Enhancing durability of CIS43 monoclonal antibody by Fc mutation or AAV delivery for malaria prevention

CIS43 is a potent neutralizing human mAb that targets a highly conserved “junctional” epitope in the Plasmodium falciparum (Pf) circumsporozoite protein (PfCSP). Enhancing the durability of CIS43 in vivo will be important for clinical translation. Here, 2 approaches were used to improve the durability of CIS43 in vivo while maintaining potent neutralization. First, the Fc domain was modified with the LS mutations (CIS43LS) to increase CIS43 binding affinity for the neonatal Fc receptor (FcRn). CIS43LS and CIS43 showed comparable in vivo protective efficacy. CIS43LS had 9- to 13-fold increased binding affinity for human (6.2 nM versus 54.2 nM) and rhesus (25.1 nM versus 325.8 nM) FcRn at endosomal pH 6.0 compared with CIS43. Importantly, the half-life of CIS43LS in rhesus macaques increased from 22 days to 39 days compared with CIS43. The second approach for sustaining antibody levels of CIS43 in vivo is through adeno-associated virus (AAV) expression. Mice administered once with AAV-expressing CIS43 had sustained antibody levels of approximately 300 μg/mL and mediated protection against sequential malaria challenges up to 36 weeks. Based on these data, CIS43LS has advanced to phase I clinical trials, and AAV delivery provides a potential next-generation approach for malaria prevention.

8 (287.3 μg/ml for CIS43LS and 280.4 μg/ml for CIS43) and was attained within 1h post mAb 151 administration (Supplemental Table 3). Although mAb titers were similar between CIS43 and 152 CIS43LS up to 14 days post infusion (Figure 2A), CIS43LS exhibited a longer T½ after this 153 timepoint, resulting in a titer of ~15 μg/ml at 98 days compared to ~1 μg/ml for CIS43. 154 Remarkably, CIS43LS remained detectable in serum for more than 140 days and persisted at 155 significantly higher levels (~3.7 μg/ml) than CIS43, which was no detectible after 126 days 156 following mAb administration (Figure 2A, p = 0.0003). We next analyzed the T½ for each mAb 157 using a two-compartment model (Supplemental Table 3 To provide an orthogonal approach for increasing antibody durability in vivo, VIP was used as a 202 gene based delivery approach (20-24). A previous study using an AAV vector encoding 2A10, a 203 mouse mAb against PfCSP, demonstrated high expression levels (> 1000 µg/ml), but only 204 partial protection following malaria challenge in mice after 8 weeks (23). Based on the enhanced 205 potentcy of CIS43 compared to 2A10, we hypothesized that VIP with CIS43 would elicit higher 206 mice and mice that passively received CIS43 (p = 0.0159) had a significant reduction (~2 logs) 220 of the liver-stage parasite burden compared to VRC01-AAV administered or untreated mice 221 when assessed two days post challenge ( Figure 3B). 2A10-AAV treated mice had a smaller 222 reduction (~1 log) in parasite burden ( Figure 3B). Remarkably, when these mice were further 223 assessed for blood stage infection 6 days post challenge, 9 out of 10 CIS43-AAV (p = 0.0005) 224 mice and 4 out of 5 mice that passively received CIS43 (p = 0.0102) had no detectable 225 parasitemia, whereas all mice treated with 2A10-AAV, VRC01-AAV or untreated mice had high 226 levels of parasitemia ( Figure 3B). Eight weeks post AAV administration, serum concentrations 227 of total human IgG were signficantly higher for 2A10 compared to the CIS43 (p = 0.0248) or 228 VRC01 (p = 0.0058) group (averages, 1114.3, 372.0, and 333.2 µg/ml, and ranges, 858.8-1302.5, 229 157.2-458.3 and 65.1-582.2 µg/ml, respectively ( Figure 3C). rPfCSP-specific Ab concentrations 230 were significantly higher for 2A10 than CIS43 (averages, 954.4 vs. 354.1 µg/ml and ranges, 231 608.5-1399.3 μg/ml vs. 186.9-470.5 μg/ml (p = 0.0409, Figure 3D). Similarly, 2A10 had higher 232 average skin concentrations of total human IgG (162.0 µg/g tissue weight) than CIS43 (61.1 µg/g 233 tissue weight, p = 0.0169) or VRC01 (56.0 µg/g tissue weight, p = 0.0051); ranges were 54.4-234 235.8 for 2A10, 38.1-110.5 for CIS43, and 6.1-94.3 µg/g tissue weight for VRC01 ( Figure 3E). 