Living Cell Study at the Single-molecule and Single-cell Levels by Atomic Force Microscopy

Xiaoli Shi; Xuejie Zhang; Tie Xia; Xiaohong Fang


Nanomedicine. 2012;7(10):1625-1637. 

In This Article

AFM Force Measurement

Owing to the limitation of AFM imaging on living mammalian cells, AFM force mode has become an essential and more popular mode than imaging in living cell studies. With AFM force mode, quantitative information on cellular interactions at the single-molecule level can be obtained.

The schematic illustration of the principle of AFM force mode is shown in Figure 1B. In principle, the tip is held over the substrate and the piezo scanner, which is installed to control the tip or the sample, makes the tip moving relative to the sample vertically up and down (approaching and retracting) in cycles. During each cycle, the piezo scanner firstly moves to let the tip press onto the sample surface. Then, after a point of maximum load is reached, the direction of motion is reversed and the piezo scanner is withdrawn to let the AFM tip leave the surface of the sample. The tip leaves the surface to complete a measurement cycle. Vertical position of the tip and deflection of the cantilever are recorded to measure the force between the AFM tip and the sample at each position. This information is converted to force-distance (F-D) curves, also known as 'force curves' or force spectroscopy. From the F-D curves, information such as adhesion force between the probe and samples, as well as sample stiffness and viscoelasticity, can be extracted by mathematical analysis.[22,23]

Single-molecule Force Spectroscopy

If the tip is functionalized by ligand molecules and the substrate by receptors, approaching of the tip to the substrate allows ligand and receptor to bind with each other, whereas their separation ruptures the ligand–receptor binding. Normally, chemical reactions are used to covalently attach the biomolecules to the AFM tips and substrates with linear poly(ethylene glycol) (PEG) chains as a spacer. The PEG spacer makes it easy to differentiate the nonspecific interaction from the specific binding between biomolecular pairs and to keep the immobilized biomoleclues in a more flexible orientation.[24,25] The ligand–receptor interactions could then be measured by the deflection of the AFM cantilever. The strength of the rupture force is calculated from the maximum cantilever deflection during the retraction process in the force curve. This is how AFM single-molecule force spectroscopy (SMFS) works to measure the noncovalent binding force between one pair of biomolecules. By changing the speed of tip-sample rupture to get dynamic force spectroscopy, valuable parameters on ligand–receptor dissociation kinetics including dissociation rate constants, the distance to the transition state and dissociation energy landscapes can be obtained.[21,26] Besides SMFS, AFM force mode also includes single-cell force spectroscopy where the AFM tip is modified with an entire living cell instead of ligand molecules.[27]

Force plays a fundamental role in biological processes. By characterization of interactions within or between molecules, SMFS has offered answers to a number of biologically and medically pertinent questions. Following the work of Moy et al.[28] and Lee et al.[29] on probing the strength of the biotin–avidin bond, SMFS has been successfully employed to measure intramolecular unfolding forces of individual proteins[30,31] and intermolecular forces of a variety of biomolecular interactions including ligand–receptor, antibody–antigen, protein–DNA and DNA–DNA.[25,26,32,33] However, these biomolecular interactions measured with the purified proteins or DNAs may be different from those in the living cells, as the conformation and function of biomolecules are highly controlled by the cellular environment. Thus, it is critical to develop SMFS directly on living cells.[34] For this purpose, AFM especially designed for integration with an inverted optical microscope is used. The optical or fluorescence images of samples are obtained to locate the AFM tips onto the target regions for SMFS. In the following sections, we will summarize some major applications of AFM living-cell SMFS to the studies in biology and medicine, such as cell adhesion, signal transduction and receptor mapping.

