Normal Synaptonemal Complex and Abnormal Recombination Nodules in Two Alleles of the Drosophila Meiotic Mutant mei-W68

Corresponding author: Adelaide T. C. Carpenter, University of Cambridge, Downing St., Cambridge CB2 3EH, United Kingdom., atc12{at}mole.bio.cam.ac.uk (E-mail)

Communicating editor: R. S. HAWLEY

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The meiotic phenotypes of two mutant alleles of the mei-W68 gene, 1 and L1, were studied by genetics and by serial-section electron microscopy. Despite no or reduced exchange, both mutant alleles have normal synaptonemal complex. However, neither has any early recombination nodules; instead, both exhibit high numbers of very long (up to 2 µm) structures here named "noodles." These are hypothesized to be formed by the unchecked extension of identical but much shorter structures ephemerally seen in wild type, which may be precursors of early recombination nodules. Although the mei-W68^L1 allele is identical to the mei-W68¹ allele in both the absence of early recombination nodules and a high frequency of noodles (i.e., it is amorphic for the noodle phene), it is hypomorphic in its effects on exchange and late recombination nodules. The differential effects of this allele on early and late recombination nodules are consistent with the hypothesis that Drosophila females have two separate recombination pathways—one for simple gene conversion, the other for exchange.

RECOMBINATION nodules (RNs; dense structures associated with synaptonemal complex and the bivalent around it during meiosis) were originally identified morphologically by conventional thin-section electron microscopy. Two types of RN have been described. The form that is present in mid-late pachytene is now called "late RN." The close correlation between numbers and distributions of late RNs and exchanges first noted in Drosophila females (CARPENTER 1975B ) has since been widely confirmed across species (ROEDER 1997 ; ZICKLER and KLECKNER 1998 ) and the hypothesis that late RNs perform roles in the exchange process itself is now widely accepted. The other form, "early RN," first described in human males (RASMUSSEN and HOLM 1978 ) and Drosophila females (CARPENTER 1979A ) as a pachytene structure appearing earlier than late RNs, is present in many species as early as earliest zygotene. (Many workers consider exact time of presence to be less important than the fact that there are two types of RN and they continue to call these zygotene structures early RNs; others choose to give them a new name, "meiotic nodules." I use early RN for both the zygotene and the pachytene structures.)

The conjecture that early RNs also have a role in recombination (RASMUSSEN and HOLM 1978 ; CARPENTER 1979A , CARPENTER 1981 ) has recently become more plausible since they have been found to be associated with enzymes thought to be involved in that pathway (ANDERSON et al. 1997 ; MOENS et al. 2001 ). Moreover, the conjectures that early and late RNs perform distinct roles and are not necessarily precursor product (POWERS and SMITHIES 1986 ; SMITHIES and POWERS 1986 ; CARPENTER 1987 ) are consistent with the results of MOENS et al. 2001 .

The functions of and relationships between RN types are addressed here from a different direction, the phenotypes of mei-W68 alleles. The hypothesis that early RNs perform a recombination function, specifically simple gene conversion (gene conversion without an accompanying exchange), is supported by the observation that mei-W68¹ homozygous females lack both simple gene conversion (MCKIM et al. 1998 ) and early RNs (this work). The hypothesis that late RNs arise de novo rather than from early RNs is supported by the observation that mei-W68^L1 homozygous females have late RNs but not early ones (this work). Furthermore, the unusual RN-like structures present in both mutant combinations, called here "noodles" in honor of the great lengths they can attain, allow sorting the amorphic (null) aspects from the hypomorphic (leaky) for the mei-W68^L1 noodle-RN phenotypes, with the conclusion that the wild-type product of this gene performs two distinct roles in recombination processes—consistent with the hypothesis that Drosophila females, at least, use separate pathways to produce the two alternative outcomes (simple gene conversion vs. exchange; CARPENTER 1987 ). mei-W68 is the Drosophila homolog of the Saccharomyces cerevisiae SPO11 topoisomerase II meiotic double-strand-break-producing enzyme (MCKIM and HAYASHI-HAGIHARA 1998 ). The mouse SPO11 homolog has also been proposed to perform two distinct roles during meiosis (ROMANIENKO and CAMERINI-OTERO 2000 ), although the precise roles may differ between flies and mice (see DISCUSSION).

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Genetics:
Crosses were done on standard cornmeal/molasses/agar medium and at 25° unless otherwise specified; parents were transferred on day 5 (day 0 is day of setup), discarded on day 10, and progeny were counted through day 17 of their vial.

The first mutant allele of the mei-W68 gene (mei-W68¹) was identified during the BAKER and CARPENTER 1972 search for X-linked meiotic mutations (B. S. BAKER, personal communication); since in that experiment the treated autosomes were specifically excluded by outcrosses, and since mei-W68 is on the second chromosome, this mutation must have occurred spontaneously either during those outcrosses or in one of those outcross stocks—which is entirely consistent with its lesion being the insertion of 5 kb of DNA of unknown origin into the second exon (MCKIM and HAYASHI-HAGIHARA 1998 ). The second allele, mei-W68^L1, was identified by Dan L. Lindsley in the late 1970s (D. L. LINDSLEY, personal communication); although he no longer remembers its origin, it was likely to have been EMS mutagenesis. Three separate mei-W68^L1-bearing second chromosomes were used in the genetic analyses of Table 1 and Table 2 and Fig 1; they are designated a, b, and c. Chromosome b was marked with al dp b pr cn; the other two were otherwise wild type. The mei-W68^L1 electron microscopy reported here was from females homozygous for the equivalent of chromosome a.

View larger version (19K): In this window In a new window Download PPT slide   Figure 1. Crossing over along chromosome 2 in various mei-W68 genotypes. The abscissa is the genetic interval in number of salivary bands between markers; the ordinate is map distance observed divided by the number of bands for that interval. Top, controls, heterozygotes, and mei-W68L1 homozygotes; curve A (crosses) is mei-W681/+, B is 18° control, C is 29° control, D is mei-W68L1 (a)/+ (solid circles), E is control from Table 1F is mei-W68L1 (b)/mei-W68L1 (c), G is mei-W68L1 (a)/mei-W68L1 (b), and H is mei-W68L1 (b)/mei-W681. Bottom, mei-W681 homozygotes (curve I) with 95% confidence intervals. Numbers of progeny for crosses not in Table 1 are: +/+ 29°, 284; +/+ 18°, 576; mei-W681/+, 293; summed mei-W681/mei-W681, 5480, including four clusters of crossovers.

Electron microscopy:
Fixation and sectioning for electron microscopy was done as reported earlier (CARPENTER 1975A ) in 1972–1980 and the initial reconstructions of pachytene nuclei and measurement of synaptonemal complex (SC) lengths were mostly completed by 1982, also as in CARPENTER 1975A . Initially I observed an electron-dense structure adjacent to the central region in a high proportion of SC images. Only after several failed attempts at analysis did I appreciate the 3-D structure they represent. I now think I understand what is happening, and I surely would not had this work been published earlier, so the long delay at least has a happy ending!