235 Skin homogenate samples from CIS43-AAV and 2A10-AAV-but not VRC01-AAV 236 administered mice, bound to rPfCSP (average, 59.0 µg/g tissue weight and range, 40.0-115.0 237 µg/g tissue weight for CIS43 vs. average, 175.9 µg/g tissue weight and range, 58.0-299.0 μg/g 238 tissue weight for 2A10), but the difference in Ab concentrations was not significant between 239 both groups ( Figure 3F). In assessing the kinetics of CIS43 in sera from a separate group of mice 240 following CIS43 AAV administration over time, the levels peaked ~2 weeks after administration 241 at a concentration of ~300 µg/ml, and remained stable up to 16 weeks (Supplemental Figure 3A). antibody expressed in vivo following AAV administration was functional, serum from mice 244 administered 2A10-and CIS43-AAV showed comparable binding to rPfCSP as recombinant 245 2A10 or CIS43 antibody by biolayer inferometry (Supplemental Figure 3C). Taken together 246 these data demonstrate that CIS43 expressed in vivo by AAV is highly protective and more 247 potent than 2A10 (17), despite 2A10 having higher serum antibody titers. 248 249 CIS43 AAV mediates long-term protection following malaria challenge 250 A potential advantage of using AAV is the ability to sustain antibody expression in vivo leading 251 to durable protection. To assess the long-term protective efficacy of CIS43-AAV, mice that 252 received CIS43-AAV, 2A10-AAV, or VRC01-AAV were challenged by the IV route 36 weeks 253 post-AAV administration. Mice that received CIS43-AAV or 300 µg of CIS43 given at the time 254 of challenge showed a significant reduction of both liver-stage parasite burden (~2 logs, p = 255 <0.0001) and blood-stage infection (p = 0.0013 for CIS43-AAV and p = 0.0235 for CIS43) 256 respectively compared to untreated mice ( Figure 4A). In contrast, mice that received 2A10-AAV 257 had a modest reduction in liver-stage parasite burden (~0.5-1 log, Figure 4A) and had 258 parasitemia similar to that seen in untreated mice ( Figure 4A). Thirty two weeks post AAV-259 administration, consistent with the findings in Figure 3, serum concentration of total human IgG 260 in these mice were signficantly higher for 2A10 (701.8 µg/ml) compared to CIS43 (186.9 µg/ml, 261 p = 0.0199) or VRC01 (184.5 µg/ml, p = 0.0279); ranges were 500.9-1000.1 µg/ml for 2A10; 262 99.8-348.4 µg/ml for CIS43; and 60.0-283.2 µg/ml for VRC01 ( Figure 4B). 263 infection over prolonged periods of time. Thus, CIS43-AAV-administered mice initially 266 challenged by mosquito bites and protected against malaria infection at week 3 ( Figure 3A) were 267 rechallenged by mosquito bites 11 weeks following AAV administration. Eight out of 9 mice 268 that received CIS43-AAV had no detectable parasitemia up to 12 days after rechallenge (p = 269 0.0001, log-rank test). In contrast, untreated mice exhibited parasitemia by day 4 ( Figure 4C). 4). Together, these data demonstrate that CIS43-AAV mediates sustained antibody production in 281 vivo and durable protection against primary and repeated malaria challenge. 282 Potent neutralizing mAbs are a promising approach for conferring high level protection 286 against malaria. While mAbs could be used to prevent malaria infection in travelers, health care 287 workers and military personnel, the greatest public health impact of this intervention would be 288 for seasonal control or elimination campaigns in Africa. Both of these indications would require 289 mechanisms to improve antibody durability. Here we provide two orthogonal approaches for 290 increasing the durability of mAb CIS43 in vivo. The present study shows that CIS43LS has a 291 longer serum T½ (38.7 ± 7.8 days) compared to CIS43 (22.3 ± 2.2 days) in NHP, with a titer of 292 ~10 g/ml at 15 weeks post-infusion. The serum T½ of CIS43LS reported here is higher 293 compared to a prior study with the HIV mAb VRC01LS (11.8 days) in NHP (18) (42). 317 318 A notable finding was that AAV can result in differential mAb expression in vivo. While it is 319 unclear why the 2A10-AAV expression was significantly higher than that of CIS43 or VRC01, 320 these data are consistent with a previous study, showing very high antibody levels by AAV 321 encoding 2A10 (23). Importantly, despite having higher antibody titers, 2A10 had more limited 322 protection than CIS43. These data highlight the primary importance of antibody potency for 323 mediating protection, which will be complemented by increased durability. 324

