The light chain of the L9 antibody is critical for binding circumsporozoite protein minor repeats and preventing malaria

L9 is a potent human monoclonal antibody (mAb) that preferentially binds two adjacent NVDP minor repeats and cross-reacts with NANP major repeats of the Plasmodium falciparum circumsporozoite protein (PfCSP) on malaria-infective sporozoites. Understanding this mAb’s ontogeny and mechanisms of binding PfCSP to neutralize sporozoites will facilitate vaccine development. Here, we isolated mAbs clonally related to L9 and showed that this B-cell lineage has baseline NVDP affinity and evolves to acquire NANP reactivity. Pairing the L9 kappa light chain (L9κ) with clonally-related heavy chains resulted in chimeric mAbs that cross-linked two NVDP, cross-reacted with NANP, and more potently neutralized sporozoites compared to their original light chain. Structural analyses revealed that chimeric mAbs bound the minor repeat motif in a type-1 β-turn seen in other repeat-specific antibodies. These data highlight the importance of L9κ in binding NVDP on PfCSP to neutralize SPZ and suggest that PfCSP-based immunogens might be improved by presenting ≥2 NVDP.


INTRODUCTION
Interventions like vaccines are needed to prevent malaria, a parasitic mosquito-borne disease that killed ~409,000 people in 2019, mostly from Plasmodium falciparum (Pf) (WHO, 2020). The most advanced malaria vaccine is RTS,S, a protein subunit vaccine that prevents malaria by inducing antibodies against the Pf circumsporozoite protein (PfCSP), the dominant surface protein on infectious sporozoites (SPZ) spread by mosquitoes (Cockburn and Seder, 2018). RTS,S adjuvanted with AS01 confers ~ 50% protection against clinical disease at one year (Olotu et al., 2013(Olotu et al., , 2016, which declines over time as anti-PfCSP antibodies wane (White et al., 2015). Thus, PfCSP-based immunogens must be improved to increase vaccine efficacy.
PfCSP is an optimal vaccine target since it is required for SPZ to infect hepatocytes (Cerami et al., 1992). PfCSP has three domains: an N-terminus, a central domain composed of repeating tetrapeptides, and a C-terminus (Bowman et al., 1999). In the Pf 3D7 reference strain, the region at the junction of the N-terminus and repeat domain commences with NPDP followed by 3 alternating NANP and NVDP repeats. This "junctional region" is followed by 35 NANP repeats, with a fourth NVDP inserted after the twentieth NANP (Cockburn and Seder, 2018).
Structural studies indicate that the motifs recognized by PfCSP mAbs in the repeat domain are DPNA, NPNV, and NPNA created by the joining of the three tetrapeptides Plassmeyer et al., 2009). Notably, RTS,S contains only 19 NANP repeats and the C-terminus (Stoute et al., 1997).
The immunodominant NANP repeats are targeted by nearly all neutralizing PfCSP mAbs reported so far (Julien and Wardemann, 2019). However, the isolation of rare and potent mAbs that preferentially bind NPDP (Kisalu et al., 2018;Tan et al., 2018) or the NVDP repeats (Wang et al., 2020) identified these subdominant epitopes as new sites of vulnerability on PfCSP. As these epitopes are not contained in RTS,S, these discoveries led to the development of nextgeneration vaccines against the junctional region (Atcheson et al., 2021;Calvo-Calle et al., 2021;Francica et al., 2021;Jelínková et al., 2021).
We recently compared the binding and potency of a panel of protective human PfCSP mAbs to determine which epitopes might improve immunogen design and select the most potent mAb for clinical development. mAb L9, which preferentially binds NVDP repeats and crossreacts with NANP repeats, was shown to be the most potently protective mAb in the panel.
Notably, L9 and other potent mAbs bound recombinant PfCSP (rPfCSP) in two binding events with distinct affinities by isothermal titration calorimetry (ITC), suggesting that this in vitro signature of "two-step binding" may be correlated with in vivo SPZ neutralization (Wang et al., 2020).
Similar to the majority of human PfCSP repeat mAbs, L9 is encoded by immunoglobulin (Ig) VH3-33 and Vκ1-5 (heavy and kappa chain variable domains) (Julien and Wardemann, 2019). Most VH3-33/Vκ1-5 mAbs preferentially bind NANP repeats and express an 8-aminoacid-long Vκ complementarity-determining region 3 (KCDR3:8) and a tryptophan at position 52 in CDRH2 (H.W52) Murugan et al., 2018Murugan et al., , 2020. Furthermore, these NANP-preferring mAbs require a minimal epitope of three NANP (Oyen et al., 2017) and rPfCSP binding is abrogated when VH3-33 is paired with Vκ1-5 with KCDR3:9 . Conversely, L9 expresses KCDR3:9 but binds rPfCSP with high affinity, preferentially binds NVDP repeats, and can only bind NANP repeats if they are sufficiently concatenated (Wang et al., 2020). These data suggest that VH3-33/Vκ1-5 mAbs can also target NVDP repeats; however, it is unclear whether the L9 B-cell lineage originated as a NANP binder that mutated to acquire NVDP affinity, as has been suggested for NPDP-reactive mAbs (Tan et al., 2018(Tan et al., , 2019, or vice versa. Here, to elucidate how the L9 B-cell lineage developed, mAbs clonally related to L9 were isolated. These mAbs preferentially bound NVDP repeats and had minimal-to-undetectable NANP reactivity. Pairing the L9 Igκ light chain (L9κ) with clonally-related IgH, compared to their original Igκ, resulted in chimeric mAbs with similar binding properties to L9 (i.e., binding two adjacent NVDP repeats, cross-reacting with NANP, and two-step binding rPfCSP) and improved SPZ neutralization in vivo. Structures of antigen-binding fragments (Fabs) from two chimeric mAbs in complex with a minimal peptide NANPNVDP showed nearly identical binding mechanisms, with the NPNV adopting a type-1 β-turn. This study provides insight into the evolution of antibodies targeting the subdominant NVDP minor repeats and the binding properties associated with the potent SPZ neutralization of L9.