Lengths of synapsed chromosomal arms and of noodles were calculated as usual, that is, using the thickness of sections transversed and linear projection as the two sides of a right-angle triangle to calculate the hypotenuse by the Pythagorean formula (see Fig 3), taking segments as required whenever slope or direction changed in the reconstructions. However, estimation of the proportion of SC where noodles would be detectable (see RESULTS) and where they were detected was done using the linear projections of arms only; this means that segments where the SC is parallel to the plane of sectioning are overrepresented in this analysis and those where the SC is perpendicular are underrepresented in these estimates only.

View larger version (119K): In this window In a new window Download PPT slide   Figure 2. Synaptonemal complex and a normal late recombination nodule from mei-W68L1 homozygotes. (A and B) Two consecutive sections through a frontal segment of SC from A173 cyst VI cell d, showing normal substructure of synaptonemal complex and a normal late recombination nodule (arrows). (C) Diagram of the SC and RN as they were sectioned; A and B show these sections as viewed from the top. Magnification x63,600; bar (in B), 100 nm.

View larger version (98K): In this window In a new window Download PPT slide   Figure 3. A long noodle from mei-W681/mei-W681 (A312 cyst VIII cell b). The first column presents the SC as it was reconstructed within the whole nucleus; this is a slanting frontal series. The indicated noodle first appears in B (arrow), continues through C–E (arrows), is obscured by chromatin in F (arrowhead), and is completed in G. The second column presents just the SC + noodle; the SC is emphasized with ink and the noodle is circled (dashed circle where the noodle cannot be distinguished in F). The third column is a diagram of this segment of SC with its noodle, as sectioned; each section is viewed from its top in the photographs of the first and second columns and the sections are rotated 90°. Section thickness is slightly exaggerated to fit the figure; in reality sections are slightly thinner than the width of SC. The indicated noodle is 1120 nm (1.1 µm) long; 6 x 0.09 µm (section thickness) = 0.54 µm, length of the figure is 22 mm/22,400 (magnification) = 0.98 µm. Square root of [(0.54)2 + (0.98)2] = 1.12 µm. A second noodle (not indicated) is along the other side of the central element in A–D. Magnification x22,400; bar (in H), 120 nm.

Brief review of relevant Drosophila oogenesis:
Drosophila females have two ovaries, each containing 10–20 blind tubes called ovarioles, with an anterior germarium and posterior vitellarium. At the anterior end of each germarium reside 1–3 (occasionally 4) germ-line stem cells; these divide asymmetrically to regenerate an anterior-residing stem-cell daughter and produce a free daughter called a cystoblast. The cystoblast then divides four times with incomplete cytokinesis to produce a cyst of 16 interconnected cells; these divisions occur after "capping" of preexisting connections ("ring canals"), so the pattern of interconnections of the 16-cell cyst is nearly invariant—there are two cells with four connections, two with three, four with two, and eight with one. One of the cells with four connections will eventually become the determined oocyte, but the nuclei of both cells with four connections initially enter meiosis, simultaneously and nearly indistinguishably (although one always seems to be a little ahead of the other), and both remain in meiosis until the latter part of pachytene, roughly at the germarial-vitellarial transition. Older cysts move down the germarium both by being pushed by younger ones and by peristaltic movements of surrounding muscles, so the germarium is basically a linear tube of cysts in developmental order. However, the germarium is wide enough for two to four early 16-cell cysts to be at the same level, and when such a logjam breaks up the first down the tube may not be the oldest; cyst position within the germarium is therefore only a guide to relative age. Moreover, 16-cell cysts do not appear to be produced at regular intervals, so the temporal interval between adjacent cysts is variable. Nevertheless, the Drosophila system does provide a lot of natural temporal information from each (lengthwise) serially sectioned ovariole.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Genetic studies:
Extensive experiments performed by Bruce S. Baker in the late 1970s to study the genetic effects of the mei-W68¹ and mei-W68^L1 alleles on meiotic chromosome behavior have never been published; with his kind permission summaries of these results are now presented in Table 1 and Table 2. Since these experiments and their interpretation are standard (see BAKER and CARPENTER 1972 ; CARPENTER and SANDLER 1974 ) they will be only minimally discussed here.

Crossing over is reduced in females homozygous for either of the mutant alleles and also in mei-W68^L1/mei-W68¹ trans-heterozygotes (Table 1); the mutations are therefore alleles, although they differ dramatically in strength. mei-W68¹ lacks meiotic crossing over (see below), yet mei-W68^L1 shows significant levels—50% of wild type; that this is simply because the mei-W68^L1 lesion is hypomorphic (leaky) for this phene is shown by the greater reduction of crossing over in mei-W68^L1/mei-W68¹ females. That there is variation in strength of the mei-W68^L1 crossover defect depending on genetic background (Table 1 and Table 2) is also consistent with its defect being hypomorphic.

Reduced crossing over means that the frequency of homologs with no exchanges at all (E₀ tetrads) goes up; for low levels of E₀ tetrads the backup distributive system ensures regular homologous segregation, but when two or more major chromosome E₀ tetrads are in the same meiosis they may segregate, via the distributive system, irregularly. Frequencies of X, fourth, and third chromosome nondisjunction are presented in Table 2. For a detailed discussion of nondisjunction in Drosophila recombination-defective mutants, see BAKER and HALL 1976 ; chromosome segregation in these mutant combinations is as expected from the levels of reduction in exchange—we need not postulate any additional meiotic defects in them to account for it.

Both mutant alleles increase mitotic recombination in somatic tissues (BAKER et al. 1978 ) and, at least for mei-W68¹ (mei-W68^L1 was not tested), in the male germ line (LUTKIN and BAKER 1979 ). In females, the mei-W68^L1 allele has too much residual meiotic exchange for any premeiotic exchanges to be detectable, but they obviously do occur in mei-W68¹/mei-W68¹: 16 of the progeny in Table 1 came from clusters of 2, 6, and 8 crossover progeny from each of three single females. Moreover, all three of these clusters included both reciprocal crossovers, so the crossovers must have occurred during stem-cell divisions (see MATERIALS AND METHODS for a brief review of oogenesis in Drosophila). There was one additional "cluster" of unrelated events: One female gave one region 1 crossover progeny and one region 4. Five of the 16 cluster crossovers were X exceptions; none of the 8 unique crossovers was. Clustering of crossovers among the progeny of mei-W68¹ females was also reported by MCKIM et al. 1998 . Clusters are considered to represent one event in calculations.

An additional experiment with mei-W68¹ involved the whole second chromosome (as y/y; al dp b pr c mei-W68¹/mei-W68¹ px sp females crossed singly to +/Y; al dp b pr c px sp homozygous males) at 18°, 25°, and 29°; mei-W68¹/+ at 29°; and controls at 18° and 29° (y/+; al dp b pr c px sp/+ + + + + + + by +/Y; al dp b pr c px sp homozygotes).

