Applications of graphene to cell biology 1

By sweeping the gate voltage to measureelectrical parameters such as Dirac peak and motility, a large shift in Diracpeak is observed when live cells are added onto the device.. 96 4.20 E

APPLICATIONS OF GRAPHENE TO CELL

BIOLOGY

NICOLAS BOICHAT BSc, MSc, EPFL

A THESIS SUBMITTED

FOR THE DEGREE OF DOCTOR OF

PHILOSOPHY

NUS GRADUATE SCHOOL FOR INTEGRATIVE

SCIENCES AND ENGINEERING

NATIONAL UNIVERSITY OF SINGAPORE

2014

I hereby declare that this thesis is my original work and it has beenwritten by me in its entirety I have duly acknowledged all the sources of

information which have been used in the thesis

This thesis has also not been submitted for any degree in any university

previously

Nicolas Boichat

13th March, 2015

1.1 Graphene 1

1.1.1 Fabrication 3

1.1.2 Electrical properties 4

1.2 Graphene in cell biology 11

1.3 Use of graphene as a sensor 12

1.4 Force sensing 14

1.5 Cell counting 16

1.5.1 Electrical methods 17

2 Graphene and cells 21 2.1 Cells on fibronectin-coated graphene 22

2.1.1 Materials and methods 22

2.1.2 Results 25

2.1.3 Discussion 27

2.1.4 Conclusion 29

2.2 Graphene as a substrate modifier 30

2.2.1 Materials and methods 31

2.2.2 Results and discussion 32

2.2.3 Potential future work 33

2.3 Graphene functionalization 40

2.3.1 Materials and methods 40

2.3.2 Results and discussion 41

2.3.3 Potential future work 47

3 Early sensor attempts 49 3.1 Capacitive 53

3.1.1 Numerical values for PDMS dielectric 53

3.1.2 Sample device 55

3.1.3 Potential future work 57

3.2 Piezoresistive film 58

3.3 Graphene as strain gauge 60

3.4 Piezoelectric film 61

3.4.1 PVDF 62

3.4.2 Experimental results 63

3.4.3 Potential future work 63

3.5 Quartz crystal micro-balance (QCM) 65

3.6 Electrical interface 70

4 Cell sensing device 73 4.1 First device design 74

4.1.1 Materials and methods 74

4.1.2 Results 87

4.1.3 Discussion 103

4.2 Inverted device design — cell counting 110

4.2.1 Material and Methods 110

4.2.2 Results 113

4.2.3 Discussion 115

4.3 Conclusion and future work 120

5 Conclusion 123 Bibliography 127 A Cell migration analysis framework 139 A.1 Segmentation and tracking 139

A.2 Motion analysis 140

A.2.1 Motion characterization 143

B Large cells: Amoeba 149 B.1 Motility and substrate affinity 150

B.2 Forces exerted on the surface 153

B.3 Fixed staining 155

B.4 Future work 157

C Effect of localized forces on cells 159 C.1 First device design 160

C.1.1 Experiments with cells 163

C.2 Second device design 165

C.2.1 Experiments with cells 167

C.2.2 AFM characterization 168

C.3 Conclusion and future work 173

In this thesis, I investigate potential uses of graphene in cell biology.Graphene is a novel 2-dimensional material, composed of carbon atom ar-ranged in a honeycomb lattice structure Graphene has a number of ex-citing properties, and, in this work, I make use of its transparency, surfaceproperties, and ability to sense, electronically, molecules on its surface

I first look into graphene as a substrate for epithelial, mammalian cells.This is an important step in order to establish if graphene can be usedfor sensing applications, which is our ultimate goal Transferring grapheneonto a flexible polymer substrate (PDMS), I show that cell motility ongraphene, coated with fibronectin, is not significantly different from thebaseline PDMS substrate However, when graphene is transferred to glass,and left uncoated, cells show a dramatic preference for glass, probably due

to the hydrophobic nature of graphene I then show how it is possible toattach molecules onto the graphene sheet using an existing pyrene-basedmethod, and quantify the number of target molecules attached as a function

of concentration

Knowing that graphene does not perturb cells in a significant manner,

I move on to the original idea of this thesis, which consists in creating agraphene-based force sensor Measuring forces exerted by cells on their

substrate is an important biological question, and current approaches allhave limitations Using graphene’s transparency and remarkable electronicproperties, I hoped to build a device that would provide electrical readout offorce information, while allowing for optical imaging of cells Five differentapproaches were considered, some did not leave the design phase, whilelarge prototypes were built for others However, all of these approacheshad significant issues, especially when scaled to lower sizes.

