Nanopatterning Protein Antigens to Refocus the Immune Response

Abstract

Vaccines for many important diseases remain elusive, and those for others need to be updated frequently. Vaccine efficacy has been hindered by existing sequence diversity in proteins and by newly-acquired mutations that enable escape from vaccine-induced immune responses. To address these limitations, we developed an approach for nanopatterning protein antigens that combines the site-specific incorporation of non-canonical amino acids with chemical modification to focus the immune response on conserved protein regions. We demonstrated the approach using green fluorescent protein (GFP) as a model antigen and with a promising malarial vaccine candidate, Merozoite surface protein 1₁₉ (MSP1₁₉). Immunization of mice with nanopatterned MSP1₁₉ elicited antibodies that recognized MSP1₁₉ from heterologous strains, differing in sequence at as many as 21 of 96 residues. Nanopatterning should enable the elicitation of broadly protective antibodies against a wide range of pathogens and toxins.

Graphical Abstract

Nanopatterning is used to shield targeted protein epitopes, enabling the immune response to be refocused to conserved regions on the protein.

An approach to protein modification that has been successfully used in the clinic involves functionalization with polyethylene glycol (PEG)¹. This approach, however, has primarily been used to eliminate protein immunogenicity. Another approach to influence protein immunogenicity, that is ubiquitous in nature, involves the modulation of glycosylation^{2, 3}. Glycans have been shown to shield neutralizing epitopes in viruses including HIV^{3, 4}. The removal or addition of glycosylation sites has also been used to modulate the immunogenicity of recombinant protein antigens^{2, 5}.

There are, however, several limitations associated with a glycosylation-based approach to modulating the immunogenicity of antigens. First, the ability to modify glycosylation is hindered by the importance of this post-translational modification for the proper folding and oligomerization of protein antigens. Secondly, glycans can themselves be part of neutralizing epitopes. For instance, in the case of HIV, many broadly neutralizing antibodies directly target glycans or bind to epitopes partly composed of glycans⁶. Finally, glycosylation is a blunt tool, with limited ability to vary the structure of glycan shields. Thus, one needs a technique that is orthogonal to glycosylation and can be applied to non-glycosylated protein antigens expressed in bacterial hosts. To that end, we developed an approach termed nanopatterning (Scheme 1), to control on the nanometer scale the chemistry and topography of the protein surface and its accessibility to components of the immune system. Nanopatterning combines the orthogonal reactivity enabled by the site-specific incorporation of non-canonical amino acids^{7, 8} with the chemical diversity provided by functionalization with synthetic molecules – including but not limited to PEG.

Scheme 1.

Approach for nanopatterning protein antigens.

While many protein antigens contain some highly conserved regions, these regions are often immunologically subdominant, whereas variable regions are often immunodominant. Nanopatterning can be used to shield these variable regions, allowing the immune response to be refocused to broadly conserved regions. As shown in Scheme 1, the approach involves the incorporation of non-canonical amino acids such as p-azido-L-phenylalanine (F*) at one or more sites near the epitope to be shielded. The reactive azide group of F* enables orthogonal site-specific functionalization with PEG by copper-free click cycloaddition⁹. Tuning the location of incorporation of F* and the size of the PEG shields enables the nanopatterning of the antigen surface.

We first tested our approach using GFP as a model antigen. The “epitope to be shielded” was chosen to be the epitope for a GFP nanobody (NB, Fig. 1A)^{10, 11}, because a structure for the GFP-NB complex has been reported (PDB ID:3OGO)¹⁰.

Figure 1. Nanopatterning allows selective shielding of targeted GFP epitope.

(A) Left: Structure of GFP in green with the footprint of Nanobody (NB) marked in magenta; Right: the structure of GFP bound to the NB. (B) Mut1-NB*, showing residues in yellow around the NB epitope that were mutated to F*; conjugation to PEG-DBCO generates nanopatterned Mut1-NB*-PEG. (C) Nanopatterned Mut2*-PEG was designed as a control to shield a site distant from the NB epitope. (D) Characterization of wild-type (wt), mutant, and nanopatterned GFP variants by SDS-PAGE. (E) Scheme for synthesizing multivalent GFP conjugates based on SpyCatcher-functionalized branched PEG scaffolds. (F) Characterization of binding of NB (pink bars) and an orthogonal anti-GFP antibody (green bars) to wt, mutant, and nanopatterned GFP variants by ELISA.

We expressed and purified wild-type (wt) GFP with an N-terminal SpyTag^{12, 13} (see Supporting Information for details). SpyTag is a 13 amino acid peptide that spontaneously forms an isopeptide bond with its protein partner (SpyCatcher)^{12, 13}. Sites for incorporating F* to shield the NB epitope – residues 175 and 204 in GFP – were selected based on factors including solvent accessibility and proximity to the epitope (Fig. 1B). Amber (UAG) stop codons were inserted into GFP at the selected sites, and E. coli were co-transformed with the plasmid encoding the GFP mutant (Mut1-NB*) and plasmid pEVOL-pAzRS2.t187 (Addgene #73544)⁷, which encodes the tRNA synthetase/tRNA pair required for the efficient incorporation of F*. As a control, we designed a second mutant GFP protein (Mut2*, Fig. 1C), incorporating F* at residues distant from the nanobody epitope (111 and 190).

