Three llamas, Whisper, Centavo, and Cowboy, were immunized with the West Nile Innovator + VEWT, an equine vaccine that includes inactivated VEEV. Plasma from each of the animals was evaluated to determine the ability to neutralize VEEV strain TC-83 as well as EEEV, WEEV, CHIKV, and West Nile virus (WNV). Results are reported in Table 1. Centavo, who had previously been immunized with a course of the West Nile Innovator + VEWT, had the strongest neutralization response of the three llamas; therefore, we chose to focus our effort on panning the phage-displayed sdAb library derived from this animal for the identification of VEEV-binding sdAb. It is unclear why the three animals had such different responses to this commercial vaccine, however Centavo likely benefited from previous inoculation. Interestingly Whisper, who showed a very high titer for West Nile virus, had been immunized with the commercial PreveNile and Recombitek equine vaccines for prevention of West Nile virus about the same time Centavo had been given his prior immunizations with West Nile Innovator + VEWT. We have observed that some animals generate more robust immune responses than others, and libraries prepared from animals possessing weak titers generally perform poorly.
We panned the Centavo library using gamma-irradiation inactivated VEEV strain TC-83. Starting with approximately the same number of phage input for each of three rounds, we obtained increasingly greater numbers of output phage; with the second round output ~ 6 times greater than the first round, and the third round ~ 60 times higher than round 1. Although not the most robust enrichment, we proceeded to monoclonal phage ELISA for rounds 2 and 3. We examined 96 clones from each round and found 22 positives from round 2 and 32 from round 3. Sequencing of 40 positive clones identified five sequence families, three with at least 7 members, and two with only one member each. Interestingly, one of the sequence families (typified by V11A1 and V2C3, Fig. 1) was also isolated from a separate library derived from Centavo after immunization with chikungunya virus like particles about one year prior to the current immunization. Through that effort we had identified clone CC3, a sdAb that strongly neutralized CHIKV, and also neutralized VEEV17,33. While the same sequence family as CC3 was obtained from the new library, a sequence identical to CC3 was not (Supplemental Figure S1).
Amino acid sequences of representative sdAb selected for ability to bind irradiated VEEV-TC-83. Sequences are given in 1-letter amino acid code. Alignment was performed using Multalin25. Red indicates high homology, while lower homology is in blue. CDR regions are defined using the IMGT definitions27.
Eight sdAb, two representatives from each of the multi-member families as well as the two single-member sequence families, were transferred from the phage display vector to the pET22b expression vector for protein production (Fig. 1). In general, good protein yields were obtained from these clones as reported in Table 2. On examination of the sequences, we observed that V3A8, the poorest producing sdAb, had a non-conserved and unpaired cysteine in framework 2. In an effort to achieve better protein production, we performed site-directed mutagenesis to revert the cysteine to the conserved arginine to produce clone V3A8f. Unfortunately, this clone produced essentially identically to the original V3A8. Undeterred, a codon optimized version of V3A8f was synthesized which had a 3 to fourfold increase in protein production yielding 10.4 mg/L.
The melting temperature and refolding ability of each sdAb was measured by circular dichroism and is reported in Table 2. Seven out of eight clones had melting temperatures between 62 and 67 °C with one clone, V2G1, melting at 73 °C. Likewise, most clones regained over 50% of their secondary structure after heat denaturation with the exception of the V8C3 and V3G9 sequence family which refolded poorly. Methods including the addition of negative charge and site-directed mutagenesis have been demonstrated to improve refolding ability and to raise melting temperature by as much as 10–20 °C, so the melting and refolding properties of these sdAb could potentially be improved by protein engineering, if desired11.
The eight clones were assessed by PRNT to determine their ability to neutralize VEEV strain TC-83. Several clones were also tested to determine if they were able to neutralize the virulent Trinidad Donkey (TrD) strain of VEEV. Results shown in Table 3 and Fig. 2 indicate that all families showed at least some ability to neutralize VEEV. In particular clones V3A8 and V2B3 showed the best neutralization, while V2G1 and V8C3 showed the weakest. Results from clones V2C3 and V11A1 were consistent with the previously isolated CC3 which we had measured a PRNT50 of 4.0 µg/mL and 1.9 µg/mL on the TC-83 vaccine strain and parental TrD strain respectively33.
PRNT of representatives from each of the five anti- VEEV-TC-83 families. Data is shown from a representative set of neutralization experiments performed in duplicate. Error bars represent the standard error.
Multiple studies have shown the advantage in using multivalent sdAb constructs for viral neutralization. Researchers have done this through genetically linked sdAb19, using tag-catcher systems34, and by fusions with an Fc region35. We chose to construct homo- and hetero-bivalent forms of the VEEV-binding sdAb by genetically linking sdAb through a flexible glycine-serine linker as detailed in the methods. Linked constructs were based on sdAb from four sequence families: V3A8f, V8C3, V2B3, and V2C3. To denote the linked constructs, we list the first sdAb followed by a hyphen and then the second sdAb. Protein yields of the bivalent constructs ranged from 2 to 9 mg/L.
