Type VI secretion system sheaths as nanoparticles forantigen displayElena Del Tordelloa,1, Olga Danilchankaa,2, Andrew J. McCluskeya,3, and John J. Mekalanosa,4

aDepartment of Microbiology and Immunobiology, Harvard Medical School, Boston, MA 02115

Contributed by John J. Mekalanos, January 11, 2016 (sent for review December 20, 2015; reviewed by Jeffery F. Miller and Rino Rappuoli)

The bacterial type 6 secretion system (T6SS) is a dynamic apparatusthat translocates proteins between cells by a mechanism analogousto phage tail contraction. T6SS sheaths are cytoplasmic tubularstructures composed of stable VipA-VipB (named for ClpV-interactingprotein A and B) heterodimers. Here, the structure of the VipA/Bsheath was exploited to generate immunogenic multivalent particlesfor vaccine delivery. Sheaths composed of VipB and VipA fused to anantigen of interest were purified from Vibrio cholerae or Escherichiacoli and used for immunization. Sheaths displaying heterologousantigens generated better immune responses against the antigenand different IgG subclasses compared with soluble antigen alone.Moreover, antigen-specific antibodies raised against sheaths pre-senting Neisseria meningitidis factor H binding protein (fHbp)antigen were functional in a serum bactericidal assay. Our resultsdemonstrate that multivalent nanoparticles based on the T6SSsheath represent a versatile scaffold for vaccine applications.

vaccine delivery | type 6 secretion system | nanoparticles | antigen |immunogen

The bacterial type 6 secretion system (T6SS) is a dynamic ap-paratus that translocates proteins between effector cells and

target cells (1–4). It is conserved in 25% of Gram-negative bacteria,including Vibrio cholerae, Pseudomonas aeruginosa and Escherichiacoli. The T6SS plays a crucial role in bacterial pathogenicity andsymbiosis, targeting either eukaryotic cells or competitor bacterialcells (5). The assembled and functional T6SS apparatus has struc-tural homology to bacteriophage T4 phage tail components and canbe divided into two distinct assemblies: a contractile phage tail-likestructure and a transmembrane complex (6). V. cholerae VipA andVipB (named for ClpV-interacting protein A and B) and ortholo-gous proteins in other bacteria build within the cytosol of effectorcells a tubular sheath structure that is anchored to the various layersof the cell envelope through its association with the T6SS trans-membrane complex (7).VipA/B sheaths are composed of six protofilaments arranged

as a right-handed six-start helix similar to early T4 tail sheaths(8). Each protofilament is formed by a VipA/B heterodimer, andthe atomic-resolution structure of a native contracted V. choleraesheath has been recently determined by cryo-electron micros-copy (9). Stable expression of VipB in V. cholerae requires thepresence of VipA, and VipA/B heterodimers can be recruitedinto assembled tubular sheath structures spontaneously (10, 11).Because both ends of VipA are exposed on the external surfaceof the sheath tubules, a C-terminal fusion of VipA protein withsuperfolded green fluorescent protein (sfGFP) is functional in T6SSsheath assembly and activity, as previously demonstrated (3).Because these tubular structures are assembled in cytoplasm and

can be purified from bacteria (3), we explored the possibility thatT6SS sheaths could be used as a new particle-based delivery systemfor vaccine antigens. It is thought that particulate structures used forvaccine formulations are efficiently targeted for uptake by antigen-presenting cells (APCs) and interact directly with antigen-specific Bcells generating humoral responses (12). Although particulate pro-tein antigens may be more resistant to degradation, they areeventually proteolytically processed, and the resulting peptides arepresented by the major histocompatibility complex (MHC) class Iand class II molecules in a process that leads to activation of

CD4+ and CD8+ T-cell helper and effector responses. Exam-ples of particulate vaccine delivery systems include lipid-basedsystems [emulsions, immune-stimulating complexes (ISCOMs),liposomes, virosomes], polymer-based structures (e.g., nano-/microparticles), and virus-like particles (VLPs), with each ofthese systems presenting their own spectrum of advantages anddisadvantages for practical use as human immunogens (13).In this work, VipA/B sheaths displaying heterologous protein

antigens on the surface were generated and tested as a particulatevaccine antigen delivery system. Our results show that sheath-likestructures displaying different antigens were immunogenic andthat antibodies elicited against one of these, theNeisseria meningitidisfactor H binding protein (fHbp), were functional in a serumbactericidal assay. The T6SS antigen delivery system demonstratespotential as a multivalent particle to deliver one or more antigenssimultaneously into the same antigen-presenting cell. Moreover,the use of heterologous VipA and VipB sheaths displaying acommon antigen in sequential vaccine booster regimens minimizesimmune responses against the delivery system itself and focuses theimmune responses against the common antigen of interest.

