Control of Ligand-Binding Specificity Using Photocleavable Linkers in AFM Force Spectroscopy.

In recent decades, atomic force microscopy (AFM), in particular the force spectroscopy mode, has become a method of choice to study biomolecular interactions at the single-molecule level. However, grafting procedures as well as determining binding specificity remain challenging. We report here an innovative approach based on a photocleavable group that enables in situ release of the ligands bound to the AFM tip and thus allows direct assessment of the binding specificity. Applicable to a wide variety of molecules, the strategy presented here provides new opportunities to study specific interactions and deliver single molecules with high spatiotemporal resolution in a wide range of applications, including AFM-based cell biology.

Over the last decades, AFM has proved to be one of the main techniques for the nanoscale characterization and manipulation of biological specimens under physiological conditions thanks to the combination of its high spatiotemporal resolution and its ability to work in controlled atmospheres.[1] Rapidly, force spectroscopy mode developments together with the ability to functionalize the AFM tip with covalently attached biological moieties opened new avenues to probe the forces involved in biointeractions down to the piconewton range and to extract the association dissociation kinetics through established biophysical models.[2] In addition, technological advances in force–distance curve-based AFM (FD-based AFM) enable us to correlate the biomolecular force characterization to simultaneous high-resolution images of biological structures under native conditions.[3] As a result, AFM has become one of the most versatile tools for ligand–receptor single-molecule force spectroscopy, which can be performed on systems ranging from biomolecule-grafted model surfaces to cellular receptors and bacterial and mammalian cells.[4] Moreover, nowadays fluorescence microscopy techniques are routinely implemented in a variety of commercial AFM setups, allowing structural identification of the biological targets by cell biology approaches and further expanding the range of applications of the technique.[5]

Notwithstanding the huge progress AFM has witnessed over the years, some of the most fundamental limitations of AFM-based force spectroscopy have not yet been overcome. Mainly, the technique still suffers from uncertainties linked to the identification of the specific interactions established between the ligand and its specific receptor, among all the inevitable nonspecific interactions.[6,7] The latter are generated from interactions of the tip and/or the immobilized ligand with surface components other than the receptor of interest (e.g., spacers, bare surface areas for model substrates, and other biomolecules in the case of more complex systems). The proportion of nonspecific interactions depends on the complexity of the investigated biological systems, i.e., from receptor-grafted surfaces to lipid membrane models and cells, in which having multiple spurious interactions becomes the rule rather than the exception. Current strategies to differentiate specific and unspecific interactions rely on control experiments using nonfunctionalized AFM tips and/or blocking experiments in which free ligands in excess are injected to prevent the specific interaction (ligand itself, antagonist, or antibody).[8,9] Nevertheless, these strategies do not always lead to convincing results and are usually time- and cost-consuming.

Here we present an optochemical approach that takes advantage of the combination of AFM and fluorescence microscopy to control the release of biomolecules from the AFM tips via a photocleavable group inserted into the poly(ethylene glycol) (PEG) spacer during the tip functionalization (see Scheme ). The insertion of a photoresponsive o-nitrobenzyl unit[10,11] within the hydrophilic PEG spacer enables us to achieve postmeasurement light-induced dissociation of the covalently attached ligand. Similar linker cleaving strategies have been proposed before, but they either require the addition of reducing agents potentially unsuitable for complex biological systems[12] or are limited to gold-coated AFM cantilevers, which restrict the range of applications.[13] In our experimental setup, the light source is a standard mercury arc lamp equipped with a G 365 bandpass excitation filter, which is routinely used for fluorescence dyes in the UV range, such as the cell nuclear stain 4′,6-diamidino-2-phenylindole (DAPI), and is therefore widely available on most commercial microscopes. This easy-to-implement strategy provides a practical way to introduce internal controls into AFM-based force spectroscopy experiments and opens new avenues to bio-AFM-based applications.

