Introduction

Cell migration plays crucial roles in many physiological processes and contributes tocancer cells invasion and dissemination. Migration strategies employed by cells changein response to the diverse environmental stimuli, such as rigidity of the substrate,molecular composition of the extracellular matrix or spatio-temporally varyingconcentrations of soluble molecules such as growth factors or cytokines. Typically,migration through/on a matrix involves the generation of cell protrusions, i.e.extensions of plasma membrane outside the cell body1. So far, differenttypes of protrusion have been identified to contribute to cell migration and invasion inspecific contexts, cell types and microenvironment2. For example,fibroblasts form either lamellipodia3 or lobopodia4according to extracellular matrix dimensionality and elasticity. Filopodia are moreexplorative structures5 and are relevant in the guidance of neuronalgrowth cones6 and endothelial tip cell during sprouting angiogenesis7. Membrane blebs instead are typical of amoeboid type of cell migrationand invasion and have been described in leucocytes8, D. discoideum9 and H. histolytica10. In lamellipodia and filopodia actinpolymerization drives forward protrusion of the plasma membrane2. Forthis reason, much emphasis has been placed on delineating molecular regulators andupstream cellular signaling of actin polymerization, which in turn control cellprotrusion formation11. However, the dynamics of cell protrusions alsoinclude their retraction. Extension and retraction should occur in a coordinated fashionin order to drive efficient cell migration12.

A challenging feature of studying protrusion dynamics is the ability to providequantitative as well as time-resolved data. The most common approach to this problem isthe use of live-microscopy on 2D adherent cells which employs different imagingtechniques such as standard wide-field, confocal or total internal reflectionfluorescence (TIRF) microscopy13,14,15. There exist advancedimplementations of these methods such as Stroboscopic Analysis of Cell Dynamics16 and fluorescent speckle microscopy, which visualizes the movement andassembly/disassembly of actin filaments in protrusive structures17.Atomic force microscopy has also been used to measure lamellipodia dynamics andthickness in adenocarcinoma cells or in migrating keratocytes18,19.

These approaches are powerful as they all allow single cell or even subcellularresolution and represent the method of choice to study protrusion dynamics. However,such methods present a few drawbacks: i) they often require complex image and/ormathematical processing to obtain quantitative results, ii) they are hardly suitable forhigh throughput studies such as biochemical functional or drug screening and iii) aresubject to cell to cell variability.

Here, we make use of a well-established technique based on the measurement of thefrequency dependent electrical impedance of cell-covered electrodes subject to a smallalternate electric current20,21. Cells adhering on the electrodes varythe impedance in a frequency dependent manner. By properly modulating the frequency ofthe current, its amplitude, the time duration of the experiment and the size andarrangement of electrodes, a number of different biological processes can bequantified21,22,23,24,25,26,27.

Here we employ the impedance reading (IR) technique to quantitatively measure protrusiondynamics and validate the results by direct comparison with quantitative data of cellsurface variation, obtained through image analysis of live TIRF microscopy. Our dataprovide insights on how lamellipodia protrusion and retraction are regulated. We presentdata directly and quantitatively linking the amplitude of the response and its kineticsto the ligand concentration. By direct comparison of microscopy data with IR data we candissect the different parts of the response into molecularly distinct events mediated byactin polymerization and myosin contraction that can be inhibited separately by specificdrugs.

Results

Impedance reading variations can be quantitatively mapped into cellprotrusion dynamics

The non-transformed mammary epithelial cells MCF10A are highly sensitive to EGFstimulation in chemotaxis28 and, when sparsely seeded on a flatsubstrate, respond to EGF stimulation by producing fast growinglamellipodia29. We investigated protrusive activity in MCF10Acells by imaging cells stably transduced with LifeAct-GFP by TIRF microscopy.Under these experimental conditions, lamellipodia were easily identifiable asrapid growing flat meshes of actin filaments1. Upon EGFstimulation, cells extended lamellipodia for about 100–400 seconds,and started then to retract, slowly reducing the adhering surface (Fig.1A, VideoS1). This behavior is clearly visible in a kymograph representation (Fig. 1B). To quantitatively measure protrusion dynamics wecalculated in each time frame the area of cells adherent to the surface. Whilethe area was stable in untreated cells, EGF stimulation caused a rapid increaseof cell area that, after reaching the maximum value, decreased (Fig. 1C).

