DNA Topology and the Initiation of Virus DNA Packaging

During progeny assembly, viruses selectively package virion genomes from a nucleic acid pool that includes host nucleic acids. For large dsDNA viruses, including tailed bacteriophages and herpesviruses, immature viral DNA is recognized and translocated into a preformed icosahedral shell, the prohead. Recognition involves specific interactions between the viral packaging enzyme, terminase, and viral DNA recognition sites. Generally, viral DNA is recognized by terminase’s small subunit (TerS). The large terminase subunit (TerL) contains translocation ATPase and endonuclease domains. In phage lambda, TerS binds a sequence repeated three times in cosB, the recognition site. TerS binding to cosB positions TerL to cut the concatemeric DNA at the adjacent nicking site, cosN. TerL introduces staggered nicks in cosN, generating twelve bp cohesive ends. Terminase separates the cohesive ends and remains bound to the cosB-containing end, in a nucleoprotein structure called Complex I. Complex I docks on the prohead’s portal vertex and translocation ensues. DNA topology plays a role in the TerSλ-cosBλ interaction. Here we show that a site, I2, located between cosN and cosB, is critically important for an early DNA packaging step. I2 contains a complex static bend. I2 mutations block DNA packaging. I2 mutant DNA is cut by terminase at cosN in vitro, but in vivo, no cos cleavage is detected, nor is there evidence for Complex I. Models for what packaging step might be blocked by I2 mutations are presented.


Introduction
Large dsDNA viruses use an ATP hydrolysis-powered motor to package DNA into preformed empty shells, called proheads (reviewed in [1][2][3][4]). For many tailed bacteriophages and the herpesviruses, replication and recombination produce concatemers, i.e., end-to-end multimers of viral chromosomes. During packaging, an endonucleolytic cut is made to generate the DNA end which is translocated into the prohead to initiate DNA packaging. After prohead filling, a observations suggest a model in which IHF and interactions with cosB facilitate tetramer formation, and that the tetramer is the active form of terminase throughout the early steps of DNA packaging. This view can be reconciled with in vivo results, as follows. Although λ forms plaques on cells lacking IHF, the virus yield is reduced to about 30% the yield in IHF + cells [33]. One possibility is that the part of λ DNA packaging that is IHF-dependent may be the fraction of tetramers that require IHF and DNA to assemble, with the remaining, IHF-independent packaging reflecting the level of tetrameric terminase assembled independently of IHF and cos. These models suggest that the terminase tetramer does not undergo major structural changes during the early packaging steps.
I2 is a ca. 33 bp segment of unknown function between cosN and cosB ( Fig 1B). An early study showed a correlation between reduction in virus yield and the size of small insertion and deletion mutations in I2, suggesting that the cosN-cosB spacing is crucial for cos function. The spacing changes affected initiation, but not termination, of DNA packaging. Both 7 bp and 11 bp deletion mutations were lethal [34]. Later work showed that altering the cosN-cosB spacing resulted in incorrect nicking, with displacement of nick positions to the right for insertions and A. Structure of cos λ . cosN is the site at which TerL endonuclease centers introduce staggered nicks to generate the cohesive ends of λ virion DNA. cosB is the complex site at which TerS binds to anchor TerL: R3, R2 and R1 are TerS binding sites, and I1 is a binding site for the E. coli site-specific DNA bending protein, IHF. I2 is located between cosN and cosB. cosQ is essential for DNA packaging termination. B. Alignment of the I2-containing left DNA ends, i.e., bp 1-70, of λ-like phages N15 (green), λ (black), 21 (blue) and gifsy-1 (red). The I2 segment extends approximately from bp 18 to 50. Approximate positions of R3 segments are underlined. C. The sequence of the left DNA end, bp 1-70, of I2 + (above) and I2 re18-50 (below). The I2 re18-50 mutation replaces the AT-rich I2 + sequence without changing the cosN-cosB spacing. Underlining highlights the poly-dA and poly-dT segments of I2 + .
to the left for deletions, indicating that proper cosN-cosB spacing positions the TerL endonuclease domains on cosN [25]. A 22-bp deletion of cosN, called cos2, abolishes nicking. I2 is strongly conserved in the λ-like phages suggesting that I2 functions as more than a spacer (Fig 1B). To study this, we investigated mutations that change the I2 sequence but not the cosN-cosB spacing.

