Direct and indirect sources of DSBs. (A) A DSB can be formed directly by, for example, ionizing radiation. This results in the production of two free DNA ends. (B) If a replication fork encounters a nick in the template strand, an arm will detach to generate a single DSB. This is commonly referred to as replication fork collapse. (C) Following replication fork stalling due to replisome defects, or because there is damage or a block in the template strands, several fates are possible (see references 127, 202, and 205). Replication may restart de novo downstream of the damage without repair. Alternatively, the replication fork may undergo reversal to form a Holliday junction. In this case, the free end of the Holliday junction can be degraded by RecBCD to reform a fork and allow replisome reassembly. Replication can also continue after DNA repair by resorbing the Holliday junction to reform a fork. Otherwise, the Holliday junction may be cleaved to produce a single-ended DSB as in B. All of these DSBs are substrates for RecBCD-dependent recombinational repair.

RecBCD-dependent homologous recombination pathways. (A) In recombination-dependent DSB repair, a broken DNA molecule containing two (previously contiguous) DNA ends is rejoined in several steps involving recombination with an intact homologous donor DNA (“ends-in” recombination) (261). The pathway shown is the classical dsDNA gap repair model invoking the resolution of two Holliday junctions in the recombination intermediates formed following DNA pairing (281). In the first step, catalyzed by the RecBCD enzyme, the DNA ends are resected to form 3′ ssDNA overhang structures terminated at the Chi sequences; the DNA upstream of Chi is degraded by RecBCD, whereas the downstream Chi-containing ssDNA is preserved due to the attenuation of RecBCD nuclease activity by Chi (see reference 269). The RecA protein is loaded onto these Chi-containing ssDNA overhangs by RecBCD and promotes the homology search and DNA strand invasion with a donor duplex. The donor DNA acts as a template for DNA synthesis, resulting in the formation of two Holliday junctions. These are resolved to yield intact duplex products. (B) In recombination-dependent replication, a single DNA end is formed upon the collapse of a replication fork. The end is processed by RecBCD as in A to create a single 3′ ssDNA overhang with Chi at its terminus. RecA catalyzes the reattachment of the broken processed DNA arm to the sister DNA duplex, reforming a fork structure. Replication restart then proceeds in several steps and eventually results in the loading of the replicative helicase in a PriA-dependent process and resumption of replication in an origin-independent manner (127). (C) Integration of linear dsDNA (e.g., in conjugation and transduction) occurs by an “ends-out” mechanism (261). Both ends of the linear duplex DNA are resected by RecBCD to form 3′ ssDNA overhangs that end at Chi sequences as in A. RecA-promoted DNA strand invasion of each processed end into homologous donor DNA primes PriA-dependent DNA replication in a bidirectional manner. The replication of the entire chromosome results in the integration of the linear dsDNA piece at a homologous locus, as shown; alternatively, the resolution of the Holliday junctions can also result in integration (not shown). Note that in all three panels, for clarity, the location of the Chi sequence is shown only during the initial stages of each pathway.

The RecBCD enzyme is a sequence-regulated DNA helicase-nuclease. Shown are details of the RecBCD-catalyzed DNA end-processing reaction. (1) RecBCD (light blue) binds tightly to a blunt (or nearly blunt) DNA end of a linear DNA duplex. (2) RecBCD couples the hydrolysis of ATP to DNA translocation and unwinding (i.e., helicase activity). The ssDNA products are cleaved asymmetrically, with the degradation of the 3′-terminated ssDNA tail being much more vigorous than the degradation of the complementary tail. (3) The enzyme continues to translocate until it pauses at a correctly oriented Chi sequence. The recognition of the Chi sequence dramatically alters the biochemical properties of the enzyme (indicated by color change to pink). (4) The enzyme continues to translocate, but the nuclease polarity is switched; the degradation of the 3′ ssDNA tail is attenuated, whereas the hydrolysis of the 5′ ssDNA tail is upregulated. After Chi recognition, RecBCD facilitates the loading of the RecA protein onto the 3′ ssDNA tail. (5) RecBCD repeatedly deposits RecA protomers, which act as nucleation points for filament growth primarily in the 5′→3′ direction. (6) The RecBCD enzyme dissociates from the DNA. The product of the enzyme is a recombinogenic nucleoprotein complex of the RecA protein bound to the 3′ ssDNA tail with Chi at its terminus. This product is able to invade homologous duplex DNA to promote the recombinational repair of a DSB or to restart DNA replication as appropriate. Note that this cartoon is not to scale; several thousand base pairs of DNA may be processed before and after the Chi sequence, and the RecA filament consists of many more protomers than are shown.

