Construction and assembly of MS2 VLPs displaying L9 peptides

Previously, we engineered the coat protein of bacteriophage MS2 so that it could tolerate the insertion of heterologous peptide sequences in an exposed loop on its surface. This display site, which is referred to as the AB loop, consists of a beta-hairpin structure that protrudes from the surface of the virus particle (Fig. 1a). Given that the core NPNV L9 epitope adopts a type 1 beta-turn³¹, we asked whether displaying this epitope in a more relevant structural context could elicit more potent antibody responses. Nucleotide sequences encoding peptides representing a minimal 8-amino acid L9 epitope centered around the core NPNV sequence (CSP amino acids 105–112), the consensus 15-amino acid epitope (amino acids 105–119, which contains two NPNV sequences), and an extended 27 amino acid epitope (amino acids 105–131) (Fig. 1b) were inserted into the downstream AB-loop of the MS2 coat protein single-chain dimer expression vector. Expression plasmids were then used to express recombinant MS2 L9 VLPs in E. coli. Because each VLP consists of 90 copies of the single-chain dimer, each recombinant MS2 VLP displays exactly 90 copies of the L9 peptide per VLP. Successful insertion of the L9 peptides was confirmed by SDS-PAGE analysis. Coat proteins from recombinant MS2 L9 VLPs displayed a higher molecular weight compared to wild-type MS2 coat protein (Fig. 1c). To confirm that the insertion of different L9 peptides into the MS2 coat protein was compatible with VLP assembly, transmission electron microscopy (TEM) was used to visualize the structures of the three different MS2 L9 VLPs (Fig. 1d). These results showed that all three of the MS2 L9 constructs assembled into particles with diameters of ~30 nm. The morphology of MS2 VLPs displaying the longer 15- and 27-amino acid L9 sequences was somewhat less regular than VLPs displaying the shorter 8 amino acid peptide. Dynamic Light Scattering analysis of the recombinant MS2 L9 VLPs revealed that longer L9 insertions were associated with modest increases in the mean VLP diameter (Supplementary Fig. 1). Of note, the 27aa L9 sequence is the largest peptide that we have successfully displayed on MS2 VLPs.

a Structure of the MS2 coat protein dimer (left) and the MS2 VLP (right). 90 coat protein dimers self-assemble into a VLP. The location of the AB-loop is highlighted in red on both structures. Images of the MS2 coat protein dimer and the MS2 VLP were generated by the authors using online tools available through the RCSB protein data bank. b PfCSP domains and the sequence of the junctional and minor repeat regions. The sequence of the three inserted L9 sequences and the location of the core NPNV epitopes (underlined) are shown. c SDS-PAGE analysis of wildtype MS2 VLPs and MS2 L9 VLPs displaying the 8, 15, and 27 amino acid L9 epitopes. The unmodified gel is shown in Supplementary Fig. 2. d Transmission electron microscopy (TEM) images of recombinant MS2 L9 VLPs. The scale bar (in white) represents 100 nm. e Binding of L9 mAb to MS2 L9 VLPs or wildtype MS2 VLPs, as measured by ELISA.

The antigenicity of MS2 L9 VLPs was assessed by measuring the binding of L9 mAb to the MS2 L9 VLPs by enzyme-linked immunosorbent assay (ELISA). All three MS2 L9 VLPs were strongly recognized by the L9 mAb (Fig. 1e), whereas wild-type MS2 VLPs were not. These findings indicate that displaying L9 epitopes in a constrained loop structure on MS2 VLPs does not affect L9 mAb binding.

MS2 L9 VLPs elicit strong anti-CSP antibody responses

Next, we compared the immunogenicity of the three MS2 L9 VLPs constructs in mice. Mice were initially immunized twice, at weeks 0 and 3, and sera were collected one week after the second immunization to measure antibody levels against full-length recombinant CSP by ELISA. All three MS2 L9 VLPs elicited strong anti-CSP IgG responses (Fig. 2a). MS2 L9(15aa) VLPs elicited the highest titer anti-CSP antibody responses. Based on their improved immunogenicity compared to the VLPs displaying the L9(8aa) epitope, MS2 L9 VLPs displaying the longer (15 and 27aa) epitopes were selected for further investigation. In each of these groups, mice received a third immunization to determine whether an additional boost would further enhance the immune response. This third immunization resulted in a substantial increase in anti-CSP antibody levels in mice that received the MS2 L9(27aa) VLPs, and a modest increase in anti-CSP antibody levels in mice immunized with MS2 L9(15aa) VLPs (Fig. 2b). After three immunizations, the antibody responses induced by the MS2 L9(27aa) and MS2 L9(15aa) VLPs were nearly identical (Fig. 2b).

