Cellular sensing platform with enhanced sensitivity based on optogenetic modulation of cell homeostasis

Abstract

We demonstrate a new biosensing concept with impact on the development of rapid, point of need cell based sensing with boosted sensitivity and wide relevance for bioanalysis. It involves optogenetic stimulation of cells stably transfected to express light sensitive protein channels for optical control of membrane potential and of ion homeostasis. Time-lapse impedance measurements are used to reveal cell dynamics changes encompassing cellular responses to bioactive stimuli and optically induced homeostasis disturbances.

We prove that light driven perturbations of cell membrane potential induce homeostatic reactions and modulate transduction mechanisms that amplify cellular response to bioactive compounds. This allows cell based biosensors to respond more rapidly and sensitively to low concentrations of bioactive/toxic analytes: statistically relevant impedance changes are recorded in less than 30 min, in comparison with >8 h in the best alternative reported tests for the same low concentration (e.g. a concentration of 25 μM CdCl₂, lower than the threshold concentration in classical cellular sensors). Comparative analysis of model bioactive/toxic compounds (ouabain and CdCl₂) demonstrates that cellular reactivity can be boosted by light driven perturbations of cellular homeostasis and that this biosensing concept is able to discriminate analytes with different modes of action (i.e. CdCl₂ toxicity versus ion pump inhibition by ouabain), a significant advance against state of the art cell based sensors.

1. Introduction

Live cells are the key components of a specialized class of biosensing platforms that monitor the response of cells toward quantitative evaluation of the changes in their microenvironment, including occurrence/presence of bioactive molecules and hence have important sensing applications in the fields of environmental safety, disease modeling, and drug effect assessment. Exposure to analytes (often noxious) prompts deviations of key variables (including membrane potential, cell volume, electrolyte concentrations, pH, reactive oxygen species - ROS) from their physiologically relevant ranges, modulates specific cellular responses (changes in state or activity of cells in terms of e.g. morphology, gene expression, enzyme production) and t riggers homeostasis restoring mechanisms (Chovatiya and Medzhitov, 2014). Failure of these mechanisms leads to cell cycle alterations and eventually to cell death, giving cell based platforms their unique ability to reflect physiologically relevant functional information concerning stimulus propensity/efficacy to elicit bio-effects, and not merely the analyte’s concentration or binding (Stenger et al., 2001).

The effervescent research in the domain of cell-based biosensing platforms (CBS) is constantly reviewed (Gupta et al., 2019; Ye et al., 2019). This reveals the sustained CBS progress from classical cellular sensors based on assessing endpoints (e.g. holistic, such as viability/death, proliferation at specific time points), to the highly effective, yet target specific, fluorescence (Hofmann et al., 2013) or luminescence (Roda et al., 2016) based biosensors and “sentinels” (e.g. Canary-Cellular Analysis and Notification of Antigen Risks and Yields (Rider et al., 2003) for detection of pathogens and toxic proteins) towards more general “physiometers” (Hu et al., 2013; Ye et al., 2019; Zou et al., 2015) such as CBS based on electrical assays.

Cellular processes, such as adherence, spreading, growth, and motility, and cellular activities, e.g. signal transduction, amplification, and analyte recognition are the principal factors involved in phenotypic assays, identification of toxic effects and screening of bioactive molecules or drugs by CBS platforms (Gui et al., 2017; Gupta et al., 2019). Accordingly, cellular sensing methods based on electrical assays (e.g. Microelectrode arrays (MEA) or Electrical Cell substrate Impedance Sensing (ECIS), alone or in combination with complementary techniques) that allow label-free, minimally invasive, multiparametric monitoring capabilities are therefore extensively used to extract the relevant information for establishing the physiological, pharmacological impact (Arndt et al., 2004; Gupta et al., 2019; Hefele et al., 2019) such as the dose dependent effectiveness of analytes in inhibiting a specific biological or biochemical function (Inhibitory Concentration, IC) or in inducing a maximal response (maximal effective concentration, EC) for specific analytes (drugs, chemicals or nanoparticles) (Atienzar et al., 2013; Lapp et al., 2017; Novellino et al., 2011; Wegener, 2015).

Despite the continuous developments of cell-based biosensors, the list of challenges remains substantial. Cell-fate decisions (e.g. whether a cell divides, differentiates or dies), can take a long time, are cell type specific and depend on, and induce, adaptation processes (i.e. transient expansion or contraction of the homeostatic range in response to exposure to sub-toxic, non-damaging, signaling molecules or events (Davies, 2016)) of cellular status. Moreover, the response characteristics are dependent on the distinct cell type used for the particular application, the cellular status (i.e. developmental stage, growth conditions) and the number, type and functionality of channel repertoire. As a result, current cell based biosensors are prone to high variability, have response characteristics both dependent on the distinct cell type or population used for the particular application and influenced by the heterogeneity in the cell population (Gupta et al., 2019). Additionaly, these general cellular biosensors (not target specific as the ones based on CANARY technologies that have high sensitivity and fast response on the expense of requiring particular sensor cells for each individual target pathogen or toxin) are characterised by long response times and limited sensitivity/specificity in assessing the subtle events that modify cell’s behavior for stimuli of reduced magnitude and the accompanying fine variations of cellular parameters (Derick et al., 2017), have limited robustness, reliability and repeatability (Gupta et al., 2019).

We address some of these challenges by advancing a new concept to increase the stability, boost the sensitivity of cell based biosensors (increase the cellular reactivity to bioactive analytes) by bestowing electrogenic (i.e. membrane potential modulation) conduct to nonexcitable cells (intrinsically more robust and having a more reduced endogenous channels repertoire than excitable cells) as a way to modulate cell homeostasis.

