|Home | About | Journals | Submit | Contact Us | Français|
The use of Lactococcus lactis to deliver a chosen antigen to the mucosal surface has been shown to elicit an immune response in mice and is a possible method of vaccination in humans. The recent discovery on Gram-positive bacteria of pili that are covalently attached to the bacterial surface and the elucidation of the residues linking the major and minor subunits of such pili suggests that the presentation of an antigen on the tip of pili external to the surface of L. lactis might constitute a successful vaccine strategy. As a proof of principle, we have fused a foreign protein (the Escherichia coli maltose-binding protein) to the C-terminal region of the native tip protein (Cpa) of the T3 pilus derived from Streptococcus pyogenes and expressed this fusion protein (MBP*) in L. lactis. We find that MBP* is incorporated into pili in this foreign host, as shown by Western blot analyses of cell wall proteins and by immunogold electron microscopy. Furthermore, since the MBP* on these pili retains its native biological activity, it appears to retain its native structure. Mucosal immunization of mice with this L. lactis strain expressing pilus-linked MBP* results in production of both a systemic and a mucosal response (IgG and IgA antibodies) against the MBP antigen. We suggest that this type of mucosal vaccine delivery system, which we term UPTOP (for unhindered presentation on tips of pili), may provide an inexpensive and stable alternative to current mechanisms of immunization for many serious human pathogens.
Pili of Gram-positive bacteria are filamentous structures that extend outward from the bacterial surface and are covalently anchored to the bacterial cell wall. They are believed to be the primary means of attachment to the appropriate environmental receptor for the organism, which, for pathogens, is within the human host. The backbone of the pilus in Gram-positive bacteria is composed of multiple covalently linked identical subunits (major pilin), to which one or more minor pilin subunits are covalently attached. Pilin proteins are synthesized with an N-terminal Sec signal, which is cleaved during transit through the cytoplasmic membrane, and a C-terminal cell wall sorting signal (CWSS), which contains an LPXTG (or similar) amino acid motif, followed by a hydrophobic region and a positively charged C terminus. Pilus assembly is catalyzed by a pilus-specific sortase family transpeptidase, which cleaves the CWSS motif between the threonine (T) and glycine (G) residues and forms a covalent bond between this T and a conserved lysine (K) residue of another major pilin subunit. As this process repeats, the pilus is polymerized until it is covalently linked to the cell wall by either the “housekeeping” sortase, which is responsible for anchoring most surface proteins of Gram-positive bacteria to the cell wall, or the pilus-specific sortase (for reviews, see references 21, 35, and 38).
We have been investigating assembly of T3 pili of Streptococcus pyogenes, an important human pathogen. In this organism, the T3 pilus locus (19) encodes the major pilin (T3) and the minor pilins Cpa and OrfB, the pilus-specific transpeptidase SrtC2, and SipA2, which is required for pilus polymerization by SrtC2 (44). Our investigations into the biogenesis of T3 pili have identified the residues of T3 and Cpa required for (i) polymerization of T3 and (ii) incorporation of Cpa into the pilus structure. We have demonstrated that lysine residue 173 (K173) (29) and the CWSS (QVPTG) of the T3 major pilin subunit (2, 29) are required for polymerization of T3. This indicates that individual T3 subunits are polymerized into the pilus structure by covalent bonds between K173 of T3 and the threonine of the CWSS (T315) of the adjacent T3 subunit. We have also demonstrated that K173 of T3, along with the CWSS (VPPTG) of Cpa, are required for incorporation of the minor pilin, Cpa, into the pilus (29). Thus, the K173 residue of T3 is required for T3-T3 linkage and is also required for covalent linkage of Cpa to the T3 pilus, demonstrating that Cpa is located at the tip of T3 pili, a conclusion supported by immunogold electron microscopy (EM) (29).
Identification of the residues required for attachment of Cpa, the tip protein, to the T3 pilus suggested to us that genetic engineering could be used to produce a Gram-positive bacterial strain in which a foreign protein would be covalently linked by the bacterium to the pilus tip in place of Cpa. In the present study, we used the Escherichia coli maltose-binding protein (MBP) as a model protein to test this idea. We identified amino acid residues of the primary structure of Cpa that are sufficient for incorporation of a foreign protein into T3 pili in vivo by SrtC2. We propose that this approach constitutes a novel technology for presentation of foreign polypeptides external to the bacterial envelope, which we call UPTOP (for unhindered presentation of polypeptides on tips of pili). We suggest that any Gram-positive bacterium can be used as the host for UPTOP. We also propose that UPTOP can be used to present vaccine antigens to the immune system. As proof of this principle, we constructed a strain of the probiotic bacterium Lactococcus lactis engineered to produce T3 pili with the model protein MBP covalently linked at the pilus tips. We show in this study that mucosal administration to mice of this vaccine strain generates both an IgG and an IgA response to the model protein.
