Electric Cell-substrate Impedance Sensing for the Quantification of Endothelial Proliferation, Barrier Function, and Motility

Setting up an experiment is fairly simple and the measurement itself is fully automatic, but as is often the case, correct analysis and interpretation of the data are crucial. In the representative data section a guide is provided for the interpretation of the most important parameters provided by impedance spectroscopy with ECIS. This will be helpful to decide for which kind of scientific problem an ECIS measurement can be useful and which parameters should be recorded.

A typical experimental read-out can generally be divided in a growth-phase after cell inoculation and a plateau-phase when cells reach confluence. Each of these phases yields information on different cellular properties. A representative measurement is shown in Figure 2 and divided into four subphases to analyze A) cell adhesion, spreading and proliferation; B) formation of a maturation EC barrier; C) cell migration after the generation of an electrical wound; and D) the cellular response to stimulation with the vasoactive agent thrombin. Phase-contrast and immunofluorescence images were acquired directly from the measurement electrode during the different subphases to illustrate the link between electrode coverage, cell status, and recorded impedance signal.

Adhesion, spreading and proliferation

Every cell type has its characteristic adhesion and growth curve that can be manipulated by varying seeding density, coating with extracellular matrix proteins or other stimuli. One of the major difficulties to study those processes is to differentiate between adhesion, spreading and proliferation. Wegener et al.¹¹ gave a detailed description for the use of a combination of resistance and capacitance to distinguish between those parameters and in Figure 3 it becomes obvious how resistance and capacitance can complement each other. It is important for the study of adhesion and spreading to understand the influence of the measurement frequency, because here the electrical field causes a polarization of the cell membrane of the cells in suspension⁸, which forces the current to flow exclusively around the cells. When the cells attach they start to restrict the current flow by spreading over the electrode and capacitance decreases in a linear fashion with the percentage of open area on the electrode (fractional area). Hence, adhesion, spreading and proliferation can be quantified best, when recording capacitance at a frequency higher than 40 kHz (Figure 3B), where the decrease in capacitance is direct proportional to the electrode coverage. Alternatively resistance at its most sensitive measurement frequency is a direct measure for the differentiation and maturation of cells into a confluent barrier (Figure 3A). In addition to the fractional area, Wegener et al. introduced two other parameters to quantify adhesion and electrode coverage: t_1/2 (half maximum capacitance), which is the time until 50% of the electrode is covered with cells (Figure 3B, insertion), and the slope of the capacitance curve (s, not shown) as the rate of cell spreading and proliferation.

Resistance as a measure of barrier quality

Resistance is the part of impedance that describes barrier quality and function best, because it neglects capacitive components from membrane, electrode and medium. Here the term barrier quality and not permeability is used, since permeability is per definition only dependent on cell-cell contacts. Resistance however is determined by the ability of cells and their cell-cell and cell-matrix interactions to block the current flow. Accordingly different cell types (clones and passages) have their specific resistance values (Figure 4A). Although resistance gives a much clearer idea about the cell barrier, the impedance signal should always be taken into consideration.

Modeling of Rb and Alpha

The ECIS software harbors a particular feature, the ability to model two parameters, which distinguish between cell-cell and cell-matrix adhesions (Figures 4B and 4C). Rb (in cm² x ohm), is the resistivity of cell-cell contacts to the current flow and thereby an inverse measure of classical permeability. High Rb implies low permeability towards the current flow. Alpha (in cm x ohm^0.5), is a measure for the impedance contributions arising from the cell-electrode junctions. The model to quantify cell-cell and cell-matrix contacts was introduced 1991 by Giaever and Keese and is based on the assumption that cells are circular, disc shaped objects that have an insulating membrane, hover over the electrode and are filled with conducting electrolyte². The insulating property of the cell membrane leaves only three pathways for the current: a) between ventral surface of the cells and electrode, b) between the cell-cell contacts, and c) through the cells (only likely when a high measurement frequency is used). A more detailed derivation and explanation can be found in the papers of Lo et al^3,12.

