The molecular machinery responsible for DNA expression, recombination, and compaction has been difficult to visualize as functionally complete entities due to their combinatorial and structural complexity. We report here the structure of the intact functional assembly responsible for regulating and executing a site-specific DNA recombination reaction. The assembly is a 240-bp Holliday junction (HJ) bound specifically by 11 protein subunits. This higher-order complex is a key intermediate in the tightly regulated pathway for the excision of bacteriophage λ viral DNA out of the E. coli host chromosome, an extensively studied paradigmatic model system for the regulated rearrangement of DNA. Our results provide a structural basis for pre-existing data describing the excisive and integrative recombination pathways, and they help explain their regulation.https://doi.org/10.7554/eLife.14313.001
Some viruses can remain dormant inside an infected cell and only become active when conditions are right to multiply and infect other cells. Bacteriophage λ is a much-studied model virus that adopts this lifecycle by inserting its genetic information into the chromosome of a bacterium called Escherichia coli. Certain signals can later trigger the viral DNA to be removed from the bacterial chromosome, often after many generations, so that it can replicate and make new copies of the virus. Specific sites on the viral and bacterial DNA earmark where the virus’s genetic information will insert and how it will be removed. Remarkably, each of these two site-specific reactions (i.e. insertion and removal) cannot be reversed once started, and their onset is precisely controlled.
These reactions involve a molecular machine or complex that consists of four enzymes that cut and reconnect the DNA strands and seven DNA-bending proteins that bring distant sites closer together. Despite decades of work by many laboratories, no one had provided a three-dimensional image of this complete molecular machine together with the DNA it acts upon.
Now, Laxmikanthan et al. reveal a three-dimensional structure of this machine with all its components by trapping and purifying the complex at the halfway point in the removal process, when the DNA forms a structure known as a “Holliday junction”. The structure was obtained using electron microscopy of complexes frozen in ice. The structure answers many of the long-standing questions about the removal and insertion reactions. For example, it shows how the DNA-bending proteins and enzymes assemble into a large complex to carry out the removal reaction, which is different from the complex that carries out the insertion reaction. It also shows that the removal and insertion reactions are each prevented from acting in the opposite direction because the two complexes have different requirements.
These new findings improve our understanding of how the insertion and removal reactions are precisely regulated. Laxmikanthan et al.’s results also serve as examples for thinking about the complicated regulatory machines that are widespread in biology.https://doi.org/10.7554/eLife.14313.002
The rearrangement of DNA, either by homologous or site-specific recombination and transposition, is a fundamental feature of evolution, genetic variation, and gene regulation; among such pathways, the integration and excision of bacteriophage λ into and out of the E. coli chromosome is one of the most thoroughly characterized. Nevertheless, even for this pathway, it has not been possible until now to visualize the entire functional structure of the recombinogenic complex, despite the wealth of genetic, biochemical, functional and structural information accumulated by many laboratories over the past 50 years.
The site-specific recombinase (Int) encoded by bacteriophage λ is the archetypical member of the tyrosine recombinase family, whose members carry out such diverse functions as chromosome segregation, chromosome copy number control, gene expression, conjugative transposition, gene dissemination, and viral integration and excision (Craig et al., 2015; Jayaram et al., 2015; Landy, 2015). All tyrosine recombinases use the same isoenergetic phosphoryl transfer chemistry and sequential strand exchange mechanism to execute DNA rearrangements via the formation and resolution of a transient four-way DNA junction, or Holliday junction (HJ), recombination intermediate (Hsu and Landy, 1984; Kitts and Nash, 1988a). Although these junctions are common intermediates in many different pathways responsible for rearranging genetic material in evolution, heredity, and gene expression, they have been particularly well-studied in the tyrosine recombinase family of reactions (Gopaul et al., 1998; Jayaram et al., 2015; Van Duyne, 2015).
In contrast to two other well-studied and highly exploited family members, Cre and FLP, where the entire bidirectional recombination reaction is executed by a recombinase tetramer (Jayaram et al., 2015; Van Duyne, 2015), Int carries out two opposing reactions that are both highly directional and tightly regulated. Each reaction requires the assembly of a unique 400 kDa multi-protein complex using accessory DNA-bending proteins and an overlapping ensemble of accessory binding sites. In response to a variety of physiological and environmental signals one complex is assembled to carry out unidirectional integrative recombination between “att” sites on the viral and bacterial chromosomes, while the other complex is formed to carry out unidirectional excisive recombination between att sites flanking the integrated viral chromosome ([Seah et al., 2014; Tong et al., 2014]; see Figures 1 and 2).
Int is a heterobivalent DNA binding protein that binds to high affinity 'arm-type' DNA sites via its small amino-terminal domain (NTD) and more weakly to 'core-type' DNA sites, where DNA cleavage and ligation take place. Int binds to core sites via a central core binding domain (CB) and a C-terminal catalytic domain (CAT); the latter two domains are referred to here as the CTD (see Figure 2; Moitoso et al., 1989). The Int subunits within a recombinogenic complex bind and bridge arm- and core-type DNA sites in patterns determined by the accessory DNA bending proteins IHF, Xis, and Fis. Differential occupancy of the 16 protein binding sites on the 240 bp att site generates two overlapping ensembles that differentiate the integrative and excisive recombination pathways ([Seah et al., 2014; Tong et al., 2014]; (see Figure 1).
Using techniques described previously (Figure 3), we trapped and purified the multi-protein HJ complex of excisive recombination and determined its structure at 11 Å resolution using single particle electron cryo-microscopy (cryo-EM). We then used the EM density, known protein-DNA crystal and NMR structures, and the known Int-mediated arm-core bridges to build an atomic model of the intact functional complex. The experimental structure for the excisive complex then allowed us to construct a model for the integrative complex. The results provide a structural basis for understanding how these complex DNA recombination machines function and how they are so tightly regulated.
Stable HJ complexes were trapped by in vitro recombination between attL and attR partners whose respective seven bp overlap sequences were designed to generate a fully paired HJ overlap sequence that would create mismatched base pairs upon reversal or progression of the recombination reaction (Matovina et al., 2010; Tong et al., 2014). Following brief crosslinking with 0.0035% glutaraldehyde, the complexes were purified by two cycles of sucrose gradient centrifugation and either frozen in liquid nitrogen or immediately prepared for cryo-EM. The complexes were analyzed for purity and homogeneity by native gel electrophoresis (Figure 3A) and negative staining electron microscopy. For details of the preparation and characterization of the complexes, see (Tong et al., 2014).