E. coli strains were cultured in LB media (34) supplemented with the appropriate antibiotic. Strains TOP10 (Invitrogen) and XL10-Gold (Stratagene) were grown at 30°C and strain BL21-CodonPlus (DE3)-RIL (Stratagene) at 37°C. L. lactis strain MG1363 was cultured without shaking at 30°C in M17 media (Oxoid) supplemented with 0.5% glucose (GM17). MG1363 was made competent by the method of Holo and Nes (14). Kanamycin and ampicillin were used at concentrations of 50 and 100 μg/ml, respectively, for E. coli. Spectinomycin was used at a concentration of 100 μg/ml for both E. coli and L. lactis.
Overlap PCR was performed according to the method of Ho et al. (12), using the primers shown in Table S1 in the supplemental material. In the first round of PCR, the template pJRS9550 (29), encoding a portion of the FCT-3 pilus gene cluster (Fig. (Fig.1)1) from S. pyogenes strain AM3, was used with the primers CpaHA_F1_BamHI and MalE_Cpa_N_O_Anti to amplify a 198-bp 5′ region of cpa which extends from 30 nucleotides upstream of the Cpa initiation codon to 168 bases past the start of the Cpa open reading frame. This region includes the Cpa ribosomal binding site (RBS) and encodes the first 56 amino acid residues of Cpa (fragment 1). Plasmid pJRS9550, along with the primers MalE_Cpa_C_O_Sense and SrtC2_R_XhoI, was used to amplify the 3′ 2,800-bp region of the FCT-3 pilus gene cluster starting from the codon for amino acid 594 of Cpa, SipA2, T3, and SrtC2 (fragment 2). Similarly, plasmid pJRS9550, along with the primers MalE_Cpa_C_O_Sense and Orf100_R_XhoI, was used to amplify the 3′ 2,102-bp region of the FCT-3 pilus gene cluster starting from the codon for amino acid 594 of Cpa, SipA2, and T3 (fragment 3). Plasmid pMalp4E (New England Biolabs) was used, along with the primers MalE_Cpa_N_O_Sense and MalE_Cpa_C_O_Anti, to amplify a 1,095-bp sequence of malE, encoding residues 31 to 393 of the E. coli MPB (fragment 4). PCRs were DpnI digested, and fragments were purified by agarose gel electrophoresis, followed by extraction from the gel slice by using a Qiagen gel extraction kit.
In the second round of PCR, purified fragments 1, 2, and 4 were combined by using primers CpaHA_F1_BamHI and SrtC2_R_XhoI, resulting in a region encoding the MBP_Cpa fusion protein (MBP*), SipA2, T3, and SrtC2. Similarly, purified fragments 1, 3, and 4 were combined by using the primers CpaHA_F1_BamHI and Orf100_R_XhoI, resulting in a region encoding the MBP_Cpa fusion protein (MBP*), SipA2, and T3. These regions were cloned into pCR2.1TOPO (Invitrogen), resulting in pEU7943 and pEU7944, respectively (Fig. (Fig.1).1). The pilus cluster regions of pEU7943 and pEU7944 were then subcloned into the shuttle vector pJRS9508 (1) using BamHI and XhoI, resulting in pJRS9565 and pJRS9566, respectively (Fig. (Fig.1).1). Expression of the MBP* pilus gene cluster regions in pJRS9565 and pJRS9566 is under the control of the strong constitutive P23 promoter (1), and it is translated from the Cpa RBS.
Overnight cultures of L. lactis strain MG1363 were washed once and concentrated 10-fold in saline. Cell wall extraction was performed using four cell units, where one cell unit/ml corresponds to 1 ml of cell culture with an optical density at 600 nm (OD600) of 2.0 (4), in lysis buffer (50 mM Tris-HCl [pH 6.8], 30% raffinose, 4 mg of lysozyme/ml, 400 U of mutanolysin/ml, and Roche complete protease inhibitors) (7) at 37°C for 3 h with gentle rotation. Cell wall and supernatant fractions were prepared for further analysis as previously described (4, 44).
SDS-PAGE and Western blot analysis were performed as previously described (29). T3 typing serum was provided by B. Beall (CDC, Atlanta, GA). Mouse monoclonal anti-MBP antibody, used at a dilution of 1:2,000, was from New England Biolabs.
A PCR fragment generated by using primers SipA2_BamHI_Sense and T3_XhoI_Anti (see Table S1 in the supplemental material) with the template pEU7657 (Fig. (Fig.1)1) was ligated in frame with the C-terminal His6 coding sequence of pET21(+) (Novagen), using BamHI and XhoI restriction endonucleases. The resulting plasmid, pEU7957 (Fig. (Fig.1),1), was transformed into BL21-CodonPlus (DE3)-RIL (Stratagene), and T3 was purified by using the B-PER His6 fusion protein purification kit (Pierce) according to the manufacturer's instructions.
