2.1.

Principles of SP-Based Spectroscopy

The SP is the surface electromagnetic wave that travels along a metal-dielectric (gold and cells in our case) interface and decays exponentially in the direction perpendicular to the interface.¹⁹ It is usually excited in the attenuated total reflection regime using a prism coupler (Fig. 1). The SP wavevector $k_{\mathrm{sp}}$ is determined by the complex dielectric permittivity of gold, ${\epsilon }_{\mathrm{Au}}$, and cells, ${\epsilon }_{\text{cell}}={(n_{\text{cell}}+i{\kappa }_{\text{cell}})}^2$. The propagation component of the SP wavevector is given by $k{(x)}_{\mathrm{sp}}=k_0{({\epsilon }_{\mathrm{Au}}{\epsilon }_{\text{cell}}/{\epsilon }_{\mathrm{Au}}+{\epsilon }_{\text{cell}})}^{1/2}$. The SP is excited at a certain angle/wavelength at which the real part of its wavevector matches the $x$-component of the incident light wavevector, $k_0$: $k^{\prime }{(x)}_{\mathrm{sp}}=n_{\text{prism}}{k}_0\sin \theta$. Here, $n_{\text{prism}}$ is the prism refractive index and $\theta$ is the angle of incidence (Fig. 1). The resonant condition is given by

Fig. 1

A double-chamber flow cell for Fourier Transform Infrared-Surface Plasmon Resonance (FTIR-SPR) measurements of living cells. The cell sample is grown homogeneously on the prism surface and is divided into two halves with a separating barrier only at the measuring stage. One chamber undergoes treatment with some stimulus while another chamber serves as control. The infrared beam is reflected from one half of the prism base and its intensity is measured. This yields information on the refractive index of the cell layer in the corresponding chamber. After each measurement, the prism is rotated 180 deg around the vertical axis, and reflectivity from the other chambers is measured (control). Then the prism is rotated $-180\mathrm{deg}$ backward to the initial position and the measurement is repeated. The minimal temporal resolution of our measurements is $\sim 1.5\min$ (40 s for one-channel measurement and 20 s for switching between the channels).

Since ${\epsilon }_{\text{cell}}$ and ${\epsilon }_{\mathrm{Au}}$ are wavelength-dependent, the resonant condition singles out a certain wavelength at which the reflectivity achieves the minimum (Fig. 2). The dielectric permittivity of the cells ${\epsilon }_{\text{cell}}$ at the wavelength corresponding to the SPR minimum can be calculated from the dielectric permittivity of the gold ${\epsilon }_{\mathrm{Au}}$, refractive index of the prism $n_{\text{prism}}$, and the angle of incidence $\theta$. This fact is the basis underlying the SPR spectroscopy—by measuring the SPR minima wavelength one finds $n_{\text{cell}}$, the refractive index of the cells adjacent to the gold.^8,20

Fig. 2

Measured infrared reflectivity spectra (black dots) and fitting using Fresnel formulae for reflectivity from a multilayer (red solid line): (a) ZnS prism/gold/cells, (b) ZnS prism/gold/water. The nominal gold film thickness is 18 nm and the angle of incidence is 35 deg. The surface plasmon (SP) and TM-waveguide mode resonances are shown by arrows.

2.2.

Studying Living Cells with an Infrared SP Wave

In the context of living cell probing by the SP wave, there are two important length scales: the SP lateral propagation length and the SP penetration depth. In the infrared range ($\lambda =0.75$ to 2.7 μm in our case), the SP propagation length is $L_x=50$ to 150 μm (Refs. 8 and 12) and it is bigger than the lateral typical cell size of 20 μm. Hence, in the confluent cell monolayer, the SP wave travels across a few cells and probes the effective refractive index of the cell layer which is an average over several cells. When the cell layer is not fully confluent, the layer probed by the SP wave consists of regions with different refractive indices (cells and growth medium). If the size of these regions is less than the SP propagation length, a single SPR corresponding to the mean refractive index of the mixture appears.⁸ In the opposite case, two separate SPRs appear, corresponding to refractive indices of media and cells.