We selected 66,033 particles from 1359 images (for example, Figure 3B) and obtained two-dimensional (2D) class averages using the ISAC procedure (Yang et al., 2012) after pixel binning (effective pixel size 5.6 Å). Following visual inspection of the class averages, we selected 52 classes that showed clear features, such as DNA loops and crosses (Figure 3C), and calculated initial maps using EMAN2 (Tang et al., 2007). The highest-scoring map was used to initialize particle alignment and three-dimensional (3D) classification in FREALIGN (Lyumkis et al., 2013). All particles were aligned against the EMAN2 map using data limited to 60 Å resolution. Starting with reconstructions calculated from six randomly selected subsets of the data, particles were 3D-classified into six classes and their alignments refined in 40 iterations in which the resolution limit was increased in regular steps to a final limit of 20 Å. The reconstruction with the highest average particle score was then used to initialize a second round of alignment and 3D classification, starting again with a resolution limit of 60 Å and increasing the limit in steps to a final 20 Å. Due to the substantial heterogeneity of the sample – only 17% of the particles ended up in the final reconstruction (Figure 3D) – we performed six such refinement/classification rounds, always initializing the next round with the reconstruction of the previous round that had the highest average particle score. The highest scoring reconstructions from each round, as well as the initial and final maps are shown in Figure 4A. Lower-scoring reconstructions showed some of the features of the highest-scoring reconstruction but were missing density in the DNA loop and/or the Int tetramer regions. This suggests that the heterogeneity is the result of particle distortions and instabilities that often arise during the grid preparation procedure, for example when particles interact with the air/water interface (Cheng et al., 2015). For the final map, we performed an additional 50 iterations of refinement and 3D classification using a resolution limit of 18 Å. The last three iterations used data with an effective pixel size of 2.8 Å. The best reconstruction contained 10,956 particles and reached a resolution of about 11 Å (Figure 4B and C).
We began by placing a complex containing the core-binding and catalytic domains of an integrase tetramer bound to core site HJ DNA into the EM envelope. This core complex was derived from a crystal structure of full-length integrase bound to HJ DNA and short oligonucleotides containing arm site sequences (Biswas et al., 2005). The assembly fit well as a rigid body, indicating that the EM and crystal structures are similar in this region. Structural models for the protein-bound P and P' arms were assembled as previously described (Seah et al., 2014), except that the Fis dimer (which is not present in the purified HJ complex used here) was omitted from the P arm model. Details of the arm constructions are provided in Materials and methods. The pre-assembled P and P' arms were spliced onto the core Int-HJ complex while maintaining standard B-DNA twist and fit into the EM envelope by applying small DNA roll angles ( ± 1°) in the regions outside of the IHF, Xis, and Int-NTD footprints. Flexibility in the IHF-induced DNA bends at H' and H2 was also allowed by testing small in-plane and out of plane deviations from the IHF/H'-DNA crystal structure (Rice et al., 1996).
Two regions in the resulting structure could not be confidently modeled based on existing crystal structures. The linkers connecting the NTD and CB domains of Int adopt distinct and partially extended conformations, a situation quite different from the compact and symmetric arrangement of linkers observed in the crystal structures of Int bound to short oligonucleotide DNAs (Biswas et al., 2005). In addition, the central base-pairs of the HJ DNA were not well-defined in the Int-HJ crystal structure. To obtain low-energy conformations for these regions within the EM envelope and to regularize the stereochemistry of the DNA splice junctions and sites of bending, we performed molecular dynamics refinement using dynamic elastic network (DEN) restraints (Schröder et al., 2007). The protein•DNA sub-complexes derived from crystal structures were tightly restrained to adopt their experimental conformations, whereas the Int NTD-CB linkers and central HJ DNA base-pairs were allowed flexibility. To restrain flexible regions of the model to remain within the EM envelope, we included an energy term for agreement with the Fourier coefficients of the EM density. As shown in Figure 5, the final structure agrees well with the EM envelope. There are no major features lacking density and only one small region of density that was not fit by the model (discussed below). The P and P' arm trajectories are particularly well defined by the EM envelope (Figure 5A), as is the core Int-HJ tetramer (Figure 5B).
The limited resolution of the cryo-EM map does not permit construction of an atomic model with accurate detail at the level of side chains and nucleotides. Consequently, the model coordinates deposited for the excisive HJ complex specify only the positions of protein domains and the paths of the DNA duplex arms. Inter-domain linkers for the Int-C', Int-B, and Int-B' subunits are provided in arbitrary conformations in the final model to improve visualization of the structure and to facilitate modeling in future experiments; no attempt was made to predict the structures of these linkers.
Three views of the excisive complex structure are shown in Figure 6 and a video illustrating several views of the complex is available as Video 1. At the center of the excisive complex an integrase tetramer is bound to a core-site HJ, where strand exchange has taken place between the B half-site of attL and the C half-site of attR (see Figures 1B and 2C). We refer to the integrase subunits by the core half-sites to which they are bound. Int-C, for example, is the integrase subunit bound to the C half-site. The most striking features of the complex are the tight bends of the P and P' arms as they emerge from the C and C' core half-sites. IHF heterodimers bend the P arm at the H2 site and the P' arm at the H' site, in each case re-directing the DNA back towards the Int tetramer. For P', the IHF bend is sufficient for the P'1 and P'2 arm-binding sites to be engaged by the Int subunits bound at the C' and B core sites, respectively (Figure 6B). These are two of the three Int bridging interactions identified using biochemical and genetic approaches (Tong et al., 2014).
For the P arm, the IHF-mediated bend at H2 is phased differently with respect to the core tetramer compared to the bend at H'. Rather than being directed towards the NTDs of Int, the H2 bend causes the P arm to run alongside the Int tetramer. Three Xis subunits provide a second change in direction of the P arm, directing it across the top of the Int tetramer (Figure 6A). As a result of the IHF and Xis-induced bends, the P2 site is positioned where it can be engaged by the Int subunit bound to the B' core site (Figure 6A and C). The B'-P2 bridge is the third bivalent interaction required for excisive recombination (Figure 2C) (Tong et al., 2014). Remarkably, the DNA bend observed in the Xis•DNA crystal structure (Abbani et al., 2007) fits well into the EM envelope, which has strong density for both the DNA and the Xis subunits. The P arm trajectory is also aided by an intrinsic A-tract sequence found in the Fis binding site. Fis was not present in the complexes analyzed by EM and as expected, there is no density corresponding to Fis at this site. We note, however that Fis can be readily accommodated by the structure (Figure 6C), which fits well with the proposed role for Fis as a non-essential enhancer of recombination (Ball and Johnson, 1991a; 1991b; Esposito and Gerard, 2003; Thompson et al., 1987a).