I2 is essential
To ask if the sequence between cosN and R3 is required for DNA packaging, we constructed a cosmid with an I2 substitution mutation, I2 re18-50 , in which the 33 bp from λ bp 18 to 50 were changed. In I2 re18-50 , the cosN-cosB spacing was retained, but not the G+C content ( Fig 1C). Isogenic cosmids differing only at I2, being I2 + [pBUC8] or I2 re18- 50 [pBC2] were subjected to in vivo cosmid packaging by a λ helper phage. In this assay, replication generates cosmid concatemers that are packaged by the helper phage DNA packaging machinery. Phages carrying linear cosmid multimers are assayed as ampicillin-transducing phages. While the I2 + cosmid gave a yield of 1.7 x 10 7 Ap-transducing phages/ml, the I2 re18-50 cosmid yield was <10 1 Aptransducing phages/ml, indicating that scrambling λ bp 18 to 50 causes a profound DNA packaging defect. The presence of the I2 + and I2 re18-50 plasmids did not affect the helper phage yield, indicating that I2 re18-50 acts in cis, at least in a plasmid background.
I2 re18-50 is cis-specific As part of cos, I2 mutations are expected to act in cis, i.e., to affect packaging of an I2 mutant chromosome, but not to act in trans. We did a phage complementation experiment to ask if I2 re18-50 is cis-specific. We first crossed the I2 re18-50 mutation into λ-P1. We used λ-P1 as wild type λ because λ-P1's prophage is a plasmid, making it useful for studies with lethal mutations such as I2 re18-50 (see Materials and Methods). λ-P1 also transduces kanamycin resistance, so that the yield of a non-plaque forming derivative, i.e., λ-P1 I2 re18-50 , can be determined. As the I2 + phage, we used λ cI857 red3, an att + phage that integrates its prophage into the bacterial chromosome. Lysogens of the I2 + and I2 re18-50 phages, and a dilysogen carrying both, were constructed using the recAhost MF3510. As expected from the above cosmid packaging experiment, I2 re18-50 reduced the yield of λ-P1 by about 5 orders of magnitude (Table 1, compare lines 2 and 4). Because I2 re18-50 is lethal, the yield of λ-P1 I2 re18-50 was determined by titering kanamycin-transducing phages. The efficiency of the kanamycin-transducing phage assay was about 50% when compared with the assay for plaque-forming phages (line 4). In the lysate of the I2 + and I2 re18-50 dilysogen, the presence of λ-P1 I2 re18-50 resulted in a modest reduction (5-fold) in the yield of λ I2 + (Table 1, compare the plaque forming phage yields in lines 1 and  3). In turn, the profoundly low yield of λ I2 re18-50 from a mono-lysogen was increased slightly (30-fold) by λ I2 + , but remained severely low (compare the kanamycin transducing phage yields in lines 2 and 3). We ascribe the modest increase in the yield of λ-P1 I2 re18-50 to the effect of λ I2 + on increased concatemer production and late gene dosage (see Materials and Methods). The yield of the λ I2 re18-50 phage is >2000-fold less than that of λ I2 + , confirming that the I2 re18-50 defect is cis-specific (line 3, compare the plaque forming phage yield of λ I2 + with the kanamycin transducing phage yield of λ-P1 I2 re18-50 ). As a control, we looked at the complementation behavior of a phage carrying a mutation, Aam42, expected to be recessive to the wild type A + allele. Aam42 is a nonsense mutation of the fifth-to-last codon of the A gene [35]. The truncated gpA of λ-P1 Aam42 lacks the C-terminus of the prohead binding domain, and the resulting terminase is defective in prohead binding. We reasoned that in a coinfection by λ A + and λ-P1 Aam42, wild type terminase would act in trans, sponsoring packaging of both A + and Aam42 DNAs. Induction of a mono-lysogen of λ-P1 Aam42 alone resulted in a very low virus yield (line 6), as expected for an amber mutant. Coinfection with λ A + increased the yield of λ-P1 Aam42 about 2000-fold (line 7), indicating that the Aam42 mutation indeed is recessive.