Isolation of a panel of mAbs clonally related to L9
L9 was isolated from a volunteer immunized with radiation-attenuated PfSPZ (Lyke et al., 2017;Wang et al., 2020). To isolate clonally-related mAbs, CD3 -CD20 -CD19 + CD27 + CD38 + plasmablasts were sorted from the same volunteer 7 days after the second and third immunizations. A total of 480 plasmablasts were isolated, from which 194 high-quality IgG VH sequences were cloned. These sequences were compared to the L9 VH (L9H) to identify clonallyrelated sequences. Two VH (D2H and F10H) from plasmablasts collected after the second immunization were clonally related to L9H. Specifically, all three VH were encoded by VH3-33*01/06, JH4*02, and had a 13-amino-acid-long HCDR3; furthermore, the HCDR3 of D2H and F10H were 85% identical to L9H (Table S1).
Regarding light chains, only F10κ was cloned. F10κ and L9κ were Vκ1-5*03, Jκ1*01, and had a 9-amino-acid-long KCDR3; furthermore, the KCDR3 of F10κ was 89% identical to L9κ (Table S1). The VH/Vκ sequences of D2, F10, and L9 were used to infer their most recent common ancestor (L9MRCA) VH/Vκ, which were respectively 0.35% and 0% diverged from germline. Notably, the VH/Vκ of L9 were more somatically mutated than those of D2 and F10 (Table S1) and all VH genes encoded tryptophan at position 52 (H.W52) in HCDR2, which has been shown to be critical for binding NANP repeats . To elucidate the contribution of Igκ to mAb recognition of PfCSP and resultant SPZ neutralization, chimeric mAbs were created wherein L9H was paired with F10κ (L9HF10κ) and F10H was paired with L9κ (F10HL9κ). As D2κ was not recovered, L9κ and F10κ were paired with D2H to respectively create the chimeric mAbs D2HL9κ and D2HF10κ ( Figure 1A).
L9MRCA had the lowest rPfCSPFL binding, likely due to its low divergence from germline (Table   S1).
L9 was reported to preferentially bind the two adjacent NPNV motifs associated with the NVDP repeats in the 15mer peptide 22 (NANPNVDPNANPNVD) and weakly cross-react with the eight concatenated NPNA motifs in the 36mer peptide (NANP)9 (Wang et al., 2020). To determine if the mAbs in the panel had similar binding properties, we measured their peptide 22 and (NANP)9 binding by ELISA ( Figure 1B). Similar to L9, all mAbs demonstrated higher binding to peptide 22 than to (NANP)9. As with rPfCSPFL, L9κ-containing mAbs had higher peptide 22 binding than F10κ-containing mAbs and L9MRCA had lower, but detectable, peptide 22 binding. The only mAbs with weak, though detectable, (NANP)9 binding were L9 and D2HL9κ.
To further characterize the mAb panel's interactions with peptide 22, alanine scanning mutagenesis was used to define the critical residues within the peptide 22 sequence bound by each mAb ( Figure 1D). Unlike L9 IgG, which clearly bound both NPNV motifs, F10 IgG mostly bound the first NPNV motif in peptide 22 and did not appear to bind the second NPNV.
Remarkably, L9κ-containing IgG clearly bound both NPNV motifs while F10κ-containing IgG had more equivocal NPNV binding, with binding being targeted more towards the first NPNV.
Furthermore, size exclusion chromatography (SEC) of Fabs bound to peptide 22 showed that L9κ-containing Fab-peptide 22 complexes eluted earlier (i.e., were larger) than F10κ-containing Fab-peptide 22 complexes, suggesting that peptide 22 is bound by two L9κ-containing Fabs but only one F10κ-containing Fab ( Figure 1E). These data show that the higher peptide 22 avidity of L9κ-containing mAbs compared to F10κ-containing mAbs is because L9κ enables high-affinity binding to two adjacent NPNV motifs instead of one.
To extend the analysis for L9, ITC was used to measure the affinity and stoichiometry of L9 Fabs binding to 15mer peptides 22, 25 (NVDPNANPNVDPNAN), and 29 (NANPNANPNANPNAN), as well as a minimal 8mer peptide (NANPNVDP) ( Table 1). L9 bound ~2 binding sites with an affinity of 13 nM on peptide 22 (two NPNV) and ~1 binding site with an affinity of 1,000 and 1,900 nM on peptide 25 and NANPNVDP, respectively (one NPNV). L9 had no detectable binding to peptide 29, which contains 3 NPNA motifs.