Due to its roles, the membrane potential is the main homeostasis regulator, acting as a liaison between the extracellular and intracellular milieus. Membrane potential modulations have exquisite roles in signal transduction and propagation, activation of intracellular processes (e.g. by regulating GPCRs -the superfamily of membrane proteins that transmit binding information of a broad spectrum of extracellular li-gands into a range of signaling pathways in the cell (Vickery et al., 2016) by directly influencing second messengers formation (Yang et al., 2013)), cell-to-cell communication, control of cellular functions (e.g. contraction, release of insulin) hence strongly shaping cell-environment interactions. Moreover, membrane depolarisation has been unequivocally related to cell proliferation and migration (Yang et al., 2013).

We use optogenetics (Deisseroth, 2015) to achieve controlled modulation of the membrane potential in nonexcitable cells. The optogenetic toolbox contains a vast array of genetic constructs of specific proteins enabling light induced changes of proteins conformation and of membrane permeability to specific ions (e.g. Channelrhodopsin for Na⁺, halorhodopsin for Cl⁻) and was proven valuable for manipulating cellular activity (including cell signalling and cell migration) with high spatial and temporal precision (Deisseroth, 2015), (Boyden, 2015; Packer et al., 2013). Traditionally associated with neurons and excitatory cells, thus with control of action potentials, optogenetics was shown capable to modulate as well the intracellular signaling and transcription (Eleftheriou et al., 2017; Rost et al., 2017) and was used in various cellular models (Mansouri et al., 2019) however, has not been so far implemented in electrical biosensing formats.

In contrast to existing cellular sensors that function under the re-striction of a specific cellular intrinsic gene expression processes or signal transduction cascade (e.g. using reporter proteins (Auslander and Fussenegger, 2016), opto-switches (Airan et al., 2009) or dyes (Derick et al., 2017)), thus are inherently slow responding ones, the proposed optogenetic integration enables pacing the membrane potential in a wide range of physiological rates (Lapp et al., 2017), an issue difficult to be achieved using electric and mechanical periodic stimulation as attempted in some cell based platforms (Ni et al., 2019; Pavesi et al., 2015).

The biosensing concept, implemented according to Scheme 1, harnesses the virtues of both optogenetics and label-free cellular sensing capabilities of fast bioimpedance assays. It involves the tuning of a cellular homeostasis restoring reaction induced by controlled stimulation (via lighting), to achieve reproducible control of cellular (quasi) steady state prior to sample/analyte exposure and highly increased cell reactivity to even minute concentrations of bioactive compounds, during short term exposures accompanying light stimulation.

Scheme 1.

Biosensing Concept. Optogenetic modification enables light modulation of membrane potential with subsequent fast and robust trigger of cellular response to restore homeostasy. Upon exposure, the modulation of membrane potential becomes stimulus dependent and the cellular response allows rapid assay of concentration dependent bioeffects. Cells are stimulated using an optimized illumination protocol and evaluated using time lapse, fast impedance assay to reveal the presence of bioactive compounds and the induced homeostasis changes based on the quantitative changes of the impedance: dynamics upon illumination - black curve, - under combined stimulation (chemical stressor and illumination) - red curve, and for mere analyte exposure (without lighting), as in typical unmodified cellular platforms - grey

Moreover, the proposed optogenetic modification (innovatively included in an analytic platform (Gheorghiu and Gheorghiu, 2018)) addresses the challenges of cell based biosensors in what concern the capacity to discriminate between different types of analytes, as well as the issue of cell population heterogeneity and response characteristics dependent on the particular cell type, by establishing stable cell lines, with a vigorous growth rate, good attachment to the surface and un-pretentious growth conditions towards point of need biosensing platforms. The versatility of optogenetic implementation demonstrated in various cell models (Mansouri et al., 2019) ensures adaptation of the proposed concept also for other cell lines or channel repertoires.

A Na⁺ influx is externally controlled using a sequence of light pulses and genetic modification of a nonelectrogenic cell line (e.g. HEK293 cells) to stably express ChR2 channels (cation permeant channels, directly gated by blue light and characterised by an inward Na⁺ current and traditionally associated with light induced membrane depolarisation (Nagel et al., 2005; Nagel et al., 2003)).

The characteristic dynamics of cellular response to light induced membrane potential depolarisation and Na⁺ influx is tuned to provide both an inner referential and to amplify the signaling cascade triggered by cell exposure to a bioactive analyte. While the reproducible control of cellular (quasi) steady state prior to sample/analyte exposure is achieved with careful adjustment of illumination parameters, the tuned periodic light stimulation determines fast and specific homeostasis modulation of the optogenetically modified cells, via controlled changes of membrane potential and intracellular Na⁺ concentration, key nodes in the otherwise overwhelmingly complex cell response network. Accordingly, the typical steps involved in cell based biosensing: a) reception the interaction of the analyte with a membrane receptor, b) the successive transduction to the intracellular side and activation of intracellular cascade (including second messengers amplification) and the subsequent, late, c) synthesis of effectors and generation of cellular response, are “rewired” to speed-up and amplify the, otherwise long term, bioeffects of cell active compounds.

