Generation and characterization of VHH nanobodies targeting PfPIMMS43

PfPIMMS43 is a 505-amino acid protein with an N-terminal signal peptide and a C-terminal GPI anchor. To generate nanobodies against PfPIMMS43, we produced a recombinant Thioredoxin-His-tagged version of PfPIMMS43, excluding the signal peptide and GPI anchor sequences (amino acids 26-481), in Escherichia coli as previously described (Ukegbu et al., 2020). This recombinant protein was used to immunize llamas, generating an immune nanobody library from which nine nanobodies (G9, E5, C12, E2, A3, A5, B11, H6, and E1) were selected. These nanobodies were chosen based on variations in their antigen-binding complementary determining regions (CDR1-3) (Fig. 1A). CDR1 and CDR2 showed median sequence identities of 25% and 22%, respectively, while CDR3, the region most involved in antigen binding, displayed a median sequence identity of less than 18% between the nine nanobodies. The length of CDR1 and CDR2 was consistently eight amino acids, except for the 7-amino acid CDR2 found in E5, E2, and A3 nanobodies. The CDR3 length varied significantly, ranging from 8 to 21 amino acids (median 15), with E5 having the shortest CDR3 (Fig. 1A). Despite the CDR diversity, the framework regions (FRs) were highly conserved, with some variations in the first and middle regions of FR3. Nanobodies are known to sometimes utilize FR regions to enhance antigen-binding diversity³².

A Multiple sequence alignment of the nine PfPIMMS43-specific nanobodies, highlighting the four framework regions (FR) and three complementarity-determining regions (CDR). Sequence conservation is represented by shades of blue, with complete conservation shown in dark blue and non-conserved regions unshaded. B SDS-PAGE gel analysis of recombinant MYC-6xHis-tagged PfPIMMS43 nanobodies expressed in E. coli and purified by affinity chromatography. C Western blot detection of recombinant thioredoxin-His-tagged PfPIMMS43 using the nine nanobodies. Binding was visualized with an HRP-conjugated anti-VHH antibody. Probing with an anti-His antibody served as a control (HIS). A Coomassie-stained SDS-PAGE gel of recombinant PfPIMMS43 prior to blotting was used as a loading control (COO). D ELISA showing nanobody binding affinities to recombinant PfPIMMS43. E Detection of endogenous PfPIMMS43 in P. falciparum NF54 ookinetes using the four highest affinity nanobodies (G9, E5, C12, and E2). Reduced cell lysates from P. falciparum-infected midguts, collected 18 h post-blood meal, were probed with nanobodies, and binding was detected using an anti-MYC antibody. An anti-Pfs25 antibody served as a positive control. The fluorescence microscopy image shows ookinetes (red, stained with anti-Pfs25) invading midgut epithelial cells (nuclei stained blue with DAPI), corresponding to midgut samples used for western blot analysis. IN, infected midguts; BF, blood-fed midguts. Arrows in the immunofluorescence image indicate Pfs25-stained ookinetes and arrowheads indicate DAPI-stained epithelial cell nuclei. Smaller nuclei correspond to other cells in the midgut epithelium. Scale bar, 5 µM.

All nine nanobodies were expressed in E. coli as Myc-His fusion proteins, which under non-reducing conditions on SDS-PAGE gels migrated between 15 and 20 kDa (Fig. 1B). To assess the ability of nanobodies to recognize recombinant TRX-His-PfPIMMS43, we performed western blot analyses using an anti-VHH antibody. All nine nanobodies successfully recognized the full-length (FL) recombinant PfPIMMS43 (~75 kDa) under non-reducing conditions (Fig. 1C, Fig. S1). Two lower bands detected were likely degradation products of recombinant PfPIMMS43, as they were also detected by the anti-His antibody.

The binding affinities of the nanobodies were evaluated using ELISA (Fig. 1D). Four of the nanobodies, G9, E5, C12, and E2, exhibited high nanomolar binding affinities to recombinant PfPIMMS43 (3, 5, 6, and 8 nM, respectively). The remaining 5 nanobodies demonstrated lower binding affinities, greater than 50 nM.

Therefore, we selected the four nanobodies with the highest affinity (G9, E5, C12, and E2) and investigated their ability to detect the ~60 kDa PfPIMMS43 protein expressed in P. falciparum ookinetes. Previous studies using polyclonal antibodies demonstrated that PfPIMMS43 is expressed on the surface of invading ookinetes 18–25 h post blood feeding (Ukegbu et al., 2020). Western blot analyses of P. falciparum NF54-infected An. coluzzii midguts at 18 h post blood feeding revealed that all four nanobodies detected a ~ 60 kDa band corresponding to PfPIMMS43 (Fig. 1E, Fig. S2). This band was absent in non-infected blood-fed midguts, which served as controls. As a stage-specific control for PfNF54 ookinete presence, Pfs25 was detected in infected midguts but not in the blood-fed controls.