Despite this drawback, the techniques I developed allowed to create amuch simpler device, that produces interesting results: a graphene-baseddevice could be used as a simple cell counter A large piece of graphene istransferred to a coverslip, connected with electrodes, and put in a Petri dishwith a platinum gating electrode By sweeping the gate voltage to measureelectrical parameters such as Dirac peak and motility, a large shift in Diracpeak is observed when live cells are added onto the device I proceed toeliminate possible reasons for this device response, and hypothesize that thedevice is detecting charged proteins produced by cells, that get adsorbedonto the graphene surface

Finally, I show that the device response depends on the number of livecells in the dish This means that the device could be used as a simplecell counter, that could measure cell metabolism and viability without anymanipulation of the cells (such as splitting, staining), or requiring directcontact with the cells

List of Figures

1.1 Graphene gate voltage vs resistance/conductance 5

1.2 Graphene back-gating 5

1.3 Graphene gate voltage vs resistance/conductance curve, with fitted parameters 9

1.4 Electrolyte gating of graphene 10

1.5 Effect of charges on graphene vs bulk material 13

2.1 Glass bottom dish with PDMS and graphene 23

2.2 IEC-6 cells on graphene/PDMS 24

2.3 Average speed on graphene vs PDMS vs glass 26

2.4 Immunostaining of IEC-6 cells on graphene vs PDMS 27

2.5 Immunostaining of IEC-6 cells on graphene vs PDMS — data analysis 28

2.6 Patterned graphene design to be written with electron beam 32 2.7 Timelapse of cell growth and motion on patterned graphene 35 2.8 HeLa cells after 32h on graphene pattern on glass 36

2.9 Cells on a 5µm glass/graphene pattern 37

2.10 Fixed HeLa cells after 48h on pattern 38

2.11 Fixed HeLa cells after 48h on high-resolution pattern 39

2.12 Emission spectrum of pyrene 42

2.13 Emission spectrum of streptavidin with Alexa 647 42

2.14 Functionalized graphene fluorescence images, 10x 44

2.15 Functionalized graphene fluorescence images, 40x 45

2.16 Pyrene concentration vs Pyrene fluorescence 46

2.17 Pyrene concentration vs Streptavidin fluorescence 47

3.1 Basic 2 layer grid sensor design 50

3.2 3D-printed “table” to apply a controlled pressure on the device 55 3.3 Capacitive readout on sample device 56

3.4 Capacitive design allowing measurements of shear forces 57

3.5 PDMS mixture with Carbon Black 59

3.6 Graphene as a strain gauge 61

3.7 Sample PVDF shock sensor from Piezotech 64

3.8 Charge amplifier circuit 64

3.9 Sample output from the sample shock sensor 64

3.10 Frequency response of a quartz crystal 66

3.11 4.433619 Mhz quartz crystal picture 68

3.12 Interface between the device and measurement equipments 71

4.1 Device fabrication steps, after graphene transfer to a coverslip 76 4.2 Electrolyte gating of graphene 77

4.3 Custom-built Processing software screenshot 78

4.4 Measurement circuit setup 79

4.5 Sample gate-voltage vs resistance curves for one experiment 81 4.6 Graphene gate voltage vs conductance curve, with fitted pa-rameters 83

