avoid including females that may not have mated. Eggs were then collected for 12 hr on plates of standard cornmeal-agar medium. Eggs were counted immediately after the egg-laying period, and the eggs that remained unhatched 48 hr later (48–60 hr AEL) were counted. Newly hatched larvae were removed from the plates during this interval and transferred to vials in order to measure viability between the first instar and adulthood.
RESULTS
Recovery of broken chromosomes
Dispensable chromosomes that form dicentrics: We constructed three chromosomes that form dicentric chromosomes by FLP-mediated sister-chromatid fusion (Figure 2). Two of the chromosomes are entirely dispensable in somatic cells. The DcY chromosome consists of a normal Y chromosome with a duplication of most of chromosome 4 appended to the tip of the YL arm. The chromosome 4 segment carries two copies of the P element P[RS5] in an inverted relative orientation, at which sister-chromatid fusion can occur. The y+·DcIV chromosome carries a duplication of the tip of the X chromosome, including the y+ gene, on the left arm. Inverted repeats of P[RS5] are carried on the right arm, in chromosome 4 material. Flies with fewer than two copies of 4 material have reduced viability, but when y+·DcIV is a supernumerary chromosome its loss has little effect. The X·DcYL chromosome carries the YL arm and chromosome 4 segment of DcY translocated onto the right arm of the X chromosome. The arm that undergoes sister-chromatid fusion is dispensable; the other arm, the X chromosome, is not.
Dicentric chromosomes in larval neuroblasts: The generation of dicentric chromosomes with DcY was demonstrated in neuroblast anaphase figures after hsFLP2B induction (Figure 3). A bridge of chromatin that spans the anaphase poles is seen. The efficiency of FLP-mediated dicentric formation is very high: approximately 70% of anaphase spreads from larvae with hsFLP2B and DcY contained a dicentric bridge (120 nuclei with bridges/166 nuclei). [Dicentric formation could not be scored in metaphase spreads as was previously done (Golic 1994) because the acentric portion generated from DcY includes only the tip of chromosome 4 and is very small; it cannot be seen in these spreads.] The frequency of dicentric chromosome formation is probably higher with 70FLP3A, which produces more FLP after heat-shock. X·DcYL should have a similar frequency of dicentric formation, because the arm that undergoes sister-chromatid fusion is the same as in DcY, and does have a qualitatively similar frequency of dicentric bridges in larval neuroblasts (not shown). With y+·DcIV we could not quantitate the frequency of dicentric formation because the small size of this chromosome precludes the recognition of either the dicentric or acentric portions.
Anaphase bridges in mitotic neuroblast cells. Anaphase figures are from controls (a) and from DcY-bearing larvae (b–d) that were heat-shocked as described in the text. The anaphase bridge in (b) clearly shows a staining pattern that is characteristic of YL and is symmetrically duplicated.
Many of the chromosome spreads that had bridges stretched between two poles of division also had condensed chromosomes at the poles, indicating that the chromosomes had just moved to the poles (Figure 3b). Occasionally, spreads were seen where a bridge of DNA was stretched between two nuclei with decondensed DNA, suggesting that the bridge remained stretched between daughter cells (Figure 3d). Other spreads appeared to contain broken bridges (Figure 3c). In fixed and squashed preparations it is not possible to determine with certainty the fate of a dicentric chromosome: we cannot tell whether an anaphase bridge would have broken, nor can we be sure that what appear to be broken bridges were not broken during preparation. We used genetic tests to make these determinations.
Dicentrics in the male germline: The markers present on the dispensable chromosomes were used to monitor their fates after the induction of FLP synthesis. In our experiments heat-shock was used to induce the 70FLP gene, and in the male germline this induction is limited to the mitotically dividing stem cells (Bonneret al. 1984; M. M. Golic and K. G. Golic 1996). The FRT-bearing insertions on DcY and X·DcYL are inserted between the chromosome 4 markers eyeless+ (ey+) and shaven+ (sv+). An FLP-mediated dicentric of this chromosome will lose the sv+ marker to the acentric portion (Figure 4). Loss of the acentric can be detected by uncovering recessive sv allele on the regular fourth chromosomes. The insertions on y+·DcIV lie slightly more proximal, between the grooveless+ (gvl+) and ey+ markers. The sparkling+ (spa+) gene is distal to the insertions, and loss of the acentric produced after dicentric formation with this chromosome can be detected by uncovering a recessive spapol allele on the other fourth chromosome.
Fragments recovered from DcY: The fate of a dicentric DcY chromosome in X/DcY males was assayed by heat-shocking males that carried 70FLP3A and also were homozygous for the chromosome 4 marker svn, and then test-crossing them to svn females. Offspring from this cross are sv+ only if they inherit the intact DcY chromosome. The progeny of the heat-shocked males included 276 shaven sons (Table 3A). All of the sons from 3 of the 71 fertile heat-shocked males were shaven; this likely indicates that dicentric chromosomes were formed in all the stem cells of those 3 males. Overall, approximately 10% of the sons from these crosses lacked the sv+ marker of DcY. Many retained more proximal markers of the DcY chromosome (described in a subsequent section). We conclude that these sons carry a broken fragment of DcY. We abbreviate fragment chromosomes derived from a dicentric chromosome as Fr chromosomes; in these cases, FrY.
In this experiment heat-shocked males produced 27% fewer sons bearing an intact DcY than males that were not heat-shocked. This reduction must also be a result of dicentric chromosome formation. Only one-third of this reduction can be accounted for by transmitted fragment chromosomes, implying that two-thirds of germline cells with dicentric chromosomes do not give rise to viable Y-bearing offspring. The loss of Y-bearing offspring must occur postmeiotically. If cells with a dicentric chromosome were eliminated premeiotically there would be an equal loss of X- and Y-bearing offspring. Instead there is a deficit of Y-bearing offspring.
