**1. Introduction**

Impedance sensing is a label-free, real-time technique used to monitor cellular function. First pioneered by Giaever and Keese, impedance sensing exposes live cells to very small electrical currents across a range of frequencies [1,2]. By measuring the impedance that the cells provide to this current, we can accurately measure the responses of the cells in realtime. As no labelling is required, measurements are non-invasive and can be carried out over extended periods to give high-resolution information in real-time [3,4]. Furthermore, this information is inherently quantitative and thus can be readily analysed statistically [5,6]. Mathematical models can also be applied to this data to allow the exploration of various cellular parameters that cannot be measured directly [7]. These advantages have triggered

A.; O'Carroll, S.J.; Unsworth, C.P.; Graham, E.S.; Angel, C.E. Comparison of Leading Biosensor Technologies to Detect Changes in Human Endothelial Barrier Properties in Response to Pro-Inflammatory TNFα and IL1β inReal-Time . *Biosensors* **2021**, *11*, 159. https://doi.org/10.3390/ bios11050159

**Citation:** Hucklesby, J.J.W.; Anchan,

Received: 22 April 2021 Accepted: 13 May 2021 Published: 18 May 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

the broad adoption of impedance sensing in a wide variety of applications, with a range of custom instruments having been developed [8–10]. However, the adoption of these systems has been limited, as the construction of customised specialist instrumentation is technically challenging. In contrast, commercially available instruments provide a turnkey solution to accessing impedance sensing. There are, however, only a few commercially available instruments including the Electrical Cell-Substrate Impedance Sensing (ECIS), xCELLigence and cellZscope platforms [11–13]. Despite the widespread use of these platforms individually to assess endothelial barriers [3,14–18] a systematic comparison of each platform's capacity to resolve changes in endothelial barrier properties has not been conducted. Therefore, in this paper, the ability of these instruments to detect changes in endothelial barrier properties in response to TNF α and IL1β were compared.

Giaever and Keese's original Electric Cell-substrate Impedance Sensing (ECIS) invention has since been commercialized by Applied BioPhysics [11]. One such instrument is the ECIS Z Θ, which can be configured to measure cellular impedance in 96-well plates with gold electrodes fabricated directly onto the base of each well that has a growth area of 0.32 cm<sup>2</sup> (Figure 1 and Supplementary Table S1). Impedance and phase measurements at frequencies ranging from 10 Hz to 10<sup>5</sup> Hz are collected by the instrument (Supplementary Table S1). Subsequently, these can be modelled computationally to indicate biologically relevant cellular parameters. Three key values are generated: Rb, Cm and Alpha (Supplementary Table S1). Rb represents the cell–cell contacts, such as those formed by junctional molecules; Cm represents the membrane resistance of the cells; whilst Alpha represents the basolateral adhesion, which is influenced by both the distance between the cells and the underlying substrate and the presence of any junctional molecules bridging this interface [7]. Together, these values allow for the in-depth analysis of biological responses [19].

More recently, ACEA Biosciences (now part of Agilent) released the xCELLigence instrument [12]. Much like ECIS, this instrument uses gold electrodes fabricated directly onto the base of wells in a 96-well plate; each well has a growth area of 0.196 cm<sup>2</sup> (Figure 1 and Supplementary Table S1). However, this instrument only collects impedance measurements at three frequencies, 10, 25 and 50 kHz (Supplementary Table S1). Although modelling cellular parameters is theoretically possible using three frequency measurements, the limited range of readings makes any results unreliable.

Finally, cellZscope is the most recent addition to the market, and is able to measure impedance across a Transwell filter with a cell growth area of 0.33 cm<sup>2</sup> (Supplementary Table S1) [13]. The Transwell is seated in a stainless steel pot that acts as an electrical conductor. A second electrode suspended over the cells makes contact with the media in the apical chamber, completing the circuit and allowing impedance to be measured (Figure 1 and Supplementary Table S1). Like ECIS, phase and impedance data are collected at a range of frequencies from 1 Hz to 100 kHz and hence, can also be modelled (Supplementary Table S1). This results in the calculation of transepithelial-endothelial electrical resistance (TER) as a measurement of the cell–cell junctional interactions, and CCL as a measure of cell layer capacitance (Supplementary Table S1) [13]. An equivalent of the Alpha value generated by the ECIS instrument is not included in this model, as the porous nature of the Transwells means that this parameter is not physically present and therefore not appropriate to infer.

Despite numerous studies using these instruments, direct comparisons between them have not been conducted. This is a critical lack of knowledge, as the inferences from the data collected from all three instruments are regularly used together to interrogate cellular responses [17,20–23]. Therefore, in this paper, we analyse the similarities and differences between these three commercially available instruments. The hCMVEC cell line was chosen due to its low overall resistance, which, although characteristic of brain microvascular endothelial cell lines [23], dictates the use of more sophisticated and more sensitive instrumentation [19].

The inflammatory cytokines TNF α and IL1β were selected for these experiments due to their well-defined biphasic response in this cell line. The response of hCMVECs to IL1β and TNFα has been explored at a molecular level and has been well-characterized using impedance instruments [17]. These responses are ideal for this study, as the cytokines first cause a decrease in resistance, followed by a substantial increase for an extended period. Therefore, both decreases and increases in resistance can be examined with the same stimulus. The transient initial decrease in resistance also showcases the high time resolution of impedance sensing, by highlighting a response that could easily go undetected between the time points of a traditional end-point assay [24]. For this study, TNFα and IL1 βconcentrations were selected to provide a robust biphasic response with which to test the impedance instruments [17].