4.7 Sample processed results for one experiment 84

4.8 Device in Biostation 86

4.9 Raman spectroscopy of graphene on glass 88

4.10 Device response to live cells 89

4.11 Control experiment: increase pH 90

4.12 Control experiment: decrease pH 91

4.13 Control experiment: pH vs Dirac peak position 91

4.14 Device response to Poly-lysine 92

4.15 Device response to Poly-glutamic acid 93

4.16 Device response to SDS 93

4.17 Device response to cells, in serum-free medium 95

4.18 Device response to FBS, then cells 95

4.19 Device response to dead cells 96

4.20 Experimental setup to verify device response to cells that are not in direct contact with graphene 97

4.21 Device response to cells that are not in direct contact with graphene 97

4.22 Comparison of device response to cells that are in direct contact with graphene, or not 98

4.23 Device response to conditioned media 99

4.24 Device response to conditioned media: concentration curve 99 4.25 Device response to conditioned media: replacement 100

4.26 Device response to cells: media replacement 101

4.27 Device response to different densities of cells 102

4.28 Analysis of device response to different densities of cells 103

4.29 Probable explanation for mobility change after adding serum 106 4.30 Probable explanation for Dirac peak after adding cells 106

4.31 Inverted device design 111

4.32 Inverted device photo 112

4.33 Results of a number of experiments on an inverted device 114

4.34 Computed precision of the cell counting device 117

A.1 Example of cell tracking using nuclei staining 140

A.2 Example of Mean Square Displacement curve 142

A.3 IEC-6 cells, graphene substrate 144

A.4 IEC-6 speed for 14 points, on 3 substrates 146

A.5 Average speed on 3 substrates 148

B.1 Images of Chaos cells on 3 different substrates 151

B.2 Images of a Chaos cell at higher magnification 152

B.3 Chaos Carolinensis on 1:80 PDMS with fluorescent beads 154

B.4 Displacement of a single bead due to substrate deformation cause by Chaos Carolinensis 155

B.5 Fixed staining of Chaos Carolinensis 156

C.1 First design of PVDF device to apply forces on cells 161

C.2 Z displacement of the piezoelectric device 162

C.3 Stitched image of the actual device 162

C.4 Hela cells on device, showing gradual death of cells 164

C.5 Second design of PVDF device: electrodes are on one side of the PVDF film only 165

C.6 Circuit used to actuate the PVDF film 166

C.7 HeLa cells seeded on the second PVDF device design 167

C.8 HeLa cells after pulses have been applied to the device 168

C.9 Deflection of the AFM tip when a pulse is applied to thedevice 170C.10 Deflection of the AFM tip vs current applied 171C.11 Initial deflection and acceleration of the AFM tip 172

List of Tables

1.1 Comparison of commercial cell counting techniques 20

3.1 Comparison of force sensor approaches 52

3.2 PDMS-based sensor capacitance vs thickness/area 54

3.3 Capacitance of PVDF-based sensor 65

3.4 Frequency shift of quartz crystal under load 69

Graphene has been studied theoretically for sixty years, and was thought

to be too unstable to exist on its own It could only be investigated as part

of larger 3D structures, for example, on top of matching lattice crystal [37]

It is only in 2004 that graphene was first isolated, using a simple tape” exfoliation technique from a block of graphite [87] This enabledthe authors to perform electrical measurement on the material, observing

“Scotch-a g“Scotch-ate volt“Scotch-age dependence of the gr“Scotch-aphene sheet conduct“Scotch-ance

Graphene has a number of remarkable electrical, optical and mechanicalproperties [37, 66, 82], and this thesis makes uses of a number of them, in

an attempt to uncover possible applications of graphene to cell biology

First, it is an excellent electricity conductor, with a sheet resistance

of 280 Ω per square [63] or lower The resistance can be modulated byapplying a voltage perpendicularly to its surface [87]: this is covered indetails in Section 1.1.2

It is also highly transparent to light, a single layer absorbing only 2.3%

of white light [82] This is very interesting for biology applications, as thislimited absorption does not perturb microscopic imaging However, it alsoexhibits a strong quenching effect on fluorescent molecules [62], which cancause problems with biology experiments (see Section 2.3)

Its thickness makes it basically invisible, topologically, especially if weconsider applications in the field of biology: a step of 3–4 ˚A is noticeableonly using sub-micrometer microscopy techniques such as AFM or elec-tron microscopy [87] Mechanically, it is also the strongest material evermeasured [66], making it a good candidate as a component of compositematerials [36]