As before, some of the crossovers from mei-W68¹/mei-W68¹ occurred premeiotically and present as identical clusters; again, each cluster is counted as only one event. There were no dramatic effects of temperature in mei-W68¹/mei-W68¹ so these results have been summed across temperatures for Fig 1, which plots the relative frequency of crossing over per unit physical for the various intervals (taken as the number of euchromatic salivary chromosome bands on the revised maps of Bridges between the markers used) with 95% Poisson confidence intervals (from tables of CROW and GARDNER 1959 ). The few crossovers recovered from mei-W68¹/mei-W68¹ females do seem to be uniformly distributed along the euchromatin, as expected if they all derive from events during mitosis. However, unlike somatic crossovers induced with X rays, the data suggest that events in mei-W68¹/mei-W68¹ females are excluded from the centric heterochromatin (if events were as likely to occur in heterochromatin in mei-W68¹/mei-W68¹ as in euchromatin, the frequency of crossovers in the pr-c interval would be 1.7 times that in any wholly euchromatic interval; instead, crossovers there are, if anything, slightly less frequent), which is like meiotic events.

Recombination data from the other genotypes have also been plotted as per unit physical in Fig 1. The control data from Table 1 (2L and centric region only) show the expected pattern of distal high, proximal low, minimum around the centromere; this is a graphic demonstration of the centromere effect. Both mei-W68^L1/mei-W68^L1 genotypes not only reduce the frequency of distal events but also flatten the curve relative to wild type and, although the frequency of events differs by a factor of two between the two mei-W68^L1 genotypes, the distributions of those events along the arm are similar: 2 x 4 contingency ² of numbers of crossovers observed in the four regions in the two genotypes = 8.5 (3 d.f.); 0.05 > P > 0.01. In mei-W68^L1/mei-W68¹ the curve is very nearly flat. This curve flattening is diagnostic of mutations defective in preconditions for exchange (events necessary before the actual initiation of the exchange process)—mutations defective in exchange itself have the same curve shape as wild type (CARPENTER and SANDLER 1974 ). The same broad trends are seen for both left and right arms in the whole-chromosome 18° and 29° control data; the effects of temperature on distribution of crossovers also present as effects on preconditions for exchange (LIFSCHYTZ 1975 ).

Both the mutant alleles mei-W68^L1 and mei-W68¹ are recessive to wild type when nondisjunction frequencies are examined (Table 2), but both have dominant effects on crossing over. The effect for mei-W68^L1 (Table 1 data) is slight but significant; 2 x 4 contingency ² of control and mei-W68^L1/+ crossovers by region (3 d.f.) gives 15.7, P < 0.01. The dominant effect of mei-W68¹ is striking. As can be seen in Fig 1, the map distances in proximal regions are considerably higher than those in the 29° control and less so or not at all in distal regions (2 x 6 contingency ² = 12.1, 5 d.f.; 0.05 > P > 0.01). Again, the overall effect is a flattening of the curve, and this indicates that the dominant effect of mei-W68¹ on exchange is a precondition defect. Strong mutant alleles of other genes necessary for normal meiotic recombination also show dominant effects on crossing over (HALL 1972 ; CARPENTER and SANDLER 1974 ): The precondition-defective mei-218¹ gives an increase that is greatest around the centromere, exchange-defective mei-9^a gives a decrease with the same distribution along the arm as wild type, and c(3)G¹ also gives an increase greatest around the centromere (HINTON 1966 ; HALL 1972 ) and therefore was presumed also to be a precondition defect; the residual exchange in a partially functional construct is polar (PAGE and HAWLEY 2001 ) so this presumption is correct. mei-P26 heterozygotes present a polar decrease (SEKELSKY et al. 1999 ). Both the recessive phenotype of the hypomorphic mei-W68^L1 allele and the dominant effect of the amorphic mei-W68¹ allele strongly suggest that the crossover function of mei-W68⁺ is a precondition, a function that is involved in establishing sites where exchange events can occur rather than a function involved exclusively in the production of those exchange events themselves.

Cytological studies:
Dynamics of meiotic progression: Four germaria from homozygous mei-W68¹ females, two from mei-W68^L1, and one from mei-W68^L1/mei-W68¹ were serially sectioned and low-magnification micrographs were taken every sixth to tenth section so that cysts and pro-oocytes could be identified by their ring-canal connections; the data for numbers of cysts of the various sizes are presented in Table 3 and for meiotic stages of 16-cell cysts in Table 4. Although there is a good deal of variability between germaria (as in wild type), the average numbers of cysts of the various sizes are well within the ranges seen in wild type; mei-W68 mutations have no effect on the dynamics of germ-line mitotic divisions (Table 3).

They also have no effect on the dynamics of entry into meiosis or exit of the losing pro-oocyte (Table 4): In particular, neither the frequency of prezygotene nor of zygotene cysts is higher than that of controls—establishing complete SC formation is not delayed even in homozygous mei-W68¹ females (contra HABER 1998 ). (Staging here is based solely on the low-magnification photographs, as were the controls; one cyst—A173 IV—was judged to be in pachytene from the low-magnification photographs but turned out to be in midzygotene once the high-magnification photographs were examined.) The one aspect in which these mutants may differ from wild type and other meiotic mutants is in the frequency of necrotic 16-cell cysts: 5/52 = 9.6% of cysts are dying in these seven germaria, whereas 9/189 = 4.8% were observed previously (CARPENTER 1979A , CARPENTER 1979B ).

Synaptonemal complex and recombination nodules: All sections of 14 mei-W68¹/mei-W68¹ nuclei were photographed at higher magnification (all of germarium A312 plus a few others) as well as all sections of 14 mei-W68^L1/mei-W68^L1 nuclei; synaptonemal complex (SC) was reconstructed by tracing its profiles from each photograph onto acetate sheets. Registration of the 40–50 sections per nucleus used all available information (SC, nuclear envelope, ring canals, and other cytoplasmic structures as needed; see CARPENTER 1975A , Figure 13, for examples of how much information is available) but always sought the minimal SC length. Lengths and other measurements were then calculated as previously described (CARPENTER 1975A ). Euchromatic lengths of arms and other observations are in Table 5 and Table 6. SC appears to be of normal morphology (Fig 2 for mei-W68^L1; see Fig 3 and MCKIM et al. 1998 for mei-W68¹) and to be between homologs: (1) Only one SC arm goes into each nucleolus or "blob" (an additional cytological landmark, see CARPENTER 1975A ), and the X (arm associated with the nucleolus) is usually the longest arm and the blob arm the shortest, as in wild type; (2) either euchromatic SC is continuous or, if it is interrupted, that interruption is not in the vicinity of any other abnormality that could be a pairing-partner switch (and the occasional short SC interruption is also seen in wild type); and (3) each SC arm occupies its own region of the nucleus without interlocking with other arms, as also observed in wild type. This regionalization could perfectly well be the mechanical result of cooperative decondensation of already somatically paired homologs after the last premeiotic division (see DISCUSSION). Only one nucleus (mei-W68¹ germarium A312, cyst VIIc) shows any evidence of synaptic difficulties (see footnotes of Table 5) but nothing suggests that any of these irregularities are related to each other. If anything synapsis is better in mei-W68¹ than in wild type (Table 5, footnote a). Developmental-age-dependent SC reorganizations are also normal; lengths decrease with age (Table 5 and Table 6) as observed in wild type (CARPENTER 1979A ); and as in wild type, as the SC shortens it also becomes thicker (data not shown but see Fig 3 and Fig 4).

