Flagellin Fusion Proteins as Adjuvants or Vaccines Induce Specific Immune Responses

Vaccination is the most efficient prophylaxis against a varietyof infectious diseases. New vaccination strategies rely on theincorporation of effective adjuvants, which stimulate the innateimmune response and, in turn, activate the adaptive immune response.It is well established that flagellin induces inflammatory responsesthrough the activation of antigen-presenting cells (APCs). Inorder to evaluate whether flagellin can serve as a carrier forthe development of adjuvants or vaccines, we prepared a flagellin-enhancedgreen fluorescent protein (EGFP) fusion protein. Our resultsdemonstrate that a flagellin-EGFP fusion protein is capableof stimulating APCs, resulting in the maturation of these cellsand secretion of proinflammatory cytokines. Furthermore, APCspulsed with the flagellin-EGFP fusion protein effectively processand present EGFP antigens. More importantly, animals immunizedwith the flagellin-EGFP fusion protein developed specific anti-EGFPT-cell responses. In contrast, recombinant EGFP was not ableto stimulate APCs, nor did it induce a T-cell response. Thus,recombinant-flagellin fusion proteins may be suitable carriersas adjuvants or vaccines for the development of new vaccinationstrategies to induce and boost immune responses against infectiousdiseases and cancer.

INTRODUCTION

Top

Introduction

Vaccination remains the most efficacious and valuable tool inthe prevention of infectious diseases (6, 7, 18). However, newvaccination strategies that induce the cellular and humoralcompartments of the immune response are needed for the developmentof effective prophylaxis against a number of infectious diseases.The immune system has two components: the innate immune response,which is a rather unspecific, first line of defense againstinfections (19), and the adaptive immune response, which representsspecific resistance to a wide variety of antigens, weak at first,but which develops in the long term into strong memory responses(15). Adaptive responses are initiated when T and B cells recognizevirtually any foreign molecule expressed on antigen-presentingcells (APCs) (37), through their unique T- and B-cell receptors,respectively. Innate responses are largely mediated by whiteblood cells such as neutrophils, monocytes, macrophages (M ${phi}$ ),and dendritic cells (DCs) that recognize common specific antigenicstructures found exclusively in microbial pathogens, termedpathogen-associated molecular patterns (PAMPs) (3).

Interestingly, the recognition of PAMPs by the innate immunesystem can stimulate the induction of adaptive immune responses(17). For example, DCs respond to some microbial product bytaking up the antigen and concurrently synthesizing a wide varietyof inflammatory mediators and cytokines that amplify the overallimmune performance. In addition, DCs can process and presentantigens, resulting in the activation of T- and B-cell responsesand the establishment of protective immunity. As such, a numberof microbial products are thought to function as effective adjuvantsdue to their effects on APCs, which in turn can boost the activationof an adaptive immune response. In this context, the identificationof safe and effective adjuvants is of crucial importance forthe development of effective vaccines. A number of novel adjuvantshave been reported during the past few years, including bacterialtoxins such as cholera toxin (25), endogenous immunomodulatorslike cytokines (33), bacterial wall components (45), and others(32).

Recent progress in understanding the innate immune responsesis beginning to yield insight into the elucidation of the receptorsand pathways involved and is leading to the characterizationof immunostimulatory molecules that enhance these processes(20, 31). More than a decade ago, Janeway postulated that recognitionof microbial PAMPs must be accomplished by specific receptorsin the APCs (21). Recent studies have demonstrated that recognitionof PAMPs by APCs is mediated by the Toll-like receptor (TLR)family (23, 28). There are currently 10 known TLRs capable ofsensing bacterial wall components such as lipopolysaccharide(LPS) (TLR-2/4), lipoteichoic acids (TLR-2/4), CpG DNA (TLR-9),flagellin (TLR-5), and others (42). TLRs are expressed in immatureand mature DCs, M ${phi}$ , and monocytes. All TLRs share a common intracellulardomain (similar to the one in interleukin-1 receptors) that,upon binding of PAMPs, initiates a signaling cascade in theAPCs through the adaptor protein MyD88 (43). The signal leadsto the activation of NF- ${kappa}$ B and mitogen-activated protein kinases(29) and ultimately triggers the maturation and activation ofAPCs (1), which include upregulation of major histocompatibilitycomplex (MHC) and costimulatory molecules and secretion of proinflammatorycytokines and chemokines (13). Maturation of APCs significantlyincreases their ability to prime naive T cells. In this way,TLRs link pathogen recognition with the induction of adaptiveimmune responses.