In terms of surface properties, graphene is hydrophobic [95], a property

I use in Section 2.2 It is normally chemically highly inert [86], but can

be functionalized using techniques borrowed from carbon nano-tubes tionalization such as pi-pi stacking of pyrene rings [49, 53] This can be ofinterest in the context of biosensors, and the functionalization process isdiscussed in Section 2.3

func-Finally, without going into details, graphene also has remarkable mal properties, allows experiments on quantum relativistic phenomena, and

ther-is of prime interest in spintronics [36] and photonics [13]

1.1.1 Fabrication

The “Scotch-tape” technique, gradually detaching thinner and thinnerflakes of graphene off a graphene block, produces high quality samples, idealfor the study of graphene properties However, only small, micrometer-sized, graphene flakes can be obtained, and even minute quantities requiresignificant manual effort, which makes the technique unsuitable for mostapplications [3]

Later advances are able to produce large scale graphene films on per [73] or nickel [63], using chemical vapor deposition (CVD) These filmsare not single-crystalline anymore [51], and do have wrinkles [73], as well

cop-as arecop-as covered by multiple layers of graphene [97] Therefore, their tronic and optical properties are not as good as exfoliated graphene [88].Despite its disadvantages, the technique allows large graphene sheets to beproduced, in a quasi-industrial manner [9], and now CVD-grown graphene

elec-is even available for purchase online [40] Thelec-is opens the door for graphene

to be used in real-world devices, e.g in high-speed electronic circuits, or

as new material for electrodes in touchscreens and solar cells, replacingindium tin oxide [3]

A different approach to this large-scale production problem is to startwith a graphene oxide solution, that is dried up on a surface, followed by

a reduction step [21] The resulting material, however, does not exhibitvery good conductivity and/or transparency (order of kΩ at best if a goodtransparency is required [3]), due to a large number of defects

Since this work does not require clean single-crystalline graphene, andcan tolerate multi-layer graphene as well as defects, all the experimentspresented in this thesis makes use of commercial, CVD-grown graphene

1.1.2 Electrical properties

The report of graphene’s discovery in Novoselov’s seminal paper [87]sparked a lot of interest, in a large part because of its electrical behaviorwhen an electric field is applied perpendicularly to its surface [3]

To probe the electronic properties of graphene, one varies the dicular electrical field, while measuring the graphene sheet resistance Theresulting bell-shaped curve (Figure 1.1) was first used to understand fun-damental properties of graphene, and is now commonly used, along withother techniques such as AFM and Raman spectroscopy, as an indicator ofgraphene’s quality [3]

perpen-There are 2 main methods to apply this perpendicular electrical field.The first one, called back-gating [90], applies a voltage through a thininsulator This method was used in Novoselov’s paper [87], and is still themost commonly used technique In this thesis, I use an alternative method,where an electrolyte is used to create charges directly on the graphenesurface This is discussed in Section 1.1.2.2

The setup used for back-gating is illustrated in Figure 1.2 It startswith a conducting n+-doped silicon wafer, covered with a layer of insulatingsilicon dioxide (300 nm thick) Graphene is then transferred to the wafer,and gold electrodes deposited using lithographic techniques [87]

Measuring the graphene sheet resistance, with a 4-probe setup to inate contact resistance, and varying the gate voltage Vg applied on thedoped silicon wafer, we can observe a large change in resistance R, asshown in Figure 1.1a

elim-Plotting the inverse to the resistance, that is, the conductance, as tion of the gate voltage (G(Vg)), we obtain a gate voltage-conductance

(b) Gate voltage vs conductance

Figure 1.1: Graphene gate voltage vs resistance and conductance, typical

Graphene1st electrode

Doped silicon wafer

Figure 1.2: Simple schematic of graphene back-gating The gate voltage isapplied to a doped silicon wafer, insulated from the graphene sheet by athin layer of silicon dioxide While the gate voltage is modified, thegraphene sheet resistance is observed (The actual setup usually provides

4 electrodes on the graphene sheet, to allow for 4-probe measurements.)

graph exhibiting a characteristic V-shape, with a minimum conductance atthe Dirac peak, and linear slopes on each side corresponding to hole andelectron mobilities [18], as shown in Figure 1.1b.

