SakmarLab

The Sakmar Laboratory

The Laboratory of
Molecular Biology & Biochemistry
at Rockefeller University

NABBs
UAAM
HTS
FTIR
SMD
Computational Methods

Dr. Sakmar and his colleagues believe that new technologies and capabilities will make significant interdisciplinary work possible to advance the understanding of how heptahelical receptors really work. One of the aims of the SakmarLab is to assemble a toolbox that will allow us to study heptahelical receptors in the normal environment of the cell membranes. The SakmarLab now has several converging technology platforms in place to study heptahelical receptor chemistry and biology

•All atom and coarse grain MD (molecular dynamics) simulations of GPCRs in membrane bilayers,
•UAAM (unnatural amino acid mutagenesis) of GPCRs using amber codon suppression technology,
•NABBs (nanoscale apolipoprotein bound bilayers) as membrane mimic support structures for GPCRs,
•SMD (single-molecule detection) of GPCRs in self-assembling oriented-tethered bilayers using TIRF (total internal reflection fluorescence) microscopy or in NABBs using microfluidics,
•HTS (high-throughput screening) or high content functional assays to characterize expressed mutant GPCRs.

These new tools in combination with a variety of analytical methods to evaluate specific receptor structural heterogeneities should provide detailed information about how receptors actually function in the cell membrane

The Sakmar Laboratory

The Laboratory of
Molecular Biology & Biochemistry
at Rockefeller University

NABBs: Nanoscale Apolipoprotein-bound Bilayers

Summary.
The Sakmar Laboratory recently developed a nanoparticle technology termed NABBs (nanoscale lipoprotein-bound bilayers), to study heptahelical G protein-coupled receptors (GPCRs) in a controlled artificial membrane environment. NABBs comprise a phospholipid bilayer disc, approximately 10 nm in diameter, surrounded by a belt of 2 interlocking apolipoprotein A1 molecules. NABBs are soluble and stable for long periods of time in solution. Members of the Sakmar Laboratory devised a rapid method to incorporate membrane proteins, including GPCRs, into NABB particles. NABBs that contain GPCRs can then be purified by affinity or size-exclusion chromatography and characterized by a variety of analytical methods, including electron microscopy and fluorescence imaging spectroscopy.

NABB Composition.
To develop a useful technology to study membrane proteins, members of the Sakmar Laboratory cloned and engineered zebrafish apolipoprotein-A1 (zap-1), which they reasoned would self-assemble rapidly in the presence of phospholipids to provide a particularly stable and homogeneous membrane nanoparticle. A typical NABB comprises a phospholipid bilayer disc, approximately 10 nm in diameter, formed by self-assembly of a belt of 2 interlocking recombinant expressed zap-1 proteins surrounding about 100 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC) lipids. NABBs self-assemble rapidly in vitro under controlled conditions and can be purified and characterized. NABB particles remain in solution for long periods of time and can be prepared in large quantities and stored efficiently. NABBs provide an effective native-like membrane environment for cell-surface receptors, including GPCRs, by maintaining crucial receptor-phospholipid interactions. GPCRs are much more stable in NABBs than in detergents.

Potential Impact of the NABB Technology.
One main advantage of NABBs is that they can be adapted to serve as a membrane-mimetic structure to study integral membrane proteins in a native-like bilayer environment. Each component of the NABB structure, including the phospholipids, zap-1 and encased GPCR, can be labeled independently. The labels can be used to quantitate NABB components or to interrogate the structure and function of the isolated GPCR. Once a GPCR is trapped inside a NABB particle, both surfaces of the receptor, intracellular and extracelluar, are accessible to the aqueous environment simultaneously. This feature provides a great advantage to study the process of signal transduction across a bilayer, which occurs when a ligand binds on one side of the receptor to trigger a conformational change on the other side of the receptor.
Members of the Sakmar Laboratory incorporated the visual pigment rhodopsin into NABB particles and showed that it functioned normally. They also showed that a single isolated rhodopsin molecule could activate G proteins in response to light and that a receptor dimer was no more efficient than a monomer in G protein activation. Recombinant expressed rhodopsin as well as expressed chemokine receptor CCR5 and the glucagon receptor, a prototypical family B GPCR, have also been incorporated into NABBs.
For example, in recent preliminary work, members of the Sakmar Laboratory have prepared and characterized NABBs that contain functional recombinant human CCR5. To detect correctly folded CCR5 in NABBs, a novel sandwich ELISA assay protocol in a 96-well plate format was devised where the CCR5-NABBs were sequentially sandwiched between two antibodies recognizing specific, distinct epitopes on the CCR5-NABB assembly. In one formulation of the assay, the capture antibody, 2D7, is a monoclonal antibody (mAb) that is known to recognize a split epitope on the extracellular side of CCR5 only when they are in proximity to each other, which would be the case in a correctly folded form of CCR5. The secondary mAb, 1D4, recognizes an engineered linear amino-acid sequence at the C-terminal tail of CCR5. The NABB technology, when combined with the ELISA assay, can be exploited to screen for conditions that preserve heterologously expressed CCR5 in its native conformation after detergent extraction and reconstitution.

