In vitro and in vivo delivery of genes and proteinsusing the bacteriophage T4 DNA packaging machinePan Taoa, Marthandan Mahalingama, Bernard S. Marasaa, Zhihong Zhanga, Ashok K. Choprab, and Venigalla B. Raoa,1

aDepartment of Biology, The Catholic University of America, Washington, DC 20064; and bDepartment of Microbiology and Immunology, University of TexasMedical Branch, Galveston, TX 77555

Edited by Michael G. Rossmann, Purdue University, West Lafayette, IN, and approved February 26, 2013 (received for review January 15, 2013)

The bacteriophage T4 DNA packaging machine consists of a molec-ular motor assembled at the portal vertex of an icosahedral head.The ATP-powered motor packages the 56-μm-long, 170-kb viral ge-nome into 120 nm × 86 nm head to near crystalline density. Weengineered this machine to deliver genes and proteins into mamma-lian cells. DNA molecules were translocated into emptied phagehead and its outer surface was decorated with proteins fused toouter capsid proteins, highly antigenic outer capsid protein (Hoc)and small outer capsid protein (Soc). T4 nanoparticles carrying re-porter genes, vaccine candidates, functional enzymes, and targetingligands were efficiently delivered into cells or targeted to antigen-presenting dendritic cells, and the delivered genes were abundantlyexpressed in vitro and in vivo. Mice delivered with a single dose ofF1-V plague vaccine containing both gene and protein in the T4 headelicited robust antibody and cellular immune responses. This “pro-gene delivery” approach might lead to new types of vaccines andgenetic therapies.

protein and DNA delivery | protein display | gene therapy | phage assembly

Delivery of recombinant genes and proteins into cells formsthe core of molecular biology and biotechnology. Although

numerous methods have been developed to deliver genes; elec-troporation (1), viral vectors (2), and microinjection (3), proteindelivery is less common (4, 5). Moreover, no platforms currentlyexist that can efficiently deliver both genes and proteins (TableS1). With the explosion of genomics and protein network data-bases, delivery of genes and proteins into target cells would beessential to generate basic understanding of mechanisms as wellas to explore novel biomedical therapies. For instance, futuretreatments of complex genetic and infectious diseases such as can-cer and AIDS might include delivery of genes and proteins in ap-propriate combinations, which we refer to as “progene delivery.”We have engineered the phage T4 DNA packaging machine

into a high-capacity progene delivery vehicle. Tailed phages usepowerful molecular machines to condense the highly negativelycharged DNA genome inside a capsid to near crystalline density(>500 mg/mL). These machines generate forces as high as 80 pNto oppose the electrostatic repulsion and bending energies thatresist DNA compaction (6). The phage T4 packaging machine,one of the fastest (packaging rate up to ∼2000 bp/sec) and mostpowerful (power density, ∼5,000 kW/m3) (7), consists of two es-sential components: an empty prohead and a molecular motorcomprised of five subunits of the large terminase protein, gp17(8, 9) (Fig. 1 A and B).Phage T4 head is decorated with two outer capsid proteins:

Hoc (highly antigenic outer capsid protein; 155 copies/head) andSoc (small outer capsid protein; 870 copies/head) (10) (Fig. 1 Cand D). The nanomolar affinity binding sites for Hoc and Socappear after the prohead has undergone “expansion,” a major con-formational change in the capsid protein, gene product (gp)23*(* represents the cleaved form). The tadpole-shaped Soc (9 kDa)assembles as a trimer at the quasi threefold axes, clamping theadjacent capsomers and forming a reinforcing cage around theshell (11) (Fig. 1 C–E). The linear fibers of Hoc (about 170 Ålong; 41 kDa) each having a string of four domains, three ofwhich are IgG-like domains, assemble at the center of hexamericcapsomers (12) (Fig. 1 C, D, and F). Hoc facilitates attachment

of phage to bacteria. Soc and Hoc are not essential for phageinfection but provide survival advantages for the virus (13).The N and C termini of Soc (Fig. 1E) and the N terminus ofHoc (Fig. 1F) are well exposed, allowing efficient display offoreign proteins on the capsid surface (14).Recently, we constructed a neckless (13 amber), tailless (10

amber), hoc and soc deletion mutant that accumulates packagedheads (15). In the classic assembly pathway, the packaging motorassembles on a prohead and after head maturation and (headful)genome packaging, themotor dissociates. Then, the neck proteins,gp13, gp14, and gp15, attach to the portal sealing off the packagedhead. Tail and tail fibers attach to the neck, producing an infectiousvirion. In the neckless mutant, the packaged heads become un-stable and release the DNA due to internal pressure, which is es-timated to be ∼6 MPa, or >10 times the pressure in a champagnebottle (6, 16). Unexpectedly, we discovered that the packagingmotor can reassemble on this fully matured, emptied phage headand refill with any DNA (15). The T4 packaging machine, thus, ispromiscuous, neither discriminating the head on which it assem-bles nor the DNA that it packages.These findings led us to ask whether the phage packaging ma-