The only component of the excisive HJ complex that is not accounted for by EM density is the NTD of Int-C (Figures 2C and 6A). This domain has no known function in excisive recombination, since only three arm-binding sites are used in the reaction (Numrych et al., 1990; Thompson et al., 1987b; Tong et al., 2014). One plausible location is near the Int-NTD bound at P2, where there is a small lobe of additional density present (Figure 5A). It is clear, however, that the Int-C NTD is not positioned directly above its connected CB domain, since the Xis molecule bound at the X1 site occupies this space. We did not fit the Int-C NTD into the P2-associated density because the domain's orientation and position could not be clearly established.
The EM structure described here provides an opportunity to re-examine the model and data recently reported in a FRET-based study of the λ excision HJ intermediate (Seah et al., 2014). In that work, 12 DNA sites were labeled with fluorophores and 28 inter-site distances were estimated using in-gel fluorescence measurements. These distances, combined with knowledge of the Int bridging interactions and crystal structures of the component sub-complexes, were used to construct a model of the excisive HJ complex. The overall architecture of our EM structure is in broad agreement with the FRET-based model. As might be expected from the small number of structural restraints available, the FRET model is largely schematic in nature. For example, the tight, parallel paths of the P and P' arms were not predicted in the FRET model, nor was the extensive Xis-Int interface that links the two arms together (see below). Indeed, the details that emerge from the EM structure provide important new insights that could not have been predicted based on distance measurements alone.
The EM structure also provides an opportunity to ask if the dye positions chosen for the FRET study were appropriate and to determine how well the EM structure agrees with the estimated distances (Table 1). Six of the locations appear to have been good choices, since they are largely free from steric interference that would compromise free rotation and/or the volumes accessible to the fluorophores. The EM structure is in good agreement with all of the distance estimates involving these sites. Two of the sites, attP- (-118) and attP- (+79), have serious steric conflicts based on the EM structure. The attP- (-118) dye on the P arm is directed into the path of the P' arm, possibly explaining why the arms diverge in the FRET model. The dye at attP- (+79) is on the P' arm and is directed into the Int-NTD bound at P2. Thus, the distances involving these sites are expected to be compromised and indeed, we see much poorer agreement with the EM structure for the distances involving these locations. The remaining four sites used for the FRET study are located on extensions of the P and P' arms that were not included in the constructs used for EM. Thus, there is excellent agreement between the EM structure and the FRET distance estimates for those dye positions that were correctly predicted to be free from steric interference.
In crystal structures of Int complexes containing simplified oligonucleotide arm-binding sites, the DNA bound by Int NTDs lies parallel to the Int tetramer and the Int NTD-CB linkers form a compact network of interactions (Biswas et al., 2005). In contrast, the P' arm in the EM structure follows a relatively straight path after the H' bend, resulting in an upwards trajectory as it passes over the Int tetramer (Figure 6B). Consequently, Int-C' and Int-B linkers must adopt extended and distinct conformations in order to bridge the P' arm and the Int CB domains. Int bridge formation at P'1 and P'2 of attL therefore does not require bending of the P' arm after the IHF-directed U-turn at H'. Instead, Int uses the inherent flexibility present in the NTD-CB linker regions to reach the P'1 and P'2 binding sites.
A primary function of Xis is to bend the P arm to stabilize a synapsis-competent attR complex. The EM structure reveals that Xis also mediates a substantial protein-protein interface between the P and P' arms. The Xis molecules bound at the X1 and X1.5 sites of the P arm are packed against the Int-B and Int-C' NTDs bound at the P'2 and P'1 sites, respectively (Figure 6C and Video 1). The Xis1.5-Int-C' interface appears to be particularly extensive, although at the resolution of the EM density, we cannot visualize the conformations or roles of individual side chains. These Xis-Int interfaces are presumably important, given the close approach and parallel trajectory of the P and P' arms in the excisive complex.
The identification of a Xis-mediated interface between the P and P' arms provides an explanation for how synapsis is facilitated in the excisive pathway. Int binds only weakly to the core sites of attL and attR and the Int-bound core sites do not form stable synaptic complexes that undergo recombination (Bushman et al., 1985). What provides the interactions that stabilize the attL x attR synaptic complex to initiate excisive recombination? Crystal structures of Int bound to core and a symmetrized mimic of arm DNA fragments suggested that the Int NTDs, along with their associated linkers, could play this role (Biswas et al., 2005). However, earlier experiments with a Cre-Int chimeric recombinase (Warren et al., 2008) and the present EM structure indicate that this is unlikely. The three Int NTDs in the excisive complex are arranged in a linear manner, where the NTD bound at P2 makes no contact with the NTDs bound at P'1 or P'2 (Figure 6C). Instead, Xis-NTD, and possibly Xis-linker interactions provide the additional glue required to hold attL and attR together and allow the initial strand exchange to take place.
There are 18 residues at the C-terminus of Xis that were missing in the Xis•DNA crystal structures (Abbani et al., 2007; Sam et al., 2004) and are not included in our excisive complex model. The C-termini of the Xis domains bound at the X1.5 and X2 sites extend into the gap between the P and P' arms (i.e., towards the viewer in Figure 6C). Six of the eighteen missing residues are arginine or lysine, suggesting that the basic C-terminal tails of Xis could interact with P and/or P' arm DNA, further stabilizing a synaptic interface. The C-terminus of Xis bound at X1 is instead directed at the Int-B NTD bound at P2, consistent with the observed cooperativity of Xis and Int binding to attR arm DNA (Bushman et al., 1985; Numrych et al., 1992; Sarkar et al., 2002). Thus, the EM structure reveals and/or explains three functional roles for Xis in excisive recombination: i) promoting Int NTD binding at the weak P2 site; ii) bending the P arm to place P2 in position for cooperative Int binding and formation of a synapsis-competent attR; iii) mediating an Xis•Int interface between the P and P' arms that facilitates attL x attR synapsis.
Although Fis is not required for excisive recombination, it does stimulate the reaction when Xis is limiting (Thompson et al., 1987a). The views in Figure 6A and C provide an explanation for this stimulation. The A-tract present in the Fis site, Xis binding at X1, X1.5, and X2, and Int binding at P2 all contribute to the formation of a properly assembled P arm, and therefore a properly assembled attR at the start of the excision pathway. Fis binding stabilizes the A-tract bend in attR (Stella et al., 2010) and facilitates Xis binding (Papagiannis et al., 2007), providing additional sources of cooperativity for P arm assembly that would enhance the activity of Xis at lower concentrations.