Location of the critical I2 segment
Scanning mutagenesis was done across I2 by constructing a series of mutations changing blocks of 6 or 12 bp in I2. The mutational changes (Fig 2) were constructed in cosmids. Preliminary cosmid packaging experiments showed that a critical segment of I2 was located in the λ bp 30-35 segment. To confirm this finding, the I2 mutation-containing cosmids were crossed with λ-P1 cos2, a mutant deleted for cosN. Since cosN and I2 are adjacent, it was expected that many of the recombinants that rescued cosN + would also co-rescue the I2 mutations. For most of these crosses, plaque-forming recombinants carrying the I2 mutations were readily obtained, with the exceptions of I2 re30-35 . For I2 re30-35 , the cross lysate was used to transduce MF1427 to Kn R , and transductants were screened for lysogens unable to produce plaque-forming phages. These recombinant prophages were found by sequencing to be I2 re30-35 mutants, confirming that I2 re30-35 is lethal. Virus yield studies showed that all of the I2 segments, except for bp 30-35, could be replaced without major effect on virus yield. In contrast, the yield of λ I2 re30-35 was reduced to~4 phage/cell, a level below that required for plaque formation (Fig 2). The results indicate the segment from bp 30-35 is crucial for virus viability. The much greater packaging defect of λ-P1 I2 re18-50 (Table 1), compared to I2 re30-35 (Fig 2), indicates that bp outside of the bp 30-35 segment play a role in I2's DNA packaging function.

A complex static bend at I2
Poly-dA tracts centered at approximately 10-11 bp intervals cause DNA to form a static twodimensional bend (reviewed in [36]). Inspection of the bp 18-to-50 segment indicated the presence of poly-dA tracts that were irregularly placed on both strands (underlined in Fig 1C). The I2 poly-dA tracts suggested that there was likely a complex, i.e., non-planar, intrinsic bend at I2. To ask about the presence of intrinsic bending at I2, the electrophoretic mobility of permuted 150 bp fragments containing a 35 bp I2 DNA insert (λ bp 18-52) was examined. As a positive control, the I1 segment extending from λ bp 65-90 was inserted in pBend [37]. As described earlier, I1 contains an intrinsic bend. Mobility was reduced for fragments carrying either I1 or I2 near the mid-point, indicating the presence of static bends in both (Fig 3). Calculations indicate the overall I2 bend angle is about 28°, and that of the I1 control was about 34°. The I2 re30-35 mutation had at most a minor effect on mobility and the calculated bend angle, but the severe I2 re18-50 mutation abolished the static bend. These results are consistent with a model in which a complex, three-dimensional bend at I2 is required for DNA packaging. The electrophoretic mobility data (Fig 3) indicate that I2 re30-35 has little effect on I2's bend, whereas the more extensive I2 re18-50 abolishes the bend. I2 re30-35 has little effect on apparent bending and is significantly less severe than I2 re18-50 . The lack of a change in apparent bending indicates that the bp 30-35 segment does not contribute significantly to the bending detected in the electrophoresis assay. I2 re18-50 abolishes bending. Taken together, the results suggest that I2 bending is produced in the I2 bp that flank the bp 30-35 segment, i.e., upstream bp 18-29 and downstream bp 36-50. How might these two flanking segments contribute to I2 bending? Stretches of poly-A base pairs result in DNA curvature [36,38,39]. Clues to bending at I2 and the effects of the I2 mutation come from examining the sequence. The wild type I2 sequence is: 20. 30. 40. 50. 5'GTTTTCGCTATTTATGAAAATTTTCCGGTTTAA-3' 3'CAAAAGCGATAAATACTTTTAAAAGGCCAAATT-5' where A 4 tracts are in bold and the bp changed by the I2 re30-35 mutation are underlined. Note the bottom strand A 4 T 4 and A 4 segments, which contribute static bends, as follows. Koo and Crothers studied static bending of a very similar sequence with A 5 and A 4 T 4 segments. When the centers of these two segments were separated by ca. 10 bp, i.e., approximately a helical turn, the bends were reinforcing, producing maximal retarded gel mobility. When these segments were separated by 15 bp, i. e., out of phase, minimal gel mobility effects were observed. In the I2 sequence, the center-to-center distance from the A 4 T 4 sequence to the A 4 sequence is an out-of-phase 17 bp. As the 17 bp spacing is not 1.5 turns, rather than cancelling, the two bends generate a skewed, 3-dimensional trajectory. I2 re30-35 alters the T 4 tract of the A 4 T 4 sequence, leaving A 4 tracts with a 19 bp center-to-center spacing, again a spacing neither in-phase or cancelling. Whereas the severe I2 re18-50 change abolishes bending, the less severe I2 re30-35 likely alters the geometry of the bending. Making the assumption that I2 re30-35 simply eliminates the A 4 tracts known to contribute to bending, we propose that I2 re30-35 alters I2 bending to produce a DNA trajectory that is unfavorable for interaction with terminase.
One alternative explanation, which we have not investigated, is that the bp 30-35 segment is flexible and permits bending at I2 that is required for proper terminase contacts. Obviously a molecular explanation for the role of I2 requires additional biophysical and structural studies beyond the scope of this report.