Figure 1
figure 1

EGF stimulation induces protrusion formation and retraction.

(A) MCF10A cells were infected with pLKO.1 LifeAct-GFP, deprived ofEGF for 6 hours and kept in a humidified chamber at37 °C and 5% CO2. Cells were then imaged by means ofTIRF microscopy. The depth of the evanescent field was kept at90 nm. Cells were imaged over a time period of1180 seconds and either stimulated or not with5 ng/ml EGF. The time at which the stimulus was added was set tot = 0 seconds. Images were acquiredevery 20 seconds (VideoS1). Four time-points are reported in the figure:−60 seconds (before EGF addition, quiescent cell),120 seconds (protrusion), 260 seconds (maximumextension) and 1120 seconds (retraction). Scale bars,10 μm. (B) Kymograph of a pLKO.1 LifeAct-GFPMCF10A cell stimulated as in A. A segment perpendicular to lamellipodium wasused to monitor fluorescence intensity at each time points and then thedifferent time points were assembled. (C) Cell surface area variationobserved by TIRF microscopy of pLKO.1 LifeAct-GFP MCF10A cells stimulatedwith EGF or unstimulated in which only the medium containing EGF was added(vehicle). Dark thick lines represent the mean values while light shadesrepresent the area included between + SD and – SD.

Full size image

This approach is powerful and quantitative but it is time expensive and biased bycell to cell variability. In order to be able to observe protrusion dynamics ina more rapid and reproducible manner, we employed a method based on IR. Toquantify the impedance change over-time, a unit-less parameter is used thatevaluates the relative change of impedance with respect to a reference impedancevalue without cells and to the impedance at beginning of the measurement. Thisparameter is called the cell index (CI). An increase in cell surface areacontacting the electrodes will result in a larger impedance and consequently inan increase in cell index. Conversely, when cell diminish their adherent surfacearea, the cell-index will decrease. To better compare IR experimentalmeasurements obtained under unstimulated and stimulated conditions, cell-indexis further post-processed to obtain a quantity called Baseline ΔCell Index. Further details are given in the Materials and Methods section.

MCF10A cells generated a reproducible and specific impedance variation pattern inresponse to EGF stimulation: immediately after EGF addition CI increasedrapidly, reaching a maximum between 100 and 400 seconds and then decreasingback close to initial values (Fig. 2A). In order toprecisely compare IR and TIRF microscopy data we defined 4 quantitativeindicators, schematically explained in Fig. 2B: the timeat which the cell-index reaches its maximum indicating how fast is the responseglobally, the value of cell-index at maximum indicating the intensity of theresponse and protrusion or retraction slopes indicating the steepness of theprotrusion or retraction phases respectively (Fig.2B).

Figure 2
figure 2

Real-time evaluation of cell protrusion dynamics by IR.

(A) Baseline Δ cell index values obtained by IR of MCF10Acells stimulated or not with EGF at t = 0.(B) Cartoon showing the indicators that we use for quantification ofthe IR response to EGF stimulation. The point where the EGF response curvereaches the highest value corresponds to Maximum Cell Index and the MaximumCell Index Time (tm). The protrusion slope is calculated asthe mean slope between t0 and tm, wheret0 is the first time point after growth factor addition.The retraction slope is calculated as the mean slope betweentm and t2m, where t2m istwice tm. (C) Maximum Cell Index Time(tm) of a number of experiments performed by means ofTIRF microscopy or IR of MCF10A stimulates with 10 ng/ml of EGF.Each point represents a separate experiment. Dashes represent the meanvalues. (D) Protrusion slope, (E) retraction slope and(F) maximum value of Baseline Δ cell index curve ofMCF10A stimulated or not with 5 ng/ml EGF.***P < 0.001.