At least 20 laboratories around the world are now using the NABB technology to study a variety of membrane proteins, including components of the nuclear pore complex and various channels proteins.

The Sakmar Laboratory

The Laboratory of
Molecular Biology & Biochemistry
at Rockefeller University

UAAM: Unnatural Amino Acid Mutagenesis

Summary.
The Sakmar laboratory has pioneered novel approaches to introduce non-perturbing chemical or spectroscopic labels into low abundance cell surface receptors such as heptahelical G protein-coupled receptor (GPCRs). Recently, in an interdisciplinary collaboration with Prof. U. L. RajBhandary at M.I.T., the Sakmar Laboratory developed an amber codon suppression strategy suitable for unnatural amino acid (UAA) mutagenesis of proteins expressed in mammalian cells. Members of the Sakmar Laboratory engineered suitable suppressor tRNAs and orthogonal amino acyl-tRNA synthetases that suppress the termination of protein synthesis at an amber codon mutation in a gene of interest and introduce a UAA. The UAA mutagenesis method was used to introduce p-azido-L-phenylalanine (azF), p-acetyl-L-phenylalanine (acF) or p-benzoyl-L-phenylalanine (bzF) into rhodopsin or CCR5 at high yield.

How Does UAAM Work?
In site-directed UAAM, engineered amino acyl-tRNA synthetase constructs are designed to bind specific UAAs and transfer them to engineered suppressor tRNAs. Together the orthogonal pairs of engineered aminoacyl-tRNA synthetase and tRNA cause the incorporation, at the site of an amber codon mutation, of a particular UAA added to the culture medium. In approximately two decades since the first demonstration of site-directed UAAM, various amino acid analogues with novel chemical or physical, properties have been incorporated into proteins expressed in E. coli or using in vitro protein translation systems. However, the methods developed for use in E. coli do not work in mammalian cells. To adapt UAAM approach for use in mammalian cells, the Sakmar Laboratory employed a cell-based luciferase screen to select for efficient amber codon suppression by testing a variety of engineered suppressor tRNA constructs and engineered amino acyl-tRNA synthetases. The mammalian expression of a particular engineered Bacillus stearathermophilus tyrosine-tRNA was optimized and paired with an engineered E. coli tyrosine amino-acyl tRNA synthetase.

As a proof-of-concept, the Sakmar Laboratory incorporated three tyrosine analogues at high efficiency in two different GPCRs – rhodopsin (the photoreceptor for the dim-light vision) and CCR5 (a chemokine receptor that has been identified to be the co-receptor for the HIV viral entry into T-cells). The strategy was to target specific sites on the receptor while preserving its function. The expression yields of functional receptors were evaluated by spectroscopic quantification of rhodopsin or calcium flux functional assay of CCR5. The results demonstrated the general utility of the novel UAA mutagenesis technology. High efficiency UAA mutagenesis of membrane receptors in mammalian cells can be used to create reactive tags tailored for the site-specific post-translational labeling application.