ResultsGeneration of Sheath-Like Structures Exposing sfGFP on the Surface.A recombinant Duet vector system was used to enable assemblyof sheath structures made by VipA, VipB, and multiple VipAfusions (see SI Materials and Methods for details). One copy ofvipB and two copies of vipA DNA constructs were generated toexpress VipB, VipA with a C-terminal 6xHis tag (VipAhis), anda second VipA fusion protein displaying sfGFP as a C-terminal

Significance

Critical aspects that limit use of nanoparticles as vaccine deliverysystems include the technical difficulty of preparing particles un-der conditions that allow controlled protein loading and folding.Biologic nanoparticles offer advantages over chemically derivedparticles that require harsh organic solvents for preparation. Here,we demonstrate that the multimeric sheath structure of the bac-terial type 6 secretion system (T6SS) can be exploited to generateparticles that display on their surface many copies of specificproteins of interest. T6SS sheaths represent a functional vaccinedelivery system with remarkable flexibility due to the ability tobiochemically modify these structures to expand or enhancepotential immunogenicity.

Author contributions: E.D.T., O.D., A.J.M., and J.J.M. designed research; E.D.T., O.D., andA.J.M. performed research; E.D.T., O.D., A.J.M., and J.J.M. analyzed data; and E.D.T. andJ.J.M. wrote the paper.

Reviewers: J.F.M., University of California, Los Angeles; and R.R., GSK Vaccines.

The authors declare no conflict of interest.

Freely available online through the PNAS open access option.1Present address: GSK Vaccines, 53100, Siena, Italy.2Present address: Epiva Biosciences, Cambridge, MA 02139.3Present address: Abbvie Bioresearch Center, Worcester, MA, 01605.4To whom correspondence should be addressed. Email: john_mekalanos@hms.harvard.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1524290113/-/DCSupplemental.

3042–3047 | PNAS | March 15, 2016 | vol. 113 | no. 11 www.pnas.org/cgi/doi/10.1073/pnas.1524290113

Dow

nloa

ded

by g

uest

on

July

23,

202

1

extension. Sheath-like structures were purified from eitherV. cholerae 2740-80 or E. coli BL21-DE3 strains, using a two-stepprotocol. The first step applies ultracentrifugation to collect largerparticulate sheath structures, and the second step uses Ni-affinitychromatography to purify proteins associated with VipAhis (Fig.1A). Purified sheaths from either V. cholerae or E. coli were an-alyzed by negative staining using electron microscopy (EM). EManalysis showed that purified sheaths are organized into 40- to200-nm tubular structures (Fig. 1B) with an appearance thatcorresponded to contracted sheaths (3). Thus, VipA and VipB canautoassemble in tubular structures in the E. coli cytoplasm withoutthe need for other T6SS components not expressed in thisbacterial strain. Furthermore, sheaths were tested for the abilityto autoassemble in cell-free in vitro conditions. To this end,recombinant VipAhis, VipBhis, and VipA-sfGFPhis were purifiedunder denaturing conditions in the presence of urea, mixed, andallowed to refold. Tubular structures similar to typical sheathswere detected by EM analysis (Fig. S1). Surface sfGFP exposureon sheath particles was determined by immunogold staining withanti-GFP antibodies and Protein A-gold particles (Fig. 1C). Theseresults indicate exposure of sfGFP on the surface and the in-clusion of VipA-sfGFP fusion protein in each assembled sheath.

Interaction of Sheath Nanoparticles with Phagocytic Eukaryotic Cells.Because the size of purified sheath structures is comparable withother particles that can be engulfed by antigen presenting cells(APCs) (14, 15), we explored their interaction with phagocyticeukaryotic cells. Sheath-like structures exposing sfGFP (sfGFPsheaths) purified from V. cholerae or E. coli were incubated withRAW264-7 macrophages as an in vitro model for antigen uptake.Compared with a negative control, after 6 h of incubation, GFP-positive RAW264-7 macrophages were detected, indicating thatsheaths purified from V. cholerae or E. coli are taken up or asso-ciated to the surface of macrophages (Fig. 2A). The number ofpositive RAW264-7 cells also increased over time (Fig. 2A). Con-focal microscopy was used to determine localization of sheaths in-side of the eukaryotic cells (Fig. 2B). Z-stack analysis showed thepresence of sfGFP sheaths, on cell surface and inside the cells.These experiments demonstrated that sheath-like structures interactwith macrophages and therefore are likely to interact with otherAPCs in the context of immunization protocols.