Full size image

Both IR and TIRF microscopy approaches revealed that the mean maximum cell indextime was at about 250 seconds (Fig. 2C), thusindicating that IR is indeed a reliable readout for adherent surface area.Furthermore, protrusion and retraction slopes and the maximum intensity ofresponse (maximum CI) were all significantly modulated by EGF stimulation andallow quantitative comparison between different experiments (Fig.2D,E,F).

EGF stimulation affects cell protrusion dynamics in a concentrationdependent manner

To assess the reliability of IR in detecting cell protrusion dynamics we firstperformed a control experiment to establish the linearity of the CI with respectto cell density. By increasing the number of plated cells, the shape of thecurves was similar while the maximum value of cell index raised (Fig. S1A). In particular we found that themaximum cell index values of EGF-stimulated curves increased linearly with thecell number in the range between1 × 103 and3 × 104 cells (Fig. S1C). Cell densities lower than3 × 103 produced maximumcell index values upon EGF stimulation almost undistinguishable from that ofvehicle treated cells (Fig. S1D).Therefore, we used 5 × 103cells/well in all experiments.

To become sensitive to EGF stimulation cells need a period of EGF deprivation. Weinvestigated how the length of EGF deprivation influenced the protrusiondynamics (Fig. S1B) and we observedthat short EGF deprivation time (0.5 hours) was sufficient to detectprotrusion and contraction in response to EGF addition. However, betterresponses in term of maximum cell index were achieved after an EGF deprivationperiod ranging between 2 and 8 hours (Fig. S1E). Interestingly, longer EGFdeprivation times reduced the cell ability to produce protrusions in response toEGF (Fig. S1E). Therefore we used infollowing experiments a starving period of 6 hours.

Then, we evaluated the effect of varying concentration of EGF on protrusiondynamics by employing 0.3 ng/ml to 30 ng/ml (Fig. 3A). Even at 0.3 ng/ml we were able todetect variations of impedance values. Interestingly, we observed thatprotrusion slope (Fig. 3B), retraction slope (Fig. 3C) and maximum cell index value (Fig.3D) raised with the EGF concentration reaching, however, highestvalues around 3-5 ng/ml of EGF. Thus, IR is a sensitive method todetect in real-time the EGF-induced protrusive activity in a bulk population ofcells over a wide range of concentrations.

Figure 3
figure 3

EGF affects protrusion dynamics in a concentration dependent manner.

5 × 103 MCF10A cells weredeprived of EGF for 6 hours and stimulated with increasingconcentrations of EGF. (A) IR of MCF10A cells stimulated with 0.3, 1,3, 10, 30 ng/ml of EGF. (B) Protrusion slope, (C)retraction slope and (D) maximum value of Baseline Δ cellindex in function of EGF concentration.

Full size image

IR is a suitable method to monitor protrusion dynamics induced bydifferent growth factors and in different cellular models

One of the drawbacks of IR based techniques is that impedance variations have tobe mapped to a known biological behavior. It is therefore of interest to verifywhether the analysis based on our 4 quantitative indicators of the curve can begeneralized to the use of other factors and cell lines. This largely depends onhow universal is the behavior in response to growth factors of different originsand biological functions. To verify this hypothesis we verified the shape andtype of response with other cell lines and growth factors.

First we studied the MCF10A response to the Hepatocyte Growth Factor (HGF) byTIRF microscopy. HGF induced the formation of large protrusions of polymerizedactin, visualized by LifeAct-GFP, in form of both lamellipodia and filopodia(Fig. 4A, VideoS2). In the same experimental conditions, but monitored by IR, HGFinduced the increase of CI, corresponding to the protrusion phase and thesubsequent reduction relative to the retraction phase (Fig.4B). Next, we evaluated by IR the response of HeLa cells to EGF andHGF stimulation. HeLa cells showed response analogous to that of MCF10A cellswith a maximum in the impedance variation followed by retraction at both5 ng/ml EGF (Fig. S2A)and 50 ng/ml HGF (Fig.S2B).

Figure 4
figure 4

IR of protrusion dynamics in different cellular models and growthfactors.