Another important length scale is the SP penetration depth ${\delta }_z$, namely the decay length of the SP electric field in the direction normal to the interface.⁸ Due to its evanescent nature, the SP wave probes the effective refractive index of the cells which is a weighted average in the direction normal to the interface

The SP penetration depth is growing with a wavelength^8,12 reaching ${\delta }_z=3\mu \text{m}$ in the infrared range, at $\lambda =2.6\mu \text{m}$. Working in the infrared range is beneficial for living cells studies. Indeed, with such a long penetration depth, the SP probes most of the cell volume. This is in contrast to the visible range where the typical SP penetration depth is much smaller, ${\delta }_z\sim 0.1\mu \text{m}$. The infrared SP wave yields information about (a) the basal part of the cells, which represents the quality of the cell–substrate attachment and (b) the apical part of the cells—complete information about the quality of the cell–cell attachment, including all kinds of junctions: adherence (responsible for cell–cell adhesion) and tight junctions (responsible for paracellular permeability).

Cells are nonhomogeneous objects that consist of water and organic content. The real part of the cell refractive index can be written as

Fig. 3

Dynamics of the real part of the IEC-6 cell refractive index $n_{\text{cell}}$ (a) and average cell height (b) measured by double-chamber infrared SP biosensor. The temporal resolution of these measurements is 3 min. Cells (${1\times 10}^6$) were seeded on the ZnS prism and cultured for 48 h. Both chambers were filled with growth medium (red line—first chamber and black line—second chamber). The difference between the refractive indices of the cells and water at SPR wavelength (2.55 μm) is presented. The variation in $n_{\text{cell}}$ and cell height during the course of experiment is associated with the cell adaptation, polarization, proliferation, and growth. The cell refractive index difference between two chambers does not exceed ${3\times 10}^{-4}$ RIU during the whole experiment, while the average cell height difference between the chambers does not exceed 0.1 μm.

Fig. 4

Short-term dynamics of IEC-18 cells refractive index $n_{\text{cell}}$ (a) and average cell height (b) in response to different concentrations of extracellular ${\mathrm{Ca}}^{2+}$. The latter was induced by the addition of three different sequentially administered concentrations of ethylenediaminetetraacetic acid (EDTA). Red line corresponds to the chamber with low ${\mathrm{Ca}}^{2+}$ (treatment) and black line corresponds to the chamber with normal concentration of ${\mathrm{Ca}}^{2+}$ (control). Duration of every EDTA treatment is 40 min. The temporal resolution of these measurements is 3 min. Inset: The difference between cell refractive index in the treatment chamber and that in the control chamber, $\mathrm{\Delta }{n}_{\text{cell}}$, versus EDTA concentration in growth medium, $C_{\mathrm{EDTA}}$ (black squares). Red line represents the quadratic dependence $\mathrm{\Delta }{n}_{\text{cell}}\sim C_{\mathrm{EDTA}}^2$.

2.3.

Studying Cell Monolayer with Waveguide Resonances

An epithelial cell monolayer cultured on a metal-coated dielectric prism and immersed into buffer medium represents a dielectric multilayer with progressively decreasing refractive indices. Such multilayer supports propagation of leaky waves (waveguide modes). The waveguide mode propagates inside the cell layer and its resonant wavelength sensitively depends on the cell layer height. Indeed, the waveguide resonance occurs when the total phase shift for round-trip propagation through the dielectric multilayer is a multiple of $2\pi$. For $p$-polarized incident light, these relations yield a set of resonant wavelengths corresponding to ${\mathrm{TM}}_1$, ${\mathrm{TM}}_{2,\dots }$ modes,

Together, the SPR, which probes the average refractive index of the cells, and the waveguide mode resonance, which probes the cell height and cell–cell attachment, make the basis of a very informative label-free sensor of cell activity.

3.1.

FTIR-SPR Experimental Setup and Measurement Protocol

The infrared multiwavelength beam is emitted from the external port of the FTIR spectrometer (Bruker Equinox 55).^8,12 The collimated beam passes through a polarizer mounted on the computer-controlled motorized rotating stage. The polarized beam is reflected from the Au-coated right-angle ZnS prism (${20\times 40\mathrm{mm}}^2$ base, ISP Optics, Inc., Irvington, New York), on which a monolayer of cells is grown above the gold layer. The infrared beam is then focused onto a liquid-nitrogen-cooled MCT detector. The $p$-polarized beam serves for SP excitation, while the $s$-polarized beam is used as a background. Each single measurement consists of recording the $p$- and $s$-reflectivity spectra (that takes approximately 40 s) and represents an average of eight scans with a $4{\mathrm{cm}}^{-1}$ resolution.