Recent evidence suggests that the flagellum, the structure thatconfers motility (5, 8, 16) to gram-positive and -negative bacteria,enhances their pathogenicity by promoting adherence to hosttissues and by directly activating host inflammatory signalingpathways (12). The component of the flagellum responsible foreliciting host immune responses is the filament protein flagellin(24). Purified or recombinant flagellin causes inflammatoryresponses on treated cells in vitro (22) or when systemicallyadministered in vivo (46). This effect is mediated by the activationof TLR-5 (14) present on target APCs. It has been reported thatflagellin is an effective adjuvant capable of enhancing T-cellresponses in vivo (9, 27). Moreover, it has been hypothesizedthat flagellin can be used as an adjuvant in vaccines to induceand boost immune responses. For example, there is evidence thatepitopes inserted into the flagellin of Salmonella and usedas live vaccines induce specific T-cell responses against theseepitopes (34, 35, 44). However, if the antigen incorporatedinto the flagellin interferes with the folding, transport, orassembly of the flagella, bacteria cannot present it on theirsurface (40), which is a major limitation to this approach.Furthermore, even the use of attenuated bacteria in vaccinationrepresents an unknown risk. For these reasons, we hypothesizedthat a recombinant fusion protein incorporating both the antigenof interest and the adjuvant (flagellin) would eliminate theseproblems. In addition, the fusion of both elements in one singlemolecule would have the advantage of targeting the antigen tothe APCs for uptake, processing and presentation. As a proofof concept we prepared a flagellin-EGFP fusion protein. We demonstratethat the flagellin-EGFP fusion protein activates APCs and iscapable of stimulating specific T-cell responses. These datasuggest that flagellin fusion proteins are suitable carriersfor the development of vaccines to induce strong and specificimmune responses against antigens of interest.

MATERIALS AND METHODS

Top

Materials and Methods

Mice and cell lines. BALB/c mice were purchased from Harlan (Indianapolis, Ind.)and housed under pathogen-free conditions. The cell lines BM-185-w.t.(wild type), BM-185 expressing EGFP (BM-185-EGFP), and BM-185expressing EGFP and CD80 (BM-185-EGFP-CD80) were kindly providedby D. Kohn at the University of Southern California, Los Angeles.All cell lines were maintained in complete RPMI medium (RPMI1640) supplemented with 10% fetal calf serum, 2 mM glutamine,5 x 10^–5 M 2-mercaptoethanol, and gentamicin (50 µg/ml).

Generation of a flagellin-EGFP fusion protein. A flagellin-EGFP chimeric protein was prepared by PCR amplificationof the entire fljB gene from Salmonella enterica serovar TyphimuriumLT2 and the EGFP gene. The products were inserted in frame intothe pRSET-A expression vector (Invitrogen) to form the chimericgene (Fig. 1). The pRSET-A-flagellin-EGFP plasmid was used totransform E. coli BL21. One selected colony was grown at 37°Cin Luria broth containing ampicillin (100 µg/ml). Stationary-phasebacteria were transferred to fresh broth, grown to an opticaldensity of $~$ 0.5 at 600 nm, and supplemented with 1 mM isopropyl-ß-D-thiogalactopyranoside(IPTG) to induce the fusion protein synthesis. After 4 h ofinduction, bacteria were harvested and resuspended in chilledSTE buffer (10 mM Tris [pH 8], 50 mM NaCl, 1 mM EDTA) containinglysozyme (100 µg/ml) and 0.2 mM phenylmethylsulfonyl fluorideand incubated 30 min on ice. Cells were lysed by two freeze-thawcycles followed by sonication, and the lysates were clearedby centrifugation (15,000 x g for 30 min). This recombinantHis-tagged flagellin fusion protein was purified by affinitychromatography on Ni-nitrilotriacetic acid agarose columns (Qiagen).Potential LPS contamination of the purified protein was neutralizedwith a polymyxin B column, and the levels of LPS were testedby the chromogenic Limulus amebocyte test. The proteins usedin the experiments described below do not contain any detectablelevels of endotoxin. Purified proteins were dialyzed againstPBS, and each preparation was analyzed by sodium dodecyl sulfate-polyacrylamidegel electrophoresis (Fig. 1) to check the size and purity ofthe protein, which was quantified by the Bradford assay. Forcontrol purposes, we prepared a recombinant EGFP protein usingthe same protocols (Fig. 1).

