Abstract
In most eukaryotes, segregation of homologous chromosomes during meiosis is dependent on crossovers that occur while the homologs are intimately paired during early prophase. Crossovers generate homolog connectors known as chiasmata that are stabilized by cohesion between sister-chromatid arms. In Drosophila males, homologs pair and segregate without recombining or forming chiasmata. Stable pairing of homologs is dependent on two proteins, SNM and MNM, that associate with chromosomes throughout meiosis I until their removal at anaphase I. SNM and MNM localize to the rDNA region of the X–Y pair, which contains 240-bp repeats that have previously been shown to function as cis-acting chromosome pairing/segregation sites. Here we show that heterochromatic mini-X chromosomes lacking native rDNA but carrying transgenic 240-bp repeat arrays segregate preferentially from full-length sex chromosomes and from each other. Mini-X pairs do not form autonomous bivalents but do associate at high frequency with the X–Y bivalent to form trivalents and quadrivalents. Both disjunction of mini-X pairs and multivalent formation are dependent on the presence of SNM and MNM. These results imply that 240-bp repeats function to mediate association of sex chromosomes with SNM and MNM.
PAIRING of homologous chromosomes is an essential step in meiosis, setting the stage for the segregation of homologs to opposite poles (Page and Hawley 2003; McKee 2004). In most eukaryotes, pairing is accompanied by high frequencies of recombination and by formation of synaptonemal complexes, proteinaceous structures that connect homologs from end to end during the pachytene stage of meiotic prophase. Segregation of homologs requires not only that they pair but also that they undergo at least one crossover. Crossovers generate cytologically visible linkers known as chiasmata, which are essential to maintaining bivalent integrity throughout late prophase I and metaphase I (Hawley 1988).
However, synaptonemal complexes, crossing over, and chiasmata are not universal prerequisites for meiotic chromosome segregation. Achiasmate segregation, i.e., segregation of homologs without chiasmata, is found in one sex or the other in several groups of eukaryotes (Wolf 1994). In Drosophila male meiosis, crossing over is normally completely absent and the homologs fail to form either synaptonemal complexes or chiasmata, yet homologs are stably conjoined during late prophase I and metaphase I and segregate from one another with great regularity during anaphase I.
Recent data show that stable conjunction and accurate segregation of all four homolog pairs in Drosophila male meiosis depends upon two proteins: stromalin in meiosis (SNM), a homolog of the SCC3/SA cohesin proteins, and Mod(mdg4) in meiosis (MNM), a BTB domain protein (Thomas et al. 2005). SNM and MNM colocalize to meiotic chromosomes throughout prophase I and metaphase I but disappear at the onset of anaphase I, suggesting that they function directly in stabilization of pairing, i.e., as substitutes for chiasmata.
Segregation of the X and Y chromosomes in Drosophila also depends upon pairing between specific sites located in the central region of the proximal X heterochromatin and on the short arm of the Y near the centromere. These sites coincide approximately with the locations of the X and Y rDNA arrays, which consist of 200–250 tandem copies of the genes for the 18S, 28S, and 5.8S ribosomal RNAs (rRNAs) (Ritossa 1976). Several findings have implicated the rDNA in X–Y pairing. X heterochromatic deletions that remove all of the rDNA cause X–Y pairing failure and X–Y nondisjunction (NDJ) during meiosis I (McKee and Lindsley 1987) but deletions that retain as few as six to eight rDNA repeats also retain pairing and disjunctional ability (Appels and Hilliker 1982; McKee and Lindsley 1987). Similarly, mini-X chromosomes generated by deletion of most of the euchromatin and variable amounts of the heterochromatin have been shown to segregate regularly from attached-XY chromosomes if they retain part of the rDNA locus but to disjoin randomly from attached-XYs if they are completely deficient for the rDNA (Lindsley and Sandler 1958; Park and Yamamoto 1995; Karpen et al. 1996).
The most compelling evidence for a specific role of the rDNA in X–Y segregation is that insertions of transgenes carrying single rRNA genes on X chromosomes deficient for the native pairing region partially restore X–Y pairing and disjunction (McKee and Karpen 1990). The first transgene to exhibit this property consisted of a single pre-rRNA transcription unit (containing the sequences for the 18S, 5.8S, and 28S rRNAs) flanked on both sides by partial copies of the intergenic spacer (IGS) region. In the native rDNA loci of the X and Y, IGS regions are located between the pre-rRNA transcription units and are usually 3–4 kb in length (Hayward and Glover 1989). They consist of several short repeated sequences, including 5–12 copies of a 240-bp repeat located immediately upstream of the pre-rRNA transcription unit. Interestingly, transgenes carrying fragments of rDNA genes that included at least six tandem 240-bp repeats proved to be competent in stimulating X–Y bivalent formation and disjunction even if most or all of the pre-rRNA transcription unit was removed. Conversely, transgenes carrying fragments encompassing most of the transcription unit proved unable to stimulate X–Y disjunction if they lacked a 240-bp repeat array (McKee 1996). These findings indicate that the 240-bp repeats, which are present in 1000–2000 copies on the X and Y, constitute the primary site for X–Y pairing and segregation.
The localization patterns of SNM and MNM indicate that they associate with the X–Y pairing site. During prophase I, MNM and SNM localize to multiple foci in the nucleolus where the 240-bp repeats also reside. Following chromosome condensation and nucleolar dissolution at prometaphase I, MNM and SNM colocalize to a dense focus associated with the X–Y bivalent and coincident with the 240-bp repeats (Thomas et al. 2005). However, the mechanism by which SNM and MNM associate with the X–Y pairing region is not understood.
The molecular function of the 240-bp repeat arrays in X–Y pairing and disjunction is also not understood. The colocalization data suggest that they might function as recruitment sites for SNM and MNM. A plausible alternative is that the 240-bp repeats mediate X–Y pairing during early prophase I, the period when homologous sequences are intimately paired, independently of SNM and MNM, and that SNM and MNM are recruited to already paired X and Y chromosomes to stabilize their association during the later stages of meiosis I. This scenario is consistent with our finding that SNM and MNM are dispensable for pairing of homologous autosomal sequences during early prophase I (Thomas et al. 2005) and with previous suggestions that cytologically visible X–Y connections (sometimes referred to as “collochores”) during prometaphase I and metaphase I are near but distinct from the rDNA arrays (Cooper 1964).
Although X–Y disjunction involving heterochromatically deficient X chromosomes has been a useful assay for identification of pairing sites, it has a significant limitation, namely that the Y chromosome pairing region cannot be manipulated and therefore remains molecularly uncharacterized. Since transgenes can be tested only for ability to pair with the native Y chromosomal rDNA, it has not been possible thus far to determine whether defined pairing sites carried on transgenes can pair and segregate from each other. Thus it is not known whether 240-bp repeats are fully autonomous with respect to pairing or whether additional components on the Y are required in addition to the 240-bp repeats. It has also not been possible to determine what role, if any, the 240-bp repeats play in recruitment of SNM and MNM to the X–Y pairing region.
