Nanobody generation and structural characterization of Plasmodium falciparum 6-cysteine protein Pf12p

Surface-associated proteins play critical roles in the Plasmodium parasite life cycle and are major targets for vaccine development. The 6-cysteine (6-cys) protein family is expressed in a stage-specific manner throughout Plasmodium falciparum life cycle and characterized by the presence of 6-cys domains, which are β-sandwich domains with conserved sets of disulfide bonds. Although several 6-cys family members have been implicated to play a role in sexual stages, mosquito transmission, evasion of the host immune response and host cell invasion, the precise function of many family members is still unknown and structural information is only available for four 6-cys proteins. Here, we present to the best of our knowledge, the first crystal structure of the 6-cys protein Pf12p determined at 2.8 Å resolution. The monomeric molecule folds into two domains, D1 and D2, both of which adopt the canonical 6-cys domain fold. Although the structural fold is similar to that of Pf12, its paralog in P. falciparum, we show that Pf12p does not complex with Pf41, which is a known interaction partner of Pf12. We generated 10 distinct Pf12p-specific nanobodies which map into two separate epitope groups; one group which binds within the D2 domain, while several members of the second group bind at the interface of the D1 and D2 domain of Pf12p. Characterization of the structural features of the 6-cys family and their associated nanobodies provide a framework for generating new tools to study the diverse functions of the 6-cys protein family in the Plasmodium life cycle.


Introduction 24
Plasmodium falciparum is the most lethal of human malaria species and responsible for 25 the majority of malaria related deaths [1]. One of the key protein families in P. falciparum is 26 the 6-cysteine (6-cys) protein family with members representing some of the most abundant 27 surface-expressed proteins across all stages of the malaria parasite life cycle [2]. In P. 28 falciparum, there are 14 members in the 6-cys protein family and they share a common 29 structural feature, the 6-cys domain or otherwise referred to as the s48/45 domain. In general, 30 the 6-cys proteins interact with specific human or mosquito proteins for entry into host tissues 31 or to evade the host immune response to promote survival of the malaria parasites [3][4][5][6]. 32 Furthermore, several members of the 6-cys proteins are involved in parasite sexual 33 development and fertilization of gametes [7,8]. 34 The 6-cys proteins are expressed during multiple stages of the parasite life cycle; 35 Pfs230, Pfs48/45, Pfs230p, Pfs47 and PfSOP12 in the sexual stages; Pf52, Pf36, PfLISP2 and 36 PfB9 in the liver stages; and Pf12, Pf12p, Pf41, Pf38, and Pf92 in the blood stages [2]. Pfs230 37 and Pfs48/45 are the leading transmission blocking vaccine candidates against malaria [9,10]. 38 Pfs230 forms a complex with Pfs48/45 on the surface of gametocytes, and the complex is 39 involved in the fertilization of male and female gametes [8,[11][12][13]. Monoclonal antibodies 40 (mAbs) against Pfs230 and Pfs48/45 are effective in blocking transmission by inhibiting 41 successful gamete fertilization [14,15]. Pfs47 is expressed on the surface of female 42 gametocytes, zygotes and ookinetes [16][17][18]. The natural selection of specific Pfs47 haplotypes 43 is consistent with the adaptation of P. falciparum to different Anopheles mosquito species. 44 Through its interaction with a specific mosquito midgut receptor protein, Pfs47 is involved in 45 a lock and key model that drives host tropism between parasite and mosquito [6]. Pf52 and 46 Pf36 are present on the surface of sporozoites and are crucial for invasion of hepatocytes and 47 the formation of a parasitophorous vacuole that envelopes the growing parasite. Entry into liver 48 cells is proposed to involve Pf36 interaction with hepatocyte receptors EphA2, CD81, and 49 Scavenger Receptor BI, which is a critical step for successful malaria infection in the human 50 host [3,19,20]. Pf12 and Pf41 form a complex on the merozoite surface and are targets of 51 naturally acquired immunity [21][22][23][24][25][26]. Pf12 is the fifth most prevalent 52 glycosylphosphatidylinositol (GPI)-anchored protein on the merozoite surface and Pf12 and 53 Pf41 have been implicated to be involved in red blood cell invasion [4,27]. Pf92 is an abundant 54 merozoite surface protein and constitutes about 5% of the total surface coat [27]. Pf92 plays a 55 role in immune evasion by recruiting human complement regulator Factor H, which is the 56 major complement regulator of the alternative pathway of complement. This recruitment serves 57 to downregulate complement activation on the merozoite surface and protect P. falciparum 58 merozoites from complement-mediated lysis [5]. 59 The 6-cys domain has two to six cysteines that form disulfide bonds and is evolutionary 60 related to the SAG1-related sequence (SRS) domain in Toxoplasma gondii [28]. The 6-cys and 61 SRS domains have been proposed to be derived from an ephrin-like precursor originating from 62 a vertebrate host protein, with a general function to mediate extracellular protein-protein 63 interactions and cellular adhesion [28]. The 6-cys domain is characterized by a β-sandwich fold 64 of parallel and anti-parallel β-strands and a conserved cysteine motif [28][29][30]. The β-sandwich 65 is formed by two β-sheets, usually termed A and B, that are pinned together by two disulfide 66 bonds. A third disulfide bond connects a loop region to the core structure and a small β-sheet 67 of two β-strands runs perpendicular along the side of β-sheet B [24,26,[28][29][30]. Between one 68 to fourteen 6-cys domains (denoted as D1, D2, D3…) are present in each 6-cys protein and 69 they are often found in tandem pairs of A-and B-type 6-cys domains [26,30]. The two types 70 of 6-cys domains differ in the number of β-strands in β-sheet A, with A-type domains usually 71 containing four and B-type domains usually containing five β-strands. The position and 72 connectivity of cysteines differ between the 6-cys domain compared to the SRS-domain in T. 73 gondii. In the 6-cys domain the disulfide bond connectivity follows a C1-C2, C3-C6, C4-C5 74 pattern, whereas in the SRS domain it follows a C1-C6, C2-C5, and C3-C4 pattern [31][32][33] crystal structures of all 6-cys proteins will be required to fully understand the diverse functions 84 of this protein family. Alpacas, llamas and their camel cousins have evolved one of the smallest 85 naturally occurring antigen recognition domains called nanobodies. Nanobodies are ~15 kDa 86 in size, display strong binding affinities to target proteins and also function as structural 87 chaperones to assist in crystal formation. Nanobodies may be used both to assist in the 88 crystallisation of 6-cys proteins and to block malaria parasite invasion by inhibiting the specific 89 functions of 6-cys proteins. 90 While several family members play critical roles in the parasite life cycle, many 6-cys 91 proteins are not well characterized and their precise functions are unknown. One of the 92 understudied 6-cys protein is Pf12p, which is a paralog of Pf12 [30]. Pf12p is predicted to 93 contain a signal peptide, two 6-cys domains and a GPI-anchor that links the protein to the 94 parasite surface [27]. Microarray data indicates that Pf12p is transcribed in blood stages and 95 mass spectrometry data suggests that this protein is also present in sporozoites but the function 96 of Pf12p is currently unknown [27,41,42]. To further characterize Pf12p using structural 97 methods and to generate antibody tools that are specific to Pf12p, we immunized an alpaca 98