View larger version (106K): In this window In a new window Download PPT slide   Figure 4. Three examples of noodle segments, illustrating the variety of presenting morphologies, all from mei-W68L1 homozygotes (A173 cyst VI cell d); magnification, x44,600; bar, 100 nm. (A) Four consecutive cross sections through SC and a long noodle (arrows); this series is from the top bend in D. (B) Three consecutive cross sections through SC with a long noodle along each side (arrows) and a diagram of the lowest section below it; this series is from the middle bend in D. (C) Three consecutive cross sections of SC and long noodle (arrows) and a diagram of the lowest section below it; the noodle image in the central section has an early-RN-like appearance. (D) Diagram of the two long noodles mentioned in text. Stippled bar, central element seen from its side; lines, noodles as detected. This series was sectioned perpendicularly to the diagram, in cross and slanting frontal section; the noodles could not be detected in the slanting frontal parts. Note that both noodles became undetectable at the same SC angles even though those photographs were of different sections.

Late RNs are now accepted to be at sites where exchange-type recombination events are being attempted; there are no late RNs in mei-W68¹/mei-W68¹ nuclei. At least 6 and possibly 10 nuclei are within the developmental range for late RN presence in wild type, to wit between the onset of cytoplasmic flow and oocyte determination (inclusive); 20–34 late RNs would have been predicted (CARPENTER 1979A ); none were observed. The role of early RNs is still unclear but, if they also function in recombination, then they must be at sites that resolve only as simple gene conversions (gene conversion without an accompanying exchange) or restorations without exchange (CARPENTER 1987 ). Early RNs are smaller than late ones and consequently slightly harder to see, but in the 5 (4 from A312, 1 from A282) mei-W68¹ nuclei at the right developmental stage (first cyst showing cytoplasmic flow plus next younger cyst) there was at most one thing that could possibly be an early RN (see Fig 6A; 14 expected, CARPENTER 1979B ). Like all other precondition recombination-defective mutations that have been examined (CARPENTER 1979B ), mei-W68¹ reduces meiotic exchange events and late RNs to the same extent (i.e., to zero); unlike them, however, it also abolishes both simple gene conversion (MCKIM et al. 1998 ) and early RNs—providing the first clear link between the presence of early RNs and simple gene conversion.

View larger version (26K): In this window In a new window Download PPT slide   Figure 5. Tracings of SC (solid lines) and long noodles (side lines) from mei-W681 A312 VIII b. Dashed lines indicate entry into heterochromatin; cross-hatching indicates SC whose plane of sectioning was such that noodles would not be detectable. NO, nucleolus; Bl, blob.

View larger version (39K): In this window In a new window Download PPT slide   Figure 6. Lengths of noodles in nanometers. x, unbounded (both ends of that noodle are known); o, bounded (one or both ends of that noodle disappear into a region where it would be undetectable from plane of sectioning); g, gapped (continuity through a short obfuscated region is assumed); E, might possibly be an early RN (see text); square, part of the noodle is early RN-like. Note breaks in abscissae. (a) mei-W681 A312; (b) mei-W68L1 A173 and A170.

Late RNs are present in mei-W68^L1/mei-W68^L1 nuclei (Fig 2) and are discussed below; only one possible early RN was found (see Fig 6B; again, 14 would have been expected in the five nuclei of the right developmental age). Unfortunately mei-W68^L1's effect on simple gene conversion is not yet known.

Noodle detection: However, both mei-W68 alleles have a novel structure, which I call "noodle," that resembles RNs in being adjacent to the central region of the SC (Fig 3 and Fig 4). Noodles are smaller than RNs, being on average only 20 nm in diameter [early RNs are 35 nm (CARPENTER 1979A ), where what are now called early RNs were called "ellipsoidal"]. They are also less dense as measured by the degree to which the various stains react with their components: Even when the SC and the noodle are perpendicular to the plane of sectioning (cross section, Fig 4) so that 90 nm of noodle length is imaged onto the same part of the photograph, the structure still appears less dense than that of an equivalent early RN—instead, it is closer to the density of chromatin (Fig 3 and Fig 4). They also exhibit morphological variability (Fig 4).

The above three attributes hamper the analysis of noodles. First, like the lateral elements of the SC (CARPENTER 1975A ), they are not detectable in all planes of section, presumably because they do not make enough of an image to be seen against background unless they reinforce themselves through at least part of the section. This is obvious in slanting frontal series, where the SC is continuous through the series but the image of the noodle is not—it becomes undetectable around the join between sections, in the region where it would be split between the two sections (Fig 3). Second, no noodles were ever observed when the SC was sectioned sagittally (sideways to Fig 2C, see CARPENTER 1975A ), either in the section with the SC or in the two flanking sections. This cannot be chance; although some noodle images might be rendered undetectable because they are in a section that also contains chromatin or even hidden under the image of the central element of the SC, most sagittally sectioned noodles must be undetectable from lack of reinforcement of the image. There are many examples where the noodle appears to end just at the point where the SC rotates into what is sagittal from the plane of sectioning. A similar problem occurs when the SC is caught in very accurate frontal section at the bottom/top of a loop; there are several examples of clear noodles before and after the (slanting frontal) loop that disappear right at its apex. The most convincing example of this (Fig 4D) is a case where there are noodles along both sides of a stretch of SC that is in cross section in the middle of this example, with both noodles visible; then as the SC bends into slanting frontal section in each direction both noodles disappear—at the same angle of SC bending even though this occurs in different sections on its two sides. One of the noodles then reappears in both directions as the SC bends back into cross section. Serial section electron microscopy (EM) alone is therefore insufficient to measure either the total number of noodles per nucleus or their maximum length (see below), although it is quite sufficient to establish minimum estimates of both. Third, noodles are sufficiently nondescript in appearance that they often cannot be unambiguously identified if they are short (less than two sections if in cross section, less than 100 nm in slanting frontal); again, this means that their numbers are underestimated. Nevertheless, quite a lot of information about them can be extracted from the data.