Dynamics of cell parameters involving: (1) morphology/surface spreading and cell proliferation (Sundelacruz et al., 2009) and of cell membrane electrical parameters (Lin et al., 2009) - the key regulated variables involved in cell homeostasis subsequent to modulation of ionic fluxes, and (2) modulation of receptors and second messengers (Eleftheriou et al., 2017; Rost et al., 2017) (typical for optogenetic control) are ideally reflected by real-time, quantitative, dynamic assessment of cell-layer impedance. Electrical Impedance Spectroscopy (EIS) platforms have gained an undisputed front place for label-free monitoring of cell substrate interaction, assessment of cell behaviour such as attachment, spreading, motility (including micromotion), growth and proliferation (Asphahani et al., 2008; Ghenim et al., 2010; Giaever and Keese, 1991; Han et al., 2007; Hong et al., 2011), as well as cellular state (Gheorghiu et al., 1999; Gheorghiu et al., 2014) providing wide spread cell based biosensing platforms (Gupta et al., 2019). As a proof of concept, fast EIS (<20 ms acquisition rate) at 1 kHz is implemented using commercial cell culture chambers and impedance analysers to monitor the dynamics of cellular responses (including morphology, membrane impedance and cellular signalling) triggered by light stimulation per se (as reference) and when accompanied by exposure to bioactive compounds.

Based on their established physiologic/toxic impact, two model test compounds were considered for proof-testing the biosensing concept: ouabain and CdCl₂. Ouabain is a glycoside (with anticancer properties (Xiao et al., 2017)) active on Na⁺, K⁺ ionic balance via the inhibition of Na⁺ pump and was selected to account for bioactive compounds exerting their effect upon direct interaction with this ubiquitous enzyme. Since cellular effects connected to ouabain induced ionic homeostasis unbalance (Nguyen et al., 2007; Russo et al., 2015) include deregulation of membrane homeostasis (and resting potential changes), cell signalling, cell surface interaction as well as cell morphology changes, cell swelling, increased vesicle exocytosis, ouabain is used also to validate the effectiveness, reliability and reproducibility of optogenetic modulation of cell homeostasis, central issue to the proposed biosensing concept.

The second compound, CdCl₂, is a model cytotoxic analyte (Jarup, 2003) of occupational and environmental concern. Concentration dependent cadmium effects at the cellular level range from genomic instability, interaction with DNA repair mechanism, generation of reactive oxygen species, modification of cell proliferation, differentiation, and induction of apoptosis. Within the proposed platform, CdCl₂ influence is addressed in a concentration range reported to elicit measurable effects only after prolonged incubations (i.e. 30 μM after 7–9 h (Mao et al., 2007)).

To prove the functionality of the biosensing concept we show that: 1) the light induced perturbation of cell homeostasis is sensitively, reproducibly reflected in impedance assays through characteristic dynamics, quantitatively associated with light stimulus parameters (magnitude and rate); 2) controlled lighting provides inner, cell culture reference dynamics; 3) this dynamics is modulated in the presence of bioactive compounds; 4) dose dependent, short term exposure to a bioactive compounds reproducibly alters the characteristic (reference) cell dynamics upon illumination, promoting high sensitivity and rapid cellular response of bio-sensing relevance - the concentration dependent cellular effects are detected in ~1 h. The capability of proposed bio-sensing platform to discriminate between compounds with different physiological impact at cell level is also addressed.

2. Materials and methods

2.1. Cell line generation and maintenance

Optogenetically modified stable cell lines were generated using the FRT genomic targeting site for the reproducible insertion of single copies of cDNAs of HEK293-Flp-In System (Life Technologies, Invitrogen - HEK293-FLPN). Site specific recombination of cDNA for (Enhanced) Yellow Fluorescent - ChR2 - fusion proteins was implemented for establishment of a light sensitive HEK293 cell line dubbed “O” (optogenetic). The cDNA for the fusion proteins was derived from the plasmid pcDNA3.1/hChR2 (HI 34R)- EYFP (a gift from Karl Deisseroth -Addgene plasmid # 20940 (Zhang et al., 2007)). This construct was subcloned in the pcDNA5/FRT expression vector and the resulting plasmid was transfected into HEK293-Flp-In cells using electroporation. EYFP fluorescence is a surrogate marker of ChR2 expression (Duda et al., 2014) thus it is used, in conjunction with patch clamp experiments to periodically evaluate the level of expression and the functionality of ChR2.

To generate a positive control cell line, cDNA for ROMKI, a 2A peptide adaptor sequence, and ChR2-EYFP were implemented in the same HEK293- Flp-ln System. The purpose of introducing ROMKI (Renal Outer Medullary Potassium channel), an inward rectifying potassium channel, was to stabilize plasma membrane potential and bestow homeostatic control of cells upon repeated illumination.

“O” cells were cultured in Dulbecco-modified Eagle medium (DMEM high glucose) with 10% fetal bovine serum (FBS) and penicillin-streptomycin (100 IU/mL-0.I mg/mL), supplemented with hygromycin (100 μg/mL) while the optogenetic, positive, control cells (dubbed “G”) were cultured in the same growth medium plus hygromycin (100 μg/mL) and geneticin (300 μg/mL). As negative controls we used HEK293-Flp-In cell without any modification.

Cells were grown in a humidified atmosphere/cell culture incubator (MCO-20AIC Sanyo, Japan) with 5% CO₂ at 37 °C.

Quantitative analysis of fluorescence data (Supplementary materials, Fig. S1A) indicates that the amount of ChR2 generated for both cell lines is comparable (82 ± 2% of O cells and 81 ± 1% of G cells display EYFP/ChR2 expression).

2.1.1. Reagents

All culture media and supplements were purchased from Invitrogen.

Fresh Ouabain octahydrate and CdCl₂ (Sigma Aldrich) solutions in deionized (Millipore) water were prepared prior to experiments from stock solutions.