For these purposes, mini-X chromosomes provide a more flexible assay system. Many mini-X chromosomes (also known as free X duplications, or Dps) carry so few genes that two or more copies can be added to a normal genome without compromising viability or fertility. Moreover, when two Dps that lack rDNA are added to a normal male genome, they segregate randomly from each other (Karpen et al. 1996). A major goal of this study was to determine whether pairing-deficient Dps carrying insertions of transgenic pairing sites such as rRNA genes or 240-bp repeat arrays are able to pair with and disjoin from one another. We show that Dps with the 240-bp repeats do disjoin nonrandomly from each other but do not form bivalents during prometaphase I or metaphase I. Instead, they are recruited into the X–Y bivalent, forming trivalents and quadrivalents. Moreover, both multivalent formation and nonrandom disjunction are dependent on the presence of the SNM and MNM proteins. Taken together, our findings indicate that the pairing activity of 240-bp repeat arrays requires direct association with SNM and MNM and suggest that these repeats serve as SNM/MNM binding sites on the X and Y chromosomes.
MATERIALS AND METHODS
Generation of clones and transgenes:
[rib7] contains one complete rDNA transcription unit flanked by IGS regions containing 11 total 240-bp repeats, 8 repeats on the 5′ side and 3 repeats on the 3′ side (Karpen et al. 1988; Figure 1, A and B). [rib7] has previously been inserted into an X chromosome deficient for the X–Y pairing site and was found to restore its competence to pair with and disjoin from a Y chromosome (McKee and Karpen 1990). The 240+x8 array was prepared from a single, subcloned 240-bp repeat derived from the X chromosomal rDNA array [subclone pBS(240+D/S), generated by DdeI digestion from p(rib,ry)7 (Karpen et al. 1988)] by self-ligation using T4 DNA ligase. The DdeI ends are asymmetric so that only head-to-tail ligations are allowed. An 8-mer fragment was identified, subcloned in pBluescript, sequenced to verify that no mutations had occurred, and cloned into the P-element transformation vector Carnegie-20 (Car-20) (Rubin and Spradling 1983) between the XhoI and SalI sites, yielding P[ry+, rib240+x8] = [240+x8] (Figure 1B).
Transgenic [240+x8] insertions were obtained by standard micro-injection and remobilization protocols using the stably inserted P transposase source Δ2-3(99B) (Robertson et al. 1988). X chromosome insertion sites were identified by in situ hybridization (McKee and Karpen 1990). Insertions on Dp1187 were generated by remobilization of X chromosome insertions using Δ2,3(99B) and were selected in two-step screens that made use of the ry+ marker in the P element and the y+ marker on Dp1187 (Figure 1C). The remobilization protocol generated two mini-X chromosomes carrying insertions of [rib7] (Dp[rib7]11 (Dp11) and Dp[rib7]12 (Dp12)), and two carrying insertions of [240+x8] (Dp[240+x8]61 (Dp61) and Dp[240+x8]62 (Dp62)) (Figure 1D). All mini-X insertions were characterized by Southern analysis as described in McKee and Karpen (1990) to ascertain copy number and determine whether internal rearrangements had occurred. These analyses indicated that Dp11 and Dp12 carry single intact [rib7] insertions, Dp61 carries a single, intact [240+x8] insertion, and Dp62 carries two complete, unrearranged insertions of [240+x8] (data not shown). Thus, Dp11, Dp12, Dp61, and Dp62 carry 11, 11, 8, and 16 wild-type 240-bp repeats, respectively.
Drosophila culture and cross procedures:
Flies were propagated on standard cornmeal–molasses agar medium at ∼23°. Crosses were carried out in plastic shell vials between single males and two to three females. Parents were discarded or transferred to fresh vials between day 7–10 and progeny were counted beginning on day 13 or 14 until day 22 when the vials were discarded. Mini-X chromosomes were maintained in stock over X chromosomes marked with y and were selected periodically for y+ flies. Markers and balancer chromosomes are described in Lindsley and Zimm (1992).
Measuring Dp transmission:
Germline transmission (T) for a Dp is defined as the probability that a Dp is successfully propagated through all of the germline divisions. In flies carrying a single Dp, perfect transmission (T = 1.0) would lead to inheritance of the Dp in exactly one-half of the offspring. Thus, T can range from 0 to 1, and the experimental measure of T is twice the frequency of Dp-bearing (y+) offspring from 1Dp males:
Determining Dp copy number:
Since flies with two or more copies of Dp1187 and its derivatives are viable and fertile, Dp stocks always contain flies with multiple copies and the copy number distribution is subject to random drift. Although copy number cannot be reliably determined by morphological or cytological criteria, it can be determined by outcrossing parental generation (P) males singly to y/y females and scoring the fraction of y+ progeny in the F1, which averages 44–49% for 1Dp males and 69–74% (or higher if significant Dp-Dp disjunction occurs) for 2Dp males, with little overlap for crosses with 50 or more progeny. To test the reliability of such classifications, we progeny tested >100 P males from the Dp1187, Dp12, and Dp62 stocks by crossing at least 15 F1 y+ males from each to y w females and scoring the y+ fraction. In a small minority of cases, a second generation of progeny testing was required to establish the genotype of the P male. From these tests, we established that males yielding y+ fractions of < ( T + 0.08) could reliably be classified as 1Dp males and y+ fractions of
T ± 0.06 could reliably be classified as 2Dp males, whereas other y+ fractions must be progeny tested to discriminate 1Dp from 2Dp or 2Dp from 3+Dp males. For the cross experiments, data from 3+Dp males were discarded. For the cytological analyses, a copy number census was carried out on the tested stocks by the above procedure near the time the males for cytological analysis were sampled.
X^Y-Dp disjunction assay:
Males heterozygous for the attached-XY chromosome, YSX.YL, In(1)EN, y B (symbolized X^Y, y B), and a Dp were generated by crossing X^Y, y B/0 males to y/y/Dp, y+ females, backcrossing their y+ B daughters (X^Y, y B/y/Dp, y+) to X^Y, y B/0 males, and then crossing the resulting y+ B sons (X^Y, y B/Dp, y+) to attached-X [C(1)RM, y2/0] females to generate stocks. Disjunction of the Dp from the X^Y was assayed by crossing X^Y, y B/Dp, y+ sons singly to two to three C(1)RM, y2/0 females. NDJ yields y2 B+ daughters [C(1)RM/O] and y+ B sons (X^Y/Dp). NDJX^Y-Dp = (y+ B males + y2 B+ females − (1 − T)N)/NT. This formula corrects for loss of the Dp. Results for each experiment were tested against the null hypothesis that NDJ = 0.5, which, from the above formula, predicts equality of the NDJ (y+ B males + y2 B+ females) and regular (y B males + y+ B+ females) classes by χ2.
Y-Dp disjunction assay:
Males of the genotype Df(1)X-1, y bb−/BSY/Dp, y+ were generated from crosses of Df(1)X-1, y bb−/FM6, y2 wa B females by y/BSY/Dp, y+ males (obtained from a cross of y/y/Dp, y+ females by +/BSY males) and mated singly with y w females to determine the distribution of the BS and y+ markers among the progeny. NDJY-Dp = (y+ BS + y B+ − (1 − T)N)/NT (males and females are combined in this formula).