Potential Impact of the UAAM Technology.
Members of the Sakmar Laboratory have optimized and validated orthogonal tRNA-amino acyl tRNA synthetase pairs to incorporate p-azido-L-phenylalanine (azF), p-acetyl-L-phenylalanine (acF) or p-benzoyl-L-phenylalanine (bzF) into heterologously-expressed proteins in mammalian cells. These particular UAAs were chosen because they contain functional groups – keto for acF and bzF, and azido (N3) for azF – not normally present on cell proteins. These functional groups, therefore, provide a specific reactive handle to label proteins with fluorophores or other informative probes. In addition, the benzyol and azido groups are photoreactive and provide the potential to carry out photochemical cross-linking experiments. Finally, the azido group has a unique infrared vibrational frequency useful for Fourier-transform infrared spectroscopy.
To label proteins post-translationally, members of the Sakmar Laboratory are using the Staudinger-Bertozzi ligation reaction and the [3+2] cyclo-addition, “click” chemistry, reaction. Preliminary results are encouraging even under physiological reactions conditions, which do not denature or disrupt the normal structure or function of the target protein. Using the UAAM procedure and post-translational labeling, possibly in combination with other methods, it should be possible to introduce multiple labels on one protein or different labels on different proteins. Thus, donor-acceptor fluorescence resonance energy transfer (FRET) pairs can be introduced to monitor intra- or intermolecular dynamic processes.

The UAAM strategy devised by the Sakmar Laboratory provides a general method to introduce specific labels at specific sites into proteins expressed in mammalian cells.

The Sakmar Laboratory

The Laboratory of
Molecular Biology & Biochemistry
at Rockefeller University

HTS: High Throughput Screening

Summary.
High-throughput screening (HTS) originally referred to an automated strategy to identify specific small-molecule drug candidates among a large library of different small molecules. HTS assays typically employ a micro-titer plate format (e.g., 96- or 384-well plates), robotic sample dispensing and automatic fluorescence/luminescence read-out. In G protein-coupled receptor (GPCR) research, HTS can be used to find ligands or drug candidates that bind to a specific receptor target. HTS methodologies have evolved over time to include cell-based assays in which binding reactions are detected indirectly by monitoring downstream signal outputs, such as increases in intracellular calcium or second messenger concentrations. The Sakmar Laboratory employs a variety of HTS platforms and novel screening technologies to study GPCR signaling, mainly focusing on chemokine receptors.

Recent Developments.
The Sakmar Laboratory has a proven track record of HTS assay development and is a major user of the Rockefeller University HTS Resource Center (HTSRC). In one established assay of GPCR activation, fluorescent calcium-sensing dyes are used to measure changes in intracellular calcium concentration as a function of time. The Sakmar Laboratory was one of the first laboratories to use this method to study the pharmacology of chemokine receptors.
Another assay called the SEAP (secreted alkaline phosphatase) assay measures intracellular cAMP levels, which are regulated by inhibitory G proteins linked to chemokine receptors. One recent innovation involves the development of a useful amber codon suppression system to carry out unnatural amino acid mutagenesis (UAAM) of GPCRs heterologously expressed in mammalian cells in culture. The UAAM method facilitates the introduction of fluorescent probes into expressed engineered receptors. The Sakmar Laboratory is also developing high-throughput and low volume assays for detecting GPCR activation using lanthanide-based resonance energy transfer (L-RET) principles.

Chemokine Receptors.
In addition to these HTS technology applications, the Sakmar Laboratory has several ongoing research projects that require the use of the HTS methods. For example, the Sakmar Laboratory is involved in an on-going research collaboration with the International Aids Vaccine Initiative (IAVI) to develop Rhesus monkey CCR5 expressing stable cell lines, which will be used as tools in the development of an AIDS vaccine. The development of stable cell lines requires careful functional characterization of the expressed receptor using HTS methods.

The role of chemokine receptor heterodimerization can be addressed using fluorescence complementation assays as well time-resolved FRET. These assays can be adapted to screen and identify drug candidates that specifically target a particular heterodimer receptor complex. The Sakmar Laboratory has identified several cellular proteins that appear to act as modulators of heterotrimeric G protein signaling pathways downstream from receptor activation itself. In order to study the role of non-canonical signal transduction pathways downstream of GPCR activation, novel assays for protein-protein interactions using FRET and polarized fluorescence assays are under development

