Structural basis for DNA recognition andloading into a viral packaging motorCarina R. Büttnera,1, Maria Chechika, Miguel Ortiz-Lombardíaa,2, Callum Smitsa, Ima-Obong Ebongb, Victor Chechikc,Gunnar Jeschked, Eric Dykemane, Stefano Beninia,3, Carol V. Robinsonb, Juan C. Alonsof, and Alfred A. Antsona,4

aYork Structural Biology Laboratory, Department of Chemistry, University of York, York, YO10 5DD, United Kingdom; bDepartment of Chemistry,University of Oxford, South Parks Road, Oxford, OX1 3QZ, United Kingdom; cDepartment of Chemistry, University of York, York, YO10 5DD, UnitedKingdom; dEidgenössische Technische Hochschule Zürich, 8093 Zürich, Switzerland; eYork Centre for Complex Systems Analysis, Departments ofMathematics and Biology, University of York, York, YO10 5DD, United Kingdom; and fCentro Nacional de Biotecnología, Consejo Superior deInvestigaciones Cientificas, Darwin 3, 28049 Madrid, Spain

Edited by Roger W. Hendrix, Pittsburgh Bacteriophage Institute, Pittsburgh, PA, and accepted by the Editorial Board November 20, 2011 (received for reviewJune 29, 2011)

Genome packaging into preformed viral procapsids is driven bypowerful molecular motors. The small terminase protein is essen-tial for the initial recognition of viral DNA and regulates themotor’s ATPase and nuclease activities during DNA translocation.The crystal structure of a full-length small terminase protein fromthe Siphoviridae bacteriophage SF6, comprising the N-terminalDNA binding, the oligomerization core, and the C-terminal β-barreldomains, reveals a nine-subunit circular assembly in which theDNA-binding domains are arranged around the oligomerizationcore in a highly flexible manner. Mass spectrometry analysis andfour further crystal structures show that, although the full-lengthprotein exclusively forms nine-subunit assemblies, protein con-structs missing the C-terminal β-barrel form both nine-subunitand ten-subunit assemblies, indicating the importance of the Cterminus for defining the oligomeric state. The mechanism bywhich a ring-shaped small terminase oligomer binds viral DNA hasnot previously been elucidated. Here, we probed binding in vitroby using EPR and surface plasmon resonance experiments, whichindicated that interaction with DNA is mediated exclusively by theDNA-binding domains and suggested a nucleosome-like model inwhich DNA binds around the outside of the protein oligomer.

bacteriophage SPP1 ∣ DNA packaging ∣ virus assembly ∣X-ray crystallography

The virus genome in tailed dsDNA bacteriophages and in theevolutionarily related herpes viruses is packaged into a pre-

formed empty procapsid (1–3). A powerful ATP-fueled molecularmachine drives the DNA with a speed of up to 1;800 bp∕sthrough the portal protein embedded in a unique vertex of theicosahedral procapsid (2–4). The molecular motor usually con-sists of the small and large terminase proteins. The small termi-nase plays a dual role in virus particle assembly: It (i) recognizesviral DNA during the initiation of packaging and (ii) modulatesthe ATPase and nuclease activities of the large terminase duringDNA translocation (5, 6). After filling a procapsid, the rest of theDNA is then docked to the portal entrance of another procapsidwhere the process of DNA translocation is repeated (7).

X-ray structures have been determined for portal proteinsfrom bacteriophages φ29 (8), SPP1 (9), and P22 (10) and also forlarge terminases from bacteriophages T4 (6), RB49 (6), and SPP1(11). Three-dimensional information on small terminases islimited to the cryo-EM structure of phage P22 small terminase(12), the NMR structure of the DNA-binding domain of phageλ gpNu1 (13), and the crystal structure of phage Sf6 small termi-nase (14). In the absence of accurate three-dimensional data forall three motor components of one particular phage, mapping offunctional information to the structure and modeling molecularinteractions between individual components is challenging. Wehave addressed this issue by extending the structural informationon Bacillus subtilis bacteriophages SPP1 and SF6, two very closelyrelated viruses of the Siphoviridae family. Here we present five

X-ray structures for several different constructs of the SF6 smallterminase. We also present mass spectrometry data on oligomericstates of the small terminase. Structural observations on the full-length protein containing the N-terminal DNA-binding domains,together with DNA-binding data and normal mode analysiscalculations, suggest a model for packaging initiation in whichDNA-binding domains are adjusted to form a periodical nucleo-protein assembly.

ResultsStructure Determination. G1P, the small terminase of bacterioph-age SF6, comprises 145 residues (Fig. 1A). We first determinedthe crystal structure of the oligomerization core domain ofG1P, residues 53–120, in which the N-terminal DNA-binding andthe C-terminal β-barrel domains were truncated (G1P53–120). Thisstructure, determined at 1.85-Å resolution by using seleno-methionine-substituted protein (SI Materials and Methods andTable S1), revealed a ten-subunit assembly (Fig. 1 B and C) andwas used as a molecular replacement model to determine struc-tures of a nine-subunit assembly (1.68-Å resolution, Fig. 1 D–F)and a second crystal form of the ten-subunit assembly (2.19-Åresolution) formed by the same truncated construct (G1P53–120).A further two structures, both revealing nine-subunit assemblies,were determined for (i) a proteolytic fragment G1P65–141 (3-Å re-solution, Fig. 2A) containing the C-terminal β-barrel but missingthe N-terminal DNA-binding domains and (ii) the full-lengthprotein (approximately 4-Å resolution, Fig. 2 B and C and Fig. S1).An overview of all constructs used in various experiments is in SIMaterials and Methods.