2.2. Light stimulation

Light stimulation was achieved using a computer controlled home built module integrating a 470 nm Rebel LED (Luxeon, Quadica Developments Inc., Canada) which delivers light pulses with power den-sities of 1.3 mW/mm² and durations selectable from 0.2 to 5 s. Within this study were considered series of stimuli comprising 10 light pulses, LP, with constant duration (set at either 1 s, 3 s or 5 s), separated by intervals of darkness (chosen at 50 s or 300 s, respectively).

2.3. Electrical Impedance Spectroscopy (EIS) measurements

The cells were cultured at a concentration of 4.5 × 10⁵ cells/mL in the individual wells of the 8W10E cultureware (IBIDI GmbH, Germany) until fully covering surface electrodes in a cell monolayer, and the experimental protocols were performed within the controlled environment of the incubator under continuous monitoring. Each well contains a circular, 250 μm diameter working electrode and a coplanar, larger counter electrode (0.49 mm² total electrode area/well). Cell seeding concentration, duration of growth and the quality of the cell monolayer formed within the wells were optimized, and checked using optical inspection (Supplementary materials, Fig. S1B) and impedance spectroscopy.

A 4294A Precision Impedance Analyzer (Agilent, Japan), interfaced with a multiplexing module (developed in house), was used for recording time series of impedance data (amplitude and phase) from the 8W 10E IBIDI culture-ware. Impedance spectra (AC potential of 100 mV amplitude with zero DC bias) corresponding to 100 Hz–100 kHz frequency range (100 frequency points with logarithmic distribution) were recorded during cell growth, before experiments and after the illumination protocol to check the degree of coverage of both working and counter electrodes and assess monolayer status. The same instrument operated in “burst mode” (i.e. externally triggered individual cycles of single frequency acquisitions at the fastest sampling rate, AC potential of 100 mV amplitude with zero DC bias) was used for single frequency fast impedance assays of cell response during illumination. The measurements were synchronized with the application of the LPs using a dedicated module developed in house. A custom developed LabView interface was used for controlling the measurement protocols, data collection and processing. This interface enabled high speed acquisition of time series of 800 measurement point cycles. To characterize cellular dynamics upon stimulation (e.g. illumination under optogenetic control and analyte exposure) that is expected to encompass morphological changes, micromotion and processes at the level of cell membrane, we performed fast, 100 Hz acquisition rate, time lapse impedance assays at 1 kHz. This frequency was proven optimal for characterization of cell-cell junctions formation (Gheorghiu et al., 2002) and morphological changes together with cellular micromotion (Arndt et al., 2004), thus was selected for the development of the sensing concept. Impedance values, modulus of the complex impedance at the selected frequency Z^* (fr, t) = |Z(fr, t)|*e^{iϕ(fr, t)} were normalized using the formula: |Z(fr, t)|_nor = |Z(fr, t)| − |Z(fr, t₀)| with t₀ corresponding to the initiation of the LPs protocol. When stated, the same normalization was applied to the imaginary part (Im[Z*(fr, t)] = − |Z(fr, t)|sin[ϕ(fr, t)]) representations: Im[Z(fr, t)]_nor = Im[Z(fr, t)] − Im[Z(fr, 0)] with t₀ corresponding to the initiation of the LPs protocol. Data analysis was achieved using OriginPro 8.5 (OriginLab, USA).

2.4. Statistical analysis

All results are expressed as mean ± SD of indicated (minimum 4) independent experiments and are represented as normalized values relative to the value corresponding to a fixed time point (e.g. onset of illumination). The statistical significance was assessed using Student’s t-test, and a p value < 0.05 was considered statistically significant (versus control).

3. Results and discussion

3.1. Real time assessment of homeostatic capacity and its controlled modulation via light pulse duration and illumination rate

The first step in the development and testing of the biosensing concept (using time lapse impedance) was to demonstrate the light control of optogenetic cells within the platform: i.e. that the application of individual LPs of distinct durations (e.g. 1 sec, 3 s, 5 s) leads to reproducible cellular responses of magnitudes dependent on illumination parameters (number of pulses, pulse duration and extent of the inter-pulse interval), as reflected by characteristic time evolutions of impedance values. According to Fig. 1, upon illumination, the characteristic dynamics contain (a) a transient decrease in impedance upon the application of the LP, followed by (b) a recovery, plateau, phase. Both depend on lighting duration: the larger the stimulus (i.e. the longer the LPs), the larger the magnitude of the transient impedance change measured versus the baseline before the LP (−50 ± 18 Ω, −85 ± 25 Ω and −125 ± 22 Ω for the 1 s, 3 s and 5 s LP, respectively) and the wider the difference in the plateau values after the LP. Scaled mirror images of the modulus are achieved for representations of the imaginary part of the impedance, thus for clarity, Fig. 1 includes only (normalized) modulus data.

Fig. 1.

Light induced cellular effects revealed by EIS data: (A) Impedance snapshots of different length illumination pulses (1 s, 3 s, 5 s) (B) Characteristic impedance response to single large (5 s) LPs. The data represent the mean evolutions for n = 7 individual pulses from independent experiments while the shaded areas represent ±SD.

The processes influencing the dynamics of impedance values upon illumination are highly complex, ranging from short time scales (e.g. associated to effects of the direct illumination of the electrodes and/or to cell membrane effects - membrane potential and/or permeability changes upon ChR2 opening), to larger time scales where cellular adaptation processes (Goold and Nicoll, 2010) occur. Since often interspersed, we performed complementary evaluations to assay the magnitude and dynamics of the individual processes. The direct illumination of electrodes placed in an electrolyte is often reported to influence electrical recordings (Kozai and Vazquez, 2015) due to the Becquerel (photovoltaic) effect. Aware of this, we evaluated the characteristics of this process for the 8W10E electrode set-up (see Supplementary Materials Fig. S2A), and confirmed that, (although present in our set-up due to unideal electrode structure), this process has limited time scales and specific, linear evolutions, distinct from the ones depicted in Fig. 1A especially in the post illumination interval.