Dp-Dp disjunction assay:
y w/Y or y/Y males carrying Dp1187, y+ or its derivatives were crossed singly to y w females in shell vials at 23°. Progeny were scored for sex and body color. The fraction of y+ progeny in each vial was used to classify the father as to copy number of Dp. For all vials in which classification was not definitive from the F1, F1 males were progeny tested, as detailed above. The frequency of y+ progeny from the 1Dp males was used to estimate the transmission fraction (T), as detailed above, and the frequency of y+ progeny from the 2Dp males
was used to calculate NDJDp-Dp.
NDJDp-Dp is defined as the fraction of meioses in X/Y/Dp/Dp (2Dp) spermatocytes in which the two Dps segregate to the same pole, giving rise to 2Dp (y+) and 0Dp (y) progeny. Regular disjunction from such spermatocytes always yields 1Dp y+ progeny. If Dp transmission were perfect, NDJ could be obtained by doubling the fraction of y progeny. However, estimation of NDJ is complicated by the fact that T is <1.0 so that some spermatocytes in a 2Dp male may have one or zero Dps and therefore yield y progeny without nondisjunction. Assuming that all loss is premeiotic and that the probability of Dp loss is independent of Dp copy number, loss will yield 1Dp spermatocytes at a frequency of 2T(1 − T) and 0Dp spermatocytes at a frequency of (1 − T)2. 0Dp spermatocytes yield only y progeny. 1Dp spermatocytes yield y and
y+ progeny. 2Dp spermatocytes will be present at a frequency of T2 and will yield y+ progeny at a frequency of (1 − NDJ) +
(NDJ) (= 1 −
NDJ) and y progeny at a frequency of
(NDJ). Thus the observed frequency of y+ progeny from a 2Dp male is given by: y+2/N2 = (T2(1 −
NDJ) + (T(1 − T))) = T −
T2 NDJ. Substituting T from above and rearranging gives:
.
The results of each Dp-Dp NDJ experiment were used to test the null hypothesis that the 2Dps assort independently by assigning a value of 0.5 to NDJ. The formulas for the expected frequencies under the null hypothesis are . The observed numbers of y and y+ progeny were then tested against the expected numbers by a χ2 test with 1 d.f.
Crosses involving snm and mnm:
For the cytological analysis, y/y/Dp, y+ females were crossed to +/Y; snmZ3-0317, st/TM3, Sb e males. The y+ Sb+ sons (y/Y/Dp, y+; snmZ3-0317, st/+) from 1Dp fathers were then crossed to y/y; snmZ3-2138/TM6, Tb e females and testes were dissected from the resulting y+ st Tb+ (y/Y/Dp, y+; snmZ3-0317, st/snmZ3-2138, st) and y+ st+ Tb+ (y/Y/Dp, y+; snmZ3-2138, st/+) sons. Similar crosses were used to generate the y/Y/Dp; mnmZ3-5578/Df(3R)T16 and sibling control males. For the NDJ measurements, y+ females from the Dp1187 and Dp62 stocks were crossed to y w/Y; Df(3R)T16, e/TM3, Sb e males and the y/Y/Dp; +/TM3, Sb e sons from 1Dp mothers crossed to y/y;snmZ3-2138, st/TM6C, Tb e males. From this cross, y/y/Dp; TM3, Sb e/snmZ3-2138, st females and y/Y/Dp; snmZ3-2138, st/+ males were selected and crossed to each other to yield 1Dp and 2Dp y/Y/Dp; snmZ3-2138, st/+ males and y/Y/Dp; snmZ3-2138, st/snmZ3-2138, st males using st vs. st+ to discriminate the snm/snm from the snm/+ siblings. The males were then crossed singly to two y w females.
Cytological methods:
The anti-ModC antibody and immunolocalization protocols are described in Thomas et al. (2005). The anti-Mod(C) antibody was diluted to 1:4000 and detected with Alexa Fluor 647 goat anti-rabbit IgG (H + L). FISH was carried out by the procedure of Balicky et al. (2002) with a probe concentration of 10 ng/μl. The 359-bp repeat (359 Rpt) FISH probe was generated by genomic PCR and PCR labeled using Chromatide Alexa Fluor 546-14-UTP (Molecular Probes, Eugene, OR) (Dernburg 2000). Primer sequences are available upon request.
RESULTS
Generation of mini-X chromosomes with and without transgenic rDNA repeats:
The organization of the X and Y heterochromatin and the structure of the rDNA loci are illustrated in Figure 1A. The mini-X chromosome Dp(1;f)1187 (referred to henceforth as Dp1187) comprises ∼1 Mb of centric heterochromatin and 300 kb of euchromatin and telomeric sequences from the distal tip of the X (Murphy and Karpen 1995). Dp1187 is devoid of rDNA but retains a substantial block of the 359-bp repeats that occupy most of the region between the rDNA and the centromere of the X (Ashburner et al. 2005). Although this mini-X chromosome carries a fully functional centromere and is transmitted efficiently through the germline mitotic and meiotic divisions, it segregates randomly from an attached-XY chromosome or from a second copy of itself in male meiosis (Park and Yamamoto 1995; Karpen et al. 1996).
Organization of sex chromosome pairing region and composition of transgenes and chromosomes used in the experiments. (A) Organization of X and Y heterochromatin (rectangular regions of X and Y diagrams). Circles, centromeres. Shaded rectangles, 359-bp repeats (1.688 g/cm3 satellite). Crosshatched rectangles, rDNA. Arrows indicate pre-rRNA transcription units; arrowheads indicate 240-bp repeats. (B) Composition of [rib7] (Karpen et al. 1988) and [240+x8] transformation vectors. [240+x8] was generated by self-ligation of a single subcloned 240-bp repeat from [rib7]. (C) Two-step scheme for remobilization of [rib7] and [240+x8] (P[rib-i, ry+]) from X to Dp1187. [Δ2,3] is a stably inserted P transposase source on a third chromosome carrying the dominant marker Sb (Robertson et al. 1988). Selection for y+ ry+ males in the first generation ensures that the transposon has hopped off the X (which is transmitted only to daughters) and that Dp1187 is recovered. Selection for cosegregation of y+ and ry+ in the second generation selects for hops to Dp1187. (D) Mini-X chromosomes carrying rDNA insertions generated by remobilization and used in this study. Diagonally cross-hatched triangles represent [rib7] trangenes; horizontally cross-hatched triangles represent [240+x8] transgenes.
Mini-X chromosomes carrying single complete rRNA genes ([rib7]) or arrays of eight wild-type 240-bp repeats [240+x8] were generated by P-element transformation and remobilization methods (Figure 1, C and D) and compared with Dp1187 for disjunctional ability. Transmission of the Dps through the male germline was measured in the progeny of crosses between X/Y/Dp, y+ males (1Dp males) and females carrying structurally normal X chromosomes marked with y or y w using the y+ marker on the Dp to monitor its presence. If a mini-X is transmitted with 100% efficiency (T = 1.0), it should be inherited by exactly one-half the progeny, yielding a 50:50 ratio of y+:y progeny. Consistent with previous results (Karpen et al. 1996), transmission of Dp1187 in our crosses was slightly less than maximal, averaging 0.954 in two replicate crosses. This slight deficit in transmission likely reflects the fact that Dp1187 provides no essential functions so that occasional spontaneous mitotic or meiotic errors leading to its loss are without phenotypic consequence. The transgenic rDNA insertions had no significant effect on mini-X transmission. Mean T values for Dp1187 with and without transgenic rDNA insertions were 0.951 and 0.954, respectively (Table 1). Thus rDNA insertions do not disrupt germline transmission of mini-X chromosomes.