Because noodles are hard to see, one needs a special tactic for finding and charting them. For early and late RNs, which are relatively conspicuous, short, dense bodies, examining all the SC profiles in each nucleus photograph by photograph detects them all; for these new structures, one must follow each arm individually as it meanders up, down, and around the nucleus (and the photographs), finding a possible noodle image in one photograph and then examining the continuation of that arm in the adjacent ones. In some cross-sectional series individual photographs have chromatin sufficiently around the top/bottom of the SC so that a noodle could not have been distinguished from it (see Fig 3F); as long as a noodle was present on both sides of this obfuscation it was counted as continuous. Many of the longer structures pass through a region of SC where they would be invisible from the plane of sectioning (see above); these too are considered to be continuous but carry the flag of being "gapped" and for most of these the gaps are relatively short. Fig 5 shows the tracings of the arms of a typical nucleus. The two gapped noodles of the bottom arm were each considered to be continuous, whereas the top right arm was scored as having three noodles. In addition, each noodle had recorded whether both ends were in regions where continuation could have been seen had it occurred ("unbounded" structures whose total length is certain) vs. one or both ends disappearing into regions where continuation could not be seen ("bounded" structures whose real total length may be greater than that measured). Finally, along each arm of the SC it was determined whether the plane of sectioning was such that a noodle could have been seen if one was present—over all nuclei, this averages to 50%. The data for lengths are presented in Fig 6 and Table 7 (lengths were measured and calculated on the three-dimensional reconstructions, not the two-dimensional projections). With but one exception, only structures that are clearly real noodles are included in Fig 6 and Table 7; additional images that for one reason or another are dubious have been excluded. However, since those are all short, even if some of them are real their absence has little effect on the following.

Noodle position and length: It is obvious from Fig 3 Fig 4 Fig 5 that a noodle can be along each of the two faces of a particular stretch of SC; when this occurs there is no obvious relationship between either the ends of the two noodles or their apparent midpoints. Moreover, the "overlap" of such pairs of noodles is no more common than expected by chance, taking into account for each arm the length (in millimeters on the two-dimensional projections) where a noodle could have been seen if present (times two for the two faces of SC) and the total length (in millimeters on the two-dimensional projections) of noodles seen on that arm. For nuclei with only a few short noodles little overlap is expected and little or none is seen; for nuclei with lots of long noodles more overlap is expected and again that is what is seen. Summing over all 10 nuclei from the A312 germarium, 198.5 mm of overlapping noodles are expected; 169.5 mm were observed. Taken together, these two observations strongly suggest that there are no obligate or even preferred regions along the arms where noodle formation initiates.

From the data in Fig 6 and Table 7 it is clear that at least some noodles lengthen; both the mean and maximum lengths increase with increasing developmental age of the pachytene nuclei. Moreover, in both genotypes new noodles are initiated through at least part of the time course assayed; the number detected increases at least up to the time cytoplasmic flow begins (time of presence of late RNs in wild type) and, up to that point, noodle numbers and lengths in mei-W68¹ and mei-W68^L1 are very similar. In mei-W68¹, noodles persist post-cytoplasmic flow and probably continue to elongate but, as is discussed below, in mei-W68^L1 significant numbers disappear once cytoplasmic flow begins. Later pachytene (definitive oocyte determined) nuclei still have noodles, although their numbers may be decreasing. There is no information here on when (or even whether) noodles disappear in mei-W68¹; all of the older but prekaryosome nuclei photographed still have them. No karyosome nuclei were photographed but noodles would not be distinguishable from the condensed chromatin of that stage anyway.

Noodles in wild type: Long noodles have not been observed at any stage of pachytene in wild type. To look for short noodles, two wild-type germaria were chosen for careful analysis (A12 and A139, CARPENTER 1979A ) because they had pre-early RN pachytene cysts; all arms of all pachytene nuclei of both germaria were examined as discussed above for possible noodles. Although as expected most nuclei did have individual images that might be possible noodles, four nuclei really do have some short ones (Fig 7A and Fig B). They are A139 cyst II cell a (1) and IV i (4); and A12 III a (4) and III f (6). These are all pre-early RN cysts. See DISCUSSION for possible relationships between noodles and RNs, but noodle presence is not due to the mei-W68 defect; only noodle persistence and the phenomenal lengths that made them detectable in the first place are.

View larger version (169K): In this window In a new window Download PPT slide   Figure 7. Other examples. (A and B) The two consecutive cross sections through a short noodle from wild type (from A12 cyst III; CARPENTER 1975A ), with tracings below; (C) a late RN between sister chromatids (from A173 mei-W68L1 cyst c) in cross-sectional view. The late RN is indicated by a single arrow; where it should be (to be between homologs) is indicated by arrowheads. Bar (in A), 117 nm.

Early and late RNs: Neither mei-W68¹ nor mei-W68^L1 has any unambiguous early RNs, although both have the occasional structure that looks a bit thicker and denser than that of a typical noodle (see Fig 4C, middle); almost all of these are continuous with bona fide noodles (either at one end or internally) and these possible intermediates are more frequent in mei-W68^L1 (9/132) than in mei-W68¹ (1/154; Fig 6, a and b). However, they are found throughout the age range monitored rather than being restricted to the earlier nuclei as in wild type, which is consistent with them being either chance spurious images or the cell's continuing futile attempt to "mature" a noodle into an early RN. A total of 28 early RNs would have been expected in the 5 mei-W68¹ and 5 mei-W68^L1 nuclei of the right developmental age. No late RNs at all were observed in the 6–10 mei-W68¹ nuclei within the late RN period of wild type; 20–34 would have been expected.

mei-W68^L1, on the other hand, does have late RNs, which first appear at the same time as late RNs in wild type (the youngest cyst with cytoplasmic flow, Table 6). Together, the three mei-W68^L1 nuclei of the right age have a total of nine late-RN-type objects, but only three of these are "normal" late RNs (Fig 2A and Fig B); of the others, two have normal morphology but are in the wrong place—off the side of the SC, between sister chromatids rather than between homologs (Fig 7C); one is between homologs but of irregular shape, two are between homologs but are small, and the last one is of normal morphology but the SC is so twisty that its location is ambiguous. Late RNs of abnormal morphology have been observed in other meiotic mutants (CARPENTER 1975B ) but I have never observed a late RN between sister chromatids in any other genotype (with the possible exception of one out in the chromatin in mei-41; CARPENTER 1979B ). Unfortunately, sister-chromatid exchange has not been examined in mei-W68^L1. However, it is extremely unlikely that these two late RNs could mediate exchange between homologs, leaving a maximum of seven late RNs/three nuclei = 2.33 ± 1.15 in mei-W68^L1 vs. 3.37 ± 0.24 in wild type (CARPENTER 1975A ). mei-W68^L1 reduces exchange to 50% of wild-type levels; unfortunately the number of mei-W68^L1 nuclei sampled is too low for meaningful comparisons. Either hypothesis—late RNs reduced to half wild-type levels (all RNs functional) vs. late RNs at wild-type levels (approximately one-half of RNs functional)—is consistent with the data.