Author contributions: C.R.B., M.C., M.O.-L., C.S., I.-o.E., V.C., G.J., E.D., S.B., C.V.R., J.C.A.,and A.A.A. designed research; C.R.B., M.C., M.O.-L., I.-o.E., V.C., G.J., E.D., S.B., C.V.R., andA.A.A. performed research; J.C.A. contributed new reagents/analytic tools; C.R.B., M.C., M.O.-L., C.S., I.-o.E., V.C., G.J., E.D., C.V.R., and A.A.A. analyzed data; and C.R.B., M.O.-L.,C.S., I.-o.E., V.C., E.D., C.V.R., and A.A.A. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission. R.W.H. is a guest editor invited by the EditorialBoard.

Freely available online through the PNAS open access option.

Data deposition: The atomic coordinates and structure factors have been deposited in theProtein Data Bank, www.pdb.org (PDB ID codes 3ZQM, 3ZQN, 3ZQO, 3ZQP, and 3ZQQ).1Present address: Department of Molecular Genetics, University of Toronto, 1 King’sCollege Circle, Toronto, ON, M5S 1A8, Canada.

2Present address: Architecture et Fonction des Macromolécules Biologiques, CentreNational de la Recherche Scientifique, Universités d’Aix-Marseille I and II, 13288 Marseillecedex 9, France.

3Present address: Bio-organic Chemistry Laboratory, Faculty of Science and Technology,Free University of Bolzano, Piazza Università, 1, 39100 Bolzano, Italy.

4To whom correspondence should be addressed. E-mail: fred@ysbl.york.ac.uk.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1110270109/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1110270109 PNAS ∣ January 17, 2012 ∣ vol. 109 ∣ no. 3 ∣ 811–816

BIOPH

YSICSAND

COMPU

TATIONALBIOLO

GY

Dow

nloa

ded

by g

uest

on

Apr

il 11

, 202

0

Oligomerization Core Domain. The conformation of individual sub-units is essentially the same in the nine-subunit and ten-subunitoligomeric states of the G1P53–120 construct (Fig. 1 B–D). A seg-ment of helix α4, present at the N terminus of this construct, andthe following peptide (residues 65–69) that link the N-terminalDNA-binding domain with the oligomerization part are locatedat the outer surface of the oligomer and do not contribute to sub-unit interactions. The main body of the oligomerization core,residues 70–120, is formed by two short antiparallel α-helices(α5 and α6) connected by a β-hairpin (β1β2) (Figs. 1B and 2D).This segment mediates multiple subunit-subunit interactionscomprising eleven hydrogen bonds, several hydrophobic contacts,and two salt bridges formed between residues His102/Glu104and Glu115/Lys112 (Fig. 1E and Table S2). Interactions are verysimilar for both oligomeric states, resulting in similar intersubunitburied surface areas of approximately 1;000 Å2. However, thetwo oligomeric states differ in the dimensions of the central chan-nel. The most constricted part of the oligomerization core do-main has a van der Waals diameter of approximately 11 Å in the9-mer and approximately 14.5 Å in the 10-mer (Fig. 1F). Thechannel reaches its maximum dimension in its middle part witha van der Waals diameter of approximately 29 Å in the 9-mer andapproximately 32 Å in the 10-mer.

The C-Terminal Segment Forms an Intersubunit β-Barrel.A proteolyticfragment, G1P65–141, comprising the C-terminal β-barrel and themain body of the oligomerization core domain, forms a nine-subunit assembly (Fig. 2A). In this structure, the N-terminal do-mains which were shown to be responsible for DNA binding (15,16) and are further on designated as DBDs (DNA-bindingdomains), were missing because of degradation occurring duringthe period required for crystal growth. However, the C-terminalsegment (residues 121–141), lacking in the previously crystallizedconstruct, is clearly visible in the electron density maps. The C-terminal segments of all subunits assemble together in a parallelβ-barrel with each subunit contributing one β-strand (β3, residues125–139). The internal van der Waals diameter of the β-barrel

varies from approximately 9.9 (measured for side-chain atomsof Gln134) to 14.7 Å (side-chain atoms of Thr132). The slopeof the β-strands with respect to the barrel axis is approximately55°, in good agreement with the theoretical value of 56.3° for thistype of β-barrel (17).

