2.1.

SD-OCPM Principle and Experimental Setup

The operation of SD-OCPM is detailed elsewhere.¹⁰ Figure 1 depicts SD-OCPM setup for label-free cell-based assay. A super-continuum laser (SC-400, Fianium Inc., Southampton, United Kingdom United Kingdom) passed through a long-pass filter (Chroma Inc., Bellows Falls, Vermont, HHQ550lp) and was directed to the fiber interferometer. Combined use of the long-pass filter and the single-mode fiber interferometer resulted in the sample light centered at $\sim 800\mathrm{nm}$ with a full width at half maximum bandwidth of $\sim 200\mathrm{nm}$. The light from the sample arm fiber was collimated with a diameter of $\sim 4.4\mathrm{mm}$, and passed through the two-dimensional galvanometric scanner (Thorlabs, Inc., Newton, New Jersey GVS-002). The beam was then focused on the top surface of a glass bottom microtiter plate with a focusing lens (Thorlabs, Inc., LSM05-BB). The light was reflected from the interfaces along the beam path, and re-coupled back into the interferometer. The interference signal was detected by a custom-built spectrometer at the detection arm with a maximum line rate of 70 kHz. The SD-OCPM was configured as an inverted microscope for live cell imaging.

Fig. 1

(a) Experimental configuration of spectral-domain optical coherence phase microscopy (SD-OCPM) for cell assay. Light from a super-continuum laser was directed into a $2\times 2$ fiber-based coupler. In the sample arm, the output beam was collimated, passed through a galvanometric beam scanner and subsequently was expanded by the beam expander. The expanded beam was then focused onto the cell-substrate interface. The reflected light was re-coupled into the fiber and detected by a spectrometer in the detection arm. (b) The back-reflected light from reference (bottom) and sample (top) surfaces interfere with each other to produce interference fringes. When the receptors on cell membrane are activated, relocation of intracellular structures proximal to the cell-substrate interface is induced, which results in a fringe shift in the interference spectrum (full-lined arrow to dashed arrow). Such fringe shifts can be quantified by examining the phase of the corresponding interference signal in the path-length resolved information.

The interference between the reflections from the top (measurement) and bottom (reference) surfaces of a substrate produces an interference spectrum as depicted in Fig. 1(b). As noted by Fang et al.⁵ and in Schröder et al.,⁷ upon activation of cell membrane receptors by external ligands, the cells undergo intracellular mass redistribution which results in alteration of cell adhesion and local scattering properties inside the cell. These changes proximal to the cell-substrate interface [red thick line on the substrate, Fig. 1(b)] lead to a fringe shift in the interference spectrum, which is in turn translated into the phase change in the optical path length domain. The coherence length of the light source determines the detection volume. In our setup, it was measured to be $\sim 1.4\mu \mathrm{m}$ in air, which is much smaller than the thickness of MCF-7. We performed confocal microscopy imaging of fluorescently labeled MCF-7s. The cells were labeled with PKH26 (Mini26, Sigma-Aldrich, St. Louis, Missouri) and were imaged by a confocal microscope ($63\times /1.2\mathrm{NA}$, Zeiss LSM 510, Germany). The thickness of MCF-7 monolayer was measured to $\sim 12\mu \mathrm{m}$ on average. The fluctuations from other regions inside the cells would be rejected due to a short coherence length.

Fig. 2

Measured phase and intensity fluctuations for the cell-free plates. The chamber was initially filled with the incubation buffer and 5-μM histamine was injected at 1 min. The arrow indicates the time for the histamine injection. No notable changes were measured for the intensity and phase signals.

In this study, we set the lateral resolution to be $\sim 20\mu \mathrm{m}$ to measure averaged cellular responses over the entire well, while maintaining the ability to resolve individual cells. Higher spatial resolution can be readily achieved with high NA lenses, and would enable to examine heterogeneous cellular responses and intracellular dynamic responses with subcellular resolution. However, it would compromise the number of cells that can be measured for a given acquisition time. In terms of label-free cell assay, it is more advantageous to examine large population of cells under the introduction of chemicals to obtain robust and reliable cellular response. Therefore, lateral resolution was set to be $\sim 20\mu \mathrm{m}$. We also cultured the cells with high confluency ($>75\%$) to minimize the empty regions, thereby performing more valid measurements over the field-of-view (FoV). SD-OCPM measurement averaged over the FoV therefore represents the collective responses of multiple cells due to the stimulants. The random dynamics such as cell membrane fluctuation would be averaged out by examining multitudes of cells.

2.2.

Data Analysis

The measured interference spectra were transformed from wavelength to wavenumber space as described in Mujat et al.¹² Inverse Fourier transform of the spectrum results in path-length resolved complex-valued information of a specimen. We located the interference signal between the reflections of the top and bottom surfaces of the microtiter glass substrate and examined its complex-valued signal. It can be seen that our measurement is greatly influenced by the reflection from the top surface of a glass substrate. The contribution of the light scattered from cell bottom membrane and other intracellular structures would be significantly smaller compared with that from the glass substrate. In order to alleviate this signal leakage problem, we adopted the strategy described in Ellerbee and Izatt.¹³ We subtracted the first image from the subsequent time-lapsed image sets, thereby measuring the changes due to cellular responses only. The amplitude and phase of the differential measurements were then averaged over the entire FoV and plotted to examine dynamic responses. A small phase drift over time was eliminated by referencing the measured phase distribution relative to the cell-free regions in the same image. The FoV of SD-OCPM image was $3.3\times 3.3{\mathrm{mm}}^2$ per well, and the image size was $500\times 500\text{pixels}$. The pixel rate was set to 10 kHz. The image acquisition was performed at every 2 min.

