Plasmonic imaging of subcellular electromechanical deformation in mammalian cells

Abstract. A membrane potential change in cells is accompanied with mechanical deformation. This electromechanical response can play a significant role in regulating action potential in neurons and in controlling voltage-gated ion channels. However, measuring this subtle deformation in mammalian cells has been a difficult task. We show a plasmonic imaging method to image mechanical deformation in single cells upon a change in the membrane potential. Using this method, we have studied the electromechanical response in mammalian cells and have observed the local deformation within the cells that are associated with cell–substrate interactions. By analyzing frequency dependence of the response, we have further examined the electromechanical deformation in terms of mechanical properties of cytoplasm and cytoskeleton. We demonstrate a plasmonic imaging approach to quantify the electromechanical responses of single mammalian cells and determine local variability related to cell–substrate interactions.


Introduction
Electricity is a fundamental driving force in biological systems. 1 It often acts in terms of a mechanical response. [2][3][4][5] Important examples include electrically regulated beating of heart cells, 6,7 auditory sensing of hair cells, [7][8][9] and conformational changes in voltage-gated ion channels. 10,11 This electromechanical coupling effect is anticipated from thermodynamics, e.g., the Lippmann equation, which relates electrical potential-induced surface tension to surface charge density. However, the underlying molecular scale mechanism, especially in biological systems, is still largely unknown. An experimental challenge in studying this effect is that the electromechanical deformation at the single cell level is on the subnanometer to nanometer scale. 4,12,13 Biologically, the cell tension is subject to many factors, such as the change of cell cycles, [14][15][16][17] cell migration status, 18 cell types, [19][20][21] pharmaceutical environment, 22 and the number may vary for more than 10 times. Atomic force microscopy (AFM) can probe the mechanical deformation, 12,13,[23][24][25] but it involves the contact of the AFM tip to the cells. AFM is also limited to the study of the top surface of cells and is unable to directly measure the bottom surface of cells, as well as the cell-substrate interactions. Newly developed technologies, including the piezoelectric nanoribbon transducer, 26 and optical methods, such as spectral-domain optical coherence phase microscopy (SD-OCT), 27 have been applied to mechanical deformation measurement. They either measure one location at a given time or the whole cell, which is insufficient to resolve subcellular responses. Quantitative phase microscopy provides subcellular imaging resolution, but the subtle potential-induced mechanical deformation at subcellular level has not been studied. 28 Here we have shown a plasmonic imaging method to quantify the subnanometer cell deformation at cell-substrate interface. We studied the cell electromechanical responses and quantified the subcellular mechanical deformation at various membrane potential polarization levels and frequencies. We further modeled the cell electromechanical responses and investigated the effect of mechanical perturbations on cytoskeletons.

Cell Culture
Wild-type HEK293T cells (ATCC, CRL-3216) were seeded on a fibronectin (Sigma, f1141)-coated gold film one day before the experiment. Cells were cultured in DMEM (Lonza) medium with 10% FBS (Life Technologies, 10437077) according to the user's manual from ATCC. After overnight incubation in 5% CO 2 -humidified atmosphere at 37°C, the cells attached to the sensor surface and were ready for experiment.

Electrophysiology
Cultured cells were recorded at room temperature in whole-cell configuration using Axopatch 200B system (Axon Instruments). Two Ag/AgCl electrodes were used for both patch electrode and bath electrode. Membrane potential was modulated sinusoidally from −60 mV to 140 mV in the voltage-clamp mode. Electrical current was simultaneously recorded. Electrical data were recorded at the sampling rate of 10,000. Glass micropipettes were pulled using a flaming puller (P-97, Sutter Instrument, California). Intracellular recording solution containing 10 mM NaCl, 135 mM K-gluconate, 10 mM HEPES, 2 mM MgCl 2 , 2 mM Mg-ATP, and 1 mM EGTA (pH 7.4) was injected to the tip of pulled micropipettes. Micropipettes with typical resistance of 3 to 10 MΩ was used for recording. The extracellular recording solution contains 135 mM NaCl, 5 mM KCl, 1.2 mM MgCl 2 , 5 mM HEPES, 2.5 mM CaCl 2 , and 10 mM glucose at pH 7.4. For the step sweep experiment, membrane potential was hyperpolarized or depolarized to levels of −80, −20, 40, and 100 mV from its resting potential of −60 mV for 10 ms (see Supplementary Material-supporting note 5 and Fig. 3(c), for more details). To depolymerize the actin structure, cells were incubated at 37°C for 30 min in 10 μM cytochalasin D-containing media before measurements. Control cells were maintained in the normal culture media. Experiments with and without CD-treated cells were performed in the normal recording media.