F10HL9κ and L9HF10κ Fabs bind NPNV motifs in an identical manner
Next, we sought to structurally determine the binding mechanism of L9 using X-ray crystallography, but diffracting crystals could only be generated in the absence of peptides and a structure of L9 Fab alone (apo-L9) was solved to 2.93 Å ( Figure S1A). To gain structural insight into the NPNV binding mechanism of L9, we crystallized Fabs of F10HL9κ and L9HF10κ in complex with the NANPNVDP peptide and solved their structures to 1.89 Å and 2.23 Å, respectively (Figure 2A,C). F10HL9κ Fab binds NANPNVDP with the NPNV motif adopting a type-1 β-turn ( Figure S1B) that the NPNA motif was shown to adopt (Ghasparian et al., 2006) and is commonly found in NPNA-preferring repeat mAbs ( Figure 2B) Oyen et al., 2017;Pholcharee et al., 2021). All three HCDRs and the KCDR1 and KCDR3 bind NANPNVDP with a total buried surface area (BSA) of ~373 Å 2 , ~203 Å 2 from the IgH and ~170 Å 2 from Igκ ( Figure 2E). The two Asn residues form a network of hydrogen bonds with both IgH/Igκ, with the Val sitting in a hydrophobic pocket formed by three aromatic residues on the KCDR1, KCDR3, and HCDR3 ( Figure 2B). Surprisingly, the structure of L9HF10κ Fab-NANPNVDP showed an almost identical binding mechanism to F10HL9κ Fab-NANPNVDP ( Figure 2D,F). The NPNV adopts the type-1 β-turn structure and is bound with a total BSA of ~364 Å 2 , ~196 Å 2 from IgH and ~168 Å 2 from Igκ ( Figure 2E).
To further investigate the differences between L9HF10κ and F10HL9κ, we aligned both structures by their shared cognate NANPNVDP peptide. The two peptides align with a root mean square deviation (RMSD) of 0.189 Å 2 over six Cα atoms ( Figure 2F). The only two residues that differ in the binding site, positions 96 (L9:Phe and F10:Tyr) and 99 (L9:Gly and F10:Ser) in IgH by kabat numbering, are both main chain interactions and are unlikely to contribute to differences between the two chimeric mAbs. The interacting residues in both Igκ are identical.
There are three differences in Igκ that may contribute to the increased binding potential of L9κ.