a Comparison of the immunogenicity of MS2 L9 VLPs displaying 8, 15, and 27 amino acid L9 epitopes. Groups of Balb/c mice (n = 5) were immunized intramuscularly with 5 µg of VLPs at weeks 0 and 3. Anti-CSP IgG end-point dilution titers were calculated by ELISA using sera collected one week after the second immunization. Results show antibody titers in individual mice, lines represent the geometric mean titer from each group. Statistical comparisons were performed using a one-way ANOVA with post-hoc analysis. b Mice immunized with MS2 L9(15aa) and MS2 L9(27aa) received a third immunization, given 7 weeks after the prime. Sera were obtained one week following the third immunization. Statistical comparisons between titers following the second dose and the third dose were performed using an unpaired t-test. c Anti-CSP antibody titers in mice immunized twice with unadjuvanted MS2 L9(15aa) VLPs compared to mice that received two doses MS2 L9(15aa) plus Cquim-MA adjuvant and mice that received a Qβ L9 prime (plus Cquim-MA) followed by an MS2 L9(15aa) boost (also with Cquim-MA). Sera were obtained one week following the second immunization. Statistical comparisons were performed using a one-way ANOVA with post-hoc analysis (Tukey’s test).

We have previously shown that the immunogenicity and protective efficacy of Qβ VLP-based vaccines targeting CSP epitopes can be enhanced by co-administration of specific adjuvants. To evaluate if adjuvants could similarly enhance antibody responses to MS2-based vaccines, we measured the anti-CSP antibody responses generated by MS2 L9(15aa) VLPs formulated with Cquim-MA, a dual TLR7/8 agonist adjuvant, that we previously have shown can enhance antibody responses to Qβ VLP-based vaccines³². Two immunizations with MS2 L9(15aa) VLPs plus Cquim-MA induced markedly stronger anti-CSP antibody responses compared with unadjuvanted MS2 L9(15aa) VLPs (Fig. 2c).

Heterologous prime-boost regimens can be an effective method for focusing antibody responses against specific epitope targets and minimizing antibody responses against the vaccine platform³³. To determine if this approach could also be used to boost anti-CSP antibody responses, mice were primed with Qβ L9 VLPs mixed with Cquim-MA and then boosted with MS2 L9(15aa) VLPs, also formulated with Cquim-MA. This heterologous prime-boost approach resulted in similar anti-CSP antibody responses compared to those induced by MS2 L9(15aa) VLPs with Cquim-MA (Fig. 2c). Taken together, these data demonstrate that MS2 L9 VLPs can elicit high-titer anti-CSP antibody responses with or without heterologous boosting, especially in combination with Cquim-MA.

MS2 L9 VLPs elicit durable antibody responses

One advantage of VLP-based vaccines is that they reliably elicit durable antibody responses³⁴, likely through the efficient induction of long-lived plasma cells³⁵. To assess the duration of the anti-CSP antibody response elicited by MS2 L9 VLPs, mice received three immunizations of MS2 L9(15aa) or MS2 L9(27aa) VLPs and antibody responses were measured for nearly a year following the prime. Both vaccines elicited long-lived antibodies, with little to no reduction in titers over 40–50 weeks following the initial prime (Fig. 3a), mirroring what we previously observed following vaccination with Qβ L9 VLPs³¹. Similarly, we compared the durability of antibody responses in mice that received two doses of Cquim-MA adjuvanted vaccines (Fig. 3b). Mice received either a homologous prime-boost with MS2 L9(15aa) VLPs or a heterologous prime-boost with Qβ L9 followed by MS2 L9(15aa) VLPs. Both vaccine regimens elicited similar peak antibody titers following the second immunization. Antibody levels slightly declined from the peak titer to the final timepoint, 38 weeks following the prime. Use of Cquim-MA with MS2 L9(15aa) VLPs slightly increased antibody titers compared to unadjuvanted vaccine at the late (38 week) timepoint, but this difference was not statistically significant.

a Groups of Balb/c mice (n = 5) were immunized intramuscularly with 5 µg of VLPs at weeks 0, 3, and 7 and were measured following each immunization and at 38–47 weeks post-prime. Anti-CSP IgG titers were calculated by end-point dilution ELISA. Geometric means plus SEM are shown at each timepoint. b Mice that received vaccine plus Cquim-MA adjuvant received two immunizations, at weeks 0 and 3.

