Laminar flow chambers

The basic principle of studying molecule interactions with laminar flow chambers consists of monitoring the trajectories of a receptor bound to a cell or particle along a ligand-coated surface in the presence of a laminar flow (Robert et al., Reference Robert, Benoliel, Pierres and Bongrand2007; Zohra, Reference Zohra2021). The cell or particle is tethered to a surface and subjected to a hydrodynamic force (see Fig. 1a). The interaction frequency (e.g. the number of binding events per second), and the detachment kinetics (e.g. the fraction of particles remaining bound as a function of time after arrest) can be obtained from this assay (Pierres et al., Reference Pierres, Benoliel and Bongrand2008). This approach resembles the in vivo conditions of several protein–protein interactions, such as the binding between endothelial cell selectins and their glycoprotein ligands on leukocytes. Selectin interactions were the first molecules studied at the single-bond level with laminar flow chambers. Here, a rapid dissociation constant of 1 s⁻¹ was reported for the binding between selectins and P-selectin glycoprotein ligand 1 (PSGL-1; Alon et al., Reference Alon, Hammer and Springer1995). This technique has been widely used to investigate cellular processes that involve protein–protein interactions (e.g. cell adhesion or migration), respond to shear forces (Polacheck et al., Reference Polacheck, Charest and Kamm2011; Shemesh et al., Reference Shemesh, Jalilian, Shi, Heng Yeoh, Knothe Tate and Ebrahimi Warkiani2015). Additionally, this method has also been implemented to characterize single-molecule interactions, for example, between streptavidin and biotin (Pierres et al., Reference Pierres, Touchard, Benoliel and Bongrand2002). See Fig. 1a.

Single-molecule force spectroscopy techniques

Biomembrane force probe. BFP is a versatile method that can be used in a wide range of forces (10⁻¹–10³ pN) and loading rates (1–10⁶ pN s⁻¹) (Gourier et al., Reference Gourier, Jegou, Husson and Pincet2008). This technique consists of using cell-sized membrane capsules like vesicles, liposomes, or red blood cells (RBCs) as force transducers (Sung et al., Reference Sung, Sung, Crimmins, Burakoff and Chien1986; Shao Reference Shao2004). The membrane capsule is held in place by a suction micropipette, and a protein-coated bead can be attached to the force probe (Ungai-Salánki et al., Reference Ungai-Salánki, Peter, Gerecsei, Orgovan, Horvath and Szabó2019). This bead can then form the bond to be studied with a target, such as another bead, a vesicle, or a living cell, also held in place. The force measurements consist of automatized retraction–contraction cycles between the ligand-functionalized probe (i.e. bead connected to the soft spring RBCs) and target-bearing receptors (Gourier et al., Reference Gourier, Jegou, Husson and Pincet2008). During the retraction step, the bond is subjected to a tension force that stretches the transducer. This deformation is measured by precise video tracking of the bead joined to the transducer. These experiments allow the measurement of parameters like dissociation rates and transition state parameters such as the distance to the energy barrier (Brampton et al., Reference Brampton, Wahab, Batchelor, Allen and Williams2011). Using BFP, several works have characterized the mechanical responses of different biomolecular interactions to external forces. For instance, Evans and coworkers used BFP to study the impact of pulling on cell surface bonds, such as how the pairing of P-selectin and PSGL-1 propagates and alters receptor-cytoskeletal linkage beneath the membrane. They described that PSGL-1 detaches from cytoskeleton prior to failure of the exterior adhesive P-selectin:PSGL-1 bond (Evans et al., Reference Evans, Heinrich, Leung and Kinoshita2005; Evans and Kinoshita, Reference Evans and Kinoshita2007). Additionally, catch-slip behavior was identified for the interaction between PSGL-1 and P-selectin (Evans et al., Reference Evans, Leung, Heinrich and Zhu2004). In another example, Perret and coworkers tested the homophilic interaction between E-cadherins molecules and opposing cells (Perret et al., Reference Perret, Leung, Feracci and Evans2004). Furthermore, BFP has been used to study the intercellular adhesion molecule-1 (ICAM-1) with recombinantly activated αLβ2 integrin by divalent cations (Evans et al., Reference Evans, Kinoshita, Simon and Leung2010). BFP also allows quantifying how force and different conformations of lymphocyte function-associated antigen-1 regulate kinetics binding parameters with ICAM-1 (Chen et al., Reference Chen, Lou and Zhu2010). See Fig. 1b.