Cell Adhesion

Cell adhesion regulates many important biological processes including embryonic development, cellular communication, inflammation, wound healing, tumor metastasis and viral/bacterial infection. The adhesion is achieved by interaction among specific cell-adhesion molecules (CAMs), and up to now, the molecular mechanisms of how CAMs work in different types of adhesions have been open to debate.[35] Many early efforts in living-cell SMFS study are focused on investigating the molecular basis of cell adhesion, including the interactions of P-selectin/PSGL-1, csA/csA (a glycoprotein as a prototype of cell-adhesion proteins).[36,37] This SMFS approach is still being widely used in the recent studies of adhesion molecules.[38,39]

Yang et al. has applied SMFS to study the conformation change of the membrane protein, macrophage differentiation antigen associated with complement three receptor function (Mac-1).[40] Mac-1 is a member of an integrin subfamily of CAMs. It has an important feature of changing ligand-binding activity in response to cellular signaling events, which enables its proper execution of different physiological functions. By measuring the single-molecule force of Mac-1/ICAM-1, it was demonstrated that after Mn2+ activation there was an increase in the ligand-binding affinity of Mac-1. This is due to the different conformations of Mac-1 between its resting and active states. They further performed dynamic force spectroscopy study of several Mac-1 mutants, which were designed to stabilize the 'open' or 'closed' conformation of the αI domain to analyze the molecular structure of the activated Mac-1. The results implicated that activation of Mac-1 was governed by the downward movement of the α7 helix in its I domain.

Atorvastatin is one of the most potent drugs for atherosclerosis, and recently it has been found to have anti-inflammatory effects. Li et al. used SMFS to investigate the effect of atorvastatin on adhesion force between ICAM-1 and CD11b, which plays an important role in inflammatory responses in the disease of atherosclerosis. The result suggested that atorvastatin did not affect ICAM-1 and CD11b interactions. This was different from the ICAM-1 monoclonal antibody, which could directly reduce the binding force of ICAM-1 and CD11b. Further results revealed that atorvastatin pretreatment decreased the TNF-α-induced ICAM-1 expression, which may contribute to its anti-inflammatory effect. The study provides a new approach for investigating anti-inflammatory mechanisms of clinical drugs.[41]

Signal Transduction

Signal transduction, a basic biological process that allows cells to perceive and respond to their microenvironments, governs cellular activities and cell actions. It has been realized that many diseases including cancers are a result of malfunctions of cell signaling pathways, such as those regulated by growth factors. This has led to intensive research on the understanding of molecular mechanism of growth factor signaling and the development of therapies based on the interception of signaling in diseased cells. Growth factor signaling is initiated by the binding of extracellular ligands (growth factors) to specific cell-surface receptors. SMFS offers a new tool for the study of the formation of ligand–receptor signaling complexes, which is the key initial step for triggering cellular signaling cascades.

Yu et al. reported the use of SMFS combined with a fluorescence microscope to investigate the interactions between TGF-β1 and its receptors.[42] The ligand TGF-β1 was immobilized to the AFM tip, whereas the receptors, which were tagged with different fluorescent proteins, were expressed on the cells. Guided by the fluorescence images, the AFM tip was moved onto the cell expressing either one or two types of TGF-β receptors to perform the measurement. Normally, there are three TGF-β receptors, type I, II and III receptors, which are involved in TGF-β1 signaling. TGF-β1 binds firstly to its specific receptor TβRII, which then recruits TβRI with the help of TβRIII. Yu et al. found that the interaction force of TGF-β1 with the TβRI–TβRII complex was stronger than that of TGF-β1 with TβRI alone. The results from single-molecule dynamic force spectroscopy also showed that the binding of TGF-β1 with its receptors was more stable after TβRI was recruited. This was the first time that the cooperative binding of TGF-β1 receptors with the ligand was demonstrated by SMFS in living cells. The formation of the stable signaling complex TGF-β1–TβRI/TβRII is advantageous for the TGF-β1 signal transduction process.

