2.1.

Experimental Setup

The sensing head for living cell measurements is shown in Fig. 1(a). The right-angle ZnS prism [$20\times 40{\mathrm{mm}}^2$ base (ISP Optics, Inc., Irvington, NY, US)] with Au coating (18 to 20 nm thickness) is attached to the 1.5-ml flow chamber equipped with the temperature controller ($\pm 0.1°\mathrm{C}$). All experiments with living cells were performed at 37°C. The medium flow in the chamber is enabled by the motorized bee syringe pump, and operated by a variable speed controller. The live cells were directly cultured on the Au-coated prism surface unless otherwise stated. Figure 1(b) and 1(c) shows two experimental setups used in this work. The angular interrogation experiments were performed in the $\theta -2\theta$ configuration¹⁹ [Fig. 1(b)] where the incident beam is fixed in space while the flow chamber and detector are rotated using two coupled rotating stages. For the kinetic measurements at fixed incident angle, the prism-flow chamber assembly was mounted on the vertical translation stage [Fig. 1(c)] whereas the cell layer on the Au-surface was simultaneously imaged using a conventional optical microscope.

Fig. 1

(a) The flow chamber for infrared reflection spectroscopy of living cells. The zoomed area illustrates the sensing region of the surface plasmon (SP) and guided modes. (b) $\theta -2\theta$ setup for angular-resolved measurements. (c) The fixed angle configuration is supplemented with optical microscopy.

2.2.

Infrared Reflectivity Measurements

The infrared beam is emitted from the external port of the FTIR spectrometer [Bruker IFS 55 Equinox FTIR], which is equipped with a tungsten lamp and KBr beamsplitter. The collimated and p-polarized infrared beam is reflected from the Au-coated prism and focused onto a liquid-nitrogen-cooled MCT (HgCdTe) detector. The measurement of a single spectrum takes approximately 25 s, and represents an average of 8 scans with $2\text{-}{\mathrm{cm}}^{-1}$ or $8\text{-}{\mathrm{cm}}^{-1}$ resolution. The $s$-polarized spectrum was used as a background. In kinetic measurements, the background $s$-polarized spectrum was measured only once, at $t=0$.

2.3.

Analysis of the Infrared Spectra

To convert the SP resonance wavelength into refractive index spectra we used an optical model based on Fresnel equations for tri-layer structure.²⁰ The procedure involves calibration measurement with known analyte (pure water) that yields the optical constants of the Au film. At the next step we used the same Fresnel equations to find the optical constants ($n_d$, ${\kappa }_d$) of the unknown analyte.²⁰

2.4.

Cell Culture and Preparation

Madin-Darby canine kidney cells (MDCK) were routinely cultured in minimal essential medium (MEM, Biological Industries, Beit Haemek, Israel), and supplemented with 5% (v/v) fetal calf serum (FCS) and 1% (v/v) antibiotics (Biological Industries). HeLa and human melanoma (MEL 1106) cells were cultured in Dolbecco’s modified Eagle’s medium (DMEM, Biological Industries, Beit Haemek, Israel), and supplemented with $4.5\mathrm{g}/\mathrm{L}$ D-glucose, 10% (v/v) FCS and 1% (v/v) antibiotics. During the measurements, the cells were bathed in a growth medium supplemented with 20 mM Hepes, pH 7.5.

Rat intestinal epithelial cells (IEC6) cells were maintained in DMEM (Sigma-Aldrich, St. Louis, MO, USA), and supplemented with 10% (v/v) FBS (Biological Industries, Beit Haemek, Israel) and 0.2% (v/v) penicillin-streptomycin-nystatin (BioLab LTD, Jerusalem Israel). The cells were cultured in 5% ${\mathrm{CO}}_2$ in a humidified atmosphere at 37°C. When the cells reached 80 to 90% coverage, the culture medium was removed, and 2.0 ml of Trypsin-EDTA (Biological Industries, Beit Haemek, Israel) solution was added. After the cell layer was dispersed (this occurs in 3 to 5 min), 4.0 ml of DMEM was added, and the cells were centrifuged. After centrifugation the cells were resuspended in 1 ml of fresh DMEM medium, and plated on the prism where they covered $\sim 80\%$ of the area ($3.5\times {10}^5\mathrm{cells}\mathrm{per}{\mathrm{cm}}^2$). During the measurements, the cells were bathed in fresh DMEM which was constantly replaced at slow flow.