Transmission blocking of P. falciparum by PfPIMMS43 nanobodies

We assessed the ability of the G9, E5, C12, and E2 PfPIMMS43 nanobodies to block P. falciparum NF54 transmission using An. coluzzii standard membrane feeding assays (SMFAs). Oocyst numbers in mosquito midguts were counted 8–10 days post-infectious blood feeding. Phosphate-buffered saline (PBS) was used as the control, as nanobodies were dissolved in this solution. No significant difference in oocyst intensity was observed when P. falciparum NF54 gametocytes were fed to An. coluzzii compared to those spiked with PBS (Table S1). We tested three different nanobody concentrations, 25, 50, and 100 µg/ml, and found that oocyst reduction was concentration-dependent, with the greatest inhibition observed at the highest concentration (Fig. 2A, B, Table S1). At the highest concentration of 100 µg/ml, G9, E5, C12, and E2 significantly reduced oocyst numbers by 86%, 99%, 89%, and 83%, respectively, compared to the PBS control (Fig. 2B, Table S1). For G9 and E2, significant TRA were also observed at both 50 µg/ml and the lowest concentration of 25 µg/ml: 70% and 58% for G9, and 73% and 66% for E2, respectively. C12 showed significant 58% TRA at 50 µg/ml but not at 25 µg/ml concentrations, while E5 showed significant 99% TRA at a 100 µg/ml concentration, the highest of all nanobodies (Fig. 2B, Table S1).

A Oocyst counts per midgut at 8–10 days post-blood meal from SMFAs using An. coluzzii N’gousso mosquitoes infected with P. falciparum NF54 gametocytes. Gametocytes were treated with G9, E5, C12, and E2 nanobodies, with PBS-treated gametocytes serving as the control. Data from three biological replicates per nanobody treatment were pooled. Red horizontal lines indicate the median oocyst count. Statistical significance was assessed using the Mann–Whitney test: ns, not significant; *P < 0.05; **P < 0.001; ***P < 0.0001. B Percent transmission-reducing activity (TRA) of the PfPIMMS43 nanobodies G9, E5, C12, and E2 relative to the PBS control, calculated from the results in A. Circles represent individual data. Error bars represent the SEM. Statistical significance was determined using an unpaired t-test with Welch’s correction: ns, not significant; *P < 0.05; **P < 0.001; ***P < 0.0001. C Oocyst counts per midgut at 8–10 days post-blood meal from DMFAs using An. gambiae Ifakara mosquitoes infected with P. falciparum gametocytes collected from children in Tanzania. Gametocytes were treated with G9 and E5 nanobodies, with PBS-treated gametocytes as the control. Data from three biological replicates were pooled. Statistical analyses of oocyst load and TRA were performed as in A, B.

Next, we tested the ability of G9 and E5, the two nanobodies with the highest affinities to recombinant PfPIMMS43, to block the transmission of natural P. falciparum isolates. Using direct membrane feeding assays (DMFAs) on blood samples from gametocytaemic children in Tanzania (Table S2), spiked with G9 and E5 nanobodies at 25, 50, and 100 µg/ml, we counted oocyst numbers 8–10 days post blood feeding of local A. gambiae mosquitoes (Ifakara strain). A. coluzzii is not endemic in Tanzania, and the two species are very closely related and often used interchangeably in both laboratory and field settings. The results were consistent with the laboratory-based SMFAs, with the highest TRA observed at 100 µg/ml for both nanobodies (Fig. 2C, Table S3). At this concentration, G9 and E5 reduced oocyst numbers by 99% and 79%, respectively, from an average of 2.3 to 0 and 0.5 oocysts per midgut, respectively (Fig. 2C, Table S3). G9 also significantly reduced transmission by 56% at 25 µg/ml and 68% at 50 µg/ml. Interestingly, E5 performed better against field P. falciparum isolates than against the laboratory NF54 strain, significantly reducing infection by 66% at 25 µg/ml and 76% at 50 µg/ml. Both nanobodies significantly reduced the mosquito infection prevalence in all concentrations tested (Table S3). Raw data obtained from infections with laboratory NF54 and natural P. falciparum isolates are presented in Supplementary Data File 1.

Epitope mapping of VHH-PfPIMMS43 interactions

The epitopes of the PfPIMMS43 protein recognized by the G9, E5, C12, and E2 nanobodies were identified using an antigen-binding and enzyme-digestion assay. Immobilized nanobodies were bound to recombinant TRX-His-PfPIMMS43, and the resulting VHH-PfPIMMS43 complexes were digested with trypsin. Eluted peptides from the bound complex were then compared to non-bound peptides and a negative control. Several unique peptides, present only in the nanobody-bound fraction, were identified (Supplementary Data File 2), and mapped to the PfPIMMS43 sequence (Fig. 3A). Some of these epitopes were bound by more than one of the nanobodies but some were unique for different nanobodies.