chine could be reconfigured to deliver genes and proteins intomammalian cells. Conceivably, each head packaging several genes(17), up to ∼170 kb, and displaying several proteins outside, up to∼1,025 molecules (14, 18), could deliver the entire “payload” intocells. Such a systemwould be attractive not only because of its largegenetic capacity but also because T4 does not infect mammaliancells, is nontoxic, and has no preexisting immunity in the host.Here, we show that combinations of reporter genes, vaccine genes,functional enzymes, and targeting ligands can be incorporated intothe T4 head and delivered into mammalian cells to near 100%efficiency. Our experiments further demonstrate that delivery canbe targeted to antigen-presenting dendritic cells (DCs) and thedelivered genes are abundantly expressed both in vitro and in vivo.Mice immunized with a single dose of “prime-boost” plague vac-cine containing the recombinant F1-V gene from Yersinia pestispackaged inside the T4 head and the F1-V protein displayedoutside elicited robust antibody and cellular immune responses.These studies established a unique phage-based mammalian geneand protein delivery system that could lead to novel vaccine andgenetic therapies.

Results and DiscussionExperimental Design for Progene Delivery. A detailed experimentalscheme was developed to quantitatively analyze progene deliveryby T4. The T4 DNA packaging machine was first assembled bybinding the gp17 subunits at the dodecameric portal (gp20) ofempty phage head (Fig. 2A). Fueled by ATP, the machinepackages DNA into the head (Fig. 2B), up to ∼170 kb. Soc- and/

Author contributions: P.T., M.M., A.K.C., and V.B.R. designed research; P.T., M.M., B.S.M.,and Z.Z. performed research; P.T., M.M., B.S.M., Z.Z., A.K.C., and V.B.R. analyzed data; andP.T. and V.B.R. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1To whom correspondence should be addressed. E-mail: rao@cua.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1300867110/-/DCSupplemental.

5846–5851 | PNAS | April 9, 2013 | vol. 110 | no. 15 www.pnas.org/cgi/doi/10.1073/pnas.1300867110

or Hoc-fused proteins/peptides were then added to decorate thecapsid surface (Fig. 2 C and D). The number of DNA moleculespackaged and the number of protein molecules displayed werequantified by agarose and polyacrylamide gel electrophoresis,respectively. The T4 nanoparticles were then added to mamma-lian cells. The particles bind to cells either nonspecifically (Fig.2E) or by recognizing a specific receptor on the cell surface (Fig. 2F and G) and are internalized, presumably by endocytosis (Fig.

2H). The surface displayed proteins (e.g., β-galactosidase) (Fig. 2I)and the encapsidated DNA (e.g., luciferase and GFP genes) (Fig.2J) would be released into the cytosol. The protein function isexpressed in the cytosol (Fig. 2I), whereas the DNA enters thenucleus (Fig. 2K) and primes transcription of cloned genes underthe control of the strong CMV promoter (Fig. 2L), leading tooverexpression of luciferase and GFP proteins (Fig. 2M).

T4 Progene Particles Are Efficiently Assembled in Vitro. We firstisolated the neckless heads fromEscherichia coli cells infected with10-amber13-amber hoc-delsoc-del mutant (Fig. 3A). Most of theseheads, ∼90%, released the viral genome spontaneously, which wasdigestedwithDNase I. The emptied heads retained an∼8-kbDNAinside, leaving the rest of the ∼162-kb space available for foreignDNA.The headswere purified byCsCl gradient centrifugation andion-exchange chromatography (Fig. S1) and mixed with gp17 toreconstitute the functional packaging machine. These machinesrefilled the heads at high efficiency, as shown by bulk and single-molecule assays (Fig. S2). CryoEM indicated that most of themachines actively encapsidated DNA. The packaging reactionswere terminated by the addition of DNase I to degrade theunpackaged DNA. The encapsidated and DNase I resistant DNAwas released by treatment with proteinase K and quantified byagarose gel electrophoresis (SIMaterials andMethods). Single (Fig.3B) or multiple plasmids, long ∼80-kb ligated DNA (up to ∼170kb) or short 2.3-kb PCR-amplified DNA (Fig. 3C) were efficientlypackaged. Quantification of the packaged eGFP (Fig. 3C Upper)and luciferase (Fig. 3C Lower) plasmid DNA bands showed thatthe number of plasmid DNA molecules packaged per head in-creased with increasing ratio of DNA molecules to head particlesin the reaction mixture (Fig. 3B; black arrows correspond to thepackaged DNA). At the highest ratio used, 30:1, approximately 10molecules of plasmidDNAwere packaged per head.When two (ormore) plasmidDNAswere present, both were packaged at roughlyequal frequencies (Fig. 3C, lane 1). Foreign peptides [cell pene-tration peptides (CPPs)] and proteins [β-galactosidase, dendriticcell specific receptor (DEC)205 monoclonal antibody (mAb),CD40 ligand (CD40L), etc.] were then arrayed on the surface byadding the Soc- and Hoc-fused recombinants to the reactionmixture containing the packaged heads, and the unbound Soc andHoc proteins were removed by centrifugation and washing withbuffer (SI Materials and Methods) (Fig. 3D and Fig. S3). Soc andHoc binding to capsid followed simple first-order kinetics, and the

Fig. 1. The bacteriophage T4 DNA packaging machine. Shown are variousstructural features of the T4 DNA packaging machine that are used in thedesign of progene delivery vehicle. (A) Structural model of the phage T4DNA packaging machine showing the pentameric motor (cyan) assembled atthe dodecameric portal vertex (red) of the head that can accommodate ∼170kb DNA. (B) Ribbon models of portal and motor. (C) Capsomer showing thearrangement of the major capsid protein gp23* (green), Soc trimers (blue),and Hoc fiber (yellow). (D) Capsomer showing the ribbon models of gp23*,Soc, and Hoc. (E) Soc trimer showing the exposed N (green dots) and C (reddots) termini. (F ) Hoc showing the exposed N terminus (green dot) at thetip of the fiber. The C-terminal Hoc domain 4 binds to capsid, but its struc-ture has not yet been solved, hence shown as a yellow mass at the base ofthe fiber.