Possible fates of a DcY dicentric bridge in mitosis. A dicentric chromosome that is stretched between the poles of mitotic division may do the following: (1) break and deliver a broken fragment chromosome to each daughter cell; (2) stretch between the daughter cells and perhaps block their further division and differentiation; (3) segregate to one pole, delivering the entire dicentric chromosome to one daughter cell; or (4) be lost from both daughter cells. The third and fourth fates both result in cells that have lost the entire DcY chromosome. The markers on DcY that distinguish between these fates are indicated. The presence of YS in cells was identified by suppression of a variegating whs insertion, as described in the text.
Production of chromosome fragments from DcY and X·DcYL males after FLP induction
Since DcY carries fertility factors required in primary spermatocytes, cells that lack substantial portions, or all, of DcY are unlikely to complete spermatogenesis. These events will not be detected with X/DcY males. Thus, the frequency of dicentric formation derived from the reduced recovery of DcY is probably an underestimate, because many dicentric chromosomes may result in cellular sterility, eliminating X- and Y-bearing gametes equally. To obtain a more accurate estimate of the frequency of dicentric chromosome formation, we also generated dicentrics with DcY in males carrying C(1;Y) so that the Y fertility factors of DcY were not required for fertility. Transmission of DcY was measured by the production of sv+ sons. There was a 48% reduction in the transmission of the intact DcY chromosome from heat-shocked males relative to the controls that lacked 70FLP, indicating that at least half of the germline cells in these animals have formed dicentrics (Table 3B). A corresponding number of shaven male progeny were recovered. Some of the shaven males recovered here retained more proximal markers from DcY and must therefore carry a FrY chromosome. Unlike the previous experiment with X/DcY fathers, in the experiment with C(1;Y)/DcY fathers, postmeiotic loss of Y-bearing progeny was not observed. Subsequent experiments will show that a portion of the fragment chromosomes exhibit dominant semilethality. The recovered fragment chromosomes that are broken within Y chromatin do not have this property. The majority of the FrY chromosomes that are broken distal to Y chromatin (within the appended chromosome 4) do exhibit this semilethality. Because the Y chromatin is needed for fertility in X/DcY males, there is a selective elimination of the fragments with breaks in Y chromatin in their germlines. We believe this accounts for the disparity: in X/DcY males most of the cells with nonlethal fragment chromosomes are eliminated in spermatogenesis, while the cells with semilethal fragment chromosomes produce an increased proportion of the functional gametes. In C(1;Y)/DcY males it is likely that semilethal fragments are also produced, but that they constitute a much smaller percentage of all FrY-bearing gametes and do not result in an obvious loss of Y-bearing progeny.
Many of the shaven sons of C(1;Y)/DcY males lacked all of the markers of the chromosome 4 duplication. These males may carry shorter FrY chromosomes or no FrY at all. To distinguish between shaven males that carried a fragment chromosome and those with no Y chromosome, we assayed for the suppressive effect of Y chromosome material on position-effect variegation. We crossed C(1;Y)/DcY males that carried 70FLP3A to females that were homozygous for the variegating P-element insertion P[>whs>]ZQa. This line carries two P elements at 84D-E of chromosome 3 that variegate for eye color because of a position effect on their white genes (Ahmad and Golic 1996). X0 males have white eyes with occasional red spots. Variegation in this line is suppressed by addition of a Y, so that XY males have eyes that are mostly red. Flies with YS only are a pale yellow color with red spots, but can still be distinguished from flies lacking any Y material. Therefore, this line provides a suitable test for distinguishing between flies that carry DcY, fragments of DcY, or flies that lack DcY entirely. By crossing C(1;Y)/DcY males to the ZQa-bearing females and evaluating variegation in sons, we found that chromosome loss was very rare following dicentric formation in germline cells of these males (Table 4). We conclude that most of the shaven male progeny from C(1;Y)/DcY males carry FrY chromosomes. Those that lack any Y are probably the result of occasional nondisjunction.
DcY chromosome loss test from C(1;Y)/DcY males
It is clear that dicentric chromosomes can be made at high frequencies in the male germline and they often break. However, one of the possible fates of a dicentric chromosome in mitosis—where daughter cells are tied together by an unbroken anaphase bridge—cannot be identified with these tests. If this occurred in the male germline we expect that cell proliferation would be blocked and reduce the production of gametes from such males. Induction of FLP in the X/DcY genotype in fact did sterilize 20% of the males, and we considered that this might result if all germline stem cells in these males had formed dicentric bridges that did not break. Alternatively, these germline cells may have lost one or more fertility factors after dicentric breakage. To determine whether the observed male sterility resulted from loss of Y fertility factors or from stretched bridges between cells, we generated males that carried DcY and a C(1;Y) that carried all of the Y fertility factors and tested their fertility after FLP induction. We used three copies of 70FLP to increase the frequency of dicentric formation. When FLP was induced, 68% of the X/DcY males were sterile—almost a fourfold increase relative to noninduced males. However, in males with C(1;Y), the induction of FLP caused no increase in sterility (Table 5). We conclude that sterility in the X/DcY genotype results from the loss of fertility factors on DcY after dicentric chromosome breakage. While we cannot absolutely rule out a low frequency of bridges that stretch but do not break, our results can be sufficiently explained by assuming that dicentric bridges break.
Fragments from X·DcYL: The Y fertility factors of X·DcYL are not required for fertility in X·DcYL/Y males, and offspring carrying fragments of X·DcYL (X·FrYL chromosomes) can be detected by the presence of the X portion and loss of sv+. When heat-shocked males were mated to attached-X females, 41% of X·YL-bearing gametes were recovered as shaven male progeny (Table 3C). These males must carry an X·FrYL chromosome. There was equal transmission of the X·YL chromosome (X·DcYL and X·FrYL) and its homolog. This confirms our observations with DcY: dicentric chromosomes can break in mitotic divisions and the centric fragments can be transmitted.
