doi:10.1016/j.jmb.2005.11.069 J. Mol. Biol. (2006) 356, 266–273

COMMUNICATION

Unlocking of the Filamentous Bacteriophage VirionDuring Infection is Mediated by the C Domain of pIII

Nicholas J. Bennett and Jasna Rakonjac*

Institute of MolecularBioSciences, Massey UniversityPalmerston North, 5320New Zealand

0022-2836/$ - see front matter q 2005 E

Abbreviations used: Amp, ampicilblue; Cm, chloramphenicol; EtBr, ethisopropyl b-D-1-thiogalactopyranosidof replication; PEG, polyethylene glysingle-stranded DNA; pIII, protein II

E-mail address of the correspondj.rakonjac@massey.ac.nz

Protein III (pIII) of filamentous phage is required for both the beginningand the end of the phage life cycle. The infection starts by binding of theN-terminal N2 and N1 domains to the primary and secondary hostreceptors, F pilus and TolA protein, respectively, whereas the life cycleterminates by the C-terminal domain-mediated release of the membrane-anchored virion from the cell. It has been assumed that the role of the C-terminal domain of pIII in the infection is that of a tether for the receptor-binding domains N1N2 to the main body of the virion.

In a poorly understood process that follows receptor binding, the viriondisassembles as its protein(s) become integrated into the host innermembrane, resulting in the phage genome entry into the bacterialcytoplasm. To begin revealing the mechanism of this process, we showedthat tethering the functional N1N2 receptor-binding domain to the virionvia termination-incompetent C domain abolishes infection. This infectiondefect cannot be complemented by in trans supply of the functional Cdomain. Therefore, the C domain of pIII acts in concert with the receptor-binding domains to mediate the post receptor binding events in theinfection.

Based on these findings, we propose a model in which binding of the N1domain to the periplasmic portion of TolA, the secondary receptor, triggersin cis a conformational change in the C domain, and that this change opensor unlocks the pIII end of the virion, allowing the entry phase of infection toproceed.

To our knowledge, this is the first virus that uses the same proteindomain both for the insertion into and release from the host membrane.

q 2005 Elsevier Ltd. All rights reserved.

Keywords: filamentous bacteriophage; phage infection; gene 3 protein; f1phage; phage display

*Corresponding author

Filamentous bacteriophages (Ff) are releasedfrom their host cells without lysis, by a processsimilar to the secretion of virulence factors or theassembly of surface filaments.1

F pilus-specific filamentous phage of Escherichiacoli (f1 or fd or M13; 98% identical) have been usedextensively (and interchangeably) as a modelsystem for protein translocation and insertion intothe membranes of Gram-negative bacteria.2–5 They

lsevier Ltd. All rights reserve

lin; BPB, bromphenolidium bromide; IPTG,e; ori, origincol; ssDNA,I; WT, wild-type.ing author:

have also been the principal workhorse of phagedisplay technology.6,7 Protein III (pIII), the subject ofthis work, is the virion protein most commonlyused as a platform for display of peptides andproteins.

The f1 virion is composed of five proteins. Themajor coat protein pVIII is present in about 2700copies. A helical array of overlapping pVIIIsubunits forms the tube-like structure that enclosesthe single-stranded (ss) DNA genome; each helicalturn contains five pVIII subunits.8,9 Four to fivecopies each of different pairs of minor coat proteins,pVII/pIX and pIII/pVI, are located at each end ofthe filament.10–12 Of the five virion proteins, four(pVIII, pVII, pIX and pVI) are short and hydro-phobic, whereas pIII is relatively large and mostlyhydrophilic.

d.

Unlocking the f1 Virion 267

pIII consists of three domains, N1, N2 and C (alsoreferred to as D1, D2 and D3), separated by flexibleglycine-rich linkers. The N-terminal domains N2and N1 bind to the primary phage receptor, the Fpilus, and its secondary receptor, the TolA protein,respectively.13–17 The N1N2 domains are tethered tothe virion via the C domain.

Binding to both of the f1 phage receptors has beenwell studied at the genetic, biochemical andstructural levels.13,15–18 Little is known, however,of the entry process itself. Similarly, there is a wealthof information about phage assembly-extrusion atthe genetic and biochemical level, and there is somestructural information on the exit port, but againmechanistic details of the assembly process are notclear. The two processes, infection and assembly,require different sets of accessory trans-envelopeprotein complexes, TolQRA and pIV/pI/pXI,respectively.14,19–24