Transmission of mini-X chromosomes in the male germline
Effects of rDNA insertions on segregation of mini-X chromosomes from full-length sex chromosomes:
Each of the mini-X chromosomes was tested for ability to segregate from an attached-XY chromosome, YSX.YL, In(1)EN, y B (X^Y), by crossing X^Y, y B/Dp, y+ males to C(1)RM, y2/O (attached-X/O) females (Table 2). X^Y-Dp NDJ yields y2 daughters [C(1)RM/O] and y+ males. In this and the other crosses reported below, the Dp transmission frequencies for the same Dp from Table 1 were used to correct the measured NDJ frequencies for Dp loss. This is important because, although Dp loss frequencies are modest, the effects of transgenic pairing sites on NDJ frequencies are sometimes modest as well. The results from two replicates of each cross showed that Dp1187 disjoined randomly from the X^Y, as expected, while all four rDNA-bearing Dps yielded NDJ values significantly <50%. A substantially lower NDJ value was observed in the Dp62 crosses than in the other crosses, suggesting that two arrays of 240-bp repeats might enhance disjunction more strongly than one, consistent with previous results for X–Y disjunction (McKee and Karpen 1990).
Disjunction of mini-X chromosomes from X^Y
To determine whether disjunction of the transgene-bearing Dps depends upon the same region of the X chromosome that X–Y disjunction depends upon, Y-Dp disjunction was measured in crosses of Df(1)X-1, y/BSY/Dp, y+ males to y w females. Df(1)X-1 is an X chromosome that carries a large deletion of proximal heterochromatin that includes all of the rDNA (Lindsley and Zimm 1992). BSY is an otherwise normal (rDNA+) Y chromosome genetically marked by a transposition of the X chromosome-derived BS allele. These two chromosomes disjoin randomly from one another because the pairing region of the X is completely deleted (McKee and Lindsley 1987), so that the progeny include XX females, XXY females, XY males, and XO males. The results (Table 3) show that transgene-bearing Dps (Dp11 and Dp62) were recovered at frequencies >50% in progeny classes that do not inherit the Y chromosome (XX females and XO males) but at frequencies <50% in the Y-bearing progeny classes, implying that these Dps segregate preferentially from the Y chromosome during male meiosis. However, Dp1187 recovery was similar among all four progeny classes, indicating that it disjoins randomly from the Y. It is instructive to compare the results in Table 3 with those in Table 1 in which both the X and Y chromosomes carried native rDNA loci. Although the Y was not marked in those crosses, X–Y disjunction is regular in these males so that the sex of the progeny is a reliable indicator of presence or absence of the Y (data not shown). The results in Table 1 exhibited no sex bias in recovery of any of the Dps, indicating that the Dps do not disjoin preferentially from either the X or the Y chromosome, irrespective of whether the Dp does or does not carry transgenic pairing sites. We conclude that mini-X chromosomes bearing insertions of either complete rDNA repeats or 240-bp repeat arrays disjoin from X and Y chromosomes in a manner dependent upon the presence of the same pairing region that mediates X–Y disjunction and that the X and Y pairing regions are equally effective in directing the orientation of such mini-X chromosomes during meiosis I.
Disjunction of mini-X chromosomes from BSY
Since rDNA insertions enable mini-X chromosomes to segregate from full-length sex chromosomes, they might also enable such mini-X chromosomes to compete with normal X and Y pairing sites in X/Y/Dp males and thereby disrupt segregation of X–Y pairs. To assess this possibility, X/BSY/Dp males carrying single copies of the mini-X chromosomes Dp1187, Dp11 or Dp62 were crossed with wild-type females, and the X-Y NDJ frequency was determined by measuring the fraction of BS (XXY) daughters and B+ (XO) sons. No X–Y NDJ was observed in males carrying either Dp1187 or Dp11, and, of 3010 total progeny, 12 NDJ progeny (0.4%) were seen in the Dp62 cross (data not shown). These data suggest that the Dp62 chromosome may act as a very weak competitor of pairing and disjunction for a normal X–Y pair, but that Dp1187 and Dp11 do not.
Can transgenic pairing sites mediate disjunction of two mini-X chromosomes?
To determine whether two mini-X chromosomes can disjoin from one another, males carrying a structurally normal X chromosome marked with yellow (y), an unmarked Y chromosome, and two copies of each Dp (2Dp males) were crossed to y w females and the fractions of y+ and y progeny measured and used to estimate the Dp-Dp NDJ frequency. Dp transmission frequencies were measured in parallel crosses of sibling 1Dp males and used to correct for Dp loss (see materials and methods for details of the assay and calculations). Each set of 1Dp and 2Dp crosses was carried out twice, using different X chromosomes.
The results of the crosses are presented in Table 4. As expected, two copies of Dp1187 disjoined at random from one another (mean NDJ over two replicates = 50.7%). Random disjunction is consistent both with previous results (Karpen et al. 1996) and with cytological evidence presented below that Dp1187 behaves consistently as a univalent. The presence of transgenic rDNA on the mini-X chromosomes led to significantly reduced NDJ frequencies. For Dp11, Dp12, and Dp61, which each carry one inserted rDNA unit or 240-bp array, the NDJ values fell in the 41–44% range. All of these values differed from random disjunction at P < 0.05 and all but one differed at P < 0.01. As in the X^Y/Dp assay, a more dramatic reduction (mean NDJ = 35.8%) was seen with Dp62, which carries two 240-bp repeat arrays. These results show that inserted 240-bp repeat arrays can promote disjunction of a pair of mini-X chromosomes that lack native rDNA and that disjunction is proportional to the number of such arrays or to the total 240-bp repeat copy number.
Dp-Dp nondisjunction in X/Y/Dp/Dp males
Dp-Dp segregation in snm mutant males:
Since X–Y disjunction is severely disrupted in snm and mnm mutants (Thomas et al. 2005), it seemed likely that disjunction of rDNA-bearing mini-X chromosomes would be similarly affected. To test this, X/Y/Dp62/Dp62 males that were either homozygous for the snm null allele snmZ3-2138 or heterozygous for snmZ3-2138 and snm+ were crossed to y w females and the progeny were analyzed as above (Table 4). Control crosses involving Dp1187 were carried out in parallel. As expected since Dp1187 disjoins randomly in wild type, the snm genotype had no effect on disjunction of Dp1187; in both snm/+ and snm/snm 2Dp males, the two copies of Dp1187 disjoined randomly from one another. The results for the X/Y/Dp62/Dp62; snm/+ controls were also similar to those for wild-type males: the two Dps segregated to opposite poles nearly twice as often as to the same pole (NDJ = 34.6%). However, preferential Dp-Dp disjunction was abolished by loss of snm function. The frequency of Dp62-Dp62 NDJ in the X/Y/Dp/Dp; snm males was 48.3%, not significantly different from 50%. The difference between the NDJ frequencies in Dp62; snm/snm and Dp62; snm/+ males was found, by a contingeny test, to be significant at P < 0.01.