Purified anti-T3 antibody was obtained from the T-typing antiserum by adsorption to a polyvinylidene difluoride (PVDF) membrane using a slight modification of the method of Ritter (31). After transfer to a PVDF membrane and overnight blocking (29), a membrane fragment containing one lane of purified T3 was stained with PVDF staining solution (50% methanol 0.05% Coomassie brilliant blue R 250) to visualize T3. After destaining (50% methanol), the membrane fragment was used to identify the location of T3 in the unstained portion of the membrane, which was excised by using a clean razor blade, followed by incubation with rabbit anti-T3 antiserum for 1 h at room temperature with orbital rotation. After four 5-min washes with TBS (20 mM Tris-HCl [pH 7.6], 137.5 mM NaCl) at room temperature, bound anti-T3 was eluted with 200 mM glycine (pH 11.5), 150 mM NaCl at room temperature for 30 min, immediately neutralized by combination with an equal volume of 1 M Tris-HCl (pH 6.5), and stored at 4°C until use.
Overnight cultures of MG1363 were washed in saline and analyzed by dot blotting with the appropriate antibody as previously described (4).
Eight cell units of MG1363 overnight cultures (prepared as described above) were resuspended in lysis buffer (50 mM Tris-HCl [pH 6.8], Roche Complete protease inhibitors, lysozyme at 4 mg/ml, and mutanolysin at 400 U/ml) and incubated for 30 min at 37°C. The samples were then sonicated twice at 4°C for 15 s each time, with 15-s pauses between sonications, followed by centrifugation at 13,000 × g for 5 min at 4°C. Supernatants were transferred to a new tube and recentrifuged to remove debris, and a sample of this supernatant, corresponding to crude lysate, was saved for later analysis. The clarified lysate was incubated at 4°C for 1 h with gentle inversion every 15 min with amylose resin (New England Biolabs), which had been washed and preequilibrated with column wash buffer (20 mM Tris-HCl [pH 7.4], 200 mM NaCl, 1 mM EDTA, 1 mM dithiothreitol). Reactions were centrifuged at 1,000 × g for 20 s at 4°C, and a sample corresponding to the flowthrough fraction was stored for later analysis. After washing the resin five times with one column volume of wash buffer at 4°C, bound protein was eluted in column wash buffer containing 25 mM maltose. Samples were then heated to 100°C for 10 min in SDS sample buffer (33).
L. lactis strain MG1363/pJRS9545 (vector control), MG1363/pJRS9550 (expressing T3 pili with an influenza virus hemagglutinin [HA] epitope-tagged Cpa), or MG1363/pJRS9565 (expressing T3 pili with the MBP* fusion) was grown as described above, harvested by centrifugation, washed with phosphate-buffered saline (PBS), and then adsorbed to polyvinyl formal-carbon-coated grids (E. F. Fullam) for 2 min and fixed with 1% glutaraldehyde for 1 min. The grids were washed twice with PBS, blocked with PBS plus 1% bovine serum albumin (BSA) and then incubated for 1 h with a 1:200 dilution (in PBS plus 1% BSA) of either the rabbit polyclonal anti-T3 antibody described above, a sheep polyclonal anti-MBP antibody (Berkeley Antibody Company), or the mouse monoclonal anti-HA (clone HA-7) antibody (Sigma-Aldrich). The grids were washed three times with PBS and then incubated for 1 h with a 1:50 dilution (in PBS plus 1% BSA) of either anti-rabbit IgG antibody, anti-sheep IgG antibody, or anti-mouse IgG antibody conjugated to 12-nm-diameter colloidal gold particles (Jackson Immunoresearch Labs). The grids were washed three times with PBS and twice with water and then negatively stained with 0.5% phosphotungstic acid (Ted Pella) for 35 s. The grids containing the negatively stained bacteria were examined on an FEI TECNAI 12 BioTwin G02 microscope (FEI) at an 80-kV accelerating voltage. Digital images were acquired with an AMT XR-60 charge-coupled device digital camera system (Advanced Microscopy Techniques).
Cells (MG1363/pJRS9545 or MG1363/pJRS9565) grown at 30°C in GM17 containing 100 μg of spectinomycin/ml, were washed, and resuspended in PBS to give 5 × 107 CFU/μl. Female CD1 mice were vaccinated intranasally (i.n.) by administration of 20 μl of cell suspension (109 CFU) into the nostril. The mice were vaccinated every 10 days with a dose of 109 CFU for three consecutive days (i.e., the animals were vaccinated on days 1, 2, 3, 14, 15, and 16 and on days 27, 28, and 29). Blood samples collected on day 39 were analyzed. The mice were sacrificed on day 39, and lung lavage fluids were obtained postmortem by inserting a nylon cannula into the exposed trachea, which was tied in place. A 1.0-ml syringe was used to inject and withdraw 1 ml of 0.9% sodium chloride solution three times, the supernatants were then stored at −80°C.