Primary structure of the RecBCD enzyme. The RecB protein is modular. The N-terminal domain (red) contains seven motifs (numbered) characteristic of SF1 helicases. The C-terminal domain (magenta) contains motifs characteristic of a diverse family of nucleases. “Nuc” marks the position of nuclease motif 3, which contains key catalytic aspartate and lysine residues (D1067, D1080, and K1082) (308, 325). The RecC protein (blue) does not show significant homology to other known proteins. A region of RecC implicated in Chi recognition is indicated (see the text). The RecD protein (green) contains the seven conserved SF1 helicase motifs (numbered). The total number of amino acids (aa) in each polypeptide is indicated in parentheses. This color scheme is used in figures of the RecBCD-DNA crystal structure.

Effect of ATP and Mg²⁺ ions on the cleavage pattern observed upon DNA processing by RecBCD. During DNA translocation and unwinding, RecBCD cleaves both nascent single strands using a single nuclease active site. The frequency of cleavage on each strand (i.e., the average distance between cleavage points) is dependent on the voracity of the nuclease domain (which increases with increasing free Mg²⁺ ion concentrations) and the translocation rate (which increases with increasing ATP:Mg²⁺ concentrations) as indicated. For example, a fast-moving enzyme with low nuclease activity will produce mostly full-length ssDNA (top DNA molecule shown), whereas a slow-moving enzyme with high nuclease activity will produce very closely spaced cleavage points in the unwound DNA (last DNA molecule shown). The relative accessibility of each strand to the nuclease domain is also an important factor, and this is controlled by Chi recognition. Consequently, RecBCD can produce a spectrum of cleavage products dependent upon the precise in vitro solution conditions. Under physiological conditions (1 to 3 mM ATP [39] and 1 to 2 mM free magnesium ion [2]), the processing of Chi-containing DNA by RecBCD results largely in products similar to those depicted in the third and fourth DNA molecules illustrated.

Unwinding polarity of SF1 DNA helicases. (A) Unwinding of DNA by a 5′→3′ DNA helicase. The enzyme moves in a 5′→3′ direction on ssDNA. To unwind duplex DNA, the helicase first binds to a 5′ ssDNA tail neighboring the duplex and then translocates along that strand and into the duplex DNA. (B) Likewise, a 3′→5′ DNA helicase can bind to a 3′ ssDNA overhang and translocate unidirectionally toward the 5′ end and into the duplex portion of the substrate. (C) RecBCD is a bipolar DNA helicase. It contains two DNA helicase subunits of opposite polarity and can initiate unwinding from a blunt duplex end. In the initiation complex, the 3′→5′ subunit (red) binds to the 3′ side of the DNA end, and the 5′→3′ helicase (green) binds to the 5′ side. Although the translocation polarities of the two DNA motors are opposite, they move in the same overall direction on the antiparallel DNA duplex. If the two DNA motors move at unequal speeds, then the faster motor will be associated with a longer ssDNA tail, and the slower motor will be associated with an ssDNA loop and a shorter ssDNA tail. This unwinding intermediate (referred to as a “loop-2-tails” structure) has been observed by electron microscopy (see the text for details).