Immunogenicity of Sheaths and Induction of an Immune ResponseAgainst the Exposed Heterologous Antigen sfGFP. Mice were im-munized with sfGFP sheaths as a proof-of-concept experiment todetermine whether a heterologous protein exposed on sheaths wasmore immunogenic compared with the soluble recombinant pro-tein alone. Groups of five BALB/c female mice were immunizedthree times biweekly with sfGFP sheaths purified from V. cholerae

or E. coli with or without alum as an adjuvant. In addition, twogroups of mice were immunized with “empty sheaths” fromV. cholerae or E. coli (i.e., formed only with VipA and VipB) mixedwith sfGFP recombinant protein to determine whether sheathshave adjuvant properties for “bystander antigens” that were nottightly associated with the particles. These experiments showedthat sfGFP sheaths were immunogenic and elicited a strong hu-moral immune response against sfGFP (Fig. 3A). No significantdifferences in total IgG ELISA titers were observed against sfGFPwhen exposed on a sheath’s surface or as a recombinant protein incombination with empty sheaths. In both cases, titers were similaror slightly higher compared with recombinant sfGFP alone.Because there was no significant difference between sheaths

isolated from the two host systems in terms of sheath structure/composition and immune response, E. coli was chosen as the pre-ferred host strain for sheath assembly and purification because it is(i) a common host for recombinant protein expression and (ii)amenable to scale-up processes that meet good manufacturingguidelines. To dissect the features of the immune response inducedby sheaths purified from E. coli, IgG subclass repertoires were de-termined by ELISA (Fig. 3B). IgG1 antibodies were the mostrepresented IgG, reflecting results obtained measuring total IgGand correlating with the use of alum (16). Interestingly, levels ofIgG2a and IgG2b were significantly higher when immunization wasdone with VipA/B/A-sfGFP compared with a recombinant sfGFPalone or in combination with empty VipA/B sheaths.Furthermore, we tested the feasibility to incorporate multiple

proteins in a single nanoparticle structure by coexpressing twofusion proteins, VipA-sfGFP and VipA-mCherry, together withVipAhis and VipB. Although the system requires optimization,our results showed that tubular structures containing both fusionproteins were assembled and were able to induce an immuneresponse against both sfGFP and mCherry (Fig. S2).

Use of Heterologous VipA-VipB Sheaths to Boost a Focused ImmuneResponse Against a Common Displayed Antigen. We explored theimmunogenicity of heterologous VipA-VipB sheaths displaying thesame antigen decoration (sfGFP) and tested whether boostingdrove the immune response selectively against the common anti-gen and not the scaffold proteins. VipA and vipB genes fromP. aeruginosa or Acinetobacter baylyi were used to express VipAhis,VipB, and VipA-sfGFP in E. coli. Resulting sheaths were analyzedby Coomassie blue staining and EM (Fig. S3 A and B). P. aerugi-nosa proteins produced homogeneous and relatively long sheathswhereas sheaths produced from A. baylyi proteins were shorter andless uniform. Flow cytometry confirmed that both sheath prepa-rations interacted with RAW264-7 cells and that a higher per-centage of GFP-positive cells was observed after incubation withA. baylyi sheaths compared with P. aeruginosa (Fig. S3C). Sheath

Fig. 1. Purification and characterization of T6SSsheaths from V. cholerae and E. coli. (A) PurifiedT6SS sheaths fused to sfGFP (1 μg) were separated bySDS/PAGE and visualized by Coomassie blue staining.All three sheath components—VipAhis, VipB, andVipA-sfGFP—were observed when purified fromeither V. cholerae (Vc) or E. coli (Ec). (B) Electronmicroscopy analysis of sheath preparations showsassembled tubular structures with a length of40–200 nm. (C) Immunogold staining by anti-sfGFPantibody at EM confirmed sfGFP exposure on thesheath surface.

Del Tordello et al. PNAS | March 15, 2016 | vol. 113 | no. 11 | 3043

MICRO

BIOLO

GY

Dow

nloa

ded

by g

uest

on

July

23,

202

1

preparations were used to immunize two groups of mice. The firstgroup was immunized with sfGFP sheaths made by V. choleraeVipA/B sequences as a reference condition. The second group wasimmunized with sfGFP sheaths prepared with V. cholerae VipA/Bfor the first injection, P. aeruginosa sfGFP-VipA/B for the secondinjection, and A. baylyi sfGFP-VipA/B for the third injection. Serawere collected 1 wk postimmunization, and ELISA was performedto detect antibodies against V. cholerae VipA and VipB (used asreference) and sfGFP. As shown in Fig. 3C, an immune responseagainst sfGFP was developed in both immunization groups. Incontrast, antibody titers against V. cholerae VipA and VipB weresignificantly lower for the group immunized with the three differ-ent sheath preparations, suggesting that the immune response wasfocused on the common displayed sfGFP antigen rather thanconserved epitopes that exist between the VipA and VipB proteinspresent in these heterologous sheaths. The initial boost withV. cholerae sfGFP-VipA/B sheaths resulted in higher antibodyresponse than boosting with P. aeruginosa sfGFP-VipA/B sheaths,suggesting that stimulation of VipA/VipB-specific CD4 T cells mayprovide B-cell help for the induction of sfGFP antibodies.