Atomic force microscopy. AFM is a technique that works in a force range of 10–10⁴ pN. It is unique in its ability to apply forces as high as nanonewtons. However, it is limited in its lower force resolution (lower than piconewtons) (Neuman and Nagy, Reference Neuman and Nagy2008). The basic principle consists of coating both the tip of a microfabricated cantilever and a substrate surface with a low density of interacting molecules of interest (Robert et al., Reference Robert, Benoliel, Pierres and Bongrand2007). The cantilever tip is brought to the surface of the substrate, so that the coated cantilever interacts with the substrate. This results in a mechanical force that will cause the cantilever to bend. Meanwhile, a laser beam that is reflected at the top of the cantilever is used to monitor the bending angle of the cantilever (Fuhrmann and Ros, Reference Fuhrmann and Ros2010), which gives a measurement of force (Meyer and Amer, Reference Meyer and Amer1988). The measurement procedure consists of repeating cycles of approach and retraction at a constant speed on a given location. When the cantilever tip is pulled apart from the substrate surface, the force required for bond rupture can be recorded (Robert et al., Reference Robert, Benoliel, Pierres and Bongrand2007). The frequency distribution of unbinding forces is obtained after enough cycles have been done (usually hundreds of approach–retraction cycles). Subsequent analysis of these rupture force values can be used to estimate the dissociation rate constants (Kada et al., Reference Kada, Blayney, Jeyakumar, Kienberger, Pastushenko, Fleischer, Schindler, Lai and Hinterdorfer2001; Lee and Marchant, Reference Lee and Marchant2001, Reference Lee and Marchant2003; Lo et al., Reference Lo, Zhu and Beebe2001; Zhang et al., Reference Zhang, Wojcikiewicz and Moy2002; Cai and Yang, Reference Cai and Yang2003; Kokkoli et al., Reference Kokkoli, Ochsenhirt and Tirrell2004; Kienberger et al., Reference Kienberger, Kada, Mueller and Hinterdorfer2005). Starting with the pioneering work on complementary DNA strands (Lee et al., Reference Lee, Chrisey and Colton1994a, Reference Lee, Kidwell and Colton1994b), biotin–streptavidin (Florin et al., Reference Florin, Moy and Gaub1994; Lee et al., Reference Lee, Kidwell and Colton1994b), and cell adhesion proteoglycans (Dammer et al., Reference Dammer, Popescu, Wagner, Anselmetti, Guntherodt and Misevic1995), this method has been applied to a wide range of interactions (Willemsen et al., Reference Willemsen, Snel, Cambi, Greve, De Grooth and Figdor2000): from complex biological intermolecular systems such as protein–nucleic acid (Bartels et al., Reference Bartels, Baumgarth, Anselmetti, Ros and Becker2003, Reference Bartels, McIntosh, Fuhrmann, Metzendorf, Plattner, Sewald, Anselmetti, Ros and Becker2007; Kühner et al., Reference Kühner, Costa, Bisch, Thalhammer, Heckl and Gaub2004; Baumgarth et al., Reference Baumgarth, Bartels, Anselmetti, Becker and Ros2005; Fuhrmann et al., Reference Fuhrmann, Schoening, Anselmetti, Staiger and Ros2009), antibody–antigen (Hinterdorfer et al., Reference Hinterdorfer, Baumgartner, Gruber, Schilcher and Schindler1996; Allen et al., Reference Allen, Chen, Davies, Davies, Dawkes, Edwards, Roberts, Sefton, Tendler and and Williams1997; Morfill et al., Reference Morfill, Blank, Zahnd, Luginbühl, Kühner, Gottschalk, Plückthun and Gaub2007), quadruplex nucleic acids (Lynch et al., Reference Lynch, Baker, Byker, Zhou and Sinniah2009), enzyme–inhibitor (Porter-Peden et al., Reference Porter-Peden, Kamper, Vander Wal, Blankespoor and Sinniah2008), and cell adhesion molecules (Fritz et al., Reference Fritz, Katopodis, Kolbinger and Anselmetti1998; Björnham and Schedin, Reference Björnham and Schedin2009; Dague et al., Reference Dague, Pons, Roland, Azaïs, Arcucci, Lachaize, Velmont, Trevisiol, N’Guyen, Sénard and Galés2022) to supramolecular examples (Eckel et al., Reference Eckel, Ros, Decker, Mattay and Anselmetti2005a) and synthetic biology (Eckel et al., Reference Eckel, Wilking, Becker, Sewald, Ros and Anselmetti2005b). See Fig. 1c.

Magnetic tweezers. MTs are a high-throughput method that can measure forces in the range of 10⁻³–10² pN. MTs can manipulate superparamagnetic beads, with the external magnetic field from a pair of small permanent magnets placed above the sample holder of an inverted microscope. In these assays, receptor proteins are bound to the superparamagnetic beads and allowed to contact their surface-bound ligand. Then the permanent magnets above the trapping chamber produce a magnetic field gradient along the axial direction, resulting in a controlled force on the bead directed toward the magnets (Neuman and Nagy, Reference Neuman and Nagy2008). MTs can be used to study several molecular interactions in parallel, offering an advantage versus other single-molecule force. For instance, this method has been used to study the prototypical ligand–receptor pair streptavidin–biotin (Claudia et al., Reference Claudia, Greenfield and Prentiss2005; Shang and Lee, Reference Shang and Lee2007; Kilinc et al., Reference Kilinc, Blasiak, O'Mahony, Suter and Lee2012). See Fig. 1d. Finally, new experimental approaches have combined AFM with MTs (Muñoz et al., Reference Muñoz, Aguilar Sandoval, Wilson and Melo2015)

Acoustic force spectroscopy. AFS uses acoustic waves to apply controlled forces on microparticles in a flow cell. The exerted forces are in the range of sub-piconewtons to hundreds of piconewtons, with loading rates between 10⁻⁴ and 10² pN s⁻¹ on thousands of protein–DNA complexes in parallel (Kamsma and Wuite, Reference Kamsma and Wuite2018). The flow cell consists of a piezo element and two glass layers with a fluidic channel in between. A protein–DNA complex attached by DNA end to a bead is allowed to contact the bottom glass coated with an interacting protein. The piezo element generates an acoustic wave that travels through the top glass to the bottom glass. The bottom glass reflects the acoustic wave, producing a standing wave over the flow cell. The bead exposed to this standing wave experiences a force along the vertical direction that applies tension to the interaction between the protein–DNA complex and the interacting protein (Lin et al., Reference Lin, Qin, Wuite and Dame2018). See Fig. 1f. AFS can be used to study DNA–protein interactions like DNA–viral capsid protein assembly (Van Rosmalen et al., Reference Van Rosmalen, Kamsma, Biebricher, Li, Zlotnick, Roos and Wuite2020), DNA–RecA filament formation, and to determine the energy landscape of digoxigenin–antidigoxigenin bonds (Sitters et al., Reference Sitters, Kamsma, Thalhammer, Ritsch-Marte, Peterman and Wuite2014). Furthermore, AFS allows to probe the mechanical properties of cells (Sorkin et al., Reference Sorkin, Bergamaschi, Kamsma and Wuite2018), the strength of molecular interactions of cells with the extracellular matrix (Kamsma et al., Reference Kamsma, Bochet, Oswald, Alblas, Goyard, Wuite, Peterman and Rose2018), and the unbinding behavior of carbohydrate binding modules to polysaccharide surfaces (Hackl et al., Reference Hackl, Contrada, Ash and Chundawat2021). See Fig. 1f.