The Dirac peak (Vmax) corresponds to the point where there is no netfield applied to the graphene sheet [87] On each side of this curve, weobserve linear regions, that can be used to compute the electron and holemobilities [116], that is, the ability of these charged particles to move ingraphene when an electric field is applied on them

From established physical models [87], we know that:

Merging the 2 equations above, we get:

G(Vg) is the conductance per unit area1 Experiments measure theconductance G(Vg) and gate voltage Vg at all times, and the conductanceper unit area can be computed based on the device geometry

In the back-gating case, Cg can easily be computed from the dielectricconstant of the insulating silicon oxide layer k, divided by its thickness d

a 2D material, area is used.

(parallel plate capacitor: Cg = k · 0/d).

This linear relationship between conductance and gate voltage, definingthe mobility, is expressed as follow:

dG(Vg)

It is only valid for certain gate voltages, away from the “puddle” regimenear Vmax, but close enough to this point as not to reach non-linear portions

of the conductance-voltage curve [116]

From the measurements of gate voltage Vg vs conductance G(Vg) (orresistance R(Vg)), 3 parameters can be fitted: Vmax, the gate voltagecorresponding to the minimum conductance (Dirac peak); and µ0h+ /e − =

Cg· µh+ /e −, mobility of the graphene times gate capacitance The mobility

is composed of 2 parts, hole (h+) and electron (e−) mobility corresponding

to fitted values respectively below and above the Dirac peak gate voltage

where R(Vg) is the graphene resistance; Vg is the applied gate voltage;

Rmax, Vmax are respectively the maximum resistance, and the gate age corresponding to the maximum resistance; and V0 is a parameter thatdepends on the mobility of the graphene, and the gate capacitance, whichbasically defines the “width” of the curve The mobility µ can be extracted

volt-from this value by taking the derivative of the conductance for large values

of Vg (Equation 1.4), ignoring contact resistance:

elec-The second technique is simpler, and fits the Dirac peak Vmax as theminimum of the conductance curve, and electron/hole mobility using linearfits of the conductance curve [18] A sample fit of experimental data isshown in Figure 1.3 Since this is the method used in this thesis, detailsare provided in Section 4.1.1.2

1.1.2.2 Back-gating versus electrolyte gating

As mentioned earlier, most electrical characterization of graphene isperformed using back-gating technique, that is, with graphene transferred

to a thin insulated layer of silicon dioxide grown on top of a conductingsilicon wafer [87] Because of the relatively small capacitance of the silicondioxide layer, large gate voltages need to be applied to obtain a full conduc-tance/gate voltage curve, with a gate voltage range close to 200 V [37, 87].However, this technique benefits from a well-characterized gate capacitance,allowing for an easy computation of the graphene mobilities [18]

An alternative to this approach is to gate the graphene sheet using anelectrolyte, from the top of the graphene [90, 92] (see Figure 1.4) In this

(c) Gate voltage vs conductance with fitted Dirac peak and mobility

Figure 1.3: (a) Graphene typical gate voltage-resistance curve (electrolytegating), with a maximum resistance at the Dirac peak position (Vmax).(b) Inverting the resistance, we obtain a gate voltage-conductance curve,that exhibits linear regions on each side of the Dirac peak, which can befitted (c) to obtain hole and electron mobilities as well as Dirac peak

position

case, charges are able to approach to a very short distance of the graphenesheet, about 1 nm [17], leading to a very large gate capacitance This meansthat a large number of charges can be generated with small gate voltages.Pachoud et al [92] use a polymer electrolyte to gate graphene to very highdoping levels, which would be impossible with conventional silicon dioxidegating Ohno et al [90] shows that it is possible to use biocompatiblebuffer solutions to gate graphene, using a Ag/AgCl reference electrode.Furthermore, they compare back-gating to electrolyte gating, and showthat the conductance/gate voltage curve can be obtained with a very smallrange of voltages, around 200 mV, compared to ≈30 V with back-gating.This technique is of particular interest in cell biology, as cell media andbuffers are always electrolyte solutions [33], and allows us to gate grapheneusing small voltages that are unlikely to perturb cells This realization isthe basis of the results presented in Chapter 4.