Use of Sheaths to Deliver Neisseria meningitidis fHbp Antigen. Totest the feasibility of using sheaths as a delivery system for morecomplex proteins, Neisseria meningitidis group B antigen factor Hbinding protein (fHbp) was used. The fHbp lipoprotein is asurface-exposed molecule characterized by a beta-barrel at its C

terminus (17), and it is currently a part of two recently licensedvaccines against meningococcal B disease (18, 19). The N ter-minus of fHbp was fused to the C terminus of V. cholerae VipAand expressed in E. coli together with VipAhis and VipB. Resultsshow fully assembled sheaths in the presence of fHbp fusionprotein (Fig. S4 A and B). Sheaths and recombinant fHbp wereused for immunization experiments, with alum as an adjuvant. Asa further control, one group of mice was also immunized withempty VipA/B sheaths mixed with recombinant fHbp protein.Total IgG ELISA titers showed that fHbp sheaths were immu-nogenic, with antibody levels comparable with the recombinantantigen alone (Fig. 4A). Interestingly, sheaths with incorporatedfHbp induced significantly higher immune response comparedwith recombinant fHbp mixed with VipA/B sheaths. IgG1 titersreflected the results obtained measuring total IgGs, as observedfor sfGFP sheaths. In the case of IgG2a and IgG2b, althoughthe differences were not statistically significant, fHbp sheathsinduced generally higher levels of antigen-specific antibodies,compared with other soluble immunogens (Fig. S4C).The functionality of antibodies raised against fHbp alone or

exposed on the surface of sheaths was tested by serum bacteri-cidal assay (SBA), which is currently used as an immunologicalcorrelate of protection for meningococcal vaccines (20). Increasingconcentrations of sera were incubated with N. meningitidis clinicalisolate NZ98/254 in the presence of an active baby rabbit comple-ment. Bacterial colony-forming units (cfu) corresponding to viable

Fig. 2. Sheath uptake by macrophages. (A) sfGFP sheathspurified from V. cholerae (Vc) or E. coli (Ec) were incubatedwith RAW264.7 macrophages for 6 or 20 h. The percentageof GFP-positive cells (highlighted green) increased overtime, as determined by FACS. (B) Confocal images of cellsincubated with sfGFP sheaths from E. coli after 20 h. (Up-per) The 3D reconstruction of the Z-stack image. Cellmembranes are visualized in red, and nuclei are in blue.(Lower) Two cross-sectional views of the vertical top orbottom section of cells, to show the presence of green dots,corresponding to sfGFP sheaths, inside the cells, as ob-served in xz and yz cross-section in smaller side panel.

3044 | www.pnas.org/cgi/doi/10.1073/pnas.1524290113 Del Tordello et al.

Dow

nloa

ded

by g

uest

on

July

23,

202

1

bacteria were counted after 1 h of incubation. The results showedsimilar SBA titers for sera obtained from mice immunized withfHbp sheaths, with or without alum as adjuvant, and higher titersthan sera obtained using recombinant protein alone or mixed withempty VipA/B sheaths (Fig. 4B). These results suggest that thedifferent IgG subclasses elicited by sheath particles were morefunctional in binding to the antigen on the N. meningitidis surface,fixing complement, and causing bacterial killing.

Modification of Sheath Surface by SrtA-Mediated Ligation. To broadenthe T6SS sheath platform for antigen delivery, we developed asystem that enables antigen fusion on preassembled, purifiedsheaths. Sortases are membrane-associated transpeptidases thatanchor proteins to the cell wall in Gram-positive bacteria (21) andhave recently been exploited to label a variety of proteins at the Cterminus, N terminus, or internal regions (22). To apply sortasetechnology to the T6SS sheath system, appropriately tagged proteinswere engineered to be substrates for sortase. First, VipA was mod-ified at its C terminus to contain the LPETGG sortase-recognitionmotif of Staphylococcus aureus sortase A (SrtA), and VipA-sort/VipB sheaths were purified from E. coli. Secondly, 6X-his-SUMO-(Gly)5-tagged versions of sfGFP, mCherry, and fHbp were generatedand purified as recombinant proteins (Fig. 5). SUMO protease wasused to cleave the SUMO tag, allowing release of (Gly)5-sfGFP (Fig.5, lane 5). When mixed with the VipA-sort/VipB sheaths, (Gly)5-sfGFP was recognized by a recombinant SrtA and ligated to theLPETGG motif exposed on the external surface of sheaths (Fig. 5,lane 6). After 10 min of incubation with SrtA, an extra band(∼45 kDa) corresponding to the predicted molecular mass of theVipA-sfGFP fusion protein was observed. This band was confirmedas VipA-sfGFP by Western blot analysis (Fig. S5A). The SrtA re-action was also performed by ligating either (Gly)5-sfGFP and (Gly)5-mCherry to the same reaction mixture, generating sheaths decorated

with both sfGFP and mCherry in equimolar amounts (Fig. 5, lane11). Similar results were obtained using (Gly)5-fHbp, demonstratingthe versatility of the reaction (Fig. S5 B and C).