Derivatization of the beads

A critical preliminary step is the binding of the relevant molecules to the beads (the order of magnitude of the diameter from ~10¹ nm to 10¹ μm; Neuman and Nagy, Reference Neuman and Nagy2008). There are many experimental approaches to attach molecules to beads, and a variety of beads to use (see Fig. 2b). Although polystyrene beads are the most common (Ramírez et al., Reference Ramírez, Rivera, Quiroga-Roger and Wilson2017; Burgos-Bravo et al., Reference Burgos-Bravo, Figueroa, Casanova-Morales, Quest, Wilson and Leyton2018), latex (Mejean et al., Reference Mejean, Schaefer, Millman, Forscher and Dufresne2009), silica (Block et al., Reference Block, Goldstein and Schnapp1990), and germanium (Sudhakar et al., Reference Sudhakar, Abdosamadi, Jachowski, Bugiel, Jannasch and Schäffer2021) beads are also available. There are several types of beads with different kinds of functionalization that can be used depending on the molecule to be studied: Ni-NTA-coated beads can be used to directly bind his-tagged proteins (Mejean et al., Reference Mejean, Schaefer, Millman, Forscher and Dufresne2009); G-protein-coated beads can be utilized to bind the molecule of interest through Fc-fusion proteins (Burgos-Bravo et al., Reference Burgos-Bravo, Figueroa, Casanova-Morales, Quest, Wilson and Leyton2018), specific antibodies (Peña-Oyarzun et al., Reference Peña-Oyarzun, Rodriguez-Peña, Burgos-Bravo and Criollo2020), or by modified-DNA linkers called ‘DNA handles’ (Ramírez et al., Reference Ramírez, Rivera, Quiroga-Roger and Wilson2017); streptavidin- or neutravidin-coated beads can also be used to bind the molecule directly (e.g. a biotinylated-protein) or through a biotinylated-DNA handle. G-protein- and streptavidin-coated beads are the most commonly used in OTs as they form strong bonds and are commercially available. The derivatization strategies of these polystyrene beads will be described below.

Fig. 2.

Bead experimental setup with miniTweezers. (a) Laminar flow chamber used in miniTweezer setup: two microchannels (1 and 3) are connected to the main channel (2) by connectors (4). Channels 1 and 3 allow the controlled entry of the functionalized beads with the proteins of interest. Bead A is injected by channel 1 and at the main channel is trapped and guided with the optical trap to a suction micropipette (5). Functionalized bead B enters by channel 2, at the main channel is trapped and guided with the optical trap in front of bead A. (b) Five examples of construct combinations that can be studied in OT. (b1) Bead A/Protein_G/anti_protein_A ↔ protein_A/cystein/S-S/DNA_handle/biotin/streptavidin/bead B. (b2) Bead A/streptavidin/biotin/DNA_handle/S-S/cysteine/Protein_A ↔ Protein_B/Halo_tag/DNA_handle/Dig/Anti_Dig/ProteinG. (b3) Bead A/proteinG/anti_protein_A/Protein_A ↔ Protein_B/Halo_ta/DNA_handle/Dig/Anti_Dig/ProteinG. (b4) Bead A/anti_protein_A/Protein_A ↔ Protein_B/Halo_ta/DNA_handle/Dig/Anti_Dig/Protein_G. (b5) Bead A/Protein_G/Fc tag/Protein_A ↔ Protein_B/Anti_proteinB/Protein_G. (c) Box with names of molecules involved.

Protein G-coated polystyrene bead/primary antibody against protein of interest. For proteins, protein G-coated beads can directly bind a fusion protein containing an IgG-Fc tag (Burgos-Bravo et al., Reference Burgos-Bravo, Figueroa, Casanova-Morales, Quest, Wilson and Leyton2018). A primary antibody against the protein of interest can be used in combination with protein G (Peña-Oyarzun et al., Reference Peña-Oyarzun, Rodriguez-Peña, Burgos-Bravo and Criollo2020; Mateluna et al., Reference Mateluna, Torres, Rodriguez–Peña, Silva, Matthies, Criollo, Bikker, Bolscher, Wilson, Zapata–Torres and Torres2022). For this option, protein G-coated beads are incubated with a primary antibody that binds the protein to be immobilized on the bead. Then, a crosslinking agent is used to covalently bind the protein G and the Fc tag or antibody (Peña-Oyarzun et al., Reference Peña-Oyarzun, Rodriguez-Peña, Burgos-Bravo and Criollo2020).

Protein G- or streptavidin-coated polystyrene beads/DNA handles. Another possibility to attach the protein of interest to polystyrene beads is using DNA handles (Ramírez et al., Reference Ramírez, Rivera, Quiroga-Roger and Wilson2017). This approach can be used for studying interactions between proteins (Ramírez et al., Reference Ramírez, Rivera, Quiroga-Roger and Wilson2017) and non-protein molecules like aptamers (Casanova-Morales et al., Reference Casanova-Morales, Figueroa, Alfaro, Montenegro, Barrera, Maze, Wilson and Conejeros2019). DNA handles are constructs between 500 and 1500 bp (Cecconi et al., Reference Cecconi, Shank, Dahlquist, Marqusee and Bustamante2008; Kaiser et al., Reference Kaiser, Goldman, Chodera, Tinoco and Bustamante2011) that act as linkers between the protein and the bead (Halvorsen et al., Reference Halvorsen, Schaak and Wong2011). These constructs separate the protein from the optically trapped bead, eliminating undesired interactions and photodamage. Shorter DNA handles (e.g. 29 bp) can be used to enhance the signal to noise ratio (Forns et al., Reference Forns, de Lorenzo, Manosas, Hayashi, Huguet and Ritort2011).