Immunization with MS2 L9(15aa) VLPs plus Cquim-MA adjuvant protects mice from malaria challenge

Next, we evaluated whether MS2 L9 VLPs could protect mice from malaria challenge. Vaccine efficacy was assessed in a model in which vaccinated mice were challenged with mosquitoes carrying transgenic Plasmodium berghei (Pb) sporozoites expressing full-length PfCSP and luciferase (Pb–PfCSP-Luc)³⁶. In this model, parasite liver loads can be quantified by measuring luciferase signal in the liver. In addition, the ability of vaccination to mediate sterilizing immunity can be determined by monitoring the development of blood-stage infection (parasitemia). In our initial challenge experiment (shown schematically in Fig. 4a), we evaluated whether immunization with MS2 L9(15aa) VLPs could protect mice from malaria infection. Mice received three immunizations with MS2 L9(15aa) VLPs, with or without Cquim-MA. To benchmark vaccine efficacy, groups of mice were also immunized three times with Qβ L9 VLPs plus Cquim-MA or with the WHO-approved vaccine RTS,S/AS01_E (using a 5 µg dose). As negative controls, groups of mice were immunized with wild-type MS2 VLPs (not displaying the L9 epitope), wild-type Qβ VLPs, or Cquim-MA adjuvant alone, or were unimmunized (naïve). Following immunization, anti-CSP antibody concentrations from individual mice were quantified. As shown in Fig. 4b, RTS,S/AS01_E, Qβ L9/Cquim-MA, MS2 L9(15aa), and MS2 L9(15aa)/Cquim-MA induced strong anti-CSP antibody responses. MS2 L9(15aa)/Cquim-MA elicited the highest anti-CSP antibody concentrations, with mean levels ~6-fold higher than unadjuvanted MS2 L9(15aa), ~3-fold higher than Qβ L9/Cquim-MA, and ~2.2-fold higher than RTS,S/AS01_E. None of the control groups elicited anti-CSP antibodies (not shown).

a Experimental timeline. C57BL/6 mice were immunized three times at weeks 0, 3, and 6, followed by a challenge with five Pb-PfCSP-Luc-infected mosquitoes at week 8. Liver luminescence was measured 42 h after the mosquito challenge, and blood smears were taken starting on day 3 post-infection. b Anti-CSP antibody levels in serum obtained following the third vaccination, prior to challenge. Horizontal lines represent the mean anti-CSP antibody concentration for each group. Groups are compared by 2-tailed unpaired t test. c Parasite liver burden measured via luminescence. Horizontal lines indicate geometric mean luminescence for each group. Statistical comparisons were performed using a one-way ANOVA in which experimental groups were compared to every other group. Multiple comparisons were corrected using the Bonferroni adjustment. d Protection from blood-stage infection, as measured by percent of blood parasite-free mice post-challenge. A log-rank test (correcting for multiple comparisons by controlling the False Discovery Rate using the two-stage step-up method of Benjamini, Krieger and Ykutieli) was used to statistically compare vaccinated groups to naïve, infected mice.

Mice were then infected by exposure to five Pb–PfCSP-Luc infected mosquitoes. As an initial determination of vaccine efficacy, liver-stage parasite burden was quantified by measuring liver luciferase activity using intravital imaging 42 h post-infection (Fig. 4c). As has been previously shown in this model³⁷, immunization with RTS,S/AS01_E dramatically reduced parasite liver burden, by ~97% relative to naïve controls. Immunization with Qβ L9/Cquim-MA also reduced parasite liver burden by ~97%. Groups immunized with MS2 L9(15aa) and MS2 L9(15aa)/Cquim-MA also displayed statistically significant reductions in parasite liver loads (by 88% and 94%, respectively). None of the negative control groups were protected from infection.