However, and astonishingly, no arm that has a late RN between homologs has an unambiguous noodle. The sole ambiguous noodle (A173, first vit arm A₃) has been left in for emphasis; the other five arms had no ambiguous ones. (The arm with a late RN of uncertain location is left out of these considerations.) The other eight arms had a total of 25 noodles, or an average of 3.1 noodles per arm, entirely consistent with the numbers of noodles seen in nuclei of equivalent age in mei-W68¹. Since numbers of noodles in mei-W68^L1 and mei-W68¹ are entirely consistent in precytoplasmic flow nuclei, this must mean that noodles are stripped off the SC around the time a late RN is formed between homologs; this requires not only action at a considerable distance (up to 10 µm) but also communication between the two surfaces of the SC—recall that noodles can and do form on both of the surfaces, yet both surfaces are clear of them along the arms that have a late RN. Moreover, the two arms that have a late RN between sister chromatids have kept their noodles; the process that recognizes the presence of a late RN (or some key enzymatic step in the exchange pathway) seems to be able to discriminate between events between homologs (remove noodles) and events elsewhere (keep them). This is remarkable!

Noodle distribution: Addressing noodle distributions along the arms is complicated by two things: the fact that the plane of sectioning is such that noodles are undetectable 50% of the time and the fact that noodles, especially in older nuclei, are not points. For the first, assume that "wrong plane of sectioning" is random along the length of each arm; this is not quite true (see below about tips of SC), but it nearly is, so the noodles detected can to a first approximation be considered to be a random sample of the total. For the second, the only solution is to plot the whole length of each noodle. Although it is clear that noodles get longer with time, we do not know whether they extend from both ends vs. unidirectionally, so taking either the apparent middle or one or the other end is making an unwarranted assumption; moreover, especially for the longer ones, noodles often disappear into a region of wrong plane of sectioning so it is unclear for those where its middle is. So the extent of each of the bona fide noodles was marked on the three-dimensional reconstructions, and their extents and locations along the arms were calculated. Since the arms shorten with increasing developmental age, each arm was then normalized to 100% of the distance between the telomere and entry into heterochromatin for the autosomal arms A₂, A₃, and A₄, which were sorted by relative length. This is reasonably reliable: One mei-W68¹ and six mei-W68^L1 nuclei had split chromocenters that separated the major autosomal arms and in all seven cases the other arm of the blob arm was A₄ (the shortest of the three arms that lack a landmark), which is reassuring in terms of reliability of arm identification by length and also permits tentative arm assignment. If the blob represents an elaboration of the clustered histone loci, which are close to the euchromatin/heterochromatin boundary on 2L, then A₄ is 2R and A₂ and A₃ are the arms of chromosome 3 with A₂ probably being 3R on the basis of relative euchromatic lengths.

The data for A₂, A₃, and A₄ are presented in Fig 8 in two ways: The outlines represent the sums of coverage, with each noodle being counted as present in each 1% it extends over, and the symbols represent the apparent midpoints of individual noodles. Each arm is presented separately and the noodles from mei-W68¹ and mei-W68^L1 have different symbols. In addition, the nuclei are sorted into two age groups: those from cysts before the onset of cytoplasmic flow (pre-cf) and those after (post-cf). This time point corresponds to the initial appearance of late RNs in wild type. More pre-cf mei-W68^L1 nuclei than mei-W68¹ were reconstructed and more post-cf mei-W68¹ nuclei than mei-W68^L1; in both cases, however, the average numbers of noodles per nucleus (or per arm, see below) are quite similar (Table 7).

View larger version (31K): In this window In a new window Download PPT slide   Figure 8. Patterns of noodle presence along autosomes A2, A3, and A4. Telomeres are at 0, entry into heterochromatin is at 100. For each arm, midpoints of noodles are given above and total coverage below. Solid triangles and solid bars, mei-W68L1 precytoplasmic flow; open circles and open bars, mei-W681 pre-cf; x's, mei-W68L1 postcytoplasmic flow; solid circles and stippled bars, mei-W681 post-cf.

If each arm has its own particular noodle pattern, then to a first approximation the distributions for each arm of mei-W68¹ and mei-W68^L1 noodles would be expected to be the same and so would distributions pre-cf and post-cf. Departures from either of these expectations would be interesting (for example, noodles initiated after cf might be distributed differently from those initiated before), but such departures are expected to be arm independent. To facilitate comparisons the numbers of noodles were summed by midpoint per one-tenth arm (to reduce granularity) and plotted pre-cf vs. post-cf and mei-W68^L1 vs. mei-W68¹. None of those distributions differed significantly—2 x 10 contingency ² tests (9 d.f.) give P values of 0.1 and up for each, meaning that for all arms pre-cf and post-cf can be considered to have been drawn from the same population—likewise for mei-W68^L1 vs. mei-W68¹. One can therefore use the summed data per arm to ask whether the numbers of noodles per tenth is random (uniform): ² (9 d.f.) for the X = 7.4 (P > 0.5), for A₂ = 11.4 (P > 0.2), and for A₄ = 16.8 (P > 0.05); the blob is marginally significantly nonrandom at 20.0 (0.05 > P > 0.01) and A₃ is highly significant at 24.0 (P < 0.01). For both the blob arm and for A₃ half the contribution to ² comes from just one interval, 0.1–0.2.

The four autosomal arms can be considered to be different samples drawn from the same population (4 x 10 contingency ² still using midpoints of noodles summed over tenths = 41.3, 27 d.f.; this gives a P value only very slightly <0.05); the total autosomal sample of 219 noodles is plotted in Fig 9A. This distribution is not uniform—² (9 d.f.) = 34.2, P << 0.01, with the subterminal interval (0.1–0.2) and the three most proximal intervals (0.7–1.0) contributing 29 of the 34. However, all of the nonuniformity derives from the post-cf sample [ Fig 9B; here, the numbers of pre-cf noodles per one-tenth arm have been adjusted down to the same number of arms (30) present in the post-cf sample to permit direct visual comparison (118 noodles across 52 arms reduced by a factor of 30/52) vs. post-cf (103 noodles across 30 arms)]. The pre-cf distribution is uniform; the post-cf distribution is highly significantly nonrandom, ² (9 d.f.) = 38.8, P << 0.01.

View larger version (20K): In this window In a new window Download PPT slide   Figure 9. Summed distributions of autosomal noodles from both mutant combinations. 0, telomere; 100, heterochromatin. (a) All noodles. (b) Noodles sorted into those from precytoplasmic flow cysts and from postcytoplasmic flow cysts. "Pre" has been adjusted to 30 arm equivalents; see text. (c) Size distributions of post-cf noodles from the 10–40% interval (solid line) and from the rest (dashed line). A total of 25% of noodles from the 10–40% interval are <300 nm long; 30% of the rest are.