On the basis of genetic and biochemical studies ofthe C terminus of pIII, we found that a rather smallpart of the pIII C terminus (83 residues) is incorpo-rated into the virion, but that a rather longer piece ofpIII (93 residues) is required for release of the phagefrom the infected cell.25 Protein pVI, which co-localizes with pIII at the same end of the virion, isalso required for release of the phage; both proteinsare also associated with the major coat protein pVIIIprior to assembly into the virion. We proposed thatupon incorporation into the virion, a conformationalchange in the C domain of pIII disrupts hydrophobicinteractions of the virion with the membrane, and inthis way catalyzes release of the phage.25 If thishypothesis is correct, then entry might be the reverseof assembly-extrusion, and a conformational changein the C domain of pIII might “unlock” the virion, i.e.expose hydrophobic membrane anchors of pIII, pVIand associated pVIII subunits, which would mediatethe insertion of the virion into the membrane. The twoprocesses, entry and release, may not necessarily besymmetrical in the terms of exact residues of the Cdomain involved, because of difference in accessoryproteins involved in the two processes, TolQRA14 andpIV/pI/pXI,26 respectively. Nevertheless, theinvolvement of the C-terminal domain of pIII inentry would implicate its active role in the infection,rather then simply linking the receptor-bindingdomains to the virion. Some earlier experimentsbear on this point. Krebber et al.27 appended thesequence encoding the single chain anti-fluoresceinantibody to the N terminus of a C-terminal domain ofpIII incorporated into the phage particle. To test aphage display method that aims to restore theintegrity of pIII by non-covalent antibody–antigeninteraction in vitro, they exposed the virions to thecognate antigen fluorescein conjugated to theC-terminal end of the N1N2 domain of pIII. However,the antigen–antibody interaction restored the infec-tion efficiency only up to 10K4 in comparison to thevirions that carried wild-type pIII. The same authorsinserted b-lactamase into the full-length pIII, betweenthe N1 and N2 or N2 and C-terminal domain of pIIIand found that the infectivity of the resulting virions

decreased about 100 to 1000-fold.27 Such trends wereobserved by other researchers developing similarphage display systems.28

Here, we show explicitly that the C-terminaldomain of pIII has an active and essential role in theinfection and propose a model for the post-receptorbinding steps of Ff phage infection.

Separation of infection and assembly/releasefunctions of pIII

Testing of the role of the C-terminal domain of pIIIin the infection required the production of phage inwhich the receptor-binding N1N2 domains (N) arecovalently linked to a short 83 residue C-terminalfragment (C83), which is incorporated into virionsbut is not competent to release phage from theinfected cell. This construct was named NC83(Figure 1(a), middle diagram). Since this C-terminalfragment cannot release the phage from the cells, asecond construct that lacked the N domains butexpressed a complete C domain, one that is fullycompetent to release phage (construct C), wasrequired for phage production (Figure 1(a), bottomdiagram). These two constructs, expressed from twoplasmids, were used to complement a phage thatlacks gIII entirely (f1d3;29 a flow-chart diagram of theexperiment is shown in Figure 1(b)). The resultingprogeny phage, which carried NC83 and C constructs(named NC83/C; Figure 1(b)) were then tested forinfectivity.

As a positive control, host cells that expressedfull-length (wild-type) pIII (WT) and C constructswere used to create particles with a mixture of WTand C (named WT/C; Figure 1(b)). As a negativecontrol, host cells carrying an empty vector (usedfor expressing NC83 and WT) and the same Cdomain construct were used to create particles withC only (named C; Figure 1(b)).

The complete C domain complements theassembly deficiency of the NC83 mutant

Ff virion length is the most significant indicator ofpIII-mediated release from the membranes. If pIII isnot functional, the virion filaments continue toelongate, and extremely long particles, carryingmore than 20 genomes, are formed.30,31 These areinitially attached to the cell; those found in thesupernatant represent filaments broken off bymechanical shearing. When subjected to nativevirion agarose electrophoresis, these very longfilaments appear as a slow-migrating smear. Todetermine whether the complete C domain of pIIIcomplemented the release deficiency of NC83, thelength of the virions obtained after the f1d3infection of NC83 only-expressing cells wascompared to those from cells expressing bothNC83 and C (Figure 2(a)). A pattern typical ofdefect in phage release was observed in the samplesassembled in the presence of NC83 only, and thenegative control, cells lacking pIII altogether(Figure 2(a), lanes 1 and 2). In contrast, the native

(b)

WT

(a)M

C

NC83

SS G GN1 N2 C

C83∆

pBADHis

C

pVIIIC83

C

pVIII

N1N2

WT/C CNC83/C

CN1N2

WT

Strain K1762Before infection

Released virions were•concentrated by PEG-precipitation•titered on K1762•quantified by densitometry

psp-gIII f1d3

+

pNJB3 f1d3

+

pJARA24

+C NC83

pNJB4 f1d3

+

pJARA24

+C

f1d3

+

pJARA24

+C

psp-gIII

Strain K1762After infection

No pIII production before infection

Phage infection-induced pIIIexpression allows multiple cycles of f1d3 infection, required to form plaques

f1d3 stock growth and plaque formation

Production of the WT/C, NC83/C and C virions

Infection

psp-gIII induciton

Assembly

and release

Virions in the stock and plaques of f1d3 carry WT pIII (encoded by the plasmid), but lack gIII