The N-Terminal DNA-Binding Domains Are Connected to the OligomerVia Flexible Linkers. The X-ray data for the full-length protein areanisotropic and complete only to approximately 6-Å resolution,although in directions perpendicular to the c axis they extend toapproximately 4 Å (Table S1). In this structure, shown in Fig. 2 Band C, the C-terminal β-barrel as well as the N-terminal DBDsof two out of three subunits in the asymmetric unit are defined inthe electron density maps (Fig. S1). The positions and orienta-tions of the two defined DBDs with respect to the oligomeriza-tion core domain differ markedly, with their centroids shiftedalong the oligomer axis by approximately 33 Å. The DBD of thethird subunit of the asymmetric unit has no defined electrondensity indicating variability in its position, which is permitted bythe lack of symmetry-related molecules in the vicinity of this do-main. Taken together, the structural observations indicate signif-icant variability in the positioning of DBDs.

To investigate further possible conformational changes in thesmall terminase and, in particular, the potential adjustments inthe orientation of the DBDs, we performed normal mode analysiscalculations. These showed low-energy modes corresponding toup-and-down movements of the DBDs (along the channel axis)as well as rotational adjustments (Movies S1 and S2). These dataindicate a large degree of flexibility in the orientation of the DBDs,consistent with the structural observations described above.

Full-Length Small Terminase Exclusively Forms Nine-Subunit Assem-blies Stabilized by the C-Terminal β-Barrel. To understand whetherthe nine-subunit state, observed for the full-length terminase,predominates in solution or whether it was selected from a mix-ture of different oligomeric states during crystallization, we per-formed mass spectrometry analysis of the full-length protein

Fig. 1. Structure of the oligomerization core domain (resi-dues 53–120). (A) Domain organization showing locationof the oligomerization core within the full-length protein.(B and C) Ribbon diagram of a ten-subunit assembly ofG1P53–120 shown in two orthogonal views. The N and C termi-ni as well as the secondary structure elements of a single sub-unit are labeled. “Linker” refers to residues 61–65 connectingα4 with the main body of the oligomerization core formed byhelices α5 and α6. (D) Ribbon diagram of a nine-subunit as-sembly of the same construct, G1P53–120, shown in three alter-nating colors. (E) Electrostatic surface potential at subunitinterface (ranging from −5, red, to þ5 kT∕e, blue) shownfor two opposing subunits with the central axis vertical asin B. (F) Channel cross section indicating internal (van derWaals) diameters. The red stars indicate the position of theMTSSL in the S106C mutant used.

812 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1110270109 Büttner et al.

Dow

nloa

ded

by g

uest

on

Apr

il 11

, 202

0

under conditions where subunit interactions are maintained (18,19). The only oligomer observed corresponded to the nine-sub-unit assembly (Fig. 3A). Partial microheterogeneity was observedat the C terminus of the protein, leading to peak splitting for the9-mer species (Fig. S2). We also used mass spectrometry to probethe oligomeric state of the protein in which the C-terminal seg-ment was removed (G1P1–120). Mass spectra for this truncatedform indicated the presence of both nine-subunit and ten-subunitspecies in an approximately 1∶1 ratio (Fig. 3B and Fig. S2).

These observations indicate that, although the C-terminal seg-ment is not essential for G1P oligomerization, it appears to sta-bilize the nine-subunit state, most probably through formation ofthe intersubunit β-barrel observed in the crystal structure. Thishypothesis is supported by the electron paramagnetic resonance(EPR) data (Table S3) showing that the construct G1P53–145 con-taining the C-terminal β-barrel forms predominantly oligomers.

Consistent with mass spectrometry analysis, variation of the sub-unit number in the crystal structures was observed only when theC-terminal β-barrel was truncated.

Comparison with Other Small Terminases. Sequence alignment of 33small terminases from phages closely related to SPP1 indicatedseveral conserved residues that are exposed on the molecularsurface (SI Materials and Methods and Figs. S3 and S4 B and C).One group of such residues (Lys5, Lys9, Arg12, and Gly32) isfound in the helix-turn-helix (HTH) motif involved in DNAbinding. Another group of conserved residues is present in thecentral channel (Glu87, Lys112, and Lys119). These residues areinvolved in electrostatic interactions that stabilize the oligomericassembly and may, potentially, also mediate interactions withDNA. No conserved residues can be found at the outer surfaceof the oligomerization core.

Structural comparisons with the small terminase of phage Sf6(14) belonging to the Podoviridae family and a putative small ter-minase from the Bacillus cereus prophage phBC6A51 [ProteinData Bank (PDB) ID code 2ao9] reveal that, despite a lack ofsequence similarity, the folds are conserved (Fig. 2D). In all threeproteins the central oligomerization motif formed by α-helicesα5 and α6 in G1P and the C-terminal β-barrel are conserved fea-tures. The fold of the DBD is also very similar (pairwise Cα rmsdof 1.8–3.1 Å for approximately 45 DBD residues) although itsposition with respect to the oligomerization core varies. G1P dif-fers from the other two terminases by the presence of an extendedβ-hairpin (β1∕β2 in Fig. 2D) that connects the conserved α-helicesα5 and α6. Interestingly, the structure of the T4-like bacterioph-age 44RR2 small terminase reported in the accompanying paper(20) also reveals an oligomerization motif comprising two α-helices, as indicated by earlier mutational analysis (21).