View larger version (29K):
[in this window]
[in a new window]
FIG. 1. Schematic representation of EGFP and flagellin-EGFP fusion protein. EGFP and flagellin-EGFP were synthesized as described in Materials and Methods. (A) Diagram of EGFP and flagellin-EGFP proteins. a.a., amino acids. (B) An aliquot of each protein was loaded onto a 10% polyacrylamide gel and subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis. Molecular weight markers (M.W.M) are indicated in thousands.

Isolation of APCs. Bone marrow cells from BALB/c mice were depleted of lymphocyteswith magnetic beads conjugated with anti-CD4, anti-CD8, andanti-B220 antibodies (Dynal, Oslo, Norway). The remaining cellswere cultured in complete RPMI medium containing 3% granulocyte-macrophagecolony-stimulating factor (GM-CSF) (supernatant from J558L cellstransfected with the murine GM-CSF gene, obtained from R Steinman,Rockefeller University, New York, N.Y.). The medium was changedevery second day, each time applying fresh RPMI medium containing3% GM-CSF. On day 8, the cells were ready to use. These cellsare a pool of activated APCs, including DCs, M ${phi}$ , and monocytes(data not shown).

Binding of flagellin-EGFP and EGFP proteins to APCs. We took advantage of the fluorescent properties of the EGFP(which are conserved in the chimeric protein upon fusion withflagellin) for detection and localization purposes. Bone marrow-derivedAPCs were recovered on day 8 and plated (10⁶) on Lab-Tek IIchambered coverglasses (Nalge Nunc Int., Naperville, Ill.).They were incubated in complete medium at 37°C for 1 h witha 10-µg/ml concentration of purified flagellin-EGFP orEGFP proteins. Then, APCs were washed twice with fluorescence-activatedcell sorter (FACS) buffer (PBS supplemented with 0.5% serumand 0.01% NaN₃), cell fluorescence intensity was analyzed byflow cytometry (FACScan; Becton Dickinson, San Diego, Calif.),and data analysis was performed using the CellQuest software.

Maturation of APCs induced by flagellin-EGFP fusion proteins. The maturation of pulsed APCs was confirmed by FACS analysisevaluating the levels of B7-1 and class II MHC expression. Forthese experiments 10³ freshly isolated APCs (not cultured withGM-CSF) were incubated with 10-µg/ml concentrations offlagellin-EGFP, EGFP, and LPS (used as control, known to inducethe maturation of APCs). After 24 h cells were washed and theexpression of class II MHC and B7.1 was evaluated by FACS analysis.To detect the expression of surface molecules, APCs were stainedwith a primary murine biotinylated antibody directed againstthe B7.1 (PharMingen, San Diego, Calif.) or class II MHC (Pharmingen)molecules, followed by incubation with fluorescein-conjugatedstreptavidin. Samples were analyzed by flow cytometry and dataanalysis was performed using the CellQuest software.

Measurement of nitric oxide (NO) and tumor necrosis factor alpha (TNF- ${alpha}$ ) production by APCs. NO^– was measured using the Griess reaction. Culture supernatants(100 µl) from stimulated APCs were added to 100 µlof 1% sulfanilamide-0.1% naphthylethylene diamine dihydrochloride-2%H₃PO₄ and incubated for 10 min at room temperature. The absorbanceat 562 nm was measured on an ELISA plate reader. A serial dilutionof NaNO₂ was used as standard. The limit of detection of thisassay was 10 µM NO₂^–. The production of TNF- ${alpha}$ wasdetermined with a commercial ELISA kit (R&D Systems) onthe same supernatants used for NO^– measurement.