The Sakmar Laboratory

The Laboratory of
Molecular Biology & Biochemistry
at Rockefeller University

FTIR: Fourier-Transform Infrared Spectroscopy

Summary.
Fourier-transform infrared (FTIR) spectroscopy can be used to measure the vibrational spectrum of a complex biological macromolecule. The Sakmar Laboratory and its collaborators, most notably Prof. Friedrich Siebert, Albert-Ludwigs-Universität, Freiburg, Germany, developed an approach to study recombinant engineered visual pigments using FTIR-difference spectroscopy. Over a period of nearly a decade, using a combination of site-directed mutagenesis, biochemical characterization and FTIR-difference spectroscopy, members of the Sakmar Laboratory were able to assign specific IR vibrational difference bands to specific amino acid side chains in rhodopsin and to show how the bands shifted upon receptor activation. In ongoing studies of rhodopsin structure and activity, the same approach is being combined with UAAM in collaboration with Prof. Reiner Vogel, Albert-Ludwigs-Universität. In a recent systems biology application of FTIR spectroscopy, Dr. Thomas Huber invented an FTIR apparatus to create and measure concentration gradients of volatile odorants in air and simultaneously record the chemotactic behavior of genetically altered fruit fly larvae.

Structural Studies of Rhodopsin.
FTIR-difference spectroscopy is a powerful method to study the structural changes that occur in rhodopsin, a prototypical G protein-coupled receptor (GPCR), as it undergoes characteristic conformational changes after photoisomerization of it 11-cis-retinal chromophore. In the mid-1990s, the Sakmar Laboratory pioneered the use of FTIR spectroscopy studies of engineered recombinant rhodopsin expressed in mammalian cells in culture. Specific vibrational bands, most notably the carbonyl stretching frequencies of protonated carboxylic acid residues, were assigned to specific amino acid side chains in the receptor. The power of FTIR-difference spectroscopy is that structural changes can be followed over time using thermal trapping methods to freeze specific protein conformations while spectra are measured. The interpretation of ongoing FTIR data has been facilitated by recent high-resolution crystal structures of rhodopsin and other related GPCRs. FTIR studies of rhodopsin and rhodopsin mutants have contributed to validating and understanding the “helix movement model of receptor activation” and the concept of “functional microdomains” in the super family of GPCRs.

Recent studies in the Sakmar Laboratory combine FTIR-difference spectroscopy with UAA mutagenesis, in which the useful IR probe azido-phenylalanine (azF) is introduced at specific sites in rhodopsin. The antisymmetric stretching vibration of azido results in a strong absorption band at around 2,100 wavenumbers that is distinct from other vibrational absorption bands typically present in proteins. The strength and position of the azido band is sensitive to subtle changes of the electric field of its local environment. In proof-of-concept studies, azF was introduced at several sites in rhodopsin. The rhodopsin mutants were then studied with FTIR-difference spectroscopy and the results were interpreted in the context of recent structural studies with the aid of computer molecular dynamics modeling.

Olfactory Chemotactic Behavior.
Chemotaxis refers to directed movement toward attractive chemicals and away from aversive chemicals. The process of chemotaxis requires continuous computation and integration of local odorant gradients and an appropriate response. How animals detect and integrate olfactory stimuli in three-dimensional space remains a mystery. Part of the problem in studying chemotactic behavior is that it is difficult to generate, control and measure odorant gradient in space. The Sakmar Laboratory recently developed a new behavioral assay where the quantity and distribution of an olfactory stimulus can be controlled and precisely quantified in space. Using the new assay, which relies on FTIR spectroscopy, we delineated the behavior and neural mechanisms underlying chemosensory orientation in genetically modified fruit fly larvae. This latest work was carried out in collaboration with Prof. Leslie Vosshall, Rockefeller University and Prof. Mathieu Louis, Centre for Genomic Regulation, Barcelona, Spain.

The Sakmar Laboratory

The Laboratory of
Molecular Biology & Biochemistry
at Rockefeller University

SMD: Single-Molecule Detection Fluroescence Spectroscopy

Summary.
The Sakmar Laboratory is committed to developing new approaches to probe GPCR dynamics and intermolecular interactions, such as the binding of a specific ligand to a specific receptor, in real time. Typical ensemble assays of receptor-ligand interactions rely on decades-old formulations of receptor theory and assume equilibrium or pseudo-equilibrium conditions. In studies of surface phenomena that occur on cell membranes, it would be preferable to measure individual binding events and then to use statistical analysis to derive kinetic rate constants and thermodynamic parameters where possible. The Sakmar Laboratory has implemented an arsenal of tools to enable single-molecule detection (SMD) fluorescence experiments of GPCRs reconstituted in solid-supported bilayer membranes with defined lipid composition. For example, members of the Sakmar Laboratory have established a method for site-specific introduction of unnatural amino acids (see section on UAA mutagenesis) in GPCRs and other proteins expressed in mammalian cells in culture. They have also demonstrated bio-orthogonal chemical modification of these UAAs in the prototypical GPCR, rhodopsin. One particularly useful method for SMD studies of GPCRs is called TIRF (total internal reflection fluorescence) microscopy. The Sakmar Laboratory recently assembled a TIRF microscopy workstation capable of SMD.