325
In summary, this report provides two approaches that could be used individually or potentially 326 together to enhance the durability of CIS43 in vivo. Based on the data reported here, a Phase I 327 clinical trial evaluating different doses of CIS43LS was initated this year to assess safety, 328 pharmacokinetics and protection against CHMI. As monoclonal antibody concentrations required 329 for protection against malaria infection in humans remains unknown, the results of the ongoing clinical trial with CIS43LS will provide important information for the future use of monoclonal 331 antibodies and potentially AAV as new modalities to prevent malaria. 332 333 334 Site-directed mutagenesis and rPfCSP probes generation 336 CIS43LS was generated through site-directed mutagenesis (GenScript). For rPfCSP production, 337 the codon-optimized synthetic gene corresponding to the full-length circumsporozoite (CS) 338 protein (NCBI gene ID: PF3D7_0304600) of 3D7, a clone of the NF54 strain of P. falciparum, 339 was used (GenScript). The 15-mer linear PfCSP peptides that overlapped by 11 residues 340 spanning the full length of PfCSP were described previously (17)  dose. Ten days following the boost, serum reactivity against CIS43 Fab was tested by ELISA. A 356 germline-reverted CIS43 mAb was generated by cloning the unmutated heavy and light chain 357 genes of CIS43. Mice that strongly reacted to CIS43 Fab but not to germline-reverted CIS43 Fab 358 were selected for cell fusion. Three days before cell fusion, mice with desired reactivity were 359 boosted with 20 μg of CIS43 Fab in a total volume of 100 µl PBS by IV injection via the tail 360 vein. Spleens were excised and purified splenocytes were fused with myeloma cells Sp2/0 361 (ATCC, Manassas, VA) in 2:1 ratio according to established fusion protocols (44). Ten days post 362 cell fusion, hybridoma supernatants were screened for binding to the CIS43 Fab but not to the 363 germline-reverted CIS43 Fab. Following three rounds of screening, positive clones were adapted 364 to serum-free hybridoma culture medium (Invitrogen) to facilitate monoclonal antibody 365 purification. Ultimately, to increase the yield, the DNA constructs encoding the heavy and light 366 chain of selected hybridomas were cloned into the mouse IgG2a expression vectors (GenScript) 367 and expressed in Expi293 cells. Antibodies were purified in-house using a recombinant protein-368 A column (GE Healthcare). Anti-idiotypic mAb 1-1 was selected for use in the downstream 369 experiments because of its potent binding to CIS43 Fab and lack of reactivity to germline-370 reverted CIS43 Fab. 371 372

Processing of skin punches to quantitate mAb
Binding of CIS43 and CIS43LS to rPfCSP or PfCSP peptides was carried out as described 383 previously (17) To quantitate CIS43, 2A10 and VRC01 induced by AAV administration in serum or 393 skin, ELISA plates were coated for 1h with 100μl of 1 μg/ml of goat anti-human IgG-Fc 394 antibody (Bethyl, cat. no. A80-104) for measuring total human IgG or plates coated with 395 100μl of 1 μg/ml PfCSP to assess PfCSP-specific mAbs. Serum or skin homogenate samples 396 serially diluted (1:100 to 1:312,500 for serum or 1:20 to 1: 62,500 for skin homogenate 397 respectively in 5-fold increments) in blocking solution were added to plate in duplicates and 398 the ELISA proceeded as described above. A horseradish peroxidase (HRP)-conjugated goat 399 anti-human kappa light chain antibody (Bethyl, cat. no. A80-115P) was used as a secondary 400 antibody. A standard curve was generated using each mAb. Antibody quantitation was made 401 from sample dillutions that fell within the linear range of the standard curve. For each 402 specimen, 6 five-fold serial dilutions starting at 1:100 (serum) or 1:20 (skin homogenate) were run in duplicate. 404 To assess the murine endogenous PfCSP antibody responses after the first malaria 405 challenge, serum samples collected 32 weeks following AAV administration, prior to the 406 second challenge, were diluted 1:100 in blocking solution and tested against rPfCSP by 407 ELISA as as described above and mouse PfCSP antibodies were detected with a peroxidase-408 labeled goat anti-mouse IgG (Invitrogen, cat. no. 62-6520).

