Affinity Is an Important Determinant of the Anti-Trypanosome Activity of Nanobodies

Background The discovery of Nanobodies (Nbs) with a direct toxic activity against African trypanosomes is a recent advancement towards a new strategy against these extracellular parasites. The anti-trypanosomal activity relies on perturbing the highly active recycling of the Variant-specific Surface Glycoprotein (VSG) that occurs in the parasite's flagellar pocket. Methodology/Principal Findings Here we expand the existing panel of Nbs with anti-Trypanosoma brucei potential and identify four categories based on their epitope specificity. We modified the binding properties of previously identified Nanobodies Nb_An05 and Nb_An33 by site-directed mutagenesis in the paratope and found this to strongly affect trypanotoxicity despite retention of antigen-targeting properties. Affinity measurements for all identified anti-trypanosomal Nbs reveal a strong correlation between trypanotoxicity and affinity (KD), suggesting that it is a crucial determinant for this activity. Half maximal effective (50%) affinity of 57 nM was calculated from the non-linear dose-response curves. In line with these observations, Nb humanizing mutations only preserved the trypanotoxic activity if the KD remained unaffected. Conclusions/Significance This study reveals that the binding properties of Nanobodies need to be compatible with achieving an occupancy of >95% saturation of the parasite surface VSG in order to exert an anti-trypanosomal activity. As such, Nb-based approaches directed against the VSG target would require binding to an accessible, conserved epitope with high affinity.


Introduction
African trypanosomes are extracellular protozoan parasites that are responsible for human sleeping sickness and trypanosomiasis in livestock. Studies of Trypanosoma brucei infections in mice have shown that these parasites cope with the host adaptive immune system through suppression of T and B cell responses [1][2][3][4] and antigenic variation of Variant-specific Surface Glycoprotein (VSG), the most abundant GPI-anchored membrane protein [5,6]. Due to stochastic genetic alterations, T. brucei changes its antigenic VSG-coat by which antibody responses mounted by the host are not protective throughout infection [7,8]. Moreover, the molecular organization of 5610 6 -10 7 VSG copies per cell as densely packed dimers [9], results in shielding of invariant surface epitopes. Recently, some trypanosome-specific Nanobodies (Nbs) were shown to exert an in vitro and in vivo activity against the T. brucei AnTat1.1 variant antigen type (VAT), independent of the complement activation pathway. Nbs are antigen-binding VHH fragments of approximately 15 kDa derived from Heavy-chain Antibodies that are present in Camelidae [10] and can be selected through phage-display technology and panning [11,12]. Experimental data suggested that the toxicity of the Nanobody depends on interference with the highly active VSG-recycling mechanism. In normal circumstances, the bloodstream form T. brucei is able to turn-over its total surface-exposed VSG pool within 12 minutes through uptake in clathrin-coated vesicles and exocytosis in the flagellar pocket area [13]. This VSG recycling would allow antibody shedding and protection against destructive host immune responses by internalization of immune complexes [14,15]. Illustrating that Nanobodies interfere with this cell trafficking mechanism, Nanobody-mediated trypanotoxicity is associated with disturbances in endo/exocytosis and a pronounced expansion of the flagellar pocket [16]. Despite these phenotypic characteristics of toxicity, the exact mode-of-action of the available AnTat1.1 VSG-specific Nanobodies remains to be elucidated. Nevertheless, it is clear that an interference with the early onset of endocytosis through the flagellar pocket is involved.
Here we report on the identification of a larger panel of T. brucei AnTat1.1 VSG-specific trypanotoxic Nanobodies from a new phagemid library and document that high affinity (nanomolar range) for VSG is a common characteristic of all identified trypanotoxic Nanobodies. In addition, we show that affinity reduction by randomization of the amino acids within the antigenbinding loops of the Nanobody or by mutation of framework residues strongly reduces the trypanotoxic potential despite retention of antigen-binding capacity. Site-directed mutagenesis to introduce humanizing mutations in framework-2 [thus replacing residues that are different between VHHs and human VHs, [17,18]] is shown to preserve trypanotoxic activity only if the affinity is not drastically reduced. We illustrate that there is a strong correlation between trypanotoxic potential and affinity (K D ) constants. We further discuss the implications of these observations for Nanobody-based anti-trypanosomal interventions.