The L9 B-cell lineage has baseline NPNV affinity
To further define the evolution of epitope reactivities in mAbs from the L9 B-cell lineage, the binding of L9MRCA, F10, and L9 to three rPfCSP constructs (FL, 5/3, ∆(NVDP)4; Figure S2A) were measured by ELISA; 317 was included as a NANP-preferring control mAb.
The binding preferences of the L9 lineage were corroborated by competition ELISA showing that a junctional peptide more potently competed mAb binding to rPfCSPFL and rPfCSP5/3 than (NANP)9, with the degree of competition correlating to the mAbs' somatic mutations ( Figure S2C). Notably, L9 was the only mAb that detectably bound rPfCSP∆(NVDP)4 and the junctional peptide more potently competed L9 binding than (NANP)9.
To extend the epitope mapping analysis, peptides 20-61 spanning the repeat domain were used to compete L9MRCA, F10, and L9 binding to rPfCSPFL ( Figure S2D). These data confirmed the markedly high affinity of L9 for peptide 22, the relatively high binding of F10 and L9MRCA to peptides 21/22/23/43 (all contain ≥1 NPNV), and all three mAbs' low/undetectable binding to peptides 27/29/61 (all lack NPNV). Furthermore, all three mAbs preferentially bound (NPNV)4 compared to (NPNA)4, confirming that the core motif recognized by these mAbs is NPNV.

L9κ confers NPNA cross-reactivity and two-step binding to rPfCSP
To elucidate how the mAb panel's relative affinities for NPNV and NPNA motifs impacted their binding to full-length protein, ITC was used to measure their binding stoichiometry and affinity to rPfCSPFL, rPfCSP5/3, and rPfCSP∆(NVDP)4. We previously reported that the most potent mAbs for protecting against in vivo SPZ challenge, including L9, bound rPfCSPFL in two binding events with distinct affinities (termed "two-step binding") and displayed high-affinity binding to the junctional region in rPfCSP5/3. Furthermore, we showed that L9 binds rPfCSP∆(NVDP)4 in a single step with lower affinity, confirming that L9 requires NPNV to two-step bind rPfCSP but can cross-react with NPNA motifs (Wang et al., 2020).
We next evaluated the ability of each mAb in the panel to lower parasite liver burden in normal C57/BL6 mice challenged intravenously (IV) with Pb-PfCSP-SPZ, a model used to quantify SPZ neutralization by PfCSP mAbs in vivo (Raghunandan et al., 2020). Passive transfer of 100 µg L9κ-containing mAbs mediated 87-96% liver burden reduction relative to untreated controls, which was significantly greater than the 21-52% reduction provided by F10κcontaining mAbs ( Figure 3B). Collectively, these data show that L9κ improves the ability of NVDP-preferring mAbs to bind and neutralize Pb-PfCSP-SPZ in vivo.

DISCUSSION
This study showed that the Igκ of L9, a highly potent human PfCSP mAb that preferentially binds the subdominant NVDP minor repeats (Wang et al., 2020), is critical for this antibody's unique binding properties and potent neutralization of SPZ. This finding was shown by pairing mAbs clonally related to L9 with different IgH/Igκ to create L9κ-containing chimeric mAbs that recapitulated the binding properties of L9 (i.e., binds two adjacent NPNV motifs, cross-reacts with concatenated NPNA motifs, and two-step binds rPfCSPFL). Conversely, pairing L9H and other VH with the Igκ of a closely related mAb, F10κ, resulted in chimeric mAbs that bound only one NPNV motif and lacked NPNA affinity and two-step binding. Furthermore, these data confirm that two-step binding is detected by ITC when a mAb exhibits two distinct affinities for the junctional region and NANP repeats on rPfCSPFL, as has been previously suggested (Wang et al., 2020).
These data also show that cross-linking two adjacent NVDP repeats, cross-reacting with NANP repeats, and two-step binding rPfCSP are binding properties correlated with higher binding to native PfCSP on Pb-PfCSP-SPZ in vitro and improved protection against in vivo Pb-PfCSP-SPZ challenge. Importantly, F10κ-containing mAbs had high (1-7 nM) rPfCSP5/3 affinity comparable to L9κ-containing mAbs by ITC, suggesting that high-affinity binding to the junctional region alone may not be sufficient for potent SPZ neutralization. Further studies with transgenic SPZ lacking NVDP repeats are required to determine whether protection mediated by L9κ-containing mAbs is due to their high-affinity divalent binding of adjacent NVDP repeats or lower-affinity multivalent interactions with NANP repeats. Collectively, this study identifies discrete binding properties associated with SPZ neutralization by NVDP-preferring mAbs and suggest that future PfCSP-based vaccines should include ≥2 adjacent NVDP repeats (NANPNVDPNANPNVDP) to induce neutralizing minor repeat antibodies.
In terms of elucidating the evolution of human minor repeat antibodies, this study showed that the L9 VH3-33/Vκ1-5 B-cell lineage has baseline affinity for NVDP repeats in the junctional region and somatically mutated to acquire NANP cross-reactivity. Specifically, the L9 lineage germline binds one NPNV and evolves to bind two adjacent NPNV motifs and concatenated NPNA motifs. This evolutionary pathway differs from a previous study showing that a large panel of VH3-33/Vκ1-5 PfCSP human mAbs had baseline NANP affinity centered on the conserved (N/D)PNANPN(A/V) core motif and evolved to acquire NPDP/NVDP crossreactivity (Murugan et al., 2020). Having shown that the core motif bound by L9 lineage mAbs is NPNV, our study suggests that some rare VH3-33/Vκ1-5 lineages initially target subdominant NVDP repeats and evolve later to cross-react with immunodominant NANP repeats. These data may assist in designing NVDP-containing PfCSP immunogens to expand naïve B-cells expressing the germline receptor for L9.
Lastly, structural studies showed that F10HL9κ and L9HF10κ both bound NPNV in the minimal peptide NANPNVDP in a type-1 β-turn that was near-identical to the NPNA type-1 βturn commonly observed for NANP-preferring mAbs Oyen et al., 2017;Pholcharee et al., 2021), suggesting that recognition of single NPNV motifs does not differ between these mAbs. The convergent conformations adopted by these subtly different tetrapeptides is in line with a previous study which also observed that cross-reactive PfCSP mAbs bind different repeat epitopes in near-identical conformations (Murugan et al., 2020).
Further structural studies on longer peptides containing ≥2 NPNV motifs and rPfCSPFL are needed to elucidate how L9κ contributes to binding adjacent NPNV motifs with high affinity and cross-reacting with NPNA motifs, as well as how these binding properties contribute to SPZ neutralization.