A DNA–protein complex is typically formed through a disulfide bond between a cysteine residue on the protein of interest and a cysteine at one end of the DNA handle (Ramírez et al., Reference Ramírez, Rivera, Quiroga-Roger and Wilson2017). Enzyme-based coupling strategies have also been used to form a covalent protein–DNA handle bond, including HaloTag fusion proteins that perform a dehalogenase reaction with the HaloTag ligand cross-linked to DNA handle (Aubin-Tam et al., Reference Aubin-Tam, Olivares, Sauer, Baker and Lang2011). By polymerase chain reaction (PCR), DNA handles are made using digoxigenin or biotin moiety-functionalized primers (Cecconi et al., Reference Cecconi, Shank, Marqusee and Bustamante2011). The process allows the binding of the DNA–protein complex to the protein G-coated spheres that have been derivatized with anti-digoxigenin (anti-dig) antibodies or streptavidin beads. The protein G or protein G/primary antibody methodology described above can also be used for refolding–unfolding and force–ramp experiments (Bustamante et al., Reference Bustamante, Sotelo-Campos, Guerra, Wilson, Bustamante and Báez2017).

The derivatization strategy to be used depends on the molecules that will be studied. When using DNA handles, at least 40% of the cysteine residue in a protein needs to be exposed to the solvent (Wilson, Reference Wilson2011), and at the opposite side of the interaction. If an intrinsic cysteine residue is not available, a mutant protein can be fabricated by replacing a serine residue with cysteine (Ramírez et al., Reference Ramírez, Rivera, Quiroga-Roger and Wilson2017). However, the retention of the native protein conformation must be confirmed. If there is a cysteine (wildtype or mutant) that meets these requirements, then the use of DNA handles is appropriate. When working with other molecules such as aptamers, the DNA handle approach can be appropriate, but there must be an SH moiety in order to bind the handle. In the case of aptamers, they can be directly bound to the DNA handles by their 3′ and 5′ ends (Cecconi et al., Reference Cecconi, Shank, Dahlquist, Marqusee and Bustamante2008). The DNA handle approach can be utilized on refolding and unfolding in trans interaction and constant force in cis interaction. For refolding and unfolding in trans interaction, there need to be two cysteine residues on the same molecule and at opposite sides. On the other hand, the protein G and protein G/primary antibody method requires the availability of a specific antibody that binds the molecule on the opposite side of the interaction. Nevertheless, a chimera of the target protein can be synthetized, for instance by adding a GFP protein (Thoumine et al., Reference Thoumine, Bard, Saint-Michel, Dequidt and Choquet2008), GSH molecule (Nicholas et al., Reference Nicholas, Rao and Gennerich2014), or Fc tag (Burgos-Bravo et al., Reference Burgos-Bravo, Figueroa, Casanova-Morales, Quest, Wilson and Leyton2018). Fusion proteins allow working with non-purified protein extract because of the binding specificity (Burgos-Bravo et al., Reference Burgos-Bravo, Figueroa, Casanova-Morales, Quest, Wilson and Leyton2018). Large molecules might be directly attached to spheres, while DNA handles are recommended for small molecules since surface imperfections on the polystyrene beads are always present at the Angstrom level. Thus, DNA handles avoid the possible interference of the bead surface when the size of the molecule is smaller than the imperfections of the polystyrene beads. The DNA handles function as spacers between the protein and the beads, keeping interactions between the tethering surfaces to a minimum (Cecconi et al., Reference Cecconi, Shank, Marqusee and Bustamante2011) and preventing photodamage of the protein of interest (Chen et al., Reference Chen, He, Srinivasakannan, Li, Yin, Peng, Guo and Zhang2019; Desai et al., Reference Desai, Frank, Lee, Righini, Lancaster, Noller, Tinoco and Bustamante2019). In this way, DNA handles minimize thermal and acoustic fluctuations, which is especially useful when there are impurities within the sample (Halvorsen et al., Reference Halvorsen, Schaak and Wong2011). These derivatization strategies can be combined in the final experimental setup. For example, DNA handles and the protein G or protein G/primary antibody approaches can be used to bind each of the molecules of study. Therefore, when choosing a strategy for binding the molecule of interest to the bead, all the factors mentioned above must be taken into consideration.

Preparing samples for OTs

As previously mentioned, the preparation of the samples is slightly different when using protein G/primary antibody versus DNA handles. In the case of protein G/primary antibody, the derivatization is performed in advance, and the binding of the protein to the beads occurs immediately before the experiment. In contrast, for DNA handles, the bead derivatization process is the first step (e.g. with anti-dig) followed by DNA handle-target molecule construct preparation (Fig. 2b). The complex is then incubated with the beads immediately before the experiment. A brief description of these two strategies will be given.

Preparation of the protein G-coated bead/primary antibody/protein complex. The derivatization of the beads is made as previously explained. After derivatization, a passivation process is recommended to prevent unwanted interactions. The passivation is carried out by incubating the beads with 1% w/v bovine serum albumin diluted in the interaction buffer (the buffer that will be used for the OT experiments) for 4 h at room temperature, then washing in the same buffer by centrifugation (Szili et al., Reference Szili, Al-Bataineh, Ruschitzka and Short2012). Passivation can also be done by adding β-casein at a final concentration of 1 mg ml⁻¹ followed by incubation while vortexing for 1 h at room temperature (Fukuda et al., Reference Fukuda, Yan, Komi, Sun, Gabizon and Bustamante2020). After passivation, the beads can be stored at 4 °C in the presence of sodium azide (0.05% w/v) for up to a year. To form the complex with the protein, the beads (0.5%, w/v) are incubated at the nanomolar concentration on ice or room temperature for 15 min. Then the beads are ready to be injected into the microfluidic chamber of the OTs (Fig. 2a).