DiscussionDespite decades of effort, there remains significant interest in de-velopment of new vaccine delivery systems that efficiently elicit aprotective immune response against pathogens and cancer, partic-ularly in multivalent formats (13). Recently, attention has beenfocused on nanoparticles where the vaccine antigen is either en-capsulated within the particles (i.e., liposome, ISCOMs, and poly-meric nanoparticles) or decorated on the surface of particles (i.e.,virus-like particles and nondegradable nanoparticles). In general,these particle-based approaches are thought to improve the im-munogenicity of antigens by increasing the influx of professionalAPCs into the injection site, promoting antigen uptake by APCs andprotecting antigens from early degradation. Critical aspects thatlimit use of these vaccine delivery systems include the technicalfeasibility of consistent particle preparation at reasonable yields,protein antigen stability during manufacture, and the toxicity asso-ciated with components present in some particle formulations.We hypothesized the structure of T6SS VipA/B sheaths could be

exploited to generate biologically derived, multivalent particles as ananoparticle delivery system for vaccine antigens. This study con-firmed the predicted utility. T6SS sheaths can be easily purifiedfrom E. coli bacterial cell lysates in a two-step protocol to generateparticulate structure with size ranging from 40 to 200 nm, optimalfor uptake by APCs (23). Sheath particles were immunogenic withincreased representation of IgG isotypes compared with solubleprotein antigens. Even in the absence of adjuvant, induction ofantigen-specific IgGs demonstrated functionality in a serum bacte-ricidal assay; this result suggests that different isotype classes havedifferent functionality despite recognizing the same antigen with

Fig. 3. sfGFP sheaths are immunogenic, and heter-ologous VipA-VipB pairs boost immune responsesagainst the displayed antigen. (A) The presence ofsfGFP-specific antibodies in serum from individualimmunized mice was determined by ELISA. Serumwas collected after immunization with recombinantsfGFP plus alum or the indicated sheath preparationwith or without alum. (B) The repertoire of IgGsubclasses in sera was measured by ELISA for miceimmunized with sfGFP sheaths from E. coli. Titers arereported as absorbency units at OD405 for serum di-lutions (dil.): dil. 1:32,000 for total IgG and IgG1; dil.1:4,000 for IgG2a, IgG2b, and IgG3. (C) Mice wereimmunized with either (i) three rounds of sfGFPsheaths containing V. cholerae VipA/VipB or (ii) se-quential injections of sfGFP sheaths containingV. cholerae VipA/VipB, P. aeruginosa VipA/VipB, orA. baylyi VipA/VipB. Sera were collected after eachimmunization (P1, post-1; P2, post-2; P3 post-3) andtested by ELISA to determine specific antibodiesagainst sfGFP, V. cholerae VipA, and V. choleraeVipB. *P value < 0.05; **P value < 0.01; ***P value <0.001; ****P value < 0.0001; ns, not statistically sig-nificant. If not specified, there was no statisticaldifference between samples.

Del Tordello et al. PNAS | March 15, 2016 | vol. 113 | no. 11 | 3045

MICRO

BIOLO

GY

Dow

nloa

ded

by g

uest

on

July

23,

202

1

different affinities. The particulate nature of T6SS sheaths is likelyresponsible for induction of a Th1-derived response, as proposedelsewhere (13). It will be interesting to dissect the immune responseinduced by sheath particles and determine the role of proin-flammatory pathways (i.e., the stimulation of NLRP3 inflamma-some) in the process. We expected sheaths to have adjuvant

properties, but our results showed that mixing recombinantprotein with empty sheaths did not result in a significant increasein immune response against the antigen. Thus, antigen fusion toVipA is critical to expose the protein with correct folding on thesheath surface to present immunogen to the immune system in amultivalent fashion similar to what happens when antigens aredisplayed on pathogen surfaces. In addition, sheath structuresmay enhance immune responses by retaining antigen at the siteof injection and reducing the rate of degradation. Althoughsuch “depotting” can also be achieved by conjugating the antigento gold or silica particles, our system has the advantage of pre-venting toxicity associated with the use of nondegradable material.Another key advantage of T6SS sheaths as a delivery system is

the potential for these structures to incorporate multiple pro-teins in a single nanoparticle and deliver them simultaneouslyinto the same APC, as we demonstrated using two VipA fusionproteins, VipA-sfGFP and VipA-mCherry. Alternatively, sheathsmay allow assembly of recombinant proteins separately purifiedunder denaturing conditions. The advantage of this approach isthat a variety of different VipA fusion proteins can in theory beexpressed and purified separately in an unfolded state, mixed inoptimal ratios based on their inherent immunogenicity, and as-sembled into multivalent sheath particles after refolding in vitro.In contrast with chemically defined nanoparticle structures, one

drawback of biologic-based vaccine delivery platforms is the like-lihood that an immune response will be generated against thedelivery system itself, potentially limiting use of the deliveryplatform to a single dose. Thus, a platform that minimizes thispredicted antisystem response would be highly desirable. We tookadvantage of the presence of T6SS in many Gram-negative bac-teria and, by using heterologous VipA-VipB pairs during boosting,we showed a focused immune response to the common displayedantigen target (i.e., sfGFP). This feature is potentially advanta-geous when selection of functional or high affinity antibodies isrequired in an application, as shown for HIV and RSV epitopesusing other display systems (24–26). It is also worth noting that,given their structural similarity to T6SS sheaths (8, 9), the sheathproteins of bacteriophage contractile tails could also be used toproduce heterologous sheaths for such sequential antigen boostapplications. It remains to be determined whether different VipA/Bpairs affect the folding of different protein antigens or preventdegradation of unstable or weak immunogens.T6SS can also be modified to carry other functional moieties.