Failure to package DNA
During DNA translocation, the λ prohead shell undergoes a structural transition to form the mature head shell [41]. The transition is thought to occur when about 30% of the DNA molecule has been translocated [42]. The transition involves an expansion of the shell due to local structural rearrangements by the major capsid protein, gpE [43]. In the electron microscope, the transition is seen as a change from the thick-walled spherical prohead to a larger, more angular and thin-walled form [44]. To ask about the effects of I2 mutations on phage assembly, we purified phage-related structures from lysates of I2 + , I2 re30-35 and I2 re18-50 . Electron microscopic examination indicated that the I2 mutants have an early DNA packaging defect, as follows. While lysates of the wild type phage contained roughly equal amounts of proheads and intact phages, the I2 re30-35 and I2 re18-50 lysates contained nearly all proheads, indicating a DNA packaging defect prior to translocation of 30% of viral chromosome (Fig 4). We next examined initial cos cleavage, a packaging step preceding translocation.

cos cleavage in vivo
In vivo cos cleavage was studied by examining restriction enzyme (AccI)-digested DNA isolated from induced lysogens. In a wildtype infection, terminase cutting at cos cleaves the cos-containing joint (J) DNA fragment, producing the nuclease susceptible mature right (R end ) and the left (L end ) chromosome end fragments (Fig 5B). The L end DNA is protected from nuclease digestion by assembly into Complex I and by packaging into the prohead. J and L end DNA fragments were detected by Southern blot analysis using a probe homologous to the left chromosome end. The experiment showed that at the time of sampling, about 80% of the DNA had been cut for the I2 + control phage (Fig 5). For the I2 re30-35 mutant, an intermediate level of cutting was observed, and for the I2 re18-50 mutant no cleavage was evident. As expected for the λ-P1 cos2 negative control phage, a cosN deletion mutant, no cleavage was found.

cos cleavage in vitro
We sought to confirm the in vivo cos cleavage results with in vitro cos cleavage studies. Linearized cosmid DNA was used as substrate for terminase cleavage reactions that were done in the  In an infection by a mutant that is able to cut cos and form complex I, but is unable package DNA, the uncut J and cut L fragments can be detected. AccI-cut total DNA was electrophoresed on a 0.8% agarose gel and transferred to a membrane for southern blotting. B. AccI digested intracellular DNAs: lane 1 is the DNA probe used for Southern blot assay (λ bp 177-2099). Lanes 2 and 3 are AccI-cut λ DNA loaded on the gel before (lane 2) and after heat treatment at 70°C for 10 min to melt the cohesive ends (lane 3). Phage DNAs from presence and absence of IHF (Fig 6). With the exception of the cosN deletion negative control DNA, each of the substrates, including I2 re18-50 and I2 re30-35 , showed efficient cos cleavage. Modest stimulation by IHF was observed at low terminase concentrations, also for all the substrates except the ΔcosN DNA used as the negative control. In our experience, in vitro cos cleavage results generally correlate well with in vivo assays, so the discrepancy between the apparant lack of cos clevage in vivo and efficient cleavage found in vitro is striking.