+

+ -

-++

-++

1.2 Graphene in cell biology

Graphene has only been the focus of limited interest as a substrate inthe context of cell biology, with most work investigating toxicity or generalcell behavior and differentiation

Graphene nanosheets have been shown to be harmful to bacteria [2],and small graphene oxide sheet cause damage to red blood cells and fibrob-lasts [76] However, the particle size modulates the toxicity [76], making itprobable that the effect is mechanical in nature: graphene in itself may not

be toxic, but the sharp edges of the graphene flakes damage cells [2] Inthis thesis, I use graphene only as a substrate, therefore these results arelargely irrelevant

Two papers indicate that a graphene substrate can induce stem celldifferentiation under specific conditions Nayak et al [83] first show thathuman mesenchymal stem cells grow normally on graphene, with simi-lar morphology and viability Furthermore, in the presence of osteogenicmedium, a graphene substrate accelerates stem cells differentiation intobone cells Lee et al [68] confirm these results, and further demonstratethat graphene acts as a “preconcentration platform”, due to its strongerability to adsorb various molecules such as adhesion proteins and growthfactors, compared to their reference substrate (PDMS) This higher con-centration of molecules is a likely explanation for the enhanced growth anddifferentiation of stem cells The same work also shows that insulin under-goes a confirmation change when bound to graphene, which inhibits stemcell differentiation to fat cells

Finally, some recent interest has been shown in depositing graphene in3D nickel foam [19] Once the nickel is etched, this leaves a 3D scaffold

that can be used for cell culture Li et al [72] show that such a scaffold isable to sustain the growth of neural stem cells, with potential applications

in tissue engineering, or basic cell biology experiments that could bettermimic in vivo environments

Chapter 2 will present the results of my own investigations in the topic,

in conditions similar to the two papers investigating stem cell tion [68, 83]: CVD-grown graphene transferred to different substrates

Graphene is of prime interest in sensing applications, as it is very sitive to adsorbed charges, and is intrinsically a low-noise material from anelectronic standpoint [103] Unlike bulk materials, its 2-dimensional naturemeans that added charges have an effect on the whole of its structure, in-stead of just the surface [3], as illustrated in Figure 1.5 Furthermore, itslow resistance and low number of crystal defects (for exfoliated graphene)help reduce the noise level, thereby increasing the signal to noise ratio [103].Even the seminal paper by Novoselov et al [87] mentions the sensitivity

sen-of graphene to its environment Adsorbed molecules dope the graphenesheet, causing a change in resistance They show that graphene is sensitive

to adsorbed water, ethanol and ammonia molecules Later on, Schedin et

al [103] show that a micrometer-sized exfoliated graphene device is able todetect adsorption and desorption of single molecules of nitrogen dioxide

Uses in biology Some interest has also been shown in using graphene as

a sensor in the context of biology Ohno et al [90] show that the position ofthe Dirac peak is sensitive to the pH of the solution used for electrolyte gat-

ing Also, adsorption of bovine serum albumin (BSA) can be detected fromthe Dirac peak shift Further steps are taken in later works, demonstrat-ing that functionalized graphene can be used to detect antibodies [81, 89],glucose levels [53], avidin and fibronectin [48], and dopamine produced byneuron cells [47].

One paper presents the detection of malaria-infected red blood cells in

a microfluidic setup, relying on difference in impedance of infected cells [5].Graphene can also be used to amplify electrical fields generated by cellsthemselves, such as cardiomyocites [20]

In another work, closer to the method described in this thesis, graphene

is able to detect bacteria (E Coli) [52] Graphene is functionalized with anantibody, bacteria binds to the antibody, and its negatively charged wallsdope the graphene, causing a Dirac peak shift Furthermore, addition ofglucose, and its metabolism by the bacteria, decreases the local pH, causinganother shift of the device Dirac peak In this work, I focus on mammalian,epithelial cells, and show that the device response is not due to a change

Graphene

Charges

+ Bulk material

+

Figure 1.5: Effect of charges on graphene vs bulk material Compared toother materials used for sensors, graphene has the advantage of atwo-dimensional structure: adsorbed charges have maximal effect on thewhole of the material structure, while, in bulk material, only the surface

is affected [3]

in pH, nor the direct interaction of cells with graphene.