ACKNOWLEDGMENTS
A.S. was supported by contract HHSN261200800001E from the National Cancer Institute, NIH.
Y.F-G. and F.Z. thank the Bloomberg Philanthropies for their continued support. We thank B.

Lead Contact
Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Robert A. Seder (rseder@mail.nih.gov).

Materials Availability
All unique reagents generated in this study are available from the Lead Contact with a completed Materials Transfer Agreement.

Data and Code Availability
The heavy and light chain gene sequences of anti-PfCSP human mAbs isolated in this study were deposited in GenBank (Accession Numbers MZ686952 -MZ686956). The structure of apo-L9, F10HL9K-NANPNVDP, and L9HF10K-NANPNVDP were deposited in the Protein Data Bank (PBD), with PDB ID of 7RQP, 7RQQ, and 7RQR, respectively.

Mice
Female 6

Cell Lines
Expi293 and 293F cells used were from Thermo Fisher Scientific.
After infection, mosquitoes were maintained in an incubator at 19-20°C and supplied with a sterile cotton pad soaked in 10% sucrose, changed every 48 hrs. SPZ were harvested 20-22 days after blood feeding.

Production of recombinant PfCSP protein constructs
Recombinant PfCSP constructs were produced as previously described (Wang et al., 2020).
Briefly, the three rPfCSP constructs (FL, 5/3, ∆(NVDP)4) used in this study were cloned into the same CMV/R-expression vectors with a C-terminal AviTag, HRV3C-processing tag, and a 6X histidine tag (GenScript). rPfCSP constructs were expressed through transient transfection in 293F cells using the Freestyle 293F expression system (Thermo Fisher Scientific) at 37°C, 8% CO2 for 6 days, and purified from culture supernatants through polyhistidine-tag affinity chromatography followed by size exclusion chromatography (SEC) on an ÄKTA TM Start (GE Healthcare). Monomer-containing fractions were pooled, concentrated, snap frozen, and stored at −80°C.

Production of PfCSP peptides
All peptides used in this study were produced by direct synthesis and biotinylated by GenScript.

Production of recombinant immunoglobulins
Recombinant immunoglobulins were produced as previously described (Wang et al., 2020).
Briefly, RNA from lysed plasmablasts was reverse transcribed to cDNA (SuperScript First-Strand Synthesis System; Thermo Fisher Scientific) and immunoglobulin variable regions heavy and kappa chains were amplified using primer cocktails (Wang et al., 2020), sequenced (ACGT), and cloned into human IgG1 expression vectors (GenScript). Sequence analysis was performed using The International Immunogenetics Information System (IMGT, http://www.imgt.org/), with clonality being defined as having the same V/J genes, HCDR3 length, and HCDR3 sequence with >80% identity. Matched heavy and light chain constructs were co-transfected into Expi293 cells using the ExpiFectamine TM 293 Transfection Kit (Thermo Fisher Scientific) and cultures were incubated at 37°C, 8% CO2 for 6 days. Supernatants were harvested and IgG was purified using rProtein A Sepharose Fast Flow resin (GE Healthcare) and buffer exchanged with 1X PBS (pH 7.4) before being concentrated using Amicon Centrifugal Filters (Millipore).
Purified IgG concentrations were determined using a Nanodrop spectrophotometer.