Determination of processing and presentation of antigens derived from flagellin-EGFP fusion proteins. Cultured APCs (10⁶) were incubated with a 10-µg/ml concentrationof flagellin-EGFP, EGFP, or adenovirus-EGFP. After 24 h of incubationAPCs were washed and the cytotoxic activity of anti-EGFP-specificT cells (previously established in our laboratory) against thesepulsed cells was measured by a ⁵¹Cr release assay. Pulsed APCswere incubated with 150 µCi of [⁵¹Cr]sodium chromate for1 h at 37°C. Cells were washed three times and resuspendedin complete RPMI medium. For the cytotoxic assay, ⁵¹Cr-labeledtarget cells (10⁴) were incubated with various concentrationsof effector cells in a final volume of 200 µl in U-bottom96-well microtiter plates. Supernatants were recovered after6 h of incubation at 37°C and the percent lysis was determinedby the formula: percent specific lysis = 100 x (experimentalrelease – spontaneous release)/(maximum release –spontaneous release).

Stimulation of T-cell responses by a flagellin-EGFP protein. BALB/c mice were immunized with a subcutaneous injection offlagellin-EGFP or EGFP (50 µg/injection). As a control,animals were immunized with an adenovirus-EGFP vector. Two weekslater, spleens from primed animals were removed and splenocyteswere restimulated in vitro with BM-185-EGFP-CD80 cells. After5 days, CTLs were assayed for lytic activity. The BM-185-w.t.,BM-185-w.t. pulsed with the H2-K^d-HYLSTQSAL peptide (a K^d-EGFP-derivedimmunodominant epitope), and BM-185-EGFP cells were incubatedwith 150 µCi of [⁵¹Cr]sodium chromate for 1 h at 37°C.Cells were washed three times and resuspended in complete RPMImedium. For the cytotoxic assay we followed the same protocolas described above. For the CD4 proliferation assays, CD4⁺ Tcells were enriched from spleens of immunized animals and incubatedfor 4 days with APCs pulsed with adenovirus-EGFP. After 3 daysof incubation, tritiated thymidine (1 µCi/well for thelast 18 h) was added, and thymidine incorporation was measured.

RESULTS

Top

Results

Binding and activation of APCs by flagellin-EGFP proteins. It is well established that flagellin interacts with APCs (14).To test the antigenic and immunostimulatory properties of theflagellin-EGFP fusion protein, we first analyzed whether theflagellin-EGFP protein has the capacity to bind APCs. Bone marrow-derivedAPCs were incubated with a 10-µg/ml concentration of thepurified flagellin-EGFP or EGFP for 1 h. As demonstrated byFACS analysis (Fig. 2), almost 50% of the APCs incubated withflagellin-EGFP were fluorescent, whereas APCs incubated withEGFP showed that only 3% of the cells were fluorescent. Moreover,after only 1 h, some flagellin-EGFP is internalized into theAPCs, as indicated by the appearance of green fluorescence inthe cytoplasm, visualized under fluorescence microscopy (datanot shown). No internalization of the recombinant-EGFP was observedafter this time. Critical for any adjuvant is the effect itwould have on APCs. Maturation of APCs is a requirement forthe optimal stimulation of a T-cell response. Therefore, weanalyzed whether flagellin-EGFP would induce the maturationof APCs. When maturation occurs, APCs increase the expressionof surface molecules such as class II MHC and B7.1 (13). Freshlyisolated APCs were incubated overnight in the presence of purifiedflagellin-EGFP or EGFP proteins or commercial LPS (used as acontrol known to induce maturation of APCs) and the expressionof class II MHC and B7.1 molecules were analyzed. As shown inFig. 3, APCs treated with flagellin-EGFP, as well as those treatedwith LPS, induced the expression of these molecules, while treatmentof APCs with EGFP did not. We also evaluated the expressionof B7.2. Interestingly, we observed that after stimulation withthe flagellin-EGFP fusion protein, APCs also expressed B7.2,but to a lower extent than B7.1 (data not shown).