The biosensor chip for living cell measurements is shown in Fig. 1. The ZnS prism with an 18-nm thick Au coating is attached hermetically to the double-chamber flow cell which is then mounted on the computer-controlled motorized rotating stage. Each chamber has a volume of about $\sim 1\mathrm{ml}$ and is $\sim 5\mathrm{mm}$ thick. The flow cell is equipped with a cooler/heater system operated by a temperature controller that maintains the same temperature in both chambers. The medium flow is controlled independently in each chamber by the double-syringe pump (NE-4000, New Era Pump Systems, Inc., Farmingdale, New York). The cells are cultured homogeneously on the prism base and form a tight monolayer. This monolayer is divided into two parts by the barrier of the double-chamber flow cell. This is done only at the measuring stage, 48 h after cell seeding.

The infrared reflectivity spectra were measured sequentially from the test and control chambers. The collimated and polarized infrared beams are incident upon one half of the prism base (Fig. 1) in order to measure the reflectivity from one chamber. After every measurement, the prism with the flow cell is rotated 180 deg around the vertical axis in such a way that the infrared beam is now incident upon the other prism half. The measurement from the sample in the second chamber is taken (control measurement) and then the prism is rotated $-180\mathrm{deg}$ backward to the initial position. Comparison between measurements from the two chambers yields direct information about stimulus-induced changes in the cells without undesired background resulting from the sample’s cellular status.

3.2.

Analysis of the SPR Spectra

To analyze the SPR spectra, we used a fitting procedure based on the Fresnel optical model for reflectivity from a multilayer.²³ The simulation algorithm consists of two parts. The first part involves two calibration measurements with two known analytes: pure water (optical constants taken from Ref. 26) and air ($n_a=1$, ${\kappa }_a=0$). The measured spectrum (black circles) and the simulated curve (red solid line) for a prism/gold/water trilayer are shown in Fig. 2(b). This part yields the optical constants of the Au film (which can differ from those of bulk gold²⁷), beam divergence, and exact values of incident angle and thickness of the Au film. The optical constants of gold were simulated using the Lorentz–Drude model, while the starting parameters for the fitting procedure were taken from Ref. 28. The second part of the analysis uses the Fresnel equations and parameters from the first part to find the optical constants ($n_d$, ${\kappa }_d$) of the cells. The measured reflectivity spectrum (black circles) and the simulated curve (red solid line) for the prism/gold/cells trilayer are shown in Fig. 2(a).

3.3.

Cell Culture, Measurements and ${\mathrm{Ca}}^{2+}$ Treatment

4.1.

Test Measurements

To test the operation of our biosensor, we conducted a trial experiment with confluent IEC-6 cells when both chambers were filled with plain growth medium under constant flow and no stimulus was applied. The purpose of this experiment was to check to what extent the cell monolayer behavior was identical in both chambers during a 24-h long measurement. The temporal resolution of these measurements was 3 min.

Figure 3 shows that at the beginning of the measurement, the refractive index of the cells in both chambers is the same, indicating that the cells grew homogeneously on the entire surface of the prism. In the course of time, the refractive index of the cells slowly changes, first due to adaptation following the buffer replacement ($t=0$ to 400 min), and then due to progressive cell polarization, growth, and proliferation (after $t=400\min$). The dynamics of cell height as measured by the wavelength of the TM waveguide resonance [Fig. 3(b)] corroborates this explanation. During the first 400 min, the cell height increases from 6.5 to 9 μm, indicating recovery after the minor shock associated with the buffer replacement. Afterwards, the cell height continues to grow, but more slowly, indicating progressive cell polarization. Notably, Fig. 3 shows that the cell height and the cell refractive index are the same in both chambers during the whole course of measurement. Such experiments were carried out with different cell lines: IEC-6 (Fig. 3) and Caco-2 (data not shown) and different culturing times of the cells on the prism (24 to 72 h). In all these experiments, we obtained consistent results.

Although the overall refractive index variation through the 24 h of measurements is ${\sim 3\times 10}^{-3}$ refractive index unit (RIU), the difference between the two chambers does not exceed ${\sim 3\times 10}^{-4}$ RIU. This means a 10-fold increase in sensitivity with respect to the single-chamber measurement. Note that the measurement uncertainty associated with the physical sources (temperature instability and its effect on the refractive index, angular misalignment arising from rotation of the prism, polarization instability of the infrared source and discrete noise associated with the spectral resolution of the apparatus) is much smaller, ${\sim 10}^{-5}$ RIU. Hence, the dominant source of the measurement variability in such an experiment is biological and not instrumental. With respect to cell height, the overall increase during 24 h is $\sim 3\mu \text{m}$, while the difference between the two chambers is $<0.1\mu \text{m}$.

4.2.