We used sortase to covalently attach protein antigens (i.e.,sfGFP, mCherry, and fHbp) to the surface of preassembled,purified VipA/B sheaths in vitro under native conditions. Othergroups have exploited sortase in vitro to label proteins withchemical moieties, such as fluorescent probes (27, 28), lipids(29), sugars (30), and proteins (31). Sortase may be applied todecorate sheaths with appropriate chemical ligands or adhesinsthat might give sheaths binding activity for particular target cells,

Fig. 4. fHbp sheaths are immunogenic, and specific anti-fHbp antibodiesare functional in SBA. (A) Serum from individual mice was collected afterimmunization with fHbp sheaths, with or without alum, or recombinantfHbp. The presence of fHbp-specific antibodies was determined by ELISA.Total mouse IgG ELISA titers are reported as absorbency units at OD405 forthe same serum dilution 1:32,000. **P value < 0.01; ***P value < 0.001; ns,not statistically significant. (B) Mice sera were used in a serum bactericidalassay (SBA) using rabbit complement serum and N. meningitidis strain NZ98/254. SBA data are reported as the percentage of bacterial survival (cfu/mLafter 60 min of incubation/cfu/mL at time 0) for each serum dilution tested.Pooled sera for each group were used. The horizontal line represents theSBA titer as the reciprocal of serum dilution resulting in a 50% decrease inbacterial cfu/mL at the end of the assay compared with control cfu/mLat time 0. sfGFP sheaths (VipA/B/A-sfGFP) were used as negative controls.Titer = 16 is the limit of detection of the assay.

Fig. 5. Modification of sheath surface by SrtA-mediated ligation. Different reaction mixtures, wheresingle components were added sequentially, wereseparated by SDS/PAGE and visualized by Coomassieblue staining. (A) sfGFP was conjugated to VipA/Bsheaths by SrtA. Lane 6 displays an extra band corre-sponding to VipA-sfGFP of ∼45 kDa. (B) SrtA conju-gation was expanded to fuse mCherry (lane 10) or acombination of sfGFP and mCherry to VipA/B sheaths(lane 11). *, This band corresponds to VipA-sortaseintermediate of ∼40 kDa.

3046 | www.pnas.org/cgi/doi/10.1073/pnas.1524290113 Del Tordello et al.

Dow

nloa

ded

by g

uest

on

July

23,

202

1

such as APC or tumor cells. In this case, the desired functionaldomains can be expressed in different cell lines than the oneexpressing sheaths (e.g., in eukaryotic cells) and subsequentlyligated to the surface of sheaths. It is predicted that fusion ofsheaths with cell-penetrating peptides (32, 33) will increase theendocytic uptake of sheath particles and facilitate their endo-some escape, as proposed for TAT-peptides conjugated to al-bumin-based lyophilisomes (34). If such an approach can deliversheaths to the cytosol of APCs, it might be possible to decoratesheaths with epitopes for processing and presentation by the classI MHC pathway, or with enzymes that can synthesize adjuvantmolecules (e.g., DncV-related enzymes that could synthesize cyclicdi-nucleotides within target cells) (35, 36). This approach wouldincrease the potential of the T6SS sheath delivery system to in-duce a cytotoxic T-cell–mediated immune response.Future experiments will be required to optimize and control the

process of antigen exposure on sheaths to consistently generateuniform particles. Nevertheless, the versatility of the T6SS sheath-based delivery system may provide for a broad array of applicationsand advantages that are difficult to achieve using other describednanoparticle systems available up to now (13).

Materials and MethodsExpression and Purification of Recombinant Proteins. V. cholerae 2740-80ΔclpVΔflgGΔvipA and E. coli BL21-DE3 Gold strains were used to coexpressVipA, VipB, and VipA-antigen fusion proteins with the Duet ExpressionSystem. Plasmids and primers used for cloning strategies are listed in TablesS1 and S2. Sheath-like structures were purified from V. cholerae and E. colihost strains using a two-step purification method: (i) ultraspeed centrifu-gation of bacterial cell lysates to concentrate the sample and remove notassembled soluble proteins; and (ii) Ni-affinity chromatography purificationto purify sheaths carrying VipAhis.