Ethics statement
The experiments, maintenance and care of mice complied with the guidelines of the European Convention for the Protection of Vertebrate Animals used for Experimental and other Scientific Purposes (CETS nu 123). Parasites were expanded in in-house bred six to eight weeks old C57B1/6 mice. Mouse care and experimental procedures were performed under approval from the Animal Ethical Committee of the Vrije Universiteit Brussel (Ref. Nr. 08-220-8).

Generation of an alpaca sVSG-immune library
Four days after the last injection, blood was collected for the isolation of peripheral blood lymphocytes and preparation of cDNA. VHH gene fragments ranging from codons from framework region 1 to framework region 4 were amplified by nested PCR as described elsewhere [21,22]. Amplicons were unidirectionally ligated using the NcoI and BstEII restriction sites of the in-house phagemid vector pMES, derived from pHEN4 [23]. The ligated material was subsequently transformed in E. coli TG1 cells and stored at 280uC in LB-medium supplemented with 50% glycerol.
Generation of a randomized library based on Nanobody 33 (Nb_An33) Residues in complementarity determining regions (CDR) 1 and 3 of the previously identified Nb_An33 [24], and putatively involved in the recognition of T. brucei sVSG, were identified based on the Nanobody crystallographic structure [18]. A Nb_An33 library with randomized codons in the paratope, at five amino acid positions either in CDR1 (Ser 29 , Thr 30 , Tyr 31 , Ser 32 and Thr 35 ) or CDR3 (Val 103 , Arg 104 , Ser 105 , Ile 106 and Arg 107 ) has been generated by PCR using degenerate primers. The codons for the residues of interest were replaced by NNK (K = G or T). Ligation of the amplicons into the pMES phagemid was performed using PstI and BstEII restriction sites. Ligated material from the randomization in CDR1 and CDR3 was pooled and used for transformation as described for the alpaca sVSG immune library (see above).

Selection of sVSG-specific Nanobodies through panning
The sVSG immune library and Nb_An33 randomized library were expressed on phages after super-infection with M13K07 helper phages. Libraries were enriched by several consecutive rounds of in vitro selection on solid-phase coated T. b. brucei AnTat1.1 sVSG using a standard procedure [22,24,25]. Phages were eluted under alkaline conditions (pH 11.0), followed by an immediate pH neutralization and amplification in E. coli TG1 cells [22]. Enrichment of the library for antigen-specificity was assessed using an AnTat1.1 sVSG-specific phage ELISA. Here, phages that are specific for AnTat1.1 sVSG were detected using a horse-radish peroxidase-anti-M13 conjugate (Amersham Biosciences). As soon as enrichment was observed in the phage ELISA, individual colonies of transfected TG1 cells were picked to evaluate the expression of antigen-specific Nanobodies upon induction with 1 mM isopropyl-b-d-thiogalactopyranoside (IPTG). Periplasmic extracts were tested for antigen recognition in ELISA using a peroxidase-conjugated mouse anti-66His IgG (Serotec).

Expression and purification of Nanobodies
T. brucei specific clones were used to purify plasmid (pMES with the VHH gene of interest). The plasmid constructs were transformed into chemocompetent E. coli WK6 (su 2 ) cells. Expression and Nanobody purification was similar as described earlier [25]. Nanobodies were eluted from Ni-NTA columns using 0.5 M imidazole/PBS and were subsequently subjected to gelfiltration on a Superdex 75 10/300 GL column (GE Healtcare) with phosphate buffered saline (PBS) as running buffer.

Surface plasmon resonance
Affinities of newly identified Nanobodies or Nb_An33 mutants were measured by surface plasmon resonance on a BIAcore T100 system. Between 1000 and 1500 RU of soluble AnTat1.1 VSG was coupled onto a CM5 chip (BIAcore) via amine groups according to the manufacturer's descriptions using EDC and NHS as cross-linking agents and ethanolamine to block free esters. For the affinity determination, Nanobody concentrations ranging from 500 to 7.8 nM were added to the antigen-coated chip at a flow-