In terms of detection methods, some papers limit themselves to usinggraphene as a conductive scaffold, and use cyclic voltametry to detect an-alytes [31, 41, 121] In that case, a alternating voltage is applied between

an electrode and the graphene sheet Plotting current versus voltage givesinsight on the nature and properties of ions in the electrolyte

Many other papers measure intrinsic electronic properties of graphene,

as it is doped by external charges, but rely either on a fixed gate [48, 53], orthe Dirac peak position only [52, 90] This is unlike this work, that performs

a full analysis of the electronic properties of graphene in terms of Dirac peakposition and mobility in Chapter 4 Only Dankerl et al [22] go further intheir analysis: their setup is able to measure carrier charge and density,

by using Hall-effect with a far more complex device geometry The simplersetup used in this thesis is unable to distinguish between capacitance andmobility changes

Finally, unlike many techniques presented in other works [5, 47, 53], themethod presented in this thesis relies on bare graphene, without function-alization

The starting idea of this thesis was to build a sensor that could sure cellular forces, using graphene Probing the interaction of cells withtheir underlying substrate is important for the understanding of cell mi-gration This process drives embryonic development, tissue repair, but alsometastasis in cancer [100]

mea-It is known that substrate rigidity guides cell motility [78] Also, it hasbeen shown that the rigidity can influence stem cell differentiation: softsubstrate, similar in stiffness to brain tissue, is neurogenic; stiffer substratedrives stem cells to differentiate into muscle cells (myogenic); while stiffestsubstrate, similar to bones, is osteogenic [28].

Understanding these phenomena requires measurements of forces erted by cells on their substrate, to understand how cells are sensing therigidity of the substrate, and the mechanism behind their motility [122].There are currently two main methods to do so, both falling under thename Traction Force Microscopy, where forces cause a substrate deforma-tion that is observed optically

ex-The first method mixes fluorescent beads with a soft, deformable strate (typically polyacrylamide gels) By observing the deflection of thebeads, and knowing the Young’s modulus of the substrate, it is possible tocompute forces exerted by cells on the substrate [15] The drawback of thisapproach is that the resulting deformation of the substrate is the result ofmany force vectors, and reconstructing the force field is complicated, andsometimes ambiguous [122]

sub-This issue can be solved by the second approach: micropatterning ofPDMS pillars Cells are seeded on top of soft pillars, and the deflection ofthe pillars is observed optically [107] This approach removes the mathe-matical complexity, as each pillar is deflected independently of the others,

so the force exerted on each pillar can be computed accurately more, the pillar radius and height can be modified to change the substratestiffness [122] However, this kind of substrate hinders cell motility, andhas limited resolution [120]

Further-Chapter 3 shows attempts at solving this problem from a different angle:

a transparent, graphene-based device would be built This device wouldstill allow imaging of the cells, thanks to graphene’s transparency, andprovide electrical read-out of the substrate deformation Several approachesare described, and even though none of these were successful, it lead to thediscovery of the effects of cells on graphene devices, described in Chapter 4

Many protocols in cell biology require counting the number of cells in

a flask or dish, and a variety of methods is available (see Table 1.1, [34]).The standard, and oldest, method, is to use a hemocytometer: cells aredetached from the flask, and a small amount of the suspension is insertedinto a thin, transparent chamber of known volume The chamber has agrid pattern on its surface to delineate a known area, and by countingmanually the number of cells, one can obtain the cell concentration [34].This method can be automated, with systems based on the same principle,but using computer vision algorithms to count the number of cells (e.g.Bio-rad TC20 Automated Cell Counter [11])

Other techniques such as flow cytometry and electronic particle ing, which flow particles through a narrow channel and detect cells viafluorescence or impedance are also available [34]

count-However, all these techniques have a major disadvantage: cells need to

be split, and separated into single cells to make sure that the cell count isaccurate Clumps of cell will cause inaccurate read-out, so will dead cells(this can be alleviated in optical methods by using a dead cell stain such as