Based on the measurement of the TGF-β1–TβRII interaction force, Yang et al. investigated the inhibition mechanism of naringenin (4',5',7'-trihydroxyflavanone) on TGF-β signaling in living cells.[43] Naringenin is a natural predominant flavanone that has the ability to reduce Smad 3 phosphorylation and expression in the presence of TGF-β1. The structure of naringenin is very different from the existing small-molecule antagonists of TGF-β signaling and the mechanism of its inhibition is unclear. Yang et al. found by SMFS that naringenin reduced the binding probability of TGF-β1 to its specific receptor TβRII, and thus inhibited the receptor dimerization and activation for the signaling complex formation (Figure 2). Naringenin and the chemical compounds with similar structures have the potential to be developed as a new class of small-molecule inhibitors of TGF-β signaling.

Figure 2.

The inhibition effect of naringenin on TGF-β1 signaling. (A) Schematic diagram of single-molecule force measurement in living cells with a TGF-β1-modified AFM tip in the absence and presence of naringenin. The tip was positioned directly above a cell expressing the desired TGF-β receptors. (B) The binding probability of TGF-β1 and TβRII before and after 50 or 100 µM naringenin treatment with R1B cells. If the AFM tip was blocked by introducing a solution of TβRII extracellular domain (5 µg/ml), the rupture peak was hardly detected and binding probability reduced dramatically.
AFM: Atomic force spectroscopy.
Reproduced with permission from [43].

SMFS could be a new approach to study the mechanisms of clinical drugs. Wildling et al. used SMFS to probe single molecular interactions between the serotonin transporter, which is the action site of antidepressants and amphetamines, and MFZ2-12 (a potent cocaine analog) in living CHOK1 cells.[44] It provided a useful framework for the further exploration of antidepressant binding. The antitumor drug herceptin is a monoclonal antibody against HER2 (a member of ErbB family), which is overexpressed in many cancers. Although herceptin has been actively used for the clinical treatment of women with HER2-overexpressing breast cancers, whether it functions through blocking the HER2 heterodimerization with other ErbB receptors is under debate because herceptin does not directly bind to the extracellular domain of HER2 where its dimerization arm is located. Shi et al. applied SMFS to study the effect of herceptin on heregulin β1 (HRG)–HER3 ligand–receptor interaction in living cells (Figure 3).[45] The results demonstrated that the presence of HER2 resulted in a more stable binding of HRG to HER3. The HER2-modulated enhancement in HRG–HER3 binding was inhibited by herceptin. Single-molecule dynamic force spectrum also confirmed that herceptin did exhibit a negative effect on the binding of HRG and HER2/HER3 complex. Different from the dynamic force spectrum of HRG–HER3/HER2 complex in the absence of herceptin, the force spectrum of HRG–HER3/HER2 in the presence of herceptin only displayed one linear region, and it was almost overlapped with that of HRG–HER3. This work provides the first experimental evidence confirming the inhibition effect of herceptin on HER2–HER3 interaction for ErbB signaling.

Figure 3.

The most probable forces of HRG–HER3 and HRG–HER3/HER2 under different conditions. Forces were measured at different loading rates for HRG-modified tips and the cells expressing only HER3 (circle) showed that the force increased linearly with the logarithm of the loading rate. In the force spectrum obtained with HRG-modified tips and the cells coexpressing HER2 and HER3 (triangle), there were two linear regions. In the presence of herceptin (square), the spectrum was similar to those measured on the cells expressing only HER3.
Reproduced with permission from [45].