Author Summary
Nanobodies, antigen binding fragments derived from a non-conventional class of antibodies in camelids, were previously shown to exert a direct activity against African trypanosomes without the need of a toxin. Their mode-ofaction relies on interference with the highly active recycling of the Variant-specific Surface Glycoprotein (VSG) that occurs in the flagellar pocket of the parasite. By expanding the panel of anti-trypanosomal Nanobodies and by modification of their binding properties through site-directed mutagenesis, we have been able to show a strong correlation between their trypanotoxic activity and affinity for the cognate antigen. From these studies it was calculated that the parasite surface saturation needs to exceed 95% in order to achieve this anti-trypanosomal effect of Nanobodies, which can be considered as a critical cut-off value for future Nanobody-based or other small molecule drug approaches against the VSG target.
rate of 30 ml/min in HBS buffer [10 mM Hepes (pH 7.5), 150 mM NaCl, 3.5 mM EDTA, 0.005% (v/v) Tween- 20)]. The flow cells were regenerated with 10 mM glycine-HCl (pH 1.5). Sensograms were fitted for a 1:1 binding model with local R max using the BIA-evaluation software version 4.1 (BIAcore), resulting in association and dissociation constants (k a and k d ) as output from which affinity (K D ) values were calculated.
Epitope mapping was performed on the same VSG sensor chip by evaluating cumulative or competitive binding onto the antigen with 10-fold higher Nanobody concentrations than the K D using all possible two-by-two Nanobody combinations.

Flow cytometry and fluorescence microscopic analysis
Binding of the different Nanobody clones was evaluated on live, bloodstream form T. b. brucei AnTat1.1 through flow cytometry following a direct or three-step labelling procedure. The direct labeling required conjugation of Nanobodies with Alexa Fluor 488 according to the manufacturer's instructions (Molecular Probes). The three-step labeling procedure relied on the detection of the surface-bound Nanobodies with a mouse anti-66His IgG and a phycoerythrin-labelled rat anti-mouse IgG. Flow cytometry analyses were performed on a FACS Canto II and histograms were prepared using the FlowJo software (Becton Dickinson, San Jose, CA). Stained trypanosomes were subsequently fixed for 30 minutes using 3% paraformaldehyde and washed five times with 0.1 M HEPES pH 7.4 in order to facilitate analysis by fluorescence microscopy as this procedure stops cell motility and VSG recycling by endocytosis. Cells were stained with DAPI by the use of the Fluoro Gel II mounting medium (Electron Microscopy Sciences).

Trypanotoxicity experiments
Trypanotoxic activity of Nanobodies was evaluated on ex vivo isolated and DEAE52-purified trypanosomes. Parasite cultures were initiated at 10 5 /ml concentrations in 200 ml HMI-9 medium supplemented with 10% decomplemented fetal bovine serum (FBS). Nanobodies were added to the parasites at concentrations of respectively 10, 20, 40 and 60 mg/ml followed by incubation at 37uC in a conditioned atmosphere with 5% CO 2 . Toxicicty was quantified by parasite counting using a Bürker hematocytometer after 18 h incubation. Percentage toxicity was calculated from at least two individual wells and relative to the condition with a nontrypanosome specific Nanobody. An additional buffer control (PBS) was included in the lysis experiments.

Data and statistical analysis
Correlation between the trypanotoxic potential of the different anti-trypanosome Nanobodies and their affinity constants k a , k d and K D was analyzed by the non-parametric Spearman correlation analysis using the GraphPad Prism 5.02. Reported data are the Spearman correlation coefficients (r) that vary between +1 and 21 depending on a positive or negative correlation between the variables. Dose response curves were obtained in the same software package by a three parameter dose response non-linear regression with a top/bottom constraint of respectively 100% and 0%. As output of this regression analysis, half maximal effective K D , k a and k d values (affinity and kinetic rate constants resulting in 50% trypanotoxicity) are reported as well as the 95% Confidence Intervals (CI). Based on the Michaelis-Menten equation assuming a single binding site and non-cooperative binding, the VSG binding site occupancy (B/B max ) at steady state for the determined half maximal effective K D values of the Nbs can be determined from the equation:
Based on information from the Nb_An33 crystal structure, residues within the Nb_An33 paratope (Ser 29 , Thr 30 , Tyr 31 , Ser 32 and Thr 35 in CDR1 or Val 103 , Arg 104 , Ser 105 , Ile 106 and Arg 107 in CDR3) were randomized by site-directed mutagenesis, combined with the generation of a phagemid library to allow selection of functional AnTat1.1 sVSG-specific Nb_An33 variants through phage display and panning. Pannings, combined with expression and sequence analysis of individual clones resulted in the selection of three Nb_An33 variants with retained antigen-binding properties. One variant was mutated in CDR1 (Nb_An33mut1: Ser29Arg/Thr30Ala/Tyr31Leu), two variants had substitutions in CDR3 (Nb_An33mut2: Val103Phe/Arg104Val/Ser105Asn/ Arg107Asn and Nb_An33mut3: Val103Asn/Arg104Asp/ Ser105Cys/Arg107Leu) ( Figure S2). Flow cytometry analysis revealed proper binding of all Nb_An33 derived variants to living T. brucei (Figure 2A). Comparative epitope mapping of Nb_An33 and the 3 mutants in SPR revealed that the identified variants bind to the same epitope or at least epitopes in a common region as no significant cumulative binding was observed onto the VSG-sensor chip for any Nanobody combination (data not shown). Affinity measurements by surface plasmon resonance revealed significant reduction in affinity of the 3 mutants, i.e. 89, 723 and 197 nM as compared to the wildtype Nb_An33 (K D = 28 nM) ( Figure 2B, Table 1). Also on-and off-rate constants were negatively affected to different extends through mutagenesis, reducing the k a from 7.41610 5 M 21 s 21 to 4.67610 5 , 3.14610 5 and 3.36610 5 M 21 s 21 and increasing k d from 0.021 s 21 to respectively 0.042, 0.227 and 0.066 s 21 ( Figure 2B).