Fig. 2. Experimental design for progene delivery.The bacteriophage T4 head is engineered to delivergenes and proteins into mammalian cells. The DNApackaging machine is assembled by binding of gp17motor to the portal of empty Hoc− Soc− phage T4head (the cut-out of the head shows both the ex-terior and the interior) (A). Using the energy fromATP hydrolysis the motor packages DNA moleculesinto the head (B). Soc-fused proteins (C) and Hoc-fused CPPs or targeting molecules (D) were dis-played on the heads. The head particles bind to cellseither nonspecifically (general delivery) (E) or througha host receptor (targeted delivery) (F and G) and areinternalized (H). The surface displayed protein (e.g.,β-galactosidase) (I) and encapsidated DNA (e.g., lucif-erase and GFP genes) (J) are released into the cytosol.The DNA enters the nucleus (K), priming transcription(L), and expression of luciferase and GFP proteins (M).

Tao et al. PNAS | April 9, 2013 | vol. 110 | no. 15 | 5847

APP

LIED

BIOLO

GICAL

SCIENCE

S

copy number of bound protein was controlled by varying the ratioof Soc- or Hoc-fusion protein molecules to binding sites (Fig. S3DandE; see below). At a ratio of 20:1, nearly all of the capsid bindingsites were occupied.

Engineered T4 Particles Efficiently Delivered Genes into MammalianCells. Initial experiments showed that T4 delivery of the lucifer-ase gene into HEK293T cells was very poor. However, when theparticles were decorated with CPP-Tat (CPP-T) or CPP-Antp(CPP-P) (Fig. 3D), delivery was efficient, as shown by the ap-pearance of high luciferase activity in the cell lysates (Fig. 3 E–H). CPPs are 20- to 30-aa peptides rich in basic amino acids thatfacilitate passage of attached cargo molecules across the cellmembrane (19). Tat and Antp refer to CPPs of HIV-1 trans-activator protein and Drosophila antennapedia homeobox pro-tein, respectively (19) (Fig. S3). Luciferase activity reached themaximum at 105 heads per cell (Fig. 3E). The activity appearedby as early as 5 h, reached a peak by about 16 h, and was sus-tained for at least 30 h (Fig. S4A).Luciferase activity increased with increasing copy number

of CPP (Fig. 3F and Fig. S4B); however, at a given copynumber, CPP-T is 3.3-fold more effective than CPP-P (Fig. 3G).Hoc-CPP-T (Hoc-T) was superior to Soc-CPP-T (Soc-T); at aHoc-T copy number of 155 per head, it showed 5.4-fold higherluciferase activity than Soc-T, even though Soc-T’s copy numberat 870 copies per head is 5.6 times greater than that of Hoc-T(Fig. 3G). This increased delivery was presumably because theHoc-fused CPP is positioned ∼170 Å away from capsid surface(height of Hoc fiber), thus having greater reach to contact thecell membrane of the mammalian cell compared with Soc-CPPthat is nearly “glued” to the capsid wall (11, 12) (Fig. 1 A and F).Combining Hoc-CPP and Soc-CPP further increased deliveryefficiency by 1.9-fold (Fig. 3H). The delivery efficiency wasroughly the same whether the packaged DNA was a plasmid,short PCR-amplified DNA, or ligated into a long concatemer(Fig. 3J). Parallel experiments using another plasmid, pEGFP-C1, gave similar results, with nearly 100% of cells showing greenfluorescence (Fig. 3I).Proheads or isometric phage heads can also be used as delivery

vehicles (Fig. S4), but the fully mature emptied phage heads,which are extremely stable and can be produced in large quantities

(∼5 × 1013 particles per liter of culture), were the heads ofchoice. Preliminary experiments using inhibitors indicated thatthe T4 head delivery was mediated by an endocytic pathway thatis energy-dependent and requires actin polymerization but notclathrin or caveolin (Fig. S5 A and B). No significant toxicity wasobserved following T4 delivery, even at a ratio of 106 head par-ticles per cell (Fig. S5 C and D).The T4 delivery efficiency was assessed to be very high,

approaching as many as 105 luminescence units per cell (Fig.3K). To determine how this efficiency compared with lipofect-amine and adeno-associated virus (AAV), the two most efficienttransfection systems currently available, luciferase gene wasdelivered into HEK293T cells under conditions that are optimalfor each of these systems. The data showed that the T4 efficiencywas similar to that of the lipofectamine but ∼3.3 times higherthan that of the AAV-DJ vector, the most efficient AAV vectorreported to date (20) (Fig. 3K).