3.1.

Surface Plasmon Wave

The surface plasmon is a transverse magnetic (TM) wave that travels along the metal ($m$)—dielectric ($d$) interface ($x$-direction), and decays in the direction perpendicular to the interface ($z$-direction).²¹

The SP wavevector $k_{\mathrm{sp}}$ is determined by the complex dielectric permittivity of the metal ${\epsilon }_m$ and of the dielectric (analyte), ${\epsilon }_d={(n_d+i{\kappa }_d)}^2$. Its lateral component, $k{(x)}_{\mathrm{sp}}$ is complex and in order to enable propagation, its imaginary part is much smaller than the real part,

The transverse component of the SP wavevector $k{(z)}_{\mathrm{sp}}$ is also complex whereas its real part is much smaller than the imaginary part,

Our experimental setup is based on Kretschmann’s configuration, which includes a high-refractive index prism coated with a thin conducting Au-film [Fig. 1(a)]. The SP is excited at certain angle/wavelength at which $k_{\mathrm{sp}}^{\prime }=k_x=n_{\text{prism}}{k}_0\sin \theta$, where $k_0$ is the incident light wavevector and $n_{\text{prism}}$ is the prism refractive index. The SP resonance can be observed using either angular [Fig. 1(a)] or wavelength [Fig. 1(b)] interrogation. The SP reflectivity dip is clearly visible and it shifts upon varying incident angle or wavelength. At fixed incident angle the resonant wavelength (${\lambda }_{\mathrm{sp}}$) of the SP is determined by the refractive index of the analyte, $n_d$,^22,23 and for a low-loss analyte (${\epsilon }_d^{\prime \prime }/{\epsilon }_d^{\prime }\ll 1$) the ${\lambda }_{\mathrm{sp}}$ is given by the inexplicit expression

The reflectivity at the SP resonance yields the imaginary part of refractive index, ${\kappa }_d$. Indeed, the reflectivity at resonance can be approximated by the Lorentzian:^21,24

The high sensitivity of the SP arises from its evanescent nature. SP is confined in a narrow layer in the vicinity of the metal-dielectric interface, and its sampling depth is determined by the decay length of the SP intensity into the dielectric media, ${\delta }_z$, $[I(z)={|E(z=0)|}^2\exp (-z/{\delta }_z)]$ (Refs. 22 and 24) where

The lateral spatial resolution of the SP equals to its propagation length in the $x$-direction, $L_x=1/2k_{\mathrm{sp}}^{\prime \prime }$.¹⁸ It means that when the analyte layer consists of the regions with different refractive indices, two separate SP resonances will appear only if the size of these regions is larger than the $L_x$.¹⁸ Otherwise, there will be a single SP resonance corresponding to the mean refractive index. However, even this “average” SP resonance is sensitive to morphological changes in cells (see Sec. 4.1).

3.2.