View larger version (16K):
[in this window]
[in a new window]
FIG. 2. APCs are bound by flagellin-EGFP but not by EGFP. Bone marrow-derived APCs were incubated for 1 h with flagellin-EGFP or EGFP (thick lines). Thin lines represent APCs left untreated, as controls. Cells were analyzed by flow cytometry for fluorescence in the FLH-1 channel. Numbers within the plots are the percentage of APCs that showed increased fluorescence. Experiments were repeated three times with similar results. One representative experiment is shown.

View larger version (17K):
[in this window]
[in a new window]
FIG. 3. Maturation of APCs is induced by LPS and by flagellin-EGFP, but not by EGFP. Freshly isolated APCs were incubated for 24 h with LPS (10 µg/ml), flagellin-EGFP, or EGFP (each at 10 µg/ml). APCs were stained to detect B7.1 (thick line) and class II MHC (broken line) molecules, or were left unstained as control (solid line), and analyzed by flow cytometry. Experiments were repeated three times with similar results. One representative experiment is shown.

A flagellin-EGFP fusion protein induces a proinflammatory response. After confirming that flagellin-EGFP binds and induces the maturationof APCs, we examined whether this adjuvant-antigen chimericprotein is also capable of stimulating APCs to produce a proinflammatoryresponse. To this end, the production of TNF- ${alpha}$ and NO was measured.TNF- ${alpha}$ and NO are known to be produced upon APC stimulation andto participate in the proinflammatory response (2). CulturedAPCs were incubated overnight with flagellin-EGFP, EGFP, andLPS. As shown in Fig. 4, APCs treated with flagellin-EGFP produceTNF- ${alpha}$ and NO to almost the same extent as APCs treated with LPS.In contrast, no production of TNF- ${alpha}$ or NO is detected after treatmentwith recombinant EGFP. Taken together these results demonstratenot only that the flagellin-EGFP fusion protein can bind specificallyto APCs, but also that this protein is capable of inducing thematuration and stimulation of APCs.

View larger version (21K):
[in this window]
[in a new window]
FIG. 4. A flagellin-EGFP fusion protein induces a proinflammatory response. Cultured APCs were incubated with a 10-µg/ml concentration of LPS, flagellin-EGFP, or EGFP or were left unstimulated (control). After 24 h the cell supernatants were collected and TNF- ${alpha}$ and NO secretion were measured. (A) TNF- ${alpha}$ production in each culture was measured by ELISA. (B) The same supernatants were used to analyze NO production, as measured by the Griess reaction. Data are the means of three independent experiments (performed with duplicates) (error bars, standard deviations). A significant (P $<=$ 0.001 [Student's t test]) difference was found between the control group and the flagellin-EGFP group.

APCs pulsed in vitro with flagellin-EGFP effectively present EGFP-derived peptides. For the induction of a specific immune response, antigens mustbe internalized, processed, and presented by APCs. To confirmthat the antigen in the flagellin-EGFP fusion protein was efficientlyprocessed and presented, the cytotoxic activity of specific-EGFPCD8⁺ T cells was tested against APCs that had been pulsed overnightwith (i) flagellin-EGFP, (ii) EGFP, or (iii) adenovirus-EGFP(as a positive control for comparison). As demonstrated in Fig.5, anti-EGFP-specific CD8⁺ T cells were able to effectivelykill APCs pulsed with flagellin-EGFP or adenovirus-EGFP vector,while no cytotoxic activity was observed on APCs pulsed withthe EGFP protein. No killing activity was observed on APCs thatwere not pulsed, indicating that the killing was specific. Theseexperiments demonstrate that antigens from flagellin-EGFP aremore effectively processed and presented by APCs than antigensfrom recombinant EGFP.

View larger version (20K):
[in this window]
[in a new window]
FIG. 5. APCs pulsed with flagellin-EGFP process and present EGFP. APCs cells were pulsed for 18 h with Ad-EGFP, flagellin-EGFP, or EGFP or were unpulsed. The cytotoxic activity of an anti-EGFP-specific CD8⁺-T-cell line was analyzed against the pulsed APCs in a 6-h ⁵¹Cr release assay at the indicated ratios of effector to target cells (E:T). Data are the means of two independent experiments (performed with triplicates) (error bars, standard deviations). A significant (P $<=$ 0.001 [Student's t test]) difference was found between the control group and the flagellin-EGFP group.