Preparation of the streptavidin-coated bead/DNA handle/protein complex and protein G-coated bead/anti-dig antibody/DNA handle/protein complex. The first step is to synthesize the DNA handles (Cecconi et al., Reference Cecconi, Shank, Dahlquist, Marqusee and Bustamante2008). The handles are synthesized via PCR in a thermal cycler using a plasmid or lambda DNA and appropriately functionalized primers (Cecconi et al., Reference Cecconi, Shank, Dahlquist, Marqusee and Bustamante2008). For synthesizing biotin-DNA-SH handles, one of the primers contains a biotin moiety and the other contains an SH moiety, so the result will be DNA handles with a biotin and an SH group at each end. In the case of digoxigenin- DNA-SH handles, one primer contains a digoxigenin moiety instead of the biotin-primer. Primers with two dig moieties can also be used in order to guarantee more stability in the binding between the molecule and DNA handle (Desai et al., Reference Desai, Frank, Lee, Righini, Lancaster, Noller, Tinoco and Bustamante2019). The biotin or digoxigenin-DNA molecules are then bound to the streptavidin or anti-dig-coated beads, respectively. Then, the DNA handle-molecule construct can be made via disulfide bonding formation (Wilson, Reference Wilson2011).

The microfluidic channels of the OT setup are filled with buffer solution before injecting the derivatized beads as described in the following section.

Trapping functionalized beads

In this step, the beads functionalized with the molecules of interest are trapped and brought in contact to promote the protein–protein interaction. The laminar flow microfluidic chamber has three independent channels with small glass tube interconnections, physically separating the two different types of beads (Fig. 2a). The central channel has a micropipette connected to the outside of the microfluidic chamber. The catching process corresponds to trapping the beads from the different channels of the microfluidic chamber and bringing them to the central channel of the chamber where the experiment is carried out. The procedure for an experiment is as follows:

1. (1) The functionalized beads are injected into the upper and lower channels.

2. (2) A dual-laser (808 or 845 nm) optical-trap (trap stiffness of 0.1 pN nm⁻¹) is used to trap the bead in one of the channels and move it to the central channels where the micropipette is.

3. (3) Suction is applied on the micropipette to hold one of the beads in place.

4. (4) The optical trap is used to bring the second bead from the other channel close to the micropipette-trapped bead.

5. (5) The beads are briefly brought into contact (less than ~3 s) to promote interaction between the molecules.

Refolding and unfolding in trans interaction

This experimental approach consists of repeated cycles to mechanically pull a molecule until it loses its natural structure and then lower the tension until the molecule refolds. The beads are brought into contact for ~1 s or less (step 5 in Section ‘Trapping functionalized beads’). Then, the optical trap is moved away at a constant speed (order of magnitude ~10–100 nm s⁻¹) (Bustamante and Smith, Reference Bustamante and Smith2007) from the bead that is fixed in the pipette. If the molecules bond successfully while the beads are in contact, the extension–force curve will be with a positive slope (Fig. 3a), otherwise, the slope is null and the beads should be brought back into contact. As the optical trap is moved away, a sharp drop (or ‘rip’) in the tension can be observed whenever a specific structure of the molecule of interest unfolds. Finally, after completely unfolding the molecule, the optical trap is returned to its initial position, causing the system to relax and letting the molecule fold again. The experiment is repeated in the presence or absence of an interacting protein or other molecule added in trans in the buffer, and the frequency of rips is observed. These data are displayed as rip force histograms, from which the lifetime histogram is obtained.

Fig. 3.

Intermolecular interaction force experiments using OT. (a) Refolding and unfolding in trans interaction: at the top, protein A (green square) is tethered between two dsDNA handles attached to a bead held by a suction micropipette and an optically trapped bead. The optically trapped bead is pulled apart at a constant loading rate (arrow direction) stretching the molecule until it unfolds. Then protein B (purple) is added to the buffer and binds to the unfolded state of protein A preventing its refolding. At the left bottom, force–extension curves of the mechanical folding and unfolding events. The unfolding cycles are shown in red, and relaxation traces with refolding events at lower forces are shown in blue. The panel to the right shows no refolding event due to the binding of the interacting partner to the unfolded state. (b) Constant force in cis interaction: at the top, beads are approached to promote the interaction between the proteins. Once the interaction is observed, force is set and sustained constant until the bond is broken. At the bottom is a force–time graph. The interaction event is observed as an increase in the axis of force. The lifetime of the interaction is obtained directly from the graph corresponding to the time between the force is set and sustained constant until the bond is broken, observed as a sudden drop in force. (c) Force ramp in cis interaction. At the top, beads are approach and brought into contact to promote the interaction between the proteins. Then beads are separated by pulling apart the optically trapped bead at a constant loading rate (arrow direction) until the bond is broken. At the bottom is a force–extension curve. In case of interaction, an increase in the axis of force is noticed. Then at a certain distance the rupture of the interaction is produced, observed as a sudden drop in force.