Mouse challenge studies with chimeric Pb-PfCSP-LUC SPZ 451
Malaria challenges to assess the antibody protective efficacy were done as previously described 452 (17 The mosquito bite challenge was done as previously described (17). Briefly, C57BL/6 mice were 474 administered with the AAV constructs or were IV injected with 300 μg of CIS43, CIS43LS, or 475 VRC01 as detailed above. Mice were anesthetized with 2% avertin (Alfa Aesar) and 10 minutes 476 later mice were exposed to 6 infected mosquitoes/mouse for ∼10 min, after which engorged 477 mosquito abdomens were visually inspected for blood to determine whether the mosquito has 478 bitten. Mouse parasitemia was monitored daily by Giemsa staining of blood smears from day 4 479 through day 12 post-infection. The mosquito bite challenge was done 3 weeks (primary 480 challenge) and 11 weeks (rechallenge) post AAV administration. After thawing, each AAV construct was diluted with PBS (pH 7.4) to 1 × 10 11 GC in a 50 μL 543 volume and injected into mice in a single injection (50 μL) into the cranial thigh muscle. Mice 544 were bled by the tail vein or skin punches were sampled bi-weekly following AAV 545 administration to assess the Ab expression. 546 547

Statistical analyses 548
All data were plotted and graphed using GraphPad Prism, version 8.4.3, unless otherwise stated. 549 For stoichiometry, errors with 95% confidence were analyzed from the fits of the data. To 550 measure CIS43, CIS43LS, 2A10 and VRC01 in macaque or mouse serum and skin, standard 551 curves were fitted with a hyperbolic parameter curve, and concentration values were 552 interpolated. For mAb in macaques, differences in serum concentrations between the CIS43LS 553 and CIS43 group over days 0-140 were calculated using the no sphericity 2way ANOVA 554 analysis. Differences in Ab concentrations between groups and liver parasite burden were 555 determined for significance using the non-parametric One-way ANOVA Kruskal-Wallis test 556 (with Dunn's correction for multiple comparisons). Parasitemias for the mosquito bite challenge 557 were analyzed by Kaplan-Meier curves using the log-rank test. A statistically significant 558 difference is indicated by P < 0.05. 559 copies (GC) AAV encoding CIS43, 2A10 or VRC01. C57BL/6 mice were challenged by 779 mosquito bites 3 weeks following AAV administration of the indicated mAb. As a control 300 780 μg-rCIS43 was administered 2h prior to challenge. Kaplan-Meier curves, analyzed using the log-781 rank test, show the frequencies of mice free of parasites. Differences between CIS43-AAV-782 administered or 300 μg rCIS43-treated mice and untreated mice are shown (P = 0.0001). wpa, 783 weeks post-AAV administration. n = 10 per group; 2A10-AAV, n = 8. 300 μg-rCIS43, mice that 784