The frequency of dicentric formation: The recovery of 48 and 41% fragment chromosome-bearing males implies that almost half of all viable germline cells had experienced dicentric chromosome formation and breakage with the DcY and X·DcYL chromosomes. Another estimate of the frequency of dicentric formation can be made from the number of male parents that transmit only the broken forms of Dc chromosomes. Four percent of fertile X/DcY males transmitted only FrY and X chromosomes: in every stem cell of these males the DcY chromosome must have undergone dicentric formation. Assuming that there are roughly 10–16 germline stem cells in each male (Lindsley and Tokuyasu 1980), and that they are equally likely to undergo dicentric formation, the frequency of dicentric formation per cell (x) can be estimated by solving the equations 0.04 = (x)10 and 0.04 = (x)16. Thus, in X/DcY males the frequency of dicentric formation is between 72 and 82%. This calculation can also be performed with the C(1;Y)/DcY and X·DcYL/Y genotypes: 11 of 30 fertile C(1;Y)/DcY males transmitted only fragment chromosomes; the estimated frequency of dicentric formation in this genotype is between 90 and 94%. Nine of 49 X·DcYL/Y males produced only fragment chromosomes, for a frequency between 84 and 90%. Thus, the frequency of dicentric formation with these chromosomes in the male germline is at least 41%, and probably 70–90%. This high frequency allows us to easily determine the consequences of dicentric formation in germline mitotic divisions and to produce many isolates of chromosome fragments.
Male sterility after DcY dicentric formation
Dicentric formation with y+·DcIV: Because structural differences between centromeres can affect the behavior of a dicentric chromosome in female meiosis (Novitski 1952), we thought it worthwhile to examine the fate of dicentric chromosomes that carried a normal centromere. Chromosome 4 can exist as a dispensable supernumerary chromosome and is suitable for such experiments.
We identified chromosome 4 lines that carried inverted repeats of P[RS5] by heat-shocking and mating males that carried hsFLP2B and a chromosome 4 with two copies of the element. We looked for loss of distal markers as an indicator of the formation and breakage of dicentric fourth chromosomes. In these crosses we followed the sparkling+ (spa+) gene as a marker for the distal part of chromosome 4. Males from 6 of the 12 lines examined produced some sparkling progeny when FLP was induced, indicating that the insertions in each of these lines are inverted with respect to one another, that dicentric chromosomes were produced, and that the tip of the chromosome was lost. The insertion line 23 was chosen for further characterization; the P elements in this line are inserted at 102C2-C5 of the right arm.
Because there is a paucity of useful markers on chromosome 4, we generated a modified chromosome 4, based on the double insertion line 23, that could be followed through dicentric formation and mitosis. To provide a marker with which to follow inheritance of this chromosome, a duplication of the tip of the X chromosome, including the y+ gene, was recombined onto the left arm of chromosome 4, generating the chromosome y+·DcIV (Figure 2). This duplication is very small and adds very little chromatin to chromosome 4. Therefore, we expect that the strength of the chromosome 4 centromere should not be altered by this duplication.
When flies that carry y+·DcIV chromosome are crossed to C(4)/C(4) flies, the y+·DcIV chromosome is dispensable in the progeny. We assayed the consequences of dicentric formation in the male germline with this chromosome, as we had with DcY and X·DcYL. Broken fragment chromosomes (y+·FrIV chromosomes) were produced (recognized as yellow+ sparkling progeny); chromosome loss (yellow sparkling progeny) was not observed (Table 6). There was also a 9% reduction in the recovery of the y+·DcIV chromosome. With two copies of 70FLP there was a further reduction in the recovery of y+·DcIV (from 91 to 71%), and there was an increase in the number of progeny with fragment chromosomes (line 3 of Table 6). However, the number of y+·FrIV-bearing progeny was much less than the number of missing y+·DcIV chromosomes in both cases. This implies that there were many fragment chromosomes that could not be recovered. There were no males that produced only y+·FrIV chromosomes.
Production of chromosome fragments from y+·DcIV males after FLP induction
It is possible that loss of chromosome 4 material decreases the recovery of fragment chromosomes from the y+·DcIV/4 males as a consequence of the resultant germ-cell aneuploidy. We tested the fate of the chromosome 4 dicentrics produced by FLP in males where y +·DcIV was present as a supernumerary chromosome 4. The transmission of the intact y+·DcIV decreased from 91%, when the homolog of DcIV was a normal 4, to 62%, when the homolog of DcIV was a compound chromosome 4 (Table 6). This suggests that in y+·DcIV/4 males many cells that made dicentrics were inviable because of chromosome 4 aneuploidy, and failed to produce any gametes. With the addition of a third copy of chromosome 4 these cells can proceed through spermatogenesis. Thus, the cells that make functional sperm now contain a higher percentage of cells with dicentric chromosomes, and the transmission of the intact y+·DcIV chromosome, compared to its homolog, is reduced. There was no significant effect of dicentric formation on fertility (Table 6).
While the presence of C(4)RM increased the measurable frequency of dicentric formation, there was no commensurate increase in the recovery of progeny carrying y+·FrIV chromosomes. Just as with the fragments of DcY, the reduced recovery of progeny carrying the y +·FrIV chromosome must result from a postmeiotic effect. The evidence provided below suggests that many y +·FrIV chromosomes are not recovered because they are lethal.