C

Figure 1. Schematic represen-tation of experimental design.(a) Diagrams of pIII variants usedin this work: top, wild-type pIII(WT); middle, NC83 mutant,which carries an internal deletionof 70 base-pairs, leaving the 83 C-terminal residues fused to the N1,N2 domain and the second gly-cine-rich linker. This mutant gIIIwas created by overlap-extensionPCR.39 Bottom, C domain of pIII(C). SS, signal sequence; N1, N1domain (green); G, glycine-richlinkers (gray); N2, N2 domain(blue); C, C domain (red); M,transmembrane helix (red); pVIII,major coat protein (black).(b) Flow-chart diagram of theexperiment and schematic rep-resentation of the pIII end of thevirions. Only three pIII molecules,instead of the predicted four orfive per virion, are shown forsimplicity; also schematized ispVIII arrangement in the virion.Top panel: strain K1762 was usedfor production of stocks and titer-ing of f1d3 phage. This straincarries the complementing plas-

mid pJARA112,29,31 in which theexpression of WT gIII is inducedby f1d3 phage infection. As pIII N1and N2 domains render host cellsresistant to infection,40,41 onlyphage-induced expression of gIIIallows multiple rounds (cascades)of infection cycles, starting from asingle infected cell within a bac-terial lawn, required for plaqueformation. Bottom panel: Pro-duction of the virions. The pIIImolecules containing N1N2domains (WT pIII and NC83)

were expressed from plasmids pNJB3 and pNJB4, respectively. Their expression was controlled by an araBAD promoterof the vector pBADHis (Invitrogen; ColEI ori; Ampr). The araBAD promoter was chosen because of its tight regulation(N1N2 domains, even when expressed at a low level, render cells resistant to f1 phage infection). Nevertheless, theinfectivity of cells before araBAD induction was reduced (50% for pNJB3 and 14% for pNJB4) in comparison to thepBADHis vector control (95%). The C domain of pIII was expressed from the tac promoter of a compatible plasmidpJARA2425 (ColD ori; CmR). The host strain was K1822 (MC1061, (hsdR, mcrB, araD139, D(araABC-leu)7697 lacI74, galV,galK, strA, thi), FC). Virions were obtained by infection of exponentially growing cells (A600 nmZ0.1) carryingappropriate plasmids in FGB (Fortified Glucose Broth; 25 g of Bacto tryptone, 7.5 g yeast extract, 6.0 g NaCl, 2 g ofglucose per l, 0.05 M Tris–HCl (pH 7.6)), 60 mg/ml of Amp, 25 mg/ml of Cm, with f1d3 (moi 100). Infection proceeded for60 min, after which the cells were washed twice with FAB medium (FGB which contained 2 g/l of arabinose instead ofglucose) to remove the infecting f1d3 phage and replace glucose with arabinose, the inducer of the pBAD promoter,which controls the production of NC83 and WT pIII. The medium also contained Amp, Cm and 0.1 mM IPTG (forinduction of the C domain expression). The resuspended culture was incubated for another 4 h at 37 8C. When theincubation was complete, virions were separated from the cells by centrifugation and precipitated in 5% (w/v) PEG and0.5 M NaCl. PEG-precipitated virions were collected by centrifugation and the pellet was dissolved in 1/100 vol. of TENbuffer (10 mM Tris (pH 7.6), 2 mM EDTA, 150 mM NaCl).

268 Unlocking the f1 Virion

virion electrophoresis of all samples carrying thecomplete C domain of pIII and/or WT pIII: WT,NC83/C, C and WT/C, have a very similar sizedistribution (Figure 2(a), lanes 3–7). These samplescontained monomeric, but also somewhat longervirions carrying two or more genomes sequentially

packaged into the same filament, which formed aladder of bands. This size distribution is typical forvirions generated by complemented f1d3 infection(Figure 2(a), lane 7).25 In conclusion, a complete Cdomain was required to complement the deficiencyof NC83 in the termination and release of phage.