SMD Using TIRF.
In TIRF microscopy, a light source is focused on the surface of a thin glass slide at a critical angle of incidence that creates an evanescent field that emanates a short distance (typically about 200 microns) above the opposite surface of the glass. A fluorescent probe with the correct photochemical properties that lies within the evanescent field becomes excited and emits a photon, which can be detected by the optical collection system of a suitable epifluorescence microscope. The TIRF method can be used for SMD experiments, Förster resonance energy transfer (FRET) applications, or imaging. A brief description of TIRF microscopy can be found by clicking here. Dr. Thomas Huber designed a novel strategy suitable for SMD-TIRF experiments of GPCRs. In preliminary experiments, expressed recombinant epitope-tagged chemokine receptors were immuno-captured in a self-assembling, oriented, tethered membrane bilayer on the surface of a TIRF chip. The custom chip surfaces can be prepared by sequential perfusion of different protein solutions in a simple semi-microfluidics chamber. Dr. Huber recorded SMD images of single fluorescently labeled chemokines (MIP-1α-Alexa-647) bound to recombinant expressed CCR5 chemokine receptor embedded in the tethered bilayer membrane. Each fluorescence spot represents the fluorescent chemokines that bind to CCR5. The membrane-receptor assembly is stable for hours and can be interrogated repeatedly using a continuous flow cell chamber. For example, the MIP-1α-Alexa-647 can be washed out of the chip flow cell and a different concentration of MIP-1α-Alexa-647, or a different labeled chemokine, can be added in turn. Competition experiments with small molecules or with chemokine mixtures can be carried out as well.

Novel SMD-TIRF-FRET Methods.
SMD-TIRF-FRET (fluorescence resonance energy transfer) experiments can be carried out using the general approach and microscope already in place, provided that an appropriate fluorescent donor-acceptor pair can be adapted. In FRET, excitation of the donor fluorophores will cause emission from the acceptor fluorophores provided that they are situated together within their characteristic Förster distance (~2-8nm distance range). To study receptor-ligand interactions, or protein-protein interactions, if two molecules of interest, one harboring the donor and the other the acceptor, form a complex that allows the fluorophores to come together within their Förster distance, then FRET will occur. If no FRET signal is observed, it suggests that the target molecules have no direct interaction. The key to making clear-cut interpretations of FRET signals relies on establishing a relevant positive control. In one particular formulation under development, a GPCR is labeled with Alexa-494 at a Cys residue through maleimide chemistry. The Sakmar Laboratory is also developing Europium(III) dye conjugates to label other signaling components. Europium(III) dyes exhibit superb photochemical properties, which will allow fluorescent imaging at the ultimate SMD level.

The Sakmar Laboratory

The Laboratory of
Molecular Biology & Biochemistry
at Rockefeller University

Computational Methods

Summary.
The long-term goal of the Sakmar Laboratory is to provide a comprehensive understanding of ligand recognition in GPCRs. Since the determination of the first high-resolution structure of a GPCR, the visual photoreceptor rhodopsin, the Sakmar Laboratory has used computer homology modeling and molecular dynamics (MD) simulations to build receptor models in native-like phospholipid bilayers. These all-atom models typically contain more than 50,000 atoms and are simulated on a sub-microsecond timescale using massive parallel processing (MPP) architectures on the TeraGrid, such as the PSC BigBen Cray XT3. When combined with biochemical and biophysical experiments to map the thermodynamics of ligand binding in rhodopsin, this computational approach provides significant insight toward understanding ligand binding and recognition in GPCRs. Recently, the Sakmar Laboratory collaborated with Dr. Xavier Periole and Prof. Siewert-Jan Marrink, University of Groningen, The Netherlands, to develop coarse-grain molecular dynamics methods to model the behavior of multiple GPCRs in bilayers, including the spontaneous self-assembly of receptor dimers and oligomers.