Chromosome fragments generated by P transposase: Although broken chromosomes are very rarely recovered after irradiation of normal males or females (Muller and Herskowitz 1954; Masonet al. 1997), such chromosomes are recovered more frequently after irradiation of mu2 females (Masonet al. 1984), or after P-element mobilization (Levis 1989). This suggests that the method by which a chromosome is broken might influence the consequences of that break. To compare the behavior of chromosome fragments generated by breakage of a dicentric chromosome with that of fragments produced by transposase, we induced terminal deficiencies with P transposase on DcY and on the P[RS5]1B chromosome 4. These were identified by loss of the distal portion of the chromosome. Transposase-induced fragments of DcY, called FrYΔ chromosomes, occurred at frequency of 0.26 events/fertile parent (12 independent events found from 46 transposase-bearing male parents). A line was established from each FrYΔ-bearing male. Other exceptional males were observed: one that was gvl+ ey sv+; two that were gvl ey sv+ and retained the whs P element; and a fourth that was gvl ey sv. These were sterile and were probably instances where interstitial portions of the chromosome had been deleted (see Engels and Preston 1984).
Fragments of chromosome 4, called FrIVΔ, were recovered at the lower frequency of 0.02 events/fertile parent (5 independent events from 270 male parents). Four of the spa male progeny were crossed to y w; C(4)RM, spapol virgins. Two were sterile and the third transmitted only the homolog to progeny. The fourth male was fertile and transmitted a chromosome derived from the P[RS5]1B chromosome. This line was named FrIVΔA; it retains a copy of P[RS5] at 102C2-5 but lacks chromosome 4 material distal to 102D1-2.
FrY and FrIV chromosomes were recovered much more readily by dicentric chromosome formation than by transposase. FrY chromosomes were produced approximately three times more frequently after dicentric formation than after exposure to transposase, and FrIV chromosomes were recovered 16 times more frequently.
Characterization of broken chromosomes
Dominant semilethality characterizes some fragment chromosomes: The recovery of fragment chromosomes from dicentric chromosomes demonstrates that a chromosome broken in this way can be viable. However, the overall deficit of Dc chromosomes suggests that some Fr-bearing gametes or zygotes are inviable. This effect was especially pronounced with y+·DcIV, where in one experiment 95% of the fragment chromosomes were not recovered (Table 6, line 5). This hypothesis is strongly supported by our observations (below) that many of the recovered fragments exhibit a dominant semilethality.
Transmission of Y chromosome fragments: FLP synthesis was induced in X/DcY males, those males were mated, and lines of independently generated FrY chromosomes were established for further characterization. Two groups of FrY lines are apparent (Figure 5). In the first group, which includes the DcY control and four FrY lines, the Y was transmitted at a normal rate. The second group, with ten FrY lines, exhibited a great reduction in the rate of FrY transmission, which is observed as a reduction in male progeny from an X/FrY father. Many of these low-transmission lines also showed sterility of some males (the y-axis of Figure 5). Males from a given line showed variability in their transmission ratios, but this variability was not heritable (not shown).
We have considered two possible explanations for the reduced transmission of the fragment chromosomes. The transmission of FrY chromosomes might be reduced by the elimination of FrY-bearing gametes during meiosis or spermiogenesis. Alternatively, some of the zygotes that receive FrY-bearing gametes might die. We distinguished these two possibilities by scoring egg-to-adult survival of zygotes from parent males carrying a low-transmission fragment chromosome. Crosses with males from the FrY1A line showed a significant level of embryonic lethality; this lethality can account for two-thirds of the total reduction in fragment-bearing progeny (Table 7).
Lethal and sterile effects of fragment chromosomes. The transmission rate and sterility associated with individual lines are indicated on the x- and y-axes, respectively. Transmission rates are given as the number of Fr-bearing progeny divided by the number of progeny with the homolog. The values indicated are the unweighted means of at least five fertile matings with single males for each independently isolated Fr chromosome. These values were measured in the first generation after the initial isolation. Transmission rates less than 0.6 were significantly different from the control (P < 0.01), as determined by a randomization test.
Fragment chromosomes of DcY that were generated by P transposase could also be divided into two groups by their rates of transmission: three lines were transmitted at a normal rate, and one line (FrYΔ14A) showed low transmission similar to that of low-transmission FrY lines (Figure 5).
We measured the transmission from 12 X·FrYL lines (Figure 5). Two showed decreased transmission of the X·FrYL chromosome (X·FrYLd26e and X·FrYLd34r).
Transmission of chromosome 4 fragments: The transmission of FrIV chromosomes and of fourth chromosomes from three control genotypes was assayed. For controls, we tested the intact chromosome y+·DcIV, the chromosome Dp(1;4)193, y+·spapol (from which the y+ duplication on y+·DcIV was derived), and the chromosome Df(4)G. Df(4)G chromosome was assayed because it is a deficiency of all chromosome 4 material distal to 102E2-10, and therefore lacks a region similar to that missing from the y+·FrIV chromosomes. It is capped with the tip of the X chromosome. All three control chromosomes were transmitted at the same rate as their homologs [either a normal 4 marked with ciD or C(4); Figure 5]. We measured transmission of y+·FrIV chromosomes from males heterozygous for C(4). Most lines exhibited reduced transmission of y+·FrIV, with transmission ratios ranging from 0.01 to 0.86 relative to the C(4) homolog (Figure 5). Egg-to-adult counts with the low-transmission y+·FrIVC72 line showed that its reduced transmission could be entirely accounted for by embryonic lethality (Table 7).
Males that carried the transposase-induced fragment chromosome FrIVΔA transmitted it at a ratio comparable to the controls. There was no reduced transmission of FrIVΔA, although it is associated with sterility when heterozygous with ciD (Figure 5). Males that carried FrIVΔA and C(4) were usually fertile (not shown), suggesting that the sterility of FrIVΔA/ciD males is caused by aneuploidy.