(b)

pIII NC83

1 2 3 4

pIII WT

pIII C

48

36

24.5

19.0

13.5

62.4

1 2 3

MW(kD)

FreessDNA

(a)

pVI(c)

pIII-terminatedvirions

Broken-offfilaments

4 5 6 7 8

Figure 2. Characterization of the virions. (a) Nativevirion electrophoresis. Virions were produced by the cellsexpressing: lane 1, NC83 only (pNJB4); 2, no pIII (vectorsonly); 3, WT (pNJB3); 4, NC83/C (pNJB4/pJARA24); 5, C(pBADHis/pJARA24); 6, WT/C (pNJB3/pJARA24); 7,WT (pJARA112); 8, purified f1d3 ssDNA (40 ng). Phagesamples were loaded onto a 0.6% (w/v) agarose gel inDNA loading buffer (1!TAE, 5% (v/v) glycerol, 0.25%(w/v) bromophenol blue).42 Electrophoresis was carriedout at 2 V/cm in TAE buffer (40 mM Tris (pH 8.3), 1 mMEDTA, 2 mM acetic acid). The position of the virionbands was revealed after alkaline-mediated in situdisassembly of the virions in the gel and staining withEtBr.30 (b) Western blot. pIII was detected using anaffinity-purified serum raised against the C-terminaldecapeptide.29 Lanes: 1, NC83/C (3.2!1010 genomeequivalents (ge)); 2, WT/C (3.2!1010 ge); 3, C (4.6!1011 ge); 4, f1 (2.5!1010 pfu). (c) Detection of pVI in thevirions using the anti-pVI antibody.43 Lanes: 1, NC83/C(2.0!1010 ge); 2, WT/C (2.0!1010 ge); 3, C (2.0!1010 ge);4, f1 (1.17!109 ge). The virions were quantified asdescribed in footnote b of Table 1.

Unlocking the f1 Virion 269

NC83 is incorporated into virions

For the investigation of the role of the C domainin infectivity, it was important to demonstrate thatthe NC83 protein fragment is incorporated intovirions efficiently. The incorporation of NC83 into

virions was monitored by Western blotting(Figure 2(b), lane 1). The NC83 band was clearlydetected (Figure 2(b)). The incorporation of NC83into NC83/C particles was about as efficient as theincorporation of WT pIII into WT/C particles(Figure 2(b), lanes 1 and 2). The antibody used todetect pIII molecules was raised against theC-terminal decapeptide of pIII29 and it does notrecognize pIII mutants in which the C terminus isaltered or blocked.25 Therefore, this experiment alsoshowed that the C terminus of the NC83 is intactand hence the membrane anchor, which may playthe crucial role in the insertion of the virion into thehost membrane during infection, is intact in NC83/C virions. The presence of pVI, which is required forphage release and forms a complex with pIII in thevirion25,32 was also examined in the NC83/Cparticles by Western blotting. NC83/C virionscarried the similar amount of pVI as did WT/Cand C virions (Figure 2(c)).

A functional C domain is required for N1N2-mediated f1 infection

To measure the ability of NC83/C virions toinfect host cells, their capacity to initiate plaqueformation was determined. The indicator strainK1762 carries phage infection-inducible wild-typegIII on a plasmid and permits plaque formation byDgIII phage31 (Figure 1(b), top panel).

The f1d3 stock that was used to prepare the virionsamples necessarily was grown on the strain K1762that provides pIII in trans (Figure 1(b), top panel).Inevitably, these three phage preparations containsome contaminating particles from the stock, whichcarry wild-type pIII. This background was w10K5

as determined by the infectivity of virions carryingonly the C domain (Table 1, third data column,second row). The infectivity of the WT/C samplewas 2500 times higher (Table 1, third data column,third row). In contrast, providing NC83 in trans onlyincreased infectivity fivefold. To achieve a lowerbackground of infectious particles, a phagemid-producing setup was used in which plasmidpJARA24 was replaced with a phagemid producingC-domain of pIII (Table 1, footnote g). In this setup,the phagemid particles instead of phage wereproduced by the cells carrying WT/C, NC83/C orC, upon the infection of the DgIII helper phageR408d3. These phagemid-carrying particles, whichharbored CmR marker, were tested for infectivity bytransduction of the CmR marker to a CmS FC host(K1822; Table 1, three lower data rows). In thissetup, the NC83/C background infectivity was aslow as that of the virions that only had the Cdomain of pIII (Table 1, third data column, fifthrow). In contrast, infectivity of the WT/C-contain-ing phagemid particles was 107 times higher. Theresidual background of infectious NC83/C and Cphagemid particles was therefore much lower thanthat obtained in the phage experiment. This back-ground may be due to either low frequency of gIIIC

recombinants in the stock of the helper phage

Table 1. Infectivity of the NC83/C, WT/C and C particles

pIII moleculesin the virions

Number of plaque-forming or transducing

particles (pfu/ml ortdp/ml)a

Quantityof virions(ge/ml)b

Infectivity(pfu/ge ortdp/ge)c

Relativeinfectivityd

Efficiency of Fpilus-independent

transduction(tdp/ml)e

PhageNC83/C 7.0G1.3!108 1.0G0.2!1013 7.0!10K5 2.1!10K3 N/AC 2.7G0.5!108 2.1G0.9!1013 1.3!10K5 4.0!10K4 N/AWT/C 8.5G2.5!1011 2.6G0.9!1013 3.3!10K2f 1 N/AWT 5.5G0.9!1012 8.4G1.3!1013 5.8!10K2 1.8 N/A