To compare the ability of standard and bivalent constructs to neutralize VEEV-TC-83, we examined the neutralization ability of the bivalent sdAb versus mixtures of the standard sdAb. We also evaluated constructs where the two sdAb were linked in both orders. Data is shown in Table 4 and Fig. 3. We found that in each case an improvement was realized with the linked sdAb versus the mixture. The most potent were V3A8f-V2B3, and V2C3-V3A8f. Based on PRNT50, the V2C3-V3A8f construct showed almost 290-fold boost over the mixed sdAb. However, even the fusion of V3A8f with V8C3 showed ~ ninefold improvement over the mixture of standard constructs despite the fact that V8C3 by itself was a poor neutralizer. We did not see dramatic differences upon changing the order of the sdAb within the multimeric constructs, and chose to only study one orientation of the V8C3 and V3A8f mix.
Representative PRNT data sets showing standard versus mixture versus linked sdAb. V2B3 and V3A8f are shown in green and black respectively. The mix of V2B3 and V3A8f is shown in red, and the linked V2B3-V3A8f is in blue. Each measurement was performed in duplicate, error bars represent the standard error.
The bivalent sdAb were also examined for their ability to neutralize the wild type TrD strain of VEEV. Results were consistent with the neutralization of VEEV-TC-83 and are reported in Table 5.
Since we isolated sdAb, such as clone V2C3, which is in the same family as previously identified CHIKV-neutralizing sdAb CC3, we also tested a subset of standard and bivalent constructs for their ability to neutralize CHIKV. The PRNT50 and PRNT80 values are shown in Table 6. The V2C3 showed comparable neutralization of CHIKV as CC3; when V2C3 was expressed as a homobivalent construct or a heterobivalent construct with CC3, neutralization improved by ~ 5–sevenfold over the standard sdAb. We examined the heterobivalent construct of V2C3 with V3A8f as we hypothesized it has the potential to effectively neutralize both VEEV and CHIKV. Indeed the construct neutralizes both viruses with PRNT50 values ~ 2 ng/mL. However, while the linked V3A8f and V2C3 was much more effective at neutralizing VEEV than the mix of the standard sdAb constructs, it was not more effective than the mix at neutralizing CHIKV. When the ability of a mixture of V3A8f and V2B3 to neutralize CHIKV was compared to the bivalent version, the linked construct was no better, and potentially worse than the mixed sdAb clones.
Selected standard and/or bivalent sdAb constructs were evaluated by PRNT to determine any cross reactivity with eastern equine encephalitis virus (EEEV) and western equine encephalitis virus (WEEV), two other new world alphaviruses. Preliminary studies showed that of the constructs tested, only CC3, V2C3-V2C3, and CC3-V2C3 neutralized EEEV with a PRNT50 of 25 or lower µg/mL (Supplemental Table S1). The bivalent constructs containing V3A8f weakly neutralized WEEV, but none neutralized greater than 30–45% at 50 µg/mL; the other linked constructs essentially did not neutralize WEEV at all. These results indicate that none of the sdAb constructs show appreciable cross-reactivity for these additional viruses. The E1 and E2 glycoproteins are the major antigenic proteins for which antibodies are produced36. Viruses with greater sequence homology in these glycoproteins have a higher likelihood of cross neutralization by antibodies that recognize conserved epitopes. A sequence comparison of the E1 glycoproteins across the alphaviruses found VEEV to be more closely related to EEEV (~ 60% sequence homology) than WEEV (~ 45% sequence homology)37. Based on the greater sequence homology between VEEV and EEEV, we expected to observe more cross neutralization with EEEV than WEEV.
We successfully met our goal of isolating sdAb that neutralized VEEV. Although it is likely that the sdAb bind to one of the envelope proteins, the exact target of these sdAb has yet to be determined. This is an interesting avenue for future experimentation which could include epitope mapping towards determining the mechanism of action of these sdAb.
The most potent of the bivalent sdAb constructs inhibited VEEV and prevented the infection of cells at low levels, thus offering the potential to be developed as therapeutics for the treatment of VEEV. In addition to providing a therapeutic for a biothreat of great concern, these sdAb may be of interest for treatment of both humans and horses during seasonal outbreaks of VEE. While horses are commonly immunized, there is currently no human vaccine, and should global warming expand the range of the mosquito vector the need for effective treatments could also likely increase. Combining the V3A8f clone with V2C3 produced a reagent with the potential to potently neutralize both VEEV and CHIKV. The next step will be to express the most promising structures in a format to provide increased serum half-life and test their effectiveness in a mouse model.