Structures representing reaction intermediates in the Cre and λ systems have led to a model for understanding how strand exchange is coordinated (Aihara et al., 2003; Biswas et al., 2005; Gopaul et al., 1998; Guo et al., 1997). The essential geometric features of this model for λ integrase are shown in Figure 7. The scissile phosphodiesters, which flank the 7-bp crossover sequence located at the center of the core sites, form a parallelogram in the HJ intermediate. The sites that are active for cleavage form the closer diagonal pair; those that are inactive are farther apart. Isomerization of the HJ intermediate between a top-strand cleavage configuration and a bottom-strand cleavage configuration involves exchanging angles within the scissile phosphate parallelogram (Figure 7A and B). In the Cre system, isomerization is coupled to changes in the angles between the core site duplexes that converge at the center of the junction. Changes in the interfaces between catalytic domain subunits that result from isomerization are responsible for activation/deactivation of active site pairs (Van Duyne, 2001).
The core HJ isomer that we fit in the EM structure corresponds to a bottom strand cleavage configuration, where the complex is poised to carry out strand exchange to form attP and attB products (Figure 7B). We considered how the structure of the alternative isomer, corresponding to a configuration where the top strands have just been exchanged to generate the HJ from the synapsed attL and attR partners, might differ in this complex (Figure 7A). Although there are significant changes near the center of the core complex where the HJ branch migrates by one bp, the ends of the duplexes remain essentially in the same position, with only small torsional changes (Figure 7 and Video 2). This finding is consistent with the pseudo-fourfold symmetry of the core HJ complex (Biswas et al., 2005) and the local four-fold symmetry of the EM density corresponding to this region (Figure 5B). Thus, at the resolution of our EM data, we cannot distinguish between HJ isomers and the complexes analyzed may be a mixture of the two isomeric forms.
The nearly isosteric model for HJ isomerization shown in Figure 7 has important implications for considering how the P and P' arms of the excisive complex might change during recombination. In the absence of structural models, it has been difficult to rationalize how the arms could accommodate, or perhaps even direct the changes in quaternary structure expected from a Cre-like isomerization. In contrast to Cre, however, the core HJ structure observed in the EM and crystal structures of the λ intermediate indicates that isomerization could occur with relatively little impact on the flanking DNA segments. We suggest that small changes in position and torsion of the C and C' core sites can be accommodated by the in-plane and out of plane bending modes of IHF, with little impact on the Xis•Int interface assembled between the P and P' arms as they cross the Int tetramer. In other words, the EM structure implies that isomerization of the HJ intermediate can occur in the absence of large changes in quaternary structure and therefore does not require coupling to large changes in the P and P' arm positions. This type of overall isosteric isomerization may be facilitated by the odd-numbered 7-bp crossover region that is found in essentially all tyrosine integrases, e.g. (Hakimi and Scocca, 1996; Kolot and Yagil, 1994; Peña et al., 1996; Williams, 2002). A stereochemical explanation for how a one-bp branch migration accomplishes this will be an interesting topic for future studies.
The integrative HJ intermediate is formed when a complex containing attP DNA, an Int tetramer, and three IHF heterodimers captures the attB site in the host genome and carries out top strand exchange between the core DNA sequences (Figure 1). We previously proposed a model for this complex based on core-arm site bridge data and insights derived from the FRET-based excisive complex model discussed above (Seah et al., 2014). The EM structure of the excisive HJ complex provides an opportunity to revise the integrative model in the context of a well-defined molecular architecture and correctly positioned P and P' arms for the excision complex. The new integrative HJ model, constructed using a similar procedure to that previously described, is shown in Figure 8.
In both the integrative and excisive complexes, IHF bending at the H' site directs the P' arm over the CB domains of the Int tetramer. In the integrative complex, the B'-NTD binds at P'3, since Xis is not present to position the P2 site to bind to this domain. A striking aspect of the new integrative complex model is that the P' arm trajectory requires no positional changes in order to readily accommodate binding of the C', C, and B' NTDs of Int (Figure 8A). Thus, the P' arm position and trajectory appear to be invariant in excisive and integrative recombination.
In the absence of Xis, the P arm is not directed towards the Int NTDs after the H2 bend. Instead, it follows an upward trajectory beyond the P' arm, where IHF binding at the H1 site reverses its course and directs it back towards the Int tetramer (Figure 8B and C). This P arm trajectory places the P1 site within reach of the Int B-NTD, a high-affinity interaction that is crucial for integrative recombination (Tong et al., 2014). Thus, the P arm crosses the P' arm to position Int-B during integration, but runs parallel to the P' arm and promotes Int-B' binding at P2 during excision.
The P-P' arm crossing that occurs during integration generates a negative DNA-crossing node, a feature that provides a structural explanation for the long-standing puzzle of why the topologies of integrative and excisive recombination are different and why supercoiling is required for efficient integration (Nash, 1975; Richet et al., 1988). Indeed, the integrative model shown in Figure 8 cannot be constructed based on IHF-phased bends at H2 and H1 alone; additional DNA-bending is required, which we suggest would occur naturally in a negatively supercoiled attP, but would be energetically unfavorable in linear substrates. The P-P' crossing also involves close DNA-DNA contacts between the arms, explaining in part the important role of polyamines such as spermidine and/or divalent cations in the integrative reaction (Kikuchi and Nash, 1978; Nash, 1975).
The most remarkable feature of bacteriophage λ excisive and integrative recombination is the precisely controlled unidirectionality of these two isoenergetic reactions which, at the DNA level, are the simple reverse of each other. While many features of the distinct but overlapping protein ensembles responsible for each reaction (and its regulation) have been worked out, the present study finally allows us to consolidate and incorporate these features into the structure of a fully functional reaction intermediate. As discussed above, the structure reveals an important new role for Xis, clarifies the role of the Int linker regions, defines the paths of the arms as they embrace the enzymatically active core region, and suggests that HJ isomerization occurs with only minor changes in quaternary structure. Equally important, the good fit between the present structure and the large body of genetic, biochemical, and biophysical data that exists (reviewed in [Landy, 2015]) makes it relatively straightforward to extend the present structure to informative models for the other intermediates involved in excisive and integrative recombination. Based on the EM structure of the excisive HJ complex and the model of the integrative HJ complex described here, our current views of the mechanisms of directionality and regulation in λ site-specific recombination are summarized below and illustrated schematically in Figure 9.
The biochemical feature defining excisive and integrative recombination as two different pathways (and not the forward and reverse of a single pathway) is the fact that both reactions are initiated by exchange of the same ('top') strands to form their respective HJ intermediates (Kitts and Nash, 1988b; Nunes-Düby et al., 1987). Examination of the models shows that the order of strand exchange in both pathways is determined prior to synapsis by the patterns of specific Int bridges between arm and core sites and by the IHF- and Xis-induced bends that enable those bridges. For example, the C' and B bridges to the P'1 and P'2 sites of attL require a specific bend of the B-C' core site, which in turn commits attL to top strand exchange in the synaptic complex with attR (Figure 9Ea). Similarly, formation of an integration-competent attP assembly requires a specific bend direction of the C-C' core site in order to form the correct bridging interactions (Figure 9Ia).