${\mathrm{Ca}}^{2+}$ Depletion Experiments

To demonstrate the advantages of the double-chamber protocol over the single-chamber one, we utilized a well-documented cell phenomenon which has been intensively studied by conventional fluorescent labeling, and we measured its kinetics in real time using a label-free SP technique. In particular, we analyzed the response of a fully confluent IEC-18 cell monolayer to different concentrations of extracellular ${\mathrm{Ca}}^{2+}$ in the bathing medium. The latter was altered by the addition of different concentrations of the EDTA chelating agent into the growth medium which contained normal ${\mathrm{Ca}}^{2+}$ levels (1.8 mM). Extracellular ${\mathrm{Ca}}^{2+}$ plays a crucial role in triggering, assembly, and sealing of adherence^29,30 and tight junctions^31–33 that are located at the apical part of the cell membrane. The structural integrity and functional polarity of epithelial cells require both cell–substrate and cell–cell adhesions, which give rise to physical and signaling stimuli for the initiation of cell polarization.³⁴ E-cadherin, which is a ${\mathrm{Ca}}^{2+}$-dependent protein, is an important component of adherence junctions in epithelial cells. Deletion of E-cadherin by siRNA methodology leads to inhibition of cell polarity establishment.³⁵ Tight junctions are protein complexes located at the cell–cell interface and are more apical than adherence junctions. They are responsible for maintaining the epithelial layer impermeable (or selectively permeable). In addition, they also play an important role in the development of epithelial cell polarity.^33,36 Previously, it was reported that cells which were exposed to low-calcium growth medium did not develop tight junctions, yet, when ${\mathrm{Ca}}^{2+}$ was added, tight junctions evolved, and cell polarization developed.^17,18 Our aim in this experiment was to track the kinetics of the morphological changes in a cell monolayer under controlled variations in the ${\mathrm{Ca}}^{2+}$ level in growth medium and to compare them with the morphological changes of identical cells under normal ${\mathrm{Ca}}^{2+}$ concentration.

In order to investigate morphological changes of a confluent cell monolayer induced by different extracellular ${\mathrm{Ca}}^{2+}$ concentrations, we conducted two kinds of experiments. In the first one, we studied a short-time exposure of the cell monolayer to the growth medium supplemented with different EDTA concentrations (40 min for each concentration) (Fig. 4). The first two replacements of the treatment medium (1.34 and 2.68 mM of EDTA) reduced the concentration of extracellular ${\mathrm{Ca}}^{2+}$ but did not deplete the growth medium completely from ${\mathrm{Ca}}^{2+}$. In contrast, the third replacement (5.36 mM of EDTA) caused a total depletion of ${\mathrm{Ca}}^{2+}$ in the medium. Figure 4 shows that the first medium replacement (1.34 mM EDTA) hardly affected the refractive index of the cells, which means that the adherence and tight junctions did not disassemble and the cells did not undergo significant morphological changes. At the second stage (2.68 mM of EDTA), the refractive index of the cells decreased slightly. We associate this decrease with the beginning of junctions opening and the decreasing height of the circumferential ring of actin filaments.^16,37 The third step (5.36 mM of EDTA) induced a fast (few minutes) breakdown of adherence and tight junctions, resulting in rapid changes in cell morphology. This process was so fast that it produced wounds in the cell monolayer²⁵ which drastically decreased cell coverage of the gold. This resulted in the sharp drop of the refractive index of the cells at the beginning of this stage. Continuation of junctions opening produced cell relaxation and spreading on the gold surface. This resulted in the increase of the cell coverage of gold and a consequent increase of the effective cell refractive index. At the end of this stage, the growth medium was replaced with fresh normal growth medium (normal ${\mathrm{Ca}}^{2+}$ concentration—1.8 mM) in order to track the buildup of cellular junctions and cell polarization development. As shown in Fig. 4, the full recovery of the cell layer takes $\sim 1000\min$.

The inset to Fig. 4 shows that the magnitude of the transient change of the cell refractive index correlates nonlinearly to the EDTA concentration, $\mathrm{\Delta }{n}_{\text{cell}}\sim C_{\mathrm{EDTA}}^2$. The nonlinearity implies the threshold-like response of the cells to ${\mathrm{Ca}}^{2+}$ depletion: the cell layer can withstand and counterbalance small levels of ${\mathrm{Ca}}^{2+}$ depletion, but cannot withstand large levels of ${\mathrm{Ca}}^{2+}$ depletion.