A 96-well EIA/RIA microplate (Costar; Corning, Inc.) was coated overnight at 4°C with 100 ng of MBP per well. The coated plate was blocked with 5% soy milk in PBS-Tween to prevent nonspecific binding. Serum (1:50 dilution) or lung fluid was reacted with the coated wells for 60 min. Antibody production was detected by using anti-mouse IgG or anti-mouse IgA secondary antibodies coupled to alkaline phosphatase (Sigma). Absorbance was measured at 405 nm after 45 min after the addition of p-nitrophenyl phosphate hexahydrate disodium salt (pNPP) tablets dissolved in diethanolamine buffer solution (KPL). Endpoint (day 39) antibody titers were determined in pooled lung lavage and serum samples by using a similar enzyme-linked immunosorbent assay (ELISA), except that the mouse samples were applied from a serial dilutions to the plate. Antibody titers were calculated as the dilution producing the same OD405 at two times the background level (the reading obtained with sera or lung fluid of nonimmunized mice).
Purified MBP and mouse anti-MBP (New England Biolabs), alkaline phosphatase (AP)-conjugated rabbit anti-mouse IgG1 and anti-IgG2a (Invitrogen), AP-conjugated anti-mouse IgG (Sigma), and purified His6-tagged T3 (described above) were used in the ELISA reactions.
The IgG subclass profile in the mouse sera was visualized by using a Rapid isotyping kit (Pierce) according to the manufacturer's instructions. In short, the serum samples were diluted (1:8,000) and applied into the cassette well. The gold conjugates embedded in the cassette form specific subclass soluble complexes with the antibody in the sample. These complexes travel the length of the membrane and are resolved on the membrane that is impregnated with anti-isotype antibody. The results are then displayed as a red band indicating the antibody isotype.
The endpoint (day 39) titers of MBP-specific IgG1 and IgG2a in pooled serum samples were determined by using the ELISA with immobilized MBP and isotype-specific secondary antibodies. The serum samples were pooled and applied from a serial dilution to the plate. Antibody titers were calculated as the dilution producing the same OD405 at two times the background level (the reading obtained with sera or lung fluid of nonimmunized mice).
Incorporation of a protein into the pilus structure requires the presence of an N-terminal Sec signal and a C-terminal CWSS (21, 35, 38). Therefore, to attach the E. coli MBP at the tip of T3 pili, the mature MBP protein (lacking its native Sec signal) was fused between the Sec signal of Cpa and the C terminus of Cpa, including its CWSS. Because mature Cpa is predicted to contain an intramolecular isopeptide bond (17), and the role of this bond in pilus morphogenesis is not known, the C-terminal region of Cpa present in our construct (encoded by pJRS9565 [SrtC2+] and pJRS9566 [SrtC2−]; Fig. Fig.1)1) includes both residues expected to form this bond (K599 and N704 of the unprocessed Cpa sequence). After processing by the leader peptidase and the pilus-specific sortase, the mature form of the resulting fusion protein, MBP*, should have the first 11 residues of the mature Cpa protein (N terminus), followed by the mature MBP protein, followed by 119 amino acids from the C terminus of mature processed Cpa. The predicted molecular mass of the mature MBP* is 54 kDa.
To determine whether L. lactis expresses the MBP* protein on its surface, intact MG1363/pJRS9565 cells spotted on membranes were reacted with monoclonal anti-MBP antibody and, separately, with polyclonal anti-T3 antiserum (“dot blots”). Strain MG1363/pJRS9566, which lacks SrtC2, was used as a negative control. The dot blots show that the intact cells reacted with both antibodies, indicating that both MBP* and T3 are exposed on the bacterial surface (Fig. 2A and B).
To evaluate incorporation of MBP* into pili on the L. lactis surface, cell wall fractions and concentrated culture supernatants of strains MG1363/pJRS9565 (MBP*), MG1363/pJRS9566 (−SrtC2), and MG1363/pJRS9545 (containing the empty vector) were analyzed by Western blotting. Monomeric MBP* (apparent molecular mass of 54 kDa) can be seen in the extracts from the SrtC2 deletion mutant, MG1363/pJRS9566 (Fig. (Fig.2C,2C, lanes 3 and 7), and it shows slight reactivity with the polyclonal anti-T3 antiserum (Fig. (Fig.2D),2D), most likely due to the C-terminal residues of Cpa in MBP* and/or to the known cross-reactivity of this antiserum with Cpa (20). A band that reacted with both monoclonal anti-MBP antibody and polyclonal anti-T3 antiserum was visible at the location expected for the MBP*-T3 heterodimer (80 kDa) in the strain expressing the T3 operon up to srtC2 (Fig. (Fig.22 lanes 1, 2, 5, and 6) and not in the SrtC2 deletion mutant derivative (Fig. (Fig.2,2, lanes 3 and 7), as expected. In extracts from two separate clones of the experimental strain, MG1363/pJRS9565, high-molecular-weight (HMW) polymers, which are characteristic of pili in Gram-positive bacteria (21, 35, 38), were also visible, whereas these were absent from the srtC2 deletion control, as expected (Fig. 2C and D). The reactivity of the HMW bands with anti-MBP indicates that MBP* was incorporated into the pilus structure. The higher-molecular-mass bands showed less reactivity with anti-MBP, and the opposite was seen with anti-T3. This is expected since there are many T3 subunits per pilus but only one MBP* subunit on the tip of each pilus (29). Similar to the cell wall extracts, concentrated culture supernatants from MG1363/pJRS9565 analyzed with anti-MBP and anti-T3 also showed the HMW ladder characteristic of pili (Fig. 2C and D, lanes 5 and 6), whereas those of MG1363/pJRS9566 (lacking SrtC2) showed only monomeric forms of MBP* (Fig. (Fig.2C,2C, lane 7).