Targeted Delivery of Genes and Proteins into DCs by T4. The antigen-presenting DCs were chosen to test whether T4 delivery can betargeted to specific cells. DCs are critical for vaccine uptake andinduction of humoral as well as cellular immune responses (21).We hypothesized that displaying a DC-specific ligand on thecapsid lattice would enable T4 to capture DCs leading to endo-cytosis and delivery of the attached cargo. To test this hypothesis,the DEC205mAb or the CD40 ligand (CD40L), which respec-tively recognize the DC-specific DEC205 receptor (22) and CD40(23), were displayed on the T4 heads. The heads were firstpackaged with luciferase and/or eGFP genes and incubated withthe Hoc-GG fusion protein plus DEC205 mAb (Fig. S6). TheGG fusion contained two tandemly linked 122–aa IgG bindingdomains of protein G (24) from Streptomyces. When attached toT4 head through Hoc, these domains captured the Fc region ofIgG and formed arrays of DEC205 mAb on the head with theirreceptor binding Fab regions well exposed (Fig. 4 A and G andFig. S6 C and D). These particles efficiently delivered thepackaged luciferase gene into mouse DC2.4 cells (25) but wereunable to do so into nonspecific cells, such as Hep3B cells, thatlacked the receptor (Fig. 4B). Delivery efficiency was nearly100% in several independent experiments (Fig. 4C). When boththe luciferase and eGFP genes were packaged into the same

Fig. 3. T4 heads efficiently deliver packaged DNAinto mammalian cells. (A) Cryo electron micrographof purified T4 heads. (B) Packaging of MluI-linear-ized pEGFP-C1 plasmid DNA (Upper) and BamHI-linearized psiCHECK2 plasmid DNA (Lower) intoHoc− Soc− T4 heads at increasing DNA-to-head ra-tios. The black arrow shows the packaged DNA. (C)Packaging of two plasmids (lane 1; psiCHECK2 plas-mid DNA containing luciferase gene under thecontrol of SV40 early promoter, blue arrow; pEGFP-C1 plasmid DNA containing eGFP gene under thecontrol of the cytomegalovirus (CMV) promoter(green arrow), concatemerized luciferase DNA (lane2, black arrow), and PCR-amplified luciferase ex-pression cassette (lane 3, orange arrow). The redarrows show the 8-kb T4 DNA present in the heads.(D) In vitro assembly of CPPs on T4 heads. The redarrow shows bound Soc-T and Soc-P, and the bluearrow shows bound Hoc-T and Hoc-P; control lanesshow purified heads. (E) Dose-dependent delivery ofluciferase DNA into mammalian cells using CPP-decorated T4 heads. (F) Luciferase DNA deliveryincreases with increasing copy number of CPP. (G)Hoc-CPP decorated heads deliver genes more effi-ciently than Soc-CPP heads. (H) Gene delivery was atits highest when the heads were decorated with both Hoc-T and Soc-T CPPs. (I) eGFP DNA packaged heads decorated with (IV) or without (I) Hoc-CPPs are usedfor delivery. Fluorescence (III and VI) and phase contrast (II and V) micrographs of HEK293T cells in the presence (V and VI) or absence (II and III) of CPP. (J)Expression of packaged and delivered luciferase gene as plasmid, concatemer, or PCR product. (K) Comparison of the delivery efficiencies of T4, AAV-DJ,and lipofectamine. (In this experiment, due to the low capacity of AAV, the shorter pLuci plasmid was used instead of the psiCHECK2). Error bars represent SD;**P < 0.01; ***P < 0.001.

5848 | www.pnas.org/cgi/doi/10.1073/pnas.1300867110 Tao et al.

head, the heads delivered both the genes to near 100% efficiencyas shown by the presence of both the luciferase and green fluo-rescence signals in the HEK293T cells (Fig. S7A), but not in thecontrol cells (Fig. S7B). Similar results were also obtained usingthe displayed CD40L fused to the N terminus of Hoc.We then tested protein delivery into DCs using the 129-kDa

E. coli β-galactosidase as the model protein. This is a stringenttest because β-galactosidase is functional only as a tetramer.Therefore, the Soc-fused protein must oligomerize into a >500-kDa complex and be efficiently displayed and delivered. A β-galactosidase-Soc recombinant was constructed and purified (Fig.S8). The fusion protein was efficiently displayed on T4 heads(Fig. 4D), and the resultant heads showed β-galactosidase activity(Fig. 4E). Hoc-fused CPP-T or DEC205 mAb was then deco-rated on the same heads (Fig. 4 F and G, I and V). These heads,when incubated with DCs, showed delivery of β-galactosidaseinto nearly 100% of the cells, as shown by the appearance of blueX-Gal signal (Fig. 4 F, IV and VIII). Control particles containingeither no targeting ligand (Fig. 4 F, III and VII) or the solubleenzyme alone (Fig. 4 F, II and VI) showed poor to no signal.Can T4 simultaneously deliver both genes and proteins into

cells? To address this question, the heads were first packagedwith luciferase and eGFP plasmids. The surface was then deco-rated with β-galactosidase as Soc fusion, and with CPP (Fig. 4H)or DEC205 mAb (Fig. 4I) as Hoc fusion. We then tested whetherthe entire payload could be delivered into DCs. As shown in Fig.4 H and I, nearly 100% of cells showed strong signals for all threemarkers, demonstrating that the T4 nanoparticles efficientlydelivered two different genes as well as large oligomeric com-plexes into DCs. The same was observed using a series of otherproteins displayed on the head, including the F1-V fusion pro-tein from Y. pestis (see below).