A flagellin-EGFP fusion protein induces specific T-cell responses in vivo. The preceding results demonstrate that a flagellin-EGFP recombinantprotein is bioactive and specifically activates APCs. Criticalfor the development of a vaccine is the induction of specificimmune responses in vivo. We next tested whether mice immunizedwith the flagellin-EGFP fusion protein developed specific T-cellresponses against EGFP. BALB/c mice were immunized with flagellin-EGFP,EGFP, and adenovirus-EGFP. For the evaluation of CD8⁺-T-cellresponses against EGFP, we utilized the identified EGFP restrictedH2-K^d-HYLSTQSAL peptide (11) and BM-185-EGFP cells (41). Asshown in Fig. 6, CD8⁺ T cells derived from mice immunized witheither adenovirus-EGFP (Fig. 6A) or flagellin-EGFP (Fig. 6B)induced comparable cytotoxic activities against both BM-185-EGFPcells and H2-K^d-HYLSTQSAL-pulsed BM-185-w.t. cells. No cytolyticactivity was observed against the parental BM-185-w.t. cells,demonstrating that the immune response is specific. In contrast,immunization with recombinant EGFP did not induce a CD8⁺-T-cellresponse (Fig. 6C).

View larger version (18K):
[in this window]
[in a new window]
FIG. 6. Induction of specific CD8⁺-T-cell responses by flagellin-EGFP fusion protein. BALB/c mice were immunized with adenovirus-EGFP (A), flagellin-EGFP (B) and EGFP (C). After 2 weeks spleens were removed, and T cells were stimulated in vitro for 5 days. Cytotoxic activity was tested against BM-185-w.t, BM-185-w.t. pulsed with the H2-K^d-EGFP-peptide, and BM-185-EGFP cells in a 6-h ⁵¹Cr release assay at the indicated ratio of effector cells to target cells (E:T). Data represent the means of results for three individually analyzed mice per group in two independent experiments (error bars, standard deviations). A significant (P $<=$ 0.001 [Student's t test]) difference was found between the control group and mice immunized with flagellin-EGFP.

It is well established that CD4⁺-T-cell responses are also criticalfor an optimal immune response. Next, we assayed whether animalsimmunized with the flagellin-EGFP fusion protein stimulate aCD4⁺-T-cell response. BALB/c mice were immunized subcutaneouslywith flagellin-EGFP or EGFP proteins. Two weeks later spleenswere removed and CD4⁺ T cells were enriched. CD4⁺-T-cell responseswere measured in a proliferation assay against APCs pulsed withadenovirus EGFP. Our data demonstrate that CD4⁺ T cells fromflagellin-EGFP-immunized mice proliferate after in vitro stimulationwhile minimal proliferation is observed on CD4⁺ T cells fromEGFP-immunized mice (Fig. 7). Taken together, these resultsdemonstrate that (i) the flagellin-EGFP fusion protein is phagocytosedwith high efficiency by APCs when compared to soluble antigen(EGFP) and (ii) the antigen fused to the flagellin was properlyprocessed and presented, which results in the induction of specificCD4⁺- and CD8⁺-T-cell responses.

View larger version (22K):
[in this window]
[in a new window]
FIG. 7. Activation of specific CD4⁺ T cells by flagellin-EGFP fusion protein. BALB/c mice were immunized with flagellin-EGFP or EGFP. After 2 weeks spleens were removed, and CD4⁺ T cells were enriched. CD4⁺ T cells were incubated for 4 days with APCs pulsed with adenovirus-EGFP or empty adenovirus. After 3 days of incubation, tritiated thymidine (1 µCi/well for the last 18 h) was added, and thymidine incorporation was measured. Data show the means of results for three individually analyzed mice per group in two independent experiments (error bars, standard deviations). A significant (P $<=$ 0.001 [Student's t test]) difference was found between the control group and mice immunized with flagellin-EGFP.