(A) MCF10A cells were infected with pLKO.1 LifeAct-GFP, deprived ofgrowth factors for 6 hours and kept in a humidified chamber at37 °C and 5% CO2. Cells were then imaged by means ofTIRF microscopy. The depth of the evanescent field was kept at90 nm. Cells were imaged over a time period of1780 seconds and stimulated or not with 50 ng/mlHGF. The time at which the stimulus was added was set tot = 0 seconds. Images were acquired every20 seconds (VideoS2). Four time-points are reported in the figure: −60seconds (before HGF addition, quiescent cell), 240 seconds(protrusion), 500 seconds (maximum extension) and1700 seconds (retraction). Scale bars,10 μm. (B) Baseline Δ cell indexvalues of MCF10A cells stimulated or not with 50 ng/ml HGF att = 0. (C) Baseline Δ cell indexvalues of HUVECs stimulated or not 10 ng/ml VEGF. (D)Baseline Δ cell index values of A431 cells stimulated or notwith 5 ng/ml EGF.

Full size image

Furthermore, the IR method was applied to non-epithelial primary cells– human umbilical vein endothelial cells (HUVECs) –stimulated with EGF (Fig. S2C) orVascular Endothelial Growth Factor (VEGF) (Fig. 4C). Boththe growth factors were able to induce a significant increase of impedance,generating curves similar to MCF10A and HeLa cells.

Finally we tested A431 epidermoid carcinoma cells, which harbor Epidermal GrowthFactor Receptor (EGFR) overexpression30. Both EGF (Fig. 4D) and HGF (Fig.S2D) were able to induce a significant change of impedance, yielding aresponse qualitatively similar to that obtained with other cell lines and growthfactors, showing a rapid protrusion and retraction phases after EGF stimulation.Interestingly the retraction phase did not simply recover the prestimulusvalues, but showed an overshooting behavior. Furthermore, we noted that HGFstimulation shows slightly different responses regarding the retraction slopeand the intensity after stimulus. This suggests that protrusive activitymonitored by IR is sensitive to the number of receptors on the cell surface andto the different receptor kinetics. This technique is thus also suitable forquantitative studies on the receptor-ligand dynamics in response to changes inexpression, receptor trafficking and temporally varying stimuli.

The effect of specific drugs on the response can be quantitated bycomparing response curve

The suitability of IR for signaling studies was evaluated by inhibiting molecularcomponent of signaling pathways activated by EGF and resulting in protrusionformation. Since the response to signaling is clearly induced by the presence ofEGF, we checked whether blocking EGFR would result in absence of response.

Cetuximab is a clinically approved humanized monoclonal antibody that binds andinhibits EGFR31. As expected, Cetuximab treatment of MCF10A cellstransduced with LifeAct-GFP completely abolished the response in TIRF microscopyexperiments (Fig. 5A, Video S3). Analogously, EGF stimulation didnot induce a noticeable variation in cell index in Cetuximab-treated cells(Fig. 5B) and both protrusion (Fig.5C) and retraction slope (Fig. 5D) weresignificantly reduced compared to untreated cells.

Figure 5
figure 5

Protrusion is regulated by a signalling pathway starting from EGFR andculminating in actin polymerization.

(A) MCF10A cells, stably transduced with pLKO.1 LifeAct-GFP, wereseeded and maintained in absence of growth factors for 6 hours,pretreated with 1 μg/ml Cetuximab and stimulatedwith 5 ng/ml EGF at t = 0. Time lapsemovie at TIRF microscope of a representative cell (Video S3) was recorded with interval of20 seconds. Scale bars, 10 μm.(B) Baseline Δ cell index values, (C) protrusionslope and (D) retraction slope of MCF10A treated or not with1 μg/ml Cetuximab during growth factors deprivationand stimulated with 5 ng/ml EGF, evaluated by IR. (E) Arepresentative MCF10A cell, stably transduced with pLKO.1 LifeAct-GFP, wasstimulated with 5 ng/ml EGF in presence of1 μM Latrunculin (Video S4). Scale bars,10 μm. (F) Baseline Δ cell indexvalues, (G) protrusion slope and (H) retraction slope ofMCF10A treated or not with 1 μM Latrunculin duringgrowth factors deprivation, stimulated with 5 ng/ml EGF and evaluated by IR.***P < 0.001.