In complementary single cell electrophysiology assays using patch clamp we confirmed, Supplementary Materials Fig. S3, both the successful cell modification and the efficient optogenetic control using LPs: i.e. specific changes of membrane properties (membrane potential and permeability) associated to ChR2 opening upon illumination. Current clamp data (Fig. S3A) confirm light sensitivity of both cell lines: light induced depolarisation is evident for both O and G cells but not for FLPN cells (parent cells, not modified with EYFP or ChR2, used as controls). These assays are periodically used as controls of the stability of ChR2 gene expression. Moreover, electrophysiology data enable assessment of the functional ROMKI channels expression in G cells (Fig. S3B), while the homeostasis restoring capacity for G cells and intracellular accumulation of Na⁺ ions and cell homeostasis perturbation for O cells are revealed by voltage clamp data (Fig. S3C).

Membrane depolarisation (Supplementary Materials Fig. S3A) and altered intracellular ionic concentrations are able to trigger complex cellular changes (including receptors and cytoskeletal reorganisations (Chifflet and Hernández, 2012)). These cellular adaptation processed (Nunes et al., 2015) are dependent on the magnitude of the induced homeostatic perturbations, thus, in the case of light perturbation, modulate the specific (nonlinear) progress of the sensing platform impedance in the “after the pulse” recovery phase. Indeed, the characteristic evolution revealed by high speed impedance time series, upon the application of single light stimuli (with LP duration ≥1 s) highlights the nonlinear cell response dynamics triggered by lighting. As presented in Fig. 1B, upon strong illumination, consequent to 5 s single LP, the cellular platform progresses towards a quantitatively different plateau value, corresponding to new cellular states reached after light stimulation, states that are dependent on the magnitude of the stimulus.

To further confirm this observation it was checked the assumption that repeated application of pulses of the same magnitude will augment the homeostatic disturbance associated to individual LPs, and will highlight the different homeostatic capacity of the stimulated cells. We applied several illumination protocols comprising a series of 10 LPs with a fixed repetition rate and monitored the related dynamics of the normalized impedance of: the parent HEK FLPN cell line (negative control cells), the positive control cells (the G cells, with a larger channel repertoire) and the ones with only optogenetic control (the O cells).

The core protocol for revealing cells’ capability to restore homeostasis and reach a physiological quasi-stationary state following light stimulation involves recording three successive phases: 1) the baseline of 5 min in the dark, 2) the controlled, repetitive light stimulation (comprising 10 LPs, with frequency of application given by controlled intervals of darkness in between the LPs) and 3) cell recovery, of 15 min duration, in the dark.

According to literature (Lin et al., 2009), the interval for full ChR2 recovery (i.e. capacity to fully respond when illuminated) is ~30 s. Therefore, in conjunction with data in Fig. 1 showing stabilization after 50 s, the interval between individual pulses was set to 50 s to ensure that after each pulse, all ChR2 channels returned to the closed state and can be activated upon subsequent illumination. As second tier, the interval between individual pulses was set to 300 s to ensure cell homeostatic restoration (if effective) in between individual pulses.

As hypothesized, following each LP, the impedance of optogeneticcally modified cells stabilizes at individual plateau values, gradually declining at each subsequent LP (−23 ± 5 Ω decline/pulse), thus after the end of the illumination phase, there is a pronounced (−130 ± 11 Ω) downward shift of impedance magnitude against the baseline. Since this specific dynamics is significantly different from the negative control cells - FLPN (parent HEK293 line, without light sensitivity) and also from the cell free electrodes (see Supplementary Materials Fig. S2C) it accurately reflects the cellular changes (due to ionic homeostasis alteration) induced by the series of 10 LP of 0.02 Hz frequency (Fig. 2A). This is further supported by Fig. S2B Supplementary Materials that highlights impedance evolutions of optogenetically modified cells upon subtraction of the dynamics of the cell free electrodes in culture medium for a 10 LP train. In optogenetically modified cells, light actuation determines an increase in intracellular Na⁺ concentrations and associated massive membrane depolarisation (Nagel et al. 2003, 2005), Supplementary Materials Fig. S3, subsequent to ChR2 mediated Na⁺ influx. At intracellular level, the increase in intracellular Na⁺ concentration is rectified via dynamic re-equilibration of K⁺ levels (outflow of K⁺ e.g. through voltage gated channels) and ATP driven Na+/K+ pump, essential components of the cell ionic homeostasis capacity. As revealed by the evolution of the optogenetic controls (i.e. the G cells, that have ROMKI additional to ChR2 thus have improved capacity to recover the resting membrane potential values after Na⁺ influx) upon illumination, the impedance changes are consistent with cell ability to balance the ionic composition (and recover the resting membrane potential values and ionic intracellular equilibria) after subsequent light pulses. The optogenetically modified cells (O cells) gradually stray with each pulse from the evolution of cells with higher homeostatic capacity (the G cells) and show, in contrast to the latter, limited capacity to recover after the LPs succession. This increased capacity for ionic homeostasis, as bestowed by design to G cells, is reflected at the end of the protocol by the notable tendency towards the initial baseline.

Fig. 2.