It was noted earlier that noodles appear to continue to be initiated and then elongate throughout at least the pre-cf segment of pachytene, so it is of interest to ask whether those initiated later have the same arm-length distribution as those initiated earlier. However, those initiating later will be doing so against a statistical background of preexisting (and elongating) noodles in the same length segment on other arms as well as noodles along different segments of the same arm. It is clear that the proximal halves of the distributions in Fig 9B are the same; the additional noodle per arm in the post-cf sample (118/52 = 2.27 pre-cf, 103/30 = 3.43 post-cf) appears to be nonrandomly added in the 10–40% length segment, although contingency ² comparison of the pre-cf and post-cf distributions (done on the original numbers) shows that they are not significantly different (² = 11.3, 9 d.f., P > 0.1). Furthermore, if all the "new" post-cf noodles initiated in the 10–40% interval, then the average length of noodles there is predicted to be significantly shorter than elsewhere (and this should be detectable, since 31/55 noodles are predicted to be new since the initiation of cf); however, as can be seen in Fig 9C, the size distribution of noodles in the 10–40% cf length interval is exactly the same as that on the other 70% of the arms, very strongly negating the hypothesis that later-initiating noodles are preferentially in either of these arm segments. An alternate hypothesis that accounts for this (nonsignificant) difference between pre-cf and post-cf noodle distributions is that their numbers are more representative of chromatin length than of SC length; SC shortening, which is continuing throughout the portion of pachytene surveyed here, is accompanied by thickening of the SC and its associated chromatin, but this thickening is much more pronounced in the more distal regions—consequently, noodles that have initiated earlier in the longer region that will become the 10–40% interval in the later nuclei will be accumulated there (see also DISCUSSION).

Last, the apparent exception to uniform distribution of noodles, the very distalmost regions, may be real but two kinds of artifact may be affecting the data. One is that longer noodles are more likely to be detectable somewhere along their length; middle regions of arms can be covered by noodles extending from either more proximal or more distal regions. Subtelomeric regions can be covered only by noodles initiating there or extending from more proximal regions. The other is that the very tip of the SC nonrandomly presents as sagittal in appearance. This may indicate a propensity for extremely distal SC regions to have different substructural organization from that of more proximal regions or it may just be chance, but regardless it means that even if noodles were present I could not detect them (see above).

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Having two mutant alleles of the mei-W68 gene, of different origins and of different strengths, is very useful in unraveling the various aspects of the wild-type allele's function. The original allele, mei-W68¹, was spontaneous and is associated molecularly with the insertion of 5 kb of DNA of unknown origin into its second exon (MCKIM and HAYASHI-HAGIHARA 1998 ); this should kill the protein and indeed, at least in terms of its effect on exchange, the mei-W68¹ allele does appear to be amorphic (null). The second allele, mei-W68^L1, was probably chemically induced and in most aspects of its phenotype (somatic exchange, BAKER et al. 1978 ; meiotic exchange and presence of late recombination nodules, this study) it behaves as a hypomorph (leaky). However, in its effects on early recombination nodules and noodles it behaves as an amorph—at least in these phenes it is indistinguishable quantitatively and qualitatively from the amorphic allele mei-W68¹. This complexity at the genetic level implies that the wild-type protein performs several functions, some of which are more drastically affected than others by the lesion(s) in mei-W68^L1.

The wild-type allele has been cloned and sequenced (MCKIM and HAYASHI-HAGIHARA 1998 ); it has homology to the same archebacterial topoisomerase II that the S. cerevisiae SPO11 gene does. However, the phenotypes of amorphic alleles in the two species differ. In the first place, the yeast spo11 amorphic condition lacks SC (ROEDER 1997 ); in mei-W68 mutant flies abundant SC not only is formed but also appears to be normal in all respects—time of initiation, between homologs, length, and length changes, etc. In the second place, the yeast spo11 protein is involved directly in the exchange process; in its absence neither recombinants nor double-strand breaks are produced. Although in flies meiotic (though not mitotic) recombination does depend absolutely on the mei-W68 protein, the genetic evidence indicates that its role is in a precondition for exchange (see RESULTS), that is, a function involved in determining the numbers and locations of exchange events rather than being exclusively part of the recombination enzymology. Although there is no a priori reason to deny that functions involved in permitting the precondition process to function normally might also be involved in DNA metabolism and therefore also function directly in exchange, or even that specific proteins might perform two or more sequential roles (in which case the phenotype of amorphic mutants would reflect the earliest role), it is possible that the wild-type mei-W68 topo-II-like protein actually functions as a normal topo II, relaxing DNA, rather than as a defective topo II like the yeast SPO11 wild-type gene, which can only make the nick, not repair it again.

Regardless of its molecular role, wild-type mei-W68 function is necessary for the occurrence of both early and late recombination nodules, and in fact the absence both of early RNs and of simple gene conversion (gene conversion without a concomitant crossover) in the amorphic mei-W68¹ allele is the first evidence that supports the hypothesis (RASMUSSEN and HOLM 1978 ; CARPENTER 1979A ) that the structure, early RN, and the recombinational outcome, simple gene conversion, are related—at least in flies. In both mutant alleles, however, long novel structures here called noodles appear instead; like RNs, noodles are located directly over the central region of the SC (thus being between homologous chromatids) but they are even smaller in diameter than early RNs. However, there are more noodles per nucleus than the sum of early + late RNs (Table 7) and noodle numbers are surely an underestimate by as much as twofold because of the problems of detecting them. Moreover, unlike RNs, which form, function, and go away, in mei-W68¹ and mei-W68^L1 females noodles form, stay, and elongate.

So what are noodles? That short ones can be detected in wild type, in pre-early RN nuclei, suggests that they are normally a transient precursor of early RNs. [This hypothesis predicts that noodle-bearing and early-RN bearing cysts should be adjacent in age; unfortunately, this early in pachytene several cysts are at the same level of the germarium, forcing the use of SC length as the only gauge of relative age. For germarium A12, this does sort the noodle cyst (III) as next younger than the first early RN cyst (V); for A139 it does not. A139 length order is cyst II, late zygotene (1, 0)–cyst IV (4, 0)–cyst III (0, 0)–cyst V (0, 0 but both early and late RNs); see CARPENTER 1979A . This could mean that wild-type noodles are not early RN precursors but I think it is more likely that SC length alone is not a sufficiently accurate indicator of relative age.]

The long mei-W68-specific noodles probably represent the unchecked development of such precursors, which in wild type are transformed into early RNs very quickly but whose transformation cannot occur in either mei-W68 allele, leaving this putative early RN precursor to carry on extending. Note that this hypothesis implies that the pachytene nucleus has a way to monitor successful early RN formation, since noodles continue to be initiated throughout at least early pachytene in mei-W68¹ and mei-W68^L1 females.

Consistent with the hypothesis that noodles are untransformed early RN precursors are the observations of the occasional short noodle segment that does look a little like an early RN (Fig 4C, middle, is the best example in the entire sample) and also the apparently random distribution of noodles along the arms.