Previously, structural data were also obtained for the isolatedDNA-binding domain of the small terminase from bacteriophageλ (13). Structural comparisons show that, although SF6 (PDB IDcode 2cmp) and Sf6 (14) small terminases share the same fold intheir DBDs, it differs markedly from the fold observed in bacter-

Fig. 2. Structure of N-terminal deletion construct (residues 65–141) and full-length small terminase. (A) Ribbon diagram of the proteolytic fragmentG1P65–141 containing the C-terminal β-barrel in addition to the main bodyof the oligomerization core domain shown with the central axis vertical, withindividual subunits in alternating colors. The indicated van der Waals dia-meter of the β-barrel (9.8 Å) corresponds to the most constricted part ofthe channel. (B) Ribbon diagram of the full-length G1P shown along the axiswith DBDs in red. Two DBDs of the asymmetric part (DBD-1 and DBD-2), thatwere observed in the electron density maps, are shown as ribbons, and theputative position of the third DBD (DBD-3) is depicted by dashed circles. (C)Models of full-length G1P shownwith all DBDs either in DBD-1 (left) or DBD-2(right) orientation. For clarity, DBDs of only five subunits are shown, with a10-bp dsDNA-DBD complexmodeled only for one subunit. The rotational andtranslational DBD movements derived from the normal mode analysis areindicated on the left. (D) Stereo figure comparing superposed individual sub-units of small terminases from Siphoviridae SF6 phage (two adjacent subu-nits, red and pink), Podoviridae Sf6 phage (blue), and the putative smallterminase from prophage phBC6A51 (yellow). The structures were alignedto fit the position of the channel axis, the first of the oligomerization α-he-lices and the C-terminal β-strand. Although the position of the two helices isconserved in the oligomer, the fold of the monomer differs in the Podovir-idae Sf6 small terminase, where helix α6 occupies the same position as α6 ofan adjacent subunit in the small terminases from the other two phages.

Fig. 3. Mass spectrometry analysis. (A) Mass spectrum recorded for full-length G1P reveals a 9-mer centered at 6;000 m∕z (green triangles). G1P hassome minor C-terminal truncations (see Fig. S2), which arise from partial clea-vage of residues 128–145. To resolve these peaks, increased acceleration vol-tage is applied leading to dissociation intomonomers (lowm∕z of 1,000–2,000,brown circles). (B) Mass spectrum recorded for the construct G1P1–120 revealsthe formation of both 9-mers (green triangles) and 10-mers (purple hexagon)in an approximately 1∶1 ratio. Gray circles correspond to monomers.

Büttner et al. PNAS ∣ January 17, 2012 ∣ vol. 109 ∣ no. 3 ∣ 813

BIOPH

YSICSAND

COMPU

TATIONALBIOLO

GY

Dow

nloa

ded

by g

uest

on

Apr

il 11

, 202

0

iophage λ small terminase (Fig. S1). Although in all three cases,the polypeptide chain folds into four α-helices, in the case of λsmall terminase the HTH motif is formed by the first and secondα-helices, whereas in the other two terminases it is formed by thesecond and third α-helices. Unlike the SF6 and Sf6 terminases,where the recognition helix (α3) is almost immediately followedby another helix (α4), in bacteriophage λ small terminase therecognition helix (α2) is followed by an extended hairpin, neces-sitating classification of this fold as “winged HTH.” Fold differ-ences result in a very different orientation of the HTH motif withrespect to helix α4, which connects the DBD with the C-terminaloligomerization part of the protein (Fig. S1). So far, no structuraldata have been obtained for the oligomerization part of thebacteriophage λ small terminase, thereby preventing furtherstructural comparison, although analytical ultracentrifugationdata obtained for a complex of large and small λ terminases in-dicated the formation of macromolecular assemblies comprisingeight subunits of the small terminase (22), thus suggesting apotential similarity with the multisubunit assemblies found inother small terminases.

Although all five small terminases (44RR2, P22, SF6,phBC6A51, and Sf6), for which oligomerization domains havebeen resolved by structural analysis, form ring-shaped oligomers,they differ in their oligomeric state with subunit numbers rangingfrom eight to eleven. Differences in oligomeric state result in dif-ferent diameters of the internal channel.

DNA Binding by G1P is Exclusively Mediated by the DNA-Binding Do-mains. The central channel as well as the presence of multiple,peripheral DNA-binding domains per oligomer posed a questionabout potential modes of interaction with DNA.We hypothesizedthat, apart from interacting with the peripheral DBDs (Fig. 4A),DNA could be accommodated in the central channel duringtranslocation (Fig. 4B). This assumption is supported by thepresence of Lys/Arg rings in the central channel (Fig. S4A) andthe intriguing finding that the geometry of the G1P β-barrelmatches the geometry of B-form DNA. Indeed, the strand angleof the β-barrel matches the helical rise of double-stranded B-DNA (Fig. 4B, Inset). We probed interaction with the two poten-tial DNA-binding surfaces (the internal channel and the surfacearound the outside of the oligomer formed by DBDs) by surfaceplasmon resonance (SPR) and EPR.