Full size image

It is well established that EGF acute stimulation activates a signaling cascadethat eventually promotes actin polymerization13. To confirm thisobservation we tested whether we could inhibit cell response by blocking actinpolymerization with Latrunculin. Latrunculin is a chemical inhibitor that bindsto actin preventing its polymerization, thus resulting in a complete disruptionof actin cytoskeleton32. Treatment of MCF10A with Latrunculincaused a rapid dissolution of stress fibers and prevented protrusion formation(Fig. 5E, VideoS4) in TIRF microscopy experiments. The same result was obtained byIR. In presence of Latrunculin the cell index did not increase upon EGFstimulation (Fig. 5F). Thus both protrusion (Fig. 5G) and retraction slope (Fig.5H) were dramatically reduced.

Different types of protrusions can be detected by IR

Cells can form different type of protrusions. Among these, lamellipodia andfilopodia are the most extensively characterized in cells grown on 2D. Inparticular, EGF-stimulated MCF10A cells preferentially produce lamellipodia29. However lamellipodia formation can be shifted to filopodiaformation by the inhibition of Arp2/3 mediated actin branching33,34. By treating cells with Arp2/3 inhibitor, we showed thatlamellipodia formation in LifeAct-GFP MCF10A stimulated with EGF was completelyabolished as shown in Fig. 6A, Video S5. However, observations in TIRFmicroscopy showed that cells extended filopodia instead of lamellipodia (Fig. 6A, videoS5). Therefore, we investigated whether filopodia dynamics wasdetectable by IR. Cells treated with CK-666 responded to EGF stimulationsimilarly to untreated cells in terms of cell index (Fig.6B). Interestingly, no changes in protrusion slope were detected(Fig. 6C), while we noticed a moderate decrease ofretraction slope (Fig. 6D). These data indicate thatretraction kinetics in filopodia might be regulated differently than inlamellipodia and provide evidence that this technique is able to detect adhesiveprotrusions of different sizes and types. In addition, these results demonstratethe advantages of comparing direct TIRF microscopy observations with IRmeasurements.

Figure 6
figure 6

IR of filopodia and the effect of myosin inhibition.

(A) MCF10A cells, stably transduced with pLKO.1 LifeAct-GFP, wereseeded and maintained in absence of growth factors for 6 hours,pretreated with 100 μM CK-666 and stimulated with5 ng/ml EGF at t = 0. Time lapse movieat TIRF microscope of a representative cell (Video S5) recorded with interval of20 seconds. Scale bars, 10 μm.(B) Baseline Δ cell index values, (C) protrusionslope and (D) retraction slope of MCF10A treated or not with100 μM CK-666 during growth factors deprivation,stimulated with 5 ng/ml EGF and evaluated by IR. (E) AMCF10A cell, stably transduced with pLKO.1 LifeAct-GFP stimulated with5 ng/ml EGF in presence of 100 μMBlebbistatin (Video S6). Newlyformed lamellipodia are indicated with white arrows. Scale bars,10 μm. (F) Baseline Δ cell indexvalues, (G) protrusion slope and (H) retraction slope ofMCF10A treated or not with 100 μM Blebbistatinduring growth factors deprivation, stimulated with 5 ng/ml EGFand evaluated by IR. **P < 0.05,***P < 0.001.