Phagemidg

NC83/C 2.4G0.9!104 4.6G1.7!1012 5.2!10K9 1.6!10K7 !10C 2.2G0.5!105 1.2G0.3!1013 1.8!10K8 5.3!10K7 !10WT/C 5.7G2.5!1010 2.8G1.7!1012 2.0!10K2f 0.6 1.89G0.05!104

a The numbers were determined from triplicate plates, each containing 200–300 plaques (phage) or CmR transductants (phagemid).The values in the column represent the average and standard deviation of the number of the plaque-forming units (pfu) or transducingunits (tdu) per ml of the concentrated sample. The indicator strains for pfu and tdu were K1762 and K1822, respectively.

b The phage or phagemid were quantified based on the amount of DNA packaged in the particles. Virions were first disassembled byincubation in DNA loading buffer containing 1% SDS for 20 min at 70 8C, followed by agarose gel electrophoresis, staining with EtBr anddensitometric analysis of the bands. Because of the non-linear relationship between band density and the amount of DNA, for everyquantification gel a series of twofold dilutions of purified f1d3 ssDNA (the concentration of which was known) was loaded, and astandard curve was determined. The mass of DNA in the samples (in ng) was determined and converted to the number of genomes,based on the molecular mass of the phage f1d3 or phagemid pDJ7 plus strand. Genome equivalent (ge) corresponds to one virioncarrying one genome, or to one genome length of the virion that carries multiple genomes. Each value was obtained from fourdensitometric measurements.

c Infectivity is defined here as a ratio of pfu or tdu to the amount of phage or phagemid DNA, respectively, expressed as genomeequivalents (ge), in the phage or phagemid preparation.

d The relative infectivity is a ratio of the infectivity of the NC83/C, C or WT particles to that of the positive control, WT/C.e The efficiency of F pilus-independent transduction was determined on FK negative strain MC1061 (parent of K1822), using

the method of Russel et al.33.f The infectivity of the WT/C virions is much below 1. However, the infectivity of the f1d3 virions, which carry exclusively WT pIII, is

similar to that of the WT/C virions. Therefore, the observed infectivity of the WT/C is not due to the interference by truncated Csubunits devoid of the receptor binding sites. Rather, it is characteristic of complemented f1d3 stocks, which always have a largepopulation of long particles containing multiple genomes, effectively decreasing the pfu/ge value.

g To generate the transducing particles (or phagemid-containing particles) an experimental setup similar to the one shown inFigure 1(b) was used, except that the strain used to generate the virions carried, instead of plasmid pJARA24, a phagemid (pDJ7; D.Jankovic & J.R., unpublished) that encoded for CmR and also expressed C domain of pIII under the control of phage infection-inducedphage-shock protein (psp) promoter. Furthermore, instead of f1d3, an isogenic interference-resistant helper phage, R408d3,29 was used toavoid the interference between the phage and phagemid origins of replication. Due to identical origins of replication, phagemid (pDJ7)and plasmid (pBADHis, pNJB3 or pNJB4) replicating in the same cell were incompatible. However, because they are maintained in thecell by selection with different antibiotics, Cm and Amp, respectively, both phagemid and plasmid were retained in 10–50% of the cellsin the culture at the end of the phagemid-producing experiment.

270 Unlocking the f1 Virion

R408d3 or a very poor efficiency recycling of theWT pIII molecules from the infecting phageparticles into the newly assembled phagemidparticles. Alternatively, it was possible that thebackground represents low efficiency infectivitytypical of the F pilus-independent infection.33 Toinvestigate the F pilus-independent infection, an Fpilus-negative indicator strain (MC1061, isogenic toK1822) was used. If the observed NC83/C back-ground level of infectious phagemid particles wasdue to F pilus-independent infection, one wouldexpect the transduction efficiency of these virionsto an F pilus-negative strain to be equal to that ofWT/C phagemid particles. However, the transduc-tion efficiency of the NC83/C particles was at leastfour orders of magnitude lower than that of theWT/C particles (Table 1, fifth data column). There-fore, low infectivity of the NC83/C is not due to theF pilus-independent infection. Consequently, theinfection by NC83/C particles is blocked, asexpected, at a step of infection after the N2 domainbinding to the F pilus.

The failure of NC83/C particles to infect hostcells is unlikely to be due to misfolding of thereceptor-binding N1 and/or N2 domains in the

NC83 protein, because these two domains arewell known to fold independently of the Cdomain.15–18,34 In summary, simple tethering ofthe N1N2 domains to the virion via a truncated Cdomain is not sufficient for f1 infection. Rather, acomplete C domain covalently linked to the N1N2domain is necessary for the f1 virion to be opened orunlocked during infection.