At least some of the reduced transmission of the y +·FrIV chromosome is caused by the accumulation of additional copies of chromosome 4 in the lines carrying y+·FrIV (materials and methods). Both viability and meiotic segregation of fourth chromosomes are affected by this aneuploidy (Grell 1972, cited in Ashburner 1989). However, this cannot account for the reduced transmission of y+·FrIV chromosomes from males with newly induced dicentrics, because these parents had not accumulated additional fourth chromosomes (Table 6). Zygotic lethality of the fragment chromosome likely accounts for the reduced recovery from the original parents as well as a portion of the reduced transmission in further crosses.
We conclude that some fragment chromosomes generated from DcY, X·DcYL, and y+·DcIV share a similar defect that causes embryonic lethality and leads to reduced transmission of the fragment chromosome. The similar behavior of these chromosomes must be due to their common feature: the broken chromosome end.
Structure of fragment chromosomes: Muller found that many chromosomes with apparently terminal deficiencies were in fact more complex rearrangements that maintained a telomere at the end. In order to determine whether the fragment chromosomes that we generated by dicentric breakage had been involved in further rearrangements, and in the hope of providing insight into the nature of the semilethal fragments, we characterized the structure of the recovered Fr chromosomes with genetical and cytological assays. [The detailed results are presented in Ahmad (1997).]
Viability of zygotes with fragment chromosomes
Genetic structure of the recovered chromosome fragments: In some of the crosses where dicentrics were made with DcY or X·DcYL, flies also were homozygous for a chromosome 4 marked with gvl eyR svn. In these cases, markers both proximal and distal to the site of the FRT insertions also could be assayed in progeny. In other cases, the positions of breakpoints were mapped in further crosses. Fragments of X·DcYL were also characterized with respect to whether they carried all the male fertility factors of YL. Many fragment chromosomes were missing proximal markers (Figure 6). In no case was a distal marker retained while a more proximal one was absent. We conclude that the site of dicentric chromosome breakage is not limited to a single location.
Mapping of fragment chromosome breakpoints. Breakpoints were mapped by complementation, by polytene cytology, and by PCR amplification. The arms of the chromosomes involved in dicentric formation are indicated as dicentric bridges, with the long arm of the Y and the attached duplication of chromosome 4 on the left, and the right arm of chromosome 4 on the right. Large circles represent the centromeres; heterochromatic segments are indicated by thick lines; euchromatic segments by thin lines. The inverted P[RS5] elements are indicated beneath each chromosome. The orientation of elements on DcY is as indicated, but the orientation on DcIV is unknown and the elements are arbitrarily drawn. The recovered chromosomes extend from the left centromere to the indicated positions. The interval where each breakpoint mapped is indicated by the horizontal lines above or below the dicentric bridge. A dashed line indicates that the exact terminus is uncertain: one P element was detected by PCR, but the possibility exists that a second element is present (see text). Fragment chromosomes were classified as normal-transmission or low-transmission as described in Figure 5. * indicates cases in which the fragment chromosome has acquired new sequences at its end. Only the Fr chromosomes for which the breakpoint was mapped and which were classified with respect to recovery are shown here.
We expected that if a dicentric bridge broke asymmetrically in mitosis one daughter cell would receive a short chromosome fragment and the other would receive a long fragment. When dicentric formation was induced in males carrying X·DcYL and homozygous for the gvl eyR svn chromosome 4, the recovery of progeny lacking gvl+ and ey+ implied that asymmetric breakage of the dicentric bridge did occur, because these flies must carry short fragments. A fly with the corresponding long fragment should be gvl+ ey+ sv; however, there were five males that produced only grooveless eyeless shaven offspring. Similarly, there was one C(1;Y)/DcY male that produced only this kind of offspring after dicentric formation. Only the short fragment chromosomes were recovered, although the corresponding long fragments must also have been generated. This led us to ask whether any of the fragment chromosomes that we recovered were long fragments. For both FrY and X·FrYL chromosomes, the breakpoints that were mapped distal to the ey+ gene could be located in one of two different regions of the chromosome. An eyeless+ shaven fly might carry a short fragment broken between ey+ and the site of sister-chromatid fusion, or a long fragment broken anywhere along the length of the YL arm, so that the recovered chromosome is deficient for material distal to the P insertions and duplicated for some proximal material. Such a chromosome will also carry two P-element insertions. This can be detected using PCR amplification with primers specific to the P element ends. We screened, by PCR, all FrY and X·FrYL lines with breakpoints that had been genetically mapped as distal to the ey+ marker. (The y+·FrIV chromosomes could not be analyzed in this way because the distance between the two P elements is too great.) Although we found many lines that carried one P element, no lines appeared to carry more than one P element. However, if DNA near the end of a chromosome cannot be easily amplified, perhaps because telomeric DNA is compacted into an unusual chromatin structure, then this result would be misleading. Therefore, we do not conclusively rule out the possibility that some fragment chromosomes carry two P elements near the terminus. However, if any fragments do include sequences of both P elements, we imagine that the break must be close to those sequences. Cytological analyses (below) confirmed this conclusion.
Cytological structure of the recovered chromosome fragments: Polytene chromosomes of larvae from fragment chromosome lines were examined in order to confirm the genetic characterization. Eleven y+·FrIV chromosomes were examined: all ended at 102C1-5 (Figures 6 and 7), which is the site of sister-chromatid fusion (i.e., the location of P[RS5] elements). Nine FrY chromosomes were examined cytologically. They exhibited breaks at various points in the chromosome 4 material (Figure 6). X·FrYL chromosomes were examined by polytene or diploid cytology (Figure 7). These ended at differing positions, with breakpoints in euchromatin or in heterochromatin. One X·FrYL breakpoint was located near the site of sister-chromatid fusion. There was no evidence of other rearrangements on these chromosomes. We conclude that the recovered chromosomes have terminal deficiencies. We also confirmed by cytology that FrIVΔA and four FrYΔ (4, 9B, 11B, 14A) fragment chromosomes induced by P transposase truly were terminal deficiencies. The other putative FrYΔ lines were not characterized.