Full size image

IR reveals the myosin role in the retraction phase of protrusiondynamics

Although it has been shown that lamellipodia have continuous cycles of protrusionand retraction12, retraction is much less characterized thanprotrusion, in particular for what concerns molecules and signaling pathwaysinvolved. For instance, it is unclear if the speed of retraction is directlydependent on the tension of the plasma membrane or whether a myosin-dependentactive contraction is involved35,36. We therefore investigatedif the retraction phase in our model was dependent on myosin activity, treatingMCF10A cell with Blebbistatin, an inhibitor of myosin-II ATPase activity37. As expected, the presence of Blebbistatin did not impairEGF-induced lamellipodia extension, which is known to be a process that does notrequire myosin activity. However, lamellipodia were not retracted after theirformation and instead were maintained extended during the observation period(Fig. 6E, VideoS6). Similarly, Blebbistatin treatment on MCF10A induced a sharpincrease of cell index monitored by IR after EGF stimulation. However, afterhaving reached the maximum value, cell index values remained stationary (Fig. 6F). Consequently, treatment with Blebbistatin did notreduce protrusion slope (Fig. 6G), while completelyabolished the retraction slope (Fig. 6H). These dataindicate that retraction phase monitored in real-time by IR is a processmediated by myosin contraction and seem to rule out actin de-polymerization asthe sole responsible mechanism for protrusion retraction.

Discussion

Lamellipodia, together with filopodia, are the most frequently observed protrusivestructures in cells migrating in a 2D environment1. Here we reportquantitative measurements of protruding activity of epithelial cells upon growthfactor stimulation by means of IR techniques. By directly comparing TIRF microscopyexperiments and IR measurements we were able to dissect the impedance response intodistinct phases that can be separately altered by specific drugs. Indeed, formationand retraction of cell protrusions exactly coincide with the increase and decreaseof IR signal, respectively. The dynamics and extent of the process can be preciselyquantified and it is analogous to that detected by TIRF microscopy or reported inliterature13,38,39.

The IR method is effective to monitor protrusive activity induced by severalpro-migratory factors, such as EGF, HGF and VEGF and is applicable to differenttypes of cells. Furthermore, this method is suitable for functional study ofsignaling pathway involved in protrusive activity and, potentially, in cellmigration. For example, an EGFR blocking antibody completely inhibited the effectsof EGF on protrusion formation.

EGF was shown to activate the small GTPases Rac1 and Cdc42 at the leading edge40 and Rac1 activation was shown to be downstream of MAPK pathway uponEGF stimulation in the formation of lamellipodia13. The ultimateeffect of these signaling cascades is the activation of actin polymerization at thefront of growing lamellipodia or filopodia. Our results confirm that impedancevariations detected after EGF stimulation are indeed totally dependent on actinpolymerization, in agreement with published evidence1. Interestingly,thanks to IR we are able to detect not only lamellipodia but also filopodia, asshown in cells treated with an Arp2/3 inhibitor.

What molecular players act in the phase of retraction of the lamellipodium is moredebated. Although lamellipodia do not contain myosin41, it has beenreported that at the peak of protrusion myosin II filaments form in lamella, a morestable region of cellular protrusions, driving the formation of actin-arc and thenshrinking protrusions12. Our results indicate that, upon growthfactors induced protrusion, retraction phase is determined by myosin contraction.This suggests that myosin activity might be important not only in the cell tailretraction but also in the leading edge dynamics. A critical role for myosin inperiodic contractions at the leading edge has been reported, where myosin II pullsthe rear of the lamellipodial actin network, causing upward bending, edgeretraction and initiation of new adhesion sites42. Here we showedthat after stimulation with growth factors, similar lamellipodia dynamics areobservable. It is worth noting that protrusion and contraction are synchronized inthe population due to the starving, thus making the response particularly clean andreproducible.

The response to a soluble cue, such as a growth factor, is particularly relevantduring directed migration of mesenchymal and epithelial cells. How lamellipodiadynamics contribute to directional cell migration is not completely known, howeververy recently it has been shown that localized Myosin II inactivation provides theasymmetry of force needed for directional migration of mesenchymal cells43.

Interestingly, it has been shown that lamellipodia are critical protrusive structuresalso in 3D migration2,4. A recent paper44 introducesthe possibility of integrating microfluidics with impedance reading to measurechemotaxis in 3D matrices. Although this kind of application of impedance basedassays are presently limited by the need of cell-electrode contact, we cannotexclude interesting future progress in this field.