To evaluate the possibility of sterilizing protection, daily blood smears were taken from the challenged mice starting at day 3 post-infection. Blood was assessed for parasitemia using Giemsa staining (Fig. 4d). While all control groups developed blood-stage parasitemia by day 4, 40% of the RTS,S/AS01_E-vaccinated group (two out of five mice) were protected, which is similar to what has been previously reported in this model³⁷ and mirrors the protection conferred by RTS,S/AS01_E in humans. 80% of the mice immunized with Qβ L9/Cquim-MA were completely protected from infection, which is an improvement compared to previous studies using Qβ L9 adjuvanted with Advax-3³⁰, suggesting that Cquim-MA may be a more effective adjuvant in combination with this vaccine. In this challenge study, immunization with Qβ L9/Cquim-MA provided stronger protection compared to RTS,S/AS01_E (80% compared to 40%), but this difference was not statistically significant. Although MS2 L9(15aa)/Cquim-MA elicited the highest anti-CSP antibody concentrations, only 20% of mice immunized with this vaccine remained free from blood-stage parasitemia.

A combination L9 VLP vaccine reduces liver parasite burden and enhances sterilizing immunity in Plasmodium-challenged mice

We performed a second, independent challenge experiment with the following goals: (1) to replicate our initial results showing that vaccination with Qβ L9/Cquim-MA provides stronger protection than RTS,S/AS01_E, (2) to evaluate the protective efficacy of MS2 L9 VLPs displaying the longer 27aa L9 peptide, and (3) to assess protection of a combination of the MS2 L9(15aa) and Qβ L9 vaccines. This experiment followed the same design as our initial study; mice were immunized with RTS,S/AS01_E, Qβ L9/Cquim-MA, MS2 L9(27aa)/Cquim-MA, or a mixture of Qβ L9 and MS2 L9(15aa)/Cquim-MA. Negative controls included a group of mice that received wildtype Qβ VLPs and a group of unimmunized (naïve) mice. As we showed previously, Qβ and MS2-based L9 vaccines elicited strong anti-CSP antibody responses, with mean levels >100 µg/mL, similar to the antibody levels induced by RTS,S/AS01_E (Fig. 5a). Immunization with Qβ L9/Cquim-MA or MS2 L9(27aa)/Cquim-MA resulted in similar reductions in parasite liver loads as RTS,S/AS01_E ( ~ 91%), but mice that received the combination vaccine plus Cquim-MA showed the greatest reduction in parasite burden (~96%), although this difference was not statistically distinct from the group immunized with RTS,S/AS01_E (Fig. 5b). Similar to our initial challenge experiment (Fig. 4), vaccination with Qβ L9/Cquim-MA resulted in sterilizing immunity in a higher percentage of mice than we observed in the group immunized with RTS,S/AS01_E (Fig. 5c), although this difference was not statistically significant. The percentage of mice that were protected from infection was lower in this experiment (43%) compared to the previous challenge study (80%). This may be explained by the higher mean parasite liver burden across all groups in the second challenge experiment (for example, it was 20% higher in the naïve group) or it could reflect the variability of parasite burden in this challenge model. Interestingly, immunization with the combination vaccine (Qβ and MS2 L9 VLPs with Cquim-MA) provided the strongest protection, with a higher percentage of mice exhibiting sterilizing immunity than in mice immunized with either RTS,S/AS01_E or Qβ L9/Cquim-MA. Taken together, these results indicate that a combination vaccine presenting the L9 epitope in multiple conformations, as an unstructured peptide on Qβ VLPs and in a structured β-hairpin on MS2 VLPs, was effective at eliciting sterilizing immunity against Plasmodium infection.

a Anti-CSP antibody levels in serum obtained following the third vaccination. Horizontal lines represent the mean anti-CSP antibody concentration for each group. No significant differences were observed between groups (p > 0.05, unpaired t-test). b Parasite liver burden measured via luminescence. Horizontal lines indicate geometric mean luminescence for each group. Statistical comparisons were performed using a one-way ANOVA in which experimental groups were compared to every other group. Multiple comparisons were corrected using the Bonferroni adjustment. c Protection from blood-stage infection, as measured by percent of blood parasite-free mice post-challenge. A log-rank test (correcting for multiple comparisons by controlling the False Discovery Rate using the two-stage step-up method of Benjamini, Krieger and Yekutieli) was used to statistically compare vaccinated groups to naïve, infected mice.