For example, the in trans interaction scheme was used to study the interaction of binding immunoglobulin protein (BiP), a major chaperone from the endoplasmic reticulum, with its protein substrate, the MJ0366 (Ramírez et al., Reference Ramírez, Rivera, Quiroga-Roger and Wilson2017). Here, BiP was added in trans, while the MJ0366 protein was tethered between two beads (Fig. 3a). Since BiP interacts with unfolded proteins (Schmid et al., Reference Schmid, Baici, Gehring and Christen1994), OTs were used to mechanically control the folded state of the substrate. The MJ0366 protein was unfolded by pulling the optically trapped bead at a constant loading rate and refolded by lowering the tension (i.e. relaxing cycle). Experiments performed in the presence of BiP showed a loss of refolding and unfolding events in the consecutive pulling–relaxing cycles (Fig. 3a) in contrast to the experiments in the absence of BiP, verifying that BiP binds to the unfolded state of MJ0366 protein and prevents its refolding. From these features in the force–extension curves, the BiP's dissociation constants were calculated by measuring the probabilities for observing rips (the presence of unfolding and refolding events) and no rips (the absence of unfolding and refolding events). The reversibility of BiP binding to its substrate was also determined due to the reappearance of refolding and unfolding events. The time intervals that BiP stayed bound (e.g. disappearance of unfolding/refolding events, τ _off) or unbound (e.g. reappearance of unfolding/refolding events, τ _on) to the protein substrate were directly calculated. The corresponding k _off and k _on were estimated by assuming an exponential distribution for the off and on lifetimes (Ramírez et al., Reference Ramírez, Rivera, Quiroga-Roger and Wilson2017). It was also determined that the binding–unbinding of BiP to its substrate depends on both the type and concentration of nucleotide, indicating that ATP hydrolysis regulates the binding and release of this chaperone to the substrate protein (Ramírez et al., Reference Ramírez, Rivera, Quiroga-Roger and Wilson2017).

Naturally, this assay can be used to study the binding between other kinds of molecules. For example, a refolding and unfolding experiment was used to study the dynamics of APTSTX1–aptamer, a single-stranded DNA molecule with a high affinity for saxitoxin (STX), a lethal neurotoxin (Casanova-Morales et al., Reference Casanova-Morales, Figueroa, Alfaro, Montenegro, Barrera, Maze, Wilson and Conejeros2019).

Constant force in cis interaction

Constant-force assays (also called force-clamp assays in the literature) have been widely used to study protein interactions. In this assay, the beads are brought into contact for ~2 s or less in order to promote the interaction between the proteins (step 5 in Section ‘Trapping functionalized beads’). Once the binding is confirmed by an increase in the tension, the optical trap is fixed at a constant force. The duration of the interaction is measured, until the bond rupture is observed as a sudden decrease in the tension. This time is related to the lifetime τ of the interaction at a specific force. Then, the experiment is repeated in order to measure the decay times as a function of the force. This strategy should be repeated a minimum of 20 times, and at least five pairs of new beads should be used for each measured force.

The range of forces that will be applied to the molecule depends on previous information and characteristics of the interaction. If this is unknown, a large range of forces can be used initially to find the most relevant region for a given interaction. It is recommended to complement the experiment with the previous force ramp experiment, which can also provide some information on τ at different forces. Using these data, it is possible to extract kinetic parameters by using the Dudko–Hummer–Szabo (DHS) model (Dudko, Reference Dudko, Hummer and Szabo2008).

For instance, the dynamic-catch bond behavior was described for the first time in the characterization of the interaction of Thy-1 with both α ₅β ₁ integrin and syndecan-4 by BFP (Fiore et al., Reference Fiore, Ju, Chen, Zhu and Barker2014). Other constant-force assays include F-actin and G-actin dissociation kinetics (Lee et al., Reference Lee, Lou, Wen and McIntire2013), kinetochore-microtubule attachments (Akiyoshi et al., Reference Akiyoshi, Sarangapani, Powers, Nelson, Reichow, Arellano-Santoyo, Gonen, Ranish, Asbury and and Biggins2010), P-selectin, and PSGL-1 binding (Marshall et al., Reference Marshall, Long, Piper, Yago, McEver and Zhu2003).

On- and off-rates, mechanical stability and affinity measurements have been measured by studying proteins directly attached to the surface of the beads (Fig. 2b). For instance, the specific interactions between glycosylphosphatidylinositol-anchored Thy-1 protein, an abundant neuronal cell adhesion molecule, and its astrocytic receptors, the transmembrane α _vβ ₃ integrin and the proteoglycan of heparan sulfate syndecan-4, was characterized using this approach (Burgos-Bravo et al., Reference Burgos-Bravo, Martínez-Meza, Quest, Wilson and Leyton2020). For this, Fc-tagged recombinant molecules were used to allow direct coupling to the beads via the strong and high-affinity Fc-protein G interactions (Fig. 2b). Specifically, the bond lifetime of the Thy-1-dependent interaction as tri-molecular complex (i.e. Thy-1-Fc/syndecan-4-Fc + α _vβ ₃-Fc) was determined by constant-force assays (Fig. 3b), where the duration before bond dissociation is measured directly under different forces. The force-dependent lifetime data were fitted using the Bell (Reference Bell1978) model showing an ordinary slip-bond behavior for the tri-molecular complex (Burgos-Bravo et al., Reference Burgos-Bravo, Martínez-Meza, Quest, Wilson and Leyton2020). Furthermore, the Bell model was also used to calculate the kinetic parameters of the unbinding process, obtaining a k _off at zero force of ~0.08 s⁻¹ and distance to the transition state of ~0.2 nm.

Force ramp in cis interaction

Another strategy to study molecular bonds are force-ramp assays. In this approach, the forces required for bond dissociation (i.e. rupture forces) are measured and used to obtain force-dependent kinetic parameters and mechanical response of molecular bonds. Rupture-force data are obtained by pulling apart the optically trapped bead at a constant loading rate after the beads are brought into contact and then an interaction is formed (also referred to as approaching and retraction cycle; see detailed explanation in Section ‘Detailed description of intermolecular interaction force experiments using OTs’). As shown in Fig. 3c, rupture force values resulting from single-molecule bindings are determined directly from the force–trap position traces with a single-step bond rupture signature (Litvinov et al., Reference Litvinov, Bennett, Weisel and Shuman2005). The beads are brought into contact to promote protein–protein interaction (step 5 in Section ‘Trapping functionalized beads’). Then the optical trap is pulled apart at a constant velocity, much like in folding/unfolding experiments. The interaction between the two molecules resists the pulling, generating a measurable tension that increases with distance. After enough force is applied, the interaction yields, and the tension force drops suddenly to zero. The maximum force achieved before the separation of the system is then recorded as the ‘rupture force’ (Fig. 3c). These approach–tension–release cycles are repeated at least 50 times for an average pair of beads. Beads are changed approximately six times in order to achieve around 300 total pulling events, with and without interaction (a number easily achievable in a single-productive day and can be sufficient for robust statistics). These data are displayed as force ramp histograms, for which the underlying probability distribution can be fitted in order to extract the kinetic parameters.

