Abstract

Over 75% of malaria-attributable deaths occur in children under the age of 5 years. However, the first malaria vaccine recommended by the World Health Organization (WHO) for pediatric use, RTS,S/AS01 (Mosquirix), has modest efficacy. Complementary strategies, including monoclonal antibodies, will be important in efforts to eradicate malaria. Here we characterize the circulating B cell repertoires of 45 RTS,S/AS01 vaccinees and discover monoclonal antibodies for development as potential therapeutics. We generated >28,000 antibody sequences and tested 481 antibodies for binding activity and 125 antibodies for antimalaria activity in vivo. Through these analyses we identified correlations suggesting that sequences in Plasmodium falciparum circumsporozoite protein, the target antigen in RTS,S/AS01, may induce immunodominant antibody responses that limit more protective, but subdominant, responses. Using binding studies, mouse malaria models, biomanufacturing assessments and protein stability assays, we selected AB-000224 and AB-007088 for advancement as a clinical lead and backup. We engineered the variable domains (Fv) of both antibodies to enable low-cost manufacturing at scale for distribution to pediatric populations, in alignment with WHO’s preferred product guidelines. The engineered clone with the optimal manufacturing and drug property profile, MAM01, was advanced into clinical development.

Discussion

Using single-cell sequencing of B cells from RTS,S vaccinees, we generated a library of CSP-specific mAbs that could be assessed and downselected to those most amenable for engineering and development as antimalaria medicines. In doing so we also uncovered important characteristics of the humoral response to RTS,S that may underlie the efficacy and durability of this vaccine.

First, we unexpectedly discovered an inverse relationship between the percentage of CSP-specific, IgG-expressing PBs and vaccinee protection status P3D (Fig. 1e–f). These data suggest that, despite the well-reported association between anti-CSP antibodies and protection following RTS,S^{56,57,58,59,60}, more cells expressing anti-CSP or anti-NANP antibodies may not drive stronger protection. Because antibodies targeting repeat regions demonstrate a range of sporozoite-inhibitory activity in vivo, the difference between protective and unprotective antibody repertoires in humans could be driven by the relative proportion of highly versus weakly effective repeat-binding antibodies. This point is exemplified by the highly efficacious mAb, AB-000317, which was expressed in the most dominant P3D lineage of a protected vaccinee and in a much less frequent lineage of an unprotected vaccinee (Fig. 1f). Competition between antibodies at the sporozoite surface (‘epitope masking’)^45,47,48 and/or within lymphoid organs³², where dominant but weakly functional antibodies outcompete subdominant, highly effective antibodies for binding to the repeat regions, could be one explanation for this inverse relationship between protection and prevalence of repeat-binding lineages (Fig. 1e). Indeed, our data indicating that P3D PB repeat-binding mAbs have lower levels of SHM than other mAbs (Fig. 1d) support the hypothesis that immature clones are preferentially activated and expanded over more protective memory clones^32,48,49. Furthermore, this hypothesis could underlie other observations about RTS,S responses: specifically, functional antibodies in sera were higher post second dose (P2D) versus P3D⁵⁶ and anti-CSP P2D, but not P3D, PB and memory B cells associate with P3D protection status⁶¹ and some vaccinees lost previous protective immune signatures P3D⁶¹. Overall, we propose that expression of potent, inhibitory antibodies by P3D PBs alone is insufficient for protection. Rather, relative levels of effective versus ineffective repeat-binding antibodies may be important in provision of consistent protection. Together, these data highlight the need for studies to correlate protection with the ratio of effective to ineffective repeat-binding antibodies circulating in sera at the time of infection.

Second, we found that sporozoite-inhibitory activity in mice does not correlate with binding kinetics to the long NANP6 peptide (Extended Data Table 1) but does significantly correlate with k_off to CSP and with binding kinetics to both the short NANP-containing peptide (NPNA3) and peptides from the minor-repeat region and JR (Fig. 3d–g and Extended Data Table 1). These data suggest that protective antibodies induced following RTS,S vaccination probably affinity mature to short NANP repeats. Because the short PNANPN sequence is contained within both major repeats (included in RTS,S) and the minor repeat and JRs (not included in RTS,S), affinity maturation against this sequence may allow these antibodies to gain and/or improve promiscuous binding activity to epitopes containing PNANPN or a portion thereof, that may be important for potency^28,39,44 but which are not fully represented in RTS,S. Consistent with this interpretation, aggregate levels of SHM in both heavy and light chains of inhibitory mAbs are correlated with binding kinetics to NANP-, NVDP- and NPDP-containing short peptides (Fig. 2c–e and Extended Data Table 1), as well as with inhibitory activity in the sporozoite-challenge model (Fig. 3h–i and Extended Data Table 1). Overall, the data are consistent with suggestions that next-generation anti-CSP vaccines contain fewer NANP repeats^37,62 and/or should include sequences from minor repeats and JRs^{16,28,34,46,63,64,65,66}.