Scheme 1

Schematic and Experimental Concept of an AFM Tip Functionalized with a Photocleavable Group, Followed by Heterobifunctional PEG Linker Attachment and Ligand Coupling to Probe the Interaction with a Cognate Receptor before and after Controlled Ligand Release via Photolysis at 365 nm

As a first proof of concept of our approach, we investigated the well-characterized avidin–biotin interaction.[14] We functionalized the AFM tips with biotin groups using the heterobifunctional PEG linker with the photocleavable group by a multistep procedure (see Figure S1 in the Supporting Information) and probed the interactions with avidin molecules physisorbed on glass (Figure a–d). In parallel, biotin-functionalized AFM tips lacking the photocleavable group were used as a negative control (Figure e–h). For both photocleavable tips (Figure b,c) and control tips (Figure e,f), rupture forces in the range of 50–100 pN were detected, in good agreement with values reported in the literature for avidin–biotin interactions[14] (examples of FD curves are presented in Figure S2). Remarkably, upon irradiation of the photocleavable AFM tips for 5 min at 365 nm in situ, the frequency of rupture forces in this range (50–100 pN) was drastically reduced (Figure d), while no decrease was observed for the binding probability of the biotin-functionalized control tips (Figure h). These results clearly indicate efficient photolysis of the photoreactive linker, leading to a controlled switch-off of the biotin–avidin interactions in situ. Notably, direct irradiation of the functionalized AFM tips through the glass substrate does not lead to significant sample damage, as evidenced by the control experiment, for which the binding probability was not altered after light exposure. A higher binding probability is observed with the photocleavable tips in comparison with the control tips before photocleavage (Figure d,h). This suggests that the linker structural change imparted by the insertion of the photocleavable spacer could allow more orientational freedom for biotin, resulting in a higher number of detected avidin–biotin interactions.

Figure 1

Force spectroscopy of single avidin–biotin interactions acquired with (a–d) photocleavable and (e–h) control AFM tips. (a) Scheme of the photocleavable tips system. (b, c) Adhesion force histogram and (inset) 5 μm × 5 μm map of the avidin–biotin interaction before (b) and after (c) photolysis at 365 nm for 5 min. (d) Box plot of the relative binding frequency before and after (dashed) photolysis, showing a significant decrease in binding. (e) Scheme of the control tips system. (f, g) Adhesion force histogram and (inset) 5 μm × 5 μm map of the avidin–biotin interaction before (f) and after (g) photolysis at 365 nm for 5 min. (h) Box plot of the relative binding frequency before and after (dashed) photolysis, showing no significant decrease in binding. Data are representative of at least n = 3 independent experiments, with a minimum of n = 10 analyzed adhesion maps before and after photolysis. ns, P > 0.05; ****, P < 0.0001; determined by two-sample t test in Origin.

Next, similar experiments were performed on a His6 peptide–tris-NTA model system (Figure ). In this setup, photocleavable (Figure a–d) and control (Figure e–h) tris-NTA-functionalized AFM tips were used to probe the interaction with His6-peptide-functionalized surfaces (see Figure S3). As for avidin–biotin, after an initial assessment of tris-NTA–His6 rupture forces and binding probability, the functionalized AFM tips were exposed to UV light in situ for 5 min to release the tris-NTA groups. Tris-NTA was found to bind the His6 peptide with rupture forces in the 50–200 pN range[15,16] and binding probabilities up to ∼80%, indicating a high peptide surface coverage under our experimental conditions. Similarly to what we observed for the avidin–biotin system, the binding probability of the tris-NTA-functionalized photocleavable tips decreased considerably after UV light exposure, while no variation was found for the tris-NTA control tips, confirming that biomolecular interactions can be efficiently suppressed in situ through controlled photocleavage of the ligand from the AFM tips.