Homology Modeling and All Atom Molecular Dynamics Simulations.
GPCRs constitute a superfamily of heptahelical receptors that respond to a variety of external stimuli such as amines, peptides, enzymes, hormones, odors, and light. Rhodopsin is the defining member of the largest family group, the rhodopsin-like class A GPCRs, with 569 possible candidates identified in the human genome. Class A GPCRs comprise the opsins and protein, peptide, and cationic amine receptor subfamilies, constituting ~90% of all known GPCRs. The sequences of related GPCRs can be used to build homology models based on rhodopsin and other GPCRs for which high-resolution structural data are available, rendering in silico studies of class A GPCRs feasible.
The Sakmar Laboratory recently performed MD simulations of the β2-adrenergic receptor and compared the results to what had been done earlier with rhodopsin. In order to investigate the binding mode of catecholamines in the context of the new structures, an MD simulation was devised to place the receptor bound to carazolol, an inverse agonist ligand, in a typical bilayer membrane environment. Next, to address the question of whether or not adrenaline, the receptor agonist, would be stable in fundamentally the same receptor conformations used to bind carazolol additional simulations were performed. The nanosecond timescale of these MD simulations is certainly not sufficient to capture the slow transition to the active receptor conformation. However, a principal component analysis of the movements of the backbone Cα atoms demonstrated a global change of the receptor structure in its transition from the carazolol to the adrenaline bound form. The change is consistent with the elongated structure of the antagonist pushing the extracellular ends of transmembrane helices 2 and 6 outward into the bilayer.

Molecular Dynamics Simulations of Ligand Binding Pathways.
Many GPCRs have hydrophobic ligands that are likely to partition into the membrane bilayer. In extended MD simulations of rhodopsin in the membrane, spontaneous structural fluctuations were observed leading to a temporary opening of the protein molecular surface to provide access to the ionone ring moiety of the 11-cis-retinal ligand. In contrast, the homologous region of a homology model of the chicken red cone pigment, iodopsin, which is known for its very fast retinal binding kinetics, revealed a continuous opening with direct view on the ionone ring of the ligand.
Careful analysis of MD models of rhodopsin in the membrane suggested that this transient pore in the membranous region of the receptor between transmembrane helices 5 and 6 might represent the primary ligand entry site. To test this hypothesis, Dr. Huber performed preliminary multi-nanosecond free energy calculations of ligand binding in a membrane model of rhodopsin based on reversible MD simulations using umbrella potentials. The weighted histogram analysis method (WHAM) was employed to eliminate the effects of the biasing umbrella potentials and to recover potential of mean force (PMF) free energy profiles. The results were compared with irreversible steering MD simulations applying the Jarzynski nonequilibrium equality, and with (multi-configurational) thermodynamic integration.
The ligand binding and un-binding simulations supported the notion that the transient pore structure might act as a channel for the ligand to enter the receptor active site. Site-directed mutagenesis experiments confirmed that substitutions of amino acid side-chains in the putative binding pocket affected the kinetics of ligand binding and release.

Coarse-Grain Molecular Dynamics (CGMD) Simulations of Complex Systems.
Coarse-grain molecular dynamics (CGMD) simulations hold considerable promise to elucidate the behavior of membrane receptors in native-like membrane environments. By reducing the degree of complexity of the system from atoms to chemical groups formed by 3–6 non-hydrogen atoms, this technique allows the required increase in system size and time scale as compared with a full atomistic approach. CGMD has been validated to assure that the physicochemical properties of the model system are conserved. The Sakmar Laboratory recently extended the CGMD approach to address the question of how the physicochemical properties of the membrane lipids might affect self-assembly of GPCRs into higher-order structures. First, simulations of rhodopsin as a monomer were used to characterize the response of both the protein and the bilayer to the presence of hydrophobic mismatch. Next, multi-copy rhodopsin simulations (16 proteins per unit cell) at a protein-to-lipid ratio 1:100 in the same bilayer environments were carried out. Rhodopsin progressively self-assembled into aggregates and ordered linear arrays. Localized adaptation of the membrane bilayer to the protein and surface complementarity were key factors that defined the rate, extent, and orientational preference of protein–protein association.