**Figure 1.** The electrode arrays used most widely in the ECIS (96W20idf plate), xCELLigence (E-plate) and cellZscope instruments differ in their electrode configuration. Both the ECIS and xCELLigence electrodes have a similar interdigitating electrode configuration, which covers a high proportion of the bottom of the well. Hence, their electrodes are directly coated with collagen and in intimate contact with the endothelial cells. In contrast, one of the cellZscope electrodes lines the lower compartment of the Transwell, whilst the second is suspended above the cell monolayer. Therefore, these electrodes are not in direct contact with the endothelial cells.

We evaluated two key parameters of the data produced: the difference in magnitude at key points in time, and the profile of the temporal measurements resulting in different curve shapes. A difference in magnitude is informative, straightforward to interpret and correlates with traditional single-time point assays [19,20]. The second characteristic, the profile of the temporal measurements or shape of the curve, is also useful. Even if two responses have the same magnitude at a key time point, they may reach that point in a very different way. This characteristic was analysed using cross-correlation with no lag, which generates a single value between 1 and −1 for each pair of curves. A value of 1 represents identical curves, 0 shows no correlation between the curves and −1 represents curves with a mirror image opposing profile or inverse correlation [25]. By assessing the magnitude and temporal profile of the response in concert, we are able to rigorously compare the measurements from all three instruments.

In this study, we ran the same experiment simultaneously on the three impedancesensing instruments. We show that, although the instruments' temporal impedance measurements have similar profiles, they differ in magnitude, demonstrating significant differences in sensitivity. Furthermore, the modelled data reinforces the differences in sensitivity between the instruments and reveals changes in endothelial barrier properties that were not evident from the overall impedance measurements. Together, this demonstrates the importance of selecting the most appropriate instrument for a particular study.

#### **2. Materials and Methods**

#### *2.1. Culture of Human Brain Endothelial Cells*

Human cerebral microvascular endothelial cells (hCMVECs) were purchased from Applied Biological Materials Inc (cat# T0259). hCMVECs were cultured in 75 cm<sup>2</sup> (T75) Nunc flasks (cat# 156499) with M199 medium containing 10% FBS, 1 μg/mL hydrocortisone, 3 ng/mL hFGF, 1 ng/mL hEGF, 10 μg/mL heparin, 2 mM GlutaMAX and 80 μM dibutyrylcAMP (later referred to as complete M199 medium) at 37 ◦C, with 5% CO2 and 100% humidity. For both hCMVEC maintenance and experiments, culture vessels were coated with 1 μg/cm<sup>2</sup> collagen I dissolved in 0.02 M acetic acid for 1 h at room temperature, before being washed 3 times with sterile MilliQ water and seeding the hCMVECs. To passage the hCMVECs, T75 flasks were washed twice with 4 mL pre-warmed PBS before being incubated with 4 mL pre-warmed TrypLE for 5 min at 37 ◦C. The TrypLE activity was then neutralized with 4 mL complete M199 and the cells were centrifuged at 100× *g* for 5 min, counted, and seeded for experiments. All experiments used hCMVECs between passages 11 to 16. All impedance instruments and experimental hCMVEC cultures were kept in dedicated incubators at 37 ◦C, with 5% CO2 and 100% humidity.

#### *2.2. Impedance Sensing Experiments*

ECIS: 96W20idf plates were treated with 10 mM cysteine for 15 min to clean the electrode and standardize the electrode impedance (as per manufacturers' instructions). The wells were then coated with collagen as described above. The hCMVECs were seeded in 200 μL complete M199 medium. The ECIS machine was run continuously in multifrequency mode using the default frequency spectra (Supplementary Table S1).

xCELLigence: E-plates (96 wells) were coated with collagen as described above. Complete M199 was added to each well and calibration was conducted. Cells were seeded in 122 μL Complete M199. Impedance was measured at 10, 25 and 50 kHz (Supplementary Table S1).

CellZscope: before the experiment, cellZscope components were cleaned with MilliQ water, 70% ethanol, and then MilliQ water again. The pots and dipping electrodes were autoclaved, whilst the remainder of the Cell Module was sterilised with 70% ethanol. Before coating, the Cell Module was assembled under sterile conditions, and each of the stainless steel pots was flooded with 900 μL basal M199 media. The assembled module was then placed in the cell culture incubator to equilibrate for at least one hour. Transwells (Corning; 6.5 mm insert, 0.4 μm pore size) were coated from the apical side, as previously described in Section 2.1. The hCMVECs were then seeded into the apical chamber in 200 μL complete M199 medium. Transwells were then transferred into the Cell Module, taking care not to trap any bubbles underneath the membrane. The Cell Module was then placed in the instrument, and the spectra were acquired at the highest resolution between 1 and 100 kHz (Supplementary Table S1). Measurements were made every 15 min, the fastest rate possible at these frequency settings.

#### *2.3. Treatment with Inflammatory Cytokines*

After seeding, the cells were cultured for 48 h to allow the barrier to fully develop and impedance to stabilise. On the day of treatment a 5× stock of TNF α and IL1β in complete M199 was prepared; once added to the corresponding culture wells this provided a final concentration of 500 pg/mL of TNF α or 500 pg/mL IL1β. For the control treatment, the 5× stock consisted of complete M199 with an equivalent amount of MilliQ water (henceforth labelled as the control). Each instrument was then paused, and the 5× stock was gently introduced to the middle of the well or apical chamber. The cultures were then returned to the respective instrument and the measurements resumed. Cell monitoring continued on all instruments for a further 27 h.