Expression and purification of nanobodies 149
Nanobodies were expressed in E. coli WK6 cells. Bacteria were grown in Terrific Broth 150 at 37 °C to an OD600 of 0.7, induced with 1 mM IPTG and grown overnight at 28 °C for 16 h. 151 Cell pellets were harvested and resuspended in 20% sucrose, 20 mM imidazole, 150 mM NaCl 152 DPBS and incubated for 15 min on ice. 5 mM EDTA was added and incubated on ice for 20 153 minutes. After this incubation, 10 mM MgCl2 was added and periplasmic extracts were 154 harvested by centrifugation and the supernatant was loaded onto a 1 ml HisTrap FF column 155 (GE Healthcare). The nanobody was eluted via a linear gradient into 400 mM imidazole, 100 156 mM NaCl, PBS. The appropriate fractions were concentrated and subjected to SEC (SD200 157 increase 10/300) pre-equilibrated in 20 mM HEPES pH 7.5, 150 mM NaCl. The affinity of Pf12 and Pf12p binding to Pf41 were measured using the method above 199 with the following modifications. His-tagged Pf41 (10 µg/mL or 20 µg/mL) was loaded onto 200 NTA sensors until a response shift of 1.8 nm. Association measurements were performed using 201 a two-fold dilution series from 16-500 nM (if loaded with 10 µg/mL of Pf41) or 31-1000 nM 202 (if loaded with 20 µg/mL of Pf41) of untagged Pf12 D1D2 and untagged Pf12p D1D2. 203

Competition binding experiment using BLI 204
For competition experiments using BLI, 150 nM untagged Pf12p D1D2 was pre-205 incubated with each nanobody at a 10-fold molar excess for one hr at RT. A 30 s baseline step 206 was established between each step of the assay. NTA sensors were first loaded with 10 µg/mL 207 of nanobody for 5 min. The sensor surface was then quenched by dipping into 20 µg/mL of an 208 irrelevant nanobody for 5 min. Nanobody-loaded sensors were then dipped into premixed 209 solutions of Pf12p D1D2 and nanobody for 5 min. Nanobody-loaded sensors were also dipped 210 into Pf12p D1D2 alone to determine the level of Pf12p D1D2 binding to immobilized 211 nanobody in the absence of other nanobodies. Percentage competition was calculated by 212 dividing the max response of the premixed Pf12p D1D2 and nanobody solution binding by the 213 max response of Pf12p binding alone, multiplied by 100. 214