The mutant proteins were expressed and purified (see Supporting Information for details) and then allowed to react with 5 kDa PEG presenting azide-reactive dibenzocyclooctyne (DBCO) end groups (mPEG5k-DBCO, Nanocs, Boston, MA). Nanopatterned proteins (Mut1-NB*-PEG and Mut2*-PEG, Fig. 1B–C) were purified and characterized by SDS Polyacrylamide Gel Electrophoresis (SDS-PAGE, Fig. 1D).

Since the multivalent presentation of antigens is known to enhance their immunogenicity^{14, 15}, we next attached the candidate antigens to branched PEG scaffolds (Fig. 1E) and characterized the multivalent conjugates by ELISA. As seen in Fig. 1F, both anti-GFP antibodies – NB and GF28R (ThermoFisher Scientific), which binds to the N-terminus of GFP – bound to wt GFP, the unpegylated mutant proteins Mut1-NB* and Mut2*, and Mut2*-PEG. However, nanopatterning selectively suppressed the binding of NB to Mut1-NB*-PEG relative to that of GF28R (Fig. 1F).

Having demonstrated the ability to modulate the recognition of antigens by antibodies in vitro, we next tested whether immunization with nanopatterned proteins could refocus the immune response in vivo. Mice (n=5 per group; 6 groups) were immunized with multivalent conjugates or scaffold alone and bled on day 70 (Fig. 2A; see Supporting Information for details). We observed a significant anti-GFP serum antibody response – expressed as area under the curve (AUC)¹⁶ – for immunization with all five multivalent GFP conjugates (Fig. 2B).

Figure 2. Nanopatterning refocuses the anti-GFP immune response.

(A) Vaccination schedule in mice. (B) Serum antibody titers for the multivalent GFP antigens and scaffold alone are expressed as area under the curve (AUC). (C) Scheme illustrating the immunodepletion to remove antibodies (green) that bind to sites on GFP other than the NB epitope, for i. anti-Mut1-NB*-PEG sera and ii. Anti-Mut2*-PEG sera. Antibodies that bind to the NB epitope are shown in pink. (D) Characterization of the binding of immunodepleted sera to wtGFP in the presence and absence of competing NB by ELISA. (****p<0.0001, ns – not statistically significant, determined by a one-way analysis of variance (ANOVA) with Tukey’s post-hoc multiple comparison between groups).

While the total anti-GFP serum antibody response to Mut1-NB*-PEG and Mut2*-PEG was comparable (Fig. 2B), it was critical to determine whether the distribution of binding sites for the elicited anti-GFP antibodies was different, i.e., whether nanopatterning had refocused the immune response for Mut1-NB*-PEG away from the NB epitope. We therefore carried out an immunodepletion, to remove from sera the antibodies that bound to sites other than the NB epitope (Fig. 2C). As seen in Fig. 2D, depleted sera from mice immunized with Mut2*-PEG bound to wt GFP (left pink bar), but the binding was reduced to baseline levels in the presence of competing NB (right pink bar). This result indicates – as expected – that immunization with Mut2*-PEG elicits antibodies that bind to the nanobody epitope. The depleted control sera (from mice immunized with scaffold alone) did not bind significantly to GFP either in the absence or presence of NB (light gray bars). In contrast, depleted sera for mice immunized with Mut1-NB*-PEG did not bind significantly to GFP either in the absence or presence of NB (dark gray bars), confirming that nanopatterning had essentially completely refocused the antibody response away from the NB epitope (Fig. 2D).

Next, we applied the nanopatterning approach to a promising malarial vaccine candidate, Merozoite surface protein 1₁₉ (MSP1₁₉), the 19 kDa C-terminal fragment of MSP1^{17, 18}. Antibodies elicited against MSP1₁₉ can inhibit the invasion of erythrocytes by malarial parasites^17–19. A major obstacle in developing an effective vaccine targeting MSP1₁₉ is the unusual elicitation of “blocking antibodies”²⁰ that inhibit the binding and action of neutralizing “inhibitory antibodies” (Fig. 3A). The variability in the MSP1₁₉ sequence can also result in a strain-specific immune response. We therefore tested the ability to use nanopatterning to shield the epitopes for blocking antibodies and focus the immune response to conserved regions of MSP1₁₉ that are epitopes for inhibitory antibodies²¹.

Figure 3. Nanopatterning PfMSP1₁₉.