I2 is Critical for a Post-recognition Step in Initiation of DNA packaging
The present results show that, in addition to providing proper spacing between cosN and cosB, I2 plays a functional role that is critical for an early step in DNA packaging by λ. The severely lethal I2 re18-50 mutation reduces virus yield by~10 −5 . I2 re18-50 is cis-acting, preventing the packaging of the mutant DNA but not interfering with DNA packaging by a co-infecting I2 + phage ( Table 1). This cis-acting behavior is expected for a site that is involved in a post-recognition interaction required for specific terminase contacts, or perhaps in providing a DNA structure that facilitates contacts with adjacent segments. The I2 DNA sequence contains poly-dATP tracts, known to create static DNA bending [38]. Gel mobility studies with circularly permuted, I2-containing DNA molecules demonstrated that I2 is bent (Fig 3). Scanning mutagenesis showed that bp 30-35 are critical for I2 function, and that the flanking segments can be mutated with only mild effects on virus yield (Fig 2). Although lethal, the I2 re30-35 mutation reduced virus yield only about ten-fold (Fig 2), and did not greatly affect DNA bending (Fig 3). For the severely defective λ I2 re18-50 mutant, lysates contained only unexpanded proheads, indicating that if there is any attempt to translocate DNA, the process fails prior to the prohead expansion event that occurs when about 30% of the DNA has been packaged [42]. For the milder I2 re30-35 mutation, reduced levels of expanded proheads and intact phages were observed, indicating I2 re30-35 does not completely abolish I2 function (Fig 4). In in vitro reactions, neither I2 mutation had a significant effect on cos cleavage, with or without IHF (Fig 5). In contrast, no evidence for cos cleavage was found in vivo for λ I2 re18-50 , and an intermediate level of cleavage was found for λ I2 re30-35 (Fig 6). The in vitro cos cleavage results indicate that I2 functions at a post-cos cleavage step of DNA packaging. Confounding this simple conclusion is the apparent lack of in vivo cos cleavage for the severe I2 re18-50 mutant, and an intermediate level for λI2 re30-35 . These contradictory results are surprising, as previous work on cos mutants has generally found a good correlation between in vitro and in vivo results [19,20,[45][46][47].
[Note: a notable exception is that cos cleavage in vivo, but not in vitro, requires assembled proheads and gpFI, as reviewed in [48]. No evidence connects the gpFI/prohead and I2 discrepancies.] We next consider models for I2's function that account for the cos cleavage conundrum.

Models for I2's function
There is a wide range of possible explanations for the behavior of I2 mutants. To explain the difference between in vivo and in vitro cos cleavage results, there are possible experimental differences in ionic strength and/or composition, that might account for the observations. Such differences would pertain only to the behavior of the I2 mutants, and not the wild type. For the present discussion, we consider models that invoke known aspects of the λ DNA positive and negative control phages λ-P1 I2 + and λ-P1 cos2 (ΔcosN), respectively are in lanes 4 and 5. In lanes 6, 7, and 8 are DNAs from λ-P1 I2 + (a second sample), λ-P1 I2 re30-35 and λ-P1 I2 re18-50 , respectively. packaging process. With this constraint, we propose several explanations for the behavior of λ I2 re mutants (Fig 7).
1. An I2 defect results in cohesive end religation. In this scenario, cosN nicking occurs normally for λ I2 re18-50 , but due to a subsequent defect in the packaging pathway, terminase   Early DNA packaging steps at which I2 might act. Four steps at which an I2 defect might interrupt DNA packaging are numbered. The step blocked for Aam42 terminase is also indicated. Model 1a: Failure to separate the newly created cohesive ends, followed by dissociation of terminase, and re-ligation that reseals the nicks. Model 1b: Reannealing of the cohesive ends followed by religation. Model 2: I2 mutations block formation of, or destabilize, Complex I. Note that accompanying Complex I formation, the R end DNA end is released and subject to exonuclease digestion; this is indicated by R end . Model 3: I2 mutations interfere with proper threading of the DNA through the motor assembly so that the DNA is translocated into the cytoplasm and is subject to exonuclease digestion. DNA represents the nuclease-susceptible virion DNA. The Aam42 defect is the absence of an intact prohead binding domain at the C-terminus of TerL, which prevents Complex I from docking on the portal and assembling an active motor [35,49,50]. dissociates from the nicked cos and the nicks are resealed by host ligase (Fig 7, Model 1a). This defect might occur if a terminase-cos interaction required for strand separation fails to occur. Alternatively, the defect might be in a post-cohesive end separation step at which a required terminase-I2 interaction is needed to prevent reannealing and religation of the cohesive ends (Fig 7, Model 1b). Models 1a and 1b are testable by directly asking if the intracellular level of the J piece, i.e., the piece with joined cohesive ends of a non-replicating I2 re18-50 cos, remains constant relative to other phage DNA segments, following late gene expression in vivo. In both models, the intracellular level of the J piece is expected to remain constant, even though terminase is active. Models 1a and 1b can be distinguished by looking for a strand separation defect in in vitro cos cleavage with the I2 re18-50 substrate DNA.
2. An I2 defect blocks Complex I formation or stability. I2 re18-50 might prevent terminase from assembling Complex I, or, if it is assembled, Complex I might be unstable (Fig 7, Model  2). In either case, both newly formed DNA ends would be subject to exonuclease attack. This model presupposes that Complex I is fundamentally different from the nicking and strand separation complexes, and requires critical interactions with the topologically complex I2. This scenario, like those of Models 1a and 1b, is testable by quantitating the amount of a non-replicating I2 re18-50 -containing J piece following late gene expression. If Model 2 is correct, the I2 re18-50 -containing J piece would decline relative to control segments of the bacterial chromosome. [Note: Attempts by us to demonstrate in vitro Complex I assembly with I2 + and I2 re18-50 DNAs were unsuccessful (unpublished).] 3. An I2 defect derails DNA threading. In this model (Fig 7, Model 3), I2 mutations do not interfere with Complex I formation or stability, but prevent proper threading of the DNA through the portal so that the DNA is translocated into the cytoplasm and is subject to exonuclease digestion. In this model, the I2 defect affects translocation, a step that occurs after Complex I formation. This model can be tested by a genetic epistasis test using a previously studied TerL λ mutation, Aam42, as follows. The Aam42 mutation is a chain termination codon such that TerL Aam42 lacks the C-terminal 5 amino acid residues [35]. Aam42 terminase is nickingcompetent. Solid genetic and biochemical evidence indicates that the C-terminus of wildtype TerL contacts the prohead's portal vertex [49][50][51]. During a λ Aam42 infection, cos cleavage is accompanied by formation of Complex I. That is, when intracellular DNA from a λ Aam42 infection was examined by restriction enzyme digestion and gel electrophoresis, the L end piece, but not the nuclease-susceptible R end piece, was observed [35].
In model 3, an I2 mutant forms a mal-functioning motor. Model 3 posits that a λ Aam42 I2 re18-50 double mutant will have the same phenotype as the Aam42 single mutant, because the I2 defect occurs after the Aam42 defect. Should this genetic test support Model 3, further in vitro packaging experiments could ask about misdirected translocation.
Note that each of the models involves a critical terminase-I2 interaction that fails in the I2 mutants. The models differ at the point along the DNA packaging pathway at which the crucial interaction occurs. Although the present study provides little structural insight into the nature of the critical interaction, the I2 mutations provide a clean block to an early packaging step that likely occurs after cos cleavage. As little is known about these steps for any DNA virus, the I2 re mutations are valuable tools for dissecting the nature of these steps in the packaging pathway.