In conclusion we have presented quantitative measurements on protrusion dynamics thatmap distinct response phases as observed in TIRF microscopy to separate regimes ofimpedance variation in IR measurements. Such clear and direct correspondence makesthe interpretation of drug treatments and genetic modifications straightforward.Furthermore, we present data on how the retraction of the lamellipodium producedupon acute growth factor stimulation is driven by myosin and rule out thepossibility that this is solely driven by actin de-polymerization.

Materials and methods

Cell lines

MCF10A (CRL-10317), 293T (CRL-11268), A431 (CRL-1555) and HeLa (CCL-2) cell lineswere obtained from ATCC resource center ( http://www.atcc.org). All experiments were performed on celllines passaged for <3 months after thawing. 293T, A431 and HeLa cellswere cultured in Dulbecco’s Modified Eagle’s Medium,DMEM, (Sigma-Aldrich, St Louis, MO, USA). The culture medium was supplementedwith 10% fetal bovine serum (Gibco, Life Technologies, Rockville, MD, USA),200 U/ml of penicillin and 200 μg/mlstreptomycin (Sigma-Aldrich, St Louis, MO, USA). MCF10A cells were cultured asdescribed45. Human EC were isolated from umbilical cordveins, characterized and grown in M199 (Sigma-Aldrich, St Louis, MO, USA)containing 20% fetal bovine serum (Gibco, Life Technologies, Rockville, MD,USA), bovine brain extract, heparin (50 μg/ml,Sigma-Aldrich, St Louis, MO, USA) and penicillin-streptomycin(200 U/ml, Sigma-Aldrich, St Louis, MO, USA) on gelatin-coatedtissue culture dishes, as previously described46. Human umbilicalcords were kindly donated by O.I.R.M- S. Anna Hospital (Agreement n.619,19/06/07) with prior written informed consent.

Reagents

Human Epidermal Growth Factor (EGF) and Vascular Endothelial Growth Factor(VEGF)-A165 were purchased from R&D Systems, Minneapolis, MN, USA. HumanHepatocyte Growth Factor (HGF) was purchased from PeproTech, Rocky Hill, NJ,USA. Cetuximab was purchased from Merck KGaA, Darmstadt, Germany. Latrunculin Aand CK-666 from Sigma-Aldrich, St Louis, MO, USA. Blebbistatin from Calbiochem,San Diego, CA, USA.

Impedance reading (IR) of protrusion dynamics

IR is a label-free non-invasive technique based on the measurement of thefrequency dependent electrical impedance of cell-covered electrodes subject toan alternate small electric current. This method was originally invented anddeveloped by Giaever and Keese20,21. IR methods have beensuccessfully applied to the study of cell adhesion22,47,48,cell proliferation and viability49, cell migration50, trans-endothelial invasion51, wound healing52,cell-cell adhesion26, apoptosis24 andcytotoxicity25.

The theoretical details of how this method works are detailed elsewhere21, here we only give a succinct description of the experimentalsystem. Briefly cells adhering on top of the electrode can vary the impedance ofthe electrode by increasing the measured voltage drop either influencingparacellular (avoiding the cell) or transcellular (passing through the cell)currents. The relative importance of these two contributions depends on thefrequency of the applied current. In the settings of the commercial system weare using (xCELLigence), the impedance is mostly due to transcellular currents.Indeed the impedance values obtained at intermediate frequencies (in the 1 to10 kHz range) depend both on the fraction of the electrode areacovered with the spreading cells and on the space between the electrode and thebasal cell membrane. On the contrary at higher frequencies (in the 10 to50 kHz range) the capacitive contribution to the impedance isdominating. Thus in this regime the measurements of impedance reflectessentially the fraction of the electrode covered with cells, thereby being alarge scale substitute for area measurements in microscopy experiments53.

There exist a number of different applications to this method in cell biology andseveral commercially available setups including ECIS (Electric Cell-substrateImpedance Sensing) from Applied Biophysics, xCELLigence (Real Time Cell ElectricSensing; RT-CES) from Acea Biosciences, the one used in this paper and Cellkey(Cellular Dielectric Spectroscopy; CDS) from Molecular Devices.

