Courtesy of Douglas K. Bishop and Denise Zickler, © Elsevier Science

The double-strand-break repair model (A) posits that during meiotic prophase I, crossovers (COs) and noncrossovers (NCOs) begin with a double-strand break (DSB) of a DNA helix. Cleavage of a structure known as the Holliday junction (HJ) ultimately generates both COs and NCOs. A newer model (B) proposes that COs still arise from HJs (right) but that NCOs come from a pathway called synthesis-dependent strand annealing (left) that does not involve HJs. (Reprinted from Cell, 117:9–15, 2004.)

Thousands of geneticists owe their livelihoods to meiotic recombination, that diversity-promoting process during egg and sperm development when homologous chromosomes from Mom and Dad swap pieces. But recombination is so dauntingly complex that only a few geneticists have dared to propose models describing its underlying mechanics. For decades, cytology and genetic analysis provided data to test these...


During early meiosis, four DNA double helices (two homologous chromosomes, each of which has already replicated) somehow come together. At each recombination site, two helices – one from Mom, one from Dad – then engage in a crossover (CO) or noncrossover (NCO) event. COs result in strands in which each end comes from a different parent. NCOs yield strands whose ends both derive from one parent but that can contain genetic material from the other parent in the middle; this material alerts geneticists that an NCO has occurred.

Introductory biology courses often illustrate recombination by an X-shaped diagram resembling a highway intersection. This picture represents an intertwining of homologous chromosomes that's known as the Holliday junction (HJ). Geneticist Robin Holliday proposed the structure in 1964, and Kleckner's lab confirmed its physical reality three decades later. The influential double-strand-break repair model, put forth in 1983, holds that recombination begins with cleavage of a DNA helix and proceeds to an HJ intermediate. Resolvase, an enzyme not yet identified in higher organisms, is then thought to cut the junction in one way to produce a CO and in another way to make an NCO.

Three years ago, Michael Lichten, a National Cancer Institute principal investigator, and his postdoc at the time, Thorsten Allers, published a study suggesting that most NCOs are generated earlier than COs and that these NCOs do not arise from HJs. The scientists conjectured that NCOs might instead be the culmination of a process already demonstrated in mitotic cells and known as synthesis-dependent strand annealing. In SDSA, one DNA strand essentially probes its partner and then pulls back.

The Lichten/Allers proposal relied on earlier work by Kleckner and on fresh experiments examining the budding yeast Saccharomyces cerevisiae. Lichten has not been able to replicate one of his findings, but he says that results involving the transcription factor Ndt80 have held up. Yeast with mutant ndt80show unresolved HJs and very few COs, while NCOs are unaffected. A new paper from Kleckner, her postdoc G. Valentin Börner, and her former postdoc Neil Hunter (now at the University of California, Davis) extends these observations to mutations in five meiosis-specific genes in yeast.1

Lichten acknowledges that some researchers remain skeptical about the idea that COs and NCOs follow different pathways. "It's still pretty new," he observes. "In this business, things don't move that fast." But he points out that the SDSA theory predicts particular NCO end-products that were reported last year by Thomas D. Petes at the University of North Carolina, Chapel Hill.4 Still, Kleckner worries that until SDSA intermediates are isolated, "you don't know for sure" that the model is correct.

Franklin W. Stahl, an emeritus biology professor at the University of Oregon in Eugene, contends that the data supporting meiotic SDSA are an artifact of mismatch repair. He hypothesizes that a structure he calls the unligated Holliday junction is cleaved to form COs and is unwound to form NCOs. But some other investigators sharply discount his theory, and even Stahl concedes a paucity of physical evidence favoring the unligated HJ. He maintains, however, that "biochemistry demands that the unligated [form] exist, at least transiently," as an intermediate leading to the indisputably real ligated form. Ligated HJs also generate some recombinants, according to his model.


During meiosis, all chromosomes must have at least one CO to ensure their proper segregation into the gametes. Yet most recombination events are actually NCOs. In mammals, for example, the NCO/CO ratio is estimated at 10 to 1. Meiosis experts don't know why one double-strand DNA break leads to a CO and another to an NCO. But one reason for NCO predominance appears to be interference, a phenomenon in which COs tend not to occur near one another. The synaptonemal complex (SC) was long thought to mediate this phenomenon.


Courtesy of Peter B. Moens

The function of the synaptonemal complex (SC), a scaffold that develops at crossover sites during meiotic prophase I, is unknown. This electron micrograph (X 70,000) depicts an SC from a grasshopper spermatocyte The structure's lateral elements (LE), 100 Å apart, flank a ladder-like central element (CE). Transverse filaments (TF) extend between the LE and CE, and chromatin (CH) surrounds the entire structure. (Reprinted from J Cell Biol, 40:542–51, 1969.)

The SC is an enormous structure resembling a railroad track that is "a remarkable thing to look at in a microscope," Keeney says. "You can see the two axial elements lying parallel to one another, separated by about 100 nanometers. Then like the crossties on a railroad track, you can see the transverse elements." The SC, whose protein composition remains ill-defined, eventually extends along the whole length of the chromosome; DNA loops are attached to the axial elements.

Data increasingly indicate that the SC nucleates at CO sites and grows outward from them.5 But mounting evidence also suggests that the SC does not regulate CO interference. Kleckner's lab reports that meiotic chronology precludes such a role for the SC.1 And Yale University's G. Shirleen Roeder, whose work on mutant yeast during the 1990s supported an SC-interference connection, recently disclosed results implying a disconnect. Her lab found that sites of future SC development already show interference; moreover, mutant yeast cells that lack Zip1, a protein needed for SC formation, display interference at those sites.2

The SC's function consequently is more mysterious than ever. In her recent paper, Roeder speculates about the existence of multiple interference mechanisms, some of which might still depend on the SC. Kleckner's thoughts, meanwhile, track in a different direction. During recombination, she notes, the axes of homologous chromosomes that the SC's twisting is "the motion that you require in order to exchange axes."

Interference, she speculates, might result from mechanical stresses that chromosomes experience when they are crammed into cellular compartments. Homologous chromosomes linked by bridges might then buckle at weak points. "That buckling will alleviate compression stress right at that point," Kleckner explains. "And that relaxation will tend to spread outward for a certain distance," preventing buckling nearby. "If bridge bucklings are crossover designations," she continues, "you won't have more crossover designations. That's basically how we think interference works."

Kleckner concedes that her ideas might be wrong and are difficult to prove. "It's much harder to study chromosomal structural components than it is to study recombination," she says. But in a testament to the potential of basic meiosis research, she adds, "Thinking about crossover interference, in fact, has led us to an entirely new view of how chromosomes could work."

Douglas Steinberg dougste@attglobal.net is a freelance writer in New York.

Interested in reading more?

Magaizne Cover

Become a Member of

Receive full access to digital editions of The Scientist, as well as TS Digest, feature stories, more than 35 years of archives, and much more!
Already a member?