Model of filamentous phage infection

Reichmann & Holliger17 have shown that inter-action of the N2 domain of pIII with the F pilusexposes the N1 domain, which then binds to thedomain III of the secondary receptor, TolA. Further-more, Click & Webster14 have shown that two otherhost proteins, which together with TolA form aninner membrane–periplasmic complex TolQRA, arerequired for virion integration into the hostmembrane. We propose a model (Figure 3) for themulti-step process that takes place after receptorbinding in which the f1 virion is opened orunlocked: Step 1, interaction of the N1 domainwith TolA generates a signal that is transmitted incis to the C domain of the same pIII molecule. Step 2,

C domain remains closed,no insertion

pVI

(a)

(b)

WT/Cvi rion

NC83/C vi rion

pVIII

TolA

1 2 3 4

1 1A

OM

IMDNATolQRA

WT pIIIIC

NC83

C83

Figure 3. Model of the Ff virion opening during infection. (a) Productive infection with the WT/C particles; (b)receptor binding, but no infection with NC83/C particles. OM, outer membrane; IM, inner membrane. The symbols forpIII domains and pVIII are the same as in Figure 1; pVI, light blue; pVII, gray; pIX, purple; TolA and TolRQ, brown.

Unlocking the f1 Virion 271

the conformational change of the C domain whichuncovers the hydrophobic anchor of pIII (andperhaps those of the associated virion proteinspVI and pVIII) is either propagated to the other pIIIsubunits in the cap complex or the conformationalchange of a single pIII subunit is sufficient. Step 3,exposed hydrophobic membrane anchors of pIII(pVI, pVIII) insert into the inner membrane, therebyanchoring the virion. Step 4, insertion of thesehydrophobic anchors triggers integration of adja-cent major coat protein into the membrane. Thisultimately leads to release of the ssDNA into thehost cytoplasm. Unlocking of the virion, membraneinsertion of the virion proteins and DNA transloca-tion into the cytoplasm in step 3 and/or step 4may be carried out with the assistance of thetransmembrane portion of TolQRA complex,14

which is also known to assist the translocationand membrane insertion of pore forming colicins(e.g. colicin A35).

In NC83/C virions, the defective C domain isincapable of undergoing the conformational changerequired to open the virion and/or to communicatethe interaction with the host receptor to other Csubunits in the virion cap or TolQRA trans-membrane complex, preventing entry (Figure 3(b),step 1A).

The C domain of pIII is able to catalyze bothintegration and release of the virion from thebacterial membranes. The direction of the reactionis likely to be determined by the accessory proteins.The assembly machinery (pI/pXI/pIV) undoubt-

edly drives the membrane release reaction, whichcompletes and “locks” the virion at the end ofassembly; the N1N2 domains of pIII are notrequired for that process. In contrast, the innermembrane complex TolQRA and interaction of theappropriately positioned N1 with TolA domain IIIare required for infection, but not for assembly andrelease. TolA domain III, which interacts with theN1 domain of pIII, is located at a set distance fromthe outer membrane, determined by the length ofthe helical TolA domain II. Shortening of thedomain II, which brings domain III closer to theinner membrane, does not increase infectivity; onthe contrary, it decreases the efficiency of infectionand integration of pVIII into the host membrane,suggesting that simple bringing of the C domaininto the proximity of the inner membrane is not themechanism by which TolQRA mediates the virioninsertion into the membrane and argues in favor ofour model.14

Given the different transmembrane machineriesused in the infection and assembly, and yet to beunderstood subtleties of the mechanism of TolQRAcomplex action in the entry step of phage infection,it is possible that the portions of C domain requiredfor the two processes are not identical. For furtherinvestigation of the mechanism of entry andsymmetry of the C domain requirements in thisprocess versus phage release, it would be necessaryto undertake a comprehensive mutagenesis of the Cdomain and test in parallel both infectivity andphage release by these mutants.

272 Unlocking the f1 Virion

If the model described above is correct, then themechanism of infection used by filamentous phageis similar to those eukaryotic viruses in whichinteraction with a secondary receptor triggers aconformational change that, in turn, exposes afusogenic peptide that inserts into the hostmembrane.36–38 However, unlike the eukaryoticviruses, which are released from the membranesby a process similar to the membrane vesiclebudding, filamentous phages, uniquely, use thesame protein domain for insertion into the mem-brane and for the reverse event, release from themembrane.

Acknowledgements

We are indebted to Marjorie Russel and PeterModel for advice, generously providing materialsand for the critical reading of the manuscript. Wethank Qing Deng for excellent technical assistanceand David Sheerin for comments on the manu-script. This work has been supported by theMarsden Fund of the Royal Society of NewZealand, contract number 02-MAU-210.