Studying Living Cells with Surface Plasmon

We focus here on the surface plasmon sensing of living cells cultured on top of the Au-coated ZnS prism. The cells are immersed in aqueous culture medium which enables their viability. Optical constants of the epithelia cell monolayer are shown in Fig. 2(a) and 2(b).²⁰ The real part of cell refractive index [Fig. 2(a)] exceeds that of the culture medium (mostly water) due to presence of up to 30% organic substances. The imaginary part of the cell refractive index [Fig. 2(b)] is determined by the absorption and scattering. The scattering losses dominate at the shorter wavelengths ($\lambda <2\mu \text{m}$) and the absorption losses dominate in the mid-infrared range ($\lambda >2\mu \text{m}$).²⁰ For nonhomogeneous media such as cells, the effective refractive index sensed by the SP wave is a weighted average in $z$-direction,

Fig. 2

Optical constants of living cells and water in the 1 to 4 μm wavelength region. (a) The red and black lines show the real parts of the cell and water refractive indices, $n$. (b) The imaginary parts of the cell and water refractive indices, $\kappa$. The gray area indicates strong water absorption band that prevents surface plasmon (SP) excitation.

The real part of refractive index $n_d$ yields the total biomass inside the SP field while the imaginary part of the refractive index ${\kappa }_d$ provides information on the cell morphology.

Using our knowledge of the cell optical constants²⁰ we could make a clever choice of the wavelength range for our living cell studies. Indeed, the infrared wavelength range enables extension of the SP sampling depth up to few microns [Fig. 3(a)] without losing the sensitivity ($\sim {10}^{-6}$ RIU).^18,19,23 This allows monitoring of cellular events not only in the cell-substrate contact region but also inside basal cell region and even in the cell nucleus. The SP propagation length, $L_x$ [Fig. 3(b)] is another parameter that should be considered in the SP-based biosensing. For nonconfluent cells there are two possible regimes: $L_x>l$ or $L_x<l$ where $l$ is the characteristic length of the cell cluster. In the first case, the SP resonance appears as a single resonance and its wavelength corresponds to the effective refractive index (e.g., the mean between the refractive indices of the cells and of the medium). Figure 4 illustrates the second case, where we observe two SP resonances corresponding to cell covered and cell-free regions. Indeed, the optical images show that HeLa cells have flat morphology with large cell-substrate contact area (cross-section approximately 50 μm) comparable to the SP propagation length ($L_x$).

Fig. 3

Characteristic lengths associated with the surface plasmon biosensing. (a) The surface plasmon (SP) sampling depth in $z$-direction. (b) The SP propagation length in $x$-direction. The gray area indicates strong water absorption band that prevents SP excitation.

Fig. 4

(a) Microscopic images of HeLa cells cultured on the Au substrate. The cell coverage of the substrate increases from left to right. (b) The surface plasmon (SP) resonance for different cell coverage. The SP resonance wavelength shifts from ${\lambda }_{\mathrm{sp}}=2.38\mu \text{m}$ in the absence of cells to ${\lambda }_{\mathrm{sp}}=2.55\mu \text{m}$ for full-cell coverage. At the intermediate cell coverage the two SP peaks, corresponding to the covered and uncovered regions, overlap. Note, the guided mode that appears in the spectrum ($\lambda =2.25\mu \text{m}$, blue line) which corresponds to full-cell coverage.

The infrared wave propagation along the cell layer can be associated with the guided mode excitation. The typical cell height is on the order of infrared wavelength; the thickness of the epithelial cell monolayers is approximately 10 μm. The correct order of the refractive indices ($n_{\text{substrate}}>n_{\text{cell}}>n_{\text{medium}}$) enables the total internal reflection at the cell-medium interface, which is mandatory for the guided mode excitation. When the propagation losses in cell monolayer are reasonably low ($\mu =\kappa k_x\sim 100{\mu \text{m}}^{-1}$) and the cell layer is uniform enough, the guided modes can be excited.

The resonant wavelength of the guided mode (${\lambda }_{\mathrm{TM}}$) depends on the cell monolayer thickness $d_{\text{cell}}$, the cell refractive index $n_{\text{cell}}$, and the incident angle. Such guided mode resonances appear in the intact cell monolayers for angular [Fig. 5(a)] and for the wavelength interrogation as well [Fig. 5(b)].

