Background

Biological Background: Protein-protein interactions (PPIs) are essential for almost all cellular processes, ranging from DNA replication, transcription, and translation over the formation of metabolic enzyme complexes and large cellular machineries, to the sensing and relay of biological signals. Given the physiological importance of PPIs, protein complexes are considered high-value targets in drug development (Milroy et al., 2014; Nevola and Giralt, 2015). However, traditional, small-molecule based targeting approaches often fail to produce potent inhibitors, mostly due to the large and often flat protein interfaces (Sperandio et al., 2010). Alternative strategies involve peptide-based inhibitors that often mimic part of the binding surface of an interaction partner (Boersma et al., 2012; Azzarito et al., 2013; Arkin et al., 2014; Pelay-Gimeno et al., 2015). Such interactions between proteins and peptides of typically 5-20 amino acids are frequently found in nature, for example with antibodies, kinases, phosphatases, MHC proteins, or epigenetic reader proteins (Stanfield and Wilson, 1995). Moreover, the specific targeting of PPIs with small peptides can help to elucidate composition and function of large and sophisticated protein complexes like the eukaryotic RNA polymerase II elongation complex (Antosz et al., 2017) or the specificity of epigenetic reader proteins (Josling et al., 2015).

Technical Background: In order to precisely and comprehensively characterize protein-protein and protein-peptide interactions with respect to affinity, kinetics, and thermodynamics, an array of biophysical techniques and methods have been developed. Well established technologies are isothermal titration calorimetry (ITC) (Pierce et al., 1999; Ghai et al., 2012), surface plasmon resonance (SPR)-based assays (Pattnaik, 2005; Ishii et al., 2013), native mass spectrometry (Heck, 2008; Yan et al., 2017), biolayer interferometry (Abdiche et al., 2008; Concepcion et al., 2009), static or dynamic light scattering (Wen et al., 1996; Kameyama and Minton, 2006; Hanlon et al., 2010), as well as fluorescence spectroscopy (Lundblad et al., 1996; Lin et al., 2014; Raines, 2015). Although some of these methods uniquely provide specific information like on- and off-rates or the precise complex stoichiometry, they are often lab-intense and complex, are often problematic due to high sample consumption, low-sensitivity, surface immobilization, mass transport limitations, as well as buffer restrictions or fail to quantify interactions involving small peptides. The innovative MicroScale Thermophoresis technology (MST) is a powerful technique that overcomes such limitations and allows for the fast, precise, cost-efficient, and quality-controlled characterization of molecular interactions with very low sample consumption, no need of sample immobilization, and a free choice of buffers or bioliquids. MST enables researchers to determine the important parameters of molecular interactions, such as binding affinity (from pico-molar o milli-molar), binding stoichiometry, and interaction thermodynamics (Wienken et al., 2010; Jerabek-Willemsen et al., 2011). Importantly, MST does not impose any limitations regarding molecular weight of the interacting molecules and thus quantifying interactions between proteins of several hundred kDa in size and short peptides with molecular weights below 1 kDa is straightforward and precise (Antosz et al., 2017).

The physical phenomenon of thermophoresis: The physical phenomenon of ‘thermophoresis’ describes the movement of molecules along a temperature gradient. This movement is dependent on the size, charge, and hydration shell of the molecules (Braun and Libchaber, 2002; Duhr and Braun, 2006). Interactions between a protein and a ligand (other protein, peptide, DNA, RNA, small molecule) alter at least one of these parameters resulting in a different thermophoretic mobility along the temperature gradient (Jerabek-Willemsen et al., 2014). Thus interactions between large proteins and small ligands, such as short peptides, which only entail negligible changes in size and charge, will significantly influence the thermophoresis due to changes in the hydration shell.

MicroScale Thermophoresis: MicroScale Thermophoresis is an innovative technology that utilizes the thermophoretic principle to study molecular interactions in solution (Duhr et al., 2004; Baaske et al., 2010; Jerabek-Willemsen et al., 2011). The thermophoretic movement of molecules is induced by generating a microscopic temperature gradient inside a very thin glass capillary using an infrared (IR) laser. The corresponding technical setup of an MST device (Monolith NT.115, NanoTemper Technologies GmbH, Munich, Germany) is shown in Figure 1A. This gradient is focused in a diameter of around 50 µM and comprises a temperature difference ∆T of 2-6 °C. In the same area, the thermophoretic movement of the molecules is tracked by fluorescence optics using either the intrinsic fluorescence of tryptophans in proteins or peptides (Seidel et al., 2012) or the fluorescence signal of an extrinsic fluorophore that is coupled to one of the interacting partners (Schubert et al., 2012; Zillner et al., 2012). Upon heating, molecules either deplete or accumulate in the center of the temperature gradient and the fluorescence optics, which can be quantified by the Soret coefficient S_T, with c_hot and c_cold representing the concentration in the heated and non-heated regions of the capillary and ∆T the temperature difference along the gradient (Duhr and Braun, 2006):