The average cell height [Fig. 4(b)] shows that before treatment the cell height in both chambers was the same—6.5 μm. This is lower than the normal cell height; hence, the cells are not fully polarized. Indeed, the cell height in the control chamber slowly grows with time to 10 μm, indicating progressive cell polarization. The EDTA treatment of the measurement chamber reduces the cell height (the cell monolayer is less rigid since the cell–cell junctions have been injured). After the treatment, the cell height recovers, but it is still lower than that in the control chamber even after 20 h.

In the second experiment, we studied the effect of long-time exposure (500 and 300 min) of the cell monolayer to the growth medium supplemented with different concentrations of EDTA (Fig. 5). At the first step, we exposed the cell monolayer to the bathing medium containing 1.34 mM of EDTA for $\sim 500\min$. Figure 5 shows that the refractive index of the cells hardly changes under this treatment. We conclude that a minor decrease in extracellular ${\mathrm{Ca}}^{2+}$ concentration induced by this concentration of EDTA hardly affected the adherence and tight junctions and thus the integrity of the cell monolayer was kept intact. At the next stage, we replaced the growth medium with that containing a higher concentration of EDTA (2.68 mM) for $\sim 300\min$. This step caused a slow persistent decrease in the refractive index of the cells. We associate this with the fact that the decrease of extracellular ${\mathrm{Ca}}^{2+}$ concentration induced by enhanced concentration of EDTA slowly breaks cellular junctions and consequently decreases the height of the circumferential ring of actin filaments. At the beginning of this stage, a sharp drop in the cell refractive index was observed, which relaxed after $\sim 50\min$. This rapid change in the refractive index resulted most likely from the mechanical stress induced by the replacement of the growth medium. Notably, there was no change in the cell refractive index in the control chamber, though the medium in this chamber was similarly withdrawn and inserted back in order to simulate equal conditions in both chambers. Although exposure of the cells to the medium with the reduced concentration of ${\mathrm{Ca}}^{2+}$ at the first stage of the experiment did not disassemble adherence and tight junctions, it weakened the cell–cell attachment, thus even a minor mechanical stress initiated monolayer integrity breakdown. Finally, when we increased the concentration of EDTA to 4.02 mM, the refractive index of the cell monolayer dropped dramatically. We associate this with a fast breakdown of adherence and tight junctions and loss of integrity of the cell monolayer.

Fig. 5

Dynamics of IEC-18 cells refractive index $n_{\text{cell}}$ (a) and average cell height (b) under prolonged exposure to different concentrations of EDTA. Red line corresponds to the chamber with low ${\mathrm{Ca}}^{2+}$ (treatment) and black line corresponds to the chamber with normal concentration of ${\mathrm{Ca}}^{2+}$ (control). Duration of the first stage (1.34 mM of EDTA) is 500 min. Duration of the second stage (2.68 mM of EDTA) is 300 min. Duration of the third stage (4.02 mM of EDTA) is 100 min. The temporal resolution of these measurements is 3 min.

The dynamics of the average cell height, as probed by the waveguide resonance [Fig. 5(b)], is consistent with this scenario. The cell layer can withstand small and intermediate ${\mathrm{Ca}}^{2+}$ depletion, but breaks down under severe ${\mathrm{Ca}}^{2+}$ depletion.

Figure 6 schematically summarizes the changes in cell monolayer morphology under ${\mathrm{Ca}}^{2+}$ depletion. At normal ${\mathrm{Ca}}^{2+}$ concentration, the cells form an intact monolayer and maintain healthy intercellular cell–cell junctions. Under low ${\mathrm{Ca}}^{2+}$ conditions, the cell–cell junctions are damaged, which leads to penetration of the growth medium into intercellular space. This process reduces the fraction of organic content $f$ [Eq. (3)] inside the layer sampled by the SP wave, resulting in a decrease in the cells’ monolayer refractive index.

Fig. 6

Schematic model for the cell monolayer response to ${\mathrm{Ca}}^{2+}$ depletion. ${\delta }_z$ indicates the SP sampling height. (a) In the presence of normal ${\mathrm{Ca}}^{2+}$ concentration (1.8 mM), the cells form an intact monolayer and maintain healthy intercellular cell–cell junctions. (b) Under low ${\mathrm{Ca}}^{2+}$ conditions, cell–cell junctions are progressively damaged, leading to penetration of growth medium with the low refractive index into intercellular space. This results in the decrease in the cell monolayer refractive index as sampled by the SP wave. The average cell height under ${\mathrm{Ca}}^{2+}$ depletion drops.