In addition to randomizing Dp62-Dp62 disjunction, loss of snm function also reduced transmission of Dp62 and Dp1187 to 79.6 and 77.9%, respectively, compared to 93.3 and 94.1%, respectively, in the snm/+ controls. The basis for this effect is not known but may be related to the chaotic anaphase I segregation pattern in snm males, which gives rise to extra mini-nuclei at telophase I that appear to carry single chromosomes and may be lost during spermatid maturation (Thomas et al. 2005 and S. Thomas and B. McKee, unpublished data). Since the effect of the snm mutations on Dp transmission is similar for the two Dps, the transmission effect cannot explain the differential effect of the snm mutations on Dp-Dp nondisjunction in the Dp62 vs. Dp1187 crosses. Thus we conclude that disjunction of mini-X pairs carrying rDNA transgenes is dependent on snm function.
Do mini-X chromosomes bearing transgenic pairing sites form bivalents?
A possible explanation for segregation of mini-X pairs that carry transgenic rDNA is that they form stable bivalents that segregate independently of the X–Y bivalent. Since Dp1187 can be visualized cytologically, it should be possible to detect Dp-Dp bivalents cytologically and to test the prediction that their formation is dependent upon transgenic pairing sites. Moreover, since snm+ is required for their segregation, SNM and MNM should associate with Dp-Dp bivalents.
To ascertain whether transgenic pairing sites lead to formation of mini-X bivalents, we examined spermatocytes from males carrying Dp1187 or Dp62 in addition to structurally normal X and Y chromosomes. The X chromosome in these stocks carried a recessive yellow (y) mutation that results in yellow pigmentation of wings, bristles, and the integument, so Dp-bearing males could be identified on the basis of their y+ (brown–black) pigmentation. Although it is not possible to determine how many Dps an individual carries on the basis of pigmentation, the frequencies of y+ males with one, two, or three Dps within each stock can be reliably determined by a cross procedure detailed in materials and methods. Copy number tests were carried out on samples of ∼200 y+ males each from the Dp1187 and Dp62 stocks (Table 5, line 1). These tests showed that the fraction of y+ males that carried two or more Dps was 31% in the Dp1187 stock and 38% in the Dp62 stock.
Results of FISH analysis of Dp1187 and Dp62 spermatocytes
The spermatocytes were hybridized with a fluorescently labeled DNA probe complementary to the 359-bp repeats and a centromere–proximal, X chromosome-specific satellite (Figure 1A; Ashburner et al. 2005) and counterstained with DAPI. The 359-bp FISH probe was chosen because although most of the 359 repeats were deleted during the synthesis of Dp1187, a residual block of repeats near the centromere generates a FISH signal strong enough to detect (Murphy and Karpen 1995). As expected, in some nuclei of both genotypes, one or more DAPI-stained masses considerably smaller than the fourth chromosomes were visible. These small DAPI masses were associated with a FISH signal much weaker than the signal from the full-length X chromosome, thus establishing their identity as mini-X chromosomes (Figure 2A).
Analysis of mini-X pairing by FISH. (A) Examples of spermatocytes from Dp62-bearing males hybridized with a 359 Rpt FISH probe with zero, one, or two Dps (arrowheads) visible. DNA stained with DAPI. See Table 5 for frequencies. Bar, 5 μm. (B) Dp62 prometaphase I spermatocyte showing two 359 FISH signals [X and Dp62 (arrow)] within the X–Y bivalent. Bar, 5 μm.
Using the FISH method, we found no evidence for independent mini-X bivalents in prometaphase I and metaphase I spermatocytes in either X/Y/Dp1187/Dp1187 or X/Y/Dp62/Dp62 males. Spermatocytes that exhibited two copies of either Dp1187 or Dp62 that were separate from the X–Y bivalent were identified in both stocks, but considerably less frequently in Dp62 males (8/105) than in Dp1187 males (13/69) for reasons explored below. In all such spermatocytes from both stocks, the two Dps appeared unpaired (Figure 2A). Although closely adjacent Dp pairs were occasionally observed in both genotypes (e.g., two nearby copies of Dp1187 in the third column in Figure 2A), in all except one of the Dp62 spermatocytes, the Dps were separated by at least 2 μm and were often on opposite sides of the nucleus. Of the eight Dp62 spermatocytes with two separate Dps, six were in metaphase I with the four regular bivalents aligned and clustered closely together. In all six, one or both of the Dps was well separated from the metaphase I chromosome mass. In no case were the two Dps aligned along the spindle axis, making it unlikely that mini-X's undergo long-distance pairing. We considered the possibility that Dp62-Dp62 bivalents, when present, are so tightly associated that they appear to be single chromosomes. If this were the case, some of the small DAPI foci and FISH signals in Dp62 spermatocytes should have been significantly larger and brighter than other such signals and larger and brighter than the signals in nuclei with two visible Dps, whereas the small DAPI foci and FISH signals in Dp1187 spermatocytes should have been more uniform in size and brightness. In addition, paired Dps would be expected to yield bi-lobed signals when viewed laterally. However, we observed no difference between Dp62 and Dp1187 1Dp spermatocytes or between Dp62 1Dp and 2Dp spermatocytes with respect to variation in size or brightness among foci and observed no bi-lobed signals in either genotype.
Staining with anti-SNM or anti-MNM antibodies enabled us to ascertain whether Dp62, when isolated from the X–Y bivalent, exhibits a mini-focus of SNM and MNM. No such mini-foci were observed, although 100% of prometaphase I and metaphase I nuclei exhibited the prominent SNM/MNM focus associated with the X–Y bivalent (data not shown). It is not clear whether this result indicates that the SNM and MNM proteins are truly absent on univalent Dps or that the amounts present fall below the detection threshold of the assay.
Do sex chromosome quadrivalents form in X/Y/Dp/Dp males?
Previous studies have demonstrated that, in spermatocytes of males carrying three sex chromosomes with native rDNA loci (e.g., X/Y/Y), the sex chromosomes regularly form trivalents in which all three chromosomes are associated at their pairing regions (Cooper 1964). Thus, an alternative scenario for Dp-Dp segregation in X/Y/Dp/Dp males is formation of a sex chromosome quadrivalent within which the Dps preferentially orient to opposite poles, either because they pair directly with each other in the quadrivalent or because each Dp pairs with different large sex chromosomes (X or Y) more often than they pair with the same large sex chromosome. If so, the mini-X chromosomes should associate at elevated frequencies with the X–Y bivalent and with its prominent SNM/MNM focus.