Cytology of fragment chromosomes. (a–h) Salivary cytology. The fragment chromosome end is indicated by an arrow. If the Fr is not paired with its normal homolog, then the same band on the regular fourth chromosomes is indicated by an arrowhead. (a) Normal chromosome 4. (b–d) Independently isolated y+·FrIV chromosomes heterozygous with a normal chromosome 4. The fragments each end at 102C1-5. (e–g) Independent FrY chromosomes: (e) FrY2A (broken at 102B2-C2); (f) FrY3F (102D2-5); (g) FrY8A (102B2-5). (h) FrYΔ14A (broken at 102D2-6). (i–k) Mitotic cytology of X·FrYL chromosomes. The fragment chromosome ends are indicated by arrows, the intact Y or X·DcYL chromosomes by arrowheads, and the breakpoint is indicated by a line between the two chromosomes. Lines in which the X·FrYL chromosome lacked a YL fertility factor were selected: (i) metaphase of X·DcYL/Y. (j) X·FrYL d5d/Y, the most distal bright block on YL is missing. (k) X·FrYL d33a/X·DcYL, the most distal bright block on YL is missing.
The breakpoints of normal-transmission fragment chromosomes were found in all the genetically defined intervals along the chromosomes, and all but one of the normal-transmission fragments lacked P[RS5] sequences. However, all low-transmission FrY, X·FrYL, and y+·FrIV chromosomes were broken near the P[RS5] elements used to generate the dicentric chromosome, and retained a P[RS5] remnant near this end (Figure 6). The coincidence of the low-transmission with the presence of P-element sequences near the broken end suggests that these sequences may play a role in determining the properties of fragment chromosomes. One y +·FrIV chromosome, which we first categorized as a low-transmission line, later changed to normal-transmission behavior. We examined the salivary cytology of this chromosome after the change in transmission frequency. The chromosome ended at 102C1-5, as do the other low-transmission lines, but did not have P-element sequences (not shown). It is possible that these sequences were present on the originally isolated chromosome but were later lost, resulting in the change in behavior.
Hybridization for telomere sequences on the recovered chromosome fragments: If a telomere is normally an essential chromosomal element, broken chromosomes should not be viable unless they have acquired a new telomere. We used in situ hybridization to examine the ends of the FrY, X·FrYL, and y+·FrIV chromosomes for the presence of HeT telomeric sequences. Fourteen of the fragment chromosome lines showed no hybridization at the end of the Fr chromosome, although hybridization to the tips of other chromosomes and to the tip of the regular chromosome 4 was observed (Figure 8). One X·FrYL chromosome did hybridize with the HeT probe and therefore acquired HeT telomeric sequences, and one FrY chromosome also showed a slight but repeatable hybridization signal and may also have an HeT telomere (Figure 8). The y+·FrIV chromosome that showed a change in behavior did not show a hybridization signal. Although in situ hybridization methods are sufficiently sensitive to detect a single copy of an HeT sequence at the end of a chromosome (Traverse and Pardue 1988), we cannot exclude the possibility that some fragment chromosomes may have truncated HeT elements that were not detected by hybridization.
Detection of HeT telomeres on fragment chromosomes. Polytene normal and fragment chromosomes after hybridization with an HeT probe are shown. The tips of Fr chromosomes are indicated with arrows; arrowheads indicate normal tips. Hybridization signals are present on the regular fourth chromosomes. (a–b) DcY viewed by (a) phase-contrast and (b) brightfield optics. HeT is present on the chromosome 4 duplication. (c–d) y+·FrIV chromosomes. No hybridization signals are present on the fragment ends. (e) FrY32D (102C-D); a weak signal is present at the tip of the fragment. (f) X·FrYL 28m (102D); a strong signal is present on the fragment chromosome end.
DISCUSSION
The fate of dicentric chromosomes in mitosis: We constructed chromosomes that are dispensable for cell viability and that form dicentric chromosomes by FLP-mediated USCE. In the experiments reported here, we deduced the fate of dicentric chromosomes in mitosis by inducing FLP in males and examining their progeny. In the male germline the 70FLP gene that we used is only induced in the mitotically dividing stem cells (Bonneret al. 1984; M. M. Golic and K. G. Golic 1996). We estimate that dicentric formation with the DcY and X·DcYL chromosomes occurs in 70–90% of germline stem cells. It appears that the predominant fate of the dicentric chromosomes in those mitotic divisions is breakage: for example, 35% of C(1;Y)/DcY males transmitted only broken forms of DcY. Loss of the dicentric chromosomes in those divisions was not observed. We cannot exclude the possibility that some dicentric chromosomes are stretched but not broken in mitosis, as the two daughter cells tied together in this way are not likely to be viable and would not produce gametes. However, this would require that mitoses with the same dicentric chromosome produce varying results.
In contrast, a dicentric X chromosome that is stretched at the reductional division in female meiosis usually does not break (Sturtevant and Beadle 1936). Novitski (1952) found that a dicentric chromosome involving X chromosomes would break when the chromosomes also carried part of the Y chromosome attached to the short arm. Novitski categorized a centromere as strong or weak by whether one centromere in a dicentric chromosome could drag a weaker one at division. While the normal X centromere is weak, compound X-Y chromosomes have strong centromeres, and this strength is attributable to the large blocks of surrounding heterochromatin (Lindsley and Novitski 1958). DcY and X·DcYL are similar in size and structure to the strong centromere chromosomes examined by Novitski, and their breakage in mitosis is consistent with having strong centromeres. While Novitski did not examine the strength of a chromosome 4 centromere, y+·DcIV is much shorter, has much less heterochromatin around its centromere than a normal X chromosome, and is therefore expected to be weak. However, dicentric chromosomes formed with y+·DcIV also frequently broke in mitosis. Breakage also seemed to be the most likely fate for dicentric chromosomes formed from a normal X chromosome in mitotic cells (Golic 1994). Our experiments revealed no obvious differences between the relative strengths of these centromeres in mitosis. Instead of supposing that all the Dc chromosomes have strong centromeres, it seems more plausible that dicentric bridges have an inherently higher likelihood of breakage on the mitotic spindle than on a Meiosis I spindle.