In Vivo T4 Delivery. In vivo T4 delivery was tested using a mousemodel. Four groups of mice were injected intramuscularly withT4 heads packaged with the luciferase plasmid. The first groupreceived heads containing no displayed ligand, whereas the sec-ond, third, and fourth groups received heads displayed withDEC205mAb, CD40L, and CPP-T, respectively. At differenttime points after injection, mice were injected with the bio-luminescence substrate D-Luciferin, and the entire body wasimaged. Unexpectedly, the strongest luciferase signal was ob-served in the first group, which received heads containing nodisplayed ligand (Fig. 5A); these same particles showed very poordelivery in vitro (Fig. 3 E and I, I–III). Furthermore, the signalappeared as early as 6 h after injection, suggesting that themuscle

cells had taken up the T4 nanoparticles and expressed the de-livered DNA efficiently at the site of injection. A time-courseanalysis showed that the luciferase signal was seen for at least 14 dusing the standard luciferase expression cassette, and for at least30 d if the cassette is flanked by AAV inverted terminal repeats(ITRs primeDNA replication of the delivered gene) (Fig. 5B andC). On the other hand, the DEC205mAb- and CD40L-displayedparticles showed weak to no luciferase signal, and the signaldisappeared rapidly: By 16 h, hardly any signal remained at thesite of injection (Fig. 5 A, compare 3 and 4 and 5 and 6 with 1 and2). These data suggested that the DCs that took up theDEC205mAb- or CD40L-displayed T4 particles migrated toother parts of the body, e.g., lymph nodes and spleen, therebydiluting the signal to below the sensitivity of bioluminescenceimaging (similar observations were made in ref. 26). The CPP-displayed group also behaved similarly (Fig. 5 A, 7 and 8), butwhether this was due to migration of DCs and/or whether othertypes of cells were also involved requires further investigation. Tofurther confirm that the lack of signal was due to migration ofDCs rather than to lack of delivery, another control experimentwas performed by delivering the same amount of luciferase DNAinto mice using heads containing either no ligand or the displayedCD40L (Fig. 5 D, 1). The CD40L group that showed no signal(Fig. 5 D, 3) induced similar titers of anti-luciferase antibodies(Fig. 5E) as the no-ligand group that showed strong signal (Fig. 5D, 2). In addition, the DC-targeted group elicited strong cellularimmune responses (see below). Taken together, these datademonstrate that T4 delivery is efficient in vivo, and the deliverycould be either localized to myocytes at the site of injection ordispersed into the body through migration of DCs. In the case ofthe former, the gene product is localized to themuscle (although itcould be secreted into the body by attaching a signal peptide),whereas in the targeted group, the gene product and the peptidesderived from it travel, but importantly, are presented to DC-interacting cells such as the T cells and B cells (21) (see below).

Single Dose of T4 Delivered Plague Vaccine Induced Robust Humoraland Cellular Immune Responses. An ideal vaccine would stimulateboth the humoral (Th2) and cellular (Th1) arms of the immunesystem (27). Delivering the vaccine in both forms, as antigen and asexpressible DNA, might prime as well as boost the immune system(“prime-boost” vaccine), stimulating the Th1 and Th2 responses(28). This prime-boost approach might be particularly importantfor combating complex infectious agents such as HIV-1, malaria,and tuberculosis (TB) for which no effective vaccines have yetbeen developed by traditional methods.

Fig. 4. Targeted delivery of genes and proteins into DCs by T4heads. (A) Display of DEC205mAb on T4 heads through GGdomain. Hoc-GG, Soc-GG, and DEC205mAb bands are markedwith green, blue, and red arrows, respectively. The top redarrow corresponds to heavy chain, and the bottom red arrowcorresponds to light chain of IgG. (B) Targeted delivery of lu-ciferase gene into DC2.4 cells but not into control Hep3B cells.(C ) Fluorescence (II and IV) and phase contrast (I and III)micrographs of DC2.4 cells transduced with GFP heads in thepresence (III and IV) or absence (I and II) of displayed DEC205-mAb. (D) Display of tetrameric β-galactosidase on T4 heads atdifferent ratios of β-galactosidase-Soc molecules to Soc bindingsites. Red arrow shows bound β-galactosidase-Soc, and the 40:0lane is the control lane showing no nonspecific binding ofβ-galactosidase-Soc in the absence of heads. (E) X-Gal cleavageactivity of displayed β-galactosidase corresponding to lanes inD. (F and G) T4 heads displayed with β-galactosidase (blue) andCPP (I; red) (F) or DEC205mAb (V; yellow) (G) delivered func-tional β-galactosidase into DC2.4 cells. (H and I) T4 headspackaged inside with luciferase and GFP DNA and displayedoutside with β-galactosidase (blue) and CPP (H; red) orDEC205mAb (I; yellow) are used for delivery. II and III show phase contrast and fluorescence micrographs, respectively, of DC2.4 cells transduced with controlpackaged heads containing no CPP or DEC205mAb; IV, stained for β-galactosidase activity with X-Gal; V, fluorescence of eGFP after staining with FITC-labeledantibody; VI, fluorescence of luciferase after staining with Rhodamine-labeled antibody; VII, merged fluorescence images of eGFP (V) and luciferase (VI).