Plasmonic Imaging
An inverted microscope (IX81, Olympus) with a 60× NA 1.49 oil immersion objective was implemented in the plasmonic imaging system. P-polarized light beam from a 670-nm superluminescent light-emitting diode (SLD-26-HP, Superlum) was introduced into the microscope via a total internal reflection fluorescence tube lens (TIRF module, Olympus) and a 50∕50 beam splitter (Thorlabs). The reflected light from the gold surface was detected by a complementary metal-oxide semiconductor (CMOS) camera (Pike, Allied Vision) with the full resolution (640 px × 480 px) recording at 106.6 frames per second (fps). To reveal the spectral response of the cell, an ultrafast CMOS camera (Phantom V310, Vision Research) with a frame rate of 1000 fps was used. For step hyperpolarization and depolarization experiment, 10,000 fps was applied to match the electrical sampling rate. Cameras were externally triggered by the patch clamp voltage controller for synchronization. Sensor chips were made from BK-7 glass coverslips by evaporated coating of ∼1.5-nm chromium layer and a ∼47-nm gold layer on top. Chips were rinsed with deionized water and ethanol before cell plating. Cells were cultured in a removable Flexi-Perm (Sarstedt) silicone cell culture chamber placed on top of the chip surface.

Deformation Calculation
Raw plasmonic images were smoothed over 5 × 5 pixels (600 × 600 nm on a Pike camera) spatially to minimize pixel noise. Fast Fourier transform (FFT) was applied to the images along time. Amplitude and phase-shift images at the modulation frequency were obtained. Applied micropipette potential was used as a reference for phase-shift calculation. The Z-axis displacement was quantified by Eq. (1) from the FFT amplitude image. For potential step experiments, image sequences were aligned with polarization steps. Multiple cycles were run on each potential step repeatedly. Image sequences were averaged over multiple cycles to enhance the signal-to-noise ratio. Local deformation profile was plotted and calculated for each potential step with 5-ms moving average smoothed over time. Spectral response was fitted with linear equation in logarithm scale.
All postacquisition analyses were carried out using customwritten MATLAB scripts.