Micromotion

It has been shown that small fluctuations in the resistance signal (Figure 4A) are due to subtle, random cell movements in confluent cell layers as well as cells moving on and off the electrode in sparse cultures². This aspect of ECIS time-course data described by Lo and Opp et al.^13,14 is often overlooked and best seen when measuring with the 1E arrays and at the most sensitive frequency to account for para- and subcellular current-paths. The calculation of this cell caused noise (micromotion) is not part of the standard measurement software, although in the newest version Fast Fourier Transformation (FFT) is integrated. Recording RTC data with a sampling frequency of 1 Hz for 1 hr at 4 kHz provides sufficient resolution for the calculation of EC micromotion. This calculation provides one single value and can be directly used for statistical analysis. An extended discussion on the use of frequency analysis for micromotion can be found elsewhere¹⁵. Alternatively to FFT other analysis strategies can be used, like variance of the increments. Variance is more sensitive towards small changes in micromotion, but because of this sensitivity the comparison of individual experiments becomes difficult. The slopes of the power spectrum are a more robust measurement of micromotion (Figure 4D). Here the area under the curve can be seen as the amount of micromotion or noise the cells create. Hence, the steeper the slope of the power spectrum the smaller the area under the curve and the smaller the micromotion.

Electric wound healing assay

To study migration of cells, usually mechanical wounds are applied by scratching cells off the culture plate with pipette tips, cell scrapers or specific stamps and follow their remigration microscopically¹⁶. This method has the obvious disadvantage that the size of the scratch varies and is usually too big to neglect the influence of cell proliferation on the wound healing process. The wounding function of the ECIS uses a high voltage, high frequency pulse to get a cell free electrode and subsequently follows remigration of neighboring cells to replace the death cells (Figure 5A). An advantage of this automated method over standard, labor intensive scratch assays is that the ECIS produces a highly reproducible wound with a precise diameter of 250 µm, that closes within a few hours, to study pure migration. Further, the electrical pulse does not remove protein coating and secreted extracellular matrix. Under the assumption that the cells remigrate on the electrode from all sides, the migration rate can easily be calculated by dividing wound radius by time required for wound closure¹⁷. As an alternative to electrical wounding, recently the so-called electric fence method was introduced (not shown). Here high current pulses prevent attachment and migration of cells on the electrodes after inoculation. The cells will grow around the measurement electrode and start migrating inwards as soon as the electric pulses are turned off. The advantage of the electric fence over the electric wounding is that the electrode surface hasn't been altered by cell growth or cell removal, which is important to measure influences of different coatings on cell migration.

Cell stimulation

In addition to the study of cell migration, impedance spectroscopy is perfectly suited to measure the effects of drugs and toxins on cells. Figure 5B presents a classical stimulation/inhibition experiment, where thrombin was used to induce hyper-permeability in the confluent EC barrier by contraction and gap formation of cells, which manifests as a transient drop in resistance. By preincubation with the Rho kinase inhibitor Y-27632 most of the thrombin response is diminished by lowering basal cell tension¹⁸ and blocking stress fiber formation¹⁹. Then the Rho kinase blocker was used at different time points after thrombin addition, leading to an immediate and total recovery of the EC barrier. Here the advantage of a continuous impedance measurement system to study acute cellular responses becomes obvious. In regular endpoint assays (e.g. immunofluorescence staining or macromolecule passage), this acute response to the used blocker would be very difficult if not impossible to detect and quantify. Important to note is that with impedance measurements the permeability towards ions is described and not towards macromolecules. Please refer to Aman et al.²⁰, where an excellent comparison between ECIS measurements and macromolecule passage experiments is provided.