Tao et al. PNAS | April 9, 2013 | vol. 110 | no. 15 | 5849

APP

LIED

BIOLO

GICAL

SCIENCE

S

Plague, also known asBlackDeath, caused byY. pestismight alsorequire the induction of antibody and cellular immune responses(29). The two principal candidates for a plague vaccine are thecapsular protein [capsule antigen family 1 (Caf1) or F1, 16 kDa]and the low calcium responseV antigen (LcrV orV, 37 kDa), a keycomponent of the type 3 secretion system. Current plague vaccinescontaining F1 and V recombinant proteins elicit strong antibodyresponses but poor cellular immune responses (29). This is acommon problem encountered by conventional subunit vaccines.We asked whether the T4 nanoparticles delivering both F1-VDNA and F1-V protein can modify the quality of immuneresponses. Furthermore, because the T4-delivered gene continuesto be expressed for weeks, we asked whether a single dose ofvaccine would be sufficient to induce strong immune responses.These questions were addressed by (i) cloning a mutant F1-V

gene that produces soluble monomeric F1-V fusion protein underthe control of the strong CMV promoter and packaging the F1-VDNA into T4 head, and (ii) fusing the mutant F1-V protein to theN terminus of Soc and displaying the fusion protein on the capsidsurface. Groups of mice were immunized intramuscularly witha single dose of F1-V vaccine in various combinations (Fig. 6A).For prime-boost vaccination (group 5), mice were injected with T4heads (∼5 × 1011 particles) packaged with ∼25 μg of F1-V DNAand displayed with ∼30 μg of F1-V protein plus ∼100 copies perhead of DEC205mAb. Control groups included naïve, F1-V pro-tein adjuvanted with Alum, heads packaged with F1-V DNA, andheads displayed with F1-V protein. No adjuvant was included inany of the T4-vaccinated groups. ELISA of sera 4 wk after im-munization showed that strong antibody responses were induced inall of the groups; end point titers as high as 7.5 × 104 were attainedin some of the mice (Fig. 6B). However, strong cellular responseswere elicited only in two of the groups: the T4 DNA group and theT4 prime-boost group that delivered both the F1-V DNA and theF1-V protein to DCs (Fig. 6C). Indeed, the latter group producedthe strongest IFNγ response. Unexpectedly, splenocytes from thisgroup showed high incidence of Elispots even without stimulationwith F1-V antigen (Fig. 6C; compare the red unstimulated columnin group 5 with the same from rest of the groups). The counts in-creased further by F1-V stimulation (Fig. 6C). These results areconsistent with our hypothesis that T4 vaccine delivery into DCsinduces abundant expression of F1-V and presentation to the im-mune system for a long period, in turn leading to continuousstimulation of T cells.

The above results demonstrated the application of T4 progenedelivery for prime-boost vaccination. A single dose of T4 vaccinewithout any adjuvant was sufficient to stimulate both arms of theimmune system. In addition, unlike the conventional vaccines, theimmune system in T4 vaccinated animals remain activated forweeks after immunization. Thus, T4 appears to be a good platformfor future development of efficacious prime-boost vaccines.

ConclusionsThe above series of experiments established the phage T4 DNApackaging machine as a unique type of gene and protein deliveryplatform. With well understood structural and mechanistic

Fig. 5. In vivo delivery by T4. (A) Mice were injected in-tramuscularly with T4 heads packaged with pLuci plasmid andits outer surface decorated either with no ligand (panels 1 and2) or with DEC205mAb (panels 3 and 4), CD40L (panels 5 and 6),or CPP (panels 7 and 8). (B) Luciferase expression cassettes used,without (pLuci) and with ITRs from AAV (pITR-luci), for pack-aging into T4 heads; in both the vectors, the luciferase gene isunder the control of the CMV promoter. (C) Luciferase signalstays for a longer period if the expression cassette is flanked byITRs. (D) Further analysis of targeted delivery into DCs. (1) Thesame amount of pITR-Luciferase was packaged into T4 headsdecorated either with no ligand (lane I) or with CD40L (lane II);lane “M” shows molecular size standards. (2) Strong luciferasesignal was seen at the site of injection with T4 heads containingno displayed ligand. (3) No signal was seen at the site of in-jection with T4 heads displayed with the DC-specific ligand,CD40L. (E) ELISA titration showed that both the no ligandgroup and the CD40L group induced the same level of anti-luciferase antibodies. y axis represents absorbance at 650 nm.

Fig. 6. A single dose of T4-delivered plague vaccine induced robust hu-moral and cellular immune responses. (A) Five groups of mice were vacci-nated by the intramuscular route (i.m.) with various formulations shown inthe table. A mutant F1 gene from Y. pestis was fused to the N terminus of Vgene and the F1-V fusion was then fused to the N terminus of Soc gene. TheF1-V-Soc fusion protein was purified as a 66 kDa monomer. For the DNAgroup (#4), F1-V was cloned under the control of the CMV promoter. Spleenswere harvested from three mice from each group at day 21 for Elispotassays, and sera were collected at day 28 for ELISAs. The ELISA (B) and Elispot(C) assays were performed according to the procedures described in Mate-rials and Methods. SFC refers to spot-forming cells expressing IFN-γ. Errorbars represent SD; **P < 0.01; ***P < 0.001.