Fig. 5

(a) The angular reflectivity spectra from the Madin-Darby canine kidney cells (MDCK) cells cultured on Au substrate. The surface plasmon (SP) resonance appears as the reflectivity dip whose position depends on the incident beam wavelength, $\lambda$. Additional shallow reflectivity dip at larger angle corresponds to the guided mode. (b) Spectral reflectance for different incident angles for the same assembly. Both surface plasmon and guided modes appear as reflectivity minima whose position depends on the incident angle, $\theta$.

4.1.

Measurement of Cell-Substrate Adhesion Rates with Surface Plasmon

The adhesion and spreading of cells on solid substrates has been studied using optical imaging techniques,^25,26 as well as optical and nonoptical biosensors (surface plasmon,^5,15,27 waveguides,^6–8 cell-impedance,^9,10 and quartz crystal microbalance^11,12). These methods established that upon initial attachment, adherent cells tend to increase their contact area with the underlying substrate. In case of simple epithelial cells, this process is accompanied by the establishment of robust intercellular contacts through the development of the cell-cell junctions until a continuous polarized cell layer is formed.²⁸

Here we track the morphology of the MDCK cell layer starting from deposition of suspended cells on substrate and ending with formation of a confluent cell monolayer. To derive the cell coverage we convert the change of SP resonant wavelength [Fig. 6(a)] into the refractive index change [Eq. (3)], $\mathrm{\Delta }{n}_d=(n_d-n_{\text{medium}})$. Thus, cell coverage $f=\mathrm{\Delta }{n}_d/\mathrm{\Delta }n$, where $\mathrm{\Delta }n\approx 0.03$ is the difference of refractive indices of cells and the extracellular medium.

Fig. 6

Real-time measurement of Madin-Darby canine kidney cells (MDCK) cell monolayer formation. (a) Surface plasmon (SP) resonance development during the cell monolayer formation. The resonance red-shift results from the increment of the refractive index ($n_d$) corresponding to the increase of the cell coverage. (b) SP resonant wavelength (${\lambda }_{\mathrm{SP}}$, upper panel) and corresponding refractive index change (right y-axis). The SP resonance depth ($R_{\min }$, lower panel) depends on the losses upon the SP propagation. This depth is proportional to total cell-medium interface length. (c) $R_{\min }$ as function of the cell coverage resolves different phases in cell morphology: I- deposition and spreading of individual cells, II- cell-cell attachment, III- monolayer closure.

In course of the monolayer formation the SP resonant wavelength ${\lambda }_{\mathrm{sp}}$ varies nonmonotonically. Eventually it shifts toward longer wavelength [Fig. 6(b), upper panel], indicating increase of the refractive index caused be the increase of the cell coverage. On the timescale of several hours the cell proliferation is negligible. This means that the number of cells does not change and the cell coverage grows as a result of cell-substrate attachment and cell spreading over the substrate.

The magnitude of the SP resonance provides information on the cell layer structure as well. The resonant reflectivity $R_{\min }$ depends on the losses upon SP propagation (i.e., scattering and absorption). The major mechanism that contributes to the $R_{\min }$ at $\lambda \sim 2.5\mu \text{m}$ is the SP scattering at cell-medium interfaces [Fig. 2(b)]. The scattering occurs due to the difference of refractive indices between the cells and the medium, $\mathrm{\Delta }n\approx 0.03$. From the dimensional considerations, the scattering cross section for the planar SP wave should scale with the length of the cell-medium interface; in other words, $R_{\min }$ measures the length of the cell-medium perimeter per unit area. At the beginning (0 to 2 h), cells are still separated one from another so the perimeter ($\sim R_{\min }$) grows proportionally to the area/cell coverage ($\sim {\lambda }_{\mathrm{sp}}$), although at the advanced stage [Fig. 6(b), phases II and III] $R_{\min }$ changes very different from ${\lambda }_{\mathrm{sp}}$. Time-excluded representation of $R_{\min }$ vs. cell coverage [Fig. 6(c)] provides better understanding of the cell morphology during monolayer formation. It pinpoints different phases of the process. Phase I corresponds to attachment and spreading of individual cells with concomitant growth of the cell coverage, and of the length of cell-medium interface. During phase II, the cell coverage continues to grow but $R_{\min }$ does not. This corresponds to the cell-cell attachment between the neighboring cells. An additional increase of the $R_{\min }$ at the end of phase II indicates that cells enlarge their projection on the substrate. We suggest that this change is associated with the development of protrusions (lamellipodia) by the marginal cells in the cell clusters. This increases the external perimeter of the clusters, similar to formation of lamellipodia at the wound margin.²⁹ Finally, at phase III cells form a contiguous monolayer (with some voids). As these voids close the cell-medium interfaces shrink, resulting in a decrease of $R_{\min }$.