Figure 2. Representative measurement. MVEC were grown on a gelatin coated 8W1E arrays with a low seeding density of 5,000 cells/cm² and MFT recording was performed over 10 days. The measurement illustrates the different read-outs of an ECIS experiment: A) adhesion, spreading and proliferation; B) formation and maturation of a confluent cell barrier; C) wound healing after the application of an electrical wound; D) stimulation with the vasoactive agent thrombin. The two arrows indicate medium refreshments, which were performed in intervals of 4 days and give an idea about the sensitivity of the measurements towards mechanical stimuli and changes in temperature and pH. The images in the top panel correspond to the measurement and were acquired directly from the measurement electrode. The phase-contrast image on the left shows the endothelial cells 24 hr after inoculation, starting to cover the electrode. The immunofluorescence pictures in the center and on the right present F-actin staining of the confluent endothelial layer before and after stimulation with thrombin (1 U/ml). Thrombin causes contraction of endothelial cells by formation of so called stress fibers and the transient opening of inter-endothelial gaps (arrows and image inserts), recognizable in the ECIS signal by a drop in resistance to baseline. Here the sensitivity of the measurement becomes obvious and how changes in the cell layer correlate with the impedance signal. Please click here to view a larger version of this figure.

Figure 3. Cell attachment and growth. MVEC from the same donor were seeded with three different densities on gelatin coated 8W10E arrays and cell growth was recorded at multiple frequencies for 60 hr. A) Resistance values for the three conditions at their optimal frequency of 4 kHz to measure barrier formation. All three seeding densities established a confluent barrier of ca. 1,200 Ω at the end of the measurement; the proliferation rates however (slopes of the curves) differ markedly. B) Measurement of capacitance at 64 kHz provides insights on adhesion and spreading. Adhesion is the initial plateau phase in the capacitance curve (between 2-8 hr). Spreading, which is in linear relation to the electrode coverage, is represented by the slope of the curve and is completed after 10-30 hr depending on the seeding density. The time point t_1/2 (half maximal coverage) can be used for statistical analysis. Please click here to view a larger version of this figure.

Figure 4. Characterization of the confluent barrier. Two MVEC donors with different passage numbers (passage 3 and 6) were seeded on gelatin coated 8W1E arrays and a MFT measurement was performed for 30 hr to characterize the EC barrier. A) Resistance measurements reveal that the younger donor 1 has a higher resistance of ca. 11,000 Ω in comparison to the ca. 7,500 Ω of the older donor 2. B-C) The modeling capabilities of the ECIS have been used to find the cause for the lower resistance. Modeling revealed a comparable cell-cell interaction Rb and indicated a weakened cell-matrix interaction Alpha as a possible cause. D) RTC was performed to quantify micromotion within the confluent cell layer by Fourier transformation, where the area under the curve is proportional to micromotion. Analyses of the spectra shows less micromotion of the older donor 2 and thereby support what can already be assumed from the resistance signal (less variance). Please click here to view a larger version of this figure.

Figure 5. Cell migration after electric wounding and effects of thrombin stimulation. Impedance recordings were performed at 4 kHz with maximal temporal resolution. A) MVEC donor 1 was used in passage 3 and donor 2 in passage 6 on gelatin coated 8W1E arrays and grown to confluence. Then an electrical wound was created by application of two 5 V pulses at 60 kHz with a duration of 20 sec each. The younger donor 1 showed better wound healing capabilities based on a faster wound closure (migration, slopes of the curves) and a higher resistance after wounding in comparison to the older donor 2. B) HUVEC were grown on gelatin coated 8W10E arrays. Upon confluence the cells have been serum starved for 1.5 hr by replacing the complete culture medium with basal medium plus 1% human serum albumin (HSA). Usage of plain medium with HSA is a critical step, since serum components block the thrombin response. As soon as a stable resistance plateau was detected, thrombin was added directly to the wells and mixed carefully with a 200 µl pipette. The maximal thrombin effect is visible after 30 min, characterized by a transient drop in resistance to baseline and a ca. 30-50% lower resistance after recovery. The cells were either preincubated with the Rho kinase blocker Y-27632 for 30 min or the blocker was added 20, 30, or 40 min after thrombin addition, which diminished the thrombin effect immediately. Please click here to view a larger version of this figure.