References

1. Webster, R. E. (2001). Filamentous phage biology. InPhage Display: A Laboratory Manual (Barbas, C. F. I.,Burton, D. R., Scott, J. K. & Silverman, G. J., eds), pp.1.1–1.37, Cold Spring Harbor Laboratory Press, ColdSpring Harbor, New York.

2. Chang, C. N., Model, P. & Blobel, G. (1978). Detectionof prokaryotic signal peptidase in Escherichia colimembrane fraction: endoproteolytic cleavage ofnascent f1 pre-coat protein. Proc. Natl Acad. Sci.USA, 75, 361–365.

3. Davis, N. G. & Model, P. (1985). An artificial anchordomain: hydrophobicity suffices to stop transfer. Cell,41, 607–614.

4. Russel, M. & Model, P. (1982). Filamentous phage pre-coat is an integral membrane protein: analysis by anew method of membrane preparation. Cell, 28,177–184.

5. Samuelson, J. C., Chen, M., Jiang, F., Moller, I.,Wiedmann, M., Kuhn, A. et al. (2000). YidC mediatesmembrane protein insertion in bacteria. Nature, 406,637–641.

6. Rodi, D. J. & Makowski, L. (1999). Phage-displaytechnology–finding a needle in a vast molecularhaystack. Curr. Opin. Biotech. 10, 87–93.

7. Zwick, M. g., Shen, J. & Scott, J. K. (1998). Phage-displayed peptide libraries. Curr. Opin. Biotech. 9,427–436.

8. Glucksman, M. J., Bhattacharjee, S. & Makowski, L.(1992). Three-dimensional structure of a cloningvector. X-ray diffraction studies of filamentousbacteriophage M13 at 7 A resolution. J. Mol. Biol.226, 455–470.

9. Wen, Z. Q., Overman, S. A. & Thomas, G. J., Jr (1997).Structure and interactions of the single-stranded

DNA genome of filamentous virus fd: investigationby ultraviolet resonance Raman spectroscopy. Bio-chemistry, 36, 7810–7820.

10. Grant, R. A., Lin, T.-C., Konigsberg, W. & Webster, R. E.(1981). Structure of the filamentous bacteriophage f1.Location of the A, C, and D minor coat proteins. J. Biol.Chem. 256, 539–546.

11. Gray, C. W., Brown, R. S. & Marvin, D. A. (1981).Adsorption complex of filamentous fd virus. J. Mol.Biol. 146, 621–627.

12. Lopez, J. & Webster, R. E. (1982). Minor coat proteincomposition and location of the A protein inbacteriophage f1 spheroids and I-forms. J. Virol. 42,1099–1107.

13. Click, E. M. & Webster, R. E. (1997). Filamentousphage infection: required interactions with the TolAprotein. J. Bacteriol. 179, 6464–6471.

14. Click, E. M. & Webster, R. E. (1998). The TolQRAproteins are required for membrane insertion of themajor capsid protein of the filamentous phage f1during infection. J. Bacteriol. 180, 1723–1728.

15. Lubkowski, J., Hennecke, F., Pluckthun, A. &Wlodawer, A. (1998). The structural basis of phagedisplay elucidated by the crystal structure of theN-terminal domains of g3p. Nature Struct. Biol. 5,140–147.

16. Lubkowski, J., Hennecke, F., Pluckthun, A. &Wlodawer, A. (1999). Filamentous phage infection:crystal structure of g3p in complex with its coreceptor,the C-terminal domain of TolA. Struct. Fold. Des. 15,711–722.

17. Reichmann, L. & Holliger, P. (1997). The C-terminaldomain of TolA is the coreceptor for filamentousphage infection of E. coli. Cell, 90, 351–360.

18. Holliger, P., Riechmann, L. & Williams, R. L. (1999).Crystal structure of the two N-terminal domains ofg3p from filamentous phage fd at 1.9 A: evidence forconformational lability. J. Mol. Biol. 288, 649–657.

19. Linderoth, N. A., Simon, M. N. & Russel, M. (1997).The filamentous phage pIV multimer visualized byscanning transmission electron microscopy. Science,278, 1635–1638.

20. Marciano, D., Russel, M. & Simon, S. M. (1999). Anaqueous channel for filamentous phage export.Science, 284, 1516–1519.

21. Marciano, D. K., Russel, M. & Simon, S. M. (2001).Assembling filamentous phage occlude pIV channels.Proc. Natl Acad. Sci. USA, 98, 9359–9364.

22. Opalka, N., Beckmann, R., Boisset, N., Simon, M. N.,Russel, M. & Darst, S. A. (2003). Structure of thefilamentous phage pIV multimer by cryo-electronmicroscopy. J. Mol. Biol. 325, 461–470.