SPR experiments were performed with a diverse range ofconstructs for G1P and the SPP1 recognition pac site DNA, asspecified in Materials and Methods. The strongest interactionwas observed for the 428-bp pac DNA (pac1, comprising pacL,pacC, and pacR) (Fig. 5A). Full-length G1P bound to this DNAwith a KD of 336� 4 nM and distinct association and dissociationphases. The binding affinity decreased when fragments of thepac site DNA were used (Table S4 and Fig. S5). Whereas the pre-sence of the C-terminal β-barrel did not increase the affinity sig-

nificantly, a significant reduction was observed for constructslacking the DBDs. We hypothesized that the residual weak bind-ing was due to nonspecific association of the DNA with basicresidues located at the outer surface of the oligomerization core,in contrast to binding in the central channel. Indeed, the mutationof solvent-exposed positively charged Lys86 (Fig. S4) resulted ina significant further drop in affinity (Table S4 and Fig. S5) ver-ifying our assumption of nonspecific binding. Therefore, the in-teraction with DNA as measured by SPR is predominantlydependent on the presence of the DBDs.

To distinguish whether the weak residual binding after remov-ing the DBDs was due to binding in the channel or whether it wasdue to nonspecific interaction with the positively charged ringspresent at the outer surface of the oligomerization domain, we

Fig. 4. Potential models for the G1P-DNA complex. (A) DNA binding aroundthe outside of the oligomer mediated by the DNA-binding domains. EachDBD in complex with a 10-bp oligonucleotide is reoriented so that DNA seg-ments can form a continuous molecule. (B) DNA binding in the channel.The Inset demonstrates the match between the geometry of the C-terminalβ-barrel and B-form DNA.

Fig. 5. DNA binding. (A) SPR experiments for full-length G1P and the pacDNA recognition site comprising 428 bp (pac1). Equilibrium dissociationconstant KD was estimated from a steady state analysis of three individualexperiments. (B) EPR experiments. Mims ENDOR spectra for G1P53–145S106C labeled with MTSSL at the inner channel surface as shown in Fig. 1F.The control experiment with 4-phosphonoxy-2,2,6,6-tetramethyl-piperidine-1-oxyl (dashed line) clearly shows phosphorous coupling, but no coupling wasobserved for the oligomer incubated with a 22-bp dsDNA (solid line).

814 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1110270109 Büttner et al.

Dow

nloa

ded

by g

uest

on

Apr

il 11

, 202

0

performed EPR experiments with the spin label attached in theinternal groove of the channel, where there is sufficient space toaccommodate the label without affecting the diameter of thechannel (Fig. 1F). Pulsed EPR measurements were performedby using the G1P53–145 S106C mutant that lacks the DNA-bindingdomains, labeled with the MTSSL (Materials and Methods). Ourmodeling indicated that if DNA is present in the channel, the dis-tance between the spin label and the nearest 31P nucleus of theDNA would be <1 nm. This distance is sufficiently short for thehyperfine interactions between the electron spin of the spin labelin the protein and the nuclear spin of the 31P nucleus in the DNAto be observed by electron nuclear double resonance (ENDOR).We used the Mims ENDOR pulse sequence at the Q band, whichwas shown to be most sensitive to longer electron-31P distances(23). Substoichiometrically labeled oligomers (containing approxi-mately one spin label per oligomer) were used to prevent spin-spin interactions between spin labels on adjacent subunits in theoligomer. The control experiment using a phosphorylated spinlabel (with ca. 0.6 nm nitroxide-31P distance) clearly showed phos-phorous coupling in the Mims ENDOR spectra with an excellentsignal-to-noise ratio (Fig. 5B). However, ENDOR spectra of sub-stoichiometrically labeled G1P oligomers incubated with a 22-bpSPP1 dsDNA oligonucleotide showed no detectable phosphorouscoupling.

The combination of SPR and EPR results demonstrates thata complex of G1P with DNA in the inner channel could not beassembled in vitro, at least under the conditions tested.

DiscussionSmall terminases play two crucial roles during DNA packaging.Initially, the small terminase binds to a specific recognition sitewithin the viral chromosome and recruits the large terminase.The assembled terminase-DNA complex then docks onto theprocapsid and the small terminase performs its second role, reg-ulating the enzymatic activities of the large terminase.

Previous models describing the interaction between the smallterminase and the pac site DNA have suggested that the DNAwraps around the circular oligomer. However, the presence ofthe central channel suggested that it could accommodate DNA(24). In this model of packaging initiation, the DNA would likelyinteract first with the DNA-binding domains and then be trans-located through the central channel of the small terminase. Thismode of DNA binding has been increasingly favored (2, 14, 24).However, the observed inner diameters of the eight-subunitassembly (14) and nine-subunit assembly (Figs. 1F and 2A) aretoo narrow to accommodate the DNA double helix. Moreover,small terminases from different phages differ in their oligomericstates, containing either eight [Sf6 (14)], nine [SF6, P22 (24),phBC6A51], or 11∕12 [44RR2 (20)] subunits, indicating that thenumber of subunits and consequently the diameter of the innerchannel are not critical for the function of the small terminase.Indeed, in the bacteriophage P22 small terminase, both nine- andten-subunit assemblies are functional for packaging in vivo (24).In agreement with these observations, the combination of ourSPR and EPR data presented here strongly indicate that theG1P-DNA interaction is mediated exclusively by DNA-bindingdomains and not by the inner channel.