Correlation between affinity and trypanotoxic potential of Nanobodies
As a result of the identification of various AnTat1.1 sVSGspecific Nanobodies and the selection of Nb_An05 variants [16] and Nb_An33 variants, a set of Nanobodies (n = 25) with different trypanotoxic potential (0-100%) and affinity-range (2 nM-2.2 mM) became available. Within this set, four specificity groups could be identified by analyzing competitive binding of all Nanobody-combinations onto the CM5/VSG chip ( Table 2). This indicates that each of these Nbs bind to one out of four different sites (epitopes) on the VSG. Plotting of trypanotoxic potential against the affinity and kinetic rate constants k a , k d and K D , revealed a very good non-linear correlation of trypanotoxic potential with the K D (R 2 = 0.91), determined both by the kinetic rate association constant k a (R 2 = 0.48) and kinetic rate dissociation constant k d (R 2 = 0.47) (Figure 3). Spearman correlation analysis provided evidence for a strong correlation (p,0.0001) between the trypanotoxic activity and the k a (Spearman correlation coefficient r = 0.71), k d (r = 20.74) and especially K D (r = 20.89). Half maximal effective (50%) affinity, k a and k d were calculated from the non-linear dose-response curves ( Figure 3

Discussion
The discovery of VHHs or Nanobodies with direct in vitro and in vivo trypanotoxic activity is a recent advancement in the development of new drug discovery approaches against African trypanosomiasis. The observed phenotype of lysis suggests that the VSG-specific Nbs interfere with the VSG recycling mechanism that -under normal conditions -is able to turn-over the total surface-exposed VSG pool within 12 minutes by endo/exocytosis in the flagellar pocket [13]. The previously identified Nanobodies, Nb_An05 and Nb_An46, were shown to lyse efficiently T. b. brucei parasites independent of the complement action, associated with a fast immobilization of the parasite and a pronounced swelling of the flagellar pocket [16]. Beside the physical stress of the enlarged flagellar pocket, an endocytosis/exocytosis malfunction might   hamper the active and passive uptake of nutrients (as evidenced by using labeled transferrin, dextran and 2-deoxy-D-glucose) and result in cell death. Similar morphological phenotypes of lysis have been reported to occur upon RNAi mediated silencing of genes involved in clathrin-mediated endocytosis [26][27][28][29]. This indicates that Nanobodies target the endocytosis/exocytosis pathway but the exact mode-of-action has yet to be elucidated. It has been shown previously that the small size of the antigenbinding fragment is crucial, supported by the generation of trypanotoxic Fab9-fragments and the abrogation of trypanotoxic activity by addition of an Fc part to the VHH [16]. Suggesting that affinity is another important parameter, all previously and newly identified trypanotoxic Nanobodies have K D -values in the nanomolar range, a recurrent feature of Nanobodies that are selected by phage display from an immune VHH library [12,25]. The association between affinity and trypanotoxicity was further supported by mutational analyses that were performed on Nb_An05, a highly trypanotoxic Nanobody [16] and Nb_An33, a trypanosome-specific Nanobody from which the structure has been resolved [18] and recognizes a cryptic and glycosylated (Man 9-5 GlcNAc 2 ) epitope on T. brucei sVSG [24]. Using the highly trypanotoxic Nb_An05 as paradigm, we have illustrated previously that modification of the binding affinities by sitedirected mutagenesis in the CDR regions resulted in strongly reduced trypanotoxic potential despite conservation of the capacity to bind to the trypanosome surface [16]. Here, we illustrate that the Nb_An33 exerts a limited trypanotoxicity over longer incubation times than those typically used for Nb_An05 and Nb_An46 [16]. This might be explained by the relatively lower affinity of Nb_An33 (28 nM) as compared to Nb_An05 (2 nM) and Nb_An46 (18 nM), although epitope-differences might also contribute to this difference. Variants in the Nb_An33 paratope (selected residues in the complementarity