Results and Discussion
We quantify the cell deformation using plasmonic imaging. The imaging system was built on top of an inverted optical microscope. Incident light was coupled into the microscope via a high numerical aperture objective. Reflection light from a gold-coated cover glass is imaged by a CMOS camera [ Fig. 1(a)]. [29][30][31] We manually tuned the incident angle close to the surface plasmon resonance angle to generate an evanescent field near the gold surface (Supplementary Materialsupporting note 4). Since the evanescent field decays exponentially into the solution from the surface with a decay length of only about 100 nm, the plasmonic imaging is extremely sensitive to the mechanical deformation perpendicular to the surface (z-direction) (for detailed theoretical calculations, please refer to previous studies [32][33][34] ). The deformation can be quantified according to the following equation: E Q -T A R G E T ; t e m p : i n t r a l i n k -; e 0 0 1 ; 3 2 6 ; 5 4 2 where ΔI I is the relative intensity change after cell deforms; Δz is the displacement in z-direction; and L is the decay length of the evanescent wave on the surface, which is ∼100 nm in our current imaging setting (Supplementary Material-supporting note 1). 35,36 We cultured HEK293T cells on top of the gold sensing surface. The evanescent field interacts with the bottom part of the cell, and the deformation of the cell is recorded in the plasmonic images. To study the membrane potential-induced deformation, we controlled the potential across the cell membrane with a patch-clamp micropipette and simultaneously recorded the plasmonic images when the membrane potential was polarized. The corresponding polarization current was also recorded electrically. Since the deformation was extremely small, we applied a sinusoidal waveform to repeatedly measure this electromechanical response [ Fig. 1(b)]. FFT was performed on the recorded images over time [ Fig. 1(c)] and the noise at frequencies other than the one at the applied alternating potential was removed (Supplementary Material-supporting note 5). The FFT amplitude represents the magnitude of the intensity change to the applied potential. From the FFT amplitude, we further determined local cell deformation using Eq. (1), and from the FFT phase, we obtained the deformation direction with respect to the polarity of the applied potential.
Figures 2(a) and 2(b) show the bright field and plasmonic images of a cell. A glass micropipette was pressed onto the cell membrane, and the membrane patch inside the micropipette tip was ruptured after a tight glass-membrane seal was established at the contact area. In this configuration, the electrode in the micropipette is in direct contact with the intracellular fluid, allowing precise and flexible control of the potential across the cell membrane. Since micropipette is beyond the penetration depth of the surface evanescent field, it is invisible on the plasmonic images. We then applied an alternating depolarization potential from −60 to 140 mV at 37 Hz to the cell  the background region without cells. A sharp peak was observed at the frequency of applied potential in the whole cell region after FFT, indicating the mechanical deformation generated from the cell by the applied potential [ Fig. 2(f)]. In contrast, no obvious response was detected in the background region. From the FFT amplitude image, we observed a much bigger response at the cell edge than that from the center, suggesting a heterogeneous electromechanical deformation at different subcellular regions [ Fig. 2(d)].
To quantify the time delay of the electromechanical deformation, we analyzed the phase shift of the recorded images with respect to the applied potential. Only in-phase (0 deg) and antiphase (180 deg) responses were observed in the FFT phase image at 37 Hz [ Fig. 2(e)]. The cell center mainly displayed an in-phase (0 deg) response (colored as cyan), whereas the edge area was antiphase (180 deg, red). This result shows that the cell bottom deforms along with the applied potential, but cell center and edge deform in opposite directions. When the cell is depolarized, the center portion of the cell moves closer to the gold surface (intensity increase) while the edge moves away from the surface (intensity decrease) [ Fig. 1(b)]. This bidirectional response rules out the possibilities that the observed plasmonic signal was from other mechanisms (Supplementary Materialsupporting note 2).
To further study the relationship between mechanical deformation and membrane potential changes, we measured the mechanical responses at different potential polarization levels. The cells were electrically hyperpolarized or depolarized in potential steps from −80 to 100 mV at an interval of 60 mV per step [ Fig. 3(c)]. Plasmonic images were simultaneously captured at the same sampling rate as 10,000 fps (see Supplementary Material-experiments and supporting note 5, for more details) [ Fig. 3(a)]. This sequence was repeated and the plasmonic images were averaged over the repeated measurement for denoising (see methods). Figure 3(b) shows a differential image of cellular response of potential depolarization from −60 to 100 mV after averaging over 94 cycles of repeated measurements. The cell center showed an increased plasmonic intensity, whereas the cell edge showed an opposite response, which is consistent with the result presented in the previous section. The similar intensity outside the cell contour region was due to the noise from the background. To accurately quantify the dynamic response of the cell deformation, we plotted the deformation responses over the cell center and over the cell edge, respectively [ Fig. 3(d)]. Both regions deformed slowly after a sudden potential polarization. However, the exact response speed is biased due to the moving average used in the denoising scheme. At the steady state, deformation magnitude showed a linear dependency on the membrane polarization for both regions [ Fig. 3(e)], with an average displacement of about 0.02 nm per 100 mV at the cell edge and 0.005 nm per 100 mV at the cell center, respectively. These numbers are about one order of magnitude smaller than that of the mechanical motions tracked at the cell edge using digital microscopy. 