A Schematic representation of the full-length (FL) recombinant PfPIMMS43 protein and its subdomains D1-D5, created through progressive C-terminal deletions. Amino acid sequences corresponding to the regions bound by each tested nanobody, as determined by LC-MS/MS following VHH-PfPIMMS43 binding and trypsin digestion, are shown. All constructs were expressed in E. coli as Thioredoxin-His fusion proteins. TRX-Thioredoxin, H-6xHis, T-Thrombin cleavage site, S-S-tag and E- enterokinase cleavage site. B Western blot detection of recombinant Thioredoxin-His-tagged PfPIMMS43 FL and its truncated variants D1-D5 using G9, E5, C12, and E2 nanobodies. Each nanobody was tested individually, and binding was detected using an HRP-conjugated anti-VHH antibody. A Coomassie-stained SDS-PAGE gel of the recombinant PfPIMMS43 proteins, served as a loading control. C Homology model of the G9-PfPIMMS43 complex. PfPIMMS43 is colored based on pLDDT confidence scores: pLDDT > 90 in blue, 90 > pLDDT > 70 in cyan, and pLDDT<70 in fading shades of silver. G9 is represented with framework regions (FR) in black, CDR1 in pale cyan, CDR2 in green, and CDR3 in red. Panels: (i) surface representation of the G9-PfPIMMS43 complex; (ii) cartoon representation of the region indicated in panel (i); (iii) zoom-in of the region indicated in the panel (ii) with side chains of interacting residues from G9 CDR2 (green), CDR3 (red) and PfPIMMS43 (blue) shown; (iv) Back side of the region shown in panel (iii) with side chains of interacting residues shown.

To validate the importance of these identified peptides for nanobody recognition, subdomains of PIMMS43 spanning different epitopes were designed, expressed and tested for binding with the nanobodies. Like the full-length PfPIMMS43 spanning amino acids 26-481 (~74 kDa), lacking the N-terminal signal peptide and C-terminal GPI-anchor sequence, these subdomains were produced as thioredoxin-His-tagged fusion proteins (Fig. 3A). They included: Domain 1 (D1) spanning amino acids 26-257 (47 kDa), D2 spanning amino acids 26-291 (51 kDa), D3 spanning amino acids 26-354 (58 kDa), D4 spanning amino acids 26-414 (65 kDa), and D5 spanning amino acids D26-K454 (69 kDa).

In western blots, compared to the full-length PIMMS43, G9 did not bind to subdomains D1 and D2 but was able to bind to D3, D4, and D5 (Fig. 3B). The binding was associated with two putative peptide epitopes: MGNDLANINISFFASEQR, found in D3, D4, and D5, and LLSQDEYIKELVK, present only in D4 and D5. E5 displayed a similar binding profile to G9, suggesting that it recognizes the same epitope bin (Fig. 3B). For C12, the strongest binding was observed with subdomain D5, which contains the putative epitope TNEQEVTISK, suggesting that this specific region is critical for C12 recognition (Fig. 3B, Fig. S3). Finally, while E2 showed some recognition of subdomains D4 and D5, which contains the epitope LLSQDEYIKELVK, the relatively weak signal to D4 suggests that the C-terminal peptide epitope KEQIIKEVK in D5 may be essential for recognition by E2 (Fig. 3B).

Homology modeling of VHH-PfPIMMS43 interactions

We used AlphaFold to generate a 3D homology model of the G9 nanobody binding to PfPIMMS43 (Fig. 3C). Initial attempts to model the interaction with the full-length PfPIMMS43 could not produce a confident 3D structure, as the protein contains large unstructured regions. Indeed, earlier attempts to analyze the structure of recombinant PfPIMMS43 using nuclear magnetic resonance (NMR) suggested that much of the protein is unfolded. Subsequently, the model was refined to focus on the most confidently modeled PfPIMMS43 region, spanning amino acids 44-367 and encompassing three of the predicted epitopes determined by the mapping assay (Fig. 3A). The resulting model of the PfPIMMS43-G9 complex had an inter-residue predicted TM-score (ipTM) of 0.81 and a predicted TM-score (pTM) of 0.51, indicating a reasonable level of confidence.

The folding of PfPIMMS43 was predicted with high confidence (pLDDT > 90) for amino acid residues 250–352, particularly for residues 292–301, 311–323, and 331–342, which together form a β-sheet structure (Fig. 3Ci). Three α-helices (amino acids 250–263, 266–282, and 344–352) showed moderate confidence (pLDDT > 70). The G9 nanobody structure displayed the classical β-sandwich immunoglobulin fold, with high confidence (pLDDT > 90) in the fold, and moderate confidence (90 > pLDDT > 70) for the CDR loops (Fig. 3Ci).

Both cartoon and surface representations of the model indicated that G9 binds to PfPIMMS43 primarily through its CDR2 and CDR3 domains, with the interaction dominated by CDR3 (Fig. 3Cii, iii). A detailed analysis of sidechain residues revealed that CDR3 interacts with residues in the high-confidence β-sheet of PfPIMMS43, including G308, N309, D310, L311, N313, Y337, L338, N340, and N342. CDR2 engages residues Y341 and N342 from the β-sheet, as well as E345 from an adjacent α-helix (Fig. 3Civ). The predicted aligned error (PAE) for these interactions was less than 5 Å, further supporting the accuracy of the predictions. Importantly, residues G308, N309, D310, L311, and N313 are part of the predicted peptide epitope MGNDLANINISFFASEQR (Fig. 3A). Attempts to model interactions of E5, C12, and E2 with PfPIMMS43 were unsuccessful.