2.3.

Cell Preparation and Chemicals

A breast carcinoma cell line (MCF-7) was used to assess the capability of SD-OCPM label-free cell-based assay. The cells were suspended in Roswell Park Memorial Institute medium (RPMI, GIBCO BRL, Gaithersburg, Maryland) at a fixed concentration of $1\times {10}^5/\mathrm{ml}$. The 100 μl of the sample was loaded on glass bottom multiwell plates (In Vitro Scientific, Sunnyvale, California, P384-1.5H-N) and incubated for 24 h prior to experiment.

We employed 2-picolinic acid (PA, Sigma-Aldrich, St. Louis, Missouri, P42800) and histamine (Sigma-Aldrich, H7125) to induce cellular changes. 2-picolinic acid (PA), a pyridine compound with a carboxyl side chain, is known to interrupt cellular metabolism. It arrests the cell cycles on $G_1$, leading to prohibition of the cell proliferation or cell death.¹⁴ On the other hand, histamine is an organic compound that regulates cell proliferation through stimulation of histamine receptors on the cell membrane.¹⁵ The cellular responses to histamine have been investigated with other sensor technologies.¹⁶

The chemicals were prepared by serially diluting 1-M stock solutions of each chemical with phosphate-buffered saline solution to obtain the concentrations of 5.0, 12.5, 25.0 mM for PA, and 1, 3, 5 μM for histamine, respectively.

2.4.

Experiment Procedure

For measurement, cell-loaded multiwell plates were placed on a custom-built SD-OCPM specimen stage. We acquired the image prior to the chemical injection as a reference. The responses of the PA-treated and histamine-treated cells were then recorded for 30 and 40 min, respectively, right after the introduction of the chemicals. Several wells in each measurement were untreated as controls.

3.1.

System Stability

Prior to measuring cellular responses to the chemicals, SD-OCPM performance was first evaluated. The microtiter well was filled with the incubation buffer, and the phase fluctuation of the interference signal between the reflections from the top and bottom surfaces of the glass substrate was recorded. The SD-OCPM beam was positioned stationary at the center of the FoV. The measured signal-to-noise ratio was $\sim 50\mathrm{dB}$, under which condition the theoretical sensitivity is $\sim 2.2\mathrm{mrad}$. The measured sensitivity in SD-OCPM was $\sim 3.3\mathrm{mrad}$ in air. The difference between the theoretical and measured sensitivities may be due to the influence of external disturbances such as vibrations during the measurement.

We also performed SD-OCPM imaging of a cell-free substrate. A cell-free glass-bottom microtiter plate was initially filled with the incubation buffer, and the 5-μM histamine was injected at 1 min to examine the effect of bulk RI variation on the SD-OCPM phase measurement (indicated by an arrow in Fig. 2). As can be noted in Fig. 2, the histamine introduction did not produce a notable change. The standard deviations of the phase and intensity signals averaged over the entire FoV were found to be $\sim 40.2\mu \mathrm{rad}$ and $\sim 0.14$, respectively. Unlike other optical cell-based assay technologies that measure the changes of RI proximal to the sensor surface, SD-OCPM measures the changes of optical thickness near the top surface of the substrate. Therefore, SD-OCPM is more immune to bulk RI variation, and rather sensitive to movement of scattering structures inside the probe volume.

SD-OCPM signal fluctuations of untreated live cells were then measured (Fig. 3). Representative SD-OCPM images of the cells are also shown in Fig. 3. It can be noted that intensity values in the cell-free regions are larger than those in the regions on which the cells adhere. This can be explained by that the magnitudes of the reflected light in the cell-free regions are greater than other areas due to the larger RI gradient. The phase distribution in the SD-OCPM phase image represents the optical path-length delay due to cellular structures within the detection volume relative to the cell-free regions. The measured phase and intensity fluctuations exhibited small variations of $\sim 427.1\mu \mathrm{rad}$ and $\sim 0.2$, respectively. Random cellular dynamics such as the intrinsic membrane motion within the probe volume may account for this small fluctuation.

Fig. 3

Measured phase and intensity fluctuations for the live MCF-7 cells, along with representative SD-OCPM phase and intensity images. Small variations in the measured signals may be due to intrinsic cell membrane fluctuation and random motion of intracellular structures inside the probe volume. (Scale bar: 1 mm).

3.2.