Size Exclusion Chromatography Binding Studies of Pf12p with Pf12 and Pf41 243
Complexation was carried out by incubating 100 µg Pf12p or Pf12 with Pf41 at a 1:1 244 molar ratio for one hr at RT. 30 µL of the sample was loaded onto an SEC column (Superdex 245 200 3.2/300) pre-equilibrated in 20 mM HEPES pH 7.5, 150 mM NaCl using a 100 µL loop. 246 The run was carried out using a 0.03 ml/min flow rate and 100 µL fraction size. Equivalent 247 amounts of Pf12, Pf12p and Pf41 were run singly for comparison of retention volumes to assess 248 complex formation. 249

Results 250
Isolation and characterization of Pf12p-specific nanobodies 251 A 10 8 nanobody phage display library was generated from an alpaca immunised with 252 recombinant Pf12p D1D2 and used to select for Pf12p-specific nanobodies. After two rounds 253 of bio-panning, we identified ten distinct nanobody clonal groups based on differences in the 254 amino acid sequence of the complementary determining region 3 (CDR3) ( Figure 1A). The 255 CDR3 regions of the nanobodies vary in at least one amino acid with lengths between 8 to 21 256 residues. One member of each clonal group was selected for further characterization and will 257 be referred to as A10, B2, B9, B12, C4, C12, D9, F7, G6, and H7. These nanobodies were 258 expressed and purified with overall yields of 1-12 mg per litre of initial culture and migrated 259 between 13 and 17 kDa on SDS-PAGE under reducing conditions ( Figure 1B). 260 To examine the specificty of these Pf12p-specific nanobodies, we used recombinant 261 Pf12p and two other recombinant 6-cys proteins, Pf12 and Pf41 in an ELISA-based assay 262 ( Figure 1C). All three recombinant proteins consist of two 6-cys domains. Pf12p shares 16.7% 263 sequence identity with Pf12 and 12.8% with Pf41. All ten nanobodies recognize Pf12p but do 264 not bind to Pf12 nor Pf41. Pf12-and Pf41-specific nanobodies, D12 and A4, respectively, do 265 not cross-react with Pf12p. Collectively, these results show that the ten nanobodies are specific 266 to Pf12p and are not cross-reactive with two other 6-cys proteins. 267 We wanted to determine whether the Pf12p-specific nanobodies are able to detect Pf12p 268 by Western blotting under reducing and non-reducing conditions ( Figure 1D). Six of the ten 269 nanobodies, A10, B9, B12, C4, D9, and G6 showed no or weak reactivity under both 270 conditions. Four nanobodies, B2, C12, F7, and H7, recognize the reduced and non-reduced 271 Pf12p to different extents. All four nanobodies above show a stronger signal with non-reduced 272 Pf12p compared to reduced protein, indicating that the presence of disulfide bonds improves 273 the recognition of Pf12p by these nanobodies using Western blotting. 274 We used bio-layer interferometry (BLI) to determine the binding kinetics and affinities 275 of the interaction between nanobodies and Pf12p ( Figure 1E and Supplementary Figure S1). 276 Nine out of ten nanobodies bind recombinant Pf12p with high affinity in the low nanomolar 277 range, with association rates around 10 5 M -1 s -1 and dissociation rates between 10 -2 and 10 -4 s -1 . 278 A10 which is the weakest binding nanobody has an affinity of ~100 nM. 279 280