Receptor Distribution

The identification and localization of specific receptors on the multicomponent heterogeneous cell membranes is of particular interest in the study of cell communication and cell function. However, the direct AFM imaging of individual receptors in living cells is difficult because of the small size and dynamic nature of membrane receptors. Using the AFM tip that functionalized with ligands, AFM could be used to map the distribution of complementary receptors on cellular surfaces. AFM force maps are 2D arrays of parameters derived from single force plots that are obtained by probing the surfaces in a raster-like fashion. Lee et al. used the anti-VEGF receptor 2 (VEGFR2) functionalized probe to image individual VEGFR and quantified the number and distribution of VEGFR.[46] Moreover, with the combination of molecular force spectroscopy of putative receptors (for koff ) and real-time image acquisition during competitive binding (for kon ), the binding kinetics were measured on an individual cell. Roduit et al. has studied glycosylphosphatidylinositol (GPI)-anchored proteins, which are involved in membrane trafficking and cell signaling under both physiological and pathological conditions, and are known to partition preferentially into cholesterol-rich microdomains.[47] They demonstrated that these GPI-anchored proteins resided within the domains that were stiffer than the surrounding membrane. By contrast, membrane domains containing the transferrin receptor, which does not associate with cholesterol-rich regions, do not display such a feature. The heightened stiffness of GPI domains is consistent with the existing data on the specific condensation of lipids. This may forge the way to unveiling the links between membrane stiffness, molecular diffusion and signaling activation.

The combination of SMFS with high-resolution AFM topographical imaging leads to the development of a pioneering dynamic recognition imaging tool, which was called 'simultaneous Topography and Recognition imaging' (TREC).[48] This technique is becoming an indispensable tool for high-resolution receptor mapping as it has been successfully demonstrated on different biomolecular model systems. It offers unprecedented possibilities for imaging the distribution of single molecules on cell surfaces.

The first TREC studies on cells were conducted with microvascular endothelial cells from mouse myocardium (MyEnd) to identify vascular endothelial (VE)–cadherin binding sites. In these cells, the adhesions play an important role to resist the hydrodynamic forces created by blood flow or blood pressure. VE–cadherin is strictly located at intercellular junctions of basically all types of the endothelium and plays a critical role in calcium-dependent homophilic cell-to-cell adhesion.[49] In this study, the appropriate fixation procedure was found to not only prevent the lateral mobility of VE–cadherin, but also to resolve topographical and recognition features with high-lateral resolution of 5–10 nm. Of course, there are still some limitations of TREC on living cells. The application of TREC on cells could be restricted to the local roughness of the cell membranes. Time resolution is also the major problem. The realization of the TREC approach in live cell studies represents a more demanding task.[48]

Force Measurements for Mechanical Properties of Living Cells

Without the specific ligand modification of the AFM tip, the F-D curves obtained after tip indentation on a cell could be used to quantify and map the mechanical properties of cells. Cellular mechanics is of prime importance in many cellular processes (e.g., cell growth and division) and disease development (e.g., cancers). AFM technique makes it possible to probe micro- and nano-mechanical properties of a single cell. Recently, a new technique, PeakForce quantitative nanomechanical mapping, has been developed in that the AFM tip is oscillated at a frequency (for example 1 or 2 kHz) far below the resonance frequency and force curves are generated each time the AFM tip taps on the sample surface. This allows simultaneously fast mapping of both topography and nanomechanical properties. It has a high potential for applications to biology and medicine.[50]

Up to now, much effort has been made to measure the mechanical properties of different types of cells.[22] Although cells exhibit both solid- and liquid-like features, in other words, they are viscoelastic, it is still a huge challenge to get a rigorous description of the cell mechanics as their viscoelastic nature is too complex.[51,52] Thus, most studies probing the stiffness or elastic modulus of cells assume that the cell is an elastic body and several theories have been developed to describe AFM tip-induced elastic deformation, such as those models proposed by Hertz, Johnson–Kendall–Roberts and Derjaguin–Muller–Toporov[23] . Here, the authors will emphasize on the study of living cell elasticity in biology and medicine, especially for the early detection of diseases.[11–14,20,53–61]

As matrix stiffness strongly influences the growth, differentiation and function of adherent cells, it is very important to understand how changes in matrix stiffness contribute to cellular biophysical properties. Recently, Shi et al. applied AFM to study the elastic moduli of cardiac myocytes and fibroblasts that were cultured on collagen-coated polyacrylamide substrates with gradient rigidity.[61] These two typical cardiac cells displayed different responses in elasticity change when the substrate stiffness was altered. While cardiac myocytes showed no evident change in elasticity on different substrates, cardiac fibroblasts displayed a non-monotonic dependence on substrate stiffness with a maximum elastic modulus. Moreover, the elasticity change of cardiac fibroblasts with substrate stiffness was found to be regulated by actin filaments. Study of the effect of substrate stiffness on cell elasticity for different cardiac cells provides new information for the better understanding of cardiac physiology and pathology.