4.2.

Cell Detachment

Trypsin is a pancreatic serine protease³⁰ which is widely used to disrupt the cell-cell and cell-substrate adhesion. Here we measured the dynamics of trypsin-mediated detachment of MDCK cells from the Au-coated ZnS prism. Figure 7 shows that the massive cell detachment starts after a 5-min delay following the introduction of trypsin solution, while complete cell detachment takes another 10 min. In contrast to cell-substrate attachment, which starts from the stage of individual sedimented cells and is followed by the stage where the cells spread on the substrate; the detachment process starts from the cell monolayer and is followed by cell detachment. The latter is initiated not uniformly but at several specific locations. The weakest cell chain is broken first, than the next one and so on. This leads to opening of voids inside the cell layer. The inset [Fig. 7] compares two SP resonances measured at the same 60% cell confluence. The first one was measured during the attachment process and the second one was measured during trypsin-induced detachment. In the first spectrum, the SP resonance appears as a single dip corresponding to some effective medium, as expected for homogeneously dispersed cells (see Sec. 3.2). In the second spectrum, a very broad SP dip indicates on the existence of two different regions: adherent cells and cell-free substrate, as expected for a disrupted cell monolayer where contiguous regions are interspersed with large voids. This interpretation is also consistent with the optical microscopy observations (not shown here).

Fig. 7

Dynamics of Madin-Darby canine kidney cells (MDCK) cell detachment following exposure to 0.25% trypsin solution, as measured from the ${\lambda }_{\mathrm{sp}}$ position. After a 5-min delay the cells start to detach one from another and from the substrate. The detachment is complete after 15 min. The inset compares the reflectivity spectra during attachment (dashed gray line) and detachment (solid black line) processes at 60% cell confluence. The narrow surface plasmon (SP) dip during attachment indicates that SP senses an effective medium consisting of cells and extracellular medium. The broad SP dip during detachment indicates that the SP dip arises from the two overlapping resonances: one from the cell-covered regions and another from uncovered regions.

4.3.

Chlorpromazine Treatment

Chlorpromazine (CPZ) which is an amphipathic, antipsychotic agent, causes membrane corrugation in cells by the modulation of the outer cell membrane, as it was shown in the studies of red blood cells,³¹ endothelial,³² and epithelial cells.³³ To track dynamics of this process through SP resonance we exposed a confluent human melanoma (MEL 1106) monolayer to 10 μM of CPZ. Figure 8(a) (upper panel) shows that ${\lambda }_{\mathrm{sp}}$ is blue-shifted after CPZ introduction. This shift corresponds to the decrease of refractive index at cell-substrate region. This result is consistent with other studies.^32,33 Explicitly, the ${\lambda }_{\mathrm{sp}}$ measures penetration of extracellular medium between the cells and the substrate. Figure 8(b) schematically shows the cell morphological change induced by CPZ. The increment of $R_{\min }$ [Fig. 8(a), lower panel] suggests increase of the cell-medium interface length, which is also consistent with the above interpretation. The effect of CPZ appears to be reversible. CPZ removal results in fast recovery of $R_{\min }$ and ${\lambda }_{\mathrm{sp}}$ (within ~45 min) and indicates that cell morphology recovers to its normal shape during this time.