We then sought to model the interaction between DNA andthe small terminase oligomer comprising multiple DNA-bindingdomains, as observed in SF6, Sf6, and phBC6A51. We did not useDNA-interaction data available for the bacteriophage λ smallterminase (13, 25) because of significant differences in DBD foldsand oligomeric states (dimeric species reported for λ are notobserved in the SF6, Sf6, and phBC6A51 small terminases) andbecause of DNA bending in λ involving an additional host factor(26). Interestingly, when modeling the DNA, guided by HTH-DNA complexes of closest structural homologues (27), the DNAorientation is roughly parallel to the oligomer axis. This mode of

interaction suggests that one DNA molecule could connect theDBDs of several small terminase oligomers. However, from themass difference accumulating to the chip surface in our SPRexperiments, we calculated that there was only about one G1Poligomer per approximately 100 bp DNA. Moreover, previousDNase footprinting experiments with G1P bound to pac DNArevealed nine protected sites within an approximately 100-nu-cleotide segment, one site every 10� 1 bp (every turn of theDNA helix); these sites were separated by DNase-sensitive re-gions (16). Taken together, these data indicate that in the com-plex an approximately 100-bp DNA fragment wraps around theprotein oligomer interacting with multiple HTHmotifs separatedby approximately 34 Å (16). The approximately 34-Å separationswould be in good agreement with the DBD separation observedin crystal structures and with their flexibility. This model suggeststhat small terminases with varying numbers of subunits would stillbe functional as long as the DBDs were appropriately spaced.

Our structure of full-length G1P has two out of three DNA-binding domains in the asymmetric unit defined, because no elec-tron density was observed for the third domain. The two DBDsfor which electron density was observed are in different positionsrelative to the oligomerization domain, which indicates signifi-cant flexibility in their position. The observed flexibility is furthersupported by normal mode analysis calculations. This flexibilitymay be essential for binding DNA because to bind DNA in theexpected circular/helicoidal orientation, each DBDmust reorientin order to match its HTH motif with the corresponding segmentof DNA. On the basis of the observed flexibility of the DBDpositions, we generated a model starting with the two extremeDBD positions defined in our crystal structure. We interpolatedthe positions of the other DBDs around the oligomer, adjustedthem to be equidistant from the oligomer and each other, andaligned the HTH motifs. Intriguingly, the two starting positionsprovide enough spatial separation to allow a helicoidal assemblywhere the DNA wraps completely around the oligomer (Fig. 6).In contrast, in a model with all DBDs adopting the same constantposition relative to the oligomerization domain, the DNA mole-cule cannot encircle the G1P oligomer without unrealistic dis-tortion.

The small terminase performs at least two distinct tasks duringDNA packaging: First, it assures the specific recognition of thepac DNA, and second, it regulates the enzymatic activities of thelarge terminase during DNA translocation, which requires thatthe small terminase is engaged in the molecular motor. We showthat the DNA recognition task involves the small terminase DBDsand that the DBDs are the dominant interaction site for the for-mation of a nucleoprotein complex. Our structure of full-lengthG1P, in which the DBD exists in two very different positions andorientations, together with normal mode analysis, DNA-bindingdata, and the DNase footprinting results (16) allowed us to model

Fig. 6. Model of the small terminase-DNA complex during DNA recognition.Positions of DBDs were remodeled so that the vertical (channel) coordinate ofthe two extreme positions corresponds to coordinates observed in the crystalstructure with all other DBDs occupying intermediate positions; orientationof each DBD was adjusted so that helix-turn-helix (HTH) motifs fit the heli-coidal DNA, with approximately 34-Å spacing between adjacent HTH motifs.

Büttner et al. PNAS ∣ January 17, 2012 ∣ vol. 109 ∣ no. 3 ∣ 815

BIOPH

YSICSAND

COMPU

TATIONALBIOLO

GY

Dow

nloa

ded

by g

uest

on

Apr

il 11

, 202

0

the interaction with DNA prior to recruiting the large terminase toperform the initial cut. We argue that the interaction must involveremodeling at the level of the tertiary structure and aligning theDBDs with the DNA to form a helicoidal, nucleosome-like assem-bly. After directing the DNA cleavage, the small terminase up-regulates ATP hydrolysis and thereby DNA translocation whileinhibiting nuclease activity (5, 28, 29). Despite the conserved over-all structure of the small terminase, no conserved residues werelocated on the exterior surface of G1P, indicating that interactionwith the large terminase or portal protein is mediated by the over-all architecture of the molecular motor and by the distinctive shapeof the small terminase.