We confirmed that the MBP protein was properly assembled into the T3 pili by using immunogold EM. L. lactis strains MG1363/pJRS9550 and MG1363/pJRS9565 (see Fig. Fig.1),1), expressing T3 pili with HA-tagged Cpa and MBP*, respectively, assembled large amounts of pilus fibers that were abundantly labeled with the anti-T3 antiserum (Fig. 3B and C). This is in contrast to the vector control strain, MG1363/pJRS9545, which lacked pili and was not labeled by the anti-T3 antiserum (Fig. (Fig.3A3A).
Labeling of MG1363/pJRS9565 (MBP* T3 pili) with an anti-MBP antiserum confirmed that MBP* was incorporated into the pili (Fig. 3H and I). In the negative control, the anti-MBP antiserum did not label bacteria expressing pili lacking the MBP* fusion (MG1363/pJRS9550; Fig. Fig.3G).3G). A polyclonal anti-MBP antiserum was used for these experiments, as the monoclonal anti-MBP antibody did not react well with the bacteria under the immuno-EM conditions. Few MBP* subunits were present in the pili compared to the number of T3 subunits (compare Fig. 3H and I to Fig. Fig.3C),3C), and the MBP* was localized at what appeared to be the pilus tips (arrows in Fig. 3H and I), although the flexible and intertwined nature of the pili prevented definitive localization. The labeling pattern of MBP* matches the localization pattern of Cpa in T3 pili expressed in S. pyogenes (29), suggesting proper incorporation of MBP* in the L. lactis T3 pili. To confirm that MBP* localized similarly to Cpa in the T3 pili assembled by L. lactis, we labeled strain MG1363/pJRS9550, expressing HA-tagged Cpa, using an anti-HA antibody. As shown in Fig. 3E and F, labeling of the HA-tagged pili by the anti-HA antibody closely matched the appearance of pili labeled by the anti-MBP antiserum (arrows in Fig. Fig.3).3). In the control, the anti-HA antibody did not label the MBP* pili, which lack the HA epitope (MG1363/pJRS9565; Fig. Fig.3D).3D). Overall, the immunogold EM results show that MBP* was correctly incorporated into the T3 pili and displayed similarly to the native Cpa minor pilin, which was previously determined to be located at the pilus tips (29).
Because it was possible that incorporation of MBP into pili would lead to its misfolding, we used the ability of pili containing MBP* to bind to an amylose resin to evaluate the activity of MBP. Lysates of MG1363/pJRS9565 (encoding MBP*), MG1363/pJRS9550 (the parental plasmid encoding intact Cpa), and MG1363/pJRS9566 (lacking SrtC2) were applied to an amylose resin, followed by analysis of the crude lysate, the flowthrough and the eluate fractions by Western blotting with anti-MBP antibody (Fig. (Fig.4A)4A) and with purified anti-T3 antiserum (Fig. (Fig.4B4B).
As expected, no signal was detected with anti-MBP in any of the fractions of strain MG1363/pJRS9550, which does not encode this protein (Fig. (Fig.4A,4A, lanes 7 to 9). In the control strain (MG1363/pJRS9566), no polymeric forms of either MBP* (Fig. (Fig.4A)4A) or T3 (Fig. (Fig.4B)4B) were detected, a finding consistent with the requirement of SrtC2 for T3 pilus polymerization and incorporation of MBP*. HMW pilus forms were detected by both the anti-MBP antibody and the purified anti-T3 antiserum in the eluate fraction of MG1363/pJRS9565, which encodes MBP*. In the control extracts from the strain with native Cpa on the T3 pili (wild type), HMW pilus forms (detected with anti-T3) were present in the crude extract and in the flowthrough fraction but were not visible in the eluate fraction, indicating that these pili had not bound to the amylose resin (Fig. (Fig.4B4B lanes 7 to 9). This shows that the binding of pili containing MBP* to the amylose resin is due to the incorporation of a functional MBP molecule and is not a result of interactions between the wild-type pilus subunits and the amylose bead matrix. Thus, we conclude that MBP* present in the T3 pili retains the activity of MBP, i.e., ability to bind amylose, which requires correct folding of the protein (36).