Sheath Characterization. For negative staining electron microscopy, sampleswere loaded on grids, stained with 0.75% (wt/vol) uranyl formate, and imagedwith a transmission electron microscope. For immunogold staining, grids wereexposed to mouse primary antibody against the exposed antigen and rabbitanti-mouse bridging antibody, followed by protein A-gold particles. To testsheath uptake by phagocytic cells, RAW 264.7 macrophages were incubatedwith sfGFP sheaths, and, after 6 or 20 h of incubation, cells were collected andanalyzed by flow cytometry to quantify GFP-positive cells. Alternatively, afterincubation, cells were stained and imaged by confocal microscopy.

Immunization Experiments and Serology Analysis. The methods for animal ex-perimentation were approved by the Harvard and Boston University MedicalCenter Institutional Care and Use Committees. Groups of five BALB/c mice wereimmunized three times intraperitoneally with 20 μg of sheath preparations, withor without alum. Other groups were immunized with 5 μg of recombinant sfGFP,mCherry, or fHbp. An ELISA was designed to measure mouse IgG1, IgG2a, IgG2b,and IgG3 antibodies specific for each exposed antigen, and V. cholerae VipA andVipB, used as scaffold for sheath assembly. Serum bactericidal activity againstan N. meningitidis strain was evaluated as previously described (37).

SrtA Sortase Reaction. VipA/B sheaths, containing an SrtA recognition motif(LPETGG), were incubated with a 10-fold excess of Gly5-sfGFP, Gly5-mCherry,or Gly5-fHbp in a sortase reaction buffer, supplemented with SrtA enzyme.Samples were analyzed by SDS/PAGE and Western blot.

Further details for all of the experimental protocols are provided in SIMaterials and Methods.

ACKNOWLEDGMENTS. We thank Davide Serruto and Mariagrazia Pizza foruseful discussion. We thank Marek Basler for a gift of sfGFP and mCherryplasmids and for the V. cholerae 2740-80 ΔclpVΔflgGΔvipA strain. We thankthe Nikon Imaging Center and Electron Microscopy Facility at Harvard Med-ical School for help with confocal microscopy and electron microscopy anal-ysis. We thank Ann Thanawastien and her team at Matrivax for support inimmunization experiments. This work was supported by National Institutesof Health Grant AI-018045 (to J.J.M.) and Novartis Vaccines.

1. Pukatzki S, et al. (2006) Identification of a conserved bacterial protein secretion sys-tem in Vibrio cholerae using the Dictyostelium host model system. Proc Natl Acad SciUSA 103(5):1528–1533.

2. Leiman PG, et al. (2009) Type VI secretion apparatus and phage tail-associated proteincomplexes share a common evolutionary origin. Proc Natl Acad Sci USA 106(11):4154–4159.

3. Basler M, Pilhofer M, Henderson GP, Jensen GJ, Mekalanos JJ (2012) Type VI secretionrequires a dynamic contractile phage tail-like structure. Nature 483(7388):182–186.

4. Basler M, Mekalanos JJ (2012) Type 6 secretion dynamics within and between bac-terial cells. Science 337(6096):815.

5. Ho BT, Dong TG, Mekalanos JJ (2014) A view to a kill: The bacterial type VI secretionsystem. Cell Host Microbe 15(1):9–21.

6. Cascales E, Cambillau C (2012) Structural biology of type VI secretion systems. PhilosTrans R Soc Lond B Biol Sci 367(1592):1102–1111.

7. Durand E, et al. (2015) Biogenesis and structure of a type VI secretion membrane corecomplex. Nature 523(7562):555–560.

8. Kube S, et al. (2014) Structure of the VipA/B type VI secretion complex suggests acontraction-state-specific recycling mechanism. Cell Reports 8(1):20–30.

9. Kudryashev M, et al. (2015) Structure of the type VI secretion system contractilesheath. Cell 160(5):952–962.

10. Kapitein N, et al. (2013) ClpV recycles VipA/VipB tubules and prevents non-productive tubuleformation to ensure efficient type VI protein secretion. Mol Microbiol 87(5):1013–1028.

11. Bönemann G, Pietrosiuk A, Diemand A, Zentgraf H, Mogk A (2009) Remodelling ofVipA/VipB tubules by ClpV-mediated threading is crucial for type VI protein secretion.EMBO J 28(4):315–325.

12. Vidard L, et al. (1996) Analysis of MHC class II presentation of particulate antigens of Blymphocytes. J Immunol 156(8):2809–2818.

13. Gregory AE, Titball R, Williamson D (2013) Vaccine delivery using nanoparticles. FrontCell Infect Microbiol 3:13.

14. Foged C, Brodin B, Frokjaer S, Sundblad A (2005) Particle size and surface charge affectparticle uptake by human dendritic cells in an in vitro model. Int J Pharm 298(2):315–322.

15. Chithrani DB (2010) Intracellular uptake, transport, and processing of gold nano-structures. Mol Membr Biol 27(7):299–311.