DISCUSSION

Top

Discussion

A major consideration in the design of vaccines is to overcomethe low immunogenicity of the antigen itself, which is commonlyachieved by coadministering antigens with an immunogenic carrieror adjuvant. In the development of efficient vaccines, a greatneed exists for the identification of safe and effective adjuvants.A number of microbial products are thought to function as effectiveadjuvants due to effects on APCs, which in turn promote theactivation of an immune response. In this regard, it has beendemonstrated that flagellin, which signals via the TLR-5 pathway(14), induces the activation of proinflammatory signals suchas the secretion of TNF- ${alpha}$ and NO (9). We hypothesize that flagellinfusion proteins can serve as adjuvants or vaccines for the inductionof specific immune responses. As a proof of concept we generateda flagellin-EGFP fusion protein. Herein, we demonstrate thatflagellin-EGFP is capable of inducing the maturation of APCsand the secretion of TNF- ${alpha}$ and NO. These proinflammatory signalselicited by flagellin-EGFP are specific and derived from theflagellin component, since treatment of APCs with recombinantEGFP does not stimulate these cells. These data demonstratethat the flagellin component of the fusion protein retains itsstimulatory effects. Moreover, we show that APCs loaded withthe flagellin fusion protein take up, process, and present theantigen, because they are specifically recognized by anti-EGFPT cells. More importantly, animals immunized with flagellin-EGFPdevelop specific CD8⁺- and CD4⁺-T-cell responses, while no immuneresponse is detected in animals immunized with recombinant EGFP.These results suggest that flagellin fusion proteins are suitablecarriers for the development of adjuvants or vaccines to induceand boost immune responses against the antigen of interest.

One of the advantages of using flagellin fusion proteins isthat it might be possible to effectively target APCs for thestimulation of specific immune responses. We do not yet understandthe mechanism through which flagellin fusion proteins are internalizedby the APCs. Sparwasser et al. (39) linked CpG-DNA to the solubleprotein OVA, and under these conditions the CpG-DNA facilitatesthe uptake of OVA by DCs through a receptor-mediated mechanismand, at the same time, these CpG-DNA complexes retain theirstimulatory potential. It is possible that not only does thebinding of flagellin-EGFP to TLR-5 on APCs trigger the activationof these cells, but also the internalization of the complexand subsequent processing and presentation of the antigen resultin the activation of an immune response. Further studies areneeded to understand the regulation and activation of APCs followinginteraction of TLR-5 with flagellin. Then it would be possibleto design more efficient vaccination strategies to achieve completeprotective immunity.

Recently, Means et al. showed that they could not detect expressionof TLR-5, by reverse transcription-PCR, on DCs isolated fromC57BL/6 mice (30). In their experiments, murine DCs were notresponsive to flagellin. However, Renshaw et al. (36) demonstratedthat murine peritoneal M ${phi}$ express TLR-5 and that treatment withflagellin induces the production of interleukin-6 and TNF- ${alpha}$ .Moreover, McSorley et al. demonstrated that flagellin is ableto boost CD4⁺-T-cell responses in vivo (27). Furthermore, Madrazoet al. found that murine osteoblasts are activated after stimulationwith flagellin (26). Sebastiani et al. also showed that TLR-5is expressed in lungs and liver (38). Taken together, theseresults indicate that TLR-5 is expressed in different tissuesand cell types, which are thus capable of interacting with flagellin.It is possible that the murine strain or the ex vivo cultureconditions of the DCs affect the expression of TLR-5 and theresponsiveness of these cells to flagellin and are responsiblefor the results obtained by Means et al. We do not discard thepossibility that murine DCs do not express TLR-5. In fact, inour experiments we used a pool of APCs derived from bone marrowthat in addition to DCs includes M ${phi}$ and monocytes, and thesecells have the capacity to respond to the stimulation with theflagellin-EGFP fusion protein. In addition, we cannot discardthe possibility that flagellin also binds to a receptor differentfrom TLR-5 and that the effect observed by the flagellin-EGFPfusion protein may not be exclusively mediated by the TLR-5.Nevertheless, our results demonstrate that the flagellin fusionprotein is active and capable of activating specific immuneresponses. We still need to understand how this flagellin-EGFPfusion protein activates the immune responses in vivo. Furtherstudies are necessary to elucidate the effect that flagellinand flagellin fusion proteins exert on the different APCs.