To assess the possibility that the presence of 240-bp repeat arrays on mini-X chromosomes induces formation of sex chromosome multivalents, we first examined spermatocytes prepared by the FISH method for evidence of multiple signals within the X–Y bivalent. We observed that, in Dp62 but not in Dp1187 spermatocytes, some nuclei (3 of 105) exhibited two 359 Rpt signals within the X–Y bivalent: a bright signal associated with the X chromosome and a much weaker signal presumably emanating from one or both Dps (Figure 2B). Although the observed frequency of such nuclei in Dp62 spermatocytes was low, this could reflect a low probability of detecting the weak Dp FISH signal in close proximity to the much stronger X FISH signal.
The FISH data also provided indirect evidence for multivalent formation in Dp62 spermatocytes: a much lower frequency of observed Dps that were separate from the X–Y bivalent in Dp62 spermatocytes (47/100 spermatocytes) than in Dp1187 spermatocytes (97/100 spermatocytes, Table 5, line 5). Spermatocytes with no visible Dps amounted to 61% of the Dp62 nuclei but only 23% of the Dp1187 nuclei, whereas nuclei with two visible Dps comprised 17.4% of Dp1187 nuclei but only 7.6% of Dp62 nuclei (Table 5, line 5). These differences cannot be due to differing frequencies of males with two or more Dps in the two stocks since the frequency of such males was slightly higher in the Dp62 stock (38%) than in the Dp1187 stock (31%) (Table 5, line 1). They also cannot be explained by differential frequencies of premeiotic Dp loss, since, as shown in Table 1, Dp1187 and Dp62 exhibit nearly identical (and very modest) germline loss frequencies. Thus, the large discrepancy in frequency of observed Dps among spermatocytes from the two stocks strongly suggests that a much larger fraction of Dps are incorporated into the X–Y bivalent in Dp62 spermatocytes than in Dp1187 spermatocytes. In a subsequent section, we use the FISH data to derive an estimate of the frequency of multivalent formation in Dp62 spermatocytes. From this procedure, we estimate that quadrivalents are present in at least 30% of 2Dp spermatocytes.
Our immunocytological (IC) staining protocol generated stronger direct evidence for formation of multivalents in Dp62 males: a distinctive and reproducible differerence in the morphology of X–Y bivalents between Dp62 and Dp1187 spermatocytes (Figure 3). Whereas the X–Y bivalents in Dp1187 spermatocytes, as in spermatocytes lacking a Dp, were roughly spherical and uniform in DAPI intensity, many of the X–Y bivalents in Dp62 spermatocytes were distinctly asymmetric in both regards. One edge of the X–Y DAPI-stained region appeared unusually broad in Dp62 spermatocytes and contained a finger-shaped region of especially bright DAPI stain, suggesting the presence of additional condensed chromatin near the periphery of the bivalent. Significantly, the broadened edge was invariably the edge nearest to or touching the SNM/MNM focus. This observation suggests that Dp62 mini-X chromosomes may be recruited into the X–Y bivalent by virtue of their affinity for MNM/SNM and are thus positioned immediately adjacent to the SNM/MNM focus. Since this unusual, claw-like morphology was exhibited by the majority of X–Y bivalents (59%, n = 27) in Dp62 spermatocytes but by none of the X–Y bivalents in Dp1187 spermatocytes (n = 23), we in-fer that the presence of transgenic 240-bp repeat arrays on the mini-X chromosome leads to regular association between mini-X chromosomes and the X–Y bivalent.
Altered morphology of X–Y bivalent in Dp62 spermatocytes. Prometaphase I spermatocytes from Dp62-bearing males and wild-type males stained with an antibody against MNM. Note the broadened edge at the upper (MNM-stained) edge of the X–Y bivalent and localized finger of strong DAPI staining in the Dp62 spermatocytes compared to the more regular appearance of the X–Y bivalent in the wild-type spermatocytes. Bar, 5 μm.
Role of SNM and MNM in localization of rDNA-bearing Dps to the X–Y bivalent:
If Dp62 localizes to the X–Y bivalent because of interactions between the transgenic 240-bp repeats and the SNM and MNM proteins associated with the X–Y bivalent, the localization pattern of Dp62 should be dependent upon wild-type function of both the snm and mnm genes. To test this, Dp1187 and Dp62 were crossed into mnm and snm backgrounds and primary spermatocytes from mutants and their heterozygous siblings were scored for visible mini-X chromosomes (Figure 4A). As expected, in the wild-type controls, separate mini-X chromosomes were observed more frequently in Dp1187 (19/22) than in Dp62 males (14/29) (P < 0.01, χ2 contingency test, 1 d.f.) (Figure 4B). However, there was no significant difference in frequency of visible mini-X's between Dp1187 and Dp62 males that were mutant for either mnm or snm. These results imply that the preferential localization of Dp62 to the vicinity of the X–Y bivalent is dependent on SNM and MNM.
(A) Representative FISH results from y/Y/Dp62; mnmZ3-5578/Df(3R)T16 and y/Y/Dp62; snmZ3-0317/snmZ3-2138 males hybridized with a 359 Rpt FISH probe. Arrowheads indicate mini-X chromosomes. Bar, 5 μm. (B) Frequencies of prometaphase I and metaphase I spermatocytes with or without a visible mini-X chromosome in snm, mnm, and heterozygous snm/+ or mnm/+ control males carrying one copy of either Dp1187 or Dp62.
Estimating the frequency of multivalents in Dp62 spermatocytes:
Although multivalents cannot be directly quantified by the FISH method, an indirect estimate of their frequency can be obtained by comparing the observed frequency of Dp62 univalents (Table 5, lines 4 and 5) with the expected frequency in the absence of multivalent formation. The expected frequency was obtained from the observed distributions of males carrying different numbers of Dps (Table 5, line 1) adjusted for germline Dp loss (line 2), assuming that all Dp loss occurs prior to meiosis and that the loss frequency is independent of the number of Dps present; i.e., that it is the same in 1Dp, 2Dp, and 3Dp males. Although Dp loss may not be entirely premeiotic, the actual timing of Dp loss is irrelevant for our estimates of multivalent frequencies as long as it does not differ greatly in Dp1187 and Dp62 males. Since loss frequencies are virtually identical in Dp1187 and Dp62 males, it seems unlikely that loss results from different mechanisms in the two stocks. We evaluate the assumption of copy-number independence below.
The necessity for a further adjustment to account for the detection efficiency (DE) of the FISH procedure is apparent from a comparison of Table 5, lines 2–4 (expected Dps), with lines 5–7 (observed Dps). Germline loss can account for only ∼3.5% of nuclei with no visible Dps (line 2) but 23% of the Dp1187 nuclei fell into this category (line 5). DE was estimated from the ratio of observed Dps in Dp1187 nuclei (97 Dps/100 nuclei, line 7) to expected Dps after adjusting for premeiotic loss (125.5 Dps/100 nuclei, line 4), yielding a DE of 0.77. The submaximal detection efficiency might result from shielding of the weak DAPI and FISH signals from the Dps by chromatin of other bivalents or by other nuclear structures or from loss of some of the Dps during slide preparation. Although we cannot rule out the alternative that Dp1187 sometimes goes undetected because it is incorporated into the X–Y bivalent, we consider this unlikely because two copies of Dp1187 disjoin randomly from one another and because we failed to detect any claw-shaped X–Y bivalents among 23 Dp1187 spermatocytes prepared by the IC method. Whatever its origin, the imperfect efficiency of the assay does not pose a serious problem as there is no reason why Dp62, when it is separate from the X–Y bivalent, should be detected more or less efficiently than Dp1187.