5850 | www.pnas.org/cgi/doi/10.1073/pnas.1300867110 Tao et al.

features (Fig. 1; the 170-kb-capacity head, the promiscuous pack-aging motor, the 1,025 molecule capacity surface lattice, Soc pro-tein with exposed termini, and ∼170-Å-long Hoc fiber to capturetarget cells), T4 offers many options to engineer the nanoparticlefor diverse applications. Unlike the existing systems that are mostlysingle featured and have limited capacity (Table S1), the T4applications range from simple in vitro delivery of genes and pro-teins to complex in vivo deliveries involving combinations of genes,proteins, oligomeric complexes, and targeting molecules to treatdiseases. A set of key experiments have been selected to demon-strate these broad applications. The data show that T4 efficientlydelivered single or multiple genes, either as plasmids, PCR-ampli-fied DNAs, or concatemerized DNA. At the same time, single ormultiple proteins were delivered, either as short peptides, full-length proteins, or large oligomeric complexes. Different parts ofthe nanoparticle were engineered to carry out different tasks butintegrated to generate a desired outcome. For instance, one part(Hoc) was dedicated to target the nanoparticles to specific cells,whereas another part (Soc) was designed to transiently expressa new function, and yet another part (capsid interior) to overexpressan antigen for an extended period. Thus, T4 offers an advancedsystem for delivery of recombinant DNA and protein molecules.The phage T4 platform offers several technical advantages that

make it easily adaptable to any laboratory. Bacterially derived, theT4 head is noninfectious, nontoxic, and has no preexisting immu-nity. The two main components of the T4 platform, heads andpackaging motor (gp17), can be prepared relatively easily fromE. coli in large quantities (∼5× 1013 heads, or 5mg of gp17 per literof culture). Both are very stable and retain activity after storage forat least 2 y at −70 °C. The assembly steps are relatively simple,performed in a single tube, and completed in about 3 h, and eachstep can be quantified. As shown, a series of customized nano-particles with different compositions of genes and proteins couldbe prepared in a single experiment.T4 progene delivery offers avenues for developing useful bio-

medical therapies. We have demonstrated a prime-boost vaccina-tion approach against pneumonic plague that elicits strong antibody

and cellular immune responses. This approach could be usedbroadly to develop efficacious vaccines against complex infectiousagents such as HIV-1, malaria, and TB. T4 heads deliveringrecombinase and targeting molecules displayed outside, and ge-nomic DNA and siRNA genes packaged inside, could be used incancer therapy, either to kill cancer cells by induction of apoptosisor to reverse the cancer phenotype by turning off certain onco-genes (30). T4 delivery might allow safe and efficient repro-graming of pluripotent stem cells by delivering transcription factorgenes and proteins such as c-Myc, kruppel-like factor 4 (Klf4), sexdetermining region Y box 2 (Sox2), and octamer-binding tran-scription factor (Oct3/4), as well as replacing the defective genewith a functional gene by in vivo delivery of a long stretch of ge-nomic DNA (31).

Materials and MethodsPackaging assays were performed using Hoc−Soc− phage T4 heads, gp17, andplasmid DNA. Protein display was carried out by adding Hoc- and/or Soc-fused proteins to the packaged heads. The heads were then incubated withHEK293 or DC2.4 cells for gene and protein delivery. After washing withbuffer, the cells were analyzed for luciferase and β-galactosidase activities.For in vivo delivery, the heads were injected into mice by intramuscularroute and imaged. These and the immunological assays such as ELISA andElispot are described in detail in SI Materials and Methods. The protocols formouse studies were approved by the Institutional Animal Care and UseCommittee of the University of Texas Medical Branch, Galveston, TX.

ACKNOWLEDGMENTS. We thank Dr. Jay Chiorini [(National Institute ofDental and Craniofacial Research/National Institutes of Health (NIH)] forcritical review of the manuscript, Mr. Victor Padilla-Sanchez [CatholicUniversity of America (CUA)] for preparing the T4 head images, Dr. MyriamGorospe (National Institute on Aging/NIH) for providing access to thefluorescent microscope, Ms. Li Dai and Mr. Vishal Kottadiel (CUA) for thesingle-molecule DNA packaging assays, and Ms. Michelle Kirtley andMs. Christina van Lier (University of Texas Medical Branch) for in vivo imagingand ELISA and Elispot assays. This work was supported by NIH/NationalInstitute of Allergy and Infectious Diseases (NIAID) Grants U01-AI082086and AI081726, the Bill and Melinda Gates Foundation (to V.B.R.), and in partby NIAID Grant AI064389 (to A.K.C.).

1. Neumann E, Schaefer-Ridder M, Wang Y, Hofschneider PH (1982) Gene transfer intomouse lyoma cells by electroporation in high electric fields. EMBO J 1(7):841–845.

2. Kay MA (2011) State-of-the-art gene-based therapies: The road ahead. Nat Rev Genet12(5):316–328.

3. Demayo JL, Wang J, Liang D, Zhang R, Demayo FJ (2012) Genetically Engineered Miceby Pronuclear DNA microinjection. Curr Protoc Mouse Biol 2:245–262.