determining region 1 and 3) and in the framework (selected residues in framework-2 that differ between VHHs and human VHs [17,18]) were generated respectively through a process of randomization by site directed mutagenesis and phage display. The selected variants exhibited lower affinities for their cognate antigen and displayed a significantly reduced or abrogated trypanotoxic activity. In the context of Nb_An33 humanization that was previously performed on framework region-2 [18], mutations that did not affect affinity, e.g. Glu44Gly/Arg45Leu, preserved the trypanotoxic activity as well. Moreover, Tyr37, a residue that is crucial for both conformational stability and antigen affinity [18], can not be substituted by the corresponding VH-residue valine without loss-of-activity as was observed for Nb_An33hum1 and Nb_An33hum2. Re-camelizing position 37 in Nb_An33hum2 (K D = 540 nM) from valine to tyrosine in Nb_An33hum3 was  sufficient to restore its affinity for VSG (K D = 23.5 nM) as well as its trypanotoxic activity. Incorporation of the affinity and trypanotoxicity data obtained from all available Nanobodies [16,18,24] confirmed a strong nonlinear correlation between trypanotoxic activity and K D (R 2 = 0.91, Spearman correlation coefficient r = 20.89). This indicates that trypanotoxic activity requires a fast association (r = 0.71) and a slow dissociation (r = 20.74) of the bound Nanobody from the trypanosome surface. Non-linear doseresponse analyses revealed apparent 50% activity at a K D of 57 nM, a k a of 3.8610 5 M 21 s 21 and a k d of 1.0610 22 s 21 . We have calculated that the 57 nM affinity (K D ) yielding a 50% trypanotoxic activity under our experimental conditions would result in an occupancy of 95% saturation of all the VSG on the parasite surface coat by the Nb. As such, these data suggest that small molecule drug approaches against VSG would require surmounting a cut-off of 95% saturation of the antigen. The functional importance of affinity has previously also been documented for tumor cell lysis by complement activation upon binding of conventional antibodies to their antigen at the cell surface [30,31]. In this case, the complement-dependent cytotoxicity was most efficient for monoclonal antibodies with slow offrates and could be enhanced by reducing antibody dissociation using a F(ab9) 2 cross-linking agent [30]. An analogous approach of generating multivalent constructs to decrease dissociation of Nanobodies from the trypanosome surface is unlikely to be successful given the cryptic nature of conserved epitopes buried in the dense variable antigenic VSG (variant-specific surface glycoprotein) coat [32]. Indeed, we have previously shown that bivalent Nb9 2 are significantly less active than the monomeric forms [16]. In fact, the VSG-specificity of the toxic Nanobodies might impose a limitation on the drug discovery given that trypanosomes constantly alter VSG-epitopes in order to evade the host immune response [7,8]. As such, there is the need to develop a Nanobody that can bind to an accessible, conserved epitope with sufficiently high affinity to achieve the required surface saturation.
As an alternative and future perspective, the procyclic developmental stage of the T. brucei parasite that occurs in the tsetse fly midgut displays a much smaller antigenic repertoire of glycosylphosphatidylinositol (GPI) anchored membrane proteins. Surface-binding lectins targeting N-glycans on procyclin were shown to be capable of inducing cell death in procyclic trypanosomes [33,34]. Given the lower antigenic complexity of the procyclic antigenic coat and the previous illustration of an antitrypanosomal activity upon binding to procyclin, this could represent an ideal target for Nbs. Therefore, we aim at generating Nbs against procyclin with kinetic parameters below the cut-off values as determined for the anti-blood stream form Nbs. Recently, a delivery system for functional Nanobodies to the tsetse fly based on its endosymbiont Sodalis has been developed [35], offering perspectives for anti-trypanosome Nbs to interfere with the trypanosome life cycle inside the insect vector.