The homeostatic capacity and cellular impact of the magnitude of the illumination stimulus and of the interval between pulses revealed by the dynamics of impedance data during optogenetic control of HEK293 cells. (A) Time series of impedance changes during a full illumination protocol (5 min baseline, 10 LPs of 3 s each with 50 s darkness interval between them and a 15 min recove1Y after the LPs) of FLPN − control, O and G (+homeostasy) cells, (B) Time series of impedance changes during a full illumination protocol (5 min baseline, 10 LPs of 3 s each with 300 s darkness interval between them and a 15 min recovery after the LPs) of FLPN (control), O and G (+homeostasy) cells; All results are expressed as mean ± SD of indicated independent experiments and are represented as normalized values relative to a value corresponding to initiation of baseline recording. The shaded areas represent ± SD for a minimum of n = 4 independent experiments.

These distinct evolutions of the optogenetically modified cell lines assayed under the same illumination conditions reveal a) sensitivity of the recovery step (either in between the pulses and at the end of the protocol) on the light stimulus induced homeostatic unbalance and b) capacity of the proposed impedance assay to reflect cellular state and homeostasis changes induced by external actuation.

To demonstrate that upon stimulation, the cells reach indeed a novel state that reflects both the magnitude of the stimulus and the homeostatic capacity of the cells, and also to support recovery between individual pulses, the interval between successive LPs has been increased up to 300 s. As revealed by Fig. 2B, within the statistical margin, new plateau values are indeed reached after each pulse for optogenetically modified cells. The stabilization at the same level, as in the case of the 50 s darkness interval, for the cells with higher homeostatic potential (the G cells) at the end of the recovery period is a further proof that this stabilization level is an appropriate indicator of the stimulus triggered cellular effects and that a longer interval between pulses is appropriate to ensure reproducible cell stimulation and cell recovery.

As evident from data in Fig. 2B although the LPs are able to induce persistent cellular changes, seldom reported for optogenetic stimulation, the longer time interval between the pulses (i.e. 300 s) is appropriate to ensure stable cellular platform response and establishment of characteristic trends and quasi-stationary levels upon light stimulation and recovery. Moreover, data in Fig. 2 highlight that the dynamics of impedance data on optogenetically modified cellular platform depends on the magnitude of the illumination stimulus, the interval between pulses and the extent of the homeostasis restoration capability.

3.2. Platform optimisation

Having established that the proposed assay is able to reveal the magnitude of cellular bioeffects (i.e. homeostasis unbalance) of a controlled stimulus, we investigated the optimal stimulation conditions to achieve cell stimulation yet provide a suitable reference signal. This reference signal is aimed for to enable live cell referencing as opposed to cell-free electrodes reference (as in ECIS assays).

We have evaluated the illumination protocols with intervals of 300 s between light pulses for different stimulus magnitude (1, 3 and 5 s respectively) on “O” cells. For clarity, the control evolutions (for FLPN) are not represented together with “O” cells, but as Supplementary Materials Fig. S2C. As evident in Fig. 3, the gradual decline of the plateau values after the illumination protocol reflects the magnitude of the stimulus, with 1 s illuminations determining the slightest overall cellular impact. It is thus confirmed that longer inter-pulses recovery intervals are effective for cells to reach a physiological quasi-steady state, if the magnitude of the stimulus is kept to a minimum, and emphasize 1 s stimulation and 300 s inter-pulse interval as optimal for biosensing platform establishment.

Fig. 3.

Platform optimisation: establishment of pulse duration and frequency for reference stimulation based on time series of impedance changes for optogenetically modified cells as function of different illumination conditions (1 s, 3 s, 5 s) and 300 s interval between individual pulses. All results are expressed as mean ± SD of indicated independent experiments and are represented as normalized values relative to a value corresponding to initiation of baseline recording. Shaded areas represent ±SD for 5 independent experiments.

Our sensing concept is thus based on controlling the whole cellular evolution following stimulation, with particular focus on recovery trends. These trends were tested and confirmed to be cell specific, dependent on cell homeostatic capacity and the extent of stimuli modulated homeostatic disturbances. These data further endorse the proposed optogenetic modification and controlled illumination conditions as able to provide an inner, cell state specific, reference signal.

3.3. Biosensing applications of the new cell platform concept

3.3.1. Cell response dynamics in the presence of bioactive compounds

3.3.1.1. Ouabain exposure.

For testing the biosensing relevance of the proposed concept, we addressed the modulation of the Na⁺/K⁺ pump, an intrinsic key regulator of cell ionic homeostasis, specifically questioning how its controlled unbalance (via an exogenous compound) is reflected in our assay. We used as model analyte, ouabain, a specific Na pump inhibitor applied at a concentration of 5 mM (relevant for full blocking of Na⁺/K⁺ pump in 90 min assays (Russo et al., 2015)). Ouabain is applied 15 min prior to illumination protocol and maintained throughout the experimental procedure (i.e. ~1 h).

As evident in Fig. 4A, the time-lapse impedance assay reflects mild cellular changes (i.e. cell swelling) induced by mere ouabain incubation (without optogenetic stimulation green trace): i.e. a moderate upward trend in impedance data that stabilizes after ~40 min (in total) ouabain incubation. At the measurement AC frequency (i.e. 1 kHz), impedance variations, due to changes of cell morphology and membrane electrical properties superpose with the ones involving modifications of cell surface adherence, and mutually contribute to shape the overall impedance of the system. Corroborated with impedance trends of cell growth on 8W10E culture-ware (data not shown), that highlight the increase of the impedance magnitude values at 1 kHz as cells adhere to and spread on the surface of the electrodes, ouabain effect reflected by impedance data is consistent with mild cell swelling (cell layer tightening), due to efficient Na⁺/K⁺ pump inhibition (Russo et al., 2015).