DNA topology in initiation of viral DNA packaging: SPP1 and λ
During progeny virus assembly, virus DNA is specifically selected. In the tailed bacteriophages, recognition involves specific interactions between TerS and the pac or cos sites. The several TerS structures that have been determined share a three-domain organization [13-15, 52, 53]. The N-terminus is a DNA binding domain (DBD), involved in viral DNA recognition. The DBD is a small α-helical bundle containing a helix-turn-helix (HTH) DNA binding motif. The DBD is tethered to a central domain consisting of two long antiparallel α-helixes that form an antiparallel coiled-coil which further oligomerizes into a hollow cylinder. TerS oligomers contain 8-12 TerS monomers, depending on the virus. At the TerS C-terminus is a β-barrel extension of the cylinder; the C-terminus also contains a specificity domain for interacting with TerL [54,55]. The DBDs of TerS oligomers are arrayed around the periphery of the central cylinder. A current model proposes that the viral DNA is wrapped around the TerS oligomer to form a nucleosome-like structure [14,52,53,56]. Strong support for nucleosome-like wrapping comes from pac phage SPP1, as follows. The SPP1initial cleavage site, pacC, is flanked by TerS SPP1 binding sites pacL and pacR sites. TerS SPP1 has an extensive footprint at pac SPP1 , including pacL and pacR [56]. pacL is intrinsically bent. Footprint experiments strongly indicate that TerS SPP1 (G1P) wraps pacL into a nucleosome-like structure. TerS SPP1 binding is specific, with moderate affinity (K app = 9 nM), and roughly two TerS SPP1 oligomers bind pac. Thus there is a role for DNA topology at the earliest step, pacC cleavage, in SPP1. The pre-cleavage complex for SPP1 possibly has a symmetric feature, as TerS SPP1 oligomers bracket pacC site. The post-cleavage events leading to docking of SPP1 terminase on the portal vertex are unclear.
[Remarkably, the DNA sequences bound by TerS SPP1 at pacL and pacR do not share identity. A recent study indicates that TerS SPP1 recognizes local DNA structure rather than a specific sequence [57].] λ's pre-cleavage nucleoprotein assemblage also involves DNA topology, in this case the IHF-enhanced intrinsic bend at I1. In contrast to SPP1, the TerS λ assemblage at cosB is asymmetric, being located on only one side of cosN. The two-fold rotational symmetry of cosN suggests that cleavage is carried out by symmetrically disposed TerL λ monomers. I2's static bend acts at a second, post-cleavage step of packaging, as discussed above. Whether there is a similar role for DNA topology in a post-cleavage packaging step for SPP1 and other DNA viruses remains to be learned.

