Core-shell microcapsules compatible with routine injection enable prime/boost immunization against malaria with a single shot

Abstract

Inadequate booster uptake threatens the success of immunization campaigns as seen with the recently rolled-out R21 malaria vaccine. The ability to administer both prime and boost immunizations with a single injection would therefore save lives and alleviate health care burdens. We present a platform for delayed delivery of the booster dose that is scalable with existing technology, easily injectable, and protective against malaria in vivo. Using chip-based microfluidics, we encapsulated the R21 malaria vaccine in polymer microcapsules that release their content weeks to months postinjection. Coinjecting microcapsules with the priming dose of the R21 vaccine elicited strong antibody responses in a mouse model and provided 85% of the protection of a standard prime/boost schedule. If confirmed in humans, these results would pave the way for rapid deployment of single-shot prime/boost vaccination, an urgently needed global health intervention.

Fig. 1. Stepped microfluidic chip design generates highly uniform PLGA microcapsules with high encapsulation efficiency.

(A) Summary of the microfluidic W/O/W emulsification process for production of microcapsules. (B) High-speed camera image of the emulsification process. (C) Evolution of the core and outer diameters of the double emulsion during continuous production (mean ± standard deviation, n > 90 per data point). (D) Core and outer (whole microcapsule) diameters after solvent extraction. CV, coefficient of variation. (E) Fluorescence microscopy images of microcapsules produced using the stepped microfluidic design, with dextran-TRITC as a model payload and 7-17 kDa 50:50 lactide:glycolide (L:G) PLGA as the shell polymer; TRITC signal and brightfield are overlayed. Inset: magnified microcapsules. (F) High-speed camera images of the emulsification process at two inner fluid pressures (left) and fluorescence microscopy images of the corresponding dextran-TRITC loaded microcapsules (right); TRITC signal and brightfield are overlayed. Full range of inner fluid pressures tested is displayed in fig. S2. All scale bars are 100 µm.

Fig. 2. Core-shell structure of microcapsules enables delayed burst release in vitro with shell thickness-dependent release kinetics and core pH.

(A) In vitro release kinetics at 37 °C from microcapsules manufactured by either the stepped microfluidic process (microcapsules) or batch emulsification (BE) using dextran-TRITC as a model payload and 7-17 kDa 50:50 L:G PLGA as the shell polymer. Mean of batch replicates (points joined by thick lines), and individual batch replicates (thin lines) are shown (n=4). (B) Evolution of microcapsule fluorescence during in vitro incubation (median and violin plot distribution, n shown above for each timepoint). AU, arbitrary units. (C) Cross-sectional fluorescence profile at days 0 and 20 of in vitro incubation. Mean (lines) and standard deviation (ribbons) of the profiles are shown (Day 0: n=2233, Day 20: n=454). (D) Fluorescence microscopy and SEM images of microcapsules containing dextran-TRITC as the model payload and PLGA 7-17 kDa 50:50 as the shell polymer at different timepoints of in vitro incubation. TRITC and brightfield are overlayed. Yellow arrows indicate holes on the shell surface. Scale bars are 100 µm (top row), 20 µm (second and third row) and 2 µm (bottom row). (E) Effect of varying the initial shell thickness on the in vitro release kinetics of dextran-TRITC/FITC at 37 °C. Mean of batch replicates (points joined by thick lines), and individual batch replicates (thin lines) are shown (n=4). (F) Evolution of core pH (mean - standard deviation, n>300 for each point) during in vitro incubation for different initial shell thicknesses.

Fig. 3. Polymer formulation controls the release kinetics in vitro.

(A) Effect of varying MW and L:G ratio of PLGA on in vitro release kinetics using 50 mg/mL 40 kDa dextran-TRITC as the payload. (B and C) Release kinetics were compared with a reference formulation (7-17 kDa, 50:50 L:G, red line) after changing payload concentration (B) or payload MW (C). The release data for the reference formulation are taken from the same experiment. Mean of batch replicates (points joined by thick lines), and individual batch replicates (thin lines) are shown (n=3 or 4).

Fig. 4. Polymer formulation determines the delayed antibody response to the encapsulated R21/LMQ malaria vaccine.

(A) Summary of the in vivo experimental protocol. Short, medium and long delay formulations (respectively µCaps-S, µCaps-M, and µCaps-L) correspond to (7-17 kDa, 50:50 L:G), (54-69 kDa, 50:50 L:G), and (24-38 kDa, 75:25 L:G) PLGA, respectively, as presented in Fig. 3A. (B) Specific antibody responses for R21/LMQ vaccine encapsulated within µCaps-S, µCaps-M, or µCaps-L, compared with non-encapsulated Prime R21/LMQ vaccine control (n=8 per group). Geometric mean for each regimen (thick lines), individual responses (thin lines), and sampling timepoints (ticks) are shown. Responses are expressed as Cterm-specific antibody titers normalized to the maximum observed titer in individual mice. Absolute titers are shown in fig. S12. (C) Summary of the experimental protocol. * indicates a fractional (1/7th) encapsulated LMQ adjuvant dose due to the fixed concentration of the supplied adjuvant. (D) Cterm-specific antibody responses for different regimens of adjuvanted R21 vaccination (n=8 per group). Regimen-level responses from GAM fits (thick lines), 95% simultaneous CI (ribbons), and individual mouse responses (thin lines) are shown. Vertical dotted lines indicate the time of the booster administration in the Prime/Boost regimen, and the lag time for µCaps-L, estimated from fig. S12. The horizontal dotted line shows the IgG mean titer at week 5 (“post-boost” titer) of the Prime/Boost regimen.

Fig. 5. A single injection of Prime+µCaps-M induces similar immunogenicity to the standard Prime/Boost regimen.

(A) Summary of the experimental protocol. (B and C) Cterm-specific (B) and NANP-specific (C) antibody titers for different regimens of dose-matched adjuvanted R21 vaccination (Week 1-6: DD Prime n=11, Prime+µCaps-M n=28, Prime/Boost n=15; Week 7-11: Prime+µCaps-M n=12, Prime/Boost n=7). The study was stopped for each mouse at the corresponding time of challenge (week 6 or week 11, Fig. 6). Regimen-level responses from GAM fits (thick lines), 95% simultaneous CI (ribbons), individual responses (thin lines), and antibody titers at time of challenge (circles) are shown. (D and E) Cterm-specific (D) and NANP-specific (E) titer fold changes of Prime+µCaps-M compared with geometric mean titers of DD Prime or Prime/Boost at Weeks 6 and 11 are plotted with the individual mouse data (open circles) and their geometric means (closed circles). Bootstrap 95% CIs are displayed as colored bands on the fold change axis.

Fig. 6. The microcapsule-delivered booster dose elicits protective efficacy against malaria in mice.

(A) Summary of the experimental protocol. Spz, sporozoites; i.v., intravenous. (B) Predicted sterile efficacy against malaria over time post priming injection, calculated from cumulative data using a log-binomial regression model with adjuvant as a covariate. Predicted efficacy for each regimen, with analysis conditional on SMNP adjuvant (thick lines), and the corresponding 83.4% confidence bands (ribbons) are shown. Sterile protection rates observed after microcapsule storage at 4 °C and -20 °C are shown as dots, for reference. Corresponding raw data are displayed in fig. S15. (C) Relative contribution of different parameters to the protective efficacy at weeks 5 and 8 from the priming injection (log-binomial regression). Estimated ratios (dots) relative to the SMNP reference adjuvant are shown, along with error bars indicating the profile likelihood 95% CIs, and p-values (p) computed from the likelihood ratio test.