4 This is probably due to the impeded motions at the cell-substrate interface of the adherent cells. The thermal energy-induced fluctuation is typically at the same level. However, these fluctuations were random, spanning across the whole spectrum. In our current detection scheme, we externally drive the motion. Different subcellular regions deform in a synchronous manner. Thus, the thermal fluctuation is suppressed by either averaging (as in Fig. 3) or limiting the detection bandwidth (as in Fig. 2).
Thermodynamics predicts that potential changes across the cell membrane will lead to a mechanical deformation. 12 The relationship between surface tension and potentials for a single interface is described as Lippmann equation: where γ is the surface tension, V is the membrane electrical potential, and q is the excess mobile charge in the double layer of the membrane surface. However, thermodynamics does not provide how much a cell will deform with a given change of membrane potential. The actual magnitude and phase of the deformation induced by the membrane surface tension is determined by the mechanical properties of the cells.
To explain the observed electromechanical response, as well as to accurately evaluate the response speed, we performed a spectral analysis on the deformation of single cells. We measured the cellular mechanical response over a wide frequency range from 7 to 312.5 Hz with a constant potential change (200 mV of depolarization from −60 mV resting potential) for each frequency. Figures 4(a)-4(e) show the responses at different modulation frequencies after FFT from the recorded plasmonic images. The amplitude is shown as the brightness and the phase-shift is encoded in color. We observed a near-in-phase and antiphase mechanical responses with respect to the applied potential modulation (0 deg and 180 deg only) over the frequencies. No other intermediate phases were detected.
The mechanical property of a cell can be modeled as E Q -T A R G E T ; t e m p : i n t r a l i n k -; e 0 0 3 ; 3 2 6 ; 7 4 1 where G is the modulus that describes the mechanical properties of the cells. 37 It is a combination of an elastic component, described by storage modulus (G 0 ), and a viscous component, or loss modulus (G 0 0 ) [ Fig. 4(f)]. The GðfÞ is examined by calculating the ratio between applied oscillatory force FðfÞ at frequency f and corresponding displacement xðfÞ, that is, FðfÞ∕ xðfÞ, in the frequency domain. Therefore, according to the FFT phase of the observed mechanical response [Figs. 4(a)-4(e)], the cell is mainly an elastic body, rather than a viscous one within the frequency range.
Electromechanical properties of the cell can be extracted by analyzing the relation between the displacement and the applied potential. We calculated the averaged deformation magnitude from both the cell center and the edge at different frequencies.  the same modulation potential, is given as jG 0 ðfÞj ∼ f β . This result is in good agreement with previous rheological measurements using intracellular microbead-or cytoskeleton-attached magnetic nanoparticle conducted on cells. 38,39 We anticipate that mechanical perturbation to the cellular structure will remarkably affect the electromechanical response and can be observed with plasmonic imaging (see Supplementary Material-supporting note 3, for full electromechanical model).
To confirm this, we treated the cells with an actin polymerization inhibitor, cytochalasin D, and examined the corresponding electromechanical responses. 40 Actin fibers are the main component of the cytoskeleton and determine the cell mechanics. 20,[41][42][43] After depolymerizing the actin frame structure, cell centers show an expanded anti-phase area (red) compared with those in the normal condition [Figs. 5(a) and 5(b), white dash lines]. The boundaries between the in-phase and the antiphase regions disappeared, and the in-phase regions appeared to be more random at different subcellular locations. To quantitatively examine the subcellular phase change, we quantified the average phase-shift over the whole cell area. A significant phase shift toward antiphase (180 deg) after cytochalasin D treatment was observed

Conclusion
We have demonstrated a plasmonic method for detecting subnanometer deformation from cell electromechanical coupling. This capability allows us to visualize membranepotential-induced cell mechanical deformation. At frequencies below hundreds of hertz, the deformation is mainly elastic. Deconstructing the actin structure revealed an important role of cytoskeleton in the observed potential-induced cell mechanical deformation. Our work shows that membrane potential-induced cell mechanical deformation can be imaged with the plasmonic method with subcellular spatial resolution and subnanometer resolution. We anticipate that the method can contribute to mechanobiological studies and shed light on the understanding of important biological functions, including mechanosensitive ion channels and protein configuration changes in voltage-gated ion channels.

Disclosures
The authors have no relevant financial interests in this article and no potential conflicts of interest to disclose.