Fig. 8

Surface plasmon (SP) follows the chlorpromazine (CPZ)-induced morphological changes in the MEL-1106 (human melanoma) cells. (a) The ${\lambda }_{\mathrm{sp}}$ dynamics (upper panel) shows that the exposure to CPZ results in the refractive index ($n_d$) decrease. The $R_{\min }$ dynamics (lower panel) indicates that the cell-substrate attachment region becomes corrugated and this results in increased SP losses. When the CPZ was washed out, both parameters recovered within 40 min. (b) Schematic illustration of the cell morphology for a CPZ treated ($+\mathrm{CPZ}$) and untreated cell ($-\mathrm{CPZ}$). The cell outer membrane is corrugated in the presence of CPZ and recovers upon CPZ removal.

4.4.

Progressive Polarization of the Epithelial Cells

The epithelium cell layer acts as an active barrier between lumen and serosa and enables transport of chemical substances between these two compartments.³⁴ This function crucially depends on cell polarization, namely the difference between the apical and basal (the side that is attached to substrate) parts of the cell.²⁸ The polarized epithelial cells typically have columnar shape, which is achieved through formation of intercellular junctions such as tight junctions.^28,34 Here we utilized combination of SP and guided modes to measure changes in cell morphology during polarization of intestinal epithelial cells (IEC6). Suspended IEC6 cells were seeded on the Au-substrate and allowed to form a continuous monolayer. For the freshly formed IEC6 monolayer the wavelength of the ${\mathrm{TM}}_1$ guided mode yielded an average layer thickness (cell height) of only 6.0 μm [Fig. 9(a), $t=0\mathrm{h}$] while the analysis of the SP resonance indicated that these cells are well attached to the substrate. Simultaneous optical microscopy showed that these freshly sedimented cells have elongated, asymmetric morphology [Fig. 9(b), $t=0\mathrm{h}$]. During the subsequent 5.5-h period, (i) the cell height grew to 12.2 μm, (ii) the ${\mathrm{TM}}_2$ mode appeared in the infrared reflectivity spectrum, indicating on high uniformity of the cell layer, and (iii) the SP resonance shifted towards shorter wavelength, suggesting entrance of the extracellular medium beneath the cell layer [see Fig. 9(b), schematic drawing]. The optical images show that by this time the cells have already adopted the epithelial-like cobblestone shape [Fig. 9(a), $t=5.5\mathrm{h}$], typical for polarized cell phenotype. Figure 9(c) shows the real-time kinetics of the whole process. During the first couple of hours the average cell height grew exponentially from 6 to 12.2 μm, whereas the refractive index of the basal cell region decreased with similar dynamics.

Fig. 9

(a) The infrared reflectivity spectra from a confluent IEC6 intestinal cell monolayer on Au substrate. Each spectrum reveals the surface plasmon (SP) and guided mode (TM) resonances. The red line stays for the freshly formed monolayer (nonpolarized cells) while the blue line stays for the same monolayer after a 5.5-h incubation period (polarized cells). (b) Upper panel: optical images of the cells taken at the same time points as in (a). The bar represents 20 μm. Lower panel: Schematic representation of the nonpolarized and polarized simple epithelium. (c) Dynamics of ${\lambda }_{\mathrm{TM}}$ (upper panel) and ${\lambda }_{\mathrm{sp}}$ (lower panel) reveal the kinetics of cell polarization. ${\lambda }_{\mathrm{TM}}$ measures the cell height. While the height of freshly formed cell monolayer is 6 μm, it takes 2 h to achieve the maximum monolayer height of 12.2 μm. The ${\lambda }_{\mathrm{sp}}$ indicates concurrent decrease of the effective cell refractive index.