(A) Structure of PfMSP1₁₉ (PDB ID: 1CEJ) with residues involved in binding to the inhibitory antibodies shown in green and residues involved in binding of blocking antibodies shown in red. (B) Characterization of wt PfMSP1₁₉ and nanopatterned mutants by SDS-PAGE. (C) Nanopatterned mutants are successful at shielding the inhibitory and blocking antibody epitopes of PfMSP1₁₉. ELISA binding expressed as absorbance (mean ± SD, n=3).

We first expressed and purified wild-type P. falciparum MSP1₁₉ (PfMSP1₁₉) as well as three mutant proteins, MutI*, MutB1*, and MutB2* (Fig. 3B); the mutant proteins each incorporated two F* residues and were designed to block the binding of the inhibitory antibody 12.8, the blocking antibody 7.5, and the blocking antibody 2.2, respectively (see Supporting Information for details)^{19, 21}. The mutant proteins were allowed to react with mPEG5k-DBCO and the PEGylated proteins were purified by SEC (Fig. 3C). Characterization by ELISA confirmed that nanopatterning PfMSP1₁₉ blocked the binding of the targeted anti-PfMSP1₁₉ antibody (Fig. 3B).

Next, we wanted to demonstrate that nanopatterning could refocus the immune response to conserved inhibitory antibody epitopes in vivo, enabling the broad (as opposed to strain-specific) recognition of MSP1₁₉. We used P. yoelii MSP1₁₉ (PyMSP1₁₉) for these experiments; PyMSP1₁₉ show significantly greater variability in sequence between parasite strains (Fig. 4A) than PfMSP1₁₉ and thus represented a more challenging test for our approach.

Figure 4. Refocusing the immune response to conserved inhibitory epitopes of PyMSP1₁₉.

(A) Structure of PyMSP1₁₉ (PDB ID: 2MGP) with residues homologous to the PfMSP1₁₉ inhibitory and blocking antibody epitopes marked in green and red respectively. Right Panel: Residues that differ between two strains of PyMSP1₁₉ (17XNL and heterologous strain N67) are marked in cyan. (B) Nanopatterned mutants, MutI*-PEG (shielding inhibitory epitope) and MutB*-PEG (shielding blocking epitope). (C) Vaccination schedule in mice. (D) Anti-MSP1₁₉ antibody response for the two immunized mutants is comparable against homologous PyMSP1₁₉ but significantly different (*p<0.05, unpaired two-tailed t-test) against heterologous PyMSP1₁₉.

While PfMSP1₁₉ and PyMSP1₁₉ are divergent in sequence (Supporting Fig. S1), the epitope for the PfMSP1₁₉ inhibitory antibody is fairly conserved and the proteins are fairly similar in structure. Since epitopes for anti-PyMSP1₁₉ inhibitory and blocking antibodies have not been mapped precisely, to guide our nanopatterning efforts, we first mapped the epitopes for the anti-PfMSP1₁₉ antibodies onto the PyMSP1₁₉ structure (Supporting Fig. S1, Fig. 4A). We expressed and purified PyMSP1₁₉ from strain 17XNL and also generated two nanopatterned mutants: the double mutant MutI*-PEG served as a control, where we shielded the conserved inhibitory antibody epitope and the quadruple mutant MutB*-PEG was designed to shield both “blocking epitopes” to refocus the immune response to the conserved inhibitory epitope (Fig. 4B). We immunized mice with these antigens (Fig. 4C) and analyzed the anti-PyMSP1₁₉ serum antibody titers by ELISA against both homologous and heterologous (Fig. 4A) PyMSP1₁₉, which differed in sequence at as many as 21 of 96 residues. The total serum antibody response against homologous PyMSP1₁₉ was comparable for mice immunized with MutI*-PEG and MutB*-PEG (Fig. 4D). Importantly, antibodies elicited by immunization with the nanopatterned MutB*-PEG recognized heterologous PyMSP1₁₉, whereas those elicited using the control MutI*-PEG failed to do so (Fig 4D). This result confirms the ability of nanopatterning to elicit strain-transcending antibodies.

Ethical statement

All the protocols that included experimental animal procedures were carried out in accordance with the US Animal Welfare Act and approved by ProSci Inc.’s Institutional Animal Care and Use Committee.

Conclusions

In conclusion, we have demonstrated the use of nanopatterning to modulate the immunogenicity of protein antigens on the nanoscale. While the current work involved the incorporation of a single non-canonical amino acid into target proteins, the ability to incorporate multiple non-canonical amino acids with orthogonal reactivities into protein antigens should provide even finer control over epitope exposure⁷. Our approach can be extended to antigens expressed in other hosts, including mammalian cells²² and to protein-resistant shields that are alternatives to PEG^23–25. This strategy should be applicable for the elicitation of broadly protective antibodies targeting pathogens ranging from malarial parasites to influenza viruses, flaviviruses, and HIV.