4. Yan M, et al. (2010) A novel intracellular protein delivery platform based on single-protein nanocapsules. Nat Nanotechnol 5(1):48–53.

5. Kaczmarczyk SJ, Sitaraman K, Young HA, Hughes SH, Chatterjee DK (2011) Proteindelivery using engineered virus-like particles. Proc Natl Acad Sci USA 108(41):16998–17003.

6. Smith DE, et al. (2001) The bacteriophage straight phi29 portal motor can packageDNA against a large internal force. Nature 413(6857):748–752.

7. Fuller DN, Raymer DM, Kottadiel VI, Rao VB, Smith DE (2007) Single phage T4 DNApackaging motors exhibit large force generation, high velocity, and dynamic vari-ability. Proc Natl Acad Sci USA 104(43):16868–16873.

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

9. Sun S, et al. (2008) The structure of the phage T4 DNA packaging motor suggestsa mechanism dependent on electrostatic forces. Cell 135(7):1251–1262.

10. Fokine A, et al. (2004) Molecular architecture of the prolate head of bacteriophageT4. Proc Natl Acad Sci USA 101(16):6003–6008.

11. Qin L, Fokine A, O’Donnell E, Rao VB, Rossmann MG (2010) Structure of the smallouter capsid protein, Soc: A clamp for stabilizing capsids of T4-like phages. J Mol Biol395(4):728–741.

12. Fokine A, et al. (2011) Structure of the three N-terminal immunoglobulin domains ofthe highly immunogenic outer capsid protein from a T4-like bacteriophage. J Virol85(16):8141–8148.

13. Ishii T, Yanagida M (1977) The two dispensable structural proteins (soc and hoc) of theT4 phage capsid; their purification and properties, isolation and characterization ofthe defective mutants, and their binding with the defective heads in vitro. J Mol Biol109(4):487–514.

14. Li Q, Shivachandra SB, Zhang Z, Rao VB (2007) Assembly of the small outer capsidprotein, Soc, on bacteriophage T4: a novel system for high density display of multiplelarge anthrax toxins and foreign proteins on phage capsid. J Mol Biol 370(5):1006–1019.

15. Zhang Z, et al. (2011) A promiscuous DNA packaging machine from bacteriophage T4.PLoS Biol 9(2):e1000592.

16. Lander GC, et al. (2006) The structure of an infectious P22 virion shows the signal forheadful DNA packaging. Science 312(5781):1791–1795.

17. Leffers G, Rao VB (1996) A discontinuous headful packaging model for packaging lessthan headful length DNA molecules by bacteriophage T4. J Mol Biol 258(5):839–850.

18. Sathaliyawala T, et al. (2006) Assembly of human immunodeficiency virus (HIV) an-tigens on bacteriophage T4: A novel in vitro approach to construct multicomponentHIV vaccines. J Virol 80(15):7688–7698.

19. Milletti F (2012) Cell-penetrating peptides: classes, origin, and current landscape.Drug Discov Today 17(15-16):850–860.

20. Grimm D, et al. (2008) In vitro and in vivo gene therapy vector evolution via multi-species interbreeding and retargeting of adeno-associated viruses. J Virol 82(12):5887–5911.

21. Steinman RM, Banchereau J (2007) Taking dendritic cells into medicine. Nature449(7161):419–426.

22. Bonifaz LC, et al. (2004) In vivo targeting of antigens to maturing dendritic cells viathe DEC-205 receptor improves T cell vaccination. J Exp Med 199(6):815–824.

23. van Kooten C, Banchereau J (2000) CD40-CD40 ligand. J Leukoc Biol 67(1):2–17.24. Akerström B, Brodin T, Reis K, Björck L (1985) Protein G: A powerful tool for binding

and detection of monoclonal and polyclonal antibodies. J Immunol 135(4):2589–2592.25. Shen Z, Reznikoff G, Dranoff G, Rock KL (1997) Cloned dendritic cells can present

exogenous antigens on both MHC class I and class II molecules. J Immunol 158(6):2723–2730.

26. Yang L, et al. (2008) Engineered lentivector targeting of dendritic cells for in vivoimmunization. Nat Biotechnol 26(3):326–334.

27. Rappuoli R (2007) Bridging the knowledge gaps in vaccine design. Nat Biotechnol25(12):1361–1366.

28. Davtyan H, et al. (2010) DNA prime-protein boost increased the titer, avidity andpersistence of anti-Abeta antibodies in wild-type mice. Gene Ther 17(2):261–271.

29. Smiley ST (2008) Current challenges in the development of vaccines for pneumonicplague. Expert Rev Vaccines 7(2):209–221.

30. Leuschner F, et al. (2011) Therapeutic siRNA silencing in inflammatory monocytes inmice. Nat Biotechnol 29(11):1005–1010.

31. Miki K, et al. (2012) Bioengineered myocardium derived from induced pluripotentstem cells improves cardiac function and attenuates cardiac remodeling followingchronic myocardial infarction in rats. Stem Cells Transl Med 1(5):430–437.

Tao et al. PNAS | April 9, 2013 | vol. 110 | no. 15 | 5851

APP

LIED

BIOLO

GICAL

SCIENCE

S