Figure 2

Force spectroscopy of single His6 peptide–tris-NTA interactions acquired with (a–d) photocleavable and (e–h) control AFM tips. (a) Scheme of the photocleavable tips system. (b, c) Adhesion force histogram and (inset) 5 μm × 5 μm map of the His6 peptide–tris-NTA interaction before (b) and after (c) photolysis at 365 nm for 5 min. (d) Box plot of the relative binding frequency before and after (dashed) photolysis, showing a significant decrease in binding. (e) Scheme of the control tips system. (f, g) Adhesion force histogram and (inset) 5 μm × 5 μm map of the His6 peptide–tris-NTA interaction before (f) and after (g) photolysis at 365 nm for 5 min. (h) Box plot of the relative binding frequency before and after (dashed) photolysis, showing no significant decrease in binding. Data are representative of at least n = 3 independent experiments, with a minimum of n = 10 analyzed adhesion maps before and after photolysis. ns, P > 0.05; ****, P < 0.0001; determined by two-sample t test in Origin.

Finally, to demonstrate that our strategy is applicable to more complex biological systems, we tested the feasibility of the photocleavable linker to study single-virus binding to living mammalian cells. To this end, we functionalized AFM tips with fluorescently labeled reoviruses with or without a photocleavable group (see Figure S4). The functionalized tips were used to probe virus interactions with Chinese hamster ovary (CHO) cells overexpressing junction adhesion molecule A (JAM-A), which we recently showed to be an efficient binding receptor for reoviruses.[17] By the use of FD-based AFM combined with laser scanning confocal microscopy (Figure ), the AFM tip was raster-scanned on the cell surface, and rupture forces were extracted for each pixel. Imaging of the AFM tip using laser-scanning confocal microscopy showed that after 5 min of UV light exposure, reoviruses were efficiently removed from the AFM tips bearing the photocleavable groups but remained attached to AFM tips having no photocleavable group (either directly irradiated (Figure a) or irradiated through the cell sample (Figure b, top row)). Furthermore, the comparison between AFM topography (Figure b, middle row) and adhesion images (Figure b, bottom row) revealed that virus release was accompanied by a significant decrease of binding event probability, from ∼12% to ∼3–4%. As a control, consecutive mapping with AFM tips lacking the photocleavable groups did not show this behavior, with a stable binding event probability of ∼10%, confirming that the virus remained attached for several consecutive maps. Indeed, although a slight decrease of the fluorescence intensity was observed, likely due to photobleaching, the control tips did not seem to significantly release reoviruses (Figure c,d, top row) or lose cell-binding capability after light exposure (Figure d, middle and bottom rows), suggesting that both phenomena are specifically related to the linker cleavage and virus release. The proof-of-concept experiments presented here confirm the versatility of the photocleavable approach. In addition, preliminary results demonstrated the possibility of releasing virions at a defined location on a cell layer upon photocleavage (Figure S5), opening new opportunities to further AFM- and fluorescence-microscopy-based cell biology applications.

Figure 3

(a, b) Proof-of-concept and (c, d) control (no photocleavable group) experiments on large-scale molecules (reovirus) interacting with specific reovirus attachment receptors (sialic acid, JAM-A) probed on living CHO-JAM-A cells (see the Supporting Information). (a) AFM tip functionalized with a fluorescently labeled (Alexa 488, green) reovirus releases most of the single viruses after photocleavage and shows a significant decrease in binding frequency after probing their interaction on JAM-A-overexpressed CHO cells (highlighted areas in b). (c) The control experiment shows almost no virus release after photocleavage as well as no significant drop in binding frequency (highlighted areas in d).

In summary, we have presented a practical and versatile optochemical approach to control the release of biomolecules from AFM cantilevers. The method has been applied successfully both to receptors immobilized on model surfaces and to more complex biological systems. The ligand release upon UV light exposure appears to be a powerful internal control for force spectroscopy experiments. Furthermore, more sophisticated biology-oriented applications for combined AFM–optical microscopy instruments, such as the possibility of releasing ligands at specific cellular locations while following the evolution of cellular responses, are envisaged in the near future.