Pf12p-specific nanobodies bind to two separate regions on Pf12p 281
To determine whether the Pf12p-specific nanobodies bound epitopes within the D1 or 282 D2 domain of Pf12p, we performed an ELISA using recombinant Pf12p D1D2 and Pf12p D2 283 proteins. Our recombinant Pf12p D1D2 protein contains both predicted 6-cys domains and 284 lacks the N-terminal signal sequence and predicted C-terminal GPI-anchor ( Figure 2A). Pf12p 285 D2 contains the C-terminal domain D2 only ( Figure 2A). Unfortunately, we were unable to 286 express the single domain D1 of Pf12p. Nanobodies B2, C4, C12, F7, and H7, bound to both 287 Pf12p D1D2 and Pf12p D2 proteins with similar binding signals showing that these five 288 nanobodies bind epitopes within the D2 domain of Pf12p ( Figure 2B). Nanobodies B9 and G6 289 showed a lower detection signal to Pf12p D2 compared to Pf12p D1D2, suggesting that both 290 the D1 and D2 domains of Pf12p may be involved in nanobody binding. A10, B12, and D9, 291 bound Pf12p D1D2 but their signal for Pf12p D2 was weaker or similar to that of the negative 292 controls ( Figure 2B). The binding sites of these three nanobodies may lay within the D1 293 domain, but a contribution of D2 cannot be excluded. 294 To determine if the Pf12p-specific nanobodies recognize similar epitopes, we 295 performed a nanobody competition experiment using BLI. As expected, all nanobodies were 296 able to compete with themselves ( Figure 2C) We determined the crystal structure of Pf12p D1D2 at a resolution of 2.8 Å by 307 molecular replacement. Two molecules are present in the asymmetric unit of our crystal 308 structure, which are nearly identical and align with a root mean square deviation (RMSD) of 309 0.4 Å. Evaluation of the interfaces using PISA [50] indicates that the protein is monomeric, 310 which is consistent with the elution profile from size exclusion chromatography ( Figure 4C). 311 In the following structural description, we will focus solely on molecule A. Our crystal 312 structure reveals that Pf12p D1D2 folds into two domains, D1 and D2, each containing six 313 cysteines ( Figure 3A). The N-terminal D1 domain adopts the fold of a typical 6-cys domain of 314 type A, which forms a β-sandwich with a 4-on-4 β-strand arrangement. The two sheets of the 315 β-sandwich consist of mixed parallel and anti-parallel β-strands and are pinned together by two 316 disulfide bonds, formed between C27 and C62, and between C76 and C144. A third disulfide 317 bond between residues C93 and C142 connects a loop to the core structure. In our structure, 318 the cysteines form C1-C2, C3-C6 and C4-C5 pairings, which is characteristic for a typical 6-319 cys domain. 320 The C-terminal D2 domain of Pf12p folds into a 6-cys domain of type B forming a β-321 sandwich with a 5-on-4 β-strand arrangement of mixed parallel and anti-parallel β-strands 322 ( Figure 3A).  Figure S3). In the case of Pf12, a surface of 461 Å 2 is buried between its two domains, 911 Å 2 338 for Pf41 and 689 Å 2 for Pf12p. The two 6-cys domains are tilted against each other in a similar 339 manner, but the relative rotation between D1 and D2 differs in the three structures ( Figure 3C). Pf41. Using BLI we show that Pf41 is able to bind to Pf12 with an equilibrium dissociation 355 constant of KD = 143.7 ± 22.6 nM ( Figure 4A). However, for Pf12p, even at the highest 356 concentration of 500 nM we were unable to detect any binding to Pf41 ( Figure 4B). Using size 357 exclusion chromatography, we observed complex formation between Pf12 and Pf41 as a higher 358 molecular weight species ( Figure 4C). In comparison, there was no indication of a complex 359 forming between Pf12p and Pf41 ( Figure 4D) Pf12p-D9 complexes which were determined to 2.0 Å and 3.25 Å, respectively ( Figure 5A, B). 370 The Pf12p-B9 structure shows that B9 forms contacts with residues on both Pf12p domains 371 ( Figure 5A, C, Table 2). All three CDR loops of B9 are involved in binding Pf12p with an 372 interaction surface of 910 Å 2 ( Figure 5C). The Pf12p-D9 structure also reveals that nanobody 373 D9 forms contacts with residues on both Pf12p domains ( Figure 5B, D,  However, the precise function of many members remains unknown and structural information 397 is not available for the majority of these surface antigens. In this study, we report the first 398 crystal structure of P. falciparum protein Pf12p with its two 6-cys domains. We also 399 characterize a collection of anti-Pf12p nanobodies for their specificity, affinities and their 400 epitope bins. Furthermore, we describe two crystal structures of Pf12p bound to distinct 401 nanobodies, both of which show that nanobodies are able to bind to regions spanning two 402 separate 6-cys domains. 403 Immunisation of Pf12p in alpacas and subsequent selection of nanobodies using phage 404 display resulted in the identification of ten distinct clonal groups of nanobodies against Pf12p. 405 These ten nanobodies were specific against Pf12p and did not show cross-reactivity towards 406 either Pf12 or Pf41, which are the closest structural homologues to Pf12p. The nanobodies have 407 affinities ranging from ~3 to 105 nM for binding to Pf12p and their CDR3 regions vary in 408 length between 8 to 21 amino acids. Using BLI, we determined that the antibodies belong to 409 two different epitope bins. One group of five nanobodies bind within domain D2 of Pf12p, but 410 we were unable to obtain crystal structures of this set of nanobodies for detailed epitope 411 determination. In the second group of five nanobodies, binding to Pf12p is partially or 412 completely abrogated in the absence of the D1 domain and we were able to determine the 413 structure of two Pf12p-nanobody complexes of this group of antibodies.   The distance measurements are based on molecules B, D (Pf12p-B9) and molecules A, B (Pf12p-D9).
Interactions and interfacing residues between Pf12p and nanobodies were determined using PISA. Pf12 + Pf41