Force-ramp assays were used to obtain rupture force data for the Thy-1/α _vβ ₃ integrin interaction (Burgos-Bravo et al., Reference Burgos-Bravo, Figueroa, Casanova-Morales, Quest, Wilson and Leyton2018). In this study, the most probable bond rupture force or bond strength of ~25 pN was determined pulling at a constant loading rate of 10 pN s⁻¹ (Burgos-Bravo et al., Reference Burgos-Bravo, Figueroa, Casanova-Morales, Quest, Wilson and Leyton2018). The force magnitude was in the previously reported range for non-covalent interactions (Weisel, Reference Weisel2003). Likewise, bond strengths between 25 and 44 pN were reported for the interactions between Staphylococcus aureus protein A and several species immunoglobulin G molecules, such as rabbit and goat, by using OT and force-ramp assays (Stout, Reference Stout2001). The kinetic parameters of the force-induced dissociation process can be characterized using the DHS model (Eq. (9); Dudko et al., Reference Dudko, Hummer and Szabo2008), which permits transforming the rupture force histograms into force-dependent lifetimes τ(F) (for more details see Section ‘Detailed description of intermolecular interaction force experiments using OTs’). This methodology permitted the characterization of the complete energy landscape (Δx ^‡ ~ 0.50 nm and ΔG ^‡ ~ 5.54k _BT) and the zero-force kinetic parameter (k _off⁰ ~ 0.051 s^–1) of the Thy-1/α _vβ ₃ dissociation process (Burgos-Bravo et al., Reference Burgos-Bravo, Figueroa, Casanova-Morales, Quest, Wilson and Leyton2018). Moreover, the force-dependent lifetime showed that force exponentially increased Thy-1/α _vβ ₃ integrin dissociation, indicating a slip-bond behavior for this molecular bond.

Applying the single-molecule OT technique, the binding specificity and direct interaction between a pair of proteins can be determined from the frequency of adhesion measurements (e.g. the probability for adhesion based on the total number of contacts) (Robert et al., Reference Robert, Benoliel, Pierres and Bongrand2007; Stangner et al., Reference Stangner, Wagner, Singer and Kremer2013). Events with and without binding are used to calculate adhesion frequency in a sequence of approaching–retraction cycles between both beads during the force-ramp experiments. This strategy was used to verify the specific interaction between syndecan-4 and Thy-1 (Burgos-Bravo et al., Reference Burgos-Bravo, Martínez-Meza, Quest, Wilson and Leyton2020). Using a similar strategy, Peña-Oyarzun and coworkers confirmed the existence of a protein complex between the mechanosensory Ca²⁺ channel polycystin 2 (PKD2) and scaffolding protein Beclin 1 (BECN1), two autophagy-related proteins (Peña-Oyarzun et al., Reference Peña-Oyarzun, Rodriguez-Peña, Burgos-Bravo and Criollo2020). Here, to perform OT experiments, GST-PKD2 and 6xHis-BECN1 fusion proteins were attached to GSH- and anti-His-coated polystyrene beads, respectively. A bond strength of ~12 pN average (loading rate of 10 pN s⁻¹) and a dissociation rate of ~0.28 s⁻¹ were calculated for the PKD2/BECN1 protein complex, suggesting a weak and transient protein–protein interaction. Moreover, an ordinary slip-bond behavior was described for this complex. By using this technique, a physical interaction was identified between PKD2 and BECN1 that is required for autophagosome formation in human cells. Recently, this strategy has been used to prove the interaction between histatin-1 and the receptor VEGFR2 (Mateluna et al., Reference Mateluna, Torres, Rodriguez–Peña, Silva, Matthies, Criollo, Bikker, Bolscher, Wilson, Zapata–Torres and Torres2022). Based on the kinetic parameters and rupture forces obtained, the authors proposed that the histatin-1–VEGFR2 interaction is specific but transient.

A tether ligand assay is another strategy that can be used to measure binding rates of ligand–receptor pairs under force. Here, the goal is to hold the same binding partners in spatial proximity under mechanical control. A fusion construct is prepared between a pair of interacting proteins joined by a flexible polypeptide linker. This allows the measurement of repeated interactions of the same pair of molecules. This strategy has provided insights into von Willebrand factor (VWF) binding of the glycoprotein Ibα subunit (GPIbα) on the surface of platelets (Kim et al., Reference Kim, Zhang, Zhang and Springer2010, Reference Kim, Hudson and Springer2015), its essential role in the platelet plug formation, and their resistance to strong hydrodynamic forces. A complex was formed between GPIbα and the A1 domain of VWF using a polypeptide linker and attached to a pair of beads by DNA handles. By moving the optically trapped bead, this construct was subjected to pulling and relaxing cycles (e.g. force-ramp assay), allowing the unbinding and rebinding of the GPIbα–A1 complex. Both unbinding and binding processes are characterized by the abrupt increase and decrease in the tether length, respectively. This experimental strategy has also been used to study the force-sensitive mechanosensory role of filamin (Rognoni et al., Reference Rognoni, Stigler, Pelz, Ylanne and Rief2012).