Measurement of Dynamic Cellular Responses

Having evaluated the SD-OCPM performance, MCF-7 cell responses to 2-picolinic acid (PA) and histamine were measured. Representative SD-OCPM signals at a PA concentration of 12.5 mM are shown in Fig. 4(a), along with SD-OCPM intensity and phase images of the MCF-7s. It can be noted that the positive phase distributions that were initially observed at 0 min decayed in time. This delay of the SD-OCPM phase signal may indicate a decrease in the number of cellular structures within the detection volume. Changes in the SD-OCPM intensity signal, on the other hand, were not significant. It may be due to that changes in the intensity signal were not larger compared with the power fluctuation of the light source. We separately performed bright-field microscopy imaging of the cells at the same concentrations, but found no marked changes (data not shown). This difference can be accounted for by that SD-OCPM measures the phase distribution only at the cell-substrate interface, while bright-field microscopy images are generated from the light transmitted through the cells. Hence, the bright-field microscopy would not be able to detect cell detachment and intracellular structural changes undergoing only at the cell-substrate interface.

Fig. 4

Dynamic cellular responses measured with SD-OCPM. (a) Representative intensity and phase profiles for 12.5mM PA treated cells, along with the corresponding SD-OCPM images (Scale bar: 1 mm). Negative signal change can be observed in the phase signal, but no significant change was detected in the intensity measurement. (b) Representative intensity and phase measurements for 5 μM histamine treated cells. The representative SD-OCPM images are shown in the insets (Scale bar: 1 mm). The phase measurement presents a positive signal change in time, whereas the intensity measurement does not show a marked signal change.

Shown in Fig. 4(b) is SD-OCPM measurement of 5-μM histamine-treated MCF-7s. Representative SD-OCPM images of the cells can also be found in the insets. Although the changes were not pronounced as in the PA-treated cells, the averaged phase over the well increased and then became stabilized in 10 min. The cellular responses were heterogeneous at a single-cell level, but SD-OCPM measured increased cellular responses averaged over the entire well. SD-OCPM intensity signal, however did not show a marked change.

It can be seen that the SD-OCPM phase signal is more sensitive to small changes of cellular responses. The motion of intracellular structures was found to generate significant phase variations, and it has been used to measure viscoelastic properties inside the cells.¹⁷ Joo and de Boer¹⁸ have also measured the protein absorption and desorption on the glass substrate based on the phase variation.

We thus used SD-OCPM phase information to examine the cellular responses as a function of PA and histamine concentrations (Fig. 5). The higher PA concentration produced more rapid and large phase changes, while the control case did not exhibit a marked change [Fig. 5(a)]. Figure 5(b) shows the magnitudes of the phase changes averaged over five wells at each concentration. SD-OCPM phase changes were found to have a linear relationship with PA concentration as $-210\mu \mathrm{rad}/\mathrm{mM}$ [Fig. 5(b)]. Figures 5(c) and 5(d) show the representative cellular responses with different histamine concentrations and the phase changes averaged over five wells for each concentration. Large phase variations were observed at the higher histamine concentration. The SD-OCPM response against histamine concentration was found to vary linearly as $+390\mu \mathrm{rad}/\mu \mathrm{M}$ [Fig. 5(d)].

Fig. 5

(a) SD-OCPM phase measurements of 2-PA treated MCF-7s at different concentrations. (b) The averaged SD-OCPM phase change versus 2-PA concentration. (c) SD-OCPM phase measurements of histamine-treated MCF-7s at different concentrations. (d) The averaged SD-OCPM phase change versus histamine concentration. Five measurements were averaged for each concentration.

Our results with PA and histamine treated cells agree to earlier publications. Fernandez-Pol et al.¹⁴ observed completed cell detachment and floating after $\sim 48\mathrm{h}$ with the 3-mM PA treatment. Cytotoxicity of PA on human skin cells has also been reported.¹⁹ Cell detachment and death result in a decrease in the optical thickness proximal to the substrate, which was detected as the negative phase changes in SD-OCPM measurement. SD-OCPM measurements with histamine treated MCF-7s are also consistent with the previous reports. For instance, Ferrie et al. measured the responses of human squamous carcinoma (A431) cells to histamine with the resonant waveguide optical sensor. They found the increased mass distribution proximal to substrate interface.¹⁶ In addition, Brischwein et al.²⁰ utilized ECIS to measure the response of HELA cells to 25-μM histamine and observed an increased impedance signal, which was stabilized within 10 min.

In Figs. 4 and 5, exponential functions were used to fit the SD-OCPM phase signals in time. Temporal responses to the chemicals have previously been modeled with exponential functions. Malleo et al.²¹ employed continuous differential impedance spectroscopy to examine HeLa cell responses to Polysorbate 20, and reported exponential behaviors in the measured signal as a function of time. Fang and Ferrie²² have also observed the exponential signal changes for the A431 cells upon the introduction of various types of chemicals with RWG sensors. Our measurements with the PA and histamine exhibited similar behaviors. For the dependence of cellular responses on the histamine concentrations, we found the linear dependence of the cellular responses on the concentration ranges of 1 to 5 μM, which is consistent with other literatures. For example, Ferrie et al.¹⁶ investigated the responses of A431 cells to histamine over the range of 2 nM to 100 μM, and observed a linear dependence on the histamine concentration in the range of 1 to 5 μM.