Determination of L9MRCA sequence
The heavy chain and kappa chain sequence of L9MRCA were determined using a previously described informatics pipeline for antibodyome analysis (Zhu et al., 2013). The variable (V), diversity (D), joining (J), and V(D)J gene recombination were assigned and analyzed by IgBLAST (Ye et al., 2013). Multiple sequence alignments of the germline gene and sequences related to L9 were aligned by mafft (Katoh and Standley, 2013) and the MRCA was computed using dnaml with default parameters (Felsenstein, 1989).

Fab production and size exclusion chromatography
Purified recombinant IgGs were mixed with LysC (NEB) at a ratio of 1 μg LysC per 10 mg of IgG and incubated at 37°C for 18 hrs with nutation. The cleaved product was incubated with Protein A resin (GoldBio) at a ratio of 1 mL resin per 10 mg of initial IgG and incubated at room temperature for 1 hr to bind any uncleaved IgG and digested Fc. The purified Fab was further purified by SEC using a HiLoad 16/600 Superdex 200 pg column (GE Healthcare). Purified Fabs were incubated with 2-fold molar excess of peptide 22 and incubated at room temp for 1 hr. Fabpeptide complexes were run on a HiLoad 16/600 Superdex 200 pg column (GE Healthcare).
PfCSP mAbs were preincubated for 2 hrs at 37 o C with varying concentrations (0 -1,000 μg/mL) of either the junctional peptide 21-25 or the (NANP)9 peptide before being added to the coated plates. Plates were then incubated for 1 hr at room temperature with 1:20,000 dilution of HRPconjugated goat anti-human IgG (Thermo Fisher Scientific). The plates were washed five times with PBS-Tween between each step. After a final wash, samples were incubated for 10 min with 1-Step Ultra TMB-ELISA Substrate (Thermo Fisher Scientific). The optical density was read at 450 nm after addition of stopping solution (2N sulfuric acid). Competition of binding to rPfCSPFL by peptides 20-61, (NPNA)4 and (NPNV)4 was performed as described above, with plates coated with 200 ng/mL of rPfCSPFL and peptide concentrations ranging from 0 -1,000 μg/mL.

Alanine scan with peptide 22 variants
Alanine scanning mutagenesis competitive ELISA was performed as previously described (Wang et al., 2020)
Association with whole IgG (serially diluted from 16.67 to 1.04 μM) was done for 300 sec, followed by a dissociation step in buffer for 600 sec. Background subtraction of nonspecific binding was performed through measurement of association in buffer alone. Data analysis and curve fitting were performed using Octet software, version 7.0. Experimental data were fitted with the binding equations describing a 1:1 analyte-ligand interaction. Global analyses of the complete data sets, assuming binding was reversible (full dissociation), were carried out using nonlinear least-squares fitting allowing a single set of binding parameters to be obtained simultaneously for all concentrations of a given mAb dilution series.

Isothermal titration calorimetry
Isothermal titration calorimetry was carried out using a VP-ITC microcalorimeter ( Crystals were cryoprotected in ML supplemented with 30% ethylene glycol. Diffraction data was collected at ALS beamline 5.0.2 at 12286keV. For L9HF10K-NANPNVDP, diffracting crystals were grown in a ML containing 0.1M MES, pH 6.5, 18% PEG 4K, and 0.6M NaCl. Crystals were cryoprotected in ML supplemented with 30% ethylene glycol. Diffraction data was collected at Advanced Photon Source beamline 19-ID at 12669keV. All datasets were processed using XDS (Kabsch, 2010)  Coot (Emsley et al., 2010) and refinement was performed in Phenix. The data collection and refinement statistics are summarized in Table S2. Structural figures were made in PyMOL (Schrodinger, LLC).

Measurement of PfCSP mAb binding to SPZ
Salivary glands containing SPZ were dissected as previously described (  Statistics for X-ray crystal structures were calculated by Phenix and MolProbity and are summarized in Table S2.