We found that a stereochemically plausible attP model can be assembled if the C-C' core site is bent in the opposite direction to that shown in Figure 9Ia (i.e., a bottom-strand cleavage configuration), but the resulting complex has a (+) crossing node and would therefore be strongly disfavored in a negatively supercoiled substrate. We conclude that only a top strand exchange integrative complex configuration is consistent with negative supercoiling.
The excisive complex structure delineates the mechanism by which Xis acts as a directionality switch between the excisive and integrative recombination pathways. During formation of a synapsis-competent attR complex, Xis enhances binding of the Int NTD at P2 through protein-protein interactions (Sarkar et al., 2002) and it enhances formation of the Int bridge between P2 and B’ by pronounced bending of the P arm. Xis interactions with the Ints bound to attL also likely play an important role in facilitating synapsis. Thus, the Xis switch functions at multiple levels involving both intramolecular (attR) and intermolecular (attL-attR) stabilization of intermediates in excisive recombination (Figure 9Ea and Eb).
The structural models also account for why the prophage doesn't rapidly excise upon completion of integration and why the excision reaction is not run efficiently in reverse. The attR and attL products of integration have zero and one intramolecular Int bridges, respectively, and are therefore unlikely to be stable (Figure 9Id). Similarly, the attP complex formed at the end of excision has only a single Int bridge (Figure 9Ec). The models also explain why the integrative reaction is inhibited in the presence of Xis (Nash, 1975): Xis-induced bending of the P arm (Abbani et al., 2007) prevents its upward trajectory in attP and thus defeats the proper positioning of the P1 site required for attB synapsis (Figure 9Ec vs Figure 9ia).
One of the especially interesting features suggested recently (Seah et al., 2014) and strengthened considerably by this work addresses the long-standing conundrum of how the large attP complex captures a naked attB site, whose core-type half-sites are arranged as imperfect inverted repeats (Richet et al., 1988). Because the openings of the two attP-bound Ints, positioned to bind attB, face in opposite directions they cannot dock attB by simple collision, i.e., the Int destined to bind the B core site must have the flexibility to wrap around the host chromosome from the opposite face. The integrative complex model shown in Figure 8 has an inherently flexible P arm which we propose can transiently swing out along with Int bound at P1 (Figure 9Ib). This would provide the dynamic binding required to embrace the bacterial chromosome and clasp onto a synapsed attB (Figure 9Ic), a structural feature that is consistent with genetic and biochemical results (Tong et al., 2014) and the especially high affinity of Int NTD for the P1 site (Sarkar et al., 2002). The structure and models are also consistent with the changes in topology that occur when integrative and excisive recombination reactions are carried out within the same circular DNA molecule (Crisona et al., 1999) and with the requirement for negative supercoiling for efficient integration (Richet et al., 1988).
The highly asymmetric arrangement of the NTDs in the structure and models reported here contrasts sharply with the tightly packed complexes seen in crystal structures of HJ-Int tetramers bound to truncated and symmetrized arm-type 'consensus' oligonucleotides (Biswas et al., 2005). Indeed, the asymmetry and inherent flexibility described here are critical elements in understanding the structural basis of the recombination mechanisms; they are both consistent with, and explanatory for, all of the extensive genetic, biochemical, and topological analyses of this recombination pathway. We suggest that, in addition to their significance for understanding regulated directionality in recombination, the results described here will also serve as exemplars for thinking about the complex regulatory machines so abundant in prokaryotic and eukaryotic biology.
The HJ complex was obtained through excisive recombination of attR and attL bubble substrates using the following conditions: 55 nM attR, 50 nM attL, 30 mM KCl, 60 mM NaCl, 0.005 mg/mL BSA, 25 mM HEPES at pH 7.5, 150 nM IHF, 350 nM Xis and 350 nM Int were incubated at room temperature for 2 hr. The reaction mixture was then crosslinked with 0.0035% glutaraldehyde at room temperature for 10 min followed by quenching with 0.3 M glycine. The mixture was concentrated using Amicon ultracell 50K filters (Merck Millipore, Billerica, MA) to a volume of 100 μL and loaded onto a 2 mL sucrose gradient (22%–40% sucrose, 10 mM Tris, pH 8, 1 M Betaine) at 4°C. The gradient was run at 47,000 rpm using a TLS-55 rotor (Beckman Coulter, Inc., Brea CA) in a TL-100 tabletop ultracentrifuge (Beckman Coulter) for 16 hr at 4°C. The gradient was fractionated into 100 μL slices which were examined on a 5% native PAGE to analyze the purity of the fractions. Pure fractions that were devoid of any secondary bands or aggregations were selected. They were further concentrated using Amicon ultracell 50K filters (Merck Millipore) to a volume of 30–40 μL and were buffer-exchanged into 10 mM Tris, 50 mM NaCl buffer using prewashed Micro Biospin P-30 columns (Bio-Rad, Hercules, CA). The concentration of the buffer-exchanged sample was determined to be about 2 mg/mL by UV absorption using a Nanodrop spectrophotometer (Thermo Scientific, Wilmington, DE). The sample was either immediately frozen in liquid nitrogen for storage or used for plunging grids.
Details and data for characterizing the HJ complex have been described previously (Tong et al. 2014). The isolated protein-DNA HJ complex (without crosslinking) is stable in polyacrylamide gels and in solution for more than 12 hr at room temperature or 4C. Low levels of crosslinking (0.0035% glutaraldehyde) are required to stabilize the complex upon dilution.
The purified Holliday junction complex was checked for homogeneity by screening negatively-stained samples on a Morgagni electron microscope (FEI, Hillsboro, Oregon). For cryo preparation, we applied 3.0 μL of complex solution to a C-flat 1.2/1.3 Cu grid (400 mesh) (Protochips, Raleigh, NC), which had been glow discharged at 20 mA for 30 s. Grids were plunge-frozen with a Vitrobot Mark II (FEI) at 85% humidity, offset -2, blot time 7 s. Images were recorded on a Tecnai F30 electron microscope (FEI) operated at 300 kV and using a liquid-nitrogen cooled 626 cryo-specimen holder (Gatan Inc., Pleasanton, CA). We used the semi-automated acquisition program SerialEM (Mastronarde, 2005) to record 1359 movies with a Falcon II direct detector (FEI) at an underfocus set between 2.7 – 4.2 μm. Each movie consisted of 25 frames, collected in a 2 s exposure of 35.5 electrons/Å2. The nominal magnification was 78,000x, corresponding to a magnification of 100,000x on the detector and 1.4 Å pixel size on the specimen. Movie frames were aligned and summed using motioncorr (Li et al., 2013).