Both force-ramp and constant-force assays have been extensively used to study the mechanical regulation of molecular interactions by experimentally measuring kinetic parameters under force. However, some differences have been described for both strategies. The force-ramp gives a higher throughput than constant-force assays. In the first strategy, the rupture forces are obtained from all binding events, while the constant force measures the lifetime from a fraction of binding events that survive the jump into the desired force. The bond lifetime values are obtained directly by the constant-force approach, without the extra data processing/analysis needed in force-ramp experiments. Moreover, when force is ramped at a constant rate, the rupture forces are measured in a dynamic condition. In contrast, when the force is clamped at a constant level, the bond dissociation is determined under static conditions, where the same force is applied to all components in the system, including interacting molecules, force sensor, and molecular linkers. Interestingly, using the constant force setup, it was possible to discover the catch (Marshall et al., Reference Marshall, Long, Piper, Yago, McEver and Zhu2003) and ideal bonds (Rakshit et al., Reference Rakshit, Zhang, Manibog, Shafraz and Sivasankar2012).

Considering the high sensitivity of this single-molecule technique, purified proteins are desirable to study unbinding parameters using OTs, for example, to reduce the rupture force resulting from non-specific interactions. The strength estimated for these non-specific binding events can reach values of ~10 pN (Weisel, Reference Weisel2003). Even when this rupture force peak might differ from the obtained for the specific dissociation between the proteins of interest, the non-specific interaction distribution partly overlaps with the lower rupture forces range produced by specific protein–protein interactions. To address this issue, a novel mathematical approach based on the law of total probabilities was designed to effectively correct the rupture force measurements for non-specific binding events, consequently obtaining an accurate estimation of the specific rupture forces for protein–protein dissociation and calculating the kinetic and transition state parameters (Burgos-Bravo et al., Reference Burgos-Bravo, Figueroa, Casanova-Morales, Quest, Wilson and Leyton2018). A complete description of the mathematical correction is given in Section ‘Detailed description of intermolecular interaction force experiments using OTs’. It has been reported that the work with purified protein still suffered from residual non-specific interactions (Burgos-Bravo et al., Reference Burgos-Bravo, Figueroa, Casanova-Morales, Quest, Wilson and Leyton2018), highlighting the importance of this mathematical correction method when single-molecule protein–protein interaction studies are performed.

Obtaining biochemical parameters through the DHS model

Once the specific force histogram is obtained, it is normalized and fitted to the DHS model (Dudko et al., Reference Dudko, Hummer and Szabo2008). First, it is necessary to calculate the lifetimes of the interaction with force dependence, τ(F):

(7)$$\tau \left({F_0 + \left[{k-\displaystyle{1 \over 2}} \right]\Delta F} \right) = \displaystyle{{\left({h_k/2 + \mathop \sum \nolimits_{i = k + 1}^N h_i} \right)\Delta F} \over {h_k\dot{F}( F_0 + [ {k-1/2} ] \Delta F}}$$

In this equation, k is an integer, with values 1, 2,…, N − 1, N indicates the number of a bar on the histogram. F ₀ is the minimum recorded force, and ΔF is the width of the bars of the histogram. Each bar has a height h _k, corresponding to the counts in bar k, C _k, normalized so that the total area is equal to: h _k = C _k/C _Tot ⋅ ΔF, with C _Tot being the total number of accounts. The loading rate $\dot{F}$ is the rate of change in force over time. When modeling the DNA handles as a worm-like chain, the loading rate corresponds to:

(8)$$\dot{F}( F ) = v\left[{\displaystyle{1 \over {k_{\rm s}}} + \displaystyle{{2\beta L_{\rm C}L_{\rm P}( {1 + \beta FL_{\rm P}} ) } \over {3 + 5\beta FL_{\rm P} + 8{( {\beta FL_{\rm P}} ) }^{5/2}}}} \right]^{{-}1}$$

where v is the constant speed of the experiment (in our case 100 nm s⁻¹). The spring constant of the miniTweezers, k _s, has a value of 0.1 pN nm⁻¹. The contour length (L _C) is the end-to-end length of the DNA handle, in this case 180 nm. The persistence length, (L _P) (Bouchiat et al., Reference Bouchiat, Wang, Allemand, Strick, Block and Croquette1999), is the characteristic length at which the chain remains rigid. In the case of double-stranded DNA, L _P has an average value that depends on the experimental conditions. In physiological buffer, it is ~50 nm (Bustamante et al., Reference Bustamante, Bryant and Smith2003). β = (k _BT)⁻¹ corresponds to the reciprocal of the thermal energy.

Then, the lifetimes of the interaction in dependence on force are given by:

(9)$$\tau ( F ) = \tau _0\left({1-\displaystyle{{\nu F\Delta x^\ne } \over {\Delta G^\ne }}} \right)^{1-1/\nu }e^{-\beta \Delta G^\ne [ {1-{( {1-\nu F\Delta x^\ne /\Delta G^\ne } ) }^{1/\nu }} ] }$$

From the previous expression, at zero force, three parameters are obtained:

1. (1) The lifetime of interaction τ ₀, whose reciprocal corresponds to the dissociation kinetic constant, k _off.

2. (2) The free energy difference to the transition state ΔG ^≠, which corresponds to the energy barrier for the dissociation process.

3. (3) The distance to the transition state Δx ^≠, which correlates with the interactions that must be broken to separate the two molecules (Schlierf and Rief, Reference Schlierf and Rief2005; Tych et al., Reference Tych, Hoffmann, Brockwell and Dougan2013; Stigler and Rief, Reference Stigler and Rief2015).

The scaling factor ν can be adjusted to values ν = 1/2 for a harmonic oscillator, ν = 2/3 for a potential with linear and cubic terms, and ν = 1 for an exponential known as the Bell equation (Bell, Reference Bell1978). For this last case, where τ(F) = τ ₀e ^−βFΔx≠, it is not possible to obtain values of ΔG ^≠.