The display of a correctly folded foreign antigen at the tip of a surface-exposed pilus by a probiotic bacterium may provide a new strategy for antigen delivery by mucosal vaccination. To examine the potential of S. pyogenes pili as vectors for antigen presentation, we investigated the immune response of mice to mucosal administration of live L. lactis expressing pili with MBP* on their tips (strain MG1363/pJRS9565) compared to the same strain carrying the empty vector (MG1363/pJRS9545). An arbitrary dosage and protocol for vaccine administration were chosen (see Materials and Methods), and the response in mucosal secretions was determined in lung lavage, which was collected 10 days after the third antigen inoculation. The presence of MBP-specific IgA in the lavage was determined by using an ELISA with immobilized MBP (Fig. (Fig.5).5). A measurable and statistically significant reaction with immobilized MBP was demonstrated in all of the mice that were inoculated with bacteria expressing the recombinant pili (P < 0.001). The IgA response to MBP in undiluted serum was at least three times higher than the background in most of the animals, and the MBP-IgA endpoint titer (day 39) in pooled lavage samples was 1:10. Only background activity (the same as in uncoated wells) was found in lung lavage from untreated mice or mice inoculated with bacteria containing the empty vector (MG1363/pJRS9545). Therefore, vaccination with L. lactis expressing the recombinant pilus elicited a specific IgA response to MBP in the mucosal secretions.
Antibody response to MBP was also investigated in the serum of the vaccinated mice. A strong reaction with the immobilized MBP was noted in serum samples from all of the mice that were inoculated with L. lactis producing the recombinant pilus with MBP* at the tip (MG1363/pJRS9565, P < 0.0001, Fig. Fig.6).6). The endpoint titer (day 39) of the MBP-specific IgG in the pooled serum samples was 1:30,000. Untreated mice or mice vaccinated with bacteria expressing the empty vector (MG1363/pJRS9545) demonstrated only background level activity.
The immune response to the T3 major pilus subunit was examined by using an ELISA with the immobilized T3 pilin protein. The T3-specific IgA antibody in undiluted lavage samples was at least 12 times higher than the background in all of the animals vaccinated with bacteria expressing the recombinant pili (MG1363/pJRS9565, P < 0.0001, Fig. Fig.7).7). Little or no reaction was seen in samples from naive mice or mice immunized with bacteria carrying the control vector. The presence of T3-specific IgG antibody was also determined in pooled sera by using the same ELISA (Fig. (Fig.8).8). As with the lung lavage samples, a strong reaction with the T3 pilin was observed in serum from mice that were immunized with bacteria expressing the recombinant pili (MG1363/pJRS9565), whereas only low reactivity with the T3 protein was found in serum from the control animals (MG1363/pJRS9545, P < 0.0001). The endpoint titer in the pooled sera is 1:120,000. Together, these results show that i.n. vaccination with L. lactis expressing T3 pili with MBP* results in a significant T3-specific IgA response in secretion and a strong systemic IgG response. The T3 pilin immunity generated both in secretions and systemically was significantly stronger than the response to MBP, probably because while there is only one MBP subunit per pilus, there are hundreds of T3 subunits.
The profile of IgG isotypes in serum from vaccinated mice was compared to that found in pooled serum from untreated mice by using isotyping cassettes containing strips impregnated with anti-IgG1, -IgG2a, and -IgG2b antibodies. This analysis demonstrated that vaccination with L. lactis expressing the recombinant pilus resulted in a shift in the serum IgG profile to an increased prevalence of IgG2a (data not shown). The levels of MBP-specific IgG1 and IgG2a subclasses in the serum were determined by ELISA using isotope-specific secondary antibody. The titer of IgG1 and IgG2a in the vaccinated mice was found to be 1:5,000 and 1:25,000, respectively. Therefore, i.n. administration of L. lactis displaying MBP* on the tip of the T3 pilus elicited a systemic IgG response that was dominated by IgG2a.
This study had two main goals: (i) development of a system (UPTOP) for presentation of a polypeptide external to the envelope of a Gram-positive bacterium and (ii) demonstration that this system could be used to generate an immunogenic mucosal vaccine delivery system. These two objectives are discussed separately below.
The use of bacteria as “nanoparticles” to present polypeptides is currently generating significant interest. Bacterial particles are being developed to present enzymes to improve bioremediation, to engineer better probiotic organisms, as additives to improve nutritional supplements for animals, and for many other uses. In addition, expression of antigens on the surface of bacteria is one approach being investigated for development of vaccine delivery vectors. Usually, these foreign proteins are attached directly to the bacterial surface: either the cytoplasmic membrane or the cell wall. However, antigens attached directly to the surface of live vaccine delivery vectors may be partially occluded by the bacterial envelope, since these proteins will not extend outwardly beyond any capsular material, S layers, or abundant protein on the cell surface and thus may have limited exposure to the environment. In agreement with this idea, a recent study using a live bacterial vector for vaccine antigen delivery determined that a stronger immune response resulted when the protective antigen was moved further from the bacterial surface by inserting a linker between it and the site attached to the bacterial cell wall (5). To overcome the limitation on exposure of the foreign protein that occurs when it is linked directly to the bacterial surface, we have developed the new technology UPTOP, which allows “unhindered presentation of a polypeptide on the tip of pili.”