Materials and MethodsProtein Production, Crystallization, and X-Ray Structure Determination. Thegene encoding full-length small terminase (G1Pfl, residues 1–145) of bacter-iophage SF6 was cloned, overexpressed, and purified as described in SIMaterials and Methods. Crystals of G1Pfl grew within 1 wk by vapor diffusionat 293 K from solution containing 30% Jeffamine 600 M and 0.1 M sodiumcacodylate, pH 6.0. The 4-Å–resolution data were collected at ID-14 beamlineat European Synchrotron Radiation Facility (ESRF). A single better-diffractingcrystal from the G1Pfl sample grew after 6 mo from solution containing 0.2 MMgCl2, 8% (wt∕vol) PEG 20,000, 8% (wt∕vol) PEG 550monomethyl ether, buf-fered with Tris·HCl, pH 8.0. The 3-Å–resolution data from this crystal werecollected in house by using a Rigaku RU200 X-ray generator with a rotatinganode. Crystals of G1P53–120-SeMet were obtained under several crystalliza-tion conditions: 0.1 M malonic acid, imidazole, and boric acid buffer, pH 7.0,25% (wt∕vol), PEG 1500 (crystal form #1); 0.2 M KCl, 50 mM Hepes, pH 7.5,35% (vol∕vol) pentaerythritol propoxylate (5/4 PO/OH) (crystal form #2) and2.7 M ðNH4Þ2SO4, 0.1 M Tris·HCl, pH 8.5 (crystal form #3). Single-wavelengthanomalous dispersion X-ray diffraction data at the selenium absorption edge(λ ¼ 0.9716 Å) for crystal forms #1 and #2 were collected at the I04 beam line(Diamond). Data for crystal form #3 were collected at BM14 (ESRF). Most crys-tallographic calculations were performed by using the CCP4 program pack-age (30). G1Pfl and G1P65–141 datasets were processed with MOSFLM/SCALA(Table S1). All other data were processed by using HKL2000 (31). A detaileddescription of structure determination is given in SI Materials and Methods.Coordinates have been deposited with the Protein Data Bank under acces-sion ID codes 3ZQM (G1P53–120, #1), 3ZQN (G1P53–120, #2), 3ZQO (G1P53–120,#3), 3ZQP (G1P65–141), and 3ZQQ (G1Pfl).

Mass Spectrometry. Thirty microliters of a solution containing G1Pfl orG1P1–120 were buffer-exchanged into 1 M ammonium acetate (pH 7.5) byusing Bio-Spin columns (Bio-Rad). MS analysis was conducted on a Q-ToF2mass spectrometer (Waters) modified for high mass detection and conserva-tion of noncovalent interactions between protein subunits (18, 19). Then 2 μLof this solution was introduced into the mass spectrometer from a gold-pla-ted capillary needle. Instrument parameters used during experiments werecapillary voltage, 1.7 kV; cone voltages, up to 110 V; extractor cone voltage,5 V; and collision energy, ranging from 30 to 100 V. Themass spectra recordedwere calibrated by using cesium iodide. All data were acquired and processedby using MassLynx v4.1 (Waters).

Surface Plasmon Resonance. On the basis of DNA-interaction regions identi-fied by DNAse footprinting of the G1P-pac DNA complex (16) with G1Pfound to bind to pacL and pacR but not to pacC, three different SPP1 pacsite variants were generated by PCR using the SPP1 genome as template(pac1: complete pac site—i.e., including all three functional sites pacL-pacC-pacR—and spans nucleotides −326 to þ108 relative to the cleavage sitein pacC, length 428 bp; pac2: ends after pacL, spans nucleotides −326 to −82,length 253 bp; pac3: starts after pacL and covers pacC and pacR, nucleotides−82 to þ108, length 190 bp). Further details are given in SI Materials andMethods.

EPR Measurements. The protein was spin labeled with 1-oxyl-2,2,5,5-tetra-methylpyrroline-3-methyl)-methanethiosulfonate (MTSSL) using a standardprotocol. Continuous wave-EPRmeasurements were carried out at room tem-perature by using an X-band spectrometer, and Mims ENDOR was carriedout by using a pulsed Q-band spectrometer at 50 K. Full details can be foundin SI Materials and Methods.

ACKNOWLEDGMENTS. We thank Bernie Strongitharm and Andrew Leech(University of York) and also John Butler and Tim Fagge (Biacore) for helpwith SPR measurements. Ralf Flaig and Martin Walsh (Diamond, Oxford) arethanked for help during data collection. We are grateful to Andrey Lebedev,Eleonor Dodson, Johan Turkenburg, and Guy Dodson (University of York) foruseful suggestions during data analysis. Ma Yun (University of York), EmmaCarter, and Damien Murphy (University of Cardiff) are thanked for help withEPR spectroscopy. This work was supported by the Welcome Trust (fellowship081916 to A.A.A.).

1. Rao VB, Feiss M (2008) The bacteriophage DNA packaging motor. Annu Rev Genet42:647–681.

2. Casjens SR (2011) The DNA-packaging nanomotor of tailed bacteriophages. Nat RevMicrobiol 9:647–657.

3. Earnshaw WC, Casjens SR (1980) DNA packaging by the double-stranded DNA bacter-iophages. Cell 21:319–331.

4. Smith DE, et al. (2001) The bacteriophage ϕ29 portal motor can package DNA againsta large internal force. Nature 413:748–752.

5. Camacho AG, Gual A, Lurz R, Tavares P, Alonso JC (2003) Bacillus subtilis bacteriophageSPP1 DNA packaging motor requires terminase and portal proteins. J Biol Chem278:23251–23259.

