DNA Molecules Recombine by Breaking and Rejoining

Genetic recombination was first defined by studies of Drosophila, on the basis of the observation that genes on different copies of homologous chromosomes can reassort during meiosis. With the subsequent discovery that genes consist of DNA, two alternative models to explain recombination at the molecular level were considered (Figure 5.28). The “copy choice” model proposed that the recombinant molecule is generated during DNA synthesis, as a result of copying first one parental DNA and then switching to copy a different template. The alternative proposal was that recombination results from the breakage and rejoining of two parental DNA molecules rather than by synthesis of new DNA.

Figure 5.28

Models of recombination. In copy choice, recombination occurs during the synthesis of daughter DNA molecules. DNA replication starts with one parental DNA template and then switches to a second parental molecule, resulting in the synthesis of recombinant (more...)

These alternatives were first distinguished in 1961 by studies of recombination between the genomes of bacterial viruses (Figure 5.29). Infection of E. coli with viruses carrying different genetic markers was known to yield recombinant progeny. To determine if this recombination involved breakage and rejoining of the parental DNAs, one of the parental viruses was grown in medium containing the heavy isotopes of carbon (¹³C) and nitrogen (¹⁵N), while the other was grown in medium containing the normal light isotopes (¹²C and ¹⁴N). The result was parental viruses having different densities, so they could be separated by equilibrium density centrifugation in a CsCl gradient. E. coli were then infected with these differentially labeled parental viruses under conditions in which replication was inhibited, and the progeny viruses produced were analyzed for both their density and their genetic characteristics. The important result was that genetic recombinant viruses were obtained that had intermediate densities, indicating that they had acquired DNA from both parents, as predicted by the breakage-and-rejoining, but not the copy choice, model.

Figure 5.29

Experimental demonstration of recombination by breakage and rejoining. Genetically distinct parental viruses were grown in medium containing either light or heavy isotopes of carbon (¹²C or ¹³C) and nitrogen (¹⁴N or ¹⁵N) to density-label their DNAs. (more...)

Models of Homologous Recombination

The finding that recombination occurs by breakage and rejoining raises a critical question: How can two parental DNA molecules be broken at precisely the same point, so that they can rejoin without mutations resulting from the gain or loss of nucleotides at the break point? During recombination between homologous DNA molecules (general homologous recombination), this alignment is provided, not surprisingly, by base pairing between complementary DNA strands (Figure 5.30). Overlapping single strands are exchanged between homologous DNA molecules, leading to the formation of a heteroduplex region, in which the two strands of the recombinant double helix are derived from different parents. If the heteroduplex region contains a genetic difference, the result is a single progeny DNA molecule that contains two genetic markers. In some cases, mispaired bases in a heteroduplex may be recognized and corrected by mismatch repair systems, as discussed in preceding sections of this chapter. Genetic evidence for the formation and repair of such heteroduplex regions, obtained in studies of recombination in fungi as well as in bacteria, led to the development of a molecular model for recombination in 1964. This model, known as the Holliday model (after Robin Holliday), has continued to provide the basis for current thinking about recombination mechanisms, although it has been modified as new data have been obtained.

Figure 5.30

Homologous recombination by complementary base pairing. Parental DNAs are broken at staggered sites, and overlapping single-stranded regions are exchanged via base pairing with homologous sequences. The result is a heteroduplex region, in which the two (more...)

The original version of the Holliday model proposed that recombination is initiated by the introduction of nicks at the same position on the two parental DNA molecules (Figure 5.31). The nicked DNA strands partially unwind, and each invades the other molecule by pairing with the complementary unbroken strand. Ligation of the broken strands then produces a crossed-strand intermediate, known as a Holliday junction, that is the central intermediate in recombination. The direct demonstration of Holliday junctions by electron microscopy has provided clear support for this model of recombination (Figure 5.32).

Figure 5.31

The Holliday model for homologous recombination. Single-strand nicks are introduced at the same position on both parental molecules. The nicked strands then exchange by complementary base pairing, and ligation produces a crossed-strand intermediate called (more...)

Figure 5.32

Identification of Holliday junctions by electron microscopy. Electron micrograph of a Holliday junction that was detected during recombination of plasmid DNAs in E. coli. An interpretive drawing of the structure is shown below. The molecule illustrates (more...)

Once a Holliday junction is formed, it can be resolved by cutting and rejoining of the crossed strands to yield recombinant molecules (Figure 5.33). This can occur in two different ways, depending on the orientation of the Holliday junction, which can readily form two different isomers. In the isomer resulting from the initial strand exchange, the crossed strands are those that were nicked at the start of the recombination process. However, simple rotation of this structure yields a different isomer in which the unbroken parental strands are crossed. Resolution of these different isomers has distinct genetic consequences. In the first case, the progeny molecules have heteroduplex regions but are nonrecombinant for DNA that flanks these regions. If isomerization occurs, however, cutting and rejoining of the crossed strands results in progeny molecules that are recombinant for DNA that flanks the heteroduplex regions. The structure of the Holliday junction thus provides the possibility of generating both recombinant and nonrecombinant heteroduplexes, consistent with the genetic data upon which the Holliday model was based.

Figure 5.33

Isomerization and resolution of Holliday junctions. Holliday junctions are resolved by cutting and rejoining of the crossed strands. If the Holliday junction formed by the initial strand exchange is resolved, the resulting progeny are heteroduplexes but (more...)