23. Russel, M. (1993). Protein–protein interactions duringfilamentous phage assembly. J. Mol. Biol. 231, 689–697.

24. Russel, M., Linderoth, N. A. & Sali, A. (1997).Filamentous phage assembly: variation on a proteinexport theme. Gene, 192, 23–32.

25. Rakonjac, J., Feng, J.-n. & Model, P. (1999). Filamen-tous phage are released from the bacterial membraneby a two-step mechanism involving a shortC-terminal fragment of pIII. J. Mol. Biol. 289,1253–1265.

26. Feng, J.-n., Model, P. & Russel, M. (1999). A trans-envelope protein complex needed for filamentousphage assembly and export. Mol. Microbiol. 34,745–755.

27. Krebber, C., Spada, S., Desplancq, D., Krebber, A.,Liming, G. & Pluckthun, A. (1997). Selectively-

Unlocking the f1 Virion 273

infective phage (SIP): a mechanistic dissection of anovel in vivo selection for protein–ligand interactions.J. Mol. Biol. 268, 607–618.

28. Duenas, M. & Borrebaeck, C. A. K. (1994). Clonalselection and amplification of phage displayedantibodies by linking antigen recognition and phagereplication. Biotechnology, 12, 999–1002.

29. Rakonjac, J., Jovanovic, G. & Model, P. (1997).Filamentous phage infection-mediated geneexpression: construction and propagation of the gIIIdeletion mutant helper phage R408d3. Gene, 198,99–103.

30. Crissman, J. W. & Smith, G. P. (1984). Gene-III proteinof filamentous phages: evidence for a carboxyl-terminal domain with a role in morphogenesis.Virology, 132, 445–455.

31. Rakonjac, J. & Model, P. (1998). The roles of pIII infilamentous phage assembly. J. Mol. Biol. 282, 25–41.

32. Gailus, V. & Rasched, I. (1994). The adsorption proteinof bacteriophage fd and its neighbour minor coatprotein build a structural entity. Eur. J. Biochem. 222,927–931.

33. Russel, M., Whirlow, H., Sun, T. P. & Webster, R. E.(1988). Low-frequency infection of FK bacteria bytransducing particles of filamentous bacteriophages.J. Bacteriol. 170, 5312–5316.

34. Martin, A. & Schmid, F. X. (2003). The foldingmechanism of a two-domain protein: folding kineticsand domain docking of the gene-3 protein of phagefd. J. Mol. Biol. 329, 599–610.

35. Bouveret, E., Journet, L., Walburger, A., Cascales, E.,Benedetti, H. & Lloubes, R. (2002). Analysis of the

Escherichia coli Tol-Pal and TonB systems by periplas-mic production of Tol, TonB, colicin, or phage capsidsoluble domains. Biochimie, 84, 413–421.

36. Furuta, R. A., Wild, C. T., Weng, Y. & Weiss, C. D.(1998). Capture of an early fusion-active conformationof HIV-1 gp41. Nature Struct. Biol. 5, 276–279.

37. Sullivan, N., Sun, Y., Sattentau, Q., Thali, M., Wu, D.,Denisova, G. et al. (1998). CD4-Induced conformation-al changes in the human immunodeficiency virustype 1 gp120 glycoprotein: consequences for virusentry and neutralization. J. Virol. 72, 4694–4703.

38. Wiley, D. C. & Skehel, J. J. (1987). The structure andfunction of the hemagglutinin membrane glyco-protein of influenza virus. Annu. Rev. Biochem. 56,365–394.

39. Ho, S. N., Hunt, H. D., Horton, R. M., Pullen, J. K. &Pease, L. R. (1989). Site-directed mutagenesis byoverlap extension using the polymerase chainreaction. Gene, 77, 51–59.

40. Boeke, J. D., Model, P. & Zinder, N. D. (1982). Effects ofbacteriophage f1 gene III protein on the host cellmembrane. Mol. Gen. Genet. 186, 185–192.

41. Stengele, I., Bross, P., Garces, X., Giray, J. & Rasched, I.(1990). Dissection of functional domains in phage fdadsorption protein: discrimination between attach-ment and penetration sites. J. Mol. Biol. 212, 143–149.

42. Nelson, F. K., Friedman, S. M. & Smith, G. P. (1981).Filamentous phage DNA cloning vectors–a non-infective mutant with a non-polar deletion in geneIII. Virology, 108, 338–350.

43. Endemann, H. & Model, P. (1995). Location offilamentous phage minor coat proteins in phage andin infected cells. J. Mol. Biol. 250, 496–506.

Edited by J. Karn

(Received 14 June 2005; received in revised form 13 November 2005; accepted 22 November 2005)Available online 9 December 2005