Applying the estimated DE value to the Dp62 data yields a corrected univalent frequency in Dp62 males of 60.4/100 nuclei (Table 5, line 8). Since this value represents only 45% (60.4/133.1) of the Dps that should be present in Dp62 spermatocytes, it follows that 55% of the Dps must be involved in sex chromosome multivalents; thus the association frequency (AF) of Dp62 is 0.55. Assuming that Dps in 1Dp, 2Dp, and 3Dp nuclei are incorporated into multivalents at the same frequency, we estimate that trivalents and quadrivalents are present in 51 and 11%, respectively, of Dp62 spermatocyte nuclei (Table 5, lines 9 and 10). Considering only spermatocytes with two copies of Dp62, we estimate that quadrivalents form in ∼30% (0.55 × 0.55 × 100) of such nuclei. We note that the resulting estimate of the frequency of multivalents, 62% (51% + 11%), is in very good agreement with the observed frequency of 59% for claw-shaped sex chromosome configurations obtained in the IC experiments. The concordance of these independent estimates of multivalent frequencies in Dp62 males strongly supports our inference that the presence of 240-bp repeats on Dp62 promotes formation of sex chromosome quadrivalents.
In the above calculations, we assumed that Dps behave similarly irrespective of their copy number, i.e., that Dp loss, DE, and AF do not differ among 1Dp, 2Dp, and 3Dp males. To test these assumptions, we compared the observed Dp distribution frequencies with the expected distribution frequencies for both the Dp1187 and Dp62 data sets (Table 6). These comparisons are meaningful because the parameters used to generate the expected distributions were estimated either from independent data sets (Dp loss, Table 1) or from the mean numbers of Dps observed in the Dp1187 spermatocytes (DE) and Dp62 spermatocytes (AF) (Table 5) and thus are independent of the frequency distributions. If either Dp loss or DE is strongly affected by Dp copy number, the observed and expected Dp1187 distributions should differ. Similarly, if Dp62 is incorporated into multivalents at frequencies that depend strongly on copy number, the observed and expected Dp62 distributions should differ. However, the observed and expected frequency distributions did not differ significantly in either the Dp1187 or the Dp62 data sets (Table 6). These results validate our assumptions about Dp behavior and, taken together with the concordance of the estimates of multivalent frequencies from the FISH and IC data, provide support for our estimate of the frequency of quadrivalent formation in Dp62 spermatocytes.
Expected vs. observed distributions of Dp1187 and Dp62 spermatocytes
DISCUSSION
The Drosophila X–Y pairing site functions to recruit SNM and MNM:
A variety of specialized chromosomal sites that function to promote homolog segregation and/or recombination during meiosis I have been reported. Several of these have been found to affect segregation by acting as local enhancers of chiasmata formation, the products of homologous crossovers that act as stable connectors between homologs (Hawley 1980; McKim et al. 1993; Villeneuve 1994; MacQueen et al. 2002, 2005; (Phillips et al. 2005; Sherizen et al. 2005). In contrast, other cis-acting sites promote homolog segregation independently of recombination. In addition to the Drosophila X–Y pairing site that is the subject of this study, sites in centromeric or pericentromeric heterochromatin have been shown to pair autonomously in Drosophila, Saccharomyces cerevisiae, and Schizosaccharomyces pombe and in some cases to provide a back-up system for segregating nonexchange bivalents in the context of chiasmate meiosis (Zhang et al. 1990; Hawley et al. 1993; Dernburg et al. 1996; Karpen et al. 1996; Matthies et al. 1999; Davis and Smith 2003; Kramer and Hawley 2003; Ding et al. 2004; Harper et al. 2004; Kemp et al. 2004; Davis and Smith 2005; Gilliland et al. 2005; Tsubouchi and Roeder 2005).
Previous studies of the Drosophila X–Y pairing site have led to the molecular identification of both the cis- and trans-acting components, making it one of the best-characterized meiotic pairing sites. Deletion of the rDNA repeats on the X chromosome leads to X–Y pairing failure and random X–Y segregation at anaphase I. Transgenic insertions of complete rRNA genes or rDNA fragments that include at least six tandem copies of the 240-bp IGS repeats from the rDNA promote X–Y bivalent stability and suppress X–Y NDJ (McKee and Lindsley 1987; McKee and Karpen 1990; McKee 1996). We recently reported that stable homolog conjunction in Drosophila meiosis is mediated by two proteins, SNM and MNM, that colocalize to the nucleolus during meiotic prophase and to the X–Y bivalent following chromosome condensation. SNM and MNM colocalize with the 240-bp repeat FISH probe on condensed X–Y bivalents, suggesting that their association with the X–Y bivalent may be mediated by direct or indirect binding to the 240-bp repeats (Thomas et al. 2005).
However, previous experiments had not tested directly for a role of the 240-bp repeats in localization of SNM and MNM to the X and Y chromosomes. Here we have shown that recruitment of mini-X chromosomes into sex chromosome trivalents and quadrivalents is dependent on the presence of both the 240-bp repeats in cis and the SNM and MNM proteins in trans. Moreover, absence of SNM and MNM randomizes disjunction of mini-X chromosomes whether or not they carry 240-bp repeats. These observations strongly support the idea that the 240-bp repeats function to mediate association of sex chromosomes with SNM/MNM and suggest that they may serve directly as binding sites for those proteins.
Repeat arrays of 240 bp mediate formation of sex chromosome quadrivalents:
Our data show that insertion of rDNA-derived pairing site sequences into an rDNA-deficient mini-X enables the mini-X to disjoin at a significant frequency from other sex chromosomes and from a second copy of itself. Preferential disjunction of mini-X chromosomes could result from formation of mini-X bivalents. However, we were unable to detect by FISH or immunocytological methods evidence for formation of stable autonomous bivalents between two mini-X chromosomes that carry rDNA transgenes. Even when two transgene-bearing Dps were visible but separate from the X–Y bivalent, they did not appear to form bivalents or to align on the metaphase I spindle, suggesting that one or two rRNA genes or 240-bp repeat arrays may not suffice to mediate formation of autonomous bivalents between chromosomes otherwise devoid of pairing sites. Moreover, no SNM or MNM signals were detected in association with isolated copies of Dp62 either by immunocytology or by native fluorescence of MNM-GFP (data not show). Since only eight Dp62 spermatocytes with two separated Dps were observed, we cannot rule out a low frequency of bivalent formation, but if such bivalents do occasionally form, their contribution to the preferential disjunction of Dp62 in 2Dp nuclei must be, at best, minor.
Our major finding is that rDNA-containing mini-X chromosomes frequently colocalized with the normal X and Y chromosomes, forming trivalents and quadrivalents, whereas non-rDNA-containing mini-X chromosomes remained separate from the X–Y bivalent. The tendency of rDNA-bearing Dps to colocalize with the X–Y pair is consistent with previous genetic and cytological evidence for regular formation of trivalents in spermatocytes of males carrying two Y chromosomes (Cooper 1964). Since both multivalent formation and preferential disjunction of mini-X chromosomes to opposite poles are dependent on SNM and MNM, we infer that formation of X/Y/Dp/Dp quadrivalents is required for preferential Dp-Dp disjunction.