The AFM measurement of cell mechanical properties has opened the possibility for developing a new strategy for the diagnostics of different pathologies at the single-cell level. Vatner et al. showed that aortic vascular smooth muscle cells from old male monkeys had a higher intrinsic stiffness compared with young monkeys. In contrast to current concepts that increased vascular stiffness of aging is simply attributable to extracellular matrix changes, primarily collagen and elastin, their study demonstrates that intrinsic mechanisms in vascular smooth muscle cells contribute to stiffness as well.[54]

The pathological changes in osteoarthritis start at the molecular level and then spread to the higher levels of the architecture of articular cartilage to cause progressive and irreversible structural and functional damage. At present, there is no treatment to cure or attenuate the degradation of cartilage. Early detection and the ability to monitor the progression of osteoarthritis are therefore important for developing effective therapies. Stolz et al. used AFM to monitor age-related morphological and biomechanical changes in the hips of normal and osteoarthritic mice.[56] Early damage in the cartilage of osteoarthritic human patients undergoing hip or knee replacements could similarly be detected (Figure 4). Changes in AFM elasticity measurement due to aging and osteoarthritis are clearly depicted before the morphological changes can be observed by current diagnostic methods. This may potentially be developed into a minimally invasive arthroscopic tool to diagnose the early onset of osteoarthritis in situ.

Figure 4.

Measurement of cell elasticity for seven osteoarthritis patients (62–96 years old) and three individuals without osteoarthritis (38, 41 and 50 years old). Each of the four linear fits represents a particular outer bridge grade. For all grades, the stiffness increased with age, possibly accompanied by reduced glycosaminoglycan content (horizontal arrow). As shown by the slopes, incremental reduction of stiffness by osteoarthritis as a function of age is greatest between grade 0 and 1, and decreases with progression to grade 2 and 3. This is correlated with the increased tendency of collagen mesh work disruption (vertical arrow).
Reproduced with permission from [56].

Early diagnosis of cancer, the leading cause of death worldwide, is of particular importance in current medical practice. In recent years, it has been demonstrated that cancer progression is accompanied by alterations in cell mechanical properties.

The AFM-based recognition of cancer cells has been successfully used to distinguish many cancer types,[57,62] and a few reports have been applied to the clinical samples.[58,59,63] Cross et al. reported that the stiffness of metastatic cancer cells was substantially 70% lower than that of healthy cells using the cells from the body fluid of patients with suspected lung, breast and pancreas cancers. The results suggested that mechanical analysis could distinguish cancerous cells from normal ones even when they showed similar morphology.[59,63] Lekka et al. studied the elastic properties of breast and prostate cancer cell lines and found that cells that originated from advanced stages of cancer had higher deformability. They further applied AFM to the detection of cancer cells in tissue slices taken from patients with various cancers.[60]

There are still some technical challenges when using AFM to measure cell mechanical properties for cancer diagnosis. Firstly, many factors such as tip geometry, indentation depth and loading frequencies could contribute to the uncertainties in the measurement of elastic modulus. It is important to standardize the measurements of patients' sample. Secondly, up to now, the Hertz model has been used in most of the evaluations of Young's modulus of cells. This model assumes the sample has a flat surface with infinitive thickness and the adhesion force is neglected, which is different from the cell surface. Thirdly, tissue sections that come from solid tumors are particularly complex. Despite the needs for further technique development, the use of the Young's modulus as an indicator of cancer stage in a patients' sample has been realized in several reports, giving hope to work out a new tool that can detect subtle changes in cell mechanic properties for effective and accurate identification of cancer.