Experiments were performed using the xCELLigence RTCA DP instrument (ACEA, SanDiego, CA, USA) which was placed in a humidified incubator at37 °C and 5% CO2. IR of protrusion dynamics wasperformed using modified 16-well plates with micro electrodes attached at thebottom of the wells (E-plate, ACEA, San Diego, CA, USA). Cells were seeded oneday before the assay in culture medium. The following day complete medium wassubstituted with 100 μl/well of culture medium withoutgrowth (DMEM supplemented 200 U/ml of penicillin and200 μg/ml streptomycin). After 6 hours ofEGF deprivation, the E-plates were placed on the RTCA DP instrument and theimpedance value of each well was automatically acquired by the xCELLigencesystem and expressed as a Cell Index value. Impedance was recorded every 5seconds for one hour. At least two technical replicates of each experimentalcondition were used in each biological replicate. After 2 minutes,growth factors were added with an additional volume of100 μl/well. Inhibitors were added one hour beforegrowth factors addition during to the EGF deprivation period and then addedtogether with growth factors to maintain the same final concentrations. In orderto correctly take into account the baseline value of impedance, the value of thefirst time-point after EGF addition was subtracted to the cell index values.Furthermore, point-by-point subtraction of the values of unstimulated cells wasperformed on each curve, to obtain the Baseline Δ Cell index values.The Slope of Baseline Δ Cell index curves in a chosen Time periodwas calculated by fitting the points to a straight line. The protrusion slope iscalculated as the mean slope between t0 and tm,where t0 is the first time point after growth factor addition andtm is the time point at which Baseline Δ Cellindex reaches the maximum value. The retraction slope is calculated as the meanslope between tm and t2m, where t2mis twice tm.

Lentivirus production

LifeAct-GFP was kindly provided by Roland Wedlich-Söldner, Max PlanckInstitute of Biochemistry, Martinsried, Germany and was inserted in pLKO.1lentiviral vector in the place of puromicin resistance sequence as previouslydescribed54. Lentivirus were produced by calcium phosphatetransfection of lentiviral plasmids (pLKO.1 LifeAct-GFP) together with packaging(pCMVdR8.74) and envelope (pMD2.G-VSVG) plasmids in 293T cells as previouslydescribed55. Supernatant was harvested 24 and48 hours post-transfection, filtered with0.45 μm filters, precipitated (19000xG for2 hours at 20 °C) and suspended in PBS at ahigher concentration. The multiplicity of infection (MOI) was determined byinfecting HeLa cells in presence of 8 μg/ml of polybreneand the quantification of GFP-positive cells was performed by flowcytometry.

LifeAct-GFP dynamics with TIRF microscopy

MCF10A cells were infected with pLKO.1 LifeAct-GFP using a M.O.I. equal to 2.Then cells were seeded at low density on glass bottom plates (Porvair, Norfolk,UK) coated with 1 μg/ml fibronectin (Sigma-Aldrich, StLouis, MO, USA). After an overnight starving in medium w/o growth factors (DMEMsupplemented 200 U/ml of penicillin and200 μg/ml streptomycin), cells were placed on aninverted microscope equipped with a 37 °C humidifiedchamber with 5% CO2 and visualized using True MultiColor Laser TIRF Leica AMTIRF MC (Leica Microsystems, Wetzlar, Germany) equipped with a 63X oil immersionobjective (HCX PL APO 63x/1.47 OIL CORR TIRF) and Hamamatsu EM-CCD cameraC9100-02. Images were acquired through Leica LAS AF6000 modular system software.The depth of the evanescent field was kept at 90 nm. Time lapsemovies were performed with a 20 seconds interval and EGF or HGF wereadded at the 4th frame. Cell area was calculated usingImageJ56.

Statistical analysis

For scatter plot representation, the central line depicts median values. All theremaining data are represented as mean value of both technical and biologicalreplicates, while error bars report the standard deviation. Statisticalsignificance was determined by Student’s t-test.

Additional Information

How to cite this article: Gagliardi, PA et al. Real-timemonitoring of cell protrusion dynamics by impedance responses. Sci. Rep.5, 10206; doi: 10.1038/srep10206 (2015).