16. McKee AS, et al. (2009) Alum induces innate immune responses through macrophageand mast cell sensors, but these sensors are not required for alum to act as an ad-juvant for specific immunity. J Immunol 183(7):4403–4414.

17. Cantini F, et al. (2009) Solution structure of the factor H-binding protein, a survival factorand protective antigen of Neisseria meningitidis. J Biol Chem 284(14):9022–9026.

18. Martin NG, Snape MD (2013) A multicomponent serogroup B meningococcal vaccineis licensed for use in Europe: What do we know, and what are we yet to learn? ExpertRev Vaccines 12(8):837–858.

19. Jiang HQ, et al. (2010) Broad vaccine coverage predicted for a bivalent recombinantfactor H binding protein based vaccine to prevent serogroup B meningococcal dis-ease. Vaccine 28(37):6086–6093.

20. Borrow R, Balmer P, Miller E (2005) Meningococcal surrogates of protection: Serumbactericidal antibody activity. Vaccine 23(17-18):2222–2227.

21. Mazmanian SK, Liu G, Ton-That H, Schneewind O (1999) Staphylococcus aureus sortase,an enzyme that anchors surface proteins to the cell wall. Science 285(5428):760–763.

22. Guimaraes CP, et al. (2013) Site-specific C-terminal and internal loop labeling ofproteins using sortase-mediated reactions. Nat Protoc 8(9):1787–1799.

23. Xiang SD, et al. (2006) Pathogen recognition and development of particulate vac-cines: Does size matter? Methods 40(1):1–9.

24. Guenaga J, et al. (2011) Heterologous epitope-scaffold prime:boosting immuno-focuses Bcell responses to the HIV-1 gp41 2F5 neutralization determinant. PLoS One 6(1):e16074.

25. Burton DR (2010) Scaffolding to build a rational vaccine design strategy. Proc NatlAcad Sci USA 107(42):17859–17860.

26. Correia BE, et al. (2014) Proof of principle for epitope-focused vaccine design. Nature507(7491):201–206.

27. Antos JM, et al. (2009) Site-specific N- and C-terminal labeling of a single polypeptideusing sortases of different specificity. J Am Chem Soc 131(31):10800–10801.

28. Popp MW, Antos JM, Grotenbreg GM, Spooner E, Ploegh HL (2007) Sortagging: Aversatile method for protein labeling. Nat Chem Biol 3(11):707–708.

29. Antos JM, Miller GM, Grotenbreg GM, Ploegh HL (2008) Lipid modification of proteinsthrough sortase-catalyzed transpeptidation. J Am Chem Soc 130(48):16338–16343.

30. Samantaray S, Marathe U, Dasgupta S, Nandicoori VK, Roy RP (2008) Peptide-sugarligation catalyzed by transpeptidase sortase: A facile approach to neoglycoconjugatesynthesis. J Am Chem Soc 130(7):2132–2133.

31. McCluskey AJ, Collier RJ (2013) Receptor-directed chimeric toxins created by sortase-mediated protein fusion. Mol Cancer Ther 12(10):2273–2281.

32. Brooks NA, Pouniotis DS, Tang CK, Apostolopoulos V, Pietersz GA (2010) Cell-penetrating peptides: Application in vaccine delivery. Biochim Biophys Acta 1805(1):25–34.

33. Hatakeyama H, et al. (2009) A pH-sensitive fusogenic peptide facilitates endosomalescape and greatly enhances the gene silencing of siRNA-containing nanoparticlesin vitro and in vivo. J Control Release 139(2):127–132.

34. van Bracht E, et al. (2014) Enhanced cellular uptake of albumin-based lyophilisomes whenfunctionalized with cell-penetrating peptide TAT in HeLa cells. PLoS One 9(11):e110813.

35. Kranzusch PJ, et al. (2014) Structure-guided reprogramming of human cGAS di-nucleotide linkage specificity. Cell 158(5):1011–1021.

36. Diner EJ, et al. (2013) The innate immune DNA sensor cGAS produces a noncanonicalcyclic dinucleotide that activates human STING. Cell Reports 3(5):1355–1361.

37. Giuliani MM, et al. (2005) The region comprising amino acids 100 to 255 of Neisseriameningitidis lipoprotein GNA 1870 elicits bactericidal antibodies. Infect Immun 73(2):1151–1160.

38. Martin DR, Walker SJ, Baker MG, Lennon DR (1998) New Zealand epidemic of meningo-coccal disease identified by a strain with phenotype B:4:P1.4. J Infect Dis 177(2):497–500.

39. Reichelt P, Schwarz C, Donzeau M (2006) Single step protocol to purify recombinantproteins with low endotoxin contents. Protein Expr Purif 46(2):483–488.

Del Tordello et al. PNAS | March 15, 2016 | vol. 113 | no. 11 | 3047

MICRO

BIOLO

GY

Dow

nloa

ded

by g

uest

on

July

23,

202

1