Pili, which are found on many bacteria, usually extend beyond the cell envelope of the organism and serve as the first contact between the bacterium and its environment. Thus, they often serve the role of adhesins that attach the bacterium to its specific niche. In Gram-positive bacteria, pili are covalently attached to the bacterial cell wall, which prevents their removal by washing even under extreme conditions. Polymerization of pili in S. pyogenes requires only one enzyme, the pilus-specific sortase, and one additional protein, SipA2 (44). In the T3 pilus of this organism, the protein Cpa is located on the pilus tip, and the threonine residue in its CWSS is linked to the major pilin subunit, T3 (29). In the present study, we capitalized on this finding to determine which residues of Cpa could be replaced by those of a model foreign protein, MBP. We showed that this model protein was covalently attached to the T3 pili expressed in the foreign host L. lactis. At the C terminus of the chimeric protein MBP*, in addition to the Cpa CWSS, we included 119 amino acids of the Cpa protein so as to retain the predicted intramolecular isopeptide bond of Cpa (17). We found that deletions in this sequence prevented efficient incorporation of the chimeric MBP-Cpa protein into pili in L. lactis (data not shown), suggesting that residues N-terminal to the CWSS may be required for incorporation of the model protein into the T3 pilus.
Both for presentation of an antigen and of an enzyme, it is essential that the foreign protein on the pilus tip be correctly folded. The mechanism used to fold proteins on the exterior of the cell wall of Gram-positive bacteria is not completely understood. Because of this, and because covalent linkage to the T3 pilus might require a non-native conformation of the foreign protein, we felt it important to evaluate the structure of the foreign protein. The use of MBP as the model foreign protein allowed us to assess this, since the active site of MBP consists of amino acids located in a cleft formed between the two different domains in the correctly folded protein (36). We determined here that the MBP on the T3 pili retains its activity, as measured by its ability to bind amylose. This indicates that MBP is most likely presented on the pilus tip in its native conformation.
L. lactis is an attractive delivery vehicle for mucosal vaccines. The majority of infections are initiated at mucosal surfaces, where some pathogens remain restricted to the mucosal membranes and others penetrate the epithelium and spread throughout the body. An effective mucosal immune response requires production of both secretory IgA and serum IgG. Mucosal IgA can form a barrier to pathogens at the mucosal surface by preventing the initial attachment of the pathogen and its infiltration of the surface layers or by binding to and neutralizing toxins that the pathogen produces. IgA defense is especially important for surfaces that cannot be reached effectively by serum IgG antibody (15, 27). Systemic IgG production supplements the mucosal defense provided by IgA and reduces the ability of the pathogen to cross mucosal membranes and spread within the body (3, 27, 43).
The properties of lactic acid bacteria (LAB) make them good candidates for live vaccine delivery vehicles. These bacteria are food-grade organisms, used in the production of fermented food products such as cheese and yogurt, and they have been safely consumed by humans for centuries. In addition, they are considered to be probiotics, i.e., live microorganisms believed to confer a health benefit on the host when administered in sufficient quantities (43). Thus, LAB, including L. lactis, have GRAS (i.e., generally regarded as safe) status and can be administered orally to people.
LAB are also attractive vaccine vectors because they have been found to have natural adjuvant activity (23, 41). In addition, L. lactis is a natural Toll-like receptor agonist (10, 18, 25, 39) that can stimulate the production of various interleukins, which can increase the antigen-specific immune response. Several mucosal immunization studies using live (7-9, 11, 22, 24, 28, 40, 42) or killed (32) LAB delivery systems expressing cytoplasmic, secreted, or cell wall-associated antigens have demonstrated both systemic and mucosal immune responses (for a review, see reference 43). Protection against challenge with pathogens was also demonstrated in a number of studies. Examples of pathogens against which protection was observed include the bacteria S. pyogenes (22), Streptococcus pneumoniae (11, 24, 28, 40), group B Streptococcus spp. (7), and enterotoxigenic E. coli (42).
LAB vaccines currently being investigated present the vaccine antigen to the mucosal surface as a cytoplasmic, cell wall anchored, or secreted protein (for a review, see reference 43). We describe here a novel method for the delivery of a vaccine antigen to the mucosal surface using UPTOP technology. We engineered the L. lactis delivery vector to produce the vaccine antigen covalently linked at the tips of pili, which are anchored covalently to the cell wall of the bacterium. Thus, although the antigen is covalently attached to the bacterium, since the pili produced are long, the antigen is exposed external to the bacterial envelope and thus is positioned to interact with maximal effectiveness with the environment, i.e., the immune system. This is particularly advantageous for B cells that can see the antigen directly, independent of processing and presentation by antigen-presenting cells. When the genes required for pilus production are expressed from a strong promoter, as we did in this work, each bacterium has hundreds of pili with the antigen on their tips, providing multiple copies of the antigen to which the immune system is exposed.
We have shown here that mucosal administration of this L. lactis strain expressing pilus-linked MBP* resulted in a detectable MBP-specific IgA response in the lung lavage of immunized mice. Therefore, this delivery system is effective in presenting protein antigens and triggering a mucosal response, at least in the upper respiratory tract. Since i.n. immunization can produce a significant IgA response in the cervicovaginal mucosa in addition to that in the upper airway mucosa (16, 26), it is possible that i.n. administration with this vaccine will induce an adaptive response in the genital tract as well.