Single-molecule analysis of the RecBCD enzyme. (A) Direct visualization of DNA translocation using fluorescently labeled DNA. A single lambda phage DNA molecule is attached to a polystyrene bead, and YOYO-1 (a fluorescent DNA binding dye) is then bound. The bead is held in an optical trap, with the DNA molecule stretched out behind it by solution flow. A single molecule of the RecBCD enzyme is able to bind to the free DNA end. Upon the addition of ATP, the RecBCD enzyme unwinds and degrades the duplex. This is observed as a progressive shortening of the fluorescent DNA in an epifluorescence microscope. (A movie of this experiment can be viewed at http://microbiology.ucdavis.edu/sklab/kowalczykowskilab.htm.) The cartoon is not to scale; the stretched lambda DNA is 48.5 kbp (∼15 to 16 μM) long and binds several thousand molecules of YOYO-1 dye. (B) Direct monitoring of RecBCD translocation by tracking of a fluorescent nanoparticle (40-nm diameter). The instrument setup is as described above for A except that the DNA is not labeled. Instead of monitoring DNA degradation, the fluorescent nanoparticle is attached to the translocating enzyme via a biotin moiety on the RecD subunit. The position of the nanoparticle relative to the position of the optical laser trap measures DNA translocation. (C) RecBCD enzyme translocation on DNA monitored by tethered-particle light microscopy. The biotinylated RecBCD enzyme is bound to a streptavidin-coated bead. The RecBCD enzyme then binds to the free end of a surface-attached DNA molecule (∼1.4 kbp), effectively tethering the bead to the surface with the DNA as a linker. The translocation of RecBCD along the DNA pulls the bead toward the surface, which places an increasing constraint on the bead's Brownian motion that can be monitored by light microscopy. (D) RecBCD enzyme translocation on DNA monitored by high-resolution optical trapping. A biotinylated RecBCD enzyme is attached to a streptavidin-coated surface (alternatively, but not shown, a second optical trap can be used). RecBCD, attached to a bead, can capture the free end of a 7-kbp DNA molecule. The bead is held in an optical trap with a force clamp. Upon DNA translocation, the RecBCD enzyme pulls on the DNA and generates force against the trapped bead. A feedback mechanism moves the surface stage toward the trapped bead to maintain a constant force. The instrument can assess the effects of applied force against translocation and measure the movement of the enzyme with nanometer resolution.

Crystal structure of a RecBCD-DNA initiation complex. (A) The RecBCD-DNA complex. Alpha-helices are shown as cylinders, and beta-sheets are shown as arrows. The color scheme is the same as that in the primary structure diagram (Fig. 4). The helicase domain of RecB (RecB^Hel) is shown in red, and the nuclease domain (RecB^Nuc) is in magenta. RecC is in blue, RecD is in green, and the hairpin DNA substrate is in orange. The hairpin end of the DNA is not visible, as it is disordered in the structure, but would occupy the space to the left of the DNA end as shown here. (B) Cutaway view of the RecBCD-DNA complex showing the tunnels running through the complex. The color scheme is described above. The protein is shown as a space-filling model, which has been cut back (dark surfaces) to reveal the ssDNA tails in the tunnel entrances. The DNA substrate has not been cut away and is shown as a partially transparent space-filling model. Architectural features on the path of the ssDNA strands as they are pulled through the complex are highlighted (see the text for details).

Crystal structure of the RecB helicase domain. (A) Cartoon representation of the RecB helicase domain showing the four subdomain structures characteristic of SF1 DNA helicases. The nuclease domain, which is attached via a long linker, is partially shown in magenta. The DNA substrate is shown in deep purple. (B) RecB helicase with the location of the seven conserved helicase “signature” motifs shown. (C) Surface representation colored as described above for A. Note the cleft at the interface of subdomains 1A and 2A, which forms the ATP-binding pocket. Also particularly evident is the 70-amino-acid linker that connects the helical domain to the nuclease (Nuc) domain (the linker is colored rose and runs upward from left to right). (D) Surface representation colored as described above for B. Note that the helicase motifs line the ATP binding cleft and extend to the ssDNA binding site that is formed across the top surface of subdomains 1A and 2A.

