Letter pubs.acs.org/ac
Drug Discovery at the Single Molecule Level: Inhibition-in-Solution Assay of Membrane-Reconstituted β‑Secretase Using SingleMolecule Imaging Anders Gunnarsson,*,† Arjan Snijder,† Jennifer Hicks,† Jenny Gunnarsson,† Fredrik Höök,‡ and Stefan Geschwindner*,† †
Discovery Sciences, AstraZeneca R&D Mölndal, SE-43183 Mölndal, Sweden Department of Applied Physics, Chalmers University of Technology, SE-412 96 Göteborg, Sweden
‡
S Supporting Information *
ABSTRACT: Inhibition-in-solution assays (ISA) employing surface-based biosensors such as surface plasmon resonance (SPR) are an effective screening approach in drug discovery. However, analysis of potent binders remains a significant hurdle due to limited sensitivity and accompanied depletion of the inhibiting compounds due to high protein concentrations needed for detectable binding signals. To overcome this limitation, we explored a microscopy-based single-molecule ISA compatible with liposome-reconstituted membrane proteins. Using a set of validated small molecule inhibitors against β-secretase 1 (BACE1), the assay was benchmarked with respect to sensitivity and dynamic range against SPR. We demonstrate that the dynamic range of measurable affinities is greatly extended by more than 2 orders of magnitude as compared to SPR, thus facilitating measurements of highly potent (Kd < nM) compounds.
M
and function of many membrane-bound targets are known to be modulated by the neighboring lipid composition.6−8 Taken together, to resolve these limitations, biophysical assays compatible with native, full-length membrane proteins embedded in a controlled lipid environment are needed. It is relevant to note that, in this context, surface-based technologies such as surface plasmon resonance (SPR) biosensor9 (e.g., BIAcore) or waveguide-based platforms10 can be used to circumvent aforementioned issues with the DBA by running so-called inhibition-in-solution assays (ISA), in which case the influence of drug candidates on the binding interaction between surface-immobilized target definition compounds (TDCs) and suspended receptors are monitored. In this way, the actual target−inhibitor interaction occurs in solution, which means that interfacial effects obscuring conventional analysis are largely reduced. While this assay configuration helps to exclude assay artifacts and enables independent determinations of the affinity (Kd) of the compound for its target, analysis of potent binders remains a challenge. This is due to sensitivity limitations since the lower limit of determinable Kd values becomes determined by the lowest concentration of protein required to achieve a detectable signal in the assay, thus defining the tight binding limit. To overcome the sensitivity limitation of conventional ISAs, we here employ an imaging methodology11−13 that enables
embrane proteins play a key role in regulating cellular activity and function. This is manifested by the fact that more than 60% of approved drugs target membrane-bound proteins1 and that they remain the dominant target class in new drug discovery projects.2 It is therefore a major concern that detailed biophysical studies of membrane protein targets are hampered by severe limitations, such as low expression levels and poor stability. Furthermore, standard biophysical/biochemical assays are often not applicable to this class of proteins owing to their hydrophobic nature, while only few technologies are compatible with having the proteins embedded in their natural lipid environment. For example, direct binding assays (DBA), using biosensor technology that directly measure interactions between surface-immobilized membrane receptors and suspended ligands, are often obstructed by protein instability, detergent interference, or low receptor abundancy, which prevents sensitive binding signals from being obtained.3 Common approaches to circumvent these shortcomings include the use of engineered and truncated versions of the target protein excluding the transmembrane domain. The thereby achieved compatibility with assays used to drive drug development campaigns comes with the associated risk of drastic changes in both the structure and function of the protein.4 As a consequence, removal of the transmembrane domain may significantly affect the outcome of screening activities and hit validation. This is particularly important for allosteric binders that bind in close proximity to the transmembrane domain and may be disregarded as nonbinders using such ectodomain-based assays.5 In addition, the activity © 2015 American Chemical Society
Received: February 24, 2015 Accepted: April 9, 2015 Published: April 9, 2015 4100
DOI: 10.1021/acs.analchem.5b00740 Anal. Chem. 2015, 87, 4100−4103
Letter
Analytical Chemistry
been shown that the proteolytic activity of BACE1 is modulated by the lipid environment surrounding the protein.8 Taken together, this strongly motivates our development of a high-sensitivity assay compatible with membrane-embedded BACE1. Full-length BACE1 (flBACE) was expressed in Sf21 and Sf9 cells using the baculovirus expression vector system26 (see Supporting Information for details). The protein was solubilized in Triton-X100 and purified using dual affinity purification (6× His-tag and immobilized peptide) to high purity, as confirmed by a single band on the SDS-PAGE gel at 56 kDa (Figure S1A, Supporting Information). The relatively broad band (confirmed by anti-His Western blot) indicates varying degrees of post-translational modifications. To assess the single-molecule methodology, which operates at orders of magnitude lower protein concentration (pM to fM) than conventional surface-based sensors, flBACE was reconstituted in liposomes (Figure S1B, Supporting Information) and a functional biosensor surface, based on the self-assembly of coblock polymers on SiO2, was prepared (Figure 1 and Supporting Information). The polymer-modified surface effectively suppress unspecific binding of liposomes while presenting the TDC (substrate analog inhibitor17) for subsequent protein binding and competition experiments. Having established the surface chemistry (Figure S2, Supporting Information), dose−response curves of a set of BACE1 inhibitors spanning a wide affinity regime were generated (Figure 2C) via linear fits to time-resolved recordings of individual liposome binding events (at ∼100 fM protein/ liposome concentration) at different compound concentrations (exemplified in Figure 2A) in a well format. Using liposomes as carriers for single proteins, the number of enzymes per liposome is difficult to determine exactly. However, at the liposome/protein ratio of 4:1 used here, only a few percent of the liposomes will have >1 enzyme. Hence, this aspect, which nevertheless translates into minor uncertainties in the affinity determination,14 was not taken into account in the analysis. Figure 2C shows that the dynamic range of measurable compound affinities with this novel single molecule ISA stretches from mM to beyond nM. In order to create a comparative data set with SPR technology, the same substrate analogue was used to design an ISA that could be employed to assess the rate of protein binding in the presence of the same set of inhibitors in a dose− response relationship using ecBACE (exemplified in Figure 2B). Note that detergent-solubilized flBACE could not be assessed in a similar manner using SPR due to nonspecific interactions, presumably via the exposed hydrophobic domain (data not shown). Linear fits to the initial rate of protein binding upon injection were converted into dose−response curves. The resulting affinities from the fitted curves in TIRF (Figure 2C) and SPR ISA (Figure 2D) are summarized in Table 1 and show good correlation for the low affinity binders. However, using the surface-based SPR ISA, the 3 most potent compounds all display similar “apparent” affinities (Figure 2D) at ∼70 nM (i.e., half the protein concentration used), reflecting the tight binding limit of the SPR assay. In contrast, using the single-molecule assay the potent compounds could be readily discriminated on the basis of their different affinities (Figure 2C). Hence, the increased sensitivity of the single molecule assay enables a robust readout at significantly lower protein concentration, thus increasing the dynamic window down to sub-nM affinities. Note that the protein concentration require-
interaction studies of membrane-embedded proteins at the level of single molecules.14 The concept, which is schematically illustrated in Figure 1, is based on time-resolved imaging using
Figure 1. Schematic illustration of the methodology. Fluorescent liposomes containing, on average,