In addition to the observed IgA response, a strong IgG response to MBP was detected in blood as a result of the course of i.n. immunization. The systemic response to MBP is very encouraging because serum antibody contributes significantly to the mucosal defense, especially in the lower respiratory and in the genitourinary pathways, where the epithelia are permeable to serum antibody (13). In addition, a serum humoral response can prevent the systemic spread of invasive pathogens. The observed mucosal and systemic response to the T3 pilin, in addition to the immunity mounted against MBP, suggests that the UPTOP system may allow the presentation of more than one antigen at a time. We have shown that insertion of the 9-amino-acid HA epitope within the T3 protein does not prevent pilus polymerization (2, 29, 44) and that mice inoculated with L. lactis expressing these HA tagged pili generate an HA-specific immune response (B. R. Quigley, Z. Eichenbaum, and J. R. Scott, unpublished results). Thus, it is possible that protective epitopes may be engineered into the T3 protein. In addition, it should be possible to use related but serologically different pili, if UPTOP is to be used for presentation of different vaccine antigens on successive occasions.
Analysis of pooled sera from our immunized mice for IgG subclass revealed a significant bias toward IgG2a (with a ratio of the titer of IgG2a to IgG1 of 5) (data not shown). The limited variation in the response of individual mice for both MBP and T3 antigens (Fig. (Fig.55 to to7)7) suggests that the IgG1 and IgG2 titers in the pooled serum are an accurate representation of the population and are not likely to be driven by the response in one or two animals. The IgG2 dominant response suggests that the adaptive T-cell response to the pilus-linked MBP could be predominantly T helper 1 (Th1) type. Similarly, a predominant Th1 response was also seen when other live L. lactis vaccines were administrated at mucosal surfaces (11, 30). Th1 cells contribute to the humoral response by supporting the production of IgG2a, while inhibiting the formation of other IgG subclasses, such as IgG1. In addition, Th1 cells produce gamma interferon (IFN-γ) and interleukin-2 (IL-2) and therefore promote a cellular immune response, which includes macrophage activation, delayed-type hypersensitivity, and T-cell cytotoxicity. Based on recent work with S. pneumoniae, the relative elevation of the IgG2a subtype is likely to produce greater protection at the mucosal surface than a predominantly IgG1 response (11). Because the IgG2a/IgG1 ratio seen in our experiments seems high, we suggest that in addition to the effect of the live L. lactis bacteria in generating an effective immune response, the properties of the pili on their surface may also have contributed to the skewing of the immune response.
The model vaccine strain described here encodes the UPTOP expression system on a plasmid that also encodes an antibiotic resistance marker. However, because the genome sequence of L. lactis is available (6) and because this organism is amenable to genetic manipulation, the UPTOP pilus gene cluster can be stably integrated into the bacterial chromosome without an antibiotic resistance marker. This would minimize the number of foreign genes introduced and avoid possible problems of the environmental spread of an antibiotic resistance plasmid.
In addition, several opportunities exist for the possible improvement of immunogenicity of our L. lactis-based vaccine. For example, increasing the number of bacteria administered per dose and/or the number of pili expressed per bacterium should result in an increase in the immune response and should increase protection. For an L. lactis-based GBS vaccine, mice were immunized with 1011 CFU/dose, i.e., 100 times the dose used in our study, without adverse effects (7). In addition, the frequency of administration of the vaccine and the fluid in which the vaccine strain is administered can be optimized, and the vaccine can also be delivered orally. It is also possible that the outcome might be improved by coexpression of IL-2 or IL-6 with the L. lactis vaccine, since this was described to increase the serum response 10- to 15-fold for the tetanus toxin fragment C antigen (37). Thus, future work will be directed at optimizing the delivery system and testing an antigen that can generate protective immunity against a mucosal pathogen.
In summary, in the present study we identified residues of the pilus tip protein that can be replaced by an antigen of choice using the E. coli MBP as a model. Replacement of these residues results in the stable attachment of the model protein to the bacterium, since the linkages are all covalent. We have also shown that a strain of L. lactis can be engineered to incorporate this model protein of 54 kDa (MBP*) into T3 pili. In addition, we demonstrated that mucosal administration of this vaccine strain resulted in a detectable MBP-specific IgG and IgA response. We conclude that UPTOP is a promising strategy for the presentation of polypeptides covalently attached to the surface of bacteria but external to the bacterial envelope and that it might be used to develop effective oral vaccines for protection against mucosal pathogens.
We thank Bernard Beall (CDC, Atlanta, GA) for providing the anti-T3 T typing serum and Kirsten Baecher and Stacey Miles for help with C-terminal deletion experiments. We thank Susan Van Horn and the Central Microscopy Imaging Center (Stony Brook University) for assistance with the EM.
This study was funded in part by a grant from the Georgia Research Alliance (J.R.S. and Z.E.) and grants AI055621 and GM062987 (D.G.T.) and AI05605 (J.R.S.) from the National Institutes of Health.
Editor: A. Camilli
Published ahead of print on 22 December 2009.
†Supplemental material for this article may be found at http://iai.asm.org/.