Post-cleavage Gymnastics
The assembled translocation motor includes a pentameric TerL ring, in both T4 [58] and Phi29 [59,60]). In both cases, the TerL molecules are asymmeteically arrayed on the prohead's portal protein. Especially in the case of λ, the transition from the pre-nicking complex with symmetrically disposed TerLs, to the translocation motor, with asymmetrically arrayed TerLs, suggests there might be a substantial rearrangement of TerLs during the transition. This line of thought contrasts with biochemical studies suggesting that TerS 2 :TerL 1 protomers assemble tetramers; the resulting tetramers are fully competent for both cos cleavage and translocation [16,17]. Electron microscopic examination indicates that the tetramers possess great structural plasticity [61]. It is thus possible that the tetramer's endonuclease domains could adopt both symmetric and asymmetric orientations without major rearrangement of the global structure of the tetramer.
The final point in the DNA packaging cycle where a rearrangement is indicated is termination. In λ, cutting of the downstream cos, to finish DNA packaging, requires cosQ [62]. In the absence of cosQ, translocation does not arrest at the downstream cos, rather packaging continues, resulting in failure to terminate the chromosome being packaged, a lethal event. Examination of the packed downstream cos shows that the top DNA strand is properly nicked but the bottom strand is not nicked [27]. It is argued that cosQ acts to present a symmetrically disposed TerL endonuclease domain to cosN's bottom strand. Again, the role of cosQ might include a major rearrangement of terminase architecture, or a local repositioning of an endonuclease domain. A third possibility is that cosQ recruits a new terminase protomer to enable cosN cleavage [27]. The one clear conclusion is that much remains to understand about terminase architecture and dynamics during DNA packaging.
Bacteria, phages and plasmids are listed in Table 2.

Induction of lysogens
In general, the following procedure was used to make lysates for the cosmid packaging assay, burst size determination, phage x plasmid crosses, and in vivo cos cleavage assay. Lysogens were grown in 5 ml of LB at 31°C to a cell density of~5 x 10 7 cells per ml, transferred to 42°C for 20 min for prophage induction, and then aerated at 37°C for a 70 min phage growth period. After CHCl 3 was added to lyse the cells, the lysate was clarified by centrifugation in a clinical centrifuge for 10 min at 4°C. To obtain counts of induced cells, dilutions of the cultures were made before induction at 42°C and plated at 30°C. Lysates were diluted in 10 mM MgSO 4 when required for titrations.

Plasmid constructions
To construct pBUC8 and pBC2, a 1248 bp fragment (SapI, bp 47,712 to MluI, bp 458) of λ DNA was cloned into pUC19 using standard methods, to create pBUC8. The sequence corresponding to λ bp 18-50 in pBUC8 was replaced with scrambled sequence to make the analogous plasmid, pBC2, containing the I2 re18-50 mutation, as follows. The middle of the I2 re18-50 mutant sequence from bp 30 to 35 is a recognition site for the PstI restriction enzyme. Oligonucleotides with the left and right halves of the I2 re18-50 mutation, including the PstI sequence, were used in PCR amplifications to produce XbaI (48442) to PstI (30) and PstI (30) to MluI (458) segments that were cloned into pUC19. The left and right half I2 re18-50 segments were subsequently ligated at the PstI site to produce the I2 re18-50 containing Xba-to-Mlu segment, which was used to replace the I2 + sequences in pBC2 ( Assay for cis-specificity: Dilysogen construction and burst size determinations In this experiment, the λ I2 + prophage was att + gam + red3 and resided in the host chromosome. The λ I2 re18-50 prophage background was λ-P1, which forms a plasmid prophage. Because λ-P1's Kn R cassette is inserted between the Sal sites at λ bps 32745 and 33244, the phage is Δ (gam-bet) and is defective in generating concatemeric, packageable DNA in a recA host, such as used here. The bet defect abolishes Red recombination, one source of concatemers, and the gam defect permits RecBCD nuclease attack of rolling circle replication, another source of concatemers [71]. In addition, the defect in producing concatemers results in a reduced late gene dosage, reducing the level of proteins involved in virion assembly. In the complementation experiment, λ I2 + provides Gam, permitting rolling circle replication by λ-P1 I2 re18-50 .
Crossing I2 mutations into phage Plasmids pOER1, pOER2, pOER3, pOER4, and pOER5 were used to transform MF1427(λ-P1:5R cI857 nin5 Sam7 cos2 Kn R ) to ampicillin resistance by standard methods. Plasmid versus phage crosses were carried out by prophage induction, and the resulting cross lysates were examined for plaque forming recombinants by plating on MF1427 in TBSA at 37°C. Plaque forming recombinants were examined for I2 markers by sequencing appropriate PCR products. The inviable I2 re30-35 recombinant was isolated by screening kanamycin resistant transductants of MF1427 for presence of a prophage unable to release plaque forming phages, followed by sequencing PCR products.