The RecB arm. Shown is a close-up view of the “arm” structure (red) formed by an auxiliary subdomain (subdomain 1B) of RecB. The arm contacts duplex DNA ahead of the translocating enzyme. A conserved patch of residues in close proximity to the minor groove of the DNA substrate (black semitransparent model) is shown in yellow. The RecC protein is in the foreground (blue). The 5′ ssDNA tail is pointing toward the viewer through a hole in the RecC protein.

The active site of the RecB nuclease domain. The active site of the RecB nuclease domain is marked out by the coordination of a Ca²⁺ ion (yellow) with conserved nuclease motifs. The three motifs characteristic of a diverse collection of nuclease enzymes are numbered (20). Motif 1 (orange) contains a glutamate residue (E1020) that may be involved in the coordination of a second divalent cation. Motifs 2 (blue) and 3 (red) contain aspartate residues (D1067 and D1080) that coordinate the bound Ca²⁺ ion. A motif specific to the “RecB-like nuclease” family is shown in green and contains highly conserved tyrosine and glutamine residues (Q1110 and Y1114). A histidine residue (magenta) that is completely conserved, but does not form part of a defined nuclease motif, also coordinates the bound Ca²⁺ ion (H956).

Locations of tunnels in the RecC protein for the nascent ssDNA strands. (A) Cartoon representation of the RecC protein revealing its subdomain structure. Four subdomains (green, red, yellow, and blue) are characteristic of SF1 DNA helicases, although the RecC protein is not active as a DNA helicase and does not possess the conserved motifs. The C-terminal domain (magenta) (“Nuc”) has a nuclease-like fold, but note that the RecC protein is not a nuclease and does not possess the conserved motifs expected for a nuclease. The 5′ ssDNA tail of the DNA substrate (black) is pointing toward the viewer and is in a tunnel in the RecC protein. The 3′ ssDNA tail runs underneath RecC (where it is in contact with the RecB helicase domain) (not shown) and toward a second small tunnel in RecC. The duplex DNA points away from the viewer below the 5′ ssDNA strand and the RecC protein. (B) Space-filling model of the RecC protein, colored as described above for A. This illustration clearly shows the two small tunnels for the nascent ssDNA strands and the large central hole that accommodates RecB auxiliary subdomain 2B. In particular, note that the 3′ tunnel is formed at the top surface of the core helicase-like domains (green and red) and that mutations (recC1004 allele) associated with altered Chi recognition specificity (black) map to this region. The 5′ tunnel is formed by the nuclease-like domain. (C) View equivalent to those of A and B with the RecC protein shown as a space-filling representation completely in blue. Cartoon representations of the RecB and RecD proteins have been added. The color scheme is described in the legend of Fig. 4. Note that the RecD protein and the RecB nuclease are positioned above the 5′ ssDNA and 3′ ssDNA tunnels, respectively. RecB subdomain 2B occupies the large hole in the middle of the RecC protein.

The methionine pin of the RecC protein. A methionine dyad (yellow) in RecC (blue) forms a pin structure at the junction of ssDNA and dsDNA (black). The translocation of the two single strands of DNA by the RecB and RecD helicase motors will split the duplex over the methionine pin. The free 3′ end of the DNA substrate is indicated. The 5′ ssDNA tail disappears into a tunnel in the RecC protein.