Although we observed that a dicentric chromosome generated by FLP-mediated USCE is not lost in mitosis, ring-X chromosome loss can occur in the early syncytial embryo or during cellular mitoses (Zalokaret al. 1980; Bachiller and Sanchez 1991). It is thought that sister-chromatid fusion generates a dicentric chromosome, and this dicentric chromosome is lost (Hinton 1959). A ring chromosome that undergoes sister-chromatid fusion will generate a dicentric ring chromosome, with a double bridge at anaphase. Perhaps the added strength of the second bridge prevents breakage and the ring is then lost.
The fact that fragment chromosomes can be recovered suggests that the broken ends have been repaired by the acquisition of telomeres. However, most of the fragment chromosomes we recovered have not acquired HeT sequences at their termini. Broken chromosomes recovered by other means also do not usually acquire HeT sequences (Levis 1989; Biessmannet al. 1990). These groups have demonstrated that Drosophila does not require specific DNA sequences at chromosome termini for viability. The fragment chromosomes that we generated by dicentric breakage provide another example.
Broken chromosomes in Saccharomyces are mitotically unstable (Sandell and Zakian 1993). With our fragment chromosomes, mitotic loss would not necessarily be apparent. For instance, the y+ marker on y +·FrIV chromosomes variegates, so it cannot be used as a marker for chromosome loss. The chromosome 4 marker gvl+ might be useful to visualize loss with FrY or y +·FrIV, but probably only if loss occurred very early in development. We saw no evidence that this occurred. Very early loss of a X·FrYL chromosome in females might produce gynandromorphs, but these chromosomes were usually kept only in males, where loss would certainly lead to cell death. It is conceivable that mitotic loss could account for the male sterility that we observed with FrY chromosomes.
Muller (1941) argued that chromosomes with a single break could not be recovered after irradiation of the male germline. Subsequent experiments have demonstrated that such chromosomes only rarely survive (Leet al. 1995; Masonet al. 1997). Levis (1989) proposed that the failure to recover broken chromosome fragments was due to the elimination of fragment-bearing cells by a checkpoint that responds to DNA damage. However, fragmented chromosomes produced by breakage of dicentric chromosomes or by P-element transposase (Levis 1989) must evade this response mechanism. The mechanism of chromosome breakage may play a role in stimulation of a checkpoint response. For example, a break induced by P transposase may escape detection if a broken end is masked by a bound transposase complex. Additionally, the number of broken ends may be crucial for adequate stimulation of a DNA damage response. Breakage of a chromosome by X rays produces two broken ends in a cell, but breakage of a dicentric chromosome in mitosis produces two daughter cells, each with a single broken chromosome end. It may be that the single broken end can escape detection.
Breakpoint distribution in fragment chromosomes: The breakpoints of the recovered fragment chromosomes show a striking nonrandom distribution. One surprising aspect is that such a large fraction of the fragment chromosomes have breaks near the site of sister-chromatid fusion (Figure 6). The question arises, is this distribution the result of selection, or of an intrinsic tendency to break near the middle of the anaphase bridge? Selection is certainly involved in the determination of the FrY breakpoints. The males from which we recovered these fragment chromosomes carried the DcY as their only Y chromosome. Since the Y is needed in primary spermatocytes for male fertility, there was undoubtedly a strong selection against chromosomes that broke within Y chromosome material and lost Y fertility factors as a consequence. The existence of this selection is confirmed by the fact that many X·FrYL chromosomes have breaks within Y chromatin. These fragment chromosomes were recovered from males with an additional free Y, so the fertility factors on X·DcYL were redundant. Still, almost half of the X·FrYL breaks are located within the chromatin of the chromosome 4 duplication, and the size of this portion of the chromosome is small in comparison to the long arm of the Y, which makes up the rest of this arm. The tendency for breakage near the site of sister-chromatid fusion is also seen with the y+·FrIV chromosomes. Those characterized in this paper were all recovered from males that also carried a compound chromosome 4 as the homolog. Thus, there should have been no selection for cells with large fragments of chromosome 4. In fact, cells that lost more of the dicentric would be more nearly euploid than cells that lost only a little. Yet, there is still a pronounced clustering of breaks very near the FRT-bearing P elements. It seems likely that breaks do in fact occur in this region in disproportionately large numbers. This property might be the result of a structural distinction in the DNA that crosses between the two fused sister chromatids. The actual frequency of breakage in this region is no doubt even higher than represented in Figure 6 because, with only one exception, all the chromosomes with a break distal to one P element are semilethal; many chromosomes like this would not be recovered.
A second clear departure from random breakage in the recovered fragment chromosomes is the apparent absence of long fragment chromosomes. Such chromosomes should have arisen as the complementary product of breakage events that produced short fragment chromosomes. Cytological analysis of the fragment chromosomes revealed none with a long inverted duplication at the end. Broken chromosomes in maize are prone to further deletion by breakage-fusion-bridge cycles (BFB cycles; McClintock 1939). In maize, telomeres are required to maintain chromosome stability and prevent the end-to-end fusions that result in BFB cycles. It is conceivable that long fragment chromosomes in Drosophila are similarly prone to further deletion. A terminal inverted duplication may engender repeated rounds of breakage in the germline. If FLP recombinase was still present in cells that received a long fragment, the inverted P[RS5] elements within the duplicated region could again recombine to generate a dicentric bridge at the next division. Alternatively, the inverted duplication of a long fragment might promote dicentric formation by recombination within a foldback structure, as discussed below. If long fragment chromosomes with inverted duplications at their ends do initiate BFB cycles in Drosophila, breakage of the dicentrics would produce short fragments at the expense of long fragments.