Determination of static bend
Permutation analysis was done with a 150 bp DNA fragment containing I2 + , I2 re30-35 , or I2 re18-50 sequences, along with I1 + as a positive control. (A) A 35 bp DNA fragment containing λ I2 + , I2 re30-35 , I2 re18-50 , or I1 + sequence was inserted into the XbaI and SalI sites of the vector pBend2 [40]. When a restriction enzyme cuts a site in the tandem repeat segment that flanks the insert, a 150 bp DNA fragment is produced which contains the I2 insert at various positions ( Fig 3A). These permuted fragments were electrophoresed in an 8% polyacrylamide gel in 1x TBE buffer (150 mM Tris, 32 mM boric acid, 1 mM EDTA, pH 8.3) for 19 hours at 150 V at 4°C. The gel was stained in ethidium bromide (1 μg/ml) for 15 min, washed in 1x TBE buffer, and then visualized under UV light. (B) Relative sizes of the permuted DNA fragments in (A) are plotted against the positions of the restriction site (permutation position). A relative mobility of 1.0 indicates the mobility of a 150 bp DNA marker. The DNA bending angles (α) are adjacent to the appropriate mobility plot. The bend angle for each permuted fragment was calculated by following an empirical relationship between the relative electrophoretic mobility retardation caused by bending and the bending angle [40].

Electron Microscopy
Pelleted phages were resuspended and applied to 400 mesh copper grids coated with Formvar and carbon. The grids were then negatively stained with 2% potassium phosphotungstate (pH 6.5) for 1 min, and examined under transmission electron microscope at 30,000X.
In vitro cos cleavage 20 μl reactions containing 10 mM pUC119 DNA having a 234 bp I2 + , I2 re30-35 , I2 re18-50 , or cos2 insert, and purified terminase were done as reported previously [72]. Terminase concentration was calculated using the A 280 and the extinction coefficient of ε = 157.2 mM -1 cm -1 [73]. Reaction products were run on a 0.8% agarose gel and stained with ethidium bromide. The inserts contain the cosN cutting site. Terminase concentrations varied from 0 to 150 nM. The percent DNA cut at cos was determined by analysis of product bands with a computer imaging program. Product band intensity was measured on a Typhoon-8600 phosphoimager (Molecular Dynamics). Parallel reactions with 50 nM purified IHF, kindly given by Helene Gaussier and Carlos Catalano, were also analyzed.

In vivo cos cleavage assay
The in vivo cos cleavage assay was performed as described in "Induction of lysogens", this section [74]. Total DNA was isolated by extractions with phenol:chloroform:isoamyl once, with phenol two times, and then with chloroform once, followed by ethanol precipitation. DNA pellets were air-dried and then resuspended in 40 μl of 10 mM Tris. DNA (10 μl) was cut with AccI for 2 h at 37°C, heated at 70°C for 10 min to melt cohesive ends, applied to a 0.8% agarose for gel electrophoresis, and then transferred to GeneScreen Plus (New England Nuclear) membrane using a vacuum blotting transfer system (American Bionics). A PCR fragment of λ bp 177 to 2099 was labeled with DIG-11-dUTP (Roche) according to the supplier's instructions and then used to probe the L end (λ bp 1 to 2190) and J (λ bp 42921 to 2190) AccI fragments. Bands were detected by chemiluminescence (Pierce). The percent cos cleavage was calculated using phosphoimaging data, according to the following formula: (L end x 100)/(L end + J), where L end and J equal L end and J fragment brightness, resp. Brightness values were corrected for background. Phosphoimaging was done using the Image Reader LAS-1000 of the Typhoon 8600 photoimager (Molecular Dynamics).