There are several reports demonstrating that certain epitopesinserted into the flagellin of Salmonella and used as live vaccinesinduce specific T-cell responses against these epitopes, whichfurther demonstrates the adjuvant abilities of flagellin (34,35, 44). The major drawback of this approach is that if theantigen incorporated into the fusion protein interferes withthe folding, transport, or assembly of flagellin, bacteria cannotexpress this molecule in its surface (40). Other advantagesfor the use of flagellin fusion proteins over epitopes insertedinto flagellin and used as live vaccines are as follows. First,fusion proteins eliminate the limitation of using known epitopesof restricted size. Instead, it will be possible to incorporatein one molecule single or multiple antigens of larger size,which promises very interesting possibilities. Second, recombinantfusion proteins will be rid of the safety risk associated withlive attenuated vaccines. Third, a very weak or no significantantiflagellin immune response appears to be developed upon administrationof flagellin (4). A major concern for the efficacy of an adjuvantis that prior or subsequent exposures to the adjuvant may inducean immuno-neutralizing response against it, resulting in a diminishedor suppressed immune response against the antigen of interest.In this regard, Ben-Yedidia and Arnon (4) demonstrated thatprior exposure to flagellin did not result in an antiflagellinresponse and did not suppress the immune responses against influenzaantigens. We have preliminary data showing that immunizationwith the flagellin-EGFP induces a minimal immune response againstflagellin (data not shown), further supporting the findingsof Ben-Yedidia and Arnon. A simple explanation as to why a strongantibody response is not induced against flagellin could bebecause hosts are tolerant or partially tolerant to flagellin.It is well established that a natural flora of bacteria existsin the body very early in life. As such, we might have developedsome level of tolerance over time whereby T- or B-cell responsesare not induced or are minimally induced against the flagellin.However, the interaction of flagellin with TLR-5 expressed onDCs, monocytes and other cells induces the activation of thesecells. Finally, another advantage of using flagellin fusionproteins is that less antigen and less adjuvant might be neededrelative to other vaccination methods, because the adjuvant(flagellin) would also target the fusion protein to the APCs.

Even though only weak responses are generated against the flagellin,it is still possible that any antiflagellin response could diminishthe full potential of flagellin-derived vaccines. Flagellincontains conserved structures at the N- and C-terminal endsand an intervening hypervariable region (10). It has been hypothesizedthat the binding of flagellin to the TLR-5 is mediated by therecognition of the conserved structures in the flagellin protein.In this manner, TLR-5 expressed on APCs can guarantee the recognitionof any flagellin. Recently, Eaves-Pyles et al. (10) demonstratedthat the constant regions of flagellin, but not the hypervariableregion, were capable of inducing a proinflammatory responsein cultured human monocytes. The results indicate that the conservedregion of flagellin contains the binding domain to TLR-5. Theidentification of flagellin minimal binding domains will facilitatethe construction of smaller effective molecules, allowing easiertissue penetration and perhaps evoking milder immune responsesagainst flagellin. The construction of other flagellin fusionproteins containing other antigens (e.g., against infectiouspathogens) will determine the capacity of these proteins toprovide protective immunity.

In conclusion, have we shown that flagellin fusion proteinsare capable of stimulating APCs and, more importantly, thatthey have the capacity to induce specific immune responses.Therefore, flagellin fusion proteins are suitable carriers forthe development of adjuvants or vaccines to induce and boostspecific immune responses.

ACKNOWLEDGMENTS

We thank D. Kohn (University of Southern California, Los Angeles)for providing the cell lines BM-185, BM-185-EGFP, and BM-185-EGFP-CD80and Lynn Zacker for administrative assistance.

This work was supported by NIH grant CA 78759 (to J.L.). M.M.was supported by NIH grants AI34829 and AI52237.

FOOTNOTES

* Corresponding author. Mailing address: Joseph Lustgarten, Sidney Kimmel Cancer Center, 10835 Altman Row, San Diego, CA 92121. Phone: (858) 410-4190. Fax: (858) 450-3251. E-mail: jlustgarten{at}skcc.org.

Editor: D. L. Burns

REFERENCES

Top