Last, multiple observations including (1) the inverse relationship between the percentage of CSP- and NANP6-binding antibodies from expanded linages and protection against CHMI (Fig. 1e); (2) the correlation between binding kinetics with NVDP- and NPDP-containing peptides absent in RTS,S (Fig. 3f–g and Extended Data Table 1) and sporozoite inhibition in vivo; and (3) the correlation between binding kinetics to the short NPNA3, but not the longer NANP6 peptide, and sporozoite inhibition in vivo (Fig. 3e and Extended Data Table 1), are consistent with a hypothesis where multiple NANP repeats act as an immune ‘decoy’^37,67 or ‘smokescreen’ ^66,68 that dilutes protective immunity^{69,70,71,72,73}. Under this hypothesis, antibody lineages that bind only to homologous epitopes are preferentially expanded over promiscuous mAbs that bind well to both homologous and heterologous epitopes. Because many of the antibodies that bind only to NANP repeats offer limited protection in vivo (Extended Data Table 1), enrichment for these antibodies dilutes the protective capacity of the broader anti-CSP repertoire^32,37,48. In contrast, promiscuously binding antibodies that can simultaneously bind to multiple^39,42,50 NANP repeats and heterologous epitopes may drive superior protection in vivo because they can saturate binding sites and further stabilize^{39,42,50,74,75} mAb interactions with both homologous and heterologous epitopes. Consistent with this idea, three of the most active in vivo mAbs, CIS43, L9 and AB-000317, can bind CSP with high stoichimetries^16,20,39 via two binding events¹⁶. Further studies that examine the mechanisms of mAb binding to junctional, minor- and major-repeat regions and the interaction with sporozoite inhibition are needed to test this hypothesis.

Our study has a number of limitations. First, we did not assess whether the antibodies we characterized from expanded PB lineages collected 1 week before infection accurately reflected the composition of sera antibodies at the time of infection. Second, the current, WHO-recommended RTS,S vaccine includes the adjuvant AS01E⁷⁶. Our study, however, examined antibodies derived from vaccinees who received RTS,S AS01B²⁴. While the two adjuvants contain the same components, they are included at different levels⁷⁷. More work will be needed to assess whether RTS,S/AS01E as used in the field produces antibody repertoires like those characterized here. Third, the correlations we observed between binding off-rates and function (Fig. 3d–i and Extended Data Table 1) were limited to mAbs from protected vaccinees. Further studies will be needed to assess whether similar correlations exist for inhibitory, repeat-binding antibodies derived from unprotected vaccinees. Finally, we tested only a small fraction of the sequences we generated. While we focused our discovery campaign on the larger PB lineages from each vaccinee, 3,334 smaller but expanded lineages remain uncharacterized.

By sequencing PBs, which represent the breadth of Ig sequence diversity that originates from lymphoid reactions following RTS,S vaccination, we deconstructed a part of the humoral response from protected and unprotected vaccinees. We identified lineages with highly protective antibodies in mouse models, screened sequence-diverse clones within those lineages for development-related properties and further engineered a clone to optimize its developability characteristics. These properties will increase the likelihood that regimens can be successfully developed for pediatric populations, which require small-volume, concentrated doses at low viscosity⁹. Given that the in vivo efficacy displayed by MAM01 is comparable to AB-000317, which in turn has activity comparable to or better than that of CIS43 (refs. ^16,34), we believe that MAM01 will be useful for individuals living in malaria-naïve and -endemic regions and may also meet the WHO’s preferred product profile⁹, including cost-effective dosing for delivery in LMICs. By focusing on properties critical for manufacture and distribution to global pediatric populations^9,23, in addition to the requirement for functional potency, the work reported here may contribute to prophylactic strategies that aid efforts in the eradication of malaria.