The fact that noodles—and very long ones as well—remain over the central element during the SC's midpachytene substructural rearrangments probably tells us that noodles (at least in mei-W68 alleles) are not bound irreversibly to chromatin. The logic is as follows. The very early pachytene bivalent—SC plus its associated chromatin—is very thin (top to bottom of Fig 2C; width of SC remains constant), but as pachytene progresses the bivalents, including the SC itself, shorten and thicken (CARPENTER 1975A ). For the SC, this probably involves simple rearrangement of SC substructural components (possibly as a unit of piece of lateral element + transverse fiber + core unit of central element + other transverse fiber + other piece of lateral element), but even if it involves local disassembly and reassembly the important point is that the chromatin goes too; chromatin is uniformly distributed across the face of lateral elements regardless of whether they are thick or thin (see CARPENTER 1975A for pictures and diagrams). In other words, the proportion of chromatin that is near the top or bottom of the lateral elements (and therefore near a noodle if one is over the central element) decreases as pachytene progresses. One could imagine that very short noodles could be holding on to a segment of chromatin, keeping it near the top of the SC during these rearrangments, but noodles can be very long; moreover, there are several cases of very long noodles on both sides of very thick SC. The rest of the chromatin would have to undergo major contortions if both those noodles maintained earlier chromatin contacts along their lengths. This reasoning by no means eliminates either transient noodle-chromatin associations or a few local, more permanent associations. It should be noted that the same logic requires that the association between a noodle and the central element be either transient or local.

It is not unreasonable to expect that topoisomerase function would be involved in a DNA-protein interaction, so the hypothesis here is that, in wild type, the mei-W68 protein mediates the initial DNA-short noodle interaction; when this is "successful" (see below) the short noodle matures into an early RN—and some sort of signal goes out along the bivalent, halting or slowing further short noodle initiations. When it is unsuccessful, as it is in both mei-W68 mutant alleles, that short noodle extends along the central element to become a recognizable long noodle and additional short noodles initiate. That there is no difference between mei-W68¹ and mei-W68^L1 in precytoplasmic flow noodle numbers or lengths strongly implies that the lesion(s) in mei-W68^L1 completely destroy this aspect of its protein's function, however it may operate biochemically.

The effects on late RNs are different. Morphologically normal late RNs are known to be present before the completion of all enzymatic reactions at sites that, in wild type, would become exchanges (CARPENTER 1979B ) and quite possibly are present very early in, or before, the DNA-metabolizing reactions. mei-W68 function is absolutely required for late RN formation and exchange because the mei-W68¹ allele lacks both; there is no alternative, redundant pathway in flies. However, although the lesion(s) in the mei-W68^L1 allele has an effect here, it is really quite mild; late RNs are present and the reduction in exchange is only to 50% of wild-type levels, implying that the mei-W68^L1 product is approximately half as efficient as wild type in its role in the exchange process (and in mei-W68^L1/mei-W68¹, where there is only one-half as much partly functional mei-W68^L1 product as in mei-W68^L1/mei-W68^L1, exchange is reduced by another 50%). This difference in the effect of mei-W68^L1 on noodles vs. late RNs/exchange implies that the wild-type mei-W68 product performs two different roles during meiosis and presumably during recombination; if it were the same role at two different times then the mei-W68^L1 lesion should be equivalently defective in both and it is not. (It is formally possible that the two roles differ only in level of efficiency required, but even if so the levels of efficiency are so different that they are functionally equivalent to different roles.) Two times of action of the mouse SPO11 homolog have also been suggested by ROMANIENKO and CAMERINI-OTERO (2000); see below.

It has been hypothesized above that, in wild type, a short noodle is normally converted with high efficiency into an early RN and that this process requires wild-type mei-W68 function. Can we deduce anything equivalent about late RNs? Only that, whatever pre-late RN structures might look like, they do not appear to be normal early RNs—because mei-W68^L1, which does present late RNs, does not present normal early ones. Although other possibilities such as very ephemeral early RNs in mei-W68^L1 cannot be excluded, this observation does strongly suggest, at least in flies, that early RNs and late RNs are independent structures and, therefore, that recombination events whose outcome will be simple gene conversion vs. those whose outcome will be exchange are separate events from the outset rather than alternative outcomes of one common set of events—as has been hypothesized previously (CARPENTER 1987 ) and is also suggested by recent results of MOENS et al. 2001 .

On the other hand, the possibility that a noodle can be converted directly into a late RN cannot be excluded. The timing of the few short noodles detected in wild type suggests that this is not the normal progression, but mei-W68^L1 is a mutant condition in which abnormal progressions might occur. What can be excluded is the possibility that the disappearance of noodles along mei-W68^L1 arms that have a late RN between homologs is due solely to their material being converted into that late RN, because noodles disappear from both sides of the SC at that time. Those on the other side have to be disassociated/removed, so it is probable that those on the same side are, too—and that the late RN forms de novo. Even less can be said about precursors of those late RNs that are found between sisters: Noodles would not be detectable there because they look too much like chromatin. Early RNs should be detectable but none were seen.

The remarkable observation that arms in mei-W68^L1 that have a late RN between homologs lack noodles, whereas arms in the same nuclei that lack homologous late RNs have, on average, as many noodles per arm as nuclei of the same age in mei-W68¹, requires some sort of recognition and signal that reaches along the entire arm. On the one hand, the long-standing observation of positive interference between exchanges on the same arm in Drosophila implies such a long-distance signal; on the other hand, the disappearance of noodles is a rather dramatic illustration of the principle. Moreover, the fact that noodles did not disappear on the two arms that have a late RN in the wrong place—between sister chromatids rather than between homologs—further emphasizes the precision required for that signal. Note that this signal, and its source, in mei-W68^L1 is presumed to occur in wild type, too (rather than as some side effect of mei-W68^L1's mutant state); and the fact that even the longest noodles, however much they deviate from anything yet observed in wild type, do respond to that signal in what is presumably the normal manner (i.e., by going away) strongly suggests that they are basically not that different from structures that are present in wild type, however hard to see/ephemeral those structures may be.

In wild type and other meiotic mutants, early RNs also disappear around the time that late ones appear, but this appears to be due to a nuclear rather than a brachial signal. On the one hand, nuclei that have both early and late RNs do have examples of both on the same arm, although this could reflect a brief developmental age that simply is not represented in the limited mei-W68^L1 sample. On the other hand, later pachytene nuclei lack early RNs entirely even on arms that do not have a late RN, although if late RNs are asynchronous in their presence some of those arms in wild type might represent cases that had had a late RN earlier. This implies that early RN disappearance is regulated at the nuclear level. This difference in disappearance control suggests that long noodles are not just oddly shaped early RNs but rather a qualitatively different structure.

A speculative model:
A hypothesis that ties everything together, which is surely wrong on some details but may provide testable predictions, builds on the observation that the gene conversion footprint in flies is longer in events without an exchange than in those associated with exchange (CURTIS et al. 1989 ; CURTIS and BENDER 1991 ) and the previous hypothesis that there are two rounds of initiation of recombination events (CARPENTER 1987 ). The first round, which in most organisms occurs early in zygotene but in flies occurs after full pachytene has been accomplished, involves early RNs and is hypothesized to resolve as simple gene conversion events only; moreover, it is hypothesized that these RNs and events are part of the homology-recognition system that is operating during synaptic initiation. The second round, which occurs during full pachytene in most organisms, involves late RNs and exchanges only.

Fly homologous chromosomes go into meiosis already aligned by the ubiquitous somatic pairing p