Locations of subdomains and motifs in the RecD protein. (A) Cartoon representation of the three-subdomain structure of RecD. The small alpha-helical N-terminal domain 1 (wheat) is responsible for contacts with the RecC protein. Subdomains 2 and 3 (green and red) form a tandem RecA fold and are equivalent to subdomains 1A and 2A of SF1 DNA helicases. The 5′ ssDNA tail (deep purple) is shown approaching these core helicase domains. A portion of the protein comprising residues 466 to 523 is not resolved in the structure. (B) Cartoon representation with the seven conserved helicase motifs indicated. (C) Surface representation colored as described above for A. (D) Surface representation colored as described above for B. Notice the location of the helicase motifs at the interface and top surface of the core helicase domains. Their general location is similar to that seen in the RecB protein. Two regions are shown for motif 1a. The lower patch (*) represents the motif as defined previously (223). However, these residues are poorly conserved in RecD, do not resemble a canonical motif 1a, and do not occupy a similar position to the motif 1a of other SF1 helicases. The upper patch of residues (L201 to A208) represents a strongly conserved region that occupies a position similar to that of motif 1a of other SF1 helicases.

Model for RecBCD enzyme mechanism (see the text for details). (A) Cartoon representation of the initiation complex as seen in Fig. 8. The three subunits are color coded as described above, with important functional regions in the structure also labeled. (B) Pre-Chi recognition. RecBCD travels along duplex DNA powered by ATP hydrolysis in the RecB and RecD helicase motors. RecD can be the faster motor, and consequently, a loop of ssDNA may form ahead of the slower RecB motor. The RecB nuclease domain is most favorably positioned to cleave the 3′ ssDNA tail, and this strand is therefore hydrolyzed much more vigorously than the 5′ ssDNA tail. Between the RecB motor and the nuclease domain, the 3′ ssDNA tail is scanned for the Chi sequence as it passes through a tunnel in the RecC protein. (C) The Chi sequence is recognized and remains tightly bound in a tunnel in the RecC protein. The complex pauses before translocation resumes at a reduced rate. Exit of the 3′ ssDNA tail is prevented, and so a final cleavage event takes place on this strand just upstream of the Chi sequence. The 5′ ssDNA strand continues to exit from the rear of the enzyme, where, in the absence of competition from the complementary strand, it is now cleaved more readily by the nuclease domain. A loop of ssDNA accumulates downstream of the Chi sequence. The RecB nuclease domain undocks, recruits RecA proteins from solution, and loads them onto the growing ssDNA loop to promote RecA nucleoprotein filament formation. This process is proposed to be triggered by the release of the RecB nuclease domain from RecC following Chi recognition.

Mutants of the RecBCD enzyme. The crystal structure of RecBCD is shown, with the main chains of RecB, RecC, and RecD shown in red/magenta, blue/cyan/grayish blue, and green, respectively. The DNA is shown in orange. This color coding is essentially the same as that described in the legend of Fig. 8, as is the orientation of the structure. Note that the RecC backbone is shown in three different shades of blue. The light blue/gray section represents residues 981 to 1084; deletion from the C terminus to within this region yields the “double-dagger” phenotype. The cyan section represents residues 790 to 922; further deletion from the C terminus to within this region yields the “dagger” phenotype. The remainder of RecC is dark blue. The locations of other individual amino acids that are mutated in several well-studied recBCD alleles are indicated by labeled surface representations of the wild-type residues shown in black. The “RecC1004” region (shown in orange) defines residues in the partial frameshift mutation that resulted in an altered specificity of Chi recognition. The “RecB motif 6” region includes R794, R800, Y803, V804, and T807. See the text for details of the phenotypes and/or in vitro enzyme properties associated with each mutation.

Primary structure of the AddAB enzyme. The AddA protein (yellow) is structurally homologous to RecB; the N-terminal domain contains seven motifs (numbered) characteristic of SF1 helicases, and the C-terminal domain contains nuclease motifs. The position of a putative catalytic aspartate residue is marked “Nuc.” The AddB protein (magenta) shows weak homology to UvrD-like SF1 DNA helicases (which include RecB and AddA) and to the RecC protein in the regions indicated. A Walker A motif involved in the binding and hydrolysis of NTP, and equivalent to helicase motif 1, is found at the extreme N terminus. Nuclease motifs equivalent to those found in RecB and AddA are present toward the C terminus. The total number of amino acids (aa) in each polypeptide is indicated in parentheses.