SCIENTIFIC REVIEW 13Keeping onscheduleReshaping assaysEvaluating hitsJoseph P. Gardner, ScientistPfizer Inc., USAThe contributions from Biacore®technology continue to have awidening impact on all aspects ofthe drug discovery process intoday’s pharmaceutical industry. As biosensor technology became apractical reality in the laboratoryduring the early 1990s, it wasreadily accepted by the scientificcommunity as a powerful analyticaltechnique for evaluating the kineticparameters of molecularinteractions, such as protein-proteinand protein-nucleic acid. Wherehistorically such measurementswere made indirectly or bychemically modifying one or all ofthe interactants, thus introducingthe potential for altering the naturalkinetic characteristics of theinteractants, a technique is nowavailable that can measure theseparameters less intrusively. And,where drug design/discovery relieson understanding structure-functionrelationships between biomolecules,results from Biacore systemscomplement structural studies withdetailed kinetic data utilizingrelatively fast methodologies andconsuming small amounts ofprecious reagents.Stay on track in drug discoveryGrowing impact of Biacore on drug discovery“...the (Biacore) system was a key contributor to keeping the project on schedule”SCIENTIFIC REVIEW 13Pfizer purchased its first Biacore system to profilethe kinetic characteristics of fusion proteinsgenerated in several of its expression labs. The application of the technology in otherfacets of the drug discovery process, however,quickly became apparent.During the early 1990s, our lab was chargedwith the development of peptide-proteincompetition assays for three differenttherapeutic targets. The format would be thesame for the three assays, the difference beingthe peptide and protein reagents. An ELISAformat was chosen to facilitate transfer to ahigh through-put screen (HTS). The protocolutilized biotinylated target peptides, boundvia steptavidin to the assay wells to capturetheir respective GST fusion proteins whichcould then be detected with an anti-GST/HRPantibody conjugate and substrate developmentstep. The GST fusion proteins were partiallypurified preparations from cell lysates.Compounds would be added to the assaywells during the peptide/fusion proteinbinding step and those compounds thatinterfered with binding would be identified bythe absence of HRP/substrate development(Figure 1). These inhibitory compoundswould be further characterized in secondaryassays.Initial assay results were frustrated by highbackground levels and replicate variabilitybetween positive signals. This variabilityappeared to be very sensitive to the washsteps. Frustration led us to recreate the assayswithin the Biacore system. The protocolinvolved capturing the biotinylated peptidesvia immobilized steptavidin to different flowcells on the surface of a Sensor Chip CM5.The GST fusion proteins were passed over thesurface and allowed to bind to their respectivetarget peptides. We observed that the affinitiesof the fusion proteins for their target peptideswere in the >200nM range. With such lowaffinities, it was suspected that this interactioncould be very sensitive to the wash steps andmight explain the observed variabilitybetween replicates.Evaluation of the data led to new peptide andprotein constructs being prepared with theintent of increasing the affinity of the interaction.The larger target peptides contained phospho-tyrosines and were difficult to synthesize. Synthesis of different batches of thephosphotyrosines peptides were not alwaysconsistent. There were solubility andphosphorylation issues to address. Simplestoichiometric calculations indicated higherpeptide bound to streptavidin, suggestingaggregation problems with peptides.Phosphorylation, unwanted dephospho-rylation, or lack of phosphorylation duringsynthesis was easily assessed by monitoringthe binding of an anti-phosphotyrosineantibody to the target peptides whenimmobilized on the sensor surface.After revising our peptide synthesis protocols,the issues of phosphorylation and solubilitywere corrected. However, we noted the loss ofbinding activity of fusion protein to the targetpeptide following only a few (~5)regenerations of the sensor surface. There didnot appear to be any change in the immob-ilization level, but there was a decrease in thebinding of anti-phosphotyrosine antibody. Infact, during experiments where the targetpeptides were incubated with partiallypurified fusion protein preparations, massspectrophotemetry showed that phosphataseactivity in the protein preparation resulted incomplete dephosphorylation of the targetpeptide within 30 minutes. This problem waseasily corrected by employing more stringentpurification conditions for the fusion proteinsand adding sodium orthovanadate to the finalprotein stocks.The fusion protein constructs were alsochanged to incorporate additional bindingsites for the target peptides. In the case of allthe fusion proteins, stoichiometriccalculations from binding data obtained fromthe Biacore system indicated that the materialbound as a dimer. GST fusion proteins areknown to dimerize and this was easilyconfirmed by chromatography.Binding activity of the purified fusion proteinswas now investigated. For a given proteinaliquot stored at 4°C for several days, the lossFigure 1ELISA format for the detection of a GST fusion protein binding to atarget peptide.SCIENTIFIC REVIEW 13of binding activity was found to be due toglutathionylation of the protein by the activeGST tag which was utilizing free glutathionepresent in the protein buffer followingpurification. Loss of binding activity was notdue to dephosphorylation of the target peptide(as assessed using anti-phosphotyrosine on theBiacore system) or degradation of the fusionprotein (as determined by chromatography).This was verified by reverse phase-HPLCfollowed by electrospray-mass spectroscopy(ES-MS). The proteins were being multiplymodified with glutathione resulting in a loss ofability to bind their target peptide. Treatmentwith DTT reversed this modification andrestored binding ability of the protein.Finally, with new reagents in hand, ELISAassays were attempted again. Two of theassays appeared to work well, however, thethird assay demonstrated high levels of non-specific binding. We had some data on thisparticular fusion protein that led us to suspectthat it could be aggregating under the assayconditions. All assay incubations had been at37°C whereas Biacore