A typical MST measurement is depicted in Figure 1B. In the beginning, the fluorescence in the capillary is measured at the fixed starting temperature for usually five seconds. Then the temperature gradient is induced by the IR laser, which results in a steep drop of the measured fluorescence signal, the so-called temperature- or T-jump. This signal jump results from temperature-dependent changes in the quantum yield of the used fluorophore. After that, a slower, thermophoresis-driven depletion of fluorescent molecules in the optical focus corresponding to the Soret coefficient results in an exponential decrease of the fluorescence signal. After a measurement time of typically 20-30 sec, the IR laser is turned off again, which results in a thermophoretic back-diffusion of fluorescent molecules into the optical focus and a concomitant reverse T-jump.

In order to determine the equilibrium dissociation constant (K_D) of the complex under study, a series of MST measurements with a fixed amount of fluorescent binding partner and increasing amounts of non-fluorescent partner are recorded (Figure 1C). The 16 MST traces are normalized with respect to their initial fluorescence. The concentration range of the non-fluorescent partner is chosen in a way that the lowest concentration results in a practically fully ‘unbound’ fluorescent partner and the highest concentration results in a fully ‘bound’ fluorescent partner. Usually 16 dilutions of the non-fluorescent partner are prepared ranging from about 10-fold above to the 10-fold below the projected K_D, supplemented with the same amount of fluorescent partner and loaded into 16 individual capillaries. The concentration of the fluorescent partner is usually kept below the projected K_D in order to allow for a precise determination of the true equilibrium affinity. The differences between the ‘cold’ and ‘hot’ states of each of the 16 MST traces is then used to determine the change in fluorescence for each trace:



Plotting these values against the ligand concentration finally results in a typical binding isotherm, which yields the K_D value of the interaction (Figure 1D).


Figure 1. MicroScale Thermophoresis. A. Technical setup of an MST device. The thermophoretic movement of molecules inside the glass capillary is induced by an infrared laser that is focused on a 50 µM wide area. In the same area fluorescence emission is used to track this movement. In total, 16 capillaries can be analyzed subsequently in one device. B. An example of MST trace. After an initial delay of five seconds, the IR laser is turned on to establish a temperature gradient. Following the T-jump phase, the thermophoretic movement leads to an exponential decrease of fluorescent molecules in the optics focus. After a measurement time of typically 20-30 sec, the laser is turned off. C. Combined traces of a typical MST experiment in which 16 capillaries are analyzed that contain the same concentration of the fluorescent interaction partner but increasing concentrations of the non-fluorescent partner. All traces are normalized to the same initial relative fluorescence value of 1. D. Final binding isotherm resulting from plotting the difference in normalized fluorescence against the concentration of the non-fluorescent binding partner (ligand). Figure modified from Entzian and Schubert, 2016; originally kindly provided by NanoTemper Technologies, Munich.

Advantages and drawbacks of MST: MST offers several integral advantages over other biophysical methods for characterizing protein-protein and protein-peptide interactions. First, the MST traces and lateral scans of the filled capillaries provide a straight-forward sample quality control as they allow for the easy detection of aggregation, precipitation, and adsorption effects. This enables the researcher to quickly alter and optimize technical and buffer conditions to increase sample stability and data quality. Importantly, these controls take less than two minutes for 16 capillaries. Also the full required measurement to get an equilibrium K_D takes less than 15 min. Moreover, MST is immobilization-free and thus allows for the determination of affinities in practically all buffers and even in complex bioliquids like lysates and sera (Wienken et al., 2010; Seidel et al., 2013). Also, MST excels by very low sample consumption, which can be a crucial issue when working with hard-to-produce eukaryotic proteins. Usually, the fluorescent interaction partner is assayed at very low nanomolar concentrations (5-20 nM) and the non-fluorescent partner usually–depending on the K_D–in the nanomolar to low micromolar range. Finally, the high dynamic range of detectable affinities (pM to mM) and non-existing size limitations allow for the characterization of a wide variety of different molecular interactions involving proteins, peptides, small molecules, DNA, and RNA, just to name a few. However, MST does not allow for determining on and off-rates of molecular interactions. Nevertheless, binding parameters derived from MST experiments agree well with established state-of-the-art methods like SPR or ITC (Ramakrishnan et al., 2012; Chen et al., 2015; Stoltenburg et al., 2015; Wan et al., 2015; Harazi et al., 2017). Finally, it has to be stated that for most MST applications one interaction partner has to be modified with a fluorophore. Thus, if neither a peptide, nor a protein can be labeled with a fluorophore (e.g., because neither contains lysine or cysteine residues for fluorophore coupling), standard MST cannot be applied. In such cases, however, label-free MST analysis can be an alternative, due to the read-out of intrinsic tyrosine and tryptophan fluorescence of the protein or the peptide.