Fig. 4.

Time series of impedance changes during ~70 min protocols (corresponding to 10 LPs of 1 s duration with 300 s darkness interval between them) for light stimulation (black), and A) light stimulation in the presence of ouabain - (red) and mere ouabain incubation (green) as well as B) light stimulation in the presence of CdCl₂ - 25 μM (red) and mere CdCl₂ - 25 μM (blue - control) - Normalized impedance modulus values. All results are expressed as mean ± SD and are represented as normalized values relative to a value corresponding to initiation of baseline recording. Shaded areas represent ± SD of 4 independent experiments.

In contrast to mere cell exposure to ouabain (green trace in Fig. 4A and 25 ± 12 Ω above the control), impedance data for the concomitant exposure to ouabain and to the series of lighting (LPs) highlight a rapid augmentation of cell swelling (sharp increase after the first pulse and subsequent stabilization after the 3rd pulse) corresponding to more effective ouabain inhibition of Na⁺/K⁺ pump.

Moreover, when comparing the data encompassing both the illumination and the ouabain exposure (red trace in Fig. 4A and 57 ± 12 Ω plateau) with illumination, i.e. control (black trace in Fig. 4A corresponding to −80 ± 16 Ω plateau), the full reversal of the downward impedance trend typical for mild (10 LPs of 1 s with 300 s pause in between) illumination conditions highlights the increased sensitivity of the proposed biosensing platform. The application of both lighting and exposure to the cell active compound, ouabain, completely changes the cell characteristic response versus the reference signal (control) and demonstrates efficient cell sensitization via periodic light actuation.

We further more used G cells, effective optogenetic controls due to additional K⁺ channels, in ouabain experiments. The evident (Supplementary Materials Fig. S4) recovery of impedance values for G cells after the train of pulses is noteworthy. It can be regarded as a further proof of the reliability of the assay: due to the functional K⁺ channels compensating the Na⁺/K⁺ pump inhibition in the presence of ouabain, the G cells are able to restore their ionic homeostasis. Accordingly, the seemingly modest ouabain dependent alteration of light induced dynamics for optogenetically modified cells (for both “O” and “G” cells) is thus demonstrated to be associated with augmented ouabain inhibition of the pump and analyte dependent modulation of the cell homeostasis.

3.3.1.2. Cadmium exposure.

Whereas the toxic potential of Cd (as CdCl₂) is well established (e.g. serving as cytotoxicity standard in RT cellular analyses assays (Garcia et al., 2013)), Cd effect (involving generation of reactive oxygen species, modification of cell proliferation, differentiation, and induction of apoptosis (Bertin and Averbeck, 2006)) is only notable at large time scales (>18 h) and for concentrations above 30 μM, as demonstrated with an optical sensor cell line probing the cadmium induced (time- and dose-dependent) cellular effects using fluorescence, classical ECIS and endpoint measurements via MTT (Hofmann et al., 2013), >8 h in RT cellular analyses assays (Garcia et al., 2013). Cd induced bioeffects for concentrations as low as 10 μM after >8 h of incubation were recently demonstrated (Luo et al., 2017) using real-time cell analysis (RTCA) system.

Indeed, the results on CdCl₂ exposure revealed by impedance assay without optogenetic stimulation (Fig. 4B - blue trace), show nearly constant impedance level for > I h incubation with 25 μM CdCl₂ concentration, both the analyte concentration and the duration of exposure being below the thresholds for eliciting measurable effects in standard ECIS assays. However, similar to the case of exposure to ouabain, the concomitant exposure to the chemical stressor and the illumination protocol (red trace Fig. 4B) characteristic for the proposed assay, is able to revert the typical cell response to light actuation (black trace) and reveal marked differences (above two times the standard deviation of control light illumination experiments) of dynamic response even for these, reportedly low (Garcia et al., 2013), (Hofmann et al., 2013), CdCl₂ concentration and duration of exposure, confirming efficient cell sensitization to cytotoxic compounds via periodic light actuation.

3.3.2. Capability to discriminate between bioactive compounds with different mechanism of action

The analysis was concentrated so far on impedance magnitude values (as closely related to Cell index in classical ECIS evaluation). However, extending the analysis to both parameters (magnitude and phase/imaginary part) provided by impedance assays, data in Fig. 5A and B reveal a unique feature of our concept, not typical for cell biosensors: the capacity to discriminate between bioactive compounds with different cell-interaction mechanisms. According to Fig. 5A, the comparison of impedance modulus data for the two stressors reveals different magnitudes of the cellular impact, with ouabain more potent than CdCl₂, to alter cellular response to light induced homeostasis disturbances at the used concentrations. However, the analysis based on the imaginary component of the impedance highlights opposite trends for ouabain and Cd respectively in comparison with reference (control), indicative (comparison between Fig. 5A and B) of different interaction mechanisms and compound induced cellular processes.

Fig. 5.

Time series of impedance changes during ~70 min protocols (10 pulses of 1 s duration with 300 s darkness interval between LPs) corresponding to light stimulation (black) and light stimulation in the presence of specific chemical stimuli ouabain (blue) and CdCl₂ (25 μM CdCl₂ red) (A) Normalized impedance modulus values; (B) Normalized imaginary values enable discrimination of stressor type. All results are expressed as mean ± SD of indicated independent experiments and are represented as normalized values relative to a value corresponding to initiation of baseline recording. The shaded areas represent ±SD for n = 4 experiments.

Whereas these results further confirm the feasibility of the proposed concept for enhancement of cell responsiveness with subsequent improvement of cell platform sensitivity, the potential of our biosensing concept to truly discriminate between stressors with different interaction mechanisms is worth, however, a study on its own.