From 1359 images we picked a total of 66,033 particles using e2boxer (Tang et al., 2007) from 4x binned images (pixel size on the specimen: 5.6 Å). We carried out 2D classification using the ISAC procedure (Yang et al., 2012) implemented in the image processing package SPARX (Hohn et al., 2007). Image defocus was determined with CTFFIND3 (Mindell and Grigorieff, 2003). We selected 52 good class averages to calculate initial maps with EMAN2 (e2initialmodel.py, [Tang et al., 2007]). The highest scoring map was used to initialize refinement and 3D classification in FREALIGN (Lyumkis et al., 2013), as described in the main text and Figure 4A. The resolution was limited to 18 Å throughout this refinement. The final reconstruction was calculated using 2x binned data (pixel size on the specimen: 2.8 Å) and had a resolution of about 11 Å (Figure 4B), as indicated by the FSC = 0.143 threshold criterion (Rosenthal and Henderson, 2003). For this estimate, the two 3D reconstructions obtained from half the data and used to calculate the FSC curve were masked with a tight mask that added a margin of about 15 Å around the reconstructed density of the Holliday junction complex. To assess how the mask affected the FSC curve, the 3D reconstructions were recalculated with particle images with randomized phases beyond 18 Å resolution and used to correct the FSC curve obtained from the masked reconstructions (Chen et al., 2013). The resolution of about 11 Å was corroborated by calculating the FSC between the map and atomic model presented in this study (Figure 4B ), and by the local resolution map (Kucukelbir et al., 2014); Figure 4C. The final map was filtered at 11 Å resolution and sharpened with a B-factor of -2500 Å2. Source data for the FSC curves shown in Figure 4B are given in Figure 4—source data 1 and Figure 4—source data 2.
Construction of an excisive HJ complex model began with a core complex derived from the 3.8 Å crystal structure of full-length Int bound to unlinked HJ and arm DNA fragments (Biswas et al., 2005). The core complex consists of the catalytic and CB domains of Int (residues 75–356) and a 4-way DNA junction containing the C, C', B, and B' core half-sites. A P' arm complex containing IHF bound at H' and Int NTDs (residues 1–55) bound at P'1 and P'2 was constructed essentially as described (Seah et al., 2014), except that the NTD•DNA segments were derived from the NMR structure of an NTD•P'2 DNA complex (Fadeev et al., 2009). A P arm complex was also constructed as previously described (Seah et al., 2014), except for the use of the NMR-based model for the NTD•P2 segment. The assembled P' arm was joined to the C' half-site and the P arm was joined to the C half-site to generate the initial excisive complex model. During all construction steps, DNA segments were joined by superimposing the flanking duplexes on an idealized B-DNA splint fragment, ensuring smooth continuity of DNA twist. DNA segments were extended by superposition of idealized B-DNA fragments of the appropriate length, using a similar overlap.
The initial complex was fit manually into EM density as a rigid body. Excellent agreement with the core complex was readily obtained, but the P and P' arms were not well-positioned in the envelope. The P and P' arms were fit into density by iterative testing of small ( ± 1°) roll angles in the protein-free regions and by varying the IHF-mediated DNA bends. Changes in bending and torsion of IHF•DNA complexes (<10°) were made by rotating the DNA duplexes about the existing kinks (in-plane) or rotating the DNA arms about an axis connecting the two kink positions (torsion). The phosphate backbones of spliced, extended, and bent DNA segments were regularized to correct small deformations by highly restrained refinement in CNS (Brunger, 2007). To complete the model, the NTD-CB linker segments (residues 56–74) for the C', B, and B' subunits were fit into the EM envelope with essentially arbitrary conformations. The final model was refined with CNS using dynamic elastic network (DEN) restraints and torsion angle dynamics to obtain reasonable linker geometries and good overall stereochemistry (Schröder et al., 2007; 2010). To restrain flexible regions of the model to remain within the EM envelope and to optimize the final model fit to density, we included an energy term for agreement with the Fourier coefficients of the EM density.
An integrative HJ model was built from the excisive HJ complex based on the different bridging interactions that form during integrative recombination (Figure 2B; Tong et al., 2014). Binding of the C', C, and B' NTDs to the P' arm could be readily accommodated by changing the NTD-CB linker conformations; no change in the P' arm was required. A new P arm was assembled in the absence of Xis as previously described (Seah et al., 2014), using the NMR-based model for Int B-NTD binding to the P1 site. The initial model, lacking the phased bend induced by Xis, clearly indicated that the P arm must pass over P', with IHF bending at the H1 site responsible for re-directing the P1 site towards the Int tetramer. Small roll angle changes and bending at H1 and H2 (< 10°) were used to obtain a P arm geometry that avoids steric clashes and positions the Int NTD bound at P1 close to the Int B CB domain. A small amount of bending of the P' site was found to better accommodate the P arm crossing. After insertion of CB-NTD linkers for each of the Int subunits (arbitrary conformations), the model was optimized using CNS to obtain reasonable linker geometries and good stereochemistry elsewhere.
Structure of the cooperative xis-dna complex reveals a micronucleoprotein filament that regulates phage lambda intasome assemblyProceedings of the National Academy of Sciences of the United States of America 104:2109–2114.https://doi.org/10.1073/pnas.0607820104
Multiple effects of Fis on integration and the control of lysogeny in phage lambdaJounal of Bacteriology 173:4032–4038.
Efficient excision of phage lambda from the Escherichia coli chromosome requires the Fis proteinJournal of Bacteriology 173:4027–4031.
The topological mechanism of phage lambda integraseJournal of Molecular Biology 289:747–775.https://doi.org/10.1006/jmbi.1999.2771
The PyMOL Molecular Graphics SystemSan Carlos: DeLano Scientific.
The Escherichia coli Fis protein stimulates bacteriophage lambda integrative recombination in vitroJournal of Bacteriology 185:3076–3080.https://doi.org/10.1128/JB.185.10.3076-3080.2003
SPARX, a new environment for Cryo-EM image processingJournal of Structural Biology 157:47–55.https://doi.org/10.1016/j.jsb.2006.07.003
An overview of tyrosine site-specific recombination: From an Flp perspectiveMicrobiology Spectrum, 3, 10.1128/microbiolspec.MDNA3-0021-2014.
The bacteriophage lambda int gene product. A filter assay for genetic recombination, purification of int, and specific binding to DNAThe Journal of Biological Chemistry 253:7149–7157.