Three alternative explanations for the apparent inability of mini-X pairs with transgenic pairing sites to form stable bivalents are as follows: (1) mini-X chromosomes bearing transgenic pairing sites are inherently incapable of forming autonomous bivalents, perhaps because successful pairing requires at least one pairing partner to carry non-rDNA sequences in addition to arrays of 240-bp repeats; (2) transgenic pairing sites are quantitatively rather than qualitatively deficient—i.e., they are too small for two such sites to interact stably with one another; and (3) transgenic pairing sites are capable of pairing with each other but are outcompeted by the vastly larger pairing regions of the X and Y chromosomes. There is no compelling evidence for explanation 1 although previous investigators have presented suggestive evidence for “collochores” flanking the rDNA (Cooper 1964; Park and Yamamoto 1995). With respect to explanation 3, it is clear from comparing the random segregation pattern of Dp62 in X/Y/Dp males to the highly biased segregation pattern in Df(1)X-1/BSY/Dp62 males that the X and Y do compete for Dp pairing sites and, moreover, that they compete on approximately even terms. However, in 8% of Dp62 spermatocytes, two Dps are present but neither Dp is visibly paired with each other or with the X–Y bivalent, suggesting that competition from the X and Y is not so overwhelming as to prevent any opportunity for Dp-Dp pairing and, moreover, that the affinity of the transgenic pairing sites for one another must be rather low. Unfortunately, it is not feasible at present to carry out a test of Dp-Dp pairing in the absence of the native rDNA loci of both X and Y, which would be the definitive test.
Quadrivalents and preferential Dp-Dp disjunction:
An important prediction of the hypothesis that Dp-Dp disjunction occurs mainly via quadrivalent formation is that quadrivalents should be present in X/Y/Dp62/Dp62 spermatocytes at a high-enough frequency to account for the reduced NDJ. The quantitative data from the FISH assay enable us to estimate that quadrivalents are present in ∼30% (0.547 × 0.547 × 100) of 2Dp nuclei in Dp62 males. Although the method that yielded this estimate is indirect, the fact that it also yielded an estimate of overall multivalent frequency very close to the direct estimate of multivalent frequency obtained from the IC analysis validates the general approach and suggests that the estimated 30% value is probably not far from the true quadrivalent frequency. Since we failed to detect any independent Dp-Dp bivalents in nuclei with two separated Dps and since there is no obvious mechanism by which formation of a trivalent plus univalent would lead to nonrandom segregation of 2Dps, we will assume that the reduced NDJ in Dp62 males results only from nonrandom segregation of Dps from quadrivalents. If Dps in a quadrivalent segregate from one another at a frequency of 100% and Dps in the remaining 70% of 2Dp nuclei segregate randomly, the expected NDJ frequency would be 35% ( × 70%). This value agrees remarkably well with the measured NDJ value of 0.359 obtained from two replicates of the Dp62 cross. This comparison is legitimate because the procedures for estimating quadrivalent frequency and Dp-Dp NDJ involved the same adjustment for Dp loss, so both were based only on 2Dp spermatocytes. Clearly, the predicted NDJ value would agree less well with the observed value if we assumed that Dp-Dp disjunction from quadrivalents occurs at any frequency significantly <100%. However, since we cannot rule out the possibility that Dp-Dp bivalents are present at a low frequency and account for some of the reduced NDJ, Dp segregation from quadrivalents might be somewhat <100%. Nevertheless, our data strongly suggest that Dp-Dp pairs that have been incorporated into X/Y/Dp/Dp quadrivalents disjoin to opposite poles the great majority of the time.
How might quadrivalent formation promote Dp-Dp disjunction?
The high frequency of Dp-Dp disjunction from quadrivalents implies either that the 2Dps nearly always pair with and disjoin from each other within quadrivalents or that they nearly always pair with and disjoin from different large sex chromosomes; i.e., if Dp1 pairs with the X, then Dp2 pairs with the Y. Since Dps do not pair with each other robustly when separate from the X–Y bivalent and since the X and Y pairing sites are so much larger and stronger than the transgenic Dp pairing sites, the latter alternative seems more likely. Figure 5 illustrates this “chain-quadrivalent” model.
Illustration of chain-quadrivalent model for Dp-Dp segregation. Dp1 and Dp2 pair with opposite large sex chromosomes and each orients to the pole opposite that of its pairing partner. rDNA loci and transgenes are cross-hatched.
Two nonexclusive mechanisms that could favor preferential pairing with opposite sex chromosomes are as follows. One possible mechanism might be that the mini-X chromosomes are too bulky to pair with the same sex chromosome. Each mini-X encompasses >1 Mb of DNA and it may therefore be more difficult for the pairing region of one large chromosome to accommodate both Dps than for each to associate with different large chromosomes. Since the X–Y pair is very compact during metaphase I and its pairing region is especially so, it seems plausible that two Dps could be more easily accommodated on opposite sides of the X–Y pairing region than on the same side.
A second possibility is a mechanism that serves to balance kinetochore orientation to equalize tension on the metaphase I spindle. With four functional kinetochores, a quadrivalent would achieve balance when two kinetochores orient to each pole. A simple kinetochore balance mechanism that took no account of relative pairing strengths would predict that the four chromosomes would orient 2 × 2 randomly. However, since our data show that the X and Y nearly always segregate away from each other in Dp-bearing males, the X and Y must orient to opposite poles irrespective of the orientation of the Dps. Since the bond between a Dp and a large sex chromosome is likely to be much weaker than that between the X and Y, it could be much more easily subject to correction by a mechanism that detects tension imbalance. Therefore, such a mechanism might promote Dp-Dp segregation by eliminating Dp-X or Dp-Y bonds in quadrivalents in which both Dps paired with the same sex chromosome and oriented to the same pole.
Other mechanisms by which quadrivalents promote Dp-Dp segregation are also possible. One is that recruitment of mini-X chromosomes into a quadrivalent could promote their direct pairing, even though separated Dps do not pair efficiently with one another. Another is that pairing and segregation occur by different mechanisms. Our data demonstrate that quadrivalent formation is dependent on rDNA sequence homology, but segregation could depend on other properties of mini-X chromosomes, such as similarity in size or in heterochromatic topology, as has been previously suggested (Karpen et al. 1996). Our finding that homology-based quadrivalent formation underlies mini-X chromosome segregation sets the stage for future investigations into the mechanism of chromosome segregation from quadrivalents.
Acknowledgments
We are grateful to Chia-sin Hong for technical assistance. We thank G. Karpen for helpful suggestions on the mini-X mobilization crosses and R. Dorn for supplying anti-ModC antibody. This work was supported by National Institutes of Health grant R01 GM40489 to B.D.M.
Footnotes
Communicating editor: J. Tamkun
- Received March 25, 2007.
- Accepted July 20, 2007.
- Copyright © 2007 by the Genetics Society of America