As a complementary test of this observation, we performed detailed optical analysis (presented in Supplementary Materials, Fig. S5) confirming the different effect of the two compounds on the number of cellular protrusions (a direct indicator of the induced cell morphology changes).

3.3.3. Concentration dependent modulation of the characteristic dynamics upon illumination

The characteristic dose-dependent effect of Cd exposure is well established (Lopez et al., 2003; Pacini et al., 2009): at high doses Cd progressively elicits cell injury, cell death, and organ failure, while at low doses it may modulate specific mechanisms without marked cellular toxicity.

Evaluation of the proposed cellular platform’s response for different concentrations of CdCl₂ (25 μM, 50 μM and 100 μM) - Fig. 6 shows concentration dependent, reproducible changes of the characteristic (reference) dynamics for up to 90 min incubation time, and confirms that the presence of bioactive compounds with toxic potential can be assessed rapidly, reliably and for concentrations below the sensitivity threshold in classical experiments (i.e. stable, statistically relevant plateau values are recorded for 25 μM even after 30 min of incubation versus >8 h in best alternative assay (Luo et al., 2017)). Moreover, whereas the impedance assay (at 1 kHz, without illumination) shows rather limited sensitivity for concentrations equal to or above 50 μM (as evident from the recorded deviation from the baseline before LPs Fig. 6A), the sensitivity of the cellular platform is dramatically improved when applying light stimulation simultaneously with exposure to the toxic analyte (Fig. 6B - calibration curve).

Fig. 6.

Time series of impedance changes (modulus) (A) during ~70 min protocols (10 LPs of 1 s duration with 300 s darkness interval between LPs) corresponding to light stimulation (black) and light stimulation in the presence of different concentrations of CdCl₂ (25 μM - green trace, 50 μM - red trace, 100 μM blue trace). All results are expressed as mean ± SD of indicated independent experiments and are represented as normalized values relative to a value corresponding to initiation of baseline recording. The shaded areas represent ±SD for a minimum of n = 4 independent (different sensors and different cell cultures) experiments. (B) Calibration curve based on modulus data for 10 μM, 25 μM, 50 μM, 100 μM CdCl₂ concentrations for stimulated cells (with LP) and for control experiments (without LP).

The proposed biosensing platform based on optogenetically modified cells reveals concentration dependent change of both the reference dynamics during LPs and of the plateau values after recovery.

As evident from the error bars within Fig. 6A and B, the reproducibility and repeatability of the assay, performed on different ECIS sensors and for different cell cultures are proving the reliable augmented biosensing capabilities of the approach. Of note, the stable optogenetic modification of cells was periodically tested in patch clamp experiments and repeated passaging were avoided. The quality of the cell coated ECIS sensors was evaluated prior to each experiments using wide frequency range (100–10⁵ Hz) impedance assays to ensure the high reproducibility of impedance values.

4. Conclusions

Harnessing optogenetics capabilities (i.e. versatile control of cellular functions with light sensitive proteins stably expressed within the cellular membrane), this study advances a light sensitive cellular platform towards a new biosensing concept. The proposed platform involves the tuning of cellular homeostasis restoring reactions induced by membrane potential pacing (via controlled light stimulation). Thus, it enables access to an inner reference dynamics, not available for nonelectrogenic cells, while providing increased reactivity to bio-active compounds.

Upon optimisation, the biosensing platform was shown to: reflect the presence of bioactive compounds (ouabain and Cd) via changes of the reference response to light stimulation, reveal within I h cellular effects currently analysed following significantly longer (~1 day) incubation times, and support concentration and compound discrimination. The latter is enabled via the parameterized dynamics of the complex impedance components and potentially represents a step forward in alleviating the poor selectivity of living cells based biosensors.

Possible applications of the proposed biosensing platform could encompass:

1. Testing of exogenous compounds of toxicological interest (including certain metals and organic compounds - e.g. mercury and organotin compounds (Vasic et al., 2008)) as well as drugs-beta-blocking agents and even Beta-amyloid (Petrushanko et al., 2016). This is supported given the choice of model compounds in platform evaluation, i.e. Cd and ouabain, that have high analytic relevance (i.e. are model compounds for a wide range of hazards and threats associated with environment, pharmaceutics and biosecurity), have different interaction mechanisms (according to literature reports and our preliminary data Supplementary Materials) and cell physiology relevant targets (such as Na⁺/K⁺-ATPase).

2. Drug discovery applications - Using light as a model stimulus, we showed that time lapse fast impedance assays are capable to reveal with exquisite sensitivity the changes of cellular state corresponding to light induced homeostasis modulation and the way specific analytes alter cellular dynamics corresponding to homeostasis restoration. The quantitative access to this information, with high repeatability and sensitivity, warrants applicability for assessing cellular, including toxic effects of drug candidates.

3. The optogenetic field: the quantitative insight on the extent and dynamics of cellular responses induced under light stimulation can be of potential relevance for other optogenetically modified cells to determine toxicity thresholds for long-term light stimulation in conjunction with pulse durations, light irradiance or duty cycles, dynamic assessment so far of limited availability in optogenetic studies.

In conclusion, the study sets the grounds for an advanced biosensing concept that enables quantitative, real time, label free assessment of model cell dynamics in respect to different levels of homeostasis perturbation, provides increased cell reactivity and enhanced cellular responses to even minute exogenous stimuli, demonstrates capacity to discriminate between different types of analytes and analyte concentrations and enables advancement of a powerful, new approach in cellular platform development for bioanalytics.