One modification of the Holliday model, proposed in 1975, eliminates a potential difficulty with the initial proposal—namely, how can both parental molecules be nicked simultaneously at the same position to initiate recombination? In this modified version, recombination is initiated by a nick in only one of the parental molecules (Figure 5.34). The nicked strand is then displaced, and the resulting single strand invades the other parental molecule by homologous base pairing. This process produces a displaced loop of DNA, which can then be cleaved and joined to the other parental molecule. The result is a crossed-strand Holliday junction, which can be resolved into recombinant or nonrecombinant heteroduplex molecules as already described.

Figure 5.34

Initiation of recombination by nicking only one parental molecule. The nicked DNA strand invades the other parental molecule by homologous base pairing, thereby displacing a single-stranded loop of DNA. This loop is then cleaved and pairs with the first (more...)

Still another modification of the Holliday model suggests that recombination is initiated by a double-strand break, rather than a single-strand nick. Exonucleases at the site of the break then generate single-stranded tails, which can invade a homologous double-stranded molecule. Again the result is a Holliday junction, which can be resolved to yield either recombinant or nonrecombinant heteroduplexes. This double-strand break repair model appears to be particularly applicable to meiotic recombination in yeasts.

Multiple alternatives may thus account for the initial stages of recombination between two DNA molecules, and the details of recombination mechanisms, particularly in eukaryotic cells, have not been fully elucidated. But the crossed-strand Holliday junction, generated by strand exchange leading to the formation of a heteroduplex region, remains the central intermediate in consideration of the recombination process.

Enzymes Involved in Homologous Recombination

Most of the enzymes currently known to be involved in recombination have been identified by analysis of recombination-defective mutants of E. coli. Such genetic analysis has established that recombination requires specific enzymes, in addition to proteins (such as DNA polymerase, ligase, and single-stranded DNA-binding proteins) that function in multiple aspects of DNA metabolism. The identification of genes required for efficient recombination in E. coli led to the isolation of their encoded proteins, which have been characterized by biochemical analysis in cell-free systems. These studies have elucidated the action of several enzymes in catalyzing the formation and resolution of Holliday junctions.

The central protein involved in homologous recombination is RecA, which promotes the exchange of strands between homologous DNAs that causes heteroduplexes to form (Figure 5.35). The action of RecA can be considered in three stages. First, the RecA protein binds to single-stranded DNA, coating the DNA to form a protein-DNA filament. Because RecA has two DNA binding sites, the RecA protein bound to single-stranded DNA is able to bind a second, double-stranded DNA molecule, forming a complex between the two DNAs. This nonspecific RecA-mediated association is followed by specific base pairing between the single-stranded DNA and its complement. The RecA protein then catalyzes strand exchange, with the single strand originally coated with RecA displacing its homologous strand to form a heteroduplex. Thus, the RecA protein is capable of catalyzing, by itself, the strand exchange reactions that are central to the formation of Holliday junctions.

Figure 5.35

Function of the RecA protein. RecA initially binds to single-stranded DNA to form a protein-DNA filament. The RecA protein that coats the single-stranded DNA then binds to a second, double-stranded DNA molecule to form a non-base-paired complex. Complementary (more...)

In yeast, a RecA-related protein, designated RAD51, is required for genetic recombination as well as for the repair of double-strand breaks. RAD51 is not only structurally similar to RecA; like RecA, it is also able to catalyze strand exchange reactions in vitro. Proteins related to RAD51 have been identified in complex eukaryotes, including humans, indicating that proteins related to RecA play key roles in homologous recombination in both prokaryotic and eukaryotic cells.

Most recombination events in E. coli also require the RecBCD enzyme, which is a complex of three proteins (RecB, C, and D). The properties of RecBCD are consistent with the hypothesis that it initiates recombination by providing the single-stranded DNA to which RecA binds. RecBCD accomplishes this task by unwinding and nicking double-stranded DNA (Figure 5.36). The RecBCD complex binds to the end of a DNA molecule and then acts as a helicase, transiently unwinding the DNA as it travels along the molecule. When it encounters a specific nucleotide sequence (GCTGGTGG, called a chi site), RecBCD acts as a nuclease to introduce a single-strand nick. It then continues to unwind the double helix, forming a displaced single strand to which RecA can bind to initiate strand exchange.

Figure 5.36

Initiation of recombination by RecBCD. The E. coli RecBCD complex binds to the end of a DNA molecule and unwinds the DNA as it travels along the molecule. When it encounters a specific sequence (called a chi site), it nicks the DNA strand. Continued unwinding (more...)

Once a Holliday junction is formed, three other E. coli proteins (RuvA, B, and C) become involved in recombination (Figure 5.37). RuvA and RuvB act as a complex to drive the migration of the site at which the strands cross in the Holliday junction, thereby varying the extent of the heteroduplex region and the position at which the crossed strands will be cut and rejoined. RuvC then resolves Holliday junctions by cleaving the crossed DNA strands. Rejoining of the cleaved strands by ligation completes the process, yielding two recombinant molecules.

Figure 5.37

Branch migration and resolution of Holliday junctions. Two E. coli proteins (RuvA and RuvB) together catalyze the movement of the crossed-strand site in Holliday junctions (branch migration). RuvC resolves the Holliday junctions by cleaving the crossed (more...)

In yeasts, Holliday junctions are resolved by a complex of RAD1 and RAD10, with RAD1 cleaving single-stranded DNA at the crossover junction. RAD1 and RAD10 are homologs of the mammalian XPF and ERCC1 DNA repair proteins and also cleave damaged DNA during nucleotide-excision repair (see Table 5.1). RAD1 and RAD10 homologs thus may have a conserved function in recombination in eukaryotic cells.