6. Sun S, et al. (2008) The structure of the phage T4 DNA packaging motor suggests amechanism dependent on electrostatic forces. Cell 135:1251–1262.

7. Black LW (1989) DNA packaging in dsDNA bacteriophages. Annu Rev Microbiol43:267–292.

8. Simpson AA, et al. (2000) Structure of the bacteriophage ϕ29 DNA packaging motor.Nature 408:745–750.

9. Lebedev AA, et al. (2007) Structural framework for DNA translocation via the viralportal protein. EMBO J 26:1984–1994.

10. Olia AS, Prevelige PE, Johnson JE, Cingolani G (2011) Three-dimensional structure of aviral genome-delivery portal vertex. Nat Struct Mol Biol 18:597–603.

11. Smits C, et al. (2009) Structural basis for the nuclease activity of a bacteriophage largeterminase. EMBO Rep 10:592–598.

12. Nĕmeček D, Lander GC, Johnson JE, Casjens SR, Thomas GJ, Jr (2008) Assembly archi-tecture and DNA binding of the bacteriophage P22 terminase small subunit. J Mol Biol383:494–501.

13. de Beer T, et al. (2002) Insights into specific DNA recognition during the assembly of aviral genome packaging machine. Mol Cell 9:981–991.

14. Zhao H, et al. (2010) Crystal structure of the DNA-recognition component of the bac-terial virus Sf6 genome-packaging machine. Proc Natl Acad Sci USA 107:1971–1976.

15. Chai S, Kruft V, Alonso JC (1994) Analysis of the Bacillus subtilis bacteriophages SPP1and SF6 gene 1 product: A protein involved in the initiation of headful packaging.Virology 202:930–939.

16. Chai S, Lurz R, Alonso JC (1995) The small subunit of the terminase enzyme of Bacillussubtilis bacteriophage SPP1 forms a specialized nucleoprotein complex with thepackaging initiation region. J Mol Biol 252:386–398.

17. Murzin AG, Lesk AM, Chothia C (1994) Principles determining the structure of beta-sheet barrels in proteins. I. A theoretical analysis. J Mol Biol 236:1369–1381.

18. Sobott F, Hernandez H, McCammon MG, Tito MA, Robinson CV (2002) A tandem massspectrometer for improved transmission and analysis of large macromolecular assem-blies. Anal Chem 74:1402–1407.

19. Hernandez H, Robinson CV (2007) Determining the stoichiometry and interactions ofmacromolecular assemblies from mass spectrometry. Nat Protoc 2:715–726.

20. Sun S, et al. (2011) The structure and function of the small terminase componentof the DNA packaging machine in T4 like bacteriophages. Proc Natl Acad Sci USA,doi: 10.1073/pnas.1110224109.

21. Kondabagil KR, Rao VB (2006) A critical coiled coil motif in the small terminase, gp16,from bacteriophage T4: Insights into DNA packaging initiation and assembly of packa-ging motor. J Mol Biol 358:67–82.

22. Maluf NK, Gaussier H, Bogner E, Feiss M, Catalano CE (2006) Assembly of bacterioph-age Lambda terminase into a viral DNAmaturation and packagingmachine. Biochem-istry 45:15259–15268.

23. Zänker P-P, Jeschke G, Goldfarb D (2005) Distance measurements between paramag-netic centers and a planar object by matrix Mims electron nuclear double resonance. JChem Phys 122:024515.

24. Nĕmeček D, et al. (2007) Subunit conformations and assembly states of a DNA-trans-locating motor: The terminase of bacteriophage P22. J Mol Biol 374:817–836.

25. Ortega ME, Catalano CE (2006) Bacteriophage lambda gpNu1 and Escherichia coli IHFproteins cooperatively bind and bend viral DNA: Implications for the assembly of agenome-packaging motor. Biochemistry 45:5180–5189.

26. Maluf NK, Yang Q, Catalano CE (2005) Self-association properties of the bacterioph-age lambda terminase holoenzyme: Implications for the DNA packaging motor. J MolBiol 347:523–542.

27. Tahirov TH, et al. (2002) Mechanism of c-Myb-C/EBP beta cooperation from separatedsites on a promoter. Cell 108:57–70.

28. Gual A, Camacho AG, Alonso JC (2000) Functional analysis of the terminase large sub-unit, G2P, of Bacillus subtilis bacteriophage SPP1. J Biol Chem 275:35311–35319.

29. Al-Zahrani AS, et al. (2009) The small terminase, gp16, of bacteriophage T4 Is a reg-ulator of the DNA packaging motor. J Biol Chem 284:24490–24500.

30. Collaborative Computational Project N (1994) The CCP4 suite: Programs for proteincrystallography. Acta Crystallogr D Biol Crystallogr 50:760–763.

31. Otwinowski Z, Minor W (1997) Processing of X-ray diffraction data collected in oscilla-tion mode. Methods Enzymol 276:307–326.

816 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1110270109 Büttner et al.

Dow

nloa

ded

by g

uest

on

Apr

il 11

, 202

0