Bacteriophage lambda site-specific recombination proceeds with a defined order of strand exchangesJournal of Molecular Biology 204204:95–107.https://doi.org/10.1016/0022-2836(88)90602-X
Position and direction of strand exchange in bacteriophage HK022 integrationMGG Molecular & General Genetics 245:623–627.https://doi.org/10.1007/BF00282225
Quantifying the local resolution of cryo-EM density mapsNature Methods 11:63–65.https://doi.org/10.1038/nmeth.2727
The λ integrase site-specific recombination pathwayMicrobiology Spectrum, 3, 10.1128/microbiolspec.MDNA3-0051-2014.
Likelihood-based classification of cryo-EM images using FREALIGNJournal of Structural Biology 183:377–388.https://doi.org/10.1016/j.jsb.2013.07.005
Automated electron microscope tomography using robust prediction of specimen movementsJournal of Structural Biology 152:36–51.https://doi.org/10.1016/j.jsb.2005.07.007
Accurate determination of local defocus and specimen tilt in electron microscopyJournal of Structural Biology 142:334–347.https://doi.org/10.1016/S1047-8477(03)00069-8
Integrative recombination of bacteriophage lambda DNA in vitroProceedings of the National Academy of Sciences of the United States of America 72:1072–1076.https://doi.org/10.1073/pnas.72.3.1072
Characterization of the bacteriophage lambda excisionase (Xis) protein: The C-terminus is required for Xis-integrase cooperativity but not for DNA bindingThe EMBO Journal 11:3797–3806.
Fis targets assembly of the Xis nucleoprotein filament to promote excisive recombination by phage lambdaJournal of Molecular Biology 367:328–343.https://doi.org/10.1016/j.jmb.2006.12.071
UCSF Chimera-a visualization system for exploratory research and analysisJournal of Computational Chemistry 25:1605–1612.https://doi.org/10.1002/jcc.20084
Positions of strand exchange in mycobacteriophage L5 integration and characterization of the attB siteJournal of Bacteriology 178:5533–5536.
Bacteriophage lambda int protein recognizes two classes of sequence in the phage att site: Characterization of arm-type sitesProceedings of the National Academy of Sciences of the United States of America 79:7724–7728.https://doi.org/10.1073/pnas.79.24.7724
Crystal structure of the excisionase-DNA complex from bacteriophage lambdaJournal of Molecular Biology 338:229–240.https://doi.org/10.1016/j.jmb.2004.02.053
Differential affinity and cooperativity functions of the amino-terminal 70 residues of lambda integraseJournal of Molecular Biology 324:775–789.https://doi.org/10.1016/S0022-2836(02)01199-3
Nucleoprotein architectures regulating the directionality of viral integration and excisionProceedings of the National Academy of Sciences of the United States of America 111:12372–12377.https://doi.org/10.1073/pnas.1413019111
The shape of the DNA minor groove directs binding by the DNA-bending protein FisGenes & Development 24:814–826.https://doi.org/10.1101/gad.1900610
EMAN2: An extensible image processing suite for electron microscopyJournal of Structural Biology 157:38–46.https://doi.org/10.1016/j.jsb.2006.05.009
Protein-protein interactions in a higher-order structure direct lambda site-specific recombinationJournal of Molecular Biology 195:481–493.https://doi.org/10.1016/0022-2836(87)90177-X
Mapping the λ integrase bridges in the nucleoprotein Holliday junction intermediates of viral integrative and excisive recombinationProceedings of the National Academy of Sciences of the United States of America 111:12366–12371.https://doi.org/10.1073/pnas.1413007111
A structural view of Cre-loxP site-specific recombinationAnnual Review of Biophysics and Biomolecular Structure 30:87–104.https://doi.org/10.1146/annurev.biophys.30.1.87
Cre recombinaseMicrobiology Spectrum, 3, 10.1128/microbiolspec.MDNA3-0014-2014.
A chimeric cre recombinase with regulated directionalityProceedings of the National Academy of Sciences of the United States of America 105:18278–18283.https://doi.org/10.1073/pnas.0809949105
Sjors HW ScheresReviewing Editor; Medical Research Council, United Kingdom
In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.
Thank you for submitting your work entitled "Structure of a Holliday Junction Complex Reveals Mechanisms Governing a Highly Regulated DNA Transaction" for consideration by eLife. Your article has been reviewed by two peer reviewers, and the evaluation has been overseen by Sjors Scheres as the Reviewing Editor and John Kuriyan as the Senior Editor.
The following individuals involved in review of your submission have agreed to reveal their identity: Andres Leschziner (peer reviewer) and Reid Johnson (peer reviewer).
The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.
The excisive intasome structure of the λ phage presented here represents the culmination of decades of biochemistry and structural biology (including some misleading X-ray structures) into an atomic model based on an 11 Å cryo-EM envelope. Lambda phage is one of the classic systems for understanding site-specific recombination-its enzymology, regulation and topology. Many structures have been reported for components of the system, as well as complexes between the integrase (Int) and shorter DNA fragments, but this is the first structure of the entire assembly and, as such, is a landmark for the field. This structure will be of interest to a broad audience, as generations of biologists grew up on the molecular biology of lambda, and one of the reviewers referred to the structure as an "Ah moment" for those in the site-specific recombination-transposition field. Therefore, there was general agreement that this paper would make an excellent contribution to eLife, and the suggestions for improvement below are all relatively minor.
Suggestions for revision:
As you can read in the detailed comments from the two original reviews that are copied for your information below, one of the reviewers wonders whether any previous data exists to address the additional postulated roles of Xis in promoting excision. This was further discussed online, and the conclusion was that it did not seem that the authors could clearly address their postulate with past data and new experiments to test this would be difficult. There was also a question regarding data behind the revision in the location of the P-P′ crossing in the integrative intasome model. However, in principal, none of the reviewers were against including this speculative model in the manuscript.
Reviewer #1 raises questions about the plausibility of the modeled conformation of the DNA, in particular within the IHF interfaces. Also this was discussed online, and the conclusion was that most likely this was caused by the relatively low-resolution of the map combined with either the manual building or the molecular dynamics procedure that was used to refine the final model. Given the relatively low-resolution of the map, reliable atomic model building is expected to be very difficult. Perhaps the authors could include a cautionary note in their revision, where the perhaps less structurally-inclined reader is warned against interpretations of the atomic model in too much detail? In addition, the limited resolution of the reconstruction probably stems from structural variability in the sample, probably in particular from flexibility in the DNA, so that the proposed model would at best be one out of many possible structures. Therefore, the addition of a local resolution map to inform about the most flexible parts of the structure was deemed informative.
In the online discussion, there were further comments on a perceived lack of clarity with which the cryo-EM image processing was described. So please take note of the suggestions by reviewer #2 to improve this. In addition, it would be insightful to include an FSC curve between the atomic model and the EM map.