We note, however, that instability of the recovered fragment chromosomes is rare. We have not detected further deletion with the recovered FrY lines after their isolation. One X·FrYL line has lost the YL arm since it was first isolated (not shown), but the mechanism of its deletion need not be related to its lack of a normal telomere. Spontaneous deletion or recombination between the X·FrYL and the Y could account for the YL loss.
Fragment chromosome lethality: While many newly broken chromosomes survive, it is equally clear that some broken chromosomes are not recovered, apparently because they cause the death of zygotes that received them. We found that many of the recovered fragment chromosomes cause partial lethality. Clearly then, many such chromosomes could also have been lost by zygotic death in the first generation after the induction of dicentric chromosome formation. And, as just discussed, fragment chromosomes with long inverted duplications are not recovered. These also may be lethal.
The flies that receive a semilethal fragment chromosome and survive appear normal with no phenotypic indication in the adult soma that abnormal cell death occurred during development. Direct counts of egg viability demonstrated that the lethal effect of a fragment chromosome is limited mostly or entirely to embryos. However, frequent male sterility was another consequence in the survivors. Lethality and sterility are not unique to the fragment chromosomes generated by breakage of a dicentric: we also isolated a fragment chromosome by destabilization of a P element that behaves similarly.
The distinguishing feature of the semilethal fragment chromosomes is that they all retain at least one of the FRT-bearing P elements where sister-chromatid fusion occurred. The nonlethal fragment chromosomes, with one exception, all lack these P elements. We suggest three models that may account for the lethality associated with low-transmission fragment chromosomes.
First, the presence of a P element adjacent to the chromosome terminus might cause the lethal effect. The P-element inverted repeats are constitutively bound by host-encoded proteins, and one of these proteins, encoded by the mus309 gene, is the Drosophila homolog of the human Ku70 protein (Beallet al. 1994; Beall and Rio 1996). In mammals, this protein is involved in forming a competent complex to signal the presence of double-strand breaks, or to direct their rejoining (Taccioliet al. 1993). The mus309 gene and the Saccharomyces homolog HDF1 have also been implicated in end-rejoining DNA repair events (Kaufmanet al. 1989; Staveleyet al. 1995; Siedeet al. 1996). The proximity of a chromosome end and mus309-binding sequences may sometimes signal the cell that a double-strand break is present, leading to a permanent cell-cycle arrest or cell death. There may be other sequences that cannot be tolerated near a chromosome terminus. A function of the normal telomeric structure of Drosophila may be to ensure that such incompatible sequences remain far from the ends of chromosomes. Arguing against this model is the fact that the broken chromosomes recovered by Levis (1989) have a P element at the terminus and are not deleterious to the flies that carry them. However, in those chromosomes, only a single P-element inverted repeat was present. Perhaps both the 5′ and 3′ ends of the P element are necessary to stimulate the response.
A second model supposes that the lethality results from sequences within the P element. The 5′ portion of the whs mini-gene is present in the P[RS5] element that remains in the fragment chromosomes. (The remainder of the whs gene was removed as a consequence of FLP-mediated excision.) If two P elements are still present, the distal element will lie so that whs transcription is directed towards the tip of the chromosome. (This is true for fragments of DcY and X·DcYL; the orientation of elements in DcIV is not known.) If the physical end of the broken chromosome is very near the P element, then transcription directed from the promoter of whs might proceed to the end of the chromosome. Perhaps this transcription destabilizes proteins that otherwise protect the end of the chromosome and initiates a cell-cycle checkpoint response. This might provide an evolutionary rationale for the silencing that is typically observed with transgenes that insert near Drosophila telomeres (Hazelrigget al. 1984; Leviset al. 1985; Karpen and Spradling 1992; Wallrath and Elgin 1995). Chromosome stability was not affected by transcription through a normal yeast telomere (Sandellet al. 1994), but this result may not apply to Drosophila telomeres. If telomeres in Drosophila can be destabilized by transcription, then a conditional telomere might be produced by placing a regulatable promoter in a similar position with respect to the end of the chromosome. Such a chromosome would have distinct advantages for the study of telomere function.
Finally, a third model is based on the possibility that the lethal chromosomes carry a small inverted duplication (if they carry two P elements) at the end of the chromosome. The process of mitotic pairing, which occurs normally in Drosophila, may cause the end of the chromosome to fold back on itself in order to associate with homologous sequences. If a broken chromosome end is recombinogenic, it may on occasion recombine with the sister chromatid to produce a new dicentric chromosome. These dicentric chromosomes may provoke a checkpoint response in early embryos.
If the events discussed in these three models occurred very early in development, when very few nuclei were present in the zygote, death of the whole animal could result. A detailed molecular study of the broken ends, along with characterization of the embryonic death, should allow us to determine if any of these models are correct. Such studies are underway.
Acknowledgments
We thank Tracy DeWolf and Stephanie Thatcher for critical comments on the manuscript, and Majid Golafshani for technical assistance. K.A. was a Howard Hughes Medical Institute Predoctoral Fellow and was also supported by training grant 5 T32 GM-07464 from the National Institutes of Health. This work was supported by grant HD-28694 from the National Institutes of Health.
Footnotes
-
Communicating editor: R. S. Hawley
- Received December 22, 1996.
- Accepted October 24, 1997.
- Copyright © 1998 by the Genetics Society of America
