Case report
3
published epitopes targeted in parallel
85%
hit rate, 39 confirmed binders in ELISA
35 & 40 nM
in a first BLI read for 2 VHHs
Contact Us Talk to our Experts
Project type
Internal validation study, ProteoGenix AI business unit
Target
GPVI (GP6), the platelet collagen receptor
Formats
VHH (single-domain) and VHVL
Field
Haemostasis and thrombosis
Epitopes targeted
3 designed and tested in parallel
Key process
GPVI is the main signalling receptor for collagen on platelets, and a target where antibody engineering has to answer a question that goes beyond binding: which site on the receptor do you aim at?
Different epitopes give different mechanisms, from direct occupation of the collagen groove to allosteric effects that leave collagen binding intact.
That makes GPVI a useful test of an epitope-targeted design platform. Rather than run one campaign and see what came out, ProteoGenix selected three published sites on the receptor and designed against all three in parallel, in the same run.
The study asked whether the platform can hold a defined epitope as a design constraint across three different site types at once, conformational and linear, and whether the resulting antibodies bind specifically.
01
Designing against three different sites in parallel, including two linear epitopes and two conformational ones
02
No reference antibody structure available in house for this antigen, unlike the PD-L1 campaign
03
Working across two formats, VHH and VHVL, in the same design run
04
Separating specific binding from background in a first ELISA screen of 46 candidates
01
Selection of three published epitopes
Three sites were chosen for their documented role in GPVI function. The collagen and CRP binding pocket on the D1 domain, defined by Val34, Leu36, Arg38, Glu40, Lys59, Arg67, Gln71, Trp76, Arg117 and Arg166. The D1 and D2 site identified through the trowaglerix venom peptides, described as two linear epitopes, QGPPGVDLYRLEK and KEGDPAPYKNPER, where blockade is non-competitive and collagen still binds. And the site adjacent to the CRP groove that was crystallised with an inhibitory nanobody in PDB 7NMU, covering Glu21, Ser45, Arg46, Tyr47, Gln48, Pro56, Ala57 and Ser61, where inhibition comes from steric hindrance combined with an allosteric change at the active site.
02
AI sequence generation and filtering
AIxplore® generated sequences against each of the three epitopes. Candidates were filtered on more than 40 parameters, including developability and manufacturability, and every output was reviewed by our bioinformaticians before selection.
03
Recombinant production across two formats
46 antibodies were expressed and purified, 35 in VHH format and 11 in VHVL format, then quality controlled before testing.
04
ELISA specificity screening
Each candidate was tested against GPVI, with a control to measure non-specific binding. The screen was designed to confirm specificity and to give a first read on the binding profile before committing candidates to affinity measurement.
05
Affinity determination by BLI
Biolayer interferometry on the best candidates is underway. It will confirm and quantify the affinities that the ELISA profiles suggest.
Of the 46 antibodies designed and produced, 39 bound GPVI in ELISA, a hit rate of 85%. Five of the 39 were VHVL, which puts the VHH set at 34 binders out of 35 designs and the VHVL set at 5 out of 11.
The ELISA results show specific binding to GPVI across the large majority of the candidates, with a low level of non-specific signal. Five candidates stand out, combining an excellent signal to noise ratio with very low background. Several antibodies reach saturation at low concentrations, which points to good affinity. The few non-specific signals observed stay limited, with the exception of one scFv showing high background.
BLI measurements are still running. At this stage two VHH give a preliminary reading around 35 nM and 40 nM. The other candidates tested so far do not appear specific in this assay. Exact values and the final ranking will come with the completed BLI dataset.
Three published epitopes on GPVI were targeted in the same design run, with no immunisation and no library. The campaign produced 46 antibodies across two formats, of which 39 bound the antigen in ELISA, with specific binding and limited background across most of the set.
For a discovery program, that means the epitope can be treated as a design decision rather than a screening outcome: several mechanisms can be pursued in parallel on the same target, and the results compared side by side before committing to one.