Finally, reviewer #2 suggests improvements to some of the figures. Please take particular note of the suggestions to improve consistency of Figure 9 and the chirality of the IHF arms in the model figures. Regarding the suggested changes to the movie and the cartoon: clearer figures and movies may indeed help to better convey the complicated organisation of your structure, but the extent to which you choose to modify the existing ones is left to your own judgement.
Detailed points by both reviewers:
Arguably, the most significant new aspect of the present cryo-EM based model is the close association of P and P' arm segments, suggesting an attractive additional function of Xis involving trans interactions with the P′ arm. (Note that in the fourth paragraph of the subsection “Structural basis for regulated directionality of integrative and excisive recombination”, the sentence should be modified to say "likely" or "are postulated to" play an important role in facilitating synapsis). The authors discuss many other features of the model that explain biochemical properties of the reaction, particularly with respect to its directional regulation.
A more speculative model for the integrative intasome assembly is also presented that somewhat modifies the 2014 Seah et al. model, particularly with respect to the location of the P-P′ arm crossing. It is not clear to me what data is driving the revision. Although I don't believe we have learned that much new from the revised integrative intasome model, I am not opposed to including it here for the sake of completeness.
There appears to be considerable modeling "freedom" in changing the path of DNA around the IHF dimers (especially H2) that seems to be necessary to generate the required writhe. The DNA paths deviate considerably from the 1IHF X-ray structure where the DNA is curved in a planar fashion as it follows along the basic sides of the protein. Is the DNA conformation within the IHF interfaces in the model structurally plausible?
In the subsection “Minimal structural changes link the two pairs of DNA strand exchange “– I am unclear on how the alternative HJ isomer (Figure 7) was modeled – what is the alternative HJ isomer structure specifically based on?
In the second paragraph of the subsection “Electron cryo-microscopy structure determination “– What is meant by 'randomly' assigned into 6 classes?
What is being referred to in the last sentence of the Abstract: "…many excisive and integrative recombination pathways, and they help explain their regulation"? I don't believe the various alternative pathways proposed for λ recombination are discussed (and don't need to be) here.
In the subsection heading 'An unexpected P' arm trajectory during excision', what is so unexpected? It seems very similar to earlier models and predicted from straightforward modeling without additional DNA bending-writhing.
In the sixth paragraph of the subsection “Structural basis for regulated directionality of integrative and excisive recombination”, what are the openings? The re-positioning of the Int-B subunit associated with the P1 site into a configuration competent to capture a naked attB seems like a very difficult step.
Would a color map displaying local resolution limits of the model be informative? Are the DNA loops as rigid as implied from the density maps?
I have listed a few minor scientific comments below. My main concerns relate to how material is presented in the manuscript. Recombination is a rather impenetrable field for the non-cognoscenti and I feel the authors could have done a better job making their results accessible to the widest possible audience. I have a section below where I expand on these concerns.
The authors speculate on the role of Xis in promoting recombination. They suggest that interactions between Xis and the Int-NTD stabilize the attL-attR complex. Is there any data supporting the role of this interaction? Are there known mutants in Xis that abolish or impair recombination whose general location agrees with the authors' model for their location in the excisive complex?
The authors also discuss the possible interaction between the highly basic (Arg, Lys) C-terminal tails of Xis and P/P' arm DNA and how this interaction could also stabilize the synapse. Have they (or anyone else) mutated these residues to test this idea?
The description of the EM data processing ("Electron cryo-microscopy structure determination") will not be easy to follow for non-experts. The distinction between 2D and 3D class averages is not made very clear in the text and people with passing familiarity with EM data processing will likely think that "particle alignment and classification" refers to 2D processing. Also, the 3D reconstructions generated during the 3D classification processing are called "classes" in the text but "reconstructions" in Figure 4A. I think both are OK as long as they are used consistently. I would still recommend adding 2D and 3D when necessary to draw distinctions.
Video 1: This movie has the potential for being the centerpiece of this paper yet all it does is show the structure rotating around a single axis. We don't even get to see the Holliday junction down its pseudo four-fold axis. Given how complicated the structure is and that many readers (particularly those not familiar with the λ system or not that structurally inclined) will have a hard time translating the cartoons shown in the figures into 3D structures, this movie could be the one point where all these complexities can be made accessible. If nothing else, the structure should rotate around both the vertical and horizontal axes. But why not go further? After introducing the entire structure, the authors could strip all the proteins out to show the path of the DNA, then add each protein component one at a time, then remove it and go to the next. (For example, show the IHFs, remove them, show the different Int monomers, remove them, show the Xis's, etc. Proteins could also be shown in pairs to highlight their interactions.) It would further help if the structure rotated while those components are displayed. The movie could end by slowly adding, and keeping, the protein components, to build back the entire complex. It is true that readers could download the structure and do this themselves but the reality is that only people with a direct interest in λ recombination would do it.
Figure 9: This figure would benefit from a little better labeling of the protein components. The arrows inside the B/C/B'/C' boxes in panel A are very hard to see. I think the Integrative Holliday junction should be shown with as much detail as the Excisive one given that an important part of the paper is modeling the Integrative complex. It would be very useful for the reader to see the different location of the Int bridges in the two complexes. I found the color scheme in panel B confusing. Why do the Int ovals change colors between the top and middle cartoons?
Figure 5: Although I realize this would take some work, a more geometrically realistic cartoon would help readers orient themselves better when looking at the structure. (For example, the IHF bound to H' is shown face-on in the cartoon while in the structure is appears edge-on, with the DNA in front.) Also, the handedness of the IHF is wrong in the cartoon in panel B (i.e. the wrong IHF arm is in front of the DNA). It would be useful if the putative density for the NTD of Int-C were indicated in this figure.
Figure 6: It would help to add the axis of rotation relating the cartoons in A and B. It may not be obvious to everyone that the axis is perpendicular to the page, rather than parallel to it as it is the case for the structures. Also, the legend should specify that the structure in C corresponds to the cartoon viewed from the front.
Figure 8: Every IHF in this figure has the wrong handedness (i.e. the wrong arm is in front of the DNA).https://doi.org/10.7554/eLife.14313.021
- Arthur Landy
- Nikolaus Grigorieff
- Arthur Landy
- Gregory D Van Duyne
- Nikolaus Grigorieff
The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.
We thank Mike Rigney for help with electron microscopy, and Christine Lank, Joanne Nelson, and Joan Boyles for help with complex preparation. This work was supported by NIH grants GM0627 and GM033928 (AL), GM108751 (GV), and P01 GM62580 (NG).
- Sjors HW Scheres, Reviewing Editor, Medical Research Council, United Kingdom
- Received: January 9, 2016
- Accepted: April 7, 2016
- Version of Record published: May 25, 2016 (version 1)
© 2016, Laxmikanthan et al.
This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.